Propylene is one of the most important building blocks in petrochemical industry [1, 2], with a global production capacity of 127 million metric tons per year. Currently, the industrial process for propylene production is mainly based on the steam cracking and fluid catalytic cracking (FCC) of naphtha, light diesel, and other oil byproducts [2, 3], which inevitably involves extensive energy consumption and significant emission of CO₂ [1, 4]. Compared with the thermal or catalytic cracking, direct catalytic propane dehydrogenation (PDH) to obtain propylene is a more economical and environmentally friendly route [5] and has been commercialized by UOP and ABB Lummus in the 1990s [6]. Three types of direct catalytic PDH technologies, CATOFIN, Oleflex and STAR, have been developed on an industrial scale in more than 14 installations [4]. In addition, many new facilities are already under construction, or their prospective constructions have been announced [7].

In commercialized PDH technologies, the CrO_x-based catalysts have been widely used because of their low price and outstanding performance [4, 8]. However, these catalysts suffer from fast deactivation by coke deposition, which has become one of the major problems in the existing PDH processes [9, 10]. The spent catalysts need to be frequently regenerated in a high-temperature oxidative atmosphere to remove coke [9]. Thus, it is of great significance and highly desirable to prepare thermally stable CrO_x-based catalysts with anti-coking properties [10]. This can be achieved by a flexible design over supports.

The catalyst support not only serves as a scaffold but also provides a complex microenvironment, which is a key factor in determining the catalytic function [3]. The catalyst support should be thermally stable enough to maintain the morphology of the catalyst, in addition to providing an open structure for the catalyst to allow easy access to the reactants and preventing the blockage of the pore by coke [4, 11]. Following these lines, porous alumina is one of the most widely used material in the industries with desirable textural properties and thermal stability [9, 11]. Unfortunately, the acidic nature of alumina always leads to coke formation and side reactions, inevitably leading to catalyst deactivation [9]. It has been reported that Lewis acids are generated after the dehydration/dehydroxylation of the surface and are present in the form of coordinatively unsaturated Al³⁺, which in turn depends on the nanostructure and morphology of alumina [12-14]. The properties of alumina depend on the initial aluminum hydrates, and the calcination step dictates the properties of the final product by controlling the stepwise removal of water and OH groups [15]. However, owing to the structural complexity and diversity of phase transition, it is still challenging to precisely figure out the relationship between the structure and preparation conditions. In our previous work, a series of alumina such as Al₂O₃ nanorods with rough surface and Al₂O₃ nanosheets rich in pentacoordinate Al³⁺ were prepared through a hydrothermal approach, which could nicely disperse and stabilize gold and platinum, respectively. When used in CO oxidation and PDH reactions, the catalysts displayed excellent activities and superior stabilities [5, 16]. Therefore, we conceive that alumina prepared by a similar approach could also be beneficial to the CrO_x-based catalysts.

Herein, we investigated the relationship between the structure and preparation conditions of Al₂O₃ and focused on adjusting the acidity of alumina by varying the calcination temperature. A series of rod-shaped porous alumina were prepared by a hydrothermal route, followed by calcination at different temperatures. The synthesized alumina, as a support for Cr₂O₃, possessed low acidity and exhibited excellent stability and anti-coking ability in the PDH process.

Typically, Al(NO₃)₃·9H₂O and CO(NH₂)₂ at a molar ratio of 1:9 were dissolved in deionized water, and the obtained solution was transferred to a Teflon-lined stainless-steel autoclave and heated at 100 ℃ for 24 h. Then, the white precipitates were washed with deionized water and anhydrous ethanol to remove the ion impurities and dried at 80 ℃. After calcination at different temperatures of 700, 800, and 900 ℃ for 2 h, the final products were obtained and named as Al₂O₃-T, where T represents the calcination temperature of alumina.

Using Cr(NO₃)₃·9H₂O and KNO₃ as the precursors, a series of catalysts were prepared using the incipient co-impregnation method, in which the theoretical amounts of Cr₂O₃ and K₂O loaded were 18 wt% and 1.5 wt%, respectively. After impregnation, the samples were maintained at room temperature for 2 h and then dried at 50 ℃ for 12 h. This was followed by calcination in 20% O₂/N₂ at 600 ℃ for 4 h and then reduced in 20% H₂/N₂ atmosphere at 600 ℃ for 2 h. The obtained unreduced and reduced samples were named as Cr-Al-T-C and Cr-Al-T, respectively. For comparison, commercial alumina (denoted as Al₂O₃-Ref) was calcined at 800 ℃ and used to prepare the reference samples, which were named as Cr-Al-Ref-C and Cr-Al-Ref.

X-ray powder diffraction (XRD) measurements were operated on a PANalytical X'Pert3 Powder diffractometer using Cu K_α radiation (λ = 0.15406 nm). The tube voltage was 40 kV, and the current was 40 mA. Nitrogen adsorption-desorption isotherms were obtained using a Micromeritics TriStar 3000 adsorption analyzer. The specific surface areas (S_BET) were calculated from the adsorption data in the relative pressure range of 0.05–0.3 using the Brunauer-Emmett-Teller (BET) method. Pore size distributions (PSDs) were determined based on the non-local density functional theory. Transmission electron microscopy (TEM) images were recorded on a FEI TECNAI F30 microscope, operating at an accelerating voltage of 300 kV. Scanning electron microscopy (SEM) investigations were carried out with a Hitachi FESEM SU8220 instrument. Ultraviolet-visible (UV-vis) spectra were recorded on a Cary 5000 UV-Vis-NIR spectrophotometers (Agilent, Japan) using diffuse reflectance spectroscopy (DRS) in the range of 200–1000 nm. BaSO₄ (AR) was used as white standards to dilute the samples to minimize the effect of the highly different extinction coefficients. Temperature-programmed reduction of hydrogen (H₂-TPR) was performed on a Micromeritics AutoChem Ⅱ 2920 apparatus with a thermal conductive detector by passing 8% H₂/Ar (flow rate of 50 mL min^-1) at a heating rate of 10 ℃ min^-1 up to 900 ℃. Before H₂-TPR, the samples were pretreated at 150 ℃ for 1 h under an Ar flow to ensure a clean catalyst surface. The system was then cooled to ambient temperature under Ar flow. Temperature-programmed desorption of ammonia (NH₃-TPD) was carried out on a Micromeritics AutoChem Ⅱ 2920 apparatus. Typically, the sample was pretreated in a flow of He (50 mL min^-1) at 500 ℃ for 1 h. After the temperature lowered to 100 ℃, the sample was pulsed by NH₃ gas six times. Then the gas phase NH₃ was removed by He purging for 1 h, after which TPD was performed in He flow (50 mL min^-1) at a heating rate of 10 ℃ min^-1. The desorbed mixture was monitored by a TCD detector, and the desorption of NH₃ was monitored by on-line mass spectrometry (MS). The amount of coke deposited was determined by the combustion of the deposited material monitored on a thermogravimetric (TG) analyzer STA 449 F3 (NETZSCH). Prior to combustion, the spent catalyst was pre-treated with flowing He (40 mL min^-1) at 200 ℃ for 1 h, and then cooled to ambient temperature under He. Finally, the sample was exposed to a mixture of 20% O₂ in N₂ flowing at 20 mL min^-1 and oxidized from ambient temperature to 900 ℃ at a heating rate of 10 ℃ min^-1. The CO₂ generated was monitored by on-line MS.

Catalytic performance was tested in a tubular fixed-bed quartz reactor (I. D. = 8 mm) at atmospheric pressure. The temperature of the reactor was controlled by a Yudian AI temperature controller (Series 708P). 100 mg of catalyst was used in each experiment. A feed gas composition of C₃H₈ at 8 mL min^-1 was passed through the catalyst bed at 600 ℃ for the reaction to take place. The products were analyzed by an on-line gas chromatograph. GDX-105 and molecular sieve 5A columns were used to analyze N₂, H_2, C₃H₈, C₃H₆, C₂H₄, C₂H₆, and CH₄ in the products.

Here, i is hydrocarbon product in the effluent gas stream, n_i is the number of carbon atoms of component i, and F(i) is the corresponding flow rate.

A first-order deactivation model was used to evaluate the catalyst stability:

Here, X_initial and X_final are the conversion measured at the start and end, respectively, of an experiment, t is the reaction time (h), and k_d is the deactivation rate constant (h^-1). Higher k_d values are indicative of rapid deactivation, and hence, low stability.

A series of Cr-Al-T catalysts were prepared, as mentioned in the experimental section, and the propane dehydrogenation reactivities over these catalysts as well as the reference catalysts were plotted (Fig. 1). All Cr-Al-T catalysts showed better activities compared with the reference catalysts (Fig. 1a). Among the three Cr-Al-T catalysts, Cr-Al-800 exhibited the optimal initial propane conversion of 33.2%, with 90.4% propylene selectivity. After 170 min of the reaction, the propane conversion and the propylene selectivity slightly decreased to 20.4% and 84.7%, respectively. For the Cr-Al-Ref catalyst, the propane conversion dramatically dropped from an initial 40.4% to 12.8%, with the propylene selectivity decreasing from 87.7% to 71.0%. The deactivation rate constant for the Cr-Al-800 catalyst (0.26%) was much lower than that for the reference catalyst (0.61%). The amount of coke deposited on the spent catalysts was determined by thermogravimetric analysis (TG) in a temperature-programmed oxidation mode (TPO) (Fig. 1b). It was observed that the amount of coke deposited on the Cr-Al-800 catalyst (3.6%) was much lower than that on the Cr-Al-Ref catalyst (15.7%). To further investigate the location of coke deposit, the concentration of CO₂ under TPO was detected by MS. Generally, the coke deposited at low temperatures is mainly the one that covers the active metal, while that deposited during combustion at higher temperatures is the one that is present on the external surface of the support [5, 17, 18]. Fig. 1d shows that the spent Cr-Al-800 and Cr-Al-Ref catalysts presented a similar desorption peak ranged between 250 and 550 ℃, which indicated that coke was deposited in the same location, although the amount of coke deposited varied greatly. Furthermore, five dehydrogenation-regeneration cycles over the fresh Cr-Al-800 catalyst were investigated to examine the regenerative ability and stability; the recycles for the PDH reactions are presented in Fig. 1c. Cr-Al-800 catalyst exhibited superior regenerative ability, with the initial propane conversions being above 33% in all recycles, and the propylene selectivity remaining almost unchanged.

To investigate the effect of different calcination temperatures for alumina and the corresponding catalysts, a series of characterization techniques were carried out. Fig. 2a shows that Al₂O₃-700 and Al₂O₃-800 have the main diffraction peaks at 37.6°, 45.8°, and 66.9°, which were assigned to the γ-Al₂O₃ phase (COD: 01-079-1158) [18]. As the temperature increased, the intensity of the reflection at 66.9° became stronger, indicating an increase in the crystallite size [15]. The change in the nature of lattice from Al₂O₃-800 to Al₂O₃-900 suggested the occurrence of phase transition, and this transition may correspond to the formation of δ-Al₂O₃ phase [15]. The diffraction peaks corresponding to the crystal phase of Cr₂O₃ were present in the XRD patterns of the Cr-Al-T catalysts (COD: 01-084-1616) (Fig. 1b). After the reaction, the spent catalysts retained their structure (Fig. S1, Supporting Information). Nitrogen adsorption-desorption isotherms (Fig. 1c) exhibited type Ⅳ adsorption isotherms with H2-shaped hysteresis loops, indicating the typical mesoporous structure with ink-bottle-like mesopores [19]. The corresponding PSDs suggested a mesopore size centered in the range of 5–10 nm. Textural properties of the samples are summarized in Table 1. Specific surface area and pore volume of alumina decreased with increasing calcination temperature, while the pore diameter roughly increased. The decrease in the surface area is because of sintering [20]. The increase in pore size and decrease in pore volume contributed to the change in the crystal structure of alumina [21]. The specific surface area and pore volume of catalysts also decreased after loading active metal oxides. The mesopore size distributions were centered at 3.8, 5.0, and 5.9 nm for Cr-Al-700, Cr-Al-800, and Cr-Al-900, respectively, which allowed easy access to the reactants and prevented the blockage of the pore by coke (Table S2). More importantly, the total acidity of the synthesized alumina was generally lower than that of the commercial alumina. A higher calcination temperature caused a significant decrease in the total acidity (Table 1).

Furthermore, the morphologies of the samples were characterized by SEM, TEM, and HRTEM. It can be clearly seen in Fig. 3a, b, and c that the representative Al₂O₃-800 sample displayed rod-shaped structure and quite a rough surface that possessed numerous small mesopores with the average sizes of 5.0 nm. The results were consistent with nitrogen sorption measurement. After the loading of Cr species, the roughness of the surface seemed to disappear, implying the uniform distribution of Cr species on the surface (Fig. 3e and f). The Al₂O₃-Ref showed large bulk particles (Fig. S4a) and a relatively smooth surface (Fig. S4d). The introduction of Cr species had little impact on the morphology (Fig. S4e). After the activity test, the spent catalysts were observed by SEM, TEM, and HRTEM images. The morphology of the Cr-Al-800 catalyst was unchanged after the reaction (Fig. 3d and g), while the particle size of Cr-Al-Ref catalyst decreased significantly (Fig. S4b and c), which indicated that the synthesized alumina as a support for Cr₂O₃ possessed excellent structural stability. The absence of any obvious difference in the surface of the catalyst after the reaction suggested that Cr species were still uniformly distributed on the rough surface (Fig. 3h and i). The unique nanostructure of the synthesized rod-shaped alumina with quite a rough surface may contribute to the outstanding stability of the Cr-Al-800 catalyst.

To identify the origin of the difference in the catalytic performance of the Cr-Al-800 and Cr-Al-Ref catalysts, a series of analyses were performed. The diffuse reflectance UV-vis spectra for the fresh catalysts were recorded to confirm the coordination state of the Cr species (Fig. 4a). All the samples exhibited four absorption bands in the range of 240–260, 350–370, 450–470, and 590–610 nm. The intense bands centered at 257 and 367 nm were assigned to the O → Cr (Ⅵ) charge transfer in isolated Cr₂O₃ with a tetrahedral symmetry [22, 23]. The band at 461 nm was assigned to the d-d transition (A_2g–T_1g) in Cr³⁺ with octahedral symmetry [23]. Another obvious band at 600 nm could be assigned to the d-d transition (A_2g–T_2g) in Cr³⁺ with octahedral symmetry, which was attributed to the Cr₂O₃ crystal [24]. The UV-vis spectra indicated that Cr-Al-800 and Cr-Al-Ref possessed the similar Cr species.

The reduction behavior of Cr-Al-800-C and Cr-Al-Ref-C was investigated by H₂-TPR experiments (Fig. 4b). Cr-Al-Ref-C exhibited two major peaks due to hydrogen consumption at 296 and 373 ℃. In the case of Cr-Al-800-C, the corresponding reduction peaks shifted to 327 and 420 ℃. The signals of hydrogen consumption in the temperature range 210–460 ℃ were attributed to the reduction of the dispersed Cr⁶⁺ to Cr³⁺ species [25]. It was worth noting that the reduction temperature of Cr-Al-800-C was higher as compared with that of Cr-Al-Ref-C, suggesting that the Cr species in the former is difficult to reduce. Also, the Cr-Al-Ref-C sample showed an obvious peak at a higher temperature, which can be attributed to the formation of dichromate phases [26, 27]. Extensive research proved that the surface acidity of Cr₂O₃ catalysts was an important feature that affected the chromia-support interaction [19, 28]. These observations indicated the presence of different metal-support interaction on Cr-Al-800-C and Cr-Al-Ref-C samples, which could be caused by the different surface acidities [28].

NH₃-TPD experiments were carried out to analyze the number and strength of the acid sites. It was observed from Fig. 4c and 4d that all the samples presented a broad desorption peak ranged between 100 and 500 ℃. The shapes of the profiles were complex and asymmetric, suggesting the presence of different acid sites. To further investigate the distribution of acid sites, we fitted the asymmetric profiles by a Gaussian function. The patterns exhibited three desorption peaks, and the peak centered between 170 to 190 ℃ was assigned to the desorption of NH₃ bonded by weak acid sites, while the peak centered between 220 to 290 ℃ was assigned to the deliberation of medium-strong acid sites. The temperature maximum (T_max) of each desorption peak and the fraction of different acid sites for each sample are listed in Table 2. The table also lists the total acidities, which were estimated from the NH₃-TPD profiles. The distribution of surface acidity was similar in Al₂O₃-800 and Al₂O₃-Ref, and the fraction of medium-strong acid sites in the former significantly decreased with the loading of active metal oxides. The decrease of these acid sites could improve the propylene selectivity [29]. It was worth noting that the total acidities of pure Al₂O₃-800 and Al₂O₃-Ref were 178 and 281 μmol NH₃ g^-1, whereas in the case of the corresponding catalysts, these values distinctly decreased to 114 and 141 μmol NH₃ g^-1, respectively. This indicated that the surface acidity could be changed considerably after loading active metal oxides [30]. The total acidity of the Cr-Al-800 catalyst (64 μmol NH₃ g^-1) was much lower than that of the Cr-Al-Ref catalyst (140 μmol NH₃ g^-1), and the fraction of medium-strong acid sites for Cr-Al-800 catalyst was lower than the reference catalyst. Previous studies have demonstrated that the coke deposits were mainly formed through polymerization of olefins during the dehydrogenation reaction, which could be catalyzed on the acid sites [7, 17, 18]. Therefore, it can be inferred that the lower acidity of the Cr-Al-800 catalyst rendered it a better anti-coking ability.

Rod-shaped porous alumina was hydrothermally synthesized, and the corresponding Cr₂O₃ catalysts were used in the PDH process to investigate their catalytic performance. The calcination temperature played a crucial role in modifying the surface-physicochemical properties of alumina, subsequently affecting the catalytic activity. The Cr-Al-800 catalyst exhibited excellent stability because of the inhibitory effect of low acidity toward side reactions and coke formation, thus leading to better anti-coking ability. Therefore, the synthesized alumina for CrO_x-base catalysts could facilitate the PDH reaction.