Electrochemical oxidation and ex-situ STM observations of ...franklin.chm.colostate.edu/bap/PDF of Papers copy/J... · Princeton Applied Research 174A polarographic ana-lyzer controlled

www.elsevier.nl/locate/jelechem

Journal of Electroanalytical Chemistry 498 (2001) 19–33

Electrochemical oxidation and ex-situ STM observations ofbis(4-dimethylamino-2-dihydroxyphenyl)squaraine dye layers on

HOPG electrodes

Norihiko Takeda, Michele E. Stawasz, Bruce A. Parkinson *Department of Chemistry, Colorado State Uni6ersity, Fort Collins, CO 80523-1872, USA

Received 13 April 2498; received in revised form 7 June 2000; accepted 7 June 2000

This paper is a contribution to this special issue honoring Professor Fred C. Anson, my mentor and friend, who has exhibited the utmostclass and integrity throughout his distinguished career in electrochemistry

Abstract

Electrochemical oxidation of insoluble highly ordered bis(4-dimethylamino-2-dihydroxyphenyl)squaraine (1-1OHSQ) dye layersadsorbed on highly oriented pyrolytic graphite (HOPG) electrodes was studied in aqueous electrolytes. Staircase cyclicvoltammetry obtained in chloride electrolytes revealed hysteresis characterized by large peak separations (100–200 mV) and sharpredox peaks (full width at half maximum (fwhm) 10–60 mV) the shape and potentials of which depended on electrolyteconcentrations. Small stochastic reduction peaks were observed at more negative potentials that are associated with the reductionof small domains of the oxidized 1-1OHSQ layers. Peak potentials and peak shapes were also dependent on the identity of theelectrolyte anion species. The results support a reaction scheme where electrolyte anions are incorporated into surface confinedone-electron oxidized dye molecular layers. From the results of peak potential shifts, the preference for ion-pairing is estimatedto be in the order of: I−\Br−\Cl−\SO4

2−:ClO4−:F−. Ex-situ STM observations were performed to reveal structural

changes upon electrochemical oxidation of 1-1OHSQ layers in LiCl electrolytes. Two polymorphs with different angles of thelong-axis of molecules to the directions of the molecular row were observed for the oxidized samples. Both polymorphs had largerpacking densities compared to the reduced form of the dye. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Surface confined redox processes; Squaraine dye; Two-dimensional surface immiscibility; Cyclic voltammetry; Scanning tunnelingmicroscopy

1. Introduction

Surface confined voltammetry of molecular adsor-bates on highly oriented pyrolytic graphite (HOPG)was first reported by Brown, Koval, and Anson for thereduction and reoxidation of aromatic compounds in-cluding phenanthraquinone [1,2]. The voltammogramsexhibited typical surface confined reversible behavior.More recent studies of surface confined electrochemicalreactions have reported systems where there is consider-able hysteresis in the voltammograms, with the oxida-

tion and reduction peaks separated by as much as 250mV or even more [3–27]. These reactions usually in-volve tightly packed two-dimensional surface phases,such as azobenzene Langmuir–Blodgett films [3] ormethyl viologen cation radicals [4], although the peakseparation in the latter was rather small. Similar behav-ior was also observed for three-dimensional conductingphases such as the organic conducting salts of tetrathia-fulvalene (TTF) [5–8,13] and 7,7,8,8-tetracyanoquino-dimethane (TCNQ) [5–7,9–13], C60 crystallites[14–23], and liquid crystal perylene diimides [24], aswell as non-conducting solid organometallic com-pounds [25–27]. Often the electrochemical reaction ofthese phases involves the incorporation of anions orcations into the oxidized or reduced phase. The largestructural changes, necessary due to the need to accom-

* Corresponding author. Tel.: +1-970-4910504; fax: +1-970-4911801.

E-mail address: [email protected] (B.A. Parkin-son).

0022-0728/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.PII: S 0022 -0728 (00 )00214 -X

N. Takeda et al. / Journal of Electroanalytical Chemistry 498 (2001) 19–3320

modate ions in the adsorbed film or molecular crystal,result in a large hysteresis in the voltammograms. Thebarrier to nucleation and growth of the new phaseformed upon oxidation or reduction was used to ex-plain and model many of the electrochemical results[4,8,11–13,23,28]. Recently Scholz et al. [29,30] havepointed out that the hysteresis in voltammograms maynot be a result of nucleation kinetics but instead is athermodynamic result due to a miscibility gap betweenthe two phases. If the oxidized and reduced surfacephases are both insoluble in the electrolyte and immis-cible then there is no mechanism for the equilibrationof the two phases. In the case where a neutral surfacephase is oxidized or reduced and ions need to beincorporated into the new phase, it is not surprisingthat the two-dimensional neutral and two-dimensionalionic phases are immiscible.

We report the investigation of a new surface confinedredox system, involving squaraine dye molecules[bis(4-dimethylamino-2-dihydroxyphenyl)squarine (1-1OHSQ), see structure below] that exhibit remarkablevoltammetry on HOPG electrodes.

In addition, we use scanning tunneling microscopy(STM) to study the actual molecular arrangements ofthese molecules on the surface of HOPG in both theiroxidized and reduced states.

2. Experimental

2.1. Chemicals

Bis(4-dimethylamino-2-dihydroxyphenyl)squaraine(1-1OHSQ) was a generous gift from the LexmarkInternational Corporation and was used as received.Dye stock solutions were prepared by dissolving 1-1OHSQ in dichloromethane (Fischer, Spectranalyzedgrade). The actual concentration of dye solutions werechecked by measuring UV–vis absorption spectra usinga HP8452A spectrophotometer (o=3.3×105 cm−1

M−1 at 636 nm [31]). Lithium chloride, sodium bro-mide, sodium fluoride, sodium iodide, and sodium car-bonate were Fischer chemical reagents andhydrochloric acid, potassium chloride, sodium chloride,and sodium sulfate were Mallinckrodt analyticalreagents, and were used as received. A 2.0 M sodiumperchlorate stock solution was prepared from dilutedperchloric acid solution and sodium carbonate.Aqueous electrolyte solutions were prepared using wa-ter purified with a Millipore Milli-Q system.

2.2. Electrode preparation

Highly oriented pyrolytic graphite (HOPG) was cutinto small pieces to prepare working electrodes. Basalplanes of the HOPG pieces were cleaved with adhesivetape (Scotch, 3M) to expose fresh surfaces. Depositionof 1-1OHSQ dye layers on HOPG substrates was car-ried out by similar methods employed in our previousphotoelectrochemical study of 1-1OHSQ on tindisulfide [32]. In this study, pieces of freshly cleavedHOPG were dipped in 10–12 mM 1-1OHSQdichloromethane solution for 2 min. For comparison,dye coated electrodes were also prepared by dropping aknown amount of 1-1OHSQ dye solution onto a freshlycleaved HOPG substrate followed by evaporation ofsolvents in ambient air. Dye stock solutions werefiltered through a membrane filter (0.02 mm pore diame-ter, Whatman anodisc 13) before use to remove anyundissolved dye particles or inert particles that couldact as nucleation sites.

2.3. Electrochemical measurements

A Teflon-made electrode holder was designed to useHOPG pieces as working electrodes and to make sam-ples transferable between electrochemical measurementsand scanning probe microscopy experiments. For elec-trochemical measurements, dye deposited HOPG pieceswere mounted on a copper plate with a silver paste (SPISupplies) inside the holder, through which electricalcontacts were made. Samples were sandwiched in atwo-piece Teflon holder with a flat Latex washer (4 mmi.d., 0.6 mm thick) with a 4 mm diameter window inone piece. The holder was held together with fourscrews to seal all but the �0.13 cm2 active area fromthe electrolyte.

Electrochemical measurements were conducted usinga conventional three-compartment electrochemical cell.A saturated calomel electrode (SCE) or a sodium-satu-rated calomel electrode (SSCE, in the case of perchlo-rate electrolyte) was used as the reference electrode andPt gauze was used as a counter electrode. Staircasecyclic voltammograms (SCVs) and staircase linearsweep voltammograms were taken with an EG&GPrinceton Applied Research 174A polarographic ana-lyzer controlled by a MacLab interface (AD Instru-ments) and a Macintosh computer. The step height was1 mV and the step width depended on the scan rate(= (step height)/(scan rate)), which was 100 ms at ascan rate of 10 mV s−1. Current signals were sampledand averaged over the last 25% of the time duration ofeach step at 0.1 ms intervals. After immersing theelectrodes, aqueous electrolyte solutions were purgedwith nitrogen gas for at least 15 min prior to measure-ments. All measurements were conducted at ambienttemperature.

N. Takeda et al. / Journal of Electroanalytical Chemistry 498 (2001) 19–33 21

2.4. Scanning probe microscopy

Molecular arrangements of both electrochemicallyoxidized and reduced (neutral form before electrochem-ical oxidation) forms of 1-1OHSQ layers adsorbed onHOPG prepared by the dipping method were investi-gated by STM. STM experiments were performed ex-situ with a Digital Instruments Nanoscope III scanningtunneling microscope operating under ambient condi-tions. Vibration isolation was provided by a bungeesystem enclosed in an environmental chamber. STMtips were cut to a sharp point mechanically from 80:20platinum–iridium 0.2 mm diameter wire (Alpha Aesar).Images of oxidized or reduced 1-1OHSQ layers wereobtained using typical tunneling parameters of −1. 4to −1.6 V (sample negative) and 40 pA with the STMoperating in the constant current mode. This range ofsample bias was slightly different from that reported forimaging of 1-1OHSQ (not electrochemically treated)under phenyloctane solvent [33].

STM samples of oxidized 1-1OHSQ on HOPG wereprepared by scanning non-oxidized samples positivelyby staircase linear sweep voltammetry (scan rate 10 mVs−1) after subjecting the samples to several potentialcycles to check their cyclic voltammetric behavior. Sam-ples were mounted in the STM within minutes of theelectrochemical experiment. Due to the ex-situ natureof this experimental setup, it was not possible to inves-tigate the same exact area of the adsorbate surface

upon exchanging the sample between the STM and theelectrochemical cell for CV cycling. Therefore it wasnecessary to rely upon statistical sampling of numerousareas of the surface to ensure that differences observedin its oxidized state (as compared to the reduced state)were truly due to electrochemical reorganization, andnot just a local orientational difference. To reduceadsorbate etching presumably caused by capillary force(described below) the tip was sometimes immersed intoa drop or two of phenyloctane (Aldrich) deposited ontothe sample surface. This procedure improved the reso-lution and reduced the noise when imaging the oxidizedform of the dye due to its negligibly small solubility inphenyloctane. Some of the reduced form of the dyedissolved in the phenyloctane increasing the imagenoise due to the higher solubility of the reduced dye inthis solvent. To prevent dissolution of the reducedsquaraine adsorbate into the phenyloctane layer, thephenyloctane was presaturated with the reduced formof 1-1OHSQ. It was then deposited on the adsorbatesurface in the same manner as the oxidized films andsucceeded in improving image quality and adsorbatelayer stability.

Atomic force microscopy was performed with a Digi-tal Instruments Nanoscope IIIa scanning probe micro-scope operated in tapping mode. Conical silicon AFMtips (Ultralevers™, Park Scientific Instruments) withspring constants of �0.26 and �0.40 N m−1 wereused. A HOPG sample with a freshly deposited 1-1OHSQ layer prepared by the dropping method wasfixed on a steel plate with a silver paste and thenmounted on an AFM scanner (type-J). All measure-ments were conducted in ambient air. After investigat-ing several areas by AFM, the sample was mounted inthe Teflon holder in the same manner as describedabove to perform electrochemical studies.

3. Results and discussion

3.1. General trends of cyclic 6oltammograms of1-1OHSQ adsorbed on HOPG

Law et al. reported electrochemical properties ofsquaraine dyes dissolved in dichloromethane using Pt asa working electrode and tetrabutylammonium perchlo-rate (TBAP) as supporting electrolyte [34]. The cyclicvoltammogram of squaraines showed two reversiblewaves with �60 mV peak separation as expected for aNernstian diffusion-controlled one-electron electro-chemical reaction [35]. For the 1-1OHSQ dye, the firstand the second formal oxidation potentials were deter-mined to be +0.41 and +0.91 V versus Ag � AgCl indichloromethane, respectively [34].

In the present study staircase cyclic voltammograms(SCVs) of the 1-1OHSQ dye layers adsorbed on HOPG

Fig. 1. Staircase cyclic voltammograms of 1-1OHSQ/HOPG elec-trodes (dipping method) in 0.1 M LiCl aqueous solutions. In eachpanel, the first scan (dotted line) and the eleventh (a) or tenth (b) scan(solid line) for two different samples are shown. Scan rate 10 mV s−1.Electrode area 0.13 cm2.


(1-1OHSQ/HOPG) electrodes prepared by the dippingmethod showed remarkably different behavior fromthat reported for solution electrochemistry. Generaltrends of observed voltammogramms are explained be-low. Fig. 1(a,b) show examples of cyclic voltam-mograms of 1-1OHSQ/HOPG electrodes obtained in0.1 M LiCl aqueous solution at a scan rate of 10 mVs−1. In each panel voltammograms obtained in the firstscan (dotted lines) and after several cycles (solid lines)are compared for two different samples prepared bysimilar procedures. Both samples revealed sharp oxida-tion and reduction peaks with a relatively large peakseparation of 140–230 mV with no significant currentsbetween them. Usually the first voltammetric scansgave considerably different peak potentials and waveshapes than those from subsequent scans. After the firstscan, the main peak potentials were relatively stable butwave shapes and peak currents occasionally changedwith repetitive scanning. In Fig. 1(a), two overlappingoxidation peaks at +0.129 and +0.159 V versus SCEcan be seen in the first scan (dotted line). In the 11thscan, shown as a solid line in the figure, these peaks hadmerged to become a single peak at +0.13790.002 Vversus SCE. A similar behavior appeared in the reversenegative scan in which a peak at around −0.01 Vversus SCE had evolved into a larger peak. In Fig. 1(b),the oxidation wave in the first scan had a prewave atthe foot of the peak (+0.185 V versus SCE) whichfused with a �50 mV more negatively shifted peak ataround +0.135 V versus SCE in the following scans.On the reduction side, a peak at −0.043 V versus SCEsplit into two comparable peaks in the tenth scan asshown. Moreover, small jagged peaks appeared at po-tentials more negative than the main peak. In general,similar current spikes often appeared especially in nega-tive scans. Sometimes the magnitude of these currentswas larger than that of the main peak. These currentspikes behaved rather stochastically, that is the poten-tial and peak heights changed from scan to scan. Thisphenomenon made the determination of reduction po-tentials more ambiguous than for the oxidation peaks.The small current spikes tended to disappear aftermany potential cycles (as seen in Fig. 1). However, insome cases multiple peaks remained after extensivecycling.

The total integrations of the positive and negativecurrents were approximately the same and were rela-tively constant with scan numbers as in Fig. 1. How-ever, in some cases the voltammogram area as well asthe peak current decreased with repetitive scanning. Itis possible that the oxidized form of the dye has a slightsolubility in aqueous solutions and desorbed fromHOPG electrodes, or that electrochemically inactiveareas could be formed. The total charge passed byelectrochemical oxidation and reduction of 1-1OHSQ/HOPG electrodes differed from sample to sample al-

though similar procedures were employed to preparethe samples. This uncertainty is presumably because theamounts of 1-1OHSQ molecules adsorbed on HOPGwere different, which could be explained by such condi-tions as the quality of the HOPG basal plane cleavage,the withdrawing process of the HOPG from the dyesolutions, or randomness in the nucleation and growthof the dye layers in the deposition processes.

Some comments about the magnitude of currentsobtained in our SCVs are needed. If nucleation andgrowth type kinetics are involved, as suggested below,current transients for each potential step would peak.Our SCV currents were averaged over the last 25% ofeach step. Due to the signal processing method em-ployed in this study, integrating the currents to deter-mine charges and surface coverage will underestimatethe actual coverage. Further coulometric studies areneeded to provide more accurate surface coverage val-ues. However, it can be safely said that multilayers ofthe dye would be formed, at least in some cases, sincethe underestimated charges and the STM results pre-sented below both confirm this.

3.2. Reaction scheme including electrolyte anionincorporation

In the present study, dye modified electrodes wereimmersed in aqueous electrolyte solutions in which1-1OHSQ is not soluble. Accordingly, the electroactive1-1OHSQ layers can be regarded as immobilized onHOPG surfaces. However, unlike typical surfaceconfined systems including physically [1,2] or covalently[36–44] attached redox molecular species, the presentsystem revealed a large hysteresis in cyclic voltam-mograms even at slow scan rates. This indicates largeactivation barriers for redox reactions of adsorbed 1-1OHSQ layers. The large splitting of redox peaks in thevoltammograms is reminiscent of other reported sys-tems such as TTF and TCNQ salts [5–13], and solidC60 immobilized on electrodes [14–23]. This behaviorhas often been attributed to large structural changesduring redox reactions. Laviron had simulated thevoltammetric shape of surface confined species, andobtained a hysteresis effect by using strong ‘interactionparameters’ between adsorbed molecules [45]. As de-scribed in the introduction, the splitting of anodic andcathodic peaks was also explained by considering theinsolubility of the adsorbed redox species and the im-miscibility of the two adsorbed redox phases [29]. Inthis case, the Nernst equation cannot be satisfied in acertain potential range (‘miscibility gap’) where the twophases are immiscible and the redox reaction is sup-pressed. If the electrode is further polarized, exceedingthe critical potential, the redox reaction is initiated anda sharp current peak appears. It should be noted that inthis case the potential region halfway between the an


Fig. 2. Staircase cyclic voltammograms of 1-1OHSQ/HOPG elec-trodes (dipping method) in 0.01 M (a), 0.1 M (b), and 1 M (c) LiClaqueous solutions. Scan rate 10 mV s−1. Electrode area 0.13 cm2.

could be initiated at nucleation sites, possibly at defectsites within the packed molecular layers or at domainedges, then rapidly expand to entire domains or layersto give the sharp redox peaks in the voltammograms.

The fact that the first scan showed a more positiveoxidation peak than successive scans (Fig. 1(b)) indi-cates that larger overpotentials are needed to incorpo-rate anions into the 1-1OHSQ layers prior toreconstruction, as was observed for another system [19].Also, the release of solvent, which might have beentrapped during the dye deposition processes(dichloromethane in the present case), could occur afterthe initial scan to alter the voltammograms. Further-more, there could be time dependent factors, especiallyfor cathodic processes, if nucleation and growth typekinetics are involved. The multiple current peaks ob-served in staircase cyclic voltammograms could be re-lated to the existence of multiple domains of 1-1OHSQlayers with different nucleation rates for anion associa-tion or expulsion. The appearance and disappearanceof the multiple current peaks could then be the result ofthe formation and annealing of small domains intolarger domains with repetitive cycling. It has beenshown that the nucleation rate limits the formation ofordered liquid crystal layers into small domains createdby controlled-size pits on HOPG [46]. Large areas ofHOPG were quickly covered with ordered layers sincethe probability of forming the critical nuclei on somepart of the surface was high [46]. Different polytypes ofpacked 1-1OHSQ domains with different formal poten-tials or overpotentials could also account for the multi-ple peaks. The large double peaks observed in Fig. 1(b)may be the result of the two polymorphs found in STMstudies discussed later.

3.3. Effect of electrolyte concentration

If the above-mentioned reaction scheme is occurring,the properties of the electrolyte will affect the voltam-metric behavior of 1-1OHSQ layers on HOPG. Theeffect of electrolyte concentration was investigated byusing three different concentrations of aqueous LiCl,NaCl, and KCl solutions as supporting electrolytes(0.01 M, 0.1 M, and 1 M). Dye modified electrodeswere prepared by the same dipping method as in thepreceding section. Changes in the electrolyte cationfrom lithium to sodium to potassium had only a smallinfluence on the voltammetric behavior, as long as theconcentrations of the electrolytes were equal. This isreasonable because reaction (2) does not involve elec-trolyte cations but anions. The electrolyte concentra-tions did affect some features of the voltammograms.Staircase cyclic voltammograms obtained in 0.01, 0.1,and 1 M of LiCl aqueous solutions after several cyclesare compared in Fig. 2 (scan rate 10 mV s−1). Thesame 1-1OHSQ/HOPG sample was used to obtain all

odic and cathodic peaks is not necessarily the value ofthe formal potential of the redox reaction [29].

We hypothesize that the redox reaction of 1-1OHSQadsorbed on HOPG proceeds as follows. A neutral1-1OHSQ molecule is oxidized by one electron to forma positively charged 1-1OHSQ radical cation (reaction(1)), followed by the association of an electrolyte anionto maintain charge neutrality (reaction (2)) as givenbelow.

(1-1OHSQ)−e−U(1-1OHSQ)+ (1)

(1-1OHSQ)++X−U(1-1OHSQ)+X− (2)

X− is the electrolyte anion species and (1-1OHSQ)+

is a one-electron oxidized 1-1OHSQ molecule. Sincethese processes are reversible, re-reduction of the oxi-dized form of 1-1OHSQ is accompanied by release ofthe electrolyte anion from the film. We assume a one-electron redox reaction because the redox peaks ofimmobilized 1-1OHSQ appeared at a similar potentialas the first one-electron oxidation of dissolved 1-1OHSQ in dichloromethane solution [34] when thesame electrolyte anion (perchlorate) was used (see Sec-tion 3.4), and the formal potential of molecular ad-sorbed species was usually found to be similar to thatof dissolved species [36,37,39,40].

Many of the observations described above can beinterpreted using the reaction scheme represented byreactions (1) and (2). Previous STM studies showedthat the squaraine dyes have strong intermolecular in-teractions and form tightly packed well ordered struc-tures on HOPG substrates [33]. A large structuralchange of the ordered 1-1OHSQ molecules occurs inorder to include or exclude the electrolyte anions uponoxidation or reduction (reaction (2)). The reaction


of these voltammograms. It can be seen that both theoxidation peak potential (Ep

ox) and the reduction peakpotential (Ep

red) depend on the concentration of thesupporting electrolyte. Furthermore, peaks becamesharper as the electrolyte concentrations increased. Val-ues of full width at half maximum (fwhm) of oxidationand reduction peaks in Fig. 2 are 48 and 42 mV in 0.01M, 23 and 34 mV in 0.1 M and 9 and 11 mV in 1 Msolutions, respectively.

Oxidation and reduction peak potentials for differentconcentrations of LiCl, NaCl, and KCl, determined ata scan rate of 10 mV s−1 after the voltammogramswere stabilized by repeated scanning, are plotted in Fig.3. In some cases multiple peaks remained after manypotential cycles. The potential of the main peaks withthe largest areas were considered to be the peak poten-tials. Regression lines shown in the figure were least-squares fits to the data obtained for each electrolyte. Ineach case, the potentials shifted negatively with increas-ing concentrations. The slopes of the lines were −70mV (Ep

ox) and −47 mV (Epred) for LiCl, −70 mV (Ep

ox)and −27 mV (Ep

red) for NaCl, and −71 mV (Epox) and

−34 mV (Epred) for KCl per decade change of concen-

tration. The magnitude of the potential shift was largerfor Ep

ox than for Epred in each case. As a result the peak

separation between the oxidation peak and the reduc-tion peak (Ep

ox–Epred) tended to become larger as the

concentration of the electrolyte was decreased. Thedifferent potential shifts could be related to the differ-ences in the oxidation and reduction processes (forinstance, the asymmetric nature of anion inclusion andexclusion). On average, there is a 53 mV shift of peakpotentials with a ten-fold increase or decrease of theelectrolyte concentration, close to a �59 mV shiftpredicted for Nernstian systems. A similar dependenceof peak potentials on electrolyte ion concentration wasobserved for other surface confined systems such asredox-functionalized self-assembled monolayers(SAMs) on gold [40–42,44] and the solid compoundsattached to graphite substrates [8,12,20], where elec-trolyte ions were involved in redox reactions. The 53mV value, close to the 59 mV Nernstian value for aone-electron process, further supports our assumptionthat the surface oxidation/reduction process is a oneelectron process. A �30 mV shift would be observed iftwo anions were involved in reaction (2).

Narrowing of peak widths with increasing electrolyteconcentrations was also commonly observed in NaCland KCl electrolytes as well as in LiCl electrolyte. Thefwhm of a one-electron redox voltammetric peak waspredicted to be 90.6 mV (at 25°C) for an ideal Nerns-tian surface confined redox system [35,37]. Theoreticaltreatments including effects of electrolyte concentrationon voltammetric shape and peak potentials have beendeveloped considering the potential distribution acrossthe redox-functionalized SAMs [47,48]. It is interestingto note that recently Ohtani et al. explained the poten-tial shift of sharply peaked voltammograms by consid-ering strong ion-pair formation with the surfaceconfined redox-active molecules, which increased thepotential drop and required a smaller polarization [48].However, in the system where a large hysteresis isobserved, the occurrence of a sharp peak, much nar-rower than the ideal Nernstian system itself, could beexplained by sudden redox reactions after passing the‘miscibility gap’ as described in the preceding section

Fig. 3. Plots of peak potentials for the oxidation (Epox, open symbols)

and reduction (Epred, closed symbols) of 1-1OHSQ dye layers on

HOPG electrodes as a function of supporting electrolyte concentra-tion. Supporting electrolyte: lithium chloride (circles), sodium chlo-ride (triangles), and potassium chloride (inverse triangles). Scan rate10 mV s−1. Lines are least-squares fits to the data.

Fig. 4. Staircase cyclic voltammograms of 1-1OHSQ/HOPG elec-trodes (dipping method) in 0.1 M NaCl (a), NaBr (b), and NaI (c)aqueous solutions. Scan rate 10 mV s−1. Electrode area 0.13 cm2.


Fig. 5. Staircase cyclic voltammograms of 1-1OHSQ/HOPG elec-trodes (dipping method) obtained in 0.1 M NaF (a), and Na2SO4 (b)aqueous electrolytes. Scan rate 10 mV s−1. Electrode area 0.13 cm2.

teresis were observed. The peak potentials were signifi-cantly shifted to more negative potentials in bromide(�150 mV) and iodide (�370 mV) electrolytes whencompared to the chloride electrolytes of the same con-centration. The differences in the magnitude of theobserved currents were probably due to the differentsurface coverages of 1-1OHSQ on HOPG that resultedfrom each sample preparation.

In Fig. 5, cyclic voltammograms obtained in 0.1 MNaF and 0.1 M Na2SO4 aqueous electrolytes afterseveral potential cycles (scan rate 10 mV s−1) areshown. In these electrolytes, broader redox peaks ap-peared at much more positive potentials than those inthe halide electrolytes of the same concentration. Inboth cases, the anodic peak in the first scan appearedmore positive than that of subsequent scans shown inFig. 5, whose behavior was also observed in halideelectrolytes. In addition, the anodic peak, observed inthe first scan, was much larger than the cathodic peak.There could be some irreversibility for the incorpora-tion of fluoride or sulfate anions into 1-1OHSQ layers.Voltammograms obtained with five successive scans in0.1 M NaClO4 aqueous electrolyte (scan rate 10 mVs−1) are shown in Fig. 6. Note that potentials are givenversus the SSCE reference electrode in this figure. Re-dox peaks appeared in a similar potential range tothose observed in fluoride and sulfate electrolytes. InFig. 6, significant decreases of the peak current withrepetitive scans can be seen. Many dye molecules aremore soluble as perchlorate salts and we suspect loss ofthe oxidized dye due to a small aqueous solubility.Voltammograms obtained in fluoride, sulfate, and per-chlorate electrolytes also resemble each other sincemeasurable currents appear between the two peaks. Thepeaks were also somewhat broader than those observedin chloride, bromide, and iodide electrolytes. The misci-bility gap concept explains the very low currents be-tween peaks in the halide electrolytes and the differentshapes and peak separations of the voltammograms ofthe adsorbed 1-1OHSQ redox reaction. One require-ment for the miscibility gap is that the two phasescannot equilibrate through the solution, thus any solu-bility of the oxidized or reduced phase would allow forthis equilibration.

All the results presented above confirm the impor-tance of the electrolyte anions in the redox reaction of1-1OHSQ on HOPG supporting the applicability of thereaction scheme proposed above (reactions (1) and (2)).The order of oxidation and reduction peak positionsobserved in 0.1 M electrolytes was I−BBr−Cl−BSO4

2−:ClO4−:F− for the anion species investigated.

As suggested in studies on other surface confined redoxreactions involving electrolyte ions [39–44,48], shifts ofredox peak potentials could be interpreted by differ-ences in the degree of association (association constant)of electrolyte anions with the oxidized form of the

Fig. 6. Staircase cyclic voltammograms of 1-1OHSQ/HOPG electrode(dipping method) obtained in 0.1 M NaClO4 aqueous electrolyte.Scan rate 10 mV s−1. Arrows indicate direction of decreasing cur-rents with repetitive cycling. Electrode area 0.13 cm2.

[29]. In such a case, the peak width is also influenced byredox or ion transfer kinetics. Broadening of thevoltammetric peak with decreasing electrolyte concen-tration observed in this study might be due to slowerkinetics such as slower diffusion of electrolyte anionsinto or out of the 1-1OHSQ layers.

3.4. Effect of electrolyte anion species

Voltammetric behavior of 1-1OHSQ deposited onHOPG by the dipping method was found to be dramat-ically affected by the electrolyte anion. Fig. 4 comparescyclic voltammograms of 1-1OHSQ on HOPG in 0.1 MNaCl, NaBr, and NaI aqueous electrolytes (scan rate 10mV s−1). Again, sharp redox peaks with a large hys-


1-1OHSQ molecules. Preferential ion-pairing of posi-tively charged molecular layers with anions would shiftthe formal potential of the redox reaction to morenegative potentials. The dramatic shift of the redoxpotential observed in this study (�600 mV differencebetween I− and F− electrolytes) suggests that the na-ture of the electrolyte anions plays an important role inthe thermodynamics of the present redox system. Fac-tors such as hydrodynamic radii of anions and Gibbsenergy differences between hydrated and dye-associatedanions would also affect the ion-pair formation. Fol-lowing the observed redox potentials, the order ofaffinity of anions would be I−\Br– −\Cl−\SO4

2−

:ClO4−:F−. The fact that similar peak potentials are

obtained in sulfate, perchlorate, and fluoride suggeststhat interactions of these anions with the oxidized1-1OHSQ layers are similar.

3.5. Effect of scan rates

An example of a scan rate dependence for staircasecyclic voltammograms of 1-1OHSQ on HOPG pre-pared by the dipping method and measured in 0.1 MKCl aqueous electrolyte is shown in Fig. 7. Scan rateswere varied from 10 to 100 mV s−1 in panel (a) andfrom 1 to 10 mV s−1 in panel (b) of the figure. Both theoxidation and reduction currents depended on scanrates but were not linearly proportional as expected fora typical surface immobilized redox species [35,37].Also, peak potentials and fwhm of peaks often de-

pended on scan rates, as can be seen in Fig. 7. As thescan rate was increased, the redox peaks becamebroader and the peak separation also became larger.Furthermore, tails after current peaks tended to becomesignificant especially at higher scan rates. It is probablethat diffusion of anions, either from the electrolyte orthe solid film would affect the current and ultimatelylimit the overall reaction rate. The current sampling ofthe staircase current may also be a factor in interpretingpeak height with such sharp peak currents. Furtherinvestigation is needed to clarify these issues.

3.6. STM imaging of reduced and oxidized forms of1-1OHSQ

Dialkylamino squaraines have been found to behighly surface active and readily amenable to investiga-tion by STM [33]. It would be useful to summarize hereseveral observations obtained in an earlier study1.Squaraine molecules appeared in most STM images asfootball-shaped areas of bright contrast measuring15.091.5 A, in length, 2.591.0 A, in width, and from0.2 to 1.5 A, in relative height. Occasionally two brightspots within the football-shaped region were resolved,each measuring �3 A, in length, with a peak to peakdistance of about 8 A, (see figure 4 in Ref. [33]). Themeasured lengths of the football shapes correspondwith the size of the chromophore of the molecule (fromthe nitrogen at one end of the molecule to the nitrogenat the other end). The width of the football correspondswell with the width of a phenyl group, not including theoxygen atoms and the hydroxyls. An explanation forwhy the oxygen atoms and hydroxyl groups are notimaged has been presented elsewhere [33]. The methylgroups in 1-1OHSQ (and longer alkyl tails for otherdialkylamino squaraines) have not been clearly re-solved. Nevertheless, the length, width, and relativeheight of the bright football-shaped areas correspondwell with a model of the molecule lying with thechromophore flat on the HOPG surface. STM mea-sured heights of 1–2 A, are typical for aromatic systemsadsorbed lying flat on the substrate surface [49]. Sincethe methyl tails are not resolved, packing of the brightchromophores must be examined quantitatively to de-duce the orientation of the methyl tails.

Different ordered structures, or polymorphs, for thesame squaraine molecules were frequently observed inthe previous study [33]. Longer-tailed squaraines tendedto produce more stable adsorbate layers with row struc-tures while the herring-bone structures formed by theshorter-tailed squaraines seemed less stable since theywere more easily stripped off the substrate surface bythe rastering motion of the tip [33]. This was often

Fig. 7. Staircase cyclic voltammograms of 1-1OHSQ/HOPG electrode(dipping method) in 0.1 M KCl aqueous electrolyte obtained atvarious scan rates of 10, 20, 50, and 100 mV s−1 (a) and 1, 2, 5, and10 mV s−1 (b). Arrows indicate direction of increasing scan rates.Electrode area 0.13 cm2.

1 In an earlier study [33], molecules were primarily adsorbed fromphenyloctane solution directly on the HOPG surface. It should alsobe noted that 1-1OHSQ was denoted as 1-1 SQ in Ref. [33].


Fig. 8. Series of 50×50 nm STM images of reduced 1-1OHSQ adsorbed in rows on HOPG showing the time-lapsed erosion of the adsorbatelayer. Image (A) shows a number of missing-molecule defects which appear as dark rectangles. Image (B) was taken 1.5 min after image (A) andshows the growth and coalescence of several of the defects. Arrows correlate the defects from image (A) which have grown and coalesced in image(B). Image (C) was taken 1.5 min after image (B) and shows the erosion of the adsorbate layer in the upper right. The direction of thermal driftis shown in the bottom right corner of the figure.

observed in this investigation of the layers of 1-1OHSQdeposited from dichloromethane solutions. Thedichloromethane solution concentrations used withinthis study often produced multiple layers of 1-1OHSQon HOPG. Layer heights were measured to be from 1.5to 4.5 A, in thickness (approximately one to threelayers). The initial STM images showed that the surfacelayers of 1-1OHSQ were smooth with domainboundaries barely visible as fine fractures. A few min-utes after engagement the rastering motion of the tipbegan to remove material from the domain boundaries.The removal of the adsorbate layer also caused theSTM image to become noisy and streaky, presumablydue to the interference of the desorbed dye moleculeswith the tunneling junction. Both oxidized and reducedforms of the dye were susceptible to tip inducederosion.

At smaller scan sizes (B100 nm) the adsorbate layerswere more resistant to tip-induced erosion and were lessnoisy as long as no missing molecule defects or domainboundaries were present. The duration of imaging aparticular area was still dictated by the removal ofmolecules in that area. At smaller scan sizes it wasapparent that removal of additional molecules wasoccurring at rectangular defects composed of a few(2–3) missing molecules. These missing-molecule sitesusually occurred within a single molecular row. As thedefect grew, molecules from adjacent rows were alsoremoved. The process of defect nucleation and growthof holes in the layer can be seen in Fig. 8. Continualscanning caused the defects to grow until neighboringholes coalesced. Capillary forces and tip collisions withdomain or defect edges are presumed to be responsiblefor the removal of the molecules from the surface.

As mentioned above, layers of reduced 1-1OHSQ

appeared smooth except for step edges and missingmolecule defects, which often revealed the presence ofmultiple layers (Fig. 9(A)). At larger scan sizes thedomain intersections are barely visible except whereerosion has occurred between the domains. On a num-ber of occasions small islands of molecules adsorbed ontop of the smooth mono or multiple layers were im-aged. These islands were 60–70 nm in length, 10–20nm in width, and 2–3 nm in height. They tended toorient with their long axis at 30 or 60° angles to eachother and along crystallographic directions, as can beseen in Fig. 9(B), where some oriented islands areparallel to a graphite step edge. At smaller scan sizes(B300 nm) molecular lamella were observed. Thewidths of the bright stripes were typically measured tobe �14 A, , a distance which corresponds to the lengthof the chromophore of the 1-1OHSQ molecule. Thelamellar domains often showed rectangular missingmolecule defects. Some experiments, in which multiplelayers of 1-1OHSQ were adsorbed on the graphitesurface, exhibited entire missing rows (see Fig. 9(C)).These row defects were measured to be one or morelayers deep.

Fig. 10 (right) shows a 15×15 nm image of the rowstructure observed for the reduced 1-1OHSQ onHOPG. The chromophores of the molecules appear asfootball-like shapes measuring �14 A, in length. Thedistance between chromophores within the same rowwas measured to be �10 A, , while the distance betweenchromophores in adjacent rows was �24 A, . The angleof the long axis of the molecule to the direction of therow was measured to be �47°. Dialkylaminophenylmoieties in squaraines have an electron-donating char-acter while the central four-membered ring has anelectron-accepting character [50]. The 47° angle of themolecule to the row aligns adjacent molecules within


the same row such that their donor and acceptor moietiesinteract (see the molecular model proposed for thisstructure in Fig. 10). The donor–acceptor interaction

between adjacent molecules within the same molecularrow has been observed for other dialkylamino hydrox-yphenyl squaraines and we believe it is the dominantinteraction driving the ordering of these squaraines onHOPG [33]. This structure is similar to the row structureobserved for 1-1OHSQ deposited onto HOPG fromphenyloctane solutions. A smearing of the electrondensity between adjacent molecules (in the same row) wasoccasionally observed and produced a bright zigzagpattern within the row (see the left image in Fig. 10). Arelatively blunt tip that smears the electron densities fromtwo adjacent aminophenyl moieties (represented by thecircled area in Fig. 10) might be responsible for thispattern. Unit cell values for both phenyloctane-depositedand dichloromethane-deposited 1-1OHSQ are shown inTable 1. Only the most commonly observed polymorphdeposited from phenyloctane from Ref. [33] (90° herring-bone) was included in Table 1. This polymorph was notobserved in the experiments reported herein and thephenyloctane-deposited row polymorph reported hereinwas not observed previously. None of the polymorphs(neither reduced nor oxidized) observed for thedichloromethane deposition reported herein were ob-served in Ref. [33].

Adsorbed 1-1OHSQ that had been electrochemicallyoxidized in 0.1 M LiCl aqueous electrolyte showedmultilayers that were not as smooth as those formed bythe reduced form of the dye. As can be seen in Fig. 11(A),the oxidized dye layers are more broken and defective.Domain boundaries are more obvious and are oftenrounded in shape as opposed to the predominantlyfaceted domains of the reduced form. Domains are alsosmaller, typically measuring 100–300 nm in diameter. Atsmaller scan sizes lamella are visible that appear to besimilar to those of the reduced form (Fig. 11(B)).Rectangular missing molecule defects are also visible.Repeated scanning eventually expands these defects anderodes away the adsorbate layer again in a similar fashionto the erosion of the reduced dye layers. Missing rowdefects like those observed in the reduced form of the dyewere found in only a very small percentage of theexperiments.

Upon closer inspection of the lamella it was found thattwo different row structure polymorphs exist for oxidized1-1OHSQ. Polymorphs are commonly observed in two-dimensional surface structures of squaraine molecules[33]. The unit cells measured for these polymorphs canbe found in Table 1. Fig. 12 shows molecular resolutionimages and proposed molecular models for both of theserow structure polymorphs. Both polymorphs exhibitsimilar distances for both adjacent molecules within thesame row and molecules in adjacent rows. The angles ofthe molecules within the row to the direction of the row,however, are different for each of the polymorphs.Polymorph 1 exhibits a �67° angle of the molecules tothe direction of the row. This angle does not offset

Fig. 9. (A) 500×500 nm STM image showing smooth layers ofreduced 1-1OHSQ on HOPG (dipping method). Multiple layers areevident at defect sites. (B) 1.5 mm×1.5 mm scan STM image ofreduced 1-1OHSQ monolayer adsorbed on HOPG terraces. Smallislands oriented at 30° or 60° to one another and parallel to HOPGstep edge are visible. (C) Several missing rows are evident in this42×42 nm STM image. Missing rows were not observed in everyexperiment and were only observed in multiple layer adsorption.


Fig. 10. A 15×15 nm scan STM image (right) showing the row orientation of reduced 1-1OHSQ. Individual squaraine chromophores are visibleas bright football-like shapes. The black box represents the unit cell, which has been enlarged and projected onto the proposed molecular model(bottom) for comparison. Donor–acceptor interactions between adjacent molecules within the same row appear to be the dominant interactiondriving this order. A 5×5 nm image (left) shows a zigzag contrast, which is sometimes observed for the reduced form of 1-1OHSQ. The blackovals overlaid on both the left image and the model illustrate the coupling of electron density between adjacent molecules within the same row,which might lead to such a zigzag structure.

Table 1Unit cell parameters for reduced a (RED) and oxidized b (OX) forms of 1-1OHSQ layers adsorbed on HOPG as determined by STM

State PolymorphSolvent c a/A, b/A, a/° Molecular area/A, 2 molecule−1

90° herringbone d 2592RED 2592Phenyloctane 9092 156919RED Phenyloctane Row e 2191 1091 3594 210916RED RowCH2Cl2 24.290.6 9.990.5 46.591.3 24099

Type 1 row 20.490.8 9.390.4CH2Cl2 67.894.0OX 19098Type 2 rowOX 20.691.0CH2Cl2 9.691.0 46.294.0 200915

a Before electrochemical treatments.b Electrochemically oxidized in 0.1 M LiCl aqueous electrolyte; chloride ions were presumably incorporated as counter ions.c Solvents used for deposition of 1-1OHSQ.d Most commonly observed polymorph in the previous STM study (Ref. [33]), but was not observed again in this study. Note that 1-1OHSQ

was denoted as 1-1 SQ in Ref. [33].e Row polymorph observed for the reduced form of 1-1OHSQ deposited from phenyloctane (same manner as in Ref. [33]) in this study was not

observed in Ref. [33].

adjacent molecules from one another enough to achieve‘optimum’ donor-acceptor interaction. In addition, thepartially-negatively charged oxygen atoms in the squarecore are still relatively close. Repulsion between thesetwo charged moieties may be mitigated by possiblehydrogen-bonding between the partially charged oxy-

gen and the hydroxyl group of the molecule adjacent toit. Polymorph 1 was observed in only �25% of theexperiments. Polymorph 2 exhibits a �46° angle muchlike the reduced dye row structure. This angle offsetsadjacent molecules within the same row such that theirdonor and acceptor moieties line up much more effec-


tively, with a minimum of Od− –Od− repulsion. Theonly difference between the oxidized polymorph 2structure and the reduced structure is the distancebetween molecules in adjacent rows. This distance is�1 A, longer in the reduced form, resulting in thereduced structure having a slightly smaller packingdensity. This is counterintuitive since it would be ex-pected that the oxidized structure would be slightlylarger to account for the presence of the counterionswithin the adsorbate layer. This contraction of theadsorbate structure may explain the increased obser-vance of missing molecule defects in the oxidized ascompared to the reduced adsorbate. Also the doublepeaks sometimes observed in the SCV (Fig. 1(b)) maybe due to different polymorphs having slightly differentredox properties.

The actual position of the chloride ion within theadsorbate unit cell is not known at this time. A specula-

tive position is shown in the molecular models of Fig.12. Four factors were taken into account in placing thechloride at this position: (1) the most probable site forthe oxidation of the squaraine molecule is the aminogroups; (2) the delocalization of this partial positivecharge is most likely over the amino nitrogen and thephenyl carbon bonded to it; (3) the close proximity ofpositive charged parts of adjacent rows to one another;and (4) the electrochemical data which show that theion is most likely associated 1:1 with a two-dimensionalphase of oxidized dye molecules rather than beingdiffusely associated with the oxidized layer. It is possi-ble that the counterion resides slightly above the planeof the adsorbate between the amino groups of adjacentrows nested above the methyl groups as if they were thebottom of a dimethylamino solvation cage around thechloride ion. There is also a possibility that water ordichloromethane solvent molecules are incorporated inthese structures and are not resolved in the STM im-ages. Loss of solvent molecules upon oxidation mayalso account for the smaller unit cell areas for theoxidized molecules.

3.7. Effect of morphologies on electrochemistry of1-1OHSQ deposits on HOPG

It has been shown that deposition of 1-1OHSQmolecules from dichloromethane solution by the dip-ping method produced electrochemically active 1-1OHSQ layers on HOPG. Alternatively, depositioncould be accomplished by dropping a known amount ofdye solution onto HOPG as described in Section 2 (�1ml, �0.8×10−10 mol cm−2 obtained from STM re-sults). However, electrodes fabricated in this way re-vealed little or no well-defined redox peaks in cyclicvoltammograms tested in 0.1 M LiCl aqueous solution.We attribute this to a morphological effect of 1-1OHSQmolecules on voltammetric behavior as discussedbelow.

Fig. 13 shows a 5 mm×5 mm scan AFM image of1-1OHSQ on HOPG prepared by the dropping method.Many bright areas attributable to deposited 1-1OHSQaggregates or small crystallites can be seen. These struc-tures have typical heights of several nanometers butsometimes heights exceed 10 nm. The aggregates appearto align along specific directions. Similar tall aggregatestructures of 1-1OHSQ were also observed on the basalplanes of SnS2 single crystals when a similar depositionmethod was used [32]. We speculate that the three-di-mensional aggregates have difficulty in incorporatingthe electrolyte anions deep into their bulk structureeven though a bias positive enough to oxidize the freemolecule was applied. In other words, the potentialdrop developed between the electrode surface and thedye in the resistive three-dimensional aggregates wasinsufficient to drive the reaction. A similar logic is true

Fig. 11. (A) 500×500 nm STM image showing broken layers ofoxidized 1-1OHSQ. Rounded domain boundaries are clearly visible.(B) three-dimensional representation of a 98×98 nm scan STMimage of an oxidized layer showing lamella and rectangular defects.


Fig. 12. (A) 20×20 nm scan STM image showing polymorph 1 type row structure. Oxidized chromophores are evident as bright football-likeshapes. The unit cell is represented by the black box overlaid on the image and projected onto the proposed molecular model. Donor–acceptorinteractions as well as hydrogen-bonding appear to drive this polytype. Possible positions of the chloride counterions are included within themodel. (B) 28×28 nm scan showing oxidized polymorph 2 type row structure. In this image oxidized chromophores appear as bright dogbone-like shapes. The slightly more acute angle of the molecular long axis to the direction of the row for this polymorph allows adjacent donorand acceptor moieties to come into closer proximity.


Fig. 13. A 5 mm×5 mm scan AFM image of 1-1OHSQ deposited onHOPG substrate prepared by dropping ca. 1 ml (�0.8×10−10 molcm−2) of the dye dichloromethane solution followed by evaporationof the solvent in ambient air. z-Range 5 nm.

teresis for oxidation and reduction of the dye layers. Thelarge rearrangement of the adsorbed dye layers, due tothe incorporation of the electrolyte anions upon oxida-tion, is responsible for the large hysteresis in the voltam-mograms.

Whether the hysteresis is explained by the miscibilitygap model [29] or if a nucleation model can explain theresults is still an open question. Nucleation does appearto be important especially in the reduction of the oxidizeddye layers. We attribute the many current spikes in theSCVs to the reduction of different domains of oxidized1-1OHSQ. STM images showed the presence of manydomains in both the oxidized and reduced forms. Theincreased stability of the ionic two-dimensional latticeperhaps is responsible for the difficulty of reducing it andexpelling the anions. If a nucleation event was limitingthe re-reduction of the layers it would be expected thatthe smaller domains would be less likely to have anucleation event and would require a larger electrochem-ical potential to initiate reduction. In most experimentsmost of the charge was present in the initial reductionpeak and after repeated scanning the small current spikeswould eventually disappear. We attribute this to anneal-ing the small domains into larger domains during thecontinual oxidation-reduction cycles.

STM images also provided information on the molec-ular arrangements of both the oxidized and reducedforms of dye layers. Differences observed between thetwo phases could be the results of structural changesassociated with anion incorporation caused by electro-chemical oxidation. However, definitive proof of anionincorporation and their location was not obtained fromthe present STM study. Further investigation of redoxprocesses of surface confined ordered molecular layers isneeded to work out the detailed mechanism of the surfacetransformation. AFM and STM experiments done in-situin an electrochemical cell will be able to shed light onthese processes as well as the mechanism of transformingthe two-dimensional reduced dye structures into thetwo-dimensional oxidized structures.

Acknowledgements

We would like to thank the Lexmark InternationalCorporation for providing the squaraine dyes used in thisstudy. This work was supported by the Department ofEnergy Office of Basic Energy Sciences under contractDE-F603-96ER14625.

References

[1] A.P. Brown, C. Koval, F.C. Anson, J. Electroanal. Chem. 72(1976) 379.

[2] A.P. Brown, F.C. Anson, Anal. Chem. 49 (1977) 1589.

for electrodes coated with polymer films containingredox active groups that cannot be permeated by coun-terions [37].

However, samples prepared by the dipping methodshowed electrochemical activity even when the surfacecoverage of 1-1OHSQ was estimated to be several layers(judging from total area of voltammetric peaks and STMobservations of the samples). In some cases a diffusiontail on the SCV peaks did indicate that diffusion wasoperating at high coverages (see Fig. 7(a)). The pho-toelectrochemical behavior of the three-dimensional ag-gregates of 1-1OHSQ was shown to be much differentthan for monolayers of this dye [32]. The red shift in thephotocurrent action spectrum for the tall aggregates isindicative of different molecule–molecule interactionsthan in the monolayer. Incorporation of the anions intothese three-dimensional structures is necessary for theoxidation of the 1-1OHSQ molecules but the three-di-mensional structure may have no mechanism for ionconduction. Thus, the electrochemically inactive aggre-gates could have different aggregate structures from theactive layers.

4. Conclusion

The potential for electrochemical oxidation of 1-1OHSQ dye layers adsorbed on the basal plane HOPGelectrodes in aqueous electrolytes has been shown to bestrongly dependent on the properties of the electrolyteanions. Staircase cyclic voltammograms (SCVs) of theseelectrodes revealed very sharp peaks and a large hys-


[3] Z.F. Liu, K. Hashimoto, A. Fujishima, Nature 347 (1990) 658.[4] R. Salas, M. Sanchez-Maestre, R. Rodrıguez-Amaro, E.

Munoz, J.J. Ruiz, L. Camacho, Langmuir 11 (1995) 1791.[5] C.D. Jaeger, A.J. Bard, J. Am. Chem. Soc. 101 (1979) 1690.[6] W.L. Wallace, C.D. Jaeger, A.J. Bard, J. Am. Chem. Soc. 101

(1979) 4840.[7] C.D. Jaeger, A.J. Bard, J. Am. Chem. Soc. 102 (1980) 5435.[8] S.J. Shaw, F. Marken, A.M. Bond, Electroanalysis 8 (1996)

732.[9] R.D. Mounts, K. Widlund, H. Gunadi, J. Perez, B. Pech, J.Q.

Chambers, J. Electroanal. Chem. 340 (1992) 227.[10] C.D. Evans, J.Q. Chambers, Chem. Mater. 6 (1994) 454.[11] J.Q. Chambers, K. Scaboo, C.D. Evans, J. Electrochem. Soc.

143 (1996) 3039.[12] A.M. Bond, S. Fletcher, F. Marken, S.J. Shaw, P.G. Symons,

J. Chem. Soc. Faraday Trans. 92 (1996) 3925.[13] K.M. Scaboo, J.Q. Chambers, Electrochim. Acta 43 (1998)

3257.[14] C. Jehoulet, A.J. Bard, J. Am. Chem. Soc. 113 (1991) 5456.[15] C. Jehulet, Y.S. Obeng, Y.-T. Kim, F. Zhou, A.J. Bard, J. Am.

Chem. Soc. 114 (1992) 4237.[16] W. Koh, D. Dubois, W. Kutner, M. Thomas Jones, K.M.

Kadish, J. Phys. Chem. 96 (1992) 4163.[17] W. Koh, D. Dubois, W. Kutner, M. Thomas Jones, K.M.

Kadish, J. Phys. Chem. 97 (1993) 6871.[18] R.G. Compton, R.A. Spackman, R.G. Wellington, M.L.H.

Green, J. Turner, J. Electroanal. Chem. 327 (1992) 337.[19] M. Nishizawa, T. Matsue, I. Uchida, J. Electroanal. Chem. 353

(1993) 329.[20] T. Tatsuma, S. Kikuyama, N. Oyama, J. Phys. Chem. 97

(1993) 12067.[21] J. Chlistunoff, D. Cliffel, A.J. Bard, Thin Solid Films 257

(1995) 166.[22] A.L. Balch, D.A. Costa, W.R. Fawcett, K. Winkler, J. Elec-

troanal. Chem. 427 (1997) 137.[23] M.F. Suarez, F. Marken, R.G. Compton, A.M. Bond, W.

Miao, C.L. Raston, J. Phys. Chem. Sect. B 103 (1999) 5637.[24] B. Gregg, R.A. Cormier, J. Phys. Chem. Sect. B 102 (1998)

9952.[25] A.M. Bond, F. Scholz, J. Phys. Chem. 95 (1991) 7460.[26] A.M. Bond, R. Colton, F. Daniels, D.R. Fernando, F.

Marken, Y. Nagaosa, R.F.M. Van Steveninck, J.N. Walter, J.Am. Chem. Soc. 115 (1993) 9556.

[27] A.M. Bond, F. Marken, J. Electroanal. Chem. 372 (1994) 125.[28] M. Sanchez Maestre, R. Rodrıguez Amaro, E. Munoz, J.J.

Ruiz, L. Camacho, J. Electroanal. Chem. 373 (1994) 31.[29] F. Scholz, M. Lovric, Z. Stojek, J. Solid State Electrochem. 1

(1997) 134.[30] F. Scholz, B. Meyer, in: A.J. Bard, I. Rubinstein (Eds.), Elec-

troanalytical Chemistry: A Series of Advances, vol. 20, MarcelDekker, New York, 1998, p. 1.

[31] K.Y. Law, J. Phys. Chem. 91 (1987) 5184.[32] N. Takeda, B.A. Parkinson, Electrochim. Acta, in press.[33] M.E. Stawasz, D.L. Sampson, B.A. Parkinson, Langmuir 16

(2000) 2326.[34] K.-Y. Law, J.S. Facci, F.C. Bailey, J.F. Yanus, J. Imag. Sci. 34

(1990) 31.[35] A.J. Bard, L.R. Faulkner, Electrochemical Methods, Wiley,

New York, 1980.[36] K.W. Willman, R.D. Rocklin, R. Nowak, K.-N. Kuo, F.A.

Schultz, R.W. Murray, J. Am. Chem. Soc. 102 (1980) 7629.[37] R.W. Murray, in: A.J. Bard (Ed.), Electroanalytical Chemistry:

A Series of Advances, vol. 13, Marcel Dekker, New York,1984, p. 191.

[38] C.E.D. Chidsey, C.R. Bertozzi, T.M. Putvinski, A.M. Mujsce,J. Am. Chem. Soc. 112 (1990) 4301.

[39] D. Acevedo, H.D. Abruna, J. Phys. Chem. 95 (1991) 9590.[40] K. Uosaki, Y. Sato, H. Kita, Langmuir 7 (1991) 1510.[41] G.K. Rowe, S.E. Creager, Langmuir 7 (1991) 2307.[42] G.K. Rowe, S.E. Creager, J. Phys. Chem. 98 (1994) 5500.[43] H.O. Finklea, M.S. Ravenscroft, D.A. Snider, Langmuir 9

(1993) 223.[44] J. Redepenning, H.M. Tunison, H.O. Finklea, Langmuir 9

(1993) 1404.[45] E. Laviron, J. Electroanal. Chem. 122 (1981) 37.[46] D.L. Patrick, V.J. Cee, T.P. Beebe, Jr., Science 265 (1994) 231.[47] C.P. Smith, H.S. White, Anal. Chem. 64 (1992) 2398.[48] M. Ohtani, S. Kuwabata, H. Yoneyama, Anal. Chem. 69

(1997) 1045.[49] T. Kawai, T. Tanaka, T. Nakagawa, Surf. Sci. 386 (1997) 124.[50] R.W. Bigelow, H.-J. Freund, Chem. Phys. 107 (1986) 159.

.

Documents

Electrochemical oxidation and ex-situ STM observations of ...franklin.chm.colostate.edu/bap/PDF of Papers copy/J... · Princeton Applied Research 174A polarographic ana-lyzer controlled