The mechanical properties and microstructure of nano- and microparticle-filled epoxy adhesives for bonded-in timber connections

Z. Ahmad Postgraduate

University of Bath Bath, BA2 7AY, UK

M. P. Ansell, University of Bath

Senior Lecturer University of Bath Bath, BA2 7AY, UK

D. Smedley

Rotafix Ltd, Rotafix House Abercraf, Swansea, SA9 1UR, UK

Summary The mechanical properties and microstructures of room temperature cured epoxy adhesives (diglycidylether of bisphenol “A”, DGEBA) supplied by Rotafix Ltd filled with different fillers have been studied. Nanosilica, liquid rubber (carboxyl-terminated butadiene and acrylonitrile (CTBN)) and microfillers (quartz, mica and bentonite) have been used as fillers in the DGEBA systems. The characterization of the properties of the adhesives was conducted using bending, tensile, compressive and block shear tests. A study of the effect of curing time on flexural strength was carried out to determine the curing time of the adhesives and it is recommended that the adhesives should be cured for at least 20 days at room temperature to develop full strength. Nanosilica and liquid rubber were observed to increase the flexural strength, tensile strength and modulus of elasticity of the adhesive systems. Tensile tests showed that the addition of quartz, mica and bentonite particles has a negative effect on the elongation at break but a positive effect on the Modulus of Elasticity. Liquid rubber filled adhesive possess the best mechanical properties. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) provide evidence of well-dispersed, separate particles in the nanosilica-filled epoxy and a clear phase separation in the rubber-filled adhesive. These well-defined, finely-divided microstructures result in higher strength than the micro-particle filled adhesive. 1. Introduction Timber members may be joined with bonded-in connectors such as rods or plates using epoxy-based adhesives. It is reported [1,2] to be an effective and attractive solution for timber to timber connections in structures. Rotafix Ltd has developed its solvent-free technology over the last three decades and offers epoxy adhesive repair material and expertise in repair of in-situ timber structural members as well as upgrading of timber components. Harvey and Ansell [2] showed that Rotafix’s CB10TSS epoxy-based adhesive develops excellent strength for bonded-in rod connections compared to some other commercial adhesives. CB10TSS is a 2-part thixotropic epoxy gap-filling adhesive which is formulated for bonding GRFP and CFRP fibre rods and plates to timber components. There is strong interest in improving the thermal and mechanical properties of this adhesive in order to improve toughness and raise the glass transition temperature so that applications in the construction market can be expanded. The properties of filled adhesives can be improved in several ways by (a) increasing the matrix toughness, (b) optimizing the interface or interphase between the filler and the matrix through the

use of coupling agents, compatibilizer and sizes, (3) optimizing the filler-related properties such as filler content, particle size and dispersion. This investigation focused on the third approach by adding fillers including nanosilica, liquid rubber and micro-sized particles such as quartz, bentonite and mica. The matrix material plays an important role in filled adhesives systems. If the adhesive matrix has high ductility, it can increase the toughness of the composite. Epoxy itself has poor impact properties. The incorporation of liquid rubber into epoxy resin has been shown to produce stable, compatible resins which develop two phases during cure [3,4,5]. In 1970, McGarry et al [6] showed that the addition of liquid rubbers made a significant improvement in the fracture toughness of the modified epoxy resins. In terms of bonding strength, Ratna and Banthia [7] reported a two-fold increase in lap shear strength using carboxyl-terminated poly (2-ethylhexyl acrylate) (CTPEHA) as the liquid rubber. The incorporation of nanofillers into thermosets has also attracted considerable interest indicated by the recent increase in the number of publications [8,9,10]. Incorporation of superfine inorganic particles play a remarkable role in modifying epoxies due to their smaller dimension, larger specific area, higher surface energy and better surface reactivity [8,11,12]. The addition of silica nanoparticles, up to 10 wt%, brings about a considerable increase in fracture toughness. Most of the work mentioned above concerned the properties of low viscosity epoxy adhesives. For glued-in timber connections, high viscosity epoxy-based resins are required so that a thicker glue line can be produced. Harvey and Ansell [2] studied glued-in glass fibre reinforced plastic (GFRP) rods used in beam to column connections with glue-line thicknesses ranging from 2 to 4.5 mm using epoxy-based adhesives including CB10TSS and found that joints formed with CB10TSS showed a higher ductility with higher glue-line thickness. Therefore in this study, a CB10TSS formulation was modified to incorporate the addition of nanosilica, liquid rubber and micro-particles (quartz, mica and bentonite) with the aim to improve the mechanical properties. A thorough study of the mechanical behaviour of the adhesives in flexure, tension and shear is presented here. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) was also employed to understand the influence of the fillers on the formation of coherent filler-matrix interfaces which affect the mechanical properties of the adhesives. 2. Experimental methods 2.1 Materials Four types of adhesives used in this study were obtained from Rotafix Ltd. The first type was a high-viscosity, two-phase toughened epoxy resin system consisting of Part–A (epoxy resin mixture of diglycidylether of bisphenol-A (DGEBA) with reactive diluents glycidlyether (monofunctional) and silica fume particles) and Part-B (hardener, mixture of polyetheramines) and this mixture (CB10TSS) is considered neat adhesive. The other three adhesives were formulated from neat adhesive but with the addition of (a) nanosilica particles, (b) liquid rubber of carboxyl-terminated butadiene and acrylonitrile (CTBN) and (c) microparticles (bentonite, quartz and mica) respectively. These adhesives are designated as Epoxy1 (neat adhesive), Epoxy2 (neat adhesive/nanosilica), Epoxy3 (neat adhesive/rubber) and Epoxy4 (neat adhesive/clay). The effect of the three different categories of fillers (nano, micro and liquid form) on mechanical properties is observed. 2.2 Specimens preparation and measurements 2.2.1 Adhesives The adhesives were manually mixed at a constant speed so as not to form a vortex as this might have entrapped air. The mixing was continued for about 15 minutes until the two components were well blended. After mixing the two components the adhesive was placed in a vacuum chamber for

about five minutes in an attempt to release air bubbles. After that the mixture was transferred into three types of mould to make flexural, tensile and shear specimens. 2.2.2 Flexural specimens The adhesive mixture was poured into a plastic rectangular mould, 2 mm thick, 500 mm wide and 500 mm long, coated with a mould release agent. Another glass plate was placed on top of the mould and weights were added to apply some pressure in order to achieve a flat surface. The adhesive was left in the mould for 6 days to cure. After remoulding, the adhesive plate was cut by diamond saw for the bending test specimens. The specimens had a span to depth ratio of 16:1. After cutting the specimens from the plates, 50 specimens were randomly selected and divided into 5 groups with 10 specimens per group for bending tests at different times following curing, namely 7 days, 10 days, 20 days, 30 days and 40 days to study the effect of curing time on flexural strength. After curing flexural tests under three-point load method were performed according to BS EN ISO 178:2003. The tests were conducted using an Instron 1122 Universal Test Machine at a crosshead speed of 1 mm/min. The load and deflection of the test specimen were recorded. 2.2.3 Tensile specimens

The remainder of the mixture was poured into dumbbell shaped mould for tensile specimens and again a glass plate was placed on top to provide a flat finish. After 20 days curing, two pairs of strain gauges, (one pair for axial deformation and the other pair for lateral deformation) were mounted on the sample at 0° and 90° orientations to determine the strain values. The tensile test was perfomed using an Instron 1185 Universal Testing Instrument equipped with a 100 kilo-Newton (kN) load cell by securing the tensile samples in the jaws of the instrument. The load was applied at a constant rate of 5 mm/min until the maximum load was reached and the load and strain gauge signal measurements were recorded.

2.2.3 Block shear specimens In this study the block was prepared from 50 mm thick Kerto laminated veneer lumber (LVL) panel manufactured from Norwegian spruce veneer. The advantage of using LVL is that it can be assumed to contain a uniform distribution of defects. Ten beam specimens; 50 mm thick, 70mm wide and 450 mm long were cut from the LVL panel. Two pieces for each adhesive type were randomly selected. The 50 mm thick LVL beam was sliced into two parts in the plane of the veneer (avoiding glue lines between veneers) to make two beams, one 22 mm thick and the other 25 mm thick as 3 mm of LVL is lost through sawing the LVL. The LVL adherends were laid side by side. Two Perspex shims of 3 mm thickness were placed at the end of the beam as thickness guides and were secured to the beam using brown tape. The 3mm shims were placed to set the required adhesive bond thickness. An aluminum plate was attached with brown tape to the bottom of the mould to prevent the adhesive from leaking out. Then adhesive was poured and spread slowly onto one of the adherends. Then the other adherend was place on top. The wood composite was clamped together by binding it with brown tape and also by using steel clamps to secure the bonding. The composite was left to cure at room temperature for 10 days. After curing for 10 days, the LVL adherends were trimmed at the edges to remove the aluminum plate and the shims. The beam specimens were cut into 50mm cubes as shown in Fig. 1. The LVL and adhesive thickness were chosen so that one adhesive to LVL interface was subjected to shear (see Fig.2). Five replicates were randomly chosen per adhesive formulation. The test specimens were stored in a humidity room kept at a temperature of 20

o C and 65% relative humidity until the time of testing. All

specimens were tested according to ASTM D 905. The shearing load was applied using an Instron 1185 test machine with a 100 kN load cell, with a continuous motion of the moveable head at 0.5 mm/min until the maximum load was reached.

2.2.4 Microscopical examination

The fracture surface of adhesives and the microparticles (quartz, bentonite, mica) were inspected in a JEOL JSN6310 scanning electron microscope (SEM) equipped with a computer image analysis system, after gold coating. Transmission electron microscopy (TEM) measurements were carried out with a JEOL 1200EX transmission electron microscope applying an accelerating voltage of 120keV. The specimens were prepared using a Leica Ultracut E (Reichert & Jung) ultramicrotome. Sections of about 100 nm thickness were sliced with a Diatome diamond knife at room temperature.

3.0 Results and discussions 3.1 Effect of curing time on bending properties Fig.3 and Fig.4 show the effect of curing time on bending strength and Modulus of Elasticity (MOE) after 7, 10, 20, 30 and 40 days. Fig. 3 shows that, except for Epoxy4, in the period from 7 to 20 days, the strength of the adhesives continue to develop, which indicates that crosslinking is still in progress. The curing of Epoxy4 is completed after 7 days or maybe earlier as testing was only started after 7 days. Fig. 4 shows that the adhesive also becomes stiffer as the process continues. Fig. 3 suggests that Epoxy1, Epoxy2 and Epoxy3 required at least 20 days before reaching a fully cured state. This has

Fig 3 The effect of curing time on bending strength.

Fig 4 The effect of curing time on Modulus of Elasticity.

3mm

Figure Error! No text of specified style in

22mm

25mm

50mm

Fig 2 Position of the block in the shear fixture

showing the edge of loading head resting on the

adhesive/LVL adherend interface.

Fig 1 Block shear test specimen.

42.63

36.45

25.45

28.13

-5

0

5

10

15

20

25

30

35

40

45

0.000 0.005 0.010 0.015 0.020 0.025

Longitudinal strain

Tensile stress (MPa)

AlbipoxNanopox

TimbersetCB10TSS

Fig 5 Typical examples of stress-strain diagram for epoxy-based adhesives stressed in tension

implications for the testing of adhesive joints cured at room temperature for less than 20 days. Great care should be taken in industrial applications where the adhesive is cured at room temperature. Based on these results, further mechanical testing was only conducted after 20 days. 3.2 The effect of fillers on mechanical properties of epoxy-based adhesive Table 1 summarizes the mechanical properties for the entire adhesive used in this study.

Tensile Properties

Bending Properties

Shear Properties

Sample code Strength (MPa)

MOE (GPa)

Strength (MPa)

MOE (GPa)

Strength (MPa)

Strain at break (%)

Mean 26.81 2.40 62.65 2.42 6.19 Std. Dev 3.79 0.47 5.43 0.13 1.20 Epoxy1 COV (%) 14.14 19.58 8.67 5.37 19.39

1.93

Mean 34.30 3.58 71.71 3.55 6.39 Std. Dev 1.45 0.39 5.97 0.18 1.16

Epoxy2 COV (%)

4.23 10.89 8.33 5.07 18.15 1.80

Mean 42.16 3.80 80.89 3.61 7.45 Std. Dev 2.27 0.26 4.68 0.18 1.50

Epoxy3 COV (%)

5.38 6.84 5.79 4.99 20.13 1.59

Mean 24.08 11.37 47.09 9.23 Std. Dev 1.27 1.39 2.73 0.95

Epoxy4 COV (%)

5.27 12.23 5.80 10.29 - 0.27

Fig.5 shows typical stress-strain behaviour from tensile tests. All specimens failed immediately after the tensile stress reached the maximum value, however, the stress-strain curves showed considerable nonlinearity before reaching the maximum stress, but no obvious yield point was found in the curve. The plots clearly show enhanced tensile properties for filled epoxies both in terms of stress at break and modulus of elasticity. As seen from Table 1, Epoxy3 has the highest bending, tensile and shear strength followed by Epoxy2, Epoxy1 and Epoxy4. The inclusion of rubber has improved the Epoxy3 bending and

tensile strength by 29% and 57% compared to standard adhesive, Epoxy1. There is also lower variability observed in the bending and tensile strength of Epoxy3, which may indicate the homogeneity of the composite. However, the bending strength is almost double the tensile strength. This is due to the density of surface flaws and the difference in the deformation mode where only the outer surfaces of the beam experiences peak stresses and strains. The mean strength of Epoxy3 is at the upper bound of values for commercial epoxy systems [13]. An increase in tensile strength suggests a more crack-resistant microstructure.

The first important requirement of rubber in the matrix is to form a separate phase when the resin is cured. If the rubber is too compatible with the resin no phase separation or limited phase separation will occur. This may lead to a small increase in the toughness but to a major decrease in mechanical strength [14]. Therefore further investigation was made on the molecular structure of the matrix using transmission electron micrograph (TEM) as shown in Fig.6 and this figure confirmed that phase separation had occurred resulting in improved tensile strength.

An improvement of about 15% and 28% in bending and tensile strength respectively were observed with the addition of nanosilica in Epoxy2 adhesive. Dispersion and aggregation phenomenon of silica within the matrix for Epoxy2 is further investigated through secondary electron imaging (SEI) and the X-ray mapping as shown in Fig. 7. In Fig. 7a the silica fume is seen as pale particles about 10 µm across but there is an even dispersion of nanosilica between the silica fume particles as seen in Fig. 7c. The white spots over the dark background indicate the location of silicon within the composites (see Fig. 7b). There was no obvious aggregation of the silica observed, which means that the silica particles are uniformly distributed. The silicon presence is further confirmed by the EDS analysis on Epoxy2. Fig 7 Micro-analysis of Epoxy2; (a) Analysis of unfractured, sectioned and polished surface using SEI (b) X-ray silicon mapping (c) TEM micrograph (bar scale dimension = 100nm). However, Epoxy4 composites containing micro size particles showed much lower bending and tensile strengths than the neat adhesive, Epoxy1, by a factor of 25% and 10% respectively. In spite of the reduction in bending and tensile strengths for Epoxy4 caused by the filler particles, the MOE for Epoxy4 is seen to be more than three times greater than for Epoxy1. These observations agree well with the results obtained for bending and tensile properties. From SEM micrographs and EDS analysis (see Fig. 8), the particles were seen to contain silicon, iron, sodium and potassium and these high modulus particles have caused the increase in the MOE.

Fig 6 Transmission electron micrograph of

rubber modified Epoxy1

Fig 8 Micro-analysis investigation of Epoxy4 adhesive: (a) Back scattered signal analysis on Epoxy4 unfractured surface, (b) EDS analysis showing the presence of potassium, silicon and aluminum elements in mica microparticles.

From Table 1, Albipox has the highest joint strength, followed by Nanopox. The value of shear strength for Albipox was 7.45 MPa which is 1.2 times the value for the neat epoxy CB10TSS (6.19MPa). This may be attributed to the improvement in the interfacial interaction of the rubber nanoparticle with the epoxy-based adhesive which provides better wetting. Failure surfaces for the block shear tests were similar with failure predominantly in the wood section and only a few blocks failed at the adherend-adhesive interface. Examples of such surfaces are shown in Fig. 9.

From an engineering point of view, elongation at break is an important parameter for describing the rupture behaviour of composite materials. Fillers with higher stiffness than the matrix can increase the modulus of the composites but generally fillers cause a dramatic decrease in the elongation at break unless there is good adhesion between the filler and the matrix [15].

Table 1 clearly indicates this is also the case when nano-silica, nano-rubber, quartz, mica and bentonite are used. This demonstrates that the addition of nanofillers is not able to increase elongation at break. The SEM micrograph for fractured surface of Epoxy 4 (see Fig. 10a), illustrates that at high magnification the particles appear to have debonded from the matrix, which suggests the presence of a weak interface between the particles and matrix.

The inhomogeneous interface and poor adhesion affect the ability of the adhesive system to absorb much fracture energy during static testing. Besides that, the quartz and mica particles are more brittle than the resin and dominate the tensile behaviour resulting in a small strain to failure. Although Epoxy3 shows a lower elongation at break, the bending and tensile strength is significantly improved. This improvement may be due to energy being absorbed by shear yielding which can be seen as white bands in the SEM image in Fig. 10b. The white bands are the result of

Fig 9 Failure mode of block shear sample showing; (a) crack propagation at the adhesive-wood interface and (b) the fracture surface of the adherend and adhesive.

Fig 10 Scanning electron micrographs of (a) fractured surface of Epoxy4 and (b) fractured surface of Epoxy3.

the contrast factor in the secondary electron image (SEI) indicating uneven ridges formed by plastic deformation. The stress-whitening zone can also be observed by the naked eye from the fractured bending test specimens. Therefore the increase in strength depends on the strength of filler/adhesive bonding while the elongation at break depends on the extensibility of the fillers and adhesives.

4.0 Conclusions

The mechanical properties and microstructure of nanofilled and microfilled adhesives have been investigated with the following outcomes.

1. Rotafix’s CB10TSS adhesive was modified by incorporating fillers, namely nanosilica, liquid rubber and ceramic microparticles with the aim of improving mechanical properties. Liquid rubber addition causes the greatest improvement in strength followed by nanosilica additions.

2. The full strength of the room temperature cured epoxy adhesives 1, 2 and 3 is achieved after approximately 20 days whereas Epoxy4 is fully cured after 7 days.

3. Epoxy1 shows higher ductility compared to the filled adhesives Epoxy2, Epoxy3 and Epoxy4 but the addition of fillers improves the flexural, tensile and shear strengths and the elastic modulus.

4. SEM and TEM provide evidence of the quality of interfacial bonding between the matrix and the filler. The clear phase-separation in the cured rubber-filled adhesive and even distribution of nanosilica in Epoxy2 has improved the strength properties of the adhesives. However, Timberset shows weak bonding which results in a brittle failure mode.

5. The incorporation of a rubbery phase increases the bonding strength significantly which is the prime target in bonded-in timber connection. The value of shear strength for Epoxy3 was 7.45 MPa which is 1.2 times the value for Epoxy1.

6. Epoxy3 has the best combination of properties with maximum flexural strength of 80 MPa, tensile strength of 42.16 MPa, and shear strength of 7.45 MPa compared to 63 MPa, 26.81 MPa and 6.19 MPa respectively for Epoxy1.

Acknowledgments The authors would like to gratefully acknowledge the Universiti Teknologi Mara and Malaysian Service Department for the award of a scholarship to Z. Ahmad. Special thanks are conveyed to Rotafix Ltd for the supply of the adhesives. We also thank Ursula Potter and Hugh Perrott for their help in performing the TEM and SEM examinations.

5.0 References

1. Alam P., Ansell M.P. and Smedley D.. 2004. Flexural properties of steel and FRP reinforced LVL composites. Proceedings of the 8

th World Conference on Timber Engineering (WCTE

2004), III, Lahti, Finland, pp. 342-6. 2. Harvey K. and Ansell M.P.. 2000. Improved timber connections using bonded-in GFRP

rods. Proceedings of 6th World Conference on Timber Engineering, Whistler, British

Columbia, pp. 04

3. Kinloch A. J..1986. Rubber-toughened thermosetting polymers in Structural Adhesives, Kinloch A.J.(ed.), Elsevier Applied Science, England.

4. Kunz S.C., Sayre J.A. and Assink R.A..1982. Morphology and toughness characterization of

epoxy resins modified with amine and carboxyl terminated rubbers. Polymer, Vol. 23, No.13, pp. 1897-1906.

5. Kinloch A.J., Shaw S.J., Tod D.A. and Hunston D. L..1983. Deformation and fracture behaviour of a rubber-toughened epoxy: I. Microstructure and fracture studies. Polymer, Vol.24, pp. 1341-54.

6. McGarry F.J.. 1970. Building Design with Fibre Reinforced Materials, Proc. Roy. Soc.

London, A319, pp.1536-59.

7. Ratna D. and Banthia A.K.. 2000. Toughened epoxy adhesives modified with acrylate based liquid. rubber. Polymer International, Vol. 49, No. 3, pp 309.

8. Zhang M.Q., Rong M.Z., Zhang H.B. and Friedrich K.. 2003. Mechanical properties of low

nano-silica filled high-density polyethylene composites. Polymer Eng. Sci. Vol.43 pp. 490–500.

9. Meguid S.A. and Sun Y.. 2004. On the tensile and shear strength of nano-reinforced

composite interfaces, Materials & Design, Vol. 25, No.4. pp. 289-296.

10. Preghenella M., Pegoretti A. and Migliaresi C.. 2005. Thermo-mechanical characterization of fumed silica-epoxy nanocomposites. Polymer, accepted for publication in Nov 2005.

11. Ratna D., Becker O., Krishnamurthy R., Simon G.P. and Varley R.J.. 2003. Nanocomposites

based on a combination of epoxy resin, hyperbranched epoxy and a layered silicate, Polymer, Vol. 44, No. 24, pp.7449-57.

12. Ragosta G., Abbate M., Musto P., Scarinzi G. and Mascia L.. 2005. Epoxy-silica particulate

nanocomposites: Chemical interactions, reinforcement and fracture toughness. Polymer, Vol. 46, No.23, pp. 10506-16.

13. Ashby M.F.. 1999. Material selection in mechanical design, Butterworth-Heinemann, 2

nd

Edn. England.

14. Kinloch A.J.. 1986. Rubber-toughened thermosetting polymers in Structural Adhesive, Kinloch A.J.(ed.), Elsevier Applied Science, England.

15. Oksman K. and Clemon C., (1998), Mechanical properties of polypropylene -wood, J.

Applied Polymer Sci., Vol.67, pp. 1503-13.