Effects of Substrate and Polymer Encapsulation on CO2 Electroreduction by Immobilized Indium(III) Protoporphyrin.

Heterogenization of molecular catalysts for CO2 electroreduction has attracted significant research activity, due to the combined advantages of homogeneous and heterogeneous catalysts. In this work, we demonstrate the strong influence of the nature of the substrate on the selectivity and reactivity of electrocatalytic CO2 reduction, as well as on the stability of the studied immobilized indium(III) protoporphyrin IX, for electrosynthesis of formic acid. Additionally, we investigate strategies to improve the CO2 reduction by tuning the chemical functionality of the substrate surface by means of electrochemical and plasma treatment and by catalyst encapsulation in polymer membranes. We point out several underlying factors that affect the performance of electrocatalytic CO2 reduction. The insights gained here allow one to optimize heterogenized molecular systems for enhanced CO2 electroreduction without modification of the catalyst itself.

Introduction

The electrocatalytic reduction of carbon dioxide (CO2RR) is a potentially efficacious strategy to tackle global energy concerns, particularly to close the carbon cycle and store renewable electrical energy in chemicals or fuels.[1] The latter is highly desired due to the intermittent character of renewable energy production. The last few decades have experienced the discovery and development of various electrocatalysts, which all lead to a diversity of products with different selectivities and activities.[2−4] In addition to heterogeneous CO2 electrocatalysis using metal, metal alloy, or metal-derived nanostructured electrocatalysts, molecular catalysis of CO2 has shown interesting properties and has undergone a striking development over the years.[5−8] Molecular catalysts are generally considered to yield high selectivity and activity and can be designed in such a way as to mimic enzymes used in nature to efficiently catalyze specific electrochemical reactions such as hydrogen evolution, water oxidation, carbon dioxide reduction, and oxygen reduction.[6,9] Although their stability and solubility in aqueous electrolytes and their poor robustness are often drawbacks, molecular catalysts are widely used to decipher mechanistic insight due to their well-known molecular structure and efficiency. Hence, many studies have been performed on intrinsic catalyst properties such as the influence of the metal center[10−12] and ligands.[13−17] The focus herein will be on metalloporphyrins, a subgroup of molecular catalysts extensively used for CO2RR research.[18,19] In previous work, we reported that cobalt(II) and indium(III) protoporphyrin IX (InPP) immobilized on pyrolytic graphite exhibit high selectivity toward carbon monoxide and formic acid, respectively.[12,20]

The usually poor solubility of molecular catalysts in aqueous media and need for large amounts of catalyst related to homogeneous molecular catalysts can be overcome by heterogenization. The molecular catalyst is immobilized on a conductive electrode, which we will refer to as “the substrate” in the remainder of this article. An additional advantage of immobilization of the catalyst is the more facile product separation, if the catalyst is utilized in a large-scale industrial process. Carbon materials are often employed as substrates owing to their relatively low cost, robustness, and inert nature toward many (electro)chemical reactions. Examples are pyrolytic graphite, glassy carbon, and more recently boron-doped diamond, carbon nanotubes, and graphene.[21−23] Carbon materials exhibit a rich surface chemistry, and their surface functionalization has been proved to play an important role in their electrochemistry.[24] Studies by Morozan et al. and Rigsby et al. have demonstrated the important influence of the carbon support of porphyrins on the selectivity of the oxygen reduction reaction.[22,25] Magdesieva et al. reported on activated carbon supports with different pore sizes for CO2RR on various porphyrin and phthalocyanine complexes, leading to different selectivities.[26] However, little systematic or comparative work has been performed hitherto, to elucidate the intrinsic influence of the substrate or surface functionalization. The choice of a specific substrate is often solely based on empirical considerations, which is not necessarily the optimal system.

Modification of electrodes started in the early 1980s with several methods such as chemisorption or covalent attachment of species on the electrode surface, encapsulation of species in polymer films, and electropolymerization of monomers directly on the electrode.[27,28] More recently, Yaghi and co-workers developed a covalent organic framework of porphyrin building blocks, which showed promising results for the CO2RR.[29] An overview of various catalyst-modified electrodes for CO2RR has been given by Sun et al.[30] From heterogeneous electrocatalysis of the CO2RR, it is known that the electrode morphology and (sub)surface structure significantly influence the activity and selectivity.[31−33] Moreover, the use of polymers has been shown to enhance CO2RR efficiency on cobalt phthalocyanines.[34−36] In the field of heterogenized molecular catalysis of the CO2RR, the effects of such substrate modifications are still unexplored. The importance of the chemical functionality on the adsorption and reactivity, as extensively discussed by McCreery,[24] is the inspiration of the idea that substrate modification may affect the reactivity, selectivity, and stability of the CO2RR. In a recent review, the importance of so-called secondary phenomena in molecular electrocatalysis has been highlighted.[37] It would be very attractive to be able to tailor the surface chemistry in such a way as to enhance CO2RR performance.

In this study we focus on extrinsic properties of the molecular catalyst, particularly related to the immobilization of the molecular catalyst. This work is a step toward a systematic investigation of the chemical functionality of carbon substrates and the chemical environment for heterogenized molecular catalysts. Herein, we study InPP immobilized on different carbon materials, basal-plane pyrolytic graphite (PG), glassy carbon (GC), and boron-doped diamond (BDD), and evince the important role of the substrate, its pretreatment, and the use of polymer membranes for immobilization in the CO2RR performance. The current work demonstrates the improvement of CO2RR catalysis on heterogenized indium(III) protoporphyrin IX by modifying the substrate, its chemical functionality, and chemical environment of the catalyst. The findings can function as a first step in trying to improve other heterogenized molecular systems as well.

Experimental Section

The electrochemical experiments were carried out with a potentiostat (IviumStat or CompactStat, Ivium Technologies), in a conventional three-electrode cell, where the working electrode (WE) and counter electrode (CE) compartments were separated by a Nafion membrane (Nafion 115). Basal-plane pyrolytic graphite, glassy carbon, and boron-doped diamond disks (5, 5, and 10 mm diameters, respectively) were used as WEs. The counter electrode and reference electrode (RE) were a platinum wire and a reversible hydrogen electrode, respectively. For correct measurements versus the RHE scale, the Luggin capillary and the RHE compartment were filled with CO2-saturated electrolyte before CO2 reduction. The electrolyte is 0.1 M phosphate buffer of pH 9.6 ± 0.1, prepared with K2HPO4, K3PO4, and ultrapure water (Millipore Milli-Q gradient A10 system, 18.2 MΩ cm). The choice for this pH was based on the enhanced HCOOH selectivity observed previously.[12] The reported current densities were always normalized by the geometric surface area of the WE, and in some cases additional normalization by the amount of active species was performed for correct comparison of the activity. The potentials were corrected for ohmic drop by the potentiostat during measurement. Generally, potentiostatic bulk electrolysis was performed at E = −1.5 V vs RHE for 90 min, with manual collection of 100 μL samples at certain times, and analyzed by high-performance liquid chromatography. The reported concentrations of liquid products, or subsequently calculated Faradaic efficiencies (FEs), are an average of three to five independent experiments with freshly prepared electrodes. Additionally, the data points for each experiment were obtained by the average of three injections of the same sample. The dominant contribution of the uncertainty in concentration/FE resulted from the different experiments. Additional experimental details can be found in the Supporting Information.

Results and Discussion

Substrate Effect

We compare the selectivity and activity for CO2RR on InPP immobilized on PG, GC, and BDD substrates. The immobilization procedure and amount of InPP dropcasted per cm2 were kept the same for all the substrates (details in the Supporting Information). From the Faradaic efficiency toward HCOOH, given in Figure a, it is observed that the substrate has a significant influence on the selectivity of the CO2RR on immobilized InPP. The PG substrate is the most selective toward HCOOH and the GC substrate the least selective. These effects cannot be ascribed to the activity of the bare substrate, as shown by control experiments in Figure S17a, which show that bare PG, GC, and BDD are not active for the CO2RR under our experimental conditions. In Figure b–d, the absolute total and partial current densities are shown for the CO2RR on the different substrates. For a correct comparison of the activity, the current density was also normalized by the indium content of the substrate (denoted by j*). As will be discussed in detail later, the electroactive coverages of InPP are not the same for the different substrates. It can be seen that there is a 1 order of magnitude difference in jtotal* and jHCOOH* on PG in comparison to GC and BDD. Note that this difference is not associated with a difference in electrochemical active surface area (ECSA), as shown in Figure S8a. The fact that GC and BDD both perform worse in comparison to PG indicates that the enhancement in CO2RR selectivity and activity cannot be ascribed to the high content of either sp2 or sp3 carbon atoms present in GC and BDD, respectively.

Figure 1

(a) Faradaic efficiency toward HCOOH, (b) absolute total current density, (c) absolute partial current density for HCOOH, and (d) absolute partial current density for H2 during CO2 reduction on immobilized InPP on different substrates in 0.1 M phosphate buffer of pH 9.6. Lines are given to guide the eye.

The results in Figure are in line with the online mass spectrometry and online HPLC experiments, depicted in parts a and b of Figure , respectively, from which we confirm that significantly more H2 is produced on InPP-GC and significantly more HCOOH on InPP-PG. The CO2 consumption on InPP-PG is also much higher in comparison to the other substrates, in agreement with higher jtotal* and the higher HCOOH production rate observed on PG in comparison to the other substrates. Although it is difficult to quantify, the onset potential for H2 and HCOOH on InPP-BDD is at a more negative potential in comparison to PG and GC, which is a general characteristic of BDD. Moreover, in Figure b we can observe differences in the peak potentials of HCOOH formation between the different substrates, which is related to the competition of the hydrogen evolution reaction (HER).

Figure 2

(a) Online electrochemical mass spectrometry and (b) online HPLC during CO2 reduction on immobilized InPP on different substrates in 0.1 M phosphate buffer of pH 9.6.

In addition to a change in selectivity and activity, there is a clear difference in stability between the substrates, since a slight decrease is observed in jHCOOH* as a function of time on PG and GC, accompanied by an increase in jH*. We define the relative FE with respect to the initial value (eq ) as a measure for the stability of the system. From a graph of this relative FE versus time (Figure a), we can compare experiments with different values of FE. A more horizontal trend indicates a higher stability as the FE does not decrease significantly in time.

Figure 3

(a) Stability of immobilized InPP on different substrates (lines are given to guide the eye), (b) XPS spectra of InPP-PG electrodes before electrolysis and after electrolysis (t = 10 min and t = 1 h) with the indium content (atom %) given in the inset, (c) Faradaic efficiency toward HCOOH (left axis) and partial current density for HCOOH (right axis) as a function of the amount InPP drop-casted on the different substrates (lines given to guide the eye), and (d) indium content of the InPP immobilized on the substrates, as estimated from XPS, as a function of the amount InPP drop-casted (lines given to guide the eye).

The tendency of the FE to decrease with time has been observed before and was associated with the detachment of InPP from the surface or deactivation of the porphyrin.[12] We have performed X-ray photoelectron spectroscopy on InPP-PG before and after electrolysis, as shown in Figure b. We observed that the indium content on the electrode, which is a measure of the actual amount of InPP adsorbed on PG, is substantially decreased after 10 min, with a negligible further decrease after 1 h, which is in agreement with the decreasing trend of FE as a function of time. For our work, the charge of redox peaks related to InPP is not an accurate measure to obtain quantitative information about the amount of InPP on the substrate, and we rely on XPS for the quantification of the indium content on the substrate. Additionally, we performed experiments under homogeneous conditions of InPP (Figure S4), in which we do not observe a decrease in selectivity with time. Therefore, we confirm that the destabilization of immobilized indium protoporphyrin is related to detachment from the surface. Moreover, the detachment from the surface takes place in the first 10 min of electrolysis. The adsorption of the porphyrin on the substrate is through noncovalent π–π interactions,[30,38,39] which is believed to be important for enhanced CO2RR performance and strongly dependent on the carbon substrate.

We believe that the substrate morphology plays an important role, as evidenced by the higher CO2 consumption observed with OLEMS, which can be explained by efficient mass transport effects due to the more porous structure of PG in comparison to GC or BDD (Figure S8a), which may also suppress the HER. These observations are similar to mass transport effects recently reported for heterogeneous electrocatalysts, showing that the mesostructure can affect the selectivity and activity of the CO2RR.[40,41] Additionally, a typical characteristic of GC is its poor permeability for gases, which likely affects the mass transport.[42] X-ray diffractometry and Raman spectroscopy of our substrates (Figures S9a and S10) indicate typical characteristics in agreement with the literature. PG and BDD exhibit sharp XRD peaks, and GC exhibits weak and broad peaks, indicating a high degree of crystallinity for PG and BDD and a somewhat amorphous structure for GC.[43] The Raman spectra indicate the characteristic D and G bands for sp2 carbons in PG and a specific peak associated with sp3 carbon in BDD. The Raman bands for GC are weaker and less sharp, which indicate disorder of the graphite lattice for GC.[44,45] The low activity and selectivity on GC may be associated with its poor crystallinity, as a high crystallinity implies enhanced charge transport, as observed in covalent organic frameworks.[29,46] Recently, crystallinity has been shown to play an important role in the selectivity of CO2RR on copper phthalocyanine catalysts.[47] However, high crystallinity alone is not sufficient for improved CO2RR catalysis, on the basis of the significant differences in CO2RR performance between PG and BDD, which are both highly crystalline.

Apart from the substrate morphology and surface structure, a plausible explanation for the enhanced selectivity and activity on PG may be related to a difference in the effective amount of the porphyrin on the three substrates. As can be seen in Figure c, a difference in the amount of immobilized InPP on the substrate leads to a difference in activity and selectivity (details of the corresponding experiments are provided in Figures S1–S3). In the case of PG this concentration effect is more pronounced and in the case of GC is almost completely absent, which can be interpreted as a “InPP saturation limit” that is reached earlier for GC in comparison with PG. In other words, the PG substrate can accommodate more catalyst in comparison to GC and BDD. These conclusions are in agreement with quantitative information obtained from XPS spectra of InPP immobilized on the different substrates (Figure S12). We also varied the amount of InPP drop-casted on the different substrates and plotted the indium content as estimated by XPS vs the amount drop-casted in Figure d. The vertical dashed lines indicate the InPP amounts used for the qualitative study shown in Figure c and Figures S1–S3. It can be seen that the indium content is the highest for PG and the lowest for GC, which is in agreement with our conclusion that a PG substrate can accommodate more catalyst in comparison to GC. Furthermore, we used the indium content of InPP immobilized on different substrates (0.65, .0.49, and 0.58 atom % In on PG, GC, and BDD, respectively) to normalize the activity to the real amount of adsorbed InPP (Figure b–d). It is noteworthy that we assumed that each immobilized InPP molecule contributes to the electrocatalytic activity, and since the difference in indium content between the substrates is small (Figure d), we obtain a reasonably accurate comparison of the activities between the substrates.

Another cause of the improved CO2RR selectivity may be a difference in HER activity between the substrates instead of solely an intrinsic substrate effect specifically related to the CO2RR. In Figure a, a comparison between the substrates for the HER and CO2RR is shown. The high overpotential for the HER on the BDD substrate does not favor the CO2RR, since the CO2RR onset potential is shifted toward more negative potentials, and the current is suppressed on InPP-BDD with CO2 in solution. The HPLC results (Figure b) confirm the formation of HCOOH at more negative potentials on BDD in comparison to GC and PG. Under argon, where only hydrogen evolution takes place, the current density on InPP-GC is smaller in comparison to InPP-PG. However, as shown by OLEMS measurements in Figure b, more H2 is produced on GC. Moreover, in the presence of CO2, the onset potential is shifted positively for InPP-PG but is almost unchanged for InPP-GC, which indicates a more efficient catalysis of CO2RR on PG with respect to GC, in agreement with the online experiments in Figure . Consequently, the GC substrate is more active toward the HER in comparison to PG. These results indicate that the competition between CO2RR and HER strongly depends on the nature of the substrate, which in turn affects HCOOH selectivity.

Figure 4

(a) Linear sweep voltammetry on immobilized InPP on different substrates in 0.1 M phosphate buffer of pH 9.6 under argon and CO2 atmospheres (scan rate 20 mV s–1) and (b) Online electrochemical mass spectrometry measurement of hydrogen evolution on InPP immobilized on different substrates.

In order to increase the impact and generality of our findings, we studied the substrate effect on protoporphyrins with Rh and Sn metal centers, which previously were found to produce HCOOH.[12] Although the HCOOH selectivity for these porphyrins is much lower than for InPP (which is an intrinsic catalytic effect), a trend can be observed for the different substrates similar to that found for InPP (Figure S5). Pyrolytic graphite substrate leads to the highest FE, while the glassy-carbon substrate is the least selective toward HCOOH.

The above results demonstrate the important influence of the substrate, which is believed to be the result of an interplay of several factors influencing the selectivity and reactivity of CO2RR, such as morphology/mesostructure (and thereby optimized mass transport effects), crystallinity, electrostatic interaction with the molecular catalyst, and activity for the HER.

Effect of Substrate Pretreatment

In addition to the nature of the substrate, modification or treatment of the substrate surface offers a means to influence the CO2 selectivity, reactivity, and stability. We investigated the influence of a cathodic and anodic electrochemical pretreatment (“Cat-PG” and “An-PG”) and a H2 and O2 plasma treatment (“H2-PG” and “O2-PG”) of the PG substrate. In Figure , it is shown that O2 plasma treatment increases the FE and jHCOOH values. These effects can be attributed to a change in chemical functionality as discussed later, instead of an increased surface area as evidenced by similar Cdl values before and after O2 plasma treatment (Figure S8b). Since the difference in amount of InPP immobilized on the three substrates is small (Figure d), we assume neglibible differences in indium content on the various pretreated PG surfaces, and a difference in surface roughness of the pretreated PG electrodes would have a larger influence on the measured current, which is taken into account in our discussion. Furthermore, the exposure time of O2 plasma seems to play a role, as shown in Figure S6, since a mild O2 plasma treatment (3 and 6 min exposure) slightly improves the CO2RR, while a harsh O2 plasma treatment (12 min exposure) worsens CO2RR performance. In Figure e, it is shown that O2 plasma treatment has negligible influence on the stability. Anodization of PG leads to lower initial FE but improves the stability dramatically. Moreover, both hydrogen treatments, cathodic and H2 plasma, decrease the selectivity, reactivity, and stability of CO2RR significantly. Since H2-PG has a lower Cdl value and thus a smaller surface area (Figure S8b), but a higher jtotal value, the observed changes in selectivity and reactivity do not result from a change in surface area of PG after H2 plasma treatment.

Figure 5

(a) Faradaic efficiency toward HCOOH, (b) absolute total current density, (c) absolute partial current density for HCOOH, (d) absolute partial current density for H2, and (e) stability during CO2 reduction on immobilized InPP on different pretreated PG substrates. Electrolyte: 0.1 M phosphate buffer with pH 9.6. Lines are given to guide the eye.

The observed differences for the investigated pretreated PG substrates highlight the role of oxygenated and hydrogenated functional groups of the substrate’s surface on the immobilization of InPP and subsequently on the CO2RR. Our results imply that hydrogen functional groups on the PG surface strongly decrease CO2RR selectivity, reactivity and stability and oxygen functionalities increase CO2RR performance. H2 plasma consists of a large amount of H atoms, which react with surface oxides, leading to C–H bonds and a decreased O/C ratio on the surface.[24,48] The chemisorbed hydrogen on the surface promotes the HER with respect to the CO2RR, which is reflected in a decrease in FE and jHCOOH and increase in jH. On the other hand, O2 plasma treatment increases the amount of oxygen functional groups (e.g., hydroxyl, carbonyl, carboxylate) on the surface, which generally leads to a more polar and hydrophilic surface and stronger adsorbate–substrate interaction. O2 plasma treated PG shows an improvement in CO2RR selectivity. A similar influence of O2 plasma treatment of the substrate has been reported before for the oxygen reduction reaction.[49]

On anodically treated PG we expected behavior similar to that after O2 plasma treatment, due to the introduction of oxygenated species, but no improvement in activity or selectivity has been observed. However, a remarkable increase in the stability is observed. On the basis of blank voltammograms of the pretreated PG (Figure S7), there is a significant difference in An-PG in comparison to the untreated and other pretreated PG electrodes. Application of strong anodic pretreatment causes a destruction of the carbon surface and formation of a thick graphite oxide layer, which contains a high amount of anionic sites and interior void volume, leading to high surface area, in agreement with the dramatic increase in Cdl for An-PG in Figure S8b.[24,50] Complementary information was obtained by X-ray diffractometry, Raman spectroscopy, and X-ray photoelectron spectroscopy. As shown in Figure S9b, the crystalline nature of PG remains intact upon electrochemical pretreatment. However, anodization of PG leads to a significant increase in the Raman peak at ∼1352 cm–1 and a slight increase in the D′ peak around 1620 cm–1 (Figure S10b). The peak at 1352 cm–1 is attributed to the presence of graphitic edges, which increases in intensity upon anodic treatment as observed in our spectra, and the peak at 1620 cm–1 indicates delamination of graphitic planes. Formation of graphite oxide increases the strain on the PG lattice, leading to fracturization.[51] In Figure S11 the ratio of the D and G band intensities for the different substrates and pretreated PG is depicted, which clearly shows a high ID/IG ratio in the case of An-PG. Anodization of PG leads to an increase in the edge plane density. Note that GC also exhibits a relatively high ID/IG ratio but showed a poor stability as seen before. Therefore, we believe a high edge plane density in combination with high crystallinity to be the reason for the strongly improved stability. The XPS spectra of the plasma and electrochemically treated pyrolytic graphite electrodes, shown in Figure S13, provide a quantitative basis for our conclusions about chemical functionality on CO2RR performance. The oxygen content is significantly higher on An-PG and O2-PG in comparison to untreated PG. Moreover, the H2 plasma and cathodically treated PG exhibit fewer surface oxygenated species in comparison to untreated PG. These results are in agreement with our interpretation of the CO2RR results and our Raman spectroscopy experiments.

Effect of Polymer Encapsulation

In addition to pretreating the substrate, another strategy aimed at improving CO2RR selectivity, reactivity, and stability is incorporation of the porphyrin in polymeric matrices. In this work, we compare the influences of didodecyldimethylammonium bromide (DDAB), Nafion, poly(4-vinylpyridine) (P4VP), and poly(3,4-ethylenedioxythiophene) polystyrenesulfonate (PEDOT:PSS). Details about the immobilization in these polymeric matrices are given in the Supporting Information. The concentration of InPP in the polymer films, and the amount of polymer drop-casted per cm2 surface were always kept the same. As depicted in Figure , interesting differences are observed between the polymer membranes. In comparison with polymer-free InPP (InPP-PG), encapsulation in DDAB, P4VP, and PEDOT:PSS shows enhanced selectivity and activity, whereas Nafion negatively affects the selectivity and activity. From Figure d it can be seen that the HER activity with P4VP is drastically decreased. A striking observation in Figure e is the enhanced stability when P4VP or PEDOT:PSS is used in comparison to the other polymers. These results underscore the importance of the nature of the polymer, as seen before for a modified bulk silver electrode.[52] Enhanced CO2RR performance by P4VP has been reported before for cobalt phthalocyanine, where the authors explain the observed effects by the presence of pyridine residues in the polymer, which influences the coordination to the catalyst.[35,36] Although Nafion-coated catalysts are widely used for various electrocatalytic systems, negative effects of Nafion on the catalytic activity have been observed previously.[53,54]

Figure 6

(a) Faradaic efficiency toward HCOOH, (b) absolute total current density, (c) absolute partial current density for HCOOH, (d) absolute partial current density for H2, and (e) stability during CO2 reduction on immobilized InPP in different polymer membranes. Electrolyte: 0.1 M phosphate buffer with pH 9.6. Lines are given to guide the eye.

Differences in polymer-dispersed catalysts are generally associated with the introduction of a hydrophobic environment, leading to the suppression of the HER.[28,34,55] However, the differences in CO2RR performance in our investigation are likely to be explained by the different chemical structures of the polymers (Figure S14), leading to different substrate/adsorbate interactions. As can be seen in the blank CVs in Figure S15, the typical InPP peaks are masked or shifted for Nafion, P4VP, and PEDOT:PSS. Changes in the voltammetric behavior after immobilization of porphyrins in DDAB have been reported before, in which the authors also suggest the possibility of porphyrin dimer formation in DDAB vesicles.[55] The absence of InPP redox peaks in the investigated potential window indicates a change in electrochemical behavior but does not rule out the presence of InPP on the surface, since reasonable amounts of HCOOH are produced. The CV of PEDOT:PSS coated PG shows a much larger double layer, which indicates the strong influence of the polymer on the chemical environment near the electrode surface. The polymers without InPP exhibit reasonable CO2RR activity, as can be seen in Figure S17b, where the same trend is observed as for InPP encapsulated in the different polymers. A detailed investigation of the electrocatalytic activity of polymers for CO2RR is beyond the aim of this work. Nonetheless, it can be concluded that the observed effects are partially due to the activity of the polymer for CO2RR, which is higher with P4VP and PEDOT:PSS. The pyridine group in P4VP and aromatic moieties in PEDOT:PSS are assumed to play an important role in this respect, since Dunwell et al. very recently reported CO2RR toward HCOOH mediated by pyridine.[56] The porphyrin exhibits a planar macrocycle with large π conjugation, which facilitates electrostatic interactions with the polymer. We believe that the increased stability is a result of the presence of aromatic building blocks in P4VP and PEDOT:PSS, which facilitates axial coordination to the indium metal center by electron donation. This effect is known for pyridine or imidazole ligands which are used for (covalently) anchoring catalysts to electrodes and are found to stabilize the coordination of CO2.[34] This interaction is absent in the case of Nafion and DDAB, leading to a inferior stability in comparison to the polymer-free InPP, P4VP, and PEDOT:PSS. DDAB shows an increase in selectivity and activity, which is associated with the suppression of the HER similarly to previous work.[52] The poor CO2RR performance with Nafion is tentatively ascribed to the low mobility of the porphyrin in the polymer and a disordered structure of the Nafion layer.[53,54] We performed electrochemical impedance spectroscopy on the different polymer-dispersed catalysts to investigate the kinetics of electron-transfer processes (Figure S16). As discussed in the Supporting Information, the charge-transfer resistance between the polymer films is in agreement with the observed activity in Figure b. We conclude that the nature of the polymer affects the rate of electron transfer during the CO2RR rather than mass transport of active species, leading to different activities among the polymer films.

The results in the present work demonstrate that a hydrophobic environment, induced by the polymer membrane in general, does not always lead to suppression of the HER activity and subsequent improvement of CO2RR. Additional effects related to the chemical structure, allowing for electrostatic interactions with the porphyrin or activity toward CO2RR by the polymer itself, play an important role. Note that the current findings strongly depend on the (molecular) catalyst under study, and one should be careful to generalize the observed polymer effects for other electrocatalytic systems. Nonetheless, we emphasize the possible negative influence of Nafion on CO2RR, as Nafion is very often used to immobilize catalysts or as a binder in the preparation of ink-containing nanoparticulate electrocatalysts.

Conclusion

This work has shown the importance of the nature of the substrate for immobilized indium(III) protoporphyrin IX for CO2RR toward formic acid. For this particular system, a pyrolytic graphite substrate outperforms glassy carbon and boron-doped diamond in terms of CO2RR selectivity and reactivity, while boron-doped diamond shows the best stability. The enhanced activity and selectivity of PG are assigned to a combination of different factors: first, to a more porous surface structure, leading to efficient mass transport, and furthermore, to an optimal interaction between substrate and InPP and a favorably low HER activity.

We have investigated two strategies to improve or alter the selectivity, reactivity, and stability of the CO2RR. Pretreatment of the substrate before catalyst immobilization and immobilization in polymeric matrices have been shown to be practical tools to fine-tune CO2 reduction performance. Hydrogenated functional groups on the surface decrease the selectivity, activity, and stability, while (mild) oxygen functionalization positively influences the CO2 reduction performance. Anodization of the graphite surface substantially increases the stability, which is believed to be related to the thick graphite oxide layer containing high edge plane density. Both P4VP and PEDOT:PSS increase the stability, which is believed to be due to axial coordination of their aromatic moieties to the indium metal center. DDAB, P4VP, and PEDOT:PSS improve the performance of the CO2RR, while Nafion affects the CO2RR negatively. These strategies are assumed to be applicable to similar macrocyclic catalysts immobilized on carbon materials.

It should be noted that the CO2RR performance is often a tradeoff among selectivity, activity, and stability, each of which can be modified by substrate pretreatment and catalyst encapsulation in polymers. Although complete mechanistic insight into the pretreatment and polymer effects is still missing, the results obtained here may help to design heterogenized molecular catalytic systems for CO2 reduction with specifically optimized properties. Other molecular catalysts may behave differently; hence, one should be careful generalizing the results obtained in this study. However, it is very likely that the substrate, its pretreatment, and catalyst dispersion in polymers will have an influence on CO2RR performance, and the insight obtained herein may be used as a starting point for further optimization of the system.