Oxidative Dehydrogenation of Propane over a High Surface Area Boron Nitride Catalyst: Exceptional Selectivity for Olefins at High Conversion.

Developing environment-friendly active and selective catalysts for oxidative dehydrogenation of propane for on-purpose propene synthesis is challenging despite tremendous industrial potential for this reaction. Herein, we report on catalytic activity of high surface area hexagonal boron nitride, toward oxidative dehydrogenation of propane. It shows remarkable selectivity for alkenes (∼70%) at very high conversion (of ∼50%) of propane. Propene and ethene selectivities as high as 53 and 18%, respectively, were obtained at a conversion of 52%. The catalytic activity is retained continuously for 5 h. Regeneration in ammonia brings back the catalytic activity to its full potential. Oxidation of surface B-N bonds in oxygen leads to the diminishing catalytic activity after 5 h which, on heating in ammonia, reduced back to their native form, regaining the indigenous activity. Remarkably, the addition of ammonia in the reaction feed showed stable activity for more than 100 h.

Introduction

Oxidative dehydrogenation (ODH) of propane for its exothermic nature, low operative temperature, and minimal coke deposition is a promising alternative to the industrial dehydrogenation reaction which is endothermic and plagued with rapid catalyst deactivation.[1−3] However, the ODH catalysts reported so far, mostly vanadium- and molybdenum-based redox metal oxides, show very poor selectivity for propene at higher propane conversion (>20%), where complete burning of hydrocarbons predominates.[3,4] The carbon nanotubes, which was also used as a catalyst for the ODH reaction, showed about 70% propene selectivity at about 5% conversion[5] with its inherent limitation to use it at higher temperature in the presence of oxygen and hydrocarbons.[6] Recently, h-BN was reported to show unusual 79% selectivity for propene but again only at a low propane conversion of 14%.[7] Boron nitride (BN) for its high thermal stability and thermal conducting property[8,9] naturally is the best catalyst to carry out the ODH reaction at higher temperatures. Nevertheless, the loss of selectivity for olefins is a major impeding factor when it is sought for high propane conversion. Herein, we report for the first time a high surface area BN catalyst displaying remarkable selectivity for olefins (∼70%) in ODH of propane even at high conversion (52%) and at a high O2/propane ratio.

Results and Discussion

High surface area BN was synthesized from boric acid (BA) and dicyanamide precursors using a reported method (see the experimental section).[10] The white-colored BN powder obtained consists of micron size particles which were porous (Supporting Information Figure S1a,b) in nature. The powder X-ray diffraction (XRD) pattern (Supporting Information Figure S1c) of the BN sample showed broad diffraction peaks corresponding to (002) and (100) planes, which confirms the hexagonal structure of the BN materials. The interplanar d-spacing value, corresponding to the (002) plane, that is, 0.354 nm, was close to that of the turbostatic BN (0.356 nm).[11] The N2 adsorption–desorption measurement shows the type IV isotherm (although the typical capillary uptake falls in the low P/P0 region with no hysteresis) along with the features for type I (Supporting Information Figure S1d) with a Brunauer–Emmett–Teller (BET) specific surface area (SSA) of 1380 m2/g.

The catalytic activity of the BN material toward the propane ODH reaction was tested by passing a mixture of propane, oxygen, and helium (propane/O2 ratio 0.02) through a quartz reactor loaded with the BN catalyst at different temperatures (390–570 °C) and analyzing the products on an online gas chromatogram (GC) system (experimental details, Supporting Information Figure S2). The low propane/O2 ratio of 0.02 was the best optimized ratio to obtain high conversion with good selectivity for olefins (Supporting Information Figure S3), and it still falls within the safety regime.[12]Figure a shows that the BN catalyst started showing activity for propane conversion beyond 400 °C, and the products observed in a significant amount were propene, ethene, and carbon dioxide (Figure a). With increasing temperature, the conversion increases sharply, and at 525 °C, it showed 52% conversion (Figure a) with 53% selectivity for propene. Taking into account the formation of 18% ethene, the total selectivity for olefins goes up to 70%. This is indeed remarkable, considering the fact that no catalyst shows such high selectivity for olefins at this high propane conversion.[4,7,12−15]Figure b compares the selectivity of propene with propane conversion over different catalysts for the ODH reaction. The carbon balance at 525 °C for this reaction was found to be around 98.8 ± 0.8%. Control reactions with a blank reactor tube and tube loaded with the packing material (quartz wool) showed conversions less than 1% (Supporting Information Figure S4), strongly supporting the catalytic role of BN in the ODH reaction of propane.

Figure 1

Catalytic activity of BN materials. (a) Propane ODH reaction over the BN catalyst. Reaction conditions: 0.1 g of catalyst, 20 mL min–1 total gas flow rate, and C3H8/O2/He = 1/50/49. (b) Propene selectivity vs propane conversion plot for the propane ODH reaction, comparing the BN catalyst with reported catalysts. Propane conversion (black squares), propene selectivity (blue circles), ethene selectivity (wine red triangles), total olefin selectivity i.e., summation of propene and ethene selectivities (green spheres), and CO2 selectivity (dark yellow pentagons). Open shapes in (b) indicate data from other works, which are cited within the figure. Methane was formed only in trace amounts (less than 1%) for the BN catalyst.

The yields of propene and total olefins (summation of propene and ethene) over the high surface area BN at 525 °C were around 28 and 38%, respectively, and a propene productivity of ∼0.625 kg per kg of the catalyst can be achieved in 10 h time. It is worth noting that the maximum propene yield obtained in this study (around 28%) is comparable to the maximum yield achieved with the metal-based catalysts (around 30%)[3] and is significantly higher than the maximum yield reported for metal-free nanocarbon catalysts (around 17%).[16] Further, as compared to other metal-free catalysts such as carbon-based nanostructures, BN materials have significant advantages in terms of their high thermal conductivity and their excellent thermal stability at high temperature (upto 800 °C, Supporting Information Figure S5) oxidizing environment.[17,18]

BN materials synthesized using BA–urea[19] and BA–melamine[20] as precursors (Supporting Information Figures S6, S7, and S8a) were also active toward the propane ODH reaction (Supporting Information Figure S8b); however, for a given propane conversion (∼50%), these samples showed less selectivity for propene. The commercial bulk BN (Sigma-Aldrich, measured BET SSA 48 m2/g) in our reaction conditions did not show any significant propane conversion (<1%) up to 550 °C (Supporting Information Figure S9).

Reactions carried out at higher residence time over BN catalysts decrease the propene selectivity (Supporting Information Figure S10) because of combustion/cracking of propene/propane. When extrapolated to zero residence time, the nonzero CO2 and ethene selectivities (also <100% propene selectivity) suggest that propane undergoes two parallel reactions, viz., combustion and cracking. Separate experiments with a propene and oxygen mixture (in a ratio of 1:50) over the BN catalyst under same conditions mainly give CO2 because of burning of propene and suggest that ethene could be formed through cracking of propane. Thus, the propene selectivity is limited by parallel (cracking/combustion of propane) and sequential (combustion of propene) reactions.

A time-on-stream (TOS) study at 525 °C shows that the activity of the BN catalyst was preserved up to 5 h (Figure a), beyond which the conversion gradually decreases and reaches below 20% in 12 h time with concomitant increase in propene selectivity up to 80%. The absence of any weight loss in thermogravimetric analysis (TGA) (Supporting Information Figure S12a) for the spent BN catalyst after the 24 h reaction (BN-24h) in the region of 200–800 °C suggests that there is no coke deposition which could deactivate the BN catalyst beyond 5 h. The XRD pattern of the BN-24h sample shows a pure hexagonal phase similar to that of the as-prepared BN sample (Supporting Information Figure S12b). The nitrogen adsorption–desorption isotherm exhibits type I behavior with a BET surface area of 1368 m2/g (Supporting Information Figure S12c). The transmission electron microscopy (TEM) image (Supporting Information Figure S12d) further confirms that the BN-24h sample was highly porous. Fourier transform infrared (FTIR) spectroscopy spectra of BN-24h, however, show weak but significant IR bands corresponding to N–B–O (850–1200 cm–1) and O–B–O (530–730 cm–1) stretching/bending modes (Figure a) which were absent in the as-prepared BN sample, suggesting partial oxidation of B–N bonds. Further, the B 1s core-level X-ray photoelectron spectroscopy (XPS) spectra of BN-24h exhibit major peaks at 192.0 and 193.8 eV, corresponding to BNO and B–O species, respectively (Figure b). Besides, the peak intensity of BN3 species (190.5 eV) has come down in BN-24h as compared to the BN material (Figure b). The presence of NBO (399.95 eV) and N–O (401.55 eV) peaks in the N 1s core-level spectra of BN-24h further supports partial oxidation of B–N bonds, leading to its catalytic deactivation with time. It is interesting to note that heating the BN catalyst in oxygen flow for 12 h did not show any IR bands corresponding to N–B–O and O–B–O stretching/bending modes (Supporting Information Figure S13a), and the oxygen-treated sample exhibited similar catalytic activity for the propane ODH reaction as that of the BN materials (Supporting Information Figure S13b). This is in accordance with the high stability of BN in an oxygen environment (Supporting Information Figure S5) and suggests that oxidation of the BN catalyst takes place only when propane is present in the reaction feed.

Figure 2

Stability of the BN catalyst for the propane ODH reaction. (a) TOS reaction with the BN sample at 525 °C. (b) Conversion and selectivity vs temperature for the regenerated BN sample, viz., BN (R). (c) TOS reaction with the BN (R) sample at 525 °C. (d) Catalytic deactivation and regeneration cycles in which the propene yield (pink spheres) is plotted at the start (0 h) and the end (24 h) of the TOS reaction. Reaction conditions: 0.1 g of catalyst, 20 mL min–1 total gas flow rate, and C3H8/O2/He = 1/50/49. Propane conversion (black squares), propene selectivity (blue circles), ethene selectivity (wine red triangles), total olefin selectivity (green spheres), and CO2 selectivity (dark yellow pentagons).

Figure 3

Spectroscopic characterization of BN, BN-24h, and BN (R). (a) FTIR spectra. (b) B 1s and N 1s core-level XPS spectra. Characteristic IR bands for BN are at around 1385 and 800 cm–1 because of B–N stretching and bending modes, respectively, whereas IR bands in the region of 850–1200 and 530–730 cm–1 are due to N–B–O and O–B–O stretching/bending modes, respectively.

The BN-24h sample was regenerated by heating it in the ammonia flow at 900 °C for 2 h. Interestingly, the regenerated BN sample, BN (R), showed a similar catalytic behavior in terms of propane conversion (Figure b), olefin selectivity (Figure b), and catalyst stability (Figure c) as the original BN materials. This confirms the complete regeneration of the catalyst. Catalyst deactivation and regeneration was carried out for 10 cycles, and after every regeneration step, total recovery of the catalytic performance of the BN material was observed (Figure d), which indicates robustness of these materials in the reaction conditions. The FTIR spectrum of BN (R) was very similar to that of the BN sample without any N–B–O and O–B–O stretching/bending modes (Figure a). Also, the B 1s and N 1s core-level XPS spectra of BN (R) were similar to those of BN, with major peaks corresponding to BN3 and NB3 units and small shoulder peaks to BNO and NH species. Also, the B–N ratio (calculated from an XPS survey scan) for BN-24h was 1.30 ± 0.06, which is nearly 30% higher as compared to that for BN (1.00 ± 0.07) and BN (R) (1.0 ± 0.1), indicating that the loss of nitrogen during oxidation of BN is getting regenerated to its original form, hence regaining its activity fully. From these results, it is evident that the surface B–N bonds are the active catalytic sites for the propane ODH reaction. Indeed, when the ODH reaction was carried out in the presence of ammonia (C3H8/O2/NH3 = 1:50:2.5; although the exact quantity required for regeneration would be very low, as ammonia may get oxidized in oxygen at elevated temperature[21]), the BN catalyst showed stable activity even after 100 h of a continuous reaction (Figure a) with a propane conversion of >50% at 540 °C (Supporting Information Figure S14b). The propene and the total olefin selectivities were >50 and >70%, respectively (Supporting Information Figure S14b). The IR spectra of the spent catalyst (after 100 h) were similar to those of the fresh BN sample and did not show any band corresponding to the oxidized B–N species (Figure b). Clearly, in the presence of ammonia in the reaction feed, the oxidized B–N bonds, during the propane ODH reaction, get reduced back at the reaction temperature, resulting in no deactivation of the catalyst.

Figure 4

(a) TOS reaction in the presence of ammonia in the feed. Reaction conditions: 0.2 g of catalyst, 60 mL min–1 total gas flow rate, and C3H8/O2/NH3/He = 0.67/33.33/1.67/64.33. Propane conversion (black squares), propene selectivity (blue circles), ethene selectivity (wine red triangles), and total olefin selectivity (green spheres). (b) FTIR spectra of fresh BN and spent BN (after the TOS reaction in NH3 for 100 h) samples.

Although it is complex to elucidate the exact mechanism, our spectroscopic (IR and XPS) studies of BN, BN-24h, and BN-R strongly suggest that surface B–N bonds are the active catalytic site and species such as BNO, NBO, N–O, and B–O (present on the surface of BN-24h) are catalytically inactive. Clearly, nitrogen (a basic site) linked with BN at the edge sites is an integral part of the active species, and the reaction is probably initiated by hydrogen abstraction from propane. This is evident from the fact that the deactivated catalyst regains its activity when it is treated with ammonia and all the N–B–O and O–B–O sites disappear as confirmed by the FTIR and XPS studies. Furthermore, the catalyst does not lose its activity when the reaction was carried out in the presence of ammonia. These observations are in contrast with the previous reports (with the oxygen to propane ratio <2 in the feed), where oxygen-rich species B–O–O–N[7] and B–O–H[22] were suggested to be the active sites. To understand the basis of these differences, additional experimental and theoretical studies are needed. Further, there could be various B–N species on the BN surface (edge)[23,24] such as B–N, N–B, B–N–B, N–B–N, and B–NH2, and for precise understanding of the active sites, thorough in situ catalyst characterization will be essential.

Conclusions

In conclusion, we report here on the catalytic behavior of high surface area hexagonal BN for the propane ODH reaction. High propene selectivity and yield obtained for the BN catalyst were comparable or even better relative to those of the reported catalysts. The catalytic activity was observed to be stable for nearly 5 h after which it decreases; however, it could be completely regained by regenerating the catalyst in the ammonia flow at elevated temperature. Oxidation of the dangling B–N bonds at the edges and in the pores was found to be responsible for the diminished catalytic activity, which reduced back to its original form on heating in the ammonia flow. This suggested that the surface B–N bonds could be the active catalytic centers for the propane ODH reaction; however, a detailed in situ characterization and mechanistic study is needed to understand the reaction mechanism. Further, by introducing ammonia in the reaction feed, the BN samples showed remarkable TOS stability for over 100 h. Currently, most (>85%) of the propene is produced as a byproduct of the ethene industry which uses heavier feedstocks such as naphtha.[2] However, the advent of shale gas (methane, ethane, propane, and butane) will be driving the ethene industry away from heavier feedstock toward low-cost ethane feedstock.[25] Thus, in future, on-purpose propene production from propane, which presently accounts for around 14% of the total propene production, would be heavily burdened and will become highly competitive.[2,25,26] In this backdrop, porous BN materials appear to be a very promising catalyst for the propane ODH reaction.

Materials and Methods

Materials

BA (H3BO3) and dicyanamide (C2H4N4) were purchased from Sigma-Aldrich. Propane (2.0%, He balance), propene (2.0%, He balance), ethene (2.0%, He balance), oxygen (99.99%), helium (99.99%), ammonia (99.99 and 5.0% in He), and carbon dioxide (10.1%, He balance) gases were purchased from Chemix, India. All the chemicals and gases were used as received from the company. Water and methanol were used as solvents wherever required.

Synthesis of BN

BA (2.74 g) and 10.1 g of dicyandiamide were dissolved in 100 mL hot water (∼85 °C) (Supporting Information Figure S15). Solution was stirred at 85 °C till the solvent evaporated completely. The resulting powder was thoroughly grinded (using a mortar–pestle) and heated in a tubular furnace (Elite Thermal Systems Limited, TSH15/50/180-2416CG) at 1000 °C for 3 h in the ammonia flow (∼20 mL min–1) (Supporting Information Figure S15). The rate of heating and cooling was maintained at 10 °C/min. A white BN powder obtained was characterized.

Characterization

Morphologies of the samples were analyzed by field emission scanning electron microscopy (FEI NovaNano SEM-600, Netherlands). TEM images were acquired with a JEOL JEM 3010 instrument (Japan) operated with an accelerating voltage of 300 kV. Powder XRD patterns were acquired at room temperature with a Bruker-D8 diffractometer employing Cu Kα. FTIR spectra were acquired on a Bruker IFS 66v/S instrument in the range of 4000–400 cm–1. The XPS measurements were carried out with an Omicron spectrometer using Al Kα as the X-ray source (1486.6 eV). On a copper tape, a thin film of BN was drop-casted, and over this film, a carbon tape was stuck. The XPS spectra were acquired close to the carbon tape. The peak positions and the binding energy shift were interpreted mainly by referring to The Handbook of X-ray Photoelectron Spectroscopy[27] also to some of the literature.[28−30] The binding energy scale was calibrated using the carbon 1s peak at 284.5 eV. The SSA of the samples was measured using N2 sorption analysis at 77 K on Autosorb iQ2. BN samples were outgassed at 150 °C under vacuum for 24 h before the N2 sorption study. TGA was carried out on PerkinElmer Pyris 1 TGA under an oxygen flow of 20 mL min–1. The heating rate was kept at 5 °C/min.

Catalyst Testing

Catalytic activity for the propane ODH reaction was tested by fixing 100 mg of the catalyst at the center of a quartz tube reactor (40 cm length and 9 mm inner diameter) with the help of quartz wool on both sides (Supporting Information Figure S2a). A mixture of 10 mL min–1 2% propane in He and 10 mL min–1 oxygen (99.99%) (C3H8/O2 = 1:50, total flow rate = 20 mL min–1) was passed through the reactor tube at 1 atm using mass flow controllers (MFC, MKS Systems Singapore) (Supporting Information, Figure S2b). Reactor temperature was controlled by heating it inside a tubular furnace (Carbolite, MTF 12/38/250). The products were analyzed using online GC (Agilent 6890N and 7890A) systems fitted with alumina (Rt-Alumina BOND/Na2SO4, 30 m × 0.32 mm × 5 μm), monolithic carbon (Agilent GS-CarbonPLOT 30 m × 0.32 mm I.D. × 3.0 μm), and molecular sieve (HP MoleSieve, 30 m × 0.32 mm × 12 μm) columns. The propane conversion and product selectivities were calculated according to the equations given below.[31]