Impact of Butyl Glycidyl Ether Comonomer on Poly(glycerol–succinate

Jan 25, 2017 - Impact of Butyl Glycidyl Ether Comonomer on Poly(glycerol–succinate) Architecture and Dynamics for Multifunctional Hyperbranched Poly...
0 downloads 9 Views 4MB Size
Article pubs.acs.org/Macromolecules

Impact of Butyl Glycidyl Ether Comonomer on Poly(glycerol− succinate) Architecture and Dynamics for Multifunctional Hyperbranched Polymer Design Jean-Mathieu Pin,† Oscar Valerio,†,‡ Manjusri Misra,†,‡ and Amar Mohanty*,†,‡ †

Bioproducts Discovery and Development Centre, Department of Plant Agriculture, and ‡School of Engineering, University of Guelph, Guelph, Ontario N1G 2W1, Canada S Supporting Information *

ABSTRACT: An original strategy is proposed to easily design functional materials from poly(glycerol−succinate) (PGS). This approach consists in the introduction of an epoxidized functional agent during the polyesterification between the glycerol and succinic acid. In order to model the effect of this epoxide group on the polymerization process and its resulting hyperbranched architecture, the butyl glycidyl ether (BGE) has been selected as comonomer agent. The theoretical potential reactions have been confronted with the topological units revealed by 2D NMR correlations. The regioselectivity against the primary alcohol and the stoichiometric balance of the system have been modified in situ by the kinetic control of parallel reactions. This had the effect to delay the gelation and increase the polyesterification conversion. The resulting hyperbranched polymers (HBPs) obtained just after gelation exhibit a temperature of glass transition (Tg) of −3.9 °C for PGS and −16.1 °C for poly(glycerol−succinate-co-butyl glycidyl ether) (PGS-co-BGE). This difference was explained by the BGE butyl tails effect which plays the role of dynamic spacer between the polymer chains during the relaxation process. The relaxation processes were investigated by the computation of the effective activation energy (Eα) through the Tg using the advanced isoconversional method and by the estimation of the βrelaxation activation energy (Eβ) by means of annealing experiments. The variation of Eα and Eβ values was discussed in terms of competition between the cooperative/noncooperative segment motions and the hindrance effect of the hydrogen-bonded network. The dynamic behavior of this system can be potentially generalizable to all the plastic glass containing a critical amount of secondary interactions.



INTRODUCTION From the primary interest of modeling natural polymers such as starch, the first exploratory work investigated by Staudinger,1 Flory,2,3 Stockmayer,4 and others on both experimental design and statistical modeling of branched polymers established the fundamental understanding of cross-linked architectures.5 The control of the fine topology of hyperbranched polymers6 (HBPs), which possess an architecture between linear and dendritic, has been one of the researchers’ main leitmotiv from the two past decades. This control is challenging in several aspects. Main efforts were done in order to prevent gelation by quenching or slow monomers addition. The limitation of intramolecular cyclization was achieved by using rigid synthons or working in melt polymerization conditions.7,8 Besides the synthesis aspects, a great focus was put to understand the physical and rheological behavior of strongly hydrogen-bonded polymers.5 It was shown that the hydrogen bonds can be able to control the rheological behavior of polymers in the melt.9 This first attempt has been extended to the case of highly © XXXX American Chemical Society

hydroxyled HBPs. It has been discovered that the glass transition temperature (Tg) is conditioned by the branching degree and the number/nature of end-groups in polymer chains.10 Alongside the fundamental aspect of matter organization understanding, HBPs were revealed to be a fascinating platform for the design of smart materials. By using functional molecules as monomeric synthon, several materials have been created with nonlinear optical (NLO) properties11 or for sensor purposes.12,13 Being easier to synthesize than their dendritic homologues, HBPs have been successfully investigated as cargo guest molecules for drug delivery systems.14,15 Since the original work of Wang et al.16 highlighting the biocompatibility of poly(glycerol−sebacate), several research groups have put efforts into the synthesis and characterizations of HBPs from Received: November 8, 2016 Revised: January 3, 2017

A

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 1. Genealogical units topology associated with the cascade of possible reactions related to the first to third generations.

glycerol and dicarboxylic acids.17 Depending on the carboxylic acid length, the resulting polymer can be either amorphous (succinic)18,19 or semicrystalline (sebacic).16,20 Different synthesis strategies for the polyesterification have been achieved, using solvents such as toluene,21,22 DMF, and DMSO, or melt condition polymerization.23 Several catalysts such as Ti(OC4H9)4,24 and dibutyltin oxide,25 or enzyme catalysts26,27 as well as different stoichiometry and experimental conditions have been tested to orient the branching topology.5 In this paper we propose to explore the fundamental aspects related to the chemistry and physics of polymers for a new approach to design functional HBPs. This strategy consists in the introduction of an epoxidized “active” agent during the polyesterification between glycerol (G) and succinic acid (SA). A myriad of molecules of interest can be easily epoxidized by using different chemical treatments such as epichlorohydrine, for instance. Furthermore, the epoxide ring-opening reaction from the glycidyl motif leads to generate the same structure as the glycerol ester derivatives which are known to be biocompatible. Hence, this strategy can have a good potential for drug delivery applications. Butyl glycidyl ether (BGE) has been chosen as model molecule to mimic the introduction of a potential epoxidized functional molecule in the polymeric system. The experimental design has been established according to the literature. A stoichiometric ratio has been chosen between G and SA, considering a functionality of two for both of them, in order to reach a maximal molecular weight.5 The melt polymerization condition was selected to avoid the intracyclization.8 The temperature of 150 °C has been preferred because it is high enough to be in the melt condition

and low enough to minimize secondary reactions such as alcoholysis.28 The BGE has been added in the polymeric media at 3 h of reaction to reduce the reaction probability with the monomers. Figure 1 depicts the possible cascade of reactions involved by the addition of BGE in the G/SA system for the three first generations and the respective theoretical branching topology from the glycerol moieties. The first generation is associated with the PGS architecture with five different types of branching. The second generation is related to the BGE addition. The versatility of the epoxide groups allows its reaction with both carboxylic acid and alcohol function. From the attack in position one or two of the oxiran ring, 17 possibilities of branching can be estimated. In this case, the BGE agent acts as simple grafting agent. Depending on the amount of BGE and the kinetics of the reactions in competition, a third generation of reaction set can occur. These reactions are associated with the hydroxyl group generated by the epoxide opening ring. This −OH can reacts with both terminal −COOH and another molecule of BGE. Depending on the epoxide selectivity, 165 theoretical branching forms can be counted. In this context, the carboxylic acid group attack allows the previously grafted BGE to act as branching agent. Beside these main reactions, several secondary reactions can occur, generating other ways of branching: epoxide homopolymerization and reaction between epoxide group and water. However, these two reactions can be considered as marginal because of the relative low synthesis temperature and the Dean−Stark system under nitrogen, respectively.29,30 B

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Molecular weight determinations were performed using gel permeation chromatography in a Viscotek GPCmax (Malvern Instruments, UK) equipped with a refractive index detector. Two Styragel columns (HR1 and HR2, Waters Corporation, USA) and one PLgel-Mixed E column (Agilent Technologies, USA) connected in series. Tetrahydrofuran was employed as solvent at a flow rate of 0.3 mL/min and a temperature of 40 °C. Prior to injecting PGS samples a calibration curve was constructed using a series of eight poly(ethylene glycol) standards with molecular weights (Mn) in the range 106− 21 300 Da (Easy Vial, Agilent Technologies, USA). PGS samples were fully dissolved in tetrahydrofuran at a concentration of 10 mg/mL and filtered through a 0.2 μm pore diameter filter prior to injection. Viscosity measurements were performed on a cone plate rheometer (CAP 2000+, Brookfield, USA) at a temperature of 100 °C and a fixed shear rate of 100 s−1. PGS samples were placed directly onto the preheated rheometer plate and allowed to equilibrate at 100 °C. Subsequently, the samples were tested over a period of 30 s to allow stabilization of the viscosity measured. Stress relaxation experiments were completed on an Anton Paar MCR 302 rheometer. The measurements were performed on a parallel plate discs with a diameter of 25 mm and a gap of 1 mm. An initial stress equivalent to 10% of strain was applied to the polymers at the constant temperature of 30 °C under an air atmosphere. Four cycles of 1120 s were done to estimate the residual stress. Differential scanning calorimetry (DSC) measurements were conducted by using a DSC 200 from TA Instruments. Around 8 mg of sample was introduced in a 40 μL aluminum pan, and the experiments were run under 50 mL min−1 of nitrogen flow. Taking into consideration the thermal history, all the samples for the Tg measurements were cooled from 40 to −70 °C at 5 °C min−1; then after an isothermal step of 5 min at −70 °C, the samples were heated up to 60 °C with three different heating rate of 10, 20, and 40 °C min−1 in order to apply the advanced isoconversional kinetic method.35 Kinetics Methods. Evolution of the Activation Energy through the Glass Transition. Vyazovkin et al.31 proposed a method to evaluate the variation of the apparent activation energy (Eα) through the glass transition. This method consists in the computation of the normalized heat capacity (CNp ) in a model free way by applying the advanced isoconversional method. As depicted in eq 1, the CNp curves were obtained by normalization of the Cp between its two extreme values corresponding to the glassy Cpg (Cp glass) and the liquid state Cpe (Cp equilibrium).36 The resulting CNp curves, assimilated to a degree of conversion α, obtained from the heating rates of 10, 20, and 40 °C min−1 for both PGS and PGS-co-BGE have been used to determine the apparent activation energy. The Eα dependency, independently calculated for each α, is got by minimizing the function presents in eq 2 using a nonlinear minimization algorithm: n is equal to three here and represents the number of different heating rates; J is related to eq 3.

All the reactions previously described present different feasibilities and kinetics. Herein, the first part of the investigation will be dedicated to chemistry of this system. The determination of the predominant reactions which drive the overall system among the theoretical possibilities exposed in Figure 1 will be investigated by nuclear magnetic resonance (NMR). The BGE addition effect will be highlighted for the polyesterification regioselectivity and conversion. Then the branching degree and the macromolecular growing associated with the viscosity of the system will be discussed with the help of correlation results from NMR, gel permeation chromatography (GPC), and viscosimetry. The second part will be allocated to the relationship between the PGS and the poly(glycerol−succinate-co-butyl glycidyl ether) (PGS-co-BGE) architectures and their relaxation processes. The variation of the apparent activation energy during the glass transition (Tg) has been calculated via the advanced isoconversional method applied to differential scanning calorimetry (DSC) measurements.31 The departure from the Arrhenius behavior through the transition has been discussed in terms of hydrogen-bonding effects and polymer segmental motion dynamics. In order to have a more in-depth explanation of the glass transition relaxation process which could be associated with cooperative and noncooperative motions, annealing experiments at temperature far below Tg were done to determine the activation energy associated with only noncooperative motion, also called the Johari−Goldstein32 or β-relaxation process.33 These results were confronted to the Angell34 classification of polymeric glass to determine their degree of fragility, i.e., estimating their structural capability to resist against a stress. In the last section the relaxation behavior investigation of these HBPs has been extended beyond Tg by means of rheological measurements.



MATERIALS AND METHODS

Tetrahydrofuran (THF) with a purity of 99.8% and succinic acid (SA) (purity ≥99.0%) were purchased from Fisher Scientific and KIC Chemicals, respectively; glycerol (G) (purity ≥99.0%) and butyl glycidyl ether (BGE) (purity 95%) were obtained from Sigma-Aldrich. The chemicals were used as received without further purification. All NMR experiments were recorded in DMSO-d6. The residual solvent signals at δ = 2.50 and 40.0 ppm for 1H and 13C nucleus, respectively, were used as a standard reference. Experiments were collected on a Bruker AVANCE III spectrometer with a 1H operating frequency of 600.0 MHz, using a 5 mm TCI cryoprobe. The sample temperature was regulated at 22 °C. 13C spectra (150 MHz) were acquired with a pulse angle of 45°, ∼2700 transients, and a relaxation delay of 20 s, for a total experiment time of ∼16 h. A separate 13C T1 measurement was performed on one representative sample to ensure 20 s was sufficient time for >99% relaxation of 13C signals between transients. Proton decoupling was active during FID acquisition to provide decoupled 13C peaks but was inactive during the relaxation delay to prevent NOE enhancements.23 For HSQC spectra, an “edited” pulse sequence was employed, such that CH2 groups have opposite phase with respect to CH and CH3 groups. Generally, spectra were collected with 1024 increments spanning 200 ppm in the indirect 13 C dimension, 8 scans per increment, and a relaxation delay of 1.5 s, for a total experiment time of 3.75 h. HMBC spectra (8 Hz coupling) were collected with 1324 increments spanning 285 ppm in the indirect 13 C dimension, 16 scans per increment, and a relaxation delay of 2.5 s, for a total experiment time of 16 h. A Thermo Scientific Nicolet 6700 Fourier transform infrared spectroscopy (FTIR) was used under attenuated total reflection mode. All the spectra have been recorded using 64 scans with a resolution of 2 cm−1, and the background was done with air.

CpN =

(Cp − Cpg)|T (Cpe − Cpg)|T n

Φ(Eα ) =

n

(1)

J[Eα , Ti(tα)] J[Eα , Tj(tα)]

(2)

⎡ − Eα ⎤ exp⎢ ⎥ dt ⎣ RT (t ) ⎦ α −Δα

(3)

∑∑ i=1 j≠i

J[Eα , T (tα)] ≡

≡α

∫t



The DSC measurements as well as the kinetic computation have been achieved by following the ICTAC recommendation.37 Activation Energy of the β-Relaxation. From the first observation of Illers,38 who detected a small endothermic peak (before Tg) after reheating poly(vinyl chloride) annealed below the Tg, Chen39 as well as Bershtein et al.40 suggested to use the annealing temperature peak shift (Tp) obtained from different heating rates (q) to estimate the activation energy of this sub-Tg relaxation (eq 4). By using annealed temperature far below Tg, Bershtein et al.40 and Vyazovkin et al.41 C

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 2. 13C NMR investigation: confrontation of PGS spectra at 7 h with PGS-co-BGE after the addition of BGE at 3.25 h until the end of the reaction at 12.25 h (A). HSQC correlation of PGS-co-BGE for the sample at 12.25 h. The purple square discriminate the 1, 2, and 3 zones where the methynes surrounded in red are respectively in the α-position from −OA, −OH, or −OE (B).

Figure 3. HSQC and HMBC spectra of PGS-co-BGE highlighting the 1JCH, 2JCH, and 3JCH coupling of the methyne hydrogen (surrounded in red) from the unit that consider the SA link to the secondary alcohol of the glycerol (A); HSQC spectrum of SBGE model with the related structures a, b, c, d, and e (B). associated this sub-Tg relaxation to the β-relaxation process, highlighting good correlations between DSC and traditional dielectric or dynamic mechanical technics results. Eβ = − R

cooled down to −85 °C and then heated at different temperature rates of 5, 7.5, 10, and 20 °C min−1 to 60 °C. PGS and PGS-co-BGE Synthesis in Melt Conditions. Poly(glycerol−succinate) (PGS). A stoichiometric ratio (0.97 mol) of glycerol (90 g) and succinic acid (115 g) has been used considering a functionality of two for both of them. The reactants were loaded in a 1000 mL reactor under a constant mechanical stirring of 100 rpm and nitrogen flow. The temperature was adjusted to 150 °C, and the formed water was collected using a Dean−Stark apparatus. The reaction was stopped at 7.18 h, just after gelation. Yield: 87.8%. Poly(glycerol−succinate-co-butyl glycidyl ether) (PGS-co-BGE). 28.2 g of butyl glycidyl ether (0.21 mol) was added dropwise in 10 min at 3 h of the PGS synthesis, respecting the conditions detailed above.

d ln q dTp−1

(4)

In this particular case, the DSC method used is associated with a first ramp to 80 °C; then after an isotherm of 5 min, the DSC sample was quenched in liquid nitrogen and then quickly put back in the DSC at the temperature of −65 °C for 3 h of annealing. This procedure was chosen in agreement with the literature: the fast rate quenching allowing shorter annealing time.39,42 After annealing, the system was D

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules The final molar ratio between the glycerol, succinic acid, and butyl glycidyl ether was 1:1:0.22, respectively. The reaction was stopped at 13.25 h after the gelation. Yield: 87.7%. Samples were collected and quenched during the synthesis every 30 min for the kinetic investigations and characterizations. It is important to precise that the t = 0 min was taken when the reactor reached the temperature of 150 °C, which takes 30 min from room temperature. Branching Molecules Model Synthesis. Glycerol and Butyl Glycidyl Ether (GBGE). 10 g of butyl glycidyl ether (7.68 × 10−2 mol) and 3.59 g of glycerol (3.84 × 10−2 mol) were mixed under stirring at 150 °C for 2 h. The stoichiometry was adjusted regarding a functionality of two for glycerol and one for butyl glycidyl ether. Succinic Acid and Butyl Glycidyl Ether (SBGE). 10 g of butyl glycidyl ether (7.68 × 10−2 mol) and 4.53 g of succinic acid (3.84 × 10−2 mol), regarding a functionality of one and two, respectively, were mixed under stirring at 150 °C for 2 h. The evaluation of the gel content has been done by swelling PGS and PGS-co-BGE polymers in THF for 48 h, then the sol content was removed with the solvent, and the gel was dried by heating at 60 °C under vacuum for 12 h. The sample weight for the gel content calculation was measured by a Mettler Toledo Excellence plus microbalance. Three replicates were made for the standard deviation.

BGE (5.00; 70.8 ppm), (4.90; 72.1 ppm), and (4.78; 74.3 ppm). The HMBC spectra of PGS-co-BGE in Figure 3A show the 2J coupling associated with the methylene in α-position of the methyne (surrounded in red). Three cross-peaks areas (in the black squares) have been assigned to the methylene link to −OA, −OH, or −OE. Finally, the 3J coupling associates the carbon ester from the secondary alcohol to the methyne proton. Based on these correlations, the units LA1,2E3/TA2E1,3/ TA2E1 predicted from the second generation seem to be found. However, the determination of these units is not that obvious. The HSQC spectrum of SBGE model (Figure 3B) exhibits also three cross-peaks in the same region 1, with chemicals shifts close from that we found for the PGS-co-BGE: (5.08; 70.4 ppm), (4.98; 71.9 ppm), and (4.85; 74.0 ppm). These crosspeaks can be associated with the methyne surrounded in red in Figure 3B related to the structures c, b, and a, respectively. The structure a should correspond to the SA attack on position 2 of the epoxide ring. The motif b can be assimilated to the attack of an −OH group on the epoxide in position 1, −R corresponding to the hydroxyl generated from an epoxide opening ring. Finally, the structure c can be associated with the attack of −COOH group on the two available sites of the functional epoxide. From these results, the three cross-peaks found in the PGS-co-BGE HSQC spectra could also be associated with the motifs a, b, and c. Because of the very high similarity between the structures from the glycerol and those from the epoxide ring-opening, it is delicate to establish a conclusion. As a result, the BGE is grafted on the PGS oligomers but also could participate in the branching by its integration in the main backbone architecture. In order to generalize, the −OE radical should correspond to the methylene link to an ether which can be associated with either the glycidyl or the tail part of the BGE. In zone 2, besides the methyne cross-peaks corresponding to the PGS units LA1,3 (3.80; 66.6 ppm) and TA1 (3.56; 69.8 ppm), two new main cross-peaks appear at (3.70; 67.8 ppm) and (3.61; 69.1 ppm). The HSQC of the SBGE mix exhibits two similar cross-peaks in the zone 2: (3.77; 67.6 ppm) and (3.66; 69.1 ppm), while for GBGE two cross-peaks appear at (3.68; 69.0 ppm) and (3.56; 70.8 ppm) (presented in the Supporting Information). One of these cross-peaks (3.61; 69.1 ppm) appears in both SBGE and GBGE cases. Thus, it could be associated with the structure d related to the attack of a hydroxyl in the position 1 of the BGE. In other words, this motif should corresponds to an −OH attack from glycerol for GBGE or the −OH attack from the BGE opened ring for SBGE. Then the cross-peak at (3.70; 67.8 ppm) for the PGS-coBGE corresponds to the motif e illustrated in Figure 3B. This hypothesis is confirmed by the GBGE HMBC spectrum (available in the Supporting Information) where the methyne proton at 3.68 ppm is coupled with only one carbon in 2J at 72.7 ppm. This unique peak is explained by the pseudoisochrony of the methylenes possessing a similar ether environment. As explained below, due to the very similar environment of the ether peak, the methyne cross-peak of PGS-co-BGE could also be associated with the primary or secondary alcohol from glycerol oligomers attack on the BGE. Another piece of information can be extracted from the inverse gated 13C NMR by relative integration. The 13C methyne peak corresponding at the structure e can be compared to the peak at 14.3 ppm (peak related to the BGE butyl tail methyl). Then, it appears that the motif e represents around 34% of the total possible motif issue from the BGE consumption (Supporting Information). This



RESULTS AND DISCUSSION Effect of the BGE Comonomer on the Polymerization Chemistry and Branching Topology. Figure 2A shows the 13 C NMR peaks evolution during the polymerization of PGSco-BGE in comparison to PGS. The PGS spectrum in pink represents the last sample before gelation at 7 h of reaction. The methyne peaks from the glycerol units, presented in the first generation of Figure 1, were assigned in agreement with the literature.23,20 The 13C chemical shifts are TA1 (69.8 ppm)/ TA2 (76.4 ppm)/LA1,2 (72.8 ppm)/LA1,3 (66.6 ppm)/DA1,2,3 (69.5 ppm). After the BGE addition at 3 h, a first spectrum was recorded at 3.25 h. The apparition of the peaks at 70.7 and 71.7 ppm are related to the methylene in the α-position of the BGE ether oxygen. The disappearance of these peaks at 5.25 h of reaction suggests the complete BGE consumption during this period of 2 h, explained by the epoxide ring-opening reaction which changes the near environment of these methylenes. Furthermore, the emergence of new peaks in this region attests of the BGE reaction with the PGS oligomers. In the next section the attention will be paid on the structural elucidation by means of 2D NMR to highlight the main reactions involving the epoxide group. Figure 2B exhibits the HSQC spectrum of the last PGS-coBGE sample before gelation (12.25 h). In comparison with the HSQC spectra of PGS (7 h) presented in the Supporting Information, a lot of cross-peaks appear, fruit of the myriad of possible cross-linking as exposed below. In order to facilitate the interpretation demarche, three zones were revealed: the different cross-peaks associated with the methynes from the glycerol structural scheme in α-position of the carboxylic acid group for the zone 1 (−OA), hydroxyl group for the zone 2 (-OH), and ether group for the zone 3 (−OE). The structural elucidation has been helped by the literature20,24 and model reactions of both SA + BGE (SBGE) and G + BGE (GBGE). After this cross-peaks partitioning, zone 1 was investigated. Figure 3A exhibits the coupling correlation between the different methyne hydrogens (encircled in red) linked to the ester bond (−OA) and these carbon neighbors in 1J, 2J, and 3J coupling positions. The motif units only related to the reaction between G and SA are confirmed to be assigned to DA1,2,3/ LA1,2/TA2 units. Three new cross-peaks (highlighted in black in Figure 2B, zone 1) appear associated with the co-reactions of E

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 4. Evolution of unit population during the polyesterification for PGS (A) and PGS-co-BGE (B); evolution of the regioselectivity at the primary alcohol site during the polyesterification calculated from eq 5 (C).

carboxylic acid/primary alcohol/secondary alcohol as well as the plural reactivity associated with the polyesterification with regard to the primary and secondary alcohol site. However, we assume that the apparent kinetic follow the general trend of kepoxide reactions > kpolyesterifications, k being the rate constant. The evolution of the population units associated only to the polyesterification (Figure 4B) for the PGS-co-BGE should indirectly reflecting the impact of the epoxide reaction. The comparison of the regions between 3 and 5 h for PGS with 3.25 and 5.25 h for PGS-co-BGE highlights a significant difference. The relative decrease of LA1,3 and the increase of LA1,2 unit (1.9% for PGS and 4.7% for PGS-co-BGE) imply a rise of the secondary alcohol selectivity for the polyesterification. From this initial observation, a perturbation in the regioselectivity has been pointed out. Based on eq 5, the polyesterification selectivity evolution was calculated based on the unit topology proportion for the overall polymerization process.43

value seems to be constant along the reaction from 5.25 to 12.25 h. Instead of zone 2 where the methyne supporting the secondary alcohols are transformed in an ester group, zone 3 could be associated with the motif where the secondary alcohols are converted to an ether group. This conversion can occur through the reaction between the secondary alcohol from glycerol and the epoxide in position 1 or by the attack of an alcohol function in position 2 of the epoxide. However, the 13C peaks intensities from zone 3 are very weak which make the integration impossible. Herein, units populations from an ether group attached in the secondary alcohol site are considered to be negligible. After the possible unit identification from the BGE addition on the PGS synthesis, we have been interested in the polyesterification kinetics. Figure 4A,B compares the evolution of the branching unit population from G pattern determined by inverse gated 13C NMR spectroscopy. The general PGS evolution of the different units is in good correlation with what have been found in the literature. The increase of the extent of conversion is related to a decrease of the monosubstituted terminal units TA1,2 and TA1,3 while the disubstituted LA1,2, LA1,3, and DA1,2,3 increase. Some deviations from this general tendency occur with the addition of the BGE. The region between 3.25 and 5.25 h is particularly interesting because it is the region where the BGE is reacting and fully consumed. In this specific zone, two mains reactions occur in parallel: the polyesterification and the epoxide reaction. In sensu stricto more than two reactions occur in the meantime: those associated with the epoxide with

primary alcohol % − regioselectivity =

2[LA1,3] + 2[DA1,2,3] + [LA1,2] + [TA1] 2[LA1,3] + 3[DA1,2,3] + 2[LA1,2] + [TA1] + [TA2]

× 100 (5)

The resulting graphic exhibited in Figure 4C clearly shows the global reduction of the selectivity at the primary alcohol site for the PGS-co-BGE polyesterification, starting from the BGE addition. This difference can be first associated with a faster reaction of BGE with the primary alcohol than the esterification and then, after BGE consumption, to a variation in the balance between the amount of primary and secondary alcohol site F

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 5. Relative amount of carboxylic acid function determined by 13C NMR integration as a function of time of reaction associated with the FTIR spectra of PGS and PGS-co-BGE polymer at the end of the reaction (A); the evolution of degree of branching during the reaction and in the inset the linear correlation between the relative amount of free carboxylic acid group vs the branching degree (B).

Figure 6. GPC chromatograms of PGS (A) and PGS-co-BGE (B) as a function of time of reaction; evolution of the average molar mass in weight (Mw) and the related polydispersity index (PDI) in the inset (C); evolution of viscosity during the synthesis of PGS and PGS-co-BGE (D).

available through the reaction conversion. Indeed, as seen in the previous section, the main reactions associated with the BGE are related to the attack of −COOH or −OH in position 1 of the epoxide ring, leading to the generation of a secondary alcohol. The difference in the reaction time, from 7.18 h for PGS to 13.25 h for PGS-co-BGE, corroborates the previous observation on the relative increase of the secondary alcohol amount. The reaction rate between the carboxylic acid and the

secondary alcohol is slower than the carboxylic acid with the primary alcohol. Beyond 5.25 h, when all the BGE has already reacted, the polyesterification becomes again the only reaction controlling the overall kinetics. The evolution of the different units reflecting that we observed for the secondary alcohol selectivity: LA1,2, TA2, and DA1,2,3 (from LA1,3 consumption) species being G

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules Table 1. Physicochemical Characteristics of the Resulting Polymers PGS PGS-co-BGE

Mna [g mol−1]

Mwa [g mol−1]

PDIa

Tgb [°C]

gel content [%]

ΔE [kJ mol−1 K−1]

m

Eβ [kJ mol−1]

1170 ± 107 1659 ± 87

4191 ± 619 19275 ± 621

3.6 ± 0.2 11.6 ± 1

−3.9 ± 0.1 −16.1 ± 1.4

30.8 ± 1.1 6.0 ± 1.4

−28.8 −19.5

130.8 119.2

≈256 ≈120

a The values of Mn, Mw, and PDI correspond to the last sample before the gelation of the system, 7 h and 12.25 h for PGS and PGS-co-BGE, respectively. bThe Tg values were measured for a heating rate of 10 °C min−1.

The previous highlights about the reactivity and topology will be confronted to the macromolecular growing, from the dimeric and sol species to the gelation of the system. Figure 6A,B exhibits the chromatograms related to the evolution of the different oligomers populations during the reaction. The PGS chromatogram reflects the monomers consumptions (retention volume of 10.1 and 10.4 mL for SA and G, respectively) as well as the polyesterification conversion by the decrease of low molecular weight oligomers and increase of the high molecular weight population. However, a broad oligomers population is represented at the end of the reaction as the literature shown for HBPs issue from G.25,43 The same tendency is observed for the PGS-co-BGE chromatogram’s evolution. The peak related to the BGE comonomer added at 3 h can be easily discriminated from the other monomers with a retention volume of 11.1 mL. After the BGE consumption, the main difference is associated with the growing of oligomer populations with higher molecular weights (retention volume of 6.4 mL) highlighting the effective reaction of BGE with PGS oligomers. The values of the different molecular mass in weigh (Mw) and number (Mn) are available in Table 1 for the last sample of PGS and PGS-co-BGE before gelation. If we look at the Mw first, the value is about 4.6 times higher for the PGS-co-BGE in comparison to the PGS (from 4191 to 19 275 g mol−1). It can be both the effect of the BGE addition and/or the polymerization degree increase. However, the Mn increase is in the range of 1.4 times higher (from 1170 to 1659 g mol−1). This disparity suggest that the BGE has been well added on the different oligomers population, but it does not significantly affect the polymerization extent. The Mn increases should be mostly associated with the higher extent of polymerization explained by the higher consumption of the carboxylic group with the glycerol moieties as explained above. The melt polymerization condition supposes a determining role of the viscosity in the reactive system. The confrontation of the Mw and the medium viscosity evolution during the polymerization (Figure 6C,D) highlights their importance and interdependency. The PGS Mw shows a constant increase until around 6 h of reaction and then reaches a plateau. This observation is correlated to the high increase of the media viscosity at 6 h associated with the gelation phenomenon.45 The gelation hampers the population of oligomers to grow: the intra-sol reactions being statistically reduced with regard to the sol/gel reaction.46 This Mw plateau is not observed in the PGSco-BGE case because the gelation is delayed. Indeed, a same viscosity of about 175 P was measured for reactions time of 6.5 h for PGS and 12.25 h for PGS-co-BGE. Furthermore, the gelation kinetic of PGS-co-BGE seems much lower as observed by the curves slopes. One direct consequence of this kinetic is the measurement of the gel content, which is higher in the case of the PGS as seen in Table 1. This macromolecular insight ends the investigation in the effect of BGE to the polymer chemistry of the system. In the following part, the discussion will focus on the influence of the

favored with regard to the PGS units evolution at the end of the reaction. Once this first clarification on the effect of the BGE addition on the polyesterification regioselectivity was done, the investigation had focused on the carboxylic acid group reactivity. Figure 5A exhibits the relative carboxylic acid function consumption during the polycondensation. These data were measured by relative integration of the 13C NMR signal. The peak massif around 29.0 ppm which correspond to the two methylene of SA moieties (associated with the free SA and the SA which had reacted with the glycerol) was compared to the peaks associated with quaternary carbon of both carboxylic acid group from free SA (174.2 ppm) and from the polymer terminal group (peak massif around 173.9 ppm). The first observation is related to the increase of the −COOH group consumption: ≈8% of groups left for the PGS over ≈2% for PGS-co-BGE. This result is relayed by the FT-IR spectra which highlight the disappearance of the broad peak in the region around 3250 cm−1 associated with the stretching of the carboxylic acid −OH. This information can be correlated to the −COOH consumption by BGE and apparent gelation delay. Indeed, in the same manner that the regioselectivity has changed, the global stoichiometry between alcohol and acid has evolved toward an excess of alcohol after the BGE consumption. Figure 5B presents the evolution of the branching degree (DB) during the polymerization, calculated from the Frey44 equation: degree of branching (DB) =

2[DA1,2,3] 2[DA1,2,3] + [LA1,2] + [LA1,3] (6)

After a brutal increase of the PGS degree of branching during the first hour of the polymerization, the DB evolution tends to reach a plateau. The same tendency is observed for PGS-coBGE. It seems that the curve is in the continuity of the PGS data. From the branching point of view, it appears that the BGE addition has a minor “apparent” effect. Consequently, the question of the BGE acting as a grafting and/or branching agent can be arising. At the molecular level, the NMR exploration showed the possibility of the BGE to participate in the branched network through the hydroxyl formed by the epoxide ring opening. Its influence seems to not be significant in the overall branching architecture topology. However, this possible participation should tend to a slightly decrease of the observed DB due to an increase of the linear moieties in comparison to the dendritic one as a consequence of the BGE bifunctionality. The inset in Figure 5B exhibits a great linearity (correlation of 0.994) between the carboxylic acid consumption and the branching degree evolution. This straight line was built from PGS and PGS-co-BGE data, confirming a good continuity in the polymerization process. Subsequently, this linearity can be an indication of the main dedication of the carboxylic group to the branching through the esterification of SA and glycerol moieties. H

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 7. DSC thermograms used for the normalized heat capacity (CNp ) (A) at 20 K/min and apparent activation energy (Eα) associated with the glass transition from the glassy to the liquid state (B). A putative scheme of the polymers architecture is depicted to highlight the hydrogen-bonding interactions.

delayer. Furthermore, in the case of HBPs, several deviations occur from the formalism exposed above. As a first attempt, Sunder et al.50 have shown that the glass transition of highly hydroxylated HBPs is affected by two main factors which are the hydrogen bonding of the terminal groups and the proclivity to generate ordered phase as meso- or crystalline phases. Because the presented HBPs are fully amorphous, the main factor impacting the Tg value should be mostly associated with the hydrogen bonding. However, other studies51,10 highlighted a non-negligible influence of the branching degree on the Tg value, especially for polar −OH-terminated HBPs: below a DB of 25% the Tg increases with the DB, but beyond this value, the Tg decreases with it. The polymers exhibit a DB of 37% for PGS and 42% for PGS-co-BGE. If we compare with the previous studies, this difference of 5% in the DB represents a variation of the glass transition of around 5 °C. The observation of a difference of 12.2 °C is an indication that the branching degree is probably not the major factor governing the Tg value in the present case. After this first qualitative explanation on the intrinsic value of the glass transition, a more in-depth investigation on its mechanism will be explored. The glass transition translates the passage from a nonequilibrium glassy state to the thermodynamically stable liquid state. This nonequilibrium state, fruit of a low kinetic of the molecular rearrangement on cooling, has a direct consequence on the dynamic of polymer chains during the relaxation.33,52 This relaxation process can be mostly

BGE in the macroscopic properties in relationship with the polymer architecture. Relationship between the Macromolecular Architecture and the Polymer Chain Dynamics. Table 1 summarizes the physicochemical properties of both HBPs. The measurements were achieved for the resulting polymer obtained just after gelation and without purification. Indeed, a previous study24 and our own experience using different solvent and temperature attest that the removal of the unreacted monomers cannot be done without taking away a nonnegligible part of the oligomers constitutive of the sol part. Herein, due to the low gel content (Table 1), this strategy was adopted to be able to make the link between the architecture topology and the macroscopic properties. The study of the chain dynamic in the relaxation process by DSC may give important information about the polymer chains interactions, directly dealing with the impact of the BGE addition. The obtained values in Table 1 are already a good indication of the networks disparity. The PGS exhibited a Tg of −3.9 °C for a molecular weight of 4191 g mol−1 while the PGS-co-BGE showed a Tg of −16.1 °C for a molecular weight of 19 275 g mol−1. This tendency is on the opposite trend that the literature describes in the case of linear47,48 or dendritic49 polymers. Indeed, for such networks, the Tg value should increases with the augmentation of the molecular weight. This particular behavior is a first indication that BGE has a real influence on the polymer architecture and not only act as a reactive gelation I

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 8. Thermograms of the PGS (A) and PGS-co-BGE (B) at several heating rates exhibiting the enthalpy relaxation recovery, after annealing at −65 °C, associated with the β-relaxation; linear fits for the estimation of the β-relaxation activation energy (C).

glassy state, the molecular dynamic can be associated with the cooperative (α-relaxation) and noncooperative (β-relaxation) motion of some groups of atoms close to a zone of mobility. Because these two processes can be strongly coupled in the early stage of the Tg, the increase of Eα could be imputed to a strong influence of the β-relaxation process. Indeed, by means of annealing experiments at different temperatures, it was shown that the sub-Tg relaxation activation energy increased by taking annealing temperature closer and closer to the Tg.59 Thus, in order to clarify this point, we investigate the βrelaxation process by using the method described in the Materials and Methods section. The regressions presented in Figure 8 exhibits a good linearity confirming the Arrhenius law obeisance of the β-relaxation process. The extracted values of Eβ calculated from the regression slopes are ≈256 and ≈120 kJ mol−1 for PGS and PGS-co-BGE, respectively (Table 1). Surprisingly, the obtained values are higher than the energy values get for the Eα dependence for both polymers case. This trend is not consistent with the typical values reported in the literature for polymers’ α- and β-relaxations.60 Indeed, the cooperative nature of the α-relaxation process should requires a high amount of energy, while the noncooperative motions are generally nonsensitive to a variation of the free volume.33 Nevertheless, in this particular case, the PGS and PGS-co-BGE chemical structures suggest that even the local atomic mobility (crankshaft motion often associated with β-relaxation for instance60) can be strongly dependent on the hydrogen bond’s strength. The utilization of G with SA as very short linker can be one of the major factors because it conditions the distance between two possible hydrogen bonds connections. Furthermore, the annealing temperature for Eβ calculations is −65 °C. At this low temperature, we assume that the hydrogen bonds are very strong: the PGS architecture could thus be assimilated to a “pseudo highly cross-linked network”. Concerning the intrinsic value of the obtained Eβ, for a same annealing temperature of −65 °C, we have Eβ(PGS-co-BGE) < Eβ(PGS). This logical trend could be explained by two phenomenon contributions. The first one is related to the

discriminate by cooperative and noncooperative mechanisms, respectively called α- and β-relaxation. While the α-process involved the conjoint motions of the closest highly hindered neighbor polymer chains, the β-relaxation process can refer to small parts of the polymer architecture that is near a “mobility island” area53 (i.e., possessing higher degree of freedom). In terms of chemical structure, that can be associated with secondary chains as ramification, dandling chains or endingchain groups. Both of these processes could occur in parallel.54 While the β-relaxation follows an Arrhenius dependence, the αrelaxation process exhibits a deviation from the Arrhenius behavior.33 Thus, the application of the advanced isoconversional kinetic method is particularly suitable because it has been shown that it is a powerful tool to detect non-Arrhenius behavior.37,55 The range of energy value variation presented in Figure 7 is related to several kJ mol−1 along the overall relaxation process. According to the literature,36,40,56 the energy variation generally associated with the α-relaxation is about hundreds of kJ mol−1, whereas that related to the β-relaxation are more close to dozens of kJ mol−1. This first observation on the energy variation, which is on the same range of the hydrogen bonds energy,57 could be an indication confirming that the hydrogen bonding governs the overall relaxation process, as suggested by the previous investigation presented in the Introduction. The second observation highlights a same evolution trend for the Eα dependence for both HBPs: an increase of Eα until a maximum (Emax) and then a decrease. In order to facilitate the interpretation, the following discussion on the potential mechanisms involved in the relaxation process will be segmented in two parts: one part related to the activation energy evolution before Emax and the other one after Emax. Since the original idea proposed by Vyazovkin et al.,31 several polymers have been investigated: PET, PS, PEN, PVP, PBMA, and PEF.58 All these polymers exhibit the same trend of a decrease of Eα with the increase of temperature, while in our case, we observe an unusual increase. Before Emax, close to the J

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Eα value and the study of the Lederer10 group which highlight the predominant role of the hydrogen bonding on the glass transition value for OH-terminated HBPs. In comparison to the hydrogen bonding, the molecular weight only has a marginal effect. Now we will focus on the region beyond Emax, related to the decrease of the activation energy. This effective energy variation can be explained in structural term by the intensification of the polymer chains motion with the enhancement of the temperature. Indeed, closest to the glassy state, few chain segments requiring a huge degree of cooperativity are able to move together. When the temperature increases, the degree of freedom increases and the chain segments are free to relax. The decrease of this energetic constriction is translated by the diminution of the apparent activation energy.58 Then, the return of an Arrhernius governance of the relaxation, sign of the equilibrium state, can be observed by a constant value of the effective activation energy. As a result, Emax could correspond to the point where take place the transition from a strongly cooperative motion mechanism to a relaxation mechanism involving the motion of more independent chain segments. This transition is possible because the reduction of the hydrogen bonding strength over the molecular motion occurs during the relaxation. The energetic gap necessary to reach this transition seems consistent with a value of around 25 kJ mol−1. Originally, Angell34 proposed a diagram build with a modified version of the Vogel−Tammann−Fulcher (VTF) equation to classify both the liquid and plastic glass depending on their departure from the Arrhenius behavior through the relaxation. Herein, “strong” liquids exhibit small deviation from the Arrhenius law while “fragile” ones present a more pronounced deviation.54 These glass-forming liquid are then classified according to their dynamic fragility “m”. Vyazovkin et al.63 introduced a variability parameter (ΔE) by eq 7 and demonstrated its good correlation with the dynamic fragility (eq 8).56

local motion of the BGE butyl tail which potentially contributes to the global β-relaxation process. The second one, related to our previous hypothesis, could be associated with the role of this butyl tails as spacer between the polymer chains. The PGSco-BGE network possesses more free volume (i.e., strength of the hydrogen bonds is reduced); thus the motions can occur easily. An extrapolation can be formulated, relative to the hindrance of the chains motions by the hydrogen bonds, encompassing the observation of both the high value of Eβ and the increasing energy until Emax for the Tg. Indeed, two antagonist processes are in competition when the temperature increases: the molecular motion associated with an increase of degree of freedom and the effect of the hydrogen bonding related to a reduction of the degree of freedom. These kinds of secondary interactions can have a huge influence on the α-relaxation process, depending on their amount and strength. This fact has been already observed in the case of phenyl π−π stacking interaction for liquid-crystalline thermoset by dynamic mechanical analysis (DMA).61,62 However, the most relevant example is related to the effective activation energy dependency of the poly(vinyl chloride) (PVC) that exhibits an increase between α = 0 and 0.2.63 By analogy with the present system, the van der Waals secondary bonds interchains (H−Cl dipole− dipole interaction)64 can participate in the motion hindrance. For the PGS dependence, the energy increases in a linear manner until α = 0.5. This suggests an apparent linearity between the increases of both the number of atoms moving cooperatively and the hydrogen bonds involved in this dynamic. Concerning the PGS-co-BGE the energy increases until α = 0.25, however, this time the increase is not represented by one linear segment but two. This could be related to two different kinds of mechanisms. In good agreement with the Eβ values, because of the BGE introduction, the neighboring atoms reorganization is affected by different “holes” structures and dynamics of distribution, according to Hirai and Eyring theory.65,66 Herein, the first linear part could be related to the motion of few atoms hindered by the hydrogen bonding as is the case for the PGS. Then when the temperature increases, more and more segments move cooperatively integrating the BGE butyl tail’s motion. Since this terminal apolar butyl tail possesses a higher intrinsic degree of freedom, it could have a synergetic effect on the entire polymer backbone motion, playing the role of a dynamic spacer. This could be an explanation for the low value of α = 0.25 for Emax in PGS-coBGE compared to α = 0.5 for the PGS. An interesting observation is that ΔEmax from α = 0.1 to α (Emax) corresponds to a same value of around 25 kJ mol−1. That could translate that the required energy to reach (Emax) corresponds to a same physical mechanism but which possess different kinetic and pathway. Another remarkable observation is the same factor two when calculating the ratio for α (Emax) (≈0.5/0.25) and Emax (≈152/75 kJ mol−1) for the α-relaxation process but also concerning the β-relaxation process Eβ (≈256/120 kJ mol−1) in both PGS systems. This observation seems to corroborate our previous hypothesis on the role of the butyl tail as dynamic spacer. Indeed, first the overall effective energy of the Tg dependence is lower in the case of the PGS-co-BGE, arguing that there is more free volume. Second, the factor two seems to indicate that there is a direct relationship between the intrachain interaction (hydrogen bonding and effect of the butyl tail) and this free volume. This last hypothesis is in good agreement with our previous assumption based on the range of

ΔE =

E0.25 − E0.75 T0.25 − T0.75

log(−ΔE ) = − 0.438 + 0.0145m

(7) (8)

As seen in eq 7, ΔE is taken as the energetic difference between α = 0.25 and α = 0.75. In the particular case of secondarybonded glass former, because of the presence of Emax, we propose to reformulate the variability parameter from αinitial ≈ α(Emax) to αterminal that can be considered as the last α value before the energy becomes constant (i.e., thermodynamic equilibrium). Herein, a linear fit has been done for PGS from α = 0.52 to α = 0.84 with a good correlation of 0.935 and from α = 0.27 to α = 0.72 with a correlation of 0.955 for PGS-co-BGE. The results are in the same order for both HBPs, with a value of ΔE = −28.8 and −19.5 kJ mol−1 K−1 and m = 130.8 and 119.2, respectively (Table 1). These data can give a qualitative estimation of the fragility degree of these hyperbranched polymers. The comparison of these values for both ΔE and m with the literature classifies these HBPs as fragile, in good correlation with what have been found for polymeric glass.63,67 In order to support the previous discussion on the apparent activation energy dependencies, stress relaxation experiments were performed at a temperature beyond Tg (at 30 °C). The first observation in Figure 9 is the same viscous response for both HBPs to the shear stress with a complete energy dissipation (the values of residual stress correspond to 0.18 ± K

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

delay the gelation phenomenon. However, the comparison of Mn and Mw argues in favor of a major BGE grafting strategy, the conversion increase being mostly associated with the reaction between G and SA. An advantage due to the fast kinetic of the epoxide reaction vs polyesterification is that the resulting architecture can be tuned by adding the comonomer at different stages of the polyesterification. As an example, a late introduction of BGE could generate core−shell structures. The BGE introductions occurring in the early stage of the polyesterification, the reaction with the oligomers allow a good repartition of the BGE pattern in the global architecture. This is reflected the presence of only one Tg at low temperature. The results obtained from the estimation of Eβ, the Tg effective activation energy variation, and the stress relaxation experiment beyond Tg highlight the dominant role of the hydrogen bonding in the governance of the overall polymer relaxation process. Indeed, the proportionality related to the energy value of Eβ as well as Emax and its extent of conversion directly suggest the mains control of the secondary interaction in the motion dynamics associated with the free volume expansion. This particular behavior may be generalizable to polymer glasses possessing a critical amount of hydrogen bonds. However, this first insight into the highly hydroxylated HBPs relaxation processes obtained by DSC needs corroborative evidence from dielectric or mechanical spectroscopies to be able to establish a precise scheme of the noncooperative and cooperative motions. Several exiting perspectives can be developed. From the fundamental point of view the utilization of a catalyst can modify the epoxide selectivity with regard to the alcohol and carboxylic acid function. Different times of BGE introduction and different amounts can also be explored to highlight theirs significance on the branching topology and polyesterification reactivity. In order to evaluate and model the effect a dynamic spacer in a HPBs hydrogen-bonded network, the utilization of different tail length for the comonomer agent can be envisaged. Finally, considering the high applicative potential of this glycidyl strategy presented in the Introduction, the design of materials with diverse properties of self-healing or NLO for instance and for several application fields as thermoplastic blending compatibilizer or nanomedicine can be strongly envisaged.

Figure 9. Stress relaxation curves for both PGS and PGS-co-BGE with their respective approximate relaxation time at a temperature of 30 °C.

0.05% and 0.05 ± 0.01% for PGS and PGS-co-BGE, respectively). According to the literature, the absence of polymer chain entanglement due to the fractal polymeric structure of HBPs68,69 associates the relaxation mechanisms control with the intramolecular reorganization of the highest molecular weight polymer fraction.70,71 By analogy with the present case, the polymer relaxation should be function of the hydrogen-bonding network rearrangement. The magnification of the relaxation curves in the Figure 9 inset clearly shows that the PGS-co-BGE takes more time to relax than the PGS. This extrapolation beyond Tg is in very good correlation with that we observed during the glass transition: this relaxation delay for the PGS-co-BGE can be associated with two mains factors which are the higher free volume and the butyl tail motion. Furthermore, the relaxation time of these HBPs can be estimated as being the time where the steady level of stress is reached. Approximate values of 90 and 180 s for PGS and PGSco-BGE were found. It seems that a same factor two between the HBPs, as observed for the value of α(Emax) and Emax, can describe the stress relaxation process. That reinforces the idea that the relaxation phenomenon of OH-terminated HBPs is dominated by the intrachain interactions during and beyond the glass transition.



ASSOCIATED CONTENT

S Supporting Information *



The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.macromol.6b02424. HSQC spectra of the PGS, the HSQC and HMBC correlation spectrum of the GBGE, as well as the 13C NMR shifts and integrations related to the carboxylic acid attack in the position 1 of the BGE (PGS-co-BGE spectrum at 5.25 and 12.25 h) (PDF)

CONCLUSION The butyl glycidyl ether comonomer has been successfully incorporated in the poly(glycerol−succinate) (PGS) architecture. The addition effect on the polymerization reaction has been evidenced from the architectural topology to the polymer chain dynamic. The determination of the predominant reactions related to the BGE epoxide group has been founded for the carboxylic acid and the primary alcohol in position 1. The BGE have been integrated in terminal position of the PGS segment via grafting strategy. This is highlighted by the two cross-peaks in zone 2 involving the reaction between a −COOH/−OH functional groups and the epoxide in position 1. Furthermore, the cross-peaks in zone 1 present the potential introduction of BGE structure via both grafting and branching. The modification of the alcohol selectivity and the stoichiometry balance between alcohol and carboxylic acid function lead to increase the polymerization conversion and



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (A.M.). ORCID

Amar Mohanty: 0000-0002-1079-2481 Notes

The authors declare no competing financial interest. L

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules



(16) Wang, Y.; Ameer, G. A.; Sheppard, B. J.; Langer, R. A tough biodegradable elastomer. Nat. Biotechnol. 2002, 20 (6), 602−606. (17) Weiss, V. M.; Naolou, T.; Hause, G.; Kuntsche, J.; Kressler, J.; Mäder, K. Poly (glycerol adipate)-fatty acid esters as versatile nanocarriers: From nanocubes over ellipsoids to nanospheres. J. Controlled Release 2012, 158 (1), 156−164. (18) Agach, M.; Delbaere, S.; Marinkovic, S.; Estrine, B.; NardelloRataj, V. Characterization, stability and ecotoxic properties of readily biodegradable branched oligoesters based on bio-sourced succinic acid and glycerol. Polym. Degrad. Stab. 2012, 97 (10), 1956−1963. (19) Valerio, O.; Horvath, T.; Pond, C.; Manjusri, M.; Mohanty, A. Improved utilization of crude glycerol from biodiesel industries: Synthesis and characterization of sustainable biobased polyesters. Ind. Crops Prod. 2015, 78, 141−147. (20) Li, Y.; Cook, W. D.; Moorhoff, C.; Huang, W. C.; Chen, Q. Z. Synthesis, characterization and properties of biocompatible poly (glycerol sebacate) pre-polymer and gel. Polym. Int. 2013, 62 (4), 534−547. (21) Maliger, R.; Halley, P. J.; Cooper-White, J. J. Poly (glycerol− sebacate) bioelastomerskinetics of step-growth reactions using Fourier Transform (FT)-Raman spectroscopy. J. Appl. Polym. Sci. 2013, 127 (5), 3980−3986. (22) Maliger, R. B.; Halley, P. J.; Cooper-White, J. J. Poly (glycerolsebacate) bioelastomers: 2. Synthesis using Brabender Plasticoder® as a batch reactor. J. Appl. Polym. Sci. 2016, 133 (1), 42852. (23) Wyatt, V. T.; Strahan, G. D.; Nuñez, A. The Lewis AcidCatalyzed Synthesis of Hyperbranched Oligo (glycerol−diacid) s in Aprotic Polar Media. J. Am. Oil Chem. Soc. 2010, 87 (11), 1359−1369. (24) Wyatt, V. T.; Nuñez, A.; Foglia, T. A.; Marmer, W. N. Synthesis of hyperbranched P poly (glycerol-diacid) oligomers. J. Am. Oil Chem. Soc. 2006, 83 (12), 1033−1039. (25) Zhang, T.; Howell, B. A.; Dumitrascu, A.; Martin, S. J.; Smith, P. B. Synthesis and characterization of glycerol-adipic acid hyperbranched polyesters. Polymer 2014, 55 (20), 5065−5072. (26) Yang, Y.; Lu, W.; Cai, J.; Hou, Y.; Ouyang, S.; Xie, W.; Gross, R. A. Poly (oleic diacid-co-glycerol): comparison of polymer structure resulting from chemical and lipase catalysis. Macromolecules 2011, 44 (7), 1977−1985. (27) Zhang, Y.-R.; Spinella, S.; Xie, W.; Cai, J.; Yang, Y.; Wang, Y.-Z.; Gross, R. A. Polymeric triglyceride analogs prepared by enzymecatalyzed condensation polymerization. Eur. Polym. J. 2013, 49 (4), 793−803. (28) Otton, J.; Ratton, S.; Vasnev, V. A.; Markova, G. D.; Nametov, K. M.; Bakhmutov, V. I.; Komarova, L. I.; Vinogradova, S. V.; Korshak, V. V. Investigation of the formation of poly (ethylene terephthalate) with model molecules: kinetics and mechanisms of the catalytic esterification and alcoholysis reactions. II. Catalysis by metallic derivatives (monofunctional reactants). J. Polym. Sci., Part A: Polym. Chem. 1988, 26 (8), 2199−2224. (29) Henry Lee, K. N. Handbook of Epoxy Resins; McGraw-Hill Book Co.: 1982. (30) Pascault, J.-P.; Williams, R. J. Epoxy Polymers; John Wiley & Sons: 2009. (31) Vyazovkin, S.; Sbirrazzuoli, N.; Dranca, I. Variation of the effective activation energy throughout the glass transition. Macromol. Rapid Commun. 2004, 25 (19), 1708−1713. (32) Johari, G. P.; Goldstein, M. Molecular mobility in simple glasses. J. Phys. Chem. 1970, 74 (9), 2034−2035. (33) Donth, E.-J. The Glass Transition: Relaxation Dynamics in Liquids and Disordered Materials; Springer Science & Business Media: 2013; Vol. 48. (34) Angell, C. Relaxation in liquids, polymers and plastic crystals strong/fragile patterns and problems. J. Non-Cryst. Solids 1991, 131, 13−31. (35) Vyazovkin, S. Modification of the integral isoconversional method to account for variation in the activation energy. J. Comput. Chem. 2001, 22 (2), 178−183. (36) Hodge, I. M. Enthalpy relaxation and recovery in amorphous materials. J. Non-Cryst. Solids 1994, 169 (3), 211−266.

ACKNOWLEDGMENTS This research is financially supported (1) the Ontario Ministry of Agriculture, Food, and Rural Affairs (OMAFRA)-New Directions Research Program (Project # 050155); (2) OMAFRA-University of Guelph Bioeconomy-Industrial Uses Theme (Project # 200283; #200359); and (3) Ontario Research Fund, Research Excellence Program; Round-7 (ORF-RE07) from the Ontario Ministry of Research and Innovation (MRI), currently known as the Ontario Ministry of Research, Innovation and Science (MRIS) (Project # 052644 and # 052665). The authors deeply thank Dr. Sameer AlAbdul-Wahid and Ehsan Behazin for their great help with the NMR experiments and the computation of the kinetic method, respectively.



REFERENCES

(1) Staudinger, H.; Schulz, G. V. Ü ber hochpolymere Verbindungen, 126. Mitteil.: Vergleich der osmotischen und viscosimetrischen Molekulargewichts-Bestimmungen an polymerhomologen Reihen. Ber. Dtsch. Chem. Ges. B 1935, 68 (12), 2320−2335. (2) Flory, P. J. Molecular Size Distribution in Three Dimensional Polymers. I. Gelation1. J. Am. Chem. Soc. 1941, 63 (11), 3083−3090. (3) Flory, P. J. Molecular Size Distribution in Three Dimensional Polymers. VI. Branched Polymers Containing ARBf-1 Type Units. J. Am. Chem. Soc. 1952, 74 (11), 2718−2723. (4) Stockmayer, W. H. Theory of Molecular Size Distribution and Gel Formation in Branched-Chain Polymers. J. Chem. Phys. 1943, 11 (2), 45−55. (5) Lederer, A.; Burchard, W. Hyperbranched Polymers: Macromolecules in Between Deterministic Linear Chains and Dendrimer Structures; Royal Society of Chemistry: 2015; Vol. 16. (6) Kim, Y. H.; Webster, O. W. Water soluble hyperbranched polyphenylene: “a unimolecular micelle?″. J. Am. Chem. Soc. 1990, 112 (11), 4592−4593. (7) Voit, B. I.; Lederer, A. Hyperbranched and Highly Branched Polymer ArchitecturesSynthetic Strategies and Major Characterization Aspects. Chem. Rev. 2009, 109 (11), 5924−5973. (8) Chen, H.; Kong, J. Hyperbranched polymers from A2 + B3 strategy: recent advances in description and control of fine topology. Polym. Chem. 2016, 7 (22), 3643−3663. (9) McKee, M. G.; Elkins, C. L.; Park, T.; Long, T. E. Influence of Random Branching on Multiple Hydrogen Bonding in Poly(alkyl methacrylate)s. Macromolecules 2005, 38 (14), 6015−6023. (10) Khalyavina, A.; Häußler, L.; Lederer, A. Effect of the degree of branching on the glass transition temperature of polyesters. Polymer 2012, 53 (5), 1049−1053. (11) Bai, Y.; Song, N.; Gao, J. P.; Sun, X.; Wang, X.; Yu, G.; Wang, Z. Y. A New Approach to Highly Electrooptically Active Materials Using Cross-Linkable, Hyperbranched Chromophore-Containing Oligomers as a Macromolecular Dopant. J. Am. Chem. Soc. 2005, 127 (7), 2060− 2061. (12) Thurecht, K. J.; Blakey, I.; Peng, H.; Squires, O.; Hsu, S.; Alexander, C.; Whittaker, A. K. Functional Hyperbranched Polymers: Toward Targeted in Vivo 19F Magnetic Resonance Imaging Using Designed Macromolecules. J. Am. Chem. Soc. 2010, 132 (15), 5336− 5337. (13) Jia, Z.; Chen, H.; Zhu, X.; Yan, D. Backbone-Thermoresponsive Hyperbranched Polyethers. J. Am. Chem. Soc. 2006, 128 (25), 8144− 8145. (14) Burakowska, E.; Quinn, J. R.; Zimmerman, S. C.; Haag, R. Cross-Linked Hyperbranched Polyglycerols as Hosts for Selective Binding of Guest Molecules. J. Am. Chem. Soc. 2009, 131 (30), 10574−10580. (15) Mohammadifar, E.; Kharat, A. N.; Adeli, M. Polyamidoamine and polyglycerol; their linear, dendritic and linear−dendritic architectures as anticancer drug delivery systems. J. Mater. Chem. B 2015, 3 (19), 3896−3921. M

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules (37) Vyazovkin, S.; Burnham, A. K.; Criado, J. M.; Pérez-Maqueda, L. A.; Popescu, C.; Sbirrazzuoli, N. ICTAC Kinetics Committee recommendations for performing kinetic computations on thermal analysis data. Thermochim. Acta 2011, 520 (1), 1−19. (38) Illers, V. K. H. Einfluss der thermischen Vorgeschichte auf die Eigenschaften von Polyvinylchlorid. Makromol. Chem. 1969, 127 (1), 1−33. (39) Chen, H. On mechanisms of structural relaxation in a Pd 48 Ni 32 P 20 glass. J. Non-Cryst. Solids 1981, 46 (3), 289−305. (40) Bershtein, V.; Yegorov, V. General mechanism of the β transition in polymers. Polym. Sci. U. S. S. R. 1985, 27 (11), 2743− 2757. (41) Vyazovkin, S.; Dranca, I. Activation energies derived from the pre-glass transition annealing peaks. Thermochim. Acta 2006, 446 (1), 140−146. (42) Vyazovkin, S.; Dranca, I. Physical stability and relaxation of amorphous indomethacin. J. Phys. Chem. B 2005, 109 (39), 18637− 18644. (43) Kulshrestha, A. S.; Gao, W.; Gross, R. A. Glycerol copolyesters: control of branching and molecular weight using a lipase catalyst. Macromolecules 2005, 38 (8), 3193−3204. (44) Hölter, D.; Burgath, A.; Frey, H. Degree of branching in hyperbranched polymers. Acta Polym. 1997, 48 (1−2), 30−35. (45) Flory, P. J. Principles of Polymer Chemistry; Cornell University Press: 1953. (46) Dušek, K.; Dušková-Smrčková, M. Modeling of Polymer Network Formation from Preformed Precursors. Macromol. React. Eng. 2012, 6 (11), 426−445. (47) Fox, T. G.; Flory, P. J. The glass temperature and related properties of polystyrene. Influence of molecular weight. J. Polym. Sci. 1954, 14 (75), 315−319. (48) Turner, D. Glass transition elevation by polymer entanglements. Polymer 1978, 19 (7), 789−796. (49) Wooley, K. L.; Hawker, C. J.; Pochan, J. M.; Frechet, J. M. J. Physical properties of dendritic macromolecules: a study of glass transition temperature. Macromolecules 1993, 26 (7), 1514−1519. (50) Sunder, A.; Bauer, T.; Mülhaupt, R.; Frey, H. Synthesis and thermal behavior of esterified aliphatic hyperbranched polyether polyols. Macromolecules 2000, 33 (4), 1330−1337. (51) Zhu, Q.; Wu, J.; Tu, C.; Shi, Y.; He, L.; Wang, R.; Zhu, X.; Yan, D. Role of branching architecture on the glass transition of hyperbranched polyethers. J. Phys. Chem. B 2009, 113 (17), 5777− 5780. (52) Angell, C. A. Insights into phases of liquid water from study of its unusual glass-forming properties. Science 2008, 319 (5863), 582− 587. (53) Johari, G. P. Intrinsic mobility of molecular glasses. J. Chem. Phys. 1973, 58 (4), 1766−1770. (54) Angell, C. A. Formation of glasses from liquids and biopolymers. Science 1995, 267 (5206), 1924. (55) Pin, J.-M.; Mija, A.; Sbirrazzuoli, N. In Star-Epoxy Mesogen to Design Multifunctional Materials: Polymerization Kinetics and Fractal Self-Assembly; EPF, European Polymer Federation Congress. (56) Vyazovkin, S. Isoconversional Kinetics of Thermally Stimulated Processes; Springer: 2015. (57) Steiner, T. The hydrogen bond in the solid state. Angew. Chem., Int. Ed. 2002, 41 (1), 48−76. (58) Codou, A.; Moncel, M.; van Berkel, J. G.; Guigo, N.; Sbirrazzuoli, N. Glass transition dynamics and cooperativity length of poly (ethylene 2, 5-furandicarboxylate) compared to poly (ethylene terephthalate). Phys. Chem. Chem. Phys. 2016, 18, 16647−16658. (59) Vyazovkin, S.; Dranca, I. A DSC study of α-and β-relaxations in a PS-clay system. J. Phys. Chem. B 2004, 108 (32), 11981−11987. (60) McCrum, N. G.; Read, B. E.; Williams, G. Anelastic and Dielectric Effects in Polymeric Solids; Wiley: 1967. (61) Ortiz, C.; Kim, R.; Rodighiero, E.; Ober, C. K.; Kramer, E. J. Deformation of a Polydomain, Liquid Crystalline Epoxy-Based Thermoset. Macromolecules 1998, 31 (13), 4074−4088.

(62) Pin, J.-M.; Sbirrazzuoli, N.; Sacarescu, L.; Mija, A. Star-epoxy mesogen with 1, 3, 5-triazine core: a model of A 4 B 3 fractal polymerization in a liquid crystalline thermoset media. Polym. Chem. 2016, 7 (6), 1221−1225. (63) Vyazovkin, S.; Sbirrazzuoli, N.; Dranca, I. Variation in Activation Energy of the Glass Transition for Polymers of Different Dynamic Fragility. Macromol. Chem. Phys. 2006, 207 (13), 1126−1130. (64) Mark, J. E. Random-Coil Dimensions and Dipole Moments of Vinyl Chloride Chains. J. Chem. Phys. 1972, 56 (1), 451−458. (65) Hirai, N.; Eyring, H. Bulk viscosity of liquids. J. Appl. Phys. 1958, 29 (5), 810−816. (66) Wunderlich, B. Study of the change in specific heat of monomeric and polymeric glasses during the glass transition. J. Phys. Chem. 1960, 64 (8), 1052−1056. (67) Beiner, M.; Huth, H.; Schröter, K. Crossover region of dynamic glass transition: general trends and individual aspects. J. Non-Cryst. Solids 2001, 279 (2−3), 126−135. (68) Suneel; Buzza, D. M. A.; Groves, D. J.; McLeish, T. C. B.; Parker, D.; Keeney, A. J.; Feast, W. J. Rheology and Molecular Weight Distribution of Hyperbranched Polymers. Macromolecules 2002, 35 (25), 9605−9612. (69) Cameron, C.; Fawcett, A. H.; Hetherington, C. R.; Mee, R. A.; McBride, F. V. Step Growth of Two Flexible AB f Monomers: The Self-Return of Random Branching Walks Eventually Frustrates Fractal Formation. Macromolecules 2000, 33 (17), 6551−6568. (70) Jahromi, S.; Palmen, J. H. M.; Steeman, P. A. M. Rheology of Side Chain Dendritic Polymers. Macromolecules 2000, 33 (2), 577− 581. (71) Irzhak, V. I. Topological structure and relaxation properties of branched polymers. Russ. Chem. Rev. 2006, 75 (10), 919−934.

N

DOI: 10.1021/acs.macromol.6b02424 Macromolecules XXXX, XXX, XXX−XXX