Study of the Diels–Alder and retro-Diels–Alder reaction between furan derivatives and maleimide for the creation of new materials

Abstract

The Diels–Alder reaction leads to a mixture of two diastereomers, one called endo and the other one exo. The cyclo-reversion temperature of the first one is lower than the exo adduct and the ratio between endo and exo adducts varies according to the substituents of the Diels–Alder partners and experimental parameters. Therefore, the influence of some reaction parameters such as the substituents of furan and maleimide derivatives, the reaction temperature and the presence of a nucleophile on the endo/exo Diels–Alder ratio and/or the retro-Diels–Alder reaction have been studied. For instance, furan and maleimide derivatives with electron withdrawing substituents induced the creation of the endo adduct preferentially. Also the presence of a far electron withdrawing substituent on furan and/or an electron attracting mesomeric substituent on maleimide resulted in a faster reversibility of the endo adduct. Finally, a high temperature and the presence of a nucleophile (thiol) also induced faster retro-Diels–Alder kinetics. Moreover, it was proved that isomerization from the endo to the exo diastereomer is preceded by a retro-Diels–Alder reaction of the endo adduct. The presence of a nucleophile in the mixture confirmed this result. This study allowed the highlighting of different parameters of the Diels–Alder reaction to obtain as much endo adduct as possible, and a fast and/or full retro-Diels–Alder reaction of this adduct.

---

1 Introduction

Self-healing materials^1–3 and materials which present post-crosslinking on demand represent a huge interest for scientists. In both cases, stimuli reversible reactions and controlled reactions (or bonds) are used. Temperature, radiation, O₂, UV, electron beam, X-ray, water, pH etc. are commonly used as external stimuli. For this study, the temperature, which is the easier way for industrialization, has been chosen. Systems such as the urethane group,⁴ nitroso group,⁵ ester group,⁶ ionene complex,^7–10 azlactone–phenol adduct^11,12 and, lastly, the Diels–Alder adduct^13–18 allow the formation of reversible covalent bonds under high temperature conditions. The Diels–Alder system was chosen for this project, in spite of existing studies,^14,19–21 because this is the most famous and the easier system. The Diels–Alder reaction already proved its significance for the formation of reversible covalent bonds^22,23 and for the synthesis of new polymers/materials which have intermittent lives. Among them, thermo-sensitive surfactants,²⁴ self-healing materials,²⁵ high and mild temperature reversible materials²⁶ and materials which have a controlled crosslinking reaction²⁷ can be found. This study on the Diels–Alder reaction will help to improve the control of the crosslinking reaction during material formation.

Indeed, the Diels–Alder cycloaddition reaction would benefit from being upgraded on blocking and unblocking temperature, particularly by studying the influence of reactant substituents on the Diels–Alder and retro-Diels–Alder reactions, to upgrade the adduct stereochemistry and the kinetics of the reaction, respectively. The selectivity of the Diels–Alder reaction between maleimide and maleic anhydride²⁸ has already been studied. Also, some study has been done on the influence of both furan and maleimide derivatives. However, those results were used to study the endergonic and exergonic behavior of the Diels–Alder reaction.²⁹ Moreover, those results were obtained by computerized simulation, and only few examples have been experimentally studied. It is also important to note that the reaction between furan and maleimide is often a concerted cycloaddition, but due to the nucleophilic character of furan derivatives, the reaction with p-electron deficient components, such as maleimide, may proceed via an asynchronous transition state and in extreme cases, a stepwise zwitterionic mechanism is also possible.^30,31

It is interesting to further investigate the structure of the reactants used for the Diels–Alder reaction and the characterisation of the cyclo-reversion reaction. The main goal of this publication is to establish the precise rules of furan and maleimide derivative substituents for Diels–Alder and retro-Diels–Alder reactions.

2 Results and discussion

It is well known that the Diels–Alder (DA) reaction leads to a mixture of two diastereomers, one called endo (the kinetic compound) and the other exo (the thermodynamic compound)^13,32 (Scheme 1). For simplicity the adduct formed after the DA reaction is named as follows: A for adduct, then endo or exo depending on the adduct obtained (if there is nothing after A, it means that it is a mixture), then F and M for the furan and the maleimide derivatives, respectively.

Scheme 1 Diastereomers obtained after the Diels–Alder reaction between furan (F) and maleimide (M).

During the reverse reaction, called retro-Diels–Alder (rDA), the endo compound is the first to be unblocked,³³ i.e. at lower temperature than the exo compound. As a consequence, the parameters affecting the diastereoselectivity (structure, experimental conditions, solvent, temperature, etc.) have to be controlled to enhance the obtainment of the endo adduct, thus giving the possibility of breaking it at average temperature and giving the maleimide double bond its reactivity back. Some studies deal with the Diels–Alder reaction between furan and maleimide derivatives but there is a lack of precision about the control of all these parameters. First, a complete modeling study has been carried out to obtain all the information required to control them. For that, the maleimide/furan derivatives system^1,13,14,34 was studied thanks to existing compounds (commercial, c) or synthesized compounds (s) (Table 1).

Table 1 Structures of different commercial (c) or synthesized (s) furan (F) and maleimide (M) derivatives

R2	R3	Blocking agent	Mass yield (%)	R1	Maleimide	Mass yield (%)
	H	F1 (c)	—		M1 (c)	—
	H	F2 (s)	95		M2 (s)	57
	H	F3 (s)	87		M3 (s)	56
	H	F4 (s)	99		M4 (s)	62
	H	F5 (s)	86		M5 (s)	64
	H	F6 (c)	—		M6 (s)	74
	H	F7 (c)	—		M7 (s)	72
	H	F8 (s)	95
		F9 (s)	88

---

All reactants were chosen to have the widest diversity of partners for the DA reaction, with various characteristics (electronic and hindering effect). It is established that the DA reaction is a spontaneous reaction between a diene, i.e. furan derivative, and a dienophile, i.e. maleimide derivative. Furan derivatives are described, in the literature, as very good dienes for the DA reaction. Indeed, the oxygen atom enables the relocation of diene double bonds and increases the reactivity of these double bonds. For maleimide derivatives, electron attracting mesomeric substituents (imides) on both sides of the double bond allow the increase of the reactivity towards dienes. In this way, furan derivatives and maleimides react quickly together at room temperature (20 °C). They are also known to react at a low retro-Diels–Alder temperature (∼110 °C).³⁵

It is worth noting that the presence of substituents on those partners can facilitate the reaction (attracting inductive substituent on the diene and electron attracting mesomeric substituent on the dienophile^36–41) or destabilize the final adduct (hindering substituent). Indeed, it is known that the addition of substituents on both partners can have a real influence on the ratio between the endo and the exo compounds, but also on the stability of the adduct obtained after the Diels–Alder reaction. That is why some R₁ substituents on maleimide, and R₂/R₃ on furan derivatives are added. These substituents could be electron donating, electron withdrawing, electron attracting and donating mesomeric or hindering substituents. Thus, Diels–Alder compounds can be formed with a range of pairs of F and M. The results are separated into two groups, the first one with various furan derivatives (Table 2) and the second one with various maleimide derivatives (Table 3).

Table 2 Resulting endo/exo adduct ratios of the Diels–Alder reactions between N-methylmaleimide (M₁) or N-dodecylmaleimide (M₂) and furan derivatives (F)

Adduct	F	M	R (%)	T (°C)	Solvent	endo^a (%)	exo^a (%)
a Determined by ^1H NMR.
AF1M1	F1	M1	94	23	DCM	77	23
AF2M1	F2	M1	92	23	DCM	71	29
AF3M1	F3	M1	96	23	DCM	73	27
AF4M1	F4	M1	90	23	DCM	70	30
AF5M1	F5	M1	95	23	DCM	71	28
AF6M1	F6	M1	95	23	DCM	61	39
AF7M1	F7	M1	98	23	DCM	87	13
AF9M1	F9	M1	99	23	DCM	73	27
AF1M2	F1	M2	85	23	THF	64	36
AF2M2	F2	M2	83	23	THF	62	38
AF3M2	F3	M2	87	23	THF	62	38
AF4M2	F4	M2	92	23	THF	63	37
AF5M2	F5	M2	88	23	THF	64	36
AF6M2	F6	M2	93	23	THF	60	40

---

Table 3 Resulting endo/exo adduct ratios (AF₁M_x) of the Diels–Alder reactions between furfuryl acetate (F₁) and maleimide derivatives (M)

Adduct	F	M	R (%)	T (°C)	Solvent	endo^a (%)	exo^a (%)
a Determined by ^1H NMR.
AF1M1	F1	M1	96	23	DCM	77	23
AendoF1M1			94	23	Ether	100	0
AF1M2	F1	M2	92	23	THF	64	36
AF1M3	F1	M3	90	23	DCM	65	35
AF1M4	F1	M4	89	23	DCM	55	45
AF1M5	F1	M5	91	23	DCM	67	33
AF1M6	F1	M6	93	23	DCM	76	24
AF1M7	F1	M7	88	23	DCM	75	25

---

After ¹H NMR analysis of all the Diels–Alder adducts, for example the AF₁M₁ adduct (Fig. 2), the endo/exo ratio was determined for each one. To calculate these ratios, the peaks corresponding to the endo or exo adduct were used. To determine them, the system was analyzed using the Karplus equation (eqn (1)) on vicinal protons H_c and H_d. The Karplus rule links the angle between the germinal protons and their coupling constant.⁴² Thus, the theoretical coupling constant of the H_c proton should be higher for the endo diastereomer. The comparison between theoretical (³J_HcHd = 5.5 Hz) and experimental (³J_HcHd = 5.8 Hz) coupling constant values for A_endoF₁M₁ enabled the attribution of the peaks for the endo and exo H_c protons (Fig. 1 and 2). The angle between the two germinal protons of the A_exoF₁M₁ adduct (83.8°), was too close to the Karplus limit to afford accurate results. Nevertheless, the value obtained, from the Karplus equation, for the exo adduct indicated that the coupling constant should be lower compared to the endo adduct and near to a 0 Hz coupling constant. Thus, the experimental value of 1.6 Hz confirmed the theoretical one (Fig. 1 and 2). Therefore, the integration of c and c1 peaks, allowed the attribution of all the other peaks by integration comparison (called x for the endo adduct and x1 for the exo adduct). Moreover, the endo/exo ratio is determined thanks to c and c1 integrations, which give the percentage of the exo or endo adduct.

Fig. 1 Molecular representation of A_endoF₁M₁, determination of the angle between H_c and H_d protons and experimental and theoretical coupling constant determination ³J_HcHd.

Fig. 2 ¹H NMR spectrum of the AF₁M₁ adduct.

The Diels–Alder reaction occurs efficiently between all furan and maleimide derivatives. Moreover, the ratio is always higher for the endo adduct which is the kinetic product, but when the temperature is increased the ratio of the exo adduct is increased at the same time. The ratio could be increased for the endo adduct if the substituent is electron withdrawing for the furan and/or the maleimide derivative. The increase of the ratio in favor of one of the diastereomers compared to the other is well known but the interest of this work is principally on the rDA reaction of each adduct and the effect of some parameters on it.

2.1. Characterization of the retro-Diels–Alder reaction

According to literature, the endo compound³³ is unblocked at a lower temperature than the exo compound. IR,^{43,44 1}H NMR^45,46 and DSC^33,47 analysis are generally used for the study of the maleimide unblocking (retro-Diels–Alder) reaction. The DSC was performed on the AF₁M₁ adduct and the spectrum (Fig. 3) shows an endothermic peak starting at 80 °C with a decrease until 110 °C (region I, rDA endo) corresponding to the endo unblocking reaction. Then another endothermic peak begins at 110 °C until a maximum point at 130 °C (region II, rDA exo) corresponding to the exo unblocking reaction.³³

Fig. 3 DSC analysis of the N-methylmaleimide (M₁)/furfuryl acetate (F₁) Diels–Alder adduct.

The DSC analysis was also performed on the adduct A_endoF₁M₁ and the analysis showed one endothermic peak with a maximum point at 120 °C corresponding to the unblocking reaction of the compound A_endoF₁M₁. Then an isothermal analysis, of the same compound, at 70 °C over 4 hours by DSC was performed and no endothermic peak was observed, proving that no unblocking reaction occurred.

After this isothermal experiment using DSC, a ¹H NMR study of the capsule content was carried out and the characteristic peaks of furan F₁ and maleimide M₁ appear in the spectra, implying that the retro-Diels–Alder reaction of A_endoF₁M₁ occurred (Fig. 4). Moreover, the peaks of the adduct A_exoF₁M₁ appear at 5.1 (c1), 4.35 (f1) and 3 ppm (e1 and c1), confirming isomerization of the endo adduct. Thus, the ¹H NMR spectrum shows that the starting material (A_endoF₁M₁) was unblocked and the free maleimide M₁ underwent a DA reaction with F₁ to give the A_exoF₁M₁ adduct.

Fig. 4 ¹H NMR spectra of the capsule content before and after isothermal DSC of the A_endoF₁M₁ adduct over 4 hours in DMSO at 70 °C.

In the literature, the Diels–Alder reaction temperature is generally determined by DSC analysis and the retro-Diels–Alder reaction temperature corresponds to the endothermic⁴⁷ peak maximum. However, the temperature of the retro-Diels–Alder reaction of the endo adduct is actually lower. Thanks to the last experiment, DSC analysis is not an appropriate analytical tool to precisely detect Diels–Alder or retro-Diels–Alder reaction temperatures. In this study ¹H NMR is used as an analytical tool to obtain precise results of the endo and exo compounds, of the Diels–Alder ratio and of retro-Diels–Alder reaction temperatures.

The results of a model study of the Diels–Alder and retro-Diels–Alder reactions for all the adducts are presented Table 4. As an example the reaction between N-dodecylmaleimide (M₂) and cyclohexancarboxylate furfuryl (F₅) was carried out using variable-temperature ¹H NMR (Fig. 5).

Table 4 Variation of the retro-Diels–Alder reaction temperature of the exo and endo adducts as a function of the blocking agents used with N-dodecylmaleimide (M₂)

Blocking agent	Retro-Diels–Alder temperature (°C)
	endo	exo
AF2M2	Between 55 and 65	Between 85 and 95
AF4M2
AF6M2
AF7M2
AF9M2
AF1M2	Between 65 and 75
AF3M2
AF5M2
AF8M2	No Diels–Alder before 120 °C

---

Fig. 5 Variable-temperature ¹H NMR spectra with overlaying of the Diels–Alder and retro-Diels–Alder reactions of the adducts of AF₅M₂.

Looking at the endo (f) and exo (f1) adduct peaks, or furan (f2) and maleimide (a1) reagent peaks, the stoichiometry variation of the endo–exo adducts and furan–maleimide reagents can be observed.

The temperature was plotted as a function of the peak intensity with the methyl(s) of M₂ as a reference (Fig. 6).

Fig. 6 Proton peak intensities of the adducts A_exoF₅M₂ ♦, A_endoF₅M₂ ■ and F₅ ▲ from ¹H NMR analysis versus temperature.

Fig. 6 shows three distinct regions: the first one (I), from 40 to 60 °C, where quantities of A_endoF₅M₂ and A_exoF₅M₂ increase and at the same time the furfuryl acetate quantity decreases, corresponding to the Diels–Alder reaction. The second region (II), from 60 to 85 °C, is where the A_endoF₅M₂ adduct quantity decreases and the A_exoF₅M₂ adduct amount increases, meaning that the retro-Diels–Alder reaction of the A_endoF₅M₂ adduct occurred from T > 60 °C, and at the same time the creation (DA) of A_exoF₅M₂ continues. In this region the quantity of F₅ slowly increases until 75 °C and then decreases until 85 °C, which is due to the retro-Diels–Alder reaction of A_endoF₅M₂ and the Diels–Alder reaction of the A_exoF₅M₂ adduct, respectively. This transformation of the endo to the exo adduct is not a simple isomerization. These variations of each adduct ratio and the change in the quantity of F₅ show that prior to transformation from the endo to exo form a retro-Diels–Alder reaction of the endo adduct occurs. In the third region (III), from 85 to 125 °C, the quantities of A_endoF₅M₂ and A_exoF₅M₂ decrease while the F₅ quantity increases. This result proves that the rDA reaction of A_exoF₅M₂ starts up to 85 °C whilst the retro-Diels–Alder reaction of A_endoF₅M₂ continues.

From the ¹H NMR, it is possible to determine, by 10 °C in our case, the rDA reaction temperature. This method was applied to determine the rDA reaction temperature of a few furan derivatives. Results from Table 4 show the influence of the addition of an R₂ substituent on the furan derivatives. If the substituent is electron withdrawing (AF₁M₂ to AF₇M₂, and AF₉M₂), the DA and rDA reactions are facilitated; on the contrary if the substituent has an electron attracting mesomeric effect (AF₈M₂) the DA reaction doesn’t occurr as expected. The mesomeric form of the compound F₈ blocks the double bonds of the furan ring which can not participate in the DA cyclization.

All these results show and confirm that the endo compound has to be predominantly obtained to reach the average unblocking temperatures.

Yet, these results do not allow the determination of the unblocking kinetics which are an important aspect for the creation of materials. That is why a complete study of the effect of some parameters on the retro-Diels–Alder reaction kinetics, i.e. temperature, the presence of a nucleophile and the partners’ backbone, has been carried out.

2.2. Influence of reaction parameters on the retro-Diels–Alder kinetics

The aim of this part is to show the link between the Diels–Alder partners’ backbones, the endo/exo ratio and the unblocking kinetics.

It is first important to increase the ratio in favor of the endo compound so as to obtain a system with an average unblocking temperature of 70 °C. According to the starting diene and dienophile used for the Diels–Alder reaction, the endo or exo adduct could be favored. Some studies have investigated the influence of solvent and pressure on Diels–Alder conversion and enantiomeric yield. However, these works⁴⁸ showed that the increase of endo content is partially improved and principally depends on the partner backbone.

The use of a catalyst could also improve the ratio in favor of the endo adduct. For instance, some metallic catalysts have been described in the literature^49–51 which allow this selectivity to be favored. However, Gandini and Belgacem⁵² showed that the addition of a Brønsted acid or a Lewis acid as a catalyst in the presence of furan derivatives with a CH₂ in the α position of the furan ring, induces the polymerization of it. Moreover, the addition of a catalyst affects the endo/exo ratio, but not the unblocking kinetics. In a nutshell, the use of a catalyst in our system is impossible.

It is well known that a decrease in the temperature during the protection (DA) reaction increases the formation of the endo adduct and on the contrary, an increase in temperature induces an increase of the exo adduct formation. Thus, the Diels–Alder reaction has to be carried out at low temperature. Nevertheless, it can be stated that the lower the temperature, the longer the blocking time and the longer the reaction time, the more the exo adduct is formed.³⁴ Therefore, it is significant to find the best compromise between temperature and blocking kinetics. In this project, room temperature (20 °C) has been chosen because it gives an endo/exo ratio largely in favor of the endo adduct and the protection kinetics (one week) are acceptable. Moreover, each diene/dienophile couple has its own rDA reaction temperature. It is therefore important to control the unblocking temperature so that it is not too high for the application. Indeed, the lower the temperature the better the material will be. Afterwards, the influence of the temperature on the unblocking reaction and the presence of a nucleophilic reagent have been studied even if in all chemical reactions, the higher the temperature, the faster the kinetics.

Thus, work under these conditions (20 °C, without catalyst, long time) to obtain predominantly the endo adduct was chosen.

2.2.1. Kinetics study of the retro-Diels–Alder reaction at different temperatures. The Diels–Alder reaction is a balance between a blocked and an unblocked form; the ratio is thus expected to reach a plateau during the retro-Diels–Alder reaction. This ratio will depend upon the temperature.

To study the influence of the temperature, the compound AF₁M₁ was used and analyzed by ¹H NMR at three different temperatures (Fig. 7).

Fig. 7 Evolution of the A_endoF₁M₁ adduct mole number during the retro-Diels–Alder reaction at different temperatures (60 °C ▲, 70 °C ♦ and 80 °C ■) obtained by ¹H NMR.

Fig. 7 shows that the higher the temperature the faster the mole number of A_endoF₁M₁ decreases, which is coherent according to kinetics laws. Moreover, this analysis shows that, after 290 min, the A_endoF₁M₁ adduct still remains and the unblocking reaction is, as predicted, not complete. Indeed, even if the unblocking is the main reaction at high temperature, a small amount of both endo and exo adduct are, nevertheless, formed (Fig. 10). The same results are observed for the A_endoF₁M₂ adduct.

The equilibrium constant K was calculated at each temperature (Table 5) and plotting ln(n) versus t allows the obtainment of straight lines with a slope of −αk (Fig. 8), where α = 1 in our case (equivalent number). The reaction has, thus, an order of one. The retro-Diels–Alder reaction rate constant could be determined for each couple enabling a comparison between them (Table 6).

Table 5 Equilibrium constants for A_endoF₁M₁ and A_endoF₁M₂ at each temperature

T (°C)	60 °C	70 °C	80 °C
KAendoF1M1	1.17 × 10^−5	1.17 × 10^−4	3.18 × 10^−3
KAendoF1M2	2.37 × 10^−4	3.56 × 10^−3	3.58 × 10^−3

---

Fig. 8 Rate constant determination of the A_endoF₁M₁ adduct at different temperatures.

Table 6 Summary of all reaction rate constants obtained for A_endoF₁M₁

T (°C)	60	70	80
k × 10^3 (min^−1)	2.8	17.1	71.1

---

The unblocking reactions were carried out at different temperatures. It is therefore possible to plot ln(k) versus (1/T), from which the slope is equal to −E_a/R. The activation energy of the unblocking reaction of A_endoF₁M₁ obtained is 156.7 kJ mol⁻¹.

Table 6 shows that the higher the temperature, the higher the rate constant. An increase of 10 °C implies an approximately five times increase of the rate constant, which is significant.

With these rate values, the activation parameters were determined from the Eyring equation (eqn (2)).

Eyring equation obtained from the plot of f(1/T) = log(k/T).

The activation enthalpy (ΔH^‡) was estimated from the plot of the logarithm of the quotient of the rate k and temperature versus the reciprocal of temperature (1/T), while the entropy of activation (ΔS^‡) was determined from the intercept with the y axis. The activation enthalpy found is 16.21 kcal mol⁻¹ and the activation entropy is −8.6 cal mol⁻¹ K⁻¹. This last value is far from the typical value for concerted [2 + 4] cycloaddition reactions. For instance, Jasiński et al.⁵³ found that the reaction between cyclopentadiene and E-2-phenylnitroethene was −28.6 cal mol⁻¹ K⁻¹ activation entropy and 17.3 kcal mol⁻¹ activation enthalpy. They proved that it was a concerted mechanism. In this case, the activation enthalpy and entropy values are close to the values obtained for the cycloaddition of 1,1-dimethoxy-1,3-butadiene and tetracyanoethylene, which is known to occur via a zwitterionic mechanism (ΔH^‡ = 10.8 kcal mol⁻¹, ΔS^‡ = −6.2 cal mol⁻¹ K⁻¹).⁵⁴ For the retro-Diels–Alder reaction of A_endoF₁M₁, it seems that a zwitterionic mechanism actually happened in the rate-determining step. Then, after the unblocking reaction was studied, the elimination of one reagent after the rDA reaction was investigated so as to verify if the reaction between a nucleophile and maleimide (Michael addition) is possible in situ from the rDA reaction and if there is an impact on the kinetics.

2.2.2. Influence of the presence of a nucleophilic agent during rDA reaction. To do this study a thiolate was added in the mixture (a good nucleophile for the Michael reaction), which could react with the maleimide after the unblocking reaction⁵⁵ and avoid a re-blocking of the maleimide’s double bond. The Michael reaction shows faster kinetics than the DA reaction kinetics, the maleimide should thus react with the thiolate preferentially.⁵⁶ Fig. 9 shows that the mole number of A_endoF₁M₁ decreases faster in presence of thiophenol. As a consequence, the retro-Diels–Alder reaction kinetics are faster when one of the partners is consumed at the same time as the unblocking reaction.

Fig. 9 Mole number of A_endoF₁M₁ as a function of time, obtained by ¹H NMR, during the retro-Diels–Alder reaction at 70 °C (left) and 80 °C (right) with ♦ or without ■ thiophenol.

Moreover, during the unblocking reaction, the nucleophile enables consumption of the unblocked maleimide and thus changes the balance of the unblocking reaction. Indeed, the retro-Diels–Alder reaction is complete whereas, without a nucleophile, there is the balanced reaction (Fig. 10). This result confirms the assumption that the formation of the exo adduct from the endo adduct is necessarily preceded by a retro-Diels–Alder of the latter. Indeed with thiophenol, no formation of the exo adduct occurs, the maleimide double bond is released and reacts, in situ, with the thiol avoiding the creation of the exo adduct. Moreover it was proved that the reaction of thiolate with the maleimide in situ of the unblocking reaction is possible.

Fig. 10 Mole number of A_exoF₁M₁ as a function of time, obtained by ¹H NMR, during the retro-Diels–Alder reaction of A_endoF₁M₁ at different temperatures.

All these results have allowed rules to be set up for the improvement of the unblocking reaction kinetics: increasing the temperature and consuming one of the partners after the unblocking reaction. It has also been confirmed that a retro-Diels–Alder reaction occurs before the transformation from the endo to the exo adduct.

This work has been done at the same concentration and, as it is known, the kinetics depend on the concentration of all the starting materials. So the influence of the ratio and the concentration on the reaction kinetics has been investigated.

2.2.3. Influence of the endo/exo ratio and of the starting material concentration. The influence of the ratio between the endo and exo adducts on the unblocking reaction of A_endoF₁M₁ was studied by ¹H NMR with both a constant concentration of starting materials (A_endoF₁M₁/A_exoF₁M₁) and a varying ratio (Fig. 11). The variation of the peak intensity of the H_f protons of the A_endoF₁M₁ adduct at 70 °C was plotted against time.

Fig. 11 Effect of the endo/exo ratio (100% ♦, 77% ▲, 61% ■) (left) and the concentration (30 mg ml⁻¹ ♦ and 50 mg ml⁻¹ ■) (right) on the rDA reaction kinetics at 70 °C.

To compare all three curves of Fig. 11 (left), as before, the rate constants (k) are used. The results show that whatever the ratio, the rate constants (k) are all equal to 0.017 min⁻¹. In conclusion, the ratio has no effect on the unblocking reaction kinetics. Also, with or without the exo adduct in the mixture, the retro-Diels–Alder reaction kinetics remain the same.

Afterwards, the effect of the variation of the A_endoF₁M₁ adduct concentration on the rate constant (k) was studied (Fig. 11 (right)) and shows that the kinetics are the same whatever the starting material concentration. The reaction kinetics are concentration dependent, and it seems less probable that the concentration does not have an influence on the unblocking kinetics. In this case, the unblocking reaction is not affected by the concentration because the reaction time is too short.

In this study, only the compound AF₁M₁ was used. However, the unblocking reaction should be dependent on the furan derivatives/maleimide couple used for the Diels–Alder reaction. Thus, in this last part, the influence of the substituents R₁ of the maleimide and R₂/R₃ of the furan derivatives on the Diels–Alder reaction was investigated.

2.2.4. Influence of the furan derivatives/maleimide couple. The reactants used are significant for the Diels–Alder reaction. Indeed, the addition of an electron donating mesomeric substituent on the diene and the addition of an electron attracting mesomeric substituent on the dienophile, considerably increase the blocking reaction kinetics and the creation of the endo adduct.

Several furan and maleimide derivatives were thus synthesized and used to determine the influence of the substituents added on the diastereoselectivity of the reaction and on the rDA reaction kinetics. This allowed the deduction of rules for the blocking reaction between furan and maleimide derivatives.

Yields and ratios of the Diels–Alder adducts were determined by ¹H NMR. All results are summarized in Table 2/Table 3 and show that the backbone of R₁, R₂ and R₃ has an influence on the ratio. Indeed, endo adducts are favored by the presence of electron donating substituents on maleimide (compound M₅) and by electron donating mesomeric substituents (played by the oxygen atom on the furan backbone) coupled with electron withdrawing substituents in the α position on the furan ring (compound F₇). Selected furan/maleimide couples enable favoring or disfavoring of one of the adducts and, as a consequence, the ability to choose the best one depending on its application. In our case, the F₇/M₅ couple is the best to obtain only the endo adduct and, thus, to obtain an average unblocking temperature.

On the one hand, to study the influence of the blocking agent (furan derivatives, F) on the unblocking reaction kinetics of the endo adduct, an ¹H NMR monitoring was performed on the N-methylmaleimide (M₁) blocked with all the furan derivatives. On the other hand, the influence of maleimide on the unblocking reaction kinetics of the endo adduct was studied by ¹H NMR monitoring on each maleimide blocked with furfuryl acetate (F₁) (Fig. 12).

Fig. 12 Influence of the N-methylmaleimide (M₁) blocking agent with all the furan derivatives (left) and of every maleimide blocked with furfuryl acetate (F₁) (right) on the unblocking reaction kinetics of the endo adduct at 70 °C.

Since the exo adduct has no influence on the unblocking reaction kinetics, it was not necessary to purify them. Adduct mole numbers are, thus, used. The rate constant for each of the blocking agents used on M₁ and each maleimide blocked with F₁ are summarized in Table 7 (M₂ has not been studied because the required amount of reactant for the kinetics was not fully soluble in 0.6 ml of DMSO D₆).

Table 7 Summary of the rate constants k for each blocking agent used on N-methylmaleimide (M₁) and each maleimide blocked with furfuryl acetate (F₁)

Blocking agent	F1	F2	F3	F4	F5	F6	F7	F9
k × 10^3 (min^−1)	16.3	17.5	22.3	12.8	21.9	74.1	11.1	30.5
Maleimide	M1	M3	M4	M5	M6	M7	—	—
k × 10^3 (min^−1)	16.3	61.4	129.1	50.0	29.7	43.0	—	—

---

Finally, the more electron withdrawing the substituents R₂ and R₃ of the blocking agent, the slower the retro-Diels–Alder reaction kinetics. Moreover the more electron attracting mesomeric R₁ on maleimide is, the faster the unblocking reaction kinetics. However, the more the ratio is in favor of the endo adduct, the slower the retro-Diels–Alder kinetics. It implies that the more the endo adduct is obtained, the higher its thermodynamic stability is which thus leads to slower kinetics. Nevertheless, there is no proportional behavior between the unblocking kinetics and the endo/exo ratio.

3 Experimental part

3.1. Materials and reagents

The furan derivative blocking agents were obtained from commercial sources or synthesized in the laboratory. Furfuryl acetate ≥98%, 2-(2-furyl)-1,3-dioxolane 98%, propionate furfuryl ≥98%, thiophenol 97%, maleic anhydride 99%, 1-methoxy-2-propylamine 95%, zinc bromide: ≥98%, hexamethyldisilazane >99%, anhydrous sodium sulfate, aniline ≥99.5%, p-anisidine ≥99%, 4-nitroaniline ≥99%, N-dodecylamine ≥98%, 2-methoxyethylamine ≥99%, alcohol furfuryl 98%, 2-furoyl chloride 95%, triethylamine ≥99%, cyclohexanecarbonyl chloride 98%, 3,5,5-trimethylhexanoyl chloride 98%, benzoyl chloride 99% and pivaloyl chloride 99%, 5-methylfuran 99%, and 5-methyl-2-furaldehyde 98% were purchased from Sigma-Aldrich. N-Methylmaleimide 97% and 5-hydroxymethyl-2-furaldehyde 98% were purchased from ABCR and used without purification.

3.2. Product characterization

All products were characterized by proton and carbon nuclear magnetic resonance (¹H and ¹³C NMR). Measurements were carried out using a Brucker Aspect 400 MHz spectrometer. The deuterium solvent used was dependent on the product and has been detailed in each case. Chemical shifts are given in ppm with an s for singlet, a d for doublet, a t for triplet and an m for multiplet.

Differential Scanning Calorimetry (DSC) analyses were carried out on a NETZSCH DSC200 calorimeter. Cell constant calibration was performed using indium, n-octadecane and n-octane standards. Nitrogen was used as the purge gas.

Informatics representation and calculation of A_endoF₁M₁ were done with MOPAC 2012 software (quantic semi-empirical calculation) and AM1 methods (geometry optimization with the convergence criteria “precise”).

3.3. Synthesis of maleimide

The method used to synthesize the maleimide model is presented in Scheme 2.⁵⁷

Scheme 2 Synthesis of maleimide model (M_x).

Maleic anhydride (1.1 eq./amine function) was reacted with amine in THF (4 ml g⁻¹) for 1 hour. Amine was added drop by drop. Then ZnBr₂ (1.4 eq./amic acid function) was added under nitrogen. After solubilization, the mixture was heated at 40 °C, and then HMDZ in THF (1.4 eq./amic acid function) was added drop by drop for 30 minutes. After total addition, the temperature was raised to 70 °C for 4 hours. At the end of the reaction the mixture was filtered off and the solvent was evaporated under low pressure. Several purification methods were carried out and have been specified for each product.

3.4. Synthesis of furan derivative

Three different methods of blocking agent synthesis were used: the first (method 1) using alcohol furfuryl (Scheme 3), the second (method 2) using furoyl chloride (Scheme 4) and the third (method 3)⁵⁸ using aldehydes (Scheme 5).

Scheme 3 Example of the synthesis of alcohol furfuryl derivatives (method 1).

Scheme 4 Synthesis of furoyl chloride derivative F₈ (method 2).

Scheme 5 Synthesis of blocking agent F₉ using an aldehyde (method 3).

All substituents R₂/R₃ have been chosen for their electronegativity and their hindering nature (Fig. 13).

Fig. 13 All furan derivative (F_y) syntheses.

For both method 1 and method 2: an acid chloride (1.3 eq./alcohol function) was solubilized in dry dichloromethane (1 ml per g of product). Then triethylamine (1.3 eq.) was added under nitrogen drop by drop to the mixture. Amine was added drop by drop. The alcohol (1 eq.) was added drop by drop, under nitrogen, at 0 °C over 15 minutes. The reaction mixture was then heated at 40 °C for 5 hours. Then the mixture was filtered off and the solvent was evaporated under low pressure. Several purification methods were carried out and have been specified for each product.

For method 3: an aldehyde (1 eq.) was solubilized in dry THF (3 ml per g of product) at 0 °C, under nitrogen. Then sodium borohydride (2 eq.) was slowly added over 10 minutes. After 2 hours the reaction mixture was neutralized with HCl (2 N). The product was extracted with ethyl acetate solvent. The organic layer was dried with anhydrous magnesium sulfate and the product was crystallized at 0 °C. A pale yellow powder was obtained.

3.5. Synthesis of Diels–Alder adduct

Maleimide M (1 eq.) was solubilized in solvent (DCM for M₁ or THF for M₂) (2 ml per g of product). Then the furan derivative F (1.1 eq.) was added. The reaction mixture was stirred for two weeks at 23 °C.

3.6. Characterization of Diels–Alder and retro-Diels–Alder reactions

3.6.1. By isothermal ¹H NMR. In an NMR tube, all the reactants for the Diels–Alder synthesis were mixed. The solvent used was DMSO D₆, because of its high boiling point at 190 °C, which enables the thermal reaction to take place directly in the NMR tube. All spectra were saved every 5 minutes with 32 acquisitions, in automatic mode, over 1 hour with temperature increment (40, 55, 65, 75, 85, 95, 105, 115 and 125 °C).

3.6.2. By DSC. In a DSC capsule (sealed when solvent was used), the Diels–Alder adduct, solubilized or not in DMSO, was weighed (between 5 and 10 mg). For the characterization, for the blocking or unblocking reaction, the analysis was a step up from 20 °C to 150 °C, 2 °C min⁻¹, then a step down to 20 °C, 2 °C min⁻¹. This cycle was repeated twice. For the characterization of the unblocking reaction in the solvent, the analysis was stepped up, 10 °C min⁻¹, to 70 °C, and then the capsule was isothermally heated at 70 °C for 4 hours and then a step down to 20 °C.

3.6.3. Characterisation of retro-Diels–Alder kinetics by isothermal ¹H NMR. In an NMR tube, the Diels–Alder adduct was solubilized in 0.6 ml of DMSO D₆. For the study in the presence of a nucleophile, a drop of thiophenol and DABCO were added. Then the tube was placed in the NMR spectrometer and was heated at constant temperature (60 °C, 70 °C, 80 °C) for 5 hours. A spectrum was saved every 5 minutes with 32 acquisitions, in automatic mode.

4 Conclusion

In conclusion, this study showed a way to increase the endo/exo ratio in favor of the endo adduct in furan derivatives/maleimide systems: the Diels–Alder reaction has to be carried out at low temperature; electron withdrawing R₂ and R₃ substituents, and electron donating R₁ substituents should be favored. Moreover, the unblocking reaction time can be adjusted by the substituents added to the Diels–Alder partners. The unblocking reaction is favored with electron withdrawing R₂ and electron attracting mesomeric R₁ substituents. If, in the mixture, there is also a nucleophile (thiol), the unblocking reaction is faster.

This study enabled the re-demonstration of general points about the Diels–Alder reaction already known, as well as presentation of the link between external factors and partner backbones with the endo/exo ratio and the retro-Diels–Alder reaction kinetics. Consequently, it is possible to create a system in which there is control of the reaction temperature, control of the unblocking reaction kinetics and a system which could react on demand with a nucleophile (release of the maleimide reactive double bond by the rDA reaction and reaction, in situ, with a nucleophile). Thanks to this study, a future publication will aim at preparing multifunctional products so as to create a new type of controlled cross-linked material containing Diels–Alder adducts and a nucleophile where it is possible to trigger the reaction by stimulation (submitted).

Acknowledgements

The authors thank the Hutchinson Company for its financial support and BOURGOGNE Cyril for the informatics calculations.

References

1. A. Gandini, Scielo, The application of the Diels–Alder reaction to polymer syntheses based on furan/maleimide reversible couplings, 2005, vol. 15, pp. 95–101.
2. Y. Zhang, A. A. Broekhuis and F. Picchioni, Macromolecules, 2009, 42, 1906–1912.
3. M. W. Keller, in Epoxy Polymers, Wiley-VCH Verlag GmbH & Co, KGaA, 2010, pp. 325–344.
4. D. A. Wicks and Z. W. Wicks Jr, Prog. Org. Coat., 1999, 36, 148–172.
5. J. F. Pazos, Polymers coupled by nitroso groups, US3872057A, 1975.
6. D. Montarnal, M. Capelot, F. Tournilhac and L. Leibler, Science, 2011, 18, 965–968.
7. J. Gordon, J. Org. Chem., 1965, 30, 4396.
8. E. C. F. Ko and K. T. Leffek, Studies on the Decomposition of Tetra-alkylammonium Salts in Solution. V. Rates and Activation Parameters of Salts Containing Substituted-benzyl or Other Active Groups, 1972, vol. 50, pp. 1297–1302.
9. K. T. Leffek and A. F. Matheson, Secondary Kinetic Isotope Effects in Bimolecular Nucleophilic Substitutions. VI. Effect of I± and I² Deuteration of Alkyl Halides in their Menschutkin Reactions with Pyridine in Nitrobenzene, 1972, vol. 50, pp. 986–991.
10. K. T. Leffek and A. F. Matheson, Secondary kinetic isotope effects in bimolecular nucleophilic substitutions. V. Role of the solvent in the reaction between methyl iodide and pyridine, 1972, vol. 50, pp. 982–985.
11. K. B. Wagener, L. P. Engle and M. H. Woodard, Macromolecules, 1991, 24, 1225–1230.
12. K. B. Wagener and L. P. Engle, Macromolecules, 1991, 24, 6809–6815.
13. O. Diels and K. Adler, Justus Liebigs Ann. Chem., 1928, 460, 98–122.
14. A. Gandini, Prog. Polym. Sci., 2013, 38, 1–29.
15. R. Gheneim, C. Perez-Berumen and A. Gandini, Macromolecules, 2002, 35, 7246–7253.
16. W. F. Bailey, N. M. Watcher-Jurcsak, M. R. Pineau, T. V. Ovaska, R. R. Warren and C. E. Lewis, J. Org. Chem., 1996, 61, 8216.
17. R. W. Roush and R. J. Sciotti, J. Am. Chem. Soc., 1998, 120, 7411.
18. O. Diels and K. Alder, Liebigs Ann. Chem., 1928, 460, 98.
19. A. Gandini and N. M. Belgacem, Polym. Int., 1998, 47, 267–276.
20. H. C. Kolb, M. G. Finn and K. B. Sharpless, Angew. Chem., Int. Ed., 2001, 11, 2004–2021.
21. C. Barner-Kowollik, F. E. Du Prez, P. Espeel, C. J. Hawker, T. Junkers, H. Schlaad and W. Van Camp, Angew. Chem., Int. Ed., 2011, 50, 60–62.
22. J. C. C. Atherton and S. Jones, Tetrahedron, 2003, 59, 9039–9057.
23. B. Gacal, H. Durmaz, M. A. Tasdelen, G. Hizal, U. Tunca, Y. Yagci and A. L. Demirel, Macromolecules, 2006, 39, 5330–5336.
24. J. R. McElhanon, B. A. Simmons, T. Zifer, G. M. Jamison, D. A. Loy, K. Rahimian, T. M. Long, D. R. Wheeler and C. L. Staiger, Thermally cleavable surfactants based on furan–maleimide Diels–Alder adducts, scheme for Gemini surfactant, and surfactant manufacture, US7022861B1, 2006.
25. H. Weizman, C. Nielsen, O. S. Weizman and S. Nemat-Nasser, J. Chem. Educ., 2011, 88, 1137–1140.
26. R. G. Gieling and A. J. H. Klunder, ARKIVOC, 2004, ii, 91–108.
27. E. E. Laborbe, B. Le Rosignol, V. Froidevaux, B. Boutevin and R. Auvergne, Cure-on-demand liquid sealant composition, polysulfide compositions and preparation thereof, and coating of substrates, US20130137817, 2013.
28. L. Rulíšek, P. Šebek, Z. Havlas, R. Hrabal, P. Čapek and A. Svatoš, J. Org. Chem., 2005, 70, 6295–6302.
29. R. C. Boutelle and B. H. Northrop, J. Org. Chem., 2011, 76, 7994–8002.
30. R. Jasiński, Comput. Theor. Chem., 2014, 1046, 93–98.
31. R. Jasiński, M. Kubik, A. Łapczuk-Krygier, A. Kącka, E. Dresler and A. Boguszewska-Czubara, React. Kinet., Mech. Catal., 2014, 113, 333–345.
32. M. W. Lee and W. C. Herndon, J. Org. Chem., 1978, 43, 518.
33. J. Canadell, H. Fischer, G. De With and R. A. T. M. van Benthem, J. Polym. Sci., Part A: Polym. Chem., 2010, 48, 3456–3467.
34. C. Goussé and A. Gandini, Polym. Int., 1999, 48, 723–731.
35. A. I. Konovalov and B. N. Solomonov, Dokl. Akad. Nauk SSSR, 1973, 211, 1115.
36. J. Sauer, Angew. Chem., Int. Ed. Engl., 1967, 6(1), 16–33.
37. Q. Tian, Y. C. Yuan, M. Z. Rong and M. Q. Zhang, J. Mater. Chem., 2009, 19, 1289–1296.
38. J. Sauer and R. Sustmann, Angew. Chem., Int. Ed. Engl., 1980, 19, 779.
39. K. N. Houk, J. Am. Chem. Soc., 1973, 4092.
40. D. Ginsburg, Tetrahedron, 1983, 2095.
41. I. Fleming, Frontier Orbitals and Organic Chemical Reactions, Wiley, 1976.
42. M. Karplus, J. Am. Chem. Soc., 1963, 85, 2870–2871.
43. S. Magana, A. Zerroukhi, C. Jegat and N. Mignard, React. Funct. Polym., 2010, 70, 442–448.
44. H. Laita, S. Boufi and A. Gandini, Eur. Polym. J., 1997, 33, 1203–1211.
45. C. Vilela, L. Cruciani, A. J. D. Silvestre and A. Gandini, Macromol. Rapid Commun., 2011, 32, 1319–1323.
46. J. R. McElhanon, T. Zifer, S. R. Kline, D. R. Wheeler, D. A. Loy, G. M. Jamison, T. M. Long, K. Rahimian and B. A. Simmons, Langmuir, 2005, 21, 3259–3266.
47. C. Jegat and N. Mignard, Polym. Bull., 2008, 60, 799–808.
48. F. Benito-Lopez, A. J. Kettelarij, R. J. M. Egberink, A. H. Velders, R. M. Tiggelaar, H. J. G. E. Gardeniers, D. N. Reinhoudt and W. Verboom, Chemistry Today, 2010, 28.
49. F. Fringuelli, L. Minuti, F. Pizzo and A. Taticchi, Acta Chem. Scand., 1993, 47, 255.
50. C.-C. Liao and J.-L. Zhu, J. Org. Chem., 2009, 74, 7873–7884.
51. E. C. Angell, F. Fringuelli, F. Pizzo, B. Porter, A. Taticchi and E. Wenkert, J. Org. Chem., 1985, 50, 4696–4698.
52. A. Gandini and M. N. Belgacem, Prog. Polym. Sci., 1997, 22, 1203–1379.
53. R. Jasiński, M. Kwiatkowska and A. Barański, J. Phys. Org. Chem., 2011, 24, 843–853.
54. R. Sustmann, S. Tappanchai and H. Bandmann, J. Am. Chem. Soc., 1996, 118, 12555–12561.
55. J. W. Chan, C. E. Hoyle, A. B. Lowe and M. Bowman, Macromolecules, 2010, 43, 6381–6388.
56. S. Billiet, W. Van Camp, X. K. D. Hillewaere, H. Rahier and F. E. Du Prez, Polymer, 2012, 53, 2320–2326.
57. T. Mizawa, K. Takenaka and T. Shiomi, J. Polym. Sci., Part A: Polym. Chem., 2000, 38, 237–246.
58. C. Zeng, H. Seino, J. Ren, K. Hatanaka and N. Yoshie, Macromolecules, 2013, 46, 1794–1802.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra01185j

---