Biobased and Aromatic Reversible Thermoset Networks fromCondensed Tannins via the Diels−Alder ReactionAntoine Duval,† Guillaume Couture,‡ Sylvain Caillol,‡ and Luc Averous*,†

†BioTeam/ICPEES-ECPM, UMR CNRS 7515, Universite de Strasbourg, 25 rue Becquerel, 67087 Strasbourg, Cedex 2, France‡Institut Charles Gerhardt, Montpellier, UMR 5253 - CNRS, Universite Montpellier, ENSCM, 8 rue de l’Ecole Normale, 34296Montpellier Cedex 5, France

*S Supporting Information

ABSTRACT: Thermo-reversible networks were obtained forthe first time from tannins, an aromatic biobased polyphenol,by reacting furan-bearing tannins and telechelic oligomers withmaleimide end groups, using the Diels−Alder (DA) reaction.Condensed tannins from mimosa (Acacia mearnsii) werefunctionalized with furfuryl glycidyl ether and thoroughlycharacterized by 1H, 31P NMR, and FTIR spectroscopy. Theaccessibility of the grafted furan groups was confirmed by amodel reaction with N-methylmaleimide. Different cross-linked networks were then obtained by DA reaction withthree PPO and PPO-b-PEO-b-PPO oligomers and evidencedby FTIR spectroscopy. The thermal properties of the obtainednetworks were evaluated with differential scanning calorimetry(DSC) and thermogravimetric analysis. Then, the reversibility of the cross-linking was shown by a quick return to the liquid stateupon heating at 120 °C. The retro Diels−Alder reaction was studied by size exclusion chromatography and DSC.

KEYWORDS: Tannins, Diels−Alder, Retro Diels−Alder, Cross-linking, Biobased

■ INTRODUCTION

With the increasing lack of certain fractions from fossilresources, the subsequent price volatility, and an increasingpressure from the civil society for a more sustainable industry,researchers, and among them chemists, are urged to findalternatives to fossil-based derivatives. Biobased aromaticderivatives are especially sought, since BTX (benzene, toluene,xylene) fractions from naphta are nowadays less and lessavailable due to the strong increase in global demand. However,aromatic monomers are largely used in polymer materials fortheir rigidity and chemical and thermal stability, as well as theirability to structure the matter by π-stacking, thus providing highperformances. Finally, biobased aromatic derivatives can alsobring some new chemical architectures.Among the aromatic biobased compounds, tannins appear as

interesting candidates in the field of materials chemistry. Afterlignins, they are the second most abundant natural source ofphenolic derivatives.1 They are found in the soft tissues (sheets,needles, or bark) of all vascular plants and exhibit variousstructures divided into three main families: hydrolyzable,complex, and condensed tannins.2 The latter represents 90%of the world production, the large majority being extractedfrom Acacia mearnsii (commonly known as mimosa or blackwattle).3 Condensed tannins are based on four types ofrepeating flavonoid units: profisetinidin, procyanidin (cat-echin), prorobinetinidin, and prodelphinidin, linked byC4−C6 or C4−C8 bonds.4,5 Thanks to their specific aromatic

structure, they could provide high thermal and chemicalstability, as well as toughness, while the numerous phenolicfunctions and nucleophilic sites may easily lead to variousfunctionalizations.1 For all these reasons, tannins or theirdepolymerized building units have been successfully used in thesynthesis of polyols,6−8 epoxy resins,9−13 acrylate resins,14,15

adhesives,1 or foams.16−20

An important concern in polymer science, especially withthermosets, is to obtain reversible cross-linking bonds. Indeed,chemically cross-linked networks cannot be recycled by heating.Thus, a reversible cross-linking that allows depolymerizing thenetwork is a desirable feature for an easy valorization at the endof life of the materials.21 It may also be a route toward stimuli-responsive, shape memory, or self-healing materials.22−25

Various parameters have been used to trigger the cross-linking/uncross-linking reactions such as temperature, radia-tions, pH, or processing.26 Among them, temperature is oftenseen as the easiest to apply, especially for industrialapplications.27 Several systems can be used to form reversiblecovalent bonds with temperature including urethane or esterbonds, nitroso or azlactone groups, ionene formation, andDiels−Alder reaction. The Diels−Alder (DA) [4 + 2]cycloaddition involves a diene and a dienophile reacting to

Received: October 28, 2016Revised: December 1, 2016Published: December 7, 2016

Research Article

pubs.acs.org/journal/ascecg

© 2016 American Chemical Society 1199 DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

Dow

nloa

ded

via

NA

GO

YA

UN

IV o

n Ju

ne 2

6, 2

018

at 2

2:07

:20

(UT

C).

Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

yield a substituted cyclohexene, which can be broken by furtherincreasing the temperature via the retro Diels−Alder reaction(rDA).28,29 The furan−maleimide system exhibits relatively lowreaction temperature for DA and rDA, around 60−70 and110−120 °C, respectively, making it compatible with thethermal stability of most polymeric materials.30

In this study, we first designed original furan-bearing tanninsvia the reaction of mimosa (Acacia mearnsii) tannins withfurfuryl glycidyl ether in various ratios. These new function-alized tannins were characterized by spectroscopic techniques(1H and 31P NMR and FTIR) and size exclusion chromatog-raphy (SEC). Then, the accessibility of the grafted furan groupswas evidenced by the study of the DA and rDA reactions withN-methylmaleimide. Original biobased networks were finallyprepared with three different bismaleimide cross-linkers basedon telechelic oligo(propylene oxide) and oligo(ethylene oxide).The efficient synthesis of the DA adducts and the reversibilityof the networks was finally evidenced by different techniquessuch as 1H NMR spectroscopy at high temperature, FTIR,SEC, and differential scanning calorimetry (DSC).

■ EXPERIMENTAL SECTIONMaterials. Condensed tannins from mimosa (MT, Acacia mearnsii,

Fintan OP) were kindly supplied by Silva Chimica (St. MicheleMondovi, Italy). Prior to analysis, they were dried in an oven at 70 °Cfor one night and stored in a desiccator. Pyridine (sequencing grade, ≥99.5%) was purchased from Fisher Scientific, acetic anhydride (ACSreagent, ≥ 97%) and furfuryl glycidyl ether (FGE, 97%) from AcrosOrganics, and methanol (laboratory reagent grade, ≥ 99.6%),dichloromethane (puriss., ≥ 99%), 2,3,4,5,6-pentafluorobenzaldehyde(98%), N-methyl maleimide (97%), 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane (95%), dimethyl sulfoxide (≥99.9%), DMSO-d6(99.9% D atoms), and CDCl3 (99.8% D atoms) from Sigma-Aldrich.The cross-linkers poly(propylene oxide) bismaleimide (DP = 3 or 6)and poly(propylene oxide)-b-poly(ethylene oxide)-b-poly(propyleneoxide) bismaleimide were purchased from Specific Polymers (Castries,France). All chemicals were used as received without furtherpurification.Syntheses. Reaction of Tannins with Furfuryl Glycidyl Ether

(FGE). Mimosa tannins (MT) were dissolved in water containingNaOH (1 mol equiv to total phenolic OH in tannins). After 30 min ofstirring, FGE (from 0.25 to 1.5 mol equiv to total phenolic OH intannins) was added dropwise, and the reaction mixture was stirredovernight at room temperature. The solution was then acidified toprecipitate the functionalized tannins, down to pH ranging from 5 to1. The precipitate was recovered by centrifugation, washed three timeswith 50 mL of acidified water (same pH as the precipitation pH), andfreeze-dried.Acetylation. Acetylation was performed in a pyridine/acetic

anhydride mixture (1:1 v/v), as previously reported on tanninsamples.6

DA Reaction with N-Methyl Maleimide. In a 10 mL round-bottomed flask, 0.1 g of furan-bearing MT was dissolved in 2 g ofDMSO. After adding 31.2 mg of N-methyl maleimide, the mixture washeated at 70 °C for 15 days under magnetic stirring. The crudeproduct was cooled, precipitated in 40 mL of toluene, and centrifugedat 3000 rpm for 3 min. After removing the liquid phase, the brownsolid was dried overnight in an oven at 40 °C under vacuum (10−2

mbar).DA and rDA with Bismaleimide Linkers. In a small flask, 0.1 g of

furan-containing mimosa tannins was dissolved in DMSO at 10 wt %,and a precise amount of bis-maleimide cross-linker was added. Aftervigorous stirring, the mixture was heated at 70 °C in an oven. Thecross-linking reaction was checked daily by turning the flask upsidedown to control the gel formation. The rDA reaction was carried outby heating the cross-linked sample at 120 °C in an oven for 1 h, thusreturning the mixture to its liquid state.

Characterizations. FTIR Spectroscopy. FTIR spectra wererecorded in the attenuated total reflectance (ATR) mode on a Nicolet380 FTIR or a PerkinElmer Spectrum 1000 spectrometer in the rangeof 400−4000 cm−1, as the average of 32 scans with 4 cm−1 resolution.The samples were deposited on the ATR crystal and carefully pressedto ensure a good contact. The background was recorded with theempty ATR crystal in air.

Size Exclusion Chromatography (SEC). SEC measurements onfuran-bearing tannins were performed in chloroform (HPLC grade) ina Shimadzu liquid chromatograph equipped with a 5 μ PL Gel Guardcolumn, two PL-gel 5 μ MIXED-C and 5 μ 100 Å 300 mm-columns,two online detectors, a Shimadzu RID-10A refractive index detector,and a Shimadzu SPD-M10A diode array (UV) detector. Molar massesand dispersity were calculated from a calibration with polystyrenestandards. Acetylated tannin samples were dissolved in chloroform andfiltered through a 0.2 μm PTFE membrane. For all analyses, theinjection volume was 50 μL, and flow rate was 0.8 mL min−1. Also, theoven temperature is set at 25 °C.

SEC evidencing the rDA reaction was recorded using a tripledetection SEC from Agilent Technologies. The system used twoPL1113−6300 ResiPore 300 mm × 7.5 mm columns with DMF as theeluent at a flow rate of 0.8 mL min−1. The detector suite comprised aPL0390−0605390 LC light scattering detector with two diffusionangles (15° and 90°), a PL0390−06034 capillary viscometer, and a390-LC PL0390−0601 refractive index detector. The entire SECsystem was thermostated at 35 °C. PMMA standards were used for thecalibration.

1H NMR. Spectra were acquired on a Bruker 400 MHz spectrometer(10 s delay, 32 scans) at room temperature or 100 °C for the rDAcharacterization. An accurately weighed amount of sample (about 20mg) was dissolved in 500 μL of deuterated solvent. For the neatsamples, DMSO-d6 was used, whereas the acetylated tannins wereanalyzed in CDCl3. For NMR titration, a precise amount of sampleand external standard were weighed: 2,3,4,5,6-pentafluorobenzalde-hyde or 2,4,6-trimethylphenol for furan or maleimide determination,respectively.

31P NMR. 31P NMR was performed after phosphitylation of thesamples, according to standard protocols.31 The spectra weremeasured on a Bruker 400 MHz spectrophotometer, with 128 scans

Scheme 1. Reaction of Tannins with Furfuryl Glycidyl Ether (FGE)

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1200

and a 15 s relaxation delay. Peaks assignations were performed basedon previous studies on tannin model compounds.32,33

Thermal Characterizations. Thermogravimetric analyses (TGA)were carried out on a TGA 51 apparatus from TA Instruments, fromroom temperature to 700 °C, at a heating rate of 10 °C min−1 undernitrogen flow.Differential scanning calorimetry (DSC) analyses were performed

under an inert atmosphere with a calorimeter DSC1 from MettlerToledo. All the samples were heated from −60 to 190 °C at a heatingrate of 10 °C min−1 and then cooled to room temperature for 1 h priorto a second run.

■ RESULTS AND DISCUSSION

Grafting of Furan Groups onto Mimosa Tannins (MT).MT were reacted with FGE to obtain a multifunctional furan-containing polymer (Scheme 1). The reaction was conductedwith various amounts of FGE, ranging from 0.25 to 1.5 equiv tototal phenolic OH in tannins, and the product was recoveredand purified after an acidic precipitation. The reactionconsumes phenolic OH groups and creates new aliphatic OHgroups, as a result of the opening of the oxirane ring. Similarreactions conducted on lignin showed that under these reactionconditions only secondary aliphatic OH groups are created,34

i.e., that the nucleophilic attack of the epoxy ring by thephenolate anion only takes place at the less hindered carbon.The content in OH of the modified tannins was measured by

31P NMR on phosphitylated samples and 1H NMR onacetylated samples (Figure 1a). Figure 1 shows the content inphenolic and aliphatic OH groups depending on the amount ofFGE used for the reaction. The decrease in phenolic groups

attests for the successful grafting (Figure 1b). Surprisingly, thecontent in aliphatic OH groups does not increase significantlywith the grafting (Figure 1c). The aliphatic OH content is evenlower for the derivatized tannins than for the initial one. Acareful study of the 31P NMR spectra of the initial tannin,shown in the Supporting Information (Figure S1), reveals thepresence of carbohydrate impurities, which give rise to somesignal in the 147−150 ppm region.8 After the reaction, thissignal disappeared, suggesting that the residual carbohydrateswere lost, probably during the washing steps in water. A newwell-defined peak at 146.3 ppm confirms the creation of thenew aliphatic OH group, generated by the opening of the epoxyring (Scheme 1).Surprisingly, the values obtained with 31P NMR are always

significantly lower than those measured by 1H NMR (Figure 1band c), whereas previous studies showed a good correlationbetween both measurements in condensed tannins.35 Bothmeasurements rely on the derivatization of the remaining OHgroups, either with a phospholane or an acetyl group. As shownon Figure 1a, the phospholane group is much bulkier than theacetyl group, and it thus seems possible that it fails to access allof the remaining OH groups present in the derivatized tannins,leading to an underestimation.The precise quantification of the grafted furan groups is

mandatory to study the DA reaction. It can directly be obtainedfrom the decrease in phenolic OH groups Δ(ϕ−OH), takinginto account the mass increase induced by the grafting (eq 1):

Figure 1. Derivatization of OH groups with 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane (top) or acetic anhydride (bottom) for thequantification by 31P or 1H NMR, respectively (a), and quantity of phenolic (b) and aliphatic OH groups (c) depending on the amount of FGE usedfor the reaction.

Table 1. Content in OH and Furan Groups in Derivatized Tannins Measured with 31P and 1H NMR

furan content (mmol g−1)

name FGE equiv residual ϕ−OH (mmol g−1)a by Δ(ϕ−OH) methodb by direct 1H NMR methodc average

MT 0 − − − −MT-F0.25 0.25 3.72 2.12 2.23 2.18 ± 0.05MT-F0.5 0.5 2.76 2.52 2.63 2.58 ± 0.05MT-F0.75 0.75 2.43 2.64 3.22 2.93 ± 0.29MT-F1 1 2.29 2.69 3.34 3.02 ± 0.33

MT-F1.25 1.25 2.02 2.78 3.09 2.94 ± 0.16MT-F1.5 1.5 1.41 2.97 2.96 2.97 ± 0.01

aBased on 1H NMR data. bDetermined using eq 1. cObtained by direct integration of furan signal in 1H NMR (Figure 2).

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1201

ϕϕ

= × Δ ‐+ Δ ‐ ×

−

MFuran(mmolg ) 1000

( OH)1 ( OH)

1

FGE (1)

where Δ(ϕ−OH) stands for the decrease in phenolic OHgroups (in mol g−1) and MFGE for the molar mass of FGE(154.2 g mol−1).The calculation has been performed taking into account the

Δ(ϕ−OH) value obtained by 1H NMR. The results arecompiled in Table 1. 1H NMR was also used to directlyquantify the amount of furan groups grafted to the tannins. TheH atoms from the furan ring give rise to two strong signals at6.4 (2H) and 7.6 ppm (1H) (Figure 2a). The latter was used to

quantify the amount of furan groups present in each samplesince it is fairly well separated from the aromatic H atoms oftannins. A good correlation between both quantificationtechniques is obtained. The amount of grafted furan groupsquickly rises with the amount of FGE used to reach a plateaufor more than 0.75 equiv.FTIR was also used to evidence the successful grafting

(Figure 2b). New characteristic signals assigned to the graftedFGE appear clearly: the C−O stretch around 1070 cm−1 or thefuran ring at 740 cm−1. The height of this last peak is a goodindicator of the grafting since it is proportional to the contentin furan groups measured by 1H NMR (Figure S2, SI). Inaddition, a signal is detected in the CO stretch region around

1710 cm−1. It was later found that its intensity increased linearlywhen the precipitation pH was reduced (Figure S3, SI), thussuggesting acid-induced side reactions occurring during theprecipitation and/or washing steps. A similar band wasobserved on the neat tannin after solubilizing in water,acidifying to pH 2 and freeze-drying (Figure S4, SI). Thissuggests that the side reaction affects the tannin backbonerather than the grafted side chain (a putative mechanism isproposed in the SI, Scheme S1).Precipitation and washings should then be conducted at an

intermediate pH to avoid important changes in the tanninstructure. However, since the recovered yield is also correlatedto the precipitation pH (Figure S5, SI), a compromise has to befound. For the rest of the study, a precipitation at pH 3 waschosen.SEC was then performed on all samples. The chromatograms

are presented on Figure 3a. The derivatization causes a clearshift toward higher molar masses. The positions of the twomain peaks then remain unchanged, whatever the amount ofFGE used for the reaction. Increasing signal in the high molarmass region suggests possible cross-linking reactions, leading toa large increase in the weight-average molar mass Mw (Figure3b).All these results show that using a large excess of FGE is not

necessary since it does not allow to significantly increase thegrafting (Table 1) nor to ameliorate the recovered yield (FigureS6, SI). For the rest of the study, MT modified with 1.15 equivof FGE was used unless stated otherwise.

Model Study of DA and rDA Reaction with N-Methylmaleimide. To study the DA reaction with furan-bearing tannins, a model reaction was carried out with N-methyl maleimide (Scheme 2). This compound has beenchosen because the expected 1H NMR signal of the methylicprotons in the synthesized DA adduct should not overlap withother tannin signals.After purification, the tannin reacted with N-methyl

maleimide was characterized by 1H NMR spectroscopy (Figure4). The success of the model reaction was confirmed by theappearance of characteristic signals of the protons of aDiels−Alder adduct at 2.9−3.0 ppm (signals k, j, and j′),3.7−4.2 ppm (d′), and 6.5 ppm (g and g′), as well as thedecrease in the signals of the furan ring at 6.4 and 7.6 ppm (eand e′, and f, respectively). However, the latter did notcompletely disappear indicating that the reaction is notcomplete. The reaction time was increased up to 15 days,and the remaining furan groups were titrated via NMR bycomparing signal f with 2,3,4,5,6-pentafluorobenzaldehyde as anexternal standard. The conversion was found to be of 50%, theDA reaction being limited, for instance, by steric hindrance.rDA reaction was monitored by 1H NMR at 100 °C by

recording spectra every 20 min for 3 h. Figure 5 displays thespectra recorded at t0 and after 3 h of heating. Severalsignificant changes can be observed including the increase inthe signals at 6.3 and 7.5 ppm (e, e′ and f) corresponding to theprotons of the furan ring, and the decrease in the signalsbetween 2.6 and 3.0 ppm (j and j′), at 3.7 and 4.2 ppm (d′),and at 6.5 ppm (g and g′) related to the Diels−Alder adduct.The 3 h heating at 100 °C was not sufficient to yield the fully“regenerated” furan-bearing tannins, as some signals fromDiels−Alder adduct are still visible, for example, at 6.5 ppm.

Cross-Linking of Furan-Bearing Tannins with Bisma-leimide Oligomers. Three bismaleimide linkers were used tocross-link the furan-bearing tannins (Scheme 3). Cross-linkers

Figure 2. Detail of the 1H NMR spectra of the initial tannin (MT) andtannin derivatized with 1 equiv FGE (MT-F1) (a) and FTIR spectra ofthe same (b).

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1202

X1 and X2 are telechelic poly(propylene oxide) of differentchain lengths, whereas X3 is based on a poly(propylene oxide)-

b-poly(ethylene oxide)-b-poly(propylene oxide) chain. Theaverage degree of polymerization and precise content in

Figure 3. SEC of the derivatized tannins (a) and evolution of Mn and Mw with the amount of FGE used for the reaction (b).

Scheme 2. Reaction of Furan-Bearing Tannins with N-Methyl Maleimidea

aR1 stands for − CH2−CH(OH)−CH2−furan and R2 stands for − CH2−CH(OH)−CH2−DA adduct.

Figure 4. 1H NMR spectra of furan-functionalized mimosa tannins (upper spectrum) and furan-functionalized mimosa tannins reacted with N-methyl maleimide (lower spectrum) recorded in DMSO-d6. R1 = −CH2−CH(OH)−CH2−furan, R2 = −CH2−CH(OH)−CH2−DA adduct.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1203

maleimide groups of the cross-linkers have been determined by1H NMR (Appendix A, SI). The results are gathered in Table 2.

Functionalized tannins were cross-linked with the threebismaleimide oligomers in a 1:1 maleimide/furan molar ratiowith the same experimental conditions (samples XMT1,XMT4, and XMT5 in Table 3). The furan molar ratio wasalso increased to 1.29 and 1.73 to determine its influence on thefinal properties of the cross-linked networks (samples XMT1,XMT2, and XMT3 in Table 3). The DA reaction was carriedout at 70 °C in an oven (Figure 6) and checked regularly byreturning bottles (pictures a and b). After 3 days, XMT5 wasalready cross-linked, whereas XMT3 and XMT4 were onlypartially cross-linked. The presence of a long PEO block in themiddle of cross-linker X3 brings additional flexibility to thechains, which may facilitate the reaction between the terminalmaleimide groups and the furans. All samples were cross-linkedgels after 7 days (picture c).rDA reaction was then carried out by heating the gels at 120

°C in an oven. After 1 h at this temperature, all samplesreturned back to their liquid state (picture d, Figure 6). Thereversibility of the system was tested by heating the solutions at70 °C again. The cross-linking reaction occurred more rapidlyas all samples were back to the gel state after only 2 days(picture e), probably because the previous rDA reaction wasnot complete.After cross-linking in solution, the gels were washed with

water to remove the DMSO and dried under vacuum. Theywere then analyzed by FTIR and compared with both thefuran-bearing tannins and the cross-linkers. Results obtained forsample XMT5 are given in Figure S7 in the SI, as arepresentative example. The DA reaction was confirmed bythe disappearance of the furan ring characteristic signals: theout-of-plane bending of the C−H bond at 738 cm−1 and thestretching of the CC double-bond at 1459 cm−1,36 as well asthose of the maleimide ring; the ring bending at 694 cm−1; theC−H bond bending at 830 cm−1; and the CC double-bondstretching at 1639 cm−1.37 Moreover, the cross-linked tanninscombine both the stretching signals of O−H, C−H, CO, andCC bonds brought by the tannins and the cross-linker X3.Other entries exhibit similar spectra indicating a successfulcross-linking for all samples.Thermal properties of the networks were then evaluated with

TGA and DSC. The thermal stability was first evaluated withTGA. The weight loss and derivative weight loss vs temperaturecurves are displayed in Figure 7, and the maximum degradationtemperatures and char amounts are gathered in Table 3. Allmaterials exhibit an important one-step weight loss with amaximum degradation around 400 °C, caused by thedegradation of the ether groups of PEO and PPO chains ofthe cross-linker, as well as FGE chains. Cross-linked materialsXMT1 to XMT4 also have a small weight loss below 300 °C,which probably arises from the evaporation of residual waterand DMSO entrapped within the polymeric network. The high

Figure 5. 1H NMR spectra recorded at 100 °C in DMSO-d6 for therDA reaction of furan-bearing mimosa tannins functionalized with N-methyl maleimide, at the beginning of the reaction (lower spectrum)and after 3 h (upper spectrum).

Scheme 3. Chemical Structures of Maleimide-TerminatedTelechelic Oligomers Used as Cross-Linkers

Table 2. Degree of Polymerization and Maleimide Contentof Bismaleimide Cross-Linkers

degree of polymerizationa

maleimide (mmol g−1) n x + z y

X1 5.00 ± 0.10 3.9 − −X2 3.09 ± 0.04 7.9 − −X3 1.60 ± 0.04 − 3.9 17.9

aSee Scheme 3 for the meaning of n, x, y, and z.

Table 3. Cross-Linking Experimental Conditions for Functionalized Tannins and Thermal Properties of Obtained Networks

TGA DSC

entry cross-linker maleimide:furan ratio Td,10% (°C) Tmax (°C) char (%) Tg (°C)a TrDA-endo (°C)

b TrDA-exo (°C)b

XMT1 X1 1:1.02 298 400 37 n.o. 126 158XMT2 X1 1:1.29 304 403 35 n.o. 133 159XMT3 X1 1:1.73 256 408 37 n.o. 129 156XMT4 X2 1:1.06 293 394 30 n.o. 127 155XMT5 X3 1:1.04 347 399 23 4 125 155

aDetermined by DSC during the first heating ramp; n.o. = not observed. bDetermined by DSC during the second heating ramp.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1204

cross-link density of these networks, caused by the short PPO

chains (Table 2), probably caused the incomplete removal of

the solvents during the drying step. On the contrary, sample

XMT5 does not exhibit any degradation before 300 °C, thusindicating a very dry material because of the longer cross-linker

chains, which allow an easy removal of solvents during vacuum

drying.

A high char amount (from 23% to 37%) is obtained becauseof the stability of the phenolic tannin backbone, being greater,for example, than a cross-linked epoxy resin based on green teatannin (21% of char at 580 °C under similar conditions).10 Thechar content decreases when increasing the cross-linker chainlength as the relative content of aliphatic units in the materialsis higher. Surprisingly, increasing the amount of cross-linker haslittle influence on the char amount (samples XMT1−XMT3,

Figure 6. Pictures of cross-linked and uncross-linked furan-bearing tannins solutions according to the applied temperature profile.

Figure 7. Mass % vs temperature (left) and weight derivative curves vs temperature (right) for cross-linked networks XMT1−XMT5. Thermogramswere recorded at 10 °C min−1 under nitrogen.

Figure 8. DSC thermograms of XMT5 recorded under N2 from −60 to 190 °C at a heating rate of 10 °C/min. The first heating step is displayed ingreen; the second heating is displayed in black.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1205

Table 3), whereas decreasing char amounts were expected as aresult of a higher aliphatic content.The networks were then analyzed by DSC, and representa-

tive results obtained on sample XMT5 are displayed on Figure8. During the first heating ramp, DSC thermograms exhibit twoendothermic peaks around 130 and 160 °C corresponding tothe rDA reaction for endo and exo adducts, respectively,38 as theendo adduct undergoes rDA at lower temperature.39 Thetemperatures of the rDA for endo and exo adducts are reportedon Table 3. They were found to be independent of the cross-linker (128 ± 3 °C for endo and 157 ± 2 °C for exo). After 1 hcooling, a second heating ramp yields peaks with similar shapearound the same temperatures, arising from an incomplete rDAreaction during the first heating step. This phenomenon isconfirmed by the absence of significant exothermic signalaround 70 °C during the first cooling and second heating steps,indicating that no DA adducts are formed back. Futhermore, iftemperature is kept at 150 °C for 4 h after the first heating step,remaining DA adducts are still observed through the character-istic endothermic peaks during the second heating. However,the reversibility of the cross-linking was confirmed by keepingthe sample at 70 °C for 6 h prior to a third heating ramp. Infact, the area of the rDA peaks was found to be higher for thislast step than during the second heating step, showing thatadducts were formed through DA reaction (Figure S8, SI).The sample prepared with the longest cross-linker XMT5 is

the only one to exhibit a glass transition temperature (Tg). Thepristine cross-linker exhibited a Tg of −53 °C, whereas thecross-linked network has a much higher Tg of 4 °C, asmeasured during the first heating ramp. This is due to thepresence of the aromatic backbone of MT, as well as thesecondary OH groups brought by the FGE ring opening, whichcan allow hydrogen bonds. During the second heating ramp,the Tg is reduced to −7 °C. This decrease is probably caused bythe rDA that occurred during the first heating ramp: the cross-link density is lowered, causing a reduced Tg. In addition, cross-linker molecules liberated during the rDA can act as plasticizer,thus further reducing the Tg.The cross-linked networks are insoluble in organic solvents,

as shown on Figure 9, where the sample XMT5 is immersed ina large excess of DMF. After performing the rDA during 1h at120 °C in DMSO, the sample became soluble and could beanalyzed by SEC. After the rDA, XMT5 exhibits a sharp signalaround 23.5 min and a broad signal between 14 and 23 min.The first corresponds to the free cross-linker X3, while thesecond has a shoulder indicating the presence of the furan-

bearing tannins. The broad signal from 14 to 20 min arises froman incomplete rDA reaction (as stated by DSC) leading tosoluble, but still partially cross-linked oligomers, which thusexhibit higher molar mass.

■ CONCLUSION

The Diels−Alder reaction between furan and maleimidemoieties was exploited to prepare new tannin-based thermo-reversible materials. In the first step, condensed tannins frommimosa (Acacia mearnsii) were modified with FGE to exposefuran groups as the chain end. The reaction was confirmed byFTIR and 31P and 1H NMR. Despite a nonquantitativereaction, derivatized tannins with a high content in furangroups were obtained, up to 3 mmol g−1. A model study for thereaction of furan-bearing tannins with maleimides was carriedout with N-methylmaleimide. The DA reaction was evidencedby 1H NMR with a conversion reaching 50%, and itsreversibility assessed with in situ 1H NMR measurements at100 °C showing the rDA. The furan-bearing tannins were thencross-linked with three telechelic oligomers of PPO or PPO-b-PEO-b-PPO terminated by maleimide groups. The DAreaction, evidenced by FTIR, led to the formation of gelsduring storage at 70 °C. The gels were able to quickly turn backto the liquid state upon exposure to higher temperatures (120°C), even though DSC and SEC experiments showed that therDA reaction was not completed in such a short time. Finally,the obtained aromatic materials exhibited high thermalstabilities and char amounts. The use of these new multi-functionnal tannins and cross-linkers may open the way fortunable, partially, or fully biobased networks with a highrecycling/reuse potential for a very large range of fields such asreversible adhesive, shape memory or self-healing materials, andbuilding, automotive, or biomedical applications.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acssusche-meng.6b02596.

31P NMR and FTIR spectra of the derivatized tannins,yields of the tannin derivatization reactions, FTIR spectraand DSC curves of the Diels−Alder cross-linkednetworks. (PDF)

■ AUTHOR INFORMATION

Corresponding Author*Phone: + 333 68852784. Fax: + 333 68852716. E-mail: luc.averous@unistra.fr.

ORCID

Luc Averous: 0000-0002-2797-226XNotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

Silvateam (Italy) and, particularly, Dr. Samuele Giovando aregratefully acknowledged for kindly supplying tannin samples.Chheng Ngov (ICPEES, Universite de Strasbourg) is thankedfor her technical support.

Figure 9. Normalized refractive index (RI) response of SECchromatograms of cross-linker X3 (red), furan-bearing tannins(blue), and crude product of XMT5 sample after rDA in DMSO.The picture on the left shows the insolubility of XMT5 in DMF priorto the rDA reaction.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1206

■ REFERENCES(1) Arbenz, A.; Averous, L. Chemical modification of tannins toelaborate aromatic biobased macromolecular architectures. GreenChem. 2015, 17 (5), 2626−2646.(2) Quideau, S.; Deffieux, D.; Douat-Casassus, C.; Pouysegu, L. PlantPolyphenols: Chemical Properties, Biological Activities, and Synthesis.Angew. Chem., Int. Ed. 2011, 50 (3), 586−621.(3) Chan, J. M.; Day, P.; Feely, J.; Thompson, R.; Little, K. M.;Norris, C. H. Acacia mearnsii industry overview: current status, keyresearch and development issues. South. For. J. For. Sci. 2015, 77 (1),19−30.(4) Khanbabaee, K.; van Ree, T. Tannins: Classification andDefinition. Nat. Prod. Rep. 2001, 18 (6), 641−649.(5) Aron, P. M.; Kennedy, J. A. Flavan-3-ols: Nature, occurrence andbiological activity. Mol. Nutr. Food Res. 2008, 52 (1), 79−104.(6) Arbenz, A.; Averous, L. Synthesis and characterization of fullybiobased aromatic polyols - oxybutylation of condensed tanninstowards new macromolecular architectures. RSC Adv. 2014, 4 (106),61564−61572.(7) Arbenz, A.; Averous, L. Oxyalkylation of gambier tanninSynthesis and characterization of ensuing biobased polyols. Ind. CropsProd. 2015, 67, 295−304.(8) Duval, A.; Averous, L. Oxyalkylation of Condensed Tannin withPropylene Carbonate as an Alternative to Propylene Oxide. ACSSustainable Chem. Eng. 2016, 4 (6), 3103−3112.(9) Nouailhas, H.; Aouf, C.; Le Guerneve, C.; Caillol, S.; Boutevin,B.; Fulcrand, H. Synthesis and Properties of Biobased Epoxy Resins.Part 1. Glycidylation of Flavonoids by Epichlorohydrin. J. Polym. Sci.,Part A: Polym. Chem. 2011, 49 (10), 2261−2270.(10) Benyahya, S.; Aouf, C.; Caillol, S.; Boutevin, B.; Pascault, J. P.;Fulcrand, H. Functionalized green tea tannins as phenolic prepolymersfor bio-based epoxy resins. Ind. Crops Prod. 2014, 53, 296−307.(11) Aouf, C.; Benyahya, S.; Esnouf, A.; Caillol, S.; Boutevin, B.;Fulcrand, H. Tara tannins as phenolic precursors of thermosettingepoxy resins. Eur. Polym. J. 2014, 55, 186−198.(12) Basnet, S.; Otsuka, M.; Sasaki, C.; Asada, C.; Nakamura, Y.Functionalization of the active ingredients of Japanese green tea(Camellia sinensis) for the synthesis of bio-based epoxy resin. Ind.Crops Prod. 2015, 73, 63−72.(13) Baroncini, E. A.; Kumar Yadav, S.; Palmese, G. R.; Stanzione, J.F. Recent advances in bio-based epoxy resins and bio-based epoxycuring agents. J. Appl. Polym. Sci. 2016, 133 (45), na.(14) Jahanshahi, S.; Pizzi, A.; Abdulkhani, A.; Shakeri, A. Analysis andTesting of Bisphenol A-Free Bio-Based Tannin Epoxy-AcrylicAdhesives. Polymers 2016, 8 (4), 143.(15) Jahanshahi, S.; Pizzi, A.; Abdulkhani, A.; Doosthoseini, K.;Shakeri, A.; Lagel, M. C.; Delmotte, L. MALDI-TOF, C-13 NMR andFT-MIR analysis and strength characterization of glycidyl ether tanninepoxy resins. Ind. Crops Prod. 2016, 83, 177−185.(16) Luo, C.; Grigsby, W. J.; Edmonds, N. R.; Al-Hakkak, J.Vegetable oil thermosets reinforced by tannin−lipid formulations. ActaBiomater. 2013, 9 (2), 5226−5233.(17) Lacoste, C.; Basso, M. C.; Pizzi, A.; Laborie, M.-P.; Celzard, A.;Fierro, V. Pine tannin-based rigid foams: Mechanical and thermalproperties. Ind. Crops Prod. 2013, 43, 245−250.(18) Basso, M. C.; Pizzi, A.; Al-Marzouki, F.; Abdalla, S.Horticultural/hydroponics and floral natural foams from tannins.Ind. Crops Prod. 2016, 87, 177−181.(19) Arbenz, A.; Frache, A.; Cuttica, F.; Averous, L. Advancedbiobased and rigid foams, based on urethane-modified isocyanuratefrom oxypropylated gambier tannin polyol. Polym. Degrad. Stab. 2016,132, 62.(20) Merle, J.; Birot, M.; Deleuze, H.; Mitterer, C.; Carre, H.;Bouhtoury, F. C.-E. New biobased foams from wood byproducts.Mater. Des. 2016, 91, 186−192.(21) Polgar, L. M.; van Duin, M.; Broekhuis, A. A.; Picchioni, F. Useof Diels−Alder Chemistry for Thermoreversible Cross-Linking ofRubbers: The Next Step toward Recycling of Rubber Products?Macromolecules 2015, 48 (19), 7096−7105.

(22) Hager, M. D.; Bode, S.; Weber, C.; Schubert, U. S. Shapememory polymers: Past, present and future developments. Prog. Polym.Sci. 2015, 49−50, 3−33.(23) Rivero, G.; Nguyen, L.-T. T.; Hillewaere, X. K. D.; Du Prez, F.E. One-Pot Thermo-Remendable Shape Memory Polyurethanes.Macromolecules 2014, 47 (6), 2010−2018.(24) Hillewaere, X. K. D.; Du Prez, F. E. Fifteen chemistries forautonomous external self-healing polymers and composites. Prog.Polym. Sci. 2015, 49−50, 121−153.(25) Liu, Y.-L.; Chuo, T.-W. Self-healing polymers based onthermally reversible Diels−Alder chemistry. Polym. Chem. 2013, 4(7), 2194−2205.(26) Tillet, G.; Boutevin, B.; Ameduri, B. Chemical reactions ofpolymer crosslinking and post-crosslinking at room and mediumtemperature. Prog. Polym. Sci. 2011, 36 (2), 191−217.(27) Froidevaux, V.; Negrell, C.; Laborbe, E.; Auvergne, R.; Boutevin,B. Thermosetting material by a thermo responsive cross-linking usingretroDiels-Alder and, in situ, Thia-Michael reactions. Eur. Polym. J.2015, 69, 510−522.(28) Sanyal, A. Diels−Alder Cycloaddition-Cycloreversion: APowerful Combo in Materials Design. Macromol. Chem. Phys. 2010,211 (13), 1417−1425.(29) Gevrek, T. N.; Arslan, M.; Sanyal, A. Design and Synthesis ofMaleimide Group Containing Polymeric Materials via the Diels-Alder/Retro Diels-Alder Strategy. In Functional Polymers by Post-Polymer-ization Modification; Theato, P., Klok, H.-A., Eds.; Wiley-VCH VerlagGmbH & Co. KGaA, 2012; pp 119−151.(30) Gandini, A. The furan/maleimide Diels−Alder reaction: Aversatile click−unclick tool in macromolecular synthesis. Prog. Polym.Sci. 2013, 38 (1), 1−29.(31) Granata, A.; Argyropoulos, D. S. 2-Chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane, a Reagent for the Accurate Determination ofthe Uncondensed and Condensed Phenolic Moieties in Lignins. J.Agric. Food Chem. 1995, 43 (6), 1538−1544.(32) Melone, F.; Saladino, R.; Lange, H.; Crestini, C. TanninStructural Elucidation and Quantitative 31P NMR Analysis. 1. ModelCompounds. J. Agric. Food Chem. 2013, 61 (39), 9307−9315.(33) Melone, F.; Saladino, R.; Lange, H.; Crestini, C. TanninStructural Elucidation and Quantitative 31P NMR Analysis. 2.Hydrolyzable Tannins and Proanthocyanidins. J. Agric. Food Chem.2013, 61 (39), 9316−9324.(34) Duval, A.; Lange, H.; Lawoko, M.; Crestini, C. Reversiblecrosslinking of lignin via the furan−maleimide Diels−Alder reaction.Green Chem. 2015, 17 (11), 4991−5000.(35) Duval, A.; Averous, L. Characterization and PhysicochemicalProperties of Condensed Tannins from Acacia catechu. J. Agric. FoodChem. 2016, 64 (8), 1751−1760.(36) Madhavaram, H.; Idriss, H. Acetaldehyde reactions over theuranium oxide system. J. Catal. 2004, 224 (2), 358−369.(37) Aguiar, E. C.; da Silva, J. B. P.; Ramos, M. N. A theoretical studyof the vibrational spectrum of maleimide. J. Mol. Struct. 2011, 993(1−3), 431−434.(38) Dolci, E.; Michaud, G.; Simon, F.; Boutevin, B.; Fouquay, S.;Caillol, S. Remendable thermosetting polymers for isocyanate-freeadhesives: a preliminary study. Polym. Chem. 2015, 6 (45), 7851−7861.(39) Froidevaux, V.; Borne, M.; Laborbe, E.; Auvergne, R.; Gandini,A.; Boutevin, B. Study of the Diels−Alder and retro-Diels−Alderreaction between furan derivatives and maleimide for the creation ofnew materials. RSC Adv. 2015, 5 (47), 37742−37754.

ACS Sustainable Chemistry & Engineering Research Article

DOI: 10.1021/acssuschemeng.6b02596ACS Sustainable Chem. Eng. 2017, 5, 1199−1207

1207