The XRD patterns of MCM-41 and iron-modified MCM-41 at low diffraction angles (2θ = 1°–10°) range, as well as high diffraction angles (2θ = 10°–60°) range, were displayed in Fig. 1. As shown in the wide diffraction angles spectra, without modification, only a broad diffraction peak at around 22° was observed, which was assigned to the amorphous feature of framework of MCM-41. On the contrary, compared with the bare MCM-41, for all iron-modified MCM-41 samples, the broad peak at 22° was hardly seen. It was reported previously that the metal-acetylacetonate precursors can interact with silica support surface by two methods (Xu et al., 2018b). Firstly, metal-acetylacetonate compounds could connect with silica surface through H-bond. Secondly, covalent metal—O—Si bond could be formed between metal-acetylacetonate compounds and silanol groups through ligand exchange. Therefore, the disappearance of the broad peak at 22° might be ascribed to the interaction between Fe(acac)3 precursor and abundant silanol groups of MCM-41 support, forming rich Fe—O—Si bond. Besides, a diffraction peak at 2θ = 35°, which was corresponding to Fe species, was clearly seen for all iron-modified MCM-41 samples, and with iron loading increasing, the intensity of this peak increased correspondingly. This phenomenon revealed that small iron oxide clusters or nanoparticles was formed either inside the mesopores or on the external surface with Fe content increasing. The bare MCM-41 and iron-modified MCM-41 samples were also characterized by Low-angle XRD measurements. As shown in the low diffraction angles spectra, compared with the bare MCM-41, the diffraction peak at around 2° of all iron-modified MCM-41 samples shifted to a lower angle region, suggesting a slight framework expansion of MCM-41 zeolite after iron modification. This framework expansion might be attributed to the electrostatic attraction interaction between iron and lattice oxygen atom, and the insertion of a fraction of iron into the silica matrix after iron modification, which affected the hexagonal unit cell parameters of MCM-41 zeolite. The diffraction peak at about 2° of all iron-modified MCM-41 samples was still observed, indicating that the mesoporous structure of iron-modified MCM-41 was preserved after the Fe species were inserted. However, the diffraction peaks lying in the 2θ range of 3°–5° of 20Fe-MCM-41 broadened, revealing a slight decrease of the structural regularity of iron-modified MCM-41 with the content of Fe increasing.
Figure 1. The XRD patterns of MCM-41 and iron modified MCM-41 in the low angle range (a) and high angle range (b)
The FT-IR results of pure MCM-41 and iron-modified MCM-41 sample (20Fe-MCM-41 as represent) are shown in Fig. 2. It is found that vibration peak at about 3500 cm–1, which was assigned to Si—OH, weakened obviously. This phenomenon proved the interaction between Fe precursor and silanol groups of MCM-41 support.
The N2 adsorption-desorption isotherms of bare MCM-41 and iron-modified MCM-41 samples, as well as the pore size distribution of bare MCM-41 and iron-modified MCM-41 samples, are displayed in Fig. 3. The textural properties of bare MCM-41 and iron-modified MCM-41 samples are summarized in Table 1. It can be seen that all iron-modified MCM-41 samples appeared the type-Ⅳ isotherms and hysteresis loops at P/P0 regions of 0.45–0.90, which was similar with pure silica support, MCM-41. It indicated that the iron modification could not destroy the well-defined meso porous structure of MCM-41 support. The pore size distribution showed that the pore size of iron-modified MCM-41 distributed widely in the range of 2–4 nm, but the pore size of pure MCM-41 distributed narrowly at around 3 nm. As summarized in Table 1, pure silica MCM-41 support exhibited high BET surface area (899 m2/g) and large total pore volume (0.84 cm3/g) respectively. However, after insertion of iron into MCM-41 support, the BET surface area and total pore volume of all iron-modified MCM-41 samples decreased significantly. In the case of 20Fe-MCM-41, the BET surface area and total pore volume decreased to 797 m2/g and 0.50 cm3/g, respectively, which might be attributed to the existence of iron oxide clusters or nanoparticles on the external surface or inside the pore of iron-modified MCM-41 with high iron loading.
Figure 3. The N2 adsorption-desorption isotherms (a) and pore size distribution (b) of MCM-41 and iron modified Fe-MCM-41
Catalyst BET surface area (m2/g)a Total pore volume (cm3/g)b MCM-41 899 0.84 5Fe-MCM-41 871 0.75 10Fe-MCM-41 834 0.67 20Fe-MCM-41 797 0.50 Notes: a, multipoint BET; b, determined from the amount adsorbed at P/P0 = 0.95.
Table 1. Textural properties of MCM-41 and iron modified Fe-MCM-41
The SEM and high-resolution TEM were employed to investigated the morphology and detailed structure characteristics of iron-modified MCM-41 (20Fe-MCM-41 as representative sample), respectively. As shown in Fig. 4, 20Fe-MCM-41 displayed an irregular, layered morphology. The TEM image showed uniform and parallel channels of samples, which confirmed that the insertion of iron species into MCM-41 support could not destroy the mesoporous structure.
The FT-IR after pyridine adsorption was employed to study the acidic properties of iron-modified MCM-41 catalysts and results are shown in Fig. 5. Compared with the bare MCM-41 samples, all iron-modified MCM-41 samples showed an obvious adsorption peak at 1445 cm–1, which was ascribed to pyridine-Lewis acid adduct. In addition, the adsorption peak of pyridine adsorbed on Bronsted acid sites was not observed at 1540 cm–1 for all iron-modified MCM-41 samples. These phenomena revealed the acidity of iron-modified MCM-41 originated mainly from Lewis acid sites. In addition, with iron loading increasing, the intensity of the adsorption peaks increased correspondingly, which indicated that the Lewis acidity of iron-modified MCM-41 sample increased with iron content of sample increasing. When iron content of sample was high, a new adsorption peak at about 1490 cm–1 was observed, which could be ascribed to the chemisorbed pyridine on both Lewis acid and Bronsted acid sites. This phenomenon suggested that the acid sites distribution of iron-modified MCM-41 changed slightly when iron content of sample was high. The result indicated that the iron-modified MCM-41 catalyst might be a potential, beneficial catalyst in aldol condensation of furfural/HMF with acetone.
The XPS was used to distinguish the chemical composition of Fe on the surface of 20Fe-MCM-41. According to previous literature, four peaks can be obtained after deconvolution of Fe 2p spectrum. As shown in Fig. 6, the binding energy (BE) of four deconvolution peaks were considered at 711.7, 717.3, 725.4 and 733.6 eV, respectively. The BE at 711.7 eV was corresponded to Fe 2p3/2 and the BE at 725.4 and 733.6 eV were associated with Fe 2p1/2, which was ascribed to the existence of iron (Ⅲ). It was widely reported that the BE of pure iron oxide was usually observed at 710.6 and 711.2 eV. Therefore, the different BE for Fe 2p3/2 in 20Fe-MCM-41 was an evidence for the interaction between Fe species and framework of silica MCM-41 support. In the meantime, the BE at 717.3 eV was an implication of the existence of isolated Fe3+ bound to surface O atoms.
The catalytic performances of two representatives of different types of catalysts were shown in Table 2. As the representative of mixed metal oxide catalysts, MgO-ZrO2 showed 59.9% furfural conversion with 100% selectivity to aldol products. The furfural conversion of 20Fe-MCM-41 was much higher than that of MgO-ZrO2, but the selectivity of 20Fe-MCM-41 was lower than that of MgO-ZrO2. This difference of selectivity between 20Fe-MCM-41 and MgO-ZrO2 could be due to the pore structure of zeolite. More side reaction might occur within the pore of 20Fe-MCM-41 compared with outer surface of MgO-ZrO2. As the representative of zeolite catalysts, Sn-MFI showed 90.0% furfural conversion, which was almost the same as 20Fe-MCM-41. However, the selectivity of 20Fe-MCM-41 was much higher than that of Sn-MFI. The possible reason was that F2Ac could escape from the mesopore of 20Fe-MCM-41 but could not escape from the micropore of Sn-MFI.
catalyst Furfural conversion (%) Product selectivity (%) 20Fe-MCM-41 86.9 86.3 MgO-ZrO2 59.9 100.0 Sn-MFI 90.0 66.6
Table 2. Catalytic performances of two representatives of different types of catalysts
Aldol condensation reaction between furfural/HMF and acetone was studied by a series of iron modified MCM-41 catalysts and results are shown in Fig. 7. It could be clearly seen that all iron modified MCM-41 catalysts display a significantly improving catalytic activity compared with pure MCM-41 catalyst. This result indicated that Fe species was a feasible Lewis acid catalytic site for aldol condensation reaction. In addition, with Fe loading increasing, the catalytic activity of iron modified MCM-41 increased correspondingly. For furfural, 5Fe-MCM-41 showed 67.7% furfural conversion with 26.8% yield of FAc and 1.5% yield of F2Ac, respectively. When Fe loading increased 20 wt%, furfural conversion achieved 86.9% with 60% yield of FAc and 7.5% yield of F2Ac, respectively. For HMF, when 20Fe-MCM-41 was used as the catalyst, 41.1% yield of HAc and 3.5% yield of H2Ac with 88.9% HMF conversion was achieved. However, when 5Fe-MCM-41 was used as the catalyst, only 25.3% yield of HAc and 0.8% yield of H2Ac with 60.5% HMF conversion was achieved. This enhanced catalytic activity of iron modified MCM-41 with higher Fe loading was ascribed to more Lewis acid sites possessed by higher Fe loading catalyst according to FT-IR after adsorbed by pyridine. In addition, iron modified MCM-41 showed preferred selectivity to aldol condensation reaction between furfural and acetone, compared with HMF and acetone. As shown in Table 3, in the case of almost same conversion, 20Fe-MCM-41 showed enhanced selectivity to condensation products of furfural compared with HMF. The 69.0% selectivity to FAc as well as 17.3% selectivity to F2Ac, 86.3% total selectivity, was achieved, which was much higher than 46.2% selectivity to HAc and 7.9% selectivity to H2Ac, 54.1% total selectivity.
Figure 7. Catalytic performance of MCM-41 and iron modified MCM-41 catalysts in aldol condensation reaction with furfural of acetone (a) and with HMF of acetone (b)
Reactant Conversion (%) FAc/HAc selcetvity (%) F2Ac/H2Ac selectivity (%) Total selectivity (%) Furfural 86.9 69.0 17.3 86.3 HMF 88.9 46.2 7.9 54.1 Notes: reaction conditions are time, 24 h; temperature, 160 ℃; catalyst loading, 0.1 g; and the molar ratio of furfural or HMF to acetone = 1꞉20.
Table 3. Selectivity to condensation products between furfural or HMF and acetone over 20Fe-MCM-41
The effect of reaction time, varying from 3 h to 24 h, on aldol condensation of furfural or HMF with acetone over 20Fe-MCM-41 catalyst was investigated. As exhibited in Fig. 8, with reaction time increased from 3 h to 24 h, furfural conversion increased from 35.1% to 86.9% and FAc, as well as F2Ac yield, increased correspondingly from 11.4% to 60.0% and from 1.0% to 7.5%, respectively. This result revealed that prolonging reaction time could effectively increase the yield of condensation product between furfural and acetone. The same trend was also observed in aldol condensation of HMF with acetone. The HMF conversion increased from 39.9% to 88.9% with HAc as well as H2Ac yield increasing from 15.4% to 41.1% and from 0.6% to 3.5%, respectively, when the reaction time was prolonged from 3 h to 24 h. It was worth noting that after 9 h, HAc yield was exceeded by FAc yield although HMF conversion was still higher than furfural conversion. It possibly was ascribed that more side reaction was occurred in aldol condensation reaction between HMF and acetone, which made that the catalyst was poisoned more seriously, compared with aldol condensation reaction between furfural and acetone. The thermo gravimetric analysis (TGA) results could certify this suppose. The mass loss of spent catalyst after reaction with furfural of acetone was 0.1%, but the mass loss of spent catalyst after reaction with the HMF of acetone was 2.2%. This result lead to the gap of selectivity of catalyst in aldol condensation between HMF with acetone and furfural with acetone was getting wider, which was obviously displayed in Fig. 9. It was seen that before 9 h, the average selectivity of catalyst was almost same, remaining at about 37%, in aldol condensation reaction between HMF with acetone and furfural with acetone. However, after 9 h, the selectivity to FAc remained at about 70% but the selectivity to HAc was only about 46%. Different with FAc (HAc) yield, the effect of reaction time on F2Ac (H2Ac) yield was not remarkable, which was mainly attributed that the competitive adsorption between FAc (HAc) and furfural (HMF) was more difficult than acetone and furfural (HMF) because of high molar ratio of acetone/furfural (HMF). Therefore, the process of production of F2Ac (H2Ac) was not a major reaction, which made that the influence of reaction time on F2Ac (H2Ac) yield was not as conspicuous as FAc (HAc) yield.
Figure 8. Effect of reaction time on aldol condensation reaction with furfural or HMF of acetone over 20Fe-MCM-41 catalyst
The effect of reaction temperature, varying from 80 ℃ to 160 ℃, on aldol condensation of furfural or HMF with acetone over 20Fe-MCM-41 catalyst was also studied. As shown in Fig. 10, furfural conversion increased from 17.8% to 86.9% and FAc yield, increased correspondingly from 1.4% to 60.0%. This result indicated that enhancing reaction temperature could effectively boost the yield of condensation product between furfural and acetone. The same trend was also obtained in aldol condensation of HMF with acetone. The HMF conversion increased from 4.2% to 88.9% with HAc yield increasing from 1.1% to 41.1%, when the reaction temperature was enhanced from 80 ℃ to 160 ℃. In addition, F2Ac and H2Ac, were only observed when reaction temperature was above 140 ℃. This phenomenon revealed that higher reaction temperature would benefit the competitive adsorption of FAc (HAc) with furfural (HMF), which led to higher F2Ac (H2Ac) yield.
Figure 10. Effect of reaction temperature on aldol condensation reaction with furfural of acetone (a) and with HMF of acetone (b) over 20Fe-MCM-41 catalyst
The reusability and regeneration research of iron modified MCM-41 in aldol condensation between furfural and acetone was investigated. Figure 11 summarized the reusability (run 2) and regeneration (run 3, 4 and 5) studies of 20Fe-MCM-41 catalyst. After the first run, attempt was made to recycle the spent catalyst via just washing with acetone thoroughly and dried at 100 ℃. As shown in Fig. 11 (run 2), the catalytic behavior of this as-treated spent catalyst decreased obviously. The furfural conversion decreased from 86.9% to 69.5% and the yield of FAc and F2Ac decreased from 60.0% to 36.9% and from 7.5% to 3.0%, respectively. The result indicated that 20Fe-MCM-41 catalyst can not be recycled through only washing with acetone after reaction. The regeneration of 20Fe-MCM-41 was performed by calcination at 550 ℃ in air for five hours after washing with acetone and drying at 100 ℃. The regenerated 20Fe-MCM-41 displayed almost the same catalytic behavior compared with the first run. The furfural conversion was 85% with a 58% yield of FAc and a 6.9% yield of F2Ac, respectively (run 3). The 20Fe-MCM-41 was regenerated by washing, drying and calcination in air (the same way as run 3) in the following two runs (run 4 and run 5). Although the furfural conversion and condensation product yield continuously decreased, the catalytic performance of 20Fe-MCM-41 catalyst after regeneration was acceptable. The reusability and regeneration research of iron modified MCM-41 in aldol condensation of HMF with acetone was also studied and the catalytic performance was similar to the result of aldol condensation of furfural with acetone.