2.1. Catalyst Characterization
Ni-based samples supported on MCM-41 and HZSM-5 were shortened to Ni/M and Ni/H, respectively. The MCM-41 supported Ni catalysts could be denoted as Ni/M xEG (molar ratio of Ni/EG was 1:x); when x equaled zero, it expressed that the catalysts were prepared by conventional wetness impregnation without EG promotion. The Ni/M(H) 1EG catalyst was prepared by co-impregnation after mixing MCM-41 with 50 wt % HZSM-5 by milling. The synthesis parameters such as drying temperature, calcination temperature, and Ni loading, which influenced the physical properties of the NiO nanoparticles, were investigated through the use of X-ray diffraction (XRD) in Figure 1
. In the diffraction patterns of the Ni/M catalysts, the broad and diffuse pattern observed clearly at around 2θ = 22.5° was attributed to amorphous silica. The samples showed diffraction lines at 37.2°, 43.2°, 62.8°, 75.3°, and 79.3°, indicating that nickel was present mainly in the form of the NiO structure after calcination. It was clear that it was pure cubic NiO (JCPDS #00-044-1159).
A shows the influence of drying temperature upon the physical properties of the NiO nanoparticles using the XRD technique. The calcination process was kept the same at 400 °C for 2 h in air with a heating rate of 2 °C/min. The drying temperature varied from 100 to 200 °C and had a clear impact on the particle size and distribution. The samples dried below 120 °C had low and diffuse peaks, confirming that extremely small NiO particles were formed and were well dispersed on the MCM-41 support; however, the samples with drying temperatures above 160 °C had slightly strong and sharp peaks, which meant that the size of NiO particles grew. The average crystal sizes of NiO increased gradually from 3.3 nm to 8.0 nm with an increase of the drying temperature from 100 to 200 °C. For the sample dried at 200 °C, it is important to note that there were two types of diffraction peaks on one kind of superimposed diffraction peak; sharp and broad peaks, which indicated two classes of NiO particles with small and large sizes. This showed that very high drying temperatures had a negative effect on the relatively homogeneous particle sizes. The effect of the drying temperature was caused by the boiling point of EG (197.3 °C). When the drying temperature approached or exceeded the boiling point of EG, the EG gradually evaporated and the amount of EG decreased substantially during the drying process. Earlier research revealed that the molar ratio of Ni/EG plays a vital role in controlling the particle sizes and dispersion of NiO on the MCM-41 support [14
]. Upon solvent evaporation at a high drying temperature, the residual amount of EG could not inhibit redistribution of the metal salt solution on the surface of the carriers, resulting in the NiO particle growth.
B displays the diffraction patterns of 20 wt % Ni/M 1EG samples calcinated at different temperatures after dried at 100 °C. The calcination temperatures had a minor impact on the particle size and distribution. The diffraction peaks of NiO could not emerge at 120 °C, due to the fact that no precursor decomposed to form the NiO crystal structure. After calcination over 150 °C, the peak intensities of NiO were very low and broad. Upon raising the calcination temperature from 550 °C to 800 °C, the Ni/M 1EG samples clearly expressed the broad and diffuse patterns of NiO with average particle sizes below 4.4 nm. This indicated that the Ni-based catalysts had favorable resistance to high temperature sintering by co-impregnation with EG. Generally, as the calcination temperature increased, the particles easily agglomerated and grew larger when using conventional wet impregnation. Thus, the Ni-based catalysts prepared by co-impregnation exhibited a wide temperature window of calcination and excellent resistance to metal sintering due to the strong metal-support interaction [29
C exhibits the XRD profiles of the Ni/M 1EG samples with different Ni loading varying from 5–40 wt % by co-impregnation. The diffraction intensity of the 5 wt % Ni/M 1EG sample was too low to be analyzed, implying that the smaller particle size of NiO (˂3.3 nm) was formed on the support due to its higher dispersion. When the metal loading exceeded 5 wt %, new dispersive diffraction patterns appeared. The diffraction peak intensity of the samples gradually increased as the metal content increased. According to the above results, NiO could be better dispersed on the MCM-41 surface using co-impregnation, and superfine NiO nanoparticles below 3.7 nm could be produced until the metal loading amount was 40 wt %. This clearly showed that the EG added during impregnation had a sufficient ability to control the relatively homogeneous particle sizes and the high dispersion for excess metal loading. This might be caused by the large surface area of the MCM-41 support, which could support excess metal even with 40 wt % Ni loading.
The pure siliceous MCM-41 lacks acidic sites, which restricts its broad application in catalysis. For the hydrorefining of bio-oil, bifunctional catalysts containing a hydrogenating metallic phase and a dehydrating acid phase are preferred to form hydrocarbons [21
]. Thus, it was particularly useful that the quantity and distribution of acidic sites on the catalyst supports containing MCM-41 was modified and improved by the strong acidity of other zeolites, such as HZSM-5. As shown in Figure 1
D, the Ni-based samples supported on MCM-41 and HZSM-5 were also investigated by XRD analysis. Typical diffraction peaks for the HZSM-5 crystalline phase centered at 2θ = 7.9°, 8.7°, 23.0°, and 23.9° were observed. The average crystal sizes of NiO supported on HZSM-5 decreased to 13.3 nm from 42.9 nm after co-impregnation with EG. In particular, the XRD patterns of Ni/M(H) 1EG also exhibited very low intensities and dissemination of the NiO diffraction peaks, suggesting that the mixed carrier containing MCM-41 and HZSM-5 could stabilize and disperse the special small metal particles as well as the single supporter of MCM-41. This could be caused by the high water-absorbing quality of its large pore volume and the high specific surface area of the MCM-41 support, resulting in the preferential adsorption of the impregnation solution by the MCM-41 in the mixed supports (Figure 2
D and Figure S1
). Consequently, the catalyst of Ni/M(H) 1EG not only retained the high hydrogenation activity of the Ni metallic phase, but also possessed the corresponding amount of dehydrating acidic sites.
The TEM images of the Ni-based samples are shown in Figure 2
A–D. Compared with conventional wetness impregnation, co-impregnation avoided NiO aggregation into even larger clusters on the MCM-41 support surface and facilitated the formation of remarkably smaller NiO particle sizes. A similar trend was observed on the HZSM-5 support. Particularly, NiO particles on the mixed supports containing MCM-41 and HZSM-5 were comparatively dispersed compared to the single MCM-41 support (Figure S1
). Table 1
summarizes the physicochemical and structural parameters of the various samples. After impregnation, the Brunauer-Emmett-Teller (BET) specific surface area clearly decreased, as well as the total pore volume. The drop in porosity and specific surface area might be ascribed to the formation of blockages on the support surface and channels. This indicated that the procedure of preparing the catalysts had no severe impact upon the molecular sieve structures and the order channels of HZSM-5 and MCM-41 (Figure S2
As illustrated in Figure 3
and Table S1
temperature-programmed reduction (H2
-TPR) experiments were performed to analyze the reducibility and reduction degrees of the Ni-based catalysts. The H2
consumption peak around 300–400 °C was associated with the reduction of bulk nickel oxide [5
]. On the other hand, the reduction peak at higher temperatures (>500 °C) could be related to the reduction of nickel oxide species, which strongly interacted with the support and appeared to be difficult to reduce [4
]. In comparison with the Ni/M 0EG catalyst, the Ni/M 1EG catalyst was quite different in reduction performance. The H2
consumption peaks were very broad, and shifted to higher temperature. According to the above results, we could consider that the existence of the nickel oxide species that were difficult to reduce was caused by the strong metal-support interaction using co-impregnation. In the case of Ni/H 1EG, the H2
-TPR profile exhibited one strong and sharp peak at around 310 °C, which showed that the reduction of NiO species relied heavily upon the nature of the support. Therefore, we can reasonably speculate that the main peak shift to 520 °C of the Ni/M(H) 1EG catalyst from 610 °C of the Ni/M 1EG catalyst was caused by the weaker interaction between the NiO and the HZSM-5 support. After reduction at 450 °C in H2
flowing for 4 h, the new diffraction peaks of Ni appeared, and the metal Ni kept the approximate crystal sizes and dispersion of NiO (Figure S3
XPS analysis was also performed to investigate the oxidation states of Ni at the outer layers of the various samples in Figure 4
. Generally, the binding energy of Ni-2p3/2
in the metallic Ni was about 852.4 eV [33
]. It was unambiguous that there was no metallic Ni present on the samples before reduction. The shoulder peak at 854–857 eV and the broad satellite centered at about 861.6 eV clearly indicated the presence of NiO [34
]. By comparison of the relevant binding energies with the observed Ni-2p3/2
satellite separation, it was determined that NiSiOx species were not formed over all the catalysts [35
]. The broad Ni-2p3/2
profiles could be deconvolved into two components, which indicated that there were two kinds of NiO species formed on the carrier surface. The peaks at around 854.5 and 856.4 eV were assigned to Ni2+
in bulk NiO and small NiO particles, respectively, corresponding to weak and strong interaction with the supports [36
]. The Ni-2p binding energy shift could be essentially explained by the particle size effect, in which the small NiO particles easily caused a charge transfer from NiO clusters to the support substrates, resulting in the strong metal-support interaction and the presence of NiO species that were difficult to reduce [37
]. Moreover, the surface area ratio of the two Ni-2p3/2
peaks elucidated the component proportion of weak and strong interactions, which was in agreement with the H2
-TPR results described above.
To characterize the overall concentration and strength of the acid sites on various samples, temperature-programmed desorption of ammonia (NH3
-TPD) analysis was performed. The NH3
-TPD spectrum of the tested samples displayed a broad distribution of acid sites with weak and strong strength, shown in Figure 5
, and the quantitative data of NH3
adsorbed for the various samples are shown in Table 2
. The low-temperature desorption peak at around 250 °C could correspond to NH3
absorbed on the weak acid sites derived from surface silanol groups. As expected, the high-temperature peak at about 467 °C suggested the presence of strong acid sites associating with Al-OH and Al-OH-Si groups in the HZSM-5 surface and structure [38
]. Nevertheless, for the Ni/M 1EG sample, there was only one very weak desorption peak that appeared at approximately 192 °C, suggesting purely siliceous MCM-41 was short of acid sites [40
]. After Ni impregnation of the HZSM-5 support, the acid amount was slightly less than that of HZSM-5 with the exception of a strong desorption peak shift to higher temperature; the temperature of the weak desorption peaks was almost unchanged, indicating that the loading of Ni mainly changed strong acid sites [41
]. Particularly, the distribution and strength of the acid sites had been changed significantly after the additional introduction of MCM-41 into the Ni/HZSM-5 1EG sample. The amount of strong acid obviously decreased, and a distinct shift of the desorption peak of the weak acid sites towards lower temperatures was observed. Normally, Brønsted acidity was ascribed to acid sites with a desorption temperature above 300 °C [42
]. Based on the NH3
-TPD studies, it could be considered that the quantity of Brønsted acid sites in HZSM-5 significantly declined after mixing with MCM-41, which was well consistent with the previous reports [25
]. The framework Al in HZSM-5 zeolite could be extracted by the silicon species in the framework of MCM-41, leading to the acidity adjustment of mixed zeolites. According to pyridine-adsorbed infrared spectroscopy in the literature, it is convincing evidence of the main existence of Brønsted acidic sites in addition to small amounts of Lewis acidic sites due to the reinforced interaction between HZSM-5 and MCM-41 when the two supports were mixed [25
]. Generally, the selective hydrogenolysis of C–O and C–C bonds was strongly dependent on the acidic properties of the catalyst supports; therefore, the weak-acid sites were conducive to the dehydration reaction in the C-O hydrogenolysis while the strong-acid sites often caused the rupture of C–C bonds. Thus, the concentration and properties of acidic sites on the mixed carrier could be regulated by mixing MCM-41 with HZSM-5, which would be beneficial for the improvement of C–O hydrogenolysis.
2.2. Guaiacol Hydrogenolysis Activity of Prepared Catalysts
To explore the hydrogenolysis activity of various catalysts prepared by co-impregnation, the Ni-based catalysts were comparatively investigated in the hydrogenation and HDO of guaiacol at different temperatures, respectively shown in Table 3
, Table 4
and Table 5
. The 20 wt. % Ni loading was determined to be selected as proper metal loading for the following tests (Table S2
). The results coincide well with the hydrogenolysis mechanism on noble metal catalysts that comprise the hydrogenation saturation of a benzene ring preferentially and the HDO of oxygenates after hydrogenation in the successive step [22
] (Scheme 1
). The main products of low temperature hydrogenation below 200 °C were methoxycyclohexanol, apart from a small amount of cyclohexanol, methoxycyclohexane, and cyclohexane. In respect to various Ni-based catalysts prepared using impregnation and co-impregnation, the hydrogenation activity below 200 °C was observed to have the following trend (Table 3
Hydrogenation activity: Ni/M (H) 1EG ≈ Ni/M 1EG > Ni/M 1EG + H > Ni/M 0EG > Ni/H 1EG.
In comparison with the Ni/M 0EG catalyst prepared by conventional wetness impregnation, the Ni/M 1EG prepared via co-impregnation presented outstanding catalytic activity in the guaiacol hydrogenation. For example, the guaiacol conversion of Ni/M 0EG increased from 15.5% to 92.6% when the reaction temperature increased from 150 to 200 °C. Correspondingly, on the Ni/M 1EG catalyst, the guaiacol conversion reached 97.4% at 150 °C. Even at 100 °C, there was 12.3% guaiacol conversion. Compared with the commercial 5% Pd/C and 5% Ru/C catalysts (Aladdin Reagents (Shanghai) Co., Ltd., Shanghai, China), the Ni/M 1EG catalyst prepared by co-impregnation exhibited similar activity with noble metal catalysts in the guaiacol hydrogenation. Thus, the catalytic activity of Ni-based catalysts supported on MCM-41 could be strongly enhanced by co-impregnation, which was essentially attributed to the high dispersion and ultra-small size of the NiO nanoparticles. Although the Ni/H 1EG catalyst had a mean NiO particle size of about 13.3 nm, it unexpectedly performed with poor hydrogenation activity, which was much lower than that of the Ni/M 1EG sample. For supported metal catalysts, it is well known that the hydrogenolysis activity of guaiacol is significantly dependent on the physicochemical properties of the support materials. The microporous HZSM-5 support presented not only the small pore volume but also the low special surface area. Moreover, the 20 wt % Ni loading dispersed on the HZSM-5 support led to the further decrease of the BET specific surface area and the blocking of the pore-channel structure. Conversely, the MCM-41 support had a high specific surface area, large pore volumes, and homogeneous hexagonal mesopore arrays, resulting in more adsorption and enrichment of the guaiacol reactant on the catalyst surface [25
]. Consequently, on the HZSM-5 support, this insufficient enrichment of guaiacol may be the cause of the low hydrogenation activity.
In addition, it is widely known that the surface acidity of solid catalysts plays a crucial role in catalytic HDO reactions. In order to improve the HDO performance of the Ni/M 1EG catalyst, the quantity of Brønsted acidity of the MCM-41 support was enhanced by additional introduction of HZSM-5. However, the guaiacol conversion decreased to 87.3% from 97.4% at 150 °C using the Ni/M 1EG catalyst together with the addition of a given mass of HZSM-5. This decreased activity could be explained by blocking of the hydrogenation active sites through coverage of the surface of the Ni/M 1EG catalyst with additional HZSM-5. Fortunately, the Ni/M(H) 1EG catalyst kept the comparative hydrogenation activity at low temperature. This was consistent with the XRD and TEM results, indicating that super small and highly dispersed NiO nanoparticles on the mixed supports containing MCM-41 and HZSM-5 also had outstanding hydrogenation activity.
The results presented above demonstrated that the distribution of products expressed a clear dependence on the reaction temperature. The HDO of oxygenated products originating from hydrogenation of the benzene ring mainly took place in the temperature range of 220–280 °C on these Ni-based catalysts. The formation of alkane products was usually considered to dissociate C–O bonds by demethoxylation and dehydroxylation [44
]. The detectable HDO products were mainly comprised of cyclohexane, methylcyclopentane, cyclopentane, methylolcyclopentane, methoxycyclohexane, and cyclohexanol. The following trend of HDO activity above 220 °C was revealed over various Ni-based catalysts (Table 4
and Table 5
HDO activity: Ni/M (H) 1EG ≈ Ni/M 1EG +H > Ni/M 1EG > Ni/M 0EG > Ni/H 1EG.
Moreover, the very small and highly dispersed NiO nanoparticles on the MCM-41 supports prepared by co-impregnation exhibited the higher HDO activity. However, the Ni/H 1EG catalyst presented both poor hydrogenation and HDO activity. For the Ni/H 1EG catalyst, the conversions of guaiacol at 220 °C, 250 °C, and 280 °C were 33.7%, 67.6%, and 98.7%, respectively. Correspondingly, other catalysts showed the full conversion at temperatures up to 200 °C. The yield of cyclohexane was increased to 73.0% on the Ni/M 1EG catalyst from 44.8% on the Ni/M 0EG catalyst at 250 °C. Obviously, the HDO activity of the Ni/M 1EG catalyst for guaiacol hydrogenolysis to cyclohexane was significantly enhanced by the addition of HZSM-5. The cyclohexane yield was rapidly raised to 52.0% from 2.2% by elevating the reaction temperature from 200 °C to 220 °C. Generally, the surface acids of heterogeneous catalysts could accelerate the breakage of C–O bonds by promoting dehydration in HDO reactions; therefore, many reaction systems adopted an acid as an efficient component in the catalysts. The cooperation between the hydrogenating metal sites and the dehydrating acidic sites could be responsible for obtaining the excellent HDO activities [22
]. Similar to the mechanical mixed catalyst of Ni/M 1EG+ H, the Ni/M(H) 1EG catalyst kept the comparative HDO activity at high temperature. Undoubtedly, this proved that the HDO reaction activity could be distinctly improved by strengthening the surface acid intensity of the catalyst supports. During the experiment, it was found that the selectivity for methoxycyclohexanol displayed an opposite trend with the increasing reaction temperature, which was caused by the C–O bond dissociations by both demethoxylation and dehydroxylation [29
]. Particularly, 84.1% yield of cyclohexane on the Ni/M(H) 1EG catalyst was achieved at 240 °C. Main byproducts were methylcyclopentane and cyclopentane, accompanied by a small amount of methyl-cyclohexane, n-hexane, n-pentane, methylpentane, butane, etc. It was noted that some guaiacol conversion schemes reported in the literature also included transalkylation reactions, isomerization reactions, ring-opening reactions, and C–C bond dissociations, which were responsible for the detected byproducts [30
In summary, well-dispersed nickel species on MCM-41 supports played a key role in both guaiacol conversion and hydrogenolysis performance, and the Ni/M 1EG catalyst synthesized using co-impregnation exhibited much higher hydrogenation and HDO activity in comparison with that of the Ni/M 0EG sample using conventional impregnation. The NiO supported on the mixed supports containing MCM-41 and HZSM-5 not only maintained the high dispersion, but also kept the comparative hydrogenation activity with noble metal catalysts at low temperature. Moreover, the HDO activity at high temperatures could be clearly improved by the introduction of HZSM-5 zeolite. Acid sites could promote demethoxylation and rearrangement of the intermediate methoxycyclohexanol. Obvious synergistic effects between the Brønsted acidic sites of supports and metallic Ni active phases were observed, which contributed to the prominent enhancement of the reaction rate in HDO of guaiacol (Table S3
, Figure S4
and Figure S5
). Meanwhile, larger BET surface area and homogeneous mesopores of MCM-41 facilitated the adsorption of guaiacol and hydrogen molecules on the catalysts, and favored the reactants’ collision, which improved the efficiency of the guaiacol hydrogenation. Therefore, one bifunctional catalyst of Ni/M(H) 1EG could be easily achieved by co-impregnation and mechanical mixing of MCM-41 and HZSM-5, and it could present higher HDO activity of guaiacol by the combination of metal-catalyzed hydrogenation and acid-catalyzed C–O bond dissociations. The observations above indicated that the hydrogenolysis of guaiacol occurred through the reaction pathway in which the hydrogenation saturation of the aromatic ring was performed to form methoxycyclohexanol as the first step, and was subsequently hydrodeoxygenated to generate cyclohexane. Consequently, the Ni/M(H) 1EG catalyst possessed similar activity with noble metal catalysts in the guaiacol hydrogenation, and it exhibited not only higher hydrogenation activity at low temperatures, with a 97.9% guaiacol conversion at 150 °C, but high HDO activity at high temperature, with an 84.1% yield of cyclohexane (C1-based) at 240 °C. In summary, this study could contribute to the development and improvement of supported metal catalysts applied in the chemical industry.