Hybrid scanning electrochemical cell microscopy-interference reflection microscopy (SECCM-IRM): Tracking phase formation on surfaces in small volumes

We describe the combination of scanning electrochemical cell microscopy (SECCM) and interference reflection microscopy (IRM) to produce a compelling technique for the study of interfacial processes and to track the SECCM meniscus status in real-time. SECCM allows reactions to be confined to well defined nm- to  m-sized regions of a surface, and for experiments to be repeated quickly and easily at multiple locations. IRM is a highly surface-sensitive technique which reveals processes happening (very) close to a substrate with temporal and spatial resolution commensurate with typical electrochemical techniques. By using thin transparent conductive layers on glass as substrates, IRM can be coupled to SECCM, to allow real-time in situ optical monitoring of the SECCM meniscus and of processes that occur within it at the electrode/electrolyte interface. We first use the technique to assess the stability of the SECCM meniscus during voltammetry at an indium tin oxide (ITO) electrode at close to neutral pH, demonstrating that the meniscus contact area is rather stable over a large potential window and reproducible, varying by only ca. 5 % over different SECCM approaches. At high cathodic potentials, subtle electrowetting is easily detected and quantified. We also look inside the meniscus to reveal surface changes at extreme cathodic potentials, assigned to the possible formation of indium nanoparticles. Finally, we examine the effect of meniscus size and driving potential


Introduction
The ability to visualise and study phase formation and change is extremely important in materials science, to understand processes such as crystallisation, precipitation, and deposition. Practical applications include the development of new pharmaceuticals, 1 crop protection agents, 2 new energy storage materials, 3 among others. Many techniques are available to track phase formation, including optical, 4 fluorescence, 5 and scanning electron and transmission electron microscopies (SEM/TEM), 6 as well as Raman and infrared spectroscopy, 7,8 or scanning probe techniques like scanning tunnelling microscopy or atomic force microscopy (STM/AFM). 9,10 Application of microscopy with electrochemical control is particularly attractive as it provides a means to trigger, monitor and track (electro)chemical processes, ultimately at the single entity level. 11,12 More generally, the use of co-located microscopy techniques that probe and correlate structure and activity in the same space have the power to reveal surface processes in quantitative microscopic detail, 13 particularly when applied in operando. [14][15][16][17] Electrochemical techniques can be used to change solution composition in a controlled manner, by either consuming/producing reagents at the surface of an electrode immersed in a solution, [18][19][20] or by precisely mixing reagents using migration and electroosmotic flow in a nanopipette. [21][22][23][24] A particular advantage of electrochemistry is that the driving force can usually be controlled via the applied potential and mass transport can be varied over a wide range, and modelled with a high degree of accuracy, [25][26][27] allowing spatiotemporal changes in solution composition to be predicted quantitatively.
Recently a hybrid approach, using interference reflection microscopy (IRM) coupled to electrochemistry, was used to visualise and study the growth of ensembles of nanoparticles at optically transparent electrodes, in situ, at different overpotentials. [28][29][30] As a surface sensitive technique, IRM is particularly suited to the investigation of phenomena at the electrode/electrolyte interface. [31][32][33][34] At optically transparent electrodes, such as indium tin oxide (ITO) immersed in solution, local phase formation and/or change at the electrode surface modify the refractive index, giving rise to a difference in contrast in the observed reflected light image. For example, this phase change can arise from the growth of nanoparticles or crystals at the electrode surface providing real-time in situ observation of the phase formation event. 31 The sensitivity of IRM, or its analogue known as iSCAT, named for interferometric scattering microscope, 35 can be pushed towards single molecule imaging, or in the context of electrochemistry to the imaging of the restructuring of the electrochemical double layer of nanoparticles. 34 Although IRM is usually performed in large electrolyte volume, IRM can be coupled with spatially resolved electrochemical techniques to provide local electrochemical control and monitoring of phase formation events.
Scanning electrochemical cell microscopy (SECCM) provides a powerful platform to create very well defined and consistent microscale and nanoscale-sized electrochemical cells on various substrates. 36,37 Since its inception, 38,39 SECCM, and the scanning micropipette contact method (SMCM) as an earlier variant was named, has been widely employed to investigate electrochemical processes at the nanoscale, revealing heterogeneous properties of many features of materials including 2D materials, 36,[40][41][42][43] nanotubes, 44,45 grain boundaries and crystallographic facets in electrodes and electrocatalysts, [46][47][48] single particles, 49-53 among a rapidly expanding range of applications. 37 For phase formation and phase change processes, the ability to confine electrochemical reactions to small volumes allows the investigation of a few or even single events. 54,55 In this context, the use of SECCM in hopping mode, 56 where the SECCM meniscus is landed at a series of spots on a surface, is particularly powerful, as it is possible to build up large datasets with the same, or different, experimental conditions applied to each spot.
In this work, we combine SECCM with the surface sensitivity of IRM to investigate phase formation in tiny droplet cells at an ITO  Raman Spectroscopy. Raman spectra were recorded using a Raman microscope (Horiba LabRam HR Evolution) fitted with a charged couple device detector and a 660 nm OPSS laser.
Reactive-Transport Modelling. The time-dependent speciation inside the meniscus and pipette were simulated by numerically solving the diffusion equation for the system using the FEM in COMSOL Multiphysics (version 5.6, COMSOL Inc., USA). A typical axisymmetric geometry consisting of the end of the pipette and the SECCM meniscus was based on optical images of the pipette and the wetted area observed by IRM. We applied the reactivetransport model we previously reported for time-dependent CaCO 3 (aq) speciation. 21 Simulation details are further described in the FEM results section below.

Results and discussion
SECCM-IRM principles. The SECCM-IRM setup is presented schematically in Figure 1. It consists of an inverted optical microscope which focuses the light source, through an objective lens, onto the back of an ITO-coated glass coverslip. The top conductive side of the coverslip is used as the WE. The SECCM probe, loaded with electrolyte solution (50 mM KNO 3 , for the study of ITO; and 5 mM CaCl 2 and 5mM NaHCO 3 (pH 2.7 HCl) for the crystallisation experiments) and containing a QRCE, was coarsely brought closer to the ITO substrate using the micromanipulators and the stepper motor. The SECCM QRCE potential was controlled during the experiments, while the ITO substrate (WE) was held at a virtual ground. The resulting current was measured at the WE with a custom-built current follower. For ease of interpretation, all potentials reported are for the WE electrode, which is the negative value of the SECCM QRCE potential. Final tip approach was performed by applying a small bias between the QRCE and WE (+0.2 V for ITO studies and -0.1 V for the crystallisation experiments). The probe movement stopped automatically as soon the current was above a threshold, set at the noise level (ranging between 100 fA and 30 pA, depending on the sensitivity of the current follower that was appropriate for each probe size). The event indicated contact between the electrolyte meniscus formed at the end of the probe and the WE, resulting in a small electrochemical cell, with dimensions related to the tip-end diameter used. This event could also be detected in the IRM images (vide infra).
The optical detection principle has been detailed in previous works. 28,31,32 It is based on the reflection of the light wave at the ITO-electrolyte interface and its possible interference with light wave scattered by objects present on the ITO surface, which disturb the local optical conditions and are highlighted within the background, with the camera acting as an interferometric detector. Optical features can show either a negative contrast, the feature appearing darker than the background, or a positive contrast, where the feature appears brighter than the background. The sign of the contrast depends on the dielectric properties of the object relative to its environment and its size. 31,61 Once the meniscus had landed on the WE (no contact from the pipette itself), a short pause allowed monitoring of the meniscus stability and provided a background for later image treatment (vide infra). The potential was then swept or stepped to drive ORR (and water reduction) at the WE, with synchronous IRM visualisation of the meniscus footprint region.
Once the potential sweep or step was finished, the potential was switched back to the approach value. The probe was then retracted, to detach the meniscus from the surface, and the WE shifted laterally and the approach recommenced. In this way, the experiment could be repeated on a fresh ITO region, allowing replicates of the experiment ( Figure 1). IRM images were acquired synchronously throughout the entire experiment. The meniscus was rather reproducible among the 17 different experiments with an overall ±5 % variation in area, measured during the short pause after meniscus landing. Figure   2c gives the evolution of the ensemble-averaged surface area of 17 menisci during the potential cycle. The meniscus was rather stable during the forward potential scan from 0 V to -1.47 V, but its surface area increased when faradaic current flowed, in the solvent breakdown region. On the reverse scan, the meniscus did not retract to its original dimension. Note that the sudden increase in the meniscus size to a stable value at the start of the scan is associated with meniscus landing that we will analyse in detail elsewhere.
The inset of Figure  From the previously reported optical and SEM images, the density of In NPs is of the order of 10 8 cm -2 . 31 Thus, in principle, the confined region on the ITO WE accessed by SECCM facilitates the study of a single NP formation and growth event. For example, within a meniscus of diameter, d < 500 nm, such as the one monitored in Figure 2a, the probability of finding one nucleation site is below 1 for an average nucleation site density lower than 4/d 2 = 5 x 10 8 cm -2 .
Although previous studies of nucleation/growth of NPs with optical microscopy monitoring could optically isolate a single event, the electrochemical information (current) associated to such event was recorded for the entire electrode, averaged over the ensemble of NPs. [28][29][30]82 In contrast, with SECCM-IRM, the current is confined to the small SECCM meniscus, allowing the synchronous electrochemical monitoring and optical imaging of a single event. 83 Should SECCM allow isolation of a single nucleation event from the electrochemical transient, the question is: can IRM also visualize this individual event?
Image background subtraction is then a helpful technique to improve analyses of noisy data sets and it is often used in image processing in IRM. 28 The IRM image captured during the last instant before the potential cycle was switched on can be used as the background image, which is taken to be equivalent  between different polymorphs. 87 The product of crystallisation is also heavily dependent on the reaction volume, with small (micron-size) volumes stabilising early-stage, less stable forms (e.g. amorphous calcium carbonate, ACC). 16 The early stages of CaCO 3 precipitation remain hotly debated, [88][89][90] although it is known to proceed via rapid formation of ACC at moderate supersaturations.
To control the precipitation of CaCO 3 we modify the local supersaturation, S: where is the activity of species i, and K SP is the solubility product of calcite, by changing the pH. [18][19][20] Starting with an acidified (ca pH 2.7, therefore undersaturated, Figure 4a procedure was repeated by means of successive approaches at different location of the ITO surface to form an array of points, typically with 5-8 repeats at each potential. Whole, ca. 8 µm wide, meniscus IRM images, after 10 s of polarisation at three different potentials, are shown in Figure 5, along with the current-time (I-t) traces recorded ( Figure 5a). The dark contrasted area delineates the meniscus region. At less negative potential (Figure 5bi), the current was negligible and only a few bright spots are observed by IRM. At -0.7 V (Figure 5bii), however, a more significant current was recorded, and multiple spots, now dark, are observed in the images. This trend continues at -0.8 V (Figure 5biii), but at -0.85 V (not shown), although the current magnitude was still increasing, no clear spots inside the meniscus were seen. Feature detection was hampered by a general darkening of the spot area at this potential and below. The I-t traces in all cases where E < -0.5 V shows non-monotonic behaviour. An initial spike to negative currents is followed by a rapid decay, and the onset of a sustained increase towards more negative currents. This is unexpected for an ORR pulse at ITO and could suggest that either the crystallisation events or a separate process at the ITO, possibly the formation of In NPs (vide infra), are increasing the surface activity.
The size of the optical features seen in the IRM images can be estimated by their FWHM intensity, which is calculated by fitting a Gaussian function to the intensity profiles.
The smallest FWHM associated with the dark features in Figure 5bii and 5biii are 185 ± 25 nm (from 30 spots -with an example shown in Figure 5bii). Optical features characterized by this low limit value of FWHM can be considered as diffraction-limited, suggesting that nanometric particles (< 200 nm) have been generated. The bright features observed at -0.5 V ( Figure 5) increase in intensity with time, in contrast to a decrease observed for the lower potentials.
Such behaviour is attributed to the difference in size of the particles formed, as was demonstrated in earlier work for gas nanobubbles: bubbles larger than ca. 300 nm appeared lighter than the background, those smaller appeared darker. 32 Despite a difference in the shape and refractive index of the CaCO 3 deposits, we might expect a similar behaviour. the ITO surface. These large negative driving potentials will lead to more rapid formation of the supersaturation condition, 19 and faster nucleation generally results in more nuclei and hence particles. Dense layers of ACC are also known to form at interfaces at appropriately high supersaturations. 91 At such negative potentials and low pH, the reductive degradation of ITO may also occur. 32 The formation of CaCO 3 crystals is supported by ITO backgroundsubtracted Raman micro-spectroscopy of a dried meniscus deposited at -0.7 V, which reveals spectra with peaks consistent with the presence of CaCO 3 particles (Figure 5c). A broad peak at 868 cm -1 is consistent with either the ν 3 mode of amorphous CaCO 3 (863 cm -1 ) or calcite (873 cm -1 ), while a small feature at 714 cm -1 is also consistent with calcite. 92

Controlled CaCO 3 crystallisation at larger meniscus.
Using pipette probes of tens of µm in diameter allows investigation of even larger areas and the production of a population of hundreds or thousands of electrochemically-induced crystallisation events. At this length scale surface heterogeneities, concentration gradients over the meniscus volume and localized fluxes may contribute to the SECCM-IRM observations. Figure 6 shows example IRM images taken after 8 s for two different applied potentials at the ITO electrode. When the experiment was repeated without CaCl 2 present in the solution (Figure 7a) a few spots were observed in the IRM images for E -1.0 V. For more negative potential, ≤ and for the considered acidic pH, it is known that ITO is unstable (with respect to reduction of In to metallic NPs), which has been observed, along with H 2 nanobubbles, by IRM previously. 32 The features observed here (Figure 7a) are different in shape and appear in much small numbers than those observed when Ca 2+ is present. To assess the effect of possible ITO degradation to the overall intensity change seen in IRM, cyclic voltammetry in acid media (HCl solution, pH 2.7) was carried out for aerated and deaerated (with nitrogen) solutions and over 2 different potential windows. Figure 7b   Two approach distances (meniscus heights -d app ) corresponding to the outer radius of the micropipette (31 m), and half of this value, were modelled. This is because the meniscus thickness impacts O 2 transport at the air-water interface (and hence the current distribution). 75 We did not measure the meniscus thickness in this work, but for this size pipette it should be achievable in the future through the use of laser scanning confocal microscopy and an appropriate fluorophore. 62 Indeed, the use of a pH-sensitive fluorophore would also confirm the pH distribution in these types of experiments. 93 To achieve a realistic estimate of the supersaturation we consider the 12 most significant species in the solution (the ions Ca 2+ , Cl -, Na + , HCO 3 -, CO 3 2-, H + and OH -, the ion pairs NaCO 3 -, CaCO 3 0 , NaHCO 3 and Ca(HCO 3 ) + and the gas phase species O 2 and CO 2 ) based on prior modelling with the PHREEQC speciation code. 94 We solved the diffusion equation (2) The equilibrium concentrations given by the PHREEQC code were applied as the boundary condition at a plane 3500 μm inside the micropipette. Given the apparent importance of spatial variation across the wetted area, we imposed atmospheric equilibrium concentrations of O 2 and CO 2 at the air/meniscus interface (273 μM and 13.6 μM, respectively). A consequence of this approximation is that the total concentration of carbon species in the meniscus is depleted relative to that in the micropipette. We note that the alternative limiting treatment of CO 2 (no exchange with the atmosphere) would result in greater buffering and therefore longer timescales for the pH swing. Using the simplifying assumptions that all current arises from the ORR, and that for a given potential, ORR is first order in O 2 , we imposed an inwards flux, of O 2 and an outwards flux of OHat the ITO surface: where we chose k such that the simulated current over the ITO area,  Both parameters have a strong spatial dependence, with change starting at the meniscus/air interface where the O 2 flux is greatest. Combined with the buffering action at the centre of the droplet due to diffusion from inside the micropipette, this explains the experimentally observed preference for nucleation to occur around the edges of the meniscus rather than at its centre. The pH buffering by atmospheric CO 2 may also contribute to the absence of particles at the very edge in the -1.0 V case (Figure 6bii). At longer times, at -1.0 V the pH of the meniscus rapidly equilibrates and the spatial variation in S and pH is lost, however at -0.8 V slight variation in S and pH across the surface is observed to persist. Closer approaches (to d app = r/2, Figure 8(d)) show a larger degree of spatial variation in pH and S. This simplified model is able to rationalise the timescale over which new features are observed and, to some extent, their spatial variation, emphasising the importance of FEM modelling in interpreting SECCM deposition experiments.

Conclusion
We simulations has highlighted the origins of the crystallisation patterns and we were able to predict the onset time and, to an extent, the location of the crystallisation events.
More generally, this work establishes a foundation for the use of SECCM-IRM in nanoscale (electro)chemistry. In future work we shall explore the landing and pull-off of the SECCM meniscus on (electrode) surfaces, during translation of the probe to and from the surface, also expanding the work on ITO to other transparent electrode materials.
Particularly, we will investigate electrode materials resistant to the electrochemically generated corroding environment, e.g. in this work the CaCO 3 electrocrystallisation and its visualisation are complicated by the electrochemical etching of ITO under some conditions.
Transparent thin Au coated glass coverslips could be used, as previously demonstrated for the optical monitoring of the electrochemistry of single NPs. 28,29,31,95 Coverslips coated with graphene-based materials are also possible electrode candidates, as shown in the IRM imaging of the electrochemistry of graphene oxide. 33,95 Finally, reflectivity microscopy can also be operated at reflective and non-transparent electrodes 96