Mesoporous Chromium Catalysts Templated on Halloysite Nanotubes and Aluminosilicate Core/Shell Composites for Oxidative Dehydrogenation of Propane with CO 2

: The oxidative dehydrogenation of alkanes is a prospective method for oleﬁns production. CO 2 -assisted propane dehydrogenation over metal oxide catalysts provides an opportunity to increase propylene production with collateral CO 2 utilization. We prepared the chromia catalysts on various mesoporous aluminosilicate supports, such as halloysite nanotubes, nanostructured core/shell composites of MCM-41/halloysite (halloysite nanotubes for the core; silica of MCM-41-type for the shell), and MCM-41@halloysite (silica of MCM-41-type for the core; halloysite nanotubes for the shell). The catalysts have been characterized by X-ray ﬂuorescence analysis, low-temperature nitrogen adsorption, X-ray diffraction, temperature-programmed reduction, temperature-programmed desorption of ammonia, transmission electron microscopy with energy-dispersive X-ray spectroscopy, and thermogravimetric analysis. The catalysts’ performance in carbon-dioxide-assisted propane dehydrogenation has been estimated in a ﬁxed-bed reactor at atmospheric pressure. The most stable catalyst is Cr/halloysite, having the lowest activity and the largest pore diameter. The catalyst, Cr/MCM-41/HNT, shows the best catalytic performance: having the highest conversion (19–88%), selectivity (83–30%), and space–time yield (4.3–7.1 mol C 3 H 6 /kg catalyst/h) at the temperature range of 550–700 ◦ C. The highest space–time yield could be related to the uniform distribution of the chromia particles over the large surface area and narrow pore size distribution of 2–4 nm provided by the MCM-41-type silica and transport channels of 12–15 nm from the halloysite nanotubes.


Introduction
Olefins are the basis of the modern petrochemical industry. The production of most bulk polymers is based on olefins polymerization. Many other petrochemical products are also olefin derivatives. Ethylene is the leading olefin, with an annual production exceeding 200 million metric tons, and is mainly used for the production of polyethylene, ethylene dichloride, ethylbenzene, and ethylene oxide. Propylene is the second largest-scale olefin, with an annual production of over 100 million metric tons. Polypropylene, acrylonitrile, propylene oxide, isopropyl benzene, and oxo alcohols are among the main products of propylene conversion [1][2][3].
While steam cracking is the exclusive method of ethylene production almost used, the main processes of propylene production are the steam cracking of hydrocarbons, fluid catalytic cracking (FCC), and propane dehydrogenation. Other processes, such as olefin removes hydrogen from the active sites by the reverse water-gas shift reaction but is unable to oxidize chromium to the inactive states Cr 5+ and Cr 6+ [6]. Other widely investigated active components are the oxides of molybdenum, vanadium, and gallium [20][21][22][23][24].
Typical supports for the dehydrogenation catalysts are SiO 2 and Al 2 O 3 [22,[25][26][27][28]. Along with the traditional supports, such as alumina and silica, ordered mesoporous silica [29][30][31][32] and zeolites [33,34] are also an object of interest. Both types of materials have a well-defined structure with narrow pore size distributions and high surface areas. The chemical nature of ordered mesoporous silicas provides excellent thermal stability [35]. The acidity of silica is lower than that of zeolites, which favors high selectivity in alkane dehydrogenation [36]. In contrast, the high acidity of zeolites prevents their direct use as dehydrogenation catalysts and demands careful dealumination to ensure high selectivity to olefins [37].
Various types of mesoporous silicas are also widely investigated as supports for dehydrogenation catalysts [29,[38][39][40][41][42][43][44]. It was shown that a high surface area of mesoporous silica favors the formation of sub-nanosized particles of the active phase of dehydrogenation catalysts [45]. Low loadings of chromium in SBA-1 lead to a high abundance of Cr 5+ species [29]. A comparison of SBA-15 as the catalyst support with Al 2 O 3 and ZrO 2 shows benefits for both oxidative and non-oxidative propane dehydrogenation [40]. Nevertheless, mesoporous silicas show the partial collapse of their structure during dehydrogenation [44]. Recently, mesoporous organo-silica has been investigated as a support of CO 2 -assisted ethane dehydrogenation [46]. Mesoporous silica was doped with Ce x Zr 1-x O 2 to promote oxygen mobility [47].
Halloysite is a mesoporous aluminosilicate comprising kaolin sheets rolled into nanotubes with a non-uniform charge distribution [48]. The nanotubes have a typical external diameter of 40-50 nm, a lumen internal diameter of 10-15 nm, and a length of up to 1500 nm [49]. Halloysite demonstrates excellent mechanical properties and may serve as a reinforcing additive to paper [50], polymer scaffolds [51], and silica gels for water shutoff [52]. The lumen of the nanotubes can be loaded with functional materials, ensuring their controlled release over usage time and, therefore, providing the deacidification of paper [50], as well as corrosion-inhibiting [53] and antibacterial properties [54]. The internal "alumina" surface is positively charged, whereas the outside "silica" surface has a negative charge. Thereby, halloysite tubes can be modified in various ways: acid or alkali etching and grafting with silanes and azines [55][56][57][58][59]. Interestingly, the asymmetry of the charge allows for modifying the exclusively internal or external nanotube surfaces, thus easing the regioselective construction of complicated composites [60] and layer-by-layer self-assembly of nanotubes into structures [61]. Halloysite nanotubes can be used as a reinforcing core for MCM-41 synthesis, thus improving the mechanical properties of the resulting composites [62,63]. Such functional materials have already been investigated in benzene hydrogenation [62] and p-xylene oxidation [64]. Another option is to use halloysite as a shell for the MCM-41 encapsulated inside nanotubes [65].
The present work is devoted to the synthesis and investigation of chromia-based catalysts supported on halloysite and halloysite-derived mesoporous core/shell materials, such as MCM-41/halloysite (halloysite nanotubes for the core; silica of MCM-41-type for the shell), and MCM-41@halloysite (silica of MCM-41-type for the core; halloysite nanotubes for the shell) in the oxidative dehydrogenation of propane, using CO 2 as an oxidant.

Results and Discussion
Three functional materials have been investigated in this work: pristine halloysite nanotubes (HNT); MCM-41, synthesized around halloysite (MCM-41/HNT); and MCM-41, synthesized inside the halloysite lumen (MCM-41@HNT). The general idea was to compare these materials' similarities by composition as catalyst supports and elucidate the influence of their morphology on catalytic performance.
The structure of the synthesized functional materials has been estimated by lowtemperature N 2 adsorption (Table 1). Pristine HNT has a BET surface area of 67 m 2 /g, a Catalysts 2023, 13, 882 4 of 15 pore volume of 0.30 cm 3 /g, and an average pore diameter (D p ) of 14 nm. The MCM-41 synthesis around HNTs increases the BET surface area up to 887 m 2 /g, a pore volume to 0.59 cm 3 /g, and decreases the pore diameter to 3 nm, which is close to the typical value for MCM-41. These dramatic changes are caused by the formation of MCM-41, not only around the nanotubes but also in the standalone phase. The synthesis of MCM-41 inside the halloysite nanotubes increases the BET surface area (324 m 2 /g) and pore volume (0.37 cm 3 /g) to a smaller extent due to the restriction of MCM-41 nucleation and decreases the pore diameter less significantly, which could indicate that not all nanotubes are filled with mesoporous silica. In this way, functional materials with high surface areas, low acidity, and high mechanical strength could be prepared based on widely spread natural halloysite clays. We used boehmite as a binder (20 %wt.) for granulating the catalysts, thus mimicking industrial-scale catalyst preparation. The binder simplifies the catalyst handling and its loading into the reactor. However, the addition of the binder may alter the textural properties of the catalyst support and complicate the microscopic assessment of the prepared catalysts. The deposition of the active phase (chrome oxide) is another possible cause for the change in textural properties. Therefore, we studied both the source composites and prepared catalysts by low-temperature nitrogen adsorption. Table 1 also shows the textural properties and chromium content of the catalysts prepared on the functional materials above. After the preparation of the catalyst supports with boehmite forming and further calcination, the BET surface areas, pore volume, and pore diameters have significantly reduced compared to the corresponding functional materials due to the relatively high content of chromia and functional material dilution with boehmite. The chromium contents are close to the nominal value. Figure 1 shows the corresponding N 2 adsorption isotherms and pore volume distribution vs. pore diameter for the catalysts. Isotherms of Cr/MCM-41/HNT and Cr/MCM-41@HNT have the typical shape specific to MCM-41 [66,67]. The pore volume distribution of Cr/MCM-41/HNT has two maxima at 2.4 and 12 nm. The first one is sharp and corresponds to the MCM-41 pores, and the second wide one could be attributed to the alumina binder. The pore volume distribution of Cr/MCM-41@HNT also has two maxima at 2.8 and 12 nm, with a similar origin. The pore size distribution of Cr/HNT is wide without a sharp maximum.
The observed pore size distributions suggest a significant increase in the catalyst support porosity with additional sharp maxima in Cr/MCM-41/HNT and Cr/MCM-41@HNT, which did not appear in the pore size distribution of Cr/HNT. This increase implies the possible improvement in the catalytic performance due to the larger surface area and availability of the surface for chromia deposition. Figure 2 shows the X-ray diffraction (XRD) patterns of the fresh calcined catalysts and pure  not show any reflections assigned to chromia, which can be due to the high dispersion and particle localization inside the hierarchical structure of the MCM-41-containing supports. Similar results have been previously observed for V/MCM-41 catalysts and could be related to the high dispersion of chromia, its amorphous state, or its incorporation in the MCM-41 phase [68]. Typically, the chromia phases appear in the XRD patterns of Cr/MCM-41 catalysts when the Cr content is at least 7-10% wt. [69,70]. The observed pore size distributions suggest a significant increase in the catalys port porosity with additional sharp maxima in Cr/MCM-41/HNT and Cr/MCM-41@ which did not appear in the pore size distribution of Cr/HNT. This increase impli possible improvement in the catalytic performance due to the larger surface are availability of the surface for chromia deposition. Figure 2 shows the X-ray diffraction (XRD) patterns of the fresh calcined cat and pure Cr2O3 (JCPDS file #38-1497). The Cr/HNT pattern includes reflections at 2θ to 24.5, 33.6, 36.2, 50.2, and 54.9, which are typical for Cr2O3. Based on the main refle (33.62°), the average particle size of the Cr2O3 phase has been estimated as 47 nm, on the Scherrer equation. The samples of Cr/MCM-41/HNT and Cr/MCM-41@HNT d show any reflections assigned to chromia, which can be due to the high dispersio particle localization inside the hierarchical structure of the MCM-41-containing sup Similar results have been previously observed for V/MCM-41 catalysts and could lated to the high dispersion of chromia, its amorphous state, or its incorporation MCM-41 phase [68]. Typically, the chromia phases appear in the XRD patter Cr/MCM-41 catalysts when the Cr content is at least 7-10% wt. [69,70].  The observed pore size distributions suggest a significant increase in the catalyst support porosity with additional sharp maxima in Cr/MCM-41/HNT and Cr/MCM-41@HNT, which did not appear in the pore size distribution of Cr/HNT. This increase implies the possible improvement in the catalytic performance due to the larger surface area and availability of the surface for chromia deposition. Figure 2 shows the X-ray diffraction (XRD) patterns of the fresh calcined catalysts and pure Cr2O3 (JCPDS file #38-1497). The Cr/HNT pattern includes reflections at 2θ equal to 24.5, 33.6, 36.2, 50.2, and 54.9, which are typical for Cr2O3. Based on the main reflection (33.62°), the average particle size of the Cr2O3 phase has been estimated as 47 nm, based on the Scherrer equation. The samples of Cr/MCM-41/HNT and Cr/MCM-41@HNT do not show any reflections assigned to chromia, which can be due to the high dispersion and particle localization inside the hierarchical structure of the MCM-41-containing supports. Similar results have been previously observed for V/MCM-41 catalysts and could be related to the high dispersion of chromia, its amorphous state, or its incorporation in the MCM-41 phase [68]. Typically, the chromia phases appear in the XRD patterns of Cr/MCM-41 catalysts when the Cr content is at least 7-10% wt. [69,70].   Figure 3 shows the TPR-H 2 profiles of the catalysts. Cr/HNT has a reduction peak with a maximum of~381 • C, which could be related to the chromate reduction on the "alumina" side of the HNTs [71,72]. Cr/MCM-41@HNT has a peak at 301 • C and could be assigned to the chromate reduction over the silica surface. Cr/MCM-41/HNT has a sharp maximum at 295 • C and a wide one at 394 • C; therefore, it could be related to both chromates in the MCM-41 pores and lumens of HNTs, correspondingly. Figure 3 shows the TPR-H2 profiles of the catalysts. Cr/HNT has a reduction peak with a maximum of ~381 °C, which could be related to the chromate reduction on the "alumina" side of the HNTs [71,72]. Cr/MCM-41@HNT has a peak at 301 °C and could be assigned to the chromate reduction over the silica surface. Cr/MCM-41/HNT has a sharp maximum at 295 °C and a wide one at 394 °C; therefore, it could be related to both chromates in the MCM-41 pores and lumens of HNTs, correspondingly.     Figure 3 shows the TPR-H2 profiles of the catalysts. Cr/HNT has a reduction peak with a maximum of ~381 °C, which could be related to the chromate reduction on the "alumina" side of the HNTs [71,72]. Cr/MCM-41@HNT has a peak at 301 °C and could be assigned to the chromate reduction over the silica surface. Cr/MCM-41/HNT has a sharp maximum at 295 °C and a wide one at 394 °C; therefore, it could be related to both chromates in the MCM-41 pores and lumens of HNTs, correspondingly.    We performed a detailed STEM/EDX study with elemental mapping to clarify the localization of chromia in the prepared catalysts. Figure 5 shows a bright-field TEM image (a), dark-field STEM micrograph (d), and the corresponding elemental mapping (b,c,e,f) of the same area of the Cr/MCM-41/HNT catalyst. The halloysite nanotube contains aluminum, which is well-defined by the aluminum EDX spectrum (Figure 5e and the blue channel in Figure 5f). Around the nanotube, the MCM-41 framework is formed, which is composed of pure silica, readily visible by its silicon spectrum (Figure 5c and the green channel in Figure 5f). Chromia is uniformly distributed over the catalyst (Figure 5b and the red channel in Figure 5f). The sharp elemental distribution can be observed in the spectral overlay (Figure 5f).
The TEM micrographs confirm the expected catalyst morphology: the presence of only tubular structures in Cr/HNT, the simultaneous presence of halloysite nanotubes and mesoporous MCM-41 matter in Cr/MCM-41/HNT, and MCM-41 matter completely packed inside the nanotubes in Cr/MCM-41@HNT. However, the localization of chrome oxide cannot be readily determined by TEM alone. Such an effect could be related to the formation of sub-nanosized particles of the active phase. [45].
We performed a detailed STEM/EDX study with elemental mapping to clarify the localization of chromia in the prepared catalysts. Figure 5 shows a bright-field TEM image (a), dark-field STEM micrograph (d), and the corresponding elemental mapping (b,c,e,f) of the same area of the Cr/MCM-41/HNT catalyst. The halloysite nanotube contains aluminum, which is well-defined by the aluminum EDX spectrum (Figure 5e and the blue channel in Figure 5f). Around the nanotube, the MCM-41 framework is formed, which is composed of pure silica, readily visible by its silicon spectrum (Figure 5c and the green channel in Figure 5f). Chromia is uniformly distributed over the catalyst (Figure 5b and the red channel in Figure 5f). The sharp elemental distribution can be observed in the spectral overlay (Figure 5f). The STEM/EDX study confirms the structure of the prepared catalysts hypothesized from the TEM images. Particularly, chromia was observed as evenly distributed on the catalyst support surface and not forming separate particles in all the studied samples. The spectral overlay (shown in Figure 5f for the Cr/MCM-41/HNT catalyst) shows the colocalization of Cr with silica (the red and green channels in Figure 5f) or with alumina (the red and blue channels in Figure 5f). Figure 6 shows the catalytic performance of the samples. The activity of the catalysts could be estimated from the conversion of propane (XC3H8) vs. temperature dependencies. In the temperature range of 550-700 °C, XC3H8 over Cr/MCM-/HNT is 0.05-0.15 higher than all the other samples. XC3H8 over Cr/HNT is the lowest. The selectivity to propylene was also the highest over Cr/MCM-41/HNT, especially at low temperatures. At 550 °C, it was 0.826 compared to 0.756 for Cr/HNT and 0.713 for Cr/MCM-41@HNT. The lowest selectivity was observed over Cr/MCM-41@HNT at any temperature in the investigated range. In terms of the space-time yield (STY), Cr/MCM-41/HNT also demonstrated the highest The STEM/EDX study confirms the structure of the prepared catalysts hypothesized from the TEM images. Particularly, chromia was observed as evenly distributed on the catalyst support surface and not forming separate particles in all the studied samples. The spectral overlay (shown in Figure 5f for the Cr/MCM-41/HNT catalyst) shows the colocalization of Cr with silica (the red and green channels in Figure 5f) or with alumina (the red and blue channels in Figure 5f). Figure 6 shows the catalytic performance of the samples. The activity of the catalysts could be estimated from the conversion of propane (X C3H8 ) vs. temperature dependencies. In the temperature range of 550-700 • C, X C3H8 over Cr/MCM-/HNT is 0.05-0.15 higher than all the other samples. X C3H8 over Cr/HNT is the lowest. The selectivity to propylene was also the highest over Cr/MCM-41/HNT, especially at low temperatures. At 550 • C, it was 0.826 compared to 0.756 for Cr/HNT and 0.713 for Cr/MCM-41@HNT. The lowest selectivity was observed over Cr/MCM-41@HNT at any temperature in the investigated range. In terms of the space-time yield (STY), Cr/MCM-41/HNT also demonstrated the highest performance. It increased from 4.3 mol C 3 H 6 /kg cat/h at 550 • C to 7.1 mol C 3 H 6 /kg cat/h at 700 • C. The STY achieved a maximum at 650 • C for Cr/HNT (5.6 mol C 3 H 6 /kg cat/h) and Cr/MCM-41@HNT (5.5 mol C 3 H 6 /kg cat/h). The maxima for both samples are due to the selectivity drop while the temperature increases. Figure 6d shows the Arrhenius plots for all the catalysts. The Arrhenius activation energies have been estimated as 84 ± 6 kJ/mol for Cr/HNT, 66 ± 5 kJ/mol for Cr/MCM-41/HNT, and Cr/MCM-41@HNT for 79 ± 4 kJ/mol. The values are close to the apparent activation energies reported earlier for the Cr/Al 2 O 3 catalysts (76-78 kJ/mol) [13] and VO x /SBA-15 (90 kJ/mol) [73].   Table 2 presents more detailed data for the product distribution, as well as the performance of the catalysts in non-oxidative propane dehydrogenation. For the latter experiment, CO2 was replaced with N2 to maintain the C3H8 partial pressure at the same level. For the experiments without CO2, one may observe lower propane conversions and higher selectivity to propylene in the same conditions. Meanwhile, the STYs were decreased, indicating lower catalyst (templated on the MCM-41-HNT composites) productivity for propylene for the non-oxidative dehydrogenation process. For the HNT-based sample, the STY values are comparable both for the oxidative and non-oxidative dehydrogenation, which could be ascribed to relatively large polychromate clusters. The same observations were reported by the researchers elsewhere [29].  Table 2 presents more detailed data for the product distribution, as well as the performance of the catalysts in non-oxidative propane dehydrogenation. For the latter experiment, CO 2 was replaced with N 2 to maintain the C 3 H 8 partial pressure at the same level. For the experiments without CO 2 , one may observe lower propane conversions and higher selectivity to propylene in the same conditions. Meanwhile, the STYs were decreased, indicating lower catalyst (templated on the MCM-41-HNT composites) productivity for propylene for the non-oxidative dehydrogenation process. For the HNT-based sample, the STY values are comparable both for the oxidative and non-oxidative dehydrogenation, which could be ascribed to relatively large polychromate clusters. The same observations were reported by the researchers elsewhere [29]. Figure 7a shows the catalyst stability estimated as the propane conversion drop with the time on stream. The most stable is the Cr/HNT sample, which could be related to the (1) lower activity and the lowest acidity (Table 1), thus, the lower coke formation rate, and (2) the wide pore size of the support (14 nm compared to 2-4 nm for MCM-41/HNT and MCM-41@HNT). The most active and selective catalyst, Cr/MCM-41/HNT, has lower stability, as X C3H8 decreases from 0.45 to 0.26 in one hour. It can be explained by the higher acidity among all the tested catalysts (0.217 mmol NH 3 /g, Table 1), boosting cracking and therefore increasing coke formation. This assumption also agrees with the results obtained   Figure 7a shows the catalyst stability estimated as the propane conversion drop with the time on stream. The most stable is the Cr/HNT sample, which could be related to the (1) lower activity and the lowest acidity (Table 1), thus, the lower coke formation rate, and (2) the wide pore size of the support (14 nm compared to 2-4 nm for MCM-41/HNT and MCM-41@HNT). The most active and selective catalyst, Cr/MCM-41/HNT, has lower stability, as XC3H8 decreases from 0.45 to 0.26 in one hour. It can be explained by the higher acidity among all the tested catalysts (0.217 mmol NH3/g, Table 1), boosting cracking and therefore increasing coke formation. This assumption also agrees with the results obtained for the Cr/MCM-41@HNT sample. The most stable Cr/HNT catalyst shows the lowest mass loss at 25% wt. Both samples supported on the MCM-41-based composites have a similar mass loss of ca. 46-48% wt (Figure 7b). It can be concluded that, from the prepared catalysts, Cr/MCM-41/HNT has the highest performance in terms of activity, selectivity, and space-time yield. The use of such materials with high surface areas and hierarchical structures allows for preparing active and selective dehydrogenation catalysts. It can be concluded that, from the prepared catalysts, Cr/MCM-41/HNT has the highest performance in terms of activity, selectivity, and space-time yield. The use of such materials with high surface areas and hierarchical structures allows for preparing active and selective dehydrogenation catalysts.

Materials
The

Synthesis of Functional Materials
The ordered functional material MCM-41/HNT, where the MCM-41 was synthesized around halloysite nanotubes, was prepared by a modification of the procedure described in [74]. In a polypropylene flask, a charge of halloysite clay (1.44 g) was dispersed in water (384 mL) by continuous stirring for 1 hour. After that, hexadecyltrimethylammonium bromide (2.00 g), propanol-2 (107.32 g), and aqueous ammonia (35.8 g) were added and stirred for an additional 1 hour. After the addition of tetraethyl orthosilicate (7.48 g), the mixture was stirred for 4 hours. Then, the suspension was filtered, and the precipitate was washed with deionized water until it was bromide-free. The washed precipitate was dried step-wise at 80, 90, 100, and 110 • C with 4 h dwell times. The final calcination was completed in an airflow atmosphere with a ramp rate of 2 • C/min at 550 • C for 6 h.
The functional material MCM-41@HNT with MCM-41, synthesized inside the halloysite nanotubes, was prepared according to the procedure described elsewhere [65]. Halloysite clay was pretreated via etching in hydrochloric acid (2 M) at 70 • C for 24 h. After filtering the suspension, the precipitate was washed and dried at 80 • C for 12 h. A charge of etched halloysite (5 g) was dispersed in a mixture of water (250 mL), hexadecyltrimethylammonium bromide (5 g), and 2-propanol (50 mL). After sonication for 60 min, the pH was adjusted to 10.5 using aqueous ammonia, and the suspension was stirred for 3 h. The gel was aged at room temperature for 24 h. Subsequent heating to 90 • C for 24 h led to the formation of a precipitate, which was washed, step-wise dried at 60, 80, 110, and 140 • C (with 2 h dwell times), and calcined with a ramp rate of 2 • C/min at 550 • C for 6 h.

Catalyst Preparation
The catalyst supports were prepared by mixing the corresponding functional material (HNT, MCM-41/HNT, MCM-41@HNT) with boehmite (20% wt.) using an aqueous peptizing agent with nitric acid (1 M) and polyethylene glycol (5% wt.). Pellets with a diameter of 2 mm were formed, dried out at room temperature for 24 h, step-wise at 80-110 • C, and calcined at 550 • C in an airflow atmosphere for 4 h. After crushing and sieving, the fraction 0.2-0.5 mm was used for impregnation.
The chromium was deposited by incipient wetness impregnation of the support with aqueous chromium (III) nitrate for 8 h with a subsequent dry-out at 100 • C for 5 h, and calcination at 550 • C in the air for 3 h. The moisture capacities of the supports were as follows: 0.51 g H 2 O/g HNT, 1.05 g H 2 O/g MCM-41/HNT, and 0.98 g H 2 O/g MCM-41@HNT. The nominal chromium loading was 5 %wt.

Catalyst Characterization
The chromium content was determined using the X-ray fluorescence spectrometer ARL Perform'X (Thermo Fisher Scientific, Waltham, MA, USA) using the standardless UniQuant ED 6.30 software.
The low-temperature N 2 adsorption isotherms were measured with a Gemini VII 2390t (Micromeritics Instrument Corp., Norcross, GA, USA). Samples (0.5 g) were randomly selected from the same batch of catalyst that was used for the catalytic performance experiments. Sample outgassing was performed at 400 • C for 4 h in a vacuum. The specific surface area (S BET ) was estimated by using the Brunauer-Emmet-Teller (BET) method in the relative pressure range of 0.05-0.30. The pore size distribution was estimated by using the Barrett-Joyner-Halenda (BJH) model.
Transmission electron microscopy (TEM) was performed with the JEM-2100 (JEOL, Tokyo, Japan) with an accelerating voltage of 200 kV. The samples were crushed and dispersed in ethanol and then distributed onto 300-mesh Lacey copper TEM grids (Ted Pella, Redding, CA, USA). The Lacey grids were chosen for their maximum clarity of TEM micrographs without a formvar background.
A temperature-programmed reduction with hydrogen (TPR-H 2 ) was conducted with the Autochem 2950HP (Micromeritics, Norcross, GA, USA) equipped with a quartz reactor and thermal conductivity detector. The reduction mixture contained 7% H 2 -Ar (balance). The measurement conditions were as follows: flow rate of 30 mL/min; heating rate of 10 • C/min.
The acidity of the supports and catalysts was evaluated using the temperatureprogrammed desorption of ammonia (TPD-NH 3 ) on the Autochem 2950HP (Micromeritics, Norcross, GA, USA) equipped with a quartz reactor and thermal conductivity detector. NH 3 adsorption was carried out at 60 • C for 30 min. Physically adsorbed ammonia was removed in a nitrogen flow at 100 • C for 30 min. The TPD-NH 3 analysis was performed upon heating the sample from 100 to 700 • C with a temperature ramp of 10 • C/min. Thermogravimetric analysis (TGA) of the spent catalysts was carried out on an STA 449F5 instrument (Netzsch, Waldkraiburg, Germany). A sample was placed in a corundum crucible and heated from 30 to 1000 • C with a heating rate of 10 K/min in an air stream (purge: 50 mL/min; protective (N 2 ): 20 mL/min). The final temperature was maintained for 2 h to ensure complete coke elimination. An empty Al 2 O 3 crucible was applied as a reference.

Catalytic Tests
The catalytic experiments were performed in an isothermal fixed-bed quartz reactor at atmospheric pressure. Before the catalysts' activity evaluation, blank experiments with quartz beds at 550-600 • C were conducted, and the propane conversion did not exceed 5.5%. The main products were C1-C2 hydrocarbons. The propane and CO 2 flow rates were controlled independently by mass-flow controllers from Bronkhorst. A catalyst sample (1 g) was mixed with quartz beds (2 g). The propane flow rate was 10 mL/min, and the molar ratio of CO 2 /C 3 H 8 was equal to 2. Before each test, the catalysts were regenerated at 650 • C for 1 h in an airflow atmosphere.
On-line gas chromatography of the feedstock and products was performed using the Chromos GC-1000 (Chromos Engineering, Dzerzhinsk, Russia) equipped with a capillary column Valco PLOT VP-Alumina Na 2 SO 4 (50 m × 0.53 mm × 10 µm) and flame ionization detector. The deviation between parallel experiments did not exceed 5%.
The conversion of propane (X C3H8 ) was calculated as and the selectivity to propylene (S C3H6 ) was according to the following equation: where C in i and C out i are the reactor inlet and outlet concentration. The space-time yield was calculated as follows: where F is the molar flow rate of the feedstock. The rate of the propane conversion was estimated as follows: where the above-mentioned designations are used.

Conclusions
Three types of chromia catalysts for propane oxidative dehydrogenation with carbon dioxide were prepared: supported on halloysite nanotubes; MCM-41, templated around halloysite nanotubes; or MCM-41, embedded in halloysite nanotubes. The silica of MCM-41-type synthesis around or inside halloysite nanotubes improves the textural properties of halloysite and the thermal stability of MCM-41. Compared to pristine halloysite nanotubes, MCM-41-decorated supports show significantly higher specific surface areas and smaller average pore diameters. The effect of specific surface area increases is especially pronounced for MCM-41/HNT. The catalyst based on MCM-41 around nanotubes shows the best performance, which could be related to the highest BET surface area (558 m 2 /g) and uniform distribution of chromia. The space-time yield exceeds 7 mol C 3 H 8 /kg cat/h (at 700 • C).
The catalytic performance of the obtained composites shows that a high surface area and thermal stability are essential for CO 2 -assisted propane dehydrogenation catalysts. Furthermore, the morphology of the catalyst particles plays an important role: the catalysts with similar composition and chromia content (namely, Cr/MCM-41/HNT and Cr/MCM-41@HNT) demonstrate remarkably different performances. The best of the tested catalysts, Cr/MCM-41/HNT, combines easily available MCM-41-type surface and larger pores that are formed by the halloysite nanotubes. Halloysite nanotubes are available in thousands of tons worldwide and could be easily scaled up for industrial applications.
Further developments for better propane dehydrogenation may involve the nanoarchitectural approach, namely, the catalysts combining macropores and ordered mesoporous materials for the maximum availability of chromia-active centers. It is worth noting that the performance of the chromia-based catalysts can also be tailored by promoting them with alkali. These improvements will help to close the "propylene gap" and utilize industrial carbon dioxide, thus facilitating the transition to greener petrochemical processes. Data Availability Statement: Data is contained within the article. Additional data may be provided on request.

Conflicts of Interest:
The authors declare no conflict of interest.