Catalytic functionalities of Pd catalysts supported on spinel MgAl2O4 for coupling of ethylbenzene dehydrogenation with nitrobenzene hydrogenation

Direct coupling of oxidative dehydrogenation of ethylbenzene with nitrobenzene hydrogenation in a fixed-bed catalytic reactor at atmospheric pressure in the temperature range of 673–823 K over MgAl2O4-supported palladium catalysts has been investigated. Simple dehydrogenation of ethylbenzene (DHEB) is reversible, endothermic and thermodynamically limited. The continuous removal of co-produced hydrogen in DHEB for in situ hydrogenation of nitrobenzene over Pd/MgAl2O4 catalyst improves the performance of the DHEB process. A series of Pd (0.25–4.0 wt%) catalysts supported on MgAl2O4 was prepared. The catalyst samples were characterized by X-ray diffraction (XRD), H2-Temperature-programmed reduction (TPR), X-ray photoelectron spectroscopy (XPS), Transmission electron microscopy (TEM), CO chemisorption, CO2–Temperature-programmed desorption and BET surface area. Dispersion and particle size of Pd were obtained by CO chemisorption. The reaction profile shows that palladium (0.5 wt%) is an optimum loading to get maximum conversion of ethylbenzene (51.8 %) and nitrobenzene (47.3 %) to give 91.4 and 100 % selectivity of styrene and aniline, respectively.


Introduction
A large number of important processes in the chemical and petrochemical industries have reactions which are, to a greater or lesser extent, limited by equilibrium considerations. Production of styrene is one of such processes. It is a major petrochemical commodity used for the production of many different polymeric materials, such as acrylonitrile-butadiene-styrene (ABS), styrene-acrylonitrile (SAN), styrene-butadiene rubber (SBR), etc. [1]. Its number of applications and demand are still growing. It is commercially produced by vapor-phase dehydrogenation of ethylbenzene (DHEB) with large excess of superheated steam over potassium promoted iron oxide (Fe-K-O) catalyst in the temperature range of 873-973 K [2]. The main drawback of this process is its strongly endothermic character and limited equilibrium conversion of ethylbenzene [3]. Several approaches were being attempted by reducing the amount of co-produced hydrogen in the product to improve the performance of styrene yield either through reactor design or process modifications [4,5]. The former approach involving mostly membrane reactors has limited applicability due to their sensitivity towards certain species present in the reaction and also a cost factor. The other approach involves the introduction of oxidant molecules such as oxygen, halogens, sulfur and more recently mild oxidant like CO 2 in the dehydrogenation of ethylbenzene to overcome thermodynamic constraints [6][7][8]. However, development of commercial process was not successful to manufacture styrene by oxidative DHEB because of loss of styrene selectivity in addition to safety factors. Towler and Lynn reported that direct coupling of endothermic and exothermic reactions had improved thermal efficiency, reduction in catalyst deactivation and reaction rate due to displacement of equilibrium [9]. Similarly, Bautista as well as Sun et al. [10,11], also reported the improved performance of conversion of ethylbenzene into styrene in the presence of nitrobenzene overcoming thermodynamic equilibrium constraints in consistent with the Le Chatelier-Braun principle. However, separation of products and recycling of reactants in direct coupling reaction may require additional unit operations. In practice, operation of coupled reaction in the same reactor to give two useful products over a single catalyst outweighs the cost required for additional units [12,13]. Therefore, coupling of DHEB to styrene with hydrogenation of nitrobenzene to aniline appears to be an interesting synergetic pair as shown in the scheme. The innovative aspect of the catalytic system discussed here is the utilization of in situ produced hydrogen as a second reagent, allowing for the integrated catalytic transformation of nitrobenzene into aniline without using an external H 2 supply. Thus, it helps suppression of the most common side reactions in the primary reaction due to cracking of side chain moiety of ethylbenzene to benzene and ethylene or hydrogenolysis to toluene and methane, coke formation that eliminates the need for periodic regeneration of the dehydrogenation catalysts. Further, aniline produced in the secondary reaction is a valuable chemical commodity for the plastic, rubber processing, herbicide, dye and pigment industries [14]. It is a challenging task to develop a sole material to catalyze direct coupling process at higher reaction temperatures, as the catalyst employed for dehydrogenation has to be typically used for hydrogenation reactions as well [15]. Palladium is considered as a primary component in many dehydrogenation catalysts due to its high activity for activating C-H bonds, coupled with an inferior activity for the rupture of C-C bonds, resulting in intrinsically high selectivity toward dehydrogenation. On a Pd surface, only low-coordination number sites (steps, kinks) are able to catalyze the C-C bond breaking, while essentially all Pd sites catalyze the rupture of the C-H bonds. Similarly, Palladium as a primary component is well known for hydrogenation catalyst in many organic transformations. Recently, Guo et al. [16] used palladium catalysts supported on carbon nanotubes, alumina and silica in the hydrogenation of benzaldehyde as well as dehydrogenation of ethylbenzene in the vapor phase. Another undesired side reactions that compete with dehydrogenation are hydrogenolysis and coke formation. Since both, hydrogenolysis and coke formation, are more sensitive to the support structure than dehydrogenation. But, conventionally used supports for Pd catalysts include alumina [17], silica [18,19], activated carbon, to a lesser extent zeolite and polymers suffer from the disadvantages such as sintering of metal particles, burial of active phase, faster deactivation and coking. The efficient support significantly improves both activity and selectivity of the reaction as observed in coupling of cyclohexanol dehydrogenation with nitrobenzene hydrogenation [20]. Interestingly, some spinel-type mixed metal oxides (AB 2 O 4 ) were observed to be interesting materials in hydrogenation and dehydrogenation reactions because of their properties like high thermal and chemical stability and high mechanical resistance. Kustrowski et a1. [21] studied the non-oxidative and oxidative DHEB over Mg-Fe-Cr ternary spinel systems. Similarly, Ji et a1. [22] also investigated Fe-doped MgAl 2 O 4 spinel catalysts in the O-DHEB reaction. Thus, Magnesium aluminate spinels with neutral as well as basic sites appear to be an efficient catalyst support for palladium active phase for ethylbenzene dehydrogenation, reducing undesired cracking and hydrogenolysis products [23]. Moreover, Magnesium aluminate spinels have a tendency to prevent sintering of precious metals due to their strong metalsupport interaction and also provide resistance to coking by reaction media owing to their spinel structure [24].
Hence, a catalytic system based on palladium supported on magnesium aluminate spinel oxide was systematically investigated for its feasibility in the coupling reaction of ethylbenzene with nitrobenzene in a fixed-bed reactor. However, to the best of our knowledge, the performance of supported Pd catalysts for the dehydrogenation of ethylbenzene has not been reported. Here, we are demonstrating the performance of dehydrogenation/hydrogenation activity of Pd/MgAl 2 O 4 catalysts aiming at higher yield of styrene shifting thermodynamic equilibrium by removing the in situ produced hydrogen for hydrogenation of nitrobenzene. This results in the formation of another useful secondary product (aniline) simultaneously. This avoids additional cost of hydrogen purchasing, transportation and storage for secondary reaction.

Catalyst preparation
Magnesium aluminate spinel (MgAl 2 O 4 ) was prepared by the co-precipitation method [24]. Stoichiometric quantities of Mg (NO 3 ) 2 Á6H 2 O and Al (NO 3 ) 3 Á9H 2 O were dissolved in distilled water and the precipitation of hydroxide was carried out at a controlled pH 9 using 1 M NH 4 OH solution at room temperature with vigorous stirring. The precipitate was washed several times with distilled water to remove the excess of nitrate ions. It was dried in an air oven at 373 K for 12 h and then calcined at 1073 K for 8 h to give a spinel phase of MgAl 2 O 4 .
A series of palladium catalysts with varying Pd loadings ranging from 0.25 to 4.0 wt% on this support was prepared by wet impregnation method using an acidified solution of PdCl 2 (Aldrich). The samples were dried at 383 K in an air oven for 24 h and finally calcined at 773 K for 5 h in air.

X-ray diffraction studies
The X-ray diffraction (XRD) patterns were recorded on a Rigaku multiflex diffractometer using Ni filtered CuKa (k = 0.15406 nm) radiation of wavelength 1.5418 Å at 2H = 2 o -80 o with a scanning rate of 20 o min -1 in a voltage of 40 kV and a current of 100 mA.

BET surface area
The specific surface areas of the support and catalyst samples were estimated using N 2 adsorption-desorption isotherm at 77 K by the multipoint BET (Brunauer-Emmett-Teller) method on an Autosorb-1 instrument (Quantachrome USA). The powder of catalytic samples was first out-gassed at 473 K to ensure a clean surface prior to construction of isotherm. A cross-sectional area (0.164 nm 2 ) of N 2 molecule was assumed in the calculation of specific surface areas by BET method.

Temperature-programmed reduction (TPR)
Temperature-programmed reduction (TPR) experiments were carried out on an Auto Chem 2910 (Micromeritics, USA) instrument. In a typical experiment, Ca. 100 mg of calcined catalyst samples was taken in a U shaped quartz sample tube. Prior to TPR studies, the catalyst was pretreated with an inert gas (He, 50 mL/min) at 473 K. After pretreatment, the sample was cooled to ambient temperature and the carrier gas containing of 5 % H 2 -Ar (50 mL/ min) was allowed to pass over the sample while heating from ambient to 873 K at the rate of 10 K/min. CO chemisorption CO chemisorption measurements were also carried out on an Auto Chem 2910 (Micromeritics, USA) instrument. Prior to adsorption measurements, Ca. 100 mg of the sample was reduced in a flow of hydrogen (50 mL/min) at 673 K for 2 h and flushed out subsequently in a pure helium gas flow for an hour at 673 K and cooled to ambient temperature under the same gas flow. CO uptake was determined by injecting pulses of 9.96 % CO balanced helium from a calibrated on-line sampling valve into the helium stream passing over the reduced samples at 673 K.

X-ray photoelectron spectroscopy (XPS)
The XPS of catalysts was measured on a Kratos-Axis 165, XPS spectrometer with Mg Ka radiation (hv = 1253.6 eV) at 75 W. The Pd 3d, Mg 2p, and Al 2p core-level spectra were recorded, and the corresponding binding energies were referenced with C 1 s line at 284.6 eV [accuracy within (0.2 eV)]. The background pressure during the data acquisition was kept below 10 -10 bar.

Transmission electron microscopy (TEM)
Transmission electron microscope (TEM) images of the catalysts were obtained using a Tecnai-12, FEI, The Netherlands at an accelerating voltage of 120 kV. The specimens were prepared by dispersing the samples in methanol using an ultrasonic bath and evaporating a drop of resultant suspension onto the carbon-coated copper grid.

Temperature-programmed desorption (TPD)
Temperature-programmed desorption of CO 2 studies was also conducted on Auto Chem 2910 (Micromeritics, USA) instrument. In a typical experiment for TPD studies, ca. 100 mg of oven-dried sample (dried at 383 K for overnight) was taken in a U shaped quartz sample tube. Prior to TPD studies, the catalyst sample was pretreated at 473 K for 30 min by passing pure helium (99.999 %, 50 mL/ min). After reduction, the sample was saturated with CO 2 in a flow of 10 % CO 2 balance He mixture at 303 K with a flow rate of 75 mL/min and was subsequently flushed at 378 K for 1 h to remove physisorbed CO 2 . TPD analysis was carried out from ambient temperature to 973 K at a heating rate of 10 K/min. The amount of CO 2 desorbed was calculated using GRAMS/32 software.

Catalytic activity studies
The simple dehydrogenation and direct coupling (mixed hydrogenation and dehydrogenation) reactions were carried out over various Pd/MgAl 2 O 4 in the range of temperature 673-823 K in a fixed-bed micro-reactor made of quartz (6 mm i.d. and 200 mm long). Ca. 0.5 g of catalyst, diluted with equal amount of quartz grains, was positioned between two layers of quartz wool at the center of reactor. The upper portion of the reactor was filled with quartz grains which served as a pre-heater and mixer for the reactants. A type-K thermocouple enclosed in a quartz thermowell of 3-mm outer diameter was positioned inside the catalyst bed for accurate measurement of the catalyst temperature. Prior to the activity test, the catalyst was reduced in H 2 (30 mL/min) at 673 K for 3 h and then passivated under N 2 flow (30 mL/min) at 673 K for 0.5 h. After bringing the reactor to required temperature in N 2 flow, a mixture of ethylbenzene (EB) and/or nitrobenzene(NB) in the molar ratio of 3:1 at a feed rate of 4.28 mmol h -1 was fed into the reactor through a microprocessor-controlled metering pump (Braun, Germany) at atmospheric pressure. The liquid products such as benzene, toluene, styrene, aniline (AN), ethylbenzene and nitrobenzene were collected for every 1 h from ice cold trap and analyzed using an off-line flame ionization detection (FID) gas chromatography (Chemito GC-8510) with a SE-30 (30 m length and 0.53 mm i.d) packed column using Iolar N 2 as a carrier gas [25]. The components of liquid products were confirmed by a GC-MS (QP 5050A, Shimadzu Instruments, Japan). Calibration curves of each analyte were constructed using relative response factor against concentration of analyte from gas chromatography signals of prepared mixtures of known composition and used for estimating the percentage of each analyte in the liquid products. In this analysis, tetralin was used as an internal standard substance [26]. A typical calibration plot for determining the percentage of ethylbenzene composition in the mixture of products is shown in Fig. 1. The GC analyzed data were further processed to calculate the conversion and selectivity as mol% using the following Eqs. (1) to (4). Carbon balances of these investigations amounted to C95 %.
Selectivity of product i ¼ product i = EB fed ÀEB recovered ð Þ ½ Â 100 where product i ¼ styrene, benzene or toluene  Table 1. The specific surface area of the pure MgAl 2 O 4 support was found to be 126.9 m 2 g -1 and it decreases as a function of palladium loading . It might be due to partial blocking of the pores of the support by crystallites of palladium oxide as evidenced by XRD.

H 2 -TPR studies
The temperature-programmed reduction profiles of various palladium supported on MgAl 2 O 4 catalysts are shown in the Fig. 3. The TPR profiles of the samples show a reduction peak at above 574 K. This peak is attributed to the reduction of Pd 2? or PdO to metallic state of Pd. This reduction temperature is found to be much higher than that of bulk PdO indicating a strong interaction between PdO and MgAl 2 O 4 , which inhibits the reduction. Francova et al. [27] observed similar nature of high temperature reduction of PdO on Pd/Mg(Al)O catalysts. The maximum reduction temperature (T max ) of this peak varied in the temperature range (574-672 K) depending upon the loading of palladium in Pd/MgAl 2 O 4 catalysts. The T max value decreases from 672 to 574 K with increasing the palladium content (0.5-4.0 wt%). This implies that at lower Pd loading, the metal particles are highly dispersed as smaller Pd particles and existence of a strong interaction between the palladium and the support.

CO chemisorption
The physical properties of catalysts such as dispersion, metal surface area and average particle size obtained from CO chemisorption are given in Table 1. The dispersion of Pd was calculated from CO chemisorption using the following equation assuming the cubic particle with five sides exposed to the gas phase. The palladium metal areas were determined using the equation S CO = nm S X m ns -1 , where S CO is the total metallic surface area, nm S is the CO consumption and X m is chemisorption stoichiometry at monolayer coverage, and ns -1 is the number of palladium atoms per unit surface area. The CO uptake values of various Pd/MgAl 2 O 4 catalysts and the other information such as percentage of dispersion, particle size and metal surface area measured at 673 K are given in Table 1. The results clearly suggest that palladium loading and dispersion follow opposite trend. It was found dispersion decreased from 51.2 to 13.4 % with increase in palladium loading from 0.5 to 4.0 wt%, respectively. The high dispersion of Pd observed at lower Pd content may be due to the strong interaction of Pd precursor with oxygencontaining groups of MgAl 2 O 4 support. It is likely that the deposition may be more on the external surface of the MgAl 2 O 4 support as Pd content increases in the catalyst. This will reduce the distance among metallic species, thereby promoting agglomerization and decreasing in dispersion. This is in good agreement with XRD results.

X-ray photoelectron spectroscopy
The binding energies of electrons determined by XPS provide useful information on the oxidation states of   Fig. 4, which evaluates the oxidation state and interaction between Pd particle and support. According to the literature, the Pd ?2 (PdO) 3d 5/2 and 3d 3/2 binding energy values are around *337.0 and *342.0 eV, respectively [28]. In the series of catalysts prepared in the present investigation, the doublet peaks of binding energy (BE) for Pd 5 These results clearly show with increase of Pd loading the metal-support interaction decreases and also dispersion decreases. Voogt et al. [29] also suggested that a chemical shift can be induced by changes in the Pd particle size.
From Fig. 4, it is clear that intensities of peak increase with increase of Pd loading, resulting increase in palladium content on the surface of the support. The present XPS results are in well agreement with the dispersion of palladium determined by CO chemisorption and TPR methods.
Transmission electron microscopy (TEM) The transmission electron microscopy shown in Fig. 5 allows to visualize the palladium distribution on MgAl 2 O 4 . The dark dots correspond to the palladium particles with heterogeneous distribution. The metal particle size determined by TEM varies from 2.0 to 8.5 nm (Fig. 5a-c) indicating that metal loading facilitated to increase the particle size slightly. It is assumed that it is a consequence of the easier availability of the metal located on the catalyst surface to form clusters. The particle size obtained by chemisorptions measurements which varied from 2.2 to 8.1 nm with increase of palladium loading from 0.5 to 4.0 wt% (Table 1) is comparable with TEM technique which concludes a good agreement between both techniques can be achieved. To study the morphology and size of metal particles distributed over fresh and used catalysts, TEM was performed on 0.5 wt%Pd/MgAl 2 O 4 catalysts (Fig. 5ad). The metal size of Pd particles in the fresh catalyst is 2.0 nm while 2.9 nm in used catalyst. The metal particle size of used catalyst has not much significantly changed during course of reaction. Apparently, the coupling reaction (in situ removal of hydrogen) would effectively inhibit the formation of coke by hydrogenolysis which can result the long life time of catalyst.
Temperature-programmed desorption of CO 2 The basicity measurements of Pd/MgAl 2 O 4 catalysts were carried out by the temperature-programmed desorption of CO 2 . The CO 2 TPD profiles of pure MgAl 2 O 4 and various Pd/MgAl 2 O 4 catalysts are shown in Fig. 6. The CO 2 uptakes by various catalysts of different basic strengths are reported in Table 2. The desorbed peak of CO 2 was deconvoluted into two temperature regions, i.e., 353-473 K consists of weak basic sites and moderate basic sites in the temperature region of 473-673 K [30]. As can be seen from Fig. 6, the TPD profiles are found to be similar for all the samples. The ''weak'' basic sites are probably associated with desorption of the molecular CO 2 , ''medium'' basic sites are corresponding to the O 2and oxygen species from M 2? -O 2pairs [23]. In the case of supported palladium catalysts, the total basicity decreases with increase of palladium loading. The decrease in basicity at higher Pd loadings might be due to the formation of palladium crystallites. These results are supported by TPR and XRD studies.  Table 3. Conversion of ethylbenzene and selectivity towards styrene on all the catalysts for coupling reactions were much higher than simple dehydrogenation. It can be attributed to in situ removal of hydrogen for participation in the coupling reaction. Similar behavior in the activity of coupling reaction was reported by Bautista et al. [10]. It is   Table 3 shows the effect of Pd loading on coupling of dehydrogenation of ethylbenzene with hydrogenation of nitrobenzene. As the Pd loading increases from 0.25 to 4.0 wt%, the conversion of ethylbenzene and nitrobenzene first increases and then decreases. It is observed that 0.5 wt% Pd/MgAl 2 O 4 shows an optimum loading for higher conversion of ethylbenzene and it is dropped from 51.8 to 15.9 % and similarly nitrobenzene conversion dropped from 47.3 to 15.6 % with further increase in the palladium loading from 0.5 to 4.0 wt%. Thus, conversions of ethylbenzene decreased progressively with increase in palladium loadings beyond 0.5 wt% are due to low dispersion of Pd and higher particle size, which is clearly observed from the CO chemisorption analysis ( Table 1).
Effect of temperature place with the increase in temperature from 673 to 823 K in simple and coupling reactions, respectively. The similar trend is reflected in the conversion of nitrobenzene which is increased from 8.1 to 47.3 % with the production of more in situ hydrogen. The increase in the conversion of ethylbenzene with rise in temperature can be attributed to shift in the equilibrium to the right favoring the forward reaction. The selectivity of styrene decreases with increase in temperature due to formation of higher amount of cracking and hydrogenolysis products such as benzene and toluene. It is interesting to observe that the selectivity of aniline maintained maximum at all the reaction temperatures.   It is observed that there is a slow decline in the conversion of nitrobenzene whereas selectivity of aniline is found to be maximum with TOS.

Conclusion
It was found that the equilibrium conversion could be greatly enhanced by the reaction coupling. When DHEB was coupled with nitrobenzene hydrogenation, the ethylbenzene conversion can reach 51.8 % at 823 K, compared with the conversion of 22.4 % at the same temperature for the simple ethylbenzene dehydrogenation. The effect of dispersion and particle size of palladium of Pd/MgAl 2 O 4 spinel catalysts is significant in the coupling reaction. It is observed that 0.5 wt% Pd on MgAl 2 O 4 has exhibited maximum catalytic activity. The reason for the higher activity is attributed to the higher dispersion and lower particle size of Pd as observed from CO chemisorptions, TEM and TPR studies. Thus, MgAl 2 O 4 -supported palladium catalyst is found to be a potential candidate in the coupling reaction for production of styrene from ethylbenzene and aniline from nitrobenzene in a single reactor.
The primary experiments on series of catalysts proved that the reaction coupling was an effective measure to improve the ethylbenzene dehydrogenation, although much more work is still necessary to develop proper catalysts for the coupling reactions.