Synthesis and Characterization of N,N,O-Tridentate Aminophenolate Zinc Complexes and Their Catalysis in the Ring-Opening Polymerization of Lactides

A series of aminophenolate ligands with various pendant groups and associated ethyl Zn complexes were synthesized and studied as catalysts for the ring-opening polymerization (ROP) of lactides (LAs). The thiophenylmethyl group (L4ZnEt) increased the catalytic activity more than the benzyl group (L1ZnEt) did, and 2-fluorobenzyl (L3ZnEt) and 2-methoxybenzyl (L2ZnEt) groups had the opposite effect. In addition, the LA polymerization mechanism proved by Nuclear Magnetic Resonance and Density Function Theory was that LA was attracted by H···O bond of an α-hydrogen of the LA molecule and the phenoxyl oxygen of the catalyst. After the dissociation of amino group from the Zn atom, the benzyl alcohol initiated LA without replacing the ethyl group of Zn complex. It is the first case where the ethyl group is regarded as a ligand and cannot be replaced by benzyl alcohol, and this information is very important for the mechanism study of ROP.


INTRODUCTION
Because petrochemical polymers such as polystyrene, polypropylene, polyethylene, and poly(vinyl chloride) can be produced easily and cheaply, they are widely used as disposable packaging materials. Because these petrochemical polymers need more than a hundred years to degrade into innocuous soil manure (Rochman et al., 2013), polymer pollution has become a serious problem (Romer, 2010;Romer and Tamminen, 2014;Ladewig et al., 2015;Zhang et al., 2017). The replacement of non-biodegradable polymers with biodegradable materials is therefore a popular field of research (Levis and Barlaz, 2011). Poly(lactide) (PLA) is a biopolymer designed to ameliorate pollution by petrochemical plastics. Owing to its biodegradability, biocompatibility, and permeability, PLA is commonly used for various purposes, such as humidity detection (Sun et al., 2007), MRI contrast agents (Patel et al., 2008), cell/tissue anti-adhesion (Lih et al., 2015), nanocomposites (Raquez et al., 2013), drug delivery (Khemtong et al., 2009), blood circulation (Ma et al., 2008), bone replacement (Simpson et al., 2007), and tissue engineering (Place et al., 2009). Ring-opening polymerization (ROP) by using metal complexes (Bellemin and Dagorne, 2014;Guillaume et al., 2015;Sarazin and Carpentier, 2015;Huang et al., 2016;Fuoco and Pappalardo, 2017;Redshaw, 2017) as catalysts is a common method for the efficient synthesis of PLA. Because no cytotoxic metal residue is required in PLA for the biomaterials, the use of a non-cytotoxic metal such as zinc (Williams et al., 2003;Romain and Williams, 2014;Yang et al., 2015;Binda et al., 2016;Ebrahimi et al., 2016;Thevenon et al., 2016;Wang et al., 2016) in lactide polymerization has been investigated widely. Ligands are crucial for the catalyst design because their catalytic activity can be increased. In a study of Zn complexes (Williams et al., 2003) bearing tridentate aminophenolate ligands, a high catalytic activity was observed during rac-lactide (rac-LA) polymerization, as shown in Figure 1A. As in previous studies, the tetradentate aminophenol ligands reacted with Zn[N(SiMe 3 ) 2 ] 2 and four coordinated Zn complexes with a non-coordinated fourth amino group were obtained, as shown in Figures 1B-D. According to the polymerization results of Figures 1B-E, the steric bulky substituents on the phenolate ring, the fourth coordinated amino groups, and chiral N-alkyl groups can increase the stereoselectivity of rac-LA polymerization, and maintain high catalytic activity. A survey of the coordination behavior of Zn complexes (Gao et al., 2013) revealed that five and six are the possible coordination numbers for these complexes, and even seven-coordinated Zn complexes were reported (Vaiana et al., 2007). It would be worthwhile to design the fourth coordinated group of tetradentate aminophenol ligands to interact with the Zn atom, and influence their catalytic activities. In this study, a series of aminophenol ligands and associated Zn complexes were synthesized, and their application in LAs polymerization was investigated.

Syntheses and Characterization
A series of N 1 -alkyl-N 2 ,N 2 -dimethylethylene-1,2-diamines was synthesized by using NaBH 4 to reduce 2-alkylideneamino-N,Ndimethylethylen-1-amines that were synthesized by condensing the aldehyde derivatives with dimethylethylenediamine in ethanol. All ligands L 1 -H-L 4 -H were prepared by refluxing a mixture of N 1 -alkyl-N 2 ,N 2 -dimethylethylene-1,2-diamine, para-formaldehyde, and 2,4-bis(α,α-dimethylbenzyl)-phenol ( Figure 2). All the ligands reacted with 1.1 equivalents of ZnEt 2 in THF at 0 • C to produce a moderate yield (74-83%) of Zn compounds after hexane washing. The Zn complex synthesis can be identified by the 1 H NMR spectrum. The peaks of the methylene groups of HO[( t Bu) 2 -Ph]CH 2 N and NCH 2 Ar from the 1 H NMR spectrum are singlet in ligands and doublet of doublets in Zn complexes, and the proton of PhO-H (10.43-10.15 ppm) disappeared after ZnEt 2 was added. The formulas and structures of the compounds were confirmed on the basis of 1 H and 13 C Nuclear Magnetic Resonance (NMR) spectra.

Ring Opening Polymerization of LAs
The catalytic activity of Zn complexes for L-LA and rac-LA polymerizations with benzyl alcohol (BnOH) as the initiator under a nitrogen atmosphere was investigated, and the results are given in Table 1. As shown by entries 1-4 in Table 1, all Zn complexes were active for L-LA polymerization at 25 • C, producing polymers with narrow polydispersity indexes (Ð, 1.08-1.13). L 4 ZnEt had the most controllability of the polymer molar mass with similar values of Mn cal , Mn NMR , and Mn GPC . However, the difference of catalytic activity of these Zn complexes was initially unclear. To determine the difference, the polymerization temperature was set at 100 • C and polymerization was terminated after 2 h (entries 5-8 in Table 1). The catalytic activity of L-LA polymerization was in the following order: L 4 ZnEt > L 1 ZnEt > L 3 ZnEt > L 2 ZnEt. The results revealed that the thiophenylmethyl group of L 4 ZnEt increased the catalytic activity of Zn complexes more than the benzyl group of L 1 ZnEt did, and that 2-fluorobenzyl and 2-methoxybenzyl groups decreased the catalytic activity of the complexes. From the polymer data (entries 1-8 in Table 1), the values of Mn NMR , and Mn GPC were smaller than that of Mn cal , and this phenomenon may be attributed to the transesterification. As shown by entries 9-12 in Table 1, the catalytic rates for rac-LA polymerization were the same as those for L-LA polymerization. The rac-PLA polymerized by L 4 ZnEt showed the highest selectivity. According to the literature, electronic donating substituents increased the catalytic activity of Zn catalysts (Chen et al., 2006Huang et al., 2006;Chuang et al., 2011;Fliedel et al., 2014a,b). In our case, the pendant-chelating substituents do not coordinate with the Zn atom in the solid state, but they increase the catalytic activity. Crystal data imply that the fourth coordinated groups, such as thiophenylmethyl, 2-methoxybenzyl, and 2-fluorobenzyl groups, still do not coordinate to Zn atom, just like the behavior of the Zn complexes (1B-1D) shown in Figure 1; however, the low stereoselectivity of rac-LA polymerization was also observed in our case. These phenomena are ascribed, possibly, to the size of the pendant substituents. The reasons will be discussed with DFT results later. In addition, the Ð values of these rac-PLAs were higher than that of PLAs (Table 1), and it may be attributed to more transesterification at a higher temperature in longer polymerization time.
The controllability of polymer molar mass by using L 4 ZnEt as a catalyst was investigated, and the results are given in Table 2    according to Equation (1)   constant and incorporated into k 1 (k 1 = k prop [BnOH] y ). The variable k obs is first assumed to be as described by Equation (2). When k obs is plotted against [L 3 ZnEt], a k 1 values of 15.034 (M −1 h −1 ) for x = 1 is obtained (Figure 6). The variable k obs is then assumed to be as described by Equation ( Table 2).

Proposed Mechanism
In order to realize the mechanism of LA polymerization by using Zn catalysts, the 1 H NMR study of the reaction of L 4 ZnEt with one equivalent BnOH was investigated as shown in Figure S19.
In Figure S19D, the 1 H NMR spectrum revealed that L 4 ZnEt did not react with BnOH. This phenomenon was very surprising because most alkyl Zn complexes could react with alcohol to form Zn alkoxide complexes (Chuang et al., 2011;Fliedel et al., 2014a,b;Chen et al., 2015).   [LA] = 0.02 M in d 8 -toluene (0.5 mL) at 25 • C) was monitored by 1 H NMR as shown in Figure 9, Figure S20, and the ethyl group was always at 0.25 ppm from the beginning to the end of the polymerization (Figure S20). When these 1 H NMR spectra revealed that the BnOH did not replace the ethyl group of the Zn complex in the LA polymerization process, we were curious about how BnOH initiated the monomer. To understand the polymerization mechanism, the interactions between the catalysts, the initiator, LA, and relative free energies of the intermediates of the catalytic reaction were studied using the DFT calculation.

DFT Calculations: Mechanistic Study of LA Polymerization by Using Zn Complexes Bearing Aminophenolate Ligands as Catalysts
In our DFT calculations, the cumyl group at the 4 position of the phenol ring L 2 ZnEt was simplified to a hydrogen atom, since it is far from the zinc catalytic center. The initiator was replaced by a methanol molecule because different alcohol molecules often show similar activity (Chang et al., 2015). The mechanism is shown in Figure 10.
When the catalyst reacted with LA, the LA molecule was initially stably bonded in an open pocket formed between the R 1 group on the 2 position of the phenol ring and the pendant R 2 group to form intermediate I as shown in Figure 11. The major interaction to stabilize this structure was found to be a unique hydrogen bond between an α-hydrogen of the LA molecule and the phenoxyl oxygen of the catalyst with a H···O distance of 2.204 Å. I then went through two possible pathways to form key catalytic intermediate IV. One of possible pathways is that the NMe 2 group dissociated from the Zn center to produce an empty space to be coordinated by a carbonyl oxygen atom with a Zn-O bond (2.383 Å) to form intermediate II. After that, the initiator came into the complex to form intermediate IV. Another possible pathway is that the initiator (ROH) bonded to the Zn complex with a coordination bond between its oxygen atom and the Zn center (2.508 Å) and a hydrogen bond between the hydroxyl hydrogen atom of ROH and phenoxyl oxygen of catalyst to form intermediate III. The NMe 2 group subsequently left the Zn atom to make a rearrangement to generate IV. In intermediate IV, the O atom of the initiator was close to the carbonyl group of LA with a distance of 2.855 Å to facilitate the ring opening reaction. Moreover, the hydrogen bond between the initiator and the phenoxyl O atom of Zn catalyst stabilized this transient intermediate structure (IV) and also activated the initiator. After the LA ring opening, a new hydroxyl group formed at one end of the polymer (or oligomer) and interacted with phenoxyl oxygen through a hydrogen bond with a coordination between the zinc center and the carbonyl group on the other end to form intermediate V. The NMe 2 group then came back to recoordinate on the zinc center to assist the leaving of the carbonyl group of the polymer to form intermediate VI and re-activated the catalyst. Finally, another LA monomer came in to reform intermediate III to continue the polymerization reaction or to    From the thermochemical data of our DFT studies as shown in Figure 12 and Table 3, it can be found that to generate IV from I, the pathway through II was a little more favored, since the relative free energy of II + MeOH was slightly lower than that of III (2.276 kcal mol −1 lower). This is because the molecular freedom (entropy) of II + methanol was much higher than that of III. After forming IV, the ring opening reaction to form V makes its relative free energy increase, due to the fact that the bulky polymer chain was restricted on the catalyst. However, after the carbonyl group dissociated from the zinc atom, the relative free energy of VI decreased dramatically. This large free energy decrease is a strong driving force to make the catalyst open the lactide ring.
From our DFT studies, some important experimental phenomena can be well-explained. First, the stability of the basic Et group on the Zn atom of the catalyst can be explained by noting that the acidic proton of the initiator (RO-H) was retained on the phenoxyl oxygen of catalyst by hydrogen bond and the crowded structures of the catalysts (III, IV, and VI). These protected the Et group on the Zn atom from the acid proton. Second, the relative activities of the four catalysts can also be rationalized, possibly by the size of the pocket for accepting LA on the catalysts found in intermediate I, besides the electronic effects on the zinc center. The size of the pocket decides the easiness of LA entering the catalyst center, and the smaller pocket size also makes monomer coordination slightly different between D-LA and L-LA to reveal the higher stereselectivity in rac-LA polymerization. The pocket sizes of the four catalysts according to the sizes of the pendant groups and the percent buried volume (%V bur ) of the ligands with a sphere radius of 7 Å measured from the optimized L'ZnEt structures (Supporting Information) are in the following order: L 4 ′ ZnEt (32.6%) > L 1 ′ ZnEt (32.8%) > L 3 ′ ZnEt (33.4%) > L 2 ′ ZnEt (33.6%) (Falivene et al., 2016). This order is coincident with that of their catalytic activities and opposed to stereselectivity.

CONCLUSION
In this study, a series of aminophenolate ligands with various pendant groups and associated ethyl Zn complexes were synthesized to investigate the effect of pendant groups on the catalytic activity of the complexes during LA polymerization. The thiophenylmethyl group (L 4 ZnEt) increased the catalytic activity more than the benzyl group (L 1 ZnEt) did, and 2-fluorobenzyl (L 3 ZnEt) and 2-methoxybenzyl (L 2 ZnEt) groups showed the opposite effect. The new LA polymerization mechanism was proven by NMR study and DFT calculation, which showed that LA was attracted by the H···O bond of an α-hydrogen of the LA molecule and the phenoxyl oxygen of the catalyst. After the dissociation of the amino group from the Zn atom, the benzyl alcohol initiated LA without replacing the ethyl group of the Zn complex. It is the first case that the ethyl group is regarded as a ligand and cannot be replaced by benzyl alcohol. This information provides researchers with another possible mechanisms of ROP by using alkyl zinc complexes as catalysts.

EXPERIMENT SECTION General
Standard Schlenk techniques and a N 2 -filled glove box were used throughout the isolation and handling of all the compounds. Solvents, L-LA, and deuterated solvents were purified prior to use. N,N-dimethylethylenediamine, 2,4-bis(α,α-dimethylbenzyl)phenol, sodium borohydride, benzaldehyde, 2-methoxybenzaldehyde, 2-fluorobenzaldehyde, 2-thiophenecarboxaldehyde, and diethyl zinc were purchased from Aldrich. 1 H and 13 C NMR spectra were recorded on a Varian Unity Inova-600 (600 MHz for 1 H and 150 MHz for 13 C) or a Varian Mercury-400 (400 MHz for 1 H and 100 MHz for 13 C) spectrometer with chemical shifts given in ppm from the internal tetramethylsilane or the central line of CDCl 3 . Gel permeation chromatography (GPC) measurements were performed on a Jasco PU-2080 plus system equipped with a RI-2031 detector using THF (high-performance liquid chromatography grade) as an eluent (flow rate 1.0 mL/min, at 40 • C). The chromatographic column was Phenomenex Phenogel 5 µm 103 Å, and the calibration curve used to calculate Mn(GPC) was produced from polystyrene standards. The GPC results were calculated using the Scientific Information Service Corporation (SISC) chromatography data solution 3.1 edition.

Synthesis of L 2 ZnEt
We used a method similar to that for L 2 ZnEt, except L 2 -H was used to replace L 1 -H. L 2 ZnEt followed the procedure of L 2 ZnEt. Yield: 2.51g (78%). 1