Dinuclear Zinc Salen Catalysts for the Ring Opening Copolymerization of Epoxides and Carbon Dioxide or Anhydrides.

A series of four dizinc complexes coordinated by salen or salan ligands, derived from ortho-vanillin and bearing (±)-trans-1,2-diaminocyclohexane (L1) or 2,2-dimethyl-1,3-propanediamine (L2) backbones, is reported. The complexes are characterized using a combination of X-ray crystallography, multinuclear NMR, DOSY, and MALDI-TOF spectroscopies, and elemental analysis. The stability of the dinuclear complexes depends on the ligand structure, with the most stable complexes having imine substituents. The complexes are tested as catalysts for the ring-opening copolymerization (ROCOP) of CO2/cyclohexene oxide (CHO) and phthalic anhydride (PA)/CHO. All complexes are active, and the structure/activity relationships reveal that the complex having both L2 and imine substituents displays the highest activity. In the ROCOP of CO2/CHO its activity is equivalent to other metal salen catalysts (TOF = 44 h(-1) at a catalyst loading of 0.1 mol %, 30 bar of CO2, and 80 °C), while for the ROCOP of PA/CHO, its activity is slightly higher than other metal salen catalysts (TOF = 198 h(-1) at a catalyst loading of 1 mol % and 100 °C). Poly(ester-block-carbonate) polymers are also afforded using the most active catalyst by the one-pot terpolymerization of PA/CHO/CO2.


■ INTRODUCTION
Salen ligands, initially derived from the condensation of salicylaldehyde and ethylenediamine, are of fundamental importance in coordination chemistry and catalysis. They have been extensively studied since the pioneering work of Jacobsen 1 and Katsuki 2 using a chiral [Mn(salen)] complex for the enantioselective epoxidation of alkenes. Since then, a range of other transition metal catalysts have been reported for reactions as distinct as oxidations, 3 polymerizations, 4 and epoxidations. 5 More recently, a novel class of salen ligands have been developed incorporating a Lewis donor functional group in the ortho position of the salicylaldehyde unit, enabling the coordination of a second metal center within the ligand scaffold. 6 Dinuclear salen complexes, where two different metals (usually a transition metal and a lanthanide) are incorporated into the ligand, have been studied for their magnetic properties 7 and are even proposed as single-molecule magnets. 8 Such complexes are also attractive catalysts for asymmetric organic reactions, where it is proposed that they show better performances due to cooperative interactions between the two metal centers. 9 Exactly this type of metal cooperation has also been proposed to be an important criterion in the preparation of highly active catalysts for the ring-opening copolymerization (ROCOP) of CO 2 and epoxides. 10 One of the first well-characterized dinuclear catalysts for CO 2 /epoxide ROCOP was a zinc β-diiminate (BDI) complex that was shown to adopt dimeric structures in the most active catalysts. 10d Since this pivotal finding, many highly active dinuclear catalysts have been developed including those based on macrocyclic, 11 BDI, 10d,12 Trost type "pro-phenolate", 4d,13 porphyrin, 14 and anilido−aldimine 15 ligand scaffolds, typically coordinated to zinc, although other transition metal systems have also been investigated. 16 Mechanistic studies and density functional theory calculations have highlighted that the catalyst activity seems to be highly dependent on the flexibility of the ligand and the distance between the two metal centers. 10a,c,11c,17 To date, one of the most active catalyst systems for the synthesis of poly(cyclohexene carbonate) (PCHC), reported earlier this year by Rieger and co-workers; 18,19 is based on a flexible dizinc complex, coordinated by two β-diiminate moieties that are linked through the phenyl rings (TOF = 155 000 h −1 at a catalyst loading of 0.0125 mol %, 100°C, and 30 bar of CO 2 ).
Mononuclear salen ligands coordinated to either Co(III) or Cr(III) centers and combined with a Lewis base/ionic cocatalyst have also been widely employed as catalysts for CO 2 /epoxide ROCOP. 4i,j,20 In this case, the separation between the metal and the cocatalyst is of key of importance for the high activity. 21 Nozaki and co-workers pioneered a new generation of "cocatalyst tethered" salen complexes, which show a significant improvement in the catalyst activity. 21 The combination of tethering together the salen catalyst and cocatalyst was subsequently successfully applied by various researchers, 22 including Lu and Lee, to prepare some of the most active catalysts reported for ROCOP (TOF > 20 000 h −1 ). 23 A further benefit is that some of these tethered catalysts can be reclaimed from the polymer and recycled without significant loss in performance. 23g Another successful strategy, also initially developed by Nozaki and co-workers, has been to join together two distinct metal salen catalysts and to combine them with cocatalysts. 24 Such dinuclear catalysts are coordinated by two salen ancillary ligands that have been covalently linked to one another; 25 the catalysts show higher activity than mononuclear counterparts, particularly at low catalyst loadings. 24,25b Given the precedence for salen ligands and the promise of dinuclear catalysts, it is perhaps surprising that there is, so far, only a single report of ROCOP catalysts where a single salen ancillary ligand coordinates two metals. 26 The catalyst comprises a multidendate bis(benzotriazole iminophenol) ligand coordinated to two Zn(II), Ni(II), or Co(II) metal centers. The best activities were achieved with the di-Ni(II) catalyst, which for the ROCOP of CHO/CO 2 shows a TOF of 53 h −1 , at 120°C and 20 bar of CO 2 . 26 These encouraging results using dinuclear salen catalysts inspired the current development of new dinuclear salen ligands and the investigation of their coordination chemistry and catalytic applications for CO 2 / CHO and phthalic anhydride (PA)/CHO ROCOP. When developing new polymerization catalysts, it is attractive to target low-cost ligands, ideally derived from renewables, which can be coordinated to earth-abundant metal centers. 27 ortho-Vanillin is an interesting candidate, as it is relatively abundant and straightforward to extract from a range of plants. In addition, ortho-vanillin salen ligands have a good precedent for the formation of dinuclear complexes, although these have not yet been applied in this area of polymerization catalysis. 28 Zinc was also selected as the most desirable metal center as it is abundant, nontoxic, and usually results in colorless complexes.

■ RESULTS AND DISCUSSION
Using ortho-vanillin as a starting point, four new ligands were synthesized, isolated, and characterized. As mentioned, orthovanillin is an attractive starting material, as it is inexpensive and commercially available; furthermore, the targeted ligands were obtained on a large scale in only 1 d, through a simple, one-pot synthetic method (Scheme 1).
First, the amination of ortho-vanillin, using the relevant diamine [(±)-trans-1,2-diaminocyclohexane (L 1 ) or 2,2dimethyl-1,3-propanediamine (L 2 )], gave the desired imine product (L 1a H, L 2a H, Scheme 1). The ligands differ according to the diamine "linker" group and were selected on the basis that such C 2 and C 3 linkers are common components of successful ROCOP catalysts. 11a,f,21,23a Subsequent in situ reduction of the imine moieties on the salen ligands was performed, using NaBH 4 , to obtain the corresponding amines and salan ligands (L 1b H, L 2b H, Scheme 1). All four ligands were obtained in near quantitative yields and did not require purification ( Figure S1−8).
Crystals suitable for X-ray diffraction studies were obtained for L 1a Zn 2 (OAc) 2 by slow diffusion of pentane into a concentrated dichloromethane solution. The molecular structure of L 1a Zn 2 (OAc) 2 shows that both the zinc centers are pentacoordinated by the ligand framework ( Figure 1 and Figure   S9, Table 1); one within the "enclosed" phenolic pocket, and the other in the "open" cavity. A κ 2 -coordinated acetate ligand provides stabilization for the dinuclear complex by bridging between the two Zn centers. The zinc center enclosed in the small pocket is coordinated via two nitrogen centers, two phenolic oxygen centers, and one acetate oxygen in a distorted square pyramidal geometry (τ = 0.16). 29

Inorganic Chemistry
Article geometry (τ = 0.34) and is coordinated by two phenolic oxygen atoms, one bridging acetate oxygen, and a terminal κ 2 acetate. In the solid state, the two methoxy groups on the salen ligand do not coordinate to the zinc in the open pocket [Zn2−O25, 2.764(2) Å; Zn2−O27, 2.592(2) Å]. The distance between the two zinc centers is 2.9539(4) Å, which is small compared to other dinuclear catalysts (∼3.1 Å). 10a,b,19 A similar molecular structure has been reported for the closely related hydrate complex, L 1a Zn 2 (OAc) 2 ·H 2 O (Figure S10), where the two zinc centers adopt distorted square pyramidal geometries. 30 The zinc center occupying the enclosed pocket of L 1a Zn 2 (OAc) 2 · H 2 O has a similar coordination mode to that of L 1a Zn 2 (OAc) 2 . However, the zinc center in the open pocket is coordinated to just one phenolic oxygen, with one methoxy group, one water, and two acetate ligands (one bridging and one terminal) completing the coordination sphere of the second zinc. As a result, the Zn···Zn distance is longer [3.358(1) Å] but still in the range of common dinuclear catalysts for epoxide/CO 2 ROCOP. 10a,b,19 It is relevant to consider that the coordination of water by L 1a Zn 2 (OAc) 2 ·H 2 O indicates that the zinc coordination geometry within L 1a Zn 2 (OAc) 2 is likely to be altered in the presence of a Lewis donor, suggesting that under the polymerization conditions, the epoxide (CHO) could coordinate to the zinc center in the open pocket. Furthermore, the different coordination geometries of the zinc in the open pocket suggest that the methoxy groups may play a role in stabilizing the active sites during the polymerization.
In the solution state, the 1 H NMR spectra of imine complexes L 1a Zn 2 (OAc) 2 and L 2a Zn 2 (OAc) 2 confirm the two metals are coordinated, with all the resonances experiencing upfield shifts in comparison to the free ligands . A particularly significant shift was observed for the imine signal in CDCl 3 [from 13.84 ppm in L 1a H to 8.23 ppm for L 1a Zn 2 (OAc) 2 ( Figure S11) and from 14.14 ppm in L 2a H to 8.08 ppm for L 2a Zn 2 (OAc) 2 ( Figure S13)]. The broadness of the methylene resonances and the loss of the phenolic proton resonance also confirm the deprotonation of the ligand and the coordination of a zinc center. The appearance of two new singlet resonances at 2.00 and 1.97 ppm are assigned to the two distinct acetate coligands. Similarly, the 1 H NMR spectrum of the amine complex L 1b Zn 2 (OAc) 2 exhibits upfield resonances in comparison to those of the free ligand (Figures S15 and S16). In d 5 -pyridine, it can be seen that the N−H resonances are broad (4.28 ppm) and that a significant shift of the benzylic resonance occurs (from 4.01 to 4.19 ppm), indicating that the zinc center is located within the small pocket. The acetate resonance at 2.11 ppm is assigned to both the zinc-coordinated acetate ligands. DOSY analysis of L 2a Zn 2 (OAc) 2 in CDCl 3 ( Figure S17) and of L 1b Zn 2 (OAc) 2 in d 5 -pyridine ( Figure S18) shows that, in each case, all the 1 H NMR resonances possess the same diffusion coefficient and give a predicted molecular weight close to the theoretical one (456 g/mol for 614 g/mol and 729 g/mol for 633 g/mol, respectively), providing further confirmation of the dinuclear complex structures. To confirm the stability of L 2a Zn 2 (OAc) 2 under typical polymerization conditions, L 2a Zn 2 (OAc) 2 was heated to reflux for 16 h in d 8 -THF. The 1 H NMR spectra remained constant over time, indicating that this species is thermally stable under the polymerization conditions ( Figure S19). Unfortunately, this test could not be performed on L 1a Zn 2 (OAc) 2 and L 1b Zn 2 (OAc) 2 as they were not soluble even in hot THF.
An alternative synthetic route was investigated using a stepwise procedure via the synthesis of a monometallic intermediate, L 2b Zn (Scheme 4), followed by its reaction with Zn(OAc) 2 , and was expected to afford the dinuclear complex L 2b Zn 2 (OAc) 2 (Scheme 5). Accordingly, when L 2b H was deprotonated, using either Zn(OAc) 2 or Et 2 Zn in dry MeOH or THF, respectively, the formation of a white precipitate was observed. The solid-state structure of L 2b Zn was determined as both the THF and methanol solvates . Although the two solvates are very distinct crystallographically (with the structure of the THF solvate containing one independent C i -symmetric complex, whereas that of the methanol solvate contains two independent C isymmetric complexes), the conformations of all three 115.57(10) Zn (2)···O (27) 2.592 (2) O(32)−Zn (2) complexes are very similar. For simplicity, only the THF solvate will be discussed here, with full crystallographic details being provided in the Supporting Information. The molecular structure of L 2b Zn reveals that it is dimeric and possesses C i symmetry, with a central {Zn 2 O 2 } ring, where one phenolic oxygen from each ligand bridges between the two zinc metal centers ( Figure 2 and Table 2). The symmetry-related zinc centers adopt a trigonal bipyramidal geometry (τ = 0.70) and are bound to three phenolic oxygens in a fac fashion. The complexation of the zinc leads to the formation of sixmembered rings, providing stabilization for the dimeric complex. The methoxy groups do not coordinate to zinc, and so the open pocket remains available for the coordination of another metal center. Multinuclear ( 1 H, 13 C) NMR spectroscopy along with two-dimensional (2D) NMR and DOSY experiments suggest that the dimeric structure is retained in solution  To investigate whether the dimer (L 2b ) 2 Zn 2 can be cleaved by the coordination of a second metal, the reaction of (L 2b ) 2 Zn 2 with 1 equiv of Et 2 Zn was performed (Scheme 5). This reaction was selected because reactions involving Et 2 Zn are generally fast and because the formation of dinuclear complexes can be monitored by following the distinctive 1 H NMR signals of the Zn-ethyl groups. The 1 H NMR spectrum of the product, in C 6 D 6 , showed no trace of the unreacted dimer, and all resonances experienced an upfield shift, consistent with the incorporation of a second zinc center within the ligand scaffold ( Figures S30 and S31). Two different sets of characteristic Zn-ethyl resonances are observed, with two triplets at 2.05 and 1.79 ppm and the corresponding quartets at 0.94 and 0.61 ppm. These Zn-ethyl chemical shifts are significantly different than those of diethylzinc [1.11 ppm (t) and 0.12 ppm (q) in C 6 D 6 ], 31 confirming the coordination of the second zinc center within the ligand scaffold. 1 H DOSY NMR spectroscopy shows that all the resonances possess the same diffusion coefficient ( Figure S32). The estimation of the molecular weight based on the diffusion coefficient, using a calibration plot, gave a calculated molecular weight of 503 g/mol, which is consistent with the formation of a monomeric L 2b Zn 2 Et 2 product (561 g/mol). Nevertheless, block colorless crystals, obtained from a saturated THF solution of L 2b Zn 2 Et 2 at 22°C, corresponded to the starting dimer (L 2b ) 2 Zn 2 ; this finding suggests that the second metal is weakly coordinated within the ligand scaffold and that an equilibrium exists between L 2b Zn 2 Et 2 , Et 2 Zn, and (L 2b ) 2 Zn 2 . To emphasize the lability of the metal centers within this ligand system, 1 equiv of free ligand was added to a solution of L 2b Zn 2 Et 2 , in C 6 D 6 (Scheme 5). The instantaneous formation of a precipitate was observed, with concomitant release of ethane. The white precipitate was isolated, dissolved in d 8 -THF, and the resultant spectrum correspond to the dimer (L 2b ) 2 Zn 2 . Finally, when the dimer (L 2b ) 2 Zn 2 was reacted with 1 equiv of Zn(OAc) 2 at room temperature, the same product mixture was obtained as when the parent ligand was reacted with 2 equiv of Zn(OAc) 2 (Scheme 5).
Further investigations were performed to quantify the ratio of the reaction products L 2b Zn 2 (OAc) 2 , L 2b Zn 3 (OAc) 4 , and (L 2 b ) 2 Zn 3 (OAc) 2 . Complexes L 2 b Zn 3 (OAc) 4 and (L 2b ) 2 Zn 3 (OAc) 2 were independently synthesized by the reaction of L 2b H with 3 equiv or 1.5 equiv of Zn(OAc) 2 in THF, respectively (Scheme 6). Colorless block crystals of L 2b Zn 3 (OAc) 4 suitable for X-ray diffraction were obtained by slow diffusion of pentane into a saturated dichloromethane solution of L 2b Zn 3 (OAc) 4 ( Figure 3, Table 3, and Figure S33). The molecular structure reveals that Zn1 and Zn2 each possess an octahedral coordination sphere, while Zn3 is tetrahedral. Zn1 occupies the enclosed phenolic pocket and is coordinated by both phenolic oxygen centers and both amine nitrogens, with two bridging acetate ligands completing its coordination sphere. Zn2 occupies the open phenolic pocket and is Scheme 4. Reaction of L 2b H with 1 equiv of Et 2 Zn and the Formation of the Dimeric Structure (L 2b ) 2 Zn 2 (i) 1 equiv of Et 2 Zn in THF, 22°C, 16 h or 1 equiv of Zn(OAc) 2 in MeOH, 22°C, 16 h. Scheme 5. Reaction of (L 2b )Zn 2 with Et 2 Zn and Zn(OAc) 2 (i) 1 equiv of Et 2 Zn in THF, 22°C, 16 h (ii) 1 equiv of L 2b H in C 6 D 6 . (iii) 1 equiv of Zn(OAc) 2 in THF, 22°C, 16 h.   (1) 1.925 (2) O (1) (21) 79.11 (7) Zn (1)···Zn (1) 3.1557 (5) O (1) Figure S34). Although the acetate resonance overlaps with the NH and methylene ligand resonances, the integration gives good agreement with the presence of four acetate coligands coordinated to the metal centers. In addition, the DOSY spectrum shows that all the resonances possess the same diffusion coefficient, suggesting that the trizinc structure is retained in C 6 D 6 solution ( Figure  S36). The estimation of the complex molecular weight, based on the diffusion coefficient, gave a calculated molecular weight of 725 g/mol, which is in close agreement with the calculated value (802 g/mol). The product is also stable, under reflux conditions in THF solution, over a period of 16 h ( Figure S37). The reaction of 1.5 equiv of Zn(OAc) 2 with L 2b H gave a convoluted 1 H NMR spectrum in C 6 D 6 ( Figure S38). No signals from the unreacted ligand, dimer (L 2b ) 2 Zn 2 , or the trimetallic species L 2b Zn 3 (OAc) 4 were observed, suggesting that a new product is formed. DOSY spectroscopy reveals that the new resonances all possess the same diffusion coefficient ( Figure S39), corresponding to a molecular weight of 978 g/ mol, which is in close agreement with the value for (L 2b ) 2 Zn 3 (OAc) 2 (1063 g/mol). Although crystals of (L 2b ) 2 Zn 3 (OAc) 2 suitable for X-ray diffraction experiments could not be obtained, it was possible to grow crystals for the closely related trizinc bis(chloride) complex, (L 2b ) 2 Zn 3 Cl 2 ( Figure S40), providing further support for the formation of a trizinc complex coordinated by the ancillary ligands. 32 Scheme 6. Reaction of L 2b H with (i) 1.5 equiv and (ii) 3 equiv of Zn(OAc) 2 (i) 1.5 equiv of Zn(OAc) 2 in THF, 22°C, 16 h; (ii) 3 equiv of Zn(OAc) 2 in THF, 22°C, 16 h.   (2) 3.2936(9) Zn(1)···Zn (2) 3.0546 (9) Zn (3)···Zn (1) 3.6005 (8)

Article
Comparison of 1 H NMR spectra of L 2b Zn 3 (OAc) 4 and (L 2b ) 2 Zn 3 (OAc) 2 with the spectrum of products obtained from the reaction of L 2b H with 2 equiv of Zn(OAc) 2 ·2(H 2 O) suggests that L 2b Zn 3 (OAc) 4 and (L 2b ) 2 Zn 3 (OAc) 2 are the two major products ( Figure S41). DOSY spectroscopy of the mixture provides further support for this observation, as two major species are observed with diffusion coefficients that correspond exactly to the values obtained for the independently prepared complexes L 2b Zn 3 (OAc) 4 and (L 2b ) 2 Zn 3 (OAc) 2 ( Figures S23, S36, and S39). A third product is also present, albeit in very low concentration, which is likely to be the dinuclear complex L 2b Zn 2 (OAc) 2 . These findings suggest that using this ligand, the formation of the dinuclear complex is not favorable and that metal rearrangement occurs to form L 2b Zn 3 (OAc) 4 and (L 2b ) 2 Zn 3 (OAc) 2 . This was further supported as L 2b Zn 3 (OAc) 2 was isolated as a white precipitate from the reaction of L 2b H and 2 equiv of Zn(OAc) 2 in dry THF at −78°C. As the dinuclear complex could not be isolated, the L 2b Zn 2 (OAc) 2 mixture, L 2b Zn 3 (OAc) 4 , and (L 2b ) 2 Zn 3 (OAc) 2 were each tested within polymerization studies.
Polymerization Catalysis. The series of dizinc complexes were tested as catalysts for the ROCOP of CO 2 and CHO, at 1 bar of pressure of CO 2 and 80°C (Table 4). As other dizinc catalysts are active under these conditions, the reaction provides a useful benchmark for comparison. 11f,13a,18 All catalysts displayed a low activity; therefore, high catalyst loadings were required to convert CHO and CO 2 to PCHC. In particular, complexes containing the L 1 ligand backbone [L 1a Zn 2 (OAc) 2 , and L 1b Zn 2 (OAc) 2 ] showed negligable TOF values (∼0.1 h −1 ; Entries 1 and 2) at a catalyst loading of 1 mol % versus CHO. These low activities are attributed to the poor solubility of the catalyst in CHO. With the L 2 backbone, the L 2b Zn 2 (OAc) 2 mixture displayed improved solubility in hot CHO and showed slightly higher TOF values of 2 h −1 (Entry 4). However, after the first 4.5 h of reaction, a white precipitate formed, and the TOF decreased to 0.3 h −1 . Similarly, L 2b Zn 3 (OAc) 4 was active during the first 6 h of polymerization (TOF = 3 h −1 , Entry 6) with similar activity to that of the trizinc macrocycle catalyst reported previously (Entry 10); 11f however, the polymers have only low carbonate linkage content, with 45% of ether linkages. Again, formation of a white precipitate was observed after 6 h, and the activity of the catalyst decreased (TOF 1 h −1 after 24 h, Entry 7). The white solid was isolated, and its 1 H NMR spectrum showed catalyst decomposition ( Figure S42). This observation suggests that the trizinc complex L 2b Zn 3 (OAc) 4 decomposes during the polymerization reaction, most likely to form the dimer (L 2b ) 2 Zn 2 and Zn(OAc) 2 , which could explain the low activity observed. Under these conditions, the catalysts could only produce oligomers with low molecular weights. However, slightly higher activities (TOF of 2 and 1 h −1 ) were obtained with the more stable L 2a Zn 2 (OAc) 2 (Entry 3) and (L 2b ) 2 Zn 3 (OAc) 2 (Entry 8). The imine complex L 2a Zn 2 (OAc) 2 also showed an improved CO 2 uptake compared to the rest of the series, producing almost perfectly alternating copolymers of low molecular weight (1500 g/mol). Despite the low activities recorded for all the catalysts, the observations from this set of experiments suggest that complexes containing the L 1 backbone tend to have lower activity compared to those with the L 2 backbone, probably due to their lack of solubility in neat CHO at high loadings.
Previously, it has been reported that some dizinc complexes show activities that are dependent on the CO 2 pressure; indeed, such effects were usually observed for complexes coordinated by flexible ligands. 17 Therefore, the catalysts were also tested at 30 bar of CO 2 pressure. In all cases, an improvement in the catalytic activity was observed, as all the complexes produced PCHC at much lower catalyst loading (0.1 mol % catalyst vs CHO). On the one hand, the activity of L 1a Zn 2 (OAc) 2 , L 1 b Zn 2 (OAc) 2 , the L 2 b Zn 2 (OAc) 2 mixture, and (L 2b ) 2 Zn 3 (OAc) 2 were still very low (Table 5, Entries 1, 2, 4, and 6, respectively). On the other hand, L 2a Zn 2 (OAc) 2 (Entry 3) and L 2b Zn 3 (OAc) 4 (Entry 5) showed a significant enhancement in their catalytic activities (TOF = 44 and 9 h −1 , respectively) and polymer selectivities. The best catalyst, L 2a Zn 2 (OAc) 2 , has comparable activity to other metal salen catalysts 4g,20e and to the dinickel salen complex that was previously reported by Ko and co-workers (TOF = 56 h −1 ). 26 At 80°C, L 2a Zn 2 (OAc) 2 formed PCHC with a very high carbonate linkage content and with much better polymer selectivities (M n = 18 800 g/mol, Đ = 1.6). These results show that the more flexible L 2 backbone gave catalysts with higher activity than those with the L 1 backbone.
The series of catalysts was also tested for the ROCOP of PA and CHO. In 2014, Lu et al. reported that L 1a Zn 2 (OAc) 2 ·H 2 O could polymerize maleic anhydride and CHO when a cocatalyst such as N,N-dimethylaminopyridine was present. 30 In this case, all the catalysts could fully polymerize PA, in 16 h, without the need for a cocatalyst (Table 6). While the imine complexes afforded the corresponding polyesters (PE) with high selectivity and no ether linkages, the amine complexes provided the polymers with a greater proportion of ether linkages. A possible explanation for this trend is that the more electron-donating imine decreases the Lewis acidity of the metal center, thereby decreasing the likelihood of successive ring opening two CHO molecules. Comparing the catalyst activities, L 1a Zn 2 (OAc) 2 (Entry 1) gave activities (TOF = 70 h −1 ) similar to some of the best catalysts based on chromium, cobalt, or aluminum salen/salophen complexes, 16a all of which require the use of a cocatalyst. However, L 2a Zn 2 (OAc) 2 (Entry 3) was significantly more active (TOF = 198 h −1 ) than the majority of other known catalysts for this reaction, under similar conditions. 33 Both Terpolymerization of PA/CHO/CO 2 . The promising results of L 2a Zn 2 (OAc) 2 toward CHO/CO 2 and PA/CHO ROCOP prompted the investigation of the terpolymerization of PA/CHO/CO 2 . The reaction was performed at 100°C and 30 bar of CO 2 pressure with a 100/1000 mixture of PA/CHO. The reaction was monitored by taking aliquots after 2 and 18 h. After 2 h, 1 H NMR analysis showed selective polyester formation, with the appearance of a distinctive resonance at 5.06 ppm (74% conversion) and with no traces of PCHC observed ( Figure S43). 1 H NMR analysis after 18 h showed full conversion of PA (with loss of the resonances at 7.84 and 7.95 ppm) and the formation of PCHC, shown by the increase of intensity of the resonance at 4.64 ppm. Between 2 and 18 h, a clear increase in the molecular weight was observed, from 3598 to 7085 g/mol, along with conservation of the polydispersity index (1.20). 13 C NMR and DOSY analysis of the crude polymer suggests that the PE and PCHC blocks are connected, as only two carbon signal can be found in the carbonate region (162.7 and 166.6 ppm); both sets of resonances have a diffusion coefficient of −9.81 × 10 −10 m 2 /s ( Figures S44 and  S45). This evidence suggests that L 2a Zn 2 (OAc) 2 can be used to synthesize poly(ester-block-carbonates) from a one-pot reaction through a PA/CHO/CO 2 terpolymerization.

■ CONCLUSIONS
A series of dinuclear zinc salen complexes that are active catalysts for carbon dioxide/cyclohexene oxide and phthalic anhydride/cyclohexene oxide ROCOP were investigated. These studies highlight the influence of the ligand structures on the catalytic activity and the stability of the dinuclear complexes. Ligands containing imine moieties resulted in complexes that were more stable than the amine analogues. Consequently, a mixture of products, deriving from metal disproportionation, was obtained with the more flexible amine L 2b H ligand, highlighting its complex solution coordination chemistry. The catalytic studies indicate that complexes with the more flexible 2,2-dimethyl propylene backbones are more active than those with the trans-1,2-cyclohexylene backbones. Further, the complexes with imine substituents showed much better selectivities, affording perfectly alternating copolymers, whereas complexes with amine substitutents gave higher etherlinkage content. The best catalyst, L 2a Zn 2 (OAc) 2 , combines imine donors with a flexible 2,2-dimethyl propylene backbone. It shows higher activity than other dizinc salen catalysts for carbon dioxide/CHO copolymerizations and higher activities than other reported catalysts for PA/CHO copolymerizations. L 2a Zn 2 (OAc) 2 was also used to successfully afford poly(esterblock-carbonate) from the one-pot terpolymerization of PA/ CO 2 /CHO, providing further evidence that this simple system is selective and is able to synthesize polymers with more complicated structures.

■ EXPERIMENTAL SECTION
All solvents and reagents were obtained from commercial sources (Aldrich and Merck) and used as received unless stated otherwise. The solvents THF, toluene, and hexane were predried over activated molecular sieves (THF) or potassium hydroxide (toluene and hexane) then refluxed over sodium/benzophenone. All dry solvents and reagents were stored under nitrogen and degassed by several freeze− pump−thaw cycles. When stated, manipulations were performed using a double-manifold Schlenk vacuum line under nitrogen atmosphere or a nitrogen-filled glovebox. Cyclohexene oxide was dried over calcium hydride, filtered, purified by fractional distillation prior to use, and stored under an inert nitrogen atmosphere. Phthalic anhydride was purified by dissolving in benzene, filtering off impurities, recrystallizing from chloroform, and then subliming. Research-grade carbon dioxide was used for copolymerization studies.
NMR. 1 H, 13 C, and 2D NMR (COSY, HMQC) spectra were recorded using a Bruker AV 400 MHz spectrometer at 298 K (unless otherwise stated). DOSY analysis was performed using a Bruker AV 500 MHz spectrometer at 298 K (see the DOSY NMR section in the Supporting Information for more details).
MALDI-TOF MS. MALDI TOF analysis was performed on Micromass MALDI micro MX spectrometer. The matrix used was trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenylidene]malononitrile, and sodium acetate was used as an additive, when necessary. The samples were prepared as follows: 10 mg/mL THF solutions of the complex, matrix, and additive were separately prepared. Then, 20 μL of the complex and 20 μL of the matrix solution were mixed, along with 10 μL of the additive solution, if required. This mixture (2 μL) was then spotted on the MALDI plate and allowed to dry.
X-ray Diffraction. Data were collected using an Agilent Xcalibur 3E diffractometer, and the structures were refined using the SHELXTL and SHELX-2013 program systems. The crystallographic data table of all the compounds can be found in the Supporting Information (Table  S4).
Elemental Analysis. Elemental analysis was determined by Stephen Boyer at London Metropolitan University.
Size-Exclusion Chromatography. Two Mixed Bed PSS SDV linear S columns were used in series, with THF as the eluent, at a flow rate of 1 mL/min, on a Shimadzu LC-20AD instrument at 40°C. Polymer molecular weight (M n ) was determined by comparison against polystyrene standards. The polymer samples were dissolved in SEC-grade THF and filtered prior to analysis.
Typical Procedure for the Amination of o-Vanillin (L 1a H and L 2a H). The appropriate diamine [(±)-trans-1,2-diaminocyclohexane (1.87 g, 0.016 mol) or 2,2-dimethyl-1,3-propanediamine (1.67 g, 0.016 mol)] in MeOH (5 mL) was added dropwise to a pale yellow solution Typical Reduction Procedure for the Synthesis of L 1b H and L 2b H. The reduction was performed in situ following the synthesis of L 1a H or L 2a H in methanol solution, by slow addition of 3.5 equiv of NaBH 4 (Caution! Exothermic). The reaction mixture was stirred for 2 h at 22°C. Water was then added until a white precipitate was obtained. The suspension was left to stir for 16 h before the product was isolated via filtration as a white powder. (Yields: L 1b H, 6.0 g, 97% and L 2b H, 6.0 g, 97%) L  ,8.078;N 7.48%. Found: C,67.19;H,7.90;N 7.49%.
Synthesis of L 2b Zn 3 (OAc) 4 . L 2b H (400 mg, 1.06 mmol) was dissolved in 20 mL of THF, and Zn(OAc) 2 (391 mg, 2.14 mmol) was subsequently added as a solid. After the mixture stirred for 16 h at ambient temperature, all solvent was removed under reduced pressure
Synthesis of L 2b Zn 2 Et 2 . Under a nitrogen atmosphere, (L 2b ) 2 Zn 2 (200 mg, 0.46 mmol) was suspended in dry THF (7 mL). A solution of ZnEt 2 (56 mg, 0.456 mmol) in dry THF (3 mL) was added dropwise into the previous solution to obtain a colorless solution. The reaction mixture was stirred at ambient temperature for 16 h, and then all volatiles were removed under reduced pressure to give L 2b Zn 2 Et 2 as a white powder (yield: 75%, 190 mg). 1 H NMR (400 MHz, C 6 D 6 , 298 K) δ 6.74 (dd, J = 7.1, 2.1 Hz, 2H, m-Ph), 6.65 (m,4H, CO 2 /CHO Polymerization Reactions at 1 bar of CO 2 Pressure. The zinc catalyst (1 equiv) was suspended in CHO (1 mL, 100 equiv) under nitrogen atmosphere in a Schlenk tube charged with a stirrer bar. The Schlenk tube was then connected to a CO 2 line, where the reaction mixture was degassed three times and then heated to 80°C under 1 bar of CO 2 pressure. The crude polymer was obtained by evaporation of the remaining CHO under reduced pressure. Polymers were dissolved in THF and then purified by precipitation from MeOH to yield a white powder. CO 2 /CHO Polymerization Reactions at 30 bar of CO 2 Pressure. The zinc catalyst (1 equiv) was suspended in CHO (6 mL, 1000 equiv) under a nitrogen atmosphere in a Schlenk tube charged with a stirrer bar. The Paar reactor was purged five times with CO 2 . The reaction mixture was then transferred into the reactor at 1 atm of CO 2 pressure. The pressure was then adjusted to 30 bar at 80°C , unless otherwise stated. The crude polymer was obtained by evaporation of the remaining CHO under reduced pressure. Polymers were purified by precipitation from a THF solution in MeOH to yield a white powder.
PA/CHO Polymerization Reactions. The zinc catalyst (1 equiv) was suspended in CHO (1 mL, 800 equiv) under nitrogen atmosphere in screw cap vial charged with a stirrer bar. PA (100 equiv) was added, and the vial was closed, sealed, and heated to 100°C. The crude polymer was obtained by evaporation of the remaining CHO under reduced pressure. Polymers were dissolved in THF then purified by precipitation from MeOH to yield a white powder.
Terpolymerization of PA/CO 2 /CHO. L 2a Zn 2 (OAc) 2 (36.6 mg, 0.059 mmol, 1 equiv) and PA (877 mg, 5.93 mmol, 100 equiv) were suspended in CHO (6 mL, 59.3 mmol, 1000 equiv) under nitrogen atmosphere in Schlenk tube charged with a stirrer bar. The Parr reactor was purged five times with CO 2 . The reaction mixture was then transferred into the reactor at 1 atm of CO 2 pressure. The pressure was then adjusted to 30 bar at 100°C. Aliquots were taken after 2 and 18 h by a depressurization−repressurization procedure at 100°C. The crude polymer was obtained by evaporation of the remaining CHO under reduced pressure. Polymers were purified by precipitation from a THF solution in MeOH to yield a white powder.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.5b02233.
The complete experimental procedures and characterization data for all new compounds and copolymers. (PDF) Single-crystal X-ray structure information in CIF format.