Electrochemical Measurement of the Surface Alloying Kinetics of

Jul 22, 2009 - Clean, smooth Au(111) surfaces yield a peak position of 0.53 V vs ... Some studies reported two peaks, one at each of these potentials,...
0 downloads 0 Views 3MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Electrochemical Measurement of the Surface Alloying Kinetics of Underpotentially Deposited Ag on Au(111) Joshua D. Snyder and Jonah D. Erlebacher* Department of Materials Science and Engineering and Department of Chemical and Biomolecular Engineering, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218 Received March 3, 2009. Revised Manuscript Received May 8, 2009 The cyclic voltammetry characterizing underpotential deposition (UPD) of Ag onto Au(111) varies in the literature with respect to the characteristic UPD peaks in both position and number. Rooryck et al.1 confirmed that the discrepancy in terms of peak position, specifically the initial UPD to which a third of a monolayer of deposition is attributed, is due to a variation in the quality of the surface. Clean, smooth Au(111) surfaces yield a peak position of 0.53 V vs Ag0/Ag+, while rough disordered surfaces yield a peak position of 0.61 V vs Ag0/Ag+. Repetitive potential cycling in the UPD region resulted in a gradual shift in peak position, with time as the deposited Ag alloyed with, and was stripped from the surface leaving vacancies. We provide a methodology for tracking the rate at which UPD Ag alloys with the Au(111) surface without the use of continuous potential cycling. A simple kinetic model is developed for the surface alloying of Ag on Au(111), from which we extract an activation barrier and attempt frequency for this process. Notably, we introduce a novel technique for the inexpensive parallel fabrication of Au(111) single crystals that allowed us to build statistics and ensured reproducibility of our data.

1. Introduction The deposition of Ag onto Au is an ideal system for the study of the early stages of metal adsorption onto a foreign metal substrate in electrolytic solutions. Ag and Au have nearly the same atomic size (thus, there are negligible deposition stresses), and Ag exhibits only one oxidation state in solution (simplifying the interpretation of the electrochemical measurements). Also, Ag undergoes clearly defined underpotential deposition (UPD) on Au(111), where UPD is the quasi-reversible process by which deposition of a monolayer (ML) at potentials positive of the reversible Nernst potential of the respective bulk phase occurs. The electrochemical UPD of Ag is a particularly well-studied system on both *Corresponding author. E-mail [email protected]. (1) Rooryck, V.; Reniers, F.; Buess-Herman, C.; Attard, G. A.; Yang, X. J. Electroanal. Chem. 2000, 482, 93–101. (2) Whelan, C.; Smyth, M.; Barnes, C.; Attard, G.; Yang, X. J. Electroanal. Chem. 1999, 474, 138–146. (3) Corcoran, S.; Chakarova, G.; Sieradzki, K. Phys. Rev. Lett. 1993, 71, 1585– 1588. (4) Corcoran, S.; Chakarova, G.; Sieradzki, K. J. Electroanal. Chem. 1994, 377, 85–90. (5) Hachiya, T.; Itaya, K. Ultramicroscopy 1992, 42-44, 445–452. (6) Mrozek, P.; Sung, Y.; Han, M.; Gamboa-Aldeco, M.; Wieckowski, A.; Chen, C.; Gewirth, A. Electrochim. Acta 1995, 40, 17–28. (7) Mrozek, P.; Sung, Y.; Wieckowski, A. Surf. Sci. 1995, 335, 44–51. (8) Uchida, H.; Miura, M.; Watanabe, M. J. Electroanal. Chem. 1995, 386, 261– 265. (9) Ogaki, K.; Itaya, K. Electrochim. Acta 1995, 40, 1249–1257. (10) Esplandiu, M. J.; Schneeweiss, M. A.; Kolb, D. M. Phys. Chem. Chem. Phys. 1999, 1, 4847–4854. (11) Michalitsch, R.; Palmer, B. J.; Laibinis, P. E. Langmuir 2000, 16, 6533– 6540. (12) Garcia, S.; Salinas, D.; Mayer, C.; Schmidt, E.; Staikov, G.; Lorenz, W. J. Electrochim. Acta 1998, 43, 3007–3019. (13) Liu, Y. Langmuir 2003, 19, 6888–6893. (14) Lee, J.; Oh, I.; Hwang, S.; Kwak, J. Langmuir 2002, 18, 8025–8032. (15) Chen, C.; Vesecky, S.; Gewirth, A. J. Am. Chem. Soc. 1992, 114, 451–458. (16) Chen, C.; Gewirth, A. Ultramicroscopy 1992, 42-44, 437–444. (17) Takami, S.; Jennings, G. K.; Laibinis, P. E. Langmuir 2001, 17, 441–448. (18) Kondo, T.; Morita, J.; Okamura, M.; Saito, T.; Uosaki, K. J. Electroanal. Chem. 2002, 532, 201–205. (19) Seo, M.; Aomi, M.; Yoshida, K. Electrochim. Acta 1994, 39, 1039–1044. (20) Santos, M. C.; Mascaro, L. H.; Machado, S. A. Electrochim. Acta 1998, 43, 2263–2272.

9596 DOI: 10.1021/la9007729

Au(111)1-18 and polycrystalline Au.19-23 The characteristic electrochemical voltammetry of the Ag UPD process is sensitive to several system parameters including anion species, electrolyte concentration, substrate surface orientation and cleanliness, sweep rate, etc. Consequently, there are discrepancies in the literature data with regards to the shape and peak positions in the voltammetry curves,1 and there is a lively debate regarding the structure of the partial Ag monolayer on the substrate surface.5-7 An important contribution to the study of Ag UPD on Au(111) was made by Rooryck et al.1 Their paper contains two important observations relevant to the present work. First, the available literature for Ag UPD on Au(111) was reviewed, and reasons for the observed variations in the cyclic voltammetry data among various groups were identified. Primarily, these variations were correlated to sample cleanliness and/or initial roughness. With this in mind, the characteristics of Ag UPD on clean, flat, smooth Au(111) were identified, i.e., the potentials associated with the deposition/stripping of various partial monolayers. We consider this measurement to serve as a standardized reference sample. Second, Rooryck et al.1 observed that, upon continued cycling of the potential through numerous deposition/stripping iterations, the magnitude of some of the UPD peaks decreased and new UPD peaks appeared. These observations were interpreted to be the result of surface alloying mediated by surface roughening caused by successive deposition/stripping cycles. The solid line in Figure 1 represents a Ag UPD curve that we have prepared on a clean Au(111) single crystal in a sulfate electrolyte. These data exactly reproduce the “standard” data of Rooryck et al.1 The voltammetry of Ag UPD on Au(111) shows the following characteristics: upon cathodic sweeping of the potential, the peaks A1, A2, and A3 are attributed to the deposition of approximately 1 ML of Ag, based on both charge integration1,12 and surface composition measurements;7,8 peak (21) Choi, H.; Laibinis, P. Anal. Chem. 2004, 76, 5911–5917. (22) Swathirajan, S.; Bruckenstein, S. J. Electroanal. Chem. 1983, 146, 137–155. (23) Sandoz, D.; Peekema, R.; Freund, H.; Morrison, C. F. J. Electroanal. Chem. 1970, 24, 165–174.

Published on Web 07/22/2009

Langmuir 2009, 25(16), 9596–9604

Snyder and Erlebacher

Article

Figure 1. UPD curve for Ag on Au(111) in 0.1 M H2SO4+1 mM Ag2SO4 with a sweep rate of 5 mV s-1. Solid line is the initial UPD curve on a clean Au(111) surface, and dotted line is the UPD curve after 50 cycles.

A4 accounts for a second ML deposited onto the surface. The respective “primed” peaks on the anodic sweep correspond to the stripping of Ag from the surface. Any further decrease in potential results in the initiation of bulk Ag deposition. A particularly noticeable inconsistency in past literature data is the reported peak potential for the initial UPD peak, A1, at the higher potentials. As reviewed in ref 1, potential values for this peak are reported anywhere from 0.51 to 0.63 V vs Ag0/Ag+. Some studies reported two peaks, one at each of these potentials, appearing on the cyclic voltammetry curves.24 There is also a variation in the reported amount of Ag deposited during this first stage of deposition. Reported values range from 0.31 to 0.56 ML via charge integration,5-7,9,10,12 electrochemical quartz microbalance measurements,8 and surface composition measurements,6,7 respectively. The dramatic variation in peak potential for the first stage of Ag UPD was first thoroughly investigated by Rooryck et al.1 They found that on a clean Au(111) surface the initial Ag UPD peak potential sat at 0.53 V vs Ag0/Ag+, but as they continued to cycle the potential over the entire UPD potential range, the peak at 0.53 V began to decline in amplitude, and this coincided with the appearance of a new peak at 0.61 V (peak B1). During this process it was also found that peaks A2 and A3 decreased until they no longer could be resolved; peak A4 also slowly declined in height as the potential was repeatedly cycled. The result of repetitive cycling in the Ag UPD potential region is demonstrated by the dotted line in Figure 1, again, reproducing Rooryck’s data.1 After 50 cycles, peak A1 has shifted higher in potential by approximately 100 mV, while peak A4 declined in height, and peaks A2 and A3 have disappeared. (24) Oyamatsu, D.; Nishizawa, M; Kuwabata, S.; Yoneyama, H. Langmuir 1998, 14, 3298–3302.

Langmuir 2009, 25(16), 9596–9604

Rooryck et al.1 attributed the gradual transformation of the Ag UPD voltammetry to the introduction of defects into the surface through the repetitive deposition and stripping of Ag. In their picture, the roughening of the surface was caused by alloying of the Ag adlayer with the Au surface, followed by stripping of the alloyed surface Ag and creation of surface vacancies. During the subsequent cathodic sweep, Ag is deposited into these vacancies as well as onto the surface, resulting in a higher degree of surface alloying when compared to the previous cycle. Continued cycling further roughens the surface, propagating the surface transformation until peak A1 has disappeared and peak B1 has fully developed. These observations led Rooryck et al.1 to attribute the UPD peak at 0.53 V to the deposition and stripping of a partial ML of Ag on the surface, and the UPD peak at 0.61 V to the deposition and stripping of a partial ML of Ag within the topmost atomic layer (i.e., in the surface). Deposition of Ag into surface vacancies to form a surface alloy is preferred over the formation of surface Ag because the coordination of the deposited atom will be higher at the defect site. Therefore, Ag deposits into these vacancies at a more positive potential than that of Ag deposition onto the surface. When sweeping the potential anodically, a higher potential is required to strip the alloyed Ag, again, because the Ag atoms have a higher coordination and a higher number of bonds with neighboring Au atoms. The formation of a surface Ag/Au alloy as well as physical roughening of the surface during potential cycling in the UPD range have been shown with in situ scanning tunneling microscopy (STM).2-4 In the present study, we have developed a technique to electrochemically track the rate at which a partial ML of Ag underpotentially deposited onto a smooth Au(111) single crystal forms a surface alloy at different temperatures and have used these results to back out an activation barrier for fundamental processes associated with the formation of a Ag/Au(111) surface DOI: 10.1021/la9007729

9597

Article

Snyder and Erlebacher

Figure 2. (a) Schematic of electrochemical setup for Ag UPD onto single crystal Au(111) foils. (b) Picture of large crystal Au(111) foil showing actual size of exposed surface during UPD experiments.

alloy. In contrast to ref 1, the alloying process in this work is not aided by the physical roughening of the surface through repetitive Ag deposition/stripping. The UPD Ag layer is allowed to rest in an acidified electrolyte at potentials that ensure that surface alloying occurs independent of any Ag dissolution. We suggest that it occurs through a site-exchange mechanism between Ag and Au terrace atoms. By stripping the alloyed Ag, we can track the change in integrated charge for the peaks at 0.53 and 0.61 V vs Ag0/Ag+ and plot their ratio over time. We have repeated this procedure at several different temperatures and used a kinetic model requiring only site-exchange to relate the change in the ratio of the charges over time.

2. Experimental Section 2.1. Au(111) Single Crystal Fabrication. Au shot (99.999%, Alfa Aesar) was melted in a graphite crucible placed within the coils of a radio frequency (RF) induction furnace (Ameritherm EasyHeat) and directionally solidified in an inert atmosphere by slowly pulling the crucible through the coils in order to drag residual impurities in the metal with the receding solidification front. After an annealing period of approximately 8 h at 1 050 °C in a furnace (Barnstead-Thermolyne 1500), the Au ingot was rolled into foils using a hand-operated jeweler’s rolling mill (Cavallin). The foil thickness was slowly decreased by cycles of cold rolling and annealing at 1 050 °C. Once a thickness of approximately 25 μm was achieved, the foil was rolled, sandwiched between two mirror-smooth tungsten foils (99.95%, 0.01 mm thick, Alfa Aesar), to a strain of no more than 1%. Annealing at 9598 DOI: 10.1021/la9007729

1 050 °C led to the growth of large grains. The rolling/annealing protocol was repeated until the grain sizes were on the order of centimeters. Figure 2b shows a representative sample. 2.2. Electrochemical Measurements. Electrochemical measurements on the Au(111) single crystals were performed in a two compartment Teflon cell, for which a schematic is shown in Figure 2a. At the base of the main compartment is a 3.2 mm diameter hole through which electrolyte contacted the Au(111) sample. This ensured that there were no variations in sample surface area among the experiments. A Teflon gasket ensured that electrolyte did not leak and contact other parts of the Au foil. A graphite (McMaster-Carr) base plate was used to hold the Au foil in place as well as make electrical contact. A silver wire (99.9%, Alfa Aesar) pseudoreference electrode was etched in concentrated nitric acid (70 wt. %, ACS, VWR) followed by flame annealing in a hydrogen flame to remove surface contamination. The counter electrode was Pt mesh (99.9%, 100 mesh woven from 0.0762 mm wire, Alfa Aesar) bonded to the end of a Pt wire (99.9%, Alfa Aesar) and was annealed in a hydrogen flame prior to each use. The reference and counter electrodes were separated from the working electrode by a Teflon barrier. Electrical contact was made using a salt bridge filled with an agar gel (powder, Alfa Aesar) impregnated with Na2SO4 (decahydrate, 99%, Alfa Aesar). UPD of Ag was conducted in a 0.1 M H2SO4 (concentrated, ACS plus, Fisher Scientific) + 1 mM Ag2SO4 (99.999%, Premion, Alfa Aesar) solution. All glassware and Teflon was soaked in a solution of concentrated H2SO4 and Nochromix cleaner (Godax Laboratories, Inc.) for at least 8 h followed by thorough rinsing in Millipore water with a resistivity greater than Langmuir 2009, 25(16), 9596–9604

Snyder and Erlebacher 18.2 MΩ cm (Milli-Q Synthesis A10). Millipore water was also used for all electrolyte solutions. Cyclic Ag UPD experiments were conducted within a potential range of 0.645 V to -0.01 V vs Ag0/Ag+ with a potential sweep rate of 5 mV s-1. For the alloying experiments, a partial Ag ML was deposited by sweeping the potential from 0.658 to 0.5 V vs Ag0/Ag+ at a sweep rate of 0.5 mV s-1. The electrolyte was then quickly switched to 0.1 M H2SO4, and the samples were allowed to alloy for the desired time at the desired temperature. The open circuit potential (OCP) in this electrolyte was monitored and found to sit near approximately 0.3 V vs Ag0/Ag+. Referring to Figure 1, we see that the OCP sits in between the partial monolayer UPD peak and the second monolayer deposition peak. Therefore, the sample is not subjected to conditions which should strip the UPD Ag nor deposit further Ag. Nonetheless, we confirmed the continued presence of UPD Ag on Au(111) samples following exposure to 0.1 M H2SO4 at OCP for extended periods of time using Auger electron spectroscopy. After resting, the partial ML was then stripped by sweeping the potential from 0.5 to 0.658 V vs Ag0/Ag+ at a sweep rate of 0.5 mV s-1. The slow sweep rate was used to limit the error in calculation of the deposition and stripping charges introduced by double layer charging. The current under the UPD peaks at 0.53 and 0.61 V was then integrated to determine the charges associated with surface Ag and alloyed Ag for each alloying time interval. Pb UPD curves were obtained in a 1 mM Pb(NO3)2 (Alfa Aesar, 99.999%)+0.1 M HClO4 (Sigma Aldrich, 70%, redistilled, 99.999%) solution by cycling at a rate of 100 mV s-1 between potentials of 0-0.8 V vs Pb0/Pb2+. Pt mesh and Pb wire (Alfa Aesar, 0.5 mm, 99.998%) were used as counter and reference electrodes, respectively. Au oxidation curves were measured in 0.1 M H2SO4 using a Hg/Hg2SO4 reference electrode (Radiometer Analytical) by cycling at a sweep rate of 50 mV s-1 within the potential range 0.5-1.8 V vs NHE. All Au(111) single crystals were kept in a furnace at 1 050 °C until they were to be used. After removal from the furnace, they were quickly placed at the base of the Teflon cell and immersed in electrolyte under potential control.

3. Results 3.1. Electrochemical and Microscopy Characterization of Au(111) Single Crystals. A notable aspect of this study is our ability to build statistics by repeating each data point several times with new Au(111) single crystal samples. Au single crystals are typically grown from seed crystals and subsequently cut and polished to the proper orientation, a time-consuming process. Single crystals produced in this manner are also rather expensive. We report here a simpler technique for the parallel fabrication of clean large Au(111) single crystals. The method is based on techniques originally used to drive grain growth in pure aluminum and has been sometimes referred to as critical strain annealing.25 The technique involves rolling a polycrystalline foil of Au to ∼25 μm thickness followed by slight rolling to impart little strain, which does not impart enough plastic deformation to drive recrystallization (i.e., nucleation of new grains), but it does depin grain boundaries and leads to copious abnormal grain growth in which low surface energy (111)-oriented grains grow significantly faster than any other grain orientation. Crystallographic orientation of the large grains was confirmed by STM, which shows a hexagonal 6-fold symmetry with interatomic distances consistent with Au(111), Figure 3b. Atomic force microscopy (AFM) of the macro-terraces reveal step trains with terrace widths ranging from 200 nm to 1 μm, separated by atomically high steps, Figure 3a. The rms roughness of the Au surface is approximately 3 A˚ over a 25 μm2 surface area. A particular feature of our Au(111) single (25) Van Lancker, M. Metallurgy of Aluminum Alloys; Wiley: New York, 1967.

Langmuir 2009, 25(16), 9596–9604

Article

Figure 3. (a) AFM on a macro-terrace of a Au(111) foil grain reveals wide, ML-high, atomically flat terraces. (b) Fourier filtered atomic resolution STM image of a terrace shows the hexagonal symmetry of the Au(111) surface with nearest-neighbor distance 2.9 ( 0.1 A˚ . Scan size 2.3 nm square.

crystals is the absence of screw dislocations. This observation contrasts with vapor-deposited Au films, which typically have a high density of screw dislocations due to the coherence mismatch between Au and the target substrate. Vapor-deposited films also possess considerable deposition stresses. Atomic probe studies of vapor-deposited Au films determined terrace widths to be on the order of 10 nm,26 which are much smaller than the terrace widths for our Au(111) single crystal foils. Particularly useful for this surface alloying study, the wide terraces gave us a large area on which Ag could alloy with the surface without the potentially complicating effect of a high step density. Repeating the experiments using numerous single crystal samples allowed us to confirm the reproducibility of our results in these very contamination-sensitive experiments and provide good data statistics. Electrochemical measurements on Au(111) single crystals are characterized by clean, sharp peaks as shown in Figures 1 and 4 for Ag UPD, Au oxidation/reduction, and Pb UPD. The UPD of Pb is particularly sensitive to surface orientation, as reported in the literature.27,28 The Pb UPD curve in Figure 4b recorded on our Au single crystals contains clean, sharp peaks for the UPD process, the shape and potentials of which agree well with literature data for Pb UPD on Au(111).29-32 The cyclic voltammograms (CVs) in both Figures 1 and 4 demonstrate the cleanliness and (111) orientation of our Au single crystal foils. Using our Au(111) single crystal foils, we were able to reproduce the transformation of the Ag UPD curve shown in ref 1 by cycling the potential in the region of the initial partial ML deposition from 0.45 to 0.65 V vs Ag0/Ag+; see Figure 5. Peak A10 , which has been attributed to stripping of approximately onethird of a monolayer from the Au surface, slowly decreased in height with each cycle and began to shift to peak C01 during the early stages of the transformation. Initially, peak C01 increased in (26) Hallmark, V. M.; Chiang, S.; Rabolt, J. F.; Swalen, J. D.; Wilson, R. J. Phys. Rev. Lett. 1987, 59, 2879–2882. (27) Hamelin, A. J. Electroanal. Chem. 1979, 101, 285–290. (28) Adzic, R.; Yeager, E.; Cahan, B. J. Electrochem. Soc. 1974, 121, 474–484. (29) Motheo, A.; Gonzalez, E.; Tremilliosi-Filho, G.; Rakotondrainibe, A.; Leger, J.; Beden, B.; Lamy, C. J. Braz. Chem. Soc. 1998, 9, 31–38. (30) Green, M.; Hanson, K.; Carr, R.; Lindau, I. J. Electrochem. Soc. 1990, 137, 3493–3498. (31) Seo, M.; Yamazaki, M. J. Electrochem. Soc. 2004, 151, E276–E281. (32) Chen, C.; Washburn, N.; Gewirth, A. J. Phys. Chem. 1993, 97, 9754–9760.

DOI: 10.1021/la9007729

9599

Article

Snyder and Erlebacher

Figure 5. Ag UPD curve on furnace grown Au(111) single crystal foils in 0.1 M H2SO4 +1 mM Ag2SO4 at a sweep rate of 5 mV s-1. Potential is cycled within the partial monolayer range of the UPD curve from 0.45 to 0.65 V vs Ag0/Ag+.

Figure 4. (a) Au oxidation curve in 0.1 M H2SO4 at a sweep rate of 50 mV s-1 and (b) Pb UPD curve in 1 mM Pb(NO3)2 + 0.1 M HClO4 at a sweep rate of 100 mV s-1 on furnace grown Au(111) single crystals.

height but eventually, after peak A01 completely disappeared, peak C01 also began to decrease as the surface continued to roughen, and Ag further alloyed with the Au surface. Peak C01 can be associated with Ag on terrace sites, but the potential is shifted slightly positive of peak A01, and thus, it is perhaps more likely that peak C01 is associated with stripping of Ag atoms at or near surface defects such as steps. Simultaneous to the decrease of peaks A1 and A01, peaks D1 and D10 quickly disappeared. The presence of these peaks is only observed on well-ordered Au(111) 9 and vanishes as the surface roughens and begins to disorder due to stripping of the alloyed Ag. The transformation of the surface to a disordered state, with the creation of more and more surface vacancies as the potential is repetitively cycled, resulted in the growth of peaks B1 and B10 , which have been attributed to the deposition of Ag into higher coordinated terrace sites and their subsequent stripping from these sites.1 After approximately 250 cycles at 5 mV s-1, the surface was fully transformed, and the entire partial ML, approximately one-third of a ML based on 9600 DOI: 10.1021/la9007729

integrated deposition charge, was alloyed with the surface. In Figure 1, the potential was cycled over the entire UPD potential range, while in Figure 5, it was cycled over the smaller range of the initial partial ML deposition. The transformation occurred faster for the data shown in Figure 1, which was complete after only 50 cycles at 5 mV s-1. This may be because more Ag was deposited onto the surface during each cycle, increasing the amount of vacancies injected per cycle and, consequently, the rate at which the surface was roughened. 3.2. Surface Alloying of the Ag Partial ML on Au(111) Single Crystal Foils. Figure 6 shows the time evolution of the surface alloying process for a partial ML of Ag deposited on the Au(111) surface without cycling the potential. In this experiment, the partial Ag ML was allowed to rest in a dilute H2SO4 solution at OCP ensuring that the alloying process was due solely to a spontaneous exchange between the Ag adlayer and Au surface atoms. The data for each alloying time were obtained on an individual clean, smooth Au(111) single crystal. In comparison of the data in Figure 6 to that in Figure 5, similar peak shifts A01 and C10 to B01 are seen. This suggests that the surface is alloying over time, even though it is not being roughened by repetitive voltage cycles. It could be argued that resting the samples in acidic electrolytes might spontaneously increase the defect density on the Au(111) surface and possibly insert surface vacancies into which the UPD Ag atoms may migrate. However, Whelan et al.2 have shown that, while cycling in a Ag+ free acid solution does slightly roughen the surface leading to very slight shifts in deposition/stripping potential for Ag UPD, it does not lead to the dramatic structural transformation caused by repetitive Ag deposition and stripping. Even if surface vacancies are created while resting the Au(111) samples in dilute H2SO4, the tendency for the surface to selfanneal and eliminate surface roughness in electrolyte would quickly smoothen out these defects. This process is enhanced in acidic electrolytes because the surface diffusion of Au is increased by several orders of magnitude, at least to 10-14 cm2 s-1,33 (33) Alonso, C.; Salvarezza, R. C.; Vara, J. M.; Arvia, A. J. Electrochim. Acta 1990, 35, 1331–1336.

Langmuir 2009, 25(16), 9596–9604

Snyder and Erlebacher

Article

Figure 6. Plot of the time evolution of the surface alloying process at 20 °C. After deposition of partial ML, surface is rested in dilute H2SO4 in order to allow the Ag to alloy with the Au(111) surface. Partial Ag ML is then stripped using linear sweep voltammetry at a sweep rate of 0.5 mV s-1 for samples allowed to alloy for 0, 5, 10, 20, 30, and 45 min. Ag stripping current density is normalized by the corresponding deposition charge for that sample because there is a slight difference in deposition charge from sample to sample because Ag UPD is highly sensitive to the surface structure.

compared to that in vacuum, 10-19 cm2 s-1,34 allowing capillary forces to quickly drive the smoothening of the surface in an attempt to reduce surface energy. For these reasons, it is unlikely that surface roughening during the alloying process occurs to a high degree in our experiments and can, therefore, be disregarded when interpreting the data. However, the same self-annealing tendency that limits the surface roughness also limits the interpretation of the kinetics of surface alloy formation by Rooryck et al.1 Because alloy formation in their experiments was induced by physical roughening of the surface, there existed a competition between surface roughening and capillary-driven surface smoothening. Increasing the temperature of the electrolyte enhanced the rate of surface smoothening by increasing the surface diffusivity of Au, and this tended to lead to the annihilation of some of the surface vacancies formed by the previous deposition/stripping cycle. These competing processes resulted in a difficult interpretation of the kinetics of surface alloying. We repeated the surface alloying experiments at 0 and 40 °C. The results of these experiments are plotted in Figure 7 and show that the ratio of the integrated charge under the surface Ag current peak at 0.53 V vs Ag0/Ag+ to the integrated charge under the alloyed Ag current peak at 0.61 V vs Ag0/Ag+ decreased with time at different rates for different temperatures. Error bars were calculated through a variance in the data for each point, each of which was an average of three tests on three distinct clean Au(111) substrates. The general trend found in Figure 7 is intuitively expected. The rate at which the Ag alloys with the Au surface increased with increasing temperature. Over a short, initial period of time, the bulk of the Ag alloyed with the Au; alloying appeared to slow at longer times.

Figure 7. Raw surface alloying data. Surface Ag charge obtained by integrating current peak at ∼0.53 V vs Ag0/Ag+, and alloyed Ag charge obtained by integrating current peak at ∼0.61 V vs Ag0/ Ag+. Experiments were run at temperatures of 0 °C (triangle), 20 °C (square), and 40 °C (diamond). Error calculated by repeating each data point three times with three different Au(111) substrates.

consistent with our observed data. By “site-exchange” we mean a mechanism by which a Ag and Au atom exchange positions without an intermediate vacancy; a knockout mechanism in which the Ag atom pushes the terrace Au atom out is an example of a particular manifestation of such a mechanism. Consider a flat (111) Au terrace with an evenly distributed UPD layer on the surface. Denoting the layer where the Ag adsorbs as surface “0” and the topmost layer of the Au electrode as surface “1”, we write the concentration of Au in each layer as C0Au and C1Au, respectively. Similarly the concentration of Ag in each layer is C0Ag and C1Ag. Assuming no desorption, the concentration of Au in each layer is related by C1Au =1- C0Au. The only way that Au may be present in layer “0” is if it exchanges with a Ag atom from that layer; consequently, the concentration of Au in that layer is C0Au= 1- C0Ag, which leads to the observation that C1Au=C0Ag. If surface alloying occurs by site-exchange, then the rate of change of the concentration of C0Ag is related to the concentrations of Au and Ag and a rate constant k via: 0 dCAg

dt

0 1 0 2 ¼ - kCAg CAu ¼ - kðCAg Þ

This equation is easily solved using the boundary condition requiring that the concentration of Ag in layer 0 be equal to unity at time t = 0. One finds 0 1 ðtÞ ¼ 1 - CAg ðtÞ ¼ CAg

0 CAg ðtÞ

4.1. Evidence for a Site-Exchange Mechanism for Surface Alloying of UPD Ag on Au(111). A simple model including only site-exchange of UPD Ag with terrace Au is

1 ðtÞ CAg

Langmuir 2009, 25(16), 9596–9604

1 kt þ 1

ð2Þ

The data in Figure 7 is presented in terms of a ratio of surface Ag to alloyed Ag, which this model predicts to be of the form:

4. Discussion

(34) Seebauer, E. G.; Allen, C. E. Prog. Surf. Sci. 1995, 49, 265–330.

ð1Þ

¼

1 kt

ð3Þ

The analysis suggests that the change in the ratio of UPD Ag to surface alloy Ag with time should scale as 1/[v exp(-Q/kBT)t] assuming that k follows Arrhenius kinetics with activation barrier DOI: 10.1021/la9007729

9601

Article

Figure 8. Data presented in Figure 7 collapsed by using the scaling relation developed in eq 3. The data are fit with a line that is forced to go through zero at infinite time by varying the activation barrier within k until a best fit to all of the data is obtained. As in Figure 7, data are obtained at temperatures of 0 °C (triangle), 20 °C (square), and 40 °C (diamond).

Q and attempt frequency v. Plotting C0Ag/C1Ag vs 1/[v exp(-Q/ kBT)t], we find that the data collapse with appropriate values of v and Q and can be fit with a straight line as shown in Figure 8. The data fit is forced to go through zero because at infinite time the entire partial ML of Ag is expected to be alloyed with the surface. The best fit to the data as shown in Figure 8 is obtained with an activation barrier of 0.40(0.03 eV and attempt frequency v ∼106 sec-1. Inherent in the relationship developed in eq 3 is the assumption of second-order kinetics with respect to the surface Ag concentration. We also consider only the event of site-exchange between an adsorbed Ag atom and a surface Au atom and not the reverse event. It has been shown for a similar system, Pb UPD on Ag(111)47, that low coverages of Pb alloyed with Ag tend to remain in the surface layer under potentiostatic conditions leading to a negligible exchange rate between Ag surface adsorbates and alloyed Pb atoms. The empirical evidence and adequate fit of our model to the data suggest that our assumption of secondorder kinetics with respect to surface Ag concentration is reasonable. The assumption of second-order kinetics contrasts with a kinetic model for surface alloying developed for Pb UPD on Ag(111) developed by Popov et al.35,36 in which the site-exchange rate is proportional only to the concentration of adsorbates. This model leads to a decaying exponential for the ratio C0Ag(t)/C1Ag(t), in contrast to eq 3. A decaying exponential does not fit our data at any time, suggesting strongly that for Ag/Au(111), siteexchange only occurs between an adsorbed Ag atom and a terrace Au atom. The value of 0.4 eV for the activation barrier associated with Ag adatom exchange with terrace Au is comparable to other literature values for similar surface processes on other facecentered cubic (fcc) metals both in experiment and simulation. For instance, a field-ion microscope was used to track Pt adatom movements on Ni(110) under UHV conditions and at temperatures above 105 K; exchange between surface Ni atoms and Pt (35) Popov, A.; Dimitrov, N.; Kashchiev, D.; Vitanov, T.; Budevski, E. Electrochim. Acta 1989, 34, 269–271. (36) Dimitrov, N.; Popov, A.; Kashchiev, D.; Vitanov, T. Electrochim. Acta 1994, 39, 957–960.

9602 DOI: 10.1021/la9007729

Snyder and Erlebacher

adatoms was observed with an activation energy of 0.28 eV.37 Molecular dynamics simulations of the displacement of a Au atom from a Au(001) surface by adsorbed Co atoms predict an activation barrier for this process of 0.40 eV. In this case, the exchange process was driven by the stress reduction due to alloying because there is a significant lattice mismatch between Co and Au.38 Patthey et al.39 used both valence-band photoemission and embedded-atom (EAM) theoretical calculations to determine activation barriers for the exchange process between Pd atoms deposited on a Ag(100) surface. These values were found to be 0.43 and 0.53 eV, respectively, showing that the experimental data are matched well by the model calculations. Tracking of radioactive Ag tracers in Ag/Au alloys have determined migration energies to be approximately 0.77 eV.40 This value, however, cannot be directly compared to our study because it was derived by tracking migration through a bulk Ag/Au alloy rather than integration of Ag atoms into the surface layer of a pure Au substrate; the higher activation barrier is reasonable for a bulk process. Stoltze41 used effective medium theory to determine activation energies for several diffusive processes on transition metal surfaces such as Au(111). These processes include vacancy diffusion (0.455 eV), adatom exchange (0.554 eV), diffusion of an atom out of a step (0.559 eV), and diffusion of an atom over a step by an exchange mechanism (0.326 eV). Again, while these reported values cannot be directly compared to our determined activation barrier due to the differences in the experimental conditions and atomic species involved, they provide values for similar surface processes that show that our measured activation barrier is reasonable. Our model does not take into account alloying at step edges,1-3,42-44 which in ref 1 was specifically shown to enhance the rate of surface alloying during potential cycling. How much of an effect might step edges have on surface alloying? It is known that removal of the alloyed Ag layer leads to roughened steps as viewed by STM.2 This result is not unexpected; the energy cost is lower to remove an atom from a seven-coordinated step edge site than a nine-coordinated (111) terrace site. However, STM investigations of UPD/stripping experiments for Ag/Au,2 Pb/Au,45 and Au/Cu46 have shown a high density of surface vacancies present after removal of the partially alloyed UPD layer, suggesting that significant alloying can occur through site-exchange with terrace atoms. When Au is deposited onto Cu, Cu atoms ejected onto the surface diffuse and agglomerate to form Cu-rich islands. If alloying were to occur strictly at step edge sites, then it is likely that these islands would not form because the Cu atoms ejected from the step would simply reattach themselves to the step.46 UPD of Pb onto Ag(111)47 also demonstrates alloying of adsorbate atoms with terrace atoms. Popov et al.47 concluded that at low coverages, incorporation of Pb atoms occurs primarily on terraces, while they have previously shown that incorporation of Pb atoms occurs primarily at step edges for coverages of 1 ML (37) Kellogg, G. L. Phys. Rev. Lett. 1991, 67, 216–219. (38) Stepanyuk, V. S.; Bazhanov, D. I.; Hergert, W. Phys. Rev. B: Condens. Matter Mater. Phys. 2000, 62, 4257–4260. (39) Patthey, F.; Massobrio, C.; Schneider, W. D. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 53, 13146–13149. (40) Mallard, W. C.; Bardner, A. B.; Bass, R. F.; Slifkin, L. M. Phys. Rev. 1963, 129, 617–625. (41) Stoltze, P. J. Phys.: Condens. Matter 1994, 6, 9495–9517. (42) Gambardella, P.; Kern, K. Surf. Sci. 2001, 475, L229–L234. (43) Vitanov, T.; Popov, A.; Staikov, G.; Budevski, E.; Lorenz, W. J.; Schmidt, E. Electrochim. Acta 1986, 31, 981–989. (44) Schmidt, U.; Vinzelberg, S.; Staikov, G. Surf. Sci. 1996, 348, 261–279. (45) Green, M. P.; Hanson, K. J. Surf. Sci. Lett. 1991, 259, L743–L749. (46) Chambliss, D. D.; Chiang, S. Surf. Sci. Lett. 1992, 264, L187–L192. (47) Popov, A.; Dimitrov, N.; Velev, O.; Vitanov, T.; Budevski, E.; Schmidt, E.; Siegenthaler, H. Electrochim. Acta 1989, 34, 265–268.

Langmuir 2009, 25(16), 9596–9604

Snyder and Erlebacher

and higher. This difference is attributed to a variation in the interaction forces among the adsorbates with surface concentration. For partial ML coverage, the interaction forces among the adsorbates are low, almost negligible, due to a larger separation distance between atoms in the less compact layer. For full ML coverage, the interaction forces are high in the close-packed layer; these strong lateral interactions hinder incorporation of the UPD atoms into terraces, preferring to alloy with the more active step sites.47 In refs 1-3 and 42-44, the surface transformations are due to deposition and stripping of at least 1 ML of adsorbate coverage. It is possible that surface alloying in those studies is limited to step edges due to strong lateral interactions of the UPD atoms. In this study, surface alloying involved only the one-third ML of Ag coverage, suggesting that we had negligible interaction between adsorbate atoms and that surface alloying preferentially occurred on terraces, not at step edges. If step edges do not play much of a role in surface alloying, perhaps lateral interactions between the Ag atoms in the adsorbed adlayer do. However, it has been shown that a partial ML of Ag on Au(111) does not preferentially agglomerate at step edges or in compact islands, but rather exists as an extended network on the surface,4 an observation that also tends to support exchange with terrace atoms due to low lateral interaction of the adlayer atoms. Ag adlayer structures on Au(111) that have been resolved √ ( 3X through STM include p(3  3) and p(5  5)6,7 as√well as √ √ 3)R30°.5,9 However, it has been argued that the ( 3 X 3)R30° structure is due to the presence of adsorbed bisulfate anions on the surface, and charge measurements indicate a p(3  3) structure.10 Theoretical considerations aside, we have also experimentally verified that there is, at most, a negligible effect of surface step density on the surface alloying kinetics of UPD Ag on Au(111). Mathur et al.48 showed that the defect density of Au(111) can be increased by exposing the surface to concentrated nitric acid for several hours. The result is a surface with many islands of single atom height, which effectively increases the step edge density of the surface. By preroughening our Au(111) single crystal surfaces in concentrated nitric acid and reproducing the surface alloy experiments, an increase in the rate of alloy formation would indicate a step edge alloying pathway, while no change or a decrease in alloying rate would indicate that alloying preferentially occurs at terrace sites as we have previously concluded. The alloying kinetics on rough versus smooth surfaces is shown in Figure 9, where it is clearly seen that there is no significant change in the rate at which the Ag alloys with the surface. (The anomalous point, at t = 10 min, represents a slower alloying rate than for the smooth surface). Surface roughening can also be achieved through repetitive Au oxidation/reduction cycles in dilute H2SO4. Au oxidation/reduction has been shown to dramatically increase the defect density of the surface,49 which in turn should shift the Ag partial ML UPD peak from 0.53 V for the smooth Au(111) surface to 0.61 V vs Ag0/Agþ for the rough Au(111) surface. However, it can be seen in Figure 10 that the peak potential does not shift, but the currents for each of the Ag UPD peaks are significantly decreased. This demonstrates that, while Au oxidation/reduction does in fact roughen the surface, it does not inject vacancies like those created by removing alloyed Ag atoms. In situ STM images during Au oxidation/reduction show that the process results in the formation of small islands and pits several atoms in diameter.49 Self-annealing of the surface, caused by the high surface diffusivity of Au in acidic electrolytes, (48) Mathur, A.; Erlebacher, J. Surf. Sci. 2008, 602, 2863–2875. (49) Nieto, F.; Andreasen, G.; Martins, M.; Castez, F.; Salvarezza, R.; Arvia, A. J. Phys. Chem. B 2003, 107, 11452–11466.

Langmuir 2009, 25(16), 9596–9604

Article

Figure 9. Plot comparing change in ratio of surface Ag charge to alloyed Ag charge with time at 20 °C for clean, smooth Au(111) (square), and Au(111) that has been roughened in concentrated nitric acid (circle).

Figure 10. Ag UPD curve on clean, smooth Au(111) (solid line) and electrochemically roughened Au(111) (dotted line) achieved through repetitive Au oxidation/reduction cycling.

acts to remove vacancies by forming Au islands after the reduction cycle, effectively increasing the step edge density. The obtained surface morphology is similar to that of the HNO3 roughened surfaces and, consequently, has the same negligible effect on the rate at which Ag alloys with the surface. 4.2. Driving Forces for Alloying UPD Ag with Au(111). Driving forces for typical surface alloy formation include stress relief, particularly in systems with a large lattice mismatch, and surface energy reduction. Systems with immiscible constituents, particularly those with a significant lattice mismatch between the adsorbate and substrate atoms, tend to form alloys that are confined to the surface layer. As atoms with a larger or smaller lattice parameter are deposited onto a substrate, a significant stress begins to build, which leaves the surface in a morphologically unstable state. It is due to this film stress that it becomes energetically favorable for the adsorbate atoms to incorporate into the surface in order to relieve the stress. However, the lattice DOI: 10.1021/la9007729

9603

Article

Snyder and Erlebacher

mismatch continues to render the species immiscible in the bulk, confining the alloy to the surface.38,50-53 The surface free energy of the individual constituents as well as the interfacial energy can play an important role in driving surface alloy formation. Siteexchange between adsorbate and substrate atoms is commonly driven by a surface energy difference between the constituent atoms where the surface energy of the adsorbate is larger, driving them into the surface and forcing the substrate atoms out, which tends to either form islands or diffuse to the nearest step edge.54,55 Alloying at the interface during heteroepitaxial growth is readily driven by surface energy reduction and interface stress relaxation; however, this typically occurs only for systems with large lattice mismatches and surface energy differences. In the case of the Ag/Au adsorbate/substrate system, neither stress relaxation nor surface energy reduction can be used to explain the surface alloy formation. The lattice parameter difference between Ag and Au is very small (aAg = 4.09 A˚, aAu = 4.08 A˚),56 therefore, it is unlikely that a significant film stress will arise during heteroepitaxial growth. We can calculate the derived stress due to Ag partial ML deposition onto Au using the relation:57 τ ¼

Y εmf dθ 1 -ν

ð4Þ

where Y is the Young’s modulus for Ag, ν is the Poisson ratio for Ag, εmf is the misfit strain, d is the height of the Ag layer, and θ is the number of MLs. Using eq 4, we find that the stress induced at the surface by adsorption of a third of a ML of Ag onto Au is only 0.030 N m-1. Comparing this value to that of the Co/Au system, which has been reported to form a surface alloy driven by stress relaxation,38 calculated to be ∼16 N m-1, it is clear that the induced stress is not sufficient to drive surface alloy formation for Ag/Au. Surface alloying also cannot be driven by surface energy because the surface energy of Ag is lower than that of Au (γAg=0.50 eV, (50) Tromp, R. M. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 7125– 7127. (51) Tromp, R. M.; Denier van der Gon, A. W.; Reuter, M. C. Phys. Rev. Lett. 1992, 68, 2313–2316. (52) Biener, M.; Biener, J.; Schalek, R.; Friend, C. Surf. Sci. 2005, 594, 221–230. (53) Tersoff, J. Phys. Rev. Lett. 1995, 74, 434–437. (54) Hernan, O.; Vazquez de Parga, A.; Gallego, J.; Miranda, R. Surf. Sci. 1998, 415, 106–121. (55) Fassbender, J.; Allenspach, R.; Durig, U. Surf. Sci. 1997, 383, L742–L748. (56) Roder, H.; Schuster, R.; Brune, H.; Kern, K. Phys. Rev. Lett. 1993, 71, 2086–2089. (57) Ibach, H. Surf. Sci. Rep. 1997, 29, 193–263.

9604 DOI: 10.1021/la9007729

γAu =0.72 eV),58 which would tend to keep Ag on the surface. The surface energies given are reported for close-packed terraces, but this is not the case with a partial ML of Ag which has a p(33) structure.6,7 It is possible that the surface energy for an extended Ag adlayer (granted, a nebulous concept) is higher than that of a close-packed Ag layer and even higher than that of Au, which could lead to incorporation of Ag into the top Au layer. However, there is no experimental or theoretical evidence to suggest that this is the case. It is most likely that the driving force for alloying of Ag UPD with Au is related to the fact that bulk Ag and Au form a uniform solid solution over the entire composition range, a phase diagram that is nearly ideal. There is an entropy gain for dissolving the surface Ag into the bulk approximately equal to ΔS = kB ln(Nb/ Ns), where Nb and Ns are the number of sites in the bulk and surface, respectively, and kB is Boltzmann’s constant. Of course, this driving force is actually driving Ag diffusion into the bulk, but bulk diffusion is slow at near-room temperatures, and therefore, a functional steady-state is obtained as an alloy that is confined to the surface layer.59

5. Conclusions We have presented a technique by which the alloying kinetics of a partial UPD ML of Ag on Au(111) can be electrochemically tracked over time at various temperatures through linear sweep voltammetry. By using a simple site-exchange model for surface alloying, we have measured an activation barrier for the exchange process of 0.40 ( 0.03 eV. Our data suggest that this exchange process occurs on terraces rather than at step edges and is consistent with the structure of the one-third Ag UPD ML, as being uniformly spread over the surface with only minimal lateral interaction between adsorbate atoms. Future work includes in situ monitoring of the Ag surface alloying process using variable temperature, electrochemical STM, and the incorporation of our measured activation barrier into Monte Carlo simulations. Acknowledgment. Funding for this work provided by the NSF under grant DMR-0705525 is gratefully acknowledged. The authors also acknowledge K. Sieradzki and L. Tang for performing AFM and STM. (58) Meyer, J.; Baikie, I.; Kopatzki, E.; Behm, R. Surf. Sci. 1996, 365, L647– L651. (59) Christensen, A.; Ruban, A. V.; Stoltze, P.; Jacobsen, K. W.; Skriver, H. L.; Norskov, J. K.; Besenbacher, F. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 56, 5822–5834.

Langmuir 2009, 25(16), 9596–9604