Desorption through a

Oct 4, 2007 - Nitrate reduction pathways on Cu single crystal surfaces: Effect of oxide and Cl −. Dennis P. Butcher , Andrew A. Gewirth. Nano Energy...
0 downloads 0 Views 834KB Size
Download PDF
16428

J. Phys. Chem. C 2007, 111, 16428-16436

Quasi-Reversible Chloride Adsorption/Desorption through a Polycationic Organic Film on Cu(100) Duc-Thanh Pham,† Sung-Lin Tsay,‡ Knud Gentz,† Caroline Zoerlein,† Simone Kossmann,† Jyh-Shen Tsay,‡ Barbara Kirchner,† Klaus Wandelt,† and Peter Broekmann*,† Institute of Physical and Theoretical Chemistry, Bonn UniVersity, Wegelerstrasse 12, 53115 Bonn, Germany, and Department of Physics, National Taiwan Normal UniVersity, Taipei 116, Taiwan ReceiVed: May 7, 2007; In Final Form: July 31, 2007

Combined cyclic voltammetry and in situ scanning tunneling microscopy studies were employed to gain information about the interfacial structure of a chloride modified Cu(100) electrode surface exposed to an acidic electrolyte solution that contained redox-active dibenzylviologens (DBV, 1,1′-dibenzyl-4,4′-bipyridinium molecules). A particular focus of this contribution lies in the structural characterization of the electrode surface under nonequilibrium reactiVe conditions, for example, during the occurrence of an electron-transfer reaction. Typically, two pairs of clearly distinguishable current waves denoted as P1/P1′ and P2/P2′ appear in the cyclic voltammogram of Cu(100) in a mixture of 10 mM HCl and 0.1 mM DBVCl2, provided the cathodic potential limit remains restricted to values of Ework > -425 mV vs reversible hydrogen electrode. Systematic variations of the DBV solution concentration and the nature of the counterion strongly suggest that P1 has to be assigned to the first electron-transfer reaction reducing the dicationic DBV2+ to the radical monocationic DBV•+ species while P1′ represents the corresponding oxidation process. Not only solution but also preadsorbed viologen species are involved in this charge-transfer reaction. Triggered by the electron transfer, the more open DBV2+ ads “cavitand” structure formed on top of the preadsorbed c(2 × 2)-Cl layer prior to the electron transfer transforms into a more compact polymeric (DBV•+ ads)n stacking phase upon reaching P1. Both the reactants and products of the electron-transfer reaction form condensed and laterally ordered 2D phases. In particular, the quite stable (DBV•+ ads)n stacking phase maintains its structural integrity during the ongoing electron-transfer reaction involving solution species. Passing P2 in the cyclic voltammogram, however, initiates an order-disorder transition within the organic film with defect lines or point defects in the (DBV•+ ads)n stacking phase acting as active sites for this structural transition. The driving force for this further phase transition is the starting chloride desorption through the (DBV•+ ads)n film. In the presence of the covering viologen film, the chloride desorption occurs at a potential that is ∆Edesorp ≈ 100 mV lower than that in the pure supporting electrolyte pointing to a significant additional activation barrier for that process. Reduced monomeric and oligomeric viologen species reveal a significantly lower lateral mobility on the metallic substrate than that on the chloride lattice. In the reverse potential sweep, chloride anions are forced to readsorb on the metallic copper substrate through the disordered viologen film resulting in a full restoration of the c(2 × 2)-Cl lattice in direct contact to the metallic copper and, in addition, in the full restoration of the ordered (DBV•+ ads)n stacking phase on top of the chloride lattice.

1. Introduction

V2+ + e- T V•+

Viologens1,2

A further electron transfer converts the radical monocation to the uncharged viologen species according to eq 2.

(1,1′-4,4′-disubstituted-bipyridinium molecules) have attracted much attention in the field of surface electrochemistry within the last few decades due to their widespread applications as chromophores, electron-transfer mediators, and most recently as gating molecules3,4 in model systems for electronic devices that are based on molecular architectures. Their main advantage lies in their low-lying LUMOs and an almost ideal reversibility of the electron-transfer reaction.1,2 Dicationic viologen species can reversibly be transformed into the corresponding radical monocations according to eq 1. * To whom correspondence should be addressed. E-mail: broekman@ pc.uni-bonn.de. † Bonn University. ‡ National Taiwan Normal University.

V•+ + e- T V0

(1)

(2)

This latter transition is supposed to be less reversible than the first electron-transfer step.1,2 Most recent studies focused on the conductivity properties of individual molecules that can be addressed independently, for instance, by the tip of a scanning tunneling microscope even at electrified solid/liquid interfaces.3,4 Quite intriguingly, monoand multilayer films of dialkylated viologens immobilized on a gold electrode via terminating thiol groups reveal a “diode”like behavior. This has been tested by use of an external redox couple in solution.5,6 While the reduction of [Ru(NH3)6]3+ to [Ru(NH3)6]2+ species is “mediated” by the bipyridinium centers,

10.1021/jp073469q CCC: $37.00 © 2007 American Chemical Society Published on Web 10/04/2007

Quasi-Reversible Chloride Adsorption/Desorption the reverse oxidation process is blocked, quite in contrast to the reversible redox reaction taking place at the unmodified electrode surface.6 Since the intra- and intermolecular stereoelectronic properties of the redox-active components in the organic film are decisive for the current/voltage response and hence for the overall reactivity of the electrode, a molecular scale characterization of these films is needed. As long as we restrict ourselves to individual molecules or monolayers of electroactive species, it appears straightforward to combine classical electrochemical methods such as cyclic voltammetry with in situ scanning probes that provide atomic-scale information on the lateral ordering and orientation of the redox-active species. In the present work, we study the structural properties of an electroactive dibenzylviologen film that self-assembles on a chloride modified Cu(100) electrode surface.7-9 In this case, viologen adsorption is not driven by covalent bonding to the electrode but by attractive electrostatic interactions between the negatively charged chloride lattice preadsorbed on the Cu(100) surface and the dicationic viologen species. By use of in situ X-ray diffraction techniques, chloride anions have recently been proven to retain, to a large extent, their charges upon adsorption on Cu(100).10 Since there is no irreversible covalent bonding between the substrate and these viologens, they are capable of undergoing pronounced positional, translational, and orientational changes at the electrode surface upon electron transfer resulting in completely new equilibrium structures.9 These fundamental differences become obvious from a comparison of the viologen structures in the reduced redox state. Viologen radical monocations immobilized by thiol functionalities prior to reduction tend to form dimers in the condensed films due to directional π-π interactions and spin-pairing effects between neighbors, whose position and orientation are largely defined by the packing arrangement in the self-assembled monolayer (SAM)-like film.11 In contrast to that, we observe a trend toward oligo- and polymerization when the viologen radical monocations exhibit a higher degree of positional and orientational freedom,9 indicating that this polymeric form is the thermodynamically favored state of the reduced radical monocations on the surfaces. In the present study, we particularly address the issue of the structural integrity of the condensed viologen films during the electron-transfer reaction. The charge-transfer reactions are superimposed by the order-disorder transitions in the viologen films.9 Their origin, however, could not be resolved unambiguously so far. In this paper, we present for the first time experimental evidence to show that the observed disordering is provoked by a quasi-reversible chloride desorption process through the polycationic layer of reduced radical monocations and not by the ongoing electron-transfer reaction of the viologen solution species. 2. Experimental Section For preparation of all solutions, high purity water (Milli-Q purification system; conductivity < 18 MΩ‚cm; TOC < 5 ppb) and reagent grade chemicals were used. All electrolyte solutions were routinely deoxygenated with argon several hours before use. All potentials given in the text refer to a reversible hydrogen electrode (RHE). A Pt wire served as the counter electrode. For the Cu(100) single crystal, a surface orientation of less than 0.5° off the (100) plane was required in order to guarantee a reproducibly smooth surface. Prior to each experiment, the copper surface had to be etched in order to remove the native oxide film which is formed in air. Therefore, the sample was

J. Phys. Chem. C, Vol. 111, No. 44, 2007 16429

Figure 1. (a) Cyclic voltammogram of Cu(100) in 10 mM HCl (supporting electrolyte). (b) Hard-sphere model of the c(2 × 2)-Cl adlayer. (c)-(f) Potential dependent series of STM images showing morphological changes upon chloride desorption, 47 × 47 nm, It ) 1.7 nA, Ubias ) 255 mV: (c) Ework ) -220 mV; (d) Ework ) -320 mV; (e) Ework ) -350 mV; (f) Ework ) -400 mV. The inset in (e) represents a 3 × 3 nm section of the lower copper terrace showing the c(2 × 2)-Cl adlayer. The inset in (f) represents a 3 × 3 nm section of the lower copper terrace showing the bare Cu(100) surface.

immersed into 50% orthophosphoric acid. Subsequently, an anodic potential of 2 V was applied between the copper electrode and a platinum foil for about 20-40 s. After being etched, the copper surface was rinsed with degassed 10 mM hydrochloric acid solution and mounted into the electrochemical cell. All experiments routinely started with an electrochemical characterization of the Cu(100) surface in the pure supporting electrolyte (10 mM HCl). The working electrolyte (10 mM HCl and 0.1 mM DBVCl2) was introduced into the system under potential control within the double layer regime (e.g., at Ework ) 0 mV) where the DBV2+ species is thermodynamically stable. All microscopic experiments have been performed using a home built electrochemical scanning tunneling microscope (EC-STM). Further information about the STM setup and the tip preparation can be found in ref 12. 3. Results and Discussion Figure 1a displays the typical steady-state cyclic voltammogram (CV) of Cu(100) in the pure supporting electrolyte. An extended double layer regime between +180 and -190 mV is confined at the anodic limit by the starting oxidative copper dissolution reaction followed by the appearance of a pronounced cathodic peak centered at +190 mV in the corresponding reverse potential sweep. This current wave can be assigned to the redeposition of the afore dissolved copper material. The exponential increase of the cathodic current starting at Ework ) -300 mV is due to the onset of the hydrogen evolution reaction (HER). From previous STM studies it is known that the chloride anions form a c(2 × 2)-Cl adlayer on Cu(100).13,14 In 10 mM HCl this adlayer is stable in the potential range between copper dissolution and close to the onset of the hydrogen evolution reaction. A hard-sphere model of this overlayer is presented in Figure 1b. As reported by Magnussen and co-workers,14 the c(2 × 2)-Cl adlayer induces a “step faceting” along the closely packed chloride rows parallel to the 〈100〉 substrate directions due to significant changes in kink and step energies in the presence of this chloride adlayer (Figure 1b), thus resulting in

16430 J. Phys. Chem. C, Vol. 111, No. 44, 2007 a new surface equilibrium morphology. In turn, the starting chloride desorption causes the corresponding “defaceting” resulting in randomly oriented substrate steps. Such a process is exemplarily shown in Figure 1c-f, which is largely consistent with the results of Magnussen et al.14 Note that in 10 mM HCl chloride desorption starts at about Ework ≈ -300 mV. With higher potential sweep rates, small cathodic and anodic current features appear in the CV at potentials before reaching the HER regime. These were assigned to the chloride adsorption and desorption processes and related order-disorder transitions.14,15 While substrate steps are preferentially aligned parallel to the 〈100〉 directions in the presence of c(2 × 2)-Cl under saturation conditions (Figure 1c-d), these steps tend to round off in Figure 1e, which already represents a chloride submonolayer coverage (θ < 0.5 ML) at Ework ) -320 mV. While Magnussen et al. correlate the step defaceting with an order-disorder transition within the c(2 × 2)-Cl adlayer,15 we still observe ordered c(2 × 2)-Cl patches on extended substrate terraces while step edges are already noticeably rounded off (Figure 1e) in this initial stage of chloride desorption. Chloride desorption obviously starts from step edges, most likely from their lower sites. The breakdown of the chloride mediated step alignment is the consequence of an enhanced mobility of copper species diffusing along these step edges in the absence of chloride. Such a step edge diffusivity is significantly lowered under chloride saturation conditions only. Similar morphological phenomena are known from ultra-high-vacuum studies dealing with the dissociative adsorption of halogens on Cu(100).16 For example, step faceting on Cu(100) was observed for bromine surface concentrations only close to saturation of Θ ) 0.5 ML although laterally ordered patches of the c(2 × 2)-Br layer were already seen in the STM experiment in the submonolayer regime at about Θ ) 0.4 ML.16 At potentials below Ework ) -350 mV, only the bare Cu(100)-(1 × 1) lattice could be imaged (see inset in Figure 1f). Recent in situ X-ray experiments indeed confirm the completion of chloride desorption at these low potentials.10 Hence, hydrogen evolution on Cu(100) in 10 mM HCl is expected to take place on an almost halide free copper surface. Adding an electrolyte that additionally contains the redoxactive viologen (working electrolyte: 10 mM HCl and 0.1 mM DBVCl2 or 10 mM HCl and 1 mM DBVCl2) leads to drastic changes in the voltammetric behavior as evidenced in Figure 2a. As long as we restrict the cathodic potential limit in the CV experiment to potentials of Ework > -425 mV, we only have to deal essentially with the first electron-transfer step of the viologen redox system, the quasi-reversible reduction of the dications (DBV2+) to the radical monocations (DBV•+), and the corresponding oxidation reaction (eq 1). Note that the HER is shifted significantly toward lower potentials due to the presence of viologens that are expected to block reactive surface sites for the HER.9 Two pairs of current waves are observed in Figure 2a. The small and rather broad peak P1 (black curve) with EP1 ) -270 mV has been assigned to the reduction of DBV2+ to DBV•+ species, while P1′ with the peak maximum at EP1′ ) -220 mV represents the corresponding reoxidation process (due to different conditions of the RHE reference electrodes the CVs reported here are shifted by -30 mV compared to those presented in ref 9). From our previous in situ STM studies8,9 it is known that the redox processes of solution DBV on Cu(100) interfere with the redox processes of the preadsorbed species and related surface phase transitions. However, extra “prepeaks” that appear at potentials more positive than the main redox-current waves are

Pham et al.

Figure 2. (a) Cyclic voltammograms of Cu(100) in 10 mM HCl/0.1 mM DBVCl2 (black line) and Cu(100) in 10 mM HCl/1 mM DBVCl2 (broken gray line), dE/dt ) 10 mV/s. (b) Cyclic voltammograms of Cu(100) in 10 mM HCl/0.1 mM DBVCl2 (black line) and Cu(100) in 5 mM H2SO4/10 mM KBr/0.1 mM DBVCl2 (broken gray line), dE/dt ) 10 mV/s. The inset shows the linear dependence of the peak current of P2/P′2 (black line) on the potential sweep rate.

not observed in this particular case. For example, spike-like prepeaks were reported for various dialkylated viologens interacting with Hg17-19 or highly oriented pyrolytic graphite (HOPG)20-22 electrode surfaces. This phenomenon was explained in terms of a Faradaic process of adsorbed viologen species coupled with 2D phase transitions between gaslike phases of the oxidized and more condensed 2D films of the corresponding reduced viologen species. The involvement of redox processes of solution species in the observed peak system P1/P1′ can be clearly proven by concentration-dependent CV measurements. Peak current densities of P1/P1′ scale up almost linearly with the viologen concentration in solution as predicted by the Randles-Sevcik equation23 (eq 3) where Jp is the peak current density, D the

nF (RT )

Jp ) -0.44nF

1/2

D1/2ν1/2c0A-1

(3)

Quasi-Reversible Chloride Adsorption/Desorption viologen diffusion coefficient, ν the potential sweep rate, c0 the viologen concentration in the bulk solution, A the area of the electrode surface, T the temperature, n the number of transferred electrons, F the Faraday constant, and R the gas constant. By increasing the viologen concentration by a factor of 10 (Figure 2a), we noted an increase in the peak maximum of P1 from JP1 ) -0.13 µA/cm2 (black curve) to JP1 ) -1.1 µA/cm2 (gray broken curve). A square-root dependence of P1/P1′ indicates the dominance of bulk solution redox processes in the voltammetric behavior in the 10 mM HCl/1 mM DBVCl2 solution (data not shown here). At this point, it should be noted that the peak heights of P2/P2′ are almost independent of the viologen concentration in solution. In the CV obtained for the 10 mM HCl/1 mM DBVCl2 solution P2/P2′ appears only as a small shoulder at the “cathodic” tail of the peak system P1/P1′. Peak positions of P2/P2′ are slightly downward shifted in the 10 mM HCl/1 mM DBVCl2 solution by about ∆E ≈ 15 mV. As it will be argued below we can attribute this effect to the slight increase of the chloride concentration (note that the DBV2+ is added as chloride salt to the solution). The use of even higher viologen concentrations, for example 10 mM HCl/10 mM DBVCl2, leads to a complete suppression of the peak system P2/P2′ due to the dominance of P1/P1′. In the following, we will discuss in detail the origin of the peak system P2/P2′. In particular, it will be demonstrated that these current waves do not correspond to the redox processes of any adsorbed or solution DBV species but to a quasi-reversible order-disorder transition coupled with chloride desorption/adsorption processes through a condensed viologen monolayer film. A first experimental hint supporting this idea comes from a comparison of the voltammetric behavior of DBV in chloride and bromide containing electrolytes (Figure 2b). Also, bromide anions form a c(2 × 2) adlayer on Cu(100).16,24,25 However, since the Cu-Br bond is stronger, bromide anions desorb from the electrode surface at more negative potentials.24 While chloride desorption in 10 mM HCl starts already at Ework ≈ -300 mV (Figure 1), bromide desorption takes place at potentials Ework < -350 mV, which is already within the regime of massive hydrogen evolution. Note that the observed downward shift of the bromide desorption potential is not only due to the stronger copper-halide interaction but it is also further promoted by the weaker trend toward solvation in the case of bromide compared to that of chloride anions. Assuming that P2/P2′ (black curve in Figure 2b) indeed involve chloride desorption/adsorption and related orderdisorder transitions of the c(2 × 2) adlayer and following the reasoning above we would not expect such a pair of peaks for the bromide containing electrolyte, at least within the given potential range. The CV obtained for the bromide containing solution (gray broken curve in Figure 2b) reveals only the peak pair P1/P1′ that corresponds to the electron-transfer reaction and there is no peak pair P2/P2′ pointing to bromide desorption/ readsorption and related order-disorder transitions. Interestingly, the peak maxima of P1 and P1′ are almost at the same potentials as those in the chloride containing electrolyte suggesting that in this particular case there is no significant impact of the nature of the halide counterion on the DBV redoxpotentials. This kind of anion effect on the viologen redox potentials is well known, for instance, from anions revealing electron donor capabilities.1,2,26 A further observation pointing to a chloride desorption/ adsorption process is the slight downward shift of P2/P2′ by ∆E ≈ 15 mV when the chloride concentration is increased upon changing the DBV solution concentration. The inset in Figure

J. Phys. Chem. C, Vol. 111, No. 44, 2007 16431

Figure 3. Cyclic voltammograms of Cu(100) in 10 mM HCl/1 mM DBVCl2 depending on the cathodic potential limit.

2b indicates a linear dependence of the peak current density of P2/P2′ on the potential sweep rate also suggesting that P2/P2′ is most likely related to a surface process. Typically, the peak position of P2 is less affected by the sweep rate (dE/dt) than that of P2′. By increasing (dE/dt) from 5 to 150 mV/s, we observe a downward shift of only 12 mV for P2 while P2′ reveals an upward shift of about 70 mV, suggesting that the latter process is even more hindered than the one represented by P2 (data not shown here). A further electron-transfer reaction is not expected for the potential regime where P2/P2′ appears in the CV. This becomes obvious from a comparison of CVs of Cu(100) in 10 mM HCl/1 mM DBVCl2 obtained for two different cathodic limits (Figure 3). Only if we decrease the cathodic potential limit, for example, from Ework ) -395 mV (black curve in Figure 3) to Ework ) -630 mV (gray dashed curve in Figure 3), we initiate the second electron-transfer reaction (see eq 2) that can be correlated with the appearance of P3 at Ework ) -440 mV (cathodic sweep). The current features in the corresponding anodic potential sweep appear more complex. P3′ corresponds to the reoxidation of the uncharged species to the corresponding radical monocations, while the huge peak P1′ has been assigned to the oxidation of the radical monocations from the (DBV•+ + Cl-)n precipitate that is most likely formed upon the conproportionation reaction between the uncharged DBV0 formed below Ework ) -440 mV and the dicationic DBV2+ solution species.1,2 Previous in situ STM studies gave clear evidence for a surface phase-transition upon reaching peak P1 (cathodic potential 8,9 While the sweep) involving the preadsorbed DBV2+ ads phase. 2+ preadsorbed DBVads species form a more open “cavitand” structure,7-9 the reduced viologen species DBV•+ ads self-asentities. This semble into compact chains of stacked DBV•+ ads surface phase-transition occurs via nucleation and growth of islands of the (DBV•+ ads)n stacking phase within the matrix of 9 the preexisting DBV•+ ads “cavitand” structure. There is even a certain potential range between Ework ≈ -150 and -230 mV

16432 J. Phys. Chem. C, Vol. 111, No. 44, 2007

Figure 4. Structural motifs of DBV adlayers on c(2 × 2)-Cl depending on the viologen redox-state: (a) cavitand phase, 8.3 × 8.3 nm; (b) stacking phase, 8.3 × 8.3 nm.

in the cathodic potential sweep in which phases of the oxidized and the reduced viologen species do coexist. Note that the coexistence regime is not identical for the cathodic and the anodic potential sweep. A quantitative analysis of these hysteresis effects will be addressed in detail in a separate paper.27 Since the DBV2+ ads cavitand phase and the related structural transitions have already been described in much detail,7-9 we present in Figure 4 only a brief overview of this process. The stacked arrangement of DBV•+ ads molecules becomes possible due to significant intramolecular changes upon electron transfer affecting in particular the interplanar dihedral angle Φ between the central bipyridine rings.1,2 Recent DFT results indicate that radical monocations typically exhibit an interplanar dihedral angle that is close to zero (ΦDBV•+ ) -5.4°) while the corresponding dicationic species show a significantly larger dihedral angle of ΦDBV2+ ) 40.1° (Figure 4).9 STM results suggest that the DBV•+ ads molecules are adsorbed with their molecular N,N-axis parallel to the surface in a so called “edgeon” or “side-on” adsorption geometry.8,9 In combination with a trans conformation of the benzyl substituents this orientation allows the DBV•+ ads entities to assemble themselves into the observed (DBV•+ ads)n oligo- or polymeric stacking chains. From our STM results it is further known that the electrontransfer reaction and the surface phase-transition take place in the presence of the ordered c(2 × 2)-Cl adlayer.9 Drastic tunneling conditions (low bias and high tunneling current) allow the local removal of the viologen layer by using the tunneling tip as an atomic-scale “brush”. The chloride lattice underneath, by contrast, remains fully intact by such a treatment and can be imaged even with high brilliance. Such an experiment is presented in Figure 5a,b. Moderate tunneling conditions allow the imaging of the stacked DBV•+ ads entities (Figure 5a) revealing an almost ideal “face-to-face” arrangement of neighboring DBV•+ ads molecules. It is assumed that the redox-active (reduced) bipyridinium moieties are imaged as bright and elongated STM dots under these tunneling conditions while the benzyl groups are supposed to lie within the dark ditches exhibiting a lower imaging contrast. On a mesoscopic length scale, these (DBV•+ ads)n polymer chains arrange themselves into extended rotational and mirror domains (Figure 5c). Figure 6 summarizes the interfacial structure right after the surface phase-transition triggered by the electron transfer. A quite compact polycationic layer covers the anionic chloride layer, both together forming a paired anion-cation layer (Figure 6a). The driving forces for the formation of the observed polymeric (DBV•+ ads)n stacking chains are intermolecular π-π

Pham et al.

Figure 5. (a) Atomically resolved STM image of the (DBV•+ ads)n stripe pattern, 4.25 × 4.25 nm, It ) 40 nA, Ubias ) 28 mV, Ework ) -200 mV (cathodic potential sweep). (b) Atomically resolved STM image of the c(2 × 2)-Cl adlayer after the tip induced removal of the (DBV•+ ads)n stripe pattern, 4.25 × 4.25 nm, It ) 9 nA, Ubias ) 1 mV, Ework ) -130 mV. (c) Rotational and mirror domains of the (DBV•+ ads)n stripe pattern, 43 × 43 nm, It ) 16 nA, Ubias ) 112 mV, Ework ) -340 mV.

Figure 6. (a) Out-of-plane structure model of the interface in the presence of the paired anion-cation layer. (b) In-plane structure model of the (DBV•+ ads)n stripe pattern on top of the c(2 × 2)-Cl adlayer as derived from Figure 5.

interactions between neighboring benzyl groups and even stronger interactions between neighboring reduced bipyridinium moieties, probably due to spin pairing.1,2 Attractive intermolecular interactions apparently overcome the repulsive electrostatic interactions of positively charged monocations within this compact film. Note that the nearest neighbor distance (NND) within the (DBV•+ ads)n stacking chains amounts only to 0.37(2)

Quasi-Reversible Chloride Adsorption/Desorption nm. This value is typical for π-stacking assemblies of aromatic systems and has also been observed, for example, for 2,2′bipyridine phases on Au(111)28 and Au(100).29 Strong attractive intermolecular interactions as the driving force for dimerization of open-shell planar aromatic species in aqueous solutions were first reported by Hausser and co-workers in 1957.30 The face-to-face orientation of adjacent radical species allows an overlap of the singly occupied π* orbitals of the neighboring aromatic ring systems. This phenomenon is known not only for viologens31 but also for reduced naphtyl,32 anthracyl,32 and pyridinium33 derivatives in aqueous solution. The formation of a dimer species is also discussed for viologen radical monocations that were immobilized before reduction on electrode surfaces via thiol or disulfide functional groups.4,11 In these SAM-like films, the dimerization is favored over the extended polymerization due to the lack of lateral mobility of the covalently immobilized viologens. It is noteworthy to mention that the presence of the anionic chloride adsorbed on the electrode surface is not mandatory for the formation of these (DBV•+ ads)n stacking chains. Similar structural motifs with almost identical NNDs of 0.37(15) nm within (DBV•+ ads)n stacking chains are observed on HOPG using the same working electrolyte.9 On this latter substrate there is no specifically adsorbed chloride layer present. Here, the phase formation of the reduced viologen species takes place directly on the unmodified HOPG surface. The main difference between the (DBV•+ ads)n chains on HOPG and those on Cu(100)-c(2 × 2)-Cl lies in their stability and robustness against external mechanical stress induced, for instance, by the tunneling tip upon scanning. While on HOPG these stripe pattern phases of adsorbed DBV•+ ads are extremely fragile and, hence, often destroyed upon scanning even at moderate tunneling conditions, we can apply, by contrast, even much more drastic tunneling conditions for the imaging of the (DBV•+ ads)n stripe pattern phases on Cu(100)-c(2 × 2)-Cl. Note that the STM image in Figure 5a was obtained by using a tunneling current of It ) 40 nA and a bias voltage of Ubias ) 28 mV. This extraordinary stability can be clearly attributed to the presence of the anionic chloride layer under the polycationic (DBV•+ ads)n stacking phase. A coadsorption of the halide anions within or between (DBV•+ ads)n stacking chains appears unlikely since identical structure motifs are observed also on the c(2 × 2) lattice of bromide anions and on the uniaxially expanded c(p × 2) lattice of iodide anions.34 It should be stressed that the electron-transfer reaction does not come to a standstill after completion of the surface phasetransition presented in Figure 4. There is an ongoing reduction of the dicationic solution species to the corresponding solution radical monocations but now in the presence of the polycationic (DBV•+ ads)n stacking phase on top of the c(2 × 2)-Cl adlayer (Figure 6a). Apparently, the ongoing electron-transfer reaction does not seriously affect the structural integrity of the (DBV•+ ads)n stacking phase on top of the chloride lattice, at least in a mixture of 10 mM HCl and 0.1 mM DBVCl2 solution. 2+ Solvated (DBV•+ solv) monomers or (DBVsolv)2 dimers as reaction products are transported back into the bulk of the solution. Their solubility in the chloride containing electrolyte is still sufficient to prevent a thick film formation of chloride salts of the radical monocations, at least at these low reaction rates. From STM experiments alone it cannot be concluded where and how this further reaction takes place, either by an electron transfer through the paired anion-cation layer at terraces ((1) in Figure 6a), which might involve electron transfer via electronic states of the viologen moiety, or at defect sites such as domain boundaries

J. Phys. Chem. C, Vol. 111, No. 44, 2007 16433

Figure 7. Series of STM images showing the reversible potential dependent decay and formation of the (DBV•+ ads)n stripe pattern: (a)(h) 29 × 29 nm, It ) 0.1 nA, Ubias ) 151 mV.

of the viologen stacking layer and substrate step edges ((2) in Figure 6a) or via an exchange process of an already reduced 2+ DBV•+ ads by an oxidized DBVsolv species followed by its reduction directly on top of the chloride layer ((3) in Figure 6a). A further decrease of the electrode potential causes an orderdisorder transition within the (DBV•+ ads)n stacking phase (Figure 7). Phenomenologically, this transition resembles potential driven 2D dissolution/melting processes reported for various condensed and laterally ordered 2D organic films adsorbed on metallic electrode surfaces.35,36 As stated by Poelman et al.,35 these processes preferentially start from pre-existing defects in the organic film such as line defects in terms of translational, rotational, or mirror domain boundaries between ordered patches of the organic film that are formed upon collision of expanding 2D islands during the initial film growth. Also, the substrate step edges can be considered as line defects. Point defects result

16434 J. Phys. Chem. C, Vol. 111, No. 44, 2007

Pham et al.

Figure 8. (a) Disordered DBV•+ layer imaged under moderate tunneling conditions, 37 × 37 nm, It ) 0.2 nA, Ubias ) 286 mV, Ework ) -415 mV. (b) Cu(100)-(1 × 1)-lattice imaged under more drastic tunneling conditions, 7 × 7 nm, It ) 1.7 nA, Ubias ) 1 mV, Ework ) -500 mV.

from the intersection of at least three domain boundaries (see white dotted circles in Figure 7a). In fact for the dissolution of the (DBV•+ ads)n stacking phase, these kinds of defects act exclusiVely as the active sites. The activation free energy for the dissolution at these defects is significantly lower than the one necessary to create new point defects within intact domains. The structural transition is completed after passing P2. In contrast to the reported order-disorder transitions of organic films, it is not a “gaseous”, an “expanded-liquid”-, or a “condensed-liquid”-like phase of laterally mobile organic molecules that is left behind after completion of the phase transition. Instead, the amorphous viologen phase seems to be composed of randomly distributed and immobilized entities (Figure 7d). A further deviation from the reported examples of defect mediated dissolution processes is the respective underlying driving force. While it is usually a temperature or potential induced reorientation of adsorbed molecules that might be associated with the partial desorption/adsorption of the organic molecules initiating these transitions,36 here, it is the starting chloride desorption though the organic film that drives the observed order-disorder transition. This hypothesis is further supported by a tip induced local removal of the viologen layer as demonstrated in Figure 8. While the disordered viologen phase could be obtained under moderate tunneling conditions (Figure 8a), it is possible to image a square lattice using more drastic tunneling conditions (Figure 8b). An NND ) 0.25(2) nm gives clear evidence for the presence of the bare Cu(100)(1 × 1) lattice under the disordered phase. Apparently, the DBV•+ ads molecules do not accompany the desorbing chloride anions into the bulk of the solution. Such an anion “carrier effect” is, for instance, known from polycationic porphyrin layers adsorbed on an ordered sulfate/water coadsorption layer on Cu(111).37 Here, the sulfate desorption leads to a concerted desorption of the cationic porphyrin molecules. In turn, sulfate adsorption initiates the coadsorption of the polycationic porphyrins also resulting in a laterally ordered paired anion-cation layer. In the present case, by contrast, the cationic organic molecules remain on the bare metal surface after anion desorption. The missing lateral order within the viologen layer right after (partial) chloride desorption points to a strongly reduced diffusivity of the radical monocations on the bare metal surface. Two reasons can be assumed for that effect: (1) Radical monocations interact much stronger with the metallic copper substrate than with the anionic chloride lattice thus preventing a high lateral mobility of the viologens. (2) It can further be assumed that the chloride desorption is not fully completed even in the presence of the disordered phase. In this scenario a disordered chloride submonolayer still in contact with the copper prevents a lateral ordering of the viologens on the metallic copper substrate (Figure 9b). How the starting chloride desorp-

Figure 9. (a) Schematic drawing showing the chloride desorption through the polycationic organic layer. (b) Schematic drawing showing the disordered DBV+• layer under reactive conditions.

tion/adsorption process affects the polycationic viologen film is illustrated in Figure 9. It should be mentioned that the electrode is still under reactive conditions even after chloride desorption. Ongoing electron-transfer processes now take place either through the disordered viologen layer or at defects in the disordered viologen film at the bare metal surface (Figure 9b). Not only the chloride desorption has an influence on the lateral order of the (DBV•+ ads)n stacking phase but also the presence of the viologen layer affects the chloride desorption/ adsorption process. Compared to the pure supporting electrolyte (Figure 1), the chloride desorption is shifted toward lower potentials in the presence of the viologen up to ∆Edesorp ≈ 100 mV. While chloride desorption in the pure supporting electrolyte has already started at Ework ≈ -330 mV,10 both the (DBV•+ ads)n stacking phase and the chloride lattice underneath are still intact at Ework ) -330 mV in the DBV containing electrolyte as evidenced in Figure 7a. The observed downward shift of the chloride desorption can simply be understood in terms of an additional activation barrier for chloride desorption when the covering viologen film is present. While chloride desorption in the pure supporting electrolyte exclusively starts at the step edges, we observe, in addition, chloride desorption from terraces in the presence of the viologen film with line and point defects in the covering (DBV•+ ads)n stacking phase, which act as active sites for that process. It should also be noted that the shift of the chloride desorption potential depends on the structural quality of the covering viologen film. A high defect density within the viologen film generally leads to a smaller shift of the chloride desorption. In general, the observed order-disorder transition is quasi-reversible with a remarkably small potential hysteresis of about ∆Ehyst ≈ 70 mV (Figures 2b and 7). In the reverse potential sweep, chloride anions are forced to readsorb

Quasi-Reversible Chloride Adsorption/Desorption onto the copper surface once potentials close to P2′ are reached. Therefore, chloride anions have to not only penetrate into the disordered viologen film but also displace DBV•+ ads molecules from the metallic copper surface and order themselves again into patches of the c(2 × 2)-Cl phase in the presence of the disordered viologen phase. In this sense, the chloride anions “crawl” under the polycationic amorphous viologen layer. Driven by the chloride readsorption two coupled disorder-order transitions take place successively. The first one affects the ordering of chloride anions that are in contact with the metallic substrate and the other one affects the viologen layer. The lateral diffusivity of DBV•+ ads molecules is again enhanced on top of the chloride lattice compared to that of the bare metal surface, thus allowing the fast recombination of DBV•+ ads molecules into the (DBV•+ ) stacking phase on top of the locally restored c(2 n ads × 2)-Cl phase (Figures 7e-g). The reorganization of the c(2 × 2)-Cl phase is considered as rate determining. Sweep-ratedependent CV measurements as discussed above suggest a kinetical hindrance that is stronger for the disorder-order transition (P2′) than for the preceding order-disorder transition (P2). In general, the chloride mediated disorder-order transition also obeys with the typical 2D nucleation and growth behavior and the coexistence of two discernible phases (disordered/2Dordered) within a certain potential regime between P2 and P2′. Ongoing nucleation during the disorder-order transition combined with a relatively slow growth of the stable 2D nuclei leads to the appearance of a high number of growing patches of the (DBV•+ ads)n stacking phase and consequently to a high density of energetically unfavorable line and point defects. However, postgrowth ripening effects tend to reduce the total length of domain boundaries (white arrows in Figure 7f-g). The role of the chloride anions in the course of the lateral ordering of DBV•+ ads molecules is not a templating one in the sense of controlling the DBV•+ ads ordering process due to strongly modulated adsorbate-substrate interactions. The chloride layer acts here more as a “buffer layer” between the DBV•+ ads film and the metallic copper substrate. This allows the DBV•+ ads entities to self-organize at the surface. Such an enhanced lateral mobility of cationic organic molecules has been reported on iodide modified metal electrodes.38 4. Conclusions In this contribution, we have demonstrated that the appearance of two well defined pairs of current peaks in the CV of Cu(100) exposed to a mixture of 10 mM HCl and 0.1 mM DBVCl2 has been attributed to an electron-transfer reaction (DBV2+/ DBV•+) involving solution and surface species (P1/P1′) followed, at more negative potentials, by a quasi-reversible orderdisorder transition in the paired anion-cation layer (P2/P2′). The latter process is triggered by the starting chloride desorption/ adsorption through a condensed film of reduced viologen species. Compared to the pure supporting electrolyte (10 mM HCl), the onset potential for chloride desorption is downward shifted. This effect can be explained in terms of additional activation barriers for chloride desorption in the presence of the viologen film. Quite surprisingly, the chloride anions can readsorb on the Cu(100) surface by “crawling” under the disordered viologen film, thereby reconstituting the c(2 × 2)Cl phase. This leads to a fast ordering of the reduced viologen species into the laterally well developed stacking phase, since the lateral diffusivity of DBV•+ ads species is much higher on the chloride lattice than that on the metal surface.

J. Phys. Chem. C, Vol. 111, No. 44, 2007 16435 Acknowledgment. This work has been supported by the collaborative research center SFB 624 of the German Science Foundation. References and Notes (1) Bird, C. L.; Kuhn, A. T. Chem. Soc. ReV. 1981, 10, 49. (2) Monk, P. M. S. The Viologens: physicochemical properties, synthesis, and applications of the salts of 4,4′-bipyridine; John Wiley & Sons: Chichester, England, 1989. (3) (a) Xu, B.; Tao, N. J. Science 2003, 301, 1221. (b) Haiss, W.; van Zaling, H.; Higgins, S. J.; Bethell, D.; Ho¨rbenreich, H.; Schiffrin, D. J.; Nichols, R. J. J. Am. Chem. Soc. 2003, 125, 15294. (c) Haiss, W.; Nichols, R. J.; van Zaling, H.; Higgins, S. J.; Bethell, D.; Schiffrin, D. J. Phys. Chem. Chem. Phys. 2004, 6, 4330. (4) Li, Z.; Han, B.; Meszaros, G.; Pobelov, I.; Wandlowski, Th.; Blaszczyk, A.; Mayor, M. Faraday Discuss. 2006, 131, 121. (5) Gittins, D. I.; Bethell, D.; Nichols, R. J.; Schiffrin, D. J. J. Mater. Chem. 2000, 10, 79. (6) Raymo, F. M.; Alvarado, R. J.; Pacsial, E. J.; Alexander, D. J. Phys. Chem. B 2004, 108, 8622. (7) Safarowsky, C.; Wandelt, K.; Broekmann, P. Langmuir 2004, 20, 8261. (8) Safarowsky, C.; Rang, A.; Schalley, C. A.; Wandelt, K.; Broekmann, P. Electrochim. Acta 2005, 50, 4257. (9) Pham, D.-T.; Gentz, K.; Zoerlein, C.; Hai, N. T. M.; Tsay, S.-L.; Kirchner, B.; Kossmann, S.; Wandelt, K.; Broekmann, P. New J. Chem. 2006, 30, 1439. (10) Huemann, S.; Hai, N. T. M.; Broekmann, P.; Wandelt, K.; Zajonz, H.; Dosch, H.; Renner, F. J. Phys. Chem. B 2006, 110 (49), 24955. (11) Tang, X.; Schneider, T. W.; Walker, J. W.; Buttry, D. A. Langmuir 1996, 12, 5021. (12) Wilms, M.; Kruft, M.; Bermes, G.; Wandelt K. ReV. Sci. Instrum. 1999, 70 (7), 3641. (13) Suggs, D. W.; Bard, A. J. J. Phys. Chem. B 1995, 99, 8349. (14) (a) Vogt, M. R.; Lachenwitzer, A.; Magnussen, O. M.; Behm, R. J. Surf. Sci. 1998, 399, 49. (b) Magnussen, O. M. Chem. ReV. 2002, 102, 679. (15) Bae, S.-E.; Gewirth, A. A. Langmuir 2006, 22, 10315. (16) Nakakura, C. Y.; Altman, E. I. J. Vac. Sci. Technol., A 1998, 16 (3), 1566. (17) Kitamura, F.; Ohsaka, T.; Tokuda, K. J. Electronal. Chem. 1993, 353, 323. (18) Kitamura, F.; Ohsaka, T.; Tokuda, K. J. Electronal. Chem. 1993, 347, 371. (19) Sagara, T.; Miuchi, K. J. Electronal. Chem. 2004, 567, 193. (20) Arihara, K.; Kitamura, Nukanobo, K. F.; Ohsaka, T.; Tokuda, K. J. Electronal. Chem. 1999, 473, 138. (21) Sagara, T.; Tanaka, S.; Fukuoka, Y.; Nakashima, N. Langmuir 2001, 17, 1620. (22) Sagara, T.; Tanaka, S.; Miuchi, K.; Nakashima, N. J. Electronal. Chem. 2002, 524-525, 68. (23) Instrumental Methods in Electrochemistry/Southampton Electrochemistry Group; Ellis Horwood Series in Physical Chemistry; Ellis Horwood: Chichester, 1985. (24) (a) Broekmann, P.; Anastasescu, M.; Spaenig, A.; Lisowski, W.; Wandelt, K. J. Electroanal. Chem. 2001, 500, 241. (b) Saracino, M.; Gentz, K.; Becker, M.; Ko¨nigshoven, P.; Zajonz, H.; Dosch, H.; Wandelt, K; Broekmann, P. Unpublished work. (25) (a) Nakakura, C. Y.; Altman, E. I. Surf. Sci. 1997, 370, 32. (b) Nakakura, C. Y.; Altman, E. I. J. Vac. Sci. Technol., A 1997, 15 (4), 2359. (c) Nakakura, C. Y.; Altman, E. I. Surf. Sci. 1998, 398, 281. (26) Sagara, T.; Maeda, H.; Yuan, Y.; Nakashima, N. Langmuir 1999, 15, 3823. (27) Pham, D.-T.; Tsay, S.-L; Tsay, J.-S.; Wandelt, K.; Broekmann, P. Unpublished work. (28) Dretschkow, Th.; Wandlowski, Th. Electrochim. Acta 1999, 45, 731. (29) Dretschkow, Th.; Wandlowski, Th. J. Electroanal. Chem. 1999, 467, 207. (30) Hausser, K. H.; Murell, J. N. J. Chem. Phys. 1957, 27, 500. (31) Imabayashi, S.-I.; Kitamura, N.; Tazuke, S.; Tokuda, K. J. Electroanal. Chem. 1988, 243, 143. (32) Hirayama, F. J. Chem. Phys. 1965, 42, 3163. (33) Kosower, E. M. In Free Radicals in Biology, Pryor, W. A., Ed.; Academic Press: New York, 1976; Vol. 2, chapter 1. (34) Gentz, K.; Pham, D.-T.; Wandelt K.; Broekmann P. Unpublished work.

16436 J. Phys. Chem. C, Vol. 111, No. 44, 2007 (35) Poelman, M.; Buess-Hermann, C.; Badiali, J.-P. Langmuir 1999, 15, 2194. (36) Wandlowski, Th. Phase Transitions in Two-dimensional Adlayers at Electrode Surfaces: Thermodynamics, Kinetics, and Structural Aspects. In Encyclopedia of Electrochemistry; Urbakh, M. I., Gileadi, E., Eds.; WileyVCH: Weinheim, 2002; Vol. 1, p 383.

Pham et al. (37) Hai, T. N. M.; Wandelt, K.; Broekmann, P. Unpublished work. (38) (a) Batina, N.; Kunitake, M.; Itaya, K. J. Electroanal. Chem. 1995, 405, 245. (b) Kunitake, M.; Batina, N.; Itaya, K. Langmuir 1995, 11, 2337. (c) Ogaki, K.; Batina, N.; Kunitake, M.; Itaya, K. J. Phys. Chem. 1996, 100, 7185.