Scratching of Materials and Applications, Volume 51 (Tribology and Interface Engineering) (Tribology and Interface Engineering)

01 Contents Chapter 1. Mechanical Analysis of the Scratching Properties of Coated Polymers (C. Gauthier et al.). Chapter...

Author: Sujeet K. Kumar Sinha

58 downloads 1692 Views 19MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

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


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


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


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.


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


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.


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

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


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


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.


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


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.


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


Rear part

.t x Frontal part ,, _ ~ .. Increase ',. ...... !...... " ,,fl i n X

--,,

h tan

"

m

o,"

~

"~ " -

X60 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 Xl). 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].


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.


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

-

+

+

+

+

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].


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].


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.


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


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.


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


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).


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 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


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


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


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


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

.

.

.

.


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.


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


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


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


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


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.


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


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


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


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,


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


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).


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


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


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


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


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


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


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


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.


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.


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.


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


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.


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


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.



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.


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


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.


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


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)


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


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.


244

Fig. 8.


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


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.


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


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


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


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


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


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-

,,=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~

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

Scratching of Materials and Applications, Volume 51 (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Tribology of Elastomers, Volume 47 (Tribology and Interface Engineering)

Tribology of Metal Cutting, Volume 52 (Tribology and Interface Engineering)

Acoustic Emission in Friction, Volume 53 (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Molybdenum Disulphide Lubrication (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Tribochemistry of Lubricating Oils, Volume 45 (Tribology and Interface Engineering)

Tribology of Interface Layers

Engineering Tribology (Tribology Series)

Engineering Tribology (Tribology Series)

Tribology of Polymeric Nanocomposites, Volume 55: Friction and Wear of Bulk Materials and Coatings (Tribology and Interface Engineering)

Coatings Tribology, Volume 56, Second Edition: Properties, Mechanisms, Techniques and Applications in Surface Engineering (Tribology and Interface Engineering)

Tribology & Biophysics of Artificial Joints, Volume 50 (Tribology and Interface Engineering)

Materials and Surface Engineering in Tribology

Materials and Surface Engineering in Tribology

Engineering Tribology, Second Edition

Engineering Tribology, Third Edition

Mechanical Tribology: Materials, Characterization, and Applications

An Intelligent System for Engine Tribological Design, Volume 46 (Tribology and Interface Engineering)

Engineering The User Interface

Luminous Chemical Vapor Deposition and Interface Engineering

Materials for Tribology

Tribology of Miniature Systems (Tribology Series)

Tissue Engineering, Volume 8: Fundamentals and Applications (Interface Science and Technology)

Materials science and engineering

Composite Materials: Science and Applications (Engineering Materials and Processes)

Interface Science in Drinking Water Treatment, Volume 10: Theory and Applications (Interface Science and Technology)

Tribology of Ceramics and Composites: Materials Science Perspective

Interface

Interface

Interface

Scratching of Materials and Applications, Volume 51 (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Tribology of Elastomers, Volume 47 (Tribology and Interface Engineering)

Tribology of Metal Cutting, Volume 52 (Tribology and Interface Engineering)

Acoustic Emission in Friction, Volume 53 (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Molybdenum Disulphide Lubrication (Tribology and Interface Engineering) (Tribology and Interface Engineering)

Tribochemistry of Lubricating Oils, Volume 45 (Tribology and Interface Engineering)

Tribology of Interface Layers

Engineering Tribology (Tribology Series)

Engineering Tribology (Tribology Series)

Tribology of Polymeric Nanocomposites, Volume 55: Friction and Wear of Bulk Materials and Coatings (Tribology and Interface Engineering)

Coatings Tribology, Volume 56, Second Edition: Properties, Mechanisms, Techniques and Applications in Surface Engineering (Tribology and Interface Engineering)

Tribology & Biophysics of Artificial Joints, Volume 50 (Tribology and Interface Engineering)

Materials and Surface Engineering in Tribology

Materials and Surface Engineering in Tribology

Engineering Tribology, Second Edition

Engineering Tribology, Third Edition

Mechanical Tribology: Materials, Characterization, and Applications

An Intelligent System for Engine Tribological Design, Volume 46 (Tribology and Interface Engineering)

Engineering The User Interface

Luminous Chemical Vapor Deposition and Interface Engineering

Materials for Tribology

Tribology of Miniature Systems (Tribology Series)

Tissue Engineering, Volume 8: Fundamentals and Applications (Interface Science and Technology)

Materials science and engineering

Composite Materials: Science and Applications (Engineering Materials and Processes)

Interface Science in Drinking Water Treatment, Volume 10: Theory and Applications (Interface Science and Technology)

Tribology of Ceramics and Composites: Materials Science Perspective

Interface

Interface

Interface

Recommend Documents