THE EFFECT OF FIBER ORIENTATION ON THE ... - ScienceDirect

(1988)

Wear, 121

127

127 - 141

THE EFFECT OF FIBER ORIENTATION ON THE ABRASIVE WEAR BEHAVIOR OF POLYMER COMPOSITE MATERIALS M. CIRINO Center

for Composite

Materials,

University

of Delaware,

Newark,

DE (U.S.A.)

K. FRIEDRICH Polymer (F.R.G.)

and

Composites

Group,

Technical

University

of Delaware,

Newark,

of Ham burg-Harburg,

Ham burg

R. B. PIPES College (Received

of Engineering, May X3,1987;

University accepted

DE (U.S.A.)

June 30,1987)

The effect of fiber orientation on the dry abrasive-dominant wear behavior of continuous-carbon-fiber-reinforced polyetheretherketone (CFPK) and continuous-aramid-fiber-reinforced epoxy (K49/3501-6) was investigated. A pin-on-flat type test apparatus was employed in determining the wear rates as a function of fiber orientation for the two materials. Different wear rate distributions were found for each material, although the optimum wear resistance occurred when the fibers were oriented normal to the sliding surface in each case. The influence of fiber-abrasive particle interactions on the wear rate distribution is discussed with regard to microscopic observations of the worn surfaces. Lastly, the wear rate distributions are described empirically allowing the prediction of the wear rate at any arbitrary fiber orientation.

1. Introduction The wear behavior of composite materials is influenced by the orientation of the fibers relative to the sliding direction. The dependence of the wear rate on the fiber orientation is difficult to model in a quantitative manner. To increase further the complexity of the matter, the effect of the fiber orientation on the wear rate also depends on the type of composite material under consideration as well as on the type of tribological system under which it operates. This anisotropic wear behavior has been investigated for several types of composite materials such as polymer matrix and metal matrix [ 1,2] composites, short fiber [ 3 - 61, continuous fiber [ 7 - lo] and woven [ll] composites and even hybrid composites [12]. These studies OQ43-1648/88/$3.50

0 Elsevier

Sequoia/Printed

in The Netherlands

128

evaluated wear rates for sliding either against smooth steel or under severe abrasion. This paper describes the ~vest~ation of the effect of fiber orientation on the dry abrasive wear behavior of composite materials consisting of unidirectional continuous fibers and polymer matrices. Wear rates as well as microscopic wear mechanisms were studied for several fiber orientations. A correlation of the microscopic observations of the wear mechanisms to the wear rate dist~butions is attempted in a qualitative manner. Lastly, empirical re~tionships of the wear data are used to predict the wear rate at arbitrary fiber orientations.

2. Experimental details

Two unidirectional, continuous fiber polymer composite materials were investigated in this study, namely, carbon-fiber-reinforced polyetheretherketone (CF-PK) (material supplied in pre-impregnated tape form (APCZ) by ICI Americas) and aramid-fiber-reinforced epoxy (AF-EP) (material supplied in pre-~preg~~ tape form by the Fiberite Coloration). Each material system was cured by the process procedures recommended by the material suppliers. Table 1 describes these materials in more detail. TABLE 1 Composite materials employed in the wear tests Abbreviation

Material

Fiber volume fraction

Density (g cm-3)

EP AF-EP PK CF-PK

3501-6 epoxy resin K49 aramid fiber-epoxy Polyetheretherketone (PEEK) AS4 carbon fiber-PEEK

-

1.28 1.34 1.27 1.56

60% 55%

All feasible fiber orientations with respect to the sliding direction can be described by two rotational parameters CYand 6. Figure 1 illustrates a fiber located at an orientation defined as (a,@). The QIrotations are defined as counterclockwise rotations about the Z-axis, and the j3 rotations are defined as counterclockwise rotations about the 3-axis. At each position, a local coordinate system can be introduced where the l-axis (also l’- and l”-axis) coincides with the longitudinal axis of the fibers. This will allow a t~~d~en~on~ description of the wear rates as a function of fiber orientations relative to the sliding direction. To facilitate the achievement of all possible fiber orientations, the (x rotations must range from -90” to +90”, and the ~3rotations must range from 0 to 90”.

129

Fig. 1. Illustration of the Q and p rotational parameters used for fiber orientation cription.

Normal lb

I p)=w)l

Parallel

I(a ,p )=WO)l

Fig. 2. The three basic fiber orientations fiber-reinforced composite materials.

des-

Antiparallel

I(a . P)=wo)1 (N, P and AP) for unidirectional continuous-

Owing to the nature of unidirectional composite materials, three basic fiber orientations are easily identified with respect to the sliding surface as depicted in Fig. 2. This terminology is commonly used in the tribobgical study of composite materials. When the fibers are perpendicular to the sliding surface plane the normal (N) orientation ((a$) = (0,O)) exists. This orientation is further categorized as an out-of-plane orientation where the plane refers to the sliding surface plane. In the same context, the other two orientations are considered as m-plane orientations. More specifically, the parallel (P) orientation ((o$) = (90,O)) exists when the fibers are in-plane and parallel to the sliding direction, and the antiparallel (AP) orientation ((a$) = (0,90)) exists when the fibers are in-plane and perpendicular to the sliding direction.

Fig. 3. Simplified

illustration

of the pin-on-flat

test apparatus.

2.2. Test procedures A pin-on-flat apparatus was employed which is illu~ra~d schematically in Fig. 3. A more detailed description of the apparatus is given in ref. 13. Specimens of cubic geometry were used for all tests whose apparent contact area A was approximately 10 mm 2. A 220 grit non-waterproof type silicon carbide (Sic) abrasive paper was used as the counterface material. A detailed description of waterproof and non-waterproof type abrasive papers is given in ref. 13. According to the man~act~er, the average Sic particle diameter D was 70 flm. The materials were tested in the three basic orientations (N, P and AP) as shown in Fig. 2. Further tests were then done on fiber orientations between the N and P orientations, between the N and AP orientations, and between the P and AP orientations. These intermediate orientations are schematically illustrated in Fig. 4. Generally, the specimens were moved in contact with the Sic paper at a constant velocity v of 300 mm mm-’ and at an apparent pressure p of

t-OO”.O”,

(.6O”P1

cl Fig. 4. Illustration and AP orientations.

1.3W.O’)

iO’,O”)

(JO’,O’)

(60’,0’1

(w-,0-1

0 of the fibers on the wear surfaces

cl for orientations

between

the N. P

131

2.2 MPa for a sliding distance L of 500 mm. Four runs in the same direction were made for each specimen. Each run was on a fresh track of Sic paper (single-pass conditions). The mass loss of the specimen was observed after each run on a Mettler analytical balance (0.1 mg) under the same environmental conditions as the test was conducted. All tests were performed in controlled laboratory air which was at about 22 “c and 35% relative humidity. The wear behavior was characterized by defining a dimensionless wear rate W which relates volume loss AV to sliding distance L and apparent contact area A in the following manner:

The volume loss can be found by using the mass loss Am and density p measurements in the following form: AV=!! P Inserting eqn. (1) into eqn. (2) yields a dimensionless wear rate which is defined by experimentally measured test parameters: Am W= PLA

(3)

Dividing eqn. (3) by the apparent normal pressure p yields the specific wear rate W, as w, =

w P

(4)

which is given the units of millimetres cubed per Newton per metre. This equation describes the physical nature of the specific wear rate as a volume loss of material by a given amount of energy input. Often, the wear resistance of a material is referred to which is simply the inverse of the wear rate (W-i or W,‘). Microscopic observations of the tested specimens were made by scanning electron microscopy (SEM). All sample surfaces which were microscopically observed under the scanning electron microscope were coated with a thin layer of gold following standard procedures for polymeric materials.

3. Results 3.1. Wear rate as a function of fiber orientation The average dimensionless wear rate distribution of the CF-PK system for various fiber orientations is illustrated graphically in Fig. 5. As the fibers proceeded from the P to AP and the N to AP orientations a relatively monotonic increase in the wear rate was observed. The wear rate distribution

132

Fig. 5. The averaged dimensionless rial for various fiber orientations.

wear rate distribution

of the CF-PK

composite

mate-

Fig. 6. The averaged dimensionless rial for various fiber orientations.

wear rate distribution

of the AF-EP

composite

mate-

behaved in a non-symmetric fashion about the N orientation where a local maximum occurred at (30,O) and (-60,O). Furthermore, the absolute minimum occurred at the N orientation for the CF-PK system. The wear rate is a function of material type as well asfiber orientation, as illustrated by the drastically different wear rate distribution of the AF-EP system shown in Fig. 6. As the fiber orientation ranged from P to AP a relatively monotonic decrease in the wear rate was observed. Unlike for the CF-PK system, the distribution of the wear rates behaved in a symmetric fashion about the N orientation. The absolute minimum occurred also at the N orientation. Furthermore, the distribution was relatively flat about the N orientation illustrating the fact that the wear rate does not significantly increase as the fibers proceed from the out-of-plane orientations towards the in-plane orientations until they are predominantly in the in-plane orientation. 3.2. Microscopy of wear mechanisms Basic wear mechanisms for unidirectional, continuous fiber-reinforced polymers in an abrasive tribological system have been identified and discussed in detail for the three basic fiber orientations (N, P and AP) by Cirino et al. [lo]. These are illustrated schematically in Figs. 7 - 9. An SEM photograph of the worn surface of a (90,30) CF-PK specimen is shown in Fig. 10. The fracture path left by a passing abrasive particle can be seen along with

Damage

Zonef

n

Fig. 7. Schematic illustration of the basic microwear mechanisms occurring in the N orientation: A, fiber slicing; B, fiber-matrix debonding; C, fiber cracking; D, fiber bending. Fig. 8. Schematic illustration of the basic microwear mechanisms occurring in the P orientation: A, interlaminar crack propagation; B, fiber cracking; C, fiber-matrix debonding; D, fiber fracturing.

Fig. 9. Schematic illustration of the basic microwear mechanisms occurring in the AP orientation: A, fiber fracturing; B, fiber-matrix debonding. Fig. 10. SEM photograph of the worn surface of the CF-PK in the (90,30) orientation. The scar (S) left by the passing abrasive particle induces damage such as fiber fracture (F), fiber-matrix debonding (D) and matrix removal (M).

the damage it has caused, e.g. fiber fracture, fiber-matrix debonding and matrix removal. The in-plane orientations are characteristic of relatively larger cut fiber pieces, whereas the out-of-plane orientations show evidence of fiber slicing which produced relatively smaller cut fiber pieces and cracks perpendicular to the surface (Figs. 11 and 12). The wear mechanisms which occurred in the AF-EP system are different from those in the CF-PK system. The AF-EP system behaved in a relatively ductile manner resulting in a relatively smoother surface appearance (Figs. 13 - 15). The in-plane orientations consisted of a high degree of fiber deformation and fibrillation as shown in Fig. 13. In this case, the fibers were cut and subsequently fibrillated by the passing abrasive particle. These damaged fibers have a tendency to remain attached to the surface and required more abrasive particle interactions before they were removed. In the out-ofplane orientations, the fiber ends show a large plastic deformation (Figs. 14 and 15). The ability of the aramid fiber to deform plastically is shown in

Fig. 11. SEM photograph S, fiber slicing. Fig. 12. SEM photograph

of the worn

surface

of the worn surface

of the CF-PK

of the CF-PK

Fig. 13. SEM photograph of the worn surface showing a high degree of fiber fibrillation. Fig. 14. SEM photograph of the worn showing highly deformed fiber ends.

surface

in the (-60,O)

in the (0,30)

of the AF-EP

of the AF-EP

orientation:

orientation.

in the (90,30)

orientation

in the (0,60)

orientation

Fig. 16. The passing abrasive particles have deformed the fiber such that on the tensile side of the fiber, a fiber fibrillation phenomenon has occurred, while on the compressive side of the fiber, kink bands are present.

4. Discussion 4.1. Differences in wear rate as a function of fiber orientation The CF-PK system was observed to exhibit higher wear rates for the --cy fiber orientations than for the +o fiber orientations. Figures 17 and 18 illustrate schematically the basic fiber-abrasive particle interactions which occurred in each case respectively. For +CXorientations, the passing abrasive particles caused the fibers to experience compression and bending. For --(II orientations, however, the fibers experienced tension and bending. The tension resulted in the damaged fibers experiencing a pull-out phenomenon. Furthermore, in this case the tip of the abrasive particle is able to indent the

Fig. 15. SEM photograph of the worn surface of the AF-EP in the (30,O) orientation showing highly deformed fiber ends. Fig. 16. SEM photograph of the worn surface of the AF-EP in the (90,30) The bent aramid fiber consists of fiber fibrillation (F) and kink bands (K).

orientation.

Si Part

Ma&x

Fibbr At,

q

Aeos

a

Makx

Fi:ber c

=-

h cos a

Fig. 17. Schematic illustration of the interactions between a fiber and a Sic particle for the +(Y fiber orientations. The fiber experiences bending and compression. An and A, are components of the bending force A imposed on the fiber by the passing Sic particle. Fig. 18. Schematic illustration of the interactions between a fiber and a Sic particle for the --(Y fiber orientations. The fiber experiences bending, tension and direct particle indentation at point D. The maximum fiber length at which direct particle indentation occurs is shown as I,.

fiber directly, thus enhancing the cracking and fracturing of the fiber. These factors are responsible for the relatively higher wear rates in the --(II orientations. More specifically, when the fibers were oriented in the +(Y direction, the (30,O) orientation case showed a slightly higher wear rate than the (60,O) orientation case. Figure 17 shows schematically that the force A applied to the fiber by the passing abrasive particle can be split into two components. The first component is directed parallel to the longitudinal axis of the fiber (A,) which places the fiber in a state of compression. The second component is directed perpendicular to the fiber (A,,) causing the fiber to bend and its magnitude is given by Ab =A COSQ

(5)

136

For the (30,O) orientation A,(30,0)

= 0.87A

(6)

and for the (60,O) orientation A,(60,0)

= 0.54

(7)

> A,(60,0)

(8)

Therefore -4,(30,0)

Theoretically, greater fiber damage is therefore expected in the (30,O) orientation case than in the (60,O) orientation case. The damaged fibers can then be subsequently removed resulting in the relatively higher wear rate for the (30,O) orientation case. When the fibers were oriented in the --(Y direction, the (-60,O) orientation showed a higher wear rate than the (-30,O) case. Assuming that the Sic abrasive particle penetrates to a constant depth h for each case (i.e. h is only dependent on the apparent contact pressure), it is shown schematically in Fig. 18 that the length of the fiber (I,) between the point at which the Sic particle contacts the fiber and the fiber end at the surface is given by h 1, = cos a

(9)

Hence, for the (-60,O) Z,(-60,O)

= 2h

and for the (-30,O) 1,(-30,O)

orientation (10)

orientation

= 1.15h

(11)

> I,(-30,O)

(12)

Therefore 1,(-60,O)

Thus, in theory, the passing abrasive particles can damage the fibers to a greater extent in the (-60,O) orientation case than in the (-30,O) orientation case. The damaged fibers can then be subsequently removed resulting in a higher wear rate for the (-60,O) orientation case. The AF-EP system exhibited relatively lower wear rates when the fibers were near the out-of-plane orientation. As in the N orientation case, the abrasive particles can only deform the fiber ends without causing any fiber fracture or debonding. When the fibers are in-plane, material removal is increased owing to the ability of the abrasive particles to bend the fibers and remove them from the surface. This is enhanced by the poor quality of the aramid fiber/polymer matrix interface. These mechanisms are not significantly effective until cy reaches values which are very close to that of the P and AP orientation; otherwise, the fibers are too far out of the sliding surface plane. Hence this is the reason for the flat area at the minimum about the N orientation. Even though each material system exhibited different

137

wear behavior, there does exist one point in common; the optimal wear resistance was exhibited in the N orientation for each material system. 4.2. Empirical wear equation Given the wear data which was ascertained for the various fiber orientations under consideration, an empirical relationship was determined for predicting the wear rate at any arbitrary fiber orientation. Since wear depends strongly on the tribological system under which it operates, the predicted wear rates are only valid for that tribological system. The idea for the development of such a relationship was inspired from studying the work of Hornbogen [ 141, who suggested that wear rates of anisotropic materials can be described, in principle, by a “wear tensor” if the wear rates in the basic directions of the material are known. For each material system, a linear Coons surface [15] was constructed from the given data. The surfaces are represented by a bivariate vector-valued function

Q(w)

= (Qol(~,~),Q~(u,~),Qw(u,u))

(13)

where the components of vector Q are illustrated schematically in Fig. 19. Initially, it is necessary to define the four boundary curves of the surface. These planar curves can be represented by a single-parameter vector-valued function which may be obtained by fixing either u or u of Q(u,u) equal to 0 and 1. The necessary data are shown in Fig. 20 as projected onto the o-0 plane. A third-order polynomial fit was performed on each boundary curve.

Fig. 19. Illustration of the Qcu, Qp and Qw component vector Q(u,u).

vectors of the bivariate surface

Two surface patches were generated for each material system; one surface (Q+) for the +o quadrant, and one surface (Q-) for the --(Y quadrant. For the CF-PK system, the surfaces for (o$,W) space are given as Q,y+(~,u)=~;

O