Critical transition of epoxy resin from brittleness to toughness by incorporating CO2-sourced cyclic carbonate

Journal of CO₂ Utilization 26 (2018) 302–313

Contents lists available at ScienceDirect

Journal of CO2 Utilization journal homepage: www.elsevier.com/locate/jcou

Critical transition of epoxy resin from brittleness to toughness by incorporating CO2-sourced cyclic carbonate

T

⁎

Jiexi Kea,b, Xiaoyun Lia,b, Shuai Jianga,b, Junwei Wanga, , Maoqing Kanga, Qifeng Lia, Yuhua Zhaoa a b

Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan, 030001, China University of Chinese Academy of Sciences, Beijing, 100049, China

A R T I C LE I N FO

A B S T R A C T

Keywords: CO2-sourced monomer Cyclic carbonate Epoxy resin Toughness Hydrogen bonding Urethane linkage

A five-membered cyclic carbonate as CO2-sourced monomer was prepared from ethylene glycol diglycidyl ether (EGDE) with CO2 by cycloaddition reaction, abbreviated as 5CC-EGDE, at 130 ℃ under 10 bar CO2 pressure in the presence of quaternary ammonium salt modified amberlyst (D296) for 20 h. The complete conversion and 98.6% selectivity were obtained, respectively. The incorporation of 5CC-EGDE into epoxy resin could reduce the viscosity during preparation process and improve the toughness effectively. It was ascribed to the ring-opening reaction of 5CC-EGDE with curing agent, and hydrogen bonding, formed between urethane groups and other polar groups. Besides, the low crosslinking density and unreacted carbonate groups remaining in epoxy network were beneficial to the movement of molecular chains and dissipation of energy during deformation process. In addition, effects of hydrogen-bonding interactions and crosslinking density on mechanical performances were also investigated by varying EGDE conversion and amount of added curing agent, respectively. The results indicated that mechanical performances of epoxy resin were promoted by combined effects of urethane linkage, hydrogen bonding, low crosslinking density and unreacted carbonate groups.

1. Introduction Carbon dioxide (CO2) is widely regarded as one of the main greenhouse gas, which is notorious for global warming. The chemical fixation of CO2 has been drawing a huge amount of attention as a result of the fact that it’s an important C1 feedstock in industry as it is abundant, easily available, non-toxic and inexpensive [1], and it quite fits well the concepts of “green chemistry” and “sustainable society” [2]. Employing CO2 as a raw material to prepare chemicals may create a positive pathway for utilization of CO2. One of the promising routes is the synthesis of cyclic carbonate via cycloaddition reaction of CO2 with epoxides [3]. Since cyclic carbonate has special properties, it is widely used as green polar solvents, electrolytes in lithium batteries as well as intermediates for organic synthesis [4]. Epoxy resin, one of the most important thermosetting polymers, is widely used as adhesives [5], coatings [6] and structural materials for high-performance composites [7] due to its outstanding performances. As thermosetting materials, epoxy resin exhibits a high degree of crosslinking density, which endows it with useful properties. However, the highly cross-linked structure also causes epoxy inherently brittle and vulnerable to cracks, thus limiting their advanced applications.

⁎

Corresponding author. E-mail address: [email protected] (J. Wang).

https://doi.org/10.1016/j.jcou.2018.05.020 Received 23 December 2017; Received in revised form 3 May 2018; Accepted 20 May 2018 2212-9820/ © 2018 Elsevier Ltd. All rights reserved.

Therefore, it is of great interest to improve the toughness of epoxy resin. Numerous attempts have been adopted in this field. The toughening strategies mainly include three categories: (1) reduce of crosslinking density [8], (2) use particles/fillers as a second phase [9], and (3) introduce plasticizers to increase plastic deformation [10]. Among these methods, incorporation of polyurethane linkage is beneficial for the formation of chemical bonding and physical entanglements such as urethane and hydrogen-bonding interactions, which can increase intraand inter- molecular forces in epoxy network [11]. In general, the traditional polyurethane linkage is incorporated into epoxy system by reacting isocyanates or isocyanate-terminated polyurethane oligomers with hydroxyl groups in epoxy resin [12]. However, the repetitive exposure to isocyanate can lead to serious health issues [13]. In addition, the synthesis of isocyanates, for instance methane diphenyl 4,4′-diisocyanate and toluene diisocyanates, requires toxic phosgene [14]. Therefore, in recent years, polyurethane prepared from non-isocyanate methods has attracted considerable attention [15,16]. One of the most attractive ways for synthesis of non-isocyanate polyurethane is that five-membered cyclic carbonate (5CC) is first synthesized from epoxide substrates and CO2, then reacts with amine. By using this “green chemistry” route, the use of toxic and expensive

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

The EGDE with different conversion was also prepared by varying reaction time for 0, 6 and 15 h.

isocyanates can be avoided. The 5CC-modified epoxy resin can be obtained either by adding cyclic carbonate [17] or by replacing only part of epoxy groups by reacting it with CO2 [18]. In previous reports, only several research groups have concentrated on 5CC-modified epoxy by incorporating cyclic carbonate. Khoshkish and co-workers have investigated the effect of CO2 fixing process producing cyclic carbonate on dynamic viscosity and kinetics of epoxy resin [19]. Rokicki et al. [17] have tremendously studied on the gel time, mechanical properties and curing kinetics of 5CC-modified epoxy resin. However, little attention has been paid to the toughening mechanism of 5CC to epoxy resin. Until recently, Ghanbaralizadeh [20] investigates the toughening mechanism by analyzing activation energy of β relaxation of chain segments. However, epoxy diluent which is adopted during the CO2 fixation into epoxy also reacts with CO2. The mixtures achieve high cyclic carbonate content which could increase viscosity causing heterogeneous dispersion. Meanwhile, the report neglects the effect of urethane linkage provided by 5CC-modified diluent on mechanical performances and toughening mechanism of epoxy resin. In this work, the epoxy resin was toughened by incorporating 5CCmodified diluent. And the toughening mechanism was further investigated. The 5CC was synthesized from CO2 and bifunctional reactive diluent, ethylene glycol diglycidyl ether (EGDE), which could be obtained from renewable glycerol that was a by-product from the preparation process of biodiesel [21–23]. Therefore, the relationship between mechanical performances and structure of modified epoxy was investigated. The influence of 5CC content on properties of epoxy composites was studied by rheology, fourier transform infrared spectra (FTIR), dynamic mechanical properties (DMA), scanning electron microscopy (SEM), etc. To study the toughening mechanism, effects of urethane linkage, hydrogen bonding, unreacted cyclic carbonate groups and crosslinking density on mechanical properties of epoxy composites were studied.

2.3. Curing reaction of BADGE/5CC-EGDE mixtures BADGE was mixed with 5 wt% 5CC-EGDE and 0.1 wt% TEDA as catalyst (compared to total weight of 5CC-EGDE and BADGE) for 10 min using a mechanical stirrer. In order to avoid bubbling, the mixture was degassed in heated oven at 75 ℃ for 20 min. Then, calculated amount curing agent, DETA, was added. After mixing, an ultrasonic bath was used to eliminate the bubbles during mixing process of the resulting mixtures. Finally, the compositions were poured into polytetrafluoroethylene mold and kept at 75 ℃ for 24 h, followed by cured at 100 ℃ for 2 h and further placed at room temperature for one week. An analogous procedure was used to prepare BADGE/5CC-EGDE composites containing 0, 10, 15, 20, 25 and 30 wt% 5CC-EGDE. There is a large gap between the reactivity of cyclic carbonate and epoxy groups with curing agent DETA. DETA is a linear molecule with two primary amines located on the chain ends and one secondary amine in the backbone. The epoxy groups of BADGE can react with primary and secondary amine. The primary amine and secondary amine has two active hydrogen atoms and one hydrogen atom, respectively, which can react with epoxy groups. Hence, the number of reactive group of DETA towards BADGE is 5. However, primary amine reacts with cyclic carbonates more efficiently than internal secondary amine [25]. It was assumed that only some of the secondary amine would react with cyclic carbonate group. Here, the effective functionality of DETA was taken to be 2.5. And the DETA amount was calculated by following Eq. (1):

CEW5CC − EGDE × m5CC − EGDE nnumber of functional groups for 5CC − EGDE EEWBADGE × mBADGE + ) × MDETA nnumber of functional groups for BADGE

mDETA = (

2. Experimental section

where CEW5CC-EGDE, m5CC-EGDE represent carbonate equivalent weight (CEW) and mass of 5CC-EGDE, EEWBADGE and mBADGE represent the epoxy equivalent weight (EEW) and mass of BADGE, and n represents the functional groups of DETA for 5CC-EGDE or BADGE. MDETA and mDETA represent the number-average molar mass and desired mass of DETA.

2.1. Materials Ethylene glycol diglycidyl ether (EGDE) (number-average molar mass = 190 g/mol) was supplied by Wuhan Yuanchen Technology Co., Ltd. Commercially available grade of Bisphenol-A diglycidyl ether (BADGE) with 5.34 mmol/g epoxy equivalent weight (EEW) was purchased from Nantong Xingchen Synthetic Materials Co., Ltd. Quaternary ammonium salt modified amberlyst (D296) was purchased from Tianjin Resin Technology Co., Ltd. Triethylenediamine (TEDA) as catalyst and diethylenetriamine (DETA) as curing agent were purchased from Sinopharm Chemical Reagent Co., Ltd. CO2 (99.99%) was kindly supplied by Shanxi Institute of Coal Chemistry, Chinese Academy of Sciences. All the raw materials were used as received without further purification.

2.4. Characterization methods Fourier Transform Infrared Spectra (FTIR) of cyclic carbonate and epoxy/5CC-EGDE composites were acquired at 25 ℃ on a NICOLET-380 Fourier transform infrared instrument (Thermo Electron Co., Ltd. USA). Spectra were recorded in the range between 4000 and 500 cm−1, with 32 scans and resolution of 4 cm−1. The ring-opening conversion of 5CCEGDE and BADGE with curing agent was determined by FTIR and calculated by following Eq. (2):

2.2. Synthesis of five-membered cyclic carbonate-based EGDE

Intensity (%) = 1−

EGDE with complete conversion into five-membered cyclic carbonate, hereafter abbreviated as 5CC-EGDE, was synthesized by coupling EGDE (100 g) with CO2 (10 bar) at 130 ℃ in the presence of catalyst (10 wt% compared to EGDE) for 20 h (see Scheme 1). Detailed preparation process of 5CC-EGDE was reported in our previous report [24].

( (

I functional group

I functional group

) )

Istandard Istandard

after reaction

before reaction

(2)

in the case of BADGE, Ifunctional group and Istandard represent the intensity of CeO at 911 cm−1 and benzene group at 1605 cm-1, respectively. For 5CC-EGDE, Ifunctional group and Istandard represent the intensity of C]O at Scheme 1. The scheme of synthesis of 5CC-EGDE.

303

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Fig. 1. (A) FTIR spectra, (B) 1H NMR curves and (C) GPC chromatogram of EGDE and 5CC-EGDE.

1790 cm−1 and –CH– at 2874 cm-1, respectively. The fraction of hydrogen bonded carbonyl (fb) of urethane groups could be calculated by the following Eq. (3): Ab

fb =

Ab

εb

Test samples were sputtered with gold for 30 s to improve the resolution for observation. Dynamic Mechanical Analysis (DMA) of modified epoxy was carried out in the three point bending mode of the equipment (DMTA-Q800 TA Instruments, New castle, USA) at a heat rate of 5 ℃/min over temperature range from 25 to 250 ℃. The corresponding thermo-mechanical properties of these materials were determined as a function of temperature. The samples were scanned at a fixed frequency of 1 Hz, strain of 0.1% and amplitude of 20 u m. The crosslinking density of the materials could also be obtained from DMA, calculated according the following Eq. (4) [28]:

εb

+

Af

εf

(3)

where Ab and Af are the area below the hydrogen bonded and free carboxyl band. The ratio of hydrogen-bonded absorptivity coefficient (εb) to free carboxyl one (εf) is defined as 1.5 [26]. 1 H Nuclear Magnetic Resonance (1H NMR) measurements were recorded on a Bruker Avance 400 MHz spectrometer (Bruker Co., Ltd., Switzerland) in CDCl3. The chemical shifts were reported in parts per million relative to tetramethylsilane (TMS). Toluene was selected as internal standard substance to determine the EEW of EGDE. CEW and selectivity of carbonate product were also determined by this method [27]. Gel Permeation Chromatography (GPC) measurements were performed by employing Agilent 1100 chromatographic system (Agilent Co., Ltd., USA) with a Gel PW column of Nanofilm GPC-150 (Sepax Technologies Co., Ltd., USA). The GPC column was eluted with a THF solution at 1 mL/min at 35 ℃. Calibration was performed using polystyrene as standard. The viscosities of BADGE, 5CC-EGDE and BADGE/5CC-EGDE mixtures were measured by using a NDJ-1 rotational viscometer (Shanghai Balance Instrument Technical Co., Ltd) with a shear rate of 25 s−1. Scanning Electron Microscopy (SEM) (2800B, KYKY, China) was operated at 25 KV to characterize the surface morphology of fracture surface of BADGE/5CC-EGDE composites after unnotched impact test.

ρ=

E′ 3RT

(4)

where ρ represents the crosslinking density per unit volume (mol/cm3), E’ is rubbery modulus (MPa) at the temperature of Tg + 40 ℃ (Tg represents the glass transition temperature of material, which is determined from the temperature at maximum peak value of the tan δ curve), R is the gas constant and T was the temperature at Tg + 40 ℃. The gel time was measured by flat-knife method. The mixtures (epoxy resin with various 5CC-EGDE content) with curing agent was placed in an oven at 75 ℃ and recorded time until the mixtures were cured. The recorded time was determined to act as gel time. Three-point bending tests were carried out according to ASTM D 790-10 standards by using an electron omnipotence experiment machine (SANS-CMT6503, Shenzhen Sans Testing Machine Co., Ltd., China) with a cross-head speed of 1 mm/min. Rectangular samples with dimension of 80 × 10 × 4 mm3 and length span of 50 mm were used. 304

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

At least five specimens were tested for each sample to get an average value. Tensile tests were performed at room temperature with a speed of 2 mm/min by using dumbbell-shaped specimens according to standard ASTM D638. Experiments were carried out by using the same instruments with three-point bending tests, and more than five specimens were tested to obtain average and standard deviation values for each sample. Unnotched impact strength was performed on a JJ-20 instrumented impact machine, according to the ASTM D4812-06 standard. Measurements were performed at room temperature. The data corresponded to the average value of five specimens. The experiments were carried out on cubic samples (80 × 10 × 4 mm3).

viscosity, exorbitant temperature would cause evaporation of curing agent and excessive curing rate. In addition, excessive curing temperature would cause stress concentration in materials which affected mechanical properties of final composites. Hence, the curing temperature was determined at 75℃. Meanwhile, from Fig. 2B, the addition of 5CC-EGDE into epoxy resin also resulted in a significant decrease in viscosity, which could effectively improve the preparation processing. Gelation is the point where there is an initial formation of a threedimensional network that is no longer completely soluble [29]. Therefore, it is very important to study the gel time from a processing standpoint. The gel time of the BADGE/5CC-EGDE composites with various 5CC-EGDE contents was shown in Fig. 3A. As the increase of 5CC-EGDE content, gel time first decreased and then increased. A shortening of the gel time was observed for BADGE with low 5CC-EGDE content. The shortest gel time was obtained in the case of composite containing 5 wt% 5CC-EGDE. Similar observation was reported for modified epoxy resin by Rokicki [17]. The reduction of gel time resulted from the different reactivity of epoxy groups and cyclic carbonate groups with curing agent DETA. Functional groups with different ring size had different ring-strain. Due to the lower ring-strain energy, the activation energy in ring-opening reaction of five-membered group was lower than that of the three-membered group [30]. In order to better understand the effect of difference of reactivity on gel time, the kinetics of ring-opening reaction of BADGE and 5CC-EGDE with DETA was studied and shown in Fig. 3B. As illustrated in Fig. 3B, in the case of ring-opening reaction of 5CC-EGDE with DETA, over half of C]O intensity of cyclic carbonate weakened within 10 min and in the case of only 30% intensity of the C]O was achieved within 40 min for reaction at 25 ℃. However, the reactivity of BADGE was much lower than that of 5CC-EGDE, in which over 60% intensity of CeO of epoxy group was kept. Furthermore, the BADGE was cured in 25 min, which was ascribed to the exothermic effect from ring-opening of 5CC-EGDE accelerating curing rate. The exothermic profiles of BADGE and 5CC-EGDE cured with DETA were also presented in Fig. S1. The lower exothermic peak and less time to reach maximum heat release were observed for 5CCEGDE as a result of smaller internal stress in five-membered rings compared to three-membered one. The heat release was beneficial to accelerate reaction. As a result, the gel time of BADGE was shortened to 13.5 min after introducing 5 wt% 5CC-EGDE when cured at 75 ℃. However, as further increase of 5CC-EGDE content, the gel time of BADGE/5CC-EGDE composites increased to 21 min after incorporating 30 wt% 5CC-EGDE. This result was markedly different from previous reports [17]. The lower crosslinking density in comparison to that of epoxy resin was a main factor for longer gel time. From Fig. 3A and B, the BADGE could be cured by DETA at 25 ℃ within 25 min and the cyclic carbonate group had remained about 40% at the same conditions. The curing of BADGE phase would limit the conversion of cyclic carbonate and reduce the complete crosslinking network of composites. On the other hand, the unreacted groups could also restrict the formation of complete network (structure and crosslinking density were confirmed as shown in followings), which prolonged the gel time of the BADGE/5CC-EGDE composites. Hence, the higher 5CC-EGDE content caused the lower crosslinking density which was responsible for the longer gel time.

3. Results and discussion 3.1. Synthesis of 5CC-EGDE The product of coupling reaction of EGDE with CO2 was investigated by using FTIR, 1H NMR, and GPC measurements. FTIR spectra of EGDE and 5CC-EGDE were shown in Fig. 1A.The spectrum of EGDE showed strong absorption bands at 911 cm−1 and 855 cm−1 that correspond to epoxy groups. These epoxy groups disappeared completely after coupling with CO2. The spectrum of 5CC-EGDE showed new intense peaks centering at 1790 cm−1 and 1174 cm-1 assigned to the stretching of C]O and CeO of cyclic carbonate unit, respectively. These results confirmed the conversion of EGDE to 5CC-EGDE. As additional evidence, the formation of cyclic carbonate was further confirmed by 1H NMR (see Fig. 1B). Peaks in the region 2.56 ∼ 3.11 ppm corresponded to H atoms of epoxy group in EGDE, which were absent entirely in 5CC-EGDE. This also proved the complete conversion of EGDE. Besides, the signals in the range of 4.41 ∼ 4.85 ppm, attributed to the –CH– and –CH2– of cyclic carbonate groups. The corresponding selectivity of product was determined to be 98.6%. According to 1H NMR spectroscopic investigation, the fixed CO2 fixation and carbonate content was 19.1 wt% and 41.5 wt%, respectively, for 5CC-EGDE (as shown in Table 1). Meanwhile, EGDE and 5CC-EGDE showed EEW and CEW of 6.6 mmol/g and 4.8 mmol/g, respectively, which were very close to those values determined by titration. Besides, GPC traces of EGDE and 5CC-EGDE were performed (Fig. 1C) to show that the curve of 5CC-EGDE was similar to that of EGDE except the position of main peak. This result further proved that high cyclic carbonate selectivity of the reaction between CO2 and EGDE was obtained. This was consistent with the result of 1H NMR. The retention time of 5CC-EGDE was slightly shorter than that of EGDE, which proved the fixation of CO2 into EGDE and higher molar mass of 5CC-EGDE. 3.2. Rheology-viscosity and gelation Effect of temperature and 5CC-EGDE content on the dynamic viscosity of BADGE/5CC-EGDE mixtures was shown in Fig. 2. It was clearly shown that BADGE had a much higher viscosity than 5CC-EGDE. Therefore, the 5CC-EGDE could act as a diluent to reduce the viscosity of mixture system, which was convenient to the process of epoxy resin. From Fig. 2A, both the viscosity decreased rapidly with elevated temperature. Although the high temperature was favorable to reduce Table 1 Properties of EGDE and 5CC-EGDE.

EGDE 5CC-EGDE a

Selectivity (%)a

Cyclic carbonate (wt%)a

Fixed CO2 (wt%)a

EEW (mmol/g)a

CEW (mmol/g)a

EEW (mmol/g)b

CEW (mmol/g)b

– 98.6

– 41.5

– 19.1

6.6 –

– 4.8

6.6 –

– 5.1

Calculated from 0% and 100% epoxy conversion detected by 1H NMR.

b

Calculated by titration method (ASTM D 1652). 305

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Fig. 2. Effect of temperature (A) and 5CC-EGDE content (B) upon the viscosity (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 3. (A) Gel time of BADGE/5CC-EGDE composites as a function of 5CC-EGDE content; (B) Kinetics of curing reaction from 5CC-EGDE and DETA and BADGE and DETA at 25 ℃.

absorption peak at 1790 cm-1 confirmed the residual cyclic carbonate in final composites and the peak intensity strengthened with the increase of 5CC-EGDE content, which verified the gel time increase as a result of inadequate crosslinking network. Since a large content of polar groups such as hydroxyl, urethane groups, cyclic carbonate and secondary amine groups existed in the final system, hydrogen bonding could be formed universally among these functional groups [31]. As we know from literature [32], the hydrogen bonding was of great importance to reinforce the mechanical performances of epoxy resin. Hence, a curve-fitting method was employed to evaluate the effect of structure and 5CC-EGDE content on the hydrogen-bonding interaction. The C]O of urethane was studied by means of FTIR-peak-differentiation-imitating analysis. The FTIR spectrum of each modified epoxy was assumed to fit closely to the Gaussian function. Fig. 5 showed representative examples (BADGE/5CC-EGDE containing 5 wt% and 30 wt% 5CC-EGDE) of the experimental and fitted C = O absorptions (Only the intermolecular hydrogen bonding was taken into account). The fraction of hydrogen-bonded carbonyl group (fb) was calculated by Eq. (3) and results of curving-fitting analysis were summarized in Table 2. Due to the large amount of polar groups, especially hydroxyl resulted from ring-opening reaction from

3.3. Structure of BADGE/5CC-EGDE composites Scheme 2 showed the possible reaction pathways of BADGE/5CCEGDE mixtures with DETA during the curing process. The epoxy groups of BADGE reacted with primary and secondary amine of DETA leading to the formation of crosslinking network. The cyclic carbonate groups could also react with DETA to form urethane group. However, partial cyclic carbonate was remained, which was ascribed to rapid curing rate of BADGE phase and caused inadequate ring-opening reaction. The FTIR spectra of these cured composites with various 5CC-EGDE contents were depicted in Fig. 4. Disappearance of the epoxy ring peak at 913 cm−1 confirmed the curing reaction occurred [18]. Formation of urethane linkage as a result of ring-opening reaction between 5CCEGDE and DETA was confirmed by the urethane peak at 1703 cm-1. The signal in the region 3100 ∼ 3500 cm-1 was attributed to the stretching vibrations of OeH or NeH groups, which further proved the ringopening reaction of five-membered carbonate and three-membered epoxy with DETA. Besides, the peaks at 970 cm-1, 911 cm-1 and 863 cm1 were ascribed to epoxy groups. Upon cured by DETA, these peaks of composites obtaining different 5CC-EGDE content disappeared completely. However, in the case of the modified epoxy resin, the 306

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Scheme 2. Reaction pathways of the BADGE/5CC-EGDE mixture with DETA during the curing process.

bonding amount continued to increase due to increased urethane linkage content. The hydrogen bonding played a vital role in obstructing the formation of physical networks, which was critical to the mechanical performances of composites.

epoxy and carbonate groups, the fb value was achieved at 78.9% when 5 wt% 5CC-EGDE was introduced into composites. It was much higher than that of traditional polyurethane [33] or poly(propylene carbonate) modified by non-isocyanate polyurethane [26]. Although the fb value gradually decreased as 5CC-EGDE content increased, the hydrogen

Fig. 4. FTIR spectra of cured BADGE/5CC-EGDE composites with different 5CC-EGDE contents. 307

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Fig. 5. Curve-fitting analysis of C=O absorptions of BADGE/5CC-EGDE with (A) 5 wt% 5CC-EGDE and (B) 30 wt% 5CC-EGDE, in which the experimental spectrum is shown by the solid red line and the fitted spectra are shown by the dotted black line.

85.2 MPa by incorporating 15 wt% 5CC-EGDE, with an increase of 18.7 MPa. Interestingly, simultaneous increase of tensile strength and elongation at break was observed in this composite system. It should be noted that both the tensile strength and elongation at break decreased when 5CC-EGDE content was over 15 wt%. Similar phenomena were observed in poly(propylene carbonate)/non-isocyanate polyurethane blend, in which the intermolecular hydrogen bonding between the modifier and the matrix was considered to contribute to the toughening phenomena [34]. The cyclic carbonate groups remaining in the composites acted as a plasticizer that contributed to the increase of elongation at break. However, based on FTIR analysis, crosslinking density of composites decreased as the 5CC-EGDE content increased. There was a correlation between the crosslinking density and mechanical performances that the decrease of crosslinking density could cause the decrease of mechanical strength [35]. The tensile strength continued to increase until 5CC-EGDE content reached 15 wt% in our case. This phenomenon could be resulted from urethane linkage, which led to the formation of hydrogen bonding for physical entanglement. The hydrogen bonding and urethane linkage could reinforce interaction of molecular chains in composites system, promoting the load transfer and dissipation. Additionally, these chemical and physical bonds could also compensate the decrease of crosslinking density due to the introduction

Table 2 Curving-fitting results of the C]O stretching vibration bands of BADGE/5CCEGDE composites. 5CC-EGDE content (wt%)

Free C = O Inter C = O fb

−1

v(cm ) v(cm−1) %

0

5

10

15

20

25

30

– – –

1723.5 1696.1 78.9

1723.2 1696.4 70.6

1723.1 1697 62.9

1723.6 1697.2 57.6

1723.2 1697.8 50.4

1723.0 1698 43.1

3.4. Static mechanical properties The tensile and flexural performances of BADGE/5CC-EGDE composites were depicted in Fig. 6. From Fig. 6A, unmodified epoxy displayed brittle fracture with an elongation at break of only 4.1%, which was more than 10% after introducing 5CC-EGDE with a content of over 15 wt%. But elongation at break of the material slightly decreased when the content of cyclic carbonate was more than 25 wt%, which might be caused by the incomplete network having excessive unreacted carbonate groups. Meanwhile, the BADGE/5CC-EGDE showed moderate toughening effect. Tensile strength increased from 66.5 MPa to

Fig. 6. Tensile (A) and flexural performances (B) of BADGE/5CC-EGDE composites. 308

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Fig. 7. DMA data of BADGE/5CC-EGDE composites: (A) E’ (storage modulus) and (B) loss factor (tan δ) as a function of temperature.

epoxy had the highest Tg at 141.9 ℃. The incorporation of 5CC-EGDE led to the shifting of Tg toward a lower temperature, indicating enhancement in the flexibility of the formulation as a result of introduction of soft ether bond and alkane segment. The internal rotation and movement of segments and segment groups required less energy. Besides, the unreacted cyclic carbonate groups in composite could act as plasticizer for BADGE. An efficient plasticizing effect was achieved by these cyclic carbonate groups, especially for composite containing 30 wt% 5CC-EGDE (Tg reduced by about 41 ℃). Meanwhile, the sparser cross-linked network (lower crosslinking density as showing in following) was also responsible for Tg decease. Specifically, due to the decrease of crosslinking density, the free volume of chain segment increased and the restriction of chain motion weakened [37], which caused the Tg decease. From Table 3, it showed that crosslinking densities of all the composites were lower than that of unmodified one due to the incorporation of 5CC-EGDE, which was consistent with the analysis by FTIR. The crosslinking density was determined from the kinetic theory of the rubber elasticity Eq. (4) [36], and the density varied with the similar tendency as E’ and Tg. The composite containing the highest content of 5CC-EGDE (30 wt%) obtained the lowest crosslinking density of only 1.5 × 10−3 mol/cm3. This low crosslinking density was responsible for the weak properties of composite compared with that obtaining 15 wt% 5CC-EGDE that had a crosslinking density of 2.1 × 10−3 mol/cm3. In addition, the area under tan δ curve (shown Table 3) was also representative of the energy dissipated in the viscoelastic relaxation and could be correlated with the impact strength [37]. The relationship between area under tan δ curve and impact strength of these BADGE/ 5CC-EGDE composites was shown in Fig. 8. All the composites exhibited larger area than unmodified one, showing a maximum impact strength increase of about 140% with 15 wt% 5CC-EGDE and minimum increase of about 14% with 30 wt% 5CC-EGDE. These results indicated that the composites enhanced the ability of energy absorption during the deformation process. Overall, these results demonstrated that the introduced 5CC-EGDE improved the mechanical performances and ability to dissipate energy of applied force effectively, and the optimal 5CC-EGDE content was in the range from 15 to 20 wt%.

of 5CC-EGDE [20]. Flexural test showed a similar trend as tensile results. As 5CC-EGDE content increased, flexural strength was improved considerably from 102 MPa to 128.3 M P with 5CC-EGDE content increase up to 20 wt% and then dropped to 109 MPa when the content of modifier was further increased to 30 wt%. However, flexural modulus dropped from 2.6 GPa to 2.3 GPa continuously in the studied range due to lower crosslinking density and plasticizing effect caused by unreacted cyclic carbonate groups. 3.5. Thermo-mechanical properties Effect of 5CC-EGDE content on dynamic mechanical properties of BADGE/5CC-EGDE composites was evaluated by using DMA. Fig. 7 showed temperature dependence of storage modulus (E’) and loss factor (tan δ) for these composites. The various parameters calculated from these DMA curves were summarized in Table 3. From Fig. 7A, it was observed that all the composites passed through three states of viscoelasticity namely glassy, leathery and rubbery state [36]. The glassy state plateau with high E’ at low temperature and rubbery state plateau with low E’ at high temperature were observed for all composites. The unmodified epoxy exhibited a relatively higher E’ than 5CC-EGDE modified ones, which was due to the presence of a higher degree of crosslinking. The glass transition temperature (Tg) was affected by several factors such as crosslinking density, plasticizer and flexible chain. These BADGE/5CC-EGDE composites had various crosslinking density and different flexible chain content as result of difference in 5CC-EGDE content. In this study, Tg was determined from the temperature at maximum peak value of the tan δ curve (shown in Fig. 7B and Table 3). The appearance of a single relaxation peak confirmed the miscibility of all the components at molecular level and was also associated to a relatively homogeneous structure. It was found that the unmodified Table 3 Summary of DMA parameter for BADGE/5CC-EGDE composites. 5CC-EGDE content (wt%)

Tg (℃)

E’ at Tg + 40 ℃(MPa)

ρ×103 (mol/ cm3)

Tan δarea (a.u.)

0 5 10 15 20 25 30

141.9 134.0 123.6 120.1 116 111.9 107.0

38.2 29.6 26.1 22.6 19.8 18.1 16.1

3.3 2.7 2.4 2.1 1.9 1.7 1.5

25.0 26.6 27.8 31.0 29.1 27.9 26.1

3.6. Effect of hydrogen bonding and crosslinking density on mechanical performances Generally, the increase of 5CC-EGDE content could improve mechanical performance efficiently. But the hydrogen bonding formed between urethane linkage and other functional groups (e.g. hydroxyl and cyclic carbonate), the residual unreacted cyclic carbonate and low 309

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

(shown in Fig. S3). Meanwhile, from Fig. S5 and Table S2, the tensile and flexural strength exhibited a similar tendency with impact strength, but the modulus and Tg of composites decreased gradually along with the increase of conversion due to the lower crosslinking density and residual cyclic carbonate (showed in Fig. S3) as previously reported in the literature [18]. Simultaneously, effect of DETA amount on performances of BADGE/ 5CC-EGDE composites was also investigated. The DETA amount had great influence on the architectures of composites such as crosslinking density and unreacted cyclic carbonate. The intensity of cyclic carbonate at 1790 cm−1 reduced gradually with the increase of DETA amount increase. Meanwhile, unreacted cyclic carbonate groups disappeared completely when 20% superfluous DETA was added (see Fig. S6). The corresponding properties exhibited distinctive tendencies. From Fig. 9B, Fig. S8 and Table S3, the impact, tensile and flexural strength, crosslinking density and Tg increased slightly and then dropped with further increased DETA amount. It was ascribed to the change of structure in the composites. The architectures of composites ranged from defective to crosslinking and then linear structures along with the increase of DETA amount, which was similar with our previous report [24]. Although the excessive DETA compensated the structural defects which were due to the unreacted cyclic carbonate, the further increase of DETA amount caused the composites with linear structure. Besides, the excessive DETA might also act as plasticizer for epoxy materials. It was clear from the results that the mechanical performances improved slightly with 10% superfluous DETA amount. However, the elongation at break was reduced by about 26.5% for this formulation (shown in Fig. S8). Meanwhile, the area under tan δ curve had no significantly increase, which demonstrated that there was almost no improvement in the impact strength. Hence, the equimolar amount of DETA was reasonable for these composites. Furthermore, a small amount of unreacted carbonate groups remaining in cross-linked structure could increase the torsional vibration of the side chains which improved the β-relaxation of chain segment and enhanced toughening performances [20].

Fig. 8. Comparison of impact strength and area under tan δ curve.

crosslinking density might have significant effect on final toughening results. Therefore, in order to analyze the influence of each factor on mechanical performances, further study was conducted. Effect of hydrogen bonding was carried out by controlling the conversion of EGDE to vary the hydrogen bonding amount. The modifier (EGDE with different conversion or CEW) content was determined as 15 wt% in composite system. The FTIR and corresponding data of EGDE with different conversions were shown in Fig. S2 and Table S1. Simultaneously, the crosslinking density was regulated by increasing DETA amount, and it also resulted in higher conversion of cyclic carbonate group into urethane linkage. Fig. 9 showed the results of effect of EGDE conversion and DETA amount on the impact strength and area under tan δ curve. From Fig. 9A, the impact strength dropped slightly ascribed to the effect of EGDE which acted as diluent and supplied aliphatic chain for epoxy resin. However, this strength, consistent with area under tan δ curve, gradually increased as the conversion of EGDE increased. This phenomenon demonstrated the improvement of ability of energy dissipation, which was ascribed to the formation of urethane linkage and hydrogen bonding in composites. The urethane groups were formed by ring-opening reaction of cyclic carbonate with curing agent. Increase of EGDE conversion provided more urethane linkages, which were illustrated by the increase of peak intensity at 1703 cm−1 of urethane

3.7. Morphology and toughening mechanism The toughness behavior of the unmodified and 5CC-EGDE modified epoxy could be explained in terms of morphology observed by SEM. The SEM images of fracture surfaces of various impact samples were recorded. Representative SEM images of neat epoxy and composites

Fig. 9. Effect of (A) conversion of EGDE and (B) DETA amount on the impact strength and area under tan δ curve (the formulation of epoxy with 100% EGDE conversion or equimolar DETA amount is the same with that of epoxy containing 15 wt% 5CC-EGDE). 310

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Fig. 10. SEM images of fracture surfaces of unnotched impact samples: (A) unmodified epoxy, (B) BADGE/5CC-EGDE with 5 wt% 5CC-EGDE, (C) BADGE/5CC-EGDE with 15 wt% 5CC-EGDE.

Scheme 3. Schematic illustration of energy dissipation mechanism in 5CC-modified epoxy.

dissipate in the network. Hence, the applied force would break down the molecular chain among networks and the matrix was destroyed from a severe failure. However, for 5CC-EGDE modified epoxy, the applied force or crack propagation gradually reduced as it passed through epoxy network which obtained chemical and physical networks. The dissipation of applied force was illustrated by the colour intensity of arrow. The hydrogen bonding, unreacted cyclic carbonate and low crosslinking density promoted movement and spacer of chain segment, which improved the damping ability. Therefore, the mechanical performances of composites were improved ascribed to this toughening mechanism.

with 5 wt% and 15 wt% 5CC-EGDE were shown in Fig. 10. The SEM micrograph of the unmodified epoxy in Fig. 10A exhibited a smooth and glassy fractured surface, showing a typical brittle failure, which accounted for the low impact strength, as there was no significant energy dissipation and plastic deformation [38]. For composites with 5 wt % 5CC-EGDE, a rough fracture surface was observed. It was well-known that large plastic deformation and crazing processes could absorb the energy significantly, thus resulted in an increase of energy needed for crack propagation and formation of new surfaces. The roughness was further increased for composites with 15 wt% 5CC-EGDE, indicating that it consumed more energy to prevent the crack propagation. Scheme 3 showed the toughening and damping mechanism of these BADGE/ 5CC-EGDE composites. In the case of unmodified epoxy, the energy transfer didn’t occur through the molecular segment and didn’t

311

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

Table 4 Comparison of increments in mechanical performances among various 5CC-modified composites. References

Ghanbaralizadeh etc. [20] Rokicki etc. [39] Parzuchowski etc. [40] Parzuchowski etc. [41] In this work

Type of cyclic carbonate

A part of 5CC groups in epoxy resin A part of 5CC groups in epoxy resin Addition of hyperbranched polyether containing 5CC Addition of soybean Oil containing 5CC Addition of bifunctional 5CC-EGDE

Viscosity of modifier at 25 ℃ (Pa·s)

Maximum increase ratio of mechanical performances Gain in impact strength (%)

Gain in tensile strength (%)

Gain in flexural strength (%)

– – ∼ 13

∼ 100 ∼ 97 ∼ 105

– Slightly decrease ∼ 25

∼ 20 – Decrease

6 2.5

∼ 117 ∼ 140

∼ 53 ∼ 28

– ∼ 26

by adjusting the EGDE conversion and DETA amount. It was demonstrated that urethane linkage, hydrogen bonding, unreacted carbonate and low crosslinking density were essential to improve the toughening effect. Thus, in this contribution, we have demonstrated the use of CO2sourced cyclic carbonate for toughening epoxy resin, obtaining increased flexural and tensile performances, elongation as well as impact strength.

3.8. Comparison of 5CC-modified epoxy resin between previous studies and this work Based on the above mechanical performances of BADGE/5CC-EGDE composites, we compared the relative increments of impact strength, tensile strength and flexural strength of various 5CC-modified epoxy resin systems after incorporation of cyclic carbonate groups [20,39–41], as shown in Table 4. The epoxy systems chosen in the table were modified by reacting with CO2 or addition of cyclic carbonate, which were comparable to the related values of the epoxy resins system used in this work. It could be found that the composites in this work exhibited the most effective increase of mechanical performances. Especially, the impact strength was much higher when compared to other 5CC-modified epoxy resins. The epoxy, modified by replacing a part of epoxy groups with cyclic carbonate, seemed to have the lower toughness results than those with addition of cyclic carbonate. It might be ascribed to the weakness of cyclic carbonate causing the urethane produce of ring-opening reaction with low molar mass and side reactions [42]. The defect in epoxy network generated by inherent carbonate groups was more serious than that caused by addition of modifier obtaining cyclic carbonate groups. In addition, Parzuchowski etc. [40,41] reported the modified epoxy resin with addition of cyclic carbonate. The improvement of mechanical performances of his reports was lower than this work as a result of high viscosity of modifiers. Although we did know that the increment of mechanical properties depended on not only the urethane linkage and hydrogen bonding, but also other factors such as dispersion of substances, which the high viscosity affected the homogeneity of epoxy and modifier. The significance of these results consisted in the fact that the 5CC-EGDE was indeed an effective toughening agent to improve the related performances of epoxy resins.

Acknowledgements The authors gratefully thank the financial support from “the Major Science & Technology Project of Shanxi Provence (No. MD2014-10 and 201603D312007)” and “Youth scientific funds of Shanxi Provence (No. 2015021043)”. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jcou.2018.05.020. References [1] T. Sakakura, J.C. Choi, H. Yasuda, Transformation of carbon dioxide, Chem. Rev. 107 (6) (2007) 2365–2387. [2] L.-F. Xiao, F.-W. Li, C.-G. Xia, An easily recoverable and efficient natural biopolymer-supported zinc chloride catalyst system for the chemical fixation of carbon dioxide to cyclic carbonate, Appl. Catal. A-Gen. 279 (1-2) (2005) 125–129. [3] T. Werner, N. Tenhumberg, Synthesis of cyclic carbonates from epoxides and CO2 catalyzed by potassium iodide and amino alcohols, J. CO2 Util. 7 (2014) 39–45. [4] M. Mikkelsen, M. Jorgensen, F.C. Krebs, The teraton challenge. A review of fixation and transformation of carbon dioxide, Energ. Environ. Sci. 3 (1) (2010) 43–81. [5] S.G. Prolongo, G. del Rosario, A. Urena, Comparative study on the adhesive properties of different epoxy resins, Int. J. Adhes. Adhes. 26 (3) (2006) 125–132. [6] J. Simal-Gandara, S. Paz-Abuin, L. Ahrne, A critical review of the quality and safety of BADGE-based epoxy coatings for cans: implications for legislation on epoxy coatings for food contact, Crit. Rev. Food Sci. 38 (8) (1998) 675–688. [7] A. Kausar, I. Rafique, B. Muhammad, Review of applications of Polymer/Carbon nanotubes and Epoxy/CNT composites, Polyme-Plast. Technol. 55 (11) (2016) 1167–1191. [8] T.D. Chang, S.H. Carr, J.O. Brittain, Studies of epoxy-resin systems. B. Effect of crosslinking on the physical-properties of an epoxy-resin, Polym. Eng. Sci. 22 (18) (1982) 1213–1220. [9] R.E. Neisiany, S.N. Khorasani, M. Naeimirad, J.K.Y. Lee, S. Ramakrishna, Improving mechanical properties of Carbon/Epoxy composite by incorporating functionalized electrospun polyacrylonitrile nanofibers, Macromol. Mater. Eng. 302 (5) (2017). [10] B. Konstantellos, E. Sideridis, The effect of plasticizer on the mechanical and acoustical properties of epoxy polymers, Plast. Rub. Compos. Pro. 22 (5) (1994) 261–267. [11] M. Kostrzewa, B. Hausnerova, M. Bakar, E. Siwek, Effects of various polyurethanes on the mechanical and structural properties of an epoxy resin, J. Appl. Polym. Sci. 119 (5) (2011) 2925–2932. [12] T. Pokropski, A. Balas, Modifications of epoxy resins and polyurethanes. part i. Polymers with oxazolidone rings, Polimery 48 (6) (2003) 417–420. [13] H. Singh, A.K. Jain, Ignition, combustion, toxicity, and fire retardancy of polyurethane foams: a comprehensive review, J. Appl. Polym. Sci. 111 (2) (2008) 1115–1143. [14] R.W. Carolyn, N.J. Anderson, D.K. Bonauto, Prevention guidance for isocyanateinduced asthma using occupational surveillance data, J. Occup. Environ. Hyg. 10 (11) (2013) 597. [15] J. Guan, Y. Song, Y. Lin, X. Yin, M. Zuo, Y. Zhao, X. Tao, Q. Zheng, Progress in study of Non-isocyanate polyurethane, Ind. Eng. Chem. Res. 50 (11) (2011) 6517–6527. [16] E. Delebecq, J.P. Pascault, B. Boutevin, F. Ganachaud, On the versatility of

4. Conclusions In this work, an effective epoxy toughening agent was synthesized by using EGDE and CO2 as a sustainable feedstock by cycloaddition reaction. This product containing five-membered cyclic carbonate was successfully synthesized with a selectivity of 98.6%. Various 5CC-EGDE contents were incorporated into epoxy resin to reduce the viscosity during preparation process and improve the performances. The sparser network structure could be achieved by introducing 5CC-EGDE into epoxy, which contributed to the movement of chain segments. Remarkable improvements in tensile, flexural and impact strength were obtained, demonstrating the strong toughening effect of 5CC-EGDE. The tensile strength and elongation at break increased by 28.1% from 66.5 MPa to 85.2 MPa and by 250% from 4% to over 10%, respectively, in case of 15 wt% 5CC-EGDE incorporated in epoxy resin. Despite modulus decrease, the maximum increase of flexural strength was achieved over 25%. The DMA results showed the ability to absorb energy was improved, which was consistent with the impact strength and area under tan δ curve. In addition, effect of hydrogen-bonding interactions and crosslinking density on the performances was also studied 312

Journal of CO₂ Utilization 26 (2018) 302–313

J. Ke et al.

[17] [18]

[19]

[20]

[21] [22] [23] [24]

[25]

[26]

[27]

[28]

[29]

Polym. Int. 53 (1) (2004) 78–84. [30] H. Tomita, F. Sanda, T. Endo, Polyaddition behavior of bis(five- and six-membered cyclic carbonate)s with diamine, J. Polym. Sci. Pol. Chem. 39 (6) (2001) 860–867. [31] V.S.G. Silverajah, N.A. Ibrahim, N. Zainuddin, W.M.Z.W. Yunus, H. Abu Hassan, Mechanical, thermal and morphological properties of poly(lactic acid)/Epoxidized Palm olein blend, Mol. Cell. Probes 17 (10) (2012) 11729–11747. [32] J.N. Clark, J.H. Daly, A. Garton, Hydrogen-bonding in epoxy-resin poly (epsiloncaprolactone) blends, J. Appl. Polym. Sci. 29 (11) (1984) 3381–3390. [33] N. Liu, Y. Zhao, M. Kang, J. Wang, X. Wang, Y. Feng, N. Yin, Q. Li, The effects of the molecular weight and structure of polycarbonatediols on the properties of waterborne polyurethanes, Prog. Org. Coat. 82 (2015) 46–56. [34] G. Ren, X. Sheng, Y. Qin, X. Chen, X. Wang, F. Wang, Toughening of poly(propylene carbonate) using rubbery non-isocyanate polyurethane: transition from brittle to marginally tough, Polymer 55 (21) (2014) 5460–5468. [35] S.K. Sahoo, S. Mohanty, S.K. Nayak, Toughened bio-based epoxy blend network modified with transesterified epoxidized soybean oil: synthesis and characterization, RSC Adv. 5 (18) (2015) 13674–13691. [36] S. Kumar, S.K. Samal, S. Mohanty, S.K. Nayak, Epoxidized soybean oil-based epoxy blend cured with anhydride-based cross-linker: thermal and mechanical characterization, Ind. Eng. Chem. Res. 56 (3) (2017) 687–698. [37] A.M. Diez-Pascual, A.L. Diez-Vicente, Epoxidized soybean oil/ZnO biocomposites for soft tissue applications: preparation and characterization, ACS Appl. Mater. Inter. 6 (19) (2014) 17277–17288. [38] B.C. Kim, S.W. Park, D.G. Lee, Fracture toughness of the nano-particle reinforced epoxy composite, Compos. Struct. 86 (1-3) (2008) 69–77. [39] G. Rokicki, M. Lewandowski, Epoxy-resins modified by carbon-dioxide, Angew. Makromol. Chem. 148 (1987) 53–66. [40] P.G. Parzuchowski, M. Kizlinska, G. Rokicki, New hyperbranched polyether containing cyclic carbonate groups as a toughening agent for epoxy resin, Polymer 48 (7) (2007) 1857–1865. [41] P.G. Parzuchowski, M. Jurczyk-Kowalska, J. Ryszkowska, G. Rokicki, Epoxy resin modified with soybean oil containing cyclic carbonate groups, J. Appl. Polym. Sci. 102 (3) (2006) 2904–2914. [42] V. Besse, F. Camara, F. Mechin, E. Fleury, S. Caillol, J.P. Pascault, B. Boutevin, How to explain low molar masses in PolyHydroxyUrethanes (PHUs), Eur. Polym. J. 71 (2015) 1–11.

urethane/urea bonds: reversibility, blocked isocyanate, and non-isocyanate polyurethane, Chem. Rev. 113 (1) (2013) 80–118. G. Rokicki, C. Wojciechowski, Epoxy-resin modified by aliphatic cyclic carbonates, J. Appl. Polym. Sci. 41 (3-4) (1990) 647–659. M. Khoshkish, H. Bouhendi, M. Vafayan, Effect of CO2 fixation reaction on the thermal and dynamic mechanical properties of tetrafunctional epoxy resin, J. Mater. Sci. 49 (19) (2014) 6813–6819. M. Khoshkish, H. Bouhendi, M. Vafayan, Modification of bifunctional epoxy resin using CO2 fixation process and nanoclay, Mater. Chem. Phys. 147 (3) (2014) 382–389. R. Ghanbaralizadeh, H. Bouhendi, K. Kabiri, M. Vafayan, A novel method for toughening epoxy resin through CO2 fixation reaction, J. CO2 Util. 16 (2016) 225–235. X.P. Gu, I. Ikeda, M. Okahara, Synthesis of glycol diglycidyl ethers using phasetransfer catalysis, Synthesis-Stuttgart 6–7 (1985) 649–651. D.G. Lahr, B.H. Shanks, Kinetic analysis of the hydrogenolysis of lower polyhydric alcohols: glycerol to glycols, Ind. Eng. Chem. Res. 42 (22) (2003) 5467–5472. J. Zhang, J.B. Li, S.B. Wu, Y. Liu, Advances in the catalytic production and utilization of sorbitol, Ind. Eng. Chem. Res. 52 (34) (2013) 11799–11815. J.X. Ke, X.Y. Li, F. Wang, S.A. Jiang, M.Q. Kang, J.W. Wang, Q.F. Li, Z.J. Wang, Non-isocyanate polyurethane/epoxy hybrid materials with different and controlled architectures prepared from a CO2-sourced monomer and epoxy via an environmentally-friendly route, RSC Adv. 7 (46) (2017) 28841–28852. B. Ochiai, J.-i. Nakayama, M. Mashiko, Y. Kaneko, T. Nagasawa, T. Endo, Synthesis and crosslinking reaction of poly(hydroxyurethane) bearing a secondary amine structure in the main chain, J. Polym. Sci., Part. A: Polym. Chem. 43 (23) (2005) 5899–5905. L.J. Chen, Y.S. Qin, X.H. Wang, X.J. Zhao, F.S. Wang, Plasticizing while toughening and reinforcing poly(propylene carbonate) using low molecular weight urethane: role of hydrogen-bonding interaction, Polymer 52 (21) (2011) 4873–4880. J.X. Ke, X.Y. Li, F. Wang, M.Q. Kang, Y.L. Feng, Y.H. Zhao, J.W. Wang, The hybrid polyhydroxyurethane materials synthesized by a prepolymerization method from CO2-sourced monomer and epoxy, J. CO2 Util. 16 (2016) 474–485. Y. Chen, S. Zhou, G. Gu, L. Wu, Microstructure and properties of polyester-based polyurethane/titania hybrid films prepared by sol–gel process, Polymer 47 (5) (2006) 1640–1648. R.J. Varley, Toughening of epoxy resin systems using low-viscosity additives,

313