Holzforschung, Vol. 66, pp. 623–631, 2012 • Copyright © by Walter de Gruyter • Berlin • Boston. DOI 10.1515/hf-2011-0171

Characterization of mixed-mode I/II fracture properties of adhesively bonded yellow-poplar by a dual actuator test frame instrument

Edoardo Nicoli 1, a , David A. Dillard 1, *, Charles E. Frazier 2 and Audrey Zink-Sharp 2

1 Engineering Science and Mechanics Department , Virginia Tech, Blacksburg, VA , USA

2 Wood Science and Forest Products Department , Virginia Tech, Blacksburg, VA , USA

* Corresponding author. Engineering Science and Mechanics Department, Virginia Tech, Blacksburg, VA 24061, USA Phone: + 1-540-231-4714 E-mail: dillard@vt.edu

Abstract

Experimental results for the fracture behavior under mixed-mode in-plane loading conditions of adhesively bonded wood specimens are reported. The material systems considered involved yellow-poplar ( Liriodendron tulipifera ), a hardwood of the Magnoliaceae family, as adherends bonded with two differ-ent adhesives, a moisture-cure polyurethane (PU) and a phenol/resorcinol/formaldehyde (PRF) resin. A dual actuator test frame permitted fi ne scanning of fracture behavior over a full range of mixed-mode I/II levels for double cantilever beam (DCB) geom-etry specimens. These tests showed that, in the considered mate-rial systems, the critical strain energy release rate, c , tends to increase as the mode-mixity of the loading increases. In particu-lar, the increase is steeper in proximity to pure mode II loading for the PRF bonded specimens. The experimental values of c obtained were fairly scattered, as is common when testing wood systems. This variability is due in part to the natural variability of wood but also to other factors such as the orientation of the grain in the bonded beams and variations of bondline thickness. In particular, measurements of adhesive layer thickness were per-formed. This analysis was implemented with microscopic exam-ination of samples cut from untested DCB specimens, where the bondline had not been disrupted by the test. Although the wood parts were power planed prior to bonding, rather large variations of the adhesive layer thickness were observed: on the order of 1 – 100 μ m for specimens bonded with the PU resin and 10 – 50 μ m for specimens bonded with the PRF resin, which showed somewhat more consistent fracture behavior.

Keywords: adhesive joints; bonded wood; double cantilever beam (DCB) geometry; dual actuator; fracture, fracture I and

a Currently at The Dow Chemical Company, Midland, MI 48667, USA.

fracture II, mixed-mode, bondline thickness; PRF resin; PU resin, yellow poplar.

Introduction

Wood has unique characteristics among natural materials in terms of its strongly anisotropic nature, providing good ratios in terms of strength to weight and stiffness to weight in the grain directions and relative ease in splitting parts perpendicularly to the grain direction. In addition, it is rela-tively easy to saw. This is the reason why wood is also a perfect construction material for homes (Wood Handbook 1999 ; Bowe 2001 ). There is a large body of literature con-cerning fracture properties of wood and bonded wood, focus-ing on mode I fracture (opening or tensile mode) (Ebewele et al. 1979, 1980 ; Triboulot et al. 1984 ; Gagliano and Frazier 2001 ; Conrad et al. 2003 ; Zheng et al. 2004 ; de Moura et al. 2008 ) or on mixed and mode II fracture (in-plane shear mode) (Cramer and Pugel 1987 ; Yoshihara and Ohta 2000 ; de Moura et al. 2006 , 2011 ; Oliveira et al. 2007, 2009 ; Singh et al. 2010 ). The fracture behavior of wood and wood com-posites was also reviewed by Stanzl -Tschegg and Navi (2009) and the modeling crack propagation was reviewed by Landis and Navi (2009) .

Fracture properties of bonded wood are affected by a large number of factors, such as wood structure, adhesive pen-etration into wood, and initial roughness of bonded surfaces as recognized early by Ebewele et al. (1979, 1980) . Since that time, this topic has remained a matter of investigation, as demonstrated by a few recent quotations concerning the adhesive penetration and the characterization of the bonding performance by various methods: scanning thermal micros-copy (Konnerth et al. 2008 ), image analysis, nanoindenta-tion (St ö ckel et al. 2010 ), air-coupled ultrasound inspection (Sanabria et al. 2011 ), or by macro- and micromechanical characterization (Follrich et al. 2010 ; Sahaf et al. 2011 ), just to mention a few.

A large scatter of values of critical strain energy release rate, c , of bonded wood has been often seen and regarded as resulting from the intrinsic variability of properties diffi cult to monitor in natural materials (Triboulot et al. 1984 ; Zheng et al. 2004 ; Kamke and Lee 2007 ). By contrast, a recent anal-ysis has identifi ed grain orientation effects as a structural, and possibly accountable, aspect as one of the factors that increase the scatter of c data (Nicoli et al. 2012a,b ).

The possibility of performing mixed-mode tests employing a dual actuator load frame (Singh et al. 2007 , 2010 ; Chaves

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

624 E. Nicoli et al.

et al. 2011 ) has facilitated mixed-mode testing of bonded sys-tems. This testing frame permits one to impose asymmetric loading conditions with two independently controlled actua-tors. Depending on the specimen type and its mounting onto the machine, loading conditions ranging from pure mode I to pure mode II can be easily achieved.

In the analysis of double cantilever beam (DCB) and other standard bonded beam tests, several improved approaches can be used for analyzing the data, including the corrected beam theory (CBT) and the experimental compliance method (ECM) described by Blackman et al. (1991, 2006) and Brunner et al. (2001) . These techniques address several limitations of simple beam theory by including root rotation at the crack tip, shear deformation of the beams, presence of load blocks, effects of beam thickness, and inaccuracies in crack length readings. These issues have been extensively described for mode I in the literature (Williams 1989 ; Blackman et al. 1991 ; Davies et al. 1992 ; Cotterell et al. 2006 ). The CBT method evaluates the effective compliance of the DCB beams and corrects the crack length reading using a linear fi t of the cube root of com-pliance vs. crack length data. The ECM similarly evaluates elastic characteristics of the tested specimens with a power law fi t. ECM and CBT for DCB specimens are both tradition-ally based on mode I tests and require experimental data from pure mode I tests, although adaptations of the techniques to pure mode II have also been developed (Wang and Williams 1992 ; Blackman et al. 2003, 2005, 2006 ). One goal of this study was to test mixed-mode conditions, but each specimen was partially tested at the same level of mode-mixity, namely pure mode I. This allowed for specimen calibration using the ECM and CBT methods, and allowed mixed mode testing to begin with a longer crack length, minimizing instability issues (Davies et al. 2001 ). With this condition, the tests proceeded through three separate phases; the fi rst and the third were con-ducted in mode I and the second phase involved mixed-mode conditions obtained by imposing different displacement rates to the two beams. As ECM and CBT fi t data of compliance with crack length, having portions of mode I tests performed with both short and long crack lengths (corresponding to the fi rst and third loading phases) was deemed to be important for an accurate evaluation of the parameters defi ned by the two fi tting procedures. Note that both CBT and ECM assume the elastic characteristic of the bonded beams to be uniform and constant along the beam length, as expected for a homoge-neous material. For wood this assumption is not always cor-rect, as explained by Liswell (2004) . Nevertheless, ECM and CBT have been traditionally applied for measuring the frac-ture properties of bonded wood (Gagliano and Frazier 2001 ; Yoshihara and Nobusue 2007 ). The effects of this assumption have been reviewed in the literature (Nicoli et al. 2012a,b ).

This paper describes a procedure that was applied for eval-uating c values in mixed-mode I/II loading conditions using DCB geometry specimens in a dual actuator load frame and the experimental results that were obtained. Yellow-poplar ( Liriodendron tulipifera ) was selected for the adherends and two different commercial adhesives were applied: a moisture-cure polyurethane (PU) and a phenol/resorcinol/formalde-hyde (PRF) resin.

Materials and methods

Experimental equipment

The dual actuator load frame (DALF) (Singh et al. 2006 , 2010 ) is a servo-hydraulic testing machine capable of providing variable mode-mixity for fracture mechanics studies by imposing asymmetric dis-placements or loads on symmetric bonded beam specimens, such as those widely used for DCB specimens. In principle, the investigation of asymmetric specimens is also possible; however, having a dual actuator, it is more practical to keep the specimen geometry con-stant and to impose asymmetric loading conditions. In the DALF, each actuator is independently controlled and is equipped with a load cell and a displacement sensor (LVDT) that provides information for quantitative evaluation of specimen behavior.

Features of DALF include the ability to test a common bonded joint specimen confi guration over the full range of mode I/II mix-ity by independently controlling the actuator motion, the ability to quickly change mode mixity on a single specimen without use of an additional fi xture or modifying the test jig, and the ability to test specimens under general loading conditions spanning a wide range of displacement speeds and in fatigue. The DALF, illustrated in Figure 1 a, was built by McGaw Technologies Inc. (Fairview Park, OH, USA). Unless the test is performed in pure mode I, the specimen is clamped in a vise at the lower end, and loads are applied to the debonded ends of the beams by pins and clevises attached to the load cells of the actuators. A controller drives the two actuators; displace-ments and forces for the two actuators are recorded as the test pro-gresses by a computer equipped with a data acquisition card (DAQ PCI 6229, National Instruments Inc., Austin, TX, USA). The crack length values of the specimen are read periodically by the operator using a magnifi cation lens and facilitated by white correction fl uid and a paper ruler applied on one side of the specimen.

The values of crack length are entered into the data acquisition system. The simplest control algorithm imposes different combina-tions of displacement rates to the two actuators as described in Nicoli (2010) . The combination of different displacement rates results in different levels of in-plane mode mixity with a standard DCB speci-men. Microscope analysis was performed with a Zeiss Axioskop (Carl Zeiss, Oberkochen, Germany) microscope, the camera for image acquisition is a DS-Fi1 (Nikon Inc., Melville, NY, USA). For preparing microscope samples, some of the DCB specimens were not tested on the DALF, but cut into microscope samples, which are 40- μ m thick slices of area approximately 5 × 5 mm. The slices were also stained with two different solutions, 0.5 % safranin O or 0.5 % toluidine blue O, to fi nd the combination that would better highlight the presence of adhesive penetrating into the sample. After several initial tests, the combinations of visible light at the microscope, sa-franin stain for parts bonded with the PU adhesive, and no stain for parts bonded with the PRF adhesive gave acceptable results.

Materials and specimens

The study focuses on yellow-poplar ( L. tulipifera ), a hardwood of the Magnoliaceae family, and two adhesives, a moisture-cure PU and a two-component PRF resin. Specimens consisted of 10-mm thick wooden adherends bonded together with one of the two adhesives. Preparation of the specimens started with two power-planed 10-mm thick boards with dimensions of approximately 250 mm × 140 mm. The boards were carefully cut from larger stock so that the orienta-tion of the wood grain was between 3 ° and 6 ° with respect to the intended bond plane (Gagliano and Frazier 2001 ). This confi guration is benefi cial in preventing the crack from moving from the bonded

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

Characterization of mixed-mode I/II fracture properties 625

layer into the adherend substrate and because the fairly constant grain orientation provides more consistent stiffness characteristics, limiting the sources of variability in the experiments. For the same reason, boards containing knots or macroscopic defects were exclud-ed from experimentation.

A 50-mm wide region of the board near one of the edges was colored with a wax crayon to limit adhesion and provide a precracked area in the fi nal bonded specimens. One board is illustrated in Figure 1b. To bond the boards, one of them was placed on a precision bal-ance and, after zeroing the load, the amount of adhesive equal to the upper limit indicated on the technical sheet was poured on the board. Then, removing the board from the balance, the adhesive was spread to obtain a uniform distribution and the second board was placed on the top of the adhesive layer. Finally, the assembly was placed in a cold press and, during this pressing phase, correct alignment of the boards to be bonded was facilitated by means of lateral constraints.

Adhesive quantity, applied pressure, and curing time were as specifi ed in the technical sheets of the adhesives. Sometimes, multi-ple stacks of bonded boards were placed in the press and compressed at the same time. Particular care was taken to complete the bonding procedure without exceeding the prescribed handling time, which was on the order of a few minutes for both adhesives. The adhesive cured at room temperature inside the press while the prescribed pres-sure, 1 MPa (150 psi) for the PF resin and 0.68 MPa (100 psi) for the PRF resin, was applied for at least 24 h.

After curing, the bonded pieces were cut into 20-mm wide speci-mens and conditioned for at least 2 weeks in a chamber at 60 % relative humidity and 21 ° C (70 ° F). Additional operations were performed on the specimens to prepare them for testing. A hole was drilled through the width direction on the debonded end of each adherend for loading clevis pins (see experimental equipment section). Next, one side of

the bonded specimens was painted with white typewriter correction fl uid to enhance crack tip detection and, fi nally, a paper ruler was bonded on the same side of the specimen to facilitate crack length measurements.

As already mentioned, attention regarding the orientation of the wood grain was given throughout the assembly procedure. The orienta-tion of the wood grain in the fi nal specimen was between 3 ° and 6 ° with respect to the bondplane, as illustrated in Figure 1c. The defi nition of axis and section directions in the wood for our specimen is illustrated in Figure 1d. Note that the longitudinal axis is not parallel to the specimen cut because of the 3 – 6 ° orientation just described.

Test method

With the dual actuator load frame, the standard DCB specimen geometry can be tested over the full range of in-plane mode-mixity. By simultaneously applying different displacement rates with the two independently controlled actuators, different levels of mode-mixity can be induced at the crack tip. Given the utility of performing part of the test in pure mode I, the displacements imposed during the tests were as shown in Figure 2 . A total of 33 specimens prepared with the PRF adhesive and 28 specimens prepared with the PU adhesive were tested. Some tests were run in pure mode I or pure mode II, but typically mixed-mode loading for the second loading phase was performed for most specimens, following profi les such as illustrated in Figure 2. The resulting critical fracture energy, c, values can each be represented as a sum of mode I and mode II components, with the relative contributions depending on the mode-mixity angle, ψ . The relations between c, its mode I and mode II components and the angle ψ are shown in Eqs. (1) and (2).

( ) ( )c c c= +I II

G G G (1)

Hydraulicactuator

Hydraulicpump

Actuatorsupport

Bondedspecimen Clamp

Servo valve

T, tangentialR, radialL, longitudinal

RL

T

140

50

250

a

b

d

c

Figure 1 Details of materials and methods. (a) Dual actuator load frame and detail of specimen connection. (b) Wooden board for specimen construction. (c) Double cantilever beam (DCB) specimen: note the grain orientation. (d) Direction of axes in DCB specimens.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

626 E. Nicoli et al.

( )( )c

c

Arctan⎛ ⎞

ψ = ⎜ ⎟⎜ ⎟⎝ ⎠II

I

G

G

(2)

These mode I and mode II components of a given c are not to be confused with the critical fracture energies for mode I and mode II, I c and I I c , respectively. Various methods for partitioning the mode I and II components of the loads have been proposed and can lead to different results. As suggested by Williams (1988) , a method based on the global analysis was chosen for this study. In particular, the partitioning of loads was based on the applied moments (Hutchinson and Suo 1991 ; Kinloch et al. 1993 ). In the case of the dual actuator load frame, the mode I and mode II components of the loads applied to each beam can be easily calculated as a function of the loads ap-plied by the two actuators (Nicoli et al. 2009 ).

The two components F I and F II are indicated in Eqs. (3) and (4), where F R and F L are, respectively, the force readings at the right and left actuator and are positive when the actuator is pulling the beam.

12

= +I R LF F F

(3)

1 -2

=II R LF F F

(4)

Results and discussion

The curves, in Figures 3 and 4 , show the results for the two material systems. The trends of the mode I and mode II com-ponents of c as well as the total c are visible. With the tests performed at constant displacement rates, the data of the sin-gle specimens are presented as a function of ψ , as the mode-mixity angle increased during the test, as analytically shown (Nicoli 2010 ).

The signifi cant data scatter in Figures 3 and 4 is often common for wood and bonded wood fracture tests (Ebewele et al. 1979, 1980 ; Triboulot et al. 1984 ). In particular, the fracture envelopes in Figure 4 illustrate that the individual specimens are relatively consistent, but that specimen-to-specimen variability is considerable. In particular, within an individual specimen the scatter seems, in most cases, to have some trends, though not necessarily consistent with

Displacement right beam

Displacement left beam

0 50 100 150 200 250

Mode I test Mode I testMixed mode test(at constant rates)

Time (s)

25D

ispl

acem

ent (

mm

)

20

15

10

5

0

-5

-10

-15

-20

Figure 2 Test procedure presented as the displacement as a func-tion of time.

ψ (°)0 30 60 90

PRF bonded

PRF bonded

a’ PU bonded

b’ PU bonded

b

a 1600

1400

1200

1000

(Gc)

I and

(Gc)

II (J m

-2)

Gc (

J m

-2)

(Gc)

I and

(Gc)

II (J m

-2)

Gc (

J m

-2)

800

600

400

200

0

1600

1400

1200

1000

800

600

400

200

0

800

900

700

600

500

400

300

200

100

0

800

900

700

600

500

400

300

200

100

0

Figure 3 c components as a function of the mode-mixity angle, ψ , for PRF- and PU-bonded specimens. (a, a ’ ) ( c ) I and ( c ) II com-ponents (× mode I component, � mode II component). (b, b ’ ) Sum of c components.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

Characterization of mixed-mode I/II fracture properties 627

the trends of other specimens tested at the same levels of mode-mixity. This aspect can be interpreted that the scatter is probably not related to random variability or noise-like disturbances during the test, but may depend on specimen or bond variability.

Moreover, the variability also seems not to be related to R-curve effects that are seen in some material systems (Irwin et al. 1954 ), as the observed trends of c are not, for example, always monotonic, i.e., are not always rigorously dependent of ψ . Figure 3b shows c for the PRF-bonded specimens. In particular, the value of c associated with small mode-mixity angles ( ψ < 30 ° ) is fairly constant, whereas for higher ψ the c increases rapidly. c for mode II loading is approximately eight times larger than that in mode I loading.

The fracture envelope of Figure 4a shows that data scat-ter is more pronounced for tests performed close to mode I loading. Note that the horizontal and vertical axes in Figure 4a have different scales. Figures 3a ’ ,b ’ , and 4b show the results of PU-bonded specimens. Also in this case, most of the observations are similar to the results of the PRF-bonded specimens including the appreciable data scatter. The latter is due to the systematic differing trends of c depending on the specimen tested rather than some random variability. Also in this case, the value of c is almost constant for ψ below 30 ° and then increases, as can be seen in Figure 3b ’ . The values of

PU bonded

PRF bonded1600

1400

1200

1000

800

600

400

200

0

00 50 100 150 200 250 300 350

100

200

300

400

500

600

700

800

900

(Gc)I (J m-2)

(Gc)

II (J

m-2

)(G

c)II (

J m

-2)

a

b

Figure 4 Fracture envelops [plots ( c ) II vs. ( c ) I ] for PRF- and PU-bonded specimens.

c associated with pure mode I loading are fairly similar for the two material systems, whereas for mode II loading the c of the PRF-bonded DCBs is approximately twice that of the PU-bonded specimens. For mixed-mode I/II (Figure 5 ) the two material systems have very similar behavior for ψ < 70 – 75 ° . One positive outcome of the results was that the grain orien-tation has successfully prevented the crack from moving into the adherends even for tests performed with mode II load-ing and all failures were observed to occur within the bond region.

Visual analysis of tested specimens

The common visual inspection of failure surfaces often per-mits the recognition of regions associated with adhesive or cohesive failure. This simple procedure is not particularly useful in bonded wood samples, however, where the adhe-sive has a layer thickness generally smaller than 100 μ m and often a color that makes it diffi cult to detect adhesive traces on the adherends. Other traditional surface analysis tech-niques, such as the X-ray photoelectron spectroscopy, may not be as helpful in studying bonded wood samples because the chemical compositions of the adhesive and adherends are similar. In this study, the failure surfaces have non-uniform characteristics, which sometimes alternate between smooth and rough areas (for the PU-bonded systems) or lighter and darker areas (for the PRF-bonded systems), suggesting dif-ferent failure mechanisms. Regions in which the adhesive plastically deforms before failing require more energy than regions in which the failure is brittle and, particularly for specimens prepared with PU adhesive, a connection was observed between the roughness of the bonded surface after failure and the recorded c. Figure 6 is an example on how c varies in a test performed with pure mode I loading on one of the specimens prepared with the PU adhesive. c var-ies between 300 and 700 J m -2 , although one would expect a fairly constant value or a value monotonically changing as a function of the crack length. The constant value for c is

1800

1600

1400

1200

1000

PRFPU

800

600

400

200

00 30 60 90

Mode I Mode II

ψ (°)

Gc (

J m

-2)

Gc=(Gc)I+(Gc)II

Figure 5 c as a function of mode-mixity angle, ψ , for PRF- and PU-bonded specimens.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

628 E. Nicoli et al.

expected considering that c is a property of the tested mate-rial system.

A monotonic trend for the failure of c can, by contrast, be an indication of the dependence of c on the crack growth rate, in view of the fact that the crack growth slows down as the crack develops, when the tests are performed at constant imposed displacement rates. In this case, other specimens have similar behavior, presenting considerable variations of c during the single test, but have different trends of the c. The only aspect that consistently correlates with the level of c is the roughness of the surface after failure: areas of high

c values correspond to sections of failed surface with accen-tuated roughness and whitening of the adhesive, whereas reduced c values correspond to smooth portions of the surface. In the example illustrated in Figure 6, particularly rough failure surface areas were obtained at different por-tions of the failed surface. The different trends in the graph are interpretable that c is not infl uenced by other factors such as the crack length or speed, which change monotoni-cally during the test.

Effect of adhesive layer thickness

Further analysis was performed on the state of the adhesive layer to detect possible causes of the scatter of experimen-tal results. The adhesion mechanisms in wood samples are generally very different from what occurs in metal or com-posite samples. Adhesive penetrates considerably deeper into the wooden beams and the bondline thickness is very thin (Wood Handbook 1999 ). Moreover, during the production of bonded wood specimens, pressure applied to the pieces to be bonded is controlled rather than the bondline thickness. However, the latter is a factor that infl uences the extension of the plastic zone and the value of c (Kinloch and Blackman 2004 ). Thus, the variability of the bondline thickness can lead to data scatter, as found in the experimental results previously illustrated.

The procedure for measuring the bondline thickness was based on methods reported in the literature (Johnson and Kamke 1992 ; Zheng et al. 2004 ; Kamke and Lee 2007 ). The analysis was performed with a fl uorescence microscope

working with transmitted light (see Material and methods sec-tion). Some of the samples for microscopic analysis are shown in Figure 7 a. The transmitted light microscope for checking bondline penetration needs thin samples that permit light to be transmitted through. This aspect complicates the measure-ments. The required microscope samples can be cut from DCB specimens that either have or have not been tested. In the fi rst case, the information regarding the adhesive layer thickness is lost because the initial thin layer has been disrupted during the test. In the second case, the cutting of the sample destroys the DCB specimen. Consequently, it is not feasible to compare the local values of c and the values of adhesive penetration and layer thickness in the same position.

For untested specimens, some of the obtained pictures are presented in Figure 7b,c. The pictures illustrate the adhesive presence in the bondline (vertical line) and in the vessels around it. For the PU-bonded specimens the presence of adhe-sive is less evident, as with the safranin stain both adhesive and wood exhibit a reddish coloration. Adhesive in the wood vessels is visible because of its slightly pink color. The results show that the values of adhesive layer thickness and penetra-tion in the substrate vary considerably, even within a single specimen. Although all the wood parts for DCB tests were power planed before bonding the specimens, the thickness of the PRF-adhesive layer varies between 10 and 50 μ m and the penetration between 0.1 and 0.75 mm. The thickness of the PU-adhesive layer varies between 1 and 100 μ m and the penetration between 0.3 and 1.5 mm. The reason for greater variation in PU-adhesive thickness is not clear, but could possibly be associated with the different viscosity of the two adhesives. The larger variability of c data from PU-bonded specimens may correlate with the more dramatic adhesive layer thickness variation.

As already mentioned, with this type of analysis both mea-sures of adhesive layer thickness and fracture properties on the same DCB specimen cannot be obtained. Nevertheless, measurements can still be performed in specimens that were adjacent in the initial board (Figure 1b). These comparisons were performed for specimens tested in mode I, where the additional perturbation given by the change of the angle of mode-mixity is not present. In these measurements, a direct

GIc (J

m-2

)

800

600

400

200

00 50 100 150

Crack length (mm)200

Figure 6 Failure surfaces and local c for PU-bonded specimen tested in mode I.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

Characterization of mixed-mode I/II fracture properties 629

relation between adhesive layer thickness and measured local value of I c was apparent, as already seen in different material systems (Kawashita et al. 2008 ; Qian and Sun 2008 ).

An example is given in Figure 8 . Here, the value of adhe-sive layer thickness is lower than 20 μ m for sections where the value of I c is around 200 J m -2 , whereas the thickness is around

a

=0.5 mm

=0.5 mm

=0.25 mm

=0.25 mm

b

c

Figure 7 Light microscopic analyses of the bonds. (a) Slices for microscope analysis ready to be tested (from left to right: toluidine stained, unstained, safranin stained). (b) Images for PRF-bonded untested DCB (unstained sample, radial-tangential section). (c) Images for the PU-bonded untested DCB (stained sample, radial-tangential section).

Crack length (mm)

600

400

200

00

50 100

Mode I test

=60 μm

=60 μm =60 μm

150 200 250

GIc (J

m-2

)

Figure 8 Local adhesive layer thicknesses and c values as a function of crack length for a PU-bonded DCB specimen tested in mode I.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

630 E. Nicoli et al.

50 μ m in sections where the value of I c peaks to almost 500 J m -2 . The dependence of the c on the adhesive layer thickness can be explained in the following way. In extremely thin layers the interfaces between adhesive and adherends infl uence the stress distribution at the crack tip and limit the amount of adhesive that is subjected to high stress levels in the area around the crack tip. With an increase of layer thickness, the I c value is expected to grow, as more adhesive is allowed to plastically deform ahead of the crack tip. The increase of I c is not indefi nite as the effect of the interfaces fades and I c reduces to the value for the bulk polymer (Kinloch and Shaw 1981 ; Kinloch and Blackman 2004 ), if the layer thickness is large enough compared to the radius of the plasticized zone. In bonded wood, the thickness of the adhesive layer is particu-larly small, as demonstrated by the measurements. Thus, it is very likely that the radius of the plastic zone is larger than the layer thickness, as local increase of layer thickness increases

I c in static tests, as shown in Figure 8.

Conclusions

This study investigates the mixed-mode I/II fracture proper-ties of yellow-poplar adherends bonded with a moisture-cure PU or a PRF resin. DCB specimen confi gurations were tested in a dual actuator load frame. The experimental procedure consists of three separate phases, two of which were mode I tests. This approach considers some aspects of the CBT and ECM in the context of the mixed-mode test results. This is particularly important when testing DCBs of non-uniform materials, where elastic properties of the adherend are not known. The curves of the components of c versus ψ and fracture envelopes show data scatter that is consistent with what is usually found in bonded wood. One of the possible reasons for the data scatter, the variability of bondline thickness, was investigated with microscope analysis and is presented in this paper. Measurements found that the PU-adhesive layer thick-ness could vary between 1 and 100 μ m and between 10 and 50 μ m for specimens prepared with PRF resin. Comparison of adhesive layer thickness and mode I critical fracture energy is interpreted that the thickness variation has a strong infl uence on c that is measured with the DCB samples. In particular, sections of reduced adhesive layer thickness were associated with lower fracture energies and vice versa.

Acknowledgements

The authors would like to acknowledge the fi nancial support provided by the National Science Foundation under contract DMR-0415840 for constructing the dual actuator load frame and by the Wood-Based Composites Center and the Sustainable Engineered Materials Institute at Virginia Tech for supporting the present study utilizing this instru-ment. We are grateful for helpful collaboration in the preparation of the specimens to Dakai Ren and David Jones of the Wood Science and Forest Products Department at Virginia Tech. We also acknowl-edge the use of facilities in the Engineering Science and Mechanics Department and the Wood Science and Forest Products Department at Virginia Tech, as well as the interdisciplinary research environment fostered by the Macromolecules and Interfaces Institute.

References

Blackman, B., Dear, J.P., Kinloch, A.J., Osiyemi, S. (1991) The cal-culation of adhesive fracture energies from double-cantilever beam test specimens. J. Mat. Sci. Lett. 10:253 – 256.

Blackman, B.R.K., Kinloch, A.J., Paraschi, M. On the mode II load-ing of adhesive joints. European Structural Integrity Society, Elsevier, 2003, vol. 32. doi: 10.1016/s1566-1369(03)80103-6.

Blackman, B.R.K., Kinloch, A.J., Paraschi, M. (2005) The determi-nation of the mode II adhesive fracture resistance, GIIC, of struc-tural adhesive joints: an effective crack length approach. Eng. Fract. Mech. 72:877 – 897.

Blackman, B.R.K., Brunner, A.J., Williams, J.G. (2006) Mode II fracture testing of composites: a new look at an old problem. Eng. Fracture Mech. 73:2443 – 2455.

Bowe, S. Wood Use and Society. Forest Facts. Series editor: Department of Forest Ecology and Management – School of Natural Resources Document. No. 93. University of Wisconsin, Madison, WI, 2001, p. 4.

Brunner, A.J., Blackman, B.R.K., Davies, P. (2001) Mode I delami-nation. In: Fracture Mechanics Testing Methods for Polymers, Adhesives and Composites. Eds. Moore, D.R., Pavan, A., Williams, J.G. European Structural Integrity Society, Elsevier. ISBN: 978-0-08-043689-0.

Chaves, F.J.P., de Moura, M.F.S.F., da Silva, L.F.M., Dillard, D.A. (2011) Numerical analysis of the dual actuator load test applied to fracture characterization of bonded joints. Int. J. Solids Struct. 48:1572 – 1578.

Conrad, M.P.C., Smith, G.D., Fernlund, G. (2003) Fracture of dis-continuous wood-adhesive bonds. Int. J. Adhesion Adhesives 23:39 – 47.

Cotterell, B., Hbaieb, K., Williams, J.G., Hadavinia, H., Tropsa, V. (2006) The root rotation in double cantilever beam and peel tests. Mech. Mater. 38:571 – 584.

Cramer, S., Pugel, A. (1987) Compact shear specimen for wood mode II fracture investigations. Int. J. Fracture 35:163 – 174.

Davies, P., Kausch, H.H., Williams, J.G., Kinloch, A.J., Charalambides, M.N., Pavan, A., Moore, D.R., Prediger, R., Robinson, I., Burgoyne, N., Friedrich, K., Wittich, H., Rebelo, C.A., Torres Marques, A., Ramsteiner, F., Melve, B., Fischer, M., Roux, N., Martin, D., Czarnocki, P., Neville, D., Verpoest, I., Goffaux, B., Lee, R., Walls, K., Trigwell, N., Partridge, I.K., Jaussaud, J., Andersen, S., Giraud, Y., Hale, G., McGrath, G. (1992) Round-robin interlaminar fracture testing of carbon-fi bre-reinforced epoxy and PEEK composites. Compos. Sci. Technol. 43:129 – 136.

Davies, P., Blackman, B., Brunner, A.J. (2001) Mode II delamina-tion. In: Fracture Mechanics Testing Methods for Polymers, Adhesives and Composites. Eds. Moore, D.R., Pavan, A., Williams, J.G. European Structural Integrity Society, Elsevier. ISBN: 978-0-08-043689-0.

de Moura, M.F.S.F., Silva, M.A.L., de Morais, A.B., Morais, J.J.L. (2006) Equivalent crack based mode II fracture characterization of wood. Eng. Fract. Mech. 73:978 – 993.

de Moura, M.F.S.F., Morais, J.J.L., Dourado, N. (2008) A new data reduction scheme for mode I wood fracture characteriza-tion using the double cantilever beam test. Eng. Fract. Mech. 75:3852 – 3865.

de Moura, M.F.S.F., Oliveira, J.M.Q., Morais, J.J.L., Dourado, N. (2011) Mixed-mode (I + II) fracture characterization of wood bonded joints. Constr. Build. Mater. 25:1956 – 1962.

Ebewele, R., River, B., Koutsky, J. (1979) Tapered double cantile-ver beam fracture tests of phenolic-wood adhesive joints. Wood Fiber Sci. 11:197 – 213.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM

Characterization of mixed-mode I/II fracture properties 631

Ebewele, R., River, B., Koutsky, J. (1980) Tapered double cantile-ver beam fracture tests of phenolic-wood adhesive joints: Part II. Effects of surface roughness, the nature of surface rough-ness, and surface aging on joint fracture energy. Wood Fiber Sci. 12:40 – 65.

Follrich, J., St ö ckel, F., Konnerth, J. (2010) Macro- and micro-mechanical characterization of wood-adhesive bonds exposed to alternating climate conditions. Holzforschung 64:705 – 711.

Gagliano, J.M., Frazier, C.E. (2001) Improvements in the fracture cleavage testing of adhesivelybonded wood. Wood Fiber Sci. 33:377 – 385.

Hutchinson, J.W., Suo, Z. (1991) Mixed mode cracking in layered materials. Adv. Appl. Mech. 29:63 – 191.

Irwin, G.R., Kies, J.A. (1954) Critical energy rate analysis of fracture strength. Welding Res. Suppl. 19:193 – 198.

Johnson, S.E., Kamke, F.A. (1992) Quantitative analysis of gross adhesive penetration in wood using fl uorescence microscopy. J. Adhesion 40:47 – 61.

Kamke, F.A., Lee, J.N. (2007) Adhesive penetration in wood – a review. Wood Fiber Sci. 39:205 – 220.

Kawashita, L.F., Kinloch, A.J., Moore, D.R., Williams, J.G. (2008) The infl uence of bond line thickness and peel arm thickness on adhesive fracture toughness of rubber toughened epoxy-aluminum alloy laminates. Int. J. Adhesion Adhesives 28:199 – 210.

Kinloch, A.J., Shaw, S.J. (1981) The fracture resistance of a tough-ened epoxy adhesive. J. Adhesion 12:59 – 77.

Kinloch, A.J., Wang, Y., Williams, J.G., Yayla, P. (1993) The mixed-mode delamination of fi bre composite materials. Compos. Sci. Technol. 47:225 – 237.

Kinloch, A.J., Blackman, B.R.K. (2004) The infl uence of adhesive bond line thickness on the toughness of adhesive joints. In: The Application of Fracture Mechanics to Polymers, Adhesives and Composites. Ed. Moore, D.R. European Structural Integrity Society, Elsevier. ISBN: 978-0-08-044205-1.

Konnerth, J., Harper, D., Lee, S.-H., Rials, T.G., Gindl, W. (2008) Adhesive penetration of wood cell walls investigated by scan-ning thermal microscopy (SThM). Holzforschung 62:91 – 98.

Landis, E.N., Navi, P. (2009) Modeling crack propagation in wood and wood composites. A review COST Action E35 2004 – 2008: wood machining – micromechanics and fracture. Holzforschung 63:150 – 156.

Liswell, B. Exploration of Wood DCB Specimens using Southern Yellow Pine for Monotonic and Cyclic Loading. Engineering Mechanics, Virginia Tech, Blacksburg, VA. Master ’ s thesis, 2004.

Nicoli, E. Characterization of Mixed-Mode Fracture Testing of Adhesively Bonded Wood Specimens. Engineering Science & Mechanics, Virginia Tech, Blacksburg, VA. PhD thesis, 2010.

Nicoli, E., Singh, H.K., Dillard, D.A., Frazier, C.E. Mixed mode fracture testing of adhesively bonded specimens using a dual-actuator load frame. In: Proceedings of the 12th International Conference on Fracture, Ottawa, Canada, 2009.

Nicoli, E., Dillard, D.A., Frazier, C. (2012a) Effects of system-atic variation of wood adherend bending stiffness on fracture properties: part 1 – estimating the infl uence of grain angle. Holzforschung, in press.

Nicoli, E., Dillard, D.A., Frazier, C. (2012b) Effects of systematic variation of wood adherend bending stiffness on fracture prop-erties: part 2 – revisiting traditional DCB analysis methods. Holzforschung, in press.

Oliveira, J.M.Q., de Moura, M.F.S.F., Silva, M.A.L., Morais, J.J.L. (2007) Numerical analysis of the MMB test for mixed-mode I/II wood fracture. Compos. Sci. Technol. 67:1764 – 1771.

Oliveira, J.M.Q., de Moura, M.F.S.F., Morais, J.J.L. (2009) Application of the end loaded split and single-leg bending tests to the mixed-mode fracture characterization of wood. Holzforschung 63:597 – 602.

Qian, H., Sun, C.T. (2008) Effect of Bondline Thickness on Mode I Fracture in Adhesive Joints. In: Proceedings of the American Institute of Aeronautics and Astronautics Inc., Schaumburg, IL. p. AIAA; ASME; ASCE; AHS; ASC.

Sahaf, A., Englund, K., Laborie, M.-P.G. (2011) Tack and shear strength of hybrid adhesive systems made of phenol-formaldehyde, dextrin and fi sh glue, and acrylic pressure-sensitive adhesive. Holzforschung 66:73 – 78.

Sanabria, S.J., Furrer, R., Neuenschwander, J., Niemz, P., Sennhauser, U. (2011) Air-coupled ultrasound inspection of glued laminated timber. Holzforschung 65:377 – 387.

Singh, H.K., Soojae, P., Ohanehi, D.C., Dillard, D.A. (2006) A design space for a novel dual actuator mixed-mode test frame. In: Proceedings of the 29 th Annual Adhesion Society Meeting, Jacksonville, FL.

Singh, H.K., Chakraborty, A., Soojae, P., Ohanehi, D.C., Dillard, D.A. (2007) Mixed-mode fracture test results from a novel dual-actuator load frame. In: Proceedings of the 30 th Annual Adhesion Society Meeting, Tampa, FL.

Singh, H.K., Chakraborty, A., Frazier, C.E., Dillard, D.A. (2010) Mixed mode fracture testing of adhesively bonded wood specimens using a dual actuator load frame. Holzforschung 64:353 – 361.

Stanzl-Tschegg, S.E., Navi, P. (2009) Fracture behaviour of wood and its composites. A review COST Action E35 2004 – 2008: wood machining – micromechanics and fracture. Holzforschung 63:139 – 149.

St ö ckel, F., Konnerth, J., Kantner, W., Moser, J., Gindl, W. (2010) Tensile shear strength of UF- and MUF-bonded veneer related to data of adhesives and cell walls measured by nanoindentation. Holzforschung 64:337 – 342.

Triboulot, P., Jodin, P., Pluvinage, G. (1984) Validity of fracture mechanics concepts applied to wood by fi nite element calcula-tion. Wood Sci. Tech. 18:51 – 58.

Wang, Y., Williams, J.G. (1992) Corrections for mode II fracture toughness specimens of composites materials. Compos. Sci. Technol. 43:251 – 256.

Williams, J.G. (1988) On the calculation of energy release rates for cracked laminates. Int. J. Fract. 36:101 – 119.

Williams, J.G. (1989) End corrections for orthotropic DCB speci-mens. Compos. Sci. Technol. 35:367 – 376.

Wood Handbook. Wood as an Engineering Material. USDA Forest Service, Forest Products Laboratory, Madison, WI, 1999.

Yoshihara, H., Ohta, M. (2000) Measurement of mode II fracture toughness of wood by the endnotched fl exure test. J. Wood Sci. 46:273 – 278.

Yoshihara, H., Nobusue, K. (2007) Mode I and mode II fracture toughness of densifi ed Sitka spruce fabricated in an airtight atmo-sphere with high-temperature steam. Holzforschung 62:82 – 85.

Zheng, J., Fox, S.C., Frazier, C. (2004) Rheological, wood penetra-tion, and fracture performance studies of PF/pMDI hybrid resins. Forest Prod. J. 54:74 – 81.

Received August 22, 2011. Accepted December 13, 2011.Previously published online February 13, 2012.

Brought to you by | Univ Libry Serials ReceivingAuthenticated | 128.173.163.217

Download Date | 10/29/12 5:17 PM