Carbon Dioxide as C-1 Block for the Synthesis of Polycarbonates

Abstract

The chapter summarizes the recent developments in catalyst synthesis, polymer analytics, and material properties of polycarbonates produced from carbon dioxide and epoxides. The physical behavior of the two most important and well-researched polycarbonates poly(propylene carbonate) (PPC) and poly(cyclohexene carbonate) (PCHC) do not meet the desired characteristics of structure polymers, yet. Terpolymerization, blending, end-capping, and increasing the molecular weight are just a few examples which are more closely covered in this chapter and could be methods for adjusting PPC and PCHC properties. NMR analytics are one of the major analytical tools for the characterization of the polymer microstructure and are also included in this chapter. Today, the focus for homogeneous catalysts is shifting toward sustainable and cheap metals, e.g., magnesium and zinc. Studies with heterogeneous and homogeneous structures indicate a strong dependence of the activity on the distance between the metal sites. Bimetallic systems try to exploit this correlation. Kinetic studies supported by DFT calculations have been performed, which support the hypothesis and show up new ways toward even higher activities in the copolymerization of carbon dioxide and epoxides.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Fossil energy sources became the dominant energy carriers at the beginning of the twentieth century. Since then, no alternatives to crude oil, gas, and coal have arisen to supplant them as a universal source of energy and raw materials for the chemical industry [1, 2]. The discovery of versatile possibilities for the use of crude oil as well as its low price and historical abundance made the steep rise in living standards in the last 100 years possible. As a result, the concentration of carbon dioxide in the atmosphere has risen significantly and reached 397 ppm in 2013 [3]. However, the finiteness of coal, gas, and especially oil reserves and the steadily increasing demand for them are likely to create a rising imbalance of supply and demand. Furthermore, the various negative implications of reliance on fossil energy carriers generate only limited incentives for industry and consumers to change their accustomed behavior. The reason for the disproportionate low price of CO₂ is that the resulting environmental damage is not or is only partially included in it. The rising generation of carbon dioxide however opens up a vast C-1 source for the synthesis of chemical raw materials. The attractive properties of CO₂ as a nontoxic, renewable, and low-cost C₁-building block are known, and it is currently used on an industrial scale in the production of urea (146 × 10⁶ t/a), methanol (6 × 10⁶ t/a), cyclic carbonate (0.040 × 10⁶ t/a), and salicylic acid (0.060 × 10⁶ t/a), among others [4]. However, the thermodynamic stability has thus far hampered its usage as widespread chemical reagent. Methods to overcome the high energy barriers are based upon reduction, oxidative coupling with unsaturated compounds on low valency metal complexes, and increasing electrophilicity of the carbonyl carbon [5].

Fig. 7.1

Alternating copolymerization of propylene oxide (PO) (a) and cyclohexene oxide (CHO) (b) with CO₂

The reaction of CO₂ with highly reactive substrates, such as epoxides, affords materials which can be produced on industrially relevant scale (Fig. 7.1). These polycarbonates from CO₂ have the potential to be the next industrial product with a high impact on the usage as commodity polymers. Consequently, the synthesis of highly active catalysts and understanding their mechanistic behavior is of upmost importance and the current developments will be discussed in this chapter.

2 Current Situation

Over 20 chemical reaction pathways for carbon dioxide have been developed over the course of the last two decades. Industrially viable reactions are scarce and only the production of urea, salicylic acid, and carbonates have reached larger volumes [6, 7]. The amount of carbon dioxide consumed by these or any other chemical processes however cannot counteract anthropologically caused climate change. They rather have the potential to supplant oil-based starting materials in various synthetic pathways and thereby open up a more eco-friendly synthesis. The production of biodegradable polycarbonates from CO₂ and epoxides is currently receiving a lot of attention [5, 8, 9]. Aside from the polymeric product, cyclic carbonates also have potential applications such as in lithium ion batteries as electrolytes and as intermediates for the synthesis of linear dialkyl carbonates [4]. In this chapter cyclic carbonates will not be discussed in detail [9–11]. The focus will instead be on the catalysis behind the synthesis of polymeric materials with the recent developments at the center. Polycarbonates such as poly(propylene carbonate) (PPC) and poly(cyclohexene carbonate) (PCHC) have promising physical properties like lightness, durability, biodegradability, heat resistance, high transparency, and gas permeability which make them possible candidates for applications in the automotive, medical, and electronics industries.

2.1 Properties of Polycarbonates on the Example of Poly(propylene carbonate)

A large variety of epoxides are commercially available and subsequently have been tested for their behavior in copolymerization with carbon dioxide or even in terpolymerization with an additional epoxide. In industry though only ethylene oxide and propylene oxide are currently produced on large scale and have therefore a commercial advantage. Terpolymerizations promise a synthetic approach toward materials with better physical and material properties. Poly(propylene carbonate) and poly(cyclohexene carbonate) do not yet meet the demands of industry in respect of their glass transition temperatures, achieved molecular weights, elasticity, and their homogeneity. This justifies an academic approach toward better understanding of the polymerization behavior and properties of the produced materials.

2.1.1 General Properties

Poly(propylene carbonate) is an aliphatic polycarbonate (Fig. 7.2). Due to irregular incorporation of carbon dioxide, polyether units can be presented in the polymer backbone. PPC is amorphous, hydrophobic, and soluble in organic solvents, e.g., dichloromethane, acetone, and cyclic propylene carbonate. The general properties strongly depend on the molecular weight, polydispersity, ratio of carbonate to ether linkages, microstructure, and possible catalyst contaminations in the polymer [12, 13]. Due to all the mentioned factors in literature, a temperature range from 25 to 46 °C for the glass transition temperature of PPC is given, and also other material characteristics like tensile strength and Young’s modulus also show some variance [14, 15]. Cyclic carbonates act as plasticizers in the polymer and have the same effect as ether units in the polymer backbone: they decrease the glass transition temperature. Additionally, high amounts of ether linkages affect the material’s thermal stability.

Fig. 7.2

Poly(propylene carbonate) (left), poly(cyclohexene carbonate) (right)

As the available analytical data is not always sufficient to correctly compare the investigated polymer samples, some properties only can be given as approximate ranges. For polymer samples with a T_g of about 40 °C, the range for the modulus was from 700 to 1,400 MPa, whereas for PPC with a T_g of 30 °C the modulus was reported to be 200 and 1,000 MPa [14, 15]. However, for strictly alternating PPC with less than 1 % cyclic carbonate content, an elongation at break of less than 50 % and a tensile strength of around 40 MPa (M_n = 180,000 g/mol) is reported in the literature [8]. One of the major disadvantages of PPC is its tendency to degrade at elevated temperatures. Earlier reports tested PPC with different molecular weights at temperatures between 150 and 180 °C. They observed severe degradation which of course has negative consequences for the material properties. Current publications suggest that the degradation is not solely induced by temperature, but that residual catalyst can enhance the degradation as well [16–20]. Already at 150 °C, a PPC sample with a molecular mass over 200 kg/mol and a PDI of 1.2 show severe degradation to cyclic carbonate after 1 h if the catalyst residue is not completely removed from the polymer [12]. An increase in thermostability to 180 °C was made possible by a thorough catalyst removal from the sample. Further increases in temperature resulted in profound degradation even for the treated polymer sample. Blending of PPC with other engineering plastics is possible but difficult due to the fact that the optimum melt processing temperature for PPC is around 100–140 °C [21]. In general, polyesters and polyamides exhibit higher melting temperatures of above 200 °C [16]. The catalysts known in the literature so far show low productivity and produce at most 38 kg polymer per gram cobalt catalyst [22]. The catalysts cannot be left in the polymer due to their coloring- and degradation-enhancing properties. Furthermore, regulations concerning the amount of toxic metals like cobalt make a thorough clean up of the polycarbonates necessary.

2.1.2 Degradation Mechanism and Thermal Stability

In the previous subchapter, the thermal degradation of PPC was briefly addressed. Now the underlying mechanisms and possibilities for the formation of more stable PPC shall be discussed in more detail.

One form of thermal degradation is chain unzipping, whereby the nucleophile for the degradation can be located in the polymer chain or through an external nucleophile from another chain (Fig. 7.3a). Terminal hydroxyl groups can perform intramolecular backbiting reactions and thereby cause the formation of cyclic carbonates. The backbiting reaction can occur via a terminal alkoxide chain end which attacks at a carbonyl carbon. The backbiting route can also be via a terminal carbonate which attacks at an electrophilic carbon in the polymer chain. Intermolecular back biting leads to an attack at a random position in the polymer and forms shorter chains. The second degradation mechanism, which is assumed to take place increasingly at higher temperatures, is random chain scission (Fig. 7.3b) [23, 24]. At elevated temperatures, a decarboxylation reaction takes place and gives rise to carbon dioxide and an olefin moiety. As can be seen in Fig. 7.3c, catalyst impurities, water residues, acids, bases, and solvents promote degradation reactions and should therefore be thoroughly excluded or removed upon completion of the reaction [16].

Fig. 7.3

Backbiting mechanisms for poly(propylene carbonate): (a) chain unzipping, (b) chain scission, and (c) residual water traces promoting the degradation

The thermal stabilities reported in the literature are difficult to compare and to relate to one another due to a lack of information concerning both the polycarbonates and the testing conditions. Generally, the chain unzipping of poly(propylene carbonate) and other carbonates can be suppressed to a certain degree by end-capping, although the potential for depolymerization via chain scission is not affected by this alteration. In the end-capping process, the nucleophilic terminal hydroxyl groups of the poly(alkylene carbonates) are replaced with less reactive functionalities (e.g., isocyanates, carboxylic acid anhydrates, acetates, urethanes, sulfonates, phosphorous containing compounds) [23–27]. Analytical data concerning the remaining hydroxyl groups have not been published so far. Hydrogen bonding has also been shown to improve the thermal stability of polycarbonates significantly. In a recent publication, octadecanoic acid was added to PPC by solution mixing. The resulting supramolecular complexes showed improved thermal stability, a thermotropic liquid crystalline character, and enhanced glass transition compared to an amorphous PPC copolymer [28]. Improvements in thermal stability were also obtained by adding wood flour [29] or nanoparticles [30–33] to polycarbonates by blending with other polymers [34–37].

The actual measured thermal stabilities are difficult to compare due to different testing conditions and the lack of information concerning the poly(propylene carbonates) themselves. It was reported that interactions between poly(propylene carbonate) and metal ions can in fact enhance the thermal stability [38–40]. Nevertheless, none of the modifications to poly(propylene carbonate) mentioned in the literature are able to inhibit degradation during processing at temperatures over 180 °C.

2.1.3 Terpolymerization

Terpolymerizations and block copolymerizations have the potential to overcome the poor properties of poly(propylene carbonate) and poly(cyclohexene oxide), respectively, and instead combine their strengths in one polymeric material. Other epoxides such as styrene oxide and epichlorohydrin offer additional opportunities as copolymers and also as substitutes for PO or CHO. Consequently, both the copolymerization and the terpolymerization of carbon dioxide and various epoxides have been investigated (Fig. 7.4) [41–54]. The incorporation of other monomers (lactones and anhydrides) in polycarbonate structures has also been studied [55–62].

Fig. 7.4

Some epoxides used in the literature [41–54]

Fundamental research on the reasons for the differing reactivities and on the inactivity of some catalysts toward some monomers is scarce. Rieger et al. published detailed kinetic studies supported by quantum chemical calculations in 2011 [63]. The homogeneous dinuclear zinc catalyst 26 of Williams et al. was chosen as a test system as it exhibits the highest activities for CHO/CO₂ copolymerization at one bar carbon dioxide pressure (TOF = 25 h⁻¹) [64]. The copolymerization of propylene oxide and carbon dioxide however is not possible with catalyst 26: instead, only cyclic carbonate is formed. Many zinc-based catalyst systems share this inactivity for PO/CO₂ copolymerization or possess only low activities compared to other metals. Therefore, Rieger et al. systematically varied the reaction conditions to identify the underlying principles. They found that neither propylene oxide nor additionally added cyclic propylene carbonate does deactivate the catalyst for the formation of PCHC, although in the presence of cyclic propylene carbonate, the activity was decreased dramatically. Terpolymerizations with different ratios of CHO to PO resulted only in PCHC and no incorporation of PO was found, which can be attributed to a very fast backbiting reaction in the case of PO. DFT calculations showed that the ring strains of CHO and PO are comparable and that kinetic barriers to polymer formation in theory favor generation of PPC over PCHC. Consequently, the authors concluded that in the case of the studied zinc catalyst, the depolymerization rate of PPC is several magnitudes faster than that of PCHC. The limitations of zinc complexes gave cobalt catalysts an advantage in the field of CHO/PO and CO₂ terpolymerizations. In 2006 the first (salen)CoX catalysts with increased activities were discovered [46]. Catalyst 1, in combination with [PPN]Cl, is able to terpolymerize at 25 °C and 15 bar CO₂ pressure equimolar quantities of PO and CHO with activities up to 129 h⁻¹. The resulting polymer had a narrow PDI of 1.24 and a high content of carbonate linkages and showed a single T_g. The authors varied the reaction conditions and were able to tune the CHO content from 30 to 60 % which gave polymers with T_g in the range of 50–100 °C. The highly alternating nature of the polymer was attributed to the inhibiting nature of CHO on PO which leaves both epoxides with matched reactivities in the copolymerization. The regioselective ring opening of PO was not disturbed by the competing binding of the epoxides to the metal center. Another cobalt-based catalyst (7) with covalently bound ionic groups was published in 2010 by Lee et al. They performed a multitude of experiments with various epoxides in different combinations [65]. Terpolymerizations of CO₂/PO/CHO, CO₂/PO/1-hexene oxide (HO), and CO₂/PO/1-butene oxide (BO) were carried out. Activities ranged between 4,400 and 14,000 h⁻¹ and no cyclic carbonate formation was observed during the reactions. Furthermore, an increase in decomposition temperature was achieved by employing a third monomer and the results demonstrated that the T_g of PO/CO₂ can be adjusted at will in a certain range (0–100 °C) by using a third monomer (CHO, HO, or BO). A linear dependency between the glass transition temperature in the terpolymer and the proportions of the third monomer was observed, which makes T_g adjustments possible. For the terpolymer CO₂/PO/HO, the temperature range was found to be from −15 to 32 °C and for CO₂/PO/BO from 9 to 33 °C. The authors also report that the incorporation of further epoxides (styrene oxide (SO), isobutylene oxide (IO), and glycidyl ether (GE)) in the PO/CO₂ copolymer failed under the tested reaction conditions due to remaining impurities in the monomers. Almost at the same time Lu et al. reported a bifunctional cobalt(III) catalyst (3) for the copolymerization of CHO/CO₂ and for terpolymerization with additional aliphatic epoxides such as PO, EO, BO, and HO (Fig. 7.5) [66].

Fig. 7.5

Cobalt catalysts for the co- and terpolymerization of carbon dioxide and epoxides [22, 46, 65–67]

The catalyst showed high activities (TOF) of 1,958–3,560 h⁻¹ (90 °C, 25 bar, 1: 1 = CHO: [PO,BO,HO,EO]) and molecular weights of the resulting terpolymers ranged between 40,000 and 60,000 g/mol (PDI ~ 1.1). The produced polymers had a content of 37–65 % of cyclohexene carbonate linkages and only one T_g was measured (32–79 °C).

Functionalized epoxides open up the possibility for post modification of the resulting carbonate structures. Already in 2007 Coates et al. used β-diiminate zinc(II) acetate catalysts ([(BDI)ZnOAc]₂) for the terpolymerization of CHO, vinylcyclohexene oxide (VCHO), and carbon dioxide to synthesize vinyl-functionalized polycarbonates [43]. The subsequent olefin metathesis using a Grubbs’ catalyst [68] made the transformation of linear polycarbonates into nanoparticles of controlled size possible. The employed epoxides had comparable reactivities, which enabled the generation of terpolymers with a VCHO content of 30 mol%. Medium molecular weights of 54,100 g/mol and low PDI (M _w/M _n = 1.20) were reached. It was observed that the cross-metathesis reaction progressed quickly in the beginning, but due to the formation of increasingly rigid polymer chains slowed down after 15 min. A remarkable increase in the T _g was observed from 114 to 194 °C at 76 % cross-linking. This increase was attributed to reduced chain mobility. A similar approach was used by Coates et al. in 2012 for the synthesis of multisegmented graft polycarbonates [69]. Norbornenyl-terminated multiblock poly(cyclohexene carbonate)s were again synthesized using a BDI zinc catalyst 8 with a norbornene carboxylate initiator (Fig. 7.6). The living nature of the copolymerization allowed the step-by-step addition of functionalized CHO. This generated variable block sequences which could subsequently be transformed to segmented graft copolymers by ring-opening metathesis polymerization.

Fig. 7.6

BDI zinc catalyst with a norbornene carboxylate initiator [69]

Propylene oxide derivatives in terpolymers can also be found in the literature. Frey et al. terpolymerized 1,2-isopropylidene glyceryl glycidyl ether (IGG), glycidyl methyl ether (GME), and carbon dioxide [70]. After the acidic removal of the protecting acetal groups, a polycarbonate poly((glyceryl glycerol)-co-(glycidyl methyl ether) carbonate) (P((GG-co-GME)C)) is received. The analysis of the polymeric material showed monomodal molecular weight distributions and PDIs between 2.5 and 3.3, although the molecular mass was comparably low (12,000–25,000 g/mol). The T _g of the unprotected polymer was 3–5 °C higher than those of the corresponding protected copolymers. Interestingly, during investigations of the degradation behavior of the P((GG-co-GME)C) copolymers in THF solution, no backbone degradation was observed over 21 days. The authors ascribe this observation to the reduced stability of the resulting cyclic carbonate structures.

As can be seen in this chapter, the possibilities for further research are far from exhausted. Apart from the synthesis of ever more refined structures, the analytics must evolve simultaneously to access the microstructure. Therefore, current developments in the field of polymer NMR techniques will be discussed in detail in the next chapter.

2.1.4 Microstructure of Poly(propylene carbonate)

Since its first discovery, poly(propylene carbonate) has been characterized via a multiple analytical methods by various research groups [71–76]. NMR has been used extensively for microstructure analysis because PPC with higher stereoregularity has the potential to increase the glass transition temperature (T _g). In general, the ring opening of PO occurs via an S_N2 type mechanism at the methylene position resulting in a predominantly head-to-tail (HT) regiostructure of the CO₂/PO copolymer. By contrast, tail-to-tail (TT) or head-to-head (HH) connections arise if the ring opening of the epoxide takes place at the methine carbon. In 2002, Chisholm and colleagues assigned ¹³C NMR spectra for different PPC copolymers using a statistical approach [17]. First, they studied the ¹³C NMR spectrum of (S)-PPC obtained from the copolymerization of (S)-PO and CO₂ using a zinc glutarate catalyst. The carbonate carbon region showed more than one resonance, providing information about the possible regiosequences that occur at the tetrad level. Consequently, Chisholm et al. proposed four regiosequences with central head-to-tail (HT) junctions that are distinguishable by NMR for the carbonate carbons (Fig. 7.7).

Fig. 7.7

The four proposed regiosequences with central HT junctions

The other carbonate carbon signals arising from TT and HH junctions at the diad level were assigned as s. Chisholm and colleagues applied the same statistical approach at the triad level for aliphatic carbon in PPC and later on they analyzed the structure of oligoether carbonates [17, 77]. Those compounds were potential models for the PPC microstructural assignments in NMR studies. It was possible to show with ¹³C NMR investigations that the carbonate carbon signals have both regio- and stereosensitivity at the diad and tetrad levels to their adjacent ether units [77]. Density functional theory calculations were applied to the tested oligoether carbonates. These calculations indicated that the carbonate groups exist predominantly in cis-cis geometries with more than one stable conformation for each molecule. In addition, the ¹³C chemical shifts predicted in the calculations were sensitive to the conformations of the molecules and the configurations of the stereocenters in PO ring-opened units. Recently, different poly(propylene carbonate) (PPC) microstructures have been synthesized by Rieger et al. from the alternating copolymerization of CO₂ with propylene oxide (PO) using chiral cobalt and (salen)Cr catalysts [78, 79].

The model which was presented in the study of Rieger et al. was based on triad structure obtained from ¹³C NMR spectroscopy (HT carbonyl region). The ¹³C NMR spectra of selected poly(propylene carbonate) samples were recorded using a 900 MHz (¹H) spectrometer, showing a previously unreported fine splitting of the carbonate resonances. Utilization of enantiopure and enantio-enriched PO allowed to apply the approach of isolated stereoerrors in the predominantly isotactic polymer to assign in detail the observed signals due to the stereo- and regioirregularities (Fig. 7.8). This assignment was performed under consideration of the direction of the polymer chain. Additional GC-analysis of the hydrolyzed polymers has shown a correlation between the extent of regioerror and the ring opening of PO with inversion of configuration at chiral atom, leading to s-HH diads. Signals of i-diads in the HH region were also observed in certain cases, which is referred to an “abnormal” PO-enchainment with retention of configuration of the stereocenters. The assignment of carbonate carbon signals at δ_C 153.8 and 154.8 ppm to HH and TT syndiotactic diads, respectively (Fig. 7.9), is consistent with the previous work of Chisholm and coworkers [80].

Fig. 7.8

¹³C (125 MHz, CDCl₃) NMR spectra (carbonate carbon region): (a) (R)-PPC copolymer, (b) (R)-PPC copolymer synthesized, (c) (S)-PPC copolymer, and (d) (S)-PPC copolymer, 1 equivalent [PPN]Cl, 1,000 eq. PO at 30 °C and 30 bar CO₂ for 20 h [(a) and (d) ((R,R)-(salen)CoCl) (b) and (c) (S,S)-(salen)CoCl] (Reprinted with permission from Ref. [78]. Copyright 2012, American Chemical Society)

Fig. 7.9

Syndiotactic HH and TT in (R)-PPC synthesized using 1, resulting from “abnormal” epoxide ring opening at the methine position [78]

However, in the HT-region, a new low-intensity peak in the vicinity of the ii-signal at the lower field was observed, which could be assigned to an effect of an HH regioerror close to the isotactic triad. Accordingly, all observed low-intensity signals in the polymers with a lower isotacticity (δ_C 154.29, 154.30, 154.35, and 154.39 ppm) correlate well with the resonances in the HH region due to the regioerror (Fig. 7.10). Such correlation was also observed in case of the polymer with a predominantly syndiotactic structure, revealing a low-intensity peak at low field close to an ss-HT-signal.

Fig. 7.10

¹³C NMR spectra of the carbonate carbon region of (a) iso-enriched PPC (125 MHz, CDCl₃), (b) iso-enriched PPC (225 MHz, CDCl₃), and (c) syndio-enriched PPC (225 MHz, CDCl₃) and the proposed assignment of signals on the tetrad level (Reprinted with permission from Ref. [78]. Copyright 2012, American Chemical Society)

Aside from the HT region, the splitting pattern of the signals in the HH region of the spectrum is informative with respect to the copolymer microstructure. The splitting of the signals in the TT region is not pronounced, therefore indicating a weak influence of the neighboring stereoconfiguration on the chemical shift of this junction. It is therefore expected that the HH junction has a stronger effect on chemical shifts in neighboring stereosequences than the TT sequence does.

In case of copolymers with a high extent of regioerrors, the described correlations are not so straightforward but still can be observed (Fig. 7.11). A greater number of peaks in the HT and HH regions are expected for such copolymers, where the model of isolated regioerrors is no longer valid.

Fig. 7.11

¹³C (75 MHz, CDCl₃) NMR spectra for PPC copolymer synthesized using catalyst (salen)CrCl with 0.5 equivalent DMAP, (a) using 75 % (S)-PO and 25 % (R)-PO (b) using 25 % (S)-PO and 75 % (R)-PO at 50 °C and 50 bar CO₂ for 20 h (Reprinted with permission from Ref. [78]. Copyright 2012, American Chemical Society)

Following the (S)-PO/(R)-PO ratios in the polymer chain during the rac-PO/CO₂ copolymerization reaction using GC and 500 MHz NMR spectroscopy provided insight into the understanding of the arising PPC microstructure. The suggested average microstructure of the copolymers shows a gradient change from iso-enriched to stereogradient microstructure with the polymerization time due to the catalyst enantioselectivity. Investigations by Nozaki et al. also led to the synthesis of stereogradient PPC which consisted of two enantiomeric structures on each end of the polymer chain [76]. The used catalyst was an optically active (salen)CoOAc with pendant ammonium arms. Thermal properties of the polymer were analyzed by differential scanning calorimetry (DSC) and thermogravimetry (TG).

Different regio- and stereoregularities did not show an influence on the glass transition temperatures (T _g ~ 33 °C) of the obtained polymers as was also confirmed by Rieger et al. [78]. But the stereogradient PPC and the stereoblock PPC exhibited higher thermal decomposition temperatures of up to 273 °C compared to PPCs with lower regio- and stereoregularities and iso-enriched PPC (T _d = 240 °C). The authors attributed their findings to a possible stereocomplex formation of a (S)-PPC block and (R)-PPC block in the same chain upon precipitation in methanol. Equimolar mixture of (S)-PPC and (R)-PPC did not show increased T _d values which might indicate a spatial dependence of the (S)-PPC and (R)-PPC for the stereoblock formation. Lu et al. achieved the synthesis of >99 % head-to-tail, >99 % carbonate linkage, and isotactic PPC with a multichiral (S,S,S)-(salen)CoNO₃ catalysts with a TBD group (1,5,7-triabicylco[4.4.0] dec-5-ene) anchored on the 5-position of one phenyl ring (Fig. 7.12) [81]. The isotactic PPC produced from (R)-PO exhibited elevated T _g values of up to 47 °C.

Fig. 7.12

(left) (salen)CoOAc 9 is capable of producing stereogradient PPC, (right) (salen)CoNO₃ 10 catalyst with pending TBD arm is capable of producing >99 % HT PPC [76, 81]

2.1.5 Microstructure of Poly(cyclohexene carbonate)

The copolymerization of meso-cyclohexene oxide and carbon dioxide produces poly(cyclohexene oxide) with different stereochemistry depending on the employed catalyst system. The ring opening of a meso-epoxide proceeds with inversion of the configuration at one of the two chiral centers (Fig. 7.13).

Fig. 7.13

Poly(cyclohexene oxide): isotactic (a) and syndiotactic (b)

Nozaki et al. were the first to report asymmetric alternating copolymerization of meso-CHO with carbon dioxide. They used an equimolar mixture of Et₂Zn and (S)-diphenyl(pyrrolidin-2-yl)methanol as chiral catalyst [82]. Through alki-treatment, the PCHC was hydrolyzed into trans-1,2-diol and the degree of asymmetric induction was measured to be 70 % ee. These isotactic PCHCs were characterized via ¹³C NMR spectroscopy using statistical methods and model compounds [83]. An improved dimeric zinc system (11) was able to generate isotactic-enriched PCHC with 80 % ee [84]. End group analysis showed that the insertion of CO₂ happens at the zinc-ethoxy bond of the catalyst. Nevertheless, the obtained stereoregular copolymer (80 % ee) showed a T _g of 117 °C which is very close to that of non-stereoregular PCHC (115 °C) (Fig. 7.14).

Fig. 7.14

Dinuclear zinc catalyst (11) (left), (salen)CoX catalyst (right) for the copolymerization of CHO and carbon dioxide [84, 85]

The first syndiotactic-enriched PCHC was reported by Coates et al. who employed (salen)CoX catalysts for the copolymerization of CHO/CO₂ [85]. Catalyst (salen)CoBr (12) and (salen)CoOBzF₅ (13) were able to produce PCHC with up to 81 % r-centered tetrads. Through Bernoullian statistical methods, the PCHC triad and tetrad sequences were assigned in the ¹³C NMR spectra in the carbonyl and methylene regions. The addition of cocatalyst ([PPN]Cl) resulted in the complete loss in syndiotacticity in the produced PCHC. The synthesis of enantiopure (salen)Co(III) catalysts (Fig. 7.15) by Lu et al. in 2012 made the synthesis of highly isotactic PCHC possible [86].

Fig. 7.15

Disymmetrical enantiopure (salen)Co(III) complexes[86]

Chiral induction agents, such as (S)-2-methyltetrahydrofuran and (S)-propylene oxide, further improved the enantioselectivity regarding (S,S)-(salen)Co(III). PCHC with up to 98:2 RR:SS could be obtained with catalyst 1d/[PPN]Cl in the presence of (S)-2-methyltetrahydrofuran. The highly isotactic PCHC was analyzed by DSC measurements and wide angle X-ray diffraction (WAXD). They found that PCHC samples with less than 90 % isotacticity did not show any crystallization. (R)-PCHC with an RR:SS ratio of 92:8 ((R)-PCHC-92) gave a T _g of 122 °C and a small endothermic melting peak at 207 °C which stems from the low degree of crystallinity of the polymer. Contrary to amorphous PCHC, the (R)-PCHC-92 sample showed sharp diffraction peaks.

Aside from defined stereo- and regioregular synthesis of optical active PPC and PCHC, the catalytic side of the polymer formation has received much attention in recent years. The current development is refocusing on new possible ligand structures with less toxic metal centers which nevertheless can achieve high activities in the copolymerization. Those developments will be presented in the next section.

3 CO₂ Utilization in Polycarbonates

3.1 Development

Before diving into the broad field of polycarbonates and their synthesis and applications, a small review of historic developments is necessary to place the subsequent work in context. The earliest mention of the synthesis of poly(ethylene carbonate) was made in a patent in 1966 by Stevens [87]. However, it was not until 1969 that Inoue et al. presented their findings concerning the activity of a ZnEt₂/water mixture which was able to catalyze the formation of poly(propylene carbonate) from carbon dioxide and propylene oxide [73]. At pressures ranging from 20 to 50 bar and at 80 °C, the heterogeneous catalyst produced low molecular weight carbonates with turnover frequencies up to 0.12 h⁻¹ and sparked a surge in developments over the next decades. In heterogeneous catalysis, Soga et al. synthesized the first well-defined catalyst system from a Zn(OH)₂/glutaric acid mixture, which is still the most active zinc-based system for copolymerization of CO₂ and propylene oxide [88]. By using different zinc carboxylates, Rieger et al. demonstrated the importance of the distance of the active zinc centers for catalytic activity. Problems arose in the evaluation of the active zinc centers, as well as the importance of crystallinity, surface, and particle size for this catalysis [10]. Furthermore, deeper understanding of the reaction mechanism was hampered by the drawbacks of heterogeneous catalysts in general. It is very challenging to define the active sites in such systems and subsequently to ascribe structural alterations to activity increases or decreases. Furthermore, the multitude of active sites in the catalysts leads to broad polydispersity indices (PDI) which makes the characterization of physical properties of the polymer difficult. Consequently, it was not until 1978 that Inoue et al. published the first homogeneous catalyst system for the copolymerization reaction of CO₂ and epoxides [89]. The aluminum tetraphenylporphyrin complexes in combination with EtPh₃PBr as cocatalyst were able to copolymerize not only cyclohexene oxide (CHO) but also propylene oxide (PO) with low PDIs. In 1995, bis(phenoxide)Zn(THF)₂ catalysts were synthesized and employed in the copolymerization of PO and also CHO and CO₂ [47, 90]. This catalyst system exhibited rather low activities and selectivities toward polycarbonate formation. The next important step toward well-defined and tunable catalysts was made by Coates et al. with the development of zinc β-diiminate complexes ((β-diiminate)ZnOR) [91]. This type of ligand allowed adjustment of the steric and electronic environment around the metal center and even small variations in the ligand framework already had drastic influences on activity and selectivity. Recently, Rieger et al. published a new dinuclear zinc catalyst system for the copolymerization of cyclohexene oxide and carbon dioxide. The catalyst 35 limits – in a certain broad pressure regime – the polymerization rate as a function of the applied CO₂ pressure. It is the first time that the rate determining step is shifted toward CO₂ insertion. This unique behavior is attributed to a flexible CH₂-tether that links both complex moieties. As a result, the highest known polymerization activities for CHO-based polycarbonates are reported.

Nowadays, development of these well-defined homogeneous catalysts and research on polycarbonates from carbon dioxide and epoxides are flourishing. Various review articles have been published in recent years in this field of research which demonstrates the growing interest in this material and the underlying mechanism [5, 9, 88, 92–97].

3.2 Mechanistic Aspects of the Epoxide and Carbon Dioxide Copolymerization

Homogeneous catalysts with the general structure L_nMX possess the great advantage of having only one defined active site. Tailoring the organic ligand L_n as well as the initiating group and the metal center provides a broad range of possibilities. Nevertheless, mechanistic investigations so far have not provided an unambiguous picture of the copolymerization of carbon dioxide and epoxides. Hence, three different possible reaction pathways shall be presented for a model salen complex.

The reaction pathway A requires interactions between two catalyst molecules and their active sites (Fig. 7.16). Thus, this mechanism is probable in the absence of a cocatalyst at low epoxide to catalyst loadings. It is rate dependent on the Lewis acidity of the corresponding metal center and on the nucleophilicity of the axial ligand X. Accordingly, the type of catalytic system used defines the mechanism. In the work of Jacobsen on the asymmetric nucleophilic ring opening of epoxides by chiral (salen)CrX complexes, this intermolecular, bimetallic pathway is a vital step in the absence of a cocatalyst [98]. The group of Rieger et al. also performed theoretical calculations concerning the chain-growth mechanism during the copolymerization of CO₂ and epoxides. The DFT calculations indicated that chain growth proceeds through the attack of a metal-bound alkyl carbonate on a precoordinated epoxide on a metal center [99]. However, other investigations indicated a bimetallic initiation step and subsequent monometallic chain propagation [71, 100–102]. Reaction pathway B is the monometallic mechanism and is comparable to an associative ligand exchange mechanism. In this scenario, the nucleophile and the epoxide are both bound to one metal center. In general, the epoxide is opened at the least hindered C-O bond but regioerrors may occur. Both the axial ligand and the nucleophile have the potential to open the epoxide and are therefore incorporated in the polymer chain as end groups. The thermally disfavored transition state makes this reaction mechanism unlikely for commonly used catalysts. Reaction pathway C takes place in the case of binary catalyst/cocatalyst systems where the added nucleophile attacks the precoordinated epoxide and opens the ring. Binary catalyst/cocatalyst systems have been under scrutiny from multiple research teams around the world in recent times [46, 85, 103–116].

Fig. 7.16

Initiation mechanism for L_nMX for the copolymerization of carbon dioxide and PO: bimetallic pathway (a), monometallic pathway (b), and binary pathway (c) (Reprinted with permission from Ref. [88]. Copyright 2011, Elsevier)

The group of Coates et al. was able to show with (β-diiminate)ZnOR catalysts that a bimetallic mechanism is at play during their reactions [117]. Furthermore, they were able to find the rate determining step as the incorporation of the epoxide. The weak nucleophilic nature of the carbonate end group makes a preactivation of the epoxide necessary. Therefore, Lewis acids in the form of cocatalysts or a second metal center have to be used. After the ring opening of the epoxide, CO₂ is inserted into the metal alkoxy bond, for which an open coordination site at the metal is not obligatory. This mechanism explains the generally observed loss of activity of cocatalyst systems at higher dilutions (Fig. 7.17).

Fig. 7.17

Bimetallic propagation and the possible concomitant side reactions

Undesired side reactions during the copolymerization are the formation of cyclic carbonates and the formation of polyether segments. The group of Darensbourg studied in detail the reasons for the product selectivity and the dependence on temperature, pressure, cocatalyst, and catalyst structure [101]. In general, both low temperatures and stronger coordination to the metal center decrease the amount of cyclic byproduct since these factors can suppress backbiting. During the backbiting reaction either the carbonato- or alkoxy-chain end of the polymer attacks at the growing chain (Fig. 7.18). This reaction is especially pronounced for aliphatic epoxides [63]. Alternatively, the growing chain end may dissociate from the metal center and more easily undergo backbiting in its unbound state [99]. The protonation of this unbound alkoxy- or carbonato-chain end is a possible strategy to reduce backbiting.

Fig. 7.18

Chain backbiting mechanism [88]

Chain transfer reactions during the copolymerization can be caused by traces of water, alcohols, or acids. Chain transfer can be exploited to tune the molecular weight by addition of, for example, adipic acid or telechelic polymers featuring alcohol end groups [118]. These chain transfer reactions lead to lower molecular weights than theoretically calculated. Furthermore, phosphoric acid and phenylphosphonic acids as additives have been tested to afford flame-retarding PPCs [42].

3.3 Catalysis

3.3.1 Bifunctional Catalysts

Owing to the desire to reach high selectivities and activities, catalyst design has shifted in recent years toward bifunctional catalysts. These complexes combine the active metal site, coordinated by an organic ligand framework, with a covalently bound cocatalyst. With these considerations in mind Nozaki et al. incorporated a piperidinium arm into a (salen)CoOAc complex. It was proposed that cyclic carbonate formation is inhibited by the protonated piperidinium arm which can protonate the growing polymer chain upon dissociation from the metal center (Fig. 7.19). At a catalyst loading of PO/catalyst = 2,000, 14 bar and ambient temperatures, a selectivity of 99 % for poly(propylene carbonate) was achieved. Furthermore, it was shown that even at elevated temperatures of 60 °C, the cyclic carbonate formation could be suppressed to a certain degree [67].

Fig. 7.19

Copolymerization mechanism for a bifunctional salenCo(III) complex with a piperidinium arm [67] (Reprinted with permission from Ref. [88]. Copyright 2011, Elsevier)

The group of Lee et al. designed a series of ionic bifunctional catalyst structures in which the cocatalyst is covalently bound and is supposed to keep the growing polymer chain close to the reactive center [22, 119]. The anions also function as initiators for the copolymerization and have a pronounced effect on polymerization activity and product selectivity [120]. These catalysts were employed in copolymerization reactions with various epoxides, for example, cyclohexene oxide, hexene oxide, 1-butene oxide, and propylene oxide [48]. Polymerizations of PO and carbon dioxide at 80 °C and at dilutions as low as 0.67 mmol% were shown to be possible while retaining high activities and selectivities for the polycarbonate formation (TOF 12,400 h⁻¹, 96 %) [22]. The highest reported TOF is 26,000 h⁻¹ (>99 %, M_n = 114 kg/mol, PDI = 1.29). It was possible to produce high molecular weight PPC at elevated temperatures with a M_n of up to 285 kg/mol and narrow molecular weight distributions of 1.18. Interestingly, the catalyst structure allowed the removal from the produced polymer after the copolymerization.

Filtration through a short pad of silica gel results in a catalyst which exhibits almost equivalent copolymerization activities even after several separation cycles. In their later work the group of Lee discovered an unusual coordination mode in their bifunctional catalysts. Though no crystal structure has yet been obtained from this proposed conformation, NMR studies and DFT calculations support the proposed structure. They postulated that the imine nitrogens of the salen ligand do not coordinate to the cobalt center (Fig. 7.20). Instead the DNPs coordinate to the cobalt, forming a negatively charged cobaltate complex. Structural variations of the ligand framework allowed the conclusion that sterically more demanding substituents in ortho-position reduce the activity. The authors attributed this to the fact that the formation of the aforementioned bidentate cobalt coordination is inhibited. High copolymerization activities were only observed on catalysts with the ability to form the cobaltate complex (Fig. 7.20 (right)) [121]. The employed 2,4-dinitrophenolate is explosive in its dry state as well as hazardous. Therefore, anion variations were performed using 2,4,5-trichlorophenolate, 4-nitrophenolate, 2,4-dichlorophenolate, and nitrate [118, 120]. It was found by chance that homoconjugated anion pairs of the phenols [X-H-Y] show increased activities and higher tolerance to water impurities leading to shorter induction periods [121]. Interestingly the rather long induction periods of the cobalt complexes during the copolymerization of epoxides and carbon dioxide are still unexplained. Further examples of bifunctional catalysts for the production of polycarbonates were published by Lu et al. These (salen)CoX complexes feature either an alkyl arm with pending quaternary ammonium salt at the 3-position of the phenyl ring or an alkyl arm bound to 1,5,7-triabicyclo[4,4,0] dec-5-ene (TBD). Catalyst 4 with the TBD group was able to polymerize with high selectivity for polycarbonate (97 %) at 100 °C and 20 bar. Activities up to 10,880 h⁻¹ at a loading of 1 mmol% were reached [122]. Through mass spectroscopy, it was possible to show that the pending TBD arm is activated after the insertion of one PO and carbon dioxide molecule. It stabilizes the Co(III) species against reduction to the inactive Co(II) by reversible intramolecular Co-O bond formation (Fig. 7.21).

Fig. 7.20

Bifunctional ionic salenCo(III) catalyst 5 and its proposed bidentate coordination structure [121]

Fig. 7.21

Bifunctional salenCo(III) catalysts [22, 66, 67]

The Co(III) catalyst (3) with an ammonium salt (-NEt₂Me⁺) and 2,4-dinitrophenolate proved to be less active with TOF of 3,900 h⁻¹ at 90 °C while nevertheless maintaining high selectivity for copolymer formation.

Bifunctional cobalt catalysts are currently the catalysts of choice for further scientific studies and improvements due to their excellent activities combined with high selectivities for the polymer formation. However, their main drawbacks cannot be solved by synthetic design as cobalt is toxic, prone to reduction, and has to be removed from the polymer after copolymerization. This puts a considerable energetic burden on the produced polymer and hampers its chances to compete with already established polymers.

3.3.2 Zinc Catalysts

Due to their colorlessness, nontoxicity, and low price, it is not surprising that there is a considerable amount of literature on Zn(II)-based catalysts. For heterogeneous catalysis, the zinc glutarate (ZnGA) system is the most widely studied one. It is easy to prepare, nontoxic, economically viable, and easy to handle. But these heterogeneous catalysts are not only interesting for industrial applications as they offer the opportunity to better understand the reaction processes. The molecular structure of ZnGA was only recently reported and showed a unique structural constitution [123]. It consists of Zn centers coordinated by four carboxyl groups with the glutarate ligands either in a bent or an extended conformation. This structural type limits the activity for the copolymerization to the surface due to restricted monomer diffusion [124]. Consequently, methods for enlarging the surface were tested but with unsatisfactory results with respect to the achieved activity. Also the replacement of GA by derivatives (e.g., 2-ketoglutaric acid, 3,3-dimethylglutaric acid) with other functionalities in the hydrocarbon chain did not bring the expected success [125]. Experiments with different alkyl lengths (succinic acid (SA), adipic acid (AA), pimelic acid (PA)) in between the Zn centers showed that these are also active [126, 127]. It was not until 2011 that Rieger et al. published a detailed study of ZnSA, ZnGA, ZnAA, and ZnPA and their respective activities in the copolymerization of epoxides and CO₂ [128]. From the copolymerization experiments, the activity gap between ZnSA and the higher homologues was identified. The molecular structure was the crucial factor as the main difference between ZnSA and the tested homologues was the Zn-Zn distance. This indicated that the defined spatial structure which is influenced by the dicarboxylic acid has direct influence on the activity of heterogeneous zinc dicarboxylate systems. The Zn-Zn distance that is necessary for copolymerization is only found on one of the main hkl-indiced plains of the solid-state structure of ZnSA. However, in the solid-state structure of ZnGA, the corresponding distance of 4.6–4.8 Å is found on each main hkl plain (Fig. 7.22). These results show that two zinc centers have to be in close spatial proximity for high copolymerization activities to be possible. Theoretical calculations suggest that the optimal Zn-Zn distance for CO₂/epoxide copolymerization is between 4.3 and 5.0 Å [129].

Fig. 7.22

Close-up view on unit cell and metal distances in zinc glutarate (Zn-Zn distance = 4.6-4.8 Å) (Reprinted with permission from Ref. [128]. Copyright 2011, American Chemical Society)

This range gives an optimum between the activation energy and product selectivity toward polymer formation. The importance of the appropriate Zn-Zn distance for homogeneous catalysts was already described in literature. The study of Rieger et al. indicates that also in heterogeneous Zn catalysts a bimetallic mechanism is at play on the surface for which the spatial separation of the metal is crucial. To gain deeper insights into the copolymerization mechanism, homogeneous single-site catalysts are required. Darensbourg et al. synthesized the first homogeneous zinc-based catalyst for the copolymerization of cyclohexene oxide and CO₂ [47]. These catalysts exhibited a strong tendency to form polyether even at elevated CO₂ pressures (55 bar). Major drawbacks of those early systems were the low activities and selectivities for polycarbonate formation. In the case of the phenoxide zinc complexes, another side reaction limited success as their phenoxide ligands were consumed as initiators during the polymerization [51, 130]. The unstable nature of the zinc phenoxide catalysts made mechanistic investigations and the defined synthesis of the ligand framework around the metal center difficult. The first breakthrough for copolymerization of epoxides and carbon dioxide was the discovery of β-diketiminato zinc catalysts (BDI) for the CHO/CO₂ copolymerization by Coates et al. which led to a more systematic catalyst design. Minor variations in the electronic and steric character of the BDI ligand framework resulted in dramatic changes in catalytic activity [47, 48]. These catalysts afforded polymers with narrow PDIs at low carbon dioxide pressures with high activities. In the copolymerization reaction of CHO/CO₂, minor changes in the backbone from R¹ = H, R², R³ = Et to R¹ = CN, R² = Me and R³ = iPr led to an activity increase from 239 to 2,290 h⁻¹ [91]. As shown in Fig. 7.23, these complexes are present in monomer/dimer equilibria which depend on sterics, electronics, and temperature.

Fig. 7.23

Monomer/dimer equilibria for the zinc-based catalysts by Coates et al. [91]

Rate studies with in situ IR spectroscopy on the copolymerization resulted in a zero-order dependence in CO₂, a first-order dependence in CHO, and a varying dependency from 1.0 to 1.8 for zinc [117]. This variation can be constituted with the appearance of the mentioned equilibrium in Fig. 7.23. All results support the assumption that two zinc centers are interacting in the copolymerization and that the rate determining step is located at the ring opening of the epoxide. Additionally, it was found that sterical variation (R¹ = Et, R² = iPr) at the aniline moiety and the introduction of electron withdrawing groups to the catalyst (20) backbone gave an active catalyst for the copolymerization of PO/CO₂ at 25 °C, 6.9 bar with a TOF of 235 h⁻¹. The resulting polymer featured a regioirregularity and a narrow molecular weight distribution [131] (Fig. 7.24).

Fig. 7.24

Zinc catalyst for copolymerization of propylene oxide and carbon dioxide [131]

Interestingly this complex structure also polymerized β-butyrolactone and β-valerolactone to afford atactic poly(3-hydroxybutyrate) (PHB) and poly(3-hydroxyvalerate) [132]. Research into the zinc BDI systems led to the discovery of ethylsulfinate as superior initiating group [133]. In 2005 Lee et al. presented an anilido-aldimine ligand for dinuclear zinc complexes (Fig. 7.25). Thereby, open (22) and closed (21–24) structures were synthesized with zinc-zinc distances of between 4.88 Å for the open structure and 4.69 Å for the closed structure. For the copolymerization of cyclohexene oxide and carbon dioxide, only the catalyst with the open structure was active. With this catalyst TOFs up to 200 h⁻¹ were achieved. Increasing the Lewis acidity of the zinc centers led to an increase of activity up to 2,860 h⁻¹ at high dilutions ([catalyst]:[CHO] = 1:50,000) [134, 135].

Fig. 7.25

Dinuclear anilido-aldimine zinc complexes [134, 135]

In 2011, Williams et al. published a rigid dinuclear zinc complex for low-pressure copolymerization of cyclohexene oxide and CO₂. Investigations of this catalyst system also revealed a zero-order dependence on CO₂ (between 1 and 40 bar CO₂ pressure) and a first-order dependence on CHO. These results confirmed the hypothesis that the incorporation of the epoxide is rate determining for their dinuclear rigid zinc catalysts [136]. This implies that at typical process conditions, an increase in CO₂ concentration does not automatically enhance polymerization activity. In subsequent work, Williams et al. replaced zinc with cobalt(II)/(III) [137] and iron [138] in the reduced Robson-type ligand structure (Fig. 7.26). The isolated complexes were all able to copolymerize CHO and carbon dioxide with good activities and high selectivities. The highest activities in the copolymerization of CHO and CO₂ were achieved with magnesium (31). Activities of up to 750 h⁻¹ and very high selectivities of >99 % for the copolymer formation were observed with this structure (12 atm CO₂, [CHO]:[31] = 1:10,000, PDI = 1.03/1.10). The data gathered from these metal screenings show very nicely how the copolymerization activity can be tuned in dinuclear catalysts by the Lewis acidity of the metal itself.

Fig. 7.26

Dinuclear reduced Robson-type complexes with Zn-, Mg-, Co-, and Fe-centers [64, 136–139]

In 2013, Rieger et al. presented a flexibly tethered dinuclear zinc complex 35, in which two BDI zinc units are linked by a tether (Fig. 7.27). The linker allows to overcome the entropically disfavored aggregation of two individual complex molecules in solution. Catalyst 35 exhibits very high activities of up to 9,130 h⁻¹ for the copolymerization of CHO and carbon dioxide (at 40 bar CO₂, 100 °C, [CHO]:[35] = 4,000:1) [140].

Fig. 7.27

Flexibly tethered dinuclear zinc complex [140]

Kinetic measurements in an in situ IR reactor revealed a first-order dependence in catalyst. Interestingly, with constant catalyst and cyclohexene oxide concentration, the order in carbon dioxide was determined to be one in the range of 5–25 bar CO₂ pressure. For 25–45 bar, this order changes from one to zero. The assignment of the order in cyclohexene oxide was performed at two different pressure regimes: 10 and 30 bar, respectively. This resulted in a reaction order of zero for 10 bar and a reaction order of one at 30 bar. The subsequent rate laws are depicted below:

Equation 7.1: Rate law of complex 35 for 5–25 bar CO₂ (a) and for 25–45 bar CO₂ (b) [140]

$$ \begin{array}{ccc}\mathrm{r}=\mathrm{k}\cdot {\left[\mathrm{CHO}\right]}^0\cdot {\left[{\mathrm{CO}}_2\right]}^1\cdot {\left[\mathrm{Catalyst}\right]}^1& 5-25\kern0.24em \mathrm{bar}\kern0.24em {\mathrm{CO}}_2& \left(\mathrm{a}\right)\\ {}\mathrm{r}=\mathrm{k}\cdot {\left[\mathrm{CHO}\right]}^1\cdot {\left[{\mathrm{CO}}_2\right]}^0\cdot {\left[\mathrm{Catalyst}\right]}^1& 25-45\kern0.24em \mathrm{bar}\kern0.24em {\mathrm{CO}}_2& \left(\mathrm{b}\right)\end{array} $$

Catalyst 35 is the first dinuclear zinc catalyst which shows a shift in the rate determining step from ring opening of the epoxide to carbon dioxide insertion for the copolymerization of cyclohexene oxide and carbon dioxide in a certain pressure regime. This unique behavior is attributed to the flexible CH₂-tethers which links both complex moieties. Considering the first-order dependence with respect to CHO of all other catalysts, this suggests that the flexible tether between the two zinc centers facilitates the cooperative ring-opening step in an unprecedented manner. These results are supported by quantum chemical calculations. The computed results for the main steps, beginning from the dicarbonato complex as the starting point, followed by coordination of CHO, epoxide ring opening to form an alkoxide and subsequent insertion of CO₂ into the coordinated alkoxide bond, are summarized in Fig. 7.28.

Fig. 7.28

Enthalpy and Gibbs free energy profile for the whole catalytic cycle of catalyst 35 (Reprinted with permission from Ref. [140]. Copyright 2013, Wiley-VCH Verlag GmbH & Co. KGaA)

As can be seen in Fig. 7.28, both elementary steps give similar computed G-values (~100 kJ/mol) which means that both transition states are equally difficult to overcome. It is notable that the enthalpy H is much bigger for the ring opening (TS(a) = 52.3 kJ/mol) compared to the H of the CO₂ insertion (TS(b) = 19.2 kJ/mol). At first glance, it appears surprising that elementary steps with such dissimilar activation energies can both be rate limiting. DFT calculations explain the similar activation energies as a consequence of Gibbs free energy which includes both activation enthalpy and entropy. The main point is that all alkoxide species are significantly higher in G if compared to the dicarbonato complex I, which is why all Gibbs free energies have to be considered in relation to complex I. As the resting state is in all cases the dicarbonato complex I, oxirane ring opening is a bimolecular reaction, i.e., the incorporation of one liquid species into the catalyst bound polymer chain. Otherwise, CO₂ insertion is effectively a trimolecular reaction, which consumes both one liquid epoxide and one gaseous CO₂. All the structural and electronic features of 35 lead to high copolymerization activities of over 9,130 h⁻¹ and selectivities of >99 % toward polycarbonate formation.

In conclusion, zinc and other nontoxic, cheap, and abundant metals have tremendous advantages compared to other transition metals, namely, cobalt and chrome, which are currently predominantly employed for the copolymerization of carbon dioxide and epoxides. In recent publications, it was nicely shown how those economic and eco-friendly elements can achieve high activities and selectivities. An additional advantage of zinc and magnesium is their colorless ions. This opens up the possibility to leave the catalyst in the formed polymer and reduce work-up costs.

4 Conclusion

Carbon dioxide as feedstock for the generation of biodegradable polycarbonates has been known in the literature for almost 45 years. Since the discovery of the copolymerization reaction of carbon dioxide and epoxides, chemistry has made significant progress, and especially in the last 10 years, the search for better and more eco-friendly catalysts has gained momentum. As the stereoselective synthesis of poly(propylene carbonate) is still in its infancy and the analytical methods used have not yet been fully exploited, NMR techniques have the potential to help interpret the structures produced by catalyst systems and to pinpoint promising catalyst candidates for further development. Though heterogeneous catalysts still dominate industrial production processes, for mechanistic investigations, homogeneous systems do have the necessary advantages to tackle the underlying questions about the mechanism and within the catalysis. Cobalt(III) catalysts with a variety of accompanying ligand structures have become the center of attention since the discovery of bifunctional structures which allow the incorporation of the cocatalyst into the complex and thereby enable high activities and the production of polycarbonates with high molecular masses, low PDIs, and >99 % carbonate content. However, cobalt as toxic and coloring metal has to be removed from the polymer after copolymerization. Any additional work-up increases the energetic burden on production and limits competitiveness of biodegradable polymers such as poly(propylene carbonate). Possible applications in food packaging also make the development of less toxic catalysts desirable. An applicable catalyst for large-scale industrial production in a continuous process has to exhibit a sufficient activity to make it economically viable. In recent years, research made great progress in the development of iron-, magnesium-, and zinc-based systems which show improved activities. It will be interesting and exciting to see the future development of catalyst systems which tolerate water contaminations and multiple epoxides and nevertheless show high activities in co- and terpolymerizations at ambient conditions.