2.1. Catalytic Activity of CuFe-Containing Catalysts in a Batch Reactor
Eight samples of the catalysts with different copper contents were studied in the process of furfural hydrogenation to furfuryl alcohol (FA). The ratio of Fe/Al oxide forms was kept constant and was about 4.6. The introduction of alumina can increase the specific surface area of the catalyst and can also lead to the formation of additional spinel structure. For convenience, catalysts were named as Fe82
, and Cu5
, where subscripts indicate the content of metal in the oxide form in the sample and textural characteristics (Table 1
The content of copper oxide in the CuFeAl samples ranged from 5 to 50 wt %; catalysts Fe82
were also prepared to determine the effect of iron and copper separately. The table also contains the surface by the BET method and CO uptake obtained by CO chemisorption for catalysts reduced at 250 °С. The reduction temperature was chosen according to literature data and our previous study [29
]. The iron-free catalyst had the largest surface due to the high content of γ-Al2
with a high surface. The decrease in Al content led to a decrease in the BET area. The minimum surface area was 22 m2
/ g and corresponds to the Cu50
sample. It should be noted that CO did not adsorb on the Fe82
catalyst. It could be concluded that adsorption takes place only on Cu particles. A stoichiometry of Cu:CO = 1:1 was assumed according to [30
]. Since furfural in its structure has an aldehyde group C=O adsorbing on catalysts’ surface during the reaction, it was assumed in the calculation of turn over frequency (TOF) that the amount of adsorbed CO was equal to the number of active sites of the catalyst. According to CO chemisorption the average amount of active sites was approximately the same among CuFe-catalysts. The low value of surface and active sites led to the need to use a high ratio of catalyst/feedstock in reaction.
CuFe catalysts were tested in a batch reactor at 100 °С and hydrogen pressure at 6.0 MPa using 60 mL 7 vol.% furfural in isopropanol to determine the optimal content of Cu. In Figure 2
the dependence of FA yield on reaction time in the presence of CuFe-containing catalysts is shown.
As can be seen from the dependence, a sample without copper does not show activity in the conversion of furfural practically (the conversion is no more than 7%), while the introduction of copper into the iron containing catalyst allows increasing the conversion of the desired product to 97%. Moreover, the most active catalysts are ones with a copper oxide content from 10% to 30%. At more than 30% of introduced copper, the activity of the catalyst was reduced, apparently due to the formation of larger particles of the active component, which were less active in the target process. The experiment without a catalyst showed no activity in furfural conversion at the same reaction condition. The repeat test for Cu20Fe66Al14 was performed to reproduce the observed data. Deviations in the received data were insignificant. Weisz-Prater criterion and the Mears criterion were calculated and showed that there were no internal diffusion limitations or external diffusion limitations.
shows the dependence of the yield of FA on the conversion of furfural. A sample that did not contain copper oxide was practically inactive, and the catalyst without iron exhibited insufficient activity in the target process. The introduction of both iron oxide and copper oxide into the catalyst made it possible to increase the activity of the monometallic copper catalyst due to the synergistic effect. Interaction of oxides of copper, iron and aluminum with the formation of spinel structure occurred. The spinel structure contributed to the formation of metallic copper particles on its surface and can be considered as additional sites for furfural adsorption. The composition of these structures will be discussed in more detail in the section of catalysts characterization. The most active sample from this series is Cu20
, the conversion of furfural and yield of FA were 98% and 97%, respectively. The selectivity for all samples was close to 100%. In addition, the selectivity did not depend on the degree of furfural conversion during the reaction time. Deviations from 100% were caused by the formation of a by-product—furfuryl hydroxyl-isopropyl ether (FHIE) formed during the interaction of furfural with isopropyl alcohol (solvent). The formation of this product was observed in our previous works [18
It was found that the experimental dependence of the concentration of furfural in the reaction mixture on the reaction time was well described by the first-order kinetic equation. Table 2
and Figure 4
present the turn over frequency (TOF) of the catalysts depending on the content of copper oxide in CuO-Fe2
. This parameter was calculated using the formula presented in the experimental part, which includes the number of active sites of the catalyst, which was determined by CO chemisorption and the constant rate of furfural conversion. The errors in Table 2
were found by minimizing the sum of squared differences of the estimated and experimental values of the mole fraction of the compound for each experimental point. According to the data, activity of the Cu-containing samples decreased in the series Cu20
The highest activity was observed for samples with a CuO content of 10 to 20 wt %. The catalyst with mixed oxide of iron and aluminum did not have activity under the process condition. A copper catalyst with alumina also did not show high activity in comparison with samples containing 10–30 wt % CuO. Apparently, the high activity of copper–iron–aluminum catalysts was associated with the formation of metal phases or mixed oxides, which determine the activity of these samples in the hydrogenation of furfural. When the content of copper oxide was more than 20 wt % a decrease in the activity of samples was observed, probably due to the formation of larger particles of copper oxide blocking the active sites of metal or mixed oxide phases.
Villaverde et al. [15
] have studied liquid-phase furfural hydrogenation over different Cu-based catalysts at 110 °C and 1.0 MPa, using iso-propanol as a solvent. The maximal TOF value among catalysts (Cu/SiO2
, CuMgAl, CuZnAl, and Cu-Cr) was 0.049 s−1
for the CuMgAl sample. In the present work, we had CuFe catalysts with twofold higher activity in furfural hydrogenation.
It is known that one of the reasons for the deactivation of catalysts is the formation of carbon deposits on its surface. An elemental CHNS analysis was performed in order to determine the amount of carbon in the composition of spent copper–iron catalysts. The results of this analysis are presented in Table 2
. It is shown that in general carbon formation was small and did not exceed 2%, while the highest carbon content corresponded to more active catalytic systems (with a mass content of copper oxide of 10%–30%). Apparently, due to the low activity of the other catalysts, carbon formation occurred not so fast (0.5% for samples Cu50
). However, along with the most active catalysts, approximately the same carbon content was observed in the case of the Fe82
catalyst. Apparently, the iron–aluminum catalyst was more prone to carbonization.
A series of experiments on the hydrogenation of furfural to FA over the Cu20
catalyst was performed in a batch reactor at the temperature in a range of 100–250 °C with a step of 30 °C to select the optimum temperature of the process (Figure 5
It was shown that the temperature range of 100–130 °C was optimal for this process and led to 97%–99% FA selectivity and 100% furfural conversion (Figure 6
). At further temperature increasing to 160 °C, a slight decrease in the formation of FA (yield of 95%) occurred, which was much more significant at a temperature of 190 °C (yield was less than 53%). With an increase in the reaction temperature (190–250 °С), the hydroxyl group of FA was hydrogenated forming 2-methylfuran (2-MF) in an amount up to 37%. At a temperature of 220 °C, a significant shift towards the formation of 2-MF occurred, and it became the main reaction product with a yield of up to 74%, while the yield of FA was only 6%. At this temperature, the further hydrogenation of the 2-MF furan ring with the formation of 2-methyltetrahydrofuran (2-MTHF) was observed followed by the breaking of the cycle with the oxygen removal with 1-pentanol formation (Figure 6
). At a reaction temperature of 250 °C, the formation of 2-MF was reduced to 52% due to deeper hydrogenation and cracking reaction and the formation of 1-pentanol with a yield of up to 22%.
Elemental analysis of the catalysts after the reaction showed that the carbon content was approximately constant and did not exceed 1.2 wt %; we could affirm that the Cu20Fe66Al14 catalyst was insensitive to the temperature under the reaction conditions in a batch reactor relative to carbon deposits.
The catalyst Cu20Fe66Al14 was chosen for further studies in selective hydrogenation of furfural to FA in the solvent-free process based on the observed data, taking into account the conversion of furfural, the yield and selectivity for FA, as well as the specific surface area. In addition, this catalyst could be considered as a catalyst for obtaining 2-MF at a temperature of 200–250 °C and hydrogen pressure 6.0 MPa. It showed the high yield of 2-MF (74%) and stability to the deposition of carbon residues on its surface.
Studying the mechanism of furfural adsorption on active sites of Cu-based catalysts Resasko et al. [17
] suggested that furfural adsorption occurs mainly through an unshared pair of oxygen electrons of the aldehyde group with formation of η1-aldehyde. In this state, the molecule of furfural is perpendicular to the catalyst surface and repulsive forces act on the aromatic ring due to overlapping of the 3D-orbital of Cu surface atoms with the aromatic furan ring. Thus, the reaction can proceed through the formation of surface alkoxide or hydroxyalkyl species.
The latter mechanism was more energetically preferable, which was explained by the stabilizing effect of the aromatic furan ring on the hydroxyalkyl intermediate (the attachment of the H atom to the O atom led to the formation of an unpaired electron on the C atom that could be delocalized by the furan ring). It should be noted that while there was a strong repulsion of the furan ring on the surface of Cu (111) the furan ring was closer to the surface due to the lower density of Cu atoms on this face on Cu (110).
2.2. Catalytic Activity of Cu20Fe66Al14 Catalyst in a Fixed-Bed Reactor in Solvent-Free Furfural Hydrogenation
On the first step, we initially optimized the temperature regime. To study the influence of the temperature on the yield and selectivity of the target product in a fixed-bed reactor, a temperature 100–200 °C was used. It should be noted that the process of hydrogenation of furfural was carried out without any solvent.
It was shown that at a reaction temperature of 160 °C, almost a complete conversion of the feedstock (99%) occurred with the formation of FA up to 96% (Figure 7
). As a secondary product, 2-MF up to 5% was identified. In the temperature range of 100–140 °С, the conversion of furfural was very low and did not exceed 30%; when the temperature was raised to 180 °С, the process shifted towards the formation of 2-MF since a relatively high temperature contributed to a more intense hydrogenation of hydroxyl group as was shown previously in a batch reactor. In addition, 2-MTHF (up to 8%) as a result of further hydrogenation of 2-MF and the self-condensation product of furfural (at a reaction temperature of 200 °C especially) with a yield of up to 30% were formed. Polymers are an undesirable product since ones reduce the activity of the catalyst by covering catalyst surface and blocking the access of the substrate to the active sites of the catalyst.
The CHNS analysis of the catalysts after the reaction showed that the carbon content increased with increasing temperature and reached 5.4 wt % at 200 °C, apparently, the formation of polymers formed from furfural molecules at this temperature covered the surface of the catalyst increasing the amount of carbon deposits. At the same time, at 160 °C the yield of FA reached 96% while the carbon content was less than 2 wt %.
Based on the data obtained, it was proposed to use a temperature of 160 °C for the maximum yield of FA. The data presented in Figure 8
demonstrate that the chosen Cu20
catalyst was actually more active than catalysts with a lower or higher copper content, studies were carried out in the presence of three catalytic systems (Cu5
, and Cu50
) in a solvent-free reaction at 160 °C, hydrogen pressure 5.0 MPa, and hydrogen flow 300 mL/min, LHSV = 1 h−1
. This was in agreement with results in a batch reactor and confirmed the right choice of the most active catalyst.
Thus with LHSV = 1 h−1
catalyst leads to an FA yield of up to 96% at 100% furfural conversion for 30 h. Moreover, the value of carbon deposits on its surface did not exceed 1.7 wt %. Nevertheless, the high stability of the catalyst could be caused by low LHSV. It was shown that with a twofold increase in LHSV in the first 10 h of the process, the conversion reached 100% (Figure 9
). The complete hydrogenation at the initial stage was provided due to the high amount of reduction sites on the catalyst surface. Then the process went to the stationary mode and conversion and, accordingly, the FA yield were reduced to 90% and remained at that level until the end of the lifetime test (75 h). It should be noted that other hydrogenated products or polymers were not found. Analysis of the spent catalyst by the method of the CHNS analysis showed that the carbon content did not exceed 3.1 wt %.
Thus, a copper–iron catalyst with a mass content of copper oxide of 20% was active and highly selective in the reaction of hydrogenation of furfural to FA, not only in a batch reactor but also in a fixed-bed reactor in the absence of a solvent. To study the composition of the active component, the catalysts were investigated by different methods such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), transmission electron microscopy (TEM), etc.
2.3. Catalysts Characterization
The XPS of the initial CuFe-containing catalysts made it possible to determine the composition of these catalysts in the surface layer (Table 3
). Thus, the atomic ratio [Cu]/[Al] for all samples was approximately constant, and the content of the forms of iron with respect to aluminum decreased; apparently, with an increase in the amount of CuO introduced into the catalyst, some iron and copper compounds were formed, which contributed to a decrease in near-surface iron.
spectra were described by one asymmetric peak with a binding energy of Cu2p3/2
in the region of 932.7–933.7 eV and the corresponding shake-up satellites in the region of 941–944 eV (Figure 10
). The shape of the spectrum and the presence and shake-up of satellites corresponded to copper oxide CuO. In the spectra of catalysts with a copper oxide content from 10 to 50 wt %, an additional peak was also observed in the region of 935.4 eV, which corresponds to copper aluminates. In the literature [33
binding energies are given for metallic copper in the range 932.5–932.6 eV, for Cu2
O the binding energy is in the same range as for the metallic state of copper; therefore, the Auger parameter must be calculated to determine the state of copper. The binding energy of Cu2p3/2
of copper in the Cu2+
state in the structure of CuO oxide was in a wider range of 933.8–934.1 eV. It should be noted that with an increase in the content of CuO there was a decrease in its amount on the catalyst surface (Table 3
) due to the formation of CuAl(Fe)Ox
The Fe2p spectra of the studied catalysts were represented by the Fe2p3/2
doublet, the integral intensities of the components of which were 2:1. To determine the state of Fe, both the position of the Fe2p3/2
mainline and the shape of the Fe2p spectrum were used. The position and intensity of the “shake-up” line of satellites depend on the chemical state of iron. In the case of the studied catalysts, the spectra of Fe2p3/2
were peaks with binding energy in the region of 711.6–712.2 eV, and shake-up satellites were observed. According to published data, iron in the composition of FeO, Fe3
, and Fe2
oxides is characterized by Fe2p3/2
binding energies in the ranges 709.5–710.2, 710.1–710.6, and 710.7–711.2 eV, respectively [34
], while the “shake-up” satellites were 5.7, 8.5, and 8.8 eV away from the main peak Fe2p3/2
. The great importance of the binding energy and the presence of “shake-up” satellites suggest that iron was in the Fe3+
state in these catalysts.
In the spectra of Al2p catalysts, a narrow symmetric peak was observed in the region of 74.5 eV, related to Al3+ in the composition of the Al2O3 support. However, the differences between Al3+ in the oxide and spinel structure according to XPS data were insignificant. Therefore, aluminum could also be included in the spinel Cu(FeAl)Ox structure.
To determine the phase composition of the studied spent catalysts, we used the XRD method, which showed that the sample Fe82
contained the hematite Fe2
was presented by aluminum metahydroxide AlOOH (Figure 11
). In both cases, the samples contained the phases based on the component with high loading.
The phase composition of copper-iron catalysts with a copper oxide content of 5 and 10 wt % was represented by magnetite Fe3
). No other phases were found in these samples either due to the presence of X-ray amorphous phases, which could not be determined by XRD. With a further increase in the content of copper oxide, the phase composition of the catalysts changes and represents reduced copper, as well as Fe3
-type spinels (for example, copper ferrite CuFe2
of mixed spinel Cu(FeAl)Ox
). It is worth noting that the smallest crystallite size of reduced copper was inherent in the sample with a copper oxide content of 20 wt %, with a further increase in the content of copper oxide, this parameter increased significantly (Table 4
). Apparently, this fact determined the highest activity of this catalytic sample. Thus, we could conclude that the active component of this catalyst was highly dispersed copper, which was formed when the content of copper oxide was 20 wt % in the presence of the Fe3
-type spinel structure.
It is worth noting that, according to the XRD data, a catalyst Cu20Fe66Al14 had the same composition both after experiments in an autoclave and in a fixed-bed reactor. Cu20Fe66Al14 consists of the metallic copper (about 8%) and 92% of spinels with the structure of Fe3O4. It could be concluded that the catalyst had high stability under the reaction conditions and did not change the composition of the active component even during prolonged testing in a fixed-bed reactor in solvent-free furfural hydrogenation.
Fresh sample reduced at 250 °C and the spent Cu20
catalyst were additionally analyzed by the differentiating dissolution (DD) method using mineral acids (nitric and hydrofluoric acids) with different concentrations. The method was adjusted in real-time in the stoichiographic titration mode according to information on the concentration of elements during DD and kinetic curves of element dissolution. The chemical dissolution curves of the elements were transformed into stoichiograms (Figure 12
). The averaged concentrations given below were calculated using several tens of design points. It was shown that the composition of the samples was represented by phases of variable composition, in which compounds with different degrees of the order were formed.
In the samples, the formation of a phase Cu0.04Fe1Al0.37 (F2) based on iron oxide with a small copper content, which dissolves under “harsh” conditions (6M HF) was observed. Presumably, on the surface of phase Cu0.04Fe1Al0.37 particles with a cationic composition Cu0.8Fe1 (F1) were found. The results of DD show that copper in the composition of the catalysts was in several forms: About 7–9% in the form of CuO or Cu and the copper in the composition of the binary and ternary compounds Cu-Fe and Cu-Fe-Al, respectively.
To study the active component formation the in situ XRD method was used. It allows tracking the changing in the phase composition in the most active catalyst in the hydrogenation of furfural to FA Cu20
during reduction at different temperature (Figure 13
Data obtained using this method showed that at room temperature the Cu20Fe66Al14 sample was a mixture of Fe2O3 (85%) and CuO (15%) oxides. In the temperature range of 30–175 °C, no changes in the diffraction pattern were observed. At a temperature of 200 °C, intense reflections of Cu0 appeared (12%). With a further increase in temperature to 250 °C, the CuO and Fe2O3 reflections disappeared and peaks corresponding to the structure of Fe3O4-type spinel appeared. When the reduction temperature reached 450 °C, metallic iron reflections began to appear (up to 16%), while a decrease in the intensity of Fe3O4 reflexes was observed (the content decreased from 83% to 68%). At a temperature of 600 °C, the phase composition was mainly represented by Cu0 and Fe0. Since no aluminum-containing phases were detected and a decrease in the lattice parameter of Fe3O4 was noted, it could be assumed that Al and Cu were embedded in the lattice of iron oxide with the formation of the mixed spinel Cu(FeAl)Ox.
The method of transmission electron microscopy allows confirming the above data regarding the phase composition with the formation of crystallites of a small size. Thus, an analysis of the reduced at 250 °C Cu20
catalyst by the TEM method showed the presence of two types of particles. The first one was CuFe2
agglomerates with 10–30 nm located on the Al2
matrix. The second one was the oxide of Cu of 2–5 nm in size (Figure 14
). It could be argued that CuO was obtained after oxidation of metallic Cu covering the surface of CuFe2
Additional studies of the Cu20
catalyst by the x-ray absorption near edge structure (XANES) method show that fresh catalyst is represented by CuO, Fe2
, and CuFe2
phases. In Figure 15
the results of approximation of the XANES spectra in the form of a linear superposition of the XANES spectra of the Cu K-edge of catalyst and standard compounds (Cu0
O, CuO, and CuFe2
) were presented. According to this analysis, copper was in the form of two phases: CuO and CuFe2
with a content of 70% and 30%, respectively. A significant part of copper was reduced to the metallic state (65%) and Cu1+
(25%), while 10% of copper remained in the CuO state when reduction at 250 °C. The spectra of CuFe2
in the reduced catalyst were not observed. It might be caused by the fact that CuFe2
could be the platform for the formation of the high active Cu0
particles. The observed data indicate that the particles Cu1+
and CuO were in the dispersed state due to these particles that were not detected by XRD while the presence of CuO was also confirmed by XPS.
Thus, the combination of physicochemical methods allowed us to make an assumption about the nature of the active sites of the catalyst and conclude that the active component of this catalyst was highly dispersed metallic copper (2–5 nm), which was formed when the content of copper oxide was higher than 20 wt % in the presence of Fe3O4-type spinel structure. Spinel could be formed during incorporation of CuO and Al2O3 in the structure of Fe3O4, leading to the formation of mixed spinel Cu(FeAl)Ox.