Water-Tolerant Boron-Substituted MCM-41 for Oxidative Dehydrogenation of Propane.

Boron-based catalysts for oxidative dehydrogenation of propane (ODHP) have displayed excellent olefin selectivity. However, the drawback of deboronation leading to catalyst deactivation limited their scalable applications. Hereby, a series of mesoporous B-MCM-41 (BM-x, B/Si = 0.015-0.147) catalysts for ODHP were prepared by a simple hydrothermal synthesis method. It was found that propane conversion was increased and the initial reaction temperature was reduced with an increase of boron content, and the optimal values appeared on BM-2.0 (B/Si = 0.062), while olefins' (ethylene and propylene) selectivity was maintained at ca. 70-80%. Most importantly, BM-1.0 (B/Si = 0.048) exhibited favorable activity, stability, and water tolerance after washing treatment or long-time operation (e.g., propane conversion of ca. 15% and overall olefin selectivity of ca. 80% at 550 °C) because its high structural stability prevented boron leaches. These features were identified by X-ray diffraction (XRD), N2 physisorption, inductively coupled plasma-mass spectrometry (ICP-MS), scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), and solid-state magic-angle-spinning nuclear magnetic resonance (MAS NMR) spectroscopy studies. The tri-coordinated B-OH species incorporated into the mesoporous silica framework are considered to be the active sites for ODHP.

Introduction

Light olefins (ethylene and propylene) are essential materials in the petrochemical industry.[1] The large-scale exploitation of shale gas, as a cheap and abundant source of light alkanes (ethane and propane), leads to the increasing proportion of light olefins’ production via the on-purpose route, namely, the direct dehydrogenation of ethane or propane (DHE or DHP).[2] Oxidative dehydrogenation of propane (ODHP) is a promising alternative to DHP on account of low reaction temperatures and inhibition of coking.[3] Recently, several boron-containing catalysts with superior selectivity to olefins (propylene is dominant, followed by ethylene), such as commercial h-BN,[4] BNOH,[5] borate-modified nanodiamonds,[6] metal borides,[7]etc., have been reported. The exciting feature of high olefin selectivity at high temperatures is not available in traditional metal oxide catalysts.[8] The oxidized boron sites formed during the ODHP reaction are considered to be the active centers.[9] As a result, the boron oxide catalysts, which were studied in the 1990s,[10] have been revitalized again.

Silica-supported boron oxide catalysts for ODHP have been investigated by Hermans’s[11] and Lu’s[12] groups simultaneously. They prepared supported boron oxide catalysts via incipient wetness impregnation on silica nanoparticles (Aerosil 300)[11] and mesoporous silica (SBA-15),[12] respectively, all of which exhibited high selectivity (>70%) to propylene. The oxidized tri-coordinated boron species are considered to be the catalytic active sites for ODHP. Meanwhile, the former pointed out that boron leaches of the catalysts with high boron loading was obvious during the activation under the feed gas,[11] and the latter found that the B2O3 content of the catalyst was decreased by 10 times after washing with water, resulting in the reduction of propane conversion reduced by 30 times at 500 °C.[12] For B2O3@BPO4 hollow sphere catalysts, the loss of B was also distinct after washing.[13] Visibly, these results show that the supported boron oxide catalysts are not water-tolerant during ODHP catalysis or washing with water. Interestingly, to solve this problem, both the groups researched the catalytic performance of microporous borosilicate zeolite MCM-22 (B-MWW) for ODHP.[14,15] Hermans’s group[14] successfully incorporated the isolated BO3 units into the zeolite framework as the major boron species in B-MWW, but the as-synthesized B-MWW did not show any catalytic activity for ODHP. Meanwhile, the aggregated boron oxides supported on B-MWW were observed to be active in ODHP. Conversely, Lu’s group[15] demonstrated that B-MWW can be an efficient catalyst for ODHP. However, the boron content of the spent B-MWW was greatly reduced from 2.48 to 0.85 wt % after washing with water. Comparing the above two studies, the structural stability of B-MWW seems to be the main difference. More recently, Zhou et al.[16] reported that the isolated tetra-coordinated boron species with a possible -B[OH···O(H)-Si]2 structure in the B-MFI zeolite framework were stable in the ODHP reaction conditions and can prevent the leaching of boron. Therefore, B-MFI displayed high activity, olefin selectivity, and durability for ODHP. The specific zeolite structure is considered to be the key factor to prevent the full hydrolysis of boron species, but the olefins’ productivity on B-MFI is still significantly lower than those of supported boron oxide catalysts.

Unlike microporous zeolites, the pore walls of ordered mesoporous silica are amorphous, making them easy to accept various heteroatom groups with excellent catalytic properties.[17−20] For example, vanadium-substituted MCM-41,[19] MCM-48,[19] and KIT-6[20] catalysts showed much higher selectivity to propylene in ODHP than that of the corresponding mesoporous silica-supported vanadium catalysts due to higher dispersed catalytic active sites in larger surface areas. However, boron-substituted mesoporous silica materials, such as B-MCM-41,[21−28] B-MCM-48,[29] and B-SBA-15,[30] as the catalysts for ODHP have never been reported yet, although they have been synthesized by several research groups.[21−30] Indeed, deboronation or structural stability of boron-substituted mesoporous silica is still the main problem to be overcome.

Herein, we focus on the preparation of B-MCM-41 as a durable catalyst for ODHP using a facile hydrothermal method. The boron content was varied to explore the effect of boron substitution on the catalytic performance and on the water tolerance of the catalysts during the ODHP reaction or after strict washing with water. The as-synthesized B-MCM-41 with a B/Si molar ratio of 0.048 (i.e. Si/B = ∼21) not only has high catalytic activity for ODHP but also shows a remarkable water tolerance. Based on catalyst characterization and theoretical calculations of the 11B isotropic shift calculated with density functional theory (DFT) methods,[31−33] the tri-coordinated B–OH species incorporated into the MCM-41 framework are considered to be the active sites and the high structural stability of B-MCM-41 makes it water tolerant.

Results and Discussion

Catalyst Characterization

The catalysts obtained are denoted as BM-x, where B stands for boron, M for MCM-41, and x = 0, 0.1, 0.2, 1.0, 2.0, or 4.0, according to the initial molar ratio of B/Si in the gel. The washed and recalcined BM-x samples are labeled BM-xW. The small-angle XRD patterns of the BM-x catalysts prepared with various molar ratios of B/Si are presented in Figure . The spectrum of BM-0 exhibits three diffraction peaks that are consistent with (100), (110), and (200) planes, respectively, indicating that a typical mesoporous MCM-41 structure was obtained.[34] The main interplanar space d100 is 3.50 nm, corresponding to a lattice parameter of a0 = 4.05 nm (Table ) if a hexagonal pore symmetry is employed. An increase in the initial molar ratio of B/Si to 0.1 results in the decrease of d100 to 3.39 nm, suggesting the presence of B in the mesoporous framework, because the B3+–O bond distance is smaller than that of Si4+–O.[35] The ordered mesoporous structure can still be formed in BM-1.0, although the intensity of the diffraction peaks decreased. However, the peaks are not well developed for BM-2.0 and almost completely disappeared for BM-4.0, indicating a poor mesoporous phase of these catalysts with excessive substitution of B to Si. In addition, a broad peak in the 2θ range of 15–30° originating from amorphous SiO2 was shown in the wide-angle XRD patterns of the above samples, and the lattice planes of B2O3 (JCPDS: 00-013-0570) were not found in all cases (Figure S1).

Figure 1

Small-angle XRD patterns of samples.

Table 1

Physicochemical Properties of Samples

sample	BET surface area (m2/g)	pore volume (cm3/g)	pore diameter (nm)	d100 (nm)a	a0 (nm)b	boron content (wt %)c	molar ratio of B/Si
BM-0	952.2	0.92	3.12	3.50	4.05	0	0
BM-0.1	982.9	0.95	3.14	3.39	3.91	0.28	0.015
BM-0.2	1026.7	0.95	2.90	3.47	4.00	0.38	0.021
BM-1.0	929.2	0.86	2.86	3.42	3.95	0.86 (0.57)	0.048
BM-2.0	817.9	0.97	2.82			1.09 (0.82)	0.062
BM-4.0	410.7	0.71	2.67			2.49 (1.47)	0.147

Calculated using the Bragg equation d100 = λ / 2 sin θ.

Calculated using the equation a0 = 2d100 /.

Determined by ICP-MS, the data in parentheses were determined after washing tests.

The N2 physisorption isotherms and the corresponding BJH pore size distributions are shown in Figures and S2. According to the IUPAC classification,[36] the isotherm of BM-0 belongs to type IV(a) shape for the mesoporous material, exhibiting an obvious capillary condensation at p/p0 of 0.30–0.45 accompanied by type H1 hysteresis loop. Correspondingly, a narrow pore size distribution with an average of ca. 3 nm is fitted by the BJH model. Then, the isotherm gradually changes to a reversible type IV(b) shape, when the initial molar ratio of B/Si increases to 1.0, indicating that the long-range ordering is increased and the uniform mesopore diameters are reduced in BM-0.2 and BM-1.0 (see Table ).[36] It means that the appropriate incorporation of B can improve the pore regularity of MCM-41, but the capillary condensation phenomenon weakens and the type H4 hysteresis loop appears for BM-2.0 and BM-4.0, illustrating that the mesoporous structures have partially collapsed, which is consistent with the above XRD results. Generally, the specific surface areas increase first and then decrease with the increase of B content. All BM-x catalysts (x = 0.1–1.0) with ordered mesoporous structures show high specific surface areas above 900 m2/g and a pore volume more than 0.85 cm3/g. In addition, the larger pore volume in BM-2.0 would be attributed to a large number of structural defects, based on its broad type H4 hysteresis loop.

Figure 2

N2 physisorption isotherms of samples.

Figure shows the SEM images of some representative samples. As seen, BM-0 is composed of irregular block nanoparticles, while the particles of BM-0.1 display a nanorod morphology dominantly with 720–990 nm in length and 450–570 nm in width, which is present even when the B/Si initial molar ratio increases to 1.0. However, the content of B continued to increase in the synthesis gel, leading to a rougher surface due to amorphous particles being generated on the surface after calcining in air.[27] These variations are in good agreement with the results observed by small-angle XRD and N2 physisorption. For further confirmation of structural regularity, TEM measurements of the typical samples were carried out, as shown in Figure . The TEM image of BM-0 presents well-defined parallel straight mesoporous channels with a pore size of ca. 3 nm. The unidirectional and hexagonal mesopore arrangement consistent with MCM-41 type materials is observed along with a frontal view in the image of BM-1.0.

Figure 3

SEM images of samples: (a) BM-0, (b) BM-0.1, (c) BM-1.0, and (d) BM-2.0.

Figure 4

TEM images of samples: (a) BM-0 and (b) BM-1.0.

As summarized in Table , the final B contents in the BM-x catalysts are significantly lower than those in the synthesis gels due to the distinct hydrolysis of B during the thorough washing process of the precursors. In addition, it was reported that a fraction of the framework tetra-coordinated boron can be converted to tri-coordinated species leading to deboronation during the calcination procedure in air.[24] Therefore, the precursors of B-MCM-41 are often calcined under flowing dry nitrogen to avoid deboronation.[24,25] However, considering the reaction conditions of ODHP, the precursors in the present work were still calcined in air. To detect the stability of B in the synthesized catalysts, the washing experiment was carried out. It is important to mention that the B content after washing is ca. 70% of that before washing for BM-1.0 and BM-2.0, but ca. 59% for BM-4.0. The specific surface area and pore volume are slightly reduced (Figures S3–S8). For instance, the specific surface area and pore volume of BM-1.0W are only ca. 10% lower than those of BM-1.0 (Figures S3 and S4). There are no obvious changes in the morphology of BM-1.0 after washing (Figure S9). Thus, it can be speculated that the remaining B species are strongly incorporated into the mesoporous framework, and a stable and water-tolerant B-substituted MCM-41 catalyst (BM-1.0) was formed. Moreover, the deboronation ratio is significantly lower than that in the literature,[35,37] which may be attributed to ammonia and ammonium pentaborate that were selected as the sources of hydroxide and boron, respectively, because it is proven that ammonium ions are beneficial to enhance the long-range order of MCM-41.[26]

Catalytic Performance

Figure shows the temperature dependence of propane conversion and propylene selectivity over the BM-x catalysts. Figure S10 shows the corresponding data of quartz sand for comparison. Table S1 lists the detailed data. As seen, no catalytic activity was found on the pure MCM-41 (BM-0) in the temperature range of 450 to 550 °C, and the propane conversion remained at ca. 3% at 600 °C due to the homolysis of propane at high temperatures,[38] which showed an identical trend on quartz sand (Figure S10). As shown in Figure a, a series of B-MCM-41 catalysts (BM-x) were active during ODHP, on which all propane conversions were increased significantly with the increase of the reaction temperature, especially after the B/Si initial molar ratio reached 1.0. Meanwhile, the propane conversion at a certain reaction temperature was improved directly by initially increasing the B content but reached a maximum value on BM-2.0. In other words, BM-4.0 showed a very similar trend to BM-2.0. For instance, 26.2, 42.5, and 39.9% of the maximum propane conversions were obtained at 600 °C on BM-1.0, BM-2.0, and BM-4.0, respectively (Table S1). However, the catalytic specific activity per mole of B (molC3H8·molB–1·h–1, see Table S2) decreased from 40.8 (BM-2.0) to 16.8 (BM-4.0), indicating that the boron oxide species incorporated into the MCM-41 framework can be responsible for the catalytic activity. This will be further confirmed by the water tolerance experiment.

Figure 5

Influence of temperature on the (a) propane conversion and (b) propylene selectivity. Reaction conditions: 100 mg of catalyst; GHSV = 28.8 L/(g·h). Feed gas: CH/O/N = 1/1/10.

As shown in Figure b and Table S1, an earlier temperature onset (from 525 to 450 °C) of olefin selectivity was realized by increasing the B content from 0.28 wt % (BM-0.1) to 1.09 wt % (BM-2.0); however, it did not continue to decrease with a further increase of B content to 2.49 wt % (BM-4.0). This is consistent with the variation of propane conversion with temperature, suggesting that the framework B sites should be the initiators of propane selective oxidation. At the same time, the initial selectivity to propylene was increased from 31.4% (at 525 °C on BM-0.1) to 61.4% (at 450 °C on BM-4.0). Olefin selectivity had a similar trend with the reaction temperature on BM-1.0, 2.0, and 4.0, and propylene selectivity showed a gradual decrease from ca. 50–60 to 40%, and ethylene selectivity increased slightly from ca. 20 to 30% with the increase of the reaction temperature and was attributed to the cleavage of the C–C bond in propylene.[16] Moreover, Figure S11 shows the variation of selectivity to propylene in ODHP as a function of propane conversion over BM-1.0 and BM-2.0 at 525 °C. As seen, the two catalysts had similar selectivity to propylene, which can be maintained at a different gaseous hourly space velocity (GHSV). Figure compares the product distributions at isoconversion (ca. 10% propane conversion) over BM-1.0 and BM-2.0. Similar product distributions are shown. The overall selectivity to olefins at 500 °C is ca. 80%, indicating similarity in the catalytic active centers and the discrepancy only in the number. In addition, the olefin selectivity of ca. 80% as the propane conversion of ca. 15%, i.e., olefin productivity of 0.40 golefins·gcat–1·h–1was gained on BM-1.0 at 550 °C, which is ca. 8 times higher than that of microporous B-MFI (0.05 golefins·gcat–1·h–1)[16] under similar reaction conditions and product distributions. The lower apparent activation energies than those of h-BN,[16] B/SiO2,[16] and B-MWW[15] were also obtained on BM-1.0 and BM-2.0 (see Figure S12 and Table S3).

Figure 6

Product distribution comparison at isoconversion (ca. 10% propane conversion) for BM-1.0 (reaction conditions: 150 mg of catalyst at 500 °C; GHSV = 14.4 L/(g·h)) and BM-2.0 (reaction conditions: 100 mg of catalyst; GHSV = 28.8 L/(g·h)) at 500 °C. Feed gas: CH/O/N = 1/1/10.

For a better understanding of the catalytic performance of the B-MCM-41 catalysts, their water tolerance tests were carried out, and comparisons of catalytic performance before and after washing are compared in Figure . Surprisingly, the propane conversion and olefin selectivity on BM-1.0W, which had an olefin productivity of 0.36 golefins·gcat–1·h–1 at 550 °C, were almost the same as that on BM-1.0. By contrast, the BM-2.0W had a ca. 3.5–1.5 times lower propane conversion from 475 to 600 °C compared to BM-2.0 and a reduction of initial propylene selectivity from 56.2 to 42.8% at 475 °C, although the B content of both catalysts after washing was decreased by ca. 30%. For BM-4.0W, despite the B content being still the highest, the propane conversion decreased more significantly than that before washing, accompanied by the loss of propylene selectivity at a low temperature. Considering the above differences in the catalyst structure, BM-1.0 has a more regular mesoporous structure (Figure and Table ). It is confirmed that the water-tolerant framework boron oxide species on the surface of the mesoporous MCM-41 structure are the effective active sites for the ODHP reaction and the ordered mesoporous B-MCM-41 framework can expose more active boron oxide sites on the surface of the pore channel leading to the maintenance of catalytic activity after washing.

Figure 7

Catalytic performance of (a) BM-1.0 and BM-1.0W, (b) BM-2.0 and BM-2.0W, and (c) BM-4.0 and BM-4.0W. The red and black lines represent fresh and washed catalysts, respectively. Reaction conditions: 100 mg of catalyst; GHSV = 28.8 L/(g·h). Feed gas: CH/O/N = 1/1/10.

To evaluate the catalytic stability of the BM-1.0 catalyst, a continuous ODHP test at 550 °C was performed, as shown in Figure . After reaction for 50 h, BM-1.0 exhibited a stable propane conversion (ca. 15%) and olefin selectivity (ca. 50% of propylene and ca. 30% of ethylene), which were almost unchanged compared with the initial stage of the reaction. This not only shows that BM-1.0 has excellent catalytic stability for ODHP but also indicates that there is no obvious induction period in the catalytic process. The perceptible leaching of boron did not occur during the reaction, as confirmed by ICP-MS analysis (0.86 vs. 0.80 wt % before and after catalysis). Moreover, the nanorod morphology and the ordered mesoporous framework structure of the spent catalyst (denoted as BM-1.0S) did not change obviously as shown in the SEM and TEM images (Figures S13 and S14), though the specific surface area, pore volume, and pore diameter were somewhat smaller than those of BM-1.0 (Figures S15 and S16). The above results show that the ordered mesoporous structure of B-MCM-41 (BM-1.0) can prevent the hydrolysis and/or dissolution of boron oxide species effectively during the ODHP reaction. Therefore, B-MCM-41 (BM-1.0) can act as a durable ODHP catalyst.

Figure 8

Stability test of BM-1.0 during a 50 h operation at 550 °C. Reaction conditions: 100 mg of catalyst; GHSV = 28.8 L/(g·h). Feed gas: C3H8/O2/N2 = 1/1/10.

Origin of Catalytic Performance of B-MCM-41

To gain insight into the original details of the activation, XPS and ssMAS NMR of the B-MCM-41 catalysts were performed. Figure S17 shows the presence of Si, O, B, and C (surface contamination and residues) elements in B-MCM-41 catalysts, and other metals or nonmetallic impurities were undetectable, which is consistent with the results of ICP-MS. The Si 2p peaks at ca. 103.7 eV (Figure S18) in all cases were identified to be Si–O in amorphous SiO2, which is in accord with the results for MCM-41,[39,40] indicating that a small amount of B incorporation hardly affected the coordination environment of Si. The O 1s signal (Figure S19) at ca. 533.1 eV is the superposition assigned to the signals of B–O and Si–O.[41,42]Figure S20 shows that the core-level B 1s spectra deconvoluted into analogous single peaks at a binding energy of ca. 193.6 eV, which were assigned to the B(III) oxidation state in tri-coordinated BO3 species, and the tetra-coordinated BO4 species were not detected.[42−44] In addition, the intensity of the B 1s peak was increased gradually with the incorporation of boron. Correspondingly, the surface atomic ratio of B/Si was increased from 0.04 in BM-0.1 to 0.18 in BM-4.0 (Table S4). It is worth noting that the ratios of BM-0.1, BM-0.2, BM-1.0, and BM-2.0 are around twice those of the bulk molar ratios of B/Si conducted by ICP-MS (Table ) but the two ratios are similar in BM-4.0. Therefore, it can be considered that the surface tri-coordinated BO3 species were responsible for ODHP and the mesoporous structure was favorable for the exposure of active BO3 species on the surface of channels, after comparison with the structural characterization and catalytic performance of the catalysts (Figures –7). Moreover, the B 1s signal and surface molar ratio of B/Si had no obvious changes after catalysis (Figure and Table S4), suggesting that the surface BO3 species was stably retained during the ODHP reaction, and both Si 2p and O 1s spectra also were not significantly modified before and after the reaction(Figures S18 and S19), which are consistent with the remaining ordered mesoporous framework, as shown in SEM and TEM images of BM-1.0S (Figures S13 and S14).

Figure 9

Core-level XPS spectra of B 1s for B-MCM-41 before and after catalysis.

The variations of the local atomic environments before and after washing or catalysis, selecting BM-1.0 as the representative, are discussed based on the ssMAS NMR spectra (Figures , 11, and S21). As shown in Figure , a broad line shape ranging of the isotropic shift ranging from 1 to 18 ppm in the 1D 11B MAS NMR spectrum of BM-1.0 is apparent, which is attributed to trigonal-planar BO3 in the mesoporous silica,[24,45] but the tetrahedral BO4 (narrow peak at 0 to −4 ppm)[46] and extra framework BO3 or B(OH)3 (18–20 ppm)[47,48] are not detected. Similar features are observed in the spectrum of BM-1.0W. These are consistent with the aforementioned XPS results but are different from microporous borosilicate zeolite B-MWW or B-MFI.[14−16] It can be regarded as the reason for the good water tolerance of BM-1.0, leading to the maintenance of the ODHP catalytic activity when BM-1.0 was strictly washed and recalcined. However, the significant broadening range of the trigonal framework B sites is not accurately simulated solely on the 1D 11B MAS NMR spectrum due not only to the quadrupolar B nucleus with a spin of 3/2[48] but also the coexistence of multiple types of trigonal framework B coordinated with both O and OH.[11,24] Therefore, based on the consistent results between the calculations (Scheme S1) and the relevant kinds of the literature,[11,12,14,42,45,49] the trigonal framework B sites can be qualitatively ascribed to three types of B species as follows: B(OSi)(OH)2 (denoted as B1), B(OSi)(OB)2-(OH) (n = 1 or 2, denoted as B2), and B(OSi)(OB)3- (n = 1, 2, or 3, denoted as B3) with isotropic shifts (the peak onset of the left side) at ca. 15.0–18, 13.5–15.0, and 10.0–13.5 ppm, respectively. It is obvious that the B1 signal is increased and other spectral features are reserved after catalysis, suggesting that the reversibility of coordination transformation of trigonal framework B sites during the ODHP reaction and the formation of B–OH should be related to the propane dehydrogenation process. Thus, the induction period in the feed gas is also not necessary on the originally existing B–OH active sites.[5] The corresponding 1H MAS NMR spectra (Figure ) will provide supporting evidence.

Figure 10

11B MAS NMR spectra of B-MCM-41.

Figure 11

1H MAS NMR spectra of samples: (a) BM-1.0, (b) BM-1.0W, and (c) BM-1.0S.

As shown in Figure a, the spectral simulation of BM-1.0 reveals the separation of four H sites at ca. 4.4, 3.7, 3.1, and 2.6 ppm. The broadening line shape of the peak at ca. 4.4 ppm originates from the presence of adsorbed water[50−52] because the samples did not undergo a dehydrated process before ssMAS NMR measurements. The peaks at ca. 3.7 ppm (denoted as BOH) and ca. 3.1 ppm (denoted as BOH) with peak areas proportion to 31.2 and 68.8% (Table S5) are associated with the chemical distributions of B–OH groups originating from B1 and B2 species, respectively.[11,16,51,53] Note that the proportion of BOH of BM-1.0W is decreased to 23.2%, indicating that B1 sites are more susceptible to deboronation than B2 sites during washing. However, it is eminently increased to 47.2% after catalysis. These results correspond to the aforementioned changes of B1 species in the 11B MAS NMR spectra and then confirm undoubtedly the transformation of B1 and B2 species during the ODHP reaction. In addition, the unchanged resonance at ca. 2.6 ppm reveals Si–OH adjacent to the boron species.[16,52] The appearance of peaks at ca. 1.9, 1.5 ppm after catalysis (Figure c) can be defined as the H2O being accessible to Si–OH.[50,51]

29Si MAS NMR spectra of BM-1.0 and BM-1.0W show analogous broad asymmetric signals ranging from −85 to −120 ppm (Figure S21), which can be divided into two peaks with chemical shifts at ca. −103 and −110 ppm for SiO3(OH) (Q3) and SiO4 (Q4) groups, respectively.[24] The percentage of Q3 is slightly decreased (54.4 vs. 42.2% for BM-1.0 and BM-1.0W, Table S6) due to the unobvious deboronation during washing and recalcination, which is consistent with the abovementioned results. All of these observations confirm that the Si–O–B bonds in the B-MCM-41 framework are fairly stable whether in the washing or in the ODHP reaction, and the trigonal framework B–OH species are closely related to the catalytic activity of B-MCM-41. Unfortunately, we cannot determine the B–OH in B2 as isolated boron or oligomer boron clusters relying solely on the 11B MAS NMR spectra. However, it is most likely that the low incorporation of boron could lead to the highly dispersed isolated framework B–OH species.[14,16,49]

Conclusions

In summary, we successfully synthesized the water-tolerant B-MCM-41 with an ordered hexagonal-like mesoporous structure (B/Si ≤ 0.048, i.e., Si/B ≥ 21), employing ammonia and ammonium pentaborate as the sources of hydroxide and boron, respectively. B-MCM-41 (BM-1.0, B/Si = 0.048) as a representative showed excellent catalytic stability for ODHP, e.g., the propane conversion of ca. 15% and olefin selectivity of ca. 80% at 550 °C were almost unchanged after strict washing or long-time reaction. It was found that the mesoporous structure had remarkable stability and the leaching of boron did not occur on it during the reaction. The multiple trigonal-planar BO3 species incorporated into the mesoporous silica framework are demonstrated to be dominant in the structure, and the surface tri-coordinated B–OH species (B1 and B2) are the likely active centers for ODHP. This work demonstrates the favorable activity and endurance of the mesoporous B-MCM-41 for ODHP and provides a new approach for overcoming the deboronation of boron-based catalysts.

Experimental Section

Materials

Tetraethoxysilane (TEOS, SiO2 ≥ 25%), hexadecyl trimethyl ammonium bromide (CTAB, 99.0%), and ammonia aqueous solution (25–28 wt %) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China), Guangfu Fine Chemical Research Institute (Tianjin, China), and Fengchuan Chemical Reagent Technologies Co., Ltd. (Tianjin, China), respectively. Ammonium pentaborate octahydrate (99%) as the boron source was provided by Aladdin Bio-Chem Technology Co., Ltd. (Shanghai, China). All of the chemicals were used without further purification.

Catalyst Preparation

A series of mesoporous B-MCM-41 (BM-x) catalysts were synthesized by a modified hydrothermal method.[25,26] The molar composition of the gels was as follows: TEOS:68H2O:4NH4OH:0.1CTAB:x/5(NH4B5O8). The B/Si initial molar ratio (i.e., x) was varied from 0 to 4.0. Typically, 1.8 g of CTAB was dissolved in 2 mol/L ammonia at 35 °C. About 9 g of TEOS was added slowly to the above aqueous solution and stirred for 30 min. Then, a required amount of ammonium pentaborate octahydrate was gradually added. The slurry (pH = 11) was stirred vigorously for 6 h. The resulting gel was transferred into a Teflon-lined autoclave (100 mL) and hydrothermaly treated in an oven at 100 °C for 48 h. Next, the white solid sample was separated and washed thoroughly with deionized water until it reached a neutral pH. Then, it was dried at 80 °C overnight and then calcined at 600 °C in air for 6 h.

To test the water tolerance of B-MCM-41 catalysts, the BM-x sample (0.5 g) was washed with deionized water (30 mL) at room temperature for 3 h, then centrifugally separated, dried at 80 °C overnight, and calcined at 550 °C in air for 4 h.

Catalyst Characterization

N2 Physisorption

The specific surface area and pore size distribution of catalysts were determined by N2 physisorption at 77 K using a Micromeritics ASAP 2020 instrument. Before the test, a sample was degassed at 100 °C under vacuum for 5 h. The specific surface area was calculated from adsorption data using the Brunauer–Emmett–Teller (BET) method in the pressure range of p/p0 = 0.05–0.30; the pore diameter distribution and pore volume were obtained from the adsorption branch of the isotherm according to the Barrett–Joyner–Halenda (BJH) model.

X-ray Diffraction (XRD) Analysis

Small-angle powder XRD patterns were recorded on a Bruker D8 Advance diffractometer with Cu Kα radiation (λ = 0.15418 nm) with 2θ ranging from 0.5 to 10° and a scan speed of 1 °/min. Wide-angle powder XRD patterns were collected using a Japanese Rigaku Ultima IV diffractometer with Cu Kα radiation (λ = 0.15406 nm). The tube voltage was 40 kV and the current was 40 mA. The data sets were acquired in the step-scan mode over the range of 5°< 2θ < 80° with a scan speed of 8 °/min.

Inductively Coupled Plasma-Mass Spectrometer (ICP-MS) Analysis

The boron and silicon contents in the catalysts were determined using an Agilent Technologies 7900 ICP-MS. Before measurement, a certain amount of accurately weighed sample was dissolved in a mixture solution of HF and HCl, which was diluted with deionized water to a certain volume.

Electron Microscopy Analysis

Transmission electron microscopy (TEM) images were taken using a Talos 200A (FEI company) instrument operated at 200 kV. Scanning electron microscopy (SEM) images were recorded on an FEI Quanta 650 FEG with an accelerating voltage of 0.2–30 kV.

X-ray Photoelectron Spectroscopy (XPS) Analysis

XPS analysis was performed on a Thermo Escalab 250Xi spectrometer equipped with a monochromatized aluminum X-ray source powered at 150 W. The binding energy (BE) was corrected by referencing the C 1s peak at 284.8 eV.

Solid-State Magic-Angle-Spinning Magnetic Resonance (ssMAS NMR)

The ssMAS NMR measurements were performed on a JEOL JNM-ECZ600R spectrometer with a 14.09 T magnet. 11B MAS NMR spectra were acquired using 3.2 mm MAS NMR probes with a spinning rate of 20 kHz. The chemical shifts were referenced to a 1 mol/L H3BO3 aqueous solution at 19.6 ppm. 1H MAS NMR spectra were obtained using a 3.2 mm MAS NMR probe with a spinning rate of 15 kHz. The chemical shifts were referenced to tetramethylsilane. 29Si MAS NMR experiments were recorded using an 8 mm MAS probe with a spinning rate of 6 kHz.

Catalytic Activity Test

The catalytic activity of the BM-x catalysts for the ODHP reaction was tested under an atmospheric pressure of 90 kPa by passing the feed gas mixture of C3H8, O2, and N2 (volume ratio = 1:1:10, gaseous hourly space velocity (GHSV) = 14.4–57.6 L/(g·h)) through a fixed-bed quartz tube reactor (internal diameter = 8 mm) loaded with 100 or 150 mg of the catalyst. The reaction temperature was varied in the range from 450 to 600 °C at 25 °C intervals. A certain amount of quartz sand was filled in the reactor to reduce the dead volume to avoid any homogeneous reactions. It should be emphasized that the catalyst has not undergone any online pretreatment or activation. The composition of gas products was analyzed after 30 min of reaction at a certain temperature with an online GC-2010 Plus gas chromatograph equipped with a Micropacked ST column (2.0 m × 1.27 mm O.D. × 1.0 mm I.D., ShinCarbon ST 80/100 mesh, column temperature: from 50 to 270 °C) and a barrier discharge ionization detector (BID). The blank runs showed that there was negligible propane conversion without a catalyst. Propane conversion and product selectivity were calculated according to the equations given in Table S1. All of the catalytic tests performed close their carbon balances within the range of 100 ± 5%.

The stability of the catalyst (BM-1.0) was studied by continuously recording the catalytic performance for 50 h online under constant reaction conditions (T = 550 °C and GHSV = 28.8 L/(g·h)).

Computational Method

All of the calculations were completed with the Gaussian 09 program.[31] Geometries of all of the substances were optimized at the level of B3LYP/ 6–31G (d, p) and the vibration frequency analysis was at the same level of geometry optimization to verify optimized geometries as minima.[32] The ruler-independent atomic orbital (GIAO) method was used to calculate the isotropic shielding tensor of boron in various chemical environments.[33]