Multiple Promotional Effects of Vanadium Oxide on Boron Nitride for Oxidative Dehydrogenation of Propane.

Featuring high olefin selectivity, hexagonal boron nitride (h-BN) has emerged recently as an attractive catalyst for oxidative dehydrogenation of propane (ODHP). Herein, we report that dispersion of vanadium oxide onto BN facilitates the oxyfunctionalization of BN to generate more BO x active sites to catalyze ODHP via the Eley-Rideal mechanism and concurrently produce nitric oxide to initiate additional gas-phase radical chemistry and to introduce redox VO x sites to catalyze ODHP via the Mars-van Krevelen mechanism, all of which promote the catalytic performance of BN for ODHP. As a result, loading 0.5 wt % V onto BN has doubled the yield of light alkene (C2-C3) at 540-580 °C, and adding an appropriate concentration of NO in the reactants further enhances the catalytic performance. These results provide a potential strategy for developing efficient h-BN-based catalysts through coupling gas-phase and surface reactions for the ODHP process.

Introduction

Propylene (C3H6) is one of the important building blocks for a large number of chemicals in the petrochemical industry.[1] Technologies that have been widely implemented to produce propylene include petroleum-derived steam cracking and fluid catalytic cracking.[1] Direct dehydrogenation of propane (DHP) has emerged as an attractive approach, as evidenced in the Oleflex process (Honeywell, UOP), which features a higher propylene yield than other technologies.[1] However, the inherent endothermicity results in concomitant thermodynamic limitations and high energy input, as well as surface coking.[2] The oxygen-assisted oxidative dehydrogenation of propane (O2-ODHP) provides alternatives to overcome these intrinsic issues, yet the overoxidation signifies the major research challenge for this reaction route.[3]

Redox-active metal oxides such as VO have been extensively studied for ODHP reaction, and the surface reaction typically proceeds via the Mars–van Krevelen (MvK) mechanism.[4−7] Hexagonal boron nitride (h-BN) has emerged as a breakthrough for the O2-ODHP reaction because it features a high selectivity of light alkenes (i.e., ethylene and propylene) with negligible CO formation.[2,3,8−17] The reaction on boron-based catalysts is proposed to proceed through Eley–Rideal (ER) mechanism, in which the radical chemistry in the gas phase is proposed mostly through computational efforts in conjunction with the progress in oxidative coupling of methane reaction.[3,8,18−21] By examining the C2/C1 ratios in products, Wang and Lin have proposed the important role of methyl radicals in forming ethylene in the gas phase.[22] Our recent work directly observed the presence of methyl radicals in the gas phase during BN-catalyzed ODHP using the synchrotron photoionization mass spectroscopy technique.[19]

The proposed active sites include mixed amorphous mixed boron oxide/hydroxide B(OH)O3- (x = 0–3)[13] and dihydroxyl boron-oxide species.[20] Due to the highly dynamic restructuring of boron species under reaction conditions, metastable boron species are also studied.[23,24] Thus, it is generally accepted that the oxidized B sites formed through the concurrent oxyfunctionalization of BN with gas-phase radical chemistry are the active sites for the ODHP reaction.[3,8] Very recently, plasma treatment was reported to regulate the local environment of h-BN by creating nitrogen defects, favoring the generation of BO active sites.[11] Efforts are also devoted to developing BO-containing advanced materials for ODHP reaction, including boron phosphate[25,26] and metal–organic framework-derived catalysts.[27]

Although the combination between VO and BN is known in the literature for propane oxidation to acrolein,[28] the role of VO in modifying the BN surface and its correlation with ODHP catalytic performance are elusive. In the present work, we hypothesize that the introduction of redox-active species such as VO could in situ tune the local chemical environment of BN by introducing oxyfunctional groups with its redox performance as in a typical supported VO system. Indeed, our results show that the redox-active VO additive can significantly enhance the catalytic performance of BN in ODHP. Characterized by Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS), scanning- ransmission electron microscopy (STEM), and synchrotron vacuum ultraviolet photoionization mass spectroscopy (SVUV-PIMS), the VO additive, on the one hand, contributes to catalyzing ODHP via the MvK mechanism; on the other hand, it is found to facilitate the oxyfunctionalization of BN to not only generate more BO active sites to catalyze ODHP via the ER mechanism but also produce NO to further induce gas-phase chemistry for enhanced ODHP.

Experimental Section

Preparation of Catalysts

h-BN with High Surface Area

h-BN was synthesized according to the procedure reported elsewhere[29] and is described briefly herein. The exfoliation was conducted in a muffle oven heated to 700 °C and maintained at this temperature, unless otherwise noted. 1 g of commercial h-BN (Sigma-Aldrich, 99%) was treated at 700 °C for 5 min. Then, the boron nitride was transferred to a 4 L Teflon beaker which contained 100 mL of liquid nitrogen, followed by covering with a lid. The treated boron nitride was carefully recollected after the liquid nitrogen was completely volatilized. The exfoliation procedure was repeated 10 times. Afterward, the obtained boron nitride was dispersed in H2O and ultrasonicated for 10 min, followed by centrifugation at a speed of 1500 rpm to remove unexfoliated materials. The upper liquid was collected and dried at 70 °C overnight to obtain h-BN with a high surface area of 72.7 m2 g–1. The as-synthesized h-BN was pretreated under the reaction atmosphere (C3H8/O2/He = 1/1/38) at 500 °C for 4 h prior to further impregnation. The pretreated materials were denoted as BN-T throughout the paper, wherein the T represents “treated” samples.

VO/BN-T Catalyst

V-loaded BN-T catalysts were prepared by impregnation. A desired amount of V precursor, namely, NH4VO3, was dissolved in 2–3 mL of ultrapure H2O at 80 °C, followed by adding BN-T (0.5 g) into the solution and stirring vigorously until water was volatilized. The mixture was dried at 100 °C overnight and calcined at 450 °C in static air for 4 h with a ramp rate of 2 °C min–1. The catalysts were denoted as xV/BN-T, wherein x represents the mass-based V loading and ranges from 0 to 1.0 wt %.

BO/SiO2, VO/SiO2, and VO/BO/SiO2 Catalysts

BO/SiO2 catalysts were prepared by impregnation. A certain amount of H3BO3 was dissolved in 10 mL of ultrapure H2O at 70 °C, followed by addition of 0.5 g of SiO2. The mixture was vigorously stirred at 80 °C for 1 h and then dried in an oven at 100 °C overnight. Calcination was conducted at 450 °C in static air for 4 h with a ramp rate of 2 °C min–1. The calcined catalysts were denoted as yBO/SiO2, wherein y represents mass-based B loading and is fixed at 5 wt %. VO/SiO2 catalysts were prepared following the same procedure as BO/SiO2, and the V loading was fixed at 0.5 and 1.0 wt % for comparison. The VO-loaded BO/SiO2 catalyst was prepared by a method similar to that of VO/BN-T, and the V loading was fixed at 1.0 wt %.

Catalytic Performance Evaluation

The activity test of the ODHP reaction was performed using an Altamira Instruments system (AMI-200). In a typical test, 100 mg of catalyst (60–80 mesh) was diluted with 300 mg of quartz sand (60–80 mesh). The mixture was loaded in a quartz U-tube (i.d. = 10 mm) with quartz wool at both ends of the mixture. Prior to the activity test, the catalyst bed was treated in the mixture of C3H8/O2/He (1/1/38) at 600 °C for 4 h with gas hourly space velocity (GHSV) = 18,000 mL g–1 h–1, followed by oxidation in 5% O2/He at 500 °C for 1 h under 30 mL min–1. Then, the activity test was initiated by switching to the mixture reaction gas of C3H8/O2/He (1/1/38) with GHSV = 18,000 mL g–1 h–1. The reaction was conducted in the temperature range of 480–600 °C with 20 °C as an interval. Each temperature was held for 2 h for data collection. For the NO-assisted O2-ODHP reaction, 0.1% NO/He or 2% NO/Ar was introduced to alter the NO concentrations at 180, 250, 350, and 3500 ppm. GHSV was fixed at 24,000 mL g–1 h–1, while the C3H8/O2 ratio remained unchanged.

Kinetic studies were conducted after the initial treatment, as described above. Then, the temperature was decreased to the desired temperature (i.e., 540 °C), followed by switching to the reaction gas mixtures using the premixed cylinders 5% C3H8/He, 5% O2/He, and Ar. The partial pressures of both C3H8 and O2 were varied from ca. 0.4 to 2.5%. While exploring the effect of the C3H8 partial pressure, the partial pressure of O2 was fixed at 2.5% and vice versa. Ar was introduced to keep the GHSV fixed at 24,000 mL g–1 h–1. The reaction order was calculated by fitting the gas composition-dependent activity data to the power law in the logarithmic relationship,[3,30,31] while the apparent activation energy was calculated by using the Arrhenius equation.[3,19]

Compositions of products were analyzed periodically using an online SRI 8610C gas chromatograph equipped with both thermal conductivity detector (TCD) and flame ionization detector (FID). MTX-WAX and Molecular Sieve 5A columns were attached to the TCD, analyzing O2, NO, CO, and CO2, while HayeSep-D and alumina were attached to the FID, analyzing hydrocarbons including C3H8, C3H6, C2H6, C2H4, and CH4. A mass spectrometer (Pfeiffer Vacuum) was also coupled with the gas chromatograph to monitor the real-time evolutions of reactants and products. The carbon balance was generally ca. 95%. The absence of mass and heat-transfer limitations were confirmed by the Weisz–Prater criterion and the Mears criterion. Detailed calculations are shown in the Supporting Information.

Characterization of Catalysts

Nitrogen (N2) physisorption was performed using a Micromeritics Gemini 2375 surface area and pore size analyzer at −196 °C. The samples were degassed for 1 h prior to measurement. The Brunauer–Emmett–Teller method was used to calculate the surface areas.

Powder X-ray diffraction (XRD) patterns were recorded using a PANalytical X’Pert Pro system with Cu Kα radiation. Diffractograms were obtained at incident angles for 2θ = 5–65°.

Scanning electron microscopy (SEM) images were collected using a Zeiss Merlin system operated at 5.00 kV. STEM images were acquired through a JEOL NEOARM microscope operated at 80 kV.

Raman spectroscopy was performed on a multiwavelength Raman system using UV 244 nm laser excitation. Raman scattering was collected via a customized ellipsoidal mirror and directed by a fiber optics bundle to the spectrograph stage of a triple Raman spectrometer (Princeton Instruments Acton Trivista 555). An edge filter (Semrock) was used in front of the UV–vis fiber optic bundle (Princeton Instruments) to block the laser irradiation. A neutral density filter was used to attenuate the laser power to 20% so that the laser power at the sample position is less than 5 mW. A UV-enhanced liquid N2-cooled CCD detector (Princeton Instrument) was employed for signal detection. The Raman catalytic reactor (Linkam CCR1000) was placed on an XY stage (Princeton Scientific, OptiScan XY system). During the acquisition, the stage translated in the raster mode, which could provide the information of heterogeneity of the samples. The fast translation and the attenuated laser power were also able to minimize the laser damage of the sample. For dehydrated spectra, the catalysts were treated in 2% O2/He/Ar with a flow rate of 30 mL min–1 at 500 °C for 30 min, followed by cooling to 120 °C and acquiring the spectra after being stabilized for 30 min. During in situ Raman measurements, the dehydrated catalysts were sequentially exposed to 2% O2/He/Ar and then to C3H8/O2/He/Ar (C3H8/O2 = 1/1) with a duration of 30 min for each at 550 °C.

XPS was performed using a Thermo Scientific (Waltham, MA, USA) model K-Alpha XPS instrument. The instrument utilizes monochromated, microfocused Al Kα X-ray (1486.6 eV) with a variable spot size (i.e., 30–400 μm). Analyses of the sample were performed with the 400 μm X-ray spot size for maximum signal to obtain an average surface composition over the largest possible area. The instrument has a hemispherical electron energy analyzer equipped with a 128-channel detector system. The base pressure in the analysis chamber was typically 2 × 10–9 mbar or lower. The samples were prepared for analysis by dispersing the powder material onto double-sided tape fixed to a clean glass slide. After transferring the samples into the analysis chamber, the survey spectra (pass energy = 200 eV) were acquired for each sample. Next, high-resolution core level spectra (pass energy = 50 eV) were acquired for a detailed chemical state analysis. All spectra were acquired with the charge neutralization flood gun (combination of low energy electrons and argon ions) turned on to maintain a stable analysis condition. The typical pressure in the analysis chamber with the flood gun operating was 2 × 10–7 mbar. Data were collected and processed using the Thermo Scientific Avantage XPS software package (v.5.96).

Detailed procedures of XAS and SVUV-PIMS are described in the Supporting Information.

Results and Discussion

Activity Performance of VOx-Loaded h-BN Catalyts for ODHP Reaction

A series of VO-loaded BN-T catalysts were prepared by impregnation. The catalyst is denoted as xV/BN-T throughout the paper, wherein x represents the loading of vanadium in wt % (ranging from 0 to 1 wt %) and the “T” in BN-T indicates that the h-BN support was treated under C3H8/O2 atmosphere prior to impregnation. The resultant activity performance of VO/BN-T catalysts is shown in Figure . All VO-loaded BN-T catalysts exhibit higher C3H8 conversion (Figure A) and light-alkene (C2–C3=) yield (Figure B) than the benchmark BN-T under the same reaction conditions, especially for 0.5V/BN-T. At 540–580 °C, the light-alkene yield of 0.5V/BN-T is doubled compared to that of BN-T. The increase in the propylene yield on 0.5V/BN-T is also nearly doubled within the same temperature range (Figure S1A). The catalyst with the optimal V loading, namely, 0.5V/BN-T, presents good cyclability in three consecutive light-off tests (Figure S2). Noteworthily, the reaction conditions in the present work used diluted C3H8 compositions. To evaluate the stability and cyclability for practical purposes, further tests under high C3H8 compositions and long duration are warranted.

Figure 1

Temperature-dependent changes in C3H8 conversion (A), light-alkene (C2–C3=) yield (B), and plotted light-alkene selectivity-C3H8 conversion relationship (C,D) on VO/BN-T catalysts for the O2-ODHP reaction. Reaction conditions: C3H8/O2/He (1/1/38) and GHSV = 18,000 mL g–1 h–1.

As shown in Figure C, the selectivity of propylene decreases with the increase in propane conversion for all catalysts (at high temperatures), while an inverted trend is observed for ethylene selectivity (Figure D). This is consistent with previous work, which demonstrates the increased contribution of gas-phase radical chemistry at higher temperatures in the formation of the secondary product ethylene.[3] V/BN-T catalysts present relatively lower C3H6 selectivity than BN-T at lower C3H8 conversions, the difference of which is particularly appreciable for 0.5V/BN-T (Figure C). Meanwhile, BN-T also outperforms the VO/BN-T catalysts in C2H4 selectivity (Figure D), resulting in the decreased net selectivity of light alkenes on VO/BN-T than that on BN-T at higher conversions (Figure S1B). As aforementioned, Wang and Lin have proposed the methyl radical-involved reaction paths in the gas phase, one of which is to produce C2 products by the coupling reaction.[22] The presence of methyl radicals is then supported by the in situ SVUV-PIMS results.[19] Thus, the decreased C2H4 selectivity on V/BN-T is indicative of the interference from the VO-induced surface contribution.[4−7] Interestingly, the C3H6 selectivity of VO/BN-T tends to coincide with that of BN-T at higher conversions. This implicates that VO plays a role in propane dehydrogenation on the surface at lower conversions, while the characteristic gas-phase contribution of BN dominates at higher conversion.

To understand the role of VO, BO, and BN in the ODHP reaction, we performed two arrays of control experiments, including the comparison between 0.5V/SiO2, BN-T, and the physical mixture of 0.5V/SiO2 and BN-T, as well as the comparison between 1V/SiO2, 5B/SiO2, the physical mixture of 1V/SiO2 and 5B/SiO2, and 1V/5B/SiO2 (prepared by sequential impregnations). Figure S3 shows the results through the comparison between 0.5V/SiO2, BN-T, their physical mixture, and 0.5V/BN-T. In general, in comparison to BN-T, the physical mixture of 0.5V/SiO2 and BN-T exhibits similar trends in activity and selectivity to 0.5V/BN-T, in which a slight increase in C3H8 conversion and decreases in C3H6 and C2H4 selectivities are evident. In particular, the physical mixture presents moderate C2H4 selectivity between BN-T and 0.5V/SiO2. This indicates that the presence of BN-T contributes to the C2H4 formation through the gas-phase reactions, therefore leading to an increase in C2H4 selectivity in comparison to 0.5V/SiO2 alone.[19,22] Meanwhile, VO induces the surface reaction, resulting in a reduction in C2H4 selectivity in comparison to BN-T.[4,8] These observations are consistent with those on V/BN-T catalysts, corroborating the contributions from both surface and gas-phase reactions with the combination of VO and BN. Notably, the enhancement of activity on 0.5V/BN-T significantly surpasses that on the physical mixture (Figure S3A,B). This demonstrates better synergism by loading a certain density of VO directly onto BN. By contrast, such an enhancement is not evident from the other array of control experiments on 1V/SiO2, 5B/SiO2, and their physical mixture (Figure S4). Nor is it observed on 1V/5B/SiO2. Besides, the physical mixture only exhibits a slightly higher selectivity toward C2H4 than 1V/SiO2 under isoconversional conditions (Figure S4D), which is, however, much lower than 5B/SiO2 or 1V/5B/SiO2. In conjunction with the observed promoting effect on V/BN-T, such a contrast between these two arrays of control experiments implies that the presence of VO on BN-T not only contributes to the ODHP reaction but may also tune the BN-T surface with positive impacts, such as oxyfunctionalization. This will be studied by XAS and XPS and is discussed in Section .

For V/BN-T catalysts, the apparent activation energy (Eapp) decreases with the increase in V loadings (i.e., ca. 250 to 200 kJ mol–1, Figure A), indicative of the surface contribution of added VO in the observed synergism on improved activity. This is also corroborated by the results of kinetic studies at different C3H8 partial pressures, in which the propane consumption rate is 1.4-order dependent on the propane partial pressure in the presence of VO (Figure B), lower than the 1.8-order on the BN support alone (Figure S5) and the reported second-order dependence on other BN alone.[3] Noteworthily, the lowest Eapp attained on 1.0V/BN-T is still higher than those reported for supported VO catalysts for O2-ODHP (i.e., 80–170 kJ mol–1).[6] Clearly, the ER mechanism, proposed for BN catalysts, still dominates despite the presence of VO.

Figure 2

(A) Variations of apparent activation energy (Ea) as a function of V loadings and (B) effect of C3H8 composition % on the C3H8 conversion rate on 0.5V/BN-T at 540 °C. Reaction conditions in (A): C3H8/O2/He (1/1/38), GHSV = 18,000 mL g–1 h–1, and C3H8 conv. <15%. Reaction conditions in (B): 0.4–2.5% C3H8/2.5% O2, balanced in He and Ar and GHSV = 24,000 mL g–1 h–1 (see Figure S6 for the effect of O2 composition % on the C3H8 conversion rate).

Characterization of Catalyst Structures

To study the catalyst structure, especially the interaction between surface BO/BN and VO species, and its correlation with the enhanced ODHP activity, the VO/BN-T catalysts were characterized by various techniques. The SEM images of VO/BN-T show a morphology similar to those of fresh BN and BN-T (Figure S7). Diffraction peaks of BN phases dominate the XRD patterns for VO/BN-T catalysts (Figure S8). No clear diffraction peaks are evident for VO phases when the V loading is below 1.0 wt %. This suggests the formation of either highly dispersed (nanocrystalline) or amorphous V-containing species. Due to the nanostructured/amorphous nature and lower V loadings, XRD is not suitable to provide structural information of VO.

The same catalysts were then characterized by Raman spectroscopy (see Figure S9 for BN and BN-T). The characteristic G band of BN, namely, the interlayer Raman active E2g mode, is observed at 1370.8 cm–1 for fresh BN, while the band slightly red shifts to 1369.8 cm–1 for 0.75V/BN-T, implying the VO-induced localized fixation effect on reduced in-plane strain (Figure S10).[32]Figure A shows the Raman spectra of dehydrated VO/BN-T catalysts. The two peaks, centered at 819 and 922 cm–1, are not due to any fundamental modes of h-BN but have been observed in previous studies with UV Raman.[33] Their assignment has not been fully resolved. Since the two bands are quite stable throughout our various experiments, they are not considered relevant to the ODHP reaction, so we focus rather on the surface vanadia species. For dehydrated catalysts, a band is observed at 1029 cm–1 for all VO/BN-T samples, signifying polyvanadate on the BN surface.[7,34] No sign of the crystalline V2O5 is observed at ca. 997 cm–1,[7,34] demonstrating that the VO is less than a monolayer coverage for all VO/BN-T catalysts. The Raman spectrum of the spent 0.5V/BN-T, collected after the O2-ODHP reaction, shows a new band at 882 cm–1 (Figure B), which can be attributed to the borate species with hydroxylated non-ring boron (B-OH).[35] For comparison, BO/SiO2 and VO/BO/SiO2 have been prepared as benchmarks, and similar peaks are also evidenced (see Figure S11 for details). This band disappears after the dehydration, and the V=O stretch appears at a similar shift as seen in the fresh and dehydrated sample, indicating the integrity of the VO species during the ODHP.

Figure 3

Raman (244 nm excitation) spectra of dehydrated VO/BN-T (A) and 0.5V/BN-T (B) under different conditions. Dehydration conditions: 2% O2/He/Ar, 30 mL min–1, 500 °C, and 30 min. The spent catalyst was collected after the ODHP activity test.

The structure of VO species is confirmed by high-angle annular dark-field STEM (HAADF-STEM) analysis, which clearly evidences the presence of VO nanoclusters (polyvanadates) in Figure . Meanwhile, some single atoms are also observed from the HAADF images, indicating the presence of a small portion of monovanadate. To our knowledge, this is one of a few cases of coupled spectroscopy and microscopy in identifying the nature of VO in powder samples, thanks to the 2D nature of BN and high contrast in STEM.

Figure 4

HAADF images of 0.75V/BN-T in different regions, along with the fast Fourier transform images of the BN substrate in the inset of (A). The arrows and circles highlight the surface monovanadate and polyvanadate, respectively.

Figure A shows the normalized pre-edge peak of the vanadium K edge XANES (X-ray absorption near edge structure) for 1.0VO/BN-T catalysts under different in situ treatments, along with those of reference compounds. Two sets of in situ measurements were performed on fresh 1.0V/BN-T. The first measurement, comprising consecutive treatments in O2, C3H8, and O2 regeneration at 500 °C, was done to verify the reversible redox cycles of VO under reaction conditions. The catalyst treated in C3H8 presents a strong decrease in the intensity of the pre-edge peak and a shift toward lower photon energy, implying a partial reduction in the oxidation state and corresponding changes in the point group symmetry of vanadium in the catalyst (Figure A). Subsequent O2 regeneration enables the recovery of the oxidation state of vanadium, as evidenced from the match between the initially O2-treated and O2-regenerated XANES. Clearly, VO can undergo redox cycling under our reaction conditions. To quantify the oxidation state and understand the local symmetry around vanadium, the normalized pre-edge peak area was plotted as a function of the pre-edge peak centroid. This method of XANES analysis has been widely used for the identification of unknown compounds in 3d metals.[36,37] The catalysts fall in a region bounded by V5+ and V4+ compounds, which is shown in Figure B. The O2-treated catalyst, the O2-regenerated catalyst, and the catalyst treated in C3H8/O2 are all very similar in oxidation state and symmetry. After regeneration, the pre-edge peak shifts up in area and down in energy position, closer to that of NH4VO3 (Td V5+). This would be consistent with a small increase in the fraction of vanadium with tetrahedral coordination. This could be caused by the formation and dispersion of polyvanadate on the surface of BN-T or on BO formed via oxyfunctionalization. The absence of inactive crystalline phase V2O5 is also confirmed from the XANES results. These observations agree with the Raman results.

Figure 5

Background-subtracted XANES spectra of vanadium pre-edge peaks for 1.0V/BN-T after consecutive treatments in O2, C3H8, and O2 regeneration (A) and corresponding normalized pre-edge intensity vs pre-edge centroid energy relative to the threshold energy of V metal (5465 eV) for 1.0V/BN-T under different treatments and for reference compounds (B). Spectra of reference compounds, as well as detailed data analysis and discussion, can be found in the Supporting Information (Tables S1, S2 and Figures S12, S13).

The second measurement includes exposing the same fresh catalyst in a mixture of C3H8/O2 at 500 °C in an effort to determine the predominant oxidation state of vanadium under steady-state reaction conditions. As shown in Figure A, the catalysts treated in O2 and C3H8/O2 present almost overlapping XANES spectra, demonstrating that a majority of V stays in +5 under reaction conditions. This is also corroborated by the similar normalized pre-edge intensity versus pre-edge centroid energy, in which the pre-edge peak area and centroid are, within error, the same as that of the O2-treated catalyst (Figure B).

Figure 6

Background-subtracted XANES spectra of vanadium pre-edge peaks for 1.0V/BN-T after the single treatment in C3H8,/O2 as well as that after O2 treatment as reference (A) and corresponding normalized pre-edge intensity vs pre-edge centroid energy relative to the threshold energy of V metal (5465 eV) for 1.0V/BN-T and for reference compounds (B). Spectra of reference compounds, as well as detailed data analysis and discussion. can be found in the Supporting Information (Tables S1, S2 and Figures S12, S13).

For the coordination environment, the Fourier-transformed k3χ(k) (extended X-ray absorption fine structure) of fresh 1.0V/BN-T, treated in situ in C3H8 and O2, shows a nearest neighbor-peak position consistent with oxygen, indicating the V–O bond of the first shell of 1.0V/BN-T (Figure S14). Combined with the spectra of pre-edge XANES, the formation of VB2 structure can be excluded under reaction conditions. Instead, VO species are anchored at the surface through the B–O–V bond. In sum, vanadium is present mostly as V5+ under reaction conditions, while the redox cycles of VO still occur with a rapid kinetics. In other words, VO contributes to the ODHP reaction via MvK mechanism with redox cycles.

To explore the effect of VO on the gas-phase radical chemistry, the gas-phase components from ODHP over 1.0V/BN-T and the support BN-T were analyzed online by SVUV-PIMS, and the results are shown in Figure . On BN-T, the methyl radical (CH3•), with the m/z value at 15.03, is observed, demonstrating the contribution of gas-phase radical chemistry in attaining such a high selectivity toward light alkenes (Figure A).[19] As shown in Figure B, adding VO leads to a notable relative increase in NO and a relative reduction in CH3• and C2H4. It is likely that the presence of VO boosts the surface reaction, and the formation of the B–O–V structure tunes the local environment of the active sites B-OH, thereby perturbing the gas-phase radical chemistry. Interestingly, the increase in NO in the gas phase is indicative of the NO release from the surface occurring concurrently with the reaction. Such a peculiar phenomenon can be associated with the dispersed VO species at the surface that facilitate the oxyfunctionalization of BN. In contrast, NO is barely observed from the SVUV-PIMS spectrum on BN-T in the gas phase, except a trace amount at 600 °C (Figure C). The blank test (Figure S15) shows no evident signals of gas-phase methyl radicals (m/z = 15) in the absence of a catalyst at 600 °C, nor does the control test present any NO (m/z = 30) signals in the presence of a catalyst at room temperature. This implies that all detected species in the gas phase on both BN-T and 1.0V/BN-T under ODHP reaction conditions are derived from the catalyst.

Figure 7

Integrated ion intensities of various components in the gas phase of O2-ODHP reaction over BN-T (A) and 1.0V/BN-T (B) at 600 °C. (C) SVUV-PIMS spectra of the gas-phase NO for ODPH reaction at 600 °C on BN-T and 1.0V/BN-T. Photon energy = 9.5 eV. Note that only a trace amount of NO was detected for BN-T at 600 °C, so that it is not included in (A).

Early work has reported the positive role of NO in the gas-phase reaction in the partial oxidation of methane to oxygenates[38,39] and ODHP reaction via both homo- and heterogeneous catalysis.[40] The evolved NO in the gas phase inspires us to perform additional activity tests of ODHP reaction in the presence of various NO concentrations. As expected, such a NO-induced enhancement in activity is evidenced with the additional NO concentration below 350 ppm. The selectivity is also slightly increased. An excess addition such as 3500 ppm is detrimental (Figures S16 and S17). Similar NO concentration-dependent trends are also evidenced on SiO2-supported BO catalysts (Figure S18). Of note, although the NO evolution is evidenced, it does not necessarily imply that this restructuring procedure of BN is continuous until all nitrogen species are stripped out. The light-off tests show the repeatability of 0.5V/BN-T for three consecutive tests under the present reaction conditions, demonstrating that VO plays an important role in the observed enhancement in activity (Figure S2).

To further verify the VO-facilitated oxyfunctionalization, XPS measurements were conducted on both fresh and spent BN-T and 0.75V/BN-T. Figure A shows the XP spectra of all samples in the region of B 1s. The major peak, centered at 190.8 eV, corresponds to B–N species on a flat surface for all samples.[11,41] Further deconvolution analysis reveals the other two B-containing surface species for fresh and spent BN and fresh 0.75V/BN-T, and they are B–O and BO centered at 191.3 and 192.2 eV, respectively.[9,41] For spent 0.75V/BN-T, the peak intensity of BO increases notably; moreover, a new peak emerges at 193.5 eV corresponding to B2O3 species at the surface.[41,42] Of note, the peak of B–O species is absent for the spent catalyst, which might be associated with the restructuring of the BN surface with the assistance of VO under the reaction conditions. Due to the formation of more BO species, a blueshift of O 1s peak for this spent 0.75V/BN-T is evidenced, signifying the substitution of nitrogen with oxygen at the BN surface (Figure S19).[14] Clearly, the presence of VO is conducive to forming BO species via oxyfunctionalization under reaction conditions, in line with the observation by the evolved NO from SVUV-PIMS. This can also be corroborated from the quantification analysis, as evidenced from the distinct higher oxygen and BO compositions in the spent samples (Tables and S3). The major band at 398.2 eV in N 1s XP spectra is the nitrogen bonded with boron in BN at the surface (Figure B).[11] A subpeak at 400.0 eV corresponds to N–H species, implying the formation of N defects,[9] and these N–H species maximize at spent 0.75V/BN-T (Table S4). Recently, Lu et al. have reported that these N defects might lead to the evolution of active BO species.[11] Clearly, combined with the observations in this work, adding VO enables the tuning of the local chemical environment by facilitating the BN oxyfunctionalization to BO, along with the release of NO into the gas phase.

Figure 8

XP spectra of fresh and spent BN-T and 0.75V/BN-T catalysts in the regions of B 1s (A) and N 1s (B).

Table 1

Surface Composition of B-containing Species by XPS Analysis for Fresh and Spent BN-T and 0.75V/BN-T

	surface composition/at. %
catalyst	B–N	B–O	BOx	B2O3
fresh BN-T	29.7	13.5	2.5
spent BN-Ta	27.7	8.4	1.7
fresh 0.75V/BN-T	26.1	11.5	1.3
spent 0.75V/BN-Ta	34.5		6.5	4.2

Spent catalysts were collected after 12 h ODHP activity test at 600 °C.

Conclusions

In summary, a significant enhancement of light-alkene yield in ODHP is achieved through the synergism between VO and BN. The VO-induced increase in activity can be attributed to the well-dispersed VO species anchored at the surface as mono- and polyvanadate via B–O–V structure, which facilitates the surface oxyfunctionalization and exposes more BO for C–H activation. Meanwhile, NO is released during the facilitated oxyfunctionalization of BN, which also contributes to the rise in activity via mediating the gas-phase radical chemistry. In addition to the above roles, VO catalyzes the reaction via the redox cycles on the surface. Adding additional NO into the feed gas leads to further enhancement in the activity. While the quantification of the contribution from gas-phase and surface reactions is challenging and warrants further investigations, our findings shed light on developing efficient boron-based catalysts for the ODHP reaction by manipulating the interplay between homogenous gas-phase catalysis and heterogeneous surface catalysis.