Langmuir 2005, 21, 6855-6865

6855

Characterization of the Deposition of Organic Molecules at the Surface of Gold by the Electrochemical Reduction of Aryldiazonium Cations Alexis Laforgue, Tania Addou, and Daniel Be´langer* De´ partement de Chimie, Universite´ du Que´ bec a` Montre´ al, Case Postale 8888, succursale Centre-Ville, Montre´ al, Que´ bec, Canada H3C 3P8 Received October 26, 2004. In Final Form: May 5, 2005 The deposition of 4-X phenyl groups (X ) NO2, COOH, N-(C2H5)2) on polycrystalline gold electrode was achieved by the electrochemical reduction of the corresponding 4-substituted phenyldiazonium tetrafluoroborate salts in anhydrous acetonitrile media. The electrochemical quartz crystal microbalance measurements evidenced a two-step deposition process: the first one is the deposition of close to a monolayer and the second one is the relatively slower growth of multilayers. In this second region, the deposition is less efficient than for the first one. The electrochemical behavior of the resulting modified gold electrode was investigated in the presence of an electroactive redox probe and these results, together with the electrochemical quartz crystal microbalance data, demonstrated significant differences in reactivity and in deposition efficiency between the diazonium salts. The characterization of the modified electrodes by cyclic voltammetry and electrochemical impedance spectroscopy, as well as X-ray photoelectron spectroscopy measurements, showed that the formation of multilayers is possible and that a significant fraction of the deposited material remained at the electrode surface, even following ultrasonic treatment. The X-ray photoelectron spectroscopy data indicate that the existence of Au-C and Au-NdN-C linkages (where C represents a carbon atom of the phenyl group) is uncertain. Nonetheless, the deposition of the aryl groups by electrochemical reduction of diazonium cations yielded a film that adheres well to the gold surface and the deposited organic film hindered gold oxides formation in acidic medium.

Introduction The derivatization of surfaces is often required to improve the performance of many materials and change their surface properties and is also often critical for some particular applications. The electrochemical grafting of specific organic molecules has become a method of choice since the deposition conditions can be easily controlled and adapted to the substrate.1-2 Moreover, in appropriate experimental conditions, they could allow the grafting of homogeneous monolayers which are rather difficult to achieve by most chemical methods. A novel technique based on the electrochemical and chemical reduction of diazonium salts has been developped in the past decade.2-16 It allowed for the covalent bonding of organic * Author to whom correspondence should be addressed. E-mail: [email protected]. Tel: (514) 987-3000 (ext. 3909). Fax: (514) 987-4054. (1) Ulman, A. Chem. Rev. 1996, 96, 1533. (2) Downard, A. J. Electroanalysis 2000, 12, 1085. (3) (a) Delamar, M.; Hitmi, R.; Pinson, J.; Save´ant, J. M. J. Am. Chem. Soc. 1992, 114, 5883. (b) Allongue, P.; Delamar, M.; Desbat, B.; Fagebaume, O.; Hitmi, R.; Pinson, J.; Save´ant, J. M. J. Am. Chem. Soc. 1997, 119, 201. (4) (a) Liu, Y.-C.; McCreery, R. L. J. Am. Chem. Soc. 1995, 117, 11254. (b) Yang, H.-H.; McCreery, R. L. Anal. Chem. 1999, 71, 4081. (c) Anariba, F.; DuVall, S. H.; McCreery, R. L. Anal. Chem. 2003, 75, 3837. (d) Solak, A. O.; Eichorst, L. R.; Clark, W. J.; McCreery, R. L. Anal. Chem. 2003, 75, 296. (e) Chen, P.; McCreery, R. L. Anal. Chem. 1996, 68, 3958. (5) Downard, A. J.; Prince, M. J. Langmuir 2001, 17, 5581. (6) (a) Kariuki, J. K.; McDermott, M. T. Langmuir 1999, 15, 6534. (b) Kariuki, J. K.; McDermott, M. T. Langmuir 2001, 17, 5947. (7) (a) Saby, C.; Ortiz, B.; Champagne, G. Y.; Be´langer, D. Langmuir 1997, 13, 6805. (b) Ortiz, B.; Saby, C.; Champagne, G. Y.; Be´langer, D. J. Electroanal. Chem. 1998, 455, 75. (c) D’Amours, M.; Be´langer, D. J. Phys. Chem. B 2003, 107, 4811. (d) Marwan, J.; Addou, T.; Be´langer Chem. Mater. 2005, 17, 2395. (8) Liu, S.; Tang, Z.; Shi, Z.; Niu, L.; Wang, E.; Dong, S. Langmuir 1999, 15, 7268. (9) Bahr, J. L.; Tour, J. M. Chem. Mater. 2001, 32, 3823. (10) Kuo, T.; McCreery, R. L.; Swain, G. M. Electrochem. Solid State Lett. 1999, 2, 288.

molecules through the formation of an aryl radical. This method has been used to modify carbons,2-10 silicium,11,12,15 metals,13-16 gallium arsenide,15 and organic materials17 such as PTFE and polyaniline. Scheme 1 describes the two-step process of this reaction which involves (1) the electrochemical reduction of the diazonium function and the formation of a phenyl radical and (2) the chemical grafting of the radical at the surface of the electrode with the formation of a covalent bond between a surface atom of the substrate and the phenyl group. The other most important technique to derivatize gold electrodes is the oxidative adsorption of thiols by a covalent (11) (a) Henry de Villeneuve, C.; Pinson, J.; Bernard, M. C.; Allongue, P. J. Phys. Chem. B 1997, 101, 2415. (b) Allongue, P.; Henry de Villeneuve, C.; Pinson, J.; Ozanam, F.; Chazalviel, J. N.; Wallart, X. Electrochim. Acta 1998, 43, 2791. (c) Allongue, P.; Henry de Villeneuve, C.; Cherouvier, G.; Corte`s, R.; Bernard, M.-C. J. Electroanal. Chem. 2003, 500, 161. (12) Hartig, P.; Dittrich, T.; Rappich, J. J. Electroanal. Chem. 2002, 524-525, 120. (13) Ahlberg, E.; Helge´e, B.; Parker, V. D. Acta Chem. Scand. B 1980, 34, 181. (14) (a) Adenier, A.; Bernard, M.-C.; Chehimi, M. M.; Cabet-Deliry, E.; Desbat, B.; Fagebaume, O.; Pinson, J.; Podvorica, F. J. Am. Chem. Soc. 2001, 123, 4541. (b) Chausse´, A.; Chehimi, M. M.; Karsi, N.; Pinson, J.; Podvorica, F.; Vautrin-Ul, C. Chem. Mater. 2002, 14, 392. (c) Adenier, A.; Cabet-Deliry, E.; Lalot, T.; Pinson, J.; Podvorica, F. Chem. Mater. 2002, 14, 4576. (d) Bernard, M.-C.; Chausse´, A.; Cabet-Deliry, E.; Chehimi, M. M.; Pinson, J.; Podvorica, F.; Vautrin-Ul, C. Chem. Mater. 2003, 15, 3450. (e) Bouderma, K.; Chehimi, M. M.; Pinson, J.; Blomfield, C. Langmuir 2003, 19, 6333. (f) Combellas, C.; Kanoufi, F.; Pinson, J.; Podvorica, F. I. Langmuir 2004, 20, 280. (g) Pinson, J.; Podvorica, F. Chem. Soc. Rev. 2005, 34, 429. (15) Stewart, M. P.; Maya, F.; Kosynkin, D. V.; Dirk, S. M.; Stapleton, J. J.; McGuiness, C. L.; Allara, D. L.; Tour, J. M. J. Am. Chem. Soc. 2004, 126, 370. (16) Hurley, B. L.; McCreery, R. L. J. Electrochem. Soc. 2004, 151, B252. (17) (a) Combellas, C.; Kanoufi, F.; Mazouzi, D.; Thie´bault, A.; Bertrand, P.; Me´dard, N. Polymer 2002, 44, 19. (b) Liu, G.; Freund, M. S. Chem. Mater. 1996, 8, 1164.

10.1021/la047369c CCC: $30.25 © 2005 American Chemical Society Published on Web 06/16/2005

6856

Langmuir, Vol. 21, No. 15, 2005

Laforgue et al.

Scheme 1. Two-step Process Deposition via the Reduction of a Diazonium Cation

S-Au bond, named self-assembled monolayers (SAMs).1,18-25 One drawback of this method is the unstability of the S-Au bond at reductive potentials which causes the desorption of the layer.20,24 Thus, it might be useful to develop a procedure that would yield a covalent bond between Au and another molecular species. Presumably, a C-Au bond would be more stable than a S-Au bond. More specifically, the grafting of aryl group by the diazonium route seems to yield robust layers that are difficult to desorb.14d Even if the grafting of aryl groups, from the electrochemical reduction of aryldiazonium salts, to a gold surface has been recently claimed,14d information about the electrochemical grafting process and the physicochemical properties of the deposited layer is limited. The purpose of this work was to study the deposition of substituted phenyl groups at the surface of polycristalline gold surface by the reduction of an aryl diazonium tetrafluoroborate salt: 4-nitrophenyldiazonium (NPDS), 4-carboxyphenyldiazonium (CPDS), and 4-diethylanilinediazonium (DEADAS). The electrochemical reduction and film deposition at a gold electrode was initially investigated by cyclic voltammetry. In addition, the electrochemical quartz crystal microbalance (EQCM) was also used, for the first time, to characterize the electrochemical deposition process. The resulting modified gold electrodes were characterized by X-ray photoelectron spectroscopy (XPS) and both cyclic voltammetry and electrochemical impedance spectroscopy in the presence of soluble electroactive species to evaluate their barrier properties for electron transfer. The resistance of the deposited layers to ultrasonic treatment was also investigated by XPS and electrochemistry. The nature of the chemical bonding between the gold surface and the aryl groups was also investigated by XPS. Finally, the ability of the deposited layer to prevent gold oxide formation under electrochemical cycling in acidic media was also evaluated. Experimental Section Reagents. Acetonitrile biotech grade (Aldrich) (H2O < 100 ppm) was used as received. All water used was Nanopure water (18 MΩ). Tetrabutylammonium tetrafluoroborate (NBu4BF4) (Aldrich) was recrystallized in anhydrous methanol/diethyl ether and dried at 80 °C under active vacuum for 15 h prior to use. Potassium ferricyanide, potassium ferrocyanide, and potassium chloride (Aldrich) were used as received. 4-Diazo-N,N-diethylaniline tetrafluoroborate (Aldrich) was used as received. The (18) Finklea, H. O. In Electroanalytical Chemistry; Bard, A. J., Rubinstein, I., Eds.; Dekker: New York, 1996; Vol. 19, pp 109-335. (19) Bain, C. D.; Whitesides, G. M. J. Am. Chem. Soc. 1989, 111, 7164. (20) Schneider, T. W.; Buttry, D. A. J. Am. Chem. Soc. 1993, 115, 12391. (21) Sabatani, E.; Cohen-Boulakia, J.; Bruening, M.; Rubinstein, I. Langmuir 1993, 9, 2974. (22) Janek, R. P.; Fawcett, W. R.; Ulman, A. Langmuir 1998, 14, 3011. (23) Boubour, E.; Lennox, R. B. J. Phys. Chem. B 2000, 104, 9004. (24) Rifai, S.; Laferrie`re, M.; Wayner, D. D. M.; Wilde, C. P.; Morin, M. J. Electroanal. Chem. 2002, 531, 111. (25) Liu, B.; Bard, A. J.; Mirkin, M. V.; Creager, S. E. J. Am. Chem. Soc. 2004, 126, 1485.

4-nitrophenyldiazonium tetrafluoroborate and 4-carboxyphenyldiazonium tetrafluoroborate salts were synthesized following a published procedure.7a The 1H NMR (DMSO) of 4-nitrophenyldiazonium tetrafluoroborate shows two doublets at 8.60 and 8.73 ppm and that of 4-carboxyphenyldiazonium tetrafluoroborate, two doublets at 8.55 and 8.97 ppm. The diazonium group was detected by infrared spectroscopy by a band at about 2290 cm-1. Electrode Preparation and Procedure. Polycrystalline gold electrodes (0.5 cm2) were cleaned using the following procedure before each experiment: 15 min of ultrasonic cleaning in 1 M HNO3 and 10 min in Nanopure H2O, 100 cycles at 1 V/s in 0.1 M NaOH, 20 cycles at 1 V/s in 0.1 M H2SO4, and 10 cycles at 100 mV/s in 0.1 M H2SO4. The electrodes were then ultrasonicated in acetonitrile for 60 min prior to modification. A onecompartment electrochemical cell was used with a three-electrode configuration. The reference electrode was a Ag/AgCl (saturated NaCl), and the platinum gauze counter electrode was flamecleaned before each experiment. The diazonium reduction experiments were carried out in a 5 mM diazonium salt/0.1 M NBu4BF4/ACN solution, which was deaerated by bubbling with ultrapure nitrogen for 10 min prior to each experiment and during the experiment. Following modification and prior to all electrochemical measurements, the electrodes were ultrasonicated in acetonitrile for 30 min. In the case of XPS measurements, the modified electrodes were ultrasonicated in acetonitrile for 60 min and dried under vacuum for 24 h prior to analysis. The barrier properties of the unmodified and the modified gold electrodes were evaluated in 5 mM Fe(CN)63-/5 mM Fe(CN)64-/ 0.1 M KCl, phosphate buffer, adjusted to pH 7 with 16 M NaOH. EQCM measurements were carried out with a 9 MHz AT-cut quartz cristal covered with gold sputtered on a Ti layer. The cleaning procedure was the same as that described above except that the ultrasonic treatment was not used to prevent the detachment of the gold layer. The geometric area of the quartz electrode was 0.22 cm2. Instrumentation and Procedure. Electrochemical experiments were performed using a potentiostat/galvanostat 263A (EG&G, Princeton Applied Research) or a multipotentiostat Solartron 1470 and a frequency response analyzer Solartron 1250 or 1255B controlled by a computer with the softwares Corrware and Zplot (Scribner Associates, version 2.6b). Electrochemical impedance experiments were performed between 65 kHz and 0.1 Hz using a signal amplitude of 10 mV. EQCM measurements were performed with a QCA922 (Seiko-EG&G, Princeton Applied Research). The “vertical cell” setup used is described elsewere.26 The counter and reference electrodes were the same as those described above. Considering that the deposited layer is rigid and that no viscoelastic changes occur at the electrode interface, the relationship between the frequency change and the mass change is given by the Sauerbrey equation:27,28

∆m ) -∆f × S

(1)

where ∆f and ∆m represent the frequency and mass changes, respectively, and S is a characteristic constant of the quartz crystal used (including the shear modulus, the quartz density and the piezoelectric active area). S is equal to 5.608 ng/cm2‚Hz in our case. Prior to the EQCM measurements, the surface area (26) Jerkiewicz, G.; Vatankhah, G.; Zolfaghari, A.; Lessard, J. Electrochem. Commun. 1999, 1, 419. (27) Buttry, D. A.; Ward, M. D. Chem. Rev. 1992, 92, 1355. (28) Qu, Q.; Morin, M. J. Electroanal. Chem. 2001, 517, 45.

Deposition of Organic Molecules at the Surface of Gold of the gold-coated quartz crystal was determined by the integration of the gold oxide reduction peak during a cyclic voltammetry at 10 mV/s in 0.1 M H2SO4.29 A calibration was performed with the lead underpotential deposition, known to yield a monolayer of Pb on the surface of gold.28,30 The roughness factor determined by this method was the same as the one found by integration of the gold oxide reduction peak when the anodic scan limit was 1.43 V vs Ag/AgCl. The roughness factor of the quartz crystal electrode was then determined by the “gold oxide” method before each experiment and ranged from 1.35 to 1.65. Subsequently, EQCM experiments were carried out to monitor the mass change of the electrode during the electrochemical reduction of the diazonium cation. The faradaic efficiency (moles of aryl deposited per mole of electron) can be determined for this process from the EQCM data and both the cyclic voltammetry and the chronoamperometry results. First, the integration of the voltammetric wave yields the charge (not background corrected) consumed to generate the aryl radicals and the fraction of these species that are deposited at the electrode surface is given by the mass increase over the same potential range. From the charge and the mass, the faradaic efficiency can be computed. Second, mass-charge plots can be generated from the corresponding chronoamperometric and mass-time curves. The slope of such curve yields the mass deposited per unit charge, and hence, the number of deposited molecule per electron can be computed. XPS was carried out with an Escalab 220i XL from VG equipped with a hemispherical analyzer and an Al anode (monochromatic KR X-rays at 1486.6 eV) used at 9 keV and 10-20 mA. The data were collected at room temperature, and the operating pressure in the analysis chamber was always below 1 × 10-9 Torr. The core level spectra were analyzed by using the peak-fitting module of Origin software (Origin Lab Corporation, version 7.0). The core level spectra were referenced to the Au 4f7/2 binding energy at 84 eV. The core level spectra were used to evaluate the atomic concentrations of the species present at the polycristalline gold electrode surface. The atomic concentrations (at. %) of each individual element were determined from the relative peak areas of the spectra and the corresponding sensitivity factors according to

at. % ) (Ai/si)/Σ(Ai/si)

(2)

where Ai is the area of the element i and si is the sensitivity factor for this element. For example, values of 1, 1.8, 2.93, and 9.79 were used for C 1s, N 1s, O 1s, and Au 4f7/2, respectively. The thickness of the layer can be estimated by the relative inhibition of the Au 4f7/2 signal, using

I/I0 ) exp(-d/λ sin θ)

(3)

where d is the layer thickness, λ the photoelectron escape depth, θ the takeoff angle, and I/I0 the ratio of the Au 4f7/2 peak intensities (modified surface/bare surface).19 In our experiments, the takeoff angle was 90° or 30° and λ was 42 Å, as reported for alkanethiols on Au (111) for the peak at 84 eV (Au 4f7/2).19

Results and Discussion Reduction of the Diazonium Cation Investigated by Cyclic Voltammetry and EQCM. Figures 1, SI 1, and SI 2 show the cyclic voltammograms and masspotential curves obtained from the EQCM measurements for three different diazonium cations: NPDS, DEADAS, and CPDS, respectively. The first cycles show the characteristic reduction peak of each diazonium function, which is observed at more- or less-negative potential depending on the electronic stabilization effect of the substituent of the diazonium.31,32 The reduction potentials are correlating precisely with the Hammett constant of the substituents (Figure SI 3). The second and third scans (29) Hoogvliet, J. C.; Dijksma, M.; Kamp, B.; Van Bennekom, W. P. Anal. Chem. 2000, 72, 2016. (30) Deakin, M. R.; Melroy, O. J. Electroanal. Chem. 1988, 239, 321. (31) Elofson, R. M.; Gadallah, F. F. J. Org. Chem. 1969, 34, 854. (32) Andrieux, C. P.; Pinson, J. J. Am. Chem. Soc. 2003, 125, 14801.

Langmuir, Vol. 21, No. 15, 2005 6857

Figure 1. Cyclic voltammograms and mass-potential curves of a gold electrode in 0.1 M NBu4BF4 + 5 mM NPDS/acetonitrile; Scan rate ) 5 mV/s.

show a drastic decrease in the cathodic peak intensity demonstrating that a blocking behavior of the gold electrode surface is achieved by the deposited species during the first scan. This observation is in agreement with previous reports dealing with the modification of carbon electrodes.2,3,7 The shape of the cyclic voltammogram for the first cycle is not that expected for a diffusioncontrolled process in that a rapid decay of the current is observed after the peak. This is due to the grafting of a blocking layer which severely inhibit further electron transfer. It should be noted that a less-pronounced decrease of the current in the presence of CPDS (see Figure SI 2) suggest a less efficient grafting (vide infra). The EQCM measurements provided compelling evidence for the reduction of the diazonium cation and the formation of a deposit at the gold electrode surface by a two-step process. The first one occurred within a potential range that is limited by the onset and the end of the reduction wave of the diazonium function. The mass increase during this initial step corresponds roughly to the deposition of a monolayer (see Tables 1 and SI 1).4a The second step occurs at more negative potentials after the reduction wave and is characterized by a slower increase of the deposited mass. This second deposition step corresponds to the completion of the first monolayer and the progressive growth of multilayers at the surface of the electrode as observed for carbon electrodes.4c,6 Following scan reversal, the mass increased slightly on the return scan of the first cycle, as well as during the two subsequent scans. During these subsequent cycles, the diazonium cations display different behaviors. A continuous mass increase is observed for NPDS, whereas the mass increase is much smaller for DEADAS and CPDS. It is worth noting that the mass increase occurred only when the potential is more negative than the onset of the reduction wave. Therefore, the mass change cannot be attributed to either a drift of the frequency of the quartz or the adsorption of solution species at the electrode surface. The continuous mass increase provides additional evidence for the formation of multilayers (Table SI 1). EQCM measurements were also performed during the potentiostatic reduction of the diazonium cations. Figure 2 presents the current-time and mass-time curves

6858

Langmuir, Vol. 21, No. 15, 2005

Laforgue et al.

Table 1. Electrochemical Impedance Spectroscopy, Cyclic Voltammetry and EQCM Data for a Gold Electrode before and after Deposition of a Thin Layera

Au Au/CP Au/DEA Au/NP

deposited massb (ng/cm2)

surface concentrationb (× 10-10 mol/cm2)

voltammetric chargec (µC/cm2)

faradaic efficiencyd (mol/e)

RCTe (Ω.cm2)

147 ( 6 197 ( 7 163 ( 5

12.2 ( 0.5 13.3 ( 0.5 13.4 ( 0.4

340 ( 20 920 ( 35 250 ( 20

0.35 ( 0.04 0.14 ( 0.01 0.52 ( 0.06

4.6 ( 0.3 9.8 ( 0.5 26 ( 1 20 ( 0.6

a Linear sweep between open circuit potential and 0.35 (NPDS), 0.2 (CPDS) and - 0.42 V (DEADAS). See Experimental Section for the composition of the solution. b By assuming a surface concentration of 10 × 10-10 mol/cm2 for a monolayer, the corresponding mass expected for a monolayer of CP, DEA, and NP is 121, 148, and 122 ng/cm2, respectively. c For the first reduction wave between the limits specified above in a. d Computed from the deposited mass and voltammetric charge (see Experimental Section). e Charge-transfer resistance evaluated from the semicircle of the Nyquist plot.

Figure 2. (a) Current-time and mass-time plots for a gold electrode in 0.1 M NBu4BF4 + 5 mM NPDS/acetonitrile; Eapplied ) -0.7 V vs Ag/AgCl; (b) Plot of mass change as a function of the charge consumed during deposition for the experiment of (a).

recorded during the reduction of NPDS at -0.7 V, while the corresponding data for DEADAS and CPDS are shown in Figures SI 4 and SI 5. The data from these potentiostatic experiments also indicate that the deposition seems to follow a two-step process. The initial stage for the deposition of 4-nitrophenyl (NP), which is characterized by a quick mass increase and decay of the current during the first 20 s represents the deposition of about four layers of aryl groups if the formation of an homogeneous and compact film is assumed. During the second step, a linear increase of the mass is seen during a 20 min deposition. In the case of DEADAS and CPDS, the first step led to the deposition of the equivalent of one or two monolayers and the second step is characterized by a slower increase of the deposited mass with time, in a manner similar to what

was observed for the deposition performed by cyclic voltammetry. The larger mass deposited for NP in comparison to the two others (4-carboxyphenyl, CP, and N,N-diethylaniline, DEA) might be explained by a difference in reactivity between NPDS and CPDS and by the larger overpotential used for NPDS (1.15 V) relative to DEADAS (0.35 V). The latter is confirmed by the smaller and similar mass (about 500 ng/cm2) deposited when the same lower overpotential was used for the electrochemical deposition of NP and DEA (Table SI 1a). The faradaic efficiency for this process can be determined from the electrochemistry and mass measurements by considering that the reduction a diazonium salt, and the deposition of the corresponding aryl group is a oneelectron process.32 First, the faradaic efficiency from the cyclic voltammetry data was 0.53, 0.36, and 0.14 for NP, CP, and DEA deposition, respectively. Second, the masscharge plots (see Figures 2b, SI 4b, and SI 5b) show three different regions that include at least two linear zones for low and high deposited masses (and charges) and a transition zone at intermediates deposited masses. The faradaic efficiency for the deposition of NP is relatively higher than that of DEA or CP and remained high for both low and high currents and also when a smaller overpotential is used for deposition (Tables 2 and SI 1). The deposition efficiency for NP is comparable to that reported (84%) for electrochemical grafting on a glassy carbon electrode.3b As mentioned above, the deposition of DEA and CP is much less efficient than that of NP. These results demonstrate that the deposition of the later aryl groups on the gold surface is less favorable and that the majority of the electrochemically generated radicals are not reacting with the electrode and, instead, are probably recombining together or reacting with molecules present in solution (diazonium salt and solvent).3b,33,34 This difference of deposition efficiency will be discussed below. For the remainder of the study, two types of organic layers were formed and characterized. First, thin films were deposited by a single potential sweep in the same conditions as in EQCM experiments, starting from open circuit potential and stopping at 0.35, 0.2, and -0.42 V for NPDS, CPDS, and DEADAS, respectively. Second, the deposition of thick films (or multilayers) was achieved by potentiostatic reduction at -0.7 V during 240 s. Barrier Effect of Thin Films. The blocking behavior of the modified gold surface was investigated by cyclic voltammetry in the presence of the ferri/ferrocyanide redox couple, and Figure 3a shows the cyclic voltamograms before and after modification of the gold electrode with a CP thin layer. The cyclic voltammogram is barely affected by the deposited layer, and the difference between the (33) Kozlovs’ka, Z. E.; Koval’chuk, E. P.; Obushak, M. D.; Rak, J.; Blazejowski, J. Electrochem. Commun. 2001, 3, 1. (34) Koval’chuk, E. P.; Kozlows’ka, Z. E.; Jozwiak, L.; Blazejowski, J. Pol. J. Chem. 2000, 74, 67.

Deposition of Organic Molecules at the Surface of Gold

Langmuir, Vol. 21, No. 15, 2005 6859

Table 2. Electrochemical Impedance Spectroscopy, Cyclic Voltammetry, and EQCM Data for a Gold Electrode before and after Deposition of a Thick Filma

Au Au/CP Au/CP Au/CP Au/DEA Au/DEA Au/DEA Au/NP Au/NP Au/NP Au/NPf Au/NPf

deposition time

deposited massb (ng/cm2)

surface concentrationb (× 10-10 mol/cm2)

faradaic efficiency (mol/e)

after 20 s after 240 s after 20 min after 20 s after 240 s after 20 min after 20 s after 240 s after 20 min after 20 s after 20 min

185 ( 7 300 ( 10 450 ( 20 220 ( 10 300 ( 15 480 ( 30 780 ( 25 1080 ( 80 1950 ( 120 190 ( 9 510 ( 15

15.3 ( 0.6 24.8 ( 0.8 37.2 ( 1.6 14.9 ( 0.7 20.3 ( 1.0 32.4 ( 2.0 63.9 ( 2.0 88.5 ( 6.5 159.8 ( 10.0 15.6 ( 0.7 41.8 ( 1.2

0.14 ( 0.01c 0.03 ( 0.01d 0.15 ( 0.01c 0.12 ( 0.01d 0.90 (0.03c 0.78 (0.02d 0.70 (0.02c 0.97 (0.03d

RCTe (Ω.cm2) 5 22 ( 1 162 ( 4 190 ( 5

a Potentiostatic deposition at -0.7 V (See Experimental Section for the composition of the solution). b By assuming a surface concentration of 10 × 10-10 mol/cm2 for a monolayer, the corresponding mass expected for a monolayer of CP, DEA, and NP is 121, 148, and 122 ng/cm2, respectively. c Computed from the deposited mass and chronoamperometric data for the deposition between 0 and 20 s. d Computed from the deposited mass and chronoamperometric data for the deposition between about 100 s and 20 min. e Charge-transfer resistance evaluated from the semicircle of the Nyquist plot. f Potentiostatic deposition with an overpotential of 300 mV with respect to the cathodic peak potential evaluated from the cyclic voltammetry data (see Figure SI 1).

Figure 3. Cyclic voltammograms at a scan rate of 10 mV/s (a) and Nyquist plot (b) for gold electrodes in a 5 mM Fe(CN)63-/4solution before and after deposition of a CP thin layer and ultrasonic (us) treatment for 60 s in water. The thin layer was deposited by linear sweep voltammetry between open circuit potential and 0.2 V.

anodic and cathodic peak potentials (∆Ep) remained constant, whereas the peak current decreases slightly (Table SI 1). A phenyl monolayer is not expected to decrease significantly the ferri/ferrocyanide electrontransfer rate.4e These results are in agreement with previous observations for thiophenyl SAMs on Au (111).21 Electrochemical impedance spectroscopy measurements

were performed to further characterize the modified surfaces. Figure 3b represents the Nyquist plots of a gold electrode before and after the modification by CP groups. The plot for the bare electrode is characterized by a semicircle at high frequency related to a RC equivalent circuit which corresponds to the combination of the chargetransfer resistance (RCT) with the double layer capacitance of the electrode and a low-frequency Warburg line at an angle of 45° representing the diffusion processes at the surface of the electrode.7a Following modification, the charge transfer resistance increased (Table 1) and remained almost unaffected by ultrasonication (see Table SI 1). The latter suggests that the thin layer is adhered strongly to the surface. The same experiments were carried out with NPDS and DEADAS, and the results are summarized in Tables 1 and SI 1. The slowing down of the electron-transfer kinetics is more evident with the NP and DEA layers, as indicated by the increase of both ∆Ep and RCT. As for the CP layer, the ultrasonication (in addition to the ultrasound treatment immediately following electrochemical grafting, see Experimental Section) does not affect the barrier effect of the NP and DEA layers, suggesting that the organic moieties are strongly attached to the surface, probably by covalent bonding (vide infra), as assumed in the second step of the reaction process. This is consistent with previous reports that have demonstrated that the covalent grafting of aryl groups can occur on a variety of noble and non-noble metals.13-16 Barrier Effect of Thick Films and Their Stability. The barrier properties of the thick films generated by potentiostatic deposition were also investigated. Figure 4 shows the cyclic voltamograms together with Nyquist and Bode plots of a gold electrode in a ferri/ferrocyanide solution before and after modification with various substituted phenyl groups during 240 s at -0.7 V. Following deposition, the electrodes were rinsed for 10 min in acetonitrile. Clearly, the barrier properties of the CP film differs significantly from those of the NP and DEA films. The CV, as well as the Nyquist and Bode, plots are slightly affected by the presence of the CP film, whereas a significant decrease of the peak current and increase of ∆Ep and RCT (see also Table SI 2) and change of the Bode plot are observed for the DEA- and NP-modified electrodes. It should be noted that these deposited aryl layers are not defect-free and do not display blocking properties such as those typically found for n-alkanethiol

6860

Langmuir, Vol. 21, No. 15, 2005

Laforgue et al.

Figure 5. XPS survey spectra of a gold electrode before and following modification by a thin layer of aryl groups. The thin layer were deposited by linear sweep voltammetry between open circuit potential and 0.35 (NPDS), 0.2 (CPDS), and -0.42 V (DEADAS).

Figure 4. Cyclic voltammograms at a scan rate of 10 mV/s (a), Nyquist plot (b), and Bode plot (c) for a gold electrode in a 5 mM Fe(CN)63-/4- solution before and after potentiostatic deposition of various aryl groups at -0.7 V vs Ag/AgCl during 240 s.

SAMs on gold.23 Indeed, the phase angle of a Bode plot (|Z| vs frequency, Figure 4c) only reached a maximum value of about 50° over a limited frequency range, which is far from the value expected (90°) for a defect-free layer.23 On the other hand, the electrochemistry data, together with the deposited masses determined by EQCM, suggest that a significant desorption of the deposited or entrapped materials occurs for NP and CP during the initial rinsing step following deposition. This stems from the fact that a smaller effect is noticeable on the CV and RCT for CP relative to DEA despite that a similar mass was deposited (see Table 2). Similarly, the desorption of entrapped species (vide infra) from the NP-modified film can explain the similar electrochemical data in comparison to DEA despite the much larger deposited mass. The stability of these thick layers to longer ultrasonic treatment was evaluated by cyclic voltammetry and electrochemical impedance spectroscopy (see Figure SI 6 for NP-modified gold electrodes). An ultrasonic treatment causes a progressive decrease of the diameter of the semicircle, suggesting that a fraction of the deposited layer and/or entrapped species are removed. A steady state is

reached after about 30 min of ultrasonic cleaning, and the resulting electrode still displays significant electrontransfer barrier properties. These results demonstrate that a significant fraction of the deposited multilayers is still present at the gold surface. A similar stability was observed for both NP- and DEA-modified electrodes (Table SI 2). On the other hand, the blocking properties of the electrode modified with CP after 10 min of ultrasonic cleaning resemble those found for a thin layer (Tables SI 1 and SI 2). XPS Measurements. XPS measurements were carried out to further characterize the deposited species after modification of the gold surface. Figure 5 presents the survey spectra of gold electrodes before and after modification with thin layers. The spectrum of the bare gold surface presents only the expected peaks, including those at binding energies of 84 and 88 eV attributed to Au4f7/2 and Au4f5/2, respectively, and is devoid of any O1s component. Following modification, the intensities of these peaks decreased significantly and other peaks appeared at 285.4 (C1s), 400 (N1s), and 533 eV (O1s). These peaks confirm the presence, at the surface of the electrode, of the functional groups associated to the corresponding diazonium species. A more detailed analysis of the surface composition of the modified electrode can be obtained from the core level spectra, and the atomic surface concentrations obtained by the integration of the core level peaks are summarized in Table 3. Figure 6 presents the Au4f core level spectra before and after modification by NPDS in various conditions. As observed on the survey spectra, the deposition of a thin layer leads to a decrease of the intensity of the Au4f peaks and the relative atomic surface concentration of gold (Table 3). On the other hand, the multilayers inhibit almost totally the Au4f signal. This is in agreement with the electrochemical data and suggests that a thick layer was deposited. The ultrasonic cleaning induces only a slight increase of the gold peaks. Table 3 also gives the thickness, evaluated from the Au4f data, of the aryl layers deposited at the gold electrode surface in various experimental conditions. Despite that the thickness of the thin layers deposited (in the range of 15 Å) is larger than expected for a monolayer (vide infra), it is consistent with the EQCM data and the presence of close to a monolayer at the surface of the gold electrode. The films obtained for longer deposition times are thicker: 112 Å in the case of NPDS and 47 Å in the case of DEADAS (Table 3). On the other

Deposition of Organic Molecules at the Surface of Gold

Langmuir, Vol. 21, No. 15, 2005 6861

Table 3. Surface Atomic Composition (at. %) after Modification by Electrochemical Reduction of the Diazonium Cations NPDS(6.8

Å)d

DEADAS (8 Å)d CPDS (6.8 Å)d

c

thin layer multilayer multilayer + us thin layer multilayer multilayer + us thin layer multilayer multilayer + us

Au

C

Na (400)

Nb (406)

O

thicknessc(Å)

25.1 2.8 3 18.9 13 14.9 26.5 22.8 25.9

55.7 77.6 72.5 62.5 69.6 67 55.6 59.6 56.8

3.6 3.9 3.2 8.3 9.8 8.8 3.4 4.2 3.7

3.9 6.1 3.9 -

11.8 15.4 17.5 10.3 7.6 8.8 13.3 13.5 13.6

15 ( 2 112 ( 2 105 ( 2 18 ( 2 47 ( 2 37 ( 2 16 ( 2 25 ( 2 11 ( 2

a Evaluated from the component at about 400 eV of the N1s spectra. b Evaluated from the component at about 406 eV of the N1s spectra. Evaluated from measurements at two different angles: 90° and 30°. d Calculated thickness for a monolayer.

Figure 6. Au4f core level spectra of a gold electrodes before and following modification by NP with a thin layer, multilayers and multilayers after ultrasonic (us) treatment.

hand, the relatively less reactive CPDS leads to a thinner film of 25 Å. Following ultrasonic cleaning, the slight decrease of the thickness of the NP and DEA layers confirmed that the multilayers are strongly attached to the surface. After ultrasonication, the thickness of the CP layer corresponds more closely to that of a monolayer. These results are consistent with the electrochemical data presented above. The N1s core level spectra were measured for gold electrodes modified with NP, DEA, and CP in various conditions and following ultrasonic treatment (Figures 7 and SI 7 and Table 3). First, no peak that would be attributed to -N2+ (at 403.8 eV) was observed on any of these spectra, indicating a transformation of the diazonium cations.15 In the case of NP, the peak observed at 406 eV is attributed to the nitro group.7a In addition, the N1s spectra always present a peak at 400 eV even for a CP-modified electrode (Figure 7a). Interestingly, this feature is observed for almost all surfaces (carbon, silicon, and metals) modified with aryl groups by reduction of diazonium salts and will be discussed below. For DEA, the N1s peak also includes the contribution of the -N(C2H5) groups at about 400 eV.7c As expected, the N1s peak is more important in the case of multilayer deposition than for a thin layer (Table 3 and Figures 7 and SI 7). Ultrasonic cleaning for 1 h in water diminished only slightly this N1s peak (Figure 7b), confirming again the electrochemistry results. For DEA, the nitrogen atomic concentration, deduced from the peak at 400 eV is about 8% (Table 3) and relatively higher than that found (3.5%) for both NP and CP. It is interesting to note that the N1s signal at 400 eV is about 4% regardless of the thickness of the deposited film (vide infra). The C1s core level spectrum of each modified electrode was carefully curve-fitted in an attempt to establish the

Figure 7. (a) N1s core level spectra of gold electrodes modified by electrochemical reduction of diazonium cations in conditions to deposit a thin layer; (b) N1s core level spectra of gold electrodes modified by NP with a thin layer, multilayers and multilayers after ultrasonic (us) treatment.

covalent nature of the bond between the deposited layer and the gold surface. The peak-fitting procedure was adapted from the one used for iron-modified electrodes.14a,e In these studies, a metal-carbon contribution on the lowbinding-energy side of the main C1s peak was found at 283.8 and 283.3 eV for NP and CP grafted on an iron electrode, respectively. Unfortunately, the C1s core level spectra for the thin layers on gold electrode (Figure SI 8) do not allow the unambiguous observation of a C-Au contribution with the exception of the spectrum of the CP-modified electrode which can be fitted with a Au-C component at about 283.5 eV. The main contribution of the C1s peak is from the aromatic carbons. The additional components of the C1s envelope depend on the nature of the substituents. The important peak at 286 eV for DEA is due to the carbon atom bound to a nitrogen atom of the DEA moiety. For NP, a contribution at 286.3 eV is due to the carbon atom bound to a nitro group and the spectra of CP can be curve-fitted with three additional peaks at

6862

Langmuir, Vol. 21, No. 15, 2005

Laforgue et al.

Figure 8. Cyclic voltammograms of gold electrodes in 0.1 M H2SO4 before and after the deposition of nitrophenyl groups (cycle no. 10 and 100). Scan rate 100 mV/s. (Modification conditions: potentiostatic reduction at -0.7 V vs Ag/AgCl during 240 s in 0.1 M NBu4BF4 + 5 mM NPDS/acetonitrile).

higher binding energies that are associated to the presence of oxygen functionalities due to -COOH.14a Aging of the Deposited Film. The cyclic voltammogram of a bare gold electrode in aqueous 0.1 M H2SO4 is characterized by anodic and cathodic peaks that are associated with the formation of gold oxide and its subsequent removal by reduction, respectively (Figure 8). Following deposition of a NP thick film, the electrochemistry of the gold surface is still observable, indicating that the organic film is not homogeneous and/or not dense enough to completely hinder the gold oxidation. It is interesting to note that the CV is little affected by repetitive cycling, suggesting that the aryl groups are strongly attached to the surface and hinder the development of the oxides. This is to be contrasted with the CV data, recorded in similar experimental conditions, for a gold electrode coated with a thiophenol layer where the thiophenol monolayer was completely removed after one gold oxide formation/removal cycle.21 On the other hand, the silicon oxide growth after the deposition of an aryl layer has already been investigated for Si (111).11b It was observed that silicon oxide growth started at defects in the grafted layer and then spread laterally by removing gradually the organic layer. Thus, in the present case, the multilayered film seems to be strongly attached in such a way to prevent the increase of the gold surface oxidation to the level observed for the bare electrode. The extent of blockage of the Au surface sites by a film can be determined by comparison of the gold oxide formation or reduction charge observed at a modified gold electrode with that of the bare electrode.35 Figure 9a and b presents the variation of the surface coverage of the films upon potential cycling in experimental conditions similar to those used in Figure 8. The surface coverage of the thin film (Figure 9a) indicates that about 25% of the surface is covered by substituted aryl groups. This seems to be in contradiction with the CV results obtained with the ferri/ferrocyanide redox system where the decrease of the intensity of the peak current was about 5% in comparison with the bare electrode (Table SI 1b). A similar behavior was found for a CP thick film. The smaller blocking effect observed with the electroactive probe (35) Xia, S. J.; Liu, G.; Birss, V. Langmuir, 2000, 16, 1379.

Figure 9. Evolution of the surface coverage of the deposited layer with the number of voltammetric cycles at 100 mV/s in 0.1 M H2SO4. The surface coverage is obtained from the ratio between the oxide reduction peak integration of the bare electrode and the oxide reduction peak integration for the modified electrode. Modification conditions: in 0.1 M NBu4BF4 + 5 mM aryl diazonium salt in ACN: (a) linear sweep between open circuit potential and 0.35 (NPDS), 0.2 (CPDS), and -0.42 V (DEADAS) and (b) potentiostatic reduction at -0.7 V vs Ag/ AgCl during 240 s.

suggests that electron transfer can occur through a thin organic layer4c,6 or that the film is not compact and fairly porous. On the other hand, a fairly good agreement is obtained between the surface coverage evaluated from the oxide reduction peak and the decrease of the ferricycanide/ferrocyanide redox waves for the DEA and NP multilayers (Figure 9b and Table SI 2). This result may seem surprising since the redox processes involved in gold oxidation and with ferri/ferrocyanide differ significantly. In contrast to the electroactive system, gold oxide formation requires the ingress of water that must reach the underlying metal and the release of the protons and the reverse processes during the reduction step.35 General Discussion. Reactivity and Deposition Efficiency. The results obtained in this study for the modification by a thin layer highlighted a variance in reactivity of the diazonium cations. For example, the electrochemical reduction of NPDS clearly led to the formation of a NP film that has better barrier properties than CP. A similar observation was previously made for NP and CP films deposited on a glassy carbon electrode.7a Several factors could be invoked to explain the difference in reactivity of diazonium cations and the resulting deposition efficiency. These factors consist in the dipole moment of the aryl radical,12 the point of zero charge (PZC) of the gold electrode, and the reduction potential of the diazonium cation. These factors, which are obviously closely interelated, especially the last two, will be discussed in relation with our data. If the reduction potential of the diazonium cation is more negative than the PZC, the deposition is favorable

Deposition of Organic Molecules at the Surface of Gold

if the charge of the radical on the para position (relative to the substituent) is positive. In the case of NP and CP, these two factors should lead to a slightly larger amount of CP being deposited, which is contrary to what was observed. On the other hand, Hartig et al. established that the dipole moment of the aryl diazonium radical is very important in the grafting process at the Si surface.12 In the case of the NP radical, the dipole moment is perpendicular to the surface of the electrode and its less negatively charged side points in the direction of the surface. This results in a very fast grafting process, and close-packed layers can be achieved.12 The deposition should be less favorable for CP due to the smaller dipole moment of the CP radical relative to that of the NP radical leading to relative poor Coulombic interaction between the radical and the electrode surface. Consequently, it can be concluded that the dipole moment seems to primarily govern the higher deposition efficiency of NP relative to CP. The effect of these factors for NP and DEA deposition is interesting because their respective deposition efficiencies are similar despite a much more negative reduction potential for the N,N-diethylaniline diazonium relative to NPDS. This observation may seem to be in contradiction with a previous report on the grafting on carbon electrodes where it was shown that the amount of deposited aryl groups and film thickness increased when the redox potential of the diazonium cation was more positive.4c However, this is not a factor for the deposition of a thin film but rather for multilayer formation (vide infra). On the other hand, Hartig et al. have demonstrated that the direction of the dipole of the DEA radical is not favorable for an efficient DEA grafting.12 This is confirmed by the larger cathodic peak current and voltammetric charge recorded during the first scan in the presence of DEADAS in comparison to NPDS and CPDS. A larger current indicates a smaller blocking effect for the reduction of the diazonium cation, by the layer being grafted, during the initial cyclic voltammetry scan. Actually, a peak current density of about 200 µA/cm2 is expected for an irreversible redox system in our experimental conditions. Therefore, the peak current can provide a rough estimate of the relative grafting efficiency. On the other hand, since the reduction potential of the DEA diazonium cation is negative of the PZC for a gold electrode in acetonitrile,36 the electrostatic interaction between the diazonium cation and the negatively charged gold surface would promote the grafting of DEA. Afterward, it seems that the reduction of the diazonium cations through the first one or two deposited layers is more efficient for NPDS. Interestingly, the increasing thickness does not seem to strongly influence the deposition or deposition efficiency for DEA and NP since the deposition efficiency only decrease slightly. The continuous deposition and the formation of multilayers, with the presence of a covalent bond between the electrode surface and the grafted group, has been explained by a polymerization-type reaction between the grafted layer and free radicals in solution,6 conductance switching4c,d and electrochemically catalyzed aromatic homolytic substitution.14f,g For the former case, the aryl radicals can be generated at the outer surface and can, afterward, react with the already grafted groups to form multilayers. In the case of redox processes at defects in the layer, the electrogenerated radicals must diffuse to the outer layer surface, indicating that the radicals should be fairly stable. The mechanism (36) Hamelin, A.; Doubova, L.; Wagner, D.; Schirmer, H. J. Electroanal. Chem. 1987, 220, 155.

Langmuir, Vol. 21, No. 15, 2005 6863 Scheme 2. Growth of Multilayers

of conductance switching involves an increase of the electron-transfer rate through the organic grafted layer upon polarization of the modified electrode at an appropriate potential.4e,f The aromatic homolytic substitution is concerned with the attack of the first grafted layer by another radical to give a cyclohexadienyl radical. Following recovery of aromaticity and repetition of this second reaction would lead to thick films.14f,g Our results tend to confirm the observations of Kariuki and McDermott on the growth of covalently attached multilayers (Scheme 2).6 Tunneling in the Presence of Electroactive Species. The electron tunneling parameters through a monolayer can be determined from the electrochemical impedance measurements in the presence of soluble electroactive redox species. The apparent rate constant, k0app, can be estimated from the charge-transfer resistance, RCT:21

k0app )

RT 1 F2 RCTC

(4)

where C is the electroactive species concentration, R the gas constant, T the absolute temperature, and F the Faraday constant. If the monolayer is considered to be compact and homogeneous, the electron transfer through this layer can be described using a tunneling mechanism.4b,5,6b The apparent rate constant depends on the layer thickness, d:

k0app ) k0 e-βd

(5)

where β is the tunneling parameter and k0 the rate constant at a bare electrode. The electron tunneling parameter, β, derived from eqs 4 and 5 and the thickness estimated from the XPS data are summarized in Table SI 3. The β value of about 0.1 Å-1 is slightly lower than the one (0.2 Å-1) obtained for aryl groups deposited on a carbon electrode.4b,5 On the other hand, the value is close to that expected (0.14 Å-1) for electron transfer through conjugated polyene spacers4b and might appear satisfactorily considering the relative uncertainties about the layer thickness and the surface coverage or compactness of the layer. For example, it should be noted that an overestimation of the thickness value used in the calculation would give a smaller β value. Finally, β reported in this work should be considered as apparent and average values and are consistent with a defective film. Comparison of the Surface Coverage as Evaluated by Several Techniques. It might be interesting to compare the surface coverage as evaluated by the EQCM, XPS, and electrochemistry of the modified electrode in the presence of soluble electroactive species and gold oxide data. The first aspect to consider is that the four techniques may not exactly probe the same thing. The EQCM gives the mass deposited at the gold surface during the potential cycling, and obviously some unbound species might contribute to the mass. The thickness determined by XPS used a gold electrode that was cleaned by sputtering inside the XPS analysis chamber. If the gold electrode is not

6864

Langmuir, Vol. 21, No. 15, 2005 Scheme 3. Growth of Mixed Multilayers

cleaned by this procedure, some carbon contaminants are found at the gold surface, and if the XPS data of this electrode are used to evaluate the aryl layer thickness, then the thickness of the layer will be smaller. Thus, the thickness reported in Table 3 might be larger than they really are because the modified electrode will likely have some contaminants that will screen the Au4f signal (see eq 3). Furthermore, the gold oxide data for the thin layer suggests that about 25% of the surface is covered, whereas the CV in the presence of Fe(CN)63- seems to point to a lower coverage (vide supra). However, since tunneling through a monolayer is possible, then an actual coverage of at least 25% and up to the complete monolayer is reasonable. Consequently, considering the uncertainties and the imprecision for the measurements and data analysis, all the data are consistent with submonolayer coverage in the conditions of thin film deposition. Peak at 400 eV in the N1s Core Level Spectra. A peak at 400 eV has previously been observed following the reduction of diazonium salts on carbon,3a,4a,7 silicon,11a boron-doped diamond,10 and metals such as iron14a,d,e and copper.16 In the case of NPDS, the reduction of nitro groups to amino groups in the XPS chamber37-39 has been proposed to explain the observation of this peak. However, this mechanism cannot explain the presence of such species for CPDS. Thus, the peak at 400 eV has been also explained by the reaction of the diazonium cation with surface groups of the carbon electrode.7a,14f An alternative explanation would involve the azophenyl radical following the one-electron reduction of the diazo functional group or the diazonium itself and the subsequent grafting of azophenyl groups directly at the surface of the electrode or on a grafted phenyl ring as described in Scheme 3.40 In these conditions, the deposited layer would consist of a blend of substituted phenyl groups and substituted azophenyl groups. It is worth noting that such mixed monolayers would have a thickness around 13-15 Å, comparable to the one estimated in this study. As the deposition proceeds for longer deposition time, further reaction with already grafted species (Scheme 3) would result in the formation of thicker multilayers. It would be interesting to determine the fraction of phenyl groups that are deposited to the gold surface via a Au-C bond, where C represents a carbon atom of the (37) Mendes, P.; Belloni, M.; Ashworth, M.; Hardy, C.; Nikitin, K.; Fitzmaurice, D.; Critchley, K.; Evans, S.; Preece, J. Chem. Phys. Chem. 2003, 4, 884. (38) Eck, W.; Stadler, V.; Geyer, W.; Zharnikov, M.; Golzhausser, A.; Grunze, M. Adv. Mater. 2000, 12, 805. (39) Adenier, A.; Cabet-Deliry, E.; Chausse´, A.; Griveau, S.; Mercier, F.; Pinson, J.; Vautrin-Ul, C. Chem. Mater. 2005, 17, 491. (40) (a) Gloor, B.; Kaul, B. L.; Zollinger, H. Helv. Chim. Acta 1972, 55, 1596. (b) Reshetnyak, O. V.; Zollinger, H. Acc. Chem. Res. 1973, 6, 335.

Laforgue et al.

aryl group, (Scheme 1) or azo linkages (Scheme 3). This can be established from the relative contribution of the N1s peaks at 400 and 406 eV, as was previously reported for the electrochemical grafting of nitrophenyl at the surface of a glassy carbon electrode.7a From the XPS data of Table 3 and by assuming that the formation of dimeric species shown in Scheme 3 is not occurring for the thin layer, this one would consist of about 50% of the phenyl groups directly bonded to gold, whereas the proportion increases to 67% for the multilayer deposit. However, the formation of multilayers was established, and therefore, the real contribution of Au-C species will be higher. These fractions of Au-C species are lower than those reported for the grafting of NP on glassy carbon.7a It is also interesting to note that the atomic composition of the nitrogen species giving rise to the N1s peak at 400 eV falls in the 3-4 at. % range for both thin layer and multilayers of CP and NP films (Table 3), similar to values previously reported for the deposition of substituted aryl groups on several metals.14d Such compositions suggest that deposition of diazonium or azophenyl species is continuously occurring at the electrode surface during the deposition process. This lends some support to the deposition process involving the formation of Au-NdN-C and Au-phenyl-NdN-C linkages, as described in Scheme 3. This is to be contrasted for Cu electrode, for which such bonding appeared to be unlikely.16 Nature of a Linkage between Au and the Aryl Group. An important issue in this study is whether a covalent bond is formed between the aryl group and the gold electrode following the generation of the aryl radicals at the gold/ solution interface. Below, the various bonding schemes, which include the formation of Au-C, Au-O-C, and AuNdN-C linkages, will be presented and discussed in light of our data. The existence of a covalent metal-carbon bond between an aryl group and a metallic substrate has been deduced by XPS for metals such as iron14d,e and copper.14d,16 This surface-sensitive technique was used to confirm the presence of a carbide bond at the metal/aryl group interface which was evidenced by a C1s component on the low-binding-energy side of the main C1s peak.14e Such XPS experimental evidence was not presented in the case of the grafting on a gold electrode, but it was shown that electroactive groups were immobilized at the gold electrode surface and that these surface groups could resist sustained ultrasonic treatment.14d In our work, we have attempted to confirm the presence of a Au-C bond by XPS. Even though the C1s XPS data for the CP-modified gold electrode suggest the presence of a gold-carbon bond (Figure SI 8c), it is obvious that such a Au-C component does not always appear to be required to get a good fit for the other modified electrodes (Figure SI 8a and b). It is interesting to note that a reaction between radicals (generated by electrochemical reduction of a diazonium cation) and a metal electrode that induce blocking of the electrode surface has been previously suggested, although no details on the nature of this blockage was proposed.13 Another possible linkage, illustrated in Scheme 3, is via the formation of a Au-NdN-C bond, as was discussed above. The formation of a covalent linkage would be consistent with the observation that a significant fraction of the deposited layer resists ultrasonic treatment. Nevertheless, the decrease of the peak at 406 eV (Table 3) after ultrasonic treatment of the multilayer NP film suggests that this film contain unbound species that desorb upon ultrasonic treatment.7c These species could be produced by recombination of two phenyl radicals and a diazonium and a phenyl radicals33,34 and eventually lead to the formation of a polymer on the surface of the electrode.

Deposition of Organic Molecules at the Surface of Gold

In a related study dealing with the electrochemical oxidation of amines, Pinson and co-workers provided some evidence for the formation of a covalent Au-N bond.41 They assigned the low-binding-energy component at 398 eV to a nitrogen directly bonded to the Au metal. The N1s spectra of our modified electrode do not show the presence of such contribution. Finally, the analysis of the Au4f core level spectrum might provide some insight into the presence or absence of a covalent bond between the gold electrode and an aryl group. First, the XPS data for the electrodes modified in the multilayer conditions revealed that the bonding through a Au-O-C linkage is unlikely because the binding energies for the Au4f peaks correspond exactly to those expected for metallic gold and not to oxidized gold. In addition, in the case of the thicker film (Figure 6, NP multilayers), the contribution of the gold electrode diminishes as most of the signal originates from the organic species and only a very thin layer of the gold electrode is probed. These conditions should favor the observation of gold oxide peaks at higher binding energies (Au4f7/2 at 85.9 eV42), but clearly this is not the case. Second, there is no obvious change of the Au4f spectrum that could revealed any bond formation between aryl groups and gold. McNeillie et al. observed a Au4f7/2 peak at 84.3 and 86.8 eV in the XPS spectra of gold (I) or (III) complexes, respectively, having a Au-C-P linkage.43 A confirmation that a shift of the Au4f peak is not occurring is obtained by the fact that the difference between the binding energies of the Au4f7/2 peak and the N1s of the nitro group at about 406 eV is the same for both the thin layer and multilayers. The invariance of the peak position of the metal XPS core level spectra seems to be the general trend for metals such as iron14e and copper.16 Thus, the XPS data and the stability of these aryl groups upon potential cycling in acidic media only allow us to conclude that adhesion of substituted phenyl groups is favorable at a gold electrode surface. At the present time, there is no definite evidence for the existence of a covalent linkage (Au-C and/or AuNdN-C) from our XPS data, but more work is clearly needed to elucidate the actual chemical linkage of the aryl groups at the gold electrode surface and to confirm the validity of the bonded species illustrated in Schemes 2 and 3. Interestingly, the formation of a gold-carbon bond has been recently reported for a gold surface exposed to a solution of diazomethane and ethyldiazoacetate that yielded surface polymerization of ester-modified polymethylene.44 On the other hand, the absence of a covalent bond between the aryl group and the gold electrode does (41) Adenier, A.; Chehimi, M. M.; Gallardo, I.; Pinson, J.; Vila`, N. Langmuir 2004, 20, 8243. (42) Pireaux, W. A.; Leihr, M.; Thiry, P. A.; Debrue, J. P.; Caudano, R. Surf. Sci. 1984, 141, 221. (43) McNeillie, A.; Brown, D. H.; Smith, W. E.; Gibson, M.; Watson, L. J. Chem. Soc., Dalton Trans. 1980, 767.

Langmuir, Vol. 21, No. 15, 2005 6865

not preclude multilayer formation since this could occur through adsorbed phenyl groups and recombination of radicals (vide supra). In addition, the formation of a polymer that deposits at the gold electrode surface by recombination of many radicals is also possible. Conclusion The electrochemical deposition of aryl groups on a gold electrode surface by reduction of diazonium salts was investigated and characterized by several techniques. EQCM experiments allowed us to distinguish at least two steps in the deposition: the first one is the deposition of about a monolayer, and the second step, which occurs at a somewhat slower rate, is the growth of multilayers most likely by the deposition of molecules on the layer already deposited. The deposition efficiency varies between 10% and almost 100% depending on the experimental conditions. The barrier properties of a thin layer and multilayers to electron transfer to electroactive redox species were investigated and showed important differences depending on the nature of the deposited groups, although the thin layers seem to have very similar properties. XPS studies confirmed the electrochemistry results: the multilayer growth of CP is difficult to achieve in contrast with DEA and NP multilayers which are easily obtained. It was shown that thin layers resist ultrasonic treatment in water whereas multilayers are partly desorbed by the same treatment. However, an important fraction remains at the gold electrode surface. The observation of a nitrogen peak at about 400 eV by XPS can be explained by the coupling of diazonium or azophenyl groups with surface phenyl groups. Finally, the aging of the deposited layer in acidic medium shows that the gold oxide electrochemistry is still observable, possibly indicating the presence of defects in the layers. In fact and contrary to the thiophenol SAM on gold,21 it was possible to use the gold oxide formation/removal to estimate the surface coverage of the aryl layers. The multilayers show a surface coverage in agreement with the diazonium reactivities and demonstrate a strong stability toward the growth of gold oxides. Acknowledgment. This work was supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada through a research grant to D.B. and an equipment grant for the XPS spectrometer. The partial financial support from UQAM is also acknowledged. Supporting Information Available: Electrochemical (cyclic voltammetry, mass-potential curves, and chronoamperometry) data for the modification of gold electrodes; cyclic voltammetry, electrochemical impedance, and XPS data for the modified electrodes. This material is available free of charge via the Internet at http://pubs.acs.org. LA047369C (44) Bai, D.; Jennings, G. K. J. Am. Chem. Soc. 2005, 127, 3048.