Ariba
Adnan
a,
Saeid
Behjati
a,
Núria
Félez-Guerrero
ab,
Kasinath
Ojha
a and
Marc T. M.
Koper
*a
aLeiden Institute of Chemistry, Leiden University, Einsteinweg 55, Leiden 2333 CC, The Netherlands. E-mail: m.koper@chem.leidenuniv.nl
bChemistry Department, Universitat Autònoma de Barcelona, Bellaterra-08193, Barcelona, Spain
First published on 1st August 2024
We examined the electric double-layer (EDL) of a Au(111) electrode in a dilute perchloric acid solution using a combination of capacitance measurements and in situ scanning tunnelling microscopy under electrochemical conditions (ECSTM). The “camel-shaped” capacitance curve of the EDL is studied with different cations and anions, including their impact on the potential of zero charge (PZC). We show that the ECSTM images of thermally reconstructed and of the potential-induced surface reconstruction of Au(111) in perchloric acid electrolyte resemble previous work in sulphuric acid, displaying a herringbone pattern for a thermally reconstructed surface. Once the reconstruction is lifted, the Au(111) forms islands with an average of 1 atomic step height. When the potential is lowered below that of the PZC, the potential-induced surface reconstruction results in a more disoriented pattern than the thermally reconstructed surface. ECSTM images at different potentials are correlated with the voltammogram to understand the time and potential dependence of the surface. This correlation has led to the development of a potential window technique that can be used to reveal the surface structure of Au(111) based on observing the changes in PZC in the voltammogram. This method provides an indirect approach to understanding the surface structure without always relying on ECSTM. From the voltammogram, we also observed that anions (SO42−, CH3SO3−, ClO4−, F−) interact more strongly with the Au(111) surface than the alkali cations. The cation capacitance peak shape does not depend strongly on the identity of the alkali metal cation (Li+, Na+, K+). However, the anion capacitance peak depends strongly on the anion identity. It suggests that some level of specific adsorption cannot be excluded, even for anions that are traditionally not considered to adsorb specifically (perchlorate, fluoride).
Since these metal cations reside in the EDL, the study of the EDL of single crystal gold electrodes is an important topic in fundamental surface electrochemistry. Regarding the low-index facets Au(110), Au(100) and Au(111), a consensus has been reached that their surfaces reconstruct; even Au(111), the most closely packed surface, is known to reconstruct.20,21 However, the surface reconstruction of Au(111) has less pronounced features in the voltammetry than the reconstruction of Au(100)22–26 Hamelin et al. hypothesised that the hysteresis observed in the positive and negative-going scan in double-layer measurements of Au(111)8,14,27 could be due to potential-induced changes in the surface structure, as had been reported for Au(100).22–26 Kolb et al. demonstrated the presence of the potential-induced surface reconstruction in Au(111) by performing ex situ reflection high-energy electron diffraction studies (RHEED). They linked changes in the potential of zero charge (PZC) in the positive and negative-going scans of a voltammogram to the lifting of the reconstruction of Au(111).21
To understand the lifting of the reconstruction and whether or not the potential-induced surface reconstruction exists, and the corresponding kinetics involved, further studies have been done by in situ scanning tunnelling microscopy (STM),24,28–34 atomic force microscopy (AFM),31,35 density functional theory (DFT) calculations,33in situ surface X-ray scattering25,36,37 and by employing structure-sensitive probe reactions.20,21,24,28,38 For Au(100), it is known that the reconstruction lifts due to anion adsorption after applying a sufficiently positive electrode potential. The reconstruction can reappear by applying an electrode potential lower than that of the PZC.28,39–41
For Au(111), the presence of the herringbone reconstruction, the lifting of this reconstruction and the occurrence of a potential-induced reconstruction have been reported previously. However, most of the work related to surface reconstruction has been done in H2SO4, i.e. with specifically adsorbing sulfate anions.28,29,38 There are relatively fewer studies of the Au(111) electrode in perchloric acid, especially concerning surface reconstruction phenomena. Au(111) in perchloric acid is the model electrode–electrolyte interface for double-layer studies and catalytic reactions like HER.18,38,42 It is still unclear to what extent perchloric acid adsorbs on gold and whether this affects the kinetics of lifting of the surface reconstruction.8,13,27,43,44 Many DFT calculations of adsorption on gold single crystals are based on an ideal (111) surface,45 which may not be the case at every potential, as the Au(111) has the tendency to reconstruct, and the atoms on a Au(111) surface are very mobile. Therefore, it is important to know exactly what the structure of the surface is at a given potential, whether or not it changes when in contact with different electrolytes, and how it affects the electrochemical fingerprint in terms of the blank cyclic voltammetry as well as the reactivity for probe reactions.
We show here that the Au(111) electrode in dilute perchloric acid shows the expected “camel shape” of the potential-dependent double-layer capacity,5,35,46–48 with the negative potential peak sensitive to cations and the positive potential peak sensitive to anions.4 However, the exact double-layer capacity and double capacity curve depend on the surface reconstruction, with a small negative shift in PZC for the unreconstructed Au(111). In perchloric acid, the lifting of the thermal reconstruction is faster than the re-emergence of the potential-induced surface reconstruction. No clear transition potential can be identified in either case. Our work provides the basis for future double-layer studies of Au(111) and other Au surfaces at different pH and electrolyte compositions with respect to the types of cations and anions.
A single crystal with a 2 mm thickness and 5 mm diameter (purity 99.9%, Mateck GmbH) was used. It was annealed and then cooled down before adding a droplet of degassed Milli-Q water. Annealing without quenching with water resulted in a very high and distorted capacitance curve in the first cycle of cyclic voltammetry, indicating the presence of impurities from air or the heating flame. It is crucial to emphasise that the Au(111) single crystal should be annealed to a point where it slightly glows, but the red-hot flame should be cooled down before making it come into contact with the water.
Cyclic voltammogram features are very sensitive to the Au(111) single-crystal preparation method when measuring the EDL. The surface of the Au(111) single crystal can be reconstructed under the influence of temperature, or the reconstruction can be induced by applying an electrode potential negative of the PZC. It has been well reported in the literature that the thermally reconstructed structure is the more stable configuration with a reconstruction with compression in [110] direction, which leads to the expulsion of adatoms, leading to islands covering ca. 4% of the surface upon lifting of the reconstruction.21,51
Fig. 2 shows a clear hysteresis between the positive-going and negative-going scans, with a shift in the potential of zero charge (PZC). Hamelin also reported this hysteresis for the Au(111) electrode by observing a shift in the adsorption peak of halide in diluted halide-containing electrolytes (halides are known to adsorb specifically on Au(111) surfaces).52 She suggested that this hysteresis was due to the re-arrangement of the surface atoms and that the potential range (be it negative or positive with respect to the PZC) did not affect this hysteresis unless the potential range where the hysteresis started to occur was eliminated.14,21 Kolb and co-workers performed experiments with Au(111) in 10 mM HClO4 acid. They linked the shift in the PZC to the re-arrangement of atoms, more specifically due to the lifting of reconstruction of the Au(111) surface at potentials positive of the PZC.21
Kolb and Schneider showed that a thermally reconstructed Au(111) surface has a more positive PZC than a surface reconstructed under the influence of negative potential after lifting the thermal reconstruction.21 It is important to note that the lifting of the reconstruction is a scan rate-dependent process. Depending on the scan rate and the positive electrode potential limit chosen, the lifting of the thermal reconstruction can vary. Also, if the single crystal has defects, that can influence the number of voltammetric cycles required before the reconstruction is lifted all over the surface. Fig. S1 (ESI†) shows EC-STM images of a case where the reconstruction of the Au(111) did not get completely lifted homogeneously throughout the whole surface, leaving patches of reconstruction visible, which in the voltammograms in Fig. S2 (ESI†) can be seen by a small peak appearing before the PZC, slowly diminishing in subsequent other cycles. It shows the presence of more than one surface structure with presumably different PZCs.
To confirm the nature of the hysteresis in the cyclic voltammetry, we have performed EC-STM experiments on Au(111) in 1 mM perchloric acid. Fig. 3 summarises the main results. The EDL of Au(111) can be divided into three regions. Starting with a freshly thermally annealed Au(111) crystal (Fig. 3a), the surface of Au(111) is reconstructed, as identified by double rows, that can form a herringbone structure.33,38,53 Once this reconstruction is lifted, at a potential positive of the PZC, the surface is an unreconstructed Au(111) covered with small Au islands caused by Au atoms expelled from the surface by the lifting of the reconstruction (Fig. 3b). The reconstruction reappears by applying or scanning the electrode potential negative of the PZC. The resulting potential-induced surface reconstruction (Fig. 3c) is more disordered compared to a thermal reconstruction (Fig. 3a), and the lines also appear broader than for a thermally reconstructed surface. On the basis of these results, we conclude that the capacitance minimum in the positive-going scan corresponds to the PZC of the reconstructed surface, whereas the capacitance minimum in the negative-going scan corresponds to the PZC of the unreconstructed Au(111) surface. Hence, the PZC of the unreconstructed Au(111) surface is slightly less positive than that of the reconstructed Au(111) surface.
To study the time dependence of the formation of these three different structures in more detail, we have performed EC-STM experiments on Au(111) in 1 mM perchloric acid at different potentials, as shown in Fig. 4. The freshly flame-annealed Au(111) is contacted with the electrolyte at 0.06 V. The surface exhibits the well-known herringbone reconstruction (Fig. 3a and 4a). The structure remains the same when the potential is increased to 0.6 V (Fig. 4b) and then to 0.7 V (Fig. 4c). If the potential is further increased in steps of 0.1 V from 0.7 (PZC) to 0.9 V, the herringbone structure disappears, and the reconstruction starts lifting. The new surface is that of a Au(111) surface with Au islands (Fig. 4d). The reconstruction gets completely lifted once the potential increases to 1.0 V (Fig. 4e). As the potential is lowered in the negative-going scan, the reconstruction takes place, and adatoms from the islands become incorporated again in the surface. This process is inherently slow and yields a reconstructed surface which looks different from a thermally annealed surface. Once the potential is decreased below the PZC, the last remaining Au islands are seen at a potential of ca. 0.65 V. After that, the reconstruction starts to appear, and the newly reconstructed surface can be seen in Fig. 4h at a potential of 0.6 V.
This potential-induced reconstruction (Fig. 3c and 4i, both measured at a potential of 0.06 V) appears more disordered with a non-uniform structure. The formation of this reconstructed surface appears to be kinetically slower than the lifting of the reconstruction, as the reconstruction lifts between 0.7 V and 1 V, whereas a larger potential window is necessary for the electrochemical reconstruction to appear, i.e. between 0.6 to 0.06 V, where the potential was held for the same duration of time at each potential. The surface was also subjected to negative potentials up to −0.1 V (Fig. S3 left, ESI†). The potential was then increased again from –0.1 V to 1 V to lift the electrochemical reconstruction (Fig. S3, right-hand side, ESI†), and then the potential was lowered back to 0.06 V. The potential-induced reconstruction features seem more pronounced under these conditions, but further experiments will be needed to address this in more detail.
To understand the role of the potential window on the surface reconstruction, we have measured the double-layer capacitance curves in so-called “window opening” experiments. Fig. 5a shows the first two cycles of the windowing experiment, starting with a freshly thermally reconstructed Au(111) single crystal. The first cycle has a capacitance minimum at 0.69 V vs. RHE (0.51 vs. SHE), and in cycle 2, the capacitance minimum in the positive-going scan has shifted to a lower value of 0.674 V vs. RHE (0.497 vs. SHE). It is important to note that in the negative-going scan, the capacitance minimum is around 0.64 V. vs. RHE (0.46 vs. SHE); it does not change within the first two cycles.
Fig. 5b shows the changes in the double-layer capacitance and PZC within potential windows with a successively less positive vertex potential. These results show that the PZCs of the potential-induced surface reconstruction remain the same. When one reduces the positive limit of the potential in successive scans, barely any change is observed in the positive-going scan, but for the negative-going scan, as the positive limit of the cycle is lowered, the lifting of the reconstruction is prohibited, and the PZC shifts to more positive potentials, as can be seen in the negative-going scan. Therefore, we conclude that the PZC of the potential-induced surface reconstruction of Au(111) is 0.674 V vs. RHE (0.497 V vs. SHE), whereas the PZC of a Au(111) (with islands resulting from the lifting of the reconstruction) is 0.64 V vs. RHE (0.46 V vs. SHE). When cycling in the region negative of the PZC (Fig. 5b and Fig. S4, ESI†), where the electrode potential stabilizes the reconstructed surface, the scans are symmetric and, hence, reversible.
A capacitance curve of the EDL region in a low electrolyte concentration generally shows a so-called “camel plot”, as illustrated schematically in Fig. 6.4,46 At low electrolyte concentrations, one expects a minimum in the capacitance at the PZC, which should correspond to the differential capacitance predicted by the Gouy–Chapman theory (CdGC).54 Positive and negative of the PZC, there are two peaks corresponding to anions and cations populating the electric double layer, respectively. The intensity of the peaks has been suggested to scale inversely with the size or partial molar volume of the (solvated) ion: the smaller the (solvated) ion, the more ions fit into the double layer, and hence the higher the differential capacitance (peak).4 The potential difference between the peaks depends on the concentration of the anions and cations, as well as on the strength of the interaction of the solvent with the surface: the peaks shift closer together (the cation peak shifting positively, the anion peak shifting negatively) at a higher electrolyte concentration, and the peaks shift further apart when the solvent molecules have a stronger interaction with the interface.4 Alternatively, the capacitance peaks have been attributed to a decrease in the local dielectric constant as a result of the high electric field; specifically, Shin et al. have ascribed the capacitance peak negative of the PZC to this dielectric saturation effect.55 At a sufficiently high electrolyte concentration, the peaks will overlap, and the capacitance curve typically takes a bell-shaped form.46
Fig. 7 shows the capacitance curves in a concentration range from 0.1 mM (pH 4) to 100 mM (pH 1) HClO4. It confirms the transition from a camel-shaped capacitance curve to a bell-shaped capacitance curve with increasing electrolyte concentration.46,48 A hysteresis is observed when comparing the positive and negative-going scans due to the surface reconstruction, as already shown in Fig. 2. If the hysteresis was only due to the difference in the work function between the different surfaces, the variation in the PZC and the anion- and cation-related peak potentials between the positive and negative-going scans should be the same. However, as shown in Fig. 8b, a clear variation is seen in the potential differences of the PZC compared to the anion peak potentials in the positive and negative-going scan in different concentrations of HClO4. The hysteresis in PZC and anion peak increases as the HClO4 concentration decreases. This could be due to the reconstructed and unreconstructed surface interacting differently with interfacial water or with the anion. The cation potential peaks have the smallest difference as the surface is supposed to be reconstructed in the corresponding potential region, in both scan directions. The anion peak at pH 4 clearly stands out from the rest while scanning in both directions. This change could be due to there not being enough perchlorate anions to facilitate the lifting of the reconstruction. The reconstruction gets lifted once the potential increases sufficiently positive, causing an increase in the surface charge. Whether or not perchlorate is or needs to be specifically absorbed to facilitate the lifting of the reconstruction is an issue that we will return to in relation to Fig. 9 and 10. One other aspect can be the onset of adsorbed hydroxyl species, but future experiments in neutral and alkaline media will be needed to elucidate this.
![]() | ||
Fig. 9 Effect of cations and anions on the peak, PZC capacitance and potential. (a) The potential-dependent capacitance of Au(111) from the negative-going scan is shown. (b) Cation peak capacitance, anion peak capacitance and PZC from negative-going scan and (c) positive-going scan. The full cyclic voltammogram is shown in Fig. S5 (ESI†). All curves were measured in 1 mM HClO4 background electrolytes in an Ar gas-saturated solution, and the scan rate was 20 mV s−1. |
Before looking into the Au(111) data with electrolytes containing different anions and cations, it is important to compare the generalized model with a voltammogram of Au(111) in 1 mM perchloric acid and highlight the impact the potential dependent surface structure can have on the cation and anion peaks. Fig. 8a shows that the surface structure influences the potential and capacitance of the anion and cation peaks. Focusing on the region positive of the PZC, the anion peak capacitance in the first two consecutive cycles is the same for both positive and negative-going scans, but the capacitance is significantly lower in the negative-going scan in comparison to the positive-going scan (ca. 61 μF cm−2vs. ca. 80 μF cm−2 peak capacitance). The reason for this discrepancy has likely to do with effect of anions on the (rate of) lifting of the reconstruction. When the lifting of the reconstruction is hindered by lowering the positive vertex potential to 0.8 V, the cation-related peak potential becomes reversible (see Fig. 8b), indicating that the surface structure is the same in both the positive and negative-going scans. However, we note that there is still a small discrepancy in peak capacitance between positive- and negative going scan (ca. 32 μF cm−2vs. ca. 28 μF cm−2), the reason for which remains unresolved.
Compared to the intensity of the anion peak, the capacitance of the cation peak is almost half that of the anion peak, suggesting that the effective molar volume of the cations is quite a bit higher than that of the anions. Fig. 8b highlights the cation peaks in the positive and negative-going scans. In the positive-going scan, when the thermally reconstructed surface (cycle 1) is compared to the potential-induced surface reconstruction (cycle 2), the cation peak is shifted to lower potential on the latter. This shift agrees with the shift in PZC shown in Fig. 5a. Note that the anion peak potential shows no shift in the positive-going scan, because the surface reconstruction is lifted in both cycles. Also, the cation (or proton) peak capacitance is somewhat lower for the potential-induced surface reconstruction for reasons that we have not elucidated. In the negative-going scan, the cation peak potential follows the shift in the PZC, the potential of which increases with less positive vertex potential.
In Fig. 9, we illustrate the effect of different cations on the capacitance against a background electrolyte of 1 mM HClO4. First of all, on the addition of the cations, the shift of the PZC is negligible (at best, it seems to shift very slightly negatively). The capacitance at PZC increases from ca. 10 to ca. 25–30 μF cm−2 by the addition of 8 mM alkali perchlorate. According to the Gouy–Chapman theory, the capacitance at 1 mM and 9 mM should be 7.2 μF cm−2 and 21.6 μF cm−2, respectively. Both the cation peak and the anion peak shift towards the PZC when 8 mM XClO4 (where X: Li, Na, K) is introduced into a solution with 1 mM HClO4, in agreement with the expectation of how the peaks in a camel capacitance plot should change with an increase in electrolyte concentration. Surprisingly, the capacitance of the cation-related peak remains essentially the same for all alkali cations, despite their different hydration radii. This could mean that changes in the interfacial dielectric constant could be a more likely explanation of the cation-related capacitance peak.55 The intensity of anion peak should remain the same with the addition of XClO4 electrolyte. It appears to very slightly decrease in both the positive going scan and negative going scan compared to the cycle with only 1 mM HClO4 background electrolyte present. This is in agreement with Fig. 7 for the anion peak in different pH of HClO4 acid, with the exception of pH 4. The difference with pH 4 is striking and reproducible, as well as somewhat puzzling. Since higher capacitance peaks could signify the effect of specific adsorption (see e.g. the curves for sulfate-containing electrolyte in Fig. 10), it seems that we cannot entirely rule out some level of perchlorate specific adsorption at higher potential. However, whether perchlorate (or anions in general) facilitate the lifting of the reconstruction by stabilizing the surface charge through electrostatic interactions or by some level of specific adsorption at potentials positive of the PZC, is also not fully clear.
Fig. 10 shows the effect of different anions on the capacitance curve in a 1 mM solution with protons as the cation. Fig. 10a shows that the cation (i.e. proton) peak potential and capacitance barely change with the identity of the anion. The anion peak potential and intensity clearly depend on the anion, with the highest capacitance and lowest peak potential for SO42−. The strong effect of sulfate is due to the specific adsorption of sulfate on Au(111), which is well-known.56 The capacitance minimum shifts to lower potential in the order of SO42− < CH3SO3−< ClO4− < F−, with the latter three anions showing only minimal shifts, suggesting their lack of strong specific adsorption, at least around the PZC. For methane–sulfonate, we expect that there could still be some weak specific adsorption.57 If we assume fluoride and perchlorate ions do not adsorb specifically, the higher differential capacitance for HF would suggest that (hydrated) fluoride has a smaller partial molar volume than (hydrated) perchlorate.
Comparing the anion capacitance peaks from Fig. 10b and c (i.e., obtained from the positive-going scan and the negative-going scan, respectively), it is evident that capacitance in the positive-going scan is always substantially higher than in the negative-going scan. From the EC-STM data we know that in the positive-going scan, the lifting of the reconstruction starts to takes place. As mentioned previously, it is generally assumed that this lifting is promoted by the specific adsorption of anions.58 We cannot exclude the specific adsorption of anions on low-coordination sites generated (or even needed) for the lifting of the reconstruction. Combined with the observation that there is a very strong discrepancy in peak capacitance for perchlorate at pH 4 (see Fig. 7), the idea that perchlorates or fluorides might specifically adsorb at higher potentials, perhaps at certain sites, can not be dismissed. Further research would be needed to study by which specific mechanism anions would facilitate the lifting of the reconstruction.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4cp02133a |
This journal is © the Owner Societies 2024 |