A New Dinuclear Cobalt Complex for Copolymerization of CO2 and Propylene Oxide: High Activity and Selectivity

Based on the ligand H4Salen-8tBu (salen-4), a new dinuclear cobalt complex (salen-4)[Co(III)TFA]2 (salen-4 = 3,5-di-tert-butylsalicylaldehyde-3,3′-diaminobiphenylamine; TFA = trifluoroacetic acid) has been firstly synthesized and characterized. It shows high catalytic activity for the copolymerization of propylene oxide (PO) and carbon dioxide (CO2), yielding regioregular poly(propylene carbonate) (PPC) with little generation of propylene carbonate (PC) by-product. It has been found that (salen-4)[Co(III)TFA]2 shows higher activity at milder conditions, generating a polymer with maximum Mn of 293 kg/mol and a narrow molecular weight distribution PDI of 1.35. The influences of reaction time, CO2 pressure, reaction temperature, nature of the cocatalyst, catalyst dosage and substrate concentration on the molecular weight, yield and selectivity of the polymer were explored in detail. The results showed that the (salen-4)[Co(III)TFA]2/[PPN]TFA catalyst system demonstrated a remarkable TOF as high as 735 h–1. In addition, a hypothetical catalytic reaction mechanism was proposed based on density functional theory (DFT) calculations and the catalytic reaction results of the (salen-4)[Co(III)TFA]2.


Introduction
With the development of social industrialization, the concentration of CO 2 in the atmosphere keeps rising, resulting in the greenhouse effect and a series of associated environmental disasters [1]. Nevertheless, CO 2 is an inexpensive, a nontoxic, rich and special renewable C1 building block [2][3][4]. So far, wide attention has been paid to the chemical conversion of CO 2 into high value-added energy, materials and chemical products. The beneficial use of CO 2 including the alternating copolymerization of CO 2 and epoxides to make degradable polycarbonates were popular (Scheme 1), which is considered one of the most potential green polymerization processes [5][6][7][8][9][10][11]. In addition, polycarbonates have become important topics because of their environment-friendly properties, such as atom-economy, energy-saving and degradation, as well as their several biomedical and pharmaceutical applications [12]. The problem is that CO2 is a highly stable and low-reactivity molecule, making it difficult to use CO2 as a reagent in a chemical reaction. Inspired by Inoue's pioneering work, researchers have developed many catalysts [13][14][15][16][17][18][19][20][21][22][23]. Nevertheless, it is still a big challenge for scientists to develop lowcost, easily prepared and highly efficient catalysts for fixation and conversion of CO2 under mild conditions. Among them, salen metal complexes were widely used due to their easy modification,

Synthesis and Characterization of (Salen-4)[Co(III)TFA]2
Based on the ligand H4Salen     Complex (salen-4)[Co(III)TFA]2 was characterized by FT-IR ( Figure 2). As is readily observed, compared with the ligand salen-4, 2957 cm −1 and 2870 cm −1 can be considered as the stretching vibration of C-H in the methyl group on the tert-butyl; 1682 cm −1 can be considered as the characteristic absorption peak of axial ligand CF3COO − . Due to the coordination of the ligand with Co(III), the characteristic absorption peak of the 1634 cm −1 v(C=N) in salen-4 shifted to the lower wavenumber by 21 cm −1 , and peaked at 1613 cm −1 . Here, 1522 cm −1 and 1484 cm −1 can be considered as the skeleton vibration of the benzene, and 1387 cm −1 can be considered as the inplane bending vibration of the C-H on -CH3. Due to the drop of the hydrogen atom on Ph-OH during the coordination process, salen-4 v(C-O) stretching vibration of Ph-OH at 1272 cm −1 is shifted by 16 cm −1 to the lower wavenumber, and the peak appeared at 1253 cm −1 . Here, 1175 cm −1 can be considered as the C-H inplane bending vibration of the benzene and 783 cm −1 can be considered as the out-of-plane Complex (salen-4)[Co(III)TFA] 2 was characterized by FT-IR ( Figure 2). As is readily observed, compared with the ligand salen-4, 2957 cm −1 and 2870 cm −1 can be considered as the stretching vibration of C-H in the methyl group on the tert-butyl; 1682 cm −1 can be considered as the characteristic absorption peak of axial ligand CF 3 COO − . Due to the coordination of the ligand with Co(III), the characteristic absorption peak of the 1634 cm −1 v(C=N) in salen-4 shifted to the lower wavenumber by 21 cm −1 , and peaked at 1613 cm −1 . Here, 1522 cm −1 and 1484 cm −1 can be considered as the skeleton vibration of the benzene, and 1387 cm −1 can be considered as the inplane bending vibration of the C-H on -CH 3 . Due to the drop of the hydrogen atom on Ph-OH during the coordination process, salen-4 v(C-O) stretching vibration of Ph-OH at 1272 cm −1 is shifted by 16 cm −1 to the lower wavenumber, and the peak appeared at 1253 cm −1 . Here, 1175 cm −1 can be considered as the C-H inplane bending vibration of the benzene and 783 cm −1 can be considered as the out-of-plane bending vibration of C-H on the benzene. Therefore, Salen-4[Co(III)TFA] 2 was successfully synthesized.

(Salen)Co(III)X Catalyzed PO/CO2 Copolymerization
The copolymerization of PO/CO2 was conducted in the presence of various catalysts, and the results are listed in Table 1. We designed a series of experiments to explore the potential catalytic performance of (salen)Co(III)X. As shown in Table 1, the catalyst and the cocatalyst had a significant effect on the copolymerization of PO/CO2. Dinuclear complex (salen-4)[Co(III)DNP]2 showed higher activity and selectivity than (salen-3)Co(III)DNP (which is one half of the dinuclear (salen-4)[Co(III)DNP]2 complexes structurally) (Entries 4 and 7, Table 1). The conversion of PO was 99%, the

(Salen)Co(III)X Catalyzed PO/CO 2 Copolymerization
The copolymerization of PO/CO 2 was conducted in the presence of various catalysts, and the results are listed in Table 1. We designed a series of experiments to explore the potential catalytic performance of (salen)Co(III)X. As shown in Table 1, the catalyst and the cocatalyst had a significant effect on the copolymerization of PO/CO 2 . Dinuclear complex (salen-4)[Co(III)DNP] 2 showed higher activity and selectivity than (salen-3)Co(III)DNP (which is one half of the dinuclear (salen-4)[Co(III)DNP] 2 complexes structurally) (Entries 4 and 7, Table 1). The conversion of PO was 99%, the selectivity of PPC was 91% and TOF was 225 h -1 . Importantly, with [PPN]TFA as the cocatalyst, the activity of dinuclear complex (salen-4)[Co(III)TFA] 2 was much higher than that of mononuclear complex (salen-3)Co(III)TFA (Entries 5 and 8, Table 1) with TOF reaching 230 h -1 , which was two times of that for (salen-3)Co(III)TFA. Interestingly, we found that as the number of tert-butyl groups with large steric hindrance on ligands of the complexes (Entries 1-3, Table 1) increased, the conversion of PO and selectivity of PPC both increased. With the rising of electron absorption capacity of axial groups of the complexes (Entries 3-5, Table 1), the conversion of PO increased continuously, so did the selectivity of PPC. Meanwhile, the dinuclear complexes (Entries 6-8, Table 1) showed the same results. By comparing mononuclear and dinuclear metal complexes of the same salen ligands and the same axial anions (Entries 3 and 6, 4 and 7, 5 and 8, Table 1), it was found that the conversion of PO and selectivity of PPC both increased slightly. The TOF also increased by two times, indicating that the dinuclear catalytic system was much more efficient than the mononuclear catalytic system for CO 2 and PO copolymerization.
Furthermore, the optimal catalyst (salen-4)[Co(III)TFA] 2 was selected. Next, the influences of reaction time, CO 2 pressure, reaction temperature, cocatalyst, catalyst dosage and substrate concentration on PO/CO 2 copolymerization were explored to improve the yield and selectivity of PPC.

Effect of Reaction Temperature
The conversion of PO increased along with the rising temperature and remained stable until it reached 25 • C. By contrast, the selectivity of PPC increased first and then decreased ( Table 2). That is because the binding of PO and metal center was blocked when the temperature was too low, which lowered the activation rate of PO, further resulting in the lower conversion of PO, the selectivity of PPC and polymerization rate. By contrast, with the rising temperature, the selectivity of PPC increased first and then decreased. That is because when the temperature is too high, the product tends to generate propylene carbonate (PC) with higher thermodynamic stability. It is known that the thermodynamic stability of PPC is lower than PC, which reduces the selectivity of PPC. Therefore, the optimal polymerization temperature of (salen-4)[Co(III)TFA] 2 /[PPN]TFA is 25 • C.

Effect of CO 2 Pressure
The conversion of PO and selectivity of PPC increased (Table 3) as the CO 2 pressure rose. When the pressure of CO 2 was too low, the combination of PO and metal center was blocked, thereby lowering the activation rate of PO, the conversion of PO, the selectivity of PPC and the polymerization rate. The yield of PPC increased along with the CO 2 pressure. When the CO 2 pressure exceeded 3.0 MPa, the yield of PPC did not increase significantly. Given that there is a potential safety hazard in laboratory operations when the CO 2 pressure is too high, thus, the optimal polymerization pressure of (salen-4)[Co(III)TFA] 2 /[PPN]TFA is 3.0 MPa.

Effect of Reaction Time
With the increase of reaction time, the conversion of PO increased gradually (Table 4). When the reaction time exceeded 4 h, the conversion of PO did not change much. The possible reason is that when the reaction time is too long, the polymer will adhere to the surface of the catalyst and reduces the contact area between the active center and PO, resulting in a slower catalytic rate. This indicates that the catalytic system basically reached equilibrium after 4 h. Therefore, the optimal reaction time of (salen-4)[Co(III)TFA] 2 /[PPN]TFA is 4 h.

Effect of Cocatalyst
It was found that cocatalysts played very important roles in binary systems. During the CO 2 copolymerization with PO, the CO 2 pressure was 3.0 MPa, the reaction temperature was 25 • C,  (Table 5). Therefore, [PPN]TFA was selected as the best cocatalyst to form a two-component catalytic system.

Theoretical Calculations of Quantum Chemistry
In this part, (salen-4)[Co(III)TFA] 2 was selected. In order to make the configuration simple and clear, the H atoms in the figure were omitted, but they were still included in the calculation process. Gaussian 16 [43] calculation program was used to calculate the (salen-4)[Co(III)TFA] 2 model (Figure 3), and the DFT B3LYP method [44] was used to optimize the geometric configuration of the (salen-4)[Co(III)TFA] 2 model at the level of pseudopotential group GENECP (the nonmetal atoms use the 6-31G** basis group [45] level, the Co atoms are heavy metal and the LanL2DZ basis group based on the effective core is adopted).
Gaussian 16 [43] calculation program was used to calculate the (salen-4)[Co(III)TFA]2 model ( Figure  3), and the DFT B3LYP method [44] was used to optimize the geometric configuration of the (salen-4)[Co(III)TFA]2 model at the level of pseudopotential group GENECP (the nonmetal atoms use the 6-31G** basis group [45] level, the Co atoms are heavy metal and the LanL2DZ basis group based on the effective core is adopted). From the optimized geometry of (salen-4)[Co(III)TFA]2 (Figure 4), it can be seen that the two central metal Co atoms are in pentacoordinated structures [46] in which two N atoms and two O From the optimized geometry of (salen-4)[Co(III)TFA] 2 (Figure 4), it can be seen that the two central metal Co atoms are in pentacoordinated structures [46] in which two N atoms and two O atoms of the ligand form the equatorial plane and the O atoms on the axial anion TFA occupy the axial position. Notably, the bond angles between the central metal Co atoms and the four coordination atoms of the ligand are about 90 • or 180 • , indicating that the Co atom is in a deformed planar quadrilateral field. The four bond angles formed by the metal Co atom and four coordination atoms on the ligand totals approximately about 360 • , that is, the metal Co atom and the four coordination atoms are substantially in the same plane. In addition, the bond angles between the O atoms of TFA and metal Co atom, and the four coordination atoms on the ligand are about 90 • . This indicates that the O atoms on the axial anion are basically perpendicular to the four coordination atoms on the ligand (Table S3).
From the energy of the frontal molecular orbitals (Table 7), it can be seen that the occupied orbitals are all negative, suggesting that the electronic state of the complexes is stable and its photoelectron spectrum is meaningful [47]. It can also be seen that the order of the LUMO orbital energy of the complexes is:

Proposed Mechanism for the Catalyst
Based on the above research, we proposed mechanisms for the dinuclear catalyst systems. In previous reports, a great deal of supported that CO 2 "insertion" actually occurs via a dissociative mechanism where a free nucleophilic chain end (alkoxide) attacks the CO 2 reversibly to give a carbonate end that is stabilized by the charge of the metal complex [16,37,41,[48][49][50][51]. As shown in Figure 5 V), which was supposed to be the key process for the bimetallic synergism. In the chain growth stage: CO 2 and PO were alternately copolymerized and M-O was broken to form poly(propylene carbonate) (VI) finally. The two active sites activated CO 2 and PO at the same time and greatly improved the activation rate, which also explained the reason why the catalytic performance of the dinuclear system was significantly better than that of the mononuclear system.
(I), and the two active centers simultaneously activated the attacking PO to form a pair of hexacoordinate structures (intermediate II). Subsequently, the nucleophile Y − attacked the C atom with a smaller steric hindrance on PO. Then, CO2 was inserted into the metal alkoxide bond (intermediate III), forming the carbonate unit (intermediate IV). The growing copolymer chain could migrate to the other mental center and attack the activated PO (intermediate V), which was supposed to be the key process for the bimetallic synergism. In the chain growth stage: CO2 and PO were alternately copolymerized and M-O was broken to form poly(propylene carbonate) (VI) finally. The two active sites activated CO2 and PO at the same time and greatly improved the activation rate, which also explained the reason why the catalytic performance of the dinuclear system was significantly better than that of the mononuclear system.

Materials and Methods
All reagents and solvents were analytical grade and were ready to use. The ligands H 2 Salen (salen-1), H 2 Salen-2 t Bu (salen-2), H 2 Salen-4 t Bu (salen-3), H 4 Salen-8 t Bu (salen-4) were synthesized following the reported procedure [41,52,53]. Propylene oxide was refluxed over a mixture of KOH/CaH 2 , and fractionally distilled under an argon atmosphere prior to use. Carbon dioxide (99.99%) and oxygen (99.99%) were purchased from Shaanxi Heping Glass Co., Ltd. (Shaanxi, China). and used as received. All other reagents were purchased from commercial sources and used as received. FT-IR spectra (Madison, WI, USA) were obtained on a VERTEX-70 Fourier transform infrared spectrometer with a band ranging from 4000 to 400 cm -1 . The elemental analysis of complexes was performed by Elementar Germany's Vario MICRO elemental analyzer (Hanau, Germany). The UV spectrum data of complexes were obtained by a Japan Shimadzu UV-2600 ultraviolet spectrophotometer (Kyoto, Japan). NMR spectra were recorded on a JEOL ECS400M spectrometer (Tokyo, Japan) with reference to the solvent signals.

Synthesis of Complexes
(Salen-1)Co(III)Cl. The ligand salen-1 (0.07 g, 0.22 mmol) and decrystallized water Co(OAc) 2 (0.05 g, 0.27 mmol) were dissolved in 10 mL anhydrous methanol. The mixture was stirred at room temperature for 12 h, and then anhydrous LiCl (0.05 g, 1.10 mmol) was added and passed in oxygen to continue the reaction for 12 h. Solvent was removed under reduced pressure and the residue was dissolved in CH 2 Cl 2 . The organic layer was rinsed with saturated aqueous Na 2 SO 4 (50 mL × 3) and NaCl (50 mL × 3), respectively. The organic layer was dried over Na 2 SO 4 , filtered and dried in vacuo. The crude product was recrystallized from CH 2 Cl 2 and anhydrous CH 3 CH 2 OH to obtain a brown-red solid [54]. were dissolved in 100 mL absolute ethanol. The mixture was heated to 80 • C for 30 min and then cooled to room temperature. The solvent was removed in vacuo and washed with cold methanol (15 mL × 3).
The solid was dissolved in 10 mL CH 2 Cl 2 and recrystallized by 200 mL n-hexane. After 24 h, a dark red crystal (salen-4)Co(II) was obtained. The (salen-4)Co(II) (0.07 g, 0.12 mmol) was dissolved in 50 mL CH 2 Cl 2 and 2, 4-dinitrophenol sodium (0.02 g, 0.12 mmol) was added. The resulting solution was stirred for 2 h at room temperature in an oxygen atmosphere. Solvent was removed in vacuo to obtain the black crude product. The crude product was recrystallized from CH 2 Cl 2 and n-hexane, respectively, and then dried in vacuo at 60 • C for 24 h [40]  (Salen-4)[Co(III)Cl] 2 . The ligand salen-4 (0.24 g, 0.22 mmol) was dissolved in 100 mL anhydrous methanol. The mixture was heated to reflux at 65 • C. After the ligand was completely dissolved, decrystallized water Co(OAc) 2 (0.09 g, 0.50 mmol) was dissolved in. Under oxygen atmosphere, anhydrous LiCl (0.02 g, 0.50 mmol) was added. After 12 h, the solvent was removed in vacuo. The residue was dissolved in CH 2 Cl 2 and recrystallized by anhydrous CH 3 CH 2 OH. The resulting solution was filtered and washed to obtain a brown-red solid [40] 2 . Under nitrogen atmosphere, the ligand salen-4 (2.16 g, 2.0 mmol) was dissolved in 20 mL CH 2 Cl 2 , and Co(OAc) 2 (0.72 g, 4.04 mmol) was dissolved in 80 mL anhydrous methanol. The mixture was stirred for 3 h at room temperature to obtain the (salen-4)Co(II) complex. Under oxygen atmosphere, (salen-4)Co(II) (1.20 g, 1.0 mmol) was dissolved in 50 mL CH 2 Cl 2 , and 2, 4-dinitrophenol sodium (0.41 g, 2.0 mmol) was added. The mixture was stirred for 3 h. Solvent was removed in vacuo to obtain the crude product. The crude product was recrystallized from CH 2 Cl 2 and n-hexane, respectively to obtain a dark green solid [40]

Catalytic Procedure
The catalyst, cocatalyst and epoxide were sequentially added into a 100 mL high-pressure reactor, and then the reactor was pressurized to a required pressure with CO 2 . The mixture was heated and stirred for a while. After completion of the reaction, unreacted CO 2 was evacuated to remove the pressure. A small aliquot of the resultant polymerization mixture was removed from the reactor for 1 H NMR and GPC analysis. The remaining polymerization mixture was then dissolved in tetrahydrofuran, quenched with 5% HCl solution in absolute ethanol, and precipitated from absolute ethanol (30 mL). The polymer was collected and dried in vacuo to constant weight, and the polymer yield was determined.

Conclusions
In summary, we have compared the mononuclear and dinuclear catalysts systems in PO and CO 2 copolymerization. The dinuclear catalyst system has higher activity, higher selectivity and PPC with higher molecular weight. A mechanism based on the conjugated dinuclear structure was proposed. At the same time, the effects of reaction temperature, CO 2 pressure, reaction time, catalyst's type and substrate concentration on the dinuclear catalytic system were studied. The DFT method was used to optimize the geometries of the complexes. Moreover, the stable structure of the complexes was further recognized in theory and its effect on the catalytic reaction was analyzed. Furthermore, an ideal catalyst model for PO/CO 2 copolymerization was constructed. The greater the number of tert-butyl groups on the ligand and the greater the anion electronegativity of the axial ligand, the higher the catalytic activity of the complex. The activity of the catalysts is expected to be further improved by changing the steric hindrances and electronic structures of the ligands, which will be further studied in the future.  Table S1: GPC of the crude product PPC,