Nanoscale Surface Structure−Activity in Electrochemistry and Electrocatalysis

Nanostructured electrochemical interfaces (electrodes) are found in diverse applications ranging from electrocatalysis and energy storage to biomedical and environmental sensing. These functional materials, which possess compositional and structural heterogeneity over a wide range of length scales, are usually characterized by classical macroscopic or “bulk” electrochemical techniques that are not well-suited to analyzing the nonuniform fluxes that govern the electrochemical response at complex interfaces. In this Perspective, we highlight new directions to studying fundamental electrochemical and electrocatalytic phenomena, whereby nanoscale-resolved information on activity is related to electrode structure and properties colocated and at a commensurate scale by using complementary high-resolution microscopy techniques. This correlative electrochemical multimicroscopy strategy aims to unambiguously resolve structure and activity by identifying and characterizing the structural features that constitute an active surface, ultimately facilitating the rational design of functional electromaterials. The discussion encompasses high-resolution correlative structure−activity investigations at well-defined surfaces such as metal single crystals and layered materials, extended structurally/compositionally heterogeneous surfaces such as polycrystalline metals, and ensemble-type electrodes exemplified by nanoparticles on an electrode support surface. This Perspective provides a roadmap for nextgeneration studies in electrochemistry and electrocatalysis, advocating that complex electrode surfaces and interfaces be broken down and studied as a set of simpler “single entities” (e.g., steps, terraces, defects, crystal facets, grain boundaries, single particles), from which the resulting nanoscale understanding of reactivity can be used to create rational models, underpinned by theory and surface physics, that are self-consistent across broader length scales and time scales.


INTRODUCTION
The structure of electrode surfaces has long been considered to have a profound effect on electrode kinetics and reaction mechanisms. Understanding structure−activity−selectivity relationships for electrocatalysts has arguably never been more important than today, with electrochemistry finding renewed interest in areas from organic synthesis to sensor technologies and being at the heart of energy storage and conversion technologies, 1 which need to be improved considerably and quickly if we are to move to a world of decarbonized energy. High-resolution microscopy has provided unprecedented views of the complexity of electrode surfaces. However, despite this acknowledged complexity, electrochemical measurements rely mainly on rather old macroscopic techniques that provide activity averaged over a wide range of interacting surface sites, thereby obscuring the nature of key elementary processes. The aim of this Perspective is to highlight opportunities for fundamental electrochemistry and electrocatalysis studies, whereby electrode activity and dynamics (electrochemical fluxes) can be visualized at the nanoscale in the form of electrochemical "activity pictures" and "activity movies", and further, where these high spatiotemporal resolution electrochemical data can be correlated directly with the underlying electrode structure and properties (electronic, chemical), obtained by using complementary high-resolution microscopy techniques in the same region of an electrode. This new age of correlative electrochemical multimicroscopy promises a much improved understanding of structural controls in electrocatalysis and will greatly advance the knowledge of electrochemical processes and facilitate rational catalyst design.
To illustrate the power of these approaches, we discuss a range of contemporary topical processes at different classes of electrodes used in electrochemistry. At the simplest level, the electrochemical processes can be divided into two categories ( Figure 1): outer-sphere redox processes, where there is little or no physical interaction between the redox species and electrode surface, and where questions relate to the influence of local electronic structure (density of states), solvent/electrolyte properties and double layer effects on electrochemical processes (Figure 1a), and inner-sphere or catalytic redox processes, where the bonding or adsorption of reactants, intermediates, and/or products to the electrode surface has a profound effect on the electrode reaction kinetics (Figure 1b). 2,3 Outer-sphere redox processes are fully described by a formal reduction potential (E°′), standard rate constant (k 0 ), and charge-transfer coefficient (α), as defined in the classical Butler−Volmer formalism of electrode kinetics, whereas inner-sphere (catalytic) processes are often benchmarked by an overpotential (η), Tafel slope (semiquantitative indicator of charge-transfer kinetics and/or mechanisms for simple processes), and/or exchange current density (j 0 , equal to the current density, j, at the equilibrium potential, which is a quantitative indicator of chargetransfer kinetics), as defined in the well-known Tafel equation. 1,3,4 The traditional approach for exploring the role of surface structure and defects in electrocatalysis has been to make use of single crystals of long-range order, prepared with a particular surface orientation (Figure 2a). Macroscopic electrochemical measurements on such surfaces have been used to elucidate how structure influences activity, particularly for noble-metal electrodes. 5−7 However, single-crystal surfaces are not perfect over large areas, and even on the best-quality surfaces it is challenging to elucidate the roles of step edges and terraces. 8 As we discuss in section 2, high-resolution electrochemical measurements at characteristic individual "single-entity" surface sites are providing new perspectives on local activity at single-crystal surfaces and layered materials.
Electrodes of practical importance are often polycrystalline and, furthermore, may show compositional variations ( Figure   2b). It is for this type of electrochemical interface that highresolution correlative structure−activity investigations come into their own, by unambiguously and directly relating local electrochemical fluxes to the corresponding local surface structure (individual grains and grain boundaries (GBs)) and composition, as we discuss in section 3. Local electrochemical measurements are also powerful in detecting the transport of reactive intermediates between neighboring active sites on a surface and the synergistic operation of catalysts and electrocatalysts. 9,10 These aspects critical to the operation of electrocatalytic systems involving nanomaterials (e.g., nanoparticles, NPs) on electrode supports (Figure 2c) are completely hidden in macroscale measurements. A further consideration is the nanoscale diffusion and interaction of electrochemically generated reactive intermediates with the support, the possible (unwanted) products thereof, and, in turn, their interaction with the electrocatalyst. The physicochemical stability of the system (e.g., NP−substrate interactions, attachment, migration, dissolution−growth−ripening, etc.) during, and as a consequence of, the electrochemical process must be considered. These key issues have brought about a diversity of different approaches to assess NP activity and stability, the relative merits of which are assessed and discussed in section 4.

WELL-DEFINED (SINGLE-CRYSTAL) SURFACES
It is widely postulated that atoms on a surface with low lattice coordination numbers, present at defects, serve as the active sites for (electro)catalytic processes. Directly identifying and characterizing the intrinsic activity of these highly localized surface sites would be valuable but is extremely challenging, due to the estimated low coverage, small size, and tendency of these sites to coarsen (restructure) under operational conditions. 7 The electrochemical scanning tunneling microscope has recently been proposed as a promising tool for identifying catalytically active sites at the atomic scale, on the basis of the In an inner-sphere (catalytic) redox process, reactants, intermediates, and/or products interact strongly with the electrode surface (specific adsorption) during the electrochemical process, which often involves the breakage or formation of chemical bonds. Examples considered herein include the Fe 2+/3+ process (in the presence of bridging ligands such as Cl − ), hydrogen evolution reaction, oxygen reduction reaction, and electrochemical CO 2 reduction. Figure 2. Dynamic (electro)chemical and/or (electro)catalytic processes in action at nanostructured interfaces. (a) Well-defined (single-crystal) surfaces, while nominally structurally and compositionally uniform, as in the terrace site (TS), possess defects such as step edges (SE) that may give rise to nonuniform reactivity and dominate the overall macroscopic response. (b) Extended heterogeneous surfaces, such as polycrystalline metals, comprise structurally (e.g., grains and grain boundaries, GBs) and/or compositionally disparate (e.g., inclusions) sites that can possess vastly different intrinsic electrochemical activities. (c) Nanoparticles (NPs), possessing size-, shape-, and structure-dependent activity, may interact physicochemically (diffusional coupling, aggregation, sintering, etc.) during electrocatalytic turnover (i), as well as undergo dynamic interaction with the support (ii).

Journal of the American Chemical Society
Perspective premise that, under reaction conditions, the electron tunneling barrier will be different over active and nonactive sites 11 (induced by local changes in electrolyte composition or adsorption/desorption processes). In practice, this has been demonstrated by monitoring changes in tunneling current noise to directly identify the active sites at metal single-crystal surfaces. For instance, elevated tunneling current noise ("activity") was identified at steplike defects in comparison to terrace sites for the hydrogen evolution reaction (HER) and oxygen reduction reaction (ORR) on Pt(111) surfaces. Furthermore, this approach revealed elevated HER activity at the boundary regions between monoatomically high Pd islands supported on an Au(111) surface, attributed to the electronic properties of active Pd atoms in the centers of the island and at their boundaries being altered dissimilarly by the catalytically inactive Au substrate. 11 A second class of materials that can be characterized by longrange order is carbon sp 2 materials (graphite, graphene, and carbon nanotubes), which are widely used and studied electromaterials. Along with the predominant basal surface (or side wall for nanotubes), there are a variety of defects, with the type, concentration, and distribution depending on the source and/or method of synthesis of the material. 12 Identifying the contribution of these different sites to the electrochemistry of carbon electrode surfaces is important, as there had been a longstanding view from classical macroscale electrochemistry that defects dominated the electrochemistry of sp 2 carbon materials, even for simple outer-sphere redox processes ( Figure  1a). The advent of nanoscale electrochemical methods, notably scanning electrochemical cell microscopy (SECCM), in tandem with complementary microscopy techniques applied to the same area (notably Raman microscopy and atomic force microscopy) enabled these established hypotheses to be tested by targeting and characterizing key surface features independently for the first time. 12 In SECCM, electrochemical measurements are performed in a series (typically many thousands) of small areas of a surface defined by a meniscus (droplet) cell created between a nanopipet probe filled with electrolyte solution (mobile electrochemical cell) and substrate (working electrode) surface. During operation, the positions of the nanopipet probe and/or substrate are precisely controlled in 3D space using piezoelectric positioners, and electrochemical (e.g., voltammetric) measurements are performed by applying a potential between a quasireference counter electrode (QRCE) located within the probe and substrate surface. 13 The QRCEs used in SECCM (and scanning ion conductance microscopy, SICM; vide infra) are predominantly comprised of electrochemically stable Ag/AgCl wires, 14 although alternative QRCEs such as palladium− hydrogen have also been used, 15−17 in addition to conventional three-electrode formats. 18 A key attribute of SECCM is that it can be viewed through the lens of classical electrochemical methodology, i.e., well-known electrochemical techniques can be applied directly, but in a format where the cell (i.e., working

Journal of the American Chemical Society
Perspective electrode area) is orders of magnitude smaller and more mobile, enabling many thousands of local (spatially resolved) electrochemical measurements to be made across a surface, which can be analyzed quantitatively. 19 Additionally, as SECCM is usually carried out under "ambient" conditions (i.e., exposed to air) or under an inert atmosphere, the probe position can be conveniently visualized pre-and/or post-experiment using a camera; this key attribute greatly facilitates colocated ex situ spectroscopic/microscopic analysis, allowing a range of techniques (correlative multimicroscopy) to implemented in the same area as SECCM scans. 12 An illustrative example of the type of measurements that can be made with correlative electrochemical microscopy (i.e., SECCM in conjunction with colocated structural analysis) is the characterization of single-, bi-, and few-layer exfoliated graphene (identified by Raman microscopy), within a single sample on an insulating oxide covered silicon support. The [Ru(NH 3 ) 6 ] 3+/2+ redox process (in aqueous solution) was shown to exhibit layerdependent electron-transfer kinetics, where the apparent k 0 value scaled with overall layer number. 13 The history (aging) of the sample was also shown to be an important factor governing the electrochemical response of graphene and graphite, with step edges showing enhanced activity in comparison to the basal surfaces on aged samples for the [Ru(NH 3 ) 6 ] 3+/2+ process, in contrast to freshly cleaved highly oriented pyrolytic graphite (HOPG), where the reaction was entirely diffusion limited on the time scale accessible by SECCM. 13,20 These observations were rationalized in terms of the local density of electronic states at these characteristic features, and kinetic data were analyzed semiquantitatively in terms of a Gerischer−Marcus model for heterogeneous electron transfer. 13,21 In a similar fashion, SECCM was able to prove that defect-free regions of the sidewalls of single-walled carbon nanotubes could support fast electron transfer for outer-sphere redox processes and that conducting and semiconducting carbon nanotubes showed contrasting behavior for certain cathodic processes. 12 While defects were not important for simple redox processes, they were shown to be critical for the ORR. 22 An important aspect of many SECCM studies of sp 2 carbon materials has been to show how nanoscale reactivity scales up rationally to explain macroscopic behavior. Consistency in electrochemical data across length scales and time scales, rationally underpinned by surface physics, is a critical theme for the future that needs to become routine in electrochemistry. 12,13,20,22 Local voltammetric measurements with SECCM have also revealed new insights on structure−activity dynamics in molybdenum disulfide (MoS 2 ), another class of layered material that has received considerable attention as an earth-abundant HER electrocatalyst. 1,4 Macroscopic electrochemical measurements on ensembles of nanostructured (exfoliated/synthesized) 2H MoS 2 have alluded to high HER activity at the edge plane, owing to the near-thermoneutral free energy of hydrogen adsorption (ΔG H ≈ 0 eV) calculated for this structural element, in comparison to the so-called "catalytically inert" basal plane. 4,23 SECCM was employed to perform current−potential (i−E) measurements on structurally well-defined natural crystals of molybdenite, which were subsequently combined to create electrochemical flux movies over a wide potential range with nanoscale spatial resolution. 15,16 Correlation with complementary structural information from SEM and/or atomic force microscopy revealed uniform HER activity on the basal plane and elevated current densities (i.e., enhanced activity) at the edge plane that scaled linearly with the number of exposed MoS 2 layers (i.e., step height), 15 in line with theoretical predictions and macroscopic electrochemical studies, above. 4,23 For example, in the synchronously obtained high-resolution topography and electrochemical activity maps in Figures 3a-i and ii, the topographical features (corresponding to multilayer step defects) of MoS 2 perfectly align with the areas of elevated surface current, directly (and unambiguously) identifying the edge plane as the active site for HER catalysis, also reflected in the pixel-resolved voltammograms, shown in Figure 3a-iii. 16 The often claimed "catalytically inert" basal plane, which incorporates some surface defects (i.e., sulfur vacancies), was shown to possess a j 0 value comparable to those of moderate HER catalysts (Au and Cu), while the edge plane possessed a j 0 value >10 times larger. 15 It is worth noting that the much higher activity found at the basal surface, particularly in comparison to previous studies on the bulk material, 4 can be explained, at least in part, by the fact that SECCM draws such small currents (typically tens of pA), meaning that it is relatively immune to bulk sample resistance, which is a major problem for macroscopic measurements on resistive semiconductor materials. 12,15,16 Furthermore, for SECCM there is no need to encapsulate the material as an electrode, because the electrochemical meniscus cell is brought into contact with the substrate (working electrode) of interest. In contrast, electrode encapsulation and preparation can also be a practical problem for conventional electrochemical studies of unusual materials (e.g., 2D materials).
These studies 11,13,15,16 and those discussed below clearly demonstrate how scanning probe techniques can go well beyond the capabilities of macroscopic electrochemical measurements to look at the heterogeneities within the surfaces of well-defined metal single crystals 11 and layered materials. 13,15,16 The techniques discussed in this section collect electrochemical and topographical information synchronously (in situ and in real time) which, in conjunction with complementary structural information, reveal nanoscale structure−activity dynamics at functional electrochemical interfaces.

EXTENDED HETEROGENEOUS SURFACES
As mentioned in section 1, electrochemical interfaces of practical importance in electrochemistry and (electro)catalysis are structurally and/or compositionally heterogeneous (see Figure 2b). To understand the overall behavior of these complex surfaces, it is essential that structure and activity can be related at the scale of surface heterogeneities. For polycrystalline surfaces, we have introduced a pseudo single-crystal approach, where SECCM is used to electrochemically interrogate individual grains and GBs on a polycrystalline surface, which is structurally characterized ex situ with electron backscatter diffraction (EBSD). 24 This approach is being adopted by other groups, 18 and we expect that it should find considerable application in electrocatalysis, as it enables the activity of a wide range of surface features, including high-index facets, which are difficult to prepare as single crystals, and grain boundaries.
This approach is well illustrated by the facet-dependent electrochemistry of Fe 2+/3+ , a well-known outer-sphere process ( Figure 1a) with (anion-mediated) inner-sphere routes ( Figure  1b), 3 on polycrystalline Pt. In weakly adsorbing electrolyte media (HClO 4 ), the individual (high-index) grains were shown to have markedly different activities, whereas in strongly adsorbing electrolyte media (H 2 SO 4 ), the individual grains exhibited comparable activities, while GBs were highly active,

Journal of the American Chemical Society
Perspective dominating the overall surface activity, suggesting that these sites dominate the macroscopic response of polycrystalline Pt. Interestingly, not all GBs exhibited enhanced activity, indicating that the character (geometry) of the GB itself is important (vide infra). 24 The pseudo single-crystal approach has also revealed grain-dependent ORR activity at polycrystalline Pt in acidic media, where SECCM mimics the three-phase boundary of lowtemperature fuel cells, with an enhanced flux of O 2 across the meniscus (droplet) cell giving rise to high rates of reactant mass transport. The individual high-index grains were shown to exhibit significantly different ORR activities, while the GBs did not exhibit any enhancement under these conditions. 25 A very recent study eloquently demonstrated the power of pseudo single-crystal SECCM for rationalizing the macroscopic response of polycrystalline Au for electrochemical CO 2 reduction. Bulk electrolysis data demonstrated increased selectivity for CO production relative to H 2 with increasing GB density (Figure 3b), implying that GBs are more active than the grains for CO 2 reduction but not for the competing HER. This hypothesis was confirmed by SECCM line scanning across a series of GBs with differing geometries (see Figure 3c), where GB-dependent, elevated current (i.e., enhanced activity) was measured at GB surface terminations for CO 2 reduction plus the HER (i.e., enhanced activity at GB relative to the grains under a CO 2 atmosphere; Figure 3c-iii) but not for HER alone (i.e., only grain-dependent activity seen, no GB-specific enhancement under an Ar atmosphere; Figure 3c-ii). The width of the region exhibiting enhanced activity (Figure 3c-iii) was commensurate with the dislocation-induced strain field associated with the each GB, while the degree of enhancement in CO 2 reduction activity was qualitatively consistent with the magnitude of the lattice microstrain, an indicator of the concentration of dislocations within the GB. Thus, GBs create strained regions by stabilizing dislocations, creating high-energy surfaces that are "kinetically trapped" under electrochemical polarization (catalytic turnover). 18 Consequently, mechanical treatments were applied to annealed (polycrystalline) Au foil to artificially increase the GB (dislocation) density, which translated into substantially increased CO 2 reduction activity at the macroscale, 18 demonstrating how a nanoscale understanding of activity can be translated into the rational design of optimal catalysts.
In addition to structural heterogeneity, (electro)materials may possess compositional heterogeneity, arising during synthesis/growth (e.g., inclusions or surface enrichment in metal alloys). SECCM is a promising technique for probing compositionally heterogeneous surfaces, as demonstrated in a recent study of local HER activity on single-crystal iron nickel sulfides (nominally Fe 4.5 Ni 4.5 S 8 ), which are highly efficient electrocatalysts in bulk form. SECCM revealed lower activity from the (111) planes of Fe 4.5 Ni 4.5 S 8 in comparison to bulk macroscale measurements, suggesting that defects in single crystals, which would be exposed in bulk measurements, are largely responsible for the observed macroscopic activity. This was confirmed by performing local measurements on a macroscopic "defect" site, which showed higher activity in comparison ot the basal (111) surface. Local composition was also found to play an important role; Fe-enriched material with segregated regions possessing Fe:Ni ratios higher than the nominal 1:1 exhibited substantially higher activity. 26 These studies 18,24−26 and others (see below) have effectively demonstrated SECCM as a powerful tool for probing structuredependent activity and rationalizing the macroscopic response of structurally and/or compositionally heterogeneous electro-des. The pseudo single-crystal approach produces results that are consistent with conventional single-crystal studies at lowindex facet electrodes, 24,25 while it also allows high-energy surfaces, such as high-index facets and GBs, to be studied. Moving forward, the high-speed, high-resolution SECCM configuration established in refs 16 and 27 opens up the possibility of probing a larger population of grains (or compositions) in greater detail, where the entire i−E characteristic can be mapped at each point of a heterogeneous electrode surface and combined to create spatially resolved electrochemical flux movies. This information, taken in conjunction with complementary information on the local heterogeneities that give rise to the high activity (e.g., GBs 18,24 or the crystallographic defects/Fe-enriched areas 26 ) that dominates the macroscopic (bulk) response of these materials, will enable a more holistic view of the structural and/or compositional controls in (electro)catalysis.

NANOPARTICLES ON SURFACES
"Real" electrocatalysts are typically nanostructured, the most common example being NPs (see Figure 2c), which serves to maximize surface area and expose particular surface sites. 1,4,7 As alluded to above, bulk measurements of NPs effectively "wash out" the unique properties of each individual entity within an ensemble, and for this reason there has been a strong drive to develop techniques capable of performing measurements at the single-entity level. Here, we highlight emerging trends in identifying key physicochemical phenomena of NP electrochemistry at the nanoscale.
4.1. Single-Nanoparticle Electrodes. A single NP electrode is the conceptually simplest approach to study electrochemistry at the single-entity level. This approach has many advantages, including well-defined and fast mass transport when the size and geometry of the NP is known: for example, through complementary SEM or transmission electron microscopy (TEM) analysis of the electrode/NP assembly. The benefits of studying single NPs are well illustrated by considering the energy storage and oxygen evolution reaction (OER) properties of individual nonfaceted Ni(OH) 2 NPs electrodeposited onto carbon nanoelectrodes. 28 By performing voltammetry at the single-NP level, it was shown that charge storage via the reversible Ni(II)/Ni(III) transformation was diffusion-limited (nonpseudocapacitive) and that the OER kinetics and mechanism (Tafel slope) were invariant with respect to NP diameter in the considered size range (diameter of NP (d NP ) 40−1000 nm). 28 Coupled in situ microscopy adds further benefits to single NP studies, as exemplified by investigations of electrodeposited Co(OH) 2 particles (d NP = 0.5−3 μm), where the overall reaction rate (inferred from electrochemical current) was linked to dark-field microscopy measurements of particle size and Co redox state. The reversible Co(II)/Co(III) transformation was shown to coincide with "electrochemical breathing", where the particle undergoes rapid volume expansion and slow volume contraction during the oxidation and reduction processes, respectively. 29 Ongoing work in this area should focus on the study of structurally well-defined, shape-controlled (faceted) NPs, which can be grown through electrosynthesis (electrodeposition) and characterized using selected area (electron) diffraction. This strategy has been illustrated in a recent study 30 and will be valuable in establishing relationships between structure (i.e., size and/or shape) and activity at the single-NP level. to be selected and probed in situ and in a high-throughput manner. The electrochemical characteristics of a single NP within an ensemble can be resolved by detecting changes in the surface refractive index (and thus scattering intensity) resulting from surface electrochemical reactions, leading to optical contrast in surface plasmon resonance images, which are used to derive electrochemical currents and construct local (spatially resolved) voltammograms. 31 This technique, originally used for single-particle electrocatalysis studies, 32 has recently been expanded to obtain the electrochemical i−E profiles and charge/discharge characteristics of single LiCoO 2 nanoplates (size of ca. 200 nm) within an ensemble, resolving phase transitions and quantifying Li + diffusion rates in situ. 33 To fully elucidate the structural controls on the electrochemical activity of individual entities within an ensemble, it would be interesting to combine plasmonic-based electrochemical current imaging with other high-resolution imaging tools, such as TEM and scanning tunneling microscopy, applied to the same particle.

Journal of the American Chemical Society
Super-resolution fluorescence techniques offer a spatial resolution that is not limited by the diffraction of light (typically tens of nanometers). 34 While they are mainly focused on life sciences applications, 34 these techniques are beginning to find use as a probe of electrode activity, as exemplified by a recent study of the HER, where H 2 nanobubbles labeled with fluorescent dye molecules were imaged at individual Au nanoplates supported on an indium tin oxide (ITO) electrode at various potentials, as shown in Figure 4. The bubble nucleation frequency was generally higher on the Au nanoplates in comparison to the ITO support (see Figure 4b), although it should be emphasized that only certain nanoplates exhibited significant HER activity (e.g., nanoplates 1 and 3 versus nanoplates 2 and 4), again highlighting the importance of single (nano)entity studies in rationalizing the macroscopic electrochemical response of an ensemble. Interestingly, a very large number of nanobubbles nucleated within the ca. 3 μm radius of the "active" nanoplates on the ITO surface, particularly at high driving potentials (Figure 4b-ii,b-iii), are clearly evident from the cumulative scatter plot in Figure 4c. This was thought to be a manifestation of the "hydrogen spillover effect", where H atoms generated on the catalyst surface migrate onto the support before undergoing nucleation to form H 2 nanobubbles. This phenomenon is well-known in the gas phase but has rarely been reported in the electrochemical context, with this study being the first example of real-time imaging of electrochemically generated nanobubbles. 35 A separate approach, and one that works at atomic resolution, is the use of identical-location (ex situ) TEM imaging to monitor structural changes within an ensemble induced by (electro)chemical perturbation. High-resolution aberration-corrected scanning transmission electron microscopy (STEM), with an electron-transparent conductive boron-doped diamond support, has recently been used to probe the initial stages (0−30 ms) of the nucleation and growth of Au through electrodeposition at high η, progressing from single Au atoms to noncrystalline nanoclusters (AuNCs) through to crystalline AuNPs (d NP ≈ 1− 3 nm). 36 Potential-induced atom movement, atom clustering, and AuNC transformation into crystalline AuNPs, which in

Journal of the American Chemical Society
Perspective contrast to classical theory occurs through either loss or gain of atoms, were observed ex situ, with monocrystalline AuNPs being the dominant structure after 30 ms of electrodeposition. Discrete entities (atoms, AuNCs, and/or AuNPs) were also shown to interact during the early stages of electrodeposition via an aggregative growth mechanism, 37 with a strong tendency for disordered AuNCs to be consumed when they were in close

Journal of the American Chemical Society
Perspective vicinity to ordered AuNPs, which transitioned through a disordered state first, before undergoing recrystallization. 36 Pipet probes have been employed to perform local ensemble measurements, as demonstrated in a recent study 38 where SECCM was used to investigate the ORR response of a small population (N < 16000) of size-selected Pt nanoclusters (PtNCs, d NC ≈ 3 nm) deposited onto a carbon-coated TEM grid support, under high mass transport rate conditions (O 2 flux across the meniscus; 25 see Figure 5a). At low surface coverages, the activity decreased with successive electrochemical (voltammetric) measurements (Figure 5b−i), attributed to poisoning of the PtNCs by carbon-and oxygen-containing moieties that are produced by the reaction of reactive oxygen intermediates (RIs) generated transiently in the ORR with the carbon support (i.e., carbon corrosion). This was less of an issue at high surface coverage, where the distance between clusters was small, meaning that RIs could be either be consumed at the same PtNC (mass transport to individual particles is lower at high coverages) or diffuse to neighboring PtNCs and undergo further reduction, rather than react chemically with the support (Figure  5b-ii). The impact energy during PtNC deposition (achieved with a cluster beam source) also drastically affected NC stability during electrocatalysis, with low-impact-energy PtNCs migrating as a result of ORR, as seen by STEM imaging (see Figure 5c). This migration was attributed to electrochemical propulsion caused by an uneven flux distribution around individual PtNCs within the ensemble, explored through finite element method (FEM) modeling (Figure 5d). The random distribution of PtNCs gives rise to uneven flux, depending on the degree of shielding by neighboring clusters (see Figure 5d), which results in nonuniform electric fields and/or chemical gradients. 38 SECCM coupled with the use of a TEM grid substrate (explored further below) is a high-throughput approach to performing local measurements on ensembles of true catalytic NPs, strengthened by the capability of performing ex situ structural characterization with STEM and complementary quantitative analysis with FEM modeling. This approach is generally applicable to any (electro)catalytic system. Future work in this area may head in many directions: e.g., to resolve structure−activity in nanostructured energy storage materials or to elucidate support effects on activity, migration, and deactivation of NPs.
4.3. Electrochemical Nanoparticle Impacts. A popular approach for observing the electrochemical properties of single NPs that has gained considerable interest, as highlighted in recent reviews, 39 is to monitor NP impact (or landing) from a solution (colloidal suspension) onto a collector electrode surface (Figure 2c-ii), with detection based, for example, on blocking of the collector current, 40 electrocatalytic amplification at the NP, 41 or oxidative dissolution (stripping) of the NP. 42 NP An important question in this field is whether NPs arriving at an electrode undergo a single-pass interaction (hit and stick or hit and bounce) or undergo multiple interactions. Multiple interactions of a single NP were first exemplified in the study of ruthenium oxide NP impacts on HOPG, detected through the electrocatalytic oxidation of hydrogen peroxide (H 2 O 2 ), where multiple i−t events associated with a single (stochastic) NP impact were detected (i.e., "multipeak" behavior on the microsecond time scale; see Figure 6a). The multipeak behavior was attributed to NPs becoming hydrodynamically trapped in the vicinity of the collector electrode (HOPG) surface and undergoing multiple impacts before release and semiquantitatively reproduced through 3D random walk simulations. 44 Highbandwidth measurements are able to probe even faster processes during electrochemical impact, as demonstrated for oxide formation and associated catalytic deactivation of AuNPs impacting with a collector electrode. The i−t transient associated with AuO x formation on AuNPs lasts ca. 500 μs (Figure 6b), and the femtocoloumb charge passed in these events is proportional to the AuNP surface area, enabling accurate electrochemical size analysis, with results comparable to those of TEM. 45 The importance of bandwidth was further emphasized in a study on AgNP dissolution dynamics. By opening of the time window, studies from three different groups revealed that AgNP stripping (oxidative dissolution) was a complex, NP-sizedependent phenomenon, where larger AgNPs undergo multiple and repetitive stripping events (impacts), resulting in often incomplete electrodissolution on the microsecond time scale (a representative example is shown in Figure 6c), 46−48 contrary to original reports. 42 Long and co-workers developed a semiquantitative 3D random walk model to rationalize the sizedependent multipeak stripping behavior of AgNPs during electrochemical impact, where the trajectories of the AgNPs were described by Brownian motion in bulk, hindered diffusion (hydrodynamic trapping) as well as electric-field-driven motion in the near-wall (electrode) region and size-dependent adsorption/desorption of the AgNP in the electron-tunneling region, where electrodissolution occurs. 48 White, Zhang, and co-workers attributed the multipeak impact behavior (e.g., Figure 6c) to a single AgNP moving in and out of contact with the collector electrode during a collision event, where the motion of an AgNP at the collector electrode/ solution interface was driven solely by Brownian motion. 47 They developed a 3D lattice random walk simulation of AgNP Brownian motion dynamics (based on mass-dependent thermal velocity) in the near-electrode region (Figure 6d), which in combination with electrochemical kinetic parameters (i.e., j 0 for the Ag/Ag + process) quantitatively reproduced the experimental

Journal of the American Chemical Society
Perspective multipeak and partial stripping behavior. The simulations reveal that tens to thousands of NP−electrode collisions (nanosecond time scale) contribute to experimentally measured peak currents, and suggest that NP thermal velocity controls the electron-transfer rate during electrochemical impact. 49 By adoption of the highest possible bandwidth measurements, the field of electrochemical impacts has progressed by providing a wealth of information that was obscured in earlier i−t collision transients. 41,42 These improved data highlight single-NP activity, 44,45 motion, 47,48 and surface-interaction dynamics, 44,46 all of which are important in the context of understanding both individual NP activity and the stability and dynamics in NP ensembles (Figure 2c). However, it has to be recognized that interpretation of i−t traces alone can be somewhat ambiguous, necessitating the implementation of complementary microscopy and/or spectroscopy techniques to understand the dynamic redox behavior at the electrode/solution interface during NP impact. For this reason, SECCM was used to deliver single NPs to a catalytically inert carbon-coated TEM grid substrate, used as a working electrode for impact studies, followed by ex situ morphological analysis of collected NPs with TEM. This approach was applied to study the catalytic properties of AuNPs for hydrazinium ([N 2 H 5 ] + ) oxidation, where the entire i−E response was measured at the single-NP level (representative example in Figure 7a). 50 An important advance has been to monitor single NP impacts in situ with a range of optical methods, including 3D superresolution holographic microscopy, where individual AgNP stripping events are observed in real time. In the absence of complexing/precipitating agents, AgNP stripping is a rapid process but may not take place immediately upon the NP entering the vicinity of the near-electrode region, with a small "lag time" often observed. In the presence of complexing/ precipitating agents such as SCN − , charge injection (Ag(s) → AgSCN(s)), detected electrochemically, was followed by slow dissolution (AgSCN(s) → [Ag(SCN) x+1 ] x− ), observed optically but "invisible" electrochemically. In other words, the onset of AgSCN(s) dissolution was delayed relative to the electrochemically detected i−t transient associated with AgNP oxidation. 51 An innovative nanochannel cell configuration has also been developed to optically monitor the dynamic collision and Topographical (i, ii) and synchronously recorded electrochemical activity images (iii, iv) obtained during catalytic [BH 4 ] − turnover at supported AuNPs (mapped area indicated by red box in (a)). The applied potential (E app ) is indicated in (c-iii) and (c-iv). These images allow important characteristics of the ensemble to be directly visualized, including geometrical effects (e.g., "ring effect" in (iii)), diffusional overlap, competition between neighboring NPs and heterogeneous NP activity (e.g., bottom-right, "blue" AuNPs in (c-iv)). (d) FEM simulations showing the distribution of ions around an isolated NP during [BH 4 ] − turnover with high and low E app (i.e., high and low η). Depletion of ions in the NP−support gap (see (b)) at low E app gives rise to the experimentally observed "ring effect" seen in (c-iii). Adapted with permission from ref 17. Copyright 2017, American Chemical Society.

Journal of the American Chemical Society
Perspective oxidation of AgNPs in situ using single-particle fluorescence microscopy. 52 A further important development in the NP impact field has been the improved delivery of NPs to electrified interfaces using nanopipets, as illustrated by the local delivery of NPs (polystyrene) using a combination of SICM for positioning (vide infra), pressure-driven flow for controlled delivery (with the delivery event sensed through a resistive-pulse method 53 ), and super-resolution fluorescence microscopy for imaging NP dynamics. The trajectory was tracked in 3D space and manipulated by balancing the attractive/repulsive forces exerted on the NP by convection and the electric field. NP diffusion kinetics were analyzed to reveal subdiffusive (hindered) motion, resulting from (potential-dependent) attractive electrostatic interactions between the NP and an ITO substrate, or superdiffusive (directed) motion, resulting from pressure-driven flow (convection) at the pipet. This was exploited to manipulate NP position in real time with tens of nanometers precision, achieved by moving the nanopipet laterally across the surface while "dragging" the NP along by fluid flow (negative differential pressure; see Figure 7b). FEM modeling was employed to quantify and visualize the distribution of forces impinged upon a near-surface NP ("trapped" by the electric field) as a result of pressure-driven flow (Figure 7c), rationalizing the real-time manipulation of NP motion. 54 This study highlights the significant progress that has been made in the study of NP impacts in terms of sophistication, control, and analysis, in comparison to the original attempts in this field. Together with other studies highlighted, there is a new course for NP impact studies that are most effective when combining high-bandwidth electrochemical measurements, NP sizing (on the fly), and/or microscopy, including in situ optical methods, in order to tease out several phenomena that contribute to the current−time− potential response.
4.4. Electrochemical Imaging. As discussed in section 3, electrochemical imaging is a powerful technique for mapping spatially heterogeneous activity at complex (heterogeneous) electrochemical interfaces. Among a limited set of techniques for nanoscale electrochemical flux mapping, scanning electrochemical probe microscopy (SEPM) techniques (i.e., those using a physical probe), most commonly scanning electrochemical microscopy (SECM), have attracted significant interest. A very recent review covers the efforts on the use of SECM and hybrid techniques in this area, and hence it will not be covered here. 55 In comparison to the solid nanoelectrodes used in SECM, which can be difficult to make and are susceptible to damage, nanopipets are easily fabricated, even with nanometric dimensions, and produce a robust (unchanging) electrochemical response over extended time periods (hours to days), making them ideal for electrochemical imaging at the nanoscale. Indeed, building on a previous study, 56 it was recently demonstrated how fine nanopipet probes could be deployed in the SICM format for synchronous nanoscale reaction-topography mapping. 17 In SICM, the tip of a nanopipet containing a QRCE (e.g., Ag/AgCl; vide supra) is immersed in bulk solution, where another QRCE is located, and a potential bias is applied between these QRCEs to induce an ion conductance current through the orifice of the pipet. The ion conductance current is sensitive to the access resistance of the nanopipet, which is governed by the tip−substrate distance (i.e., topographical mapping in conventional SICM) in addition to surface charge, 57,58 as well as ion flux arising from (electro)chemical reactions at the interface. 17,56 Through careful control of the SICM probe bias, it is possible to map topography and surface activity simultaneously.
In SICM reaction mapping, 17,56 the probe acts as a "mobile ion conductance sensor", mapping the heterogeneous flux arising from an electrochemical reaction. This aspect is exemplified through studies of the oxidation of [BH 4 ] − at carbon-fiber-supported AuNPs (see Figure 8a,b), which consumes OH − , leading to a local change in ionic composition, while also mapping topography with high fidelity (see Figure 8ci,c-ii). Ion flux heterogeneities arising from geometrical effects (i.e., NP−support gap, Figure 8c-iii), diffusional overlap, competition for reagent between neighboring NPs (arising due to random distribution of NPs in the ensemble 38 ), and differences in NP activity (e.g., bottom-right AuNPs in Figure  8c-iv) were visualized and quantified. FEM modeling ( Figure  8d) was employed to rationalize the heterogeneous ion fluxes measured experimentally, with hindered mass transport at the NP−support gap (Figure 8b) giving rise to the characteristic "ring" effect in the activity maps at low driving force (E app , e.g., see Figure 8c-iii). 17 Nanopipets can also be deployed in the SECCM format to perform direct high-speed, high-resolution electrochemical imaging, as demonstrated for mapping the OER activity of iridium oxide (IrO x ) NPs on a HOPG support. Heterogeneity in OER activity was observed, with some IrO x NPs possessing considerable activity before the onset of detectable current from the whole ensemble (all NPs plus the support). 59 The highspeed, high-resolution voltammetric SECCM approach established in refs 16 and 27 (e.g., see Figure 3a) has also been used to elucidate the spatially resolved i−E behavior of individual nonfaceted polycrystalline AuNPs (d NP ≈ 350 nm) grown on a glassy-carbon support. While different individual AuNPs exhibited very similar overall catalytic activity toward [N 2 H 5 ] + oxidation (at the interparticle level), electrochemical reaction rates varied significantly across the surface of individual AuNPs (at the intraparticle level), attributed to structural (crystallographic) heterogeneities, demonstrating that these single entities cannot be considered uniformly active. 16 Recent advances in the speed and/or resolution of nanopipetbased SEPM (e.g., SICM 17,56 and SECCM 16,27,59 ) has enabled high-fidelity synchronous topographical/electrochemical mapping with true nanoscale (tens of nanometers) resolution. In all of the imaging studies highlighted above, 16,17,27,59 nonuniform electrochemical activity was identified within populations of nominally identical NPs, alluding to structural and/or compositional heterogeneities within the ensemble. Looking to the future, applying nanometer-resolution SEPM with complementary high-resolution spectroscopy/microscopy, e.g., through the use of TEM grid substrates in SECCM, 38,50 will be a powerful approach for (nano)material structure−activity determination, which is a crucial step in rational (electro)catalyst design and synthesis.
Single-particle reactivity mapping (at the inter-and intraparticle levels) has also been achieved with a range of superresolution (vide supra) optical approaches, which imply (electro)chemical reactivity from the "on/off" states of redoxreporter molecules (oxidized/reduced forms switch between "emissive" and "dark" states) and separately use supplementary techniques to characterize morphology. For example, superresolution surface-enhanced Raman scattering (SERS) has been used to probe local surface potentials on colloidal Au aggregates labeled with Nile Blue A. Correlation of potential-dependent

Journal of the American Chemical Society
Perspective emission centroids with morphological information from SEM revealed that there were discrete locations, typically situated at NP−aggregate junction regions, associated with the reporter molecules that were most difficult to reduce/easiest to oxidize. In other words, at high potentials, when all Nile Blue A molecules are in the emissive (oxidized) state, the emission centroid is initially located at the geometric center of the AuNPaggregate (Figure 9a-i), before spreading out (Figure 9a-ii) and becoming localized at specific junctions (Figure 9a-iii) at increasingly negative potentials. From this, it was hypothesized that reporter molecules at certain locations experience sitespecific electrochemical environments (Figure 9a-iv) or location-dependent redox potentials. 60 Super-resolution single-molecule fluorescence microscopy has been employed for (electro)chemical imaging, e.g., for mapping spatiotemporal fluctuations in single-molecule catalysis events on Sb-doped TiO 2 nanorods, where the middle of the nanorod was shown to be more active initially for OER photocatalysis, before deactivating (but sometimes recovering in a "self-healing" mechanism), causing the two ends of the rod to dominate the overall activity in the time-integrated response, despite lower intrinsic activity. 61 The same technique has also been used to probe intraparticle photoactivity on pristine and catalyst-modified single-crystal rutile TiO 2 nanorods supported on ITO, under an applied potential bias. Hole-and electroninduced surface reaction rates were shown to be strikingly nonuniform along the surface of single nanorods, with a strong spatial correlation between the hole (Figure 9b-i)-and electroninduced (Figure 9b-ii) photocatalytic "hot spots", revealed to be structural defects or impurity sites from complementary ex situ TEM/SEM analysis. The photocurrent associated with OER (i.e., photocatalytic activity for water oxidation), measured through the ITO substrate upon local laser illumination, revealed that the catalytically active sites coincide with the hole-and electron-driven "hot spots" (Figures 9b-i,b-ii, respectively), indicating that these sites effectively mediate both oxidation and reduction reactions. The TiO 2 nanorods were subsequently decorated with an OER catalyst (cobaltborate) through local photodeposition (Figure 9b-iii), which enhanced hole-/electron-driven photoelectrocatalytic activity at initially low activity sites and/or lowering the onset potential of the OER, especially at sites with initially high onset potentials. These two types of sites are not necessarily the same, highlighting the challenge of engineering efficient photoanodes with minimal amounts of catalyst, and on this basis, a block− deposit−remove type strategy to yield optimally located catalysts was proposed. 62 Super-resolution optical approaches have revealed active "hot spots" directly at the single-molecule level 60−62 of NPs in the hundreds of nanometers size range, as well as enabling studies of the time evolution of (electro)catalytic activity 62 and catalytic cooperativity in real time. 10 These techniques are undoubtedly very powerful, achieving unprecedented spatial (tens of nanometers) and temporal (sub-microsecond) resolution, but are restricted to the use of certain materials (i.e., plasmonic materials in surface plasmon resonance) and probe molecules (e.g., redox fluorophores in single-molecule fluorescence), as well as optically transparent supports. In contrast, scanning probe methods, such as SICM and SECCM outlined above, are more generally applicable to any class of (electro)catalytic system and provide synchronous topographical (morphological) information but are generally carried under true catalytic

Journal of the American Chemical Society
Perspective turnover conditions, rather than the single-molecule level, making cross-correlation between temporally neighboring catalytic events 10 impossible. While significant recent advances in this area have enabled nonuniform (electro)chemical fluxes to be measured at the sub-100 nm level, electrochemical imaging techniques alone generally cannot unambiguously reveal the compositional and/or structural origin of the nonuniformity, which must come through correlative multimicroscopy, in which electrochemical images and movies are related to colocated images of surface structure, chemistry, and/or electronic properties. This versatile philosophy, highlighted and emphasized throughout this Perspective, needs to be adopted more widely in the future in order to resolve the relationship between surface structure and activity in electrochemistry and electrocatalysis.

SUMMARY AND OUTLOOK
In this Perspective we have highlighted recent approaches to studying fundamental electrochemistry and electrocatalysis, whereby colocated information on structure and activity is collected on commensurate scales, ranging from hundreds of nanometers all the way down to the atomic level. Some of the different techniques that we have discussed provide electrode topography and activity synchronously, and although this alone may reveal a wealth of information, there is often the need to further employ complementary ex situ high-resolution microscopy/spectroscopy, which for some techniques (e.g., SECCM) is easier to implement due to the ease of obtaining a wide field optical view pre-/post-experiment but for others may require sample marking to assist in the use of colocation techniques. This correlative electrochemical multimicroscopy approach avoids the ambiguity inherent to classical macroscopic or bulk electrochemical techniques, where the electrochemical response is averaged over a large population of interacting surface sites, obscuring the nature of key elementary processes.
We have covered well-defined (single-crystal) surfaces (section 2), structurally and/or compositionally heterogeneous extended surfaces (section 3), and complex NP/support ensemble-type electrodes (section 4). While each configuration presents a unique set of challenges in terms of structure−activity resolution, one important aspect we have emphasized throughout is the importance of targeting the characteristic "single entities" that make up these electrochemical interfaces, to enable multiscale predictions that link the microscopic and macroscopic worlds. Another important aspect borne out of these studies is the importance of applying theory and developing computational models/simulations to rationalize experimental data, a combined approach that is becoming more accessible with the availability of powerful commercial software packages. Indeed, it is the consistency between experiment and theory at the single-entity (nanoscopic) level that allows macroscopic (bulk) behavior to be explained and further predicted, enabling materials discovery and the rational design of improved functional materials for technologically important applications such as energy conversion and storage.
Future avenues for high-resolution structure−activity measurements in electrochemistry that promise a holistic view of electrode dynamics will include improved in situ and in operando imaging techniques that incorporate electrochemical microscopy alongside other forms of microscopy, the integration of spectroscopic capability into imaging probes to enhance chemical information, the use of "intelligent probes" that make use of AI and machine learning, and methodologies that draw on the convergence in the scale of measurements and molecular dynamics and density functional theory simulations. There is no doubt that the techniques and technologies will continue to push the state of the art in terms of spatiotemporal resolution, measurement speed, and multifunctionality, enabling the routine study of phenomena on smaller length scales: for instance, catalytic sites at the atomic level (e.g., single-atom step defects 11 ) or mapping of local electrical double-layer properties (i.e., surface charge 57,58 ). Multifunctional, spatiotemporally resolved, and multilength scale studies inevitably generate large volumes of data, also necessitating the development and adoption of big data mining protocols in this field. We have only just scratched the surface of what is possible with single-entity electrochemistry and correlative electrochemical microscopy, but it is important not to lose sight of the importance of "ease of use" and "user friendliness"; future developments will also need to make the technologies accessible to the "general user" in order to facilitate the widespread adoption of these powerful techniques and methodologies.