Tailoring Copper Foam with Silver Dendrite Catalysts for Highly Selective Carbon Dioxide Conversion into Carbon Monoxide.

The present study outlines the important steps to bring electrochemical conversion of carbon dioxide (CO2) closer to commercial viability by using a large-scale metallic foam electrode as a highly conductive catalyst scaffold. Because of its versatility, it was possible to specifically tailor three-dimensional copper foam through coating with silver dendrite catalysts by electrodeposition. The requirements of high-yield CO2 conversion to carbon monoxide (CO) were met by tuning the deposition parameters toward a homogeneous coverage of the copper foam with nanosized dendrites, which additionally featured crystallographic surface orientations favoring CO production. The presented results evidence that Ag dendrites, owing a high density of planes with stepped (220) surface sites, paired with the superior active surface area of the copper foam can significantly foster the CO productivity. In a continuous flow-cell reactor setup, CO Faradaic efficiencies reaching from 85 to 96% for a wide range of low applied cathode potentials (<1.0 VRHE) along with high CO current densities up to 27 mA/cm2 were achieved, far outperforming other tested scaffold materials. Overall, this research provides new strategic guidelines for the fabrication of efficient and versatile cathodes for CO2 conversion compatible with large-scale integrated prototype devices.


Introduction
Re-utilization of carbon dioxide (CO 2 ) is deemed as one of the key challenges to lower the atmospheric CO 2 levels and steer towards a more sustainable future. In this regard, the electrochemical CO 2 reduction reaction (CO 2 RR) comes to the fore as an alluring way to convert CO 2 into fuels, chemical and materials. Simultaneously, such an electrochemical process can furthermore allow using on demand excess energy from renewable energy plants. 1 However, to make this technology available at industry level, i.e. to bridge the gap between academic innovation and commercialization, versatile and cost-effective processes at large-scale are still in dire need. In addition to that, highly selective and efficient electroreduction of CO 2 into value-added products is a crucial requirement. The numerous CO 2 RR products range from two electron products, such as carbon monoxide (CO) and formic acid (HCOOH), [2][3][4][5] to multi-electron products such as methane (CH 4 ), 6 methanol (CH 3 OH), 7 ethylene (C 2 H 4 ), 8 and ethanol (CH 3 CH 2 OH), 9 as well as other compounds. 10 Due to the complex reaction pathways of CO 2 RR, high yield selective product generation remains challenging, and, so far, has only been achieved for CO and formic acid in aqueous solutions. CO is of particular interest, as it can be efficiently separated from other reaction products (separation of liquid products is much more energy-intensive) and because together with hydrogen (H 2 ) it can be valorized to synthesis gas. From this versatile gas mixture, liquid transportation hydrocarbon fuels and various other value-added chemical products can be produced via the well-known and established Fischer-Tropsch synthesis. 11 As reported to date, high yield and selective production of CO from CO 2 relies on the utilization of precious metal catalysts, typically gold (Au), [12][13][14] or requires the usage of ionic liquids, [15][16][17] which however impair the process stability. Investigating systems and processes enabling inexpensive, selective and stable CO 2 RR is therefore a crucial requirement for making this process economically viable, as it will be presented in this work. Furthermore, in contrast to the previous laboratory-scale systems, which only demonstrate small-scale sub-assemblies performing CO 2 RR, we outline the important steps towards a functional and scalable process to selectively produce CO at large scale and high yield in neutral pH aqueous solution.
In particular, we propose a CO 2 RR cathode based on commercial and large-scale metallic foam, which was specifically adapted to meet the stringent requirements for selective CO 2 -to-CO conversion. Due to their three-dimensional (3D) structure, metallic foams offer the advantage of high surface area, which is necessary to circumvent the solubility limitations of CO 2 gas in liquid electrolytes and thus enabling efficient CO 2 RR catalysis. In this regard, it is important to guarantee a high number of threephase interfaces, i.e. interfaces where CO 2 gas-liquid electrolyte-solid electrode coincide. The tailoring of the 3D foam consisted firstly in the homogenous coating with optimized silver (Ag) catalysts and secondly, in the optimization of the morphological and structural characteristics of the deposited catalyst. Ag is an attractive alternative to electrodes made of Au for CO 2 RR as it exhibits similar selectivity towards CO under moderate overpotentials and is significantly cheaper than most other noble metals. 18,19 Furthermore, nano-or microstructured Ag catalysts can lower even more the required overpotentials 3 while increasing at the same time the density of highly active sites for CO 2 RR. In this regard, dendritic structures are of particular interest and were targeted in this study. Besides numerous examples of Ag based materials for CO 2 RR to CO, such as nanoplates, 20 nanoporous material and nanoparticles, [21][22][23][24][25] nanowires, 26,27 nanoclusters 28 and porous foams, 29 , dendritic structures offer large active surface areas paired with facile and non-expensive synthetization. [30][31][32] In this regard, we selected the 3 electrodeposition method to deposit the Ag catalyst. On account of its functionality and cost effectiveness, this deposition process is highly viable for large-scale industrial applications, e.g. the coating of large-scale metallic foams, and therefore promising to fulfil the requirements of commercial viability. [33][34][35] We selected copper (Cu) as the scaffold material because it is highly conductive and cost-effective paired with high technical maturity, and because Cu surfaces favor the growth of dendritic structures applying electrodeposition method. This paper has a twofold purpose: to understand how the electrodeposition process can be systematically adapted in order to tailor (i) the coverage and (ii) the catalytic properties towards CO 2 RR of the deposited Ag catalyst. A complete coverage of the Cu foam with the Ag catalyst is vital to avoid Cu to take also part in the catalysis. The purpose of tuning the catalytic properties will be achieved by featuring crystallographic surface orientations that are conducive to selective CO production. Theoretical studies have proven that (110) and (220) Ag planes, respectively, particularly enhance the catalytic activity towards CO production. 36 It was shown that these facets facilitate the activation of CO 2 through a strong binding of the surface-adsorbed *COOH species (1 st reaction step) before releasing CO due to weak binding of the *CO species (2 nd reaction step). These findings, however, are supported only by a few experimental studies. 30,37 Therefore, the present study aims to further generalize the results by employing conventional and practical methods and materials to implement the desired crystallographic orientation of the catalyst surface at an industrial scale. The herein presented results evidence that the high active surface area of the Cu foam paired with the optimized facet orientations of the deposited Ag dendrites give rise to superior CO 2 conversion productivities, significantly enhancing the CO current density per area and mass, respectively, in comparison to other scaffold materials.
Overall the findings of this study provide a platform for the design of a highly active and selective cathode material for CO 2 -to-CO conversion which can be employed in processes at large-scale industrial scale.

Catalyst Deposition
The Ag catalyst deposition was conducted by an electrodeposition process on Cu foam employing a conventional electrochemical three-electrode set-up (VMP3 Bio-Logic multi-channel potentiostat). A sheet of Cu foam with a size of 40×40 mm 2 (2.5 mm thick, Recemat BV Cu-4753.04, average pore diameter: 0.4 mm) was used as the catalyst substrate (working electrode). Before the deposition, the foam was firstly cleaned in acetone, 2 M HCl, and deionized water. For the counter electrode a platinum mesh was used and Ag/AgCl/KCl (3M) (E 0 =0.197 V NHE ) was used as reference electrode. The electrodeposition was conducted in a plating solution containing 0.1 M AgNO 3 , 0.5 M NH 4 OH, and 0.1 M NH 4 NO 3 . 38 The cathode optimization was performed by varying the deposition parameters at ambient conditions, i.e. the applied potential and the deposition time. The last cathode preparation step consisted in rinsing of the coated Cu foam with deionized water and drying at room temperature. The average Ag loading on the Cu foam was measured by weight difference. For comparison, silver and copper foils (Alfa Aesar, 99.998%, 0.1 mm thick) cathodes (with a mass of 104.9 g for 10 cm 2 of 0.1 mm Ag foil) and nickel (Ni) foam (1.4 mm thick, Recemat BV Ni-1116) were also employed. 4

Cathode Characterization
Experimental details on the structural (X-ray diffraction (XRD)), morphological (scanning electron microscopy (SEM), transmission and high resolution transmission electron microscopy (TEM and HRTEM)), and compositional (electron energy loss spectroscopy (EELS)) characterization methods applied herein to analyze the asdeposited Ag catalyst on the Cu foam have been provided in Ref. 39. X-ray photoelectron spectroscopy (XPS) was applied to analyze the electronic structure of the cathode surface by using a PHI instrument model 5773 Multitechnique with A1 Kα radiation (1486.6 eV). The electrochemical performance of the Ag dendrite coated Cu foams towards CO 2 reduction was assessed using an adapted flow cell set-up (Micro Flow Cell, Electrocell A/S), which is schematically depicted in Figure 1 and which was already introduced as scalable (photo)-electrochemical device in our previous works. [39][40][41] Figure 1. Sketch of the experimental set-up used to assess the CO 2 RR performance of the Cu-Ag cathode. The set-up consisted of a filter-press flow cell with three inlets (catholyte, anolyte, and CO 2 ) and two outlets (catholyte + CO 2 and anolyte). Catholyte and anolyte compartments were separated by a Nafion membrane. The Cu-Ag cathode was mounted onto a metallic holder defining a geometric surface area of the cathode of 10 cm 2 . The CO 2 gas and electrolyte pathways within the flow cell are indicated by the red and blue arrows, respectively. A 0.5 M KHCO 3 aqueous electrolyte solution (CO 2 saturated) was used as the electrolyte. The reference electrode was a Ag/AgCl (3.4 M KCl) electrode and a dimensionally stable anode plate (DSA) served as counter electrode.
The flow rates of CO 2 gas and electrolyte (0.5 M KHCO 3 ), respectively, within the flow cell were kept at 20 ml/min. As illustrated in Figure 1, the CO 2 gas was directly introduced through the macroporous 3D Cu-Ag foam cathode (geometric surface area: 10 cm 2 ). When using silver or copper foil, CO 2 was flown into the catholyte solution, as the gas, in this case, could not be injected directly through the cathode, which may affect the CO 2 RR under operation in comparison to direct flow through the cathode. For details on the cyclic voltammetry (CV) and gas chromatography (GC) regarding the assessment of current-voltage behavior and the faradaic efficiency for gaseous CO 2 RR 5 products, respectively, the reader is referred to Ref. 39 and the Supporting Information (SI), respectively.

Catalyst deposition and characterization
We initiated our study with the coating of Cu foam substrate with Ag dendrite catalysts by means of electrodeposition. In the potentiostatic electrodeposition process, we investigated the effect of two parameters on the coating of the Cu foam: the deposition time and the applied potential. Figure 2 compares the morphologies of Ag coating on Cu foams after 15, 30, and 60 seconds at constant potentials of -0.05, -0.2, -0.35, and -0.5 V RHE , respectively in the plating solution. At the lowest deposition potential (-0.05 V RHE ), the Cu foam surface was barely coated with Ag after 15 s and 30 s, respectively. After 60 s a continuous film of Ag started to grow but the dendrite structure was not observed at this applied potential. The electrodeposition at -0.2 V RHE , on the contrary, led to a thin coating film with small dendrite like structures, with average dendrite sizes below 1 µm. However, after 15 and 30 s of deposition relatively large parts of the surface of the Cu substrate were still exposed and presumably accessible to the electrolyte. As a direct consequence, the Cu foam substrate, in this case, could also be active in the CO 2 RR catalysis, which would be to the detriment to CO 2 -to-CO conversion efficiency, because Cu surfaces are known to catalytically promote liquid chemicals as CO 2 reduction products, such as formic acid and methanol. 42 A high CO selectivity therefore requires an excellent coverage of the Cu foam with the Ag catalyst. The SEM image of the 60 seconds electrodeposition process revealed that the coverage of the foam substrate significantly augmented along with an increase of the Ag dendrite structure size. The dendrite branches ranged from 500 nm to 3 µm (or even bigger for a few distinct structures as observable in the respective quadrant in Figure 2) featuring a highly random coating of high surface area catalyst structures, which is conducive to the exposure of more active sites for CO 2 RR and permeation of the catholyte. As apparent from Figure 2, the coating for -0.35 V RHE applied potential, in average, led to bigger dendrite structures (branches and stem) for the respective deposition times. However, the coverage of the Cu foam seemed to be less dense compared to the one at -0.2 V RHE . Both observation were confirmed for the deposition at -0.5 V RHE , as the dendrites increased in dimension but significant portion of the Cu foam surface remained uncovered also after 60 s of deposition. This tendency was related to mass transport limitation of Ag ions at the electrode surface under increasing negative deposition potentials. As a result, existing dendrites continued growing, rather than new nucleation, and thus fewer dendrites appeared. Consequently, in this case, the coverage of the Cu foam was reduced. Besides the analysis of the morphologies, shown in Figure 2, in general it is also important to determine the active surface area of the catalyst coated scaffold. For this purpose, we estimated the electrochemical active surface area (ECSA) for one of the Ag dendrite coated Cu foams in comparison to the ECSA of bare Cu foam. In fact, the ECSA can be estimated from the electrochemical double-layer capacitance C dl at the solid/liquid interface. 25,27 The ECSA calculation process is presented in detail in the SI. From Figure S1a and S1b it can be deduced that the Ag dendrite coated Cu foam cathode exhibited a higher C dl than bare Cu foam, evidencing a significantly higher active surface area, which was estimated as 92.50 cm 2 for the Ag dendrite coated Cu foam and 30.25 cm 2 for the bare Cu foam (tested geometric surface area for ECSA: 2 cm 2 ). The next step in the investigation consisted in the analysis of the structural surface properties of the deposited Ag catalyst on Cu foam. The crystalline structure of the asdeposited Ag dendritic structures was investigated by XRD. Figure 3a exemplarily shows the XRD diffraction pattern for the sample deposited for 60 s at -0.2 V (see Figure 2). No significant differences in the XRD measurements were detected between the other prepared samples in Figure 2. The analysis revealed that the Ag dendrite electrocatalyst mainly crystallized in the cubic phase, while a small percentage of the hexagonal phase was measured as well (< 10%). The diffraction peaks are in perfect agreement with the reference pattern for cubic Ag (JCPDS card number 00-001-1164), namely Ag (111) at 38.1º, Ag (200) at 44.3º, Ag (220) at 64.5º and Ag (311) at 77.4º. The Ag (101) peak (*) corresponds to the hexagonal Ag phase with primitive lattice (JCPDS Card number 00-041-1402). The peaks from the Cu foam substrate corresponding to the Cu (JCPDS card number 01-070-3038) were also discernible in the diffraction pattern of the Cu-Ag system, as expected (Cu (111) at 43.2º, Cu (200) at 50.4º and Cu (220) at 74.1º). Figure 3b plots the XPS spectrum of Ag 3d 3/2 and Ag 3d 5/2 double peaks from the substrate (again sample deposited for 60s at -0.2 V), which were centered at 374.3 and 368.3 eV, respectively. This was in perfect agreement with those of elemental Ag, 43,44 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 7 and demonstrates that there was no corrosion of Ag dendrites during the electrodeposition procedure. The deposited Ag dendrites were further investigated by means of high resolution TEM (also for sample deposited at -0.2 V for 60 s). Figure 4a displays the HRTEM image of one analyzed dendrite edge. The shape of the edge was random without sharp exposure to crystalline planes, indicating the random faceting planes of the dendrite structures. The highly crystalline feature of the Ag catalyst is persuasively depicted in Figure 4b. The fast Fourier transform (FFT) spectrum in Figure 4c (red squared region in Figure  4a) indicates that the edge nanostructures crystallized in the Ag cubic phase. The XRD measurement (Figure 3a), however, also identified a small hexagonal phase contribution. In this regard, Figure S2a in the SI shows the HRTEM image of a different dendrite edge, from which a polytypic behavior is observable. In these dendrites, periphery regions could be detected that presented many defects, such as twins and stacking faults. The FFT of Figure S2b furthermore reveals that there were two sets of diffraction patterns from the twinning domains (cubic), which shared the same planes of {1-11} (indicated by the white arrow in Figure S2a). The presence of a high density of twins, as those reported in Figure S2a, may induce the formation of polytypic structures, with alternating cubic and hexagonal domains in the nanostructure, as already typically observed in semiconductor nanostructures (e.g. Si, 45 GaAs, 46 ZnTe, 47 ). In fact, this could be the origin of the cubic and hexagonal signals observed in the XRD patterns ( Figure 3a).  The crystallographic orientation of the dendritic structures was assessed hereafter. The low magnification view of the Ag-Cu foam samples in Figure 5a shows the dendritic structure comprising the stem and the dendrite branches. As apparent from Figure 5a), the branches followed <111> directions and an angle between the stem and the branches of 109.5º could be measured. This implies that the stem of the Ag dendrites grew along the [001] direction. Based on this finding and based on the FFT result shown in Figure  S2b, a simulated diffraction pattern (inset in Figure 5a) could be deduced. This information allowed performing a simulated scale model of the Ag dendrite branch along the [1][2][3][4][5][6][7][8][9][10][11]    To investigate the dendrite structural composition in more detail, EELS spectrum images (compositional maps) were obtained at the nanoscale. The acquired composition map of an Ag dendrite branch is shown in Figure 6. The analysis evidenced that the dendrite branches were mainly composed of crystalline Ag, which dominated the core part of the catalyst structures. In the thin periphery region, oxygen, copper, and carbon were detected. The latter element was detected as residual from the hexane dispersion solution (see Experimental 2.2), while Cu may have occurred from the residual of Cu foam in the dispersion solution used for the TEM measurements (see Experimental 2.2). The evidenced oxygen is supposed to stem from the oxidation of copper and carbon residuals because the crystalline Ag dominates the dendritic matrix based on the HRTEM characterization. Therefore, we can assume that the residual induced periphery layer (down right in Figure 6) was not present in the as-deposited Ag catalyst structure, and thus would not affect the electrochemical characterization.

Electrochemical reduction of CO 2
The electrochemical performance of the Cu-Ag cathodes regarding CO 2 RR was assessed in the electrochemical flow cell (three-electrode configuration), as shown in Figure 1. A commonly employed 0.5 M KHCO 3 aqueous electrolyte solution saturated with CO 2 (pH = 7.5) was used as catholyte. 21,24,26,29 The fluidic characteristics of the employed cell design allowed to recirculate the catholyte solution (flow rate of 20 ml/min), while the CO 2 gas was directly injected through the macroporous 3D Cu-Ag  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 cathode. Such continuous-flow operation designs allow alleviating mass transport limitations for CO 2 RR catalysis and thus improving partial current densities of reaction products, as shown by Kenis et al. and Newman et al. 50,51 In one of our prior studies, we identified a gas-to-electrolyte flow ratio of 1 as optimum for the flow reactor design, which is why we also employed this ratio in the present study. In particular, it was verified that a ratio of 1 ensures the perfect balance between offering the right amount of catholyte and CO 2 gas at the cathode surface. 40   As expected, the bare Cu foam surface (0 seconds of electrodeposition time) did not promote the generation of CO. The data shows that H 2 was the major detected gaseous product with a maximum faradaic efficiency of 87.8 % at -0.4 V RHE . The selectivity for CO production was below 5 % for the entire investigated potential range and the total faradaic efficiency for CO and H 2 ranged between 61 % and 91 %. This suggested the formation of other products (not measured here), for instance formic acid, as already reported elsewhere. 42  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 11 The thin coating film provided after 15 seconds of electrodeposition immensely affected the catalytic behavior of the tested cathode, as the selectivity towards CO products drastically increased (Figure 7, upper right). The CO faradaic efficiency significantly augmented for the majority of the investigated potentials from around 5 % without Ag dendrites (0 s) to a maximum value of 64.3 % at -0.8 V RHE for the Ag dendrite coated Cu foam cathode (15 s). However, the total faradaic efficiency remained between 81 % and 90 % for the different potentials, suggesting that the uncovered Cu foam areas take part in the CO 2 RR catalysis to produce other products (as expected and shown in Ref. 42). A more dense coverage of the Cu foam including increased dendrite structures, which was obvious after 30 seconds of electrodeposition (see Figure 2 for -0.2 V RHE ), further drastically increased the selectivity towards CO formation, as can be seen from the lower left plot in Figure 7, with a maximum faradaic efficiency of 85.5 % (at -0.8 V RHE ). This entails that the total faradaic efficiency for CO and H 2 surpassed 90 % for the majority of the investigated potentials and reached values up to 97 %. The best selectivity for CO and H 2 was however exhibited by the cathode prepared for 60 s (at -0.2 V RHE , Figure 2). As apparent from the lower right graph in Figure 7, CO production was amplified for all investigated potentials, especially in the potential range between -0.4 and -1.1 V RHE . In this range, total CO/H 2 faradaic efficiencies close to 100 % were measured along with a maximum CO faradaic efficiency of 95.7 % at -0.8 V RHE applied bias, implying an excellent coverage of the Cu foam with the Ag dendritic catalyst. For lower potentials (< -0.4 V RHE ) we expect the formation of low amounts of liquid products (not measured here), as already reported elsewhere for Ag catalysts. 29 Deposition times beyond 60 s were also considered in this study. In the SI, the morphology and the faradaic efficiency of the Cu foam cathode, respectively, after 2 and 3 min of electrodeposition are shown ( Figure S3). As apparent the coating of the 3D foam was highly dense in both cases. Additionally the overall selectivity towards gas production (CO + H 2 ) slightly increased, and reached nearly 100 % faradaic efficiency in the investigated potential range. However, the maximum faradaic efficiency for the targeted CO (in both cases at -0.8 V RHE ) could not be augmented (94.7 % and 92.3 %, respectively). Figure 8 depicts the maximum CO faradaic efficiency (FE CO ) as a function of the electrodeposition time and the applied potential in the potentiostatic process for all investigated samples (shown in Figure 2). This graph enables to examine the impact of the complete set of parameters on the catalytic performance studied in this work. The values for maximum faradaic efficiency in Figure 8 were acquired at -0.8 V RHE , as the highest CO selectivity was detected at this cathodic potential for the samples deposited at -0.2 V (see Figure 7). From Figure 8 it becomes obvious that, except for the -0.05 V samples (grey triangles), deposition times beyond 60 s could not lead to higher maximum CO selectivities. The values for FE CO did not increase (even decreased slightly) for longer deposition times for the cathodes deposited at -0.2 (black squares), -0.35 (purple circles), and -0.5 V RHE (blue diamonds), respectively. This observation suggests that after 60 s no new nucleation could start (because of complete coverage or ion diffusion limitation) and that prolonged deposition times solely increased the dendrite dimensions. Hence the CO selectivity was slightly impaired, as it is well accepted that nano/microsized catalysts propel catalytic activities more than macroscopic structures. 3 We therefore concluded that plating processes longer than 60 s are not beneficial for this process with respect to high CO selectivity. Moreover, this result was in line with the aim to minimize process time and loaded catalyst material in view of large scale commercial application. The highest faradaic efficiency measured in this study was 95.7   Figure 9a represents the cyclic voltammogram of the champion Cu foam based cathode (deposited at -0.2 V for 60 s) from Figure 7 and Figure 8 (black curve). For comparison, the same measurement was also conducted for a flat Cu foil with Ag catalyst (green curve) and a flat Ag foil cathode (orange curve). The Cu foil sample was coated with Ag employing the same electrodeposition process as for the Cu foam cathode. As apparent from the SEM image shown in the SI, in Figure S4a and S4b, the same deposition process did not lead to the growth of dendritic structures on Cu foil, but to Ag agglomerates. Fluctuations in the CV curves may stem from the characteristics of the used flow-reactor set-up.   Figure S6 in the SI.

Comparison of Cu Foam with Other Substrates
For the Ag foil, it can be seen from Figure 9a that no evident voltammetric response was detected before the potential -0.6 V RHE , indicating the negligible CO 2 RR activity. Inversely, the CO 2 RR current of the Ag dendrite coated Cu foam appeared at the earliest potential of -0.3 V RHE owing to its large catalytically active 3D surface and the direct injection of gaseous CO 2 at the cathode surface, which favors the CO 2 RR kinetics. The onset potential measured for the Cu foil coated with the Ag agglomerates was around -0.4 V RHE . From Figure 9a it can therefore be concluded that the performance of the Ag coated Cu foam cathode exceeded the performance of the Ag foil and Cu-Ag foil regarding CO 2 RR current density in the entire investigated potential range from 0 to -1.0 V RHE . Only for potentials above -1.0 V RHE the Ag foil and Cu-Ag foam cathode provided similar current densities. The superior electrochemical performance of the Ag coated Cu foam cathode regarding CO 2 RR was further evidenced by comparing the performance of the sample measured under pure argon (Ar) flow. As apparent from Figure S5, the cathode performance was significantly enhanced under CO 2 flow compared to Ar flow, regarding onset potential for cathodic current and high current densities.
Following the CV measurement, we subsequently assessed the CO 2 RR selectivity of the Cu-Ag foam cathode and of the two other electrodes towards reaction products using gas chromatography. Figure 9b displays the measured faradaic efficiencies for CO which were already presented in Figure 7 for the 60s sample. CO and H 2 partial current densities of the CO 2 RR are plotted in Figure S7. As already apparent from Figure 7 (for 60 s), a very high selectivity for the generation of the two gaseous products CO and H 2 was exhibited from the Cu-Ag foam cathode (total faradaic efficiency close to 100 % for potentials from 0 to -1.1 V RHE , see Figure 9b and Figure S6). At the highest faradaic efficiency of CO of 95.7 % (at -0.8 V RHE ) this cathode exhibited an eminent CO current density of 9.5 mA/cm 2 ( Figure S7). A maximum CO partial current density of 27.3 14 mA/cm 2 was measured at -1.0 V RHE . Further increase in applied potential did not result in higher CO yield because mostly H 2 formation was promoted (see black squares in Figure S6). Similar observations in other reports were assigned to mass transport limitations of CO 2 at high current densities and not to the intrinsic activity of the Ag catalysts. 19,52 Nevertheless, the herein demonstrated CO production yield (at low potentials < -1.0 V RHE ) surpasses the performance of Ag dendrite and other Ag based systems for CO 2 RR at ambient conditions. 20,29,30,51,53,54 Therefore, our findings clearly demonstrate the superior electrochemical characteristics of the presented metallic foam supported cathode. Figure 9b furthermore provides evidence that the Ag dendrite based foam cathode outperformed the Ag and Cu foil scaffold cathode with respect to selectivity for CO production. The CO faradaic efficiencies measured for the Ag foil (orange triangles) and Cu foil (green diamonds), respectively, were below the values of the Cu-Ag cathode (black squares) in the entire investigated potential range with an inferior maximum faradaic efficiency of 55.7 % and 72.5 %, respectively (both at -0.8 V RHE ). Notably from Figure 9b, the CO selectivity of the Cu foam based cathode also started to augment at lower potentials than for the Ag and Cu foil based cathodes, respectively. This behavior exactly mirrors the respective CV curves in Figure 9a, i.e. the shift in the onset potentials for CO 2 RR of both tested cathodes. In fact, the enhanced CO 2 RR catalytic activity can be ascribed to the combination of nano/microsized dendrites with the 3D macroporous structure of the supporting Cu foam itself. This marriage resulted in a significantly increased contact area between catalysts and electrolytes and provided a greater amount of active sites where CO 2 RR could take place (see ECSA estimation in the SI, Figure S1). Furthermore, the flow cell design of the applied reactor ensured an improved transport of reactants (CO 2 and electrolyte) and products (CO and H 2 ). Moreover, the presented Ag dendrite catalyst offers distinct advantages regarding large scale applications due to its facile deposition and high catalytic performance in delivering significantly higher (4 orders of magnitude) mass specific CO current density than flat Ag foil (see corresponding plot in Figure S8).
Besides CV and GC data, another crucial parameter of electrocatalytic activity is the corresponding Tafel slope. In Figure 10 the logarithm of the CO partial current density is plotted against the potential of the Ag dendrite coated Cu foam and Ag foil cathode. The Tafel slope is acquired from the range of linearity at the low-voltage scope. As listed in Figure 10, in this region, the Tafel slope of Cu-Ag cathode was 89.7 mV/dec, and thus, as expected, much lower than Ag foil (198.7 mV/dec) and also lower compared to other Ag structures for CO 2 RR. 20,26 This result verifies that the Ag dendrite catalyst promotes a fast CO 2 RR rate. The high-current data in Figure 10 reflects the data shown in Figure S7 and evidences that the Cu-Ag foam cathode, in this region, exhibited much higher CO production current densities than the flat Ag foil. The observable decay in the CO production was caused by an increase of the hydrogen production and occurred at lower potentials in the case of the Ag foil, preventing a better catalytic performance.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58 59 60 Figure 10. Applied potential versus CO production partial current density for Ag dendrites (black squares) and Ag foil (orange triangles).
In particular, the Tafel evaluation gives insights in the CO 2 RR reaction kinetics. It is well accepted that the first step of the reaction consists in the transfer of one electron to a CO 2 molecule to form a CO 2 radical absorbed on the catalyst surface. This step is rate determining for the entire reaction process, because it proceeds (as suggested by literature) at much more negative potentials than subsequent steps, in which the CO 2 intermediate species reacts with two protons and another electron to form a CO and a H 2 O molecule. The much lower Tafel slope for the Ag dendrite based cathode, as depicted in Figure 10, therefore suggests that the Ag dendrite surfaces are able to stabilize the CO 2 intermediate more efficiently than a flat surface, and thus propel the formation of CO molecules subsequently. This observation is of highest importance for CO production and was already studied theoretically and experimentally, respectively. 36,37 Durability is a vital appraisal criterion to estimate a practical electrolysis application. Figure 11 displays the examined long-term stability of the Ag dendrite coated Cu foam cathode integrated in our flow cell reactor. For comparison, the potentiostatic measurement of the Ag foil was also included. At an applied potential of -0.8 V RHE (potential of maximum FE CO ), the Cu-Ag cathode exhibited a current density of around 10 mA/cm 2 which stabilized around this value over the whole reaction session, without the observation of a significant decay (95% retained after 15 hours of continuous operation). Similarly, the CO faradaic efficiency was also stabilized around 94 % for the whole duration of 15 hours of testing. Measured current density and faradaic efficiency are in accordance to the data presented in Figure S7 and Figure 9b, respectively. Thus, the stability of our prototype reactor was in good agreement with previously reported stabilities for Ag catalyst based electrodes. 21,29 The operation robustness of the Cu-Ag cathode was additionally confirmed by SEM, TEM, and XRD analysis, respectively (see Figure S9, in the SI). As presented in Figure S9a and S9b, the Ag dendrites preserved their morphology and structural properties, respectively, after long-term electrolysis at -0.8 V RHE . The XRD diffraction peaks were identical before and after the 15 hours electrolysis operation ( Figure S9c). For the sake of completeness, we also assessed the stability of the Ag foil ( Figure 11), which, as expected, also exhibited a very stable operation (tested for 5 hours) with a lower CO 2 RR current density of around 3.4 mA/cm 2 and a CO selectivity of around 41 %, which was in line with the data shown in Figure 9a and 9b, respectively.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58 59 60 Figure 11. Stability of total CO 2 reduction current density (left ordinate) for Ag dendrites (black curve and squares) and Ag foil (orange curve and triangles) at -0.8 V RHE in 0.5 M KHCO 3 . The CO faradaic efficiency of both tested cathodes as a function of operation time is plotted on the right ordinate.
Besides the comparison with flat metallic foil surfaces, the Cu foam was also checked against a Ni 3D metallic foam, as high surface area catalyst scaffold. Two-dimensional scaffold materials, such as carbon cloth or titanium mesh were not investigated in the present study, mainly because of the lower surface area and lower electrical conductivity compared to Ni foam. 55,56 In the SI, the comparison with Ni foam, regarding Ag catalyst deposition and selectivity towards CO production is presented. In contrast to the Cu foam surface, the Ni foam surface did not give rise to dendritic catalyst growth, but the Ag catalyst was deposited in the form of particles of 1 µm in average (see SEM images in the SI, in Figure S10a and Figure S10b). The results on partial CO and H 2 current densities, respectively, presented in Figure S11, furthermore evidence that the Cu-Ag cathode showed a significantly higher selectivity towards CO production than the Ni-Ag cathode, which confirms the outstanding performance of the developed Cu-foam based cathode.

Discussion
The reason for the discrepancy in the activity towards CO production between the investigated cathodes, observable in Figure 9b and Figure S11, respectively, can be explained by several factors. (i) First, the morphology of the deposited Ag catalyst is a crucial parameter regarding CO 2 conversion activity. Based on the SEM images it can be deduced that the dendritic structures showing the highest activity towards CO production (i.e. deposited for 60 s at -0.2 V RHE , Figure 2) possessed more active surface area compared to the other Ag catalyst morphologies obtained on Cu foam (e.g. deposition at -0.05 V RHE , Figure 2), Cu foil ( Figure S4b) and Ni foam ( Figure S10b). (ii) Secondly, the better coverage of the Cu foam with the dendritic catalysts (see deposition at -0.2 V RHE in Figure 2 and Figure 6) ensured that only the Ag catalyst was active in the CO 2 RR. As explained earlier, surfaces such as Cu and Ni favor other products than CO. Cu favors alcohol production and Ni surfaces are known to favor the hydrogen evolution reaction. The latter is in line with the data on partial current density shown in Figure S11, which reveals that the Ni-Ag sample showed a higher activity towards H 2 production than the Cu-Ag sample. (iii) Another important parameter influencing the 17 CO 2 RR activity towards CO is the catalyst surface morphology. As revealed by experiment and simulation ( Figure 5), the as-deposited Ag dendrites had a high number of planes. It is well accepted that (110) planes own a high amount of stepped sites, which are highly active sites for CO 2 -to-CO conversion. 36,37,49 To directly measure the ratio of (110) planes vs. more flat sites is however very challenging for 3D materials in general. To quantify the relative amount of (110) surface, a more convenient approach would be to calculate the ratio of (110) and (111) intensities with the help of XRD measurements. Please note that the diffraction peak of Ag (110) was not observable in the XRD pattern of the Cu-Ag foam (Figure 3a) because in face centered cubic structures this reflection is not allowed. Nevertheless, it can be assumed that the (110) structure exists in an amount proportional to that of (220). Consequently, we took the ratio of the Ag signals (220)/(111) as approximation for the quantity of stepped vs. planar sites, as already described elsewhere. 30 In Figure 12, the values for the intensity ratios for four investigated samples are plotted as a function of the respective maximum CO faradaic efficiency. The comparison was conducted for four samples owing different morphologies of the deposited Ag catalysts: Ag dendrites deposited on Cu foam (for 60 s at -0.2 V RHE ), which will be called "optimized" hereafter, because the highest CO productivity was achieved with this sample (morphology: Figure 2 and faradaic efficiency: Figure 8); Ag catalyst deposited on Cu foam (for 60 s at -0.05 V RHE ), which will be called "non-optimized", because the corresponding CO productivity of this sample was significantly below the optimized one (morphology: Figure 2 and faradaic efficiency: Figure 8); Ag catalyst deposited on Cu foil (morphology: Figure S4b and faradaic efficiency: Figure 9b) and Ag catalyst deposited on Ni foam (morphology: Figure S10b, CO productivity: Figure S11). Ag on Cu foil and on Ni foam, respectively, was deposited with the same deposition process as for the optimized sample. From the data in Figure 12, a clear trend of increasing FE CO with increasing (220)/(111) ratio can be deduced. The XRD measurements from which the (220)/(111) ratios are deduced, can be found in the SI in Figure S12.
As apparent, the optimized Ag dendrite catalyst on Cu foam, exhibiting the highest FE CO of 95.7%, also showed the highest (220)/(111) intensity ratio of 27.5%. The lowest ratio of 22.1% was found for the Ag catalyst deposited on Ni foam, for which also the lowest FE CO among the investigated samples of 47.5% was measured. Hence, the experimental findings of this study support the theoretical calculations done by Back et al. showing that Ag surfaces owing a high amount of (220) planes are prone to high CO productivity. 36 Overall, the present study successfully generalizes the theoretical findings on Ag catalyst behavior towards CO 2 RR by employing practical concepts to implement the desired crystallographic orientations of the catalyst surface at large scale.

Conclusions
In summary, we successfully tailored large-scale metallic foam for highly selective CO 2 conversion into CO at high current and low overpotentials. The experimental observations manifested that the combination of Ag dendritic catalysts with highly conductive 3D Cu foam effectively ameliorated the electrocatalytic performance towards CO production and gave rise to a stable maximum faradaic efficiency of 95.7 % at -0.8 V RHE . A maximum operation current density of 27.3 mA/cm 2 was measured at -1.0 V RHE . The performance boost, in comparison with other catalyst substrates, was attributed to mainly two features of the Cu foam based cathode: (i) reduced mass transport limitations due to the amplified 3D electrochemically active contact area between the homogeneously coated Cu foam and the electrolyte and (ii) the demonstrated high selectivity of the deposited Ag catalyst towards CO production, which could be attributed to the unique dendrite surface orientations favoring a high density of active sites where CO 2 RR to CO could take place (high (220)/(111) XRD intensity ratio). Overall, this work exploits a novel pathway for the fabrication and application of robust, efficient, and highly selective metallic foam based electrodes in prototype reactor devices, which are viable for scalable electrochemical reduction of CO 2 to CO at large scale.