Synthesis and study of properties of dental resin composites with different nanosilica particles size

Synthesis and study of properties of dental resin composites with different nanosilica particles size

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835 available at www.sciencedirect.com journal homepage: www.intl.elsevierhealth.com/journals/dema...

907KB Sizes 1 Downloads 19 Views

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

available at www.sciencedirect.com

journal homepage: www.intl.elsevierhealth.com/journals/dema

Synthesis and study of properties of dental resin composites with different nanosilica particles size Maria M. Karabela, Irini D. Sideridou ∗ Laboratory of Organic Chemical Technology, Department of Chemistry, Aristotle University of Thessaloniki, Thessaloniki GR-54124, Greece

a r t i c l e

i n f o

a b s t r a c t

Article history:

Objectives. The aim of this work was the synthesis of light-cured resin nanocomposites using

Received 30 August 2010

nanosilica particles with different particle size and the study of some physical–mechanical

Received in revised form

properties of the composites.

31 March 2011

Methods. Various types of silica nanoparticles (Aerosil) with average particle size of 40,

Accepted 18 April 2011

20, 16, 14, and 7 nm, used as filler were silanized with the silane 3-methacryloxypropyltrimethoxysilane (MPS). The total amount of silane used was kept constant at 10 wt% relative to the filler weight to ensure the complete silanization of nanoparticles. The silanizated

Keywords:

silica nanoparticles were identified by FT-IR spectroscopy and thermogravimetric analysis

Dental nanocomposites

(TGA). Then the silanized nanoparticles (55 wt%) were mixed with a photoactivated Bis-

Particle size

GMA/TEGDMA (50/50 wt/wt) matrix. Degree of conversion of composites was determined

Bis-GMA/TEGDMA

by FT-IR analysis. The static flexural strength and flexural modulus were measured using

Aerosil OX50

a three-point bending set up. The dynamic thermomechanical properties were determined

Degree of conversion

by dynamic mechanical analyzer (DMA). Sorption, solubility and volumetric change were

Flexural strength

determined after storage of composites in water or ethanol/water solution 75 vol% for 30

Sorption

days. The TGA for composites was performed in nitrogen atmosphere from 30 to 700 ◦ C.

Dynamic mechanical properties

Results. As the average silica particle size decreases, the percentage amount of MPS attached

Thermogravimetric analysis

on the silica surface increases. However, the number of MPS molecules attached on the silica surface area of 1 nm2 is independent of filler particle size. As the average filler particles size decreases a progressive increase in the degree of conversion of composites and an increase in the amount of sorbed water is observed. Significance. The prepared composites containing different amount of silica filler, with different particle size, but with the same amount of silanized silica and organic matrix showed similar flexural strength and flexural modulus, except composite with the lowest filler particle size, which showed lower flexural modulus. © 2011 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

The introduction of resin-based composite technology to restorative dentistry was one of the most significant contri-



butions to dentistry in the last century. Nowadays, esthetic tooth-shaded dental restorations are well-accepted and becoming more and more popular over metallic dental amalgams. Dental resin composites are consisted of a polymeric matrix admixed with silane reinforcing inorganic filler. Silane

Corresponding author. Tel.: +30 210 2310997825; fax: +30 210 2310997769. E-mail address: [email protected] (I.D. Sideridou). 0109-5641/$ – see front matter © 2011 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.dental.2011.04.008

826

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

provides a crucial link between the matrix and the filler that can have a significant effect on the overall-performance of composite [1]. The organic matrix is based on methacrylate chemistry, especially cross-linking dimethacrylates [2–7], while the fillers have different type (silica, ceramic, etc.), size, shape and morphology [1,6,8]. Because of the major influence of the fillers on the physical properties, the classification of dental filling composites is based on the type and the particle size of fillers [1,5,6,8,9]. In general, two types of composites are available on the market, the microfill and the hybrid composites. Microfill composites were formulated having fillers with an average particle size in the range of 0.01–0.05 ␮m and were launched in the market to overcome the problems of poor esthetic properties. However, the mechanical properties are considered low for application in regions of high occlusal forces. Hybrids, on the other hand, offer intermediate esthetic properties, but excellent mechanical properties by the incorporation of fillers with different average particle sizes (15–20 ␮m and 0.01–0.05 ␮m) [8]. In dentistry, posterior restorations (class I or II) require composites that show higher mechanical properties, while anterior restorations (class IV–V) need composites that have superior esthetics. The resin composite that meets all the requirements of both posterior and anterior restorations has not emerged yet. Therefore, the great interest in resin composite’s research is nanotechnology. Dental nanocomposites or nanofilled dental composites are claimed to combine the good mechanical strength of the hybrids [5,10,11] and the superior polish of the microfills [12]. Because of the reduced dimension of the particles (0.1–100 nm) and their wide size distribution, an increased filler load can be achieved technically resulting in reducing the polymerization shrinkage [10,14] and increasing the mechanical properties, such as tensile strength, compressive strength and resistance to fracture. On the other hand, the small size of filler particles can improve the optical properties of resin composites, because their diameter is a fraction of the wavelength of visible light (0.4–0.8 ␮m), resulting in the human’s eye inability to detect the particles [11]. Furthermore, the wear rate is diminished and the gloss retention is better [11,13,15]. As a consequence, manufacturers now recommend the use of nanocomposites for both anterior and posterior restorations. Nowadays, nanocomposites are available as nanohybrid types containing milled glass fillers and discrete nanoparticles (40–50 nm); and as nanofill types containing both, nanosized filler particles called nanomers, and agglomerations of these particles described as “nanoclusters” [11]. The nanoclusters provide a distinct reinforcing mechanism compared with the microfill or nanohybrid systems resulting in significant improvements to the strength and clinical reliability [16]. This study was designed to examine the effect of nanosilica particle size, which is silane reinforced, on some selected physico-mechanical properties of light-cured composites based on Bis-GMA/TEGDMA (50/50 wt/wt) matrix. Silica nanoparticles, used as filler, have a controlled size from 7 to 40 nm. Their surface was treated with the silane coupling agent 3-methacryloxypropyltrimethoxysilane (MPS).

2.

Materials and methods

2.1.

Materials

Propylamine, 99+% (Lot no 0.4419MS), cyclohexane, 99+% (Lot no S22455-155), and MPS, 98% (Lot no. 09426EC-265) used in the silanization of nanoparticles of silica were received from Sigma–Aldrich GmbH (Deisenhofen, Germany). The monomers used, i.e. 2,2phenyl]propane bis[4-(2-hydroxymethacryloxypropoxy) (Bis-GMA) (Lot no 07210BB) and triethyleneglycol dimethacrylate (TEGDMA) (Lot no 09004BC-275) were provided also from Sigma–Aldrich. The photoinitiator systems were camphorquinone, 97% (CQ) (Lot no S12442-053) and ethyl 4-dimethylaminobenzoate, 99+% (4EDMAB) (Lot no: 90909001) from Sigma–Aldrich. The fillers (Aerosil OX50, Aerosil 90, Aerosil 130, Aerosil 150, and Aerosil 300) used were from Degussa AG (Hanau, Germany). They are fumed amorphous silica with different average specific surface area (determined by the BET gas adsorption technique; BET comes from initial letters of authors’ names who proposed it, i.e. Brunauer, Emmett, and Teller) and average particle diameter. Their specifications are listed in Table 1. All the materials used in this study were utilized as received, without further purification.

2.2.

Silanization of silica nanoparticles

The Aerosil OX50, Aerosil 90, Aerosil 130, Aerosil 150 and Aerosil 300 silica nanoparticles were silanized with MPS, following the method of Chen and Brauer [17]. In all cases, the amount of organosilane was kept constant at 10% wt/wt relative to silica, a ratio that was more than enough to completely cover the surface of the silicas and to provide a durable interphase [18]. The amount of silane in grams (g) to obtain a minimum uniform coverage of the filler particles (X) is given by the following equation [18,19]: X=

A ω

f

In which A is the surface area of the filler (50, 90, 130, 150, and 300 m2 /g), ω is the surface coverage per gram of silane MPS (ω = 2525 m2 /g) [20] and f is the amount of silica (g). The silica (5.0 ± 0.05 g), the silane (0.50 ± 0.01 g), the solvent (100 ml cyclohexane) and n-propylamine (0.1 ± 0.01 g) were stirred at room temperature for 30 min and then at 60 ± 5 ◦ C for additional 30 min at atmospheric pressure. The mixture was then placed in a rotary evaporator at 60 ◦ C for removing the solvent and volatile by-products. The powder was then heated at 95 ± 5 ◦ C for 1 h in a rotary evaporator and finally was dried at 80 ◦ C in a vacuum oven for 20 h. The silanized silica nanoparticles were identified by FTIR spectroscopy (FTIR Spectrum One, Perkin Elmer, resolution 4 cm−1 , 32 scans, 4000–1350 cm−1 ). Spectroscopic grade potassium bromide (KBr, Sigma–Aldrich, Cas no 7758-02-3) and silica powder were pressed together into a pellet using a KBr palletizer. Thermogravimetric analysis of the silanized silica nanoparticles was performed using a Pyris 1 TGA (Perkin Elmer) thermal analyzer utilizing about 5 mg of each sample.

827

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

Table 1 – Types of fumed amorphous silica (Aerosil) fillers used in the five composites investigated and their specifications. Aerosil OX50 Specific surface area (BET) [m2 /g] Average primary particle size [nm] Tapped density [g/l] Moisture, 2 h at 105 ◦ C [wt%] Ignition loss, 2 h at 1000 ◦ C, based on material dried for 2 h at 105 ◦ C [wt%] pH, in 4% dispersion SiO2 -content based on ignited material [wt%] Materials no. Lot No

Aerosil 90

150 ± 15 14 ∼50 ≤0.5 ≤1.0

300 ± 30 7 ∼50 ≤1.5 ≤2.0

3.8–4.8 ≥99.8

3.7–4.7 ≥99.8

3.7–4.7 ≥99.8

3.7–4.7 ≥99.8

3.7–4.7 ≥99.8

23.8112-0020.99 3155-092345

23.8010-0000.99 3155-052516

23.8012-0000.99 3155-080115

23.8013-0000.99 3155-100211

23.8016-0000.99 3155-101812

Preparation of nanocomposites

Thermogravimetric analysis

The thermal degradation of the composites and the determination of the weight percentage of fillers were studied by thermogravimetric analysis. Weight changes as a function of time and temperature were evaluated with a thermal program by heating from 30 to 700 ◦ C at the heating rate of 10 ◦ C min−1 in nitrogen atmosphere (flow 20 ml/min) and cooling (10 ◦ C min−1 ) at room temperature. Thermogravimetric analysis was performed on a Pyris 1 TGA (Perkin Elmer) thermal analyzer using about 5 mg of each sample.

2.5.

Aerosil 300

130 ± 25 16 ∼50 ≤1.5 ≤1.0

The resin matrix was consisted of Bis-GMA/TEGDMA mixture (50:50 wt/wt) which contained the photoinitiator system CQ (0.2 wt%) and 4EDMAB (0.8 wt%). Bis-GMA was first heated in an ultrasonic bath at about 40 ◦ C for 10 min and then TEGDMA containing the photoinitiating system was added. Then the silanized silica (55 wt%) was mixed with the resin by hand spatulation, as suggested in Refs. [18,21–23]. Once the powder was completely wetted with the resin, the composite pastes were sheared against a glass surface with a Teflon spatula until the pastes were semi-transparent to assure maximum particle dispersion in the resin. Then the composite pastes were placed into an ultrasonic bath for 3 h. This employment of high shear strength has shown to help the silanized Aerosil to form stable sol with the dimethacrylate resin [24]. Then the pastes were placed in the vacuum oven to remove air bubbles which probably were entrapped inside the paste.

2.4.

Aerosil 150

90 ± 15 20 ∼80 ≤1.0 ≤1.0

The particles were heated to 700 ◦ C under nitrogen atmosphere flow (20 ml/min) with a heating rate of 10 ◦ C min−1 .

2.3.

Aerosil 130

50 ± 15 40 ∼130 ≤1.5 ≤1.0

Degree of conversion

The FT-IR analysis was conducted in a FT-IR spectrometer, Spectrum One of Perkin-Elmer. Spectra were obtained over 4000–600 cm−1 region and were acquired with a resolution of 4 cm−1 and a total of 32 scans per spectrum. A small amount of each composite was placed between two translucent Mylar strips, which pressed to produce a very thin film. The FT-IR spectrum was recorded at zero time and immediately after exposure to visible light (for 80 s), utilizing a XL 3000 dental

photocuring unit from 3M-ESPE Company (St Paul, MN, USA). This source consisted of a 75-W tungsten halogen lamp, which emits radiation between 420 and 500 nm with a maximum peak of 470 nm. For each spectrum, it was determined the height of aliphatic C C peak absorption at 1637 cm−1 , and the aromatic C C peak absorption at 1580 cm−1 , utilizing a base line technique which proved the best fit to the Beer–Lambert law [25]. The aromatic C C vibration is used as an internal standard. The percent monomer conversion of the cured specimen, which expresses the percent amount of double carbon bond reacted, is determined according to the following equation:



Degree of conversion (%) = 100

1−

(A1637 /A1580 )polymer



(A1637 /A1580 )monomer

2.6. Sorption of water and ethanol/water solution (75 vol%)–solubility–volumetric change Sorption and solubility tests were determined according to the method described in ANSI/ADA Specification No. 27-1993 for resin based filling materials. Specimen discs were prepared by filling a Teflon mold (15 mm in diameter and 1 mm in thickness) with the unpolymerized material. The samples were irradiated for 80 s on each side, using the XL3000 dental photocuring source. The curing unit was used without the light guide at the contact with the sample. Four specimen discs were prepared for each composite material. All the specimens were immersed in water or ethanol/water solution 75 vol%, at 37 ± 1 ◦ C, for 30 days. The percentage amount of water or ethanol/water solution sorbed (WS (%) or EWS (%)) and desorbed (WD (%) or EWD (%)), the solubility (SL (%)) in these liquids, the volumetric change (VI (%)) and the fraction of the liquid contributing to an increase in swelling (f) were determined according to the method described in details in our previous work [26].

2.7.

Mechanical properties

Mechanical properties were measured in accordance with International Standard Organization (ISO) Specification No. 4049. Bar specimens were prepared by filling a Teflon mold with unpolymerized material, taking care to minimize entrapped air. The upper and lower surface of the mold was

828

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

overlaid with glass slides covered with a Mylar sheet to avoid their adhesion with the unpolymerized material. The completed assembly was held together with spring clips and irradiated by overlapping technique, as recommended in ISO4049, using the XL 3000 dental photocuring unit. This unit was used without the light guide in contact with the glass slide. Each overlap was light-cured for 80 s. The samples were irradiated on both sides. Then, the mold was dismantled and the composite was carefully removed by flexing the Teflon mold. Five specimen bars were prepared for each composite.

2.7.1.

Flexural strength and flexural modulus

For flexural tests, bar-specimens were prepared by filling a Teflon mold (2 mm × 2 mm × 25 mm) and storing them in distilled water at 37 ± 1 ◦ C in dark for 24 h, immediately after curing. The specimens were bent in a three-point transverse testing rig with 20 mm between the two supports (3-point bending). The rig was fitted to a mechanical testing machine (Instron, model 3344, High Wycombe, England). All bend tests were carried out with a constant cross-head speed of 0.75 ± 0.25 mm/min until fracture occurred. The load and the corresponding deflection were recorded. The flexural modulus (E), in MPa, and the flexural strength (), in MPa, were calculated using the following equations: E=

F1 l3 4bdh3

and  =

3Fl 2bh2

where F1 represents the load in Newtons exerted on the specimen, F is the maximum load in Newtons exerted on the specimen at the point fracture, l is the distance in mm between the supports, h is the height of specimen in mm measured immediately prior to testing, b is the width of specimen in mm measured immediately prior to testing and d is the deflection corresponding to the load F1 .

2.7.2.

Dynamic mechanical analysis (DMA)

For DMA tests, bar specimens were prepared by filling a Teflon mold (2 mm × 2 mm × 40 mm). The bar-shaped specimens were divided into four groups of four samples each. The first group consisted of dry samples measured 1 h after preparation. During this time they were remained in a desiccator at 37 ± 1 ◦ C in a dark environment. The second, third and fourth group consisted of samples, which had been stored in distilled water at 37 ± 1 ◦ C in dark for periods of 1, 7 and 30 days correspondingly. The samples of groups II–IV were immersed in water at 37 ± 1 ◦ C immediately after curing. DMA tests were performed on a Diamond dynamic mechanical analyzer (Perkin Elmer Inc., Waltham, MA, USA) using a dual-cantilever clamp. A frequency of 2 Hz was applied (approximately average chewing rate) and at amplitude of 10 ␮m. A temperature range of 25–185 ◦ C at a heating rate of 2 ◦ C min−1 was selected to cover mouth temperature and the materials glass-transition temperature (Tg ). Elastic modulus (E ), viscous modulus (E ) and tangent delta (tan ı) were plotted against temperature over this period. After the DMA run was completed, the sample was allowed to cool naturally at room temperature and the values of E , E and tan ı at various temperatures were noted. This method was used for each of the samples and the mean values were calculated.

2.8.

Statistical analysis

The values reported in all following tables and figures represent mean values ± standard deviation of replicates. One-way analysis of variance (ANOVA) test followed by a multiple comparisons Tukey’s test used to compare between means were utilized to determinate significant differences at p = 0.05 significance level set. This was performed separately for each of the different properties or parameters.

3.

Results and discussion

For clarity, the composites described herein will be abbreviated according to the type of silica used as filler and the silane surface treatment used on the silica particles. For example, a composite filled with silica Aerosil OX50 silanized with MPS will be referred to as the “OX50-MPS composite”, and the composite filled with silica Aerosil 300 silanized with MPS will be referred to as the “300-MPS composite”.

3.1.

Characterization of silanized silica

To improve the interfacial adhesion between the silica nanoparticles and resin matrix and to uniform filler dispersion within the resin, a silane coupling agent has been used. Many methods have been used to treat the filler particles with a silane coupling agent and a general consensus does not emerge from the literature regarding the best method of silanization. In this work silanization of silica nanoparticles was performed by mixing MPS silane with cyclohexane in the presence of n-propylamine used as catalyst. The use of npropylamine accelerates the silanization of the silica [27,28] and yields the most stable interfacial silicon–silane bonds [17]. This silanization procedure was followed by Antonucci et al. [18,21,22] in their attempt to study the interphase effects in dental nanocomposites. According to this mechanism, socalled direct condensation mechanism, the silane molecules would chemically bond to the surface of filler particles via direct condensation of –OCH3 groups of the silane with the surface hydroxyl groups of the filler, to form a covalent bond (oxane bond formation, Si–O–Si) [27,28]. The –OCH3 groups on adjacent silane molecule can also be condensed with each other to form a polymeric siloxane polymer film on the silica surface (Fig. 1) [29]. The FTIR spectra of the five types of silica nanoparticles showed the following absorption bands: (a) a broad band at 3600 cm−1 because of the presence of surface silanol groups ( SiOH), (b) at 1990 cm−1 and 1870 cm−1 , due to the siloxane linkages ( Si–O–Si ) in the bulk of the silica, which are unaffected by the surface treatment and (c) at 1627 cm−1 which is attributed to the hydroxyl groups (–O–H) bending vibration of the bulk silica moisture. The FTIR spectra of silica treated with MPS showed the absorption bands: (a) at 1640 cm−1 due to the stretching vibration of the C C bond in silane, (b) the overlapped absorption bands at 1722 and 1706 cm−1 due to C O modes; the band at 1722 cm−1 is due to the free carbonyl (–C O) stretching vibration, and the peak at 1706 cm−1 is a characteristic of the carbonyl groups which form hydrogen bonds with the silica or adjacent silane hydroxyls, (c)

829

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

Fig. 1 – A simplified representation for idealized silanization of silica-filler particles according to the direct condensation mechanism (R: CH2 C(CH3 )COO(CH2 )3 –).

at 1470 cm−1 , because of methylene group (CH2 ) of silane, and (d) at 2850–3050 cm−1 , due to symmetric and asymmetric stretching vibration of the –CH3 and –CH2 – groups of MPS. These spectra indicate that the silane molecules have been successfully grafted onto the surface of silica nanoparticles forming a layer around the particles. Fig. 2 illustrates representative FTIR spectra of silica Aerosil 150 and silanized silica with the MPS. The amount of silane MPS attached on each silica surface was quantitatively determined by TGA and is presented in Table 2. The TGA of the samples showed one-step weight loss, indicating that only chemically bound silane molecules remained on the surface of the silica Aerosil OX50, Aerosil 90, Aerosil 130 and Aerosil 150, since the physically adsorbed silane is removed at lower temperatures (50–180 ◦ C) [19] and no such a weight loss was observed in TGA curves at that temperature range. The drying process applied during the preparation of silanized silica, 1 h at 95 ◦ C and 20 h at 80 ◦ C under vacuum, seems to remove the physically adsorbed silane simply by evaporation. In the case of Aerosil 300 silanized with MPS, TGA curves showed two-step weight loss. The first step, at low temperatures with maximum degradation rate at 196.8 ◦ C is due to removal of a small amount of physically adsorbed silane that remained on the surface of silica particles, despite of drying process applied during the preparation of silanized silica. The second degradation step corresponds to removal of chemically bound silane molecules remained on the surface of the silica. TGA curves showed also a small weight loss at approximately 500 ◦ C which is due to condensation of surface

silanols [19]. TGA data from Table 2 show that, as the average primary silica particle size decreases, the percentage amount of silane MPS attached on the silica surface increases. However, the number of MPS molecules attached on 1 nm2 of the silica surface is independent of filler particle size (3.04 ± 0.07 molecules of MPS/nm2 of silica), since silanization process took place under the same conditions, except silica Aerosil OX50 (2.22 molecules MPS/nm2 silica). According to the previous works, MPS silane molecules oriented parallel to the surface area of filler occupy 0.83 nm2 of filler surface per molecule [30], while MPS molecules oriented perpendicularly occupy 0.24 nm2 of filler surface per molecule [31]. These values correspond to 1.20 molecules MPS/nm2 of silica in case of parallel coverage and 4.17 molecules MPS/nm2 of silica in the case of perpendicular coverage. Comparing these values with the number of MPS molecules attached on the silica surface (per nm2 ) determined by TGA of each type of silanized silica (Table 2), it is concluded that MPS molecules must have a random (parallel and perpendicular) orientation relative to the silica surface, in all types of the silica used (Aerosil OX50, Aerosil 90, Aerosil 130, Aerosil 150 or Aerosil 300).

3.2.

Thermogravimetric analysis

The TGA curves and the corresponding differential TGA (DTGA) curves of all prepared composites containing different filler particles size filler indicated a three-step degradation process (Fig. 3). The first two steps of the thermal degrada-

Absorbance

Aerosil 150 + ΜPS Aerosil 150

4000

3600

3200

2800

2400

2000

1800

1600

1400

Wavenumber (cm-1)

Fig. 2 – FT-IR spectra of silica Aerosil 150 and MPS silane treated silica Aerosil 150 surface.

830

2.22 2.95 3.04 3.12 3.06 3.68 4.90 5.05 5.18 5.08 4.57 10.95 16.30 19.31 37.81 – – – – 0.07 – – – – 0.12 – – – – 0.93

wt%

3.68 4.90 5.05 5.18 5.20 4.57 10.95 16.30 19.31 38.73 Aerosil OX50 Aerosil 90 Aerosil 130 Aerosil 150 Aerosil 300

2.34 5.60 8.35 9.89 19.35

wt%

molecules/nm2 ␮mol/m2

Total

2.22 2.95 3.04 3.12 3.13

molecules/nm2 ␮mol/m2

Chemical absorbed

wt% molecules/nm2 ␮mol/m2

Physically absorbed

Silane TGA weight loss (%) Type of silica

Table 2 – Amount of silane (MPS) attached on the silica surface determined by TGA of each type of silanizated silica, in nitrogen atmosphere.

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

Fig. 3 – Effect of nanosilica particle size (40, 20, 16, 14, and 7 nm) used as fillers on the TGA and DTGA curves of prepared nanocomposites, in nitrogen atmosphere.

tion of the composites are related to the degradation of neat polymeric matrix, Bis-GMA/TEGDMA, while the third step is due to condensation of surface silanols of silica [19]. Comparison of the TGA/DTGA curves of the polymer matrix and the curves of the composites is shown in Fig. 4. The appearance of two thermal degradation steps in the matrix consisted of Bis-GMA/TEGDMA is attributed to inhomogeneities in the network structure mainly due to the formation of primary cycles during photopolymerization. The first degradation step, at low temperatures, corresponds to bond breaking near the cycling points of polymer network, while the second corresponds to bond breaking of the main network [32–34]. From thermogravimetric curves of the prepared composites the values of contained inorganic filler were determined, which are 53.5 wt% (OX50-MPS), 51.4 wt% (90-MPS), 50.2 wt% (130-MPS), 49.6 wt% (150-MPS) and 46.3 wt% (300-MPS). It is observed that as particle size of inorganic filler decreases, the fraction of inorganic filler of the corresponding composites decreases. As the particle size of silica decreases, the specific surface area increases and therefore, the percentage

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

to high viscosity. Therefore, the termination step becomes diffusion-limited, resulting to low degree of conversion values. Additionally, the smaller filler particle entails higher light scattering, and hence higher values of conversion [23,35]. According to Ruyter and Øysæd [36], the maximum degree of conversion occurs when the filler particle size is near half the wavelength of activating light. In this study, the visible light wavelength used was 420–500 nm for photocuring the composites and average filler particle size was 40, 20, 16, 14 and 7 nm. Therefore, the increased light scatterings during photocuring caused an increase of degree of conversion. The low degree of conversion values of prepared composites indicates that some amount of unreacted monomers (free or/and trapped) remained. The orientation of the silane molecules within the interfacial layer, as well as the highly condensed silane interphase limit mobility of the silane methacrylate and hence, the degree of conversion of composites which expresses the number of double bonds of the monomer methacrylate groups which react.

Fig. 4 – TGA/DTGA curves of Bis-GMA/TEGDMA polymer matrix, in nitrogen atmosphere.

of MPS coupling agent that required for complete and uniform coverage of silica surface increases and the filler content decreases. However, no significant differences were found to exist between the courses of thermal decomposition of composites. A very interesting point is the weight loss during the third step of decomposition, which is due to the condensation of surface silanols of silica; this weight loss follows the order: 300-MPS-composite (6.6 wt%) ≈ 150-MPS-composite (6.5 wt%) > 130-MPS-composite (6.2 wt%) ≈ 90-MPS-composite (6.1 wt%) ≈ OX50-MPS-composite (6.1 wt%).

Degree of conversion

3.3.

The correlation between degree of conversion of composites and average particle size of their filler is shown in Fig. 5. A progressive decrease in the degree of conversion of composites is observed by increasing the silica particle size. From TGA of composites, a progressively higher percentage filler in the composites is noted with increasing filler particle size. The increase in filler percentage leads to gradually restricted mobility of macroradicals as the polymerization proceeds, due 66

53

65

52

64

51

63

50 62

49 48

61

47

60

46 5

10

15

20

25

30

35

40

45

Degree of conversion (%)

Inorganic filler (%)

54

831

59

Average particle size of silica (nm)

Fig. 5 – Effect of nanosilica particle size (40, 20, 16, 14, and 7 nm) used as fillers on the degree of conversion of the corresponding composites.

3.4. Sorption of water or ethanol/water solution (75 vol%)–solubility–volumetric change The values of the determined parameters for sorption of water and ethanol/water solution by the prepared composites, after storage in liquids (37 ± 1 ◦ C) for 30 days, are reported in Table 3. The 300-MPS-composite sorbed the highest amount of water followed by the other four composites which follow the 150-MPS-composite > 130-MPS-composite > 90-MPSorder: composite ≈ OX50-MPS-composite (p ≤ 0.05). According to TGA of the prepared composites (Table 3), the 300-MPS-composite contains the highest percentage of organic phase (matrix and coupling agent), while the other materials follow the order: 150-MPS-composite > 130-MPS-composite > 90-MPScomposite > OX50-MPS-composite. This order corresponds to the order of the hydrophilicity of composites. Additionally, the crosslinking between the polymer chains, which makes the polymeric network more heterogeneous, can also, contribute to the composite’s hydrophilicity. The heterogeneity entails to the existence of more microvoids in the polymeric network, in which can accommodate molecules of water. Therefore, the highest water uptake value of 300-MPS-composite is probably due to the high degree of conversion value (Fig. 5). As also revealed in Table 3, free surface silanols of silica (-SiOH) continue to exist despite the presence of chemisorbed layer of silane at the silica surface. These silanols form hydrogen bonds with sorbed water molecules, contributing to an increase to the composite’s hydrophilicity. As a result, the 300 MPS-composite sorbed the highest amount of water, since it contains the highest percentage of free surface silanols. In summary, it is manifested that as the filler particles size of composites decreases, the amount of sorbed water increases. This behavior is associated with: (a) an increase of the amount of hydrophilic coupling agent MPS, (b) heterogeneity of the polymeric network because of crosslinking and (c) increase of amount of free surface silanols of silica, while the average filler particle size decreases. A similar behavior was observed for the water desorption. The amount of water desorbed was no statistically different (p ≥ 0.05) from that sorbed.

832

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

Table 3 – Sorption/desorption parameters of water or ethanol/water solution (75 vol%) at 37 ◦ C for the five composite materials, after storage in liquids for 30 days [mean (S.D.)]* , n = 4. Composite

Sorption (wt%)

Water 2.07 (0.05)a,A OX50-MPS 90-MPS 2.09 (0.02)a,C 130-MPS 2.13 (0.03)E 150-MPS 2.26 (0.03)G 300-MPS 2.47 (0.07)I Ethanol/water solution 75 vol% 3.96 (0.12)B OX50-MPS 3.23 (0.10)c,D 90-MPS 130-MPS 3.14 (0.19)c,F 150-MPS 3.33 (0.16)c,H 300-MPS 5.37 (0.03)J ∗

Desorption (wt%)

Solubility (wt%)

% volume increase

Swelling, f

2.08 (0.03)b,A 2.11 (0.04)b,C 2.16 (0.03)E 2.24 (0.04)G 2.46 (0.04)I

0.31 (0.03)e 0.27 (0.02)e 0.31 (0.03)e 0.39 (0.03) 0.78 (0.05)

1.45 (0.49)f 2.26 (0.45)f 1.97 (0.47)f 2.18 (0.20)f 1.98 (0.50)f

0.44 (0.14)h 0.68 (0.14)h 0.59 (0.15)h 0.65 (0.05)h 0.54 (0.14)h

3.76 (0.10)B 3.07 (0.08)d,D 3.00 (0.15)d,F 3.27 (0.16)d,H 5.33 (0.05)J

0.27(0.04) 0.08 (0.05) 0.18 (0.05) 0.41 (0.07) 1.3 (0.07)

4.89 (0.14)g 4.96 (0.24)g 5.07 (0.18)g 5.51 (0.21) 7.04 (0.16)

0.66 (0.03) 0.82 (0.03)i 0.86 (0.03)i 0.89 (0.03)i 0.76 (0.03)

Common corresponding lowercase letters in a given column indicate no significant difference (p ≥ 0.05). Common corresponding uppercase letters in a given row indicate no significant difference (p ≥ 0.05).

The results obtained for the sorption and desorption of the ethanol/water solution by the prepared composites are also presented in Table 3. The 300-MPS-composite with the smallest filler particles size, sorbed/desorbed the highest amount of solution, as in the case of water, followed by the other composites: OX50-MPS-composite > 90-MPS-composite ≈ 130MPS-composite ≈ 150-MPS-composite (p ≤ 0.05). The amount of water desorbed was no statistically different at p ≥ 0.05 from that sorbed. All composites showed higher values sorbed liquid after immersion in ethanol solution when compared with those in water. This behavior is due to different chemical affinity between sorbed liquid and the polymer matrix [37]. The amount of unreacted monomer that was extracted by water or the ethanol solution (37 ± 1 ◦ C) during the 30 days of immersion, known as the solubility (SL) of the composite, is presented in Table 3. As it can be seen, the 300-MPS-composite shows the highest solubility value, during the immersion in water and in ethanol solution. According to TGA of the composites, in the interphase of 300-MPS-composite above the first silane layer, which is covalently bonded to silica surface, a second layer which is hydrogen bonded to the first layer is formed. This second layer may weaken the composite, since the silane molecules may not link the filler to the matrix, in consequence the highest solubility of 300-MPS-composite may result. The other four composites, in which the interphase consists of a silane layer (only chemically bound silane molecules), showed statistically (p ≥ 0.05) the same values of solubility. The chemically bound MPS molecules of the first layer can react with the dimethacrylate monomers during polymerization via methacrylate groups creating a strong coupling between the filler and polymer matrix phase, which results in decrease of the solubility of polymeric matrix. In the case of water, no significant statistical difference (p ≥ 0.05) was found between the volume increase (VI) values of composites, while in the case of ethanol/water solution, they follow the order: 300-MPS-composite > 150-MPS-composite > 130-MPScomposite ≈ 90-MPS-composite ≈ OX50-MPS-composite (p ≤ 0.05) (Table 3). Finally, the fraction (f) of sorbed solution which is accommodated between polymer chains and contributed to swelling of composite is shown in Table 3. The fraction f of water is independent (p ≥ 0.05)

on the amount of filler and the filler particle size. However, the fraction of ethanol solution follows the OX50-MPS-composite < 300-MPS-composite < 150order: MPS-composite ≈ 130-MPS-composite ≈ 90-MPS-composite (p ≤ 0.05). Comparing the volume increase and swelling of composites in both solvents it is noted that all composites show extremely high values in ethanol/water solution. This behavior is due to the highest amount of ethanol solution than water sorbed by the composites.

3.5.

Mechanical properties

The influence of inorganic filler particles size of composites on flexural strength (3-point bending) and flexural modulus is shown in Table 4. The results shown in the table depend on the competitive actions which are caused from: (a) different particles size of inorganic filler, (b) different amount of inorganic filler and (c) aging of composites in water of composites for 24 h. It is well known that, initial fracture of resin composites occurs at the filler/matrix interphase, because the rigidities of the silica filler and matrix resin are different [38]. The interphase contributes to stress distribution that developed under loading. As the filler particles size decreases, their surface area increases relative to their volume and therefore higher surface energy at the interphase was observed. Thus, the stress concentration at the filler/matrix interphase decreases with

Table 4 – Flexural strength and flexural modulus of the composites, after storage in water (37 ± 1 ◦ C) for 24 h [means (S.D.)]* , n = 5. Composite OX50-MPS 90-MPS 130-MPS 150-MPS 300-MPS ∗

Flexural strength (MPa)

Flexural modulus (GPa)

90 (8)a 95 (8)a 100 (7)a 103 (6)a 92 (5)a

5.37 (0.07)b 5.24 (0.08)b 5.49 (0.08)b 5.43 (0.09)b 4.45 (0.10)

Common corresponding lowercase letters in a given column indicate no significant difference (p ≥ 0.05).

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

decreasing particles size of the filler, having as a result higher values of flexural strength of the corresponding composites. However, the composites of the present study have different amount of filler and particularly it decreases with decreasing silica particle size, resulting in weakening of the composites strength. Moreover, the aging of composites in water causes either a decrease in their strength (plasticizing ability of water molecules, possible hydrolysis of filler-matrix bonds), or an increase in the strength (extraction of the unreacted monomers, existence of post-curing reaction). The extent of above competitive actions and which of these actions will prevail, determines the final behavior of composites in flexion. According to Table 4, it was not observed change (p ≥ 0.05) in the flexural strength of composites as the particles size of inorganic filler decreased. On the other hand, the flexural modulus values of composites with different filler sizes indicated no significant difference (p ≥ 0.05), except the 300MPS-composite, which showed the lowest value of modulus. This composite showed the highest degree of conversion. Usually a higher degree of conversion of the crosslinking monomers results in a higher density of the polymer network. The higher the crosslinking density the higher is the flexural modulus of the material. However, the 300-MPS composite showed the significantly lowest value of the flexural modulus. This can be explained by the low filler load of this composite. Because the filler content also strongly influences the modulus of elasticity of a composite. Also the flexural modulus depends on the stress transmission between the filler and the matrix in composite [38]. The stress-transfer ability between the filler and matrix depends on the interface, which is typically derived from silane coupling agent. In 300-MPS-composite the interphase adhesion between the filler and the matrix is weak, because the silane molecules form a two-layer interphase (chemically and physically absorbed silane). In other composites, the silane molecules are covalently bonded to matrix (copolymerization via methacrylate groups) creating a strong bond. In these composites, the number of MPS molecules attached on 1 nm2 of the filler surface area is independent of filler particle size (Section 3.1). Thus, there is no significant difference in interphase bond between the matrix and the filler of composites and as a result in flexural strength values. Fig. 6 presented the effect of filler particles size on dynamic mechanical properties of composites, immediately after curing. A parameter for characterizing the crosslinking extent and the heterogeneity of a polymeric network is the tan ı peak width [39–41]. During polymerization of multifunctional monomers polymeric networks with great heterogeneity are formed, which contain areas with high number of crosslinks and less crosslinked regions. In this study, the peak width of tan ı peak of materials follows the order: 300-MPS-composite > 150-MPS-composite > 130-MPScomposite > 90-MPS-composite > OX50-MPS-composite. The 300-MPS-composite polymer network, in which the interfacial adhesion between filler and matrix is weak (the interphase consists of two layers: the layer with chemically sorbed silane and the layer with physically sorbed silane), showed the greater heterogeneity. The glass transition temperature, Tg , values of the prepared composites which were determined by peak tan ı curves, are

833

Fig. 6 – Effect of nanosilica particle size (40, 20, 16, 14, and 7 nm) used as fillers on the dynamic mechanical properties (a) E , (b) E , (c) tan ı of the corresponding composites, immediately after curing.

834

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

Table 5 – Values glass transition temperature, Tg , (◦ C) of the composites, immediately after curing [mean (S.D.)]* , n = 4. Composite

OX50-MPS 90-MPS 130-MPS 150-MPS 300-MPS ∗

Average particle size of filler (nm)

Tg (◦ C)

tan ı (×103 ) at Tg

40 20 16 14 7

136.3 (0.7) 140.0 (0.4)a 143.6 (0.3) 139.9 (0.5)a 121.1 (0.3)

130 (0) 106 (2) 100 (1) 90 (1) 80 (1)

Common corresponding lowercase letters in a given column indicate no significant difference (p ≥ 0.05).

presented in Table 5. However, it was reported [42] that the feature that is most sensitive to the structure of the interphase is the height of the tan ı at the Tg and is recommended as the best parameter for characterizing variations in interfacial structure of composites. It was observed that as the particles size of fillers decreased, the values of tan ı at the Tg , of the corresponding composites decreased. The low value tan ı at the Tg indicates better interfacial adhesion between filler and matrix. Therefore, the small particle size of silica, modified with sufficient amount of coupling agent, ensures better interfacial adhesion between filler and matrix and contributes to uniform dispersion of particles in polymeric network. When the composites are soaked in water, the viscoelastic properties are affected by the water sorption, extraction of unreacted components from water, post curing and possible hydrolysis of filler-matrix bond. Storage modulus of composites, immediately after curing, decreased as the average particles size of the inorganic filler decreased. This behavior is the result of the competitive actions that come from different particles size of inorganic filler and different amount of inorganic filler. Aging of composites in water caused a significant increase (at p ≤ 0.05) in storage modulus. Comparing the values of E of composites after 30 days immersion in water, it is not observed significant difference (p ≥ 0.05); except of 300-MPS-composite in which the interfacial adhesion between filler and matrix is weak. All composites during immersion in water showed a significant decrease (at p ≤ 0.05) of Tg . Water acts as a plasticizer of the resin matrix and decreases the Tg , analogously to the absorbed water amount. Also, it was noted the differences on Tg among composites with different filler particles size, immediately after curing and after immersion in water for 1, 7 and 30 days.

4.

Conclusions

After modifying all types of silica particles of different sizes with MPS silane coupling agent, the MPS molecules are successfully grafted onto the surface of silica nanoparticles, forming a layer around them, in which silane molecules must have a random (parallel and perpendicularly) orientation relative to the silica surface. As the average silica particle size decreases, the percentage amount of MPS attached on the silica surface increases. However, the number of MPS molecules

attached on the silica surface area 1 nm2 is independent of filler particle size (3.04 ± 0.07 molecules MPS/nm2 silica). By introducing silica particles into the dimethacrylate monomers, the pastes of composites are prepared. As the particle size of inorganic filler decreases, the fraction of inorganic filler of the corresponding composites decreases. A progressive decrease in degree of conversion of composites is observed by increasing the silica particles size. This occurs as a result of an increase of filler percentage in the composites and light scattering during curing. Also, as the filler particles size of the composites decreases, the amount of sorbed water increases. This behavior is associated with: (a) an increase of the amount of hydrophilic MPS coupling agent, (b) the heterogeneity of the polymeric network because of crosslinking and (c) an increase of the amount of free surface silanols of silica, while the average filler particle size decreases. The influence of inorganic filler particles size of composites on flexural strength (3-point bending) and flexural modulus contributes to the competitive actions that come from: (a) different particles size of inorganic filler, (b) different amount of inorganic filler and (c) aging in water of composites for 24 h. Finally, the small particle size of silica, modified with sufficient amount of coupling agent, ensures better interfacial adhesion between filler and matrix and contributes to uniform dispersion of particles in polymeric network.

references

[1] Ferracane JL. Current trends in dental composites. Crit Rev Oral Biol Med 1995;6(4):302–18. [2] Peutzfeldt A. Resin composites in dentistry: the monomer systems. Eur J Oral Sci 1997;105:97–116. [3] Ruyter IE, Øysæd H. Composites for use in posterior teeth: composition and conversion. J Biomed Mater Res 1987;21:11–23. [4] Moszner N. New monomers for dental application. Macromol Symp 2004;217:63–75. [5] Moszner N, Klapdohr S. Nanotechnology for dental composites. Int J Nanotechnol 2004;1:130–56. [6] Moszner N, Salz U. In: Shalaby SW, Salz U, editors. Polymers for dental and orthopedic applications. Boca Raton/London/New York: CRC Press Taylor and Francis Group; 2007. [7] Sideridou ID, Tserki V, Papanastasiou G. Effect of chemical structure on degree of conversion in light-cured dimethacrylate-based dental resins. Biomaterials 2002;23:1819–29. [8] van Noort R. Introduction to dental materials. 2nd ed. Mosby; 2002. p. 101. [9] Klapdohr S, Moszner N. New inorganic components for dental filling composites. Monatsh Chem 2005;136:21–45. [10] Moszner N, Salz U. New developments of polymeric dental composites. Prog Polym Sci 2001;26:535–76. [11] Mitra SB, WU D, Holmes BN. An application of nanotechnology in advanced dental materials. J Am Dent Assoc 2003;134:1382–90. [12] Turssi CP, Saad JR, Duarte SL, Rodrigues AL. Composite surfaces after finishing and polishing techniques. Am J Dent 2000;13:136–8. [13] Terry DA. Direct applications of a nanocomposites resin system. Part 1. The evolution of contemporary composite materials. Pract Periodontics Aesthet Dent 2004;16(6):417–22.

d e n t a l m a t e r i a l s 2 7 ( 2 0 1 1 ) 825–835

[14] Chen MH, Chen CR, Hsu SH, Sun SP, Su WF. Low shrinkage light curable nanocomposite for dental restorative material. Dent Mater 2006;22:138–45. [15] Yap AU, Tan CH, Chung SM. Wear behaviour of new composite restoratives. Oper Dent 2004;29:269–74. [16] Curtis AR, Palin WM, Fleming GJP, Shortall ACC, Marquis PM. The mechanical properties of nanofilled resin-based composites: the impact of dry and wet cyclic pre-loading on bi-axial flexure strength. Dent Mater 2009;25:188–97. [17] Chen TM, Brauer GM. Solvent effects on bonding organo-silane to silica surfaces. J Dent Res 1982;61:1439–43. [18] Wilson KS, Zhang K, Antonucci JM. Systematic variation of interfacial phase reactivity in dental nanocomposites. Biomaterials 2005;26:5095–103. [19] Liu Q, Ding J, Chambers DE, Debnath S, Wunder SL, Baran GR. Filler-coupling agent-matrix interactions in silica/polymethylmethacrylate composites. J Biomed Mater Res 2001;57:384–93. [20] Söderholm KJM, Shang SW. Molecular orientation of silane at the surface of colloidal silica. J Dent Res 1993;72:1050–4. [21] Wilson KS, Antonucci JM. Interphase structure–property relationships in thermoset dimethacrylate nanocomposites. Dent Mater 2006;22:995–1001. [22] Wilson KS, Allen AJ, Washburn NR, Antonucci JM. Interphase effects in dental nanocomposites investigated by small-angle neutron scattering. J Biomed Mater Res 2007;81A:113–23. [23] Halvorson RH, Erickson RL, Davidson CL. The effect of filler and silane content on conversion of resin-based composite. Dent Mater 2003;19:327–33. [24] Pflug K, Weber Ch. Dental materials having a nanoscale filler. U.S. Patent 6,399,037; 2002. [25] Rueggerberg FA, Hashinger DT, Fairhurst CW. Calibration of FTIR conversion analysis of contemporary dental resin composites. Dent Mater 1990;6:241–9. [26] Sideridou ID, Karabela MM, Vouvoudi ECh. Volumetric dimensional changes of dental light-cured dimethacrylate resins after sorption of water or ethanol. Dent Mater 2008;24:1131–6. [27] Pluedemann EP. Silane coupling agents. New York: Plenum Press; 1982. [28] Kaas RL, Kardos JL. The interaction of alkoxy silane coupling agents with silica surfaces. Polym Eng Sci 1971;11:11–8. [29] Antonucci JM, Dickens SH, Fowler BO, Xu HHK, McDonough WG. Chemistry of silanes: interfaces in dental polymers and composites. J Res Natl Inst Stand Technol 2005;110:541–58. [30] Sideridou ID, Karabela MM. Effect of the amount of 3-methacryloxypropyltrimethoxy-silane coupling agent on

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

835

physical properties of dental resin nanocomposites. Dent Mater 2009;25:1315–24. Miller JD, Ishida H. Quantitative monomolecular coverage of inorganic particulates by methacryl-functional silanes. Surf Sci 1984;148:601–22. Elliott JE, Lovell LG, Bowman CN. Primary cyclization in the polymerization of Bis-GMA and TEGDMA: a modeling approach to understanding the cure of dental resins. Dent Mater 2001;17:221–9. Lovell LG, Berchtold KA, Elliott JE, Lu H, Bowman CN. Understanding the kinetics and network formation of dimethacrylate dental resins. Polym Adv Technol 2001;12:335–45. Achilias DS, Karabela MM, Sideridou ID. Thermal degradation of light-cured dimethacrylate resins. Part I. Isoconversional kinetic analysis. Thermochim Acta 2008;472:74–83. Kim S, Jang J. The measurement of degree of conversion for Bis-GMA/silica composites by FTIR spectroscopy. Polym Test 1996;15:559–71. Ruyter IE, Øysæd H. Conversion in different depths of ultraviolet and visible light activated composite materials. Acta Odontol Scand 1982;40:179–92. Kao EC. Influence of food-simulating solvents on resin composites and glass-ionomer restorative cement. Dent Mater 1989;5:201–8. Tanimoto Y, Nishiyama N, Nemoto K, Ben G. Effect of filler content on bending properties of dental composites: numerical simulation with the use of the finite-element method. J Biomed Mater Res 2004;71B: 188–95. Lu H, Stansbury JW, Nie J, Berchtold KA, Bowman ChN. Development of highly reactive mono-(meth)acrylates as reactive diluents for dimethacrylate-based dental resin systems. Biomaterials 2005;26:1329–36. Kannurpatti AR, Anseth JW, Bowman ChN. A study of the evolution of mechanical properties and structural heterogeneity of polymer networks formed by photopolymerizations of multifunctional (metha)acrylates. Polymer 1998;39:2507–13. Lu H, Lovell LG, Bowmann CN. Exploiting the heterogeneity of cross-linked photopolymers to create high-Tg polymers from polymerizations performed at ambient conditions. Macromolecules 2001;34:8021–5. Wang J-Y, Ploehn H. Dynamic mechanical analysis of the effect of water on glass bead-epoxy composites. J Appl Polym Sci 1996;59:345–57.