Surface & Coatings Technology 206 (2011) 1860–1866

Contents lists available at SciVerse ScienceDirect

Surface & Coatings Technology

j ourna l homepage: www.e lsev ie r .com/ locate /sur fcoat

Interface characterizations of diamond-coated tools by scratch testingand simulations

P. Lu a, X. Xiao b, M. Lukitsch b, A. Sachdev b, Y.K. Chou a,⁎a Mechanical Engineering Department, The University of Alabama, Tuscaloosa, Alabama, USAb Research & Development Center, General Motors Corporation, Warren, Michigan, USA

⁎ Corresponding author at: The University of Alabama,Mment, 290 Hardaway Hall, 7th Ave., Tuscaloosa, Al 35480044; fax: +1 205 348 6419.

E-mail address: kchou@eng.ua.edu (Y.K. Chou).

0257-8972/$ – see front matter © 2011 Elsevier B.V. Alldoi:10.1016/j.surfcoat.2011.08.022

a b s t r a c t

a r t i c l e i n f o

Available online 28 August 2011

Keywords:Diamond-coated carbide toolsFinite elementInterfaceScratch testing

In this study, micro-scratch tests were conducted on diamond-coated tungsten-carbide substrates to investi-gate coating adhesion. During the scratch testing, high intensity acoustic emission (AE) signals can be clearlydetected when the coating delamination occurs. It is also found that the tangential force increase graduallywith the normal force, but fluctuates significantly when the critical load of coating delamination is reached.A three-dimensional (3D) finite element (FE) model with a cohesive-zone interface was developed tosimulate the scratch process, and by comparing with the delamination critical load from the experiment,the interface characteristic length, the maximum strength and the fracture energy can be obtained. The pre-liminary results indicate that it is feasible to use the FE simulation combined with scratch tests to evaluatethe coating interface behaviors.

echanical Engineering Depart-7-0076, USA. Tel.: +1 205 348

rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Applying hard coatings such as diamond on cutting tools is a novelapproach for enhancing the tool lifetime and improving the machiningperformance because of diamond's exceptional tribological properties.Comparing to synthetic polycrystalline diamond (PCD) tools, whichare commonly used in the manufacturing industry, diamond-coatedtools can bemore cost-effective, also flexible in various tool geometriesand sizes, e.g., micro-scaled drills [1]. However, in machining applica-tions, diamond coating delamination remains the primary wear modethat often results in catastrophic tool failures [2,3]. Coating delamina-tion is due to insufficient adhesion between the coating and substrateand excess thermo-mechanical loads from services such as machining.Different substrate surface treatments have been developed to enhancediamond coating adhesion. On the other hand, reliable methods tocharacterize the coating interface behaviors are necessary in order toquantitatively compare the adhesion of differently treated diamond-coated tools, and to further optimize interface approaches.

There have been several techniques used to examine the adhesionperformance of coatings [4]. Scratch testing is one of the mostpractical methods of evaluating the adhesion of a hard and thincoating on a substrate [5,6], since it is reliable and simple to performwith no special specimen shape or preparation required. Adhesion is

measured when a critical normal load is reached at which coating fail-ure occurs. Assuming that the failure is an adhesive mode, this criticalload is considered as a measure of the coating–substrate adhesionstrength [7,8]. During scratch testing, a spherical indenter tip slidesover the surface of the coating to generate a groove under an incremen-tal or constant normal load. The tangential force can be measured dur-ing the test and the morphology of the scratches is typically observedsimultaneously or afterwards. When the average compressive stressover an area in the coating exceeds a critical value, the coating detachesfrom the substrate to lower the elastic energy stored in the coatingwhile subject to the loading [9]. The work of adhesion at the interfacebetween the coating and substrate is equal to the energy release ratefrom coating detachment and this rate is a function of the meancompressive coating stress over the detaching area at the instant of de-tachment. Thus, the critical average compressive coating stress respon-sible for the detachment could be a measure of coating–substrateadhesion. However, diamond coatings are very brittle. While a coatingmaywithstand compressive stresses induced by the indenter to a cer-tain extent, it may fracture if a high tensile or shear stress field is in-duced simultaneously, in particular, at the interface such asdelamination [10].

The objective of this study is to better understand the adhesioncharacteristics of diamond-coated carbide tools using togethermicro-scratch testing and finite element simulations. The criticalload for coating delamination and the corresponding process signalswere focused. In addition, a three-dimensional (3D) finite element(FE) model was developed to simulate the scratch process, with em-phasis on the interface characteristics, and to evaluate the adhesionbetween the coating and the substrate.

Fig. 1. Digital microscopic images of scratch grooves on the sample.

1861P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866

2. Experimental details

For the diamond-coated specimens, the substrates were made ofcobalt cemented tungsten carbide (WC-Co)with 6% Co and submicroncarbide grains. Before the coating process, the substrates were cleanedand chemically etched, using the Murakami agent, to remove the sur-face cobalt. Diamond filmswere deposited, conducted by University ofSouth Florida, on the pre-treated inserts by a hot filament chemicalvapor deposition (HFCVD) system (Blue Wave Semiconductors)using a CH4/H2 gas flow ratio of 0.05 (5.0 SCCM CH4 flow rate),pressure of 20 Torr and about 800 °C substrate temperature, and a fil-ament voltage of 85 V in a two filament arrange. The deposition timewas 9 h and resulted in a coating thickness around 4 μm. The surfaceroughness of the samples was measured by a profilometer and wasabout 0.38 μm Ra. The as-received coated samples were cleaned bysolvent in an ultrasonic bath before scratch testing. A Micro-scratchtester from CSM Instruments, model Micro-Combi, was used for theexperiments at the room temperature. The diamond indenter had a

(a) Overall model

(b) Details around the ind

Coating

Substrate

3000 m

R=50 m

4m

60

m

Fig. 2. 3D FEA model for the scratch test: (a) overa

tip radius of 50 μm. The scratch speed was 2 mm/min with the pro-gressive loading method. The scratch length for each test was 5 mm.During the scratch test, tangential forces, acoustic emission (AE) sig-nals, and the depth of the scratched track were acquired. A KEYENCEdigital microscope (VHX-600X)was used to observe the scratch tracksand delamination areas after the test. In addition, a white-light inter-ferometer (WLI) was used to acquire the morphology of scratchgrooves. To determine the critical load of the tested diamond-coatedtools, scratch tests were carried out with a progressive load. In addi-tion, different maximum normal forces were applied to examine theloading rate effect. Fig. 1 shows the overall images of 5 scratch groovesat the corresponding load (1: 10 N, 2: 15 N, 3: 20 N, 4: 25 N and 5:30 N). A few very light rubbing marks were noted on the samplesurfaces. However, they did not appear to affect the scratch responsebased on the forces data collected. In addition, the Indenter tip wasalso checked after each scratch and show slight increase in the tipdiameter and very little visible damage.

3. Scratch process simulations

In order to study the interface behavior, a 3D FE model was devel-oped to study the scratch response of diamond-coated carbides byusing ABAQUS 6.8. A half model was used because of the symmetry ofthe problem. The geometric model of the specimen shown in Fig. 2awas 3000 μm in length and 200 μm in width, which was smaller thanthe actual sample size in order to be computationally efficient. Thethickness of the diamond coating and the carbide substrate were 4 μmand 60 μm, respectively. The diamond indenter had a tip radius of50 μm and was assumed rigid because of its relatively larger size com-pared to the coating. Fig. 2b shows the details of the geometric modelaround the indenter. The smallest mesh size underneath the indenterwas 0.6 μm for the coating and 5.0 μm for the substrate. Further, acohesive layer with zero thickness was included to model themechanical behavior of the interface between the coating and thesubstrate.

The materials of the specimen, both the coating and the substrate,were modeled as an elastoplastic solid with isotropic hardening.

enter

Indenter

Cohesive Zone

200

m

ll model, and (b) details around the indenter.

Table 1Properties of the substrate and coating for the diamond-coated cutting tools.

Part Material Young'smodulus/GPa

Poisson'sratio

Yieldingstrength/GPa

Hardeningexponent

Substrate WC–6 wt.% Co 619.5 0.24 3.605 0.244Coating Diamond 1200 0.07 26.7 0.23

1862 P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866

According to the Ramberg–Osgood law, the stress-strain relation can bedescribed as

σ ¼Eε

σyssy

!nε ≤ εy

ε ≥ εy;

8><>:

where εy ¼ σy

E, and the strain-hardening exponent is generally in the

range of 0 to 0.5 [11]. The yield strength of WC–6 wt.%Co has beenreported in the range of 1.45 to 5.76 GP [12], and the averagedvalue, 3.61 GPa, was assumed in this study. Other properties of thecarbide substrates are identical to those in a previous study [13].The yield strength and hardening exponent of the diamond wereobtained by the method proposed by Suresh [14–16] using the experi-mental data from nanonindentation testing on diamond coatings [17].In this study, the Young's modulus of the coating was assumed as1200 GPa; other properties of the coating and substrate are listed inTable 1. As the coating surface roughness will affect the contact condi-tion during scratching [18], the frictional coefficientwas set as 0.1 to ac-count for the coating surface roughness. The cohesive zone propertiessuch as the maximum normal strength and characteristic length werederived from a previous study [19], nominally, 543 MPamaximumnor-mal strength, 1.41 μm characteristic length and 384 J/m2 fracture ener-gy density. The interface properties would later be adjusted toapproximate the critical load observed from the experiments. Duringsimulations, the spherical indenter was first pressed downward in con-tact with the coating, then slid along the surface of the coating, at aspeed of 500 μm/s, with a linear increase in depth, and the sliding dis-tance was 500 μm. The final load was about 15 N. At the end, the sliderwas raised upward until fully unloaded.

4. Results and discussion

As an example (20 N maximum load), Fig. 3 shows the AE signaland tangential force (Ft) vs. the applied normal load (Fn) during thescratch test. From the figure, it is observed that the tangential force in-creases smoothly when the normal force is less than 4.3 N, but variesconsiderably once the load exceeds 4.3 N. It is also noted that anabrupt amplitude increase of AE signals (at Spot 1) appears at theload around 4.3 N, followed by a series of continuous high-amplitude

Spot 1

Fig. 3. Acoustic emission (AE) and tangential force (Ft) vs. normal load (Fn) for a maxi-mum load of 20 N.

AE peaks. This implies coating delamination initiation occurred around4.3 N.

After testing, the scratched groove was observed in the digital mi-croscope at 500×. From the relation between the load and the scratchdistance, the location corresponding to the high force variation (atSpot 1) can be calculated and verified whether coating delaminationhas initiated at that point. Fig. 4a shows the digital microscopicimage at Spot 1 and the corresponding normal load is around 4.3 N.It can be clearly confirmed that at such a force, coating delaminationhas initiated, visibly exposing the substrate layer of WC, near Spot 1.Fig. 4b shows the digital microscopic image at the end of the scratchtest. It is shown that coating delamination continued, once initiated,to the end of the final load, with an increased delamination width.

Using the approach described earlier, the corresponding criticalload of delamination for every scratch condition (different maximumloads) can be estimated. It is found that the critical load of delamina-tion initiation is around 4 to 6 N for all tested loads, indicating thatthe delamination critical load is not sensitive to the load rate appliedin the experiment. Thus, it can be concluded that the evaluateddiamond-coated substrates with a coating thickness of 4 μm has a de-lamination critical load in the range of 4 to 6 N. Note that the studiedsample was form a trial condition, and thus, the critical load valuedoes not necessarily represent common diamond-coated carbides.

The white-light interferometer has also been used to obtain quan-titative geometric information of the scratched tracks. Fig. 5a shows a

Fig. 4. The digital microscopic images of (a) Spot 1 around 4.3 N, and (b) the end of thescratch for scratch test under maximum load of 20 N.

1863P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866

3D surface topography of part of the scratched track, at about 3 mmdistance from the beginning (corresponding to 12 N for the 20 Nmaximum load). It shows coating delamination propagated outward.Fig. 5b is the 2D profile analysis of the scratched track at the samelocation. It is estimated that the delamination width is about 88 μmand the depth of the track is about 7 μm.

In the FE simulations, the ideal cohesive zone properties, estimatedfrom the substrate toughness and obtained from a previous study [19],were first tested. For the current study, to approximate the criticalload measured from the experiments, it needs to point out that thecohesive-zone properties were adjusted (1/3 of the upper-bound frac-ture energy and characteristic length). The cohesive-zone characteristicswere estimated to be 0.43 μm characteristic length and 117 J/m2 frac-ture energy density, whichmay be used to represent the interface prop-erty related to the delamination behaviors. Such interface characteristicsmay be utilized to evaluate the adhesion performance of differentdiamond-coated tools. Moreover, the interface characteristics may beincorporated into thermo-mechanical simulations of different processesthat diamond-coated tool may be further subject to, e.g., machining.

(a) 3D morphology

(b) 2D profile analysis

Fig. 5. WLI analysis of scratch track at 3 mm scratch distance for 20 N maximum load

Fig. 6 shows (a) themaximumprincipal stress and (b) the equivalentplastic strain in the coating-substrate specimen at the onset of coatingdelamination. It can be observed that high compressive stresses, around20 GPa, occurred in the coating near the slider and the maximumequivalent strain is 0.082. Moreover, high tensile stresses appear in thewake of the slider. The corresponding load at this location and this in-stant is about 4 N, and the delamination has just initiated, to be shownnext.

Fig. 7 illustrates the normal stress distribution in the cohesive-zone layer at different times: (a) initial scratch, (b) onset of delamina-tion and (c) end of scratch. At initial scratching, though the normalstress of some cohesive-zone elements has reached the maximumnormal strength (543 MPa), the damage evolution has not completedto result in the cohesive zone failure. Thus, the interface is still intact.On the other hand, at around 4 N of normal load, the cohesive-zonelayer has shown some damage initiated (Fig. 7b); the normal separa-tion of some cohesive-zone elements has reached its characteristiclength (0.43 μm). Fig. 7c shows a large delamination area (the areawhere elements disappear) at the end of the scratch simulation. The

, showing coating delamination: (a) 3D morphology, and (b) 2D profile analysis.

(a) Maximum principal stress (unit: 1000 GPa)

(b) Equivalent plastic strain

200

m

64

m

Fig. 6. (a) Maximum principal stress (unit: 1000 GPa) and (b) Equivalent plastic strainat the onset of delamination.

1864 P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866

coating delamination area has spread in both the sliding and trans-verse directions.

Fig. 8 further shows the cohesive-zone behavior (normal stress vs.normal separation) traced along the scratch time at two different lo-cations, specified in Fig. 8a. Fig. 8b is for a location where the interfaceremains attached. The normal stress is first compressive when theslider is approached to that location and quickly changes to tensileand reaches the interface maximum normal strength. Then, the stressis reduced in a short time, but with the normal separation increased.Finally, as the slider moves away from this location, the stress is re-duced, so is the separation, but does not return to zero at the end.On the other hand, Fig. 8c is for a location where the interface hasbeen delaminated. The normal stress vs. separation curve followsthe typical traction-separation behavior and the normal separationhas reached the length characteristic and resulted in the interfacefailure.

5. Conclusions

Scratch testing was conducted on diamond-coated WC-Co sub-strates using a micro-scratch tester. During the scratch tests, the nor-mal force, the tangential force, the acoustic emission signals and thepenetration depth were acquired. After the tests, the scratch trackswere also observed in a digital microscope and analyzed by awhite-light interferometer. Moreover, to characterize the interfacemechanical behavior, a 3D FEmodel including a cohesive-zone inter-face was developed to simulate the scratch response of diamond-coated carbides. The results are summarized as the following:

(1) Coating delamination can be clearly detected by AE signals. Itwas observed that the abrupt AE peak jumps followed by severalcontinuous AE high-amplitude peaks are associatedwith coatingdelamination. The critical load can also be confirmed from thedelamination-onset location and the load-location history.

(2) The delamination critical load of the tested diamond-coated WCtools was in the range of 4 to 6 N, which has been confirmed byrepeated tests. In addition, the critical was not sensitive to theloading rate in the range applied in this study.

(3) From the FE simulations, the interface of the tested diamond-coated carbide substrates has 0.43 μm characteristic length and117 J/m2 fracture energy density. The interface characteristicscan be used to represent the interface properties related to coat-ing delamination. Thus, it is feasible to use the developed FEmodel, combined with scratch tests, to obtain coating interfacecharacteristics, whichmay be further applied to evaluate the ad-hesion performance of different diamond-coated tools.

Acknowledgements

This studywas conducted under anNSF-funded joint project (CMMI0928627) — GOALI/Collaborative Research: Interface Engineered Dia-mond Coatings for Dry Machining, between The University of Alabama,General Motors and University of South Florida. The diamond-coatedsamples were provided by the University of South Florida.

References

[1] J. Hu, Y.K. Chou, R.G. Thompson, Trans. NAMRI/SME 35 (2007) 177.[2] P. Lu, Y.K. Chou, R.G. Thompson, Proc. 2009 Int. Manuf. Sci. and Eng. Conf, 2009,

MSEC2009-84372.[3] F. Qin, J. Hu, Y.K. Chou, R.G. Thompson, Wear 267 (2009) 991.[4] A.J. Perry, Thin Solid Films 107 (1983) 167.[5] S.J. Bull, in: W. Gissler, H.A. Jehn (Eds.), ECSC, EEC, EAEC, 1992, p. 31.[6] H. Ollendorf, D. Schneider, Surf. Coat. Technol. 113 (1999) 86.[7] R. Jaworski, L. Pawlowski, F. Roudet, S. Kozerski, F. Petit, Surf. Coat. Technol. 202

(2008) 2644.[8] S. Nakao, J. Kim, J. Choi, S. Miyagawa, Y. Miyagawa, M. Ikeyama, Surf. Coat. Technol.

201 (2007) 8334.[9] F.X. Liu, F.Q. Yang, Y.F. Gao, W.H. Jiang, Y.F. Guan, P.D. Rack, O. Sergic, P.K. Liaw,

Surf. Coat. Technol. 203 (2009) 3480.[10] Y. Xie, H.M. Hawthorne, Surf. Coat. Technol. 155 (2002) 121.[11] Z. Xu, X. Li, Acta Mater. 53 (2005) 1913.[12] K.L. Mittal, V.S.P. Intl Sci. Netherland. (2001).[13] A. Renaud, J. Hu, F. Qin, Y.K. Chou, Int. J. Mechatron. Manuf. Syst. 2 (2009) 490.[14] A.E. Giannakopoulos, S. Suresh, Scr. Mater. 40 (1999) 1191.[15] U. Ramamurty, S. Sridhar, A.E. Giannakopoulos, S. Suresh, ActaMater. 47 (1999) 2417.[16] T.A. Venkatesh, K.J. Van Vliet, A.E. Giannakopoulos, S. Suresh, Scr. Mater. 42

(2000) 833.[17] S. Chowdhury, E. de Barra, M.T. Laugier, Diamond Relat. Mater. 13 (2004) 1625.[18] F. Deueler, O. Lemmer, M. Frank, M. Pohl, C. Heßing, Int. J. Refract. Met. Hard

Mater. 20 (2002) 115.[19] J. Hu, Y.K. Chou, R.G. Thompson, Surf. Coat. Technol. 203 (2008) 730.

(a) Initial scratch

(b) Onset of delamination

(c) End of scratch.

200

m

Fig. 7. Cohesive zone stress distribution (unit: 1000 GPa) at the interface: (a) initial scratch, (b) onset of delamination, and (c) end of scratch.

1865P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866

(a) Plot to indicate locations of different stress-separation curves

(b) Location without delamination

(c) Location with delamination

for (b)for (c)

Top view of the cohesive -zone interface

200

m

Fig. 8. Normal stress vs. separation behavior along the scratch time: (a) Plot to indicate locations of different stress-separation curves, (b) location without delamination, and(c) location with delamination.

1866 P. Lu et al. / Surface & Coatings Technology 206 (2011) 1860–1866