A. Morphologic aspects of the GDL/composite catalyst

The GDL initially had a paper consistency with thickness: 325 μm, mass: 110 g/m², porosity: 80% and air gas permeability: 1.5 s. Thus, Fig. 1 shows SEM images of the back [Fig. 1(a)], top of the GDL before [Fig. 1(b)] and after the growth of a Cu₂O–CuBr (N-ED30) film [Fig. 1(c)]. The back surface of the GDL clearly was very open [Fig. 1(a)]. Comparison of the bare GDL top with that supporting the Cu₂O–CuBr (N-ED30) film [Figs. 1(b) and 1(c)] clearly demonstrated the change from hydrophobic to hydrophilic character triggered by the electrodeposited film (see contact angle photos of a water drop on top of them). We will further address this important feature in the results section below. The surface morphology of the Cu₂O–CuBr (N-ED30) film was composed of compact large flat crystals partially covered with small dark entities (see Table I). It is worth noting that the openness of the GDL facilitated CO₂ gas passage from the back to its front side to be electroreduced at the hybrid Cu₂O–CuBr/electrolyte interface.

FIG. 1.

SEM images of the gas diffusion layer (Sigracet 35BC GDL) used as support for the electrodeposited composite films. The back and top faces of the GDL are shown in images (a) and (b) respectively, while an electrodeposited Cu₂O–CuBr (N-ED30) film on the top face of the GDL is shown in image (c). Contact angle photos of a water drop on top of the bare GDL before (b) and after (c) the growth of the Cu₂O–CuBr (N-ED30) film, clearly demonstrated the change from hydrophobic to hydrophilic character triggered by the electrodeposited film.

Lactate species have been commonly deployed for Cu₂O electrodeposition. ^{Reference Golden, Shumsky, Zhou, VanderWerf, Van Leeuwen and Switzer25,Reference Wang, de Tacconi, Chenthamarakshan, Rajeshwar and Tao26} The complexing properties of lactate are required during electrodeposition to maintain soluble Cu²⁺ ions (to be reduced to Cu⁺) and avoiding chemical precipitation of Cu(OH)₂ triggered at any alkaline pH. For L-type films (i.e., grown in the presence of lactic acid), the molar ratio of lactic acid to Cu²⁺ ion precursor was adjusted to 10, and used at two chosen pHs (9 and 7). The choice of these pH values reflected the facts that (i) the surfactant properties of the lactic acid at pH ≤ 9 are known to promote Cu₂O electrodeposited films with preferential (100) crystallographic orientation ^{Reference Wang, de Tacconi, Chenthamarakshan, Rajeshwar and Tao26} ; and (ii) a neutral pH promotes a dominance of n-type semiconductor behavior with high electronic conductivity which is expected to benefit at negative potentials. ^{Reference Wu, Yin, He, Huang, Zhou and Zhang27}

Representative SEM images of films grown from an electrolytic bath containing 0.2 M CuSO₄·xH₂O + 0.1 M NaBr + 2 M lactic acid at pH 9 are shown in Fig. 2. The pristine Cu₂O (L9-ED 15) is shown in Fig. 2(a) to compare with hybrid films prepared sequentially [Fig. 2(b)] and simultaneously [Figs. 2(c) and 2(d)]. The pristine Cu₂O film displays intersecting cubic crystals all over the film while the sequential film growth [Fig. 2(b)] shows some crystal erosion as indicated by uneven or rounded corners of the cubic crystals. On the other hand, the simultaneous electrodeposition of Cu₂O and CuBr performed at two electrodeposition times [Fig. 2(c) for 15 min and Fig. 2(d) for 20 min] shows that the resulting films were not composed of cubic nanostructures. In fact, Fig. 2 is clear evidence that the {100} cubic lateral planes were only maintained in pristine Cu₂O [Fig. 2(a)] and to some extent in the sequentially grown Cu₂O–CuBr (L9-ED 15-5) composite [Fig. 2(b)]. Interestingly, no intact cubic crystals (with {100} lateral sides) remained when the bromide precursor was in the electrolytic bath at electrodeposition times ≥15 min [Figs. 2(c) and 2(d)]. Therefore, when the growing cubic Cu₂O crystals are exposed during the entire electrodeposition time to the etching effect of bromide in the electrolyte, the resulting films are made of rounded grains which become larger as the electrodepositing time increases [Figs. 2(c) and 2(d)].

FIG. 2.

SEM images showing the progressive transformation of Cu₂O cubic crystals by exposure to 0.1 M NaBr solution during different electrodeposition times. (a) Cu₂O grown for 15 min in 0.2 M CuSO₄·xH₂O + 2 M lactate (b) Cu₂O/CuBr containing pre-grown Cu₂O (EDτ 15 min) followed by 5 min in electrodeposition bath with bromide (c) Cu₂O–CuBr grown for 15 min in 0.2 M CuSO₄·xH₂O + 2 M lactate + 0.1 M NaBr and (d) same as (c) but grown for 20 min. All electrolytic solutions were adjusted to pH 9 and the deposition potential was −0.4 V.

The relative amounts of Cu₂O and CuBr in the composite film can be simply tuned by pH control of the lactate bath (see Table I). Thus formation of CuBr could be enhanced by adjusting the pH of the lactic electrolytic bath to 7 instead of 9. The SEM characterization results for L7 films are presented in Fig. 3. These films consisted of macro-sized perfectly perfectly-grown cubic structures [Fig. 3(d)] and contained nanocrystalline rectangular plates incrusted on the surface [Fig. 3(e)]. Besides, L7 films show the striking morphological changes as the EDQ increased [Figs. 3(a)–3(c)]. Compositional assays by EDX also revealed differences in the relative amounts of oxide and bromide relative to the bulk grains but importantly, these films also contained a significant amount of copper unlike the L9 films (see Table I). Other morphological differences between nitrate- and lactate-grown films are worthy of note. In the former case, (111) faces of compact CuBr prevailed [Fig. 1(c)] while in the latter case, perfect cubic entities exposing (100) planes to the electrolyte (Fig. 3) were the norm. XRD data on selected samples are contained in Supplementary Material (see Figs. S3–S5).

FIG. 3.

SEM images obtained during the growth of composite Cu₂O–CuBr (L7-EDτ) films for selected electrodeposition charges: after EDQ = 2 (a, d), 4 (b) and 5 C/cm² (c). Films were electrodeposited on the Sigracet GDL (35 BC) at −0.4 V (60 °C) using a solution of 0.2 M CuSO₄ + 0.1 M CuBr + 2 M lactate adjusted to pH 7. (e) Amplifying the morphology of the platelet incrustations on the cube surfaces.

B. Electrolysis results

The current efficiency for CO₂ reduction is known to diminish with decreasing CO₂ pressure and at the expense of an increase of H₂ evolution. Therefore, in the experiments below, CO₂ was constantly forced through the back of the Sigracet GDL at a flow rate in the range: 1–10 mL/min. All data were obtained with the carrier gas flowing through the back channels of the working electrode and thus in contact with the GDL back side which was hydrophobic and free of the hydrophilic Cu₂O–CuBr film [Figs. 1(b) versus 1(c)]. The difference in hydrophobicity at the back and hydrophilicity at the top (due to the presence of Cu₂O–CuBr film) enhanced transport of freshly-generated hydrocarbon gases to the back of the GDL. This important feature circumvents their accumulation at the electrode top surface which otherwise would have blocked essential catalytic sites thus decreasing the rate of the targeted electrochemical reaction. ^{Reference de Tacconi, Chanmanee, Rajeshwar and Dennis28} Moreover, as aqueous electrolytic media were used, the hydrophobic/hydrophilic properties of the GDL/Cu₂O–CuBr cathode also precluded electrolyte ingress to the gas channels located at the back side of the GDL. ^{Reference de Tacconi, Chanmanee, Rajeshwar and Dennis28}

Figure 4 contains representative current density-potential data [Fig. 4(a)] and product evolution profiles as monitored by on-line GC [Fig. 4(b)]. The best performance for ethylene formation occurred at the most negative potentials, although formation was also seen at less negative potentials where CO formation was also detected (data not shown). This latter observation has mechanistic implications as elaborated later. The current was seen to increase with more negative potentials, passing by a slow increase and then growing exponentially at potentials negative of −1.2 V [Fig. 4(a)]. Most interestingly, the amount of C₂H₄ formation increased significantly in the potential range between −1.0 V and −2.0 V, while H₂ formation was seen to decrease proportionally as the potential reached approximately −1.5 V [Fig. 4(b)]. Interestingly, in the −0.2 V/−1.0 V region the current showed almost a plateau [Fig. 4(a)] reflecting a slowdown of H₂ formation [Fig. 4(b)].

FIG. 4.

Current density/potential (a) and product selectivity/potential profiles (b) for composite Cu₂O–CuBr (N-ED 15) performing CO₂ electrolysis in 1 M KOH solution. These data were obtained with a Sigracet/GDL as support (geometric area = 4.52 cm²) and with a CO₂ flow rate of 1 mL/min.

Other relevant findings obtained from the on-line GC detection of products during CO₂ electrolysis are as follows:

1. (i) When the carrier gas was nitrogen, the only product detected was hydrogen independently of the gas flow rate; this finding is as expected because there is no CO₂ in contact with the working electrode and therefore no formation of hydrocarbons is possible.

2. (ii) Most efficient electrochemical CO₂ reduction to ethylene was found in potassium hydroxide as electrolyte and at a concentration of 1 M or higher. Clearly, alkaline media inhibited/delayed hydrogen formation at the cathode. Other electrolytes, such as NaHCO₃ and Na₂SO₄, were found to be less efficient.

3. (iii) Potentials more negative than −0.5 V versus RHE were required to obtain dominance of ethylene among the hydrocarbon products reaching the on-line GC instrument. Other contributor products were: CH₄, C₂H₆, and C₃H₆. All composite Cu₂O–CuBr films also generated H₂ gas and the dominance of ethylene versus hydrogen depended on the formulation of the composite film and electrolyte conditions (as elaborated below).

4. (iv) Nitrate-derived (“N”) films were more electrocatalytic for CO₂ electroreduction than “L9” films when both were compared in NaHCO₃ electrolyte at room temperature. This trend underlined the importance of the CuBr component in the composite film (see, Table I).

Figure 5(a) compares the product selectivity, as established by on-line GC detection, during 120 min of CO₂ electrolysis at room temperature and −1.9 V (versus RHE) using three types of films: Cu₂O (L9-ED30), Cu₂O–CuBr (L9-ED20), and Cu₂O–CuBr (N-ED15). Although the best current performance (∼60 mA/cm²), was found for a Cu₂O film electrodeposited from a solution containing lactate at pH 9, the electrochemical reaction generates H₂ at 89% mass efficiency and 11% of ethylene [Fig. 5(a)]. When the electrodeposition bath for Cu₂O was dosed with bromide in the lactate electrolytic bath, the resulting L9-composite film was found to perform with lower current density (∼36 mA/cm²) and lower formation of hydrocarbons. Finally, the N-hybrid films were found to generate up to 19% mass of hydrocarbons and 81% mass of H₂ thus showing improved performance with respect to the Cu₂O–CuBr (L9-ED20) film.

FIG. 5.

Comparison of performance for (a) product selectivity at room temperature, (b) respective hydrocarbon generation and (c) product selectivity at different temperature of three composite Cu₂O–CuBr films (L9, L7, and N) performing CO₂ electroreduction at −1.9 V (versus. RHE) in 1 M KOH electrolyte at room temperature and at 3 °C. Average current densities for selected films during 120 min of electrolysis are provided in Fig. 5(a). The CO₂ flow was 1 mL/min in all cases.

Likewise, Fig. 5(b) shows the hydrocarbon selectivity for the films shown in Fig. 5(a), i.e., leaving aside the amount of hydrogen formed. Interestingly, all films produced ethylene as the major hydrocarbon, but the composite L9 film generated ethane and propylene as minor components while the counterpart film from nitrate showed only ethane as minor component.

A balanced amount of Cu₂O and CuBr in the composite film as well as a lower electrolysis temperature (e.g., 3 °C) further promotes ethylene formation from CO₂ electrolysis; Fig. 5(c) includes relevant data for L7 type film (see, Table I). The maximum amount of ethylene reached ∼68% for the L7 films at the beginning of the electrolysis and with an average value of approximately 51% throughout the 120 min of electrolysis. Moreover, the average C₂H₄ production increased for L7 films from 51 to 83% when the electrolysis temperature is kept at 3 °C [Fig. 5(c), last bar]. Thus, a comparison of four electrode compositions for selective ethylene formation is shown in Fig. 5(c). The comparison includes three composite Cu₂O–CuBr films of different compositions and a pristine Cu₂O film. Thus, on Cu₂O (L9), Cu₂O–CuBr (L9) and Cu₂O–CuBr (N) cathode films, the major gaseous product was hydrogen. However, for the new hybrid Cu₂O–CuBr (L7) film, the main product was ethylene which was formed at a current density of ∼53 mA/cm².

C. Durability over longer electrolysis times

The performance and durability of a GDL/Cu₂O–CuBr (L7) cathode was analyzed by emulating durability tests developed for fuel cells. With this aim, a high CO₂ flow rate (8 mL/min) through the channels at the back of the GDL was used during continuous CO₂ electrolysis for 10 h at −1.9 V versus RHE in 1M KOH electrolyte cooled to 3 °C. The electrolyte flow was kept at 1 mL/min and product formation was sustained remarkably well (Fig. S6). Nevertheless, a slow decrease in electroactivity was observed during this 10 h period of CO₂ electrolysis (Fig. S6).

D. Pulsed electrochemical activation (PECA)

On the thesis that the slow deactivation seen for the cathode (Fig. S6) stems from progressive conversion of the oxide phase to the metallic state, a PECA protocol was developed based on anodic regenerative pulses. The pulse routine (Fig. 6) consisted of rectangular pulses applied periodically (every y hours, with y = 1 to 24) to recover the cathodic current during CO₂ electrolysis. The time period (y) in between application of the pulses must also be optimized, as it is necessary to evaluate if it is better to wait long times (i.e., 24 h) and then to do many pulses or to decrease the number of pulses (applied for x min) and to do them more frequently (every hour for instance). The pulse duration and symmetry need to be adjusted as well (from 0.3 to 20 s range). For PECA, the MCFCR was directly connected to a custom-built power supply (instead of a potentiostat) in a two-electrode configuration; i.e., the reference electrode was omitted.

FIG. 6.

Scheme of the rectangular pulses to be applied periodically to the cathode for re-activation of cathode electroactivity during CO₂ electrolysis. Low and high voltage limits for the PECA are defined as V _L and V _H respectively.

Based on the observation that the highest decrease in current occurred during the first 1–2 h of the process, the PECA routine was applied every hour (instead of after several hours as schematized in Fig. 6) and using 4 pulses with same negative voltage (V _L = −5.5 V) and different positive V _H (= 1.5, 2.0, 2.5, and 3.0 V) respectively. The results are contained in Fig. S7. Four identical pulses were applied with symmetrical waveforms (10 s at V _H and other 10 s at V _L). The −5.5 V/1.5 V routine yielded the most satisfactory results in terms of maintaining the film electroactivity for CO₂ reduction to ethylene.

Finally, Fig. 7 shows the stability of current density from a composite Cu₂O–CuBr (L7-ED30) film subjected to PECA protocol as well as the GC analysis of gases at the back of the GDL cathode. Figure 7(a) shows the recorded current/time profile which includes the current transients due to the PECA protocol performed every hour. Importantly, the current density was quite steady because of the PECA protocol and the mass% of ethylene was dominant and stable during the 10 h electrolysis [Fig. 7(b)].

FIG. 7.

Current/time (a) and gas products formed (b) during CO₂ electrolysis at 3 °C with a Cu₂O–CuBr (L7-ED30) as cathode. The PECA routine was −5.5 V/1.5 V with four (10 s/10 s) pulses.