Revisiting Optical Reflectance from Au(111) Electrode Surfaces with Combined High-Energy Surface X-ray Diffraction

We have combined high-energy surface X-ray diffraction (HESXRD) with 2D surface optical re ﬂ ectance (2D-SOR) to perform in situ electrochemical measurements of a Au(111) electrode in 0.1 M HClO 4 electrolyte. We show that electrochemically induced changes to Au(111) surface during cyclic voltammetry can be simultaneously observed with 2D-SOR and HESXRD. We discuss how small one atom high 1x1 islands, accommodating excess atoms after the lifting of the surface reconstruction, can lead to discrepancies between the two techniques. The use of HESXRD allows us to simultaneously detect parts of the truncation rods from the (1 × 1) surface termination and the p x √ 3 electrochemically induced surface reconstruction, during cyclic voltammetry. The presence of reconstruction phenomena is shown to not depend on having an ideally prepared surface and can in fact be observed after going to very oxidizing potentials. 2D-SOR can also detect the oxidation of the Au surface, however no oxide peaks are detected in the HESXRD signal, which is evidence that any Au oxide is X-ray amorphous

Understanding the structure of electrodes is important for many potential future energy technologies such as fuel cells and water splitting electrolyzers, where we seek to optimize the: activity, selectivity, cost, and stability of electrocatalysts.In most cases, electrocatalysts are complex materials with multiple phases, surface structures, pores, and particle sizes. 1 Systematic studies of how structure impacts an electrocatalytic reaction can be made by using model electrodes, such as single crystals.One drawback to such an approach is that the known model structure under investigation can restructure and alter under reaction conditions.Therefore, in situ studies of electrocatalysts under reaction conditions are essential for understanding the impact of structure upon catalytic performance. 2lectrochemical techniques such as: cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), Galvano/Potentiostatic, and differential capacitance measurements have been extensively used to understand processes such as: adsorption, reaction rates, interfacial charge, reaction onsets, and catalyst degradation.However, these macroscopic techniques give an average response of the whole electrode and contain little direct information on any structural changes that occur at the sample surface.To characterize atomic scale structural changes at electrode interfaces in situ, tools such as electrochemical scanning tunnelling microscopy (EC-STM) and surface X-ray diffraction (SXRD) have been largely successful.More recently, high-energy surface X-ray diffraction (HESXRD) has been shown to allow more rapid mapping of reciprocal space under reaction conditions, 3,4 and has also been combined with electrochemistry to study oxide formation on Pt single crystals. 5][13][14] It is well-known that the topmost atomic surface layer of Au(111) can reconstruct.Where the atoms are slightly compressed, such that the atoms in the center of the unit cell sit closer to the hexagonallyclosed-packed (HCP) sites instead of the face-centered-cubic (FCC) sites with a ∼22 × √3 unit cell, this produces "stripes" of raised atoms.The reconstruction is often referred to as the herringbone reconstruction (HB) because there are three rotated domains, and the stripes from the different domains produce a herringbone-like motif that is visible in STM measurements. 15In SXRD the HB is observable as super lattice rods (SLRs) forming a hexagon pattern around the crystal truncation rods (CTRs). 162][13] However, in the electrochemical environment there is also some variability in the length of the unit cell (p), which has been reported to be between 22.5 and 27, depending on: the sample used, electrolyte, and the preparation conditions. 12,17At more positive potentials, the reconstruction is "lifted" and only the 1 × 1 CTRs are observable.The stability of the surface reconstruction is strongly linked to the surface charge on the Au atoms, where the lifting is thought to be caused by charge transfer from anion adsorption (i.e.specific adsorption). 7,129][20] The thermal-induced reconstruction (i.e. from flame annealing) lifts at higher potentials whereas the more defect rich potential-induced reconstruction lifts at slightly lower potentials. 20,21n the 1980s Kolb and Schneider measured phase transitions on Au (100) using "electroreflectance", 21 a term that seems to have fallen out of fashion.In this technique a rapid-scan spectrometer was used to measure the reflectance of linearly polarized light as a function of different wavelengths.However, they were unable to detect substantial differences between the Au(111)−1 × 1 and Au(111)-p × √3 structures.In contrast, in situ second harmonic generation (SHG) was able to detect a surface transition when going to more anodic potentials, since rotation of the sample yields different responses for the 1x1 and p x √3 terminations, due to the differences in symmetry. 22Then in the 1990s, SXRD was used to directly confirm the presence of the HB in electrochemical conditions-only the potentials did not agree with the SHG measurements.The CV and SXRD suggested that the reconstruction lifts over a narrow range, z E-mail: gary.harlow@sljus.lu.se *Electrochemical Society Member.
Journal of The Electrochemical Society, 2021 168 096511 whereas the SHG indicated that the process occurred over several hundred mV, calling into question the interpretation of the SHG results.
2D Surface Optical Reflectance (2D-SOR) is an operando technique that can follow surface morphology changes during reactions in gaseous conditions, [23][24][25][26][27] giving a 2D map of the reflected light from a surface.We recently combined it with an electrochemical cell that could also be used for SXRD. 28In this paper, 2D-SOR has been combined with HESXRD.The simultaneous combination allows us to correlate changes in the measured 2D-SOR intensity with surface sensitive changes in the diffraction pattern, such as the lifting of the HB reconstruction.We find that 2D-SOR is sensitive to the increased density of the HB reconstruction, resulting in a maximum optical reflectivity at cathodic potentials.By combining our measurement with high-energy SXRD we are simultaneously sensitive to both the CTR intensity and the HB intensity, providing complementary information about the lifting process.To understand the relationship between the intensities of the CTRs and the HB we simulated models with and without the HB reconstruction, which demonstrate that the decrease in the CTR intensities is due to the mixed fcc and hcp sites in the top-most layer of the HB reconstruction.As reported previously 28 we also confirm that the oxidation and reduction peaks in the CV are closely correlated to a change of the 2D-SOR signal, we show this corresponds to a reduction in the CTR intensity.

Experimental
The experiments were conducted at beamline P21.2 at PETRA III, DESY, Hamburg, Germany, and the experimental setup is illustrated in Fig. 1.The HESXRD patterns were detected with two 432 × 432 mm Perkin-Elmer flat panel X-ray detectors, constructed of CsI scintillators on amorphous silicon photo diodes.These detectors are sensitive to X-rays with an energy above 20 keV and have a resolution of 2880 × 2880 pixels, with a pixel size of 150 × 150 μm.To protect the detector from the high intensity Bragg reflections of the Au sample, the positions of the Bragg reflections on the detector were covered by circular beam stops.This gives rise to the circular black spots in the detector images, shown in Figs. 1  and 3.During the experiments, the X-rays impinged onto the sample at a grazing incidence angle of 0.0638°with an energy of 70.3 keV.
The electrochemical cell used was a modified version of a previously designed electrochemical PEEK cell, 29 in which the top-most part has been replaced by a glass (fused silica) window.The cell design and modifications are described in previous papers. 28,30The glass window enables the imaging of the surface of the sample with 2D-SOR.The 2D-SOR technique consists of a setup including: a LED, lenses, apertures, and a CMOS camera (illustrated in Fig. 1).The LED light is reflected by the Au(111) sample and images are captured with a CMOS camera through a series of lenses.The camera collects a 2D image of the entire sample surface.Our 2D-SOR setup has been presented in previous papers, [25][26][27][28] and recently we developed a 2D-SOR microscope. 31he single crystal Au(111) working electrode, purchased from MaTeck, was polished to 0.01 μm and had a miscut of less than 0.1°.A gold rod (diameter 2 mm, length 5 cm) was used as the counter electrode and an Ag/AgCl electrode (eDAQ, ET072) as the reference electrode.All presented potentials are reported against RHE (where 0 V = −2.05V Ag/AgCl).The CV measurements were performed using a potentiostat (Metrohm Autolab PGSTAT204).The Au(111) single crystal was prepared through flame annealing with a butane flame, followed by cycling the potential between 0.0 V and 0.9 V in the 0.1 M HClO 4 electrolyte, a procedure known to produce a clean surface with large terraces. 19,21However, due to previous experiments with the crystal, where very positive potentials were applied, the crystal is not perfect and the surface has been roughened, which is indicated by the presence of powder rings in the HESXRD pattern.The glassware used for the experiment were cleaned with a Piranha solution (3:1 98% H 2 SO 4 + 30% H 2 O 2 ), while the electrochemical PEEK cell was cleaned in 70% HNO 3 .The glassware and the cell were then rinsed 5 times in ultra-pure water (18.2MΩ.cm, PURELAB Chorus 1, Elga labwater), after which they were boiled in ultra-pure water.After the boiling they were once more rinsed 5 times in ultra-pure water before use.The 0.1 M HClO 4 electrolyte solution was prepared from 99.999% trace metal HClO 4 (Sigma-Aldrich).The electrolyte was deoxygenated and bubbled with 5 N argon during the whole experiment.The PEEK cell was not completely gas tight in this experiment and there is a small oxygen reduction current wave in the CV.The cell design also exposes the sides of the sample to the electrolyte, which means the CV has contributions from has different facets than the (111) oriented surface.However, the HESXRD and 2D-SOR measurements are independent of this.A fixed rotational angle was chosen for the HESXRD measurement, such that parts of both a CTR from the substrate and the SLR for the surface reconstruction could be detected in the HESXRD pattern simultaneously.

Results
The potential was cycled between 0.00 V and 1.95 V in 0.1 M HClO 4 at scan rate of 10 mV s −1 , while 2D-SOR and HESXRD images were recorded continuously.Figure 2a shows the HESXRD pattern registered on one of the two detectors, the image shows the maximum intensity each pixel registered across a whole CV cycle.A magnified section of the detector, indicated by the white box, is shown in Fig. 2b.The intensity of the CTR and the SLR were extracted by summing the pixels within the respective regions of interest (ROIs).Figure 2c shows the 2D-SOR image from the Au (111) crystal surface during the experiment, where the blue ROI marks the area used for extracting any intensity changes in the 2D-SOR ROI during the CV measurements.The darker straight line across the surface is from the interaction of the X-ray beam with the sample, so called "beam-damage."Supplementary Fig. S1a (available online at stacks.iop.org/JES/168/096511/mmedia)shows a 2D-SOR measurment of the sample.The three ROIs indicate the beam on the sample (black), earlier beam damage (green) and a region with no beam damage (blue).The beam damage appears as a straight black line and the X-ray beam on the sample appears as a brighter line across the sample.In Fig. S1b the intensity integrated from each separate ROI is plotted as a function of applied potential.The general trend of the SOR intensity is similar for all three regions as the CV is scanned, with clear features matching those we observe later (see Fig. 4), correlating to the surface (HB) reconstruction, oxidation and reduction of the sample.The X-ray beam damage is is a straight line across the sample because the sample was held still at a constant position, where both the SLR and HB were visible in the HESXRD image, during the whole measurement.The earlier beam damage is due to previous experiments where the X-ray beam hit the sample in that direction.All three regions seem to show similar features, except in the long term behaviour where the reflectance seems to grow or reduce after a number of cycles.The reason for this increase in reflectance could be the beam inducing changes in the surface that lead to higher reflectance, for example promoting the redeposition of dissolved gold.Since the features are reproducicable and we see no significant difference in the HESXRD results between the earlier cycles (where all the areas of the sample had similar SOR signals) and the later cycles where they diverge, we assume that our HESXRD results are valid in this case.We note that in subsequent (not yet published) experiments we have been able to significantly mitigate this beam damage problem by using attenuators, fast shutters and rotating the sample quickly.A useful feature of the 2D-SOR camera, is therefore that beam-damage can be detected and hopefully mitigated.The larger circular features at the edges of the image are due to bubble formation during the CV, as we go to oxygen evolution reaction (OER) potentials.
A representative CV curve from one of middle cycles is shown in Fig. 3a.Position 1 is the cathodic limit of the CV scan where there is a small ORR current, 2 is the surface oxidation peak, 3 is the oxygen   The pattern is a superposition of the maximum intensity registered for each pixel during a CV between 0 V and 1.95 at 10 mV s −1 .The Bragg reflections (labelled in hexagonal surface units 14 ) are blocked by beam stoppers.The bright spot between (012) and ( 105) is likely a spot in a powder ring and not part of the CTR.Note: the no orientation matrix was generated so the Bragg peaks could be equally indexed as their symmetry equivalents.b) A magnification of the area indicated with a white box in a), in this area the CTR and SLR signal is detected, depending on the applied potential.The white ROI shows the area used to extract the intensity for the CTR and the red ROI is used for the intensity of the SLR, in Fig. 4. c) The 2D-SOR image of the Au(111) single crystal surface, measured at 0 V.The blue ROI is used to extract the 2D-SOR signal in Fig. 4. The dark stripe is a result of beam-damage and is discussed in the SI.} matches the CV where electrochemical processes that significantly alter the surface occur. 28The numbered features are correlated to simultaneously measured 2D-SOR and HESXRD images in Fig. 3b.To enhance the changes in the signals, the 2D-SOR images are normalized (divided by) to the image at 0.00 V and the HESXRD measurements have been normalized to the HESXRD image at 1.95 V, where the CTR and SLR both have a low intensity.In Fig. 3b, the HESXRD image 1 shows the SLR from the HB reconstruction, which demonstrates that the Au(111) is reconstructed, despite the sample having previously been oxidized and there being some beam damage.At the same time there is a maximum reflectance in the 2D-SOR, either due to the higher density of the reconstructed surface or its increased flatness.At the positions 2 and 4, corresponding to the Au oxidation and reduction peaks in the CV, only the CTR signal in the HESXRD can be detected.This clearly shows that at these potentials the HB surface reconstruction is lifted, and the unreconstructed (1 × 1) surface is present.At position 3 of the CV, where the current corresponds to the oxygen evolution reaction, there is neither a CTR nor a SLR signal present in the HESXRD image, at the same time the 2D-SOR reflectance is at its lowest.This must be because the surface of the Au(111) electrode is roughened at these potentials.
The trends of the 2D-SOR, CTR, SLR signals, and current as a function of measurement time for 5 cycles are plotted in Fig. 4a, showing clearly reproducible behavior and significant correlation between the 2D-SOR and the HB reconstruction.Figure 4b shows the same data (averaged across all cycles) as a function of potential.In the top panel the intensity of the CTR is compared to the CV current, and shows a clear increase in the CTR intensity around 0.55 V. islands are assumed to form on the surface, since the HB is lifting the excess 3%-4% atoms have been shown to form small one atom high 1 × 1 islands. 7Then a (1 × 1) surface is formed, removing the destructive interference of HB and increasing the CTR intensity.The increaseAt this potential d distance between dislocations also means more of the surface is (1 × 1) terminated. 7There is also a sudden drop in the CTR intensity around 1.5 V where the initial stages of Au oxidation occur.The middle panel directly shows the intensity of the SLR due to the surface reconstruction, and this sharply decreases around 0.55 V. However in the bottom panel, which shows the 2D-SOR signal, there seems to be a more gradual transition when the surface reconstruction lifts, it also decreases during Au oxidation.The lifting as observed by the HESXRD then occurs around 0.55 V which is consistent with a the sharp peak typically observed in cyclic voltammograms at the same potential.At this potential STM studies showed complete change in the HB reconstruction and new more complicated dislocation network was found. 7 model of the HB reconstruction was constructed and is shown in the inset in Fig. 5a with a (22 × √3) unit cell assumed.Here, the well-known striped features from the fcc-hcp transition can be seen.The structure is approximated by a transition from a normal fcc terminated domain and a faulted domain with atoms sitting on hcp sites.The transition between the domains is made smooth using an error function.The width of the faulted domain, of the domain walls, and the overall unit cell size can be varied in the model.From these parameters, the CTR profiles, their magnitude, and the in-plane location are calculated by using the standard formula for the structure factor.Finally, a 3D model of the surface was constructed with a hard sphere relaxation model to find the height of the atoms in the model.The corresponding in-plane reciprocal pattern using only 1 of the 3 rotated domains is shown in Fig. 5a.The CTR rods of (0,2) and (−22, 1) as well as the SLR rod (−23, 1) are indicated on the inplane reciprocal space pattern.In Fig. 5b the calculated diffraction intensities using 3 domains for the (−22, 1) CTR for both the reconstructed and unreconstructed surfaces are shown, for L-values between 0 and 5 reciprocal lattice units (r.l.u).From the calculations it is clear the presence of the HB reconstruction results in sharp drop of the diffracted intensity between the Bragg peaks (where the diffraction signal is most surface sensitive) compared to an unreconstructed (111) surface.This due to the destructive interference of the atoms at HCP and bridge sites in the HB reconstruction.Therefore, the intensity of the CTR should decrease as the intensity of the SLR increases, which matches our experimental results.

Discussion
The HB reconstruction represents the most dense Au(111) surface, giving rise to an increase in the 2D-SOR reflectance and also the SLR intensity.However, the HB reconstruction consists of mixed hcp and fcc sites in the top-most layer of the Au(111), resulting in strong interference effects decreasing the intensity of the CTR, especially at the minimum of the CTR.When the HB reconstruction is lifted, small Au islands are formed on the (1x1) unreconstructed surface due to the Au density being 3-4% higher in the HB reconstruction compared to the layer below. 7,21,32This is effectively a roughening of the surface.This is detected as a continuous decrease in the 2D-SOR signal until 1.00 V, which extents much further than any (above background) SLR intensity in the SXRD measurements.Around the peak at 1.50 V (position 2) and higher potentials, the gold oxidizes. 33,34This oxidation will lead to further roughening of the Au(111) surface 34 or perhaps a change Journal of The Electrochemical Society, 2021 168 096511 in the dielectric constant.When this occurs, the CTR disappears abruptly and the 2D-SOR signal decreases further.The fact that this oxide is not detectable as new peaks or rings in the HESXRD measurements suggests that a Au oxide without any long range order (i.e. is X-ray amorphous) is forming or a very rough Au surface.However, there is an abrupt increase in intensity around the potential of the reduction peak (position 4) in the 2D-SOR and CTR ROI showing the (1 × 1) the surface rapidly regains its order over a very small potential window.Au and other metal surfaces such as Pt have been shown to become more rough after CV cycling, [34][35][36][37][38][39] which should theoretically be detected as a decrease in intensity of the 2D-SOR after a CV cycle.However, no clear decrease in the maximum 2D-SOR intensity between cycles could be detected, which is probably because we had already cycled the sample to oxidizing  potentials in earlier experiments and the sample was therefore already rough.Despite such non-ideal conditions, we are still able to detect (a weak) surface reconstruction signal with HESXRD.Often it is assumed that such surfaces need to be well prepared for surface reconstruction phenomena to occur or be detectable, however our results indicate such effects need to still be considered on quite rough and previously oxidized samples, even after beam damage.
It is clear the changes in the 2D-SOR intensity correlate well with the disappearance of the SLR due to the reconstructed surface.However, the change in 2D-SOR intensity seems to be more gradual than that from HESXRD which is a similar situation to how early SHG measurements compared to SXRD. 22By using high-energies and a large detector we have the advantage of being able to capture accurate intensity values even if the SLR moves in reciprocal space, whereas the measurements that were previously compared to SHG were performed by only monitoring one place in reciprocal space (the SLR peak).We are also able to simultaneously detect the CTR and SLR signals.The 2D-SOR signal continues to decrease even after the CTR signal has reached its maximum intensity, one possible explanation for this could be a gradual increase in the roughness of the Au islands present on the surface after lifting of the reconstruction.2D-SOR is also sensitive to the oxidation of the Au surface, but not for instance to reactions such as oxygen reduction and evolution that do not significantly modify the electrode structure, in this way 2D-SOR can be used to determine the origin of currents measured during cyclic voltammetry.The implementation of 2D-SOR is relatively straightforward and inexpensive (hundreds to a few thousand US dollars) and therefore through correlation with HESXRD measurements atomic level surface sensitivity should be available in most laboratories.Due to the very fast detection rate possible with 2D-SOR (it is an optical camera) one could envisage making kinetic studies with potentialstep experiments.For instance, the rate of the hydrogen evolution reaction, which is affected by the presence of the surface reconstruction, was recently used to determine how anion adsorption affects the formation of the surface reconstruction at different potentials. 40owever, from our measurements it seems like there is less hysteresis in the 2D-SOR signal (surface density) than the HESXRD signals (in-plane order) when the reconstruction is reforming, therefore it is the transition in surface order that takes longer.So it is possible that high-frequency combined 2D-SOR and HESXRD experiments could offer new understanding about the kinetics of this very interesting system.

Conclusions
We have combined HESXRD and 2D-SOR, and have shown that 2D-SOR is not only able to detect the appearance and disappearance of the herringbone surface reconstruction on Au(111) but also provide complementary information related to the existence of 1x1 islands on the surface and the ordering kinetics.We have also discussed and explained apparent discrepancies between early SXRD and optical measurements.No evidence for new diffraction peaks or rings were detected by HESXRD measurements after oxidation of the electrode surface.Thus any oxide forming on the Au (111) sample during the CV has little or no long-range order.To investigate this, other techniques such as electrochemical X-ray photoelectron spectroscopy (EC-XPS) could be used to study the electrooxidation of Au further.The high sensitivity and the low cost of 2D-SOR makes it an interesting tool for surface studies as well as a complimentary tool for HESXRD measurements.Finally, 2D-SOR was shown to be a useful tool to directly image and explore the effect of X-ray induced beam damage on the sample, which could be significant in an aqueous environment at high-energies.
Journal of The Electrochemical Society, 2021 168 096511 evolution reaction and 4 is the surface oxide reduction peak.The blue line shows the derivative of the 2D-SOR signal, which closely

Figure 1 .
Figure 1.Illustration of the experimental setup.The HESXRD pattern from the diffracted X-rays and the 2D-SOR images are detected simultaneously.The LED light shines onto the sample from the top of the electrochemical cell.The X-rays are directed onto the sample at a grazing incidence angle to increase the surface sensitivity.

Figure 2 .
Figure 2. Detector images and regions extracted.a) The detected HESXRD pattern for the Au(111) single crystal.The pattern is a superposition of the maximum intensity registered for each pixel during a CV between 0 V and 1.95 at 10 mV s −1 .The Bragg reflections (labelled in hexagonal surface units14 ) are blocked by beam stoppers.The bright spot between (012) and (105) is likely a spot in a powder ring and not part of the CTR.Note: the no orientation matrix was generated so the Bragg peaks could be equally indexed as their symmetry equivalents.b) A magnification of the area indicated with a white box in a), in this area the CTR and SLR signal is detected, depending on the applied potential.The white ROI shows the area used to extract the intensity for the CTR and the red ROI is used for the intensity of the SLR, in Fig.4.c) The 2D-SOR image of the Au(111) single crystal surface, measured at 0 V.The blue ROI is used to extract the 2D-SOR signal in Fig.4.The dark stripe is a result of beam-damage and is discussed in the SI.}

Figure 3 .
Figure 3. Potentials of interest.(a) CV curve between 0.00 V and 1.75 V for the Au(111) crystal measured during the experiment.The sweep rate was 10 mV s −1 .The numbers in the curve represent potentials of interest.1 is the cathodic limit of the CV scan, 2 is the Au oxidation peak, 3 is current due to OER and 4 is the reduction of Au oxide.The blue line shows the derivative of the 2D-SOR signal.(b) Images from the 2D-SOR and HESXRD measurements at the potentials indicated in (a).The 2D-SOR images have been normalized by dividing the image with the corresponding images at 0 V.The HESXRD images have been divided with an HESXRD image measured at 1.75 V, where the SLR and CTR have a low intensity, to visualize the intensity changes better.The SLR is present in the HESXRD at the cathodic limit of the CV.The CTR signal is present at the oxidation and reduction peaks of the CV.At the oxygen evolution neither the CTR nor the SLR are present.

Figure 4 .
Figure 4. HEXSRD and SOR intensity during cyclic voltammograms.(a) The trend of the integrated ROI's from Fig. 2 are plotted for 5 CV scans between 0.00 V and 1.95 V.The red line is the CTR intensity (topmost panel), the black line the SLR intensity (middle panel), and the blue line is the 2D-SOR intensity (bottom panel), and the green curves are the measured CV current.The numbers correspond to those in the text and Fig. 3. (b) The average signals from all cycles plotted as a function of potential, the colors match those used in a).The scan rate was 10 mV s −1

Figure 5 .
Figure 5. Theoretical calculation of the intensity of CTR for the unreconstructed and reconstructed Au(111) surface.(a) The in-plane diffraction pattern for 1 rotational domain for the HB reconstructed Au(111) surface.The inset shows the model of the HB reconstructed Au(111) surface with the (22 × √3) unit cell indicated, the same unit cell is indicated in the in-plane diffraction pattern.(b) The calculated (−22, 1) rod for the HB reconstruction and the unreconstructed Au (111) surfaces.