01 Contents Chapter 1. Mechanical Analysis of the Scratching Properties of Coated Polymers (C. Gauthier et al.). Chapter 2. Mechanical Analysis of the Scratching of Metals and Polymers at Moderate and Large Strains (E. Felder, J.L. Bucaille). Chapter 3. Friction, Damage and Stick-Slip in the Scratching of Polymers (S.L. Zhang). Chapter 4. Nanoscratch and Interfacial Friction of Poly(Amide) Fibres (J. Cayer-Barrioz et al.). Chapter 5. Metrology for Characterizing the Scratch Resistance of Polymeric Coatings Throughoptical Scattering (Li-Piin Sung et al.). Chapter 6. Scratching of Polymers: Deformation Mapping and Wear Modeling (S.K. Sinha). Chapter 7. An Overview of the Potential of Quantitative Coating Adhesion Measurement by Scratch Testing (S.J. Bull, E.G-Berasetegui). Chapter 8. Characterization of Mar/Scratch Resistance of Coatings with a Nano-Indenter and a Scanning Probe Microscope (Weidian Shen et al.). Chapter 9. Evaluating the Cohesive Strength of a Surface Material by Controlled Scratching (Yongsong Xie, H.M. Hawthorne). Chapter 10. Mechanical Characterization of Nanostructured TiB2 Coatings using Microscratch Techniques (N. Panich, Y. Sun). Chapter 11. Damage Identification of Dlc Coating by Microscratch Test (A. Djamai et al.). Chapter 12. Correlation between Adhesion and Wear Behaviour of Commercial Carbon Based Coating (K.H. Lau, K.Y. Li). Chapter 13. The Study of the Adhesion of a TiN Coating on Steel and Titanium Alloy Substrates Using a Multi-Mode Scratch Tester (J. Stallard et al.). Chapter 14. Studies on Friction and Transfer Layer using Inclined Scratch (P.L. Menezes). Chapter 15. Scratch Resistance of High Nitrogen Austenitic Stainless Steels (A.P. Tschiptschin et al.). Chapter 16. Abrasion of Engineering Ceramics, AlMgB14?tiB2 Composite and other Hard Materials (S. Bahadur, A. Ahmed).
Page 1
PREFACE
The idea of the need for a compiled source book in the area of scratching of materials came when I was writing a review paper with Professor Brian J. Briscoe for a special issue. After much search in the literature, it soon became clear to us that the scratching field, despite being among the oldest of all mechanical tests, does not have a source book which brings together important findings in this area. I thank Brian for providing the spark of the idea that has led to the completion of this edited book. Twelve of its sixteen chapters were first published as journal papers in a special issue of Tribology International (Vol. 39(2), 2006). This book contains some of the important, and authoritative, works that are being conducted in the area of scratch testing for a variety of materials. This area has grown extensively since the earlier days of the Mohs Scale for ranking minerals according to their relative scratch resistance. It is rather surprising that it took more than 100 years for this test to really mature into an engineering tool. To date, this test has been used on metals, ceramics, glasses, polymers and coatings of various types and thicknesses. The important aspect of this test is that it can elucidate surface mechanical responses of a material by simply scratching the surface with a well-defined tip. However, in this simplicity lies the very complex nature of the stresses around a tip that makes it very difficult to interpret data in terms of the basic mechanical properties of the material such as the yield and fracture strengths, elastic modulus, interfacial friction, fracture toughness and so on. Perhaps, like in many other mechanical complexities, computer simulations such as finite element could bring the required solutions to the problem. As the editor of this book, I take great pride and privilege in providing you the works of some of the world's leading researchers in materials, tribology and surface mechanics fields. As you read through the book you will find names of the authors who have devoted a considerable amount of their research career in the area of scratching of materials. Though initially not intended, I have grouped the chapters according to the type of the engineering materials used. The beginning chapters relate mostly to bulk polymers, which are followed by different types of coatings (hard wear resistant to the diamond-like carbon coatings) and finally, chapters on the application of scratching technique to metals and ceramics are included at the end of the book. Thus, the book covers a fairly wide spectrum of engineering materials which are useful to engineers and researchers. l hope the readers of this book will find the chapters interesting and useful for their understanding of scratching technique as applied to different materials. 1 thank all the authors of for contributing their excellent works. Their hard work in preparing the manuscripts is greatly appreciated. Any shortcomings or errors in the book are entirely due to my incapability and I hope to receive much critical feedback on this book from the readers in due course. 1 would like to acknowledge the help of my graduate student Mr. Myo Minn in putting together the content and index of this book. Sujeet K. Sinha Department of Mechanical Engineering National University of Singapore Personal website: http://www.geocities.com/sujeet2020
CHAPTER 1
MECHANICAL ANALYSIS OF THE SCRATCHING PROPERTIES OF COATED POLYMERS
C. GAUTHIER, A.-L. DURIER, C. FOND and R. SCHIRRER Universit~ Louis Pasteur, Institut Charles Sadron, CNRS UPR 22,6 rue Boussingault, F-67083 STRASBOURG, FRANCE. E-mail."
[email protected]
ABSTRACT In the case of polymer scratching, at present there is no model which can take into account the viscoelastic viscoplastic behaviour of the material and the ability of polymers to strain harden or soften. Progress has now been made using numerical simulation and a new experimental set-up. When a viscoelastic contact generates a viscoelastic groove, the recovery is sensitive to the high local strain introduced by a geometric discontinuity such as the roughness of the grooving tip or the angle between two faces of the tip. The scratch resistance conferred by a coating is evident on both the macroscopic scale of the contact and the local scale of the roughness of the tip. On the local scale, the coating prevents the roughness of the tip from creating micro-scratches at the surface of the macro-groove. Therefore, since the absence of micro-scratches is a condition for relaxation of the macro-groove, the thickness of the coating must be greater than the roughness of the tip. On the macroscopic scale, the mechanical behaviour of the contact is modified by the decrease in the friction coefficient.
KEYWORDS coating, polymer, scratching, friction, roughness, recovering, groove
LIST OF SYMBOLS
E V
(:Tyield
ap V,,p F~ F, T a
(o
Young's modulus Poisson's ratio yield stress relaxation peaks sliding speed normal load tangential load temperature contact radius rear contact angle
2
Scratching of materials and applications
Rtip
tip radius total roughness
Rt
R0
concave radius of the groove at t=0s
R(t)
concave radius of the groove
l
length representative mean strain in the matter around the groove
E
strain or mean contact strain scratching contact strain strain rate or mean contact strain rate
~ lapp
apparent friction coefficient
/z
true friction coefficient
i~local
local friction coefficient A, B, C and D local pressure and shear elementary action integrals t* life time of a micro scratch C
ratio of the contact pressure to the yield stress
p(T~) or P~ contact pressure
Cvp Cvp
generalised viscoplastic strain rate
k Or
consistency thermal coefficient
m
strain hardening coefficient sensitivity to the strain rate
generalised viscoplastic strain
INTRODUCTION Most polymeric glasses are sensitive to scratching and resistance against marring or scratching is desirable for applications. Increasing the scratch resistance is equivalent to introducing an elastic contribution into a fully plastic behaviour or to increasing the elastic component in an elastic-plastic behaviour. There are three ways to improve the scratch resistance [ 1]: (a) by decreasing the ratio E/o'yield, where E is the Young's modulus and (~.iela the yield stress, although this cames the major risk of decreasing the Young's modulus with subsequent loss of the macroscopic mechanical properties of the structure. One may note that an elastomeric material, which has a low E/~,eta ratio, is not sensitive to scratching but only to cutting, cracking and wear. (b) by introducing a strain-hardening effect into the stress-strain relationship of the bulk material, which is a means of increasing the elastic unloading in an elastic-plastic strain [2]. Such polymers are generally brittle and sensitive to the influence of a local geometrical flaw.
Mechanical analysis of the scratching properties of coated polymers
3
(c) by coating the material. Coating is a common way of improving the scratch resistance of polymeric glasses. The first solution found to reduce the scratch sensitivity was to deposit a mineral coating on the surface of the polymer. This procedure had however little success, at least partly due to the large difference between the elastic strain domains of the substrate and coating. A second generation of coatings used polysiloxane and acrylic materials, where the scratch resistance is given by the hardness of the coat and the coatings have elastic strain domains in the same range as the substrate. The most recent generation of protective coatings has employed nano-materials, in which an organic matrix is filled with nano-sized particles of silica. The idea behind this strategy is to associate the large elastic domain of an elastomeric polymer with the hardness of the filling. The majority of existing studies describing the behaviour of coatings explore the boundary between sliding or ductile scratching and brittle contact. They generally use the concept of the critical load and the models developed to analyse the cracking after passage of a sliding tip may be viewed as improvements on work done in the 1960's by Lawn et al. [3-10]. Thus, mechanical analyses have been performed assuming that the interface is submitted to shear stress and the coating to compressive and buckling stress, but the adhesion of the coating has not been very successfully correlated with the critical load. Several scratch-adhesion models proposed in the literature were recently compared with experimental data obtained for polymeric coatings [ 11 ]. Two of the models seem to give a reasonable description of the dependence of the critical load on the friction during scratch-adhesion testing at constant sliding speed. These models always predict that the first damage will appear behind or in front of the contact area and in most cases the normal load is linked to the crack energy, sometimes also taking into account the strain energy of the substrate. Still more recently, Bertand-Lambotte et al. [ 12] have proposed that the transition from ductile to brittle scratching of a coating is dependent on a double condition: a fracture energy criterion and a size criterion. Since the mechanical properties of polymers are time and temperature dependent, a single value of the critical load cannot describe the overall mechanical behaviour of a coating. Demirci et al. [ 1] have shown that the mechanical behaviour of a coating on a viscoelastic material should not be analysed in terms of the critical load, but in terms of the shape of the stress field, modified by the effect of the local friction between a scratching tip and the coat, where this local friction will depend on the roughness of the tip. Polymeric materials display complex behaviour and a more refined analysis than for other classes of materials is required to understand the influence of material properties on the scratch resistance [13-20]. Briscoe and Thomas [13] and Gauthier and Schirrer [14] have shown that an analysis of the viscoplastic behaviour of the surface of a material during contact with a sliding tip requires an evaluation of the strain and strain rate. The average value of the mean contact strain rate k may generally be simply estimated as the tip speed divided by the groove width [13] or the contact width [14],
dg ldt=V.pl2a
(1)
where Vtip is the sliding speed and a the contact radius. The mean contact strain is proportional to the ratio of the radius of the surface contact area to the radius of the tip as originally defined by Tabor [16]. The mechanical properties of polymeric materials are usually stress and temperature activated and follow an Arrhenius law at temperatures below the glass transition [14]. It was previously shown [15] that the rear and front contact areas can be
4
Scratching of materials and applications
predicted in the case of plastic and elastic-plastic contacts on a polymer surface. The rear contact area is due to the elastic recovery of the polymer and depends on the plastic deformation around the contact. Even for a moving tip, the rear area can be almost identical to the front area if the contact is almost elastic and the mechanical loss factor low. All previous studies have focused on the behaviour of the surface during the contact time. On the other hand, the major difference between polymeric and other classes of materials is the capacity of the groove left on the surface to recover and this capacity is one way to improve the scratch resistance of polymeric surfaces. The recovery has a time and temperature dependency and may be accelerated if the glass temperature has been crossed during the life time of the groove [ 12]. The analysis of this recovery is recent and complicated to perform due to the difficulty of measuring the geometry left at the start of the life of the groove, just after contact. Conversely, the existence of creeping during normal contact is well known and experimental studies [ 16, 21-23], mechanical analyses [24-27] and more recently a numerical analysis [28] exist. Historically, while the first creeping experiments were carried out on a macroscopic millimeter scale [21], the characteristic length of the contact was later decreased by using a nanoindenter [24, 29]. The creeping function was generally derived from data obtained by recording the vertical motion of the tip. This vertical motion is however not exactly equal to the depth of the imprint, which was found to be dependent on the contact behaviour and highly sensitive to an exact knowledge of the tip shape. An analysis of the recovery of an indentation imprint is more recent and more difficult to perform [29]. The major difficulty is the choice of a probe to quantify a phenomenon confined under a contact and the latest experiments use AFM or SPM probes [23, 29]. Analysis of the recovery of the groove left on the surface has begun [30-31 ]. The aim of this chapter is to present a mechanical analysis of the scratch behaviour and the recovery after grooving of organic coatings deposited on organic glasses.
EXPERIMENTAL PROCEDURES
Experimental set-up The experimental device for the scratch test, called the 'microvisioscratch', has been described previously [14,15]. It consists of a commercial servomechanism beating a small, temperature controlled transparent box which contains the sample and the scratching tip. Control of the moving tip and recording of the normal F, and tangential Ell loads, scratching speed Vtip and temperature T are computer driven. A built-in microscope allows in situ observation and measurement of the groove left on the surface. Scratching over a wide range of speeds (1 to 104 ~tm/s) and within a temperature range coveting the polymer relaxation peaks (-70 to +120 ~ are the main innovative features of the system. The normal load F, applied to the moving tip can be selected from 0.05 to 35 N by adjusting the compression of a spring of low stiffness. In the present experiments, performed at 30~ the speed of the tip was kept constant at 0.03 mm/s. Two cone-shaped diamond tips with a spherical extremity were used, the first having an apex angle of 60 ~ a tip radius Rtip of 116 Jam and a total roughness Rt of 0.6 ~tm and the second an apex angle of 90 ~ a tip radius of 110 ~tm and a total roughness of 2.5 lam. A standard procedure was used to carry out the friction tests. After cleaning the tip and the sample with alcohol and drying them, a preliminary test was performed to age the surface of the tip with the polymer, which is necessary to obtain reproducible measurements. The experiment was then carried out, starting at the lowest normal load and varying it stepwise in the range 0.1 to 2 N, within a single groove, in as many steps as required to explore the entire range of strain sensitivity. At each loading step and throughout the scratching process, in situ photographs
Mechanical analysis of the scratching properties of coated polymers
5
were taken to record information on the shape of the true contact area and the beginning of the life span of the groove left on the surface. As the contact width was not constant in these tests, the mean strain rate ranged from 0.15 to 1 s ~ . Stress/strain curves were determined in compression tests. The experimental device for these tests is based on the moving cross head of an Instron 4502 tensile machine and the whole apparatus is enclosed in an Instron environmental chamber. Compression tests over a wide range of strain rates (10 4 to 10-~ s~), within a temperature range covering the a and fl relaxation peaks of common polymers (-70 to +120 ~ and measuring the longitudinal and radial strain are the main characteristics of this system. The longitudinal strain was limited to 20% during tests.
Materials The organic glass was an amorphous thermoset polymer (diethylene glycol bis(allyl carbonate)) called CR39. The Young's modulus of this resin is typically 2 GPa at 20 ~ and 1 Hz. Cylindrical samples 12 mm long and 5 mm in diameter were used for compression tests while scratch test samples were plates a few millimetres thick. The coating was a spin coating of a nano-composite material, a thermoset matrix filled to about 20 % of its volume with sub-micron silica particles (about 10 nm in diameter). The Young's modulus of this coating is about 4 GPa at 20 ~ and 1 Hz. Since it is partially filled with mineral particles, it does not have a very marked time or temperature dependency. Coatings of different thicknesses (1.1 and 4.38 l.tm) were selected to include the highest degree of roughness.
EXPERIMENTAL RESULTS Contact area Figure 1 shows representative photographs of uncoated and thinly and thickly coated specimens scratched with the two tips. At a normal load of a few tenths of a Newton, the smooth tip slides over the surface of the polymeric materials and leaves a slight residual viscoelastic groove. As the normal load increases, the importance of the coating becomes clear: at a median load the recovery o f the groove is more marked on the coated samples, while at the highest normal load the coated samples always display less lateral and frontal pad formation than the uncoated sample. The rough tip immediately reveals the role of the coating: after contact with the tip under a normal load of a few tenths of a Newton, the uncoated sample has a blemished surface with a few micro scratches. Unlike for the 1.1 ~tm coating, for the 4.38 lam coating there is no difference between scratching by the two tips. The importance of the ratio of the roughness of the tip to the thickness of the coating has been demonstrated previously [ 1] and is confirmed here at higher normal loads: the coating prevents the roughness of the diamond tip from creating micro-scratches at the surface of the macro-groove. Therefore, as the absence of microscratches seems to be a condition for relaxation of the macro-groove, the thickness of the coating must be greater than the roughness of the tip. Since if cracking appears the geometry of the contact area may be modified, the following analysis concerns only contacts without cracks. The contact area between a moving tip and a polymer surface has a front side and a rear side and its shape changes with the strain. A contact area which is entirely plastically deformed will be called for simplicity a 'viscoplastic contact'. If the contact area is not entirely plastically deformed it will be called a 'viscoelastic-plastic contact'. It was previously shown [ 15] that the shape of the contact may be simply described by the rear angle or by the ratio of the rear length to the front length. The rear contact angle co is due to the elastic recovery of the polymer and
6
Scratching of materials and applications
depends on the plastic deformation around the contact. This angle has been drawn on the upper fight image of Fig. 1.
Fig. 1. True contact areas between a spherical tip and the surface of polymer samples and images of the grooves left immediately after passage of the tip. f
7t/2
,
I
t
I
-
.....
I
.
~
1
\,,
f
',
.
Strata hardening
J I
EPcontact model ,~plastic 1 ~d o m a m , I I
Elastic domain
I I
w..
L,.. Plastic dolnam wr~
i
e
P
~;~ afRti p Fig. 2. Rear contact angle as a function of the mean contact strain. This angle decreases in the case of a viscoelastic-plastic or viscoplastic contact and increases if the material shows strain hardening.
Mechanical analysis of the scratching properties of coated polymers
7
Figure 2 gives a schematic representation of the evolution of the rear angle as a function of the mean contact strain G. The strain is assumed to be simply proportional to the ratio of the contact radius to the tip radius for a spherical tip and one may note that there exists no clear definition of the scratching mean strain linking the geometry of the contact and the friction coefficient. Se and cp denote the contact strains at the end of the elastic contact area and the beginning of the plastic contact area, respectively. In the case of a viscoelastic contact the rear angle co is equal to n:/2 for elastic deformation. If the material displays strain hardening, the contact strain decreases and co increases [32]. As the strain rate and temperature vary, even under constant loading, the contact area may also vary considerably and the strain near the contact may change from viscoelastic to viscoplastic.
W2 1.4 --" ~ o o I ~'~. 1.2 "O "v
9 o
1.0
oOooo
0.8
s
~
"\
0.2
E
p cR39 S m o o t h tip
0.0 . . . . . . . . . . . 0.0 0.1 0.2 0.3 0.4
0.5
9 . . . . . . . . 0.6 0.7 0.8 0.9
a/ati
F J 2
9
,
~
|
9
,
9
,
9
,,,
9
|
s
9
o 9
1.0 I 0.8
"
1.0
p
~ . 1.21.4 ~ ~ ~ " t
,-~
uncoated CR391 coating4.38prn
s
i
9
,
9
1
9
uncoated CR39 coating 4.381Jm
coating 4.38pm
s
o.6-
I ~ [] o [] [] oo ~o OO~
0.4.
|
o.2t
~p c R 3 9
0.0, 9 . . . . .
9
,". . . . .
Rough
tip
9, 9, 9
o.o o~ 02 03 0.4 o s 0.6 07 0.8 o.g
o
aJRti p
Fig. 3. Rear contact angle versus contact strain for uncoated CR39 and samples with a 4.38 lain coating. Plastic contact seems to begin at a higher contact strain in the presence of a thick coating.
The evolution of the rear contact angle as a function of the contact strain for polymer samples with no coating and a 4.38 ~tm coating is presented in Fig. 3. Whatever the roughness of the tip, the end of the elastic domain seems to remain constant. In contrast, the elastic-plastic domain
8
Scratching of materials and applications
seems to expand if a thick coating has been deposited on CR39, with the result that the plastic domain begins at a higher contact strain for thickly coated samples. The thin coating presents a less clear evolution (data not shown). Contact pressure The contact pressure is the ratio of the normal load to the true contact area, which is the sum of the front and rear areas. It may be called the scratch hardness only in the case of plastic contact. Figure 4 shows the mean contact pressure as a function of the contact strain for all three types of polymer sample. As previously seen, there is no large difference in the responses to the two tips. On the uncoated material, scratched by the smooth tip, the contact pressure increases from 90 to 140 MPa and then seems to decline as the ratio a/Rti p exceeds 0.4, whereas on the 4.38 mm coating the contact pressure increases continuously. In view of the evolution of the rear angle, one may assume that the contact behaviour of the thickly coated sample is essentially elasticplastic. 160
.
.
.
.
.
.
ur~coated' CR39 coating 1.101Jm coating 4.38pm"
: =
140
(1~ 120
Q. v
100
C~ 80
9 0.0
160
0.1
0.2
1
0.3
. . . . . . - - n - - uncoated CR39
140-
= --
0.4 a/Rtip
.
.
S m o o t h tip] o.s 0.6 0.7 0.8
.
.
coating 1.10pm coating 4.38pm
100.
~e 80.
/e /
60
o.o
o'.~o'.2
R o u g h tip o'.a o'.4 o'.s o'.6 o'.7o.8 a/Rtip
Fig. 4. Contact pressure as a function of the contact strain.
Mechanical analysis of the scratching properties of coated polymers
9
Friction coefficients Figure 5 shows the evolution of the apparent friction coefficient
~Uapp--FtlFn
(2)
as a function of the ratio a/R,p for the two tips and all samples. In the case of the smooth tip the effect of the thickness of the coating on the apparent friction coefficient is clear: the friction decreases as the thickness increases. The same tendency is observed for the rough tip provided the thickness is greater than the roughness of the tip.
E: 0.7
9
..._
.o 0.6 0 0 E 0
.m
~
.
...
~
r
,
,
.
,
- - o - - uncoated CR39 --~,-- coating 1 .lOpm + coating 4.38pm
0.5 0.4.
0 "E
0.3
E
0.2.
S m o o t h tip ELO.1
<
011
0.o
._o 06
o_
013'014
,
v
v
,
0.8
-
|
9
,
|
~ o
'
~
9
n~o-~~J~~
R o u g h tip
(D 0 0.50 cO 0.4.m ,,i.,,a 0 "E. 0.3
r
012
v 9 9 0.5 016 017
a/Rti p
(-- 0.7
E
"-
~
uncoated
o
CR39
---~-- coating 1 .lOpm ; coating 4.38pm
0.2 .
.
0.0 0'1
<
.
.
.
012 013 014 0'.~ 016 0'.7 0.8 aJRti p
Fig. 5. Experimental values of the apparent friction coefficient as a function of polymer samples and the two tips.
a/Rap for all
If the ratio of the local shear to the local pressure is termed the "true friction coefficient"/./, then the apparent friction coefficient may be written as [39]"
Ft ~].lappmC-[-V~.l Fn
A+Bp
(3)
10
Scratching of materials and applications
Solution of this relation between the true and apparent friction coefficients requires calculation of the four integrals A, B, C and D, which are the local pressure and shear elementary action integrals, together with a knowledge of the rear angle c0, the real contact area and the geometry of the tip. A, B, C and D take into account the macroscopic contact shape. However, the true friction coefficient refers to a smooth tip. If the tip is rough, this coefficient of the true friction between tip and surface must be called the "local friction coefficient" because the roughness effect cannot be removed. Figure 6 presents the local friction coefficient as a function of the ratio a]Rtip for all samples and the two tips. At higher values of the contact strain (greater than 0.4), the local friction coefficient does not significantly depend on the contact strain, is greater for the rough tip and smaller when the thickness of the coating is greater than the roughness of the tip. 0.6 - - o - - uncoated CR39 t-O O.S ---~-- coating 1.10pm to 9 coating 4.38pm 0.4
o
to 0.3 cO ._,., '*'-' 0.2 to _.
0.1
03
to O 0.0 -,J 0.0
011
o'.ao'.+
Smooth tip o++ o'.+ o'.+o'.7 o.+
a]Rti p 0.6 r 0.5
Rough tip
to (1) 0 . 4
o
o~
tO 0.3 tO "-~ O.2 to _
_
0.1
03 to O 0.0
..-+
o.o
-o
~~176176176176176
---~-- uncoated CR39 - - ~ - - coating 1.10pm ---o-- coating 4.38pm
o'.~ o'.2 0'3 0'4 o's 0'6 o'7
08
aJRti p
Fig. 6. Estimation of the local friction coefficient as a function of samples and the two tips.
a/R,p for
all polymer
Groove recovery Since if cracking appears the recovery of the groove may be hindered, this section relates to grooves which stay ductile after scratching. Comparison of the recovery of different grooves was thus made in a particular configuration: the normal load was fixed at 0.64 N to generate scratches with a minimum yield, so as to have the time to record the groove profiles without having a fully plastic contact. The tests selected were uncoated CR39 scratched by the two tips
Mechanical analysis of the scratching properties of coated polymers
11
and CR39 with a 1.10 gm coating scratched by the rough tip. Under this normal load, the contact strain lies in the range 0.4 to 0.45, the local friction coefficient has its asymptotic value of 0.3 to 0.35 and cracking never appears. The recovery of samples with a 4.38 gm coating was not tip sensitive and it can be clearly seen in Figure 1 that for such samples one cannot justify recording the groove profile for a minimum period of time. Experiments were performed at 30~ the first profile was recorded 9 minutes after scratching and 6 further profiles were recorded during the next 6000 minutes. Figure 7 shows the in-situ contacts and the grooves initially created on the three samples. As previously, the number of micro scratches on the imprint of the groove decreases if the sample has been coated, as does the size of the frontal push pad. Two important points should be noted: under the chosen experimental conditions, the width of the groove remains constant and its shape does not resemble a flame. Micro scratches generated by the roughness of the tip can recover and the number of micro scratches decreases with time (see arrows drawn on Figure 8). The life time of a micro scratch as recorded on this photograph is the ratio of the scratch length to the sliding speed
t*=/__/__ V,ip
(4)
and is typically about 10 s. After scratching, the size and shape of the groove left on a sample was analysed using a commercial mechanical profile recorder and the radius of the section of the groove was estimated by fitting the profile data. This radius increases with the life time of the groove as seen in Fig. 9. If at 10 minutes, the sections generated by the rough tip are of the same order of magnitude, at 6000 minutes the radius of the groove section is clearly higher on the coated sample. At both time points, the radius of the groove section is greater for the smooth tip. Figure 9 also shows interferometric profiles of the grooves at two time points. These profiles clearly demonstrate that micro scratches on the coating are more open than micro scratches on the uncoated samples and that the radii increase with time.
12
Scratching of materials and applications
Fig. 7. In-situ images of the contact areas and initial grooves on the three samples. At the top uncoated CR39 scratched by the smooth tip, in the centre the same sample scratched by the rough tip and at the bottom CR39 with a 1.10 lam coating scratched by the rough tip.
Mechanical analysis of the scratching properties of coated polymers
13
Fig. 9. Radius of the groove section versus life time of the groove. The radius increases with time as the groove recovers, while increasing the roughness of the tip slows this phenomenon.
14
Scratching of materials and applications
DISCUSSION
Analysis of contact results In the case of the smooth tip and high contact strain, the local friction coefficient decreases as the thickness of the coating increases, while for the rough tip the tendency would seem to be the same, but is not as marked. During these tests at constant temperature and normal load, the mean strain rate varies over nearly one decade.
100 ~
,
,
,
.....
80
n 60
dddt
E o
b
40
o
8) :~
1E-1
9
1E-2
o
1E-3
9
1E-4
20
t,--
I--
T=20~ 0~
o.oo
o.~5
S "1
o
0.~ 5
0.i0
"
-
0.20
True strain
O_
200
v
175 ~
"o
,,
._ 150~
o
9
(D
~9
o
125-
_ o
=
"
-,-"
9
o
v~
50
.
-4r~---
8) L
,
o
+90~
,,
+70oc
o
+60~
v
+20~
,-,
_20oc
100 -c 75-
c'} E
function exp. datas datas .
.............................................
1E-5
"
~
v
---
1E-4
b
O-
1E-3
0
....
0.01
o
o
0.1 1
1
••
10
d~/dt ( s )
Fig. 10. True compressive stress versus true strain (top) and comparison of the extrapolated function with the experimental yield stress as a function of the strain rate over a range of temperatures and strain rates (bottom). The experimental data and extrapolated function correlate over a wide range of temperatures.
It is well known that below the glass temperature, the yield stress of polymeric materials increases linearly with the logarithm of the strain rate as predicted by Eyring's law. Just as the scratch hardness was classically written as a function of the yield stress, the contact pressure may be written as a function of the yield stress with a factor c depending on the contact strain:
Mechanical analysis of the scratching properties of coated polymers
15
pyield:c(~)
(5)
O'y This ratio of the contact pressure to the yield stress, called the normalised contact pressure, is time and temperature independent and depends only on the strain in the contact area [34]. The major assumption made here is the decoupling of the strain dependency and the temperature and velocity dependency. In the case of an elastic or plastic contact the factor c is linked to the ratio E.a
O'yield~ip
. The contact pressure p(T,~) should thus be normalised by the yield stress cry(T,~) for
the same values of(T,s). Figure 10 shows on the left the stress-strain curves at one temperature and several strain rates and on the right the yield stress estimated from these tests as a function of the strain rate. Using these values, an extrapolated function was derived which allows one to estimate the yield stress to a good approximation over a wide range of temperatures and scratching velocities. Thus, the yield stress was fitted with a second degree polynomial law to estimate the values at strain rates comparable to those in scratch tests ( 10 -2 to 102s -~ ) :
o-r (~',T)=a( T)+b( T)loge+c(T)(loge) 2 2.0
.
,
.
,
,
9 ,
,
,
.,
,
(6)
9
1.8
"~ 9 "~ 1.2
~"
...
1.0
........
0.8 ~
0.6
0.4
D
0.2
uncoated CR39 coating 1.10pm coating 4.381Jm
9
Smooth tip
0.0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
a/Rtio
2.0
1.8 1.6 1.4 -o 1.2 ~
1.o
~o
0.8~
r
0.6~
o.42 o.22
0,0~
0,0"
/
a
unc0ate d CR39 coating 1 10pm
9
coating 4.38pm
0~.1 " 0'.2
0'.3 " 0'.4 a/Rti
Rough tip. 0'.5
0'.6 " 0'.7 "
),8
p
Fig. 11. Evolution of the normalised contact pressure as a function of the mean contact strain for the two tips. For the 4.38 ~tm coating, the yielding seems to be retarded: the normalised contact pressure reaches a value of about 1.6 for a large value of the mean contact strain.
16
Scratching of materials and applications
A thin polymeric coating deposited on a polymeric substrate (typically a few microns thick for a contact width of about 80 ~tm) will not modify the global mechanical response to an indenter. O'Sullivan et. al. [36] have shown that for a spherical tip sliding over a layered elastic halfspace, the radius of the contact zone and the pressure under the centre of the indenter differ significantly from the Hertzian case only when the Young's modulus of the coating differs significantly from that of the substrate. Since in our experiments the ratio of the contact width to the thickness of the coating was greater than 10 for the lowest normal load and the Young's moduli of the two materials were of the same order of magnitude, the coating should not have influenced the bulk behaviour, contact geometry or contact pressure. Consequently, all the contact pressures may be normalised by the extrapolated function of the yield stress for CR39 under compression. During elastic static contact, yield occurs when the normalised contact pressure is equal to 1.1 for a Poisson's ratio of 0.3 and perfectly elastic-plastic behaviour [35]. This result may be used as an upper boundary for a sliding test where the friction favours yielding. When the normalised contact pressure is plotted against the contact strain (Figure 11), it is clear that the coating prevents yielding. At a given contact strain, the normalised contact pressure is lower if the sample has been coated. The estimated local friction coefficient shown in Figure 6 displays a large variation (0.15 to 0.4). At a ratio a/Rtip of 0.4, the contact width is 90 ~tm and the depth of penetration of the tip almost 10 ~tm. As the coating has a Young's modulus only twice that of the substrate, its improvement of the scratch resistance must be investigated primarily as an effect due to decreasing the friction coefficient.
Numerical simulation to identify the friction dependency Although indentation or scratch tests permit easy determination of a mean contact pressure, interpretation of the stress-strain curves is otten difficult. Hence the contact behaviour was modelled using the CAST3M 9 code. The mesh procedure Demete version 2.0. beta 9 (CEA/SEMT - P Verpeaux) was employed and the finite element mesh was a right-angled parallelepiped. The domain elements were three-dimensional meshes with ten-node tetrahedra and the mesh was refined under the contact area. Elliptical contact pressure and shear stress distributions were used to model the contact between a spherical tip and the surface. During calculation, the normal load was increased step by step while the contact radius stayed constant. Despite the fact that the elasticity of polymers is often non linear at a given temperature and strain rate, the elastic behaviour was modelled by a linear incremental law defined by Young's modulus E and Poisson's ratio v, both taken to be constant. E and v were determined in a compressive test. The flow stress was described by a G'sell-Jonas law [33]:
o-=k e(aT/ T~l-e-W~vp~mvl~hge~p where
kvp and
Cvpare respectively the generalised viscoplastic strain rate and strain and
the consistency, a r a thermal coefficient,
hg
(7)
k
is
the strain hardening coefficient and m the
sensitivity to the strain rate. In the formalism of G'sell and Jonas, the term (1 -e-~"") describes the viscoelastic behaviour under loading but does not model the elastic unloading of the deformation and hence this term was not considered in the present work (w=5000 and 1 - e -w~ ~ 1 ). In our simulation, the elastic recovery was directly related to the ratio of the
Mechanical analysis of the scratchingproperties of coated polymers
17
flow stress o'. to Young's modulus E, thermal effects were neglected and a r was equal to 0. Eq. (7) then becomes: 2
Cry=ks. m
o- =k ~mvt~hgg~p whichimplies
(8)
The three parameters,k, hg and m have been described previously [15] and were determined by an inverse method adapted to large deformations and based on interpretation of the force-penetration curves in indentation tests with two indenter shapes: m = 0.078 and
hg =
4.5. In our case, the strain rate was higher than in [15] and the consistency was adjusted: k- 87 MPa.s m. Figure 12 compares the experimental compressive stress-strain data with the numerical function used for the present simulations. In the case of perfectly elastic-plastic behaviour, this contact model has been validated for coated and uncoated materials [38] by comparing it with all well known elastic solutions [35-37]. 400 ~,
350
13.
300
v
250
i
d J d t s -1 o
E O
#
fJ
200
13
150
::3 t._
100
I--
1E-1
- - , , - - modelled law
50
~
0,. 0.0
,.,.~_
# ii
~---Z~.,,0~'~176
o'.2 0:4 True strain
0.6
Fig. 12. Comparison of the stress-strain function used for the present simulations with the experimental data
Numerical simulations were performed to locate the boundaries between elastic and elasticplastic contact and between elastic-plastic and plastic contact. The first boundary could be simply related to the first finite volume having a strain higher than the elastic strain, while the second was defined to occur when all the matter contained in the half spherical volume under the contact area flowed plastically. These two boundaries appear in Figure 13, where the normalised contact pressure is plotted against the local friction coefficient. The results obtained for the first boundary agree with those reported by Johnson [35]. Experimental data for the smooth tip are plotted on the same figure and the in-situ photographs clearly show that for a given normalised contact pressure, the size of the push pads and the contact yielding depend on the local friction coefficient.
18
Scratching of materials and applications
Thus, for a given tip radius, the yielding of the contact depends on both the normal load which governs the 'geometrical contact strain' (i.e. the ratio a/Raip ) and the local friction coefficient. As a result, the 'scratching contact strain' must be defined as a function of these two variables:
~scratching = f l a / R~p ). g(p~o~, )
(9)
If these two components are decoupled, for a given ratio a/R,~p, the 'scratching contact strain' may be considered to be simply proportional to the local friction coefficient. As our recovery experiments were performed at a constant ratio a/Rap, the recovery of the groove must be analysed with regard to the local friction coefficient and the ratio of the thickness of the coating to the roughness of the tip.
Fig. 13. Boundaries between elastic and elastic-plastic contact and between elastic-plastic and plastic contact obtained by numerical simulation. Experimental data for the smooth tip are also plotted and the in-situ photography (F,=l.58N) clearly shows that for a given normalised contact pressure, the size of the push pads and the contact yielding depend on the local friction coefficient.
Mechanical analysis of the scratching properties of coated polymers
19
Recovery of the groove It was not easy to measure the radius of the groove section just after contact, at the beginning of the life span of the groove. Although the contact displayed yielding, if the unloading after contact was elastic and hence reversible, a second passage of the scratching tip along the groove with a section of concave radius R0 did not increase the yielding. The following analysis may be considered to transfer to scratching the results of previous work on indentation [35]. Provided the elastic sliding may be predicted with the elastic Hertz theory, the radius of the groove can be related to the radius of the tip through the equation:
1_ 1 3F. Ro-R, ip 4E*a 3
(10)
where E* is the Hertz contact elastic modulus. During recovery, the edges of the groove lie parallel. The mean strain in the matter around the groove may be defined as:
e(t)=Ra(t)
(11)
and the recovery of the groove is described by the ratio: e ( t ) _ R0 m
go
R(t)
(12)
This recovery is plotted in Figure 14. During contact, the recovery process was not freely stressed because contact between the tip and the surface existed. The start of recovery was defined to occur after contact, i.e. after the maximum strain had been imposed at the maximum contact width for a time to : to~ a---aVii,
(13)
Results are in agreement with the previous analysis. A comparison of the two curves for the uncoated material indicates that the recovery increases if the tip is smooth, which is related to the fact that the local interfacial strain decreases if there are no micro scratches. A comparison of the two curves for the rough tip indicates that the recovery increases if the sample has been coated. This is linked to a decrease in the yield in the elastic-plastic strain under the contact area as the local friction coefficient decreases.
Scratching of materials and applications
20
- :,~"
':
..... rough'tiP'-"l~'R39 .............. !
"~.
o
roughtip - coating 1.1 pm
v
rr rr S-to=aJVtip
0.1
l o.o~
.'
.......
r
o.~
,,
.
.
.
....~
.
.
.
.
.
'i'o
.
.
.
;5o
"io'oo'ioooo
life time (min) Fig. 14. Recovery of the groove left on the surface as a function of time. The recovery increases if the tip is smooth or if the local friction coefficient is low (coated sample).
CONCLUSIONS Deposition of a scratch-resistant coating is a common way to improve the scratch behaviour of a polymeric surface. However, a thin scratch-resistant coating cannot prevent yielding on the macroscopic scale of the contact. In-situ photographs show macro grooves with parallel edges, which is an indication of the occurrence of yielding during contact. The mechanical behaviour of a coating on a viscoelastic material is easily described by considering the normalised contact pressure and should be analysed in terms of the shape of the stress field, modified by the effect of the local friction between a scratching tip and the surface, where this local friction will depend on the roughness of the tip and the presence of the coating. A coating decreases the yielding in the elastic-plastic behaviour of the contact if the local friction coefficient between the surface and the tip is low. The major benefit is the reduction of the "scratching contact strain". The ratio of the thickness of the coating to the roughness of the tip is confirmed to be a critical parameter, which enables one to increase the scratch resistance in the case of a thin coating. The recovery of the groove left on the surface must be analysed in relation to the contact behaviour. If the local friction coefficient is low and there are no micro scratches along the macro groove, the recovery may be fast.
REFERENCES
,
Demirci, I., Gauthier, C. and Schirrer R. (2005) Thin Solid Films 479, 207 Bucaille, J.L., Felder, E., Hochstetter, G. (2004), Journal of Tribology 126, 372. Lawn, B.R., (1967), Proceeding of the Royal Society London Ser. A 299, 307. Velkamp, J.D.B., Hattu, N., Snijders, V.A.C., In: Fracture Mechanics of Ceramics 1/ol. 3: Cracks formation during scratching of brittle materials, pp273-301 Ed. R.C. Bradt, D.H.P. Hasselman and F.F. Lange, Plenum Press, N.Y. (1978). Malzbender, J., de With, G., (2000), Surface and Coatings Technology 135, 60.
Mechanical analysis of the scratching properties of coated polymers
.
7. 8. ~
10. 11. 12.
21
Steinmann, P.A., Tardy, Y., Hintermann, H.E., (1987), Thin Solid Films 154, 333. Bumett, P., Rickerby, D.S., (1987), Thin Solid Films 154, 403. Bull, S.J., Rickerby, D.S., Matthews, A., Legland, A., Pau, A.R., Valli, J., (1988), Surface and Coatings Technology 36, 503. Thouless, M.D., (1998), Engineering Fracture Mechanics 61, 75. Malzbender, J., de With, G., (2002), Surface and Coatings Technology 154, 21. Blees, M.H., Winkelman, G.B., Balkenende, A.R., Den Toonder, J.M.J., (2000), Thin Solid Films 359, 1. Bertrand-Lambotte, P., Loubet, J.L., Verpy, C., Pavan, P., (2002), Thin Solid Films 420421, 281.
13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
Briscoe, B.J., Thomas, P.S., (1995), Tribology Transactions 38, 382. Gauthier, C., Schirrer, R., (2000), Journal of Materials Science 35, 2121. Gauthier, C., Lafaye, S., Schirrer, R., (2001), Tribology International 34, 469. Tabor, D., (1970), Review of Physics in Technology 1, 145. Briscoe, B.J., Sinha, S.K., (2003), Materialwissenschaft und Werkstoffiechnik 34, 989. Zhang, S.L., Nishizoe, K., (2004), Tribology Letters 16, 73. Jardret, V., Morel, P., (2003), Progress in Organics Coatings 48, 322. Wong, J.S.S., Sue, H.J., Zeng, K.Y., Li, R.K.Y., Mai, Y.W., (2004), Acta Materialia 52, 431. Darlix, B., Montmittonnet, P., Monasse, B., (1986), Polymer Testing 6, 189. Syed Asif, S.A., Pethica, J.B., (1998), Journal Adhesion 67, 153. Basire, C., Fretigny, C., (1997), C.R. Acad. Sci. t. 325, Serie II b, 21 I. Lee, E.H., Radok, J.R.M., (1960), Journal of Applied Mechanics 27, 438. Hill, R., (1992), Proceedings of the Royal Society Math. and Phys. Sci. 436, 617. Ting, T.C.T., (1966), Journal of Applied Mechanics Trans ASME, 35, 845. Graham, G.A.C., (1967), International Journal of Engineering Sciences 5, 495. Yue, Z.Y., Eggeler, G., Stfckhert, B., (2001), Computational Materials Science 21, 37. Karapanagiotis, I., Evans, D.F., Gerberich, W.W., (2002), Polymer 43, 1343. Shen, W., Smith, S. M., Ye, H., Jonnes, F., Jacobs, P. B., (1998), Tribology Letters 5, 75. Knapicka, A., Johansson, B., Johansson, M., Hult, A., (2003), Progress in Organic Coatings 48, 14. Bucaille, J.L., Gauthier, C., Felder, E., Schirrer, R., (2005), Wear in press doi: 10. I 016/j.wear.2005.04.007 G'Sell, C., Jonas, J.J., (1979), Journal of Materials Science 14, 583. Gauthier, C., Schirrer, R., The viscoelastic viscoplastic behaviour of a scratch on a polymeric surface, Proceedings of the 2nd World Tribology Congress 14~C2001, Vienna Austria September 2001, CDRom, ISBN 3-901657-09-6 Johnson, K.L., Contact Mechanics. Cambridge University Press, 1985. O'Sullivan, T.C., King, R.B., (1988), Journal of Tribology 110, 235. Hamilton, G.M., Goodman, L.E., (1966), ASME Journal of Applied Mechanics 33, 371. Demirci, I. Phd, Louis Pasteur University- ULP Strasbourg France, 2004. Lafaye, S., Gauthier, C., Schirrer, R., (2005), Tribology International 38, 113.
22
CHAPTER 2
MECHANICAL ANALYSIS OF THE SCRATCHING OF METALS AND POLYMERS AT MODERATE AND LARGE STRAINS
Originally published in Tribology hlternational vol 39. February 2006
BP 207
E. FELDER and J. L. BUCAILLE Centre de Mise en Forme des Mat~riaux (CEMEF) UMR 7635 CNRS-Ecole des Mines de Paris F 06904 Sophia Antipolis Cedex (France). E-mail."
[email protected]
ABSTRACT Scratch test provides a convenient means to study the surface mechanical properties and the tribological performances of materials. The representative strain of the material in this test increases with the attack angle 13 of the indenter and so, for a conical indenter, the strain increases as its apical angle 20 decreases. But the mechanical analysis of this test by analytic models is very intricate. First we perform a preliminary discussion of the various aspects of the problem by considering the plane strain scratching of materials by wedges. Second, we present the conditions of the numerical simulations of the scratch test with conical indenters with a three-dimensional (3D) finite element code. These simulations provide the scratch geometry (contact surface, elastic recovery), the plastic strain map and the volume average plastic strain, the scratch hardness and the force ratio, the apparent friction coefficient ~t0=Ft/W. We compare the behaviour of polymeric and metallic materials in scratch test at low and large strain and relate their difference in scratch resistance to their rheological properties. Polymers develop higher elastic strains at yielding than metals, a phenomenon which is characterised by the yield stress to Young's modulus ratio, ee = O'y/E. For 0=70.3 deg where pure ploughing occurs for all materials, we study the scratching of elastic-perfectly plastic solids with various values of ee under zero friction. Some comparisons with the behaviour in indentation are performed and we study the influence of friction in the scratching of workhardened steel with the same cone. At high strain the main rheological difference is the workhardening behaviour: it is described by a power law for metals and an exponential law for polymers. For 0 decreasing from 70.3 to 20 deg we compare the behaviour of a cold-worked steel to the behavour of polycarbonate, a thermoplastic polymer: a transition from ploughing to ploughing-cutting occurs only for steel.
KEYWORDS Scratch, metals, polymers, cone, representative strain, hardness.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
23
LIST OF NOTATIONS frontal contact radius lateral contact radius area of the contact surface projected on the sample surface width of the residual groove C shape ratio (hJh) E Young's modulus tangential force Ft h penetration depth h~ (lateral) contact depth frontal contact depth hfr hg strain hardening coefficient H indentation hardness (true) scratch hardness (W/A) Hs apparent scratch hardness (8W/[rtb2]) Hsl/2 m strain rate coefficient contact pressure P V scratch speed W normal force X indentation index ([E/ay]COt0) (0x,y,z) rectangular coordinate system (at the cone tip) Oy axis parallel to the scratch speed 0z axis normal to the material surface
af
al A b
Greek symbols
I;
.2_"
oe E
l.t ~to 20 O'y (TO 1;
rear contact angle wedge attack angle strain strain rate elastic strain at the yield stress in tensile testing (t3y/E) volume average deformation generalized (plastic) strain rate generalized (plastic) strain (true) Coulomb friction coefficient (at the indenter/material interface) (z/p) apparent friction coefficient (Ft/W) apical angle of the conical indenter yield stress flow stress strength coefficient friction shear stress
Scratching of materials and applications
24 INTRODUCTION
The scratch test is a very old experimental procedure used to study the mechanical properties of materials near their surface. As early as 1722 Reaumur [ 1] developed a scratch scale which was a measure of the position on an end quenched steel bar which could be scratched by the metal specimen. Exactly a century later (1822) Mohs proposed ten minerals in increasing order of scratch hardness: each mineral will scratch the one on the scale below it but will not scratch the one above it. Another century later (1954) Tabor [2] demonstrates that a metal surface of (indentation) hardness H1 will be scratched by a point of hardness H2 if H2 > 1.2 Hi and that each Mohs standard is approximately 60 % harder than the preceding one. The Tabor's study
demonstrates clearly the connections between the (indentation) hardness H and the scratch hardness Hs, a problem related to the the models of contact and friction between solids [3], but the accurate relations are still not well understood even for metals [4,5]. Another problem related to the scratch test is the transition between ploughing (or rubbing) and cutting first investigated to our knowledge by Mulheam and Samuels [6]. A main parameter for this problem is the attack angle which is defined in Fig.l; they state precisely by experiments the existence of a critical attack angle 13c below which ploughing occurs without any material removal and above which cutting occurs with the formation of a chip. This very complex problem is in context with the matter balance during the scratch test and the understanding of the abrasive machining and the wear of materials [7]. Finally, scratch tests have been applied to polymers in the last 10 years [813] because polymers coatings are often used to increase the scratch resistance in many applications: paintings for automobile industry or optical devices. Because of the large elastic recovery of polymers and their complex rheological behaviour, the analysis and the interpretation of this test is far more difficult than for metals. For example, to compute scratch hardness Hs, Briscoe et al. [8] have considered that for all scratch conditions (scratch speed V, angle of the indenter 0) the elastic recovery is complete at the rear face of the indenter (Fig. 1); by building a new apparatus to observe the real contact surface under load Gauthier and Schirrer [ 13] demonstrated that in fact the contact surface at the rear part decreases as the attack angle 13 increases. front view
side view
B-B
A-A
a_.,..j
B F! I-*--
under ]oad
hc=ch
I k..___
_.~AI
:
......
i.~.s
b
residual groove I
! ............
._~/
Fig. 1. Geometrical and mechanical parameters during a scratch test performed with a conical indenter with apical angle 20.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
25
The scratch hardness and the surface deformation mechanisms of materials depend in particular on the rheology of the material, the indenter geometry and the friction at the interface. But the mechanical analysis of this test by analytic models is very intricate and requires many simplifying assumptions especially in three dimensional (3D) conditions [5, 14-17]. First, after having defined our notations we perform a preliminary discussion of the various aspects of the problem by considering the two dimensional (2D) case: the plane strain scratching of materials by wedges. After, we present the conditions of the numerical simulations of scratching with conical indenters with a three-dimensional (3D) finite element code. So we compare the behaviour of polymeric and metallic materials in the scratch test at low and high strain and relate their difference in scratch resistance to their rheological properties. In addition, we discuss the reliability of many assumptions commonly used in the interpretation of the scratch test and compare the indentation and scratch testing.
PRELIMINARY: THE SCRATCHING WITH CONICAL INDENTERS
Representative Strain and Indentation lndex In scratch test, as in order of magnitude distance in relation strain in indentation
indentation, the normal component of the material displacement Uz is of the of the penetration depth h and this displacement is accommodated on a with the contact radius a=bc/2 (Fig. 1). So the order of magnitude of the ei or scratching es is:
ei ~ es J
OUz
~, ..
h
9-~ - ~ cot 0 = tan ,8 Oz a
(1)
Thus the strain increases with the attack angle 13 or as the semiapical angle 0 decreases. However, we must notice that the representative strain depends on the investigated problem (indentation or scratching, hardness or volume average value...) and its relation with cot0 can be rather complex as we shall demonstrate later. For example for metal indentation, following the Tabor's model of indentation pressure with spherical indenter [ 18], Johnson proposed that the indentation hardness of metals is related to the value of the flow stress o0 at the strain 0.2 cot0 [ 19]: H = Co'0(0.2cot0 )
(2)
So, a first rheological parameter of interest is the yield stress (yy to the Young's modulus E ratio: O'y oee = ~ E
(3)
ee defines the order of magnitude of the elastic strain of the material at the point of plastic yielding which is same as the elastic recovery after plastic strain: it is generally very small for metals (0.1-0.5 %) whereas it is much higher for polymers (1-10 %). We may say that the elasticity of the material increases if ~e increases; rigid plastic materials correspond to the limit case ee =0. If the flow stress of the material does not depend on the strain or the strain rate, case which corresponds to perfectly plastic materials for which the flow stress is always equal to the
26
Scratching of materials and applications
yield stress o0 = Oy, the behaviour in cone indentation or scratch test at constant temperature is characterised by I~i t o ~;e ratio, a quantity which we call the indentation index X: X = - -6i =~
ge
E cot0 Cry
(4)
We can expect that the elastic effects become more and more negligible as X increases. For example, for a given value of 0, in indentation, the constrain factor C (relation (2)) increases steadily with X and tends toward a limit which depends only on 0 (full plastic regime) ca(0), ca(0) being an increasing function of 0 under zero friction [ 19]. Following the approach of Hill et. al. [20] for indentation, we define a quantity which characterises the contact geometry: the shape ratio (Fig. 1): hc
c = m - - - > b c = 2chtanO h
(5)
For elastic contact (very low X) the contact conditions are described by the Hertz's theory: the contact is limited by a full circle, with the radius a=(2/rt)h tan0 (the material sinks in), c=2/n~0.637. We expect that as in indentation (see below) c is an increasing function of the indentation index. Scratch Hardness and Representative Strain
On the indenter are applied a normal force W and a tangential force Ft (Fig. 1). Consider the area of the true contact surface projected on the sample surface A and the width b of the residual groove which is the distance between the tops of the shoulders or the scratch pileups (Fig. 1). We can define two types of scratch hardness related to the normal force: w A
Hs =~
Hs
1/2
8W =~ ztb2
(6)
Hs is the average contact pressure if friction between material and indenter is negligible; we name Hs the (true) scratch hardness. The quantity Hsl/2 defined in [4] is easily measured in experiments. Hsl/2-Hs if bc---b and if the contact is restricted on the frontal part and limited by half of a circle (so this requires particularly lx:---hfc). So we name H s m the apparent scratch hardness. If it is assumed in addition that the contact pressure is uniform, the force ratio or apparent friction coefficient/zo is related to the apical angle (Fig. 1) [3]: /1o . .F,. . 2 cot 0 W n"
(/~ = 0)
(7)
This relation is in good agreement with the experimental results of lubricated scratching of metals for 30< 0 < 70 deg [4]. These assumptions on contact geometry are commonly valid for metals because ec is very small. So the elastic effects are assumed negligible and the contact is considered as restricted to the frontal part [4]. They are not verified for polymers [ 13]. If, on the
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
27
contrary, the strain of the material is totally elastic (X very low), Hsl/2=2Hs=(E/(1-o2))cotO and la0=0. So in the general case we can expect a relation of the form: Ft
2
l,o --if-- r
(~ =0)
(8)
0_< ~'(X)< 1
where ~(X) is an increasing function of the indentation index. We will discuss later the reliability of these relations and the influence of the (real) friction coefficient ~t.
ANALYSIS OF THE WEDGE SCRATCHING (2D CASE) Before considering the three dimensional (3D) scratching, it is instructive to review briefly the two dimensional (2D) case where analytic and experimental results are available. Even in this apparently "simple" case there remain some open questions. This is not surprising because in all cases scratching involves free surface with unknown shape, and so the uniqueness of the theoretical solution of the equations is not insured. Case of a Rigid Plastic Material under Zero Friction We consider first the rigid perfectly plastic (RPP) material (ge=0; (~0"-~-~y). This case can be analysed by the slip line field (SLF) method. Since the pioneering work of Challen and Oxley [21 ] it is now well established from an experimental and theoretical point of view that for low attack angle the wedge produces a ploughing flow and is supported by a frontal edge, the plastic wave, where the material is plastically sheared (Fig. 2a).
~--.-~--z..
of flow V ~
R/I w
1
,
P
Material * /
Material a) The ploughing flow : the plastic wave [21 ].
b) The cutting flow (~t=0).
Fig. 2. Flow patterns in plane strain scratching; Interaction between a rigid wedge and a rigid perfectly plastic body.
If w and l (=DE) are respectively the width and the length of the contact, the force balance and the power balance are:
28
Scratching of materials and applications
Ft V =
Oo~p e V + l r Au
)
(9)
Au is the sliding velocity along the interface, e and % are respectively the thickness of the strained layer and the average plastic strain in this layer (Fig. 2a). So we can deduce from the current SLF model [21 ] an estimation of the mean plastic strain induced in 2D ploughing. Under zero friction where la0--tan[3, the contact pressure p which is uniform and ep are:
f
P = ~3 (2+rr-4fl)
(u =o)
(lo)
sin fl
For high attack angle, it is sure that cutting occurs, but various complex SLF models have been proposed. We consider here the very approximate, but simple model of Merchant where the chip is formed by a single velocity discontinuity with the shear angle ~ [22]. Under zero friction =13/2 and the chip has the same thickness than the feed e (Fig. 2b). It provides a very simple expression for the plastic strain of the chip: ~ =~
cot
O, = o)
(I I)
We see in Fig. 3 that the ploughing strain % increases with the attack angle as expected, but this increase is much higher than tanl3: for example for [3=20deg for which tan[3=0.36, %--2--5.56tan [3. This strain becomes infinite for 13=45 deg which is the highest attack angle for this SLF model. On the contrary the cutting strain % is a decreasing function of 13 and becomes lower than ep for [3>30 deg. This very simple model suggests the existence of a critical attack angle [3c-30 deg. This conclusion is in rather good agreement with more elaborate models: kinematic models of ploughing flow [23] cannot be built for close to 13>15 deg and this suggests a change in flow pattern for higher attack angles. Petryk [24] has built various SLF models of wedge scratching and has demonstrated that the plastic wave flow dissipates higher power than a composite flow inducing cutting and ploughing for 13>20 deg. This result suggests that the critical attack angle in the plane strain scratching of RPP solids under zero friction is: flc~20deg
for
2D scratching o f RPP solids
and
/1=0
(12)
Mechanical analysis of the scratching of metals and polymers at moderate and large strains 10
c~
8
o. o.I=, u
I
1
29
,/, ~p I
I
-
6
Friction ~Elasticity /~ ~. / Work/ ~ , / harden!ng
-
f.o
4.3 4
~J
2-
_
.. j
Ploughing Cutting
"
/ 0
. I
o
I
1
lO
20
13c= 3 0
t 4O
13 (deg)
Fig. 3. Plane strain scratching of a RPP body: Evolution of the cutting ec and ploughing go strain and evaluation of the critical attack angle; Schematic evolution of gc and go with friction and rheology of the body.
This conclusion is, to a first approximation, in agreement with various experimental results where a transition between pure ploughing and a cutting-ploughing flow is observed and for work-hardened metals this occurs at between 15 and 30 deg [25, 26].
Influence of Friction, Elasticity and Work hardening Figure 3 provides a very simple mean for discussing the influence on the critical attack angle of the various rheological characteristics of real materials: elasticity ge, strain sensitivity 0or0 /0~ and strain rate sensitivity0cr 0 /c3or At first insight we do not expect that such characteristics have a very significant influence on the cutting strain go(13). On the contrary, an increase in these quantities produces a very significant increase in the strained depth e (Fig. 2a) and so decreases gr,(13). The effect of the workhardening exponent 01nor 0 /01ng has been demonstrated with a kinematic model of ploughing [27]; numerical simulation will demonstrate the effect of elasticity below. So we can expect that the critical attack angle increases with elasticity, strain and strain rate sensitivity (Fig. 3). Recent numerical simulations [28] of 2D scratching of silicon (ge---3.4 %) demonstrate that the critical attack angle is comprised between 22.5 and 30 deg, this higher value than 20 deg (relation (12)) could be due to elasticity. Experiments suggest that a prior workhardening of metals which reduces their strain sensitivity decreases 13cdespite the fact that ge is increased [25]. It is difficult to predict without any calculations the influence of friction because an increase in friction produces a very marked decrease in the thickness e of the strained layer (Fig. 2a) and in the shear angle r (Fig. 2b) and finally a simultaneous increase in go(13) and gp([3) (Fig. 3). This
Scratching of materials and applications
30
question remains open. Experiments suggest that 13c increases with friction which would promote ploughing flow [26]. But from a theoretical point of view an increase in friction reduces the highest value of the attack angle for which the plastic wave SLF model (Fig. 2a) can be built [21, 24]. Recent numerical simulations [28] demonstrate that for an increase in friction the plastic wave and the chip degenerates into an adhesive edge (or built up edge) which is similar to the SLF "wear" model [21 ] where the frontal edge is a dead zone separated from the material by a velocity discontinuity line. Finally, notice that in ploughing flow the contact pressure decreases with an increase in the attack angle or in friction [21 ].
Application to the three dimensional scratching All these conclusions are certainly qualitatively correct in three dimensional scratching but it is difficult to compare directly both the configurations. For example we are not able to define a shape ratio c (relation (5)) in 2D scratching. For the other quantities, to a first approximation, we can propose an equivalence through the apparent friction coefficient: for a conical indenter with a semiapical angle 0, the attack angle 132D of a wedge inducing the same apparent friction coefficient under zero friction is: fl2D =
a tanI2cotO /
(13)
In w 5-6 we consider the cone with 0=70.3 deg; it is the equivalent cone with equal volume for a given penetration depth to the Berkovich and Vickers pyramids in indentation. This cone would be equivalent to a wedge with the attack angle 132D--12.8 deg well below the critical attack angle and for which ep-~l.12 and p---2.45 o0. but for the same criterion the equivalent wedge attack angle of the Vickers (Berkovich) pyramid,would be 132D-~15.9 (13) deg for edge scratching and 132D-~22 (24.7) deg for face scratching, all values higher than that related to the previous cone and for face scratching higher than 13c. For the Rockwell C indenter (0=60 deg) the equivalent wedge attack angle is 132D"-20.2 deg, a value near 13c. for which ep---2.05 and p~2.15 or0. Finally the more acute indenter considered (0=20 deg) corresponds to 132D"60.2 deg. These results are of interest for the w 7.
CONDITIONS OF THE NUMERICAL SIMULATIONS OF SCRATCHING AND INDENTATION Computer Code and Computation Conditions Scratching simulations are modelled using the Forge3 * implicit code using an automatic remeshing procedure. The domain is a right-angled parallelepiped. Fig. 4 shows half of the finite-element mesh corresponding to the region x>0, with the plane x-0, a symmetry plane. The displacements of the mesh in the other directions were prevented by two planes y=0 and z=0 which are also considered as symmetry planes. The size of the domain was generally chosen (Table 1) so that boundary effects do not influence the results.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
31
Fig. 4. Three-dimensional view of the mesh used for the simulation of scratching. The mesh box placed near the indenter tip contains small sized elements and moves with the indenter along the y axis. The indenter is a rigid cone of semiapical angle 0 (20-70.3 deg).
Table 1. Size of the domain and the elements near the indenter in the scratch simulations. Maximum penetration h . . . . .
Width
Length
Height
Size o f elements near the indenter ,
1
8
50
10
0.04
The indenter is rigid and modelled as an axisymmetric cone with a semiapical angle 0 (20-70.3 deg). The displacement of the indenter is along the y axis with a scratch speed V=0,2 ~tm.s~ for polycarbonate, its penetration depth was constant, and is equal to h=0.4 ~tm for polycarbonate (w 7); this corresponds to an apparent strain rate i:ap ~ V/h = 0.5 s ~. In other cases V and h are kept constant, but their absolute values have no influence on the results because the indenter is a quasi-perfect cone and the materials are elastic perfectly plastic. Elements of the domain are three-dimensional meshes with four-node tetrahedra. Far from the indenter, elements have a typical length of about h. With the Forge3 * software, parallepiped boxes are used, where the mesh is refined and 20 nodes are at least in contact with the indenter on the generatrix in the plane x=0. For example, a scratch simulation on polycarbonate, a rather elastic solid (ce---4.2 %), require 11,000 nodes and 45,000 elements, about 35 hours of CPU time on a 400 MHz quadriprocessor SUN (manufacturing year 1998), and a remeshing procedure every five increments. The calculations require more CPU time and remeshings as the elasticity of the material decreases. For each time increment, normal and tangential forces were computed. A post processing procedure gives the average of several geometrical parameters such as the
Scratching of materials and applications
32
scratch width, b, the frontal pile-up height, hbf, the residual depth, hr (Fig. 1) and so on. The indentation test (w 5) is modelled with Forge2 | a two dimensional axisymmetric finite element code. A two-dimensional rectangular mesh incorporating six-node elements is constructed. Elements have a length of 0.04hm~x near the indenter and of 3hmax far from the indenter. The rigid indenter is modelled as an axisymmetric cone with a semiapical angle 0=70.3 deg. The CPU times required for the calculations of indentation are very much smaller than those required for the simulation of scratch test. More details concerning simulation of the scratch test and indentation test are given in [29,30]. Examinations of the results of the simulations of scratching demonstrate that despite our cautions they involve some scatter and errors due to numerical error. In addition, the numerical simulations described in w 5-7 have been performed in different steps of the study and we observed some difference in results related to the same nominal condition (X= 100, ~t=0). This is probably due to some difference in meshing-remeshing procedures which have not been always optimal. But it is reminded that at present time such a calculation of the scratch test remains a very difficult task and despite these limitations, we can draw some very clear conclusions about the main aspects of the studied problems.
Constitutive equations All materials are considered homogeneous. The inertial forces are assumed negligible. We assume that at each time the strain rate tensor is the sum of an elastic strain rate tensor and a plastic strain rate tensor (elastoplastic material):
= ~el + ~pl
(14)
The elastic behaviour is modelled by a linear law with two constant parameters: Poisson's ratio, t~, and Young's modulus, E; it means that the viscoelasticity of polymers is neglected. The yield condition is given by the von Mises yield criterion with the flow stress o0 and the associated flow law. We define the generalised strain rate and the generalised strain by the classical relations:
=
~lPjl~l.
g = I~ dt
(15)
For the polymer, the flow stress is described by a simplified G'sell-Jonas law [31]: - m exp ( hgE 2 ) o. 0 = o. loe
(16)
where 13' 1 is the strength coefficient, hg is the strain hardening coefficient and m the sensitivity to the strain rate. Isotropic hardening is assumed and so no Bauschinger effect (decrease in flow stress) occurs as the stress is reversed. The values of the rheological parameters used in the numerical simulations are summarised in table 2.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
33
Table 2 Mechanical properties and values of the friction coefficient used in the numerical simulation of the scratch test Material E (GPa) Material (w 5) E Steel (w 6-7) 210 Polycarbonate (w 7)__ 2.4
o 0.3 0.3 0.35
O'l (GPa) -0.35.103-0.35 E 0.75 (re-q3.35 %) 0.102 (~e-4.25 %)
m 0 0 0.053
h~ 0 0 0.5
la 0 0-0.2 0.3
The values of the plastic parameters of the thermoplastic polymer polycarbonate considered in w 7 have been deduced by an inverse method based on the interpretation by numerical simulations of the force-penetration curves of nano-indentation tests; in order to identify the various quantities these testings have been performed at various indenter speeds and with various indenter shapes [32]. They are in good agreement with the results of compression testings. According to the value of the apparent strain rate ~ap ~0.5 s ~, the effective yield stress of polycarbonate is about: O'y "" 102(0.5) ~176 --- 98 MPa---ol; so we will normalise for clarity the values of the hardness by Crl. For 0=70.3 deg, its indentation index is X---8.4 whereas workhardened steel considered in w 6 and 7 has an indentation index X-100. All thermal effects are neglected (cf. w 7). Coulomb friction with coefficient la is imposed at the contact surface between material and indenter (Table 2). The value used for scratching the polycarbonate has been deduced from friction tests performed with a spherical tip [29]. Calculations with friction are very much time consuming especially for material with low elasticity (typically re" 0.35 %). In addition as it requires much more remeshings higher numerical errors are expected. For more elastic material, such as polycarbonate, calculations are easier and quicker.
ELASTIC EFFECTS IN LOW STRAIN SCRATCHING AND INDENTATION OF ELASTIC PERFECTLY PLASTIC SOLIDS We study here the influence of the elasticity of the material characterised by ee during the scratch test performed by a cone with the semiapical angle 0=70.3 deg under zero friction (la=0); the variation of ee (table 2) induces a variation of the indentation index X between 1 and 103. The results are described in details in [33] and we shall here summarise the main conclusions. Notice that in [33] we name the shape ratio c2; we use here the notation c p =hffh proposed by Storakers and Larrson [34] for an indenter whose profile is described by a power law with an exponent p: a cone corresponds to p= 1.
34
Scratching of materials and applications
Rear part
.t x Frontal part ,, _ ~ .. Increase ',. ...... !...... " ,,fl i n X
--,,
h tan
"
m
o,"
~
"~ " -
X
i
"
-..
&X ~
Frontal part
~ " ,, Increase [_.. ........... . . . . . . ~ ~ t a X
',
""" %
Rear part
_r - t -t
I ~ ' . . Z ..... ~, ~" X>lO
T li ~ ..": / - " ~ L i n e of b d 2 = l I".. . . . ..xx-'" / / equal ch tane ] ] ....... " ~ s t contact
a) X< 60: Decrease of the rear contact angle ~ b) X > 60: Contact on the frontal part of the and increase of the contact height, indenter and increase of the height (hfc is 10 % greater than h~). Fig 5. Scratching elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Schematic evolution of the contact surface (top view) with the indentation index X for a given penetration depth h.
Flow Pattern and Contact Geometry As expected in w 3 the indenter induces in all materials a ploughing flow. Fig. 5 describes schematically for a given penetration depth h the evolution with the indentation index X of the contact geometry, one of the main unknown quantities of this problem. This evolution comprises three main steps: 9 For 1 <X
1 < X < 60 X>60
ct = 78.229 - 103.14L + 60.931L 2 - 15.387L 3 deg a=0
L = log X
( 17)
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
.
.
.
.
.
.
.
.
(x (deg)
'\
1
,
,
,
.v.,
,
,,i
9
.
.
.
.
.
.
35
.
60
"~ \'~
40
L=log(X) or.=78.229-103.14*LL: ~
20
1
0
,
,
9
....
v
1
01
10
x
100
Fig 6. Scratching elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution of the rear contact angle ot with the indentation index X.
A tentative extrapolation of the results suggests that tx = 90 deg for X---0.8 (Fig. 6)" it means that the contact would be limited by a full circle for X< 0.8. A first consequence of this evolution of the rear contact is that the apparent friction coefficient follows equation (8): Ix0 increases with X and for X>60 it attains a limiting value very close the value 0.228 predicted by the classical equation (7) (Fig. 7). This result is in agreement with the experimental values related to lubricated scratching of metals by cones (for which X>60) [4]. The validity of the relation (7) for high value of the indentation index is explained by the fact that according to the numerical simulation in first approximation the contact pressure p decreases linearly with the radial distance r from the tip of the indenter toward the frontal limit (Fig. 5b). In [33] we demonstrated that the contact model of Fig. 5a provides a correct value of ~t0 by the equation: /~0 =
cot 0
1 + (2a/n"
ct in rad
So the equations (17) and (18) provide the evolution of Ix0 with X.
(18)
36
Scratching of materials and applications O.25
0.228 = 2hr cote
~to
A ~kl----A
......
..,
O.2
0.15
x o.%
,'o
I 100
.....
1000
Fig 7. Scratching elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution of the apparent fiction coefficient l,t0 with the indentation index X.
Before considering more in details the frontal geometry in scratching, it is instructive to remember the geometry of the indentations in the same materials calculated by RamondAngelelis (Fig. 8) [30]. Under load the material sinks in for X<30 (c
30 (c>l). A very important result of these calculations is that elasticity has some influence in the whole range o f values of the indentation index because the indentation profile is not yet constant between X=200 and 1000. Even for X=1000, where the shape ratio c--1.25 we observe some slight elastic recovery during unloading. After unloading we observe in all cases an indent with a bulge. If the indent is not very marked for the quasi-elastic case X =1 and even X=5, the bulge is apparent for X > 10 and the radial distance to the indentation axis of the residual bulge top a' is related to the contact radius a under load by a'/a=l +8(X) with ~(X) a decreasing function of X which tends toward 0; in addition 5(X) < 10 % for X > 10: so the measurement o f the residual indent radius a 'provides an under-estimation of the true indentation pressure H=W/(rca 2) with an error lower than 20 %. i i
9
after unloading i i
_~0..o.2 - ._.. XX == IJ xX =: 5, ~ .0
-0.4
~-o.e
.c:
"15
x - - lo
X ---- 1 0
--
X x=2o =20
x X ==3 3o 0
-
-
.-. -0.8
~
-.
"--
",X,X\
v,,, _- w e.~ x=6o
-4
-3
I
\
",~',X\ \ ',~,~\ " ,,' I~v,~~1\ ~
X x ==8 0so x ==1 0lOO X 0 xX == 2OO 200
x='1000 ""/
X=5
"', 9
-'-"
under load
i N!
',.~,~\ i ",~,~
',,~,~
-2
-I
0
I
2
3
4
5
indentation radius
Fig 8: Indentation of elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution with the indentation index X of the indent profile under load and after unloading [30].
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
37
We observe very similar results in scratching (Fig. 9) with two main differences: first as noticed previously the material piles up for close to X>I 0 and not 30, and for the same value of the indentation index the frontal edge is higher and very much steeper. In addition we see that the elastic recovery of the material in the scratch symmetry plane (x=0) is significant for X < 30, but the initial slope of the profile behind the indenter tip is not sufficiently high to maintain the contact between the tip of the indenter and the material for X>I. Notice however that the calculation of the elastic recovery for higher values of X is not very accurate. Nevertheles the residual groove comprises always two shoulders more or less marked and their distance is very near the contact width: b~bc (Fig. 1). We describe in w 7 the profile of the groove under load in plane y=0. I
l
I
Y
x=,ooo
-
....
\
X=l
- - X = 5 --
X=lO
--
X=20
-.
X=30
X=60 9- - X = 8 0 ~
X=lO0
---
X-200
--
X = 1000
_
I
i
,._
iO y
0
-10
Fig 9. Scratching elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution with the indentation index X of the indent profile along the symmetry plane x=O [29]. 2
scratch
C
1.5
indentation
1
2/'71:=0"636 X [ i,
0.5 1
,
,
= ,,,,H
10
100
1000
Fig 10. Indentation and scratching of elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution of the shape ratio c with the indentation index X.
Scratching of materials and applications
38
The evolution with X of the shape ratio in indentation and scratching is reported in Fig. 10. This figure states precisely that the shape ratio is higher in scratching than in indentation; in addition we see two interesting points: 9 for X=I, c-~0.75 a value already significantly higher than the value c=2/n---0.636 related to a full elastic contact (see below); 9 for X=1000, c attains a high value c---1.78, but it is not sure that it is the limiting value related to the RPP solid. The evolution with the indentation index X of the shape ratio in scratching can be fitted by a polynomial regression in two steps (similar equations have been proposed for indentation in [30]): I0.74052 + 0 . 1 6 5 8 L - 0.01118L 2 + 0.12741L 3 c = 1,0.26092 + 0.98257L - 0.15885L 2
l<X<30
with L= logX
(19)
30 < X < 103
The first relation must be used with some caution for X<5 because we have simulated only the cases X=I and X=5. Nevertheless extrapolations for X 0.8, there would exist a range of values of the indentation index, 0.3 <_X<0.8, where the contact is limited by a full circle (quasi-elastic contact), but where in fact some plastic strain of the material occurs and where the contact radius is greater than its elastic value, for a given penetration depth. This quasielastic contact regime could be due to the gradual increase in size of the plastic core around the indenter tip.
Volume Average Plastic Strain and Scratch Hardness All these previous results suggest that scratching induces higher plastic strain than indentation. In order to state more precisely this point we have estimated the average plastic strain ev in the deformation volume V:
1 ~g dV EV =-Vv
(20)
The numerical procedure is described in details in [33]. We see in Fig. 11 that in first approximation for 2 < X _<200, ev increases in direct relation with logX and that it is 66 % about higher in scratching than in indentation. For higher values of the indentation index, cz becomes constant despite the fact that the contact geometry is not yet steady and it attains in scratching 0.936 a value very close to the value of 1.12 suggested by the 2D analogy for RPP solids (w 3). Notice however that for low elasticity materials with high indentation index, ev for indentation and scratching is much higher than the representative value 0.071 for indentation pressure suggested by Johnson (relation (2)).
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
39
~V
e, = 0.936
.
0.8
jt
e v = 0.562
1 01
. . . . . .
..,I,
9
I
I
i LI t,lt
10
10o
I
i
i
!
i
ll|l
1000
X = E/o0 cote Fig 11. Indentation and scratching of elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution of the volume average plastic strain ev with the indentation index X
Finally, Fig. 12 presents the evolution with the indentation index X of the indentation hardness H (true average contact pressure W/(na2)) and the scratch hardness Hs and Hsl/2. At first sight we observe that all these quantities have similar evolution with X, but in fact: 9 the indentation hardness H is an increasing function of X and as expected tends toward a limiting value -2.6 c~0 which corresponds to the limiting value of the constrain factor C"(70.3 deg)--2.6 (w 2); 9
9
the (true) scratch hardness Hs increases in a first time, attains a maximum Hsmax ---2.5 or0 for X-20, then decreases slowly toward 2.25 cr0 for X=103; so the ratio Hs/H is a slowly decreasing function of X varying from 1.22 toward 0.87 about as X increases from 1 to 103, the value 1 being obtained for X-22; the evolution of the apparent scratch hardness Hsl/2 is very similar to that of Hs. Because the area of the semi-disk with diameter the groove width b underestimates in all cases the real contact area, Hs]/2 is always greater than Hs, but Hsl/2/I-Is is first a quick decreasing function of X from 1.57 (X=l) to 1 (X=I 0) because the rear contact angle decreases quickly (Fig.6); for X >_ 20 it is about constant - 1.1 because the frontal edge is a little higher than the lateral edge (Fig. 5b): in the range X>_ 10 under zero friction the apparent scratch hardness Hst/2 is a rather good estimation of the true contact pressure Hs. It is interesting to notice that for X > 30 the relative difference between H and Hsl/2 is lower than 6 % and so in this large range of values of the indentation index the indentation hardness is very close to the apparent scratch hardness as it is commonly observed in testing at 20~ metals with high melting temperature [4].
Scratching of materials and applications
40
0
0
0
X
X
xH O
X l 10
,
i
Hslt2
zx H s
, i,,.l
.
tO0
i
|
. ..1, I000
Fig 12. Indentation and scratching of elastic perfectly plastic solids with a 70.3 deg cone under zero friction: Evolution of the indentation hardness H, the scratch harness Hs and Hsl/2 with the indentation index X.
These results confirm the great similarity between scratching and indentation commonly observed especially for X> 10, i. e. for common metals, the difference in scratch and indentation hardness becoming rather high for X<10 as for some polymers. An unexpected difference however is the decrease of the scratch hardness at high values of the indentation index. This decrease is limited and so this result would require some additional work for confirmation. But from a physical point of view we think that it is real and due to the increase with X of the slope of the free surface of the edge (Fig. 9): 9 the contact pressure for the equivalent wedge scratching (w 3) is p--2.45 G0, a value very close to Hsmax ---2.5 G0 obtained for X-~20 where the slope of the edge is very close to the slope of the indenter, as in the 2D scratching; 9 for higher values of X the slope of the edge increases and for the 2D scratching this produces a decrease in the contact pressure.
FRICTION EFFECTS IN LOW STRAIN SCRATCHING AN ELASTIC PERFECTLY PLASTIC SOLID Here we study the effect of the friction in the scratching of a work hardened steel (table 2) with the same conical indenter (0=70.3 deg). This material does not work harden and its indentation index is X = 100; the friction coefficient la increases from 0 to 0.2 in steps of 0.05.
Contact Geometry and Volume Average Plastic Strain For this value of the indentation index the material is in contact only with the frontal part of the indenter (see w 5). As la increases, the flow remains of the ploughing type and Fig. 13 describes schematically the evolution of the contact geometry with friction for a given penetration depth: in first approximation the contact width bc and the contact height he remain constant, an unexpected feature. On the contrary the frontal height hie increases significantly (Fig. 14). So
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
&X
Rear part
41
Frontal part
%
~
Increase
0 bd2-1.53h t a n O ~
Fig. 13. Influence of the friction coefficient ~t on the contact surface (top view) in scratching an elastic perfectly plastic material (ee-~0.35 %) with a conical indenter (0=70.3 deg-> X-~100)- h constant.
the top view of the contact is about half of an elliptic disk whose eccentricity increases with the friction (Fig. 13). So the friction produces an accumulation of the material on the frontal surface of the indenter and in addition we see on Fig. 14 that the slope of the frontal edge increases, a feature very similar to what occurs in 2D scratching [21, 24]. Unfortunately we have not observed more in details the flow field in the frontal edge in order to see if the increase in friction promotes the formation of a dead zone as in the SLF models of ploughing under high friction proposed by Petryk [24] and in the 2D numerical simulations [28]. So for 3D scratching this point requires further work.
Scratching of materials and applications
42
Z
p=0.2 ,,//,,0.15
,','z/
0.I
.
9
/
.
.
.
y
Fig. 14. Influence of the friction coefficient la on the profile of the frontal edge (side view, plane x=O) in scratching an elastic perfectly plastic material (ee-~0.35 %) with a conical indenter (0=70.3 deg--> X---100).
2.2 hfc/h
-~
-
2 1.8 _ J
h/h
-
I
1.6 1.4
1.2 -
~
1 y/ 0.8
1 0
0.05
~ 0.1
1 0.15
P 0.2
Fig.15. Influence of the friction coefficient la on the volume average plastic strain cv and the frontal and lateral edge height hfc and hc in scratching an elastic perfectly plastic material (ce-0.35 %) with a conical indenter (0=70.3 deg--) X---100) As a first approximation the frontal height to the penetration depth ratio hfc/h is a linear function of the friction coefficient la (Fig. 15). The analogy with the 2D scratching suggests that such an evolution corresponds to a reduction in the depth of the strained volume (w 3.2) and an increase in the average plastic strain in the material. We observe effectively that the volume average plastic strain ev increases markedly with ~t from 0.95 to 1.61 as la increases from 0 to 0.2 (Fig. 15) and this increase is very similar to that of the frontal edge height hfc.
Apparent Friction Coefficient and Scratch Hardness As expected, the apparent friction coefficient ~t0 increases with the (real) friction coefficient ~t (Fig. 16): This increase is linear as commonly assumed and is well described in first
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
43
approximation by the classical relation of Bowden and Tabor for modelling the friction between solids [3]" 2 /~o =--cotO +/~
(21)
7/"
0.45
Po 0.4
0.35
0.3
0.25
0.2 0
~ 0.05
J _ ~ 0.1
l
~ 0.15
P 0.2
Fig. 16. Influence of the friction coefficient la on the apparent friction coefficient ~t0 in scratching an elastic perfectly plastic material (ee---0.35 %) with a conical indenter (0=70.3 deg--) X-~100).
So our numerical simulation validates the model of Bowden and Tabor for 3D asperities similar to blunt cones or spheres with small penetrations. This model is different from the model proposed by Goddard and Wilman for acute cones [35] where the adhesion term is pondered by 2/(nsin0) and which is based on three main assumptions: 9 the contact pressure is uniform: in fact it is sufficient in first approximation that the lines of equal contact pressure are parallel to the contact boundary (Fig. 5b) and it is effectively the case in the numerical simulation; 9 the contact is limited on the frontal part by half of a circle and the flow lines at the contact surface are contained in planes parallel to the initial material surface. In fact the increase in frontal edge height produces only a slight increase in the ploughing term; therefore the form of the adhesion term suggests that the flow lines at the contact surface are contained in planes parallel to the plane x=0: it is not surprising in fact that this blunt cone does not induce a very significant sliding velocity component perpendicular to the scratch speed. An unexpected result is that for a given penetration depth the normal force W does not vary significantly with the friction; as the contact width bc---b is also constant, it implies that the apparent scratch hardness nsl/2 and the residual groove profile is quasi-independent on the friction for a given value of the normal force. This surprising result is in fact in agreement with experiments: the apparent scratch hardness of metals does not depend on the conditions of lubrication [4] and the residual groove profile does not vary significantly with the apparent friction coefficient [36].
44
Scratching of materials and applications
3
i
t
.
.
.
.
.
.
.
.
.
1
.
.
2
.
.
.
.
.
.
.
.
.
~
H
.
.
.
.
.
.
.
.
.
Io
i
1
I
0
"
p l o ~ (2D s c r a t c h i n g ) l 0.05
1 0.1
L 0.15
" i'1
I k
0.2
Fig.17. Influence of the friction coefficient p on the scratch hardness Hs in scratching an elastic perfectly plastic material (~e---0.35 %) with a conical indenter (0=70.3 deg--) X--100); Comparison with the contact pressure p in 2D equivalent scratching.
It is interesting to note that for such a blunt indenter and a less elastic solid (X=IOJ), the indentation force also does not vary significantly with friction, but the increase of friction from 0 to the maximal value (friction stress equal to the yield shear stress =or 0 / ~
) produces a
decrease of the shape ratio c from 1.25 to 1.16 and an increase of the contact pressure from 2.6 o0 to 3.12 o0 [30]. On the contrary, in scratching, because the frontal edge height increases, the (true) scratch hardness decreases as friction increases (Fig. 17): this decrease is initially limited, but becomes more marked for p > 0.1. Such an evolution is in agreement with the theory of the junction growth of Tabor [3]. It is easy to deduce from the current SLF model of ploughing [21] the evolution of the true scratch hardness for the equivalent wedge and the corresponding line is reported on Fig. 17 as p/o0: The decrease of p with p is linear and so it appears a shift between this line and the numerical results for la < 0.1, but for la > 0.1, the slope of the p line is very close to the slope of the line passing through the numerical values of Hs.
COMPARISON OF SCRATCHING OF WORKHARDENED POLYCARBONATE AT LOW AND HIGH STRAINS
STEEL
AND
Here we study the scratching ofworkhardened steel and polycarbonate (Table 2) performed with cones whose semi-apical angle 0 decreases from 70.3 to 20 (or 30) deg. One of the main objectives is to state precisely a transition from a ploughing flow to a cutting flow.
Scratching of Workhardened Steel As expected in w 3 we observe such a transition for scratching of steel under zero friction (Fig. 18). For 0=70.3 and 60 deg, the material piles up on the frontal surface of the indenter, the reduction in 0 producing only a significant increase in the frontal edge height. But for 0 = 45 and 30 deg, this frontal edge becomes clearly a chip whose thickness decreases with the cone angle.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
45
In all cases as expected the elastic recovery is small and the contact remains restricted on the frontal part of the indenter inside about half of a circle. So despite the change in the flow pattern the apparent friction coefficient la0 is well predicted by the relation (7). 303..~ : ,w 45 ~
- /:: [../[ .. ".
70.3~
/,-~
: _of {"/"" ".. , / ~'\ .-~t "~,: ..........."-. .':'..~.... : " .-"~ - ".:,'_..',~,,,pr
- .., ....~: t ..""
/...'" s
~""
,//
///
/
o
Y
2
4
'
~,
'
8
Fig. 18. Influence of the semi-apical angle 0 of a conical indenter on the scratch geometry (side view; plane x=0) an elastic perfectly plastic solid (ee'--0.35 %) (~t=0).
The apparent scratch hardness Hsl/2 is always greater than the (true) scratch hardness Hs (Fig. 19). But Hs decreases steadily with the cone angle and attains for 0=30 deg a value --1.12 or0, a little greater than the flow stress; Hs~/2 decreases in a first time, but remains constant and equal to --1.6 or0 for 0= 45 and 30 deg where a chip is formed (Fig. 19). Brookes et. al. [4] observed such an evolution for workhardened metals and reported that some chip is formed for 0 <45 deg. It is interesting to see that to a first approximation the evolution of Hsl/2 is very close to the evolution of the contact pressure p for the equivalent wedge scratching (relations (10) and (13)) for0> 45 deg (Fig. 19).
3
1
Cuttingploughing 2.5
Ploughing ,oo
-
9
Hslt2/~o .o
.~
Hs/C ~
.-"...""
,.,.. r-"
/;,Oo (2D
scratching)
1.5
0 (deg) j
1 30
40
50
60
.
i
70
80
Fig. 19. Influence of the semi-apical angle 0 of a conical indenter on the scratch hardness of an elastic perfectly plastic solid (eel-0.35 %) (la=0); Comparison with the contact pressure p in 2D equivalent scratching.
46
Scratching of materials and applications
Fig. 20 provides the groove profile under load (plane y=0) and its plastic strain map for the four cone angles. We must notice again first that the calculations of such quantities are particularly difficult and some errors are expected. But it is clear that for 0 = 45 and 30 deg the grooves contain very significant lateral bulges and that the flow pattern is in fact a mixture of ploughing and cutting as it is most ot~en observed in experiments [4,7] and theoretically predicted in 2D SLF scratching models [24]. The lines of equal value of the plastic strain are in first approximation parallel to the indenter surface and the line g = 0.6 which is already a high plastic strain defines more or less the boundary of the lateral bulge in all cases; so we must not forget that very significant plastic strain occurs below this line in the bulk of the material. This strain map is very different in shape and intensity of the map of plastic strain in indentation of elastoplastic solids [ 19] where a radial field occurs and where the plastic strains are very much lower. So scratching a low elasticity material (ee ---0.35 %) induces a high level of plastic strain a result which is very clearly illustrated by the maximal plastic strain gmax : it is attained near the indenter tip as expected and already for 0= 70.3, ~-max-~1.5 a value significantly higher than the volume average plastic strain ~v---0.936 (fig. 11). gmax increases dramatically as the indenter angle decreases in first approximation as cot0 in agreement with the relation (1): grnax ~. 1.5 4- 5.6 - 8.4 for 0 = 70.3 - 60 - 45-30 deg.
Fig. 20: Influence of the semiapical angle 0 of a conical indenter on the groove profile under load (plane y=0) and its plastic strain map in an elastic perfectly plastic solid (ee-0.35 %) (la=0)
Finally we can notice that some preliminary numerical simulations of the scratching of workhardened steel with the 30 deg cone under high friction (la=l) suggest that the chip transforms into a build up edge as in numerical simulation of 2D scratching [28].
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
47
Scratching of Polycarbonate It appears clearly on the side view of the scratch (Fig. 21) that no such transition occurs in polycarbonate even for the lowest cone angle 20 deg; the flow pattern is always of ploughing type with a frontal edge whose height does not vary significantly with 0. The origin of this difference with steel however requires some discussion because the simulations related to polycarbonate have been performed with a friction coefficient ~t=0.3 whereas the ones related to steel with zero friction. In addition the rheology of the two materials presents very great differences (Table 2). 9 We have already briefly discussed the influence of the strain rate sensitivity which does not seem very important (w 4) because the strain rate exponent m is rather low (0.052). 9 For 0=70.3 deg the apparent index of indentation of polycarbonate is rather low: X=(E/Ol)COt0---8.4; examination of the contact surface demonstrates effectively that there exists a rear contact surface and so the apparent friction coefficient is ~to---0.527 a value little lower than the value provided by the relation (21) : 0.546; so these results are not very different from the results related to the elastic perfectly plastic solids studied in w 5 for the same value of the indentation index. However we observe in Fig. 21 the formation of a frontal edge which is not in agreement with the results of Fig. 9; the real scratch hardness is effectively lower. The friction level explains these facts: we have observed that for polycarbonate scratched by a conical indenter with a semiapical angle 30 deg and having a spherical tip ("rounded indenter A") the frontal edge height increase with friction as for the workhardened steel (Fig. 14) [29]. 9 Finally, we have very clearly demonstrated that the very great stability of the ploughing flow for acute cones is due to the high workhardenability of the polycarbonate: if we limit the increases in flow stress at a plastic strain level 1.5, which corresponds to 3 times increase of the strength according to the relation (16) and the values of table 2, we observe the formation of a chip during scratching by the "rounded indenter A" under the same friction coefficient ~t=0.3, despite the fact that this kind of indenter is less acute than the perfect cone with the same semiapical angle [37]. z (/~m)
300 45~ 60~ 70.3 ~ 20~ " ' ' ~ / ....... ,,--, 9
; o
/
/
i/ / / /
:
/
,
ii
#
I / I II -
.:/ ./
~"'~.-,:-.-.:..-.s |
#
y (~m) ,
0
I I
,
I 2
3
Fig. 21. Influence of the semiapical angle 0 of a conical indenter on the scratch geometry (side view; plane x=0) ofpolycarbonate (~e~-4.25 %; hg=0.5) (~t=0.3) At this time, it is very instructive to observe the groove profile under load (plane y=0) and its plastic strain map (Fig. 22) for the four cone angles. They are very significantly different from those related to steel (Fig. 20). First a residual groove exists in all cases, but it is very much
48
Scratching of materials and applications
diffuse than for steel; careful examination demonstrates that the residual groove width b is greater than the contact width bc especially for the most acute cones (Fig. 22); so the groove width in scratching polymers with acute cones does not provide a reliable estimation of the real contact geometry. We observe that the plastic strain field is more of the radial type and its intensity is much lower: for example, the line E =0.15 defines the limit of the residual edge against 0.6 for steel. This lower level of plastic strain is clearly due, in the main, to the high elasticity of the material (4=4.25 %) because for 0=70.3 deg the plastic strain map is not very different from the one related to elastic perfectly plastic material with the indentation index X---10 and It=0, the higher level of plastic strain being due to friction
g m a x 1.5 against
approximately 1 [29]. gmax increases slowly as the cone angle decreases because it is about 2.5 for the most acute cone. Clearly the high work hardenability of polycarbonate insures a dilution o f the displacement on a greater depth and so reduces the plastic strain level and this effect is especially important as the cone angle decreases: for this reason the maximal plastic strain does no more increase in direct relation with cot0.
Fig. 22. Influence of the semiapical angle 0 of a conical indenter on the groove profile (plane y=0) under load and its plastic strain map in polycarbonate (ee'--4.25 %; hg=0.5) (~t=0.3).
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
49
For 0=70.3 deg the apparent friction coefficient Ix0 is well predicted by the relation (21) as we have previously noticed, but due to the complex evolution with the cone angle of the flow and the contact geometry this relation predicts less and less well Ix0 as the cone angle decreases; the discrepancy being maximal for the most acute cone 0 = 20 deg. In this case, the numerical simulation provides the values la0=l (ix=O) and 1.26 (ix=0.3) against ix0=1.74 (ix=0) and 2.04 (ix=0.3) for the relation (21). So the contribution of the adhesion term remains correct, the discrepancy being due to the ploughing term which is more and more overestimated as 0 decreases because the rear contact surface increases as the cone angle decreases: the elastic strain increases due to the increase in the contact pressure as we see it below. The fact that the flow lines remain in planes parallel to the plane x=0 in first approximation as suggested by the adhesion term of the apparent friction coefficient is another confirmation of the increasing influence of elasticity which restricts the deviation of the flow lines from the scratch speed.
\ ~Hs/G \
-
1
\
\ -
"~ \
4
2
\ \
-
\-
H
lo
i
e (deg) 0
l
2O
30
I
...... 1
l
l
40
50
60
70
80
Fig. 23. Influence of the semiapical angle 0 of a conical indenter on the scratch hardness of polycarbonate (ge-4.25 % ; hg=0.5) (IX---O.3). Figure 22 provides the evolution of the scratch hardness with the cone angle. Due to the high hardenability of polycarbonate and despite the decrease in contact pressure observed for perfectly plastic steel as 0 decreases (Fig. 19), Hs increases steadily and attains a very high value ---10 o~ for 0=20 deg. On the contrary the apparent scratch hardness Hs~/2 which is higher than Hs for the blunter cones increases initially as 0 decreases, is maximal ---3.5 ol for 0=45 and 30 deg and then decreases a little as observed in experiments [8]. This difference in evolution of the two scratch hardness values is due to the difference between the residual groove width and the effective contact width (Fig. 22). These results explain the high scratch resistance of polymers which can be improved by changing their structure and so increasing their strain hardening coefficient hg [29, 37].
Discussion of the Results and Nature of Further work It is now time to perform an estimation of the limits of such a work on scratching of metals and polymers. First, we shall recognise again that because metals have low elasticity, they undergo high plastic strain in scratching, and so the level of numerical error is certainly higher than for high elasticity materials such as polymers. In addition this numerical error increases certainly as the cone angle
50
Scratching of materials and applications
decreases or the friction increases. Nevertheless the comparison with available experimental results demonstrates clearly that the main conclusions are still valid. It remains to perform more accurate comparisons by taking into account more detailed rheological laws. For metals this task seems relatively easy because their elastic strains remain in all cases very low and well described by the classical theory of elasticity. Their flow stress at constant temperature can be described in a large range of strain and strain rate by power laws: (22)
O.0 = O.1 (E 0 + oF)n ~ m
Notice that such a rheological behaviour induces some reduction in the plastic strain level with respect to perfectly plastic solids and so reduces the numerical error level. Nevertheless some difficulties remain and require some consideration. The increase in temperature T could not be negligible and in this case the influence of T on the parameters of (22) must be included, the energy equation must be integrated simultaneously with the appropriate boundary conditions. So the problem becomes rather complex. But it is interesting to remark that the intensity of the thermal effect can be characterised by the Peclet's number Pe:
Pe= ~=Vbxb 2 ( ~ )
with
strain rate
~ = V
b thermal diffusivity tr
(23)
The loss of the mechanical energy involved in the plastic strain of the material by heat conduction restricts more and more the increase in temperature as Pe decreases. So reducing in proportion the penetration depth h (and so the groove width b) and the scratch velocity V in order to maintain the order of magnitude of the strain rate of the material permits to reduce the thermal effects without changing the level of strain and strain rate: we thus observe all the interest of the nanoscratch testers which can impose very low penetration depths. For example for a steel (~:~ 10 mm2.s-l) with h=l ktm and V = 1 lam.s1, ~ ~ 1 s! and Pe --- 10 "7 ! For this very low value of Pe the scratching can be considered as isothermal with a very good approximation. Despite their very much lower thermal diffusivity (K:- 0.1 mm2.s -l) this approximation remains valid at this scale of testing for polymer. So the nanoscratch test appears as a mean to study large strains of metals and polymers at high strain rates under quasi-isothermal conditions. However due to the inhomogeneous state of strain and strain rate interpretation of the test requires at present time numerical simulations. We must notice however that at these small penetration depths rheological description of even metals is not so easy as flow stress is affected by the size effects due to the discontinuous character of the plastic deformation which involves the movement of dislocations [38]. For polymers the problem appears as still more complex because their rheological behaviour even at macroscopic scale is very complex [31, 39-41]. One question emerges: Is the effect of the plastic strain on polymers well described by relation (16) ? At present time, the answer is "yes". We have demonstrated that saturation in the hardening produces for acute cones the formation of a chip which is not observed experimentally. Another interesting validation is provided by the experimental results obtained by scratching another thermoplastic polymer, the poly(methylmethacrylate) (PMMA), by the
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
51
Berkovich pyramid [42]. The scratchings are performed on a nanoindentation device on which it is possible to scratch the material with a rather high normal force (which corresponds to a penetration depth h---1.4 gm, the scratching velocity being V---10 gm.s -~) and after to measure the residual groove profile and the local nanohardness of the material with a much lower normal force and penetration depth (for a penetration depth h---30 nm). So it appears that the residual groove depth hr (Fig. 1) is very low: 0,1 h and 0,23 h for respectively edge and face scratching; in parallel the residual groove height hb increases. The nanohardness before scratching is about H0--,0.3 GPa; the nanohardness of the material of the tops of the residual groove is equal to H~---0.5 GPa and 0.7 GPa for edge and face scratching respectively. It is possible to deduce from such values an estimation of the equivalent residual plastic strain of the material with the following relation suggested by the rheological equation (16):
(24)
In
e-res ,~
The rheological law of PMMA is very similar to the one of polycarbonate (it has an equal value for the workhardening coefficient hg [29]). Application of relation (24) suggests that the residual plastic strain attains respectively 1 and 1.3, values in agreement with the results of the numerical simulation of edge and face scratching with Berkovich pyramids [29]. In addition, the numerical simulation demonstrates that the plastic strain in the groove centre for face scratching is still higher about 1.8. So although the residual penetration depth is very low the residual mechanical properties of the polymer are well affected as described by the relation (16). Finally it is instructive to try to evaluate the elastic strain component, a representative plastic strain and a total scratch strain starting from the numerical values of the scratch hardness of polycarbonate. For this purpose we propose the following relations: el Hs raP= E .
Cap =
_
_
In
(25) .rn
(7"10eap el oesap = Cap + Cap
We choose a constrain factor C=1.5, independent on the cone angle, a debatable assumption, but in fact the conclusions that we can draw from the numerical application of these relations (Table 3) remain valid if we change this value. We add the evaluation gap =1.5 cot0 for two reasons: Firstly this expression permits to understand the evolution of the apparent scratch hardness of metals with the cone angle (power hardening law (23) with m-0), secondly the relation (25) does not provide a reliable evaluation of the apparent plastic strain for the bluntest cone where the strain hardening has no significant influence on the scratch hardness according to (16) as we can easily verify it.
Scratching of materials and applications
52
From the results obtained with these formulas (Table 3) we can draw the following conclusions" 9 First the elastic component of deformation increases greatly as 0 decreases and attains 41 % for 0=20 deg; clearly the description of such a (visco)elastic deformation would require more elaborate models than the classical theory of (small) elastic strain. 9 We see in addition that for 0=60 deg, both the evaluations of the apparent plastic strain 6--ap are in rather good agreement and so the representative plastic strain proposed for 9
9
metals whose hardenability is very much lower seems reliable for 0=70.3 and 60 deg. But for the most acute cones 0=45-30-20 deg which tend to induce higher plastic strains it is the relation (25) which provides the best estimation because the high hardenability of polymers restrains the level of plastic strain as previously seen (see Fig. 22). For the three bluntest cones 0=70.3-60-45 deg the part of the elastic component is nearly constant -~10 % whereas it increases dramatically for the most acute cones 0=30-20 deg. So at present time it is rather difficult to propose a simple model of scratching of polymers.
Table 3. Evaluation of the elastic and plastic component (the most reliable estimation is in bold character) of deformation deduced from the results of the numerical simulation of scratching of polycarbonate (C=1.5- eap=l sl). 0 (deg)
70.3
60
45
30
el rap
0.064
0.089
0.138
0.29
0.41
~ap (25)
0
0.82
1.25
1.74
1.93
1.5cotO
0.53 0.6
0.86 0.9
1.5 1.4
2.6 2
4.12 2.3
10,7
9.7
10
14.3
17.5
OOsap .--
el eap/ O~ap (%)
,
20
Another limitation of our theological description of polymers is the assumption of isotropic hardening. After reversing the stress for example in shear, yield occurs in fact for lower stress [31]. Such a phenomenon certainly plays a role during the recovery at the rear part of the contact: our calculations overerestimate greatly the residual penetration depth hr (Fig. 1). Recently Kermouche et. al. [43] have demonstrated by numerical simulation of the scratching with a sphere at low penetration depth that the recovery increases greatly for a material having kinematic hardening. Finally, the real conical indenters have always a tip defect which is very commonly assumed spherical. The study of the influence of this indenter tip radius R which can be significant in nanoscale testing is to our knowledge well developed in indentation but not in scratching. It is easy to apply the current way usually used in analysis of indentation in order to define some equivalence between conical and spherical indenters [19]. To a first approximation, the spherical tip reduces the effective attack angle 13 of the indenter for a penetration depth h lower than the spherical cup height h*, so we have:
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
a ~ tan p.-.-~--
-->X.-.- E ~ O'y
h < h* = R(1 - sinO)
53
(26)
So the increase in penetration depth is equivalent to an increase in the indentation index. This formulation provides a simple way for applying the present results to real indenters. Evidently, further work is required in order to evaluate the reliability of this equivalence law and refine it.
CONCLUSIONS Despite some limitations in the comparisons with the experimental work, the present work demonstrates the power of the numerical simulation in order to develop the mechanics of the scratch test of materials. 9 First we propose a very simplified 3D cone scratching model based on an analogy with 2D wedge scratching which provides some very simple explanations of the results of the numerical simulation in the case of low elasticity material (metals): level of volume average plastic strain, influence of friction on contact pressure and influence of the cone angle on the flow pattern and the scratch hardness. 9 We state precisely the similarities on the influence of elasticity of materials in indentation and scratching: the real contact pressures are nearly equal in indentation and scratching of elastic-perfectly plastic material under zero friction for the axisymmetrical cone equivalent to the Vickers pyramid. 9 We also elucidate the influence of the indentation index on the rear contact surface, a problem specific to scratching: As a first approximation, the contact surface is limited partly by a circle; as the indentation index X increases, i. e. for less and less elastic materials, the contact radius increases for a given penetration depth whereas the rear contact angle decreases and this decrease induces an increase in the ploughing friction. 9 Numerical simulations demonstrate that scratching induces higher plastic strain than indentation. They validate many assumptions commonly adopted in the analysis of scratching of metals: contact geometry (half of a circular disk), relation between cone angle, friction and apparent friction coefficient. 9 We confirm also some other common assumptions about the main differences between indentation and scratching especially on the influence of friction: For low elasticity materials and blunt indenter the increase in friction produces a reduction in contact radius and an increase in contact pressure in indentation whereas the inverse occurs in scratching. The increase in friction produces mainly an increase in the frontal edge; so the frontal part of the contact is limited by half of an ellipse whose minor axis length is nearly constant and the apparent scratch hardness does not vary. 9 Finally, we demonstrate that low cone angles induce a composite flow cutting-ploughing in scratching elastic perfectly plastic materials as workhardened metals whereas the high hardenability of polymers stabilises the ploughing flow till the lowest cone angle investigated (20 deg), restrains the plastic strain and increases greatly the contact pressure. These results explain the high scratch resistance of polymers which can be improved by changing their structure and so increasing their strain hardening coefficient hg.
Scratching of materials and applications
54 ACKNOWLEDGEMENTS
Essilor INTL Coatings Research and Development is acknowledged for its interest for the results and its financial support for part of this work. We wish especially to express our thanks to Dr. G. Hochstetter and M. A. Jimenez for hepful discussion during the course of this work.
REFERENCES
13.
de Rraumur, R. A. F. L'Art de Convertier le Fer Forge en Acier, Paris 1722, as referred in Petty E. R. (1971). In Measurements of mechanical properties, Techniques of metal research, Vol V Part 2, pp.158-220, Bunshan R. F. (Ed), J. Wiley & Sons, New York. Tabor D. (1954) Proc. Phys. Soc. B67, 249. Bowden P. and Tabor D., (1950-1964). The friction and lubrication of solids, Clarendon Press, Oxford. Brookes C. A., Green P., Harrison P. H. and Moxley H. B. (1972). J. Phys. D: Appl. Phys. 5, 1284. Williams J. A. (1996). Trib. Int. 29, N~ 675. Mulhearn T. O. and Samuels L. E. (1962). Wear 5, 478. Buttery O. and Archard J. F. (1970-71). Proc. Instn Mech. Engrs 165 (43/71) 537. Briscoe J., Pelillo E. and Sinha S. K. (1996). Polym. Eng. Sci. 36, 2996. Ni B. Y. and Le Faou A. (1996). J. Mater. Sci. 31, 3955. Jardret V., Zahouani H., Loubet J. L. and Mathia T. G. (1998). Wear 218, 8. Gauthier C. and Schirrer R. (2000). J. Mater. Sci. 35, 2121. Bertrand Lambotte P. (2001). Sur les m~canismes de rayure des vernis de finition automobile, Th~se de Doctorat en Mrcanique, Ecole Centrale de Lyon. Gauthier C., Lafaye F. and Schirrer R. (2001). Trib. Int. 34, 469.
14. 15. 16. 17. 18. 19.
Childs T. H. C. (1970). Int. J. Mech. Sci. 12,393. De Vathaire M., Delamare F. and Felder E. (1981). Wear 66, 55. Gilormini P. and Felder E. (1983). Wear 88, 195. Torrance M. A. (1988). Wear 123, 87. Tabor D., (1970). Rev. Phys. Technol., 145. Johnson K. L. (1970). J. Mech. Phys. Solids 18, 115.
20. 21. 22. 23.
Hill R., Storakers B. and Zdunek A. B. (1989). Proc. R. Soc. Lond. A 423, 301. Challen J. M. and Oxley P. L. B. (1979). Wear 53, 229. Lee E.H. and Shaffer B.W_ (1951) J. Appl. Mech. 18,405. Lebouvier D., Gilormini P. and Felder E. (1985) Proc. EUROTRIB 85, Ecully, Elsevier, Amsterdam, Article 5-3.7, 10 p. Petryk H. (1987). In Friction, lubrication and wear, pp. 987-994, Int. Conf. Tribology, London, I. Mech. E., C140/87,. Black A. J., Kopalinsky E. M. and Oxley P. L. B. (1988). Wear 123, 97. Kailas S. V. and Biswas S. K. (1993). Wear 162-164, 110. Samper V. and Felder E. (1990) In Mechanics of Coatings, pp. 271-279, Tribology Series 17, Dowson D., Taylor C. M., Godet M. (Eds) Elsevier Amsterdam. Felder E. and Rouault de Coligny P. (2004).pp. 203-210, Proc. 2"d International Conference on Tribology in Manufacturing Processes ICTMP 2004, Nyborg, Denmark.
.
3. .
.
6. 7. 8. 9. 10. 11. 12.
24. 25. 26. 27. 28.
Mechanical analysis of the scratching of metals and polymers at moderate and large strains
29.
30.
31. 32. 33. 34. 35. 36. 37 38. 39. 40. 41. 42. 43.
55
Bucaille J. L. (2001). Simulation num~rique de l'indentation et de la rayure des verres organiques, Th~se de Doctorat en Sciences et G6nie des Mat6riaux, Ecole des Mines de Paris. Ramond-Ang6161is C. (1998). Analyse m~canique des essais d'indentation sur mat~riaux ~lasto-plastiques homogbnes ou multi-couches. Application ~ la caract~risation de la rhdologie et de la tenue m~canique des films minces, Th~se de Doctorat en Sciences et G6nie des Mat6riaux, Ecole des Mines de Paris. G'Sell C. (1995). In Introduction ~ la m~canique des polym~res pp. 141-168, G'Sell C., Haudin J. M. (Eds) Institut National Polytechnique de Lorraine. Bucaille J. L., Felder E. and Hochstetter G. (2002). J. Mater. Sci. 37, 3999. Bucaille J. L., Felder E. and Hochstetter G. (2001 ). Wear 249, 422. Storakers B. and Larsson P. L. (1994). J. Mech. Phys. Solids 42, N~ 307. Goddard J. and Wilman H. (1962). Wear 5, 114. Moore A. J. W. and McG. Tegart J. (1952). Proc. R. Soc. Lond. A 212, 452. Bucaille J. L., Felder E. and Hochstetter G. (2004). ASME J. Tribol. 126, 372. Gane N. and Skinner J. (1973). Wear 24, 207. Boyce M. C. and Arruda E. M. (1990). Polym. Eng. Sci. 30, 1288. Lucas B. N., Oliver W. C., Pharr G. M., Loubet J.-L. (1997). Mat. Res. Soc. Symp. Proc. 436, 233. Hochstetter G., Jimenez A. and Loubet J. L. (1996). J. Macromol. Sci .Phys. B38, N ~ 56, 681. Adams M. J., Allan A., Briscoe B. J., Doyle P. J., Gorman D. M. and Johnson S. A. (2001). Wear 251, 1579. Kermouche G., Barge M. and Bergheau J. M. (2003).In Assemblages : des mat~riaux la structure, pp. 321-324, Actes du Colloque National MECAMAT, Aussois.
56
CHAPTER 3
FRICTION, DAMAGE AND STICK-SLIP IN THE SCRATCHING OF POLYMERS
S.L. ZHANG
Department of Mechanical Engineering, University of Rochester, Rochester, NY 14627 Present address: Xerox Corporation, Bldg. 111-30N, 800 Phillips Road, Webster, NY 14580 E-mail: [email protected] ABSTRACT This chapter presents results of friction, scratch morphology and stick-slip characteristics during the scratching of a selection of polymers (uncrosslinked and crosslinked poly(n-butyl acrylate) (PnBA) coatings, poly(dimethylsiloxane) (PDMS) coatings, and a bulk styrene-acrylonitrile (SAN) copolymer). The chapter describes in detail the dependence of the horizontal force or friction force on the normal load and driving speed, scratch damage and mechanism, critical normal load for scratch damage, and slip process of stick-slip motion. During the scratching of PnBA and PDMS coatings, the horizontal force appears proportional to the contact area that depends on the normal load, and the scratching is a rate process in which the viscoelastic property of the coatings influences their frictional behavior. Below a critical normal load, crosslinked PnBA and cured PDMS coatings recover elastically after being scratched so that there are no scratch marks left behind. The study on the slip process during the scratching of SAN copolymer indicates that during slip the scratch velocity starts from zero, increases to a maximum and then decreases to zero again. The scratch groove made during slip shows a nonuniform depth, which increases with the decrease of scratch velocity.
KEYWORDS Scratching, friction, poly(n-butyl acrylate) (PnBA), poly(dimethylsiloxane) (PDMS), styreneacrylonitrile (SAN), coatings, elastomers, crossslinking, JKR theory, adhesion, damages, cracking, delamination, stick-slip, slip velocity.
INTRODUCTION Polymer coatings have important tribological applications. Their friction and wear properties are of practical importance and of considerable scientific interest. There are many parameters that can be expected to affect the frictional properties of polymer coatings. However, this subject has not received sufficient attention to date [ 1]. Scratch testing is an important tool for the characterization of the tribological properties of materials. In this test, a hard object such as an indenter or a stylus is pressed onto the material surface under a normal load and moved by a horizontal force to produce a scratch in the
Friction, damage and stick-slip in the scratching of polymers
57
specimen surface. Scratch testing is one of the more frequently used methods for determining the mechanical resistance of coatings. It is a relatively quick and efficient way of acquiring information on the dominant damage mechanisms of a coating. So far, mechanical studies of this kind have been largely confined to those coatings that are relatively hard and tough [ 1]. A major area of interest is their resistance to detachment, that is, for the determination of the critical normal load for de-adhesion of the coating from the substrate. In contrast, more compliant coatings, such as elastomers, have received much less attention. For such systems, an area of interest is their tolerance to cohesive damage, which may affect the physical and chemical properties of the coatings. It is well known that several possible failure modes can occur [2, 3], and only some of these are the cause of detachment at the coating-substrate interface; other failure modes may depend on plastic deformation and fracture within the coating. The friction force and the scratching resistance of a material could be influenced to varying degrees by attractive forces between the two contact surfaces. In a careful study [4] of contact between two rubber spheres, the area of contact was found to be larger than that predicted by the Hertz theory [5, 6]. This was attributed to the van der Waals forces attracting the rubber spheres to one another. The van der Waals attraction adds an additional force to the applied normal load, thereby increasing the area of contact. One characteristic feature of elastomers is their ability to recover from large deformations. Their friction coefficient is generally high when they are sliding under a dry condition on another material [ 1]. Hysteresis phenomena may occur in their frictional behavior, and stick-slip otten takes place [7]. When a stylus or an asperity slides across the surface of an elastomer, the material is compressed in the front of the contact and extended at the rear. A raised lip is otten formed in the front of the contact. The associated subsurface cyclic compressive and shear deformations lead to viscoelastic hysteresis losses, which create a frictional resistance to sliding but result in no permanent damage if the yield stress is not exceeded. Therefore, elastomers usually have good erosive and abrasive wear resistance because of their ability to deform elastically [8]. In this chapter, friction and damage characteristics are studied for poly(n-butyl acrylate) (PnBA) and poly(dimethylsiloxane) (PDMS) coatings scratched by a spherical indenter. Crosslinked PnBA and cured PDMS behave as elastomers. This study will shed some light on the scratching behavior of elastomeric coatings. Stick-slip processes between two sliding surfaces are commonly observed and occur in a wide range of length scales from atomic to macroscopic [9-11]. The stick-slip motion is a frictioninduced oscillation, which is a result of the interaction of an elastic system with a frictionally slipping surface and observed usually at slow sliding speeds. It is undesirable in precision movements and quiet operations. Sometimes the severe vibration can even cause damage and failure of machine components. The widely accepted cause for the stick-slip phenomenon is that the static friction coefficient exceeds the kinetic friction coefficient [ 11, 12]. While static friction is the force required to break away from a stick condition, kinetic friction is the force needed to maintain a relative speed of motion. The stiffness of the loading system may play a crucial part in the dynamics of stick-slip motion. Stick-slip is expected to occur if the loading system is soft and the relative
Scratching of materials and applications
58
sliding velocity is low [ 13]. A dynamic analysis [ 10] shows that there exists a crucial velocity beyond which no stick-slip motion occurs. During scratching, if the sample or the indenter is driven by elastic members such as a spring or a metal rod, stick-slip phenomenon may be observed also; i.e. the sample and the indenter may stick together until the elastic forces become sufficient to jerk them loose and then sliding occurs until they become stick again when the scratching velocity becomes zero. Because the scratch is formed during slip, it is necessary to investigate the nature of the slip process to answer the question as to what is happening during the slip stage. In the last part of this chapter, an experimental study of the motion during the slip process in the scratching of a bulk styreneacrylonitrile (SAN) copolymer is presented.
SCRATCH TEST The apparatus for scratch testing was built in-house [14]. A schematic representation of the scratch tester is shown in Fig. 1. Scratches were made by moving the specimen under a vertically loaded stationary indenter. The specimen was mounted on an air-bearing slider with almost frictionless motion (friction coefficient < 0.01 by experimental determination). The displacement of the slider was driven by a step motor (with a step size of 16 nm/step) that could be controlled to move at a constant speed (0.4 ~trn/s to 2 mrn/s) through a spring system.
1111111
Counter Balance Weight
[~ ~ Sliier~ - - 1 ~
,StepM o t o ~ [ ~ e a r i n g (
WeightBalance )
[ Orri i Condi] tionerI Horizontal D,,~isplacem ent ~ Fig. 1. Schematic diagram of the scratch tester.
..... I
i1 ~-
Normal Force
Indenter Specimen
Friction, damage and stick-slip in the scratching of polymers
59
The indenter holder was attached to a double diaphragm made of two layers of six springs, which provided rigidity in the X-Y plane, but allowed free vertical motion of the indenter. The actual normal load applied on the sample was measured by a balance on which the slider was positioned. The horizontal force was measured by a load cell attached to the motor. The output from the load cell was amplified, low-pass filtered, and then sampled by an A/D converter. The scratching experiments began with the lowering of the indenter onto the sample surface and preloading it to a specified normal load. The motor was then triggered to move the specimen at a certain speed for a certain time period. Both the normal load and horizontal force were recorded by a computer. The indenter tip was cleaned before each test for the removal of any debris accumulated on the tip during scratching. The scratches were made at different normal loads and various driving speeds. Scratch testing and other experiments were all done at the ambient temperature.
FR/CTION AND DAMAGE IN THE SCRATCHING OF POLY(N-BUTYL ACRYLATE) COATINGS Acrylic elastomers such as poly(n-butyl acrylate) (PnBA) are widely used in coating applications because of their inherent thermal stability, oil resistance and adhesive properties [ 15]. These same features make acrylic elastomers attractive for fundamental studies of scratch deformation. Although crosslinked PnBA behaves as an elastomer, uncrosslinked PnBA is a soft plastic with a sticky surface. A scratch study of this coating would also help us to understand the scratch resistance of generally soft and sticky materials, such as photographic film in photoprocessing solutions. In this section, we describe the experimental results for the friction and damage characteristics of PnBA coatings scratched by a spherical sapphire (76-~tmradius) indenter. The coating (15-235 ~tm thick) was coated on a polyethylene terephthalate (PET) substrate (100 ~tm thick). Some PnBA coatings were crosslinked by the addition of an aziridine crosslinking agent at a concentration of 10 wt% based on the total weight of the coating. The coatings were applied from solution with a 70/30 acetone/Dowanol PM solvent system, air-dried 5 min at 50 ~ and then dried in a convection oven for an additional 15 min at 100 ~ to remove the residual solvent and enable crosslinking. The samples were cut in the shape of a rectangle of 15x10 mm 2. The morphology of the scratches was examined with optical microscopy and scanning electron microscopy (SEM). Because the PnBA coating is nonconductive, the sample for SEM was coated with a very thin gold layer (ca. 10 nm thick) by thermal evaporation.
Scratching Behavior of Uncrosslinked PnBA Coating Horizontal Force during Scratching: Figure 2 shows typical experimental plots in which the horizontal forces are shown as a function of time. The horizontal force during scratching displays significant fluctuations, especially at low driving speeds and high normal loads, where stick-slip patterns are observed. The horizontal force builds up to a maximum during the stick stage and falls rapidly during the slip stage. Since the sample was driven through an elastic mechanism, the sample and the indenter stuck together until the elastic forces became sufficient to jerk them loose, and then sliding occurred until they stuck together again when the sliding
Scratching of materials and applications
60
velocity decreased to zero. The fluctuations are much reduced at high driving speeds and low normal loads.
50
40 z E v
9
.-o ;~'o
30
~~
Normal
Load
Driving
Speed
= 58.8 m N = 10 lam/sec
o ,~ _-o _. -o o.=7 ~ ~ o ~
o'-o
.~Oo _Y g"
-~
O It_
20
o
.
,
Id
O . _N
/
~
o
/
-1 -1
O
"-1'10
0
Normal
Load
Driving
Speed
100
= 25.6 mN = 20 p m / s e c
200
_~ _~
300
400
500
Time (sec)
Fig. 2. Horizontal force versus time during the scratching of uncrosslinked PnBA coating [ 16].
35
,-,
{
~
,
,
,
,
,
,
r
,
,
,
,
~
,
,
,
'
i
,,
,
,
....
15
....
5
'
_ ................
Driving
o
. . . .
0
i
10
. . . .
i
. . . .
20 Normal
Speed
i , ,
30 Load
--.4 _1
= 20 lam/sec ,
t
40
.
.
.
J .
50
(mN)
Fig. 3. Normal load effect on the horizontal force during the scratching of uncrosslinked PnBA coating (200 l.tm thick) [16].
Figure 3 shows the normal load effect on the average horizontal force (error bar represents the standard deviation) during scratching. It is seen that the horizontal force decreases with decreasing normal load but has a residual value (--- 6 mN) as the normal load approaches zero. Clearly, the attractive force between the coating and the indenter contributes significantly to the scratch force at low normal loads.
Friction, damage and stick-slip in the scratching of polymers
61
To understand the experimental results of Fig. 3, we can simply use the Johnson-KendallRoberts (JKR) theory [4] directly to calculate the contact area due to the normal load and assume that the horizontal force is proportional to the contact area. In fact, by doing just that and approximating the coating as a half space, we found in a previous study [ 17] that the JKR equation fitted Fig. 3 nicely. Indeed, the contact area between the indenter and the coating is finite at zero normal load because of adhesion between the indenter and the coating. This appeared satisfactory except that the interfacial energy found between the indenter and the coating, 3.9 J/m 2, is too high. Therefore, ABAQUS finite element software was used to analyze the problem. To introduce the adhesion energy of the contact area into the numerical analysis with available commercial software, the following methodology was adopted. In a load control situation, let the applied load be P0. Several contact radii ( a ) were chosen, and for each value of a , the depth or displacement (6") and the elastic strain energy in the coating were calculated with ABAQUS. Elastic equilibrium under the rigid indenter was assumed. Then, the total energy (E) of the system for each contact radius was obtained as a sum of the strain energy (Es), the adhesion energy (-rff a 2, where V is the adhesion energy per unit area), and the mechanical potential of the applied force (-P06); that is, E = E , - rcya 2 _ Por. The adhesion energy ( ~ is defined as the energy decrease per unit contact area due to adhesion when compared with the two surfaces before touching: Y~a + Y2o- Y~2, where y~a is the interfacial energy between the indenter and air, Y2ois that between the coating and air, and Y~2 is that between the indenter and the coating. The equilibrium contact radius at a loading P0 was determined by minimization of the total energy with respect to the contact radius a. The program was first checked against the JKR equation for the half space and then applied to a coating. The contact radii at four different normal loads were calculated with the finite element method (FEM) described previously. Young's modulus of 5.5 MPa and Poisson's ratio of 0.499 [ 18] for PnBA coating were used in the analysis. The results are compared with experiments in Fig. 3. Again, we found proportionality between the horizontal force and the contact area for different normal loads, except that the adhesion energy used was 0.6 J/m 2 instead of 3.9 J/m 2, which was used in our previous publication [ 17]. The adhesion energy obtained from the thincoating results with the JKR equation was too large, as pointed out by Li [19]. By knowing the contact area and the horizontal force required for scratching, we calculated the shear stress to be 2.67 MPa. This can be compared with the shear modulus, which is 1.83 MPa. The reason why the shear stress is so much larger than the shear modulus of the coating must be the shear strain to failure is large. The damage caused by the scratching process is extensive, as shown later. In Fig. 3, it is clear that the attractive force between the coating and the indenter contributes to the horizontal force or friction. The ratio of horizontal force to normal load, which is the usual definition of the friction coefficient, rapidly increases when the normal load decreases. At zero applied load, it becomes infinity. In fact, at some negative normal load (tension), the ratio is negative. Therefore, the traditional definition of the friction coefficient is no longer valid here. Instead, the slope of the curve of the horizontal force versus the normal load maybe more meaningful as a measure of the frictional resistance. Even then, the slope may be infinite at a critical negative normal load at which the indenter separates from the coating so that the friction force becomes zero.
Scratching of materials and applications
62 7O L . . . .
6O
!
. . . .
!
(a)
. . . .
~ . . . .
;
. . . .
I
:
t-~
a~ 50
o
I!
O N .~.
"rO
30 !
:
Normal Load = 58.8 mN J
20 0
500
1000
1500
2000
2500
Driving Speed (pm/sec)
7o
..... ib) ........
! ................
................
i ............................
60 E
50
o
u. u
,.
;.
~
........
40 . . . . . . . . . . . . . . . . . ::. . . . . . . . . . . . . . . . . . . . .9 ,,r . . . . . . . . . . . . . . . . . . . . . . . . . . -~;
"I-
20
1
1
l
i
i
~ o . L
'
I
: Normal Load = 58.8 mN
IIIL
1
10
t
,
i
IlJt
,
A
100
I
[
~iZ~l
1000
. . . .
1,,,]
104
Driving Speed (pm/sec)
Fig. 4. Driving speed effect on the horizontal force during the scratching of uncrosslinked PnBA coating [ 16].
Figure 4 (a) represents the horizontal force as a function of the driving speed (average scratching speed). With increasing driving speed, the horizontal force increases but the rate of increase decreases. Figure 4 (b) shows the same data plotted in the logarithmic scales in which a linear correlation is shown. The results suggest that the scratching of PnBA coating is a rate process. The viscoelastic property and/or the kinetics of adhesion and de-adhesion influence the frictional behavior. The frictional energy dissipation reflects not only the strength of interfacial bonds between the indenter and the material but also the viscoelastic loss properties of the material itself. Roberts [20] showed that the viscoelasticity could greatly magnify the work involved to overcome surface adhesive forces. Friction forces generated by polymer surfaces generally do not obey the classical Amonton's and Coulomb's laws [21 ], in which the frictional force is proportional to the normal load and independent of the sliding speed. Because of the viscoelastic nature of polymers, the friction force depends on the sliding speed and temperature. The experiments done by Grosch [22] suggested that two mechanisms contribute to the friction of rubber materials: the adhesion
Friction, damage and stick-slip in the scratching of polymers
63
between two contacting surfaces and the energy loss arising from the deformation of the rubber by surface asperities. Grosch's friction force results can be represented by a single master curve with the Williams-Landel-Ferry (WLF) equation [23]. The master curve was derived from the experimental results by horizontal translation of the curve of the coefficient of friction versus the velocity at a temperature T by an amount log~o a r to fit the reference curve at a temperature To. The horizontal shift log~o a r can be expressed by the WLF equation [23]:
-
logio a r =
8.86(T - TO) 101.5+T-T 0
(1)
where To is a reference temperature equal to Tg + 50 ~ and Tg is the glass-transition temperature. Two peaks on a master curve [22] were distinguishable in certain cases, which were presumed to arise from those two basic mechanisms.
Fig. 5. Scratch morphology of uncrosslinked PnBA coatings. The normal loads were (a) 9.8 mN and (b) 88.2 mN, and the driving speed was 100 I.tm/s for both cases [ 16].
Morphology of the Scratch Tracks: In uncross-linked PnBA coatings scratched by a spherical sapphire indenter, scratch morphology changes significantly with increasing normal load. In Fig. 5 (a), it is shown that materials can be pulled up from the surface by the indenter. This phenomenon was observed only at very small normal loads (<40 mN). For larger normal loads
Scratching of materials and applications
64
(>60 mN), the scratch track displays long and deep pits accompanied by relatively short and shallow gashes, as shown in Fig. 5 (b). The horizontal force decreases during slip when the indenter moves quickly, whereas it increases when the indenter is temporarily blocked in the stick stage by the undamaged part of the coating material. As a result, the center of a scratch track oscillates considerably up and down.
Scratching Behavior of Crosslinked PnBA Coating Horizontal Force during Scratching: Unlike uncross-linked PnBA coating, the surface of crosslinked PnBA coating is not sticky. Figure 6 shows the horizontal force during scratching of crosslinked PnBA coating (two thicknesses 57 and 235 ~m). Here, the horizontal force starts from zero at no normal load in contrast to the uncrosslinked PnBA coating, for which the horizontal force has a residual value (- 6 mN) as the normal load approaches zero. This indicates that crosslinking has dramatically decreased the surface tackiness of PnBA coating; therefore, the interfacial energy between the coating and the indenter is almost zero. For the 235-1am coating, the horizontal force can be fitted by the Sneddon equation [24] for normal loads of less than 100 mN (see Fig. 6) if the horizontal force is assumed to be proportional to the contact area. The constant of proportionality between the horizontal force and the contact area was found to be 10.85 MPa, which is about 2.5 times the shear modulus. The Sneddon equation, which is applicable to larger contact areas than the Hertz equation, shows the relation between the normal load P and the contact radius a
E ((R2+aZ)lnR+a 2aR) R-a
P = 2 ( 1 - v 2)
(2)
or the displacement 6 :
6=aln ~ 2 R-a
(3)
where R is the indenter radius and E and v are Young's modulus and Poisson's ratio, respectively, for the half space. The penetration 5 of the indenter into the coating at a normal load of 100 mN is estimated to be about 61 ~m, which is about 26% of the thickness of the coating. Young's modulus E of 12.7 MPa (determined by nanoindentation measurement) and Poisson's ratio of v = 0.5 were used in the calculation. A finite element analysis [18] was performed for the elastic contact between a rigid spherical indenter (76 ktm in radius) and a soft thin film coated on a rigid substrate. The contact radius and displacement were calculated for two thicknesses (100 and 250 lam) of the thin film. The results of the analysis indicated that the film thickness has little effect on contact radius despite a significant effect on displacement. The discrepancy of the contact radius between the thin film and a half-space medium is less than 5% for a contact radius of up to 0.8R. The change in contact radius with film thickness from 250 to 100 ~tm is less than 2%. Therefore, the Sneddon equation derived for a half space can be used for the 235-~m coating.
Friction, damage and stick-slip in the scratching of polymers
500 I~''''' "9'
'57'1pm ....
T: . . . .
~. ,---,~-'~ ~ . " ~
400
~
65
..... :
300
........
i .........
......
.
:
,(
"
.
.
.
.
.
.
.
.
.
.
.
.
.
d
E 200 ~
100
:,: ,.i--..:.-.-..,'-.: ..........i. .............. , . g "...... " : ;
Dnving Speed = 20 wn/sec 0
100
200
300
400
500
Normal Load (mN)
Fig. 6. Horizontal force during the scratching of crosslinked PnBA coating. The solid line is the curve fitting for the 235-pm-thick coating with the Sneddon equation [ 16].
Mechanisms for Scratch Damages: Because cross-linked PnBA coating behaves as an elastomer, below a critical normal load, which depends on the thickness of the coating, the crosslinked coating recovers elastically after being scratched and leaves no discernible damage. Above the critical load, the coating is damaged and, depending on the coating thickness, shows two distinct damage mechanisms. Eight different thicknesses ranging from 15 to 235 ~tm were examined. For thin coatings (<70 pm), the scratch damage is believed to be a combination of delamination and spallation (called type I damage hereafter). Figure 7 clearly shows that the coming is delaminated under the shear stress ahead of the indenter and removed by spallation. This coming stripping process was explained by Benjamin and Weaver [25] and Weaver [26] as follows: The shearing effect between the coating and the substrate is greatest at the tip of the indentation, and this is where the rupture of adhesive bonds would first occur at the critical load. Because of the horizontal motion of the indenter, the rupture of adhesive bonds between the coating and the substrate would extend, and the sheared material would be pushed away by the indenter. Figure 6 (see the solid circles) shows that the critical load (433 mN) coincides with an abrupt variation in horizontal force. There is an increase in horizontal force when the coating is damaged. Type I scratch damages offer opportunities for the study of adhesion between a coming and its substrate [25, 27-29]. For soft thin coating/soft substrate systems, the indenter sinks much deeper into the surface and a failure mechanism such as buckling due to compressive strain ahead of the indenter may instead occur [30]. In thick coatings (>120 pm), the scratch damage appears as periodic partial tings that are concave with respect to the scratching direction as shown in Fig. 8 (a) (called type II damage hereafter). In contrast to thin coatings (<70 pm) in which the horizontal force increases when the coating is damaged, there is a decrease in horizontal force (see the open squares in Fig. 6).
66
Scratching of materials and applications
Fig. 7. Scratch damage in thin crosslinked PnBA coating (57 ktm thick) at a normal load of 433 mN and a driving speed of 20 ktm/s [ 16].
Fig. 8. Scratch damages in thick crosslinked PnBA coating (235 ~tm thick): (a) periodic partial tings and (b) enlarged view of a ring or a surface crack. The normal load was 102 mN, and the driving speed was 20 btm/s [ 16].
Friction, damage and stick-slip in the scratching of polymers
67
when such damage begins to occur in thick coatings. The drop in horizontal force may be caused by a sudden increase in the local compliance of the sample in front of the indenter when such cracking occurs, which, in turn, results in an abrupt drop of the load cell output. Similar cracking patterns were found in other organic and inorganic materials [31-35]. Scanning electron micrographs reveal that those partial tings are surface cracks [Fig. 8 (b)]. The driving force for this type damage is mainly the horizontal force. The tensile stress produced in the coating behind a moving indenter due to the presence of a friction force is well known [36]. The application of a horizontal force to the normally loaded surface adds a compressive stress in front of the sliding sphere and a tensile stress at the rear. The result of this is to decrease the critical normal load for tensile cracking and produce in the sliding track a series of parallel curved cracks [37]. It appears that the mechanism of the type II damage corresponds to tensile cracking. If so, the primary material property governing the damage is the fracture toughness. The critical normal load increases with the coating thickness regardless of the damage mechanisms, as shown in Fig. 9. Similar thickness dependence of the critical normal load was observed by Gupta and Bhushan [38] in amorphous carbon coatings in which the failure mechanisms were delamination and spallation in thin coatings (_< 30 nm) and tensile cracking in thick coatings (_> 100 nm).
500
,7,-,
i' 1 - , ' ~ .
9 ,
_ 9
'
'
'
'
9
9
9
I
'
9
'
9
OO
. J
;~2oo
o
_.i_____.____.__...______._
......
...........................
.
i ..........i.........o.
100
L...~.-.., . . . . . 0
5o
i .... 100
i ........ 150
Film Thictcness
2o0
250
(.urn)
Fig. 9. Thickness dependence of the critical normal load for the crosslinked PnBA coatings: ( I ) thin coatings and (D) thick coatings. The driving speed was 20 ~rn/s [I 6].
For type I damage in thin coatings, the explanation could be that the stresses around the indenter fall off with the depth [36]; the application of correspondingly greater loads would then be required to produce the critical interfacial shear stress for the coating to be delaminated from the substrate. For thick coatings, the results could be explained by the realization that the horizontal force for tensile fracture in the coating should be proportional to the thickness of the coating. More experiments are needed to understand the transition between the two mechanisms in Fig. 9 as a function of thickness, namely, whether the transition is sharp or more general so as to have a mixed region.
68
Scratching of materials and applications
SCRATCHING BEHAVIOR OF ELASTOMERIC POLY(DIMETHYLSILOXANE) COATINGS Poly(dimethylsiloxane) (PDMS) is widely used in many industrial applications such as coatings, seals, gaskets and adhesives [39]. PDMS, which is a very mobile polymer with a glass-transition temperature (Tg) o f - 1 2 5 ~ maintains its thermal stability and mechanical, chemical, and electric properties between-70 and 250 ~ These same features make PDMS attractive for fundamental studies of scratch deformation. In this section, we present experimental results for friction and damage characteristics during scratch tests of PDMS coatings on stainless steel. The cured PDMS was coated on a 90-~tm-thick substrate of stainless steel shim. The samples were cut in the shape of a rectangle (15x 10 mm2). Scratch tests were performed with a spherical sapphire (76-~tm-radius) indenter unless specified otherwise. The morphology of the scratches was examined with scanning electron microscopy (SEM). Samples for SEM were made by coating the specimen surface with a very thin gold layer (ca. 10 nm thick) by thermal evaporation. For the purpose of detecting the steel substrate exposed because of scratching damage, an original (as-scratched) sample without a gold layer was also examined. The PDMS coating at a comer of the sample was intentionally peeled off for exposure of the substrate. Surprisingly, this sample worked as well and produced quality photos with sharp contrast. To confirm the inherent viscoelastic loss property of the PDMS material, we conducted dynamical mechanical test for bulk cured PDMS with a Rheometric Solid Analyzer (RSAII) in a compression mode. In the dynamical test, oscillating strains were applied to the sample for measurement of the storage modulus (E') and loss modulus (E"), as well as the loss tangent (E"/E'). E' quantified the elastic response of the sample. E"/E' indicated the level of viscous energy loss experienced in the test. The sample (2.4 mm thick) was cut in the shape of a circular plate with a radius of 8.5 mm. The compressive storage modulus and loss tangent were determined simultaneously at frequencies ranging from 0.159 to 15.9 Hz. A static precompression (corresponding to a strain of 0.5%) was applied so that the sample was always in compression. An oscillating strain with amplitude of 0.45% was imposed for implementation of the measurement.
Effect of Normal Load on the Horizontal Force during Scratching Below a critical normal load (see the Scratch Damage and Mechanism section), the PDMS coating recovers after being scratched. Above the critical normal load, the coating is damaged. This section describes the features of horizontal force during scratching in which the PDMS coating recovers and leaves no visible damage behind. Figure 10 shows the effect of the normal load on the average horizontal force (error bar represents the standard deviation) during the scratching of cured PDMS coatings (135 ~tm thick). The horizontal force starts from zero at zero normal load in contrast to uncrosslinked PnBA coatings in which the horizontal force has a residual value (---6 mN) as the normal load approaches zero. Moreover, the friction coefficient decreases with increasing normal load. The results in Fig. 10 can be understood as follows: during scratching, the contact area between the indenter and the coating is determined by the elastic contact, and the frictional force is proportional to the contact area. For a spherical indenter of radius R, the contact area between
69
Friction, damage and stick-slip in the scratching of polymers
the indenter and the material being scratched is given as a function of the normal load P by the Hertz theory [5, 6]. If the frictional force is assumed to be proportional to the contact area, the horizontal force f and friction coefficient/a would depend on P according to the following relations: f = CP ~/3 (4) (5)
l.t = f / P = CP -'/3
where C ItS[RIg] 2/3 ~-'[3(l-v2)R/4E] 2/3, in which S is the interfacial shear strength (force per unit area) between the indenter and the material; v and E are Poisson's ratio and Young's modulus of the material, respectively; and K is defined as shown. =
=
In Fig. 10, the curve fittings with Eqns. (4) and (5) are also shown. The fitting parameter S / K 2/3 found is 31.28 (Pa) 1/3 for Fig. 10 (a) and 31.33 (Pa) 1/3 for Fig. 10 (b). This agreement and the good fittings imply that, within the range of normal loads employed in Fig. 10, the contact between the indenter and the PDMS coating obeys the Hertz theory, and the friction force is proportional to the contact area. The constant of proportionality between the horizontal force and the contact area is 0.78 MPa for both Fig. 10 ((a) and (b)). This proportionality constant can be compared with the shear modulus, which is 0.73 MPa. This comparison implies that the 1 ~ 0.9
~
,
~- i
~- ::.~
9
'--,
, ~
,
,
, ~
70
(a)
o, L t
i ~176 =
....
Q)
"
..-
~ o.4 ~_
;.,.-
-4
....... ~ {
~2o
O.3 ~.." 9
Driving S p e e d = 20 ~ m / s , .
.
q
150
200
0.2
0 0
50
100 Normal Load (mN)
~,.,
,,
~ ....
r ....
~ ....
2.4 ~- ::
i ....
(b)
".., ' '
'._-I 150
~.{125
.~
2
'".
~-
_~
~~
.~-
~ ' . . ..
.
0.8 [ " ' ' ' 0
75
.
.
.
.
.
.
{
25
. . . . . . . . .
50
o
"n
.
~..... ~
Driving S p e e d = 20 lam/s "•
"I" o 4
5 o
~.
.'"
-~ 100
.
;.-~ !
~
. - }~
9
,.~
. " .... ~
L
75
....
-~ 2s ..... i,~
i .... 100
i .... 125
1 0 150
Normal Load (mN)
Fig. 10. Horizontal force and friction coefficient as functions of the normal load during the scratching of cured PDMS coating (135 ~m thick): (a) 76-~tm-radius indenter and (b) 381-I.tmradius indenter. Dotted lines are curve fittings with the Hertz theory [40].
Scratching of materials and applications
70
interfacial fracture strain is about 1.07 if Hooke's law holds for this elastomer. For the same fracture strain in tension, it means an elongation of 2.9 times. For simplicity, we have adopted the Hertz theory in the calculation of the contact area. Since the Hertz theory is for the contact of two spheres or a sphere with a half-space, it is not exactly the situation here. However, according to the finite element analysis [ 18] mentioned in previous section, the coating thickness has little effect on contact radius despite a significant effect on displacement.
Effect of Driving Speed Figure 11 represents the friction coefficient as a function of the driving speed (nominal scratching speed). The friction coefficient falls to small values at low speeds. With increasing driving speed, the friction coefficient increases. The results suggest that the scratching of the PDMS coating is a rate process. The viscoelastic property of the coating influences its frictional behavior.
o.ot 0.8
1E .~_ o
0.7 f
o 00 f =
"= I..1_
0.5
.4
fF i-
Normal Load = 98 mN
Z
0.3 L~_._~
10
1O0
1000
104
Driving Speed (tam/sec)
Fig. 11. Driving speed effect on the friction coefficient during the scratching of cured PDMS coating (135 ~tm thick) [40].
Schallamach [41] suggested that the frictional sliding of rubber is a thermally activated rate process; the main reason for this assumption is the observed temperature dependence of frictional sliding. No detailed molecular model was given, but the general picture was that rubber molecules move due to thermal agitation at the sliding interface. The rate of jumps that the molecules make depends on temperature. In the absence of external stress, their jump direction is completely random. An applied frictional stress directs the jump in the direction of stress. Figure 12 shows that both storage modulus and loss tangent increase with frequency in a dynamic mechanical measurement. We consider that the driving speed dependence of the friction force directly corresponds to the frequency dependence of the material mechanical property [42]. The frequencyfcan be estimated as follows: f = V/2a (6)
71
Friction, damage and stick-slip in the scratching of polymers
1.05
10.07
0 0
1
7,"
o8
v
--1
-o
i
O
tll n
o o
0.95
0.06
0.05 g
o
--t
oo o
oo
O
o o o
~
r
o
0
N o.9
Oo
Q
OOOOO
o~
o
-
0.85
i
,
0.1
i
I
I
l:l[
i.
.
.
.
i,i,i
1
9 ~
0.03
0.02
100
10
Frequency
0.04
(Hz)
Fig. 12. Frequency dependence of the storage modulus and loss tangent of cured PDMS material [40].
where a and V are the contact radius and the moving speed of the indenter, respectively. According to Hertz's theory, a can be expressed as follows:
a=
-
.... 4E
PR
(7)
where P is the applied normal load, R is the indenter radius, v and E are Poisson's ratio and Young's modulus, respectively, of the PDMS material. Therefore, the relationship between f and V can be expressed as follows [42]:
f=
2 t, RP)
(8)
The scratch speeds of 10, 100 and 1000 lam/s in Fig. 11 correspond to 0.03, 0.3 and 3 Hz, respectively. Figure 12 shows that between frequencies of 0.3 and 3 Hz, the loss tangent increases by 41%. This can be compared with Fig. 11, in which between speeds of 100 and 1000 lam/s, the friction coefficient increases by 39%. Therefore, it is likely that the friction, in addition to its proportionality with the contact area, has its origin in the anelasticity of the material.
Scratch D a m a g e and Mechanism
As mentioned earlier, below a critical normal load, the PDMS elastomer coating recovers after being scratched and leaves no visible damage. Above the critical normal load, the coating is damaged by a combination of through-thickness cracking and delamination at the coating/substrate interface.
72
Scratching of materials and applications
Figure 13 (a) shows that at a normal load of 294 mN the coating is cracked and the scratch track thereby formed is not straight. Figure 13 (b) reveals that the coating is delaminated from the substrate under the shear stress ahead of the indenter, removed by spallation, and thrown up at the end of the track. The delamination of the coating is evidenced by the exposed substrate in the SEM view. In Fig. 13 (c), the lower left comer (the region with white contrast) is stainless steel substrate exposed by intentional peeling of the coating, and so the areas with the same white contrast in scratch tracks must represent the substrate.
Fig. 13. Scratch damage in cured PDMS coating (135 ~tm thick): (a) cracking, (b) spallation, and (c) exposed substrate (normal load - 294 mN; driving speed = 100 ~m/s) [40].
When the coating is damaged, there is an increase in the friction coefficient, as shown in Fig. 14. A critical load about 290 mN coincides with an abrupt increase in the horizontal force. Figure 15 shows two examples of the horizontal force versus time for the same driving speed of
Friction, damage and stick-slip in the scratching of polymers
73
100 ~rn/s. For a normal load of 278 mN, which is below the critical load, the film is intact, and the horizontal force varies only slightly (+ 5%). However, for a normal load of 294 mN (above the critical load), the coating is damaged, and the horizontal force displays significant fluctuations (+10%). When cracking and spallation occur, the horizontal force drops because the indenter dives abruptly down and moves quickly along the substrate surface. Then the indenter is temporarily blocked by the undamaged coating material ahead, so the horizontal force increases and builds up to a maximum to cause damage again, and the cycle repeats.
Critical Normal Load for Scratch Damage Obviously, it is important to measure the coating/substrate interfacial strength and to characterize the factors influencing it. The critical normal loads measured in scratch tests have been used in assessing the adhesion in the coating/substrate systems. The critical normal load is defined as the load at which a well-defined failure event occurs. If this failure event represents coating detachment, the critical normal load can be used as a semiquantitative measure of coating/substrate adhesion [25, 27, 28]. The critical load can also be determined from the observation of a sudden change in the slope of the curve of the horizontal force versus the normal load. There are discussions of various factors that can influence the critical normal load of delamination. For example, critical normal loads may be affected by coating thickness (as seen in previous section on crosslinked PnBA coatings) and scratching speed. These are discussed below. 150
120
..~-1
t
4 2
,.D m
60 ..4
i t1"
,,"
Ddving Speed = 100 pnVsec
ii
0
~
1~
1~
2~
2~
300
3~
Normal Load (mN)
Fig. 14. Horizontal force before and alter coating is damaged during the scratching of cured PDMS coating (135 ~tm thick) [40].
The critical normal load increases with the coating thickness, as shown in Fig. 16. It is known that the stress around the indenter falls off quickly with depth [36]. Therefore, for delamination of the coating from the substrate, larger forces are required for thicker coatings to produce the same critical interfacial shear stress as for thinner coatings. However, there is no relation available between the interfacial energy and the critical load.
Scratching of materials and applications
74
Normal Load = 278.3 mN
'~176 f
Normal Load = 294 mN
e.
140
Z
E
.._..
~
N "~_ 0
120
~oo
-i-
f
-i
Driving Speed = 100 !arn/sec
i
6O 0
20
40
60
80
100
Time (sec)
Fig. 15. Horizontal force versus time during the scratching of cured PDMS coating (135 ~m thick; normal load = 278.3 and 294 mN; driving speed = 100 lam/s) [40].
It is also interesting to know how the scratching speed affects the occurrence of scratch damage in PDMS coatings. Figure 17 compares two curves of the horizontal force versus time obtained at the same normal load of 294 mN but at different driving speeds. The PDMS coating was damaged at a scratching speed of 100 lam/s, but it remained intact at a speed of 250 ~trn/s. For the latter speed, a normal load of 319 mN was required to damage the coating. As also shown in Fig. 15, the horizontal force displays much more fluctuation when the coating is damaged. 250
-!,
200 :7
-!,
m
150
0
100
0 ..1
z
.e-
50
Driving Speed = 20 ~jrn/sec
0
50
100
150
Coating Thickness (pm)
Fig. 16. Thickness dependence of the critical normal load for cured PDMS coating [40].
Friction, damage and stick-slip in the scratching of polymers 150
,
,
,
i
,
,
,
w
r
,
,
T
,
,
,
Speed
= 100 larn/sec
Speed
= 250 larn/sec
75
w
,
,
,
1
,
140 ---o--130 z
g ~
120
~
110
.g -r
100
90
Normal Load = 2 9 4 mN
I
80
F 0
20
40
60
80
100
Time (sec)
Fig. 17. Horizontal force versus time curves during the scratching of cured PDMS coating (135 ~m thick; normal load = 294 mN; driving speeds = 100 larn/s and 250 larn/s [40].
Figure 18 shows that the critical normal load increases with the driving speed. Since it is possible that time is needed to nucleate damage, the coatings are more easily damaged at a lower scratching speed than at a higher speed. Steinmann et al. [29] found that the critical load of a chemical-vapor-deposited TiC coating on steel decreased with increasing scratching speed. The positive scratching speed dependence of critical load for PDMS coating may be unique for viscoelastic materials. Ni and Le Faou [35] found that the critical horizontal force for poly(ethylene terephthalate) (PET) film increases with increasing scratching speed; the mechanism of scratch damage is taken as surface shear yielding.
400
z E
........
l
........
l
350
',
' '-
? yTV
t1I o .d
T
Damage
T
-~ 300 E o Z
w It
m
No D a m a g e
t~
..=o _
250
O T
200
t
1
i
i
i
ii1[
i
100
i
i
i
i
|11[
1000
Driving Speed (gm/sec) Fig. 18. Driving speed dependence of the critical normal load for cured PDMS coating (135 pm thick) [40].
Scratching of materials and applications
76
Figure 18 serves as a map, with the line separating regions of damage and no damage. A rule for good coating development is to push the line upward and to the let~.
SLIP PROCESS OF STICK-SLIP MOTION IN THE SCRATCHING OF STYRENEACRYLONITRILE COPOLYMER Styrene-acrylonitrile (SAN) copolymer supplied by Dow Chemical Company was chosen as the testing material because it undergoes only plastic deformation (no cracking) when being scratched, and its plastic properties have been studied [43, 44]. The samples were cut in the shape of a rectangular parallelepiped of 15x10x5 mm 3. The surfaces of the sample were mechanically polished, first with 1 pm and then 0.05 ~tm alumina slurries. A diamond indenter (136 ~ conical) was used in scratch test. The scratches were made at low speeds ( ~ 0 p.m/s), so that heating is negligible. The morphology of the scratches was examined in an atomic force microscope.
General Features of the Slip Process A spring-block model [45] is otten used to represent schematically the sliding bodies where a slider block is pulled with a spring against a counterface at a constant velocity. Such a model can well explain the occurrence of stick-slip phenomenon. During the stick stage of stick-slip motion, the block is fixed and the spring elongation expresses the storage of elastic energy. When the shear stress between the block and the counterface reaches contact strength, a sudden slip occurs. During this slip stage, the spring is shortened, which releases a part of the previously stored elastic energy. l_ x
Vo
f /7"1
k Fig. 19. A simple model of the scratch system [46].
Figure 19 shows a simple spring-block model of the scratch system. The effective spring constant k of the mechanism connecting the step motor (moving at a constant speed V0) to the slider can be obtained from the slope of the horizontal force versus time curve (Fig. 20) during the stick stage. The total mass of the slider and sample is m. L is the normal load applied to the specimen and x is the real scratch distance moved by the indenter. The force balance in the sliding direction is k(Vot - x) - f = rnJi
(9)
77
Friction, damage and stick-slip in the scratching of polymers 600
,
I
I
1
'
I
,
l
I
1
'
I
1
1""
'
/
500 400 Z E
-~ 300 o ~"
200 9
-
100 0
0
200
400
600
800
Time (sec)
Fig. 20. Horizontal force, k(Vot - x), measured by the load cell [46].
During scratching, the horizontal force measured by the load cell can be represented by k(V0t x), where (V0t - x) is the real extension of the spring being stretched, f is the force needed to deform plastically the material in front of the indenter (f is referred to as the scratch force hereafter). As mentioned before, the actual normal load during scratching was measured by a balance on which the slider was positioned. Even though it was a dead load system, there appeared some variation in normal load, about 2%, due to the vertical movement of the indenter. A plot of the horizontal force, k(V0t - x), versus time for a scratching process is shown in Fig. 20, revealing a sawtooth wave form characteristic of stick-slip behavior. It was observed that the scratching motion proceeded by jerks instead of a smooth path, that is, scratching motion is uneven, consisting of stick and slip stages. During the stick stage, there is no relative motion between the indenter and sample. The time of the stick (tens of seconds, depending on the normal load and driving speed) was much longer than that of the slip (< Is, usually is in the order of 1/10 of a second), and the speed of slip was much higher than the driving speed of the step motor. The variation of horizontal force, k(V0t - x), for a single slip stage is represented in Fig. 21, the curve being approximately symmetrical about its point of inflection. In principle, the computer records enough information so that the real scratch distance (x), velocity (V), and scratch force (D at any instant during slip can be found. The static friction is equal to the restoring force of the spring just before slip. The scratch distance, x, calculated from the horizontal force, k(V0t - x), is also exhibited in Fig. 21. To find the velocity and acceleration, one may differentiate the distance-time data once and twice, respectively. If the derivatives are calculated by the numerical process of taking differences, the errors of measurements in the original data are considerably magnified by each differentiation. Especially, the scattering of points of acceleration so calculated is of the same order of magnitude as the acceleration itself. Consequently some method of treating the data had to be used. The actual procedure followed was curve-fitting the original data. The results are affected by an error in the determination of the instant when the slip motion begins, which, unfortunately, is difficult to ascertain.
Scratching of materials and applications
78
Figure 22 (a) represents the scratch velocity versus time during slip, which is obtained by fitting the horizontal force data first and then taking differentiation. During the slip stage, the scratch velocity increases from zero to a maximum and then decreases to zero to enter the stick stage again. It can be seen that the velocity profile seems symmetric and the maximum velocity, Vrmx, attained in the point of inflection can be more than 103 times higher than the driving speed of the motor. If the velocity is plotted against the scratch distance, as shown in Fig. 22 (b), it is clear that for most of the scratch distance, the velocity is within (0.5 - 1)Vm~x, i. e. the slip is all fast except at the beginning and end. By twice differentiating the horizontal force, k(Vot - x), with respect to t, the acceleration and then the scratch force (/) through Eq. (9) can be calculated. Fig. 23 (a) is a plot o f f against time. Several facts are striking. At the beginning of slip, the scratch force drops abruptly from the static to a kinetic value - - in fact, no intermediate value was found by this method. The scratch force does not return to its starting value when the slider comes momentarily to rest at the end of slip, but instead increases somewhat. However, it does not return to the beginning value before the slip stage (at the end of stick stage) either. This is interpreted by the increase of adhesion between the indenter and the material during the stick stage. Iffis plotted against x, as shown in Fig. 23 (b), the scratch force, quite surprisingly, is almost constant for most of the distance in the slip stage. 550
'''
r'''
I''','''
I ' ' ' , ' ' '
500 -~'~+++++++++++ 450
.. ++++++
."
~-.I-F-
z vE
400
~
350
, ' ' '
E~
0o ~
3
9 -t.-_4.. ,,, 9o 99
+4_++
2
,..., 3 3
300 250 99 ~ ~
++++++++4+.
i
200 0
0.02
0.04
0.06 Time
0.08
O. 1
O. 12
O. 14
(sec)
Fig. 21. The time traces of horizontal force, k(Vot-X), and scratch distance (x). Normal load = 882.2 mN and driving speed-- 20 ~trn/s [46].
Friction, damage and stick-slip in the scratching of polymers 50
79
~ ' ' 1 ' ' ' 1 " ' ' t ' ' ' 1 ' ' ' 1 " ' ' ' 1 ' ' ' '
(a )
+++~+44+ H~+++++++
~
40 r CD
+
+
+
-- 30 E E
+
+
+
+
+
+
+
+
~ 20
+
+
._ 0
+
+
>
-
+
+
+
+
10
+ +
O F J , , I , ~ I L , ~ I , , , I L , , I , , , t , , ~
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
Time (sec) 50
40 IT
(b) '-'
+++'
'++.
+
0
i _:
'++++
++
~
0
++++++++++++++++ +t' '++_%
-
+
1
-
"
2 3 Distance (mm)
4
+"
Fig. 22. Scratch velocity during slip: (a) versus time, (b) versus scratch distance. Normal load = 882.2 mN and driving speed = 20 ~trn/s [46]. 550
+ Z
(a)
5OO
E
9 450
/
0
tl
r'-
400
r
o~ 350
_
/
300 0
_++++++++44+~ 0.02
0.04
~~++e~++++++ 1 0.06
0.08
0.1
0.12
0.14
Time (sec)
Fig. 23. Scratch force during slip: (a) versus time, (b) versus distance. Normal load = 882.2 mN and driving speed = 20 ~tm/s [46].
Scratching of materials and applications
80 550 5OO g
E m 450 o It.
" 400 .,..,
I
S!
09 350
300 0
1
2
3
4
Distance (mm)
Fig. 23. (Continue from the previous page).
Fig. 24. An example of the scratch track; scratch direction: up right [46].
Scratch Morphology and Depth Profile An example of a scratch made on SAN is shown in Fig. 24. The scratch track looks like a bamboo stem with periodic joints. It is not smooth and contains deep indents connected by shallower grooves of non-uniform depths. The bamboo-like morphology of the scratch track corresponds to the stick-slip motion. By comparing the morphology of the scratch with the horizontal force curve, it could be deduced that a 'joint' was formed during the stick stage and during slip a 'stem' was made. In Fig. 24, the left arrow points to the stick stage and the right arrow points to the slip stage. Figure 25 is the cross section of the scratch groove showing pileups on the two sides. The areas above the surface appear to be the same as the area below the surface, which means material volume is conserved.
Friction, damage and stick-slip in the scratching of polymers
81
Figure 26 is the center-line profile of the scratch track. It is noticeable that the depth of the stem was not uniform. At the joint, the depth was the deepest because the time of stay of the indenter at one location was much longer than any time during the slip stage so a static indentation was made during the stick stage. As soon as slip commenced, the indenter climbed out from the indent and made a stem. As described above, the scratch velocity during slip can be obtained by fitting the horizontal force data first and then taking differentiation. It is interesting to see what is the correlation between the scratch depth and slip velocity. Because the scratch depth recovered to some extent as the indenter moved away, the depth h can be calculated only from the groove width and the shape of the indenter. Since the in situ groove width during slip was not available, and earlier studies [43, 44] of indentation recovery showed that the depth recovered much faster than the width, the residual groove width was then used in the calculations. The measured (residual) depth is found only about 40% of the calculated one, showing a large extent of indentation recovery. _
_
J
,,,~
.
.
.
.
-
_
,
.
.
.
.
9
qtmt ,
0 ~?~.,~,-~..~
4.---
~--
I
\
Z
, T
'
I
~
. . . . . . .
|
?5
50
!
1 O0
Fig. 25. Cross-section profile of the scratch groove in the middle of the slip stage [46]. UTI
0
8
.~
7 0
! 25
I
I ?S
SO
I ..... 100
pm
Fig. 26. Center-line profile along the scratch track; scratch direction: right [46].
Scratching of materials and applications
82
~ ~ ~ .
yj= 2.167 - 0.6913X R= 0.92075
2.5
2 E v
{:z (1) D
1.5 9
9
9o
t
Calculated
9
Measured
9 9
i
9 9 9 9
9 9 9 9
OoO O 9
Normal Load = 75.95 mN Driving Speed = 10 lam/sec
0.5
0
0.25
0.5
0.75
1
Velocity (mm/sec)
Fig. 27. Scratch depth versus velocity during slip. Normal load = 75.95 mN, driving speed = 10 ~trn/s [46].
Figure 27 is a plot showing the calculated depth (or the actual depth during scratch) gradually decreases with the increase of velocity. It appears that the shallower the groove depth, the faster is the scratch velocity. Let us consider a very small distance Ax along the scratch path, the time period needed for the indenter to go over this distance is Ax/V where V is the velocity of the indenter. Analogous to static indentation, the scratch depth is expected to increase with the time period Ax/V, i. e. to decrease with increasing velocity.
SUMMARY For uncrosslinked PnBA coatings, the horizontal force increases with increasing the applied normal load and begins with a residual value at zero normal load. The relation between the horizontal force and normal load can be understood by finite element computation based on the JKR theory. With increasing driving speed, the horizontal force shows a power relation with speed indicating a rate process. For crosslinked PnBA coating, the horizontal force approaches zero at zero normal load. Below a critical normal load, which depends on the thickness of the coating, the crosslinked coating recovers elastically after being scratched. Above the critical load, the coating is damaged and, depending on the coating thickness, shows two distinct damage mechanisms. During the scratching of PDMS coatings, the contact area is essentially due to elastic deformation by the normal load, and the horizontal force appears proportional to the contact area. Above a critical normal load, which increases with coating thickness, the coating is damaged due to a combination of delamination at the coating/substrate interface and throughthickness cracking. When the coating is damaged, there is an increase in the friction coefficient,
Friction, damage and stick-slip in the scratching of polymers
83
and the horizontal force exhibits large fluctuations. The critical normal load increases with the scratching speed; this implies that time is needed to nucleate damage. The scratch velocity of the indenter in the slip stage is approximately symmetric about its point of inflection that is in the center of the slip distance. The kinetic scratching force remains rather constant for most of the distance traveled in the slip stage. The scratch groove made during slip shows a non-uniform depth, which varies with velocity; the faster the scratch velocity, the shallower is the groove depth.
REFERENCES o
2. 3. 4. o
6. 7. 8. 9. 10. 11.
12. 13. 14. 15.
16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.
Holmberg, K. and Matthews, A. (1994). Coatings Tribology. Elsevier, Amsterdam. Bull, S. J. (1991) Surf. Coat. Technol. 50, 25. Bull, S. J. (1997) Tribol. lnt. 30, 491. Johnson, K. L., Kendall, K. and Roberts, A. D. (1971) Proc. R. Soc. London Ser. A 324, 301. Hertz, H. (1882) J. Reine. Angew. Math. 92, 156. Hertz, H. (1896). Miscellaneous Papers. Macmillan, London. Greenwood, J. A. and Tabor, D. (1958) Proc. Phys. Soc. 71,989. Moore, D. F. (1975). Principles and Applications of Tribology. Pergamon Press, Oxford. Khurana, A. (1988) Phys. Today 41, 17. Gao, C., Kuhlmann-Wilsdorf, D. and Makel, D. D. (1994) Wear 173, 1. (a) Bowden, F. P. and Tabor, D. T. (1986). The Friction and Lubrication of Solids. Clarendon Press, Oxford. (b) Bowden, F. P. and Tabor, D. T. (1982). Friction, An Introduction to Tribology. Krieger Publishing Company, Florida. Gao, C., Kuhlmann-Wilsdorf, D. and Makel, D. D. (1993) Wear 162-164, 1139. Heslot, F., Baumberger, T., Perrin, B., Caroli B., and Caroli, C. (1994) Phys. Rev. E 49, 4973. Li, K., Ni, B. Y. and Li, J. C. M. (1996)J. Mat. Res. 11, 1574. Starmer, P. H. and Wolf, F. R. (1985). In: Encyclopedia of Polymer Science and Engineering, pp. 306, Mark, H. F., Bikales, N. M., Overberger, C. G., Menges, G. and Kroschwitz, J. I. (Eds). Wiley, New York. Zhang, S. L., Tsou A. H. and Li, J. C. M. (2002) J. Polym. Sci. Part B: Polym. Phys. 40, 585. Zhang, S. L., Tsou A. H. and Li, J. C. M. (1998) Mat. Res. Soc. Symp. Proc. 522, 371. Zhang, X. Z. (2001). PhD Thesis, University of Rochester, Rochester, NY. Li, J. C. M. (2001) Mat. Sci. Eng. A 317, 197. Roberts, A. D. (1977) J. Phys. D Appl. Phys, 10, 1801. Moore, D. F. (1972). The Friction and Lubrication of Elastomers. Pergamon, Oxford. Grosch, K. A. (1963) Proc. R. Soc. London Ser. A 274, 21. Williams, M. L., Landel, R. F. and Ferry, J. D. (1955) J. Am. Chem. Soc. 77, 3701. Sneddon, I. N. (1965) Int. J. Eng. Sci. 3, 47. Benjamin, P. and Weaver, C. (1960) Proc. R. Soc. London. Set. A 254, 163. Weaver, C. (1975) J. Vac. Sci. Technol. 12, 18. (a) Burnett, P. J. and Rickerby, D. S. (1988) Thin Solid Films 157, 233. (b) Burnett, P. J. and Rickerby, D. S. (1987) Thin Solid Films 154, 403. Laugier, M. T. (1984) Thin Solid Films 117, 243.
84 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.
Scratching of materials and applications
Steinmann, P. A., Tardy, Y. and Hintermann, H. E. (1987) Thin Solid Films 154, 333. Von Stebut, J., Rezakhanlou, R., Anoun, K., Michel H. and Gantoils, M. (1989) Thin Solid Films 181,555. Hedenqvist, P., Olsson M., Jacobson, S. and Soderberg, S. (1990) Surf. Coat. Technol. 41,31. Bethune, B. (1976)./. Mater. Sci. 11,199. Leu, H. J. and Scattergood, R. O. (1988) J. Mater. Sci. 23, 3006. Wu, T. W. (1991) J. Mater. Res. 6, 407. Ni, B. Y. and Le Faou, A. (1996) J. Mater. Sci. 31, 3955. Hamilton, G. M. and Goodman, L. E. (1966) J. Appl. Mech. 33, 371. Lawn, B. R. (1967) Proc. R. Soc. London Ser. A 299, 307. Gupta, B. K. and Bhushan, B. (1995) Thin Solid Films 270, 39 I. Moore, G. and Kline, D. (1984). Properties and Processing of Polymer for Engineers. Prentice-Hall, Englewood Cliffs, NJ. Zhang, S. L., Tsou A. H. and Li, J. C. M. (2002)J. Polym. Sci. Part B: Polym. Phys. 40, 1530. Schallamach, A. (1953) Proc. Phys. Soc. London Sect. B 66, 386. Kajiyama, T., Tanaka, K. and Takahara, A. (1997) Macromolecules 30, 280. Chang, B. T. A. and Li, J. C. M. (1979) Scripta Metall. 13, 51. Chang, B. T. A. and Li, J. C. M. (1980) J. Mater. Sci. 15, 1364. Rabinowicz, E. (1965). Friction and Wear of Materials. Wiley, New York. Zhang, S. L. and Li, J. C. M. (2003) Mat. Sci. Eng. A 344, 182.
85
CHAPTER 4
NANOSCRATCH AND INTERFACIAL FRICTION OF POLY(AMIDE) FIBRES J. CAYER-BARRIOZ (~), D. MAZUYER t~), A. TONCK (~), Ph. KAPSA (~), and A. CHATEAUMINOIS t2) COLaboratoire de Tribologie et de Dynamique des Syst~mes, UMR 5513, Ecole Centrale de Lyon, 36 avenue Guy de Collongue - 69130 Ecully FRANCE 12;Laboratoire de Physico-Chimie des Polym~res et des Milieux Dispersds, UMR 7615, ESPCI, 10 rue Vauquelin- 75231 Paris FRANCE E-mail : Juliette. [email protected]
ABSTRACT The analysis of the wear resistance of polymeric fibres requires a better understanding of both their abrasive scratch behaviour and their frictional response. These aspects have been investigated at the nanometre scale using the resources of a modified surface force apparatus. In an attempt to simulate abrasive wear losses, nanomachining experiments have been carried out which consist in the repeated scratching of a portion of the fibre surface by the rigid indenter. However, an analysis of the resulting surface topography indicated a significant plastic grooving of the fibre surface with no evidence of wear losses as it was observed at the macroscopic scale. Single pass nanoscratch experiments realised at various sliding speeds also allow discussing the relative contributions of both the material viscoplasticity and the tip/material local interactions on the frictional response. When the sliding speed was incrementally changed during a scratch experiment, it was observed that the associated friction variation was accommodated on a 50 nm distance, independently of the sliding speed.
KEYWORDS Apparent friction, interfacial friction, nanoscratch, nanotribology, poly(amide) fibre
INTRODUCTION Over the past years, the tribological behaviour of polymeric materials, and particularly their frictional response, has drawn a considerable interest. Bowden and Tabor [1] first tried to describe the friction phenomena by taking into account the mechanical contact and adhesion between microscopically rough surfaces. This model, developed for perfectly elastic - plastic materials, has been refined through the years and introduced into statistical physics. Such a statistical model was recently used to describe polymer- polymer friction [2]. However, no effective and comprehensive model for the evolution of the friction force with distance or time has been advanced to date. In order to overcome the difficulties inherent to macroscopic multi asperity contacts, model experiments that attempt to simulate the friction induced by a single
86
Scratching of materials and applications
asperity contact are often considered [3]: for instance, nanoscratching experiments have long been recognized as a potential route to mimic and characterize in a more controlled manner the deformation and frictional response involved in asperity engagements [3] since this method does not require statistical models. The early description of Bowden and Tabor [1 ] assumed that the tangential load necessary to move a grooving tip laterally under a constant normal load is the sum of two separate components, namely an interfacial shear component (often denoted to as the 'adhesive' component of friction) and a bulk ploughing or deformation term. The former is associated with the shearing of the adhesive junctions formed between contacting microasperities within the true contact, while the latter results from mechanical losses induced by the interpenetration of the surface irregularities [4]. Therefore, during a nanoscratch experiment at constant sliding speed, the apparent friction coefficient, ~l.app, c a n be defined as: Fx Fadhesive"t- Fp/oughing = Fadhesive ~ + ~H T A T Papp = ~ = Fz Fz HNAN
(1)
where Fx is the tangential load, Fz is the normal load, Fadh is the 'adhesive' component, Fploughingis the 'ploughing' component, HT and HN are respectively the tangential and normal hardness [5-6], AT and AN are the projected contact areas seen in a direction respectively normal and parallel to that of the relative motion of the indenter [5-6]. The deformation term remains difficult to estimate since most existing models describing the scratch properties of materials do not take into account the viscoelastic and/or viscoplastic effects which can be significant in the case of polymeric materials [ 1, 6-7]. The analysis of this component is critically dependent upon the evaluation of the hardness and also upon the estimate of the size and shape of the contact area. The true contact area between the moving tip and the material is generally considered to be the front half of the part of the tip in contact with the surface but is somewhat difficult to predict since it clearly depends on the material and on the existence of elastic recovery [8]. Calculi presented in the Appendix confirm that the friction component due to ploughing is function of geometric aspects and on the material. Lately, this led to the development of in situ visualization devices [8]. The interfacial shear term results from intricated effects of speed, contact pressure and time taking place near the interface. These effects, strongly correlated to the specific rheological response of the confined interfacial layer of the material, may lead to transient phenomena [913]. The frictional work is assumed to be dissipated in two distinct regions: an interface zone and a subsurface zone. The processes occurring in one are considered not to influence those occurring in the other [ 14-15]. Therefore, the two components are supposed to be non interactive. Recent scratching experiments using the resources of in situ visualization demonstrated, however, that strong interactions between the interface and ploughing components can be involved during the scratching of viscoelastic polymers [8]. Such effects are especially demonstrated through the investigation of the dependence of the shape and size of the contact area on parameters such as sliding speed or temperature. Nevertheless, in situations such as scratching, extensive bulk
Nanoscratch and interfacial friction of poly(amide) fibres
87
ploughing mechanisms are involved and it is extremely difficult to accurately estimate the interfacial friction component from the apparent friction. Numerical models have been developed but they usually assume an Amontons-Coulomb's friction law at the interface [ 16] and stationary sliding conditions. In the context of polymeric materials, two attempts to estimate the interfacial friction component are noteworthy: Bucaille et al[ 17] adapted the Tabor's model whereas Lafaye [18] proposed a method based on a three dimensional flow line description. However, these latter require an accurate knowledge of the elastic recovery of the material. This analysis highlights the strong interactions existing between the two components of the friction, as it is illustrated in literature on bulk polymers. However, no attention has been paid to the frictional behaviour of oriented polymers in spite of the relevance of friction and wear properties in the field of textile applications. The goal of this study is to investigate the deformation modes and the friction processes involved during the scratching of oriented polymeric fibres at the nanoscale. An innovative approach was used, where the study of scratch formation was combined with an analysis of the frictional response of the polymeric fibres. A method associating imaging procedures with nanoscratch experiments was developed on the basis of a modified Surface Force Apparatus [ 19]. This technique was first used to investigate the nanomachining of the fibre surface and to quantify the associated wear at the nanoscale. In a second step, an analysis of the frictional response of the fibre was carried out by means of nanoscratch tests performed at various sliding speeds. This leads to a discussion on the role of the interfacial friction and to a proposal on the interpretation of the contact behaviour in terms of interfacial rheology.
EXPERIMENTAL DETAILS Materials The polymeric fibres investigated in this study were supplied by R.hodia (Saint-Fons, France). Specimens, made of thermoplastic semi-crystalline poly(amide) 6, were elaborated by melt spinning followed by an additional hot drawing step [20] in order to achieve the required draw ratio of 3. This manufacturing process is associated with the development of a microfibrillar structure which can be described using the morphological "swiss-cheese" model proposed by Prevorsek [21-22]. In this model, the fibres are composed of periodic series of crystallites and amorphous domains, called microfibrils, which are embedded in an oriented amorphous matrix, as illustrated in Fig. 1. The molecular weight, M,, of the specimens measured using size exclusion chromatography in dichloromethane is about 19 kg/mol. The fibre round section has a mean diameter of 42 jam. The mechanical properties of the fibres (Young's modulus in the radial and the tangential directions and the hardness) were determined thanks to nano-indentation experiments [23] at 300 K and a zero relative humidity: the reduced Young's modulus in the radial (respectively tangential) direction is about 1.8 x 109 Pa (respectively 2.7 x 109 Pa) while the hardness reaches 108 Pa. The fibre glass transition temperature, estimated by differential scanning calorimetry at 1.6 K/s is around 333 K at a zero relative humidity.
Scratching of materials and applications
88
1-3 nm <__> +
2.5nm / " .
.
.
.
.
.
.
.
.
Microfibril
.
Crystallites
~=enmJq/
2L 3 I!!!!!111~/)11111111111 !!1111111 Extendednon-crystalline molecules
I
~
~
, Disordereddomains
Fig. 1. The Prevorsek's "Swiss-cheese" structural model of polyamide 6 fibres (from ref. [2122]). The fibre axis is vertical. Crystallites are periodically organised to form microfibrils which are embedded in an amorphous oriented matrix. Molecular parameters such as the diameter of the microfibrils, their periodic lengths and the width of the amorphous disordered domains are taken from the literature [22] and SAXS measurements.
Nanoscratching using a Surface Force Apparatus The nanoscratch experiments were carried out using the Ecole Centrale de Lyon Surface Force Apparatus (SFA) which was already described in the literature [19, 23]. The principle of the SFA is shown schematically in Fig. 2. A diamond tip can be moved towards and away from a plane sample holder. The use of the expansion and the vibration of three piezoelectric actuators, controlled by three specifically designed capacitive sensors, allows accurate displacement control along the three axes x, y (parallel to the plane sample holder) and z (normal to the plane sample holder): the sensitivity of the displacements is 10.2 nm in each direction. High resolution and compliant (up to 2 x 10-6 m/N) capacitive sensors equip double cantilever sensors which are supporting the sample holder. The latter allow measuring the quasi-static normal and tangential forces (respectively Fz and Fx) with a resolution up to 10s N. Three closed feedback loops are used to control the high voltage amplifiers associated with the piezoelectric actuators. Two displacement closed feedback loops allow controlling the tangential displacements x and y while the operations in the normal direction z can be carried out either in displacement or normal force control [24]. The scratch experiment can then be made either at constant penetration depth or at constant normal load. Using the z feedback control in the constant force mode, the surface topography can be imaged with the diamond tip before and at2er scratching.
Nanoscratch and interfacialfriction of poly(amide) fibres
89
Fig. 2. Schematic diagram of the Surface Force Apparatus.
Experimental Methodology Nanoscratch experiments were performed at room temperature. A trigonal diamond tip with an angle of 90" between edges was used. The tip defect, determined according to a precise calibration procedure detailed in [25-26], was estimated to be about 16 nm. The studied fibres were maintained along their longitudinal axis on a fiat sample holder by means of a carbon pellet. Prior to the experiments, the tip must be accurately located at the top of the curved fibre surface. In such a situation, the normal displacement axis, z, corresponds to the radial direction of the fibre and the tangential direction, x, is parallel to the fibre longitudinal axis. Two kinds of experiments were performed: (a) first, the elementary wear processes of the fibre were investigated by nanomachining of the surface: an area scratching, consisting in 256 parallel scratches of 1 ~tm long with a spacing of 4 nm between each other was performed; this area scratching procedure was repeated for four times over the same portion of the fibre. The principle of the experiment is illustrated in Fig. 3. The scratching speed was 400 nm/s and the scratching was made in the constant penetration mode with a penetration depth of 100 nm with the tip faced forward [ 19]. The in-situ imaging of the nanomachined surface allows quantifying the wear volume generated by the repeated sliding of the tip. During the nanomachining procedure, both the tangential and the normal forces were continuously recorded. The combined measurements of forces and worn volume allow the calculation of the dissipated energy. (b) single scratch experiments (parallel to the tangential direction, x) of the edged forward tip on the fibre surface at a controlled normal force of 24 laN and at low speed (between 0.7 and 14 nm/s) were carried out in order to study the friction between the
Scratching of materials and applications
90
tip and the polymeric fibre. During the tests, the normal, Fz, and tangential, Fx, forces were recorded, which allowed to calculate the apparent friction coefficient, l.tapp, defined as in Eq. 1.
Fig. 3. Schematic description of the nanomachining procedure used to investigate the nanoscale wear processes of the fibres.
The mechanical behaviour of poly(amide) materials is known to be highly sensitive to water plasticization effects [27-30]. During water diffusion, the plasticization of the poly(amide) amorphous phase induces a significant decrease in elastic and plastic properties, by virtue of a shif~ of the glass transition below room temperature [28]. In order to reduce the effects of variable moisture on poly(amide), the fibres were first dried for 12 h under vacuum at 10 .9 bar and then exposed to nitrogen at 1 bar during the experiments.
RESULTS AND DISCUSSION The first part of this section presents the analysis of nanowear processes as determined from the nanomachining experiments. The results regarding the frictional behaviour of the polymeric fibres and the influence of the scratching velocity are discussed in the following parts.
Nanowear Analysis In order to estimate the microscopic wear rate associated with the repeated sliding of an asperity, four successive nanomachining scans were performed at the surface of the polymeric fibres. Figure 4 presents an image of the resulting fibre surface. A large frontal pile-up can be observed as well as some more limited lateral pile-up. The wear volume, Vwear, defined as the difference between the grooved volume, V_, and the pile-up volume, V+, can be estimated [31] from topographic measurements along the sliding direction, x, as shown Fig. 5, and along the direction y. V+ and V. are defined from a reference plane which corresponds to the initial
Nanoscratch and interfacial friction of poly(amide) fibres
91
surface of the fibre before the machining test. Both front, rear and lateral pile up were taken into account in the calculation of V+. The resulting wear volume, Vwear, atter four successive nanomachining experiments of the surface is close to zero. The polymeric material is clearly plastically deformed and displaced around the tip but is not abraded by the scratching process.
Fig. 4. 3-dimensional topographic image of the fibre surface after four successive nanomachining paths. The sliding direction is along the x direction. Large frontal and minor lateral pile-up are observed.
Fig. 5. Calculation of the wear volume, Vwear, from the knowledge of the surface profile. V+ (respectively V_) represents the volume of material in the frontal pile-up (respectively the grooved volume).
Scratching of materials and applications
92
During a nanomachining experiment, the order of magnitude of the dissipated energy, Ev, and the order of magnitude of the energy used for the plastic deformation, Ep, can be calculated from: Et=FxxLxN=3xl0 Ep=HxV. =3x10 l~
sJ
(2)
(3)
where Fx is the average tangential force (an appropriate order of magnitude is 30/aN), L is the length of the scratch (i.e. 1 /am), N is the number of scratches (i.e. 1024) and H is the hardness of the polymer. Since Eo << ET, it can be deduced that the major part of the energy is probably dissipated in the interfacial friction process. At the macroscopic scale, a wear process is noticed, associated with major material losses [32-33]. Therefore, it may be supposed that at the nanoscale, the dissipated mechanical energy is too low to result in material removal, although all the energy is not only used for the plastic deformation but also in the interfacial friction process. Moreover, even if the trigonal indenter is well representative of the geometry of the abrasive asperities, the strain rates are much lower than those used in the macroscopic wear process (due to the strong difference between the sliding speeds at both scales). Before and after each nanomachining experiment, nano-indentation tests at 10 /aN were performed on the polymer surface at a penetration rate of 1 nrn/s. The nano-indentation procedure is described in detail in [23]. From these tests, it was concluded that the mechanical properties of the polymer surface in terms of modulus and hardness are not significantly modified by the nanomachining process. However, nano-indentation tests reveal a modification of the fibre surface: although no adhesive pull-off force was detected on the original surface, a low adhesive pull-off force (about 0.05 /aN) appears during the unloading stage following the first nanomachining experiment. This indicates that dissipative and/or adhesive properties of the polymeric surfaces are modified due to the sliding of the asperities.
Nanofriction Experiments at Constant Sliding Speed A preliminary nano-indentation step was realised at 30/aN and for a loading speed of 1 nm/s in the controlled displacement mode. After the relaxation of the indentation force, nanoscratch experiments were conducted at 14 nm/s, with the edged forward tip in the tangential direction, x, at a controlled normal force of 24 ~tN. Figure 6 shows the evolution of the tangential force, Fx, and of the tip penetration depth, h, versus the sliding distance normalized with respect to the contact half length, a. By virtue of geometric considerations, the latter is estimated as follows for a trigonal tip: a=0.90xh
(4)
In order to compare the observed behaviour to other typical scratching responses, a schematic evolution of the tangential force and of the tip penetration depth, for a purely plastic and a viscoplastic isotropic materials is also presented in Fig. 6.
Nanoscratch and interfacial friction of poly(amide) fibres
93
Fig. 6. Evolution of both the tangential force, F,, and the penetration depth, h, versus the ratio of the sliding distance to the contact half length during a nanoscratch realised with the edged forward tip and for a normal controlled force Fz of 24 p.N. (A) Fx, (o) h for the viscoelastoplastic PA6 fibres. Theoretical curves of the tangential force and of the penetration depth are also given for perfectly plastic and viscoplastic materials. (grey--) Fx for perfectly plastic material, (black --) Fx for viscoplastic material, (~ penetration depth for perfectly plastic and viscoplastic materials.
The behaviour of the viscoelastoplastic fibres strongly differs from that of plastic and viscoplastic isotropic materials: in the case of the viscoelastoplastic fibres, the tangential force reaches a maximum and then decreases while it should remain constant within the frame of a plastic or viscoplastic behaviour. Moreover, the tip penetration depth also varies with the sliding distance: the tip is pushed out of the surface, which is unexpected since the creep of the contact would rather make the penetration depth increase. Neither the penetration depth nor friction reaches a stationary value within the considered sliding distance. During nanoscratch experiments, the tangential force results from an interface frictional component characterizing the relative sliding of two surfaces without any bulk material deformation and from a ploughing component attributed to the viscoelastic and/or viscoplastic flow of the material around the tip. As a consequence, the variations of the tangential force and of the penetration depth can be due either to the viscoelastoplasticity of the bulk poly(amide) or to the variation of the interfacial frictional response. One can also wonder the effects of the interactions between bulk viscoplasticity and interface friction on the fluctuations of the penetration depth. During the incipient stages of sliding, the polymeric material in the contact zone undergoes contradictory effects:
Scratching of materials and applications
94 -
-
-
an increase in the mean deformation rate which is likely to be associated with enhanced elastoplastic properties. This in turn will induce a decrease in the contact area, and hence, a decrease in the penetration depth. during scratching, polymeric material flow is forced downwards and outwards into the surrounding hinterland which expands. It can be assumed that the energy dissipated to push the material is higher than the one required to climb over the polymeric "wave". This also results in a diminution of the penetration depth. in contrast, the creep of the contact leads to an augmentation of the penetration depth.
These effects are interactive and they result in variations of the penetration depth, as illustrated in Fig. 6.
Identification of the Preponderant Contribution In this paper, an attempt is made to identify the relative contributions of interface shear and bulk ploughing to the friction force by conducting scratch tests at variable sliding speeds. Nanoscratch tests were carried out in a range of sliding speeds varying between 0.7 and 14 nm/s with the edged forward tip. The sliding speed initially equal to 0.7 nm/s was increased to 3.5 nm/s and then to 14 nm/s before decreasing following the opposite procedure. The friction experiments were conducted by keeping the normal force, Fz, constant at 24 ~tN. The variations of the tangential force, Fx, and of the tip penetration depth, h, with time are shown in Fig. 7. Due to the very low tangential compliance of the apparatus and the large displacement resolution [9], the tangential compliance of the material itself and the variations of the tangential force, Fx, are detected. Figure 7 gives rise to three main observations: the tangential force fluctuations are accommodated by penetration depth variations, the tangential force depends on the sliding speed, which is in striking contrast with the Amontons-Coulomb laws of friction. the shape of the transient peaks generated at each speed change differs as a function of the considered sliding distance, in particular below and beyond a sliding distance corresponding to one contact half length (approximately 300 nm). The following paragraphs present in more details these transient peaks and their interpretation as a function of the sliding distance.
Sliding Distance Inferior to One Contact Half Length: During this first stage, the evolutions of the tangential forces as a result of a velocity change are related to the polymer viscoplasticity (Fig. 7): the speed increase is associated with a significant augmentation of the tangential force (which is typical of a viscoplastic effect) and then with a slower variation related to the pile-up formation around the tip. The apparent friction coefficient reaches 0.64 (Fig. 7), which approximately corresponds to the ploughing of the surface (tan 13 = 0.7 where 13 is the attack angle presented by the edged forward indenter). The small difference between theoretical and experimental values can be attributed to a slight rotation of the tip axis relative to the direction of the fibre axis. This clearly demonstrates the preponderance, during this first stage, of the bulk viscoplasticity- a time accommodated process - on the tip/material local interactions.
Nanoscratch and interfacial friction of poly(amide) fibres
95
Fig. 7. Influence of the sliding speed on the tangential force and the penetration depth during a nanoscratch at an imposed normal force Fz = 24 ~tN, with the edged forward tip. The sliding speed varies from 0.7 nrn/s to 14 nrn/s. (A) tangential force Fx, (t~) penetration depth h, (--) sliding speed.
Sliding Distance Greater than One Contact Half Length: Beyond one contact half length, the evolutions of the forces as a result of a change in the sliding speed can no longer be attributed to the bulk viscoplasticity: the associated transient peak in the tangential load exhibits a first direct effect, consisting of a decrease in the tangential force with decreasing speed, followed by a relaxation step to reach a stabilized value (Fig. 7). In an attempt to differentiate between the contributions of bulk ploughing and interface shear to the velocity dependence of the frictional force, two successive scratches, with the edged forward indenter, were realised inside a previously formed scar. This entails a better characterization of the interfacial friction between the tip and the surface since the formation of the viscoplastic groove and the associated pile-up is mainly achieved during the first scratch. At first, the sliding speed was maintained at 14 nm/s. Beyond a sliding distance corresponding to one contact half length, the speed was decreased to 3.5 nm/s, then 0.7 nm/s before being increased following the opposite procedure. Figure 8 shows the evolution of the tangential force and of the penetration depth versus the sliding distance, for a controlled normal force of 24 ~tN during the third scratch.
96
Scratching of materials and applications
Fig. 8. Changes in the tangential force (A) and of the penetration depth (o) as a function of the sliding distance, during the third nanoscratch carried out inside the same groove on the fibre surface (edged forward tip, normal force of 24 laN). The sliding speed ranges from 0.7 nm/s to 14 nm/s. Beyond a sliding distance of one contact half length (approximately 300 nm), the tip/material local frictional interactions are preponderant on bulk viscoplastic ploughing. At each velocity change, an accommodation of the frictional processes is observed over a distance of about 50 nm independently of the sliding speed (insert).
Beyond one contact half length, the shape of the transient peak associated with a speed modification (Fig. 8) remains similar to that observed during the first scratch. Moreover, it is noteworthy that the relationship between the speed and the tangential force is reversible since it is observed during both a sequential speed increase or decrease procedure. Such a nonCoulombic behaviour can thus be interpreted in terms of an interfacial accommodation of the sliding speed [ 10]. An accommodation distance, Do, independent of the sliding speed and close to 50 nm, can be ascribed to this process. Similar accommodation stages were also observed during scratch tests carried out with the faced forward tip. As a consequence, they are independent of the tip attack angle and therefore of the ploughing contribution. According to Baumberger et al [10-12], this accommodation distance, Do, can be attributed to the average memory length necessary to slip to refresh the microcontacts population in a sliding multi-asperity contact. The accommodation time is then attributed to the microcontact ageing. However, similar effects have also been observed for single asperity contacts [9, 13] which indicates that they must be essentially considered as a signature of the interface rheology, independently of considerations associated with surface topography
Nanoscratch and interfacialfriction of poly(amide) fibres
97
Analysis of the Interface Rheology Nanoscratch experiments show that the friction coefficient does not follow the AmontonsCoulomb's laws of friction and depends in particular on the sliding velocity, V. It is known from many studies [10-12, 34] that, in the low sliding velocity regime (typically V < 100 ~trn/s), the steady sliding of a multicontact interface is characterized by a "velocity weakening" behaviour. For most studied materials, one measures, to a good approximation: ~app =~to-13dXl n ( / ~ o)
(5)
where ~to is the friction coefficient corresponding at a sliding velocity Vo, and 13a is the slope. The friction coefficient was measured beyond a sliding distance of one contact half length at two different sliding speeds during the second or third consecutive scratch: ~
[.lapp(V = 14 nm/s) = 0.42, ~app(V = 0.7 nm/s) = 0.5
Therefore, the relation (5) is experimentally fitted for a slope 13d - 0.26 at room temperature, which is in good agreement with the value found in the literature [10]. From this relation, the apparent friction coefficient can be calculated as a function of the sliding velocity (Fig. 9). In this low sliding velocity regime, the relation [.lapp ( V ) with a negative slope induces an amplification of every fluctuation of speed around its stationary value: a deceleration leads to an increase in the friction force, hence to further deceleration.
Fig. 9. Linear evolution of the apparent friction coefficient, ktapp, with the sliding speed in logarithmic scale, typical of a velocity weakening regime. The slope of the linear relationship, 13a, is about 0.026 which is in good agreement with the literature [10].
Despite the monocontact nature of the interface, it may be assumedthat this relation can also be applied to the case investigated in this paper: friction phenomena are found to be accommodated on a distance, Do, independent of the sliding speed and the existence of this accommodation distance is mentioned in the literature for mono or multicontact interface and various materials.
98
Scratching of materials and applications
CONCLUSIONS An investigation of the wear of poly(amide) 6 fibres at the nanometre scale was performed by nanoscratch experiments between a single fibre and a trigonal indenter using a Surface Force Apparatus. The nanomachining of the polymeric surface resulted in the plastic flow of the material and did not imply any material loss at this scale. This difference with the abrasive processes observed at the macroscopic scale was tentatively attributed to the much lower strain rates achieved during nanoscratching. Although the nanomachining process did not induce significant changes in the polymer mechanical properties, it modified the surface dissipative and/or adhesive properties. Energetic calculations showed that a major part of the frictional energy was dissipated by interface frictional process. Nanoscratch experiments clearly demonstrated that the apparent friction coefficient did not follow the Coulomb's laws since they gave strong evidence of a correlation between the indenter penetration, the sliding speed and the tangential force. In spite of complex interactions between bulk viscoplasticity and interfacial friction, it was possible to determine the relative contributions of these processes to the measured friction coefficient, even if the stationary friction regime was not completely achieved. Whereas the viscoplasticity, a time accommodated process, was preponderant during pile-up formation, the interfacial friction became prevalent beyond one contact half length, after formation of the pile-up. During this stage, changes in the sliding velocity were accommodated within a characteristic length of about 50 nm, independently of the sliding speed. This accommodation length was attributed to the rheology of the interface zone. These effects linked to the friction dynamics have been observed for different systems, from dry contacts to lubricated ones, but remain hardly understood. However, despite the prior work to explain the friction phenomenon in terms of the mechanics of relative motion of rough surfaces in contact, only sporadic attention has been paid to the question of friction as a dynamical process evolving in the contact. The need for such an understanding is becoming now more urgent than ever since it is crucial for the knowledge and control of systems that involve rubbing elements, from machines to earthquakes.
REFERENCES
.
3. 4. 5. 6. 7. 8. 9. 10. 11. 12.
Bowden, F.P. and Tabor, D. (1951). Friction and lubrication of solids. Oxford University Press, London. Baumberger, T., Berthoud, P. and Caroli, C. (1999) Phys Rev B 60, 3928. Briscoe, B.J. (1998) Tribology Int. 31,121. Lancaster, J.K. (1973) Plastics and Polymers 12,297. Briscoe, B.J., Delfino, A. and Pelillo, E. (1999) Wear 225-229, 319. Williams, J.A. (1996) Tribology Int. 29, 675. Rabinowicz, E. (1965) Friction and wear of materials. J. Wiley and Sons. Gauthier, C., Lafaye, S. and Schirrer, R. (2001) Tribology Int. 34, 469. Georges, J.M., Tonck, A. and Mazuyer, D. (1994) Wear 175, 59. Bureau, L. (2002) PhD Thesis, Universit6 Paris VII, France. Baumberger, T. (1997) Solid State Communications 102, 175. Bureau, L., Baumberger, T. and Caroli, C. (2002) Eur Phys J E 8, 331.
Nanoscratch and interfacialfriction of poly(amide) fibres 13.
14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.
99
Mazuyer, D., Tonck, A., Bec, S. and Loubet, J.L. (2001) In: Nanoscale surface rheology in tribology. Tribology Series, Tribology Research : From model experiment to industrial problems. 39 pp. 273-282. Elsevier, Oxford. Briscoe, B. (1992) Friction of Organic Polymers, Fundamentals of Friction: Macroscopic and Microscopic Processes. Singer and Pollock Editors. Briscoe, B. (1981) Tribology International 14, 231. Georges, J.M. (2000) Frottement, usure et lubrification. CNRS Editions, Paris. Bucaille, J.L., Felder, E. and Hochstetter, G. (2001) Wear 249, 422. Lafaye, S. (2002) PhD Thesis, Universit6 de Strasbourg, France. Tonck, A., Bec, S., Mazuyer, D., Georges, J.M. and Lubrecht, A.A. (1999). Proc Instn Mech Engrs 213 Part J, 353. Penning, J.P., Van Ruiten, J., Brouwer, R. and GabriElse, W. (2003) Polymer 44, 5869. Bukosek, V. and Prevorsek, D.C. (2000) Int J Polymeric Mater 47, 569. Prevorsek, D.C., Harget, P.J., Sharma, R.K. and Reimschuessel, A.C. (1973) J Macromol Sci Phys BS, 127. Cayer-Barrioz, J., Tonck, A., Mazuyer, D., Kapsa, Ph. and Chateauminois, A. (2005) J. Polym. Sci: Part B: Polym. Phys. 43, 264 Bec, S., Tonck, A., Georges, J.M., Georges, E. and Loubet, J.L. (1996) Philosophical Magazine A 74, 1061. Briscoe, B.J., Sebastian, K.S. and Adams, M.J. (1994) J Phys D: Appl Phys 27, 1156. Odoni, L. (1999) PhD Thesis, Ecole Centrale de Lyon, France. Hernandez, R.J. and Gavara, R. (1994) J Polym Sci: Part B: Polym Phys 32, 2367. Kohan, M.I.. (1995) Nylon plastics handbook. Hanser, Munchen. Stuart, B. and Briscoe, B.J. (1995) Polym Int 38, 95. Valentin, D., Paray, F. and Guetta, B. (1987) J Mater Sci 22, 46. Kato, K. (1997) In: Wear mechanisms: Proceedings of the 1st World Tribology Congress. pp.39-56. Mechanical Engineering Publications. Cayer-Barrioz, J., Mazuyer, D., Kapsa, Ph., Chateauminois, A. and Bouquerel, F. (2003) Wear 255, 751. Cayer-Barrioz, J., Mazuyer, D., Kapsa, Ph., Chateauminois, A. and Robert, G. (2004) Polymer 45, 2729. Bureau, L., Baumberger, T., Caroli, C. and Ronsin, O. (2001) CR Acad Sci 2(IV), 699. Cayer-Barrioz, J. (2003) PhD Thesis, Ecole Centrale de Lyon, France.
APPENDIX: An Estimate of the Ploughing Component of Friction for Various Materials When the adhesive component can be neglected, Eq. (1) becomes: .
app • ~l ploughing
-
HrAr
(6)
H NA N
In the case of plastic materials, for a trigonal indenter, the contact areas are calculated from the tip penetration depth and the attack angle, 13, of the sliding indenter [6]. The ratio of the contact areas is then purely geometric and only depends on the attack angle. As a consequence, for a purely plastic material, the apparent friction coefficient is: HT
/.tapp (plastic material) = ~
tan/3
(7)
1O0
Scratching of materials and applications
In the case of viscoplastic materials, hardness depends on the strain rate, k, which is proportional to the sliding speed. The apparent friction coefficient becomes:
Papp(viscoplatic material) =
HT (~----~) HN (~ ) tan p
(8)
In the case of viscoelastoplastic materials, a contact between the tip and the polymeric surface can occur in the trailing zone of the contact by virtue of the elastic recovery of the material. In such a situation, the ploughing term of the friction force only depends on the attack angle (the asperity slope angle) of the tip and on this partial elastic recovery [35]. Assuming that the height of the elastic recovery is z] and the total penetration depth is z, the modified contact areas can be calculated as illustrated in Fig. 10.
Fig. 10. Evaluation of the projected contact areas for a viscoelastoplastic material a) edge forward, b) face forward. Real contact areas are hachured and the elastic recovery is schematically represented in black.
In the case of edged forward tip,
AN = V~
(Z 2 + Z]2.tan 2 P)
(9)
A T = ,~" (Z 2 / 2 - Z,2.tan 2 ,B)
(lo)
AT / AN _- tan ft.(1 - r 2 ) 1 + r 2. t a n 2 p
(11)
with r = Z~/~ Z While in the case of faced forward tip,
Nanoscratch and interfach~l friction of poly(amide) fibres
AN
=--~/'3 t a Z2 n P +Z" 2
3 .~/(--~.Z2 + 9.Z,2.tan2,6' - 94~ Z.Z,.tan,B)
A T = Z (Z-Z,).~/'3/2 AT I AN
= f(Z, Zl,tan r)
101
(12)
(13) (14)
Therefore for viscoelastoplastic materials at a constant sliding speed, the apparent friction coefficient remains constant with the sliding distance, once the stationary friction is achieved: [lap p (viscoelastoplastic material) = f(Z~ ,Z, fl, HT ,H N )
(15)
These calculi show that the friction component due to ploughing is only dependent on the attack angle and on the elastic recovery of the material.
102
CHAPTER 5
M E T R O L O G Y F O R C H A R A C T E R I Z I N G THE SCRATCH RESISTANCE OF P O L Y M E R I C COATINGS T H R O U G H OPTICAL S C A T T E R I N G
LI-PIIN SUNG l' *, PETER L. DRZAL 2, MARK R. VANLANDINGHAM 3, and AARON M. FORSTER I
1National Institute of Standards and Technology, Building and Fire Research Laboratory, 1O0 Bureau Dr, Mail Stop 8615, Gaithersburg, Maryland 20899-8615 2pPG Industries, Inc., Resin and Coatings Research and Development, 4325 Rosanna Drive, Allison Park, Pennsylvania, 15101 3Multifunctional Materials Branch, U.S. Army Research Laboratory, Aberdeen Proving Ground, MD 21005-5069 *E-mail: lipiin.sung@nist, gov ABSTRACT Scratch and mar resistance is an attribute of great practical importance to polymeric materials. Physical measurements of scratch and mar resistance have been conducted, but the ability to link these resistance to other properties of a polymer and to the customer-perceived appearance of a surface has remained elusive. Thus, optimization of material properties enhancing scratch and mar resistance is not possible. Over the years, a multitude of scratch test devices and protocols has been developed, but the large amount of data generated has made it very difficult to compare and pool data, and ultimately to standardize any test protocol. Also, the relevancy of physical measures of scratch and mar are continually being raised, since a link between physical measurements of scratch and mar and consumer perceived failure caused by scratch and mar has not been established. Establishing a connection between physical and appearance characterizations of scratch and mar are difficult, since appearance measures of scratch and mar are almost always based on qualitative, as opposed to quantitative, assessments of damage. These appearance measurements are made through visual inspection, gloss changes, and changes in gray scale level or lightness. These qualitative assessments are seldom either repeatable or reproducible, and thus a reliable standardized test method for assessing scratch and mar from a consumer's perspective is also not currently available to the polymeric materials community. In this chapter, progress is reported toward the development of a scientifically based standardized test method for quantifying scratch resistance based on optical scattering metrologies. A scratch protocol to impart a standard scratch deformation is described. Optical scattering measurements are conducted to identify the onset of plastic deformation by analyzing specular and off-specular intensities. The angular-resolved optical scattering reflectance profiles at various incident angles are measured using a custom-designed optical scattering instrument at the National Institute of Standards and Technology (NIST). Angular-resolved optical scattering from surfaces having a single or multiple scratches are compared to gloss measurements taken
Metrologyfor characterizing the scratch resistance of polymeric coatings
103
at 20 ~ and 60 ~ with a commercial glossmeter. By analyzing the specular and off-specular scattered intensities, scratch damage can be quantitatively assessed, even when the results from gloss measurements are indistinguishable. Additionally, the correlation between surface roughness and gloss measurements is presented. Finally, there is commentary herein on future research directions that includes implementing metrologies for linking appearance-based scratch resistance measurements to nanomechanical material properties measured through instrumented indentation.
KEYWORDS Appearance, gloss, laser scanning confocal microscopy, mechanical properties, optical scattering, scratch and mar, surface morphology
INTRODUCTION Damage to the appearance of polymeric materials by surface deformation, such as a finger nail scratch or gouge, remains a major challenge for commercial applications of these materials. As a result, considerable scientific and engineering efforts have been expended to assess, improve, and predict the appearance durability or "scratch resistance" of plastic materials. Typically, surface deformation that negatively affects appearance can be categorized as either scratch or mar. Marring usually occurs under less severe conditions compared to scratching, and the depth of mar damage is less than that of a scratch. Scratching, on the other hand, is associated with a much lower density of larger, deeper scratches, sometimes even a single scratch. In both cases, the result is a decrease in the appearance of the polymeric surface, the extent of which qualitatively depends on a customer's perception of what he or she considers a defect. To understand scratch and mar damage, a number of commercial and custom instruments have been developed to impart scratch or mar damage. These instruments can be quite sophisticated, providing instrumented signals such as displacement, scratching force, frictional force, and many others. Coinciding with many of these instruments [1-7] are numerous test methods to quantify or rank the scratch resistance of polymeric materials. The wide variety of scratch methods and instrumentation presents many challenges in the standardization of a scratch protocol, because the same material scratched under different tip geometries, normal force, and velocity will give dramatically different damage morphologies. Equally difficult to standardize is the assessment and measurement of scratch resistance. Scratch resistance is commonly measured by assessing appearance changes brought about by scratch damage. Scratch damage can range from plastic grooving in a ductile material, to cracking and chipping in a brittle material. The severity of the scratch appearance will not only depend on the damage morphology of the scratch region but also on other variables such as surface roughness, color of the plastic part, and index of refraction of the surface. Therefore, a scratch resistance ranking methodology based on a single damage event, like cracking, does not adequately capture the materials resistance to scratch visibility. In the coating industry, specular gloss measurements using commercially available glossmeters remain the mainstream measurement tool for assessing the appearance and durability of coated objects. Commercial glossmeters can be used to measure the change in gloss (decrease in specular reflection) due to scratch or mar damage occurring in laboratory or field testing of a
104
Scratching of materials and applications
material. As will be demonstrated, specular gloss measurements have significant limitations relative to assessing scratch and mar damage. For example, a severe scratch or damage from multiple scratches on a coating surface results in a strong suppression of the specular gloss. A commercial glossmeter is only sensitive to the overall reduction in specular gloss and not the underlying scratch morphology that actually reduced the gloss value. Additional assessment methods such as gloss change, change in gray scale level, or lightness have been created as useful adjuncts to visual inspection to help distinguish between two highly damaged surfaces. However, these techniques often provide only a relative answer, such as whether the surface is scratched or not. For example, a single scratch in a coating or surface produces only a slight change in specular scattering, but it produces a large increase in diffusive scattering intensity (off-specular). Commercial glossmeters, gray scale level, or lightness measurements are not sensitive to these minute decreases in specular gloss nor are they sensitive to the rise in diffusive scattering intensity. Thus, specular techniques cannot be used to quantitatively describe the scratch damage morphology and its resulting scratch visibility for a single scratch surface. More quantitative approaches, such as described in a recent study by Rangarajan et al. [8], have used optical imaging techniques to quantify the optical contrast of a scratch on a glossy polymer surface. The total optical contrast was a function of scratch dimensions and the contrast in both specular and off-specular scattering. Therefore, it provided a more robust measurement of the relationship between damage and visibility. Similarly, a strong correlation between the total optical contrast and scratch visibility was proposed by the industrial appearance perception study [9]. However, this study did not include discussion of the relationships between scratch appearance and the material properties or surface characteristics (e.g., elastic modulus, frictional coefficient, yield stress, and surface roughness). A full angular-resolved optical scattering profile along with characterization of the surface mechanical properties is needed to fully understand the structure-property relationships that govern the scratch durability over time. Currently, relationships between appearance attributes and surface deformation associated with scratching and marring are ambiguous. This lack of connectivity is one of the major barriers to the development and acceptance of standard measurement techniques for determining scratch and mar resistance. In order to successfully implement a scientifically based standardized test method for quantifying scratch resistance, it is vital to understand the relationships between material/mechanical properties, morphology, and appearance (optical properties) of surface and sub-surface deformation. A three-step methodology is proposed to provide the information required to draw conclusions about structure-property relationships that govern scratch resistance. A scratch is first generated in a material using a well-controlled scratch measurement protocol [ 10]. Second, an optical imaging technique is utilized to identify the "onset" of plastic deformation. At this point, the onset may be linked to the mechanical properties. Finally, the scratch damage is characterized with respect to the background signal from the undamaged surface by measuring both the specular and off-specular scattering intensities. The ratio of these two scattering signals includes information on surface roughness, substrate color, incident wavelength, and angle, and is used to evaluate the visibility of the scratch. The combination of absolute physical and optical measurements permits the quantitative evaluation of scratch resistance for each material. Therefore, we may objectively relate performance (durability, appearance) to material/mechanical properties.
Metrology for characterizing the scratch resistance of polymeric coatings
105
This chapter describes an optically-based scratch test methodology developed through a National Institute of Standards and Technology (NIST) partnership with industry (Polymer Interphase Consortium (PIC)) and its application to a series of model polymeric materials. The reader is expected to gain a better understanding of (1) how scratch variables like tip geometry, scratch speed, and method of force application (constant or progressive) influence the onset of scratch visibility, (2) the contribution of surface roughness to scratch visibility, (3) the importance of off-specular scattering and the limitations of specular gloss measurements, and (4) the effect of multiple scratches and a single-scratch on surface appearance.
EXPERIMENTAL SECTION*
Materials Materials used in this study included a set of model amine-cured epoxy (EP, prepared at NIST) coatings, 2 component polyurethane clear coatings (PU, provided by PPG), black pigmented poly(methyl methacrylate) (PMMA, a commercial product), and black pigmented highly crystalline poly(propylene) (PP, provided by Dow Chemical). A brief description of sample preparations along with mechanical characterization is described below.
Influence of Roughness:
A set of epoxy (EP) coatings, each having a different surface roughness, were fabricated to study the link between surface morphology and optical reflectance properties such as gloss [11,12]. The epoxy coatings were cast between a smooth black glass substrate (Schott Glass NG-1, 3 mm thickness) and a mold embossed with a well-characterized surface roughness. Mold fabrication and sample preparation details are reported elsewhere [ 11 ]. The final coatings, designated as EPI5 through EP45, had decreasing RMS roughness values ranging from 800 nm (EP15) to 100 nm (EP45), as estimated from mechanical (stylus) profiling data [13].
Multiple Scratch Visibility: Two different polyurethane clear coat formulations (PU-A and PUB, coating thickness --40 ~m, prepared by PPG) were cast on NG-1 black glass substrates. Multiple scratches were generated on the cured coatings using 5 cycles from an AMTEC Kistler (AK) scratch tester (performed by PPG). Non-scratched polyurethane samples were also prepared for reference.
Single Scratch Visibility: Injection-molded plaques of highly crystalline poly(propylene) (PP) (3 mm thickness) were provided by Dow Chemical. Poly(methyl methacrylate) PMMA samples (3.8 mm thickness) were obtained from a commercial source. An instrumented indentation and scratch testing system (MTS Nanoindenter XP equipped with lateral force measurementinstrument described elsewhere [14]) was used to create and characterize single scratches on these materials. Scratch tests were performed at NIST using 45 ~ semi-apical angle diamond cone indenters with tip radii of 1 lam and 10 ~tm. Scratches were generated by either progressive-force or constant force scratch methods. In a progressive force scratch test, the applied normal force linearly increases over the length of the scratch to a maximum applied force. In a constant-force scratch test, the applied normal force is maintained at a constant value over the length of the scratch. Regardless of the test method, the scratch and residual depths,
Certain instruments or materials are identified in this paper in order to adequatelyspecify experimentaldetails. In no case does it imply endorsement by NIST or imply that it is necessarily the best product for the experimental procedure.
Scratching of materials and applications
106
friction coefficient, and residual roughness are measured during scratch testing. Scratch lengths were either 500 ~tm or 1000 ~tm. For all testing, the samples described above were used as received, with no further annealing or modification. A representative indentation modulus was obtained for each material using an MTS DCM Nanoindenter, and the results are listed in Table 1.
Table 1" Modulus values of samples measured using a MTS DCM Nanoindenter and evaluated at an indentation depth of 1 lam. Each average modulus value is followed by a + symbol and another value representing plus or minus one standard deviation (k = 1), estimated from 10 individual indentations. Specimen Modulus (GPa)
EP-Series 4.16__.0.06
Pu-A 3"88 +0.03
PU-B 4.00 + 0.01
" PMMA "5.11 +0.08
1 PP " ! 1.33 +0'.07
Surface Roughness and Scratch Morphology Characterization A Zeiss model LSM510 reflection laser scanning confocal microscope (LSCM) was employed to characterize surface morphology (topographic profile) and scratch damage. A detailed description of LSCM measurements can be found elsewhere [15, 16]. The laser wavelength used in this study was 543 nm. LSCM images presented in this paper are 2D intensity projections (an image formed by summing the stack of images over the z direction, 512 pixel x 512 pixel), or 3D topographic profiles of the coating surface. The 2D intensity projection images are effectively the sum of all the light scattered by different layers of the coating, as far into the coating (approximately 100 lam in the 5X configuration) as light is able to penetrate. The pixel intensity level represents the total amount of back-scattered light. Darker areas represent regions scattering less light than lighter colored areas. Figure 1 shows an example of (a) epoxy coating surface morphology, and (b) a scratch profile generated by a constant-force scratch test method. The scratch width was defined as the peak-to-peak distance perpendicular to the scratch length, and is indicated in Figure 1b. The Root-Mean-Square (RMS) surface roughness Sq is calculated using a surface tilt correlation across an automatic plane fit of the 3D topographic image. The plane fit is a single polynomial fit to the intensity data across the image. The polynomial fit is subtracted from the image to remove any tilt in the image. Since the subtraction occurs over the entire image, the plane fit does not corrupt the topographic data. The RMS surface roughness is calculated without a numerical filter according to the following formula:
Sq-
.N~'" E " E " Ez(x, , y j ) - , , 7 , ] i-1
s,
(1)
j=l
,.,
Here, Nx, Ny indicate the number of pixels in X- and Y-directions. Estimated uncertainties of roughness measurements were one standard deviation from the mean determined from 5 different locations on each sample
Metrologyfor characterizing the scratch resistance of polymeric coatings
107
Fig. 1. LSCM images of (a) a rough surface- 2D intensity projection (top, left) and 3D topological presentation (bottom, left); (b) a scratch generated by a constant-force scratch method: 2D intensity projection (top, right) and height profile (bottom, fight).
Angular-resolved Optical Scattering Measurements Optical scattering measurements using a newly constructed optical scattering instrument were conducted at various incident angles in the specular, off-specular, and out-of-plane scattering configurations on each polymer sample with and without scratches. The apparatus consists of a laser light source, a five-axis goniometric sample stage, and a two-dimensional (2D) detector mounted in a concentric ring around the sample stage (see Figure 2a). The incident laser wavelength was 633 nm, and the beam was vertically polarized (90 ~ and focused on the sample with a spot size diameter of 1 mm. The sample rotation stage and the detector ring position determine the incident angle of the beam on the sample and the viewing angle of the detector. The bottom illustration of Figure 2a presents the optical geometry, where 0~ and 0s are the incident and scattered angles measured with respect to the sample normal. The quantity, ct (omitted for clarity here), is the out-of-plane scattering angle, which is important for characterizing the scattering profile due to the scratch damage. Specular reflection is measured when Os = 0,-. A detailed description of the instrument will be reported elsewhere [ 17]. Results are presented in terms of the 2D angular distribution of light scattered from a scratched surface at incident angles of 20 ~ 45 ~ and 60 ~ One-dimensional (1D) angular-resolved optical scattering profiles are obtained by circularly integrating the 2D scattering intensity (Figure 2a) as a function of the scattering angle. The scratch scattering signal is compared to the undamaged coating background signal, and the ratio of these two signals is used to evaluate the visibility of the scratch. Figure 2b illustrates the angular distribution of scattered light with respect to gloss (specular angle) for smooth and rough surfaces in surface appearance measurements. The surface roughness contributes more off-specular intensity and reduces the specular intensity. Since commercial glossmeters measure only specular scattering, information related to roughness or scratch morphology that is contained within the off-specular scattering is absent.
108
Scratching of materials and applications
Fig. 2. (a) Photo of the incident laser beam, goniometric sample stage, and 2D detector, and the top view of the layout and optical geometry for the incident, 0~, and scattering, 0s, angles, respectively. A 2D scattering image is also presented here. (b) A diagram illustrates the surface appearance measurements for smooth and rough surfaces. The collection angular-range of scattered intensity in the gloss and haze measurements is also shown.
Gloss Measurements Gloss measurements were made using a hand-held commercial glossmeter (Minolta, MultiGloss model 268). Measurements conform to the ASTM D 523 standard measurement protocol. The reflectance area for 20" and 60 ~ gloss measurements was 9 mm x 9 mm, and 9 mm x 18 mm, respectively. The collection angle is + 0.9" from the specular angle, as illustrated in Figure 2b. All data presented in this chapter are the average of 6 measurements obtained from each sample. The estimated uncertainties in the gloss measurement presented are one standard deviation from the mean value for these 6 measurements.
APPEARANCE-BASED SCRATCH RESISTANCE STUDIES Measurement Protocol for Scratch Test Researchers from NIST and industry have proposed a methodology to quantitatively relate surface deformation (scratch morphology) to appearance attributes and, eventually, to quantitatively evaluate the scratch resistance of polymer coatings and plastics. This work was conducted through a NIST partnership with a number of industrial partners under the auspices of the Polymer Interphase Consortium (PIC) [18], and the research conducted within this consortium has generated a measurement protocol, called the Polymer Interface Consortium Scratch Test Protocol (PICSTP). A brief description of this measurement protocol follows: (1) A series of progressive-force scratch tests are used to impart a number of scratches coveting a range of severities; (2) LSCM (or a high-resolution optical microscopy) is used to characterize the resulting surface deformation and identify the chosen onset of scratch visibility; (3) constant force scratch tests are conducted at force levels both above and below the previously determined onset force (from Step 2) for plastic deformation; (4) LSCM is used to analyze the constant force scratches to precisely identify the force that corresponded to the onset of scratch visibility; (5) important scratch features such as scratch width, yield coefficient of friction, scratch depth,
Metrologyfor characterizing the scratch resistance of polymeric coatings
109
and residual depth at the onset of plastic deformation are identified from LSCM and scratch profilometer data; and (6) this quantitative assessment of scratch damage is then correlated with visual inspection and optical scattering (appearance) measurements.
Scratch Test for the PMMA System Figure 3 is an example of the PICSTP applied to PMMA. Figure 3a corresponds to the scratch profile produced by a progressive-force test using the 1 jam cone. The scratch load ranged from 0 mN to 30 mN over a total scratch length of 500 lam using a scratch velocity of 1 lam/s. Plastic deformation and a concave cracking pattern were observed along the scratch direction. This cracking pattern was typical of materials that have undergone brittle failure [ 19]. The onset of scratch visibility was determined to be very close in proximity to the onset of cracking. This deformation behavior is a consequence of the small diameter tip producing large contact pressures under relatively small forces. Nevertheless, the onset of the readily perceivable cracking pattern in PMMA can be determined both optically and through the instrumented scratch apparatus. The optically determined cracking onset is highlighted in Figure 3a. In Figure 3b, the scratch topography was determined using the 1 ~tm cone as a stylus, and the onset of cracking was defined as the initial increase of the residual roughness level above the undamaged surface roughness. Both methods resulted in similar onset points for the PMMA sample. Constant force tests (Figure 3c) at force values slightly above and slightly below the "critical load" were conducted to identify the onset point more precisely. The critical load was determined to be (3.8 + 0.2) mN.
Fig. 3. (a) LSCM image of a scratch on PMMA produced by a progressive-force test; the onset is defined (arrow) as the point where cracking occurred. (b) The residual roughness level measured from the progressive-force test in (a) along the scratch; the onset is defined by the significant increase in the residual roughness level above the undamaged surface. (c) Two scratches produced by constant force tests below (3 mN) and above (5 mN) the onset load.
110
Scratching of materials and applications
Figure 4 shows the scratch penetration data generated by the instrument during progressiveforce scratch tests on PMMA. Variations in the penetration profiles resulted from "stick-slip" behavior corresponding to the formation of the cracking pattern that had been previously identified through optical inspection. The corresponding residual depth and scratch width that signal the onset of cracking were estimated to be (500 + 20) nm and (6 + 1) l.tm, respectively. At the onset point, the elastic recovery was determined to be 56 %.
0L
Residual
Depth
A
E C ,w t
> L "!
-2000
"', .
(J
. '~
.
~
.
..
,;
"~ "
.
~
~,
.
v
f
'
~, ,
"v ! , H ~
,
~!
',,,
15/Pi
t~O
-3000
,
.
,-
r I-
.~ ,
, "..-.
G) t.-
-4000
Scratch
Depth
.,:;
.
G)
, ,.
G. -5000
' 9
0
,
,
,
*
100
. . . .
|
200
9
,
,
,
I
,
j
,
300
Scratch Distance
,
i
400
. . . .
i
500
(l~m)
Fig. 4. Plot of penetration data recorded by the indentation instrument during the progressiveforce scratch tests described in Figure 3 for PMMA. The lower curve represents the penetration depth during the scratch, whereas the upper curve represents the unrecovered depth (residual depth) remaining just after scratching. The estimated uncertainty is one standard deviation (k= 1) in the data, which is about 3 %.
Scratch velocity is known to be a critical parameter in the creation of scratch surface damage (depth, width, and the onset force). Figure 5a shows the scratch damage obtained with a conical indenter at a constant 4 mN of normal force but at different scratch velocities. The measured scratch width was approximately 30 % wider at 1 larn/s than it was at 100 larn/s. The velocity dependence of scratch deformation demonstrates how the viscoelastic properties of a polymer influence scratch resistance. At the high velocities, the material is stiffer, and, thus, resulting plastic deformation is less. Similar trends were observed in the scratch and residual depth measurements when varying scratch velocities under different constant-force values. Figure 5b and 5c show the semi-log plot of scratch/residual depth and scratch width as a function of scratch velocity, respectively. The residual depth decreased from = 570 nm at 1 ~m/s to = 400 nm at 100 larn/s, and the elastic recovery changed from 56 % at 1 l.trn/s to 63 % at 100 ~m/s, for a constant scratch force of 4 mN. In both plots, a linear relationship (inversely correlated) was observed in the semi-log plot, i.e. depth (or width) oc log (velocity). The complex interplay between polymer viscoelasticity and the velocity dependence of scratch resistance illustrated the challenge in comparing scratch rankings when the scratch velocity is not constant between laboratories. Scratch rankings can be further convoluted when different tip geometries or an onset of deformation (cracking) is used to compare dissimilar materials. These issues are explored on the model PP materials in the next section.
Metrology for characterizing the scratch resistance of polymeric coatings
111
Fig. 5. (a) The scratch profiles of PMMA samples at 4 mN for five different scratch velocities. Semi-log plots of (b) scratch and residual depths and (c) scratch width vs. scratch speed for two different scratch forces. The lines are the linear fit to data in the semi-log plots. The error bars represent one standard deviation (k= 1).
Scratch Test for the PP System The PICSTP methodology applied to the PP samples using the same 1 ~tm radius conical indenter at a velocity of 1 larn/s is summarized in Figure 6. Noticeably, the scratch morphology of the PP system was quite different from the PMMA. In this case, a convex cracking pattern was observed. This scratch pattern is typical for tough materials such as polyolefins. When compared to the PMMA, the initial scratch damage was more severe and occurred at lower normal force values. The cracking onset force obtained from the LSCM image (Figure 6a) and the residual roughness level data (Figure 6b) of a progressive-force scratch test (0 mN to 30 mN) were difficult to determine, and were estimated at 1.2 mN and 1.8 mN, respectively. Subsequent constant-force scratch tests conducted at normal forces below the critical load, shown in Figure 6c, continued to generate both significant plastic deformation and cracking within the PP. Again, the high contact stresses due to the small diameter cone made the identification of the cracking onset force difficult to isolate given the available force resolution. A larger radius cone, 10 p,m radius, was used to better resolve the forces at which plastic deformation occurred and is shown in Figure 7. Although force resolution improved with this tip geometry, the onset of plastic deformation remained difficult to isolate from a progressive force scratch. The onset of cracking was shit~ed to longer scratch lengths with the larger radius tip. Onset values determined from LSCM images and the residual roughness level were restricted by instrumental limitations and are shown in Figure 7a and Figure 7b. Ultimately, a series of constant-force scratches using the 10 ~tm cone, shown in Figure 7c, provided better identification of the cracking onset near 5 mN. The critical force is affected not only by scratch velocity, but tip geometry. Therefore, it is reasonable to have deformation occur at a higher force for a larger radius tip. Additional information describing the elastic recovery of the PP was also measured. The deformation measured during testing with the 10 ~tm radius cone and the residual damage from both progressive and constant force tests are shown in Figure 8. The
112
Scratching of materials and applications
elastic recovery at low scratch loads (less than 3 mN) was almost 100 %. The recovery drops to 86 % at 4 mN and further to 48 % at 30 mN. Note that the progressive and constant force deformation data overlap quite well at the same scratch velocity.
Fig. 6. (a) LSCM image of a scratch (on PP) produced by a progressive-force test using 1 ~m indenter tip. (b) The residual roughness level corresponding to the scratch progressive-force test in (a) as a function of scratch distance. (c) Three scratches produced by constant force tests, below and near the onset load.
A comparison of the critical loads determined with the two different tips demonstrates the advantage of the constant-force scratch test method and the correct tip geometry selection to better identify the elastic-plastic transition. Determination of this transition is a challenge for experimentalists, because the onset of deformation should be unambiguous. In light of better force and displacement identification corresponding to the deformation onset, we have selected the 10 ktm 90 ~ cone as our standard scratch cone. This tip size and geometry affords us the greatest versatility for most polymeric materials evaluated to date. Further, we have designated the onset of scratch visibility, as determined by LSCM, to be our critical scratch criterion. We define this onset as the point at which 500 nm of residual depth exists in the scratch deformation. This has been selected for a number of reasons. Primarily, we are interested in appearance criteria, and 500 nm provides a general limit of what is perceivable to the human eye. Secondarily, we have selected this criterion because it provides an onset point that is independent of material-dependent transitions such as elastic-plastic or ductile-brittle. Severe deformation events such as fracture are highly variable depending on scratch parameters such as tip geometry, scratch velocity, force application, and polymer type. Lastly, the displacement, force, and cone tip geometry corresponding to the onset of scratch visibility presents an opportunity to define a critical strain or critical stress for appearance loss as other researchers have suggested [20].
Metrology for characterizing the scratch resistance of polymeric coatings
113
Fig. 7. (a) LSCM image of a scratch (on PP) produced by a progressive-force test at a velocity of 1 ~m/s using the 10 ~tm cone indenter tip. (b) The residual roughness level corresponding to the scratch progressive-force test in (a) as a function of scratch distance. (c) Three scratches produced by constant force tests: 2 mN, 4 mN, and 5 mN of scratch force. Scale bars represent 10 ~tm. Onset !
|
|
0
~
-2000
~ i u a l D e p t h
~-~r .=
"~~--~#
Scratch Depth -6000 -8000 ,
0
1
5
9
|
10
,
1
i
15
20
,
i
25
,
1
30
Force (mN) Fig. 8. Scratch and residual depths as a function of scratch force for a PP sample and a 10 #m radius conical indenter. The lines are data produced from progressive-force tests, and symbols are data generated by constant-force tests. Scratch velocity was 1 pm/s.
Appearance Measurements The scratch protocol described previously was utilized to identify the general conditions under which scratch deformation is optically perceivable. Optical perception is the metric that coating suppliers and academic researchers have invested considerable effort in quantifying and predicting to improve a material's resistance. Coating researchers wish to answer the questions: "When is a scratch visible?" and "How may mechanical properties be tailored to minimize scratch and mar damage?" Currently, there has been no definitive link established between material properties, scratch geometry, and appearance. Efforts to develop this link are
114
Scratching of materials and applications
complicated because appearance is a visual measurement affected by many physical variables such as illumination (incident angle and wavelength, orientation of perception) scratch geometry, surface roughness, scratch density, gloss, and coating color. Recent efforts at NIST [11, 12, 21] to link surface morphology and subsurface microstructure to optical reflectance properties (appearance) in coated materials have shown promise. This appearance measurement methodology utilizes a ray scattering model and precise, multi-angle illumination of a coating to quantify the optical reflectance (related to gloss values) for a given surface morphology/microstructure. Applying this methodology, we can directly link the scratch morphology to optical reflectance properties, and ultimately to the visibility of a scratch. This section on appearance measurements demonstrates how light scattering techniques, such as these developed at NIST, are objectively able to characterize scratch damage by considering the above variables.
Surface Topography, Optical Scattering, and Gloss Measurements Figure 9 shows the 2D intensity projections of LSCM images taken from an EP coating series. The RMS surface roughness, Sq, calculated from the LSCM generated 3D topographic data are listed in Table 2. The RMS roughness value measured by LSCM was consistent with RMS roughness values determined from interferometric microscopy measurements [ 11 ].
Fig. 9. LSCM 2D intensity projection images of EP-coatings of two different magnifications: 5X (top row, 1842.7 ~tm x 1842.7 ~tm) and 20X (bottom row, 460.7 p.m x 460.7 ~tm).
One-dimensional (1D) angular-resolved optical scattering reflectance data at an incident angle of 20 ~ are presented in Figure 10. The gray shaded region indicates the specular scattering angular range measured by commercial glossmeters. Clearly, the rougher surface (EP 15) has a broader angular scattering profile and a lower specular scattered intensity. The surface roughness has been shown to be a major source of off-specular scattering [11 ]. Similar optical scattering results were obtained at the incident angle of 60~ for these two coatings, and the specular reflectance values were higher at 60 ~ than those values at 20 ~ [ 11 ]. The corresponding 20~ and 60 ~ gloss values of EP series are also listed in Table 2. Here the gloss values were obtained using a commercial handheld glossmeter, which collected only specular reflectance and excluded off-specular components. Therefore, the gloss measurements do not represent the total scattering from a rough surface as shown in Figure 10.
Metrology for characterizing the scratch resistance of polymeric coatings
115
300000
~
"
o
250000
v v v v
200000
V
.#
9
9
EP15
9
EP45
V v
150000 v
v
m 9
100000
v
v 9
50000 (/)
-30
-25 -20 Scattering Angle
-15
-10
(deg)
Fig. 10. Optical reflectance data as a function of scattering angle for incident angles of 20 ~ for two EP-coatings: EP 15 and EP45. The gray shaded area shows the angular range for the gloss measurements.
Table 2. The Root Mean Square (RMS) surface roughness, Sq, measured by LSCM for the EP coating series and their corresponding 20 ~ and 60 ~ gloss values. Error bars represent one standard deviation (k=l), estimated from 5 measurements for surface roughness data and 6 measurements for gloss measurements. Specimen
'q
S (5x), p.m sq (20x), lam
20 ~ gloss 60~
Black glass
EP45
EP35
EP30
EP20
EP15
1.10 0.07 92.0 96.0
1.21 +0.03 0.16 + 0.01 90.2 + 1.0 96.8 + 0.5
1.48 + olb2 0.14 + 0.03 74.0 + 1.1 95.4 + 1.0
2.02 0.43 52.1 87.6
2.66 0.66 29.3 74.4
5.18 +0.28 0.83 + 0.03 15.9 + 0.5 53.3 + 1.3
+0.01 + 0.01 + 0.8 + 0.5
+0.10 + 0.03 + 2.0 + 1.5
+ 0.55 + 0.02 + 1.5 + 1.3
Correlation between Surface Roughness and Gloss Measurements Figure 11 displays the 20 ~ and 60 ~ gloss values as a function of RMS surface roughness for two different measured surface areas: 5X objective (1842.7. ~m x 1842.7. lam) and 20X objective (460.7. ~tm x 460.7 ~tm). Figure 11 clearly demonstrates that specular 20 ~ gloss values correlate linearly to RMS surface roughness at a 20X field of view, see Figure 1 lb. This linear relationship is similar for the larger 5X field of view, although it fails for the roughest surfaces measured (EP15). Note that the larger 5X field of view is much closer to the area measured using a commercial glossmeter. Typically, 20 ~ gloss measurements are applied to high gloss surfaces (low surface roughness), which have a strong specular reflectance. In the steep slope of the curve (the linear part of Figure 1 l a), the differences between glossy samples are clearly measured, while in the fiat part (EP 15, higher surface roughness) the measurement geometry no longer correlates with visual observation. Therefore, the 20 ~ gloss measurement is not sensitive enough to distinguish between semi-gloss surfaces (when the 20 ~ gloss value is less than 20), and a 60 ~ measurement geometry is selected for a semi-gloss surface due to the higher specular reflectance at this angle. This phenomenon reflects a linear correlation between 60 ~ gloss measurements and RMS surface roughness throughout the EP series at the larger field of view (Figure 1 l c). A deviation from linearity for the 60 ~ measurements at the smaller field of view (5X) and smoothest EP surface is observed. This result highlights how the correlation between
Scratching of materials and applications
116
surface roughness and gloss measurements is a function of the size of the area measured and the optical geometry. 100
(a)
\
80
\
6O
\
60
0
0
\
\ o 0 04
40
o
40
20 \
lg
.~
.
,
,
.
,
.
.~\
,o
,
.
100
(c)
,~
,o
0 o
-
.
,
,
(d)
~
0 70
o
60
70
60
50
~ 0
I
' 2
3
4
Sq- LSCM-5X (~m)
5
6
50 0.0
0.2
0.4
0.6
O.8
iO
Sq- LSCM-20X (l~m)
Fig. 11. Comparison between RMS surface roughness values and 20 ~ and 60~ gloss values of EP-coatings for two different surface measurement sizes: 5X objective (1842.7 gm x 1842.7 gm) and 20X objective (460.7 gm x 460.7 gin). The lines represent the best fits for the data: polynomial fit to 20~ gloss values at 5X LSCM magnification and linear fit for the rest of plots.
However, gloss measurements quantify only the specular contribution to the roughness scattering intensity. There is still information about roughness present in the off-specular scattering intensity. The off-specular scattering contribution may be quantified by the ratio of off-specular intensity to specular intensity. For example, if the area under the off-specular scattering intensity (the intensity outside the gray area in Figure 10) is integrated and normalized by the specular scattering intensity (the equivalent 20 ~ gloss value from inside the gray area in Figure 10, a ratio of 69 % for EP45 (smooth) and 37 % for EP 15 (rough) coatings is calculated. Note that for these samples the 20 ~ gloss values decrease more than 30 %. Therefore, gloss values may represent the differences in surface roughness when the offspecular contribution is weak. When considering scratch damage on a surface, the off-specular scattering contribution becomes significant, and should be considered to evaluate surface damage.
Characterization of Damage from Mar or Multiple Scratches A high density of shallow scratches on a coating causes marring. Marring results in an overall reduction of surface gloss and, correspondingly, an increase in haze. Traditionally, mar is only quantified as either a damaged or undamaged surface from gloss measurements. This is an acceptable methodology for two main reasons. First, due to the inherent diffusive scattering from a marred surface, it is difficult to link an optical gloss value to the type and degree of mar damage (RMS roughness). Therefore, it is only possible to rank material performance after gross marring is observed. Second, for a coating exposed to light marring, the only method to visualize the scratches is by evaluating the off-specular reflection, as an observer would do by tilting a sample. Commercial glossmeters are not sensitive to this information, and therefore
Metrology for characterizing the scratch resistance of polymeric coatings
117
cannot discern light marring from an undamaged surface. In this section, the technique for characterizing mar and off-specular scattering is demonstrated to show that off-specular scattering is not required to characterize mar damage.
Fig. 12. LSCM images of PU coatings: (a) non-scratched surface (same for coating PU-A and PU-B), (b) PU-A, and (c) PU-B with multiple scratches created by 5 cycles of the AK scratch test. The bottom row represents the 2D optical scattering profiles (at 0~= 20 ~ 0s = 23~ and the numerical values under the scattering graphs are the 20 ~ gloss values measured using a commercial glossmeter. The boxes indicate the specular reflection region.
For this measurement, two different polyurethane coatings were tested. The modulus for PU-A is slightly lower than PU-B (see Table 1) as measured from instrumented indentation experiments. Single scratch measurements show that 500 nm of residual roughness occurs at 4.5 mN and ~ 6 mN of force using a 10 ~tm conical tip. Also, the friction coefficient of PU-A is higher (0.47) than PU-B (0.22) at 6 mN normal force. Figure 12 shows the surface morphologies (LSCM) and optical scattering profiles for two polyurethane clearcoats with and without multiple scratches. Scratches were generated through 5 cycles of an AMTEC Kistler (AK) scratch tester. LSCM scratch density images and 20~ gloss values confirm that the PU-A coating exhibited less damage than did the PU-B coating, i.e. the scratch resistance of PU-A is higher than PU-B coating. The correlation of surface roughness from multiple scratch damage to the gloss value for the PU coatings was examined. The RMS roughness at 5X of non-scratch PU, scratched PU-A, and scratched PU-B are (0.08 + 0.01) ~tm, (0.08 + 0.01) ~tm, and (0.11 + 0.01) lam, respectively. There was little change in RMS surface roughness and no strong correlation was found between RMS surface roughness, and 20 ~ or 60~ gloss values in the PU scratched coatings. For the PU coatings, the scratches do not affect the RMS surface roughness because they are small and shallow. Gloss loss is more likely due to the optical grating effects, which can shift scattering off-specularly and generate destructive interference and incoherent diffusive scattering at certain viewing angles. Diffusive and destructive intensity is observed in these conditions, as shown in the optical scattering profiles in Figure 12 and Figure 13 (the same offscattering intensity profiles at longer exposure time). It clearly indicates that PU-A has a higher
118
Scratching of materials and applications
scratch resistance ranking than PU-B, and the specular gloss measurements (20 ~ or 60 ~ gloss values for PU-A: 81.3 vs. PU-B: 64.8) are sufficient to properly characterize the scratch damage for a multiple scratch surface. While gloss is useful for mar measurements, a single, deep scratch on a surface provides unique challenges for the traditional gloss measurement.
Fig. 13. Off-specular optical scattering (at 0i = 20 ~ 0~= 24.5 ~ from scratched (a) PU-A and (b) PU-B at higher exposure time. (c) The relative optical geometry of the laser with respect to the scratch profile and orientation.
While gloss measurements were used to detect the specular reflectance changes in the multiple scratch surface (similar to a current field test), when the scratch intensity decreases and/or the scratches are shallow (ie. marring), the scattering intensity is weak in the specular direction, as shown in Figure 12. Thus, the specular gloss measurements are not able to differentiate mar resistance at low scratch densities. In the next section, the extreme case of a light, single scratch, which causes minimal changes in the specular reflection, is evaluated. In order to detect this type of scratch, the observer must tilt the sample to recognize the out-of-plane scattering intensity at a particular viewing angle.
Single Scratch Measurement At low scratch densities or for scratches that minimally scratch damage the surface of a coating, the full scattering profile caused by the damage is evaluated to accurately quantify scratch resistance. In this section, the scratch behavior of PP is discussed. Figure 14a shows the LSCM image of a scratch (only the initial 200 ~m) on PP produced by a progressive-force test using a 10 lam radius conical indenter tip. The scratch damage (visibility) increases with scratch force. As discussed previously, the LSCM scratch profile provides an estimate of the onset of plastic deformation, as shown in the circle of Figure 14a. As discussed earlier, the onset of plastic deformation was difficult to isolate for a progressive scratch test. A series of constant-force scratch tests was required to identify the onset of plasticity at ~. 4 mN at a velocity of 10 lam/s for a 10 ~m radius conical tip. A similar onset value is estimated from the scratch depth and residual depth profiles (Figure 14b). If the shaded area represents 100 % elastic recovery, then the onset of plastic deformation is clearly visible on the plot.
Metrologyfor characterizing the scratch resistance of polymeric coatings
119
Fig. 14. (a) LSCM image of a single scratch (only the first 200 ~m portion) on PP produced by a progressive-force test from 0 mN to 30 mN at a velocity of 10 ~tm/s using 10 ~tm indenter tip. The total scratch length was 500 lam. The visibility of the scratch increases with increasing scratch force, and (b) the scratch and residual depths as a function of scratch force for a progressive scan. The shaded area indicates the 100 % recovery area.
To link the onset of the deformation to the visibility of the scratch, the scattering profiles from two scratches were compared against the original non-scratched surface. Figure 15a shows the scratch morphology measured by LSCM of the unscratched surface (ns) and two 3 mm-long single scratches (sl and s2) in the PP sample. The corresponding optical scattering measurements are also included as Figure 15b. Scratch s l was made with a scratch force of 1 mN less than the onset load, while scratch s2 was made at a scratch force of 2 mN greater than the onset load. The relative optical geometry of the laser with respect to the scratch profile and its orientation are illustrated in Figure 13, for an incident angle of 45 ~. By visual inspection, scratch s l was hardly visible, while scratch s2 was readily visible. The 20 ~ specular gloss measurements of all three surfaces using a commercial glossmeter were indistinguishable, 56.4 + 1.0 for all three surfaces. However, the scattering profiles from the unscratched surface and two scratches are distinguishable (Figure 15b is specular, and Figure 15c is off-specular). Table lists the calculated scattered intensity for specular gloss intensity, and the total intensities from the scattering profiles shown in Figure 15b and Figure 15c. The specular gloss intensity was obtained by integrating the scattered intensity within the angular range of 45~ 0.9 ~. Similar to the results from the commercial glossmeter, there was little difference in specular gloss intensity for the three surfaces. In order to distinguish the difference in visibility of two scratches, the non-specular intensity must be considered. As shown earlier, scratch parameters, such as size, shape, depth of the scratch, pile-up, and roughness of the unscratched surface have a strong impact on the total scattered intensity distribution. The total scattered intensities listed in Table 3 of scratch s2 are greater than the values of ns and sl for both near-specular (0s = 43 ~ ct = 0.5 ~ and off-specular (0s = 40 ~ a = 0.5 ~ configurations. Here, 0~ and ct are the scattering angle (as defined in Figure 2) and the out-of-plane scattering angle, respectively. This preliminary
120
Scratching of materials and applications
result indicates that the onset of a visible scratch may be resolved from optical scattering experiments that account for both specular and off-specular scattering intensity. Current research has been dedicated to replicating these measurements on scratches with different features that affect scratch visibility, such as surface roughness, subsurface microstructure, and color.
Fig. 15. (a) LSCM images of the non-scratch surface (ns) and two single scratches (sl, s2) of a PP sample and their corresponding scattered intensity patterns from (b) at near-specular (0s = 43 ~ a = 0.5 ~ configuration and (c) at off-specular (0s = 40 ~ or= 0.5 ~ configuration. Here 0s and ct are the scattering angle and the out-of-plane scattering angle, respectively.
Table 3: The total scattered intensity from the unscratched surface (ns) and two single scratches (sl, s2) in Figure 15b and Figure 15c. Error bars represent an estimated standard deviation (k=l Gloss intensity* Total intensity from Total intensity from (b) (specular angle __+ (c) (x 106 counts) Location 0.9 ~ (x 106 counts) (x 10 6 counts) 2.20 + 0.06 6.91 __+0.06 0.15 _ 0.04 ns sl 2.16 + 0.06 6.92 + 0.06 0.10 + 0.05 0.38 + 0.05 s2 2.14 + 0.06 8.80 + 0.06 *Gloss intensity was obtained by integrating the scattered light intensity from the scattered profile in Figure 15b within the angular range 45 0__+0.9~ This value is similar to a specular gloss measurement. ..,
,
_
Metrologyfor characterizing the scratch resistance of polymeric coatings
121
SUMMARY AND FUTURE DIRECTIONS The purpose of this chapter was to demonstrate how optical scattering provides the critical information needed to determine the appearance-based scratch resistance of polymeric materials. Characterization of the scratch resistance of polymeric materials included optical scattering experiments that distinguished the severity of the damage while accounting for the impact of surface roughness, gloss, damage type (mar or scratch) over the full 2D angular space. The optical evaluation, including deformation pattern, scratch width/depth, and the transition from elastic to permanent deformation, was determined for PMMA and PP commercial samples using a proposed measurement protocol (PICSTP). The deformation patterns observed in each material were quite different. The critical forces for plastic deformation and cracking were determined to depend on both the scratch velocity and the tip shape. The use of a 10 p.m 90 ~ cone provided the best onset force and displacement resolution for the model materials evaluated. LSCM was used to identify the position along the scratch that yielded a residual depth of 500 nm. The defined onset of scratch visibility provided a perception-based limit that was independent of deformation mode and provided the critical force required to distinguish the severity of the scratch using optical scattering.
Fig. 16. Schematic of the iterative methodology used in the appearance based scratch resistance metrology under development through the NIST/Industry Polymer Interphase Consortium.
The research efforts of the NIST/Industry Polymers Interphase Consortium continue with a second phase and the correlation of appearance-based metrology with mechanical properties and modeling. Figure 16 illustrates a schematic of the iterative design used to determine the material structure-property relationships that govern the appearance-based scratch resistance. Material and mechanical characterization utilizing instrumented indentation have been used to measure material properties such as elastic modulus, hardness or yield stress, visco-elastic properties [22,23], and frictional coefficients over length scales that are relevant to the scratch resistance
Scratching of materials and applications
122
and material heterogeneity. These research results will be reported in the literature at a later date. Future research will expand to include polymeric materials and their appearance-based scratch resistance as a function of environmental aging.
ACKNOWLEDGEMENTS The authors gratefully acknowledge funding support from the NIST-Industry Polymer Interphase Consortium (PIC). PIC Industrial members include: Visteon Corporation, Dow Chemical, PPG industries, MTS system Corporation, Arkema Inc., and Eastman Chemical. The authors give special thanks to PPG industries and Dow Chemical for providing the samples reported here.
REFERENCES
.
3. .
.
6.
10. 11. 12.
13. 14. 15. 16. 17.
VanLandingham, M. R. (2005). In: Proceedings of 3rd International Symposium on Service Life Prediction of Coatings, pp. 305-330, Martin, J.W., Ryntz, R.A., and Dickie, R. A. (eds.), Federation of Societies for Coating Technology, Blue Bell, PA. Betz, P. and Bartelt, A. (1993) Progress in Organic Coatings 22, pp. 27-37. Ryntz, R.A., Abell, B.D., Pollano, G.M., Nguyen, L.H., and Shen, W.C. (2000) J. of Coatings Technology 72 (904), pp. 47-53. Lin, L., Blackman, G.M., and Matheson, R.R. (2001) Materials Science and Engineering A317, pp. 163-170. Ryntz, R.A. and Britz, D. (2002) 3'. of Coatings Technology 74 (925), pp. 77-81. Krupicka, A., Johansson, M., Wanstrand, O., and Hult, A. (2003) Progress in Organic Coating 48, pp. 1-13. Courter, J.L. (1997) J. of Coatings Technology, 69 (866), pp. 57-63. Rangarajan, P., Sinha, M., Watkins,V., Harding, K., and Sparks, J. (2003) Materials Engineering and Science 43 (3), pp. 749-758. Fernholz, K., Sinha, M., Gardner, M., and Watkins, V. (2004). In: Proceeding of 7th International Coatingsfor Plastics Symposium, June 7-9, 2004, Troy, MI. Sung, L., Drzal, P., VanLandingham, M. R., Wu, T.-Y., and Chang, S.-H. (2005) JCT Research 2 (8), pp. 583-589. McKnight, M.E., Marx, E., Nadal, M., Vorburger,T.V., Barnes, P.Y., and Galler, M.A. (2001 ) Applied Optics, 40 (13), pp. 2159-2168. Hunt, F.Y., Marx, E., Meyer, G.W., Vorburger, T.V., Walker, P.A., and Westlund, H.B. (2002). In: Service Life Prediction: Methodology and Metrologies, J. W. Martin and D. R. Bauer (eds.) ACS Symposium Series 805, Oxford University press. ASME B46.1 (1995) Surface Texture, Amer. Soc. Mech. Engrs., NY. VanLandingham, M.R., Sung, L., Chang, N.-K., Wu, T.-Y., Chang, S.-H., and Jardret, V.D. (2004)JCTResearch, 1 (4), pp. 257-266. Corle, T.R. and Kino, G.S. (1996) Confocal Scanning Optical Microscopy and Related Imaging Systems, Academic Press. Sung, L., Jasmin, J., Gu, X., Nguyen, T., and Martin, J.W. (2004) JCT Research 1 (4), pp. 267-276. Sung, L., Garver, J., Embree, E., Dickens, B., and Martin, J.W. (2005) NIST internal report.
Metrology for characterizing the scratch resistance of polymeric coatings 18. 19. 20. 21. 22. 23.
123
NIST-Industry Polymer Interphase Consortium (PIC) - more details: visit consortium web site slp.nist.gov. Briscoe, B. J., Pelillo, E., and Sinha, S.K. (1996) Polymer Engineering and Science 36 (24), pp. 2996-3005. Jardret, V. and Morel, P. (2003) Progress in Organic Coatings 48, pp. 322-331. Sung, L., Nadal, M.E., McKnight, M.E., Marx, E., and Laurenti, B. (2002) J. of Coatings Technology, 74 (932), pp. 55-63. VanLandingham, M.R., Chang, N.K., Drzal, P.L., White, C.C., and Chang, S.-H. (2005) ,I. Poly. Sci. B - Poly. Phys., 43 (14), pp. 1794-1811. White, C.C., VanLandingham, M.R., Drzal, P.L., Chang, N.K., and Chang, S.-H. (2005) J. Poly. Sci. B - Poly. Phys., 43 (14), pp. 1812-1824.
124
CHAPTER 6
SCRATCHING OF POLYMERS: DEFORMATION MAPPING AND WEAR MODELING
SUJEET K. SINHA
Department of Mechanical Engineering National University of Singapore 9 Engineering Drive 1, 117576, SINGAPORE E-mail: [email protected] ABSTRACT Scratching as a material testing tool has been around for well over 180 years [ 1] but despite many advantages, this technique has only recently found general acceptance by the engineers. Often the problem lies not in the testing method but in the deconvolution of the test results due to a lack of proper understanding of the stress field around a scratching tip and the boundary condition between the tip surface and the material. In this chapter, we will present some of the successes in the application of scratching for bulk polymer where this technique has provided insight into the materials surface damage characteristics. Examples are drawn from this author and coworkers' current and past works on the scratching of mainly bulk solid polymers. The first application example is taken from the work on scratching maps for polymers. Scratching technique has shown that often materials' deformation behavior, when faced with a sharp tip, is a strong function of the tip attack angle, normal load, scratching velocity and temperature. The second part of this chapter will focus on more recent results on the use of nano-scratching to understand the wear debris generation process for polymers. It is shown that cyclic intersecting scratching provides a hitherto unknown clue as to why some polymers, despite their low tensile strength, are more wear resistant than other stronger polymers. KEYWORDS: Scratching maps, intersecting scratching, polymers, abrasive wear modeling
INTRODUCTION Scratching technique for (bulk) polymers has several advantages. For example, materials deformation characteristics for a range of imposed conditions (load, speed, temperature etc.) can be obtained by this simple test [2,3]. Scratching technique can also be used to understand the friction models such as plowing and sliding contributions to friction [4]. Recently, this technique has also been used to model abrasive wear of metals and polymers [5,6]. Although the stress field around a moving tip in a scratching event is very complex and requires the computational technique to understand and estimate stress [7], much of the materials specific data, for practical applications, can be obtained quite conveniently in a scratch test by simple measurements of the scratch force and post-scratch geometry of the groove. Historically,
Scratching of polymers: Deformation mapping and wear modeling
125
the scratch hardness value has been obtained by a simple ratio of the load divided by the projected contact area [2]. This way of calculating scratch hardness is quite similar to the calculation of indentation hardness, however, difficulty in scratching is that the calculation of the projected contact area is, at best, only an estimate. The scratching event is marred by elastic and plastic deformations and recoveries coupled with cracking, fracture and chip formation. These deformation and fracture events are directly related to parameters such as the material's mechanical properties, the indenter attack angle, applied load, scratching speed and the prevailing temperature. Thus, the actual projected contact area is difficult to estimate for scratching. Scratch hardness is useful as a surface property data when ductile metals or ductile polymers are involved, and also, the general material surface deformation behavior during scratching can be studied by this test. In this chapter we will present two examples of the materials deformation and failure studies conducted by scratching. The examples are the scratch deformation mapping and the abrasive wear modeling. The studies mentioned here are related only to polymers, as metals and ceramics may present different set of trends.
SCRATCH DEFORMATION MAPPING FOR POLYMERS The scratch deformation mapping study has been conducted to understand the condition under which a scratch would be damaging to the surface of a polymer. It is more related to the visual perception of a polymer surface than a real mechanical damage to its strength. Often many polymers are used for the appearance (such as transparency) or optical aesthetics of an engineering component and for such applications the quality of the surface is judged by visual appearance. The visual appearance can be quantified by measuring optical reflectivity or optical scatter of a surface [8,9]. Thus, when a material deforms on its surface by only plastic deformation or by only brittle fracture, the optical reflectivity or scatter values of the deformed surface for these two extreme cases will be different. The other purpose of knowing the distinction between different deformation mechanisms is applicable in predicting failure as a result of surface damage. Whatever the purpose and intended final application, the main objective of this type of work is to understand the surface deformation characteristics, as perceived by some optical means, under a range of operational parameters such as the imposed strain, normal load, scratching speed and the ambient temperature. A change of any of these operational parameters, as will be seen later, brings a drastic change in the deformation characteristics of the polymers. It may be noted that the surface deformation characteristics of a material decide much of the other measured performances of the materials such as the coefficient of friction, wear rate and surface fatigue failure. Polymers are known to demonstrate a wide range of deformation behaviors for a slight change in the operational conditions. The variety in the materials responses has been shown schematically for a change in one factor, strain, characterized by the included angle of a conical tip; Fig. 1. This figure, though only a qualitative guideline, also provides an excellent definition of the nature of deformation encountered during scratching of polymers. Elastic deformation would be said when the scratching process does not leave behind an optically visible scratch mark. This is akin to pure elastic sliding often encountered when the imposed strain is extremely low coupled with low loading or stress condition. Ironing is the term used when we observe a type of mark on the surface of the polymer that is indicative of smoothening of the tall asperities on the surface but no observable plastic deformation on the real surface of the polymer. This type of deformation may be defined as a stage between the elastic deformation and the ductile plowing. Thus,
Scratching of materials and applications
126
ductile plowing is defined as plastic deformation of the polymer surface by the scratching tip. The scratch groove does not show any presence of micro-cracks inside or along the scratch mark. Often a ductile polymer would show pile-ups or plastically deformed bulges formed on the sides of the scratch when the plowing is deep. Cracking would be encountered in and around the scratch if the material behaves in brittle manner. Lastly, brittle machining is recognized when there is the production of chips and loose debris as a result of crack initiation and propagation. Brittle machining is one important mechanism by which wear of the polymer material by hard asperities takes place. Often there is a combination of the above-mentioned deformation types which are recorded while constructing a scratch deformation map. Response (Pictorial)
Generic
or. 180"
150"
_
.
.
.
_
.
.
120"
Ploughing 90"
Ductile
Machining
+ Cracking 60"
30" MIchlnll~ O"
Fig. 1. A qualitative representation of the influence of cone angle upon the type of damage produced [4]. The figure only shows the effects and is not drawn to any scale. For clarity, the pictorial view does not show the change in the cone angle.
Figures 2 (a) and (b) show scratching maps for PMMA for different strains imposed at varying normal loads (Fig. 2 (a)) and temperatures (Fig. 2 (b)). The nominal strain was varied by changing the included angle of the conical indenter from 30 ~ to 170 ~ The strain in this case is calculated by the parameter 0.2 tan 0; where 0 is the attack angle of the cone (0 = 90 ~ - 0.5 x (included cone angle)). The map was constructed by observing the deformation patterns after scratching under a scanning electron microscope. It is seen that at normal temperature, PMMA starts to show brittle behavior at a strain of around 0.12 depending upon the normal load. Though not a very marked effect, high load has tendency to change the deformation mechanism to brittle machining and cracking. It is probable that both ductile deformation and brittle fracture can get pronounced at higher load just because deep penetration of the surface would create much higher stress field (both tensile and compressive) around the indenter than would be possible at lower load. Such change in the mechanism from ductile to brittle has been
Scratching of polymers: Deformation mapping and wear modeling
127
observed for poly(carbonate) as well, though only for lower cone angles or high strains [ 10]. It is interesting to note that both PMMA and PC would be expected to fail in brittle manner under normal mechanical testing (such as in tensile testing) as both are amorphous polymers with their glass transition temperatures approximately 100 and 149 ~ respectively. A very ductile polymer may show more of deep grooving and plastic deformation as the load is increased (see Fig. 3 for the case of UHMWPE).
Fig. 2. Scratch deformation maps for PMMA. (a) The diagram shows the dependence of the observed deformation modes upon nominal contact strain and the applied load. The scratches were produced at a constant scratching velocity of 0.004 mm s ~, at ambient condition in dry contact condition. (b) The dependence of the observed deformation mode upon nominal contact strain and the bulk temperature of the polymer. The scratches were produced at a constant scratch velocity of 0.004 mm s-~, under a constant normal load of 1 N in dry contact condition. (Ref. [2])
128
Scratching of materials and applications
Fig 3. Scratching mode map for UHMWPE. The diagram shows the dependence of the observed scratch deformation mode upon the nominal contact strain and the applied normal load. The scratches were produced at a constant scratching velocity of 0.004 mm sl, at ambient temperature (20 ~ and under dry contact condition (Ref. [2]).
At lower strains, all polymers tested showed only ductile plowing effects as the load was increased. In general, lower strain has milder effect on the deformation which changes from elastic for very low strain to ironing and to ductile plowing as the strain is increased progressively. Among many other parametric effects, we have also constructed maps for the effects of temperature and scratching speed. Effect of temperature on all polymers would be to change the deformation to ductile and visco-plastic plowing. Figure 2(b) shows a scratch map for PMMA when the bulk temperature and the strain were varied. For all strains, the deformation changes to ductile visco-plastic as the temperature is increased. Ductile viscoplastic behavior is characterized as the mode where there is delayed or time-dependent recovery after plastic deformation. Although the temperature tested was still below the glass transition temperature of PMMA, there is a marked change in the way materials behave when the temperature is Close to the Tg. The change in scratching speed did not produce any marked change in the mode of deformation during scratching. There is very slight indication that the polymer may behave in more ductile manner suppressing brittle failure when the scratching speed is increased. This is despite the fact that polymers when deformed at a high strain rate in a tensile test show brittle type of behavior and hence, by analogy, ought to show brittle nature when scratched at high scratching speed. It has been found that during scratching, a large amount of frictional and plastic works, which are very localized around the tip, can actually transform into heat with high temperature rise locally. Temperature up to 100 ~ has been recorded for PMMA scratched at a speed of 1.5 m s~ or more by 90 ~ and 120~ conical indenters [11 ]. The frictional heating effect can in fact suppress brittle deformation mode and promote ductile and visco-plastic deformation mode due to thermal softening of the polymer. This effect
Scratching of polymers: Deformation mapping and wear modeling
129
is more pronounced in the case of polymers as they are thermal insulators and their melting/sottening temperatures are much lower in comparison to those for metals and ceramics. Summarizing, there is a clear indication that scratching characteristic of polymers is a very complex phenomenon and the concept of scratching map can be used to present such data which are otherwise difficult to comprehend. Surface deformation characteristics can be an initiator of a whole range of tribological phenomena such as friction, wear and surface fatigue in addition to the visual aesthetic aspect (gloss) of a polymer surface.
SCRATCHING AS A MEANS TO UNDERSTAND ABRASIVE WEAR A number of attempts have been made in the past, both for metals and polymers, to devise means of predicting wear performance (mainly abrasive). Scratching basically models one hard asperity of a surface if the shape and size of the scratching tip is well defined and they are similar to those of the asperities on the surface [12,13].
l::
0.25-
!
r l PS
0.2-
<
l::: 1::: "~ 0.1519 !.. !.._
0.1PPS+PT]~
ABS u tj t9
D
0.05-
POIVI+Fq'FE
[]
[2I
BR
o,,
u)
"
0 0
r O UHMW.PE
100
200
PE]~K
D ~ !
l
300
400
500
Scratch hardness, MPa Fig. 4. Specific wear rate PMMA is from ref. [15] softest polymer with very wear rate (Reprinted from
in abrasive wear tests as a function of the scratch hardness (Data for whereas all other data are from ref. [6]). UHMWPE is among the low wear rate while PS is among the harder polymers with very high ref [ 16] with publisher's permission).
The work on modeling wear by scratching for polymers have been prompted by several evidences that the indentation or normal hardness does not correlate with the wear performance for polymers. One example may be cited here. PMMA is roughly 5-6 times harder in normal indentation than UHMWPE but the wear rate of PMMA is approximately 85 times greater when compared to that of UHMWPE [14]. Similarly, Budinski's [6] work has also shown that the wear rate of polymers furnishes no correlation with the scratch hardness as seen in Fig. 4. The general conclusion is that a harder polymer is not necessarily more wear resistant than a softer polymer when they are slid against the same metallic surface and under the same experimental
130
Scratching of materials and applications
conditions. This is counter to the well-accepted Archard's model where wear volume is presented as inversely proportional to the hardness [ 17]. Archard's model is well applicable to metals. Thus, it seems there is a gap in the understanding of wear modeling for polymer. There is another fact about the hardness values (scratch and normal) for metals and polymer that for metals there is a reasonable linear correlation between the two parameters whereas they do not show any relation for polymers (Fig. 5). Thus, abrasive wear of polymers is neither related to normal hardness nor to scratch hardness in any reasonable manner.
Fig. 5. (a) A comparison of the scratch hardness and normal hardness values for metals. The data are taken from ref [5]. (b) Scratch hardness as a function of Shore Hardness D for engineering polymers. Data are taken from ref [6].
Scratching of polymers: Deformation mapping and wear modeling
131
The classical work on the modeling of polymer wear have looked at the macroscopic wear tests and then plotted wear rate as a function of some combination of material properties. The most useful work in this area has been carried out by Ratner and coworkers [18] and then by Lancaster [19]. Ratner and coworkers, in their classic work of sliding polymers against metal gauges, looked at the wear of polymers as a surface fatigue phenomenon and related specific wear rate to the mechanical properties by a relation [ 18], Specific wear rate = const. [(la/H of e)]
(1)
where kt is the coefficient of friction, H is the indentation hardness, of is the breaking stress and e is the % elongation to break. Equation (1) includes indentation hardness in the form of inverse proportionality. Interestingly, many later workers have not included indentation hardness as one variable in their wear models [16]. One other important parameter, which is the roughness of the counterface, is not included in equation (1). The roughness of the counterface has been found to affect the wear behavior of polymers whether it is adhesive or abrasive wear [20]. The actual effects of asperities, their shapes, sizes and of other characteristics are very complex and the reader is referred to ref. [21 ] for further explanation on these effects. Recognizing the importance of tensile strength, the counterface roughness parameter and fatigue property of polymers, Hollander and Lancaster gave a model for wear as [21 ], W
O~ O f n r -2n/3
(2)
where Of is the failure stress of the polymer, r is the mean tip radius of the asperities on the abrasive surface and n is the fatigue exponent, a material property that can be derived from fatigue test on bulk material. One important finding in the work by Hollander and Lancaster is the realization that it is important to consider a suitable roughness parameter such as, r, which takes into account the actual tip shape (especially the attack angle), rather than a statistical mean value of roughness such as Ra. Many further studies on the wear modeling for polymers have provided equations for wear, however, a large majority of them have been tested on only one polymer system which limits the general applicability of the models. One notable observation, perhaps as a way of data presentation, has been found that the specific wear rate when plotted against the reciprocal of the product of ultimate tensile strength (OUTS) and % elongation at failure (e) furnishes a linear relationship [22]. Figure 6 gives such a plot where the data are obtained from literature. We may observe here that, regardless of the counterface roughness condition and the prevailing coefficient of friction for individual polymer, a linear relation is obtained.
Scratching of materials and applications
132 3.0E-04 -
XM
2.5E-04 A
O'1
K 2.0E-04 L
c~ <
E E
1.5E-04 J X
L. I,,.,
1.0E-04 - G H X><x I 5 . 0 E - 0 5 - E~
0.0E+00
C
B 0.0
~
' 0.1
" , 0.2
'
' 0.3
0.4
l/(~f e) (Mpa^-l) Fig. 6. A plot of wear rate (mm 3 mm l kg !) as a function of the reciprocal of the product of ultimate tensile stress and elongation to fracture. The data are taken from literature. A poly(ethylene) [20]; B - Nylon 66 [20]; C - PTFE [20]; D - poly(propene) [20]; E - High Density Poly(ethylene) [23]; F - Acetal; G - poly(carbonate) [20]; H - poly(propylene) [23]; I poly(ethyleneterephthalate glycol) [23]; J - poly(vinyl chloride) [23]; K - PMMA [23]; L poly(styrene) [19]; M - PMMA [23]. The plot shows reasonable linearity regardless of the polymer type. (Reprinted from ref [ 16] with publisher's permission)
Thus, several previous works have pointed out that the wear rate of bulk polymers is, in some way, strongly related to the ultimate tensile strength and bulk toughness related parameter such as % elongation at failure. This is valid when the counterface is a hard surface. The above macroscopic observations have been further proven by conducting experiments at nanoscale scratching using single diamond tip [24,25]. The polymers were scratched by a diamond tip of 90 ~ included angle and tip radius of 1 micron. The scratching was performed in cyclic fashion on a single track or on intersecting tracks. It was observed that for a very wear resistant polymer, such as ultra-high molecular weight poly(ethylene) (UHMWPE), there were no wear particle formation during scratching performed on a single track. Even repeated scratching on the same track had almost no effect on the wear particle generation. This is because most of the work-done was accommodated in the plastic deformation of the polymer making groove and side pileups in the very first scratching pass. Further scratching passes on the same track only helped in expanding the scratch width but only by a slight margin in the case of UHMWPE. We did not observe any wear particle even atter 10 cycles of scratching on the same track. In contrast, for a harder polymer, such as PMMA, even single pass scratching gave many loose wear particles which were seen either along the scratch or attached to the tip surface. When UHMWPE is subjected to intersecting scratching, where the scratching direction is changed by
Scratching of polymers: Deformation mapping and wear modeling
133
90 ~ alter every scratching pass, we observed the formation of wear particle after 10 cycles. The location of the formation of wear particle was at the comers of the intersection of the scratching directions. A new phenomenon of"wall" formation was identified as the main initiator of wear particles [24]. We observed localized regions of high plastic deformation at the four comers of the intersection. In addition, a small amount of material, which constituted the "wall", was shuffled from one scratching direction to another while undergoing low cycle fatigue process. The polymer material of the "wall" was the first to form a wear particle. Materials at the intersection also underwent low cycle fatigue (Fig. 7). Thus, the nanoscale scratching experiment gave the first real clue as to why a polymer, which may be soft but has high toughness property, can be wear resistant than a harder but less tough polymer.
Fig. 7. Intersecting scratching on PMMA, UHMWPE and PTFE surfaces. The figure shows the phenomenon of "wall" formation when one scratch is intersected by an orthogonal scratch. (Reprinted from ref [25] with publisher's permission)
A wear process, where a polymer is slid against a metal, may be approximated as a statistical event of intersecting scratching at various angles of intersection by large number asperities that are present on the hard surface. The process of wear particle generation will largely depend upon the ultimate tensile stress and the % elongation at fracture or the fracture toughness of the polymer as has been shown in macroscopic wear and nanoscratching experiments. Thus, the wear rates of polymers are not related to the indentation hardness which is only an indicator of the yield property of the materials in a largely compressive type of loading. The scratching technique may be used as an indicator of the wear resistance of polymer if we compare polymers' relative ability to produce wear debris rather than their computed scratch
134
Scratching of materials and applications
hardness. Further work is now underway to show that by estimating the wear particle surface area or volume during cyclic scratching (single track or intersecting) we can rank polymers according to their wear resistance. This ranking has almost linear correlation with the specific wear rate data obtained from a pin-on-disk test for the same polymers against a rough (R~ -~ 1.35 micorn) steel counterface [ 14]. The above explanations on the use of scratching for modeling abrasive wear behavior of polymers has shown the capability of this versatile test. Though scratch hardness as a mechanical property parameter does not relate to wear rate, the quantification of the wear debris produced during scratching is related to wear performance. It may be clarified here that scratch hardness is calculated based on the width of the scratch which is not necessarily produced by wearing of the material. For polymers scratched by low attack angle tip, most of the scratch width is formed as a result of the plastic deformation leading to side pileups. This type of plastically deformed pileup is not wear particle as it is still part of the bulk of the materials. This is the main reason why scratch hardness value does not relate to wear performance. Thus, the actual wear debris particles produced during scratching is the real indicator of wear and should be taken into account while using scratch test for abrasive wear modeling.
CONCLUDING REMARKS This chapter has looked at the usefulness of scratch testing for the tribological characterizations of bulk polymers. The examples drawn here are for two important aspects, scratch deformation mapping and abrasive wear modeling. It is shown that the scratch mapping of polymers can be used as a tool for understanding deformation characteristics of polymers for a wide range of scratching conditions of normal loads, scratching velocity, imposed strain (or indenter attack angle) and bulk temperature. This information can be particularly important where polymers' surface optical characteristic such as glossiness is important. Surface related properties such as friction, wear and fatigue are also related to the surface deformation characteristics of the polymer. The abrasive wear of polymers and scratch hardness do not show any correlation. However, by the new technique of conducting cyclic (single track or intersecting) scratch test, where the amount of wear debris generated in scratching is taken into account, it is possible to rank polymers according to their abrasive wear resistance using scratching data. This ranking correlates well with the specific wear data obtained in a pin-on-disk type wear test. Further, the nanoscratching technique has clearly demonstrated why the macroscopically observed wear rate for polymers shows strong relation to the ultimate tensile stress and the % elongation at failure. The abrasive wear process for polymers, when investigated at micron to nano scale, has shown that wear particle formation is a localized low cycle fatigue process.
ACKNOWLEDGMENTS The author would like to acknowledge the works carried out by many of his former and current coworkers that helped understand the fascinating area of scratch testing. Many of the stimulating discussions were extremely helpful. In particular, I would like to thank Professor Brian J. Briscoe, Dr. P. D. Evans, Dr. E. Pelillo, Mr. Brian K. P. Wong, Dr. Kaiyang Zeng, Ms. Joyce, P. Y. Tan, Mr. Rosli B. M. Sani, Mr. Mark C. W. Lup and Professor S. C. Lim.
Scratching of polymers: Deformation mapping and wear modeling
135
REFERENCES
.
3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
Mohs, F. Grundriss der Mineralogie, 1824 (English translation by W. Haidinger: Treaties on Minerology, Constable, Edinburgh, 1825). Briscoe, B. J., Evans, P. D., Pelillo, E. and Sinha, S. K. (1996) Wear, 200, 137. Jardret, V., Zahouani, H., Loubet, J. L. and Mathia, T. G. (1998) Wear, 218, 8. Briscoe, B. J. (1998) Tribology International, 31, No. 1-3, 121. Williams, J. A. (1996) Tribology International, 29, No. 8, 675. Budinski, K. G. (1997) Wear, 203-204, 302. Felder, E. and Bucaille, J. L. (2006) Chapter 2 in this book. Briscoe, B. J., Pelillo, E. and Sinha, S. K. (1997) Polymer International, 43, 359. Sung, L-P., Drzal, P. L., Vanlandingham, M. R. and Forster, A. M. (2006) Chapter 5 in this book. Briscoe, B. Pelillo, E. and Sinha, S. K. (1996) Polym. Eng. and Sci., 36(24), 2996. Bonne, M., Briscoe, B. J., Manimaaran, S. and Allan, A. (2003) Wear, 254, 55. Zum Gahr, K. H. (1987) in Chapter 5, Microstructure and Wear of Materials, Elsevier Science Publishers B. V. Briscoe, B. J. and Sinha, S. K. (2003) Materialwissenschaft und Werkstofftechnik, 1O-11, 989. Sinha, S. K., Lup M. C. W. and Lim, S. C. (data to be published elsewhere). Trezona, R. I. and Hutchings, I. M. (1999) Wear, 233-235, pp. 209-221. Sinha, S. K. and Briscoe, B. J. (2006) in Encyclopedia of Polymer Science & Technology, John Wiley & Sons, NY. Archard, J. F. (1953) J. Applied Physics, 24, 981. Ratner, S. N., Farberoua, I. I., Radyukeuich, O. V. and Lure, E. G. (1964) Soviet Plastics, 7, 37. Lancaster, J. K. (1969) Tribology Conv. 1969, Institute of Mechanical Engineers, London, pp. 100. Briscoe, B. J. (1981 ) Tribology International, 231. Hollander, A. E. and Lancaster, J. K. (1973) Wear, Z5, 155. Lancaster, J. K. in Encyclopedia of Polymer Science and Engineering (editor-in-chief: Jacqueline I. Kroschwitz), 1990, Wily, NY, 2"d Edition, 1. Shipway, P. H. and Ngao, N. K. (2003) Wear, 255, 742. Wong, K. W. P., Sinha, S. K., Tan, J. P. Y. and Zeng, K. Y. (2004) Trib. Letters, 17(3), 613. Sani, R. B. M., Sinha, S. K., Tan, J. P. Y. and Zeng, K. Y. (2005) Phil. Mag., 85(19), 2101.
136
CHAPTER 7
AN O V E R V I E W O F T H E P O T E N T I A L OF Q U A N T I T A T I V E C O A T I N G A D H E S I O N M E A S U R E M E N T BY S C R A T C H T E S T I N G Originally published in Tribologyhnernational vol 39, February 2006 S.J. BULL and E.G-BERASETEGUI
School of Chemical Engineering and Advanced Materials University of Newcastle, Newcastle-upon- Tyne NE1 7RU, U.K. E-mail: [email protected]
ABSTRACT The scratch test has been used to assess the adhesion of thin hard coatings for some time now and is a useful tool for coating development or quality assurance. However, the test is influenced by a number of intrinsic and extrinsic factors which are not adhesion-related and the results of the test are usually regarded as only semi-quantitative. The stress state around a moving indenter scratching a coating/substrate system is very complex and it is difficult to determine the stresses which lead to detachment. Furthermore the interfacial defect state responsible for failure is unknown. However, by a careful analysis of the observed failure modes in the scratch test (not all of which are related to adhesion) it is possible to identify adhesive failures and in some cases these occur in regions where the stress state is relatively simple and quantification can be attempted. Ideally engineers would like a material parameter (such as work of adhesion or interfacial toughness) which can be used in an appropriate model of the coating-substrate system stress state to determine if detachment will occur under the loading conditions experienced in service. This data is not usually available and the development of such models must be seen as a long term goal. In modem indentation and scratch systems the work of friction (or indentation) can be directly measured and the relationship between this parameter and adhesive failure can be demonstrated in some cases. This chapter reviews the main adhesion-related failure modes and the stresses responsible for them and indicates where quantification is possible illustrating this with results from hard coatings on steel, thermally grown oxide scales and optical coatings on glass. The use of empirical calibration studies, directly measured work of friction and quantification by finite element methods is discussed. KEYWORDS Coatings, thin films, adhesion, critical load.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
137
INTRODUCTION The scratch test has been used for many years to provide a measure of coating/substrate adhesion [1-6]. In the normal configuration of the test a diamond stylus is drawn across the coated surface under an increasing load until some well-defined failure occurs at a load which is often termed the critical load, Lc. Many different failures are observed which include coating detachment, through-thickness cracking and plastic deformation or cracking in the coating or substrate [7-10]. In fact it is usual that several different failure modes occur at the same time and this can make results of the test difficult to interpret. The failure modes observed in the scratch test depend on many factors and are most easily characterised in terms of the hardness of both substrate and coating (Figure 1). In the case of a typical Rockwell 'C' diamond indenter (120 ~ cone with 200ktm hemispherical tip), for soft coatings and soft substrates the test is dominated by plastic deformation and groove formation and little or no cracking is observed except at very high loads. For hard coatings on soft substrates deformation of the substrate is predominantly plastic whilst the coating may plastically deform or fracture as it is bent into the track created by plastic deformation of the substrate. Soft coatings on a harder substrate tend to deform by plastic deformation and some extrusion of the coating from between the stylus and the substrate may occur. Considerable thinning of the coating by plastic deformation will occur before plastic deformation and fracture of the substrate becomes significant. For hard coatings on hard substrates plastic deformation is minimal and fracture dominates the scratch response.
-
'
~
" Bulk
Interfacial
r,~
Failure Plastic
.
Deformation
\ ~
r~
,.Q Through-thickness Fracture
_
Coating hardness, H c
Fig. 1. Schematic showing the various scratch test failure modes which dominate as a function of coating and substrate hardness (Hc and H~, respectively).
As the indenter becomes sharper, plastic deformation becomes more localised to the surface and it is easier to prevent plastic deformation of the substrate. In such cases the results of the test are
138
Scratching of materials and applications
easier to analyse and quantify, particularly for more modem depth sensing indentation and scratch systems. However, damage to the diamond stylus during the test becomes much more significant as its sharpness increases. The choice of stylus thus represents a compromise between damage and ease of data analysis - for industrial hard coatings of reasonable thickness (>ll.tm) the Rockwell 'C' stylus has proved very successful whilst for sub micron coatings a conical indenter with a tip radius of a few microns is more suitable. The scratch test is not well-suited to measure the adhesion of sott coatings but can give some information if the interfacial shear strength is less than the shear strength of either the coating or substrate. In general, the scratch test is most effective if the substrate does not plastically deform to any great extent. In such cases the coating is effectively scraped from it and the uncovering of the substrate itself can be used as a guide to adhesion. However, it is difficult for this to be quantified. Detection of the uncovered substrate may be a problem unless post facto chemical analysis methods can be employed. However, some success is possible by measuring the change in friction during the scratch if the coating and substrate behave differently. For instance, an increase in friction may be observed if a high friction coefficient substrate in uncovered during the test. The scratch adhesion test is much more useful for hard coatings, particularly when these are deposited on softer substrates. For a harder coating on a soft substrate the spallation and buckling failure modes arise from interfacial detachment [8,9,10] and can thus be used as the basis for an adhesion test. Both may be quantified in some circumstances and are discussed in this paper. The origin of these failure modes and the theoretical basis for analysing them is introduced in the next section, together with finite element results aimed at improving quantification.
OVERVIEW OF SCRATCH ADHESION TESTING
Requirements for quantification of the scratch test If the scratch test is to be fully quantitative it must deliver a parameter which is representative of the state of adhesion of the interface but is not related to the other properties of the coating/substrate system such as hardness. The best parameter for this is work of adhesion which is a measure of the chemical bonding across the interface. However, most adhesion tests do not measure this basic adhesion but produce a practical adhesion measurement conflating basic adhesion with other factors which can be specific to a given material pair or test method [11-13]. Film adhesion is otten characterised by the strain energy released per unit increase in delamination area, G, which is sometimes referred to as the interfacial fracture energy and can be used to generate an interfacial fracture toughness, K~. For most practical purposes this measurement of practical adhesion is sufficient but it should be corrected for method-specific factors to ensure that the test is widely applicable and the data produced can be compared with that from other test methods. The scratch test is usually only regarded as semi-quantitative as there are a number of intrinsic and extrinsic parameters which are known to affect the measured critical load (Table 1). Many of these intrinsic factors are instrument-specific and require a careful calibration approach if results are to be compared between instruments. However, the extrinsic factors such as coating thickness and substrate hardness must also be known if the results of the test are to be understood. These parameters, together with the residual stress in the coating and its Young's
An overview of the potential of quantitative coating adhesion measurement by scratch testing
139
modulus, are an important requirement for the models of the failure mode used to generate interfacial fracture toughness. Table 1: Intrinsic and Extrinsic factors in the scratch test. Intrinsic Extrinsic Loading rate [3, 39, 40] Substrate properties (hardness, elastic modulus) [40] Scratching speed [3, 39, 40] Coating properties (thickness, hardness, modulus, residual stress) [39, 40] Indenter tip radius [3, 39] Friction coefficient [3, 24] Indenter wear [39] Surface roughness [40] Machine stiffness/desi~;n
Fig. 2 (a) Scanning electron micrographs (backscattered image) of scratch tracks in a 100nm aluminium coating on 304 stainless steel showing the stripping of the coating at the critical load with a sharp change (top scratch) and a more gradual change as the aluminium coating is thinned (lower scratch). (b) Associated friction traces showing a sharp and gradual transition in the friction coefficient.
Scratching of materials and applications
140
There are thus four requirements for a quantitative scratch adhesion test:1) An adhesion-related failure mode, 2) A well-defined failure mechanism, 3) A method to identify that adhesion failure has occurred, where it is located and the size of the failure, 4) A method of determining the stresses which cause failure. These will be discussed in more detail in the following sections.
Scratch test failure modes Soft coatings: In general soft coatings (hardness<5GPa) fail by plastic deformation whether deposited on softer or harder substrates. Coatings deposited with a porous microstructure (e.g. the open columnar structures produced by vapour deposition at low temperatures) may also show some evidence of fracture but this is not widespread. The scratch test is not very useful for assessing adhesion unless the interfacial shear stress is less than the shear strength of the softer component. In such cases stripping or flaking of the coating may occur if the adhesion is very poor but often there is little to see but a plastic groove after the test is complete. When a soft coating is deposited on a very different, harder substrate, such as aluminium or gold on glass, the detection of interfacial failure is much easier. As the load is increased in the scratch test the soft coating is progressively plastically deformed until at the critical load the substrate is uncovered. This can be detected by a colour change or by the use of surface analysis techniques such as x-ray photoelectron spectroscopy (XPS) which are surface sensitive. However, XPS analysis is not always practical since failure does not occur exactly at the interface - in such cases x-ray mapping or backscattered imaging in the SEM can be used to determine a critical load but this does not represent the load for interfacial detachment (Figure 2a), rather this is the load at which the coating has been scraped off the substrate. Unless there is a sharp transition when the coating is stripped, indicating some adhesive failure, selecting a critical load for coating detachment is almost impossible in such cases but the scratch test critical load may give an indication of the resistance to scratch damage of the coated surface. An alternative method to detect the appearance of the substrate is to analyse the friction traces developed during the scratch test. In the case of aluminium coatings on 304 stainless steel the friction coefficient increases when the substrate is uncovered (Figure 2b). The sharpness of the friction transition in the plot of friction coefficient versus load mimics the sharpness of the transition observed in backscattered electron images in the scanning electron microscope. However, such clear results are not often observed. The earliest attempts at scratch test quantification by Benjamin and Weaver [7] are most applicable when thin coatings are plastically deformed in the scratch test. According to these authors the critical shearing force for coating removal, x, is a function of the scratch geometry, the substrate properties and the frictional force on the stylus. Thus, for a stylus of radius, R,
r =
kAH ( R 2 - A2) ~
(1)
An overview of the potential of quantitative coating adhesion measurement by scratch testing
141
where the radius of the contact A=(Lc/nH) ~/2, Lc is the critical load, H is the hardness of the substrate material and k is a constant varying between 0.2 and 1.0. The critical shear stress increases as the substrate hardness increases which agrees with experiment. This model assumes full plastic deformation (which is only applicable in a limited range of cases) and does not show the influence of coating thickness. For soft polymeric coatings on harder metallic substrates the shear stress applied to the coating during the scratch test can lead to regions of delamination extending ahead of the stylus. In such cases a fracture mechanics model has been developed to assess adhesion based on the assumption that the stress field around a moving indenter can be given by the Boussinesq solution [14, 15]. This is clearly not a complete solution as it does not deal with elastic mismatch at the coating substrate interface but generates strain energy release rates comparable to those obtained by different adhesion test methods. However, the method requires knowledge of the area and geometry of delamination which is not always easy to determine if the coating is not transparent and the same mechanism of failure is not often observed for other coating systems. For this reason the model is not widely applicable. In general only semi-quantitative measurements of adhesion of soft coatings can be achieved by scratch testing and alternative adhesion test methods are preferred (e.g. tensile and peel tests [16], blister tests [17], superlayer tests [18]). Since soft coatings are usually quite ductile and may be mechanically manipulated without failure such mechanical tests are relatively easy to perform. The main problem is that the work done is not solely governed by the energy expended in detachment and deconvoluting the measurements in a way which separates the work of adhesion from other energy absorbing mechanisms is difficult [19]. Hard coatings: In the context of this chapter hard coating refers to coating materials with a hardness of greater than 5GPa. The failure modes can be broadly split into four categories: Through-thickness cracking - including tensile cracking behind the indenter [8,20], conformal cracking as the coating is bent into the scratch track [8,20] and Hertzian cracking [8]. These cracks may extend into the substrate if it is sufficiently brittle but are usually stopped at the interface in a hard coating on a softer substrate. Coating detachment - including compressive spallation ahead of the indenter [8,20], buckling spallation ahead of the indenter [8] or elastic recovery-induced spallation behind the indenter [8, 21 ]. .
.
Chipping within the coating- usually observed for thick coatings on a softer substrate. The scratch test cannot practically measure the adhesion of coatings greater than 50~tm thick in its conventional form since it is impossible to generate sufficiently large stresses at the interface before chipping of the coating occurs. Chipping within the substrate - for brittle coatings on brittle substrates where the adhesion is good the system tends to behave in the same way as a brittle bulk material and unless the coating is sufficiently thick chipping of the substrate will occur.
The type of failure which is observed for a given coating/substrate system depends on the test load, the indenter radius, the coating thickness, the residual stress in the coating and the
142
Scratching of materials and applications
substrate hardness and interfacial adhesion. Generally the critical load at which a failure mode first occurs, or occurs regularly along the track, is used to assess the coating though there is a distribution of flaws and hence of failures in most cases [6]. Comparisons between different coatings are only valid if the mechanism of failure is the same which requires careful post facto microscopical examination for confirmation. The adhesion related failures which are the basis of the scratch adhesion test for hard coatings are buckling and spallation [9,10] and are described in more detail in the next section.
Failure mechanisms related to adhesion for hard coatings Buckling: This failure mode is most common for thin coatings (thickness typically <10~tm) which are able to bend in response to applied stresses. Coatings much thicker than this limit will tend to show through-thickness fracture at stresses lower than those necessary to cause buckling and will fail by wedge spallation (see next section). Failure occurs in response to the compressive stresses generated ahead of the moving indenter (Figures 3a and 3b). Localised regions containing interfacial defects allow the coating to buckle in response to the stresses and individual buckles will then spread laterally by the propagation of an interfacial crack. Spallation occurs when through-thickness cracks form in regions of high tensile stress within the coating. Once the buckle has occurred, the scratch stylus passes over the failed region crushing the coating into the surface of the scratch track formed in the substrate. Coating removal can be enhanced at this point or the failure may disappear completely depending on its size and the toughness of the coating. Buckling failures typically appear as curved cracks or patches of damage extending to the edge of the scratch track or beyond. They are often delineated by considerable coating fragmentation and have major crack planes perpendicular to the coating/substrate interface. In most cases buckles form in the region of plastic pile-up ahead of the moving indenter (Figure 3c). The size of the buckle is typically less than or equal to the extent of pile-up. This would imply that the pile-up process controls the buckle failure mode to a great degree. This explains, to a large extent, the increase in critical load with substrate hardness for titanium nitride tool coatings on steel which is often reported [ 1] since in such coatings the buckle failure mode dominates. As the steel hardness increases plastic pile-up ahead of the indenter is reduced and the bending stresses induced in the coating by the pile-up are limited. A higher normal load is needed to develop equivalent pile-up and bending stresses and thus the critical load increases. The correlation between buckle diameter and pile-up diameter is very close for alumina scales on the oxide-dispersion strengthened alloy MA956 or TiN coating on stainless steel [10]. For TiN coatings on steel, changes in buckle diameter can be produced by changes in interfacial structure and adhesion but within limits defined by the size of pile-up.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
143
(a)
Fig. 3. Buckling failure mode in the scratch test; (a) pile-up ahead of the moving indenter and (b) interfacial failure leading to buckling. Through-thickness cracking results in removal of coating material. (c) Scanning electron micrograph of buckle failures in TiN coated stainless steel.
144
Scratching of materials and applications
According to Evans [22] the critical buckling stress Ob is given by:
(2)
where Ec and vc are Young's modulus and Poisson's ratio of the coating, t is coating thickness and R is the radius of the buckled region. This predicts that the critical buckle stress increases with coating thickness as is mostly observed. However, this equation assumes a planar interface which is not always the case. For a curved interface, Strawbridge et al [23] have shown that a tensile stress, Or, is generated normal to the interface by the action of the stress in the plane of the surface, o0, and the magnitude of this stress at the interface is given by: t O"t = O ' 0 ~
(3)
Ri where Ri is the radius of curvature of the curved interface and t is the coating thickness. In the scratch test the applied stress (i.e. the sum of any residual stress and the stresses introduced by the scratch stylus) determines o0 and, in the case of a coating bent over the pile-up ahead of the moving indenter, Ri represents the radius of curvature of the pile-up. Since the amount of pileup depends on the hardness of the substrate, the critical load in the scratch test should thus be proportional to Ri and inversely proportional to t. As the hardness of the substrate increases so Ri tends to increase and this behaviour is maintained for a fixed value of t but the critical load is not inversely proportional to thickness. This is due to the fact that Ri is actually a function of t the extent of pile-up decreases as t increases as mentioned previously. In fact for thin TiN coatings on a range of steels experimental results indicate that Ri is proportional to t 2 which would imply that the critical load is in fact proportional to coating thickness which is close to what is observed. However, much more data are necessary to determine the validity of this observation.
Fig. 4. Schematic diagram of the geometry of a chipped segment of coating.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
145
Buckling failures have been observed around static indentations and scratches in hard coatings on hard substrates produced by facetted indenters (e.g. by a Berkovich tip commonly used in nanoindentation testing). In such cases the detached buckle is often bounded by radial cracks and plastic deformation in the substrate is limited. This process can be analysed based on model suggested by Thouless [24] which assumes that: (i) the in-plane load on the delaminated sector due to indentation causes the growth of the delamination area, and (ii) the coating chips at the moment of buckling of the sector due to the same in-plane load. According to den Toonder et al [25] the interfacial fracture energy can be calculated using:
a_+
I
Et5 L
F/= 1.42-~--
t(1 - v)cr~ + - - + E
a.+ L
fl~'~
3.36(1- v)t 3orr L -ff~ L2 a+fl~. L
(4)
where E is the Young's modulus of the coating; t is the thickness of the coating; v is the Poisson's ratio of the coating; ~r is the residual stress in the coating and a, L and [3 define the geometry of the chipped piece (Figure 4). Reasonable values of interfacial toughness for hard films on silicon, glass and other hard substrates are produced by equation (4). This approach may be applied to cracks associated either with static indentations or scratches but the fracture energy is generally different in the two cases which implies that there is a frictional contribution to the failure in the scratch test which has not been considered.
Wedge Spallation: For thicker (>101am) coatings where bending is less common the buckling failure mode is not observed. In fact the coating can suppress the formation of a narrowly defined pile-up region (Figure 5d) and the stresses ahead of the indenter are less complex. Adhesive failure now occurs by a different mechanism (Figure 5a-c). Initially compressive shear cracks form some distance ahead of the indenter through the thickness of the coating. These propagate to the surface and interface and generally have sloping sides which can act like a wedge. Continued forward motion of the indenter drives the coating up the wedge causing an interfacial crack to propagate. As the extent of interfacial failure increases the wedge lifts the coating further away from the substrate creating bending stresses within it. Large enough displacements will cause a region ahead of the indenter to be detached in response to the tensile bending stresses created. When this happens the scratch diamond tip can drop into the hole lef~ by removal of the coating (Figure 5d) and there is a dramatic increase in scratch width and scratch depth. Pile-up is then often seen beside the track until the stylus climbs up the wedge and out of the hole. Whereas such large failures are often observed for alumina scales on MA956, much smaller failures are often produced for vapour deposited TiN coatings and it is rare that the stylus drops into the hole left by the spalled coating. In this case the stylus passes over the edge of the spalled region creating considerable microfracture in the coating as it passes. The wedge spallation failure mode depends on two distinct processes occurring [22]. Firstly a compressive shear crack must form and then interfacial detachment occurs. According to Evans [22] the biaxial stress necessary to cause the wedge crack, r is given by
Scratching of materials and applications
146
(a) ~,~ ---...~~
(b)
,,,
_J
Fig. 5. Wedge spallation failure mode in the scratch test; (a) wedge crack forms some way ahead of the moving indenter; (b) continued forward motion drives the coating up the wedge opening up an interfacial crack; (c) through-thickness cracking close to the indenter leads to spallation. (d) Scanning electron micrograph of a wedge spallation failure in an alumina scale on MA956 oxidised at 1250~ for 100h.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
<"
4E
147
(5)
where Gf is the coating fracture energy and ~. is the width of the wedge spalled region. The biaxial stress to produce the spall, Crsp, after shear cracking has occurred is given by
E
(6)
where Gi is the strain energy release rate for a crack in the interfacial region (i.e. the interfacial fracture energy) which, in an ideal case, can be equated with the crack surface energy, "/, and hence basic adhesion. However, for all real interfaces other energy dissipation mechanisms are likely to be operating (e.g. plasticity, microfracture, heat generation, etc.) so Gi can be taken as a measure of practical adhesion. Since to get a visible wedge spall both the through-thickness and interfacial cracks must be formed the total failure stress, OF, is given by the sum of equations (4) and (5). As Equation 5 has a 1/-4t- dependence the critical load for wedge spallation is expected to decrease as coating thickness increases. There is no requirement that the crack propagates exactly along the interface in this analysis though this is often the case if the adhesion is poor and the interface is sufficiently planar.
STRESSES RESPONSIBLE FOR FAILURE The stresses around a moving indenter sliding across a coating/substrate system are complex and no analytical model exists which fully describes what is observed. Some progress has been made with finite element modelling, particularly in cases where both coating and substrate remain elastic, but this work is a simplification compared to what usually occurs in a real scratch test. A number of improvements to the approach are required and still need to be addressed:1. Realistic materials models are required which include the elastic properties of coatings and substrate, their yield and fracture strength and work hardening characteristics. 2. Cracking and the modification of the stress state by the presence of cracking needs to be implemented for hard coatings or substrates. 3. Modification of the stress field by changes in indenter/coating friction requires data for each coating/substrate system of interest. 4. Roughness of the surfaces and interface causes localised stress concentrations which are often ignored in basic finite element models and can lead to considerable reductions in the normal load at which failure occurs. The yield, work hardening and fracture properties of many coatings are not well known and getting good input data for finite element models can be problematic. Combinations of finite element modelling and experiment are attempting to address this problem. For instance, Jiang et al have developed finite element models of scratches using more realistic elastic-plastic materials models [26] and Holmberg and co-workers have developed a method for determining the fracture properties of coatings from the tensile cracking which occurs behind the indenter in the scratch test [27]. However, much more progress is needed, particularly as adhesion failures
148
Scratching of materials and applications
often arise after through-thickness fracture has already occurred and the modifications which this generates to the stress field around the moving indenter need to be incorporated in any finite element model if accurate failure stresses are to be determined. Also, any failure mode which occurs close to the indenter is likely to be in a region where the stress field is changing rapidly and it will be difficult to determine exactly which stresses are responsible for failure. Therefore, finite element modelling can currently only be regarded as providing a guide to the stress fields around the moving indenter but can help to identify those failure modes which are generated by simple stress conditions. If finite element modelling is to be used to predict failure in the scratch test a suitable failure criterion needs to be developed. Developing stresses is one part of this but the defect distribution in the coating and interfacial region will also need to be known if accurate predictions are to be made. Relatively little work has been done in this area to date.
DETECTION OF FAILURE A number of techniques have been used to identify the onset of adhesion failure in the scratch test including post facto microscopy, acoustic emission and friction analysis. Perhaps the most useful and reliable is imaging the scratch by an appropriate microscopy technique. For the conventional scratch test where a Rockwell 'C' stylus is used to generate failure it is relatively easy to identify failure using reflected light microscopy or scanning electron microscopy as the failures are usually tens of microns across. In such cases adhesive failure detected by acoustic emission or friction changes usually correlates well with the failures visible by microscopy. However, it is ot~en observed that acoustic emission detects failures which cannot be seen in the microscope, perhaps failures of the interface which are not associated with chipping. AE signals result either from the sudden release of elastic energy or from surface interactions such as friction and adhesion. Sudden energy release occurs during unstable crack growth, highspeed phase transformations, and plastic instabilities. A transducer to measure the acoustic emission is attached to the scratch slider and its output can be correlated to events which occur along the scratch track. At constant load, as the scratch is created a baseline AE response is established due to plastic deformation and friction/adhesion and individual AE responses from fracture are superimposed upon this. The magnitude of the acoustic emission signal depends on the size of the crack produced since the energy in the AE signal scales with the energy released in the process [28]. For coatings thicker than a micron, the area of adhesive failure is large compared to the area of through-thickness cracks; a large jump in the acoustic signal is thus a reasonable indication of adhesive failure. However, as coating thicknesses reduce,.the size of the adhesive and through-thickness failures also reduce and are more comparable. The acoustic signal is reduced and it becomes much more difficult to determine what sort of failure has occurred. Indeed, at small scales the generation of acoustic emission can represent the emission of bursts of dislocations as well as fracture [29] and acoustic emission measurements from nanoindentation and nanoscratch tests can identify that these mechanisms are operating when combined with careful microscopy. When nanoscratch testing is applied to coatings considerably less than l~tm thick, light microscopy is no longer suitable and other techniques are essential to assess failure mechanisms. High resolution scanning electron microscopy or atomic force microscopy can offer some
An overview of the potential of quantitative coating adhesion measurement by scratch testing
149
information but there is a limit to what can be seen - adhesion failure may be visible but through-thickness cracking is often not unless the crack opening is significant. In such cases, changes in the indentation load-displacement or friction curves are a good indication that failure has occurred but determining the failure mode still requires careful microscopy. The advantage of using load-displacement or friction-displacement curves to determine the onset of failure is that this enables quantification of the failure in terms of the work done. This will be discussed in more detail later in the chapter.
EXPERIMENTAL METHODS In this study the scratch adhesion behaviour of relatively thick hard coatings on soft substrates (coating hardness >10GPa, Substrate hardness < 5GPa) and thin hard coatings on hard substrates has been investigated (coating and substrate hardness >7GPa). Similar samples have been investigated in previous work [ 10] but in this case the residual stress in the coating has been carefully determined and the quantification is expected to be more accurate.
Hard Coatings on soft substrates Samples of 304 stainless steel and the oxide dispersion strengthened (ODS) alloy MA956 (composition in Table 2) were cut into 20 x 10 x 2mm sections, polished to a l l.tm diamond finish, and cleaned and degreased in isopropyl alcohol prior to use. The stainless steel coupons were coated with TiN by sputter ion plating (SIP [30]) or arc evaporation [31] in commercial coating equipment at a temperature of 500~ Coatings with thicknesses in the range 1 to 25~tm were deposited with a 120 nm titanium interlayer to promote adhesion. The alloy samples were isothermally oxidised in air at temperatures between 1150~ and 1300~ for times up to 1400h to produce alumina scales up to 20ktm in thickness. The thickness of all scales or coatings was measured by both ball cratering and metallographic cross sections. Table 2: Composition and properties of the materials investigated in this stud), Alloy Composition Coating Young's Poisson's ratio modulus (GPa) ,..
304 stainless MA956 Float glass
Fe-18Ni-9Cr1Ti Fe-20Cr-5A10.4Ti-0.5Y203 73%SIO2, 15%Na20, 10%CaO+ traces MgO, A1203 and K20
SIP TiN Arc TiN ~-A1203
450-J:35 500-a:62 388•
0.28 0.28 0.26
TiOxNy
122+ 15
0.25
Scratch testing was performed using a CSEM scratch tester fitted with a Rockwell 'C' diamond (2001.tm tip radius). This is a dead-loaded machine where a separate scratch is made for each applied load. A 3mm scratch was made at each load. For the tests reported here scratches were made at 2N intervals starting at 2N. Care was taken to place the scratches sufficiently far apart
150
Scratching of materials and applications
so that their deformation regions did not overlap. Critical loads for each failure mode were determined by post facto microscope examination of the scratch tracks. The critical load criterion used was the lowest load at which the failure occurred more than twice along the scratch track. Since the total number of wedge cracks produced was low it was not possible to perform a full Weibull statistics analysis [6].
Hard coatings on hard substrates 400nm thick titania coatings were deposited on glass by sputtering. The 10cm by 10cm float glass samples were 4mm thick and were coated on the air side. Titania was sputtered from a metal target with a pulsed DC mode (2.43KW) in an argon/nitrogen/oxygen mixture and is referred to as TiOxNy hereafter. Nanoscratch testing was performed using a Hysitron Triboindenter- scratch tests were performed with a Berkovich indenter (200nm tip end radius) in the edge leading configuration. The total scratch length was 10~m. Along the length of the scratch the load was ramped to the maximum (10mN) at lmN/~tm over 30 seconds. At the end of the test it was ramped down to zero in the same time with the indenter nominally stationary. AFM traces were carried out with the same tip which made the s c r a t c h - the relatively blunt tip means that these images have limited resolution when compared to conventional AFM however this technique does make it very much easier to find the scratches and image them successfully. Scratches were viewed by scanning electron microscopy and energy dispersive x-ray microanalysis was used to determine if spallation of the coating around the scratch had occurred. It was only possible to find those scratches where spallation had occurred in the SEM as through-thickness fracture is almost invisible even at the highest practical magnification. The image quality is poor due to the effects of charging- gold or carbon coating could not be used as this covers the cracks and reduces their visibility even further.
Finite Element Modelling Previously finite element modelling of the scratch test for 21am TiN on stainless steel had been undertaken by the authors using DYNA3D [32] but this analysis has been repeated using better substrate mechanical properties obtained from tensile tests on small samples cut from the substrate after coating. A two-dimensional plane strain model, which models the indentation as a cylinder rather than a sphere, was implemented in the ANSYS commercial Finite Element code in order to achieve reasonable run times. The mesh was chosen to be symmetric about the y axis and was refined in the region below the indenter. The coating thickness was set at 2p.m and reasonable materials properties were used for both substrate and coating. For the stainless steel substrate plastic deformation and work hardening was allowed (yield stress, t~y=450MPa; work hardening exponent, n=0.26) whereas for the TiN coating deformation is elastic up to 8GPa with no work hardening. The indenter/coating friction coefficient was fixed at 0.15. This is an approximation to the mechanical response of the system because it is expected that the TiN coating will undergo fracture if relatively modest tensile stresses (-500MPa) are generated. As in the previous work models were run for comparison: a static indentation where the maximum vertical indenter displacement was 2pm and a simulated scratch where the indenter was allowed to indent to 2pm and was then moved tangentially 101am. The new FE results are very similar to those obtained previously. Given the uncertainties about the difference between cylindrical and spherical indentations, as well as questions about the quality of the materials
An overview of the potential of quantitative coating adhesion measurement by scratch testing data, the absolute stress values generated must be questionable. between static indentation and scratches is instructive.
151
However, the difference
Data for Modelling In order to calculate the practical adhesion in terms of the interfacial fracture energies Young's modulus and Poisson's ratio of the coating are necessary. For all the scales and coatings investigated on soft substrates in this study the Young's modulus was determined by nanoindentation testing on coatings that were at least 81am thick using a Berkovich indenter using the method of Oliver and Pharr [33]. In order to reduce the scatter in the data the sample surface was polished prior to testing at a maximum load of 10mN; under these test conditions the contribution from the substrate is expected to be minimal. Properties for the coatings on glass were obtained from very low load indentation testing (<100~N) on the 400nm coating using a very sharp cube comer indenter (tip end radius <50nm). Young's modulus was again extracted from the unloading curve by the method of Oliver and Pharr [33] - the measured moduli values were extrapolated to zero depth to account for the effect of the substrate. Indents were imaged after testing using the tip that made them to confirm that no detectable pile-up occurred and that the measured area of the indent was close to that determined from the Oliver and Pharr approach - in such circumstances the Elastic Modulus determined by the Oliver and Pharr method is the same whether a well-calibrated Berkovich or cube comer indenter is used. However, if significant pile-up occurs it can cause errors in the data obtained using either indenter and this may be worse for the cube-comer indenter. Correction of the areas for pile-up is essential in such circumstances. Quoted values in Table 2 are the average of ten measurements and are similar to what is expected for such coatings. There was a small variation in Young's modulus of the oxide scale on MA956 depending on the oxidation temperature but this was not significant in the temperature ranges investigated here. Handbook values have been used for Poisson's ratio in all cases.
RESULTS Hard Coating on Soft Substrate Scratch Test Failure Load Regimes: For all coatings investigated the critical load for buckle formation increases as the coating thickness increases (Figure 6a and 6b). Wedge spallation does not occur until higher coating thicknesses and the critical load for wedge spallation decreases as thickness increases. This is exactly the same as has been observed previously [9,10] and is broadly in agreement with the theoretical predictions in Section 2.3. The critical load for buckle failure for the are evaporated TiN is lower than that for the sputtered coatings which might imply poorer adhesion but the wedge spallation critical loads are more comparable.
Finite Element Results: The main stress components in the coating at the coating/substrate interface have been extracted from the finite element data for both static indentation and scratching and are plotted in Figure 7. g• (parallel to the surface) is tensile beneath the indenter in static indentation due to the stretching and bending of the coating as it is dragged into the impression as the substrate plastically deforms beneath it. At the edge of the contact the bending is in the opposite sense and compressive stresses are observed. This is exacerbated by pile-up, gxx quickly falls to zero outside the pile-up region. On moving the indenter the compressive stress is increased ahead of the indenter and reduced behind it, probably due to changes in the amount of bending in the coating. Well outside the contact region a compressive
Scratching of materials and applications
152
(a) MA956 4000
. . . . . . . .
!
. . . . . . . .
9
3500
~
................................... ,'-.................. _/%1 .......................... 7 ..... ~
3000
!
............
. . . .
1150~
Buckle
9
1250~
Buckle
9
1150~ Spall
. 15 t~ o t _
_.=
Spall
2500
2000
ii 1500
1000
...... 9 ........ "..................................................... 9
500
.
.
.
.
.
0
.
.
.
.
5
.
.
.
.
.
10
.
! ................
.
i
15
Thickness
. . . .
20
25
(pm)
(b)
PVD TiN 4500
4000
................................ --~......................................................
3500
....................... [ ] ......... i . ~ - ............
:: ~ -
-o o
9 []
SIP , SIP ----0-- Arc Arc
i ~
TiN-Spall T i N - Buckle TiN- Spall TiN - Buckle
I |.. I |
3000
o . m
2500
c-
2000
O9
1500
i
1000 .
500
. 0
.
.
.
. 5
.
.
.
.
.
.
.
10 Thickness
Fig. 6.
.
.
.
.
.
.
15
.
.
20
25
(pro)
V a r i a t i o n o f c r i t i c a l l o a d for w e d g e or b u c k l e f o r m a t i o n
t h i c k n e s s for (a) A l u m i n a o n M A 9 5 6 a n d (b) T i N o n s t a i n l e s s steel.
as a f u n c t i o n o f c o a t i n g
An overview of the potential of quantitative coating adhesion measurement by scratch testing
153
(a) Load Scratch., directioff
( ,
iy
r
x
(b) 1 104
'
'
o
5000
'
'
!
'
'
"
'
T
.
.
.
.
Indent Scratch
' 9
-~--
~ i
................................ ,,%
i
I:1.
~
t;
, ~i/
.......
I
-5000
,
-1
104
-100
-50
0
50
100
x (pm)
Fig. 7. Stress components in the coating next to the coating/substrate interface determined for static indentation and scratching using the Finite Element code ANSYS (a) axis and scratch directions (b) 6xx, (Continue to the next page)
Scratching of materials and applications
154
(c)
lOOOi
. . . . . . . .
,,
~!
. . . .
!
-1000
..................... i .......
' i / ..... ....................................
1
"~" 12.
.
-2000
....
~ t
....................................
>,
t3
-3000 .................................... -4000
...............................
~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
'
-5ooo
.....................
I
~
: l,'J .. . . . . . . . . . 8 . . . . . . . . . . . . . . . . . . . . . . .
~ ---e--
Indent Scratch
J...4
I
1
'
I
-6000
I -100
-50
0
50
100
x (pm)
(d) 1500
--,
,
,
,
[
~
,
-, -I
1000
......................
.
. . . . . . . . . . . . '~ . . . . . . . ~.I.
,
,
~'
-
. . . . .
qt 1'.
,.~
-!, . . . . . . . . . . . . . . . . . . . . . . . . . . . .
,
---;
-!,
i
13.. v >., x
0 ~~~~.t -5oo
!/~~ ......~......,,,,i,,~~,
............................... :
~--. . . . . . . . . . . . . . . . . . . .
~
...............................
7
I l II '
.......................,. . . . . . . . -1000
:i . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~
--4~---Indent --~--Scratch
I-~ | -'
L - 1500 -100
-50
0
50
100
x (~tm)
Fig. 7. (Continue from the previous page) Stress components in the coating next to the coating/substrate interface determined for static indentation and scratching using the Finite Element code ANSYS (c) ~yy and (d) X~y.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
155
stress exists ahead of the indenter which approximates to a state of pure compression. In the region of bending at the edge of the contact the shear stress component Xxyis also significant and the values at the leading edge of the indenter are increased by sliding. The stresses perpendicular to the interface Oyy are compressive in the contact region as expected with tensile stress regions just outside in the bending zone. These tensile stresses are much reduced in the scratch case. Clearly, a very complicated stress state exists in the pile-up region close to the indenter where bending of the coating occurs. This is the region where buckling occurs and it makes the relationship between Orb and L~ difficult to define. The stress state well ahead of the indenter where wedge cracking occurs is much simpler and a linear relationship between ~ and OF is expected for a given coating/substrate/indenter combination. The compressive stress is reduced the further ahead of the indenter it is determined but the rate of change is relatively small. For a precise knowledge of the stresses responsible for detachment an interface fracture criterion is required. The FE results here imply that failure will occur just ahead of the moving indenter where the compressive stresses are highest. Two distinct stress maxima are observed in Oxx, one at the edge of the contact where bending stresses and compression ahead of the moving indenter combine which is where buckle failure is most likely and a second maximum some distance ahead of the contact where wedge spallation is the preferred failure mode. For the 21am thick coatings examined here the buckling failure is most likely as the stresses at the contact edge are highest as expected. For thicker coatings where the second moment of area is larger and bending is reduced the compressive stress maximum further away from the contact is largest and wedge spallation occurs. At intermediate coating thicknesses, the stresses are more comparable and either failure mode is possible, depending on the interfacial defect distribution. Alumina Scales on MA956 4.3 13.
v(.9
4.2
O)
1
l
]
]--
~
I
]
!
I
.
~
4.1 '
~9 rr ~
'
l
m ...... j .......... ! .......... < ......... E...... \
N
1
1
1
1
I
'
T
,
1
]
9 9
8'lam scale "1 121am scale ~~
"
201am scale I f
~
,
4
:
\
19=0.39MPa
_-
3.9
t_
~ E
3.8
t=
3.7
1g=O.45M/Sa
\:
n
3000
3500
0 0 n" 3.6
1500
2000
2500
W e d g e Spallation Critical Load,
Lc (g)
Fig. 8. Variation of critical load with residual stress for 81am and 201am alumina scales grown on MA956 at a range of oxidation temperatures.
Scratching of materials and applications
156
Quantification of Failure Stresses: The stresses responsible for coating detachment, err, are a combination of the residual stresses remaining in the coating at room temperature, CrR, and the stresses introduced by the scratch stylus, Os. Thus cr r = o" R + a s
(6)
OR Can be measured for both TiN and alumina coatings by x-ray diffraction using the sin2~ method [34]. The stresses induced by the indenter have been determined empirically. The critical load for coating detachment is known to decrease as the residual stress in the coating increases for a wide range of coatings such as TiN [35]. In the case of TiN coatings the residual stress can be increased by increasing the energy or flux of ion bombardment during deposition [36]. Equating the change in scratch test critical load with the difference in measured residual stress enables a calibration factor to be determined. For sputtered TiN 1g normal load in the scratch test equates to a 2.3 MPa compressive stress ahead of the indenter whereas for Arc TiN, this rises to 4MPa with an error of about 10% in both cases. For the alumina scales grown on MA956 the residual stress can be changed by altering the oxidation temperature. Plotting the measured residual stress against critical load allows a calibration coefficient to be determined from the slope of the graph (Figure 8). Experiments have been performed at three different oxide thicknesses. The calibration constant is almost the same in each case and is effectively constant within experimental error. Thus, for this material lg normal load in the scratch test equates to an average value of 0.40a:0.04 MPa. In all cases a linear relationship between critical load and stress is assumed which appears reasonable in these systems but may not be valid in all cases. 20
,
l
~
l
l
'
T
"
'
'
~ '
'
T
] ~
r
'
1
I"
]
'
'
T
j
!
4
~
.~
d
10
.
.
.
.
.
.
.
.
8
.
.
Ie
-. 0
~
0
I
,
5
,
,
,
I
10
,
,
,
,
I
15
~
,
,
,
~
20
L
" ,
,
,.;
25
Thickness (lam)
Fig. 9. Variation of compressive failure stress with coating thickness for alumina on MA956 and TiN on 304 stainless steel. The scratch induced stress has been added to the residual stress in the scales and coatings.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
157
Calculated failure stresses for alumina scales on MA956 and TiN on stainless steel are shown in Figure 9 where the stress introduced by the scratch stylus has been added to the residual stress in the coating to give the failure stress. It is clear that the failure stresses for arc evaporated TiN are now higher than for sputtered material since the scratch calibration constant is higher and the residual stress in the arc coatings is also much higher (7.0GPa compressive as opposed to 2.1GPa compressive for the sputtered TiN film). The wedge failure stresses in Figure 9 can then be used to determine the work of adhesion by plotting the calculated OF against the reciprocal of the square root of coating thickness (Figure 10). It is then possible to separate the two components contributing to wedging failure (Equations 5 and 6); the slope of this curve can be used to determine the interfacial fracture energy, Gi, using the coating data in Table 2, whereas the intercept gives a measure of coating fracture strength, Ow. These values are presented in Table 3. It is clear that the fracture strengths of the coatings or scales are quite similar but that there is a much greater difference in the interfacial adhesion. Table 3: Wedge fracture stress and interfacial fracture energy determined from the scratch test. Interfacial Wedge Substrate/coating fracture energy, Gi fracture stress (J/m 2) (GPa) ,,,
SIP TIN/304
538
4.64
Arc TIN/304
2578
4.89
AI203/MA956 1150~
35.0
7.59
1250~
16.9
8.06
It is important to determine if failure is really interfacial if the scratch test is to be used for adhesion assessment, so Auger Electron Spectroscopy (AES) was used to identify the locus of interfacial failure for both materials. In the case of the alumina scales there is always a thin layer of oxide on the uncovered substrate at the bottom of the wedge-spalled pit. However, this may have been formed atter scratch testing due to the exposure of the bare metal substrate to the atmosphere. There is no evidence for substrate material on the underside of any spalled debris that was collected. It is therefore reasonably certain that failure occurs at or very near to the interface once the wedge crack reaches the interfacial region.
Scratching of materials and applications
158
(a) MA956
n (.9 v
91
tO .D
r
1150~
II
1200~
<~
1250~
B-
Q.
O3 "O
,',
"O c-
.
89 -x--
1300~
_
88
"O O
.
"
~
o
x
87
C
~
86
~ = 7.59 + 0 . 0 0 4 3 / t 1/2 R= 0 . 9 8 3
......... G = 8.01 + 0 . 0 0 2 6 / t 1/2 R= 0 . 9 9 8 85 9 o
.....
o'= 8.06 + 0 . 0 0 2 7 / t 1/2 R= 0 . 9 9 7
.....
(~ = 8.28 + 0 . 0 0 2 2 / t 1/2 R= 0 . 9 1 7
84
O
200
220
240
260
280
300
320
340
360
~
I
l i t 1/2 ( m -1/2)
(b) Wedge Spallation in PVD TiN 20
. . . . . . . .
i
. . . .
~ '
. . . . . . .
i
. . . .
!
. . . .
. . . .
.o_ t~ ~
16
I~1
'
~ 9
c_
o3
,
.
Arc evaporated Sputtered TiN
TiN
12
__=
u.
lo
.....................
~,. . . . . . ~
.................................
0 .~ ~,_
250
260
270
280
290
300
310
320
330
1/t 1/2 (m -1/2)
Fig. 10. Variation of critical failure stress, ~F with the reciprocal of the square root of coating thickness, t, for (a) alumina scales on MA956 at different oxidation temperatures, (b) TiN on stainless steel.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
159
Assessing the failure locus of the TiN coatings is more complex since a thin (---100nm) titanium interlayer was used to promote coating adhesion which dissolves a considerable amount of carbon and oxygen from the substrate surface in the early stages of deposition [37]. In the coating processes used ion bombardment of the growing coating is employed to promote adhesion by forming a pseudo-diffusion zone in the interface region giving a metallurgical bond with no well-defined interface plane. The gold TiN was clearly removed at the bottom of wedge spalls but there was still considerable titanium present on the surface of the substrate. SIMS images of the surface showed that the nitrogen content of this surface layer is very low compared to the carbon and oxygen levels. It thus seems likely that failure has occurred within the titanium interlayer. For both materials the interfacial fracture energies are higher than that expected from the fracture energy of the coating (--- 1J/m2) but lower than or comparable to typical substrate values (~- 103j/m2). This also indicates that the failure crack is propagating at or near the interface with at least some crack tip plasticity occurring within the substrate. As Gi increases the effective coating/substrate adhesion increases so the results here indicate that the TiN/stainless adhesion is better than that for alumina/MA956 since the mechanical properties of the substrates, and hence the energy dissipated in crack-tip plasticity processes, are very similar. Since the TiN coated stainless steel has much smaller spalled regions and the coating is considerably more resistant to detachment during abrasion than the alumina scale the relative values of the fracture energies are as expected. S c r a t c h t e s t i n g of M A 9 5 6 50
-
,
,
,
,
i
. . . .
i
~
~
~
'
1
I (,, We E
. . . .
~-'
,
~
9
stresslT
40 8.5
0"~
t
T
C.9
CD
b
30
m .
-
tQ
c_..
U.I
11 \\,"
o
20
9
,"
\
"
i
i
-
8
6o "-"t
6O 6O
.L
I ,
\
"13
i,,,.
In
I! 7.5
0
10 t-
~0
1100
---0--Interfacial
1
Energy
h
1150
1200 Oxidation
1250 Temperature,
1300
1350
(~
Fig. 11. Variation of interfacial fracture energy with wedge fracture strength of alumina scales on MA956 as a function of oxidation temperature.
160
Scratching of materials and applications
The higher value of Gi for the arc evaporated TiN compared to the sputtered material indicates that the adhesion of the arc coatings is actually much better than their sputtered counterparts, a conclusion which is counter to what might be drawn from the scratch test critical loads. The coating flux in arc evaporation is highly ionised compared to sputtering and the bombardment of the substrate with these highly ionised particles in the early stage of deposition is much more effective at cleaning the surface and forming strong chemical bonds [31 ]. The interfacial fracture energy for alumina on MA956 is reduced as the oxidation temperature increases (Figure 11) but the fracture strength of the coating is actually increased. During the long exposures necessary to grow thick scales on the alloy at low temperatures void-like defects are known to grow in the scale. These will act as the crack nucleation sites that lead to failure [38]. If the scale has a constant toughness then the more defective low temperature scales would be expected to fail at a lower stress level. The better adhesion of the low temperature scales is more difficult to explain but may be due to different chemistry of the interfacial regions.
Hard coating on hard substrate For the TiOxNy coating on glass the AFM image does not show a clear sign of adhesion failure though some damage at the side of the track is visible (Figure 12a). The edge on orientation is not perfect - more of the indenter face is involved in forming the left side of the scratch. There are no obvious through-thickness cracks in the AFM images, but the sharp edges delineating the scratch in the SEM image imply that through thickness cracking has occurred (Figure 12b). Energy dispersive x-ray microanalysis in the SEM confirms that coating detachment has occurred. There is a smooth increase in friction force with scratch length (Figure 13a) up to a normal load of 3.2mN when some oscillations in the friction trace are observed. Such oscillations are also visible in the friction coefficient trace (Figure 13b). However, the clearest indication of failure is seen in the work of friction plot (Figure 13c) when jumps are observed associated with each failure event. The work of friction, which is the integral of the friction coefficient up to the given displacement, represents the irreversible work done during scratching. The smooth increase in work of friction is related to the increased work necessary to plastically deform the coating/substrate system to create the scratch track. However, the rapid increase in work of friction correlates with the onset of coating fracture. Both through-thickness and adhesive failure can generate such jumps. For through thickness cracking the crack length is typically of the order of a few microns and the coating thickness is 400nm giving a crack area of---1012m 2. Given typical surface energies of 1J/m 2 the size of a jump in the work of friction trace might be expected to be of the order of l pNm which is much smaller than observed in Figure 13c. Given that the area of adhesion failure is typically two orders of magnitude greater it is likely that these jumps are due to adhesive failure. The adhesion failure is visible in the scanning electron micrograph (Figure 12b) and it appears that this failure occurred in several stages from the jumps in the work of friction trace.
An overview of the potential of quantitative coating adhesion measurement by scratch testing
161
Fig. 12. (a) AFM image ofa 10mN scratch in TiOxNy. (b) SEM image of the same scratch.
Scratching of materials and applications
162 (a) 0
t~
I
1
1
l
1
I
I
-500
Z
-1000
v
o ii
-d
-1500
!
tl:l
t__
-2000 ' __1 -2500
-3000 0
I
I
1
1
1
2
3
4
5
6
7
Lateral Displacement (IJm) (b) 0.5
I
I
1
I
1
I
1
0.4 ill)
o..,m ~
o O t.mo
0.3
off
,8.
0.2 4
.m t__
Id..
0.1
I
1
I
1
I
1
1
1
2
3
4
5
6
7
Lateral Displacement (~m) (Continue to the next page)
8
An overview of the potential of quantitative coating adhesion measurement by scratch testing
163
(c) 1 104
1
9000
-
8000
-
7000
-
r~
6000
-
0
5000
-
4000
-
3000
-
E
1
T
t
Z c~ v
tO o..., ..,,.
0
/
/
//
~
r
1
/
/
2O00 3.5
4
4.5
5
5.5
6
6.5
7
7.5
Lateral deflection (lum) Fig. 13. (Continue from the previous page) Variation of (a) friction force, (b) friction coefficient and (c) work of friction with scratch displacement for 400nm TiOxNy on float glass.
The total work represented in the jumps is 855pNm and the area of delamination is approximately 18~tm2. This would predict an interfacial fracture energy of 47.5J/m 2 which is comparable to results obtained for brittle coatings on brittle substrates by other methods. However, without improvements in the measurement of area this value must be regarded as tentative. The work of friction could, in theory, be used to characterise the failure of hard coatings on soft substrates. However, given the size of the failures and the likely interfacial fracture energies the work of friction resolution of conventional scratch testing equipment is insufficient to resolve the failures.
CONCLUSIONS The scratch test is a good method for quality assurance/quality control testing of the adhesion of hard coatings and is useful in the development of new coatings for process optimisation. However, at present, quantitative adhesion data from the test which might be used as the basis of a performance model is difficult to extract from the test data. In most cases the stresses around the moving indenter are too complicated to be predicted accurately and therefore the
Scratching of materials and applications
164
stresses driving coating failure are not known. However, some failure modes occur sufficiently far away from the indenter where the stress state is simpler and quantification is a possibility. The two main adhesion related failure modes in the scratch testing of hard coatings are wedge spallation and buckling. Buckling occurs for thin coatings which are able to bend in response to applied stresses. The stresses responsible for failure are complex due to the fact that buckling is confined within the region of pile-up close to the indenter. For thicker, stiffer coatings wedge spallation becomes the dominant failure mechanism. This occurs well ahead of the moving indenter and the stresses which are responsible for failure approximate to a state of pure compression. Wedge spallation stresses can, therefore, be quantified by calibration enabling the interfacial fracture energy to be determined. To derive the maximum benefit from the scratch test better theoretical models for the stress fields associated with a moving indenter in a coating/substrate system are needed. These are most likely to be based on finite element analysis, but the modelling approach would need to include a large number of factors if the true stress state is to be predicted accurately enough. The use of good constitutive equations for coating and substrate, the incorporation of a suitable fracture model and a mechanism for handling interfacial and surface roughness will be essential if this is to be achieved. If a model is to be developed which can predict the onset of fracture then a method of representing the defect distribution in the system is also necessary. Furthermore the incorporation of residual stresses into the model is essential for accurate results. A considerable amount of development and validation work is thus required for any new system under investigation. With the emergence of depth sensing indentation and scratch systems with high resolution of applied load, friction force and lateral displacement, it is possible to make more direct measurements of interfacial fracture energy for thin coatings. This approach needs investigating on a wider range of systems and requires a very accurate and critical assessment of the delamination area but offers a more quantitative evaluation method when compared to the traditional scratch test.
ACKNOWLEDGEMENTS The authors would like to thank Ian Gilbert for some SEM images and colleagues at Newcastle University for useful discussions.
REFERENCES
Perry A. J. (1981 ) Thin Solid Films 78, 77. Steinmann, P. A. and Hintermann, H. E. (1985) Jr. Vac. Sci. Technol._ A3, 2394. Valli, J. (1986) J. Vac., Sci. Technol. A4, 3001. Hintermann, H.E. (1984) Wear 100, 381. Perry, A.J. (1983) Thin Solid Films 107, 167. Bull, S.J. and Rickerby, D.S. (1990) Surf Coat. Technol. 42, 149. Benjamin, P., and Weaver, C. (1960) Proc. Roy. Soc. Lond., Set A 254, 163. Bull, S.J. (1991) Surf. Coat. Technol. 50, 25.
An overview of the potential of quantitative coating adhesion measurement by scratch testing 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
33 34 35 36 37 38
39
40
165
Bull,S.J. (1995) Materials at High Temp. 13, 169. Bull, S.J. (1997) Tribology Int. 30, 491. Mittal, K.L. (1978) in Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, pp5-17, K.L. Mittal (Ed.), STP No. 640, ASTM Philadephia. Mittal, K.L. (1976) Electrocomponent Sci. Technol. 3, 21. Mittal, K.L. (1995) in Adhesion Measurement of Films and Coatings, ppl-13, K.L. Mittal (Ed.), VSP, Utrecht, The Netherlands. Venkataraman, S., Kohlstedt D.L., and Gerberich, W.W. (1992) J. Mater. Res. 7, 1126. Kriese, M.D., Boismier, D.A., Moody, N.R. and Gerberich, W.W. (1998) Eng. Fracture Mechanics 61, 1. Kim, K.S. and Aravas, N. (1988) Int. J. Solids Struct. 24, 417. Jeong, H.S. and White, R.C. (1993) J. Vac. Sci. Technol. All, 1373. Bagchi,A., Lucas, G.E., Suo, Z. and Evans, A.G. (1995) J. Mater. Res. 9, 1734. Wei, Y., and Hutchinson, J.W. (1998) Int. J. Fracture 93, 315. Bumett, P.J., and Rickerby, D.S. (1987) Thin Solid Films 154, 403. Arnell, R.D. (1990) Surf Coat. Technol. 43/44, 674. Evans, H.E. (1994) Materials at High Temperature 12, 219. Strawbridge, A., Evans, H.E., and Ponton, C .B. (1997) Mat. Sci. Forum 251/252, 365. Thouless, M.D. (1998) Eng. Fract. Mech. 61, 75. den Toonder, J., Malzbender, J., de With, G., and Balkenende, R. (2002) J. Mater. Res. 17,224. Jiang, X.Y., Lauke, B., and Schueller, T. (2002) Thin Solid Films 414, 63. Holmberg, K., Laukkanen, A., Ronkainen, H., Wallin, K., and Varjus, S. (2003) Wear 254, 278. Tymiak, N.I., Daugela, A., Wyrobek, T.J., and Warren, O.L. (2003) J. Mater. Res. 18, 784. Crepin, J., Bretheau, T., Caldemaison, D., and Ferrer, F. (2000) Acta Materalia 48, 505. Rickerby, D.S. and Newbury, R.B. (1988) Vacuum 38, 161. Mattox, D.M. (1994) in ASM handbook Volume 5: Surface Engineering, pp582-592, ASM International, Materials Park, Ohio, USA. Bull, S.J., (1997) P r o c . 4 th Int. Conference on Advances in Surface Engineering Volume 1, pp274-285 P.K. Datta and J.S. Burnell-Gray (Eds) Royal Society of Chemistry, London. Oliver, W.C., and Pharr, G.M. (1992) J. Mater. Res. 7, 1564. Rickerby, D.S., Jones, A.M., and Bellamy, B.A. (1989) Surf. Coat. Technol. 37, 111. Bull, S.J., Rickerby, D.S., Matthews, A., Leyand, A., Pace, A.R., and Valli J. (1988) Surf. Coat. Technol. 36, 503. Rickerby, D.S., and Bull, S.J. (1989) Surf. Coat. Technol. 39/40, 315. Chalker, P.R., Bull, S.J., Ayres, C.F., and Rickerby, D.S. (1991) Mat. Sci. Eng. A139, 71. Wilber, J.P., Nicholls, J.R., and Bennett, M.J. (1997) in Microscopy of Oxidation 3, pp207-220, S.B. Newcomb and J.A. Little (Eds), Institute of Materials book 675, London. Bull, S.J., Rickerby, D.S., Matthews, A., Pace, A.R., and Leyland, A. (1989) in Plasma Surface Engineering Volume 2, pp1227-1235, E. Broszeit, W D Munz, H Oechsner, K-T Rie and G K Wolf (Eds.), DGM Informationsgesellschatt, Oberursel. Steinmann, P.A., Tardy, Y., and Hintermann, H.E. (1987) Thin Solid Films 154, 333.
166
CHAPTER 8
C H A R A C T E R I Z A T I O N OF M A R / S C R A T C H RESISTANCE OF COATINGS WITH A NANO-INDENTER AND A SCANNING PROBE MICROSCOPE Originally published in TribologyInternationalvol 39. February 2006 WEIDIAN SHEN, LAN MI, and BIN JIANG
Surface Science and Nano-Tribology Laboratory Department of Physics and Astronomy, Eastern Michigan University Ypsilanti M1 48197, USA. E-mail: [email protected]
ABSTRACT Mar/scratch resistance characterization techniques are presented in this paper, in which a Nanoindenter and a Scanning Probe Microscope (SPM) are used to measure the micro mar resistance (MMR), different responses of coatings to the marring stress, and critical forces for rough trough, cracking, delamination, and chipping, quantitatively. It provides a full spectrum of the mar/scratch resistance behavior of the tested samples. To comprehensively evaluate a coating/material in a specified application, introduction of a quantitative index is very useful. The concept of the index and the procedure to calculate it are described. In addition, to meet the variety of coatings/materials properties and the requirements in their applications, some complementary test methods are discussed.
KEYWORDS Wear Resistance Measurement, Mar Resistance Measurement, Scratch Measurement, Nano Indentation and Scratching, and Critical Force for Cracking
Resistance
INTRODUCTION Mar/Scratch resistance is crucial for coatings/materials in many applications, e.g., the polymer topcoats used in the automotive industry [ 1-8]. Mars/marring refer to the light surface damages encountered in the real field that are usually shallow and narrow while scratches/scratching refer to the medium to severe damages. The majority of the damages to the topcoats applied on the automobile bodies belong to the mar category. Scanning Probe Microscope (SPM) was used to examine the mars encountered in the real field. The depth of most mars ranges from a couple of dozen nanometers to several hundred nanometers, while the width ranges from a couple of hundred nanometers up to 2 micrometers. A single mar may not be readily noticeable; however, the existence of a group of such mars does degrade the appearance of coatings. Mar resistance is a measure of a material's ability to resist small-scale mechanical stresses. However, some empirical test methods for mar/scratch resistance used a decade ago were not pertinent and appropriate, such as the crockmeter test, which was fairly popular in the paint industry in the
Characterization of mar/scratch resistance of coatings
167
1980s and early 1990s. In the crockmeter test, a clear topcoat to be tested is applied over a black basecoat on a rigid panel and cured. The panel is immersed in dry Bon Ami cleaning powder and is secured on a test bed. To perform the test, a test probe covered with a fresh green felt pad is moved back and forth over a portion of the panel in ten double strokes so the panel is marred in the area abraded by the probe. The panel then is cleaned in a stream of cold tap water and gently dried with a so~ paper towel. The gloss is measured using a Byk 200 pocket gloss meter by slowly moving the meter across the panel, measuring gloss of both the unmarred and marred sections. The result of the resistance against scratching and marring is reported as percent of gloss retained. SPM was used to examine the surface of a coat that had undergone the crockmeter test and compared it with the surface of the same coat that contained real mars and scratches made in the application field, as shown in Fig. 1. Obviously, the images show quite different configurations, indicating that the surface in the crockmeter test suffered different stresses and damages than actually encountered in the field. Thus, development of new dependable and reliable test methods is necessary. Quite a few groups have made great contributions, using various approaches to address the mar/scratch resistance characterization of polymeric coatings [9-16].
Fig. 1. Comparison of the surface of a coat that has undergone the crockmeter test with the surface of the same coat that contains real mars and scratches made in the application field.
168
Scratching of materials and applications
A technique of using a modified Scanning Probe Microscope (SPM) to study tribological properties of mild-steel with and without lubricant and clutch frictional material in automotive transmission fluid was developed in the early 90s [ 17, 18], as well as to measure the micro mar resistance of surface coatings quantitatively [ 19, 20]. In the commercial SPM, the spring constant of the probe ranges from 0.06-0.58 N/m [21 ]. In the modified SPM, the probe was replaced by a customized probe, in which a diamond shard in micrometer size, picked from a smashed diamond chip, was glued, using epoxy, to a rectangular-shaped cantilever made of a tungsten foil of 12.7, 25.4, or 50.8 ~tm thick. The spring constant of the customized probes is three to four orders of magnitude larger than the commercial one, which is capable of making artificial mars similar to the mars encountered in the real field. Thus, it can be used to measure micro mar resistance (MMR) and different responses of coatings to the marring stress, i.e., elastic recovery, plastic deformation, and abrasive wear, quantitatively [22]. The detailed procedure of the measurement will be described in the following sections. Since the shape of the shard was not predictable and reproducible, the test results were only good for comparison of coatings tested with the same tip and under the same conditions. With the rapid development of nanotribology, the commercial micro diamond tips with well-defined shapes, conical or three or four faced pyramid-shaped, became available in the second half of the 1990s [23]. They were used to replace the shard in the customized probes [24, 25]. While great achievements were made with the modified SPM, designed originally for nondestructive imaging, limitations were encountered. The force generated by deflecting the customized cantilever was limited to a few milli-newtons, and the penetration depth was limited to a couple of micrometers, the z range of the scope, which was not enough for study of some specimens. As the demand for new instruments grew, a new generation of nano indenter/scratcher appeared [26]. A Nano-Indenter XP, made by MTS System Corp., was acquired in the late 90s. The indenter can perform both nanoindentation and nanoscratch tests with a single head, with a normal force up to 650 mN and a penetration depth up to 2 mm. It has greatly enhanced the capability of carrying out tribological measurements in a much larger range. The Nano-indenter can scratch the tested surfaces under a constant, increasing, or incremental load, and the SPM is used to examine the damaged surfaces and analyze the marring/scratching mechanism [27]. Before the test, the samples will be washed in an ultrasonic bath with a mild solvent-free detergent, rinsed in a stream of cool tap water, gently dried with a soft tissue, and then blown dry with high-pressure nitrogen gas to remove any dust and grease on the surface. During the scraping, as the normal force applied to the indenter increases, the mar gets deeper and wider, and a fairly neat mar may become a rough trough; cracks, delamination and chipping may also take place. The damages that the most coatings suffered in the practical applications can be roughly classified into five different modes; mar, rough trough, cracking, delamination, and chipping. Depending on the properties of the coatings and the application conditions, tested coatings may or may not show all the five modes. The combined use of nano-indenter and scanning probe microscope allows measurement of critical force for rough trough, at which point the mar switches to the rough trough, as well as critical forces for cracking, for delamination, and for chipping, if any, with good accuracy. These measured critical forces, along with the micro mar resistance, give a full spectrum of mar/scratch resistance behavior of tested samples.
Characterization of mar~scratch resistance of coatings
169
To further comprehensively evaluate mar/scratch resistance of a material in a specified application, thus directing the development of new high mar/scratch resistant material for the application, introduction of a quantitative index is very useful. The index depends on the properties of the material as well as on the application environment. Intense study of the samples, which have been used in the application environment, investigating the application conditions, and determining their equivalent laboratory test conditions, is a necessary first step. This is followed by quantification of damages made by different scratching modes, by using a Haze-gard plus instrument, which will be described in the section of Quantitative Index. Samples with mars, rough troughs, cracks, delamination, and chipping on the surfaces are evaluated by the instrument for the haze level increases caused by these different damage modes. Then, such a quantitative index can be calculated, either through summation or through integration, which is useful in developing new materials and avoiding a time-consuming field test. Due to the variety of the materials and various requirements in their applications, the micro mar resistance and critical force measurement may not be adequate for characterization of all of them. Some alternate test methods will also be discussed, including crack density measurement, repeated scratching test, and cross-scratching test. The present paper is focused on the introduction of mar/scratch resistance characterization techniques, not on the chemistry of the coatings/materials or the tribological properties of specific samples.
MAR RESISTANCE MEASUREMENT Mar resistance is a complicated issue, and it cannot be characterized with a single quantity. The test results vary with the applied load during the scraping, penetration depth of the tip, and other marring conditions. A Nano-indenter, made by MTS Systems Corp., equipped with a 900 conical-shaped diamond tip with a radius of 1 or 2 ~m at its apex is used to scrape the surface of the sample in the majority of our tests. The indenter can scratch the surface under pre-selected conditions for a distance up to 30 mm at a scraping speed up to 10 mm/s. In performing the scratching, the tip first makes a pre-scan under a light load of about 10 ~tN or less to measure the surface profile along the line to be tested. The surface profile is stored and used to automatically correct subsequent data. During the scratching procedure, lateral motion, applied load, real-time penetration depth, and the frictional force encountered by the tip are recorded. Following the scratching, the tip will make a post-scan to measure the residual depth of the scratch. Curves of applied load, real-time penetration depth, residual depth, and frictional force versus the lateral movement of the tip can be plotted. In the mar resistance measurements, the Indenter will scrape the surface under five to ten preselected different constant loads, before the mar switches to the rough trough, for a distance of 100-150 ~tm at a speed of 10-20 ~trn/s. It produces a group of five to ten parallel mars, made under the different loads, with a pre-selected spacing of 10-15 ~m on the surface. After the marring, the sample will be washed again, but without the detergent, to remove any broken material, then the mars will be examined with the SPM. Figure 2 (a) shows an image of ten mars at the surface of a hard-coated polycarbonate system, a candidate for automobile windows, made under the 10 different constant loads from 4 to 15mN, respectively, and their cross-section profiles. The plot of the profiles is made by the software in the SPM, based on the average
Scratching of materials and applications
170
values over about 400 selected data points along the mar. It allows us to measure the dimensions of the mars with great accuracy, thus calculating the micro mar resistance, MMR, quantitatively. MMR is defined as the normal force applied during the marring divided by the cross-section area of the trough as shown below,
Micro Mar Resistance 1000
800
6oo
System B
r162
, :~
4o0
4b 4b
2OO
4
9
14
N o r m al F o r c e (m N)
(b) Fig. 2. (a) An image of ten mars, made under the 10 different constant loads from 5 to 14mN, respectively, and their cross-section profiles. The micro mar resistance (MMR) is then calculated as MMR = FN / At~o,gh. (b) MMR versus the applied normal forces of two glazing hard-coated polycarbonate systems; System A and System B.
MMR varies with the applied load, i.e. penetration depth, so a group of values obtained under the different loads is needed to present it. Figure 2 (b) is a plot of MMR versus the applied normal forces of two systems. System A is the system used in Fig. 2 (a). MMR of system B was much better than A's under the light normal forces, but it decreased more rapidly with the increasing normal force than A's. MMR of system B was about the same as A's under the large normal forces. The systems A and B are multi-layered polycarbonate based glazing systems. In the preparation of them, an acrylic coating of about 0.5 micron was deposited on top of
Characterization of mar~scratch resistance of coatings
171
polycarbonate first to enhance the adhesion, followed by a siloxane layer of about 6 microns for ultraviolet protection. On top of the siloxane was a plasma-enhanced chemical vapor deposition (PECVD) of inorganic-organic hybrid hard coating containing Si, O, H, and C, with a thickness of about 2-3 microns, which improved the mar/scratch resistance. System A and System B have slight difference in the components of the top coating as well as in the processing procedure. For most of our tested cross-linked polymeric clear topcoats, frequently, MMR, as well as the micro indentation hardness, is large at the very top layer, and decreases as the load increases and the tip begins to penetrate into the surface. This suggests the existence of a hard crust at the top layer of these cross-linked polymer coatings, where the crosslink density may be higher due to the weathering effect. Analyzing the high-resolution images of the mars, the different responses of the coatings to the marring stress could be identified, thus studying the different marring mechanisms was possible. Figure 3 shows two different configurations of mars. Plastic deformation dominates in (a); two big shoulders sit on both sides of the ditch, indicating the material was displaced from the ditch to build these two shoulders during the marring. Abrasive wear, i.e. mass loss, dominates in (b); there are no shoulders and the material dug out from the ditch was broken from the surface and was washed away in the subsequent cleaning.
Fig. 3. Two different mar configurations, (a) plastic deformation dominates, and (b) abrasive wear (mass loss) dominates.
Plastic deformation conserves volume, i.e. if the material of the shoulders can be placed back in the ditch. An interesting test was carried out at the surface of a plastic dominant material, in which 512 vertical mars were made from the left side to the right side of an area of 70 ~m by 70 pm. During the marring, the left shoulder of the second mar filled up the ditch of the first mar, and the ditch of the second mar was made on the top of the fight shoulder of the first mar, and so forth. The "healing" followed the "damaging". When the area marring finished, only the left shoulder of the first mar and the ditch and the right shoulder of the last mar were left at the surface. The remaining area was almost completely restored, as shown in the image of Fig. 4(a) and its profile of 4(b).
172
Scratching of materials and applications
Fig. 4. After 512 vertical mattings from left to right, only the left shoulder of the first mars and the ditch and right shoulder of the last mars remained at the surface. The upper and lower horizontal ditches in the image were made by the turning of the 512 marrings.
Most tested coatings showed a mixture of the responses, as shown in Fig. 5. The total crosssection area of the two shoulders is less than the cross section-area of the ditch. In this case, the area of the two shoulders reflects the plastic deformation, and the difference, considering the compressibility of the tested coatings is no more than 5%, between the total area of two shoulders and the area of the ditch reflects the abrasive wear, i.e. mass loss. Figure 6 is an illustration of how to calculate the micro mar resistance (MMR) and three different responses of coatings to marring stress, based on the dimensions of the mar. The largest inverted triangle represents the cross-section area of the tip that penetrated the surface during the marring, which was calculated based on the real-time penetration depth during the marring and the shape of the tip. The difference between it and the cross-section area of the residual ditch reflects the immediate elastic recovery. The calculation of plastic deformation and abrasive wear was mentioned above. In the MMR calculation, the cross-section area of the ditch was used first to divide the applied force. Later, it was replaced by the cross-section area of the trough, which is the cross-section area of the ditch plus the area between two shoulders, if any, based on the following consideration.
Fig. 5. Mixture of responses, plastic deformation and abrasive wear (mass loss). The crosssection area of the two shoulders is less than the cross-section area of the ditch.
Characterization of mar/scratch resistance of coatings
173
Am:
Cross-section area during scratching Adit" Cross-section area Ashs: Cross-section area At~gh: Cross-section area
of penetration of residual ditch of two shoulders of trough
Micro Mar Resistance: FN/Atrgh
\ / \/
Elastic Recovery: (Arc.- Adit) / Apen* % Plastic Deformation: Ashs/Ape. * % Abrasive Wear:
(Adit- Ashs) / Ap,~ * %
Fig. 6. Illustration of how to calculate the micro mar resistance (MMR) and three different responses of coatings to marring stress.
174
Scratching of materials and applications
Suppose two mars possess the same size of ditch, but one has two shoulders and the other has none. Due to the larger topographic fluctuation of the surface, the damage of the first sample will be more visible. To make the MMR more consistent with the visual judgment and other optical evaluations, the cross-section area of the trough was used to replace the cross-section area of the ditch in the calculation of MMR. A large portion of the tested coatings showed a self-healing to different extents, mainly depending on their glass transition temperature, Tg [28-30]. This is attributed to viscoelastic recovery. Viscoelastic recovery is different from elastic recovery. It results in partial or complete recovery of a marred surface within a time frame from several minutes to several hours [31, 32], while the elastic recovery occurs immediately after the marring tip moves over the surface. A study of viscoelastic recovery using SPM was reported [33], as well as a study of the strain hardening, micro cracking, and surface fatigue [34], using this technique. Although the Nano-indenter equipped with the diamond tips can make artificial mars/scratches with the dimensions similar to those encountered in the field, the scraping speed is much slower than most scrapings in the real field. The speeds used in the tests are limited by the data-taking rate of the Indenter, which is 5 data points per second. To acquire enough data points of the realtime penetration depth, encountered frictional force, etc. along the scratch for analysis, we keep the scraping speed slower than 20 ~m/s in the majority of tests, which is the main limitation of the micro- and nano-scale measurements.
CRITICAL FORCES MEASUREMENT Mar is the light damage, made by the scraping under low normal force. It is usually fairly neat, consisting of a ditch with a smooth bottom and two, if any, well-shaped shoulders on both sides of the ditch. In this stage, the micro mar resistance (MMR) is a reasonable characterization of the coatings' ability to resist small-scale mechanical stresses. Scraping the surface with an increasing normal force, the bottom of the ditch, as well as the ridges of the two shoulders, becomes rough. The neat mar becomes a rough trough. MMR is no longer an appropriate characterization since the cross-section area of the trough, used in the calculation of MMR, begins to change erratically along the rough trough. As the normal force increases further, cracking may occur in the surface of the coatings. Under the continuously increasing normal force, delamination may take place if the penetration depth of the tip reaches the interface and the stress generated by the scraping tip exceeds the adhesion strength. Increasing the force further more may result in the delaminated top layer being chipped off, piece-by-piece, from the surface. Figure 7 shows the five typical distinguishable damage modes, commonly seen in the marring/scratching tests, as well as encountered in the applications [35-38]. Most scratching damages on the surfaces of the coatings used in the real field can be roughly classified into these five modes. Depending on the properties of the coatings, as well as application environments, the coatings may or may not show all the modes. To characterize the coatings' ability to resist the medium to severe damages, critical forces were used [27, 39, 40]. In the present study, the critical force for rough trough, at which a neat mar switches to a rough trough, as well as the critical force for cracking, for delamination and for chipping, if any, were measured, using the Nano-indenter operated under an increasing normal load. As the damage mode switches to the next more severe mode, the real-time penetration depth, as well as the depth of the residual ditch, becomes rougher, and the frictional force encountered becomes more fluctuated, which provide the
Characterization of mar~scratch resistance of coatings
175
Fig. 7. Five distinguishable damage modes, mar, rough trough, crack, delamination and chipping.
evidences of the switching points. The mar/scratch will be further examined by SPM to confirm the switches, and determine the corresponding critical forces. Usually, five to eight mars/scratches were made under a selected increasing force. The average values of the measured critical forces will be used in the results. In the development of a glazing material, the critical force measurements were made on an inorganic-organic hybrid hard coating, containing Si, O, H, and C, produced by plasma enhanced chemical vapor deposition (PECVD) on siloxane/acrylic/polycarbonate composite, which is a potential candidate to replace glass windows in the automobile industry due to its much lighter weight and extremely high impact resistance. The measured values of the critical forces in six sets of scratches were well within a deviation of 3% or less, which, in return, verified the validity of the measurements.
QUANTITATIVE INDEX To comprehensively evaluate mar/scratch resistance of a coating/material in a specified application, introduction of a quantitative index is very useful [41 ]. It consists of three steps: (1) Investigating the application environments and determining the equivalent laboratory test conditions by intensive statistical study of the samples, which have been used in the application environment, (2) Quantification of damages made by different scratching modes, and (3) Calculation of index. To explain the concept of the index and describe the procedure of obtaining it, an illustration is used below. In the development of the above-mentioned coated glazing material for automobile windows, the used windows from the cars and trucks driven in the different US states, as well as abroad, were collected, and samples were cut from different parts of the windows. An optical microscope, equipped with a CCD camera, and a Scanning Probe Microscope were used to examine the damaged surfaces of the samples. For simplicity, all the damages were classified into five categories; mar, rough trough, cracking, delamination, and chipping, as described above. Based
Scratching of materials and applications
176
on the intensive statistical study, the weights of the occurrence of the five distinguishable damages are 75%, 15%, 6%, 3%, and 1%, respectively. Carrying out a laboratory test with the Nano-indenter and SPM on the material of the used windows, as described above, the measured critical forces for rough trough, for cracking, for delamination, and for chipping were 20, 30, 40, and 50 mN, respectively. Thus, the distribution of the forces encountered by the windows in the real field could be equivalent to the laboratory forces plotted in Fig. 8.
0.10 0.08 0.06 0.04
0.02
0.00
0
10
20
t
3O
~
m
40
50
F(mN)
Fig. 8. Force distribution plot based on study of a sample, which has been used in the field with the weights of the occurrence of mar, rough trough, cracking, delamination, and chipping being 75%, 15%, 6%, 3%, and 1%, respectively, and its critical forces for rough trough, cracking, delamination, and chipping being 20, 30, 40, and 50 mN, respectively, in the laboratory test.
The area under the line between F=O to F=20mN is 0.75, i.e., 75%; the area under the line between F=2OmN to F=3OmN is 0.15, i.e., 15%; and so forth. The total area under the line is 1, i.e., 100%, and the force distribution function P is normalized. Here, an assumption was made for simplicity that the forces distribute uniformly between F=0 to F=2OmN, distribute uniformly between F=20mN to F=3OmN, and so forth. If the damages are classified into more and more modes, by measuring the weights of occurrence of each mode and their corresponding critical forces, a smooth force distribution curve can be obtained eventually, l:'(~)
r
F
Fig. 9. Force distribution curve; if we classify the damages into more and more modes, and measure their corresponding critical forces, ideally we will obtain a smooth curve of the force distribution.
Characterization of mar/scratch resistance of coatings
177
such as shown in Fig. 9. The curve is very useful. If it is heavily weighted in the light side, it indicates material in the applications will mostly suffer from light invading forces, having only narrow and shallow mars at its surface, i.e. marring will dominate; it will not be necessary to enhance its resistance against the severe damages. On the other hand, if the force curve is heavily weighted in the heavy side, it indicates material will mostly suffer from severe damages, i.e. cracking, delamination, and chipping; it will make no sense to concentrate on improving its micro mar resistance. For a new sample, carrying out a test in the lab to measure its critical forces for different damages, the weights of each damage mode that will occur on the surface of the sample, if it is used in the real field, can be predicted by using the curve. Suppose the critical forces of the new sample for rough trough, for cracking, for delamination, and for chipping are measured to be F~, F2, F3, and F4, respectively. The area under the curve between 0 and F1 is the occurrence weight of mars, the area under the curve between Fl and F2 is the occurrence weight of rough trough, and so forth, as shown in Fig. 10, which avoids a time-consuming field test. The larger the critical forces, the more light damages the material will experience.
P(F Wmar
Wtto Wcrack
~'1
i
i
F2
F3
~
Wc~,D ~'~ " ~ , . , ~ F4
F
Fig. 10. Prediction of the weights of different damages that will occur in the application for a sample, whose measured critical forces for rough trough, cracking, delamination, and chipping are F~, F2, F3, and F4, respectively, in the laboratory, test.
Quantification of damages depends on the requirements in the applications. For the development of a material used for windows, considering that an object will be clearly sharp if viewed through an undamaged window and will be milky or cloudy if viewed through a damaged window, using the haze level increase measurement to quantify the different damage modes is adequate. On the other hand, if it is the development of new topcoat applied to the automobile bodies, using an optical reflection measurement will be adequate. A Haze-gard plus instrument made by BYK Gardner, as shown in Fig. 11, was used to evaluate the degrading visibility of different damage modes by measuring the haze level increase. In the measurement, the transparent specimen is illuminated at normal incidence, and the transmitted light is measured photo-electrically by an integrating sphere. Haze is caused by wide-angle scattering. According to ASTM D 1003, haze is the percentage of transmitted light that deviates from the incident beam by more than 2.50 on the average. When the total transmittance is measured, the sphere's normal outlet is closed, and when haze is measured, the normal outlet is opened. Increase of haze of a transparent sample reduces the contrast of an object viewed through the transparent sample and results in a milky or cloudy appearance of the object.
Scratching of materials and applications
178
Fig. 11. A Haze-gard plus instrument made by BYK Gardner used to measure the haze level increase, thus evaluating the degrading visibility and quantification of different damage modes.
The measured haze levels of the samples with mars, rough troughs, cracks, delamination, and chippings were 0.04, 1.49, 3.52, 6.98, and 16.6 percents, respectively, as shown in Fig. 12. As the damage mode switched from mar to trough to cracking to delamination and to chipping, the haze level increased dramatically, which indicated the cracking, delamination, and chipping made much more severe damage visible to eye-examination. We used 0.04, 1.49, 3.52, 6.98, and 16.6 as quantitative damage level of mar, rough trough, cracking, delamination, and chipping, respectively.
15-,.-.., ,.,_., N
10
--
-to ~D q) L_
0
mm
--
i
Mar
-
~f
Trough
II
mm
Crack
Oelaminatfon
mm 'f
i i
mm !
Chipping
Fig. 12. The measured haze levels of the mars, rough troughs, cracks, delamination, and chippings were 0.04, 1.49, 3.52, 6.98, and 16.6 percents, respectively.
As an illustration, the quantitative mar/scratch resistance index of the material of the used windows is calculated below. In the five damage category cases, the weights of occurrence of mar, rough trough, cracking, delamination, and chipping are 75%, 15%, 6%, 3%, and 1%, respectively, and their quantitative damage levels are 0.04, 1.49, 3.52, 6.98, and 16.6, respectively. Thus, the index can be calculated through a summation, 5
Q = ~'Wi* i=1
=0.8395
D i = 0.75 x 0 . 0 4 + 0.15 x 1.49 + 0.06 x 3.52 + 0.03 x 6.98 + 0.01 x
16.6
Characterization of mar/scratch resistance of coatings
179
The smaller the index, the better the material performs in mar/scratching resistance behavior. If the damages are classified into more and more modes, the index will be, eventually, calculated through integration, Q = ~P(F)O(F)dF, F=0
where P(F)dF is the weight of the damage caused by a force between F and F + dF, and D(F) is a function of F, representing the quantitative damage level.
SOME COMPLEMENTARY TEST METHODS Due to the variety of the coatings/materials properties and their application requirements, micro mar resistance and critical forces measurement may not be an adequate technique of characterization of mar/scratch resistance behavior for all of them. Continuous development of complementary test methods is needed.
Crack Density Measurement Thermoplastic olefin (TPO) is used more and more as interior and exterior material in automobiles, recently [28, 30]. It is a fragile material, and the surface of it is relatively rough with fluctuations of up to several hundred nanometers. Due to the roughness of the surface, equivalent to the depth of most mars, it is hard to make a neat mar and calculate the micro mar resistance of it. Due to its fragile nature, the surface will be cracked as soon as the tip sticks into the surface and begins to scrape. To characterize its mar/scratch resistance behavior, an alternate method was proposed, in which the Nano-indenter is used to scrape the surface of TPO under a constant normal load of a couple of raN. After the scraping, SPM is used to examine the scratch and measure the number of cracks per unit length. The density of the cracks correlates to the material's toughness against the fracture qualitatively. It also depends on the penetration depth during the scratching and the scraping speed. Figure 13 is a 25 lam by 25 ~m image of a scratch made under 3mN normal force, showing the cracks distributed on both sides of the ditch.
Fig. 13. Image of a scratch made under 3 mN normal force, showing the cracks distributed on the both sides of the ditch.
180
Scratching of materials and applications
More than one dozen TPO samples with various components, prepared with different processing procedures, were examined. The density of cracks, the average value from about a dozen images examined for each sample, varies from about 8 cracks per 25 ~m to about 16 cracks per 25 pm. Usually, the less the crack density, the stronger the toughness against the fracture.
Repeated Scraping Test For the materials that may be subjected to the repeated scraping along the same trench in the applications, this test can provide a good measurement of their mar/scratch resistance behavior in the applications. For example, the glazing material used for automobile side windows, the hard dust particles trapped between the window and the rubber seals may repeatedly mar and scratch the surface in the same trench when the window is rolled up and down. In the repeated scraping test, the Nano-indenter is used to scrape the sample surface along the same trench with a relatively small and constant load for several selected times. After the scrapings, the images of the scratches are taken using the SPM for configuration study, and the depths of the scratches are measured. The test results of two glazing materials, labeled as A and B, potential candidates for windows, are presented as examples. In the test, their surfaces were scraped along the same trench under a constant load of 5mN, repeatedly for 5, 10, 20, 40 and 60 times at a scraping speed of 20 ktm/s. The measured depths of the mars after 5 and 10 scrapings at the surface of A were 60 and 90 nm, respectively; the measured depths of the mars after 5, 10, and 20 scrapings at the surface of B were 100, 125, and 135 nm, respectively, before the damages switched to the rough trough. The depths of mars at the surface of A were shallower than those at the surface of B after the same number of scrapings, which might be an indication that sample A is harder than B at the top layer. It was confirmed by the micro indentation hardness test that A, under a light normal force of 1.5 or 3.0 mN, was harder than B by about 35%. The images of the scratches showed that the damages of sample A belonged to the category of "mar" after 5 or 10 scrapings, became "rough trough" after 20 scrapings, and the chipping took place after 40 or 60 scrapings. In contrast, the damages of B stayed in the category of "mar" beyond 20 scrapings. However, when the chipping occurred after 40 and 60 scrapings, the damages at the surface of B were more severe than those at the surface of A; the chipped spots were deeper and wider. The selected images for illustration are shown in Fig. 14. The results indicate while B is soft, it is more ductile, so it stayed in "mar" longer. On the other side, sample A is hard, but it may be brittle and was vulnerable under the increasing load, and it switched to rough trough earlier. The results are consistent with the results of the micro mar resistance and critical force measurements of A and B. Under the light loads, MMR of sample A is better than B, since B is soft and ductile, and the mars in the surface of B were deeper and wider. However, the critical force of switching from mar to rough trough of sample A was lower than B's, since A is brittle, and its surface is easily broken under the increasing load. Examining the dimensions of the chippings further suggest that the vulnerable layer of sample A, where the adhesion/cohesion was weak and it was easy to be chipped off, might be in the depth of 600-800 nm, while B's might be in the depth of 1.4-1.8 pm. When the chipping took place in both surfaces, B suffered more severe damages than A; see the scratch images after 40 and 60 scrapings in Fig. 14.
Characterization of mar/scratch resistance of coatings
181
Fig. 14. (a) After 20 scrapings, surface A had a rough trough, in the left, while surface B had a mar, in the right. (b) After 40 scrapings and (c) after 60 scrapings, both surfaces were chipped, but the chipped area at the surface B, in the right, is bigger and deeper than A's, in the left.
Cross Scratching Test For the materials that may be subjected to the cross scraping in the applications, this test can provide a good measurement of their mar/scratch resistance behavior in the applications. For example, in the glazing material used for automobile windshields, the overlap area swept by two rubber wipers may be cross-scratched by the hard dust particles trapped between the windshield and the rubber blades. Again, the sample A and B were used as examples to describe the cross-scratching test. The 1000 ~tm long scratches were made with the Nano-indenter under a linearly increasing force from 0 to 30 mN at a scraping speed of 20 lam/s. Two groups of a total of 20 scratches were
Scratching of materials and applications
182
made in a pair of orthogonal directions, which made a matrix with a spacing of 100 ~tm as shown in Fig. 15. Atter the scratching, all the intersections were examined by the SPM, and a mark "x" was used to indicate the observation of the chipping at the intersections.
o
t
~
6
g
}
x
~2
~s
la
2~
Z4
2r
3omN
21
~9
0
Sample A
Sample B
Fig. 15. Two groups of a total of 20 scratches, each of them was l000 pm long made under a linearly increasing force from 0 to 30 mN, in a pair of orthogonal directions, made a matrix with a spacing of 100 pm. A mark "x" was used to indicate the observation of the chipping at the intersections.
There were more chipped intersections in the surface of A than that in the surface of B. The chipping began to take place under the lighter loads in surface A than that in surface B. However, as soon as the chipping began to take place at the intersections of surface B, the damage was more severe than that of surface A. Figure 16 shows two pairs of images. One was taken at the intersection of (12 mN, 15 mN), where the chipping took place only in the surface of A; the other was taken at the intersection of (27 mN, 24 mN), where the chippings occurred in both surfaces, and the chipped spot in the surface of B was bigger and deeper. The results gave further support to the repeated scraping test. Due to the brittleness of the top layer of sample A and ductility of sample B, the intersections in the surface of A began to be chipped first. Under the increasing loads, chipping took place at the intersections of surface B, too. The vulnerable layer of B is deeper, thus the chipped pieces in the surface of B are thicker and bigger than A's. The Taber test is commonly used in the industries. In the test, the abrasion is produced by the contact of a test sample, turning on a vertical axis, with the sliding rotation of two abrading wheels, which results in an abraded ring-shaped area formed by crossed scratches. Most severe damages occur at the intersections of the crossed scratches [42]. The cross-scratching test with the Nano-indenter and SPM could be useful in detailed analyzing of the results of Taber test.
Characterization of mar/scratch resistance of coatings
183
Fig. 16. (a): Intersection of(12mN, 15mN) at surface A began to be chipped (b): Intersection of (12 mN, 15 mN) at surface B showed no chipping. (c) and (d): Intersections of (27 mN, 24 mN) at both surfaces were chipped, but the damage on surface B shown in (d) is more severe than A shown in (c), the chipped area were bigger and deeper.
SUMMARY A Nano-indenter and a Scanning Probe Microscope are used to characterize the mar/scratch resistance of coatings and other materials. The indenter is used to make artificial damages, scratching the surface of the samples in pre-designed ways, e.g., scratching under a constant load or an increasing load, making a single scraping or repeated scrapings, making parallel scratches or cross-scratches, etc., and the SPM is used to examine the damages, make dimensional measurements of the scratches, and analyze the wear mechanisms. Combined use of the two instruments, the micro mar resistance (MMR), the different responses of coatings/materials to the marring stress, i.e., elastic recovery, plastic deformation, and abrasive wear, as well as the critical forces for rough trough, cracking, delamination, and chipping, can be measured quantitatively with good accuracy. It provides a full spectrum of the mar/scratch resistance behavior of the tested samples. To comprehensively evaluate mar/scratch resistance of a coating/material in a specified application, a quantitative index is introduced. The index depends on the properties of the coating/material as well as on its application environment. To obtain the index, investigating the application conditions, thus establishing an equivalent laboratory force
Scratching of materials and applications
184
distribution curve is a necessary first step, followed by quantification of the different damage modes. Then, the index can be formed either through summation or integration. To meet the variety of coatings/materials properties and the requirements in their applications, some complementary test methods, such as the crack density measurement, repeated scratching test, and cross-scratching test are discussed.
ACKNOWLEDGEMENTS This work is supported by the National Science Foundation Industry/University Cooperative Research Center in Coatings at Eastern Michigan University (NSF UU CRCC at EMU) and the Graduate School of EMU. The authors are grateful to Dr. Frank Jones, the long-time collaborator in coatings research, and Dr. Rose Ryntz and Dr. Alex Scholten for their helpful discussions in developing and improving the techniques.
REFERENCES
.
3. 4. 5.
10. 11. 12. 13. 14. 15. 16. 17. 18.
Gregorovich, B.V. and McGonigal, P.J. (1992) Proc. Adv. Coat. Technol. Conf. ASM/ESD, Materials Park OH, 121. Kahl, L., Halpaap, R. and Wamprecht, C. (1993) Surf Coat. Int. / JOCCA 76(10), 394. McGinniss, V.D. (1994) Paint & Coat. Ind. 34. Courter, J.L. (1997) Journal of Coatings Technology, 69(866), 56. Andrei, D.C., Keddie, J.L., Hay, J.N., Yeates, S.G., Briscoe, B.J. and Parsonage, D. (2001) Journal of Coatings Technology 73(912), 65. Schulz, U., Wachtendorf, V., Klimmasch, T. and Alers, P. (2001) Progress in Organic Coatings 42 (1-2), 38. Bertrand-Lambotte, P., Loubet, J.L., Verpy, C. and Pavan, S. (2002) Thin Solid Films 420, 281. Ryntz, R.A. and Britz, D. (2002) Journal of Coatings Technology 74(925), 77. Briscoe, B.J., Pelillo, E. and Sinha, S.K. (1997) The 10th International Conference on Deformation Yield and Fracture of Polymers, Churchill College, Cambridge, UK. Wagner, G. and Osterhold, M. (1999) MATERIALWISSENSCHAFT UND WERKSTOFFTECHNIK 30(10), 617. Lin, L., Blackman, G.S. and Matheson, R.R. (2000) Progress in Organic Coatings 40(14), 85. Bertrand-Lambotte, P., Loubet, J.L., Verpy, C. and Pavan, S. (2001) Thin Solid Films 398, 306. Osterhold, M. and Wagner, G. (2002) Progress in Organic Coatings 45(4), 365. Krupicka, A., Johansson, M. and Hult, A. (2003) Progress in Organic Coatings 46(1), 32. Schlesing, W., Buhk, M. and Osterhold, M. (2004) Progress in Organic Coatings 49(3), 197. Tahmassebi, N. and Moradian, S. (2004) Polymer Degradation and Stability 83(3), 405. Everson, M.P., Gangopadhyay, A.K., Jaklevic, R.C., Scholl, D. and Shen, W. (1994) NANO Advanced Study Institute. Slough, C.G., Everson, M.P., Jaklevic, R.C., Melotik, D.J. and Shen, W. (1996) Tribology Transactions 39(3), 609.
Characterization of mar/scratch resistance of coatings 19. 20. 21. 22. 23. 24.
25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
40. 41. 42.
185
Shen, W., Ji, C., Jones, F., Everson, M.P. and Ryntz, R.A. (1996) Polym. Mater. Sci. Eng. 74, 346. Shen, W., Ji, C., Jones, F., Everson, M.P. and Ryntz, R.A. (1996) Surface Coatings International 79(6), 253. See contact mode Nanotips of Digital Instruments, Veeco Metrology, Group. Shen, W., Smith, S.M., Jones, F.N., Ji, C., Ryntz, R.A. and Everson, M.P. (1997) Journal of Coatings Technology 69(873), 123. For example, Imetra, Inc. in Elmsford NY, it supplies the diamond tips with various shapes and sharpness, whose manufacturer is in Switzerland Shen, W. and Jones, F.N. (1999) In: Microstructure and Microtribology of Polymer Surface: Investigation of Tribological Properties of Polymeric Surface Coatings with Scanning Probe Microscope, pp. 439-454, Tsukruk, V.V. and Wahl, K.J. (Eds) The American Chemical Society. Shen, W.C., Jiang, B. and Jones, F.N. (2000) Journal of Coatings Technology 72(907), 89. For example, Nano-Indenter XP made by MTS, Nano Scratch Tester by CSEM, TriboScope by Hysitron, etc. Shen, W., Sun, J., Liu, Zh., Mao, W., Nordstrom, J.D., Ziemer, P.D. and Jones, F.N. (2004)Journal of Coatings Technology Research 1(2), 117. Ryntz, R.A., Abell, B.D., Pollano, G.M., Nguyen, L.H. and Shen, W. (2000) Journal of Coatings Technology 72(904), 47. Hara, Y., Mori, T. and Fujitani, T. (2000) Progress in Organic Coatings 40(1-4), 39. Yaneff, P.V., Adamsons, K., Ryntz, R.A. and Britz, D. (2002) Journal of Coatings Technology 74(933), 135. Betz, P. and Bartelt, A. (1993) Progress in Organic Coatings 22, 27. Sano, S., Yamada, K. and Ishihara, M. (1994) Toso Kagaku 29, 475. Shen, W., Smith, S.M., Ye, H., Jones, F. and Jacobs, P.B. (1998) Tribology Letters 5(1), 75. Shen, W., Ye, H. and Jones, F. (1998) Tribology Letters 5(2), 197. Bull, S.J., Rickerby, D.S., Matthews, A., Leyland, A., Pace, A.R. and Valli, J. (1988) Surface and Coatings Technology 36, 503. Bull, S.J. (1997) Tribology International 30(7), 491. Ronkainen, H., Kosinen, J., Varjus, S. and Holmberg, K. (1999) Tribology Letters 6, 63. Shen, W., Jiang, B., Gasworth, S.M. and Mukamal, H. (2001) Tribology International 34(2), 135. Lin, L., Blackman, G.S. and Matheson, R.R. (1999) In: Microstructure and Microtribology of Polymer Surface: Micromechanical Characterization of Scratch and Mar Behavior of Automotive Topcoats, pp. 428-438, Tsukruk, V.V. and Wahl, K.J. (Eds) The American Chemical Society. Courter, J.L. and Kamenetzky, E.A. (1999) The 5th Nurnberg Congress: "Creative Advances in Coating Technology ", Nurnberg, Germany. Shen, W., Jiang, B., Scholten, A., Schwenke, R., Mi, L., Seal, C. and Wang, P. (2004) Tribology Letters 17(3), 637. Sun, J., Mukamal, H., Liu, Zh. and Shen, W. (2002) Tribologv Letters 13(1), 49.
186
CHAPTER 9
E V A L U A T I N G T H E C O H E S I V E S T R E N G T H OF A SURFACE M A T E R I A L BY CONTROLLED SCRATCHING
YONGSONG XIE and HOWARD M. HAWTHORNE
Institute for Fuel Cell Innovation, National Research Council Canada,3250 East Mall Vancouver, BC V6T 1W5, Canada. Email: [email protected]
ABSTRACT Controlled scratching is a convenient technique for evaluating the cohesive strength of a coating or surface material. Single or parallel overlapping scratching by a hard spherical indenter under increasing load induces increasing contact deformation (strain) in a controlled manner. Thus, plastic deformation is accumulated incrementally, and micro-fracture is induced, on the specimen surface. Cohesion criteria based on the extent of acoustic emission, micro-fracture and tensile cracking in the scratch groove, as well as the scratch groove area, are described and results are presented that correlate reasonably well with the wear characteristics of a number of different materials.
KEYWORDS Single scratch, parallel overlapping scratch, cohesive strength, microstructural integrity, microfracture, critical plastic strain.
INTRODUCTION The use of a harder material to scratch a softer one dates back at least 7000 years [1] and possibly even to pre-historic times, as evidenced by the examples of rock or cave carvings in many parts of the world. However, it is only much more recently that scratch testing is used systematically to provide some measure of the "strength" of a solid material. Scratch testing has been used for measuring the hardness of a surface, assessing the adhesion between a coating and the substrate, simulating abrasive wear, etc [2]. In addition, scratch testing can be used for evaluating the cohesive strength of a coating or surface material. The term cohesive strength is used here to indicate the microstructural integrity, or resistance to micro-fracture under deformation, of a surface material. A material's cohesive strength depends upon many factors such as, the bonding between different phases in a composite material, inter-splat bonding in thermal sprayed coatings, inter-crystallite boundaries in PVD coatings, and microcracks or other weakening imperfections or defects within different materials. It is this material characteristic that we investigate below via controlled scratching techniques as distinct from the study of coating - substrate adhesion that is the more usual province of scratch testing.
Evaluating the cohesive strength of a surface material by controlled scratching
187
While the resistance of a material to plastic contact deformation is measured by hardness testing and its ability to deform plastically without macro-fracture is assessed by tensile ductility testing, its behaviour under plastic contact deformation with micro-fracture can be evaluated by scratching. A scratch indenter can induce localized, severe plastic deformation within a thick coating or on the surface of a bulk material which leads to micro-fracture. It has been found that, along with abrasion and erosion with large, hard particles, scratch testing is the most promising technique for evaluating the cohesive strength of plasma sprayed coatings and other wear resistant materials [3,2].
THEORETICAL DEVELOPMENT Single Scratching When a hard indenter is drawn across a surface that is relatively soft, or, at least, sufficiently compliant as not to be brittle, if the contact pressure between the indenter and the surface exceeds the latter's elastic limit, some of the surface material is plastically displaced by the indenter and a scratch groove is formed. The plastically displaced material is pushed sideways, and sometimes upwards, along the indenter. When sideways-flowing material forms side ridges on the scratch groove it is called ploughing. When upward-flowing material forms chips that leave the surface this is called cutting. Whether, and how much of, the plastically displaced material becomes cutting chips depends mainly on the indenter geometry, the strain hardening ability of the surface and the interfacial friction. Only sharp indenters with a large attack angle (angle between the leading side of the indenter at the contact edge and the direction of sliding) may cause cutting. Here, we discuss further only the situation of ploughing. In ploughing, the extent of the associated plastic strain is mainly determined by the geometry of the indenter. With increased indenter sharpness, the plastic strain increases as shown in Fig. 1. Directly under the indenter, the localized hydrostatic pressure is very high and inhibits fracture. However, near
Fig. 1. SEM images of the ends of scratches made on solid Co-Cr-Ni-Mo-Fe-W alloy by conical diamond indenters with included angle of 120~ and spherical tip radius of (a) 200 l.tm at normal load of 42 N, and (b) 110 ~tm at normal load of 22 N. Extensive shear steps in (b) indicate that the plastic strain produced by the sharper indenter is more severe although the applied load is smaller.
Scratching of materials and applications
188
the free surface on the ridges of the scratch groove, the hydrostati c pressure decreases to near zero and, thus, fracture occurs easily. It has been found that, during single ploughing and repeated unidirectional ploughing on both ductile materials and many nominally brittle materials, almost all of the debris that is removed from the surface is detached from the side and frontal ridges of the scratch groove [4]. The microstructural mechanisms of the debris formation are mainly growth of voids in metallic materials and coalescence of micro "wing cracks" in ceramic materials [5]. Both mechanisms essentially involve micro-fracture driven by shear. The contact geometry to cause such micro-fracture and the extent of the micro-fracture vary from material to material. There is no complete mechanical solution for calculating the plastic strain induced by a scratch indenter [6]. However, the plastic shear strain associated with the ploughing by a pyramidal indenter can be estimated by an approach based on the upper bound theorem in plasticity. The upper bound model has been described more fully elsewhere [7,8], so only a brief description is given here. When a rigid pyramidal indenter moves across a flat surface with an edge leading, all of the displaced material on the surface appears as ploughed ridges. Figure 2 shows the idealized arrangement of the deformation in the symmetric situation (D is the pyramid apex and CDFE is the central plane of the symmetric deforming zone), which allows us to consider only one side of the indenter. The deformation is deemed to be due to successive shear taking place on a series of subsurface shear planes FBE, FBC, BDC, BDF and BAD. The exact positions of A, B, C, E and F are found by minimizing the rate of energy expended at (a) the shear which takes place at the shear planes, and (b) the friction which takes place at the indenter-specimen interface. Provided that the shear flow stress, k, of the material is given by the plastic stressstrain relation
k=Ko?,"
i
n
d
normal load e n ~ p
(1)
loug~
surface
E
Fig. 2. General arrangement of subsurface tangential velocity discontinuities in the upper bound model.
Evaluating the cohesive strength of a surface material br controlled scratching
189
where y is the shear strain, and Ko and n are material constants, the total plastic shear strain associated with the ploughing, together with the positions of A, B, C, E and F, is determined by minimizing the rate of expended energy. An exact calculation of this total plastic shear strain is complicated since only a numerical solution can be obtained, and the strain hardening exponent of a specimen material, n, needs to be known. However, an approximation can be made as follows. In static indentation with a spherical tipped asperity, the plastic strain is approximately equal to 0.2 a/r, where a is the contact radius and r is the asperity radius [9]. In scratching, the plastic strain is much larger due to material flow around the indenter. Nevertheless, it is reasonable to expect that the strain would still be proportional to a/r. It is known that plastic strains caused by a pyramidal indenter, or a conical indenter that displaces the same volume at the same penetration depth, are close, as are the strains caused by conical and spherical indenters which have the same angle at the edge of contact with the ploughed surface, [ 10]. Hence, a spherical indenter can be converted into a pyramidal indenter to calculate the plastic strain produced when scratching a surface, and vice versa. Using this conversion, the shear strains of two materials with different strain hardening exponents calculated from the upper bound model and the value of a/r, are compared, Fig. 3. This figure shows that the shear strain remains reasonably close to a/r values since the strain hardening exponent of most engineering materials, including metallic and ceramic materials, is between 0.05 and 0.2.
0.6
o 1o 0.4 r E o f f 0.2
,s 0 0
0.2
0.4
0.6
a/r Fig. 3. Comparison between the shear stain calculated from the upper bound approach and from a/r.
Based on this comparison, it is not unreasonable to use a/r as a measurement of the indenter induced plastic strain. Therefore, if an increasing plastic deformation results from scratching a spherical tipped indenter on a specimen surface at progressively increasing normal load, and the induced micro-fracture initiation can be detected, the corresponding value of a/r can be obtained and used for determining the critical plastic strain to initiate micro-fracture. More analyses of the single scratching can be found in [11 ].
Scratching of materials and applications
190
Parallel Overlapping Scratching For materials with good ductility, the plastic strain induced by a single scratch may not be sufficient to cause measurable micro-fracture. In this case, parallel overlapping scratching can be used to accumulate the plastic strain, thereby inducing a significant amount of micro-fracture that can readily be measured. Repeated parallel overlapping scratching has been used by the authors for evaluating the cohesive strength of thick coatings or surface materials. As shown in Fig. 4, five or more parallel scratches are made on the finely polished top surface of a specimen by a spherical tipped indenter with a fixed separation distance between adjacent scratches. At light load the scratch groove is narrow and the extent of plastic deformation (proportional to the ratio of groove width to indenter tip radius as mentioned in the previous section) is small. With increasing normal load along each scratch, the scratch groove becomes wider. This results in an increase in the ratio of groove width to indenter tip radius, but also in the extent of interaction between scratches so that subsequent scratches are made on an already plastically deformed and damaged surface. Hence, although the extent of plastic deformation and damage may be small after the first scratch on the virgin surface, it becomes much more severe after the five parallel scratches. The extent of this surface damage on the specimen surface can then be used as a measure of the surface material's cohesive strength, or resistance to micro-fracture.
Separating distance between adjacent scratches
Sep~ating distance between ad acent scratches
Scratch direction
P
f
_
"
~
n
._Jr
/
Top view Fig. 4. Schematic illustration showing parallel (interacting) scratching at increasing load to induce repeated and increasing plastic deformation.
APPLICATIONS
Single Scratching As mentioned in the previous section, it is reasonable to use a/r as a measurement of the indenter induced plastic shear strain in single scratching. Therefore, if an increasing plastic deformation results from drawing a spherical tipped indenter across a specimen surface at progressively increasing normal load, and the onset of the induced micro-fracture can be detected, the corresponding value of a/r can be obtained and used for determining the critical plastic strain to initiate micro-fracture. Some examples of such application are given below.
Evaluating the cohesive strength of a surface material by controlled scratching
191
A REVETEST scratch tester (CSM Instruments, Switzerland) with a load range from 0 to 200 N was used as the test rig. Two conical diamond indenters with an included angle of 120~ and spherical tip radii of 110 ~m and 200 ~tm, respectively, were used. (These values were measured by a non-contacting, optical 3-D surface imager (WYKO NT-2000, Veeco Instruments, USA) on nominally 100 lam and 200 p.m tips. The difference on the smaller tip may be due to manufacturing tolerance and/or tip wear). Scratch tests were performed with the indenter axis normal to the finely polished (Ra < 0.05 ~m) specimen surface. During a test, the indenter was slid on the specimen surface at a constant speed and the normal load on the indenter was smoothly and linearly increased from initial to final pre-set values. The specimen test surface and the indenter tip were wiped with ethanol prior to each test, which was carried out in an unlubricated condition. An acoustic emission (AE) transducer attached to the indenter holder detected the elastic energy, consisting mainly of cracking and fracturing energy, released from the specimen surface during scratching. The normal load, resulting tangential force and the AE signal counts were recorded continuously by computer during each test. Microscopic examination of the scratches revealed that, at relatively light load, there was no debris produced. After the normal load was increased beyond a critical value, Pc, wear debris caused by micro-fracture began to appear on the scratch sides on all tested ceramic materials and some metallic materials, Fig. 5. With further increase in normal load, progressively more debris appeared. The initiation of the micro-fracture was accompanied by a clearly seen increase in AE signal per unit load on all materials except those with high porosity where a noisy AE signal was recorded. The AE signal increase was more gradual when scratching ceramic specimens but sudden when scratching metallic specimens, Fig. 6. With an increase in the amount of wear debris, the AE signal increased further. The recorded coefficient of friction
Fig. 5. SEM images of the scratches made by 200 p.m radius indenter on (a) A1203/13% TiO2 coating under a normal load of 42 N, and (b) Ni-W-Cr-Si-Fe-B-C coating under a normal load of 110 N. Scratch direction was from top right to bottom left.
Scratching of materials and applications
192
0.4
0.3 .................................................................................... AI203/13% TiO2 coating u I,
0.3~ ~
0.2 it t_
o.2-~ -o
8
u ~o.1 o.,,
o
0.1~
0
Ill
0
, 30
20
, 40
0 50
Load, N
(a) 0.4
0.5
Ni-W-Cr-Si-Fe-B-C coating
1~ ULI
i ~
0.4 "=E
.Co 0.3
"5 o.a~
"6
.~ 0.2 0.2 ~
0.1
o.~
0 80
100 120 Load, N
140
0 160
(b) Fig. 6. AE signal per unit load and coefficient of friction as a function of normal load for a 200 lam radius diamond indenter scratching (a) A1203/13% TiO2 coating, and (b) Ni-W-Cr-Si-Fe-BC coating.
It, (the ratio of tangential force to normal load) increased gradually with the normal load. There was no correspondence between the initiation of the micro-fracture and any increase in friction coefficient, Fig. 6. The increase in AE signal is a good indication of the initiation of microfracture, it was thus used as the main criterion to determine the initiation of micro-fracture in the tests. In such tests, the contact width at the position where micro-fracture initiated was measured by optical microscopy (with an uncertainty < 1 ~m) and then the critical plastic strain to microfracture, ao/r, was determined.
Evaluating the cohesive strength of a surface material by controlled scratching
193
Scratching parameters The single scratch tests were carried out at various sliding speeds and rates of load ramping. It was found that, although the loading speed and sliding speed have little effect on the critical groove width for micro-fracture initiation within the available tester ranges (3 to 35 mm/min. and 32 to 400 N/min., respectively), the measured data were less scattered when using lower sliding and loading speeds. A balance between the sliding speed and the rate of load increase is needed to achieve a suitable load increment per unit scratch length that allows accurate measurement of the critical groove width. Therefore, all scratch tests were carried out at the minimum loading speed of 32 N/min. Sliding speed was chosen to ensure that micro-fracture occurred about 3 to 8 mm from the beginning of the scratch, depending upon specimen size. This resulted in sliding speeds between 3 and 10 mm/min.
Suitability of materials for measurement by the scratch method Various materials, including metallic materials and ceramic materials in both solid and coated forms, were tested. All coatings were relatively thick (> 200 p.m atter grinding and polishing) and, thus, the plastic deformation zone remained entirely within the coating, as estimated based on Johnson's hemi-spherical core model of contact plasticity [ 10]. It was found that, on more brittle materials such as sintered alumina, magnesium oxide and quench-hardened 52100 steel, surface or subsurface macro-cracks developed before the normal load reached the critical load for micro-fracture. Recorded AE signals were too variable to allow detection of micro-fracture initiation. Wear debris adjacent to the scratches sometimes consisted of macro- and microfracture sized material, which also made it difficult to determine when micro-fracture initiated. On the other hand, for materials with good ductility, including most solid metallic materials and many metallic coatings, no micro-fracture took place even when the plastic strain, a/r, reached the maximum value of 0.5 when using the 120~ included angle indenters. For such ductile materials, parallel, overlapping scratching is needed to accumulate sufficient unidirectional plastic strain to eventually produce micro-fracture. It was also found that, for porous materials, such as thermal sprayed alumina with a porosity above 5 %, the AE signal was high and noisy, and micro-fracture initiation could not be detected using this method. In such cases, microfracture initiation was determined solely by microscopic observation after testing.
Effect of tip radius Various ceramic and metallic coatings were scratched by the two indenters with spherical tip radii of 110 p.m and 200 p.m. It was found that, when scratching some metallic coatings using the 110 p.m radius indenter, a few AE signal peaks arose that did not correspond with any micro-fracture on scratch edges. It is believed that these resulted from sub-surface cracking driven by tensile stress, since the sharp indenter is capable of generating and propagating such cracks while scratching coatings containing pre-existing defects. To avoid the interference of such cracking on micro-fracture detection, the 200 p.m radius indenter was used to evaluate metallic coatings. The effect of tip radius on the critical plastic strain to micro-fracture was studied by scratching two plasma sprayed ceramic coatings, A1203 and A1203/13% TiO3, because these have a unique microstructure that has been characterized previously [12] for studies of their damage mechanisms. These coatings contain a network of vertical micro-cracks and many horizontal micro-pores. This microstructure effectively deflects any single dominant macro-crack so that brittle fracture is suppressed, but it cannot withstand large plastic strain and, thus, microfracture readily takes place.
Scratching of materials and applications
194
Table 1 shows the test results. The tests were carried out at a sliding speed of 10 mm/min, and load ramping speed of 32 N/min. These results show that the radius of the spherical tip has little effect on the critical plastic strain to micro-fracture because of the geometrical similarity in the stress-strain field as long as the contact dimension remains very much larger than the microstructure scale (e.g., splat sizes) of the coatings. Scratch testing has demonstrated some success in the following practical uses of the method.
Evaluating the cohesive strength of coatings and comparison with their abrasion resistance An alumina coating was sprayed under different processing conditions from 18 ~m mean particle size, 99.9 % pure mono-crystalline a-alumina, onto a grit-blasted mild carbon steel substrate using an axial plasma spray torch (Axial III System, Northwest Mettech Corp., Richmond, BC, Canada). Micro hardness measurements were made on the top surfaces of the coatings using a Vickers indenter at 3 N load. Abrasive wear resistance of the coatings was evaluated in a pin-on-drum test. In this a 5 mm wide beam specimen was pressed under 12.8 N load against a 200 mm diameter drum (rotating at 0.26 m/s) that was covered with course (200 i.tm) SiC abrasive paper. Details of the effects of different processing conditions on the composition, microstructure and properties of the alumina coatings have already been reported [13,14]. Reproduced in Table 2 is a summary of relevant data on processing conditions, microhardness and abrasive wear results. The ductility of the coatings under model abrasive wear conditions, as represented by the critical plastic strain values derived from the single scratch experiments, are listed in the last column of the table. Table 1: Critical plastic strains to micro-fracture of two ceramic coatings
Material
Indenter Tip Radius r (~tm) 110
A1203 coating 200 110 A1203/13% TiO3 coating
200
Critical Normal Load
Pc (N) 17 16 18 56 58 59 12 14 35 33 36 36
Scratch Width at Critical Load 2ac (~tm) 51 49 52 90 93 96 42 44 75 73 76 75
Critical Plastic Strain
ao/r 0.23 0.22 0.24 0.23 0.23 0.24 0.19 0.20 0.19 0.18 0.19 0.19
Evaluating the cohesive strength of a surface material by controlled scratching
195
Table 2" Processing conditions and test results for thermal sprayed alumina coatings
Spray Run
Vol.% H2
1 2 3 4
10 10 20 20
Coatin$ Preparation Relative Torch Heat Input Nozzle To Plasma Size (mm) Low 12.5 Low 14.1 High 12.5 Hig~h 14.1
Relative Residence Time In Plasma Low High Low High
Coatin $ properties Hardness Abrasive Critical HV3N Wear Plastic (GPa) Rate, Strain, ac/r (g/Nm) 10.7 0.315 0.18 10.4 0.287 0.19 10.7 0.231 0.22 14.5 0.232 _ 0.21
Clearly, the higher critical plastic strain to micro-fracture values of the coatings from spray runs 3 and 4 correlate with their previously measured greater abrasive wear resistance. Microstructural examination revealed that the coatings from spray runs 1 and 2 contained a few incompletely melted alumina particles, and that their porosity and horizontal-to-vertical microcrack ratios were higher, all due to the lower heat input during spraying [14]. These factors resulted in better microstructural integrity of the coatings from spray runs 3 and 4. This work shows that using the relatively simple, but well-controlled, scratch test to evaluate the plastic strain to micro-fracture can provide a similar relative ranking of the cohesive strength of the coatings. Screening candidate materials for wear application Several metallic materials, candidates for potential erosion-resistant applications, have been evaluated by microhardness, controlled scratching and slurry erosive wear tests. The latter test has been described in detail previously [ 15]. In essence, a 5 mm diameter jet of concentrated slurry of crushed alumina (200 ~tm mean size) in tap water was directed at 15 m/s against specimen surfaces under 90~ nominal impingement angle for 1 hour. The specimens and the test results are given in Table 3. Although the high velocity oxyfuel (HVOF) sprayed Co/Cr based alloy is about 25 % harder than the bulk material of identical composition, their slurry erosion wear rates are almost identical since the bulk material has better cohesive strength (as revealed by their critical plastic strain values) than the coating. On the other hand, identical processing of different alloy compositions produced material specimen 3 approximately 50 % harder than specimen 4 but it had much lower cohesive strength (according to the critical plastic strains) and, hence, it was only about 20 ~ more wear resistant. These results indicate that, along with hardness, the critical plastic strain to microfracture is another important material property that can be used as a guide to the relative abrasive and erosive wear resistance of materials.
Scratching of materials and applications
196
Table 3" Hardness, critical plastic strain and wear volume of tested specimens Specimen Number
Specimen Type
Nominal Composition
1
Solid material
Co bal. 26Cr 9Ni 5Mo 3Fe 2W 0.8Mn 0.3Si 0.08N 0.06C Co bal. 26Cr 9Ni 5Mo 3Fe 2W 0.8Mn 0.3Si 0.08N 0.06C
2
3
HVOF sprayed coatin8 Laser clad coating Laser clad coating
HV3N(GPa)
Hardness
Critical Plastic Strain ao/r
Wear Volume (mm 3)
4.1
> 0.5*
18.4
5.2
0.33
18.0 .....
50Cr 39Fe 8B 3Si 10.1 0.29 12.8 32Cr 28Fe 19Ni 9Co 4 4Mo 4B 2.4 Cu 6.8 > 0.5* 15.4 1.4Si *: 0.5 is the maximum plastic strain when using indenter with included angle of 120~
Fig. 7. SEM image of two parallel scratches made on a cobalt-based alloy. The radius of the scratch indenter was 200 lam. The plastic shear trains induced by the two scratches were 0.26 and 0.30 (at the end of the second scratch at which acoustic emission signal jump had been detected), respectively.
Double Parallel Overlapping Scratching A spherical indenter can only induce limited plastic strain without producing detached debris from the surface because, at a high plastic shear strain of a/r greater than about 0.5, some of the plastically displaced specimen material may be pushed upwards along the indenter surface. When single scratching is not sufficient to cause measurable micro-fracture, a subsequent, parallel overlapping scratch can be added to accumulate the plastic strain so that the critical plastic strain can be reached. Figure 7 shows an example of double parallel overlapping scratching. In the scratch testing, a spherical diamond indenter was drawn over the specimen surface. An initial scratch was made on the specimen surface at a constant load under which no micro-fracture was detected. Then, a subsequent scratch was made parallel to the previous one
Evaluating the cohesive strength of a surface material by controlled scratching
197
but displaced from it by a fixed separating distance at increasing load starting from a small value until a critical normal load was reached at which a jump in the AE signal was detected. The cohesive strength of a series of cobalt-based alloys was ranked using this double parallel overlapping scratching technique [ 16].
Multiple Parallel Overlapping Scratching Multiple parallel overlapping scratching has been used by the authors for evaluating the cohesive strength of thick coatings or surface materials. As shown in Fig. 4, five or more parallel scratches are made on the finely polished top surface of a specimen by a spherical tipped indenter with a fixed separation distance between adjacent scratches. At light load the scratch groove is narrow and the extent of plastic deformation is small. With increasing normal load, the scratch groove becomes wider. This results in an increase in the ratio of groove width to indenter tip radius, but also in the extent of interaction between scratches so that subsequent scratches are made on an already plastically deformed and damaged surface. Hence, although the extent of plastic deformation and damage is small after the first scratch on the virgin surface, it becomes much more severe after the five parallel scratches. The extent of this surface damage on the specimen surface can then be used as a measure of its cohesive strength, or resistance to micro-fracture. Since this material property is an important factor in determining the resistance of a material to ploughing wear, as well as evaluating its cohesive strength, parallel overlapping scratching can also be used for assessing the material's abrasive wear resistance. In the multiple parallel overlapping scratching method, cohesive strength in the surface of the scratched specimen is evaluated and compared using the following four parameters: a. Material lost at. specified load. The difference between the volume below and that above the original surface (volumes of scratch groove and scratch side ridges, respectively). b. Acoustic emission (AE) signal at specified load. The AE signal, recorded continuously by computer, relates directly to the elastic energy released during cracking and fracturing in the scratch test. c. Extent of micr0-cracking in brittle ~ains. Figure 8 is an example of such cracking. This depends on both grain strength reduction due to material processing, poor bonding between the hard grains and binder material, etc. It can be estimated qualitatively from SEM examination of scratches at specific load positions. d. Tensile cracks on the bottom of scratch zrooves. Since there is a tensile stress field at the trailing edge of the contact between a scratch indenter and the surface, long cracks perpendicular to the scratch direction may appear if the bonding strength of the scratched material is relatively weak, or, if the pre-existing micro-cracks grow and connect with each other, as shown in Fig. 9. _
198
Scratching of materials and applications
Fig. 9. SEM image of parallel overlapping scratches on a WC-Co-Cr cermet coating made by a spherical indenter with radius of 100 ktm at 24 N. Scratch direction was from right to left.
The material lost from the surface of the scratched specimen consists of material detached from the surface as wear debris, plastic compaction and material accommodated elastically in the far field. The latter is small for most engineering materials. More analyses and some applications of the multiple parallel overlapping scratching can be found in [ 17].
Evaluating the cohesive strength of bond coatings Thermal barrier coating (TBC) systems, consisting of a yttria partially stabilized zirconia top coat and a metallic bond coat deposited onto a superalloy substrate, are ot~en used for protecting hot section components in advanced gas turbine engines. These can withstand higher inlet temperatures than the uncoated alloy and thus improve overall engine performance. The performance of such TBC systems is dictated by thermo-mechanical interactions between the four constituent materials: substrate, bond coating, thermal grown oxide (TGO) formed between the bond coat .and the top coat by depletion of aluminum from the bond coat, and the ceramic
Evaluating the cohesive strength of a surface material by controlled scratching
199
top coat. One of the important factors that govern TBC system performance is the microstructural integrity of the adjacent bond coat and TGO because, during thermal cycling, the imperfections in the vicinity of the TGO localize the misfit stress and induce a strain energy release rate, and cause small cracks and separations to nucleate. Failure occurs when cracks from neighboring imperfections coalesce and detach the layer over a sufficient area to cause large-scale buckling or edge delamination [ 18]. Thermal spray coating is the most widely used method to fabricate TBC systems. However, thermal sprayed coatings exhibit porosity, microcracks, partly melted particles and oxide inclusions, all to various extents. Even the best highvelocity oxyfuel sprayed coatings can have defects such as imperfect inter-splat or interlamellar (inter-pass) bonding and thermal cracks. Because of such defects, and the residual stresses resulting from the spray process, TS coating properties are usually inferior to those of corresponding sintered materials. The extent and type of defects in a TS coating depend upon the spray torch and the precurser powder used, and the spraying condition. Good control of these parameters can ensure a minimum number of detrimental microstructural and compositional defects and the resulting good coating cohesive strength, or microstructural integrity. The cohesive strength of TS coatings depends on all factors that determine how well they "hold together" during use. Although some of these factors are difficult to measure individually, a material's cohesive strength is closely related to its resistance to contact deformation and its ability to deform without fracture. In principle, therefore, by inducing severe contact deformation on a specimen surface and then measuring the extent of damage, the cohesive strength of TS coatings can be evaluated. Parallel overlapping scratching was used to rank the relative cohesive strength of thermal sprayed bond coatings prepared using powders with the same composition but from different powder manufacturers, and using different thermal spray torches and different spray parameters. Twenty thermal sprayed CoNiCrAIY bond coatings were tested. The torches, powders and spray conditions used to prepare coatings are listed in Table 4. The REVETEST scratch tester and a 120~ conical diamond indenter with a 200 l.tm tip radius were used. Scratching speed was 12 mrn/min. Five parallel scratches were made on the finely polished top surface of a specimen with a separation distance of 50 ~m between adjacent scratches, as shown schematically in Fig. 4. In order to have essentially the same extent of plastic deformation on each sample, the width of scratch grooves was controlled to be about 132 ~m at the end of the first scratch for each sample. This scratch width ensured that the plastic deformation was fully constrained within the thickness of all of the bond coatings. Because the hardnesses of the samples were different, the applied maximum normal loads to produce the 132 ~m scratch grooves were different from sample to sample. This maximum load value was predetermined from a separate single scratch on each sample under a ramping load. These maximum load values for each coating are listed in Table 4. Incidentally, the value of scratch hardness, Hs, was also calculated and listed in Table 4. The scratch hardness is defined as the ratio of normal load to projected load-beating area (assumed to be the area of a semi-circle) and was determined from the width of the scratch groove as follows 8P H, =n.b2
(2)
where P is the normal load and b is the scratch width. An acoustic emission (AE) transducer was used for detecting the surface energy released during scratching.
200
Scratching of materials and applications
The cohesive strength in the different coatings was evaluated and compared using the following four scratch damage measures. a. Scratch-induc...ed surface damage by differential image analysis. Examination of SEM images of the multiple scratched traces revealed various amounts of damage on most coatings, mainly in the form of partially detached debris at the scratch sides. There was little or no evidence of debris being totally removed from the surfaces, however. Coating C-II-2, which was prepared using the torch C/powder II/spray parameter 2 combination, showed only smooth, plastically deformed scratch traces with no obvious evidence of micro-fracture at their sides. Ranking the relative extent of the microfracture damage on the other coatings was obtained by a differential image analysis procedure. The SEM images of the multiple scratched traces were subjected to image analysis using Optimas software. The image of the C-II-2 coating scratches was subtracted from that of the corresponding image of each of the other coatings and the area fraction of the remaining surface damage features calculated to give a scratch damage severity index that excluded any plastic deformation contributions. For clarity, this differential image analysis procedure is illustrated in Fig. 10. b. Percentage of lost material, LM. This is given by L M = V~ - V 2 x 1 0 0 %
z,
(3)
where I:1 and V2 are, respectively, the volumes below, and above, the original surface (equal to the appropriate cross-sectional areas multiplied by the scratch length). The volumes were measured using the non-contacting optical 3-D surface imager at a position near the scratch end on each sample aider the scratched surface was cleaned of loose debris by a compressed air jet. c. Acoustic emission signal in t.he last (5 th) scratch. In the test, the AE signal (in arbitrary units) was recorded continuously by computer. Besides the extent of cracking/fracturing, the magnitude of the AE signal is also determined by the yield strength of the surface, the density of the sample, initial residual stress, etc. The harder and denser a material, or, the higher the initial residual stress, the higher the AE signal. Hence, in this case, the AE signal was normalized by
AE = i=1
(4) n
where Pi is the normal load when recording the AE signal AEi, and n is the total number of data points. Nevertheless, the A E can only be used for ranking samples with closely similar hardness (proportional to yield strength) and LM. When a coating had a relatively high porosity (> -- 4%), the detected AE signal was very low. In this case, the AE data was not relevant to micro-fracture.
Evaluating the cohesive strength of a su(ace material by controlled scratching
201
Fig. 10. Figure illustrates the differential image analysis procedure for ranking the relative extent of microfracture damage in different coatings after scratching. Top lett is the SEM image of a coating with significant damage; top right is that of the least damaged coating, C-II-2; bottom image shows the remaining surface damage features after subtracting the C-II-2 coating image from that of the significantly damaged coating.
All of the measured scratch damage parameter values are listed in Table 4. Figure 11 shows the SEM images of the high load ends of the parallel overlapping scratches made on five of the thermal sprayed CoNiCrA1Y coatings which were the best-performing ones from each torch/powder set combination. It was found that the ranking from the scratch testing showed an excellent correlation with that from other characterizations. The correlation between the measured "lost material" and porosity (obtained separately) in Fig. 12, shows that the lost material is strongly related to the coating porosity, regardless of the methods by which the coatings were prepared. Material volume could be lost in three ways: (1) detached from the surface; (2) accommodated by the elastic strain field in the relatively far distance which can be approximately related to the ratio of hardness to Young's modulus of the surface; and (3) plastic compaction. Since the portion of detached material was found to be negligibly small for all the coatings and the coatings should all have similar hardness to Young's modulus ratio because they are the same kind of material, the difference in LM between these coatings was mainly due to differences in plastic compaction, which was mainly determined by the porosity of the material. This indicates that the scratch induced plastic compaction can be used as a measure of porosity of these samples. The higher the plastic compaction, the more porous the sample is. Microstructural examinations of polished cross-sections of these bond coatings revealed that the C-II coatings, which were the best-performing ones, resulted from fully melted powders, the DIII, A-I and E-I coatings contained partially molten and/or non-molten particles, and the B-II
Scratching of materials and applications
202
coatings which were the worst-performing ones contained partially molten and non-molten particles and oxide inclusions. Table 4: Torches, powders and spray conditions used to prepare coatings, and all scratch damage parameter values
AE Surface (arbitrary LM Maximum Hs Damage Index units) Load (GPa) (% feature (%) (N) area) 2.76 1 40 5.90 13.7 0.67 2.18 2 41 5.91 10.1 0.66 6.33 A I 3 37 5.44 17.0 0.61 5.27 4 45 6.51 8.3 0.65 2.64 6 37 5.44 10 0.75 1 24 3.34 14.8 0.83 0.05 B H 8 21 3.04 14.1 0.82 0.04 5 25 3.59 13.6 0.77 0.17 1 44 6.42 0.7 0.47 4.89 C 2 33 4.79 0 (ref) 0.31 0.82 H 3 37 5.38 1.1 0.43 5.93 1 26 3.86 6.1 0.76 1.42 2 39 5.70 8.1 0.56 0.96 6 27 3.90 8.7 0.79 2.13 D III 7 37 5.46 12.0 0.55 2.20 8 33 4.79 5.4 0.64 1.08 3 37 5.35 4.4 0.51 1.79 9 24 3.46 7.4 0.88 0.11 1 27 3.85 8.9 0.84 0.06 E I 3 24 3.56 8.9 0.84 0.06 Note: Bold numbers indicate the data of the best-performing coatings from each torch/powder set combination. Torch
Powder
Spray Parameter
.
.
.
.
Evaluating the cohesive strength of a surface material by controlled scratching
203
Fig. 11. SEM images of the high load ends of the parallel overlapping scratches made on five thermal sprayed CoNiCrA1Y coatings. In each micrograph the top scratch was the last one made.
Scratching of materials and applications
204 90 0
0
0 0
80
9
70
~X
o~ 60 ._1
50 40 30 20 0
2
4
6
8
10
Porosity (%) Fig. 12. Correlation between the measured "lost material" and porosity obtained from separate measurements. The symbols represent different spray and powder combinations.
Evaluating the wear resistance of thermal sprayed tungsten carbide-based coatings Thermal sprayed (TS) tungsten carbide (WC)-based coatings are widely used for highly abrasive, or erosive, wear-resistant components. However, such coatings can exhibit porosity, micro-cracks, partly melted particles, oxide inclusions and carbide degradation. It has been found that compositional and microstructural variables, such as phase content and carbide particle size largely determine the sliding, abrasive and erosive wear behaviour of WC cermet coatings, but defects and residual stresses also have an influence [ 19]. Both high-stress abrasion of WC coatings [ 19], and high-energy impact erosion of WC/Co/Cr coatings with dry sand [20] fractures the carbide phase. Defect-related damage mechanisms were also found important in both dry alumina particle and slurry erosion behaviour of various HVOF cermet coatings [21 ]. During erosion or abrasion, surfaces sustain repeated interactions with a hard abrasive or erodent medium, such that plastic deformation is accumulated locally, resulting in many microfracture events on the worn specimen surface. Scratch testing has been used to induce repeated plastic deformation in a controlled manner on surfaces and this was shown to approximate closely the contact deformation during abrasion and solid particle erosion [7,4]. The multiple parallel overlapping scratching should, therefore, be capable of evaluating the cohesive strength that relates to the wear resistance of a material. Two commercially available tungsten carbide-12 weight% cobalt (WC-12Co) powders (M5 and C1) were used as the starting materials for preparing the samples. Coatings were deposited on low-carbon steel substrates by thermal spraying using a HVOF system (JP5000, Tafa-Praxair Inc., Concord, NH, USA). A wide selection of fuel (kerosene) and oxygen flow settings was employed to produce a range of in-flight particle characteristics (temperature and velocity) and, hence, coatings exhibiting differences in their wear properties. Coatings were deposited at two stand-off distances, 20 cm and 38 cm. A total of eighteen different powder-spray parameter combinations were employed. The details of the coating preparation were provided in ref. [22]. Single pass, depth-sensing scratching at relatively small load was carried out first to estimate how much material would be "lost" due to plastic compaction. A micro scratch tester (CSM Instruments, Switzerland) was used for this initial scratching. A 200 ~tm radius, spherical-tipped
Evaluating the cohesive strength of a surface material by controlled scratching
205
diamond indenter was drawn over the testing surface at a constant normal load of 15 N. The indenter axis was normal to the top surface of the finely polished coating. The indenter geometry and the load ensured that no material was removed from the surfaces and no obvious cracking occurred. The scratch speed was 4 mm/min and the scratch distance was 6 mm. The percentage of lost material caused by plastic compaction, LM, in the scratching was measured by the non-contacting, optical 3-D surface imager and calculated using the Equation 3 given in a previous section. Incidentally, the average value of scratch hardness, Hs was also calculated using Equation 2 given previously. Here the scratch width b was derived from the penetration depth under load, the residual depth after the indenter had been removed and the indenter radius, based on contact mechanics calculation. Both the penetration depth under load and the residual depth after the indenter had been removed were continuously recorded by computer during the scratching. The REVETEST scratch tester was then used with a 200/am radius, spherical-tipped diamond indenter to make five overlapping, parallel scratches, 50 ~tm apart, on the top surface of the finely polished specimen surfaces. Normal load was ramped from 0 to 100 N over 60 seconds in each 8 mm long scratch. This induced greater coating damage, as plastic deformation accumulated with each overlapping scratch and with increasing load. In this case, the wear resistance of different coatings was evaluated and compared using the following four parameters: a. The amount of material removed as detached debris RM, from the surface by this multiple scratching which is given by R M = ( V~. - V2p)(l - L M )
(5)
where Vlp and V2p are, respectively, material volumes below and above the original surface per unit scratch length in this parallel scratching. The same as the measurement for LM, RM was determined using the optical surface imager at a position near the scratch end on each sample after the scratched surface was cleaned of loose debris by a compressed air jet. b. Total acoustic emission count in the last (5 th) scratch. c. Tensile cracks on the bottom of scratch grooves. d. Extent of surface damage from examination o.f SEM images of the scratches. The multiple scratched area (end) of each coating specimen was examined by SEM and the severity of damage was estimated visually. All of the measured scratch damage parameter values, along with the scratch hardnesses, are listed in Table 5. Figure 13 shows the scratched surfaces of the best-performing coatings and one of the poorer performing coatings from each powder. The performance of the various coatings when subjected to scratching was compared with their three-body abrasive wear behaviour. The performance of the coatings in three-body abrasive wear was evaluated using the standard dry sand rubber wheel abrasion test that involved feeding silica sand between a rotating rubber-coated wheel turning in contact with a WC-12Co coated sample. The test was run over an equivalent lineal distance of 4309 m. The amount of coating removed (volume of the wear scar) during the test was determined using optical profilometry. A comparison of the performance of the various coatings when subjected to three-body dry abrasion and to scratch tests is presented in Fig. 14. In general, the agreement in identifying the
Scratching of materials and applications
206
best-performing coatings using these two tests is very good. The overall best-performing coating was found to be the same in the two tests. It is also interesting to look at the results for the different groups of coatings. For powder 1 (M5), there is agreement for the two techniques in identifying the two best-performing coatings produced at a spray distance of 38 cm and the two best coatings sprayed at a distance of 20 cm. Similarly, for powder 2, there was agreement between the two techniques regarding the best-performing coating at a spray distance of 38 cm and the one at 20 cm. Another approach for comparing the results is to set a performance level as a criterion for accepting or rejecting a coating. For example, in Fig. 14, lines have been drawn to indicate the point where the material removal by abrasion or scratching is 1.3 times that exhibited by the best-performing coating (C1-38-6). Using this arbitrary performance criterion, it can be seen that the scratch test would have identified five coatings as falling within the acceptance window (i.e., the five coatings with RM values falling below the solid line). The abrasion test would have identified these same five coatings, as well as two additional coatings, as being acceptable (i.e., the seven coatings for which the volume loss in abrasion falls below the dashed line). This comparison demonstrates that there is good agreement between the results obtained in the standard three-body dry abrasion tests and in controlled scratch testing. Not surprisingly, the relationship is not perfect and the absolute ranking of the performance of the samples as determined using the two techniques does not always agree; however, the general trends correlate and both approaches identify the best-performing coatings.
Table 5: Measured scratch damage parameter values
Hs (GPa)
Sample .
.
.
LM (%) .
.
.
RM (p.m3/lam) .
.
.
.
AE (arbitrary unit)
.
M5-38-2 8.19 14 2545 M5-38-17 8.69 40 4125 M5-38-18 10.68 -~0 772 M5-38-20 10.92 --- 0 681 M5-38-21 8.33 18 1242 M5-20-5 10.57 - 0 642 M5-20-8 10.42 --- 0 726 M5-20-9 10.60 11 1069 M5-20-10 9.94 3 1044 C1-38-21 7.97 15 3557 C1-38-20 9.92 --- 0 1042 C1-38-16 9.06 ~- 0 1145 C1-38-15 9.01 --- 0 1613 C1-38-14 8.52 -- 0 1202 C1-38-6 9.89 --- 0 682 C1-20-1 10.04 ~- 0 1493 C1-20-5 11.48 --- 0 1056 C 1-20-10 9.28 --- 0 2095 * 1 is the least, and 5 is the most damage.
1693 9740 1352 1271 1736 770 1180 1740 1248 1668 712 660 1429 1077 839 645 647 693
Tensile Cracks
Extent of Surface ..... Damage 9 Many 4 Very many 5 None 2 None 2 Many 3 None 1 None 2 Some 3 Some 3 Very many 5 None 2 None 2 Some 3 None 2 None 1 None 3 None 3 A few 3
Evaluating the cohesive strength of a surface material by controlled scratching
207
Fig. 13. SEM images of coatings showing the scratched surfaces of the best-performing coatings and one of the poorer performing coatings from each powder. The micrographs were taken at a load of either 100 N or 50 N. In each micrograph the top scratch was the last one made. The direction of scratching was from right to left.
208
Scratching of materials and applications
Fig. 14. Performance of coatings in abrasion and scratch tests. Sample designation refers to the powder-spray distance (in cm)-trial number, where M5 is powder 1 and C1 is powder 2. The best-performing coatings in each group are identified with circles. The thick horizontal lines designate an arbitrary acceptance criterion based on a performance level of 1.3 times that of the best-performing coating: the dashed line represents the cut-off point for the abrasion performance and the solid line that for the scratch test performance.
CONCLUSIONS Controlled scratching, under conditions that favour ploughing and ductile micro-fracture, of specimen surfaces is a convenient and promising method of assessing the relative cohesive strength of thick coatings or surface materials. The critical plastic strain to micro-fracture is a good measure of the ductility of the coatings/surface materials. The scratch induced volume loss due to plastic compaction has a good correlation with the porosity of the thermal sprayed coating samples. The wear area measured on the scratched surface is a good determinant of the overall cohesive strength of such materials. Tensile cracking and cracking in hard phases in the scratch groove are indicators of poor cohesion. A sudden jump in the AE signal when scratching a surface indicates that it is likely to fracture under severe contact deformation conditions, for example, in components subject to coarse abrasion. The findings presented in this work comparing results from controlled scratch testing and abrasion and erosion testing indicate the value of controlled scratch testing as a tool in evaluating the wear characteristics of thick coatings or surface materials. The scratch test technique described provides a relatively easily applied experimental method that is suitable for quick screening tests in industrial applications and is complementary to wear tests.
Evaluating the cohesive strength of a surface material by controlled scratching
209
REFERENCES
.
~
5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
16. 17. 18. 19. 20. 21. 22.
Dowson, D. (1979). History ofTribology. Longman Group, London. Hawthorne, H.M. and Xie, Y. (1999). Proceedings of the 2"a International Conference on Surface Engineering Wuhan '99, Wuhan, China, pp. 18-24. Erickson, L.C., Westerg~rd, R., Wiklund, U., Axrn, N., Hawthorne, H.M. and Hogmark, S. (1998) Wear 214, 30. Xie, Y. and Williams, J.A. (1993) Wear 162-164, 864. Lawn, B.R. (1998) Journal of The American Ceramic Society 81, 1977. Kapoor, A., Johnson, K. L. and Williams, J. A. (1996) Wear 200, 38. Williams, J.A. and Xie, Y. (1992) Wear 155, 363. Xie, Y. (1994). Ph.D. Thesis, University of Cambridge, UK. Tabor, D. (1951 ). Hardness of metals. Clarendon Press, Oxford. Johnson, K.L. (1985). Contact mechanics. Cambridge University Press, Cambridge. Xie, Y. and Hawthorne, H.M. (2000) Wear 240, 65. Xie, Y. and Hawthorne, H.M. (1999) Wear 233-235, 293. Erickson, L. C., Troczynski, T., Ross, D., Tai, H. and Hawthorne, H. M. (1997). Abstracts of papers, the 1 st Worm Tribology Congress, London, pp. 146. Erickson, L. C., Troczynski, T., Hawthorne, H. M., Tai, H. and Ross, D. (1998). Proceedings of the 1 5 th International Thermal Spray Conference, Nice, pp 791-796. Arsenault, B., Legoux, J.G. and Hawthorne, H.M. (1997). In: Thermal Spray: A United Forum for Scientific and Technological Advances, pp. 97-106, Bemdt, C.C. (Ed). ASM, Materials Park, Ohio. Yao, M.X., Wu, J.B.C. and Xie, Y. (2005) Materials Science and Engineering: A 407, 234. Hawthorne, H.M. and Xie, Y. (2001) Meccanica 36, 675. Mumm, D.R., Evans, A.G. and Spitsberg, I.T. (2001) Acta Materialia 49, 2329. Chen, H. and Hutchings, I.M. (1998) Surface and Coatings Technology 107, 106. Wood, R.J.K., Mellor, B.G. and Binfield, M.L. (1997) Wear 211, 70. Hawthome, H.M., Arsenault, B., Immarigeon, J.P., Legoux, J.G. and Parameswaran, V.R. (1999) Wear 225-229, 825. Marple, B.R., Hawthorne, H.M. and Xie, Y. (2004). Proceedings of 2004 International Thermal Spray Conference and Exposition (ITSC 2004), Osaka, Japan.
210 CHAPTER 10
MECHANICAL CHARACTERIZATION OF NANOSTRUCTURED TIB2 COATINGS USING MICROSCRATCH TECHNIQUES Originally published in Tribologyhlternationalvol 39. February 2006 N. PANICH !'2 and Y. S U N 3
1School of Materials Science and Engineering, Nanyang Technological University, Singapore. 2Metallurgy and Materials Science Research Institute, Chulalongkorn University, Thailand 10330. 3School of Engineering & Technology, De Montfort University, Leicester LE1 9BH, UK. Email: [email protected] ABSTRACT TiB2-based nanostructured coatings were fabricated on high-speed steel by magnetron sputtering technique. Mechanical characterization of the resultant coating-substrate systems, such as coating adhesion, friction and scratch resistance, was conducted by microscratch technique. The multi-pass scratch with linearly-increasing load mode of microscratch test was studied to determine the most effective and informative testing conditions and to determine the critical load (Lc) for coating failure. The mode of failure was examined by high resolution SEM and AFM. In order to gain a better understanding of the scratch behaviour during the test, a three-dimensional finite element (FE) model was developed to simulate the scratch process. The developed FE model was able to demonstrate the elastic and plastic behavior of the coating and substrate around the contact area during scratch test. Good agreement has been observed between the FE analysis results and experimental investigations.
KEYWORDS Microscratch, titanium diboride coating, adhesion, finite element analysis, critical load.
INTRODUCTION Titanium diboride, TiB2, is well known as a ceramic material with a hexagonal structure which presents various attractive properties, such as high hardness, excellent corrosion, thermal oxidation and wear resistance [1-2]. TiB2 is easy to be deposited by various approaches; one of the most commonly used is magnetron sputtering due to the low cost of operation. Although many attempts have been made to utilize this material as a protective coating, its real and commercial applications have been very limited owing to the difficulties in producing TiB2 coatings with good mechanical integrity [3]. The major problem in producing high quality TiB2 coatings is that the adhesion of TiB2 coating is poor for the coating-substrate system. Actually, the adhesion of the coating is the most critical
Mechanical characterization of nanostructured TiB 2 coatings
211
aspect of the coating-substrate system particularly for tribological applications. Recently, Berger et al. [4] developed a method to fabricate TiB2 coatings by d.c. magnetron sputtering by applying a positive substrate bias which could improve the adhesive strength without deteriorating coating hardness, while this could not be done by using negative substrate bias. On the other hand, Panich and Sun [5] addressed that the deposition process can be controlled to produce a TiB2 coating with both high hardness and good adhesion strength. This is achieved by introducing substrate sputter-cleaning and then biasing (rf negative) for the early stage of deposition, followed by deposition without biasing. In order to determine and analyze the adhesion strength of a coating or thin film to a substrate, scratch test is widely used by research laboratories as well as the industry. In the present work, attempts were made to enhance the adhesion of TiB2-based nanostructured coatings onto highspeed steel substrates with high coating hardness, by applying a small substrate bias and controlling the sputter-cleaning time. This paper discusses further the effect of sputter-cleaning time, with respect to hardness and scratch resistance of the resultant coatings. In order to gain a better understanding of the scratch behaviour during the test, attempts have also been made to simulate the scratch process of the coating-substrate system using the finite element (FE) method. The plastic deformation behaviour indented by a rigid conical indenter has been analyzed to study the elastic-plastic behaviour and the stress-strain of the coating during the scratch process.
EXPERIMENTAL DETAILS High-speed steel (SECO WKE45, Sweden) in fully hardened and tempered condition was chosen as the substrate in this study. HSS was cut into 12 m m • 12 mm • 3 mm pieces. The specimen's surface was prepared by grinding and polishing and was then ultrasonically cleaned for 10 minutes before charging into the deposition chamber. High-purity argon gas was introduced into the deposited chamber after it was evacuated to below 5• 10-4 Pa. The Ti target was powered in the direct current (dc) mode and TiB2 target was powered in the radio frequency (rf) mode. In order to clean up the targets, the targets were pre-sputtered for 10 minutes with the target shutters closed. The working table where specimens were placed was rotating at 6 rpm during the process. The substrate to target distance was held constant at 10 cm for dc target (Ti) and at 5 cm for rf target (TiB2). All the experiments were conducted at a constant working pressure of 0.65 Pa and at a total gas flow rate (Ar) of 20 sccm. The substrate temperature was 400 ~ for all depositions. A rf power biased to the substrate was used before deposition in order to sputter-clean the substrate surface by using a power of 150 W and a rf power of 30 W was also provided to the substrate as bias during the first hour of deposition. The remaining two hours of deposition was conducted without substrate bias. A thin pure Ti interlayer (about 50 nm) was deposited first, by sputtering the Ti target for 10 minutes with a dc power of 200 W. This was followed by sputtering of the TiB2 target for 3 hours with a rf power of 200 W. Process variation was studied with reference to the effect of sputter-cleaning of substrate before deposition as summarized in Table 1. The thickness of the deposited coatings was examined by ball crater, which summarized in Table 1. Nanoindentation test was performed using the NanoTest TM (Micro Materials Limited, UK), with a Berkovich diamond indenter. For the purpose of statistics and reliability, five to ten loading/unloading curves were made in each experiment to find the average results. In the present work, all experiments were performed at a constant loading and unloading rate of 0.1 mN/s and to a maximum depth of 50 nm (controlled depth mode). The hold time at maximum
212
Scratching of materials and applications
load was 30 s, and thermal drift correction was set at 60 s of holding period at 80% unloading. The unloading curves were used to derive the hardness and modulus values by the analytical technique developed by Oliver and Pharr [6]. The microscratch test was performed using the multi-pass microscratch mode available in the NanoTest TM device with a conical diamond indenter (Rockwell type, 120 ~ diamond cone) with spherical tip-end o f 25 ~tm in radius. A new test method was employed in this work, as described by Xia et. al. [7] and detailed below with reference to Fig. 1.
9erties of studied samples Table 1" Summary o f deposition conditions and Coating Hardness Reduced Yieldload Materials Sputter-cleaning Substrate Thickness ( G P a ) modulus Lu (raN) bias (W) of substrate (rim) (GPa) 98.5 700 28.4 307.2 Sample 1 No sputter-cleaning No bias 216.9 650 19.2 235.5 Sample 2 No sputter-cleaning 30 W for the first hour only 650 20.8 242.5 276.4 Sample 3 rf 150 W 30 min 30 W for the first hour only 298.3 600 23.6 275.1 Sample 4 rfl50 W 60 min 30 W for the first hour only 550 26.4 283.2 345.8 Sample 5 rf 1'50W 90 rain 30 W for the first hour only
S: before scratch I
Critical load Lc (raN) 353.7 529.1 595.4 648.5 875.5
sS
liDs: during scratch
Lc sS s S
| [AS: after scratch E E ~
z ~
.
~
-
A
B
s S
BS
Y
S S
s
S
initial
load
Sliding Distance (B1nl Fig. 1. Schematic of the multi-pass microscratch test, showing the before-scratch (BS) profile, the during-scratch (DS) profile, the linearly increasing load during scratch and the after-scratch (AS) profile.
For each test, a set of surface profiles along the track was measured. Firstly, the initial track profile before scratch or surface topography (see BS profile in Fig. 1) was measured by scanning across the full length of distance to be scratched with a small load of 0.25 mN. Then, the scanned length was scratched by applying a linearly increasing load at 5 mN/s after
Mechanical characterization of nanostructured TiB 2 coatings
213
prescanning the initial 300 ~tm distance under a small initial load of 0.25 mN (from A to B in Fig. 1). During scratching, the friction force on the indenter and the surface profile along the full length of the scratched track were measured continually, such that a friction force versus scratching distance (or load) curve and a during-scratch profile (DS profile in Fig. 1) were obtained. The critical load for coating failure (Lc) was determined by the sudden change in friction force, which also led to a sudden change in the DS profile as schematically shown in Fig. 1. After the scratch test, a third profile was measured along the full length to obtain an after-scratch profile of the track (AS profile in Fig. 1). The difference between BS and DS profiles indicates the total scratch depth including elastic and plastic deformation and surface damage during the scratch process. The difference between BS and AS profiles represents the depth of the scratch groove remaining on the coating surface after the scratch test. Obviously, the difference between AS and DS profiles provides important information on elastic recovery of the coated surface aider the scratch test. The surface profiles obtained from the scratch test outlined above can thus be divided into several regions (Fig. 1). Firstly, the region from point A to B is the pre-scan region under a small load of 0.25 mN, where no scratch damage occurred and the three profiles (BS, DS and AS) are the same. In the region from B to Y, DS profile increases with increasing load, but there is no plastic deformation or material loss in this region because the scratch depth is fully recovered after the scratch test, as evidenced by the same BS and AS profiles in this region. After point Y, plastic deformation or material loss starts, which leads to continually increased AS profile until point C, at which the DS and AS profiles change suddenly due to the failure of the coating. The load at point Y is hence crucial because it is a maximum scratch load that the coating can sustain without plastic deformation. This critical load is called the "Yield Load (Lv)" [7], and was measured for the TiB2 coatings studied in this work, as listed in Table 1. The critical load at point C (Lc) is a measure of the coating-substrate adhesion strength, as further confirmed by microscopic examination of the scratch track, and was also measured for the coatings studied (Table 1).
FINITE ELEMENT MODELLING The scratch process was designed as three steps ie. loading (indentation), sliding and unloading. Although scratch process which involves elastic-plastic deformation has been modelled and simulated for bulk materials by the finite element method by some investigators [8-9], very few have modeled the scratch process of coating systems [10]. It is recognized that the scratch process in relative motion is a very complex system neither easy to understand nor to predict. In the present study, a three-dimensional (3-D) model has been developed by using the capacities of the ABAQUS finite element (FE) code [ 11 ]. The scratch process under consideration involves a hard coating (TiB2) on a substrate (HSS) scratched by a rigid conical indenter. The conical has spherical end form of 25 ktm in radius. Accordingly, the scratch process can be modelled with the finite element mesh shown in Fig. 2. A very fine mesh was used in the coating and the substrate adjacent to the contact zone. The mesh was continuously coarsened further away from the contact area. Two element sets, one corresponding to the coating and the other to the substrate, were used to define separately the elastic and plastic properties of the coating and the substrate materials.
214
Scratching of materials and applications
In order to detect the contact between the coating surface and the rigid indenter, the contact constraint was defined by choosing the indenter as the master surface and the coating surface as the slave surface, where only the master surface can penetrate into the slave surface. The contact or separation between the master surface and slave surface nodal points is automatically detected and monitored in the program. Since strain-hardening is not considered in this work, all the coating and substrate materials were assumed to be isotropic, linear elastic-perfectly plastic materials. In addition, no coating residual stresses are considered in the model. For isotropic materials, elastic deformation ceases and yielding commences when the von Mises yield criterion is satisfied. To simplify the analysis, the adhesion between the stylus and the coating surface is ignored, such that the tangential force induced during sliding is caused only by the ploughing acting of the stylus. Table 2 lists all the material properties used in the FE calculation.
Fig. 2. FE mode of scratch process.
The above FE model is oversimplified of the real scratch test, in that the real continuous test was simulated by a model with step-wise process and frictionless interface. The results of simulation may only be compared qualitatively with experiments.
Table 2: The mechanical properties used in the FE model Material Young's Modulus Yield Strength Coating (TiB2) Substrate (HSS)
E (GPa)
Y (GPa)
350 140
7 3
Poisson's ratio
0.25 0.30
RESULTS AND DISCUSSION Nanoindentation test There has been a wide range of hardness and modulus values reported for TiB2-based coatings produced under different conditions. It is well known that for real application of hard coatings,
Mechanical characterization of nanostructured TiB 2 coatings
215
the hardness required is around 20-40 GPa, which can be achieved in TiB2 coatings. It is not necessary to produce coatings that are too hard, and the adhesion strength which is the subject of this work must also be seriously considered. In order to assess the intrinsic mechanical properties of the coatings, all specimens were tested at 50 nm penetration depths to minimize effect of substrate. Figure 3 shows the typical load-displacement curves of samples 1, 2, 3 and 5. The hardness and modulus values as measured by nanoindentation are summarized in Table 1. From Table 1, it can be seen that the coatings produced with substrate bias have relatively low hardness and modulus and experience significant plastic deformation during the indentation process (e.g. sample 2 in Fig. 3 (a)). On the other hand, the coatings produced without substrate bias experience significant elastic recovery during the unloading stage (e.g. sample 1 in Fig. 3 (a)) and possess much higher hardness, around 30 GPa and higher modulus between 300-320 GPa. It can be seen that samples 2 -5 produced with substrate bias 30 W for the first hour only have relatively low hardness and modulus (Table 1), compared with those produced without biasing (sample 1). Indeed, substrate biasing during deposition causes a drop in coating hardness and modulus.
Fig. 3. Load-displacement curves of samples 1 and 2 (a) and samples 3 and 5 (b) extracted at the penetration depth of 50 nm.
However, substrate cleaning helps to improve the hardness and modulus. With increasing sputter-cleaning time, the hardness and modulus increase significantly as shown in Fig. 4 (a). As confirmed by load-displacement curves (Fig. 3 (b)), for shorter sputter-cleaning time, sample 3, the coating experiences significant plastic deformation during the indentation process, but the coating produced with longer sputter-cleaning time experiences significant elastic recovery during the unloading stage (sample 5 in Fig.3 (b)), possesses much higher hardness of around 26.4 GPa and higher modulus of 283.2 GPa, as well as better coating-substrate adhesion (Fig. 4 (b)), as discussed below. Microscratch test
From Fig. 4 (b), it is obvious that with increasing sputter-cleaning time, the critical load for coating adhesion failure (Lc) increases significantly. For instance, the Lc of sample 5 increases
216
Scratching of materials and applications
by almost 3 times of sample 1 (no sputter-cleaning and no substrate bias). It indicates that substrate cleaning helps to improve the critical load or coating adhesion strength.
Fig. 4. The effect of sputter-cleaning time on (a) hardness and reduced modulus, and (b) critical load.
Figure 5 shows the typical multi-pass scratch curves with linearly increasing load of sample 1. It can be seen that DS profile during scratch increases linearly with increasing load. The profiles in Fig. 5 can be divided into several regions. Firstly, the region from A to B shows the pre-scan (300 ~m) under the small load of 0.25 mN, where no scratch damage occurred, since BS, DS and AS have the same profile. Secondly, in the region from B to C, DS profile increases almost linearly with increasing load. However, there is no plastic deformation or material loss that took place in this period which is shown by the elastic recovery of TiB2 coating represented by AS and BS having the same profile. It is noted that after point Y, plastic deformation or material loss starts, which the load at point Y is identified as yield load (Lv), which was described in section of experimental details.
Fig. 5. Experimental multi-pass microscratch curves of deposited TiB2 coating (sample 1).
Mechanical characterization of nanostructured TiB 2 coatings
217
The Ly values for all coatings are also listed in Table 1. It is noted that by increasing the time of sputter-cleaning of substrate, the increment of Ly values are observed. This result suggests that sputter-cleaning of substrate helps to improve elastic recovery of coatings since the oxide layers or contaminants on the substrate surface were removed by sputter-cleaning, which resulted in enhancing coating adhesion to substrate and adhesion strength. From the region Y to C, all profiles are relatively smooth until point C, where fluctuations in DS profile and AS profile start. This indicates the damage of TiB2 coating. Therefore, the scratch load at point C can be taken as the critical load (Lc) for coating failure. This phenomenon was further verified by SEM examination of the resultant scratches to identify the scratch mode of coating failure as shown in Fig. 6. The Lc values for all coatings are also listed in Table 1. It can be seen that Lc is much enhanced by sputter-cleaning as expected due to the increment of Ly. SEM examination revealed that coating failure of sample 1 (Fig. 6 (a)) produced without sputter-cleaning and bias, is typical of the compressive spallation failure mode, which shows poor adhesion behaviour due to brittle coating [12]. However, after introducing substrate bias, although the hardness and modulus reduce, the coating adhesion is improved which can be seen from the increase of critical load value. It is also noted that the failure mode has been changed from compressive spallation to wedge spallation (Fig. 6 (b). The wedge spallation occurs because of the accumulation of residual stress during the stylus motion [ 10]. Wedge spallation occurs instead of compressive spallation because the coating adhesion is strong enough to bear the stress. In addition, compressive shear cracks are formed in the coating, which led to interfacial detachment (Fig. 6 (b)).
Fig. 6. SEM images showing the scratch tracks and mode of failures of (a) sample 1, (b) sample 2, (c) sample 3 and (d) sample 5.
Figure 6 (c) shows the failure mode of sample 3 produced with sputter-cleaning for 30 min and substrate bias 30 W for the first hour. The failure mode is also the wedge spallation with smaller damaged areas and less wearing out of the coating. When sputter-cleaning time is longer, the wedge spallation mode is changed to micro-cutting (no coating failure) (Fig. 6 (d)), which
218
Scratching of materials and applications
occurs due to the physical interaction between the stylus and the coating surface. It thus indicates the tremendous improvement in coating adhesion strength. In order to find the critical point of sample 5, the scratch distance was extended from 1,000 ~tm to 1,500 ~tm, and the experimental critical load of sample 5 was found to be 875.5 mN, almost 3 times that of sample 1. Figure 7 shows the microscratch track of a TiB2 coating examined under AFM. It is obvious that there is a certain amount of material removal during scratch test. AFM also reveals that the scratch track exhibits the formation of material pile-up along the sides, which is confirmed by FE simulation. The pile-up was caused by plastic deformation of the coating and substrate, which was grooved by the indenter tip. In addition, the orientation of the surface of the scratch track is around 45 degree, which is in line with the maximum shear stress direction.
Fig. 7. AFM images showing the orientation of surface of scratch track of deposited TiB2 (sample 1).
FE simulation of scratch process In order to gain a better understanding of scratch behaviour during the testing process, FE simulation has been employed. To simplify the problem, the scratch process was simulated with constant applied load (normal load) mode. Fig. 8 shows the yon Mises stress contours in the TiB2 coating/HSS substrate system during the scratch process.
Mechanical characterization of nanostructured TiB2 coatings
219
Fig. 8. Development of plastic zone in the scratch process (transverse section view): (a) loading process at penetration depth of 0.1 ~tm, (b) during scratch process.
Figure 8 (a) shows the step of indentation into the coating, in which there is no sliding movement but only a normal load applied by the indenter. At small indentation depth (0.1 ~tm, Fig. 8 (a)), plastic deformation occurs around the indenter tip region, which propagates both vertically and laterally as round shape. It is noted that plastic deformation occurs at the indenter tip region, not only in the top layer (stress level of 7 GPa), but also propagates to the substrate (stress level of 3 GPa). With increasing penetration depth, plastic deformation propagates further below the surface and at the substrate. Plastic deformation also propagates both
220
Scratching of materials and applications
vertically and laterally both in the coating and the substrate. The interface between the coating and substrate can be identified clearly by the different level of stresses (as shown in the figure in different colours), which resulted from the difference in material properties. Figure 8 (b) shows the early stage of scratching process. The indenter slides on the surface under the applied normal load. This results in the ploughing of the material, and a tangential force due to this ploughing action. As the stylus is moving, the plastic deformation zone is moving in the same direction but the plastic deformation zone is no longer located in the center of the indenter tip region. It is moved towards the leading edge of the indenter, which can be seen clearly from the top view as shown in Fig. 9 (a). When the scratch test finishes, the stylus moves away from the coating as unloading step. It can be seen that compressive residual stress are present within the groove and material pile-up is identified (Fig. 9 (b)). In order to compare the simulation results with experiment ones, constant load scratch test was conducted at a load of 240 mN. Fig. 10 compares the measured tangential (friction) force with that from FE simulation. It can be seen that the friction force from the experiment is higher than the calculated one. It arises from the facts that the measured friction force is the combined effect of the ploughing action of the stylus and the adhesion between the stylus and the coating surface, whilst the calculation tangential force is the result of the ploughing action alone, since stylus-coating adhesion was not considered in the FE model. Thus, the difference between the measured values and calculated values should be the adhesion force between the stylus and the coating surface, which is difficult to measure by experiment, but can be estimated through simulation using the present model. For example, in this case study, the experiment shows the average friction force values of 48.41 mN, whilst the FE model calculates the average tangential force values of 22.39 mN. The adhesion force between the stylus and the coating surface should be around 26.02 mN.
(Continue to the next page)
Mechanical characterization of nanostructured TiB2 coatings
221
Fig. 9. (a) 3-D stress contours during scratch process, and (b) scratch track after unloading step.
140
+
~, 120
FE calculated friction force (raN) Expetimantal friction force (naN)
I00 8o 60 4O
20 m
"Ira
I
0
3
I
1
I
I
6
9
12
15
18
Scratch distance (micron) Fig. 10. Comparison of simulated tangential force with experimental results at constant load of 240 mN and scratch distance of 15 p.m.
222
Scratching of materials and applications
CONCLUSIONS (1) Sputter-cleaning of substrate helps to improve TiB2 coating hardness and adhesion strength. (2) The TiB2 coatings produced with substrate bias possess low hardness, but good adhesion with the substrate. With biasing at the early stage of deposition, the failure mode is changed from compressive spallation to wedge spallation, which obviously shows the improvement in coating adhesion strength. (3) A FE model has been developed to study and simulate the scratch process. The model is applied under assumption of step-wise loading, and scratch, and can be used to simulate the loading, scratching and unloading stages, the plastic deformation behaviour of the coating and the substrate during the scratch test. (4) The present model can be used for estimation of the adhesion force between the stylus and the coating surface.
REFERENCES o
2. .
4. 5. 6. 7. 8. 9. 10. 11. 12.
Munro, R.G. (2000) Jr. Res. NatL Inst. Stand. Technol. 105, 709. Cutler, R.A. (1991). Engineering properties of borides. Engineering Materials Handbook: Ceramic and Glasses. ASM International, Vol. 4. Berger, M., Coronel, E., Olsson E. (2004) Surf Coat. Technal. 185, 240. Berger, M., Karlsson, L., Larsson, M., Hogmark, S. (2001) Thin Solid Films 401,179. Panich, N., Sun, Y. (2005) Surf. Coat. Technal. 198, 14. Oliver W C, Pharr G M. (1992) J. Mater. Res. 1992;7:1564-1583. Xia, J., Li, C.X., Dong, H., and Bell, T. (2004) J. Mater Res. 19, 291. Xia, J., Li, C.X., Dong, H., and Bell, T. (2004) J. Mater Res. 19, 291. Wong, M., Lim, G.T., Moyse, A., Reddy, J.N., Sue, H-J. (2004) Wear 256, 1214. Holmberg, K., Laukkanen, A., Ronkainene, H., Wallin, K., Varjus, S. (2003) Wear 254, 278. Hibbitt, Karlsson and Sorensen Inc. (1998) ABAQUS. User's Manual. Pawtucket. RI, Ver 6.3. Salas, O., Keams, K., Carrera, S., Moore, J.J. (2003) Surf Coat. Technal. 172, 117.
223
CHAPTER 11
DAMAGE IDENTIFICATION OF DLC COATING BY MICROSCRATCH TEST
A. DJAMAI*, H. ZAIDI*, K. J. CHIN* and T. MATHIA ** * Universit~ de Poitiers, Laboratoire LMS (UMR-661 O- CNRS), SP2MI, T~l~port 2, Boulevard Marie et Pierre Curie, BP 30179, 86962 Futuroscope Chasseneuil Cedex, France. ** Ecole Centrale de Lyon. Laboratoire de Tribologie et Dynamique des Systbmes (UMR-5513CNRS),36, Av. Guy de Collongue, 69134 Ecully, France. E-mail . [email protected]
ABSTRACT In order to characterize the adherence of DLC coatings (Diamond Like Carbon), scratch testing was performed on a unit equipped with sensors for normal and tangential forces, and an acoustic detector to detect the nucleation and the propagation of cracks. The system is also equipped with a microscope permitting observation of each event on the scratch according to the tangential friction force signal or the acoustic signal. The local microscopic observation allows identification of the damage with respect to the normal load. The test was performed with a Rockwell C indenter at the relative displacement speed v = 10 mm/min under a progressive normal load from 5 to 55N. Coating failure appears in various modes, particularly the following : propagation of the cracks along the longitudinal edges of the scratch; propagation in front of the indenter; rupture along the maximum principal stress lines; and. detachment in the subsurface by shearing of the coating. The microscopic analysis of the evolution of the scratch under a progressive normal load permits identification of the various traces and the damage mechanisms of the coating. In this study, experimental results are shown for the scratch tests on bulk glass and DLC coating. Various modes of crack initiation, damage and rupture of these materials according to the critical normal load are presented. The analysis of the contact stress field distribution in bulk glass enables identification of the crack initiation and its propagation in the coating. KEYWORDS DLC coating, Adherence, Scratch testing.
INTRODUCTION To take advantage of the remarkable mechanical and thermal properties of diamond, many research teams deposit diamond type layers on supports such as the coating on cutting tools, etc. The difficulty is to create the thermal conditions for existence of the diamond phase out of equilibrium (at low pressure). Diamond is formed only in extreme conditions of temperature and pressure.
224
Scratching of materials and applications
The artificial diamond synthesizers make diamond grow in conditions out of thermodynamic equilibrium so that, locally, the parameters of diamond formation are favored compared with those which lead to graphite formation. During the deposition phase, there is always graphite which settles in great quantity (98 %) and diamond at a very low rate (< 1%). To consume the deposited graphite, we operate in the presence of a more erosive gas toward graphite (X 1000) than that for the diamond structure, such as hydrogen or oxygen. Bachmann [ 1] described on a triangular diagram (Carbon, Hydrogen, Oxygen) the zones likely to favor the diamond structure out of thermodynamic balance for a combination of hydrogen and carbon. The zone with best practical probability is situated around pure hydrogen with a low carbon concentration. However, in a carbon / oxygen mixture, the growth of diamond is probable in a broad range of concentration for the gas constituting the mixture. Being placed in zones where the composition of carbon, oxygen and hydrogen is in a favorable ratio, it is possible to obtain diamond crystals or DLC with nanostructure sp3/sp2. The physicochemical and mechanical properties of the coating strongly depend on the sp3/sp2 ratio. One of the principal requirements for films on the surface of a substrate in the majority of the technological applications is that the film adheres to the substrate. If adherence is insufficient, premature failures may result from detachment of the film by interfacial fractures. Many measurement techniques for thin and hard film adherence have been proposed, among which the techniques of scratching and marking after application of a load are the most widespread. During scratch testing DLC, we observed practically the same figures of damages as those on a brittle solid bulk like glass; therefore, we considered it useful to compare the surface damage phenomena of DLC with that of glass. We performed experimental and theoretical tests on these two materials to better understand the surface cracking of DLC during scratching. The modeling of the stress fields in solid bulk enabled us to explain the mode of initiation and propagation for cracks in DLC coatings. Many studies of the contact mechanics have been published since Hertz established the theory of elastic contact for isotropic materials. Concerning the punch/plane indentation, the stress field was solved by Barquins and Maugis [2]. The determination of a stress field for a spherical contact sliding on a plane was established by Hamilton and Goodman [3]. With regard to brittle materials, Lawn [4], Mouginot [5], Mathia [6] and Zeng [7] studied cracking during a spherical indentation. In addition, Lawn [4] and Frank [8] used the energy criterion of Irwin to anticipate the points of crack initiation during scratching. In a sliding contact, where low friction coefficient is involved, these cracks propagate in front of the indenter, forming opened rings wrapping the contact zone [4].
EXPERIMENTAL EQUIPEMENT The coating adherence characterizes the mechanical resistance of the interface between the coating and the material on which it is deposited. Currently, the only test possible where a quantitative measure of adherence (critical load) is obtained involves the destructive test of scratching.
Damage identification of DLC coating by microscratch test
225
The scratch tests were carried out on the Teer Coatings ST-3001 scratch tester equipped with sensors for normal and tangential forces, and an acoustic detector to visualize the nucleation and propagation of cracks. The system is also equipped with a microscope for observing each event on the scratch according to the friction tangential force signal or the acoustic signal. On glass, the scratch test is carried at length of 10 mm, at the relative displacement speed of 10 mm/min under progressive normal load. The loading rate during a test is 110 N/min. The indenter was a Rockwell C 0.8 mm. The glass used was ordinary glass soda lime of dimension 3cm x 3cm x 1cm with Young's modulus E = 60 GPa. The DLC specimen was deposited on stainless steel 304L (2cm x 2cm) in a PVD reactor with CH4-H2 gas mixture. The chosen deposition parameters during the growth tests are: temperature 800 ~ gas pressure 30-40 mBar, total flow rate 300 cm3/s, the gas ratio CHa/H2 = 1.5 % and time deposition is 1 hour. The substrate was polished to roughness Ra = 13 nm by diamond powder. The coating thickness measured by scanning electronic microscopy (SEM) is 3/zm. The applied load varies from 5 N to 55 N, the loading rate is 100 N/min and the displacement speed of the specimen is 10 mm/min. The indenter is the Rockwell C 0.2 mm.
EXPERIMENTAL RESULTS Scratching on bulk glass The experimental results relating to the scratching on glass are shown in Fig. 1. Curve in Fig. l(a) gives the evolution of the normal load in time, curve in Fig. l(b) corresponds to the tangential force during the sliding according to the normal load, curve Fig. 1(c) is the derivative of the tangential force according to the normal load, and the curve Fig. 1(d) corresponds to the acoustic emission during scratching with respect to normal load. The rupture or the flaking of glass (on the surface) leads to a discrete acoustic emission with high amplitude produced by strong release of elastic energy stored. The process is followed recording of the associated acoustic emission which allows determination, in a more reliable way, the points corresponding to failure. According to the curve in Fig. l(d), we note that at the beginning of the test, the acoustic emission is around 40 dB. This noise corresponds to the background noise of the scratch tester to which the noise of the sliding is added. Around a normal load of 30 N, there is a rise in the curve reaching the value of 85 dB. This abrupt increase suggests the beginning of deteriorations in glass. On the other hand, at this stage, the curve in Fig. l(b) does not yet show a notable change in the friction force. We do not observe a variation in the derivative of the friction force (curve in Fig. 1(c)) for the load of 30 N. The critical normal load is related to the abrupt rise in the acoustic emission. This critical load is determined by visualization of the scratch, using a microscope. We find the position of the crack and get the corresponding load by direct observation (Fig. 2(a)) and (Fig. 1). Experimental results give a load of 30 N for the first damage corresponding to a pressure of 8 GPa in the center of the contact zone. As the normal load increases, the curves in Fig. l(b) and l(c) show increased fluctuations. Indeed, the damages tend to be more important with increasing normal load. From a normal load
226
Scratching of materials and applications
of 89 N, there is a peak on curve in Fig. 1(c), as well as an apparent change in the linearity of the friction force on the curve in Fig. 1(b). Visual examination using the microscope corresponding to the track position of this peak confirms the complete deterioration of surface and the formation of closed crack rings (curve 2(c)). From the normal load of 89 N onward the acoustic emissions remain around 95 dB indicating constant cracking and complete failure on glass until the end of the test.
Fig. 1. Scratch test on glass with the indenter Rockwell C 0.8 mm. (a): evolution of applied normal load N versus time, (b): firction force according to N, (c): derivative of the tangential force according to N, (d): acoustic emission intensity versus N.
The friction coefficient of the scratching is 0.061 through out the test. The critical load of crack initiation calculated for an average of 10 tests is 29 N. The load for complete deterioration is obtained in the same way which is 85 N. In the beginning of scratching (Fig. 2(d)), when the normal load reaches the critical load of crack initiation, the first circular crack is formed as anticipated in the theoretical model of Hamilton and Goodman at the rear of the contact zone on the sliding axis. This crack propagates in a quasi circular manner toward the front of the indenter, following the direction of the second principal stress in the surface. As the indenter continues to advance, it will inevitably pass by the first crack, at which point another phenomenon will occur. Two other cracks (symmetrical compared to the direction of displacement) will be propagated step by step starting from the crack formed previously (see Fig. 2(d)) which are initiated by the back of the indenter. They will propagate under the effect of the first principal stress in these points. The directions that they take are in the direction of the second principal stress trajectories, perpendicular to the first crack. When the indenter advances further, it is possible for a third crack to form just after the second. When the propagation of the third crack joins the second one, there is then a flaking of the surface since the three cracks form a closed body facilitating flake removal. As the normal force increases and thus the tangential force, the cracks in the edge trace tend to propagate
Damage identification of DLC coating by microscratch test
227
farther. If the two symmetrical cracks (compared to the sliding axis) on the edge of the trace meet, the complete deterioration of surface and thus the formation of closed circular cracks results (in this case for a normal load of 89 N). It is noticed that as soon as the first damage appear at the beginning of the test, the phenomenon described above will reproduce continuously and in a repetitive manner through to the end of the test. We note that the value of the maximum principal stress on these points is lower than that taken at the rear of the same contact zone. On the other hand, it is sufficient to cause crack propagation, the initiation being already realized. In parallel, this does not prevent seeing the cracks formed at the rear of the indenter, which exist less visible, these are observed in images shown in Fig. 2(a) and 2(b).
Fig.2. Optical observation of the scratch on glass.
Scratching on DLC coatings Figure 3 gives the damage evolution on DLC film during scratching under a progressive applied normal load from 5 N to 55 N. Figure 3(a) shows the general scratch observed on the DLC coating after the test. Figures 3(b), 3(c) and 3(d) are the partially magnified images of Fig. 3(a). On Fig. 3(b), we observe the first damage mark in the edge of the scratch line. The image in Fig. 3(c) corresponds to the appearance of the first flaking, and the image in Fig. 3(d) presents the appearance of the complete failure of the DLC on the surface.
228
Scratching of materials and applications
The first critical normal load, C~, relating to the first damages (Fig. 3(b)) is at 12 N. The critical load, CR, for complete coating failure is at 24 N. This load is characterized by the formation of the first closed circle (Fig. 3(d)). Additionally, in comparing the damages on bulk glass (Fig. 2) to those on DLC coating (Fig. 3), we observed that the failures present on the DLC coating are very similar to those observed on the scratch track of glass. Indeed the observation of the first cracks in edge of the track Fig. 3(b) does not indicate that there was no damage beforehand Oust like on glass). It only indicates on the other hand the place where we can optically locate them.
Fig. 3. Optical observation of the scratch on DLC coating.
Figure 3(c) and Fig. 2(d) show flaking in both cases. On glass, we could prove that this flaking was due to the propagation of the cracks formed beforehand. According to the optical observation, it is the same on the DLC film. In these two materials, the complete surface failure (Fig. 3(d) and Fig. 2(c)) is characterized by circular damages formed by the union of the two symmetrical cracks in the edge of the scratch track.
MODELING AND DISCUSSION By using the equations of Hamilton and Goodman [3], we obtain under static condition that q, (the largest among the three principal stress) acts following a radial direction (Fig. 4), where P is the indentation load, a is the radius of contact area, and x, y are the Cartesian coordinates. The tops of this stress curve on all the plane z - 0 form a ring which would be responsible for cracking, according to Frank and Lawn's research [8]. However, work of Mouginot [5] (using the energy criterion of Irwin) shows that cracking does not take place on the contact zone but just outside of it.
Damage identification of DLC coating by microscratch test
229
Fig. 4. Distribution of the maximum principal stress c l undimensionned with respect to the maximum pressure on the surface for a friction coefficient la = 0.
The effect of the tangential friction force is to increase the compressive stress in front of the contact zone and to intensify the traction at the rear of the contact zone when the friction coefficient # increases. The stress traction value at the rear of contact is [3] (1)
3P [ 1 - 2 v 2 a , = ~27ra 2 3
+/Jx-
4+v2] 8
where P2 is the Poisson's ratio. Figure 5 represents the distribution of the maximal traction stress in the surface for a friction coefficient of 0.06 (experimental value of the friction coefficient in our tests).
Fig. 5. Distribution of the maximal principal stress ~! on the surface for a friction coefficient ~t = 0.06.
Scratching of materials and applications
230
Fig. 6. Isostatics of the first and the second principal stresses; (a): Static indentation, ~ = 0; (b): friction coefficient ~ = 0.1.
During propagation on the surface, the crack follows the direction of the second principal stress (propagation perpendicular to the stress having caused the failure; dark lines). Figures 6(a) and 6(b) represent the virtual paths by which the crack will be propagated for four various initiating points at the rear of the indenter. In static mode, the crack will follow the directions of propagation corresponding to perfect circles of radius equal to the distance between the initiating point of cracking and the center of the contact zone (dark line on Fig. 6(a) and 6(b)). In relative sliding (Fig. 6(b)), the trajectories of crack propagation tend to open and move away from the contact zone when the friction coefficient ~t. increases, which corresponds to our experimental observations. These microscopic observations of crack initiation and its propagation show that the distributions of the contact stress on the surface of DLC coating are similar to those of bulk glass. We can explain this similarity by the high hardness of the coating.
CONCLUSIONS The experimental study of the scratching on bulk glass and DLC coating permits the following conclusions: The first critical load C~ of bulk glass is 29 N and the second critical load CR is 85 N (indenter; Rockwell C 0.8 mm). The first critical load Cl of DLC coating is 12 N and the value of CR is 24 N (indenter; Rockwell C 0.2 mm). The exploitation of modeling based on Hamilton's study concerning the contact stress field on the surface for a brittle bulk material enabled us to explain the failure modes for DLC coating. The cracks optically observed on glass and DLC coating correspond well to the maximum principal stress on the surface. The propagation of the cracks depends on the applied normal load and the friction coefficient of the contact.
Damage identification of DLC coating by microscratch test
231
The propagations of the cracks observed on DLC coating are similar to those observed on bulk glass.
REFERENCES
.
7. 8.
Bachmann, P. K., Leers, D. and Lydtin, H. (1991), Diamond Relat Mater., 1. p. 1. Barquins, M. And Maugis, D. (1982),JMecanique Theor Appli, 1, p. 331. Hamilton, G. M. and Goodman, L E. (1966), JAppl. Mech. 33, p. 371-376. Lawn, B. R. (1967),Proc R. Soc, vol 299,, p. 307-316. Mouginot, R. (1988), Fracture of fragile elastic materials under indentation of plane and spherical punches, Thesis, University Paris 6. Mathia, T. G. and Encrenaz, E. (1981), Wear, Vol. 73, No 1, p. 77-81. Zeng, K. and Breder, K. (1992), Acta metal. Mater. vol 40, No 10, p. 2601-2605. Frank, F. C. and Lawn, B. R. (1967), Proc. R. Soc. London, Vol. 299, p. 292-306.
232
CHAPTER 12
CORRELATION BETWEEN ADHESION AND WEAR BEHAVIOUR OF C O M M E R C I A L CARBON BASED COATING Originally published in TribologyInternationalvol 39. FebruarT 2006
K.H. LAU and K.Y. LI
Advanced Coatings Applied Research Laboratory (ACARL) Department of Manufacturing Engineering and Engineering Management City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong SAR, China. E-mail: [email protected]
ABSTRACT Adhesion and wear behaviour of hard coatings are two important properties that concern researchers. We need to use different tests to obtain these two values and the correlation between adhesion and wear is not always obvious. The purpose of this chapter is to find the relationship between the wear rate and the critical load for commercial hard coatings. Commercial carbon based coatings were deposited on M42 high speed steel by magnetron sputtering. Different lengths of pre-sputtering time were used to change the properties of the coatings. Single pass scratch tests and multi-pass bi-directional wear tests were used to obtain the critical load and wear rate values, respectively. Results showed that there was a correlation between wear and adhesion. It was found that adhesion increases and wear rate decreases when the pre-sputtering time increases. It was also observed that adhesion strength varies with pre-sputtering time in a step function manner. To a large extent, the adhesion behaviour is related to the wear behaviour as far as the commercial carbon based coating is concerned. However, as different coatings have different microstructure and properties, further work should be done to decide whether a similar relationship can be applied in other cases.
KEYWORDS Adhesion; wear; scratch tests; pre-sputtering time
Correlation between adhesion and wear behaviour of commercial carbon based coating
233
INTRODUCTION Adhesion and wear behaviour of hard coatings are two major properties that always concern researcher and several studies have been done on them. Adhesion of the coatings is an important property. The greater the adhesion between the film and substrate, the more difficult it is to break the coating. As drastic coating failure, such as spallation, will occur if the adhesion is poor [ 1], many researchers also focused on studying the adhesion of coatings [2 4], Wear behaviour [5, 6] is an important factor also, as it can help us to determine some specific coating properties such as durability. There are various methods for assessing the adhesion of hard coatings to a substrate [7, 8], but most researchers agree that the scratch test is the most practical approach [9]. The result of scratch test can be identified by the critical load (Lcl and Lc2, which is defined as the load where initial coating failure is detected and total delamination of the coating from the substrate respectively). Due to the high reliability of scratch test, many researchers use this method to assess the adhesion of coating, see Refs [3, 10]. As far as wear performance of coating is concerned, a hard counter-body material (e.g. tungsten carbide) is commonly used to carry out sliding wear. One common test is Pin-on-disc wear test [11]; in this paper a multi-pass bidirectional wear test is used. Sometimes it is not feasible to carry out both the adhesion and wear test for a given coating sample due to the availability of machine, time and cost, etc. Therefore, if it is possible to find a general relationship between the two tests, one less test could then be used to identify the basic mechanical properties of the film. Up to now, few papers pay close attention to this topic, and even those researchers involved cannot observe any obvious relationship between them. For instance, S. J. Bull [12] had done similar work using TiN coatings deposited on 304 stainless steel, and he found that the correlation was poor. His results showed that when the critical load (Lc) increased, the wear resistance of some samples increased, but generally there was no change. Therefore, he concluded that the correlation between adhesion and wear was poor as there were many other factors which influenced the wear performance. The aim of this paper is to investigate the correlation between wear and adhesion of a commercial carbon based coating by using scratch tests and multi-pass bi-directional wear tests.
EXPERIMENTAL DETAILS Deposition
A Teer UDP450 closed field unbalanced magnetron sputter ion plating (CFUBMSIP) system was used to deposit the coatings onto silicon wafer, as well as hardened (hardness-8 GPa) and
234
Scratching of materials and applications
non-hardened (hardness---2.5 GPa) M42 high speed steel. The steel substrates were polished to roughness (Ra) about 0.03#m. DC power and pulsed DC were used on the targets and the substrate respectively. To obtain coatings with different properties, five different pre-sputtering times [13] (3, 10, 20, 35 & 60 minutes) with -450V bias voltage on the substrate were used before depositions. Afterwards, -80V bias voltage with a frequency of 50 kHz was used on the substrate during deposition for two hours. The coating thickness was around 2#m for all samples (see figure 1). Scratch test
The scratch tests were carried out using a Teer ST-3001 scratch tester. A Rockwell C diamond tip (120 ~ 200#m radius) was used as an indenter. During the test, the indenter was drawn over the coated surface as the applied normal load increased continuously until coating failure occurred. The load was increased from 10 N to 80 N by 100 N/min and sliding speed was constant at 10 mm/min. The scratch tests were performed in air at temperatures between 22 ~ and about 68% relative humidity (RH). The scratch tracks were observed under microscope attached to the scratch tester and scanning electron microscope (SEM). The failure modes of the scratch tracks on different samples were obtained and the critical loads were determined by locating the position at which the coatings started to fail. Wear test
Besides testing adhesion, the ST-3001 was employed to perform the reciprocating wear (multi-pass) test. In this test, a 5 mm diameter tungsten carbide (WC-6% Co) ball applied a constant load to the sample. Once the initial load was reached the sample moves backwards and forwards over a pre-selected distance until the pre-determined number of cycles was completed (one cycle means one complete reciprocating movement). The ambience conditions were around 22 ~ and 68% RH. As the samples have different wear behaviour, depending on the pre-sputtering time, the number of cycles for each of the sample was carefully selected to ensure the indenter will not wear through the coating and reach the substrate and therefore lead to an incorrectly wear rate calculation. Parameters of the test are shown in Table 1. After each test, a Talysurf profilometer was used to measure the cross sectional area of each wear track in order to calculate the wear rate.
Correlation between adhesion and wear behaviour of commercial carbon based coating
Table 1"
I-
235
Test parameters of multi-pass bidirectional wear test Constant load (N)
30
[_Linear displacement (mm) I Linear velocity (mm/min)
300 Number of cycles
Pre-sputtering time (mins)
Hardened steel substrate
Non-hardened steel substrate
3
750
500
10
750
500
20
750
500
35
1250
500
60
1250
500
RESULTS AND DISCUSSION Microstructure
The carbon based coating is a solid lubricant coating which is different from other metallic coating such as TiN, as the coating does not show any columnar structure along the film growth direction, under SEM [14]. Figure 1 shows the cross section of the carbon based coating. No proper pattern or microstructure can be seen from this figure. The grain size of the coating must therefore be so small that it cannot be observed by SEM.
Fig. 1.
Scanning electron micrograph of cross section of carbon based coating on steel
236
Scratching of materials and applications
Scratch tests Scratch failure modes: Figure 2 shows examples of the scratch test failures modes that occurred in the coatings. Figure 2 (a) shows the failure mode in lower critical load (Lcl). Slight spallation has occurred at a load equal to 18 N, so this is the Lcl value of this coating. The spallation occurs due to the compressive stress generated by the indenter, which causes coating deformation. Such deformation can be seen in figure 2 (a), which shows some shallow diagonal lines in the track. Those diagonal lines are formed when the coating is deformed by the indenter in its moving direction. If the adhesion of the coating is not adequate, the deformation will cause the coating to detach in order to release the deformation energy. Spallation becomes more serious as the load increases. Figure 2 (b) shows the scratch track when indenter load increases to 32 N. It shows more spallation than figure 2 (a) and as the spallations dominate the track, diagonal lines are not obvious in this figure. Figure 2 (c) shows failure mode in upper critical load (Lc2). The white region of the track is the substrate, which can be seen after the film delaminates. If the adhesion of the coating is not good, the coating undergoes serious spallation and starts cracking when the load is high. Eventually it piles off, when the compressive stress in front of the indenter reaches a certain limit. In our experiment, however, this occurred only in the sample which used non-hardened steel as a substrate and three minutes pre-sputtering time, meaning this sample has the worst adhesion strength.
(Continue to the next page)
Correlation between adhesion and wear behaviour of commercial carbon based coating
Fig. 2.
237
Scratch failure pictures of the coating on non-hardened steel substrate and under 3
minutes pre-sputtering time at (a) Lcl (b) failure track (c) coating pile off, Lc2
Adhesion versus pre-sputtering time: Lower critical load (Lcl) value had been used as
adhesion strength of coatings for analysis. Figures 3 (a) & (b) show that the adhesion strength (Lcj) increases when the pre-sputtering time increases, for both hardened and non-hardened steel substrates.
Scratching of materials and applications
238 7570 65 6O "-I ..... 55 50 45
I ~
9 9
............................
9
~" 4o 9,-- 35 030 ,.--I 252O 15 10 5-" O
.
.
o
.
lb
.
.
2'0
(a)
.
3'0
4'0
~'o
do
Pre-sputtering Time (min)
25I..--
20
-! .... 1 ....... "--, 15 zv
o 10
..1
~b)
o
'
~'o
'
2b
'
3'o
'
.'o
'
~'o
" 6b
Pre-sputtering Time (min)
Fig. 3.
Adhesion (Lcl) versus pre-sputtering time for (a) hardened steel substrate (b)
non-hardened steel substrate
They vary in a step function manner, that is to say they have similar Lc values for under 20 minutes pre-sputtering time, then the values suddenly increase for 20 to 35 minutes pre-sputtering time, but after 35 minutes, further increasing the time has insignificant effect on
Correlation between adhesion and wear behaviour of commercial carbon based coating
239
the adhesion. The adhesion on hardened steel is much better than on non-hardened steel (70 N versus 23 N). In the increasing range of the Lc value, they have similar percentage increase which is about 27%. This shows that the change of pre-sputtering time has a similar effect on the adhesion regardless of the substrate used. A possible explanation for the effect of pre-sputtering time on adhesion is that ion cleaning before deposition reduces surface contaminants to an acceptable level. It is expected that the change of adhesion is caused by the change of the interfacial bonding, and that impurities can have a substantial effect on this interfacial bonding. According to Koski [ 13], contaminants on the surface will separate the coating and substrate, weakening the interfacial bonds between them and directly decreasing the adhesion energy. So the adhesion can be improved by removing contaminants such as oxide and carbon at the interface. When the pre-sputtering time increases, the percentage of the contaminated surface layers removed increases and thereby enhances the adhesion strength. The adhesion strength will not keep increasing with pre-sputtering time, and H. Ronkainen et al. [ 15] showed that when he coated a-C:H film with TiC intermediate layer on cemented carbide, after the pre-sputtering time reaches the optimum point, further increase in the time will cause the critical load values to slightly decrease.
Fig. 4.
Cross section area of the wear track measured by profilometer (Sample: carbon based
coating on hardened steel and 3 minutes pre-sputtering time)
Wear Rate versus pre-sputtering
The wear rate is calculated by first using the Talysurf profilometer to measure the cross sectional area of the wear track (see example Fig. 4), then by using equation 1"
w = _ v~
DF
(1)
where W is wear rate (m3/Nm), V is the wear volume (m3), D is total travelling or sliding distance (m) and F is the constant normal load (N) apply on the sample.
Scratching of materials and applications
240
Figure 5 shows that the wear rate of the coating decreases when the pre-sputtering time increases. In figure 5 (a), the wear rate initially decreases from 2.5x10 15 to 1.8x10 -15 m3/Nm when increasing the pre-sputtering time from 3 to 20 minutes, the wear rate drops suddenly to less than 0.5x10 ~5 m3/Nm when the time increased beyond 35 minutes. Further increase in time to 60 minutes has little effect on the wear rate. The coatings on non-hardened steel substrates show different behaviour. Figure 5 (b) shows that the wear rates decreases gradually from 10 x l 0 15 to 9 x l 0 -z5 ma/Nm for the same pre-sputtering time range. It is believed that ion cleaning removes the contaminants on the surface of the substrate and hence strengthening the interfacial bonds between coating and substrate; therefore the coating provides better wear resistance.
3.0 |
\x
~" 2.5
"i
E 2.0
\,| I'..
& 1.5 t~ 1.0 ~
0.5
(a)
9
0.o
i
Pm-s~ttenng time (min)
10
-Jr.... | ....... I_
E
z
~
.............
J...
8
E b T--
6
rr" 1_ m
~2 0
(b) Fig. 5.
o
9
i
lo
~
'"
~
,~
~
6'o
Pre-sputteringTime (min)
Wear Rate versus pre-sputtering time (a) hardened steel substrate (b) non-hardened
steel substrate
Correlation between adhesion and wear behaviour of commercial carbon based coating
241
Impact o f the substrate microstructure on the wear rate and adhesion
When non-hardened steel substrates are used, the coatings show poor adhesion and high wear rate (see figures 3 (b) & 5 (b)). This is because the hardening process changes the microstructure of the steel. According to J. Petersen et al. [ 16], the hardening process improves the uniformity of the ferritic-pearlitic microstructure, and this fine grained structure induces the steel have both high strength and good toughness.
As the hardened steel gives stronger
support to the coating, the deformation of the coating is lower. And this will decrease the tensile stress imposed on the film and this results in good adhesion and low wear rate. Besides, as the coatings may deform with the substrate, the larger deformation of the soft substrate leads to incorrectly calculated wear rate values of the coatings. Therefore, the calculated wear rates will be overestimated. See Ref [ 17] for details. In short from the results of adhesion and wear tests, it was found that a minimum pre-sputtering time of 35 minutes, and hardened steel substrate should be used to obtain commercial carbon coating with superior properties. Correlation between adhesion test and wear test Scratch and wear tracks: Figures 6 and 7 show the cross sectional pictures of the coatings and
substrates after the scratch test.
Fig. 6.
Cross section of scratch track when loading at 30 N (a) whole track (b) comer of the track
(Sample: carbon based coating on non-hardened steel and 3 minutes pre-sputtering time)
Scratching of materials and applications
242
Fig. 7.
Cross section of scratch track when loading at Lc2 --- 60 N (a) whole track (b) comer
of the track (Sample: carbon based coating on non-hardened steel and 3 minutes pre-sputtering time)
The scratch track at Lc2 (Fig. 7 (a)) has larger deformation than at indenter load equal to 30 N (Fig. 6 (a)) because higher loading was applied in Lc2. Moreover in figure 6, some coating remains on the track, but in figure 7, the coating is completely removed by the indenter. As Lc2 is defined as the point of total delamination of coating from the substrate, figure 7 (b) shows that the film remains on the substrate until it reaches the edges of the track, the film breaks down and no longer remains on the track. Referring to figure 8 (a), it shows the cross section at the middle of the wear track, which exhibits little plastic deformation with a magnification of 450. The same spot (figure 8b) is then magnified to 6500 times. It shows the original carbon coating is still intact after the wear
Correlation between adhesion and wear behaviour of commercial carbon based coating
243
test. When compared to the deformation on the scratch track at 30N (Fig. 6 (a) & 8 (a)), the deformation of the coating in the scratch track is larger than wear track. This is because in the scratch test, the indenter applies higher Hertzian contact stress on the sample than in the wear test. Hertzian contact stress can be calculated by equation 2:
Po
3P 27~/2
(2)
where P0 is peak contact pressure (GPa) used as the stress acting on the specimen, P is the indenter load (N), a is the radius of the circle of contact (m) which is calculated by equation 3:
I
(3)
R is the indenter radius (m) and the E is the Young modulus of the materials (GPa). As the indenter size for the scratch test (R - 200#m) is much smaller than that for the wear test (R = 2.5 mm), the calculated Hertzian stress applied on the coating in scratch tests (Using the coatings on non-hardened steel as an example.) is 16.13 GPa, while it is only 2.99 GPa in wear tests. The deformation will increase with the Hertzian contact stress, therefore, deformation of the coating and substrate in the scratch test is larger.
(Continue to the next page)
244
Fig. 8.
Scratching of materials and applications
Cross section of wear track (a) whole track (b) part of the track (Sample: carbon
based coating on non-hardened steel and 3 minutes pre-sputtering time; Normal load = 30 N, Linear displacement = 2 mm, Linear velocity = 300 mm/min, Number of cycle = 500)
Wear rate against critical load: Using the above results, the relationship between wear and adhesion can be estimated. Figure 9 (a) shows their correlation which includes all the results from hardened and non-hardened steel substrates. It clearly shows that the wear rate decreases when the adhesion strength increases. The wear rate decreases from l l x l 0 15 to 0.15x10 15 m3/Nm as the critical load increases from 17 to 72 N. Figure 9 (a) can be considered as having three groups of data. Group A is the results obtained from the coatings using non-hardened steel as substrates; group B and C are results obtained from the coatings using hardened steel. From these three groups of data, we find that the substrate has larger effect on the coatings properties. Group A has much higher wear rate and lower critical load than group B and C (more than 40 N difference in critical load and almost one order of magnitude in wear rate). Furthermore, group B and C show the effect of the pre-sputtering time. The coatings using pre-sputtering time more than 35 minutes (group C) have better adhesion and wear resistance than using less than 35 minutes (group B). However this effect is not obvious in group A. The details of each group are shown in figures 9 (b) and (c). Figure 9 (b) shows the wear rate decreases from 11 x 1015 to 8.5x 1015 m3/Nm gradually when the critical load increases from 17 to 24 N. In the case using hardened steel substrate (Fig. 9 (c)), the effect ofpre-sputtering time is more obvious. The critical load increases from 53 to 74 N. Also, the data can be further separated into two groups (group B and C in Fig. 9 (a)). A definite transition of wear rate is found from about 2.5• 1015 to less than 0.2x 1015 m3/Nm at critical load around 62.5 N. However, it can be seen that the wear rate remains constant and independent of the critical load in group B and group C. It shows that if the pre-sputtering time
Correlation between adhesion and wear behaviour of commercial carbon based coating
245
is less than 35 minutes, even increasing the time will not have significant effect on the interface between coating and substrate. When the sputtering time reaches 35 minutes, pre-sputtering on the substrate greatly enhances the strength of interfacial bonds between coating and substrate so that the coating has lower wear rate and higher critical load. After 35 minutes, as the substrate has reached its optimal clean level, further increase in pre-sputtering time will not improve the substrate surface quality much. Therefore, no correlation between wear and adhesion is found in group C also. In each group of data in figure 9 (a), the data also scatter to a large extent which has variation about 10 N in critical load. The correlation between wear and adhesion is poor if we focus on each group of data individually. As many factors will affect the testing results especially in wear tests, such as deformation and temperature, variation of the data is expected and acceptable. Therefore, if the range of the critical load and wear rate results in the experiment is not large enough, it is possible that the relationship cannot be demonstrated. As the change of the coatings or any other experimental details will affect the results, the range of the variation may be different in other cases. However, it shows that if the range of indenter load is too narrow (say 20N), a poor correlation will be found.
Scratching of materials and applications
246
i
12 '1~
•z'•
3
I
i
I
:..Group A
9
,
',
8 ......
.....
Orou
/
4 2
~
"',
|
:"~~
/
, I f \
l
',-: . . . .
C~ou0c =,,
,' ,
, ~
9:--- ',
,, . . . . .
0 1
(a)
3.0
10 z
~
z
8
E 2.0. "2
E b x
II19 "\,
~'2.5
E
9
9
"-.. \
6-
x 1.5 \
4-
~10.
rr
~2. 0
~0.5 i
17
.
|
18
(b)
Fig. 9.
.
I
19
.
i
,
20
u
21
Lo~(N)
,
:
22
9
!
23
9
u
24
",
,. ,
i
25
(C)
0.0 50
,
5'5
,
6'(3
6'5
7'0
7'5
t.c I{N)
Wear Rate obtained from wear tests against adhesion (Lc~) obtained from scratch tests
(a) all results (b) coatings on non-hardened steel substrate only (c) coatings on hardened steel substrate only
Correlation between adhesion and wear behaviour of commercial carbon based coating
247
To explain why wear resistance increases with adhesion strength, scratch and wear tracks pictures can be used. Figures 10 (a) & (b) show the wear tracks of the films with 1 hour and 3 minutes pre-sputtering time, respectively. When we compare the two figures, figure 10 (b) shows a few different failure modes on the wear track, such as a little spallation at the side of the track and some diagonal cracks in the middle, while figure 10 (a) has a clear wear track without any failure. Those failure modes can also be found in the scratch track, shown in figure 11 (a) & (b). Failures similar to wear track, such as spallation (Fig. 11 (a)) and crack (Fig. 11 (b)), are found, the only difference is that they exhibit more serious damage. Therefore, it can be concluded that the failures in wear are mainly caused by weak adhesion strength, which will accelerate the wear
Scratching of materials and applications
248
of the film. As the adhesion results show that the coating with 1 hour pre-sputtering time has better adhesion (see Fig. 3), the failure does not occur in figure 10 (a). In addition, failure due to weak adhesion will increase the amount of wear debris, which increases the chance of adhesive wear between the coatings and wear debris. In this paper, we found that the adhesion behaviour is related to the wear behaviour, as far as the commercial carbon based coating is concerned. However, further work should focus on different coatings and conditions for several reasons: 1) Different coatings have different microstructure and properties. 2) The substrate material used may affect the coating properties. Shima et. al. [ 18] had found that using different substrate material directly affected the coating wear behaviour because of the substrate elastic modulus, coating adhesion and local asperity deformation.
CONCLUSIONS The adhesion of the carbon based coatings was found to increase with the pre-sputtering time, while the wear rate will decrease. After combining these two results, the correlation between wear and adhesion was found; it showed that the wear rate will decrease when the adhesion increases. This relationship is better demonstrated over a large range of the critical load and wear rate. To apply this result to other cases, more work should be done, as only carbon based coating were used in this research.
ACKNOWLEDGEMENT KH Lau acknowledges the support of the research scholarship from the City University of Hong Kong. The authors appreciate the use of the facilities in Advanced Coatings Applied Research Laboratory (ACARL), which is supported by the Innovation Technology Fund of Hong Kong. In addition, a grant from the Research Grant Council of the HKSAR, China (Project No. CityU 1173/03E) also supports this work.
REFERENCES Larsson M., Olsson M., Hedenqvist P., Hogmark S. (2000) Surface Engineering, 16, no. 5, p 436-444. .
Harry E., Rouzaud A., Julieta E, Pauleau Y. (1999) Thin Solid Films, 342, no. 1-2, Mar, p
Correlation between adhesion and wear behaviour of commercial carbon based coating
249
207-213. PerryA. J. (1983) Thin solid films, 107, no. 2, p 167-180. Wang R. (1993) Materials Research Society Symposium Proceedings, 308, Thin Films: Stresses and Mechanical Properties IV, p 227-233. Staia M. H. Castillo EJ, Puchi ES, Lewis B, Hintermann HE. (1996) Surface and
Coatings Technology, 86-87, p 598-602. Bienk E. J., Reitz H., Mikkelsen N. J. (1995) Surface & Coatings Technology, 76-77, no. 1-3 pt 2, p 475-480. Schwarzer N., Richter F. (1995) Surface & Coatings Technology, 74, no. 1-3 pt 1, p 97-103. Chalker P. R., Bull S. J., Rickerby D. S. (1991) Materials Science & Engineering A:
Structural Materials: Properties, Microstructure and Processing, A140, no. 1-2, p 583-592. ~
Xie Y., Hawthorne H. M. (2001) Surface and Coatings Technology, 141, no. 1, p 15-25.
11.
Laugier M. T. (1987) J. Vac. Sci. Technol. A5 (1), p 67-69. Farhat Z. N., Ding Y., Alpas A. T., Northwood D. O. (1997) Journal of Materials
12.
Processing Technology, 63, no. 1-3, p 859-864 Bull S. J. (1999) Wear, 233-235, p 412-423
13.
Koski K., Holsa J., Emoult J., Rouzaud A. (1996) Surface & Coatings Technology, 80, no.
10.
1-2, Symposium H on Advanced Deposition Processes and Characterization of Protective
Coatings, p 195-199. 14.
Yang S., Camino D., Jones A. H. S., Teer D. Ct (2000) Surface and Coatings Technology, 124, no. 2, p 110-116.
15.
Ronkainen H., Vihersalo J., Varjus S., Zilliacus R., Ehrnsten U., Nenonen P. (1997)
16.
Surface & Coatings Technology, 90, no. 3, p 190-196. Eisenhuttenleute V. D., (editor) (1993) Steel- A handbook for materials research and engineering, p 35.
17.
Shum P. W., Zhou Z. F., Li K. Y. (2004) Wear, 256, no. 3-4, p 362-373.
18.
Shima M., Okado J., McColl I. R., Waterhouse R. B., Hasegawa T., Kasaya M. (1999)
Wear, 225-229, no. I, p 38-45.
250 CHAPTER 13
THE STUDY OF T H E ADHESION OF A TiN C O A T I N G ON STEEL AND TITANIUM A L L O Y SUBSTRATES USING A M U L T I - M O D E SCRATCH T E S T E R Originally published in Tribology hlternational vol 39, February 2006 J. STALLARD, S. POULAT and D.G. TEER
Teer Coatings Ltd, West Stone House, Berry Hill Industrial Est., Droitwich, Worts., WR9 9AS, U.K. E-mail:[email protected]
ABSTRACT A titanium nitride (TIN) coating was deposited by magnetron sputter ion plating onto steel and titanium alloy polished substrates. The adhesion of the coating on each substrate material was investigated using a newly developed multimode scratch tester. Progressive loading scratch tests, constant load scratch tests, multiple scratch tests in the same track and indentation tests were performed. It was shown that the modified scratch tester can be used to identify not only coating detachment during progressive load scratch tests, but also other failure events such as cracking and cohesive damage to the coatings. By using the additional modes of operation, it was possible to study the fracture mechanisms in more detail i.e. chipping in the scratch track was cohesive for the TiN coated steel and adhesive for the TiN coated Ti alloy. KEYWORDS Thin coating, Adhesion, Scratch, TiN, Titanium alloy
INTRODUCTION Ceramic coatings are the usual choice for increasing the wear lifetime of industrial components, of which TiN is the most widely accepted. The coatings possess different properties for each specific application but their overall performance is very dependent on the adhesion between the coating and the substrate material. Adhesion is measured as the force or the work required to detach a coating from the substrate [ 1]. Adhesion has been measured using various test methods for which the applications and limitations have been reviewed [2-4]. The Scratch Test is the most popular method for measuring adhesion because it is one of the few tests that can be simply and quickly used to assess relatively well-adhered surface coatings. The coatings studied by scratch testing span a wide range of applications from wear resistant coatings on cutting tools to optical coatings on glass. The scratch tester is now an essential tool for use in industry for quality control purposes or research laboratories for studying the mechanical strength of coatings on machine components. The conventional scratch test procedure involves drawing a diamond stylus across the coated surface under increasing load until adhesion failure is detected. The critical load (Lc) is defined as the load at which the coated film is removed from the substrate [5]. It is influenced by many factors such as substrate hardness, film thickness, interface bonding, and intrinsic properties of
The study of the adhesion of a tin coating on steel and titaniunl alloy substrates
251
the deposited film. Many authors [6-8] have reported that the critical load increases linearly with the substrate hardness. This general behaviour has been explained in terms of the increasing load beating capacity of the substrate as its hardness increases. The various test parameters (scratching velocity, stylus properties, etc.) and coating-substrate composite properties (hardness and surface roughness, etc.) all affect the critical load value [9-12]. Other failure events, such as cracking or cohesive failure are also equally important in determining the behaviour of the coating [13]. The critical load and any failure event can be detected and observed using friction force, acoustic emission (AE) and examination of the scratch track under optical microscopy and Scanning Electron Microscopy (SEM). In recent years, following the continuing improvement of coatings and their properties, numerous efforts have been made to improve available scratch testers. Updated instrumentation and extended operating capabilities have been added. The Teer Coatings ST3001 scratch tester was designed and developed as a computer controlled multi-mode scratch tester to incorporate these new developments. This work was completed as part of the European project entitled "Multimode scratch testing (MMST): Extension of operation modes and update of instrumentation" [ 14]. To illustrate how these new developments can be used for quality control or research activities a testing programme was implemented: the aim was to identify the failure mechanisms of a PVD TiN coating deposited onto two different substrate materials.
EXPERIMENTAL DETAILS Substrate preparation ASP23 powder metallurgy steel and TA 46 titanium alloy (032 mm and height 6mm), ground and polished to a surface roughness of Ra = 0.02 lam were used as substrate materials for the coating. The substrates were placed in an ultrasonic bath in acetone for 15 minutes and then dried in hot air to remove the residual solvent. The substrates were then placed on a precleaned sample holder for coating. A TiN coating was deposited in an industrial closed field unbalanced magnetron sputter ion-plating (CFUBMSIP) system with titanium targets. The substrates were Ar plasma ion cleaned using pulsed DC bias prior to deposition and a thin, approximately 0.1 ~tm, adhesion promoting Ti layer was first deposited by DC magnetron sputtering, again with a pulsed DC bias on the substrates. Nitrogen, under controlled flow conditions, was introduced into the chamber after deposition of the initial Ti interlayer to produce the stoichiometric TiN coating, the reactive deposition conditions being maintained via an automatic feedback control optical emission monitoring system. Characterisation of the films A standard hardness tester (Wilson / Rockwell B503-R) using a 150 kgf load was also used to assess the adhesion of the coatings, using the HF1 to HF6 scale [15]. The microhardness was measured using a Fischerscope H100 ultra-microhardness tester with a load of 50 mN. For indentation depths of more than 10% of the coating thickness a composite hardness value was obtained [ 16, 17]. Coating thickness was assessed using the ball crater taper-section technique [ 18,19]. Optical microscopy was used to examine and measure the coatings. Progressive load scratch test procedure Adhesion was measured using a Teer Coatings ST3001 scratch tester with a 0.2 mm tip radius Rockwell diamond indenter. The diamond tip was drawn across the coatings with a loading rate of 100 N min ~ and a sliding speed of 10 mm min ~. The increasing load scratch test method used has been accepted as a European standard [8]. A start load of 5 N was used in order to identify the start of the scratch track and the test was stopped after a dramatic increase in friction occurred, which corresponds to the substrate being exposed. Using optical microscopy
Scratching of materials and applications
252
examination at x200 magnification the critical loads Lcl, Lc2, Lc3 and Lc4 were assessed. During each test a computer recorded the normal load, friction force and AE signals.
Constant load scratch test procedure To assess the homogeneity of the coating along the coating surface and to test the repeatability, constant load scratch tests were completed. A 0.2 mm tip radius Rockwell diamond indenter was loaded to the critical load Lc2 determined during the progressive load scratch test and was then drawn across the coating surface with a sliding speed of 10 mm min -1. The number of failure events along the length of the scratch track was then analysed. During each test a computer recorded the normal load, friction force and AE signals.
Multiple scratch test procedure To assess the wear resistance of the coating, uni-directional multiple scratch tests in the same track were performed. A 0.2 mm tip radius Rockwell diamond indenter was loaded to the critical load Lc2 value determined during the progressive load scratch test and was then drawn across the coating surface with a sliding speed of 10 mm min l . During each test and for each pass a computer recorded the normal load, friction force and AE signals.
Indentation procedure To assess the fracture mode of the coatings, some medium load indentations were completed. A 0.2 mm tip radius Rockwell diamond indenter was loaded from 5N to a final load value, which was chosen following the analysis of the progressive load scratch test data. The loading rate was 100 N min l . The final load was then maintained for a selected duration and, finally, it was removed with an unloading rate of 100 N min -~. Different final loads were selected. During each test, for the three stages, loading, pause and unloading, a computer recorded the AE signal.
RESULTS The results of the thickness, hardness and Rockwell indents are shown in Table 1. Due to the substrate's influence on the hardness result the TiN coating on the Ti alloy substrate had a composite hardness lower than for the coating on the steel substrate. An example of a crater taper section used to measure the thickness of the TiN coating on the steel sample is shown in Fig. 1 (a). A comparison of Rockwell indents on the coated steel and Ti alloy is shown in Fig. 1 (b) and (c). No coating failure was observed for the Rockwell indent on the coated steel sample (HF1) but cracking and small edge chips were observed for the coating on the Ti alloy (HF3), this was expected as the Ti alloy was a softer substrate and would produce more deformation under load.
Table 1: Thickness, hardness and Rockwell adhesion of the coatings Substrate
Steel Titanium alloy
Total Thickness / ~tm 2.0 2.0
Composite Hardness / GPa 35.3 34.1
Rockwell C Indent (HF1 to 6) 1 3
The study of the adhesion of a tin coating on steel and titanium alloy substrates
253
Fig. 1. (a) Ball crater on TiN coating on steel, (b) Rockwell indent on TiN coating on steel and (c) Rockwell indent on TiN coating on Ti alloy.
Progressive load scratch test results
Three progressive load scratch tests were performed on each sample and observation of the scratch tests showed that there were four main failure events, which were identified and labelled as I-cl, Lc2, I-,c3 and Lc4. Micrographs and a description of the failure events corresponding to each critical load are shown in Fig. 2 and Table 2 respectively. The progression of the scratch is shown from lett to fight.
Fig. 2. Micrographs of failure events used in scratch testing.
Table 2: Description of failure events used in scratch testing Critical load Lcl Lcz Lc3 Lc4
Description of failure event Semicircular coating cracks inside the scratch track " Adhesiv e chipping at irack edges Initial failure of coating Total faiiure of coating (substrate completely exposed)
The scratch results for each coated sample are shown in Table 3. The TiN coating on the Ti alloy had lower critical load failures than the same coating on the steel substrate. As previously stated, many authors have shown that the critical load in a scratch test increases with increasing substrate hardness [6-8]. They suggest that when the substrate reaches a critical deformation under the effects of the scratch indenter the coating failure in the scratch test occurs. The failure modes Lcl and Lc2 are clearly identified with the increase in the AE signal; the failure modes
254
Scratching of materials and applications
Lc3 and Lc4 correspond to the change in friction force and the peaks in the first derivative of the friction force as depicted in the graph shown in Fig. 3.
Table 3: Failure mode results for the progressive load scratch tests Substrate Steel
Scratch No. 1
2 3 Ti ailoy
1
2 3
Failure modes (N) Lcl LC2 Lc3 Lc4 76 35 41 68 75 35 40 68 36 42 67 72 36 15 15 17 36 9 9 21 10 10 17 38 .,
. .
, ,
Fig. 3. Graph of friction force, first derivative of the friction force and AE for a progressive load scratch test on a TiN coated steel substrate, plotted versus the applied load (N).
The observed AE during the scratch test corresponds to the released elastic energy generated by the propagation of cracks during scratching. For the coating on the Ti alloy, during the scratch test Lcz semicircular coating cracks due to buckling inside the scratch track were accompanied by 1-<:2 edge chipping, as shown in Fig. 4. The comparison of the results gained by optical evaluation, frictional force and AE measurement showed that the failure modes for the same
The study of the adhesion of a tin coating on steel and titanium alloy substrates
255
TiN coating occurred at different loads for the two substrate materials. However, measurements of the residual track width at the total failure point identified that the failures happened at a similar residual track width and therefore indent depth (Fig. 5).
Fig. 4. Progressive load scratch track for TiN coating on Ti alloy showing combined Lcl and Lc2 failure.
Fig. 5. Total failure in progressive load scratch tracks for the TiN coated (a) steel and (b) Ti alloy.
A repeat of a progressive load scratch test was performed on each of the coated substrates until an AE event of amplitude greater than 75 dB was recorded. At this point the test automatically stopped so that the failure mode corresponding to the AE event could be identified. Figure 6 shows the graph of the failure event for the coating on the Ti alloy. The first failure event occurred under a load of 15 N, as found previously. Figure 7 shows that the first failure event of the TiN coating is cohesive (Lcl) on the steel substrate and that it is adhesive (Lc2) on the Ti alloy substrate. Both events occurred at similar residual track widths of-- 85 jam. Constant load scratch tests were performed on the coated steel and Ti alloy substrates under the loads at which each first failure occurred to observe more of the failure event.
256
Scratching of materials and applications
Fig. 6. Graph showing the first AE event greater than 75 dB stopping the test for the TiN coated Ti alloy.
Fig. 7. Failure mode corresponding to first AE event greater than 75 dB for progressive load scratch tracks on the TiN coated (a) steel and (b) Ti alloy.
The study of the adhesion of a tin coating on steel and titanium alloy substrates
257
Constant load scratch test results To confirm the nature of the failure events, constant load tests were performed under loads o f 40N and 15 N on the coated steel and Ti alloy substrates respectively. These loads corresponded to a residual track width of-- 85 lam. At this track width, cracks were visible for both coated substrates (Fig. 8). However, the cracks were more severe and accompanied more often by adhesive chipping events for the coated Ti alloy. Both of the coated substrates show an initial cracking failure but the coating on the Ti alloy shows that the stress caused by the cracking is enough to remove the coating from the substrate rather than a cohesive failure observed for the coated steel substrate.
Fig. 8. Cracking and chipping for a scratch track width of --- 85 lam on a TiN coated (a) steel and (b) Ti alloy substrate.
Multiple scratch test results Constant load scratches were made repeatedly over the same track at a load of 40 N and 15 N on the coated steel and Ti alloy substrates respectively. Ten passes were completed for each coated substrate material, subsequent scratch tracks are shown in Fig. 9. Both tracks show that, as well as edge flaking and chipping, adhesive failure occurred. These adhesive type failures were very similar to the initial failure of the coating defined as Lc3 during the progressive load scratch tests. These failures occurred at a much higher load during the progressive load scratch tests. Exposure of the substrate in the track produced a high friction coefficient for the system. Figure 10 shows the progression of the friction force along the scratch track for each pass. During the repeated scratches the coefficient of friction generally decreased with increasing number of passes. When the substrate was exposed in the track a local maximum of the friction coefficient occurred. More exposure of the substrate was observed for the coated Ti alloy than for the coated steel; this is clearly presented in the 3D friction maps shown in Fig. 10. It seems that adhesive failure occurred after the third and first pass for the coated steel and Ti alloy substrate respectively. This shows that the 3D map technique is a very useful tool for identifying the amount of failure produced during a multiple-scratch test, and would also be useful for monitoring a progressive wear process.
258
Scratching of materials and applications
Fig. 9. Scratch tracks after 10 passes for the (a) coated steel under a load of 40 N and (b) coated Ti alloy under a load of 15 N.
Fig. 10. 3D friction coefficient map of 10 constant load scratches over the same track for the (a) TiN coated steel under a load of 40 N and (b) TiN coated Ti alloy under a load of 15 N.
Indentation results Indentations at loads of 15 N, 40 N and 100 N were carried out on the coated Ti alloy. These loads were used to compare the failure modes around the indent with those seen for the progressive load and constant load scratch tests. After a load of 15 N no failure cracks were observed around the indent, although this indent had the same width as the scratch tracks at the same load. Only very fine cracks were observed around the indent after a load of 40 N, which was the same load and penetration depth at which the coating failed for the progressive load scratch tests. Even an indent produced after a load of 100 N showed only very slight cracking within the indent and at its edge, and no chipping or exposure of substrate (Fig. 11). A graph of load and AE versus time shows that under the load of 100 N the highest AE signal event happened at a relatively low value of 50 dB (Fig. 12).
The study of the adhesion of a tin coating on steel and titanium alloy substrates
259
Fig. 11. An indent produced after a load of 100 N on a TiN coated Ti alloy showing (a) cracks inside the indent and (b) cracks at the indent edge.
Fig. 12. Graph of load and AE versus time for a 100 N load indent on TiN coated Ti alloy substrate.
260
Scratching of materials and applications
DISCUSSION A TiN coating was deposited onto a steel and a Ti alloy substrate in the same coating run. Each substrate had similar surface polished finish and the same coating thickness applied. Hardness measurements showed that the Ti alloy substrate had a greater influence on the composite hardness than the steel substrate, which is not surprising given the comparative hardness of the two materials. Also, a Rockwell indent performed on each sample showed slight chipping around the indent on the coated Ti alloy substrate. Each coated substrate was tested for adhesion using a conventional progressive load scratch test under standard conditions. For each substrate it was observed that the coating failure events were of a similar type. It was also noted, firstly, that the friction coefficient recorded just before the failure point was higher for the Ti alloy than for the steel, secondly, the AE showed that the failure event occurred while the sample was under load and not during the elastic recovery of the coating / substrate system and, finally, that the coating on the softer Ti alloy substrate failed under a lower load than the coating on the steel substrate. The latter observations could be due to the fact that the shear forces applied on the coating are more important in the case of soft substrates, as shown by the higher recorded friction force. This could in turn be explained by the fact that the ploughing component of the friction force is more important in the case of soft substrates. Even if the failure modes were of similar types, stopping a progressive load scratch test at the first significant AE event identified that, on the steel substrate, the coating showed cohesive failure and, on the Ti alloy substrate, the failure was adhesive. Constant load tests were performed on each coated substrate at the same load for which the first failure event had been previously seen (15 N and 40 N for the Ti alloy and steel substrate respectively), so that more of the scratch track could be analysed. For the coating on the Ti alloy considerable cracking and adhesive failure was observed, but for the steel substrate only slight cracking and cohesive failure was observed. Multiple constant load scratch tests, performed at 15 N and 40 N for the Ti alloy and steel substrate respectively, showed that, after 10 passes, the coating failed in the track exposing the substrate. The Lc3 type failure of the coating happened at a lower load than previously seen for the progressive load scratch tests. This could be explained by the fact that the coating was weakened during the first few passes by the occurrence of cracking events, as shown during the constant load scratch tests. As chipping can originate from a surface or an internal crack and these cracks can appear at very low loads it could explain how it was possible to observe Lc3 type adhesive failure of the coating during the multipass test, even when it was carried out at a load lower than the Lc3 critical load determined during progressive load scratch test. Indentations in the coating on the Ti alloy substrate showed that at a load of 15 N the test did not show any cracking inside or around the subsequent indent. Indents were also performed at loads of 40 N and 100 N. Only the 100 N load indent showed slight cracking inside and at the edge of the indent. As it was possible to observe cracking events at loads lower than 15 N during progressive load scratch test, it can be concluded that the failure events are not only due to the deformation of the substrate but also to the shear forces applied to the coating / substrate system.
CONCLUSIONS It has been shown that, even if the residual indent depth is similar for scratches on the TiN coated Ti alloy substrate and the TiN coated steel substrate, more failure events can be observed in the case of scratches on the Ti alloy substrate. This may be due to the interface bond strength of the deposited Ti layer being less important with the Ti alloy than with the steel. It could also
The study of the adhesion of a tin coating on steel and titanium alloy substrates
261
be due to the fact that the elastic recovery for the Ti alloy substrate is more important than for the steel and so the actual indent depth in the loaded state was more important on the Ti alloy coated sample than on the steel coated sample, hence generating a higher ploughing effect. It has also been shown that adhesive failure events appear at lower load for the TiN coated Ti alloy. This could be explained by the fact that, because of the greater deformation observed in the case of the Ti alloy substrate, the shear forces are applied at the interface between the coating and the substrate.
REFERENCES l.
2. 3. 4. 5.
10. 11. 12. 13. 14.
15. 16. 17. 18. 19.
Chapman, B. N. (1974) J. Vac. Sci. Technol. 11, 106 Chalker, P. R., Bull, S. J. and Rickerby, D. S. (1991) Mater Sci. Eng., A, (40), 583 Jacobson, R. (1976) Thin Solid Films, 34, 191 Valli, J. (1986) J. Vac. Sci. Tchnol., A 4 (6), 3007 Hedenquist, P., Olsson, M., Jacobson, S. and Soderberg, S. (1990) Surf. Coat. Technol., 41,31 Perry, A. J., (1983) Thin SolidFilms 107, 167. Steinmann, P. A. and Hintermann, H. E. (1985) J. Vac. Sci. Technol., A3, 2394 Ichimura, H. and Rodrigo, A. (2000) Surf. Coat. Technol., 126, 152 Bumett, P. J. and Rickerby, D. S. (1988) Thin Solid Films 157, 233 Steinmann, P. A., Tardy, Y. and Hintermann, H. E. (1987) Thin Solid Films 154, 333 Bull, S. J., Rickerby, D. S., Matthews, A., Leyland, A., Pace, A. R. and Valli, J. (1988) Surf. Coat. Technol. 36, 503 Valli, J. and M/ikel/i, U. (1987) Wear 115, 215 Jindal, P. C., Quinto, D. T. and Wolfe, G. J. (1987) Thin SolidFilms 154, 361 European commission- Standards, Measurements and Testing Programme, Project 'Multi-mode Scratch Testing (MMST): Extension of Operation Modes and Update of Instrumentation', contract SMT4-CT97/2150 Heike, W., Leyland, A., Matthews, A., Berg, G., Friedrich, C. and Broszeit, E. (1995) Thin Solid Films 270, 431 Antunes, J. M., Cavaleiro, A., Menezes, L. F., Simoes, M. I. and Femandes, J. V. (2002) Surface and Coatings Tech. 149, 27 Saha, R. and Nix, W. D. (2002) Acta Materialia 50, 23 ASTM standard E 1182-93 Walls, J. M., Hall, D. D. and Sykes, D. E. (1979) Surface and Interface Analysis, 1 (6), 204.
262
C H A P T E R 14
STUDIES ON FRICTION AND TRANSFER LAYER USING INCLINED SCRATCH Originally published in TribologyInternationalvol 39. February 2006 PRADEEP L. MENEZES, KISHORE
Department of Metallurgy & SATISH V. KAILAS
Department of Mechanical Engineering lndian Institute of Science Bangalore 560 012 INDIA. E-mail: satvk@mecheng, iisc.ernet.in
ABSTRACT Friction influences the nature of transfer layer formed at the interface between die and sheet during forming. In the present investigation, basic studies were conducted using Inclined Scratch Tester to understand the effect of surface topography on friction and transfer layer formation. A tribological couple made of an A1-Mg alloy pin against steel fiat was used in the tests. Tests were conducted at a sliding speed of 2 mrn/sec in ambient conditions under both dry and lubricated conditions. The inclination angle of the steel fiat was 1.0 • 0.10. Normal loads varied from 0 to 135 N during the scratch test. Two surface parameters of steel fiats- roughness and t e x t u r e - were varied in tests. Care was taken to ensure that the surface roughness, measured along the scratch direction, had similar values for different textures, namely Unidirectional 8-ground, and Random. Grinding the EN8 fiat in a uni-directional manner and a criss-cross manner on emery sheets produced the uni-directional and 8-ground surfaces. While the random surfaces were produced by polishing the EN8 fiats using various abrasive powders. Scanning Electron Microscopy was used to reveal the pin damage and morphology of the transfer layer formed on flats. It is observed that the coefficient of friction, and the transfer layer formation, is controlled by the nature of surfaces and is independent of surface roughness. Moreover, the coefficient of friction, which has two components - the adhesion and plowing, is controlled by the nature of surfaces. The plowing component of friction was highest for the surface that promotes plane strain conditions near the surface and was lowest for the surface that promotes plane stress conditions near the surface.
KEYWORDS Coefficient of Friction, Nature of surface, inclined scratch, surface roughness, surface topography, Transfer Layer
Studies on friction and transfer laver using inclined scratch
263
INTRODUCTION The knowledge of various parameters, which control the friction forces are important in metal working operations. Friction is one such important parameter which controls the tool load, product quality (geometry, tolerance and surface finish) and tool wear. The coefficient of friction, if controlled properly, could generate the required stresses to deform the metal to the required shape. It could also lead to fracture of the sheet if not controlled properly. Since the pioneering work by Bowden and Tabor [ 1] various aspects that contribute to the friction have been extensively studied [2-7]. Kim and Suh [8] studied frictional forces generated by plowing of surfaces, where they concluded that it is extremely difficult to eliminate micro structural changes caused by mechanical interactions between the surfaces at a microscopic scale. Lovell et al. [9] studied the variation of sliding friction as a function of normal load by sliding a hard pin on a sot~ surface. They found that the coefficient of friction increases with apparent contact pressure due to increased plowing effects. The initial rise in friction was found to be rapid, due to change from elastic to plastic contact, and then levels off once all the contacting asperities deform plastically. Many researchers have carried out experiments on friction to characterize the metal forming process and to study the effect of surface roughness on friction during metal forming. Bello and Walton [10] studied the interaction of surface roughness and lubrication at the tool-metal interface in sliding contact. In their experiment, strips of commercial pure aluminum were pulled through steel dies designed to give partial simulation of the conditions, which exist in the flange and die radius profile regions of the deep drawing process. They found that the conventional surface roughness parameters do not provide a satisfactory functional characterization of the surface in the context of the friction developed in sliding contact. Rasp and Wichern [11] studied the effect of surface topography on frictional resistance using different kinds of surfaces. In their experiment, the specimen surfaces were as received, mirror polished, chemically etched, abrasively scratched parallel and perpendicular to the simulated rolling direction. They found that the arithmetic roughness value and lubrication regime had greater influence than the directionality. The influence of surface topography of the sheet material on the frictional characteristics of 3104 A1 alloy sheet were investigated by Saha et al. [12] by stretching a strip around a cylindrical pin. They found that friction increased with the strain occurring during the contact, which supports the model relating friction to flattening of strip asperities and real area of contact. They also found that the coefficient of friction depends on the rolling direction of the strip. Schedin [13] conducted experiments using U-bending test and strip drawing test to study the formation of transfer layer during forming processes. His experimental conditions resemble the contact conditions in sheet metal forming, where a hard and smooth tool surface will make repeated contact with a soft and rough sheet surface. He concluded that it is impossible to completely avoid sheet metal transfer in sheet metal forming operation but the growth of the transfer layer could be controlled. But in all the above cases the test material is deformable, so the surface topography of deformable material cannot explain true friction values during metal forming, and thus it is important to have knowledge about the surface topography of harder material. Considerable amount of work has also been done to study the effect of surface topography of harder material on sot~er deformable material during sliding in sheet metal forming operations. Lakshmipathy and Sagar [ 14] studied the influence of die grinding marks directionality on friction in open die forging under lubricated conditions. They used commercial pure aluminium as the work piece material and H11 steel as the die material. Two sets of dies, one with unidirectional grinding marks and the other with criss-cross grinding marks, were used. It was found that, for the same
264
Scratching of materials and applications
percentage of deformation, the criss-cross ground dies required lesser forging loads compared with the die of uni-directionally ground ones. The friction factor was also lesser during the forging process when the die with the criss-cross surface pattern was used. They concluded that the lubrication breakdown tendency is more when pressing is done with unidirectional ground die than with criss-cross ground die. The relation between friction and surface topography using various lubricants was studied by Hu and Dean [15]. They found that a random smoother surface could retain more lubricant and reduce friction. Maatta et al. [ 16] studied the friction and adhesion of stainless steel strip against different tool steels. They concluded that the composition of the tool steel does not have a marked effect on the friction between the tool and the work piece. However, the surface roughness and topography of the tool have a marked effect, where polishing of the tool surface reduces the surface roughness which in-turn reduces the friction between the tool and the work piece. Xie and Williams [ 17] proposed a model in order to predict the value of the overall coefficient of friction and wear rate, when the sott surface slid against a rough harder surface. This model points to both friction and wear to depend essentially on the roughness characteristics of the harder surface, the mechanical properties of both surfaces, nominal contact pressure or load, and the state of lubrication. Malayappan and Narayanasamy [ 18] studied the bulging effect of aluminium solid cylinders by varying the frictional conditions at the interface between cylinder and fiat die surfaces. Different machining processes like grinding, milling, electro-spark machining, and lathe turning with emery finish were produced on the fiat dies to vary the surface roughness which in turn varied the frictional conditions. They concluded that the barreling depends on friction and thus surface finish. Many researchers used scratch test to study the effect of various parameters on coefficient of friction [19-21]. Liu et al. [22] carried out experimental and analytical study of plowing and friction for commercially available metals using a nano-indenter, where the indenter is used to make scratches on the surface of metals under different normal loads. They found that the hardness of the scratched surface dominates the plowing friction mechanism and the contribution of the plowing component to the total friction coefficient is predominant. Considerable amount of work has also been done by FEM and other computational methods to study the effect of various parameters on the contact between two surfaces [23-27]. The experimental and computational results are vast. Many fundamental issues have been reviewed, and new issues "discovered" [28-35]. Most of the experiments were based on variation in roughness values than the nature of surfaces, which cannot explain the true friction values during metal forming processes. The exact description of the contact surface and the nature of surfaces are important to understand the tribological system. The characterization of technical surfaces with traditional surface roughness parameter is insufficient to describe the tribological behavior. To mitigate this inadequacy, surfaces are to be simulated to those witnessed in actual conditions. Therefore, inclined scratch test is used here to study the effect of surface roughness and nature of surfaces on the coefficient of friction. The investigation is conducted by sliding an AI-Mg alloy pin on EN8 steel fiat. The surface morphology of the fiat is varied using various grinding and polishing methods. In the following sections, we present the experimental results and discuss the nature of contact and friction.
Studies on friction and transfer layer using inclined scratch
265
EXPERIMENTAL DETAILS Experiments were conducted using an inclined scratch-testing machine, the schematic of which is shown in Fig. 1.
Load cell Vertical slide EN8
Pin
Pin EN8 Flat slide
Holder ,% j,
Direction of Motion of EN8 Flat
Fig. 1. Schematic diagram of scratch testing machine and inclined sample.
Description of scratch tests The scratch-testing machine has a vertical slide and a horizontal slide, which are driven by stepper motors with step size of 2.5/am and 10/am respectively. A 2-D load cell is mounted on the vertical slide to measure both the normal and traction forces. A LVDT with a resolution of +1 /on mounted on the vertical slide was used to measure the angle of inclination of EN8 fiat. The pins were made of an AI-3.5%Mg alloy with a total Fe, Mn, and Si content of about 1 wt.%. The pins were 10 mm long, 3 mm in diameter with a tip radius of 1.5 mm. EN8 steel fiats were of dimensions 28 mm x 20 mm x 10 mm (thickness). The pins were first machined, and then were electro-polished to remove any work-hardened layer that might have formed during machining. The EN8 steel flat was ground against emery papers of grit size 220, 400, 600, 800 or 1000 to generate 2 kinds of surfaces with varying roughness. For the first surface, care was taken so that the grinding marks were unidirectional in nature. 8-ground surfaces were generated by moving the EN8 steel flat on the emery papers in a path having profile of the number "8" for about 500 times. The third kind of surfaces with random grinding marks were generated using a polishing wheel with abrasive medium as SiC powder (600 or 1000 grit), A1203 powder (0.017 microns), or diamond paste (1-3 microns). Figures 2 (a) and 2 (b) show 3-D profiles of surfaces generated by uni-directional grinding and random grinding, respectively. Before each experiment, the pins and EN8 steel flats were thoroughly cleaned in a soap solution and then in an ultrasonic cleaner with acetone. Then EN8 steel fiat and a pin were mounted on the horizontal slide and vertical slide respectively. Tests were performed to obtain five parallel scratches on the same fiat. It was observed that the initial sphere-on-flat contact essentially became a flat-on-flat type contact even before the end of the first scratch. But at the same time, it was observed that scratch width varies considerably in first three tests. Hence, all the results presented were of the fourth scratch. It was observed that the coefficient of friction did not vary much for all these five scratches. For the lubricated tests, a drop of commercially available engine lubricant was applied on the surface before the experiment. Both the dry and lubricated
266
Scratching of materials and applications
tests were done on the same flat so that the results of the dry and lubricated experiments will exclude variations during preparation of the flats.
Fig. 2. 3-Dimensional images of (a) uni-directional ground (b) randomly ground EN8 steel flat.
The procedure employed to measure the angle of inclination was as follows: The pin was moved towards the EN8 steel fiat surface slowly using the vertical slide. Once the pin touched the EN8 steel fiat (indicated by load cell registering an increase in normal load), it was stopped and the height was recorded. The pin was then retracted and the EN8 steel fiat was moved in the horizontal direction by a distance of 5 mm, and the pin was again made to touch the fiat surface. Knowing the difference in the heights at which the pin touched the EN8 steel fiat and the horizontal distance, the angle of the EN8 steel fiat could be accurately calculated. This angle was kept within 1~ + 0.1 ~ The waviness of the EN8 steel fiat was not considered, which in any case was low.
Studies on friction and transfer laver using inclined scratch
267
The EN8 steel fiat was moved to a position where a minimum scratch length of 10 mm could be obtained. The tester was programmed such that pin would be lowered till it touches the fiat, then the horizontal slide was moved at a speed of 2 mm/sec. The horizontal and vertical forces can be registered once the pin touches the surface, which increases as the pin slides on the EN8 steel fiat. The normal and tangential forces are continuously acquired using a personal computer. The coefficient of friction was calculated using formula given by equation (1). T N
-
Fnsin~176176
cosO- Fr
(I)
sin 0
where '0' is the angle of inclination of the EN8 steel flat, Fr is the recorded traction force and FN is the recorded normal force at any instant. The pins were slid at perpendicular and parallel to the uni-directional grinding marks generated by emery papers on EN8 steel fiat. For the 8-ground and randomly polished surface, the direction of sliding was not important. The profiles and roughness parameters of the EN8 steel fiat were measured with a standard contact-type profilometer in the direction of the sliding on the bare surface away from the scratches. Later, the pins and EN8 fiats were observed using a scanning electron microscope (SEM) to study the surface morphology.
RESULTS The average surface roughness (Ra) values of all EN8 steel flats were measured for all four kinds of surfaces and are presented in table 1. It can be seen that surface roughness values for different textured surfaces are comparable with each other when they are ground against same grinding media. Figure 3 (a) shows a typical variation of the normal load and tangential load obtained in the inclined scratch experiments. The recording is for an experiment carried out under dry conditions with an EN8 fiat with uni-directional grinding marks (Ra = 0.2359 lam) and the scratch direction being perpendicular to the uni-directional grinding marks. Figure 3(b) shows the variation of coefficient of friction with the sliding distance. It can be seen that the coefficient of friction does not vary much with the normal load, within the present load range in which the tests are conducted.
268
Scratching q[ materials and applications
Table 1" The surface roughness values for different textured surfaces. IUI,
Surface ni-directional Perpendicular
8-Ground
Uni-directional Parallel
Random
Grinding media Ra (microns) 220 grit emery paper 0.4816 400 grit emery paper 0.2754 600 grit emery paper 0.2359 800 grit emery paper 0.1823 1000 grit emery paper 0.1171 220 grit emery paper 0.3965 400 l~rit emery paper 0.2539 600 grit emery paper 0.1681 800 ~xit emery paper 0.1055 1000 grit emery paper 010911 220 grit emery paper 0.2028 400 grit emery paper 0.1993 600 grit emery paper 0.1685 800 grit emery paper 0.1167 1000 grit emery paper 0.0845 Diamond paste (1-3~tm) 0.0237 A1203 powder (0.017 ~tm) 0.1043 600 grit SiC powder 0.2253 1000 grit SiC powder 0.1993
[ [ I [ ]
....
160 140 120
Z
lOO Normal
co ~
80
/
60
" * " - " Tractional Force
40
20
0
2
4
6
8
10
12
Sliding Distance (mm) Fig. 3(a). Variation of forces with distance for A1-Mg alloy pin on EN8 steel flat microns). Scratch direction is perpendicular to uni-directional grinding direction.
(Ra =
0.2359
Figure 4 shows the variation of coefficient of friction for scratches conducted perpendicular to the uni-directional grinding marks with various roughnesses. Tests conducted under dry and lubricated experiments are shown on the same graph. It can be seen, for the dry experiments, that the coefficient of friction does not vary much with roughness (within the present range of
Studies on friction and transfer laver using inclined scratch
269
roughness) and the values crowd around a coefficient of friction value of 0.7. For the lubricated experiments the coefficient of friction drops to a value of around 0.3. Here too, the friction values are independent of the roughness value.
1.0 0.9 0.8
r 9-~
0.7
9~.
0.6 0.5
0.4
r
.~.
0.3
E
ao
0.2 [. 0.1 ~L 0.0 L 0
1
9
,
I
2
J_
,
4
.
.
.
1
.
6
. . . .
8
10
Sliding Distance (mm) Fig. 3 (b). Variation of coefficient of friction with distance for A1-Mg alloy pin on EN8 steel fiat. (Ra= 0.2359 microns).
1.0 ~ =
0.9 f 0.8
--o----<>--
r .~ o. 7 9r.
0.6
= .~.
0.4!
~ R =0.4816
~
R = 0.2359
--..a,--- R = 0.1823
a
R =0.2754 a
R =0.1171 Dry
/
a
0.5
~,
E~ ~ 0.2
Lubricated
0.1
I
0,0 t
0
.
~
2
_.,.
I
4
,
J
6
9
,
8
_
..
I
10
12
Sliding Distance (ram) Fig. 4. Variation of coefficient of friction with sliding distance for A1-Mg alloy pins on EN8 steel flat of different roughness (Ra in microns) under dry and lubricated conditions. Scratch direction is perpendicular to uni-directional grinding marks.
Figures 5, 6 and 7 show the variation of friction for the scratches carried out on the 8-ground EN8 fiats, parallel to the uni-directional grinding marks and for the randomly ground surfaces. In all the above experiments it can be seen that, for a given kind of surface, the coefficient of
Scratching of materials and applications
270
friction does not vary drastically with roughness and that the coefficient of friction, as expected, is lower for the lubricated experiments.
1 1110 / 0.9 !I-I" 0.8
=
R
r
R a = 0.1681
~
= 0.3965
---o---
R = 0.2539
--"
R a = 0.1055
"
----o---- R = 0 . 0 9 1 1
.6
. ~
0.5
~
0.4 0.3
~'~ 0.1 t 0.0 ~
Lubricated
'
0
'
'
'
2
'
4
'
'
'
6
'
J
'
10
8
12
Sliding Distance (mm) Fig. 5. Variation of coefficient of friction with sliding distance for AI-Mg alloy pins on 8ground EN8 steel fiat of different roughness (Ra in microns) under dry and lubricated conditions.
1.0 --
0.9
R
0.8
=0.2028 =0.1685
=
R
~
R
a
R
a
• 9~
a
-----o---- R
=0.1993 =0.1167
a
= 0.0845 a
0,7
"~, 0.6
Dry
~*~ 0.5 "~
0.4 0.3 I 9
_
~ -
-__
~
--
_
"~
0,2t
\
Lubricated
0.1 0.0
,
0
I
2
,
I
4
,
I
6
,
l
8
,
I
10
9
12
Sliding Distance (mm) Fig. 6. Variation of coefficient of friction with sliding distance for AI-Mg alloy pins on EN8 steel fiat of different roughness (R~ in microns) under dry and lubricated conditions. Scratch direction is parallel to uni-directional grinding marks.
271
Studies on friction and transfer laver using inclined scratch 1.0 0.9
= R = 0.0237 tl ------<>----Ra = 0.2253
0.8 o
= Ra = 0.1043 --" Ra = 0.1993
,~ 0.7 ;,r,, 0.6
r
Dry
0.5
0.3
i0.2 ted
0.1
0.0
,
I
o
i
2
I
,
4
l
,
I
6
I0
e
1
12
Sliding Distance (mm) Fig. 7. Variation of coefficient of friction with sliding distance for AI-Mg alloy pins on randomly ground EN8 steel fiat of different roughness (Ra in microns) under dry and lubricated conditions.
1.0 0.9
0.8
r .~0.7
9
Coefficient of Friction .......... Fractal Dimension ~ Roughness 9",'" Dry condition oo~oo 9 !
!
"~ 0.6
..L.
;,r., ~" r 0.5
2.0 1.9
.... i
0.8
1.8
.-~
i
0.7
_~ 1.7
:
:
r 0.4
~
1.0 0.9
9 0.6 ~~ r,
0.4
1.6 P~
u~ 1.4
0.3
r 0.2 0.1
0.0
i
I
,
U-PD
8-Ground
U-PL
Nature
o.1 ~
1.1
0.0
1.0
Random
of Surfaces
Fig. 8. Variation of coefficient of friction, surface roughness (Ra) and fractal dimension with nature of surfaces.
The range in which the coefficient of friction values fall for the various surfaces under both the dry and lubricated conditions are plotted in Fig. 8. The range of the roughness values and fractal dimension for each of the surfaces is also plotted in the figure. Many of the engineering surfaces are identified as fractal in nature and have a fractal dimension [36], a quantity that has been related to the friction and wear characteristics of surfaces. The fractal dimension is
272
Scratching of materials and applications
calculated using the method described by Hasegawa, et. al. [37]. The roughness value and fractal dimension range are seen to fall within a particular range for all the different surfaces. It can be seen that the coefficient of friction, for the dry experiments, is high for the unidirectional perpendicular experiments and reduces monotonically for the 8-ground, unidirectional parallel experiments and the random surfaces. A similar trend is seen for the lubricated experiments. A large amount of transfer layer is seen on the surface scratched in the direction perpendicular to the uni-directionally ground EN8 fiat for the dry experiment (Fig. 9a). Figure 9b shows the surface of the pin slid on the surface shown in Fig. 9a. The damage on the pin is extensive and considerable shear damage can be seen on the pin. The amount of transfer layer drops drastically when a lubricant is added on the surface (Fig. 9c). The damage observed on the pin also reduces drastically, though some scratches are still seen on the surface (Fig. 9d). A similar trend was observed for other EN8 steel fiats with different roughness values.
Fig. 9. Scanning Electron Micrographs of EN8 flat (Fig. (a)) and AI-Mg alloy pin (Fig. (b)) under dry condition (R~ = 0.2359 pan). Figures (c) and (d) show the Scanning Electron Micrographs of EN8 flat and AI-Mg alloy pin under lubricated condition (Ra = 0.2359 lam). Scratch direction is perpendicular to the uni-directional grinding marks. The arrow indicates the sliding direction.
Studies on friction and transfer laver using inclined scratch
273
The morphology of the transfer layer on the EN8 surface and the morphology of the pin for the random surface, which shows the least coefficient of friction for the lubricated condition, are shown in Fig. 10 (a-d). Here it can be seen that there is considerable amount of transfer layer on the surface under dry conditions and the damage observed on the pin surface is also severe. On addition of a drop of lubricant the amount of transferred layer on the EN8 flat is almost nil and the damage of the pin drops drastically. A similar trend was observed for other EN8 steel fiats with different roughness values. In addition, the damage to the pin was highest, when A1-Mg alloy pin slid on randomly ground EN8 fiat with lowest P-~value.
Fig. 10. Scanning Electron Micrographs of randomly ground EN8 flat (Fig. (a)) and AI-Mg alloy pin (Fig. (b)) under dry condition (Ra = 0.0237 ~tm). Figures (c) and (d) show the Scanning Electron Micrographs of randomly ground EN8 fiat and A1-Mg alloy pin under lubricated condition (Ra = 0.0237 ~tm). The arrow indicates the sliding direction. Figure 11 shows the scanning electron micrograph of randomly ground EN8 flat with R~=0.1993 microns. Comparing Fig. 9 (b), 10(b) and 11, it was observed that the damage to the pin was highest for randomly ground fiats under dry conditions when compared to other kind of surfaces of close-ranged roughness values.
Scratching of materials and applications
274
Fig. 11. Scanning Electron Micrographs of randomly ground EN8 flat under dry condition (Ra = 0.1993 Itm). The arrow indicates the sliding direction.
DISCUSSION Comparing the coefficient of friction data (Fig. 8), scanning electron micrographs of EN8 flats (Fig. 9 (a) and (c) and Fig. 10 (a) and (c)) and the scanning electron micrographs of the pins (Fig. 9 (b) and (d) and Fig. 10 (b) and (d) and Fig. 11) the following points and observations can be m a d e 1) The coefficient of friction for both the dry and lubricated scratch carried out perpendicular on the uni-directionally ground EN 8 flats is the highest. 2) The coefficient of friction is lowest for the experiments carried out on the randomly ground flats. 3) The material transfer is high for the dry experiments for all flats tested. 4) The material transfer is low under lubricated conditions for all the flats tested. 5) The damage on the pin is high for all the dry experiments and reduces drastically for the lubricated experiments. The reduction in damage is highest for the randomly ground surface. From the fundamental work done on friction by Bowden and Tabor [1] the coefficient of friction, It, can be considered to have two components, the adhesion component, ~ , and the plowing component, l.t0, such that;
}.ta, the adhesion component, depends on the material pair and also on the real area of contact, while ~ , the plowing component, depends on the "degree" of plastic deformation taking place at the asperity level. The adhesion component of friction, It,, can be minimized, if not eliminated, by the addition of a lubricant between the contacting surfaces. This would be true in a lubricant where an "extreme pressure additive" is added. The lubricant used in the present case is commercially available engine oil that is expected to have an extreme pressure additive such as ZDDP. Further, the speed of the present set of experiments, which is 2mm/sec, would ensure that hydrodynamic lubrication is absent. Thus, one can say that the coefficient of friction recorded for the lubricated experiments (Fig. 8) would be in the boundary lubrication region and is basically the plowing component of friction. It can be seen that the plowing component of friction is highest for the scratch done perpendicular to the uni-directionally ground EN8 fiat
Studies on friction and transfer layer using inclined scratch
275
and reduces monotonically for the 8 ground, uni-directional parallel scratches and the randomly ground EN8 surface. The figure also clearly shows that the adhesion component of friction, which is the difference between the coefficient of friction for the dry experiments and the lubricated experiments, is more or less constant but least for the randomly ground EN8 fiats. When a soft material (AI-Mg alloy) is pressed and slid on a harder surface (EN8 steel) the sorer material will deform plastically to follow the contour of the harder material. The higher the normal pressure the closer would the sorer material follow the contour of the surface of the harder material. For the sliding experiments carried out perpendicular to the uni-directionally ground EN8 fiats, which has a wave like surface, the sorer aluminium alloy material will have to climb over the asperities. But, for the randomly ground surfaces, which has a hill and valley surface the softer aluminium alloy material could flow around the asperities. Only the material that comes directly in the central path of the asperity flows over the asperity. A schematic of the possible flow pattern is shown in Fig. 12.
More Plane Strain Condition
More Plane Stress Condition
Fig. 12. Schematic of flow pattern of a soft material over a hard cylinder and a sphere.
Flow over the uni-directionally ground EN8 flats is shown as flow over cylindrical asperities and flow over the randomly ground EN8 fiat is shown as flow over spherical asperities. The flow pattern for sliding experiments carried out perpendicular to the uni-directionally ground EN8 fiats induces a flow pattern that is more constrained and a state of stress that is more towards plane strain conditions. For the randomly ground EN8 fiats the flow pattern is more unconstrained and has a state of stress that is more towards plane stress conditions. The flow pattern for the sliding experiments carried out on the 8-ground EN8 flats and parallel to the unidirectionally ground EN8 fiats is, then, expected to fall in between these two extremes. This is because the constraint for flow induced by the 8-ground surface would be lower than the unidirectional perpendicular experiments and would further reduce for the uni-directional parallel experiments. The results shown in Fig. 8 confirm this trend. For a given Ra value, it can be expected that the cylindrical asperity height would be the same as that of the spherical asperity height. In such a situation the shear stresses ahead of the asperities
276
Scratching of materials and applications
while flowing over a cylindrical asperity (uni-directionally ground EN8 flats) would be much higher than the shear stresses generated ahead of the asperities while flowing around a spherical asperity (randomly ground EN8 fiat). This again means that one could expect a higher plowing component of coefficient of friction for the uni-directionally ground EN8 fiat, of similar R~, when compared to the randomly ground EN8 fiat. This is in agreement with the results of the present set of experiments. Further, the shear stresses generated for the dry experiments would be higher ahead of the asperities, which could lead to sheafing of the material ahead of the asperities. This would lead to collection of transfer layer ahead of the asperities. From Fig. 9a, which shows the transfer layer formed on the uni-directionally ground EN8 fiats, it can be clearly seen that the transfer layer forms ahead of the asperities. Here too, the results for the sliding experiments carried out on the 8-ground EN8 fiats and parallel to the uni-directionally ground EN8 fiats would fall between these two extremes. Adhesion component of friction was characterized by the difference between coefficient of friction for the experiments carried out under dry and lubricated conditions, respectively. It is seen that the adhesion component of friction is almost same for all the surfaces. Real area of contact is high when adhesion component is high. This would increase the damage of the pin. If it is the adhesion component that is causing the damage, a drop of lubricant on the surface should reduce the damage. This is because the adhesion component of friction would be minimized. This is in fact what is observed when comparing the damage on the pin (compare Figs. 9b, 9d and 10b, 10d). This is evident from the fact the amount of transfer layer on the EN8 fiat (Fig. 10 (a)) and damage on the pin (Fig. 10(b)) is high for the dry experiments. If, it is the adhesion component of friction that causes the large amount of transfer layer and damage of the pin one would then expect the transfer layer on the EN8 fiat and the damage on the pin to be low under lubricated conditions. In fact, the amount of transfer layer is almost non-existent for the lubricated EN8 flat (Fig. 10(c)) and the damage of the pin reduces drastically (Fig. 10(d)). Thus one can conclude that the plowing component of friction is controlled by the constraint to flow induced by the surface. The higher the constraint to flow the higher would be the plowing component of friction. Finally, though the roughness, as given by Ra, and fractal dimension does not have a relation to the coefficient of friction other hybrid parameters that define a surface might have a relation. This has to be checked and work is progressing in this direction. In the simulations done for sheet metal forming and other forming operations [24-28] the coefficient of friction is taken to be a constant or can be set at different values at various locations. The criterion for choosing different values of coefficient of friction at different locations is arbitrary or based on intuition. The present results give a better basis for choosing the coefficient of friction at various locations. In fact, the results can be used to choose a particular finish at different locations of the die so that the coefficient of friction could be varied according to the requirement. In locations where the coefficient of friction need to be high a uni-directionally ground surface with the flow perpendicular to the grinding marks can be machined and in locations where the coefficient of friction need to be low a randomly ground surface can be machined. Previous work carried out on dies with various surface roughness [913] did not differentiate between the surfaces that offer various levels of constraint to flow. The work conducted by Lakshmipathy and Sagar [14] clearly showed that the friction factor, as estimated by ring tests, was lower for the die with a criss-cross surface when compared to a die with uni-directional surface. They, however, did not differentiate the role of constraint to flow which results in the reduction in friction factor. The control on the coefficient of friction under lubricated conditions could be based on this criterion but more work needs to be done to understand the exact reasons that control the adhesion component of friction. The role of adhesion has been summarized by Urbakh et al. [35] at much lower loads and speeds
Studies on friction and transfer laver using inclined scratch
277
encountered in forming operations and could form the basis for further study. This is important because in forming operations such as extrusion, a lubricant will not be present in the beating channel, but, in other forming operations for example, sheet metal forming, a lubricant is expected be present between the die and the sheet being formed.
CONCLUSIONS In the present work an inclined scratch is used to study the process of transfer layer formation while sliding an A1-3.5Mg alloy pin on EN8 fiats. The usefulness of this test is that from a single experiment, the effect of load on the coefficient of friction and transfer layer could be studied. The conclusions based on the experimental results are as follows 1) The normal load up to the test limit of 135 N does not have any effect on the coefficient of friction, for a particular "kind of surface" against the A1-Mg pin. The coefficient of friction is constant within the present test range. 2) The coefficient of friction is much higher, under lubricated conditions, for the unidirectionally ground flats with scratch direction perpendicular to uni-directional grinding marks than for the randomly ground flats. The results of the 8-ground EN8 flats and for the experiments carried out parallel to the uni-directionally parallel flats fall in between these two extremes. 3) The higher coefficient of friction for the uni-directionally ground flats with scratch direction perpendicular to uni-directional grinding marks is attributed to the constrained nature of flow for the sof~ material. This constrained flow of the sot~ material induces a higher degree of plane strain conditions to exist near the surface, which increases the plowing component of friction. 4) For the randomly ground flat, the coefficient of friction under lubricated conditions is lower as the flow is unconstrained. This promotes a stress condition closer to plane stress near the surface. This reduces the plowing component of friction. 5) For the randomly ground EN8 flat, damage to the pin is severe and even more severe when the flat surface is the smoothest under dry conditions. 6) The coefficient of friction for the 8-ground flats and for experiments carried out parallel to the uni-directionally ground flats fall in between the uni-directional perpendicular and the randomly ground flats. This is because the constraint to flow offered by the 8ground and uni-directional parallel flats is expected to fall in between the uni-directional perpendicular surface and the random surface. 7) The coefficient of friction is found to be independent of surface roughness (as given by Ra) and the fractal dimension in the present test range and depends on "nature of surfaces." Finally, the coefficient of friction and the amount of soft material transferred onto a hard surface at "ambient temperatures" is dictated primarily by the degree of plane strain conditions existing near the surface.
278
Scratching of materials and applications
ACKNOWLEDGEMENTS The authors would like to thank General Motors R&D Center, Warren. Michigan, USA for funding the work. One of the author, P. L. Menezes, would also like to thank GM R&D group for the scholarship during the initial stages of the work. The authors would also like to thank Mr. H. S. Shamasundar, Department of Mechanical Engineering, Mr. K. R. Kannan, Senior Scientific Officer, and Mr. K. Sathyanarayana, Technical Assistant, Materials Research Centre and Mr. Gurulinga, Technical Assistant, Department of Metallurgy, Indian Institute of Science, Bangalore, for their help.
REFERENCES
.
3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.
Bowden, F. P. and Tabor, D. (1954). The Friction and Lubrication of Solids. Clarendon, Oxford. Hehn, A. H. and Kimzey, J. H. (1968). d'. Lubr. Eng. 274. Sargent, L. B. (1977). ASLE Trans. 21,285. Suh, N. P. and Saka, N. (1987). Ann. CIRP. 36, 403. Suh, N. P. (1986). Tribophysics, Prentice-Hall, NJ. Moeller, C. E. and Noland, M. C. (1967). ASLE Trans. 10, 146. Komvopoulos, K., Saka, N. and Suh, N. P. (1986). J. Tribology 108, 313. Kim, D. E. and Suh, N. P. (1991). Wear 149, 199. Lovell, M. R., Deng, Z. and Khonsari, M. M. (2000). Transactions of ASME 122, 856. Bello, D. O. and Walton, S. (1987). Tribology International 20, 59. Rasp, W. and Wichem, C. M. (2002). Journal of Materials Processing Technology 125126, 379. Saha, P. K., William, R. D., Wilson and Timsit, R. S. (1996). Wear 197, 123. Schedin, E. (1994). Wear 179, 123. Lakshmipathy, R. and Sagar, R. (1992). International Journal of Machine Tools and Manufacture 32, 685. Hu, Z. M. and Dean, T. A. (2000). International Journal of Machine Tools and Manufacture 40, 1637. Maatta, A. Vuoristo, P. and Mantyla, T. (2001). Tribology International 34(11) 779. Xie, Y. and Williams, J. A. (1996). Wear 196, 21. Malayappan, S. and Narayanasamy, R. (2004). Int. ,1. Adv. Manuf Technol. 23, 636. Nieminen, I., Andersson, P. and Holmberg, K. (1989). Wear 130(1), 167. Xie, Y. and Hawthorne, H. M. (2000). Wear 240 (1-2), 65. Jardret, V., Zahouani, H., Loubet, J. L. and Mathia, T. G. (1998). Wear 218, 8. Liu, Z., Sun, J. and Shen, W. (2002). Tribology International 35(8), 511. Zhang, S., Hodgson, P. D., Duncan, J. L., Cardew-Hall, M. J. and Kalyanasundaram, S. (2002). Wear 253, 610. Subhash, G. and Zhang, W. (2002). Wear 252, 123. Wang, F., Lacey, P., Gates, R. S. and Hsu, S. M. (1991 ). Journal of Tribology 113 755. Sahoo, P. and Chowdhury, S. K. R. (2002). Wear 253, 924. Varadi, K., Neder, Z. and Friedrich, K. (1996). Wear 200, 55. Yoshioka, N. (1997). Tectonophysics 277, 29. Williams, J. A. (1996). Tribology International 29, 675. Jardret, V., Zahouani, H., Loubet, J. L. and Mathia, T. G. (1998). Wear 218, 8. Kailas, S. V. and Biswas,S. K. (1993). Wear 162-164, 110. Wang, Y. and Hsu, S. M. (1998). Wear 217, 104.
Studies on friction and transfer layer using inclined scratch
33. 34. 35. 36. 37.
279
Tichy, J. A. and Meyer, D. M. (2000). International Journal of Solids and Structures 37, 391. Blau, P. J. (2001). Tribology International 34, 585. Urbakh, M., Klafier, J. D., Gourdon and Israelachvili, J. (2004). Nature 430, 525. Feder, J. (1988). Fractals, Plenum Press, New York. Hasegawa, M., Liu, J., Okuda K. and Nunobiki, M. (1996). Wear 192, 40.
280
CHAPTER 15
SCRATCH RESISTANCE OF HIGH NITROGEN AUSTENITIC STAINLESS STEELS
A. P. TSCHIPTSCHIN 1, C.M. GARZON 2 and D.M. LOPEZ !
1Metallurgical and Materials Engineering Department, University of Sdo Paulo, USP Av. Prof. Mello Moraes 2463, CEP 05508-900, Sdo Paulo, Brazil 2Brazilian Synchroton Light Laboratory, LNLS Caixa Postal 6192- CEP 13084-971, Campinas, SP, Brazil E-mail : [email protected]
ABSTRACT High nitrogen stainless steels (HNSS) are being considered a new promising class of engineering materials. When nitrogen is added to austenitic steels it can simultaneously improve fatigue life, strength and wear and localized corrosion resistance. In this work a single pass pendulum scratch test was used to study the effect of nitrogen on the scratch resistance of an UNS $30403 austenitic stainless steel. Samples with increasing nitrogen contents at the surface were obtained trough high temperature gas nitriding. The thermo-chemical treatments were performed at 1473 K in (N2+Ar) gas atmospheres for 10 hrs, obtaining fully austenitic cases (surface nitrogen contents up to 0.5 wt-%) circa 1.5 mm in depth. The scratch tests were performed in a single-pass pendulum, equipped with strain gages to measure normal and tangential forces during scratching. The specific absorbed energy was calculated as the ratio between the measured absorbed energy and the amount of mass removed from the specimen. An increase of the specific absorbed energy with increasing nitrogen content was observed. The results of the scratch tests were analyzed taking into account the stress- strain behavior during depth sensing indentation tests and the energy absorbed during Charpy impact tests. The improvement in scratch resistance due to nitrogen alloying was attributed to the strong hardening effect of nitrogen in solid solution, which does not affect significantly work hardening and toughness. A comparison between the scratch resistance and the cavitationerosion resistance, measured in previous work, was made too.
KEYWORDS High nitrogen steels, scratch resistance, abrasive wear, machinability
Scratch resistance of high nitrogen austenitic stainless steels
281
INTRODUCTION Stainless steels (SS) have been widely used in different components working under corrosive effects. Investigations have shown that nitrogen is an important alloying addition to stainless steel improving simultaneously localized corrosion resistance, fatigue life, creep resistance, tensile strength and wear resistance [1-6]. These characteristics may be important in many applications, particularly where wear and corrosion mechanisms acting together in mechanical parts such as valve seats and artificial hip joints [7]. Some reasons for these favorable effects are related to the increase of the metallic component of interatomic bonds and the promotion of a short range ordering of atoms in contrast to the clustering of carbon atoms. One of the practical consequences of that is a better corrosion response [8]. The production of such kind of steels is not trivial in view of the limited solubility of nitrogen in liquid metal. Production routes of High Nitrogen Stainless Steels (HNSS) by alloying, pressure metallurgy, powder metallurgy and solid-state diffusion have been studied [9-11]. A low cost thermochemical treatment, which allows alloying the surface and near surface regions of a conventional SS with nitrogen, was developed recently [6, 11-17]. This High Temperature Gas Nitriding treatment (HTGN) is performed exposing steel parts to still N2 gas atmospheres in the range 1273 - 1473 K. Microstructures without nitride precipitation and nitrogen contents at the surface ranging from c.a. 0.2 to 1.0 wt-% can be obtained if adequate nitriding temperatures and N2 partial pressures are used [ 17-19]. Typical case depths obtained vary between c.a. 0.5 to 2.0 mm after 5 to 12 hrs nitriding-treatments. Thus, this HTGN treatment is quite different from conventional nitriding, usually performed between - 750 and 850 K, in which intense chromium nitride precipitation occurs, greatly increasing the hardness, but impairing the corrosion resistance of nitrided parts. Improvements of wear resistance (cavitation, erosion, sliding and fretting) and localized corrosion resistance (pitting and crevice) of martensitic, austenitic and duplex SS after HTGN have been reported [5-6, 11-15, 18-19]. One of the problems in optimizing the HTGN treatment conditions is the considerable amount of performance tests needed to evaluate the processing vs properties relationship. Therefore, in the present work, the possibility of using a single pass pendulum scratch test to assess the tribological properties of high temperature gas nitrided SS is discussed. Single scratch sclerometric technique is a cheap and quick test, which has been mainly used for calculating dynamic hardness and assessing abrasion resistance of different bulk and coating materials [20-23]. The single pass pendulum scratch test has also been used to evaluate the machinability of materials without performing time consuming and expensive in-field machinability tests [24]. The aim of this work is to study the effect of nitrogen addition - through high temperature gas nitriding - on the scratch resistance of an UNS $30403 austenitic stainless steel, by using a single-pass pendulum scratch test. The scratch resistance is evaluated through the specific energy and the results of scratch tests are analyzed taking into account the s t r e s s - strain behavior during depth sensing indentation tests and values of energy absorbed during Charpy impact tests. In addition, a comparison between scratch resistance and cavitation-erosion resistance, measured in previous work [19], is made.
Scratching of materials and applications
282 EXPERIMENTAL PROCEDURE
High Temperature Gas Nitriding and Solution-Annealing Treatments Samples for hardness, scratch and impact tests were cut and machined from a 6 mm thick sheet of UNS $30403 hot-rolled austenitic SS. The chemical composition is given in Table I. Table 1" Chemical composition of the UNS $30403 austenitic stainless steel (wt- %) Cr
Ni
Mn
Si
Mo
C
Ti
18.1
8.5
1.3
0.5
0.04
0.025
0.0056
,
,
i
,
,,,,
9
,
Four sets of samples were high temperature gas nitrided aiming to obtain different nitrogen contents at the near surface region. One set of samples was solution-annealed aiming to obtain specimens without nitrogen additions. The HTGN treatments were carried out in a tubular furnace. The specimens were heated up to 1473 K, under 0.13 Pa, and then exposed to a high purity Ar + N~ atmosphere during 10 hrs, at 0.02, 0.05, 0.1 and 0.17 MPa N2 partial pressures. After nitriding the samples were direct quenched in water. The solution-annealing treatments were performed in an Ar atmosphere using the same thermal cycle reported for the HTGN treatments.
Scratch Test Rectangular prismatic tests samples (6 mm x 9 mm x 55 mm) were metallographically polished up to 1 p.m diamond paste and submitted to the single scratch test in a modified Charpy impact pendulum, with 50 J of maximum capacity. The sintered tungsten carbide scratching stylus, was a tnmcated square-base pyramid of 40 ~ apex angle, with a 0.5 x 0.5 mm fiat top. The initial pendulum height was set to accumulate a potential energy of 35 J, resulting in a velocity of the scratching stylus, at the beginning of stylus-sample contact, of 3.16 m/s. The depth of the scratches, controlled by vertical adjustment of the specimen holder, was set at 105 ~m. Four scratches were made in each of two specimens obtained for each treatment condition. The absorbed energy (E) was measured with a readability of 0.01 J. The removed mass (W) was measured using a Scientech SA120 scale with a readability of 0.1 mg. The specific absorbed energy (e) - which represents the energy consumed during the removal of 1 g of material - was calculated from the measured absorbed energy (E) and the removed mass (W) using equation (1) [20].
E
e=--W
(1)
The system was equipped with strain gages to measure tangential and normal forces developed during the test.
Charpy Impact Tests Notched bar (5 x 10 x 50 mm) sub-sized specimens were tested at 300 K in a pendulum machine with 300 J of maximum capacity, according to the ASTM E23 standard. The depth of the notch was 2 mm. Three samples for each treatment condition were tested.
Scratch resistance of high nitrogen austenitic stainless steels
283
Specimens Characterization Optical microscopy and scanning electron microscopy were used to analyze the microstructure and topography of the scratched specimens and of the turns. SEM was performed in a Philips XL30 microscope. An Oxford WDX600 X-ray spectrometer was used to analyze the nitrogen contents at the surface of the nitrided samples. Depth-sensing indentation tests on top of rectangular (6 x 10 x 10 mm) samples were carded out in a Fisherscope HIO0 apparatus, using a Vickers indenter tip. The maximum load was 250 mN, the loading and unloading times were 90 s and the dwell time at peak load was 20 s. The indentation data were analyzed using the method proposed by Oliver and Pharr [25]. Each data point is an average of 11 measurements. The indentation data analyzed were: hardness (H), total indentation work (Wt), irreversible indentation work (Wir), reversible indentation work (We), loading slope (S1), unloading slope (S) and strain-hardening coefficient during indentation (n). The n coefficient was calculated introducing a correction function, fin) [26], to the Oliver and Pharr procedure, expressed by equation 2 for a Vickers indenter:
s, w, s~
fin) = 1.202 - 0.857n + 0.302n2 = ____
(2)
RESULTS
Characterization of the As Treated Samples Both the solution-annealed and the nitrided samples had austenitic microstructure free of precipitates. The hardened case depth, determined by hardness measurements in the transverse section of the nitrided samples, was between 1.2 and 1.6 mm. Table 2 gives the results of depth-sensing indentation tests, WDS chemical analysis and Charpy impact tests. The nitrogen content at the surface of the samples increases with increasing N2 partial pressure. One can see a strong hardening effect of nitrogen alloying, associated to a tenuous decrease of both the strain-hardening coefficient (n) and the Wir/W t ratio (ductility index). Increasing the nitrogen content up to 0.5 wt-% increases hardness from 2.05 to 3.2 GPa and decreases n from 0.155 to 0.13, Wt from 185.1 to 156.4 nJ and W i l t
from 0.899 to 0.854.
Scratching of materials and applications
284
Table 2: Results of WDS chemical analysis, depth-sensing indentation tests and Charpy impact tests.
PN2 (MPa)
Nitrogen Content (wt-%)
Depth-sensing indentation tests
Charpy impact energy
H (GPa)
Wt (nJ)
Wi~ t (nJ/nJ)
(J)
n
0.0
0.0
2.05 -+01o
0.155
185.1 53
0.899
171.5 +10
0.02
0.05 _+0.o5 2.25 -+o.13
0.145
181.5 54
0.895
171.5 _+!5
0.05
0.11 _+0.07 2.35 5o.15
0.135
178.2 _+4
0.892
165.0 5 l0
0.10
0.31 _+0.o5 2.85 5o10
0.125
165.5 _+25
0.885
159.0 _+15
0.17
0.5 _+0.o7
3.20_+o.10 0.130
156.4 _+2.5
0.854
The Charpy samples suffered large macroscopic plastic strain before fracturing in a fully ductile manner. Table 2 shows that the HTGN treatments almost did not affect the impact resistance of the UNS $30403 steel. Increasing the N 2 partial pressure decreases slightly the impact energy from 171.5 to 159 J. The tenuous decrease in the absorbed impact energy is due to a slight decrease of both the work hardening ability and the ductility index.
Scratch Resistance Figure 1 shows mean values of the parameters analyzed during the scratch experiments: absorbed energy (E), removed mass (W), specific absorbed energy (e), maximum normal force (NF) and the ratio of the tangential force to the normal force (TF/NF). Fig. 2 shows a representative force-distance scratch curve, as well as the ratio TF/NF during the scratching event, for each studied treatment condition. The specific absorbed energy increases with nitrogen content whereas both the absorbed energy and the removed mass decrease with increasing nitrogen contents. When the nitrogen content of the specimens increases the mass removed during the scratch test decreases more rapidly than the absorbed energy, leading to an increase in the specific absorbed energy.
Scratch resistance of high nitrogen austenitic stainless steels
285
0.0025
4500 4000
0.0020 3500 30O0
0.0015
. . . . . . . . i. . . . .
.~ 2500 2000
0.0010
1500
0.0005
1000 5O0
I
!
I
I
t
0
0.1
0.2
0.3
0.4
-
I
0.5
0.0000
.
I
.
4
0
0.6
.
.
0.2
600
I
.
t
0.3
.
:
0.4
0.5
0.6
(wt-%)
N-content,
(wt-%)
N-content,
i
O. 1
0.9
500
0.8
400 [~ 0 . 7 "
A Z
~
!
300
0.6-
200 0.5-
100 0
t
I
I
I
I
0
0.1
0.2
0.3
0.4
N-content,
(wt-%)
.
! 0.5
0.4 0.6
* 0
.
~ 0.1
.
I
9
0.2 N-content,
:
:
~,
0.3
0.4
0.5
0.6
(wt-%)
Fig. 1. Effect of the nitrogen additions on the results of the scratch experiments.
Additionally, one can see in Fig. 1 that the maximum normal force and the mean TF/NF ratio remain almost constant. The maximum normal force almost did not vary because it is mainly influenced by the scratch geometry (scratching stylus and depth of the scratch). It is affected in a lesser extent by the plastic properties of the material. On the other hand, the minor variation in the mean ratio TF/NF is mainly due to the dominance of the plowing term during the tests: the total tangential force is equal to the sum of the adhesion and plowing terms, and it has been reported that for ductile materials the plowing term is larger than the friction term [27]. The large values of the plowing term have been attributed to the relative ease with which the penetration depth increased with the imposed depth of cut [27]. In the force-distance scratch curves in Fig. 2 one can see that the maximum tangential force takes place in the midway of the scratch, where the maximum depth of cut occurs, while the maximum normal force takes place in the last third part of the scratch. A high frequency oscillation in the tangential force after the first third part of the scratch is observed. The oscillation is probably due to contact instability while the indenter is plowing the specimen. This behavior, observed in several other materials, has been associated to a stick and slip mechanism [20, 28]. Vingsbo and Hogmark reported that these oscillations in the tangential
Scratching of materials and applications
286
force o c c u r for metallic materials when the maximum scratch depth overtakes a critical value, which vary with material [20].
,500]
0.0N wt-%
]
1.4
"
400
0.0 N wt-%
1.2 1.0-
0.8
~,300 w.,,
u_ 0.4 100 0.2
0 0
4
8
12
0.0 I , 4 0
16
~ ~ : : : I ' : ~ ~ ' ' ' t 4 8 12 16
Distance, (mm) 50O
Distance (mm)
1.4
0.05 N wt-% TF NF
......
400 t
0.05 N wt-%
1.2. 1.0.
A300 Z
0.8
e" ,o 200 o i,,i.
0.6 0.4
100 0.2 ] 0 0
4
8
12
t . 16
0.0 0
4
0.11 N wt-%
400
12
1 16
Distance (ram)
Distance, (rnm) 500
8
:- .... . . . . . . .
T~ NF
!
I
1.4
0.11 N wt-%
1.2 1.0
~,300
0.8
LL 0.4 100 0.2
0 0
. . . . . . . . . . . . . 4 8 12
0.0 16
0
Distance, (mm)
(Continue to the next page)
4
8
12
Distance (ram)
16
Scratch resistance of high nitrogen austenitic stainless steels
500 "l
287
1.4 --
0.31 N wt-%
0.31 N w1-%
1.2--
,oo I
...... ~
~oo~ ,r
~~AA"' ......."
/
"* ~ o /
o.~_
"
'.
.o 2O0 q-
,.o.
"
=~
O.6o4 0.2 --
'"
0 o
9
0
4
8 Distance,
12
16
(ram)
'
'
:
~
~
~
I
~
;
;
~ 1 7 6 o.: , , :
;
16
nee
500
0.50 N wt-%
400
. . . . . I~
1.4 -,
0.50 N wt-%
1.0~,300
. . . .
'-
0.8-
200
~1,. 0.6-
100 I
~
0.2-0.0
0
4
8 Distance,
12 (ram)
16
I
0
4
'
:
:
~
~
,
t
8 Distance
I
;
12
16
;
(ram)
Fig 2. (Continue from the previous page) Force vs. distance scratch curves for each treatment condition.
288
Scratching of materials and applications
Figure 3 shows the scratched surfaces of two samples containing 0.0 and 0.31 N2 wt-%. The scratch walls at the beginning of the scratching event are smoother than midway, where the stick and slip mechanism could be identified, in accordance with the high frequency oscillation observed in the tangential force after the first third part of the scratch event (Fig. 2).
Fig. 3. SEM micrographs of the ditch walls at the initial parts, (a) and (c), and midway, (b) and (d), of the scratch for samples with 0.0 and 0.31 N2 wt-%.
The high nitrogen steel show smaller and fewer tearing and pull outs compared to the steel containing no nitrogen. This qualitative observation was not quantitatively measured in the TF/NF profiles, probably, due to the dominance of the plowing term in the tangential force. In Fig. 4 it can also be seen that a part of the affected volume of the groove was displaced to the edges. Extensive plastic deformation without inducing micro-cracking was observed aside the ditches, indicating that the increase in nitrogen content did not embrittle the material. The removal of mass occurred mainly by a micro-cutting mechanism.
289
Scratch resistance of high nitrogen austenitic stainless steels
Fig. 4. SEM micrograph from a region aside the scratch in a sample with 0.0 N2 wt-%. Intense plastic deformation with associated slip lines and ductile fractured surfaces are observable.
Correlation between Scratch Resistance and Cavitation-Erosion Resistance Figure 5 shows the variation of the machinability index (inverse of the specific energy) and of the mass loss rate measured during vibratory cavitation, in a previous work [17], both as a function of the nitrogen content in the near surface region of HTGN $30403 SS. Both, the machinability index and the mass loss rate, vary with nitrogen content in a similar manner. The linear correlation between scratch resistance and cavitation-erosion resistance is probably due to the fact that in both wear tests the performance of the nitrided cases is controlled by the resistance to elasto-plastic deformation. dwtdt, (mg/ks)
e't, (g/J)
1.o5
6xlo ~
0.85 5x10 "4 0.65
0.45
4x10 "~
0.25 3xlo"
0.05 2x104 0
0.1
0.2
0.3
0.4
0.5
0.6
N-content, (wt-%)
Fig. 5. Machinability index and mass loss rate during CE as function of the nitrogen content.
290
Scratching of materials and applications
DISCUSSION It has been well recognized that the response of material surfaces to scratching involves elastic recovery, plastic deformation, and abrasive wear including sheafing and micro-cracking. The experimental results obtained in the present work showed that, despite the slightly lower toughness of the HTGN samples compared to the solution-annealed ones, nitrogen alloying in the near surface region led to fully ductile fractures during both scratch tests and Charpy impact tests. Besides, the nitrided cases exhibited enhanced resistance to plastic deformation associated to slight decrease in the work hardening rate. Therefore, the increase in specific energy consumed in the scratch tests is mainly due to a decrease in the amount of work done by the contact force when deforming the materials elasto-plastically. This hypothesis is in accordance with the experimentally observed relationship between the specific energy and the total indentation work during depth-sensing indentation tests, which is shown in Fig. 6. 4500
.
.
.
.
.
.
.
.
.
.
.
.
.
4000
3000 "~
"~""~-..
2500
0 31 N ~ ~
,
~ 2000
~ 0.
1500 1000 500
N
1:t2= 0.84 9
140
fT.x0.11N
I
150
9
!
160
9
I 170
"
9
I 180
9
I 190
200
WT, ( n J ) Fig. 6. Relationship between specific energy measured during scratch tests and total indentation work measured during depth-sensing indentation tests.
Regarding the suitability of the single pendulum scratch test to assess the tribological properties of high temperature gas nitrided SS, the experimental results allow pointing out three major qualities: (i) the scratch tests gives a good indication of the elasto-plastic behavior of nitrided cases submitted to contact and cutting stresses, (ii) the results obtained in the analyzed region have a great statistical significance due to its large size (the analyzed area on the sample's surface was circa 0.7 x 17 mm) and (iii) the depth of the scratch is around just ten percent of the hardened case depth.
Scratch resistance of high nitrogen austenitic stainless steels
291
CONCLUSIONS 1- High temperature gas nitriding strongly improves the scratch resistance of UNS $30403 austenitic SS. Increasing the nitrogen content in solid solution up to 0.5 wt-% leads to an increase in the specific energy from 1770 to 3540 J/g. This can be attributed to the strong hardening effect of nitrogen in solid solution, which does not affect work hardening and toughness significantly. 2 - A linear correlation between the mass loss rate during vibratory cavitation, measured in a previous work, and the machinability (inverse of the specific energy) of high temperature gas nitrided UNS $30403 austenitic SS was observed. 3 - The single pass scratch test is suitable to assess the performance of case hardened HTGN SS, submitted to dynamic contact and cutting stress allowing to measure the abrasion resistance and to estimate the cavitation erosion resistance of these steels.
REFERENCES .
2. .
4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
Gavriljuk V. G. (1996) ISIJ Int 36 (7), 738-745. H~inninen H., Romu J., Ilola R., Tervo J. and Laitinen A. (2001 ) J. Mater. Process Tech. 117 (3): 424-430 Tervo J.,Mater. Sci. Forum 1999; 318-320:743 - 750. Harminen H. (1999)Mater. Sci. Forum 318-320:479-488. Betas H., Eul U., Heitz E. and Juse R. (1999) Mater. Sci. Forum 318-320, 517-522. Gavriljuk V. G., Betas H. (1999) High Nitrogen Steels. Berlin: Springer-Verlag. Thomann, U. and Uggowitzer, P. (2000) Wear 239, 48-58 Gavriljuk, V. Nitrogen in iron and steel. ISIJ Int 1996; 36 (7): 738-745. Betas H. (1996) ISIJ lnt 36 (7): 909-914. Toro A., Alonso-Falleiros N., Rodrigues D., Ambrosio filho F. and Tschiptschin A. P. (2002) Trans. Indian Inst. Met. 55 (5), 481-487. Betas H., Siebert S. (1993) Proceedings of lnt. Conf. on High Nitrogen Steels- HNS, 1993, Kiev: Institute of Metal Physics, p. 566-571. Siebert S. Doctoral thesis, Bochum: Ruhr-University, 1994. Berns H. and Siebert S. (1996) ISIJInt 36 (7), 927-931. Berns H., Bouwman J. W., Eul U., Izaguirre J., Juse R., Niederau H., Tavemier G. and Zieschang R. (2000) Mat.-wiss. U. Werkstofftech 31 (2),152-161. Toro A., Misiolek W., Tschiptschin A. P. (2003) Acta Mater 51 (12), 3363-3374. Garz6n, C. M. and Tschiptschin A. P. (2004) Mat. Sci. and Tech. 51(12), 915-918. Garzon, C. M . , Toro A. and Tschiptschin A. P. (2002) Trans. Indian Inst. Met. 55 (4), 255-263. Mesa D. H., Toro A. and Tschiptschin A. P. (2003) Wear 255 (1-6), 139-145. Dos Santos J. F. , Garz6n C. M. and Tschiptschin A. P. (2004) Mat. Sci. and Eng.- A 382 (1-2), 378-386. Vingsbo O. and Hogmark S. (1984) Wear 100 (1-3), 489-502. Jiang J., Yao M., Sheng F. and Gao X. (1995) Wear 181-183 (5), 371-378. Williams J. A. (1996) Tribology Int 29(8), 675-694. Hu W., Li S., Li S., Sun X. and Guan H. (1999) Tribology Int 32 (3), 153-160. Paro J., Hanninen H. and Kauppinen V. J. Mater. Process Tech. 2001, 119:14-20. Oliver W. C. and Pharr G. M. (1992) J. Mater. Res. 7 (6), 1564-1583.
292
Scratching of materials and applications
26. 27. 28.
Malzbender J., De With G. (2002) Jr. Mater. Res. 17(2), 502-511. Liang Y. N., Li S. Z., Li D. F. and Li S. (1996) Wear 199 (1), 66-73. V61ez, J. M., Tanaka, D. K., Sinatora, A. and Tschiptschin A. P. (2001) Wear 251(112), 1315-1319.
APPENDIX
Fig. A1. Modified Charpy pendulum used for carrying out the experiments, b) Sample holder and micrometric positioning table.
Fig. A2. Schematic representation of the pendulum.
Scratch resistance of high nitrogen austenitic stainless steels
293
Fig. A3. The stylus geometry; Square-based 40 ~ truncated pyramid with a 0.5x 0.5mm flat top.
294
CHAPTER 16
ABRASION OF ENGINEERING CERAMICS, AIMgBI4-TiB2 COMPOSITE AND OTHER HARD MATERIALS
S. BAHADUR and A. AHMED
Department of Mechanical Engineering, lowa State University Ames, 1A 50011-3020, USA. E-mail: [email protected] ABSTRACT The abrasion resistance of hard materials makes these materials particularly suited for harsh situations such as earth moving, mining, rock drilling etc. where contact with abrasive materials is involved. In view of their commercial importance, the abrasion behavior of engineering ceramics such as alumina, silicon carbide and silicon nitride, a-BN, and A1-Mg borides, which are an emerging class of materials, is presented. The latter also include compositions modified by the addition of TiB2 which, when optimized for microstructure and composition, approach the hardness of B-BN. The abrasion results reported pertain to single point scratching and belt abrasion methods which are described. The material removal in scratch tests was estimated by profilometry and in belt abrasion test by gravimetric measurements. The variation of material removal from abrasion with process variables such as load and belt speed is presented. The mechanisms of abrasion in these materials have been studied with the help of scanning electron microscopy.
INTRODUCTION Ultra-hard materials are commonly used for abrasion-resistant applications and cutting tools. Such materials are needed in many applications such as earth moving, mining, rock drilling etc. where they come in sliding contact with abrasive materials. In view of the commercial significance of abrasion, many researchers have studied the mechanisms involved in abrasion. Others have studied them with the objective of increasing material removal rates in abrasive machining of hard materials. The problem with enhancing the material removal rates is the surface and subsurface damage that occurs and is detrimental to mechanical properties. Abrasion is a complex phenomenon affected by hardness, elastic modulus, yield strength, crystal structure, microstructure, and composition. The early work by Khrushchov and Babichev [1 ] on pure metals showed that abrasion rates were inversely proportional to hardness. With reference to microstructure, it has been shown that austenite and bainite of equal hardness are more abrasion-resistant than ferrite, pearlite, or martensite in steels. In order to be effective for abrasion resistance, the particles precipitated or externally added need to be larger than the abrading particles. The additive particle characteristics that work best for abrasive wear resistance are hard, tough, and blocky. The latter are more effective in terms of reduced crack propagation and breakage as compared to the plate or rod shaped particles. Zum Gahr [2] has
Abrasion of engineering ceramics, AIMgB 14--TiB2 composite and other hard materials
295
demonstrated that the orientation, size, modulus of elasticity, relative hardness, and brittleness of the second phase in composites are important factors for wear resistance. Fisher et al. [3] performed abrasion studies on a series of zirconia samples with constant hardness but varying toughness. They found that the abrasive wear decreased with the fourth power of toughness. This fourth power law does not apply to all materials. Quercia et al. [4] conducted micro-abrasion test and found that there was a linear variation of wear volume with sliding distance. Mao et al. [5] studied the abrasion behavior of advanced A1203-TiC-Co ceramics with varying proportions of the constituents. They found that abrasion resistance depended mainly on fracture toughness, while hardness had merely a secondary effect. The factors that affect abrasive wear in belt abrasion are the type of abrasive and its characteristics, such as hardness, toughness, angularity, and size [6, 7], speed of contact, unit load of abrasive on the material, humidity, and temperature. The shape of the abrasive particle, together with the load, influences the shape of the groove produced in the material, and the transition from elastic to plastic contact. The mechanisms of abrasive wear that have been proposed are chipping [8], delamination [9], ploughing, flake formation, and the generation of powdery fragments. The mechanisms depend on contact stress [ 10, 11 ] and grain boundary microfracture characteristics of the material [ 12, 13]. In brittle materials, a transition in wear mechanism occurs with increasing load and/or particle size [ 14]. At low loads or with small particles, fracture may be suppressed and abrasive wear may occur by plastic processes. At higher loads or with larger particles, brittle fracture occurs leading to a much higher wear rate. Gee [ 15] reported that in the case of hard metals fracture occurred on a fine scale, but in ceramics, fracture occurred on a larger scale often removing large fragments of the material. Diamond and cubic boron nitride (13-BN) are currently the only established bulk materials with hardness greater than 40 GPa. A new class of competing materials has recently emerged. These are the complex borides of aluminum and magnesium, A1MgBI4, prepared with sub-micron sized second-phase additives. These boride composites have hardness values ranging from 3046 GPa, depending on the size and distribution of the phases. Thus these materials may complement diamond and 13-BN in applications where high hardness is needed. The boride composites may also offer a cost advantage over diamond and 13-BN, if suitable large-scale manufacturing technologies are developed [ 16-18]. In the present chapter, the abrasion of hard materials is discussed and specific results are presented from our earlier works on the abrasion of ceramics and ultra-hard A1MgB~4 composites.
ABRASION TESTS There are many kinds of tests that have been used for abrasion studies. For specific information on such tests, reference to the ASTM Standards Vol. 03.2 is recommended [ 19]. This has six standards described for abrasion studies. Since the contact configuration and the contact stresses in these tests vary, the results from different tests are not always in agreement.
Scratching of materials and applications
296
The two test methods that are recommended for single-pass and multiple-pass abrasion studies are described below. These were used in the abrasion studies on ceramics and A1MgB~4 materials reported here.
Single-Point Scratch Test
Direction of Motion ~ /
I
Bushing
Weight r Spindle Coupling [
Motor
IIII
!
!
' J Rotating --,~/ / .... threaded ,, , ~ screw .... ' / ~t __ Diamond 1..........]....,~rlillllllllllllllllltlllll l ...................1... i ] indentor ] rlllllllllll,,
Specimen _~ _ _ ~, i ~ ~
Fig. 1. Schematic arrangement of the experimental setup for single-point scratch test.
This test method is similar to the ASTM G171 Standard [ 19] but not identical. The experimental set-up for the test is shown in Fig. 1. It consists of a Rockwell C 120 o spherocone diamond indenter with a 200 pm tip radius that is secured to the end of a vertical spindle which slides freely in a long bushing. The friction between the bushing and the spindle is minimized by lubricating the sliding surfaces with lithium grease. The bushing assembly is welded to a nut which traverses linearly as the screw rotates in place; this makes the indenter traverse linearly. The screw is coupled to a motor shatt with a flexible coupling. Limiting switches are installed so as to set the indenter travel. The scratches are spaced apart so as to minimize the effect of damage from adjoining scratches.
Abrasion of engineering ceramics, AIMgB I4--TiB 2 COnlposite and other hard materials
M
0
1
'f Ill :1
297
D E P T H
J |
J GROOVE WIDTH
Fig. 2. Schematic record of a surface profile with scratch by profilometer.
In order to measure the groove width, scratch profile is recorded by a stylus profilometer to a high resolution. The software routine calculates the surface roughness (Ra) and assigns the zero line of reference which is displayed on the computer screen. The width of the groove is measured with reference to the zero line. The distance between the points where zero line intersects the two sides of the groove is taken to be the width of the scratch, as shown in Fig. 2. The width of each scratch is measured in many locations. The mean scratch width is reported as an indicator of the abrasion resistance of the material.
Belt Abrasion Tests Load ~ + Spindle - - - ~ ~ Sample Holder
Container
\
'
' /
Motor Bushing
t
.
Sample
C)
I
Lubricant Level
L
Dai~ BIt
,
, ,
,
n
Fig. 3. Schematic arrangement of the experimental setup for belt abrasion test.
This is one of the configurations of the ASTM G132 standard [19]. The set-up for multipoint abrasion test consists of a diamond abrasive belt mounted on two rollers which are driven by a variable speed motor (Fig. 3). The specimen is loaded on the rotating belt surface. It is secured
Scratching of materials and applications
298
to a spindle which is constrained by a bushing but is free to move in the vertical direction. The friction between the spindle and bushing is minimized by lubrication. In order to avoid clogging of abrasive belt with cutting particles, the whole set-up is positioned in a container filled with a lubricant/coolant.
MATERIALS AND PROPERTIES Table 1. Hardness and Fracture toughness of hard materials studied for abrasion resistance.
Material
Hardness (GPa)
Fracture Toughness (MPa~n)
A1203
13
5.5 ....
Si3N4
15
6.8
22-25
2.5-2.9
. .
SiC .
.
.
.
.
.
.
.
.
.
_
WC+Co
18
fI-BN
45
20
8.5
10.5
50
6.0
7.5 _
.
.
.
.
A1MgBt4 .
.
.
.
AIMgBI4
.
.
_.
.
.
28 .
.
.
.
.
.
.
.
3.7 • 0.20
70 wt.% TiB2
37
4.1 • 0.21
AIMgBt4 - 80 wt.% TiB2
36
.
A1MgBI4
.
.
.
.
.
.
.
3.0 • 0.19
33
.
30 wt.% TiB2
.
.
_
.
,
,
3.4 • 0.17
The hardness and fracture toughness of the materials covered here are given in Table 1. In the case of hard materials, these are the most relevant properties for abrasion resistance. Of the materials given in Table 1, A1203, SiC and Si3N4 are the most common ceramic materials used where high hardness is needed. Cubic boron nitride (fl-BN) is a super hard material with good toughness but is fairly expensive. Included in the discussion is a newer material A1MgB~4 which is made harder by the addition of TiB2 while still having reasonable fracture toughness. The interesting point to note here is that with 70 wt% TiB2 in A1MgB~4 along with the increase in hardness, fracture toughness is also increased. The hardness of AIMgBI4-70 wt.% TiB2 is within 10-20% of that of 13-BN and is much higher than that of WC+Co or SiC. Since TiB2 proportion in the material with optimum properties is greater than that of AIMgB14, one could also consider it as an enhanced TiB2 material.
Abrasion of engineering ceramics, AIMgB I4-TiB 2 composite and other hard materials
299
ABRASION BEHAVIOR OF MATERIALS
Fig. 4. Material removal rate as a function of load for alumina and silicon nitride in air and water.
Scratch tests were done on alumina and silicon nitride with loads ranging from 5 to 50 N in 5 N increments [20]. In the tests at the lowest load of 5 N, microchips could be seen but surface profilometry was not able to resolve these scratches. This indicated that the critical depth for chip formation was barely reached as elastic recovery was able to overcome plastic deformation in the grooved track at this low load. The load below which the volume of material removed could not be measured using profilometry was in the range of 5 to 10 N in the case of alumina and 10 to 15 N for silicon nitride. Figure 4 shows the volume of material removed per unit sliding distance as a function of load for alumina and silicon nitride in air and water [20]. It may be seen that the material removal rate is small at low loads. It increases with increasing load and after about 30 N load, the increase is much more rapid. The material removal rate for silicon nitride is lower than that for alumina for the same load. This difference could be accounted for in terms of the relative hardness of these materials (Table 1). Furthermore, the material removal in water is lower than in air for both materials.
300
Scratching of materials and applications
Fig. 5. Material removal in abrasive machining performed on alumina specimens in various media [21 ].
Fig. 6. Material removal in abrasive machining performed on silicon nitride specimens in various media [21 ].
In belt abrasion tests, surface velocity is much higher than that used in scratch tests and so temperature rise becomes an additional factor to reckon with. The temperature rise on specimen surface accompanied by cooling is conducive to residual stresses of tensile nature which would be expected to contribute to microcracking. Figures 5 and 6 show the material removal rates as a function of load in belt abrasion tests on alumina and silicon nitride in the environments mentioned [21 ]. In all the cases, material removal rates increased with load and were higher in lubricants than in air. The material removal rate at any load was the highest for alumina in mineral oil and for silicon nitride in mineral oil with 1% chlorinated paraffin. The observations on higher material removal rates for belt abrasion in lubricants than in air are in contradiction to
Abrasion of engineering ceramics, AIMgBI4--TiB 2 composite and other hard materials
301
those from scratch tests where material removal rates were the highest in dry condition [2]. This is attributed to the problem of clogging of abrasive belt with cutting particles in dry abrasion which reduces the abrasion efficiency. The lubricants promote dislodging of abraded particles from the belt. The bonding of abraded fragments to the cutting particles as in the dry condition also increases the rake angle which reduces the cutting action. In single-point scratch tests, these were not the pertinent factors.
260
-
240 220 200
-
=L
,,,
180 -
"~
160 -
,,~
140 120
r~
A1MgB14+ 0%TiB2 ---O-- AIMgB14+ 10%TiB2 A1MgB14+ 30%TiB2 ---'V-- AIMgB14+50%TiB2 ----II-- AIMgB14+ 70%TiB2 ----t3-- AIMgB14+ 80%TiB2
-
100 -
8060 40 10
i
r
T
i
20
30
40
50
60
70
80
Load, N Fig. 7. Variation of scratch width with load for A1MgBI4 with different TiB 2 proportions by weight.
Figure 7 shows the variation of scratch width with load for different TiB 2 proportions in A1MgB14 [22]. There are two observations that deserve to be noticed. Firstly, for any material, the scratch width increases almost linearly with load. Secondly, for any load, the scratch width decreases with increasing TiB 2 proportion up to 70 wt.%. For 80 wt.% TiB 2, scratch width is observed to increase considerably. For this composition, both hardness and fracture toughness are seen to decrease. This behavior supports the generally accepted relationship that a decrease in hardness or fracture toughness leads to the loss of abrasion resistance. At low loads of 20 and 30 N, microchips could be seen on the scratched surface but the critical depth for chip formation was barely reached, as elastic recovery was able to overcome plastic deformation in the grooved track.
Scratching of materials and applications
302 300 WC + co -,--O--- SiC 70% TiB.
250
---q-- C
=-
200
"0 . ~
~
~"
150
100
50
i
0
|
|
|
,
,
,
|
i
,
,
,
|
,
,
i
i
|
,
,
,
,
,
,
t
i
!
!
!
!
|
20
30
40
50
60
70
i
,
9
80
Load, N Fig. 8. Comparison of scratch resistance of AIMgB~4-70 wt.% TiB 2 with reference materials listed in Table 1.
Figure 8 provides the comparison of scratch resistance of A1MgB~4-70 wt.% TiB2 with the reference materials (SIC, I3-BN, and WC+Co) given in Table 1 [22]. The boride material is considerably superior in scratch resistance to all the materials except [3-BN. Figure 9 shows the variation in material removal rate as a function of belt speed in the belt abrasion tests for A1MgB~a-TiB2 with 0, 30, and 70 wt.% TiB2 [22]. Included in this figure are also the plots for WC-Co and 13-BN. It should be noted that wear rate decreases with the increase in belt speed for any given load. With the increase in belt speed from 0.42 to 0.65 m/s (A to B; as shown in Fig. 9(a)), there is a large decrease in wear rate for some materials but later from B to C the decrease occurs gradually for all the materials. At A, the belt speed is fairly low and so the heating effect at the interface is negligible. Thus cutting action by the abrasive particles is quite efficient resulting in high wear rate. With the increase in speed to 0.65 m/s corresponding to B, the temperature rise at the interface becomes significant so that the abrasive particles do more plowing and gouging than cutting. As the speed is increased from B to C, there is continuing temperature rise at the cutting interface so that the efficiency of abrasive action decreases. Thus, wear rate continues to decrease from B to C. The A1MgB14 materials with 30 and 70 wt.% TiB2 along with 13-BN exhibit higher abrasion resistance than the other two materials. This is because of their higher hardness (Table 1) which offers more resistance to indentation under load. Apart from the difference in hardness, the large difference between the wear rates of boride materials with and without boride is because of the TiB2 phase which is harder than AIMgB]4. The wear rate of WC-Co is the highest because its hardness is the lowest of all the materials.
Abrasion of engineering ceramics, AIMgB I4-TiB 2 composite and other hard materials
0.014 -
I
WC + Co ----O-- AIMgB14 + 0 wt.% TiB 2
0.012 E
~-BN AIMgB~4 + 30 wt.% TiB 2
I
E E
O.OLO-
rt._
0.008 -
303
-..--II-- AIMgB14 + 70 wt.% TiB 2 I I I
t._
I
I I I
0.006
I
>
,,=1=
t~ t._
I
I
I
0.004
I
I
I
I
I
I
=
0.002
0.000
0.2
(a)
0.4
0.6
0.8
1.0
Belt speed,
1.2
1.4
1.6
1.8
m/s
Fig. 9. Variation in material removal rate with belt speed for A1MgB~4 with different TiB2 proportions and the reference materials in Table 1 when specimens were loaded with (a) 5 N, (b) 10 N, and (c) 20 N loads on their 9 mm x 3 mm surface. (Continue to the next page)
304
Scratching of materials and applications 0.025 WC + Co - - O - - AIMgB14 + 0 wt.% TiB 2
E
0.020
~ ~
-
[3-BN AIMgB14 + 30 wt.% TiB 2
- - - I - - AIMgB14 + 70 wt.% TiB 2
E E
. . ,
0.015 I,.
(g (9 m
0.010
-
0.005
-
!._
'~
0.000
,
,
,
,
,
I
0.2
0.4
A
,
,
i
I
i
l
l
A
I
0.6
i
=
i
.
I
0.8
.
.
.
,
I
1.0
,
,
,
I
1.2
A
,
,
,
I
1.4
1.6
1.8
Belt speed, m/s
(b)
0.06
WC+Co - - O - - AIMgB~4 + 0 wt.% TiB 2 0.05 -
13 -BN AIMgB~4 + 30 wt.% TiB 2
E E E
AIMgB~4 + 70 wt.% TiB 2
0.04-
t~ i_
"-E------______Q
t._
m
0.03 -
~ 9 r
0.02
I,,,
J~
<1: 0.01 w
,
0.00
,
|
, i
0.2
(c)
,
0.4
i
0.6
|
|
,
i !
0.8
J
i
,
,
,
,
w
i
,
i
l
|
i
1
i
1.0
1.2
1.4
1.6
Belt speed, m/s
Fig. 9. (Continued from the previous page)
,
r
,
.
.
1.8
Abrasion of engineering ceramics, AIMgBI4--TiB2 composite and other hard materials
305
0.06
0.05
WC + Co AIMgB~4 + 0 wt.% TiB 2 ~-BN AIMgB~4 + 30 wt.% TiB 2
0.04
AIMgB~4 + 70 wt.% TiB 2
E E
. ~ J
E =; L
0.03 -
L
0= (9
0.02
-
0.01
-
0.00
-
, i
L_
,
0
,
L
i
,
i
j
|
,
i
i
j
.
.
.
.
.
.
.
,
,
i
i
i
i
i
5
10
15
20
25
i
,
,
30
Load, N
Fig. 10. Variation in material removal rate with load for A1MgBI4 with different TiB2 proportions and the reference materials in Table 1 at a belt speed of 1.1 m/s.
Figure 10 shows the effect of load on wear rate for a sliding speed of 1.1 m/s [22]. It should be noted that wear rate increases continuously with the increase in the load for any material. The rate of increase in wear rate is the greatest for WC-Co which has the lowest hardness. For the next harder material, which is A1MgB~4, the wear rate is much lower. This indicates that wear rate is dependent upon the embedment depth of abrasive particles into the material surface. This is further supported by the observations that wear rate increases continuously with the increase in load.
306
Scratching of materials and applications
ABRASWE WEAR MECHANISMS
Fig. 11. Scanning electron micrographs of low load scratch tests on a-SiC ceramic: (a) single pass with 50 ~tm indenter; (b) 50 passes with 200 ~tm indenter; (c) 100 passes with 200 ~tm indenter.
Gee [ 15] performed multiple scratch tests on a-SiC ceramic materials. Arrays of seven scratches separated by lmm were performed on the samples (an initial single scratch under the test load of 4N, followed by scratches with 2, 5, 10, 20, 50 and 100 repeats). Three scratches
Abrasion of engineering ceramics, AIMgB I4-TiB2 composite and other hard materials
307
corresponding to a single pass with 50 lam indenter, and 50 and 100 passes with 200 ~tm indenter are shown in Fig. 11. It can be seen that severe fracture damage occurred just after a single pass with 50 ~tm indenter (Fig. 1 l a). The damage increased as the number of repeats increased. By contrast, the first isolated pockets of significant fracture damage for the 200 lam indenter occurred only after 50 repeat passes (Fig. 11b). This damage extended along the length of the scratch and became more marked as the number of repeats increased (Fig. 11 c).
Fig. 12. Schematic representation of the classical modes of subsurface damage in brittle materials: (a) under a concentrated contact, (b) in scratching of brittle materials, showing the side and end views.
Indentation and scratching in brittle solids produce cracks as shown in Fig. 12 [9, 23]. Pennyshaped median cracks are generated beneath the plastic deformation zone perpendicular to the surface during loading. During unloading, lateral cracks form below the surface in the plastic zone and propagate parallel to the surface. They are formed by residual elastic stresses caused by relaxation of the deformed material around the region of contact. These cracks may extend upwards to the surface and join the radial cracks. The latter are generated from the edge of the contact impression perpendicular to the surface during unloading at low loads. During scratching, lateral cracks form continuously in the wake of the indenter while median cracks form just ahead of the point of contact [9], as shown schematically in Fig. 12b.
308
Scratching of materials and applications
Fig. 13. Micrographs showing the subsurface damage in silicon nitride specimens scratched in air at (a) 15 N, and (b) 50 N loads.
Desa and Bahadur [20] examined the subsurface damage by bonding two-halves of the samples with finely polished surfaces in the thickness direction and scratching perpendicular to the bonded edge. After the test, the polished surfaces were separated and examined in a scanning electron microscope after standard preparation. The specimens were sometimes tilted in the microscope, so that the top part of the figure showed a portion of the scratched surface and the lower part showed the subsurface directly below the point of contact. In all the cases, the grooved surface exhibited plastic deformation but the subsurface microcrack formations varied. In the case of silicon nitride scratched at 15 N load (Fig. 13a.), a lateral crack is seen emanating from a point directly below the contact surface and progressing on the left and right sides towards the surface. Figure 13b shows the area directly underneath the scratch profile for 50 N load. There appear to be two lateral cracks emanating from a point below the surface and a median crack extending from the same point towards the surface. Plastic deformation extended
Abrasion of engineering ceramics, AIMgB 14--TiB2 composite and other hard materials
309
to some depth below the scratch surface. The lower left portion shows a few fine cracks about the size of a grain which are not fully developed.
Fig. 14. Micrographs showing the transverse cross-sections of alumina samples scratched in air at (a) 10 N, and (b) 50 N loads.
Figure 14 shows the transverse cross-sections of alumina scratched at 10 N and 50 N loads. The groove produced by scratching is seen on the top. The substrate deformation for 50 N load is much greater than for 10 N load. In the case of 50 N load, there is a median crack (shown by an arrow) which seems to have originated in the plastic zone near the surface. There is extensive plastic deformation below the point of contact because no grain structure can be seen.
310
Scratching of materials and applications
Fig. 15. SEM micrograph showing deformation and other surface features at different magnifications in AIMgB14-30 wt.% TiB2 scratched at 20 N and 70 N loads.
Figure 15 shows the buildup of damage at different magnifications in A1MgB14-30 wt.% TiB2 when scratched at 20 N and 70 N loads [22]. The lett column of the diagram shows surface features for 20 N load and the right side for 70N load, and each successive figure below shows features at higher magnifications. At the load of 20 N, the deformation at surface asperity peaks is seen and elsewhere the contact is merely superficial. There is local yielding occurring because of the hydrostatic stress state at indenter tip contact. At the higher load of 70 N, a groove of about 150 ~tm width is formed. The groove boundaries are sharp and plastic deformation is noticed on the entire surface (Fig. 15e). In the high magnification micrograph of Fig. 15f, cracking at the location marked by arrows is seen.
Abrasion of engineering ceramics, AlMgB I4-TiB 2 composite and other hard materials
311
Fig. 16. SEM micrographs showing the subsurface of A1MgB~4 at 70N load in air (a) with 0 wt.% TiB2, (b)with 70 wt.% TiB2.
Figure 16 shows deformation on the grooved surface of A1MgB14 sample with 0 and 70 wt.% TiB2 scratched at 70 N load [22]. Basically plastic deformation along with fragmented material is seen on the grooved surface of A1MgB14 sample with 0 wt.% TiB2. In contrast to this, profuse cracking occurs on the scratched surface of A1MgBi4-70 wt.% TiB2. This is due to the presence
312
Scratching of materials and applications
of TiB2 brittle phase. Thus, whereas the presence of TiB2 in AIMgBt4 makes it more resistant to scratching, it also makes the material prone to cracking at high indentation loads.
Fig. 17.
Scanning electron micrographs of the surface features on the groove surface in 30 wt.% TiB2 for two speed and load combinations: (a) 5 N load and 0.42 rn/s belt speed and (b) 20 N load and 1.67 m/s belt speed.
A1MgBI4
-
Abrasion of engineering ceramics, AIMgBI4--TiB2 composite and other hard materials
313
Fig. 18. SEM micrograph of AIMgB14-70 wt.% TiB2 abraded at 20 N load and 1.67 belt speed.
Figure 17 provides a comparison of the surface features of A1MgB14 - 30 wt.% TiB2 when abraded at two different combinations of loads and speeds in a belt abrasion test [22]. In Fig. 17a, which corresponds to low load and low speed, no evidence of any kind of surface damage is seen and the abraded surface looks smooth and uniform. Figure 17b shows the surface abraded at higher load and higher speed. In addition to plowing, as evidenced by a groove at the top, it exhibits microfragmentation of material. It appears that microcracks initiate at discrete asperity locations due to stress reversal under the action of load and moving abrasive particle. Whereas cracking can be seen at many locations, there is a large area affected by intense cracking. Similar microfragmentation was observed on the abraded surface of AlMgB14-70 wt.% TiB2 as well, as shown in Fig. 18. Here the cracks appear to be mostly intergranular. Because of very high hardness of the material, the grooving features are so shallow that they are almost absent and the abrasive wear is essentially being contributed by microfragmentation.
314
Scratching of materials and applications
Fig. 19. SEM micrograph showing the surface of 13-BN scratched at 70 N load in air.
Figure 19 shows surface damage for 13-BN scratched in air [22]. Here plastic deformation occurs similar to that observed for A1MgB~4 but there is no indication of cracking. In view of the intense cracking observed in case of AIMgBI4-70 wt.% TiB2, it would be concluded that 13-BN is superior to A1MgBI4-70 wt.% TiB2 both in scratch resistance and surface damage by cracking.
Abrasion of engineering ceramics, AIMgBI4--TiB2 composite and other hard materials
315
Fig. 20. SEM micrographs of the surface of/3-BN abraded at 20 N load and 1.67 rn/s belt speed.
Figure 20 shows the features on the surface of 13-BN abraded in a belt abrasion test at 20 N load and 1.67 rn/s belt speed. There is no indication of microfracture on the surface. This is so because the fracture toughness of this material is much higher than that of TiB2-modified AIMgB14. There is some indication of plastic deformation. These two factors account for a very low material removal rate in the abrasion of this material.
DISCUSSION Both in single-point scratch test and Abrasive machining test, it was found that there is a minimum load below which material removed could not be measured. The minimum loads differed for every material, showing a dependence on material properties. The latter also accounted for different material removal rates which were higher for alumina than for silicon nitride. The two mechanisms suggested in the literature [24, 14] for material removal are based on plastic deformation and fracture mechanics. The model for plastic deformation assumes the penetration of an asperity into the surface and provides a linear relationship between wear rate and load [ 14] as below: V/L = kP/H,
316
Scratching of materials and applications
where V is the volume of material removed, L the length of travel, P the load, and H the hardness of material. Here, k is the wear coefficient which should remain constant unless the mechanism of wear changes. This would imply a linear relationship between the material removal rate and load. Fig. 4 shows that this is not the case. A model based on the fracture-dominated process of material removal was proposed by Xu and Jahanmir [24]. They modeled the material removal from scratch tests in ceramics based on fracture mechanics. They considered that microcracks initiated from grain boundary flaws as a result of loading and that the driving force for propagation of these microcracks was the local grain boundary stress. They derived the following equation: dV/dn = A C(E4/5/Hg/5)p2 t m, where dV/dn is the volume of material removed per pass, H the hardness, P the load, E the modulus of elasticity, l the grain size, and A and C are constants that depend on the experimental conditions such as relative humidity, lubrication, sliding speed, surface roughness, and indenter shape. This equation implies that material removal rate is directly proportional to the load squared if other variables are held constant. The curves in Fig. 4 were fitted according to the relationship dV/dn ot 1>2. The coefficients of determination R 2 ranged from 83.33% to 98.5%. The subsurface damage in alumina in Belt abrasive machining was also greater than in silicon nitride. In addition to the behavior discussed above, temperature rise during abrasive machining is a major factor. Increased contribution to damage would be expected from the thermal effects during machining of the low thermal conductivity material. The role of lubricant in the subsurface damage is significant, as they bring down the temperature at the tool material interface. The lubricants also contribute to higher material removal rates.
CONCLUSIONS In scratch tests, the material removal rates for alumina and silicon nitride increased with increasing load and after a certain load the increases became much more rapid. The material removal rate for silicon nitride was lower than that for alumina both in scratch test and belt abrasion test. With the increase in belt speed in belt abrasion test, the material removal rate decreased because of heating at the contact surface. In both scratching and belt abrasion of A1MgBl4-70wt.% TiB2, the mechanism of failure was micro-fragmentation. There was no indication of cracking in 13-BN in both scratching and belt abrasion, instead plastic deformation was exhibited on the surface. The abrasion resistance of A1MgB~4-70wt.% TiB2 was found to be superior to that of WC+Co and SiC but lower than that of 13-BN. The existence and the nature of the cracks in ceramics are controlled by the environment and the mechanical properties of the material.
Abrasion of engineering ceramics, AIMgB I4-TiB2 composite and other hard materials
317
REFERENCES ~
2. 3. 4. .
6. 7. .
9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.
Khrushchov, M. M. and Babichev, MA. (1958) Friction and Wear in Machinery 12, 1. Gahr, K. H. Z. (1987) Microstructure and Wear of Materials. Elsevier, Amsterdam, 180. Fisher, T. E., Anderson, MP. and Jahanmir, S. (1989), J. Am. Ceram. Soc. 72, 252. Quercia, G., Grigorescu, I., Contreras, H., Di Rauso, C. and Gutierrez-campos, D. (2001), J. Refractory Metals and Hard Materials 19, 359. Mao, D. S., Li, J., Guo, S. Y. and Mao, Z. Y. (1997), Wear 209, 153. Nathan, G. K. and Jones, W. J. D. (1966) Proc Instn Mech Engrs 181, 215. Avery, H. S. (196 l) The Measurement of Wear Resistance. Case Report 340- l 0, Dept. Report 9-AE-134, American Brake Shoe Company. Lawn, B. R. (1975), Wear 33, 369. Swain, M. V. (1975), Wear 35, 185. Hokkirigawa, K. (1991), Wear 151,219. Zhang, B., Tokura, H. and Yoshikawa. (1988), M. J. Mater. Sci 23, 3214. Ajayi, O. O. and Ludema, K. C. (1988), Wear 124, 237. Moore, M. A. and King, F. S. (1980), Wear 60, 123. Hutchings, I. M. (1992) Tribology: Friction and Wear of Engineering Materials, CRC Press, Boca Raton, FL. Gee, M. G. (2001), Wear 250, 264. Cook, B. A., Harringa, J. L., Lewis, T. L., Russell, A. M. and Lee, Y. (2004), J. Adv. Mater 36, 56. Cook, B. A., Harringa, J. L., Lewis, T. L. and Russell, A. M. (2000), Scripta Materialia 42, 597. Lewis, T. L., Cook, B. A., Harringa, J. L. and Russell, A. M. (2003), Mater Sci Eng A 351, 117. Annual Book of ASTM Standards. Vol. 03.02, Wear and Erosion; Metal Corrosion, ASTM Intl., West Conshohoken, Pennsylvania. Desa, O. and Bahadur, S. (1999), Wear 225-229, 1264. Desa, O. and Bahadur, S. (2001), Wear 251, 1085. Ahmed, A. (2005) MS Thesis, Mechanical Engineering Department, Iowa State University, Ames. Ahn, H., Wei, L. and Jahanmir, S. (1996), J. Eng. Mater. Tech., ASME Trans. 118, 402. Jahanmir, S. and Xu, H. H. K. (1995), J. Mater. Sci. 30, 2235.
318 INDEX
Abrasive: wear, 129-31,134, 271,273,280, 290 wear modeling, 124, 125 Adherence, 223 Adhesion, 56, 57, 61, 62, 65, 73, 78, 136, 138-42, 147-51, 157-60, 163, 164, 166, 210, 211,213-18, 220, 222,232, 233, 238, 239, 250, 251,276 Apparent friction, 85 Appearance, 103, 113, 124 Coating, 1, 3, 5, 6, 8-16, 18, 19, 21,23, 56, 57, 59, 63, 65, 67, 68, 73-5, 82, 136, 140, 142, 149, 164-6 Cohesive strength, 186, 187, 190, 194, 195, 197-200, 204, 208, 210 Cone, 22, 23, 25, 30-7, 39, 40, 43-9, 51,53 Contact angle, 1-12, 14-23, 25 Cracking, 56, 67, 71-3, 76, 82 Critical: force for cracking, 166 load, 136--44, 147, 149, 151,152, 155, 156, 158, 159, 164, 210, 213, 215-18 plastic strain, 186, 189-97, 208 Cross-linking, 56 Damages, 56, 65, 66 Delamination, 56, 65, 67, 71-3, 82 DLC coating, 223-5,227,228, 230, 231 Elastomers, 56, 57, 59 Finite element analysis, 210 Friction, 56-9, 61-3, 67-72, 77, 82, 85-7, 89, 90, 92-5, 97-101 coefficient, 1-5, 8, 10-12, 16, 19-23, 25,262 Gloss, 102-5, 107, 108, 114-21 Groove, 1, 2, 4-6, 12, 13, 15, 21-3, 25 Hardness, 22-6, 33, 39, 40, 43-5, 47, 49,51,53
High nitrogen steels, 280 JKR theory, 56 Laser scanning confocal microscopy, 103 Machinability, 280, 281,289, 291 Mar resistance measurement, 166, 167, 169, 174, 175, 186 Mechanical properties, 103-5, 113, 121 Micro: fracture, 186-97, 199, 200, 204, 208 scratch, 210, 215, 223 Microstructural integrity, 186, 188, 195, 199, 204 Nano: indentation and scratching, 166, 168, 169, 175, 177, 180-84, 186 scratch, 85-8, 92, 93, 95, 96, 98 Nanotribology, 85 Nature of surface, 262 Optical scattering, 102-5, 107-10, 114, 116-21
Parallel overlapping scratch, 186, 190, 193, 196-99, 201,203-5 Poly(amide) fibre, 85 Poly(dimethylsiloxane) (PDMS), 56 Poly(n-butyl acrylate) (PnBA), 56, 57, 59 Polymers, 124, 125, 127-34 Pre-sputtering time, 232, 234, 236, 238-45, 248, 249 Recovering, 1 Representative strain, 22, 25, 38, 51,52 Roughness, 1, 2, 4-6, 9-11, 13, 15, 21, 23 Scratch, 22, 26, 38, 42, 56, 250, 254, 258, 261 inclined, 262, 264, 265,267,277 resistance, 1-5, 9, 13, 15-7, 19, 23, 24, 166, 186, 280, 281,289, 291 resistance measurement, 166
C~ ~
i~m ~l ~
~
.i