Surface & Coatings Technology 258 (2014) 1–16

Contents lists available at ScienceDirect

Surface & Coatings Technology

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

Toward hard yet tough ceramic coatings

Yu Xi Wang a, Sam Zhang b,⁎a Tum Create, #10-02 Create Tower, 1 Create Way, Singapore 138602, Singaporeb School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore 639798, Singapore

⁎ Corresponding author.E-mail addresses: yuxi.wang@tum-create.edu.sg (Y.X.W

(S. Zhang).

http://dx.doi.org/10.1016/j.surfcoat.2014.07.0070257-8972/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 14 May 2014Accepted in revised form 2 July 2014Available online 8 July 2014

Keywords:Hard yet toughCeramic coatingsFilmsCoatings

Over the past decades, hard and super hard ceramic coatings have been developed and widely used in variousindustrial applications. Meanwhile, an increasing number of studies have realized that the toughness is just ascrucial, if not more, than hardness especially for ceramic coatings. However, hardness and toughness do not gonaturally hand in hand. In other words, hard coatings usually are brittle and less durable while toughened coat-ings are of lower strength. For practical engineering applications, it is more desirable to have coatings with highhardness without sacrificing toughness too much. In this article, a review is presented on continuous progress torealize hard-yet-tough ceramic coatings from an angle of hardening as well as toughening.

ang),MSYZHANG@ntu.edu.sg

© 2014 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12. Toward hard-yet-tough ceramic coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.1. Ways toward hardening of ceramic coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1.1. Hardening via grain size refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1.2. Hardening via grain boundary reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22.1.3. Hardening via solid-state solution and precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.1.4. Hardening via ion bombardment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42.1.5. Hardening via multilayering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.2. Ways toward toughening of ceramic coatings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2.1. Toughening through introducing a toughening agent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2.2. Phase transformation toughening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2.3. Compressive stress toughening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2.4. Toughening through optimization of coating architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3. Toughness evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134. Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1. Introduction

Ceramic coatings have been widely used in various engineering sys-tems, for instance, protecting structuralmaterials in harsh environment,prolonging life of manufacturing tools by improving wear/corrosion

resistance and enhancing efficiency in energy storage and/or conver-sion. However, a low toughness limits the use of ceramic coatings. Themost typical example is the sudden failure of machining tools withhard (H N 20 GPa) or super-hard coatings (H N 40 GPa) when they arein contactwith large foreign impact [1]. To solve this problem, a numberof works have been put forth to understand the origin of cracks in hardor super-hard coatings [2–5]. Many toughening methods have beendeveloped to obtain coatings of both improved hardness and toughness[6,7]. In this article, we reviewed the academic journey targeting hard-yet-tough ceramic coatings in the past decades.

2 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

2. Toward hard-yet-tough ceramic coatings

The foremost feature of a machining tool is high hardness. Hardnessis the resistance of amaterial against plastic deformation. Experimental-ly, hardness is measured using a stiff and hard indenter. In most cases,the indenter is a diamond pyramid or cone of a given shape, which ispressed into the surface of the material with a given load P. After theload is removed, the remnant contact area A can be observed with amicroscope. The load P over this contact area P / A is referred to as theindentation hardness H or, H = P / A. Nowadays, an automatic load–depth-sensing indentation instrument automates this process and thehardness is determined from analysis of the loading–unloading curve(i.e. Oliver–Pharr method [8]). It should be emphasized that the hard-ness values measured using indentation are sensitive to a number offactors: indenter geometry, tip rounding, indentation size effect andsubstrate condition in terms of surface roughness, surface oxidationand surface piling-up/sinking-in [9]. Toughness, on the other hand,measures the resistance to crack propagation or energy consumed tofracture a pre-cracked sample. If the fracture is abrupt (i.e., little plasticdeformation) thematerial is referred to as brittle. If the fracture requiresconsiderable plasticwork and is accompaniedwith steady drop in stressbefore complete separation, the material is ductile, in other words,tough. To evaluate the toughness, a stress intensity factor KIC (with aunit of MPa m1/2) is generally used and can be readily determined ac-cording to ASTM standards [10,11]. The subscript “IC” stands for themode I crack opening, where the crack opens under a normal tensilestress perpendicular to the plane of the crack. Over a long history inthe development of fracture mechanics theory, many other fracture pa-rameters, in terms of energy release rate, J-integral and crack-tip open-ing displacement/angle, are also important in experimental evaluationof toughness. The values of these parameters are technically influencedby factors, for instance, loading-rate, crack-tip constraint, fracture insta-bility and environmental temperature [12].

An ideal coatingwould be hard yet sustaining a sudden impactwith-out catastrophic failure. That iswhyhard-yet-tough ceramic coating hasbeen the focal point of research for the last few decades. Needless to say,fabricating such a coating is difficult because of the natural conflict be-tween hardness and toughness, or an increase of hardness usuallygoes at an expense of toughness. Fig. 1 shows the typical dilemma inhard ceramic coatings [6] and the engineer's dream (“hard yettough”). To realize a “hard yet tough” ceramic coating, one needs to con-sider ways toward both hardening and toughening.

2.1. Ways toward hardening of ceramic coatings

Hardness is defined as the resistance of a material to plastic defor-mation. For coarse-grained ceramics, plastic deformation occurs

Fig. 1. Schematic of the current status of hard ceramic coatings.Re-plot from [6].

predominantly through dislocation. Under this circumstance, increasingthe resistance to the dislocation movement is the essence of hardening.However for fine-grained ceramics (e.g. ceramics with grain size lessthan 10 nm or amorphous phase), deformation by grain boundarymicrocracking and sliding (i.e. quasi-plasticity) is the major cause ofstrength decline [13]. As of yet, the evolution of cracking can be ob-served using advanced transmission electron microscopy (TEM) [14].Currently, several major strengthening mechanisms are active in hard-ening ceramic coatings: (i) grain size refinement, (ii) grain boundary re-inforcement, (iii) solid solution hardening, (iv) multilayer hardeningand (v) ion bombardment/stressing hardening. It should be mentionedthat some of the above hardening mechanisms are not applicable toamorphous coating systems, for instance, diamond like carbon (DLC),which is an important coating for industrial application due to its intrin-sically high hardness. Hardening of DLC lies on the proper doping of for-eign elements (e.g. Si, B, N, W, Mo, Ti and Ni), microstructuraloptimization (i.e. gradient ormultilayer) and good control of hybridizedcarbon bond ratio (i.e. sp3/sp2). Limited by the length of the article, it isrecommended to followother related reviewpapers or book chapters tobetter understand the mechanisms involved [15,16].

2.1.1. Hardening via grain size refinementThe dominantmechanismof plastic deformation of a crystallinema-

terial is the generation and motion of dislocations. Under an appliedstress, existing dislocations and dislocations nucleated mostly fromFrank–Read sources will move through the crystal structure until agrain boundary is encountered, where the large atomic mismatch be-tween different grains creates a repulsive stress field to oppose contin-ued dislocationmovement. Asmore andmore dislocations propagate tothis boundary, a “pile-up” occurs. These dislocations will generate re-pulsive stress fields, countering the energy barrier to cross the bound-ary. As the energy barrier is overcome finally, dislocations move acrossthe boundary, leading to a further deformation in thematerial. Decreaseof grain size, however, decreases the amount of possible pile-ups at theboundary but increases the threshold of applied stress to move a dislo-cation across a grain boundary, thus increases strength. Theoretically,the stress needed for generation and motion of dislocations increasesin inverse proportion to the distance of the pinning points in the dislo-cation network, and the strength increases with decreasing crystallitesize are well governed by the Hall–Petch relation [17,18]

H ¼ H0 þ kd−1=2

where H0 is the intrinsic hardness, d is the grain size and k is a constantparameter for a given material.

2.1.2. Hardening via grain boundary reinforcementThe Hall–Petch effect governs the coarse-grained materials (i.e.

grain size d N ca. 30 nm). However, as the grain size is decreaseddown to the order of a few tens of or even a few nanometers, this ruleceases to function perfectly. Many researchers have reported an abnor-mal behavior and correlated it to an inverse or reverse Hall–Petch effect.As seen in Fig. 2, a maximum hardness is achievedwhen d is close to ca.10 nm [4]. It is suggested that the traditional view of dislocation-drivenplasticity in polycrystalline materials needs to be revisited and the wayof achieving further hardening needs to be reconsidered. Several factorssuch as grain boundary sliding, creep diffusion, triple junctions, and im-purities could contribute to reverse Hall–Petch effect. Among them, thegrain boundary sliding via a certain accommodation mechanism is con-sidered the most primary cause. When grain sizes are below a criticalvalue (i.e. d b ca. 10 nm [19]), nanocrystallites contain a large fractionof atoms at interfaces. In this case, pile-up of the dislocations againstgrain boundaries are hardly expected to occur since the size of aFrank–Read source is smaller than the grain size. With a phenomeno-logical mesoscopic model, Hahn et al. have predicted a critical grainsize at which the grain boundary sliding becomes dominant [20,21].

Fig. 2. Schematic illustration of coating hardness as a function of the size d of grains [4].

3Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

Coupled with the numerical study, molecular dynamics (MD)modelingwas employed to predict thenanocrystalline plasticity in the nanograinswith size below 10 nm [22]. In this study, the primary cause for thenanograin deformation was verified as localized sliding of atoms inthe grain boundary, triggered by atomic shuffling and stress-assistedfree volume migration.

To overhaul the size limitation in hardening, Veprek and Reiprichproposed a ternary nanocomposite, formed by thermodynamically driv-en spinodal segregation. This system has a type nc-MenN/a-Si3N4,where Men represents transition metal like Ti, W, V, and Zr, nc- repre-sents nanocrystalline and a- represents amorphous. It should be pointedout that “a-” refers to X-ray amorphous only with no signal of Bragg re-flections [23,24]. The use of transition metals lies on the large cohesiveenergy and high bulk modulus (Kb), allowing formation of carbides, ni-trides, borides or oxides with very low compressibility and highstrength. In a typical TiSiN nanocomposite (Fig. 3), the crystallites oftransition metal nitride (d b 10 nm) were separated by about one ortwo monolayers of a-Si3N4. The thin layer of Si3N4 serves as strong“glue” amongMeN nanocrystals, effectively suppressing the occurrenceof grain boundary sliding and thus the super high hardness (i.e. HN 50 GPa) [25]. For instance, the nc-TiN/a-Si3N4 and the nc-W2N/a-Si3N4 nanocomposite appeared to increase hardness with decreasingcrystallite size down to 4 nm. Although the claimed ultra-high hardnessvalue ranged from 80 to 105 GPa has evoked an intensive debate [26,27], the concept brings about a remarkable progress in pursuing coat-ings with super high hardness.

Similar results have been found in other ternary systems, for exam-ple nc-TiN/a-BN, nc-TiN/a-BN/a-TiB2 [28], nc-TiC/a-C [29,30], andnc-WC/a-C [31], and in quandary systems [1] such as nc-CrAlN/a-SiNx,

Fig. 3. Schematics of the nanostructure of nc-TiN/a-Si3N4/a and nc-TiSi2 [25].

and nc-TiAlN/a-SiNx. According to the design concept, themost promis-ing option in creating a strong and hard material is the realization of anisotropic nanocomposite consisting of two strongly immiscible mate-rials which are generated from a strong thermodynamically drivenspinodal phase segregation. In addition, an extremely thin interfaciallayer (i.e. close to one monolayer as reported in the nc-TiN/a-Si3N4 sys-tem) is beneficial to further minimize internal weakening in an interfa-cial material and diminish misfit dislocations associated with theinterface. The crucial effect of the interfacial layer thickness has been re-cently explained in terms of the critical thickness of a pseudomorphic orheteroepitaxial growth [32]. The experimental results of S derberg et al.[33] and theoretical calculations of Zhang et al. [34] and Hao et al. [35]confirmed that one interfacial monolayer was the most stable configu-ration in a coherent TiN–Si3N4 system. In this system, the de-cohesionstrength was found to be greater than that of a single crystal Si3N4. Itwas also found that a thicker interfacial layer would lead to the forma-tion of real amorphous SiNx, resulting to the loss of structural coherencebetween TiN and SiN and thus the loss of strength or hardness. Fig. 4 il-lustrates the hardness tendency of TiSiN as a function of Si3N4 content[36]. Estimation of the mean grain separation at the composition of15–20 at.% Si3N4 (inset B) shows only a fewmonolayers of silicon nitridein the interfacial layer. At high silicon nitride fractions, the mean grainseparation becomes greater, leading to the large degradation of coher-ence between TiN and SiN, and eventually the hardness drop.

2.1.3. Hardening via solid-state solution and precipitationThe development of solid-state solution (or solid solution) can be

traced back to bronze-age when the use of alloy emerged. The insertionof atoms of an alloying element into the interstitial or substitutional po-sition of the matrix atoms leads to physical homogeneity disruptionwith crystal lattice distortion. This distortion introduces a local stressfield, impeding the motion of dislocations and thus increases thestrength. In thin films and coatings, this principle works in the sameway. Insertion of B into TiN proved strength improvement, since thegliding of the dislocations formed inside the crystallites was hinderedby the lattice distortion resultant strain [37]. In bulk materials, solidsolutioning can be achieved through heat treatment while in thinfilms and coatings, the solid solutioning can be realized through non-equilibrium growth in physical/chemical vapor deposition. In reactivearc evaporation, a highly ionized flux of ions was used to synthesize apromising metastable super saturated Ti–B–N solid solution. The maxi-mumhardnesswas obtained at 34.5 GPa. An effectiveway of optimizinghardness of solid solution is changing the soluble content of the foreignelement. Itwas once reported that amaximumof 17.4 at.% of B rendered

Fig. 4. Hardness of nc-TiN/a-Si3N4 as a function of silicon nitride fraction. Insets illustratethe schematic nanostructure for different compositions [36].

4 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

an increased hardness up to 43 GPa [37]. Other examples are TiCN [38],CrZrN [39], TiAlN [40], and CrAlN [41]. TiAlN and CrAlN are intensivelystudied in recent years since the insertion of Al in TiN or CrN can provideimproved thermal stability in addition to enhanced strength [42].

Unlike solid solution hardening which relies on the process ofmerging two or more different elements into one crystalline structure,precipitation hardening comes from the decomposition of multiple ele-mental composite, and can be achieved by altering solid-state solubilitythrough heat treatment (i.e. annealing) [43]. Supersaturated phases canbe easily obtained by PVD or CVD due to limited atomic kinetic energy.As-deposited (TiAl)N [44] and Ti(BN) [45] exhibited a dense columnarmicrostructure of supersaturated NaCl structure, in which the Ti posi-tion in fcc-TiN was occupied by Al and the N position was substitutedby B, respectively. During annealing, they underwent spontaneousspinodal decomposition and transformed into coherent cubic phase do-mains with nanograins, eventually leading to hardness increase.

2.1.4. Hardening via ion bombardmentThe energetic bombardment in magnetron sputtering comes from

the energy of the sputtered atoms, the flux of energetic neutrals andflux of gas ions from the plasma impinging on the biased substrate[46,47]. Studies have demonstrated that ion bombardment during de-position at low temperature can be used to increase coating density[48], and modify coating morphology [49,50] in addition to improvingthe adhesion between coating and substrate [42]. Normally, increasingion bombardment produces re-sputtering on the film surface andnudges the sputtered atoms to fill the “valley” and “hump” that natural-ly occur from the shadowing effects, and thus increases film density.Also, bombarding ions can retard the grain growth, increase the nucle-ation sites and allow the formation of nanocrystalline microstructure.This process has been interpreted in terms of the Thorton–Messierstructure-zone diagram [51,52]. In parallel withmicrostructural change,biaxial compressive stress is built up and is mainly attributed to the lat-tice defect arrangements induced by energetic ions [53,54]. During de-position, incoming ions with enough kinetic energy will knock atomsin thefilms out of their original positions if the recoil energy transfer ex-ceeds about 25 eV, creating a sequence of intensive collisions (i.e. colli-sion cascades). Usually, the initial collision leads to the most energytransfer and exchange among the atoms. And the resultant atomic mo-tion along the trajectory contributes to the eventual rearrangement ofthe lattice atoms [55].

The development of hard coatings utilizing a synergetic effect of thedecrease of grain size, densification of grain boundaries and compres-sive stress upon ion bombardment is a subject of intensive research.Common methods to study the effect of ion energy and ion flux bom-bardment are manipulating plasma [46,49] or applying a negative sub-strate bias [56,57]. Lin et al. demonstrated the influence of ion energieson the microstructure and the mechanical properties of CrAlN coatingsprepared via pulsed closed field unbalanced magnetron sputtering. Ata lower ion energy of 72 eV, the coating favored in (200) for minimiza-tion of the surface energy. With increased ion energy and ion flux, thepreferential orientation turned into (111) in order to lower the strainenergy. A super high hardness of 48 GPa was achieved when the totalion energies were about 177–200 eV. The increased hardness was at-tributed to the high residual stress and lattice distortion, which restrict-ed the plastic deformation and dislocation movement [46]. Wang et al.[58] prepared the CrAlN coating using magnetron sputtering at variousnegative bias voltages. It was found that with increased bias voltage(from 50 to 260 V), columnar grains were transformed to glassy densestructure; thereby hardness was increased monotonically up to26GPa. Similar results can be also found in preparing other nitride coat-ings, such as TiN [59]. However, the resultant hardness improvementmay be annealed out upon heat treatment at a certain temperaturedue to the stress relaxation [60]. On the other hand, too much stresswill cause coating delamination when coating thickness increases.

2.1.5. Hardening via multilayeringMultilayer and/or supper-lattice coatings are materials prepared on

atomic scales, with structures consisted of layers in thickness of tensor only a few nanometers. In 1970, Koehler demonstrated that byemploying alternate ultra-thin layers of materials with higher andlower elastic constants, materials with superior mechanical propertiescan be achieved [61]. In 1987, Helmersson et al. showed that single-crystal transition metal nitride multilayers presented 2–3 times thehardness of single layered nitrides. A maximum hardness value of56 GPa was obtained for the TiN/VN multilayer with a bilayer thicknessof 5.2 nm [62]. The choice of component materials for the constituentlayer and the thickness of the individual layers are essential in obtainingthe desired properties. Combinations ofmetal/metal, metal/ceramic andceramic/ceramic have been investigated by many researchers. The in-creased interest in the ceramic/ceramic multilayer coatings is drivenby their great potential applications in diverse fields. Nowadays, themost technologically important ceramic/ceramic nitride multilayersystems are: TiN/NbN [63–65], CrN/NbN [66], TiN/VN [62,67], TiN/CrN[68,69], TiN/AlN [70], TiN/TiAlN [71], CrN/TiAlN [72–74], CrN/CrAlN[75], TiAlN/CrAlN [76], CrAlN/VN [77], multilayering DLC [78,79], andTiC/DLC [80], to name but a few.

The improved hardness is attributed to several reinforcementmech-anisms, including dislocation blocking by layer interfaces, Hall–Petch ef-fect, dislocation motion in an alternating strain field and supermoduluseffect. Among them, dislocation blocking by layer interfaces is the moststudied and considered effective in predicting hardness improvement.Koehler explained the dislocations moving across layers by introducingimage force [61]. Theoretically, the dislocation motion would beinhibited in a multilayered structure due to the image force on disloca-tions created by the different dislocation line energies in each layer. Themodel assumed that themultilayers consisted of layers with two differ-ent shear moduli, a lower one (layer A) and a higher one (layer B).Under this circumstance, a shear stress of the order of GA/100 (G standsfor the shear modulus) would be required to drive dislocations throughthemultilayers given that the individual layer is thin enough. Assumingthat the dislocation glide plane and the interface were orthogonal,Pacheco et al. derived the shear stress on the dislocation on the basisof a Peierls dislocation model [81]. The maximum shear stress actingon a screw dislocation occurs when it is located at the interface and isgiven by

τmax ¼ sin θ GB−GAð Þ=π

where τ is the shear stress, and θ is the angle of dislocation slip planewith respect to the interface.

In this model, the dislocationwould be constrained in layer A unlessa shear stress of τA + τmax was applied to the system, where τA is theshear stress required to move the dislocation in material A. With thepredictable value of shear stress, the yield stress can be obtained usingSchmid's law, as given by

σ ¼ τ=m

whereσ is the yield stress, τ is the shear stress andm is the Taylor factor.For transition metal nitrides in the NaCl-type structure, m ≈ 0.3 [61].The estimated hardness from the yield stress can be then obtained using

H≈ 3σ:

Therefore, the maximum expected hardness of a multilayer due toimage effects in the system is

Hmax ¼ HA þ 3 sinθ GB−GAð Þ=mπ2

where, HA is the hardness of layer A, θ is the angle of dislocation slipplane with respect to the interface, G is the shear modulus and m isthe Taylor factor.

Fig. 5. (a) Hardness of monolayer and multilayer DLC coatings as a function of substratebias and substrate bias ratio. (b) Residual stress inmonolayer andmultilayer DLC coatingsas a function of substrate bias and substrate bias ratio [78].

5Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

This simple model gives a good approximation for polycrystallinematerials and can be further improved by taking into considerationmany other factors, in terms of layer thickness, layer numbers and inter-face diffusion. Koehler mentioned that the layer thickness must be thinenough so that dislocation generation cannot occur within the layers.While if the individual layer thickness is large, dislocations will be con-finedwithin the layer without encountering an interface, thus no signif-icant image force effect would occur. With this consideration, Chu andBarnett proposed a model on the basis of Kzanowski's assumption forimage force effect [82]. With this model, they managed to predict thetendency of hardness versus individual layer thickness, so as to findout the optimal layer thickness to inhibit dislocation motion across theboundaries between the layerswithout havingdislocationmotionwith-in a layer. To highlight, this model matched well with the experimentaldata of TiN/NbN multilayers [82].

For multilayers, another significant factor that should be consideredis the residual stress, since large accumulated stress may lead to layerexfoliation, detrimental to the coating system. To solve this problem, at-tempts have been made on optimizing the multilayer structure. Li et al.prepared a multilayered diamond like carbon (DLC) with alternate softand hard carbon layers by varying bias voltage during magnetronsputtering [78]. The hardness of the sub-layer layers was adjusted as afunction of the sp3 site fraction. Increased bias voltage resulted in in-creased sp3 site fraction thus increased hardness. By optimizing thesub-layer thickness and total thickness, the hardness of multilayer DLCwith soft/hard sub-layers was improved from 18.9 GPa to 21 GPawhile the residual stress decreased from 6.0 GPa to 2.8 GPa. Fig. 5 dem-onstrates the comparison of hardness and residual stress betweenmonolayer and multilayer DLC coatings fabricated using various biasvoltage combinations.

2.2. Ways toward toughening of ceramic coatings

Toughness is the measurement of the resistance to crack propaga-tion, indicative of the amount of energy needed to break the material.In principle, materials would be tougher if more energy has to be con-sumed during crack propagation. Design of tough ceramic coatingstherefore is realized via (i) introducing a toughening agent, includingmetal ductile phase and carbon nanotube additive, (ii) utilizing phasetransformation, (iii) inducing compressive stress and (iv) optimizingthe coating architecture. Herein, it should be highlighted that in thethin film community, there is not yet a universally accepted toughnessmeasurement methodology [2,3]. Current measurement methods aredivided into two groups, i.e. qualitative and quantitative methods [6].Qualitatively, toughness can be estimated by the plasticity, micro-hard-ness dissipation parameter and scratch crack propagation resistance.Quantitatively, toughness can be measured via bending, buckling, in-dentation and tensile elongation [5]. Owing to the different measure-ment methods used, the following discussion on toughening effectwould be also different in terminology.

2.2.1. Toughening through introducing a toughening agent

2.2.1.1. Ductile phase toughening. Incorporating a ductile phase isthe most straightforward route in toughening hard ceramic coatings.The toughness increment comes from extra work consumed duringplastic deformation, in two primary mechanisms (Fig. 6): 1) the relaxa-tion of the strain field around the crack tip through the ductile phaseand 2) the yielding and bridging of cracks by ligaments of the ductilephase. More often, metals are the promising candidates as ductilephases due to their good ductility.

In ceramic coatings, the ductile phasemay serve asmatrix (i.e. metalmatrix composite) or dopant (i.e. ceramic matrix composite). In studiesof metal matrix composites, less attention on toughening effect hasbeen given but more are focused on microstructure optimization to ob-tain the maximum hardness, as reported in ZrTiN/Cu [83], ZrN/Cu

[84,85], TiN/Cu [86], CrN/Cu [87], TiN/Ag, CrN/Ag, ZrN/Ag [88,89],Mo2N/Ag [90], ZrN/Ni [91] and etc. For the ceramic matrix composite,toughening effect via a metallic dopant is much reported. In the caseof TaN/Cu deposited tool steel [92], the crack resistancewas determinedby the content of Cu diffused into the TaN matrix via the rapid thermalannealing process. The coating with 1.4 at.% Cu exhibited the leastamount of cracks after indentation test, which was attributed to theCu network. Similar results can be found in Cu into ZrOx [93]. The en-hanced toughness was attributed to the amorphous structure and com-pressive macrostress introduced.

As mentioned, nc-MenN/a-SiNx is a significant hard/super-hardnanocomposite coating and has been widely used in industrial applica-tions [1]. Recently, the intrinsic lower toughness of this nanocompositecoating has been increasingly recognized. To toughen the nc-TiN/a-SiNx

nanocomposite, Ni was introduced via co-sputtering of Ti, TiNi andSi3N4 targets in an Ar/N2 atmosphere [94,95]. When Ni ranged from0 to ~40 at.%, the toughness increased from 1.15 to 2.6 MPa m1/2.Wang et al. did a similar work but investigated the toughening effectof Ni in a more complicated nanocomposite, nc-CrAlN/a-SiNx [96].Fig. 7 illustrates the bright field high resolution transmission electronmicroscope (HRTEM) image of the nc-CrAlN/a-SiNx with 12 at.% Ni. It

Fig. 6. Schematic diagram of ductile phase toughening through (1) ductile phase deforma-tion or crack blunting, and (2) crack bridging [7].

6 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

wasobserved that thenanocrystallites (ca. 3 nm)were embedded in thea-SiNx and Ni network, which limited the grain growth. An analysis ofthe selected area diffraction (SAD) pattern indicated that thesenanocrystallites were polycrystalline CrAlN in face centered cubic struc-ture. The measured hardness and scratch toughness (Fig. 8) suggestedthat doping 4.3 at.% Ni led to the best balance in strength and crack re-sistance. The increase of toughness is from 1) relaxation of the strainfield around the crack tip through the ductile phase (metallic Ni) defor-mation or crack blunting, whereby the work for plastic deformation isincreased and 2) a network of amorphous Ni atoms together with a-SiNx surrounding the CrAlN nanocrystallites. Pei et al. once proposed atoughening theory based on crack localization in an amorphous matrix.Under this circumstance, crack nucleationmay be suppressed by the in-troduction of nanoparticles [97].

Diamond-like Carbon (DLC) is another extensively used coating dueto its high hardness, low friction coefficient, superior wear resistanceand chemical inertness [15,98–101]. However, a large compressivestress from the ion bombardment during deposition and an intrinsicbrittle feature make DLC less durable when exposed to sudden impact.To overcome these problems, Tay et al. doped Al in amorphous carbon(a-C) to decrease the accumulated stress and at the same time improvethe toughness [102]. In spite of the amelioration of the residual stressand toughness, an inevitable significant drop of hardness was observed.With the increase of Al from 1 to 10 at.%, the hardness decreased from27 to 18 GPa accordingly. Bui et al. utilized a similar methodology andperformed a systematical investigation on the microstructure, residualstress, hardness and scratch toughness of a-C/Al [103]. Results from

Fig. 7. HRTEM image (a) and corresponding selected area diffrac

XRD and TEM indicated that Al is in X-ray amorphous state.With differ-ent contents of Al, mechanical properties of a-C/Al were varied as ex-pected. Without Al, the hardness of pure DLC was 32.5 GPa with ahigh residual stress of 4.1 GPa.When Al content reached 19 at.%, the re-sidual stress declined largely to 0.2 GPa while the hardness was only7.8 GPa. In parallel, the addition of Al significantly improved the coatingtoughness and adhesion. Without Al, a lower critical scratch force atonly 118 mN can lead to a brittle delamination with severe chippingand buckling. In contrast, with 19 at.% Al, the lower critical scratchforce can reach 180 mN without rendering brittle cracking. No coatingfracture or interfacial failure was observed even when the scratch loadis up to 455 mN, suggesting that the coating is of enhanced toughness.

Doping ductile phase is an effective way of toughening, but it coststoo much hardness. To achieve a hard-yet-tough coating, hardnessmust be restored, which is themotivation of doping another reinforcingphase. Following this concept, nc-TiC nanoparticles were incorporatedinto a-C/Al to form nc-TiC/a-C(Al) nanocomposite coatings [103,104].The randomly oriented TiC nanograins (~5 nm) surrounded by a-C/Alphase was achieved. With the addition of nc-TiC, the hardness was re-covered at a large magnitude. Fig. 9 illustrates the difference of scratchtests in comparing a-C, bias-graded a-C, nc-TiC/a-C and nc-TiC/a-C(Al).Among these coatings, nc-TiC/a-C(Al) presents a hard-yet-tough fea-ture. To qualitatively determine the toughness, a related parameter-plasticity was used to show the benchmark and a scratch test was car-ried out to verify. Plasticity here is defined as the ratio of plastic strainover the total strain. Considering that plastic deformation is the majorsource of stress relaxation, a larger plasticity should be associated with(but not equal to) higher toughness inherent in the material. In suchnc-TiC/a-C(Al) nanocomposites, the plasticity was as high as 55%. As acomparison, the super-hard nc-TiN/a-Si3N4 shows almost no plasticityand the nc-TiC/a-C coating of 30 GPa hardness had a plasticity of ~40%.

2.2.1.2. Carbon nanotube toughening. Since discovered in 1991 [105,106],carbon nanotubes (CNTs) have been considered versatile due to theirunique features in mechanical, electronic and chemical aspects. Carbonnanotubes compose one (i.e. single-walled nanotube) or several (i.e.multi-walled nanotube) graphite basal planes rolling into a cylinder[107]. Rolled graphite basal planes with co-axial structure make CNTsone of the stiffest and most robust materials, since the constituentcarbon–carbon bond on the basal plane is predicted as one of thestrongest in nature. Single-walled nanotubes (SWNTs) present tensilestrength 100 times that of steel with weight of only 1/6. In particular,SWNTs have shown tensile strength greater than 65 GPa (predictedvalue up to 200 GPa) and modulus of elasticity in the magnitude of1000 GPa, allowing them to withstand 10%–30% elongation beforebrittle breakage, plastic deformation or bond rupture [108]. TheYoung's modulus of multi-walled nanotube has been measured up to

tion patterns (b) of nc-CrAlN/a-SiNx with 12.0 at.% Ni [96].

Fig. 8. Plot of hardness versus scratch toughness and fracture behavior after scratch of nc-CrAlN/a-SiNx with varied Ni contents [96].

7Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

~1800GPa, which ismuchhigher than that of their predecessors, name-ly carbon fiber (~680 GPa) and carbon whisker (~800 GPa). Reversibil-ity of deformation in multi-walled nanotube is also good and has beenobserved with the aid of TEM [109]. Similar results regarding the out-standingmechanical properties of CNTs have been reported by differentresearch groups [110,111], indicatingCNTswould be the ideal candidatefor the toughening agent.

To date, not many studies have been carried out on CNTs toughenedceramic coatings. The difficulty, primarily, is the uniform dispersion ofnanotubes into the ceramic matrix. In-situ growth of CNTs is one wayout as reported in the case of CNTs/Al2O3 composite with thickness of400 μm [112]. With the aid of CVD and plasma spray, well-dispersed

Fig. 9. Scratch tracks on (a) a-C H = 28.1 GPa, (b) bias-graded a-C H = 25.1 GP

CNT was obtained. Fig. 10(a) demonstrates the interface between CNTand Al2O3, where a 0.5 nm thick amorphous interlayer can be identified.It was believed that such interlayer acts as a buffer layer to accommo-date imperfect surface termination. And because of the tube anchoring(Fig. 10(b)) in the Al2O3 phase, the composite presented balance be-tween hardness (~9 GPa) and toughness (~4.62 MPa m1/2).

However in most CNT embedded composites, nanotube tangling isthemajor cause of declined toughening effect. To optimize the strength-ening effect of CNTs, Xia et al. prepared Al2O3 coatings (20–90 μm)withhighly oriented CNTs [113]. The CNTs grew in-situ (CVD assistant) fromthe anodized aluminum template with a pore diameter of 30–40 nmand Co or Ni catalyst particles at the bottom. Microscopic analyses

a, (c) nc-TiC/a-C H = 27.4 GPa, and (d) nc-TiC/a-C(Al) H = 19.6 GPa [103].

Fig. 10. (a) HR-TEM image showing 0.5 nm amorphous interface and embedded Al2O3 crystallite in CNT with lattice spacing of 2.49 Å and 1.70 Å; (b) CNT anchored with Al2O3 crystallite[112].

8 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

confirmed the formation of MWNTs inside the pore walls, as seen fromFig. 11. And the nanotube toughening behaviorswere observed in termsof crack deflection at the tube/matrix interface, crack bridging from tubeand tube pull-out from the image shown in Fig. 12.

In favor of uniformly dispersing CNTs in a ceramicmatrix, a two-stepfabrication process aiming at in-situ growth of CNTs was proposed.Firstly, carbon source (i.e. graphite), transition metal and catalyst ele-ment (e.g. Ni) were co-sputtered. Afterwards, a rapid-thermal anneal-ing was applied to as-deposited coatings. Depending on whether H2

was introduced or not, partial carbon atoms would be formed in carbonnanotubes or carbon nanofibers [114]. However to date, no results areavailable to demonstrate the toughening effect, but some of the recentworks in understanding the growth mechanisms under this circum-stance to optimize the growth condition have been published [115].

Fig. 11. Top view of as-fabricated highly orientated CNT/Al2O3 composites [113].

2.2.2. Phase transformation tougheningPhase transformation process readily toughens brittle ceramics,

since the transformation consumes a large amount of energy due tothe dimensional variation in crystalline structure. In previous studies[116,117], transformation occurred under stress, leading to crystalliteschanging from tetragonal phase to monoclinic phase. This change wasaccompanied by a 4% volume expansion,which releases the stress accu-mulated and consumes the fracture energy from the resultant strain.ZrO2 is one typical material with such characteristics and ZrO2 tough-ened ZrB2 is one of the examples [116]. In this case, retention of thehigh-temperature tetragonal phase ZrO2 is essential, since it has a hightoughness against the brittle fracture behavior [117]. Stress-inducedphase transformation was observed by other researchers [118]. Zirconi-um oxide thin film was deposited using radiofrequency magnetronsputtering. The as-deposited thin films contained a mixture of tetrago-nal and monoclinic zirconia phases. With increased bias voltage duringsputtering, an increase in lateral defects was observed, which was be-lieved to be the cause of phase transformation [119]. However, withouta proper method of toughness evaluation, the toughness was not mea-sured. But recently, partially stabilized zirconiawas applied to dental ce-ramic materials. The toughness of zirconia thin film treated substratewas increased by 55% as compared to that of the non-treated one [120].

2.2.3. Compressive stress tougheningOpen surfaces of cracks are generally initiated by the tensile stress.

Therefore, the toughnesswill be likely increased if compressive residualstress is properly induced in the coatings, in which extra energy isrequired to overcome that stress before a crack is driven in tension.When a coating has a high compressive residual stress to start with,the coating is able to take more tensile strain before a fracture. In prac-tice, there are many ways to introduce stress into the coatings. For in-stance, Halitim et al. did it by ion implantation [121] while Abe et al.by phase change [122,123]. In the case of NiAl/Al2O3, the formation ofNiAl2O4 was the source of stress as a result of volume change. The in-duced 127MPa compressive stress led to the large increase of toughnessfrom 3.91 to 6.22 MPa m1/2.

Fig. 12. (a) Crack deflection at CNT/Al2O3 interface; (b) CNT bridging longitudinal Al2O3 matrix and tube pullout [113].

9Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

Compressive stress induced high toughness can be also found in coat-ing systems such as TiN, DLC,MoS2, TiC and CrAlN [58,59,124,125].Wanget al. measured the toughness of TiC with different compressive stressesoriginating from the bias voltage during sputtering. The increased biasvoltage led to the increase of stress as well as the restrained columnarstructure, and thus the increased toughness from 1.19 to 1.89 MPa m1/2

[59]. However, too much compression is detrimental, resulting in thepoor adhesion, delamination and micro-cracks in the as-deposited coat-ings and in turn renders decrease of toughness [126]. Similar resultswere reported in CrAlN coating system [58]. Themeasured toughness in-creased from 1.67 to 2.02MPam1/2 when bias voltage increased from 50to 210V (Fig. 13(a)), but further increasing bias voltage led to the drop of

Fig. 13. (a) Fracture toughness of CrAlN prepared at different bias voltages via sputtering [58];load, white arrow higher critical load) (unpublished data based on [58]).

toughness and adhesion. Fig. 13(b) illustrates the fracture behavior ofsuch CrAlN coatings prepared at varied bias voltages when they are sub-jective to the scratch test. A poor adhesion along with severe chippingand buckling was observed when the bias voltage was at 260 V.

2.2.4. Toughening through optimization of coating architecture

2.2.4.1. Gradient structure toughening. In practical applications, require-ments on a coating can vary throughout the coating thickness (fromthe coating/substrate interface all the way up to the top of the coatingsurface). For example, a good adhesion is required at the coating–substrate interface, while at the coating surface a very stable material

(b) scratch tracks of CrAlN prepared at different bias voltages (black arrow: lower critical

Fig. 14. Cross-sectional image of heterogeneously gradient CrASiN nanocomposites, chemical compositions are listed accordingly in the unit of at.%.

10 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

is needed (against penetration, scratching, corrosion and etc. [127]).However, in a homogeneous single layer coating, these desired proper-ties are rarely achieved simultaneously. Therefore, it is apt to applygraded coatings, consisting of a sequence of coatingmaterials, to satisfydifferent requirements at different coating thicknesses.

In the gradient design, the substrate is firstly coated with a high ad-hesion layer and then the coating constituents are allowed to vary ho-mogeneously or heterogeneously while the coating thickness buildsup. The gradually changed coating constituents can be achieved by vary-ing deposition conditions. Wang et al. adjusted the target power duringsputtering and fabricated CrAlSiN nanocomposite with a heteroge-neously gradient structure (Fig. 14). An obvious structure change canbe identified from the bottom to the top. The toughness was increasedby 300% as compared to the monolayer CrAlSiN while the hardnesscan be maintained at around 25 GPa.

By gradually changing the bias-voltage, Zhang et al. [128] prepared a1.5 μm thick a-C gradient coating on the tool steel withmoderately highhardness (25 GPa) and high toughness (plasticity ratio approximately58%). As the negative bias voltage varied from 20 to 150 V, the sp3/sp2

ratio of the carbon gradually increased from the bottom layer to the

Fig. 15. (a) Hardness of graded DLC coatings; (b) scratch adhesion strength of the non-hydroge

top surface. The lower sp3/sp2 ratio ensured good adhesion at the coat-ing–substrate interface, while the higher sp3/sp2 ratio gave rise to highhardness, as well as the coating of improved tribological performance.The graded coatings became hard yet tough (Fig. 15). In other gradientcoating systems, such as commercial grading TiCN [129,130], TiN-TiCN-TiC-DLC [79] and Ti-TiC-DLC [80], high hardness with enhanced adhe-sion was also achieved. Gradient structure is beneficial for the suppres-sion of crack initiation due to the absence of sharp interface.

2.2.4.2. Multilayering structure toughening. Multilayering structure isconsidered another effectiveway to improve toughness of ceramic coat-ings. Themain mechanisms include crack deflection at interface amongthe sub-layers, interface delamination, ductile interlayer ligamentbridging and crack tip blunting because of nanoplasticity, which willcause extra energy consumption and dissipation during crack propaga-tion [131] (Fig. 16).

These toughening mechanisms have been validated by a number ofexperimental observations. Suresha et al. [132] demonstrated the crack-ing patterns in monolayer TiN and multilayered TiN/AlTiN at an inden-tation load of 20N. Severe edge crackingwas observed in themonolayer

nated DLCs deposited under various constant bias voltages or bias-graded conditions [128].

Fig. 16.Multilayering structure toughening mechanisms.Redrawn from [131].

11Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

while much less cracking was found in the multilayers (Fig. 17). In par-ticular, TiN/AlTiN with a bilayer thickness of 650 nm showed two edgecracks while that with a bilayer thickness of 130 nm showed only one.Fig. 18 illustrates the cross-sectional images of the fracture regionunder the indentation with a normal load of 5 N. Continuous shearcracks were evident in the monolayer TiN. Multilayers of bilayer thick-ness of 650 nm displayed discontinuous shear cracks, while such crackswere absent in themultilayers of bilayer thickness of 130 nm. These re-sults indicated the positive effects of the layered structure in preventingcrack propagation.

For the same multilayer system, Lee et al. [133] investigated thetoughening effect using scratch test in addition to nanoindentation.Multilayers presented great resistance to crack initiation and propaga-tion, in great contrast with the monolayer, which exhibited extreme

Fig. 17. Surface view of the indentation with a load of 20 N showing dependence of edge cracki650 nm and (c) multilayer TiN/TiAlN with bilayer thickness of 130 nm [132].

Fig. 18.Cross-sectional images of the indentation under 5N showing inclined shear cracks: (a)mTiN/TiAlN with bilayer thickness of 130 nm [132].

brittleness accompanied by severe coating chipping off at the fringe ofthe groove track (Fig. 19). Nanoindentation tests further confirmedthe advantage of themultilayers in fracture resistance. Only a few cracksspread out radially within the multilayers, whereas delamination wasobvious in the monolayer, together with radial cracks.

Karimi et al. [134] fabricated a variety of nanostructured TiAlN(Si, C).A summary of the results for several samples is given in Fig. 20, inwhich“M” stands for multilayers. It was shown that multilayers (4) exhibitedthe best performances, both in hardness and toughness. A similar resultwas obtained in the case of CrN/AlN multilayer, subjected to the Rock-well C-Brale indentation test [135]. Practically, the number of the inter-layers, the thickness of each sub-layer, and the thickness ratio ofdifferent sub-layers all affect the toughness. The dependence of thetoughness on these factors is evaluated in coating systems such asTiC/TiB2 [136], TiC/TiB2 [137], TiC/CrC [138], CrN/AlN [139] and TiN/Ti[140]. In general, increased number of the interface leads to improvedtoughness. Besides the dimensional effect of the sub-layers, studieshave been carried out on the crystalline structure of the constituentlayers. Yau et al. [141] investigated the multilayers with the bilayers ofnanocrystalline/amorphous phases, such as nc-TiAlN/a-Si3N4. Maxi-mum critical scratching load of the multilayers was increased by 165%as compared to monolayer TiAlN. The improvement was attributed tothe crack deflection at the layer interfaces (Fig. 21). In the more recentresearches of TiN/SiNx [142,143] and CrAlN/SiNx multilayers [144,145],the crystallite structure of the SiNx layers were found to be dependenton their thickness. At very small thickness, for example a few tens ofAngstrom, the SiNx was in an epitaxial crystalline structure. However,when the thickness was larger than a certain threshold, the epitaxialcrystalline structure could transform into an amorphous phase. Thehardness was varied accordingly due to the phase change of the SiNx.Although the tendency of the toughness is not reported, it is believedthat the epitaxial crystalline structure will influence the ability of thecoating in preventing crack growth and propagation.

By investigating the hardness and toughness of homogeneouslygradient CrAlSiN nanocomposite [146] and polycrystalline CrAlN

ng on layer spacing: (a) monolayer TiN, (b) multilayer TiN/TiAlN with bilayer thickness of

onolayer TiN, (b)multilayer TiN/TiAlNwith bilayer thickness of 650nmand (c)multilayer

Fig. 19. Scratch channels ((a) and (b)) and indent marks ((c) and (d)) of monolayer (Ti0.5Al0.5)N and TiN/(Ti0.5Al0.5)N multilayer coatings using [133].

12 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

(pc-CrAlN) [58], Wang et al. integrated them into a multilayer structureconsisting alternative layers of nanocomposite CrAlSiN and pc-CrAlN(Fig. 22). The interfaces shown by thewhite arrows in Fig. 22(a) are en-larged in Fig. 22(b). The continuous growth of the columnar pc-CrAlNgrainswas periodically interrupted due to the insertion of the nanocom-posite layer, as also observed in Ref. [147]. No epitaxial relation wasfound to take place within the adjacent layers. However, the interfacewas not sharp. At the interface, pc-CrAlN layerwith large grains (around10 nm)was observed. The pc-CrAlN layer combinedwith the nanocom-posite layer (with nanocrystallites around 5 nm as seen in Fig. 22(c)),leading to complex interface boundaries (Fig. 22(d)). In the nanocom-posite layer, only a few nano-crystallites, whose lattice planes were ori-ented parallel to the electron beam, appeared in the lattice resolutiondue to random orientation.With such structure, themultilayer coatings

Fig. 20. Variation of hardness, critical load for cracking and fracture toughness for a num-ber of coatings: 1. TiN; 2. TiCN, 3. TiAlN, 4. TiAlN(M), 5. TiAlSiN(M), 6. TiAlCN [134].

(the open circles) do appear toward the hard yet tough direction invarying extents at varying period thicknesses (Λ) [147]: Λ = 10 nm,the hardness of themultilayer coatingmatches that of the nanocompos-ite monolayer, meanwhile, its toughness is better (moving toward rightin the hardness vs toughness space). At even larger period thickness,Λ = 60 nm, the multilayer's hardness drops with increased toughness;at Λ = 40 nm, hardness returns with much improved toughness(7 times as much as that of the nanocomposite monolayer); at Λ =20 nm, the hardness further improves, and even surpasses that of thenanocomposite (from less than 30 to 33 GPa) at a slight drop in tough-ness from that at Λ = 40 nm. With this architecture (Λ = 20 nm), themultilayer coating's toughness increased 5 fold from that of the hardnanocomposite (the open triangle, Fig. 23).

Fig. 21. Cross-section TEM micrograph showed the propagation of nanocracks innc-TiAlN/a-Si3N4 nanolaminate coatings with Λ = 20 nm [141].

Fig. 22. (a) HRTEM image of the multilayer coating with period thickness of 20 nm, (b) an enlarged interface at the white arrows in (a), (c) HRTEM image of the nanocomposite(i.e. nc-CrAlN in a-SiNx) sub-layer and (d) the schematic of the sandwich structure.

13Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

3. Toughness evaluation

In studies of toughening of thin films and coatings, it is inevitablethat one needs to evaluate toughness. However, a universally acceptedmethod is not yet established for thin films or coatings. Toughnessmea-surement of bulkmaterials has been achieved by the generous use of in-dentation method [148], but obtaining good toughness estimation ofthin films remains challenging [149]. By adopting conventional indenta-tion, researchers found that the accuracy of toughness measurement forthin films is influenced by factors in terms of the substrate, the interfaceand the precise measurement of crack geometries. To provide a quick

Fig. 23. Hardness versus scratch toughness of multilayer and monolayer CrAlSiN [147].

benchmarking of toughness enhancement endeavor, Zhang et al. pro-posed a facile way of using scratch test [149] to examine the “scratchpropagation resistance”. Scratch test was first developed to evaluatethe strength of the adhesive bonds between the coating and the sub-strate. In scratch test, a diamond stylus subjected to a linearly increasingload is drawn across the coating surface. When coatings break, parts ofthem would be detached (i.e. either delamination or spallation) andthat will be registered as a disturbance of the friction coefficient, thusis discernible. The minimum load at which the onset of crack occurs isdefined as the “lower critical load” (Lc1). The load associated to the com-plete delamination is termed as “higher critical load” (Lc2), as illustratedin Fig. 24.With these two parameters, a “scratch propagation resistance”(CPRs),which is equal to Lc1(Lc2− Lc1), is used to semi-quantitatively as-sess the coating toughness. More recently, nano-indentation has proved

Fig. 24. A typical scratch test profile [149].

Fig. 25. Schematic of the testing configuration: a rectangular silicon wafer substrate con-taining an edge crack and two pin-holes; a series of film strips just ahead of the tip of sub-strate crack [150].

14 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

to be an effectiveway out to assess toughness of thinfilms. An extremelylow load (in a scale of mN or less) is applied via indenter to generateshallow indent marks on thin films. The toughness value is obtainedvia post-mortem analysis of a load–unload curve and crack patternsupon stress-based or energy-based models. A more detailed discussioncan be found in a review paper [2] in great length. However, themethods mentioned are carried out on a coating–substrate systemthus the influence from the interface and/or substrate should be careful-ly considered and calculated to make accurate estimation of coatingtoughness.

To solve this problem, the direct measurement of free-standing thinfilms seems to be the logical way [5], but difficult to realize because ofthe minute load needed and the difficulty in clamping the free-standing testing film. One recently proposed method or the micro-tensile testing of film micro-bridges via macro-tension to a substrate[150] cleverly avoided both difficulties while realizing tensile testingof “free-standing” films. In this method initial edge crack was intro-duced to a rectangular Si substrate using a diamond cutter. Before pat-terned film strips were made via physical vapor deposition withpatterning mask on ZnO sacrificial layers. A pre-crack was introducedto each and every film strip using Vickers indentation at the adjacentsubstrate. When the underneath ZnO sacrificial layer is etched awaythe film strips become “micro bridges” anchored to the Si substrate.The initial edge crack was driven forward by pulling away the twopin-holes via a manually controlled micro-apparatus. As edge crackpropagated underneath the film strips, the strips were subjected tothe tensile stress when the edge crack opening became larger(Fig. 25). The stress intensity factor KIC at the tip of the pre-crack wasobtained via the expression

KIC ¼ σ fπaF a=Wð Þ

where F(a /W) is a polynomial functionwith a boundary that a /W≤ 0.6,wherein, W is the width of the film bridge and a is the length of thepre-crack.

4. Summary and outlook

This review provides a comprehensive discussion on the hardeningand toughening mechanisms of ceramic coatings. Effective hardeningstarts from the design of the nanostructured coating. The decrease of

the crystallite size (at ca. 10 nm) and increase of the grain boundarieshinder dislocation movements, and therefore provide Hall–Petchstrengthening. A further strength enhancement is achievable byintroducing a strong interfacial layer with crystallite size below 10 nm(i.e. nanocomposite hardening). Multilayering renders hardening byblocking dislocation movement between the layers. With manipulationof the layer thickness, total number of layers and interfaces hardeningeffect can be optimized. The toughening follows the similar way vianano-scale engineering. Individual toughening technique in terms of in-troducing a toughening agent (e.g. metallic additives and carbon nano-tubes) and optimizing coating structure (e.g. gradient structure andmultilayering) have shown great potential in enhancing toughness.The key of toughening lies in the creation of complex coherent bound-aries at interfaces, by whichmore energy is consumed during crack ini-tiation and propagation. This principle can be applied to any coatingsystem. The challenge now is how to properly integrate these hardeningand toughening mechanisms in one, so as to make hard-yet-tough ce-ramic coatings practical. Nano-multilayers containing nanocompositelayers have switched on a beacon light in this direction.

It is exciting to note that over the past decades, the research for ce-ramic coatings with combination of hardness and toughness keeps itsmomentum. The future of hardening and tougheningwould be, withoutdoubt, based on further understanding of materials in nano-scale. Withthis view, we'd also like to stress that:

• A further understanding of the coating deformation in nanoscale isdesired. Although theoretical design of hard-yet-tough coatingshas been forged toward the scale of nanosize, to date, not many stud-ies in such scale are available regarding the characterization of dislo-cation formation and crack evolution. One potential solution isintroducing the micro-electro mechanical systems into the in-situmeasurement of coating mechanical properties.

• High temperature application should be taken into considerationwhen designing hard-yet-tough coating systems. To date, very fewdiscussions of toughening at elevated temperatures (N1500 °C oreven ~2000 °C) are documented. Moving beyond the temperaturebarrier would be a significant breakthrough for practical industrialapplications.

• A universal methodology is badly needed for thin film and coatingtoughnessmeasurement. There are two aspects in this regard: tough-ness measurement of a free-standing film and that of a coating on asubstrate. Understanding the energy dissipation during crack propa-gation and the interfacial effect between coating and substrate is in-dispensable in the establishment of the measurement methodology.

References

[1] S. Veprek, M.J.G. Veprek-Heijman, Surf. Coat. Technol. 202 (2008) 5063–5073.[2] J. Chen, J. Phys. D. Appl. Phys. 45 (2012).[3] J. Chen, S.J. Bull, Thin Solid Films 517 (2009) 2945–2952.[4] J. Musil, Surf. Coat. Technol. 207 (2012) 50–65.[5] S. Zhang, X. Zhang, Thin Solid Films 520 (2012) 2375–2389.[6] S. Zhang, H.L. Wang, S.E. Ong, D. Sun, X.L. Bui, Plasma Process. Polym. 4 (2007)

219–228.[7] S. Zhang, D. Sun, Y. Fu, H. Du, Surf. Coat. Technol. 198 (2005) 2–8.[8] D.S. Harding, W.C. Oliver, G.M. Pharr, Cracking during nanoindentation and its use

in the measurement of fracture toughness, MRS Bulletin 356 (1995) 663–668.[9] A.C. Fischer-Cripps, Nanoindentation, Springer, New York, 2004.

[10] American Society for Testing and Materials, Philadelphia, PA, 1997.[11] G. Cherepanov, Mechanics of Brittle Fracture, McGraw-Hill International Book Co.,

New York, 1979.[12] X.K. Zhu, J.A. Joyce, Eng. Fract. Mech. 85 (2012) 1–46.[13] J. Schiøtz, F.D. Di Tolla, K.W. Jacobsen, Nature 391 (1998) 561–563.[14] J.M. Cairney, M.J. Hoffman, P.R. Munroe, P.J. Martin, A. Bendavid, Thin Solid Films

479 (2005) 193–200.[15] J. Robertson, Mater. Sci. Eng. R 37 (2002).[16] J.-M. Ting, W.-Y. Wu, S.P. Sharma, J.C.-M. Sung, M.-C. Kan, in: S. Zhang (Ed.), Nano-

structured Thin Films and Coatings Mechanical Properties, Taylor & Francis Group,New York, 2010, pp. 357–427.

[17] E.O. Hall, Proc. Phys. Soc. Sect. B 64 (1951) 747–753.[18] N.J. Petch, J. Iron Steel Inst. 174 (1953) 25–28.

15Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

[19] A. Winkelmann, J.M. Cairney, M.J. Hoffman, P.J. Martin, A. Bendavid, Surf. Coat.Technol. 200 (2006) 4213–4219.

[20] H. Hahn, P. Mondal, K.A. Padmanabhan, Nanostruct. Mater. 9 (1997) 603–606.[21] H. Hahn, K.A. Padmanabhan, Philos. Mag. B: Phys. Condens. Matter; Stat. Mech.

Electron., Opt. Magn. Prop. 76 (1997) 559–571.[22] H. Van Swygenhoven, D. Farkas, A. Caro, Phys. Rev. B 62 (2000) 831–838.[23] S. Vepřek, S. Reiprich, Thin Solid Films 268 (1995) 64–71.[24] S. Vepřek, S. Reiprich, L. Shizhi, Appl. Phys. Lett. (1995) 2640.[25] S. Veprek, A. Niederhofer, K. Moto, T. Bolom, H.D. Männling, P. Nesladek, G.

Dollinger, A. Bergmaier, Surf. Coat. Technol. 133–134 (2000) 152–159.[26] A.C. Fischer-Cripps, S.J. Bull, N. Schwarzer, Philos. Mag. 92 (2012) 1601–1630.[27] S. Veprek, J. Nanosci. Nanotechnol. 11 (2011) 14–35.[28] S. Veprek, M.G.J. Veprek-Heijman, P. Karvankova, J. Prochazka, Thin Solid Films 476

(2005) 1–29.[29] A.A. Voevodin, S.V. Prasad, J.S. Zabinski, J. Appl. Phys. 82 (1997) 855–858.[30] A.A. Voevodin, J.S. Zabinski, J. Mater. Sci. 33 (1998) 319–327.[31] A.A. Voevodin, J.P. O'Neill, S.V. Prasad, J.S. Zabinski, J. Vac. Sci. Technol. A 17 (1999)

986–992.[32] S. Veprek, M.G.J. Veprek-Heijman, Surf. Coat. Technol. 201 (2007) 6064–6070.[33] H. Söderberg, M. Od́n, J.M. Molina-Aldareguia, L. Hultman, J. Appl. Phys. 97 (2005).[34] S. Veprek, A.S. Argon, R.F. Zhang, Philos. Mag. Lett. 87 (2007) 955–966.[35] S. Hao, B. Delley, S. Veprek, C. Stampfl, Phys. Rev. Lett. 97 (2006) 086102.[36] M. Diserens, J. Patscheider, F. Levy, Surf. Coat. Technol. 120–121 (1999) 158–165.[37] P.H. Mayrhofer, M. Stoiber, C. Mitterer, Scr. Mater. 53 (2005) 241–245.[38] J.E. Sundgren, H.T.G. Hentzell, J. Vac. Sci. Technol. A 4 (1986).[39] G. Kim, B. Kim, S. Lee, J. Hahn, Surf. Coat. Technol. 200 (2005) 1669–1675.[40] F.H.W. Löffler, Surf. Coat. Technol. 68–69 (1994) 729–740.[41] J.C. Sanchez-Lopez, D. Martinez-Martinez, C. Lopez-Cartes, A. Fernandez, M.

Brizuela, A. Garcia-Luis, J.I. Onate, J. Vac. Sci. Technol. A 23 (2005) 681–686.[42] S. PalDey, S.C. Deevi, Mater. Sci. Eng. A 342 (2003) 58–79.[43] P.H. Mayrhofer, C. Mitterer, L. Hultman, H. Clemens, Prog. Mater. Sci. 51 (2006)

1032–1114.[44] P.H. Mayrhofer, A. Hörling, L. Karlsson, J. Sjölén, T. Larsson, C. Mitterer, L. Hultman,

Appl. Phys. Lett. 83 (2003) 2049–2051.[45] J.G. Wen, P.H. Mayrhofer, C. Mitterer, J.E. Greene, I. Petrov, Microsc. Microanal. 12

(2006) 720–721.[46] J. Lin, J.J. Moore, B. Mishra, M. Pinkas, W.D. Sproul, J.A. Rees, Surf. Coat. Technol. 202

(2008) 1418–1436.[47] B. Window, Surf. Coat. Technol. 81 (1996) 92–98.[48] D.M. Mattox, J. Vac. Sci. Technol. A 7 (1989) 1105–1114.[49] S.K. Karkari, A. Vetushka, J.W. Bradley, J. Vac. Sci. Technol. A 21 (2003) L28–L32.[50] M. Audronis, A. Leyland, P.J. Kelly, A. Matthews, Surf. Coat. Technol. 201 (2006)

3970–3976.[51] J.A. Thornton, D.W. Hoffman, Thin Solid Films 171 (1989) 5–31.[52] R. Messier, A.P. Giri, R.A. Roy, J. Vac. Sci. Technol. A 2 (1984) 500–503.[53] W.D. Sproul, Surf. Coat. Technol. 81 (1996) 1–7.[54] H. Ljungcrantz, L. Hultman, J.E. Sundgren, L. Karlsson, J. Appl. Phys. 78 (1995)

832–837.[55] B. Chapman, Glow Discharge Processes: Sputtering and Plasma Etching, JohnWiley

& Sons, New York, 1980.[56] P.H. Mayrhofer, F. Kunc, J. Musil, C. Mitterer, Thin Solid Films 415 (2002)

151–159.[57] J. Musil, S. Kadlec, J. Vyskočil, V. Valvoda, Thin Solid Films 167 (1988) 107–120.[58] Y.X. Wang, S. Zhang, J.W. Lee, W.S. Lew, B. Li, Surf. Coat. Technol. 206 (2012)

5103–5107.[59] H. Wang, S. Zhang, Y. Li, D. Sun, Thin Solid Films 516 (2008) 5419–5423.[60] P. Karvánková, H.D. Männling, C. Eggs, S. Veprek, Surf. Coat. Technol. 146–147

(2001) 280–285.[61] J.S. Koehler, Phys. Rev. B 2 (1970) 547–551.[62] U. Helmersson, S. Todorova, S.A. Barnett, J.E. Sundgren, L.C. Markert, J.E. Greene, J.

Appl. Phys. 62 (1987) 481–484.[63] H.C. Barshilia, K.S. Rajam, Surf. Coat. Technol. 183 (2004) 174–183.[64] X. Chu, S.A. Barnett, M.S. Wong, W.D. Sproul, Surf. Coat. Technol. 57 (1993) 13–18.[65] M. Shinn, L. Hultman, S.A. Barnett, J. Mater. Res. 7 (1992) 901–911.[66] P.E. Hovsepian, D.B. Lewis, W.D. Müunz, A. Rouzaud, P. Juliet, Surf. Coat. Technol.

116–119 (1999) 727–734.[67] X. Chu, M.S. Wong, W.D. Sproul, S.A. Barnett, J. Mater. Res. 14 (1999) 2500–2507.[68] P. Yashar, S.A. Barnett, J. Rechner, W.D. Sproul, J. Vac. Sci. Technol. A 16 (1998)

2913–2918.[69] Q. Yang, C. He, L.R. Zhao, J.P. Immarigeon, Scr. Mater. 46 (2002) 293–297.[70] A. Madan, I.W. Kim, S.C. Cheng, P. Yashar, V.P. Dravid, S.A. Barnett, Phys. Rev. Lett.

78 (1997) 1743–1746.[71] A. Knutsson, M.P. Johansson, P.O.Å. Persson, L. Hultman, M. Od́n, Appl. Phys. Lett.

93 (2008).[72] H.C. Barshilia, M.S. Prakash, A. Jain, K.S. Rajam, Vacuum 77 (2005) 169–179.[73] D.B. Lewis, I. Wadsworth, W.D. Münz, R. Kuzel Jr., V. Valvoda, Surf. Coat. Technol.

116–119 (1999) 284–291.[74] Q. Luo, W.M. Rainforth, W.D. Münz, Wear 225–229 (1999) 74–82.[75] H.C. Barshilia, B. Deepthi, N. Selvakumar, A. Jain, K.S. Rajam, Appl. Surf. Sci. 253

(2007) 5076–5083.[76] H.C. Barshilia, B. Deepthi, K.S. Rajam, Thin Solid Films 516 (2008) 4168–4174.[77] Y. Qiu, S. Zhang, J.W. Lee, B. Li, Y. Wang, D. Zhao, Appl. Surf. Sci. 279 (2013)

189–196.[78] F. Li, S. Zhang, J. Kong, Y. Zhang, W. Zhang, Thin Solid Films 519 (2011) 4910–4916.[79] A.A. Voevodin, J.M. Schneider, C. Rebholz, A. Matthews, Tribol. Int. 29 (1996)

559–570.

[80] A.A. Voevodin, M.A. Capano, S.J.P. Laube, M.S. Donley, J.S. Zabinski, Thin Solid Films298 (1997) 107–115.

[81] E.S. Pacheco, T. Mura, J. Mech. Phys. Solids 17 (1969) 163–170.[82] X. Chu, S.A. Barnett, J. Appl. Phys. 77 (1995) 4403–4411.[83] J. Musil, R. Daniel, Surf. Coat. Technol. 166 (2003) 243–253.[84] J. Musil, P. Zeman, H. Hrubý, P.H. Mayrhofer, Surf. Coat. Technol. 120–121 (1999)

179–183.[85] J. Musil, P. Zeman, Vacuum 52 (1999) 269–275.[86] J.L. He, Y. Setsuhara, I. Shimizu, S. Miyake, Surf. Coat. Technol. 137 (2001) 38–42.[87] J. Musil, I. Leipner, M. Kolega, Surf. Coat. Technol. 115 (1999) 32–37.[88] P.J. Kelly, H. Li, P.S. Benson, K.A. Whitehead, J. Verran, R.D. Arnell, I. Iordanova, Surf.

Coat. Technol. 205 (2010) 1606–1610.[89] S.M. Aouadi, M. Debessai, P. Filip, J. Vac. Sci. Technol. B 22 (2004) 1134–1140.[90] W. Gulbiński, T. Suszko, Surf. Coat. Technol. 201 (2006) 1469–1476.[91] S.M. Aouadi, T. Maeruf, M. Debessai, P. Filip, Vacuum 79 (2005) 186–193.[92] J.H. Hsieh, P.C. Liu, C. Li, M.K. Cheng, S.Y. Chang, Surf. Coat. Technol. 202 (2008)

5530–5534.[93] M. Jirout, J. Musil, Surf. Coat. Technol. 200 (2006) 6792–6800.[94] S. Zhang, D. Sun, Y. Fu, Y.T. Pei, J.T.M. De Hosson, Surf. Coat. Technol. 200 (2005)

1530–1534.[95] S. Zhang, D. Sun, Y. Fu, H. Du, Thin Solid Films 447–448 (2004) 462–467.[96] Y.X. Wang, S. Zhang, J.-W. Lee, W.S. Lew, B. Li, Appl. Surf. Sci. 265 (2013) 418–423.[97] Y.T. Pei, C.Q. Chen, K.P. Shaha, J.T.M. De Hosson, J.W. Bradley, S.A. Voronin, M. Čada,

Acta Mater. 56 (2008) 696–709.[98] J. Robertson, Surf. Coat. Technol. 50 (1992) 185–203.[99] Y. Lifshitz, Diam. Relat. Mater. 5 (1996) 388–400.

[100] Y. Lifshitz, Diam. Relat. Mater. 8 (1999) 1659–1676.[101] A.A. Voevodin, J.S. Zabinski, Thin Solid Films 370 (2000) 223–231.[102] B.K. Tay, Y.H. Cheng, X.Z. Ding, S.P. Lau, X. Shi, G.F. You, D. Sheeja, Diam. Relat.

Mater. 10 (2001) 1082–1087.[103] S. Zhang, X. Lam Bui, X.T. Zeng, X. Li, Thin Solid Films 482 (2005) 138–144.[104] S. Zhang, X.L. Bui, Y. Fu, Thin Solid Films 467 (2004) 261–266.[105] E.T. Thostenson, Z. Ren, T.-W. Chou, Compos. Sci. Technol. 61 (2001) 1899–1912.[106] S. Iijima, Nature 354 (1991) 56–58.[107] R.H. Baughman, A.A. Zakhidov, W.A. de Heer, Science 297 (2002) 787–792.[108] Y. Guo, W. Guo, J. Phys. D. Appl. Phys. 36 (2003) 805–811.[109] P.M. Ajayan, Chem. Rev. 99 (1999) 1787–1799.[110] D.T. Colbert, Plast. Addit. Compounding 5 (2003) 18–25.[111] D. Srivastava, C. Wei, K. Cho, Appl. Mech. Rev. 56 (2003) 215–229.[112] K. Balani, T. Zhang, A. Karakoti, W.Z. Li, S. Seal, A. Agarwal, Acta Mater. 56 (2008)

571–579.[113] Z. Xia, L. Riester, W.A. Curtin, H. Li, B.W. Sheldon, J. Liang, B. Chang, J.M. Xu, Acta

Mater. 52 (2004) 931–944.[114] H. Wang, S. Zhang, S.E. Ong, J. Guo, Nanosci. Nanotechnol. Lett. 3 (2011) 491–493.[115] F.J. Li, S. Zhang, G.Z. Sun, J.H. Kong, R.S. Rawat, U. Ilyas, B. Li, Nanosci. Nanotechnol.

Lett. 4 (2012) 1194–1202.[116] B. Basu, T. Venkateswaran, D.Y. Kim, J. Am. Ceram. Soc. 89 (2006) 2405–2412.[117] D.E. Ruddell, B.R. Stoner, J.Y. Thompson, Thin Solid Films 445 (2003) 14–19.[118] S. Sprio, S. Guicciardi, A. Bellosi, G. Pezzotti, Surf. Coat. Technol. 200 (2006)

4579–4585.[119] J.R. Piascik, J.Y. Thompson, C.A. Bower, B.R. Stoner, J. Vac. Sci. Technol. A 24 (2006)

1091–1095.[120] R.N. Chan, B.R. Stoner, J.Y. Thompson, R.O. Scattergood, J.R. Piascik, Dent. Mater. 29

(2013) 881–887.[121] F. Halitim, N. Ikhlef, L. Boudoukha, G. Fantozzi, J. Phys. D. Appl. Phys. 30 (1997)

330–337.[122] O. Abe, Y. Ohwa, Solid State Ionics 172 (2004) 553–556.[123] O. Abe, S. Takata, Y. Ohwa, J. Eur. Ceram. Soc. 24 (2004) 489–494.[124] K. Holmberg, H. Ronkainen, A. Laukkanen, K. Wallin, S. Hogmark, S. Jacobson, U.

Wiklund, R.M. Souza, P. Ståhle, Wear 267 (2009) 2142–2156.[125] Y. Chunyan, T. Linhai, W. Yinghui, W. Shebin, L. Tianbao, X. Bingshe, Appl. Surf. Sci.

255 (2009) 4033–4038.[126] S.H.N. Lim, D.G. McCulloch, M.M.M. Bilek, D.R. McKenzie, Surf. Coat. Technol.

174–175 (2003) 76–80.[127] R. Fella, H. Holleck, H. Schulz, Surf. Coat. Technol. 36 (1988) 257–264.[128] S. Zhang, X.L. Bui, Y. Fu, D.L. Butler, H. Du, Diam. Relat. Mater. 13 (2004) 867–871.[129] S.J. Bull, D.G. Bhat, M.H. Staia, Surf. Coat. Technol. 163–164 (2003) 499–506.[130] S.J. Bull, D.G. Bhat, M.H. Staia, Surf. Coat. Technol. 163–164 (2003) 507–514.[131] H. Holleck, V. Schier, Surf. Coat. Technol. 76–77 (1995) 328–336 (Part 1).[132] S.J. Suresha, S. Math, V. Jayaram, S.K. Biswas, Philos. Mag. 87 (2007) 2521–2539.[133] D.-K. Lee, S.-H. Lee, J.-J. Lee, Surf. Coat. Technol. 169–170 (2003) 433–437.[134] A. Karimi, Y. Wang, T. Cselle, M. Morstein, Thin Solid Films 420–421 (2002)

275–280.[135] J. Lin, J.J. Moore, W.C. Moerbe, M. Pinkas, B. Mishra, G.L. Doll, W.D. Sproul, Int. J.

Refract. Met. Hard Mater. 28 (2010) 2–14.[136] H. Holleck, M. Lahres, P. Woll, Surf. Coat. Technol. 41 (1990) 179–190.[137] D.E. Wolfe, J. Singh, K. Narasimhan, Surf. Coat. Technol. 165 (2003) 8–25.[138] D.E. Wolfe, J. Singh, K. Narasimhan, Surf. Coat. Technol. 160 (2002) 206–218.[139] J. Lin, J.J. Moore, B. Mishra, M. Pinkas, W.D. Sproul, Surf. Coat. Technol. 204 (2009)

936–940.[140] K.J. Ma, A. Bloyce, T. Bell, Surf. Coat. Technol. 76–77 (1995) 297–302 (Part 1).[141] B.-S. Yau, J.-L. Huang, H.-H. Lu, P. Sajgalik, Surf. Coat. Technol. 194 (2005) 119–127.[142] L. Hultman, J. Bareno, A. Flink, H. Soderberg, K. Larsson, V. Petrova, M. Oden, J.E.

Greene, I. Petrov, Phys. Rev. B Condens. Matter Mater. Phys. 75 (2007).[143] M. Kong, W. Zhao, L. Wei, G. Li, J. Phys. D. Appl. Phys. 40 (2007) 2858–2863.[144] C.-H. Lin, Y.-Z. Tsai, J.-G. Duh, Thin Solid Films 518 (2010) 7312–7315.

16 Y.X. Wang, S. Zhang / Surface & Coatings Technology 258 (2014) 1–16

[145] S.H. Tsai, J.G. Duh, Thin Solid Films 518 (2009) 1480–1483.[146] Y.X. Wang, S. Zhang, J.-W. Lee, W.S. Lew, D. Sun, B. Li, Surf. Coat. Technol. 231

(2013) 346–352.[147] Y.X. Wang, S. Zhang, J.-W. Lee, W.S. Lew, D. Sun, B. Li, Nanosci. Nanotechnol. Lett. 4

(2012) 375–377.[148] P.J. Blau, B.R. Lawn, A.S.f. Testing, M.C.E.-o. Metallography, I.M. Society,

Microindentation Techniques in Materials Science and Engineering: A Symposium

Sponsored by ASTM Committee E-4 on Metallography and by the InternationalMetallographic Society, Philadelphia, PA, 15–18 July 1984, American Society forTesting Materials, 1985.

[149] S. Zhang, D. Sun, Y. Fu, H. Du, Surf. Coat. Technol. 198 (2005) 74–84.[150] X. Zhang, S. Zhang, Nanosci. Nanotechnol. Lett. 3 (2011) 735–743.