Facet Engineering in Constructing Lewis Acid-Base Pairs for CO2 Cycloaddition to High Value-Added Carbonates

Cycloaddition of epoxides with CO2 to synthesis cyclic carbonates is an atom-economic pathway for CO2 utilization with promising industry application value, while its efficiency was greatly inhibited for the lack of highly active catalytic sites. Herein, by taking BiOX (X = Cl, Br) with layered structure for example, we proposed a facet engineering strategy to construct Lewis acid-base pairs for CO2 cycloaddition, where the typical BiOBr with (010) facets expose surface Lewis acid Bi sites and Lewis base Br sites simultaneously. By the combination of in-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) and theoretical calculations, the oxygen atom of the epoxide is interacted with the Lewis acid Bi site to activate the ternary ring, then facilitates the attack of the carbon atom by the Lewis base Br site for the ring-opening of the epoxide, which is the rate-determining step in the cycloaddition reaction. As a result, the BiOBr-(010) with rich surface Lewis acid-base pairs showed a high conversion of 85% with 100% atomic economy in the synthesis of cyclic-carbonates without any cocatalyst. This study provides a model structure for CO2 cycloaddition to high value-added long chain chemicals.


Introduction
The over-use of fossil fuels results in massive emissions of the greenhouse gas (CO 2 ), which has a profound impact on the global abnormal climate change and energy crisis [1,2]. However, from a chemical point of view, carbon dioxide is an important C1 resource due to its inexpensive, abundant, nontoxic, and recyclability [3][4][5]. The development of approaches to activate and convert CO 2 into high valueadded chemicals, especially in the carbon chain growth reactions of organic synthesis has become a research hotspot nowadays [6][7][8]. One of the representative applications of carbon dioxide in chemical synthesis is coupling with epoxides to produce cyclic carbonates or polycarbonates, which shows a 100% atomic economy and promising industry application value [2,[9][10][11]. Cyclic carbonates are notable compounds in industrial production, which are widely used as fuel additives, electrolytes in lithium batteries, aprotic high-boiling polar solvents, degreasers, monomers in the synthesis of polycarbonates, organic synthetic intermediates, and biosynthetic precursors [12]. Of note, conventional industrial production of cyclic carbonates is based on diols with toxic phosgene [13], or using ester exchange, which increases the risk of accidents and unnecessary energy consumption [14]. In contrast, using CO 2 as C1 feedstock, the direct synthesis of cyclic carbonate from epoxide and carbon dioxide is atomic economy and environmentally-friendly.
As is generally accepted that the ring-opening of the epoxide is the rate-determining step in cycloaddition reaction, it is of great importance to rationally design catalysts that can effectively activate and accelerate the ring-opening of epoxides [15,16]. In recent years, a large number of homogeneous catalysts have been reported for CO 2 cycloaddition into cyclic carbonate, including ionic liquids [17], Schiff bases [18], alkali metal halides [19], and quaternary ammonium or phosphonium salts [20], etc. Nevertheless, the development of homogeneous catalysts is greatly limited by the inherent limitations of product separation and catalyst recovery, despite their relatively high catalytic efficiency [21]. Recently, to solving the above shortcomings, heterogeneous catalysts including metal oxides, MOFs were developed for the CO 2 cycloaddition to various epoxides, while complicated synthetic processes, harsh reaction conditions and requiring the assistance of homogeneous cocatalysts (Tetrabutylazanium bromide (TBAB), etc.) restrict their industrial application [22,23]. Although the cocatalysts with free halide ions can act as nucleophilic reagents (Lewis base) to attack carbon atoms and accomplish ring opening of epoxide, they would corrode stainless steel at the same time, increasing the risk of chemical leakage [24]. Thus, it is imperative to design efficient catalysts for the CO 2 cycloaddition with epoxides under relatively mild conditions and non-use of cocatalysts, in particular containing free halide ions.
Given that the epoxide ternary ring contains oxygen atom with lone pair of electrons, meanwhile carbon atoms are attackable by nucleophilic reagents, it is reasonable to construct Lewis acid-base pairs that can simultaneously act as electron acceptors and donors to activate epoxide. Further, Lewis acid-base pairs could also be beneficial to the adsorption and activation of CO 2 . Based on this, we proposed to construct Lewis acid-base pairs for effective ringopening of epoxides, particularly with nonfree halide ions as Lewis base sites. Considering the unique confined layered structure of bismuth oxyhalides (BiOX, X = Cl, Br, I) with alternate positive [Bi 2 O 2 ] layer and double negative halogen layers [25], the active surface can be obtained by appropriate design containing nonfree nucleophilic halide ions, which can cooperate with the Bi sites to form Lewis acid-base pairs to catalyze the CO 2 cycloaddition of epoxides reaction. For example, bismuth oxybromide (BiOBr) holding the structure of [Bi 2 O 2 ] 2+ layers interleaving with double layers of bromine atoms becomes an ideal model for our study. As shown in Scheme 1, only Lewis base Br sites are exposed in BiOBr with (001) crystal facets. In comparison, Lewis acid Bi sites and Lewis base Br sites are exposed simultaneously for (010) crystal facets, where Bi atom can accept electrons from oxygen atoms in epoxides, and Br atom can provide electrons to carbon atoms in epoxide ternary rings. Thus, it is anticipated that effectively CO 2 cycloaddition reaction could be accomplished by facet engineering on BiOBr nanocrystal to construct Lewis acid-base pairs.

Results and Discussion
To get deeper insights into the role of Lewis acid-base pairs in epoxide and CO 2 activation, density functional theory (DFT) was employed to investigate the adsorption behavior of propylene oxide (PO) and CO 2 on different facets of BiOBr models, as shown in Figure 1 and S1. Compared to the slab of BiOBr-(001), the slab of BiOBr-(010) with the exposed Lewis acid-base sites exhibited more negative adsorption energy for both PO and CO 2 (Table S2), indicating its good adsorption ability for reactants. Meanwhile, the Bader charge analysis revealed that almost no electron transfer occurred in the activation of PO and CO 2 on the slab of BiOBr-(001) (Table S3), while the product propylene carbonate (PC) had barely any electron transfer on the slab of BiOBr-(010), indicating that the reaction was selective up to the PC, as well as facilitating the detachment of the product.
In order to study this issue, herein the BiOBr nanoplates with exposed (001) and (010) crystal facets were obtained by a modified hydrothermal method, which donated as BiOBr-(001) and BiOBr-(010), respectively (See detail in Supporting Information). As a well characteristic method for determining the orientation of crystal growth, X-ray diffraction (XRD) patterns display that both BiOBr samples exhibiting peak positions could be indexed to the tetragonal phase BiOBr with high purity and no other impurity peaks (Figure 2(b)). To better compare the differences in XRD peak intensities of samples with different crystal facets, the intensity ratios of several key crystal facets were calculated (Figure 2(c)). [26] It is remarkable that the peak intensities of (00 L) facets in BiOBr-(001) sample are prominently higher than that of BiOBr-(010) sample. Furthermore, for BiOBr-(010), the intensity ratios of (102), (110), and (200) to (001) are distinctively higher except (002) to (200) than those of BiOBr-(001) sample, indicating that BiOBr-(001) and BiOBr-(010) have distinctly diverse growth directions. The full spectral sweep of the X-ray photoelectron spectroscopy (XPS) shows that both samples are free of impurity elements, further illustrating the high purity of the phases ( Figure [27,28]. For comparison, HAADF-STEM, HRTEM, and SAED of BiOBr-(001) confirmed the dominant exposed surfaces to be (001) facets ( Figures S5  and S6). In summary, BiOBr nanoplates with different exposed facets were successfully prepared.
Due to the surface atomic structure and coordination environment of the catalyst directly determining its properties, the surface structure of the (001) and (010) facets of BiOBr were then investigated. The (001) facet shows a closed structure consisting of the most outer layer of Br atoms, whereas the (010) facet has alternating Bi, O, and Br atoms at the outermost layer (Figure 2(a) and S7) [26]. Thus, the Lewis acid-base pairs can be constructed simultaneously by controlling the exposure of the (010) facets of the BiOBr nanoplates.
With abundant Lewis acid-base pairs on the surface, BiOBr-(010) would exhibit excellent PO activation ability, which facilitates efficient catalytic CO 2 cycloaddition reaction. To investigate the role of Lewis acid-base pairs in 2 Research CO 2 cycloaddition, BiOBr-(001) and BiOBr-(010) nanoplates were employed for the synthesis of PC from CO 2 and PO. The cycloaddition reaction of carbon dioxide with PO was carried out in a closed pressurized reactor under 8 bar CO 2 at 423 K. As shown in Figure 3(a), it is a surprise that the conversion yield of BiOBr-(010) with rich Lewis acid-base pairs achieved up to 85% under cocatalyst-free condition, which was~1.8-fold higher than that of BiOBr-(001), suggesting that Lewis acid-base pairs play a positive effect in the CO 2 cycloaddition reaction. To further study the differences in the catalytic property of the samples, a series of catalytic reactions were tested at different temperatures while maintaining other factors unchanged. It is clearly seen from Figure 3(b) that BiOBr-(010) nanoplates shows a greater catalytic reactivity with higher conversion at all temperature conditions. The TOF number was calculated by controlling the reaction time to obtain the conversion of the reaction at the initial stage for a more accurate comparison of the catalytic activity. Moreover, the activation energy (E a ) values were calculated based on the Arrhenius plots, which were obtained by linearly fitting the curve of ln(TOF) to 1000/T. As shown in Figure 3(c), BiOBr-(010) exhibited a much lower activation energy (E a = 57:94 ± 0:58 kJ/mol), which was only about half of that for BiOBr-(001) (E a = 105:89 ± 1:17 kJ/mol). Therefore, BiOBr-(010) can lower the activation energy of the reaction more effectively and greatly enhance the catalytic activity without cocatalyst. As mentioned above, BiOBr-(010) possesses a higher conversion rate and a lower activation energy of the reaction under the same reaction conditions. In addition, to verify the robust stability of the sample during the reaction, a catalytic cycle stability test was performed on BiOBr-(010). After five catalytic cycles, BiOBr-(010) still maintained its original catalytic activity with no decrease in conversion and selectivity (Figure 3(d)). In addition, the XRD and TEM after the reaction displayed that the original morphology and structure of BiOBr-(010) was still maintained ( Figure S12). Besides, the scope of BiOBr-(010) [001] BiOX nanocrystal (X = Cl, Br) X atoms exposed only Bi, X atoms exposed simultaneously Lewis acid Bi sites attack oxygen Propylene oxide Lewis base X sites attack carbon Scheme 1: Illustration of CO 2 cycloaddition with epoxides on different facets of BiOX.

Research
catalyst in other epoxy-based substrates was examined to further evaluate its substrate tolerance and versatility. As shown in Figure 3(e) and Table S1, epichlorohydrin, epibromohydrin, 1,2-epoxybutane, and styrene could be effectively transformed into corresponding cyclic organic carbonates under cocatalyst-free condition, excepting for cyclohexene oxide. The relatively low conversion of cyclohexene oxide was attributed to the steric effect, which blocked the interaction of the epoxide with the active site on the catalyst surface. More importantly, the conversion and selectivity for BiOBr-(010) significantly surpassed previously reported catalysts (Figure 3(f) and Table S4). [22,29] Therefore, BiOBr-(010) with abundant Lewis acidbase pairs were efficient catalysts for the CO 2 cycloaddition with epoxides in lack of cocatalysts. The same facet engineering method is also work for BiOCl samples, where the conversion of PO via BiOCl-(010) is up to 86%,~1.9 times greater than that of BiOCl-(001) ( Figure S17).
With the enrichment of Lewis acid-base pairs on the surface, BiOBr-(010) exhibits outstanding catalytic performance in CO 2 cycloaddition reaction, and it is essential to investigate the mechanism of the reaction to gain a clearer and more definite understanding of the overall catalytic reaction. To further study the crucial role of Lewis acidbase pairs in the ring-opening of PO and the CO 2 cycloaddition reaction, in-situ diffuse reflectance infrared Fouriertransform spectroscopy (DRIFT) was performed to study the reaction intermediates by observing bonding alterations in the catalytic reaction. As shown in Figure 4 Figure 4: Reaction mechanism study. (a) In-situ DRIFTS spectra for CO 2 cycloaddition with PO for BiOBr-(010). (b) 2D plot of in-situ DRIFTS spectra for CO 2 cycloaddition with PO for BiOBr-(010). (c) In-situ DRIFTS spectra for the catalysts with CO 2 and PO for 30 min. d) Proposed mechanism for the CO 2 cycloaddition with PO catalyzed by BiOBr-(010). 6 Research vibration of C-Br, C-O, C-O-C, carbonate, and C = O, respectively [30][31][32][33][34]. Of note, the formation of C-Br demonstrated that Br atoms acted as a Lewis base sites to attack the carbon atoms in the epoxide, which played a key role in the ring-opening of the epoxide. The intensity of each peak strengthened as time increased. Meanwhile, comparing the signals collected at 30 min for both samples, the all above peaks of BiOBr-(010) nanoplates were significantly stronger than that of BiOBr-(001), suggesting the high catalytic activity of BiOBr-(010). More importantly, despite BiOBr-(010) had lower density Br sites compared to BiOBr-(001), the C-Br peak of BiOBr-(010) was much higher than that of BiOBr-(001), which forcefully confirmed Lewis acid-base pairs in (010) facet of BiOBr were efficient for the activation of epoxide. Further, DFT calculations were conducted to illustrate the origin of catalytic activity on BiOBr models with different facets. As seen from the calculated density of states (DOS), compared to (001) facet of BiOBr, the disappearance of the O peak near the Fermi energy level in the DOS of (010) facet is accompanied by a decrease in the area of the nearest neighbouring Bi nonbonded peak, which is caused by the electron injection into Bi from the O of PO ( Figure S15). Furthermore, combined with the Bader charge analysis (Table S3), it was found that PO lost 0.04 electrons on (010) facet of BiOBr, while almost no electrons were transferred on (001) facet, suggesting that the Bi sites on (010) facet act as Lewis acid sites. In other words, the simultaneous presence of the Bi and Br sites on (010) facet of BiOBr achieved the construction of Lewis acid-base pairs. Considering the unique ternary ring structure of PO, the first step in the activation of PO is actually the electron transfer from O atom of PO to Bi site, which weakens the epoxide bond, and then Br site attacks C atom as a nucleophilic reagent (Lewis base), so that effective ring-opening can take place under relatively mild conditions. It is worth mentioning that the adsorption of CO 2 on the (010) facet of BiOBr is easier than that of (001) owing to the existence of Lewis acid-base sites ( Figure S9). Subsequently, the oxygen in the activated ringopened epoxide interacts with CO 2 to form a carbonate structural intermediate, which is ultimately converted to the corresponding cyclic carbonate via a ring-closing step (Figure 4(d)). The above analysis further explains the predominance of constructing Lewis acid-base pairs in PO activation and CO 2 cycloaddition reaction. In contrast, for the (001) facet of BiOBr with only the Lewis base Br sites, more harsh reaction conditions would be required. Benefitting from the synergy of the dual active sites, BiOBr-(010) can effectively catalyze the CO 2 cycloaddition with epoxides in a moderate and cocatalyst-free condition.

Conclusion
In summary, a facet engineering strategy was proposed to construct Lewis acid-base pairs for CO 2 cycloaddition, where BiOX (X = Cl, Br) with exposed specific facet could effectively activate CO 2 and PO in producing high valueadded cycle carbonates. By taking BiOBr nanoplates as a model system, both theoretical calculations and experiments demonstrated that BiOBr-(010) with Lewis acid-base pairs has a stronger CO 2 and PO activation capacity in comparison with BiOBr-(001). In detail, the synergistic interaction between the Lewis acid Bi sites and the Lewis base Br sites coexisted in BiOBr-(010) promotes the ring opening of the epoxide as well as the efficient activation of CO 2 without any cocatalyst as evidenced by in-situ DRIFT. To go further, the mechanism investigation based on Bader charge calculations and DOS evidenced that the electrons of O atom in PO was transferred to Lewis acid Bi sites, accompanied by the attack of Lewis base Br sites to C atom of PO, leading to effective ring-opening under relatively mild conditions. Benefiting from the simultaneous exposure of Lewis acid Bi sites and the Lewis base Br sites, BiOBr-(010) showed a high conversion yield of 85% with almost 100% selectivity in the catalytic CO 2 cycloaddition reaction with epoxides without cocatalyst. This work not only provides insight into the design of a cocatalyst-free catalyst in CO 2 cycloaddition, but also gives a deep understanding of the mechanism of CO 2 cycloaddition reactions.

Materials and Methods
Samples synthesis and additional characterization are included in the Supplementary Materials (available here).

Data Availability
All data needed to evaluate the conclusions in the paper are presented in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
Association of CAS (Y2021123), the Fundamental Research Funds for the Central Universities (WK2060000039), and the University Synergy Innovation Program of Anhui Province (GXXT-2020-005). The numerical calculations in this paper have been done on the supercomputing system in the Supercomputing Center of University of Science and Technology of China. We thank the Infrared Spectroscopy and Microspectroscopy Endstation (BL01B) in the National Synchrotron Radiation Laboratory (NSRL) for characterization help. Figure S1. CO 2 absorption on the surface of (a) BiOBr-(001) and (b) BiOBr-(010). Figure S2. The components and chemical state revealed by the XPS survey scan spectrum of BiOBr-(001) and BiOBr-(010) samples, revealing the purity of the samples. Figure S3. SEM image of as-synthesized (a) BiOBr-(001) and (b) BiOBr-(010) samples. Figure S4 Figure S7. Atomic structure of BiOBr-(001) surfaces of the side view. Figure S8. N 2 adsorption-desorption isotherm of (a) BiOBr-(001) and (b) BiOBr-(010). Figure S9. CO 2 adsorption isotherms of BiOBr samples. Figure S10. Representative 1 H-NMR spectra of the products for Cycloaddition of Epoxides with CO 2 in dimethylformamide (DMF) solution after 12 h reaction. Figure S11. In-situ DRIFTS spectra for CO 2 cycloaddition with PO with TBAB for 30 min. Figure S12. Characterization of BiOBr-(010) sample after 5 catalytic cycles. (a) XRD pattern and (b) TEM image after catalytic cycles up to 60 hours. Figure S13. The XPS spectra of Bi 4f for BiOBr-(010) before and after catalysis. Figure S14 Figure S16. XRD patterns of the BiOCl samples. Figure S17. Performances of BiOCl samples for CO 2 cycloaddition with PO. Reaction condition: CO 2 pressure of 8 bar at 423 K. Figure S18. HRTEM, SAED, and corresponding EDS mapping images of BiOCl-(001) (a-c) and BiOCl-(010) (d-f). Table S1. Cycloaddition of Epoxides with CO 2 .