Introduction

Carbon dioxide and plastics are products of industrial activity and intersect with respect to sustainability concerns as well as opportunities for decarbonization. [1][2][3][4] The environmental impact of anthropogenic greenhouse gas emissions has incentivized the development of CO 2 sequestration and conversion technologies. Incorporation of captured CO 2 into commercially useful materials has the potential to generate economic value. Plastics generation currently relies on precursors (e.g., ethylene) derived from petroleum with substantial associated CO 2 emissions (Fig. 1). For example, for each ton of ethylene, the most used monomer in polymer synthesis, 1.6-2 metric tons of carbon dioxide emissions are generated. 5,6 Incorporation of CO 2 into plastics is appealing if it can be done so efficiently from waste CO 2 given the large scale of polymer production. 4,7,8 Several approaches to CO 2 incorporation into polymers have been demonstrated, such as CO 2 copolymerization with epoxides, diols, and with olefinic compounds including dienes and vinyl ethers. [9][10][11][12][13][14][15][16][17][18][19][20] However, the rheological properties of the resulting polymers, such as low melting and glass transition temperatures, limit their use in common applications. 15 Additionally, some catalytic systems suffer from low activity due to the high thermodynamic stability of CO 2 . 21,22 Therefore, other pathways of CO 2 incorporation into polymers are explored. 16,22,23 One such strategy that has been explored recently is coupling electrochemical CO 2 reduction (eCO 2 R) using electricity from sustainable sources [24][25][26] into the process of (co)-monomer production, whereby more common polymerization precursors, such as CO and C 2 H 4 , are generated as intermediates. Tandem electrochemical-thermochemical systems capable of CO 2 conversion to useful products such as of C 3 oxygenates, butane, and polyketones via eCO 2 R have been demonstrated in the past few years. [27][28][29][30] A two-step process capable of converting CO 2 and ethylene to CO-ethylene copolymers -polyketones -has been established, sourcing of one comonomer, CO, from CO 2 by eCO 2 R; the C 2 H 4 incorporated into the monomer, however, was obtained from feedstock petroleum. 29 To establish a coupled catalytic system that eliminates the use of petroleum-derived carbon precursors, we targeted the generation of the comonomers from eCO 2 R . Polyketones are an appealing initial target as both CO and C 2 H 4 can be generated from eCO 2 R. Polyketones have been identified for a wide range of promising applications. 31 For instance, the chemical resistance of polyketones against hydrocarbons exceeds that of many widely used polymers; 32 they also display excellent strength, fatigue and elongation resistance. 33,34 Also, polyketones can be blended with other polymers to create materials with desirable properties. 35,36 Finally, polyketones have demonstrated photodegradation under UV light, which leads to reduced environmental persistence of these plastics. 37,38 Ethylene generation can be accomplished from eCO 2 R by Cu electrocatalysis on gas diffusion electrodes in aqueous ele ctrolytes. 39,40 Such systems display high (~40%) Faradaic efficiency (FE) towards ethylene, while simultaneously producing CO in substantial quantities, when electrolysis is performed on bare Cu surfaces in electrolytes of pH < 8.5 (1 M KHCO 3 ) at a potential of ~ -1 V vs. RHE. [41][42][43][44] Furthermore, CO can also be produced from CO 2 with FE >90% with Ag electrocatalysts. 45 The selectivities, surface properties of these electrocatalysts, their surface oxidation states and local environments under eCO 2 R conditions, as well as reaction mechanisms have been studied previously in different conditions, establishing a theoretical basis for the design of an eCO 2 R system optimized for polyketone production. [46][47][48][49] A two-reactor system, with gaseous products from eCO 2 R being transferred to another reactor for CO-ethylene copolymerization by a homogeneous catalyst, separates the electrolyte mixture of the electrochemical process from a thermochemical organometallic catalyst. However, even in the gas phase, aqueous eCO 2 R results in product streams with impurities such as H 2 and water vapor; the latter is known for reducing the activity of CO-ethylene copolymerization catalysts. 50,51 Furthermore, due to current density limitations of Cu GDE systems (~100-400 mA cm -2 at -1.1 V vs. RHE at electrolyte pH < 8.5), 44 the concentrations of ethylene and CO in the gas product streams of single-pass Cu GDE systems are often low at flow rates above 10 standard cubic centimeters per minute (sccm), leading to low copolymerization activity. Finally, performing eCO 2 R on GDEs at elevated pressures (>30 bar), as typically used for CO-ethylene copolymerization, 52 is challenging due to electrode flooding or gas penetration through the electrode. This occurs when a pressure gradient is established across the GDE during start-up of high-pressure systems, significantly hindering GDE performance. [53][54][55] Thus, performing the copolymerization reaction at or near atmospheric pressure is desirable.

Herein, we demonstrate a tandem catalytic system that achieves the preparation of polyketones entirely from CO 2 by coupling electrochemical CO 2 reduction and thermocatalytic CO-ethylene copolymerization with an overall CO 2 conversion utilization of 14% and current-to-polymer efficiency of 51%.

Results and Discussion

To assess the possibility of CO-ethylene copolymerization using the gaseous product mixture of eCO 2 R, which is not expected to contain high concentrations of CO and ethylene at substantial flow rates, we measured the performance of a Pd copolymerizaton catalyst at reduced concentrations of CO and ethylene, diluted with CO 2 . The catalyst, [(dppp)Pd(CH 3 )(NCCH 3 )][BAr F 24 ] (Pd-PP, Table 1; BAr F 24 is tetrakis[3,5-bis(trifluoromethyl)phenyl] borate), was selected for two reasons. First, Pd-PP is reported to exhibit a high activity towards CO-ethylene copolymerization. 56 Second, Pd-PP has been previously demonstrated to have activity for COethylene copolymerization in the presence of contaminants introduced by CO 2 reduction to CO, such as water vapor at high ethylene partial pressures, or at atmospheric pressure with 0.5 atm each of CO and ethylene. 29 Third, Pd-PP is not expected to react with CO 2 , since CO 2 insertion into Pd-alkyl bonds has only been observed in Pd complexes with Pd-B or Pd-C bonds in the trans position to the Pd-alkyl bonds. 57,58 Fourth, Pd-PP catalyzes the production of perfectly alternating CO-ethylene copolymers at a wide range of CO-ethylene ratios, allowing for the tuning of the eCO 2 R system to produce a mixture containing CO and ethylene with a 1:1 molar ratio to match the reaction stoichiometry and, therefore, maximize CO 2 incorporation. To further increase copolymer yields by improving mass transport, a flow-through glass frit reactor (Fig. S129-S130) was employed instead of a conventional pressurized tank reactor. The former enables efficient interfacial mass transport via a swarm of rising bubbles, increasing the gas-liquid interfacial area and improving substrate mixing in the liquid phase. Using this reactor makes high-pressure operation impossible; however, given that eCO 2 R is performed at atmospheric pressure, high-pressure polymerization downstream of eCO 2 R is undesirable.

Test trials with mixtures of CO and ethylene diluted with CO 2 (Table 1) showed solid product formation even with feed concentrations of CO and ethylene as low as 5% each. In these experiments, the gas mixture was passed through 10 cm 3 of a 1 mM solution of Pd-PP in 1,1,2,2-tetrachloroethane (TCE) at a rate of 20 sccm. After 16 hours, the solution was mixed with methanol, acidified with HCl, and stirred under atmospheric air. Resulting solid products were examined by 13 C and 1 H NMR spectroscopy, dissolved in a mixture of 1,1,1,3,3,3-hexafluoropropan-2-ol (HFIP) and C 6 D 6 with a 4:1 volume ratio. Spectra obtained were consistent with a perfectly alternating CO-ethylene copolymer (Fig. S1-S81). 29,56 Specifically, peaks are observed in the 13 C NMR spectra at 213 and 35.6 ppm, corresponding to the carbonyl and methylene groups, respectively. A major CH 2 signal is observed at 2.49 ppm, in addition to minor signals assigned to CH 3 and CH 3 CH 2 chain ends. These results suggest that polyke-

Table 1. Polymerization of CO and C2H4 with reagent gas dilution. ‡ CO + C 2 H 4 25°C, 1 atm 1,1,2,2-TCE C O C H 2 tones can be produced at reactant concentrations low enough to be applicable to coupling CO-ethylene copolymerization using Pd-PP and eCO 2 R product mixtures. Moreover, the use of a glass-frit reactor affords a three-fold increase in copolymerization activity at atmospheric pressure compared to a tank reactor. 29 This increase in activity is explained by improved mass transport, which is especially desirable when the concentration of reagents in the feed is low.

Next, the performance of the eCO 2 R system was assessed for its suitability for tandem operation. To determine the concentrations of CO and C 2 H 4 that can be practically obtained by eCO 2 R, as well as to find the FE distribution of the Cu GDE system, single pass eCO 2 R experiments were performed using a pure CO 2 feed and a 1 cm 2 Cu GDE in a standard, 1 cm 2 GDE cell (Fig. S114).

In these experiments, eCO 2 R was performed at a constant current density of -170 mA cm -2 in aqueous 1 M KHCO 3 . This achieved a cathodic potential close to -1.1 V vs. RHE, a potential that produces the highest FE towards ethylene on Cu. 39 The resulting FE distribution (Fig. 2a) largely matched published data; 41,42,59 however, the concentrations of CO and C 2 H 4 in the product stream were below 1 mol % each throughout the experiment (Fig. 2b). These concentrations are significantly lower than the lowest concentrations where measurable CO-ethylene copolymer could be obtained under the conditions tested (Table 1). Furthermore, the performance of the Cu GDE after 60 minutes of electrolysis toward CO and ethylene production decreased, which can be attributed to electrode flooding and salt deposition. 60,61 As an additional limitation, after 60-70 minutes of electrolysis, significant deposition of water condensate was observed on the exhaust tubing of and inside the GDE cell. This observation indicates that by that time significant amounts of water were incorporated into the product gas, which could poison a Pd-PP catalyst in a downstream reactor. 50,51 To maximize CO and ethylene production, minimize the introduction of unwanted compounds into the product stream, and obtain a system that returns reproducible results, limiting the electrolysis time to 60 minutes was considered optimal for this GDE setup. Since the polymerization time scale, especially at low CO and ethylene concentrations, is greater than 60 minutes, CO 2 reduction and CO-ethylene copolymerization units could not be placed in series directly. Hence, a gas recirculation loop was incorporated to generate a gas mixture with higher CO and ethylene concentrations, as well as to separate CO 2 reduction and polymerization processes in space and time. To model the product distribution in recirculatory eCO 2 R system, we placed two 1 cmfoot_1 Cu GDEs in series in a twin GDE cell (Fig. S115) and performed the same single pass experiment with a pure CO 2 feed. The experimental conditions remained the same as before: -170 mA cm -2 , aqueous 1 M KHCO 3 electrolyte. We found that the concentrations of CO and ethylene in the product stream increased roughly twofold (Fig. 2b). However, the obtained concentration of CO was lower than the concentration of C 2 H 4 , which we attributed to the ability of CO to undergo reduction further, 62,63 while C 2 H 4 does not undergo further reduction on Cu. 47 This conclusion is further supported by the fact that in the two-electrode system, the average FE towards C 2 H 4 increased, while the FE towards CO formation decreased, compared to the one-electron cell (Fig. 2a). Therefore, additional tuning of the electrochemical process to generate supplementary CO was necessary to reach a 1:1 molar ratio of CO to ethylene, necessary for maximum CO 2 conversion to polymer and current-topolyketone efficiency, since Pd-PP only produces perfectly alternating polyketones under various ethylene-CO ratios, and repeated insertions of ethylene are strongly disfavored. 29,64 To this end, Ag GDEs were used, which can furnish a high FE (~90%) towards CO in eCO 2 R. 45 Thus, a small 0.25 cm 2 Ag GDE cell (Fig. S113) was installed in series with the twin Cu GDE cell, and the same experiment was carried out, applying a -160 mA cm -2 current density to the Ag GDE (operating at a cathodic potential close to -1.1 V vs. RHE on the Ag GDE). As expected, the concentration of CO in the effluent stream increased substantially, but the ethylene concentration remained the same (Fig. 2c); the average FE towards CO also increased (Fig. 2d). The single-pass CO concentration in the effluent gas was greater than that of ethylene, which was deemed acceptable because recirculatory eCO 2 R using this layout is expected to result in some CO consumption by Cu GDEs. 62,63,65 The production of CO and ethylene in recirculatory eCO 2 R was then evaluated using the two-loop recirculation apparatus (Fig. 3a, S132-S133), which featured a 100 cm 3 gas tank evacuated and charged with CO 2 to a pressure of 40 psig at the beginning of each experiment, an MFC to control the gas flow rate, and a peristaltic pump to move the eCO 2 R product gas into the highpressure tank. 60-minute recirculatory eCO 2 R experiments were performed using a twin (2×1 cm 2 ) Cu GDE cell, as well as a chain of one twin (2×1 cm 2 ) Cu and one 0.25 cm 2 Ag GDE cells. All experimental conditions were the same. When only Cu GDEs were used, the product gas at the end of the experiment contained 3.1% CO and 5.5% ethylene (Fig. 3b), significantly exceeding the results of single-pass eCO 2 R. However, as expected, a CO deficit was observed, supporting the need for a Ag GDE to generate more CO. When a combination of Cu and Ag GDE cells was used, the concentration of ethylene slightly increased to 6.5%, while the CO concentration increased to 8.9%. (Fig. 3b) Reinstalling new GDEs, recharging the gas tank with CO 2 , and repeating recirculatory eCO 2 R for another 60 minutes raised the CO and C 2 H 4 concentrations to 14% and 11%, respectively (Fig. 3b). A decline in tank pressure from 40 to 16 psig observed during these experiments was attributed to CO 2 crossover, which has been reported. 66,67 Such a significant decline cannot be explained by the reduction in the molar quantity of gas due to C-C coupling in course of eCO 2 R on Cu or a leak; in absence of applied current, the tank pressure dropped by 1 psig after 60 minutes of recirculation. Having found the optimal eCO 2 R system configuration and assessed the performance of Pd-PP with diluted reagent feeds, the glass frit polymerization reactor was combined with the recirculatory eCO 2 R system in a two-loop recirculation system (Fig. 3a, S132-S133). In this system, the gas from the tank could be pumped through the GDE cells or the reactor; as a result, polymerization and eCO 2 R could be performed independently for different periods of time. To test the performance of the tandem system, the above mentioned recirculatory eCO 2 R experiments were repeated, and the produced gas was pumped through the polymerization glass-frit reactor loaded with a 5 mM solution of Pd-PP in TCE for 16 hours at a rate of 20 sccm. To compare the performance of Pd-PP with an eCO 2 R product gas feed against a mixture of pure CO, C 2 H 4 and CO 2 , a series of experiments was performed, where the gas tank was charged with mixtures of pure gases. After the same work-up procedure as in the single-pass experiments, pure gas-derived polyketones were generated via eCO 2 R and isolated.

The eCO 2 R-derived polyketone yields were close to the expectation deduced from pure gas experiments based on their CO and C 2 H 4 content (Fig. 3c-d). For example, with pure gas mixtures containing 10% CO and 10% C 2 H 4 , 39.4 mg of polymer were obtained, while the eCO 2 R gas mixture from the optimized setup afforded 57.9 mg of polymer. Additionally, in recirculation mode, the introduction of water vapor into the gas mixture via Cu and Ag GDEs without applying a potential did not result in a reduction in observed polymerization activity (Fig. 3c). This observation is explained by the fact that the total amount of water introduced into the gas mixture is low due to the small volume of recirculated gas, which did not produce a high enough concentration of water in the Pd-PP solution in the polymerization reactor to affect activity, matching previous reports. 51 This conclusion is supported when the maximum possible water content in the Pd-PP solution after eCO 2 R is calculated: the mass fraction of water is expected to be approximately 10 -4 , significantly below the reported 1% threshold. 51 Comparison of different GDE combinations showed that the current-to-polymer efficiency (CPE) increased greatly when CO deficit is overcome with the addition of small Ag GDEs: when only Cu GDEs were used, the CPE was 12.0%, compared to 43.8% and 51.0% CPE with Ag GDEs after 1-and 2 hour eCO 2 R (Fig. 3d). The corresponding CO 2 utilization figures are 2.39%, 8.76%, and 13.8%, respectively, where substantial CO 2 loss is, again, attributed to CO 2 crossover, as well as the fact that CO 2 was also converted into other products other than ethylene and CO in eCO 2 R.

The identity of the products was proven to be perfectly alternating polyketones using 13 C and 1 H NMR spectroscopy (Fig. 4, S82-S106). To demonstrate that CO 2 was the carbon source in the polyketones, a tandem eCO 2 R copolymerization was performed using 13 C-labeled CO 2 as the eCO 2 R feed. The 13 C NMR spectrum of the solid product shows intense peaks at 213 and 35.6 ppm, corresponding to carbonyl and methylene C atoms of the polyketone chain, respectively. 29,56 These data establish that labeled CO 2 is the source of carbon, as do additional smaller peaks corresponding to the chain ends, assigned using 1 H- 13 C HSQC and HMBC NMR spectroscopy (Fig. 4, S72-S73). The 1 H NMR spectrum displays a C-H splitting of the methylene peak with J C-H = 125 Hz further supporting this claim (Fig. S106). 29,56 Furthermore, NMR end group analysis suggests that termination only occurred during the acidified methanol workup, which is indicated by a triplet and a quartet at 0.82 and 2.21 ppm, respectively, consistent with a terminal ethyl group produced via protonolysis of the Pd-CH 2 bond; no peaks consistent with βhydrogen elimination, which is the only possible termination mode in TCE, are observed. In addition, no evidence of incorporation of H 2 and ethanol or propanol generated during eCO 2 R has been observed. The former is indicated by similar molecular weights of polyketones obtained from pure gas mixtures and eCO 2 R products determined by size exclusion chromatography (SEC; Table S3, Fig. S107-S112) and NMR spectroscopy, indicating no chain transfer by dihydrogen that has been reported previously. The latter is indicated by the absence of -C(=O)OR peaks in the 13 C NMR spectrum of the 13 C-labeled polyketone, which are expected in the 170-180 ppm region relative to TMS. 56 Moreover, matching our expectations based on published studies, no evidence of CO 2 incorporation into the polymer has been observed by NMR spectroscopy. Finally, analysis of polymer samples by SEC indicates that CO 2 -derived polyketones are quite narrowly dispersed with a polydispersity index of 1.1.

Conclusion

This report presents a novel approach to CO 2 utilization via conversion to polyketones by coupling electrochemical CO 2 reduction and thermochemical CO-C 2 H 4 copolymerization catalyzed by an organometallic catalyst. Through careful consideration of reactor and GDE cell design, as well as the optimal polymerization conditions and organometallic catalyst selection, we were able to obtain polyketones derived entirely from CO 2 at an appreciable activity, current utilization, and CO 2 conversion, with all processes being performed under mild conditions at room temperature and atmospheric pressure. Nevertheless, a few serious challenges remain. For example, comparatively low CO 2 -to-polyketone conversion occurs due to CO 2 crossover. This factor makes the system presented herein less competitive than polyketone generation systems utilizing inexpensive petroleum-derived ethylene and CO 2 -derived CO obtained with higher FE on Ag catalysts, 29 which can result in higher overall CO 2 conversion. This approach, however, would introduce fossilized carbon into the atmosphere, given the photodegradability of polyketones and the CO 2 footprint of petroleum-derived ethylene, 5,6,37,38 unlike the system presented in this paper. Additionally, at the low current densities towards CO and C 2 H 4 and corresponding low concentrations of these substrates, low polymerization activity is observed, and the molecular weight of CO 2 -derived polyketones is low, raising concerns about their material properties in comparison to industrially produced polyketones. Nevertheless, since termination only occurred during the acidified methanol workup, by carrying out more CO 2 reduction and, thus, generating more CO and ethylene, the polymer molecular weight can be increased. Moreover, the lack of GDE stability limiting the eCO 2 R time to 1-2 hours poses a challenge to industrial implementation of this system. However, multiple strategies such as careful pressure control to minimize GDE flooding, as well as GDE rinsing to mitigate salt deposition have been studied to prolong their lifetimes. 60,[68][69][70] As a result, while future engineering designs are warranted to address these shortcomings, the ability to derive polyketone entirely from CO 2 via eCOR 2 , as demonstrated here, provides an important starting point.