Surfactant monolayers on electrode surfaces: self-assembly of a series

Gomez, Jing. Li, and Angel E. Kaifer. Langmuir , 1991, 7 (8), pp 1797– .... Hannoch Ron, Sophie Matlis, and Israel Rubinstein. Langmuir 1998 14 (5),...
0 downloads 0 Views 1MB Size
Langmuir 1991, 7, 1797-1806

1797

Surfactant Monolayers on Electrode Surfaces: Self-Assembly of a Series of Amphiphilic Viologens on Gold and Tin Oxide Marielle Gomez, Jing Li, and Angel E. Kaifer’ Department of Chemistry, University of Miami, Coral Gables, Florida 33124 Received October 27, 1990. I n Final Form: February 25, 1991 The aggregation of a series of N-alkyl-”-ethyl viologens, with alkyl chains from 10 to 18 carbons long, N-allyl-N’-hexadecylviologen,N-allyl-N’wtadecyl viologen,and NJV’-diheptylviologen at the electrodesolution interface was surveyed on Au in order to establish correlations between the structural features of the viologen amphiphiles and the electrochemical properties of the interfacial assemblies. The octadecyl viologens l and 7 were found to self-assemble on gold from aqueous solutions, forming compact monolayers with a limiting coverage of 4.1 X 10-10 mol/cm2. The hexadecyl derivatives 2 and 6 also self-assembled on gold but the measured limiting coverages were lower. The symmetric viologen 8 was not found to self-assemble appreciably under identical experimental conditions. Aggregation was also observed on SnOz surfaces. Compact interfacial monolayer assemblies formed by compounds 1 and 7 mediate the heterogeneous electron transfer reaction between the water-soluble complex Ru(NHs)Pand the underlying electrode surface. In contrast, no mediation was observed for the hexadecyl analogues. A zwitterionic derivative of ferrocene was also used to probe the degree of packing and determine the barrier properties of the viologen monolayers. The results indicate that the blocking effects arise mainly from the lipophilic nature of the monomers’ alkyl chains.

Introduction The self-assembly of amphiphilic molecules leads to various types of supramolecular aggregates, such as micelles, vesicles, monolayers, bilayers, and others. The morphology and properties of these aggregates have been the subject of extensive research work in many areas of science. Electrochemistry has been no exception and, during the last 5 years, the preparation of highly organized monolayers or multilayer5 at the electrode solution interface has generated strong interest. The main reason for this interest is the precise molecular architecture that can be accomplished with these systems which, in turn, may provide an unparalleled degree of control on the electrochemical reactivity of the underlying electrode surface. Deposition of Langmuir-Blodgett monolayers on electrode substrates’-16 constitutes one of the most common methods to build interfacial amphiphilic structures. Another approach is based on the self-assembly of lipophilic molecules ending in hydrolytically unstable groups, such as -Sicla, or sulfur functional groups, such (1) Memming, R. Discuss. Faraday SOC.1974,58, 261. (2) Fromhen, P.; Arden, W. J. Am. Chem. SOC.1980,102,6211. (3) Memming, R.; Schroppel, F. Chem. Phys. Lett. 1979, 62, 207. (4) Daifuku, H.; Aoki, K.; Tokuda, K.; Matauda, H. J . Electroanal. Chem. Interfacial Electrochem. 1982, 140, 179. (5) Park, H. G.; Aoki, K. A,; Tokuda, K.; Matauda, H. J. ELectroanal. Chem. Interfacial Electrochem. 1985, 195, 157. (6) Daifuku, H.; Aoki, K.; Tokuda, K.; Matauda, H. J . Electroanal. Chem. Interfacial Electrochem. 1985, 183, 1. (7) Daifuku, H.; Yoshimira, I.; Hirata, I.; Aoki, K.; Tokuda, K.; Matauda, H. J. Electroanal. Chem. Interfacial Electrochem. 1986,199,47. (8) Aoki, K.;Tokuda,K.;Matauda,H. J. Electroanal.Chem.Interfacial Electrochem. 1986, 199, 69. (9) Mateuda,H.; Aoki, K.;Tokuda,K.J. Electroanal. Chem.Interfacial Electrochem. 1987,217, 1. (10) Fujihira, M.; Pooeittisak, S. J. Electroanal. Chem. Interfacial Electrochem. 1986,199,69. (11) Fujihira, M.; Araki, T. J. Electroanal. Chem. Interfacial Electrochem. 1986,205, 329. (12) Fujihira, M.; Araki, T. Bull. Chem. SOC.Jpn. 1986,59, 2375. (13) Lee, C.-W.;Bard, A. J. J. Electroanal. Chem. Interfacial Electrochem. 1988,239,441. (14) Zhang,X.; Bard, A. J. J. Phys. Chem. 1988,92,5566. (16) Zhang,X.; Bard, A. J. J . Am. Chem. SOC.1989,111,8098.

0743-7463/91/2407-1797$02.50/0

as thiol or sulfide (generally to take advantage of their strong adsorption on Au surfaces).’Surfactant molecules lacking sulfur or trichlorosilane functional groups may also self-assemblea t the electrodesolution interface. This phenomenon has been used to build monolayers of redox-active amphiphiles by simply exposing a clean electrode surface to a dilute aqueous solution (usually in the micromolar range) of the redoxactive surfactant. In this case, the self-assembly is essentially driven by hydrophobic effects. For instance, Facci has studied the monolayer behavior on Pt of a fer(16) Porter, M. D.; Bright, T. B.; Allara, D. L.; Chidaey, C. E. J. Am. Chem. SOC.1987,109,3559.

(17) Finklea,H.O.;Robinaon,L.R.;Blackburn,A.;Richter,B.;Wara,

D.: Bright, T. Lonamuir 1986.2, 239. .(18)Pkklea,H.-O.; Avery,.S.i Lynch, M. Langmuir 1987,3,409. (19) Finklea, H. 0.;Fedyk, J.; Schwab, J. In Electrochemical Surface Science; Soriaga,M., Ed.;ACS SymposiumSeries378; American Chemical Society: Washington, DC, 1988; p 431. (20) Lee,K. A. B.; Mowry, R.; McLennan, C.; Finklea, H. 0.J. Electroanal. Chem. Interfacial Electrochem. 1988,246,217. (21) Sabatini, E.; Rubinatein, I.; Maoz, R.; Sagiv, J. J. Electmnal. Chem. Interfacial Electrochem. 1987,219, 365. (22) Sabatini, E.; Rubinatein, I. J. Phys. Chem. 1987,91,6663. (23) Rubinstein,I.; Steinberg,S.;Tor, Y.; Shanzer,A.; Sagiv,J. Nature 1988,332,426. (24) Maoz, R.; Sagiv, J. J. Colloid Interface Sci. 1984, 100,466, (25) Gun, J.; Iscovici, R.; Sagiv, J. J. Colloid Interface Sci. 1981,101, 201. (26) Cohen, S. R.; Naaman,R.;Sagiv, J. J. Phys. Chem. 1986,90,3054. (27) Maoz, R.; Sagiv, J. Langmuir 1987,3, 1034. (28) Maoz, R.; Sagiv, J. Langmuir 1987, 3, 1045. (29) Bain, C. D.; Whitesides, G. M. J. Am. Chem. SOC.1988,110,3665. (30) Bain, C. D.; Whitesides, G. M. Science 1988,240, 62. (31) Bain, C. D.; Troughton, E. B.; Tao, Y.-T.; Evall, J.; Whiteaides, G. M.; Nuzzo,R. G. J. Am. Chem. SOC.1989,111,321. (32) Strong, L.; Whitasides, G. M. Langmuir 1988,4, 548. (33) Nuzzo,R. G.;Zegarski, B. R.; Dubois, L. H. J. Am. Chem. Soc. 1987.109.733. (34) Diem, T.; Czajka, B.; Weber, B.; Regen, 5.L. J. Am. Chem. SOC. 1986,108,6094. (35) Nuzzo, R. G.; Allara, D. L. J. Am. Chem. SOC.19lw,106,4481. (36) Nuzzo, R. G.; Fueco, R. A.; Allara, D. L. J. Am. Chem. Soc. 1987, 109, 2358. (37) Troughton, R. B.; Bain, C. D.; Whitesides, G. M.; Nuzzo,R. C.; Allara, D. L.; Porter, M. D. Langmuir 1988,4,365. (38) Li, T. T.-T.; Weaver, M. J. J. Am. Chem. SOC.1984,106,6107. (39) Lee, K. A. B. Langmuir 1990,6,709. (40) De Long, H. C.; Buttry, D. A. Langmuir 1990,6,1319.

0 1991 American Chemical Society

1798 Langmuir, Vol. 7, No. 8,1991

Chart I

COMPOUND

Ri

1

CHzCH,

2

CHzCH:,

3

CHzCH,

4

CHzCH,

5

CHzCH,

6

CHzCH=CHz

7

CHzCH-CHz

0

(CH2)6CH3

rocene surfactant derivative.41 Very recently, Donohue and Buttry have surveyed the adsorption and micellization effects of a related series of ferrocene-containing amphiphiles by using quartz crystal microbalance techniques.‘2 Lee and Bard have also reported on the selfassembly of a surfactant viologen on glassy carbon.13 In our laboratory, the self-organization of the surfactant viologen 1 into monolayers on the surface of Pt and Au electrodes was also d e m ~ n s t r a t e d . ~ ~ Using self-assemblymethodology,Majda and co-workers have explored the properties of a unique type of electrode based on the modification of planar surfaces with an A1203 layer containing a high number of pores perpendicular to the surface. The oxide surface of the pores is covered with a monolayer of octadecyltrichlorosilane(OTS) that, in turn, drives the formation of a second monolayer of electroactive surfactant. The surface of the electrode can then be used to assess lateral electron transfer rates across the bilayer ~ t r u c t u r e Majda . ~ ~ has also recently reported on the electrochemical behavior of an amphiphilic ferrocene derivative at the air-solution interfacea and the monolayer self-assemblyof a surfactant viologen derivative on the surface of gold electrode^.^^ The self-assembly of certain sulfur-containing lipids on Au has been addressed by Regen and co-workers.w Mallouk et al. have described the adsorption of ordered zirconium phosphonate multilayer films on Si and A U . ~ ~ In spite of the recent activity which characterizes this field, the effects of structural features of the amphiphilic monomers on the electrochemical response of the interfacial aggregates have not been addressed in detail. Very little is known about this issue because most investigations have been performed with only an individual monomer for monolayer preparation and it is quite difficult to establish comparisonsbetween data obtained by different groups under usually diverse experimental conditions. We have initiated a systematic survey of the electrochemical properties of interfacial monolayers formed from redoxactive amphiphilic monomers exhibiting structural vari(41) Fecci, J. S. Langmuir 1987, 3, 525. (42) Dpnohue, J. J.; Buttry, D. A. Langmuir 1989,5,671. A.; Knifer, A. E. J. Electroanal. Chem. Interfacial Elec(43) h, trochem. 1988, 249, 333. (44) Miller, C. J.; M ‘da,M. J. Am. Chem. SOC.1986,108, 3118. (45) Miller, C. J.; W k g , C. A.; Charych, D. H.; Majda, M. J. phy8. Chem. 1988,92, 1928. (46)C y C. A.; Miller, C. J.; Majda, M. J. Phye. Chem. 1988,92,1937. (47) Mdler, C. J.; Majda, M. Anal. Chem. 1988,60, 1168. (48)Widrig. C. A.; Miller, C. J.; Majda, - - M. J. Am. Chem. SOC.1988, 110,2009. (49) Widrig, C. 9.; M ’da, M. Langmuir 1989,5, 689. (50) Fabianoweki, W.;%oyle,L. C.; Weber, B. A.; Graneta, R. D.;Caatnor, D. C.; Sadownik, A.; Regen, S. L. Langmuir 1989,5, 35. (61) Lee,H.;Kepley, L. J.; Hong,H. 0.;Akhter, S.;Mallouk, T. E. J. Phys. Chem. 1988,92,2597.

Comez et al. ations. It is hoped that this approach will allow us to establish correlations between the structural parameters of the monomers and the electrochemical properties of the interfacial monolayer assemblies. This should afford an improved foundation for the rational design of interfacial monolayer structures. In this paper we report on the aggregation from aqueous solution of a series of amphiphilic viologens on gold and SnOzsurfaces. The eight redox-activemonomers surveyed are shown in Chart I. We also report data concerning the degree of blocking of the gold surfaces by the self-assembled long chain viologen structures as demonstrated by the voltammetric behavior of two hydrophilic solutionspecies: the positively charged complex hexaamineruthenium(III),Ru(NH&P, and a zwitterionic (at neutral pH) ferrocene derivative, FC

Experimental Section Materials. Compounds1-5 were prepared in two steps. First, the monoquatemizedprecursor l-ethyl-4-(4’-pyridyl)pyridinium bromidewas synthesizedby miring 1.0g of 4,4’-bipyridine(Fluka) with an excess (5 mL) of bromoethane (Aldrich) in 10 mL of refluxingbenzene for approximately15 h. The reaction mixture, which shows a substantial white or pale green precipitate, was then diluted in 100 mL of toluene and stirred for at least 1 h to remove unreacted starting materials. The off-white solid was collected by filtration and recrystallized from hot acetonitrile. Alternatively, the solid may be dissolved in boiling chloroform and filtered, as the diquaternized product is not soluble in the hot solvent. Upon drying under vacuum, the monoquaternized product was characterized by 400-MHz lH NMR and DCI-MS. The isolated yield was about 45%. The asymmetric compound 1 was synthesized from this product and bromooctadecane (Aldrich) in refluxing acetonitrileovernight. The resulting yellow solid was stirredin chloroformand fiitered severaltimes to remove residual bromooctadecane(confirmedby DCI-MS) and then recrystallized from a minimum of ethanol. The final product (approximately60% yield) was again characterized by 400-MHz lH NMR (in DMSO-d6) and DCI-MS. Compounds 2-6 were synthesized by a similar procedure from the same monoquaternized precursor and the appropriate alkyl bromide analogue. The bis(hexafluoroph0sphate) salt of 1 was prepared by precipitation from aqueous solution using NHQFe. Compound 7 was also synthesizedin two steps. Firstthe monoquaternized derivative l-octedecyl-4-(4’-pyridyl)pyridmiumbromide was obtained from an equimolar mixture of 4,4’-bipyridine and bromoctadecanerefluxing in THFovernight. In this medium the product appeared in the reduced form, resulting in a deep blue solution that returned to colorless or pale yellow upon exposureto air, and developed a white precipitateupon cooling. After filtration, the solid is dissolved in hot DMF and filtered again to separate any residual diquaternized product. Addition of ether to this filtrate yielded a white powder (30%),characterized by DCI-MS and 400-MHz lH NMR. This product was then stirred in allylbromide (Fluka)for 2 h to yield the asymmetric viologen 7 (90%). Compound 6 was prepared by the same procedure but with bromohexadecane(Aldrich)instead of bromooctadecaneas the startingmaterial in the fiit quaternization step. 6 and 7 were recrystallized twice from methanol-diethyl ether mixtures. Structural confirmation was obtained by 400MHz lH NMR spectra in DMSO-& and DCI-MS analysis. The zwitterionic ferrocene derivative, FC, wae prepared by refluxing 4.0 mmol of 3-bromopropionicacid (Aldrich)with 5.0 mmol of (ferrocenylmethy1)dimethylamine(Aldrich) in 40 mL of benzene for 16 h. The precipitated crude product waa filtered off and washed with a small volume of acetone to remove any

Langmuir, Vol. 7, No. 8, 1991 1799

Surfactant Monolayers on Electrode Surfaces unreacted starting materials. The resulting yellow-orange solid showed two spots in TLC (MeOH/CHzCls [1:1] on silica). The = 0.25) could be bromide salt of the protonated form of FC (4 simply isolated by treating the mixture with dichloromethane in which the second, unwanted product (Rf= 0.11) was readily soluble. The structure of FC was verified by DCI-MS (MH+ peak at 397) and titration with NaOH, which yields a neutralization equivalent of 389.2 (calculated = 396.1). The pK, was found to be 3.5. The structure of the compound was also verified by 400-MHz 1H NMR spectroscopy (in DzO) but the integral values corresponding to the cyclopentadyenylprotons were lower than expected probably because of partial oxidation of the ferrocene group by dissolved oxygen (the acidic nature of the compound favors thie process). Compound 8 was obtained from Aldrich and used as received. Methyl viologen was also purchased from Aldrich and used without further purification. a-Cyclodextrin (a-CD) was purchased from Fluka and used without further purification. Ru(NH3)&13 was purchased from Strem Chemicals (Newburyport, MA). All solutions were freshly prepared with distilled water passed through a four-cartridge, Barnstead Nanopure water purification system. Equipment. Electrochemical experiments were performed with a Princeton Applied Research (PAR) Model 175 universal programmer, a Model 173 potentiostat, and a Model 179 digital coulometer equipped with positive feedback circuitry for IR compensation. Voltammograms were routinely recorded on a Houston Instruments Model 2000 X-Y recorder. Fast scan voltammograms were acquired with an €BM PC/XT Hyundai microcomputer through a Metra-byte DAS-lG(F) A/D board using Basic software developed around binary routines provided by the board manufacturer. NMR spectra were recorded in a Varian VXR-400-S spectrometer. Mase spectra were obtained by using direct chemical ionization (DCI), with ammonia as the carrier gas, in a VG Trio-2 spectrometer. Procedures. In a typical experiment gold flag working electrodes were cleaned by immersing briefly in hot concentrated nitric acid, rinsing thoroughly with purified water and then annealing to form a round bead in a gaa-oxygen flame. This pretreatment is known to produce a surface with a low roughness factor. This point was verified several times by comparing the actual electrode surface area (from the charge passed in the oxidation of a monolayer of chemisorbediodine)‘% the geometric area obtained from diffusion-controued experiments with Ru(NHs)e3+. The former was found to be larger by approximately 10-15%. Typical values for the geometric area of the bead electrodes ranged from 0.10 to 0.14 cml. Immediately after annealing, the electrode was immersed in fresh supporting electrolyte solution (0.50 M NazSOJ. After a satisfactory background was obtained, aliquot additions fram a 1.0 mM stock solution of the amphiphilic viologen (in the same supporting electrolyte) were made to adjust the concentration for studies in the 30-80 pM range. Separate solutions were prepared by dilution of the stock solution to obtain concentrations in the 0.10-0.50 mM range. At each concentration, at least 10 min was allowed before recording any voltammograms to permit the equilibration of the system. SnOlelectrodes were purchased from Delta Technologies(Stillwater, MN) and consisted of a pyrolitically deposited antimonydoped tin oxide (ATO) coating (300 nm of thickness, 100 Q/square) on float glass. The ATOs were mounted in a horizontal (parallel to the airsolution interface) configuration and dipped vertically onto the solution, but not fully immersed into it. The electrical contacts were provided by silver epoxy and covered by a liquid electrical tape. The glass sides of the electrode were sprayed with a thin coating of Teflon spray (Fluoroglide,Norton) to prevent aggregation on any plain glass surface that remained in contact with the solution. The electrodes were cleaned with Alconox and rinsed with large amounts of absolute ethanol and purified water before being mounted in the electrochemical cell. (62) Rodriguez, J. F.;Mebrahtu, T.;Soriaga, M. Chem. Interfacial Electrochem. 1987,299, 283.

P.J. Electrwrrcrl.

Po-,

V vs SSCE

Figure 1. Cyclic voltammetric response on Au of a 30pM solution of 1 in 0.5 M Na#04. Scan rate = 200 mV/s. In these experiments the supporting electrolyte solution was 0.50 M phosphate buffer (pH = 7). Viologen surface coverages were determined from the integration of the f i i t reduction wave. The integration was performed graphically for both slow and fast scan voltammograms. The error mlirgin estimated in the integrations is ca.5%.

Results and Discussion Interfacial Self-Assembly of 1-7 on Gold. A cyclic voltammogram abtained with a clean gold electrode immersed in a salution containing 30 pM 1 and 0.50 M NazSOr is shown in Figure 1. The voltammogram reveala two redox couples, exhibitinghalf-wavepotentials of 4 . 4 7 and -0.81 V vs SSCE, which correspond to the two consecutive one-electron reduction processes characteristic of the viologen subunit. The shape of the waves clearly suggests that the electroactive species is adsorbed on the electrodesurface. This is supported by the small potential differencebetween the cathodicand anodic peak potentials of each couple. Furthermore, the cathodic peak current of the first reduction process depends linearly on the scan rate, as we have already reported in our preliminary account of these phenomena.a The voltammogram in Figure 1shows a striking difference between the waves of each of the viologen redox processes. While the first reexhibits well-developed cathodic dox process (PQa+/PQ+) and anodic waves, the waves corresponding to the second redox couple (PQ+/PQ)are not well developed and preaent clear deviations from the Nernstian shape for surfaceconfimed redox processes. The reasons for these distortions appear to be related to sizable differences in the stability of the interfacial structure as the adsorbate naolecules are switched amongtheir three pmible redoxstates. Initially, the interfacial structure is brought about by viologen di-

Gomez et al.

1800 Langmuir, Vol. 7, No. 8, 1991 2.00

50 pM ClOVC2

t

~

f

1.60

4 0

I5

t

0.80t

t

0.00 0 0

50 mVh

5

.10

15

20 4 25

30

35 0

~

SCAN RATE, V/s

Figure 2. Scan rate dependence of the integrated chargefor the first cathodic peak on Au of a 0.1 m M solution of 1 in 0.5 M NaBO4.

cations (PQ2+)which self-assembleat the electrodesolution interface. We have done experiments in which the potential of the working electrode was stepped to -0.7 V vs SSCE,kept there for variable periods of time, and then linearly scannedback to the initial value of -0.2 V vs SSCE. These experimentsdemonstrated that the coverage of the electrode surface (as measured from the integration of the anodic peak) increased with the residence time at -0.7 V. In other words, when the interfacial structure is switched to the PQ+redox state, reduction of solution Pa2+species (a diffusion-controlled process that occurs slowly due to the low Pa2+solution concentration) results in a gradual accumulation (or precipitation) of PQ+ at the interface. Similar phenomena take place at the potentials of the PQ+/PQ couple, thus explaining the departure of this couple’s waves from Nernstian shapes. However, the reasons for the lower currents observed for this couple are still obscure. If the potential window is restricted to values capable only of driving PQ2+/PQ+ interconvRrsions Nernstian (reversible) voltammetric waves of the surface-confined type are recorded for this couple by using scan rates of 50 mV/s or faster. We performed a scan rate study with 1 to define experimental conditions that assure recorded currents which essentiallyminimize solution contributions. Since diffusion-controlledcurrents vary linearly with the square root of the scan rate and currents produced by adsorbed electroactive material are directly proportional to the scan rate, increasing the scan rate should enhance the relative current contribution from the surface-confined species. In fact, it can be easily proven that

where and ipa&represent the peak current densities (A/cm resulting from the dissolved and adsorbed electroactive species, respectively, l’o is the electroactive surface coverage (mol/cm2),Do iathe diffusion coefficient (cm2/s),COis the solution concentration of the electroactive species (mol/cm3), and Y stands for the scan rate used in the voltammetric experiment(V/s). This equation indicates that the current contribution from the freely diffusingspeciesdecreasesas the scan rate increases.Thus, even for concentrations close to the millimolar range, fast scan rates will minimize solution contributions. Figure 2 illustrates this point for a 0.10 mM solution of 1, plotting the values measured for the cathodic charge, determined from the integration of the first cathodic peak, as a function

-03

-0.7

FQTENTUL.,,V vs SSCE

Figure3. Cyclic voltammetric responseon Au of a 60pM solution of 5 in 0.5 M NazSO4. of the voltammetric scan rate utilized. This plot shows that the cathodic charge decreases initially in the !i!OOmV/s to 2.0 V/s scan rate range. However, at scan rates faster than 5.0 V/s, the cathodic charge is essentially constant within experimental error. This suggests that the current contribution from viologen dications in solution is very small under our experimental conditions,i.e. for scan rates faster than 2.0 V/s. Several reasons can be advanced to explain why eq 1is not exactly followed by our experimental results. For instance,eq 1is only based on peak currents and disregards the potentials at which the diesolved and adsorbed species undergo electron transfer with the electrode surface. Therefore, the half-wave potentials for the diffusioncontrolled and the surface-confined electrochemical processes may differ enough that their waves will not overlap. This fact is demonstrated by experiments with 8, which show (see Figure 3) differept reduction potentials for the surface-confined and diffusing species. In addition, the current contributed at very fast scan rates by the freely diffusing species is small even for submillimolar concentration levels and may go undetected, especially where the capacitive background currents are large. This is confmed by control experimentsin which the amphiphilic viologen was replaced by methyl viologen. No viologen current waves were detected at concentrations below 50 pM in these experiments. At scan rates faster than 1V/s, the currents were negligible for methyl viologen concentrations in the 5W2OO pM range, clearly indicating that the diffusion-controlled process does not generate detectable currents under our experimentalconditions.Thus, surface coverages were determined from the integration of the first voltammetriccathodic wave recorded with the electrode contacting a 50 wM-0.50 mM solution of the corresponding amphiphilic viologen, at scan rates of 1.0 V/s or faster. Surface coverage values are extremely useful since they provide indications about the type of structure formed on the electrode surface. Isotherm data for derivatives 1and 2 are plotted in Figure 4A as a function of concentration. The plot clearly shows that both compounds reach a limiting surface coverage in the concentration range surveyed. The octadecyl derivative reaches saturation

Langmuir, Vol. 7, No. 8,1991 1801

Surfactant Monolayers on Electrode Surfaces 0.50

A

.

0.40t 0. A

0.30.. A*

A /m -

A

0.20..

4 4

4

.

4 4

0.10--

.+

m

/ A m i -

Figure 5. Schematic representation of the self-assemblyof am0.20

0.00

0.60

0.40

0.80

1.00

SURFACTANT CONCENTRATION, "4

cu

E

0.50

B

Y

t

2

8Y

0

0.40-' *v vv 0.30'9

4

0

.

7,

v

v

0.20-

0.00 0.00

0.20

0.40

0.60

0.80

1.00

SURFACTANT CWCENTRATION. mM

Figure 4. Concentrationdependence of the surface coverage on flat Au electrodes of (A) the ethyl viologen derivatives, 1 ( O h 2 (A),3 (e),and 4 (W, and (B)the allylviologen derivatives,6 (VI and 7 (0). surface coverage within the same concentration range as the hexadecyl derivative (ca. 0.10 mM). The limiting coverage for the octadecyl derivative is, however, higher (4.1 X 10-10 mol/cm2) than that measured for the hexamol/cm2). Similar results decyl derivative (3.1 X were obtained with viologen derivatives6 and 7 (see Figure 4B),indicating that the replacement of the ethyl group by an allyl group in the short end of the asymmetric viologen structure does not have any substantial effectson the self-assembly of these amphiphilic dications on gold surfaces. It should be noted however that some differences between the ethyl and allyl systems were observed with regard to the stability of the monolayers formed, as discussed below. The limiting surface coverage obtained with both octadecyl derivatives can be also expressed in terms of surface area per molecule, ca. 41 A2/molecule. This value corresponds well to the cross section of the viologen group, suggesting that the octadecyl derivatives form a compact monolayer on gold in which the amphiphilic dications are oriented perpendicularly to the electrode surface. The interfacial area occupied by these viologen derivatives is in good agreement with values previously reported by Lee and Bard13 and more recently by Widrig and Majda4gfor structurally related viologen amphiphiles. It is interesting to note that the cross-sectional area of an alkyl tail is ca. 20A2,implying that the packing of the monomers in these viologen-based interfacialmonolayers is completely limited by the packing of the viologen groups. Therefore, at limiting coverage the lipophilic region of the monolayer

phiphilic viologens at the Au-solution interface.

assembly is not very compact and, thus, is not expected to act as a completely impermeable barrier toward hydrophilic species. While the surface coverage data provide strong evidence in favor of a perpendicular alignment of the octadecyl viologen derivatives 1 and 7 in the interfacial region, the question of orientation with respect to the electrode surface is more difficult to address. Do these amphiphilic derivatives adsorb with the viologen group (heads-in configuration) or with the aliphatic chain (tails-in configuration) pointing to the electrode surface? In principle, the fact that reversible viologen electrochemistry is observed suggeststhe proximity of the viologen groups to the electrode surface. However, recent results on a redox amphiphile of analogous length by Bard and Zhangl6 demonstrated reversibleelectrochemistry in a LangmuirBldgett monolayer,built in a tails-in configuration, across a 19 carbon atom alkyl framework. However, our own results with 7 provide some evidence for a heads in configuration. Monolayers of this compound exhibited some what enhancedstability at the electrode solution interface (vide infra). This stability is assumed to result from the interaction of the allylic moiety with the gold surface,which is consistent with heads-in adsorption of the viologen groups. Moreover, recent SERS/SERRS work by Cotton and co-workers also substantiates this adsorption configuration, albeit for a different metal substrate (Aghm The similarity in the resulting electrochemistry for 1and 7, along with the agreement in surface coverage values, support heads-in as the most likely adsorption configuration for both the octadecyl viologens. A schematic representation of the proposed model for the interfacial monolayer is given in Figure 5. This model is similar to the one postulated by Widrig and Majda for the adsorption of N-methyl-N'-octadecyl viologen dichloride on A U . ~ However, at certain viologen concentrations these authors found evidence for a second monolayer of viologen monomers by using a thin-layer electrochemical cell. Our data are consistent with a simple monolayer structure but we cannot rule out the existence of more complicated, bilayer or multilayer, structures. If this were the case, our data indicate that only the first monolayer, adjacent to the Au surface, is capable of exchanging electrons with the electrode surface. Since the adsorption of halide ions on Au and Pt is very strong, the possibility that bromide adsorption on the Au surface may precede and subsequentlydrive the adsorption of the amphiphilic viologen dications should be taken into account. Therefore, we prepared the bis(hexafluoropho8phate) salt of 1 in order to investigate its interfacial self-

-

(53) Lu,T.; Cotton, T.M.;H w t , J. K.;Thompeon, D.H. P.J. Phys. Chem. 1988,92,6978.

Gomez et al.

1802 Langmuir, Vol. 7, No. 8,1991 350 I

r

k 5

0.00

0.20

0.40

0.60

0.80

1.00

Concentration. m o ~ c m sx io7 Figure 7. C/r v8 C langmuir plots for 1 (A),2 (e),and the

hexafluorophosphate salt of 1 (m).

0

I

14 Mnfl.ll(ll

-0.70

POTEHIUL, V nSSCE

21

28

35

!A$

0.0

Figure 6. Cyclicvoltammetricreaponseon Au of a 15 pM solution of the bis(hexafluorophoephate)salt of 1in 0.5 M NazSO, at acan rates of 100-500 mV/s. Inset: Concentrationdependence of the surface coverage on flat Au electrodes recorded with this viologen salt. assembly in a halide-free medium. The results clearly indicate that self-assembly of the amphiphilic viologen dications does not require the presence of halide anions. Figure 6 shows an example of the voltammetric response recorded in these experiments. The inset shows a plot of the measured surface coverage as a functibn of concentration of the bis(hexafluorophosphate)salt. This system reaches saturation coverage at lower solution concentrations than the analogue bis(bromide) salt. This is probably a reflection of the decreased aqueous solubility of the former compound that shifts the equilibrium toward adsorption. The hexadecyl derivatives 2 and 6 exhibit lower saturation surface coverages than their octadecyl counterparts 1 and 7. The specific values are 3.1 X 1O-IO mol/cm2 for 2 and 3.3 X 10-lo mol/cm2 for 6. These can be easily converted into interfacial areas of 53.5 and 50.3 A2 per molecule, respectively. These values are iarger than the area occupied by an amphiphilic viologen dication adsorbed with its main axis perpendicular to the electrode surface and smaller than the area needed to accommodate adsorption in a "flat" conformation. This finding can be rationalized if one assumes that a small fraction of the hexadecyl viologen derivatives adopt a flat conformation while the rest remain adsorbed perpendicularly to the electrode surface. Alternatively, tilting of the viologen head groups or the presence of pinholes in the monolayer may also be invoked to explain the larger area per molecule measured with the hexadecyl derivatives. The lower molecular area observed with the octadecyl derivatives probably reflects the larger magnitude of favorable van der Waals interactions among the hydrophobic tails of these longer chain viologens. The less organized structure of the monolayers based on hexadecyl derivatives will also presumably affect access of hydrophilic electroactive species to the electrode surface (vide infra). The surface coverage vs concentration data for these systems can be treated, assuming that the adsorption

process follows the simple Langmuir model, according to the equation -=-+c 1 C (2)

r

Kr-

rm=

where C is the bulk concentration of amphiphilic viologen, r is the experimental surface coverage, I'max is the maximum surface coverage, and K is the equilibrium constant for the adsorption process. Plots of C/I'vs C were linear (correlation coefficientsabove 0.998, see Figure 7) for 1, 2, and the bis(hexafluorophosphate)salt of 1. From the slopes we determined r- values of (4.2 f 0.1) x 10-*0,(3.3 f 0.1) x 10-10, and (4.3 f 0.1) X 10-lo mol/ cm2, respectively, which are in all cases in excellent agreement with the experimentally observed saturation coverage values. The aalculated K values were (1.4 f 0.3) x 106, (1.4 f 0.3) x W , and (1.5 f 0.3) x 1Oe L/mol, respectively, from which Ace& values of approximately -29 kJ/mol were obtained for 1 and 2, while -35 kJ/.mol was obtained for the hexafluorophosphatesalt of 1. These values are in good agreement with values reported €orthe adsorption of a similar vio10gen~~ or a longchain ferrocene derivative.4l It is well-known however, that a good fit of the experimental data to eq 2 does not necessarily mean that the system obeys the assumptions of the Langmuir isotherm. In this case, the assumption of no interactions among the adsorbates appears to be particularly dubious, since favorable van der Waals interactions among the long alkyl tails seem to be a cqucial factor in the aelf-aseembly of the interfacial structures (vide infra). These interactions should play a strong role at coverage values close to the saturation limit. The inadequacies of the Langmuir model may account for the lack of discrimination in the A G d values obtained for some of these systems. This is most notable in the instance of 1 and 2 whose similar AG.& are inconsistent with the different saturation coverages measured with these two systems. Because of these inconsistencies we did not attempt to extend this analysis to the remaining viologens. Results for the shorter chain compounds appear to be analogous to those for the hexadecyl system and show gradually decreasing surface coverages as the length of the alkyl chain decreases. The limit for surfacsconfined behavior without a substantial diffusional component appears to be an alkyl chain of approximately 12 carbons units, as the decyl viologen shown in Figure 3 shows two sets of waves in the PQ2+/PQ+potential range, presumably resulting from adsorption and diffusion-controlled com-

Langmuir, Vol. 7, No. 8, 1991 1803

Surfactant Monolayers on Electrode Surfaces Scheme I. Different Degrees of Monolayer Permeation by Ru(NIia)'a* Resulting from the Imperfect Packing of the Alkyl Tails

ponents as the intensity of the latter (more negative redox peak) was relatively diminished at faster scan rates. In contrast to the above results, the symmetric viologen 8 does not aggregate appreciably on gold electrodes under identical experimental conditions. Hence,an asymmetric, amphiphilic structure and a long alkyl tail seem crucial for the interfacial self-assembly of the viologens derivatives. The long alkyl tail plays an important role in promoting interfacial aggregation,as evidenced by simple experiments based on the use of cyclodextrins. We have demonstrated recently that viologens 1 and 2 form inclusion complexes with a-CD in aqueous In these complexes,the alkyl chain of the viologen is included in the hydrophobic cavity of the cyclodextrin host.M Therefore, complexation with a-CD substantially decreases the lipophilic character of the amphiphilic viologen derivative due to the hydroxyl groups on the outer surface of the cyclodextrin,thus reducingthe driving force for Self-assembly. Addition of a-CD (final concentration = 1 mM) to a 0.20 mM solution of 1 in 0.50 M NazSOl results in the complete disappearance of the surface viologen waves at fast scan rates (25 V/s), indicating clearly that complexation by a-CD eliminates the tendency of 1 to aggregateat the electrode-solution interface. Therefore, the interfacial self-assembly of 1 seems to be largely dependent on the sizable hydrophobic interactions among the alkyl tails of the viologen derivatives. The degree of packing and the level of organization of the interfacial monolayer are thus enhanced by the increasing length of the alkyl tail. Effects of Interfacial Viologen Monolayer Assemblies on the ElectrochemicalBehaviorof Hydrophilic Species. The derivatization of the electrode surface with highly organized lipophilic structures may, in principle, completely block the electrode if the structure is sufficiently well packed, i.e., the electrode becomes unresponsive to the direct oxidation or reduction of water-soluble, electroactivespecies. The hydrophilic speciesmust diffuse to the electrode surface, a process hindered by the amphiphilic interfacial assembly,for the electron transfer to take place. In practice, however, a lack of molecular organization, or the presence of structural defects such as pinholes in the interfacial assembly, may allow the direct heterogeneous electron transfer reaction, as the solution species can directly contact the electrode surface in the region of such defects. A number of groups have investigated the degree of blocking of the electrode surface (54) Diaz, A.; Quintela, P.A.; Schuette, J. M.;Kaifer, A. E.J. Phys. Chem. 1988,92,3537. (55) Kaifer, A. E.; Quintela, P. A.; Schuette, J. M.J. Inclueion Phenom. 1989, 7, 107. (58) Ienin, R.; Yoon, H.R.; Vargae, R.; Quintela, P.A.; Kaifer, A. E. Carbohydr. Rea. 1989,192,357.

--

-O*'O

POTENlU&VnSSCE

Lo

4.70

pcr"&vn=

Lo

Figure 8. (A) Cyclic voltammetric response (at 200 mV/s) on a Au electrode of a 1.0 mM solution of Ru(NH&C&in 0.5 M NazSOl in the absence (dotted line) and in the presence (continuousline) of 0.1 mM 2. (B) Similar cyclic voltammograms in the absence (dottedline)and in the presence (continuous line) of 0.1 mM 1.

caused by monolayers prepared by using either selfassembly16J811 or Langmuir-Blodgett16 methods. In the case of the viologen monolayers,although the long aliphatic tails provide the driving force for the interfacial structure, it seems that the electroactivehead groupslimit the extent of packing of the monolayer. Thus, it may be possible for a diffusing speciesto permeate the monolayer in the loosely packed alkyl region but encounter a completely blocked surface in the region of the head groups. Such a situation is depicted in Scheme I. In this case, it would be possible for the diffusing species to exchange electrons with the viologen head groups. In summary, the viologen-based monolayers surveyed in this work may present a barrier for water-soluble species which is created by the looeely packed alkyl tails of the monomers. In addition, the positive charge on the viologen head groups can create an electrostaticbarrier for positively charged solutionspecies. In order to assess the relative importance of lipophilic and electrostatic interactions, we utilized both a positively charged and a zwitterionic speciesin experiments designed to probe the degree of blocking of the electrode surface by surfactant viologen monolayers. Ru(NH3)sS+was selected as the cationic hydrophilic probe because of its fast and accessible 3+/2+ redox couple.67 Its cationic nature prevents ion exchange interactions with the network of viologen subunits in the proximity of the electrode surface. Ion exchange into the monolayer would be the predominant factor in the case of an anionic hydrophilic probe such as ferrocyanide.47*a In a typical experiment the voltammetric behavior of a 1 mM solution of Ru(NH3)sCla was recorded and then sufficient viologen was added to bring the solution to the desired concentration of viologen. The system was slowly allowed to reach equilibrium, usually for a minimum of 15 min. The subsequently recorded scans showed viologen and Ru(II1) waves for the hexadecyl system 2, as shown in Figure 8A. The dotted line voltammogram represents the diffusion-controlled reduction of Ru(III), while the solid line voltammogram corresponds to the Rumviologen monolayer system. As is easily noted from the figure, (57) Dominey, R. N.;Lewis, N. S.;Bruce,J. A.; Bookbinder, D. C.; Wrighton, M. 5.J. Am. Chem. SOC.1982,104,487. (58) Our own experimenta with monolayer-covered electroden contacting ferrocyanide solutions a t micromolar concentration levebindicate extensiveion exchangeof ferrocyanideanionainto the networkof poeitively charged viologen head group in the monolayer.

Gomez et al.

1804 Langmuir, Vol. 7, No. 8, 1991

Table I. Voltammetric A 4 Values (in mV) for the Ru(N&)6*++/*+ Couple Measured on a Au Electrode in the Absence and in the Presence of Several Amphiphilic Viologens at the 0.1 mM Concentration Level electrode surface 0.1 VIS 0.2 VIS 0.5 VIS naked Au 62 65 67

Scheme 11. Monolayer-Mediated Reduction of Ru(NHa)6'+

-Ya

2-covered Aua 6-covered Aub 4-covered Auc

solutim

both redox couples are seen, although the Ru(II1) redox couple appears to suffer in reversibilitydue to the presence of the interfacial viologen structure. In contrast, the electrochemistry obtained for a similar system composed of the more organized octadecylviologen 1 is substantially different, as shown in Figure 8B. The shape of this voltammogram indicates that the direct heterogeneous electron transfer of the Ru(II0 complex with the Au surface is so hindered that the reduction of this species is mediated by the viologen redox sites of the monolayer. This indicates an excellent level of blocking by the viologen monolayer. Mediation by the interfacial structure, as represented in Scheme 11, becomes possible when the reduction peak of the Ru(II1)speciesis sufficientlyshifted to potentials as negative as those necessary to drive the conversion from PQz+ to PQ+ in the monolayer. The mediation mechanism is clearly evidenced by the greatly diminished viologen anodic peak on scan reversal which reflects the disappearance of the reduced viologen species PQ+ by the reaction

PQ'

+ Ru(NH,)F

-

PQz++ Ru(NH,)F

(3)

This reaction is thermodynamically favored because the reduction potential of the Ru complex is higher than that of the surfactant viologen. Nonetheless, the onset of mediation requires a very compact and organized interfacial structure that must be substantially free from defects or pinholes and capable of exerting a substantial kinetic hindrance on the direct heterogeneous electron transfer reaction between the Ru(II1) species and the electrode surface. The voltammogram obtained with the 2-Ru(III) system is analogous to those reported by Bard and Zhang for a system composed of a Ru2+-based,Langmuir-Blodgett monolayer and a cationic species, 0s(bpy)s2+,as a diffusional probe.15 It is important to note that while their results were obtained with a longer aliphatic system (C1& the experiments were performed with IT0 electrodes, which generally have a high roughness factor that may have marked effects on monolayer organization as noted below. We wish to emphasize that our results with the 2-Ru(III) system clearly demonstrate that direct Ru(I1I) solution electrochemistry across the positively charged layer is possible for the interfacial viologen structures composed of amphiphiles with CUor shorter chains which presumably lead to less organized, more loosely packed monolayers. Therefore, electrostatic effects alone do not interfere to suchan extent as to block the electrochemistry of the solution species. To further demonstrate this point we examined the other hexadecyl system 6 along with a shorter viologen system 4 under similar conditions. The data are shown in Table I. The Upvalues exhibit the expected magnitude for a fast diffusion-controlled electrochemical couple in the absence of viologen (60 mV at 20 mV/s) but increase substantiallly in the presence of 0.10 m M viologen, a concentration that ensures the formation of an interfacial monolayer. All three viologen derivatives appear to slow down the Ru(NH&3+/2+ interconversions,but the hexadecyl derivatives 2 and 6 seem

110 135 75

125 155

80

185 250 80

a Surface coverage = 3.1 X m o l / c m 2 . Surface coverage = 3.3 x 10-10mol/cm2. c Surfacecoverage = 1.2 x Womol/ cm2.

to be more efficient in this respect, as indicated by the larger AE,values,while the peak separation in the presence of 0.10 mM 4 shows only a slight increase in AE, values in comparison to the values obtained with the bare Au electrode. These data support our previous proposal that shorter alkyl chains result in less organized structures on the electrode surface and decrease the tendency of the viologen derivative to self-assembleat the interface, hence increasing the ease by which the solution species can undergo electron transfer with the underlying surface. The greatest degree of molecular organization and packing is thus observed when the electrode surface is exposed to solutions of the octadecyl derivatives 1 and 7. However,severalexperimentswere performed to ascertain the experimental conditions necessary to observe monolayer-mediated reduction of the Ru(II1) complex. It was found, for instance, that if the solution concentration of 1 is taken below 60 pM, then no mediated reduction was observed and the recorded voltammograms were similar to that shown in Figure 8A. This is consistent with the data of Figure 4A which indicate that at concentrations of 1below60 pM the monolayerhas not reached ita limiting coverage and thus its barrier or blocking properties are not as effective as those attainable at saturation coverage (fully packed monolayer). A related issue has to do with the smoothness of the electrode surface. Can equally blocking monolayers be self-assembled on substrates having very differentroughnessfactors? In order to answer this question qualitatively, we performed some experiments on polished disk gold electrodes, which have substantially higher roughness factors (about 2) than the smooth, flame-annealed gold electrodes otherwise employed in this work. As expected the voltammetric behavior observed on the rougher polished electrodes is similar to that shown in Figure 8A, without any indications of monolayer mediation. This finding indicates that molecular organization and packing of amphiphilic structures self-assembled at the electrode-solution interface are adversely affected by a rough electrode surface. Of course, this is not completely unexpected but clearly emphasizes the importance of a smooth substrate in preparing highly organized and compact, self-assembled monolayers from amphiphilic monomers. In order to better discriminate between electrostatic and lipophilic effects on the barrier properties exhibited by these interfacial monolayers, we performed a similar set of experiments with the ferrocene derivative, FC, that exhibits a zwitterionic structure at neutral pH values. Then, the oxidation of this species should not be affected by electrostatic barrier effects. Of course, the oxidative nature of the chemistryof this probe makes the observation of mediationby the viologen monolayer impossible. Figure 9 shows the voltammetric behavior recorded with a 1.0 mM solution of FC (buffered at pH = 7.03)on a bare Au electrode and on the same electrode covered with a monolayer of the octadecyl derivative 7. The voltammetric

Langmuir, Vol. 7, No.8, 1991 1805

Surfactant Monolayers on Electrode Surfaces

POTENTIAL,V va SSCE

Figure 9. Cyclic voltammetric response on Au of a 1.0 mM solution of FC in 0.5 M NazSO, (bufferedat pH = 7.03 with 12 mM phosphate) in the absence (dottedline) and in the presence (continuous line) of 0.1 mM 7. Scan rate = 250 mV/s. behavior for the oxidation of the ferrocene moiety in FC is reversible on bare gold, as expected, consistently exhibiting AE, values between 60 and 65 mV for scan rates in the 20-500 mV/s range. The half-wave potential was determined to be +0.39 V vs SSCE. However, in the presence of 0.1 mM 7, the voltammetricbehavior indicates decreased reversibility with AE, values ranging from 90 to 130mV(for 20-500mV/s). Similar resultsare obtained in the presence of 0.1 mM 1, with AE, values going from 90 to 110mV in the same range of scan rates. Thus, monolayers of both octadecylderivatives substantially decrease the apparent rate of heterogeneous electron transfer for the FC+/FC couple. These data strongly suggest that a positively charged, self-assembled viologen monolayer hinders the electron transfer reaction between hydrophilic species and the underlying electrode surface even for species having no net positive charge like the zwitterionic derivative FC. This can be explained by the lipophilic barrier that the alkyl tails present to diffusing hydrophilic species. Thus, electrostatic effects appear to be of secondary importance although they can reinforce the lipophilic effects generated by the alkyl tails of the amphiphilic monomers. To confirm our assessment about the relative importance of the alkyl tail's length in the blocking effects, we performed a series of experiments with similar surface coverages of 1 and 2. This ensures that the electrostatic effects are held constant and that only the organization of the hydrophobic tails varies. Thus,we selected asurface coverage of about 3.1 X 10-10 mol/cm2,which is below the maximum coverage of 1 and at the maximum value we have observed for 2. In the case of 1, this level of coverage was enough to cause the mediated reduction of the Ru(111)complex. In contrast, at this coverage of the hexadecyl derivative, 2, no mediation was observed. This is significant since this is the maximum coverage attainable with this derivative. This clearly indicates that hydrophobic effects appear to be the dominant factor in the barrier properties. Although hydrophobic effects dominate this scenario, electrostatic contributions still exist. For instance, careful analysis of Figure 9 reveals the influence of electrostatic interactions on the overall blocking effects: in the presence of the viologen monolayer, the shape of the anodic wave (for the FC FC+conversion) is comparativelymuch less FC distorted than the cathodic wave (for the FC+ conversion). This can be interpreted as the result of the added hindrance in the approach of the electroactive species to the electrode surface, due to the electrostatic repulsion between the viologen monolayer and the oxidized ferrocene species FC+. In summary, the results of these experiments are consistentwith the predominance of lipophilic interactions

-

-

-io

-6s

-as

-a4

.dz

(0

POTENTIAL, V vs SSCE Figure 10. Cyclic voltammetricresponse on SnOz of a 0.10 mM solution of 1 in 0.5 M phosphate buffer (pH = 7). Scan rate = 1.0 v f 8.

in the creation of interfacial barriers by the self-assembly of surfactant viologen monomers. Electrostatic effects do not seem to play a prominent role but they do enhance the barrier properties toward positively charged hydrophilic species. Interfacial Self-Assembly of 1,2,6, and 7 on SnO2. All of the asymmetric viologens presented were found to aggregate on SnO2 to a varying degree. Our primary interest was, however, in describing the interfacial selfassembly of viologens 1,2,6, and 7, as these were known to form more organized layers on gold. A typical cyclic voltammogram of 1 on SnO2 is shown in Figure 10. It shows the two successive one-electronreduction processes expected for the surface-confined viologen groups. As was the case on gold, the second reduction couple (PQ+/PQ) exhibits broad and poorly developed current waves compared to the f i t reduction couple (PQ2+/PQ+).Departure from reversible surface waves and precipitation effects were observed when the potential of the SnO2 surface was kept negative from the PQ2+/PQ+half-wave potential for periods of time longer than those needed to complete the scans. This was also seen on gold; however, the voltammetric behavior appears to be more complicated on SnO2, particularly at slow scan rates. Voltammograms recorded at slow scan rates exhibit an anodic shoulder at more positive potentials than the main anodic peak. This shoulder disappears as the scan rate is increased. This behavior was never observed on gold surfaces. Nonetheless, the cathodic peak current vs scan rate plot was linear as expected for a reversible surface-confined redox couple. The surface coverage values on SnO2 were again determined from the integration of the first cathodic peak in the cyclic voltammograms recorded with electrodes immersed in buffered solutions (0.5 M phosphate, pH = 7) containing concentrations of the amphiphilic viologens in the 50 pM to 0.20 mM range (Similar results were obtained in 0.50 M NazSO,.) The geometric area was determinedby using the Randles-Sevcik equation on peak currents obtained with Ru(NHs)P, which exhibits reversible, diffusion-controlled electrochemistry on SnOz. The determination of accurate surface coverage values on SnO2 was problematic due to the tendency of the amphiphilic viologen to aggregate on glass4@and the high

1806 Langmuir, Vol. 7, No. 8,1991

roughness factors as estimated by the manufacturer. The former necessitated the design of the electrodes described in the Experimental Section in order to control the area available for aggregation. The latter factor cannot be circumvented at the present time, and thus, the coverage values presented here are quite high because they are based on uncorrected geometric areas. The octadecyl derivatives 1 and 7 both yielded similar surface coverage values in the range of 1 X lo4 mol/cm2 at a solution concentration of 0.10 mM, indicative of the considerable roughness of the surface when compared to the value obtained on smooth gold at the same concentration which was lower by a factor of approximately 2.5. Similar high values were found for the hexadecyl derivatives as well. To further demonstrate that the high coverage values obtained are related to the surface roughness, experiments with 1.0 mM Ru(NH&& and 0.10 mM 1 showed no mediation by the viologen layer of the Ru(II1) reduction, an expected result since the high roughness factor of the electrode surface impedes the formation of a compact and organized monolayer at the interface. Our results clearly evidence the self-assemblyof 1-7 on gold and SnO2. Previously, we have reported that 1 also self-assembles on platinum.43 Preliminary experiments on silver also confirmed the aggregation of 1 on this surface. It is then evident that the interfacial self-assembly of asymmetric amphiphilic viologensfromaqueous solutions is rather general. Indeed, this is in good agreement with the hydrophobic interactions among the alkyl tails being the driving force for the interfacial aggregation of these asymmetric viologens (vide supra). However, it must be pointed out that the roughness of the electrode surface plays an important role in determining the level of organization of the interfacial assembly;i.e., a very rough surfacewill disrupt the molecular organization of the mone layer. Interfacial Stability of Self-Assembled Monolayers of 1 and 7. An important advantage of monolayers prepared by the self-assemblyof amphiphilic viologens 1 and 7 is the simplicity of their preparation. Additionally, it would be highly desirable that these supramolecular aggregates maintain their structure and organization for long periods of time and in a variety of chemical environments. Therefore, we decided to investigate the stability of the monolayers in the absence of amphiphilic viologen in the contacting solution. We exposed a clean gold electrode to a 0.10 mM solution of 1 in 0.5 M Na2S04, detected the self-assembledmonolayer, and measured the electroactive surface coverage. Then, we transferred the

Gomez et al. electrodeto a 0.5 M NafiO, solutionlacking 1and recorded the surface coverage as a function of time. The transfer step must be performed carefully to prevent monolayer disruption. We have found that if the electrode surface is rinsed during the transfer, the majority of the monolayer assembly is washed away. However, if the electrode is removed from the viologen solution and immersed, without rinsing, into a pure supporting electrolytesolution, surface viologen waves are observed after the transfer, although the coverage was found to decrease gradually as a function of time. After 20 min, less than 50% of the initial coverage could be detected. As mentioned above, a somewhat greater degree of stabilization was noted with the allylic viologens. These experiments clearly demonstrate that these monolayers show relatively poor stability at the electrode solution interface, even on very smooth gold surfaces. In fact, a micromolar concentration of the corresponding amphiphilic viologen must be maintained in the contacting solution in order to keep the supramolecular organization of the viologen dications at the interface. Similar findings have been reported by Widrig and Majda for structurally related viologen m~nolayers.~B We have recently developed a chemical scheme that greatly increasesthe stability of these interfacial assemblies by the use of anionic polyelectrolytes such as poly(styrenesulfonate).m Conclusions This work has shown that amphiphilic asymmetric viologens 1 and 7 self-assemble at the electrode solution interface forming monolayers in which the viologen head groups are packed tightly while the hydrophobic alkyl tails exhibit rather loose packing. The driving force for the interfacial self-assembly seems to be the hydrophobic interactions among neighboring alkyktails in the monolayer structure. These monolayers hinder the heterogeneous electron transfer between a water-soluble Ru(II1) complex (or a zwiterionic ferrocene derivative) and the underlying electrode surface. Acknowledgment. The authors are grateful to Maida Betancourt, Jose Delgado, Abigail Dim, Orlando Garcia, Rahimah Isnin, and Dr. Carmen A. Vega for performing preliminary experiments. Financial support from the National Science Foundation (CHE-9000531)is gratefully acknowledged. The purchase of the NMR and MS spectrometerswas made possible by grants from NIH (RR03351 and RR-04680, respectively). (59)Gomez, M.;Li,J.; Knifer, A. E. Langmuir, in press.