Benoit
Briou
*a,
Lucas
Jégo
a,
Thomas
De Dios Miguel
b,
Nicolas
Duguet
*b and
Sylvain
Caillol
*c
aOrpia Innovation, CNRS, Bâtiment Chimie Balard, 1919 Route de Mendes, 34000 Montpellier, France. E-mail: b.briou@orpiainnovation.com
bUniv Lyon, Université Claude Bernard Lyon1, CNRS, INSA-Lyon, CPE-Lyon, Institut de Chimie et Biochimie Moléculaires et Supramoléculaires, ICBMS, UMR 5246, Equipe CAtalyse, SYnthèse et ENvironnement (CASYEN), 1 rue Victor Grignard, 69100, Villeurbanne, France. E-mail: nicolas.duguet@univ-lyon1.fr
cICGM, Univ Montpellier, CNRS, ENSCM, Montpellier, France. E-mail: sylvain.caillol@enscm.fr
First published on 16th May 2023
Phenolic thermosets are mainly synthesized from two very toxic and petroleum-based molecules, phenol and formaldehyde. This study aims at replacing these two species with a unique bio-based phenolic molecule bearing an aldehyde function. For this purpose, cardanol, a phenolic component extracted from cashew nut shells, has been chemically modified to give 8-(3-hydroxyphenyl)octanal. Two synthetic routes were investigated, each one consisting in three steps and sharing the same key intermediate. The first route consists in the epoxidation of cardanol, followed by hydrolysis to form diols, and finally an oxidative cleavage of these diols using NaIO4. A second route has been considered consisting in an epoxidation, followed by the formation of β-hydroxy hydroperoxides with H2O2, and finally a thermal cleavage. The green potential of these two methods was compared and discussed with other routes described in the literature. The resulting aldehyde was subsequently used as an AB monomer to obtain phenolic network by direct step growth homopolymerization, enabling the substitution of both phenol and formaldehyde at the same time. The resin formed thereby shows a low Tg due to the C8 alkyl chain of 8-(3-hydroxyphenyl)octanal as well as a good thermal stability.
Sustainability spotlightPhenolic thermosets have found numerous applications thanks to their high thermal and chemical resistance and their rigid mechanical behavior. However, most of them are prepared from phenol and formaldehyde, that are fossil-based reagents classified as CMR (Carcinogenic Mutagenic Reprotoxic). To address these concerns, we have prepared a unique bio-based platform molecule, incorporating both a phenol and an aldehyde moiety, which was synthesized from cardanol, a phenolic compound extracted from cashew nut shells. Several routes were investigated and discussed based on their safety score and green metrics. The resulting monomer was used to prepare phenolic networks with improved sustainability profile (bio-based origin, no CMR compounds). |
It is acknowledged in organic chemistry that a double bond of a molecule could be cleaved, shortening the alkyl chain involved.20 These reactions can also be applied to cardanol. In fact, cleavage of the cardanol double bonds was already reported by Graham et al.21 in 2002 via reductive ozonolysis and permitted to obtain 8-(3-hydroxyphenyl)octanal with high yield (96%). This molecule has a C8 alkyl chain and the advantage to bear both a phenol and an aldehyde moiety. The presence of these two functional groups on the same molecule would allow a step growth homopolymerization without using formaldehyde or other aldehydes. However, ozonization is a dangerous method due to the intrinsic toxicity of ozone and the explosive nature of ozonide intermediates. Another synthetic pathway was described through an oxidative cleavage of diols but it led to a low yield.21
In this context, it would be highly desirable to find alternative methods to produce 8-(3-hydroxyphenyl)octanal. Given the structural similarities of cardanol and unsaturated vegetable oils, it might be possible to transpose some selective (towards the formation of aldehydes) methods reported on vegetable oil derivatives to cardanol. For instance, it was reported that fatty diols can be cleaved into aldehydes using NaIO4.22 Recently, it was also demonstrated that fatty epoxides can be cleaved into aldehydes, through the formation of β-hydroxy hydroperoxides and their subsequent thermal cleavage in flow.23 To the best of our knowledge, such methods of forming aldehydes have never been reported on cardanol or CNSL derivatives.
In this paper, we report a greener and safer synthesis of 8-(3-hydroxyphenyl)octanal from cardanol epoxides, through the formation of either diols or β-hydroxy hydroperoxides.24 The different methods to obtain 8-(3-hydroxyphenyl)octanal are compared and discussed based on green chemistry metrics. The 8-(3-hydroxyphenyl)octanal homopolycondensation and the thermal properties of the resulting phenolic thermosets were explored (Fig. 1C).
![]() | (1) |
E-factor is calculated following the eqn (2):
![]() | (2) |
The atom efficacy is calculated following the eqn (3):
Atom efficacy = yield × AE | (3) |
Evaluation of the safety score (SC) was established using the safety data sheets (SDS) of the products involved in the synthesis of the molecule of interest, whether reagents, catalysts or solvent, during each step: reaction, extraction, purification. A score is assigned for each category of hazard considering the number of sub-categories of this hazard. The maximum hazard category score is the category number +1. The minimum score is therefore +1 if it is the lowest category, 0 if there is no such danger. In our view, the following hazards: corrosive, irritant and STOT, present a less significant hazard than the other hazards, their rating is therefore halved. In contrast, because of the increased danger for humans and the process, an additional note is assigned to the following products: +2 for flammable gas, +3 for radioactive and/or explosive. The SC of a product is then the sum of the notes of each category of danger present on its SDS.
First, cardanol was epoxidized using 30%-H2O2 (1.5 equiv./alkene) in the presence of formic acid. Under these conditions, the overall epoxidation of double bonds reached 82%. In details, the conversion of internal double bonds was almost complete while the epoxidation of the terminal alkene only reached 55%. This clearly demonstrates that internal double bonds are more reactive than terminal ones, as previously reported28 (Fig. 3). The uncomplete epoxidation of the terminal double bond does not represent a problem considering that the aim of this approach is to cleave the cardanol's alkyl chain at the first internal double bond (position 8). In fact, this uncomplete epoxidation limits the quantity of side products that could be produced during the oxidative cleavage sequence. In the following step, the mixture of cardanol epoxides (CxEp) was hydrolyzed to CxDOH in the presence of phosphorous acid to give the corresponding cardanol diols with 98% yield. The presence of hydroxyl moieties has been confirmed by NMR. On the 1H NMR spectrum of the cardanol diol (CxDOH) (Fig. 3), the signals j,k,l (3.0–3.2 ppm) of the epoxides were shifted to the o,m,n signals (3.4–4.2 ppm), which is consistent with a diol chemical shift.
In addition, it was clearly noticeable in the 13C NMR spectrum that the signal of epoxy carbon shift from 45–60 ppm to the region of carbon bringing hydroxy function at 65–75 ppm, confirming the previous inference. Finally, the cleavage of alkyl chain by oxidation of the diol function was performed using NaIO4 (1 equiv./diol function). The calculation of the quantity of NaIO4 involved all the internal diols able to react such as about 1.5 moiety by molecule. The cleavage yield was 95% but the crude product included 8-(3-hydroxyphenyl)octanal and a mixture of other aldehydes. After purification by column chromatography, 8-(3-hydroxyphenyl)octanal was obtained with a yield of 52%. The loss in yield could be explained by the difficulty to separate each aldehyde from the other and some 8-(3-hydroxyphenyl)octanal has been sacrificed to obtain pure molecules. A purification by distillation could avoid such loss, as previously demonstrated by Graham et al.21 The structure of 8-(3-hydroxyphenyl)octanal was confirmed by 1H NMR spectrum (Fig. 3). Aromatic signals between 6.5 and 7.5 ppm and the hydroxy signal stated that the compound brings a phenol moiety. Alkyl chain signals between 1.2 and 3.0 ppm was still noticeable but their number has decreased. The ending CH3 signal at 0.8–0.95 ppm from CxDOH has disappeared. A new signal at 9.7 ppm validates the presence of an aldehyde function. These changes of signal were corroborating with the cleavage of the cardanol alkyl chain, replacing the CH3 by an aldehyde moiety at the end of shorter alkyl chain (C8 instead of C15). No traces of epoxy, diol or double bonds were detected. This 3-step synthesis gave 8-(3-hydroxyphenyl) octanal with a total yield of 74% (42% after purification by column chromatography).
In this work, 8-(3-hydroxyphenyl)octanal was prepared from cardanol in 3 steps following an epoxidation/hydrolysis/oxidative cleavage sequence (Scheme 1, route b). Despite that good yields can be achieved through this route, the key step requires the use of NaIO4 as an oxidant (1 equiv./diol function), thus resulting in the production of large quantities of waste. Moreover, NaIO4 is a harmful reactant and is, at some concentration, toxic for both humans and the environment. In this context, we have also investigated an alternative and potentially greener route (Scheme 1, route c).
We have previously reported that epoxidized fatty derivatives can be cleaved into the corresponding aldehydes through the formation of β-hydroxy hydroperoxides.23 These species were formed by ring-opening of epoxides with H2O2 and were next thermally cleaved in flow conditions to give the desired aldehydes.23 Based on these results, we envisioned that a similar strategy could be applied to produce 8-(3-hydroxyphenyl)octanal from cardanol, that is a much more complex substrate (Scheme 2). The first step of epoxidation was performed in similar conditions that was previously described in the route b. Then, the mixture of cardanol epoxides CxEp in t-amyl alcohol was treated at 20 °C for 16 h with aqueous 50%-H2O2 (1.1 equiv./epoxide function) in the presence of phosphotungstic acid (PTA, 0.1 mol%/epoxide function) to give a mixture of cardanol β-hydroxy hydroperoxides CxHP. The formation of hydroperoxide functions can be observed in 1H NMR spectrum by the presence of a broad peak around 11.5 ppm. The broad peak indicates the formation of many regioisomers and diastereoisomers as could be expected for cardanol derivatives. Finally, the mixture of β-hydroxy hydroperoxides was thermally cleaved at 300 °C under flow conditions (residence time of about 1 min). Satisfyingly, the formation of the desired aldehyde CAl was detected in GC along with other cleavage products. Formaldehyde was not detected under these reaction conditions. However, its formation could not be excluded considering that the terminal double bond of cardanol was partially converted to the epoxyde in the first step.
![]() | ||
Scheme 2 Preparation of 8-(3-hydroxyphenyl)octanal from cardanol through thermal cleavage of β-hydroxy hydroperoxides. |
The purification of the mixture by column chromatography afforded 8-(3-hydroxyphenyl)octanal with 23% yield. This relatively low yield can be explained by the fact that the ring opening of epoxides with aqueous 50%-H2O2 is not completely selective. Indeed, the competitive ring opening of epoxides with H2O (from aqueous H2O2) leads to the formation of the corresponding diols, that can not be cleaved under the thermal conditions.
As observed with vegetable oil derivatives, the selectivity towards β-hydroxy hydroperoxides is higher than 90% when the fatty chain bears only one double bond (i.e. using methyl oleate derivatives). However, this selectivity drops to about 50% when the fatty chain bears multiple double bonds (i.e. using methyl linoleate derivatives).23 Considering that cardanol is actually a mixture of compounds with a fatty chain bearing none C0 (5–8%), one C1 (48–49%), two C2 (16–17%) or three double bonds C3 (29–30%), a lot of diols were produced in that case. This could explain why only a small amount of the desired aldehyde was produced. To support this hypothesis, the mixture of epoxidized cardanol CxEp was purified by column chromatography (see ESI† for details) (Scheme 3). A fraction of cardanol mono-epoxides was isolated with 19% yield. From 1H NMR, it was found that this fraction of mono-epoxides contains 87% of saturated chain (C1Ep:0), 3% of one double bond (C1Ep:1) and 10% of two double bonds (C1Ep:2) (Fig. 4, top). The formation of these 3 cardanol mono-epoxides was also validated by HRMS (see ESI†). The mixture of mono-epoxides was treated with 50%-H2O2 under the same conditions to give the corresponding β-hydroxy hydroperoxide C1HP with 43% yield after purification by column chromatography. In that case, the formation of the β-hydroxy hydroperoxide function was assessed by 1H NMR with the presence of a sharp peak at 11.2 ppm (Fig. 4, middle). The GC analysis of compound C1HP shows that it can be readily cleaved at 300 °C (temperature of the GC injector) to give the desired aldehyde CAl and heptanal (Fig. 5).
![]() | ||
Fig. 5 GC chromatogram of cardanol mono-β-hydroxy hydroperoxide (C1HP), showing its thermal cleavage to heptanal and 8-(3-hydroxyphenyl)octanal (CAl) (GC injector temperature = 300 °C). |
Finally, β-hydroxy hydroperoxide C1HP was thermally cleaved at 300 °C in a preparative way under flow conditions to give the desired 8-(3-hydroxyphenyl)octanal with 32% yield (Fig. 4, bottom). It should be noted that, in order to simplify the work-up of the reaction, heptanal was not recovered but was removed by distillation (rotary evaporator).
These results show that cardanol β-hydroxy hydroperoxide C1HP can be readily cleaved into 8-(3-hydroxyphenyl)octanal CAl with high selectivity. Therefore, it confirms our hypothesis that the presence of diols prevents the formation of the desired aldehyde. Indeed, the overall selectivity is plummeted by the fact that the ring-opening of the mixture of cardanol epoxides is not selective due to the use of an aqueous solution of H2O2. This problem could be probably solved by using an ethereal solution of H2O2 but this raises safety problems, especially when working on the large scale. The in situ generation of H2O2 would afford – by far – a safer option.
In their paper, Graham et al. described ozonization as an effective method for obtaining 8-(3-hydroxyphenyl)octanal by focusing only on the yield of the reaction. They compared another non-optimized oxidative cleavage methods using KIO4. We decided to include these synthetic pathways in our study. We are aware that in order to achieve an oxidative cleavage, strong oxidants should be used, which often leads to some drawbacks such as hazard and toxicity. Therefore, it is important to identify their impact and find promising alternatives or compromises.
Before starting the discussion on the various methods to obtain 8-(3-hydroxyphenyl)octanal, it is worth pointing out that the starting material used in these reactions is cardanol. Cardanol, on its own, is non-toxic, bio-sourced and comes from the recycling of inedible agricultural waste. The use of a cardanol derivative, in addition, as a mono-component, for the synthesis of phenolic resin is already in itself an eco-responsible advance for the substitution of phenol and formaldehyde, both classified as CMR, in the production of these materials.
About the validity of the synthesis pathway, the aim is to find an alternative to ozonization. Although effective, ozone shows a significant potential for danger and needs economical investment to be used without harm. Table 1 compares the routes according to the dangerousness of the molecules used (see ESI† for details). The first thing that can be noticed is that the most dangerous part does not come from reaction solvents (average SC of 12.75) but from reactants (average SC of 30.75). To reduce the SC score of the pathway, the best options are either to use water as a solvent or to use the same solvent for all the steps, if possible. Among reactants, the oxidants are the most dangerous chemicals, the most reactive being ozone (SC 21). It is therefore necessary to find an alternative to the use of ozone in this process of cleavage of cardanol double bonds. However, when comparing each pathway based on their total safety score, the route a exhibits the lowest value (SC of 34) by comparison with other routes that are relatively close (SC of 42–51). In contrast, when comparing the total safety score per number of chemicals, the routes b1, b2 and c (SC/chemicals of 5.3–6.4) appear by far safer than the route a (SC/chemicals of 11.3). This means that the hazard is much more distributed throughout the process for the routes b1, b2 and c than for the route a, where the hazard is concentrated on only one (ozonization) step.
Entry | Route | Steps | SC of reactants | SC of solvents | Most hazardous molecule (pts) | Number of products | Final SC | Final SC/Nbr of products |
---|---|---|---|---|---|---|---|---|
a Synthesis with NaIO4. b Synthesis with KIO4. | ||||||||
1 | a | 2 | 28 | 6 | Ozone (21) | 3 | 34 | 11.3 |
2 | b1a | 3 | 40 | 11 | Sodium periodate (15) | 8 | 51 | 6.4 |
3 | b2b | 3 | 33.5 | 13 | H2O2 (9.5) | 10 | 46.5 | 4.3 |
4 | c | 3 | 21.5 | 21 | H2O2 (9.5) | 8 | 42.5 | 5.3 |
In Table 2, each synthesis is compared using different metrics commonly used in green chemistry such as the atom economy (AE), the overall yield, the atom efficacy and the E factor of the synthesis process. Each method is also compared to itself depending on whether or not the aldehyde co-products and the solvents are recycled. Not surprisingly, the metric scores evolve positively when solvents are recycled, which confirms the interest of such action. Despite the concerns of dangerousness of the process, the synthesis method using ozone (route a) is the one that obtains the best scores. Synthesis in a few steps (2), its moderate atom economy (54%) is mainly due to the formation of heptanal and other aldehydes during oxidative cleavage. If recovered at the end of the reaction, considered as co-products and no longer a waste, the AE value of this route would then increase to 89% (Table 2, entry 1).
Entry | Route | AE (%) | Overall yield (%) | Atom efficacy (%) | E factor | ||||
---|---|---|---|---|---|---|---|---|---|
w/o recov. aldehydes | With recov. aldehydes | w/o recov. ald. | With recov. ald. | Standard | With recov. solvents | With recov. solvents & aldehydes | |||
a Synthesis with NaIO4. b Synthesis with KIO4. | |||||||||
1 | a | 54 | 89 | 95 | 51 | 85 | 40 | 6 | 3 |
2 | b1a | 26 | 43 | 42 | 11 | 18 | 447 | 13 | 8 |
3 | b2b | 17 | 28 | 34 | 6 | 10 | 721 | 28 | 19 |
4 | c | 49 | 82 | 3 | 1.5 | 2.5 | 8285 | 70 | 46 |
It should also be noted that all other metrics are also positively impacted by the recovery of aldehyde co-products, mainly heptanal (Table 2, entries 1–4). Indeed, heptanal is a molecule of interest that is used in the perfume industry either as such or as a starting material for the preparation of jasminaldehyde.29–31 Moreover, it can also be used as an alkylating agent for the aldolization32,33 or reductive etherification of polyols34–37 and sugars38–41 to give biobased surfactants. Based on the results presented in Table 2, the method using NaIO4 (route b1) appears to be the best compromise between performances (yield = 42%; AE of 43%; E factor of 8) and safety (SC of NaIO4 = 15), provided the solvents are recycled. The thermal cleavage of β-hydroxy hydroperoxides (route c) is currently the least economically interesting because of its low overall yield (3%). Moreover, the requirement of two purifications by column chromatography on the laboratory scale negatively affects the E factor of the route. However, looking at the atom economy of the route, it is the only one that would be able to compete with reductive ozonolysis (82% vs. 89%). In addition, this route only requires H2O2 as an oxidant, which has be determined as the safer oxidant (SC of 9.5) and the whole process would be amenable to continuous flow synthesis. To conclude, the route c could be envisioned as the most promising solution but the technology is not mature enough. Therefore, future optimization would be necessary.
The change of state (liquid to solid) and of colour (colourless to black) showed that a reaction occurs. The same phenomenon happened, but faster, using CAl mixed with para-toluene sulfonic acid (p-TSA, 0.05 equiv., 5 mol%), a catalyst already used for the synthesis of novolac resins.12 The two reactions were again performed and monitored by differential scanning calorimetry (DSC) from room temperature to 150 °C, at 20 °C min−1. As seen in Fig. 6, for CAl alone, an endothermic peak appears from 100 °C to 140 °C with a peak apex at 129 °C. A similar behavior occurs at lower temperature (from 85 °C to 110 °C, with a peak apex at 110 °C) for CAl mixed with PTSA. This endothermic peak could be explained by the formation and the evaporation of water molecules produced during the polycondensation, confirming that, with or without catalyst, CAl has the ability to homopolycondensate.
![]() | ||
Fig. 6 DSC curves from ambient temperature to 150 °C at 20 °C min−1, monitoring the homopolycondensation of [1] CAl alone, [2] CAl mixed with PTSA (5 mol%). |
![]() | ||
Fig. 7 DSC curves of HCAlR for the determination of the glass transition temperature from −100 °C to 50 °C (20 °C min−1). |
Phenolic networks are well-known for their good thermal stability often up to 400 °C for 10% weigh loss.6 Even if the dangling chain of CAl are shorter than cardanol, as well as the induced plasticizing effect, it could impact the thermal behaviour of the phenolic material. Thermal properties of HCAlR were determined by thermogravimetric analysis (TGA) under air from 25 °C to 500 °C at 10 °C min−1 as illustrated in Fig. 8.
The TGA and DTG curves show a first low degradation, 10% of weight lost from 100 °C to 305 °C. This degradation could be due to the unfinished polycondensation of CAl and its water evaporation. A second and major degradation step occurs around 400 °C due to the actual degradation of RCAl resin which is a close result to the thermal degradation of cardanol-nonanal-based resin previously reported.12
The use of CAl as a single-component building-block for the synthesis of phenol and formaldehyde-free phenolic thermosets has been demonstrated and followed by differential scanning calorimetry. It led to a promising thermally stable (until 400 °C) and soft material (Tg of −22 °C). A deeper investigation of the formulation of CAl-based novolac resins and their resulting properties is ongoing and will be reported in a near future.
Footnote |
† Electronic supplementary information (ESI) available: Procedures, detailed calculations of green metrics and safety score, 1H and 13C NMR spectra. See DOI: https://doi.org/10.1039/d3su00058c |
This journal is © The Royal Society of Chemistry 2023 |