3.3.2 Optimization of the esterification process using synthetic broth
Before performing the esterification of organic acids in the real fermentation broth, a synthetic broth containing similar compositions of the desalinated fermentation broth (Table3 ) was prepared and used for optimization of the esterification conditions. To achieve better separation and recovery of catalyst from the reaction mixture, a heterogeneous catalyst Amberlyst-15 resin, which is reported to have an excellent catalytic activity in many esterification reactions (Dange et al., 2014 ), was used for the formation of methyl esters instead of using homogeneous catalysts such as sulfuric acid or para-toluene-sulfonic acid. As shown in Fig.5A and5B , the conversion of both acetic and butyric acid increased markedly as the reaction temperature was increased from 25 to 70oC. An equilibrium maximum conversion of 99.54 ± 0.14% in 90 min reaction time for acetic acid (XeA), and 98.27 ± 0.03% in 150 min for butyric acid (XeB), were reached at the boiling temperature of 70oC, respectively. In addition, it can be also seen that the XeB decreased to 62.02 ± 0.80% when the temperature was reduced to 25oC, while a relatively high XeA of 90.88 ± 4.4% was maintained at the same reaction temperature, indicating that the activation energy needed for the formation of MB is higher than that for MA. The reaction temperature was fixed at 70oC with a reaction time of 2.5 h for the rest of the esterification experiments in order to ensure a high conversion of both acetic and butyric acid and thus a high recovery yield of the organic acids from the fermentation broth.
Catalyst loading was also observed to have positive influence on the conversion of acetic and butyric acids (Fig5C ). Regardless of whether the catalyst loading was 10 or 20%, the conversion of both acids were over 98%. A reduction to 5% loading led to a similar acetic acid conversion, whereas the butyric acid conversion slightly decreased to 96%. By contrast, in control samples without catalyst less than 10% acid conversion was reached, indicating an excellent catalytic activity of Amberlyst-15 for simultaneous synthesis of MA and MB. We also examined whether the high catalyst loading of 20% can accelerate the reaction rate for reaching the same maximum acid conversion yield, but no significant reduction of the time was observed (data not shown). These findings illustrate that a 10% loading of Amberlyst-15 is enough for the efficient conversion of acetic and butyric acids in the synthetic broth.
In contrast to catalyst loading, the initial water content presented in the synthetic broth negatively affected the acid conversion (Fig.5D ). The XA and XB under the optimal reaction conditions decreased to 94.08 ± 0.69% and 87.21 ± 1.11% with 10% water, which was 5.5% and 11.3% lower than those achieved in reactions without water, respectively. Thus the initial water content in the fermentation broth should be carefully controlled before starting the esterification reaction. In our designed DSP, water from the fermentation medium can be completely removed by vacuum distillation. However, some water will be inevitably introduced to the fermentation broth during the acidification step using HCl solution. To remove the trace amount of water and improve the equilibrium conversion in the real fermentation broth, addition of molecular sieves or silica gel that has good water absorption can be considered in the DSP.
In principle, alcohols other than methanol, such as ethanol and 1-propanol can also be esterified with the free organic acids in the fermentation broth, forming corresponding ethyl ester or propyl esters of acetic and butyric acid, which are also high value-added products widely used in manufacturing perfumeries and flavoring food (Dange et al., 2015 ; Xue et al., 2020 ;Nemati et al., 2021 ). In fact, we found that ethanol and 1-propanol were also effective for salt crystallization from the fermentation broth. Therefore, the reactivity of ethanol and 1-propanol with acetic and butyric acids in the synthetic broth was also investigated. As shown in Fig.6A , both XA and XB slightly decreased to around 96% when the alcohol was changed to same molar of ethanol or 1-propanol. The total consumption of acids by the esterification reaction maintained at a similar level regardless of the carbon chain length of the used alcohol. However, there are large differences in the esterification selectivity of the organic acids to the different alcohols including PDO and glycerol under the same reaction conditions (Fig.6B ). The content of PDO and glycerol decreased markedly after the esterification reaction as the number of carbons in the linear alkyl chain of alcohol increased, indicating more acids were esterified with the polyols when ethanol or 1-propanol was used. In particular, the reduction of PDO in all three different alcohol based synthetic broths was almost 2 times higher than the corresponding reduction of glycerol. This may be attributed to the increased steric hindrance and thus reduced reactivity between the acids and the ortho-hydroxylated glycerol compared to the meta-hydroxylated PDO (Cui et al., 2017 ). Although esterification between the acids and the polyols is inevitable, both acetic and butyric acids exhibited the highest esterification selectivity to methanol in the synthetic mixture. The reacted polyols (PDO + glycerol) in the methanol based reaction was 1.6 and 5.4 times lower than those in the ethanol and 1-propanol based reactions, respectively (Fig 6B ). More than 95% of the reacted acetic and butyric acids were successfully converted to the corresponding methyl esters (data not shown). As discussed above, less PDO derived ester impurities facilitate the separation of acids and the rectification of highly pure PDO. Therefore, considering the ester recovery and the PDO purification efficiency, methanol is a good choice for the esterification conversion of organic acids in the fermentation broth.