Xiao
Chen‡
ab,
Yong
Zou‡
ac,
Mingkai
Zhang
c,
Wangyan
Gou
a,
Sai
Zhang
*a and
Yongquan
Qu
*ac
aKey Laboratory of Special Functional and Smart Polymer Materials of Ministry of Industry and Information Technology, School of Chemistry and Chemical Engineering, Northwestern Polytechnical University, Xi'an 710072, China. E-mail: zhangsai1112@nwpu.edu.cn; yongquan@nwpu.edu.cn
bSchool of Materials Science and Engineering, North China University of Water Resources and Electric Power, Zhengzhou 450045, China
cFrontier Institute of Science and Technology, Xi'an Jiaotong University, Xi'an, 710049, China
First published on 15th December 2021
Epoxides are versatile intermediates in the production of a diverse set of chemical products. The direct epoxidation of alkenes using O2 represents an environmentally friendly and economical process to replace the use of expensive H2O2 or organic peroxides for the synthesis of epoxides. Herein, single-atom Co anchored on N-doped carbon supports with a multiscale porous structure (Co1/NC-h) has been successfully constructed by etching CoZn-ZIF metal–organic framework precursors before a carbonation process. Benefitting from the high intrinsic activity of Co–Nx active sites and the more accessibly exposed reactive sites of multiscale porous structures, the Co1/NC-h catalysts can achieve the epoxidation of cyclooctene with 95% yield of 1,2-epoxycycloheptane at 140 °C and 0.5 MPa O2. This research opens up opportunities for designing high-performance single-atom catalysts towards applications in diverse heterogeneous catalysis.
Owing to the unique atom utilization and satisfactory selectivity, the application of single-atom catalysts has received growing attention.21,22 Lately, carbon-based materials have been proven to be an enormous family of single-atom catalysts because of their specific features including superb chemical and mechanical stability, high specifical surface properties and area, low production cost, easy handling, and so on.23–26 Inspired by their behaviours in catalysing peroxide activation and efficiencies in advanced oxidation processes, carbon-based single-atom catalysts provide a promising opportunity for efficiently activating O2 for alkene epoxidation. Particularly, transition metal (M)–nitrogen carbon materials with atomically dispersed M–Nx active sites are considered as the most encouraging breakthrough case owing to their widely proven capacity to effectively active O2 in various oxygen reduction reactions.27–30 Generally, metal–organic frameworks (MOFs) as precursors with specific morphologies can assist the preparation of carbon-based single-atom catalysts via a carbonization process at high temperature (>700 °C).31–33 However, the internal diffusion of reactants in these catalysts is inevitably weakened by the narrow passage in heterogeneous catalysis. As a result, only <10% of the M–Nx sites in carbon-based single-atom catalysts are involved in catalytic reactions.34,35 Moreover, this phenomenon is further exacerbated in liquid heterogeneous catalysis with relatively large reactant molecules. Therefore, constructing multiscale porous structures is particularly desirable for carbon-based single-atom catalysts to achieve the efficient utilization of the M–Nx moieties, consequently resulting in the improvement of their catalytic ability.
Herein, single-atom Co anchored on nitrogen-doped multiscale porous carbon (Co1/NC-h) with more exposed active sites has been successfully prepared to achieve high intrinsic activity for alkene epoxidation with O2. The multiscale porous structure was constructed by etching CoZn-ZIF MOF precursors before the carbonation process. For cyclooctene epoxidation, Co1/NC-h catalysts with a multiscale porous structure obtained a 95% yield of 1,2-epoxycycloheptane at 140 °C and 0.5 MPa O2. Mechanism studies revealed that the Co–Nx active sites exhibited a high intrinsic activity for epoxidation, while the constructed multiscale porous structure could facilitate the exposure of more Co–Nx sites in the multiscale pores of catalysts. Therefore, the catalytic activity of Co1/NC-h was greatly improved.
Then, the morphology of carbon-based single-atom catalysts was characterized by using a scanning electron microscope (SEM) and transmission electron microscope (TEM). Compared with CoZn-ZIF precursors, the as-prepared Co1/NC catalysts retained the initial octahedral shape (Fig. 1a). Particularly, no Co nanoparticles were observed both on the surface and/or in the bulk of the Co1/NC catalysts from the SEM (Fig. 1a) and dark field TEM images (Fig. 1b). Meanwhile, the X-ray diffraction (XRD) pattern for Co1/NC showed no characteristic peaks of Co or cobalt oxide crystals (Fig. S3†). However, inductively coupled plasma mass spectrometry (ICP-MS) confirmed that the content of Co was 1.2 wt% in the Co1/NC catalysts. The distribution of the Co element was well consistent with the C and N elements from the energy dispersive spectroscopy (EDS) mapping analysis, revealing that the Co atom was highly distributed on the N-doped carbon supports (Fig. 1c). Meanwhile, according to the image from the aberration corrected high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM), the brighter spots were assigned to the Co atoms, further exhibiting the atomic dispersion on the Co1/NC catalysts (Fig. 1d). Therefore, single-atom Co anchored on the N-doped carbon supports could be successfully obtained via the carbonization of CoZn-ZIF.
Meanwhile, the etched CoZn-ZIF precursors with undestroyed phase were also calcined with the same process. TEM images revealed that the as-synthesized Co1/NC-h catalysts also exhibited an approximate octahedral shape but with obvious irregular porous features (Fig. 1e), compared with the surface of Co1/NC catalysts. Also, a relatively rough surface of Co1/NC-h catalysts was observed from the SEM image, compared with the surface features of Co1/NC (Fig. S4†). Thus, some macropores were formed on the Co1/NC-h catalysts. As for Co1/NC catalysts, the Co nanoparticles were also not revealed from the TEM image (Fig. 1e) and XRD pattern (Fig. S3†). Meanwhile, the atomically dispersed Co atom in the Co1/NC-h catalysts could be confirmed by combining the results of HAADF-STEM (Fig. 1f). ICP-MS analysis suggested a Co percentage of 1.1 wt% in Co1/NC-h.
The pore distribution of the Co1/NC and Co1/NC-h catalysts was further analyzed by the Brunauer–Emmett–Teller (BET) method. As shown in Fig. 2a, the N2 adsorption–desorption curves revealed that the adsorption capacity of the Co1/NC catalysts was higher at low pressure and lower at high pressure, compared with that of Co1/NC-h. Thus, the Co1/NC catalyst has a larger number of micropores but a smaller number of macropores than the Co1/NC-h catalysts. Besides, the larger specific surface of Co1/NC (766 m2 g−1) also proved that it has a larger number of micropores compared with Co1/NC-h (613 m2 g−1). As shown in Fig. 2b, the proportion of pores lager than 2 nm was 23.8% on Co1/NC, which was obviously smaller than that of Co1/NC-h (45.1%). These characterization results were in accordance with the previous statement that the Co1/NC-h catalysts exhibited a larger number of relatively large pores. In addition, the Co1/NC-h catalysts also exhibited a reduced pore size in the range of 1–2 nm compared with the Co1/NC catalysts (Fig. 2c). Therefore, the increased number of macropores and the reduced number of micropores could contribute to the formation of a multiscale porous structure in the Co1/NC-h catalysts.
Fig. 2 (a) Adsorption–desorption curves of N2, (b) pore volume and (c) pore size distribution of the Co1/NC and Co1/NC-h catalysts. |
Meanwhile, the superior catalytic activity of single-atom Co catalysts would be further demonstrated when normalizing the reaction rate of cyclooctene epoxidation based on the total content of Co in each catalyst. As shown in Fig. 3b, the reaction ratio of Co1/NC catalysts was 30 mmol mmolCo−1 h−1, which was 63 times higher than that of Co/NC (0.48 mmol mmolCo−1 h−1). Obviously, the single-atom catalysts exhibited nearly two orders of magnitude increase in catalytic activity compared with nanocatalysts. Also, the Co1/NC-h catalysts with a multiscale porous structure exhibited a further increased reaction rate from 30 mmol mmolCo−1 h−1 to 106 mmol mmolCo−1 h−1. Therefore, owing to the single-atom Co and the multiscale porous structure, the Co1/NC-h catalysts yielded an excellent catalytic performance for cyclooctene epoxidation with O2 as the oxidant.
Except the intrinsically active Co atoms, the different environments of Co atoms could also result in totally distinct catalytic activity between Co/NC and Co1/NC-h. As shown in Fig. S6† and 4b, the Co/NC catalysts exhibited a similar multiscale porous structure and defect of N-doped carbon to those of Co1/NC-h. However, the different catalytic performance of two catalysts revealed that other factors such as the differences in the coordination mode of N atoms could play a crucial role. In previous reports, the Co–Nx active sites have been successfully confirmed in the single-atom Co-anchored on the N-doped carbon supports with the same carbonization process.32,40 For the Co1/NC-h catalysts, the Co–Nx active sites were detected from the XPS analysis of N 1s peaks (Fig. 4c). The corresponding valence state of Co atoms was +2 in the Co1/NC-h catalysts (Fig. 4d). Although the Co–Nx active sites were also detected (Fig. 4c), the fraction of Co0 was 27.4% from the XPS analysis of Co 2p in the fresh Co/NC catalysts (Fig. 4d). As shown in Fig. S7,† these metallic Co atoms could be easily oxidized along with the formation of a Co–O bond during the epoxidation of cyclooctene.
Therefore, the different fraction of Co–Nx active sites in Co1/NC-h and Co/NC catalysts led to their different catalytic performance of cyclooctene epoxidation. The O2 activation of the Co–Nx and Co–O active sites directly determined the intrinsic catalytic activity of Co1/NC and Co/NC catalysts. To explore their intrinsic activity, DFT simulation was employed to verify the adsorption behavior of O2 molecules and reactive oxygen atoms. As previously reported, a Co atom coordinated with three N atoms and one C atom (Co–N3) was selected as the model local structure of single-atom Co1/NC-h catalysts, as referenced from previous reports.36,37 Meanwhile, the CoO surface was the model surface owing to the existence of Co2+ in the used Co/NC catalysts from the XPS analysis (Fig. S7†). As shown in Fig. 5a and S8a,† the Co–N3 and CoO surfaces exhibited a similar adsorption strength for O2, resulting in a similar degree of activation.
During the epoxidation process, the Co active sites could be occupied by O2 molecules (Fig. 5a) to give a Co(III)OO˙ superoxo complex (I), as shown in Fig. S9.† The superoxo complex facilitates interaction with the double bond of the alkene and then undergoes a migratory insertion to form a cobalt peroxo complex (II), followed by the formation of a four membered cyclic radical intermediate (III).19,20,41–44 Finally, the four membered cyclic radical intermediate reacts with another alkene molecule to give the epoxide product (Fig. S9†). Thus, the Co–O bond breaking will further affect the desorption of the cobalt peroxo complex to give the four membered cyclic radical intermediate. Due to their different coordination environments, the O atom exhibited greatly stronger adsorption on the CoO surface than on the Co–N3 active sites (Fig. 5a and S8b†). Therefore, the weak interaction between the O atom and single-atom Co further improves the desorption of the cobalt peroxo complex, while the strong adsorption of the reactive O atom on CoO leads to weak activity for epoxidation.
Meanwhile, 3,5-di-tert-butyl-4-hydroxytoluene was selected as the model molecule to investigate the generation of reactive oxygen species. As shown in Fig. 5b, the consumption rate of 3,5-di-tert-butyl-4-hydroxytoluene was 107.7 mmol mmolCo−1 h−1 for the single-atom Co1/NC-h catalysts, while only a 1.4 mmol mmolCo−1 h−1 consumption rate was obtained with the Co/NC catalysts under the same reaction conditions. The huge gap between Co1/NC-h and Co/NC showed that the Co–Nx active sites could generate a larger number of reactive oxygen species compared with Co–O. Thus, the intrinsic catalytic activity of Co–Nx sites was demonstrated by combining DFT simulation with the consumption rate of 3,5-di-tert-butyl-4-hydroxytoluene.
Then, the influence of the multiscale porous structure on the catalytic activity could be obtained by comparing the reaction rate of Co1/NC and Co1/NC-h catalysts. The same structural features of carbon supports and the same coordinated environment as well as the same Co valence revealed that the Co1/NC and Co1/NC-h catalysts exhibited the same Co–Nx active sites for cyclooctene epoxidation (Fig. 4b–d). However, due to the multiscale porous structure, Co1/NC-h can expose more accessible active sites for reactants, resulting in the increase of overall catalytic activity. The exposed number of active sites could be correlated with the electrochemically active surface area, which was investigated by measuring the electrochemical double-layer capacitance (Cdl). The Cdl was obtained through cyclic voltammetric curves recorded at various scan rates from 10 to 60 mV s−1 (Fig. S10†). As shown in Fig. 5c, the Cdl value of Co1/NC-h catalysts was calculated to be 33.8 mF cm−2, which was 1.5 times higher than that of Co1/NC catalysts (22.0 mF cm−2). Therefore, the enhanced Cdl value indicated that more active sites were exposed on Co1/NC-h catalysts.
Meanwhile, the consumption of 3,5-di-tert-butyl-4-hydroxytoluene was also determined to further verify the difference in the number of exposed Co–Nx active sites in the Co1/NC-h and Co1/NC catalysts. With the same total number of Co–Nx active sites, the consumption rate of 3,5-di-tert-butyl-4-hydroxytoluene on Co1/NC-h was 0.063 mmol h−1, which was 1.7 times higher than 0.037 mmol h−1 obtained when catalyzed by Co1/NC. The improved consumption rate of Co1/NC-h also proved that more Co–Nx active sites were involved in the catalytic reaction, consistent with the electrochemically active surface areas of the two catalysts as shown in Fig. 5c.
Based on the above analysis, the single atom Co active sites in N-doped carbon supports exhibited a high intrinsic activity for O2 activation. Meanwhile, the multiscale porous structure in the catalysts resulted in the increase of the available number of exposed active sites for catalysis. Therefore, it is not surprising that Co1/NC-h with single-atom Co and a multiscale porous structure exhibited excellent catalytic performance for cyclooctene epoxidation.
In addition, catalytic stability is also a critical factor to evaluate the catalytic performance of the present catalysts. After the reaction, the Co1/NC-h catalysts could be easily recycled by centrifugation and reused for the next cycles without any treatment. As shown in Fig. 6, the final yield of 1,2-epoxycycloheptane was maintained in the range of 89%∼94% without an obvious decrease for 6 cycles. The morphology of the used Co1/NC-h catalysts was well preserved during the recycling process (Fig. S11a†). Meanwhile, there was no metallic Co and/or cobalt oxide phases in the spent Co1/NC-h catalysts, as revealed by the XRD pattern (Fig. S11b†). The XPS analysis of the Co 2p peak on the fresh and used Co1/NC-h catalysts also suggested the preserved electronic structure of Co in the catalysts after catalytic reactions (Fig. S11c†). Determined by ICP-MS, the Co ions were not detected in the reaction solution. Taking all together, the un-degraded catalytic activity as well as the unchanged morphological features of the Co1/NC-h catalysts demonstrated the excellent catalytic stability of the catalysts.
Fig. 6 Catalytic stability of the Co1/NC-h catalysts for cyclooctene epoxidation. Reaction conditions: cyclooctene (0.5 mmol), catalysts (5 mg), DMF (2 mL), 140 °C, 0.5 MPa O2 and 9 h reaction. |
Co-ZIF and Zn-ZIF were synthesized following the same preparation procedures by using a single metal salt of Co(NO3)2·6H2O (3.75 mmol) and Zn(NO3)·6H2O (3.75 mmol), respectively.
The Co1/NC, Co/NC and NC catalysts were obtained by the pyrolysis of CoZn-ZIF, Co-ZIF and Zn-ZIF at 900 °C for 2 h under N2 with the same process, respectively.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ta09227h |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2022 |