Electrocrystallization of Rhodium Clusters on Thiolate-Covered

5124

Langmuir 2005, 21, 5124-5133

Electrocrystallization of Rhodium Clusters on Thiolate-Covered Polycrystalline Gold Steven Langerock,† Hugues Meńard,‡ Paul Rowntree,‡ and Luc Heerman*,† Department of Chemistry, Molecular Design and Synthesis Group, Katholieke Universiteit Leuven, B-3001 Leuven, Belgium, and De´ partement de Chimie, Centre de Recherche en Electrochimie et Electrocatalyse, Universite´ de Sherbrooke, Que´ bec, Canada J1K 2R1 Received January 11, 2005. In Final Form: March 9, 2005 This study reports on the electrochemical deposition of rhodium metal clusters on a polycrystalline gold electrode, modified with a monolayer of dodecanethiol through self-assembly from solution. The deposition process was investigated using cyclic voltammetry, chronoamperometry, and electrochemical quartz crystal microbalance. It is shown that the presence of the thiol monolayer drastically alters the nucleation and growth mechanism compared with the mechanism on the bare gold electrode. The small uncovered gold domains, located at the imperfections in the thiolate monolayer which are induced by the gold nanoroughness, act as nucleation sites for small rhodium clusters. At longer times, these clusters can outgrow the organic monolayer. The resulting surface morphology was characterized by scanning electron microscopy. Rhodium electrocrystallization on the bare gold substrate resulted in an ensemble of a very large amount of very small clusters that are difficult to distinguish from the gold roughness. In contrast, in the presence of a self-assembled monolayer (SAM) of dodecanethiol covalently attached to the gold electrode, the resulting deposit consisted of an ensemble of hemispherical particles. The size distribution of the rhodium particles obtained by using double step chronoamperometry was compared to the ones obtained with cyclic voltammetry and “classical” chronoamperometry. It is shown by X-ray photoelectron spectroscopy that the SAM is still present after rhodium deposition on the thiolate-covered gold substrate. Because the rhodium clusters are directly attached to the gold substrate and can thus easily be electrified, the resulting interface could be used as a composite electrode consisting of a random array of gold supported rhodium nano/ microparticles separated from each other by an organic phase. On the other hand, it is shown that the SAM is easily removed by electrochemical oxidation without dissolving the rhodium clusters and, thus, leaving a different array of rhodium clusters on the gold surface compared with the topography obtained in the absence of the SAM. From this point of view, substrate modification with such “removable” organic monolayers was found to be an interesting tool to tune the nano- or microtopography of electrochemically deposited rhodium.

Introduction Only a couple of studies were performed to explore the electrochemical metal deposition on thiolate-covered Au(111) electrodes in acidic solutions. In 1995 SondagHuethorst and Fokkink studied the galvanostatic copper deposition on thiol-modified gold electrodes.1 The presence of the thiol monolayer on the gold substrate during copper deposition resulted in the formation of hemispherical nuclei as characterized by scanning electron microscopy (SEM). This surface morphology was different than the morphology on the bare gold substrate obtained by deposition under the same conditions. It was ascribed to the larger difference in surface tension between copper and the thiol monolayer compared to the difference between copper and gold. Also, a larger deposition overvoltage was needed in the presence of the selfassembled monolayer (SAM). These authors claimed that nucleation occurred on top of the thiol layer when the SAM was highly ordered. The research group of Kern studied the process of copper deposition on Au(111)2-4 * To whom correspondence should be addressed. Tel: +32-16327337. Fax: +32-16-327992. E-mail: luc.heerman@ chem.kuleuven.ac.be. † Katholieke Universiteit Leuven. ‡ Universite ´ de Sherbrooke. (1) Sondag-Huethorst, J. A. M.; Fokkink, L. G. J. Langmuir 1995, 11, 4823. (2) Gilbert, S. E.; Cavalleri, O.; Kern, K. J. Phys. Chem. 1996, 100, 12123. (3) Cavalleri, O.; Gilbert, S. E.; Kern, K. Chem. Phys. Lett. 1997, 269, 479.

using cyclic voltammetry combined with in situ scanning tunneling microscopy (STM) to gain more information about the interaction between the alkanethiol SAM and the copper deposit. They observed the birth of a large number of copper nano-islands in the underpotential deposition (UPD) range without a clearly measurable current.2 Further growth in the overpotential deposition (OPD) range on long thiol chains was prevented while the growth mode on small chains was strongly dependent on their length. A remarkable current decrease compared with deposition on the bare gold substrate was observed in this potential range. The highly reduced growth kinetics and the strong growth inhibition caused by thick monolayers pointed, according to these authors, to the electrochemical nucleation of copper on top of the alkanethiol monolayer which acts as an electron-transfer barrier. The research group of Yoneyama investigated the initial stages of copper and silver electrodeposition on thiolate-covered gold.5-7 They observed the insertion of a UPD layer of these metals between the gold and the thiol monolayer, proceeding initially at molecular defects in the SAM. However, such penetration behavior of metal ions through an organic monolayer was only observed with rather short (4) Cavalleri, O.; Gilbert, S. E.; Kern, K. Surf. Sci. 1997, 377-379, 931. (5) Nishizawa, M.; Sunagawa, T.; Yoneyama, H. Langmuir 1997, 13, 5215. (6) Oyamatsu, D.; Nishizawa, M.; Kuwabata, S.; Yoneyama, H. Langmuir 1998, 14, 3298. (7) Oyamatsu, D.; Kuwabata, S.; Yoneyama, H. J. Electroanal. Chem. 1999, 473, 59.

10.1021/la050078z CCC: $30.25 © 2005 American Chemical Society Published on Web 04/09/2005

Rh Electrocrystallization on SAM/Au

alkanethiols on Au(111) substrates. Kolb and co-workers explored the copper8 and silver9 deposition on thiolatecovered Au(111) substrates by means of cyclic voltammetry and in situ STM. In their study, the cyclic voltammograms (CVs) did show a cathodic current response for deposition in the UPD region, which corresponded to the insertion of metal under the SAM. For Cu UPD, the amount of Cu deposited during cyclic voltammetry depended markedly on the defect density in the SAM.8 Furthermore, the same research group also observed smaller vacancy islands for long-chain SAMs compared to short-chain SAMs on Au(111) substrates in the Ag UPD range.9 This might explain the reduced deposition kinetics observed for short chains compared to long chain SAMs because the Ag UPD seems to start at such vacancy islands.6,7 Interestingly, in the OPD region, even at very low overpotential values, they observed that three-dimensional nucleation at large substrate defects was predominant. Thus, imperfections in a SAM seem to play a special role in the very initial stages of electrocrystallization on modified electrodes. This has also been concluded by other researchers.10 However, it should be mentioned that these research groups sometimes used different operational conditions, such as different preparation conditions for the substrates, thiol molecules with different chain lengths and functionalities, and so forth. This may partially explain the sometimes divergent experimental observations as described above. Subramanian and Lakshminarayanan studied the effect of roughness on the self-assembly of thiol monolayers on polycrystalline noble metal electrodes and established a direct correlation between the surface roughness and the number of pinholes or defect sites in the SAMs11 (as investigated by electrochemistry measurements for the ferricyanide redox species). The first paper related to metal deposition into defect sites in a SAM was published by the group of Crooks.12 They used electrochemical metal deposition of small islands to monitor the individual defect sites in a thiolate-covered Au(111) substrate. Very recently, Azzaroni et al. took advantage of a dodecanethiolate-covered polycrystalline copper substrate to initiate deposition on the copper substrate itself by penetration through the thiolate layer at defect sites.13 Metal-organic-metal sandwich structures might have considerable applications as catalysts14 or molecular electronics15 or for the fabrication of thin standing-free films.13 Obviously, application possibilities of the resulting interface strongly depend on the specific interactions between the SAM and the deposit and, thus, on the presence of defect sites in the SAM. In particular, the exact location of the deposited metal atoms (on top of the SAM, attached to the gold in defect sites, or buried under the SAM) is of extreme importance. For example, considerable metal penetration through the organic film can lead to short circuits in nanometer-scale electronic devices.15 The formation of either two- or three-dimensional deposits depends on the difference in interaction energy (8) Hagenstro¨m, H.; Schneeweiss, M. A.; Kolb, D. M. Langmuir 1999, 15, 7802. (9) Hagenstro¨m, H.; Esplandiu, M. J.; Kolb, D. M. Langmuir 2001, 17, 839. (10) Eliadis, E. D.; Nuzzo, R. G.; Gewirth, A. A.; Alkire, R. C. J. Electrochem. Soc. 1997, 144, 96. (11) Subramanian, R.; Lakshminarayanan, V. Curr. Sci. 1999, 76, 665. (12) Sun, L.; Crooks, R. M. J. Electrochem. Soc. 1991, 138, L23. (13) Azzaroni, O.; Schilardi, P. L.; Salvarezza, R. C. Electrochim. Acta 2003, 48, 3107. (14) St.-Pierre, G.; Chagnes, A.; Bouchard, N. A.; Harvey, P. D.; Brossard, L.; Meńard, H. Langmuir 2004, 20, 6365. (15) Herdt, G. C.; King, D. E.; Czanderna, A. W. Z. Phys. Chem. 1997, 202, 163.

Langmuir, Vol. 21, No. 11, 2005 5125

between metal atoms of the same nature and an atom of the metal under consideration with the underlying substrate atoms. Most reports on metal deposition on modified substrates are limited to “classical” systems such as Cu and Ag electrodeposition because the deposition of these metals on bare gold substrates is extensively documented. Several reports have recently been made on the electrodeposition of noble metals, such as rhodium,16-22 on different unmodified substrates. These metals are known for their (electro)catalytic performance in many chemical reactions. Rhodium often shows good catalytic activity23-27 and, thus, offers some additional advantages. It has a higher melting point and a lower density than platinum. It has a high reflectance and a low electrical resistance, and it is exceptionally hard and durable.28 According to a mechanistic theory of the research group of Meńard,14 composite electrodes consisting of electrified noble metals twodimensionally surrounded by an organic phase might show an enhanced performance toward electrocatalytic hydrogenation. This study explores the possibility to deposit small amounts of rhodium on thiolate-covered gold substrates to make such composite electrodes. To promote direct nucleation of rhodium on the gold instead of on top of the thiol monolayer, polycrystalline sputtered gold electrodes with nanoroughness are used. SAMs on such electrodes are known to be less ordered than on atomically flat monocrystals and consequently allow greater penetration of the rhodium ions through the pinholes or defect sites of the SAM toward the gold electrode. Special attention will be addressed to the contact between the rhodium deposit and the thiolate-covered gold electrode, the deposition mechanism, and the resulting topography of the interface. Experimental Section Plating Solution and Chemicals. Plating solutions were prepared from Na3RhCl6 (Alfa Aesar). Perchlorate was avoided as supporting electrolyte since Horanyi29 discovered the electroreduction reaction of perchlorate to chloride at rhodanized electrodes in a potential region were rhodium can be deposited. Because of the aquation reaction of the sodium hexachlororhodate complex that leads in a natural way to the occurrence of chloride ions in solution, NaCl (Fisher Chemicals) was chosen as the electrolyte. The pH of the solution was adjusted at a value of 4 to avoid interference of the hydrogen evolution reaction. Sulfuric acid p.a. and ferricyanide were provided by Merck. All solutions were prepared with Milli-Q water. Before each electrochemical experiment the solution was deoxygenated with purified nitrogen. (16) Pletcher, D.; Urbina, R. I. J. Electroanal. Chem. 1997, 421, 137. (17) Pletcher, D.; Urbina, R. I. J. Electroanal. Chem. 1997, 421, 145. (18) Kibler, L. A.; Kleinert, M.; Kolb, D. M. J. Electroanal. Chem. 1999, 467, 249. (19) De Dios, F. J. G.; Gomez, R.; Feliu, J. M. Electrochem. Commun. 2001, 3, 11. (20) Arbib, M.; Zhang, B.; Lazarov, V.; Stoychev, D.; Milchev, A.; Buess-Herman, C. J. Electroanal. Chem. 2001, 510, 67. (21) Langerock, S.; Heerman, L. J. Electrochem. Soc. 2004, 151, C155. (22) Oliveira, R. T. S.; Santos, M. C.; Bulhoes, L. O. S.; Pereira, E. C. J. Electroanal. Chem. 2004, 569, 233. (23) Thomas, J. M.; Thomas, W. J. Principles and practice of Heterogeneous catalysis; VCH: Weinheim, 1997. (24) Somorjai, G. A. Introduction to surface chemistry and catalysis; Wiley-Interscience: New York, 1994. (25) Bond, G. C. Heterogeneous catalysis. Principles and applications, 2nd ed.; Clarendon press: Oxford, 1986. (26) Kellogg, G. L. Phys. Rev. Lett. 1985, 54, 82. (27) Dube´ P.; Kerdouss, F.; Laplante, F.; Proulx, P.; Brossard, L.; Meńard, H. J. Appl. Electrochem. 2003, 33, 541. (28) Weast, R. C. Handbook of Chemistry and Physics, 52nd ed.; Chemical Rubber Publishing Company: Cleveland, 1971. (29) Wasberg, M.; Horanyi, G. J. Electroanal. Chem. 1995, 385, 63.

5126

Langmuir, Vol. 21, No. 11, 2005

Langerock et al.

Figure 1. CV for rhodium deposition from 5 mM Na3RhCl6 + 0.1 M NaCl on a polycrystalline sputtered gold electrode. (Conditioning and initial potential, 0.2 V; scan rate, 0.020 V/s.) Inset: Steady-state CV of the obtained rhodium deposit in 0.1 M H2SO4. (Scan rate: 0.050 V/s.) Gold Substrates. Gold evaporated on a silicon wafer (with a Ti layer of 10 nm) was used as the disposable working electrode for voltammetric or chronoamperometric experiments. The wafer was cut in 2 × 2 cm2 squares that fitted into an electrochemical cell. Prior to each experiment, the gold substrate was rinsed in a boiling sulfuric/nitric acid equimolar mixture. Caution: such mixtures exhibit strong oxidizing characteristics and therefore react strongly with several kinds of organic substances. Such a cleaning procedure should be performed very carefully. Afterward the gold substrate was mounted in an electrochemical cell with only a fraction of the gold surface in contact with the electrolytic solution. It was activated in 0.1 M sulfuric acid by sweeping its potential value between 0.1 and 1.50 V versus saturated calomel electrode (SCE) at a sweep rate of 50 mV/s until no changes between two subsequent voltammograms were observed. Afterward, this solution was removed, the plating solution was inserted, and rhodium depositions were performed using cyclic voltammetry and chronoamperometry. Rhodium metal cannot be dissolved anodically. Therefore, each deposition was performed on a new and freshly cleaned gold substrate. SAM Formation (SAM/Au). The dodecanethiol assemblies were spontaneously formed by immersing the gold substrates in a dilute dodecanethiol (10-3 M) solution of methanol. The contact time between the gold substrate and the thiol solution was 12 h unless clearly indicated in the text or figure captions. Cyclic Voltammetry and Chronoamperometry. All electrochemical measurements were performed in a classical threeelectrode configuration with the gold substrate as the working electrode, a SCE as the reference electrode, and a platinum electrode with large real surface area as the counter electrode in a separate compartment. The electrochemical cell was constructed in such a way to ensure linear diffusion to the working electrode. The potentiostat was an EG&G PAR model 273A. Electrochemical measurements were performed at room temperature (20 ( 2 °C). Electrochemical Quartz Crystal Microbalance (EQCM). EQCM measurements were performed with a home-built system. The working electrode was a polycrystalline gold electrode attached to a polished AT-cut crystal (Elchema) with a resonant frequency at 10 MHz. The mass sensitive area was 0.196 cm2. All observed resonant frequency changes were interpreted as mass changes as described by the Sauerbrey equation (∆f ) k∆m). The calibration factor k was determined by both galvanostatic and potentiostatic deposition of copper and was found close to its theoretical value (0.226 Hz cm2 ng-1). The counter electrode consisted of a platinum wire and the reference electrode was a Ag/AgCl electrode, but potentials were converted to the SCE to be able to compare with all the other electrochemical experiments reported in this paper. The voltammetric measurements were conducted with a PAR 174 A polarographic analyzer with the working electrode at hard ground. The frequency signal from the HP 53123a universal counter had a precision of 0.1 Hz and reached a stability of less than 0.5 Hz in the measured time

interval. A personal computer with a GPIB and a data acquisition board (National instruments) was used for the simultaneous measurement of the frequency and current with a sampling time of 10 ms. X-ray Photoelectron Spectroscopy (XPS). XPS spectra were measured with a VG Escalab MKII with a Mg source (KR). Samples were loaded in the vacuum chamber of the XPS apparatus 3 h after sample preparation. Measurements were performed once a pressure of less than 10-9 Torr was achieved. The measured surface was 2 × 3 mm, and the analysis depth was 50 Å. Because of minor SAM degradation during these 3 h, XPS was only performed to achieve qualitative information and no curve fitting was done. All measured binding energies were corrected by referencing the binding energy of the C(1s) level to 284.8 eV. Scanning Electron Microscopy-Energy Dispersive X-ray Spectrometry (SEM-EDX). SEM was performed on a JEOL JSM-840A with a secondary electron detector and with “Quartz imaging” software for the image recording and URSA EDS as the analysis system. High-resolution images in the nanometer range were obtained with a HITACHI S4700 using both secondary electron and backscattered electron detectors. Specific operational details (working distance, accelerating voltage, magnification, etc.) are indicated on the images.

Results and Discussion Rh Nucleation on Au: CV and SEM. Figure 1 shows a CV at a sweep rate of 20 mV/s of a polycrystalline gold electrode in contact with an aqueous solution containing 5 × 10-3 M Na3RhCl6 + 0.1 M NaCl at pH 4. The voltammogram was started at 0.2 V versus SCE and was scanned in the negative potential direction. The current response in the forward scan showed a small current increase around 0 V at the foot of a large peak beginning at -0.175 and with a peak position at -0.250 V. A pronounced nucleation loop is seen when the scan direction was reversed. This observation is generally accepted as a proof for a nucleation and growth phenomenon and was attributed to the nucleation of rhodium. This is in good agreement with the massograms of the EQCM experiments which showed a sudden mass increase around -0.175 V corresponding to the beginning of the current peak in the voltammogram, just after a minor mass gain before this value. The mass change measured before -0.175 V corresponded only to submonolayer coverages of rhodium. A constant relation between the mass increase (as calculated from the frequency shift through the Sauerbrey equation) and the cathodic charge increase was observed over a broad potential range after the current


Langmuir, Vol. 21, No. 11, 2005 5127

Figure 2. (a) Scanning electron micrograph of the polycrystalline sputtered gold substrate prior to rhodium deposition. (b) Scanning electron micrograph of the resulting deposit on the gold electrode after rhodium deposition by cyclic voltammetry. (c) Scanning electron micrograph of the resulting deposit after rhodium deposition by cyclic voltammetry on the thiolate-covered gold substrate. (Back-scattered electrons: dark gray represents the electrodeposited rhodium, light gray represents the gold substrate.) (d) Scanning electron microscopic closeup of a rhodium cluster after rhodium deposition on a thiolate-covered gold substrate.

peak. The proportionality corresponded to 33.9 g per mole electron. This value is in good agreement with the expected value for a three-electron reduction process for rhodium metal deposition from the hexachlororhodate complex. However, at potential values more negative than -0.450 V, large deviations from this proportionality were observed due to hydrogen evolution. When the potential was scanned to more positive values in the reverse scan, the current passed through zero and became anodic; the massograms did not show any mass losses that would accompany anodic stripping of the rhodium. The small mass increases at values higher than 0.150 V corresponded to the formation of a passivating (hydr)oxide layer on the rhodium deposit. The formal potential of the redox couple RhCl63-/Rh has been estimated as 0.500 V versus standard hydrogen electrode and differs somewhat from the estimation of the equilibrium potential under the experimental conditions by comparison with the crossover potential. The electrocrystallization of rhodium on the unmodified gold substrate was possible in the overpotential range only. The absence of the formation of a rhodium monolayer at underpotential values is not surprising in view of the rather large difference in lattice parameters between gold and rhodium (for (111) substrates: 4.08 and 3.80 Å, respectivily).30 This overall

picture of rhodium deposition as studied by cyclic voltammetry corresponds well with the findings of Kibler et al.18 who studied the rhodium deposition on Au(111) from 0.1 mM Na3RhCl6 in 0.1 M H2SO4 + 10 mM HCl by combination of cyclic voltammetry and in situ STM. The amount of rhodium deposited here (25.6 mC/cm2) corresponded to an equivalent of more or less 38 monolayers as the electrochemical formation of 1 monolayer of Rh(111) takes 667 µC/cm2. The presence of rhodium during cyclic voltammetry was confirmed by taking a CV of the resulting rhodium-covered gold substrate in 0.1 M H2SO4 in the double layer region of the bare gold substrate in this same electrolyte. The inset of Figure 1 shows this CV. The CV shows the typical surface features associated with a polycrystalline rhodium electrode.31 The hydrogen adsorption/desorption on rhodium was seen around -0.2 V as a reversible couple. Part of this electrochemical process was masked in the CV by the hydrogen evolution reaction on rhodium. The hydroxide layer formation and stripping was seen as broad peaks at positive potential values. SEM was used to investigate the surface morphology of the resulting Rh/Au deposit. A micrograph of the bare gold substrate before rhodium deposition is shown in Figure 2a. The topography obtained in earlier studies with

(30) Donnay, J. H. D.; Donnay, G.; Cox, E. G.; Kennard, O.; King, M. V.; Crystal data; Polycrystal book service: Pittsburgh, 1963.

(31) Florit, M. I.; Bolzan, A. E.; Arvia, A. J. J. Electroanal. Chem. 1995, 394, 253.

5128


Langerock et al.

atomic force microscopy (AFM) of this kind of gold substrate revealed a nanoroughness.21 The micrograph represented in Figure 2b was taken ex situ after the CV shown in Figure 1 and is representative for the whole deposit. The presence of rhodium was confirmed with EDX. This picture resembled closely the bare gold substrate before deposition. Close inspection with AFM in the nanometer range of a resulting deposit after rhodium deposition on similar gold substrates revealed a nanofibrous structure with an increased roughness and an increased average depth of the surface features compared to that of the bare gold substrate.21 Stability Range of the SAM. A gold electrode similar to the one used in the experiment described in Figure 1 was modified with a monolayer of dodecanethiol by selfassembly from a dilute methanol solution. The SAM was chemically characterized with XPS. The good structural qualities of the SAM were examined by cyclic voltammetry experiments with ferricyanide. Upon potential sweeping, a thiolate monolayer does not remain stable over the entire electrochemical window of an aqueous solvent. SAMs consisting of thiol molecules can be electro-oxidized to alkyl-sulfonates in an aqueous environment. This typically occurs around potential values where metal oxides are formed. In our case, the gold electrode modified with the SAM will be used in the cathodic region. The stability of a thiolate monolayer in this region is limited by the reductive electrodesorption.32 The potential value of this process has to be determined experimentally because it is not only pH sensitive but also strongly dependent on several different interactions between the alkanethiolate, the substrate, and the solvent. Figure 3 shows a CV of a bare gold substrate (dashed line) and a dodecanethiolatecovered gold substrate (full line) in contact with the background electrolyte of the plating solution used for rhodium deposition. The current increase at -0.450 V on the bare gold substrate corresponds to hydrogen evolution on gold. The thiolate-covered gold substrate exhibits a strong blocking behavior as evidenced by the strongly reduced current between -0.450 and -0.750 V compared with an unmodified gold substrate. The CV on the thiolatecovered gold substrate shows a sharp sudden current increase at about -0.750 V. Upon reversal of the scan direction, a larger current was measured than in the forward scan. Lowering the potential value of the scan

reversal (to more negative values) resulted in a larger difference in current between the forward and backward scans. This points to SAM electrodesorption at -0.750 V that is immediately followed by hydrogen evolution on the freshly exposed gold substrate. The current in the reverse scan does, however, not reach the same current density as in the case of cyclic voltammetry on the bare gold substrate because the electrodesorbed thiol molecules do not dissolve easily in water and might form micellelike structures33 on the gold surface or might reabsorb on the gold substrate which both lead to partial liberation of the gold electrode from thiol molecules. A sharp peak according to thiol desorption is seen only in aqueous solutions with high pH values. By lowering the pH of the solution to a value of 4, this peak was completely masked even at high sweep rates. In any case, the cathodic potential limit for the thiolate-covered gold electrode is clearly limited to -0.750 V. Rh Nucleation on SAM/Au: CV and SEM. The full line in Figure 4 represents a CV for rhodium deposition under the same conditions as the CV in Figure 1 on a gold substrate premodified by dodecanethiol through selfassembly from a dilute methanol solution (12 h of contact time). The current response shows a steep rise at about -0.325 V in the forward scan. In the reverse scan, a pronounced nucleation loop is seen and the current falls to zero at about the same potential value as observed in the case of rhodium deposition on the bare gold. It is obvious that a larger overpotential for nucleation is needed when the gold substrate was modified with a SAM of thiol molecules. The overpotential for rhodium nucleation on the thiolate-covered gold substrate diminished somewhat when deposited on a less-ordered SAM. This less-ordered SAM was formed by decreasing the contact time between the gold substrate and the thiol solution (15 min, dashed line). When the contact time was more than 12 h, the same CV profile as depicted with the full line in Figure 4 was always observed. The total mass of rhodium metal deposited during cyclic voltammetry on the thiolate-covered gold substrate with 12 h of immersing time was the equivalent of 20 monolayers. A CV of the resulting deposit in the double layer region of the thiolate-covered gold electrode in 0.1 M H2SO4 is shown in the inset of Figure 4. It shows the same typical surface features associated with a Rh/Au electrode. However, the charge associated with the H-UPD process on the Rh/SAM/Au electrode (280 µC/cm2) was less than the charge measured on the Rh/Au electrode (880 µC/ cm2), even taking into account the different quantities of rhodium involved (respectivily an equivalent of 20 and 38 Rh metal monolayers). Because the amount of adsorbed hydrogen is directly associated with the magnitude of the rhodium surface exposed to the electrolytic solution, it is already clear from this experiment that the morphology of the rhodium deposit would be drastically different in the case of rhodium deposition on a thiolate-covered gold surface compared to on an unmodified gold surface. Figure 2c shows a SEM micrograph of the Rh/SAM/Au surface after the CV. The gold substrate is represented in light gray color while rhodium metal is shown in a dark gray color by using backscattered electrons. The resulting deposit consists of a random array of three-dimensional nano/microparticles of different sizes. As already suggested from CV experiments, this picture is indeed totally different from Figure 2b, the resulting deposit after CV on the bare gold under the same conditions. A closeup of

(32) Kawaguchi, T.; Yasuda, H.; Shimazu, K.; Porter, M. D. Langmuir 2000, 16, 9830.

(33) Kondo, T.; Sumi, T.; Uosaki, K. J. Electroanal. Chem. 2002, 538, 59.

Figure 3. CV of the bare gold electrode in the supporting electrolyte (9) and of the thiolate-covered gold substrate (full line) in the supporting electrolyte. (Conditioning and initial potential, 0.1 V; scan rate, 0.020 V/s).


Langmuir, Vol. 21, No. 11, 2005 5129

Figure 4. CV for rhodium deposition from 5 mM Na3RhCl6 + 0.1 M NaCl on a dodecanethiolate-covered polycrystalline sputtered gold electrode. (Conditioning and initial potential, 0.2 V; scan rate, 0.020 V/s). The full line was taken after a 12-h contact time between the gold substrate and the dilute dodecanethiol solution, and the dashed line was taken after a 15-min contact time. Inset: Steady-state CV of the obtained rhodium deposit in 0.1 M H2SO4. (Scan rate: 0.050 V/s.)

a rhodium particle is presented in Figure 2d. EDX analysis of the particle itself revealed the presence of both rhodium and gold, while only gold was present between two neighboring rhodium clusters. Conventional SEM and other observation techniques such as STM or AFM are unable to to discriminate unambiguously between a spherical and a hemispherical cluster attached to a substrate. To gain more information about the exact geometry of the rhodium particles, the mother substrate after deposition has been modified. The 2 × 2 cm2 mother substrate was carefully cut according to the crystallographic axes of the underlying silicon, in such a way to obtain a smaller square, which had been completely in contact with the electroplating solution. This part of the mother substrate was then mounted on a specially fabricated sample holder that tilted the substrate with respect with the normal sample holder. Figure 5a shows an inclined scanning electron microscopic side view of the border of this daughter substrate, revealing a population of hemispherical particles. Because of the application of mechanical stress during cutting of the substrate, some of the particles are detached from their original position. Figure 5b shows a closeup view of the bottom side of a hemispherical cluster. Because the potential interval for rhodium deposition from the electroplating solution used fitted perfectly in the stability region for the SAM, the SAM was expected to be present after rhodium deposition. This was confirmed by XPS. Figure 6a shows a typical high-resolution spectrum of the S 2p region. The S 2p region shows a doublet structure with a spin-orbit splitting of 1.2 eV. The 2p3/2 level is situated at 161.8 eV, and the 2p1/2 level is situated at 163 eV. These binding energies correspond to the values observed in the literature for a gold-sulfur bond.34 Although a small shoulder arises around 163.8 eV that masks to some small extent the theoretical 2:1 area ratio for the 2p doublet structure, the presence of the goldsulfur bond remains very clear. The extremely small amount of unbound thiol species corresponding to the small shoulder was due to sample transportation before measurement as the same phenomenon was also observed in thiolate-covered blank gold substrates. The intense peak in the C 1s region (at 284.8 eV) in Figure 6b is consistent with the presence of the aliphatic carbon chains of the dodecanethiol molecules. In the onset of Figure 4, the hydrogen adsorption and desorption on the Rh/SAM/Au surface was shown. The (34) Duwez, A. S. J. Electron Spectrosc. 2004, 134, 97.

Figure 5. (a) Scanning electron micrograph after rhodium deposition on a thiolate-covered gold substrate. The sample was lifted under an angle to show the hemispherical shape of the particles. (b) Scanning electron micrograph of the same sample showing the bottom side of the detached rhodium hemispheres.

potential value for this process was the same on the Rh/ Au surfaces. This means that the rhodium hemispheres electrodeposited on thiolate-covered gold were not covered by thiol molecules which excludes the migration of the thiol molecules from the gold substrate toward the rhodium particles and which also excludes the deposition

5130


Langerock et al.

Figure 7. Potentiostatic transients for rhodium deposition at different potentials from 5 mM Na3RhCl6 + 0.1 M NaCl on a polycrystalline sputtered gold electrode covered with dodecanethiol. (Conditioning and initial potential: 0.2 V.)

Figure 6. High-resolution X-ray photoelectron spectrum of (a) the S 2p region and (b) the C 1s region.

of rhodium underneath the SAM. Thus, the resulting interface after rhodium electrodeposition on a thiolatecovered gold electrode consists of a random array of hemispherical rhodium particles two-dimensionally surrounded by thiol molecules covalently bound to the gold substrate. Rh Nucleation on SAM/Au: Chronoamperometry. To gain more information about the nucleation and growth parameters (the nucleation rate constant A, the active site density N0, and the diffusion coefficient D) involved in this deposition process, rhodium electrocrystallization was investigated at constant potential values, that is, fixed supersaturation. It was shown by EQCM that, for potential values less negative than -0.450 V, there was no contribution of hydrogen evolution, so that the measured current could safely be interpreted as the rate of rhodium deposition. All potentiostatic current transients for rhodium electrocrystallization on thiolate-covered polycrystalline gold electrodes when the potential is stepped from 0.2 V to a more negative value show the same behavior. Figure 7 shows a set of transients recorded at different potential values up to -0.450 V. They all start with a rising part and pass through a maximum to obtain the same current values at longer times. The time scale of the initial rising part diminishes with higher overpotentials (i.e., more negative potential values) which is typical for a nucleation and growth process. At long times, the current behavior of all transients fitted well with a 1/xt relationship which points to a diffusion-limited current to the electrode surface area. Scharifker and Mostany35 described the current response after a potential step for the nucleation with diffusion-controlled growth of hemispherical metal particles using the concept of planar diffusion zones. The initial rising part reflects the growth of nuclei. Because of the development of diffusion fields around growing nuclei, their growth slows down and the current

passes through a maximum and attains at longer times the diffusion-controlled current when the electrode is fully covered with a uniform diffusion layer. The resulting topography after a chronoamperometric experiment performed at -0.450 V for 30 s is represented in Figure 8. The dimensionless plots, (j/jmax)2 versus t/tmax, of the experimental transients for rhodium deposition on SAM/ Au are shown in Figure 9. Such plots are customary for the characterization of nucleation transients.35,36 The full line represents the theoretical lines for the limiting cases of instantaneous and progressive nucleation with diffusion-controlled growth. The case of instantaneous nucleation corresponds to the limiting situation that the number density of nuclei quickly reaches a constant value. When nucleation is progressive, new nuclei are formed continuously in time if the deactivation of active sites by the growing diffusion zones can be neglected. This corresponds to the experimental conditions of low concentration (c) of the electroactive species. In the dimensionless plot, the experimental points fall between the theoretical lines, suggesting an intermediate situation between instantaneous and progressive nucleation. The experimental transients were interpreted according to the theoretical model for three-dimensional nucleation with diffusioncontrolled growth of Heerman and Tarallo.36 This model, also derived by Mirkin and Nilov,37 is a correction of the

(35) Scharifker, B. R.; Mostany, J. J. Electroanal. Chem. 1984, 177, 13.

(36) Heerman, L.; Tarallo, A. J. Electroanal. Chem. 1999, 470, 70. (37) Mirkin, M. V.; Nilov, A. P. J. Electroanal. Chem. 1990, 283, 35.

Figure 8. Scanning electron micrograph of the resulting collection of rhodium clusters obtained by chonoamperometry at -0.450 V. The SEM was performed after 30 s of growth to better visualize all the clusters.


Langmuir, Vol. 21, No. 11, 2005 5131

Figure 9. Dimensionless plots of (j/jmax)2 versus t/tmax for the transients of Figure 7. The solid lines are the theoretical lines for instantaneous (upper line) and progressive (bottom line) three-dimensional nucleation with diffusion-controlled growth.

Figure 10. Comparison between the chronoamperometric transient recorded at -0.450 V on a bare gold electrode (circle) and on a thiolate-covered gold electrode (square). The full line represents the Cottrell line.

model of Scharifker and Mostany.35 The current density is given by

deposition under the same conditions as for the experiments described in Figure 7 revealed a surface morphology comparable to Figure 2b and confirms this finding. The resulting configuration after chonoamperometry at -0.450 V for 30 s (Figure 8) represents a collection of rhodium hemispheres of clearly different sizes leading to a broad particle size distribution. Size monodispersity of three-dimensional metal clusters electrodeposited on a foreign substrate is often hard to obtain. First of all, the nucleation is often “progressive”, such that the size of particles that are born at earlier times have larger sizes. Second, individual metal clusters can only grow independently during a limited time range after their birth. The research group of Penner investigated the proximity effects of diffusion-controlled electrochemically growing neighboring particles in a random arrangement.39-46 They observed by Brownian dynamics simulations and experimental depositions of different metals that the particle size distribution was due to “interparticle diffusional coupling”. The underlying cause of this phenomenon is the fact that two particles closely positioned near each other share a flux of metal ions from solution and grow at a slower rate than a metal particle located in isolation from other particles. To overcome this problem, a double pulse chronoamperometric technique was used. First, a short nucleation pulse of 0.5 s at -0.600 V was applied. The high overpotential of this pulse was chosen to obtain instantaneous nucleation without stripping of the thiol molecules. Afterward, the ensemble of nuclei was slowly grown at -0.1 V for 30 s, a potential where no more nucleation occurs, and mixed kinetics (i.e., charge transfer and diffusion control) governs the growth regime of the clusters. The deviation of pure diffusion control ensures a nonzero concentration of electro-active species at the interface of the growing nuclei with the solution so that the growth of individual particles is less influenced by the depletion layer of its closest neighbors. A closeup of an ensemble of neighboring hemispherical rhodium particles

j(t) )

zFDc Φ {1 - exp[-RN0(πDt)1/2t1/2Θ]} (1) (πDt)1/2 Θ

with

R ) 2π

(2MDc F )

1/2

1 - e-At , and At e-At Φ)1(At)1/2

, Θ)1-

∫0(At)

1/2

2

eλ dλ

M is the molar mass, and F represents the mass density. Because the nucleation and growth behavior did not seem to correspond to one of the limiting cases of nucleation, the use of the “single-point” method, proposed by Scharfiker et al.,35 to extract information on the nucleation parameters was avoided. Instead, a fitting was performed on the whole transient. The fitting used the Levenberg-Marquardt nonlinear least squares algorithm with three parameters (N0, A, and D). The analysis of the transient recorded at -0.450 V yielded N0 ) 106 cm-2, A ) 7 s-1, and D ) 1.1 × 10-5 cm2 s-1. The diffusion coefficient is in close agreement with the value reported by Arbib et al.20 However, SEM (Figure 8) did reveal a cluster density much larger than predicted by the theoretical model described by eq 1. The same behavior was found when the less accurate model of Scharifker was used. The validity of the model proposed by Heerman and Tarallo36 has been verified with digital simulations by Cao et al.38 The model was found in good agreement with the computer simulations. So, the poor agreement between the particle densities observed with SEM and the extracted information from the experimental transient profile should be associated with the basic assumptions for the use of this model. The model supposes that the clusters grow hemispherically (a) at all times and (b) under pure diffusion control. Figure 10 compares the rhodium deposition on the bare gold substrate with deposition on the thiolate-covered gold. The shape of both curves is typical for three-dimensional nucleation and growth. However, the nucleation parameters for these transients seem drastically different. A SEM picture of the resulting deposit after rhodium (38) Cao, Y.; Searson, P. C.; West, A. C. J. Electrochem. Soc. 2001, 148, C376.

(39) Zoval, J. V.; Stiger, R. M.; Biernacki, P. R.; Penner, R. M. J. Phys. Chem. 1996, 100, 837. (40) Hsiao, G. S.; Anderson, M. G.; Gorer, S.; Harris, D.; Penner, R. M. J. Am. Chem. Soc. 1997, 119, 1439. (41) Zoval, J. V.; Lee, J.; Gorer, S.; Penner, R. M. J. Phys. Chem. B 1998, 102, 1166. (42) Stiger, R. M.; Gorer, S.; Craft, B.; Penner, R. M. Langmuir 1999, 15, 790. (43) Fransaer, J. L.; Penner, R. M. J. Phys. Chem. B 1999, 103, 7643. (44) Liu, H.; Penner, R. M. J. Phys. Chem. B 2000, 104, 9131. (45) Penner, R. M. J. Phys. Chem. B 2001, 105, 8672. (46) Liu, H.; Favier, F.; Ng, K.; Zach, M. P.; Penner, R. M. Electrochim. Acta 2001, 47, 671.

5132


Langerock et al.

Figure 11. Scanning electron micrograph (5 × 5 µm2) showing the narrow size distribution between neighboring clusters by application of a nucleation pulse at -0.600 V for 0.5 s followed by slow growth at -0.1 V for 30 s.

obtained in this way in presented in Figure 11 and shows an extremely narrow size distribution compared with that of the deposits obtained by cyclic voltammetry and chronoamperometry. Further Characterization of the Rh/SAM/Au Interface. Concerning the exact nucleation position for rhodium on SAM/Au, neither the potentiostatic transients nor the SEM pictures shown above could reveal convincing details about the exact Rh/SAM/Au configuration. From the results of the SEM it was impossible to decide whether the rhodium hemispheres were attached directly to the gold substrate or whether they were located on top of the dodecanethiol monolayer. However, close inspection of the CV shown in the inset of Figure 4 suggests nucleation of rhodium on the gold substrate. Indeed, it is shown in this figure that the surface reaction of hydrogen adsorption on the rhodium metal is situated around the same potential value for the reaction on rhodium deposited on the bare gold substrate and for rhodium on the thiolatecovered gold substrate. This points to rhodium directly attached to the gold substrate, because the long aliphatic carbon chains on the gold substrate induce an extra charge transfer barrier, compared with electrochemical phenomenon on the bare gold substrate. To test this hypothesis of a direct attachment of the rhodium clusters to the gold substrate, we tried to remove the SAM from the gold substrate anodically and to show the presence of rhodium on the gold substrate afterward. Figure 12a shows a CV in 0.1 M H2SO4 at 100 mV/s of a Rh/SAM/Au electrode made by rhodium deposition on a thiolate-covered gold electrode at -0.450 V for 10 s. The first part of the voltammogram in the negative direction shows the hydrogen adsorption on the rhodium metal which is proportional to the rhodium surface in contact with the acidic electrolyte. In the reverse scan, formation of rhodium (hydr)oxide species is seen as a broad peak before the sharp current increase around 1.3 V. When the scan direction is reversed, a small peak around 0.85 V is seen, which corresponds to the reduction of a gold oxide layer. The conclusion that a gold oxide layer was electrochemically dissolved in the cathodic direction requires that it was first formed in the anodic scan, which is only possible if the SAM was first or at the same time removed. The clear observation of the gold oxide layer formation is masked by the rhodium (hydr)oxide formation. The peak starting from 0.25 V at the end of the first scan is the reduction of surface rhodium (hydr)oxide species and overlaps in this voltammogram with the hydrogen adsorption peak during the second scan. The XPS spectrum of a Rh/SAM/Au electrode after SAM removal is shown in Figure 13. The SAM removal was done by cyclic voltammetry as shown in Figure 12a. The high-resolution XPS spectrum of the S 2p region does not

Figure 12. (a) CV in 0.1 M H2SO4 of the deposit obtained by 10 s of electrodeposition at -0.450 V on a gold electrode covered with dodecanethiol. The dashed line represents the beginning of the second scan. (Scan rate: 0.100 V/s.) (b) Expanded voltammograms in 0.1 M H2SO4 solution at the H-UPD region of a rhodium deposit obtained by 10 s of electrodeposition at -0.450 V. Circles (O) show the voltammogram before thiolate stripping from the underlying gold substrate, and the full line represents the voltammogram after thiolate stripping. (Scan rate: 0.020 V/s.)

show any evidence of the S 2p doublet structure of the sulfur-gold bond of the covalently attached dodecanethiol molecules. This safely excludes the readsorption of the thiol molecules at potential values when the gold surface is reduced. No oxidized sulfur species such as sulfonates (S 2p binding energy above 166 eV) on the surface were identified. A high-resolution XPS spectrum in the Au 4f and Rh 3d region is also shown to prove the presence of these metals when the SAM is removed. To compare the number of rhodium clusters before and after SAM removal, a separate CV before and after SAM removal was taken in a limited potential range of the hydrogen adsorption/desorption region so that the hydrogen adsorption and the reduction of (hydr)oxide species did not overlap. Neither the position nor the magnitude of the hydrogen adsorption peak in the CV was changed after SAM removal so that the number of rhodium particles, as represented in the CVs by the quantity of hydrogen adsorbed, was unaffected by the SAM desorption. Counting of the number density of rhodium particles before and after SAM removal with SEM confirmed this conclusion. SAMs consisting of thiol molecules on substrates which are not monatomically flat consist of many pinholes or defect sites in the SAM. A strong nanoroughness of a gold substrate induces very small pinholes because dode-


Langmuir, Vol. 21, No. 11, 2005 5133

It should be noticed that no ohmic loss was detected for the mushroomlike rhodium particles when cyclic voltammetry in the H-UPD region on Rh/SAM/Au electrodes was compared with that on the Rh/Au electrodes (see inset of Figure 1 and inset of Figure 4). This indicates that the resulting rhodium deposit after electrocrystallization on a thiolate-covered gold electrode can be used as a composite electrode consisting of rhodium clusters and an organic phase. Conclusions

Figure 13. High-resolution X-ray photoelectron spectrum of (a) the S 2p region followed by (b) the Au 4f and the Rh 3d regions.

canethiol molecules are very strongly ordered in small domains because of the strong intermolecular van der Waals forces. This fact was illustrated by the very small current measured in cyclic voltammetry experiments of the thiolate-covered gold electrode in contact with a ferricyanide solution. Nucleation takes place in these pinholes because the metal clusters were found to be attached to the gold substrate. Arbib et al.20 concluded that the critical nucleus for rhodium electrodeposition on polycrystalline gold was zero atoms. This means that every rhodium atom deposited on a polycrystalline gold surface is a stable nucleus and can proceed with further growth. In a first stage, the nuclei are believed to grow in a cylindrical way. In a further stage, they grow out of the SAM like a mushroom to expose their hemispherical shape on top of the SAM. It is obvious that the stem of the rhodium particle could not clearly be seen in the SEM pictures because it has only the dimensions of the aliphatic carbon chain of the dodecanethiol molecules. In this study, the clusters were grown so that they could all be observed with SEM. Large concentrations and high time scales tend to larger rhodium particles. It is evident that for certain purposes, the growth could easily be limited to the nanometer range by choosing appropriate concentration and growth times. The deviation from the hemispherical shape of the clusters in the beginning of the deposition process in the defect sites probably contributes to the ambiguity of the potentiostatic transients. This offers in any case a clear warning against the uncritical interpretation of potentiostatic transients with theories that seem appropriate at first sight for the extraction of nucleation parameters.

A composite interface was fabricated consisting of a random array of rhodium particles which are all directly attached to a gold substrate and which are two-dimensionally surrounded by an organic insulating (dodecanethiol) phase chemically attached to the gold substrate. The realization of this new interface is based on the use of electrocrystallization to nucleate stable rhodium clusters in the defect sites of the thiolate SAM on a gold electrode. The defect sites in the SAM are believed to be induced by the nanoroughness of the sputtered polycrystalline gold substrate. At longer times, these stable clusters grow as mushrooms out of the SAM, to expose their hemispherical shape on top of the organic phase. An additional advantage of electrocrystallization is that the growth of the cluster can readily be controlled from the nano- to micrometer range. Using the “double step” method with slow growth in the second step for the electrodeposition of metal clusters, narrower particle size distributions could be obtained compared to classical (single step) chronoamperometric depositions. Because the rhodium particles attached to the underlying gold substrate do not dissolve anodically, the thiol monolayer can easily be stripped without dissolving the random array of rhodium clusters. The resulting topography (rhodium clusters attached to the gold substrate) differs drastically from the topography obtained by direct rhodium deposition on the bare gold substrate under identical conditions. In this way, the premodification of a rough gold substrate with a SAM before rhodium deposition followed by removal of the SAM afterward was found to be an efficient tool for the tuning of micro/nanorhodium structures. Further optimizing by varying the physical properties of the gold substrate (e.g., the roughness) or the chemical properties of the SAM is expected to open new opportunities for the control of the specific arrangement of the material structures prepared in this way. Acknowledgment. The authors thank Ireńe KelseyLe´vesque and Sonia Blais of the Universite´ de Sherbrooke (IMSI) for their help with the microscopy studies and useful discussions. L.H. and S.L. thank the FWO-Vlaanderen for financial support. S.L. is indebted to the FWOVlaanderen for the award of a doctoral fellowship (Aspirant FWO-Vlaanderen) and for a travel grant to study at the “Universite´ de Sherbrooke”. FQRNT (Quebec) and NSERC (Canada) are acknowledged by H.M. and P.R. for financial support. LA050078Z

Electrocrystallization of Rhodium Clusters on Thiolate-Covered

Recommend Documents