3.1. Trans-Esterification of D-fructose
Lipases (triacylglycerol ester hydrolase, EC 184.108.40.206) are very commonly used in synthetic organic chemistry [24
]. Beside the ubiquitous ability to catalyze the hydrolysis of fats and oils, lipases can catalyze enantioselective derivatization of chiral compounds, esterification, trans-esterification, and inter-esterification reactions [29
]. Although lipases from different sources are able to catalyze the same reaction, their catalytic activity might be very different even under the same reaction conditions [31
]. Table 1
presents the results when three types of commercially available lipases were used for the trans-esterification of D-fructose in various solvents, with vinyl acetate as the acyl donor. Since trans-esterification is a reversible reaction, the use of vinyl acetate as an acyl donor would make it become irreversible, as the by-product, vinyl alcohol, is labile and immediately converts to acetaldehyde [29
]. It should be noted that lipase was chosen as the catalyst because the enzyme’s characteristics favor the trans-esterification of primary alcohol groups, which would lead to the formation of the desired di-ester derivation of D-fructose. The three particular types of immobilized lipase used for this trans-esterification were Novozym 435 (1000 PLU/g), Lipozyme TL IM (250 IUN/g), and Lipozyme RM IM (300 IUN/g). It was noticed that all three types of lipase could catalyze the trans-esterification when an ether-type solvent (THF or dioxane) was used as media for the reaction. When Novozym 435 together with THF or dioxane was used, the formation of DAF was most favorable (Entry 5,6—Table 1
; 96.2% and 90.4% of DAF yield in THF and dioxane, respectively). Meanwhile, water and solvents containing –OH functional groups unsurprisingly prohibited the reaction (Entry 1,2,3—Table 1
) since the –OH group of these solvents interferes with the exchange of the acetyl group from the acyl donor and primary –OH groups of D-fructose. On the other hand, solvents with a medium and low polarity index gave little or no yield of DAF (Entry 4,8,9,10,11—Table 1
) because they can hardly dissolve D-fructose, thus the reaction remains stagnated when D-fructose could not be dissolved and came into contact with the immobilized lipase. It was also observed that using an excess amount of acyl donor (5 equiv. of vinyl acetate) did not improve the yield of DAF (Entry 12,13—Table 1
). This screening result is consistent with the literature [24
], and therefore, Novozym 435 was chosen as the source of catalyst while both THF and dioxane were used as solvents for the scaling-up procedure.
Scaling up of the trans-esterification reactions (×10 to ×600 time) was performed based on the screening results. After the scaling-up reactions, the catalyst was filtered off and recovered, the solvent was removed with rotary vaporization and vacuum, and DAF was then obtained in a yellow syrup form and stocked up as material for the dehydration step. Results in Table 2
show that scaling up the procedure had very little effect on the conversion of D-fructose as well as the formation of DAF. With some minor adjustments of the reaction time and stirring speed, we managed to achieve an 88.0% yield of DAF with 92% of D-fructose conversion (Entry 11—Table 2
) in a 600 time scale-up reaction. This result indicates that trans-esterification of D-fructose with immobilized enzyme catalyst is a quite stable procedure and can be utilized for sustainable production of DAF. One major setback of this procedure is that when the concentration of the starting material was increased (from 1% w/v
to 5% w/v
), the yield of DAF started to decrease accordingly (a 22.9% yield drop in THF and 21.9% yield drop in dioxane was recorded). In other words, it was hard to achieve a high yield of DAF without using a large amount of solvent to fully dissolve D-fructose at the beginning state of the reaction. This was due to the low solubility of D-fructose in organic solvents and was hard to compromise. However, almost all of the solvent could be recovered and reused after the reaction.
Experiments to investigate the reusability of the immobilized lipase were also performed in THF. After each cycle, the enzyme in the small bead form was simply filtered off, washed with THF, and dried at room temperature for 12 h before reuse. It was observed that Novozym 435 can be reused for up to five cycles with less than 9% of weight loss while still giving a stable yield of DAF (less than 5% decline in DAF yield—Figure 1
). Further recycling of the enzyme led to a higher drop in DAF yield (10.4% and 13.7% drop in DAF yield at the sixth and seventh recycling circle, respectively) as well as over a 10% loss of the catalyst amount. The loss of catalyst could be explained by the partial fragmentation of catalyst beads during the long reaction and these fragments were lost in the recovery and transferring process between the cycles. As observed in Figure 1
, the increment in catalyst loss correlated well with the drop in DAF yield.
3.2. Dehydration of DAF
DAF could be converted to AMF by removing three molecules of water and one molecule of acetic acid under acidic conditions as depicted in Scheme S1
). The DAF obtained from the prior trans-esterification step was dissolved in organic solvent media prior to the dehydration reaction. Based on previous studies of catalytic dehydration of sugars into furan compounds [9
], Amberlyst 15 was chosen as the heterogeneous acidic catalyst for the dehydration step. Multiple experiments were conducted in order to optimize the reaction parameters, such as the reaction time, catalyst ratio, and reaction media.
In order to study the course of the reaction, a time-dependent experiment was performed. DAF diluted in dioxane was heated at 120 °C and samples were taken at certain time intervals for HPLC analysis. It was observed that both AMF and HMF were formed under this condition, with AMF being the major product. After 8 h of reaction, we achieved the highest yield of AMF with 23.6% and 8.2% of HMF (Figure 2
). Further prolongation of the reaction showed no increment of the AMF quantity. We suspected that the relatively low yield of AMF could be improved when reaction conditions, such as the solvent media or catalyst amount, were to be adjusted. In addition, the simultaneous formation of AMF and HMF in the reaction media suggested that both dehydration and hydrolysis were catalyzed in the presence of Amberlyst 15 (Scheme S2
Since DAF is synthesized and easily dissolved in THF or dioxane, these solvents were tested to see if they are suitable for the dehydration of DAF. DMSO was also examined because AMF, unlike HMF, can be easily separated from DMSO by liquid-liquid extraction [20
]. The results in Table 3
show that THF was inferior to dioxane and DMSO as a solvent for dehydration. It was understandable since THF possesses a low boiling point (66 °C) and therefore would not be very effective for the dehydration of DAF, which was performed at a much higher temperature (120 °C). Compared to the time-dependent experiment, we increased the amount of Amberlyst 15 used in this experiment to 50 mg, but the yield of AMF in dioxane was not significantly improved (23.6% compared to 29.2%). At the same time, the yield of HMF in dioxane, when increasing the amount of Amberlyst 15, was quite noticeable (from 8.2% to 15.6%). This could be an indication that the hydrolysis reaction in dioxane is a little more favorable when more acid catalyst is added.
The AMF yield was the highest in the case of DMSO, which has been widely and effectively used as a solvent for the dehydration of ketose-type sugars into furan compounds [33
]. Several researches also mentioned that DMSO could stabilize HMF, and reduce side reactions during dehydration [36
]. Thus, it was understandable that DMSO gave a higher total yield of HMF and AMF. On the other hand, the yield of AMF in dioxane was not so far behind when compared to in DMSO (29.2% to 38.7%), especially considering dioxane has a much lower boiling point than DMSO (101 °C compared to 189 °C). Therefore, we purposely combined DMSO and dioxane as a mixture solvent for further investigation, with the aim of maximizing the yield of AMF while limiting the formation of HMF. As the results show in Figure 3
, using DMSO as a co-solvent greatly enhanced the formation of AMF while still keeping HMF at low yields. The best condition for solvent was at 15% DMSO in dioxane, which resulted in a 55.8% and 9.0% yield of AMF and HMF, respectively. It was not expected that when we increased the ratio of DMSO in dioxane to over 20%, the yield of AMF started to decline. As we mentioned that AMF and HMF were simultaneously formed during the reaction, we highly suspected that when DMSO was mixed in as co-solvent at a specific ratio, the combined solvent media had positive effects on the dehydration of DAF into furan compounds. Firstly, it was clear that the added DMSO possessed the benefit of increasing the boiling point of the reaction media, which made the combined solvent more suitable for dehydration at high temperature (120 °C). Secondly, the presence of DMSO was suggested to have a catalytic activity in the removal of H2
O molecules from the furanose ring [33
] as well as reducing the formation of by-products and stabilizing the formed AMF and HMF in the reaction mixture [39
Since Amberlyst 15 is a cation exchange resin in the H+
form with a macro-reticular pore structure [40
], dehydration of DAF in this case is catalyzed heterogeneously at hydrogen ion sites located throughout the resin beads. It is expected that higher amounts of catalyst during long reaction times will lead to a better yield of AMF. With that in mind, we investigated the effect of the catalyst ratio on the dehydration reaction. Upon varying the mass ratio between catalyst/substrate while maintaining the composition of reaction media at 15% DMSO in dioxane (Figure 4
), it was observed that when the catalyst/substrate ratio ranged from 0.6 to 1.0 (w/w
), the formation of AMF was significantly improved and stabilized (over 50% yield of AMF) compared to using a lower catalyst/substrate mass ratio. This means that when optimizing the procedure for sustainable production, we can slightly adjust the amount of catalyst in order to lower the cost of the reaction.
Various scaling-up dehydrations of DAF were also performed based on the best reaction media (15% DMSO in dioxane) with the ratio of Amberlyst 15/DAF (RCat/Sub
) at 0.6. The results in Table 4
confirm that this procedure could be effectively scaled up to 100 times without any effect on the yield of the desired AMF product (Entry 5). It was noticed that since the amount of DAF and Amberlyst 15 was increased while the solvent volume was kept at a constant value, the stirring speed had to be slightly adjusted (ranging from 300 to 500 rpm) for better heat regulation during the entire reaction course.
The recycling of catalyst would have an impact on the cost of the reaction. Therefore, we conducted an experiment to test the reusability of Amberlyst 15 in this particular reaction since it could affect the possibility for commercialization of this procedure. After each cycle of dehydration, Amberlyst 15 was filtered, washed with acetone, and dried in a vacuum oven (25 °C) overnight before being reused. The results showed that Amberlyst 15 can be reused for up to five cycles without a remarkable change in the yield of AMF and HMF (Figure 5
). It should be noted that all the cycles in the reusability test were performed using 15% DMSO in dioxane as the reaction media.
Additional experiments using different acidic catalysts were also performed as an attempt to improve the yield of AMF from DAF dehydration. Two homogeneous acid catalysts and one heterogeneous acid catalyst were used, and the results are shown in Table 5
. Concentrated sulfuric acid (H2
) and p
-toluenesulfonic acid (pTSA) displayed relatively high catalytic activity for the dehydration in DMSO solvent (over 73% of AMF yield, Entry 4, 5, 6, and 12—Table 5
). On the other hand, alumina (Al2
—acidic form) possessed quite low activity even when used at a higher molar ratio and longer reaction time (DAF conversion was lower than 50% and AMF yield was lower than 40%, Entry 13–20 in Table 5
). Overall, DMSO proved to be a better solvent for DAF dehydration since most of the experiments performed in DMSO gave a higher conversion of DAF as well as a higher yield of AMF. When applying a mixed solvent system (15% DMSO in dioxane), only the case of Al2
displayed a noticeable improvement in AMF yield (entry 23—Table 5
). It could be concluded that the use of a co-solvent would have little effect on dehydration with homogeneous acid catalyst. We also noticed that when the amount of homogeneous acid catalysts (from 25 mol% to 50 mol% of DAF) was increased, the yield of AMF was not much improved, or even decreased (in the case of Entry 5 and 6 in Table 5
) so we conducted blank experiments to examine the stability of AMF in the presence of these acids. The results showed that when the catalyst amount was doubled, the formation of HMF from AMF was tripled (Entry 21–24 in Table 5
) and a small amount of levulinic acid was detected in the reaction mixture.