Ionic Liquid-Modified Porous Organometallic Polymers as Efficient and Selective Photocatalysts for Visible-Light-Driven CO2 Reduction

In the photoreduction of CO2 to CO, the competitive H2 evolution is always inevitable due to the approximate reduction potentials of H+/H2 and CO2/CO, which results in poor selectivity for CO production. Herein, imidazolium-type ionic liquid- (IL-) modified rhenium bipyridine-based porous organometallic polymers (Re-POMP-IL) were designed as efficient and selective photocatalysts for visible-light CO2 photoreduction to CO based on the affinity of IL with CO2. Photoreduction studies demonstrated that CO2 photoreduction promoted by Re-POMP-IL functioning as the catalyst exhibits excellent CO selectivity up to 95.5% and generate 40.1 mmol CO/g of Re-POMP-IL1.0 (obtained by providing equivalent [(5,5′-divinyl-2,2′-bipyridine)Re(CO)3Cl] and 3-ethyl-1-vinyl-1H-imidazol-3-ium bromide) at 12 h, outperforming that attained with the corresponding Re-POMP analogue without IL, which highlights the crucial role of IL. Notably, CO2 adsorption, light harvesting, and transfer of photogenerated charges as key steps for CO2RR were studied by employing POMPs modified with different amounts of IL as photocatalysts, among which the CO2 affinity as an important factor for POMPs catalyzed CO2 reduction is revealed. Overall, this work provides a practical pathway to improve the CO2 photoreduction efficiency and CO selectivity by employing IL as a regulator.


Introduction
Visible-light-driven reduction of CO 2 to fuel and/or chemicals provides a promising solution to solve both the environmental problems and the increased energy requirements [1][2][3][4][5]. As a typical product of CO 2 reduction reaction (CO 2 RR), carbon monoxide (CO) is especially valuable in chemical industry because of its usage as the feedstock for a variety of carbon-based fuels [6,7]. For photoreduction of CO 2 to CO, a proton source is always involved, producing the competitive H 2 evolution accompanied with the CO generation. In fact, the H 2 formation derived from protons trapping electrons is more favorable than the CO production from CO 2 RR by comparing their reduction potentials [3]. Therefore, the development of photocatalysts capable of disfavouring the H 2 evolution in an attempt to enhance the efficiency and CO selectivity in CO 2 RR is urgently needed.
For CO 2 photoreduction to CO, rhenium complexes of fac-[Re I (N^N)(CO) 3 X] (N^N=diimine, X=Cl, Br) as photo-catalysts have gained long-term attention because of their excellent performance [8][9][10][11][12][13][14][15][16]. However, the notorious instability of fac-[Re I (N^N)(CO) 3 X] deriving form bimetallic decomposition greatly limits their application. Porous organic polymers (POPs), a type of polymer porous materials with micropores and/or mesopores formed by covalent bonds of organic structural units, feature with advantages of good chemical stability and easily tailored structures [17][18][19][20]. To date, several rhenium-metalated conjugated microporous polymers have been designed and demonstrated to be stable photocatalysts for CO 2 photoreduction [21][22][23][24][25]. Even though, the reported Re(I)-metalated porous materials, commonly fabricated by post modification, namely, polypyridine linkers are firstly constructed and then coordinate with Re(CO) 5 Cl [21][22][23][24][25]. Unfortunately, with post modification, incomplete coordination and random metal anchoring are difficult to avoid [26], resulting in compromised catalytic performance and/or inevitable H 2 evolution in most cases. Comparatively, direct polymerization of metal complexes to prepare porous organometallic polymers (POMPs) can maximally maintain the structures of molecular catalysts, meanwhile, ensure uniform dispersion of metal sites in the porous matrix [26]. On the other hand, as the photocatalysts for CO 2 RR, characteristics with good CO 2 adsorption, remarkable light-harvesting property, and efficient separation of photogenerated electrons and holes will be desirable [27]. However, it will be a time-consuming work to design an efficient photocatalyst for CO 2 RR, because there is no panacea to meet all of these features. Hence, locating the key factors to CO 2 RR will be helpful to solve this problem.
Ionic liquids (ILs), composed of cations and anions, have been widely applied in CO 2 capture and conversion [28][29][30][31][32][33]. For facilitated CO 2 absorption, IL has been directly added into the photocatalytic CO 2 systems, which are proven to accelerate the process of CO 2 to CO conversion and inhibits H 2 generation [34,35]. In addition, the ionic group on the second coordination sphere, i.e., ionic second coordination sphere, has been proven to be promoting CO 2 photoreduction [16]. Therefore, we speculate ILs as regulators that can be covalently introduced into porous materials to enhance their photocatalytic performance.
Herein, the novel IL-modified Re-based POMP (Re-POMP-IL) through the simple and scalable copolymerization of Re bipyridine complex and IL was designed and synthesized as shown in Figure 1. By revealing the performance by employing Re-POMP-IL as the photocatalyst for CO 2 reduction to CO, the promoting effect of IL on photocatalytic activity and CO selectivity is highlighted. In addition, the influence of CO 2 adsorption ability, optical property, and lifetime for photogenerated charges on the performance of these polymers in photocatalytic CO 2 RR was systematically evaluated by tuning the amount of IL in the preparation of Re-POMP-IL, in which the important factor of efficiency enhancement ascribed to the easy CO 2 adsorption of Re-POMP-IL was exposed.

Results
As shown in Figure 1, Re-based polymers used in this work can be easily prepared through the solvothermal copolymerization of [(5,5′-divinyl-2,2′-bipyridine)Re(CO) 3 Cl] (1a) and 3-ethyl-1-vinyl-1H-imidazol-3-ium bromide (2a). For Re-POMP-IL1.0, an equivalent 1a and 2a were used. For comparison, Re-POMP without IL moiety was also synthesized by self-polymerization of 1a. The synthesized polymers, attained as yellow solids, are stable in air and insoluble in water and organic solvents (e.g., CH 3 CN). The chemical structures of Re-POMP and Re-POMP-IL1.0 were characterized by FT-IR and solid-state 13 C CP/MAS NMR technique. As exhibited in Figure 2(a), signals between 2025 and 1870 cm -1 are ascribed to the stretching vibration of CO ligand coordinated with Re(I). Compared with Re-POMP, the occurrence of the new peak at 1157 cm -1 in the FT-IR spectrum of Re-POMP-IL1.0 arises from the C-N bond vibration of covalently linked 2a. In the solid-state 13 C CP/MAS NMR spectra (Figure 2(b)), chemical shifts ranging from 154 to 123 ppm indicate the signals of bipyridyl moieties, and the disappearance of signals of vinyl groups accom-panying with the appearance of a new peak at 39 ppm demonstrates the successful polymerization of monomers [36]. The other peaks at 58, 46, 32, 18, and 15 ppm are assigned to the carbons of alkyl chains of polymerized 2a. Similar morphologies of Re-POMP and Re-POMP-IL1.0 were observed by scanning electron microscopy (SEM) ( Figure S4, see Supplementary Materials) and transmission electron microscopy (TEM) ( Figure S4, see Supplementary Materials) measurements, which indicates the addition of IL unit does not affect the morphology of Re-POMP. The thermal stability of Re-POMP and Re-POMP-IL1.0 was measured through thermogravimetric analysis (TGA), and no obvious weight loss was found until temperature was above 300°C ( Figure S5, see Supplementary Materials). Additionally, power X-ray diffraction (PXRD) pattern of Re-POMP-IL1.0 exhibits the broad diffraction peak ( Figure S6, see Supplementary Materials), revealing the amorphous character of the polymer.
Subsequently, the elemental compositions of Re-POMP and Re-POMP-IL1.0 were identified by X-ray photoelectron spectroscopy (XPS). As shown in Figure 2(c), elements including C, N, O, Cl, and Re exist in the survey spectrum of Re-POMP. For Re-POMP-IL1.0, element of Br is also detected because of the introduction of 2a. The highresolution Re 4f spectra display two peaks belonging to Re 4f 5/2 and Re 4f 7/2 , respectively ( Figure 2(d)). Similar chemical environment of Re in 1a, Re-POMP and Re-POMP-IL1.0 was observed, indicating that the polymerization does not destroy the coordination of Re(I) complex. Compared with Re-POMP, both of Re 4f peaks shift to lower binding energies for Re-POMP-IL1.0, indicating the increased Re electron density after the installation of IL. The high electron density on Re atom may facilitate the interaction between Re and CO 2 (vide infra). The accurate metal loading in polymers was further measured by inductively coupled plasma-optical emission spectroscopy (ICP-OES), and the Re contents of 31.0 and 24.4 wt% in Re-POMP and Re-POMP-IL1.0 were detected, respectively, which are basically consistent with the theoretical results. It is noteworthy that this is the highest Re content detected among the reported Re-based porous materials used in CO 2 RR [21][22][23][24][25] (Table fS1, see Supplementary Materials), and such high concentrations of bpy-Re were conceived to dramatically improve their activity in CO 2 photoreduction.
The pore structures of polymers were investigated via N 2 physical sorption at 77 K. As exhibited in Figures 3(a) and 3(b), the N 2 adsorption/desorption isotherms of Re-POMP and Re-POMP-IL1.0 showed similar characteristics with a combination of type I and type IV patterns [37] with predominant pore sizes ranging from 0.5 to 10 nm, indicating their hierarchical porosity with micro-and mesopores. In fact, hierarchical porous polymers have remarkable advantages over those with microporous structures, profiting from the existence of mesopores facilitates the mass transfer process [17]. Porosity parameters of Re-POMP and Re-POMP-IL1.0 are summarized in Table 1, the Brunauer-Emmett-Teller (BET) surface areas are 452 and 326 m 2 ·g -1 , respectively, and total pore volumes are 0.33 and 0.24 cm 3 ·g -1 , respectively. Clearly, the introduction of 2a    leads to the decrease in BET surface area and total pore volume, possibly due to the blocking of intrinsic pores by flexible chains or Brof IL [38]. The CO 2 uptake capacities of Re-POMP and Re-POMP-IL1.0 were also examined ( Figure 3(c) and Table 1). A slightly reduced adsorption capacity of Re-POMP-IL1.0 relative to Re-POMP was observed, probably owing to the obvious reduced BET surface area and total pore volume after the introduction of IL. The interaction between polymers and CO 2 was further evaluated by using the Clausius−Clapeyron equation to calculate the isosteric heat of adsorption (Q st ). As described in Figure 3(d) and Table 1   being attributed to the electrostatic interaction between IL and CO 2 [39,40] and the increased Re electron density in Re-POMP-IL1.0. The optical absorption and band gaps of Re-POMP and Re-POMP-IL1.0 were studied by UV/Vis diffuse reflectance spectra as illustrated in Figure 4(a), in which the good visible-light absorption of polymers was observed. Compared with Re-POMP, Re-POMPIL1.0 shows about 50 nm blue shift in absorption onset, probably being ascribed to the existence of IL reduce the conjugated structures [41]. According to their absorption edges, the optical band gaps of Re-POMP and Re-POMP-IL1.0 were obtained from the Tauc plots, in which the band gap is broadened as the presence of IL unit ( Figure S7, see Supplementary Materials). In addition, charge separation behavior was investigated through steady-state time-resolved fluorescence spectroscopy ( Figure S8, see Supplementary Materials). Fitting results of the fluorescence attenuation curves showed that lifetimes of 80 and 93 ns were calculated for Re-POMP and Re-POMP-IL1.0 (Table 1), respectively, which are sufficient to enable the polymers to exhibit good charge separation. Specifically, the fluorescence lifetime of the polymer is prolonged after IL addition, meaning that Re-POMP-IL1.0 have more free charge to participate in the photoreaction than Re-POMP.
The semiconductor properties and the electronic band positions were estimated by the Mott-Schottky measurements at frequencies of 500 and 1000 Hz ( Figure S9 their flat band potentials [21], which are the intersection values on the abscissa obtained by depicting the plots of C -2 values relative to applied potentials. As shown in Figure 4 Collectively, the installation of IL into Re-POMP has a little effect on the morphology, while the added IL leads to a decrease in BET surface area so as to affect the CO 2 adsorption capacity, and simultaneously impede the visible-light harvest. On the other hand, the significantly enhanced CO 2 affinity and prolonged lifetime of photogenerated charges are found because of the introduction of IL unit. Upon acquiring these features, the performance of IL-modified Re-based porous material as photocatalyst for CO 2 reduction to CO was evaluated and the decisive factor affecting the CO 2 RR was further elucidated. With Re-POMP-IL1.0 as the photocatalyst, the photocatalytic experiment was firstly carried out under a pure CO 2 atmosphere (1.0 atm, 298 K) in MeCN solution with triethanolamine (TEOA) as the sacrificial agent and 500 W longarc Xenon lamp (λ ≥ 400 nm) as the light source. As shown in Figure 5(a), CO 2 can be effectively reduced to CO with a TON value of 30.9 and CO selectivity of 95.4%. In fact, the produced CO amount of 40.1 mmol/g Re-POMP-IL1.0 at 12 h is more than those of reported Re(I)-metalated porous polymers [21][22][23][24][25] (Table S1, see Supplementary Materials), probably benefiting from the relatively high Re content and the hierarchical porosity of Re-POMP-IL1.0. No other reduction products such as HCOOH and MeOH were found, and H 2 was detected as a by-product owing to the competitive proton reduction [22]. For comparison, employing Re-POMP as the photocatalyst under the consistent conditions only give TON for CO (TON CO ) of 16.1 and selectivity of 69%. These results can clearly demonstrate that IL as a modifier has the ability to enhance the performance of Re(I)-based porous material. In addition, self-polymerization of 2a was conducted, in which the polymerized ionic liquid named as PIL was obtained and characterized ( Figure S10 and S11, see Supplementary Materials). However, physically mixing of Re-POMP and PIL or IL 2a cannot achieve the comparable activity to that of Re-POMP-IL1.0 and individual PIL or 2a as the photocatalyst does not work, meaning that the superior properties of IL-modified porous organometallic polymers are the key to enhance the photocatalytic performance.
In addition, it is noteworthy that these synthesized Re-POMPs exhibited even better activity than the homogeneous one, i.e., 1a (Figure 5(a)), demonstrating the superiority of pore materials which may provide a unique reaction microenvironment and has the ability to increase the CO 2 concentration around the catalytic sites [17,18]. Particularly, POMP exhibits good photostability and Re-POMP-IL1.0 as the photocatalyst for CO 2 reduction keeps working even if the irradiation time is prolonged to 60 h ( Figure 5(b)). At this time, TON CO value is 56 producing CO amount of 73 mmol/g of Re-POMP-IL1.0. Further investigation of the necessity of each component in the catalytic system showed that CO cannot be produced if any of Re-POMP-IL1.0, light source, TEOA, or CO 2 was absent (Table S2 in Supplementary Materials). The isotopic labeling experiment by employing 13 CO 2 for CO 2 RR affords 13 CO as the major product determined by GC-MS analysis, confirming the CO source originating from CO 2 ( Figure S12, see Supplementary Materials).
For CO 2 RR, light harvesting, transfer of photogenerated charges as well as CO 2 capture and conversion are key steps [27]. In fact, slightly reduced CO 2 adsorption capacity, shortened optical absorption range, significantly enhanced CO 2 affinity, and prolonged lifetime of photogenerated charges of Re-POMP-IL1.0 relative to Re-POMP were observed in this study (see Figures 3(c) and 4(a), Table 1, and Figure S8). The better catalytic performance of Re-POMP-IL1.0 than Re-POMP suggests that the processes of interaction with CO 2 and separation of photogenerated charges may be closer factors than CO 2 uptake amount and optical property for POMPs catalyzed CO 2 RR.
In view that the amount of IL in polymer has an effect on the properties, which may affect its photocatalytic activity, Re-POMP-IL0.4 and Re-POMP-IL1.5 were prepared by adjusting the amount of added 2a in copolymerization using the same synthetic method with Re-POMP-IL1.0. Through ICP-OES, Re contents in Re-POMP-IL0.4 and Re-POMP-IL1.5 were measured as 29.3 and 22.6 wt%, respectively. The N 2 adsorption/desorption isotherms, pore size distributions, porosity properties, CO 2 uptake capacities, and excitedstate lifetimes of Re-POMP-IL0.4 and Re-POMP-IL1.5 were measured in comparison with those of Re-POMP and Re-POMP-IL1.0 in Figure S13 and Table S3 (see  Supplementary Materials). Re-POMP-IL1.5 exhibits much lower surface area than other POMPs, being attributed to the significantly reduced meso-and micropores probably because of the blocking of intrinsic pores by more IL [38]. The higher CO 2 uptake amounts of Re-POMP-IL0.4 than Re-POMP further highlight the positive effect of IL on CO 2 absorption, while the reduced CO 2 uptake amounts of Re-POMP-IL1.0 and Re-POMP-IL1.5 than those of Re-POMP could be probably due to their obviously reduced BET surface area and pore volume. Accompanying with the introduction of IL, the gradually increasing Q st values in the order of Re-POMP-IL1.0>Re-POMP-IL0.4>Re-POMP were observed, suggesting the enhanced CO 2 capture with more IL installed. On the other hand, CO 2 absorption behavior is also affected by porous properties; thus, a slight reduced Q st value of Re-POMP-IL1.5 compared with Re-POMP-IL1.0 may be attributed to its significantly reduced mesoand micropores. The low to moderate Q st values of POMPs allow reversible CO 2 absorption and relatively easy CO 2 desorption, indicating major physical adsorption rather than chemical adsorption of POMPs for CO 2 uptake was performed [38,44]. Moreover, the continually prolonged excited-state lifetimes of POMPs accompanied with the increase of added IL amount were discovered.
Then, Re-POMP-IL0.4 and Re-POMP-IL1.5 were applied to the photocatalytic CO 2 reduction and both of them exhibits higher activity and CO selectivity than Re-POMP ( Figure 5(c)), further illustrating the promoting effect on photocatalytic performance by chemically introducing IL into polymer and simultaneously proving the rationality of the design. In addition, the amount of IL in polymer also has an influence on the activity. Increased TON CO and selectivity of CO were found when advancing the amount of 2a from 0 to 1.0 equivalent relative to 1a, probably profiting from the increased IL in polymer can reduce its energy consumption for CO 2 capture and simultaneously lead to the prolonged excited-state lifetime. However, continuing to increase the amount of 2a to 1.5 equivalent relative to 1a, the produced CO began to decrease even though Re-POMP-IL1.5 owns the longest excited-state lifetime among these polymers. On the other hand, the enhanced photocatalytic activity in the order of Re-POMP<Re-POMP-IL0.4<Re-POMP-IL1.5<Re-POMP-IL1.0 was observed ( Figure 5(c), Table S3), being consistent with their Q st values. These results further elucidate the important effect of CO 2 affinity on CO 2 RR using POMPs as the photocatalysts. This finding is different from the previous work, in which the high CO 2 adsorption capacity, broadened optical absorption, and/or facile charge separation are always highlighted [21,23,25].

Discussion
In conclusion, the IL-modified Re-based porous organometallic polymers (Re-POMP-IL) have been designed and developed as effective photocatalysts for CO 2 reduction to CO. Specifically, the installation of IL (3-ethyl-1-vinyl-1Himidazol-3-ium bromide used in this work) can obviously enhance the photocatalytic activity and CO selectivity. Re-POMP-IL can be obtained through radical copolymerization, which is simple and scalable. Characteristics on pore features, optical properties, redox abilities, and excited-state lifetimes of Re-POMP-IL demonstrated that they have high CO 2 uptake capacities, visible-light absorption, suitable redox potential, and long enough excited-state lifetimes for CO 2 reduction. Catalytic activity tests showed that Re-POMP-IL behaves better than Re-POMP in promoting CO 2 to selectively produce CO, and the competitive H 2 evolution can be effectively suppressed with CO selectivity up to 95.5%. Investigation on the effect of IL content on the catalytic performance of POMPs shows that Re-POMP-IL1.0 with the largest CO 2 isosteric heat of adsorption exhibits the best catalytic performance, emphasizing the important role of CO 2 affinity for POMPs catalyzed CO 2 RR. In a word, these results demonstrate the feasibility of using IL as a regulator to develop efficient and selective photocatalyst for CO 2 reduction.
2a [46]: white solid ( , and DMF (7 mL) were added to a 25 mL autoclave. Then, the autoclave was sealed and heated at 100°C for 24 h. When finished, the autoclave was cooled to room temperature and the resulting mixture was transferred to a 50 mL conical flask, followed by adding 20 mL ethanol to soak overnight. Then, the obtained mixture was filtered and washed with ethanol until the filtrate was colorless. After the filter cake was dried under vacuum at 70°C for overnight, the yellow solid (185.7 mg) was obtained and named as Re-POMP.

Synthesis of Re-POMP-IL.
Take the synthesis of Re-POMP-IL1.0 as an example, 1a (200 mg, 0.39 mmol), 2a (78.8 mg, 0.39 mmol), AIBN (10 mg, 0.06 mmol), and DMF (7 mL) were added to a 25 mL autoclave. Then, the autoclave was heated at 100°C for 24 h. When finished, the autoclave was cooled to room temperature and the resulting mixture was transferred to a 50 mL conical flask, followed by adding 20 mL ethanol to soak overnight. Then, the obtained mixture was filtered and washed with ethanol until the filtrate was colorless. After the filter cake was dried under vacuum at 70°C for overnight, the yellow solid (268.3 mg) was obtained and named as Re-POMP-IL1.0.
For the synthesis of Re-POMP-IL0.4 and Re-POMP-IL1.5, their synthetic procedures are similar to that of Re-POMP-IL1.0, except that the amount of 2a added was adjusted to 0.16 mmol and 0.59 mmol, respectively. 4.5. Synthesis of PIL. 2a (505 mg, 2.5 mmol), AIBN (24.6 mg, 0.15 mmol), and ethanol (5 mL) were added to a 50 mL Schlenk flask under Ar atmosphere. Then, the reaction was conducted at 80°C for 16 h. When finished, the resultant mixture was cooled to room temperature, followed by adding 20 mL ethanol to soak overnight. Then, the obtained mixture was filtered and washed with ethanol (5 × 20 mL). After the filter cake was dried under vacuum at 70°C for overnight, PIL (420.2 mg) was obtained as white solid. 4.6. General Procedure for Photocatalytic CO 2 Reduction. To a 25 mL Schlenk tube, photocatalyst (1 mg), TEOA (1.12 g, 1 mL) and MeCN (3 mL) were successively added. Then, the reaction mixture was sonicated to allow the polymer to disperse evenly in the whole solution. Air in the Schlenk tube was replaced by CO 2 through the freeze-pump-thaw method. Then, the reaction tube was sealed and placed under a 500 W long-arc Xenon lamp (λ ≥ 400 nm) for 12 h at room temperature. The light source was cooled by flowing cooling water before irradiating the reaction tube and the number of moles of photons absorbed by the photocatalyst was measured as 1:4 × 10 −8 einstein/s (see Supplementary Materials). After the reaction, partial gaseous products (1 mL) were taken from the tube using a syringe and then analyzed by gas chromatography with a TCD detector.