Aa

JJa

b

c

a

ARRAA

KAACSC

1

rrmbbn

dsstbp

0h

Journal of Materials Processing Technology 214 (2014) 1467–1481

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

jo ur nal ho me page: www.elsev ier .com/ locate / jmatprotec

dhesion phenomena in the secondary shear zone in turning ofustenitic stainless steel and carbon steel

. Gertha, F. Gustavssona, M. Collinb, G. Anderssonb, L.-G. Nordhc,

. Heinrichsa, U. Wiklunda,∗

Tribomaterials Group, Uppsala University, Box 534, SE-751 21 Uppsala, SwedenAB Sandvik Coromant, R&D Materials and Processes, 126 80 Stockholm, SwedenUddeholms AB, 683 85 Hagfors, Sweden

r t i c l e i n f o

rticle history:eceived 26 October 2012eceived in revised form 10 January 2014ccepted 27 January 2014vailable online 3 February 2014

eywords:dhesionustenitic stainless steelarbon steelurface roughness

a b s t r a c t

This paper aims to increase the understanding of the adhesion between chip and tool rake face by studyingthe initial material transfer to the tool during orthogonal machining at 150 m/min. Two types of workmaterial were tested, an austenitic stainless steel, 316L, and a carbon steel, UHB 11. The tools used werecemented carbide inserts coated with hard ceramic coatings. Two different CVD coatings, TiN and Al2O3,produced with two different surface roughnesses, polished and rough, were tested. The influences ofboth tool surface topography and chemistry on the adhesion phenomena in the secondary shear zonewere thus evaluated. Extensive surface analyses of the inserts after cutting were made using techniquessuch as Scanning Electron Microscopy (SEM), Energy Dispersive Spectroscopy (EDS), X-ray PhotoelectronSpectroscopy (XPS), and Transmission Electron Microscopy (TEM). As expected, cutting in the stainlesssteel resulted in a higher amount of adhered material, compared to cutting in the carbon steel. Remnants

utting tools of built-up layers were found on the surfaces of the 316L chips but not on the UHB 11 chips. Moreover,it was shown that for both materials the tool roughness had a profound effect, with the rougher surfacescomprising much higher amounts of adhered material than the polished ones. Non-metallic inclusionsfrom both types of workpiece steels accumulate in the high temperature area on the inserts. The generaltendency was that higher amounts of transferred material were found on the TiN coating than on theAl2O3 coating after cutting.

. Introduction

The adhesion properties between work material and tool mate-ial in the secondary shear zone are of significant importance bothegarding the energy spent in the cutting process and the perfor-ance of the cutting process itself. Although these properties have

een studied for many years, and significant new knowledge haseen gained on the subject, the details of the processes are stilleither fully understood nor exploited in tool design.

Austenitic stainless steels are generally known to be moreemanding to machine than plain carbon steels. The austenitictainless steel adheres strongly to the tool and chips often remaintuck to the tool after cutting. A high tendency to work harden,

ogether with a low thermal conductivity compared to plain car-on steels, cause a high temperature in the flow-zone, whichromotes more extensive chip–tool interactions. The temperature

∗ Corresponding author. Tel.: +46 18 4713092; fax: +46 18 4713572.E-mail address: urban.wiklund@angstrom.uu.se (U. Wiklund).

924-0136/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.jmatprotec.2014.01.017

© 2014 Elsevier B.V. All rights reserved.

distribution on the tool rake face has, however, the same characteras when cutting other steels, i.e. a comparatively cool region at thecutting edge and a temperature maximum some distance from thecutting edge, as described by Trent and Wright (2000). Adding oxideforming elements such as Ca, or other elements such as S, to stain-less steels have been shown to improve their machinability. Fangand Zhang (1996) described the role of Ca in a turning tests with aTiC carbide tool cutting in calcium deoxidised free cutting stainlesssteel (Ca content < 0.01% and S content ≤ 0.1%). It was shown thatan adhered layer was formed on the tool surface by the extrusion ofnon-metallic inclusions in the steel. The layer consisted mainly of(CaO, MgO, MnO)–Al2O3–SiO2 and grew thicker with cutting lengthand increased cutting speed. The formation process of the layer wassummarized as “viscous and extruding, adhering and coating, hard-ening and thickening, and then relatively stable”. After becomingstable, it was assumed to protect the tool from abrasive, adhe-sive and diffusive wear, and to lower the temperature in the flow

zone.

There are a number of models of the chip–tool contact, and themajority nowadays assumes a stagnant seized layer at the tool–chipinterface, with more or less full contact between chip and tool

1 cessin

m(cntcmicaatWsrlTadpctIolcts

sccrwms

afofitTizmbbqtTattt

(efabaedi(

468 J. Gerth et al. / Journal of Materials Pro

aterial, as proposed by Trent and Wright (2000). Qi and Mills1996) proposed a new dynamic flow zone model based on theoncept of a “cutting interface”. This interface is not located at theormal interface where the speed of the chip is zero, but some dis-ance into the chip, where the maximum shear strain rate of thehip is found. This dynamic model was proposed to explain the for-ation of transfer layers at the tool–chip interface and was applied

n a case study of three different austenitic stainless steels; a cal-ium deoxidised free cutting austenitic stainless steel (Ca < 0.01%nd S ≤ 0.1%), a resulphurized austenitic stainless steel (0.17% S),nd an ordinary austenitic stainless steel. An adhered layer, con-aining high concentrations of Al, Ca, Si, and S, was found on the

C–TiC–Co tool after machining the calcium deoxidised stainlessteel while a MnS layer adhered to the tool when turning in theesulphurized stainless steel. When machining the ordinary stain-ess steel, a layer of work material was generated on the tool surface.he layer formed by the calcium deoxidised steel was stable alsot higher cutting speeds and offered protection from adhesive,iffusive and abrasive tool wear. The MnS layer formed by the resul-hurized stainless steel was found to be viscous and thin at higherutting speeds, implying that the “cutting interface” was located athe tool surface which accelerated the abrasive and diffusive wear.t did, however, prevent welding and seizure of workpiece materialnto the tool which reduce the adhesive wear. The easily shearedayer could also act as a solid lubricant, reducing friction force andutting temperatures. In contrast, the work material adhering to theool when machining the ordinary stainless steel suggested risks ofevere adhesive wear.

Katayama and Hashimura (1995) studied the interfacial adhe-ion between different cutting tool materials and free-cutting lowarbon steels. They found that the frictional forces measured duringutting were connected to the amount and type of adhered mate-ials, Fe and/or MnS, found on the tool surface. The tool materialas shown to influence the adherence of elements and hence theeasured friction force. Little attention was, however, given to the

urface finish of the tools.Wiklund et al. (2012) studied the initial material transfer of

ustenitic stainless steel onto a TiN coated cutting tool by per-orming quick-stop tests, freezing the chip forming process, afternly a short cutting distance. Three different zones were identi-ed when studying the tool and chip after the quick-stop. Close tohe cutting zone, tool fragments were found sticking to the chip.his implied very strong adhesive forces between tool and chipn this zone during cutting. This zone was followed by a stagnantone further away from the cutting zone where no signs of relativeovement between chip and tool surface was found. The adhesion

etween the tool and the chip was weaker than the cohesion ofoth the tool and the chip material, allowing the separation afteruick-stop to occur at the tool–chip interface. Despite this, indica-ions of an extremely thin layer of adhered steel remaining on theiN surface were found. The third and final contact zone, furthestway from the cutting zone, was described as a region with onlyemporary contact between the chip and the tool. An abundance ofransferred steel, initiated at protrusions in the coating surface andhereafter successively built-up, was found in this region.

Klocke and Krieg (1999) stated that physical vapour depositionPVD) and chemical vapour deposition (CVD) are the two most rel-vant groups of deposition processes for cutting tools. In differentorms their dominance prevails, as illustrated by, e.g. Dobrzanskind Pakuła (2005) and Sokovic et al. (2009). Coatings manufacturedy CVD are often used for high-speed metal cutting where there is

high demand of wear resistance, as described by, e.g. Bouzakis

t al. (2012), while PVD coatings are suitable for applications withemands on edge-line toughness such as milling, as exemplified

n, e.g. the industrial survey on gear cutting tools by Gerth et al.2011). In general, PVD coatings are thinner than coatings applied

g Technology 214 (2014) 1467–1481

by CVD. The CVD technique is commonly employed since it is theonly deposition method that allows economical large-scale produc-tion of high quality alumina coatings as discussed by Ruppi (2005).From this reasons, choosing two CVD tool coatings for this study ismotivated.

Moreover, the CVD tool coatings TiN and Al2O3 are both fre-quently used in industrial machining of steels and they are bothknown to have high wear resistance as described by Holmberg andMatthews (1994) and in studied detail by, e.g. Dearnley (1985).Compared to TiN, Al2O3 is thermally stable, shown by Trinh et al.(2009), has a lower thermal conductivity at high temperatures,as shown by Cahill et al. (1998) and a lower solubility in steel athigh temperatures, which limits the chemical wear as describedby Dearnley, (1985). All this suggest it is the more suitable coat-ing in applications where very high temperatures are involved.Another aspect which could influence the wear characteristics ofthe coating is the material transfer to the tool rake face. This paperreports on several aspects of the adhesion between the chip and therake face of the tool by studying the initial material transfer duringorthogonal machining of two different work materials. Two verydifferent kinds of steels were used; a commercial machinability-improved grade of the austenitic stainless steel 316L and a plaincarbon steel, UHB 11. These were chosen to represent challengingmaterials, 316L, and steels characterized by a good machinability,UHB 11. The influence of the surface roughness of the tool on theinitial material transfer, which has often been neglected in earlierworks, was also studied.

2. Experimental procedure

2.1. Experimental set-up

Cutting tests were performed in a lathe with a setup shown inFig. 1. The workpiece consisted of a hollow cylinder with an innerdiameter of 154 mm and an outer diameter of 160 mm, resultingin a wall thickness of 3 mm. By cutting from the side of the cylin-der (longitudinal turning), the width of cut became equal to thewall thickness. The cutting speed was 150 m/min and the feedrate was 0.154 mm/rev. The cutting and feed forces were mea-sured using a KISTLER 3-component dynamometer (type 9257A).This sensor has a maximum range of 10 kN in the z direction (sen-sitivity −3.5 pC/N) and 5 kN in the x and y directions (sensitivity−7.5 pC/N). The rake and clearance angles were 0◦ and 11◦, respec-tively. Rapid engagement and disengagement of the tool, with theworkpiece rotating at full speed, was used. Data from the force mea-surements proved this method to be capable of controlling the testsdurations at an accuracy of about 0.1 s. Tests with two differentdurations were executed, 1 s and 5 s, and all tests were repeatedtwice.

2.2. Materials

The tools tested were cemented carbide inserts (SPUN 120308,WC + 10.5 wt% Co, hardness 1350 HV3kg) with two different CVDcoatings. The coatings were composed of approximately 3 �mTi(C,N) as an inner layer, and 3 �m of either TiN or �-Al2O3 asa top layer. The inner Ti(C,N) layer was deposited onto the sub-strates at a temperature of 885 ◦C using TiCl4 as Ti precursorreacting with a gas mixture of CH3CN:H2:N2. The TiN top layer wasdeposited at 1010 ◦C using TiCl4 as Ti-precursor reacting with agas mixture of H2:N2, while the �-Al2O3 top layer was depositedat 1010 ◦C using AlCl3 as Al precursor reacting with a gas mixture

of CO2:H2S:H2. The depositions were performed in a Bernex 325reactor.

Coated inserts with two different surface topographies wereprepared. One referred to as “standard rough” where the blanks

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1469

F sert to the right, showing the directions of measured forces and the location of studieda

wacorpdwsToi1

oai1

Table 1Hardness and surface parameters of the coatings.

Hardness (GPa) Surface parameters (�m)

Standard rough Polished

TiN24.3 ± 2.4 Ra 0.37 ± 0.02 0.19 ± 0.01

Rz 7.0 ± 2.0 2.8 ± 0.5Al2O3

ig. 1. Setup of the cutting tests to the left and schematic pictures of the cutting inreas.

ere polished before coating and then used with the surfacess coated. These inserts thus have a surface representing that ofommercial cutting inserts. The other type of inserts were polished,n the rake faces, both before and after coating deposition, and areeferred to as polished. The final polishing of these coatings waserformed using a 9 �m diamond suspension followed by a 1 �miamond suspension. The surface parameters presented in Table 1ere measured on 1 mm2 areas. The appearances of the different

urfaces of TiN and Al2O3 are shown in Figs. 2 and 3, respectively.he hardness of each coating was measured using nanoindentationn polished coatings and the results are presented in Table 1. Thendents, fifty on each sample, were made to a constant depth of00 nm using a Berkovich diamond tip.

The work materials used were a machinability-improved grade

f a 316L stainless steel and a plain carbon steel, UHB 11. Theustenitic 316L had a hardness of 210 HB (∼2.16 GPa) and the nom-nal chemical analysis (intervals or max values): 0.03% C, 0.1% N,6–18% Cr, 10–14% Ni, 2–3% Mo, 2.0% Mn, 0.045% P, 0.03% S, 0.75%

Fig. 2. Surface appearance of the TiN coated inserts w

Fig. 3. Surface appearance of the Al2O3 coated inserts w

32.0 ± 2.2 Ra 0.34 ± 0.02 0.19 ± 0.02Rz 5.5 ± 0.4 3.1 ± 0.9

Si, Fe bal. The specific 316L steel used in this work also contained0.0030% Ca, 0.005% Al, and 0.06% V. The UHB 11 had a ferrite-

pearlitic structure and a hardness of 200 HB (∼1.96 GPa), and thenominal chemical analysis: 0.46% C, 0.2% Si, 0.7% Mn, Fe bal. Bothwork materials were hot-rolled into bars and then machined intohollow cylinders.

ith (a) standard rough and (b) polished surfaces.

ith (a) standard rough and (b) polished surfaces.

1470 J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481

F nsertsc

2

(tX2TtCspawAm

3

3

hi

ig. 4. Measured cutting (a and b) and feed (c and d) forces for standard rough iombination.

.3. Analysis of used tools

The analyses were made using Scanning Electron MicroscopySEM, Zeiss 1550, Zeiss DSM 960A), Energy Dispersive Spec-roscopy with an acceleration voltage of 20 kV (EDS; Oxford Aztec-max,), X-ray Photoelectron Spectroscopy (XPS; PHI Quantum000), Transmission Electron Microscopy (TEM and EFTEM; FEIecnai F30 ST, operated at 300 kV), Electron Energy Loss Spec-roscopy in Scanning-TEM mode (STEM-EELS, FEI Titan 80-300ubed equipped with a probe corrector and a HR post-columnpectrometer, operated at 200 kV) and white light interferencerofilometry (WYKO NT 1100, VSI mode with 736 × 480 samplesnd a sampling size of 1.56 �m). TEM samples were preparedith a Focused Ion Beam (FEI Strata DB235) using in situ lift out.s a last step the samples were cleaned with a 5 kV ion beam toinimize the amount of beam damage.

. Results

.1. Force measurements

Both cutting forces and feed forces were generally slightlyigher when turning in UHB 11 than in 316L. The cutting forces

nitiated at about the same level for the two work materials. They

(a and c) and polished inserts (b and d) when turning 316L. Two tests for each

began to decrease after only a fraction of a second when cuttingin 316L, as shown in Fig. 4a and b, while they remained high forUHB 11 during the 5 s duration of the cutting tests. There were onlyminute differences between the forces when using the differentcoatings, i.e. TiN and Al2O3, polished and unpolished.

Turning to the feed forces, there were, however, a few differ-ences. For 316L with standard rough inserts the feed force wasinitially high and decreased with cutting time, as shown in Fig. 4c.The corresponding polished inserts initially gave a lower feed forcethat also changed less during the tests, see Fig. 4d. The polishedTiN gave a feed force that decreased slightly, while the feed forcefor the polished Al2O3 appeared to increase very slightly from aninitially low value.

For the polished inserts, the observed trends hold for cuttingboth 316L and UHB 11. However, the standard rough inserts did notshow the same consistent trends when turning UHB 11 as whenturning 316L. It should also be noted that substantial vibrationsinfluenced the raw data and these have been filtered from the datapresented in Fig. 4 to more clearly show the general trends. Filtering

of the raw data, acquired at a sampling frequency of 10,240 Hz, wasmade in two steps. First by smoothing based on up to 200 adjacentpoints and after that using non-parametric spline smoothing, alloffered by the software.

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1471

Fig. 5. SEM images showing adhered work material, bright areas, on the rake faces of the cutting inserts, after turning in 316L and UHB 11, for 1 and 5 s (SEM, acc. voltage20 kV, backscatter compositional mode). Inserts after machining 316L, (a) standard rough TiN 1 s, (b) polished TiN 1 s, (c) standard rough TiN 5 s, (d) polished TiN 5 s, (e)s lishedT , (n) p

3

3

wpiaeanlf

tandard rough Al2O3 1 s, (f) polished Al2O3 1 s, (g) standard rough Al2O3 5 s, (h) poiN 1 s, (k) standard rough TiN 5 s, (l) polished TiN 5 s, (m) standard rough Al2O3 1 s

.2. Surface analyses of the cutting inserts

.2.1. OverviewSurface analysis using SEM revealed that a higher amount of

ork material adhered to the standard rough inserts than to theolished inserts after turning in either 316L or UHB 11. By compar-

ng the images in Fig. 5, this can be seen to hold for both the TiNnd the Al2O3 coated inserts. The area with adhered work materialxtends further away from the cutting edge after 5 s of turning than

fter 1 s of turning for all material and surface topography combi-ations; compare, e.g. Fig. 5e and g. This means that the contact

ength and contact area increased when the turning time increasedrom 1 to 5 s. However, no consistent differences in contact length,

Al2O3 5 s. Inserts after machining UHB 11, (i) standard rough TiN 1 s, (j) polishedolished Al2O3 1 s, (o) standard rough Al2O3 5 s, (p) polished Al2O3 5 s.

caused by the different types of inserts, could be resolved in theSEM.

The observed contact area between the chip and the rake facecould be divided in to three different regions based on differencesin surface appearance, material transfer and cutting conditions. Theclassification follows a system suggested by Wiklund et al. (2012).A1 is the area adjacent to the cutting edge, which is subject to highcompressive stresses during cutting. A2 is an area about 300 �min on the rake face where the temperature during cutting nor-

mally reaches a maximum. Finally, A3 denotes the area where thechip separates from the rake face and plenty of oxygen is presentduring cutting. It is reasonable to believe that seizure occurredin regions A1 and A2, while A3 was dominated by sliding, which

1472 J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481

F sert ab

wabAoap

iTbdcesm3r

ig. 6. Comparison of the three areas A1, A2, and A3, between polished TiN coated inackscatter compositional mode).

ould be in accordance with work by Trent and Wright (2000) and recent quick stop study by Wiklund et al. (2011). A comparisonetween these areas on polished TiN coated inserts and polishedl2O3 coated inserts after turning for 1 s in 316L, showed that lumpsf work material were present mainly on the TiN coating and inll three areas, see Fig. 6. The largest difference between the twoolished coatings was found in A3.

The same regions on standard rough and polished TiN coatednserts after turning in 316L, now for 5 s, are compared in Fig. 7.he standard rough TiN coating was almost completely coveredy work material in the entire contact area. There were still smallifferences visible between A1, A2, and A3. On the polished TiNoating there was a considerably smaller amount of work material,specially in A2. However, fairly large lumps of work material were

till adhering in A3. Chunks of work material were also found in A1,ost often associated with pits in the surface. The pits were up to

�m deep, but there were no signs of exposed substrate mate-ial, see Fig. 8. This particular type of coating pits was only found

nd polished Al2O3 coated insert after 1 s of turning in 316L (SEM, acc. voltage 20 kV,

on polished TiN which had been machining 316L. Coating detach-ments at the cutting edges and/or notch wear were, however, seenon several other inserts as well. In fact, all inserts showed areas ofexposed cemented carbide at the very cutting edge after 5 s of tur-ning in 316L. It was most severe on the standard rough TiN coatedinserts where also remnants of the chip were found on the cuttingedge. In contrast, the cutting edges of the inserts having machinedUHB 11 were all intact, even after 5 s of cutting.

Clearly, a high roughness, cf. Figs. 2a and 3a, promoted adhe-sive transfer of work material. There was, however, work materialsticking to smaller surface irregularities on the polished surfaces aswell. This is exemplified in Fig. 9 showing an area located withinA2 of a polished Al2O3 coated insert after turning in 316L.

Each lump of work material can be assigned to an underlying

crater in the surface. Moreover, a pattern of ridges and valleys wasclearly visible along the chip flow direction, see Fig. 9. More orless all inserts showed indications of such a pattern, but the fea-ture was most striking on the Al2O3 coated inserts. Topography

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1473

F and pb

mb

3

ieccataywAFaw

ig. 7. Comparison of the three areas A1, A2, and A3, between standard rough (s)ackscatter compositional mode).

easurements suggested that the top to valley distances wereelow 50 nm.

.2.2. Chemical analysesEDS analyses showed that the bright areas in the SEM images

n Fig. 5 contained predominantly work material constituents. Thexceptions were the small areas of high brightness at the veryutting edge of the inserts, which show positions with exposedemented carbide. EDS element spectra were obtained by scanningreas of 0.1 mm2 in the regions A1, A2, and A3 on each type of cut-ing insert after 5 s of turning in 316L and on the polished insertsfter 5 s of turning in UHB 11. These showed that Fe and steel allo-ing elements, in approximately the same proportions as in theork materials, were transferred to the cutting inserts in regions

1 and A3. In region A2, however, some elements were enriched.or inserts turning in the 316L, these elements were mainly Al, Si,nd Ca, as evident when comparing the relative peak height of Feith those of Al, Si, and Ca, in the representative spectra in A1, A2,

olished (p) TiN coated insert after 5 s of turning in 316L (SEM acc. voltage 20 kV,

and A3 shown for polished TiN in Fig. 10. The A3 spectrum mostclosely represents the work material itself. The standard rough TiNcoated insert proved to be an exception, showing high amount ofFe in all regions. The spectra obtained on the polished and standardrough Al2O3 coated inserts, both showed elemental ratios similarto that on TiN, apart from the fact there is a very large contributionto the Al peak from the Al2O3 coating itself, see Fig. 11. For insertsturning in the UHB 11, the enrichments of the steel alloying ele-ments S, Mn, and Al were found in A2 on the polished TiN coatedinsert, see Fig. 12. On the polished Al2O3 coated insert, only verysmall amounts of adhered material were found, and no significantdifference in relative amounts of adhered elements in the threedifferent regions could be resolved.

XPS depth analyses were performed on the polished TiN and

Al2O3 coated inserts that had been turning for 5 s in 316L. Theanalyzed areas had diameters of 150 �m, and depth profiles wereobtained by alternating analysis and material removal by sput-tering using Ar-ions with an energy of 1 kV. The measurements

1474 J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481

Fig. 8. (a) SEM image highlighting coating damage on the rake face of the polished TiN coated cutting insert after machining 316L for 5 s. The insert is orientated with thec ograpc

srStiSi

IA

Fau

utting edge upwards in the picture (acc. voltage 3 kV, secondary electrons). (b) Tophip flow direction).

upported the findings of enrichments of Al, Si, and Ca in the A2egion. The Al2p peak was located at a binding energy 75.3 eV, thei2p at 102.5 eV, and the Ca2p3/2 at 348.3 eV. This indicates thathese elements were chemically bound as oxides, halides or sil-cates, rather than metallically bonded (as metals: Al2p 72.9 eV,i2p 99.3 eV, Ca2p3/2 346.7 eV), considering stated peak positions

n Moulder et al. (1995).

Fe and Cr were found in all areas but differently oxidized.n the outermost surface, Fe and Cr were oxidized in all areas.fter sputtering for some time, metallic Fe and Cr were found

ig. 9. SEM image (acc. voltage 3 kV, secondary electrons) of Al2O3 coated insertfter 1 s of turning in 316L. Here the insert is orientated with the cutting edgepwards in the picture.

hy measurement of an area comprising such coating damages (an arrow indicates

even though oxides were still present. This is clear when com-paring the Cr 2p3/2 peak obtained after only 0.2 min of sputtering,intended to remove surface contaminations from the air, andthe peak obtained after 8 additional minutes of sputtering, seeFig. 13.

After 0.2 min of sputtering, the peak had its maximum at 576 eV,corresponding to the position of Cr in the form of Cr2O3 as shown byLatha et al. (1997). The broadening of the peak at the high energyside indicates the presence of CrO3, which according to Moulderet al. (1995), has a peak just above 578 eV. After the more extensivesputtering, the peak became broader with contributions from bothmetallic Cr (peak around 574 eV, again described by Latha et al.(1997)) and oxidized Cr. The relative contribution from metallic Crto that of oxidized Cr was higher in the area A3 than in A1 andA2, after 8.2 min of sputtering, cf. Fig. 13a and b. The Fe 2p3/2 peakshowed the same evolution upon sputtering, although curve fittingand decomposition of the peaks were more challenging. The spectraobtained in areas A2 and A3 are shown in Fig. 14a and b, respec-tively, with binding energies for metallic Fe, Fe2+ in FeO and Fe3+

in Fe2O3 indicated at positions stated by Moulder et al. (1995) andMarcus and Olefjord (1988). After 0.2 min of sputtering, the peakcould mainly be attributed to Fe2O3. After 8.2 min of sputtering itcontained contributions from metallic Fe, FeO and Fe2O3. As for Cr,the metallic Fe contribution to the peak after 8.2 min of sputteringwas largest in the A3 area. The XPS results presented are repre-sentative also for the Al2O3 coated inserts. However, the signalsobtained from adhered elements on the Al2O3 coated insert imme-

diately became considerably weaker upon sputtering, in contrast tothe corresponding signals obtained on the TiN coated insert. Thisshows that the adhered material in the analyzed areas on Al2O3were thinner than that on TiN.

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1475

Fig. 10. EDS spectra obtained in areas A1, A2 and A3 on polished TiN coated insert after 5 s of turning in 316L (*Si-escape peak of Ti).

3 coat

3

a3at

f

Fig. 11. EDS spectra obtained in areas A1, A2 and A3 on polished Al2O

.2.3. Surface cross-section analyses using TEMCross-sections for TEM analyses were prepared, using a FIB, in

rea A2 of polished TiN and Al2O3 inserts that had been machining16L for 1 s. The areas, from which the TEM samples were extracted,re indicated in Fig. 15. Both cross-sections were made parallel to

he cutting edge, i.e. perpendicular to the chip flow direction.

STEM-EELS spectrum imaging (SI) was performed at the inter-ace of TiN where an adhered steel particle was found. The

Fig. 12. EDS spectra obtained in areas A1, A2 and A3 on polished TiN co

ed insert after 5 s of turning in 316L (*sum peak of Al + O and Al + Al).

elemental maps in Fig. 16 show that the particle, consisting mainlyof Fe and Cr, had an oxide on its surface exposed to air, but noton its surface facing the coating. Ni was not included in the EELSanalysis shown in Fig. 16 due to the selection in energy range.However, a STEM-EDS analysis of the sample showed that the dis-

tribution of Ni correlated well to the distribution of Fe, i.e. Ni wasonly found in adhered steel particles. Some oxygen was, however,seen in parts of the interface between the particle and the coating,

ated insert after 5 s of turning in UHB 11 (*Si-escape peak of Ti).

1476 J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481

Fig. 13. The Cr 2p3/2 spectrum obtained in area A2 (a) and area A3 (b) on polished TiN after 5 s of turning in 316L, after minute sputtering (0.2 min) and more extensives 3.

aCottbioiFoct

Fs

puttering (8.2 min), with curve fitting for metallic Cr, Cr3+ in Cr2O3 and Cr6+ in CrO

t positions where also Cr, V, and Ca were present. Interestingly,r was also detected outside the particle, apparently smeared outver the surface. V and Ca were also seen covering the surface ofhe coating. Matching of the elemental maps shows strong correla-ions between V, Ca, and O under the particle and strong correlationetween Cr and O at the side of the particle. No correlation at all

s found between Ti and O. The pattern shown in Fig. 9, consistingf ridges and valleys parallel to the chip flow direction, on the pol-shed Al2O3 coating is very striking in cross-section, as shown in

ig. 17. A featureless layer of varying thickness (10–100 nm), with-ut any crystal information, was found in the very surface region,learly distinguishable from the crystalline Al2O3 further down inhe cross-section. EFTEM and STEM-EELS-SI showed areas with Ca

ig. 14. The Fe 2p3/2 spectrum obtained in area A2 (a) and area A3 on polished TiN afteputtering (8.2 min), with binding energies for Fe, Fe2+ in FeO and Fe3+ in Fe2O3 indicated

covering the surface, see Fig. 18a, although the major part of thesample showed no elements adhered on top of the Al2O3 coatingas exemplified in Fig. 18b.

3.3. Surface analyses of chips

By studying the surface of the chip, i.e. the surface that has beenin contact with the tool rake face, an obvious effect of the tool sur-face roughness was revealed. Chips produced by the standard rough

inserts comprised ridges and grooves running parallel to the slidingdirection of the chip while the surfaces of the chips produced by thepolished inserts appeared much smoother, cf. Figs. 19 and 20. Eventhough it was not verified that the topography on the chip was a

r 5 s of turning in 316L (b), after minute sputtering (0.2 min) and more extensive.

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1477

Fig. 15. Areas, within area A2, where the TEM samples were extracted. Polished TiN insert (a) and polished Al2O3 insert (b) after 1 s of turning in 316L.

F iN inseu S edge

di3

oi

F1o

ig. 16. STEM-EELS-SI acquired around an adhered steel particle on the polished Tpwards. Cr, Ca, O, and V are present outside the particle, on top of the coating. EEL

irect replica of the insert topography, it is obvious that a roughnsert surface induced a rough chip surface. On chips consisting of

16L, surface irregularities of a different character were also found.

These were tongues of material, less than 0.5 �m thick, smearedut in the sliding direction of the chip, see Figs. 19 and 21. Thisndicates that shearing off of built-up layers occurred frequently.

ig. 17. TEM cross section of the polished Al2O3 coating after machining 316L for s. The surface was coated with gold which is seen as a black discontinuous coatingn the surface. To the right in the figure an adhered steel particle can be seen.

rt (cf. Fig. 15a). The particle, containing Fe and Cr, is oxidized at the surface facings used for extraction of maps were Cr L, N K, Fe L, Ti L, O K, Ca L, and V L.

This type of irregularities were not found on the chips consisting ofUHB 11, see Fig. 20. Another fundamental difference between thechips of the different work materials was that those consisting of316L had a wave pattern with ridges and grooves perpendicular tothe chip flow direction, see Fig. 21. This pattern is associated to thesegmentation of the chip and was not seen on the chips consistingof UHB 11.

4. Discussion

Together, the obtained results provide an extensive picture ofthe conditions on the different rake faces and show just how dif-ferent they can be.

Even though it is not obvious from Fig. 5, there is a trend ofcorresponding inserts showing higher amounts of adhered workmaterial, after turning in the austenitic stainless steel than in thecarbon steel. With higher amounts, both thicker and higher degree

1478 J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481

F n in th1 f a pari sugge

oenlrlptcrls

F(

ig. 18. (a) EFTEM maps of a part of the Al2O3 coating surface where Ca can be see0–100 nm thick, where no crystal information was discovered. (b) STEM-EELS-SI o

mages consist of Al and O, and the fine structure of the Al edge in the EELS spectra

f area coverage is understood. Although not shown here, this wasvident when studying micrographs of the inserts at higher mag-ification. For polished inserts, the amount of adherence varied

argely between tests. However, it was always less than on the cor-esponding standard rough inserts. The variation was likely due toocally different densities of surface irregularities remaining on theolished inserts, cf. Figs. 2b and 3b . As work material was proneo accumulate in these craters, see Fig. 9, variations in density of

raters would imply variations in the amount of transferred mate-ial. A more defect free surface of the inserts would thus render eveness adhered work material. EDS and XPS showed that there wereignificant differences between the adhered materials in different

ig. 19. SEM pictures (acc. voltage 20 kV, backscatter topographical mode) of the surfaceb) TiN coating, and standard rough (c) and polished (d) Al2O3 coating.

e surface. Above the thin dotted line in the TEM image, is a featureless layer aboutt of the surface where no Ca was found. It is clear that the high peaks seen in thesests that the whole area was Al2O3.

areas. In areas A1 and A3, the analyses complied fairly well withwhat would be adhered steel. However, in area A2, where the max-imum temperature during cutting is normally reached, there wereconsiderable enrichments of specific elements from the workpiecematerials.

For the austenitic stainless steel, these were mainly Al, Si, and Ca.XPS analyses suggested that these elements were chemically bondas oxides, halides, or silicates. These results are in accordance with

the literature, e.g. the work cited above, Fang and Zhang (1996), Qiand Mills (1996) and Katayama and Hashimura (1995), and stemsfrom the adherence of non-metallic inclusions from the workpiecesteel. Interestingly, the depth profiles obtained with XPS suggested

s of the 316L chips that have been sliding against standard rough (a) and polished

J. Gerth et al. / Journal of Materials Processing Technology 214 (2014) 1467–1481 1479

F rfaces(

tt

tafTVipfnti

Fp

ig. 20. SEM pictures (acc. voltage 20 kV, backscatter topographical mode) of the sub) TiN coating, and standard rough (c) and polished (d) Al2O3 coating.

hat the adhered material was thinner on the Al2O3 coated inserthan on the TiN coated insert, in all analyzed areas.

The EELS analysis of the polished TiN insert, cf. Fig. 16, revealedhat Cr could be found on the tool surface in areas not containingny Fe. That Cr precipitates from the chip and sticks to the tool sur-ace could be a result of a higher chemical affinity between Cr andiN than between Fe and TiN, for the prevailing contact conditions.

was also seen in the TEM analysis, actually in a large part of thenterface between the steel particle and the TiN surface. When com-aring the positions of different elements, and possible compounds

ormed, the thickness of the sample must be kept in mind. Its thick-ess of about 50 nm is actually of the same order of magnitude ashe size of the particle, so it is in this case difficult to say if overlapsndicate true compounds, or merely elemental regions positioned

ig. 21. The wavy appearance of the topography on the chip consisting of 316Lroduced by a polished TiN coated insert.

of the UHB 11 chips that have been sliding against standard rough (a) and polished

behind each other. It should also be noted that the specific analy-sis shown in Fig. 16 was conducted in a 100 nm wide region andthe results are indicative, but far from statistically verified. The factthat this type of preferential adherence of Cr and V was not foundon the TEM sample prepared from the Al2O3 coated insert does notnecessarily mean that it does not exists. In fact, the area from whichthe Al2O3 sample was extracted was chosen to study surface ridgesand not steel lump adherence.

The TEM analyses of the polished Al2O3 insert after turning in316L for 1 s revealed that the ridges found on the surface, predom-inantly in area A2, consisted of Al2O3 and not an adhered layerof workpiece material. Patches of calcium oxide could, however,sometimes be seen smeared out on the top surface of the patternformed. Dearnley (1985) noted similar type of ridges forming onAl2O3 coated inserts after turning in steel, and proposed discreteplastic deformation of surface asperities to be the cause. This maycontribute also to the pattern found in the present work, althoughthe surface asperities in this case were very much smaller. In theupper part of the Al2O3 coating, comprising the ridges, no crys-tallinity was detected in TEM, in contrast to the clearly crystallinematerial further down. Most likely the outermost part of the coat-ing had been deformed into this amorphous surface layer. It shouldbe noted, though, that the Al2O3 coating had been polished prior touse and it is not clarified whether the polishing process affects thecrystallinity of the sub surface region. The ridge pattern, however,was undoubtedly resulting from the cutting operation.

These results lead to the question of why less adhered materialwas found on the Al2O3 tool surface than on the TiN tool surface.Two scenarios can be put forward. The first one is the one commonly

referred to, i.e. simply a lower chemical affinity between the steelconstituents and the coating for Al2O3 than for TiN. The second sce-nario is more complex. Due to the lower thermal conductivity of theAl2O3 coating, heat is more efficiently retained at the Al2O3 surface

1 cessin

tsroAorfphclptr

ppcwuceprna

wttattwttab

wMcttidpprucTsmibgoltlhi

c

480 J. Gerth et al. / Journal of Materials Pro

han on the TiN surface, and higher temperatures at the Al2O3 toolhould be expected. This could lead to relocation of the lowest shearesistance in the contact from within the chip to a position close to,r even in, the surface of the Al2O3. The plastic deformation of thel2O3 surface, evidenced in TEM, might actually be a consequencef the latter. Even though non-metallic inclusions in the work mate-ial were likely extruded onto the Al2O3 surface, they would veryrequently be sheared off. The deformation in the Al2O3 surface,ossibly involving minute wear of the coating, would effectivelyinder large quantities of transferred material to be found afterutting. We suggest the first scenario is largely responsible for theower amount of steel in the A3 region, where pressure and tem-eratures are comparably low, and the second scenario contributeso a lower accumulation rate of non-metallic inclusions in the A2egion on Al2O3 than on TiN.

Interestingly, already after 5 s the two Al2O3 surfaces, rough andolished, have similar appearances in the A2 areas, suggesting thatolishing of Al2O3 prior to use had very limited influence on theonditions in that area. This means that, if this area control theear rate of the coating, polishing of the surface prior to use is annnecessary step. In contrast, the area A1 on the rough Al2O3 stillontained un-deformed surface asperities with abundant adher-nce of steel, whereas the polished insert did not. In this regionolishing clearly had an effect. For TiN coated inserts, polishingesulted in very different appearances, in all regions, after 5 s tur-ing. The rough TiN insert showed much more steel transfer, with

surface more or less entirely covered by adhered steel.Both TEM and XPS depth profiling showed that the adhered steel

as not completely oxidized, although the steel surfaces facinghe environment were covered by oxides. The XPS analyses fur-her indicated that both the Fe and Cr oxides were thicker in thereas A1 and A2 than in the area A3. This might seem contradictiveo the fact that A3 was more exposed to the oxygen from the airhan the other two areas during cutting. However, the temperatureas much higher in A1 and A2, which would give a higher oxida-

ion rate upon disengagement. In addition, vibrations, associatedo the chip segmentation process, could allow air to access the A1nd A2 regions during cutting and small amounts of oxygen maye present both in the steels and on the as-deposited coating.

Lower amount of steel transfer to the polished tools was seenhen cutting UHB 11 than when cutting austenitic stainless steel.oreover, areas of exposed cemented carbide were found on the

utting edges turning in 316L but not when cutting UHB 11. Anotherype of coating damage, located on the rake face close to the cut-ing edge, was also found on polished TiN which had been turningn 316L. The latter strongly suggests that coating segments wereetached and removed by the chip, resulting in the formation ofits that subsequently were filled with new work material. Thisrocedure, likely a result of excessive adhesive forces, occurredepeatedly during cutting and not only when the chip was removedpon disengagement. All this indicate higher adhesion between thehip and the tool when turning 316L than when turning UHB 11.his could be effects of the high amount of Ni and Cr in the austenitictainless steel, enhancing the chemical affinity between the workaterial and the tool. In fact, the TEM study of the polished TiN coat-

ng suggests enrichments of Cr on the tool surface. Related studiesased on intermittent sliding, Gerth et al. (2012), and on industrialear hobbing, Gerth et al. (2009) also reported on the enrichmentf specific alloying elements in the very contact zone, and it is veryikely this influences the subsequent tribological performance. Fur-hermore, a high ability to strain harden, in combination with aow thermal conductivity, of the austenitic stainless steel promote

igh temperatures in the contact zone, which, per se, can endorse

nteractions between the stainless steel and the tool material.The machined surface on the produced austenitic stainless steel

hips gave evidence of shear in and frequent shearing off of built-up

g Technology 214 (2014) 1467–1481

layers, whereas no such phenomena were apparent on the carbonsteel chips, cf. Figs. 19 and 20. It could thus be concluded that thetool rake face interacts fundamentally different with an austeniticstainless steel and a simple carbon steel. This is well known inpractice and consistent with what was found by, e.g. Trent andWright (2000). The fact that the stagnant layer close to the cut-ting edge was continuously detached and renewed when cutting316L may be a consequence of the chip forming process. Astakhov(2006) stated that the ductility of the austenitic stainless steel pro-motes the formation of saw-toothed chips where the size of the“teeth” becomes prominent. This chip forming process can inducesubstantial vibrations to the cutting process and hence influencethe state of stress acting on the cutting edge and cause the detach-ment described. The wavy pattern seen on the machined surfaceon the 316L chips, cf. Fig. 17 is an indication of this.

A final remark is that even though different types of coatings androughnesses showed large differences in appearances after turning,there were surprisingly small deviations in the measured cuttingand feed forces. The influence on feed force is larger than that oncutting force, which is expected as the former is most related to thefriction between the chip and the tool. Polished and standard roughtools show about the same feed forces after 5 s of cutting. Arguably,at this time the different tools have developed steady-state surfacesnot too different from each other, although the route to get therehad been different. Much work initially goes into smoothening ofthe standard rough surfaces of both TiN and Al2O3, and to someextent for polished TiN. These three also show trends of decreasingfeed forces, cf. Fig. 9, whereas the polished surface of Al2O3 can startdeveloping the steady state surfaces right from the very beginning,and shows the least varying feed force.

5. Conclusions

Short term turning tests in austenitic stainless steel, 316L, andcarbon steel, UHB 11, were made using CVD Al2O3 and CVD TiNcoated cemented carbide inserts. The inserts were prepared to twodifferent surface roughnesses, standard rough (as deposited) andpolished. The results can be summarized as:

• Roughness

The high roughness of standard rough inserts caused higheramount of work material to adhere than on corresponding polishedinserts. Even at polished tool surfaces work material stick to rem-nant surface irregularities. Furthermore, the surface roughness ofthe inserts strongly influenced the roughness of the rake face sideof the chips.

• TiN and Al2O3

Less work material adhered to the Al2O3 coated inserts thanon the TiN coated inserts. Accumulation of non-metallic inclusionsfrom the 316L was found in the high temperature area of both typesof coatings, although more pronounced on the TiN coating. Fiveseconds of turning in 316L resulted in the formation of pits on therake face, near the cutting edge, in the TiN coating while no suchdamages could be seen in the Al2O3 coating.

• Al2O3 deformation

The Al2O3 coatings showed signs of shallow deformation

in the outermost surface in the high temperature areas. Thisresulted in very similar appearances of the standard rough and thepolished Al2O3 coating in this particular area. This deformation,possiblyinvolving minute wear, is also suggested to counteract

cessin

ts

•

c1m1

•

pnt

A

Mon

R

A

B

C

D

J. Gerth et al. / Journal of Materials Pro

he adherence of steel and non-metallic compounds to the Al2O3urface.

Built up layers on chips

Remnants of built-up layers were found on the surfaces of thehips consisting of 316L, but not on the ones consisting of UHB1, was evidence of stronger adhesive forces between the coatingaterials and 316L than between the coating materials and UHB

1.

Layers on tools

Chromium and vanadium from the stainless steel showed higherreferential adherence to the polished TiN surface than iron. Also,on-metallic inclusions from both types of work piece steels con-ribute to the layers in the high temperature area on the inserts.

cknowledgements

The authors are grateful for financial support from the Ångströmaterials Academy. Parts of the experimental work were carried

ut at the Canadian Centre for Electron Microscopy (CCEM), aational facility supported by NSERC and McMaster University.

eferences

stakhov, V.P., 2006. Tribology of metal cutting. In: Briscoe, B.J. (Ed.), Tribology andInterface Engineering Series, vol. 52. Elsevier Ltd., UK.

ouzakis, K.-D., Michailidis, N., Skordaris, G., Bouzakis, E., Biermann, D., M’Saoubi, R.,2012. Cutting with coated tools: coating technologies, characterization methodsand performance optimization. CIRP Annals – Manufacturing Technology 61 (2),

703–723.

ahill, D., Lee, S.-M., Selinder, T., 1998. Thermal conductivity of �-Al2O3 and �-Al2O3

wear-resistant coatings. Journal of Applied Physics 83 (11), 5783–5786.earnley, P.A., 1985. Rake and flank wear mechanisms of coated cemented carbides.

Surface Engineering 1 (1), 43–58.

g Technology 214 (2014) 1467–1481 1481

Dobrzanski, L.A., Pakuła, D., 2005. Comparison of the structure and properties of thePVD and CVD coatings deposited on nitride tool ceramics. Journal of MaterialsProcessing Technology 164–165, 832–842.

Fang, X.D., Zhang, D., 1996. An investigation of adhering layer formation during toolwear progression in turning of free-cutting stainless steel. Wear 197, 169–178.

Gerth, J., Heinrichs, J., Nyberg, H., Larsson, M., Wiklund, U., 2012. Evaluation ofan intermittent sliding test for reproducing work material transfer in millingoperations. Tribology International 52, 153–160.

Gerth, J., Larsson, M., Wiklund, U., 2011. Survey of damage mechanisms on PVDcoated HSS hobs used in Swedish gear manufacturing industry. Tribologia 30(1–2), 37–50.

Gerth, J., Larsson, M., Wiklund, U., Riddar, F., Hogmark, S., 2009. On the wear ofPVD-coated HSS hobs in dry gear cutting. Wear 266, 444–452.

Holmberg, K., Matthews, A., 1994. Coatings tribology: properties, mechanisms,techniques and applications in surface engineering. In: Briscoe, B.J. (Ed.), Tribol-ogy and Interface Engineering Series, vol. 56, second ed. Elsevier, Amsterdam,Netherlands.

Katayama, S., Hashimura, M., 1995. Study on interfacial adhesion between cuttingtool and microstructures of free-machining steel. International Journal of theJapan Society for Precision Engineering 29 (1), 36–41.

Klocke, F., Krieg, T., 1999. Coated tools for metal cutting – features and applications.CIRP Annals – Manufacturing Technology 48 (2), 515–525.

Latha, G., Rajendran, N., Rajeswari, S., 1997. Influence on alloying elements on thecorrosion performance of alloy 33 and alloy 24 in seawater. Journal of MaterialsEngineering and Performance 6 (6), 743–748.

Marcus, P., Olefjord, I., 1988. A Round Robin on combined electrochemical andAES/ESCA characterization of the passive films on Fe–Cr and Fe–Cr–Mo alloys.Corrosion Science 28 (6), 589–602.

Moulder, J.F., Sobol, W.F., Bomben, K.D., 1995. Handbook of X-ray PhotoelectronSpectroscopy. Physical Electronics Inc., MN, USA.

Qi, H.S., Mills, B., 1996. On the formation of adhered layers on a cutting tool. Wear198, 192–196.

Ruppi, S., 2005. Deposition, microstructure and properties of texture-controlled CVD�-Al2O3 coatings. International Journal of Refractory Metals and Hard Materials23, 306.

Sokovic, M., Barisic, B., Sladic, S., 2009. Model of quality management of hard coatingson ceramic cutting tools. Journal of Materials Processing Technology 209 (8),4207–4216.

Trent, E., Wright, P., 2000. Metal Cutting, fourth ed. Butterworth–Heinemann, USA.Trinh, D.H., Back, K., Pozina, G., Blomqvist, H., Selinder, T., Collin, M., Reineck, I.,

Hultman, L., Högberg, H., 2009. Phase transformation in �- and �-Al2O3 coatingson cutting tool inserts. Surface & Coatings Technology 2013, 1682–1688.

Wiklund, U., Rubino, S., Kádas, K., Skorodumova, N.V., Eriksson, O., Hedberg, S., Collin,M., Olsson, A., Leifer, K., 2012. Experimental and theoretical studies on stainlesssteel transfer onto a TiN-coated cutting tool. Acta Materialia 59, 68–74.