Oxidative dehydrogenation of propane on silica-supported vanadyl sites promoted with sodium metavanadate

Abstract

The promotion of silica-supported vanadyl species [VO₄]/SiO₂ (1) by α-NaVO₃ or β-NaVO₃ enhances the specific rate of the propene formation in oxidative dehydrogenation of propane (ODP) by, respectively, 30 and 125% at 450 °C and ca. 1 V nm⁻² nominal coverage. The increased rate of propene formation is offset only moderately by a decreased selectivity to propene, which declines by 10 and 15% relative to 1 (74%) in α-NaVO₃/1 and β-NaVO₃/1, at 5.8 and 8.2% propane conversion. The structural characterization of the promoted catalysts by Raman mapping, X-ray absorption near edge structure (XANES), transmission electron microscopy (TEM) and solid-state nuclear magnetic resonance (⁵¹V and ²³Na MAS NMR) allowed for associating the higher specific activity of β-NaVO₃/1 with a higher dispersion of vanadium sites on the silica support, while the agglomeration of these sites with the concomitant formation of a poorly dispersed Na_1+xV₃O₈ phase is related to a decreased catalytic activity. Surprisingly, solid-state ⁵¹V NMR and Raman spectroscopies reveal that the α-NaVO₃/1 and β-NaVO₃/1 catalysts contain the metastable β-NaVO₃ phase, explained by a more favorable interaction of Na_1+xV₃O₈/SiO₂, formed after calcination in both materials, with β-NaVO₃ than with α-NaVO₃.

---

Introduction

Propene, a fundamental building block in the production of bulk chemicals and polymers,¹ is typically obtained as a byproduct from fluid catalytic cracking (FCC) and steam cracking of naphtha.² However, the ongoing replacement of naphtha by shale gas³ decreases the propene production through this route, which occurs simultaneously with a growing demand for propene.^3,4 Industrial, “on-purpose” propene production technologies via propane dehydrogenation (PDH) rely currently on CrO_x/Al₂O₃ or Pt–Sn/Al₂O₃, both catalysts promoted with Na/K (Catofin and Oleflex processes, respectively).⁵ These technologies have drawbacks, including coking, high energy demand (), low conversions (at the thermodynamic equilibrium, 25% at 527 °C), high price of Pt, and toxicity of Cr^VI.^2,5–7 An alternative to PDH is the oxidative dehydrogenation of propane (ODP) that exothermally converts propane and oxygen to propene and water () at lower temperatures (ca. 450 °C). Propane conversion is not limited by thermodynamics in ODP and coking is avoided due to the use of oxygen, thus providing usually a stable catalytic performance. Despite its potential, no ODP process has yet been industrialized, primarily because of the insufficient selectivity to propene at high propane conversions.⁶

At low vanadium loadings (such as those used in this work), dehydrated vanadia on oxide supports features mostly site-isolated, surface-grafted tripodal vanadium oxo sites, (–O)₃V=O, often denoted [VO₄] sites.^8–20 These species are among the best-performing ODP catalysts.^2,6,8,21,22 Alkali dopants were reported to improve propene selectivity of supported VO_x catalysts, which was, however, associated with a lower catalytic activity than in the undoped catalysts.^23,24 It was argued that the basic alkali doping decreased the strong acidity of the undoped catalysts,^23–25 in addition to weakening the V=O bond.^24,26 Furthermore, Na doping can improve the dispersion of VO_x sites on a silica support by increasing the reactivity of surface OH groups of the silica support.^26–28

Recently, we have shown that a catalyst derived from silica supported sodium decavanadate (Na₆V₁₀O₂₈) provides 65% selectivity to propene at 6% propane conversion at 450 °C.²⁹ When heated under air to 600 °C, Na₆V₁₀O₂₈ decomposes on the silica surface to the metastable β-NaVO₃ phase along with a Na_1+xV₃O₈ phase interacting with the silica support (Na_1+xV₃O₈/SiO₂). The formation of β-NaVO₃ in the calcined Na₆V₁₀O₂₈/SiO₂ material under these conditions was surprising because the transformation of bulk and silica-supported β-NaVO₃ to α-NaVO₃ proceeds at notably lower temperatures than 600 °C and therefore an α-NaVO₃ phase would have been expected.^30,31 The presence of β-NaVO₃ in the calcined Na₆V₁₀O₂₈/SiO₂ suggests that Na_1+xV₃O₈/SiO₂ plays a role in stabilizing β-NaVO₃.²⁹

With the objective of improving our understanding of the interaction between the NaVO₃ phases and the vanadyl sites on the silica surface, we incipient wetness impregnated (IWI) an aqueous solution of either α-NaVO₃ or β-NaVO₃ onto a SiO₂ support, followed by overnight drying at 100 °C and an IWI of an aqueous solution of NH₄VO₃. This procedure gave, after calcination, α-NaVO₃/1 and β-NaVO₃/1 materials with a similar nominal vanadium loading of ca. 1 V nm⁻² and a Na/V ratio of ca. 0.6. We find that the phase of the NaVO₃ promoter used for the impregnation influences the increase of the initial specific rate for propene formation of the reference catalyst 1, i.e. an increase by 30 and 125% is observed, respectively, for α-NaVO₃/1 and β-NaVO₃/1. Interestingly, solid state ⁵¹V NMR and Raman spectroscopy suggest that α-NaVO₃/1 and β-NaVO₃/1 contain the metastable β-NaVO₃ phase. Yet the catalytic activity of α-NaVO₃/1 and β-NaVO₃/1 and their deactivation with time on stream (TOS) are different, with the rate of propene formation decreasing after 4 h by 12 and 21%, respectively. Agglomeration of Na and V species in the used catalyst was identified as the driving force for the deactivation. We explain the higher activity of β-NaVO₃/1 compared to α-NaVO₃/1 by the higher dispersion of β-NaVO₃-Na_1+xV₃O₈/SiO₂ species in β-NaVO₃/1 (Scheme 1).^6,20,32

Scheme 1 Silica-supported materials and vanadium species prepared in this work. (IWI stands for incipient wetness impregnation).

Results and discussion

Incipient wetness impregnation of ammonium metavanadate^6,8 was used to prepare [VO₄]/SiO₂ (1, 2.1 wt% V by ICP, Table S1†) containing ca. 1 V nm⁻², which is below the monolayer coverage for SiO₂ support.¹⁷ Vanadyl sites on silica have been characterized in details in previous reports.^{13,14,17,28,29,33–36} Since the nature of the supported VO_x species changes with hydroxylation of the support,^17,32,33 the materials discussed below were treated under synthetic air (500 °C, 1 h, 30 ml min⁻¹) and stored in a glovebox (H₂O and O₂ < 0.5 ppm), indicated by the respective subscript notation, for instance 1_(500-air). To prepare α-NaVO₃/1 and β-NaVO₃/1 materials, silica was impregnated first with aqueous solutions of α-NaVO₃ or β-NaVO₃ (0.6 V nm⁻²) followed by overnight drying at 100 °C and a subsequent IWI of NH₄VO₃ (0.4 V nm⁻²). Calcined α-NaVO₃/1 and β-NaVO₃/1 contained 1.9 and 2.0 wt% of vanadium and a molar ratio of Na/V of 0.63 and 0.62 (by ICP), respectively, corresponding to a nominal silica coverage of ca. 1 V nm⁻². Although the nominal vanadium loading of the as impregnated α- and β-NaVO₃ promoted catalysts were similar to that of the benchmark catalyst (1 V nm⁻², ca. 2 wt%), the surface density of V calculated from the ICP-determined V loading and the specific surface area of the material (according to BET N₂ physisorption measurements) was notably higher for the promoted catalysts compared to 1 (i.e. 1.6 and 2.0 V nm⁻²vs. 1.1 V nm⁻², see Table S1,† entries 1–4). This is explained by a reduced surface area of the silica support due to the etching effect caused by Na-containing precursors.²⁵ We have therefore optimized the loading of vanadium precursors in order to obtain a comparable surface density of V in the promoted catalysts and in 1. This was achieved for α- and β-NaVO₃ promoted catalysts with a lower nominal vanadium loading (i.e. 0.7 V nm⁻², denoted in a subscript), resulting in 1.2 and 1.0 V nm⁻², respectively, after the calcination (Table S1,† entries 5–6). Lastly, note that the dissolution of β-NaVO₃ in water gives dihydrate species, NaVO₃·(2H₂O) that transform, above ca. 34 °C, to β-NaVO₃.^31,37 The irreversible transformation of β-NaVO₃ to α-NaVO₃ was reported to occur at 403–405 °C.³⁰

The Raman spectrum of 1_(500-air) features a characteristic sharp peak at 1039 cm⁻¹ owing to the vanadium oxo stretching vibration^29,33,38,39 that is significantly reduced in intensity in β-NaVO₃/1_(500-air) and α-NaVO₃/1_(500-air) (Fig. 1a and S1†). The latter materials contain also bands of β-NaVO₃ and α-NaVO₃ at 945 and 954 cm⁻¹, respectively, but with diverging intensities. The characteristic Raman band of β-NaVO₃ is minor and the band of α-NaVO₃ is major in β-NaVO₃/1_(500-air). In α-NaVO₃/1_(500-air), the band of β-NaVO₃ is more intense than the band of α-NaVO₃, although both bands are less intense than in β-NaVO₃/1_(500-air). In addition, two broad peaks at 804 and 737 cm⁻¹ that match the peak positions in the Na_1+xV₃O₈/SiO_2(500-air) reference are observed in β-NaVO₃/1_(500-air) and, to a larger extent, in α-NaVO₃/1_(500-air) (Fig. 1a).²⁹ In this Na_1+xV₃O₈/SiO_2(500-air) material, the Na_1+xV₃O₈ phase interacts with the SiO₂ support, as evidenced by Raman, ⁵¹V and ²³Na NMR data, although the exact nature of the formed sites is currently unclear.²⁹ The Raman peaks at 804 and 737 cm⁻¹ in α-NaVO₃/1_(500-air) are more intense in comparison to β-NaVO₃/1_(500-air), probably due to the lower dispersion and increased long-range order of the Na_1+xV₃O₈ phase in α-NaVO₃/1_(500-air); besides, the β-NaVO₃ peak has a lower intensity in β-NaVO₃/1_(500-air). Considering that α-NaVO₃/SiO_2(500-air) (i.e. the catalyst made by IWI of α-NaVO₃ on silica at a nominal vanadium loading of 1 V nm⁻²) does not feature peaks at 804 and 737 cm⁻¹,²⁹ these bands must have been formed owing to an interaction between [VO₄]/SiO₂ and α-/β-NaVO₃. These species may feature different degrees of dispersion and/or crystallinity, which leads to different intensities in Raman spectra (vide infra).^33,40

Fig. 1 (a) Raman, (b) ⁵¹V and (c) ²³Na MAS NMR spectra of the studied materials (see labels in the panels); (d) and (e) are linear combination fittings (LCF, see Table 1 for details) of the V K-edge XANES of α-NaVO₃/1 and β-NaVO₃/1 dehydroxylated at 500 °C as well as f) V K-edge XANES spectra of the reference materials. Subscript TOS in hours indicates a used catalyst that was cooled down to room temperature while flowing the ODP gas mixture and handled in pristine conditions. Side bands of the NMR spectra are marked by asterisks; a spinning rate of 15–18 kHz was used. Notations α, β, and α′ indicate α-NaVO₃, β-NaVO₃ and α′-NaV₂O₅ phases, respectively. Characterization of Na_1+xV₃O₈/SiO_2(500-air) was reported by us previously and is reproduced here for comparison purposes.²⁹

It is conceivable that the melting of α-NaVO₃ on the SiO₂ surface upon calcination and its subsequent recrystallization during cooling yielded the metastable β-NaVO₃ polymorph, owing to the more favorable interaction of this polymorph with the V-based, supported species. To test this hypothesis, we calcined α-NaVO₃/1 to ca. 600 °C in situ in a Raman cell (Linkam CCR1000) under flow of synthetic air (30 ml min⁻¹). By recording spectra from the various regions of the specimen heated to ca. 600 °C and then cooled down to room temperature, we observed an inhomogeneous distribution of vanadium species, which is possibly related to heat transfer gradients in the in situ Raman cell. Specifically, two distinct areas were found, viz. areas with peaks of α-NaVO₃ and isolated vanadyl sites, as well as areas containing predominantly peaks of β-NaVO₃ and Na_1+xV₃O₈/SiO₂ (beam spot size was ca. 1.6 μm, Fig. S2†). In a control experiment, calcination of α-NaVO₃/1 in a muffle furnace at 600 °C for 4 h with the subsequent exposure to air gave a more homogeneous material that predominately features peaks of β-NaVO₃ and Na_1+xV₃O₈; only occasionally areas with α-NaVO₃ and [VO₄] sites are found (Fig. S3†). These experiments suggest that α-NaVO₃ may react with [VO₄]/SiO₂ to give Na_1+xV₃O₈/SiO₂ and β-NaVO₃. This mechanism for the formation of metastable β-NaVO₃ does not necessarily require recrystallization of the molten NaVO₃. Indeed, a differential scanning calorimetry (DSC) experiment of the calcination of α-NaVO₃/1 reveals no clear features due to melting and recrystallization (Fig. S4†).

⁵¹V magic angle spinning (MAS) NMR spectra of β-NaVO₃, α-NaVO₃, and 1_(500-air) give signals at −515, −577, and −710 ppm, respectively. In the ²³Na NMR spectrum of β-NaVO₃, one peak is observed ca. −13 ppm while two signals centered at −5 and −20 ppm are observed for α-NaVO₃ (Fig. 1b and S5†).^11,29,36,41 In line with Raman spectroscopy results, peaks due to β- and α-NaVO₃ are observed in the ⁵¹V NMR spectra of α-NaVO₃/1_(500-air) and β-NaVO₃/1_(500-air); the signal from β-NaVO₃ is more intense for α-NaVO₃/1_(500-air). A broad feature at ca. −610 ppm is observed for both promoted catalysts and Na_1+xV₃O₈/SiO_2(500-air). At least in part, this broad feature may be due to the vanadyl sites interacting with a Na⁺ cation, which induces a downfield shift by 100 ppm compared to that in 1_(500-air) (Fig. 1b).^26,28,29 Note that a broad shoulder at the same position of ca. −610 ppm is also observed for α-NaVO₃/SiO_2(500-air), and it is likely due to a partial decomposition of silica-supported α-NaVO₃ to vanadyl sites interacting with the nearby sodium cations on surface siloxides ((–O)₃V=O⋯Na⁺, Fig. S6†). Consistent with our inferences from the Raman and ⁵¹V NMR data, ²³Na MAS NMR of the promoted catalysts shows peaks that can be ascribed to α-, β-NaVO₃, and Na_1+xV₃O₈/SiO₂ species (Fig. 1c). Interestingly, the feature due to β-NaVO₃ centered at −13 ppm is more prominent in α-NaVO₃/1_(500-air) while two features due to α-NaVO₃ (centered at −21 and −5 ppm) are more prominent in β-NaVO₃/1_(500-air) and are not noticeable in α-NaVO₃/1_(500-air). ²³Na MAS NMR spectra of Na_1+xV₃O₈/SiO_2(500-air) and α-NaVO₃/1_(500-air) are similar, with a ca. 5 ppm upfield shift of peaks in the latter material. This indicates that Na_1+xV₃O₈/SiO₂ is a major phase in α-NaVO₃/1_(500-air) and that this material has nearly no Na atoms in the environment of α-NaVO₃. Features of Na_1+xV₃O₈/SiO₂ are less prominent in β-NaVO₃/1_(500-air) (assessed by the peak at −28 ppm) and this is offset by more intense features of α- and β-NaVO₃.

The intensity of the pre-edge peak in V K-edge XANES depends on the symmetry of the ligand sphere around the vanadium atom such that a more centro-symmetric environment gives lower pre-edge peak heights in the order: tetrahedral (T_d) > distorted tetrahedral > square pyramidal (SP) > distorted octahedral (O_h) > octahedral (O_h, Fig. S7†).^17,29,42 The spectra of 1_(air-500) is consistent with V sites in a T_d coordination.²⁹ The V K-edge XANES spectra of Na_1+xV₃O₈/SiO₂ and 1 (exposed to air or dehydroxylated) are similar (Fig. S8 and S9†), indicating structural similarities of the V sites in those materials. To quantify the different V species present in the promoted materials we used linear combination fitting (LCF) of the V K-edge XANES spectra of the α- and β-NaVO₃/1_(500-air). In this analysis, we used the well-defined material 1_(air-500) as one of the references, as well as α-NaVO₃ and β-NaVO₃. LCF analysis yielded a slightly higher fraction of α-NaVO₃ than β-NaVO₃ in β-NaVO₃/1_(500-air) (24 and 20%, respectively, Table 1, entry 1) and a moderately higher fraction of β-NaVO₃ in α-NaVO₃/1_(500-air) in comparison to α-NaVO₃ (28 and 16%, respectively, Table 1, entry 2, and Fig. 1d–f). These obtained phase percentages are consistent with the Raman and MAS NMR observations described above. Similar values for 1_(500-air) were obtained for both promoted materials (56%).

Table 1 Linear combination fitting (LCF) results of the V K-edge XANES spectra of the promoted catalysts

Entry	Material	β-NaVO3	α-NaVO3	1 (500-air)
a Representing Na1+xV3O8/SiO2(500-air) (see Fig. S8 and S9†).
1	β-NaVO3/1(500-air)	20	24	56
2	α-NaVO3/1(500-air)	28	16	56
3	β-NaVO3/1(TOS-4h)	28	21	51

---

To investigate the dispersion of the α-NaVO₃, β-NaVO₃ and Na_1+xV₃O₈/SiO_2(500-air) phases in α-NaVO₃/1_(500-air) and β-NaVO₃/1_(500-air), Raman maps were collected. The freshly calcined materials were sealed in quartz capillaries under an inert atmosphere and Raman maps acquired from in total 225 points (15 × 15) separated by 4 μm (the laser spot size was ca. 1.6 μm). The intensities of the characteristic Raman peaks at 954, 945, 804 cm⁻¹ (±2 cm⁻¹) were used to map α-NaVO₃, β-NaVO₃, and Na_1+xV₃O₈/SiO₂, respectively (Fig. 2a and S10†). We observe that α-NaVO₃ is less uniformly dispersed than β-NaVO₃ in α-NaVO₃/1_(500-air) or β-NaVO₃/1_(500-air). In addition, Na_1+xV₃O₈/SiO₂ appears more abundant and less well dispersed in α-NaVO₃/1_(500-air) relative to β-NaVO₃/1_(500-air). Yet for both promoted materials, the distribution of the intensities of β-NaVO₃ and Na_1+xV₃O₈/SiO₂ in the Raman maps is relatively similar, i.e. these phases are found similarly dispersed, which suggests an interaction between these two phases. Furthermore, EDX mapping of β-NaVO₃/1 and α-NaVO₃/1 (exposed to air during the sample transfer) shows a more uniform dispersion of Na and V in β-NaVO₃/1 compared to α-NaVO₃/1 that shows agglomerates of a Na/V rich phase (Fig. 2b).

Fig. 2 (a) Raman mapping of α-NaVO₃/1_(500-air), β-NaVO₃/1_(500-air) and β-NaVO₃/1_(TOS-4h) cooled down in an ODP atmosphere; the spectra shown correspond to the areas marked with A, B, and C (bottom left panel). (b) EDX mapping of α- and β-NaVO₃/1 as well as β-NaVO₃/1_(TOS-4h) (exposed to air).

The reducibility of supported vanadium-based catalysts for the oxidative dehydrogenation of propane and methanol (with vanadium loading below the monolayer coverage) was previously correlated with the turn over frequency (TOF) of those catalysts, such that higher reducibility is typically associated with higher activity.^6,43,44 However, a counter example is crystalline V₂O₅ on Al₂O₃ promoted with molybdenum that showed higher conversions and selectivities in ODP with decreasing reducibility of vanadium, as assessed by the temperature corresponding to the maximum of H₂ consumption (T_max) in the H₂ temperature-programmed reduction (TPR) experiment.⁴⁵ This indicates that the activity and selectivity of the V-based catalysts for ODP does not depend solely on their reducibility^6,43 but is influenced by other factors, for instance, the interaction with the support,^20,43 dispersion of the active phase,^29,46,47 V–O binding energy,⁴⁵ and acid–base properties.⁴⁸

Considering that α-NaVO₃/1 and β-NaVO₃/1 contain similar species, but show distinct catalytic activity (vide infra), we were interested to compare H₂-TPR profiles of these two catalysts. A slightly lower T_max was observed for the more active catalyst, β-NaVO₃/1, as compared to α-NaVO₃/1 (555 and 580 °C, respectively, Fig. S11†), while T_max of the less active Na_1+xV₃O₈/SiO₂ and 1 catalysts were centered at 602 and 473 °C, consistent with the different nature of vanadium species in studied catalysts.

In summary, Raman spectroscopy, ⁵¹V and ²³Na MAS NMR data as well as LCF of XANES spectra show that α- and β-NaVO₃ as well as Na_1+xV₃O₈ interacting with the silica support are present in α-NaVO₃/1_(500-air) and β-NaVO₃/1_(500-air) materials, albeit in different relative amounts. The dispersion of Na and V species is higher in β-NaVO₃/1_(500-air) compared to α-NaVO₃/1_(500-air) according to Raman and EDX mapping. By NMR and Raman spectroscopies, a higher fraction of Na_1+xV₃O₈/SiO₂ and β-NaVO₃ is found in α-NaVO₃/1_(500-air) relative to β-NaVO₃/1_(500-air). The unexpected formation of the metastable β-NaVO₃ polymorph from the thermodynamically stable α-NaVO₃ is likely due to the stabilizing interaction between β-NaVO₃ and Na_1+xV₃O₈/SiO₂ species, as compared to the respective interaction with α-NaVO₃. This is supported by the fact that the calcination of β-NaVO₃ on silica without vanadyl sites leads to α-NaVO₃ (Fig. S12†).²⁹ The materials Na_1+xV₃O₈/SiO₂ and α-NaVO₃/SiO_2(500-air) likely contain structurally similar (–O)₃V=O⋯Na⁺ sites, as suggested by the characteristic broad feature in the respective ⁵¹V MAS NMR spectra at −610 ppm (Fig. 1b and S6a†). However, Raman bands at 737 and 804 cm⁻¹ due to V–O–V bonds, diagnostic for Na_1+xV₃O₈/SiO_2(500-air), are not observed for α-NaVO₃/SiO_2(500-air).^17,29 Studies to refine our understanding of the nature of sites in Na_1+xV₃O₈/SiO₂ go beyond the scope of this work.

When the weight loadings of vanadium in the promoted catalysts were similar to that of the benchmark catalyst 1 (ca. 2 wt%, Na/V = 0.6), a higher initial specific activity was obtained for α- and β-NaVO₃/1 catalysts (3.4 and 5.7 mmol C₃H₆ mol V⁻¹ s⁻¹) than for 1 (2.5 mmol C₃H₆ mol V⁻¹ s⁻¹), albeit the propene selectivities of α-NaVO₃/1 and β-NaVO₃/1 (64 and 59%) were lower than of 1 (74%, Fig. 3a). Yet α-NaVO₃/1 and β-NaVO₃/1 deactivate with TOS, after 4 h by 12 and 21%, respectively. However, the specific activity of α-NaVO₃/1_(0.7V) and β-NaVO₃/1_(0.7V), i.e. materials with a similar vanadium surface density to that of 1 (ca. 1.1 V nm⁻² obtained at ca. Na/V ratio of 0.3, Table S1†) were 2.5 and 3.5 mmol C₃H₆ mol V⁻¹ s⁻¹, respectively, while a similar selectivity of 74% was observed for all three these catalysts. Noteworthy, α-NaVO₃/1_(0.7V) and β-NaVO₃/1_(0.7V) did not deactivate with TOS after 240 min (at ca. 2.9 and 4.0% conversion, respectively, Fig. S14†). These results demonstrate that the surface density of Na and V influence catalyst activity, selectivity and stability.

Fig. 3 (a) Initial catalytic activity (TOS = 30 min), (b) propene selectivity and productivity vs. propane conversion for the studied catalysts (see Table S1†). WHSV was varied between 5.1–13.6 h⁻¹ by changing the total feed flow (15.8–42 ml min⁻¹). (c) Changes with TOS of β- and α-NaVO₃/1 under ODP conditions (C₃H₈: air = 2: 5, total flow of 21 ml min⁻¹, 450 °C).

By increasing the contact time, the conversion increases, yet the selectivity of the benchmark catalyst (1) drops with a higher rate compared to both promoted catalysts (Fig. 3b). For a nominal V loading of 1 V nm⁻², the benchmark catalyst 1 shows a higher selectivity to propene compared to the promoted catalysts at similar conversions that did not exceed 8% (Fig. 3b). Our promoted catalysts show higher propene selectivities at conversions exceeding 10%, i.e. 58, 51 and 41% at 13, 15 and 13% propane conversion for β-NaVO₃/1, α-NaVO₃/1 and 1, respectively (Fig. S13†). Overall, these values translate into higher initial productivities (within 30 min) for the promoted catalysts relative to 1.

Furthermore, the productivity of β-NaVO₃/1 is ca. 2.4 times higher than that of 1 at similar propane conversion (0.39 vs. 0.16 kg_C3H6 kg_cat⁻¹ h⁻¹ at 4.7 vs. 4.3%, respectively, Fig. 3b). Interestingly, Na_1+xV₃O₈/SiO₂ showed the highest initial selectivity to propene among the studied catalysts, reaching 80% at a 2.2% propane conversion, i.e. slightly higher than the sodium-free benchmark catalyst 1 (77% at 3.6% propane conversion at 450 °C). In addition to propene, Na_1+xV₃O₈/SiO₂ and 1 produced up to ca. 3% C₂H₄ while α-NaVO₃/1 and β-NaVO₃/1 only gave propene and CO_x, with an initial propene selectivity of 64 and 59% at 5.8 and 8.2% conversions, respectively. At similar reaction conditions (WHSV = 6.8 h⁻¹), the Na₆V₁₀O₂₈/SiO₂ catalyst reported by us previously²⁹ showed 65% propene selectivity at 6% propane conversion.

Notably, while the ODP activities of catalysts 1 and Na_1+xV₃O₈/SiO₂ are stable (Fig. S15,† typical for V-based ODP catalysts),²⁹ the catalytic activity of both promoted materials decreases with TOS (Fig. 3c). We observe a decline of the specific activity by 12 and 21% within 4 h TOS for α-NaVO₃/1 and β-NaVO₃/1, respectively. Comparison of the TEM images and EDX mapping of β-NaVO₃/1 after 4 h TOS to the fresh catalyst reveals agglomeration of the Na and V species on the silica surface after the ODP reaction, which is a likely reason for deactivation (Fig. 2b). The Raman spectrum of the reacted β-NaVO₃/1 (denoted β-NaVO₃/1_(TOS-4h); the material was handled without exposure to air) shows a tangible increase in the intensity of the peaks at 737 and 804 cm⁻¹ compared to the fresh catalyst. This can indicate the formation of a less dispersed Na_1+xV₃O₈ phase compared to the fresh catalyst and would be in line with the TEM analysis. The formation of three-dimensional V₂O₅ crystals has been reported to deactivate VO_x-based ODP catalysts.^6,20 Furthermore, the strongly reduced intensity of the 954 cm⁻¹ peak of α-NaVO₃ might be due to its reaction with [VO₄] sites and transformation to Na_1+xV₃O₈/SiO₂ with TOS that decrease the number of active sites (possibly, (–O)₃V=O⋯Na⁺ sites). That being said, no notable change was observed in the characteristic peak of β-NaVO₃ (945 cm⁻¹, Fig. 1a). In addition, a low-intensity peak at 968 cm⁻¹ due to α′-NaV₂O₅ is also observed in β-NaVO₃/1_(TOS-4h) (Fig. 1a). Note that we have previously shown that α′-NaV₂O₅ forms on the silica surface under inert conditions owing to the reaction between β-NaVO₃ and Na_1+xV₃O₈/SiO₂; α′-NaV₂O₅ is poorly active for ODP.²⁹ In agreement with the Raman data, ⁵¹V MAS NMR of β-NaVO₃/1_(TOS-4h) shows a decreased intensity of α-NaVO₃ signatures and an increased intensity of β-NaVO₃ that also broadens (Fig. 1b).⁴⁹ MAS NMR data on ²³Na nucleus shows a new peak centered at −48 ppm for β-NaVO₃/1_(TOS-4h), due to the formation of the α′-NaV₂O₅ phase. A decreased intensity of the α-NaVO₃ phase (peaks at −21 and −5 ppm, Fig. 1c) is also observed.

We discussed above that Raman mapping of β-NaVO₃/1_(500-air) shows inhomogeneous distribution of α-NaVO₃ in this material, while β-NaVO₃ and Na_1+xV₃O₈/SiO₂ are more homogeneously dispersed compared to α-NaVO₃ (Fig. 2a). After 4 h TOS, i.e. β-NaVO₃/1_(TOS-4h), Raman peaks of the α-NaVO₃ phase have decreased notably, while peaks of β-NaVO₃ and Na_1+xV₃O₈/SiO₂ phases have increased (Fig. 2a). As mentioned, this might be due to a reaction of the remaining vanadyl sites on silica with α-NaVO₃ forming crystalline (three-dimensional) Na_1+xV₃O₈ that is associated with a decreased catalytic activity. Raman maps in Fig. 2a also show that the increased intensity of signals related to Na_1+xV₃O₈/SiO₂ correlates with an increased intensity of the β-NaVO₃ peak, which is consistent with the aforementioned hypothesis of an increased stabilization of the metastable β-NaVO₃ by Na_1+xV₃O₈/SiO₂, in preference to the formation of α-NaVO₃, which is consumed with TOS in β-NaVO₃/1_(500-air) (Fig. 2a).

Conclusions

[VO₄]/SiO₂ promoted with α- or β-polymorphs of NaVO₃ shows an increase in the initial rate of propene formation for ODP by 30 and 125%, respectively, at similar vanadium loadings (ca. 1 V nm⁻² nominal coverage and 2 wt%), albeit offset by a 10 and 15% decrease in propene selectivity. Both catalysts lose activity with TOS due to the agglomeration and formation of less well dispersed (and possibly crystalline) Na_1+xV₃O₈. Calcination of vanadyl sites promoted by β-NaVO₃ leads to the α-NaVO₃ phase, while promotion of vanadyl sites with α-NaVO₃ gives notable amounts of the β-NaVO₃ phase, which is unexpected considering that β-NaVO₃ (both bulk and silica-supported) transforms completely to α-NaVO₃ already at ca. 400 °C. The stabilization of the metastable β-NaVO₃ on [VO₄]/SiO₂ is associated with the formation of dispersed Na_1+xV₃O₈ that interacts with the silica support. The interaction of [VO₄]/SiO₂ with β-NaVO₃ seems to lead to Na_1+xV₃O₈ with a higher dispersion on silica than when promoted by α-NaVO₃, resulting in a higher catalytic activity for the β-NaVO₃ promoted catalyst. We are currently exploring other ways (such as support effect) to maintain a high dispersion of Na_1+xV₃O₈.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We gratefully acknowledge the Scientific Centre for Optical and Electron Microscopy (ScopeM) of ETH Zürich for providing access to electron microscopes. The Paul Scherer Institute (PSI), SuperXAS beamline is acknowledged for providing access to the synchrotron X-ray facility.

Notes and references

1. J. S. Plotkin, The changing dynamics of olefin supply/demand, Catal. Today, 2005, 106, 10–14.
2. J. J. H. B. Sattler, J. Ruiz-Martinez, E. Santillan-Jimenez and B. M. Weckhuysen, Catalytic Dehydrogenation of Light Alkanes on Metals and Metal Oxides, Chem. Rev., 2014, 114, 10613–10653.
3. E. McFarland, Unconventional Chemistry for Unconventional Natural Gas, Science, 2012, 338, 340–342.
4. J. T. Grant, J. M. Venegas, W. P. McDermott and I. Hermans, Aerobic Oxidations of Light Alkanes over Solid Metal Oxide Catalysts, Chem. Rev., 2018, 118, 2769–2815.
5. M. M. Bhasin, J. H. McCain, B. V. Vora, T. Imai and P. R. Pujadó, Dehydrogenation and oxydehydrogenation of paraffins to olefins, Appl. Catal., A, 2001, 221, 397–419.
6. C. A. Carrero, R. Schloegl, I. E. Wachs and R. Schomaecker, Critical Literature Review of the Kinetics for the Oxidative Dehydrogenation of Propane over Well-Defined Supported Vanadium Oxide Catalysts, ACS Catal., 2014, 4, 3357–3380.
7. Y. Wang, H. Su, Y. Gu, X. Song and J. Zhao, Carcinogenicity of chromium and chemoprevention: a brief update, OncoTargets Ther., 2017, 10, 4065–4079.
8. R. R. Langeslay, D. M. Kaphan, C. L. Marshall, P. C. Stair, A. P. Sattelberger and M. Delferro, Catalytic Applications of Vanadium: A Mechanistic Perspective, Chem. Rev., 2019, 119, 2128–2191.
9. H. Eckert, G. Deo, I. E. Wachs and A. M. Hirt, Solid state ⁵¹V NMR structural studies of vanadium(V) oxide catalysts supported on TiO₂ (anatase) and TiO₂ (rutile). The influence of surface impurities on the vanadium(V) coordination, Colloids Surf., 1990, 45, 347–359.
10. O. B. Lapina, V. M. Mastikhin, L. G. Simonova and Y. O. Bulgakova, Characterization of surface species of supported V₂O₅-Al₂O₃ catalysts by ⁵¹V NMR, J. Mol. Catal., 1991, 69, 61–73.
11. N. Das, H. Eckert, H. Hu, I. E. Wachs, J. F. Walzer and F. J. Feher, Bonding states of surface vanadium(V) oxide phases on silica: structural characterization by vanadium-51 NMR and Raman spectroscopy, J. Phys. Chem., 1993, 97, 8240–8243.
12. N. R. Jaegers, J.-K. Lai, Y. He, E. Walter, D. A. Dixon, M. Vasiliu, Y. Chen, C. Wang, M. Y. Hu, K. T. Mueller, I. E. Wachs, Y. Wang and J. Z. Hu, Mechanism by which Tungsten Oxide Promotes the Activity of Supported V₂O₅/TiO₂ Catalysts for NO_X Abatement: Structural Effects Revealed by ⁵¹V MAS NMR Spectroscopy, Angew. Chem., Int. Ed., 2019, 58, 12609–12616.
13. N. R. Jaegers, C. Wan, M. Y. Hu, M. Vasiliu, D. A. Dixon, E. Walter, I. E. Wachs, Y. Wang and J. Z. Hu, Investigation of Silica-Supported Vanadium Oxide Catalysts by High-Field ⁵¹V Magic-Angle Spinning NMR, J. Phys. Chem. C, 2017, 121, 6246–6254.
14. T. Tanaka, H. Yamashita, R. Tsuchitani, T. Funabiki and S. Yoshida, X-ray absorption (EXAFS/XANES) study of supported vanadium oxide catalysts. Structure of surface vanadium oxide species on silica and γ-alumina at a low level of vanadium loading, J. Chem. Soc., Faraday Trans., 1988, 84, 2987–2999.
15. E. L. Lee and I. E. Wachs, In Situ Raman Spectroscopy of SiO₂-Supported Transition Metal Oxide Catalysts: An Isotopic ¹⁸O−¹⁶O Exchange Study, J. Phys. Chem. C, 2008, 112, 6487–6498.
16. S. T. Oyama, G. T. Went, K. B. Lewis, A. T. Bell and G. A. Somorjai, Oxygen chemisorption and laser Raman spectroscopy of unsupported and silica-supported vanadium oxide catalysts, J. Phys. Chem., 1989, 93, 6786–6790.
17. X. Gao, S. R. Bare, B. M. Weckhuysen and I. E. Wachs, In Situ Spectroscopic Investigation of Molecular Structures of Highly Dispersed Vanadium Oxide on Silica under Various Conditions, J. Phys. Chem. B, 1998, 102, 10842–10852.
18. X. Gao, M. A. Bañares and I. E. Wachs, Ethane and n-Butane Oxidation over Supported Vanadium Oxide Catalysts: An in Situ UV–Visible Diffuse Reflectance Spectroscopic Investigation, J. Catal., 1999, 188, 325–331.
19. X. Gao and I. E. Wachs, Investigation of Surface Structures of Supported Vanadium Oxide Catalysts by UV−vis−NIR Diffuse Reflectance Spectroscopy, J. Phys. Chem. B, 2000, 104, 1261–1268.
20. C. A. Carrero, C. J. Keturakis, A. Orrego, R. Schomacker and I. E. Wachs, Anomalous reactivity of supported V₂O₅ nanoparticles for propane oxidative dehydrogenation: influence of the vanadium oxide precursor, Dalton Trans., 2013, 42, 12644–12653.
21. C. Carrero, M. Kauer, A. Dinse, T. Wolfram, N. Hamilton, A. Trunschke, R. Schlogl and R. Schomacker, High performance (VO_x)n-(TiO_x)m/SBA-15 catalysts for the oxidative dehydrogenation of propane, Catal. Sci. Technol., 2014, 4, 786–794.
22. J. T. Grant, C. A. Carrero, F. Goeltl, J. Venegas, P. Mueller, S. P. Burt, S. E. Specht, W. P. McDermott, A. Chieregato and I. Hermans, Selective oxidative dehydrogenation of propane to propene using boron nitride catalysts, Science, 2016, 354, 1570–1573.
23. A. A. Lemonidou, L. Nalbandian and I. A. Vasalos, Oxidative dehydrogenation of propane over vanadium oxide based catalysts: Effect of support and alkali promoter, Catal. Today, 2000, 61, 333–341.
24. G. Garcia Cortez, J. L. G. Fierro and M. A. Bañares, Role of potassium on the structure and activity of alumina-supported vanadium oxide catalysts for propane oxidative dehydrogenation, Catal. Today, 2003, 78, 219–228.
25. S. Takenaka, T. Tanaka, T. Yamazaki, T. Funabiki and S. Yoshida, Structure of Active Species in Alkali-Ion-Modified Silica-Supported Vanadium Oxide, J. Phys. Chem. B, 1997, 101, 9035–9040.
26. J. T. Grant, A. M. Love, C. A. Carrero, F. Huang, J. Panger, R. Verel and I. Hermans, Improved supported metal oxides for the oxidative dehydrogenation of propane, Top. Catal., 2016, 59, 1545–1553.
27. S. Irusta, A. Marchi, E. Lombardo and E. Miré, Characterization of surface species on V/SiO₂ and V, Na/SiO₂ and their role in the partial oxidation of methane to formaldehyde, Catal. Lett., 1996, 40, 9–16.
28. J. T. Grant, C. A. Carrero, A. M. Love, R. Verel and I. Hermans, Enhanced Two-Dimensional Dispersion of Group V Metal Oxides on Silica, ACS Catal., 2015, 5, 5787–5793.
29. M. Nadjafi, P. M. Abdala, R. Verel, D. Hosseini, O. V. Safonova, A. Fedorov and C. R. Müller, Reducibility and Dispersion Influence the Activity in Silica-Supported Vanadium-Based Catalysts for the Oxidative Dehydrogenation of Propane: The Case of Sodium Decavanadate, ACS Catal., 2020, 10, 2314–2321.
30. I. Lukács and C. Strusievici, Über die Polymorphie von Natriummetavanadat, Z. Anorg. Allg. Chem., 1962, 315, 323–326.
31. S. Seetharaman, H. L. Bhat and P. S. Narayanan, Raman spectroscopic studies on sodium metavanadate, J. Raman Spectrosc., 1983, 14, 401–405.
32. I. E. Wachs, Catalysis science of supported vanadium oxide catalysts, Dalton Trans., 2013, 42, 11762–11769.
33. S. Xie, E. Iglesia and A. T. Bell, Effects of Hydration and Dehydration on the Structure of Silica-Supported Vanadia Species, Langmuir, 2000, 16, 7162–7167.
34. A. M. Love, C. A. Carrero, A. Chieregato, J. T. Grant, S. Conrad, R. Verel and I. Hermans, Elucidation of Anchoring and Restructuring Steps during Synthesis of Silica-Supported Vanadium Oxide Catalysts, Chem. Mater., 2016, 28, 5495–5504.
35. K. C. Szeto, B. Loges, N. Merle, N. Popoff, A. Quadrelli, H. Jia, E. Berrier, A. De Mallmann, L. Delevoye, R. M. Gauvin and M. Taoufik, Vanadium Oxo Organometallic Species Supported on Silica for the Selective Non-oxidative Dehydrogenation of Propane, Organometallics, 2013, 32, 6452–6460.
36. B. Schimmoeller, Y. Jiang, S. E. Pratsinis and A. Baiker, Structure of flame-made vanadia/silica and catalytic behavior in the oxidative dehydrogenation of propane, J. Catal., 2010, 274, 64–75.
37. K. Kato and E. Takayama, Das Entwasserungsverhalten des Natriummetavanadatdihydrats und die Kristallstruktur des β-Natriummetavanadats, Acta Crystallogr., Sect. B: Struct. Sci., 1984, 40, 102–105.
38. M. D. Argyle, K. Chen, A. T. Bell and E. Iglesia, Effect of Catalyst Structure on Oxidative Dehydrogenation of Ethane and Propane on Alumina-Supported Vanadia, J. Catal., 2002, 208, 139–149.
39. G. T. Went, S. T. Oyama and A. T. Bell, Laser Raman spectroscopy of supported vanadium oxide catalysts, J. Phys. Chem., 1990, 94, 4240–4246.
40. P. M. A. Sherwood, Vibrational spectroscopy of solids, CUP Archive, 1972.
41. J. Skibsted, N. C. Nielsen, H. Bildsoe and H. J. Jakobsen, Magnitudes and relative orientation of vanadium-51 quadrupole coupling and anisotropic shielding tensors in metavanadates and potassium vanadium oxide (KV₃O₈) from vanadium-51 MAS NMR spectra. Sodium-23 quadrupole coupling parameters for α- and β-NaVO₃, J. Am. Chem. Soc., 1993, 115, 7351–7362.
42. J. Wong, F. W. Lytle, R. P. Messmer and D. H. Maylotte, K-edge absorption spectra of selected vanadium compounds, Phys. Rev. B: Condens. Matter Mater. Phys., 1984, 30, 5596–5610.
43. A. Dinse, B. Frank, C. Hess, D. Habel and R. Schomäcker, Oxidative dehydrogenation of propane over low-loaded vanadia catalysts: Impact of the support material on kinetics and selectivity, J. Mol. Catal. A: Chem., 2008, 289, 28–37.
44. G. Deo and I. E. Wachs, Reactivity of Supported Vanadium Oxide Catalysts: The Partial Oxidation of Methanol, J. Catal., 1994, 146, 323–334.
45. S. Chen, L. Zeng, R. Mu, C. Xiong, Z.-J. Zhao, C. Zhao, C. Pei, L. Peng, J. Luo, L.-S. Fan and J. Gong, Modulating Lattice Oxygen in Dual-Functional Mo–V–O Mixed Oxides for Chemical Looping Oxidative Dehydrogenation, J. Am. Chem. Soc., 2019, 141, 18653–18657.
46. E. V. Kondratenko, N. Steinfeldt and M. Baerns, Transient and steady state investigation of selective and non-selective reaction pathways in the oxidative dehydrogenation of propane over supported vanadia catalysts, Phys. Chem. Chem. Phys., 2006, 8, 1624–1633.
47. P. Gruene, T. Wolfram, K. Pelzer, R. Schlögl and A. Trunschke, Role of dispersion of vanadia on SBA-15 in the oxidative dehydrogenation of propane, Catal. Today, 2010, 157, 137–142.
48. I. Rossetti, G. F. Mancini, P. Ghigna, M. Scavini, M. Piumetti, B. Bonelli, F. Cavani and A. Comite, Spectroscopic Enlightening of the Local Structure Of VO_X Active Sites in Catalysts for the ODH of Propane, J. Phys. Chem. C, 2012, 116, 22386–22398.
49. A. A. Zvyagin, Temperature dependence of the electron paramagnetic resonance linewidth in NaV₂O₅, Phys. Rev. B, 2001, 63, 172409.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details, XRD, NMR, Raman, H₂ temperature programmed reduction (TPR), XAS, DSC, and TEM data. See DOI: 10.1039/d0cy01234c

---