A Review on the Potential and Limitations of Recyclable Thermosets for Structural Applications

Abstract

The outstanding performance of conventional thermosets arising from their covalently cross-linked networks directly results in a limited recyclability. The available commercial or close-to-commercial techniques facing this challenge can be divided into mechanical, thermal, and chemical processing. However, these methods typically require a high energy input and do not take the recycling of the thermoset matrix itself into account. Rather, they focus on retrieving the more valuable fibers, fillers, or substrates. To increase the circularity of thermoset products, many academic studies report potential solutions which require a reduced energy input by using degradable linkages or dynamic covalent bonds. However, the majority of these studies have limited potential for industrial implementation. This review aims to bridge the gap between the industrial and academic developments by focusing on those which are most relevant from a technological, sustainable and economic point of view. An overview is given of currently used approaches for the recycling of thermoset materials, the development of novel inherently recyclable thermosets and examples of possible applications that could reach the market in the near future.

Keywords:

• Thermoset
• recycling
• polymer matrix
• dynamic covalent cross-links
• circular economy

1. Introduction

The increasing amount of fossil-based plastic waste ending up in our environment is one of the most pressing issues of this decade. The installment of a circular economy, as described by the Ellen MacArthur Foundation,¹ could contribute to a more sustainable future for generations to come. Two key aspects of this concept are that the vast majority of plastic materials should be recyclable and that those materials entering the chain should be bio-based. Although the recycling of plastic materials is currently achieved only for a very small percentage of all plastics produced (global average in 2016 was 10%¹), the recycling and reprocessing of thermoplastic materials can generally be performed relatively easily from a practical and technological point of view.^2–5 Thermoset materials, on the other hand, are classified amongst the most difficult materials to recycle, and for a long time this was even considered impossible. The defining characteristic of a thermoset is the presence of covalent intermolecular chemical cross-links that create increased strength and stiffness, and reduce the susceptibility to creep as compared to their thermoplastic counterparts. As a result, they become less susceptible to damage by thermal and chemical impulses from the nearby environment which makes them highly suitable for use in structural and protective (composite) applications (e.g. aerospace materials and wind turbines). However, a direct consequence of this high thermal and mechanical stability is that thermosets are very difficult to recycle. Major thermoset resin classes are isocyanates, unsaturated polyesters, formaldehydes, epoxies, and alkyds.^6,7 These are usually applied as multicomponent reactive formulations to bind (functional) particles and reinforcing fibers to create, after curing, strong lightweight materials.

Currently, the majority of thermoset plastic waste is being managed by landfilling, which is, according to the Environmental Protection Agency’s (EPA) guidelines for solid waste management (1989), the least preferred waste management approach. Other industrial ways to handle thermoset waste are mainly limited to grinding and combustion.^8,9 Ground thermosets are re-used, typically in the form of fillers, in lower quality applications without separation of the specific constituents and without changing its molecular structure. Combustion mainly targets to regain the more valuable reinforcing fibers and to recover energy by burning off the thermoset matrix, which is considered to be the lowest class of (partial) recycling.¹⁰ However, in order to increase the composites’ contribution to the circular economy, it is critical that not only the fillers or fibers, but also the thermoset matrix itself is regained and re-used.

A more promising route to achieve recyclable thermosets is by the introduction of degradable or dynamic covalent bonds within their molecular structure. The dynamic response of these bonds is triggered by an external stimulus which can be of thermal, chemical or optical nature.^11,12 In the past decades, polymers containing such linkages have been reported under a variety of nomenclatures, such as recyclable, reworkable, remendable, and self-healing materials.^12–17 The majority of the materials reported in these studies are industrially irrelevant due to complex expensive chemistry or a lack of relevant mechanical properties. Nevertheless, this new class of thermosets is slowly moving out of the academic spectrum, and the first examples of recyclable thermosets are approaching the market.^18–20 In the last decade, extensive reviews were published that either focus on the industrial approaches for thermoset recycling^8,9,21 or on the wide range of available chemistries for dynamic covalently cross-linked polymers.^11–13 This review aims to connect these industrial and academic developments by focusing on those which are the most relevant from both an economic and sustainability point of view. In doing so, an overview is given of the currently applied solutions for the recycling of thermoset materials, the development of novel recyclable thermosets and examples of possible applications that could reach the market in the near future.

2. Recycling of conventional thermoset materials

Due to an increasing number of energy-resourceful applications that use lightweight thermoset materials and restrictions on landfill solutions throughout the EU, the development of thermoset recycling strategies has accelerated in recent years.^8,22,23 Another important driver is the value of the regained products. Regarding thermosets, most attention is given to the recycling of fiber-reinforced composites, since the fibers are generally more valuable than the matrix material, especially when carbon fibers are used. As indicated by the European Union’s Waste Framework directive, the most favorable method of recycling is the direct re-use of components in similar or lower performance applications without any form of reprocessing.²⁴ Therefore, a first consideration in the recycling of high-quality demanding thermoset products should always be to verify the possibility of direct re-use in applications where a lower quality is acceptable. Examples of these applications can be found in housing and sports equipment.^25,26 However, since the options for re-use are limited and waste material often has complex non-adaptable geometries, the direct re-use of thermoset products will most likely remain a specific niche market. Therefore direct re-use will not significantly contribute to solving the thermoset waste problem. Waste treatment technologies that are capable of processing thermosets via a more universal approach were therefore studied thoroughly in the past decades. The presently available techniques that are already commercial or are approaching a commercial status can be divided into mechanical, thermal and chemical processing.⁹ A schematic overview of conventional thermoset composite waste processing and recycling routes is shown in Figure 1.

Figure 1. Schematic overview of conventional thermoset composite waste processing and recycling routes.

Mechanical recycling most often involves a cutting and grinding step in order to transform scrap products of different shapes into uniformly sized flakes with a typical size of 50–100 mm using a speed cutting or crushing mill.¹⁰ This form of mechanical recycling is applied to both thermoset and thermoplastic materials. The fundamental difference is that thermoplastics can be thermally reprocessed from this point onwards, while thermoset flakes need more advanced grinding procedures. These procedures use hammer mills or high-speed mills to grind the scrap flakes into finer products of several millimeters down to sizes within the micrometer range. For fiber-reinforced thermosets, a large disadvantage of this approach is that the matrix and the fibers are not treated separately and as a result, any orientational advantage of the fiber material is reduced or lost. This consequently decreases the superior mechanical properties and value of the product. Therefore, most investigations on mechanical recycling have targeted the processing of glass fiber-reinforced polymer (GFRP) products instead of their more expensive carbon fiber-reinforced polymer (CFRP) counterparts. Furthermore, the higher strength at break and stiffness of the carbon fibers also requires higher forces and more resistant machinery. The small-sized (<50 µm diameter) powder obtained after mechanical grinding can generally only be used as a filler in new thermoset and thermoplastic materials.²⁷ Unfortunately, these powders cannot compete economically with conventional filler material such as calcium carbonates and silicates.²⁸ To have any added value as filler, the ground products should therefore give an additional form of reinforcement as was intended in the original application. An example was given by Palmer et al., who developed a method to maintain the fibrous structure of the recyclates after grinding, which were successfully used as reinforcing filler in sheet molding compounds.²⁹

Thermal recycling can be divided into aerobic and anaerobic combustion where the latter is often defined as pyrolysis. The presence of oxygen is directly connected to the recycling output. The most straightforward aerobic thermal treatment is incineration, solely leading to energy recovery. However, as no material is actually recovered this is not classified as recycling and therefore not addressed in this review. The fluidized-bed combustion process developed by Pickering et al. is an example of an aerobic composite recycling process for GFRPs that is capable of fiber recovery while the thermoset matrix is broken down in CO₂ and water vapor which is converted into energy.³⁰ A second promising example of aerobic thermal recycling is the use of scrap GFR thermosets as feedstock for cement processing in cement kilns. Again, the matrix material is converted into gas and energy, but the glass, mainly consisting of borosilicate and calcium carbonate, is used as resource material for the cement clinker which is later ground to cement. The European Composites Industry Association recommends GFR thermosets to be recycled via this approach as it is currently the only technically and economically feasible alternative to landfill.²⁸

Compared to aerobic combustion, pyrolysis will break down the thermoset matrix into lower molecular weight organic compounds of which the chemical nature is correlated to the original material.³¹ The applied temperatures lie within the range of 300–800 °C and the resulting products are gaseous, liquid and solid carbon-based compounds that, contrary to those of aerobic combustion, can potentially be used as feedstock for further chemical processing.^8,32 Another advantage is that, under anaerobic conditions, carbon fibers in CFRPs are not affected by oxidation which improves the recycling quality of these products. Compared to processing into cement kilns, pyrolysis gives rise to the option of re-using the thermoset product components more than once. With the assistance of specific catalysts, the thermoset pyrolysis recycling output can be enhanced. This is exemplified by the work of Ciprioti et al. who reported on improving the recycling yield of (thermoplastic) waste electrical and electronic equipment (WEEE) products by using fly ash (FA) as catalyst.^33,34 As FA is a waste product originating from coal plants it actually adds to the circularity of the process and does not heavily impact the costs. They also showed that the thermoset recycling yield can be enhanced by modification of the catalyst prior to pyrolysis. Their concept is further illustrated in Figure 2. These results add to the fact that, from a circular economy point of view, pyrolysis is considered to be the most optimal industrially available thermal recycling treatment at present. However, even though industrial pyrolysis processes are currently being developed, their economic feasibility is still questionable.³⁵ Continued research efforts on aspects such as reduction of pollutant emission³⁶ and the required process energy³⁷ are crucial before pyrolysis can have a significant impact on the thermoset waste issue.

Figure 2. Overview of the FA-catalyzed pyrolysis process of WEEE plastics leading to exploitable fuels and feedstocks. The figure shows that the recycling output can be improved by enhancing the FA catalyst via acid (FAMA) and base (FAMB) treatments. Reprinted with permission from Benedetti et al.^[34] Copyright 2017 SpringerLink.

Solvolysis (often referred to as chemical recycling) is a recycling strategy in which a liquid medium is used to degrade and dissolve the polymer matrix so that the reinforcing fibers or particles can be regained. Similar to thermal recycling, the main driver for development of solvolysis for thermosets is the potential to regain carbon fibers from thermoset composites. Additionally, as with pyrolysis, the degraded and dissolved organic compounds originating from the matrix can be reclaimed from the solvent and be re-used as molecular building blocks. Based on the applied dissolution medium, solvolysis can be divided into hydrolysis,^38,39 glycolysis,^40–42 and acid digestion.^8,43,44 The selected solvent determines which type of molecular bonds can be broken. These solvolysis treatments were successfully reported for a wide range of polymer networks including those with amine-epoxy,^45,46 anhydride-epoxy,⁴⁷ polyester,^40,48,49 and polyurethane⁵⁰ linkages. A recent trend is the use of (near) supercritical fluids as recycling medium. These serve as good solvents for degradation chemistry due to their high mass transport coefficients, high diffusivities, and low viscosities. An additional advantage of using supercritical fluids over pure liquid solvents is that the solvent power, and thereby the reaction rates and selectivity, can be controlled by adjusting the applied pressure.^51–54 The main advantage of solvolysis over pyrolysis is that, at least conceptually, no thermal energy is required, which makes the process more energy-efficient. However, the majority of solvolysis methods investigated in the past decades are performed at temperatures above 200 °C and therefore they should be considered as thermo-chemical recycling.⁹ Another issue is that the frequent use of hazardous solvents often results in poor life cycle analysis scores. Hence, the environmental impact of chemical recycling remains questionable.⁵⁵ Nevertheless, the concept of chemical recycling remains promising, provided that future research efforts focus on methods that are energy-efficient and produce organic components that can easily be reclaimed from the selected solvent. This topic and the, mainly academic, progress made so far is further discussed in the next chapter.

3. Development of thermosets with inherent recyclability

The aforementioned routes to recycle conventional thermoset polymers (e.g. epoxy-, polyurethane-, and polyester-based composites) most often require extensive amounts of energy and are generally not regenerating the polymer itself as is shown in Figure 1. Therefore, there is a high demand for novel thermosets with polymer architectures that allow for low energy molecular debonding that enables easy matrix removal or recycling. This molecular debonding can be obtained by stimuli-triggered degradation or by introducing dynamic covalent bonds in the networks. These concepts can be subdivided into different mechanisms which are schematically depicted in Figure 3. Stimuli-triggered degradation is achieved by modifying the conventional cross-linked resin via inclusion of labile bonds. It can result in different degradation products depending on the amount and location of the labile linkages within the polymer structure (Figure 3): (a) total degradation of the polymer network into lower molecular weight organic substances (gases and char), b) degradation of the polymer backbone to yield monomers or oligomers, and (c) degradation of cross-linking bonds, resulting in a thermoplastic polymer.

Figure 3. Schematic overview of the existing concepts to develop recyclable thermoset polymers.

Even though stimuli-triggered degradation can lead to an improved thermoset recyclability, the original polymer architecture is irreversibly destroyed and as a consequence, the polymer recyclate needs to be resynthesized or be used in lower performing applications (i.e. down-cycling). This is illustrated in Figure 4(a), which shows the molecular mechanism of a degradable acetal linkage indicating the irreversible nature of this type of bond. This drawback can be overcome by the use of dynamic covalent networks which function via either associative or dissociative mechanisms. The mechanism type affects the physical properties of the material during the recycling process, but both mechanisms yield a cross-linked structure with properties similar to the original thermoset. This is illustrated in Figure 4(b,c) which show the molecular mechanisms of a associative (transesterification) and dissociative (Diels-Alder [DA] reaction) linkage, respectively. The concepts depicted in Figures 3 and 4 their practical implications and limitations are each further described in the upcoming sections.

Figure 4. Examples of molecular mechanisms describing the main concepts of recyclable thermosets; a degradable acetal linkage (a), an associative transesterification reaction, (b) and a dissociative Diels-Alder reaction (c).

3.1. Stimuli-triggered degradable thermosets

In essence, the vast majority of all industrially available routes that deal with the end of life of thermosets (e.g. grinding, pyrolysis, and solvolysis) are based on polymer degradation. However, these techniques focus on the degradation of polymers that were actually designed to remain stable under extreme conditions which leads to a very high energy consumption. Alternatively, thermoset polymers can be designed in such a way that low energy requiring external stimuli can be used to trigger degradation on demand, while maintaining their outstanding performance. This is achieved by the introduction of stimuli-triggered cleavable linkages in the polymer network. Examples of such linkages are esters, orthoesters, sulfur-, nitrogen-, and phosphorus-containing linkages, carbonate, acetal, hemiacetal/hemiketal ester, olefinic, tertiary ether and peroxide bonds, vicinal tricarbonyl, and (retro) DA addition bonds. The specific chemistry that is responsible for the degradative behavior of these bonds is not subject of this review but can be found in the extensive work by Ma and Webster.¹⁷

Based on the type, amount, and location of the stimuli-responsive molecular moieties within the polymer (main or side) chain, the extent of degradation will differ. This leads to mechanisms that break down the polymer matrix into either (i) non-usable components (e.g. char and gases) or (ii) deliver directly reusable monomer/oligomer/polymer systems on the other side. This section reviews different classes of degradable thermosets based on the type of external stimulus that is required to trigger the degradation and intends to critically assess the current methods to develop degradable thermosets with respect to how efficient they are from the energy consumption and materials recovery point of view.

3.1.1. Triggered degradation without matrix recycling

The cumulative goal of stimuli-triggered degradable thermoset matrices is to decrease the energy required to degrade the polymers. This can be achieved by including more sensitive thermally cleavable bonds that lower the degradation temperature within the thermoset or by introducing chemical bonds that are cleaved upon exposure to other, alternative stimuli (pH, UV, and catalysts). However, the majority of the studies reported do not describe the attempt (or possibility) to recover the matrix. Nevertheless, these systems contribute to an enhanced thermoset (product) recyclability and are therefore covered in this review.

3.1.1.1. Thermally cleavable linkages

Thermosets with a lowered degradation onset temperature within the range of 200 °C up to 300 °C are often called thermally reworkable.^14,56–58 Within this class of polymers, the main challenge is to have low degradation temperatures without affecting the material stability and lifetime during application.

An early example of degradation temperature reduction by introducing reworkable bonds in epoxy-based thermosets was given by Wang et al. who reported two series of reworkable cycloaliphatic diepoxides containing thermally cleavable carbamate⁵⁷ and carbonate⁵⁸ linkages. These materials undergo curing reactions with cyclic anhydride in a similar fashion as a commercial cycloaliphatic epoxide, except that the carbamate group within the diepoxides can act as the internal catalyst. Once cured, these thermosets decompose between 200–300 °C (as compared with 350 °C for the commercial cycloaliphatic epoxide). A second example is reported by Liu et al. who synthesized phosphate and sulfite-containing epoxides based on commercially available resins with tunable degradation temperatures within the reworking temperature range 180–300 °C by adjusting the ratio of the used monomers.^59–62 They demonstrated that the C–O bonds in O = P–O–C– and O = S–O–C are less stable than in the O = C–O–C– linkage, which suggests that the stability of C–O bond is strongly affected by the electron-withdrawing effect of the neighboring group. Another example of modification of commercially used thermoset monomers was reported by Zhang et al. who prepared, via a one-step reaction, a series of maleimides containing acetal ester linkages that can be used for electronic chip recycling. Their results showed that the degradation of the acetal ester linkage initiated from ∼225 °C which resulted in a significant decrease of adhesion strength.⁶³

Much lower degradation onset temperatures (T_deg) were obtained by incorporating thermally cleavable secondary and tertiary ester linkages in epoxies^14,64–77 and (meth)acrylates.^78–82 While primary ester linkages have a typical T_deg of 340 °C, Yang et al. managed to decrease this temperature to 120 °C by introducing the tertiary ester linkages into cycloaliphatic epoxies. The cured epoxies with tertiary esters retain the room temperature mechanical behavior of conventional epoxies comprising primary esters, while having lower mechanical properties at elevated temperatures, thereby offering the possibility of easier thermoset removal. Secondary ester linkages reduced the T_deg with only 20 °C,¹⁴ while hyperbranched aliphatic poly(ester-amides) showed a reduction of 100 °C.^75,77 When air is used as the heating medium instead of nitrogen, T_deg was further decreased by an additional 20 °C (tertiary ester linkages) and 85 °C (secondary esters), as is described by the work of Yang et al.¹⁴ and Chen et al.,⁶⁵ respectively. The latter study also showed that it is possible to increase the content of non-cleavable linkages (from 36 mol.% to 50 mol.%) without impacting the degradation temperature. In order to prevent the uncontrolled hydrolysis of these ester linkages, Li et al. introduced benzoic rings into diepoxies without compromising T_deg.⁶⁷ A bio-based approach to modify epoxy thermosets with polylactic acid was reported by Acebo et al. However, compared to earlier described concepts, the reported degradation temperature range (300–400 °C) is high, thereby potentially negating the initial gain in sustainability.⁸³

The energy required to achieve degradation of the cleavable linkages in degradable thermosets can be further decreased by using catalysts. In this respect, González et al. synthesized poly(ether esters) using a conventional epoxy resin (DGEBA) with 7,7‐dimethyl‐6,8‐dioxaspiro[3.5]nonane‐5,9‐dione catalyzed by waterproof environmentally friendly earth triflates. They found that an increase in Lewis acidity leads to a drop in degradation temperature down to a T_deg of 214 °C.^84–86 Significantly lower degradation temperatures of tertiary esters are obtained when using photo-generated acids acting as catalysts as is shown by the work of Okamura et al.^82,87,88 They developed photo-cross-linkable reworkable methacrylate and thiol-ene resins with an onset T_deg of 110 °C. The main drawback of these approaches is that the catalysts used in these syntheses are either not commercially available (9‐fluorenylideneimino-p‐toluenesulfonate)⁸⁷ or very expensive (di(tert-butylphenyl) iodonium trifluoromethanesulfonate).⁸⁸

3.1.1.2. Cleavability via alternative (non-thermal) triggers

Besides thermal activation, polymer degradation can be triggered by alternative stimuli such as acidic and basic environments. Many reworkable thermosets are shown to degrade via acidolysis, such as those containing acetal, ketal,^89,90 and Schiff base⁹¹ linkages which are reported to degrade in aqueous HCl in combination with various solvents. An interesting comparison between thermal and alkaline degradation can be found in lactone-modified epoxies based on bisphenol-A diglycidyl ether (DGEBA) with tertiary ester groups. These thermosets are shown to degrade thermally⁹² or by alkaline hydrolysis.^93,94 However, the alkaline hydrolysis proceeds relatively slow, measuring only around 5% weight loss after two months immersion.⁹⁴ It can be facilitated by hydrolyzing with reflux at elevated temperature (80 °C), where 25 wt.% becomes soluble after only 24 hr, as found by Giménez et al.⁹³

The low-temperature cleavage of C=C bonds via ozonolysis in anhydride containing copolymers was reported by Tsujii et al.^95,96 These polymers are thermally stable up to ∼300 °C but can be degraded at temperatures ranging from −73 to 30 °C via ozonolysis in organic solvents. They showed that the degradation efficiency is inversely proportional to the temperature and that degradation occurs within minutes at low temperatures. Cross-linked poly(vinyl alcohol) and poly(vinyl acetate) have also shown ozone-triggered degradability, as reported by Lou et al. Advantageously, in these polymers, ozone degradation was performed in water, thus avoiding organic solvents.⁹⁷ When the use of organic solvents and cryogenic temperatures is minimized, ozonolysis is considered an environmentally friendly and up-scalable alternative to heavy metal oxidation reactions, as it gives high recyclate yields and does not require additional separation steps to remove unreacted reagents.⁹⁸

Gallagher et al. reported the synthesis of two new dimethacrylate polymers that contain rigid core structures derived from sugars, glucose and mannose.⁹⁹ Thermally initiated free radical polymerization of these substrates produced highly cross-linked bio-based thermosets with mechanical properties comparable to those reported for commercially available poly(dimethacrylates). These thermosets completely degrade in aqueous alkaline conditions after 17 days, while remaining stable in aqueous neutral and acidic conditions. Another bio-based stimuli-triggered degradable thermoset was reported by Ma et al., who cross-linked multifunctional carboxylic acids found in fruit juices with epoxidized sucrose soyate in aqueous environment without the use of catalysts. These thermosets, with T_g values in the range 9–96 °C and potential applications in coatings, could be degraded and completely dissolved in aqueous alkaline solutions within 13 min.¹⁰⁰ Bio-based thermosets that start degrading upon entering a specific biological environment (e.g. marine, freshwater, soil) could contribute to the thermoset waste problem of certain typical applications such as fishing gear and agricultural tools. However, it is questionable whether it is desirable and achievable to combine the typical high-performance behavior of thermosets with biodegradation. Nevertheless, examples of biodegradable thermosets are reported in soft materials, such as hydrogels and elastomers. Their degradation is often triggered by acidic, lysosomal, and endosomal conditions and these concepts could serve as a starting point for the development of biodegradable thermosets for structural applications.^101–106

3.2.1. Triggered degradation with matrix recycling

From a circular point of view, it is highly favorable to engineer a degradable thermoset in such a way that it can be recovered in the form of the original monomers, chemical building blocks or thermoplastic polymers. In this respect, a promising study on a family of recyclable poly(hexahydrotriazine) (PHT) thermosets which can be degraded by strong acid digestion to yield bisaniline monomers was reported by García et al.¹⁰⁷ More recently, a new PHT thermoset was prepared via the reaction of 2,2-bis[4-(4-aminophenoxy)phenyl]propane (BBAP) and formaldehyde by Yuan et al., which can be degraded via gentle depolymerization in special diluted acid solution (involving HCl and THF) within 18 hr. The breakdown process demonstrates two stages, dissolution and precipitation, which correspond to depolymerization into soluble oligomers and further generation of monomers,¹⁰⁸ as demonstrated in Figure 5. The acid-catalyzed reversible cross-linking of bifunctional spiro orthoesters synthesized from terephthalic acid, succinic acid, and 1,4‐cyclohexanedicarboxylic acid were reported by Yoshida et al.¹⁰⁹ They found that the resulting thermoset could be depolymerized to regenerate the original monomers. Similar structures containing bicyclo orthoester moieties were explored by Hitomi et al.¹¹⁰ By grafting the functional groups on the polymer side chains, they were able to de-cross-link the thermosets leading to regeneration of the starting linear polymers.

Figure 5. Illustrative example and kinetics of the triggered thermoset degradation in a 1 M HCL/THF solution with recovery of the BAPP-PHT matrix. (a) The resin’s degradation status at different times; (b) the resin’s first-stage depolymerization time (t1) and the contact angles of their corresponding H₂O/THF solutions; (c) the depolymerization kinetics of the BAPP-PHT and a PHT-based CFRP, (d) ¹H NMR spectra of the depolymerized products of the BAPP-PHT at different times; (e) the regeneration curve of BAPP. Reprinted with permission from Yuan et al.^[108] Copyright 2017 Nature Publishing Group.

A clear disadvantage of the aforementioned degradable thermosets with matrix recovery is that harmful solvents are used as degradation triggers. A more sustainable alternative is reported by Ogden and Guan who developed a vitrimer-like thermoset based on boroxine networks. In addition to unusual mechanical properties (strong and highly malleable), another important attribute was that, besides reprocessability, the polymer can be recycled completely back to its monomer upon disintegration and dissolution in boiling water. Subsequently, the monomer precipitated upon the cooling of the solution. A clear disadvantage of this polymer is that, due to its regeneration in aqueous environments, it shows a high sensitivity to moisture which largely affects the mechanical performance.¹¹¹

3.2. Recyclable thermosets based on dynamic covalent networks

Although the aforementioned routes to recycling via stimuli-triggered degradation can lead to increased thermoset circularity, the original polymer architecture is destroyed and as a consequence the polymer recyclate needs to be resynthesized or be used in lower performing applications (i.e. down-cycling). Alternatively, dynamic covalent chemistry enables the recycling of cross-linked polymers while maintaining the overall polymer structure. In addition, these polymers and their composites are often capable of damage healing or can be reprocessed prior to recycling. Comparable to the earlier described degradation mechanisms, dynamic covalent bonds are often (thermal, pH, light, moisture) stimuli-triggered. Reported covalent chemistries include carbon–carbon, carbon–nitrogen, carbon–oxygen, disulfide and boron–oxygen bonds.^11,12,112 Furthermore, the family of dynamic covalently bonded polymers can generally be divided into those with dissociative and those with associative exchange mechanisms, as schematically depicted in Figure 3.^113,114

The associative exchange mechanism involves the simultaneous exchange of one reaction partner for another, in which the original cross-link is only broken when a new covalent bond to another position is formed.¹¹⁴ On the other hand, dissociative covalently bonded linkages are de facto broken before they are reconnected to another reactive site. This implies that resins containing dissociative networks are capable of a dramatic on-demand viscosity drop, which gives them a distinct benefit over associative networks: matrix and fiber/filler separation and recycling. Nevertheless, both mechanisms have been studied extensively in recent years and thus advanced the field of self-healing, reprocessable and recyclable thermoset materials.

3.2.1. Thermally reversible covalent bonds

As was observed for thermosets with inherent on-demand degradability, the most common trigger for dynamic covalent bonds is heat, and for that reason matrices containing thermally responsive linkages are described separately from the other available mechanisms.

Associative cross-link chemistries are frequently used to create healable and reprocessable thermosets. These materials are often called vitrimers, a hybrid class of polymers which show both thermoplastic and thermoset like behavior based on the pioneering work of Leibler et al.,¹¹⁵ which is illustrated in Figure 6. A clear mini review about different chemistries for vitrimers was already provided by Du Prez and coworkers in 2016.¹¹⁴ In the present work, the aim is not to describe the mechanisms of different chemistries to obtain vitrimers, but to focus on the sustainability and technical feasibility of their usage in the industrial applications. The type of associative thermally reversible chemistries involve transesterification,^116,117 transamination^15,114,118, a selection of disulfide exchange mechanisms,^119–123 and alkene metathesis.^124,125

Figure 6. Illustrative example of the vitrimer concept and capabilities focusing on the flow and insolubility properties of an epoxy network with 5 mol% Zn(Ac)₂ catalyst. (A) Swelling during immersion in trichlorobenzene. (B) Normalized stress relaxation at different temperatures. (C) A cross-linked sample broken into pieces is reprocessed in an injection machine to recover its initial aspect and properties. (D) A fusilli-shaped elastomer made by local heating from a cross-linked ribbon of length 10 cm is reversibly deformed by a weight of 1.4 kg. Reprinted with permission from Montarnal et al.^[115] Copyright 2011 The American Association for the Advancement of Science.

When vitrimers are heated above a certain temperature (100 °C < T_v < 200 °C), the reversible bonds cause network reorganization resulting in macroscopic flow without risking structural damage. Often catalysts are used to regulate the activation energy and T_v. For thermosets that work via transesterification a potential major drawback is that large amounts of a specific type of catalyst are used, which makes it difficult to tailor the desired properties.^114,116,118 Furthermore, these catalysts often deactivate over time which restricts the material lifetime. However, as long as the catalyst is present and active, the reversible nature enables welding, molding, reshaping, and recycling of fully cured materials. A new catalyst-free system based on the transamination in vinylogous urethanes was proposed recently. Interestingly, these vinylogous urethanes are stable towards hydrolysis, and can even be formed quantitatively in water as a solvent.¹¹⁸ To increase the circularity of these thermosets bio-based monomers can be used. Partial¹²⁶ and full^{100,127–131} bio-based vitrimers are already reported in recent literature.

Associative networks do not depolymerize upon heating as they retain cross-link density and hence insolubility at all temperatures. Even though the viscosity reduces following the Arrhenius relationship upon temperature increase, it remains at a rather high value of typically 1010 Pa·s at 200 °C.¹¹⁵ This relatively high viscosity makes these chemistries very suitable for repairing the composite and reprocessing with methods that do not require a low matrix viscosity such as compression molding and (ultrasonic) welding. However, in the case of composite materials, a sole exposure to thermal stimuli is not sufficient for the separation of the fiber or filler material from the resin.

A seemingly more feasible alternative to obtain thermosets with an on-demand low viscosity is the use of dissociative reversible covalent thermosetting chemistries. Due to a net decrease in molecular linkages upon exposure to a thermal trigger, these materials can achieve very fast topology rearrangements (stress relaxation and flow). This temporary de-cross-linking typically results in a sudden viscosity drop, analogous to the melting of thermoplastics. Cross-links are re-formed upon cooling, which reestablishes the required thermosetting properties such as high moduli and insolubility. As a result, these dynamic cross-links could allow for thermal (re-)processing of polymeric networks as is typical for thermoplastic systems (e.g. via injection molding and extrusion).¹¹⁴

One of the most extensively explored dissociative reversible exchange mechanisms is the [4 + 2] cycloaddition of a diene and a dienophile, also known as the DA reaction. The DA-reaction takes place, depending on the type of the diene and dienophile, usually at temperatures below 50 °C.¹³² More interestingly, the DA-adduct is thermally unstable which results in its dissociation (i.e. retro-DA) back into diene and dienophile at elevated temperatures. This reversibility is very attractive for healable and/or recyclable thermosets.^133–137

The temperatures of the DA and retro-DA transitions largely depend on the choice of diene and dienophile. In literature, the furan-maleimide system is most often selected due to its suitable temperature profile for reversible bonding.^133,134,138 In these systems, the DA reaction takes place at temperatures below 40 °C while the retro DA occurs at temperatures above 140 °C. By adjusting the molecular weight and ratio of the individual constituents, the mechanical properties and the cross-linking density can be tuned, which in turn alters the viscosity of the retro-DA mixture. This is exemplified by the work of Billiet et al. who introduced DA-based thermoreversible 1,2,4-triazoline-3,5-dione groups into polyurethane and polymethacrylate materials, thereby creating a recyclable thermoset with a tunable dynamic behavior. The fast thermal recycling of their thermoset is illustrated in Figure 7.¹³⁹

Figure 7. A polyurethane thermoset with DA-based 1,2,4-triazoline-3,5-dione groups that allows for the recycling of broken pieces by molding at elevated temperature and pressure. Reprinted with permission from Billiet et al.^[139] Copyright 2014 Springer Nature.

Although DA-adducts are promising candidates for recyclable thermosets, two important characteristics must be accounted for; DA adduct formation is slow compared to conventional curing of thermosets and the reaction continues as long as there is some mobility in the polymer matrix. This results in an uncontrolled increase in cross-link density which changes the mechanical properties of the composite over time, usually manifested as polymer embrittlement.

3.2.2. Reversibility via alternative (non-thermal) triggers

Polyimine thermosets based on Schiff base reversible chemistry with heat or water-driven malleability were developed from a combination of diamine, dialdehyde, and triamine cross-linkers through imine condensation in the absence of any catalyst.^140,141 It was demonstrated that the reversibility of imine bond formation and temperature‐dependent rate of the imine exchange render such malleable cross-linked polyimines.

The dynamic nature of sulfur–sulfur linkages and of disulfides, in particular, has been extensively studied because of the great interest for vulcanized rubbers in the chemical industry.¹¹⁴ Recent reports show that the behavior of covalently exchanging disulfide bonds is a rather complex process and involves several mechanisms (both associative and dissociative) that also depend on the used conditions and substitution patterns of the disulfides. In the simplest form, disulfides can be reduced to two thiols and then oxidized again in rather mild conditions (involving low concentration of solvents such as HCl and iodine solution in diglyme).^142–144 Alternatively, dissociative disulfides can be reversible under UV,¹²² shear, and/or heat triggers.^145,146

Alternatively to the thermally induced [4 + 2] cycloaddition in DA reactions, photo-induced [2 + 2] and [4 + 4] cycloadditions are used for the synthesis of linear polymers¹⁴⁷ and reversible cross-linked networks.¹⁴⁸ The main advantage of photo-induced cycloaddition, compared to DA, can be found in the high rate of dimerization. For example, whereas DA adduct reformation typically spans over several hours, the cross-linking of cinnamoyl groups takes place within 10 min.¹⁴⁹ The photo-induced cycloadditions take place at other wavelengths than the reverse reaction, which enables tunable healing and recycling conditions.¹⁴⁹ Currently, the practical usage of photo-induced cycloaddition in composite applications remains unreported. Presumably, the main limitation for this application is the insufficient penetration depth of light into multicomponent applications.

4. Application perspective

Compared to thermoplastics, the market for thermoset resins is much more fragmented. The number of producers of thermoset resins is large and contains, besides a few multinational industries, many small- and medium-sized manufacturers. Hence, reliable data on market sizes of thermoset resins are difficult to find, and estimate vary considerably. Based on publicly accessible data, it is estimated that the global production volume of thermoset resins is currently somewhere between 40 and 50 million tonnes, which is about 10–15% of total ‘plastics’ production.⁶ The global market size of thermoset products is expected to grow by 40% in the upcoming decade.²⁸ About 4–5 million tonnes (10%) are used to make FRP composites.⁷ Glass fiber is used predominantly (>90%) due to cost-effectiveness. High-performance carbon fibers are also used, although the trend here is towards carbon-fiber/thermoplastic composites due to easier recycling and reclaiming of the carbon fiber.⁷ In general, due to the intrinsic higher recyclability, the use of thermoplastic polymers in applications that were traditionally reserved for thermosets is increasing. Even though this could be a feasible approach for the recycling of a multitude of thermoset applications, this approach will not solve the thermoset recycling issue as a whole. As a result, due to continuously growing demand of high-performance light-weight materials, the overall number of thermoset applications is expected to grow regardless.

The increasing usage of thermoset materials will logically result in an increased amount of waste and higher demand for recycling options. As addressed in the preceding sections, the currently used waste treatment techniques require high amounts of energy and rarely take recycling of the polymer matrix itself into account (Figure 1). In order to tackle this issue, many new academic concepts were developed in the past decades that focus on actual recycling of the thermoset. A closer look at these concepts reveals that there is no universal solution; every technology has its own unique selling points and disadvantages. In order to improve the recyclability of a specific product, it is crucial that individual application-based requirements are taken into account when selecting the most suitable recycling concept. This is further illustrated in Figure 8 which shows the circularity of the different concepts for thermosets with inherent recyclability (as shown in Figure 3) in combination with some application examples. Various concepts of both degradable and dynamic covalent systems are being translated to more application-oriented studies and some are already approaching the market. Examples of application-driven studies are found in typical thermoset fields such as structural composites, rubber, adhesive and electronic applications.^{11,13,150,151}

Figure 8. Schematic overview of the circularity of thermoset materials with inherent recyclability.

In line with the current state of the art in thermoset recycling, the initial driver for application-oriented research often remains the ability to reclaim the more valuable fibers and therefore most developments are targeted towards structural composite applications. An example of stimuli-triggered degradation that is already approaching the market is based on the work performed by Pastine and coworkers. They developed a series of acetal-containing amine hardeners for cross-linking of commercially available epoxy resins which can be impregnated in glass and carbon fiber fabrics at conventional composite processing conditions. The resulting composites show mechanical performances comparable to commercially available composites. In addition, the acetal linkages degrade upon immersion into an acidic solution and the remaining linear polymer backbone can be regained and used as a thermoplastic polymer in different applications. Based on this technology Connora^® was founded which focuses on the development of recyclable resins for sporting goods, automotive, and wind turbine applications.^{18,19,152–154} A schematic overview of their technological concept is shown in Figure 9(a).

Figure 9. Conceptual representations of two recently founded startup companies that focus on the development of recyclable thermoset for composite applications. Connora (a) uses degradable linkages to obtain thermoplastic recyclates. Mallinda (b) uses polyimine networks that can be separated by dissolution.^[18,20]

Even though the described degradable thermoset concepts show a potential for thermoset composite recycling that was not yet realized till this date, the resulting matrix recyclates are still degraded versions of the original polymer. To obtain a higher quality of the recyclates, the infusion of dynamic covalent networks into fibrous composites was studied by many researchers. For instance, Chabert et al. transferred the vitrimer concept into GFRP composites by resin transfer molding of an epoxy-anhydride containing resin at 50 °C.¹⁵⁵ The resulting composites showed industrially relevant mechanical properties (i.e. Young’s modulus >20 GPa and strength at break >500 MPa) and good healing characteristics. However, due to the associative nature of vitrimer linkages the matrix-fiber separation and thereby the overall recycling of such composites will remain a challenge. A similar statement can be made for the recently developed epoxy-based FRPs with incorporated disulfide linkages. These composites (Young’s modulus >10 GPa) could be prepared by conventional vacuum infusion processing and are capable of low-temperature (70 °C) repair of cracks and delamination¹⁵⁶ and reprocessing at temperatures above 200 °C.¹⁵⁷ However, fiber separation could only occur via solvolysis in thiol-containing solvent mixtures without any matrix recycling.¹⁵⁷ Taynton et al. reported another example of dynamic covalent linkages as a binder for CFRP composites by using a polyimine-based polymer. Via a wet lay-up approach, they prepared composites with near-commercial mechanical properties (Young’s moduli >10 GPa) that could be recycled by dissolution in an ethanol-based solution.¹⁵⁸ The development of this technology resulted in the founding of Mallinda, a recent startup company that focuses on the production of catalyst-free malleable CFRPs with short cycle times.²⁰ The circular philosophy of Mallinda is visualized in Figure 9(b). Unfortunately, a clear disadvantage of these polyimine-based thermosets is that their mechanical performance is highly affected by moisture. Consequently, it is expected that their application field will most likely be limited to niche applications in low humidity conditions.

Thermosets with dissociative covalent linkages, where bonds are actually broken while being stimuli-triggered, can theoretically be designed in such a way that the drop in viscosity is sufficient for easy matrix removal. This technology has already been patented yet no recyclable composites that utilize dissociative bonds to separate matrix from fibers have reached the market up till this date.¹⁵⁹ The reason for the absence of such a polymer is most likely the complexity involved in designing a system that combines sufficient mechanical integrity with the required drop in viscosity. With this respect, DA reversible linkages seem to be the most promising mechanism for recyclable dissociative thermosets as they can be molecularly built into conventional composite resins (e.g. epoxy, polyurethane) with relative ease.¹⁶⁰ Furthermore, it is possible to graft DA groups (e.g. furan and maleimides) onto the reinforcing fibers in order to create repairable polymer–fiber interfaces.^161–164 However, processing of the DA functional resins into fiber fabrics tends to be very difficult since they are highly viscous or even in the solid state. Nevertheless, several researchers managed to produce FRPs-containing reversible DA linkages via injection molding,^165,166 vacuum-assisted autoclave molding,¹⁶⁷ and compression molding.^133,138 The produced composites showed good healing and easy repair capabilities, but full separation and recyclability of fibers and matrix were not yet reported.

Compared to FRP composites, the academic interest in the development of recycling methods via dynamic covalent linkages of other applications is rather limited. For example, recyclable cross-linked rubber products (e.g. car tires and sealing rings) are less investigated. This class of thermosets has intrinsically different mechanical properties compared to typical composite resins, but due to their permanent covalent cross-links the challenges for recycling are comparable. However, the fact that applications containing cross-linked rubbers are generally not hindered by continuous fiber networks makes that recycling is potentially easier to achieve.¹⁶⁸ As a result, the recycling of thermoset rubber tires has been developed to a greater extent, but similar to FRPs, high-energy-intensive mechanical and thermal recycling are also indicated as the most economically feasible options for this class of thermosets.^169,170 The most straightforward approach to utilize dynamic covalent linkages in cross-linked rubbers seems to be via disulfide linkages which are already present in cross-linked (natural) rubbers due to sulfur vulcanization.^171,172 By tuning of the molecular architecture and addition of CuCl₂ catalyst, healing and easy repair of these rubbers are possible, but due to the associative nature of the disulfide linkages, the reprocessing and recycling is just as complex as it is for conventional rubbers.¹⁵¹ In that respect, dissociative DA linkages show a higher recycling and reprocessing potential. Therefore, they account for the most investigated type of reversible bonds in rubbers, and it is expected that these concepts will approach the market in the upcoming decade.^173–175

The development of recyclable thermosets for applications in electronics is a relatively new field of interest. In this case, contrary to structural composites and rubbers, functional properties are more important than mechanical performance. Hence, recycling technology should focus on these aspects. For example, thermal interface materials were developed that show easy repair capacity of the thermal conductivity using associative disulfide linkages.^176,177 Furthermore, flexible electronics that are capable of restoration of the electronic conductivity due to the presence of dynamic DA^178,179 and polyimine¹⁸⁰ linkages were reported. The repair in these concepts is often triggered by a thermal stimulus that is already applied due to intrinsic use and that the functional polymers are often shielded from environmental hazards such as moisture. As a result, less constraints need to be overcome and therefore the commercial development of these products is expected to arrive in the upcoming years.

5. Conclusions

Besides the issues and complications that exist for the recycling of thermoplastic materials, the recycling of thermosets suffers from additional technical complications due to the covalently cross-linked networks that are present in their molecular architecture. In other words, their intended high thermal stability and chemical resistance directly result in very limited recycling options. As a consequence, landfill is currently the most used end-of-life option for thermoset products. Nowadays, recycling methods such as reprocessing in cement kilns, pyrolysis, grinding, and solvolysis are growing towards an industrial level, but they typically require a high energy input and are de facto merely disposing instead of recycling the polymer. As recycling is crucial in realizing a circular economy, there is a high demand for reduced-energy solutions that allow for efficient thermoset recycling. One route to achieve this is by the introduction of degradable linkages or dynamic covalent bonds within the thermoset molecular structure. Unfortunately, regarding thermosets with degradable bonds, most research efforts end as soon as successful degradation is shown and no attention is given to the industrial exploitation of these concepts. Therefore, the development of low-energy methods for re-obtaining the polymer phase in the form of monomers, oligomers, or thermoplastic polymers should be a focal point in the upcoming years. Actual thermoset recycling can more easily be achieved when dynamic covalent bonds are implemented. From a recycling perspective, these systems should focus on the development of dissociative exchange mechanisms as they can allow a viscosity drop that is sufficient for (thermoplastic) reprocessing and separation of the matrix in polymer thermoset composite products. Degradable and dynamic covalent thermosets are gaining increasing attention in the application-oriented research and some concepts are slowly penetrating the market. In this respect, it has to be noted that there is no universal approach to energy-efficient thermoset recycling; for each application an individual assessment needs to be made as to which end-of-life option fits best with required product performance.

Additional information

Funding

This research project has been carried out by Wageningen Food & Biobased Research and funded by the Dutch Ministry of Agriculture Nature and Food Quality, in the context of “Total Use of Resources for Food and Materials.”