The extraordinary stability imparted to silver monolayers by chloride

Electrochimica Acta 56 (2011) 1652–1661

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

The extraordinary stability imparted to silver monolayers by chloride Erin V. Iski a , Mahnaz El-Kouedi a , Camilo Calderon b , Feng Wang b , Darin O. Bellisario a , Tao Ye c , E. Charles H. Sykes a,∗ a

Department of Chemistry, Tufts University, 62 Talbot Ave., Medford, MA 02155-5813, United States Department of Chemistry, Boston University, Boston, MA 02215-2521, United States c School of Natural Sciences, University of California, Merced, CA 95343-5001, United States b

a r t i c l e

i n f o

Article history: Received 27 May 2010 Received in revised form 25 October 2010 Accepted 25 October 2010 Available online 4 November 2010 Keywords: EC-STM UPD Cyclic voltammetry Metal halide film Ambient stability

a b s t r a c t While single crystal surfaces facilitate the majority of surface studies, only a handful of these materials are stable under ambient conditions and at extreme temperatures. Therefore, there is a continued interest in the development of robust and ordered surfaces that can be studied under realistic conditions. Electrochemical scanning tunneling microscopy (EC-STM) revealed that in a chloride-free electrolyte, Ag forms an ordered monolayer on Au(1 1 1) with a structure that could be atomically resolved. However, upon removal from the cell, these chloride-free Ag monolayers were subject to degradation by air and high temperatures. Interestingly, if the Ag layer was formed in the presence of chloride, the resulting AgClx layer was stable both in air and at high temperatures. X-ray photoelectron spectroscopy was used to characterize the system and ambient-, low temperature-, and EC-STM revealed that even after exposure to extreme temperatures, the AgClx layer remained intact. Density functional theory (DFT) indicated that the equilibrium coverage of Cl on the Ag monolayer was ∼0.5 ML, and that the barrier for surface reorganization of the overlayer was low. It is proposed that this facile mobility of the overlayer imparts a protective property that allows the AgClx layer to withstand extreme temperatures and attack by oxygen. © 2010 Elsevier Ltd. All rights reserved.

1. Introduction Atomically flat, clean, and ordered surfaces are vital for the detailed study of surface and interface phenomena. Single-crystal metal and semiconductor surfaces, which have well-defined atomic lattices that are continuous with few or no defects, are widely used in surface science. However, in order to analyze most metal and semiconductor surfaces, complex ultrahigh vacuum (UHV) or electrochemical procedures must be used to keep the surfaces clean and ordered at the atomic-level. In contrast to the vast number of surface facets of metals, oxides, and semiconductors available to the UHV surface scientist, only a relatively small set of atomically flat and ordered surfaces exist that remain stable and contamination-free under ambient conditions. A fairly exhaustive list of these substrates includes: highly oriented pyrolytic graphite [1], graphene on SiO2 [2], sulfur on Mo(0 0 1) [3], TaSe2 [4], TaS2 [4], NbSe2 [5], iodine on Pt and Au [6–8], carbon on Ni [9–10], Cr2 O3 [11], and self-assembled thiol monolayers (SAMs) supported on noble metal surfaces [12]. These systems offer many applications in

∗ Corresponding author. Tel.: +1 617 627 3773. E-mail address: [email protected] (E.C.H. Sykes). 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.10.060

self-assembly [13–14], sensing [15], lubrication [16], X-ray optics [17], and corrosion protection [18]. Additionally, silver halide films have shown to be stable at room temperature. Kawasaki and Uchiki reported the formation of AgCl and AgBr layers on thin Au films that were formed via solid phase material transfer [19]. These layers were stable under ambient conditions, however, the thermal stability beyond room temperature of such films was not explored [19]. This paper reports the discovery of an ultra-stable AgClx layer on Au(1 1 1) that is not only atomically ordered and contamination free at ambient pressures and temperatures, but remains intact after heating cycles as high as 1000 K. In order to generate this atomically thin, metal halide film, under potential deposition (UPD) was employed [20–21]. UPD is a useful procedure as it allows for the controllable and reproducible deposition of up to a few atomic layers of a metal before the onset of bulk deposition, thus enabling the study of atomically thin, metal layers which may otherwise be difficult to generate outside the UHV environment [22–25]. The UPD of Ag on Au(1 1 1) has been studied extensively [20–21,26–30]. Gewirth and co-workers carried out one of the first atomic-scale studies on this surface using in situ atomic force microscopy (AFM) [21]. The UPD structure was found to depend on the composition of the electrolyte used in the deposition procedure. In sulfuric acid the Ag adlayer adopted a (3 × 3) overlayer, while in

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

nitrate- and carbonate-containing electrolytes a (4 × 4) overlayer was observed. Although it was not possible to identify the exact structure of the overlayer in perchloric acid, it was determined that the structure was not close-packed like the sulfate and nitrate systems. The same system was studied with electrochemical scanning tunneling microscopy (EC-STM) by Ogaki and Itaya in 1995 [31]. In this study, the authors found that the UPD of Ag occurs in three steps on Au(1 1 1) in both sulfuric and perchloric acid. For the first √ √ adlayer of Ag on Au(1 1 1) in sulfuric acid solution, a ( 3 × 3)R30◦ structure was observed. In the potential range between 0.85 V and 1.10 V vs. RHE, an open (4 × 4) structure of the Ag was formed in the perchloric acid electrolyte. The authors were also able to observe a (1 × 1) close-packed structure at less positive potentials. Furthermore, it was possible to detect the onset of bulk Ag deposition, which ultimately resulted in a Ag(1 1 1)-(1 × 1) structure. Kwak and co-workers also studied this system both with and without chloride in an effort to determine how chloride altered the Ag structures; however, no data on thermal stability was reported [32]. In addition to these studies, many STM experiments (UHV, ambient, and EC) have also been performed on the adsorption of Cl on bulk Ag(1 1 1) crystals [33–37]. Importantly, none of the aforementioned studies describe the high thermal stability observed in the following experiments. In this paper, we describe the UPD of Ag on Au(1 1 1) with and without the presence of chloride using EC-STM, ambient STM, low-temperature STM, and ex situ X-ray photoelectron spectroscopy (XPS). We show how the presence of chloride on the Ag adlayer greatly affects both the structure and the properties of the AgClx film in very unexpected ways. Surprisingly, AgClx films were remarkably stable; they remained intact on the Au surface even after the surface was heated in air to a temperature of 1000 K. Highresolution STM imaging revealed that the atomic-scale structure after flame annealing was actually more perfect than the structure of the as-prepared film. On the other hand, when Cl was not present in the electrolyte solution, the Ag adlayer was not maintained after the high temperature treatment, thus pointing to the presence of chloride as the origin of the film’s stability. The system was modeled using density functional theory (DFT) calculations, which predicted the equilibrium stoichiometry of the surface AgClx layer to be Ag and Cl in a ∼9:5 ratio. DFT calculations also indicated that the barrier for translation of the Cl layer over the Ag monolayer was extremely low, thus pointing to a protective mechanism as the key to the unprecedented stability.

2. Experimental 2.1. EC-STM setup and measurements The working electrode was a Au(1 1 1) single crystal with a 9 mm diameter (MaTecK). The surface was cleaned in pirana solution (1:3 H2 SO4 /H2 O2 ) to remove any organics and boiled in concentrated HNO3 (ACS grade) to remove any excess Ag after every experiment. The sample was rinsed in ultrapure water and cyclic voltammograms (CVs) were performed to ensure sample cleanliness. The electrodes and fluid cell were cleaned in a similar manner. Prior to each experiment, the single crystal was flame annealed in a H2 (g) flame at 1000 K for 2 min, cooled to room temperature, and then quickly covered with a drop of ultrapure water (18.2 M cm, Millipore-Q) to protect the surface against contamination. The electrode was then transferred to the STM electrochemical cell, which was subsequently filled with 0.1 M HClO4 (OPTIMA grade, Fisher). In the electrochemical cell, Pt and Pt/Ir wires were used for the quasi-reference and counter electrodes, respectively. All potentials were subsequently quoted against a Ag/AgCl reference electrode. A PicoScan Molecular Imaging STM was used for all experiments. STM

1653

imaging was performed in situ using etched Pt/Ir wire tips (80/20, 0.25 mm diameter) coated in Apiezon wax to reduce Faradaic currents at the tip/electrolyte interface below 70 pA (Agilent Tech.). All images were recorded using the constant current mode. Prior to each experiment, the condition of the Au(1 1 1)-(1 × 1) surface was verified with CVs and EC-STM imaging. Ag was deposited onto the Au surface through the addition of either 1 mM AgClO4 or 0.1 ␮M saturated AgCl solution to the STM electrochemical cell depending on the type of monolayer to be studied. All solutions were prepared from Sigma ACS grade chemicals and ultrapure water. 2.2. Electrochemical measurements Cyclic voltammetry was performed in an external EC cell using Pt wire mesh as the counter electrode and either Ag wire for the chloride-free data or Ag/AgCl for all other data as the reference electrode. The solutions were purged with N2 (g) prior to running the electrochemical measurements. For all of the Ag deposition experiments, the scan rate was 0.1 V/s. Control of voltage sweeps was performed with a CH Instruments 830 Electrochemical Analyzer. 2.3. UHV STM study The ultrahigh vacuum (UHV) STM experiments were performed in a low-temperature (LT-UHV) microscope built by OmicronNanotechnologyTM . The AgClx layer was first formed electrochemically, then flame annealed in air with a H2 (g) flame for 15 min, and finally transferred directly to the UHV STM (<5 × 10−10 mbar) chamber for imaging. In approximately 30 min, the sample cooled from room temperature to 78 K. All images were recorded using the constant current mode at 78 K with etched W tips, and voltages refer to the sample bias. 2.4. XPS measurements In order to ascertain the chemical composition of the surface, ex situ X-ray photoelectron spectroscopy (XPS) was employed. Two XPS instruments at independent user facilities were used for the XPS experiments. The instruments had load-lock entry, and the UHV analysis chambers were maintained at a base pressure below 5 × 10−10 mbar during experiments. Table 1 shows the relevant details describing the XPS instruments and experiments. For the spectra taken at the MIT facility, a grazing take-off angle of 70◦ was used. For the data taken at Harvard, a take-off angle of 55◦ was used. All peaks were adjusted using a Shirley type background. Accurate binding energies (BE) were determined by referencing to the Au 4f peak at 84.0 eV. Atomic ratios were computed from peak intensity ratios and reported atomic sensitivity factors [38] by the CasaXPS software package (version 2.3.15). 2.5. Computational details DFT calculations were performed using the Vienna Ab inito Simulation Program (VASP) [39–42]. Vanderbilt ultra-soft pseudopotentials (USPP) [43–44] were used to describe the core electrons. Unless otherwise specified, the electron exchange and correlation effect was modeled using the Perdew and Wang 91 (PW91) exchange and correlation functional [45]. The simulation box contained 4 layers of Au with 9 Au atoms in each layer in a 3 × 3 super cell arrangement. On top of the Au layer, one full monolayer of Ag was adsorbed in the face-centered threefold hollow (FCC) positions of the underlying Au surface. The Cl overlayer was modeled in the plane above the Ag layer. The bottom two layers of Au were fixed at their bulk locations, and the z dimen˚ A 2 × 2 × 1 k-point mesh sion of the simulation box was 28.98 A.

1654

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

Table 1 Experimental details of XPS experiments. Facility

Instrument

X-ray source

Detector

Pass energy

Spot size

Harvard CNS MIT CMSE

ESCA SSX-100 Kratos AXIS Ultra

Al K␣ Al K␣

Hemispherical electron energy analyzer Hemispherical electron energy analyzer

150 eV, 25 eV 160 eV, 20 eV

600 ␮m2 400 ␮m × 700 ␮m

CNS = Center of Nanoscale Systems, CMSE = Center for Materials Science and Engineering.

and a kinetic energy cutoff of 300.0 eV were employed. The electron density at the Fermi level was smeared using the method of Methfessel and Paxton [46] with a smearing width of 0.1 eV. All of the calculations were performed with the spin-restricted approach. A few representative configurations were recalculated with spinunrestricted wavefunctions. Identical results were obtained as the converged spin density distribution was always everywhere zero. ˚ Using PW91, the lattice constant for Au was predicted to be 4.183 A, which is in good agreement with the experimental lattice constant of 4.078 A˚ and previously published results [47]. Lattice constants for the other exchange correlation functionals tested in this work are reported in Table S2 of the Supplemental Information. Equilibrium DFT lattice constants are used for all theoretical modeling. Several methods exist that can be used to perform a population analysis with a plane-wave basis set [48–50]. In this work, we chose to use the Bader population analysis method [51] in order to determine the partial charges. Although Bader population analysis is considered to be one of the most rigorous methods for assigning partial charges [52], due to the quantum mechanical nature of electrons, any method of population analysis is somewhat arbitrary in assigning continuous electron densities to atoms. We postulate that

the trend in partial charges when the surface is modified is more trustworthy than the actual values of the partial charges. While LDA, PW91, and PBE calculations predicted different absolute values for the partial charges, the same trend was observed for all of the functionals. 3. Results and discussion 3.1. Experimental STM results 3.1.1. UPD and STM of “chloride-free” layers Fig. 1a shows the topography of the clean Au(1 1 1) surface prior to Ag deposition. As described by Weaver and others, the morphology of the Au(1 1 1) surface could be manipulated through potential control (see Fig. S1) [53–55]. At negative sample potentials, the sur√ face adopted the herringbone or (22 × 3) reconstruction which contains approximately 4.5% more surface atoms than a flat (1 × 1) surface. As the potential was increased, the herringbone reconstruction lifted, and the extra Au atoms expelled formed Au islands. If the positive sample potential was maintained for some time, the Au islands coalesced with one another and the step edges, and a flat

Fig. 1. (a) EC-STM image of clean Au(1 1 1)-(1 × 1) in 0.1 M HClO4 . Tunneling current (It ) = 8 nA, sample potential (Vsample ) = 0.47 V, tip potential (Vtip ) = 0.1 V with respect to Ag/AgCl. (b) CV in 1 mM AgClO4 + 0.1 M HClO4 on the Au(1 1 1) crystal with a scan rate of 5 mV/s. The anodic desorption (A1 , A2 , A3 ) and cathodic adsorption (C1, C2 , C3 ) peaks are labeled. (c) Atomically resolved EC-STM image in a solution containing 1 mM AgClO4 + 0.1 M HClO4 . Imaging conditions: It = 5 nA, Vsample = 0.47 V. (d) Schematic showing the observed (4 × 4) unit cell of Ag on Au(1 1 1) in a chloride-free environment.

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

1655

(1 × 1) surface could be obtained. The flat Au(1 1 1)-(1 × 1) surface in Fig. 1a was obtained by scanning the sample bias from −0.1 V to 0.2 V at a rate of 0.01 V/s in 0.1 M HClO4 . Fig. 1b shows a CV of the UPD of Ag on a Au(1 1 1)-(1 × 1) surface in 1 mM AgClO4 and 0.1 M HClO4 . The peaks in the CV closely resemble those observed by Ogaki and Itaya in 1995 [31]. The CV contains three anodic and three cathodic peaks representing desorption and adsorption, respectively. As the UPD peaks are mostly symmetric, it can be said that the UPD of Ag on Au in a chloride-free environment is reversible [31]. The change in the surface topography as a result of the UPD procedure can be seen in Fig. 1c, which is an atomically resolved EC-STM image of the surface after deposition of Ag at 0.47 V. This image reveals a periodic array of Ag atoms. The distance between each Ag atom is 0.37 ± 0.01 nm. Measurements indicate that the Ag atomic rows are aligned with the atomic rows of the underlying Au lattice, such that the adlayer structure corresponds to a (4 × 4) unit cell, which is depicted schematically in Fig. 1d. This structure is the same as that observed by Itaya and Kwak at the corresponding sample potentials [31,32]. The four corners of the unit cell are Ag atoms adsorbed on atop sites of the Au surface. The open structure indicates the participation of perchlorate anions and/or solvent molecules in the stabilization of the adlayer [31]. As a control experiment for the following work, after the UPD formation of the chloride-free Ag adlayer, the single crystal was removed from the cell into air and annealed in a H2 (g) flame at 1000 K for 15 min as depicted in Fig. 2a. The sample surface temperature during the annealing procedure was monitored with an optical pyrometer. After annealing, the single crystal was returned to the EC cell for imaging. The result of the high temperature treatment can be seen in Fig. 2b. In this case, the previously observed array of Ag atoms was unsurprisingly not maintained after the flame anneal. STM images of the surface revealed depressions indicative of the presence of a more insulating species on the surface. Although speculative, the appearance of insulating species suggests that the Ag atoms may aggregate to form clumps of Agx O during the high temperature treatment as illustrated by the schematic in Fig. 2c. It was found that in all Ag UPD systems in which no chloride ions were present, the Ag monolayers were destroyed by H2 (g) flame annealing as expected. 3.1.2. UPD and STM of “chloride-containing” layers To examine the effect of chloride on the structure and properties of Ag films, the UPD of Ag on Au(1 1 1) in the presence of chloride was investigated. Once again, the Au surface was held under electrochemical control until a flat (1 × 1) structure was observed as seen in Fig. 3a. Fig. 3b shows a CV of such a surface obtained in a 0.1 ␮M saturated AgCl solution in 0.1 M HClO4 . The arrow in Fig. 3b represents the potential at which the AgCl solution was added to the fluid cell (0.47 V). Three desorption and adsorption peaks are present as labeled in the figure. In this case, the large peak separation, specifically between the A2 /C2 and the A3 /C3 peaks, indicates the irreversible behavior of this system. The asymmetric pairs of peaks are indicative of the low solubility of AgCl and the competitive adsorption and desorption of chloride anions on the Ag/Au surface [32]. The coadsorption of anions is a general phenomenon in UPD systems. The UPD of silver reduces the work function of the substrate and renders the surface more positively charged, thus promoting the adsorption of chloride anions [56,57]. Upon the addition of the saturated AgCl solution, the atomic-scale structure of the Au surface was immediately altered, and a representative image is shown in Fig. 3c. EC-STM images revealed a periodic array of atoms covering the surface. In comparison to Fig. 1c in which the unit cell dimensions of the Cl-free structure are 0.37 ± 0.01 nm, the unit cell dimensions in Fig. 3c are 0.29 ± 0.02 nm, thus indicating that a different structure has formed in the presence of Cl. The differ-

Fig. 2. (a) After the UPD of Ag, the sample was removed from the EC cell and annealed in a H2 (g) flame for 15 min at 1000 K. (b) 50 nm × 50 nm EC-STM image of the Au(1 1 1) surface in 0.1 M HClO4 after flame annealing. The arrows point to depressions on the surface which may represent insulating clusters of Agx O. It = 1 nA, Vsample = 0.47 V. (c) Schematic showing how the Ag atoms may aggregate to form Agx O clusters on the surface during the flame annealing procedure.

ence between the adsorbed structures is evident in the STM images when the inter-row spacings are compared for systems prepared under different conditions. In order to determine the stability of the AgClx adlayer, the single crystal was removed from the EC cell and annealed in a H2 (g) flame at 1000 K for 15 min. The sample was then placed back into the EC cell for imaging. Fig. 4a shows an atomically resolved EC-STM image of the layer after flame annealing indicating a well-ordered

1656

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661 Table 2 Chart indicating the percent atomic concentrations and ratio of Cl to Ag present in the first 3–5 layers of the Au(1 1 1) crystal after flame annealing as measured by ex situ XPS. Results are from a representative trial and take account of the relative sensitivity factor (RSF) for each element. Element

% Atomic concentration

Cl/Ag ratio

RSF values

Cl 2p Ag 3d Au 4f

11.5 19.1 69.4

0.6

2.29 18 17.1

and continuous array of atoms. In order to investigate the stability of the post-flame annealed AgClx structure and accurately measure the periodicity of the atomic-scale features of the overlayer, the sample was transferred into vacuum and imaged at 78 K in a LTUHV STM. Fig. 4b shows a representative LT-UHV STM image of the AgClx film after flame annealing in which the structure of the AgClx layer is clearly seen. A larger scale LT-UHV STM image is shown in Fig. S2. The protrusions visible in Fig. 4b form a hexagonally packed structure with a spacing of 0.29 ± 0.02 nm with a 0◦ rotation with respect to the underlying Au lattice. This spacing corresponds to a (1 × 1) unit cell in good agreement with our EC-STM measurements. Surprisingly, after flame annealing the AgClx adlayer was not only preserved on the surface, but it also became more ordered and contained fewer defects, like depressions. The fact that the layer became more ordered after flame annealing is evidenced by the presence of continuous, well-defined atomic rows across the Au surface. (Refer to Fig. S3 for a comparison of the pre-annealed AgClx structure.) 3.2. Ex situ XPS study

Fig. 3. (a) 250 nm × 250 nm EC-STM image of the flat Au(1 1 1)-(1 × 1) surface in 0.1 M HClO4 . It = 8 nA, Vsample = 0.47 V, Vtip = 0.1 V. (b) CV in saturated AgCl solution in 0.1 M HClO4 on the Au(1 1 1) crystal. Scan rate = 0.1 V/s. The arrow indicates the potential at which the AgCl was added to the fluid cell before imaging. (c) Atomically resolved EC-STM image in a solution of saturated AgCl in 0.1 M HClO4 . It = 1 nA, Vsample = 0.47 V, Vtip = 0.1 V.

In an effort to corroborate the existence of the AgClx adlayer after flame annealing, ex situ XPS was performed on a representative group of samples (see Fig. S4 for an example of a survey XP spectrum). This allowed quantification of the elemental composition of the top ∼5 layers of the surface. Data from a typical XP spectrum is summarized in Table 2. Not included in the table are the large amounts of C and O that exist on the surface (with the absolute values being dependent on factors such as exposure to air), which is due to the inevitable contamination of using an ex situ technique. The presence of Ag and Cl is indicated in the high resolution XP spectrum in Fig. 5. In total, the average Cl:Ag ratio observed over multiple scans is 0.9 ± 0.3. The uncertainty in the Cl:Ag ratio arises from the fact that a variety of ordered AgClx structures can be formed electrochemically with slight changes in the UPD potential. An example of a lower density AgClx structure can be seen in Fig. 6. It was determined that each of the ordered structures was stable to elevated temperatures under ambient pressure. While the majority of the experiment and theory in this paper is focused on the simplest (1 × 1) structure, the fact that many different structures/coverages exist explains the wide variance in the XPS data. (See Fig. S5 for more images of the lower density structures.) An important point to mention is that while a control XP spectrum was taken on a “Cl-free” sample, due to the nature of the electrolyte (0.1 M HClO4 ) used in the experiments, the XPS data always shows a residual Cl signal. This is not unexpected as a reactive Ag film would bind strongly to the ions of the electrolyte [31]. However, the Cl:Ag ratio in the control sample was substantially lower than that observed for the “Cl-containing” samples (i.e. c.a. 0.1 vs. 0.9 ± 0.3). Furthermore, the two take-off angles (55◦ and 70◦ ) of the different XPS instruments did not produce significantly different results. The slight differences are also included in the error bar of the Cl:Ag ratio. The procedure of electrochemically forming the AgClx layer on the Au(1 1 1) single crystal, flame annealing at 1000 K, imaging with STM, and then performing XPS analysis was repeated several times.

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

1657

Fig. 4. (a) Following the formation of the Ag adlayer in the presence of Cl, the surface was flame annealed in air in a 1000 K H2 (g) flame for 15 min. Despite the high temperature treatment, the adlayer remained intact on the surface as observed in this atomically resolved EC-STM image. It = 5 nA, Vsample = 0.4 V. (b) Similar images with the same atomic spacing were also observed under UHV conditions, It = 1.6 nA, Vtip = 0.2 V, 78 K.

These results indicate that the structures forming the (1 × 1) unit cell originate from the periodicity of the Ag atoms which are known to grow epitaxially on Au(1 1 1) [31] and that Cl is present in a lower concentration. If the 1 × 1 features originated from Cl, the Cl–Cl distance would be 0.29 nm which is too close to be energetically feasible. The closest reported packing of Cl in the literature is 0.35 ± 0.05 nm [32].

Considering all possible coverages allowed on a 3 × 3 surface, there are a total of 23 unique initial structures if all of the Cl resides on the FCC positions, and there are another 23 possibilities if all of the Cl resides initially on the HCP positions. All 46 initial structures were optimized, and the structure with the lowest configuration energy at each coverage was obtained. The surface formation energy was defined as the potential energy change for the following reaction:

3.3. Density functional theory calculations

Ag9 Au36 (slab) + N/2Cl2 (g) = ClN Ag9 Au36 (slab)

In order to quantify the structure and energetics of the AgClx formation, a computationally manageable 3 × 3 surface periodicity was used for all DFT calculations. The energies of formation of a range of Cl coverages on an epitaxial 1 × 1 layer of Ag on a 4 layer thick Au(1 1 1) slab were calculated. At lower coverages, Cl was most stable at the face-centered cubic threefold hollow (FCC) positions of the Ag surface. The energy of Cl at the hexagonally close-packed threefold hollow (HCP) locations was slightly higher, whereas the energy for the bridge and atop positions were substantially higher [58]. In order to obtain the optimal structure at each coverage, all of the possible initial configurations with Cl at FCC locations and HCP locations were enumerated. With Cl at the FCC or HCP locations, the number of unique initial structures depends on coverage. Whereas there is only one possibility for the 9/9 or 8/9 coverages, 4 possibilities exist for either the 6/9 or 5/9 coverages.

Fig. 7 summarizes the geometries with the lowest energy of formation. While adding the first ½ Cl2 to the 3 × 3 surface is exothermic by 1.72 eV, the process becomes endothermic with a value of 0.63 eV when ½ Cl2 is added to a surface with 5/9 coverage. Thus, at 0 K, a 5/9 Cl/Ag ratio is thermodynamically most stable. The average Cl–Cl distance at this coverage is 0.38 nm. If entropy is taken into consideration, a slightly lower coverage may become more stable at room temperature. Since the XPS data showed some variability of the Cl/Ag ratio, only a rough estimate of the optimal surface coverage was needed, and it was decided not to carry out computationally expensive phonon calculations to estimate the entropy. This is further justified because the restriction of a 3 × 3 surface periodicity is likely to introduce an error to the estimate at a magnitude equal to or greater than that introduced by the neglect of entropy itself. The reliability of the PW91 exchange correlation

Fig. 5. High resolution XP spectra of Ag and Cl with peak fits indicating their presence in the first 3–5 layers of the surface.

1658

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

Fig. 6. Larger-scale EC-STM images of a lower density AgClx structure. Atomic spacings are 0.33 ± 0.01 nm. The potential at which these structures were formed was more positive than the potential used to form the (1 × 1) structure, leading to a lower coverage of Ag being deposited and the formation of a lower density structure. (a) It = 7 nA, Vsample = 0.1 V, Vtip = −0.05 V. (b) I = 2 nA, Vsample = 0.1 V, Vtip = −0.05 V. Inset shows atomic resolution of the structure, I = 5 nA, Vsample = 0.4 V, Vtip = −0.05 V.

functional was verified using the Perdew, Burke, Ernzerhof (PBE) [59] and Local Density Approximation (LDA) functionals [40,41]. Although slightly different energies of formation were obtained, all three functionals confirmed the 5/9 coverage to be most stable at 0 K. The lattices constants, binding energies, and Cl–Cl distances calculated using both PBE and LDA functionals are reported in Table S2. We note that at about half monolayer coverage, the true Ag–Cl binding energy could be rather different from what is predicted by any of these functionals. We only want to establish that all of the functionals tested predict the most stable surface coverage to be 5/9 of a monolayer under the 3 × 3 surface super-cell constraint. This indicates that the experimental coverage should be close to half of a monolayer. In order to rule out the possibility that the system consisted of a layer of Cl sandwiched by bulk Au and a monolayer of Ag, a series of Au–Cl–Ag configurations were investigated. The configuration energies of the Au–Cl–Ag arrangements were approximately

Fig. 7. DFT results for the lowest energy configurations for each Cl coverage on an epitaxial 1 × 1 layer of Ag on a 4 layer thick Au(1 1 1) slab with a 3 × 3 surface periodicity. The number below each configuration is the energy of formation. The most stable configuration has Cl and Ag in a ratio of 5/9.

0.8 eV per Cl higher than the corresponding Au–Ag–Cl arrangements according to PW91. The same trend is also observed for LDA and PBE. Thus, it is very unlikely that the Cl layer is sandwiched by Au and Ag. Furthermore, we find that the ratio of Cl to Ag is higher when the XPS is taken at a grazing angle (70◦ ). This supports the hypothesis that Cl is in the topmost layer. The possibility of Cl and Ag both adsorbed in the same plane on top of the Au surface in the FCC positions forming a 2 × 2 pattern with each Ag atom surrounded by four Cl atoms was also investigated. Geometry minimization using PW91 resulted in a clear difference in the Ag and Cl corrugation, thus it is not likely that such a structure could be mistaken as a 1 × 1 pattern by STM. While the DFT calculations suggest a Cl surface coverage close to half of a monolayer as being the most favorable, the observed experimental corrugation indicates a 1 × 1 surface arrangement, which is postulated to originate from the underlying 1 × 1 Ag layer. Both EC- and LT-UHV STM imaging showed no features with a greater periodicity that could be attributed to Cl. This is not unexpected if the barrier for Cl translocation is small enough that Cl atom motion can occur on a timescale faster than that of STM imaging. In order to quantify the barrier for Cl translocation, the potential energy penalty associated with moving a Cl atom from a FCC hollow toward a nearby HCP hollow was calculated as indicated in Fig. 8. During the potential energy scan, all of the atoms in the top 4 layers of the super cell were allowed to relax. It is clear from Fig. 8 that the potential energy surface is extremely flat. The potential energy penalty for moving a Cl atom by 1.1 A˚ was calculated to be ∼0.03 eV. It is proposed that this potential energy scan represents just one of a vast number of possible pathways through which surface diffusion can take place. The 3 × 3 periodicity enforced by the modeling cell used in these calculations is expected to cause an appreciable overestimation of the barrier due to finite size effects in that other Cl atoms in the cell have less freedom to avoid the moving Cl. Thus, it is reasonable to expect that the potential energy barrier for diffusive motion in the experimental case would be even lower. Even so, the calculated penalty for diffusive motion is comparable to the thermal energy (kT) at room temperature, 0.026 eV. The Cl atoms are therefore able to diffuse to the extent that they are not imaged with the STM. Even at 78 K, with an assumed attempt frequency of 1012 Hz, the Cl hop rate would be ∼1010 Hz. 3.4. Comparison and correlation of experimental and theoretical data Both EC- and UHV-STM measurements indicate that a AgClx structure with a 1 × 1 periodicity was present on the Au surface

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

1659

Fig. 8. The potential energy profile associated with the translocation of one Cl atom in the AgCl5/9 overlayer. The upper panel depicts a trial diffusive pathway associated with moving a single Cl atom (marked with an arrow) from the FCC hollow (left) to the HCP hollow (right). The lower panel reports the potential energy associated with such a translocation. The x-axis is the displacement of the Cl atom with respect to the FCC initial location.

under ambient conditions and even after being subjected to temperatures up to 1000 K. The XPS studies reveal the presence of both Ag and Cl in the top layers of the ultra-stable samples and that Cl was present in a lower concentration than Ag. According to our DFT calculations, the surface formation energy for the 5/9 arrangement would be −6.29 eV for a slab with Cl adsorbed on a layer of Ag vs. −3.23 eV for a slab with Cl adsorbed on a layer of Au. This indicates that the Ag–Cl interaction is more favorable than the Au–Cl interaction by a factor of 0.61 eV per Cl atom (i.e. slab energy difference/5 Cl atoms). This is consistent with previous theoretical studies indicating that Cl binds stronger to bulk Ag than to bulk Au [58]. Therefore, it is postulated that the stronger Ag–Cl interactions prevent the Ag atoms from mixing into the Au layer at elevated temperatures as would be expected given that the two metals are completely soluble. This conclusion is supported by experiments which have shown that when a thin Au film (film thickness ∼5 nm) on a Ag(1 1 1) single crystal was treated with dilute KCl, a AgCl monolayer formed on the surface as a result of a material transfer of Ag through the Au film to the solid/liquid interface [19]. This result clearly indicates the preferred Ag–Cl interaction over Au–Cl interaction. Additional experimental observations have yielded similar results indicating that a Ag–Cl bond is stronger than a Au–Cl bond [60]. Specifically, the dissociation energy of Au–Cl is ∼20 kcal/mol lower than the dissociation energy of a Ag–Cl bond [61]. In terms of resistance to oxidation, our DFT calculations indicate that the barrier to Cl diffusion above the Ag layer is very low. This result is consistent with the work of other groups who have also

reported that Cl can have a low diffusion barrier on (1 1 1) metals [32,60]. Furthermore, Cl has a significantly higher electronegativity than Ag. Thus, the Ag–Cl bond is highly polarized. As the average partial charge on the Cl atoms is −0.44e0 (e0 = elementary charge unit), there is a strong electrostatic repulsion between neighboring Cl atoms (see Table S1). Surface bound Cl atoms have never been reported closer than 0.35 ± 0.05 nm apart [32]. Our hypothesis is that it is the repulsive interactions between the Cl atoms with negative partial charges that force them to spread out on the surface. This, coupled with the low barrier to diffusion, leads to the stability of the AgClx system. The gaps left in the 0.5 ML coverage of Cl on Ag (∼0.4 nm) are not sufficiently large to allow oxygen or other contaminants to react with the Ag layer. The fact that the Cl atoms repel each other and can move easily to cover any vacancies in the layer leads to the protective character of the film. The thermal stability of the AgClx layers reported here is surprising, especially when compared to the sublimation temperature of bulk AgCl in UHV at ∼670 K and the desorption temperature of AgCl from Ag(1 1 1) in UHV at 750 K [62]. Our XPS measurements and DFT calculations, however, demonstrate that the Cl/Ag ratio is not unity in this system as it is in bulk AgCl. Since the Cl is adsorbed on a single layer of Ag, it is not expected that 1 ML of Ag would behave the same as the bulk. While the layer is supported on Au, differences in the work function between Ag and Au lead to a different interaction strength with Cl. The average partial charge of the Ag atoms in the proposed Ag/Cl 9:5 structure is 0.31e0 . On the other hand, the Ag atoms in the topmost layer will have an average

1660

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661

partial charge of 0.23e0 , if the same surface arrangement forms on a bulk Ag substrate. It is proposed that a combination of the fact that Cl bonds more strongly to Ag than to Au, and that Cl can diffuse over Ag monolayers with a very low barrier gives rise to the unprecedented thermal stability. This is a case in which the surface properties of a 2D material are substantially different from those of the bulk properties. This atomically thin and ultra-stable layer which is also resistant to oxidation may find applications in a variety of fields. Due to their ease of preparation and well-defined nature, substrates like Au/mica and graphite (HOPG) are ubiquitous to fields like molecular self-assembly, molecular electronics, sensing, and surface catalysis. AgClx films grown on inexpensive Au/mica substrates may offer a similarly well-defined surface with different properties. It may also be possible that this highly stable, protective AgClx film could be used for select anti-corrosion applications in which a thin, stable coating of a metal is required. 4. Conclusions In situ EC-, ambient-, and LT-UHV STM combined with ex situ XPS were used to investigate the UPD of Ag on a Au(1 1 1)-(1 × 1) surface with and without the presence of chloride. When the Ag adlayer was formed in a chloride-free solution, the structure was not maintained after a high temperature treatment in a H2 (g) flame. Surprisingly, when the UPD of Ag took place in a saturated AgCl solution, the AgClx adlayer was maintained after flame annealing at 1000 K for 15 min and could be imaged under solution, in air, and under LT-UHV conditions. The presence of AgClx after flame annealing was confirmed through the use of ex situ XPS, which indicated that both Ag and Cl were present on the post flame-annealed Au crystal. DFT calculations revealed that the equilibrium surface coverage of Cl was close to half of a monolayer (x ∼5/9). Calculations also indicated a very low diffusion barrier for the Cl on the Ag adlayer, which allows facile diffusion of the Cl and enables a protective property through which the AgClx film can resist oxidative or thermal damage. Supplementary data EC-STM images of the potential control of the Au(1 1 1) sur√ face 22 × 3 reconstruction, larger-scale UHV STM images of the post-annealed AgClx surface, large-scale EC-STM images of the preannealed AgClx surface, and EC-STM images of the lower density AgClx structure are included. A survey XP spectrum and a table of the Bader charges for the top 3 atomic layers for the most stable Ag/Cl 9/5 surface arrangement is also included. The lattices constants, binding energies, and Cl–Cl distances calculated using both PBE and LDA functionals are reported as well. Acknowledgements ECHS thanks the Research Corporation, the NSF (Grant 0717978), and the Beckman Foundation for support of this research. EVI thanks the DOEd for a GAANN fellowship. TY thanks the support from UC Merced and the Petroleum Research Foundation (Grant 48335-G5). CC and FW want to thank the National Center for Supercomputing Applications for allocation TG-CHE070060 and the generous computer allocation from the scientific computing and visualization center at Boston University. The XPS work was performed in part at the Center for Nanoscale Systems (CNS), a member of the National Nanotechnology Infrastructure Network (NNIN), which is supported by the National Science Foundation under NSF award no. ECS-0335765. CNS is part of the Faculty of Arts and Sciences at Harvard University. The XPS work, per-

formed at MIT, was supported in part by the MRSEC Program of the National Science Foundation under award number DMR0819762 Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.electacta.2010.10.060. References [1] S.I. Park, C.F. Quate, Appl. Phys. Lett. 48 (1986) 112. [2] E. Stolyarova, D. Stolyarov, L. Liu, K.T. Rim, Y. Zhang, M. Han, M. Hybersten, P. Kim, G. Flynn, J. Phys. Chem. C 112 (2008) 6681. [3] B. Marchon, P. Bernhardt, M.E. Bussell, G.A. Somorjai, M. Salmeron, W. Siekhaus, Phys. Rev. Lett. 60 (1988) 1166. [4] R.V. Coleman, B. Drake, P.K. Hansma, G. Slough, Phys. Rev. Lett. 55 (1985) 394. [5] K. Kajimura, H. Bando, K. Endo, W. Mizutani, H. Murakami, M. Okano, S. Okayama, M. Ono, Y. Ono, H. Tokumoto, F. Sakai, K. Watanabe, S. Wakiyama, Surf. Sci. 181 (1987) 165. [6] B.C. Schardt, S.L. Yau, F. Rinaldi, Science 243 (1989) 1050. [7] R.L. McCarley, A.J. Bard, J. Phys. Chem. 95 (1991) 9618. [8] W. Haiss, J.K. Sass, X. Gao, M.J. Weaver, Surf. Sci. 274 (1992) L593. [9] S. Harada, M. Goda, J. Vac. Sci. Technol. A 8 (1990) 308. [10] S. Yamazaki, M. Tanaka, S. Tanaka, M. Fuginami, R. Uemori, D. Fujita, T. Homma, M. Ono, J. Vac. Sci. Technol. B 9 (1991) 883. [11] N.M.D. Brown, H.X. You, Surf. Sci. 233 (1990) 317. [12] C.A. Widrig, C.A. Alves, M.D. Porter, J. Am. Chem. Soc. 113 (1991) 2805. [13] Z.J. Donhauser, B.A. Mantooth, K.F. Kelly, L.A. Bumm, J.D. Monnell, J.J. Stapleton, D.W. Price, A.M. Rawlett, D.L. Allara, J.M. Tour, P.S. Weiss, Science 292 (2001) 2303. [14] A. Hatzor, P.S. Weiss, Science 291 (2001) 1019. [15] S. Flink, F. van Veggel, D.N. Reinhoudt, Adv. Mater. 12 (2000) 1315. [16] X.D. Xiao, J. Hu, D.H. Charych, M. Salmeron, Langmuir 12 (1996) 235. [17] I.G. Grigorieva, A.A. Antonov, X-Ray Spectrom. 32 (2003) 64. [18] K. Itaya, Prog. Surf. Sci. 58 (1998) 121. [19] M. Kawasaki, H. Uchiki, Chem. Phys. Lett. 254 (1996) 98. [20] R. Michalitsch, B.J. Palmer, P.E. Laibinis, Langmuir 16 (2000) 6533. [21] C.H. Chen, S.M. Vesecky, A.A. Gewirth, J. Am. Chem. Soc. 114 (1992) 451. [22] Y.G. Kim, J.Y. Kim, C. Thambidurai, J.L. Stickney, Langmuir 23 (2007) 2539. [23] A.A. Gewirth, B.K. Niece, Chem. Rev. 97 (1997) 1129. [24] E. Herrero, L.J. Buller, H.D. Abruna, Chem. Rev. 101 (2001) 1897. [25] O.M. Magnussen, Chem. Rev. 102 (2002) 679. [26] L.B. Rogers, D.P. Krause, J.C. Griess, D.B. Ehrlinger, J. Electrochem. Soc. 95 (1949) 33. [27] L.B. Rogers, ACS Symp. Ser. 390 (1989) 396. [28] M.J. Esplandiu, M.A. Schneeweiss, D.M. Kolb, Phys Chem. Chem. Phys. 1 (1999) 4847. [29] S. Garcia, D. Salinas, C. Mayer, E. Schmidt, G. Staikov, W.J. Lorenz, Electrochim. Acta 43 (1998) 3007. [30] T. Kondo, J. Morita, M. Okamura, T. Saito, K. Uosaki, J. Electroanal. Chem. 532 (2002) 201. [31] K. Ogaki, K. Itaya, Electrochim. Acta 40 (1995) 1249. [32] J. Lee, I. Oh, S. Hwang, J. Kwak, Langmuir 18 (2002) 8025. [33] B.V. Andryushechkin, K.N. Eltsov, V.M. Shevlyuga, V.Y. Yurov, Surf. Sci. 407 (1998) L633. [34] J.H. Schott, H.S. White, J. Phys. Chem. 98 (1994) 291. [35] J.H. Schott, H.S. White, Langmuir 10 (1994) 486. [36] M. Kawasaki, H. Ishii, Langmuir 11 (1995) 832. [37] G. Aloisi, A.M. Funtikov, T. Will, J. Electroanal. Chem. 370 (1994) 297. [38] C.D. Wagner, L.E. Davis, M.V. Zeller, J.A. Taylor, R.H. Raymond, L.H. Gale, Surf. Interface Anal. 3 (1981) 211. [39] G. Kresse, J. Furthmuller, Phys. Rev. B 54 (1996) 11169. [40] G. Kresse, J. Furthmuller, Comput. Mater. Sci. 6 (1996) 15. [41] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251. [42] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558. [43] G. Kresse, J. Hafner, J. Phys. Condens. Matter 6 (1994) 8245. [44] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892. [45] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C. Fiolhais, Phys. Rev. B 46 (1992) 6671. [46] M. Methfessel, A.T. Paxton, Phys. Rev. B 40 (1989) 3616. [47] M. Mavrikakis, P. Stoltze, J.K. Norskov, Catal. Lett. 64 (2000) 101. [48] M.D. Segall, C.J. Pickard, R. Shah, M.C. Payne, Mol. Phys. 89 (1996) 571. [49] D. Sanchezportal, E. Artacho, J.M. Soler, Solid State Commun. 95 (1995) 685. [50] W.A. Harrison, Electronic Structure and the Properties of Solids, The Physics of the Chemical Bond Dover Publications, 1989. [51] R.F.W. Bader, Atoms in Molecules: A Quantum Theory, Clarendon, Oxford, 1990. [52] F. Jensen, Introduction to Computational Chemistry, Wiley, 1999. [53] X.P. Gao, A. Hamelin, M.J. Weaver, J. Chem. Phys. 95 (1991) 6993. [54] A. Hamelin, J. Electroanal. Chem. 407 (1996) 1. [55] T. Dretschkow, A.S. Dakkouri, T. Wandlowski, Langmuir 13 (1997) 2843.

E.V. Iski et al. / Electrochimica Acta 56 (2011) 1652–1661 [56] P. Zelenay, L.M. Rice-Jackson, A. Wieckowski, J. Gawlowski, Surf. Sci. 256 (1991) 253. [57] C. Sanchez, E.P.M. Leiva, Electrochim. Acta 45 (1999) 691. [58] A. Migani, F. Illas, J. Phys. Chem. B 110 (2006) 11894. [59] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.

1661

[60] W.W. Gao, T.A. Baker, L. Zhou, D.S. Pinnaduwage, E. Kaxiras, C.M. Friend, J Am. Chem. Soc. 130 (2008) 3560. [61] G.N. Kastanas, B.E. Koel, Appl. Surf. Sci. 64 (1993) 235. [62] M. Bowker, K.C. Waugh, Surf. Sci. 134 (1983) 639.