Wear, 107 (1986) 1 - 12

EROSIVE WEAR OF COMPOSITE MATERIALS

K. V. POOL, C. K. H. DHARAN and I. FINNIE

Department of Mechanical Engineering, University of California, Berkeley, CA 94720
(U.S.A.)
(Received April 22, 1985; accepted July 16, 1985)

summary
The erosive wear behavior of selected polymer matrix composite mate-
rials was investigated using an erosion wear tester. Scanning electron micros-
copy was used to characterize the eroded surface. The results show that the
erosive wear rates in these materials are at least an order of magnitude
greater than that of low carbon steel. Of the composites tested, continuous
graphite fiberepoxy composites showed erosive wear that is typical of
brittle materials (maximum wear rate when the impingement is normal to
the surface), while continuous aramid fiber-epoxy and chopped graphite
fiber-polyphenylene sulfide showed quasi-ductile behavior (maximum wear
rate at 25” - 45” impingement angle). These results are discussed in terms of
the observed failure modes.

1. Introduction

The wear behavior of composite materials has received much less atten-
tion than that of conventional materials. However, as composites are utilized
to an increasing extent in the aerospace, transportation and process in-
dustries, their durability may become a prime consideration. In the present
investigation the wear behavior of certain composites is contrasted with that
of metals for erosive wear.
In erosion, material is removed by an impinging stream of solid parti-
cles. Studies to develop an understanding of the mechanisms of erosive wear
have been motivated by reduced lifetimes and failures of mechanical compo-
nents used in erosive environments, e.g. in pipelines carrying sand slurries, in
petroleum refining [l, 21 and in aircraft gas turbine compressor blades
[3,4]. In addition to these studies, which were conducted to understand
erosion behavior in isotropic materials, there is increasing interest in under-
standing the erosion behavior of anisotropic materials. Because of their very
high specific stiffness and strength, composites are now used extensively in
aircraft structures. The understanding of erosive wear behavior is obviously
important for such structures, e.g. helicopter rotor blades. While polymeric

0043-1648/86/$3.50 0 Elsevier Sequoia/Printed in The Netherlands

2

TABLE 1
General factors influencing erosion

Effect of erosion
_ __

Eroded surface properties

Hardness Ductile: E is inversely proportional to the
Vickers hardness
Brittle: very little correlation
Stress level Ductile: little effect on E
Brittle: more effect on E
Surface finish Rougher surfaces raise E (this is a tran-
sient initial effect)

Eroding particle properties

Size Ductile: no effect for particle diameters
2 100 pm; lower E for particle diam-
eters 5 100 /Jrn
Brittle: ductile behavior for particle
diameters > 10 /_tm
Shape Angular particles produce more wear
Hardness Harder particles produce more wear (they
also tend to be more angular)

Flow and environmental conditions

Angle of impingement Ductile: maximum erosion at about 20”
Brittle: maximum erosion at about 90”
Particle velocity Ductile: E 0: U02q3
Brittle: E 0: U03-s
Particle flux (mass per time) Generally small effect on E
Temperature Less effect than predicted from corre-
sponding change in hardness (for tem-
peratures less than half the melting
point in kelvins)

E is the mass removed divided by the mass of particles.

coatings have been developed to protect composite aircraft structures from

rain erosion [ 5,6 J, there is little understanding of the mechanisms of erosive
wear in these materials. In this paper the results of a preliminary study
conducted to determine the effects of fiber orientation, lay-up, matrix
properties and fiber properties on the erosion behavior of selected composite
materials are reported.
For ductile metals and brittle solids, Table 1 lists the factors influencing
the erosion of these (isotropic) materials. By contrasting these factors with
the erosive behavior of composite materials, it is possible to draw some
generalizations. This is one of the objectives of the present investigation.
For polymers and composite materials, Tilly and Sage [ 71 investigated
the influence of velocity, impact angle, particle size and weight of impacted
abrasive for nylon, carbon-fiber-reinforced nylon, epoxy resin, poly-
propylene and glass-fiber-reinforced plastic. Their results showed that, for
 3

the particular materials and conditions of their tests, composite materials

generally behaved in an ideally brittle fashion (i.e. maximum erosion rate
occurred at normal impact). Fiber reinforcement may improve or worsen the
resistance to erosion, depending on the type of fibers used. In addition, the
erosion rates in composites continued to increase with particle size, in
contrast with the independence of erosion rate on particle size found in steel
with particle diameters greater than about 100 I.trn [7, 81.
Zahavi and Schmitt [9] performed erosion tests on a quartz-polyimide
composite and a quartz-polybutadiene composite and again determined
their behavior to be like that of nearly ideally brittle materials. One
interesting result was the behavior of an E-glass-reinforced epoxy composite
which exhibited erosion rates that were less than those of the other com-
posites by a factor of 5. This was attributed to better adhesion between the
matrix and the fibers and the lower porosity of this composite in comparison
with the others studied. The E glass-epoxy composite exhibited semiductile
erosion behavior with a maximum weight loss at an impingement of 45” -
60”, while the others eroded in a brittle manner with the maximum weight
loss occurring at 75” - 90”. It should be noted here that the volume erosion
rates for all the composite tests found in the literature are greater than that
of low carbon (AISI 1020) steel by at least an order of magnitude.
The erosion process in fiber composite materials, as described by
Zahavi and Schmitt, can be characterized as follows.
(1) There is local removal of resin material from the impacted surface
which results in exposure of the fibers to the erosive environment.
(2) Sand particles impact on the fibers and cause fibers to break be-
cause of the formation of cracks perpendicular to their length. These cracks
are presumably caused by fiber-bending stresses due to the impact of parti-
cles on the unsupported fibers.
(3) Further damage results when the interfaces between the broken
fibers and the matrix resin are degraded until the fibers are removed by
subsequent impacts.
The approach taken in this study is to determine whether the above
proposed mechanisms for erosive wear in fiber composites can be verified
by scanning electron microscopy (SEM) studies of advanced erosion in uni-
directional and woven graphite fiberepoxy, in woven aramid-epoxy and in
chopped graphite fiber-polyphenylene sulfide (PPS).

2. Experimental procedure

A schematic diagram of the erosion tester used is shown in Fig. 1. The

abrasive particles are fed into an air stream at room temperature by a
vibrating hopper. The frequency of vibration and the inclination of the
hopper determine the feed rate. The rate of flow of the air, which deter-
mines the velocity of the particles, is easily controlled by adjusting the
nozzle inlet pressure. The abrasive stream is accelerated by air flowing
 CARRIER GAS

PARTICLE AND GAS

MIXING CHAMBER

12 N LENGTH

-SPECIMEN

2 Y2.x 3/dx 3/l $

w
I I

EXHAUST

Fig. 1. Schematic diagram of the erosion tester.

through a nozzle 0.3 m (12 in) long and of 5 mm (0.19 in) internal diameter
and impinges on a sample which may be rotated to give any desired angle
of impingement as shown in the figure. Velocity calibration is accomplished
by using an apparatus which consists of two parallel plates rotating at a
constant’ angular velocity [lo]. The plane of the plates is perpendicular to
the direction of particle flow. Some particles pass through a radial slit in the
top plate and impinge on the bottom plate. By knowing the separation
distance and the angular velocity, and by measuring the angular displacement
of the erosion mark from the radial slit, one can determine the particle
velocity. In this work a velocity of 31 m s-l was used. The velocity deter-
mination was estimated to be accurate to within 10%. Since 30 g of abrasive
were used to calibrate the velocities, and since varying the loading rate does
seem to have some effect on the particle velocity, every sample was eroded
with 30 g of sand at a time.
Sand was selected as the abrasive in this study for several reasons. It is
the most common form of abrasive to which materials are exposed in
engineering applications and it is used in the solid particle erosion studies on
composites reported in the literature. The sand used in this study is com-
mercially sized silica sand of 155 pm diameter (120 mesh). The size distribu-
tions before and after erosion are shown in Table 2.
 5

TABLE 2
Size distribution of sand used in erosion tests

As received
Less than 250 mesh 2.3%
Between 250 and 200 mesh 5.1%
Between 200 and 170 mesh 10.0%
Between 170 and 140 mesh 10.0%
Between 140 and 100 mesh 62.2%
Between 100 and 80 mesh 7.4%
Between 80 and 65 mesh 1.9%
Greater than 65 mesh 1.1%
After erosion
Less than 200 mesh 7.1%
Between 200 and 170 mesh 4.7%
Between 170 and 140 mesh 38.3%
Between 140 and 120 mesh 31.0%
Greater than 120 mesh 18.9%

TABLE 3
Mechanical properties of fibers and matrix materials

Property Fibers Matrices

Thomel300 Kevlar 49 PPS EPOXY Polyimide

Density (g cm”) 1.75 - 1.95 1.45 1.34 1.1 - 1.4 1.59

Fiber diameter (pm) 7.0 - 9.7 11.9 - - -
Tensile modulus (GPa) 345 124 3.1 2.5 3.2
Tensile strength (MPa) 2500 2800 70 80 70
Elongation to 0.5 2.2 - 2.8 3-4 1-3 5
breakage (%)

Four composite materials were tested for erosive wear: (1) a unidirec-
tional continuous-graphite-fiber-reinforced polyimide laminate (Union
Carbide’s Thomel-300 graphite fiber in Hexcel’s F-173 bismaleimide
polyimide resin), (2) a woven (O/90) graphite-fiber-reinforced epoxy lami-
nate (Thornel 300 graphite fiber in Fiberite’s 934 epoxy resin), (3) a woven
aramid-fiber-reinforced epoxy laminate in a quasi-isotropic (O/90/*45) sym-
metric lay-up (Kevlar fiber in Fiberite’s 934 epoxy resin) and (4) a chopped-
graphite-fiber-reinforced thermoplastic (LNP Corporation’s 40% graphite
fiber in PPS, injection molded).
The properties of these materials are listed in Table 3. After erosion
testing, the surfaces of the materials were examined under a scanning elec-
tron microscope to characterize the surface morphology and to establish
possible mechanisms for material removal. A thin solid film (200 A) was
sputter coated onto the surfaces to prevent charge build-up in the scanning
electron microscope.
3. Results and discussion

Figure 2 is a plot showing the erosion rates of all the materials tested as
a function of the angle of impingement (for U, = 31 m s-l). This plot was
obtained by determining the steady state slope of the weight loss uersus mass
of sand impinged for each material and angle of impingement tested. As can
be seen from this figure, the erosion rates of the composites tested are much
higher than that of steel. Also, since the density of steel is approximately
five times greater than that of the composites, the relative volume erosion
resistance of the steel is even greater.

0 30 60 90
Angie of Impingement (0~~)
Fig. 2. Erosion rate us. angle of impingement for various composite materials tested
(impingement velocity, 31 m s-l).

The aramid (Keviar) laminate and the chopped graphite fiber-PPS

composite showed the least erosion rates, while the perpendicularly oriented
Thornel-300 graphite-polyimide showed the greatest erosion rates.
Figures 3 - 10 are SEM micrographs of surface features in the materials
tested. Figures 3 and 4 are of material 1, with fibers oriented at 90” to the
direction of the airstream with the particles impinging at an angle of 30” to
the surface, and show both the initial stages of resin removal that expose the,
fibers to breakage and the latter stages of fiber breakage and subsequent
fiber removal. Figure 5 is of a representative portion of the eroded surface
fibers parallel to the stream impinging at 30”, showing some longer unsup-
ported fibers than exist in the eroded surface of the sample shown in Fig. 6
 7

Fig. 3. Unidirectional continuous-graphite-fiber-reinforced polyimide eroded at 30”

impingement angle to the surface and perpendicular to the fibs ers.

Fig. 4. As in Fig. 3, showing the later stages of fiber breakage.

(impingement normal to the surface). This suggests t;hat some of the stre fsses
due to the impacting particles are taken up in the fiber axial direction . In
Fig. 7 one can see that the intersection of the crc bss-weaves in the (0, ‘90)
graphite-epoxy fabric seems to be a preferential spot for erosion (the ho1.e in
the upper right-hand corner should be noted).
Fig. 5. Unidirectional graphite-fiber-reinforced polyimide (Thornel 300-Hexcel F-173
bismaleimide) eroded at 30’ impingement angle to the surface and paruZZe2to the fibers.

Fig. 6. Unidirectional graphite-fiber-reinforced polyimide eroded normal to the surface.

Figure 8 (chopped graphite fiberPPS, eroded at an impingement angle

of 20”) shows fairly uniform erosion as one would expect in an isotropic
material. Figures 9 and 10 show the eroded surface of the aramid-epoxy
 9

Fig, 7. Woven graphite-fiber-reinforced epoxy (woven Thornel 300-Fiberite 934) eroded

at 50” impingement angle to the surface and parallel and perpendicular to the woven
fibers.

Fig. 8. Chopped graphite-PPS eroded at 20” impingement angle to the surface.

Fig. 9. Woven aramid-fiber-reinforced epoxy (woven Kevlar 49~-Fiberite 934) eroded at
40” impingement angle to the surface.
Fig. 10. As in Fig. 9, at greater magnification showing the extensive fibrillation undergone
by the aramid fibers.

(Kevlar) laminate eroded at an impingement angle of 40”. They clearly show

the large degree of fibrillation which occurs in the Kevlar fibers before they
break.
The curves in Fig. 2 show a strong dependence of the erosion behavior
on the fiber orientation in composites. The difference in the curves for
erosion perpendicular and parallel to the fiber direction in the unidirectional
graphite-polyimide composite is substantial. This can be explained by
reference to Figs. 3 - 5. In Figs. 3 and 4 the fibers are always perpendicular
to the direction of impacting sand, while in Fig. 5 the fibers are aligned
nearly parallel to the direction of the jet. Consequently, any bending stresses
present will be of greater magnitude in the perpendicular case than in the
parallel case. This explanation is further backed by noting the longer unsup-
ported fibers that exist in the specimen (Fig. 5).
The larger reductions in erosion rates are found for quasi-isotropic
composites (e.g. the Thornel-300 (O/90) fabric, the Kevlar [(O/90)(+45/
-45)], and the chopped fiber composite). This can be explained by refer-
ence to the steps in the erosion process outlined previously. In unidirectional
composites, as the resin is removed essentially nothing remains to support
the exposed fibers. In the cross-ply fiber composites, however, local fiber-
bending stresses are resisted by the underlying cross-fibers which are still
held firmly in place by the as-yet undamaged resin material surrounding
 11

them. This implies that the thickness of the composite should have an effect
on the erosion rate, up to a critical minimum thickness below which the
surface fibers are not supported.
In the chopped fiber composite (Fig. 8) several factors contribute to
the comparatively low erosion rate. In a randomly oriented short fiber com-
posite a reasonable proportion of the fibers will be oriented such that they
are nearly aligned with the direction of the impinging particles. The fibers
that are not favorably oriented will still derive support from the underlying
fibers. In addition, PPS is a thermoplastic. At the high temperatures known
to occur in solid particle erosion [ 31, softening of the matrix could occur,
thereby accounting for this material’s ductile behavior. Tilly [7,8] found
similar ductile behavior in both glass- and graphite-reinforced nylon.
The aramid-epoxy (material 3, woven aramid-epoxy) also showed
essentially ductile behavior as seen in Fig. 2. The generally low erosion rate
may be due partly to the quasi-isotropic lay-up; however, the exhibited
ductile behavior cannot be attributed to thermally induced softening of the
epoxy matrix since epoxy is thermosetting. At temperatures above the glass
transition temperature (about 210 “C for the Fiberite 934 resin matrix)
epoxy loses its strength but remains brittle, i.e. it does not flow. Thus,
erosion should continue to occur in a brittle manner.
A plausible reason for the ductile-type erosive behavior of aramid-
epoxy lies in the behavior of the aramid fiber which fibrillates during failure,
thereby absorbing significantly more energy than brittle fibers such as
graphite and glass. Similar semiductile erosion behavior has been observed in
an E glass-epoxy composite and has been attributed to good adhesion
between the matrix and the fiber [ 61.

4. Conclusions

(1) The volume erosion rates in the composites tested are greater than
those of steel by at least an order of magnitude.
(2) The highest erosion rates (32 X 10e5 g g-‘) were obtained for the
unidirectional graphite fiber composites at 90” impingement compared with
erosion rates of 5 X 10e5 g gg’ for the woven aramid and the chopped
graphite-PPS composites.
(3) In comparing woven graphite and woven aramid reinforcements in
the same epoxy matrix, the aramid composite exhibited a maximum erosion
rate of 5 X lo-’ g g-l compared with the woven graphite composite’s
maximum erosion rate of 12 X lop5 g g-l.
(4) The maximum erosion rate occurred at an angle of impingement of
90” for the unidirectional and woven graphite composites signifying brittle-
type erosion behavior. The woven aramid composites exhibited a maxi-
mum erosion rate between 34” - 45” indicating semi ductile behavior.
Interestingly, the injection-molded short fiber graphite-PPS material showed
a maximum erosion rate at 25”, also indicating ductile behavior.
12

(5) From this preliminary study it appears that the factors governing
erosion rates in composites are influenced by (a) the brittleness of the fibers,
(b) whether the matrix is thermosetting or thermoplastic and (c) the inter-
facial bond strength between the fibers and the matrix. Thus well-bonded
ductile fibers in a thermoplastic matrix should exhibit the lowest erosion
rates. The determination of the relative importance of fiber ductility and
interfacial bond strength on erosion rates requires further study.

Acknowledgment

This work was supported by the Director, Office of Basic Energy

Research, Office of Basic Energy Sciences, Materials Sciences Division of the
U.S. Department of Energy, under Contract W-7405-ENG-48.

References

1 I. Finnie, An experimental study of erosion, Proc. Sot. Exp. Stress Anal., 17 (1960)
65 - 70.
2 I. Finnie, Erosion of surface by solid particles, Wear, 3 (1960) 87 - 103.
3 C. E. Smeltzer, M. E. Gulden and W. A. Compton, Mechanisms of material removal
by impacting dust particles, J. Basic Eng., 92 (1970) 639 - 654.
4 G. P. Tilly, Erosion caused by airborne particles, Wear, 14 (1969) 63 - 79.
5 G. P. Schmitt, The erosion behavior of polymeric coatings and composites at subsonic
velocities. In A. A. Fyall and R. B. King (eds.), Proc. 3rd Int. Conf. on Rain Erosion
and Associated Phenomena, Royal Aircraft Establishment, Farnborough, 1970,
pp. 107 - 128.
6 J. Zahavi and G. F. Schmitt, Jr., Solid particle erosion of polymeric coatings, Wear,
71 (1981) 191- 210.
7 G. P. Tilly and W. Sage, The interaction of particle and material behaviour in erosion
processes, Wear, 16 (1970) 447 - 465.
8 G. P. Tilly, Sand erosion of metals and plastics: a brief review, Wear, I4 (1969) 241 -
248.
9 J. Zahavi and G. F. Schmitt, Jr., Solid particle erosion of reinforced composite mate-
rials, Wear, 71 (1981) 179 - 190.
10 A. W. Ruff and L. K. Ives, Measurement of solid particle velocity in erosive wear,
Wear, 35 (1975) 195 - 199.