Anal. Chem. 2003, 75, 4534-4540
Fabrication and AFM Investigation of the Temperature-Dependent Surface Morphology of Au (100) Single Crystal Ultramicroelectrodes Vladimir Komanicky and W. Ronald Fawcett*
Department of Chemistry, University of California, Davis, California 95616
A previously reported method for preparation of gold (111) single-crystal utramicroelectrodes (scumes) has been used to fabricate gold (100) scumes with effective diameters from 20 to 50 µm. Cyclic voltammograms for these ultramicroelectrodes obtained in perchloric acid show similar features to gold (100) single-crystal electrodes of more conventional sizes, but with differences that are a result of different surface preparation methods used before the measurement. The gold crystals used to prepare Au (100) scumes were grown at room temperature so that the cyclic voltammetric characteristics reflect the unique properties of a room temperature-ordered surface. The AFM images of the (100) facets of these crystals are also presented. The effects of annealing Au (100) at 800 °C for a short time were studied both electrochemically and using AFM and are discussed with respect to the data obtained for crystals grown at room temperature. The gold (100) surface has attracted considerable attention from the surface electrochemistry community because of its unstable nature. Numerous groups have studied reconstruction phenomena by LEED,1 He-scattering,2 and STM.3-5 The conventional method of preparing the gold (100) ordered surface for electrochemical experiments requires heating the crystal prior to the measurement. This step has to be done in order to restore atomic order in the top atomic layers disturbed in mechanical polishing. The generally accepted protocol in electrochemistry, first introduced for platinum single-crystal electrodes by Clavilier,6 and extended for gold electrodes by Hamelin,7 requires annealing of the polished gold single crystal at ∼800 °C in a methane or * Corresponding author. Phone: 530-752-1105. Fax: 530-752-8995. E-mail:
[email protected]. (1) Van Hove, M. A.; Koestner; R. J.; Stair, P. C.; Biberian, L.; Kesmondel, L.; Bartos, I.; Somorjai, G. A. Surf. Sci. 1981, 103, 189. (2) Rieder, K. H.; Engel , T.; Swendsen, R. H.; Manninen, M. Surf. Sci. 1983, 127, 223. (3) Nichols, R. J.; Magnussen, O. M.; Hotlos, J.; Twomey, T.; Behm, R. J.; Kolb, D. M. J. Electroanal. Chem. 1990, 290, 21. (4) Xiaoping, G.; Hamelin, A.; Weaver, M. J. Phys. Rev. Lett. 1991, 67 (5), 618. (5) Xiaoping, G.; Hamelin, A.; Weaver, M. J. Phys. Rev. B 1992, 46, 7096. (6) Clavilier, J.; Faure, R.; Guinet, G.; Durand, R. J. Electroanal Chem. 1980, 107, 205. (7) Hamelin, A. Double-Layer Properties at sp and sd Metal Single-Crystal Electrodes. In Modern Aspects of Electrochemistry; Conway, B. E., White, R. E., Bockris, J. O’M., Eds.; Plenum Press: New York, 1985, Vol. 16, Chapter 1.
4534 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
propane flame. This produces a clean and well-oriented gold surface, as can be easily confirmed by cyclic voltammetry in sulfuric or perchloric acid. The oxide formation region on the cyclic voltammogram, which serves as a “fingerprint” for the given crystallographic orientation, has been widely used to confirm the quality of the single crystallinity of a given gold surface. It should be mentioned at this point that in many cases, these basal plane surfaces are assumed to be ideal, which means that the surface corresponds to one as free from defects as is possible. Evaluation of the contribution of defects to the electrochemical response of the given basal plane is not possible unless one makes a similarly oriented surface by an alternative method. In this paper, a different approach to the preparation of a gold (100) surface in which gold crystals are grown at room temperature in sodium silicate gels is presented. AFM images of the (100) facet of a crystal grown at room temperature are compared to the ex situ AFM image of the same surface annealed at 800 °C. In addition, cyclic voltammograms of a gold (100) scume are compared with those for the corresponding gold macroelectrode prepared by the conventional flame-annealing method. The differences in the fingerprint region are discussed and correlated with the AFM observations. The aim of this paper is not to discuss atomic scale phenomena, such as reconstruction, but to look at an area of micrometer dimensions and evaluate the contribution of the larger-scale defects, such as steps to the electrochemical response of the gold (100) surface. The fabricated (100) scumes were a few tens of micrometers in the largest dimension. Because of the small size of the crystals used in this study, it is possible to scan a considerable portion of the electrode area and to evaluate with high confidence the overall density of steps and defects on the surface of the scume. We also want to present a method for the fabrication gold (100) scumes. Ultramicroelectrodes are a unique tool in electrochemistry because of their small size. The mass transfer conditions at an ultramicroelectrode in a typical voltammetry experiment are very different from those at an electrode of normal dimensions. This fact has been used to facilitate the measurement of the kinetics of very fast electron-transfer reactions at solid ultramicroelectrodes fabricated from precious metals, such as gold or platinum.8 Well-known examples are the experiments used to determine the rate constant for the electrooxidation of ferrocene in an acetonitrile solution, one of the fastest electron-transfer (8) Fawcett, W. R.; Opallo, M. Angew. Chem. 1994, 106, 2239; Angew. Chem., Int. Ed. 1994, 33, 2131. 10.1021/ac034406g CCC: $25.00
© 2003 American Chemical Society Published on Web 07/30/2003
reactions known. At the same time, there is an interest in carrying out kinetic measurements at well-defined solid surfaces so that double layer effects may be assessed.9 The availability of singlecrystal ultramicroelectrodes (scumes) would permit these studies to be extended to fast electron-transfer reactions. In the case of a fast electron-transfer reaction, the inner and outer sphere contributions to the Gibbs energy of activation are small. However, the fraction of the total Gibbs barrier to the electron-transfer process due to the double layer effect is large and more important in a relative sense for the fast reaction. Thus, the rate of the fast reaction is controlled to a large extent by the double layer effects. By studying the reaction at Au electrodes of different crystallographic orientations, an improved estimate of the double layer contribution to the rate of the reaction can be made.10,11 The method for fabrication of gold (100) scumes presented in this paper makes use of gold crystallites grown in a silicate gel, although crystals of other metals grown by some other technique12 could also be used. Scanning electron and atomic force microscopy images of the crystallites are also presented. EXPERIMENTAL SECTION The procedure used to prepare the sodium silicate gel was described earlier.13 The first gold crystallites were observed in the vicinity of the anode after 24 h, but they were only harvested after 2 weeks when more crystals had been formed. Usually the gold anode was inserted at the top of the cell after the gel had already formed, and the gel was stirred slightly to promote formation of a higher number of smaller crystals. To promote the formation of more robust crystals containing (100) facets, the anode was made from braided gold wire. The higher surface area of this electrode promotes the growth of more robust crystals. Chlorine gas generated at the anode reacts with the gold, producing a high local concentration of the dichloroaurous anion in the vicinity of the anode. This is thought to promote the formation of more robust crystals. On the other hand, when the anode is placed in undisturbed gel at the bottom of the cell, the number of nucleation sites in the silicate gel is smaller, and formation of fewer larger crystals is observed. The size of the triangular or hexagonal crystals grown this way can reach a few millimeters, but their thickness is usually only a few micrometers. The growth of clusters of gold crystals on the cathode was also observed. Details about the mechanism and chemistry involved in the formation of gold crystals in the silicate gel were published earlier.13 The crystals in the vicinity of the anode were harvested after dissolving the gel in 5 M KOH. They were rinsed with ultrapure water with a resistivity of 18 MΩ cm (Barnstead), which was also used for preparing of all solutions throughout this experiment. The surface of the crystal contained some residues of the gel, so further rinsing with concentrated hydrofluoric acid was necessary. The crystals were then rinsed again a few times with ultrapure water and immediately pipetted onto a glass slide protected with a strip of celophane tape. The celophane tape was used because the gold crystals do not stick (9) Fawcett, W. R. Double Layer Effects in the Electrode Kinetics of Electron and Ion Transfer Reaction in Electrocatalysis; Lipkowski, J., Ross, P. N., Eds.; Wiley-VCH: New York, 1998, Chapter 8. (10) Hromadova, M.; Fawcett, W. R. J. Phys. Chem. A 2000, 104, 4356. (11) Hromadova, M.; Fawcett, W. R. J. Phys. Chem. A 2001, 105, 104. (12) Rau, H.; Rabenau, A. J. Cryst. Growth 1968, 3-4, 417. (13) Komanicky, V.; Fawcett, W. R. Angew. Chem., Int. Ed. 2001, 40, 563.
Figure 1. Schematic diagram of the micromanipulator: 1, microscope lens; 2, metal tip used with epoxy; 3, gold crystal; 4, copper tubing; 5, epoxy resin or glass suspention; 6, a stand fixed on microscope table; and 7, 8, holders for copper wire; Two positioning screws on the microscope table allowed movement of the device in the x and y directions. Tip 2 was fixed on the objective of the microscope (1) in a way such that the end of the tip was exactly in focus. The upper part of the stand could be rotated about 360°, allowing insulation of the crystal from all directions.
to it and can be easily lifted from the surface when needed. The crystals were then stored like this for weeks without any visible changes. Storing of the crystals in water was avoided, because flocculation of the crystals would occur, and also their surfaces would gradually accumulate dirt from the solution. To evaluate the background currents in the presence of epoxy, gold-covered glass slides were cut into rectangles 1 × 0.5 cm. Electrical contact was made to one side of the slide, and the whole conductive side was covered with the given epoxy resin. The thickness of the insulating epoxy resin was 1 mm. After curing of the epoxy, 0.5 cm of the electrode length was immersed in the supporting electrolyte and polarized in the chosen potential range. All epoxy resins showed quite a large resistive current, except the 2-Ton epoxy (Devcon) chosen for this study. Some epoxy resins, such as 5-Minute epoxy (Devcon) and Torr Seal epoxy (Varian Vacuum Technologies) were not completely impermeable to the electrolyte and became more conductive with prolonged exposure to the electrolyte. Further work involved development of a technique for incorporating the gold crystals into a Au (100) ultramicroelectrode. The procedure is very similar to that published earlier,13 but it differed in a few details as a result of the different geometry of the crystals that contain (100) facets. The gold crystal was rotated in a micromanipulator until the (100) facet was exactly horizontal. It was then slowly lowered onto an almost cured epoxy resin so that it just touched the resin. After 24 h, the remaining uninsulated part of the crystal was covered with epoxy resin using the tip of a micromanipulator (see Figure 1). An optical microscopic image of a gold (100) scume with an area of 4.0 × 10 -6 µm is shown in the Figure 2a. The epoxy wall is almost at the same level as the (100) facet of the crystal (see Figure 2b). After the epoxy was completely cured, the electrode was washed with acetone and allowed to dry in the air. To investigate the changes induced on the crystal surface by annealing, the gold microcrystal was attached to a gold microwire with a diameter of 25 µm using gold organometalic paint (Engelhard, Hanovia). The crystal could then be annealed at a given temperature in an electric oven. The crystal was then oriented using the micromanipulator so that the (100) facet was in a horizontal orientation. This was checked by reflection of the microscope light, which was strongest when the chosen facet on Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4535
Figure 3. Scanning electron microscope images illustrating the variety of the gold crystals grown in an electrochemical cell containing silicate gel.
Figure 2. (a) Optical microscopic image of a Au (100) electrode with a geometric area of 4.0 × 10-6 cm2. The gold microwire with a diameter 10 µm, which is attached to the crystal in order to make electrical contact, is clearly visible in the background. The crystal is embedded in epoxy resin, leaving only one facet of the crystal exposed. (b) AFM image of the Au (100) scume sealed in epoxy resin. The crystal was not annealed prior to mounting into the ultramicroelectrode. The difference between the height of the epoxy wall and crystal facet is ∼200 nm.
the crystal was exactly horizontal. The crystal was then lowered into a droplet of a suspension of ground sodium glass in water. The droplet had been previously placed on an alumina substrate. The glass suspension was then left to dry and the crystal could be then placed in the AFM. The reason for using the sodium glass suspension was to provide a glass insulation background instead of epoxy. Such a procedure was developed in the fabrication of platinum (111) scumes.14 As shown in ref 14, the sodium glass (14) Komanicky, V.; Fawcett, W. R. J. Electroanal. Chem. 2003, in press.
4536 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
does not pose a severe contamination problem in the case of the very sensitive platinum surface. All electrochemical experiments were carried out in a waterjacketed three-electrode glass cell with a gold counter electrode and a calomel reference electrode filled with 0.05 M KCl. The cell was placed in a grounded Faraday cage. The body of the working electrode was not otherwise shielded, but coaxial cables were used to connect all three electrodes to a potentiostat. All measurements were done at 25 °C. Electrolyte solutions were prepared from HClO4 (Aldrich, 99.999%) diluted to 0.01 M with ultrapure water, then purged with argon for 10 min and kept under an argon atmosphere during the measurement. The reference electrode was placed in a Luggin capillary filled with the same solution of HClO4 as in the main cell compartment. Cyclic voltammograms were recorded using an EG&G 283 potentiostat/ galvanostat connected to a PC computer by a GPIB IEEE-488.2 interface. The data were acquired using the EG&G research electrochemistry software (4.23, model 217/250). Atomic force microscopy images were acquired with an AFM-4 deflection type microscope controlled by a Nanoscope E control station interfaced with a PC computer. The silicone nitride cantilevers were made by Park Scientific, Inc. (Sunnyvale, CA) with a force constant of 0.1 N.m-1. All images were acquired using the contact mode under ambient laboratory conditions. RESULTS AND DISCUSSION Shape and Morphology of the Crystals. Figure 3 shows the morphologies of the gold crystals found in the gel around the
Figure 4. Atomic force microscopy images of the gold crystal grown in the electrochemical cell in sodium silicate gel. A 5 × 5-µm area of the surface of gold crystal which had been washed with 5 M KOH and then with ultrapure water is shown.
gold anode. The crystals were mainly triangular and hexagonal plates with a (111) orientation of gold atoms on the top and bottom side of the crystal. The sides of the triangular crystals had also a (111) orientation. (100) facets occur on the triangular crystals with cut corners (Figure 3b) and can be easily recognized by their rectangular shape. Figure 3d shows a triangular gold microcrystal ∼25 µm in size with a thickness of ∼0.5 µm. Visual perfection of the crystals is considerably affected by the density of the gel used in the growth of the crystals. In very dense gels, formation of deformed clusters of crystals with imperfect and corrugated surfaces has been observed. Low pH values with an optimal value of 1 are important for obtaining clean well-developed crystals. At neutral or slightly acidic pH values the surfaces of the crystals became contaminated with the hydrolysis products of the gold(III) ion. Figure 4 shows an AFM image of the (100) surface of a rectangular gold crystal extracted from the gel with 5 M KOH. The surface is covered with small clusters of hydrated silica. The same crystal cleaned with concentrated HF, and rinsed with ultrapure water is shown in Figure 5. The topographic image of the surface clearly resolves monatomic (110) steps intersecting at right angles. Atomic resolution has been obtained on a (111) facet cleaned with concentrated HF and annealed at 400 °C for 10 min. This is an indication that the surfaces of the gold crystals grown in the sodium silicate gel can be prepared in a very clean condition. For the study of the influence of annealing on the morphology of the (100) surface, a facet with an approximate width of 20 µm and length of 100 µm was chosen. Several AFM scans along the crystal were taken to characterize the whole surface of the (100) facet. This study was performed with a 14 × 14 µm scanner, positioning of the tip in lateral directions being achieved by means of two screws mounted on the scanning head of the AFM-4
Figure 5. (100) surface of the gold crystal washed with concentrated HF, then with ultrapure water.
Figure 6. AFM image of a gold (100) facet grown at room temperature. Steps at the ends of the large square terraces were from 30 to 120 atoms high. The large terraces were very flat with visible monatomic steps intersecting at a 90° angle.
microscope. A representative area of the surface is shown in Figure 6. The whole area of the (100) surface of the crystal contained large squarely shaped terraces with a step height at the edge of the terrace from 30 to 120 atoms high. The terraces themselves were very flat with monatomic (110) steps running at a 45° angle with respect to the larger edges of the terrace. This is an indication that large steps at the end of the terrace are either (100) or (111) steps. The angle measurement with AFM could not be used to distinguish between (100) and (111) steps because of the tip convolution and also because quite a large area of the Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4537
Figure 7. AFM image of the gold (100) facet annealed at 800 °C for 2 min. Square terraces disappeared completely from the surface.
facet was scanned. As a result, the vertical distortion was quite large, and the topographic images had to be flattened. The same crystal was placed in an electrical furnace and annealed at 800 °C for 2 min. Annealing at this temperature is commonly used in the preparation of a (100) surface prior to electrochemical measurements. The AFM image of the annealed (100) facet is shown in Figure 7. As in the previous case, almost the whole (100) surface of the crystal was scanned, and the result shown in Figure 7 is a representative sample. There is a striking difference between the images in Figure 6 and Figure 7. Square terraces no longer exist, and the surface is composed of large steps from 10 to 30 atoms high. These large steps intersect at a 45° angle with smaller steps from 1 to 4 gold atoms high. It is evident that the square symmetry of the surface defects disappeared after annealing. The larger steps on the annealed (100) surface were most likely created from the large steps at the end of the square terraces present on the (100) surface grown at room temperature. This fact is obvious when Figures 6 and 7 are compared. The density of the larger steps on Figure 7 is higher than on Figure 6 but the height of these steps is smaller. When the crystal is being annealed, atoms on the surface become more mobile. The first atoms, which are released from the gold lattice, are the ones in the kinks or the corners of a terrace. This is the reason large square terraces no longer exist on the annealed surface. The large square terraces with (100) or (111) steps on the room temperature grown surface of (100) crystal are converted into larger (110) steps upon annealing. In addition, the smaller, atomically high, square kinks disappear, and a small diagonal step between two long vertical steps is formed, as shown in Figure 8. Electrochemical Behavior of the Microelectrodes. The (100) Scume. Figure 9a shows a cyclic voltammogram recorded at a scan rate of 50 mV/s at a Au (100) single-crystal ultramicroelectrode with an area of 4.0 × 10-6 cm2. The number of peaks in the gold oxide formation region is clearly the same as that observed at a Au (100) electrode of normal dimensions, shown 4538 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
Figure 8. AFM image showing a zoomed view of a smaller area of the annealed crystal. Large diagonally running steps are intersected at 45° angle with smaller vertical steps. In addition, the smaller, square kinks have disappeared, and small diagonal steps between two long vertical steps have formed.
on Figure 9b. Four anodic peaks attributed to the formation of gold oxide and one large cathodic peak attributed to its reduction, typical for (100) gold surfaces, were observed, but the behavior of the microelectrode during the cycling was slightly different from that of a macroelectrode. The cycling of the ultramicroelectrode through the hydrogen evolution region helped to clean the ultramicroelectrode surface. When the ultramicroelectrode was not cycled through the hydrogen evolution region, the cathodic peak for reduction of gold oxide to gold became smaller each cycle, indicating possible accumulation of contaminants on the electrode surface. Typical steady cyclic voltammograms were observed after a few hours of cycling, and they did not change further, even after several more hours of cycling. Sometimes, to obtain typical “steady state” cyclic voltammograms, the scume was rinsed with ultrapure water after 8 h of cycling, left in the air for 12 h, and then cycled again. Initially, after the first cycle, only one gold(I) oxide formation peak was observed at +1.15 V. This peak slowly diminished and shifted toward a more negative potential and eventually split into two peaks labeled 3′ and 4′. In addition, after continuous cycling, two peaks labeled 1′ and 2′ developed eventually. A large cathodic peak grew during the cycling, sometimes reaching about a 30% higher value than the original peak observed on the first scan. This behavior can be attributed to self-cleaning of the scume as well as a changing of the surface topography during the cycling through the oxide formation region. The separation among the peaks corresponds to the values for the macroelectrode under the same conditions. In addition, the double layer region was flat, indicating no leakage of electrolyte through the epoxy-gold seal. Although the number of gold oxide formation peaks are the same in case of the scume and the macroelectrode, there are some distinct differences. The intensity of the peak labeled 2 is clearly
Figure 9. Cyclic voltammograms of (a) Au (100) single-crystal ultramicroelectrode in 0.01 M HClO4, A ) 4 × 10-6 cm2. (b) Au (100) singlecrystal macroelectrode in 0.01 M HClO4, A ) 19 × 10-2 cm2.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4539
higher than the intensity of peak 2′. This difference is attributed to a higher density of (110) steps on the annealed surface of the (100) macroelectrode. Hamelin15 presented four cyclic voltammograms for four crystallographic orientations between the (100) and (110) faces. It is evident from the oxide formation region of the cyclic voltammograms that transition from the (100) orientation to (110) results in an increase in the length and density of (110) steps so that the peak labeled 1 on Figure 9b gradually disappears. Only one very sharp peak, at the potential where peak 2 is, remains in the case of a (110) single-crystal electrode. According to these data, the higher intensity of peak 2 compared to peak 2′ is attributed to the more (110)-like character of the annealed (100) macroelectrode. This hypothesis is also supported by the AFM data. The annealed (100) facet compared to the (100) facet grown at room temperature clearly shows a higher density of (110) steps. CONCLUSIONS A method for fabrication of Au (100) single-crystal microelectrodes with an area in range of 4.0 × 10-6 cm2 has been developed. This area corresponds to a square shaped (100) scume with a side length of 20 µm. Ultramicroelectrodes with these areas are routinely used to probe fast electron-transfer kinetics, such as the reduction of Ru(NH3)+3 in aqueous media or the oxidation of ferrocene in acetonitrile. We measured the cyclic voltammograms of these microelectrodes in perchloric acid and compared them to those of normal sized gold single-crystal electrodes to demonstrate that this method yields well-ordered surfaces. The difference in the gold oxide formation region between a Au (100) scume and a Au (100) macroelectrode is ascribed to a different density of (110) steps on the two surfaces. This demonstrates that room temperature grown single crystals provide a very interesting alternative to conventionally prepared single crystals. Utilization (15) Hamelin, A.; Martins, A. M. J. Electroanal. Chem. 1996, 407, 13. (16) Goldschmidt, V. Atlas der Krystallformen; Carl Winters Universita¨tsbuchhandlung: Heidelberg, 1918; Band IV.
4540
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
of these crystals in electrochemistry should facilitate a new understanding of results obtained at conventionally prepared single crystals. Some new geometries of ultramicroelectrodes arose from these studies, namely, the square and rectangle in the case of the (100) orientation and the hexagon and triangle with the Au (111) scume orientation.13 The mass transfer conditions for these geometries have not yet been described in the literature. The proposed method for fabricating (111) and (100) gold scumes can also be used to prepare gold microelectrodes of other crystallographic orientations or single-crystal microelectrodes of other metals. The crucial point is to choose a suitable method and conditions to prepare a crystal of the desired morphology. Goldschmidt16 lists 117 different shapes of gold crystals found in nature with 15 different crystal faces from which at least six types of crystals show only one type of facet. A very important step in the fabrication of an ultramicroelectrode is the choice of a suitable insulating material. Five brands of commercially available epoxy resins were tested, but only one (2-Ton, Devcon) showed acceptable insulating properties. Cyclic voltammograms of the fabricated scumes did not show excessive capacitive currents in the double layer region. This indicates that adhesion of the chosen epoxy resin to gold was very good, resulting in negligible leakage of the electrolyte between the gold and the epoxy. ACKNOWLEDGMENT Financial support was provided by the Petroleum Research Foundation of the American Chemical Society (Grant ACS-PRF no. 37192-AC5) and by the National Science Foundation (Grant no. CHE 0133758).
Received for review April 18, 2003. Accepted June 18, 2003. AC034406G
