Journal of Materials Science

, Volume 47, Issue 4, pp 1621–1630

The influence of Zr layer thickness on contact deformation and fracture in a ZrN–Zr multilayer coating

Authors

    • Department of Materials EngineeringIndian Institute of Science
  • Vikram Jayaram
    • Department of Materials EngineeringIndian Institute of Science
E-MRS MACAN

DOI: 10.1007/s10853-011-6001-y

Cite this article as:
Verma, N. & Jayaram, V. J Mater Sci (2012) 47: 1621. doi:10.1007/s10853-011-6001-y

Abstract

In order to understand the influence of ductile metal interlayer on the overall deformation behavior of metal/nitride multilayer, different configurations of metal and nitride layers were deposited and tested under indentation loading. To provide insight into the trends in deformation with multilayer spacings, an FEM model with elastic-perfect plastic metal layers alternate with an elastic nitride on top of an elastic–plastic substrate. The strong strain mismatch between the metal and nitride layers significantly alters the stress field under contact loading leading to micro-cracking in the nitride, large tensile stresses immediately below the contact, and a transition from columnar sliding in thin metal films to a more uniform bending and microcracking in thicker coatings.

Introduction

Metal/nitride composite multilayer coatings can offer advantages over monolithic nitrides if they display higher toughness without compromising on hardness significantly [1, 2]. Metal interlayers are reported to enhance adhesion by reducing residual stress levels of the overall coating owing to shear deformation of metal, corrosion resistance by interrupting the corrosion path through the nitride at each interlayer and also possibly change the texture, in the case of TiN, to minimize the interfacial energy with the interlayer [39]. Different Ti through thickness distributions were checked for their influence on residual stress with constant, increasing, and decreasing metal content toward coating/substrate interface. Adhesion is not only dependent on total amount of metal but also dependent on the metal distribution and it is best for increasing metal amount near coating/substrate interface [6]. These coating are deposited by many processes such as closed field unbalanced magnetron sputtering, dual ion beam sputtering and pulsed laser deposition. Many such systems have been studied, including metal/nitride systems such as TiN/Ti, ZrN/Zr, TiN/Ag, Al/SiC, ZrN/W, CrN/Cr, CrN/Cu, AlN/Al, TaN/Ta, and Al/Al2O3 [418]. The first metal/nitride coating was TiN/Ti [11]. The microstructure of these coatings are mostly columnar with sharp interfaces and the nitride renucleates at each interface except in case of epitaxial ZrN/W. Nitrogen is believed to be present in the metal layer depending on its solubility and very thin layers degrade the multilayer mechanical properties due to the lack of sharp interfaces. In case of CrN/Cr multilayers, the microstructure coarsens as the layer becomes finer and less defective metal layers form due to the higher diffusivity of metal [11]. AlN/Al shows that microstructure become coarser with larger layer thickness of AlN and better crystallization of layers which increases the surface roughness of film with increase in layer thickness. A preferred orientation (111) is seen in case of ZrN and TiN multilayers with their respective metals due to the large number of interfaces that relieve residual stresses. This orientation (111) is seen to show the best mechanical properties. Friction coefficient gets reduced in case of metal/nitride coatings and wear life increases due to metal ductility [13]. ZrN/W show roughening of layers owing to kinetic effect of residual stress relaxation [13]. Delamination of W from underlying ZrN layers is seen due to compressive stress in W. With decrease in layer spacing increase in residual stress is observed. Metal/nitride systems are believed to be more stable at high temperature as they can be non isostructural and immiscible compared to other nitride/nitride systems which degrade due to diffusional intermixing of layers [19].

Nanoindentation is being extensively used to extract the hardness and modulus of these coatings [24, 25] and correlating it with bilayer periodicity [26, 27] with the advantage of probing a small volume to extract only the properties of coating. Hardness and modulus always show dependence on volume fraction and bilayer periodicity [16]. Hardness in all these systems are shown to increase with lower volume fraction of metal and with finer layer spacing it gets enhanced to above 40% more than the rule of mixture value [15]. Modulus is mostly a function of volume fraction and shows no dependence on bilayer spacing [17]. Different metals were also used to study the influence of different crystal structure and nitrogen reactivity on mechanical properties. Al(FCC), Ti(HCP) both having reactivity to nitrogen and Cu(FCC) with no reactivity to nitrogen were chosen for the study. All coatings were columnar and columns become continuous with finer layer spacing. One interlayer is sufficient to reduce the hardness and Ti interlayer did not reduce the hardness which might be due to N absorption during deposition. The residual stress in case of Al and Ti increases with number of interfaces but the opposite happens in case of Cu due to less reactivity of Cu with N. Hence Cu can deform and reduce compressive residual stress [29].

Tensile test on Al/Al2O3 showed that the strength was related to bilayer spacing and followed the Hall–Petch relation, with yield strengths ~μ/70 where μ is shear modulus of Al [20]. The modulus is reported to be lower than the weighted average of layers and believed to be due to lower density and microcracking in the alumina layer. In case of indentations in Al/SiC multilayers, void formation in Al layers and shear banding was reported which was attributed to the constraint offered to deformation of soft phase by the harder SiC [21]. It is shown in case of these coatings that the nitride grows with columnar grains and metal does not allow the columnar grains of nitride to grow throughout the whole coating. Columnar sliding is the common mode of deformation in these coatings which is greatly affected by the presence of metal interlayers as the columns are obstructed by the interlayer from sliding and so provide more resistance to deformation [22]. There are reports on TiN/Al multilayers where the fraction of metal nitride was varied and coating of higher metal content was tested. The deformation was seen to be cooperative and the stiff nitride layer was also shown to undergo huge bending with equal volume fraction of nitride and metal and at very small length scales. In all other combinations of metal and nitride, the metal plastically deformed and caused the nitride to deform by cracking at higher layer spacing and shear banding at thinner layer spacing with unequal volume fractions of phases [23].

FEM has been used to evaluate the complex stress state under indentation in these composite coatings successfully. For example, it has been reported [21] that plastic flow in the metal during unloading can lead to changes in the calculated modulus using standard techniques. Metal plastically deforms and the modulus cannot be evaluated simply by unloading curve. Numerical simulations using HAFILM codes, which can simulate large plastic deformation and rotation, have reported the influence of interlayer properties and thickness on hardness, modulus and the ratio of H/E which is a measure of the propensity for cracking. The hardness and modulus decreases with increasing number of interlayers and H/E have no influence of number of interlayers for Ti interlayer [28, 29]. FEM calculations of nanoindentations on Al/SiC [30] revealed that the composite elastic properties may be obtained beyond an indentation depth of 8–10 bilayer spacings. It is reported that in case of a multilayer of alternate soft and hard layers the major shearing take place in soft layers and the hard layers merely slide over each other and so do not experience large bending stresses [31]. Thus, the metal/nitride coating can show better performance over a single hard monolayer.

Despite extensive numerical analyses, the links with detailed microscopical observations are fewer [32, 33]. In the transition metal mononitride system, a further complicating factor is the tendency to form secondary nitrides, e.g., Ti2N, Cr2N, etc. One of the rare combinations that are stable is Zr–ZrN. Consequently, we have chosen this particular system to examine the role of multilayering on contact deformation supplemented with FEM calculations to understand the role of the stress field.

Experimental details

The finite element model consists of alternating Zr and ZrN layers on a SS 304 substrate (Fig. 1a). The model mimics the actual configuration of multilayers tested experimentally. A schematic of the model, along with the finite element mesh near the indentation site, are shown in Fig. 1a. The top layer is ZrN in contact with indenter and the bottom layer at film substrate interface is Zr. A spherical indenter with radius of 5 μm is used. The model used for calculations is axisymmetric to reduce the computation time. The sample dimensions were chosen to be sufficiently large (50 × 50 μm), including the 5 μm film on the top, to exclude edge effects. The Zr and ZrN layer thicknesses were varied from 30 to 110 nm and 200 to 250 nm, respectively to match with the experimental configuration. The left end boundary and the symmetric axis were restricted to move along ‘x’ (Fig. 1) and the bottom boundary was restricted in both directions. The top surface was restricted to follow the indenter geometry. A very fine mesh was introduced near the indenter–film contact to get better resolution of stress in film (Fig. 1b). The convergence was checked with the effect of mesh refinement on stress values. The finite element program ABAQUS version 6.5 was used to carry out the analysis. The Young’s modulus and Poisson’s ratio for Zr and was taken to be 91 GPa and 0.34 from literature [34] and 325 GPa and 0.25, respectively for ZrN. The modulus for ZrN was evaluated by nanoindentation. The yield stress for two different grain sizes of Zr was taken from the literature [35] and extrapolated to the current film thicknesses using the Hall–Petch relation. The yield stress for Zr was assumed to be constant as there is no literature available at fine scale for Zr, and the reports for nanocrystalline Ti also shows that there is not much strain hardening [36]. The plastic flow behavior of Zr was taken to follow Von-Mises criterion.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig1_HTML.gif
Fig. 1

a Model representation of multilayer and substrate with spherical indenter in contact with top surface, b enlarged view of region near contact showing refinement of mesh

The ZrN is assumed to be elastic as the yield strength of these materials is very high compared to that of metal. The indenter is assumed to be perfectly rigid and all interfaces are perfectly bonded between all the layers. The yield stress for substrate used was taken to be 333 MPa and the strain hardening behavior was taken from literature [37]. The FEM indentation was done under load control as done in the real situation with a maximum of 0.2 N in all cases.

Multilayer coatings with varying thickness of Zr and ZrN were deposited by sputtering (TEER Coating, UK). The coating thickness of 250/50, 220/110, and 200/30 nm layer thickness of nitride/metal were analyzed after microindentation using FIB (focused ion beam, Strada 200xP, FEI Inc., USA) to see the subsurface deformation. The microstructure after deposition was analyzed through TEM (Tecnai F-30, FEI Inc., USA). The coatings do not exhibit any preferred orientation as seen from diffraction pattern which show more or less continuous rings characteristic of nanocrystalline grains (Fig. 2). Grains are columnar in nitride and continuous in case of monolayer (Fig. 2a) and get interrupted at each interface in the case of multilayers as a result of nitride renucleation at each metal nitride interface (Fig. 3). The fine scale structure of the nitride in multilayer is evident from Fig. 4a while the corresponding diffraction pattern in Fig. 4b pattern establishes the presence of metal phase in the coating.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig2_HTML.jpg
Fig. 2

a TEM dark field image of monolayer, the white contrast representing grains show the columnar microstructure and b the corresponding indexed diffraction pattern having continuous ring pattern showing absence of any preferred orientation or texture

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig3_HTML.jpg
Fig. 3

a TEM dark field image of multilayer where bright grains are seen to be discontinuous throughout the coating thickness and b corresponding diffraction pattern where part of first and second ring is used to take the dark field image

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig4_HTML.jpg
Fig. 4

a TEM bright field of multilayer consisting of ZrN (darker) and Zr (brighter) layers showing nano size grains and b corresponding diffraction pattern of both phases with two indexed rings of Zr phase confirming presence of both phases and revealing absence of texture

Results and discussion

Indentation cross-section

Figures 5, 6, 7, 8 show low magnification views of the cross-sections of indentations of a monolayer, 220/110, 250/50, and 200/30 nm ZrN/Zr coatings, all of 5 μm thickness. The principal features of the deformation are now summarized. Monolayers show characteristic columnar sliding where the sliding event is continuous from substrate to free surface (arrows are marked to indicate the steps at the top surface and bottom interface in Fig. 5. The 30 nm metal layers also show sliding but with a dispersal of the displacement across many events as the free surface is approached with associated microcracking at each interface (Fig. 6). In contrast, the coating with 50 and 110 nm show no continuous sliding into the substrate but a more uniform bending with nanocracks at each interface (Figs. 7, 8). Higher magnification views (Fig. 7b) reveal that the cracks in the nitride originate at the lower interface below the indentation centre and at the upper interface toward the indentation periphery. In the case of 110 nm metal layers, fracture of the nitride appears to have caused shear of the metal across and into these cracks and an intermingling of the two phases as partly evidenced by the reduction in metal layer thickness in these regions (Fig. 8b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig5_HTML.jpg
Fig. 5

FIB cross-section of indented monolayer where arrows indicate the step due to columnar sliding at top surface and interface showing columnar sliding to be continuous throughout thickness of coating

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig6_HTML.jpg
Fig. 6

FIB cross-section of indented multilayer with 30 nm metal layers demonstrating the presence of cracking giving rise to multiple sliding at the top surface converging to single shearing events at the bottom interface with the substrate

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig7_HTML.jpg
Fig. 7

Cross-section of indented multilayer with 50 nm metal layers: a number of cracks seen near the top surface are more compared to bottom interface with substrate, b microcracks in the nitride originate from the bottom interface of each bilayer at the center of indent and from each upper interface at the edge of the indentation (enclosed within the circular and rectangularinset, respectively). Cracks are confined to a single nitride layer and deviate before reaching the opposite interface

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig8_HTML.jpg
Fig. 8

Multilayer with 110 nm metal layers: a Extensive cracking under indent squashes the whole coating near top surface and cracks in the nitride go completely through the layers, b enlarged view of top center portion of deformed zone where area inside circles show that the metal has disappeared or thinned down along with the presence of cracking

FEM

Stress contours obtained from the FEM analysis are shown in Fig. 9 for the radial stress Sxx and in Fig. 10 for the shear stress τxy. It is of interest to note that the radial stress (Fig. 11) and shear stress (Fig. 12) oscillates within each nitride layer and, in sharp contrast to the monolithic hard coating, can even become tensile directly below the indentation (Fig. 11). Within the nitride layer the maximum tensile stress drops as the metal thickness decreases while the minimum stress also changes from tensile to weakly compressive. It may be noted from Fig. 13, which compares the Sxx variation below and outside the indentation, that the location of the maximum is reversed from the lower interface (r = 0.5a) to the upper interface (r = 1.5a). The compressive stress, on the other hand, is no more concentrated only at the center of the indent, as in the case of a monolayer, but resides largely in the metal layer. The maximum level of tensile stress inside the film in case of a monolayer is 5.6 GPa. The multilayer exhibits 19, 14, and 13 GPa for 110, 50, and 30 nm metal interlayers, respectively. The compressive stress level in the monolayer is 3.8 GPa whereas in the case of the multilayer, the maximum compressive stress present (in the Zr) are 16, 20, and 23 GPa for 110, 50, and 30 nm metal interlayers, respectively. The mean shear stress peaks (Fig. 15) slightly below the surface. The thickness averaged shear stress in case of monolayer is 4.67 GPa and in multilayer case it is 2.9, 2.79, and 2.98 GPa for 30, 50, and 110 nm metal layer multilayer, respectively. The plastic strain in the metal is given in Fig. 17a–c. The highest value of equivalent plastic strain inside metal scales inversely with the metal layer spacing as it is 3.4 for 30 nm, 2.71 for 50 nm, and 2.16 for 110 nm.
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig9_HTML.gif
Fig. 9

Stress contours for Sxx (GPa) × 1012. a monolayer, b 30 nm metal, c 50 nm metal, and d 110 nm metal. Encircled region shows the distance over which Sxx is significant

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig10_HTML.gif
Fig. 10

Stress contours for Sxy (GPa) × 1012. a monolayer, b 30 nm metal, c 50 nm metal, and d 110 nm metal. The highest shear stress is near metal/nitride interface and there is gradation in stress within individual ZrN layer

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig11_HTML.gif
Fig. 11

Variation of Sxx at r = 0 (centre of indent) and with respect to depth for (left to right) 110, 50, 30 nm metal layers. Notice residual tension immediately below indentation center in nitride (reference line kept for zero stress)

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig12_HTML.gif
Fig. 12

Shear stress (Sxy) variation inside film with respect to depth for (from left to right) 110, 50, and 30 nm

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig13_HTML.gif
Fig. 13

Variation of Sxx with respect to depth at (left) r/a = 0.5 (under indent) and (right) r/a = 1.5 (outside indent) for 50 nm metal multilayer. Note reversal of location of maximum tension from lower interface (left) to upper interface (right)

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig14_HTML.gif
Fig. 14

Profile of Sxx at surface for a multilayers and b monolayer. Note the much wider zone of high tension in (a)

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig15_HTML.gif
Fig. 15

Weighted average of shear stress at r/a = 1 in a multilayers and b monolayer with approximate variation of multilayers from a superimposed as a straight line

https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig16_HTML.gif
Fig. 16

Weighted average of tensile stress Sxx in each layer for all multilayers at r/a = 1

Discussion

Radial stress Sxx in the classical Hertzian case is responsible for ring cracking and the so-called bending cracks at the interface due to strain mismatch with the substrate [38], shear stress τxy drives columnar sliding [39]. The stress reversal of Sxx exactly matches with the experimental results (Fig. 7b). Despite the larger tensile stresses in the nitride, cracking is interrupted by the high compression in the metal and one sees distributed damage rather than a single long crack, though, as mentioned above, the weakening of the nitride leads to extensive plastic flow of metal into these cracked regions for the thickest metal layer for which the tensile stress is also the highest. Thus, microcracking of the nitride is more extensive. Another feature of the FEM analysis is the distribution of the tensile stress at the surface nitride layer for multilayers. As seen in Fig. 14, Sxx remains high over a greater radial distance (~5 μm) for the case of multilayers than for monolayers (<1 μm) or the Hertzian case. Where such cracking combines with column sliding, it facilitates a more uniform distribution of columnar sliding as seen earlier in Fig. 6, while a greater density of cracks is also seen in Fig. 7 where sliding is absent. From Fig. 15, it is also evident that the thickness averaged shear stress in the nitride is greatly reduced by multilayering, thus contributing to a reduced driving force for column sliding. While there does not appear to be a major dependence on metal film thickness, when combines this reduction with the normal stress at the contact periphery r = a, it is seen (Fig. 16) that the average tensile stress in the nitride for 200/30 nm is much less than the 250/50 or 220/110 nm case. The weighted average shear stress over full thickness of multilayer is slightly higher in case of 200/30 nm compared to other multilayers. Moreover, average tensile stresses inside nitride due to plastic flow of metal is reduced in case of 200/30 nm compared to other multilayers (Fig. 15a). Both these factors can lead to sliding rather than cracking in this particular case. The deformation in thick metal layers is dictated by plastic flow of metal whereas in case of thin metal layer plastic flow is restricted due to small size; nitride dictates the metal deformation which would be sliding of columns similar to monolayer. The metal layer plastically flows without any cracking and is visible in all the metal layer coatings Figs. 6 and 7. The peak in the plastic strain is almost at the edge of contact where we see maximum bending taking place in the coating (Fig. 17).
https://static-content.springer.com/image/art%3A10.1007%2Fs10853-011-6001-y/MediaObjects/10853_2011_6001_Fig17_HTML.gif
Fig. 17

Stress contours within metal layer for equivalent plastic strain for metal layer thickness of a 110 nm, b 50 nm, and c 30 nm. Plastic strain is maximum near edge of indent and is restricted only within few top layers for the case of thicker metal layers

Conclusion

  • Nitride–metal multilayers display a transition with metal layer thickness in the mode of deformation beneath indentations which can be captured using FEM. Only columnar sliding is seen in case of monolayer and 30 nm metal layers while thicker layers display increasing amounts of microcracking in the nitride.

  • Plastic flow in the metal can generate very high tensile stresses in the nitride. These lead to tensile stresses even directly under the indentation where monolayers generally display high compressive stresses. The interface from which microcracks in the nitride initiate reverse as one moves toward the indentation periphery in agreement with the predictions from FEM.

Acknowledgement

The authors are grateful to Defense Research & Development Organization (Govt. of India) for financial support.

Copyright information

© Springer Science+Business Media, LLC 2011