3.1. OCL Synthesis and Analysis
In the preliminary studies, the tests of BV oxidation of cyclohexanone by various peracids were performed (
Scheme 4). Peracids were synthesized according to the literature [
15]. Linear carboxylic acids (C
8, C
9, C
10, C
12) were stirred with 95% of sulfuric acid at 10 °C, and next 50% H
2O
2 was added dropwise, keeping the temperature below 30 °C. After the reaction, water with ice was added in order to cool the mixture and precipitate the solid peracid. Then, diethyl ether was added to dissolve the peracid, and the formed biphasic system was separated. The organic phase was next collected and dried under anhydrous MgSO
4. The product was recrystallized from petroleum ether at −18 °C and then the corresponding peracid was obtained via vacuum filtration with >99% purity (iodometric titration).
The main goal of the preliminary studies was to identify the reaction conditions that promote the formation of OCL. To compare the effect of the solvent on the CNON conversion and the kind of obtained products, toluene, cyclohexane or excess of cyclohexanone were used. Various peracids were introduced to the process to determine their influence on the amount and kind of synthesized OCL. The experiments were carried out at 25–55 °C up to 4 h with molar ratio of cyclohexanone to peracid (CNON/peracid) ranging from 1:0.5 to 1:2. The conversion of cyclohexanone was analysed by GC, while the selectivity to OCL was calculated by
1H NMR spectroscopy (
Figure 1).
As shown in
Table 1 the reaction time and temperature have a significant influence on the selectivity of the process. After 4 h of reaction at 45 or 55 °C in cyclohexane or toluene, using perC
10 as an oxidant, almost complete conversion and selectivity to OCLs were achieved. It is assumed that under these conditions, in the presence of peracid most likely the ring opening of ε-caprolactone, formation of the hydroxy acid and then its oligomerization successively occurred. This confirms that the oligomers can be obtained by a one-pot two-step BV oxidation-oligomerization process (
Scheme 4). Extending the reaction time from 60 min to 240 min resulted in an increase in selectivity to OCLs at 45 or 55 °C in cyclohexane or toluene, using perC
10. In turn, only 9% selectivity towards OCLs was reached at 25 °C in cyclohexane after 1 h, while at 45 °C it was already 62%. However, the reaction cannot be carried out at temperatures higher than 65 °C due to the thermal decomposition of the peracid. The most effective peracid for proposed one-pot OCL synthesis is perC
10 (58% selectivity to OCL after 60 min), which is additionally in agreement with the safety of the process. In our previous work the stability tests for several peracids (perC
6-C
12) were performed concerning sensitivity to mechanical impulse (shock and friction), electrical (spark), and thermal sensitivity (temperature and heat of decomposition) and perC
10 was selected as the safest peracid [
15]. For other peracids, selectivity to OCLs did not exceed 19% after 60 min. Regardless of the solvent, the use of an excess of peracid in relation to CNON increased both the conversion of CNON and the selectivity to OCLs. For toluene, 58% selectivity to OCLs was achieved with a two-fold molar excess of peracid (90% CNON conversion), while an equimolar amount of CNON:perC
10 allowed for obtaining only 28% selectivity and 74% CNON conversion after 60 min. A similar phenomenon was observed for cyclohexanone. The kind of solvent did not strongly affect the conversion nor selectivity. Preliminary studies also confirmed that the use of CNON in excess, and thus the elimination of additional solvents, leads to only traces of OCLs (up to 5%).
In order to determine the structure of obtained OCLs, the oxidation of cyclohexanone was scaled up four times. The experiments were carried out with deficiency of peracid (molar ratio CNON:perC
10 1:0.5) at 55 °C in cyclohexane or in toluene, approaching the almost 100% selectivity to OCLs. The reaction mixtures were first concentrated using a rotary evaporator to get rid of the solvent and unreacted cyclohexanone. Next, methanol was added, the flask was cooled and kept overnight at −18 °C to precipitate the OCLs. The procedure was repeated twice, achieving the purity of the product over 99% (
Figure 2). For the characterization of the product,
1H NMR spectroscopy and MS MALDI TOF spectrometry were exploited.
It is worth noticing that the esterification of OCLs with decanoic acid during the reaction as well as with methanol during crystallization of OCLs can occur. In the
1H NMR spectrum (
Figure 3) for the process carried out in toluene, the triplet visible at approximately 4.05 ppm corresponds to the CH
2 group protons that are attached to the ester group by a covalent bond with oxygen. The higher the ratio of this signal to the signals at approximately 3.65 ppm, the longer the OCLs, assuming no other end groups in the molecule. In this case, a very weak signal present in the range of 3.50–3.70 ppm indicates that there are a few compounds with a CH
2 group attached to a free -OH group. The alcohol chain ends, however, may have been esterified with decanoic acid, which would confirm the presence of an asymmetric signal at approximately 4.05 ppm and the methyl group signal of the aliphatic acid chain at 0.85 ppm. It is also possible that cyclic oligomers are present in the sample, which would also not give signals in the range of 3.50–3.70 ppm with significant signal amplification at 4.05 ppm. Therefore, the NMR spectrum is not suitable for unambiguous information about the chain length of the oligomer and the number of oligomers of different chain lengths in the sample.
Hence, MS MALDI TOF analysis of OCLs was performed. In the spectrum (
Figure 4), the most intense signal is 1108 m/z (one repeating unit of oligo(ε-caprolactone) possesses a molecular mass of 114 Da). This signal is a part of one set of peaks and does not correspond to a cyclic oligomer (multiplying the repeating unit n times by 114 Da), which suggests the presence of linear oligomers with end groups. It was assumed that OCLs were singly ionized and m/z ratio corresponds to the mass in Da. On the other hand, the signal corresponding to the mass 1006 m/z does not belong to the same sets of peaks, and therefore it corresponds to the oligomers obtained as a result of incorporation of successive repeating units into the compound with a different end group. In-depth analysis of the spectrum confirms that there were two major sets of oligomers with different chain lengths in the spectrum. The signal from the longest oligomer 2364 Da corresponds to 19 repeating units in the oligomer molecule terminated with an -OH group esterified with decanoic acid, after ionization of carboxyl group by sodium coming from the glass under analysis conditions. Analysis of signals in the spectrum, together with the
1H NMR spectrum, showed that the second set of oligomers most likely corresponds to the same end groups but is ionized with potassium coming from the matrix. The result of the MS MALDI TOF analysis explains why no signals from CH
2 groups at the free -OH group were observed in the
1H NMR spectrum—such groups do not exist in the sample due to the esterification with decanoic acid.
MS MALDI TOF analysis of the OCLs isolated from the reaction carried out in cyclohexane (
Figure S1) was performed. In the obtained spectrum, the most intense signal equals 1222 m/z, which corresponds to 9 repeating units terminated with a sodium-ionized carboxyl group and hydroxyl group esterified with decanoic acid. The final signal corresponds to a mass of 2821 Da as assigned to the 23 repeating units of oligo(ε-caprolactone). The
1H NMR spectrum (
Figure S2) confirms the results of the MS MALDI TOF analysis. Based on the results of these analyses, the proposed structure of OCLs is presented in
Scheme 5.
Next, the influence of the molar ratio of CNON:perC
10 ranging from 1:0.5 to 1:2 on the structure of OCLs was studied. Due to the negligible effect of the solvent the processes were carried out in cyclohexane as it is characterized by a higher vapor pressure and thus easier working up of the reaction mixture.
1H NMR spectra of products from these processes indicate that oligo(ε-caprolactone) with a similar structure was obtained (
Figure 5,
Figures S3 and S4). The ratio of the signals at approximately 4.05 ppm and 3.65 ppm reached similar values (approximately 13.5). The average molecular weight of the obtained oligomers was approximately 1500 Da, however it is not known whether the signals corresponded to one or several sets of OCLs nor whether there were any other compounds in the sample that could give similar signals (such as cyclic oligomers or the presence of the group esterified with decanoic acid in the sample). The signal at 4.05 ppm is not a symmetrical triplet, confirming the presence of a decanoic acid ester as a chain end in at least one of the resulting oligomer series. The presence of decanoic acid or its ester is clearly evidenced by the signal at approximately 0.85 ppm, coming from the methyl group. On the basis of the obtained spectra, it is suspected that the peracid to substrate ratio at the stage of the oxidation process does not significantly affect the structure of the obtained OCLs, which must be confirmed by further MS MALDI TOF analyses.
MS MALDI TOF of OCLs were performed in the range 600–4000 m/z (
Figure 6,
Figures S5 and S6)). The average molecular weight is in each case approximately 1100–1300 Da, which is a slightly lower value than that estimated by interpreting the
1H NMR spectra. This suggests that in MS spectra it is necessary to find signals from OCLs with a free terminal -OH group, and that cyclic oligomers could also arise, which would increase the integral value of the proton signal in the
1H NMR spectrum at approximately 4.05 ppm in relation to the proton signals at approximately 3.65 ppm. The higher the ratio of CNON:perC
10 used in the oxidation step, the shorter the oligomers obtained. This is most likely due to a lower concentration of peracid in the reaction mixture leading to incorporation of successive units into the structure of already formed oligomers than into the structure of monomer and short-chain oligomers that start the next chain. Apparently, there is not enough decanoic acid to esterify the -OH group and to complete the formed oligomer chain. In the case of a higher concentration of perC
10 in the mixture, subsequent peracid molecules take part in the oxidation of cyclohexanone faster, which increases the reaction course and hence short-molecule oligomers are formed. Oligomers can be terminated more rapidly with ester groups due to a higher amount of decanoic acid in the reaction system, which is the process that completes the incorporation of subsequent mers into the oligomer structure. Moreover, two main sets of oligomers were observed in the sample coming from the process conducted with a deficiency of perC
10. Less peracid in the system leads to OCL being obtained more selectively. The difference in the structure of OCL obtained with the perC
10 deficiency process compared to the others was observed. For these OCLs, one sets of peaks corresponds to a group of oligomers terminating with an esterified -OH group from one side and with a carboxyl group ionized with potassium from another one. Another set corresponds to oligomers with a free -OH group at the end of the chain and a lithium-ionized carboxyl group. The most intense signal in this spectrum—1124 Da—corresponds to the oligomer concerning 8 repeating units and belongs to the sets of peaks ionized by potassium. In the case of samples from processes carried out with a higher molar ratio of perC
10:CNON, an additional set of peaks coming from cyclic oligomers was observed. The other two series correspond to an OCLs terminated with an -OH group esterified with decanoic acid or non-esterified and with a carboxyl group ionized with lithium or potassium. The only difference is that shorter oligomers were obtained in a process using an excess of peracid. The similar ratio of the signals at 4.10 to 3.65 were caused by the fact that in the first case the longest oligomers were obtained, and in the second case cyclic oligomers did not give a signal at 3.65 ppm in the
1H NMR spectrum, which increased the value of the estimated molecular weight of the oligomers. In the case of the oxidant deficiency process, the highest molecular weight of the oligomer (2952 Da) observed on the spectrum corresponds to the 24 repeating units incorporated into OCL, with the -OH ending group esterified with decanoic acid. In the case of the oxidation process carried out with an equimolar amount of perC
10, the highest molecular weight of the oligomer is 2509 Da, which corresponds to the 22 repeating units incorporated into cyclic OCL, and in the case of the oxidation process carried out with a double excess of oxidant, the OCL with the highest mass was obtained (1694 Da), which corresponds to only 13 repeating units in the structure of OCL with an -OH esterified ending group in the molecule. Hence, it is possible to obtain εCL oligomers more selectively, which form linear esterified or non-esterified molecules or cyclic oligomers, depending on the implemented conditions.
Therefore, the following conclusions followed from performed experiments: (a) CNON: perC
10 ratio affects both: the type and molecular weight of the obtained εCL oligomers; (b) the higher concentration of perC
10 in the reaction system, the shorter oligomers were obtained after the same time; and (c) in the case of conducting the cyclohexanone oxidation process with an insufficient amount of oxidant, cyclic oligomers were hardly obtained. The structure of the created OCLs were shown in
Scheme 6 and a proposed oligomerization mechanism is presented in
Scheme 7.
3.2. Analysis of OCLs Separated during Distillation
In the industrial oxidation of cyclohexanone by perC10 acid to εCL, the product is isolated by vacuum distillation. Therefore, in the next step, it was checked whether exposing the post-reaction mixture to a high temperature during distillation under reduced pressure influences the type of end groups, dispersibility, and average molecular weight of the byproduct OCLs. The oxidation process was carried out in toluene (molar ratio CNON:perC10 1:0.6, 55 °C, 4 h). After the reaction, the mixture was sent to a distillation process to evaporate the solvent at a pressure of 500 mbar. The mixtures were then preheated to 100 °C, the pressure was lowered to approximately 15 mbar, and next mixtures were heated gradually to 200 °C, collecting the rest of the solvent, cyclohexanone, and a small amount of the lactone and 6-hydroxycaproic acid at 10 mbar. To characterize the OCLs after distillation, the distillation residue was dissolved in methanol and left overnight. The obtained oligomers were then analyzed by NMR, MS MALDI TOF, and TGA.
Analyzing the
1H NMR spectrum (
Figure 7a) of the precipitate obtained after distillation of mixture from the process carried out in toluene, it can be seen that its main component is oligo(ε-caprolactone). The characteristic peaks at approximately 4.10 ppm and 3.65 ppm (signal ratio approximately 34) may indicate that only a small amount of chain terminated with free -OH groups are present in the oligomer. Most likely decanoic acid is incorporated into the chain endings. Such a high value of this ratio may be also due to the presence in the system a significant amount of oligo(ε-caprolactone) in the cyclic form or oligomers with longer chains than those obtained so far because these compounds stayed at high temperature for a much longer time. Based on the results of the experiments described above, it is predicted that at this molar ratio of CNON: perC
10 acid, such a high value of signal ratio is mainly due to the presence of the high-molecular weight oligomers and not cyclic compounds. Moreover, it is apparent that decanoic acid is most likely incorporated into the chain ends, which would not be possible with only cyclic oligomers.
13C NMR spectrum (
Figure 7b) also shows that the basic component of the obtained precipitate is oligo(ε-caprolactone) but gives no additional information.
MS MALDI TOF analysis (
Figure 8a) shows that the sample was composed of oligomers with different end groups in the chain, giving two sets of signals. The most intense 1068 Da signal suggests that it is a 9 repeating units incorporated into OCLs terminated with -OH and -COOH groups ionized with sodium. The signal of 2935 Da belongs to the same series of signals, corresponding to 24 repeating units terminated in the same way. The second signal set visible on the spectrum, but present in a much smaller amount, is the sodium-ionized oligo(ε-caprolactone) peaks with the -OH group esterified with decanoic acid, which includes, for example, the signal at 1107 m/z corresponding to the 8 repeating units in the structure of oligomer. The spectrum shows that oligomers with a molecular weight of up to 3000 Da were obtained in the process. The interpretation of the MS MALDI TOF spectrum is therefore consistent with the preliminary analysis of the NMR spectra.
TGA-DSC analysis (
Figure 8b) was also performed. The obtained spectra show that the loss of mass in the sample only occurs in the temperature range from approximately 310 to 480 °C and it corresponds to the range of oligo(ε-caprolactone) presented in the literature [
20,
21]. The absence of other effects and the shape of the curve indicate that decanoic acid is present in the sample as a chain ending rather than free acid. Therefore, it can be concluded that the obtained oligo(ε-caprolactone) consists of two series of oligomers with chains terminated with carboxyl groups and an alcohol or ester, with a high purity and an average molecular weight in the range of 1200–1500 Da.
The experiments of post-reaction mixtures after distillation under conditions similar to the industrial show that: (a) it is possible to obtain an oligo(ε-caprolactone) mixture with a molecular weight of up to 3000 Da, (b) at a high temperature, mainly linear oligomers terminated with a free -OH group or an -OH group esterified with decanoic acid, and (c) it is possible to isolate oligo(ε-caprolactone) with a purity of approximately 99% with a satisfactory efficiency (69% of the mass of cyclohexanone introduced into the process).
Scheme 8 shows the structures of oligo(ε-caprolactone) isolated from the performed process.