Promoting Li/MgO Catalyst with Molybdenum Oxide for Oxidative Conversion of n-Hexane

In this work, molybdena-promoted Li/MgO is studied as a catalyst for the oxidative conversion of n-hexane. The structure of the catalysts is investigated with X-ray Diffraction (XRD) and Raman spectroscopy. The MoO3/Li/MgO catalyst contains three types of molybdena-containing species, the presence of which depend on molybdena loading. At low Mo/Li ratios (i) isolated dispersed [MoO4]2− anionic species are observed. At high Mo/Li ratios, the formation of crystalline lithium molybdate phases such as (ii) monomeric Li2MoO4 and tentatively (iii) polymeric Li2Mo4O13 are concluded. The presence of these lithium molybdates diminishes the formation of Li2CO3 in the catalyst. Subsequently, the catalyst maintains high surface area and stability with time-on-stream during oxidative conversion. Molybdena loading as low as 0.5 wt % is sufficient to induce these improvements, maintaining the non-redox characteristics of the catalyst, whereas higher loadings enhance deep oxidation and oxidative dehydrogenation reactions. Promoting a Li/MgO catalyst with 0.5 wt % MoO3 is thus efficient for selective conversion of n-hexane to alkenes, giving alkene yield up to 24% as well as good stability.


Introduction
Catalytic oxidative conversion of alkanes to alkenes has gained interest over the years for on-purpose alkene production. The main challenge is that alkenes are highly reactive and susceptible to further oxidation in the presence of gaseous oxygen via adsorption on the catalyst surface. Therefore, to maintain high alkene yields in oxidative conditions, the use of basic catalysts is essential.
Li/MgO is a promising catalyst for the oxidative conversion (dehydrogenation/cracking) of lower alkanes to alkenes [1][2][3][4][5][6][7][8][9][10][11][12]. This catalyst has no formal redox character, i.e., Li + and Mg 2+ are not susceptible to oxidation state changes during the above reactions, and together with its inherent strong Brønsted basicity, it minimizes the re-adsorption and sequential combustion of formed alkenes [1][2][3][4][5][6][7][8]. Thus, the catalyst results in high selectivity to alkenes, which is highly desirable in the oxidative conversion of alkanes. This makes Li/MgO a better catalyst for oxidative reactions compared to acidic or redox-type catalysts, such as alumina or vanadia. It has been established through the work of Lunsford [13][14][15][16][17][18] on the oxidative coupling of methane that [Li + O − ]-type defect sites are responsible for catalytic activity. The nucleophilic [O − ] site is a strong hydrogen abstractor and initiates alkane activation via homolytic scission of the C-H bond in the alkane forming a radical. However, contradicting this, more recently Schlögl and co-workers [19,20] concluded, supported by quantum chemical calculations, that [Li + O − ] is not the active site. Calculations on cluster models illustrated that both Li/MgO and MgO possess the same nature of active sites; i.e., low coordinated Mg 2+ O 2− reaction with another [Li + O − ] site is converted into the more stable Li 2 CO 3 [18].
In order to improve both the activity and stability of Li/MgO, we promoted the catalyst with low amounts of oxides with redox properties [6]. Our goal was to add slight redox activity to the catalyst to enhance C-H bond scission and eventually n-hexane conversions. Of the different transition metal oxides attempted, molybdena-promoted Li/MgO showed the best alkene yields. Significantly, the presence of molybdena also prevented deactivation and catalyst stability was restored [6,7].
Various molybdenum oxide systems have been reported, e.g., supported on MgO, ZrO 2 , Al 2 O 3 , TiO 2 , and SiO 2 [32,35,36,[39][40][41][42][43]. The presence of molybdena influences the physiochemical properties of the oxide support. In general, molybdena content, state of molybdena species on the surface of the support, and calcination temperature influence both the textural and acidic properties of the support. For example, in alumina-supported molybdena, in addition to Lewis acid sites, Brønsted acid sites are detected, the concentration of which depends on molybdena loading and calcination temperature [44]. In zirconia-supported molybdena, the increase in acidity with an increase in molybdena loading is correlated to the formation of Mo-O-Zr surface species as precursors for crystalline Zr(MoO 4 ) 2 [45]. Moreover, molybdena enhances the surface area of zirconia through stabilizing the tetragonal zirconia phase [46,47].
Generally, the performance of molybdena-based catalysts is related to the extent of crystallinity and chemical structure of the oxidic molybdena species on the support, e.g., free MoO 3 , monomeric MoO 4 2− or polymeric Mo 6 O 19 2− , Mo 7 O 24 −6 units [32,35,41]. Hence, structure-performance correlations for supported molybdena catalysts have been of continuous interest. Magnesium oxide-supported molybdena (MoO 3 /MgO), in particular, is reported to be an efficient and selective catalyst for the oxidative dehydrogenation of C 3 -C 4 alkanes to the corresponding alkenes [33,34]. The activity of the MoO 3 /MgO for alkane activation depends on the molybdena loading [37,38]. In the oxidative dehydrogenation of propane over a MgMoO x catalyst, for example, Yoon et al. [37,38] reported that optimal molybdena loadings are necessary to maintain high propane conversions. Mg 0.95 MoO x crystallites with a slight excess of molybdenum (Mg/Mo = 0.9-1.0) showed the highest activity for propane conversion (22% conversion at 515 • C). Above these optimal loadings oxidation activity became significant. The high oxidation activity in these catalysts is due to the facile redox properties of molybdena, where the cation easily undergoes a change in the oxidation state (e.g., Mo 6+ to Mo 4+ ) [25]. In relation to the chemical structure of the molybdates on magnesium oxide, extensive literature has been published on the topic [39][40][41][42][43]. Bare and co-workers [41,42] reported that, in the case of MoO 3 /MgO catalysts, the structure of molybdena species depends on the molybdenum coverage of the support. For sub-monolayer coverages, dispersed species are observed. These consist of highly distorted octahedral molybdena species, e.g., MoO 6 , at low molybdenum loadings and regular octahedrally-coordinated polymolybdate species, e.g., [Mo 7 O 24 ] 6− , at high molybdenum loadings [41]. For coverages exceeding the monolayer, crystalline magnesium molybdate (MgMoO 4 ), in which molybdenum is tetrahedrally coordinated, is observed as the dominant species. Raman spectroscopic studies [39,40] showed that for magnesium oxide-supported catalysts, surface molybdena species are sensitive to hydration. Upon exposure to water, octahedrally-coordinated molybdena species transform to tetrahedrally-coordinated MoO 4 2− species.
In this work, we present a detailed study on the role of molybdena in improving the performance of Li/MgO. A detailed characterization of the MoO 3 /Li/MgO catalysts is presented in order to identify the chemical structure of the different molybdena species (MoO x ) present on surface of Li/MgO. Our objective is to determine the influence of different molybdena species on (i) n-hexane conversion, (ii) alkene versus combustion selectivity, and (iii) catalyst stability. This is expected to help establish guidelines for developing an optimal catalyst for the oxidative conversion of n-hexane. propane conversion (22% conversion at 515 °C). Above these optimal loadings oxidation activity became significant. The high oxidation activity in these catalysts is due to the facile redox properties of molybdena, where the cation easily undergoes a change in the oxidation state (e.g., Mo 6+ to Mo 4+ ) [25]. In relation to the chemical structure of the molybdates on magnesium oxide, extensive literature has been published on the topic [39][40][41][42][43]. Bare and co-workers [41,42] reported that, in the case of MoO3/MgO catalysts, the structure of molybdena species depends on the molybdenum coverage of the support. For sub-monolayer coverages, dispersed species are observed. These consist of highly distorted octahedral molybdena species, e.g., MoO6, at low molybdenum loadings and regular octahedrally-coordinated polymolybdate species, e.g., [Mo7O24] 6− , at high molybdenum loadings [41]. For coverages exceeding the monolayer, crystalline magnesium molybdate (MgMoO4), in which molybdenum is tetrahedrally coordinated, is observed as the dominant species. Raman spectroscopic studies [39,40] showed that for magnesium oxide-supported catalysts, surface molybdena species are sensitive to hydration. Upon exposure to water, octahedrally-coordinated molybdena species transform to tetrahedrally-coordinated MoO4 2− species.

Catalytic Tests
In this work, we present a detailed study on the role of molybdena in improving the performance of Li/MgO. A detailed characterization of the MoO3/Li/MgO catalysts is presented in order to identify the chemical structure of the different molybdena species (MoOx) present on surface of Li/MgO. Our objective is to determine the influence of different molybdena species on (i) n-hexane conversion, (ii) alkene versus combustion selectivity, and (iii) catalyst stability. This is expected to help establish guidelines for developing an optimal catalyst for the oxidative conversion of n-hexane.   It has similar problems with Figure 1    More specifically, Figure 2 presents n-hexane conversions as a function of Mo/Li ratio after 6 h of time-on-stream. The catalyst promoted with 0.5wt % molybdena (Mo/Li = 0.03) resulted in the highest activity, presenting significant improvement in activity compared to undoped Li/MgO. The influence of varying loadings of molybdena on the performance of Li/MgO during the oxidative conversion of n-hexane at 575 °C is shown in Table 1. Initial conversions and corresponding product selectivities after 30 min time-on-stream are reported. It is important to note that product selectivities remained constant with time. 0.5MoO3/Li/MgO showed similar conversion and selectivity to products as Li/MgO, however catalysts with higher molybdena loadings resulted in lower conversion and more combustion products. Moreover, in Figure 3 we compare selectivities to products at the same n-hexane conversion of ca. 10 mol %. The same n-hexane conversions were achieved through changing contact time with the The influence of varying loadings of molybdena on the performance of Li/MgO during the oxidative conversion of n-hexane at 575 • C is shown in Table 1. Initial conversions and corresponding product selectivities after 30 min time-on-stream are reported. It is important to note that product selectivities remained constant with time. 0.5MoO 3 /Li/MgO showed similar conversion and selectivity to products as Li/MgO, however catalysts with higher molybdena loadings resulted in lower conversion and more combustion products. Moreover, in Figure 3 we compare selectivities to products at the same n-hexane conversion of ca. 10 mol %. The same n-hexane conversions were achieved through changing contact time with the catalyst. The 0.5MoO 3 /Li/MgO catalyst showed a similar selectivity pattern to that of Li/MgO, while both the 3.6MoO 3 /Li/MgO and 7.1MoO 3 /Li/MgO catalysts resulted in the formation of more combustion products. Significant formation of C 6 alkene (hexene) was observed in the case of both the 3.6MoO 3 /Li/MgO and 7.1MoO 3 /Li/MgO catalysts. Oxygen conversion was almost same (~35 mol %) over both the Li/MgO and 0.5MoO 3 /Li/MgO catalysts, while it increased to 43 mol % and 53 mol % over 3.6MoO 3 /Li/MgO and 7.1MoO 3 /Li/MgO, respectively. Figure 4 shows the selectivities to C 2 -C 4 alkenes as a function of n-hexane conversion. These results clearly indicate that the 0.5MoO 3 /Li/MgO catalyst preserved the characteristics of the unpromoted catalyst and maintained the high alkene selectivity, even with increasing n-hexane conversion. However, both the 3.6MoO 3 /Li/MgO and 7.1MoO 3 /Li/MgO catalysts exhibited typical performance of redox catalysts, i.e., decreasing selectivity to alkenes with increasing conversion. catalyst. The 0.5MoO3/Li/MgO catalyst showed a similar selectivity pattern to that of Li/MgO, while both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts resulted in the formation of more combustion products. Significant formation of C6 alkene (hexene) was observed in the case of both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts. Oxygen conversion was almost same (~35 mol %) over both the Li/MgO and 0.5MoO3/Li/MgO catalysts, while it increased to 43 mol % and 53 mol % over 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO, respectively.  Figure 4 shows the selectivities to C2-C4 alkenes as a function of n-hexane conversion. These results clearly indicate that the 0.5MoO3/Li/MgO catalyst preserved the characteristics of the unpromoted catalyst and maintained the high alkene selectivity, even with increasing n-hexane conversion. However, both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts exhibited typical performance of redox catalysts, i.e. decreasing selectivity to alkenes with increasing conversion.   both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts resulted in the formation of more combustion products. Significant formation of C6 alkene (hexene) was observed in the case of both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts. Oxygen conversion was almost same (~35 mol %) over both the Li/MgO and 0.5MoO3/Li/MgO catalysts, while it increased to 43 mol % and 53 mol % over 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO, respectively.  Figure 4 shows the selectivities to C2-C4 alkenes as a function of n-hexane conversion. These results clearly indicate that the 0.5MoO3/Li/MgO catalyst preserved the characteristics of the unpromoted catalyst and maintained the high alkene selectivity, even with increasing n-hexane conversion. However, both the 3.6MoO3/Li/MgO and 7.1MoO3/Li/MgO catalysts exhibited typical performance of redox catalysts, i.e. decreasing selectivity to alkenes with increasing conversion.

Surface Area and XRD
The characteristics of the studied catalysts are presented in Table 2. It was observed that Li/MgO suffered a dramatic decrease in surface area upon calcination at 600 • C (15 m 2 /g), while in the presence of molybdena, high surface areas (70-82 m 2 /g) were retained. In the case of MgO, the loss in surface area upon calcination at 600 • C was less severe. Nevertheless, promotion with ≥3.3 wt % of MoO 3 resulted in even higher surface areas than the unpromoted sample calcined at 600 • C (148 m 2 /g). The MoO x surface coverage (θ) of molybdena-promoted catalysts shown in Table 2 were calculated based on the surface area of each catalyst and using 22 Å 2 as the mean surface area occupied by one Mo 6+ oxide unit (MoO 3 ) [5]. Values indicated sub-monolayer coverage for all samples studied. Figure 5 presents the XRD patterns of the catalysts. All catalysts exhibited characteristic peaks for crystalline MgO. No XRD peaks corresponding to crystalline MgMoO 4 or MgMo 2 O 7 phases were observed, which would be expected at 2theta values between 20-300 [38]. The absence of such peaks confirms that molybdena phases were finely dispersed and no crystalline phases were observed. Our results are in good agreement with literature findings; in supported molybdenum oxide systems with sub-monolayer coverages the presence of dispersed molybdena anionic species such as MoO 4 2-, In addition, no peaks corresponding to any crystalline Li phases, e.g., Li 2 CO 3 , were observed in any molybdena-promoted catalysts. The observed shift in the MgO peak positions in the MoO x -containing samples indicates the formation of solid solutions between MoO x and Li/MgO. The observed shift was Catalysts 2020, 10, 354 6 of 16 stronger in the 0.5MoO 3 /Li/MgO and 3.6MoO 3 /Li/MgO samples than in 7.1MoO 3 /Li/MgO, suggesting that the formation of lithium molybdates is dominant in the latter. This was further investigated by characterization of the catalysts with Raman spectroscopy.  Figure 5 presents the XRD patterns of the catalysts. All catalysts exhibited characteristic peaks for crystalline MgO. No XRD peaks corresponding to crystalline MgMoO4 or MgMo2O7 phases were observed, which would be expected at 2theta values between 20 -300 [38]. The absence of such peaks confirms that molybdena phases were finely dispersed and no crystalline phases were observed. Our results are in good agreement with literature findings; in supported molybdenum oxide systems with sub-monolayer coverages the presence of dispersed molybdena anionic species such as MoO4 2-, Mo6O19 2− , and Mo7O24 6− are commonly reported [32,35,36,41].
In addition, no peaks corresponding to any crystalline Li phases, e.g., Li2CO3, were observed in any molybdena-promoted catalysts. The observed shift in the MgO peak positions in the MoOxcontaining samples indicates the formation of solid solutions between MoOx and Li/MgO. The observed shift was stronger in the 0.5MoO3/Li/MgO and 3.6MoO3/Li/MgO samples than in 7.1MoO3/Li/MgO, suggesting that the formation of lithium molybdates is dominant in the latter. This was further investigated by characterization of the catalysts with Raman spectroscopy.     Figure 7 presents results of TPD experiments for both fresh (a) and used (b) catalysts. The TPD of Li/MgO exhibited a broad CO2 desorption peak with a maximum at 860 °C, which is typical for Li2CO3 [18]. The presence of Li2CO3 (formed with CO2 from ambient) is an inherent property of Li/MgO [4,26]. The TPD of the molybdena-promoted catalysts exhibited similar Li2CO3 peaks, however with lower intensity, decreasing with increasing molybdena loading. This confirms the gradual decrease in amounts of Li2CO3 in Li/MgO with increasing molybdena loading.

Temperature Programmed Desorption (TPD)
Similarly, the CO2 desorption peak observed for the used 0.5MoO3/Li/MgO is significantly  Figure 7 presents results of TPD experiments for both fresh (a) and used (b) catalysts. The TPD of Li/MgO exhibited a broad CO 2 desorption peak with a maximum at 860 • C, which is typical for Li 2 CO 3 [18]. The presence of Li 2 CO 3 (formed with CO 2 from ambient) is an inherent property of Li/MgO [4,26]. The TPD of the molybdena-promoted catalysts exhibited similar Li 2 CO 3 peaks, however with lower intensity, decreasing with increasing molybdena loading. This confirms the gradual decrease in amounts of Li 2 CO 3 in Li/MgO with increasing molybdena loading.

Raman Spectra
All MoO3/Li/MgO catalysts were characterized by Raman spectroscopy (Figure 8b). For better assignment of Raman bands, MgO promoted with comparable molybdena loadings was prepared and characterized (Figure 8a). At first, we assign the Raman bands of MoO3/MgO. Figure 8a presents the Raman spectra of MoO3/MgO samples. The Raman band at 800 cm −1 was characteristic in all samples and also appeared in the Raman spectra of Li/MgO in (Figure 8b). This band was not observed in any of the Raman spectra of MgO reported in literature [48]. Thus, the presence of such a band in our catalysts should relate to the sol-gel synthesized MgO support. This band was used as an internal standard when comparing intensities of Raman bands corresponding to the presence of molybdena species.
Bare et al. studied the surface chemistry of magnesium oxide-supported molybdenum oxide [41]. They reported that the structure of the molybdate species on MgO support after calcination at 600 °C in dry air depends on the molybdenum loadings. Below 3.3 wt %, isolated distorted octahedrally-coordinated molybdate species are present, which upon exposure to water saturated air transform to tetrahedrally-coordinated species. Above 6.7 wt % molybdenum, octahedrallycoordinated polymolybdate species are formed. Only above 20 wt % molybdenum is bulk magnesium molybdate formed from the reaction of MoO3 with MgO.  Similarly, the CO 2 desorption peak observed for the used 0.5MoO 3 /Li/MgO is significantly smaller than that of used Li/MgO. The broad desorption peak observed between 600 and 700 • C in the used catalysts is attributed to the desorption of weakly adsorbed CO 2 on [Li as illustrated earlier by Lunsford [18].

Raman Spectra
All MoO 3 /Li/MgO catalysts were characterized by Raman spectroscopy (Figure 8b). For better assignment of Raman bands, MgO promoted with comparable molybdena loadings was prepared and characterized (Figure 8a). At first, we assign the Raman bands of MoO 3 /MgO. Figure 8a presents the Raman spectra of MoO 3 /MgO samples. The Raman band at 800 cm −1 was characteristic in all samples and also appeared in the Raman spectra of Li/MgO in (Figure 8b). This band was not observed in any of the Raman spectra of MgO reported in literature [48]. Thus, the presence of such a band in our catalysts should relate to the sol-gel synthesized MgO support. This band was used as an internal standard when comparing intensities of Raman bands corresponding to the presence of molybdena species.
Bare et al. studied the surface chemistry of magnesium oxide-supported molybdenum oxide [41]. They reported that the structure of the molybdate species on MgO support after calcination at 600 • C in dry air depends on the molybdenum loadings. Below 3.3 wt %, isolated distorted octahedrally-coordinated molybdate species are present, which upon exposure to water saturated air transform to tetrahedrally-coordinated species. Above 6.7 wt % molybdenum, octahedrally-coordinated polymolybdate species are formed. Only above 20 wt % molybdenum is bulk magnesium molybdate formed from the reaction of MoO 3 with MgO. The Raman spectra of our MoO3/MgO catalysts show close resemblance to those of MoO3/MgO calcined at 600 °C and exposed to water-saturated air, reported by Bare et al. (see Table 3) [41]. The Raman band 445 cm −1 appearing in all MoO3/MgO samples is a characteristic for the lattice vibration and Mg-O stretching vibration in Mg(OH)2 [41]. The 0.5MoO3/MgO exhibited a Raman band at 908 cm −1 , which, with increasing the molybdena amount, increased in intensity and shifted to 912 cm −1 and 917 cm −1 for 3.3MoO3/MgO and 7.9MoO3/MgO, respectively. Thus, in agreement with Bare and coworkers [41], we assigned the Raman bands at 912 cm −1 , 874 cm −1 , and 320 cm −1 in the 3.3MoO3/MgO and the bands at 917 cm −1 , 874 cm −1 , and 325 cm −1 in the 7.9MoO3/MgO to the symmetric, antisymmetric stretching, and bending modes, respectively, of terminal Mo-Ot in tetrahedrallycoordinated isolated MoO4 2− species. These assignments are summarized in Table 3. Usually in MoO3/MgO systems, a Raman band at 220 cm −1 is characteristic of heptamolybdates [41,49]. The absence of this band in all the spectra (Figure 8a,b) indicates the absence of well-defined polymolybdate species in our systems. In the Raman spectra of 4.5-8.9 wt % molybdenum supported on MgO, calcined at 600 °C in dry air, Bare and co-workers [41]   The Raman spectra of our MoO 3 /MgO catalysts show close resemblance to those of MoO 3 /MgO calcined at 600 • C and exposed to water-saturated air, reported by Bare et al. (see Table 3) [41]. The Raman band 445 cm −1 appearing in all MoO 3 /MgO samples is a characteristic for the lattice vibration and Mg-O stretching vibration in Mg(OH) 2 [41]. The 0.5MoO 3 /MgO exhibited a Raman band at 908 cm −1 , which, with increasing the molybdena amount, increased in intensity and shifted to 912 cm −1 and 917 cm −1 for 3.3MoO 3 /MgO and 7.9MoO 3 /MgO, respectively. Thus, in agreement with Bare and co-workers [41], we assigned the Raman bands at 912 cm −1 , 874 cm −1 , and 320 cm −1 in the 3.3MoO 3 /MgO and the bands at 917 cm −1 , 874 cm −1 , and 325 cm −1 in the 7.9MoO 3 /MgO to the symmetric, antisymmetric stretching, and bending modes, respectively, of terminal Mo-O t in tetrahedrally-coordinated isolated MoO 4 2− species. These assignments are summarized in Table 3. Usually in MoO 3 /MgO systems, a Raman band at 220 cm −1 is characteristic of heptamolybdates [41,49]. The absence of this band in all the spectra (Figure 8a,b) indicates the absence of well-defined polymolybdate species in our systems. In the Raman spectra of 4.5-8.9 wt % molybdenum supported on MgO, calcined at 600 • C in dry air, Bare and co-workers [41] (Figure 8b). These were absent in the corresponding 7.9MoO 3 /MgO catalyst. To investigate the nature of these phases we subtracted the Raman spectrum of the 7.9MoO 3 /MgO catalyst from the corresponding spectrum of the 7.3MoO 3 /Li/MgO catalyst. Figure 9 shows the difference spectrum. The majority of the bands appearing in the difference spectrum of the 7.  Additionally, the difference spectrum shows two sharp Raman bands at 289 cm −1 and 891 cm −1 . The quasi-binary phase diagram of Li2O and MoO3 reported in literature [53] shows that in MoO3rich conditions (MoO3/Li2O > 0.5) at T > 600 C, the formation of Li2Mo4O13 is favored. We also observed the formation of such species by XRD ( Figure 6). We tentatively assign the Raman bands at 289 cm −1 and 891 cm −1 to the bending and symmetric stretching modes of the terminal Mo-Ot bond in a well-defined Li2Mo4O13 phase, while other Raman bands expected for Li2Mo4O13 (160-420 cm −1 , 420-746 cm −1 , and 845-991 cm −1 ) [54] were not clearly detected. We hence suggest in catalysts with 7.1 wt % MoO3 the possible presence of Li2Mo4O13 polymolybdate species. Li2Mo4O13 is reported in literature [55] to have a regular derivative structure of V6O13, in which Mo is octahedrally-coordinated. Table 4 summarizes the Raman bands and corresponding molybdena phases in molybdena-promoted Li/MgO.
In conclusion, the molybdena-promoted Li/MgO catalyst contains three types of molybdate species, the formation of which depends on the molybdena loading. At low Mo/Li ratios, in both the 0.5MoO3/Li/MgO and 3.6MoO3/Li/MgO catalysts, isolated dispersed [MoO4] 2− anionic species were detected. At high Mo/Li ratios, in the 7.1MoO3/Li/MgO catalyst, the formation of crystalline lithium Additionally, the difference spectrum shows two sharp Raman bands at 289 cm −1 and 891 cm −1 . The quasi-binary phase diagram of Li 2 O and MoO 3 reported in literature [53] shows that in MoO 3 -rich conditions (MoO 3 /Li 2 O > 0.5) at T > 600 • C, the formation of Li 2 Mo 4 O 13 is favored. We also observed the formation of such species by XRD ( Figure 6). We tentatively assign the Raman bands at 289 cm −1 and 891 cm −1 to the bending and symmetric stretching modes of the terminal Mo-O t bond in a well-defined Li 2 Mo 4 O 13 phase, while other Raman bands expected for Li 2 Mo 4 O 13 (160-420 cm −1 , 420-746 cm −1 , and 845-991 cm −1 ) [54] were not clearly detected. We hence suggest in catalysts with 7.1 wt % MoO 3 the possible presence of Li 2 Mo 4 O 13 polymolybdate species. Li 2 Mo 4 O 13 is reported in literature [55] to have a regular derivative structure of V 6 O 13 , in which Mo is octahedrally-coordinated. Table 4 summarizes the Raman bands and corresponding molybdena phases in molybdena-promoted Li/MgO. In conclusion, the molybdena-promoted Li/MgO catalyst contains three types of molybdate species, the formation of which depends on the molybdena loading. At low Mo/Li ratios, in both the 0.5MoO 3 /Li/MgO and 3.6MoO 3 /Li/MgO catalysts, isolated dispersed [MoO 4 ] 2− anionic species were detected. At high Mo/Li ratios, in the 7.1MoO 3 /Li/MgO catalyst, the formation of crystalline lithium molybdate phases such as (i) monomeric Li 2 MoO 4 , and tentatively, (ii) polymeric Li 2 Mo 4 O 13 were detected. These lithium molybdate phases were present at low concentrations, below the detection limit of XRD and their formation became significant with increasing Mo/Li atomic ratio.

Discussion
The results presented above indicate that the promotion of Li/MgO with sub monolayer coverages of molybdena introduces structural changes, which influence both physical properties and the performance of the catalyst in the oxidative conversion of n-hexane. XRD and Raman spectra confirmed the presence of well-dispersed molybdena species in addition to lithium molybdates at high Mo/Li ratios, influencing catalyst performance, as discussed below.

Textural Properties and Stability of the Catalyst
The presence of Li 2 CO 3 , as confirmed by TPD results (Figure 7a) and previously reported [2,3,18], is an inherent property of Li/MgO. Trionfetti et al. [4] reported that only~40% of the lithium incorporates into the MgO lattice as [Li + O − ] sites during the sol-gel preparation of Li/MgO, whereas the rest stays as Li 2 O which forms Li 2 CO 3 via reaction with ambient CO 2 . The presence of Li 2 CO 3 leads to detrimental changes in surface area upon exposure of the catalyst to high temperatures, as observed from the BET results in Table 2. Molybdena-promoted Li/MgO catalysts, however, exhibit less presence of Li 2 CO 3 , clearly observed from the TPD and Raman results, and consequently maintain a higher surface area even after exposure to high temperature treatment (Table 2). Hence, we conclude that the presence of molybdena reduces the amount of Li 2 CO 3 in Li/MgO.
We suggest that during catalyst calcination at 600 • C, MoO 3 , formed as a result of the decomposition of ammonium molybdate, reacts with the Li 2 CO 3 present in Li/MgO, forming lithium molybdate mixed oxide phases. Indeed, the XRD pattern ( Figure 6) of an equimolar mixture of Li 2 CO 3 and (NH 4 ) 2 MoO 4 calcined at similar conditions during catalyst preparation (600 • C for 5 h) resulted in characteristic peaks of Li 2 MoO 4 and Li 2 Mo 4 O 13 . The absence of characteristic peaks of MoO 3 confirms the reactivity of MoO 3 with Li 2 CO 3 in the catalyst, leading to the formation of lithium molybdate phases [47]. In our catalysts, these molybdates, as detected by Raman in Figure 8b, were present as dispersed species at low Mo/Li ratios (0.5MoO 3 /Li/MgO and 3.6MoO 3 /Li/MgO) and well-defined crystalline Li 2 MoO 4 and tentatively Li 2 Mo 4 O 13 at high Mo/Li ratios.
In addition to Li 2 CO 3 , inherently present in the catalyst, Li 2 CO 3 is formed during oxidative reaction from the reaction of the product CO 2 with [Li + O − ] sites of Li/MgO. The formation of Li 2 CO 3 in Li/MgO during the catalytic reaction was confirmed by the TPD of the used catalyst in Figure 7b. Li 2 CO 3 sinters the catalyst at reaction temperatures [56], and consequently, this leads to degradation of defect sites and loss of catalytic activity.
The  (Figure 7b), for example, confirmed that the formation of Li 2 CO 3 was clearly partly suppressed. With less Li 2 CO 3 formation during catalytic reaction, the catalyst preserves its high surface area, and hence catalytic activity.

Activity and Selectivity
The 0.5MoO 3 /Li/MgO catalyst exhibited similar initial n-hexane conversion to Li/MgO, while catalysts with higher molybdena loading (3.6 and 7.1MoO 3 /Li/MgO) exhibited even lower n-hexane conversions than Li/MgO (Figure 1). Increase in molybdena loading led to a clear increase in selectivity to CO x , accompanied with a significant decrease in selectivity to C 2 -C 4 alkenes. In addition, a noticeable formation of hexene in the case of both 3.6 and 7.1 MoO 3 /Li/MgO catalysts was observed (Figure 3).
The product selectivity patterns observed over the molybdena-doped catalysts in Figure 3 suggest that three competing parallel reaction pathways exist: (1) surface-initiated gas-phase oxidative conversion of n-hexane, (2) oxidative dehydrogenation (ODH) of n-hexane, and (3) surface-enhanced deep oxidation.
Hereby, we briefly describe, based on previous literature findings, the mechanisms of the above-mentioned three reaction pathways and the significance of their occurrence at various Mo/Li ratios.
At low Mo/Li ratios, surface-initiated oxidative conversion of n-hexane was significant, leading to high selectivity to light alkenes. At these conditions dispersed molybena species were present, which did not contribute to catalytic activity. The conversion of n-hexane was initiated via heterolytic C-H bond scission and hexyl radical formation on the O 2− of low coordinated Mg 2+ LC O 2− LC sites, in analogy to the suggestion by Sclögl and co-workers [19,20]. It is well established in the oxidative conversion of alkanes over Li/MgO that alkane activation proceeds via surface radical initiation [2,[15][16][17][18][19][20]. The alkyl radical formed would undergo a series of gas-phase reactions in the presence of molecular oxygen, leading to the observed product composition for Li/MgO and 0.5MoO 3 /Li/MgO catalysts in Figure 3.
Adversely, high Mo/Li ratios led to the formation of crystalline Li 2 MoO 4 and Li 2 Mo 4 O 13 phases which enhanced the oxidative dehydrogenation of n-hexane to hexene and deep oxidation reactions. The ODH reaction occurs through subsequent hydrogen abstraction from the n-hexane via O 2− of MoO x [25,31,32]. This, in agreement with literature [38], was most significant in catalyst with 7.1 wt % MoO 3 , in which the presence of crystalline Li Figure 3.
In the ODH of light alkanes, ethane and propane, it is well established that reactions on molybdena follow the Mars-van-Krevelen redox mechanism; i.e., molybdenum is reduced via loss of lattice oxygen and is then re-oxidized via molecular oxygen [25,31,32]. Consequently, in our catalysts with high molybdena loadings, excessive redox reactions resulted in complete consumption of molecular oxygen, hence limiting the n-hexane conversion route via molecular oxygen and inhibiting the extent of gas-phase radical chain reactions. Therefore, although C-H bond scission from n-hexane and ODH activity increased with the increase in molybdena loading, overall n-hexane conversions remained lower than in Li/MgO and 0.5MoO 3 /Li/MgO catalysts.
The performance of 0.5MoO 3 /Li/MgO mimicked that of Li/MgO, with similar initial activity as well as selectivity ( Figure 4); moreover, the catalyst preserved activity with time-on-stream. The optimal Mo/Li ratio (Figure 2) is thus essential in order to, on one hand, maintain the non-redox nature of Li/MgO, and hence the desired high selectivity to alkenes (Figures 3 and 4), and on the other hand, to stabilize the catalyst activity ( Figure 1).

Catalyst Preparation
The MgO and Li/MgO catalysts used in this study were prepared according to the method described in detail earlier [4,7]. A methanol solution containing Mg(OCH 3 ) 2 (0.4 M) was mixed at room temperature with another methanol solution containing water (0.8 M) to form a sol. For Li/MgO, the required amount of LiNO 3 was added to the solution to obtain~1 wt % Li. The solution was allowed to stay for gelation for 24 h. The gel formed was dried at 50 • C in vacuum for 7 h, and calcined at 600 • C in a flowing air of 50 mL/min for 1 h with a heating rate of 5 • C/min. Modified MoO 3 /Li/MgO catalysts were prepared by wet impregnation of the sol-gel synthesized Li/MgO (calcined at 500 • C) using an aqueous solution of the ammonium molybdate. These were then dried at 50 • C in vacuum for 7 h and calcined at 600 • C in a flowing air of 50 mL/min for 5 h with a heating rate of 5 • C/min. Similarly, MoO 3 /MgO catalysts were prepared with wet impregnation of the sol-gel synthesized MgO. Molybdena-promoted samples are denoted as xMoO 3 /Li/MgO and xMoO 3 /MgO, where x is wt % of MoO 3 . Table 1 presents the list of the catalysts prepared.

Catalyst Characterization
The Brunauer-Emmett-Teller (BET) surface area of the catalyst was determined with nitrogen physisorption using a Micro-metrics Tristar instrument (Micro-metrics, USA).
X-ray Diffraction (XRD) patterns were recorded with a Philips PW 1830 diffractometer (Philips, Netherlands) using Cu Kα radiation, λ = 0.1544 nm. Spectra were registered in the 2θ range of 35 to 50 with step size of 0.01 and integration time of 1 s per step. Elemental composition of the catalysts was determined with atomic absorption spectroscopy (AAS). Li content in all the catalysts was 0.86 wt %.
Temperature programmed desorption (TPD) experiments were performed to decompose the Li 2 CO 3 inherently present in both fresh and used catalysts. A total of 100 mg of the catalyst was pretreated in O 2 /He at 600 • C for one hour to decompose any MgCO 3 present. After cooling the catalyst to 100 • C in helium, TPD was conducted from 100 • C to 950 • C, with an increment of 10 • C/min, with helium flow of 10 mL/min as a carrier gas.
Raman spectral measurements were conducted with a SENTERRA instrument (Bruker Optics, Netherlands) equipped with a cooled charge-coupled device (CCD) detector (−60 • C). The samples were excited with a 785 nm red laser of 100 mW power. Spectra were recorded at room temperature from 100 to 1000 cm −1 , at a resolution of 3 cm −1 and a 5 min integration time.

Catalytic Tests
The catalytic tests were carried out at atmospheric pressure and isothermal conditions in a fixed-bed reactor [6]. An alumina tube reactor of 4 mm internal diameter was used. Powder catalyst was pressed, crushed, and sieved to particle size of 0.4-0.6 mm before use. A total of 10-100 mg of catalyst sample (to obtain varying contact times) was diluted in quartz particles and placed in the isothermal region of the furnace (1 cm). An alumina rod of 3 mm internal diameter was placed right below the catalytic bed to reduce the post catalytic volume in order to minimize homogenous gas-phase reactions. A Chromel-Alumel thermocouple inside a quartz tube was inserted above the catalytic bed to record reaction temperature. The temperature of the furnace was controlled by a second thermocouple placed outside the reactor tube within the isothermal zone of the tubular furnace.
Reactions were studied at 575 • C. Feed (100 mL/min) consisted of 10 mol % of n-hexane vapor, 8 mol % of oxygen, and balance helium. Similar n-hexane conversions were achieved by varying the weight hourly space velocity (WHSV). Before each catalytic test, the catalysts were pretreated at 625 • C in 50% O 2 /He (60 mL/min) for 1 h. For analysis of the product, samples of outlet gas stream were injected into two micro gas chromatographies (micro GCs) (Varian CP4900, Netherlands) every 30 min during a period of 6 h. The first micro GC was a quad system consisting of four channels for the separation of O 2 , N 2 , CH 4 CO, CO 2 , H 2 O, C 2 -C 4 hydrocarbons (alkanes and alkenes). The second micro GC was a dual system consisting of two channels for the separation of He, H 2 , and C 6 -C 8 hydrocarbons (alkanes and alkenes).
n-Hexane conversions were calculated on carbon mol basis; i.e., (C 6 in moles − C 6 out moles) / C 6 in moles × 100%. The carbon balance closed between 100% and 105%. Selectivity to individual products was also calculated based on the number of moles of carbon contained in the products, divided by the total number of moles of carbon in the product mixture excluding unconverted feed; i.e., (n i C i / n i C i ) × 100%

Conclusions
Doping of Li/MgO catalyst with molybdena improves the performance in oxidative conversion of n-hexane. Promotion of Li/MgO with molybdena loading as low as 0.5wt % is sufficient to bring considerable improvements in the stability of the catalyst. Moreover, the catalyst exhibits similar activity for n-hexane conversion and selectivity to alkenes as the unpromoted Li/MgO.