A self-healing polymer network based on reversible covalent bonding

Reactive & Functional Polymers 73 (2013) 413–420

Contents lists available at SciVerse ScienceDirect

Reactive & Functional Polymers journal homepage: www.elsevier.com/locate/react

A self-healing polymer network based on reversible covalent bonding G. Scheltjens ⇑, M.M. Diaz, J. Brancart, G. Van Assche, B. Van Mele Vrije Universiteit Brussel, Physical Chemistry and Polymer Science (FYSC), Pleinlaan 2, B-1050 Brussels, Belgium

a r t i c l e

i n f o

Article history: Available online 6 July 2012 Keywords: Self-healing coatings Diels–Alder Kinetics Gel-point Repeatability

a b s t r a c t A self-healing polymer network for potential coating applications was designed based on the concept of the reversible Diels–Alder (DA) reaction between a furan functionalized compound and a bismaleimide. The network allows local mobility in a temperature window from ca. 80 °C to 120 °C by shifting the DA equilibrium towards the initial building blocks. Changing the spacer length in the furan functionalized compound leads to tailor-made properties. Elastomeric model systems were chosen to evaluate the kinetic parameters by Fourier transform infrared spectroscopy. For the DA reaction a pre-exponential factor ln(ADA in kg mol1 s1) equal to 13.1 ± 0.8 and an activation energy (EDA) of 55.7 ± 2.3 kJ mol1 are found. For the retro-DA reaction, ln(ArDA) and ErDA are 25.8 ± 1.8 s1 and 94.2 ± 4.8 kJ mol1, respectively. The enthalpy and entropy of reaction are calculated as 38.6 kJ mol1 and 105.3 J mol1 K1. The kinetic results are validated by micro-calorimetry. Non-isothermal dynamic rheometry provides the gel-point temperature of the reversible network. The sealing capacity is evaluated by atomic force microscopy for micro-meter sized defects. Repeatability of the non-autonomous healing is checked by micro-calorimetry, ruling out side-reactions below 120 °C. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction In the field of organic coatings, thermosetting materials are widely used for corrosion prevention. These densely cross-linked structures possess advantageous barrier and mechanical properties compared to thermoplastics [1]. However, their life-time span is limited due to a number of factors, such as loss of adhesion, degradation by photo-initiated oxidation, and hydrolysis [2]. In addition, changes in temperature and humidity may result in cracking due to a difference in expansion coefficient between substrate and coating. When micro-cracks are formed through the brittle network coating, the underlying metal is left exposed, leading to an accelerated corrosion process [3]. Self-healing materials are promising systems to deal with this kind of coating failure. Over the last few years, strong research has been performed on this subject, giving rise to the development of numerous self-healing materials [4]. Two main groups can be identified based on their activation trigger: autonomic (or intrinsic) and non-autonomic (or extrinsic) self-healing systems [4,5]. In case of autonomic systems, the mobile phase responsible for healing is embedded in the coating matrix. This means no intervention is required as the material holds the substance that allows regeneration [6]. Therefore, the self-healing efficiency, and also repeatability, is limited to the amount of regenerating substance available. ⇑ Corresponding author. Tel.: +32 2 629 32 76. E-mail address: [email protected] (G. Scheltjens). 1381-5148/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.reactfunctpolym.2012.06.017

For the non-autonomic self-healing systems, different external triggers are available, such as photo, mechanical, chemical and heat stimuli [7]. One specific type of a non-autonomic self-healing design is based on reversible cross-links using heat as an external trigger. This category comprises hydrogen bonds, ionomers, coordination bonds, and reversible covalent bonds [8]. The reversible [4 + 2] cyclo-addition, also known as the Diels–Alder (DA) reaction, has been introduced into thermosetting polymeric systems in order to synthesize thermo-remendable and recyclable materials [9–11]. In earlier work, we also demonstrated the feasibility of a reversible polymer network based on the DA chemistry between a bismaleimide and a furan functionalized compound (see Scheme 1). The reaction equilibrium is shifted from diene and dienophile towards cyclo-adduct (and backward) by decreasing (increasing) the temperature. Healing of macroscopic and microscopic defects takes place in a temperature window ranging from 80 to 130 °C [12,13]. The advantage of this particular self-healing system lies in its flexible network design by using spacers in the furan functionalized compound of different length. Networks with shorter spacers lead to a higher cross-link density and thermomechanical properties suited for coating applications. It was shown that higher concentrations of cross-links shift the thermo-reversibility towards higher temperatures [13]. The focus of this paper is on the thermal evaluation of the DA reaction between a bismaleimide and a 4-functional furan compound. Model systems with longer spacers (less suited for applications) are chosen based on their low glass transition temperature in order to avoid any vitrification during the kinetic evaluation.

414

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

O

O

+

O

R1

Furan (F)

N

R2

O

Δ

O

Maleimide (M)

N

R2

O

R1

Diels-Alder Adduct (A)

Scheme 1. Diels–Alder equilibrium reaction between furan and maleimide functional groups.

Furan and maleimide conversions as a function of time at different curing temperatures are obtained by temperature-controlled Fourier transform infrared (FTIR) spectroscopy. The kinetic model is validated through micro-calorimetry. The determination of preexponential factors, activation energies, molar heat of reaction and entropy of reaction will allow a controlled optimization of the healing procedure. In addition to kinetic modelling, rheological information is essential to take into account the thermo-mechanical behavior during healing. The self-healing mechanism under study is based on a temporary increase in local mobility due to a decrease in viscosity upon heating. The change in viscosity and viscoelastic behavior should be limited in order to maintain the mechanical integrity of the coating on the substrate. The flow behavior of the model network is studied by dynamic rheometry and allows determination of a gel-point temperature. In addition, sealing of microscopic scratches through viscous flow is determined by atomic force microscopy at elevated temperatures. After sealing any damage at higher temperature in the thermoreversible temperature window, the recovery of initial properties takes place in a subsequent cooling with reformation of covalent bonds by DA cyclo-addition. An important concern is the repeatability of the reversible DA reaction. In order to ensure repeatable healing, the concentration of reactive groups must be maintained. No leaching of reactive component should occur and no irreversible side-reaction should take place during the healing procedure. Repeatability of the exothermic DA reaction of the model system is studied by micro-calorimetry at ambient temperature to rule out any side-reaction in the thermo-reversible temperature window used in this study.

O

(A)

H2N

NH2 n

O O

(B)

O

O

O

(C) N

N O

O

Scheme 2. Monomers used for the synthesis of the reversible network system: (A) Jeffamine D-series Jx (x = 230, 400, 2000, 4000), (B) FGE, and (C) DPBM.

furan functionalized compound (FGE-Jx) was mixed with DPBM in a stoichiometric ratio to obtain the reversible polymer network through DA reaction. Prior to the DA reaction, the DPBM was dissolved in chloroform to obtain a homogeneous reversible network. In order to create bulk samples, the DA curing reaction between FGE-Jx and DPBM was carried out at 25 °C by solvent casting, and traces of chloroform were removed under vacuum.

2. Experimental 2.3. Thermal analysis techniques 2.1. Materials Poly(propylene glycol) bis(2-aminopropyl ether) (Jeffamine Dseries) with average degree of polymerization n (determined by nuclear magnetic resonance spectroscopy) of 2.9 (trade name Jeffamine D-230 or abbreviated J230, M n = 240 g mol1), 7.0 (J400, M n = 478 g mol1), 39.2 (J2000, M n = 2345 g mol1), 99.4 (J4000, M n = 5839 g mol1), furfuryl glycidyl ether (FGE, 96%), 1,10 -(methylenedi-1,4-phenylene)bismaleimide (DPBM, 95%), and triethylamine were purchased from Sigma–Aldrich. All products were used as received (see Scheme 2). 2.2. Remendable polymer network synthesis The synthesis of the remendable network was performed in two steps. In a first step, FGE was irreversibly bonded to Jx (x = 230, 400, 2000, and 4000) through an epoxy–amine reaction, yielding a furan functionalized compound. This reaction was performed at 80 °C without any solvent until completion. In a second step, the

2.3.1. Differential scanning calorimetry (DSC) DSC experiments were performed on a Q2000 from TA instruments, equipped with a liquid nitrogen cooling device. Tzero calibration was performed at 2.5 °C min1 and 20 °C min1 with sapphire disks and temperature and enthalpy calibration were performed using an indium standard. A heating rate of 20 °C min1 was applied to evaluate initial building blocks, Jx and FGE-Jx. In order to characterize the synthesized remendable polymer networks, modulated temperature DSC experiments (MTDSC) were performed at 2.5 °C min1. 2.3.2. Micro-calorimetry The micro-calorimeter used was a thermal activity monitor TAMIII from TA instruments, with a precision of 200 nW. In a glass ampoule of 4 ml a bulk sample was heated to a maximum temperature of 150 °C for 15 min. Immediately after heating, the material was quenched in liquid nitrogen. In the micro-calorimeter, the formation of covalent bonds by the exothermic DA reaction was stud-

415

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

ied isothermally at 20 °C. Repeatability of the DA reaction was verified at 25 °C in consecutive cycles, after quenching the material that was thermally treated at 120 °C in an oven for 20 min in each cycle.

Table 1 Evolution of Tg measured by DSC for the consecutive synthesis steps.

2.4. FTIR spectroscopy Spectroscopic studies were performed using a Thermos Scientific Fourier-Transform Infra-Red (FTIR) spectrometer equipped with a horizontal transmission temperature-controlled heating stage. Samples for IR spectroscopy were produced by spincoating a thin film on a KBr crystal by a SPIN150-v3 spincoater from SPSEurope. A freshly made 15 wt% chloroform solution containing FGE-J2000 and DPBM in a stoichiometric ratio (model system) was spincoated at 4000 rpm for 1 min. In succession, a spin drying step was programmed to remove traces of chloroform at 8000 rpm for 40 s. In general, the elapsed time between mixing and the initial spectrum collection was less than 3 min. The reaction during this period is assumed to be negligible as the DA reaction of the model system between FGE-J2000 and DPBM proceeds slowly at ambient temperature. Initial conversions of the maleimide functionalities were determined at different curing temperatures (Tcure) by the (decreasing) evolution of a maleimide absorption peak, representative for DPBM, centred at 1146 cm1 [14]. The DA/retro-DA equilibrium conversion (as a function of isothermal Tcure) was determined from the absorption peak centred at 1174 cm1, representative for the DA adduct [12]. This spectroscopic study was repeated for the DA reaction between FGE-J4000 and DPBM. 2.5. Dynamic rheometry Rheological experiments were performed on a TA instruments AR-G2 rheometer equipped with electrical heating plates and aluminum parallel plate geometry with a diameter of 5 mm and a gap size of 500 lm. A thermal sweep was performed ranging from 60 °C to 110 °C at a heating rate of 2 °C min1. During the thermal scan, the frequency was switched between 0.01 Hz, 0.32 Hz, and 1 Hz. 2.6. Atomic force microscopy Atomic force microscopy measurements were performed on an Asylum Research MFP-3D AFM equipped with an Olympus AC160TS-E2 AFM tip for imaging (tapping mode) and nanolithography (calibrated spring constant of 30.34 N m1). In situ thermal treatment of polymer coatings was performed with an Asylum Research polymer heater. 3. Results and discussion 3.1. Glass transition of DA polymer networks The glass transition temperature throughout the different synthesis steps was characterized by DSC for all materials having a different spacer length. When comparing the Tg’s of the initial Jeffamine D-series with those of the furan functionalized series, the conclusion can be drawn that the end-group effect of the irreversibly linked FGE on Tg is highest for J230 (see Table 1). This endgroup effect decreases with increasing spacer length, resulting in an almost negligible end-group effect for FGE-J4000. From Table 1, it can be concluded that with decreasing spacer length, or increasing cross-link density, the Tg of the network increases. The cured FGE-J230+DPBM as well as FGE-J400+DPBM are behaving like a brittle thermoset (at ambient temperature). Because of the fact that the Tg of the polymerizing network exceeds

Spacer Jx

Tg of Jx (°C)

Tg of FGE-Jxa (°C)

Tg of FGE-Jx+DPBMb (°C)

J230 J400 J2000 J4000

87.0 75.3 69.1 67.8

28.8 36.0 58.1 65.8

63.9 55.5 55.3 64.6

(thermoset) (thermoset) (elastomer) (elastomer)

a

Synthesis of FGE-Jx was performed at 80 °C until completion. Synthesis of FGE-Jx+DPBM was performed through DA curing reaction at ambient temperature. b

the ambient curing temperature (Tcure), mobility limitations become significant. In this case, vitrification of the network interferes, impeding the equilibrium to be reached. As a consequence, it should be noted that the measured Tg for FGE-J230+DPBM and FGE-J400+DPBM is not the ultimate attainable Tg. In contrast, both cured FGE-J2000 and FGE-J4000 with DPBM have lower cross-link densities, resulting in a rubbery behavior at ambient temperature. The final Tg of these elastomers (55.3 and 64.6 °C, respectively) is well below the ambient curing temperature, meaning that vitrification will not occur during this DA curing reaction. For this reason, J2000 and J4000 are chosen as spacer in the reversible network and used as model systems to study the equilibrium and kinetics of the DA reaction between a 4-functional furan compound (FGE-J2000 and FGE-J4000) and a bismaleimide (DPBM). 3.2. Kinetics and equilibrium of DA cyclo-addition of a model polymer network The kinetics of the Diels–Alder reaction as well as the DA/retroDA equilibrium was studied by FTIR spectroscopy. Typical spectra are shown in [12]. The Diels–Alder adduct, composed by reversible linkage between FGE-J2000 and DPBM is favored at ambient temperature (see Scheme 1). The recorded initial absorbance spectra for the different isothermal curing temperatures can be compared in a quantitative way after correction against an internal reference peak centred at 1513 cm1, representing a CAC stretch in the phenyl groups of DPBM [14]. The maleimide peak height at 1146 cm1 is therefore corrected with the height of the internal reference peak, resulting in an absorbance ratio (ARDPBM). This AR is used to calculate the conversion (x(t)) of maleimide functional groups.

xðtÞ ¼ 1 

ARDPBM ðtÞ ARDA ðtÞ ¼ ARDPBM ð0Þ ARDA ð0Þ

ð1Þ

In order to determine the DA/retro-DA equilibrium as a function of Tcure, the reacting system is allowed sufficient time to reach equilibrium, leading to constant spectra in time. Instead of the decreasing maleimide peak height at 1146 cm1 with time, the peak height at 1174 cm1, representing the generated DA adduct, is used to calculate the absorbance ratio ARDA and thus the equilibrium conversion (see Eq. (1)). This procedure leads to an optimal quantification of the IR spectra for determination of both initial reaction rate and equilibrium of the DA/retro-DA system studied. Maleimide conversion, determined by FTIR, as a function of time for FGE-J2000+DPBM copolymerization at several curing temperatures is shown in Fig. 1. For clarity, furan conversions are not shown as maleimide and furan functionalities are consumed at equal rates. The temperature dependence of the rate and equilibrium of the DA reaction is consistent with the fact that a decrease in curing temperature leads to a decrease in conversion rate and an increase in equilibrium conversion (see Fig. 1). As the curing temperature increases, the retro Diels–Alder reaction becomes more prominent, shifting the equilibrium towards higher concentrations of the diene (FGE-J2000) and dienophile (DPBM).

416

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

Fig. 1. Experimental maleimide conversions (x(t)) for stoichiometric mixtures of FGE-J2000 and DPBM, at 27.3 °C (h), 42.9 °C (), 49.6 °C (D), 58.4 °C (s) and 69.2 °C (+). Both initial conversions (up to 60 min) and equilibrium conversions are shown. The solid lines represent the modelled maleimide conversions.

The overall reaction rate of maleimide (M) in the reversible DA reaction with furan (F) is written as

d½M ¼ kDA ½F½M þ krDA ½A dt

ð2Þ

with [M] and [F] the concentration of maleimide and furan functional groups, [A] the concentration of cyclo-adduct, and kDA and krDA the reaction rate constants of the forward and reverse DA reaction. For stoichiometric mixtures of maleimide and furan, Eq. (2) can be re-written in terms of conversion (x(t))

dxðtÞ ¼ kDA ½M0 ð1  xðtÞÞ2  krDA xðtÞ dt

ð3Þ

With [M]0 the initial maleimide concentration in the reaction mixture. This allows non-linear model-fitting by a least sum of square differences between the measured and calculated data (see Fig. 1) [15]. Using the Arrhenius equations of kDA and krDA this modelfitting leads to the determination of pre-exponential factor and activation energy of both forward and reverse DA reaction, as shown in Table 2 with experimental error estimation. Using the Arrhenius equations of kDA and krDA, the reaction enthalpy DrH0 and the reaction entropy DrS0 can be calculated via the van’t Hoff equation (see Fig. 2 and Table 2):

lnðKÞ ¼ ln

!

x ½M0 ð1  xÞ

2

 ¼ ln

kDA krDA

 ¼

Dr S0 Dr H0  R RT

ð4Þ

with K the equilibrium constant and x the maleimide equilibrium conversion.

Table 2 Optimized kinetic (ADA, EDA, ArDA, and ErDA) and calculated thermodynamic (DrH0 and DrS0) parameters for the DA reaction between furan and maleimide functional groups. An experimental error estimation is provided for FGE-J2000+DPBM. The reaction rate constants, kDA and krDA, at 60 °C are given for comparison. Parameter

Units

FTIR:FGEJ2000+DPBM

FTIR:FGEJ4000+DPBM

Literature [16]

ln(ADA) EDA ln(ArDA) ErDA DrH0 DrS0 ln(kDA,60 °C) ln(krDA,60 °C)

kg mol1 s1 kJ mol1 s1 kJ mol1 kJ mol1 J mol1 K1 kg mol1 s1 s1

13.1 ± 0.8 55.7 ± 2.3 25.8 ± 1.8 94.2 ± 4.8 38.6 105.3 7.0 8.3

13.1 56.4 25.8 95.2 38.8 106.0 7.3 8.6

9.7 48.0 22.4 88.0 40.0 106.0 7.6 9.3

Fig. 2. Natural logarithm of the equilibrium constant (K) from the optimized kDA and krDA for FGE-J2000+DPBM (solid line) and FGE-J4000+DPBM (short dashed line) as a function of reciprocal curing temperature. The experimental points are given for comparison: final values of the conversion profiles for FGE-J2000+DPBM (open symbols; see Fig. 1) and FGE-J4000+DPBM (j). The long dashed line represents ln(K) values from literature [16]. All studies are shown in their respective T-range.

The DA reaction between FGE-J4000 and DPBM was also studied by FTIR, following the same procedure as for the J2000 spacer, leading to very comparable optimized kinetic parameters (see Table 2) and similar equilibrium constants K (see Fig. 2) within experimental error of the FGE-J2000+DPBM system. The parameter set for FGE-2000+DPBM, however, was determined using more accurate analytical data in a wider temperature range. It is therefore chosen as representative for all FGE-Jx+DPBM systems, and will be used as such in following discussions in this paper. The representative set is in good agreement with literature values (see Fig. 2 and Table 2). The kinetic and thermodynamic parameters of Adzima et al. [16], however, predict lower reaction rate constants for both forward and reverse DA reactions (compare the values at 60 °C in Table 2). The effect is more pronounced for the reverse reaction, leading also to slightly lower values of the equilibrium constant than the ones observed in the temperature window of this study (see Fig. 2). The observed difference might be explained by the fact that in literature a 3-functional furan compound, pentaerythritol propoxylate tris(3-(furfurylthiol)-propionate) (PPTF), was used to characterize the DA reaction between furan and maleimide functional groups. While the FGEJ2000+DPBM and FGE-J4000+DPBM model systems resulted in elastomeric polymer networks, the 3-functional furan compound PPTF described in literature leads to a reversible thermosetting material, vitrifying around 45 °C [16]. The proximity of Tg of these thermosetting networks might slow down the reaction rates of the reversible DA reactions by diffusion effects, i.e. mobility restrictions, in the temperature window studied. The representative set of kinetic parameters obtained from isothermal FTIR data enables the simulation of non-isothermal experiments, predicting the reaction conversion during heating or cooling at different rates. Fig. 3 clearly shows that with increased heating rate, the system is deviating further from equilibrium, especially at lower temperatures due to too slow reaction rates. The reaction rate constants, kDA and krDA, increase with temperature according to Arrhenius law, enabling the system to regain equilibrium at elevated temperatures. In a subsequent cooling, similar effects are noticed. At a cooling rate of 0.2 °C min1, equilibrium is lost below approximately 70 °C, while the reaction conversion remains quasi constant over the full temperature span during cooling at 1000 °C min1. The fact that the equilibrium condition at 150 °C can be frozen in

417

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

Fig. 4 shows a good agreement between the experimental reaction rate at 20 °C (heat flow) and the simulated rate according to the kinetic and thermodynamic parameters from FTIR. Moreover, it can be seen that the exothermic partial heat of reaction (DrHpart) is measured to be 17.2 J g1. This value corresponds to the reformation of 73% of covalent bonds when achieving an equilibrium conversion of 0.77 at 20 °C. The simulated DrHpart is equal to 16.1 J g1, representing 68% of covalent bonds being reformed. It is clear that the kinetic parameters from FTIR are suited to model the thermal effects associated with the kinetics and equilibrium of the reversible DA reaction.

3.4. Gel-point temperature and conversion of a model polymer network coating

Fig. 3. Simulated equilibrium conversion (solid line) and simulated conversions during a heating scan (long dashed lines) and subsequent cooling (short dashed lines) at various rates for the DA reaction between FGE-J4000 and DPBM based on the kinetic parameters determined by FTIR. It is assumed that the reversible network is in equilibrium at the start of the heating at 20 °C and at the start of the cooling at 150 °C.

by quench-cooling is important to validate the FTIR results for kinetics and equilibrium of the model network by means of micro-calorimetry at ambient temperature. 3.3. Validation of kinetics and equilibrium of DA cyclo-addition The kinetic and thermodynamic parameters obtained by FTIR spectroscopy in the previous paragraph were validated by independent micro-calorimetric measurements, as illustrated for the DA cyclo-addition of FGE-J4000 and DPBM. The high sensitivity and excellent baseline stability of the micro-calorimeter allow an accurate investigation of the slow cyclo-addition of this model system. In a first stage, the remendable model network was heated to 150 °C for 15 min in an oven. This thermal treatment is necessary to drive the equilibrium towards the reactants (see Fig. 3), leading to a lowered maleimide equilibrium conversion of 0.09, according to the representative kinetic parameter set determined by FTIR. Secondly, after quenching in liquid nitrogen, the exothermic DA reaction was monitored in the micro-calorimeter at a constant curing temperature of 20 °C, leading to the reformation of covalent bonds by cyclo-addition (see Fig. 4).

The gel-point temperature (Tgel) and corresponding conversion (xgel) is the temperature (conversion) at which an incipient network is formed in a cross-linking system. It can be determined using dynamic rheometry as the cross-over between G0 (storage modulus) and G00 (loss modulus) at a loss angle of 45°. Tgel is independent of the oscillatory frequency. The conversion at Tgel, the gel-point conversion (xgel), can be predicted by the Flory-Stockmayer equation for step-growth polymerization:

1 xgel ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rð1  fM Þð1  fF Þ

ð5Þ

with r the mixing ratio, and fM and fF the functionality for the maleimide and furan functionalized monomers, respectively [17,18]. If Eq. (5) is valid for the reversible DA cyclo-addition, the gel-point conversion for the reaction between FGE-Jx (fF = 4) and DPBM (fM = 2) is predicted to be 0.58. Using Eq. (4), the gel-point temperature can be estimated for the different reversible networks (see Table 3). The theoretical gel-point temperature obtained for the model network FGE-J4000+DPBM was evaluated experimentally by dynamic rheometry (see Fig. 5). At relatively low temperatures, below 80 °C, the reversible model system exhibits an elastic response which is in the same order of magnitude as for soft rubbery materials [19]. At elevated temperatures, above 100 °C, the loss angle reaches 90°. At these high temperatures, the reversible network structure gets fully destroyed, shifting the DA/r-DA equilibrium towards viscous flow of the initial reactants. For the FGE-J4000+DPBM model system, the cross-over between G0 and G00 leads to a gel-point temperature of 88 °C (see Fig. 5). The loss angle remains frequency independent between ca. 85 °C and 95 °C. It is clear that these values are not corresponding to the theoretical gel-point temperature, ca. 49 °C, as shown in Table 3. However, the non-isothermal simulation of the maleimide conversion at 2 °C min1 (as in dynamic rheometry), shows that the FGE-J4000+DPBM system goes out of equilibrium until a temperature of approximately 120 °C is reached (see Fig. 3). This means that the gel-point temperature for FGE-J4000+DPBM determined by dynamic rheometry (ca. 88 °C) is not corresponding to an equilibrium condition. Based on the simulation in Fig. 3, a gel-point

Table 3 Gel-point temperatures predicted by the Flory-Stockmayer equation (see Eq. (5)) for the DA reaction between FGE-Jx (fF = 4) and DPBM (fM = 2).

Fig. 4. Exothermic DA reaction between FGE-J4000 and DPBM monitored isothermally at 20 °C by micro-calorimetry, after a heat-treatment at 150 °C for 15 min in an oven. The monitored heat flow (s) and heat of reaction (D) are simulated with the kinetic parameters for FGE-J2000+DPBM of Table 2 (solid lines).

FGE-Jx+DPBM

Predicted Tgel (°C)

FGE-J230+DPBM FGE-J400+DPBM FGE-J2000+DPBM FGE-J4000+DPBM

82.9 82.4 65.0 49.1

418

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

0.1

1.E+06 Frequency increases

90

(d)

60

1.E+02

30

Depth / μm

1.E+04

Loss angle / °

G' / Pa, G" / Pa

-0.2

(c)

-0.5

-0.8

T gel = 88 °C 1.E+00 60

70

80

(a)

90

100

110

-1.1

0

Fig. 5. Dynamic rheometry measurement for FGE-J4000+DPBM at 0.01 Hz, 0.3174 Hz, and 1 Hz, performed at a heating rate of 2 °C min1.

temperature of more than 75 °C for a gel conversion of 0.58 is predicted for FGE-J4000+DPBM at 2 °C min1. Simulations at reduced heating and cooling rates (e.g. 0.2 °C min1) show that the DA/retro-DA equilibrium condition for the model systems is never met at temperatures around Tgel (see Fig. 3 and Table 3). In order to validate the Tgel  xgel relationship according to Eq. (5), a rheological procedure using a stepwise heating to allow the reversible network to reach equilibrium at a certain isothermal temperature, followed by a frequency sweep, is promising [16]. This procedure is currently elaborated in ongoing research for the FGE-Jx+DPBM systems. 3.5. Sealing capacity of a model polymer network coating In order to verify the self-healing ability of the DA material, microscopic scratches were manufactured on the spin-coated FGE-J2000+DPBM model system through nanolithography [13]. The resulting micro-damage was analyzed by atomic force microscopy, revealing micro-scratches of approximately 1 lm wide and 1 lm deep. In a first experiment, the scratched coating was kept at 40 °C for 3 days (see Fig. 6).

0.3

0.0

Depth / μm

-0.3

-0.6

-0.9

-1.2

initial scratch 40 °C for 3 days

0

0.5

1

1.5

2

0

0.4

0.8

1.2

1.6

2

X-axis / μm

Temperature / °C

-1.5

(a) initial scratch (b) 70 °C for 5 min (c) 80 °C for 5 min (d) 80 °C for 10 min

(b)

2.5

3

3.5

X-axis / μm Fig. 6. AFM line-measurements of a scratched FGE-J2000+DPBM coating before (solid line) and after a thermal treatment at 40 °C for 3 days (dashed line).

Fig. 7. AFM line-measurements showing sealing of a scratched FGE-J2000+DPBM coating as a function of thermal treatment.

Out of Fig. 6, it can be concluded that no noticeable sealing of the microscopic damage occurs at 40 °C over a period of 3 days. In these temperature–time conditions, no sufficient mobility was achieved to initiate the self-healing mechanism. The experiment also rules out the possibility of elastic recovery in a time-scale of 3 days for the micro-sized defects. Fig. 7 demonstrates that as the temperature is increased to 70 °C for 5 min, the shape of the produced micro-scratch changes, indicating that local mobility is increased to initiate the self-healing process. Complete crack-sealing is achieved after 10 min at 80 °C, showing that sufficient local mobility is created and sealing of the DA coating is achieved in these conditions. If sealing can be achieved at temperatures below Tgel, the network architecture is partly maintained. This property is important for practical coating applications, guaranteeing sufficient thermo-mechanical properties of the coating on its substrate during healing. It should be noted that the used temperature–time conditions in the AFM study of Fig. 7 are indicative, as the equilibrium conversion in each sealing step is reached with delay due to the kinetics of the reversible network system (see Fig. 3). 3.6. Self-healing repeatability of a model polymer network through thermal cycling In order to verify the repeatability of the self-healing capacity of the polymer network in a thermal cycle, the FGE-J4000+DPBM model system was repeatedly studied at low temperature by micro-calorimetry after an ex situ thermal treatment at high temperature. In a first step, the upper limit of the temperature window should be determined in order to avoid undesired side-reactions in consecutive thermal cycles. At high temperatures above 150 °C, and long residence times, an irreversible maleimide homopolymerization might occur [20]. The homopolymerization sidereaction of DPBM was investigated with DSC. In Fig. 8, DPBM was heated at 2.5 °C min1 in its pure state and also in the presence of a tertiary amine catalyst, triethylamine. Fig. 8 shows that the exothermic homopolymerization reaction of pure DPBM occurs with a peak maximum around 220 °C (see Fig. 8(II)), after the endothermic melting of DPBM (see Fig. 8(I)). When triethylamine is added to DPBM in a ratio giving the same tertiary amine equivalent as in the DA mixture of FGE-J4000 with DPBM, a shift of the homopolymerization towards lower temperatures is observed, with a peak maximum around 170 °C (see Fig. 8(IV)). This indicates that the DPBM homopolymerization is

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

419

4. Conclusions DPBM

1

DPBM + triethylamine

Heat Flow / W g

-1

0.8 (I)

0.6 (II)

0.4 (III)

0.2 (IV)

endo up

0 50

100

150

200

250

Temperature / °C Fig. 8. DSC-experiment of DPBM and a mixture of 78 wt% DPBM and 12 wt% triethylamine, at a heating rate of 2.5 °C min1.

catalyzed by the tertiary amine, triethylamine. In conclusion of these measurements, a temperature of 120 °C was chosen as the upper limit of the temperature window for avoiding side-reactions during thermal cycling. In a next step, the repeatability of successive ex situ thermal treatments at 120 °C is evaluated by micro-calorimetry at 25 °C (see also Section 3.3). As the equilibrium is shifted towards the initial building blocks (FGE-J4000 and DPBM) at 120 °C, the restoration of initial properties is measured through the exothermic DA bond formation at 25 °C. In Fig. 9, the reaction rate (heat flow) and the resulting DrHpart for the DA cyclo-addition of the first and fourth consecutive healing cycle is given. Out of Fig. 9, it can be concluded that consecutive healing cycles (at 120 °C and 25 °C) are repeatable as the reaction rate and the exothermic partial heat of the DA cross-linking at 25 °C remain almost unchanged throughout the different cycles. In case of an irreversible maleimide homopolymerization at 120 °C, a decrease in both the rate and DrHpart is expected as the concentration of maleimide functional groups would decrease after each thermal treatment. The fact that the measured rate and DrHpart remain stable implies that the furan and maleimide concentrations remain constant after repeated thermal treatments at 120 °C, ruling out side-reactions and ensuring the repeatability of the self-healing system in the applied temperature window of thermal cycling.

0.E+00

-1.E-04 -1

-3 -2.E-04 -5

-3.E-04

-4.E-04 0

Δ rH / J g

Heat Flow / W g-1

-1

A reversible polymer network was formed by the DA crosslinking reaction between a synthesized furan functional compound and a bismaleimide. Using the Jeffamine D-series enables us to vary the spacer length in the furan functional compound, leading to tailor-made properties, such as cross-link density and glass transition temperature. High cross-link densities were obtained using Jeffamine D-230 (J230) and J400, forming vitrified thermosets with a minimum Tg of 55 °C. An increased spacer length leads to the formation of reversible elastomers with Tg’s well under the ambient curing temperature, completely avoiding mobility limitations during synthesis. For mainly this reason, FGE-J2000+DPBM and FGE-J4000+DPBM were chosen as model systems to characterize the experimental kinetic parameters of the reversible DA cyclo-addition. The pre-exponential factor and activation energy of both forward and reverse DA reaction were obtained through temperature controlled FTIR. The obtained kinetic parameters were validated by micro-calorimetric experiments. A simulation by means of the representative set of kinetic parameters for FGE-J2000+DPBM showed that, even at slow heating and cooling, the reversible model systems go out of equilibrium. It implies that experimental results from non-isothermal testing are not straightforward comparable to equilibrium model predictions. As an example, non-isothermal dynamic rheometry of FGEJ4000+DPBM at a heating rate of 2 °C min1 revealed a gel-point temperature (Tgel) around 88 °C, while a theoretical prediction of ca. 49 °C was made by the Flory-Stockmayer equation based on the kinetic parameters obtained by temperature controlled FTIR. Nanolithography combined with AFM was used to manufacture microscopic scratches onto the coating’s surface in order to verify the self-healing ability of the DA material. It was concluded that a complete crack-sealing is attained at 80 °C for 10 min. This implies that sufficient local mobility is created and sealing of the DA model coating can be achieved. The main condition to be met to ensure a (theoretical) infinite repeatability is that the concentration of functional groups remains constant throughout the total life-cycle of the self-healing coating. This means that no leaching or irreversible side-reaction between functional groups should occur. A micro-calorimetric study of the exothermic formation of covalent bonds through the forward DA reaction led to the conclusion that an irreversible side-reaction in the FGE-J4000+DPBM model system can be ruled out when healing temperatures of 120 °C are not exceeded. In order to design a repeatable procedure for optimized sealing with restoration of initial properties after damage, controlled temperature–time profiles for an accurate determination of sealing/ healing against Tgel and xgel are important. In this respect, kinetic/ thermodynamic modelling in combination with dynamic rheometry, a damage procedure with controlled scratch geometry, and temperature controlled AFM protocols will be elaborated in future work.

-7

12

24

36

48

60

-9 72

Time / hours Fig. 9. Repeatability of healing procedures of FGE-J4000+BPBM at 120 °C studied by micro-calorimetry at 25 °C. For clarity, only the 1st (dashed) and 4th (solid) cycle are presented.

Acknowledgement The ‘Instituut ter bevordering van het Wetenschappelijk Onderzoek en de Innovatie van Brussel’ (INNOVIRIS) is gratefully acknowledged for financial support in the framework of the ‘Impulse Programme-Environment 2008’. References [1] D.Y. Wu, S. Meure, D. Solomon, Prog. Polym. Sci. 33 (2008) 479–522. [2] R. Benthem, W. Ming, G. With, Self healing polymer coatings, in: S. Van der Zwaag (Ed.), Self Healing Materials, Springer, The Netherlands, 2008, pp. 139– 159. [3] S.H. Cho, S.R. White, Adv. Mater. 21 (2009) 645–649.

420

G. Scheltjens et al. / Reactive & Functional Polymers 73 (2013) 413–420

[4] S.J. Garcia, H.R. Fischer, S. van der Zwaag, Prog. Org. Coat. 72 (2011) 211–221. [5] S.K. Ghosh, Self-healing materials: fundamentals, design strategies, and applications, in: S.K. Ghosh (Ed.), Self-Healing Materials, Wiley-VCH, Weinheim, 2009, pp. 1–28. [6] S.D. Bergman, F. Wudl, J. Mater. Chem. 18 (2008) 41–62. [7] K.A. Williams, D.R. Dreyer, C.W. Bielawski, MRS Bull. 33 (2008) 759–765. [8] E.B. Murphy, F. Wudl, Prog. Polym. Sci. 35 (2010) 223–251. [9] X. Chen, M.A. Dam, K. Ono, A. Mal, H. Shen, S.R. Nutt, K. Sheran, F. Wudl, Science 295 (2002) 1698–1702. [10] Y.L. Liu, C.Y. Hsieh, Y.W. Chen, Polymer 47 (2006) 2581–2586. [11] M. Wouters, E. Craenmehr, K. Tempelaars, H. Fischer, Prog. Org. Coat. 64 (2009) 156–162. [12] G. Scheltjens, J. Brancart, I. De Graeve, B. Van Mele, H. Terryn, G. Van Assche, J. Therm. Anal. Calorim. 105 (2011) 805–809.

[13] J. Brancart, G. Scheltjens, T. Muselle, B. Van Mele, H. Terryn, G. Van Assche, J. Intel. Mater. Syst. Str., in press [14] S.F. Parker, Spectrochim. Acta A. 63 (2006) 544–549. [15] S. Vyazovkin, A.K. Burnham, J.M. Criado, L.A. Pérez-Maqueda, C. Popescu, N. Sbirrazzuoli, Thermochim. Acta. 520 (2011) 1–19. [16] B.J. Adzima, H.A. Aguirre, J.C. Kloxin, F.T. Scott, N.C. Bowman, Macromolecules 41 (2008) 9112–9117. [17] P.J. Flory, J. Am. Chem. Soc. 63 (1941) 3083–3090. [18] W.H. Stockmayer, J. Chem. Phys. 11 (1943) 45–55. [19] J.C. Ortiz-Cisneros, F. Belmontes, C.A. Mendose, L.F. Ramos de Valle, Dynamic mechanical analysis of natural rubber crosslinking reaction kinetics, in: Society of Plastics Engineers, , 2011 (accessed 07.11). [20] M.J. Grenier-Loustalot, J. Da Cunha, Eur. Polym. J. 34 (1997) 95–102.