2.1 Materials and catalyst synthesis

2.2 Catalyst characterization

2.3 Catalytic measurements

CO hydrogenation reactions were performed using a high-pressure fixed-bed microreactor (stainless steel 316#, 200 mm length and 4 mm internal diameter). First, 0.5 g catalyst was reduced in situ at 673 K (heating rate of 1 K min⁻¹) for 16 h in a flow of H₂/N₂ (1:9) at a rate of 100 cm³ min⁻¹. The reactor was then cooled down and the gas flow was switched to 75 cm³ min⁻¹ (or 60 cm³ min⁻¹ or 45 cm³ min⁻¹) of syngas (molar ratio H₂/CO = 2). The system was pressurized at 1.0 MPa and heated to the desired temperatures at a heating ramp of 10 K min⁻¹. Temperature was measured with a type-K thermocouple buried in the catalytic bed. Flow rates were controlled using a Brooks 5850 TR Series mass flow controllers. All experimental variables were carefully controlled to ensure identical reaction conditions when testing catalysts. Some recommendations and criteria proposed by Pérez-Ramírez et al. [13] were followed to ensure accurate measurements.

Product analysis was performed online with a gas chromatograph (GC2014) equipped with TCD and FID detectors and two in-series fused silica capillary columns: SPB-5 (60 m × 0.53 mm) and Supel-Q Plot (30 m × 0.53 mm). It was possible to analyze inorganic gases such as H₂, CO and CO₂ etc., C₁–C₂₀ hydrocarbons, C₁–C₁₀ alcohols, and other oxygenated compounds. The identification of reaction products and gas chromatograph calibration were accomplished using gas chromatography-mass spectrometry (GC-MS) and quantitative standard solutions from AccuStandard.

3.1 CO hydrogenation results

Manganese loading plays a crucial role in promoting the yield of oxygenated compounds, as shown in Table 1. Only a trace of ethanol was produced on the unpromoted Rh/SBA-15 catalyst, although high CO conversion was observed. It indicated that a completely promoter-free rhodium catalyst could not produce ethanol. When the Mn/Rh ratio increased from 0.5 to 2, the average CO conversion increased from 8.2 to a maximum 21.9% at Mn/Rh = 1, then decreasing to 9.3%. The yield of liquid products also presented the same trend, increasing from 120.6 to a maximum 508.4 g (kg-cat h)⁻¹ at Mn/Rh = 1, then decreasing to 182.4 g (kg-cat h)⁻¹.

The effects of flow rate of syngas on the conversion and selectivity with the 5 wt% Rh-5 wt% Mn/SBA-15 catalyst are summarized in Table 2. As expected, CO conversion decreased with increasing flow rate. Obviously, the product selectivities are also affected. Table 2 illustrates that increasing syngas flow rate from 60 to 90 cm³ min⁻¹ decreased the average CO conversion from 25.0 to 17.6% while simultaneously increasing the ethanol selectivity from 8.7% at a flow rate of 60 cm³ min⁻¹ to a maximum 12.9% at 75 cm³ min⁻¹ and then decreasing to 11.6% at 90 cm³ min⁻¹.

The effect of Rh loading on the activity and selectivity of Mn promoted Rh/SBA-15 catalysts (Mn/Rh weight ratio = 1) is shown in Table 3. As expected, CO conversion over the catalysts increased with increasing Rh loading. However, ethanol selectivity reached a maximum for 5 wt%Rh-5 wt%Mn/SBA-15, the formation of ethanol being suppressed at higher Rh loading. Fig. 1 illustrates how the CO conversion changed with time over a 13-h time interval. It should be noted that the catalyst 7 wt% Rh-7 wt% Mn/SBA-15 exhibits excellent activity. CO conversion is more than 90% in the first 3 h, and average CO conversion is 64.3%, as shown in Table 3. The increased catalytic performance is attributed to the increase in the amount of reducible oxide which resulted in an enhancement of the perimeter of the Rh-MnO interface when the Mn/Rh weight ratio is about 1, and the enhanced ability of rhodium to dissociate and to hydrogenate the carbon monoxide molecule by the addition of manganese. This point will be discussed in following sections.

3.2 Characterization of the catalysts by XRD

Fig. 2 presents the XRD patterns for the SBA-15-supported Rh samples with different loading amounts of manganese. In the case of parent SBA-15, three well resolved diffraction peaks, which are indexed as the (1 0 0), (1 1 0) and (2 0 0) reflections, associated with p6 mm hexagonal symmetry [8], are observed. With increasing loading of manganese, although peaks (1 1 0) and (2 0 0) became unobservable, indicating the de-organization at long range of the mesoporous structure, the peak (1 0 0) was clearly observed, which indicated that the hexagonal patterns were still retained on the impregnated catalysts.

3.3 Characterization of the catalysts by N₂ adsorption-desorption isotherms

Table 4 gives the textural properties of parent SBA-15 and 5 wt%Rh-5 wt%Mn/SBA-15 by the N₂ adsorption-desorption isotherms. Impregnation of SBA-15 with RhCl₃ and Mn(NO₃)₂ produces a decrease in specific surface area which is attributed to clogging support pores by RhCl₃ and Mn(NO₃)₂ species that makes them inaccessible for nitrogen adsorption. This phenomenon was already observed by Laugel et al. [14]. After RhCl₃ and Mn(NO₃)₂ introduction, the average volume of the catalysts becomes smaller but the average pore diameter does not change. These results suggest that the RhCl₃ and Mn(NO₃)₂ species are probably located inside the mesoporous channels of SBA-15 rather than dispersing on the wall surface. This is consistent with the results of Wang et al. [15], who examined the SBA-15-supported iron phosphate catalyst for partial oxidation of methane to formaldehyde, and believed that iron phosphate (FePO₄) species with loading amount lower than 40 wt% were located inside the mesoporous channels of SBA-15.

3.4 Characterization of the catalysts by TEM

TEM observations further support the encapsulation of RhCl₃ and Mn(NO₃)₂ species within the mesopores. As typical example, TEM images of the 5 wt%Rh-5 wt%Mn/SBA-15 are shown in Fig. 3. The hexagonal array of mesoporous channels is clearly observed. Moreover, small clusters encapsulated inside the mesoporous channels can also be discerned when Figs. 3B and C are compared. No large particles locating outside the mesopores are observed in the case of 5 wt% Rh-5 wt% Mn/SBA –15 (Figs. 3A and B).

3.5 Characterization of the catalysts by TPR

Fig. 4 shows TPR profiles for the catalysts investigated. The Rh-free Mn-supported SBA-15 sample exhibits two small reduction peaks at about 456 and 783 K, which can be ascribed to the reduction sequence MnO₂ → Mn₃O₄ → MnO [16]. This is based on thermodynamic arguments and experimental evidence from the literature [17]. The Rh/SBA-15 catalyst without promoter gives a reduction peak centered at 370 K, which corresponds to the reduction of Rh³⁺ to give metallic rhodium. The addition of the Mn promoter changed the reducibility of Rh and Mn oxides. Upon increasing the manganese loading, the reducibility of the rhodium was decreased, shifting the maximum of reduction peak from 370 K for the unpromoted catalyst to 400, 410 and 410 K for the samples with Mn/Rh = 0.5, 1 and 2 (Fig. 4C, D, E), respectively. At the same time, a remarkable shift to a lower reduction temperature of the manganese oxide is observed. The manganese oxide reduction peaks seem to overlap with that of the rhodium oxide in the case of 5 wt%Rh-5 wt%Mn/SBA-15 (Fig. 4C). The ability of a noble metal oxide to accelerate the reduction of a second surface metal oxide species with which it is in intimate contact has been reported previously [18]. Metallic rhodium provides sites for hydrogen adsorption and dissociation with extremely low activation energy. If MnO_x particles are close enough to the metal sites, their reduction by atomic hydrogen takes place at a lower temperature than if molecular hydrogen has to interact with MnO₂, and hence the reducibility of Mn oxides is enhanced by the close proximity of Rh [16]. These findings clearly support that a close interaction between Rh and Mn oxides existed in our catalysts.

With increasing manganese loading, the H₂ consumption increased to maximum at Mn/Rh = 1, and then decreased. The enhancement of H₂ consumption (Fig. 4D) was related to the increase in the amount of reducible oxide [19], which resulted in an enhancement of the perimeter of the Rh-MnO interface responsible for CO insertion [1]. It was one of reasons for the increased catalytic performances when the Mn/Rh weight ratio is about 1. This can be explained by taking into account two aspects: the amount of promoter added and the size of the MnO particles deposited. When the Mn content is low, the MnO particles deposited over the rhodium surface are rather small, and any increase in promoter loading would be reflected as an enhancement of the perimeter of the Rh-MnO interface, thus increasing the number of active CO insertion sites. However, when the Mn/Rh ratio is higher than 1, large MnO patches are deposited over the rhodium surface, and the Rh-MnO interface perimeter is diminished. So there exists an optimum promoter loading, that is Mn/Rh (weight ratio) = 1, at which the Rh-MnO interface has the highest perimeter.

3.6 Characterization of the catalysts by TPSR

The reactivity of chemisorbed CO under H₂ flow provides useful information about the ability of rhodium to dissociate and to hydrogenate the carbon monoxide molecule. Since the hydrogenation of carbon into methane is very fast on Rh-Mn based catalysts, methane production can be used as a tool for measuring the CO dissociation over such catalysts. Fig. 5 shows the production of CH₄ when the hydrogenation of chemisorbed CO is carried out from room temperature to 900 K.

Two effects were observed when manganese was added to the unpromoted Rh/SBA-15 catalyst. The first observation was that the temperature at which methane started to be formed and the corresponding maximum were shifted toward lower temperatures with increasing promoter loading. The methane formation peak shifted from 490 to 450, 440 and 430 K for Mn/Rh = 0.5, 1 and 2. This indicates that the addition of manganese increased the ability of rhodium to dissociate the CO molecules, which can strongly affect catalytic performance. This is consistent with the literature [20]. It can be explained by taking into account electronic effects. According to the well known model of CO adsorption over metal surfaces first postulated by Blyholder [21], the carbon monoxide-to-metal bond is considered to result from a charge transfer from the HOMO of CO (5σ) into a free σ symmetry d orbital of the metal and an electron back-donation from an occupied π symmetry d orbital of the metal into the unoccupied 2π* antibonding orbital of CO. This back-donation to the CO antibonding orbital weakens the CO bond. As an electron acceptor, the oxophilic promoter Mn ions coadsorbed CO with Rh on the same surface [20], therefore the back-donation to the antibonding molecular orbital is enhanced and, in turn, the ability of rhodium to dissociate the CO molecule is increased. The second observation was the increase in the methane peak area from B to D. In principle, it would be expected that increasing the amount of promoter could partially block the rhodium surface, thereby diminishing the number of active sites. Accordingly, the increment in the CO dissociate would be due to a higher catalytic activity per rhodium site, i.e., a higher turn over frequency. Sachtler and Ichikawa [20] believed that on the Rh-promoter interface the adsorbed CO could be readily dissociated. This effect became stronger as the Mn/Rh weight ratio increases.

So the increased ability of rhodium to dissociate the CO molecules by the addition of manganese was another reason for the improved catalytic performances of Rh-Mn/SBA-15.