2.1 Electrode Preparation

CuO powder particles delivered by Johnson Matthey were used for the preparation of Cu-based GDEs. A CuO ink was prepared with 40 mg of CuO powder, 160 mg of 5 wt% Sustanion XA-9 anionic ionomer in ethanol solution, and 4 mL of isopropanol (99.9%, Sigma Aldrich). The ink was sonicated in an ultrasonic water bath for 30 min and immediately drop cast onto a 10 cm × 5.5 cm carbon-based gas diffusion layer (GDL) from Freudenberg (H23C2), which consisted of two layers: a carbon-based microporous layer containing hydrophobic PTFE binder and a macroporous layer consisting of carbon fibers. Together with the drop-cast catalyst layer, the GDE therefore had three layers. Later, the electrode was dried overnight at room temperature under a N₂ atmosphere with a flow of 50 sccm. The dried electrode batch was cut into two pieces with a 48 mm × 44 mm cutting template. The active area that took part in the electrolysis was 10 cm². The catalyst loading was 0.73 mg cm⁻².

2.2 Electrode Surface Analysis

A Bruker D2 PHASER diffractometer (Cu K radiation, scan rate of 0.02° s⁻¹) was used to perform ex situ X-ray diffraction (XRD) analysis on electrodes pre- and postelectrolysis. Scanning electron micrographs (SEM) of the GDEs pre- and postelectrolysis were taken at different magnifications with a high-resolution field-emission scanning electron microscope (JSM-7500F, JEOL). Both XRD and SEM results are given in the Supporting Information.

2.3 Experimental Set-Up

An internally designed flow cell was used in all experiments (Figure S5 and S6, Supporting Information). The cell consisted of three compartments where gas, catholyte, and anolyte flow in and out, respectively. PTFE and PEEK materials were used to determine the fluid flow patterns, to ensure cell tightness and provide better mechanical stability for the GDE and membrane. An IrOx-coated electrode with a 10 cm² active area (Electrocell) was used as the anode and the CuO GDE described above as the cathode. Anode and cathode chambers were separated from one another by a cation exchange membrane (CEM) (Nafion 117, Ion Power). 250 mL of 1 m KHCO₃ (+99.7% in dry basis, Alfa Aesar) was circulated in the system as electrolyte between the gaps of cathode–membrane (catholyte) and anode–membrane (anolyte). The electrolyte flows circulating in both anode and cathode gaps were mixed in an electrolyte chamber external to the cell and pumped continuously through the system with microdiaphragm liquid pumps (NFB 25 KPDCB-4A, KNF) operated at a constant flow of 100 mL min⁻¹. The gas chamber was fed with CO₂ (+99.998%, Linde) which was humidified by a custom-made water bubbler at room temperature. Gas flow rates in the system were controlled with mass flow controllers (SFC5400 Sensirion). In contrast to the electrolyte, the gas head spaces of catholyte and anolyte were kept separate. The gas outlet from the cell was mixed with catholyte head space. The setup and cell details are shown in the Supporting Information.

2.4 Electrochemical Measurements

2.5 Product Analysis

The catholyte gas head space was connected to a gas chromatograph (GC) for quantitative product analysis (7890B Agilent, Santa Clara, USA). N₂ was mixed with the product stream before the GC inlet as an internal standard for quantification. He is the carrier gas enabling the detection of H₂ as a negative peak. A thermal conductivity detector (TCD) and three serially connected columns were used: a HayeSeP Q-column, a Porapak Q-column for the separation of CH₄, CO₂, and C₂H₄, and a molecular sieve 5 Å column to separate N₂, O₂, and CO. 1 mL of the gas product stream was automatically injected every 20 min from the cell to the GC during the course of the electrolysis.

¹H nuclear magnetic resonance (NMR) was used for quantification of liquid products. Electrolyte circulated constantly in a separated closed system between reservoir and cell sampled from a septum. The NMR measurements were performed in a 500 MHz Bruker (Bruker Bio-Spin, Karlsruhe, Germany) following the method described by Cuellar et al.^{[ 45 ]} The water peak was suppressed by a presaturation sequence. An aliquot of the KHCO₃ electrolyte-containing liquid products (300 μL) was mixed with sodium fumarate as an internal standard (50 μL) in D₂O (250 μL) for quantification.

3.1 Cell Setup and Application of Back Pressure

Back pressure experiments were performed on carbon-based GDEs embedded with a commercial CuO catalyst, positioned in a two-gap flow cell architecture (see Experimental Section and Figure S1–S4, Supporting Information, for electrode description and characterization). The cells consist of two electrodes separated by a CEM and two gaps that accommodate electrolyte flow (1 m KHCO₃, flow rate = 100 mL min⁻¹; Figure 1). The third compartment, namely, the gas chamber where humidified CO₂ (atmospheric gas inlet pressure) flows, comes into contact only with the cathode and therefore does not reside between the two conductive electrodes. A membrane valve is positioned at the gas outlet stream that carries mainly unconverted CO₂, H₂, and CO₂RR gas products. The absolute pressure in the gas compartment is increased and controlled by the membrane valve.

Pressures in the gas compartment and in the catholyte chamber were monitored with pressure sensors. The back pressure (ΔP) applied on the GDE is defined as the difference between P1 (gas) and P2 (catholyte). This was set to 70 and 130 mbar at high gas flow rates and 40 mbar in low-gas flow rate experiments. It must be noted that higher back pressures can lead to so-called “flow-through” operation, where the pressure difference enables the gas to be pushed through the porous electrode to the catholyte chamber on the other side. For this reason, back pressure must be set to a value that maintains the desired “low-by” operation. This threshold value for flow-by strongly depends on the electrode architecture and properties, employed flow rate, catholyte properties, and cell design and size. We stress that optimum back pressure values can vary in differing cell systems and should therefore be optimized accordingly.

3.2 HER Suppression Under Back Pressure

The stability of the CuO electrodes was tested at an applied current density of 200 mA cm⁻² in two series of experiments: base case without back pressure and with ΔP = 130 mbar. In Figure  2 , the FE for H₂ and C₂H₄ is depicted versus time for both cases. HER as the competing reaction to CO₂RR is a good indicator for a decrease in CO₂RR catalytic activity. ${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ is chosen as an indicator of CO₂RR stability, as it is one of the more valuable CO₂RR products. At a back pressure of 130 mbar, ${\text{FE}}_{{\mathrm{H}}_2}$ is suppressed below 25% for 72 h, in comparison with the base case where H₂ selectivity is clearly higher already from the outset at 25–40% for the first 48 h and only becomes more dominant over time (FE_HER > 50%). Two separate phenomena are observed here: 1) Back pressure significantly decreases the selectivity for H₂. At 24 h, FE_HER is 15.1 (±1.0)% under ΔP = 130 mbar, whereas in the base case it is almost double this at 30.7 (±5.6)% in the base case; and 2) Back pressure provides longer stable CO₂RR operation. ${\text{FE}}_{{\mathrm{C}}_{\text{2}}}{}_{{\mathrm{H}}_{\text{4}}}$ remained above 20% for only 36 h in the base case, while back pressure of 130 mbar prolonged its stability to 72 h, corresponding to a stability increase by a factor of two when 20% is assumed as the stability threshold.

According to the adsorption equilibrium equation, adsorbed CO₂ species coverage (${\text{CO}}_2^{*}$) increases with increasing CO₂ concentration in the gas phase, assuming that CO₂ adsorption is molecular and first order.^{[ 46, 63 ]} Figure  3 schematically illustrates the increasing surface coverage of ${\text{CO}}_2^{*}$ with increasing partial pressure of ${\text{CO}}_2$ provided by application of back pressure. As ${\text{CO}}_2$ adsorption (${\text{CO}}_2\to {\text{CO}}_2^{*}$) is the rate-determining step of CO₂ reduction to <2e⁻ products and the reaction from ${\text{CO}}_2^{*}$ to CO* is faster, a higher ${\text{CO}}_2$ partial pressure would accordingly translate into an abundant CO* surface coverage.^{[ 8 ]} Based on our results, it can be assumed that increased back pressure leads to higher CO* coverage and therefore HER suppression.

On the second phenomenon, two explanations for longer stability under back pressure are possible. As mentioned, catalysts producing more C₂₊ relative to C₁ products may be more stable due to a possible graphitic intermediate in the C₁ route.^{[ 35, 36, 47, 59-63 ]} In mechanistic studies conducted by Rahaman et al.^{[ 43 ]} and Akhade et al.,^{[ 58 ]} the poisoning C* species is postulated to be formed via reduction of COH*, which is an intermediate of CO*-to-CH₄ reduction. As discussed in the following subsections, abundant CO* coverage provided by back pressure suppresses C₁ hydrocarbon production (CH₄) and promotes C₂₊ selectivity. This may have resulted in the observed longer stability by avoiding poisonous C* intermediate formation, as this route is not preferred under high CO* coverage. Another possibility is that back pressure keeps the capillary pressure (local pressure difference between wetting and nonwetting media) at a sufficiently high value that minimizes the flooding of the catalyst and GDLs, an otherwise highly likely cause of deactivation.^{[ 60-62 ]} When flooding is minimized, the salts present in the aqueous electrolyte do not crystallize in the pores of the electrode and do not increase its hydrophilicity. We hypothesize that flooding occurs to a smaller degree or can be delayed under back pressure. Mass transport limitation on CO₂ availability therefore remains low and can be avoided for a longer period.

3.3 C₁ Product Formation under Back Pressure

FEs of three main C₁ products, CO, COOH⁻, and CH₄, obtained under CO₂RR conditions at an applied current density of 200 mA cm⁻² and at three different back pressure conditions are shown in Figure  4 . Methanol is detected only in trace amounts and has not been included in the results.

The key intermediate CO is produced in significantly larger amounts when back pressure is applied, indicating that CO* desorbs before being reduced further to >2e⁻ products. FE_CO = 18.3 ± 0.6% in the base case compared with 26.0 ± 2.1% under 70 mbar and 26.7 ± 1.8% under 130 mbar back pressure. Therefore, a correlation between increasing back pressure and CO desorption may be established. A similar trend of increased CO production at higher pressures is described in the literature, the cause of which is given as the increased CO* surface coverage under higher pressures.^{[ 36, 43, 63 ]} It has also been claimed that abundant ${\text{CO}}_2^{*}$ coverage reduces the CO* binding energy to the catalyst surface via adsorbate–adsorbate repulsion, which, in turn, causes CO to bond weakly and therefore its desorption.^{[ 68 ]} Additionally, we hypothesize that shorter residence times due to a stronger driving force for diffusion through the catalyst layer at higher back pressures deprive ${\text{CO}}_2$ of sufficient residence time to react to >2e⁻ products.

The production of formate, the only product that is formed exclusively from ${\text{CO}}_2^{*}$ and not through CO*, is also slightly greater under back pressure, enhanced by the higher ${\text{CO}}_2^{*}$ surface coverage under such conditions. Notably, this increase in FE_COOH ⁻ is not as high as that of CO. This might suggest that the proportion of CO₂ atoms bound to the Cu surface via the C atom (leading to CO pathways) becomes significantly higher under back pressure, whereas the number of CO₂ atoms bound to Cu surface by O atom(s) (leading to HCOO– pathway) does not increase as much.

FE for CH₄ is observed to decrease from 0.6% in the base case, to 0.1% at 70 mbar back pressure and to 0% at 130 mbar back pressure. The lower selectivity toward CH₄ in the base case can be attributed to the intrinsic catalytic behavior of CuO. Oxide-derived Cu catalysts are known to demonstrate low selectivities for CH₄ ^{[ 43, 69 ]} and the presence of several Cu₂O facets is proven to diminish methane formation.^{[ 43 ]} The gradual decrease of ${\text{FE}}_{{\text{CH}}_4}$ with increasing back pressure and the complete suppression of CH₄ at 130 mbar may be correlated to abundant ${\text{CO}}_2^{*}$ coverage, which prohibits H* surface coverage and therefore protonation of CO* intermediates required for CH₄ formation.

3.4 C₂₊ Product Formation under Back Pressure

C₂₊ products from CO₂RR on Cu surfaces typically possess either 2 or 3 carbon atoms. It is logical to divide >2e⁻ CO₂RR products into two groups as hydrocarbons and oxygenates as they follow distinct potential dependencies, mechanistic pathways, and surface coverage dependencies. Ethylene and methane constitute the hydrocarbons group whereas methanol, ethanol, n-propanol, allyl alcohol, acetaldehyde, propionaldehyde, and acetate constitute the oxygenates group. Ethylene is categorized separately from the others as it is the only C₂₊ hydrocarbon molecule, the rest being oxygenates. Acetaldehyde, propionaldehyde, and propionate are detected only in trace amounts and have not been included in the results, while allyl alcohol and acetate are observed in very small amounts and have been included within the C₂₊ and oxygenate groups.

${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ does not change significantly at different back pressure levels and accounts for ≈23–25% in all cases (Figure  5a). FE for total C₂₊ products is however enhanced from 36.9% in the base case to 43.0% at ΔP = 70 mbar and 43.9% at ΔP = 130 mbar, and clearly stems not from ethylene but rather from oxygenate formation (Figure 5b). The ratio of >2e⁻ oxygenates to hydrocarbons can be tuned between 0.53 and 0.85 at 0 < ΔP < 130 mbar. An enhancement of FE_oxygenates from 13% to 20% (accounting for an improvement of 54% in oxygenate selectivity) is driven mainly by ethanol and propanol formation and can be achieved simply by varying the back pressure.

The observed correlation between back pressure and C₂₊ selectivity can be explained with the predicted second-order kinetic dependency of CO* coverage for C–C coupling,^{[ 37 ]} leading to accelerated C₂₊ production kinetics with higher CO* coverage. Nevertheless, we do not exclude the effect of possible changes in local pH when local CO₂ concentrations are higher due to back pressure. The selectivity between ethylene and oxygenates may be explained by the effects of high CO* coverage on the reaction mechanism, as Li et al. discussed with both density functional theory calculations and experimentation:^{[ 46 ]} *CHCOH intermediates tendentially deoxidize to *CCH under lower CO* coverage, leading to the ethylene reaction pathway versus tending to hydrogenate to *CHCHOH under higher CO* coverage and hence the oxygenate pathway. Our experimental results confirm this. Oxygenate selectivity improvements (and correspondingly improved C₂₊ selectivity) achieved with higher CO₂ or CO feed concentrations have similarly been reported by others.^{[ 3, 15, 51 ]} We therefore demonstrate that such trends may also be influenced by applying a back pressure.

In parallel with the aforementioned stability of C₂H₄ formation, FE_C2+ drops only minimally from 43.9% to 40.5% between the 3rd and 48th h of electrolysis when under ΔP = 130 mbar. In stark contrast to this, ${\text{FE}}_{{\mathrm{C}}_{\text{2}\text{+}}}$ diminishes drastically from 36.9 to 22.8% within same period if no back pressure is applied. A detailed breakdown of product selectivities over time is given in Table S4, Supporting Information.

3.5 Deconvolution of Back Pressure Effects on Stability and Selectivity

Control of ${\text{CO}}_2^{*}$ coverage over catalyst surface and capillary pressure control circumventing the electrode flooding have been mentioned as the possible underlying reasons of the stability and selectivity effects of back pressure application. To decouple these two phenomena from each other, two various experiment series both with ΔP = 130 mbar (P _abs = 1.13) back pressure application at different two different feed compositions were performed: 100% CO₂ feed ($P_{{\text{CO}}_{\text{2}}}$= 1.13 bar) and CO₂: N₂ feed in volumetric ratio of 88.5:11.5 ($P_{{\text{CO}}_{\text{2}}}$= 1.00 bar). The diluted feed experiments with ΔP = 130 mbar have the same CO₂ partial pressure of 1.00 bar as the control experiments with no back pressure in the base case (P _abs = 1 bar, $P_{{\text{CO}}_{\text{2}}}$= 1 bar).

C₂H₄ and H₂ FEs over time are shown as the stability indicators in Figure  6 . N₂-diluted CO₂-fed experiments at ΔP = 130 mbar exhibit C₂H₄ stability for ≈48 h, almost identically to the base case with no back pressure. It is significantly inferior to the 72 h stability of pure CO₂-fed experiments at ΔP = 130 mbar. H₂ was not as suppressed with $P_{{\text{CO}}_{\text{2}}}$= 1.00 bar as it was with $P_{{\text{CO}}_{\text{2}}}$= 1.13 bar, its FE was on constant rise after 48 h, similar to the base case with no back pressure. In the light of these results, back pressure can provide H₂ suppression and stable C₂H₄ formation primarily by means of increased ${\text{CO}}_2^{*}$ coverage and not necessarily through capillary pressure preventing electrolyte incorporation into the catalyst layer, as the same ΔP = 130 mbar with diluted CO₂ feed could not maintain stable C₂H₄ formation and H₂ suppression.

Figure  7 sets out the effects of $P_{{\text{CO}}_{\text{2}}}$ on product distribution under the same back pressure of ΔP = 130 mbar. CO selectivity did not differ significantly and failure of H₂ suppression at $P_{{\text{CO}}_{\text{2}}}$= 1.00 bar was discussed thoroughly above. ${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ was greater at $P_{{\text{CO}}_{\text{2}}}$= 1.00 bar with diluted CO₂ feed (29.6%) compared to the $P_{{\text{CO}}_{\text{2}}}$= 1.13 bar case (24.3%). COOH⁻ was formed substantially in smaller amounts at lower $P_{{\text{CO}}_{\text{2}}}$. Oxygenate selectivity was significantly improved at higher $P_{{\text{CO}}_{\text{2}}}$. In line with the previous findings, we confirm that high ${\text{CO}}_2^{*}$ coverage at $P_{{\text{CO}}_{\text{2}}}$= 1.13 bar promotes COOH⁻ and oxygenate selectivities. $P_{{\text{CO}}_{\text{2}}}$= 1.00 bar condition with diluted CO₂ feed (88.5 vol%) possibly enabled a milder ${\text{CO}}_2^{*}$, shifting the oxygenate–hydrocarbon (C₂H₄) selectivity toward the hydrocarbon.

Base case experiments (P _abs = 1 bar) and dilute CO₂ feed experiments at ΔP = 130 mbar back pressure (P _abs = 1.13 bar), both employing $P_{{\text{CO}}_{\text{2}}}$= 1 bar, showed similar product distribution (Figure S8, Supporting Information). The slight differences favoring CO and C₂H₄ over H₂ in the dilute CO₂ feed at ΔP = 130 mbar may be attributed to the prevention of electrolyte incorporation to the catalyst layer by the presence of back pressure.

3.6 Combined Effect of Back Pressure and Low Flow Rates

Despite the improvements obtained in terms of stability and selectivity toward C₂₊ oxygenates through back pressure applications, the increase observed in CO production remains (Figure 4) and limits the overall selectivity for C₂₊ that can be achieved as it implies lost opportunity for its further reduction. Additionally, improved C₂H₄ selectivity has not been achieved despite abundant surface coverage of ${\text{CO}}_2^{*}$ provided by back pressure.

Shorter residence times at the catalyst-coated surface of the electrode is a possible reason for the above observations. A series of experiments at a lower CO₂ feed flow rate were conducted to investigate whether a further reduction of CO* may be achieved by a longer residence time and a more optimum surface coverage for C₂H₄. Feed flow rates are expressed in terms of λ, which is the ratio of actual CO₂ flow rate to the CO₂ flow rate theoretically required for full utilization of electrons for ethylene formation (${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ = 100%; see Equation (2) in the Experimental Section). A CO₂ feed of 50 sccm, equivalent to λ = 10.78, is supplied for high flow rate experiments and 14 sccm CO_2, equivalent to λ = 3.01, is fed to the electrolyzer for low flow rate experiments. CO₂RR was conducted under four different conditions: high flow rate experiments under ΔP = 130 and 0 mbar back pressure and low flow rate experiments under ΔP = 40 and 0 mbar back pressure. The back pressures applied in each of these flow rate conditions correspond to the maximum achievable back pressure if a flow-by operation were to be maintained; a pressure difference exceeding these values for the two cases caused flow-through operation. It is worth noting that the maximum achievable back pressure for flow-by operation strongly depends on electrode characteristics, cell design, catholyte wetting properties and gas flow rate, and the values reported here are therefore highly specific to the experimental materials and design employed.

It is important to emphasize that CO₂ availability in the catalyst layer of the electrode changes fundamentally under low flow rate conditions. The bulk CO₂ gas concentration is reduced from 81.9 to 33.3 mol%, when CO₂ is fed into the electrolyzer cell inlet at 50 and 14 sccm, respectively (see Table S2, Supporting Information). The local CO₂ gas concentration at the catalyst is predicted to be even lower in the low λ case, as a thicker gas boundary layer with higher mass transfer resistance can be expected. Therefore, the difference between the bulk and locally available CO₂ gas concentrations is predicted to be greater at lower flow rates. Tan et al.^{[ 63 ]} discussed the same trend for CO₂ local availability in CO₂RR environments at low flow rates.

As shown in Figure  8a, FE_CO declines from 18.3% to 7.0% when the CO₂ flow rate was reduced from λ = 10.78 to 3.01. Interestingly, an even lower FE for CO was observed under combined back pressure and low λ conditions (from 26.7% to 4.4%). Back pressure enhanced CO production at higher λ, whereas it appeared to facilitate the opposite at lower λ. This suggests that the surface coverage at combined back pressure and low flow rates offers an environment where CO* does not desorb but further reacts to C₂₊ products. Lower λ conditions showed H₂ FEs similar to that of high λ, at the ≈25% range, independent of back pressure. HER was not suppressed at combined low λ and back pressure (${\text{FE}}_{{\mathrm{H}}_2}$ = 25.0%) as much when compared with high λ and back pressure case (${\text{FE}}_{{\mathrm{H}}_2}$ =14.9%). The interpretations from CO and H₂ production signal that CO₂ surface coverage reaches a smaller extent under lower flow rates, leading to lower CO* coverage and more abundant H*. ${\text{FE}}_{{\text{CH}}_4}$ going up to 1.83% at low λ confirms that H* availability is higher at lower flow rates.

Most interestingly, a lower CO*/H* coverage ratio and shorter residence time obtained by combined back pressure and low λ boosted C₂H₄ production. ${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ is increased from 24.3% to 38.5% when a low flow rate is used in combination with back pressure. A smaller improvement to the base case, to ${\text{FE}}_{{\mathrm{C}}_2{\mathrm{H}}_4}$ of 31.0%, is observed at low flow rates without back pressure, this would suggest that back pressure contributes significantly to overall improvement and that the combined effects of optimum CO*/H* coverage and longer residence time in the catalyst layer favor C₂H₄ formation. Together with boosted C₂H₄ production, ${\text{FE}}_{{\mathrm{C}}_{\text{2}\text{+}}}$= 60% was achieved by combining back pressure and low flow rates, which is significantly greater than the 37% achieved in the base case (Figure 8b).

Finally, Figure  9 shows how C₂₊ conversion efficiency depends on feed flow rate and back pressure. A combination of low feed flow rate and back pressure delivers the highest C₂₊ conversion efficiency at 19.8%, compared to 3.5% in the base case. Significantly lower levels of unconverted CO₂ at low feed flow rates and higher selectivity towards C₂₊ with back pressure and low flow rate enable this remarkable improvement in the conversion efficiency. C₂₊ conversion efficiency is defined in the Equation (3) in the Experimental section.