nanocomposite layers of ceramic oxides and metals prepared by reactive

7/30/2019 Nanocomposite Layers of Ceramic Oxides and Metals Prepared by Reactive

1/7

Surface and Coatings Technology 179 (2004) 279285

0257-8972/04/$ - see front matter 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0257-8972(03)00865-X

Nanocomposite layers of ceramic oxides and metals prepared by reactivegas-flow sputtering

M. Birkholz*, U. Albers, T. Jung

Fraunhofer Institut fur Schicht-und Oberflachentechnik, Bienroder Weg 54 E, 38108 Braunschweig, Germany

Received 21 February 2003; accepted in revised form 14 July 2003

Abstract

Thin nanocomposite layers of ceramic oxides in conjunction with a second metallic phase were prepared by reactive gas-flowsputtering. The two nanogranular phases were dispersed into each other by co-deposition from two separate targets. TiO and2Al O were chosen as ceramic phase, respectively, which was obtained by sputtering one of the metals and reacting with oxygen.2 3In order to investigate the influence of a hard and a ductile metal, either tungsten or copper was chosen as the second metallicphase. Highly compact thin films could be prepared as was revealed by SEM. A low hydrogen content of 12 at.% was measuredby SIMS indicating a low sample porosity and an effective residual gas displacement from the surface of the growing film. X-Ray diffraction revealed the TiO -derived films to contain rutile-phase grains of average grain size of less than 10 nm. Mechanical2and tribological properties of the 3- to 4-mm thick coatings were determined by a micro-indenter operating with a Vickers-typediamond and by a pin-on-disk tester. The plastic microhardness HU amounted up to 24.1 GPa in case of TiO -derivedpl 2nanocomposite films and up to 14.8 GPa in case of Al O -based films. Also, with respect to tribological properties TiO :Cu and2 3 2TiO :W coatings were superior over Al O -derived ones. A wear coefficient of 0.35=10 mm yNm was obtained for optimisedy6 32 2 3TiO films against an alumina sphere under unlubricated sliding conditions. In lubricated pin-on-disk tests wear coefficients on2the order of 0.1=10 mm yNm were obtained for both metal-doped TiO and Al O films.y6 3 2 2 3

2003 Elsevier B.V. All rights reserved.

Keywords: Vickers hardness; Pin on disc; X-Ray diffraction; Reactive sputtering; Cermets and composites; Nanocomposites

1. Introduction

Hard coatings for wear and abrasive protection aremainly fabricated from nitride compounds and carbon-based compounds w1x. Their usage has enabled pro-longed lifetimes of a large variety of sliding mechanicalassemblies like engine working parts, rolling elements

such as bearings, cutting tools and others w2x. For manyapplications in air or under humidity it would bedesirable, however, to dispose of hard coatings thatbetter withstand oxidation or corrosion under chemicalattack. Oxides would accordingly represent an idealclass of materials for these purposes since they arestable at high temperatures and in chemically aggressiveenvironments. Various oxide systems are currently underinvestigations with many activities focussing on the Al

*Corresponding author. Tel.: q49-531-2155626; fax: q49-531-2155900.

E-mail address: [email protected] (M. Birkholz).

O system w3,4x and the ZrO system w5x. Two strategiesmight be applied to overcome the brittleness of ceramicoxides, which is the main obstacle for their use intribological applications: one approach makes use of thechange of mechanical properties with decreasing grainsizes down to the nanometer range w6x, while in asecond approach the dispersion of a second metallic

phase into the granular ceramic material is used. In thiswork, we report on experiments that make equally useof both approaches by preparing nanogranular disper-sions of metaloxides and metals by virtue of the gas-flow sputtering (GFS) technique w7,8x. This depositiontechnique appears well suited for the preparation ofnanogranular particles and coatings, since it operates ina pressure range where high collisional probabilitiesbetween the active species leads to the formation ofnanometer-sized particles already in the gasplasmaphase w9x. The GFS process may be equally applied tothe deposition of a large variety of metallic and metal


2/7

280 M. Birkholz et al. / Surface and Coatings Technology 179 (2004) 279285

Fig. 1. Scheme of deposition geometry for a co-deposition experiment.

Table 1Investigated ranges of deposition parameters during GFS co-deposition process for the preparation of nanocomposite oxide layers

Global parametersTotal pressure 0.20.9 mbarSubstrates CZ-(100) Si wafer and 100Cr6

Individual parameters Ceramic particle phase Metallic matrix phase

Material TiO , Al O2 2 3 Cu, WTarget powder densityyWycm2 3050 014Cathode voltagesyV 400800 01000Cathode currentsyA 3.56.5 01Argon flowysccm 5002000 50800Oxygen flowysccm 180 Source-substrate distanceycm 620 1015

oxide coatings and high deposition rates of some 10mmyh may be obtained for coatings of metal oxides.The technique is well suited for industrial applications,since it requires only a comparatively simple vacuumequipment and can be up-scaled to large areas w10x. Forthe study on nanocomposites presented here, TiO and2

Al O were chosen for the oxide phase, while Cu and2 3W were chosen for the metallic matrix phase. The focusof this work was on the region of high oxide concentra-tions combined with less amounts of the metallic matrixphase. It will be shown that metaloxidemetal coatingswith high mechanical hardness and advantageous tribo-logical properties can be prepared by the GFS technique.

2. Deposition process and preparation of nanoparti-

cles by GFS

The GFS process operates in a pressure range of 0.11 mbar, which is 12 orders of magnitude higher thanin the usual magnetron sputtering. This pressure regimeis associated with short mean free paths of gas atoms inthe 100-mm range and a viscous flow of gas, becauseof which the process has been named gas-flow sputteringw11x. The deposition technique makes use of the hollowcathode effect, which is associated with a high plasmadensity of 10 10 cm w12x. Thin films as presented12 13 y3

in this work were produced with cylindrical targetsthrough which a flow of argon gas was passed, see Fig.1. These cylinders ( 40=60 mm) were operated ashollow cathode, i.e. as sputter cathode by applying a

d.c. power in the range between 2 and 4 kW. The gasstream was conducted through the tubes to the substrate,which was positioned at a distance of 10 cm behind thetube openings. Most of the experiments were performedunder a total pressure of 0.4 mbar. Pure oxygen gas wasintroduced into the gas stream after it had left the hollow

cathode for the formation of metaloxide nanoparticles.The preparation of composite layers may be per-

formed via different routes with the GFS technique. Ourinvestigation concentrated on the co-deposition tech-nique, which operates by running two hollow cathodeswith different tube materials like Ti, Al, Cu and W, seeFig. 1. Cu-doped TiO coatings, for instance, were2prepared by directing one Ar stream through a Ti-tubeGFS source and a second Ar stream through a Cu sourceto the substrate. Thin films were prepared on both CZ-Si (100) wafers and 100Cr6 steel shims. Before depo-sition the substrates were cleaned by a plasma-assisted

process by biasing the substrate (y400 V) and passingan argon flow of 80 sccm through the chamber yieldinga total pressure of approximately 0.05 mbar. The sub-strate was first coated with pure metal Ti or Al,respectively, to improve the adhesion of the functionallayer. In a second step the oxygen flow was opened andthe second hollow cathode was switched on for thepreparation of the nanocomposite layer. In order tofacilitate hard and compact coatings, the depositionprocess was supported by biasing the substrate (y100V). No intentional heating of the substrate was applied,but temperatures up to 200 8C were measured at theback of the Si substrates during the process by virtue ofan electrically isolated thermocouple. The ranges ofdeposition parameters investigated with the co-deposi-tion technique are listed in Table 1. For most of theTiO - and Al O -derived nanocomposite coatings report-2 2 3ed in the following, the Ti target or Al target waspowered with 2.5 kW (equivalent to 33 Wycm ), which2

was associated with voltages in the range between 650and 700 V.

Transmission electron microscopy (TEM) was appliedto study the ability of GFS for the preparation ofnanogranular particles. For this purpose, very short


3/7

281M. Birkholz et al. / Surface and Coatings Technology 179 (2004) 279285

Fig. 2. (a) Transmission electron micrograph of tungsten particles produced by a 1 s gas-flow sputtering experiment, substrate: graphite-coated

TEM grid. The bar in the lower left corner measures 200 nm. (b) Particle size distribution function of tungsten particles that was obtained bygauging the shape of 480 grains. The frequency of calculated diameters was well accounted for a by Gaussian function with peak centroid at 19nm and a width of 11 nm.

deposition experiments of a duration of 1 s wereperformed in order to deposit the nanoparticles formedin the gasplasma phase as isolated islands on thesubstrate. Such capture experiments are frequentlyapplied to prepare a set of countable and measurablenanoparticles w13x, from the measurement of which theparticle size distribution function (PSDF) can bederived. Graphite-coated TEM grids (3.05 mm, 200mesh) were positioned 12 cm in front of the targets

aperture and 1200 sccm Ar were conducted through aW target that was powered with 3 kW. The sputteredmetal atoms were swept out of the cathode with the Argas stream forming clusters during their transport to theTEM grids due to a high probability for direct collisionunder an exerted total pressure of 0.53 mbar. Fig. 2adisplays a TEM micrograph from one of the gridsshowing a set of few dozen particles. Their chemicalnature was investigated by electron spectroscopic imag-ing (ESI), unequivocally identifying the captured clus-ters to consist of tungsten. It is therefore concluded thatW atoms had indeed formed nanoclusters during the

transport from the sputter cathode to the substrate. Theparticle distribution function was derived by sizing theprojected surface of 480 particles and calculating anaverage diameter for each. The obtained PSDF is shownin Fig. 2b. It was often observed that PSDF of nano-meter-sized grain ensembles are well accounted for bya log-normal distribution function w14x. However, in thecase presented in Fig. 2b a Gaussian function yieldedsmaller estimated S.D. and an increased reliability valuewhen both model functions were fitted to the data. Theaverage particle diameter is accordingly 18.7"0.2 nm,while the width of the PSDF amounts to 11.0"0.5 nm.This experiment clearly demonstrates the capability of

the GFS technique to prepare nanogranular particles.Since all experiments were performed under comparabledeposition conditions, it can be assumed that particlegrowth occurred in the gas plasma phase yielding ananogranular composition in the films prepared. Themain deposition parameters by which the PSDF mightbe tailored are the total pressure, the electrical powerconducted to the target and the distance between targetand substrate.

3. Composition, morphology and mechanical

properties

The stoichiometry of prepared samples was evaluatedwith an electron microscope analysis (EPMA) systemby exciting with 12 keV electrons, and determining theintegral intensity of the metals and oxygen Ka emissionline, against defined standards. By properly adjustingthe deposition parameters metal oxide films close to theideal stoichiometry were obtained, yielding for instancea OyTi ratio of 2.02"0.02 in case of pure titanium

dioxide films. The concentration of the metals matrixphase was varied in the range from 0 to 5 at.%. Fig. 3displays a SEM cross-section micrograph of a slightlycopper-doped TiO film on Si, portraying a highly2compact morphology without a noticeable columnargrowth structure. Pure oxide films without matrix inclu-sions were optically transparent. SIMS measurements ofCu-doped and pure TiO samples revealed a low hydro-2gen content in the 12 at.% range. This result isinterpreted to indicate an effective residual gas displace-ment from the surface of the growing film, sinceotherwise, higher amounts of hydroxyl groups wouldhave been built-in to the growing film. The expulsion


4/7


Fig. 3. SEM cross-sectional micrograph of copper-doped TiO film2

prepared by reactive gas-flow sputtering. The film is recognized to behighly compact without columnar growth structures.

Table 2

Mechanical and tribological properties of nanocomposite coatings prepared by reactive gas-flow sputtering (hardness and associated values:maximum values, pin-on-disk results: optimum range)

TiO -based2 Al O -based2 3

HardnessyGPa 24.1 14.8Equivalent to Vickers hardnessyHV 2060 1300Elastic modulus EyGPa 268 179Elastic share of indentation work W yW y%el tot 59 53

Tribological pin-on-disk test vs. 4 mm Al O sphere (sliding velocity 4 cmys, load 2 N)2 3Unlubricated Coefficient of friction 0.40.5 0.60.8

Wear coefficient Ky10 mm yNmy6 3 0.30.4 810Lubricated (0W30) Coefficient of friction 0.130.14 0.110.13

Wear coefficient Ky10 mm yNmy6 3 -0.1 0.050.3

of residual gas from the active area is due to the argonstream onto the substrate and might be considered as akey element for the preparation of low-contaminatedfilms in a rough vacuum environment. Moreover, thesmall hydrogen amount measured by SIMS, signals alow-porosity such that ex-situ H O contaminations from2the ambient were successfully expelled and not incor-porated into the films.

Hardness H and elastic modulus E were measured

with a Fischerscope microindenter FH 100 according tothe standard technical rules w15x. For this purpose aVickers-shaped diamond pyramid was pressed into thecoating by increasing the loading to a maximum forcebetween 30 and 50 mN. The maximum indentationdepth during the measurement was: (i) larger than 200nm; (ii) smaller than or approximately 1y10 of the filmthickness; (iii) exceeded the 20-fold R roughness ofathe coating w15x. The hardness was evaluated from thetangent to the unloading curve (plastic universal hard-ness HU ), so that Vickers hardness values HV couldpl

be derived from an established relationship w16x and thevalues obtained can directly be compared with otherhard coatings, see for instance, for the nitride systemsw17,18x. The hardness of the Al O -derived composite2 3films was found in the 1415 GPa range, which com-pares well with pure Al O thin films prepared by other2 3

deposition techniques under the same temperature con-ditions w3x. The hardness values in the range of up to24.1 GPa were obtained in the case of TiO -based2coatings, which corresponds to a Vickers hardness of2060 HV. This value was equally obtained for TiO2coatings prepared on steel shims and on silicon wafers.Table 2 lists the mechanical and some other propertiesof the prepared films. It can be seen that the microhard-ness is approximately 1y10 of the elastic modulus,which is typically observed for hard coatings and canbe taken as justification for the reliability of obtainedvalues. To our knowledge, such hard TiO coatings have2

not been prepared before. Up to now, our investigationsdid not reveal a hardness enhancement caused by theinclusion of a metallic matrix phase as has generallybeen established to occur in the case of materials fromthe nitride system w17,18x. The highest hardness valueswere exhibited by pure oxides, from which especiallythe TiO coatings show remarkable high hardness. The2tribological properties of one set of the oxide layers(group of Al O -derived coatings), however, sensitively2 3depended on the inclusion of a second metal matrixphase.

Tribological investigations of nanocomposite films

were performed under dry-sliding and under lubricatedconditions by use of a pin-on-disk tester. For thispurpose, a stationary alumina (99.7%) sphere of 4.0-mm diameter (Saphirwerk, CH, grade 10) was pressedagainst the plane surface of the rotating specimen (30rpm) by a normal force of F s2 N and a sliding speedNof 4 cmys. The sliding distance was Ds200 m at aradius of 26 mm, resulting in more than 2500 cycles.Tests were performed under ambient conditions, i.e.under varying relative humidity that ranged from 44 to62% for the different tests. The wear coefficient Ks


5/7


Fig. 4. Coefficient of friction vs. wear distance for lubricated (lowercurve) and unlubricated (upper curve) pin-on-disk tests of TiO -based2nanocomposite layers.

Fig. 5. X-Ray diffraction pattern of TiO yTiysteel measured in grazing2incidence geometry, as0.38, with CuKa radiation. The sticks indicatepositions and integral intensities of rutile and hcp-Ti diffraction pat-terns according to ICDD.

V yDyF was derived from the worn volume V asw N wdetermined from step profiling the wear track. Practical-ly, for all tests under dry sliding conditions the samecourse of the coefficient of friction (COF) as a functionof testing time was observed. The COF started with avalue around or less than 0.2, but increased after some100 cycles. The set of all samples decomposed into twogroups, for the first of which the COF stabilized atconstant values, while for the second set the coatingbecame totally grinded after less than the total testingtime. We interpret this phenomenon in terms of anabrasive wear mechanism with the wastage from the

wear track acting as abrasive agent.The coatings could be considered to decompose into

a first group that withstand the wear from its ownabraded material, while the second group does not. PureAl O coatings turned out to belong to the second group.2 3In their case, the integration of Cu or W nanodispersionsled equally to an enhanced stability against abrasivewear. Also, the extreme sensitivity of pure Al O coat-2 3ings against moisture was reduced by the inclusion ofone of the metal matrix components. The coefficient offriction for Al O :Cu and Al O :W coatings were found2 3 2 3in the 0.60.8 range, while the wear coefficients for the

measuring conditions given above amounted to 810=10 mm yNm. In case of TiO -derived nanocom-y6 3 2posite coatings, improved tribological properties couldbe demonstrated as was expected because of theirenhanced mechanical hardness. For coatings made frompure titanium dioxide the initial coefficient of friction(COF) amounted to 0.2, but raised to 0.5 after 50 mand stayed constant for the remaining sliding distance,see Fig. 4. The wear coefficient was determined to Ks0.35=10 mm yNm, which is at the low edge ofy6 3

measured values for unlubricated sliding ceramicceramic couples w19x. In case of the lubricated measure-ment with a standard motor oil (0W30), the wear

coefficient was hardly measurable and only an upperlimit of 0.1=10 mm yNm can be specified.y6 3

The coefficient of friction of titanium oxides has beenreported to decrease with increasing sliding velocity andincreasing temperature w19x. Compared to the citedliterature, the thin TiO films presented in this work2

were examined with rather low sliding velocity of 4cmys and this may offer the potential for further tribo-logical applications, where sliding velocities in the mysrange are often applied.

4. Structural characterisation

In order to gain more insight into the nanogranularmorphology associated with gas-flow sputtered thinfilms and into the microscopic structure of such hardand wear resistant coatings the samples were examinedby X-ray diffraction. The focus for these investigations

was on TiO -based coatings because of their superior2mechanical and tribological properties. For this purpose,a Seifert XRD 3000 PTS diffractometer equipped witha secondary graphite monochromator and operated witha copper anode wl(Ka )s154.0598 pmx at 40 kV and130 mA, was utilised in asymmetric grazing incidence(GI) geometry by setting the angle of incidence to as0.38. Diffractograms were measured in the step scanmode and a step size of 0.058. Fig. 5 displays thediffraction patterns as obtained for a 4.2 mm thin filmdeposited on a 100Cr6 steel shim. Inserted histogramsdisplay Bragg reflection positions and relative intensitiesfor TiO rutile and hexagonally-closed packed Ti powder2samples according to ICDD (International Centre forDiffraction Data) card numbers 21-1276 and 5-682. Themeasured reflections are seen to be clearly accountedfor by the rutile phase. The small deviations in scattering


6/7


angle position may be understood from residual stresseswithin the film, but will not be evaluated here explicitly,because of the small precision in position measurementand the variation of diffraction vector orientation in thegrazing incidence geometry. The metallic matrix phasewas not detected, most probably because of its low

concentration and its amorphous structure.It can be seen from the figure that Bragg reflections

caused from the TiO -layer exhibited a large broadening.2In order to derive the grain size from this effect, peakprofiles were fitted by pseudo Voigt model functionsand the profile parameters like peak centroid 2u , full0width at half maximum (FWHM) 2v and integralintensity I were determined. Grain sizes were evaluatedby simply applying Scherrers equation, NL Ms0.94 lyV2v cosu w20x, from which a volume-averaged column0height NL M is obtained, which is a rough measure forVthe average grain size. Contributions of microstrain to

the peak broadening were neglected, because the com-bined determination of grain size and microstrainsrequires a set of free standing, non-overlapping peaks tobe evaluated, for instance, in terms of the WilliamsonHall scheme w21x. Since microstrains are mainly causedby dislocations, this approximation may be justified inthe case of grains that are too small for dislocation tooccur. It should be remembered that dislocation-inducedstrain fields usually extend over tens of nanometers w1x,and that the dislocation core generally measures approx-imately 1 nm. The determination of NL M values for theVdiffractogram given in Fig. 5 were then found to liebetween 2.4 and 7.8 nm. It is well known that the size

averaging according to Scherrers equation yields ratherlarge values when compared with the median or meanvalue of diameters of grain populations w22x. The medianof the diameter distribution function will accordingly besmaller than the values given above. It, therefore,appears appropriate to consider NL M values as upperVlimits for the grain size and it may safely concludedthat the size of rutile grains in TiO -based coatings is2less than 10 nm. This might also be taken as justificationfor neglecting the effect of dislocation-induced micro-strain on peak broadening.

5. Discussion

In a recent work, Bendavid et al. have demonstratedthe preparation of titanium dioxide films in the rutilemodification by filtered arc deposition for which micro-hardness values up to 18 GPa could be obtained w23x.Such high hardness values could also be demonstratedby Bally et al. w24x and Lapostolle et al. w25x for TiO .2We note, that these results for TiO films clearly sur-2mounted the bulk hardness of rutile, which is given as1100 HV in the literature w1x, but still differ from the24 GPa obtained for the GFS-prepared films in thiswork. Bally et al. and Lapostolle et al. even succeeded

in the preparation of hard TiO coatings in the 20 23xGPa range by reactive magnetron sputtering, but thesefilms were suboxidic and deviated strongly from theideal TiO stoichiometry. Bally et al. reported that2substoichiometric films were golden-metallic to darkblue and intransparent for optical wavelength. In a

temper experiment with an as-deposited TiO film,1qxthe O content was varied to arrive at almost stoichio-metric TiO , which was accompanied by a hardness2drop from 23 to 21 GPa w24x.

Regarding a possible explanation for the high hard-ness obtained we consider two different line of argu-ments by which the phenomenon may be understood.Firstly, the nanogranular morphology of the coatingsmay impede the occurrence and the motion of disloca-tion within the coatings w26x. The high hardness of thelayers could then be understood in terms of a reduceddislocation mobility leaving any plastic deformability

only to the comparatively more elaborate grain boundarydisplacement. It should be considered in this context,however, that the familiar Hall Petch rule has to bemodified when grain sizes in the nanometer range areconsidered w27x. The advantageous mechanical proper-ties would accordingly be caused by the ability of theGFS technique to produce nanogranular particles. Sec-ondly, the exceptional hardness of TiO -derived coatings2prepared by GFS may be understood from the ionbombardment during growth. Ion currents at the sub-strate holder were specified during the deposition pro-cess by the bias generator that was run in the constantvoltage mode. According to this measurement the ion

current density amounted to 4 6 mAycm during the2

deposition of the oxide layer. These current densitiesindicate a strong ion bombardment when compared toother deposition experiments as performed by magnetronsputtering, for instance. It has to be remembered, how-ever, that this large ion flux to the substrate is associatedwith ions of much lower energy when compared withother sputtering processes. This is due to the highpressure and the small average free paths of atoms andions in the plasma of the gas-flow sputtering process,leading to a fast thermalisation of atoms and ions w28x.The ion bombardment would then be associated with a

large number of ions, but of low energy. We concludethat both possible explanations of the high hardnessthe nanometer-size hypothesis and the ion bombardmenthypothesesrely on the special conditions the growingfilm encounters during the GFS process.

It is of interest to compare the results of this workfor TiO with those obtained for other hard oxide2coatings. High hardness values of up to 24 GPa have,for instance, been achieved in the case of Al O prepared2 3by pulsed magnetron sputtering w3x or pulsed d.c.PACVD processes w4x. But to arrive at hardness valuesof approximately 2024 GPa the heating of the substrateto temperatures in excess of 600 8C and even more have


7/7


to be applied. The effect is understood from the severalphases existing in the Al O material system, from whichonly the high-temperature a-Al O phase is suited for2 3hard coatings, while the hardness associated with theamorphous and the g-phase only amounts to values of10 GPa or less than 20 GPa, respectively. For many

mechanical working parts the thermal load associatedwith such high temperatures is undesired and hardTiO coatings prepared at less than 200 8C would offer2a useful technical alternative. Other applications for thetitanium dioxide films presented here will be derivedfrom their useful optical and electrically isolating prop-erties. Most importantly, in contrast to the classical hardcoatings from the nitride material system like TiN, AlN,etc., a hard TiO coating is transparent for visible light.2It, therefore, offers one of the very few technicalsolutions for a transparent coating for protection againstscratching and abrasing loads, which is of technical

importance for a large variety of applications.

6. Conclusion

In conclusion, it has been shown that nanocompositecoatings based on TiO and Al O as ceramic phase in2 2 3combination with either Cu or W as metal matrix phasecan be prepared by the GFS technique. The gas flowsputtering technique was demonstrated by capture exper-iments in combination with TEM of being capable toproduce nanometer-sized particles by nucleation andgrowth of sputtered atoms during the transport from thetarget to the substrate. Prepared films were highly

compact and contained few in- and ex-situ infiltratedcontaminants. X-Ray diffraction revealed the titaniumdioxide to exist mainly in the rutile phase with graindiameters of less than 10 nm. TiO -derived films turned2out to be superior when compared to Al O -based ones2 3and exhibited a high mechanical hardness up to 24.1GPa and small wear coefficients under unlubricatedsliding of 0.35=10 mm yNm. These advantageousy6 3

properties were achieved without intentionally heatingthe substrate under a low thermal budget regime.

Acknowledgments

This work was supported by the Bundesministeriumfur Bildung und Forschung ( 03N3086B), Robert

Bosch GmbH and Volkswagen A.G. We gratefullyacknowledge the close cooperation with our industrialpartners. We also thank Dr H. Bremers, T.U. Braun-

schweig, for help in XRD work, Dr U. Gernert, TUBerlin, for SEM micrograph, Dr S. Maas, M.A.S. Frei-burg, for taking the TEM pictures, C. Steinberg forEPMA and P. Willich for SIMS investigation.

References

w1x M. Ohring, The Materials Science of Thin Films, AcademicPress, San Diego, 1992, p. 12, Chapter 12 and Table 1.

w2x R.F. Bunshah, Handbook of Hard Coatings, William AndrewPublishing, Norwich, 2001.

w3x O. Zywitzki, G. Hoetzsch, Surf. Coat. Technol. 9495 (1997)303.

w4x C. Taschner, B. Ljungberg, V. Alfredsson, I. Endler, A. Leon-

hardt, Surf. Coat. Technol. 108109 (1998) 257.w5x A. Rosidian, Y. Liu, R.O. Claus, Adv. Mater. 10 (1998) 1087.w6x J. Karch, R. Birringer, H. Gleiter, Nature 330 (1987) 586.w7x T. Jung, A. Westphal, Surf. Coat. Technol. 59 (1993) 171.w8x T. Jung, T. Kalber, V. von der Heide, Surf. Coat. Technol. 86-

87 (1996) 218.

w9x K. Ishii, T. Ohba, T. Hara, Mater. Sci. Eng. A217-218 (1996)232.

w10x M. Hofer, A. Jung, T. Jung, H.-U. Kricheldorf, F. Schmidt,

43rd Annual Technical Conference of the Society of VacuumCoaters, Society of Vacuum Coaters, Denver, 2000.

w11x K. Ishii, J. Vac. Sci. Technol. A7 (1989) 256.w12x T. Kalber, HohlkathodenGasflusputtern zur Ver-

schleischutzbeschichtung von Kunststoffen, Doctoral Thesis,TU Braunschweig, 1998.

w13x D.L. Peng, T.J. Konno, K. Wakoh, T. Hihara, K. Sumiyama,Appl. Phys. Lett. 78 (2001) 1535.

w14x W.C. Hinds, Aerosol Technology, Wiley, New York, 1999.w15x Universalharteprufung-Teil 1: Prufverfahren, Deutsches Institut

fur Normung, Berlin, 1997.w16x H.-H. Behnke, Hart. Techn. Mitt. 48 (1993) 3.

w17x S. Veprek, Thin Solid Films 317 (1998) 449.w18x J. Musil, J. Vlcek, Surf. Coat. Technol. 142-144 (2001) 557.w19x M. Woydt, Tribol. Lett. 8 (2000) 117.w20x B.E. Warren, X-Ray Diffraction, Addison-Wesley, Reading,

1969.w21x J.I. Langford, D. Louer, E.J. Sonneveld, J.W. Visser, Powder

Diffr. 1 (1986) 211.w22x C.E. Krill, R. Birringer, Philos. Mag. A 77 (1998) 621.w23x A. Bendavid, P.J. Martin, H. Takikawa, Thin Solid Films 360

(2000) 241.w24x A.R. Bally, P. Hones, R. Sanjines, P.E. Schmid, F. Levy, Surf.

Coat. Technol. 108-109 (1998) 166.w25x F. Lapostolle, A. Billard, J. von Stebut, Surf. Coat. Technol.

135 (2000) 1.

w26x R. Hauert, J. Patscheider, Adv. Eng. Mater. 2 (2000) 247.w27x D.A. Konstantinidis, E.C. Aifantis, Nanostruct. Mater. 10

(1999) 1111.w28x K. Ishii, H. Hamakake, 43rd Annual Technical Conference of

the Society of Vacuum Coaters, Society of Vacuum Coaters,Denver, 2000.

nanocomposite layers of ceramic oxides and metals prepared by reactive

Documents