Multi-band Metasurface-Driven Surface-Enhanced Infrared Absorption Spectroscopy for Improved Characterization of in-Situ Electrochemical Reactions

Surface-enhanced spectroscopy techniques are the method-of-choice to characterize adsorbed intermediates occurring during electrochemical reactions, which are crucial in realizing a green and sustainable future. Characterizing species with low coverage or short lifetimes has so far been limited by low signal enhancement. Recently, single-band metasurface-driven surface-enhanced infrared absorption spectroscopy (SEIRAS) has been pioneered as a promising technology to monitor a single vibrational mode during electrochemical CO oxidation. However, electrochemical reactions are complex, and their understanding requires the simultaneous monitoring of multiple adsorbed species in situ, hampering the adoption of nanostructured electrodes in spectro-electrochemistry. Here, we develop a multi-band nanophotonic-electrochemical platform that simultaneously monitors in situ multiple adsorbed species emerging during cyclic voltammetry scans by leveraging the high resolution offered by the reproducible nanostructuring of the working electrode. Specifically, we studied the electrochemical reduction of CO2 on a Pt surface and used two separately tuned metasurface arrays to monitor two adsorption configurations of CO with vibrational bands at ∼2030 and ∼1840 cm–1. Our platform provides a ∼40-fold enhancement in the detection of characteristic absorption signals compared to conventional broadband electrochemically roughened platinum films. A straightforward methodology is outlined starting with baselining our system in a CO-saturated environment and clearly detecting both configurations of adsorption. In contrast, during the electrochemical reduction of CO2 on platinum in K2CO3, CO adsorbed in a bridged configuration could not be detected. We anticipate that our technology will guide researchers in developing similar sensing platforms to simultaneously detect multiple challenging intermediates, with low surface coverage or short lifetimes.


Introduction
The study of intermediates occurring during electrochemical reactions remains a challenge due to low surface coverages and short lifetimes.Raman and IR spectroscopy are powerful in-situ optical characterization techniques that can detect the rotational or vibrational modes of molecules.During the CO2 reduction reaction (CO2RR) a key intermediate is CO, which is detectable with IR spectroscopic techniques. 1 However, as IR characterization techniques suffer from high losses by aqueous electrolyte, two geometries are typically used to acquire IR spectra of molecules adsorbed at electrode surfaces.Firstly, the external reflectance approach consists in squeezing a thin layer of electrolyte between the working electrode and a prism to reduce the optical losses. 2,3The advantage of the external reflectance approach is a high degree of freedom in choosing the morphology and thickness of the studied material.However, although the electrolyte layer is thin it still suffers from dampened signals due to light absorption from the electrolyte.Moreover, due to the requirement of a thin electrolyte layer diffusional processes are strongly limited, making the external reflectance approach an inadequate technique for in situ studies of electrocatalytic reactions, such as the CO2RR.
3][4] As light is totally reflected on a surface, evanescent waves with an exponential decay probe the other side of the reflective surface.Due to the evanescent approach, the losses stemming from the electrolyte are minimized.This so-called Kretchmann configuration leaves the electrode freely accessible for transport processes to and from the electrolyte.However, it requires thin film electrodes as the quickly decaying evanescent waves must be able to penetrate the electrolyte past the metal/electrolyte interface to probe the analyte.But, as the metal layer is decreased in thickness the electrochemical stability decreases. 3In addition, ATR infrared absorption spectroscopy at plane electrodes yields only low IR signals.This disadvantage is overcome when rough film electrodes are prepared, e.g. by electrochemical roughening.Then, the electron density of nanoparticles with a linear dimension of the order of the wavelength of IR irradiation can come into resonance with the electromagnetic wave, leading to an enhanced vibrational signature. 5The latter methods is usually referred to as ATR surface enhanced infrared absorption spectroscopy, or, in short as ATR-SEIRAS.However, with ATR-SEIRAS besides the stability of the film electrodes, reproducibility of the acquired spectra becomes an issue.The intensity of the vibrational bands strongly depends on the distribution of the nanoparticle size resulting in an uncontrolled and random signal enhancement making this technique difficult to reproduce and often unreliable. 6ecently, a promising nanophotonic-electrochemical platform based on ATR-SEIRAS was designed employing a nanostructured platinum surface to characterize CO during CO oxidation. 7e precisely nanostructured platinum metasurface that integrated SEIRAS with cyclic voltammetry for the study of electrochemical interfaces provided a clear and reproducible method to produce SEIRAS-active electrodes.Specifically, the electric near-field enhancement produced by the platinum-based nano-slot metasurface was shown to amplify the in-situ generated signal traces of the vibrational mode of linearly adsorbed CO (COlinear) at 2033 cm -1 .Changes in the reflection intensity based on the coupling of the metasurface-driven resonances and the vibrational modes of the adsorbed species controllably enhanced their characteristic signal traces.However, the reported nanophotonic-electrochemical platform only featured one resonance and could therefore only spectrally target and enhance one molecular vibrational mode.
Here, we develop multi-band metasurface-driven SEIRAS for the improved characterization of in situ electrochemical reactions (Figure 1a).We demonstrate its successful operation during the CO2RR where the molecular signals of two configurations of adsorption of CO on platinum are resolved: on-top (COlinear) and bridged bound (CObridge) CO molecules, respectively (Figure 1b).So far, little attention has been paid to the involvement of bridge site configurations of CO on Pt during the CO2RR.][10][11] A fundamental and systematic study of bridge site configurations during the CO2RR in alkaline media is still missing.We use a platinum nano-slot metasurface on a CaF2 substrate featuring two arrays that were each numerically modeled and tuned to spectrally target one of the two aforementioned characteristic molecular vibrations of the CO2RR.For each vibrational mode, a unique array with a spectrally targeted resonance was fabricated.Each array locally enhanced the electric near fields and enhanced the corresponding molecular signal traces.Our multiband approach can be extended to multiple vibrational bands.SEIRAS was performed in an ATR geometry (Figure 1c) to maintain free accessibility of the electrode surface and minimize the contribution of the electrolyte to the IR spectrum.We validated the nanophotonic-electrochemical platform by following the oxidation of CO into CO2 during a cathodic polarization, measuring simultaneously the top and bridge site adsorption of CO.The clear detection and enhancement of the top and bridge adsorption configurations of CO on Pt were confirmed by observing the typical Stark shift in the molecular signal traces.Then, we followed the reduction of CO2 in situ to study the same adsorption sites.The contrast between the CO saturated signal traces compared to those measured during the CO2RR provided conclusive insights into the adsorption characteristics of intermediates emerging during the CO2RR.Our results suggest that the CO2RR in alkaline environments proceeds via CO molecules adsorbed mainly as COlinear and not as CObridge.Finally, we established a methodology to implement similar multi-band nanophotonic-electrochemical platforms, providing a framework for future research in this area.stemming from two metasurface arrays on Pt.The shared geometrical parameters were h=30 nm, w=180 nm, py=1420 nm, px-l=230 nm with the slot length being swept from (blue) l=1370 nm to (red) l=1580.(e) Sketch of the unit cell.A 1 nm thick Ti adhesion layer was utilized in the structure's fabrication for improved adhesion of Pt on CaF2.

Numerical Simulations.
The numerical simulations in this study utilized CST Studio Suite (2021) which uses the finiteelement frequency-domain Maxwell solver.For the simulation of CaF2, a refractive index of 1.4 was used, while the surrounding medium was represented by water with a refractive index of 1.33.
Platinum was modeled using its experimental complex refractive index data. 12To introduce linearly polarized light at an angle of incidence of 72° into the system, an impedance-matched open port with a perfectly matched layer was employed.Since light experiences total internal reflection at this angle at the CaF2-Pt interface, the boundary opposite the open port was set as a perfect electric conductor.The unit cell was defined and simulated as an infinite periodic array using Floquet boundaries.

Analytical analysis of resonances.
The resonances were characterized in terms of their radiative (γrad) and intrinsic (γint) damping rates, from which their total Q-factor could be determined as = ( ) , where ν0 is the central wavenumber.We employed temporal coupled mode theory 13 according to Ref. 14 describing a resonator with a single port that supported reflected waves and a single resonance that coupled to the far-field via the coupling constant = 2 .Additionally, an intrinsic loss channel introduced damping to the resonance at a rate γint.The reflectance spectra R were fitted by where ν is the wavenumber.

Multi-band Metasurface Fabrication.
The multi-band metasurface fabrication was similar to the protocol provided in Ref. 7 Instead of one array, two arrays were fabricated ca.500 µm apart to fit within the window of the focal plane array of the IR-spectrometer.The arrays were each ca.3×2 mm in size showing that large arrays could be made.The arrays had to be large as the Fourier transform IR spectroscopy device did not have and did not need a focusing microscope objective.We chose CaF2 as the substrate due to its transparent properties within the mid-IR spectral range, as well as its high chemical stability and low solubility.Before the experiments, the substrate underwent a thorough cleaning process, involving an acetone bath in an ultrasonic bath, followed by oxygen plasma treatment.
Subsequently, the substrate was spin-coated with an adhesion promoter (Surpass 4000), followed by a layer of negative-tone photoresist (ma-N 2403).The photoresist was baked at 100 °C for 60 s, and a conducting layer (ESpacer 300Z) was deposited using spin coating.The metasurface patterns were generated by defining a unit cell and replicating it in both the x and y directions.
Electron-beam lithography (Raith Eline Plus) was employed to write the patterns, utilizing an acceleration voltage of 30 kV and an aperture of 20 µm.The exposed resist was developed in ma-D 525 for 70 s at room temperature.Thereafter, a titanium adhesion layer (1 nm at 0.4 Å s −1 ) and a platinum film (30 nm at 2 Å s −1 ) were deposited on the patterned surface using electron-beam evaporation (PRO Line PVD 75, Lesker).Finally, the fabrication process was completed with an overnight lift-off process in mr-REM 700.For the in-situ SEIRAS measurements, a pure 30 nm thick platinum film on a 1 nm titanium layer on CaF2 was utilized as a reference.
Surface-Enhanced Infrared Absorption Spectroscopy Measurements.The behavior of the nano-slots was characterized in CO-saturated electrolyte.After bubbling CO for 2 hours, cyclic voltammetry was performed from +1650 mVRHE to -85 mVRHE, using a scan rate of 0.25 mV.s -1 .

SEIRAS was conducted using a
Finally, we saturated the electrolyte with CO2, which decreased the pH to approximately 8.After 2h of gas bubbling, cyclic voltammograms were measured with a slow scan rate of 0.25 mV s −1 , from the OCP to + 1425 mVRHE and then back to +25 mVRHE.SEIRAS spectra were acquired in intervals of 100 mV.

Results and Discussion
Numerical Design of Catalytic Multi-Band Nano-Slot Metasurface.
We start the implementation of our catalytic multi-band nano-slot metasurface by defining the fundamental unit cell for the numerical simulations consisting of a solitary slot within a continuous platinum film immersed in water on CaF2 (Figure 1e).The geometrical parameters of the unit cell can be changed to tune the resonance strength and position of the metasurface.As the goal of our investigations was multi-band signal enhancement, the unit cell parameters were separately tuned to two resonance frequencies matching the vibration frequencies of two different adsorption configurations of CO on platinum.Specifically, the goal was a straightforward approach to create two adjacent nanostructured arrays on platinum that would each enhance one of the two vibrational frequencies.
A common approach to achieving this is to scale all geometrical dimensions of the system at the same time, used for example in biosensing 15 and catalysis 16 .However, constraints in the fabrication with negative resists limited the unit cell length in y to a minimum of 1.4 µm due to proximity effects.Proximity effects arise due to the scattering of electrons in the resist and substrate due to exposure of an electron beam. 17We found that the slots merged and the quality of the lift-off procedure decreased as the unit cell length in y was decreased below 1.4 µm.Therefore, we simplified the resonance tuning protocol allowing only the slot length to change and taking into account a shift of ca.80 cm -1 between simulation and experiment.By increasing the slot length from 1370 nm to 1580 nm while leaving all other parameters unchanged the metasurface-driven resonance was redshifted by ca. 150 cm -1 (Figure 1d), corresponding to the spectral separation of the vibrational modes.
The Rayleigh anomaly was tuned to the smaller frequency side of the resonance to ensure optimal sensing performance.To improve the characterization of the resonances, the fitting model was adapted specifically for total internal reflection (see Experimental section).The Q-factor and coupling ratio γe/γi were obtained by fitting the simulated resonance in reflectance using temporal coupled mode theory (see Experimental Section).Based on our simulations, the multi-band nanoslot metasurface achieved a modulation of ca.84% and 88% in reflection, a Q-factor of 4.3 and 5.0, and a ratio of external to intrinsic coupling of 2.6 and 2.1 for the higher and lower frequency resonances, respectively.Our system can be further optimized by maximizing the modulation in reflection or absorption to push it toward its critical coupling condition.

Multi-band Metasurface Characterization.
According to the literature, the vibrational modes of the COlinear and CObridge are expected to occur at ca. 2050 cm -1 18-22 and 1850 cm - 1 18,19,21,23 , respectively.First, we characterized the optical properties of our multi-band metasurface in electrolyte saturated with Ar and CO using SEIRAS.
An example of the heat map obtained by integrating the IR spectra collected by the focal plane array detector is provided in Figure 2a.The two pink-colored areas correspond to the two metasurface-arrays designed to enhance CO detection.The quality of the fabricated slots was verified by scanning electron microscopy images (Figure 2b and 2c).The signal received by the high-frequency array (Figure 2a, bottom array) was averaged, resulting in a resonance spectrally positioned at ca. 2030 cm -1 in the Ar saturated electrolyte (Figure 2d).On the other hand, the average signal from the low-frequency array (Figure 2a, right array) produced a resonance located at ca. 1860 cm -1 (Figure 2e).In both cases, the system presents a near critically-coupled behavior between the metasurface-driven resonances and the vibrational modes of CO, 7 leading to a dip in the high and low-frequency resonances at 2046 cm -1 and 1848 cm -1 , respectively.To extract the signal of the COlinear and CObridge vibrational modes more clearly, then, the reflectance spectra with, R, and without, R0, adsorbed CO were converted into their differential absorbance, − ln , to separate the metasurface-driven resonance from the CO signals (Figure 2f and 2g).Furthermore, a comparison of the differential absorbance in the regions of the COlinear and CObridge, obtained on an unstructured Pt film (30 nm thick) and with our multi-band nano-slot metasurface, indicates that the high-frequency array exhibited a signal enhancement of 41.This signal enhancement is higher than the enhancement reported in our previous work, 7 attributed to an improved liftoff procedure.The signal enhancement provided by the low-frequency array cannot be reliably estimated, as the CObridge signal was not clearly distinguishable on the unstructured Pt film.The CObridge signal could only be observed here with the multi-band metasurface due to its high signal enhancing properties.Behavior in CO saturated electrolyte.
As a proof-of-concept, the behavior of the multi-band nano-slot metasurface in 0.5M K2CO3 saturated with CO is provided here by using electrochemical voltammetry.A cathodic scan was conducted with a rate of 0.25 mV.s -1 starting at 1650 mVRHE, as shown in Figure 3a.As already observed in the literature, 7,24,25 the increase in the current around 950 mVRHE followed by a plateau corresponds to the oxidation of CO into CO2, which is limited by CO diffusion from the bulk of the electrolyte to the Pt surface.Then, the following decrease in current at 350 mVRHE indicates the end of the region where CO was oxidized.Finally, the slight drop in current around 0 mVRHE can be attributed to the hydrogen evolution reaction.SEIRAS spectra were acquired at intervals of 100 mV during a cathodic scan.The potential regions in which CO was or was not detected are shown as red and blue regions, respectively.
The evolution of the IR spectrum with the electrical polarization of the high (Figure 3b) and lowfrequency (Figure 3c) arrays shows the successful detection of the COlinear and CObridge, respectively.The vibrational modes appeared as peaks in the differential absorbance spectra.
Regarding the behavior of the COlinear, a spectral shift of approximately 63 cm -1 .V -1 was observed (Figure 3d) from 450 mVRHE to -50 mVRHE, which is in line with the literature 7,[26][27][28][29] and can be attributed to either a higher π-back-donation from the metal to CO 27,30 and/or to the Stark effect. 28,30,31Additionally, a significant spectral redshift attributed to the decrease of the dipoledipole interactions as the coverage decreases 32,33 was resolved from 650 to 450 mVRHE (Figure 3d), thanks to the high resolution achieved with our platform.Concerning the behavior of the CObridge on the low-frequency array, a distinct peak was resolved exhibiting a similar behavior as the COlinear.However, the observed Stark shift in this case was smaller, resulting in approximately 21 cm -1 .V -1 .This difference in Stark shift between COlinear and CObridge has been observed in some other literature. 28,29However, the origin of this difference is still controversial as other authors suggested that both configurations should provide the same Stark shift. 34,35Our observations can be explained by the smaller IR cross-section of the CObridge compared to the COlinear. 34When CO is adsorbed linearly on the platinum surface, it strongly binds to the Pt atoms, 36 which could explain the larger Stark shift.Furthermore, the electric field from the metal surface could have affected the COlinear more significantly, leading to a larger change in its vibrational frequency.On the other hand, the configuration of the CObridge may have resulted in a weaker interaction with the metal surface. 36In that case, the electric field from the metal surface could have had a smaller impact on this configuration, resulting in a smaller Stark shift.
At higher potentials, the redshift with decreasing coverage was observed next to a broadening of the peak from 550-750 mVRHE, which needs further investigation.A Fano-type asymmetric peak was observed at the same position as the COlinear due to the off-resonance coupling between the resonance of the low-frequency array and the vibrational mode of the COlinear. 37Interestingly, the area of the COlinear peak slightly decreased from 650mVRHE to -50mVRHE, while the area of the CObridge peak showed a continuous increase (Figure 3e).The area and intensity of the CO peaks are related to the coverage of the COlinear and CObridge, suggesting a transfer from a COlinear to a CObridge configuration as the cathodic potential increased.According to some authors, 29,36 the competition between CO and hydrogen adsorption on Pt in the cathodic region could have been responsible for this transition.Furthermore, the barrier for CO diffusion from a top site to a bridge site was theoretically predicted to be very small. 38These findings can explain the increase in the area of the peak attributed to the CObridge to the detriment of the COlinear peak area, which decreases.
To conclude, our multi-band nano-slot metasurface selectively and simultaneously enhanced and detected with a high accuracy the behavior of the COlinear and CObridge.In the next section, the study focused on the behavior of adsorbed CO during the CO2 reduction reaction.

Reduction of CO2.
0][41] A second peak appeared around 250 mVRHE, which is attributed to CO2 reduction. 22The drop in the current starting at 0 mVRHE is due to the hydrogen evolution reaction and the positive current peak observed after reversing the scan direction stems from the hydrogen oxidation reaction. 28Finally, a small oxidative peak was observed around 550 mVRHE and is attributed to the oxidation of the previously formed CO. 42 Looking at the IR spectra of the high-frequency array optimized for the COlinear during the cathodic scan (Figure 4b), a peak was observed around 200 mVRHE.This peak indicates the presence of adsorbed CO, which is directly correlated to the reduction peak observed in the voltammogram.Moreover, this peak became more pronounced and more defined with higher cathodic polarizations.
Moving to the low-frequency array optimized for the CObridge detection, at the highest applied cathodic polarization (0 mVRHE) a small peak became discernible at around 1785 cm -1 in the IR spectra (Figure 4c).This peak is hardly above the detection limit despite the high resolution achieved with our multi-band nanophotonic-electrochemical platform and appeared at significantly higher cathodic potentials (200 mV difference) than the COlinear peak obtained with the high-frequency array.Our findings suggest that the CObridge is not significantly involved in the CO2 reduction process on Pt.Some authors 19,43 have suggested that the favorable configuration of adsorption of CO on Pt is the COlinear.Moreover, the literature suggests that the CObridge formation occurs once the CO coverage approaches its maximum limit where a transfer from the COlinear to the CObridge configuration takes place. 29,36ring the anodic scan, the COlinear peak maintained a constant amplitude (Figure 4d) but exhibited a classic Stark shift between 0 to 400 mVRHE, followed by an intensity decrease and a small redshift attributed to reduced coverage around 500 mVRHE.Then, the peak disappeared, which can be directly correlated with the oxidation peak observed in the voltammogramm (Figure 4a).In contrast, the peak attributed to the CObridge disappeared at 100 mVRHE (Figure 4e), supporting our conclusion that the CObridge is not significantly involved in the CO2 reduction on platinum in an alkaline environment.The distortion of the baseline at around 2000 cm -1 is attributed to the strong off-resonance coupling between the metasurface-driven resonance of the low-frequency array and the vibrational mode of COlinear. 37Thanks to the high resolution and signal enhancing properties provided by our nanophotonic-electrochemical platform, Figure 4 suggests that the CObridge is not significantly involved in the reduction of CO2 in alkaline electrolyte.][10][11] The difference in the CO adsorption behavior between alkaline and acidic media could be due to a modification of the competition with the hydrogen evolution reaction.5][46] The Heyrovsky reaction involves the formation of hydroxide ions near the Pt surface, which could interact with adsorbed CO2 and CO to favor the formation of the COlinear.These considerations are in agreement with our observation that CObridge did not significantly participate in the CO2RR in K2CO3, i.e., in an alkaline environment.

Conclusion
We have developed a multi-band nanophotonic-electrochemical platform enabling enhanced simultaneous in-situ characterization of two adsorption configurations of CO on Pt during the electrochemical reduction of CO2.Our platform provided an enhancement over conventional systems by a factor of over 40.Crucially, our platform was able to detect the CObridge configuration, which could not be detected here with an unstructured Pt film.Using a straightforward and easily reproducible methodology, we numerical modeled, fabricated, and tested our platform.The COlinear and CObridge configurations were characterized in CO saturated electrolytes, highlighting a transition from top to bridge site configurations at high cathodic potentials and demonstrating the high resolution provided by our platform.The vibrational modes of CO were confirmed via their typical Stark shift.Our final experimental tests focused on the characterization of the CO2RR.
Interestingly, we found that during the CO2RR in an alkaline environment, the CObridge configuration does not play a significant role, whereas the COlinear was successfully detected.This finding could be attributed to the competition with the hydrogen evolution reaction in alkaline environments.We anticipate that our multi-band nanophotonic-electrochemical platform provides a new strategy to study electrochemical reactions with low coverage or transient features by providing a higher resolution than conventional systems for now limitless IR-active vibrational modes.

Figure 1 .
Figure 1.Concept and numerical design of the multi-band nanophotonic-electrochemical platform.(a) Schematic of the multi-band platinum-based nano-slot metasurface to study in-situ CO2 reduction.(b) Schematic showing the chemical structure of the two adsorption configurations of CO on platinum.(c) Schematic illustrating the metasurface in an electrochemical chamber filled with electrolyte that was illuminated from below in an ATR geometry.(d) The resonances Vertex 80 spectrometer equipped with an IMAC Focal Plane Array macro imaging accessory from Bruker.A specular reflection unit (VeeMax III from PIKE Technologies) paired with a CaF2 prism and light polarizer introduced light at 72° w.r.t. an electrochemical jackfish cell, thereby enabling the attenuated total internal reflection geometry.A focal plane array detector (64 × 64 MCT detectors) was used to characterize the optical properties of the multi-band nano-slot metasurface.By integrating the measured (absorbance) signal across the spectral range corresponding to the metasurface-driven resonances (1600-2800 cm −1 ) a pixelated two-dimensional heat map of the sample was created to identify the pixels corresponding to the nanostructured arrays (Figure2a).Then, the pixels corresponding to each array were averaged to construct the final spectra.The samples were cleaned via electrochemical cycling.Prior to each measurement, an initial background was acquired using p-polarized light.The metasurface-driven resonances were measured in situ using s-polarized light.Each spectrum was acquired at a resolution of 4 cm −1 and by averaging 10 scans.The data was treated by applying a baseline correction and Savitzky-Golay filter.The enhancement provided by the multiband metasurface was determined by comparing the area of the peaks associated with the vibrational modes of the COlinear and CObridge configurations to the corresponding measurement performed on an unstructured Pt film (30 nm).Electrochemical Measurements.A classical three-electrode system was implemented by using the Pt multi-band metasurface as the working electrode, a Pt wire as counter electrode and a saturated calomel electrode (E = 0.244 VSHE) as reference electrode.The electrolyte was a 0.5 mol.L -1 K2CO3 solution (pH 11.9), saturated with either Ar, CO, or CO2.Before the first characterization, the cleanliness of the electrode surface was verified by performing a cyclic voltammogram at a scan rate of 20 mV s −1 in Arsaturated electrolyte.

Figure 2 .
Figure 2. Characterization of the multi-band nano-slot metasurface.(a) Heat map of the metasurface obtained by integrating the SEIRAS signal from 1600 to 2800 cm -1 .(b-c) SEM pictures of the (b) high and (c) low-frequency arrays, indicated in (a) via blue and red dashed boxes, respectively.(d-e) Resonances with and without CO for the (d) high and (e) low-frequency arrays in 0.5M K2CO3.(f-g) Comparison of the differential absorbance obtained with the (f) high and (g) low-frequency array with an unstructured Pt film (30 nm thick, inset) and with the multiband nano-slot metasurface.

Figure 3 .
Figure 3. Cathodic polarization of the multi-band metasurface in 0.5M K2CO3 saturated with CO.(a) Evolution of the current density during the cathodic polarization at 0.25 mV.s -1 .(b-c) Evolution of the IR spectra with the potential acquired by the arrays optimized for the (b) COlinear and (c) CObridge detection.(d-e) Evolution of the (d) position and (e) area of peaks with the potential.

Figure 4 .
Figure 4. Cyclic voltammetry of the multi-band metasurface in 0.5M K2CO3 saturated with CO2.(a) Evolution of the current density during the polarization at 0.25 mV.s -1 .A zoom-out of the current density is shown.(b-c) Evolution of IR spectra with the potential acquired on the (b) highfrequency array optimized for the COlinear and (c) low-frequency array optimized for the CObridge detection during the cathodic scan.(d-e) Evolution of IR spectra with the potential acquired on the