CO₂, which is widely known as one of the main greenhouse gases, could potentially be used as building block for the construction of useful aliphatic polycarbonates. Aliphatic polycarbonates are becoming increasingly important because of their extensive range of applications in the production of polyurethanes, coating materials, biodegradable surfactants and biomedical materials ^{[1, 2]}. Aliphatic polycarbonates are mainly prepared by transesterification ^[3] and CO₂ coupling ^[4] reactions. A wide variety of catalytic systems have been developed during the last three decades for the efficient copolymerization of CO₂ with epoxides, including ZnEt₂-protic compounds ^{[5, 6]}, zinc glutarate or adipate ^{[7, 8, 9]}, rare-earth catalysts ^{[10, 11, 12]}, metal-porphyrins ^{[13, 14, 15]}, zinc bis(β-diiminates) ^{[16, 17, 18]}, salen-metal systems, metal microporous polymers ^{[19, 20, 21, 22, 23, 24, 25]} and double metal cyanide (DMC) catalysts ^{[26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37]}. DMC catalysts in particular have promising industrial prospects because of their high activity, cost-effectiveness and insensitivity to moisture.

DMC catalysts were initially used in the production of polyether and later utilized in the copolymerization of propylene oxide with CO₂ ^{[24, 25, 26, 27, 28, 29, 30]}. DMC catalysts are typically prepared via a precipitation process with a heavy organic complexing agent such as tert-butyl alcohol (t-BuOH), followed by multiple washing and centrifugation stages to allow for the removal of the excess potassium ions ^{[26, 27, 28, 29]}, which can have an adverse impact on the copolymerization of CO₂ and propylene oxide ^[29]. A new process was recently developed to avoid the production of K⁺ where H₃Co(CN)₆ was prepared according to an ion exchange method using K₃Co(CN)₆, followed by precipitation with ZnCl₂ in methanol to give [ZnCl]⁺₂[HCo(CN)₆]^2- without any K^{+ [29]}. This material was subsequently used in the copolymerization of CO₂ with propylene oxide to give a copolymer product with a high molar fraction of CO₂ (about 60%). Sebastian et al. ^[30] prepared a series of Zn-Co DMC catalysts with K⁺ contents in the range of 0.41%-2.0% without the inclusion of an additional washing step. However, the highest catalytic efficiency achieved with these Zn-Co DMC catalysts for the copolymerization of CO₂ and cyclohexene oxide was only 52.8 g polymer/g catalyst. In light of the problems posed by the presence of K⁺ in these catalysts, significant research efforts have been directed towards the development of new techniques for the removal of free potassium ions from heterogeneous DMC catalysts. However, the traditional methods for the preparation of DMC catalysts require multiple washing and centrifugation (up to seven times), making these processes time consuming, inefficient and poorly reproducible ^{[26, 27, 28, 31, 32]}.

It is noteworthy that 18-crown-6 ether has an appropriate hole size (0.26-0.32 nm) for the binding of K⁺ (0.266 nm). Herein, we describe the results of our study towards the preparation of DMC catalysts in the presence of 18-crown-6 with the aim of simplifying the washing and centrifugation steps. The resulting crown ether complexing Zn-Co DMC catalysts were subsequently evaluated as catalysts for the copolymerization of CO₂ with propylene oxide.

Potassium hexacyanocobaltate(III), 18-crown-6, zinc chloride and tert-butyl alcohol were purchased as the analytical grade from Aladdin Reagent (Shanghai, China) and used without further purification. Propylene oxide was refluxed over calcium hydride and then stored over 3Å molecular sieves prior to being used in the copolymerization reactions. CO₂ (>99.9%) was purchased from Kedi Air Chmical (Foshan, China) and used without further purification.

The methods used for the preparation of the catalysts are summarized in Table 1. CDMC1 was prepared according to Ref. ^{[28, 32]} with minor modifications. Briefly, a solution of K₃[Co(CN)₆] (1.66 g, 5 mmol) in distilled H₂O (25 mL) was added in a dropwise manner to a vigorously stirred suspension of ZnCl₂ (2.5 g, 18.4 mmol) in a mixture of distilled H₂O (75 mL) and t-BuOH (40 mL) at 45 °C over 30 min, and the resulting mixture was stirred for 2 h. 18-Crown-6 (8 g, 30.3 mmol) was then added to the reaction in a single portion, and the resulting mixture was stirred for 3 h to allow for the complete complexation of the K⁺ ions. The resulting white precipitate was isolated by centrifugation (25 °C, 15 min), suspended in t-BuOH (140 ml) and stirred for 2 h at 45 °C. The mixture was then purified by centrifugation (25 °C, 15 min) to give the catalyst, which was dried under vacuum at 50 °C to constant weight, resulting in a cake. The top and bottom portions of the CDMC1 cake were labeled as CDMC1-T and CDMC1-B, respectively. All of the other catalysts prepared were defined in a similar manner. Elemental analysis of CDMC1 revealed 21.6% Zn, 8.1% Co and 1.2% K.

Table 1 Conditions used for the preparation of the different DMC catalysts a. Catalyst Times of washing Crown ether:K+ molar ratio CDMC1 1 2 CDMC2 1 1 CDMC3 2 2 DMC0 1 0 DMC1 1 0 DMC2 7 0 a K3[Co(CN)6] (1.66 g) was used at 45 °C. ZnCl2 (2.5 g) was used in CDMC1, CDMC2, CDMC3, DMC1 and DMC2 catalysts. A smaller amount of ZnCl2 (1.02 g) was used in DMC0.

Table 1 Conditions used for the preparation of the different DMC catalysts a.

CDMC2 and CDMC3 were prepared according to a procedure similar to CDMC1 but with different crown ethers and washing times with a 1:1 (v/v) mixture of t-BuOH and water.

DMC1 and DMC2 were synthesized according to a similar procedure to that used for CDMC 1 but without addition of the crown ether ^{[28, 32]}. In the case of DMC2, the remaining K⁺ ions were removed by washing the precipitate seven times with a mixture of distilled H₂O and t-BuOH whilst decreasing the quantity of H₂O and increasing the amount of t-BuOH. Elemental analyses revealed the following results: DMC1 (Zn 25.14%, Co 8.53%, K 1.43%); DMC2 (Zn 30.85%, Co 9.81%, no K⁺ ions were detected). DMC0 consisted of pure Zn₃[Co(CN)₆]₂, which was prepared in the absence of a complexing agent.

Fourier transform infrared spectra (FTIR) were measured on an Analect RFX-65A FTIR spectrophotometer (Analect, USA). The surface morphologies of the DMC catalysts were observed by scanning electron microscopy (SEM) on a JSM-6360LV scanning electron microscope (JEOL, Japan). Prior to scanning, the top and bottom portions of the DMC and CDMC cakes were coated with a thin layer of gold under high vacuum conditions. X-ray powder diffraction (XRD) patterns were collected on a Bruker D8 Advance diffractometer (Bruker, Sweden) using monochromated radiation. The chemical compositions of the catalysts were obtained by elemental analysis using a Vario EL CHNS elemental analyzer (Elementar, Germany) equipped with a PE5100 ICP-AES instrument (Perkin Elmer, USA). A Thermogravimetric analysis/infrared spectra (TGA-IR) were recorded on a TG-209/Vector^TM-22 system (Netzsch, Germany). ¹H-NMR spectra of the copolymers were recorded on a DRX-400 spectrometer (400 MHz) using CDCl₃ as a solvent. The molecular weights and molecular weight distributions of the copolymers were determined by gel permeation chromatography using tetrahydrofuran as a solvent with on a Waters 515-410 system (Waters, USA).

Several copolymerization experiments were conducted in a 130 mL autoclave to evaluate the catalytic performance of the catalysts. The autoclave was loaded with a catalyst, placed under vacuum at 25 °C and purged with N₂ three times to allow for the removal of any moisture. The autoclave was then cooled to room temperature and charged with propylene oxide using a syringe before being pressurized with CO₂ to the desired pressure. Upon completion of the reaction, the autoclave was cooled to room temperature and the CO₂ released.

The C≡N groups in the FTIR spectra of Zn-Co DMC catalysts generally give strong stretching vibrations (ν_C≡N) around 2196 cm^-1 following the coordination of t-BuOH, whereas the corresponding ν_C≡N value of pure Zn₃[Co(CN)₆]₂ is 2185 cm^{-1 [32]}. As shown in Fig. 1, the FTIR spectra of CDMC1 and CDMC3 still contained a peak at 2195 cm^-1 following the addition of 18-crown-6. Structural transformations of this type are generally thought to originate from the effective coordination of Zn with O for the polymerization process. Notably, the stretching vibration of the Co-C bond in pure Zn₃[Co(CN)₆]₂ (DMC0), which is generally observed at 451 cm^-1, was appeared at 471 and 472 cm^-1in CDMC1 and CDMC3, respectively. Furthermore, the peaks at 1107 and 1109 cm^-1in CDMC1 and CDMC3 were attributed to the stretching vibrations of the C-O-C bonds of 18-crown-6. Although different amounts of 18-crown-6 were added to CDMC1 and CDMC3, the IR curves of these materials were almost identical except for minor differences in the stretching vibrations of the C-O-C and Co-C bonds.

Fig. 1. FTIR spectra of CDMC1 (1), CDMC3 (2) and DMC0 (3).

The FTIR data for the stretching vibrations of the C≡N, Co-C and C-O-C bonds of the CDMC and DMC catalysts are shown in Table 2. These results revealed that the FTIR spectra of the top and bottom surfaces of CDMC1 and CDMC3 cakes, which were prepared with a crown ether to K⁺ molar ratio of 2:1, were the same. This indicated that the cakes of the CDMC1 and CDMC2 materials were totally uniform in terms of their composition. In contrast, the FTIR spectra of the top and bottom surfaces of the DMC1 and DMC2 cakes, which were prepared without any crown ether, revealed that the peaks corresponding to the stretching vibrations of the C≡N bonds were different, which indicated that the materials were unevenly distributed in terms of their composition. Polymers containing ether groups [e.g., poly(tetramethylene ether glycol) and polyethylene glycol] can be used as complexing agents to improve the activity of DMC catalysts ^{[32, 33]}. For the CDMC catalysts, it was envisaged that the addition of 18-crown-6 would not only lead to the binding of the K⁺ but that it would also coordinate Zn²⁺. In this way, the 18-crown-6 could perform as a co-complexing agent, which could improve the homogeneity of the catalysts. FTIR analysis of CDMC3 revealed stretching vibrations for the C-O-C bonds of the 18-crown-6 at 1107 cm^-1, whereas the corresponding band appeared at 1109 cm^-1 in CDMC1. This result therefore illustrates the differences in the complexation states of the different catalysts. In fact, the stretching vibrations of the Co-C bonds in all of the catalysts prepared in the current study except for DMC0 were in the range of 471-472 cm^-1. The similarity in these values could be attributed to the Co-C bonds being positioned inside of the molecules so that they would not be affected by changes in the chemical environment. In summary, these results show that the addition of 18-crown-6 led to significant improvements in the homogeneity of catalysts.

Table 2 IR wavenumber data of νC≡N, νCo-C and νC-O-C in Zn-Co DMC catalysts. Catalyst νC≡N (cm-1) νCo-C (cm-1) νC-O-C (cm-1) CDMC1-T 2195 471 1109 CDMC1-B 2195 471 1109 CDMC2-T 2195 471 1109 CDMC2-B 2195 471 1109 CDMC3-T 2195 472 1107 CDMC3-B 2195 472 1107 DMC1-T 2197 472 — DMC1-B 2195 472 — DMC2-T 2197 472 — DMC2-B 2195 472 — DMC0 2185 451 —

Table 2 IR wavenumber data of νC≡N, νCo-C and νC-O-C in Zn-Co DMC catalysts.

The surface morphologies of the different catalysts were clearly observed by SEM (Fig. 2). The surface of the CDMC3 catalyst prepared in the presence of the crown ether was relatively dispersive with a large surface area. For the CDMC3-T and CDMC3-B catalysts, their surfaces were disordered and multilayered, which indicated that the CDMC3 was even. However, the surface structures of the top and bottom portions of the DMC1 and DMC2 catalysts were found to be different, which indicated that these materials were not uniform, were consistent with those of the FTIR spectra.

Fig. 2. SEM images of CDMC3-T (a), CDMC3-B (b), CDMC3-T (c), CDMC3-B (d), DMC1-T (e), DMC1-B (f), DMC2-T (g) and DMC2-B (h).

TGA-IR spectra of the thermal decomposition products resulting from CDMC1 and DMC1 were recorded at different stages of the heating process to confirm the binding states of the t-BuOH and crown ether ligands in these systems based on weight loss, and the results are shown in Table 3.

Table 3 Thermal decomposition stages of DMC1 and CDMC1. DMC1 CDMC1 Binding state Decomposition products Temperature range (°C) Weight loss (%) Temperature range (°C) Weight loss (%) 20-123 2.4 20-110 0.9 Free H2O and t-BuOH 123-357 15.4 110-390 17.8 Coordinated t-BuOH and alkene 357-775 26.2 390-770 20.4 Coordinated or bonding Alkene, cyanide and alkyne Ash 44.0 Ash 40.2 — Metallic oxide

Table 3 Thermal decomposition stages of DMC1 and CDMC1.

The decomposition profiles of DMC1 and CDMC1 were divided into three stages. The first of these stages occurred from 20 to about 120 °C. DMC1 and CDMC1 contained peaks at 3670, 2960, 1375 and 1059 cm^-1, which were assigned to the ν_O-H, ν_C-H, ν_C-H and ν_C-O vibrations of t-BuOH (Fig. 3). These results indicated that free H₂O and t-BuOH were being released during the first stage of the thermal decomposition process.

Fig. 3. FTIR spectra of gas mixture derived from thermal decomposition of DMC1 (a) and CDMC1 (b). (1) 94 °C; (2) 206 °C; (3) 326 °C; (4) 500 °C; (5) 625 °C. (c) TGA curves of DMC1 (1) and CDMC1 (2).

The second stage of the thermal decomposition process occurred from 120 to about 360 °C and corresponded to the loss of coordinated ligands (i.e., t-BuOH and/or 18-crown-6). The presence of an alkene analogue was confirmed by FTIR analysis, which revealed peaks at 3080 and 1646 cm^-1 corresponding to ν_=C-H and ν_C=C at 123 °C above. Further evidence of an alkene was provided by an increase in the intensity of the characteristic peaks of the alkene above 326 °C, as shown in Fig. 3(a). The t-BuOH molecules present in the DMC1 and CDMC1 catalysts would undergo a dehydration reaction during the second stage to give 2-methylprop-1-ene. Figure 3 also revealed the formation of peaks at 1149 and 1021 cm^-1, which were assigned to the asymmetrical and symmetrical stretching vibration of the ether bond of the crown ether.

The final stage in the thermal decomposition profiles of the DMC1 and CDMC1 catalysts occurred from 360 to 770 °C. This stage was attributed to the decomposition of the cyano groups and the other organic ligands (i.e., t-BuOH and 18-crown-6) attached to the active metallic center. The change in the stretching vibration of the C≡N bond (2278 cm^-1) in Fig. 3(a) (curve 4) implied that the cyano group was being converted to free cyanide during the final stage of the thermal decomposition process. At the same time, the characteristic adsorption peaks (3278 and 2180 cm^-1) of an alkyne could also be seen. The decomposition residues of the DMC1 and CDMC1 catalysts were determined to be metallic oxides with weight ratios of 44.0% and 40.2%, respectively. The compositions of these metal oxides were found to be in agreement with the results of the elemental analysis (CDMC1: 21.6% Zn, 8.1% Co and 1.2% K; DMC1: 25.1% Zn, 8.6% Co and 1.4% K).

As shown in Table 3, the metal oxides composition of CDMC1 (about 40.2%) was less than that of DMC1 (about 44%) because of the extra crown ether co-complexing agents. The decomposition products of the organic ligands (alkene and alkyne) could be observed by FTIR until the temperature reached 750 °C, which indicated that these ligands were coordinated to the metal center. After 290 °C, the weight of the CDMC1 material diminished rapidly, whereas the weight loss from the DMC1 catalyst occurred at a much slower rate (Fig. 3 (c)). This difference in the weight loss profiles of the two catalysts could be attributed to the highly dispersive structure of CDMC1 compared with the more compact DMC1, as shown in Fig. 2. Furthermore, the loss of the cyano ligand from the CDMC1 material at 390 °C occurred at a much higher temperature than that of the DMC1 catalyst (357 °C). This difference was attributed to an enhanced level of coordination from the O atom of the crown ether to the Zn ion in the CDMC1 catalyst, which would have led to the weakening of the bond between the cyano group and Zn ion, whilst strengthening the interaction between the cyano group and the Co metal center. The overall weight loss from the CDMC1 material (59.2%) was higher than that of the DMC1 catalyst (56.0%) between 20 and 770 °C because of the complexation of the crown ether.

The XRD patterns of all of the catalysts are shown in Fig. 4. As expected, DMC0 catalyst (pure Zn₃[Co(CN)₆]), prepared in the absence of the ligand and excess ZnCl₂ contained sharp peaks with d-spacing values of 5.75, 5.07, 3.59, 2.54, and 2.28 Å, and exhibited a typical cubic lattice structure ^[36]. In general, the addition of an organic ligand such as t-BuOH can disrupt the crystal structure and lead to a reduction in the crystallinity ^[32], which would be consistent with our XRD results. Regardless of whether the crown ether was added or not, other CDMC and DMC catalysts contained very broad and highly dispersed signals, which implied rather weak crystalline structures. For example, the XRD pattern among CDMC1 catalysts was the flattest of the all of the catalysts most likely because it contained the largest amount of crown ether. The signals at 12° and 21° in the DMC1 and DMC2 increased gradually, which suggested that the inclusion of multiple washing steps was leading to the generation of a new crystalline phase.

Fig. 4. XRD patterns of the CDMC and DMC catalysts.

CDMC2-1, CDMC2-2 and CDMC2-3 were prepared in the same manner using 18-crown-6 as the CDMC2. As shown in Table 4, the Zn/Co molar ratios of these materials varied slightly from 1.77% to 1.86% for three independent experiments. The C and N contents of these catalysts were found to be almost equal because of the robust and highly reproducible nature of the processes involved in the preparation of these catalysts.

Table 4 Elemental analyses of the CDMC catalysts. Catalyst K (at%) Zn/Co molar ratio C (wt%) H (wt%) N (wt%) CDMC2-1 1.26 1.80 29.84 3.277 13.41 CDMC2-2 1.13 1.86 30.03 3.416 13.19 CDMC2-3 1.14 1.77 30.09 3.359 13.06 K content from the EDS; C, H and N content from elemental analyses.

Table 4 Elemental analyses of the CDMC catalysts.

The coupling reactions of CO₂ and epoxides can occur under mild conditions ^[38]. The coupling and copolymerization of CO₂ and propylene oxide can occur simultaneously according to Scheme 1. The amount of the copolymer and the coupling product can be counted by ¹H-NMR as shown in Fig. 5. The presence of K⁺ in catalysts can have an adverse impact on their selectivity. 18-Crown-6 was used to coordinate to the K⁺ in an attempt to simply the procedure for the synthesis of uniform active heterogeneous Zn-Co DMC catalysts, which appear to provide the most promising combination of metals with high levels of activity and selectivity in copolymerization reactions. The results in Table 5 clearly show that the CDMC1 catalyst showed catalytic activity as high as 21889 g/g Zn, which was much higher than those of the DMC1 and DMC2 catalysts (16300 and 16400 g/g Zn, respectively). The polymer prepared in the presence of CDMC3 had a similar carbonate linkage content (47.8%) to that of DMC2 (48.3%), but much higher catalytic activity (32600 g/g Zn). The DMC1 catalyst prepared without 18-crown-6 and more washing steps, which was shown to contain 1.43% K by ICP analysis, was less active and selective than all of the CDMC catalysts and the DMC2 catalyst with seven washing steps. These results therefore imply that the free K⁺ would be harmful to the copolymerization reactions of CO₂ and epoxides. No K⁺ was detected in the DMC2 catalyst. Both of the CDMC1 and CDMC2 materials were found to be much more active than DMC1. This result indicated that the coordination of the K⁺ to 18-crown-6 was beneficial to the copolymerization reaction. DMC0, which was prepared without any organic ligands, had hardly any activity, which indicated that an active metallic center must be formed between a metal and some organic ligands to allow for the copolymerization of CO₂ with epoxides.

Table 5 Results of the copolymerization catalyzed by various DMC catalysts a. Catalyst fCO2 b (mol%) WPC c (wt%) Efficiency d g/g Zn (g/g cat.) Mn PDI CDMC1 40.6 3.3 21900 (4728) 31700 4.02 CDMC2 42.0 2.6 27500 (4683) 32500 3.56 CDMC3 47.8 1.5 32600 (5122) 38100 3.29 DMC1 29.2 3.3 16300 (4100) 32000 4.10 DMC2 48.3 2.4 16400 (5073) 38600 3.41 DMC0 — — 2.4 — — a Copolymerization conditions: catalyst 7 mg, propylene oxide 30 mL, CO2 3.0 MPa 16 h, 80 °C. b Molar fraction of carbonate calculated based on the 1H-NMR peak area using the equation fCO2= (A5.0 + A4.2 - 2A4.6)/(A5.0 + A4.2 - 2A4.6 + i>A3.5). c Weight fraction of the cyclic carbonate byproduct confirmed by referring to the 1H-NMR peak area using the equation WPC = 102A1.5/ [102(A5.0 + A4.2 - 2A4.6 + A1.5) + 58A3.5]. d Grams of polymer per gram of Zn and grams of polymer per gram of catalyst in parentheses.

Table 5 Results of the copolymerization catalyzed by various DMC catalysts a.

Scheme 1. Synthesis of PPC.

Fig. 5. 1H NMR spectrum of the PPC copolymer.

Several well-known DMC catalysts have been used during the past two decades to catalyze the polymerization reactions of epoxides and the copolymerization reactions of CO₂ with epoxides. Once the long induction time or activation stage is over, the polymerization reaction will proceed rapidly and reach termination in a short time. The final product of these reactions has two hydroxyl end groups, as confirmed by Zhang et al. ^[33]. It was envisaged that the polymerization reaction described in the current study proceeded in two stages, the slow activation stage and the rapid propagation stage, as depicted in Scheme 2. The reaction would begin with the coordination of PO to Zn and the subsequent nucleophilic ring-opening of the PO by a hydroxyl group coordinated to the active Zn center. Given that the amount of catalyst would be small relative to the monomer, the hydroxyl group coordinated to the active Zn center would not be regenerated, and the speed of the activation stage would therefore be restricted. However, during the propagation of the intermediate, the resulting Zn-OR species could trigger a further polymerization reaction and be regenerated through a chain transfer, which would lead to the auto acceleration of the polymerization process. In addition, Lee et al. ^[39] found that the use of small linear aliphatic alcohols was preferred over bulky t-BuOH in terms of shortening the induction period. It is possible that the reduction in steric hindrance resulting from the use of smaller alcohols would lead to an increase in the chain transfer, and therefore accelerate the activation period .

Scheme 2. Proposed mechanism of copolymerization.

18-Crown-6 has been used for the first time to coordinate K⁺ in an attempt to generate uniform and highly active Zn-Co DMC catalysts. Based on the FTIR and SEM results, it was clear that the addition of crown ether had a significant impact on the surface morphologies of the catalysts, which became even and uniformly dispersed compared to the layered and compacted surfaces of the catalysts prepared in the absence of the crown ether. TGA-IR analysis showed that the active metallic center was formed from several metals coordinated to organic ligands (t-BuOH and 18-crown-6). In spite of the large excess of K⁺, the crown ether complexing Zn-Co DMC catalyst prepared without the inclusion of extra washing steps (CDMC1) showed a high catalytic efficiency, which was similar to that the DMC2 catalyst prepared with multiple washings but in the absence of the crown ether. The catalytic efficiency of this material as high as 5122 g/g catalyst (32600 g/g Zn) was obtained in the presence of a Zn-Co DMC with the addition of double the theoretical amount of the crown ether.