Epoxy

Subscriber access provided by RUTGERS UNIVERSITY

Article

Effect of Interfacial Bonding on Interphase Properties in SiO2/ Epoxy Nanocomposite: A Molecular Dynamics Simulation Study Zhikun Wang, Qiang Lv, Shenghui Chen, Chunling Li, Shuangqing Sun, and Songqing Hu ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.5b11810 • Publication Date (Web): 01 Mar 2016 Downloaded from http://pubs.acs.org on March 2, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Applied Materials & Interfaces is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Effect of Interfacial Bonding on Interphase Properties in SiO2/Epoxy Nanocomposite: A Molecular Dynamics Simulation Study Zhikun Wanga, Qiang Lva, Shenghui Chena, Chunling Lia,b, Shuangqing Suna,b, Songqing Hua,b* a

College of Science, China University of Petroleum, Qingdao 266580, Shandong Province, P.R.

China b

Key Laboratory of New Energy Physics & Materials Science in Universities of Shandong (China

University of Petroleum)

ABSTRACT: Atomistic molecular dynamics simulations have been performed to explore the effect of interfacial bonding on the interphase properties of a nanocomposite system consisted of a silica nanoparticle and the highly cross-linked epoxy matrix. For the structural properties, results show that interfacial covalent bonding can broaden the interphase region by increasing the radial effect range of fluctuated mass density and oriented chains, as well as strengthen the interphase region by improving the thermal stability of interfacial van der Waals excluded volume and reducing the proportion of cis conformers of epoxy segments. The improved thermal stability of the interphase region in the covalently bonded model results in an increase of ~21 K in the glass transition temperature (Tg) compared to that of the pure epoxy. It is also found that interfacial covalent bonding mainly restricts the volume thermal expansion of the model at temperatures near or larger than Tg. Furthermore, investigations from mean-square displacement and fraction of immobile atoms point out that interfacial covalent and non-covalent bonding induces lower and higher mobility of interphase atoms than that of the pure epoxy, 1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

respectively. The obtained critical interfacial bonding ratio when the interphase and matrix atoms have the same mobility is 5.8%. These results demonstrate that the glass transitions of the interphase and matrix will be asynchronous when the interfacial bonding ratio is not 5.8%. Specifically, the interphase region will trigger the glass transition of the matrix when the ratio is larger than 5.8%, whereas restrain the glass transition of the matrix when the ratio is smaller than 5.8%. Keywords: interphase; nanocomposite; epoxy; glass transition; molecular dynamics simulation

1. INTRODUCTION Nanoparticles (NPs) are expected to be excellent candidates for improving the thermodynamic, mechanical or viscoelastic properties of cross-linked epoxy matrices.1-6 It has been reported that the resulting nanocomposite materials (NCMs) with novel properties have great potential for a variety of applications in electronic, coating, automobile, aerospace etc.7-13 Overall, the properties of NCMs depend on a combined effect of the polymer matrix and the filler since both cross-linked epoxy resins and NPs are a diverse class of materials, and the specific properties of NCMs vary widely within these diversities. In the NCMs, matrix polymers in the vicinity of NPs inevitably experience the structural and dynamical perturbations caused by NPs. Almost all current researches attribute such perturbations to the existence of a layered polymer structure (i.e., interphase) around the NP, manifested as peaks of enhanced density, bond orientation and restrained atom mobility.14-20 2


Page 2 of 42

Page 3 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In order to predict and improve the overall properties of NCMs, it is essential to find out how the structural and dynamical properties of the interfacial region depend on the characteristics of the filler and the matrix. Covalent grafting, in particular, is regarded as a promising way to improve the load transfer efficiency by implementing bonded interactions between the filler and the matrix. Some independent experiments have verified that the interphase in NCMs often has longer relaxation time than the matrix, which is mainly caused by the stronger adhesion of matrix chains to the surface.21-25 On the contrary, some experimental studies also reported that the relaxation time of interfacial polymer chains in NCMs can be reduced.21-23,26,27 These studies indicate the complexity of the NCMs. On the other hand, the altered properties of polymer NCMs have been extensively studied by atomistic or coarse-grained (CG) molecular dynamics (MD) simulations. Some of these simulations have focused on the question of how deep the structural and dynamical properties of the interphase are affected by the shape, size and grafting density of the NPs and the entangled or unentangled polymer matrix.8,17,28-30 Also, a variety of simulation methods have been proposed to investigate the spatially resolved structural, mechanical and rheological properties of NCMs.15,31-36 The available results indicate that the interphase region extends only a few nanometers from the filler surface, beyond such a distance all dynamical or structural properties show bulk-like behavior. Only a few recent studies have attempted to explore the variations of the spatial resolved properties of the polymer matrix in the vicinity of NPs during the glass transition processes of NCMs. 3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Most current research efforts have a goal of providing a quantitative description of the temperature dependent properties of the interphase region. The glass transition temperature (Tg) of the NCM is an important signature of the effect of NP inclusion on the thermodynamic property of the polymer matrix. Tg is affected by the characteristics of the NP (shape, size, surface modification and extent of dispersion), the polymer matrix (specific chemistry, conversion degree), the additives (surfactants, residual solvent, impurities), etc. Briefly, it was found that5,6,37-40: (1) the creation of a denser or more dynamically restricted interphase between polymer matrix and NP contributes significantly to the increase of Tg, (2) unfavorable interaction between NP and polymer matrix causes a depression of Tg, (3) surface functionalization of NP results in a stronger interfacial interaction and an increased Tg compared to bare NP, (4) more compact conformation of chain segments (e.g., tangential orientation) and fewer free volumes in the spatially resolved interphase lead to a larger Tg. Generally speaking, the strength and proportion of interfacial regions often contribute to the fluctuation of Tg. Some recent efforts also focused on elucidating the influence of the dynamical heterogeneity between interphase region and bulk region on the glass transition behavior of NCMs, especially for glassy materials exhibiting the coexistence of the mobile and immobile domains in the vicinity of Tg. Some experiments and simulations have demonstrated glass transition to be a dynamic behavior which strongly depends on the heterogeneous character of the system. Forrest et al.21-23 studied the relaxation time between the polymer matrix and its free surface, and 4


Page 4 of 42

Page 5 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


revealed that the interface region exhibits higher molecular mobility than the matrix when the temperature is smaller than Tg, and the system transitions from ‘whole-film’ flow to ‘surface localized’ flow over a narrow temperature range near the bulk Tg. Long and Lequeux 41 proposed that the glass transition is governed by the percolation of the immobile domains which are generated by density fluctuations in the system. These studies are conducive to better understand the macroscopically observed peculiarities in NCMs during the glass transition process. The present study aims to answer the following two questions: (1) To what degree does the interfacial bonding status influences the volumetric, structural and dynamical properties of the interphase region in the NCMs? (2) What effects does the interfacial bonding status has on the glass transition of the interphase region in the NCMs? To answer the first question, we created two nanocomposite (NC) models (i.e., interfacial bonded - IB and interfacial nonbonded - INB) consisting of a functionalized silica nanoparticle (FSNP) and the cross-linked epoxy matrix and then compared these two models in the volumetric, structural and dynamical properties. To answer the second question, we performed a dynamic cooling process on the NC models and analyzed the temperature dependent properties of the volumetric and dynamic heterogeneity properties during glass transition. In addition, a pure cross-linked epoxy model was studied as a reference. All the investigations were performed based on all-atom MD simulation method. 2. SIMULATION METHOD 2.1 Molecular models and force field 5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The epoxy matrix consists of diglycidyl ether of bisphenol A (DGEBA) as the epoxy monomer and isophorone diamines (IPD) as the cross-linker, while a functionalized silica nanoparticle (FSNP) is used as the filler (Figure 1). All MD simulations in this work were carried out using the Materials Studio software (Accelrys Inc.). The COMPASS42 force field was applied to describe both inter- and intramolecular interactions. A velocity Verlet algorithm with a time integration step of 1 fs was used for atom motion equations. The nonbonded vdW (van der Waals) interactions were truncated at 12.5 Å. Atom-based and Ewald43 summation methods were used for vdW and coulomb interactions, respectively. In addition, Nosé-Hoover44 thermostat and Berendsen45 barostat were used to control the temperature and pressure in all NPT (isothermal-isobaric ensemble) simulations, respectively.

Figure 1. Chemical structures of DGEBA (epoxy monomer) (a), IPD (cross-linker) (b) and FSNP (functionalized silica nanoparticle) (c). In (c), oxygen, nitrogen, carbon, silicon and hydrogen atoms are in red, blue, gray, yellow and white, respectively. 6


Page 6 of 42

Page 7 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2.2 Structure building procedure The FSNP was trimmed from bulk α-quartz in the form of spherical atom cluster with a radius of 10 Å. All silicon free radicals of the spherical nanosilica were first grafted by oxygen atoms to mimic the surface oxidation process. Then, 12 of 104 (the grafting ratio is 11.5%) oxygen free radicals on the FSNP surface were randomly chosen to covalently bond with –Si(OH)2(CH2)3NH2 (Figure 1c), i.e., the linker between the nanoparticle and the epoxy matrix. The remaining oxygen free radicals were saturated with hydrogen atoms. After building the FSNP model, the Amorphous Cell module in Materials Studio software was used to construct the unit cell of NC, in which one FSNP was embedded in the center of the cell and surrounded by 160 DGEBAs and 80 IPDs. This model contains 11244 atoms in total, including 10560 atoms of the epoxy matrix and 684 atoms of the FSNP. Periodic boundary conditions were applied to xyz directions of the model structure. This original model was equilibrated at 700 K and 1 atm through 1000 ps (picosecond) of NPT ensemble simulation before it was used to build the IB and INB NC models. Two successive cyclic dynamic processes were used to build the IB model: (1) Interfacial bonding process. All close contacts between the linker nitrogen atoms on FSNPs and the epoxy terminal carbon atoms on DGEBAs were first monitored and the closest distance was detected. If the closest distance was within a pre-defined threshold distance (d, ranging from 4 to 8 Å), a new covalent bond will be formed manually according to the real chemical reaction process46. Otherwise, a 30 ps NPT ensemble simulation at 700 K was performed on the structure and a larger threshold 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

distance (d+1 Å) was used for the same contact detection. This process was repeated until all available contacts were chemically bonded. (2) Epoxy cross-linking process. We conducted a widely used method46-49 to construct the highly cross-linked epoxy matrix. Firstly, distances between all epoxy terminal carbon atoms on DGEBAs and amino nitrogen atoms on IPDs were monitored and the closest distance was detected. When the distance is smaller than d=4 Å, a new covalent bond will be formed manually between the epoxy and amino groups, and a 30 ps NPT ensemble simulation at 700 K was subsequently conducted on this model to remove the internal stresses and geometric distortions. This process was repeated until d is too small for further cross-linking. Then, a larger distance, d+1 Å, was used as the new threshold distance for the next cross-linking cycle. The whole process will be stopped when d equals 8 Å and no new contact can be found anymore. During the two processes mentioned above, ring spearing (i.e., the case that a benzene ring is crossed by a chemical bond) must be checked and avoided right after the formation of each new chemical bond. The final conversion degree of the epoxy matrix of the IB model is 89.7% (i.e., 287 of 320 epoxy groups reacted) at the largest

d. To build the INB model, as there is no chemical bond between the FSNP and the epoxy matrix, only the ‘bulk epoxy cross-linking’ process was performed on all DGEBA and IPD monomers in the box. Moreover, a pure epoxy model without embedding FSNP was also constructed based on the method mentioned above. It should be noted that the three models contain same number of DGEBA and IPD monomers, as well as same number of reacted epoxy and amino groups on these two 8


Page 8 of 42

Page 9 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


monomers. Therefore, the final conversion degrees of the epoxy matrices in the three models are all 89.7%. 2.3 Simulation procedure The initial configurations (pure epoxy, INB and IB models) were first equilibrated by NPT ensemble simulations for 3000 ps at 700 K. Then, they were gradually cooled down to 275 K by reducing the temperature by step of 25 K/1000 ps. After cooling down, the last 500 ps of the computed trajectory files at each temperature was used to study the volumetric and dynamical properties as a function of temperature. Moreover, we also conducted an additional 3000 ps NPT simulation on the output configuration at temperatures of 600 K and 300 K to further analyze the structural and dynamic properties of the three models. It is important to note that atoms in the FSNP were not constrained during all NPT ensemble simulations. In order to minimize the random error in the simulation, the whole process was repeated three times by rebuilding the cross-linked models and averaging the calculation results. 3. RESULTS AND DISCUSSION 3.1 Structural properties 3.1.1 Mass density In order to study the aggregation behavior of the epoxy atoms under different interfacial bonding statuses, the spatially resolved density of the epoxy around the FSNP was calculated in this work. The normalized mass densities of the epoxy in the INB and IB models were plotted as a function of the radial distance from the mass center of the FSNP at two representing temperatures, 300 K and 600 K (Figure 2a). 9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The linker molecules are not included in the density data. Results show that there is a low-density region when the radial distance is slightly larger than 10 Å. As the radial distance increases, the densities increase rapidly and peak at about 13~15 Å, after which the densities exhibit apparent fluctuations and finally reach the bulk average density. The whole region from the NP surface to the radial position where the density reaches the bulk average density was defined as the interphase region. A structural representation of the interphase region around the FSNP is shown in Figure 2c. The effect of interfacial bonding on spatial mass density distribution is mainly discussed through the thicknesses of the vdW excluded region and interphase region in this work.

10


Page 10 of 42

Page 11 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Aggregation behavior of the interfacial epoxy atoms in the IB and INB models. (a) Normalized radial mass densities at 300 K and 600 K; (b) Thickness of the vdW excluded region as a function of temperature; (c) Illustration of the interphase structure around the FSNP.

(1) Thickness of the vdW excluded region In this work, the initial low-density region is thought to associate with the interfacial vdW excluded volume.8,50 The thickness (δ) of this region was precisely defined by the equation below based on the density profiles,51,52 ∞

3 (rNP + δ )3 - rNP = 3∫ r 2 (1 − ρr ρ0 )dr rNP

(1)

where ρr/ρ0 is the normalized radial density at a given distance r from the mass center of FSNP. The δs of two models were also plotted as a function of temperature, as shown in Figure 2b. Results show that the initial δ in the INB model is smaller than that in the IB model. This may be due to the fact that the obtained normalized density at the peak position of the INB model is much larger than that of the IB model (300 K, Figure 2a), and a larger normalized density will make the integrand of equation (1) smaller. However, as the temperature increases, the δ in the INB model grows faster than that in the IB model. When the temperature reaches 350 K, the δ in the INB model starts to surpass 11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the IB model at the same temperature. The slower growth rate of the δ in the IB model indicates that the interfacial covalent bonding in the IB model restrained the thermal expansion of the vdW excluded region than that in the INB model. Previous studies5,6,20 also reported that the INB model with non-covalent bonding between the NP and the polymer matrix can shrink or stretch more easily than the IB model, leading to a larger amplitude of the interphase density as the temperature changes. In addition, four representing δs for the two models at 300 K and 600 K were calculated (i.e., INB: 2.54 (300 K) and 3.17 (600 K), IB: 2.62 (300 K) at and 3.01 (600 K)). These δs are supposed to be related to the corresponding density peaks as shown in Figure 2a. Note that, the density peak of the INB model is higher at 300 K but lower at 600 K than that in the IB model. We speculate from these results that: (1) at lower temperatures (350 K), the interfacial covalent bonding in the IB model can reduce the thermal expansion tendency of the vdW excluded region and result in a smaller δ than that in the INB model, thus increase the density peak. (2) Thickness of the interphase region In this work, the thickness of the interphase region (h) was also calculated based on the density profiles. The results show that the h of the IB model is ~10 Å at 300 K and ~13 Å at 600 K, which is slightly larger than that of the INB model at the same temperature. The approximate value of h is found to be in line with previous studies 12


Page 12 of 42

Page 13 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


employing all-atom simulations on similar NC models.2,4 Considering the difference between the two NC models, we deduced that the interphase region near the FSNP can be broadened by the interfacial covalent bonding between the FSNP and the epoxy matrix. This will be further discussed from other properties of the considered models in the present work. 3.1.2 Chain orientation We next consider how the interfacial bonding can influence the alignment distribution of epoxy segments. The orientation order parameter (S) has been used to analyze the spatial orientation of the representing segment (i.e., phenyl ring) on the epoxy monomer at 300 K and 600 K. S is calculated by the following equation: S = 1 (3 cos 2 θ − 1) , where θ represents the orientation angle of the segment. S 2 ranges between -0.5 and 1.0, corresponding to tangential or radial orientation of the phenyl ring with respect to the FSNP surface, respectively. A random orientation is characterized by S=0. The results are shown in Figure 3. The inset illustrates the definition of θ . The obtained results are accumulated in 0.5 Å thick bins, and the geometry center of the phenyl ring is used to determine the radial distance.

13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 42

Figure 3. Radial order parameter S of the phenyl rings on epoxy segments in the IB and INB models at 300 K and 600 K. The inset indicates the definition of the orientation angle ( θ ): the

r

r

angle between the normal vector ( u1 ) of the NP surface and the vector ( u2 ) connecting C1 and C4 carbon atoms on the phenyl ring.

Our results show that the S of the nearest epoxy segment to the FSNP surface is about -0.4. As the separation gets larger, S starts to increase rapidly and comes to a peak at ~0.2 when the radial distance increases to 15~18 Å, after which S starts to decrease and gradually reach about 0 showing the random orientation. The results indicate that the epoxy segments tend to align tangentially to the particle surface, and the tendency will decrease as the separation increases. This observation has been confirmed by some atomistic10,17,52 and coarse-grained16,53 MD simulations on various NCMs. For this tangential preference, it is widely accepted that the excluded volume effect and the interfacial attraction can flatten the chain configuration against the particle surface to approach it closely.17 In addition, Figure 3 also indicates that the S in the IB model needs larger separation to reach the average bulk value than that in the INB model at 14


Page 15 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


the same temperature. Even though the torsion considered here is only an representing torsion on the epoxy segment, we can learn that the interfacial covalent bonding in the IB model have a larger influence range on the segmental alignment of the epoxy matrix, i.e., the IB model has a longer interphase region dimension. 3.1.3 Chain conformation Another question is how the interfacial bonding affects the conformer of the epoxy segment. The cis conformers of a representing torsion on the epoxy segment (Cphenyl-O-C-C, the most flexible torsion48) from the MD trajectories were counted and the proportion to the total conformers (% cis) was used to reflect the conformer variation. Figure 4 shows the % cis as a function of the radial distance from the mass center of FSNP at 300 K and 600 K. The conformer of a torsion is defined to be cis if the torsion angle is within [-180º, -120º] or [120º, 180º], and the midpoint of bond O-C is used to determine the radial distance in the present work.

15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Radial % cis distributions of the Cphenyl-O-C-C torsion angle in the IB and INB models at 300 K and 600 K (a), and an example of the Cphenyl-O-C-C torsion angle distribution in the pure epoxy matrix at 300 K (b).

It can be observed that all 4 curves have a low the % cis in the vicinity of the FSNP surface. This is reasonable since the torsional motions of epoxy segments near the NP surface will be hindered by the epoxy-surface attraction and the steric hindrance effect.8,50 The result also indicates that the increasing temperature and interfacial covalent bonding induce a larger reduction of % cis. However, as the separation increases, the % cis goes up and reaches an average bulk value of about 80% at around 15 Å, which suggests that the FSNP inclusion has a smaller influence range on the chain conformer (~5 Å) than the mass density (10~13 Å) and the chain orientation (8~12 Å). Barbier33 have reported that although the NP inclusion results in longer relaxation time for the polymer matrix to reach an equilibrium state, the conformations of polymer chains will remain basically the same if the polymer matrix is rigid enough. Therefore, the limited influence range of NP inclusion on % cis is mainly attributed to the rigid characteristic of the highly cross-linked epoxy matrix. 3.2 Volumetric properties 3.2.1 Overall glass transition In order to explain the influence of the interfacial bonding status on the volumetric properties of the epoxy matrix, the temperature dependence of the specific volumes (i.e., reciprocal of densities) of the simulated models was investigated (Figure 5). Glass transition can be observed as the variation in the slope of the plot during the 16


Page 16 of 42

Page 17 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


cooling process. The temperature ranges of 275~400 K and 575~700 K are used to determine the linear least-squares fitting lines of the glassy and rubbery states, respectively, and the specific temperature at the intersection of the two fitting lines is taken as Tg.

Figure 5. Specific volumes of the pure epoxy (triangles), INB (squares) and IB (circles) models as a function of temperature. The solid lines are the linear least-squares fitting lines in the high and low temperature ranges. For clarify, error bars are shown only at 275 K and 700 K.

Figure 5 shows that the inclusion of a high-density silica into the pure epoxy matrix greatly reduces the specific volume of the entire system. As for the two NC models, the specific volume of the IB model is smaller than that of the INB model in the rubbery state and near Tg. However, when the temperature is lower than 425 K, the specific volumes of the two models become coincide with each other. Since the only difference between the two NC models is the interfacial bonding status, it is speculated that the interfacial covalent bonding mainly plays a restriction effect on the thermal expansion of the system volume near or above Tg, while in the glassy state, 17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

and there is no obvious difference between the effects of the interfacial covalent and non-covalent bonding status. The results also indicate that the Tg of the pure epoxy model is 462 ± 9 K, which is 26 K higher than the experimental value (436 K) measured by differential scanning calorimeter (DSC).54 This difference is mainly attributed to the faster cooling rate used in this work (25 K/1000 ps) than that in the DSC test (7.5 K/min ), which is supported by many previous studies for both linear vinyl polymers55,56 and cross-linked epoxy resins57,58. In addition, the different volume-temperature relations of the two models have generated different Tgs. The obtained Tg of the INB system (456 ± 11 K) is very close to that of the pure epoxy within statistical uncertainties, while the obtained Tg of the IB system (483 ± 7 K) is found to be 21 K higher than that of the pure epoxy. These observations highlight that the strength of the interfacial connections has an essential effect on the thermal motion of epoxy atoms. Stronger interfacial interactions (i.e., the covalent bonding) in NCMs will lead to a larger Tg, while weak interfacial interactions (i.e., the non-covalent bonding) may even induce a large decrease of Tg.4-6,39 3.2.2 Local glass transition To explore the local volumetric property of the NC model under different interfacial bonding statuses, the volume-temperature relations of two spherical shells, i.e., 10~17 Å (inner shell) and 17~24 Å (outer shell) from the mass center of FSNP (Figure 2c), were studied and shown in Figure 6. The shell width (7 Å) is smaller than the interphase thickness estimated by the radial mass density distribution in 18


Page 18 of 42

Page 19 of 42

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


section 3.1.1, and the outer shell lies in the transition region from the interphase to the matrix.

Figure 6. Specific volumes of the inner and outer shells of the INB (squares) and IB (circles) models as a function of temperature. The solid lines are the linear least-squares fitting lines in the high and low temperature ranges. For clarify, error bars are shown only at 275 K and 700 K.

Overall, the specific volume of the inner shell is larger than that of the outer shell at the same temperature due to the existence of the vdW excluded volume at the FSNP surface. For the inner shell, results show that the IB model has lower specific volume than the INB model in the higher temperature region. However, as the two models cool down, the difference in the specific volume of the two models gradually decreases and becomes nearly zero when the temperature is lower than 425 K, right after the glass transitions of the two models. For the outer shell, the specific volume of the IB model is very close to that of the INB model within the entire temperature range investigated. Considering the difference in the volume-temperature relation between the inner and outer shells, it indicates that the covalent bonding effect only 19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 42

works on a very small region (

Epoxy

Recommend Documents