Scratch Testing of Metals and Polymers

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Wear 266 (2009) 76–83

Contents lists available at ScienceDirect

Wear
journal homepage: www.elsevier.com/locate/wear

Scratch testing of metals and polymers: Experiments and numerics


Fredrik Wredenberg, Per-Lennart Larsson ∗
KTH Solid Mechanics, Royal Institute of Technology, S-10044 Stockholm, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: An experimental and numerical study of the scratch test performed on metals and polymers was con-
Received 28 September 2007 ducted. The materials tested, being both metallic and polymeric, were related to the well known Johnson’s
Received in revised form 28 April 2008 parameter, often used to correlate indentation experiments. The aim was to determine whether it was
Accepted 27 May 2008
possible to use the numerical approach presented by Wredenberg and Larsson [F. Wredenberg, P.-L.
Available online 21 July 2008
Larsson, On the numerics and correlation of scratch testing, Journal of Mechanics of Materials and Struc-
tures 2 (2006) 573–594] to describe the scratch mechanism and of course also to investigate whether
Keywords:
or not important scratch quantities can be determined with sufficient accuracy from standard scratch
Scratch test experiments
Finite element analysis
experiments.
Contact © 2008 Elsevier B.V. All rights reserved.
Metals
Polymers

1. Introduction been gained over the years regarding the mechanical behaviour
at indentation. Sharp indenters are most often used, for practical
The understanding of the mechanical behaviour of scratch- reasons, at least when modern experimental devices such as the
ing is not as developed as that for indentation testing, although nanoindenter (or other types of instrumented indentation devices)
early mechanical analyses concerning different aspects of scratch- are at issue. For such indenters semi-empirical relations for mate-
ing exist and are presented by, for example, Goddard and Wilman rial characterisation were derived and used already in the late 1940s
[2], Childs [3], Vathaire et al. [4] and Gilormini and Felder [5]. and early 1950s, in particular for metals and alloys, cf., e.g. Tabor [7].
It is possible to distinguish between two main different types of In short, from comprehensive experimental investigations Tabor [7]
mechanism at scratching. The first one, which here is termed mild derived a relation
scratching, is when the behaviour does not involve or is not sig-
H = Crepr , (1)
nificantly influenced by any interface properties. If the material is
sufficiently tough the scratch deformation will be governed by the between the indentation hardness H, here defined as the mean con-
bulk constitutive properties, typically plastic for metallic materi- tact pressure at indentation and scratching, and the material yield
als or non-linearly viscoelastic for polymers. In this context, tough stress repr at a representative value on the accumulated (effective)
meaning without chipping, spalling or delaminating. The second plastic strain, repr . Furthermore in Eq. (1), C is a constant that only
type, termed s evere scratching, occurs if the material toughness is depends on the geometry of the indenter. For a Vickers indenter
sufficiently low and the formation of cracks takes place. Alterna- Tabor [7] determines the values C ≈ 3 and repr ≈ 8% while Atkins
tively, when scratching of coatings is at issue, delamination occurs. and Tabor [8] find C ≈ 2.54 and repr ≈ 11% for a conical indenter
It appears likely, at least if no cracks develop, that the scratch pro- with an angle of ˇ = 22◦ (similar to the Vickers indenter) between
cess is very similar to conventional indentation testing and that the indenter and the undeformed surface (see Fig. 1). These results
thus the scratch process can be well described by conventional for sharp indenters have lately been validated and further improved
material properties. In fact, scratch tests have been used to measure (and also extended to other types of constitutive behaviour) using
elastic and plastic properties [6]. finite element simulations, cf., e.g. [9–12].
Indentation testing and scratch testing show many similar fea- Based partly on the above discussed results, further progress is
tures. When it comes to indentation a great deal of knowledge has achieved by Johnson [13,14], who shows from theoretical consid-
erations that indentation testing on different materials can be well
correlated by using a parameter,

∗ Corresponding author. Tel.: +46 8 790 60 00; fax: +46 8 411 24 18.
E tan ˇ
= , (2)
E-mail address: pelle@hallf.kth.se (P.-L. Larsson). (1 − 2 )repr

0043-1648/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2008.05.014
F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83 77

as indentation parameters. Also, Bucaille et al. [16] find, as could be


expected, that the strain levels at scratching are much higher than
at indentation but this finding is not, as only perfectly plastic mate-
rial behaviour is at issue, explored in connection with the concept of
representative strains. Recently, Wredenberg and Larsson [1] pre-
sented a numerical approach for modelling the scratch test (from
a mechanical point of view) using the finite element method. For
one thing, these authors find that there exists a representative level
of plastic strain of approximately 35% at frictionless scratching.
Furthermore, as the scratch test is often used to determine charac-
teristics of polymeric coatings this issue is also under investigation,
Fig. 1. Cone scratching. cf., e.g. Briscoe et al. [17] and Gauthier et al. [18]. Due to the com-
plexity of the boundary value problem at scratching, high accuracy
of results can only be obtained if numerical methods, preferably the
where E is Young’s modulus and  is Poisson’s ratio. As regards finite element method (FEM), is relied upon. Recently quite a few
details Johnson [13,14] also suggests that indentation properties for such analyses have been presented. Further progress based on FEM
various materials will fall into one out of three levels as indicated in analyses of the scratch test is achieved also by Bucaille et al. [19] for
Fig. 2 In short, these levels can be characterised as follows: in level I, polymeric materials modelled by using standard elastoplasticity,
 < 3, very little plastic deformation occurs during the indentation corresponding to stage II rheology (see Fig. 2). Further, Felder and
test and all global properties can be derived from an elastic analysis. Bucaille [20] investigate the scratch behaviour of a work hardening
In level II, 3 <  < 30, an increasing amount of plastic deformation polycarbonate.
is present and both the elastic and the plastic properties of the When performing scratch experiments a number of phenomena
material will influence the outcome of a hardness test according to may complicate the evaluations of the experiments. Estimating the
Johnson [13,14]. Based on the fact that the stress field just beneath scratch depth can prove difficult due to difficulties in detecting ini-
the indenter in such a situation is almost hydrostatic, the process is tial contact, at least when using a conical stylus. When scratching
very similar to the case of expansion of a spherical cavity in a large of polymeric materials time dependence may come into effect [17]
solid due to an internal pressure and consequently the formula both during the actual scratching and afterwards when the residual
  deformation is to be measured. Also at scratching of polymers, the
2 E tan ˇ
H = repr 1 + ln , (3) deformation is to a great extent elastic, making the estimate of the
3 3(1 − 2 )repr
real contact area based on the residual groove difficult [17].
is derived for the hardness. (It should be emphasized that in the With the above in mind, the present analysis is devoted towards
analysis performed presently the results by Johnson [13,14] are only understanding of some fundamental mechanical issues at scratch-
used for a qualitative classification of the materials. For a quan- ing. Accordingly, experiments have been performed in this work in
titative description more recent results by e.g. Gao et al. [15] are conjunction with numerical simulations in order both to verify the
more suitable.) Finally, in level III,  > 30, plastic deformation is accuracy of previous simulations by Wredenberg and Larsson [1]
present all over the contact area and elasticity no longer influences and to try to clarify the behaviour of scratch quantities such as real
the hardness value of the material. This is also the region pertinent contact area, representative strain and the ploughing coefficient
to most standard engineering materials, such as steel and many of friction, which cannot be directly observed from experiments,
aluminium and copper alloys just to mention a few, and it is also as referred to global scratch quantities such as normal/tangential
the region where Eq. (1) applies. hardness and the apparent coefficient of friction. In particular, it is
In the same manner as for indentation testing, Bucaille et al. the goal of this investigation to clarify the mechanical behaviour
[16] analyse cone scratching of perfectly plastic materials and their at scratching for a wide-spread of materials (here represented by
results also indicate that the Johnson parameter in Eq. (2) can, metals and polymers). Commonly a conical stylus is used at scratch-
indeed, be used for the correlation of scratch parameters as well ing and out of simplicity, but not of necessity, such a stylus was
considered in the present analysis.

2. Background

The present analysis of scratching using a sharp conical stylus


concerns a problem where quasi-static and steady-state conditions
are assumed to prevail. It should be emphasised that substantial
efforts were devoted towards achieving such a situation during the
experiments. When these conditions are fulfilled the problem is, as
when normal indentation is at issue, self-similar with no charac-
teristic length present. Consequently, the normal hardness,
Fnorm
Hnorm = (4)
Anorm
and the tangential hardness,
Ftan
Htan = , (5)
Atan

as well as a ratio h/ A (where h is the scratch depth), will be con-
Fig. 2. Sketch of the characteristic behaviour of indentation hardness [14]. The
indentation hardness H divided by the representative stress repr is plotted against stant during the loading sequence of a scratch test and stresses
the non-dimensional strain parameter . and strains will be functions of the dimensionless variable (xi being
78 F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83

Cartesian coordinates) as shown in Fig. 4 and material properties, enabled the measurement of the residual scratch depth, which
alone. √ could later be related to the contact area.
In this context, A should be interpreted as a representative
contact length and the indices norm and tan represent the nor- 3.2. Specimens
mal and tangential components of the scratch quantities. The fact
that the normal and tangential scratch hardness are constant dur- The experiments were performed on two aluminium alloys,
ing a cone scratch test, is valid at classical elastoplastic material three polymers and one stainless steel. These materials were cho-
behaviour, as assumed here, but fails at for example strain gradient sen in order to achieve a reasonable spread in Johnson’s parameter
material behaviour as then a characteristic length is present in the  and in strain hardening. The material properties were deter-
constitutive equation, cf., e.g. Fleck and Hutchinson [21]. mined by uniaxial tensile tests, alternatively by compression for
Epoxy and PMMA (see Table 1). For these materials the results for
3. Experimental analysis the compression test rather than for the uniaxial tensile test was
used since then it was possible to reach higher strain levels without
In conjunction with the scratch experiments uniaxial tensile fracture. Also polymers have a tendency for higher yield stresses in
tests were performed in order to characterise the scratched speci- compression than in tension and since the material predominantly
men materials. Also purely frictional measurements were made to undergoes compression during the scratch test the compression
determine the diamond–specimen coefficient of friction. All exper- test was considered suitable. To extract the plastic material param-
iments were performed at room temperature. eters from the compression test an inverse analysis was performed
using ABAQUS and MATLAB where the materials were assumed to
be elastic perfectly plastic as indicated by the compression tests.
3.1. The scratch setup
The polymeric materials in Table 1 were to some limited extent
rate dependent. Thus the yield stress and consequently , would
Scratching experiments were performed by dragging a conical
depend on the rate of which the specimens were loaded. The speci-
diamond (with the angle ˇ = 22◦ ) over the surface of a specimen,
mens were therefore loaded slowly in a quasi static manner, similar
using a MTS 66202A-01 bi-axial servo-hydraulic machine with an
to the situation at scratching. During scratching at different veloc-
Instron 8500 control unit. The normal forces were measured with
ities no significant difference in the results were found and this
a MTS load cell and the tangential forces were measured by a load
matter was therefore not dwelled upon further.
cell built in-house. During the test normal and tangential forces
were recorded and steady-state conditions were ensured. The rota-
tional motion of the actuator was transformed to a translational 3.3. Indentation
motion by a carriage sliding on a rail (see Fig. 3). During scratch-
ing the normal load was held constant to allow for the stylus to In order to be able to correlate the scratch quantities to
follow the contours of the specimen. All specimens were scratched the indentation counterparts, indentation experiments were per-
with a speed of 0.4 mm/s and 0.9 mm/s and in some cases with a formed on the scratch specimens. The same conical indenter as
higher speed of 1.8 mm/s. No significant difference in global scratch for scratching was used also for indentation and was pressed into
parameters between the different scratch speeds could however be the specimens by a servo hydraulic MTS machine. The residual
detected. The normal load was held at constant at 250 N or 500 N indent (1–2 mm in diameter) was measured using an optical micro-
for the polymers while the metallic specimens were scratched with scope and the normally projected contact area was calculated. The
a normal load of 500 N, 1 kN or 1.5 kN. indentation experiments on the polymers were performed with a
After the specimen was scratched the residual groove was mea- maximum load of 500 N and with a hold time of 5 s. The on and off
sured both using an optical microscope and a profilometer (Form loading was virtually instantaneous. For the metallic materials, the
Talysurf Mk1) with a stylus radius of 0.2 ␮m. The profilometer maximum load was varied between 1.0 kN and 3.5 kN.

3.4. Friction

A similar setup as for scratching was used for frictional measure-


ments. Now a stylus with a spherical tip of the radius of 20 mm was
used. The normal load was adjusted so that the contact radius would
be small compared to the radius of the stylus, giving an essentially
flat contact surface. The normal load was held constant and the tan-
gential load was measured as the spherical tip was dragged across
the surface at approximately 1 mm/s. As the frictional results from
scratching did not indicate any dependence on scratch speed, the

Table 1
Material properties determined by uniaxial/compression testing

Material E (GPa) Y (MPa) ( = 35%) (MPa) 

Vinyl ester 3.5 108 108 15


PMMA 2.9 110 110 12
Epoxy 3.1 97 97 14
Al 7050 70 a 560 790 39
Al 4120 70 a 102 260 120
Stainless steel 200 a 560 760 120
Fig. 3. The experimental scratch setup with a carriage (green) sliding on a rail. Above
are two force transducers (shear and normal). The specimen is placed in the center E is Young’s modulus, Y the initial yield stress and ( = 35%) the yield stress at a
and in the vice (purple). (For interpretation of the references to color in this figure plastic strain of 35%. The Johnson parameter  was evaluated using ( = 35%).
a
legend, the reader is referred to the web version of the article.) From literature [22].
F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83 79

Table 2 integrity. The mesh was composed of some 30,000 eight node linear
Results from frictional measurements (standard deviation)
elements and is shown in Fig. 4. These elements were chosen since
Material i they show a faster convergence with respect to mesh refinement
Vinyl ester 0.14 (21 × 10−3 )
than tetrahedral elements and do not have the inherent contact
PMMA 0.33 (35 × 10−3 ) problems of quadratic elements [23]. The stylus was assumed
Epoxy 0.11 (33 × 10−3 ) to be perfectly rigid and Coulomb friction was assumed when
Al 7050 0.09 (9 × 10−3 ) appropriate.
Al SS 4120 0.14 (40 × 10−3 )
As regards boundary conditions, the surface outside the con-
Stainless steel SS 2333 0.08 (9 × 10−3 )
tact area was assumed traction free and within the area of contact
At least four experiments were performed for each material.
unilateral kinematic constraints, given by the shape of the conical
indenter/stylus depicted in Fig. 1, had to be accounted for.
frictional measurements were limited to one sliding speed. The nor-
mal load was also varied between the different measurements in 5. Results and discussion
order to investigate its effect on the friction. The polymer specimens
were washed with a mild detergent and the metallic with 2- Below, experimental and numerical results are presented, eval-
propanol (as was the stylus) to remove any grease from the surfaces. uated and compared in relation to different important features of
The results from the frictional measurements are shown in Table 2. substantial interest at scratching. In all subsequent figures when
nondimensionalised parameters are introduced, a representative
4. Numerical analysis stress, repr , was used, being the stress at a representative level
of plastic strain (repr = 35%) at scratching found by Wreden-
In order to verify the model developed by Wredenberg and Lars- berg and Larsson [1] in their numerical analysis of the problem.
son [1] each scratch experiment was numerically simulated using The level of the representative strain is not of fundamental
finite element methods implemented in the commercial FEM pack- importance in the present analysis as repr is only used for
age ABAQUS [23]. a qualitative classification of the materials mainly through the
As for details of the numerical analysis (including validation of parameter , Eq. (2). Cracking of the material was in generally not
the procedure) we refer to Wredenberg and Larsson [1] and here observed at scratching of metals and polymers. However epoxy,
it should just be mentioned that regarding the constitutive specifi- occasionally showed crazing at the very bottom of the residual
cation the incremental, rate independent Prandtl–Reuss equations groove.
for classical large deformation von Mises plasticity with isotropic In Figs. 7–13 the error bars indicate the standard deviation. All of
hardening according to the presented results are pertinent to a perfectly sharp conical sty-
lus. In order to ensure a relevant comparison between experimental
(p ) = Y + 0 np , (6) and numerical results, experimental scratch depths were chosen in
were implemented. In Eq. (6), (p ) is the flow stress, Y the ini- such a way that any influence from the stylus tip defect was neg-
tial yield stress, p the effective plastic strain and n the hardening ligible (as investigated carefully in each case through numerical
exponent. At elastic loading, or unloading, a hypoelastic formu- simulations).
lation of Hooke’s law, pertinent to the first elastic part of the
Prandtl–Reuss equations, was relied upon. Obviously, within the 5.1. Contact area
present setting, kinematic hardening effects were not included
in the analysis. Such effects could certainly have influenced the When the scratch hardness is to be determined, it is in gen-
outcome of scratch test but would also have increased the num- eral easy to acquire the load, the contact area however is more
ber of required numerical computations substantially (due to an challenging. Unlike normal indentation where the contact area is
increased number of constitutive parameters) and would have relatively accurately estimated from the residual indent, cf., e.g.
made a straightforward interpretation of the results more diffi- Tabor [7], the scratch contact area can only be roughly estimated
cult. For this reason, it was thought advisable, as a first attempt, to by the residual groove and requires assumptions about the con-
restrict the analysis to classical von Mises plasticity with isotropic tact area shape to be made. Below, the contact was assumed only
hardening. In particular so, as mainly the loading part of the scratch to occur on the front face of the indenter. Also the elastic effects
test was of primary interest here. As the material experienced very were assumed to be small, i.e. the contact width was the same as
large strains, adaptive meshing was used to maintain the element the residual groove width. The latter assumption is again based on
corresponding indentation results by Tabor [7].
As the real normal contact area was not easily measurable it was
(when not stated otherwise) approximated using
 w 2 
Anorm = , (7)
2 2

were w is the residual groove width [2]. For materials undergoing


predominantly plastic deformation (large  materials) this was a
fairly good approximation. This effect can be seen from numerical
results shown in Fig. 5, where the real normal contact area, Areal ,
is divided by Anorm and plotted against , defined in Eq. (2), for
different levels of hardening (where n is the hardening exponent in
a standard isotropic power law hardening model according to Eq.
(6)). The resolution of the width w was not sufficiently high when
Fig. 4. FE mesh used in the simulations along with the Cartesian coordinates xi obtaining the results for Fig. 5, although the general behaviour is
(following the tip of the stylus). Scratching is performed in the x2 direction. still clear.
80 F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83

ing this matter are presented in Fig. 5 and in many of the following
figures and tables.

5.2. The apparent coefficient of friction

Friction plays a critical role at scratching, influencing all scratch


quantities to some degree. When the scratch experiment was per-
formed two quantities, the normal and tangential forces (Fnorm and
Ftan , respectively) were acquired. The ratio of Fnorm /Ftan was in a
standard manner defined as the apparent coefficient of friction 0 .
It was assumed that the ploughing part p and interfacial part i
of the apparent coefficient of friction were separable [24](also sup-
ported by Felder and Bucaille [20] through simulations of scratching
of elastic perfectly plastic materials) and that the apparent coeffi-
Fig. 5. The real contact area divided by the contact area estimated by Eq. (7). Numer- cient of friction 0 could be described by
ical simulations with the hardening exponent according to Eq. (6).
0 = i + p (). (9)
The scratch tangential hardness can be approximated in a similar The interfacial friction was simply defined as the friction between
manner by the surface of the diamond stylus and the surface of the specimen
whereas the ploughing coefficient of friction was the contribution
w2
Atan = tan ˇ. (8) from the contact pressure on the stylus walls. Thus the plough-
4
ing friction is an effect of large displacements. Through numerical
In Eq. (8) ˇ is the stylus angle as defined in Fig. 1. The estimates in simulations it is found that the ploughing part p depend on John-
Eqs. (7) and (8) will be more accurate for large  materials as the son’s parameter  alone, regardless of hardening characteristics
effect of elastic recovery then diminishes as do the contact area on [1]. This behaviour was experimentally investigated and the results
the rear face of the stylus. are shown in Fig. 7 along with the numerical results. The plough-
Due to pileup around the stylus a purely geometric approach, i.e. ing friction was found by measuring the apparent coefficient of
where the contact radius is calculated from the scratch depth and friction 0 through scratch experiments and subtracting the inter-
angle of the stylus instead of measurements of the groove width, facial coefficient of friction i found by frictional measurements
yields only a very rough estimate. Far less accurate than the esti- described earlier. The importance of interfacial friction on global
mates described by Eqs. (7) and (8). This is so also at indentation scratch quantities is shown by Bellemare et al. [25] numerically and
testing. experimentally by scratching pure nickel, assuming an interfacial
Furthermore, the amount of interfacial friction will influence the coefficient of friction of 0.15.
shape of the contact area, stretching it in the scratch direction (see The experimental results for ploughing friction in Fig. 7 showed
Fig. 6), making the estimate of Anorm less accurate. reasonable correlation with the numerical prediction, except for
Following this discussion it should be stressed once again that Al 4120 which had a surprisingly high ploughing coefficient of fric-
there are many features that will make contact area estimates tion. This may be possibly explained by the formation of aluminium
according to Eqs. (7) and (8) inaccurate. This matter is of great oxide, causing the frictional test described earlier to measure the
importance when evaluating scratch experiments and was inves- interfacial friction between aluminium oxide and diamond rather
tigated and accounted for throughout the analysis. In particular, than aluminium and diamond, and/or due to the inadequacy of
the real normal contact area at loading (denoted Areal above and Coulomb friction to describe the frictional interaction of Al 4120
below), determined from numerical simulations, were compared and the diamond stylus.
to the experimentally determined normal contact area (denoted
Aestimated below), estimated using Eq. (7). Pertinent results concern- 5.3. Scratch hardness and the concept of a representative strain

Wredenberg and Larsson [1] suggest a level of a representative


plastic strain at frictionless scratching of 35% with a C ≈ 2.5 (see

Fig. 6. Contact area at scratching of Al 4120 (simulated). Interfacial friction i = Fig. 7. Experimental results for ploughing friction. Numerically simulated results
0.18. The scratch direction is indicated by the arrow. are also shown.
F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83 81

Fig. 8. Scratch normal hardness divided by the representative stress repr . Plotted Fig. 10. Scratch tangential hardness divided by the representative stress repr . Plot-
against Johnson’s parameter . Experimental results. ted against Johnson’s parameter . Experimental results.

Eq. (1)) through numerical simulations. According to the exper-


imental results in Fig. 8 it is clear that the constant C is larger
than 2.5. It may be me mentioned in this context, that Jardret and
Morel [26] find the representative level of plastic strain at ≈ 34%
for large contact radius scratching of PMMA with ˇ = 30◦ . Simi-
larly Bellemare et al. [25] report of a representative strain of 15.5%
or 33.6%, depending on the level of strain hardening, at scratch-
ing with an attack angle of ˇ = 19.7◦ . However, one must bear in
mind the influence of friction on the hardness, mainly due to an
underestimation of the contact area. In order to compensate for
this effect the “real” contact area was extracted through numeri-
cal simulations (see Section 5.4). These results are shown in Fig. 9
(where “adjusted contact area” indicates that estimated contact
area using Eq. (7) is multiplied by a ratio of Areal /Aestimated found
through numerical simulations of the material in question), which
indicates a value of C much closer to the predicted 2.5. Felder and Fig. 11. Scratch tangential hardness divided by the representative stress repr with
adjusted contact area. Plotted against Johnson’s parameter . Experimental and
Bucaille [20] show very similar results for elastic perfectly plastic numerical results.
frictionless scratching.
The tangential hardness divided by the representative stress
friction measurements were not accounted for when determining
(seen in Fig. 10) shows the curve levelling out at approximately
the standard deviation in Figs. 10–12.
4. The same data is shown in Fig. 11, but here the contact area
In Fig. 12 the scratch normal hardness divided by the tangential
is adjusted in the same way as in Fig. 9. The interfacial frictional
hardness shows good agreement between the experimental and
part of the tangential hardness in Figs. 10–12 has been compen-
the numerical results found by Wredenberg and Larsson [1] above
sated to allow for a more direct comparison to the numerical results
and in many of the following figures and tables. Although it does
for frictionless contact, found by Wredenberg and Larsson [1]. The
not level out to a value of 1.
interfacial frictional part of the tangential hardness was removed
Fig. 13 shows the ratio of normal scratch hardness to indentation
by subtracting the interfacial frictional force i Fnorm from the mea-
hardness, Hind , for the different specimens. As expected; in the case
sured tangential force Ftan . The uncertainties of the interfacial
of metals, the material with the highest degree of strain hardening

Fig. 9. Scratch normal hardness with adjusted contact area. Plotted against John- Fig. 12. Scratch normal hardness divided by tangential scratch hardness, plotted
son’s parameter . Experimental and numerical results. against Johnson’s parameter . Experimental and numerical results.
82 F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83

Fig. 13. Scratch normal hardness divided by indentation hardness, plotted against
Johnson’s parameter . Experimental results.

Table 3 Fig. 14. Typical cross section of the remaining groove (simulated). The groove width
Scratch quantities at simulated scratching w is indicated.

Material Hnorm (MPa) Htan (MPa) Areal /Aestimated p


ing the highest point as the groove boundary would produce lower
Vinyl ester 330 300 1.8 0.23
PMMA 520 1100 2.1 0.23 scratch hardness values than can be seen in Table 3. The ratio of
Epoxy 310 400 1.7 0.22 real to estimated normally projected contact area (see Table 3)
Al 7050 2800 3800 1.6 0.26 was used to calculate the “real” contact area when evaluating
Al SS 4120 640 1200 1.3 0.29 the scratch experiments (see Figs. 9 and 11 with adjusted contact
Stainless steel SS 2333 2500 3400 1.4 0.27
area). A quantity which would otherwise not have been avail-
Here Areal and Aestimated denote the normally projected real contact area and the able. In comparison, Jardret [27] reports of elastic recovery ratios
estimated contact area by Eq. (7), respectively.
of the scratch depth to remaining groove depth of approximately
1.25–2.8, when scratching polymer surfaces with a Berkovich
(stainless steel) showed the largest Hnorm /Hind ratio. It should be indenter.
noted in passing that the dependence of  in Fig. 12 is a result of the As for the present numerical and experimental results it can
fact that the approximation of the contact area being a half-circle be directly concluded, when it comes to normal and tangential
(Eqs. (7) and (8)) is only accurate at fully plastic contact. scratch hardness, that the two sets of results are relatively close,
giving some confidence as regards the reliability of the numer-
5.4. Experimental results vs. numerical results ical approach. However, as is also evident from the results in
Tables 3 and 4 the situation is worse for the ploughing coefficient
One of the most important aims of the present analysis concerns of friction where the values differ to some extent. It is clear though
the reliability of the numerical approach presented by Wredenberg that this quantity is significantly influenced by the interfacial fric-
and Larsson [1]. This has already been discussed above for a gen- tion (i ) and remembering the difficulties involved in a proper
eral set of numerical results, but of more immediate interest is of characterisation of this feature, the difference between the two
course to compare explicitly experimental and numerical results for sets of results (as regards the ploughing coefficient of friction) is
the actual materials constitutively described in Table 1, accounting not alarming.
explicitly for frictional effects in the numerical calculations. The In summary then, the comparison conducted in the present (and
resulting scratch hardness quantities are shown in Tables 3 and 4. also the results shown in previous sections) section of this paper
It should be mentioned in this context that the scratch quanti- indicates that the numerical approach designed by Wredenberg
ties Hnorm and Htan were calculated not using the real contact area and Larsson [1] is a reliable tool for simulation of mild scratch-
but estimated by Eq. (7) and (8) so that the numerical results could ing. It should be immediately emphasised though that the results
be directly compared to the experimental results. In the numeri- also suggest that an accurate and detailed understanding of the
cal analysis the groove width was determined by identifying the frictional behaviour is of utmost importance for a proper determi-
first element with a substantially different slope (starting from the nation of important scratch quantities.
centre of the groove). The position of the first node of that ele-
ment was then considered to be the boundary of the groove (see 6. Conclusions
Fig. 14). This value may not coincide with the highest point. Choos-
In the present investigation a combined experimental and
Table 4 numerical study of the scratch test of homogeneous material was
Scratch quantities at experimental scratching (standard deviation) conducted. The most important results can be summarised as fol-
lows:
Material Hnorm (MPa) Htan (MPa) p

Vinyl ester 400 (20) 480 (70) 0.14 (0.06)


• The interpretation of scratch experiments rely heavily on
PMMA 590 (50) 1100 (40) 0.14 (0.04)
Epoxy 370 (50) 420 (50) 0.22 (0.09) assumptions about the size of the contact area between the mate-
Al 7050 3200 (610) 4000 (330) 0.30 (0.04) rial and the stylus and also about the frictional behaviour at
Al SS 4120 640 (60) 1500 (130) 0.48 (0.04) scratching. In many situations standard assumptions regarding
Stainless steel SS 2333 2800 (110) 4200 (120) 0.26 (0.09)
these features are not very accurate and it is of utmost impor-
The values are averages of at least four experiments. tance to use also numerical tools in order to understand the
F. Wredenberg, P.-L. Larsson / Wear 266 (2009) 76–83 83

mechanical behaviour at scratching. This is particularly so for [8] A.G. Atkins, D. Tabor, Plastic indentation in metals with cones, Journal of The
polymeric materials where, for example, the standard assump- Mechanics and Physics of Solids 13 (1965) 149–164.
[9] A.E. Giannakopoulos, P.-L. Larsson, R. Vestergaard, Analysis of vickers indenta-
tion of a contact area restricted to only the front of the stylus tion, International Journal of Solids and Structures 31 (1994) 2679–2708.
leads to inaccurate results. [10] P.-L. Larsson, A.E. Giannakopoulos, E. Söderlund, D.J. Rowcliffe, R. Vestergaard,
• The numerical approach presented previously by Wredenberg Analysis of berkovich indentation, International Journal of Solids and Structures
33 (1996) 221–248.
and Larsson [1] is a reliable tool for investigating scratching in [11] A.E. Giannakopoulos, P.-L. Larsson, Analysis of pyramid indentation of pressure-
a detailed manner. However the present study also shows that an sensitive hard metals and ceramics, Mechanics of Materials 25 (1997) 1–35.
accurate characterisation of the frictional behaviour at scratching [12] P.-L. Larsson, A.E. Giannakopoulos, Tensile stresses and their implication to
cracking at pyramid indentation of pressure-sensitive hard metals and ceram-
is essential at interpretation of many relevant scratch quantities. ics, Materials Science and Engineering A 254 (1998) 268–281.
[13] K.L. Johnson, The correlation of indentation experiments, Journal of The
Acknowledgements Mechanics and Physics of Solids 18 (1970) 115–126.
[14] K.L. Johnson, Contact Mechanics, Cambridge University Press, UK, 1985.
[15] X.-L. Gao, X.N. Jing, G. Subhash, Two new expanding cavity models for indenta-
The authors want to acknowledge the support through grant tion deformations of elastic strain-hardening materials, International Journal
621-2005-5803 from the Swedish Research Council. The authors of Solids and Structures 43 (2005) 2193–2208.
[16] J.L. Bucaille, E. Felder, G. Hochstetter, Mechanical analysis of the scratch test on
also wish to thank Professor Fred Nilsson for providing valuable
elastic and perfectly plastic materials with three-dimensional finite element
advice and discussions, and for reading and commenting on the modeling, Wear 249 (2001) 422–432.
manuscript, Professor Anders Hult and Professor Mats Johansson [17] B.J. Briscoe, P.D. Evans, S.K. Biswas, S.K. Sinha, The hardness of poly
for their help in choosing and acquiring polymer materials, Mr. (methylmethacrylate), Tribology International 29 (1996) 93–104.
[18] C. Gauthier, S. Lafaye, R. Schirrer, Elastic recovery of a scratch in a polymeric
Hans Öberg for his help and advise in the laboratory and Messrs. surface: experiments and analysis, Tribology International 34 (2001) 469–479.
Bertil Dolk and Kurt Lindqvist for manufacturing specimens and [19] J.L. Bucaille, E. Felder, G. Hochstetter, Experimental and three-dimensional
experimental equipment. finite element study of scratch test of polymers at large deformations, Journal
of Tribology 126 (2004) 372–379.
[20] E. Felder, J.-L. Bucaille, Mechanical analysis of the scratching of metals and
References polymers with conical indenters at moderate and large strains, Tribology Inter-
national 39 (2006) 70–87.
[1] F. Wredenberg, P.-L. Larsson, On the numerics and correlation of scratch [21] N.A. Fleck, J.W. Hutchinson, A phenomenological theory for strain gradient
testing, Journal of Mechanics of Materials and Structures 2 (2006) effects in plasticity, Journal of the Mechanics and Physics of Solids 41 (1993)
573–594. 1825–1857.
[2] J. Goddard, H. Wilman, A theory of friction and wear during the abrasion of [22] T. Lyman (Ed.), Metals Handbook, 8th edition, American Society for Metals,
metals, Wear 5 (1962) 114–135. 1961.
[3] T.H.C. Childs, The sliding of rigid cones over metals in high adhesion conditions, [23] ABAQUS, ABAQUS Manual, v. 6.4, 2004.
International Journal of Mechanical Sciences 12 (1970) 393–403. [24] F.P. Bowden, D. Tabor, The Friction and Lubrication of Solids, Clarendon Press,
[4] M. De Vathaire, F. Delamare, E. Felder, An upper bound model of ploughing by 1950.
a pyramidal indenter, Wear 66 (1981) 55–64. [25] S. Bellemare, M. Dao, S. Suresh, The frictional sliding response of elasto-plastic
[5] P. Gilormini, E. Felder, Theoretical and experimental study of the ploughing of a materials in contact with a conical indenter, International Journal of Solids and
rigid-plastic semi-infinite body by a rigid pyramidal indenter, Wear 88 (1983) Structures 44 (2006) 1970–1989.
195–206. [26] V. Jardret, P. Morel, Viscoelastic effects on the scratch resistance of polymers:
[6] Y. Xie, H.M. Hawthorne, A controlled scratch test for measuring the elastic prop- relationship between mechanical properties and scratch properties at various
erty, yield stress and contact stress–strain relationship of a surface, Surface and temperatures, Progress in Organic Coatings 48 (2003) 322–331.
Coatings Technology 127 (2000) 130–137. [27] V. Jardret, Understanding and quantification of elastic and plastic deformation
[7] D. Tabor, Hardness of Metals, Oxford University Press, UK, 1951. during a scratch test, Wear 218 (1998) 8–14.

You might also like