Photocatalytic cyclohexane oxidation and epoxidation using hedgehog particles

Inorganic particles are effective photocatalysts for the liquid-state production of organic precursors and monomers at ambient conditions. However, poor colloidal stability of inorganic micro- and nanoparticles in low-polarity solvents limits their utilization as heterogeneous catalysts and coating them with surfactants drastically reduces their catalytic activity. Here we show that effective photo-oxidation of liquid cyclohexane (CH) is possible using spiky particles from metal oxides with hierarchical structure combining micro- and nanoscale structural features engineered for enhanced dispersibility in CH. Nanoscale ZnO spikes are assembled radially on α-Fe2O3 microcube cores to produce complex ‘hedgehog’ particles (HPs). The ‘halo’ of stiff spikes reduces van der Waals attraction, preventing aggregation of the catalytic particles. Photocatalysis in Pickering emulsions formed by HPs with hydrogen peroxide provides a viable pathway to energy-efficient alkane oxidation in the liquid state. Additionally, HPs enable a direct chemical pathway from alkanes to epoxides at ambient conditions, specifically to cyclohexene oxide, indicating that the structure of HPs has a direct effect on the recombination of ion-radicals during the hydrocarbon oxidation. These findings demonstrate the potential of inorganic photocatalysts with complex architecture for ‘green’ catalysis.


Inventory of Supporting Information
Source Data for Figures 3,4, and 5 is included in a supplementary source data file (.xls) Figure S1: Emulsion polymerization of styrene and Hedgehog Particles (HPs) Table S1: Particle Dimensions of HP Catalysts and their components Table S2: Surface area of catalysts determined using Brunauer-Emmett-Teller (BET) theory Figure S2: Oxidation data from Figure 3 unnormalized or normalized by surface area. Figure S3: Product yield as a function of the amount of water present in the reaction Figure S4: SEM of ZnO/SiO2 HP catalysts and UV-Vis extinction and photoluminescence spectra. Figure S5: Product yield for select catalysts normalized by ZnO mass fraction. Figure S6: Characterization of ZnO nanoparticle decorated Fe2O3 microcube and unnormalized data from Figure 4 Figure S7: X-ray diffraction of HP Catalysts including nanodisc functionalization and components. Figure S8: Epoxide selectivity of catalysts in cyclohexane oxidation Figure S9: Product concentration as a function of HP catalyst concentration Figure S10: Photoluminescence emission spectra of microcube and nanorod mixture and nonnormalized data from Figure 4d. Table S3: Cyclohexene yield for select single experiments at low and high oxidant concentrations. Figure S11: Product yield of HP catalyst with designated concentration of products added to reaction flask. Figure S12: TEM images and UV-Vis extinction spectra of Au-NP/ZnO/Fe2O3 HP Figure S13: Cyclohexane oxidation results with recycled HP Catalyst Figure S14: XPS spectra and SEM of recycled HP catalyst Figure S15: Light intensity for X-Cite Series 120 lamp used in cyclohexane oxidation experiments Figure S16: Cyclohexane oxidation product distribution from various light and heat conditions Figure S17: GC-MS mass spectrum identification of cyclohexene oxide in solution after reaction 2

Fig. S1: Emulsion polymerization of styrene with Hedeghog Particle and Nanorods
To visualize the interaction of particles with the bulk organic phase in the event that the emulsion broke, in one experiment, the emulsion was initiated and then agitated gently to break the organic droplets and polymerize the bulk organic phase. The results are shown in Fig. S1. In the bulk polymer, hedgehog particles (HPs) (Fig. S1a, c) show good dispersion within the organic layer while ZnO nanorods (NRs) self-assemble into aggregated chains (Fig. S1b, d), confirming that even when the emulsion is broken, HPs are engineered to retain dispersion and remain wetted by the organic phase and therefore retain higher available surface area for catalysis. SEM images were taken on a TESCAN RISE under low vacuum conditions without any conductive sputter coating. mass fraction of ZnO per sample, assuming 200 spikes/particle and densities of 5.26 g/cm 3 for hematite, 5.61 g/cm 3 for zinc oxide, and 2.65 g/cm 3 for silica.

Figure S5: Product yield for select catalysts normalized by ZnO mass fraction
We argue that the physical structure of the catalyst is the most important factor for the catalytic enhancements we have seen. However, we cannot completely rule out the possible electronic influence of the heterostructure of the HP catalyst. [1][2][3][4][5][6][7] The core material of a large HP structure likely does not absorb light due to light scattering within the particles, 8 so photo-induced transfers may be limited to those from ZnO to Fe2O3. Because of the influx of holes to the hematite core, electrons may be displaced toward ZnO. This theory is supported by the product yield and selectivity for various catalysts when normalized by the mass fraction of ZnO (Fig.  S5)  ZnO-ND/ZnO/Fe2O3 HP were synthesized using an adapted literature procedure 12 to purposefully encapsulate ZnO spikes with concentric hexagonal nanodiscs. Briefly, 1 g 3.7-ZnO/Fe2O3 HPs were dispersed in 1.6 L DI water containing 25 mM zinc nitrate hexahydrate, 200 mM hexamethylene tetramine, and 0.2 mM sodium citrate tribasic dihydrate. The solution was heated to ca 90°C and sonicated using a ultrasonic pilot-scale Hieslcher UIP 1000HdT reactor for four hours. After sonication, the ND HPs were purified by allowing the ND HPs to settle and removing the excess reaction effluent and loose ZnO particles produced during synthesis. Purification continued until all excess ZnO was removed.    The PL spectroscopy shown in Fig. 4d supports our experimental findings that larger spikes lead to higher product yields consistent with previous work 13 and a large reduction of recombination, except for the 1.9-ZnO/Fe2O3 HPs. These HPs had lower product yield but showed the lowest e -/h + recombination. With these HPs, we suspect that hematite core can absorb light (yet still not contribute to the photocatalytic reaction) unlike HPs with thicker and longer nanorods, evidenced by its deep red color like hematite cubes (Fig. 1). With epoxide formation dependent on ZnO light absorption, we suspect the hematite core light absorption lowers the photon efficiency by hindering epoxide formation in this case. Additional PL spectroscopy is shown below in Fig. S10.

Figure S11: Product yield of HP catalyst with designated concentration of products added to reaction flask
Another potential pathway considered for epoxidation was the dehydration of CHol to CHene, which has been shown previously on Cu-ZnO catalysts and is dependent on the acidic sites on the catalytic surface. 9 CHol dehydrogenation to CHone is also widely reported. 9 Adding CHol to the reaction mixture did not lead to quantifiable CHene production and decreased CHO and CHone production compared to the results of Fig. 3a. At high CHol concentrations (5 mM), we see an increase in the selectivity and overall yield of CHone compared to Fig. 3a, indicating that CHol can be further oxidized to CHone at those conditions. 9 However, at lower cyclohexanol concentrations (3 mM, 1 mM), there is no increase in CHone or CHO production but there is obvious consumption of the cyclohexanol added, indicating that cyclohexanol is likely fully oxidized to CO2 at our reaction conditions rather than dehydrated to CHene (Fig. S11). 10,11 As seen in Fig. S11, CHone and CHO were also both added to the reaction mixture. CHone is also consumed in the reaction and inhibits the production of CHO, while added CHO is not consumed but does inhibit the production of CHone. This further indicates competition of these species, possibly for adsorption sites on the catalyst surface.

Figure S12: TEM images and UV-Vis extinction spectra of Au-NP/ZnO/Fe2O3 HP
Au-NP/ZnO/Fe2O3 HPs were synthesized using deposition-precipitation of gold chloride with urea. 14 500 mg of 3.7-ZnO/Fe2O3 HP were dispersed in 50 mL of ultrapure water containing 4.2 mM gold chloride trihydrate and 0.42 M urea. The mixture was stirred at 80°C under reflux in the dark for 16 hours. The particles were then centrifuged and washed with water 3x at 1000 RPM to remove excess precursor. The particles were then lyophilized and calcined at 400°C for 4 hours in air. Finally, the particles were washed with water 3x and lyophilized again. Gold was confirmed using EDX. Fig. S12 contains TEM images and additional spectroscopic characterization of the Au-NP/ZnO/Fe2O3 HP.      . Cyclohexene oxide was also confirmed and quantified by running a series of calibration standards in GC-FID to confirm identical retention time to a product peak seen after reaction.