CO2 hydrogenation to dimethyl ether over In2O3 catalysts supported on aluminosilicate halloysite nanotubes


 This work presents results on CO2 hydrogenation to dimethyl ether (DME) over bifunctional catalysts consisting of In2O3, supported on natural clay halloysite nanotubes (HNT), and HNT modified with Al-MCM-41 silica arrays. The catalysts were characterized by TEM, STEM, EDX-mapping, NH3-TPD, XRD, low-temperature nitrogen adsorption, TPO, and H2-TPR techniques. Catalytic properties of In2O3/HNT and In2O3/Al-MCM-41/HNT in the CO2 hydrogenation to DME were investigated in a fixed-bed continuous flow stainless steel reactor at 10–40 atm, in the temperature range of 200–300°C, at GHSV = 12,000 h−1 and molar ratio of H2:CO2 = 3:1. The best catalyst for CO2 hydrogenation was In2O3/Al-MCM-41/HNT that provided DME production rate 0.15 gDME·(gcat·h)−1 with DME selectivity 53% and at 40 bar, GHSV = 12,000 h−1, and T = 250°C. It was shown that In2O3/Al-MCM-41/HNT exhibited stable operation for at least 40 h on stream.


Introduction
Currently, many efforts of various researchers around the world are being made to solve environmental problems. Air pollution is considered one of the main such problems. Various options are offeredthe use of alternative energy sources, such as solar [1], wind [2], and biofuel [3]. However, it is also necessary to pay great attention to the utilization of CO 2 . The ever-growing production capabilities of different countries lead to an increase in carbon dioxide emissions into the atmosphere. This is the reason for the increase in the so-called "greenhouse effect," which leads to an increase in the global temperature of the planet and, accordingly, climate change. This, as well as the fact that CO 2 is an inexpensive, readily available compound, requires the search for new technologies, methods, and ways of processing carbon dioxide. Recently, more and more attention of researchers from all over the world has been attracting the study of the reaction of CO 2 hydrogenation into various compounds such as methane [4], methanol [5][6][7][8], dimethyl ether (DME) [9][10][11], or hydrocarbons [12]. Among these compounds, DME attracts attention as a multipurpose productit is used in the synthesis of methyl acetate, dimethyl sulfate, various petrochemical compounds, as a feedstock for powering fuel cells [13][14][15]. Due to its propertieshigh cetane number (55)(56)(57)(58)(59)(60), low autoignition temperature, and high oxygen content (∼35%), DME is considered as an alternative to diesel fuel or LPG. In terms of its physicochemical properties, DME is close to LPG, which allows its simple storage and transportation.
Commonly, DME is synthesized according to a twostage scheme through the synthesis of methanol (MeOH) from synthesis gas on Cu/ZnO/Al 2 O 3 (CZA) catalyst and its subsequent conversion into DME on a solid acid catalyst. However, the direct synthesis of DME is thermodynamically more favorable than the synthesis of methanol [16], which attracts more attention to the study of this process. Catalysts for the direct synthesis of DME by hydrogenation of CO 2 are divided into two types: first, a mechanical mixture of catalysts for the synthesis of methanol and a catalyst for its dehydration; second, a catalyst called "bifunctional" which contains both types of necessary catalytic sites on its surface. Typically, first type systems are prepared by mixing or grinding of methanol synthesis and acidic components. Such method has some disadvantages like disintegration of its components during the reaction, mass, and heat transfer limitations [17]. So, recently, bifunctional catalysts have attracted much attention of scientists [10]. Another challenge is to perform direct DME synthesis with high selectivity without formation of CO. Industrial Cu/ZnO/Al 2 O 3 methanol synthesis catalyst is also known to be active in reverse water gas shift (RWGS) reaction causing hydrogen losses due to formation of CO.
According to the recent density functional theory calculations [18], it is possible to obtain methanol with high selectivity via the hydrogenation of CO 2 on indium oxide. The reaction proceeds by the cyclic mechanism of the formation of oxygen vacancies and subsequent activation of CO 2 on them. Later, these calculations were experimentally confirmed. It was shown that methanol with ∼100% selectivity is achieved on bulk In 2 O 3 at low CO 2 conversions [19]. These works gave an impetus to further extensive studying of catalysts on indium oxide in the hydrogenation of CO 2the effect of various supports, preparation methods, the structure of indium oxide, and various additives on the catalytic activity [20][21][22][23][24]. Thus, indium oxide as a catalyst for methanol synthesis looks promising. An acid component is required for the design of a bifunctional DME direct synthesis catalyst. Usually, γ-Al 2 O 3 or various zeolites -H-ZSM-5, Y, MOR, FERare used as an acid catalyst in a two-stage process [17,[25][26][27][28][29][30][31][32][33].
Nowadays, core-shell structure catalysts, such as Cu-ZnO-Al 2 O 3 @HZSM-5 [51] or CuO-ZnO-Al 2 O 3 @SiO 2 -Al 2 O 3 [52], are intensively studying in direct DME synthesis from CO 2 . These systems prevent metal particles from sintering [53] and deactivation due coke formation by side reactions [54]. So, in the literature there are works on the modification surface of halloysite with MCM-41 [36] to make core-shell structure. This core-shell halloysitebased aluminosilicate composite is promising for catalytic applications due to high specific surface area and enhanced thermal and mechanical properties [39]. However, pure MCM-41 doesn't have the required acid sites on its surface. Therefore, before using it as a support, the surface was modified with aluminum in order to increase acid sites. The resulting Al-MCM-41 has a large amount of acid sites, which are necessary to produce DME by CO 2 hydrogenation.
In this work, we present novel bifunctional In 2 O 3 catalysts supported on natural clay nanotubes (10 wt% In 2 O 3 /HNT) and composite with structured mesoporous silica (10 wt% In 2 O 3 /MCM-41/HNT) for CO 2 hydrogenation to DME. Actual In 2 O 3 loadings in the catalysts were determined by inductively coupled plasma atomic emission spectrometry (Optima instrument; Perkin-Elmer).
Transmission electron microscope (TEM) JEOL JEM-2100 (UHR) operated at 200 kV (the lattice resolution of 0.19 nm) and equipped with LaB6 gun was employed to investigate structure, morphology, and chemical composition of the obtained samples. The samples for the TEM analysis were prepared by the dispersing in ethanol. The as-prepared dispersed solution was dropped onto carbon-coated formvar TEM Cu grid (300 mesh, Ted Pella, Inc.). The acquisition of TEM/HRTEM images was performed in TEM mode using Olympus Quemesa 11 megapixel CCD camera. The collection of each EDX map was performed in STEM mode with help of EX-24065JGT energy dispersive X-ray (EDX) analyser.
The specific BET surface areas (S BET ) and pore volume (V p ) of the support and the catalysts were determined using the low-temperature N 2 -adsorption method using a TriStar3000 apparatus. Before experiment, all samples were outgassed in vacuum at 300°C, then nitrogen adsorption/desorption isotherms were recorded at −196°C. The specific surface area of the samples was calculated by the Brunauer-Emmett-Teller (BET) equation. The pore volume was evaluated in accordance with Barrett-Joyner-Halenda model. NH 3 temperature-programmed desorption (NH 3 -TPD) was used to evaluate the acid properties of the samples. The catalyst was saturated by mixture of NH 3 and N 2 at 100°C for 30 min. After that, the sample was purged with a stream of nitrogen to remove physisorbed ammonia at same conditions. Then NH 3 -TPD curve was recorded up to 700°C with a rate of 10°per minute.
X-ray structural analysis (XRD) of the samples was recorded on a Bruker D8 Advance (Bruker, Germany) diffractometer (CuKα) in the 2θ range of 8°-63°with a step 0.05°per 4 s. Analysis of the obtained diffraction data was carried out using the PowderCell 2.4 programme using the JCPDS international diffraction database as a reference.

Catalyst testing
Catalytic experiments on CO 2 hydrogenation were studied in a fixed-bed continuous-flow stainless steel reactor (inner diameter 8 mm) at a 10-40 atm pressure in the temperature interval 200-300°C, at GHSV = 12,000 h −1 and molar ratio H 2 :CO 2 = 3:1. Prior to the reaction, all the catalysts (V cat = 2 cm 3 , m cat = ∼1.4 g for In 2 O 3 /HNT and ∼0.5 g for In 2 O 3 /Al-MCM-41/HNT, particle size of 0.5-1 mm) were pretreated at 300°C for 1 h in helium flow. The temperature was measured using a chromelalumel thermocouple, which was placed in the middle of the catalytic bed. The results were obtained after multiple catalytic experiments. The catalysts were tested in several temperature increasing/decreasing cycles. At each temperature, the catalyst was kept for 1-2 h. Thus, total time onstream under CO 2 hydrogenation conditions was not less than 10 h. The catalytic performance during this period remained stable. The compositions of the inlet and outlet gas mixtures were analyzed by a gas chromatograph (Chromos-1000) equipped with TCD and FID detectors and molecular sieve (5A) and Carbowax columns. Argon was used as a carrier gas. The detection limits for CO, CO 2 , CH 4 , DME, and methanol were 5 × 10 −3 vol%. The carbon imbalance in all catalytic experiments was ±5%.
CO 2 conversion (X CO2 ), MeOH, and DME selectivity (S MeOH , S DME ) were calculated as follows:  Figure A1 in Appendix) areas of In 2 O 3 and HNT are quite similar and equal 68 and 71 m 2 ·g −1 , respectively. After the impregnation of In 2 O 3 on the HNTs' surface, morphological characteristics are practically unchanged: S BET and pore volume slightly decreased to 62 m 2 ·g −1 and 0.13 cm 3 ·g −1 , respectively. The most likely reason is blocking of some pores by the indium oxide particles. Formation of MCM-41 phase on HNT leads to significant increase of surface area due to ordered structure of silica arrays. After deposition of indium oxide, specific surface area decreased by 100 m 2 ·g −1 , for the same reason as on the In 2 O 3 /HNT catalyst. Acidity parameters of the catalysts and supports calculated from NH 3 -TPD method are listed in Table 2 ( Figure A2). Based on the desorption spectra, the acidity was classified as weak and medium (amount of ammonia (μmol·g −1 ) desorbed below 300°C) and strong sites (amount of ammonia (μmol·g −1 ) desorbed above 300°C). Table 2, the acidity of unmodified halloysite is seriously lower than that of HNT/Al-MCM-41 due to the fact that modified with aluminum MCM-41 has strong acid sites on the surface [43]. The total amount of acid sites on the surface of In 2 O 3 supported catalysts is reduced in comparison with supports. This fact can be explained by partial blocking of the pores by the indium oxide particles. For the catalysts after the experiment, we can say that total amount of acid sites remained the same. Figure 1 shows the XRD patterns for fresh and used catalysts. According to XRD data, we can say that indium oxide on the surface 10% In 2 O 3 /HNT has cubic crystal phase structure [21] with crystallite size of 13 nm. In case of 10% In 2 O 3 /Al-MCM-41/HNT, the crystallite size of the indium oxide is slightly smaller -10 nm. We assumed that this is due to the higher dispersion of indium oxide particles. As we can see from the diffraction patterns of the used catalysts (curve 3 and 5 on Figure 1), there are no significant changes in the number and composition of the peaks. We only note that for the used catalysts, the   peaks related to indium oxide are slightly smaller compared to fresh catalysts. Both used catalysts don't have any considerable changes in the crystal structurepore volume and CSR remained almost the same. This tells us that indium oxide particles are stable on the catalysts surface. The In 2 O 3 /HNT and In 2 O 3 /Al-MCM-41/HNT catalysts were studied by TEM, STEM, and EDX techniques. Figure 2 shows the TEM images of the fresh and used catalyst In 2 O 3 /Al-MCM-41/HNT, which are obviously similar.

As shown in
HNT were observed in both samples, and the images show that the nanotubes remained stable under reaction conditions. The same results were obtained for the In 2 O 3 /HNT catalyst. Also, in Figure 2b and d, we can see the structure of the mesoporous MCM-41 type silica deposited on the outer surface of HNTs. Some agglomerates of fresh and used In 2 O 3 /Al-MCM-41/HNT catalysts were studied by STEM and EDX-mapping. Results are shown in Figure 3. It is seen that the STEM images (Figure 3c and d) of both catalysts are similar. It can be noted that the indium particles are located mainly in the same place as the silicon particles in the case of both catalysts. Thus, it can be concluded that In 2 O 3 /Al-MCM-41/HNT catalyst contains two types of active sites on the surfaceindium oxide particles, supported on silica, and alumina oxide particles.
Catalysts were examined by H 2 -TPR ( Figure 4) to study reducibility of catalysts. We can see that pure indium oxide isn't reduced in the investigated temperature range. For the In 2 O 3 /Al-MCM-41-HNT, hydrogen consumption occurs only at ∼170°C, indicating that the indium oxide nanoparticles are mainly localized on mesoporous silica of Al-MCM-41 type. This peak can be correlated to reduction of In 2 O 3 surface and can also be attributed as indirect evidence of the formation of oxygen vacancies on the surface of indium oxide [57,58]. For the In 2 O 3 /HNT, there is also a peak at ∼170, which could be assigned to reduction of In 2 O 3 surface particles on the outer (SiO
Among the tested catalysts, In 2 O 3 catalyst exhibited the lowest CO 2 conversion, but the highest methanol selectivity over the entire temperature range. This can be explained by the fact that no DME was observed in reaction products. It seems to be quite obvious, since this catalyst does not have required acid sites on its surface. We can see that CO 2 conversion increases with increasing temperature from 1% at 200°C up to 4% at 300°C. The selectivity for methanol, on the contrary, decreases with increasing temperature from 99% at 200°C to 65% at 300°C due to the CO formation by RWGS reaction.
In contrast to bulk In 2 O 3 for the supported In 2 O 3 /HNT and In 2 O 3 /Al-MCM-41/HNT catalysts, DME appears in reaction products, due to the presence of acid sites. Temperature dependence for DME selectivity is similar to methanol in case of bulk In 2 O 3the curve decreases with increasing temperature. There are some reasons for that. First, methanol dehydration is exothermic reaction, so increasing temperature leads to decrease of DME/MeOH equilibrium ratio. Second, at high temperatures, RWGS reaction (which is endothermic) contributes more to product distribution. So, it should be an optimal temperature, where the combination of formation rate of DME and CO 2 conversion will be maximum. Over the temperature range, we can see that on In 2 O 3 /Al-MCM-41/HNT there are higher values of CO 2 conversion and DME selectivity than on In 2 O 3 /HNT. Most likely, it is connected with higher surface area and more acid sites on In 2 O 3 /Al-MCM-41/HNT catalyst. Figure 6 shows temperature dependencies of DME production rate on In 2 O 3 , In 2 O 3 /HNT, and In 2 O 3 /Al-MCM-41/HNT catalysts.
The highest DME formation rate of 0.15 g DME ·(g cat ·h) −1 was observed at 250°C on In 2 O 3 /Al-MCM-41/HNT. Further, the performance of the most active and selective catalyst In 2 O 3 /Al-MCM-41/HNT was studied in more detail. It is well-known that with increasing pressure, the equilibrium of the CO 2 hydrogenation reaction shifts towards the products according to the Le Chatelier principle. So, we studied the pressure influence on catalytic activity in DME direct synthesis from CO 2 and H 2 . The results are shown in the Figure 7. The experiments were carried out at T = 250°C, GHSV = 12,000 h −1 .
As expected, with increasing pressure, CO 2 conversion and DME selectivity also increase, while MeOH passes through the maximum at 20 atm and then decreases. This is in accordance with thermodynamic equations for this system [9]. The highest value of the DME formation rate is observed at 4 MPa. Note that at temperatures higher than 250°C, W DME decreased, despite an increase in the CO 2 conversion due to a significant drop of DME selectivity.
One of the key properties of the catalyst, in addition to activity, is the stability under reaction conditions. A series of experiments were carried out to investigate this aspect. Figure 8 shows the effect of time-on-stream on the outlet product concentrations and CO 2 conversion. The In 2 O 3 /Al-MCM-41/HNT catalyst was tested at 250°C, the inlet mixture H 2 :CO 2 = 3:1, and GHSV = 12,000 h −1 . Under these conditions, only CO, MeOH, and DME were detected as reaction products; methane appeared only in trace amounts. During 10 h on stream, no significant changes were observed either in the conversion of CO 2 or in the MeOH and DME selectivity. No significant changes in the selectivity of DME were observed after 8 h of the experiment. After that, catalyst remained in operation conditions for 24 h, and after that catalyst activity was recorded. All the parameters, such as methanol, DME selectivity, and CO 2 conversion, remained the same. In our view, these results mean that indium oxide particles remained stable as well as acid sites of modified HNT. In addition, the spent catalyst was tested by TPO. No carbon deposition was observed. This means that acidic properties of HNT (strength and number of acidic sites) are optimal for DME synthesis reaction and do not induce condensation reactions.
Up to now, almost no works of using indium oxide for direct synthesis of DME can be found in literature; only   one work devoted to the study Cu-In-Zr-O catalyst mixed with SAPO-34 zeolite for direct DME synthesis is available [59]. Basically, all works on the direct synthesis of DME from CO 2 and H 2 are devoted to the study of copper catalysts mixed with zeolites. So, we compared the best In 2 O 3 /Al-MCM-41/HNT catalyst with literature data. Table 3 shows comparative data, in particular, experimental conditions (temperature, pressure, flow), CO 2 conversion, and DME selectivity. Since a fairly large number of works devoted to the hydrogenation of CO 2 to DME are currently presented in the literature, the table shows those with similar experimental conditions with this work, in particularpressure of 10-50 atm, temperature of 200-300°C, and inlet composition H 2 :CO 2 = 3:1. Catalytic activity of In 2 O 3 /Al-MCM-41/HNT is lower than literature data, but study of these systems is at the very beginning. Such systems look very promising due to the following factors: the possibility of a significant increase in CO 2 conversion and selectivity for DME after optimization of the catalyst composition, its dispersion, the method of preparation, and adding of promoters. According to the literature data [60], catalysts based on indium oxide make it possible to obtain methanol with a selectivity of about 100%, and with the appropriate selection of the acid component, high DME yields can be achieved. It is also important that In 2 O 3 /MCM-41/HNT catalyst shows good stability, due to the fact that indium oxide particles do not sinter during the reaction, and acid sites remain stable in presence of water. There is a wide field for further catalyst improvement, including optimization of In 2 O 3 morphology and interaction with the support, tuning acidic properties, doping by metals active in CO hydrogenation, such as Cu, Pd, Ga, and even Ni and Co. These points will be the subject of our further studies.

Conclusion
Indium oxide catalysts, bulk and supported on aluminosilicate HNTs and modified HNTs with ordered Al-MCM-41 silica arrays, were studied in CO 2 hydrogenation to DME. Based on data from physicochemical methods, such as XRD, S BET , FTIR, and H 2 -TPR, we can suggest that these catalysts have two types of active sitesindium oxide particles, which are responsible for methanol formation, and acid sites of HNT, which are responsible for methanol dehydration to DME. The influence of temperature and pressure was studied. The best catalyst for CO 2 hydrogenation was In 2 O 3 /Al-MCM-41/HNT that provides 4% CO 2 conversion with DME selectivity 53% and DME production rate 0.15 g DME ·(g cat ·h) −1 at 40 bar, GHSV = 12,000 h −1 , and T = 250°C. It was shown that this catalyst didn't lose activity after 40 h of experiment. So, it is very promising systems, based on new material for direct hydrogenation of CO 2 to DME.