catalysts Dispersion and Stabilization of Supported Layered Double Hydroxide-Based Nanocomposites on V-Based Catalysts for Nonoxidative Dehydrogenation of Isobutane to Isobutene

: Nonoxidative dehydrogenation of isobutane is one of the sustainable strategies for pro-ducing high value added isobutene. As alternatives for the commercial Pt- and Cr-based dehydrogenation catalysts, supported V-based catalysts are worthy of study. In this work, a series of VO x /mMgAlO-R catalysts (m = 10, 15, 20, 25 and 30) were designed and prepared by loading VO x on mMgAlO composite oxide supports derived from mesoporous Al 2 O 3 -supported layered double hydroxide (LDH) nanocomposites. The calcined and reduced catalysts were characterized by X-ray diffraction (XRD), Raman spectra, Ultraviolet-visible diffuse reﬂectance (UV-Vis) spectra, NH 3 temperature-programmed desorption (NH 3 -TPD), Temperature-programmed reduction (H 2 -TPR), X-ray photoelectron spectroscopy (XPS), thermogravimetric analysis (TG) and low temperature N 2 adsorption–desorption isotherms. The as-synthesized VO x /20MgAlO-R with appropriate Mg addition exhibits superior activity (43–56% conversion and 77–81% selectivity), excellent stability and coking-resistance for the isobutane dehydrogenation. The structure–performance relationship reveals that the formation of VO x species conﬁned in the reconstructed LDH interlayer and porous MgO facilitates dispersing and stabilizing the VO x species. The low polymerization degree and higher proportion of V 4+ ion for VO x species, strong acidity of medium acid sites and low concentration of strong acid sites are responsible for the excellent anti-coking and catalytic performance. The strong VO x –support interaction is beneﬁcial for enhancing the stability of the catalysts. F.L. and resources, L.Z.; writing—original preparation, F.L.; writing—review and Y.X. and L.Z.; supervision, L.Z.; project administration, funding


Introduction
Isobutene is an important raw material to produce highly valued chemicals, such as methyl-tert-butyl ether, methacrylate polymers, butyl rubber and polyisobutylene [1]. In recent years, the demand for isobutene has been rapidly growing worldwide, and its production process has received more and more attention [2]. Compared with nonoxidative dehydrogenation, the reactants and products are further oxidized even in the presence of a small amount of O 2 due to the strong oxidizing properties of O 2 during oxidative dehydrogenation, thus reducing the selectivity of the products [3]. The nonoxidative dehydrogenation of isobutane is one of the significant ways to produce isobutene. For the process, the most important work is to develop an efficient and stable catalyst [1,4,5]. Although Pt-and CrO x -based catalysts have been widely used in industry, the high cost and toxicity of Pt-and CrO x -based catalysts has prompted some scientists to focus their work on the less costly and less hazardous non-noble metal oxide catalysts, such as VO x , GaO x , ZnO and FeO x [3,[6][7][8][9][10][11][12].
The high dehydrogenation performance and regeneration stability have made VO xbased catalysts one of the important alternative non-noble metal oxide catalysts widely used in oxidative dehydrogenation [13][14][15] and nonoxidative dehydrogenation of alkanes [3,10,11]. Usually, the catalytic performance and coking behavior are closely related to the surface acid-base properties and chemical state of VO x species. As for the valence states of VO x species, it is usually accepted that V 3+ , V 4+ and V 5+ ions could appear together, but only V 3+ and V 4+ ions are often considered as the active species [10,[16][17][18]. Some reports have emphasized that only the isolated V 3+ and V 4+ could be identified as dehydrogenation active sites [19,20]. Certainly, the single V 3+ ion was also found to be more active for dehydrogenation under certain conditions [21,22]. On the other hand, the polymerization degree (namely aggregation degree) of VO x also plays an important in promoting dehydrogenation and resisting coke formation [10,17,20,23]. Generally, the isolated and low polymerized VO x are more active than the aggregated V 2 O 5 [24,25]. However, the aggregated VO x species have also been reported to perform better than isolated VO x species [11]. Therefore, it is still necessary to investigate the structure-performance relationship of the VO x -based catalysts under different conditions.
As a commonly used support, Al 2 O 3 has been extensively studied in the dehydrogenation process due to the high specific surface area, excellent thermal stability and adjustable physical and chemical properties [11,[25][26][27]. Undoubtedly, the phase structure of Al 2 O 3 could influence the surface acidity, structure and dispersion of VO x species [26,27]. The slight difference in the surface VO x species dispersion depends on the different density of surface hydroxyl groups on γ-Al 2 O 3 and δ-Al 2 O 3 [27], while the crystallization temperature of Al 2 O 3 also can influence the proportion of polymerized VO x species and V 4+ state, which is associated with the surface acid sites [11]. The use of alkaline promoters is another way to modify the dehydrogenation performance of catalysts [13][14][15]. The addition of Mg can not only change the distribution and strength of surface acid sites, but also improve the stability and dispersion of VO x species [25]. The alkaline earth hydroxyapatite-supported V 2 O 5 displayed good dispersity and high selectivity to the desired product in the oxidative dehydrogenation of alkane [13][14][15]. Recently, there have been reports on catalysts made of combined V, Mg and Al in the nonoxidative dehydrogenation of alkane [25]. Besides Al 2 O 3 , the layered double hydroxide (LDH)-derived MgAlO composite oxide is also regarded as an excellent support applied in dehydrogenation reaction, because of its appropriate acid-base properties, easy reconstruction due to the memory effect and atomic-scale uniform distribution of metal species arising from the topotactic transformation [28][29][30][31]. However, most reports focus on the LDH derivative-supported Pt-based catalysts, while the studies on LDH-derived VO x -based catalysts are seldomly reported in the dehydrogenation reaction.
In this study, the in situ grown MgAl-LDH derivatives on the surface of Al 2 O 3 were prepared as composite supports to disperse and stabilize VO x species by reconstructing a brucite-like structure and forming porous MgO. The physicochemical properties of these catalysts and precursors were investigated by various methods, and the catalytic activity was evaluated in non-oxidative dehydrogenation of isobutane to isobutene. The target of this study is to investigate the effect of supported MgAl-LDH derivatives on the chemical states of VO x species, surface acid-base properties, coking behavior and catalytic performance by changing the Mg content. To confirm the phase structure of support precursors mMgAlO-P (m = 10, 15, 20, 25 and 30), composite supports mMgAlO, supported catalysts VO x /mMgAlO and the corresponding precursors VO x /mMgAlO-P, their X-ray diffraction (XRD) patterns are presented (see Figure 1). In Figure 1a, the strong characteristic diffraction peaks of the LDH phase can be detected at 2θ of ca. 11 (012), (015), (018), (110) and (113) planes of the LDH phase (JCPDS file No. 89-0640), respectively. There are no diffraction peaks of other phases, except the LDH phase can be recognized. This indicates the single LDH can be in situ formed on γ-Al2O3.

Results and Discussion
After calcination, all diffraction peaks of the LDH phase disappear for the composite supports mMgAlO (see Figure 1b) due to the layered structure collapse of the LDH phase. Consequently, the characteristic diffraction peaks of the MgO phase (JCPDS file No.  can be found at 2θ of ca. 42.9° and 62.3°, indexed as (200) and (220) planes, respectively, in addition to the diffraction peaks of γ-Al2O3 (JCPDS file No. 50-0741). In addition, the diffraction peak intensities of MgO have an obvious increase for 25MgAlO and 30MgAlO, with high Mg content, which means the excessive Mg addition can result in the growth of MgO crystallite size, even covering the partial VOx species. After calcination, all diffraction peaks of the LDH phase disappear for the composite supports mMgAlO (see Figure 1b) due to the layered structure collapse of the LDH phase. Consequently, the characteristic diffraction peaks of the MgO phase (JCPDS file No.  can be found at 2θ of ca. 42.9 • and 62.3 • , indexed as (200) and (220) planes, respectively, in addition to the diffraction peaks of γ-Al 2 O 3 (JCPDS file No. 50-0741). In addition, the diffraction peak intensities of MgO have an obvious increase for 25MgAlO and 30MgAlO, with high Mg content, which means the excessive Mg addition can result in the growth of MgO crystallite size, even covering the partial VO x species.
As shown in Figure 1c, [34]. This can be confirmed by the results of Figure 1d, Figure S1 (Supplementary Materials) and Table 1 below. In addition, no diffraction peaks can be attributed to VO x species, but a weak diffraction peak can be found at 2θ of ca. 9.6 • for VO x /25MgAlO-P and VO x /30MgAlO-P. This can be mainly assigned to the decavanadate [V 10 O 28 ] 6− intercalated LDH phase (LDH-[V 10 O 28 ] 6− ) [34]. This indicates that the calcined LDH structure has a different chance to be reconstructed along with the formation of MgC 2 O 4 ·2H 2 O, which depends on Mg content. In other words, the VO x species can be introduced into the interlayer of the reconstructed LDH for all samples, although there is no obvious diffraction of LDH-[V 10 O 28 ] 6− in the XRD patterns for VO x /10MgAlO-P, VO x /15MgAlO-P and VO x /20MgAlO-P, which is owing to the poor crystallinity arising from low Mg content.  Figure 1d). The variation trend of MgO diffraction peaks with Mg content is consistent with that of the MgC 2 O 4 ·2H 2 O precursor and different from that of initial MgO formed on mMgAlO supports ( Figure 1b). This further indicates that the formation of MgC 2 O 4 ·2H 2 O can influence the MgO texture, as discussed above. Another important phenomenon is that diffraction peaks of VO x species still cannot be detected for all samples of VO x /mMgAlO, which are calcined VO x /mMgAlO-P; this is related to the high dispersion of VO x species, and the possibility of low V loading amount can be excluded in this work. In fact, the formation of lattice-confined VO x species and porous MgO is beneficial for dispersing and stabilizing the VO x species on the surface of mMgAlO [35].
Considering the possible influence of MgC 2 O 4 ·2H 2 O formation on the texture, the low temperature N 2 adsorption-desorption isotherms and pore size distribution (PSD) curves of fresh catalysts VO x /mMgAlO with different Mg content are given in Figure S1; the corresponding BET special surface area (S BET ), total pore volume (V T ), average pore size (D AP ) and pore size distribution percentage in different range (D P ) are listed in Table 1. As shown in Figure S1a, all catalysts show the typical type IV isotherms with H3 hysteresis loops. This indicates that all catalysts possess the typical slit-shaped mesoporous feature related to a layered structure [29,36], which arises from the topological transformation of the LDH structure [29]. The pore size distribution curves in Figure S1b also confirm the mesoporous feature of all catalysts.
From Table 1, it can be seen that the S BET values increase from 102 to 161 m 2 ·g −1 , with the Mg content increasing from 10 to 20%, and then significantly decrease when the Mg content further increases. It is obvious that VO x /20MgAlO possesses the largest S BET value of 161 m 2 ·g −1 among all catalysts. At the same time, the continuous introduction of Mg has no influence on the V T value of catalysts; on the contrary, it enhances this value from 0.31 m 3 ·g −1 of VO x /10MgAlO to 0.51 m 3 ·g −1 of VO x /25MgAlO, except for the lowest V T value of 0.23 m 3 ·g −1 for VO x /30MgAlO. As seen in the D P values in the range of 7.5-17.0 nm, they display a similar trend to that of the S BET values, changing with Mg content. These results suggest that the VO x /mMgAlO-P precursors with more MgC 2 O 4 ·2H 2 O microcrystals enable more porous MgO formation, with higher S BET and D P values in the middle size range. This results agrees rather well with those discussed for XRD patterns above.

Polymerization Degree of VO x Species
To analyze surface VO x species, the catalysts VO x /mMgAlO with different Mg content were analyzed by Raman spectroscopy. As shown in Figure 2, no sharp characteristic Raman band can be found at 998 cm −1 for all catalysts, indicating that no V 2 O 5 crystal particles are formed on the surface of these catalysts [11,27]. The results are consistent with those of XRD patterns above and ultraviolet-visible diffuse reflectance (UV-Vis) spectra below. It is worth mentioning that only sample VO x /20MgAlO exhibits a stronger band at 1070 cm −1 , which is attributed to the V=O bond stretching mode, suggesting the presence of more isolated tetrahedral-coordinated VO x species on the surface of VO x /20MgAlO [11]. All samples display a weak signal at 488 cm −1 and a broad band in the range of 700-1000 cm −1 , which are caused by V-O-V and/or V-O-Al antisymmetric and symmetric stretching vibrations [3,10,11]. The appearance of these bands indicates the presence of polymerized VO x species on the surface of catalysts. With the increase of Mg content, the broad band shifts from 941 to 895 cm −1 , and the red shift of this band indicates that the polymerization degree of VO x species decreases [17]. We can also observe a doublet peak for VO x /30MgAlO; the peak at 872 cm −1 is related to the existence of microcrystal Mg 2 V 2 O 7 [37,38]. At the same time, this broad band grows in intensity significantly from VO x /10MgAlO to VO x /15MgAlO and VO x /20MgAlO, and then weakens obviously for VO x /25MgAlO and VO x /30MgAlO, which reflects the quantity change of surface polymerized VO x species. These differences indicate that the formation of moderate lattice-confined VO x species and abundant porous MgO facilitates dispersing the VO x species on the surface of mMgAlO [35,39]. of 161 m 2 ·g −1 among all catalysts. At the same time, the continuous introduction of Mg has no influence on the VT value of catalysts; on the contrary, it enhances this value from 0.31 m 3 ·g −1 of VOx/10MgAlO to 0.51 m 3 ·g −1 of VOx/25MgAlO, except for the lowest VT value of 0.23 m 3 ·g −1 for VOx/30MgAlO. As seen in the DP values in the range of 7.5-17.0 nm, they display a similar trend to that of the SBET values, changing with Mg content. These results suggest that the VOx/mMgAlO-P precursors with more MgC2O4·2H2O microcrystals enable more porous MgO formation, with higher SBET and DP values in the middle size range. This results agrees rather well with those discussed for XRD patterns above.

Polymerization Degree of VOx Species
To analyze surface VOx species, the catalysts VOx/mMgAlO with different Mg content were analyzed by Raman spectroscopy. As shown in Figure 2, no sharp characteristic Raman band can be found at 998 cm −1 for all catalysts, indicating that no V2O5 crystal particles are formed on the surface of these catalysts [11,27]. The results are consistent with those of XRD patterns above and ultraviolet-visible diffuse reflectance (UV-Vis) spectra below. It is worth mentioning that only sample VOx/20MgAlO exhibits a stronger band at 1070 cm −1 , which is attributed to the V=O bond stretching mode, suggesting the presence of more isolated tetrahedral-coordinated VOx species on the surface of VOx/20MgAlO [11]. All samples display a weak signal at 488 cm −1 and a broad band in the range of 700-1000 cm −1 , which are caused by V-O-V and/or V-O-Al antisymmetric and symmetric stretching vibrations [3,10,11]. The appearance of these bands indicates the presence of polymerized VOx species on the surface of catalysts. With the increase of Mg content, the broad band shifts from 941 to 895 cm −1 , and the red shift of this band indicates that the polymerization degree of VOx species decreases [17]. We can also observe a doublet peak for VOx/30MgAlO; the peak at 872 cm −1 is related to the existence of microcrystal Mg2V2O7 [37,38]. At the same time, this broad band grows in intensity significantly from VOx/10MgAlO to VOx/15MgAlO and VOx/20MgAlO, and then weakens obviously for VOx/25MgAlO and VOx/30MgAlO, which reflects the quantity change of surface polymerized VOx species. These differences indicate that the formation of moderate lattice-confined VOx species and abundant porous MgO facilitates dispersing the VOx species on the surface of mMgAlO [35,39]. All the catalysts VOx/mMgAlO are also characterized by the UV-Vis spectra to further determine the dispersion state of VOx species on the composite supports mMgAlO, and the corresponding spectra are displayed in Figure 3. As shown in Figure 3a, an intense and broad UV-Vis adsorption band related to the O 2− →V 5+ ligand to metal charge transfer All the catalysts VO x /mMgAlO are also characterized by the UV-Vis spectra to further determine the dispersion state of VO x species on the composite supports mMgAlO, and the corresponding spectra are displayed in Figure 3. As shown in Figure 3a, an intense and broad UV-Vis adsorption band related to the O 2− →V 5+ ligand to metal charge transfer (LMCT) can be found at ca. 275 nm, with a shoulder on the higher wavelength side for all samples. This means that the surface VO x species are dominated by highly dispersed isolated VO x species and low-polymerized VO x species with tetrahedral coordination [10,11,40]. The weak absorption bands at 350-400 nm indicate that there are few high-polymerized tetrahedral and distorted octahedral coordinated VO x species on the surface of the catalysts [11]. In this region, a slight decrease in intensity can be found from VO x /10MgAlO and VO x /15MgAlO to others with high Mg content. It can be deduced that there are more isolated and low-polymerized VO x species formed on VO x /20MgAlO than that on VO x /15MgAlO. Furthermore, the absent bands between 400 and 600 nm indicate that there are no V 2 O 5 crystals formed on the surface of catalysts [11], which is consistent with the results of XRD and Raman. The band at 208 nm is associated with O 2− →V 4+ LMCT [41], and the absorption band increases slightly with the increase of Mg, indicating that more V 4+ species are formed. (LMCT) can be found at ca. 275 nm, with a shoulder on the higher wavelength side for all samples. This means that the surface VOx species are dominated by highly dispersed isolated VOx species and low-polymerized VOx species with tetrahedral coordination [10,11,40]. The weak absorption bands at 350-400 nm indicate that there are few highpolymerized tetrahedral and distorted octahedral coordinated VOx species on the surface of the catalysts [11]. In this region, a slight decrease in intensity can be found from VOx/10MgAlO and VOx/15MgAlO to others with high Mg content. It can be deduced that there are more isolated and low-polymerized VOx species formed on VOx/20MgAlO than that on VOx/15MgAlO. Furthermore, the absent bands between 400 and 600 nm indicate that there are no V2O5 crystals formed on the surface of catalysts [11], which is consistent with the results of XRD and Raman. The band at 208 nm is associated with O 2− →V 4+ LMCT [41], and the absorption band increases slightly with the increase of Mg, indicating that more V 4+ species are formed. The change of edge energy (Eg) corresponds to the change of polymerization degree of VOx species on the catalyst surface [22]. As shown in Figure 3b, the Eg value slightly increases with increasing Mg content from 10% to 30%, further indicating that the polymerization degree of VOx species gradually decreases and that more isolated and low-polymerized VOx species are formed, which is consistent with the result of the Raman spectra above.

Surface Acidic Properties of Catalysts
Besides the aim to disperse VOx species, another important aspect is the modification of Mg on the acidity of catalysts; therefore, the NH3 temperature-programmed desorption (NH3-TPD) profiles of catalysts with different Mg content were used to analyze the acid sites of the catalysts, and the curves are drawn in Figure 4. Each curve can be fitted into three peaks, centered at ca. 160 °C (Peak I), 250 °C (Peak II) and 390 °C (Peak III), which are associated with the NH3 desorption from the weak, medium and strong acid sites, respectively [26,42,43].The peak area fractions and peak temperatures (Tmax) related to the relative concentration and strength of acid sites are collected in Table 2.
It can be noted that there is no obvious difference in the concentration and strength of the weak acid sites among these catalysts, which suggests that the influence of Mg on the weak acidity can be ignored. However, the concentration of medium acid sites presents a rising trend, with increasing Mg content from 10% to 25%, and then falls back to a The change of edge energy (E g ) corresponds to the change of polymerization degree of VO x species on the catalyst surface [22]. As shown in Figure 3b, the E g value slightly increases with increasing Mg content from 10% to 30%, further indicating that the polymerization degree of VO x species gradually decreases and that more isolated and low-polymerized VO x species are formed, which is consistent with the result of the Raman spectra above.

Surface Acidic Properties of Catalysts
Besides the aim to disperse VO x species, another important aspect is the modification of Mg on the acidity of catalysts; therefore, the NH 3 temperature-programmed desorption (NH 3 -TPD) profiles of catalysts with different Mg content were used to analyze the acid sites of the catalysts, and the curves are drawn in Figure 4. Each curve can be fitted into three peaks, centered at ca. 160 • C (Peak I), 250 • C (Peak II) and 390 • C (Peak III), which are associated with the NH 3 desorption from the weak, medium and strong acid sites, respectively [26,42,43].The peak area fractions and peak temperatures (T max ) related to the relative concentration and strength of acid sites are collected in Table 2.
the acid strength drops to a constant level. Obviously, the coverage of excessive Mg species on the acid sites is an important reason to decrease the acidity of VOx/25MgAlO and VOx/30MgAlO [25]. Generally, the different surface acidity of catalysts can affect the isobutane dehydrogenation. The enhancement of medium acidity could facilitate dehydrogenation behavior by establishing a balance between the adsorption of light alkanes and the desorption of alkenes; accordingly, the side reactions and even coke formation can be inhibited, which favors prolonging the lifespan of catalysts [10,[42][43][44].   It can be noted that there is no obvious difference in the concentration and strength of the weak acid sites among these catalysts, which suggests that the influence of Mg on the weak acidity can be ignored. However, the concentration of medium acid sites presents a rising trend, with increasing Mg content from 10% to 25%, and then falls back to a lower value than that of VO x /20MgAlO, as the Mg content further increases to 30%. At the same time, an opposite trend can be found for the concentration change of strong acid sites with Mg content increasing. According to the result, it can be deduced that the introduced Mg species preferentially interact with strong acid sites on the Al 2 O 3 support; therefore, the greater the Mg content is, the more the strong acid sites are neutralized and the larger the relative amount of the medium acid site that remains. Additionally, the strength of medium and strong acid sites can be enhanced by increasing the Mg content from 10% to 20%. This is related to the improving dispersion of VO x species, which is promoted by forming specific LDH derivatives as described above. Once the Mg content exceeds 20%, the acid strength drops to a constant level. Obviously, the coverage of excessive Mg species on the acid sites is an important reason to decrease the acidity of VO x /25MgAlO and VO x /30MgAlO [25]. Generally, the different surface acidity of catalysts can affect the isobutane dehydrogenation. The enhancement of medium acidity could facilitate dehydrogenation behavior by establishing a balance between the adsorption of light alkanes and the desorption of alkenes; accordingly, the side reactions and even coke formation can be inhibited, which favors prolonging the lifespan of catalysts [10,[42][43][44].

Reducibility and Surface Chemical State of VO x Species
The reduction behavior of the supported VO x species was evaluated by temperatureprogrammed reduction (H 2 -TPR), and the profiles are displayed in Figure 5. As shown in Figure S2, no reduction peak can be found in the TPR profile of the bare support. This indicates the Al 2 O 3 support cannot be reduced in the temperature range. All the catalysts present a single peak at around 550 • C, with a broad frontal peak. This must be related to the reduction of VO x species with various V-O bands [22,26]. Usually, the reduction process of VO x species is proposed from V 5+ to V 4+ and to V 3+ [13][14][15]. With the increase of Mg content, the reduction peak temperature increases from 531 • C of VO x /10MgAlO to 535 • C of VO x /15MgAlO and to 555 • C of others; meanwhile, the VO x polymerization degree presents a decreasing tendency, as illustrated by the Raman spectra above. It can be deduced that the higher reduction temperature suggests a stronger VO x -support interaction and a better dispersion of VO x species on the support. This must be related to the promoting role of reconstructed LDH intercalated with [V 10 O 28 ] 6− anions in the stability and separation of VO x species [39]. Another reason for the poor reducibility for VO x /25MgAlO and VO x /30MgAlO can be attributed to the cover of excessive MgO. Although the strong VO x -support interaction is beneficial to stabilizing active VO x species, it can result in the loss of isobutene selectivity to some extent, as a consequence of decreasing the electron density of V-O-Al [10,26]. In order to deeply investigate the reduction behavior of catalysts, these TPR profiles are deconvolved into three peaks; the detailed parameters, including peak temperature, peak area percentage, consumed hydrogen amount and average oxidation state (AOS a ) of VOx, are summarized in Table S1. From low to high temperature, three peaks can be assigned to the reduction of isolated monomeric VO x species (Peak I), oligomeric VO x species (Peak II) and high polymeric VO x species (Peak III), respectively [13][14][15]45]. The percentage of oligomeric VO x species is more than twice that of isolated monomeric VO x species, while the percentage of high polymeric VO x species is so small that it can be ignored. This indicates that the VO x species mainly belong to oligomeric and isolated states. Even VO x /25MgAlO presents the highest percentage of isolated monomeric VO x species. This further proves the deduction drawn from the results of Raman and UV-Vis spectra above. Additionally, the AOS a values of main oligomeric VO x species show a slight increase with the increase of Mg content. The decreasing reducibility of VO x species can be used to explain this phenomenon. However, the low-valence VOx species are considered to be the active sites [10,11,22]; therefore, the valence distribution and the average oxidation state (AOSb) of surface VOx species are further evaluated by X-Ray photoelectron spectroscopy (XPS). All XPS spectra of the reduced catalysts are displayed in Figure 6, and the deconvolution results are summarized in Table 3. Only one broad peak can be found in the V 2p3/2 region, and the corresponding binding energy (BE) increases from 516.6 eV to 517.0 eV in the following order: VOx/10MgAlO-R < VOx/15MgAlO-R < VOx/20MgAlO-R ≈ VOx/25MgAlO-R ≈ VOx/30MgAlO-R. This further proves the decrease of the VOx species electron density and enhancement of VOx-support interaction along with the Mg increasing to 20%, which is consistent with the H2-TPR result. Additionally, the original peak can be deconvoluted into three peaks related to V 3+ , V 4+ and V 5+ ion [10,11,26]. It can be found that the AOSb of VOx species and the fraction of V 5+ ion calculated by XPS deconvolution results exhibits a slightly rising tendency with increasing Mg content, which is also in agreement with the reducibility of catalysts analyzed by H2-TPR. This suggests that the fraction of low valence V ions, including V 3+ and V 4+ ions, tends to be reduced with increasing Mg content, However, the low-valence VO x species are considered to be the active sites [10,11,22]; therefore, the valence distribution and the average oxidation state (AOS b ) of surface VO x species are further evaluated by X-ray photoelectron spectroscopy (XPS). All XPS spectra of the reduced catalysts are displayed in Figure 6, and the deconvolution results are summarized in Table 3. Only one broad peak can be found in the V 2p 3/2 region, and the corresponding binding energy (BE) increases from 516.6 eV to 517.0 eV in the following order: VO x /10MgAlO-R < VO x /15MgAlO-R < VO x /20MgAlO-R ≈ VO x /25MgAlO-R ≈ VO x /30MgAlO-R. This further proves the decrease of the VO x species electron density and enhancement of VO xsupport interaction along with the Mg increasing to 20%, which is consistent with the H 2 -TPR result. Additionally, the original peak can be deconvoluted into three peaks related to V 3+ , V 4+ and V 5+ ion [10,11,26]. It can be found that the AOS b of VO x species and the fraction of V 5+ ion calculated by XPS deconvolution results exhibits a slightly rising tendency with increasing Mg content, which is also in agreement with the reducibility of catalysts analyzed by H 2 -TPR. This suggests that the fraction of low valence V ions, including V 3+ and V 4+ ions, tends to be reduced with increasing Mg content, especially for the V 3+ ion. However, there is more V 4+ ion for VO x /15MgAlO-R and VO x /20MgAlO-R than the others. In the dehydrogenation process, the coke is more conducive to forming on V 3+ sites than that on V 4+ sites [35]. This suggests that the serious coking behavior is related to the relatively low AOS b of VO x species in catalysts, which can be tuned by changing the Mg content.
coking behavior is related to the relatively low AOSb of VOx species in catalysts, which can be tuned by changing the Mg content.

Coking Behavior on the Used Catalysts
Carbon deposition usually occurs with the dehydrogenation process. The deposited carbon is one of the most important causes of catalyst deactivation. Therefore, it is necessary to evaluate the carbon deposition behavior on the catalysts by thermogravimetric analysis (TG) and XRD. As shown in Figure 7, the TG curves show significant mass loss at about 550 °C for VOx/mMgAlO-R (m = 10, 15, 20, 25 and 30). With the increase of Mg content, the mass loss presents a sharp drop from 21% of VOx/10MgAlO-R to 18% of

Coking Behavior on the Used Catalysts
Carbon deposition usually occurs with the dehydrogenation process. The deposited carbon is one of the most important causes of catalyst deactivation. Therefore, it is necessary to evaluate the carbon deposition behavior on the catalysts by thermogravimetric analysis (TG) and XRD. As shown in Figure 7, the TG curves show significant mass loss at about 550 • C for VO x /mMgAlO-R (m = 10, 15, 20, 25 and 30). With the increase of Mg content, the mass loss presents a sharp drop from 21% of VO x /10MgAlO-R to 18% of VO x /15MgAlO-R and to 13% VO x /20MgAlO-R. Meanwhile, the mass loss for catalysts with high Mg content, from 20-30%, is similar. Combined with the characteristic results of fresh catalysts, the coke accumulation is dependent on the amount of strong acid sites, the polymerization degree and the AOS of VO x species. Certainly, the poor dehydrogenation activity as a consequence of covering VO x species by excessive MgO also cannot be excluded as a reason for the lower amount of coke deposition of VO x /25MgAlO-R and VO x /30MgAlO-R. It can be deduced that the excellent coking resistance of VO x /20MgAlO-R is a result of the relatively high concentration of strong acid sites, low polymerization degree of VO x species and high AOS of surface VO x species, which is close to 4+ state (see Table 3) [10,22,46]. Instead, heavy coke deposition occurs on VO x /10MgAlO-R and VO x /15MgAlO-R, which could result in quick deactivation. VOx/15MgAlO-R and to 13% VOx/20MgAlO-R. Meanwhile, the mass loss for catalysts with high Mg content, from 20%-30%, is similar. Combined with the characteristic results of fresh catalysts, the coke accumulation is dependent on the amount of strong acid sites, the polymerization degree and the AOS of VOx species. Certainly, the poor dehydrogenation activity as a consequence of covering VOx species by excessive MgO also cannot be excluded as a reason for the lower amount of coke deposition of VOx/25MgAlO-R and VOx/30MgAlO-R. It can be deduced that the excellent coking resistance of VOx/20MgAlO-R is a result of the relatively high concentration of strong acid sites, low polymerization degree of VOx species and high AOS of surface VOx species, which is close to 4+ state (see Table 3) [10,22,46]. Instead, heavy coke deposition occurs on VOx/10MgAlO-R and VOx/15MgAlO-R, which could result in quick deactivation.  Figure 8. It can be seen that the samples still retain the peaks of MgO and γ-Al2O3 after a long reaction time, but there is no peak of carbonaceous species for all samples after the reaction, which means that the deposited coke is amorphous, thus making them easy to be regenerated in the next step.   Figure 8. It can be seen that the samples still retain the peaks of MgO and γ-Al 2 O 3 after a long reaction time, but there is no peak of carbonaceous species for all samples after the reaction, which means that the deposited coke is amorphous, thus making them easy to be regenerated in the next step. VOx/15MgAlO-R and to 13% VOx/20MgAlO-R. Meanwhile, the mass loss for catalysts with high Mg content, from 20%-30%, is similar. Combined with the characteristic results of fresh catalysts, the coke accumulation is dependent on the amount of strong acid sites, the polymerization degree and the AOS of VOx species. Certainly, the poor dehydrogenation activity as a consequence of covering VOx species by excessive MgO also cannot be excluded as a reason for the lower amount of coke deposition of VOx/25MgAlO-R and VOx/30MgAlO-R. It can be deduced that the excellent coking resistance of VOx/20MgAlO-R is a result of the relatively high concentration of strong acid sites, low polymerization degree of VOx species and high AOS of surface VOx species, which is close to 4+ state (see Table 3) [10,22,46]. Instead, heavy coke deposition occurs on VOx/10MgAlO-R and VOx/15MgAlO-R, which could result in quick deactivation.  Figure 8. It can be seen that the samples still retain the peaks of MgO and γ-Al2O3 after a long reaction time, but there is no peak of carbonaceous species for all samples after the reaction, which means that the deposited coke is amorphous, thus making them easy to be regenerated in the next step.

Catalytic Performance
The curves with the time on stream of isobutane dehydrogenation on the reduced catalyst VO x /mMgAlO-R (m = 10,15,20,25,30) are displayed in Figure 9. The isobutane conversions follow a decreasing trend from VO x /20MgAlO-R to VO x /15MgAlO-R to VO x /10MgAlO-R to VO x /25MgAlO-R and to VO x /30MgAlO-R in most reaction periods. The enhancement of isobutane conversion from VO x /10MgAlO-R to VO x /15MgAlO-R and to VO x /20MgAlO-R is related the decreasing polymerization degree of the VO x species, elevating the concentration and strength of medium acid sites and increasing AOS close to a 4+ state. However, the poor dehydrogenation activity of VO x /25MgAlO-R and VO x /30MgAlO-R mainly arises from the partial surface VO x covered by excessive MgO, because the addition of excessive Mg can lead to the overflow of MgO [25]. Additionally, all catalysts present a decreasing activity with time on stream in the reaction, and the decline degree is correlated well with the amount of coke deposition. The dramatic activity loss of VO x /10MgAlO-R and VO x /15MgAlO-R is likely due to the severe coke formation during the whole reaction [22]. The stable dehydrogenation behavior of the others is related to their strong VO x -support interaction, which can prevent the aggregation of VO x species. In particular, the catalyst VO x /20MgAlO-R shows maximum activity, but its isobutane conversion only shows a slight decrease from the initial 56% to 43% after reaction of 7.5 h.
conversions follow a decreasing trend from VOx/20MgAlO-R to VOx/15MgAlO-R to VOx/10MgAlO-R to VOx/25MgAlO-R and to VOx/30MgAlO-R in most reaction periods. The enhancement of isobutane conversion from VOx/10MgAlO-R to VOx/15MgAlO-R and to VOx/20MgAlO-R is related the decreasing polymerization degree of the VOx species, elevating the concentration and strength of medium acid sites and increasing AOS close to a 4+ state. However, the poor dehydrogenation activity of VOx/25MgAlO-R and VOx/30MgAlO-R mainly arises from the partial surface VOx covered by excessive MgO, because the addition of excessive Mg can lead to the overflow of MgO [25]. Additionally, all catalysts present a decreasing activity with time on stream in the reaction, and the decline degree is correlated well with the amount of coke deposition. The dramatic activity loss of VOx/10MgAlO-R and VOx/15MgAlO-R is likely due to the severe coke formation during the whole reaction [22]. The stable dehydrogenation behavior of the others is related to their strong VOx-support interaction, which can prevent the aggregation of VOx species. In particular, the catalyst VOx/20MgAlO-R shows maximum activity, but its isobutane conversion only shows a slight decrease from the initial 56% to 43% after reaction of 7.5 h.
As for the isobutene selectivity and yield, the trends with reaction time are similar to those of isobutane conversion for all catalysts, except the isobutene selectivity of VOx/20MgAlO-R, which is slightly lower than that of VOx/15MgAlO-R but not less than 72%. The result agrees well with the deduction from H2-TPR. That is to say, the most likely reason for the lower isobutene selectivity of VOx/20MgAlO-R is related to the lower electron density of V-O-Al arising from the stronger interaction between the VOx and support, which makes isobutene more difficult to desorb from the surface of VOx/20MgAlO-R, thus resulting in the side reactions and further low selectivity. Although there is slight loss of isobutene selectivity, the stability of VOx/20MgAlO-R is effectively improved, and the superior isobutene yield, more than 33%, is achieved as expected.

Materials
The required materials include MgCl2·6H2O (98.0% purity, Tianjin Kermel Chemical Reagents Co., Ltd., Tianjin, China), urea (99.0% purity, DAMAO Chemical Reagents Co., As for the isobutene selectivity and yield, the trends with reaction time are similar to those of isobutane conversion for all catalysts, except the isobutene selectivity of VO x /20MgAlO-R, which is slightly lower than that of VO x /15MgAlO-R but not less than 72%. The result agrees well with the deduction from H 2 -TPR. That is to say, the most likely reason for the lower isobutene selectivity of VO x /20MgAlO-R is related to the lower electron density of V-O-Al arising from the stronger interaction between the VO x and support, which makes isobutene more difficult to desorb from the surface of VO x /20MgAlO-R, thus resulting in the side reactions and further low selectivity. Although there is slight loss of isobutene selectivity, the stability of VO x /20MgAlO-R is effectively improved, and the superior isobutene yield, more than 33%, is achieved as expected.

Composite Oxide Support Preparation
The MgAl composite oxide support precursors with different Mg content were prepared by the hydrothermal method as described in our previous work [29]. First, MgCl 2 ·6H 2 O and urea were dissolved in 65 mL deionized water with Mg/urea of 1/4 under vigorous stirring at room temperature. Then, 1.5 g γ-Al 2 O 3 powder was added into the mixture solution. After fully stirring, the mixture was transferred into a 100 mL Teflon autoclave and heated at 120 • C for 20 h. Then, the composite precursor was obtained by filtering, washing and drying the mixture at 120 • C for 12 h, and it was marked as mMgAlO-P (m = 10, 15, 20, 25 and 30), where "m" is the theoretical weight percentage of MgO relative to γ-Al 2 O 3 . After calcining the precursor at 550 • C for 4 h, the corresponding γ-Al 2 O 3 -supported composite support was obtained and denoted as mMgAlO.

Catalyst Preparation
The composite supports mMgAlO-supported VO x catalysts (VO x /mMgAlO) with a V loading amount of 7 wt% were prepared by the traditional incipient impregnation method. First, the as-synthesized supports mMgAlO were impregnated with an aqueous solution containing NH 4 VO 3 and oxalic acid, with the molar ratio of 1:2. The wet solids were aged in air at room temperature for 12 h, dried at 120 • C overnight, and finally calcined in a muffle furnace at 550 • C for 4 h with a heating rate of 2 • C·min −1 ; the corresponding precursors were designated as VO x /mMgAlO. Before reaction, these calcined samples were subjected to a high temperature pre-reduction treatment in 5 vol% H 2 /N 2 (30 mL·min −1 ) at 600 • C for 2 h, and the products were labeled as reduced catalysts VO x /mMgAlO-R.

Catalyst Characterization
X-ray diffraction (XRD) characterization was carried out on D-Foucas (Bruker, Karlsruhe, Germany) equipped with a Cu Kα radiation (λ = 0.15418 nm), and the spectral scanning angle and scanning speed were 10-90 • and 8 • /min, respectively. A low-temperature N 2 adsorption-desorption test was performed on an automated analyzer (QUA211007, Quantachrome, Boynton, FL, USA). Raman spectra were measured at ambient conditions by a DXR Raman spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with an excitation wavelength of 532 nm. Ultraviolet-visible diffuse reflectance (UV-Vis) spectra (200-800 nm) were acquired on a Shimadzu UV-3600 spectrophotometer (Kyoto, Japan) in the region of 200-800 nm, Normalized UV-Vis spectra were drawn according to the Kubelka-Munk equation: F(R) = (1 − R) 2 /2R, where R is reflectivity. A temperature-programmed reduction (H 2 -TPR) experiment was carried out on automatic multipurpose adsorption apparatus (tp5080 XQINSTRUMENT Co., Ltd., Tianjin, China). The NH 3 temperatureprogrammed desorption (NH 3 -TPD) measurements were taken using a Micromeritics AutoChem II 2920 apparatus (Micromeritics, Norcross, GA, USA). The X-ray photoelectron spectroscopy (XPS) measurement was carried out using a Thermo ESCALAB 250Xi (US) (Thermo Fisher Scientific, Waltham, MA, USA), using Al Kα radiation and a C1s level with a binding energy (BE) of 284.8 eV as an internal reference to obtain information on the chemical composition and chemical states of the elements located on the surface of the reduced catalysts. The coking behavior of the used catalysts was analyzed by thermogravimetric analysis (TG, HITACHI STA7300, HITACHI, Tokyo, Japan), with a heating rate of 10 • C·min −1 from room temperature to 900 • C in air.

Catalytic Performance Test
The isobutane dehydrogenation reaction was carried out in a fixed bed reactor at 600 • C and atmospheric pressure; the inner diameter of the reaction tube was 8 mm, the amount of catalyst was 0.5 g (40-60 mesh). The reduction pretreatment was carried out in an atmosphere of 5% H 2 and 95% N 2 . After that, the isobutane and hydrogen were mixed into the reaction tube at a molar ratio of 1:1. The weight hourly space velocity (WHSV) of isobutane was 3 h −1 , and the concentrations of all hydrocarbons involving C 4 H 10 , C 4 H 8 , C 3 H 8 , C 3 H 6 , C 2 H 6 , C 2 H 4 and CH 4 , were analyzed by an online gas chromatograph (GC) equipped with a flame ionization detector (FID). The carbon balance ranged between 96% and 102%, which was evaluated by comparing the number of moles of carbon in the outlet gases to the number of moles of carbon in the inlet C 4 H 10 gas and calibrating by taking into consideration the coke amount determined by the TG measurement. The isobutane conversion, isobutene selectivity and yield were calculated as follows:

Conclusions
The composite oxide-supported VO x -based catalysts VO x /mMgAlO-R (m = 10, 15, 20, 25 and 30) were successfully prepared and applied in the nonoxidative dehydrogenation of isobutane to isobutene. The composite oxide supports can be obtained by calcining in-situ synthesized MgAl-LDH/Al 2 O 3 . The catalytic performance strongly depends on the Mg content. The correlation between structure and catalytic performance reveals that the appropriate addition of Mg can help the formation of the reconstructed brucite-like lattice-confined VO x species and abundant porous MgO, which facilitates dispersing and stabilizing the VO x species by strongly interacting with the support, and thus promotes the formation of more isolated and low-polymerized VO x species and a higher proportion of V 4+ ions. These factors lead the catalyst VO x /20MgAlO-R to exhibit superior activity and stability and excellent coke resistance, without significantly impairing isobutene selectivity. Additionally, the increase of Mg content could increase the consumption of the strong acid sites and enhance the strength and concentration of medium acid sites of the Al 2 O 3 surface, thus resisting coke formation and improving dehydrogenation performance. The excess MgO formation could cover the surface of VO x species and result in the dramatic loss in activity for VO x /25MgAlO-R and VO x /30MgAlO-R.