Recent advances in metallo/organo-catalyzed immortal ring-opening polymerization of cyclic carbonates

Abstract

The ring-opening polymerization of cyclic carbonate monomers derived from biomass feedstocks constitutes an attractive way to prepare polycarbonates that are potentially useful as high-tech materials or commodity “bioplastics”. This process can be catalyzed by inherently different systems ranging from simple basic organocatalysts, simple Lewis acidic metallic salts such as triflates, or more sophisticated discrete metallo-organic complexes derived from oxophilic metals (zinc, yttrium). In this perspective, the most recent achievements in this chemistry are reviewed. The quite different performances within reach of those catalytic systems, which are assessed in terms of intrinsic reactivity, robustness and regioselectivity towards dissymmetric monomers, are highlighted.

---

1. Introduction and general considerations

Aliphatic polyesters, including polycarbonates, are synthetic polymers which are nowadays attracting increasing interest. In light of the benefits that they offer through their functionality (fit for purpose specialty polymer materials) and environmental overall performances (biocompatibility, (bio)degradability), various applications or product segments are currently exhibiting high growth rates. These range from specialty biomaterials like drug/gene delivery systems (nanovectors) or tissue/bone engineering substrates (orthopedic screws and pins, (resorbable) sutures, scaffolds for soft tissue regeneration, implants such as stents, etc.) to commodity polymer materials such as carrier bags, compostable waste bags, film packaging for food, rigid packaging such as containers or bottles, or biodegradable mulching films.^1,2

Polyesters/polycarbonates are typically prepared from the condensation of a diol and a diacid/phosgene or dialkyl carbonate, respectively. Although industrially used and suitable with a large choice of monomer feedstocks, this step-growth polymerization suffers from major drawbacks: high temperatures are required and the leaving group/side-product (generally water) constantly has to be removed; also, long reaction times are required for the reaction to be completed. Such operating conditions unfortunately favor undesirable side-reactions, which in turn result in the lack of chain length control, no tuning of chain-end functions and limited access to copolymers.² In comparison, the ring-opening polymerization (ROP) of cyclic esters/carbonates, most usually driven by the relief of the ring strain,³ is a much better controlled chain-growth polymerization process allowing the synthesis of well-defined and finely tuned (co)polymers. ROP investigations are being extensively developed academically and are now gaining significant industrial considerations, in particular with poly(L-lactide) (PLLA) which occupies a prominent place with a worldwide production of ca. 150 000 metric tons p.a.⁴ In contrast to polycondensations, ROP reactions require milder operating parameters (lower temperature and/or shorter reaction time) to reach high molar mass polymers displaying controlled molecular characteristics (well-defined and tuned microstructure, regio/stereo-selectivity, end-group fidelity, predictable M̄_n and narrow M̄_w/M̄_n values) with limited side-reactions. Tailored (co)polymers are thus accessible with various topologies: linear, star, branched, block, random, etc.² Cationic, anionic and pseudo-anionic “coordination–insertion” ROP processes have been developed, with the latter catalytic route providing better controlled and “living” polymerizations. The catalytic systems required for ROP are based on either organometallic, inorganic or organic derivatives. The compliance requirements of the catalytic system ideally include cheap, easily accessible, robust and non-toxic component(s) (metal, ligands and co-catalyst), high activity, productivity and selectivity, and minimized promotion of undesirable side-reactions like transesterification or epimerization (in the case of chiral monomers/polymers). Of upmost interest for a variety of reasons (high productivity, low residues), the optimized catalytic system should be fully effective when used in minimized loadings. In this regard, we have recently revisited the “immortal” ring-opening polymerization (iROP) pioneered by Inoue,⁵ to extend and generalize it to the truly catalytic polymerization of cyclic esters such as lactides, β-lactones and carbonates.⁶

Such an “immortal” ROP process is clearly differentiated from a “classical” ROP (Scheme 1). In a “classical” ROP mediated by an organometallic catalyst, the number of growing polymer chains is given by the number of initiating functions, that is at most the number of nucleophilic groups (typically alkoxide moieties) attached to the metal center; i.e., it cannot exceed the largest oxidation state of the metal, namely four for the elements usually involved in such a coordination–insertion approach. Thus, in a “classical” ROP, the molar mass of the polymer is, in principle (i.e., for a “living” polymerization), anticipated from the monomer-to-initiator ratio. This therefore significantly limits the productivity of this route. Another issue associated with this approach is that impurities eventually present in the reaction medium (originating from the monomer, solvent, etc.) may irreversibly deactivate part (or all) of the initiating alkoxide moieties, which will therefore affect the final molar mass of the polymer. The other major weakness is the possible traces of metallic/ligand residues recovered in the final polymeric material, which could be a significant issue for a variety of applications (especially microelectronics, biomedical, food packaging).

Scheme 1 Schematic illustration of “classical” ROP vs. “immortal” ROP.

In comparison, “immortal” ROP makes use of a true catalyst (the metal or organic species) and an initiator acting also as a chain transfer agent (CTA). The latter initiator/CTA is a protic source, typically an alcohol or more seldom an amine, introduced in a large excess relative to the catalyst species. In such iROP processes,⁵ provided initiation is faster than propagation, termination reactions are avoided, and quite importantly, chain transfer reactions between dormant and active macromolecular chains are reversible and fast relative to propagation, the “immortal” polymerization is also “living”. Thus, the polymer molar mass is dictated only by the initial monomer-to-initiator/CTA ratio, independently of the catalytic loading. So, the larger the amount of initiator/CTA introduced, the lower the molar mass of the polymer. Nevertheless, greater loadings in monomer units still allow the preparation of high molar mass (in practical terms, desirable) polymers, thereby improving significantly the productivity relatively to “classical” ROP processes. Therefore, in iROP procedures, the catalyst is introduced in minor loadings already at the early stage of the iROP, i.e. in catalytic amounts vs. both the monomer and the initiator/CTA, and it is present during the propagation also in catalytic amounts vs. the polymer chains as well. As a consequence, catalytic residues are even less likely to be found in the final precipitated polymers, thus allowing to alleviate possible issues revolving around toxicity or contaminations. The major challenge in iROP yet remains the identification of a catalyst that stays stable relative to the large excess of initiator/CTA, while still active.⁶

Whereas polyesters such as polylactones and especially polylactides have been broadly studied,⁴ polycarbonates have been less commonly investigated.² However, within the last decade, they have drawn a significant regain of attention, especially in light of their possible bio-sourced nature (Scheme 2). In particular, trimethylene carbonate (1,3-dioxane-2-one, hereafter abbreviated TMC), the most common cyclic carbonate, can be derived from glycerol, a by-product formed during the production of biodiesel originating from vegetable oils and animals fats and cheaply available in large volumes, and more generally from bio-sourced 1,3-propanediol (Scheme 2).⁷ Other six- and seven-membered ring cyclic carbonates have also been synthesized from the biomass. Among these, α-methyl-substituted TMC (rac-4-methyl-1,3-dioxan-2-one, α-MeTMC) can be similarly prepared via 1,3-butanediol derived from ethanol,⁸ whereas 2,2-dimethoxytrimethylene carbonate (2,2-dimethoxypropane-1,3-diol carbonate, TMC(OMe)₂, a masked hydroxy-TMC equivalent potentially leading to hydrophilic polycarbonates) originates from the naturally occurring 1,3-dihydroxyacetone (DHA; Scheme 2).⁹ Likewise, the more recently developed seven-membered ring carbonates, 4-methyl- and 5-methyl-1,3-dioxepan-2-one (α-Me7CC and β-Me7CC, respectively), can be similarly formed upon cyclization of the corresponding α,ω-diols, namely 1,4-pentanediol and 2-methyl-1,4-butanediol, themselves arising from the bio-sourced levulinic and itaconic acids, respectively.¹⁰ Actually, glycerol, ethanol, levulinic or itaconic acids all belong to the US Department Of Energy's initial “top 10” or latter “top 10-revisited” value chemical opportunities from carbohydrate-based biorefineries.¹¹ While DHA is not included in these classifications, it can be produced from glycerol.^7c Noteworthy, all these chemicals presently under consideration are issued from non-edible biomass, allowing a non-controversial economical valorization.

Scheme 2 Some bio-sourced six- and seven-membered cyclic carbonates studied in (i)ROP.

Therefore, when such polymers are being eco-conceived thanks to sustainable polymerization approaches such as bulk (i.e. solvent-free) iROP, they are sometimes referred to as “green” polymers or “bioplastics” that could also provide a safe alternative to the production of commodity polymers presently originating from depleting fossil resources. In this context, our ongoing efforts aim at developing original poly(carbonate)s potentially derived from the biomass.

This contribution is an account of our most recent results in the field of iROP of cyclic carbonates derived from biomass resources, with a variety of catalysts including metallo-organic, metallic and organic examples, highlighting their respective catalytic performance/efficiencies and overall rewards. Possible outlooks of this work are also given.

2. Controlled iROP of six- and seven-membered cyclic carbonates by (metallo)-organic catalyst systems

Various catalytic systems composed of metallo-organic, simple metallic salts or organic compounds have been successfully evaluated in association with benzyl alcohol (BnOH) acting as initiator and CTA in the iROP of TMC,^6,12–14 TMC(OMe)₂,¹⁵ α-MeTMC,⁸ α-Me7CC and β-Me7CC¹⁰ (Scheme 2, 3).⁶ The dissymmetry in α-MeTMC and α-/β-Me7CC monomers offers opportunities for assessing the regio/stereo-selectivity of the catalysts, to access eventually microstructurally-controlled materials with tunable properties.^8,10

Scheme 3 Metal- and organo-catalyzed iROP of cyclic carbonates.

A first class of catalysts that were investigated in depth are the discrete metallo-organic complexes such as the zinc-{β-diiminate} amido complex [(BDI^iPr)Zn(N(SiMe₃)₂)] ((BDI^iPr) = 2-((2,6-diisopropylphenyl)amido)-4-((2,6-diisopropylphenyl)-imino)-2-pentene) or the yttrium-{amino-alkoxy-bis(phenolate)} amido complex [(ONOO^tBu)Y(N(SiHMe₂)₂)(THF)] (Scheme 4).^8,10,12,15 The selection of these catalyst systems was based on their prior performances in the ROP of lactides as well as of the more reluctant to polymerize β-butyrolactone, which include high activity and regio- and stereo-selectivity.^16,17 In particular, while the yttrium compound proved to be highly stereoselective for the formation of the syndiotactic corresponding poly(hydroxyalkanoate)s, the zinc-based initiator afforded atactic polymers.¹⁶

Scheme 4 Some metal-based and organic catalysts used in iROP of cyclic carbonates.

On the other hand, robust catalytic systems were highly desirable to better withstand drastic industrial applications with technical-grade monomer batches. Therefore, other classes of catalysts were also examined, including simple metallic salts, notably some metallic triflates, (M(OTf)₃, M = Sc, Al, Bi),^6,13 and some organocatalysts, essentially of the amine, guanidine or phosphazene type such as 4-N,N-dimethylaminopyridine (DMAP), 1.5.7-triazabicyclo-[4.4.0]dec-5-ene (TBD) or 2-tert-butylimino-2-diethylamino-1,3-dimethylperhydro-1,3,2-diazaphosphorine (BEMP), respectively (Scheme 4).^6,12c,14,15 These were picked for their known resistance (definitely better than that of the metallo-organic complexes) towards impurities as well as for their recognized catalytic ability in several organic transformations, including polymerization.^18,19

Although the former metallo-organic systems operate through a coordination-insertion mechanism (CIM), the latter metal triflates usually act through an activated-monomer mechanism (AMM);^4a,6 for organo-catalyzed ROP, a variety of mechanistic pathways are encountered: nucleophilic/electrophilic activation of the monomer, basic activation of the initiating/propagating alcohol, or bifunctional activation of the monomer and alcohol.¹⁹ The significant benefit imparted by the metallo-organic complexes like the above mentioned zinc and yttrium derivatives is their high activity and especially their high selectivity, as evidenced in the iROP of asymmetric six- and seven-membered carbonates (vide infra, Scheme 5). This latter remarkable feature relies on the stereo-electronically favorable surrounding of the bulky β-diiminate or alkoxy-amino-bis(phenolate) ligands that allows a better control of the regioselectivity in the ring-opening of the monomer units, as further demonstrated in the stereoselectivity control with other cyclic monomers, especially with chiral racemic β-lactones and lactides.¹⁶

Scheme 5 Possible regioselectivities for the ROP of a dissymmetric cyclic carbonate illustrated in the case of α-MeTMC.

For comparison purposes, the bulk iROP of each carbonate monomer with these different catalyst systems has been investigated at a given ratio [carbonate]₀/[catalyst]₀/[BnOH]₀ = 500 : 1 : 5. Selected results obtained at a temperature ranging from 20 °C to 150 °C are reported in Table 1. It is noteworthy that these experiments were carried out using rigorously purified monomers, unless otherwise stated.

Table 1 Bulk (i)ROP of TMC, TMC(OMe)₂, α-MeTMC, α-Me7CC and β-Me7CC promoted by various catalyst/BnOH systems at [carbonate]₀/[catalyst]₀/[BnOH]₀ = 500 : 1 : 5^8,10,12–15

Carbonate	[Catalyst]	Temp./°C	Reaction Time^a/min	Conv.^b (%)	M̄ ntheo ^c/g mol^−1	M̄ nSEC ^d/g mol^−1	M̄ w/M̄n^e	TOF^f/h^−1
a Reaction times were not necessarily optimized. b Monomer conversion determined by ^1H NMR. c Calculated from [carbonate]0/[BnOH]0 × monomer conversion × Mcarbonate + MBnOH, with MTMC = 102 g mol^−1, MTMC(OMe)2 = 162 g mol^−1, MMeTMC = 116 g mol^−1, MMe7CC = 130 g mol^−1, MBnOH = 108 g mol^−1. d Number average molar mass determined by SEC vs. polystyrene standards and corrected by a factor of 0.73 for TMC^26 or uncorrected for polycarbonates derived from TMC(OMe)2, α-MeTMC and α,β-Me7CC. e Molar mass distribution calculated from SEC traces. f Apparent turnover frequency expressed in molcarbonate molcatalyst^−1 h^−1.^21 g Reaction run with unpurified TMC at [TMC]0/[catalyst]0/[BnOH]0=10 000 : 1 : 10. h Reaction run with unpurified TMC at [TMC]0/[catalyst]0/[BnOH]0 = 100 000 : 1 : 100. i Reaction run with purified TMC at [TMC]0/[catalyst]0/[BnOH]0=10 000 : 1 : 20. j Reaction run at [α-Me7CC]0/[catalyst]0/[BnOH]0 = 200 : 1 : 1. k Reaction run at [β-Me7CC]0/[catalyst]0/[BnOH]0 = 100 : 1 : 1.
TMC	Al(OTf)3	110	60	96	9900	13 800	1.62	480
TMC	Al(OTf)3	150	5	98	10 100	10 950	1.55	5880
TMC^g	Al(OTf)3	150	20	92	94 050	61 200	1.42	27 600
TMC^h	Al(OTf)3	150	15 × 60	87	88 850	41 400	1.66	5800
TMC	(BDI^iPr)Zn(N(SiMe3)2)	60	7	99	10 200	12 400	1.55	4240
TMC	(BDI^iPr)Zn(N(SiMe3)2)	110	3	100	10 320	11 750	1.77	10 000
TMC^i	(BDI^iPr)Zn(N(SiMe3)2)	60	180	89	45 500	43 300	1.90	2967
TMC^j	(BDI^iPr)Zn(N(SiMe3)2)
TMC	BEMP	60	60	85	8780	8250	1.25	425
TMC	BEMP	110	5	80	8270	6950	1.43	4800
TMC^i	BEMP	110	60	60	30 710	29 400	1.40	6000
TMC^i	BEMP	150	30	65	33 260	30 200	1.61	13 000
TMC^h	BEMP	150	15 × 60	95	97 000	24 300	1.51	6330
TMC	TBD	60	30	99	10 210	10 900	1.85	990
TMC	TBD	110	5	99	10 210	12 700	1.52	5940
TMC^h	TBD	150	15 × 60	91	92 930	25 400	1.54	6067
TMC^i	TBD	110	120	98	50 090	44 850	1.52	4900
TMC^i	TBD	150	10	82	41 930	19 300	1.65	49 200
TMC	DMAP	60	30	50	5210	5050	1.18	500
TMC	DMAP	110	15	97	10 000	13 200	1.53	1940
TMC^i	DMAP	110	120	97	49 580	42 050	1.56	4850
TMC^i	DMAP	150	10	93	47 540	30 050	1.47	55 800
TMC(OMe)2	Al(OTf)3	60	330	28	4 650	—	—	25
TMC(OMe)2	Al(OTf)3	90	330	96	15 660	10 600	1.18	87
TMC(OMe)2	Al(OTf)3	110	330	0	—	—	—	0
TMC(OMe)2	(BDI^iPr)Zn(N(SiMe3)2)	60	90	95	15 660	14 450	1.28	320
TMC(OMe)2	(BDI^iPr)Zn(N(SiMe3)2)	90	60	93	15 200	17 000	1.25	465
TMC(OMe)2	BEMP	90	180	100	16 300	14 300	1.53	167
TMC(OMe)2	TBD	90	180	99	16 150	14 100	1.71	165
TMC(OMe)2	DMAP	90	330	45	7400	6800	1.23	41
α-MeTMC	Al(OTf)3	110	150	81	9500	9300	1.18	162
α-MeTMC	Al(OTf)3	150	15	77	9100	5700	1.18	1540
α-MeTMC	(BDI^iPr)Zn(N(SiMe3)2)	60	7	94	11 000	12 600	1.28	4028
α-MeTMC	BEMP	60	180	72	8450	6000	1.19	120
α-MeTMC	TBD	60	10	93	10 900	10 000	1.19	2790
α-MeTMC	TBD	110	5	99	11 600	8500	1.55	5940
α-MeTMC	DMAP	60	180	30	3600	2500	1.25	50
α-Me7CC	Al(OTf)3	40	85	84	11 050	5950	1.43	296
α-Me7CC	Al(OTf)3	110	15	100	13 100	4900	1.15	2000
α-Me7CC	(BDI^iPr)Zn(N(SiMe3)2)	60	30	94	12 300	8050	1.27	940
α-Me7CC^i	(BDI^iPr)Zn(N(SiMe3)2)	20	10	100	26 100	8600	1.68	1200
β-Me7CC	Al(OTf)3	110	15	100	13 100	22 700	1.18	2000
β-Me7CC	(BDI^iPr)Zn(N(SiMe3)2)	20	15	17	2300	—	—	340
β-Me7CC^k	TBD	110	60	100	13 100	10 350	1.40	100
β-Me7CC^k	BEM	110	60	10	13 100	7800	1.20	100
β-Me7CC^k	DMAP	110	60	87	11 400	8500	1.21	87

---

The overall control of the molecular features of the polycarbonates produced by iROP, that is (i) the match of the experimental molar mass determined by SEC (M̄_nSEC) and/or by NMR (M̄_nNMR) with the one calculated based on the initial [monomer]₀/[initiator/CTA]₀ ratio and on the monomer conversion (M̄_ntheo), and (ii) the relatively narrow molar mass distribution values (M̄_w/M̄_n),²⁰ was quite good. Some undesirable side-reactions occurred, as expected to be more likely for a bulk process as opposed to a solution one, however to a rather limited extent. In the iROP of TMC, the monomer we most extensively investigated,^6,12–14 the range of M̄_w/M̄_n values generally remained within 1.10 to 1.90 for molar mass values ranging up to M̄_nSEC = 185 200 g mol⁻¹.^12a,b

Studies on all other carbonates were focused on the identification of viable catalytic systems effective and possibly regioselective (in the case of α-MeTMC and α/β-Me7CC) in their controlled iROP, rather than on the optimization of the catalyst performances in terms of activity and productivity. For TMC(OMe)₂, polycarbonates of molar mass as high as M̄_nSEC = 70 200 g mol⁻¹ and of M̄_w/M̄_n values in the range 1.12–1.71 were thus prepared.^15,20 Regarding the methyl-substituted six- and seven-membered cyclic carbonates α-MeTMC, α-Me7CC and β-Me7CC, our efforts were aimed at establishing some relationship between the substitution site and the regioselectivity of the catalytic system in ring-opening the cyclic monomer at the oxygen–acyl bond close or remote from the Me group position. The corresponding polymers feature molar mass values in a range allowing NMR investigations which were required to assess the regioselectivity, namely M̄_nSEC ≤ 28 300 g mol⁻¹ and 1.12 < M̄_w/M̄_n < 1.55 for α-MeTMC,^8,20 and M̄_nSEC ≤ 17 000 g mol⁻¹ and 1.11 < M̄_w/M̄_n < 1.73 for α/β-Me7CC.^10,20

In each carbonate monomer, the expected α-hydroxy, ω-alkoxyester polymer chain-ends were always observed, as evidenced by detailed NMR and MALDI-ToF-MS structural analyses of the polycarbonates. Whichever the mechanism involved (coordination–insertion mechanism or activated-monomer mechanism, vide supra),^4a,6 formation of such telechelic polymers was expected (Scheme 2). The alkoxyester end-capping group always originates from the alkoxide moiety of the alcohol used as initiator/CTA. On the other hand, as a result of a different mechanistic approach of the ROP which depends on the catalytic system selected, the origin of the hydroxyl end-group differs based on whether a coordination–insertion mechanism, an activated-monomer mechanism or a mechanism involved for an organocatalyst takes place. In the case of a coordination–insertion mechanism, the hydroxyl chain-end is generated while quenching (i.e., during the final deactivation workup) the polymerization upon protonation of the M–O(alkoxide) bond of the active metal–polymeryl propagating species. On the contrary, the terminal hydroxyl end-function is formed earlier in an activated-monomer mechanism or an organo-catalyzed procedure: as soon as the first (activated) carbonate monomer is ring-opened by the exogenous nucleophilic protic initiator, the carbonate segment to be grown next already bears the terminal hydroxy unit. This reactive α-hydroxy group is then maintained throughout propagation and eventually recovered in the final polymer.⁶

3. Comparative efficiency of the (metallo)-organic catalyst systems in the iROP of six- and seven-membered cyclic carbonates

3.1. iROP of TMC

3.1.1. iROP of TMC using purified monomers. Regarding the iROP of purified TMC under typical operating conditions ([TMC]₀/[catalyst]₀/[BnOH]₀ = 500 : 1 : 5, bulk reaction medium), the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH system afforded the best activity at 110 °C (TOF = 10 000 h⁻¹),^21,22 as compared to the metal triflates (TOF = 480 h⁻¹)¹³ or organic bases (BEMP: TOF = 4800 h⁻¹; TBD: TOF = 5940 h⁻¹; DMAP: TOF = 1940 h⁻¹)¹⁴ (Table 1). The Al(OTf)₃/BnOH system was less active and required a temperature of 150 °C to reach activities similar to those obtained from organocatalysts (TOF = 5880 h⁻¹).¹³ The highest turnover frequency we ever obtained with purified TMC was achieved with the latter organic-based systems DMAP and TBD at 150 °C (TOF = 55 800 and 49 200 h⁻¹, respectively), using a [TMC]₀/[catalyst]₀/[BnOH]₀ ratio of 10 000 : 1 : 20.¹⁴ The best productivity with purified TMC was obtained with a bis(morpholinomethyl)phenoxide zinc system (TON = 96 000 in 8 h).^6,23 However, upon raising either the initial amount of TMC or BnOH (both purified), the BEMP and TBD based systems resist, better than the zinc systems, the problems caused by the potential residual impurities being inherent to such large quantities of reagents. In fact, at a [TMC]₀/[catalyst]₀/[BnOH]₀ ratio of 10 000 : 1 : 20, the {β-diiminate}–zinc system is already less efficient (apparent²¹ TOF_60°C = 2967 h⁻¹)^12a,b than that based on either TBD (TOF_110°C = 4900 h⁻¹), DMAP (TOF_110°C = 4850 h⁻¹) or BEMP (TOF_110°C = 6000 h⁻¹).¹⁴ Also, increasing the alcohol content to 200 equiv. significantly altered the activity of the {β-diiminate}–zinc system (TOF_60°C = 1165 h⁻¹ at [purified TMC]₀/[catalyst]₀ ratio of 5000 : 1),^12a,b as compared to the TBD (TOF_110°C = 10 000 h⁻¹) or BEMP (TOF_110°C = 9800 h⁻¹) systems evaluated with an amount of unpurified TMC even twice as large ([TMC]₀/[catalyst]₀ ratio of 10 000 : 1 : 200).¹⁴ Obviously, these data reflect that part of the metallo-organic {β-diiminate}–zinc (pre)catalyst is irreversibly deactivated in the presence of large amounts of monomer and initiator/CTA, in contrast to the robust organocatalysts.

Several other metallic complexes including Mg[N(SiMe₃)₂]₂, Ca[N(SiMe₃)₂]₂(THF)₂, Y[N(SiMe₃)₂]₃, [(BDI^iPr)Fe(N(SiMe₃)₂)], Fe[N(SiMe₃)₂]₂, Fe[N(SiMe₃)₂]₃, Zn[N(SiMe₃)₂]₂, and ZnEt₂ were also successfully combined with an alcohol such as isopropanol or benzyl alcohol into bi-component catalytic systems for the bulk iROP of TMC at 60–110 °C.^12b The magnesium, calcium and yttrium amido derivatives investigated, which are more active than [(BDI^iPr)Zn[N(SiMe₃)₂], remain yet more sensitive and undergo rapid deactivation when the TMC loading is increased, while the iron amido systems are less active. Thus, among all these metallo-organic precursors, the {β-diiminate}–zinc amido complex provides an ideal compromise affording the overall best performances combining both a high activity and a relatively good stability. Under optimized conditions, as much as 50 000 equiv. of purified TMC could be fully converted from as little as 20 ppm of this metallic precursor, with as many as 300 growing polycarbonate chains, allowing the preparation of a PTMC of a molar mass as high as 185 200 g mol⁻¹ with a relatively narrow molar mass distribution (M̄_w/M̄_n = 1.68).^12a,20 Also, a double monomer feed experiment carried out with this same {β-diiminate}–zinc/BnOH catalytic system further proved the “living” character of the polymerization.^12b

3.1.2. iROP of TMC using unpurified monomer. When technical quality (that is, unpurified commercial) TMC was used, the more sensitive {β-diiminate}–zinc-based system was rapidly decomposed by the impurities present in such a lower grade monomer; it was thus simply completely ineffective. Yet, Al(OTf)₃, BEMP, TBD and DMAP were still efficient under such technical conditions, affording very high activities and productivities even at higher [unpurified TMC]₀/[catalyst]₀/[BnOH]₀ ratios: Al(OTf)₃ (ratio = 10 000 : 1 : 10, TOF_150°C = 27 600 h⁻¹; 100 000 : 1 : 100, TOF_150°C = 5800 h⁻¹, TON_150°C = 87 000),¹³ BEMP (ratio = 100 000 : 1 : 100, TOF_150°C = 6330 h⁻¹, TON_150°C = 95 000) and TBD (ratio = 100 000 : 1 : 100, TOF_150°C = 6067 h⁻¹; TON_150°C = 91 000; Table 1).¹⁴

The overall efficiency of these robust metal triflates and organocatalyst systems towards technical grade TMC is particularly relevant and significant for industrial applications which do not provide high purity reagents/operating conditions. Remarkably, under our experimental conditions, the growth of as many as 800 polymer chains from the catalyst, introduced at loadings as low as 10 ppm (i.e., 2 mg of BEMP or TBD), was possible by converting quantitatively 74.4 g of technical-grade TMC, which resulted in a PTMC with a molar mass as high as 61 200 g mol⁻¹.

With unpurified TMC, some of the impurities (supposedly protic reagents, such as water or residual 1,3-propanediol) present in the monomer in larger amounts with larger loadings were demonstrated to behave as additional initiators/CTAs in a similar manner to the purposely added exogenous alcohol. In that case, the total number of growing polymer chains is then determined by the sum of added alcohol CTA molecules and “transfer active” impurities. This increase in active species in turn, at a given conversion, proportionally decreases the molar mass of PTMCs resulting from such polymerizations. We established that an accurate theoretical molar mass of the polymer () can be determined upon taking into account the amount of these “transfer active” impurities X̄ (X̄ = 0.056% with the TMC batch used).^13,14 This approach further allowed us to predict and thereby to tune “at will” the polymer molar mass from a given loading of TMC, based on the [TMC]₀/[Al(OTf)₃]₀/[BnOH]₀ ratio and the X̄ content of “active transferring” impurities contained within all reagents (monomer and initiator/CTA).

3.2. iROP of TMC(OMe)₂

The addition of a dimethoxy acetal group onto the TMC ring significantly altered the activity of the ROP catalyst systems (Table 1).¹⁵ In fact, the activity of the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH system observed in the iROP of TMC (TOF_60°C = 4240 h⁻¹)^12a,b dramatically dropped down to TOF_60°C = 320 h⁻¹ with TMC(OMe₂). The Al(OTf)₃/BnOH system was also found to be less efficient towards the iROP of TMC(OMe₂) (TON_90°C = 87 h⁻¹)¹⁵ than of TMC (TON_110°C = 480 h⁻¹).¹³ Regarding the organocatalyst systems, these all remained weakly effective at 90 °C, with yet the guanidine (TOF_TBD = 165 h⁻¹) and phosphazene (TOF_BEMP = 167 h⁻¹) being both more active than the amine (TOF_DMAP = 41 h⁻¹),¹⁵ especially in light of the performances established in the iROP of TMC at an even lower temperature (60 °C; TOF_TBD = 990 h⁻¹; TOF_BEMP = 425 h⁻¹; TOF_DMAP = 500 h⁻¹).¹⁴

The decreased polymerization activities observed with TMC(OMe)₂ (as compared to TMC) for all these catalysts most likely result from unfavorable interactions of the additional methoxy functionalities with either the Lewis acidic metal (Zn, Al) centers or the organocatalysts, which compete with the effective activation of the carbonate moiety.

3.3. iROP of α-MeTMC

In the iROP of α-MeTMC, at [carbonate]₀/[catalyst]₀/[BnOH]₀ = 500 : 1 : 5, all the catalytic systems successfully afforded the corresponding poly(α-MeTMC)s with quite good control and activities (Table 1).⁸ Both the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH (TOF_60°C = 4028 h⁻¹) and the TBD/BnOH (TOF_60°C = 2790 h⁻¹) catalytic systems proved particularly active at 60 °C, while the other organocatalyst systems, BEMP (TOF_60°C = 120 h⁻¹) and DMAP (TOF_60°C = 50 h⁻¹) were less effective.⁸ The catalyst based on Al(OTf)₃ polymerized α-MeTMC at 110/150 °C (TOF_α-MeTMC = 162 h⁻¹ at 110 °C and 1540 h⁻¹ at 150 °C), yet less efficiently than TMC under the same conditions (TOF_TMC = 480/5880 h⁻¹, respectively).^8,13 In fact, the overall catalytic efficiency always remained higher in the iROP of TMC compared to that of α-MeTMC, irrespective of the catalyst considered. This apparent reactivity trend can be accounted for by the presence of the α-methyl substituent which may sterically disfavor monomer coordination or nucleophilic attack of the initiator, as opposed to the naked cyclic carbonate analogue.

3.4. iROP of α-Me7CC and β-Me7CC

Similarly, in the iROP of the seven-membered methyl-substituted carbonates, namely α-Me7CC and β-Me7CC, at [carbonate]₀/[catalyst]₀/[BnOH]₀ = 500 : 1 : 5, all catalytic systems successfully afforded the corresponding poly(α-Me7CC)s and poly(β-Me7CC)s with quite good control and activities.¹⁰ Whichever the seven-membered cyclic monomer, relatively high activities were achieved with [(BDI^iPr)Zn(N(SiMe₃)₂)] at room temperature (TOF_{α-Me7CC/20°C} = 940 h⁻¹, TOF_{β-Me7CC/20°C} = 340 h⁻¹) systems, while the Al(OTf)₃/BnOH system again required higher temperatures to be operative (TOF_110°C = 2000 h⁻¹ for both monomers). The latter Lewis acid catalyst is more active with the larger carbonates than with TMC: at 110 °C, whereas a TMC conversion of 88% was achieved after 30 min,¹³ quantitative conversion of α-Me7CC and β-Me7CC was achieved within 15 min.¹⁰ Also, in the iROP of the one-carbon smaller α-MeTMC, Al(OTf)₃/BnOH exhibited at 110 °C a lower activity (TOF_α-Me7CC = 2000 h⁻¹vs. TOF_α-MeTMC = 162 h⁻¹).^8,10

The ROP of β-Me7CC using the organocatalysts DMAP, TBD or BEMP in the presence of BnOH ([carbonate]₀/[catalyst]₀/[BnOH]₀ = 100 : 1 : 1) was found to be effective only at higher temperatures: whereas no polymer was formed at 20 or 60 °C, the polymerization proceeded smoothly at 110 °C. All three organocatalytic systems then exhibited rather similar activities (TOF ≤ 87–100 h⁻¹).^8,10

3.5. Regioselectivity of the metallo-organic catalyst systems in the iROP of six- and seven-membered cyclic carbonates

The dissymmetry of α-MeTMC, α-Me7CC and β-Me7CC raises the question of the regioselectivity of the catalyst during the ring-opening and the enchainment of monomer units (Scheme 5). Detailed microstructural analyses of the poly(α-MeTMC)s, poly(α-Me7CC)s and poly(β-Me7CC)s using ¹H and ¹³C{¹H} NMR and MALDI-ToF-MS techniques allow us to gauge the effect of the methyl substituent position within the carbonate cycle and the performance of the different catalyst systems.

¹³C NMR investigations of the microstructure of the polycarbonates revealed particularly informative. The intensity of the three resonances observed in the carbonyl region proved diagnostic of the monomer diad sequences, allowing us to differentiate and to quantify the enchainment of two monomer units, i.e. the regioregularity of cleavage of either of the two O–C(O)O bonds in α-MeTMC, α-Me7CC or β-Me7CC. The intensity of the two smaller upfield and downfield resonances (B and C, arising from regioirregular enchainment of the monomer units) relative to the central one (A and D, arising from regioregular enchainment of the monomer units) enabled to calculate the regioregularity X_reg in the ROP.²⁴ This extent of regioregularity provided by the catalytic system was independently confirmed with the resonances in the methine (OCHCH₃) and methylene (OCH₂) regions of the ¹³C{¹H} NMR spectra.^8,10

Considering poly(α-MeTMC)s, the X_reg values illustrated that the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH system affords the best regioselectivity (X_reg > 0.98), as opposed to the Al(OTf)₃/BnOH and TBD/BnOH systems which are less or hardly regioselective (X_reg = ca. 0.62 and 0.35–0.39, respectively).⁸ The same trend in regioselectivity values was observed in the iROP of α-Me7CC but to somewhat different extents: the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH system achieved the most regioselective (X_reg = 0.71) iROP, while [(ONOO^tBu)Y(N(SiHMe₂)₂)(THF)]/BnOH and especially Al(OTf)₃/BnOH were less regioselective (X_reg = ca. 0.58 and 0.27, respectively).¹⁰ Apparently, the smaller ring size of α-MeTMC (as compared to α-Me7CC) and the bulkier structure of the zinc (vs. aluminium) catalyst favored a regioselective process. The lack of regioselectivity of TBD in the iROP of α-MeTMC, although inherently different in nature, is in line with the poor stereoselectivity that this and most other organocatalysts have been thus far able to achieve in ROP of chiral cyclic esters.²⁵ Finally, we observed that, in contrast to poly(α-Me7CC), the ROP of β-Me7CC always proceeds in a perfectly statistical fashion (i.e., X_reg = 0), irrespective of the operating conditions and catalyst systems used.¹⁰

Overall, these microstructural investigations revealed: (i) the higher regioselectivity in the ROP of α-MeTMC and of α-Me7CC—with preferential ring-opening at the most hindered oxygen-acyl O–C(O)O bond, i.e. closest to the α-Me substituent—of the zinc-based system, most likely as a result of the favorable steric hindrance of this bulky catalyst; (ii) the absence of regioselectivity of the Al(OTf)₃ or TBD catalyst systems in the ROP of α-MeTMC and α-Me7CC; and (iii) the absence of regioselectivity in the ROP of β-Me7CC, whichever the catalyst system, most likely as a result of the OC(O)O-further remote substitution site.

4. Conclusions and outlooks

Various well-defined bioresourced polycarbonates, including the common PTMC, the less familiar substituted/functionalized analogues poly(α-MeTMC) and P(TMC(OMe)₂), and the even less studied polycarbonates derived from the seven-membered rings α-Me7CC and the novel β-Me7CC, are accessible by (i)ROP of the corresponding cyclic monomers using several bi-component systems based on metallo-organic, inorganic or organic (pre)catalysts. All the catalytic systems surveyed were efficient, yet at different temperatures and with quite different activities and/or productivities.

The overall order of catalytic efficiency is [(BDI^iPr)Zn(N(SiMe₃)₂)] > Al(OTf)₃ ≈ TBD > BEMP ≈ DMAP, whichever the carbonate monomer, but provided a rigorously purified sample is used. While the [(BDI^iPr)Zn(N(SiMe₃)₂)]/BnOH system features high activity and productivity at typically 60–110 °C when using purified TMC loadings, Al(OTf)₃ and all organocatalysts remain by far the most effective catalysts with technical grade TMC batches. The robustness and achievements of these metallic triflates and organocatalysts are remarkable and of high relevance for industrial applications. They are most effective at higher temperatures in the range 110–150 °C, more appropriate from industrial considerations. However, they lack selectivity and, as especially pointed out in recent studies, they do not impart regioselectivity in the ring-opening of dissymmetric cyclic carbonates such as α-MeTMC, α-Me7CC and β-Me7CC, as opposed to the more selective but much more sensitive {β-diiminate}–zinc and {alkoxy-amino-bis(phenolate)}-yttrium precursors.

According to the desired achievements and applications, one may thus select one catalyst system over another in order to reach the best compromise between sensitivity, activity and selectivity. Obviously, the design of an ideal catalytic system combining selectivities and high effectiveness still remains a challenge. One first approach might be considered in light of the recently developed, so-called “dual organic/organometallic ring-opening polymerization” which provides a three-component catalytic system: a metallic Lewis acidic center (a discrete cationic {diaminophenolato}–zinc complex) that activates the monomer (lactide), and an organic base (a tertiary amine) that activates the initiating/propagating alcohol.²⁷ It is envisioned that the role of the Lewis acidic catalyst in inducing selectivity in such a dual polymerization [which can even proceed under immortal and living ROP conditions] could be promoted by the surrounding ancillary ligand(s).

Acknowledgements

We gratefully thank all co-workers who contributed to this research and whose names appear in the reference section. This research has been gratefully supported in part by Total Petrochemicals Co. (PhD grant to M. Helou), ANR “BIOPOLYCAT” (CP2D-08-01, PhD grant to P. Brignou), the CAPES-COFECUB (program 556/07; CAPES grant to M. Priebe Gil), the Région Bretagne (ACOMB research program, SMG; PhD grant to C. Guillaume), Rennes Métropole (SMG), the CNRS, and the Institut Universitaire de France (JFC).

Notes and references

1. (a) D. J. A. Cameron and M. P. Shaver, Chem. Soc. Rev., 2011, 40, 1761–1776; (b) B. D. Ulery, L. S. Nair and C. T. Laurencin, J. Polym. Sci., Part B: Polym. Phys., 2011, 49, 832–864; (c) A. Lendlein, Macromol. Biosci., 2010, 10, 993–997; (d) T. Artham and M. Doble, Macromol. Biosci., 2008, 8, 14–24; (e) S. Slomkowski, Macromol. Symp., 2007, 253, 47–58; (f) S. Y. Wong, J. M. Pelet and D. Putnam, Prog. Polym. Sci., 2007, 32, 799–837; (g) R. Auras, B. Harte and S. Selke, Macromol. Biosci., 2004, 4, 835–864; (h) A.-C. Albertsson and I. K. Varma, Biomacromolecules, 2003, 4, 1466–1486; (i) Y. Ikada and H. Tsuji, Macromol. Rapid Commun., 2000, 21, 117–132.
2. (a) A. P. Dove, Chem. Commun., 2008, 6446–6470; (b) C. Jérôme and P. Lecomte, Adv. Drug Delivery Rev., 2008, 60, 1056–1076; (c) S. Penczek, M. Cypryk, A. Duda, P. Kubisa and S. Slomkowski, Prog. Polym. Sci., 2007, 32, 247–282; (d) A.-C. Albertsson and I. K. Varma, Adv. Polym. Sci., 2002, 157, 1–40; (e) M. Okada, Prog. Polym. Sci., 2002, 27, 87–133; (f) G. Rokicki, Prog. Polym. Sci., 2000, 25, 259–342.
3. (a) W. Saiyasombat, R. Molly, T. M. Nicholson, A. F. Johnson, I. M. Ward and S. Poshyachinda, Polymer, 1998, 39, 5581–5585; (b) S. Strandman, J. E. Gautrot and X. X. Zhu, Polym. Chem., 2011, 2, 791–799.
4. (a) O. Dechy-Cabaret, B. Martin-Vaca and D. Bourissou, Chem. Rev., 2004, 104, 6147–6176; (b) J. M. Becker, R. J. Pounder and A. P. Dove, Macromol. Rapid Commun., 2010, 31, 1923–1937.
5. S. Inoue, J. Polym. Sci., Part A: Polym. Chem., 2000, 38, 2861–2871.
6. N. Ajellal, J.-F. Carpentier, C. Guillaume, S. M. Guillaume, M. Helou, V. Poirier, Y. Sarazin and A. Trifonov, Dalton Trans., 2010, 39, 8363–8376.
7. (a) A. Behr, J. Eilting, K. Irawadi, J. Leschinski and F. Lindner, Green Chem., 2008, 10, 13–30; (b) C.-H. Zhou, J. N. Beltramini, Y.-X. Fan and G. Q. Lu, Chem. Soc. Rev., 2008, 37, 527–549; (c) D. T. Johnson and K. A. Taconi, Environ. Prog., 2007, 26, 338–348.
8. P. Brignou, J.-F. Carpentier and S. M. Guillaume, Macromolecules, 2011, 44, 5127–5135.
9. A. N. Zelikin, P. N. Zawaneh and D. Putnam, Biomacromolecules, 2006, 7, 3239–3244.
10. P. Brignou, M. Priebe Gil, O. Casagrande Jr., J.-F. Carpentier and S. M. Guillaume, Macromolecules, 2010, 43, 8007–8017.
11. (a) T. Werpy, G. Petersen (Principal Investigators), A. Aden, J. Bozell, J. Holladay, J. White and A. Manheim (Contributing Authors), Results of Screening Potential Candidates from Sugars and Synthesis Gas, US DoE report, Washington, DC, 2004, vol. 1; (b) J. J. Bozell and G. R. Petersen, Green Chem., 2010, 12, 539–554; (c) P. J. Alaimo and A.-L. Marshall, Chem.–Eur. J., 2010, 16, 4970–4980.
12. (a) M. Helou, O. Miserque, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, Chem.–Eur. J., 2010, 16, 13805–13813; (b) M. Helou, O. Miserque, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, Adv. Synth. Catal., 2009, 351, 1312–1324; (c) M. Helou, O. Miserque, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, Macromol. Rapid Commun., 2009, 30, 2128–2135; (d) M. Helou, J.-F. Carpentier and S. M. Guillaume, Green Chem., 2011, 13, 266–271.
13. M. Helou, O. Miserque, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, ChemCatChem, 2010, 2, 306–313.
14. M. Helou, O. Miserque, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, Chem.–Eur. J., 2010, 16, 13805–13813.
15. M. Helou, J.-M. Brusson, J.-F. Carpentier and S. M. Guillaume, Polym. Chem., 2011, 2, 2789–2795.
16. (a) A. Amgoune, C. M. Thomas, S. Ilinca, T. Roisnel and J.-F. Carpentier, Angew. Chem., Int. Ed., 2006, 45, 2782–2784; (b) C. Guillaume, J.-F. Carpentier and S. M. Guillaume, Polymer, 2009, 50, 5909–5917; (c) C. Guillaume, N. Ajellal, J.-F. Carpentier and S. M. Guillaume, J. Polym. Sci., Part A: Polym. Chem., 2011, 49, 907–917; (d) M. Helou, G. Moriceau, Z. W. Huang, S. Cammas-Marion and S. M. Guillaume, Polym. Chem., 2011, 2, 840–850.
17. L. R. Rieth, D. R. Moore, E. Lobkovsky and G. W. Coates, J. Am. Chem. Soc., 2002, 124, 15239–15248 and references therein.
18. (a) S. Kobayashi and K. Manabe, Acc. Chem. Res., 2002, 35, 209–217; (b) S. Kobayashi, M. Sugiura, H. Kitagawa and W. W.-L. Lam, Chem. Rev., 2002, 102, 2227–2302.
19. (a) M. K. Kiesewetter, E. J. Shin, J. L. Hedrick and R. M. Waymouth, Macromolecules, 2010, 43, 2093–2107; (b) N. E. Kamber, W. Jeong, R. M. Waymouth, R. C. Pratt, B. G. G. Lohmeijer and J. L. Hedrick, Chem. Rev., 2007, 107, 5813–5840; (c) F. Nederberg, B. G. G. Lohmeijer, F. Leibfarth, R. C. Pratt, J. Choi, A. P. Dove, R. M. Waymouth and J. L. Hedrick, Biomacromolecules, 2007, 8, 153–160.
20. A purely “immortal” polymerization is anticipated to generate polymers with very narrow molar mass distributions, ideally approaching a value of 1.0. The polycarbonates obtained in the processes reported in this manuscript usually had larger M̄_w/M̄_n values, in the range 1.1–1.9. This discrepancy is accounted for by the fact that most of those polymerizations were performed in bulk monomer, thus inducing side reactions.
21. Apparent turnover frequencies expressed in mol_monomer mol_catalyst⁻¹ h⁻¹ and calculated from the total loading of (pre)catalyst initially introduced; these values are lower estimates as they are derived from high conversion values and from reaction times that were not necessarily optimized.
22. The iROP of TMC was carried out over 3 min at 110 °C with a ratio of [TMC]₀/[Al(OTf)₃]₀/[BnOH]₀ = 500 : 1 : 5. SEC analysis of the resulting PTMC (TMC conversion = 100%; M̄_n theoretical = 10 320 g mol⁻¹) showed M̄_nSEC = 11 750 g mol⁻¹ and/M̄_w/M̄_n = 1.77.
23. V. Poirier, T. Roisnel, J.-F. Carpentier and Y. Sarazin, Dalton Trans., 2009, 9820–9827.
24. The degree of regioregularity was calculated by the formula X_reg = 1 − (relative intensity of the related regioirregular resonances), the resonance for regioregular enchainment being set up to 1. The degree of regioregularity, X_reg, is opposed to the degree of randomness, X_rand = 1 − X_reg). For a perfectly statistical ROP process X_reg = 0 (i.e., X_rand = 1): the probability of formation of these latter diad sequences is equal, and a 1 : 2 : 1 distribution for the three carbonyl resonances is expected. For a more regioregular ROP process X_reg → 1 (i.e., X_rand → 0): the high and low field carbonyl resonances are expected to progressively decrease such that only the central resonance should be observed for a perfectly regioregular (X_reg = 1) ROP.
25. (a) M. K. Kiesewetter, E. J. Shin, J. L. Hedrick and R. M. Waymouth, Macromolecules, 2010, 43, 2093–2107; (b) M. J. Stanford and A. P. Dove, Chem. Soc. Rev., 2010, 39, 486–494.
26. I. Palard, M. Schappacher, B. Belloncle, A. Soum and S. M. Guillaume, Chem.–Eur. J., 2007, 13, 1511–1521.
27. (a) E. Piedra-Arroni, P. Brignou, A. Amgoune, S. M. Guillaume, J.-F. Carpentier and D. Bourissou, Chem. Commun., 2011, 47, 9828–9830; (b) P. Brignou, S. M. Guillaume, D. Bourissou, T. Roisnel and J.-F. Carpentier, submitted.

---