The four different zeolite supports applied in this study (H-ZSM-5, H-beta, H-mordenite, and H-Y) were characterized by nitrogen physisorption and ammonia temperature-programmed desorption (NH3-TPD). The results are compiled in Table 1. As can be seen from the results, zeolites H-Y and H-beta possessed the largest surface areas (730 and 680 m2 g−1, respectively), whereas the lowest total acidity (amount of absorbed ammonia) was presented by H-beta (1008 μmol g−1). This result correlates very well with the fact that H-beta has the lowest Al content of the examined supports. In all cases, the surface area decreased upon impregnation of the supports to obtain the 10 wt % V2O5/H-zeolite catalyst (see Table 1). At the same time, the amount of absorbed ammonia increased considerably as a result of the acidity of the vanadia. This increase was especially remarkable in the case of zeolite H-beta, which increased from 1008 to 1604 μmol g−1. Although comparison of zeolites with different pore frameworks, acidity, and surface area is not straightforward, the rise in acidity of H-beta may be a result of improved dispersion of the vanadia over the support. This zeolite H-beta also showed the best catalytic results, as will be discussed later; further characterization studies were performed with this support.
Table 1. Characteristics of the zeolite supports and 10 wt % V2O5 catalysts.
| ||Zeolite support||10 wt % V2O5/zeolite|
|Catalyst||Si/Al||Surface area[a] [m2 g−1]||Total acidity[b] [μmol g−1]||Surface area[a] [m2 g−1]||Total acidity[b] [μmol g−1]|
To study the effect of V2O5 loading, 1, 3, and 10 wt % V2O5/H-beta samples were synthesized and characterized. X-ray powder diffraction (XRPD) analysis of the prepared V2O5 catalysts with H-beta support (1, 3, and 10 wt % V2O5)25 revealed only diffraction peaks of the support, indicating an amorphous structure and high dispersion of the deposited vanadium species in the three catalysts. Similar results were also found for the catalysts obtained using the other zeolite supports (results not shown). The results from SEM coupled with energy dispersive spectrometry (SEM-EDS) and BET analyses of the prepared 1, 3, and 10 wt % V2O5/H-beta catalysts are presented in Figure 1 and Table 2.
Table 2. Characteristics of V2O5/H-beta catalysts.
|Material||Surface area[a] [m2 g−1]||Total acidity[b] [μmol g−]||Experimental V2O5 content[c] [wt %]||Surface density[d] [V atoms nm−2]|
|1 wt % V2O5/H-beta||680||1039||0.97||0.1|
|3 wt % V2O5/H-beta||502||1083||3.08||0.5|
|10 wt % V2O5/H-beta||401||1604||10.16||1.5|
As shown in Table 2, the surface area of the vanadia loaded catalysts remained high and decreased in the order 1>3>10 wt %, as expected. The vanadia contents determined by EDS analyses were in good accordance with the desired weight loadings in all of the prepared catalysts. Moreover, the total acidity of the final catalyst was also measured. The acidity increased with the vanadia loading, from 1008 to 1604 μmol g−1, in the case of pure support and the 10 wt % V2O5/H-beta catalyst, respectively. It has been reported that deposited vanadium surface species mainly present Brønsted acidity.26
Raman measurements were also performed on the H-beta catalysts with the aim of characterizing the molecular structures of the surface VOx species present in the 1, 3, and 10 wt % V2O5/H-beta catalysts. In the Raman spectra, the bands corresponding to VO vibrations appear in the range of 1100–900 cm−1.27–29 It has been reported for vanadia supported on silicas, aluminas, and zeolites that a band at approximately 1030 cm−1 corresponds to the VO symmetric stretching mode of vanadium at isolated sites of tetrahedral coordination. In the case of crystalline V2O5, a sharp band is located at approximately 990 cm−1.30–33
If visible laser light was used (532 and 785 nm), no bands were observed. The visible excited Raman spectra are more dependent on polymeric vanadyl species than monomeric species, so the band corresponding to isolated species is difficult to detect under these conditions.28 Fluorescence of the support and thermal instability of the zeolites, limiting the applicable power of the exciting laser beam, make detection of the sought bands difficult.
To overcome these drawbacks, UV Raman spectroscopy was employed.32 The excitation laser wavelength was selected to 244 nm to obtain a resonance Raman effect. Figure 2 shows the UV Raman spectra of the samples without dehydration treatment. The weak band at 1555 cm−1 corresponds to atmospheric oxygen (Rayleigh scattered Raman light) and indicates that the peak positions of the Raman shifts in the spectra were correct.33 No significant differences were found among the three samples. In all the V2O5/H-beta catalysts, one band corresponding to isolated tetrahedral V sites was identified (at ≈1026 cm−1). The width of this band indicated that the V species were well dispersed. The presence of a crystalline V2O5 phase (a band at 989 cm−1) was not detected in the prepared catalysts, in good agreement with the XRPD results.
Figure 2. UV Raman spectra of the V2O5/H-beta catalysts, the H-beta support, and pure V2O5 (λexcitation=244 nm, power at sample=3 mW).
Download figure to PowerPoint
The calculated surface density of vanadium on the catalysts (V atoms per nm2) is included in Table 2. In the case of the 1 and 3 wt % catalysts, the V loading was below the monolayer surface coverage for silica (0.7 atoms nm−2),39 which corroborates the assumption of absence of crystalline V2O5. The monolayer surface coverage for alumina is much higher (7.3 atoms nm−2), so it is likely that even the 10 wt % catalyst had a vanadium loading below the monolayer.
HMF oxidation at ambient pressure
The prospects of using vanadia supported on different microporous zeolites (H-beta, H-ZSM-5, H-Y, and H-mordenite) as catalysts were explored. Samples of 10 wt % V2O5 deposited on each zeolite were prepared as described in the Experimental Section and employed in the aerobic oxidation of HMF in N,N-dimethylformamide (DMF) as solvent. The substrate conversion and DFF yield as a function of the reaction time are presented in Figures 3 a and 3 b. The DFF yield as a function of the HMF conversion is presented in Figure 3 c.
Figure 3. a) HMF conversion and b) DFF yield as a function of reaction time, and c) DFF yield as a function of HMF conversion in the aerobic oxidation of HMF in DMF with 10 wt % zeolite-supported V2O5 catalysts. Reaction conditions: 0.1 g HMF, 5 mL DMF, 0.01 g of catalyst (1.4 mol % V), O2 flow (1 bar), T=100 °C.
Download figure to PowerPoint
The reaction progress data shows that the HMF conversion increased continuously within the examined time period (Figure 3 a), whereas the yield of DFF decreased slightly at prolonged reaction time (1440 min) for at least three of the tested catalysts (Figure 3 b). Thus, using V2O5/H-mordenite catalyst, the yield of DFF decreased from approximately 17 to 14 % after 330 and 1440 min of reaction, respectively. A similar tendency was observed if supported H-beta and H-ZSM-5 catalysts were employed. In the reaction with the V2O5/H-Y catalyst, no decrease was observed.
The highest selectivity to DFF at relatively high HMF conversion level was found if V2O5/H-beta, the catalyst with the highest acidity, was employed (Figure 3 c). Notably, all four tested zeolites were on H-form (i.e., possessed Brønsted acidity detrimental for the stability of HMF),12 although H-beta has a lower aluminum content compared to the other supports. It has been suggested by Carlini et al. that the Lewis type surface acidity, usually associated with the presence of coordination unsaturated Al sites,40 can promote side reactions leading to undesired by-products.17 This can explain why H-beta, the support containing the lower amount of Al in the structure (lowest Lewis acidity), and the highest acidity owing to V2O5 (Brønsted acidity), and, consequently, the highest metal dispersion, provided the highest DFF yield. The employment of a catalyst consisting of V2O5 supported on Na-beta zeolite (obtained by means of ion-exchange) did not, however, result in improved DFF selectivity. In contrast, both the HMF conversion and DFF yield decreased (31 and 11 % after 330 min, respectively, compared to 51 and 18 % for the H-form, as shown in Figure 3 a and b). This unambiguously indicates the influence of support acidity on the activity of the catalysts; Brønsted acidity enhances the overall catalytic activity, whereas Lewis acidity has a dominant role on side-reactions of HMF (i.e., on the DFF selectivity). Thus, the observed difference in catalytic behavior of the different zeolite-supported V2O5 catalysts could possibly be related to the observed, albeit not drastic, difference in the acidity (see Table 1).
In preliminary reactions, a TiO2-supported V2O5 catalyst was found to exhibit a high degree of leaching of catalytic species.41 Hence, the homogeneous contribution to the activity of the four 10 wt % zeolite-supported catalysts was investigated. In these experiments, the catalysts were kept under stirring in DMF for 330 min at 100 °C under a flow of oxygen. Then the catalysts were filtered off hot and the filtrate used as solvent for the aerobic oxidation of HMF. The conversions of HMF found for the experiments are presented in Figure 4.
Figure 4. HMF conversion (total and contributed by lixiviated catalytic species) in the aerobic oxidation of HMF in DMF with 10 wt % V2O5/zeolite catalysts. Reaction conditions: 0.1 g HMF, 5 mL DMF, 0.01 g catalyst (1.4 mol % V), O2 flow (1 bar), T=100 °C, t=330 min.
Download figure to PowerPoint
The obtained data shows that for V2O5/H-ZSM-5 and V2O5/H-mordenite catalysts over 60 % of the total catalyst activity is a result of the catalytic species dissolved from the solid catalyst during the pretreatment. The homogeneous contribution was especially high in the case of zeolite H-Y as support. Nonetheless, this value decreased to 45 % with the H-beta catalyst. Importantly, if the reaction was performed without catalyst and with pure H-beta zeolite, both the conversion of HMF and the yield of DFF remained under 1 % after 330 min. This indicates that under the applied reaction conditions the conversion of HMF observed in the homogeneous test (Figure 4) can be fully related to the presence of dissolved catalytic species.
The apparent higher HMF conversion arising from the homogeneous contribution with the V2O5/H-Y catalyst (Figure 3 a and 4) might suggest that the activity of the soluble species is superior to the solid sites and that the dissolution process of vanadia from the H-Y support requires some time. Therefore, in the leaching test, in which the catalytically active vanadium species have already been dissolved (Figure 4), the oxidation reaction proceeds faster than the reaction with the solid catalyst (Figure 3 a).
As the V2O5/H-beta catalyst was found to provide superior DFF yield in the HMF oxidation reaction (see Figure 3 b) as well as the lowest leaching of catalytically active species (resulting in the lowest homogeneous contribution to the catalytic activity of ≈45 %; Figure 4), this catalyst was chosen for further investigation. The possible correlation between the leaching of the active phase and the vanadia content of the employed catalyst was explored. In this context, the activity of the 1 and 3 wt % V2O5/H-beta catalysts was measured and compared to the activity obtained with the 10 wt % V2O5/H-beta catalyst. Importantly, the vanadia/substrate ratio was held constant in the experiments to allow direct comparison. The total HMF conversions achieved with the catalysts and their homogeneous contribution to the activity are presented in Figure 5.
Figure 5. HMF conversion in the aerobic oxidation of HMF in DMF with 1–10 wt % V2O5/H-beta catalysts as a function of reaction time: a) total and b) contributed by lixiviated species. Reaction conditions: 0.1 g HMF, 5 mL DMF, 1-10 wt % V2O5/H-beta catalyst (1.4 mol % V), O2 flow (1 bar), T=100 °C.
Download figure to PowerPoint
The observed HMF conversions after 330 min were found to be 51 and 54 % for the 10 and 3 wt % catalysts, respectively, whereas the 1 wt % catalyst exhibited a much lower activity corresponding to 21 % HMF conversion (Figure 5 a). This difference in catalytic performance is likely an effect of the vanadia loading, which may direct the formation of different active species. Moreover, these active sites can have different reactivity and accessibility in the porous channel network of the zeolite support.
Interestingly, the homogeneous contribution to the total catalyst activity (related to the HMF conversion) diminished drastically if the vanadium loading on the zeolite was lowered (Figure 5 b). Indeed, in case of the 1 wt % V2O5/H-beta catalyst no activity from dissolved species was found suggesting that leaching of the active phase was essentially avoided. This clearly confirmed that the catalytic activity of this catalyst was provided entirely by the active sites on the surface of the solid. Thus, indicating that at 1 wt % loading, the vanadium-containing species deposited on H-beta zeolite are less prone to lixiviation from the surface of the support. In contrast, the homogeneous activity contribution with the 10 wt % V2O5/H-beta catalyst constituted approximately half of the total activity. This is the first time that this kind of study of the leaching and homogeneous contribution of supported vanadium catalyst has been performed for this reaction.17, 32, 42
Following the established negligible homogeneous contribution of the 1 wt % V2O5/H-beta catalyst in DMF, the catalyst was further employed to study the effect of different solvents. Hence, the HMF aerobic oxidation was performed in toluene, α,α,α-trifluorotoluene (TFT), methyl isobutyl ketone (MIBK), acetonitrile, and dimethylsulfoxide (DMSO). The results are shown in Figure 6. For comparison, the results of the HMF oxidation in DMF are also shown in Figure 6 (solid line).
Figure 6. a) HMF conversion and b) DFF yield in the aerobic oxidation of HMF in organic solvents with 1 wt % V2O5/H-beta catalyst as a function of reaction time. Reaction conditions: 0.1 g HMF, 5 mL solvent, 0.1 g catalyst (1.4 mol % V), O2 flow (1 bar), T=100 °C.
Download figure to PowerPoint
Notably, HMF conversion of <2 % was observed if the reaction was performed in acetonitrile (at 80 °C) or DMSO (at 100 °C) under the flow of oxygen.
It can be seen from the data that although the conversion of HMF reached approximately 70 % if the reaction was performed in toluene or TFT after a reaction time of 330 min, the yield of DFF remained around 5–10 %, clearly indicating the dominance of side-reactions or reactions leading to the formation of humins, undetectable by HPLC or GC–MS. In the case of MIBK, HMF conversion constituted approximately 50 % after 330 min. Notably, at the same time, the yield of DFF was found to be almost as high (45 %). Thus, the reaction selectivity to DFF and DFF yields were found to be significantly higher after 330 min in MIBK solvent compared to DMF. Also, it can be seen that the DFF yield increased with the reaction progress, in contrast to the reaction in DMF (Figure 6 b).
It is generally assumed that solvent polarity affects the activity of the catalyst, although there is no general agreement regarding the influence that this phenomenon has on the conversion and selectivity.42 If vanadium-based catalysts were utilized, the conversion of HMF appeared to raise with increased solvent polarity.17 In our case, different solvents with increasing polarity were used: MIBK, acetonitrile, DMF and DMSO (polarity indexes of 4.2, 5.8, 6.4. and 7.2, respectively43). Very low values of conversion (<5 %) were observed in acetonitrile and DMSO under applied conditions, whereas MIBK provided higher conversion and selectivity values compared to the equivalent runs in DMF. In addition to this, the selectivity to DFF was also improved by using MIBK as a solvent (Figure 6). According to these results, the polarity of the solvent does not have a distinct effect on the catalyst activity, but it is noticeable that the medium plays an important role in the development of the reaction.
Further, the possibility of recycling the catalyst in two solvents, DMF and MIBK, was explored (Figure 7). It is seen from the results that the rate of DFF production per gram of catalyst remained constant if DMF was used as a solvent, even after being recycled four times. In the case of MIBK, an apparent deactivation of the catalyst took place after the second run. This could, among other reasons, be a result of deposition of carbonaceous residues in the pores of the catalyst, blocking the access to the active sites. Indeed, the color of the used catalyst turned to brown and an increment in the weight was observed. Furthermore, if the homogeneous contribution test was performed in MIBK as solvent, high DFF yield (60 %) was observed immediately after the addition of the substrate (HMF) to the reaction medium containing the leached species from the vanadium catalyst. Since the yield did not increase over time, the high conversion might be attributed to the presence of some very active oxidant species soluble in the medium, formed as a consequence of interactions between the solvent and the catalyst.
Figure 7. Rate of DFF formation per gram of catalyst in the recycling of 1 wt % V2O5/H-beta catalyst in DMF and MIBK. Reaction conditions: 0.1 g HMF, 5 mL solvent, 0.1 g catalyst (1.4 mol % V), O2 flow (1 bar), T=100 °C, t=330 min.
Download figure to PowerPoint
HMF oxidation at elevated pressures
Under the applied conditions (100 °C and atmospheric pressure) the reaction in DMSO resulted in a very low HMF conversion (see above). A test experiment at higher temperatures was conducted. The reaction in DMSO at 150 °C afforded a 30 % yield of DFF at 70 % HMF conversion after 24 h (1440 min). Hence, it proved possible to increase the DFF yield in DMSO with the 1 wt % H-beta-supported vanadia catalyst, albeit with a low DFF selectivity. This, together with the proven durability and recyclability of the 1 wt % V2O5/H-beta catalyst in DMF, motivated us to investigate the possibility of improving the DFF yield and selectivity by performing the reaction at elevated pressures. For this, the aerobic oxidation of HMF was performed in DMF under 2.5 bar (1 bar=100 kPa) of dioxygen pressure at 100 °C. DFF yield and HMF conversion are shown in Figure 8.
Figure 8. HMF conversion and DFF yield in the aerobic oxidation of HMF in DMF with 1 wt % V2O5/H-beta catalyst as a function of reaction time. Reaction conditions: 0.1 g HMF, 5 mL solvent, 0.1 g catalyst (1.4 mol % V), 2.5 bar of O2, T=100 °C.
Download figure to PowerPoint
Evidently, both HMF conversion and DFF yield increased with the oxidant pressure. Approximately 90 % HMF conversion and 25 % DFF yield were obtained after 330 min at 2.5 bar of O2, in contrast to 20 and 5 %, respectively, at ambient pressure (see Figure 5). Notably, the DFF selectivity did not change with increasing pressure, and remained around 25 % in both cases. However, the only product observed by GC–MS and HPLC analyses was DFF, thus suggesting the formation of humins, possibly through polymerization or overoxidation.
Importantly, if a reaction under identical conditions was performed in MIBK (2.5 bar O2, 100 °C), HPLC analysis of the products after 90 min of reaction revealed large amounts of formic acid (≈20 % yield at 90 % HMF conversion). Concurrently, substantial amounts of 5-hydroxymethyl-2-furancarboxylic acid (HMFCA) and 5-formyl-2-furancarboxylic acid (FFCA) were observed by both GC–MS and HPLC analyses, whereas DFF yield constituted only 5 %. This fact confirms that low selectivity is obtained for the reaction in MIBK under elevated pressures.
The results of the oxidation experiments performed with elevated oxygen pressure in DMF and DMSO with 1 wt % V2O5/H-beta catalyst are presented in Table 3. The obtained data show that the reaction in DMF at higher oxygen pressures (Table 3, entries 1 and 2) is clearly affected by the temperature. The DFF yield increased from 7 to 23 % at 60 and 80 °C, respectively, and the HMF conversion increased as well. However, the selectivity of the reaction towards DFF formation was reduced as the temperature increased (from 90 and 41 %, for 330 min at 60 and 80 °C, respectively). For comparison, the DFF selectivity after 330 min constituted only approximately 26 % under 2.5 bar of O2 at 100 °C, although the DFF yield was comparable (see Figure 8).
Table 3. HMF conversion and DFF yield and selectivity in the aerobic oxidation of HMF in organic solvents with 1 wt % V2O5/H-beta catalyst under elevated pressure.[a]
|Entry||P [bar]||T [°C]||t [min]||HMF conversion [%]||DFF yield [%]||DFF selectivity [%]|
Furthermore, the reaction in DMSO at 150 °C and 2.5 bar afforded high yields and selectivities of DFF at high HMF conversion values (Table 3, entries 3 and 4). DFF yield and selectivity reached 68 and 69 %, respectively, at full HMF conversion already after 330 min of the reaction time under 2.5 bar of O2 (entry 4). In contrast, the respective values were found to be only 9 and 19 % if the reaction was performed in DMSO at 150 °C under ambient pressure.
These results, together with the improved conversion and yield of the reaction in DMF at 10 bar pressure, prompted us to perform the aerobic oxidation of HMF in DMSO under 10 bar of O2 pressure. Here, the reaction was performed for 180 min at varying temperatures (Table 3, entries 5, 6, and 9). It is seen from the data that the conversion increased with the temperature, whereas the DFF yield reached a maximum at 125 °C. Indeed, the selectivities were found to be approximately 99, 98, and 74 % at 100, 125, and 150 °C, respectively. Notably, although providing high selectivity towards the desired product, the reaction at 100 °C afforded only a very low HMF conversion (Table 3, entry 6). However, at prolonged reaction times, an increase in HMF conversion and DFF yield were observed together with a decrease in the DFF selectivity: from >99 to 80 % after 180 and 1200 min, respectively (entries 6 and 8). A similar tendency was observed at 125 °C. Here, the amount of DFF and consequently the DFF selectivity decreased with the reaction propagation from 82 % after 180 min (84 % HMF conversion) to 71 % (>99 % HMF conversion) (Table 3, entries 9–11). The fact that the amount of DFF decreased in this case suggests that a side-reaction leads to gradual product degradation as the HMF conversion augments.
An increase of the oxygen pressure from 10 to 40 bar at 100 °C (Table 3, entries 7 and 12) allowed us to double the HMF conversion and DFF yield, from 21 and 20 %, respectively at 10 bar to 44 and 41 % at 40 bar, with approximately equal selectivity (≈95 %). At the same time, the reaction at 40 bar of O2 at 125 °C resulted in no improvement of the DFF yield compared to that achieved under 10 bar (entry 9, 82 and 81 %; at 10 and 40 bar respectively). Nevertheless, since the HMF conversion increased at higher pressures, the DFF selectivity diminished in this latter case (from 98 to 89 %; entries 9 and 13).
Further, a homogeneous test was conducted for the reaction with the highest achieved values of both substrate conversion and DFF selectivity (see Table 3, entry 9; 10 bar O2, 125 °C, 180 min). The results are presented in Figure 9.
Figure 9. Contribution to the HMF conversion and DFF yield of the 1 wt % V2O5/H-beta catalyst, the leached species, the H-beta support, and blank experiment in the aerobic oxidation of HMF in DMSO. Reaction conditions: 0.1 g HMF, 5 mL solvent, 0.1 g of 1 wt % V2O5/H-beta catalyst (1.4 mol % V) or H-beta, 10 bar of O2, T=125 °C, t=180 min.
Download figure to PowerPoint
The obtained data showed that an equally high HMF conversion (approximately 84 %) was reached in the oxidation reaction with the leached catalytic species, as well as if pure H-beta was introduced in the reaction. The conversion in the absence of the catalyst (a “blank” experiment) resulted in approximately 70 % conversion, indicating instability of HMF under the applied conditions. However, the yield of DFF was only 46 % if the lixiviated species were present in the medium, compared to the 82 % obtained with the supported catalyst. The use of the pure zeolite support afforded a 24 % DFF yield, and a low, but not negligible DFF yield (10 %), was observed in the absence of catalyst, thus making a direct evaluation of the homogeneous contribution difficult. However, it can be inferred that the solubilized species were not entirely responsible for the full catalyst activity. This fact implies that the active sites on the surface of the solid catalyst still had a major role in the catalytic reaction, although the extent of leaching cannot be ignored under these conditions.