Investigation of Biophotocatalytic Conversion of Used Cooking Oil into Olefins

Abstract

The possibility of biophotocatalytic conversion of used cooking oil was investigated. The targeted transformation of used cooking oil into the desired olefin products was demonstrated using a modification of a previously published protocol, which resulted in the yield of the desired olefin products of 36.1% (±6.1%). Subsequently, the possibility to replace the originally used solvent dichloromethane with two more environmentally benign solvents 2-methyltetrahydrofuran and tert-butyl methyl ether was investigated, and tert-butyl methyl ether was identified as a potential replacement solvent. The yield of the desired olefin products for the reactions conducted using tert-butyl methyl ether was 41.2% (±2.6%). Furthermore, the potential of recycling the catalytic system was investigated both in dichloromethane and in tert-butyl methyl ether. Recycling was possible only when tert-butyl methyl ether was used, as loss of enzyme activity after the initial run of the reaction was observed in dichloromethane. Potential for recycling of the entire catalytic system was demonstrated as well. Scaled up experiments exhibited reduced reaction yield, which was ascribed to the need for extension of reaction time. The feasibility of the reaction was also demonstrated with waste pork fat. Finally, GC-MS analysis indicates that the product makeup corresponds to the fatty acid composition of the starting materials.

1. Introduction

Alkenes are a vital component of the petrochemical industry, serving as fundamental building blocks for an array of chemical products. The utility of alkenes stems from their facile functionalization, which gives rise to a number of valued products [1,2,3,4]. Alkenes of various lengths are traditionally made by the oligomerization of ethylene derived from coal or petroleum [5,6,7,8,9]. The non-renewable fossil fuel origin (e.g., coal or crude oil) of the majority of bulk scale olefins presents an environmental and resource issue [10,11]. This is further compounded by the energy requirements of the processes involved (e.g., cracking).

Fats, oils, and fatty acids appear as interesting potential precursors for sourcing commodity alkenes from renewable sources, given the fact that they contain largely unfunctionalized hydrocarbon chains of various lengths [10,11,12]. Furthermore, significant amounts of fat-based waste materials (e.g., waste cooking oil) are being produced by human activities [13]. These waste materials could present an even more sustainable source to produce olefins than pristine oils and fats produced directly for this purpose. This is further augmented by the fact that improper disposal and storage of waste fats and oils can present danger for health and the environment [13].

While the direct conversion of fats and oils to olefins has received limited attention, the interesting possibility of obtaining alkenes from fatty acids via the dehydrodecarboxylation reaction was investigated using both chemical and biological approaches. Over the last 60 years, several approaches for converting fatty acids into olefins have been developed. These mainly utilize homogeneous transition metal complexes (Pd, Rh, Ir, Fe, Ru, and Ni). Palladium-catalyzed reactions have received most of the interest so far. Tsuji and co-workers first reported such reaction in 1968 when they presented the decarbonylation of aliphatic acyl chlorides to olefins by a Pd catalyst [14]. The reaction proceeds at 200 °C and provides a mixture of positional double bond isomers.

While other precious metals (e.g., Rh, Ru) have been used for this reaction as well, it is important to note that the reaction can also be catalyzed by more common first row transition metals. Several examples of alkene formation from carboxylic acids catalyzed by Ni(COD)₂ (COD = 1,5-cyclooctadiene) have been reported [15,16]. Similarly, the possibility to carry out this reaction with Fe-based catalysts has been demonstrated as well [17]. Unfortunately, high temperatures were required in both cases.

Biocatalytic approaches to the conversion of carboxylic acids to alkenes have been developed as well [11,12,18]. The first report of decarboxylation of long chain aliphatic fatty acids to alpha-olefins was reported by Rude and coworkers in 2011 [19]. An interesting enzymatic system capable of the conversion of carboxylic acids to alkenes is the P450 enzyme OleT_JE, which catalyzes the decarboxylation of various fatty acids.

Photocatalysis, which has seen a renaissance in recent years due to the development of photoredox processes, presents another possible approach to the conversion of carboxylic acids to alkenes [20]. This approach has been developed only recently, with one of the first examples being published by Tlahuext-Aca and coworkers in 2018 [21].

The examples described in the previous sections focus on the conversion of carboxylic (fatty) acids into olefins. The possibility of directly converting triglycerides into olefins is of interest as well. Interestingly, Nguyen and coworkers have reported a method combining these two steps into a single process in 2019 [5]. Their approach utilizes lipase-based hydrolysis of triglycerides into fatty acids and glycerol and a photocatalytic dehydrodecarboxylation step utilizing acridine- and cobalt-based catalytic system. Various reports on the use of chemo-enzymatic cascades exist in the literature [22,23,24,25]. Furthermore, Cui and coworkers published a report describing a biophotocatalytic cascade used for acylation of N-heterocycles in 2022 [26]. However, the report by Nguyen and coworkers was the first describing this cascade to produce alkenes from bio-based oils.

Fats and oils constitute significant portion of waste generated by human activity [27]. Thus, these materials are available on a large and meaningful scale. However, the quality of these materials, for example the presence of significant amounts of free fatty acids alongside triglycerides, complicates their valorization [27,28]. One of the most dominant fat waste materials is used cooking oil, which is produced on an annual scale reaching 190 million metric tons [13,29]. Several approaches to the valorization of used cooking oil are already being investigated and implemented. The main among these is its conversion to biodiesel, with conversions to urethanes, lubricants, or bioplastics being other less developed options [13]. Direct conversion of used cooking oil to olefins has received minimal attention so far. Therefore, in this work, we focused on further developing the above-mentioned method of converting triglycerides to olefins by applying it to used cooking oil, a common fat-based waste [5]. In addition, we explored the possibility of replacing dichloromethane used in the original procedure with a more environmentally friendly solvent [30]. Finally, we also explored the possibility of reusing a part of the catalytic system to improve the effectivity of the reaction.

2. Results

2.1. Initial Experiments

This investigation was carried out in order to extend the utility of a previously reported method for the direct biophotocatalytic conversion of oils into olefins [5]. This report has demonstrated the feasibility of this reaction with oils (e.g., sunflower oil) and oil containing biomass (e.g., sunflower seeds). Therefore, the aim of this work was to attempt to investigate the possibility of employing used cooking oil as a typical fat-based waste material in this reaction. Furthermore, we also aimed at replacing dichloromethane originally used in this reaction with a more environmentally friendly solvent (e.g., 2-methyltetrahydrofuran, tert-butyl methyl ether).

Scheme 1 illustrates that the biophotocatalytic conversion of used cooking oil, a fat-based waste material composed mainly of triglycerides, which was investigated in this work, proceeds in two distinct steps. The first step is an enzymatic hydrolysis of the triglycerides, as well as other partial glycerides, yielding fatty acids. This is then followed by the second step where the fatty acid intermediates are converted to the desired olefin products. This second reaction uses a combination of acridine and cobaloxime catalysts. A simplified scheme of the mechanism of the photocatalytic step can be found in the Supplementary Materials (Scheme S1).

We selected ¹H-NMR spectroscopy as the key analytical technique for investigating the performance of the reactions. A list of the key peaks monitored in this work is provided in Table S1. Therefore, we started the work with the acquisition of ¹H-NMR spectra of a sample of used cooking oil, which is shown in Figure 1. The ¹H-NMR spectrum of used cooking oil in Figure 1 demonstrates that the major components of this material are triglycerides. Firstly, the peak observed at 5.26 ppm (c) corresponds to the presence of hydrogen atom on the central carbon of the glycerol backbone (sn-2). Secondly, the peaks at 4.30 ppm (e) and 4.15 ppm (f) correspond to the protons on the terminal carbons of the glycerol backbone (sn-1 and sn-3). Finally, the spectra contain signals of the alkyl chains of the fatty acid with peaks around 2.3 ppm (h, α-to the carbonyl function), 2.0 ppm (i, β-to the carbonyl function), 1.60 ppm (k, γ-to the carbonyl function), 1.10–1.30 ppm (l, intermediate methylene groups on fatty acid residues), and 0.9 ppm (m, terminal methyl group of the fatty acid chains). The spectra also contain a broad peak between 5.30 and 5.40 ppm (b, corresponding to the internal double bonds found in unsaturated fatty acids).

Having obtained the reference spectrum of the used cooking oil starting material, we proceeded with performing initial iterations of the reaction. The ¹H-NMR spectrum of the crude reaction mixture carried out with used cooking oil is shown in Figure S1. This spectrum indicates that the reaction has taken place. The evidence for this is the disappearance of the glyceride signals at 4.30 ppm (e) and 4.15 ppm (f) and the appearance of the signals for terminal alkenes at 4.95 ppm (d) and 5.85 ppm (a).

Subsequently, the crude product mixture was purified using column chromatography on silica where the alkene products were eluted using pure hexane. Figure 2 shows the ¹H-NMR spectrum of the purified product obtained from used cooking oil. The spectrum indicates that the purification step has succeeded in removing the impurities present in the crude products, such as the acridine catalyst.

The original report upon which this work was based used a reaction time of 60 h. Therefore, reaction time of 60 h was used in the reaction described above, which provided data presented in Figure S1 and Figure 2. Using this reaction time in our experiments resulted in low isolated amount of the desired product. Therefore, we tested the effect of the reaction time on the amount of the product that could be isolated and established 144 h as a suitable reaction time, which was used in further experiments. Subsequently, the reaction was repeated in triplicate with reaction time of 144 h and the average isolated olefin product weight was 25.3 mg (±4.3 mg). This corresponds to an average yield of the product of 36.1% (±6.1%).

2.2. Selection of a More Suitable Organic Solvent

As mentioned above, one of the issues with the original reaction protocol, which was also used in the experiments described above, is the use of dichloromethane as a solvent, as dichloromethane is hazardous both to humans as well as to the environment. Therefore, one of the goals of this work was an attempt to carry out the reaction using alternative solvents. Two alternative solvents, 2-methyltetrahydrofuran and tert-butyl methyl ether, were selected based on the literature precedent [30].

Test reactions using these solvents instead of dichloromethane were carried out. The ¹H-NMR spectra of the crude reaction mixtures for reactions carried out using 2-methyltetrahydrofuran and tert-butyl methyl ether can be seen in Figures S2 and S3, respectively. The ¹H-NMR spectrum for the reaction carried out in 2-methyltetrahydrofuran (Figure S2) demonstrates that the substitution of dichloromethane resulted in a reaction that does not produce the desired alkene product. This is evident from the fact that the spectrum does not contain the characteristic peaks of terminal olefins at 4.95 ppm (d) and 5.85 ppm (a). The spectrum in Figure S2 also indicates that the first step of the reaction, enzymatic hydrolysis of the oil into fatty acids and glycerol, proceeded well, while it is the second step of converting the fatty acids into olefins that did not take place. On the other hand, the ¹H-NMR spectrum in Figure S3 contains the signals at 4.95 ppm (d) and 5.85 ppm (a), which indicates that the desired olefin product was formed in the reaction carried out using tert-butyl methyl ether as the substitute solvent.

The crude product from the reaction carried out using tert-butyl methyl ether as a substitute solvent was purified using column chromatography and the olefin products were isolated after elution with hexane. The ¹H-NMR spectrum of the purified product is shown in Figure 3, and it is very similar to the spectrum of the purified product isolated after the reaction carried out with dichloromethane as the main organic solvent (Figure 2), which provides further evidence for the possibility of using tert-butyl methyl ether as a substitute solvent.

The conversion of used cooking oil in a reaction where tert-butyl methyl ether was the main organic solvent was carried out in triplicate and the average isolated product weight from these experiments was 28.8 mg (±1.8 mg). This resulted in an average yield of 41.2% (±2.6%). This indicated that the change in the solvent from dichloromethane to tert-butyl methyl ether did not affect the reaction extensively and might even be beneficial in terms of product yield.

2.3. Recycling Experiments

While performing the above experiments, we observed that the reaction mixtures were biphasic when both dichloromethane and tert-butyl methyl ether were used as the main organic solvent. This led us to the decision to attempt to recycle the enzyme catalyst contained in the aqueous portion of the reaction mixture after isolating the product from the organic layer. Thus, a repeat of the reaction was carried out using the aqueous buffer solution from the initial run, while fresh acridine and cobalt catalysts were used. The repeat reactions were carried out using both dichloromethane and tert-butyl methyl ether.

The results of this experiment performed in duplicate are shown in Figure 4. The data in Figure 4 shows that the amount of olefin products isolated from reactions carried out with recycled enzyme in dichloromethane and tert-butyl methyl ether were 5.5 mg (±4.9 mg) and 25.4 mg (±1.8 mg), respectively. Thus, there was a drastic reduction in the amount of olefin product isolated after the reaction with recycled enzyme catalyst when dichloromethane was used as the main organic solvent, while only a small performance reduction is observed in the case of tert-butyl methyl ether being used.

To explain this difference in the performance of the reactions with reused enzyme catalyst, we performed ¹H-NMR analysis of the materials isolated from the column purification of the products using a highly polar mobile phase made of ethyl acetate (70% v/v) and methanol (30% v/v). The ¹H-NMR spectra for the reactions performed using dichloromethane and tert-butyl methyl ether are shown in Figures S4 and S5, respectively. The spectrum for the reaction carried out using dichloromethane (Figure S4) exhibits the characteristic glycerol backbone peaks in triglycerides at 4.30 ppm (e) and 4.15 ppm (f). These peaks are absent in the ¹H-NMR spectrum for the material obtained from the reaction performed using tert-butyl methyl ether as the main organic solvent (Figure S5). This indicates that in the reactions where dichloromethane is used as the solvent, the lipase enzyme loses activity when used after recycling, as the starting triglycerides are observed at 4.30 ppm (e) and 4.15 ppm (f, Figure S4) in the materials isolated from the reaction mixture. This explains the low yield of the olefin product obtained in this reaction. On the other hand, the absence of triglycerides in the products of a rection carried out in tert-butyl methyl ether with recycled lipase catalyst (Figure S5) indicates that the lipase remains active after recycling, which leads to the activity being maintained to a large degree in these reactions.

Having shown the possibility of recycling the enzyme catalyst once, we proceeded to an experiment to investigate whether the recycling of the enzyme in further reaction repeats would be possible. The results of this experiment are shown in Figure 5 (blue bars), which shows the percentage of olefin products isolated in each reaction round, relative to the initial reaction. In this experiment, the weight of olefin products isolated after the first round of enzyme recycling is 78% relative to the weight of the products isolated in the initial run of the experiment. The relative product weight then drops to 38% in the second recycling cycle.

To understand this outcome, we carried out ¹H-NMR analysis of the material obtained from the column using the polar eluent described above after the second enzyme recycling step (Figure S6). The spectrum contains the characteristic signals of triglycerides at 4.15 ppm (f) and 4.30 ppm (e), which indicates incomplete hydrolysis of the starting material. This is likely due to loss of enzyme activity during the second recycling step as no triglycerides were observed in this fraction after the first recycling step (Figure S5).

The ¹H-NMR analyses of the materials obtained from the column chromatography purifications of the reaction mixtures in the previous experiments using highly polar solvent systems (e.g., Figure S6) indicated the potential presence of the other components of the catalytic system, especially acridine, in this fraction. Therefore, we decided to test the possibility to carry out recycling experiments where the fraction of the materials isolated from column chromatography using polar eluent would be returned to the reaction and no new catalyst would be added. The results of this experiment are shown in Figure 5 (orange bars). In this instance, the weight of the olefin products isolated after the first recycling cycle is 89% relative to the initial reaction and this then drops to 43% in the second recycling cycle. The materials isolated from column chromatography using the polar eluent after the second recycling cycle were analyzed by ¹H-NMR spectroscopy (Figure S7). In this case, the spectrum did not contain the typical triglyceride peaks observed at 4.15 ppm (f) and 4.30 ppm (e). However, the presence of peaks at 4.11 ppm could indicate the presence of products of incomplete hydrolysis (e.g., diglycerides, monoglycerides). This would again suggest that loss of activity by the enzyme is part of the reason for the drop in the isolated weights of the olefin products. However, the loss of activity in the photocatalytic part of the reaction sequence, likely due to loss of the catalysts during the repeated column purification steps, is also likely a contributing factor.

2.4. Scale up Experiments

Subsequently we attempted to conduct the reaction with its size scaled up three times. The desired olefin products were successfully isolated from the scaled-up reaction as demonstrated by the ¹H-NMR spectrum shown in Figure S8. The product of the scaled-up reaction was further characterized by ¹³C-NMR spectroscopy, which showed peaks consistent with the presence of the olefin products (Figure S9). The spectrum contains peaks for alkyl carbons between 10 and 35 ppm, peaks for internal olefins originating from unsaturated fatty acids present in the used cooking oil between 125 and 135 ppm, and peaks of terminal olefins formed in the reaction at 114 and 139 ppm. Furthermore, the composition of the product olefin mixture was also investigated using GC-MS (Figures S10–S13). Figure S10 is the TIC chromatogram, which shows the presence of two major peaks with retention times of 12.585 min and 12.659 min and a minor peak at 10.590. The EIC chromatogram for m/z 234.23 (Figure S11) indicates that the peak with retention time of 12.585 min represents (8Z, 11Z)-1,8,11-heptadecatriene, which is expected to form in the investigated reaction from linoleic acid. On the other hand, the EIC chromatogram for m/z 236.25 (Figure S12) shows that the peak at 12.659 min represents (8Z)-1,8-heptadecadiene, which corresponds to the expected product from oleic acid. Finally, the EIC for m/z 210.23 (Figure S13) shows that the minor peak at 10.590 min corresponds to 1-pentadecene, which is the expected product from palmitic acid. The results of the GC-MS analysis are consistent with the fact that soybean and sunflower oil were the key constituents of the used cooking oil starting material in this work, while the material also contained a minor portion of palm oil [31].

The scaled-up reaction was carried in three repeats, and the average isolated product weight was 49.1 mg (±3.9 mg), which corresponds to an average yield of 23.4% (±1.8%). This is markedly lower than the 41.2% yield observed in the reactions conducted on the original scale. The residue of the reaction mixture isolated from column chromatography using the polar eluent described above was analyzed by ¹H-NMR (Figure S14). The spectrum shows that the enzymatic hydrolysis of the triglycerides has run to its completion. Therefore, the lower yield, in comparison to the smaller scale reaction is ascribed to the progress of the second photocatalytic step of the reaction.

To gain further insight into this issue, the composition of the reaction mixture of the scaled-up reaction was investigated at 48 h intervals by ¹H-NMR spectroscopy. The ¹H-NMR spectra obtained from these analyses can be seen in Figures S15–S19. The ¹H-NMR spectrum obtained after the first 48 h (Figure S15) of the reaction indicates that the enzymatic hydrolysis of the triglycerides in the starting material was complete within this timeframe. On the other hand, the spectrum in Figure S15 contains only weak signals at 4.95 ppm (d) and 5.85 ppm (a) corresponding to the product, which indicates that the second step of the catalytic conversion has progressed only marginally at this time point. The ¹H-NMR spectrum in Figure S15 also contains a peak around 2.35 ppm, which corresponds to the α-position relative to a carbonyl group (h). This further supports the fact that the enzymatic hydrolysis of the triglycerides into fatty acids has been completed at this time point. The integrals of the peaks of the terminal olefins (a, d) increased relative to the integral of the internal olefins at 5.35 ppm (b) at the latter time points (Figures S16–S19). Concomitantly, the relative intensity of the peak at 2.35 ppm (h) decreased during the course of this reaction, indicating the continued conversion of the enzymatically produced fatty acids in the photocatalytic part of the reaction. However, this peak is still present in the spectrum acquired at 240 h (Figure S19), which signifies that the photocatalytic reaction did not run to its completion. The experiment has been carried out in duplicate, and we used the integrals to calculate the percentage of reaction progress, which was plotted against reaction time (Figure 6). The data in Figure 6 shows that the rate of formation of the olefin product is steady up to the 144 h datapoint. After this, the reaction appears to slow down. These observations indicate that the reaction continues to progress beyond 144 h, which was the stopping point in the above-mentioned experiments. Thus, the relatively low product yield observed in these experiments could be attributed to the reaction time being too short for the scaled-up reaction.

Subsequently, the scaled-up reaction was carried out in triplicate with a reaction time of 240 h. The average isolated product weight was 64.3 mg (±8.0 mg), which corresponds to an average yield of 30.6% (±3.8%). This represents an improvement in comparison to the 23.4% yield observed in the original scaled up experiment, which could be assigned to the extended reaction time. However, this yield is still lower than that observed in the smaller scale reaction. This might, in part, be attributed to the lesser suitability of the reaction vessel for the scaled-up reaction.

To further demonstrate the utility of this process, the modified scaled up reaction was carried out with waste pork fat. This average amount of olefin product isolated from this reaction carried out in triplicate was 50.5 mg (±6.2 mg), which corresponds to an average yield of 24.0% (±2.9%). This result indicates that waste pork oil can also act as a starting material in this reaction. However, the reaction will need to be optimized further to achieve better outcomes. The ¹H-NMR spectrum (Figure S20) for the product obtained from the conversion of waste pork oil is consistent with the formation of the expected olefin product. Furthermore, the product was analyzed by GC-MS (Figures S21–S25). Figure S21 is the TIC chromatogram, which shows the presence of two major peaks with retention times of 12.661 min and 10.596 min and two minor peaks at 12.589 min and 12.895 min. The EIC chromatogram for m/z 236.25 (Figure S22) shows that the peak at 12.661 min represents (8Z)-1,8-heptadecadiene, which corresponds to the expected product from oleic acid. The EIC for m/z 210.23 (Figure S23) shows that the peak at 10.596 min corresponds to 1-pentadecene, which is the expected product from palmitic acid. The EIC chromatogram for m/z 238.27 (Figure S24) indicates that the minor peak with retention time of 12.895 min corresponds to 1-heptadecene, which is the expected product from stearic acid. The minor peak with retention time of 12.589 min was assigned to (8Z, 11Z)-1,8,11-heptadecatriene, which is expected to form in the reaction from linoleic acid. This assignment was made by comparison with the results discussed above for the GC-MS analysis of the products from the conversion of used cooking oil. This was because the EIC chromatogram for m/z 234.23 from the product of conversion of waste pork fat did not exhibit strong enough signal (Figure S25). These results are consistent with the results obtained for the used cooking oil discussed above and with the fatty acid composition of pork-based fats such as lard [32].

3. Discussion

We have undertaken this study to test the possibility of expanding the utility of a previously reported procedure for the conversion of fats and oils into olefins [5]. The work has pursued two main ideas. Firstly, we aimed to investigate the possibility of carrying out this transformation with waste material rather than clean oil. The reason for this was the fact that fat-based waste materials are being produced on a significant scale, which makes their conversion into new useful chemicals desirable from the point of sustainability. We specifically chose used cooking oil as it is one of the most abundant fat-based wastes [13]. Secondly, we tried to determine whether it would be possible to replace dichloromethane, which was used as the main organic solvent in the original report, with an eco-friendlier solvent [30].

As mentioned above (Scheme 1), the conversion of oils and fats in the explored reaction proceeds in two distinct steps. The first step is the enzymatic hydrolysis of triglycerides into glycerol and fatty acids, while the second step is the photocatalytic dehydrodecarboxylation of the fatty acids into the olefin products catalyzed by acridine and cobaloxime. The first step is a typical enzymatic hydrolysis step, which is mechanistically unremarkable. On the other hand, the photocatalytic dehydrodecarboxylation is less common. Therefore, we will briefly discuss its mechanism. A simplified diagram of the mechanism, which is based on the findings of Nguyen and coworkers, can be found in Scheme S1 [5]. The mechanism contains distinct acridine and cobaloxime catalytic cycles. The reaction starts in the acridine catalytic cycle with the formation of a hydrogen-bonded complex of acridine (I) and the carboxylic (fatty) acid (II). This complex (III) then undergoes light-promoted decomposition in which the acridine abstracts the shared hydrogen atom and forms a radical species (IV). The carboxylic acid is initially turned into carboxylic acid radical, which spontaneously decomposes into carbon dioxide (V) and an alkyl radical (VI). The hydrogen atom is subsequently abstracted from the alkyl radical by a Co^II species (VII) within the cobaloxime catalytic cycle. This results in the formation of the alkene product and (VIII) a Co^III hydride intermediate (IX). The acridine catalytic cycle can be closed via different pathways, which can either involve a carbocation (X) or a dihydro acridine species (XI). These steps in the acridine regeneration pathways are interconnected with the steps closing the cobaloxime (XII) catalytic cycle and result in the formation of molecular hydrogen (XIII).

Initial attempts at performing the conversion of waste cooking oil, which were carried out with a slight modification of the published procedure while still using dichloromethane as solvent, demonstrated the possibility to convert waste cooking oil into the desired olefin products.

Subsequently, we tested the possibility of replacing dichloromethane with an eco-friendlier solvent. Two candidate solvents, tert-butyl methyl ether and 2-methyl tetrahydrofuran, were chosen based on suggestions from the literature. These experiments have shown that the reaction could be carried out using tert-butyl methyl ether but not 2-methyl tetrahydrofuran as no product formation was observed when the latter solvent was used. Gratifyingly, better product yield was observed using tert-butyl methyl ether (41.2% (±2.6%)) than when dichloromethane was used (36.1% (±6.1%)).

The observation that the reaction mixtures, when using either dichloromethane or tert-butyl methyl ether, separate into two phases suggested the possibility of recycling of the catalytic system, which was then explored. The initial focus was on recycling of the enzyme catalyst, which remains in the aqueous portion of the reaction mixture, which is not processed during the isolation and purification of the products. The initial recycling experiments were performed using both dichloromethane and tert-butyl methyl ether as solvents. The results have shown that when dichloromethane is used as the principal organic solvent, substantial decrease in the performance of the reaction is already observed in the first repeat of the reaction. Investigation of the reaction mixture by ¹H NMR (Figure S4) suggests that this is due to the loss of the activity of the enzyme. On the other hand, reaction performance is markedly better in the first repeat of the reaction with recycled enzyme when tert-butyl methyl ether is used as a solvent, albeit with some activity lost with respect to the initial run of the reaction. These results further support the idea that tert-butyl methyl ether can be used as a replacement solvent for dichloromentane. To investigate further the possibility of recycling of the catalytic system, we decided to also attempt to recycle the photocatalysts used in the reaction, which we believed could be recovered during the column chromatography purification of the olefin products. A comparison between recycling the enzyme catalyst and the whole catalytic system was performed in this experiment. A modest drop in activity was observed in both cases during the first reaction repeat. This was, unfortunately, followed by a steeper activity decline in the second reaction repeat. This was attributed to loss of enzyme activity in the case of enzyme recycling only, while a combination of both enzyme activity decreases and degradation of effectivity of the photocatalysts was observed in the case of recycling of the complete catalytic system. These results indicate that there is potential for recycling the catalytic system. However, further investigation is needed to achieve better outcomes.

As mentioned in previous sections, one of our aims was to find a greener alternative to dichloromethane as a solvent in this reaction. The importance of this goal is clearly demonstrated by the recent expansion of bans on dichloromethane use instituted by the EPA in the USA [33]. Furthermore, the drive towards dichloromethane replacement is also in line with principle five (safer solvents and auxiliaries) of the twelve principles of green chemistry [34]. Of the two alternative solvents chosen in this work, 2-methyltetrahydrofuran and tert-butyl methyl ether, the former was a better alternative as it can be produced from cellulose [35]. Unfortunately, this solvent proved to be unsuitable for the reaction. Tert-butyl methyl ether still presents a better alternative to dichloromethane. This is demonstrated by the fact that in the CHEM21 solvent guide its environmental score of Tert-butyl methyl ether is five, with dichloromethane being worse in this aspect with a score of seven [36]. The fact that tert-butyl methyl ether represents a better solvent choice is also demonstrated by the facts that the company Pfizer considers it a usable alternative to dichloromethane or that its use in industrial settings in Germany does not carry a special administrative burden [30,37]. However, the results of this work demonstrate that the replacement of dichloromethane with tert-butyl methyl ether has an additional benefit from the point of view of sustainability. The improved performance in terms of recyclability, even if now limited to a single recycling step, does make the reaction in tert-butyl methyl ether more efficient. This can be expressed in improvement in the process mass intensity (PMI) parameter as defined in the literature [34]. Thus, in the case of a single recycling of the enzyme catalyst, the PMI value for the reaction improves from 184 when carried out in dichloromethane, to 72, when carried out in tert-butyl methyl ether. Finally, the results obtained in this work with tert-butyl methyl ether as a solvent raise the possibility that more environmentally friendly ether solvents, such as cyclopentyl methyl ether or tert-butyl ethyl ether, could be used in this reaction [35,36].

Subsequently, we carried out the reaction on a scale that was three times that of the previous experiments. The desired product was observed. However, the yield was decreased (23.4% (±1.8%)) in comparison to the previous reactions carried out on a smaller scale. Therefore, we investigated the reaction progress using ¹H-NMR, which indicated that the scaled-up reaction will require an extended reaction time, which resulted in improvement of the yield (30.6% (±3.8%)), albeit to values still lower than those observed in the smaller scale experiments. These experiments have shown that the enzymatic hydrolysis of the starting triglycerides runs to completion within the first 48 h of the reaction, while the photocatalytic conversion of the intermediate fatty acids proceeds at a slower rate and does not reach completion even after 240 h. These experiments demonstrated that it is the photochemical conversion of the intermediate fatty acids into olefins that is the bottleneck of this reaction. The discrepancy between the yields of the small scale and scaled-up reactions, which results from issues in the second photochemical step, might be due to the fact that the same reactor (20 mL glass vial) has been used for both scales of the reaction. This might be of importance as the solvent volumes were scaled with the same factor as the reagents. Thus, the heights of the individual reaction phases within the reactor are different. Furthermore, as identical stirrers and stirring rates were used for both scales of the experiment, the extent of intermixing of the reaction phases for the two scales of the reaction is likely different. This might affect factors such as mass transfer between the phases or even light propagation within the reactor, which in turn could result in the observed negative impact on the yields in the scaled-up reaction. Thus, there is a need to further improve the photocatalytic step of the reaction in the future, where one of the avenues should be the optimization of the photoreactor.

Finally, to further demonstrate the utility of this process we carried out the scaled-up reaction with waste pork fat as an alternative starting material. The desired olefin products were observed in this reaction as well (24.0% (±2.9%)). The products from the scaled-up conversion of both used cooking oil and waste pork products were investigated using GC-MS to provide insights into the identity of the major olefin products in the mixture. This analysis has shown that the product from used cooking oil mainly contains (8Z, 11Z)-1,8,11-heptadecatriene, (8Z)-1,8-heptadecadiene, and 1-pentadecene. On the other hand, the product from waste pork oil was shown to mainly contain 1-pentadecene, 1-heptadecene, and (8Z, 11Z)-1,8,11-heptadecatriene. These key olefin products are in line with the expectations based on the fatty acid compositions of the starting waste materials.

4. Materials and Methods

4.1. Material and Chemicals

Acridine and 2-methyltetrahydrofuran were obtained from Thermo Fisher Scientific Incorporated, Waltham, MA, USA. Acetonitrile, dichloromethane, di-Potassium hydrogen phosphate trihydrate potassium dihydrogen phosphate ACS-ISO, and tert-Butyl methyl ether were sourced from CARLO ERBA Reagents (DASIT Group, Cornaredo, Italy). Amano lipase PS from Burkholderia cepacia and Chloro(pyridine)bis(dimethylglyoximato)cobalt (III) were procured from Sigma-Aldrich Corporation, Burlington, USA. Chloroform-D was acquired from Cambridge Isotopes Laboratories Incorporated, Tewksbury, USA. Ethyl acetate and Hexane were supplied by RCI Labscan Limited, Bangkok, Thailand. Silica gel was received from Supelco Incorporated, Bellefonte, USA. The waste oil materials were obtained from domestic use. ¹H-NMR spectra were acquired on a Bruker Avance-400 spectrometer (Bruker, Billerica, MA, USA) sing CDCl₃ as a solvent. Used cooking oil, a mixture of soybean oil, sunflower oil, and palm oil, was obtained from domestic consumption. Waste pork fat was obtained from domestic frying of pork belly.

4.2. GC-MS Analysis

The composition of the product olefin mixture was analyzed using an Agilent 8890GC/7000DGC/MS Triple Quad instrument (Agilent Technologies, Santa Clara, CA, USA) equipped with the HP-5ms column (30 m × 250 μm × 0.25 μm). The samples were injected at 230 °C. The carrier gas (He) flowrate was 1 mL/min. The initial oven temperature was 50 °C for 1 min, then increased to 100 °C at 15 °C/min, then increased to 210 °C at 10 °C/min and held for 1 min, then increased to 310 °C at 5 °C and held for 8 min. Samples were analyzed in full-scan mode (m/z 100–500).

4.3. Initial Procedure for Biophotocatalytic Conversion of Used Cooking Oil

Used cooking oil (90 mg), acridine (12 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), were mixed with dichloromethane (1.5 mL) and acetonitrile (0.15 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (200 mg) was dissolved in phosphate buffer (1 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 60 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with dichloromethane (2 × 2 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.4. Procedure for Biophotocatalytic Conversion of Used Cooking Oil with Extended Reaction Time

Used cooking oil (90 mg), acridine (12 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), were mixed with dichloromethane (1.5 mL) and acetonitrile (0.15 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (200 mg) was dissolved in phosphate buffer (1 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 144 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with dichloromethane (2 × 2 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.5. Procedure for Selection of Alternative Solvent for Biophotocatalytic Conversion of Used Cooking Oil

Used cooking oil (90 mg), acridine (12 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), were mixed with 2-methyltetrahydrofuran (or tert-butyl methyl ether) (1.5 mL) and acetonitrile (0.15 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (200 mg) was dissolved in phosphate buffer (1 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 144 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with the solvent originally used in the reaction mixture (2 × 2 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.6. Procedure for Enzyme Recycling in Biophotocatalytic Conversion of Used Cooking

Used cooking oil (90 mg), acridine (12 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), were mixed with dichloromethane (or tert-butyl methyl ether) (1.5 mL) and acetonitrile (0.15 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (200 mg) was dissolved in phosphate buffer (1 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 144 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with the organic solvent originally used for the reaction (2 × 2 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. A repeat of the reaction was set up using the recovered aqueous phase from the first run of the reaction to which used cooking oil (90 mg), acridine (12 mg), chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), dichloromethane (or tert-butyl methyl ether) (1.5 mL), and acetonitrile (0.15 mL) were added. The repeat reaction was conducted, worked up, and analyzed in a manner identical to the initial run. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.7. Procedure for Enzyme Recycling in Biophotocatalytic Conversion of Used Cooking Oil

Used cooking oil (90 mg), acridine (12 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (12 mg), were mixed or tert-butyl methyl ether (1.5 mL) and acetonitrile (0.15 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (200 mg) was dissolved in phosphate buffer (1 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 144 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with the organic solvent originally used for the reaction (2 × 2 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The other components of the crude reaction mixture were obtained from the column using a mixture of ethyl acetate (70% v/v) and methanol (30% v/v). A repeat of the reaction was set up using the recovered aqueous phase from the first run of the reaction, which were added to a solution of the materials obtained from the column using the polar mobile phase dissolved in tert-butyl methyl ether (1.5 mL), and acetonitrile (0.15 mL). The repeat reaction was conducted, worked up, and analyzed in a manner identical to the initial run. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.8. Procedure for Scaled Up Biophotocatalytic Conversion of Used Cooking Oil

Used cooking oil (270 mg), acridine (36 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (36 mg), were mixed with tert-butyl methyl ether (4.5 mL) and acetonitrile (0.45 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (600 mg) was dissolved in phosphate buffer (3 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 144 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with tert-butyl methyl ether (2 × 5 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.9. Procedure for Monitoring Reaction Progress of Scaled Up Biophotocatalytic Conversion of Fatty Acid Waste Materials

Used cooking oil (270 mg), acridine (36 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (36 mg), were mixed with tert-butyl methyl ether (4.5 mL) and acetonitrile (0.45 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (600 mg) was dissolved in phosphate buffer (3 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Samples (1 mL) of the organic portion of the reaction mixture were collected every 48 h. The samples were evaporated to dryness, analyzed by ¹H-NMR in chloroform-d. The percentage of reaction progress was calculated using Equation (1).

$$Reactionprogress(\%)={\displaystyle \frac{\raisebox{1ex}{$I_{5.80ppm}^t$}\!\left/ \!\raisebox{-1ex}{$I_{5.35ppm}^t$}\right.}{\raisebox{1ex}{$I_{5.80ppm}^p$}\!\left/ \!\raisebox{-1ex}{$I_{5.35ppm}^p$}\right.}}\times 100$$

(1)

where $I_{5.80ppm}^t$, $I_{5.35ppm}^t$, $I_{5.80ppm}^p,$ and $I_{5.35ppm}^p$ are the integral of the signal at 5.80 ppm in the reaction at a given time t, the integral of the signal at 5.35 ppm at a given reaction time t, the integral of the signal at 5.80 ppm in the product, and the integral of the signal at 5.35 ppm in the product, respectively.

Subsequently, the material used for the analysis was recovered, dissolved in tert-butyl methyl ether (1 mL), and returned to the reaction mixture.

Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with tert-butyl methyl ether (2 × 5 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The crude and purified products were analyzed by ¹H-NMR in chloroform-d.

4.10. Procedure for Scaled Up Biophotocatalytic Conversion of Waste Pork Fat

Waste pork fat (270 mg), acridine (36 mg), and chloro(pyridine)bis(dimethylglyoximato)cobalt (III) (36 mg), were mixed with tert-butyl methyl ether (4.5 mL) and acetonitrile (0.45 mL) in a clear glass vial. Subsequently, Amano lipase PS from Burkholderia cepacia (600 mg) was dissolved in phosphate buffer (3 mL, pH 7.0) and added to the reaction mixture. The reaction mixture was stirred vigorously for 240 h under the irradiation of two UV lights (LED Light, 395 nm, 50 W, Dreamcast light, Samut Prakan, Thailand). Subsequently, the organic and aqueous phase of the reaction mixture were separated, and the aqueous phase was extracted with tert-butyl methyl ether (2 × 5 mL). The combined organic fractions were evaporated to dryness. Purified olefin products were obtained after column purification on silica using hexane as the mobile phase. The purified products were analyzed by ¹H-NMR in chloroform-d.

5. Conclusions

The possibility of utilizing a recently reported method for conversion of fats and oils into olefins for the conversion of waste materials was investigated for used cooking oil. The possibility to perform the reaction using tert-butyl methyl ether as a replacement for originally used dichloromethane was demonstrated. This solvent replacement resulted in enhancement of the reaction yield from 36.1 to 41.2% and positively affected the possibility of recycling the catalysts. The reaction was demonstrated in a scaled-up version, albeit with a lower yield (23.4%). This was ascribed to the need for extension of the reaction time in comparison to the smaller scale reaction. The extension of the reaction time resulted in the improvement of the yield to 33.2%. The reaction was also demonstrated with waste pork fat as a second fat-based waste material. GC-MS analysis of the products indicated that the composition of the product olefin mixture corresponded with the fatty acid makeup of the starting materials. The possibility to utilize the studied reaction to convert used cooking oil to olefins in a more environmentally friendly solvent was demonstrated. However, further optimization of the reaction, especially in terms of photoreactor development, is needed to achieve better performance.