Potential‐Dependent Morphology of Copper Catalysts During CO2 Electroreduction Revealed by In Situ Atomic Force Microscopy

Abstract Electrochemical AFM is a powerful tool for the real‐space characterization of catalysts under realistic electrochemical CO2 reduction (CO2RR) conditions. The evolution of structural features ranging from the micrometer to the atomic scale could be resolved during CO2RR. Using Cu(100) as model surface, distinct nanoscale surface morphologies and their potential‐dependent transformations from granular to smoothly curved mound‐pit surfaces or structures with rectangular terraces are revealed during CO2RR in 0.1 m KHCO3. The density of undercoordinated copper sites during CO2RR is shown to increase with decreasing potential. In situ atomic‐scale imaging reveals specific adsorption occurring at distinct cathodic potentials impacting the observed catalyst structure. These results show the complex interrelation of the morphology, structure, defect density, applied potential, and electrolyte in copper CO2RR catalysts.


As-prepared: Comparison of electropolished and UHV-prepared Cu(100) after air exposure
Electropolished samples show increased mesoscopic roughness and thus an increased step density with smaller terraces. Nevertheless, large flat surface areas can be found on these samples, Figure S1a. Step heights are multiples of the Cu(100) interlayer spacing of 180.5 pm as indicated by dotted lines in Figure S1b. The terraces show a similar structure as in the ultrahigh vacuum (UHV) case ( Figures S1c,d). Granular structures cover the entire surface area evenly and in favorable cases are even smaller than those observed on some UHV-prepared samples, as shown in Figure S1. However, due to the altered surface morphology, the electropolished surface shows typically larger step bunches and only a few monoatomic steps.
For comparison, Figures S1c,d shows a UHV-prepared sample with flat surface morphology and evenly distributed terraces of several 100 nm width and step heights down to monoatomic steps (180.5 pm). Even exposure to air moisture for extended periods of time (hours-days) preserves this morphology. The grain sizes observed depend slightly on time and on AFM tip radius (tip convolution effect).
In both cases, the surface passivates immediately and comprehensively upon preparation. The surface morphology of the metallic copper underneath is preserved to a large extend. The resulting terrace roughness is below 0.5 nm. The granular passivation features can be understood from the heterogeneous nucleation of planar, linear and three dimensional oxide phases on Cu(100) at 100 mbar oxygen pressure. [2] Figure S1. Comparison of as-prepared electropolished and UHV-prepared surface terraces in air. (a) AFM image of an electropolished surface after 1.5 h in air. Mono-atomic steps are observed besides step bunches several times that height. Granular structures cover all terraces. (b) AFM image of a UHV-prepared Cu(100) surface after air exposure. Terraces separated by steps of atomic height dominate the surface. Again the entire surface is covered by granular structures rather evenly. The typical terrace roughness is below 0.5 nm in both cases. Equidistant dotted lines represent the Cu(100) interlayer spacing of 180.5 pm. Both images 500 nm × 500 nm.

Height scale of morphological changes
While AFM topographies do not generally provide a direct measure to gauge the depth to which wet oxidation affects the surface, the comparison of certain morphological features can give some understanding. Figure S2 shows the (a,b) as-prepared and (c,d) as-reduced state of a sample, which had been prepared in UHV after suffering some roughening during electropolishing. The latter resulted in cascades of massive step bunches, up to several 10 nm in height, with accurately aligned step edges. After growth and reduction of the usual hydrothermal oxide layer in the electrolyte. Only the large step bunches are reminiscent of the initial morphology, while the terraces comprise several nanometer or even 10 nm deep pits. This leads us to the conclusion that the observed height variations due to wet oxidation is 15-20 nm. This is not too different from the observed thicknesses of 6-10 nm for passive films grown on copper in other electrolytes.

In situ AFM on electropolished Cu(100) in deaerated H2O
Submerging the electropolished or UHV-prepared copper surfaces within the aqueous electrolyte results in the drastic morphology changes described in the main text. Exposure to pure H2O produces a similar effect. The in situ AFM images in Figure S3 show an electropolished Cu(100) surface in contact with pure water. The granular structures observed and loss of morphological details, like single atomic steps from the as-prepared state, are strikingly reminiscent of the corresponding images in an aqueous bicarbonate solution. Surface roughness is consequently increased to several nanometers also in the case of pure water. Although our electrolytes were composed of highest purity ingredients and deaerated by bubbling with 5N argon for several tens of minutes to many hours (and subsequently with CO2, if not indicated otherwise), we cannot fully exclude residual oxygen which could oxidize the sample surface without an applied electrochemical potential. Figure S3. In situ AFM image of a Cu(100) single crystal electrode recorded in pure H2O. Image size 500 nm × 500 nm.
As many electrocatalytic experiments evolve over time scales much longer than a series of EC-AFM images, i.e. sometimes many hours or days rather than half an hour, we show a time series in Figure S4 of the morphological development from the initial reducing cathodic potential step to 1.5 h afterwards while being held continuously at the same reducing potential. These in situ images witness a rapid change from the nanoparticulate as-wetted surface morphology to the reduced states at three different potentials. Arrows in the images indicate the direction of the AFM and EC scans. The location of their base coincides with the application of the cathodic potential step. Transformations from the as-wetted to the reduced state occur during only a few scan lines, i.e. within several seconds, but completion may take several images, i.e. 5-10 minutes, presumably depending on the thickness of the initial passivating layer. However, the latter impression could also result from poor tip states, which may prohibit good resolution until the interaction with (mobile) surface species might alter again the tip and result in better resolution. After the swift initial transition, the surface reaches a steady state with characteristic morphological features for a given potential. This steady state morphology is stable at least over the course of 1.5 hours.

Effect of specific adsorption on surface morphology
As described in the main text, there are specifically adsorbed anions known to affect the surface structure and morphology. By virtue of strong chemisorption bonds, ordered surface adsorbates can stabilize certain substrate geometries as shown for copper in early EC-STM work. [3] When atomic resolution is not achieved, this knowledge may enable deduction of the surface adsorption state from the surface morphology as illustrated in Figure S6. For non-deaeraed electrolyte and -0.5 VRHE for example, islands and terrace structures with step edges aligned along Cu<100> axes are observed, while deaerated electrolytes produce smoothly curved edges at the same cathodic potential ( Figure S6a+b). This is rationalized by pinning of the copper substrate by adsorbates (presumably atomic oxygen) along lattice axes which are energetically favorable for the adsorbate-covered surface, while less so for the bare copper. In the deaerated electrolyte the stabilization through the adsorbate is missing and electrochemical annealing results in rounded surface shapes. This is in line with different step edge orientations observed for Cu(100) in the deaerated electrolyte at more cathodic potentials. Under these conditions, no oxygen should be adsorbed, in fact, every adsorbate except for possibly hydrogen and CO will have desorbed. This results in crystallographically-oriented step edges aligned with the Cu<110> axes as expected for a bare copper surface ( Figure S6c).

Gas evolution in the EC-AFM cell during imaging
Gas evolution within the EC-cell is a major problem for most, if not all in situ / operando characterization methods. Gas bubbles (i) alter the electrode-electrolyte interfacial properties, (ii) those of the electrolyte or (iii) simply obscure the electrode-liquid interface. In EC-AFM all three aspects are relevant. Confinement of the educt-depleted electrolyte may alter the double layer region and the presence of gas may change the local electrolyte composition (e.g local pH). Additionally, gas bubbles can block the optical paths of the AFM excitation and detection lasers which prevents imaging during the passage or residence time of the bubble. Figure S8a shows an optical micrograph of the EC-AFM cell during imaging in the presence of a bubble emerging from underneath the cantilever chip, which eventually altered the laser path sufficiently to prevent detection. Figure S8b shows the case of small stationary bubbles in the vicinity of the cantilever which do not interfere with the detection and would allow imaging. In certain cases of continuous gas evolution, imaging precedes despite bubbles emerging near the cantilever. The temporarily blocked optical path, resulting in image segments without signal, is visible as stripes with no contrast parallel to the fast scan direction (x-or horizontal axis), see Figure S4b. Sometimes imaging resumes after a bubble perturbance without visible effect, other times the piezo creep leads to artifacts in the image.