Langmuir 1991, 7, 2362-2369
2362
Electrochemical Measurements of Electron Transfer Rates through Zirconium 1,2-Ethanediylbis(phosphonate) Multilayer Films on Gold Electrodes Hun-Gi Hong and Thomas E. Mallouk' Department of Chemistry and Biochemistry, The University of Texas at Austin, Austin, Texas 78712 Received December 7, 1990. I n Final Form: May 3, 1991 The rates of electron transfer between gold electrodes and solution-phaseor covalently attached redox molecules were measured as a function of the thickness of an intervening metal phosphonate thin film. Multilayer films of zirconium 1,2-ethanediylbis(phosphonate)(ZEDP) were prepared on gold electrodes by sequential adsorption of aqueous zirconyl chloride and 1 2-ethanediylbis(phosphonicacid) solutions. Ellipsometryshows a stepwise increase in film thicknessof 8 A per ZEDP layer. A logarithmicdependence of the heterogeneous electron transfer rate constant for the oxidation of Fe(CN)& in 0.1 M NaClOe-on the thickness of the ZEDP multilayer film was obtained from Tafel plots. The low electron tunneling constant (0.43A-l) obtained suggests that defect sites in the ZEDP multilayer film dominate the electrochemical response. Films containing covalently bound ferrocene-terminatedphosphonate groups and ZEDP spacers were studied by the same technique. Similar values of the tunneling constant were obtained, even after most of the ferrocene units were replaced by electroinactive benzylphosphonate spacers. However, potential step chronoamperometry revealed kinetic heterogeneity in the ferrocene layer even at a ferrocene mole fraction of 0.2. This indicates that there is a heterogeneous distribution of ferrocene sites in the monolayer and that electron hopping between ferrocene units may be important in mediating charge transport between the least accessible electroactive groups and the metal electrode.
Introduction Organic films on solid substrates have been the focus of considerable research effort, because of the important role that rationally designed interfacial structures could potentially play in wetting, adhesion, biocompatibility, nonlinear optics, catalysis, and numerous other applications. Self-assembly and related techniques have provided a means to fabricate, at ambient pressure and in an ordinary laboratory environment, interfaces of very welldefined structure. Films of n-alkylorganosulfur compounds such as sulfides,6and disulfides6on gold surfaces, alkylsilanes covalently attached to glass' and silica: and fatty acids adsorbed on oxidized aluminums and silver'o have recently been explored in an effort to understand and control the properties of solid interfaces. (1) (a) Finklea, H. 0.; Robinson, L. R.; Blackburn, A.; Richter, B.; Allara, D. L.; Bright, T. Langmuir 1986,2,239. (b) Finklea, H. 0.;Avery, 5.; Lynch, M.; Furtach, T. Langmuir 1987, 3,409. (2) (a) Bain, C. D.; Whitesides, G. M. J. Am. Chem. SOC.1988,110, 3665. (b) Bain, C. D.; Whitesides, G.M. J. Am. Chem. SOC. 1988,110, 5897. (c) Bain, C. D.; Whitesides, G.M. J. Am. Chem. SOC.1988,110, 6560. (d) Waeserman, S. R.; Whitesides, G. M.; Tidswell, I. M.; Ocko, B. M.; Pershan, P. S.; Axe, J. D. J. Am. Chem. SOC.1989, 111, 5852 and references therein. (3) Nuzzo, R. G.;Dubois, L. H.; Allara, D. L. J. Am. Chem. SOC.1990, 112, 558. (4) Porter, M. D.; Bright, T.B.; Allara, D. L.; Chidsey, C. E. D. J. Am. Chem. SOC.1987, 109,3559. (5) Troughton, E. B.; Bain, C. D.; Whitesides, G. M.; Nuzzo, R. G.; Allara, D. L.; Porter, M. D. Langmuir 1988,4, 365. (6) (a) Nuzzo,R. G.;Allara, D. L. J. Am. Chem. SOC.1983,105,4481. (b) Nuzzo, R. G.;Zegmki, B. R.; Dubois, L. H. J . Am. Chem. Soc. 1987, 109,733. (c) Nuzzo, R. G.;Fusco, F. A.; A h a , D. L. J. Am. Chem. SOC. 1987. - - - ., -10.9. - -,2RAR (7) Sagiv, J. J. Am. Chem. SOC.1980, 102, 92. (8) (a) Maoz, R.; Sagiv, J. J. Colloid Interface Sci. 1984,100,465. (b) Wasserman, S. R.;Tao, Y.-T.; Whitesides, G.M.Langmuir 1989,5,1074.
(c) Tillman, N.; Ulman,A.; Schildkraut, J. S.;Penner, T.L. J. Am. Chem. SOC.1988,110, 6136. (d) Maoz, R.; Sagiv, J. Langmuir 1987,3, 1034-
-.,--.
1 nfii
(9) (a) A h a , D. L.; Nuzzo, R. G.Langmuir 1985, 1, 45. (b) Allara, D. L.; Nuzzo, R. G. Langmuir 1985, 1,52. (10) Schlotter, N. E.; Porter, M. D.; Bright, T.B.; Allara, D. L. Chem. Phys. Lett. 1986, 132, 93.
Recently, we have found" that stable, well-ordered multilayer films based on metal phosphonates can be prepared on a variety of surfaces via sequential adsorption techniques (Scheme I). These films are prepared by alternately adsorbing tetravalent or trivalent metal ions and a,u-bis(phosphonic acids) from aqueous solution. The technique has two essential advantages: first, that the thickness of the film can be controlled to angstrom resolution by choosing the appropriate phosphonic acid chain length and number of layers; second, that films prepared in this way are chemically and mechanically stable, free of pinholes, compact, and well-ordered. Since these films are electronically insulating, it is of interest to see if they can be used as spacer layers for the measurement of molecule-to-electrode electron tunneling rates. Such measurements provide information which is relevant to the design of electronic devices which might employ these films as insulators and also provide a molecular probe of structural defects in the films. Li and WeaverI2 found that the electronic coupling between an electrode and an electron acceptor (a cobalt(111) ammine complex) decreased exponentially with the number of atoms in the bifunctional ligand connecting the Co(II1) center to electrode surface. They measured an electron tunneling constant, 0, of 1.4A-1, for saturated alkyl spacer groups. 0 is defined according to eq 1, in
which dl and dz represent acceptor-electrode distances with two different spacer chains. Attempts to measure (11) (a) Lee, H.; Kepley, L. R.; Hong, H.-G.; Mallouk, T. E. J. Am. Chem.Soc. 1988,110,618. (b) Lee,H.;Kepley,L. J.;Hong,H.-G.;Akhter, S.; Mallouk, T. E. J. Phys. Chem. 1988,92,2597. (c) Akhter, S.; Lee, H.; Hona, H.-G.; Mallouk, T. E.; White, J. M. J. Vac. Sci. Technol., A 1989, 7 , la8. (12) Weaver, M. J.; Li, T. T.-T.J . Am. Chem. SOC.1984,106,6107. (b) Li, T. T.-T.; Weaver, M. J. J . Electroanal. Chem. Interfacial Electrochem. 1985,188,121. (c) Weaver, M. J.;Li,T. T.-T.J.Phys. Chem. 1986, 90,3823.
0743-7463/91/2407-2362$02.50/0 0 1991 American Chemical Society
Langmuir, Vol. 7, No. 10, 1991 2363
Measurement of Electron Transfer Rates
Gold
4
cj:
Scheme I. Growth of ZEDP Layers on Au by Sequential Adsorption of EDPA a n d Zr(1V) =4'%;im3"2
. Fg2.C
"20JPCZH4
PqHi
(3)
long-range electron tunneling through seemingly wellordered alkylthiol and alkylsilane monolayers to redox couples in solution have not been s u c c e ~ s f ubecause l ~ ~ ~ of the presence of defects within the monolayers. Because of the exponential dependence of rate on distance, rapid electron transfer at a low concentration of defect sites can make a dominant contribution to the measured electron transfer current. Chidsey et al.13 reported Nernstian electrochemical behavior from monolayers of alkanethi01s terminated with polar ferrocene groups diluted with unsubstituted n-alkanethiols on gold. However, they found that the apparent electron transfer rate constant decreased and obtained a value for the electron tunneling constant on the order of 1A-l, when electron transfer at defect sites was removed through exchange of ferroceneterminated thiols with unsubstituted thiols. In the first part of this work reported here, we have used zirconium 1,2-ethanediylbis(phosphonate) (ZEDP) layers on gold as insulating spacers, to control the rate of electron transfer between the electrode and ferrocyanide ions in solution. Scheme I shows the procedure followed for preparing ZEDP multilayers on gold substrates. Ellipsometric examination of these multilayer films reveals that they grow in a stepwise fashion, according to Scheme I, with the stacking axis of the layered phosphonate salt oriented perpendicular to the surface. The well-defined structure and variable thickness of the ZEDP multilayers provide a direct means of probing the distance dependence of heterogeneous electron transfer across the insulating film. In the second part, we have used a ferroceneterminated phosphonate as the electroactive probe molecule in a covalentlybound monolayer, in order to eliminate the complication of a diffusingsolution-phase redox couple. Insertion of a ZEDP layer then allows one to probe the distance dependence of electron tunneling directly. By preparing mixed monolayer films containing the ferroceneterminated phosphonate and an electroinactive "spacer" phosphonate, we have attempted to minimize the effects of charge transfer diffusion within the monolayer, which might be expected to lead to unrealistically low values of the tunneling constant j3. Experimental Section P r e p a r a t i o n of 4-Mercaptobutylphosphonic Acid (MBPA). Glassware and syringes were dried in an oven at 120 "C overnight. A 100-mL, three-neck round-bottom flask was fitted with an air-cooled condenser which was connected to a water-cooled condenser. The flask was heated to 80 OC on an oil bath and purged with nitrogen. Forty milliliters of 1,2-dibromobutane (0.34 mol) was injected into the flask via a syringe and was stirred under nitrogen. The flask was heated to 150 OC, and 10 mL of triethyl phosphite (0.06 mol) was added dropwise via a syringe under nitrogen. The gaseous ethyl bromide evolved from the reaction wascondensed, collected,and measured. When ita volume was ca. 4.5 mL (0.06 mole), the reaction was stopped. The reaction flask was removed from the oil bath and cooled to room temperature. Excess dibromobutane was removed by ~~
(13) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am. Chem. SOC. 1990,112, 4301.
vacuum evaporation at 50 OC, and the product waa separated by column chromatography with silica gel (60-200 mesh). Trace dibromobutane was removed by eluting with hexane, and the diethyl 4-bromobutylphosphonate was eluted with ethyl acetate. After the removal of ethyl acetate, 13.5 g of diethyl 4-bromobutylphosphonate was obtained (yield ca. 85%1. lH NMR (Varian EM 390) (CDCls), 6 1.3 ppm (t, 6 H, -OCHzCHs), 6 1.5-2.0 (m, 6 H, -CH2CHzCHzP-), 6 3.4 (t, 2 H, BrCHz-), 6 4.1 (q, 4 H, -OCHz-). A solution of diethyl 4-bromobutylphosphonate (13.5 g, 0.05 mol) in DMSO (20 mL) was added dropwise to a DMSO solution of NaSH (4.5g, 0.08mol)under nitrogen. The mixture was heated at 70 OC for 4 h on an oil bath. During the reaction, NaBr formed as a white precipitate. The reaction mixture was cooled to room temperature and the precipitate was removed by filtration. The DMSO solution was extracted 10 times with 20-mL portions of diethyl ether. Evaporation of the ether gave ca. 9.8 g of diethyl 4-mercaptobutylphosphonate(yield ca. 90%). lH NMR (Varian EM 390) (CDClS), 6 1.3 ppm (t, 6 H, -OCHzCHs), 6 1.5-1.9 (m, 4 H, -CH2CH2P-), 6 2.3-2.7 (m, 4 H, HSCHzCHz-), 6 4.1 (quintet, 4 HI -OCH2-). Concentrated (35%)HC1,15 mL, was added to 9.8 g of diethyl 4-mercaptobutylphosphonate and refluxed at 100-110 OC for 8 h. The reaction flash was cooled to room temperature, and a white crystallineprecipitate appeared. After vacuum evaporation of water, the white powder was washed with HPLC grade acetonitrile and dried in a desiccator over phosphorus pentoxide. 4-Mercaptobutylphosphonic acid was identified by 1H NMR. 1H NMR (GE QE300) (DMSO-de), 6 1.4-1.7 ppm (m,6 HI -CH2CH&H2P-), 6 2.6 (t, 2 H, HSCH2-). Preparation of Potassium 4-(Ferrocenecarboxamido)benzylphosphonate (FcBP). All glassware was dried in an oven at 120 OC overnight. Ferrocenecarboxylic acid (0.47 g, 2 mmol) was dissolved in the dry methylene chloride (10 mL) in a threeneck round flask equipped with dropping funnel. The system was immersed in the ice bath. Oxalyl chloride (1mL, 10 mmol) dissolved in dry methylene chloride (10 mL) was added dropwise to the ferrocenecarboxyl acid solution was stirring for 30 min under nitrogen. The ice bath was removed after the addition was complete. The reaction mixture was stirred for 1h at room temperature and refluxed for 10min. After vacuum evaporation of the methylene chloride, a dark red solid remained in the flask. This residue was extracted with dry boiling pentane 3 times. Removal of the pentane gave 0.47 g of red microcrystalline ferrocenoyl chloride (C&FeC&HCOCl),yield 95%. Diethyl (p-aminobenzy1)phosphonate(0.45 g, 1.85mmol) was dissolved in 10 mL of dry THF. The solution was bubbled with nitrogen to expel dissolved oxygen, and dry triethylamine (0.3 mL, 2 mmol) was added with a syringe. A solution of ferrocenoyl chloride (0.47 g, 1.9 mmol) in 15 mL of dry THF was added dropwise to the stirred solution under nitrogen at 0 OC. The solution was then refluxed for 3 h in an oil bath. EtsNHCl was formed as a white precipitate in the THF solution. The precipitate was filtered and the remainingsolutionwas evaporated under vacuum. A light yellow crystalline powder was left in the flask. This powder was washed with cold ether (40 mL) twice, with aqueous NazCOs (5%) solution (50 mL) twice, and then thoroughly with deionized water. The fine yellow powder (0.79 g, 94 % yield) was identified as diethyl[4-(ferrocenecarboxamido)benzyl]phosphonate by 1H NMR. The diethyl group of this compound was then replaced by bis(trimethylsily1)because of the ease of subsequent hydrolysis under mild conditions following Rabinowitz's method." Diethyl[4-(ferrocenecarboxamido)benzyl]phosphonate(0.46 g, 1mmol) was dissolved in dry CHCls (20 mL) under nitrogen. Trimethylbromosilane (0.42 mL, 2.5 mmol) was added to the solution via a syringe. The solution was stirred for 8 h at room temperature. The solvent and excess BrSiMe:, were evaporated in vacuo, and the dark red powder remaining in the flask was identified as bis(trimethylsilyl) [4-(ferrocenecarboxamido) benzyl] phosphonate by lH NMR and mass spectrometry (M+ = 543.0). This silyl ester (0.54 g, 1mmol) was dissolved in 50 mL of methanol; (14) (a) Morita, T.; Okamoto, Y.;Sakurai, H. Bull. Chem. SOC. Jpn. 1978, 51, 2169. (b) Rabinowitz, R. J. Org. Chem. 1963, 28, 2975. (c) Degenhardt, C. R.; Burdsall, D. C. J. Org. Chem. 1986,51,3488.
2364 Langmuir, Vol. 7,No. 10,1991 the volume of the solution was decreased by evaporation to 10 mL, and the product was precipitated by addition of methanolic KOH solution. The mixture was filtered, and the precipitate was washed with methanol and dried under vacuum. This gave 0.42 g of the potassium salt of [4-(ferrocenecarboxamido)benzyl]phosphonic acid (yield 90%). lH NMR (Varian EM390) (DpO),6 2.77 (d, 2 H, -CHz-, J(CH2P) 20 Hz), 6 4.27 (8, 5 H, -CsHs-), 6 4.57 (8,2 H, -CH=CH-), 6 4.88 (8,2 H, -CHCON-), 6 7.3 (s, 4 H, phenyl). Materials. Methylene chloride and chloroformwere refluxed over PtOs and distilled freshly before use. Pentane was refluxed over CaClz and distilled. THF was distilled from CaHp. Ferrocenecarboxylicacid, diethyl(4-aminobenzyl)phosphonate,trimethylbromosilane, sodium hydrosulfide, triethyl phosphite, zirconyl chloride octahydrate, and 1,4-dibromobutane were used as received from Aldrich Chemical Co. 1,2-Ethanediylbis(phosphonic acid) (EDPA) was used as received from Alpha Chemicals. All other matsrials were reagent grade, obtained from commercial sources, and used without purification. Instrumentation. For the preparation of gold thin film electrodes on glass, a 8620Sputtering System (Materials Research Corp.) was used. The sputtering conditions were as follows: rf forward power, 150 W;rf peak-to-peak voltage, 1.8 kV, pressure, 1W2 Torr. Preparation of Au Electrodes a n d Zirconium Phosphonate Films. Gold was sputtered onto borosilicate glass slides (1in. x 3 in.) by using an aluminum mask with seven uniformly spaced circular holes (s/la in. diameter) and narrow slits for electrical contact. The slides were precleaned by immersingthem in a warm chromic-sulfuric acid solution for 2 h, rinsing with copious amounts of deionized water, and drying prior to transfer to the sputtering chamber. The glass was etched first with argon plasma for 30 s; an approximately 50 A thick layer of chromium was deposited, followed by a 1500-A layer of gold. Chromium underdeposition was used to increase adhesion between the gold and the glass surface. The area of the individual gold disk electrodes was typically 0.178 cml. In order to make electrical contact, the end of the gold slit was covered with silver paste and clamped with an alligator clip. The electrodes were cleaned prior to mono/multilayer deposition in hot 30% H202/98% H2SOI (volume ratio 1:5) for 10 s and were cycled electrochemically between 0.25 and +1.40 V vs SCE in 0.5 M HzSO, until the typical cyclic voltammogram of gold was obtained. All pretreatments were followed by rinsing with deionized water and drying in a stream of nitrogen. The dried and pretreated gold electrodes were soaked for 1 day in a 2 mM aqueous solution of (4-mercaptobuty1)phosphonic acid (a little methanol was added to increase the solubility). The electrode was washed with methanol first, then with water thoroughly and was immersed in a 5 mM aqueous solution of zirconium chloride octahydrate for 4 h. In order to grow ZEDP films, the electrodes were immersed in 10 mM aqueous EDPA for 5 h. ZEDP multilayers were prepared according to Scheme I, by sequential adsorption of Zr'+ and EDPA. Ferroceneterminated thin films were prepared by first adsorbing (4-mercaptobuty1)phosphonic acid and zirconium ion; the electrodes were then immersed in a 3-4 mM aqueous solution of the potassium salt of FcBP for 1day (Scheme11). The FcBP-modified electrodes were washed with methanol and then thoroughly with deionized water. Mixed ferrocene/benzylphosphonate layers were prepared by coadsorbing the two phosphonic acids on electrodespretreated with (4mercaptobutyl)phosphonicacid and zirconium ion, and, in some cases, with a ZEDP layer (Schemes I1 and 111). For mixed layers, a 5 mM total phosphonic acid solution was usually used. The mole fraction of FcBP in the solution relative to totalphosphonic acid (denoted Xp,) was varied between 0.1 and 1.0. Electrodes were removed from these adsorption solutions after 1day to a few days and were washed sequentially methanol and water. The amounts of electroactive material adsorbed were not significantly changed for samples soaked for different periods of time. All of the metal phosphonate layer deposition steps were performed under ambient conditions. Electrochemical Measurements. An EG&G/PAR 173POtentiostat and a 175programmer were used with a Kipp & Zonen recorder for cyclic voltammetry. For rapid scan BD91 X-Y-Y'
Hong and Mallouk voltammetric measurements, a NORLAND 3001/DMX (Norland Corp.) data acquisition system coupled to a NORLAND 3106R digital oscilloscope was used. A PAR 273 potentiostat coupled to an IBM PSI1 computer with Model 270 Electrochemical Analysis software were used for potential step chronoamperometry. Electrodes were mounted in a conventional three-electrode cell. The electrolyte solutionswere prepared with deionized water purified to a resistivity of 18.3 Mi2 cm with a Barnstead Nanopure I1 system (Sybron Corp.) and purged with argon before use. All potentials were measured and reported with respect to a SCE reference electrode. All of the electrodes which were modified with ZEDP multilayers and FcBP were used in the electrochemical measurements after ellipsometric measurements. Since the largest currents measured were on the order of 1 mA, uncompensated solution resistance effects in 1.0 M aqueous NaClO4 were assumed to be negligible. Ellipsometry. Ellipsometric measurements were made with a Rudolf 437 ellipsometer equipped with a RR-2000 rotating analyzer detector and 632.8-nm (He-Ne laser) analyzing light. The angle 0 between the incident beam and surface normal was 70'. The data were processed with a HewlettPackard Model 9816 desktop computer equipped with an H P 7470A plotter. All data were recorded in air without controlling the humidity. Au disk electrodes were immersed in hot HzO2/HBO4 (volume ratio 1:5) for 10 s to remove organics, washed with deionized water, and dried. The ellipsometric parameters, Q and A, for these precleaned Au substrates were measured as quickly as possible to avoid contamination. Ellipsometric parameters were remeasured following each EDPA adsorption step in the cycle shown in Scheme I. The thickness of ZEDP multilayer fiims was extracted from Q and A, using the previously determined optical constants of the gold films and an estimated real refractive index of 1.76 for the ZEDP multilayer. For thickness measurements of electrodes modified with FcBP according to Scheme 11,the ellipsometric parameters were measured at the precleaned, bare Au, measured again after adsorbing (4-mercaptobuty1)phophonic acid, and measured again after the adsorption of FcBP. For electrodes modified with FcBP according to Scheme I11 (i.e., with a ZEBP spacer), the ellipsometric parameters were measured for bare Au and after steps 1,3, and 4. For FcBP films, the total thickness was determined by using a real refractive index of 1.45, which is appropriate for organic films such as alkanethiola. Even though the presence of the ferrocene group may alter the refractive index somewhat, this measurement gives a qualitative indication of the relative thickness of the film.16
Results and Discussion Growthof ZEDP Multilayer Films. Scheme I shows the procedure used to prepare multilayers of ZEDP on electrode surfaces. We have previously reported the growth of similar films on electrodes using tri- and tetravalent metal ions and decanediylbis(phosphonic acid)." With t h e latter, well-ordered films are formed which appear to be structurally similar to the corresponding bulk-layered compounds MN(03PC~oH2oP0s) and M'"(HO3PCloH~P03). Interestingly, even with phosphonic acids as small as H203PCzHdP03Hs (EDPA), layered solids can be formed by reaction with metal ions.16 As this is the smallest bis(phosphonic acid) available to us for use as an electron tunneling spacer, we attempted to grow thin films with it according to Scheme I. The ellipsometric parameters \k and A were measured after each complete adsorption cycle in Scheme I. The value is in general not sensitive to the growth of the ZEDP layers, while A increases by ca. lo per layer in an almost linear fashion. The best fit to the data was obtained by using a real refractive index for the film of 1.76, and i t gave 67 f 5 A as the thickness of a seven-layer film. The error in A for t h e bare gold electrode gives an almost constant error in thickness of 1 A, which is very significant (15)Chidaey, C. E. D.; Loiacono, D. N.Langmuir 1990, 6,682.
Langmuir, Vol. 7, No.10,1991 2365
Measurement of Electron Transfer Rates
a
3v vs. SCE
Number of ZEDP Layers
Figure 1. Plot of the total thickness of the ZEDP multilayer
film, calculatedfrom ellipsometricmeasurements, vs number of ZEDP layers. "Zero"layers refers to an MBPA layer on gold.
for the zeroth layer but only about 1.5% of the total for the seven-layer measurement. The total thickness of the ZEDP multilayer is not sensitive to small changes (10.04) in refractive index used, and outside of this range the fit becomes considerably worse. Figure 1 shows the film thickness calculated from @ and A after adsorption of each ZEDP layer. The plot is linear, and from the slope an average layer thickness of 8.0 A is calculated. This is in good agreement with the layer distance, 7.0 A, found by X-ray powder diffraction for bulk Zr(OsPCzH903). The intercept in Figure 1(ca. 9 A) provides a measure of the thickness of the MBPA anchoring layer. The linear increase in thickness with number of layers, high refractive index, and close correspondence of the repeat distance to the bulk layer spacing, collectivelyindicate that a compact layered film is growing with the stacking axis normal to the electrode surface. It is interesting that such films can be grown with only a two-carbon spacer, since a chain length of approximately nine carbon atoms is needed to produce an ordered n-alkanethiol monolayer.' With two carbon atoms, alkyl chain self-assemblyforces can be expected to be minimal. This result underscores the structure-directing character of the highly polar metal-oxygen-phosphorus network, which is the same in the structuresls of a-Zr(O3POH)2-H20and MW(03PR)z (R = alkyl, aryl), and only slightly different in M1I(HO3PR)2(M = Ca, Cd), independent of the nature of the R group provided that its cross-sectional area is 124 A2. Electrochemistry of Ferro/ferricyanide. Figure 2 showscyclicvoltammetry of clean gold and ZEDP modified electrodes in aqueous solutions containing equimolar (1 mM) Fe(CN)& and Fe(CN)sS-. Ferro/ferricyanide was selected as an electrochemical probe because it is an electrochemicallyreversible,one-electron outer-sphere redox c0up1e.l~At the bare gold electrode the shapes of the i-E curves and the 60-mV separation between anodic and cathodic peak potentials are indicative of a diffusionlimited or electrochemically reversible one-electron process.'* In Figure 2b, it can be seen that the presence of a 9-A MBPA layer causes the separation of the peak potentials to increase to ca. 300 mV. This can be ascribed (16) (a) Dines,M. B.; DiGiacomo, P. Inorg. Chem. 1981,20,92. (b) Diner, M. B.; DiGiacomo, P.; Callahan, K. P.; Griffith, P. C.; Lane, R.; Cooksey, R. E. In Chemically Modified Surfaces in Catalysis and Electrocatalyst; ACS Sympoeium Series 192; Miller, J., Ed.;American Chemical Society: Washington, DC, 1982;p 223. (c) Dines, M. B.; Cookeey, R. E.; Griffith, P. C. Inorg. Chem. 1985,22,1003. (d) Clearfield, A,; Smith, G. D. Inorg. Chem. l%9,8,431. (e) Cao, G.;Lynch, V. M.; Swinnea, J. S.;Mallouk, T. E. Inorg. Chem. ISSO, 29, 2112. (17) (a) Conway, B. E.; Currie, J. C. J. Electrochem. Soc. 1978, 125, 257. (b) Weber, J.; Samec, 2.;Marecek, V. J. Electroanal. Chem. Interfacial Electrochem. 1978,89, 271. (18) Bard,A. J.; Fnulkner, L. R. Electrochemical Methods: Fundamentals and Applicatione; Wiley, New York, 1980.
W
Figure 2. Cyclic voltammograms of modified gold electrodes in
1 mM Fe(CN)&, 1 mM Fe(CN)a*, 0.1 M NaClO4: (a) bare Au, scan rate 50 mV/s; current S = 56 rA/cmz; (b) Au with MBPA, 20 mV/s; S = 28 pA/cm2; (c) Au with MBPA 1 layer of ZEDP, 20 mV/s, S = 11 pA/cm2; (d) Au with MBPA/2 layers of ZEDP, 20 mV/s, S = 11 pA/cm2.
. .: -
I
_._I
,
*
,
,
,
,
,
,
,
2366 Langmuir, Vol. 7,No. 10,1991
Hong and Mallouk
Table I. Linear Regression Analysis of Data in Figure 3
*
log k ( a m ) -3.2 -5.2 -6.3 -7.6
electrodea MBPA 1 ZEDP 2 ZEDP 3 ZEDP
(1 - a)' 0.61 0.59 0.48 0.37
Scheme 11. Au Electrode Modified with FcBP/ Benzylphosphonic Acid
a Final adsorption state of the gold electrode. b k(app) = apparent rate constant (cm/s) for oxidation of Fe(CN)e4-. The electron transfer symmetry factor (1 - a) calculated from the Tafel slopes.
ZEDP film thickness(.&)
Figure 4. Plot of log (kapp)vs the total thickness of the ZEDP multilayer film for the oxidation of ferrocyanide in 0.1 M Nac10,.
slope decreases smoothly with increasing film thickness, giving transfer coefficients of 0.61 with a single MBPA layer and 0.37with four ZEDP layers. From Marcus theory for heterogeneous electron transfer reactions, it is expected that as the driving force of the oxidation increases, the Tafel slope should approach zero according to eq 2lS2l (Y
= 0.5 f nFv/2X
(2)
where n is the number of electrons transferred, F is Faraday's constant, 7 is the electrode overpotential, and X is the reorganization energy of the redox couple. This systematic decrease of the transfer coefficient was demonstrated over a wide range of anodic overpotentials at mercury electrodes by Weaver and Tyma.21 Having extracted the kapp values from the Tafel plots, they can be plotted against film thickness to give a value for the tunneling constant 0. If tunneling across the film is the dominant mechanism of electron transfer, then the oxidation current a t any potential should decrease exponentially with distance according to eq 3,where io is the
i = io exp(-pd)
(3)
current observed at a bare electrode and d distance of closest approach of the diffusing redox molecule, Le., the film thickness. Likewise, a plot of In (kapp)vs d should give a straight line with a slope of -0.Such a plot is shown in Figure 4. It is interesting that a linear relation is found for 0-3 ZEDP layers and that the intercept obtained is within experimental error, the same as the reported rate constant for Fe(CN)s4- oxidation a t a clean gold electrode in this medium.22 The value of the tunneling constant we obtain, however, from the plot is 0.43 A-l, which is quite significantly smaller than values (0.5-1.4 A-*) measured (19) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985,811,265. (20) Gratzel, M. Heterogeneous Photochemical Electron Transfer; CRC Press: Boca Raton, FL, 1989; Chapter 1, p 36. (21) Weaver, M. J.; Tyma, P. D. J . EZectroanal. Chem. Interfacial Electrochem. 1980, 111, 195. (22) Mareak,V.;Samec,Z.;Weber, J. J.Electroanal. Chem.Znter/acial Electrochem. 1978, 94, 169.
for electron tunneling through saturated spacers in a variety of experiments.12~26~27 It should be noted that the presence of defects or pinholes in the layer could be entirely responsible for the measured current a t high overpotential, because of the strong dependence of the electron transfer rate on the thickness of the ZEDP film. If the multilayer film has pinholes exposing the bare electrode surface or collapse sites1 in the film, then electron transfer will take place by diffusion of electroactive molecules to these sites. Fortunately, we can assess the effect of pinhole defects by using the model proposed by Amatore et al.23for the redox kinetics of partially blocked electrodes. According to this model, if the ZEDP-modified electrode has defects which are separated by distances greater than the characteristic diffusion length of the experiment (a few micrometers for the cyclic voltammetry used here), sigmoidal i-E waves (from an array of microelectrode^^^^^^) should be observed near E O ' of the redox couple. Our voltammetry does not show this behavior. If the electrode is covered by defects separated by smaller distances, however, one still might get a linear log kapp vs d curve, provided the film thickness at the defect sites grows linearly with the number of layers. In this case the average film thickness, as measured by ellipsometry, would be significantly greater than the thickness at defect sites such as multilayer and substrate grain boundaries, where electron transfer would occur most readily. Ferrocene-ContainingMonolayers. The problem of film defects may be overcome, at least in principle, by using a covalently bound, rather than freely diffusing, redox probe molecule. Scheme I1shows the procedure used to prepare monolayers containing a covalently bound ferrocene group a t the outer surface of the film, and Scheme I11 shows a modification of the procedure where an additional ZEDP layer is inserted between the ferrocenecontaining monolayer and the electrode. (23) Amatore, C.; Saveant, J. M.; Teesier, D. J.%lectroanol. Chem. Inter/acial Electrochem. 1983,147, 39. (24) Ikeda, T.; Schmehl, R.; Denisevich, P.; Willman, K.; Murray, R. W. J. Am. Chem. SOC.1982,104, 2683. (25) (a) Wightman, R. M. Anal. Chem. 1981,53,1127A. (b) Dayton, M. A.; Brown, J. C.; Stutts, K.J.; Wightman, R. M. Anol. Chem. 1980, 52, 946. (26) (a) Close, G. L.; Calcaterra, L. T.; Green,N. J.; Penfield, K. W.; Miller, J. R. J. Phys. Chem. 1986,90,3673. (b) Beratan, D. N.J. Am. Chem. SOC.1986,108,4321. (c) Oevering, H.; Paddon-Row, M. N.; H e g pener, M.; Oliver, A. M.; Cotaaris, E.; Verhoeven, J. W.; Hush, N. S. J. Am. Chem. Soc. 1987,109,3258. (d) Paddon-Row, M. N.; Oliver, A.M.; Warman, J. M.; Smit, K.J.; De Haas, M. P.; Oevering, H.; Verhoeven, J. W. J. Phys. Chem. 1988, 92, 6958. (e) Kuhn, H. Pure Appl. Chem. 1979,51, 341. (0Mobius, D. Ber. Bunsen-Ges. Phys. Chem. 1978,82, 848. (27) Polymeropoulos, E. E.; Sagiv, J. J. Chem. Phys. 1978,69, 1836.
Langmuir, Vol. 7, No. 10,1991 2367
Table 11. Thickness of Sequentially Deposited Layers on Au as Measured by Ellipsometry layering sequence thickness,A MBPA 9h1 MBPA-ZFEDPA 19*2 MBPA-ZFFCBP 26 f 2 MBPA-ZFEDP A-ZPFCBP 34 f 4
The stepwise growth of these films can again be followed by ellipsometry. Adsorption of the MBPA, ZEDP, and FcBP layers causes a stepwise increase in A and little change in q, as with the pure ZEDP layers. In the case of FcBP films the best fits to the ellipsometric data were obtained with values of 1.45-1.50 for the real part of the refractive index. Changes in thickness following the sequential deposition of layers are shown in Table 11. The thickness of the MBPA-Zr-FcBP layer is 26 f 2 A, and the thickness of the MBPA-Zr-EDPA-Zr-FcBP layer is 34 f 4 A. The 8-A difference is as expected for a single ZEDP layer. Estimated errors in these thicknesses arise mainly from the measurements of A and the blank (zeroth layer) measurements. While the thickness measurements are not exact, they are in semiquantitative agreement with the expected thicknesses. The FcBP molecule in an extended configuration is about 17.5 A long, and CPK models show that these molecules are relatively rigid; assuming that the P-C bond is roughly normal to the layer plane, these molecules should tilt so that their height in the monolayer is about 13.5-15.5A. Hence, a 26-AMBPAZr-FcBP layer is quite reasonable allowing about 9 A for MBPA and 2 A for the Zr layer. Also, since our purpose is to find the value of the tunneling constant 8, the absolute thicknesses are of less importance than their difference (see eq l),which is simply the thickness of the ZEDP layer. Figure 5 shows typical cyclic voltammograms of Au electrodes modified with MBPA-Zr-FcBP in 1M aqueous NaC104. In Figure 5a the film was grown from a solution containing only FcBP and no benzylphosphonate. For these films, the oxidation and reduction peaks are quite symmetric, and the separation of peak potentials was 1& 30 mV, for several electrodes, independent of scan rate between 5 and 100mV/s. The charge passed in the anodic and cathodic waves indicates coverages of electroactive material in the range (2-4) X 10-lo mol/cm2, or (1.2-2.4) X 10“ molecules/cm2,also independent of scan rate. These data indicate that electron transfer between the electrode and the ferrocene groups is reversible a t these scan rates28 and that most or all of the ferrocene is electrochemically accessible. From crystallographic datamthe diameter of ferrocene is 6.6 A, and the surface density a t maximum coverage13 should therefore be about 2.7 X 1014/cm2. However, in (28) Murray, R. W. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker: New York, 19EM; Vol. 13, pp 191-368. (29) Seiler, P.; Dunitz, J. D. Acta Crystallogr. 1979, B35, 1068.
1
0.7
\,vI
0.5
,
0.3
J
0.1
I
0.7 V vs. SCE
1
0.5
0.3
0.1
Figure 5. Cyclic voltammogramsof modified gold electrodes in 1 M NaClOd: (a) Au/MBPA/FcBP (xpC = 1.0);(b)Au/MBPA/ FcBP ( X F ~= 0.1) and benzylphosphonate. Scan rates were 10, 20,50,100,and 200 mV/s.
the zirconium phosphonate structure, each Zr or phosphonate group occupies an area of 24 A2, so we would expect a maximum coverage of 4.1 X 1014/cm2if every available site were occupied by a FcBP molecule. Clearly, it is not possible to fit one molecule per site, and the observedcoverage shows that only about half the available Zr sites are occupied, making a fairly dense monolayer of FcBP molecules. In this situation we would expect a highly disordered layer. In order to overcome this problem, most of the FcBP molecules were replaced by electroinactive benzylphosphonate spacer molecules. Figure 5b shows cyclic voltammetry for an electrode prepared in a solution where X F ~ was 0.1. Again the electrochemistry is reversible a t the scan rates shown, although the peak width at halfmaximum is 120-150 mV, significantly larger than the Nernstian value of 90 mV. This extra width (also found in the X F =~ 1.0 samples) signals either an interaction between the ferrocene groups or a heterogeneous distribution of electroactive groups in the film. Integration of charge (correctingfor background current) gives a coverage of 3.6 X 1013/cm2, which is about 13% of monolayer coverage for ferrocene groups. The benzylphosphonate groups are small enough to occupy the remaining 3.7 X 1014available sites/cm2. Interestingly, attempta to exchange out more of the ferrocene-containing groups by soaking these electrodes for several days in benzylphosphonic acid solutions led to no measurable loss, within experimental error, of electroactivity. This implies that covalent binding of FcBP to Zr at X F =~ 0.1 is quite irreversible, even at defect sites. Electrodes prepared according to Scheme I11 (with one ZEDP spacer layer) are similar to those prepared according to Scheme 11,in that the ratio of anodic and cathodic peak currents is maintained as unity and peak currents are proportional to scan rate at slow scan rates. However, as the scan rate is increased from 5 to 100 mV/s, these electrodes begin to show quasi-reversible behavior, and the peak potential separation increases from 30 to 100 mV. At higher scan rates both seta of electrodes show increasing peak potential splitting, and the splitting may
Hong and Mallouk
2368 Langmuir, Vol. 7, No. 10,1991
Table 111. Dependence of Rate Constants and Transfer Coefficients on ZEDP Spacers and Xp,
0.8
5'
8
spacer layep without ZEDP with ZEDP
0.6
g 3
B
5.
0.4
w" 0.2 -1
0
1
= 0.1' anodicb d 54f3 46 f 2 58+ 2 44 f 1 e 0.23f 0.01 0.23 f 0.03 0.60+ 0.0 0.43f 0.01 d 2.5f 0.1 2.2 f 0.2 1.5 f 0.1 e 0.61 f 0.06 0.37 f 0.05 0.77 f 0.03 XF,
cathodicb
1.w anodicb
XFc
cathodicb
Spacer layer between the Au surface and FcBP. Electrode reaction. FcBP mole fraction. d The heterogeneouselectron transfer rate constant (8-l). e The electron transfer symmetry factor (cathodic, a;anodic, 1 - a).
2
Log V (Voltlsec)
Figure 6. Plot of anodic (E ) and cathodic (Ep)peak potentials vs log (scan rate) (V/s) for f u electrodes modified with MBPA/ FcBP (xpC = 1.0) in 1 M NaC104.
-1' 0.0
"
0.2
"
0.4
"
0.6
"
0.8
Time(sec) 0.3
'
-1
0 1 Log V ( Volt/Sec )
Figure 8. Plot of log (anodiccurrent) vs time for an Au/MBPA/ FcBP electrode ( X F ~= 0.2). The applied potential was stepped from E = 0.2 V vs SCE to E = 0.5 V (which is near the FcBP formal potential) in 1 M NaClO4.
2
Figure 7. Plot of anodic (E and cathodic (Epc)peak potentials vs log (scan rate) (V/s) for u electrodes modified with MBPA/ FcBP (xpC = 0.1) in 1 M NaC104.
2
be wed to extract kinetic data (transfer coefficients and rate constants).goAccordingtoLaviron's procedureFlthese quantities may be obtained in a straightforward manner from eqs 4 and 5 under totally irreversible conditions, i.e.
Epa= Eao'- (RT/a!nF)In (RTk"/anFu,) E, = E,"'
- ( R T / ( l - a!)nF)In ( R T k o / ( l- a)nFv,)
(4) (5)
at scan rates where the peak potential separation is 1200 mV. U p and uc are critical scan rates which are obtained from extrapolating the linear portion of the E, vs log u plots to the formal cathodic and anodic potentials E,"' and E,,"'. The latter are the cathodic and anodic peak potentials observed under reversible conditions (i.e,, at slow scan rate). The slopes of the linear portions of the E, vs log u curves are -2.3RTlanF for the cathodic branch and 2.3RT/(1- a)nF for the anodic branch. The heterogeneous rate constant k" is given by anFu,/RT for the anodic process and ( 1- a!)nFu,/RTforthe cathodicprocess. Figure 6 shows a plot of EW and E, vs log (scan rate) for a XF, = 1.0 electrode prepared according to Scheme 11. From the slopes of the two branches, the transfer coefficients a! and 1 - a were evaluated as 0.23 and 0.24. These values differ considerably from 0.5, and their sum is not unity; similarly low values appear in the literature for surface bound ferrocene.32 These apparently low transfer coefficientsare consistent with a highly defective film structure, which was inferred above from packing models. Finklea and co-workerslb have shown that low transfer coefficients are obtained when film defects carry (30) (a) Sharp, M.;Petemon, M.; Edstrom, K. J. Electroanal. Chem. Interfacial Electrochem. 1979,96, 123. (b) Sharp,M.;Petemon, M.; Edstrom, K. J. J.Electroanal. Chem.InterfacialElectrochem. 1980,109,
271.
(31) Laviron, E.J.Electroanal. Chem. Interfacial Electrochem. 1979, 101,19. (32)Lane, R. F.;Hubbard, A. T. J. Phys. Chem. 1973, 77, 1401.
most of the current at high overpotential. In this case the most reasonable mechanism would involve charge transfer diffusion between ferrocene sites, allowing most of the current to be collected at "collapse" sites where a ferrocene group is relatively close to the gold surface. The k o values found for the cathodic and anodic processes 154 and 46 s-l, respectively) are in reasonable agreement for XF, = 1.0 electrodes and did not vary greatly from electrode to electrode. The Xp, = 0.1 electrodesgave plots like that shown in Figure 7, from which we obtain a and 1 - a values of 0.60 and 0.43, respectively. The increase in the transfer coefficients, relative to the Xp, = 1.0 electrodes, and a s u m near unity are consistent with a reduction in lateral charge transfer diffusion; as mentioned above, lateral charge transfer will carry current to defect sites in competition with direct electron tunneling across the monolayer. Surprisingly, the k" values measured at these electrodes are essentially the same as those found at XF,= 1.0 (Table 111). The low values of the transfer coefficients obtained for XF,= 1.0 electrodes suggest that in this case the system does not fulfill the assumptions of the model, and therefore we are inclined to disregard the measured k" values in this case. Potential step chronoamperometry (Figure 8)was used to assess the kinetic heterogeneity of these films at Xp, = 0.2. In principle, a film with all ferrocene units the same distance from the electrode should give a singleexponential current decay following an abrupt potential step, as demonstrated by C h i d ~ e y .In ~ ~the present case the current does not decay via a single exponential in the time regime where faradaic processes are dominant, indicating a distribution of ferrocene sites. Therefore, the measured k" for these electrodes should be regarded as an average value, for a distribution of electroactivesites. Adding an 8-A ZEDP spacer layer, as shown in Scheme 111, significantly slows the rate of electron transfer for both the XF,= 1.0 and XF,= 0.1 electrodes. Figures 9 and 10 show peak potential vs scan rate plots for these electrodes, and kinetic data extracted from these plots are (33)Chidsey, C. E.D.; Science, in press.
Langmuir, Vol. 7, No. 10, 1991 2369
Measurement of Electron Transfer Rates
7
O**
0.2 4
-2
-1 Log V
0 (
I
1
Volt I Sec )
2
Figure 9. Plot of anodic ( E and cathodic (+) peak potentials v8 log (scan rate) (V/s) for u electrodes modified with MBPA/ ZEDP/FcBP (xpC = 1.0) in 1 M NaClO,. 0.5
.
Epe
5: g
0.4
w" ".J
-2
-1
0
1
Log V (Voltlsec)
Figure 10. Plot of cathodic (Ep) peak potentials vs log (scan rate) (V/s) for Au electrodes modified with MBPA/ZEDP/FcBP (xpC = 0.1) in 1 M NaClO,.
summarized in Table 111. For the X F =~ 0.1 electrodes, the anodic wave was not sufficiently well-resolved from anodic background current to give a plot on the anodic branch. We find again the ko is relatively independent (within a factor of 2 or so) O f XFCfor electrodes grown with a ZEDP spacer layer. Using just the cathodic data for the X F =~0.1 electrodes with and without ZEDP spacers, we can make get a rough estimate of the tunneling constant j3,which from eq 1is 2.3[log (58/1.5)]/8A = 0.46 A-l. This value is surprisingly low, in light of a considerable body
of literature123*27@which finds higher values (0.51.4 A-1) for intermolecular electron transfer through a variety of saturated spacers. In the present case, we cannot discount the possibility that the heterogeneity of electroactive sites is a source of error, and we are currently synthesizing other electroactive probe molecules which are expected to make more highly ordered monolayers. Conclusions This study has shown that multilayer metal phosphonate films can be grown on gold electrodes by sequential adsorption reactions of Zr'+ and an alkanediylphosphonic acid, even when the latter contains only two carbon atoms. The regular stepwise growth of these layers is rather remarkable, because it shows that alkane packing forces play a minimal role in the formation of these well-ordered layered films. Multilayer structurescan be grown by using these ZEDP spacer layers and ferrocene-containing monolayers, although the ferrocene layers appear to be disordered and heterogeneous in terms of the kinetics of electron transfer to the gold electrode. The electrochemical techniques used to characterize these f i i give an estimate of the electron tunneling constant (3. The value obtained for j3 is uncharacteristically low for a saturated spacer, and a systematic error in ita measurement may arise from the disorder of the ferrocene-containing layers. Studies involving other covalently bound redox-active probe molecules, which may resolve these issues, are currently in progress. Acknowledgment. This work was supported by grants from the National Institutes of Health (lROlGM43844Ol), the National Science Foundation (PYI Award CHE8657729), and the Robert A. Welch Foundation. T.E.M. also thanks the Camille and Henry Dreyfus Foundation for support in the form of a Teacher-Scholar Award. Registry NO. EDPA, 6145-31-9; ZEDP, 75406-94-9; MBPA, 135865-74-6; Au, 7440-57-5;ZrC120, 7699-43-6; NaClO,, 760189-0; ZrCb, 10026-11-6; ferrocyanide, 13408-63-4; potaeeium 4-(ferrocenecarboxamido)benzylphosphonate,135865-75-7;ferricyanide, 13408-62-3. (34)Auweraer,M.V.; Verschuere,B.; Biasmans, G.;De Schryver, F. C.; Willig, F. Langmuir 1987, 3, 992.