The nickel-support interaction as determining factor of the selectivity to ethylene in the oxidative dehydrogenation of ethane over nickel oxide/alumina catalysts

Applied Catalysis A, General 623 (2021) 118242 Available online 6 June 2021 0926-860X/© 2021 Elsevier B.V. All rights reserved. The nickel-support interaction as determining factor of the selectivity to ethylene in the oxidative dehydrogenation of ethane over nickel oxide/ alumina catalysts Yousra Abdelbaki a , d , Agustín de Arriba a , Benjamín Solsona b , *, Daniel Delgado a , e , Ester García-Gonz´ alez c , Rachid Issaadi d , Jos´ e M. L´ opez Nieto a , * a Instituto de Tecnología Química, Universitat Polit` ecnica de Val` encia-Consejo Superior de Investigaciones Científicas, Avenida de los Naranjos s/n, 46022, Valencia, Spain b Departament d′Enginyeria Química, Universitat de Val` encia, C/ Dr. Moliner 50, 46100, Burjassot, Valencia, Spain c Departamento Química Inorg´ anica, Facultad de Ciencias Químicas, Universidad Complutense, 28040, Madrid, Spain d Hydrogen Energy Applications Laboratory, Faculty of Technology, University of Blida 1, Algeria e Department of Inorganic Chemistry, Fritz-Haber-Institute of the Max-Planck-Gesellschaft, 14195, Berlin, Germany 1 ARTICLE INFO Keywords: Ethylene Oxidative dehydrogenation of ethane ODH Nickel oxide Alumina Catalyst characterization (XPS, HREM, UV raman and UV–vis-DRS) ABSTRACT Nickel oxides supported on γ-alumina (Ni-loading from 5 to 30 wt% NiO) have been synthesized and tested in the oxidative dehydrogenation (ODH) of ethane in order to determine the importance of the NiO-support interaction. The best performance was achieved by the catalyst with 15 wt% NiO; higher NiO-loadings lead to the formation of unselective bulk-like NiO and lower Ni-loadings present high proportion of free alumina surface sites. The presence of oxalic acid and/or niobium in the synthesis gel resulted in the formation of NiO particles with similar size, but higher crystallinity and reducibility than the standard 15 wt% NiO catalyst. The obtained results have revealed that, in addition to NiO crystal size, the nickel oxide-support interaction determines the catalytic performance of these catalysts. 1. Introduction The recent discovery of shale gas reserves has caused a decrease in the price of natural gas, encouraging its use as feedstock for the production of valuable chemicals [1]. Ethylene is, by far, the most important chemical feedstock for Petrochemistry, being directly used in the production of a wide number of commodity chemicals [2]. Currently, the main industrial route to obtain ethylene is steam cracking, which is an energy intensive and non-catalytic process [2,3]. In fact, steam cracking is considered as the most energy consuming process of the chemical industry, due to its endothermic character and the need for high reaction temperatures. Moreover, the absence of catalysts leads to the formation of many reaction products so that the separation costs are also high. Among all the possibilities, the oxidative dehydrogenation (ODH) of ethane is established as one of the most interesting alternatives to steam cracking, being an exothermic process with lower energy requirements and no thermodynamic limitations [3–5]. The energy consumption of ODH is expected to be substantially lower than any of the current alkene production technologies due to its exothermic nature. Furthermore, the deposition of coke is prevented provided that the presence of oxygen can oxidize coke to form carbon oxides. Despite the large amount of research efforts, industrial scale application of the ODH of ethane has not been implemented up to date due to the relatively low ethylene selectivity shown by the catalysts currently available. The main problem with most of the catalysts studied for the ODH of ethane is the excessive formation of carbon oxides (CO x ) which limits the selectivity to ethylene [3–6]. In this sense, among all catalytic systems based on reducible metal oxide catalysts, the most promising ones are multicomponent MoV(Te,Sb)NbO mixed oxides [7,8] and modified NiO materials [9–21]. In the latter case, it has been reported that pure NiO exhibits an important formation of CO 2 and low ethylene selectivity [9–20]. However, the role of promoters [9–15] and/or supports/diluents [16–25] in NiO based catalysts is still under discussion. In this way, several investigations have shown the effect of different promoters on * Corresponding author. E-mail addresses: [email protected] (B. Solsona), [email protected] (J.M. L´ opez Nieto). 1 Present address. Contents lists available at ScienceDirect Applied Catalysis A, General journal homepage: www.elsevier.com/locate/apcata https://doi.org/10.1016/j.apcata.2021.118242 Received 16 April 2021; Received in revised form 26 May 2021; Accepted 3 June 2021 Applied Catalysis A, General 623 (2021) 118242 2 the catalytic behavior of NiO-based materials [9–25], with the presence of many promoters reducing the formation of carbon dioxide. Nevertheless, at this moment, Nb-promoted catalysts are the most effective ones [9–15], although the presence of other elements, such as Sn 4+ , W 6+ , Zr 4+ , Ti 4+ [10–15], with dopant contents lower than 10 at.%, favor small changes in the characteristics of NiO particles, thus leading to the best catalytic performance (selectivity to ethylene up to 80–90 %). Alternatively, supported/diluted NiO catalysts also show a high ethylene selectivity, especially by using Al 2 O 3 [16–19], porous clays [20] or other supports based on transition metal oxides [21–25]. In this case, after the incorporation of NiO contents of ca. 10−30 wt%, changes in both physico-chemical characteristics and catalytic performance are observed. These changes have been related to a decrease in crystal size of NiO particles but, in addition, some interaction between NiO and the support (decreasing the reducibility of Ni-O bonds) could be also necessary. Thus, it is known that the decrease of the NiO crystallite size [9,21], the elimination of non-stoichiometric oxygen species, a decrease in the reducibility of Ni n+ sites, an increased Lewis acidity [15,26,27] or a lower electron conductivity [28] can give rise to high selectivity to ethylene during the ODH of ethane [9,11,12,17]. Interestingly, a similar increasing effect in the selectivity to the olefin in the ODH of ethane has been observed when oxalic acid is incorporated during the preparation procedure [15]. Generally, nickel oxide catalysts with small NiO particle size (below 20 nm, although less than 10 nm is preferred) present optimum catalytic performance in the ODH of ethane [15]. However, the drastic change observed in the redox properties of NiO must rely on the modification of the chemical nature of NiO (coordination, surface environment, oxidation state) [9,15]. This could be achieved by decreasing NiO particle size. However, it is possible to increase the selectivity to ethylene by optimizing the active phase-support interaction, without substantially decreasing NiO particle size, as observed in TiO 2 -supported nickel oxide catalysts [28]. Thus, and according to these observations, it seems that a small particle size might not be a sufficient requirement to improve the catalytic performance of NiO-based catalysts in ODH. In order to shed some light into the chemical nature of selective NiO catalysts for the oxidative dehydrogenation of ethane, we have synthesized Al 2 O 3 -supported nickel oxide catalysts, but with varying degrees of nickel oxide-support interaction by using modifying the catalyst preparation procedure. In this way, it will be possible to determine the influence of the interaction between nickel oxide and the support, minimizing the possible interference of both the NiO crystal size effect and the support employed. Accordingly, we have followed two different synthetic approaches for a series of NiO/Al 2 O 3 catalysts: i) addition of oxalic acid as an organic additive to NiO/Al 2 O 3 system; and ii) incorporation of Nb 5+ as a dopant during the preparation of the Al 2 O 3 -supported materials. In the latter case, the synthesis of Nb 5+ -promoted NiO/Al 2 O 3 catalysts has been carried out by incorporating Nb 5+ in one or two steps: a) using a Ni 2+ /Nb 5+ -containing solution to be directly impregnated on Al 2 O 3 ; and b) Nb 5+ is firstly impregnated on Al 2 O 3 , and a Ni 2+ -containing solution is subsequently added on the NbO x -Al 2 O 3 support. The results are discussed in terms of the specific chemical and structural features found in selective and unselective materials. 2. Experimental 2.1. Catalysts preparation Al 2 O 3 -supported NiO catalysts were prepared by wet impregnation of γ-Al 2 O 3 (S BET =210 m 2 /g, ABCR) with aqueous solutions of nickel nitrate Ni(NO 3 ) 2 ⋅6H 2 O (Sigma-Aldrich, 99 %). The catalysts are named as xNi/AL, where x is the NiO wt%. Alternatively, oxalic acid was added to the nickel nitrate solution, with Ni/oxalic acid molar ratios of 1/1 and 1/3 (i.e. 15Ni/AL-o1 and 15Ni/AL-o3 catalysts, respectively). For comparison, a 15 NiO wt% Al 2 O 3 catalyst was prepared by a mechano-chemical procedure, by mixing and grinding the corresponding amounts of nickel oxide and alumina in an agate mortar (i.e. NiþAL(PM) sample). Nb-containing alumina-supported nickel oxide catalysts were prepared by following two strategies, using an aqueous solution of C 4 H 4 NNbO 9 ⋅xH 2 O (Sigma–Aldrich): i) direct impregnation of γ-Al 2 O 3 by an aqueous solution of promoting compounds and subsequent impregnation with an aqueous solution of nickel nitrate (15 wt% NiO) named as Ni/Nb/AL; ii) γ-Al 2 O 3 -supported Ni-Nb-O mixed oxides were prepared by wet impregnation method using aqueous solutions of nickel nitrate and the niobium compound, with a Nb/(Ni +Nb) atomic ratio of 0.1; which has been named as (NiþNb)/AL. All catalysts were dried overnight at 100 ◦C and finally calcined at 500 ◦C for 2 h (5 ◦C/min). The characteristics of these catalysts are shown in Table 1. 2.2. Catalytic tests Catalytic tests were carried out under steady state conditions in a fixed bed quartz reactor (i.d. 20 mm, length 400 mm) at temperatures in the 300−500 ◦C range. Feed consisted of an ethane/O 2 /He mixture with 3/1/26molar ratio. The total flow and the catalyst weight were varied (25−100 mL min −1 , 0.1–1.0 g of catalyst and 0.3–0.5 mm particle size) in order to achieve several contact times (W/F). For some selected experiments an ethane/O 2 /He mixture with 3/3/24molar ratio was employed. Reactants and products were analyzed by gas chromatography using two packed columns [20]: (i) molecular sieve 5A (2.5 m); and (ii) Porapak Q (3 m). 2.3. Catalysts characterization N 2 -adsorption isotherms were recorded in a Micromeritics ASAP 2000 device. The materials were degassed in vacuum at 300 ◦C prior to N 2 adsorption. Surface areas were estimated by the Brunauer-EmmetTeller (BET) method. X-ray diffraction patterns were collected in an Enraf Nonius FR590 diffractometer with a monochromatic CuK α 1 source operated at 40 keV and 30 mA. Raman spectra were obtained in an inVia Renishaw spectrometer, equipped with an Olympus microscope, using a wavelength of 325 nm (UV-Raman), generated with a Renishaw HPNIR laser with a power of approximately 15 mW. UV–vis diffuse reflectance spectroscopy measurements of the solids were carried out within the 200−800 nm range using a Varian spectrometer model Cary 5000. The value of band gap Eg is calculated by extrapolating the linear fitted region at [F(R(∞))h υ ] 2 =0 in the plot of [F(R(∞)) h υ ] 2 versus h υ . Additional information in supporting information. Temperature-programmed reduction experiments (H 2 -TPR) were performed in an Autochem 2910 (Micromeritics) equipped with a TCD detector. The reducing gas composition was 10 % H 2 in Ar, with a total flow rate of 50 mL min −1 . The materials were heated until 800 ◦C, with a heating rate of 10 ◦C min −1 . X-ray photoelectron spectroscopy (XPS) measurements were performed on a SPECS spectrometer equipped with a Phoibos 150 MCD-9 detector using a monochromatic Al K α (1486.6 eV) X-ray source. Spectra were recorded using an analyzer pass energy of 50 eV, an X-ray power of 100 W, and an operating pressure of 10 −9 mbar. Spectra treatment was performed using CASA software. Binding energies (BE) were referenced to C 1s at 284.5 eV. Selected area electron diffraction (SAED), high-resolution transmission electron microscopy (HRTEM) and Scanning-TEM (STEM)-Energy-Dispersive Spectroscopy (EDS) maps were performed on a JEOL JEM300 F electron microscope by working at 300 kV (point resolution of 0.17 nm). Crystal-by-crystal chemical microanalysis was performed by energy-dispersive X-ray spectroscopy (XEDS) in the same microscope Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 3 equipped with an ISIS 300 X-ray microanalysis system (Oxford Instruments) with a detector model LINK “Pentafet” (resolution 135 eV). Samples for transmission electron microscopy (TEM) were ultrasonically dispersed in n-butanol and transferred to carbon coated copper grids. 3. Results and discussion 3.1. Catalytic results in the ODH of ethane The catalytic performance of supported nickel oxide catalysts during the oxidative dehydrogenation (ODH) of ethane at 400 ◦C is summarized in Table 1. As mentioned in the experimental section, the catalysts tested consist of a set of nickel oxide materials supported on γ-alumina, with Ni-loadings from 5 to 30 wt% NiO. In addition, catalysts with a 15 wt% NiO prepared with different amounts of oxalic acid and promoted with Nb 5+ were also evaluated. The ethane conversion and the selectivity to ethylene strongly depend on NiO-loading and the catalyst preparation procedure. For comparison, it has been also included the catalytic results of a mechanochemical NiO-Al 2 O 3 mixture, named as NiþAL(PM). In all cases, ethylene was the main reaction product. In addition, CO 2 was the only product detected from oxidation reactions. The variation of the selectivity to ethylene with the ethane conversion during the ethane ODH on representative Al 2 O 3 -supported nickel oxide catalysts (xNi/AL series) is shown in Fig. 1a. On the other hand, Fig. 1b allows the comparison of catalytic results over catalysts with 15 wt% NiO with or without additional synthetic modifications: i) Nbpromoted catalysts, i.e. Ni/Nb/AL and (NiþNb)/AL samples; ii) unpromoted catalysts prepared from synthesis gels containing oxalic acid (i.e. 15Ni/AL-o1 and 15Ni/AL-o3 samples). As it can be observed in Fig. 1a and Table 1, the selectivity to ethylene has a clear dependence on the NiO-loading. Thus, the selectivity to ethylene gradually increases with NiO-loading (in the range of 88–95 %), reaching a maximum of ca. 95 % over sample 15Ni/AL (i.e. 15 wt% NiO). On the contrary, further increasing NiO-loading on Al 2 O 3 (up to 30 wt% NiO) leads to a decrease in the selectivity to ethylene during the ODH of ethane (down to ca. 72 %). In the case of Nb-promoted catalysts, the (NiþNb)/AL sample (synthesized in one step) presents a slight decrease in the selectivity to ethylene (ca. 86 %) compared to that achieved using the reference sample, 15Ni/AL. Moreover, Ni/Nb/AL catalyst (prepared in two steps) presents a remarkably lower selectivity (ca. 71 %). The method used to prepare the catalysts, and especially the presence of oxalic acid in the synthesis gel, may also have an important influence in the catalytic performance. Fig. 1b shows the variation of selectivity with ethane conversion of samples with 15 wt% of NiO. Lower selectivity to ethylene is observed over samples prepared with oxalic acid in the synthesis gel (see 15Ni/AL-o1 and 15Ni/AL-o3), especially notorious when high oxalic acid contents are used. Thus, both Nb-promoted NiO/γ-Al 2 O 3 catalysts and Nb-free NiO/γ-Al 2 O 3 catalysts Table 1 Characteristics of catalysts and their catalytic results in the ODH of ethane. Catalyst NiO (wt.%) S BET (m 2 g −1 ) TPR (H 2 - uptake) e Eg (eV) f Ethane conversion (%) Selectivity to ethylene (%) Selectivity to ethylene (%) g Catalytic activity h Catalytic Activity/NiO i STY rate of product formation j 5Ni/AL 5 202 17 4.50 0.92 88 88 14.6 293 13.7 10Ni/AL 10 189 49 4.08 1.2 94.7 88 19.1 191 16.7 15 Ni/AL 15 171 71 3.81 3.4 95.0 93 54.1 361 47.3 20 Ni/AL 20 166 86 3.70 4.4 91.5 90 70.1 350 59.8 30 Ni/AL 30 152 126 3.60 7.7 70.9 72 123 409 81.1 Ni/Nb/AL 15 151 77 3.58 2.1 78.6 71 33.4 223 24.5 (Ni+Nb)/ AL 15 184 63 3.50 4.5 88.5 86 71.6 478 59.2 15Ni/ALo1 b 15 195 50 3.25 6.8 78 78 108 722 78.8 15Ni/ALo3 c 15 195 72 3.55 5.3 63.6 63 84.4 563 50.1 Ni+AL (PM) d 15 173 65 3.48 8.6 39.4 39 137 913 50.4 a) At 400ºC and a contact time, W/F, of 20.5 g cat h mol C2H6 −1 ; b) Oxalic acid/nickel molar ratio in synthesis gel of 1; c) Oxalic acid/nickel molar ratio in synthesis gel of 3; d) mechano-chemical mixture; e) hydrogen-uptake during the TPR experiments, in mL H2 /g cat ; f) Edge energy in eV, determined from DR-UV–vis spectra; g) Selectivity to ethylene at 400ºC and isoconversion of 10 %; h) Catalytic activity (in g C2H6 kg cat −1 h −1 ); i) Catalytic activity per loading of NiO (in g C2H6 kg NiO −1 h −1 ); j) Space-time yield, STY, is the formation rate of ethylene (in g C2H4 kg cat −1 h −1 ). Fig. 1. Variation of the selectivity to ethylene with the ethane conversion on: a) xNi/AL series; and b) on catalysts with 15 wt% NiO, with or without additional synthetic modifications. Note: reaction conditions in the main text, reaction temperature 400-450 ◦C. Catalysts: 5Ni/AL ( ); 15Ni/AL ( ); 30Ni/AL ( ); 15Ni/AL-o1 (); 15Ni/AL-o3 ( ); (NiþNb)/AL ( ). Data corresponding to the mechano-chemical mixture, NiþAL(PM) (), have been included for comparison. Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 4 (where the oxalic acid has been directly added) show a poorer selectivity to ethylene than the corresponding catalyst of xNi/Al series (i.e. 15Ni/ AL sample). The selectivity to ethylene for catalysts prepared with a 15 wt% NiO decreases according to the following trend: 15Ni/AL > (NiþNb)/AL >15Ni/AL-o1 >Ni/Nb/AL >15Ni/AL-o3 >NiþAL (PM). Fig. 2 shows the variation of the space-time yield for ethylene formation of studied catalysts. For xNi/AL series, the formation of ethylene per nickel site increases with the NiO-loading, achieving its maximum value for 20Ni/AL catalyst (Fig. 2). However, the highest space-time yield for ethylene formation was observed for samples 15Ni/AL-o1 and (NiþNb)/AL (Fig. 2). This is so because of the higher C 2 H 6 reaction rates for the latter cases (Figure S1, supporting information). The present study has been undertaken using a low concentration of oxygen (C 2 H 6 /O 2 /He =3/1/26 M ratio) so achieving high conversions without running out of oxygen is complicated. Then, we have carried out a few experiments with the optimal catalyst 15Ni/AL using more oxygen in the feed (C 2 H 6 O 2 /He =3/3/24) and higher contact times to reach higher conversions. Thus, an ethane conversion of 51.2 % and a selectivity to ethylene of 71.7 %, so that the ethylene yield obtained was 36.7 %, was obtained when working at 450 ◦C and a contact time, W/F, of 205 g cat h (mol C2H6 ) −1 (Table S1, supporting information). This yield could be enhanced by optimizing the reaction conditions. 3.2. Characterization of Al 2 O 3 -supported nickel oxide catalysts In order to understand their catalytic performance, the catalysts were characterized by several physicochemical techniques. Table 1 summarizes their main physicochemical characteristics. Fig. 3 displays the XRD patterns of Al 2 O 3 -supported NiO catalysts synthesized in the absence or the presence of increasing amounts of oxalic acid or Nb-oxalate in the synthesis gel. For comparison, XRD pattern of the mechano-chemically mixed NiþAL(PM) sample is presented in Figure S2-A (supporting information). All the materials show diffraction lines corresponding to facecentered cubic NiO phase, space group Fm-3m (JCPDS: 78−0643), and γ-alumina (JCPDS: 10−0425). However, depending on the catalyst preparation procedure (the NiO-loading and/or the presence/absence of Nb 5+ as promoter), differences in the relative intensity of diffraction maxima of the two crystal phases were observed. X-ray diffraction patterns of the catalysts from xNi/AL series show a progressive increase in the relative intensity of NiO diffraction maxima as the nickel oxide loading increases (see Fig. 3, patterns a to e). For the same nickel oxide content (we have considered the optimal Ni/Al ratio), the incorporation of increasing amounts of oxalic acid leads to an increase in the relative intensity of the NiO peaks (Fig. 3, diagrams c, h and i). Likewise, both the use of niobium as a promoter and the method followed to incorporate it, modify the relative intensity and peak width of NiO maxima, for a fixed Ni/Al content (again, we have considered the optimal NiO content of 15 wt%). Thus, NiO diffraction maxima appear to be narrower in Ni/Nb/AL sample than in (NiþNb)/AL (Fig. 3, patterns f and g, respectively), which could be related to differences in NiO crystal size and/or degree of crystallinity, depending on the extent of interaction with γ–Al 2 O 3 support. In order to get further insight into the crystalline nature of NiO-based catalysts, these materials were analyzed by high-resolution transmission electron microscopy. Catalysts of the xNi/AL series are made up of γ–Al 2 O 3 and NiO crystallites of 5−10 nm in size up to the composition 15Ni/AL, NiO showing a good dispersion over the support (Figs. 4a and 4b). In the 20Ni/AL catalyst, platelet-like NiO crystals of 30−50 nm begin to appear (Fig. 4c), which are very abundant, and in the form of agglomerates in 30Ni/AL sample (Fig. 4d). The addition of oxalic acid tends to modify the microstructure of the catalysts. After adding oxalic acid in 1:1 Ni/oxalic acid molar ratio (catalyst 15Ni/AL-o1), crystallinity, the degree of dispersion on the support and the average size of the NiO crystallites, remain almost unchanged with respect to that observed in catalyst 15Ni/AL. However, for 1: 3 Ni/oxalic acid molar ratio (i.e. 15Ni/AL-o3 sample), NiO particles display higher crystallinity, even though the average crystallite size hardly changes (Fig. 4e). Note that despite the similarity between X-ray patterns of 15Ni/AL-o3 and 30Ni/AL catalysts (Fig. 3, patterns i and e, respectively), their microstructure is quite different. In the case of niobium promoted catalysts, microstructural differences are found depending on the preparation method. The Nbpromoted NiO/γ-Al 2 O 3 catalyst prepared in two steps (Ni/Nb/AL sample) is formed by NiO crystallites of 10−15 nm distributed on the support, as well as agglomerates of large crystals of ~ 50 nm. EDS maps (Figure S3) show a homogeneous distribution of niobium on the γ-Al 2 O 3 support and the absence of niobium in areas where NiO crystals are observed. Interestingly, well-dispersed crystallites of approximately 5−10 nm size with a characteristic d spacing of approximately 7 Å can be also observed in this catalyst. Provided the chemical composition of the sample and the elemental distribution, this periodicity is compatible with the (002) d spacing of Ni 2 O 3 (d 002 =7.29 Å). These crystallites are clearly visible in Fig. 4f. It is important to mention that this is the only catalyst in which this type of crystals have been observed, and the details of the preparation method in this particular case must be the origin of the formation of this species. For Nb-promoted NiO/γ-Al 2 O 3 prepared in one step, (NiþNb)/AL, NiO crystallites of 10−15 nm distributed on the support with agglomerates of larger NiO crystals of ~ 50 nm were observed. These large crystals are less abundant than in the catalyst prepared in two steps. EDS maps performed (Figure S4) show that, although in low concentration, niobium is effectively distributed and associated with nickel on the support. Al 2 O 3 -supported NiO catalysts were also characterized by UV Raman spectroscopy (using an excitation wavelength of 325 nm) in order to elucidate the spin-phonon interaction in the materials [29,30]. Fig. 5 shows the UV Raman spectra of alumina-supported nickel oxide catalysts. For comparison, the UV Raman spectrum of NiþAL(PM) sample is presented in Figure S2 (pattern B). As reported in the literature, the UV Raman spectra of bulk NiO are characterized by the presence of five Raman bands [29,30]: i) two bands at ca. 510 and 580 cm −1 , assigned to one-phonon (1P) transverse optical (TO) and longitudinal optical (LO) modes; ii) two weak bands at ~740 cm -1 and ~900 cm −1 and an intense band at ca. 1100 cm −1 related to two-phonon modes, i.e. the second-order transverse optical mode (2TO), the combination of TO +LO modes and the second-order longitudinal optical (2LO) modes, respectively. Among all these bands, the most intense ones are those located at 580 and ca. 1100 cm −1 . We must notice that, when NiO is antiferromagnetically ordered or defect-rich, the intensity of one-phonon scattering (1P, LO and TO modes) increases significantly [29,31]. In addition, a very low intensity Fig. 2. Variation of the space time yield ethylene formation rate per nickel oxide site (vertical bars) on NiO/Al 2 O 3 catalysts of the xNi/AL series and catalysts with 15 wt% NiO with additional synthetic modifications. Reaction conditions: 400 ◦C, different contact times and C 2 /O 2 /He =3/1/26 M ratio. Variation of selectivity to ethylene at an ethane conversion of 10 % (reaction temperature, 400 ◦C) ( ). Note: Selectivity of Ni þAL(PM) is very low (ca. 39 %). Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 5 of the band at ca. 1124 cm −1 has been observed in silica-supported nickel oxide (with 3 wt% NiO), which has been attributed to the presence of very small NiO crystals [32]. In addition, the presence of only one band at ca. 570 cm -1 has been recently reported for NiO supported on Nb 5+ -containing siliceous porous clay heterostructure catalysts [33]. This observation was attributed to a high NiO dispersion as a consequence of an effective active phase-support interaction, what would lead to a decrease in NiO particle size and/or the generation of defects. As expected, the intensity of the most characteristic bands (1P LO band at ~580 cm −1 and 2P 2LO band at ~1100 cm −1 ) increases when the nickel oxide loading increases (Fig. 5). However, the relative growth of both bands differs depending on the specific structural and chemical features of the catalyst. Thus, an increase in the intensity of 2P 2LO band (I 1100 ) higher than that of 1P LO band (I 580 ) is observed at high NiOloading. Therefore, the relative intensity of 1P LO band simultaneously increases with the decrease of NiO particle size and/or the presence of oxygen defects [31]. This fact would mean that the increase in the nickel oxide loading favors an increase of the NiO crystal size and/or a decrease in the concentration of oxygen defects. UV Raman spectra of catalysts with the same NiO-loading (with or without Nb 5+ or oxalic acid in the synthesis gel) are comparatively shown in Fig. 5 (spectra c, f to i). Differences are observed in the relative intensity of the LO band (I 580 ) and 2LO band (I 1100 ) depending on the catalyst preparation procedure. Catalysts 15Ni/AL and (NiþNb)/AL present an I 580 /I 1100 ratio higher than 1; whereas Ni/Nb/AL, 15Ni/ALo1 and 15Ni/AL-o3 catalysts present I 580 /I 1100 ratios lower than 1. These results could be partly explained in terms of the presence of NiO particles with different crystal sizes. However, the distribution and the mean crystal size of NiO in the reference catalyst (15Ni/AL) and sample 15Ni/AL-o1 are very similar, despite being the relative I 580 /I 1100 ratio much higher in the reference catalyst. In the same way, Nb-promoted NiO/Al 2 O 3 catalysts prepared in either one or two synthesis steps ((NiþNb)/AL and Ni/Nb/AL, respectively) also present significant differences in their UV Raman profiles (Fig. 5, spectra f and g, respectively). Despite showing similar NiO crystal size, both catalysts display different relative I 580 /I 1100 ratio, being higher in the material prepared in one-step (i.e. (NiþNb)/AL sample). In addition, this would also underline the low capability of niobium oxide in the dispersion of NiO [34]. The DR-UV–vis spectra of prepared catalysts are shown in Fig. 6, in which spectra a to e correspond to Al 2 O 3 -supported NiO catalysts with different Ni-loadings, while spectra f to i are those corresponding to catalysts prepared in the presence of oxalate anions, Nb-promoted and unpromoted catalysts. For comparison, DR-UV–vis spectrum of NiOAl 2 O 3 mixture, NiþAL(PM) sample, has been also recorded (Figure S2C). Bulk NiO shows bands at 715 nm and 377 nm, which can be assigned to octahedrally coordinated Ni 2+ species in the NiO lattice [17,19,23,35, 36]. Additionally, a band at 510 nm can be also assigned to charge transfer in NiO crystals [36,37]. On the other hand, it has been reported that supported nickel oxide catalysts, such as NiO-Al 2 O 3 [38,39], NiO/Silica-Alumina [39] or NiO/Al 2 O 3 [17,19], can also present a doublet (at 600−645 nm) and a band at 416–430 nm, which were attributed to Ni 2+ species with tetrahedral (Td) and octahedral (Oh) coordination, respectively. Nevertheless, a band at 630 nm has been also observed in NiO supported on TiO 2 /γ-Al 2 O 3 [40], which has been attributed to the absorption of surface-dispersed nickel oxide species in tetrahedral coordination. We must note that the accommodation of Ni 2+ species in both tetrahedral and octahedral coordination could lead to nickel aluminate as surface spinel phase in samples calcined at higher temperatures [17,24,36,38–40]. According to our results, the DR-UV–vis spectra of samples of xNi/AL series with NiO-loading up to 10 wt% NiO suggest the presence of Ni 2+ species with tetrahedral (doublet at 600 and 640 nm) and octahedral (band at 410 nm) coordination, without the appearance of the band at 715 nm, typical of bulk NiO (Fig. 6 a–b). These results could indicate the existence of highly dispersed nickel species with high interaction between support and part of Ni-containing crystallites (Ni 2+ tetrahedral diffused into the γ-Al 2 O 3 lattice) [17,24,36,38,39]. At this point it is important to mention that NiAl 2 O 4 spinel was not detected neither by X-ray diffraction nor by electron microscopy. When nickel loading increases, the interaction with the support reaches a maximum (15 wt% NiO or below), and from there, NiO crystals begin to grow, reaching larger size. Then, the bands corresponding to Ni 2+ in octahedral coordination in NiO begin to be observed in the spectrum of the 15Ni/AL catalyst, thus indicating the simultaneous presence of nickel sites linked to the support as well as bulk NiO (Fig. 6, spectra c–e). The above analysis is in strong agreement with what was observed by electron Fig. 3. XRD patterns of Al 2 O 3 -supported NiO catalysts: a) 5Ni/AL; b) 10Ni/AL; c) 15Ni/AL; d) 20Ni/AL; e) 30Ni/AL; f) Ni/Nb/AL; g) (NiþNb)/AL; h) 15Ni/AL-o1; i) 15Ni/AL-o3. Diffraction lines of NiO ( ) and γ-Al 2 O 3 ( ) have been marked. Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 6 microscopy, where NiO crystals of 5−10 nm size are observed in 10Ni/AL and 15Ni/AL, in the last one co-existing with agglomerates of larger polygonal NiO crystals. In Nb-promoted catalysts, the nature of Ni 2+ -species depends on the catalyst preparation procedure. Spectrum of Ni/Nb/AL sample shows an intense band at 715 nm, Ni 2+ (Oh), in addition to a small band at 630 nm, Ni 2+ (Td) (Fig. 6, spectrum g). Thus, the presence of niobium species on the surface of support, as in the Ni/Nb/AL catalyst, limits the NiOsupport interaction, thus favoring NiO crystallization. However, the distribution of niobium associated to nickel in (NiþNb)/AL sample (one-step synthesis), limits the growth of NiO crystals as the band at 715 nm shows lower intensity (Fig. 6, spectrum f). Accordingly, agglomerates of large NiO crystals are more abundant in Ni/Nb/AL sample, as observed by electron microscopy. Bands corresponding to NiO at 377 and 715 nm are particularly intense in catalysts 15Ni/AL-o1 and 15Ni/AL-o3, indicating that crystallized NiO has a weak interaction with the support. This fact is in agreement with electron microscopy data, where an increase in the crystallinity of NiO is observed although the size of the crystallites does not increase significantly. Fig. 4. High resolution transmission electron micrographs corresponding to crystallites of catalysts: 15Ni/AL (a and b), 20Ni/AL (c), 30Ni/AL (d), 15Ni/AL-o3 (e), and Ni/Nb/AL (f). Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 7 Fig. 5. UV-Raman spectra of Al 2 O 3 -supported NiO catalysts: a) 5Ni/AL; b) 10Ni/AL; c) 15Ni/AL; d) 20Ni/AL; e) 30Ni/AL; f) (NiþNb)/AL; g) Ni/Nb/AL; h) 15Ni/ AL-o1; i) 15Ni/AL-o3. Fig. 6. Diffuse reflectance UV–vis spectra of Al 2 O 3 -supported NiO catalysts: a) 5Ni/AL; b) 10Ni/AL; c) 15Ni/AL; d) 20Ni/AL; e) 30Ni/AL; f) (Ni þNb)/AL; g) Ni/ Nb/AL; h) 15Ni/AL-o1; i) 15Ni/AL-o3. Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 8 Table 1 summarizes the energy band gaps (E g , in eV) of supported nickel oxide catalysts calculated from Kubelka-Munk function (Figure S5). In general, E g values decrease when increasing the Niloading. In the same way, for a fixed NiO loading (15 wt%), the lowest band gap values are observed in Nb-containing catalysts (3.50 and 3.58 eV) and catalysts prepared in the presence of oxalic acid in the synthesis gel (3.25 and 3.55 eV). Samples from xNi/AL series with NiO contents in the range 5−15 wt% display the highest E g among the samples analyzed (4.05−3.81 eV) (Table 1), which are also the most selective catalysts in the ODH of ethane. In our case, small differences are observed for all the catalysts, however, these band gap values alone cannot explain their catalytic properties as it may be influenced by various factors such as crystallite size, structural parameter, carrier concentrations, presence of impurities and lattice strain [41–44]. In order to study the reducibility of catalysts and NiO-support interaction, TPR-H 2 experiments were performed. Fig. 7 shows the TPR-H 2 profiles of xNiO/AL catalysts with different NiO-loading (Fig. 7, patterns a to e) and catalysts with 15 wt% NiO, with or without promoters, and synthesized by different preparation procedures (Fig. 7, patterns f to i). For comparative purposes, TPR-H 2 profile of a mechanochemical mixture, NiþAL(PM) sample (Fig. S2), is also included. The shape of reduction profiles depends on the strength of NiOsupport interaction. In the prepared catalysts, profiles show three main features corresponding to (in order of decreasing reduction temperature) (Fig. 7): i) the reduction of small and highly dispersed NiO particles for which the above interaction is strong [17–19,36], which gives a peak at ca. 520−550 ◦C (observed in all catalysts); ii) the reduction of NiO particles that tend to form small agglomerates with medium strength interaction with the support, that originates a reduction peak slightly above 450 ◦C (observed in sample with Ni-loading of 15 wt%); and iii) the reduction of large NiO crystals with weak interaction with the support, which gives a peak at ca. 340 ◦C, similar to that observed in pure NiO [17–25] (observed in sample 30Ni/AL). Additionally, differences in reducibility can be clearly observed as a function of the catalyst preparation procedure when comparing catalysts with 15 wt% NiO (Fig. 7, patterns c, f-i). In this sense, Nb-promoted catalysts show a peak of high reducibility at 330 ◦C, which is more intense and constitutes the main feature of the reduction profile in Ni/ Nb/AL. On the other hand, (NiþNb)/AL catalyst shows lower reducibility, with the main peak appearing at 530 ◦C. These results are consistent with the structural characterization. Thus, higher reducibility of Ni-species in Ni/Nb/AL catalyst can be interpreted on the basis of a low interaction between NiO and support (NbO x /Al 2 O 3 ), which facilitates the growth of crystals and the formation of agglomerates that are more easily reduced. In this way, it has been proposed that Nb 2 O 5 has not shown good properties as a NiO diluter/support, being unable in these conditions to eliminate a large proportion of non-selective sites [34]. In catalysts prepared with oxalic acid in the synthesis gel, the reduction profiles show a unique peak around 400−450 ◦C, in contrast with the xNi/AL series with Ni-loading below 15 wt% NiO, that presents a reduction peak at ca. 520 ◦C. According to TEM data, catalysts with 15 wt% of NiO and different amounts of oxalic acid (i.e. 15Ni/AL-o1 and 15Ni/AL-o3) present a similar size distribution of the NiO crystals, although crystallinity clearly improves compared to 15Ni/AL when increasing amounts of oxalic acid are used. A poor interaction with the support is the reason of this change as well as the decrease in reducibility. This is also in agreement with results from DR-UV–vis spectra. It is worth mentioning that nickel aluminate-like species present a reduction peak at ~ 800 ◦C, as reported in the literature [17,38]. This peak is not observed in the reduction profiles of the catalysts under study. This is in agreement with the structural and microstructural characterization of the samples, where the formation of NiAl 2 O 4 has not been observed in any case. However, the possible presence of this phase cannot be completely ruled out, although, if present, it should be in low concentration in samples with low NiO-loading. Fig. 7. TPR-H 2 profiles of Al 2 O 3 -supported NiO catalysts: a) 5Ni/AL; b) 10Ni/AL; c) 15Ni/AL; d) 20Ni/AL; e) 30Ni/AL; f) Ni/Nb/AL; g) (NiþNb)/AL; h) 15Ni/ALo1; i) 15Ni/AL-o3. Y. Abdelbaki et al. Applied Catalysis A, General 623 (2021) 118242 9 According to the TPR-H 2 experiments (Fig. 7), the catalyst 15Ni/AL presents the highest NiO-support interaction among the samples with a Ni-loading of 15 wt% NiO, also showing the maximum relative intensity of LO (1 P) band in UV Raman spectra (Fig. 5). Fig. 8 shows Ni 2p 3/2 core-level XPS spectra for selected Al 2 O 3 -supported NiO catalysts, whereas Figures S4 and S5 displays the XPS results of additional NiO-based materials. Ni 2p 3/2 core level spectra present the characteristic features of NiO, i.e. a main peak (ca. 856 eV) together with two satellites at 1.5–2.0 and 7.0 eV over the main peak (Sat I and Sat II, respectively) [13,15,33]. Sat I can be attributed to the presence of a wide variety of defects, such as Ni 2+ vacancies, Ni 3+ species or surface Ni 2+ -OH species; while Sat II is usually assigned to ligand-metal charge transfer. Changes in the relative intensity of Sat I signal have been related to the variations in the concentration of defects or in the particle size, which can be favored when an effective NiO-support interaction takes place [9,15,17–21,33]. Unfortunately, not a clear relationship between the relative intensity of Sat I / Main peak and the selectivity to ethylene has been observed. O 1s core level spectra are shown in Fig. 9 and Figure S6. In general, the O 1s signal of alumina appeared at higher binding energy (532.2 eV) than the signal for nickel oxide (530.5 eV) in all cases, as seen elsewhere [45]. The contribution of the alumina O 1s signal is bigger for the catalysts with Ni-loading lower than 20 wt% (Fig. 9), with a symmetric display of the peak. However, a shift of the band to lower binding energy is observed for catalysts with 30 wt% of NiO (Fig. S6), samples prepared with Nb (especially for Ni/Nb/AL sample (Fig. 9)) or a mechano-chemically treated sample (Fig. S6), in agreement with a worst dispersion of the NiO. A strong interaction of the oxygen anions with the Ni 2+ cations for these catalysts was also suggested as a shift to lower binding energy occurred. Al 2p and Ni 3p spectra for the catalysts with different Ni amount (Fig. 10), show a single peak at 74.8 eV for Al 2p, corresponding to the Al 3+ species in octahedral coordination, together with a signal at 68.5 eV, attributed to Ni 3p core level [45]. This latter Ni 3p core level peak increases in intensity as the NiO-loading increases. On the other hand, a small shift to lower binding energies when increasing the Ni loading is observed, likely associated with a progressively higher interaction of Al 3+ -bonded oxygen sites with Ni 2+ cations, resulting in a distortion of the Al 2 O 3 octahedral network [45]. In Nb 3d XPS spectra (Fig. S7), Nb-containing catalysts presented a classical doublet with a split spin-orbit of the components of 2.78 eV which is related to the unique presence of dispersed Nb 5+ [33]. 3.3. A determining factor of the selectivity to ethylene The TPR experiments undertaken confirm the close relationship between reducibility of nickel oxide species and the interaction of NiO particles with the support, which seems to determine the selectivity to ethylene during ethane ODH. Fig. 11 plots the relative hydrogen consumption of the band at 330 ◦C (reduction degree), related to NiO with low interaction with the alumina support, and the selectivity to ethylene at isoconversion conditions. It can be observed that the presence of NiO with low interaction with the alumina support must be avoided in order to achieve high ethylene formation, since an inverse relationship between the hydrogen consumption of the NiO reduction peak and the selectivity to the olefin has been observed. Accordingly, a higher interaction between NiO particles and support decreases the reducibility of Ni n+ species, and the concentration of electrophilic oxygen species, thus favoring a more controlled oxygen supply during catalytic cycles, and a higher selectivity to the olefin. This NiO-support interaction determines not only NiO crystal size, but also the definition/crystallinity of the crystals (and the concentration of defects in the active phase). As discussed in the UV Raman spectra, an increase in the relative intensity of the 1 P LO band (I 580 ) with respect to that for 2 P 2LO band (I 1100 ) means that the NiO crystallite size decreases and/or the amount of oxygen defects increases. As the crystal size in all the catalysts is not constant, an accurate estimation of the number of defects cannot be undertaken through this technique. Interestingly, as observed by TEM, the reference catalyst 15Ni/AL and those with oxalic acid (samples 15Ni/AL-o1 and 15Ni/AL-o3) present very similar NiO crystal size (Fig. 4). However, the I 580 /I 1100 ratio in the corresponding UV Raman spectrum of 15Ni/AL (Fig. 5, spectrum c) is remarkably higher than that observed for 15Ni/AL-o1 and 15Ni/AL-o3 catalyst (Fig. 5, spectra h and i). Accordingly, a different concentration of oxygen species can be proposed for the reference catalyst (15Ni/AL). Similarly, both Nb-containing catalysts present similar NiO crystal Fig. 8. Ni 2p 3/2 core-level XPS spectra: xNi/Al catalysts (left); selected Al 2 O 3 -supported NiO catalysts with 15 wt% NiO (right). Y. Abdelbaki et al.