Use of titanocalix[4]arenes in the ring opening polymerization of cyclic esters†

The known dichloride complexes [TiCl2LIJO)2IJOR)2] (type I: R = Me (1), n-Pr (2) and n-pentyl (3); LIJOH)2IJOR)2 = 1,3-dialkyloxy-p-tert-butylcalixij4]arene), together with the new complexes {[TiLIJO)3IJOR)]2IJμCl)2}·6MeCN (R = n-decyl (4·6MeCN)), and [TiIJNCMe)ClIJLIJO)3IJOR))]·MeCN (type II: R = Me, 5·MeCN) are reported. Attempts to prepare type II for R = n-Pr and n-pentyl using [TiCl4] resulted in the complexes {[TiLIJO)3IJOn-propyl)]2IJμ-Cl)IJμ-OH)} 6·7MeCN and {[TiLIJO)3IJOn-pentyl)]2IJμ-Cl)IJμ-OH)}·7.5MeCN (7·7.5MeCN), respectively; use of [TiCl4IJTHF)2] resulted in a co-crystallized THF ring-opened product [TiIJNCMe)IJμ3-O)LIJO)4TiClIJOIJCH2)4Cl)]2–2ijTiClIJNCMe)IJLIJO)3IJOn-Pr))]·11MeCN (8·11MeCN). The molecular structures of 2·2MeCN, 4·6MeCN, and 5·MeCN together with the hydrolysis products {[TiLIJO)3IJOR)]2IJμCl)IJμ-OH)} (R = n-Pr 6·7MeCN; n-pentyl, 7·7.5MeCN, 9·9MeCN); R = n-decyl 10·8.5MeCN) and that of the ring opened product 8·11MeCN and the co-crystallized species [Ti2IJOH)ClIJLIJO)3IJOR))]ijLIJOH)2IJOR)2] ·2.85IJC2H3N)·0.43IJH2O) (R = n-pentyl, 11·2.85IJC2H3N)·0.43IJH2O)) are reported. Type I and II complexes have been screened for their ability to act as catalysts in the ring opening polymerization (ROP) of ε-caprolactone (ε-CL), δ-valerolactone (δ-VL), ω-pentadecalactone (ω-PDL) and rac-lactide (r-LA), both with and without benzyl alcohol present and either under N2 or in air. The copolymerization of ε-CL with δ-VL and with r-LA has also been investigated. For the ROP of ε-CL, all performed efficiently (>99% conversion) at 130 °C over 24 h both under N2 and in air, whilst over 1 h, for the type I complexes the trend was 3 > 2 > 1 but all were poor (≤12% conversion). By contrast, 5 over 1 h at 130 °C was highly active (85% conversion). At 80 °C, the activity trend followed the order 5 ≈ 4 > 3 > 2 > 1. For δ-VL, at 80 °C the activity trend 5 ≈ 4 > 1 > 2 > 3 was observed. ROP of the larger ω-PDL was only possible using 5 at 130 °C over 24 h with moderate activity (48% conversion). For r-LA, only low molecular weight products were obtained, whilst for the co-polymerization of ε-CL with δ-VL using 5, high activity was observed at 80 °C affording a polymer of molecular weight >23000 Da and with equal incorporation of each monomer. In the case of ε-CL/r-LA co-polymerization using 5 either under N2 or air, the polymerization was more sluggish and only 65% conversion of CL was observed and the resultant co-polymer had 65 : 35 incorporation. Complex 5 could also be supported on silica, however this system was not as active as its homogeneous counterpart. Finally, the activity of these complexes is compared with that of three benchmark species: a di-phenolate Ti compound {TiCl2IJ2,2′-CH3CHij4,6-(t-Bu)2C6H2O]2)} (12) and a previously reported NO2-containing titanocalix[4]arene catalyst, namely cone-5,17-bis-tert-butyl-11,23dinitro-25,27-dipropyloxy-26,28-dioxo-calixij4]arene titanium dichloride (13), as well as [TiIJOi-Pr)4]; the parent calixarenes were also screened.


Introduction
Metallocalixarenes have attracted a significant amount of interest in the catalysis field, and have found application for a variety of processes. 1 In terms of titanocalixarenes, which date back to the 1980s, 2 Frediani, Sémeril et al. employed the 1,3-di-n-propyloxycalixĳ4]arene, 1,3-LĲOH) 2 Ĳn-PrO) 2 , containing complex [1,3-LĲO) 2 Ĳn-PrO) 2 TiCl 2 ], in conjunction with MAO (methylaluminoxane), to afford ultrahigh molecular weight polyethylene. 3 The active species, as determined by 1 H NMR spectroscopy, was determined to be a titanium methyl cation, as observed in earlier work by Proto et al. 4 Subsequently, Taoufik, Bonnamour et al. extended these studies and investigated the effect of varying the 1,3-dialkyloxy R groups (for R = methyl, ethyl, n-propyl and i-butyl) on the catalytic activity during ethylene polymerization. In the same study, a couple of 1,2-dialkoxycalixĳ4]arene derivatives were also prepared, namely the methyloxy and a complex containing a chelating siloxide SiMe 2 , together with a number of depleted calix [4]arene complexes bearing 1,2-or 1,3-titanium dichloride motifs. 5 Whilst the behavior of the 1,2-and 1,3-systems was different, it was determined that within each series, increasing the number of alkyloxy groups present was detrimental to the catalytic activity. On variation of the R group from methyl to isobutyl in the 1,3-systems, there was no clear structural activity trend. A number of other, less welldefined, titanium calix[n]arenes have also been employed for ethylene polymerization. 6 We also note that this type of titanocalix [4]arene has been grafted onto silica and employed in olefin epoxidation by the group of Katz. 7 With regard to the ring opening polymerization (ROP) of cyclic esters, reports using titanocalix [n]arenes are scant. Frediani,Sémeril et al. employed the complex {[1, )LĲO) 2 Ĳn-PrO) 2 -1,3]TiCl 2 }, under solvent-free conditions, for the well-controlled ROP of lactide. 8 The resultant polymer was highly isotactic, whilst the addition of n-butanol led to an increase in both the rate of polymerization and transfer with the monomer. The same group also employed complex [1,3-LĲO) 2 Ĳn-PrO) 2 TiCl 2 ] for the ROP of rac-lactide (r-LA), initiating the catalyst either by use of microwave radiation or heat. 9 Although the rate of polymerization was enhanced by using microwaves, this was to the detriment of control. More recently, McIntosh et al. reported preliminary studies on the use of the complex [Ti 4 L 2 ĲO) 8 ĲOn-Pr) 8 ĲTHF) 2 ] (where L 2 ĲOH) 8 = p-tert-butylcalixĳ8]arene) as a catalyst for the ROP of r-LA at 130°C. 10 We have also investigated the use of metallocalix [n]arenes for the ROP of cyclic esters such as ε-caprolactone, and have reported how, for a series of tungstocalix [6 and 8]arenes, different sized rings and their associated conformations can drastically affect the catalytic activity. 11 Molybdo-and tungstocalix [4] arenes were found to be less active. 12 Herein, we have screened several titanocalix [4]arenes, namely [TiCl 2 LĲO) 2 -ĲOR) 2 ] (type I: R = Me (1), n-Pr (2) and n-pentyl (3)), the dimeric compound {[TiLĲO) 3 ĲOR)] 2 Ĳμ-Cl) 2 } (R = n-decyl (4)), and [TiĲNCMe)ClĲLĲO) 3 ĲOR))]·MeCN (type II, R = Me, 5·MeCN, R = n-propyl 6 and n-pentyl 7) for the ROP of ε-CL, δ-VL and r-LA (the copolymerization of ε-CL and r-LA was also investigated) (Chart 1).
In the case of R = n-decyl, prolonged standing at 0°C afforded large red prisms suitable for X-ray crystallography. The molecular structure (CCDC 1954693) is shown in Fig. 2, with selected bond lengths and angles given in the caption. The structure of 4 is a chloro-bridged dimer {[TiLĲO) 3 ĲOR)] 2 Ĳμ-Cl) 2 } (R = n-decyl), in which each titanium center is best described as distorted octahedral. The asymmetric unit contains two half molecules both on inversion centers as well as six molecules of crystallization (MeCN). There are weak C-H⋯O hydrogen bonds between the first CH 2 group of the n-decyl chain and a phenolate oxygen on the symmetry related calix [4]arene. The two half molecules differ in two respects. Firstly, the n-decyl chains have different conformations at their tails and one is disordered here, while the other is not. Secondly, the number of MeCN molecules in the calixarene cavities is different.
The structure (CCDC 1954694) is shown in Fig. 3, with selected bond lengths and angles given in the caption. The structure is non-merohedrally twinned via a 180°rotation about the direct axis (1 0 0). There are two titanocalix [4]arene complexes (related via mirror symmetry) and two acetonitrile molecules in the asymmetric cell. The coordination geometry about the distorted octahedral titanium center comprises cis chloride and acetonitrile ligands together with a monomethoxycalix [4]arene, the latter adopting a 'down, down, down, out' conformation. The acetonitriles of crystallization reside with their methyl groups inside the calix [4]arene cavities.
We note that a route to monofunctionalised calix [4]arenes via removal of an ether functionality using TiCl 4 has been reported by Floriani and coworkers. 13 Floriani et al. identified CH 3 Cl in their reaction and suggested this was the result of capture of the leaving group by the Cl − nucleophile. We also noted dealkylation when studying the air stability of the type I complex with R = n-pentyl. 14 In the case of 5, this can be viewed as the acetonitrile trapped titanium intermediate enroute to the formation of the monoalkoxycalix [4]arene. Surprisingly, on extension of this methodology to the systems derived from LĲOH) 2 ĲOR) 2 (R = n-propyl or n-pentyl), the corresponding monochloride complexes [TiĲNCMe)ClĲLĲO) 3 -ĲOR))]·MeCN (R = n-propyl and n-pentyl) were not accessible (Scheme 2b). It was postulated that the de-alkylation of the calix [4]arenes bearing n-propyl and n-pentyl moieties does not proceed as per their methoxy analogue due to the nature Scheme 1 Synthesis of complexes 1-4. 5    Fig. 5, with selected bond lengths and angles given in the caption. Each titanium center is distorted octahedral with the coordinated provided by a mono-pentoxy calix [4] arene and bridging hydroxyl and chloro ligation. In the packing, the calixarene cavities approach each other forming fairly large solvent accessible volumes, leading to the observed disorder in the MeCN of crystallization.
This type of reaction was further complicated when [TiCl 4 ĲTHF) 2 ] was employed as the metal precursor. 15 In this case, on attempting to isolate the n-propyl analogue of 5, the orange/red co-crystallized complex [TiĲNCMe)Ĳμ 3 -O)LĲO) 4 TiClĲOĲCH 2 ) 4 Cl)] 2 -2ĳTiClĲNCMe)ĲLĲO) 3 ĲOn-Pr))]·11MeCN (8·11MeCN) was isolated in moderate yield. The molecular structure (CCDC 1954697) is shown in Fig. 6, with selected bond lengths and angles given in the caption. The asymmetric unit comprises half a tetrametallic complex, one monometallic complex and 5.5 MeCNs. The tetrametallic complex lies on a centre of symmetry, with octahedral metal centres bound by either a calix [4]arene which adopts a cone conformation with a metal-bound MeCN in the cavity and a μ 3 -oxo {Ti (2) and Ti(2A)}, or in the case of Ti(3) and Ti(3A) by an oxygen from each L, the two μ 3 -oxos and a chlorinated butoxy ligand (OĲCH 2 ) 4 Cl). The calix [4]arenes have lost all of their lower rim bound n-propyl groups. By contrast, in the monometallic complex, the calixarene adopts a 'down, down, down, out' conformation and retains an n-propyl group.   We have also investigated the air/water stability of a number of these systems, given that often during catalysis small amounts of water are present, which can lead to often unwanted transesterification processes. Furthermore, studies have revealed that the water content of hydrophilic monomers such as ε-CL can varying greatly with temperature and time. 16 Thus, given the likely presence of water in our ROP studies, we re-visited the type I complex with R = n-pentyl, for which we had previously structurally characterized the complex {[TiLĲO) 3 ĲOR)] 2 Ĳμ-Cl)Ĳμ-OH)}·71/ 4MeCN, as well as the and n-decyl system. 14 Following initial exposure to air (2 h), work-up was conducted under an inert atmosphere of nitrogen (Scheme 3a). In the case of R = n-pentyl, the complex isolated, namely {[TiLĲO) 3 ĲOn-pentyl)] 2 Ĳμ-Cl)Ĳμ-OH)}·9MeCN 9·9MeCN, differed only in the degree of solvation from 7·7.5MeCN and from that which we reported previously, see ESI, † Fig. S2 (CCDC 1954689); all geometrical parameters were very similar.
In the case of the n-decyl system, crystallization from acetonitrile again afforded a chloro/hydroxyl-bridged complex, namely {[TiLĲO) 3 ĲOn-decyl)] 2 Ĳμ-Cl)Ĳμ-OH)}·8.5MeCN (10·8.5MeCN). The molecular structure (CCDC 1954690) of 10·8.5MeCN is shown in Fig. 7, with selected bond lengths and angles given in the caption; alternative views are given in the ESI † (Fig. S3). Each titanium is distorted octahedral, and is bound by a calix [4]arene ligand bearing only one decyl chain. One decyl chain, on O(1), is fully extended, but the other, on O(5), is only straight until the 6th C atom, at which point it is somewhat folded. Molecules of 10 pack into layers in the b/c plane, but are generally well-separated with MeCNs of crystallization in crevices between molecules. MeCNs at N(3), N(7) and N(10) reside mostly with the calixarene cavities and with Me groups furthest inside, the others lie exo.
We observed that if the entire procedure above was conducted in air, then the result was precipitation of the parent calixarene ligand. However, we found that conducting a controlled hydrolysis reaction of the type I complex with R = n-pentyl, i.e. addition of 0.5 equivalents of H 2 O to a toluene solution of [TiCl 2 LĲO) 2 ĲOn-pentyl) 2 ] under reflux, followed by extraction into acetonitrile and then crystallization (still under nitrogen) led to the isolation of orange, plate-like crystals in moderate yield (Scheme 3b). The molecular structure (see Fig. 8) revealed a co-crystallized species (11).
One Ti 2 complex plus half the non-coordinated cocrystallized bis-n-pentyl ligand, plus the solvent of crystallization in the asymmetric unit (CCDC 1954691). Each Ti adopts a distorted octahedral conformation. The bis-npentyl ligand lies on a center of symmetry at the body center of the unit cell. The non-coordinated calix [4]arene molecule retains two phenol hydrogens which form intramolecular   H-bonds to the neighbouring oxygens. This molecule adopts a 1,2-alternate conformation. The existence of the cocrystallized bis-n-pentyl ligand demonstrates that this is the species added to the reaction and that the loss of one n-pentyl group occurs upon coordination. The coordinated calix [4]arenes are essentially eclipsed when the complex molecule is viewed end-on (see Fig. S4, ESI †). For catalytic comparison purposes, a titanium complex {TiCl 2 Ĳ2,2′-CH 3 CHĳ4,6-(t-Bu) 2 C 6 H 2 O] 2 )} (12) bearing a diphenolate ligand derived from the diphenol 2,2′-CH 3 CHĳ4,6-(t-Bu) 2 C 6 H 2 OH] 2 was synthesized according to the procedure reported by Aida et al. (Scheme 4).
Plate-like, orange crystals suitable for X-ray analysis were obtained from a saturated hexane solution of the complex upon standing at room temperature for 2 days. The molecular structure of 12 (CCDC 969053) is shown in Fig. 9. A tetrahedral Ti 4+ is coordinated to two chlorides and two diphenolate oxygens. The complex is chiral but only one enantiomer is present. The fold angle between rings C(1) > C(6) and C(17) > C(22) is 64.63Ĳ7)°. No disorder or solvent of crystallization are observed.

ROP screening
We have examined the ability of the complexes prepared herein to act as catalysts for the ROP of ε-CL both under nitrogen and in air; the ROP of ε-CL under air when using the titanium tetraalkoxides TiĲOi-Pr) 4 and TiĲOn-Bu) 4 has been reported. 16 Firstly, looking at the series [TiCl 2 LĲO) 2 ĲOR) 2 ] (type I: R = Me (1), n-Pr (2) and n-pentyl (3)), at 130°C over 24 h with a ratio of 500 : 1 and in the presence of two equivalents of benzyl alcohol, all three are efficient catalysts and afford >99% conversion (runs 1-4, Table 1). Complex 1 as its MeCN solvate exhibits the best control, whilst in all cases, observed molecular weights are far lower than calculated values suggesting extensive transesterification is occurring (Scheme 5). 8 Over 1 h (runs 5-7), some differentiation is possible with highest conversion achieved using the pentyl catalyst system 3. Interestingly, if the ROP runs are conducted under air (runs 8-10), the catalysts remain efficient with conversions >99%, with comparable molecular weights for the polymers isolated using 1 and 2 as those obtained under N 2 , though that from 3 was somewhat lower.
The control in air was generally better than that observed under N 2 . On lowering the temperature to 80°C, over 24 h (runs [14][15][16], complex 3 was again the most active, with both systems 1 and 3 affording polymers with good control (M w / M n = 1.1). In the case of system 2, only oily oligomers were isolated (360 Da). Over 1 h at 80°C (runs 20-22), the systems were inactive. When runs were conducted in the absence of benzyl alcohol (runs 29-31), all systems were again active (>99% conversion), with observed molecular weight higher than for comparable runs in the presence of external alcohol. This could be ascribed to the lack of the chain-transfer to the co-activator (see Scheme 5). Over 1 h at 130°C in the absence of alcohol, the systems were inactive (runs [26][27][28]. It was noteworthy that complete conversion was achieved in the ROP conducted in air in the presence of complex 4·MeCN (run 11). The same outcome was achieved at 80°C over 24 h (run 17). Although only 11% conversion was obtained after 1 h (run 17), the complex proved to be more active than its congeners 1-3.
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

View Article Online
Catal. Sci. Technol., 2020, 10, 1619-1639 | 1625 This journal is © The Royal Society of Chemistry 2020 24 h under N 2 (runs 18,19) conversions were >99% and over 1 h (run 24) conversions were >37%. The analysis of the polymer terminal groups was performed by means of 1 H NMR spectroscopy. The spectra of the PCLs synthesized in the presence of the co-activator, both under nitrogen and in air, exhibited signals for the aromatic and benzylic protons of the benzyloxy-end group (δ = 7.35 and 5.10 ppm, respectively) in a 5 : 2 ratio (see the ESI, † Fig.   S5 and S6, respectively). As expected, such signals were not observed in the spectrum of the polymer isolated in the absence of benzyl alcohol (see the ESI, † Fig. S7). In all cases, the triplet corresponding to the terminal -CH 2 OH group (δ = 3.65 ppm) was also observed. For the polymers isolated with BnOH, the ratio between the areas of the benzyl-and -C H 2 OH groups was found to be far from the theoretical value (5 : 2 : 2 vs. 5 : 2 : 6 and 5 : 2 : 10 for the run under N 2 and air, Scheme 3 Formation of compounds 9-11 from the parental complexes 3 and 4.  (3) S10a). Noteworthy, the formation of cyclic polyesters occurring in the absence of BnOH has recently been reported. 19 A distribution of α-hydroxyl-ω-(carboxylic acid)-terminated polymers became predominant for higher fractions (see ESI, † Fig. S10b and c). Interestingly, no Cl-terminal groups were detected, suggesting the inability of the Ti-Cl species to promote ROP reactions. 17,18 Also in this case, the active species was thought to be a Ti-OH compound derived from the hydrolysis of the pre-catalyst.

δ-Valerolactone (δ-VL)
For the ROP of δ-VL (Table 2), using 1-3, at 80°C over 24 h (runs 1-3) only moderate activity was observed with conversions ≤50% (activity followed the trend 1 > 2 > 3), but with good control and observed molecular weights much lower than calculated values. Higher conversion (78%) was obtained in the presence of 4 (run 4). Interestingly, the molecular weight observed was close to the calculated value.
In the case of the reaction performed over 1 h at 80°C, systems 1-3 were inactive, while only 4% conversion was observed with complex 4 (runs 7-9 and 10, respectively). In the case of [TiĲNCMe)ClĲLĲO) 3 ĲOMe))]·MeCN (5·MeCN), good activity (81% conversion) was observed at 80°C in the presence of either one or two equivalents of BnOH over 24 h (runs 5 and 6). Over 1 h, 5·MeCN exhibited low activity ≤14%. Moderate activity (62% conversion) was also observed on conducting the reaction under air over 24 h (run 7). Also in this case, the 1 H NMR spectra of selected polymer samples highlighted the presence of benzyl-and -CH 2 OH terminal groups (see the ESI, † Fig. S11 and S12, for the run performed under N 2 and in air, respectively). (11) (most H atoms are omitted for clarity). Selected bond lengths (Å) and angles (°): TiĲ1)-OĲ5) 2.396 (2), TiĲ1)-OĲ6) 1.8106Ĳ19), TiĲ1)-OĲ7) 1.839 (2), TiĲ1)-OĲ8) 1.804 (2), TiĲ1)-OĲ9) 1.975 (4), Scheme 4 Synthesis of the di-phenolate complex 12. 17 Attempted polymerization of r-LA was less successful. At 80°C over 24 h, none of the systems proved to be active (runs 1-4, Table 3). On increasing the temperature to 130°C , almost complete conversion was achieved with complexes 1-3 and 5 (runs 5-7 and 9, respectively), while moderate activity was observed in the case of 4 (65% conversion, run 8). In the case of systems 1-3 and 5, the molecular mass observed was lower than the calculated values and there was poor control. It is noteworthy that a monodisperse polymer (M w /M n = 1.04) with molecular mass observed close to the calculated value was obtained with complex 4. It was hypothesized that in the presence of such catalysts, the rate of transesterification processes was lower than that of the chain growth allowing for the occurrence of a living polymerization. By contrast, 1-3 and 5 are subject to increased transesterification versus the other systems. Interestingly, complete conversion was achieved in air by using complex 5 (run 10). None of the systems was active over 1 h (runs [11][12][13][14][15][16]. The syndiotactic bias was determined by 2D J-resolved 1 H NMR spectroscopy, investigating the methine area (ca. 5.2 ppm) of the spectra (see ESI, † Fig. S13-S17). 20 The peaks were assigned to the corresponding tetrads according to the literature reports. 21 While complexes 1 and 2 afforded almost heterotactic polymers (P r 0.46 and 0.42, respectively), isotactic materials were isolated in the case of systems 3, 4, and 5 (Chart 2).

ω-pentadecalactone (ω-PDL)
In the case of the larger cyclic ester ω-PDL, only catalyst 5 was successful, achieving moderate activity at 130°C over 24 h (run 1, Table 4). Low activity (11% conversion) was observed when the reaction was carried out in air (run 3, Table 4). In both cases, molecular weights lower than the expected values and compatible with oligomeric species (12-13 units) were observed. Also in this case, the benzyl end-group was observed by 1 H NMR spectroscopy on the polymer sample (see the ESI, † Fig. S18).
Interestingly, the triplet corresponding to the -CH 2 OH group was not observed. However, a rigorous analysis of the terminal groups could not be performed due the presence of the signals of residual monomer which could not be separated from the polymer. Co-polymerization of ε-CL with δ-VL At 80°C using 5, high activity (99% conversion) was observed over 24 h affording a polymer of molecular weight >23 000 Da ( Table 5, run 1). The ratio CL/VL was found to be 1 : 1, as highlighted by 1 H NMR spectroscopy (see the ESI, † Fig. S19). Interestingly, a similar outcome was observed in the case of the reaction performed in air at higher temperature (  Fig. S20). 22 Based in the integrations of these signals, the number-average sequence length was found to be 2.22 and 1.82 for CL and VL, respectively, compatible with a "random-type" co-polymer (randomness degree, R = 0.90, see ESI, † eqn (S1)-(S3)). The thermal properties of the co- This journal is © The Royal Society of Chemistry 2020    Fig. S21). No transitions were observed between 23 and 100°C, confirming the irregular co-monomer distribution within the polymer chain leading to lower melting points. This is in agreement with the observations recently disclosed by Hu et al. 22a Co-polymerization of ε-CL with r-LA At 130°C using 5 and a 10 : 1 ratio of ε-CL/r-LA, high activity (99% conversion) was observed over 24 h affording a polymer of molecular weight >5000 Da (Table 6, run 1) which comprised mostly PCL (see the ESI, † Fig. S22). Use of a 1 : 1 ratio of CL/r-LA afforded a copolymer of molecular weight >20 500 Da (Table 6, entry 3), for which the 1 H NMR spectra indicated only 65% of CL was converted (see the ESI, † Fig. S23). A similar outcome was obtained by performing the reaction in air (Table 6, run 5). The co-polymer compositions was determined by analyzing the carbonyl range of the 13 C NMR spectrum. 23 The LA/CL ratio was found to be 75 : 35 and the average sequence length was 3.04 and 2.42 for CL and LA, respectively (ESI, † Fig. S24 and eqn (S3) and (S4)). Noteworthy, no peaks corresponding to the CL-LA-CL triad at 171.1 ppm was observed. Such signals arise from the transesterification of the cleavage of the lactyl-lactyl bond in the lactidyl unit. 23 Concerning the thermal properties of the co-polymer, no transitions accountable to CL blocks were detected on the DSC curve of the sample (ESI, † Fig. S25). Nevertheless, rather intense endothermic peaks were observed between 150 and 170°C, suggesting the existence of crystalline PLAmicrodomains. 23a Comparison with other Ti-based catalysts Since complex 5 proved to be the most performing catalyst for the ROP of cyclic esters, its activity was compared to that of other Ti-based species, namely the novel di-phenolate complex 12 and the nitro-containing titanocalix [4]arene species cone -5,17-bis-tert-butyl-11,23-dinitro-25,27-dipropyloxy-26,28-dioxo-calixĳ4]arene titanium dichloride 13, previously reported by Toupet et al. (Chart 3). 8 These complexes, along with [TiĲOi-Pr) 4 ] were employed as catalysts in the ROP of different cyclic esters (Table 7). In all cases, 5 outperformed complexes 12 and 13, both in terms of conversion and polymer M n (cf. for example runs 1-3 and 5-7). Although [TiĲOi-Pr) 4 ] was shown to promote the ROP of ε-CL under solvent-free conditions, 16 it was found to be completely inactive under our reaction conditions (toluene, [monomer] = 0.9 M). Notably, only complex 5 proved to be active in the ROP of the larger ω-pentadecalactone.
The polymerization of ε-CL under solvent-free conditions was next considered (Table 8). By performing the reaction in the presence of the monochloride titanocalix [4]arene complex 5, 50% conversion was achieved within 10 minutes at 80°C (run 1). Under the same reaction conditions, no reaction was observed neither with the bi-phenolate complex 12 nor with the dichloro-species 13 (runs 2 and 3). Notably, 29% monomer conversion was obtained by using [TiĲOi-Pr) 4 ] (run 4). In order to explain the inactivity of 12 and 13, the inefficient activation of the dichloro-based species in the absence of the solvent was postulated. Differently, the presence of the labile acetonitrile ligand in complex 5 would allow for the prompt formation of the catalytically active species, even under solvent-free conditions. Moreover, higher performance of catalyst 5 over [TiĲOi-Pr) 4 ], both in terms of conversion and polymer M n , highlighted the beneficial effect of the calix [4]arene ligand on the ROP process.
Moreover, the performances of complexes 5, 12 and 13 were compared in the ROP of ε-CL under aerobic conditions (Table 9). At 130°C in the presence of 5 and 12, good and   1 and 2). Interestingly, full monomer conversion was observed by performing the reaction with complex 13 (run 3). Nevertheless, higher polymer M n were achieved in the presence of 5 (19 kDa vs. 11 kDa for 5 and 13, respectively). By carrying out the reaction for 24 h, complete monomer conversion was observed with both 5 and 13 (runs 4 and 5). The M n of the samples were found to be in a rather narrow range (spanning from 4600 to 5400) albeit better control was exhibited by the monochloride titanocalix [4]arene complex. Finally, given the interest in metal-free ROP catalysts, 20 the parent 1,3-dialkoxycalixarenes LĲOH) 2 ĲOR) 2 were also screened under the conditions herein. These metal-free species proved to be inactive.

Kinetics
From a kinetic study of the ROP of ε-CL using 1-5 and 12-13, it was observed that the polymerization rate exhibited first order dependence on the ε-CL concentration (Fig. 9, left), and the conversion of monomer achieved over 60 min was >75% (100% for 5 and 13). Fig. 10 indicates that the rate order is 13 > 5 > 12 > 3 > 2 > 1 > 4. This suggested that a labile acetonitrile ligand (for 5) or electron-withdrawing groups on the calixarene backbone (for 13) are beneficial. The induction period of ca. 10 min observed for complexes 1-4 could be ascribed to the longer time required for the formation of the catalytically active species (allegedly a Ti-bisĲbenzyloxide) compound) from a dichloro-precursor.
A kinetic study of the ROP of δ-VL using 1-5 and 12-13, again revealed first order dependence on the monomer concentration (Fig. 11, left). Conversions over 60 min were lower than observed for ε-CL with all ≤60% (ca. 70% for 13). The activity trend in this case revealed that 13 was the most active and then 4 = 5 >3 > 1 > 2.
The dependence of the M n and molecular weight distribution on the monomer conversion in the reactions catalyzed by 5/BnOH was also investigated (Fig. 12). For the ROP of ε-CL, the polymer M n was shown to increase linearly with the conversion, while the M w /M n was found to be rather constant at ca. 1.14 throughout the reaction (Fig. 12, left). A similar outcome was also observed in the reaction involving δ-VL (Fig. 12, right).
Finally, the kinetic behaviour of complex 5 in the ROP of ε-CL was compared to that of catalysts 12 and 13 both under inert and aerobic conditions (Fig. 13). In the presence of the calix [4]arene-based complexes 13 and 5, the reaction rate of the polymerization conducted in air was shown to be similar to that of the runs preformed under inert atmosphere. On the contrary, a detrimental effect was observed when the biphenolate complex 12 was employed in air compared to the reaction carried out under inert atmosphere.

Silica immobilization of complex 5
Complex 5 was immobilized on pre-treated silica by refluxing in toluene affording structure Si-5; the silica was heated at 200°C under dynamic vacuum for 2 h. X-ray photoelectron spectroscopy (XPS) analysis of the silica surface was performed Chart 3 Structure of the benchmark Ti-complex 13. 8 This journal is © The Royal Society of Chemistry 2020 (Fig. 14). The region area ratios of the C-1s to Ti-2p indicated a Ti content of 1.89% with respect to C. This was slightly lower than the calculated value (2.17%) and allegedly due to adventitious carbon bound to the surface. However, this evidence suggested that no decomposition occurred on binding the complex to the surface. Moreover, the Ti-2p peak was consistent with the presence of only one titanium species.

Polymerization of cyclic esters catalyzed by Si-5
The supported version of complex 5 was tested in the ROP of cyclic esters (Table 10). For each run, 50 mg of catalyst were employed. Based on the Ti content found by XPS analysis, the monomer/Ti ratio was found to be 400. Si-5 proved to be very active in the reaction involving ε-CL.
Complete conversion was observed after 24 h at 130°C both in the presence and in the absence of 1 equiv. of alcohol (runs 1 and 2). Interestingly, higher molecular mass and better control was observed in the experiment conducted in the absence of the co-catalyst. While high conversions were achieved with the homogeneous complex 5 after 1 h (85%, cf. Table 1, entry 13), lower activity was observed in the case of its supported version (runs 3 and 4). The catalyst proved to be active also in the ROP of δ-VL and rac-LA albeit with moderate activity (runs 5 and 6). While low control was observed for the former, narrow polydispersity was achieved for the latter (2.48 vs. 1.15, respectively). Unlike its homogeneous analogues, system Si-5 was not active in the ROP of the larger monomer ω-pentadecalactone (run 7).   This journal is © The Royal Society of Chemistry 2020 In conclusion, we have isolated and structurally characterized a number of titanocalix [4]arene complexes which are capable of the ROP of cyclic esters. In particular, we find that a mono-methoxy complex [TiĲNCMe)ClĲLĲO) 3 -   , which is thought to be due to the lability of the metal-bound acetonitrile under ROP conditions. Importantly, the ROP of both ε-CL and δ-VL can be conducted under air without loss of activity. It also proved possible to ROP the larger cyclic ester ω-PDL with moderate conversation (ca. 50%). Complex 5 was able to co-polymerize ε-CL and δ-VL with the same activity, while higher poly-lactide content was observed in the ε-CL/rac-LA co-polymerization. We have also investigated the air and moisture stability of some of these titanocalix [4] arenes, and have isolated and structurally characterized bridged OH − /Cl − species.

General
All manipulations were carried out under an atmosphere of dry nitrogen using conventional Schlenk and cannula techniques or in a conventional nitrogen-filled glove box. Hexane and toluene were refluxed over sodium. Acetonitrile was refluxed over calcium hydride. All solvents were distilled and degassed prior to use. IR spectra (nujol mulls, KBr windows) were recorded on a Nicolet Avatar 360 FT IR spectrometer; 1 H NMR spectra were recorded at room temperature on a Varian VXR 400 S spectrometer at 400 MHz or a Gemini 300 NMR spectrometer or a Bruker Advance DPX-300 spectrometer at 300 MHz. The 1 H NMR spectra were calibrated against the residual protio impurity of the deuterated solvent. Elemental analyses were performed by the elemental analysis service at the London Metropolitan University and in the Department of Chemistry, the University of Hull. The precursor [TiCl 4 ĲTHF) 2 ] was prepared by the literature method. 24 The pro-ligands LĲOH) 2 ĲOR) 2 were prepared as described previously. 25 Complex 13 was synthesized according to the literature procedures. 8
This article is licensed under a Creative Commons Attribution 3.0 Unported Licence.

View Article Online
Preparation of silica supported complex (Si-5) SiO 2 pretreatment: silica gel (60 Å, mesh 35-70 μm, 5.0 g) was heated at 200°C under vacuum for 2 h. Immobilization of the complex: a solution of 5 in toluene (0.07 g in 10 mL) was added to a suspension of the pretreated silica in toluene (500 mg in 10 mL). The mixture was stirred at reflux for 16 h and then allowed to cool down to room temperature. The mixture was filtered, and the solid residue was washed with warm toluene (2 × 10 mL) and dried under reduced pressure at room temperature for 2 h affording a yellow fine powder (0.46 g).

Kinetic studies
The polymerizations were carried out at 130°C under N 2 atmosphere. The molar ratio of monomer to catalyst to initiator was fixed at 500 : 1 : 2. At appropriate time intervals (10 minutes), 0.5 μL aliquots were removed and quenched with methanol. The solvent was removed in vacuo and the percent conversion was determined by 1 H NMR spectroscopy in CDCl 3 .

X-ray photoelectron spectroscopy
A small amount (∼1 mg) of powdered sample was pressed onto adhesive carbon tape. High resolution XPS core level measurements were performed with a Specs NAP-XPS (operating under UHV conditions), equipped with a SPECS Phoibos 150 NAP hemispherical analyser. XPS experiments were carried out using a high intensity monochromated Al-Kα source (1486.6 eV) operated at 14.5 kV and 25 mA. Detailed spectra were recorded using a pass energy of 30 eV. The energy scale of the spectrometer was calibrated to the C-1s at 284.8 eV.

Crystal structure determinations
Diffraction data for all crystal structures were collected at low temperature on modern diffractometers equipped with hybrid pixel array detectors and rotating anode X-ray sources. 27 Full details are given in Table 11, in the ESI, † and in the deposited cif files. Data were corrected for absorption and Lp effects. 27 The structures were solved using a dual-space, charge-flipping algorithm and refined on F 2 . 28,29 In common with many calixarene crystal structures there are were often some difficulties to overcome during the refinement. Some t Bu groups needed to be modelled with the methyl groups or the whole moiety split over two sets of positions. MeCNs of crystallization were often disordered and needed to be modelled with split positions, or as diffuse electron density via the Platon Squeeze procedure. 30 The number of MeCN molecules of crystallization should only be taken as approximate. Where disorder has been modelled, restraints were applied to anisotropic displacement parameters and the geometry of the affected and immediately adjacent atoms. Hydrogen atoms were placed in geometrically determined positions, except for those on hetero atoms in structures with good data, where coordinates were refined.  2·2MeCN, 4·6MeCN, 5·MeCN, 6·7MeCN, 7·7.5MeCN,  8·11MeCN and 12, respectively.

Conflicts of interest
There are no conflicts of interest.