K. L. Rutherford 1 and L M.

Hutchings 1

Theory and Application of a Micro-Scale Abrasive Wear Test

REFERENCE: Rutherford, K. L. and Hutchings, I. M., "Theory and History of the Method
Application of a Micro-Scale Abrasive Wear Test," Journal
of Testing and Evaluation, JTEVA, Vol. 25, No. 2, March 1997,
pp. 250-260. One of the first reports of a scientific abrasion test was by
Mikhail Lomonosov, who was a member of the Moscow Academy
ABSTRACT: A micro-scale abrasion test has recently been developed of Sciences between 1745 and 1765 [1]. The test consisted of a
that allows measurement of the wear resistance of the surface regions grindstone, 0.46 m in diameter, that was turned by a hand crank
of a material. The typical penetration depth is less than 30 Ixm. The on a horizontal axle. The test was used to study the abrasion of
test uses a simple mechanical and optical system and involves rotating
a hard steel sphere against a specimen in the presence of small abrasive metals, minerals, and glasses that were held in a lateral wooden
particles. The method has been used to investigate the wear resistance arm and pressed against the top of the stone rotating under a dead
of thin PVD coatings (1 to 5 ~m), metallic glass ribbons, and paint load weight. This type of wheel abrasion test was refined in 1921
films in addition to bulk samples of metals, ceramics, and glasses. The by Brinell 121, who replaced the stone with a 100-mm-diameter
associated theory has been extended so that results may be obtained
from any curved surface. This furthers its applicability to practical steel wheel and fed quartz particles into the contact with the
surface-engineered components such as twist drills, bearings, turbine specimen. The next most significant change was by Haworth 13]
blades, and biomedical prostheses. A detailed characterization of the who identified a number of advantages in reducing the forces at
experimental procedure has been undertaken to provide an understand- the specimen. He achieved this by applying a rubber rim to the
ing of the repeatability and sensitivity of the test. In a study of cutting- wheel to produce the first rubber wheel abrasion test in 1949.
tt×)l coatings, the wear resistance measured by this method has been
shown to correlate with scratch test response and with the performance This was also the year in which the first micro abrasion test
of coated end-mills in cutting tests. was described by Grodzinski [41. His apparatus consisted of a t
inch (25 mm) diameter cast iron wheel of a double conical shape
KEYWORDS: abrasion, wear test, coatings, surface engineering and an included angle of 110°. Abrasive particles (0 to 2 l~m
diamond) were fed into the contact and the wheel was pressed
against a specimen under a load of 20 to 500 g three and at a
Surface engineering is consistently identified as offering the
rotational speed of up to 20 000 rpm. The resulting wear crater
potential for major improvements in manufacturing technology.
was then measured by the multiple beam interferometry techniques
However, for widespread introduction into industrial processes,
of Tolansky and Austin [5]. This test was used by Grodzinski and
robust test methods are required that can assess the performance of
surface-engineered components. The current lack of such methods Stern [6] and Wilks [7] to investigate anisotropy in the wear
results from the problems encountered in devising wear tests that resistance of diamond surfaces. These authors pointed out that the
do not rely upon measurements of very small mass losses or test permitted the examination of small material volumes and
dimensional changes. One of the most widely used tests for the of small but hard samples. Problems were, however, caused by
quality assurance of thin films is the scratch test, but even this is eccentricity in the rotation of the wheel.
generally acknowledged to provide poor simulation of most service In 1956 Happ and Shockley [8] used the same principle, with
conditions since the test involves severe plastic deformation of the controlled rotation of a cylinder, to generate wear craters to
the substrate. The lack of suitable methods has meant that current profile the depth of diffusion layers in silicon. This technique was
understanding of the wear behavior of surfaces is relatively poor. refined in 1962 by McDonald and Goetzberger [9], who also
This paper provides a summary of the current theory and practi- showed how the thickness of a thin coating could be derived from
cal understanding of the ball-cratering method used as a small- optical measurements of the wear scar. A spherical counterbody
scale abrasive wear test. The history of micro-scale abrasion testing was first used by Thompson, Hintermann, and Chollet [10]. They
is reviewed and the current capabilities of the ball-cratering test are used a sphere coated with a standard diamond lapping paste to
summarized. These include the ability to test very small samples, prepare specimens for depth profiling by scanning Auger Electron
its applicability to specimens with compound curvature, and the Spectroscopy. In this geometry the worn region takes the form
capacity to determine the properties of thin coatings independently of a spherical cap. A more detailed study of ball cratering was
of their thickness and adhesive strength. These aspects are then undertaken in 1985 by Valli et al. [1i] who determined the optimum
illustrated in the context of four case studies. test conditions for accurate coating thickness measurement. The
rotating sphere apparatus is now commercially available and rou-
tinely used for coating thickness determination; an ASTM standard
Manuscript received 5/23/96; accepted for publication 12/18/96.
~Universityof Cambridge, Department of Materials Science and Metal- test procedure is available for making the measurement of coating
lurgy, Pembroke Street, Cambridge CB2 3QZ, United Kingdom. thickness by radial sectioning (El 182-87).
© 1997 by the American Society for Testing and Materials
250
Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
 RUTHERFORD AND HUTCHINGS ON MICRO-SCALE TESTING 251

In 1991 Kassman et al. [12] made a significant development in where S is the total distance slid by the sphere relative to the
the wear testing of thin coatings by using a dimple grinder with specimen surface, N is the normal force between sphere and sample,
a crowned cylindrical wheel, originally designed for thinning trans- V is the wear volume, K is the wear coefficient, R is the radius of
mission electron microscope specimens. These authors modeled the sphere, and b is the external diameter of the wear crater, as
the combined wear behavior of a thin coating and its substrate illustrated in Fig. la. The penetration depth h, for a particular
using an extended form of the Archard wear equation. This enabled, wear crater diameter, is determined by the radius of the ball.
for the first time, independent wear coefficients to be measured
b2
for both a coating and its substrate. They demonstrated the capabili- h "~ - - (2)
ties of the test on 4-1xm titanium nitride and 7-txm titanium carbide 8R
coatings on high speed steel and cemented carbide substrates Figure 2 shows Eq 2 plotted for three different values of the ball
respectively, radius, R. This illustrates how, with a ball of radius 12.7 mm, the
In 1993 Nothnagel [13] applied the same basic principle to test can be used to examine the behavior of only the upper 30 Ixm
sliding wear by a rotating sphere but did not consider the combined of material from measurement of a wear crater with a diameter
wear of a coating and substrate. In 1994 Pensaert et al. [14] of 1 to 2 mm.
presented a calculation and experimental results for a crossed Equation 1 may be extended to model the combined wear of a
cylinder test for the wear resistance of coatings on thin wires. In coating and substrate, each with an independent wear coefficient
1995 Silva et al. [15] returned to using a crowned steel disc in an [12]. The extended equation is most simply expressed as:
investigation of dry sliding against 25 to 260 nm diamond-like
carbon (DLC) films on silicon. These authors used profilometry
traces to rank the relative wear performance of different films with
reference to uncoated silicon. In 1996 the present authors [16]
used a ball-cratering apparatus and extended the work of Kassman where the suffices c and s refer to the coating and substrate,
et al. [12] by reformulating the extended Archard equation so that respectively. The wear volumes Vc and V~ may be calculated from
the independent wear coefficients could be calculated from the the diameters a and b of the imposed wear craters (as shown in
combined wear behavior of a coating and its substrate in a single Fig. lb).
test, The capabilities of the test were later extended in order to
study surfaces with general compound curvature, and the test has Analysis of Experimental Data
been applied to a wide range of thin coatings and small samples
[17-19]. Analysis of the experimental measurements of wear crater diam-
G~thlin et al. [20] have recently investigated the effect of a eter b versus sliding distance S for bulk homogeneous materials
number of different test parameters on the test method employing is trivial since Eq 1 can be used to produce a linear plot (of ('rrb4/
a crowned disc. The wear craters generated in this type of apparatus 64R) against SN) the gradient of which is determined by the wear
can deviate from the idealized geometry of a spherical cap leading coefficient t<. Analysis of the data for a coated specimen is more
to large errors in the measured wear resistance of a material. complex. One method involves fitting an expression, equivalent
Coating thickness measurements derived from such apparatus are to Eq 3, to a curve in order to derive values for the two free
therefore prone to significant errors. The likely reasons for the variables that represent the wear coefficients of the coating (Kc)
imperfect geometry are machining inaccuracies in the disks and and substrate (Ks). However, Kassman et al. [12] who first
misalignment of the axes of rotation of the specimen and the employed this method concluded that the errors in Kc determined
sample. It is interesting to note that these problems are similar to in this way were reduced if K, was measured independently in a
those encountered in the first microabrasion test of 1949 [4]. Ghhlin separate test.
et al. have proposed a revised method of analysis based upon a This analysis was later refined by Hedenqvist et al. [21]. They
"misfit" parameter to correct for the geometrical inaccuracy; the again used an independent measurement of K, but rearranged the
method requires profilometry measurements to be made at the end expression to generate a linear graph, the gradient of which repre-
of each test. sented the wear coefficient of the coating:
There are therefore strong advantages in using a ball-cratering
geometry that eliminates the problems of misalignment and uses ( S N - V~.)= KcVc (4)
spberical balls that are routinely manufactured to very precise
dimensional tolerances.
It is possible, however, to rearrange the imposed wear crater for-
mula in two different ways and to derive both K~ and K, simultane-
Theoretical Basis ously, also by plotting a linear function:
The rotation of a sphere against a specimen in the presence of
small abrasive particles generates a wear crater with an imposed = + Vr (5a)
spherical geometry within the material. Figure 1 shows optical L\ K,K,, /
micrographs of a selection of wear craters in uncoated and coated
plane and cylindrical samples. For homogeneous bulk materials,
it can readily be shown that a simple model for abrasive wear, = + vr (5b)
Lk K,K~ /
completely equivalent to the Archard equation for sliding wear,
leads to: The wear volume terms can be defined in terms of the coating
thickness t and either the wear crater diameter b or the penetration
SN = V l (Trb4~ depth h. If the appropriate expressions are substituted into Eqs 5a
--£ ~ -£ \64R] for b << R (1)
and 5b, a simple rearrangement gives equations that can be used

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
252 JOURNAL OF TESTING AND EVALUATION

FIG. l--Optical micrographs of imposed wear craters in (a) zirconia: (b) 3.8 ~m TiZrN coated ASP23 tool steel; and (c) 4.1 p~m 7~ZrN coated
cylindrical (radius 6.35 ram) M2 tool steel

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
 RUTHERFORD AND HUTCHINGS ON MICRO-SCALE TESTING 253

expression for the abscissa term in the second equation of each

~.~ 40 . set is equal to 1 minus the abscissa term in the first equation.
" Hence, the condition under which Eq 5a leads to a smaller error
~- 30
compared to Eq 5b, is when the abscissa values are to the left of
the midpoint of the abscissa coordinate, in other words, when Eq
2o 8 is satisfied for the smallest wear crater b.

(16Rtb2~64RZt2)<0.5 (8)
g. i0

0 The same argument means that Eq 7a leads to less uncertainty

0.4 0.8 1.2 1.6 than Eq 7b when, for the smallest value of h:
Wear crater diameter [mm]
FIG. 2--Relationship between the depth of penetration h and the diame-
ter of the wear crater b fi~r three ball radii R (Eq 2).

Equations 8 and 9 are plotted in Figs. 3a and b. These show the

to generate straight line graphs for which the intercept and gradient combinations of coating thickness and crater diameters for which
are determined by the values of Kc and Ks. For example, for a flat the different equations lead to the smallest uncertainty. In short,
sample where the equations are expressed in terms of the outer for coatings < 2 p~m thick, Eqs 6a and 7a lead to the smallest
wear crater diameter b: uncertainty while for coatings > 7 txm thick Eqs 6b and 7b tend
to be most suitable. Where the range o f measured crater diameters/
SA64R] 64RZtZ) (l'~q penetration depths overlap the two regions, the equation corres-
""t b'l = i_t )t V 7 ti_lZ + (6a) ponding to the area in which most of the data lie should be used.
However, all the methods of analysis described above can be
applied only to the case of a uniform thin coating on a plane
N(64R) = [(K~
K --
- "l(l_16Rt+64R2t2~ substrate. This restriction limits the applicability of the method,
S -~ L\ KsKc 1\ b2 b4 ) since many coated components such as twist drills, turbine blades,

+ (6b) 2
=

or in terms of the penetration depth h: 1.6 Use equatio

1.2

i 0.8

"rrRh2 L\ KsKc 1\ Z (7b)

0.4
This method of data analysis offers significant advantage over
separate evaluation of Kc and K,. It ensures that the value of K~ is 0
truly representative of the substrate beneath the coating at the point 0 2 4 6 8 10
of measurement. It also avoids the need to remove a separate area Coating thickness [gm]
(a)
of the coating to determine K, thereby minimizing the extent of
damage associated with the testing, and furthermore halves the
number of separate tests required for the determination of the
coating wear coefficient Kc. 35 i
The uncertainty in K,. and K.~ depends upon which of the two _ 30
Eqs 5a or 5b is used. Their relative suitability depends upon the ~_ 25 Use equatio
coating thickness, the size of the wear craters, and whether the
crater diameter b or penetration depth h is measured. The suitability .=o 20
can be assessed by studying the relative uncertainties in the inter-
cept and gradient.
In both sets of Eq 6a,b and 7a,b, the ordinate term is identical g- 10
and the absolute error in the value of the abscissa is the same.
The uncertainty in the gradient is therefore the same for both
0
equations in each set. The relative uncertainty in the wear coeffi- 0 2 4 6 8 10
cients is therefore controlled by the position of the experimental Coatingthickness[lain]
points along the abscissa coordinate since this controls the magni- (b)
tude of the uncertainty in the intercept for a particular error in the
FIG. 3--(a) Equations 6 and (b) Equations 7 plotted for typical values
gradient. In the two sets of Eqs 6 and 7 the value of the abscissa of the wear crater diameter and penetration depth to show the relative
is always greater than zero and less than unity. Furthermore, the suitability of the different methods of analysis.

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
254 JOURNAL OF TESTING AND EVALUATION

slurry
and biomedical prostheses have curved surfaces and it is much feed
less satisfactory to perform the test on separately-coated plane
coupons rather than on the actual shaped components. specimen i
A test performed on curved components will yield an elliptical
wear crater in which its dimensions along the major and minor
axes can be defined as b and a, respectively, as shown in Fig. l c.
Hence, for the intersection of a sphere (radius R) with a general
surface with compound curvature (r~, rb), the total volume Vr
of the imposed wear crater must be expressed in the form of FRONT
VIEW
an integral:
slurry feed
Vr = T dhl + ~ dh2 (10)

where a/b = x/r,(R + rh)/rb(R + r,,), b: = 8Rhj = 8h2rh and hi,

I
h2 are the penetration depths of the ball into the specimen and the
specimen into the ball respectively as defined in Ref 18.
If the maximum penetration depth h and the coating thickness drive
t are much smaller than any of the radii of curvature R, r, and rb, shaft
then to a good approximation:
Trb4(ra(R-t-r,,)~'/2(l l) SIDE
VIEW
Vr ~ --~ \r,(R + ~,,)1 \-R + ~ (1 1)
FIG. 4---Themicro-scaleabrasiontest.
"rrb'4(r.(R+rb)ll/2(I
Vs "~ - - ~ \r,(R + ~.)] ~-R + ~
l) (12)
The relative sliding speed was fixed at 0.05 m s- t and the normal
force on the sample was between 0.05 and 0.22 N, depending
Vc = V T - Vs (13) upon the sample, and was measured by a sensitive load cell. An
alternative method by which this force can be calculated from the
where b' is the inner major axis of the ellipse formed by the geometry of the test apparatus was described previously 116]. The
intersection of the counterbody with the substrate, as shown in test is interrupted periodically and the size of the resulting wear
Fig. Ic. The approximation will always give an overestimate of crater is measured with a calibrated optical microscope.
the true volume but the error is less than 0.1% for typical values
of b, r~. and R.
Test Sensitivity
Equations 11-13 allow the volume terms in Eq 3 to be calculated
by measurement of the major axis of the elliptical wear scar. Hence, A good understanding of the effects of variations in all the test
provided that at least one of the principal radii of curvature of the variables is fundamental in establishing the robustness of this
specimen is known, the volume of the intersection between the abrasion test. Three of the key parameters are the concentration
sphere and the surface may be calculated from Eq 11. Indeed, the of the abrasive slurry, the sliding speed of the ball, and the effect
local radii of curvature of the specimen may be determined by of interrupting the test to measure the size of the wear scar. Figure
performing two micro-scale abrasion tests with spheres of different 5 shows the effects of these parameters in controlled tests on
radius R. The different aspect ratios derived from the two tests polished copper and soda-lime glass specimens.
would allow values of both radii of curvature r, and rb tO be It was found that the measured wear behavior was effectively
calculated by solving two simultaneous equations based on the independent of variations in the slurry concentration (Fig. 5a) and
elliptical ratio as defined for Eq 10. in the number of interruptions (Fig. 5b). The first result can be
This relatively simple method facilitates calculation of the vol- explained if it is appreciated that only a small fraction of the very
ume of intersection between any two bodies with known local large number of abrasive particles present are dragged through the
radii of curvature. For example, it can be shown to give the same contact. For example, 1 g of the abrasive contains between 109
result for the intersection of crossed cylinders as that resulting and 10 m individual particles. Hence, provided a minimum presence
from a lengthy calculation published by Pensaert et al. [14,22] in of particles is maintained near to the ball, a steady-state condition
their recent study of the abrasive wear resistance of wires. will be achieved at the specimen. The insensitivity to the number
of interruptions confirms that the test apparatus allows accurate
and consistent repositioning of the ball. A similar result was shown
Test Procedure
previously for the abrasion test based on the dimple grinder appara-
The principle of the micro-scale abrasion test is illustrated in tus [20].
Fig. 4; a hard steel sphere is rotated against the test specimen in Figure 5c shows that although the wear resistance of the soda-
the presence of a pool of slurry of small abrasive particles that is lime glass and copper were largely independent of the sliding
maintained and replenished at the contact region by a slow drip speed, the wear rate of the glass did increase significantly at speeds
feed. In this work a 25.4-mm-diameter sphere was used under a below 0.02 m s-~. This may be associated with the specific wear
controlled contact toad derived only from its weight. The abrasive mechanism of this material. For example, a stress-corrosion pro-
slurry was an aqueous suspension of small (4.5 Ixm) silicon carbide cess, to which inorganic glasses are known to be susceptible in
abrasive particles with an initial concentration of 0.75 g cm -3. aqueous environments, might become more significant under these

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
 RUTHERFORD AND HUTCHINGS ON MICRO-SCALE TESTING 255

20
only the scale of the deformation being influenced by the hardness
Z of the surface. Three typical worn surfaces are shown in Fig. 6.
% 16
These indicate a three-body abrasive wear mechanism involving
e,l
Soda time glass
plastic deformation, with the particles rotating within the contact,
12
x
resulting in multiple indentation and small-scale cutting and
ploughing on a scale comparable with the particle size. Long
8
grooves in the direction of sliding were not seen on either the test
o
sample or the spherical counterbody, indicating that the abrasive
4
~ ~ } ¢ ~ Copper particles had not become embedded in the wearing surfaces.
For soda-lime glass (Fig. 6c) the nature of the worn surface
0
0.3 06 0.9 L2 1.5 l.g 2.1
suggested that the action of these small hard abrasive particles
Slurry concentration [g ( S I C ) c m ' Z ( H 2 0 ) ]
was the same as that described above; however, the wear mecha-
nism also involved small-scale brittle fracture in addition to some
20 ' i ' I ' i ' I • I ' I ' i • I ' I ' !
plastic cutting. This suggests that, under a total normal force of
0.20 N, the load on each abrasive particle is close to the critical
~z
% 16 value below which brittle fracture will not occur [23].
In all cases, the wear mechanism was the same over the total
12 area of the wear crater. Furthermore, no directionality was apparent
E
in the surface damage despite the rotation of the sphere being
8 about a single axis parallel to the drive shaft.
The analysis of the rotating sphere test used above relies upon
4 a spherical wear crater geometry being maintained throughout the
Co/~ino
ntcff~.tplcd
sliding sliding
test. Experimental evidence for the imposed crater geometry and
0
I , ! • I • I * I . I , I • I , I • I • the negligible effect of ball wear is shown in Fig. 7. This shows
0.2 0.4 0.6 08 1.0 stylus profilometry traces for a virgin ball, a used ball, and the
Irderval fraction
shape of a wear crater in a thinly coated material. Figure 7b
30 ' ' ' 1 ' ' " 1 ' ' ' 1 ' ' ' 1 ' ' '
confirms that the geometry closely conforms to the ideal shape of
the sphere and that the large deviations from a spherical geometry,
'z 25 highlighted by Ghhlin [20], for wear craters produced using a
%
crowned disc are eliminated by the use of the ball-cratering
,Q 2o
Soda-lime glass geometry.

Case Studies
~ '°
The micro-scale abrasion test has now been applied to a wide
Ii m •
Copper range of thin coatings and bulk materials. Several material systems
0
are discussed below in the context of four case studies to illustrate
0 0.02 0.04 0.06 O.0g 0. I the wide applicability of the test.
Sliding speed [ m s "1 ]

FIG. 5--Variation in the wear coefficient of soda-lime glass and copper

as a function of the (a) concentration of the abrasive slurry; (b) fraction Metallic Glass Ribbons
of the total test time at which sliding was interrupted; (c) sliding speed
of the ball. Amorphous metals (metallic glasses) can offer novel properties
such as superplasticity and high magnetic coercivity. Hence,
although the development of these materials is relatively recent
conditions and would explain a dependence on speed of the form they are already used in technological components such as mag-
observed. In future investigations it is therefore important to recog- netic sensors, read-write heads, and valves where a knowledge of
nize that the sliding speed might affect the measured wear rate their mechanical properties and wear behavior is important. The
and to control and report this aspect of the test. rapid cooling rates essential for the preparation of these materials
In summary, the micro-scale abrasion test has been found to be means that samples are invariably small in at least one dimension.
largely insensitive to three of the key test variables. Furthermore, For example, one of the most efficient ways of producing metallic
the repeatability of the measured wear resistance of identical steel glasses is by the process of melt-spinning in which a thin ribbon,
and ceramic bulk specimens was consistently better than 7% and a few tens of micro-meters thick, is produced [24], The micro-
for thin ceramic coatings it was better than 10%. These results scale abrasion test can be used to measure the wear resistance of
provide confidence in the robustness of the test. This behavior is these materials, to which more conventional abrasive wear tests
quite typical of three-body abrasion tests because of the large can not be applied.
number of individual particle interactions that contribute to the Amorphous aluminum alloy ribbons (2 mm by 5 mm by 40
wear process. p~m) with the composition AlsvNit0Ce3 were prepared by a melt
spinning technique. Systematic variation of the mechanical proper-
ties and micro-structure of the ribbon was then achieved by con-
Wear Mechanism
trolled devitrification of small samples in a differential scanning
SEM examination of worn samples shows the wear mechanisms calorimeter (DSC). A single length of ribbon was divided into 5-
of thin PVD coatings and metallic samples to be identical, with mm lengths which were then subjected to different heat treatments.

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
256 JOURNAL OF TESTING AND EVALUATION

FIG. 6--SEM images of the worn surfaces of (a) amorphous A187NiloCe3; and (b) PVD Ti/TIN coated ASP23 tool steel," (c) soda-lime glass.

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
 RUTHERFORD AND HUTCHtNGS ON MICRO-SCALE TESTING 257

20 ,i , ,, l , , , i , , ~l ,, , l~ ¢"~1 , ~, l ,
~ Theoretical 7
16
E

o:
12 ° i
AI1111~
t~
8
"6
£
FOLL raANS-
8 4 Foa~ArloN I I I I I I -- I

20 30 40 50 60 70 80 90

0 Angle 20
-0.8 -0.6 -0,4 -0.2 0 0.2 0.4 0.6 0.8 FIG. 8--X-ray diffraction traces for a series o f progressively devitrified
metallic glass (Als7NitoCe3). X-ray diffraction was performed using a
Distancefrom centre [ram] Phitips 1050 vertical diffractometer with Cu Kct radiation. The samples
(a) were mounted on a silicon substrate that was oriented to give a single
(400) refiection.
2
4 ' , • i ' ' • ~ ' ' • i , ' ' , ~ 600
_= B o u n d a r y of Boundaryof
1.5 wear crater wear crater
3 W e a r resistance 1

. 2.5 , ~ 5o0

: 450
0.5 ca 1.5 "I
400 9 <

.i~ - Wear crater • .i: ' ",.' 350

0.5
o
0 , , i , , , i . . . I . . . I . , 300

8 20 40 60 80 100
~- -0.5 ,,I , , , I , , , I , , , l , , , l , , l l i i ~ [ , = % transformed
-o.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 FIG. 9--Micro-scale abrasion and micro-hardness measurements as a
Distance from centre [ram] function o f the extent o f devitrification o f a metallic glass with composition
The error bars represent the spread of 2 or 3 repeated tests
A 1 8 7 N i l o C e 3 .

(b) on nominally identical samples.

FIG, 7--(a) Measured surface profiles o f a virgin ball, a worn ball and
a wear crater in a 3.2 ~m thick 7iN coating on a ASP23 tool steel substrate
and comparison with the theoretical spherical geometry; (b) deviation of lized before rising to a peak when 90% of the transformation
the measured profile for the wear crater from the perfect geometry o f a was completed.
sphere with 12.7-mm radius,
The deformation in both the hardness indentation and the abra-
sion tests is thought to be inhomogeneous, involving shear on
highly localized bands. As devitrification proceeds, and the number
All the samples were heated at a constant rate of 10°(2 min-~ to of nano-crystalline precipitates increases, the likelihood of their
final temperatures between 160 and 300°C. This final temperature interaction with these shear bands increases. At the same time,
range achieves different extents of a transformation in which ccp- the concentration of solute in the amorphous matrix increases
AI is precipitated in the amorphous matrix. All of the wear tests and, for similar materials, this has been shown to influence the
were performed from the same side of the ribbon samples (the side mechanical properties separately from any shear band interaction
that did not make contact with the wheel during melt-spinning). effects [26]. Hence, the increased hardness and wear resistance
The samples were also characterized by Vickers micro-hardness might be explained by a combination of these effects as the number
measurements and X-ray diffraction (XRD) with a vertical diffrac- of precipitates increases during devitrification.
tometer (Philips 1050 with Cu-Kc(radiation). This does not, however, explain why the hardness increased
The micro-scale abrasion test allowed wear tests to be performed linearly while the wear resistance showed a more complicated
on the ribbon samples with dimensions smaller than 2 mm X 5 mm dependence on the extent of devitrification. A possible explanation
X 40 ~m. Figure 8 shows XRD traces illustrating the progressive is provided by a size effect. On the scale of a hardness test, a
nucleation and growth of nanocrystalline ccp-aluminum precipi- systematic increase in the concentration of precipitates would be
tates within the metallic glass. Figure 9 shows the hardness and expected to result in a corresponding decrease in the ease of large-
abrasive wear resistance plotted as a function of the extent of the scale shear band formation and hence lead to decreased ductility
transformation. The error bars represent the repeatability of each and increased hardness. The linear relationship observed is consis-
test, with each result based on a minimum of two tests on the tent with trends reported previously [27]. However, a crucial differ-
same specimen. ence in the deformation induced during the abrasion test is that
The precipitation of nanocrystalline aluminum by controlled the scale of each abrasive scratch is smaller by an order of magni-
devitrification resulted in a significant increase in both the hardness tude. This leads to much higher strain rates and means that the
of the samples and in their measured wear resistance. However, total volume of material involved in each elementary abrasion
while the hardness increased approximately linearly, the wear resis- interaction is considerably smaller. Under these conditions a large
tance was relatively constant until 60% of the material had crystal- number of very short shear bands will thus be required to accommo-

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
258 JOURNAL OF TESTING AND EVALUATION

100 . . . . . . . . ! " ....... I . . . . . . .

date high plastic strains. A relatively low concentration of precipi-
tates might therefore be expected to have little effect, since the
proportion of the shear bands affected would be small and they "FLEXIBLE"
CLEARCOATS
would tend to operate in the regions between precipitates. This
may therefore explain the relatively constant wear rate over the ~ E i0
initial stages of devitrification. However, as the concentration of
the particles begins to saturate the amorphous matrix, and their
spacing becomes comparable with the length and dilated width of CLEARCOATS

the shear bands, they would be expected to have a detectable effect,

....... I . . . . . . . . I . . . . . . .
leading to the progressive rise in wear resistance after ~ 6 0 % of 10 100 1000
the transformation. The subsequent peak and decrease in the wear Energy to failure [mJ turn -2]
resistance may be accounted for by the precipitates either sup- FIG. 10--Ratner-Lancaster plot of the micro-scale abrasion resistance
pressing shear to such an extent that an increased fraction of versus the energy for tensile failure of five commercial automobile
the abrasive scratch volume is removed in cutting or that the "'clearcoat" polymer coatings.
precipitates, acting as stress raisers, promote cracking and the
removal of material.
Further measurements of the hardness and wear resistance of deposited (PVD) coatings. Three tests were used in the investiga-
fully amorphous Al~NiyCe~ alloys, as a function of composition, tion: micro-scale abrasion, single-point scratching, and accelerated
are needed to identify the relative influence of shear band, chemi- milling. The coatings were all deposited, by standard commercial
cal, and size effects in these results. procedures, on M2 high speed steel (HSS) substrates. Two types
of substrate were used: square sectioned bars 6.35 by 6.35 by 100
mm, and 12.7-mm-diameter end mills. In each case, both types
Polymeric Paint Coatings
of substrate were coated in the same deposition run. The samples
Automobile paint coatings are typically composed of up to included two titanium nitride (TIN) coatings, produced by reactive
seven distinct layers. There is a general need for an improved evaporation and by sputtering. The evaporated TiN coating was
understanding of the factors controlling both their individual wear also deposited onto two HSS substrates that had undergone differ-
resistance and their collective durability. However, the industry ent heat treatments prior to coating, intended to produce different
lacks quantitative tests and it has even been claimed that appro- interfacial oxide layers to modify the interfacial adhesion proper-
priate testing and test methods are now a more important part of ties. The other four materials were second-generation coatings
coatings technology than resin and paint formulations [28]. of titanium carbo-nitride (Ti(C,N)), chromium nitride (CrN), and
Each layer plays a particular role, for example, in corrosion titanium aluminum nitride ((Ti,AI)N). The Ti(C,N) coatings were,
protection, ultra violet absorption, adhesion promotion, and aes- like the TiN, deposited by both evaporation and sputtering
thetic qualities, but the uppermost layer is invariably a --50-~m techniques.
thick transparent coating known as the "clear coat." A major func- Figure 11 shows that the micro-scale abrasion test discriminated
tion of this is to provide resistance to mechanical and corrosive clearly between the different coatings and yielded a statistically
damage. The capability of the micro-scale abrasion test to examine consistent value of wear coefficient Ks for the substrates (mean
only the uppermost 30 Ixm of a surface enables the abrasive wear value = 0.94 X 10 - l / m z N -I, standard deviation = 0.05 X
performance of this layer to be measured independently of the 10 -J2 m z N-l). The standard deviation is comparable with the
underlying material. A large number of different formulations of 7% repeatability of the method for replicated tests performed on
clear coat materials exist commercially and their properties are polished steel. These results confirm that the test method can be
normally optimized for different levels of substrate flexibility (such used to measure independent wear coefficients for both coating
as metallic or polymeric substrates). and substrate from a single test. It is also noteworthy that this test
Micro-scale abrasion tests have been performed on five different method, expected to sample the intrinsic abrasion resistance of a
clear coat materials, 20 to 40 I~m thick, on polymeric substrates.
Three of the coatings had been formulated for use on flexible
substrates, and two for rigid materials. All of the coatings had 20
glass transition temperatures in the range 38 to 57°C. Separate test
samples of the coating films were prepared on substrates specially E
treated so that the coating could be easily removed, to allow the ._o 15
:1=
coating material alone to be tested in tension.
The micro-scale abrasion test was found to discriminate strongly ~ Z

between the coatings designed for flexible and rigid substrates; lO

the clear-coats that had been formulated for use on polymeric f,=
substrates were consistently more wear-resistant than those
designed for steel. Furthermore, Fig. 10 shows a correlation similar
to the Ratner-Lancaster correlation [29] for bulk polymers between 9 5
the wear resistance of the coatings in the abrasion tests and the
energy required to break the coatings in the tensile tests.
0
I' I I I
Thin Hard PVD Coatings TiN TiN TiCN CFN TiAIN
evapo~ted spullece~J ev~por=ted sl~ttete0
Micro-scale abrasion tests have been performed during the wider FIG. l 1--The micro-scale abrasion resistance of a series of thin PVD
tribological characterization of six different physically vapor coatings on identical M2 tool steel substrates,

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
 RUTHERFORD AND HUTCHINGS ON MICRO-SCALE TESTING 259

coating independently of its adhesion to the substrate, did indeed Hard Coatings on Curved Substrates
yield similar wear coefficients for the three nominally identical
An important practical use of thin coatings is in the surface
TiN coatings on the substrates with differently treated interfaces.
engineering of drill and milling cutters, the bodies of which often
Valuable comparisons can be made between the results of the
have a cylindrical geometry. To illustrate the use of the extended
micro-scale abrasion test, the scratch test, and the performance of
model for curved surfaces, the equations given in section 3 have
the coatings during milling. Figure 12a shows the results of the
been solved for the specific case of a coated cylindrical specimen.
scratch test and the micro-scale abrasion test. The correlation
For the case of a sphere (radius R) interacting with a cylinder
between these results suggests that the two tests measure similar
coating properties. The micro-scale abrasion test may therefore
(ra = r, r b = ~ ) Eqs 11 and 12 simplify to give:
have a role in both quality assurance testing and the evaluation of ~b4 { r ~1,2
coating durability, purposes for which the scratch test is extensively v~ = ~ \2-~-;1 (14)
used. Indeed, the excellent repeatability of the micro-scale abrasion
test coupled with its applicability to very small specimens suggest
that it may be superior to the scratch test for some types of
Vs = ~ \~--~--;r/ (15)
assessment.
Figure 12b compares the micro-scale abrasion wear coefficients
with the flank wear rate of the coated end mills. For these coatings, In terms of the quantities measured in a micro-scale abrasion test:
performance in the micro-scale abrasion test correlates with the SN {R + r~ t/2 = (K, - Kc~{ ~rt 7rRtZ~
relative wear behavior of the coated tools in cutting tests. This -b4 ~ ' - - ~ ) \ K, Kc )\4b 2 b4 )
suggests that the factor that determined the life of the milling
cutters in these tests was the abrasive wear of the tool flank. A
similar result has been shown by Santner and Meier zu K0cker (16)
[30] who compared wear in a pin-on-disc tribo-simulation test
with the life of coated milling cutter inserts. They used similar which can be written as:
coatings to those used in the present work, and found a similar
ranking of abrasive wear performance. y = mx + c (17)
These results demonstrate the potential of the micro-abrasion where b is the major diameter of the elliptical wear scar, t is the
test as a simple low-cost characterization technique with the ability coating thickness, R is the radius of the rotating sphere, and r is
to predict the performance of coatings under service conditions. the radius of the cylinder.
The use of the test to pre-select coatings prior to in-service trials Three different coatings of TiNbN, TiZrN, and ZrNbN were
could lead to significant financial savings. deposited via a commercial PVD (reactive arc evaporation) tech-
nology on three different substrate materials. These were plane-
7O0O polished samples of ASP23 powder metallurgical tool steel and
TiCN
600o AISI 304 austenitic stainless steel, and cylindrical M2 tool steel
drill blanks (radius 6.35 mm) that had been ground, hardened, and
5000 tempered prior to coating. Each coating was deposited on the
TL~IN TLN
40o0 TiCN different substrates in the same deposition cycle. The coatings
were nominally 4 to 5 ~m thick.
._o 3000 CrN
• TiN The wear resistances measured for the coatings on each of the
"I-tN eTi N substrates are shown in Fig. 13. Typical wear craters in the TiZrN
~ lol)o
coated flat and cylindrical samples are shown in Figs. lb and lc.
Wear resistance
High ~ ~ Low The wear resistance of each coating and substrate was found to
, = , I , , , I , , , I , , , I , , ,
o be consistent and was independent of the other component of the
4 8 12 16 20
Micro-abrasion wear coefficient [x 10 "13 m 2 N-t]
coated system. Comparison of the results from the tests performed

40 ' ' ' I ' ' ' I ' ' ' I , , ,

TiN •
35 • M2 (drill)
~ 30 [] ASP23
8 1.5 [] AISI 304
~ 25
~ 2o TiN TiCN

TiAIN

0.5
TiCN Wear resistance
5 E zi E
• High ,,~ ~ Low

i i | i , • ! • , * | * , , i i ! i
c)
4 8 12 16 20
TiZrN NbriN ZrNbN
Micro-abrasion wear coefficient ~c [x 10"13 m 2 N-1 ] Coating material
FIG. 12--Comparison between (a) scratch test failure loads (63.2% FIG. 1 3 - - M i c r o - s c a l e abrasion wear resistance o f three PVD coatings
cumulative failure probability); and (b) the rate of flank wear o f coated on different substrates. In each case the coating thickness is shown. The
end mills in accelerated milling tests and the micro-scale abrasion wear M2 drill samples were cylindrical (6.35-mm radius) while the other samples
coefficients for a range of PVD coatings. were fiat plates.

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.
260 JOURNAL OF TESTING AND EVALUATION

on the cylindrical drill blanks with the results from the same Using Cylindrical Grooves," Bulletin of the American Physical Soci-
coatings on the plane specimens confirms that the method of ety, Vol. 1, 1956, p. 382.
[9] McDonald, B. and Goetzberger, A., "Technique for Determination
analysis described above can be applied successfully to the case of the Depth of Diffused Layers," Journal of the Electrochemical
of a coating on a cylindrical surface. Extension of the method of Society, Vol. 109, 1962, pp. 141-144.
analysis to a substrate with compound curvature (such as a turbine [10] Thomson, V., Hintermann, H. E. and Chollet, L., "The Determination
blade or biomedical prosthesis) is straightforward. of Composition Depth Profiles Using Spherical Erosion and Scanning
Auger Electron Spectroscopy," Surface Technology, Vol. 8, 1979,
pp. 421--428.
[11] Valli, J., Palojarvi, J. and Makela, U., "Measurement of Coating
Conclusions Thickness by Using a Ball Cratering Method," Technical Research
Centre of Finland, Report No. 435, t985.
The history, theoretical basis and practical understanding of ball- [12] Kassman, A., et al., "A New Test Method for the Intrinsic Abrasion
cratering as a micro-scale abrasion test has been summarized. Four Resistance of Thin Coatings," Surface and Coatings Technology, Vol.
50, 1991, pp. 75-84.
case studies have been presented to illustrate the current capabilities [13] Nothnagel, G., "Wear Resistance Determination of Coatings From
of the test, through studies of the abrasive wear resistance of Cross-Section Measurements of Ball-Ground Craters," Surface and
the following: Coatings Technology, Vol. 57, 1993, pp. 151-154.
[14] Pensaert, S., Van Peteghem, A. and Debuyck, F., "A New Method
1. Metallic glass ribbons in which the sample dimensions were of Evaluating the Abrasion Resistance of Wires. Part I: Presentation
smaller than 5 mm by 2 mm by 40 p~m. of the Device and Mathematical Background," Wear, Vol. 171, 1994,
pp. 99-107.
2. The uppermost surface layer (total penetration depth of less [15] Silva, S. R. P., Kapoor, A. and Amaratunga, G. A. J., "Tribological
than 40 p,m) in multi-layered automotive polymer paint coatings. Properties of Diamond-Like Carbon Films Deposited on Silicon
3. Thin PVD coatings (1 to 10 Ixm) on tool steels, for which Using R.E Plasma Enhanced CVD," Surface and Coatings Technol-
the independent wear resistance of both the coating and the sub- ogy, Vol. 73, 1995, pp. 132-136.
[16] Rutherford, K. L. and Hutchings, !. M., "A Micro-Abrasion Test, with
strate was assessed from their combined wear behavior in a single Particular Application to Coated Systems," Surface and Coatings
test. The results are also independent of the interfacial adhesion Technology, Vol. 79, 1996, pp. 231-239.
and coating thickness. [17] Rutherlbrd, K. L., et al., "'Laboratory Characterisation of the Wear
4. Bulk and thinly coated samples with surfaces of general Behaviour of PVD-Coated Tool Steels and Correlation with Cutting
compound curvature. Tool Performance," Surface and Coatings Technology, Vol. 80, 1996,
pp. 176-180.
[18] Rutherford, K. L. and Hutchings, I. M., "Micro-Scale Abrasive Wear
The method shows considerable potential for use both as a research Testing of PVD Coatings on Curved Surfaces," Tribology Letters,
tool and for quality assurance testing of surface engineered, coated, Vol. 2, 1996, pp. 1-11.
and painted components. [19] Rutherford, K. L., et al., "The Abrasive Wea~- Resistance of TiN/
NbN Multi-Layers: Measurement and Neural Network Modelling,"
Surface and Coatings Technology, Vot. 86-7, 1996, pp. 472~J,79.
{20] Gfihlin, R., et al., "The Crater Grinder Method as a Means for Coating
Acknowledgments Wear EvaluationIState of the Art," submitted to Surface and Coat-
ings Technology, 1996.
This work was supported via the CASE studentship scheme by [21] Hedenqvist, E, et al., "Mechanical and Tribological Characterisation
the Engineering and Physical Sciences Research Council and Brit- of Low Temperature Deposited PVD TiN Coatings," Surface and
ish Gas plc, UK. We are grateful to P. Hatto (Multi-Arc, Consett, Coatings Technology, Vol. 63, 1994, pp. 115-122.
[22] Pensaert, S., Van Peteghem, A. and Debuyck, E, "A New Method
UK) and E. D. Doyle (Surface Technology, Thomastown, Austra- of Evaluating the Abrasion Resistance of Wires and Experimental
lia) for provision of the PVD coated samples and to A. C. Rama- Work on Coated and Uncoated Wires," Wear, Vol. 171, 1994, pp.
murthy (Ford Motor Co., Dearborn, MI, USA) for provision of 109-113.
the automotive clear-coats. [23] Hutcbings, I. M., "Development of Wear Maps for the Abrasion and
Erosion of Ceramic Materials," in Advanced Ceramicsfor Structural
and Tribological Applications, H. M. Hawthorne and T. Troczynski,
Eds., 1995, Montreal, pp. t27-138.
References [24] Greet, A. L, "Metallic Glasses," Science, Vol. 267, 1995, pp.
1947-1953.
[t] Mcnshutkin, B. N., Russia's Lomonosov, Princeton University Press, [25] Donovan, P. E. and Stobbs, W. M., "'On Dilation and Plastic Flow
Princeton, NJ, 1952. in Some Metallic Glasses," Philosophical Magazine A, Vol. 47, 1983,
[2] Brinell, J. A., "Unders/3knig r6rande jarns och stals samt en del andra pp. 537-546.
kroppars f6rm~ga att motst~ n6tnig," Jernkontorets Annaler, Vol. 76, [26] Zhong, Z. C., Jiang, X. Y. and Greer, A. L., 1996, "Nano-Crystallisa-
1921, pp. 347-398. tion in AI-Based Amorphous Alloys," Philosophical Magazine B
[3J Haworth, R. D. J., "The Abrasion Resistance of Metals," Trans. ASM, (in press).
Vol. 41, 1949, pp. 819-869. [27] Kim, Y. H., et al., "Crystallisation and High Mechanical Strength of
[4] Mott, B. W., Micro-lndentation Hardness Testing, Butterworths Sci- AI-Based Amorphous Alloys," Materials Transactions, Vol. 35, 1994,
entific Publications, London 1956. pp. 293-302.
[5] Tolansky, S. and Austin, E., "Abrasion Tests on Diamonds-- [281 Schoff, C. K., "New Trends in Coatings Technology," Standardization
Interferometric Examination," Nature, Vol. 164, 1949, pp. 194-195. News, American Society for Testing and Materials, West Consho-
[6] Grodzinski, P. and Stern, W., "Abrasion Test on Diamonds," Nature, hocken, PA, October 1995, pp. 24-27.
Vol. 164, 1949, pp. 193-194. [29] Lancaster, J. K., "Abrasive Wear of Polymers," Wear, Vol. 14, 1969,
[7] Wilks, E. M., "An Iuterferometric Investigation of the Abrasion pp. 223-239.
Hardness Properties of Diamond," Philosophical Magazine, Vol. 43, [30] Sanmer, E. and Meier zu K6cker, G., "Utility and Limitations of
1952, pp. 1140-1145. Tribosimulationfor Quality Control and Material Preselection," Wear,
[8] Happ, W, and Shockley, W.,"Diffusion Depths in Silicon Measured Vol. 181-183, 1995, pp. 350-359.

Copyright by ASTM Int'l (all rights reserved); Mon Jan 28 13:36:52 EST 2019
Downloaded/printed by
INSTITUTO POLITECNICO NACIONAL (INSTITUTO POLITECNICO NACIONAL) pursuant to License Agreement. No further reproductions authorized.