1-s2.0-S0022311515301239-main

lable at ScienceDirect

Journal of Nuclear Materials 466 (2015) 69e79

Contents lists avai

Journal of Nuclear Materials

journal homepage: www.elsevier .com/locate/ jnucmat

Mechanical and tribological properties of crystalline aluminum nitridecoatings deposited on stainless steel by magnetron sputtering

R.K. Choudhary a, *, S.C. Mishra a, P. Mishra a, P.K. Limaye b, K. Singh c

a Materials Processing Division, Bhabha Atomic Research Centre, Mumbai 400085, Indiab Refuelling Technology Division, Bhabha Atomic Research Centre, Mumbai 400085, Indiac Fusion Reactor Materials Section, Bhabha Atomic Research Centre, Mumbai 400085, India

a r t i c l e i n f o

Article history:Received 10 April 2015Received in revised form22 May 2015Accepted 24 July 2015Available online 26 July 2015

Keywords:Aluminum nitrideWurtziteMagnetron sputteringAdhesionHardnessWear

* Corresponding author.E-mail addresses: [email protected]

(R.K. Choudhary).

http://dx.doi.org/10.1016/j.jnucmat.2015.07.0360022-3115/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

Aluminum nitride (AlN) coating is a potential candidate for addressing the problems of MHD pressuredrop, tritium permeation and liquid metal corrosion of the test blanket module of fusion reactor. In thiswork, AlN coatings were grown on stainless steel by magnetron sputtering. Grazing incidence X-raydiffraction measurement revealed that formation of mixed phase (wurtzite and rock salt) AlN wasfavored at low discharge power and substrate negative biasing. However, at sufficiently high dischargepower and substrate bias, (100) oriented wurtzite AlN was obtained. Secondary ion mass spectroscopyshowed presence of oxygen in the coatings. The highest value of hardness and Young's modulus were14.1 GPa and 215 GPa, respectively. Scratch test showed adhesive failure at a load of about 20 N. Wear testshowed improved wear resistance of the coatings obtained at higher substrate bias.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Reduced activation ferritic/martensitic (RAFM) steel and vana-dium alloys have been accepted as the candidate structural materialfor various types of solid and liquid based tritium breeding blanketdesign concepts, proposed by the participating countries in Inter-national Thermonuclear Experimental Reactor (ITER) programme.Accordingly, it has seen the development of RAFM steels such asEurofer97, F82H, Optifer IVa, Manet I, CLAFM and, V-4Cr-4Ti alloy[1e3]. Similarly, the Test Blanket Modules (TBMs) based on liquidmetal breeder blanket concept have been proposed by many of theparticipating countries. This includes LeadeLithium CooledCeramic Breeder (LLCB) TBM proposed by India [4], Dual CoolantLead Lithium (DCLL) TBM by the USA [5], Dual Functional Lith-iumeLead (DFLL) TBM by China [6] and liquid lithium self-cooledbreeder blanket design by the Russian Federation [7]. These blan-ket designs require flow of liquid metal through the TBM channelswith finite velocity. However, when the conductive liquid metalflows through an electrically conductive duct surrounded by a

m, [email protected]

strong magnetic field, electric current of high magnitude is inducedin the liquid metal depending upon its velocity. In ITER, ToroidalMagnetic field of around 5.3 T at major radius and Poloidal mag-netic field of around 4 T are present. Interaction of this inducedcurrent with the external magnetic field present can give rise to thelarge Magneto-hydro-dynamic (MHD) pressure drop due to theresulting Lorentz force [8,9]. The MHD effect not only increases thepumping power requirement for liquid metal circulation but alsoaffects its flow distribution pattern. This in turn can impact the heattransfer, tritium transport and corrosion related issues. In the caseof Li/V blanket design proposed by the Russian Federation [7], thepressure drop due to the MHD effect can reach to prohibitivelyunmanageable values if the flow channels are not electricallyinsulated. Moreover, besides theMHD effect, there are several othermaterial related critical issues like permeation of tritium throughthe structural material and also the liquid metal induced corrosionof the structural material [10].

Problems of MHD pressure drop, tritium permeation and liquidmetal corrosion can be addressed by electrically insulating thestructural material of TBM in contact with the flowing liquid metal.This can be achieved either by applying suitable ceramic coatingsor, by way of introducing silicon carbide fiber reinforced siliconcarbidematrix composite (SiCf/SiC composite) flow channel inserts.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e7970

If the above mentioned problems are addressed by the use ofcoatings, the coating must have sufficiently high electrical re-sistivity, adequate tritium permeation resistance, chemical stabil-ity, mechanical integrity and high resistance against radiationdamage. Here, it can be pointed out that most of the concertedeffort for the development of electrical insulation coatings has beendirected towards the self cooled Li/V TBM concept. Electricalinsulation coatings for addressing the MHD effect in this case is anintegral part of the design. Large number of ceramic coatings suchas alumina (Al2O3), yttria (Y2O3), calcium oxide (CaO), magnesiumoxide (MgO), beryllium oxide (BeO), erbium oxide (Er2O3), boronnitride (BN) and silicon nitride (Si3N4) were initially considered butAlN, Er2O3, Y2O3 and multilayer coatings (ceramic coatings with aprotective metallic alloy layer) are currently being focussed upon[11]. Development of electrical insulation coatings on RAFM steelhas been mostly limited to Al2O3, originally oriented as a tritiumpermeation barrier. However, India and China have envisaged itsuse as MHD coating also.

In our previously reported work [12], adherent alumi-naealuminide coatings were formed on ferritic-martensitic T91 steelby a two step oxidation of aluminium layer, deposited electrochem-ically using AlCl3-1-ethyl-3-methyl imidazolium chloride ionic liquid.Similarly, Rayjada et al. [13] optimized the process for depositingEr2O3 coatings on P91 alloy by direct current reactive magnetronsputtering, for DEMO relevant fusion blanket module applications.Zhang et al. [14] from China reported the formation of g-Al2O3/(Fe, Cr,Ni)x Al double layered structure on 321 stainless steel and demon-strated their potential application as tritium permeation barriercoating. They electrodeposited 20 mm thick aluminium coating on321 stainless steel tubes using ionic liquid bath and subjected them toheat treatment for aluminide formation followed it by selectiveoxidation for getting g-Al2O3 top layer. Recently, Zhan et al. [15] re-ported the formation of a-alumina coatings on CLAM steel byoxidation of the pack aluminized layers. In addition to its high tritiumpermeation reduction factor (TPRF) and self healing properties, FeAl/Al2O3 composite coating on RAFM steel is also very effective in pre-venting corrosion of blanket structural material in flowing PbeLi athigh temperature [16].

AlN coatings have shown excellent compatibility results in thecase of Li/V system [17]. Although, the individual solubility ofAluminum and Nitrogen is high in liquid Lithium, the value of theirsolubility product for Lithium/AlN system is very small, suggestinglittle dissolution of AlN in liquid lithium [18]. Now, lithium in Pb-17Li will be even far less reactive due to its large negative devia-tion from ideal behavior. This will further enhance the stability ofAlN in contact with Pb-17Li eutectic. Hubberstey [19] calculated thevalues of thermodynamic functions for Pb-17Li eutectic at 773 Kand obtained gLi ¼ 7.24 � 10�4, aLi ¼ 1.23 � 10�4 andGLi ¼ �57:8kj=mol. This highly negative partial Gibbs free energy ofLi in this alloy makes it much less reactive. From these values theycalculated [19] that the co-existence of LiH, Li3N and even Li2C2with Pb-17Li eutectic is not possible and only oxygen can react withPb-17Li to form Li2O. Therefore, although carbon, hydrogen, nitro-gen and oxygen have high solubility in Pb-17Li, most of the carbideand nitride coatings are compatiblewith Pb-17Li except b-SiC, Si3N4and CrN and some oxides (viz. Fe2O3, NiO, Cr2O3, B2O3 and LiCrO2)[19]. Similarly, Valls et al. [20] reported about þ300 kj/mol freeenergy changes required at 773 K for the reaction of AlN with Pb-15.7Li eutectic. This energy barrier is sufficiently high to preventany reaction between AlN and Pb-15.7Li eutectic. But, impuritylevel in Pb-17Li eutectic particularly with respect to oxygen have tobe controlled in order to avoid the possibility of formation of binaryor ternary oxides such as LiAlO8 and LiAlO2.

Besides having same order of electrical resistivity, AlN possessesmuch higher thermal conductivity in comparison to Al2O3. Despite

having such favorable properties and thermodynamic stability, ithas not received themuch needed attention for application asMHDcoating on RAFM steel. In other words, the role of AlN as a candidatecoating material for Pb-Li/RAFMS system has still not beenconfirmed practically. There is hardly any detailed study reportedon the growth of crystalline AlN coatings on steel substrate despitethe fact that crystalline AlN coatings are more stable in liquidmetals [21e23]. However, numerous works have been reported onthe growth of crystalline AlN coatings on silicon substrate forelectronic, optoelectronic and surface acoustic wave device appli-cations [24e26]. Moreover, there is very little information availableon the mechanical and tribological properties of AlN coatingsgrown on steel. Adequate mechanical and tribological propertiesare necessary requirements since the coating has to withstand thestresses generated by the high pressure (~1.2 MPa) PbeLi liquidcoolant. To the best of our knowledge only Cabrera et al. [27] re-ported some results on the mechanical and tribological propertiesof AlN coatings grown on AISI D3 steel and silicon substrates but,information about the crystalline structure of the AlN coatingsgrown on steel is again missing.

Therefore, the present work discusses in detail the deposition ofcrystalline AlN coatings on stainless steel substrate and also itsmechanical and tribological properties. Austenitic stainless steel304L has been chosen as a substrate because of its ready availabilityand the ease of sample preparation. Its selection in place of RAFMsteel is not expected to make any significant effect on the results ofthe current studies with respect to the performance of AlN coatingsdue to the presence of Al interlayer that bonds well with bothsubstrates.

2. Experimental details

2.1. Coating deposition

AlN coatings were deposited in a custom built balanced planarmagnetron sputtering system in sputter down configuration. Themagnetron was powered by a 5 kW high frequency asymmetricbipolar pulsed DC power supply (ENI RPG-50). High purity(99.999%) aluminum disc (Ø75 mm � 10 mm) was used as asputtering target. Annealed AISI stainless steel (SS) 304L specimens(Dimension: 15 mm � 15 mm � 2 mm) were used as the substrate.Prior to deposition, the specimens were metallographically pol-ished using SiC abrasive papers of 240, 320, 400, 600, 800 and 1200standard ANSI grit sizes followed by cloth polishingwith submicronsize polycrystalline diamond paste. The polished substrates werecleaned in a proprietary alkaline solution at 75 �C for 10 min fol-lowed by acid dip in 15% HCl for 2 min at room temperature.Substrates were rinsed with distilled water before and after theacid dip step. They were subsequently cleaned in an ultrasonic bathusing analytical grade isopropyl alcohol. The cleaned substrateswere dried, weighed and mounted on the sample holder situatedinside the deposition chamber. The weight of the substrate wastaken using an electronic balance having an accuracy of better than2 mg. Substrates were finally cleaned in situ in argon plasma for30 min at a negative bias voltage of 800 V and a pressure of 10 Pa.Although the bias voltage applied to the substrate was �800 V, theion current density during ion cleaning was in the range of200e300 mA/cm2. This value of current was not sufficient to causeany detectable change in the weight of the small sized substrateused during experiments, and was insignificant in comparison tothe weight gain due to the coating process. This was confirmed byweighing the substrate before and after the ion cleaning step.Before of actual deposition, pre-sputtering of the Al target wascarried out in argon plasma for about 10e15min to remove the thinoxide layer present on the surface. During pre-sputtering, a

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e79 71

manually operated shutter was placed between the magnetroncathode and substrate. Just prior to AlN deposition, a thin layer ofaluminum (approximate thickness 0.1 mm) was sputtered on thesubstrate to enhance the bonding of the subsequent AlN layers. AlNdeposition was carried out at a working pressure of 0.25 Pa, in ni-trogen and argon plasma. Numerical values of the deposition pa-rameters are listed in Table 1. Before being fed into the depositionchamber, the sputtering gases were purified to the moisture andoxygen level of less than 10 ppb. During AlN deposition nitrogenconcentration in the gas mixture, cathode discharge power andsubstrate negative DC bias voltagewere varied. The deposition timefor different experiments was adjusted in such away that a uniformcoating thickness of around 1.5 mm was obtained, considering theeffect of deposition parameters on the deposition rate of AlN. In thepresent work, the deposition time for AlN was varied between 2and 4 h.

2.2. Characterization of deposited coating

The thickness of the coatings was calculated by the weight gainmethod assuming the density of the deposited AlN to be same asthat of bulk AlN (3.26 g/cm3). The thickness of AlN coatings re-ported in this work is not very accurate as the density of a coatingmay deviate significantly from the bulk due to the effect of porosity,non stoichiometry and presence of interstitials. However, in view ofthe small size of the substrate used in the present study, AlNcoatings were initially deposited also on large size circular foils ofSS (diameter: 75 mm) under exactly the same deposition conditionas employed for the smaller SS substrates. The calculated thicknessof AlN coatings deposited on SS foil was found to be very close tothat of the coating deposited on the actual substrate of smallerdimension. A grazing incidence X-ray diffractometer (GIXRD; X'PertPROMRD; model: PANalytical B.V.) was used to evaluate the crystalstructure of the coatings. GIXRD measurements were carried outusing monochromatized CuKa radiation (wavelength: 0.154 nm) atan incidence angle of 1�. To obtain the GIXRD pattern of the verythin Al interlayer a lower incidence angle of 0.3� was used. Theobtained GIXRD pattern was matched with the Joint Committee onPowder Diffraction Standards (JCPDS) database (Card No.: 25-1133,46-1200) to identify the reflections of AlN. The chemical compo-sition of the coatings was measured by a custom designed sec-ondary ion mass spectroscope (SIMS). The specimen wasbombarded by a pulsed liquidmetal ion source (69Gaþ), at energy of20 keV. For depth profiling studies, an argon ion etch gun at 5 keVwas interlaced with the primary ion gun. The secondary ion spectrawere acquired from an area of 150 mm � 150 mm. The hardness andYoung's modulus of the coatings were measured by nano-indentation method using Berkovich diamond indenter (UNHT,CSM Instruments, Switzerland) at a predefined load of 1 mN. Ascratch adhesion test facility (CSM, RST S/N: 27-0497) was used toprovide a measure of the coating/substrate adhesion. It was oper-ated in progressive mode with a minimum normal load of 0.9 N atthe point of contact to amaximum normal load of 20 N at the end of

Table 1Deposition conditions for the growth of AlN coatings.

Parameters

Base pressure 1.5 � 10�3 PaTotal gas (Ar þ N2) flow rate 10 sccmNitrogen concentration 40, 50, 60, 70, 80 and 90%Substrate temperature 250 �CSubstrate biasing 0, �25, �50, �75 and �100 VDischarge power 150, 200, 250, 300, 350, 400 and 450 WDuty cycle 75%Pulse frequency 125 kHz

the scratch length of 3 mm. Other parameters of adhesion test aregiven in Table 2. The applied normal load, frictional force andacoustic emission data were recorded online during scratch test.The resulting scratch impression was analyzed under an opticalmicroscope. Wear studies were performed using a computercontrolled reciprocating sliding machine (PLINT TE-77) that pro-duces a linear relative oscillating motion with a ball-on-flatconfiguration, similar to ASTM: G133-05. A spring loaded TungstenCarbide (WC) ball of diameter of 6 mm, confirming to AFBMAGrade-10wasmade to oscillate against the AlN coated specimens atdifferent normal loads and sliding frequencies, keeping theamplitude and sliding duration constant. Wear test parameters arelisted in Table 3. All wear measurements were carried out in anambient atmosphere at room temperature (30 ± 2 �C) with arelative humidity (RH) of 40 ± 5%. The variation of the coefficient offriction was recorded online and reported. The estimation of ballwear volume was based on the wear scar dimension of the ballmeasured by an optical microscope. From the measured ball wearvolume the specific ball wear rate [wear volume/(load � slidingdistance)] was calculated. Tests were repeated to confirm the dataconsistency of the results and the average values of the coefficientof friction and wear rate were reported.

3. Results

3.1. Crystal structure and chemical composition

Fig. 1a shows GIXRD pattern of the aluminum nitride coatingsdeposited under different N2 concentrations (40e90%) in thesputtering gas mixture, keeping the discharge power constant at150 W. For comparison, JCPDS data of the wurtzite AlN is alsoincluded in the figure. From Fig. 1a, it can be observed that the AlNhas crystallized in rock salt cubic crystal structure belonging toFm3m (225) space group and, at any value of N2 concentrationpeaks for wurtzite AlN were not present. Formation of cubic AlN isevident by the presence of diffraction peaks from the (111), (200),(220) and (311) crystallographic planes. However, the unresolvedbaselines indicate that there is some amorphous phase also presentin the coatings. Fig. 1b shows the variation of the intensity ratio ofthe (200) and (111) reflections of cubic phase AlN as a function of N2

concentration. It can be observed that the intensity ratio variedalmost linearly with N2 concentration. Moreover, the value of in-tensity ratio at 90% N2 is 57% higher than that at 40% N2. GIXRDpattern of the AlN coatings deposited at increasing discharge powerlevels but at a fixed N2 concentration of 60% is presented in Fig. 2a.It can be seen that as the discharge power is increased several broadpeaks from the (100), (002), (102), (110) and (112) planes ofwurtzite phase AlN belonging to P63mc (186) space group appearedin the pattern, in addition to the reflections from cubic AlN. How-ever, the peaks remained broad up to the discharge power of 350W.Another observation that can be derived from the diffractionpattern (Fig. 2a) is that up to 250 W of discharge power reflectionfrom the (200) plane of cubic AlN is the most prominent whereas athigher powers the (111) plane becomes dominant. Fig. 2b shows

Table 2Parameters of scratch adhesion test.

Parameters

Indenter type RockwellIndenter material DiamondIndenter tip radius 200 mmIndenter speed 4.7 mm/minLoading rate 30 N/minAE sensitivity 70%

Table 3Parameters of wear test.

Parameters

Normal load 3, 4 and 5 NSliding frequency 8, 10 and 13 HzSliding speed 20 mm/sTest duration 180 s

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e7972

GIXRD pattern of the AlN coatings obtained at higher dischargepowers along with substrate biasing. The combined effect of thesetwo parameters lead to the formation of wurtzite AlN preferentiallyaligned along the (100) plane. It can be observed that at a constantbias of �50 V the increase in discharge power leads to the gradualsuppression of the reflections of cubic AlN. Finally, at 450 W powerand �100 V bias the deposited coating was predominantly (100)oriented wurtzite AlN.

Fig. 3a shows SIMS compositionedepth profile of the AlNcoating deposited at 60% N2, 150 W discharge power and 0 V sub-strate bias. Counts for Al, AlN and AleO species are nearly constantacross the thickness, suggesting uniform composition of thecoating. Also, Fig. 3a shows relatively higher concentration of ox-ygen up to the initial depth of 20 nm. Fig. 3b and c shows thenormalized ratio of the counts of oxygen bonded with Al to the freeoxygen (O), for the coating deposited at 60 and 80% N2, respectively.It is clear that there is very little free oxygen present in the coating.However, oxygenwas observed in the form of aluminum oxide. It isto be noted that due to vast difference in the relative sensitivityfactor (RSF) for different species, the quantitative estimation is notconsidered to be very accurate by SIMS unless reliable standardsare available [28]. Therefore in this work, only qualitative chemicalcomposition of AlN coatings has been presented.

3.2. Hardness

Fig. 4 shows loadedisplacement curves from nanoindentation

Fig. 1. (a) GIXRD pattern of AlN coatings deposited on SS 304L at different N2 concentrationcubic AlN as a function of N2 concentration. Other deposition parameters; discharge power:cubic phase of AlN.

measurement carried out on the AlN coatings grown at 0 and �75 Vbias, keeping the N2 concentration and discharge power fixed at 90%and 450W, respectively, in both cases. As expected, the hardness (H)of the coating improved significantly on increasing the bias voltage.The measured values of hardness and Young's modulus (E) for AlNcoatings deposited at different bias voltages are shown in Fig. 5. It isclear that as the bias voltage is increased the corresponding values ofhardness and Young's modulus increases significantly. In Fig. 6, theH3/E2 ratio (plastic deformation resistance) [27] and R (elastic re-covery) are plotted for the AlN coatings obtained at different biasvoltages. The figure shows that an increase in bias voltage signifi-cantly increased the H3/E2 ratio (0.024 at 0 V to 0.061 at �100 V) aswell as R (42.21% at 0 V to 63.91% at �100 V).

3.3. Adhesion

Optical images of scratch impressions (left after scratch adhe-sion tests) on the AlN coatings deposited at 0, -25, �50, �75and �100 V bias are shown in Figs. 7e11. In Fig. 12, the critical loadfor adhesive failure and the approximate percentage delaminatedarea are plotted. The percentage delamination was calculated at15 N load using mesh method. From these figures it is clear thatwith increasing bias voltage the adhesion of the coating improvedsignificantly. The coating deposited without biasing delaminatedpartially at 7 N (Fig. 7) whereas, negligible delamination wasobserved up to 20 N for the coating grown at �100 V (Fig. 11). Theimprovement in adhesion is evident also from the gradual increasein load for adhesive failure (Fig. 12) and the decrease in delami-nation (%) occurring to the coating (Fig.12). The lateral spread in thescratch impressions is also getting reduced with increasing sub-strate bias, indicating decrease in plasticity of the coating.

3.4. Wear and friction

Fig. 13aed shows optical micrographs of the wear scar formed

s and, GIXRD pattern of Al interlayer. (b) Intensity ratio of the (200) and (111) planes of150 W, substrate temperature: 250 �C. Here, the symbol ‘c’ is being used to denote the

Fig. 2. GIXRD pattern of AlN coatings deposited on SS 304L at: (a) various discharge powers and (b) different combinations of discharge power and bias voltage. Other depositionparameters; N2 concentration: 60%, substrate temperature: 250 �C. Here, the symbols ‘c’ and ‘h’ are being used to denote the cubic and hexagonal phases of AlN, respectively.

0 20 40 60 80 100 120 140 160101

102

103

104

105

106

Cou

nts

(s-1

)

Depth (nm)

Al AlN Al-O

(a)

0 20 40 60 80 100 120 140 1600.0

0.2

0.4

0.6

0.8

1.0

stnuoc)O/

OlA(

dezilamro

N

Depth (nm)

(b)

0 20 40 60 80 100 120 140 1600.0

0.2

0.4

0.6

0.8

1.0

stnuoc)O/

OlA(

de zil amro

N

Depth (nm)

(c)

Fig. 3. (a) SIMS compositionedepth profile of AlN coatings deposited at 60% N2 and normalized AlO/O counts for AlN coating deposited at (b) 60 and (c) 80% N2.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e79 73

Fig. 4. Loadedisplacement curves for AlN coatings deposited at two different substratebias voltages.

Fig. 5. Hardness, Young's modulus and specific ball wear rate of AlN coating as afunction of substrate bias voltage.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e7974

on the tungsten carbide balls after sliding against the AlN coatingsdeposited at various bias voltages. It can be observed that the ballwear scar diameter was larger against the AlN coatings deposited athigher bias voltages. This is also corroborated by the specific ballwear rate data shown in Fig. 5. Plots of coefficient of friction vs.

Fig. 6. H3/E2 ratio and elastic recovery (R) for AlN coatings deposited at differentsubstrate bias voltages.

sliding distance for wear test carried out at three different normalloads (at constant sliding frequency) are shown in Fig. 14a. Thecoating in this case was deposited at 450 W discharge power, 60%N2 concentration and 0 V substrate bias. It is clear that at constantfrequency, the coefficient of friction decreased marginally withincrease in the normal load. Moreover, the sliding frequency (atconstant load) had no appreciable effect on the coefficient of fric-tion (Fig. 14b). The variation of the coefficient of dynamic frictionwith substrate bias for wear test conducted at three differentnormal loads is plotted in Fig. 15. It can be found that at each load,the coefficient of dynamic friction increased continuously withincrease in the bias voltage. Nearly in all cases, the value of thecoefficient of dynamic friction for AlN coatings was in the range of0.4e0.6.

4. Discussion

Formation of c-axis oriented wurtzite aluminum nitride coat-ings/films by DC/RF magnetron sputtering on silicon substrate hasbeen widely reported by many authors [24e26]. However, limitedliterature is available on the deposition of cubic (zinc blende or rocksalt) AlN coatings by magnetron sputtering. In our previous re-ported study [28], it was brought out that the cubic AlN coating canbe deposited on SS 304L substrate by employing balancedmagnetron configuration, low discharge power and zero substratebias. Several other authors have also reported the formation ofcubic AlN but only as a minor constituent along with the pre-dominant wurtzite phase AlN. For example, Khanna et al. [29]found weak XRD peaks from the (111) and (200) planes of cubicAlN together with the major wurtzite phase AlN grown on Si (111),by AC reactive magnetron sputtering. They attributed the presenceof cubic AlN to the diamond cubic crystal structure of the silicon.Similarly, Cai et al. [30] reported the (200) peak of cubic AlN alongwith the wurtzite phase AlN, for the coating grown on LaAlO3substrate by RF reactive magnetron sputtering. Here, again thereason can be attributed to the cubic perovskite structure of theLaAlO3 helping formation of cubic AlN. Recently, Shanmugan et al.[31] reported mixed phase AlN coatings grown on the metallicsubstrates like copper and aluminum by DC magnetron sputtering.The appearance of sharp XRD peaks of cubic AlN in the presentwork for the coatings obtained at various N2 concentrations can beexplained as follows. The cubic crystal structure with matchinglattice parameter of the aluminum interlayer, the low dischargepower and the balanced configuration of magnetron used forsputtering set the conditions required for the growth of cubic phaseAlN [28]. It is well known that in the case of balanced magnetronsputtering, the ion current drawn at the substrate is insufficient tomodify the structure of the film. Therefore, the depositing specieshave a tendency to retain the structure of the substrate up to a fewatomic layers, unless a high substrate bias voltage is applied[28,32]. The influence of substrate on the crystal structure of AlNcoatings is further supported by the work of Ababneh et al. [33],who reported c-axis oriented wurtzite AlN coatings on titanium.Titanium having hexagonal crystal structure promoted the growthof wurtzite AlN. The (111) and (200) diffraction peaks of cubic AlNwere also observed by Garcia-Mendez et al. [34], for their coatingdeposited at a relatively higher pressure by magnetron sputtering.They argued that, at low deposition pressure the sputtered speciespossess sufficient energy to form the stable wurtzite AlN. It can alsobe argued that the XRD peaks attributed to cubic AlN in this workcan be instead from the Al interlayer, as the cubic AlN and Al havevery close lattice size matching therefore, their diffraction peaksappears at nearly the same angle. However, this possibility can beruled out from the fact that the condition for the deposition of Alinterlayer was exactly same for all experiments. Hence, it is highly

Fig. 7. Optical images of scratch impression on AlN coated SS 304L. Coating was deposited at: 0 V substrate bias, 250 �C temperature, 450 W discharge power and 60% N2

concentration.

Fig. 8. Optical images of scratch impression on AlN coated SS 304L. Coating was deposited at: �25 V substrate bias, 250 �C temperature, 450 W discharge power and 60% N2

concentration.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e79 75

unlikely that the intensity ratio of the (200) and (111) reflectionswill vary so much as described in the result section, if it is assumedthat these reflections belonged to Al only. However, it is more likelythat the (111), (200), (220) and (311) planes of cubic AlN have

Fig. 9. Optical images of scratch impression on AlN coated SS 304L. Coating was depositeconcentration.

grown over the (111), (200), (220) and (311) planes of aluminuminterlayer, respectively. The increase in the intensity ratio of theXRD reflections from the (200) and (111) planes is due to the changein growth rate of these planes with N2 concentration. Many authors

d at: �50 V substrate bias, 250 �C temperature, 450 W discharge power and 60% N2

Fig. 10. Optical images of scratch impression on AlN coated SS 304L. Coating was deposited at: �75 V substrate bias, 250 �C temperature, 450 W discharge power and 60% N2

concentration.

Fig. 11. Optical images of scratch impression on AlN coated SS 304L. Coating was deposited at: �100 V substrate bias, 250 �C temperature, 450 W discharge power and 60% N2

concentration.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e7976

have pointed out the change in preferential orientation of AlN onincreasing the N2 concentration during reactive sputtering [35,36].Hence, we can arrive at the conclusion that a mixture of amorphous

-100 -80 -60 -40 -20 0

6

8

10

12

14

16

18

20

22 Critical load for adhesive failureDelaminated area (%) at 15 N load

Substrate bias voltage (V)

)N(

daollacitirC

Del

amin

ated

are

a (%

)

0

5

10

15

20

25

30

Fig. 12. Critical load for adhesive failure for AlN coatings obtained at different sub-strate bias voltages and delaminated area (%) of the coating surface along the scratchimpressions at 15 N load.

and cubic AlN was formed over the Al interlayer at various N2concentrations. The appearance of broad peaks corresponding tothe (100) plane of wurtzite AlN with increase in discharge powercan be attributed to the increased fluence [37,38] of the sputteredAl atoms on the substrate surface. In our earlier reported work [28],an increase in discharge power resulted in the appearance of the(002) broad peak of wurtzite AlN. The difference in orientation canbe correlated to the difference in the characteristics of DC andpulsed DC plasma. Formation of the (100) oriented wurtzite AlN onthe negatively biased substrate at higher discharge powers can beattributed to the further increase in the mobility of reactive species[39]. The application of bias voltage results in drawing of some ofthe Arþ ions from plasma towards the substrate. The energetic Arþ

ions transfer large amount of momentum to the reactive species.Furthermore, an increase in bias voltagemakes the bombarding Arþ

ions more energetic. Thus gradual increase in bias goes on raisingthe mobility of the reactive species on the substrate surface, help-ing formation of stable wurtzite AlN.

The presence of oxygen in AlN coatings as detected by SIMS ismainly due to the residual oxygen present in the depositionchamber, the oxygen impurity transported to the depositionchamber by the sputtering gases and also the oxygen present in the

Fig. 13. Optical images (�10) of the worn surface of tungsten carbide balls used for wear test of AlN coatings deposited on SS 304L at (a) 0, (b) �50, (c) �75 and (d) �100 V substratebias.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e79 77

target itself used for sputtering. The higher oxygen at the surface ofthe coating is due to the ingress of oxygen from the atmosphere.

As mentioned earlier, the application of greater bias voltageleads to increased number of heavier Arþ ions drawn towards thesubstrate, resulting in increased bombardment of the reactivespecies. Hence, the reactive species becomemore energetic and AlNcoatings with better crystallinity are obtained. In addition, an in-crease in bias also leads to increase in compressive stress and defectdensity in the coating. Vacandio et al. [40] reported about1.3e2.98 GPa compressive stresses in the AlN coatings obtained bysputtering on stainless steel substrate. Chun [41] reportedincreased hardness of the TiN coatings grown on negatively biasedSi (100) substrate by sputtering. He attributed the increased hard-ness to the improved crystallinity and increased level of compres-sive stress and defect density in the coatings. Similarly, Medjaniet al. [42] reported an increase in compressive stress of AlN coatingswith increasing substrate bias, for the coating deposited bymagnetron sputtering on silicon. The values of hardness andYoung's modulus for AlN coatings reported in this work are inagreement with the values reported for AlN coatings in the litera-ture [43].

Due to large difference in the coefficient of thermal expansionand lattice size of AlN and SS 304L, high stresses are generated inthe coating-substrate interfacial region. The quantity of stress is sohigh [40] that the brittle AlN coating is not stable on SS 304L. In fewof the experiments where AlN was deposited directly without Alinterlayer resulted in complete delamination or even crumbling ofthe coating. On the other hand, introduction of a thin layer ofductile Al reduces the interfacial stresses substantially [44] since,the lattice size of Al matches well with that of AlN and, Al bondswell with SS 304L. Hence, the coating becomes adherent. Theadherence of the coating is further improved on application of thenegative bias voltage to the substrate. This happens due to knock-ing off loosely bonded particles by the bombarding Arþ ions. Inaddition to this, the bombarding Arþ ions remove the adsorbedcontaminants such as oxygen from the substrate surface and impart

kinetic energy to the reactive species thereby enhancing thediffusion process. This results in better adhesion and crystallinity ofthe growing film. However, the bias voltage cannot be increasedindefinitely since it induces residual stresses in the coating thatagain leads to poor adhesion. Therefore, in the present work, amaximum bias voltage of �100 V was applied to the substrate. Anincrease in bias voltage significantly increased the value of Young'smodulus for AlN coatings. Therefore, the load required for plasticdeformation increased. This was supported by the reduced lateralspread of the scratch impressions created in the coating duringscratch test. The reduced lateral spread can also be correlated to thehigherH3/E2 ratio observed for the AlN coatings deposited at higherbias voltages, during nanoindentation measurement. A higher H3/E2 ratio means increased resistance to plastic deformation [27] orlow contact pressure [45], leading to reduced lateral spread in thecoating.

An increase in bias voltage could make the AlN coatings hardand therefore difficult to deform. This fact is supported by thehigher plastic deformation resistance (H3/E2) and elastic recovery(R) values for these coatings, measured during nanoindentationtest. Hence, the coating becomes difficult to be worn by the tung-sten carbide balls. Guo et al. [45] reported significant increase in thewear resistance of AlN coatings by incorporating Cu atoms into AlN.Similarly, Jagannadham et al. [46] reported a lowwear rate for hardAlN coatings obtained by pulsed laser deposition technique onsilicon.

The higher normal load increased the frictional force betweenthe ball and the coating, during wear test. This could result in moreheat dissipated on the coating surface leading to softening of thecoating. Hence, the coefficient of friction decreased. The increase inthe coefficient of friction for the AlN coatings deposited at higherbias voltages may be due to the increased hardness of AlN. Duringwear test, the wear debris also acts as an abrading material alongwith the abrading ball. The wear debris of a relatively hardercoating could also result in the increased value of the coefficient offriction. The coefficient of friction for AlN coatings reported in this

0 500 1000 1500 2000 2500 30000.0

0.1

0.2

0.3

0.4

0.5

0.6

noitcirfci

manydfotneiciffeoC

Sliding distance (mm)

3 N 4 N 5 N

(a)

0 500 1000 1500 2000 2500 30000.0

0.1

0.2

0.3

0.4

0.5

noitcirfci

manydfotneiciffeoC

Sliding distance (mm)

(b)

Fig. 14. Coefficient of dynamic friction for AlN coating at (a) 3, 4 and 5 N loads and 8 Hzsliding frequency and (b) 5 N load and 13 Hz sliding frequency, during wear test; thecoating was deposited at 250 �C temperature, 450 W discharge power and 60% N2

concentration.

-110 -100 -90 -80 -70 -60 -50 -40 -30 -200.42

0.44

0.46

0.48

0.50

0.52

0.54

0.56

0.58

noitcirfci

manydfo tneiciffeoC

Substrate bias voltage (V)

3 N 4 N 5 N

Fig. 15. Coefficient of dynamic friction for AlN coatings deposited at various biasvoltages, during wear test conducted at various normal loads, 8 Hz sliding frequencyand 180 s test duration. Other deposition parameters; N2 concentration: 60%, dischargepower: 450 W, substrate temperature: 250 �C.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e7978

work is in excellent agreement with the same reported by Cabreraet al. [27] for sputter deposited AlN coatings.

5. Conclusions

This work has shown that it is possible to obtain relatively highcrystalline and adherent aluminum nitride coatings on stainlesssteel substrate by reactive magnetron sputtering. Introducing a thininterlayer of Al improved the adhesion of AlN coatings significantly.Discharge power and substrate negative biasing had large effect onthe crystal structure of the growing film and, at sufficiently highdischarge power and bias voltage, wurtzite AlN oriented along the(100) plane was obtained. However, at low discharge power andbias voltage, the grown coating was a mixture of wurtzite and rocksalt AlN. Nitrogen concentration in the sputtering gas mixture hadlarge influence on the preferred orientation of the cubic phase AlN.Some amount of oxygen bonded with Al was also detected in theAlN coatings. Substrate biasing improved the adhesion, hardnessand wear resistance of AlN coatings. On the other hand, the coef-ficient of frictionwas higher for the coating deposited with a highersubstrate bias. The coefficient of friction decreased with increase innormal load whereas, negligible change in the coefficient of frictionwas observed at various sliding frequencies.

References

[1] S.E. Wulf, W. Krauss, J. Konys, Fusion Eng. Des. 89 (2014) 2368e2372.[2] P.F. Zheng, T. Nagasaka, T. Muroga, J.M. Chen, Fusion Eng. Des. 89 (2014)

1648e1652.[3] L. Huang, X. Hu, W. Yan, W. Sha, F. Xiao, Y. Shan, K. Yang, Mater. Des. 63 (2014)

333e335.[4] P. Chaudhuri, S. Ranjithkumar, D. Sharma, C. Danani, H.L. Swami, R. Bhattacharya,

A. Patel, E.R. Kumar, K.N. Vyas, Fusion Eng. Des. 89 (2014) 1362e1369.[5] C.P.C. Wong, M. Abdou, M. Dagher, Y. Katoh, R.J. Kurtz, S. Malang, E.P. Marriott,

B.J. Merrill, K. Messadek, N.B. Morley, M.E. Sawan, S. Sharafat, S. Smolentsev,D.K. Sze, S. Willms, A. Ying, M.Z. Youssef, Fusion Eng. Des. 85 (2010)1129e1132.

[6] Q. Huang, Q. Wu, S. Liu, C. Li, B. Huang, L. Peng, S. Zheng, Q. Han, Y. Wu, FDSTeam, Fusion Eng. Des. 86 (2011) 2611e2615.

[7] T. Muroga, Mater. Trans. 46 (2005) 405e411.[8] P.J. Bhuyan, K.S. Goswami, IEEE T. Plasma Sci. 36 (2008) 1955e1959.[9] X. Zengyu, P. Chuanjie, J. Weihong, W. Wenhao, L. Wenzhong, Q. Jiapu, Fusion

Eng. Des. 39e40 (1998) 793e798.[10] J. Konys, W. Krauss, N. Holstein, Fusion Eng. Des. 85 (2010) 2141e2145.[11] D.L. Smith, J.H. Park, I. Lyublinski, V. Evtikhin, A. Perujo, H. Glassbrenner,

T. Terai, S. Zinkle, Fusion Eng. Des. 61e62 (2002) 629e641.[12] R.K. Choudhary, V. Kain, R.C. Hubli, B, J. Min. Metall. Sect. B-Metall 50 (2)

(2014) 165e170.[13] P.A. Rayjada, N.P. Vaghela, N.L. Chauhan, A. Sircar, E. Rajendrakumar,

L.M. Manocha, P.M. Raole, Fusion Sci. Technol. 5 (2014) 194e198.[14] G.K. Zhang, C.A. Chen, D.L. Luo, X.L. Wang, Fusion Eng. Des. 87 (2012)

1370e1375.[15] Q. Zhan, W. Zhao, H. Yang, Y. Hatano, X. Yuan, T. Nozaki, X. Zhu, J. Nucl. Mater.

464 (2015) 135e139.[16] W. Krauss, J. Konys, N. Holstein, H. Zimmermann, J. Nucl. Mater. 417 (2011)

1233e1236.[17] I.V. Vitkovsky, A.V. Gorunov, V.I. Engelko, I.R. Kirillov, V.M. Khoroshikh,

N.D. Kraev, A.P. Ogorodnikov, V.Y. Petrovsky, A.E. Rusanov, Fusion Eng. Des.61e62 (2002) 739e743.

[18] K. Natesan, C.B. Reed, D.L. Rink, R.C. Haglund, J. Nucl. Mater. 258e263 (1998)488e494.

[19] P. Hubberstey, J. Nucl. Mater. 247 (1997) 208e214.[20] E.M. de les Valls, L.A. Sedano, L. Batet, I. Ricapito, A. Aiello, O. Gastaldi,

F. Gabriel, J. Nucl. Mater. 376 (2008) 353e357.[21] A. Sawada, A. Suzuki, T. Terai, T. Muroga, J. Nucl. Mater. 329e333 (2004)

1411e1413.[22] T. Watanabe, M. Kondo, T. Nagasaka, A. Sagara, J. Plasma Fusion Res. 9 (2010)

342e347.[23] A. Suzuki, T. Muroga, B.A. Pint, T. Yoneoka, S. Tanaka, Fusion Eng. Des. 69

(2003) 397e401.[24] F. Martin, P. Muralt, M.A. Dubois, A. Pezous, J. Vac. Sci. Technol. A 22 (2004)

361e365.[25] Kuo-Sheng Kao, Chien-Chuan Cheng, Ying-Chuang Chen, Jpn. J. Appl. Phys. 40

(2001) 4969e4973.[26] A.V. Singh, S. Chandra, G. Bose, Thin Solid Films 519 (2011) 5846e5853.[27] G. Cabrera, J.C. Caicedo, C. Amaya, L. Yate, J.M. Saldana, P. Prieto, Mater. Chem.

Phys. 125 (2011) 576e586.

R.K. Choudhary et al. / Journal of Nuclear Materials 466 (2015) 69e79 79

[28] R.K. Choudhary, P. Mishra, R.C. Hubli, Surf. Eng. 30 (2014) 535e539.[29] A. Khanna, D.G. Bhat, J. Vac. Sci. Technol. A 25 (2007) 557e565.[30] W.X. Cai, L. Kang, P.H. Wu, S.Z. Yang, Z.M. Ji, Supercond. Sci. Technol. 15 (2002)

1781e1783.[31] S. Shanmugan, D. Mutharasu, P. Anithambigai, N. Teeba, I.A. Razak, J. Ceram.

Process. 14 (2013) 385e390.[32] P.J. Kelly, R.D. Arnell, Vacuum 56 (2000) 159e172.[33] A. Ababneh, M. Alsumady, H. Seidel, T. Manzaneque, J. Garcia, J.L. Rojas,

A. Bittner, U. Schmid, Appl. Surf. Sci. 259 (2012) 59e65.[34] M. Garcia-Mendez, S. Morales-Rodriguez, R. Machorro, W. De La Cruz, Rev.

Mex. Fis. 54 (2008) 271e278.[35] A. Kale, R.S. Brusa, A. Miotello, Appl. Surf. Sci. 258 (2012) 3450e3454.[36] S. Cho, J. Cryst. Growth 326 (2011) 179e182.[37] Z. Zhang, C. Bao, S. Ma, L. Zhang, S. Hou, J. Aust. Ceram. Soc. 48 (2012)

214e222.[38] M.A. Moreira, I. Doi, J.F. Souza, J.A. Diniz, Microelectron. Eng. 88 (2011) 802e806.

[39] M. Clement, E. Iborra, J. Sangrador, A. Sanz-Hervas, L. Vergara, M. Aguilar,J. Appl. Phy. 94 (2003) 1495e1500.

[40] F. Vacandio, Y. Massiani, P. Gergaud, O. Thomas, Thin Solid Films 359 (2000)221e227.

[41] Sung-Yong Chun, J. Korean Phys. Soc. 56 (2010) 1134e1139.[42] F. Medjani, R. Sanjines, G. Allidi, A. Karimi, Thin Solid Films 515 (2006)

260e265.[43] Q. Wei, X. Zhang, D. Liu, J. Li, K. Zhou, D. Zhang, Z. Yu, Trans. Nonferrous Met.

Soc. China 24 (2014) 2845e2855.[44] J.C. Olivera, A. Cavaleiro, M.T. Vieira, Surf. Coat. Technol. 151e152 (2002)

466e470.[45] J. Guo, H. Wang, F. Meng, X. Liu, F. Huang, Surf. Coat. Technol. 228 (2013)

68e75.[46] K. Jagannadham, A.K. Sharma, Q. Wei, R. Kalyanraman, J. Narayan, J. Vac. Sci.

Technol. A 16 (1998) 2804e2815.