You are on page 1of 12

Journal of the Mechanics and Physics of Solids 60 (2012) 379–390

Contents lists available at SciVerse ScienceDirect

Journal of the Mechanics and Physics of Solids


journal homepage: www.elsevier.com/locate/jmps

Fracture scaling relations for scratch tests of axisymmetric shape


Ange-Therese Akono, Franz-Josef Ulm n
Massachusetts Institute of Technology, Cambridge MA 02139, United States

a r t i c l e i n f o abstract

Article history: Scratch testing and scratch test analysis continues to gain momentum in Applied
Received 22 March 2011 Mechanics, due to the possibility offered by this method to assess fracture properties at
Received in revised form very fine scales. In this paper, we derive general scratch force scaling relations for
8 December 2011
axisymmetric scratch probes defined by single variable monomial functions. These
Accepted 14 December 2011
relations are used to define fracture criteria with and without consideration of the
Available online 22 December 2011
development of shear stresses at the probe–material interface. The approach is
Keywords: illustrated for common scratch probe geometries: conical probe, flat punch, and
Scratch test analysis hemi-spherical probe. Application of the proposed method to micro-scratch tests on
Axisymmetrical probe
two materials (an aluminum alloy and a thermoplastic polymer) using a Rockwell probe
Fracture mechanics
(a conical probe ending in a hemi-spherical shape) illustrates the versatility of the
Scaling
Fracture toughness approach: First, the scratch force-depth scaling relations provide a means to determine
the degree of the homogeneous function characterizing the scratch probe. Second, the
fracture criteria enable an experimental assessment of the fracture toughness. The good
agreement between the fracture toughness determined by scratching and values
reported in the open literature show the potential of the proposed method for
determining fracture properties of materials at even smaller scales.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Scratching a weaker material with a tougher one is no doubt the most elemental conceptualization of a mechanics-of-
materials test ever conceived by mankind. In use as a tool to compare the relative hardness of two materials since ancient
time, the first abstraction of scratch resistance into a quantitative metric of material classification is due to Carl Friedrich
Christian Mohs (1773–1839), who in 1824 put the ability of one mineral sample to scratch another on an ordinal (rank-
ordering) scale, the Mohs scale of mineral hardness (Bowden and Tabor, 1964). Scratch force criteria as a basis for
comparison of scratch resistance of materials emerged throughout the 20th century, and namely the scratch hardness, HT,
that links the horizontal scratch force, FT, required to move the scratch probe, to the contact area between the probe and
the scratched material projected in the scratch direction, ALB:
def
F T ¼ HT ALB ð1Þ

All this led to the development of instrumented scratch tests in the early 1970s, in which the applied forces and acoustic
emissions generated by microcracks are measured simultaneously (Wilshaw and Rothwell, 1971). In turn, these advances
in instrument development were the stepping stone to link a measurable mechanical event, i.e. force or hardness, to
material behavior, i.e. plastic or fracture behavior. They enabled theoretical works in scratch test analysis that ultimately

n
Corresponding author.
E-mail address: ulm@mit.edu (F.-J. Ulm).

0022-5096/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jmps.2011.12.009
380 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

aim at linking the scratch force or hardness and scratch test geometry to mechanical properties of the scratched material;
like strength, friction and fracture properties (Williams, 1996; Schei et al., 2000; Randall et al., 2001; Atkins, 2005;
Bellemare et al., 2007; Patel et al., 2009; Akono and Ulm, 2011; Bard and Ulm, in press; Ulm and James, 2011). The
overarching theme of these contributions is that rationalizing scratch test force measurements into intrinsic material
parameters is highly dependent on the geometry of the test set-up. These observations motivate this paper. In particular,
with a focus on fracture properties, this paper aims at developing a framework amendable to the scaling of scratch force
measurements for arbitrary scratch probes of axisymmetrical shape; similar to developments over the last decade or so in
the field of indentation analysis (Borodich and Galanov, 2002; Borodich et al., 2003; Cheng and Cheng, 2004).

2. Method development

2.1. Geometric description of axisymmetric probe

Consider an axisymmetric shape of a scratch probe. The origin of the coordinate system is at the tip of the probe (Fig. 1).
Analogous to earlier contributions to indentation analysis (Borodich et al., 2003; Ma et al., 2002), we consider the shape of
the probe to be defined by a single variable monomial functions of the form:
z ¼ Br E ð2Þ
1E
where B (of dimension L ) is the height at unit radius, and E is the degree of the homogeneous function. The degree E and
the proportionality factor B for common probe shapes considered is given in Table 1. Some geometric relations will turn
out useful for the following developments. These are the outward unit normal of the axisymmetric scratch probe:
EBrE1 cos f 1
n ¼  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e r þ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e z ð3Þ
E1 2
1 þ ðEBr Þ 1 þðEBr E1 Þ2

for r 2 ½0,ðd=BÞ1=E  and f 2 ½p=2, p=2; and the differential line element, ds, and surface element, dS:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ds ¼ 1þ ðEBr E1 Þ2 dr ð4aÞ

dS ¼ r df ds ð4bÞ

Fig. 1. Axisymmetric scratch probe geometry. x is the scratch direction due to the application of the scratch force F T . The scratch probe is maintained at a
depth d through application of a vertical force.

Table 1
Degree d of the homogeneous function and proportionality factor B for several scratch probes.

Probe type E B

Conical 1 cot y
Spherical 2 1=ð2RÞ
Flat Punch -1 1=ðRE1 Þ
A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390 381

Derived geometric quantities of interest are the contact area projected in the scratch direction nx ¼ n  e x ,
Z  ð1=EÞ þ 1
2BE d
ALB ðdÞ ¼  nx dS ¼ ð5Þ
ðSÞ Eþ1 B
and the perimeter p of the probe projected onto the scratch direction (Fig. 1):
Z  1=E
d
pðdÞ ¼ ds ¼ b ð6Þ
B
with b a noteworthy dimensionless parameter:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z 1  2=E
d
b¼2 1 þðEdÞ2 x2E2 dx ð7Þ
0 B

2.2. Contour integral method

We hypothesize the existence of crack planes surrounding the scratch probe, G ¼ p‘, with p the perimeter of the probe
defined by (6), and ‘ the crack length in the scratch direction x. The theoretical framework employed in our study is an
extension of the contour integral method for the evaluation of the energy release rate (Rice, 1968). Yet, since there are
some minor, still important differences in the adaption of this method to the scratch test, it is useful to recall some recent
developments first that deal with the application of the contour integral method to scratching as a fracture phenomenon
(Akono et al., 2011). More precisely, we are interested in evaluating the dissipation rate from the spontaneous change of
potential energy, Epot, due to the creation of fracture surface G:
dEpot d‘ @Epot
D¼ ¼ Gp ; G¼ ð8Þ
dt dt @G
where G is the energy release rate. True to the original idea of the contour integral method, we estimate the change in
potential energy during the fracture process from the perspective of an observer that is attached to the propagating crack.
In a displacement driven test, the observer thus defined witnesses two sources of potential energy change: one due to the
change in free energy density c in a material volume V inclosing the crack tip; the other due to the energy release that is
_ x passed the (fixed) observer:
convectively transported at a speed V  n ¼ ‘n
Z Z
dEpot @c _ x dA
¼ dV c‘n ð9Þ
dt V @t A

where A is the closed boundary of V. For a linear elastic material, for which c ¼ 12sij ui,j (with sij the symmetric stress tensor,
ui,j the displacement gradient), the divergence theorem allows a change of the volume integral into a surface integral:
Z Z Z
@c @ui,j @u
dV ¼ sij dV ¼ t i i ‘_ dA ð10Þ
V @t V @t A @x
where t i ¼ sij nj are surface tractions, and where we made use of the fact that in a displacement-controlled fracture test the
displacement rate seen by the moving observer is equal to ui,t ¼ ‘_ ui,x . A comparison of Eqs. (8)–(10) provides the following
expression of the energy release rate:
Z  
1 @u
G¼ cnx t i i dA ð11Þ
p A @x
Compared to the classical form of the J-integral (Rice, 1968) for planar cracks, in which the fracture perimeter coincides
with the fracture width, dA¼p ds, we have chosen to consider a difference between these lengths, in order to employ the
technique for different scratch geometries. For the scratch problem at hand, we choose a closed volume that includes the
probe–material interface, the stress-free surface at the top (nx ¼ 0; t i ¼ 0), the (stress free) fracture surfaces in prolongation
of the scratch probe surface (nx ¼ 0; t i ¼ 0), and closing material surfaces far removed from the surfaces (c ¼ 0; ui,x ¼ 0); so
that the only contribution to the surface integral comes from the probe–material interface S:
Z  
1 @u
G¼ cnx ti i dS ð12Þ
p ðSÞ @x

Physically speaking, the energy release so defined can be associated with the energy stored, prior to chipping, into a
material domain in front of the scratch probe.

2.3. Fracture criterion

In a first approach (to be refined later on), we consider that the sole stress contributing to energy changes ahead of the
scratch device is sxx , which is related to the applied scratch force by:
Z
FT ¼ sxx nx dS ¼ HT ALB ð13Þ
ðSÞ
382 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

The free energy is c ¼ ð1=2Þks2xx =E and the work of the stress vector along @ux =@x ¼ Exx is sxx nx Exx ¼ ks2xx nx =E (E ¼ Young’s
modulus; k ¼ 1 in plane stress and k ¼ 1n2 in plane strain application; n ¼ Poisson’s ratio). The energy release rate thus
obtained is the quadratic stress average over the projected contact area nx dS ¼ dSx :
Z Z
k k
G¼ s2xx nx dS ¼ s2 da ð14Þ
2pE ðSÞ 2pE ðALB Þ xx

The problem to be solved amounts to combining the linear stress average of the boundary condition (13) with the
quadratic stress average (14). For instance, assuming a constant stress field over the probe surface, we have:
k kALB
G¼ F 2T ¼ H2T ð15Þ
2pALB E 2pE
where we note that ALB ¼ Sx is the nominal area of the projected surface area defined by (5), and p the perimeter defined
by (6). Then using these expressions in (15), we obtain the following expression of the energy release rate for any
axisymmetric probe:
 ð2=EÞ þ 1  
k 1þE B k BE d
G¼ ðEBÞ1 F 2T ¼ H2T ð16Þ
E 4b d E bðE þ 1Þ B
Finally, when the fracture propagates, the energy release rate equals the fracture energy, Gc, which in turn relates to the
fracture toughness by Gc ¼ ðk=EÞK 2c . We thus obtain the following criterion for fracture propagation:
 ð1=EÞ þ ð1=2Þ  1=2
d E
F T rF c ¼ 2K c bB ð17Þ
B 1þE
or in terms of the scratch hardness (1):
 1=2
Fc 1=2 1þ E
HT r Hc ¼ ¼ Kcd b ð18Þ
ALB E
The previous relations thus reveal that the scratch force and scratch hardness, upon fracture propagation (i.e. F T ¼ F c ,
HT ¼ Hc ), scale with the scratch depth d according to:
 
F c ðldÞ bðldÞ 1=2 ð1=EÞ þ ð1=2Þ
¼ l ð19Þ
F c ðdÞ bðdÞ
 
Hc ðldÞ bðldÞ 1=2 1=2
¼ l ð20Þ
Hc ðdÞ bðdÞ
k
Due to the lack of geometrical self-similarity of the dimensionless perimeter function (7), that is bðldÞal bðdÞ, there is no
reason that scratch force or scratch hardness should obey to self-similar transformation rules, except for special yet highly
relevant cases; as illustrated later on.

2.4. Theory refinement: account for shear stresses

Arguably, the assumption that the axial stress solely contributes to energy changes may be an oversimplification of the
actual stress field that develops ahead of the scratch probe. A refined analysis will in addition consider (at least) shear
stresses at the probe–material interface due to the presence of a vertical force, FV, that maintains the scratch probe at
depth d (Fig. 1). The focus of this section is to incorporate this effect. Instead of Eq. (13) we thus consider the following
force–stress condition:
Z
F ¼ F T e x F V e z ¼ ðs  nÞ dS ð21Þ
ðSÞ

with
Z
FT ¼ ðsxx nx þ sxz nz Þ dS ð22aÞ
ðSÞ

Z
F V ¼ ðsxz nx Þ dS ð22bÞ
ðSÞ

The stresses, sxx and sxz , are estimated by an Airy stress function approach (Akono and Ulm, 2011):
 
3
jðx,zÞ ¼ bx z3  zd2 þcz2 ð23aÞ
4

@2 j
sxx ¼ ¼ 6bxzþ 2c ð23bÞ
@z2
A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390 383

 
@2 j 3 2
sxz ¼  ¼ b 3z2  d ð23cÞ
@x@z 4

@2 j
szz ¼ ¼0 ð23dÞ
@x2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where ðb,cÞ are constants that are determined from using (23) in (22) [with dS ¼ r df 1þ ðEBr E1 Þ2 dr]:
 
ð2E þ1Þð3E þ 1Þ B 1=E F V
b¼ ð24aÞ
6E2 d d
3

 1=E
Eþ1 B FT
c¼ ð24bÞ
4E d d

The second required refinement is an estimate of the displacement gradient at the probe–material interface, for which
we use a potential function V that satisfies DV ¼ 0 and V, xz ¼ Dj ¼ sxx :
 4 
b x þz4
V ¼  þ 3ðxzÞ2 þ2cxz ð25aÞ
2 2
 
@ux 1 @2 j @2 V k
¼ ð1 þ nÞ 2 þ k ¼ sxx ð25bÞ
@x E @x @z@x E
 
@uz 1 @2 j @2 V 1
¼ ð1 þ nÞ þ k 2 ¼ ðð1 þ nÞsxz þ kV, xx Þ ð25cÞ
@x E @z@x @x E

where ðb,cÞ are still defined by (24). Use of the previous expressions in (12) simplifies to:
Z  2 
k sxx
G¼ nx þ sxz ðsxx nz þ V, xx nx Þ dS ð26Þ
Ep ðSÞ 2

After (somewhat hefty) integration the overall expression of the energy release rate we thus obtain is quite similar to
expression (16):
 ð2=EÞ þ 1
k 1þE B k
G¼ ðEBÞ1 F 2eq r Gc ¼ K 2c ð27Þ
E 4b d E

where an equivalent force, Feq, accounts for both the scratch force, F T , and the indentation force, FV:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 ð4E2 þ E þ 1Þð2E þ1Þð3E þ 1Þ 2
F eq ¼ F 2T þ FV ð28Þ
2 EðE þ1Þð4E þ 1Þð5E þ 1Þ

The fracture criterion is of a similar form as Eq. (17):


 ð1=EÞ þ ð1=2Þ  1=2
d E
F eq r F c ¼ 2K c bB ð29Þ
B 1þE

The most important result of the refined analysis is that a consideration of shear stresses at the probe–material interface
does not affect the scaling relations, but only the scratch force definition, Eq. (28), the weighting function of the vertical
force being a function of only the degree of the homogeneous function, E.

3. Application to common scratch probes

To illustrate the previous developments, we consider common scratch probes defined in Table 1.

3.1. Conical scratch probe

Consider a conical scratch probe, E ¼ 1; B ¼ cot y. Application of Eqs. (5)–(7) readily gives:
Z 1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 2 d 2
ALB ¼ d ; p¼ b; b ¼ 2 1þ ðcot yÞ2 ds ¼ ð30Þ
cot y cot y 0 siny

The energy release rate is given by:


k cos2 y F 2T k
G¼ rGc ¼ K 2c ð31Þ
E 4 sin y d3 E
384 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

Whence the fracture criteria:


pffiffiffiffiffiffiffiffiffiffiffi
3=2 sin y 1=2 1
F T rF c ¼ 2K c d ; HT r Hc ¼ 2K c d pffiffiffiffiffiffiffiffiffiffiffi ð32Þ
cos y sin y
Thus, due to the invariance of the b-function w.r.t. the scratch depth d of the conical probe, the scratch-force and scratch
hardness obey to self-similar scaling relations:
3=2 1=2
F c ðldÞ ¼ l F c ðdÞ; Hc ðldÞ ¼ l Hc ðdÞ ð33Þ

with l 2 R þ . Finally, the force-scaling relations remain valid if one replaces F T by Feq as defined by (28):
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3
F eq ¼ F 2T þ F 2V ð34Þ
5
Last, to put these developments in context, it is interesting to note that a similar scaling of the fracture force with depth
3=2
d has been obtained in the past from LEFM analysis of the pull-out force of axisymmetric anchors (Ballarini et al., 1985;
Karihaloo, 1996). Given the difference in load and boundary conditions, and associated stress fields between a pull-out test
and the scratch test, this scaling is readily attributed to the particular axisymmetric geometry.

3.2. Flat punch

The perimeter of the flat punch is


 
d
p ¼ 2R 1 þ ð35Þ
R

Written in form of monomial function, it is ðE-1; B ¼ R1E Þ:


 1=E
d
p ¼ lim 1 E b ð36Þ
E-1 R

with (see Appendix A):


sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Z 1  ð2=EÞ  
2 d d
b ¼ lim 2 1 þ E2 d x2E2 dx ¼ 2 1þ ð37Þ
E-1 0 B R

Thus,
 ð1=EÞ þ 1
2ER1E d
ALB ¼ lim ¼ 2Rd ð38Þ
E-1 E þ1 R1E
Expression (16) thus permits evaluating the energy release rate as:
k 1 2 k k
lim G ¼  F ¼   d H2T rGc ¼ K 2c ð39Þ
E-1 E 8R2 d 1þ d T 2E 1 þ Rd E
R

Whence the fracture criteria for the flat punch:


sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  rffiffiffiffiffiffiffiffiffiffiffiffi
d 2K c d
F T rF c ¼ 2R 2d 1þ K c ; HT r Hc ¼ p ffiffiffi 1þ ð40Þ
R d R

Thus, for small depth-to-radius values only, the fracture force and hardness obey to self-similarity:
1=2 1=2
ðd=RÞ 51; F c ðldÞ ¼ l F c ðdÞ; Hc ðldÞ ¼ l Hc ðdÞ ð41Þ

It need to be emphasized that the LEFM solution here derived neglects by design any plastic zone ahead of the sharp corner
of the punch, which may well affect the scaling even in a brittle material. The consideration of combined plastic and
fracture phenomena in scratching goes beyond the focus of this paper.

3.3. Spherical scratch probe

The loss of self-similarity found for the flat punch is even more pronounced for the (hemi-)spherical scratch probe,
which is classically approximated by a parabolic shape, E ¼ 2,B ¼ 1=ð2RÞ, for which:
2 pffiffiffiffiffiffiffiffiffi
ALB ¼ ð2RdÞ3=2 ; p ¼ 2Rd l ð42Þ
3R
  pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffi
d 1
b x¼ ¼ 1 þ2x þ pffiffiffiffiffiffi arc sinhð 2xÞ ð43Þ
R 2x
A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390 385

1 3
0.9 Normalized y = 0.8984x 2.9
0.8 Force R = 0.9984 2.8
Beta (d/R)
0.7 Linear Fit 2.7

Fc /(4KcR3/2)
0.6 2.6
0.5 2.5

β
0.4 2.4
0.3 2.3
0.2 2.2
0.1 2.1
0 2
0 0.2 0.4 0.6 0.8 1
d/R

Fig. 2. Scaling of the normalized fracture force and the perimeter function, b, of the spherical scratch probe. (The linear fit of the normalized force is
almost indistinguishable from the actual curve within d=R 2 ½0; 1.)

The energy release rate is given by:


k 3 F 2T k
G¼ r Gc ¼ K 2c ð44Þ
E 16b Rd2 E
Whence the fracture criteria:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi 1 d
F T r F c ¼ 4K c d R b ð45Þ
3 R
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
Kc 3 d
HT rHc ¼ pffiffiffi b ð46Þ
d 2 R
1=2
Except for small values of d=R5 1, for which b ¼ 2, and for which F T ðldÞ ¼ l F T ðdÞ and HT ðldÞ ¼ l HT ðdÞ, there is strictly
no self-similar relation for the parabolic scratch probe. On the other hand, for all practical ranges of hemi-spherical scratch
probes (that is, d=R 2 ½0; 1), bðd=RÞ varies little between 2 and 2:5425, and does little affect the scaling (Fig. 2); so that the
following self-similar scaling relation can be used in a first, but very good approximation:
1=2
F c ðldÞ  lF c ðdÞ; Hc ðldÞ  l Hc ðdÞ ð47Þ
Finally, replacing FT by F eq as defined by (28) provides a means to account for shear force transmission over the interface:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
665 2
F eq ¼ F 2T þ F ð48Þ
1188 V

4. Materials application

By purpose of application of the fracture scaling relations, we consider a series of scratch tests carried out with a
Rockwell micro-scratch probe, a conical probe with a 2y ¼ 1201 included angle blending tangentially with a spherical tip of
R ¼ 200 mm radius. The popularity of the Rockwell probe in scratch testing relates to this geometry: the smooth spherical
shape for small depths ensures the longevity of the probe compared to sharp probe shapes (Randall et al., 2001). The
transition from the sphere to the cone occurs at a radius of r ¼ 100 mm (Ma et al., 2002), corresponding to a depth-to-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sphere radius ratio of d=R ¼ 1 1ðr=RÞ2 ¼ 0:13. In a first step we will explore the force scaling relations, before
determining the fracture toughness.

4.1. Scratch force scaling

The first application is to a series of three scratch tests on Aluminum Alloy 2024 (AA 2024). In each test, the vertical
force, FV, is increased from zero to 200 N generating a variable depth of up to 90 mm along a 6 mm scratch path. The scratch
force, FT, required to move the scratch is recorded. Fig. 3(a) and (b) displays respectively FT and FV in function of d=R, of
three scratch tests. While the vertical force (Fig. 3(b)) displays an almost linear scaling with the scratch depth, the scratch
force (Fig. 3(a)) exhibits an interesting scaling in accordance with the preceding theoretical analysis. In fact, for low depths
ðd=R o 0:1Þ for which the response is representative of the spherical scratch probe, we find a linear scaling of the scratch
386 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

1000
Power (d/R > 0.1)
Linear (d/R < 0.1)
100

FT /N
10

0.1
0.001 0.01 0.1 1
d/R

1000

100
FV /N

10

1
0.001 0.01 0.1 1
d/R
Fig. 3. Forces in three scratch tests on aluminum alloy 2024 (AA 2024) obtained with a Rockwell scratch probe (R ¼ 200 mmÞ: (a) scratch force, F T ,
(b) vertical force, FV, vs. depth-to-radius ratio. In each test, the vertical force is linearly increased along the scratch path of 6 mm to a maximum value of
200 N; the scratch force, FT, and depth, d, are recorded. The grey domain represents the envelope of measured forces in three tests.

force with d, F T  d; in accordance with Eq. (47). In return, for larger depths ðd=R 4 0:1Þ, the horizontal force converges
g
towards a power scaling relation, F T  d , with an exponent g ¼ 1:49 that is not far off the theoretical value of 3=2,
characteristic of a conical scratch probe; see Eq. (33). The scratch force scaling relations confirm relevance of the fracture
analysis in the considered force-depth range.

4.2. Fracture toughness determination

The force scaling relations are indicative of a fracture process during scratching. We thus can explore relations (17) and
(29) to estimate the fracture toughness for AA 2024. Given the agreement of the force scaling for d=R 4 0:1 with the conical
scratch probe scaling,
pffiffiffiffiffiffiffiffiffiffi
ffi we use relation (33) for determining the fracture toughness. This is displayed in Fig. 4(b) in a plot
3=2
F T =ð2d po
sin y=cos yÞ vs. d=R. In return, given the linear relation for lower values of d=R 0:1,
ffiffiffip we
ffi consider scaling relation
ffiffiffiffiffiffiffiffi
(47) of the hemi-spheric probe to determine the fracture toughness in a plot of F eq =ð2d R b=3Þ; with Feq defined by (48);
see Fig. 4(a). In the application of F eq we correct the scratch force FT by the off-set value of the linear fitting relation shown
in Fig. 3(a). The most important result that emerges is that both scaling relations representative of two different probe
geometries converge, within their depth range of application, toward the same fracture toughness value for the tested
A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390 387

60

50

Feq /(4dR1/2/MPa.m1/2)
40

30
ε =2
20

10

0
0 0.1 0.2 0.3 0.4 0.5
d/R

60

50
FT /(2d3/2√sin /cos ) /MPa.m1/2

40

30
ε =1
20

10

0
0 0.1 0.2 0.3 0.4 0.5
d/R

Fig. 4. Fracture toughness determination from the scratch response of AA 2024 with a Rockwell probe using the scaling relation of (a) the spherical
scratch probe for small d=R [Eq. (47)], and (b) the conical scratch probe [Eq. (33)] for greater values of d=R. The grey domain represents the envelope of
measured forces in three tests.

pffiffiffiffiffi
material: K c ¼ 34:4 7 3:0 MPa m. This value is in very close agreement with reported fracture toughness values of
pffiffiffiffiffi
K Ic ¼ 3237 MPa m for AA 2024 determined from classical fracture toughness tests (http://asm.matweb.com/search/
SpecificMaterial.asp?).

4.3. Application to a thermoplastic polymer

We have tested the fracture toughness determination procedure for several materials. By way of further application,
Fig. 5 displays the fracture force scaling relation and the fracture toughness determination for a thermoplastic engineering
polymer, Delrin (Grade 150, Producer: Dupont), manufactured by the polymerization of formaldehyde. In comparison to
the aluminum alloy, Delrin is relatively soft. The results shown in Fig. 5 represent three tests carried out by linearly
increasing the vertical force from 0 to 200 N along a scratch path of 3 mm. While there are too little recorded experimental
values for low d=R ratios to explore the spherical scaling relation, we find that the recorded scratch force converges toward
a power relation with an exponent close to the one representative of the conical probe (Fig. 5(b)). The so determined
pffiffiffiffiffi
fracture toughness of K c ¼ 2:5 70:2 MPa m (Fig. 5(b)) is in excellent agreement with reported values for Delrin obtained
pffiffiffiffiffi
by classical notched beam testing, K Ic ¼ 2:43:4 MPa m (http://www2.dupont.com/Plastics/en_US/assets/downloads/
design/DELDGe.pdf).
388 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

100

Power (d/R > 0.3)


Linear (d/R < 0.2)
10

y = 26.366x

FT /N
R = 0.9344
1

y = 17.099x - 1.6921
0.1 R = 0.819

0.01
0.01 0.1 1
d/R

2.5
FT /(2d3/2√sin /cos ) /MPa.m1/2

1.5
ε =1
1

0.5

0
0 0.2 0.4 0.6 0.8
d/R

Fig. 5. Response and analysis of three scratch tests on a thermoplastic polymer (Delrin 150) carried out with a Rockwell probe: (a) scratch force scaling;
(b) fracture toughness determination. In each test, the vertical force is linearly increased along the scratch path of 3 mm to a maximum value of 200 N;
the scratch force, FT, and depth, d, are recorded. The grey domain represents the envelope of measured forces in three tests.

5. Conclusion

Scratch testing and scratch test analysis continues to gain momentum in Applied Mechanics, due to the possibility
offered by this method to assess material properties at very fine scales. This paper aimed at contributing to these
developments for fracture properties accessible by axisymmetric scratch probes:

1. The proposed LEFM approach allows rationalizing scratch force responses obtained with axisymmetric scratch probes
into distinct force-depth scaling relations. The only input parameter to using these scaling relations is the degree of the
homogeneous function, E. These scaling relations provide a convenient way to check whether fracture is the dominating
mechanism at play in the scratching response, and to determine the fracture toughness.
2. Application of the proposed method to a Rockwell scratch probe illustrates the versatility of the approach. The method
consists of two steps: First, from force-depth scaling relations, one determines whether fracture is the phenomenon
dominating the scratch response, together with the degree of the homogeneous function, E, for the particular scratch
probe. Second, using the derived solutions one determines the fracture toughness. We illustrated this procedure
through application to scratching of an aluminum alloy. The analysis of the experimental force-depth scaling allowed
A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390 389

us to detect two characteristic scaling relations reminiscent of the particular Rockwell probe geometry: for low scratch
depths where the hemispheric shape characterizes the response, the spherical scaling relation ðE ¼ 2Þ is confirmed,
while for greater depths the conical scaling relation is retrieved ðE ¼ 1Þ. Both situations can be accommodated by the
developed LEFM approach. While the particular scaling in both cases is consistent with the fracture model, most
importantly, application of the derived fracture criteria yields for both probe geometries the same fracture toughness;
thus validating the approach. Specifically, for low-depth scratching with the spherical probe, we find an influence of
shear stresses transmitting a part of the applied vertical force via the probe–material interface. In return, for greater
depths and conical probe, these shear stresses appear to be negligible, and the fracture response is driven by axial
stresses in the scratch direction. We confirmed this trend for several materials, and illustrated this further through the
application to a thermoplastic polymer, Delrin.
3. There are some worth mentioning limitations of the approach, which inherently relate to the assumption of a linear
elastic behavior of the scratched solid. One consequence of this assumption is that the approach cannot capture the
effect of large (eventually inelastic) deformation on the stress–strain response that occur ahead of the scratch probe
(Bellemare et al., 2007); thus restricting the domain of application of the model to elastic-brittle materials. Such
deformations may also affect the actual value of the probe–material contact area, as well known in e.g. indentation
analysis (Cheng and Cheng, 2004). A similar remark could be made for time-dependent deformations that give rise to
rate effects affecting the scratch response. It is expected that an extension of the proposed approach to such
deformation—in combination with an appropriate calibration procedure—will remove these restrictions, and make the
scratch test even more appealing for the determination of fracture properties for an even larger range of materials.

Acknowledgments

The authors gratefully acknowledge the financial support of this study by the Total scholarship for graduate studies of
ATA at MIT, and by discretionary funds of the George Macomber Chair for FJU at MIT. We are grateful to NX. Randall (CSM,
Needham, MA) for making available the CSM Micro Combi Testing System for the microscale scratch tests.

Appendix A. Flat punch perimeter function

The perimeter expression (6) is developed for the flat punch ðE-1; B ¼ R1E Þ in the form:
 1=E  1=E
d d
p ¼ lim lim b; lim ¼R ð49Þ
E-1 B E-1 E-1 B

where the dimensionless perimeter function is given by:


vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!2
Z 1u
u  ð1=EÞ
b t d E
¼ lim 1 þ Ed x 1 dx ð50Þ
2 E-1 0 B

We evoke the triangle inequality together with the Euclidean norm:


qffiffiffiffiffiffiffiffiffiffiffiffiffi
8y; 1 þ y2 r1 þ JyJ ð51Þ

Applied to the integrand of Eq. (50) this means:


vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!2
u  ð1=EÞ  ð1=EÞ
u d d
t
1 r 1þ Ed x E 1 r 1þ Ed xE1 ð52Þ
B B

It follows:
b d
1r r1þ ð53Þ
2 R
where the lower bound is relevant for small d=R ratios, while the upper bound for larger depth-to-radius values of the
flat punch.

References

Akono, A.-T., Ulm, F.-J., 2011. Scratch test model for the determination of fracture toughness. Eng. Fract. Mech. 78 (2), 334–342.
Akono, A.-T., Reis, P.M., Ulm, F.-J., 2011. Scratching as a fracture process: from butter to steel. Phys. Rev. Let. 106 (20), 204302.
Atkins, A.G., 2005. Toughness and cutting: a new way of simultaneously determining ductile fracture toughness and strength. Eng. Fract. Mech. 72 (6),
849–860.
Ballarini, R., Shah, S.P., Keer, L.M., 1985. Failure characteristics of short anchor bolts embedded in a brittle material. Proc. R. Soc. London A404, 35–54.
Bard, R., Ulm, F.-J., Scratch hardness strength solutions for cohesive-frictional materials. Int. J. Numer. Anal. Methods Geomech, in press, doi: 10.1002/
nag.1008.
390 A.-T. Akono, F.-J. Ulm / J. Mech. Phys. Solids 60 (2012) 379–390

Bellemare, S., Dao, M., Suresh, S., 2007. The frictional sliding response of elasto-plastic materials in contact with a conical indenter. Int. J. Solids Struct. 44
(6), 1970–1989.
Borodich, F.M., Galanov, B.A., 2002. Self-similar problems of elastic contact for non-convex punches. J. Mech. Phys. Solids 50, 2441–2461.
Borodich, F.M., Keer, L.M., Korach, C.S., 2003. Analytical study of fundamental nanoindentation test relations for indenters of non-ideal shapes.
Nanotechnology 14, 803–808.
Bowden, F.P., Tabor, D., 1964. The friction and Lubrication of Solids. Clarendon Press, Oxford.
Cheng, Y.-T., Cheng, C.-M., 2004. Scaling, dimensional analysis and indentation measurements. Mater. Sci. Eng. R. 44, 91–149.
Karihaloo, B.L., 1996. Pull-out of axisymmetric headed anchors. Mater. Struct. 29, 152–157.
Ma, L., Zhou, J., Lau, A., Low, S., deWit, R., 2002. Self-similarity simplification approaches for the modeling and analysis of Rockwell hardness indentation.
J. Res. Natl Inst. Stand. Technol. 107 (5), 401–412.
Patel, Y., Blackman, B.R.K., Williams, J.G., 2009. Determining fracture toughness from cutting tests on polymers. Eng. Fract. Mech. 76 (18), 2711–2730.
Randall, N.X., Favaro, G., Frankel, C.H., 2001. The effect of intrinsic parameters on the critical load as measured with the scratch test method. Surf. Coat.
Technol. 137 (40212), 146–151.
Rice, J.R., 1968. A path independent integral and the approximate analysis of strain concentration by notches and cracks. J. Appl. Mech. 35, 379–386.
Schei, G., Fjær, E., Detournay, E., Kenter, C.J., Fuh, G.F., Zausa, F., 2000. The Scratch Test: An Attractive Technique for Determining Strength and Elastic
Properties of Sedimentary Rocks. SPE Annual Technical Conference and Exhibition, 1–4 October 2000, Dallas, Texas, paper No. 63255-MS,
doi:10.2118/63255-MS.
See, for instance /http://asm.matweb.com/search/SpecificMaterial.asp?S bassnum ¼ MA2024T4.
See, for instance /http://www2.dupont.com/Plastics/en_US/assets/downloads/design/DELDGe.pdfS.
Ulm, F.-J., James, S., 2011. The scratch test for strength and fracture toughness determination of oil well cements cured at high temperature and pressure.
Cement Concr. Res. 41 (9), 942–946.
Williams, J.A., 1996. Analytical models of scratch hardness. Tribol. Int. 29 (8), 675–694.
Wilshaw, T.R., Rothwell, R., 1971. Instrumented scratch test for measuring fracture behavior of strong solids. Nat. Phys. Sci. 229, 155–157.

You might also like