**1. Introduction**

In recent decades, anthropogenic activities have dramatically increased the concentration of atmospheric CO2; this concentration was found to be higher than 400 ppm in 2017 [1]. The rapid increase in the CO2 content in the atmosphere makes CO2 a major greenhouse gas. Consequently, the development of efficient and safe methods for capturing and sequestering CO2 is garnering increased attention. Furthermore, the development of methods and processes for converting CO2 into value-added chemicals is of paramount interest because CO2 is a cheap, non-toxic, and abundant carbon feedstock [2,3]. Recently, a grea<sup>t</sup> deal of research has been devoted to capturing and utilizing CO2 to synthesize a variety of value-added products [4–17].

One of the most sustainable strategies for utilizing CO2 is the copolymerization of CO2 with epoxides to produce poly(alkylene carbonates). These materials are commercially viable owing to their vast number of applications, such as in adhesives, packing and coating materials, and ceramic binders [18–22]. Furthermore, these polycarbonates are biodegradable and are useful in biomedical applications [19]. The alternating copolymerization of CO2 with epoxides was first reported by Inoue et al., who used a diethylzinc–water system as a catalyst [23,24]. Subsequently, numerous homogeneous and heterogeneous catalytic systems have been developed. For example, homogeneous catalysts, such as metalloporphyrins, β-diiminate Zn complexes, and metal salen complexes have been found to be highly active for the copolymerization of CO2 and epoxides [25–27]. Nevertheless, the industrial utilization of these homogeneous catalysts is limited because of their complicated syntheses, the use of toxic metals like chromium, and the difficulties in separating the catalyst/product mixtures. However, heterogeneous catalysts are preferred for industrial scale applications owing to their low cost,

easy synthesis, facile separation from products, and reusability [28–31]. Among the heterogeneous catalysts, zinc dicarboxylates, Zn-Co double metal cyanide complexes, and ternary rare-earth complexes are found to be particularly active for the copolymerization of CO2 and epoxides. Among them, the zinc glutarate (ZnGA) is widely applied in industry as a catalyst for the copolymerization of propylene oxide (PO) and CO2 because it is economic, non-toxic, easy to synthesize, and, most beneficially, it yields copolymers with high molecular weights [18,32]. Based on a recent life cycle assessment study, for each one kg of CO2 incorporated into polycarbonate polyols, up to three kg of CO2-equivalent greenhouse gas emissions could be reduced [33]. Therefore, improving the productivity of ZnGA becomes necessary in order to meet the increasing economic and environmental requirements.

Despite a growing need and the continued efforts by researchers, the catalytic activity of ZnGA has improved only marginally over the past few decades. Rieger et al. reported on the considerable improvement of the catalytic activity of ZnGA via the introduction of zinc-ethylsulfinate initiator groups to its surface [34]. However, this post-modification has limited industrial viability because of its procedural complexity and use of expensive precursors. Ree et al., demonstrated the effects of different zinc and glutarate sources in the synthesis of ZnGA and consequently the effects of different morphologies of ZnGA on its catalytic activity for CO2/PO copolymerization [35]. Through continuous research effort, the catalytic activity of ZnGA has been found to be dependent on its surface area and crystallinity [18,36–38]. Recently, the precise structure of ZnGA was obtained by single-crystal X-ray diffraction studies, thus revealing that its catalytic activity mainly originates on the outer surfaces of the Zn-dicarboxylate particles [39]. It should nevertheless be noted that the crystal structure of ZnGA features a 3D-network structure of glutarate ligands and Zn atom of the molecule in which, each Zn atoms tetrahedrally bind to the oxygen atoms of four different glutarate units.

According to the recent definition of metal organic framework (MOF) by Seth and Matzger, ZnGA can be considered a MOF [40]. It is noteworthy that most other Zn-based MOF systems have been reported to yield cyclic carbonate as the predominant product in the reaction of CO2 and an epoxide, whereas the ZnGA system produces the poly(alkylene carbonate) as the major product [41–47]. In recent years, most studies on ZnGA have focused on increasing its surface area and crystallinity using different supporters, amphiphilic templates, and ultrasonic treatment [48–51]. Furthermore, a recent mechanistic study on CO2/PO copolymerization using ZnGA indicated a bimetallic mechanism involving sequential insertions of CO2 and epoxide into Zn-alkoxide and Zn-carboxylate initiator groups on the surface of the catalyst [52]. The optimal separation between two adjacent Zn atoms was suggested to be in the range 4.3–5.0 Å, which results in the optimal activation energy required for copolymerization. Recently, we explored the catalytic ability of ZnGA for the copolymerization of CO2 with a relatively reluctant epoxide, epichlorohydrin, and the application of ZnGA as a catalyst for the terpolymerization of CO2, PO, and β-butyrolactone [53,54].

As a part of our continued research effort to develop heterogeneous catalysts for CO2 conversions, this study reports a facile method for preparing a surface modified ZnGA and its enhanced catalytic activity in the CO2/PO polymerization. In order to improve the catalytic activity of ZnGA, the number of active sites on the surfaces of the catalyst particles must be increased. It has been reported that ZnGA has protruding glutarate and hydroxyl groups on its surface that act as initiators for CO2/PO copolymerization [55,56]. These glutarate and hydroxyl groups can ligate additional incoming metal ions and create oxygen-bound metal centers on the surface of the ZnGA. Therefore, we hypothesized that surface modification of ZnGA by treatment with Lewis metal ions to form metal-treated ZnGA catalysts (ZnGA-M) would increase the number of active metal centers on the surface of the ZnGA, as depicted in Figure 1. This would provide catalysts with improved cooperative bimetallic properties for CO2/PO copolymerization (Scheme 1). Indeed, the pore structure of some MOFs were modified by the installation of additional ligand motif to coordinate additional metal centers in a step called "post synthetic modification" (PSM) [57–61]. However, in the case of ZnGA frameworks, the protruding glutarate groups may be considered for this purpose. To test this hypothesis, different metal chlorides, i.e., FeCl3, AlCl3, ZnCl2, and CoCl2, were selected for the preparation of ZnGA-M catalysts based on their activities in homogeneous complexes species. The resulting catalysts were subsequently assessed for their activities in CO2/PO copolymerization.

**Figure 1.** Schematic representation of the formation of ZnGA-M.

**Scheme 1.** Copolymerization of CO2 and PO using ZnGA-M.

#### **2. Results and Discussion**

#### *2.1. Synthesis and Characterization of Catalysts*

Initially, standard ZnGA (std-ZnGA) was prepared according to a published procedure with slight modifications [62]. Figure S1 compares the powder X-ray diffraction (PXRD) pattern of the resultant white precipitate with the pattern calculated from the crystal structure and confirms the formation of std-ZnGA in its pure form with relatively high crystallinity (Figure S1). Fourier-transform infrared (FT-IR) spectroscopic analysis shows typical peaks for ZnGA. The CH2 scissoring and CH stretching bands were observed at 1445 cm<sup>−</sup><sup>1</sup> and 2955 cm<sup>−</sup>1. The bands at ~1585 cm<sup>−</sup><sup>1</sup> and ~1405 cm<sup>−</sup><sup>1</sup> correspond to COO− antisymmetric stretching frequencies. The COO− symmetric stretching band was observed at 1538 cm<sup>−</sup><sup>1</sup> (Figure S2a). Figure S3a shows a scanning electron microscopy (SEM) image in which the std-ZnGA has taken the form of typical platelet-shaped particles. These analyses collectively confirm the formation of std-ZnGA according to the previous reports in the literature.

The ZnGA-M catalysts were then prepared in anhydrous THF through treatment with different metal chloride solutions at different ratios, as shown in Table 1. The copolymerization of CO2 and PO requires sequential insertion of CO2 and PO into the Zn-alkoxide and Zn-carboxylate initiator groups on the surface of the catalyst. Therefore, the metal treatment should be kept as mild as possible to effect synchronized cooperative catalysis because a thick coating of metal chloride would block the monomers from approaching the catalytic sites. Thus, the ratio of metal to Zn was maintained in the range of 10−3–10−<sup>4</sup> equivalents in the metal treatment step.

The metal-treated ZnGAs were analyzed via PXRD, FT-IR spectroscopy, and SEM and TEM analyses. The FT-IR spectra of the ZnGA-M catalysts are shown in Figure S2a. The FT-IR analysis shows that the metal treatment does not affect the original glutarate binding, since the amount of each metal salt used is too small. As shown in Figure S2b, the amounts of metal ions used are too low to cause any phase changes in the crystal lattice or structural deformations. Thus, the PXRD

patterns of all the catalysts are similar. Figure S3 shows the SEM images of the ZnGA-M samples and reveals that the metal treatments do not significantly change the morphology. The metal ions cannot be detected via SEM energy-dispersive X-ray spectroscopy (SEM-EDS) analysis when fewer than 10−<sup>2</sup> equivalents of the metal ions are used. However, the scanning transmission electron microscopy (STEM) and energy-dispersive X-ray spectroscopy (EDS)-assisted elemental mapping of Al and Zn in ZnGA-Al-10−<sup>3</sup> reveal the homogeneous distribution of Zn and Al metals in the ZnGA-Al-10−<sup>3</sup> (Figure 2). In addition, the actual coated amount of Al in ZnGA-Al-10−<sup>3</sup> was obtained via inductively coupled plasma optical emission spectroscopy (ICPOES), in which the ratio of Zn to Al was found to be 1:1.1 × 10−3. Similarly, the ratio between Zn and Fe in ZnGA-Fe-10−<sup>3</sup> was found to be 1:1.0 × 10−3. These analyses confirmed the attachment of the different metal ions after metal treatment.



a Conditions: 0.20 g catalyst; 20.0 mL PO; 2.0 MPa CO2; 60 ◦C and 40 h. b Treated mol ratio of ZnGA and metal salt. c TON = weight of polymer formed per gram of catalyst. d Values represent the increment in productivity of the metal-treated catalysts in comparison to std-ZnGA. e Fco2 of the polymers was determined from 1H NMR spectra of the polymers. f Obtained from 1H NMR spectra of the polymers; PC = propylene carbonate. g Mn, Mw, and PDI values of the polymers were determined using GPC with polystyrene standards in THF. h Determined from DSC.

**Figure 2.** STEM image and EDS mapping. (**a**) STEM image of ZnGA-Al-10−3; STEM-EDS mapping of (**b**) C (purple), (**c**) O (blue), (**d**) Zn (green), (**e**) Al (red), and (**f**) Cl (cyan) elements in ZnGA-Al-10−3; (Scale bar: = 300 nm).

#### *2.2. Catalytic Activity Studies*

To see the e ffect of metal treatment in the copolymerization of CO2 and PO, the catalytic activities of the metal-treated ZnGAs were assessed and compared with those of std-ZnGA. All the copolymerization reactions were performed using 0.20 g of catalyst and 20.0 mL PO under 2.0 MPa CO2 at 60 ◦C for 40 h. The results are summarized in Table 1. The productivity or the turnover number (TON) is given as grams of PPC formed per gram of catalyst (g PPC/g catalyst). The carbonate content (Fco2) values of the poly(propylene carbonate)-co-polyethers produced were determined using 1H NMR spectroscopy according to the following equation:

$$\text{Fco}\_2 = \left[ (\text{A}\_{5.0} + \text{A}\_{4.2}) / (\text{A}\_{5.0} + \text{A}\_{4.2} + \text{A}\_{3.8-3.5}) \right] \times 100\tag{1}$$

where A represents the integral area of the corresponding protons in the 1H NMR spectrum.

In an initial trial, the std-ZnGA produced poly(propylene carbonate) (PPC) with 93.9% Fco2 and a turnover number (TON) of 72.4 g of PPC/g of catalyst. The resultant polymer has a molecular weight of 156.4 kg/mol and a polydispersity index (PDI) of 3.5 (entry 1, Table 1). Then the ZnGA-M catalysts were used as catalysts for the copolymerization of CO2 and PPO. The results are summarized in Table 1. From Table 1, it is evident that the metal-treated catalysts show a significant increment in TON for CO2/PO copolymerization, and the activity increments are in the range of 1.6–38.3%. Among the metal-treated ZnGA catalysts, ZnGA-Al, ZnGA-Fe, and ZnGA-Zn show dramatic improvements in productivity compared to that of std-ZnGA, with TONs ranging from 80.7 to 100.1 (entry 2–5, 7, and 8, Table 1). It is noteworthy that treatment with 10−<sup>3</sup> equivalents of ZnCl2 results in a TON of 100.1, which is ~38% higher than that of std-ZnGA (entry 7, Table 1).

For AlCl3 and ZnCl2, the catalytic activities increase following a rise in the amount of metal from 10−<sup>4</sup> equivalents to 10−<sup>3</sup> equivalents (entry 4, 5, 7, and 8, Table 1). Conversely, FeCl3 shows a slight decline in the TON after an increase in the metal treatment amount from 10−<sup>4</sup> equivalents to 10−<sup>3</sup> equivalents (entry 2 and 3, Table 1). In addition, when increasing the ZnCl2 amount to 10−<sup>2</sup> equivalents, the observed productivity is lower than that of std-ZnGA (entry 6, Table 1). These results sugges<sup>t</sup> that there is a threshold limit for the amount of metal ions employed, beyond which the metal ions cover the whole surface of the ZnGA and block the monomers from approaching the active Zn-metal bimetallic sites. This masking e ffect is repeated when using CoCl2 as the metal salt, because Co2+ ions are inactive when used in homogeneous salen complexes and they need to be oxidized to Co3+ for higher catalytic performance [63]. Importantly, increasing the amount of CoCl2 has a negative e ffect on copolymerization, and the productivity of ZnGA-Co-10−<sup>3</sup> is decreased by ~0.5% (TON = 72.0).

As an extreme example, std-ZnGA was treated with 1.0 equivalent of AlCl3 for 24 h and the resultant white particles were probed using SEM analysis. The SEM image shows that the surface of the ZnGA is completely covered with a thick layer of AlCl3, with the AlCl3 also having round-shaped edges (Figure 3). TEM-EDS mapping had shown the homogeneous distribution of Zn and Al atoms throughout the sample (Figure S4). The ratio of Zn to Al as shown by TEM-EDS is 1 to 7.4, suggesting a thick layer of Al-motif covered the surface. Additionally, the FT-IR spectrum of ZnGA-Al-1 shows a broad hydroxyl stretching in the range of 3000–3500 cm<sup>−</sup>1. This is di fferent from other ZnGA-M samples (Figure S5). Therefore, as shown in Figure S6, when increasing the amount of metal salt beyond a specific limit, the excess metal salt form a thick layer resembling the bulk metal chloride, thus rendering the catalyst less active (entry 11, Table 1). This fact accounts for the reduced activity of ZnGA-Zn-10−<sup>2</sup> (entry 6, Table 1). These results clearly show that the presence of other metal ions influences the reactivity of the Zn atoms on the surface of the ZnGA catalyst. Additionally, the added metal on the surface may create a Zn-O-M bimetallic site with optimal distance between the surface metal centers. As reported for a number of homogeneous di-nuclear or bimetallic catalysts, the reaction pathway may follow a bimetallic-cooperative mechanism, wherein one metal may selectively bind with epoxide and ease the ring opening, while the other metal activates the CO2 and attacks the activated

epoxide to form the metal carbonate bond [52]. Alternate additions of epoxide and CO2 results in a polymer chain growth.

**Figure 3.** SEM image of ZnGA-Al-1 showing thick layers of AlCl3 on the ZnGA.

#### *2.3. Properties of Polymers*

The polymers formed using the different ZnGA-M samples were then characterized with 1H NMR spectroscopy, differential scanning calorimetry (DSC), thermogravimetric analysis (TGA), and Gel Permeation Chromatography (GPC). The 1H NMR spectrum of the polymer exhibits the methyl, methylene, and methine peaks of the polycarbonate at 1.3 ppm, 4.2 ppm, and 5.1–4.9 ppm, respectively (Figure S7).

It is worth mentioning that all these catalysts produce copolymers with significantly high TONs at a relatively low CO2 pressure of 2.0 MPa, while the Fco2 values are in the range 90–95% with a very small amount of polyether linkages. The microstructure of the polymers was analyzed by 13C NMR analysis. This revealed that the polymers are formed with a predominantly head-to-tail (HT) connectivity (Figure S8). The molecular weight distributions of the prepared polycarbonates were then analyzed using GPC with polystyrene standards in THF and the GPC elugrams. Some of the selected polymers are shown in Figure S9. The results demonstrate that the cooperative bimetallic catalysts actually help to produce polymers with very high molecular weights. As seen in the Table 1, all the catalysts except ZnGA-Co-10−<sup>4</sup> and ZnGA-Zn-10−<sup>2</sup> produce polymers with high Mn values and PDI values close to 2.0. Interestingly, ZnGA-Fe-10−<sup>4</sup> affords a polymer with the very high Mn of 262.4 kg/mol. The PDIs of the polymers varied from 1.8 to 3.5.

The TGA results of the polycarbonates show that the 5% weight loss temperatures (T5%) of the polymers are around 230 ◦C and complete decomposition occurs in the range 330–350 ◦C. However, the glass transition temperatures (*Tg*) of the resulting polymers are slightly lower than that of the polycarbonate produced using std-ZnGA and vary from 33–38 ◦C (Figure S10). These values are in the accepted range typically reported for PPC prepared from ZnGA under different conditions.
