Electrochemical conversion of pressurized CO2 at simple silver-based cathodes in undivided cells: study of the effect of pressure and other operative parameters

Electrochemical reduction of pressurized CO2 is proposed as an interesting approach to overcome the main hurdle of the CO2 electrochemical conversion in aqueous solution, its low solubility (ca. 0.033 M), and to achieve good faradaic efficiency in CO using simple sheet silver cathodes and undivided cells, thus lowering the overall costs of the process. The effect on the process of CO2 pressure (1–30 bar), current density, nature of the supporting electrolyte and other operative conditions, such as the surface of the cathode or the mixing rate, was studied to enhance the production of CO. It was shown that pressurized conditions allow to improve drastically the current efficiency of CO (CECO). Furthermore, at relatively high pressure (20 bars), the utilization of simple sheet silver cathodes and silver electrodes with high surfaces gave similar CECO. The stability of the system was monitored for 10 h; it was shown that at a relatively high pressure (15 bar) in aqueous electrolyte of KOH using a simple plate silver cathode a constant current efficiency of CO close to 70% was obtained.


Introduction
Nowadays, in a world struggling to curb the global warming, electrochemical conversion of carbon dioxide could be considered one of the main strategies for both synthesis of chemicals and decrease of CO 2 emissions [1][2][3]. Recently, Bushuyev et al. [4] have discussed the current state of emerging technologies for the catalytic conversion of carbon dioxide into various chemical products, keeping in consideration the economics of the process. They visualized at least six potentially disruptive CO 2 catalytic conversion technologies (i.e. electrocatalysis, photocatalysis, biohybrid, nanoporous confinement, etc.), envisioning a timeline for their implementation on a large scale and reaching the conclusion that the closest technologies to commercialisation with start-up and entrenched companies (i.e. Opus-12, Dioxide Material, and Carbon Recycling International) are based on CO 2 electrochemical conversion. More in general, the Carbon Capture and Utilization (CCU) technologies become more attractive as the renewable energy price continues to decline (getting to 2 ¢/kWh in some jurisdictions) and the process could utilize the excess electric energy from intermittent renewable sources to transform CO 2 into carbon-based chemicals, storing electric energy in the form of chemical energy [2,4,5]. Electrochemical conversion of CO 2 is challenging because of (i) low stability, (ii) high overpotential, (iii) low solubility of CO 2 in aqueous electrolytes at ambient conditions (ca. 34 mM), (iv) in some cases high catalyst cost, and (v) low productivity often associated to high products selectivity [6,7]. To date, among the wide range of chemical products produced by the CO 2 electrochemical reduction in water solvent [8][9][10][11][12][13][14], the highest current efficiencies (CEs) and current densities (j) are usually reported for CO and formate/formic acid (2e − reduction products) [3,6]. Furthermore, some studies have shown that CO, formic acid and other carboxylic acids can be produced in aprotic solvents by CO 2 reduction [15,16]. In aqueous electrolytes, several catalysts are known to selectively lead to the production of CO (i.e. Ag, Zn, Au, Cu) [17]; in particular, Ag and Au based electrodes allow to achieve quite high CE in CO (CE CO ) in short electrolyses and they are characterized by relatively low overpotential values for the hydrogen evolution reaction (HER) allowing to obtain both CO and H 2 . Among the noble metals, the Ag electrode is considered one of the more promising materials for the scale-up of the CO 2 reduction process due to its lower market cost with respect to Au and good faradic efficiency [18].
While, at the beginning, the formation of CO from CO 2 reduction was investigated using conventional plate electrodes [17], in last years, many researchers have focused their attention on the engineering of better electrodes, catalyst, electrolyte composition and configuration of electrolyser [6,19,20] to improve the performances of the process. In particular, attention was devoted on overcoming one of the main issues of the electrochemical conversion of CO 2 in water solution: its low concentration, resulting in low CO 2 conversion rate (limited by mass transport limitations) and in low productivity or/and low current efficiencies.
Several works have shown that the implementation of electrodes with high active surface allows operating at higher current intensities than that achieved at simple plate electrodes improving the performances of the process [21][22][23][24][25][26][27]. For example, Lu [23] and Ma [22] and co-workers have synthesized, respectively, a nanoporous (np-Ag) and an oxide-derived nanostructured (OD-Ag) silver catalyst (i.e. both structures characterized by a larger area for catalytic reaction than that of polycrystalline silver) which allow, at low overpotentials, to achieve a j at least 20 times higher than that reached at planar Ag and increase drastically the CE CO from less than 4% using a plate silver to 80 or 92% using OD-Ag or np-Ag electrode, respectively, in a CO 2 saturated aqueous KHCO 3 electrolyte. Electrodes with high active surface for low overpotentials are characterized by a lower value of the Tafel slope (i.e. 58 and 77 mV dec −1 for np-Ag and OD-Ag, respectively) than the polycrystalline silver electrode (~ 130 mV dec −1 ), that indicates a fast initial electron transfer to a CO 2 molecule speeding up the reduction rate; however, at relatively high overpotentials, the Tafel slope increases, thus showing that, also for this kind of electrodes, the CO 2 electrochemical conversion could be limited by the mass transport of CO 2 to the surface of the electrode [6,22,23,28].
Another strategy to reduce mass transfer limitations due to the low concentration of CO 2 dissolved in aqueous electrolyte at atmospheric pressure is the utilization of gas diffusion electrodes (GDEs), that make faster the reactant transport and distribution [3,13,[29][30][31]. Ma et al. [24] have shown that it is possible to achieve a nominal current density up to 350 mA cm −2 with high faradic efficiency (CE CO ~ 95%) and an energy efficiency of CO (EE CO ) of about 45% working with an Ag-GDE at 3 V in alkaline media (1 M KOH) for a short time. In addition, under adopted conditions, the utilization of a Ag-GDE can guarantee a quite high conversion ratio up to 32% per pass maintaining a CE CO close to 80% [32]. To date, although the implementation of GDEs gave very promising results in term of CE CO , j and productivity [3,11,33,34], there are some challenges that should be solved. Most of the GDEs are suffering from (i) loss of the catalyst possibly due to the low adhesion of the common-used Nafion solution as a binder of the catalyst to the support; (ii) low stability with the time; (iii) high overall costs; and, (iv) salt deposition on the gas diffusion layer determining blocking of CO 2 access in the GDE [35,36]. Furthermore, a crucial factor, using a GDE-type cell, is the pressure balance between the gas and liquid compartments to avoid that the gas can go through the gas diffusion layer, not allowing the essential three-phases contact (among the solid catalyst, liquid electrolyte and gas reagent), or flooding of the gas compartment due to the electrolyte.
First data about high pressure CO 2 electrochemical conversion at silver cathodes in water electrolyte were reported by Hara et al. [37] in 1995, showing that it is possible to obtain a quite high CE CO (~ 75%) at 30 bar using an Ag plate electrode in a divided cell equipped with a Nafion 117. In spite of these very promising results, few studies were devoted later on the potential utilization of pressurized CO 2 for its electrochemical conversion to CO mainly using a GDE with a promising result for short electrolysis. In particular, Dufek and co-workers [41,45] have shown an improvement in CE CO (~ 80%) and in reduction rate (~ 220 mA cm −2 ) with a corresponding suppression of hydrogen evolution reaction by working in K 2 SO 4 aqueous electrolyte at an elevate pressure of about 20 bar for 1 h. More recently, it has been demonstrated by Gabardo et al. [46] that an increase in the performances of the process can be achieved due to the synergetic effect of very highly alkaline reaction environment (1-7 M KOH aqueous electrolyte), which decreases the overpotentials (thus enhancing the energetic efficiency), and pressurized CO 2 system (up to 7 bars), which inhibits the production of other CO 2 reduction products (thus increasing the selectivity). Conversely, other studies have shown that the utilization of the pressure change remarkably the CO 2 product distribution; for example, Hara [38] found that an enhancement of the CO 2 pressure and/or mixing rate at Cu-based electrodes can selectively drive the electrochemical reduction to CO, HCOOH, H 2 or hydrocarbons. However, due to the problems and costs of GDEs based electrodes, it would be interesting to focus on the utilization of simpler cathodes under pressurized conditions. In particular, we would like to focus on the possibility to use pressurized carbon dioxide to convert it effectively to CO using relatively simple and cheap apparatuses such as undivided cells (in order to avoid the economic penalties given by the separator) and simple plate silver cathodes (in order to avoid the costs connected with more complex electrodes) by studying the effect of different operative parameters. In this frame, it is important to remind that (i) the utilization of not too high pressures involves a quite low operative and capital costs [59] (ii) pressurized condition could aid the industrial integration of the CO 2 reduction process with up-and downstream processes and, (iii) in spite the wide literature on the CO 2 reduction to CO, up to our knowledge, no other studies are focused on the investigation of several operative parameters under pressurized conditions at flat Ag on the performances of the process. Hence, in this work the electrochemical conversion of CO 2 to CO was studied at simple silver electrodes with the aim to better evaluate the effect of the pressure (1-30 bar) and of various operative conditions (current density, nature of the supporting electrolyte, surface of the electrode, etc.) under pressurized conditions on the performances of the process. As previously mentioned, the utilization of electrodes with high active surface for CO 2 reduction could be an interesting alternative to improve the performances of the process; hence, the synergetic utilization of a cathode with high surface and pressurized CO 2 was also studied, comparing the results with that achieved with a silver plate cathode. Eventually, it was studied the stability of the CO production at relatively high pressure (15 bars) using different electrolytes for 10 h.

Reagents, materials and analyses
An electrolytic solution composed of bi-distilled water as solvent and 0.2 M K 2 SO 4 (purity > 99.0%, Janssen Chimica), 0.5 M KOH (purity > 86.7%, VWR Chemicals), 0.5 M KHCO 3 (purity > 99.7%, Aldrich) or 0.5 M KCl (purity > 99.5%, Fluka) as supporting electrolyte has been employed. The solution pH was measured with a Checker® pH Tester (HI98103) supplied by HANNA® instruments. To feed the system, CO 2 (99.999% purity) supplied from Rivoira was employed. Silver plate (plate-Ag) and a high surface silver electrode (hs-Ag) based on commercial nanoparticles supported on carbon fiber paper (PTFE treated to 5%) were used as working electrode. The hs-Ag electrodes were prepared via hand-painting techniques according the procedure described in [60]; the catalyst ink was prepared by mixing Ag catalyst (unsupported Ag nanoparticles, < 100 nm particle size, 99.5% trace metals basis, Sigma-Aldrich), deionized H 2 O (200 µL), Nafion ® perfluorinated resin solution (5 wt% in lower aliphatic alcohols and water, ρ = 0.874 g/ mL, Sigma Aldrich; 10:1 catalyst-to-Nafion ratio) and isopropyl alcohol (200 µL). Subsequently, the ink was sonicated for 20 min by a LabSonic FalcSonicator, hand-painted using a paintbrush to cover with catalyst a total geometrical area of 1.5 cm 2 and dried under an atmosphere of N 2 (99.999% purity; supplied by Air Liquide). The actual catalyst loading was of 0.5 mg cm −2 . Agilent HP 1100 HPLC fitted with Rezex ROA-Organic Acid H + (8%) column (Phenomenex) at 55 °C and coupled with a UV detector working at 210 nm was used to evaluate the formic acid concentration. The used mobile phase is a 0.005 N H 2 SO 4 water solution at pH 2.5 eluted at 0.6 mL min −1 . To calibrate the HPLC instrument, a pure standard of HCOOH (99-100%, supplied by Sigma-Aldrich) was used. Gas chromatography analyses were made to characterize the gas products compositions (H 2 , CO, CO 2 , O 2 , N 2 and CH 4 ). A gas-tight syringe (supplied by Hamilton GASTIGHT®) was used to take a gas sample of 250 µL from a pierceable septum located in the effluent gas stream. Agilent 7890B GC equipped with a Carboxen-1000 (60/80) column and a Thermal Conductivity Detector (at 230 °C) was used. Helium (Air Liquide 99.999% purity) was used as carrier gas with constant pressure set at 1 bar; the column temperature program was an isotherm at 35 °C for 5 min followed by a 20 °C min −1 ramp up to 225 °C and by an isothermal step for 45 min.

Electrochemical characterization
Linear sweep voltammetry (LSV) characterizations were carried out using: (i) a conventional three-electrode cell with a SCE reference (Radiometer analytical) and a Pt wire counter electrode (system I); and (ii) cylindrical AISI 316 stainless steel cell able to work at high CO 2 pressure (system II), described in detail in previous work [44]. The latter was utilized to perform pseudo-polarization curves at relatively high pressure of CO 2 (range 1-30 bar) as a function of the overall cell potential. The geometric area of the working electrode was 0.1 cm 2 . Stirring of the solution was made with a magnetic stirrer. Characterizations were performed at 0 and 500 rpm. Prior to all characterization, the solution was purged for 25 min by either N 2 (99.999% purity; supplied by Air Liquide) or CO 2 (99.999% purity; supplied by Rivoira). LSVs were performed under N 2 and CO 2 atmosphere with a scan rate of 0.005 V s −1 using an AutoLab PG-STAT12. The jCO 2 -N 2 was calculated under the assumption that the current of the H 2 evolution and CO 2 reduction can add up and H 2 O reduction is the only competitive reaction to the CO 2 conversion at the cathode surface.

Electrolysis
Electrolysis were carried out using two different systems. System I was a conventional batch undivided glass cell equipped with a gas inlet, plate-Ag (3 cm 2 ) or hs-Ag (1.5 cm 2 ) as cathode, a compact graphite (from Carbon Lorraine) or a Ti/IrO 2 -Ta 2 O 5 sheet anode (from ElectroCell AB) as anode and a SCE reference electrode (Radiometer analytical). System II was an undivided cylindrical AISI 316 stainless steel cell able to work at high pressure of CO 2 , described in detail in previously work [44]. Briefly, it was equipped with a gas inlet, a pressure gauge, a dip tube connected to a pressure relief valve and two electrical line connections for the electrodes. The system worked with a continuous supply of carbon dioxide without accumulation of gases. CO 2 was used to pressurize the system at a flow of about 100 mL min −1 and the operative pressure was controlled using a pressure reducer. Stirring of the solution (V = 0.05 L) was performed with a magnetic stirrer. The electrode gap was about 1 cm for both systems. Before each electrolyses, plate-Ag electrode was firstly polished with an alumina powder suspension (size 1 µm), sonicated in bi-distillate water for 10 min, and then rinsed with bi-distillate water. Compact graphite was mechanically scrubbed with sandpaper (P-800) and sonicated in bi-distillate water until the water resulted in transparently. The DSA® anode was polished by ultrasound bath in bi-distillate water. The current density was calculated as the ratio between the current intensity and the wet surface area of the cathode exposed to the anode. Electrolyses were performed under galvanostatic condition (Amel 2053 potentiostat/galvanostat) at room temperature. Experiments were repeated at least twice with a reproducibility within 6% of the results. The instantaneous current efficiency (CE), energetic efficiency (EE) and the energetic consumption (Wh/ mol CO ) were defined, as follows: where F is the faradaic constant (96,485 C mol −1 ), n i mole of i product and I the current intensity, t the time, E°c ell is the standard cell potential: − 1.33V for the electroreduction Energetic consumption = I * ΔV∕n CO for CO 2 to CO coupled to O 2 evolution, ΔV the applied cell potential. The CO production rate, r CO , (mol h −1 m −2 ) was calculated as follow: where the CE CO is the current efficiency of CO, j the current density (A m −2 ).
The limiting current density j lim = z F (D/δ) c b CO 2 (where z is the number of electrons involved in the CO 2 reduction, F the faraday constant (96,485 C mol −1 ), D the CO 2 diffusion coefficient, δ the thickness of the stagnant layer and c b CO 2 the bulk CO 2 concentration) was computed considering the solubility of CO 2 in water, c b CO 2 , at 25 °C [61], D = 1.94 10 -9 m 2 s −1 [42] while δ was previously estimated in our laboratory [62] for each adopted mixing rate through a well-known diffusion limiting current technique using a very stable redox couple Fe(II)/Fe(III).

Electrochemical conversion of carbon dioxide to CO at various pressures
A first series of experiments was performed at a silver plate cathode using K 2 SO 4 as supporting electrolyte, since this salt according to the literature [36,63,64] can give quite good results in terms of CO production and it grants a quite high conductivity. In particular, a series of polarizations and pseudo-polarizations was achieved both at 1 bar and under pressure to achieve information on the process. Afterwards, a series of electrolyses was performed at different CO 2 pressures, current densities and mixing rates. Figure 1a reports some polarizations recorded under nitrogen or carbon dioxide atmosphere (1 bar) at a silver plate electrode at 0 and 500 rpm. Under nitrogen atmosphere, current density starts to increase significantly at a working potential close to − 1.65 V vs. SCE as a result of the hydrogen evolution reaction (HER). When N 2 is replaced with CO 2 at 0 rpm, the current density starts to increase significantly at a working potential close to − 1.45 V vs. SCE, thus showing that there is a range of potential of about 0.2 V where CO 2 reduction can potentially take place with a very limited hydrogen evolution. For working potentials more negative than − 1.7 V vs SCE (when hydrogen evolution takes place in a relevant way), the addition of CO 2 gives rise to a decrease of the overall current density. The difference between the overall current and the current recorded in the absence of CO 2 was called jCO 2 -N 2 and it is reported in Fig. 1b as a function of the working potential. It is seen that jCO 2 -N 2 presents positive values for potentials up to − 1.77 V (with a maximum at − 1.67 V) and negative values for more negative potentials. Hence, it can be inferred that the addition of CO 2 reduces the rate of hydrogen evolution at least for very negative potentials. According to Hori [17] and Ma [22], the reduction of carbon dioxide at silver cathode proceeds trough the formation of adsorbed CO 2 − (Eqs. (1)-(4) where * indicates a free active site).
Conversely, Rosen and co-authors [65] considered that CO 2 and CO 2 − present no stable interactions with Ag, thus making unlikely that these species can be adsorbed at Ag cathodes. Hence, they proposed the direct formation of COOH ad in the presence of HCO 3 − : Hence, the decrease of the current in the presence of CO 2 may be related to the fact that the reduction derivatives of carbon dioxide are adsorbed on the silver surface (CO ad , CO 2-ad and/or COOH ad ), thus reducing the active sites for hydrogen evolution reaction.
When polarization was repeated at 500 rpm, an increase of current density was obtained for potentials more negative than − 1.6 V (Fig. 1a); this is probably due to the fact that a faster mass transfer of CO 2 and protons favors the CO production. To evaluate the effect of the aqueous concentration of CO 2 on its reduction, current densities were recorded as a function of cell potential ΔV in the presence of N 2 or with different pressures of CO 2 in the range 10-30 bar. As shown in Fig. 1c, current density increases upon enhancing (5) CO 2 + HCO − 3 + e − + * → COOH ad + CO 2− 3 . . V = 0.05 L the CO 2 pressure, thus showing that the rate of carbon dioxide reduction strongly depends on the concentration of CO 2 in solution. Hence, the process is likely to be kinetically limited by the mass transfer of CO 2 from the bulk to the electrode surface and/or its reduction (Eqs. (1) or (5)). However, the limiting current density for direct CO 2 electroreduction can be estimated to be more than 400 mA cm −2 at 20 bars under operative conditions adopted in our electrochemical cell. Hence, it can be ruled out the role of mass transfer as r.d.s. at relatively high pressures of carbon dioxide. As shown in Fig. 1d, the plot jCO 2 -N 2 vs. Δ V can be divided in two regions: (I) a first area for Δ V less negative than 2.5 V where the CO 2 evolution takes place in the presence of a small hydrogen evolution and jCO 2 -N 2 tends to a plateau value for more negative potentials; (II) a second area for cell potentials more negative than − 2.5 V (where hydrogen evolution starts to take place with a high slope in Fig. 1c) where jCO 2 -N 2 starts again to increase with the potential reaching a maximum value, where carbon dioxide and water reduction coexist and compete. A series of galvanostatic electrolyses was performed at a silver plate cathode at 12 mA cm −2 , 500 rpm and different CO 2 pressures (from 1 to 30 bar) to evaluate the effect of CO 2 concentration on the production of CO and other byproducts. As shown in Fig. 2a, according to the literature, at 1 bar CO was produced with a very low CE (~ 5%), which corresponds to a current density devoted to CO production lower than 0.6 mA cm −2 . Indeed, various authors have previously reported that quite low faradic efficiencies and current densities related to CO production are obtained at 1 bar at Ag foil cathodes [66,67]. The presence of other products of CO 2 reduction, including formic acid, was checked with no results, while a high amount of hydrogen was generated. This result explains once more why several researchers have tried to find alternative solutions such as the use of high surfaces electrodes or GDE. However, when the pressure was increased, a strong enhancement of the CO production was observed. Indeed, the CE CO increased to 16, 26 and 67% enhancing the CO 2 pressure at 10, 15 and 30 bar, respectively, thus confirming that the enhancement of the CO 2 concentration in the bulk is effective to assist the electrocatalytic properties of silver for the CO generation. Furthermore, at 15 and 30 bars, the presence of small amounts of formic acid in the liquid phase was observed.
One of the main disadvantages of the utilization of an undivided cell is the potential degradation of cathodic products to the anode surface; in this frame, it is important to highlight that our apparatus works in continuous gas-flow mode avoiding gas products accumulation and the potential cathodic reduction of CO. Indeed, the sum of the current efficiency of the cathodic products (CO, H 2 and HCOOH) was close to 100%

Effect of current density and mixing rate with pressurized carbon dioxide
The effect of current density was evaluated at both 1 and 30 bars, carrying out a series of electrolyses at 7, 12 and 30 mA cm −2 and 500 rpm. As shown in Fig. 2b, at 1 bar a very low CO production rate was achieved at all adopted current densities. Conversely, at 30 bar a strong effect of the current density was observed; indeed, the enhancement of the current density from 7 to 12 mA cm −2 resulted in a drastic increase of the CO production (from 0.2 to 1.5 mol h −1 m −2 ). However, a further enhancement of the current density from 12 to 30 mA cm −2 resulted in a slightly increase of CO production accompanied by a drastic decrease of CE CO (from about 65 to 38%). The effect of the current density can be due to the fact that the increase of the current density to 30 mA cm −2 involves quite high ΔV (~ 2.6 V) that enhance significantly the hydrogen evolution (Fig. 1c) and that, according to some authors, can limit the CO production since at these potentials CO could be more strongly adsorbed to the silver surface due to the back-donation of an electron to the CO specie [47].
To evaluate the effect of mixing on the performances of the process, a series of electrolyses was carried out at 30 bar and 12 mA cm −2 at different mixing rates (namely 0, 150, 300, 500 and 700 rpm). In this case, experiments were prolonged up to 3 h. It was shown that CE CO reaches a maximum for an intermediate value of 300 rpm (Fig. 3). Indeed, for higher values of rpm, CE CO decreases as a result of a higher generation of hydrogen. Conversely, for the lower values of the mixing rate, lower current efficiencies for hydrogen are achieved and it was observed a little amount of methane with a current efficiency of about 4%. These results show that under particular operative conditions the reduction of CO 2 at silver can give rise to the formation of methane. The formation of methane at silver plate cathode was previously reported by He [68] and Kuhl [69] in a divided cell using KHCO 3 as supporting electrolyte in the catholyte with current efficiencies lower than 4%. Other studies have shown that, at high pressure, the mixing rate can affect the process; indeed, working without mixing rate gave at copper-based cathodes higher CE in light hydrocarbons [38]. In order to understand the effect of mixing rate on the selectivity of the process, it is relevant to observe that under adopted operating conditions (high pressures of CO 2 and moderate current densities), the reduction of carbon dioxide is not expected to be affected by the mass transfer of CO 2 from the bulk to the cathode surface. Conversely, the mixing rate could affect the mass transfer of CO from the cathode to the bulk. Hence, at low stirring rate, the cathode should present a higher concentration of CO, thus favoring follow-up reduction paths, including the conversion of CO to methane. High stirring rate, furthermore, should enhance the mass transfer of protons from the bulk to the cathode surface, thus decreasing the local pH, and favoring the hydrogen evolution.

Effect of the supporting electrolyte at different CO 2 pressure and of the anode
The effect of the nature of the supporting electrolyte on the cathodic reduction of CO 2 at silver-based cathodes was evaluated at atmospheric pressure by several authors [31,70,71]. In particular, Verma and co-authors [71] have studied that effect of the nature of the supporting electrolyte on the reduction of CO 2 on Ag based GDEs using KOH, KCl and KHCO 3 and have found the higher production of CO with KOH. To evaluate the effect of the nature of the supporting electrolyte on the reduction of pressurized CO 2 to CO at silver cathode, some electrolyses were performed replacing K 2 SO 4 with KOH, KCl and KHCO 3 that, according to the literature, allow to achieve good productivity of CO. In the case of KOH some polarizations were also performed. Table 1 reports the results achieved after 2 h at 1 and 15 bars with the different adopted supporting electrolytes using a silver plate cathode under amperostatic conditions (12 mA cm −2 ). As shown in Table 1 (entries 1-3) at 1 bar, for all adopted electrolytes, using a compact graphite anode as counter electrode very low current efficiencies in CO were recorded (< 5%). At the end of electrolyses, the presence in solution of small particles of compact graphite, generated by the deterioration of the anode, was observed.
Hence, the electrolyses with KOH and KCl were repeated using a dimensional stable electrode Ti/IrO 2 -Ta 2 O 5 as anode. In both cases, higher CE CO were obtained (Table 1, entries 4 and 5), even if always lower than 10%. When experiments were repeated at 15 bars (Table 1, entries 6-8), the productions of CO increased for all the adopted supporting electrolytes, thus confirming the beneficial effect of CO 2 pressure on the cathodic reduction of carbon dioxide.
Furthermore, a strong effect of the nature of the supporting electrolyte on the performances of the process was observed. In particular, with K 2 SO 4 and KCl, at 15 bars (Table 1, entries 6 and 7) quite similar CE CO were obtained (44 and 39%), while with KOH and KHCO 3 ( Table 1, entries 8 and 9), a higher production of CO was obtained (CE CO of 60 and 65%). together with a lower cell potential. In particular, the energetic consumption decreased of about 50% using KOH and KHCO 3 (0.24 and 0.27 kWh/mol CO for KOH and KHCO 3 , respectively) instead of the other two supporting electrolytes (about 0.4 kWh/mol CO ), because of both the higher CE CO and the lower cell potential recorded with these salts. In our system, where carbon dioxide is continuously Counter electrode: compact graphite. V = 0.05 L fed to the system, H 2 CO 3 is expected to react with KOH to give KHCO 3 , thus giving rise to the formation of a buffer solution. Hence, similar results are expected using KOH or KHCO 3 . Indeed, these two electrolytes gave the highest values of CE CO . The slightly higher production of CO achieved in the presence of KOH could be due to various reasons such as transient effects, even if the saturation of the solution with CO 2 for 30 min before the electrolysis should reduce the not-stationary period, or a surface modification of silver electrode when it is immersed in KOH aqueous solution.
The reduction of CO 2 at silver in the presence of KOH was further investigated by polarization studies. Figure 4a reports the polarizations achieved using KOH or K 2 SO 4 under nitrogen and carbon dioxide atmosphere (1 bar). Similar forms of the polarizations were achieved at the two electrolytes. However, with KOH higher current densities were recorded both under N 2 and in CO 2 atmosphere (Fig. 4a,  b). In particular, when KOH is used the replacement of N 2 with CO 2 gives rise to an increase of the current density for working potentials up to − 1.7 V vs. SCE, when the hydrogen evolution is limited: hence, there is a range of potential (from − 1.3 to − 1.7 V) where hydrogen evolution is very limited and CO 2 reduction can take place. Conversely, for working potential more negative than − 1.7 V (when hydrogen evolution takes place in a relevant way), the addition of CO 2 gives rise to a decrease of the overall current density. As previously observed for experiments performed with K 2 SO 4 , this is probably due to the fact that the derivatives of carbon dioxide reduction are adsorbed at silver, thus reducing the active sites for water reduction, or to the depletion of protons at the cathode surface driven by CO formation. As shown in Fig. 4c, when the pressure of CO 2 is enhanced from 1 to 20 bar, a significant increase of the current density is observed, thus showing that the rate of carbon dioxide reduction depends on the concentration of CO 2 also using KOH. However, when the pressure is increased from 20 to 30 bar, no a significant change of the current densities was observed, thus showing that at these relatively high pressures, the process is no more affected by the concentration of CO 2 in the bulk. It was also shown that higher current densities are recorded for KOH with respect to K 2 SO 4 for each value of the pressure and of the cell potential (Figs. 1c and 4c), according to the above-mentioned results and considerations.

Effect of the nature of cathode
The silver plate cathode (named plate-Ag) is characterized by a quite low specific surface. Hence, in order to increase the performances of the process some polarizations, pseudopolarizations and electrolyses were repeated in the presence of a high surface Ag based electrode (hs-Ag).

Polarizations and pseudo-polarization measurements
Figure 5a reports some polarization curves recorded with the two cathodes. Under nitrogen atmosphere, high apparent current densities are recorded for hs-Ag, probably for the higher surface due to the occurrence of pore structure. However, at the more negative working potentials the polarizations achieved at the two electrodes become more similar. This behavior could be due to the fact that for very negative working potential, the massive hydrogen evolution could fill the pores, limiting the active surface available for water reduction. Under carbon dioxide atmosphere, the CO 2 reduction starts at similar potentials at the two electrodes, but a strong increase of the current density is recorded for hs-Ag with respect to plate-Ag, thus showing that the higher surface of hs-Ag can be exploited for CO 2 reduction. Figure 5b reports the value of jCO 2 -N 2 achieved at the two cathodes in the absence and in the presence of mixing. It is shown that at both electrodes, jCO 2 -N 2 presents a slight increase in the presence of the mixing, thus showing that at 1 bar the process is partially limited by the mass transfer of CO 2 to the cathode surface for both electrodes. Figure 5c reports the pseudo-polarizations achieved at the two cathodes at various CO 2 pressures. It is shown that (i) hs-Ag gives higher current densities with respect to plate-Ag at all adopted pressures and (ii) a smaller effect of pressure is observed for hs-Ag, thus showing that at this electrode the rate determining step, at least at pressure higher than 10 bar, depends in a limited way on the CO 2 concentration in the bulk.

Electrolyses
As shown in Table 2, the replacement of the plate-Ag cathode with the hs-Ag one allowed to achieve a strong increase of CO production by amperostatic electrolyses at 1 bar and 12 mA cm −2 ; indeed, the CE CO were about 10 and 31% at In particular, at hs-Ag cathode the same value of current density was achieved with a lower value of the working potential, according with the pseudo-polarization curves reported in Fig. 5, thus allowing to reduce the impact of water reduction.
To evaluate the effect of current density on the production of CO at the hs-Ag cathode, a series of amperostatic electrolyses was carried out at 1 bar and 12, 36 and 50 mA cm −2 . As shown in Table 2, at all adopted current densities, the replacement of plate-Ag (entries 1 and 2) with hs-Ag (3 and 4) gave rise to an enhancement of about three times of the CE CO and allowed to achieve a significant decrease of working and cell potentials. This is due to the fact that at low concentrations of carbon dioxide, the high surface of hs-Ag allows to speed up the process. Focusing on hs-Ag cathode, the enhancement of the current density from 12 to 36 mA cm −2 ( Table 2, entries 3 and 4) resulted in a strong increase of the CO production (from 0.69 to 1.61 mol h −1 m −2 ) as a result of the higher amount of charge passed, even if with a decrease of the CE CO from 31 to 24%, probably due to an increase of the hydrogen evolution due to the more negative working potential involved. When the current density was further increased to 50 mA cm −2 ( Table 2, entry 5), the production of CO decreased, with a dramatic reduction of CE CO to 9%, due to the occurrence of a quite negative working potential (− 2.35 V vs SCE) that can favor the HER and decrease the CO desorption, as above discussed [47].
In order to evaluate the effect of the pressure on the process at the high surface cathode, a series of electrolyses was performed at different pressures (1, 10, 20 and 30 bar) and current densities (12, 36 and 50 mA cm −2 ) at hs-Ag cathode. At 12 mA cm −2 , an increase of the pressure from 1 to 20 bars did not affect the performances of the process (Table 3, entry 1 and 2), according to the pseudo-polarization curves; indeed, a value of CE CO of about 30% was achieved at both 1 and 20 bars, probably due to the fact that, under adopted operative conditions, the r.d.s. does not involve the mass transport of CO 2 or its cathodic reduction. Conversely, at 36 and 50 mA cm −2 , a strong effect of the pressure was observed; indeed, as an example, an increase of the pressure from 1 to 10 and 20 bar at 36 mA cm −2 resulted in a simultaneously reduction of the cell potential of about 115 and 550 mV, respectively, and in an enhancement of CE CO from 24 up to 40 and 57% (Table 3, entries 3-5), probably due to the fact that at high current densities and working potentials the process becomes kinetically limited by the concentration of CO 2 . Indeed, as shown in Fig. 5b, at these values of current density and cell potentials, the current increases enhancing the pressure from 10 to 20 bars.
However, a further increase of the pressure from 20 to 30 bars resulted in a slightly decrease of the CE CO (from 57 to at 50%) and in a very small change of the total CO 2 reduction efficiency (computed taking in account the CO and the formic acid formation), which was close to 59% at both 20 and 30 bars, according to the pseudo-polarizations reported in Fig. 5b, because the concentration of carbon dioxide in the bulk at these high pressures becomes sufficient to sustain the mass transport and the cathodic reduction of CO 2 and the process is likely to be limited by a following step, similarly to what observed at low current densities.
The performances of hs-Ag and plate-Ag at high pressures are compared in Table 2. The increase of the pressure from 1 to 20 bar (entries 2, 4, 8 and 9) allowed to increase drastically both the CO production and the corresponding CE CO at both electrodes (from 8 to 50% at plate-Ag and from 24 to 58% at hs-Ag). However, at high pressures, the benefits of using hs-Ag is less important (CE CO 50% at plate-Ag and 58% at hs-Ag), because the high solubility of CO 2 achieved at high pressures makes less relevant the surface of the cathode. These results seem quite important because it shows that relatively high pressures can allow to avoid the use of more complex and expensive electrodes.

Time stability of performances at high pressure
To evaluate the stability of the performances of the process, two long amperostatic electrolyses were performed at plate-Ag with both KOH and K 2 SO 4 electrolytes at 15 bar and 12 mA cm −2 (Fig. 6). It was shown that the process takes place under very stable conditions in terms of production of CO, CE CO and cell potential using KOH as supporting electrolyte; indeed, a CE CO close to 70% was recorded for 10 h for a very stable value of the cell potential. Conversely, for K 2 SO 4 , the CE CO presented a small decrease from 53 to 45% after 2 h, a period of stability for about 6 h and a decrease for the last 3 h to reach a value close to 20% after 10 h and a correspondent increase of the cell potential up to 3.2 V,

Conclusions
Electrochemical conversion of CO 2 to CO at silver electrodes was studied in detail using a liquid phase electrolyzer, investigating the effect of several operating parameters, including CO 2 pressure (1-30 bar), nature of supporting electrolytes and current density. It was found that a simple and relatively cheap undivided cell equipped with simple silver sheet cathodes can allow to obtain interesting current efficiency in CO (CE CO ) using pressurized conditions at medium current densities. For example, at silver plate electrode and 12 mA cm −2 , the current efficiency of CO dramatically increases from really low values (< 4%) at 1 bar to 67% at 30 bars. Furthermore, it was found that: -the effect of the surface of the electrode strongly depends on the adopted CO 2 pressure. At 1 bar the increase of the surface allows to improve strongly the performances of the process; as an example, the replacement of the plate Ag electrode with a high surface one allows to increase the CE CO from 8 to 24% at 36 mA cm −2 . However, at 20 bar the effect of the surface is reduced; indeed, the replacement gave rise to a quite small improvement of the CE CO from 50 to 57% at plate and high surface Ag electrode, respectively, at 36 mA cm −2 ; -there is an important effect of the nature of the supporting electrolyte at high pressure. Indeed, at 15 bars, a higher production of CO with a quite high current efficiency of about 60-65% was reached by replacing K 2 SO 4 or KCl (both CE CO ~ 40%) with KOH or KHCO 3 at silver plate cathode and 12 mA cm −2 ; -at high pressure, the mixing rate can affect the product distribution even if mass transfer control of CO 2 electroreduction can be excluded; in particular, under particular operative conditions, the reduction of CO 2 gives rise to the formation of CH 4 even if with low current efficiency (4%); -the effect of current density strongly depends on both the adopted pressure and surface of the electrode. In particular, the productivity of the system can be increased, by enhancing the current density, using (i) high pressure of CO 2 or (ii) high surface electrodes.
Furthermore, the stability of the performances highly depends on the adopted operating conditions. In particular, the process was stable with the time maintaining a quite high current efficiency for CO of about 70% for 10 h using mild  On the overall, the results reported in this manuscript demonstrate that the utilization of carbon dioxide at mild pressure allows to operate with simple electrodes in undivided cells by maintaining quite good results. In this context, it is worth to mention that, according to several authors [58,59], the utilization of pressurized conditions up to 20 bars does not impact remarkably on the global costs of the process, involving just a small increase of operative and capital costs with respect the other main costs. Conversely, it has been shown the implementation of CO 2 reduction technologies on large scale is mainly limited by the electrolyzer costs (almost 50% of the overall costs) [72][73][74], predominantly due to the energy consumption and GDE costs (e.g. estimated GDE cost of about 7700 € m −2 [75]).
From the industrial point of view, the utilization of pressurized conditions could be an interesting alternative to the direct integration of the pressurized products' streams (i.e. syngas) as a feedstock with other conventional industrial processes currently performed at high pressure. In this framework, further investigations will be necessary to improve the scalability on large scale of this apparatus. Indeed, this process may suffer of the potential formation of an explosive mixture of H 2 and O 2 and of a low CO 2 utilization rate; hence, further researches will be focused on this topic and various strategies will be tested, including: (i) the implementation of sacrificial anodes to avoid the O 2 evolution reaction; (ii) the utilization of higher-selective CO 3D-cathodes (mesh, perforated, foam) to suppress the H 2 production and improve the CO productivity at high pressure; (iii) the development of a novel reactor configuration operating in continuous mode, in order to prevent the mixing of potentially formed cathodic co-product, hydrogen, with the anodic product, oxygen and to increase the conversion rate. It is worth to mention that a similar alternative to the latter was successfully evaluated by some authors [36,55], which have estimated an improvement of the CO 2 conversion rate up to 26% and an enhancement of the selectivity towards CO using a divided semi-continuous mode system under pressurized conditions.