Hydrogenation of Nitrobenzene Over Au/MeOx Catalysts—A Matter of the Support

The heterogeneous hydrogenation of substituted nitrobenzenes is a reaction of great interest, because aniline and its derivates are valuable substances in the chemical industry for the production of polymers, pharmaceuticals, herbicides, and dyes.1 The state-of-the-art catalysts are mostly active metals, such as Pt, Pd, Ni, Cu, and Ir, which are supported on various materials, such as activated C, CaCO₃, and SiO₂, depending on their application.2 To achieve high selectivity to substituted anilines in the presence of other reducible groups and to prevent arylhydroxylamine accumulation in the reaction mixture, state-of-the-art catalysts are often modified with environmentally harmful additives, such as Pb and V promoters and Fe salts.3 Since the discovery that Au, when present as nanoparticles in the range of 1–3 nm, catalyzes CO oxidation, more and more reactions have been shown to be catalyzed by Au,4 among them the hydrogenation of nitrobenzene.5 Hydrogenation of nitroaromatics containing additional unsaturated groups over unmodified Au/TiO₂ and Au/Fe₂O₃ shows a high selectivity to the nitro group. Thus, Au/MeO_x (Me corresponds to a metal) catalysts have been presented as a “green” alternative in reactions where a high selectivity under moderate reaction conditions is required.

Haber proposed a reaction scheme (Scheme 1) for the electrochemical hydrogenation of nitrobenzene and its derivates in 1898;7 however, there is an ongoing debate about the reaction mechanism over heterogeneous catalysts. Haber proposed two main reaction routes, namely the “direct” (left hand side) and the “condensation” route (right hand side). In the direct route, nitrobenzene is reduced to nitrosobenzene, then to phenylhydroxylamine, and finally to aniline (Steps I–III). A variation of the direct route is the “no-nitroso route” (Step IV), in which nitrobenzene directly reacts to phenylhydroxylamine and then to aniline.8 The condensation route occurs when the two intermediates nitrosobenzene and phenylhydroxylamine condensate to form azoxybenzene (Step VI). This species is then hydrogenated to aniline in consecutive steps via the intermediates azobenzene and hydrazobenzene (Steps VII–IX).

Scheme 1. Possible reaction pathways for the hydrogenation of aromatic nitro compounds to the corresponding anilines. NB: nitrobenzene, NSB: nitrosobenzene, PHA: phenylhydroxylamine, AN: aniline, AOB: azoxybenzene, AB: azobenzene, HAB: hydrazobenzene. Adapted from Ref. 6.

Another possible step in the transformation of nitrobenzene to aniline is the decomposition of phenylhydroxylamine into nitrosobenzene and aniline (Step V). Aniline is produced by the disproportion of phenylhydroxylamine.9 The nitrosobenzene generated by the disproportion reenters the catalytic cycle and is subsequently transformed into phenylhydroxylamine. These findings are based on measurements of nitrobenzene hydrogenations over Ir/C poisoned by Hg.

Azoxybenzene is the first intermediate that is formed in the condensation route, which is observed when the reactions are performed in the presence of a base. Azoxybenzene can also be detected at slow reaction rates, for example, over Pd/SiO₂ in methanol at 25 °C.10

Recently, the selective catalytic hydrogenation of functionalized nitroarenes has been reviewed.11 The authors describe precisely the tailoring of selective catalysts by using organic and inorganic modifiers and their application for different catalytic problems. Also, the effect of solvent, particle size, and support are discussed. The discussion on the influence of the support focuses on selectivity, activity, and stabilization of the metal nanoparticles. Other reports detail the effect of the composition of the reaction mixture,12 the noble metal,13 and the support on selectivity, activity, or stability. Activity and selectivity of p-chloronitrobenzene hydrogenation over Pt/MeO_x catalysts is affected by the support.14 Furthermore, the dehalogenation of halogen-modified nitroaromatics using BaCO₃ as support has been measured.15 Metal oxides, such as TiO₂ and Fe₂O₃, which form a strong metal–support interaction, perform better in the preferential hydrogenation of nitrobenzene compared to “inert” supports, such as SiO₂ and activated C.16 To the best of our knowledge, there are no reports on how the support actively influences the mechanism of the reaction.

In this communication, we show that the support directly influences the reaction route in the hydrogenation of nitrobenzene in the liquid phase. Figure 1 shows the concentration profile during the reaction of nitrobenzene to aniline at 100 °C under 10 bar (1 bar=10⁵ Pa) of H₂ over Au/TiO₂ (Figure 1 a) and Au/CeO₂ (Figure 1 b). Over Au/TiO₂, the profile is dominated by the substrate nitrobenzene and the final product aniline. Only traces (<1 %) of the intermediate azoxybenzene were detected, which is in good agreement to previous results.5, 17 The evolution of the nitrobenzene concentration proceeded as a mirror image of the aniline concentration. A slight variation in the C-balance was observed at the beginning of the reaction (between 0 and 15 min); however, the C-balance was always higher than 95 % throughout the reaction. Full conversion was achieved after 60 min. Hydrogenation of nitrobenzene over the Au/CeO₂ catalyst shows a very different behavior. In addition to the substrate nitrobenzene and the final product aniline, two intermediates of the condensation route, azoxybenzene and azobenzene, were also detected. The C-balance dropped within the first 20 min of the reaction and recovered as the formation of azoxybenzene and aniline was detected, probably because of adsorption of intermediates and/or product on the surface of the catalyst; the C-balance exceeded 90 % at all times. Azobenzene was first detected after 30 min of reaction. The concentrations of azoxybenzene, azobenzene, and aniline increased with reaction time. The azoxybenzene concentration started to decrease after 120 min, and coinciding with full conversion of nitrobenzene, the concentration of azoxybenzene was zero at approximately 180 min. At this point, only azobenzene and aniline were present in the solution, and the first compound was eventually completely converted into the second. The occurrence of condensation products for Au/CeO₂ and their absence for Au/TiO₂ catalysts was observed for both, different Au loadings and different particle sizes (see the Supporting Information).

Figure 1. Evolution of substrate, intermediates, product, and C-balance in the liquid-phase hydrogenation of nitrobenzene (0.8 mmol substrate, 10 bar H₂, 100 °C) over a) Au/TiO₂ and b) Au/CeO₂ (y-axis relative to the stoichiometric coefficient of aniline); (•) nitrobenzene, (▾) azoxybenzene, (□) azobenzene, (▪) aniline, and (○) C-balance.

The evolution of the species in solution for a reaction of azoxybenzene over Au/CeO₂, is shown in Figure 2. The amount of azoxybenzene decreased exponentially until azoxybenzene was fully converted after 250 min. Azobenzene was the intermediate product in the formation of aniline. Both azobenzene and aniline were detected in the liquid phase already after 5 min of reaction. The azobenzene concentration reached a maximum at approximately 90 min, after which it steadily dropped until the end of the reaction. At 250 min, the azobenzene concentration was approximately 25 % of its maximum concentration at 90 min. The C-balance slightly fluctuated in the first 15 min of the reaction and remained above 95 % throughout the whole reaction.

Figure 2. Evolution of substrate, intermediates, product, and C-balance in the liquid-phase hydrogenation of azoxybenzene (0.4 mmol substrate, 10 bar H₂, 100 °C) over Au/CeO₂ (y-axis relative to the stoichiometric coefficient of aniline); (▾) azoxybenzene, (□) azobenzene, (▪) aniline, and (○) C-balance.

The evolution of the different species detected in the liquid phase using nitrosobenzene as the starting material is shown in Figure 3 a for Au/TiO₂ and in Figure 3 b for Au/CeO₂. Because of earlier results, which showed that a high nitrosobenzene concentration poisoned the reaction,6 we decreased the concentration of nitrosobenzene in this experiment from 0.8 to 0.4 mmol, which did not result in catalyst poisoning. All other reaction conditions were the same as in Figure 1. Over Au/TiO₂ (Figure 3 a), the nitrosobenzene concentration dropped to zero within the first 10 min of the reaction. After 5 min, azoxybenzene, azobenzene, and aniline were already detected in the liquid phase. Both azoxybenzene and azobenzene reached maxima after 5 min of reaction. Azoxybenzene was already completely converted after 15 min, whereas the amount of azobenzene constantly dropped throughout the reaction and reached 0 mmol after 120 min. The aniline concentration increased steadily during the reaction. The C-balance strongly decreased during the first 20 min of the reaction to 70 %, but then increased with a similar slope as the aniline concentration, and finally reached a value exceeding 90 %.

Figure 3. Evolution of substrate, intermediates, product, and C-balance in the liquid-phase hydrogenation of nitrosobenzene (0.4 mmol substrate, 10 bar H₂, 100 °C) over a) Au/TiO₂ and b) Au/CeO₂ (y-axis relative to the stoichiometric coefficient of aniline); (⧫) nitrosobenzene, (▾) azoxybenzene, (□) azobenzene, (▪) aniline, and (○) C-balance.

The conversion rate of nitrosobenzene was lower over Au/CeO₂ (Figure 3 b). Again, the maximum azoxybenzene concentration was reached after 5 min. In contrast to Au/TiO₂, the azobenzene concentration reached its maximum after 15 min, and azobenzene was slowly converted into aniline. The C-balance had an unconventional evolution, as it started at 60 % and rose throughout the reaction to 100 % at the end of the reaction. Intermediates probably formed at the beginning of the reaction, which were not detected because adsorption on the catalyst led to an imperfect C-balance in the liquid phase. Nitrosobenzene is a candidate for such an adsorbed intermediate.

The concentration profile of the nitrobenzene hydrogenation over Au/TiO₂ with addition of CeO₂ is shown in Figure 4 a. In contrast to the CeO₂-free reaction, azoxybenzene and azobenzene were detected in a similar time dependence as the reaction of nitrobenzene over Au/CeO₂ (Figure 1 b). The C-balance was >95 % throughout the reaction, except after 5 min, when it dropped to approximately 80 % at the onset of azoxy- and azobenzene production. In contrast to the reaction over pure Au/TiO₂ (Figure 1 a), the nitrobenzene and aniline concentrations did not mirror each other, which would be indicative for the formation of reaction intermediates. Adding TiO₂ to Au/CeO₂ did not lead to any changes in the concentration profile of pure Au/CeO₂ (not shown).

Figure 4. a) Evolution of substrate, intermediates, product, and C-balance in the liquid-phase hydrogenation of nitrobenzene (0.8 mmol substrate, 10 bar H₂, 100 °C) on a Au/TiO₂ catalyst with the addition of an equal mass amount of pure CeO₂ (y-axis relative to the stoichiometric coefficient of aniline); (•) nitrobenzene, (▾) azoxybenzene, (□) azobenzene, (▪) aniline, and (○) C-balance. b) Evolution of the nitrobenzene concentration in the liquid-phase hydrogenation of nitrobenzene (0.8 mmol substrate, 10 bar H₂, 100 °C) on a Au/TiO₂ catalyst with (

• 

) and without the addition of pure CeO₂ (

• 

).

The nitrobenzene concentration profiles of the experiments with (hexagons) and without (pentagons) the addition of CeO₂ are shown in Figure 4 b. The nitrobenzene concentration dropped faster in the first 20 min when CeO₂ was added to the mixture. The enhanced conversion of nitrobenzene after addition of CeO₂ correlated quantitatively with the amounts of azo- and azoxybenzene, suggesting that CeO₂ initiated an additional parallel reaction channel in the reaction. Hydrogenation of all the known intermediates in the direct and condensation route over bare TiO₂ and CeO₂ were tested to identify the differences between the two supports. The only difference was found in the reaction of phenylhydroxylamine, which is a rather unstable species and which decomposed faster into nitrosobenzene and aniline in the presence of CeO₂ than in the presence of TiO₂. For the hydrogenation of nitrobenzene, both direct and condensation routes were observed. Azoxybenzene could be formed through the condensation of nitrosobenzene and phenylhydroxylamine. Thus, for the direct route to occur, nitrosobenzene and phenylhydroxylamine should be present in low concentrations to prevent azoxybenzene formation. However, the intermediates phenylhydroxylamine and nitrosobenzene can also form azoxybenzene by themselves. Phenylhydroxylamine easily disproportionates into nitrosobenzene and aniline.18 The so-formed nitrosobenzene can then condensate with another molecule of phenylhydroxylamine to form azoxybenzene. In the case of nitrosobenzene, two molecules can form a dimer and then react with another molecule of nitrosobenzene to form azoxybenzene and nitrobenzene.19 Thus, for the condensation route to occur, accumulation of nitrosobenzene and/or phenylhydroxylamine is necessary to form azoxybenzene according to one of the ways described above, which is subsequently converted into aniline via azobenzene. The direct route dominates over Au/TiO₂,17b, 18 and accumulation of phenylhydroxylamine on the surface of the catalyst has been observed.6 We found that hydrogenation was sufficiently fast over Au/TiO₂ to prevent condensation of phenylhydroxylamine with nitrosobenzene, which was formed either from nitrobenzene or by phenylhydroxylamine decomposition, but slower than over Au/CeO₂. Phenylhydroxylamine has been found as an intermediate on the surface of the catalyst during the liquid-phase hydrogenation of nitrobenzene by using liquid-phase attenuated-total-reflectance (ATR) IR measurements.6 FTIR measurements have revealed that nitrosobenzene reacts to aniline via phenylhydroxylamine in the gas phase.17a In contrast, we identified azoxybenzene as the primary product by using liquid-phase experiments, which was transformed into aniline via azobenzene over both Au/TiO₂ and Au/CeO₂. Most likely, the condensation route was suppressed because the molecules that form azoxybenzene could not condensate in the gas phase. Thus, the reaction over Au/TiO₂ can also proceed through the condensation route when the concentration of nitrosobenzene is high enough.

Au/TiO₂ rapidly converts nitrosobenzene, which is also converted rapidly over Au/CeO₂; however, the large loss of C-balance suggests that initially a large amount adsorbs on CeO₂, which leads to a high surface concentration. Hydrogenation of nitrobenzene over Au/CeO₂ (Figure 1 b) proceeds through the condensation route and at a considerably slower rate than over Au/TiO₂. This lower hydrogenation rate and the fast decomposition of phenylhydroxylamine could lead to a buildup of nitrosobenzene molecules on the surface of CeO₂, which allows for the condensation route to occur. Adsorption of large amounts of nitrosobenzene can be assumed because the C-balance is reduced directly after exposing CeO₂ to nitrosobenzene (Figure 3 b). Adding CeO₂ to Au/TiO₂ yields the condensation products azoxybenzene and azobenzene, which are probably formed through nitrosobenzene accumulation on the CeO₂ surface. The nitrobenzene concentration drops faster when CeO₂ is added to the reaction mixture when Au/TiO₂ is used as the catalyst. In addition to the direct route that occurs over Au/TiO₂, CeO₂ catalyzes the condensation route through accumulation of phenylhydroxylamine and its decomposition into nitrosobenzene or through accumulation of nitrosobenzene directly. The result is a reaction that occurs through the condensation route. The direct route is not strongly affected, probably because the surface concentration of phenylhydroxylamine on Au/TiO₂ remains sufficiently high to poison the reaction.6 The hydrogenation of nitrobenzene over Au/TiO₂ proceeds through the direct route, whereas the hydrogenation reaction over Au/CeO₂ proceeds through the condensation route. For the condensation route to occur, a high (surface) nitrosobenzene concentration is necessary. In the case of Au/TiO₂, nitrosobenzene is rapidly converted into phenylhydroxylamine, which accumulates on the surface and is then transformed to aniline. The concentration of nitrosobenzene is never high enough to form azoxybenzene. For Au/CeO₂, the rate of hydrogenation is considerably lower, and the conversion of nitrobenzene and nitrosobenzene are slower; as a result, nitrosobenzene can accumulate and form condensation intermediates. An additional path to nitrosobenzene is the decomposition of phenylhydroxylamine, which is especially fast over the CeO₂ support. Furthermore, the CeO₂ support catalyzes the condensation. The support has a direct impact on the reaction mechanism and actively changes the reaction route.

Experimental Section

Synthesis of materials

TiO₂ (Aeroxide P25) was supplied by Acros, and CeO₂ by MicroCoating Technologies. A 1.5 wt % Au/TiO₂ (actual Au loading: 1.48 wt %) catalyst was supplied by the World Gold Council (WGC, reference catalyst Type A: 1.5 wt % Au/TiO₂). 1 wt % Au/CeO₂ (actual Au loading: 0.7wt %) catalyst was synthesized by the deposition/precipitation of Au onto the surface of CeO₂ using urea according to 20. Prior to synthesis, CeO₂ was calcined at 500 °C to remove all residuals remaining from synthesis. In a teflon container, the support (3 g) was dissolved in water (300 mL) containing urea (0.9 g). HAuCl₄⋅3 H₂O (50.97 mg; Au: 49 %, 99.9 % metal basis, ABCR-Chemicals) was added to this solution. Subsequently, the mixture was stirred at 80 °C for 16 h under the exclusion of light. The precipitate was filtered, washed three times with water to prevent Cl⁻ contamination, and dried in vacuum over night. The resulting catalysts were referred to as “as-prepared” and were pretreated as described in the paragraph Kinetic measurements.

By using a Varian SpectrAA 220 FS spectrometer, atomic absorption spectrometry (AAS) analysis was used to determine the loading of Au (wt %) in each catalyst. To achieve that, the catalyst (100 mg) was dissolved in aqua regia overnight (6 mL), and then water was added to obtain a total volume of 50 mL. This solution was compared to a standard calibration series.

Kinetic measurements

All catalysts were pretreated as follows before the reaction was performed. The catalyst (200 mg) was heated to 300 °C (heating rate: 2 °C min⁻¹) in a stream of 5 % (v/v) H₂ in He (flow rate: 100 mL min⁻¹). After reaching 300 °C, the catalyst was kept at that temperature for 60 min and then cooled down to room temperature in the same stream. We refer to this method as 300H₂.

Kinetic measurements were performed in 50 mL Premex autoclaves. A typical reaction composition consisted of toluene (25 g), internal standard (120 mg; mesitylene, puriss>99.0 %, Sigma–Aldrich), catalyst (100 mg), and substrate (0.8 mmol: nitrobenzene, ≥99.5 %, Sigma–Aldrich, and n-phenylhydroxylamine, ≥98 %, Sigma–Aldrich; 0.4 mmol: azoxybenzene 97 %, Sigma–Aldrich, and nitrosobenzene 98 %, ABCR-Chemicals). This mixture was filled into an autoclave and purged three times after sealing by using H₂ (5 bar; 1 bar=10⁵ Pa) and while stirring. The autoclave was then pressurized to 10 bar by using H₂ and heated up to 100 °C under constant stirring, during which the reaction started. Samples were taken at fixed time intervals by means of a sample tube. The samples were filtered and analyzed by using GC using an Agilent 7820 A with an apolar 30 m HP 5 MS column. The temperature program consisted of 20 °C min⁻¹ heating steps from 80–300 °C at a split ratio of 1/10.

Acknowledgements

We thank Andreas Dutly and his staff for support in GC analysis. This work was funded by the SNF grant Nr. 200021_124458.