Effect of substrate and film thickness on the growth, structure, mechanical and optical properties of chromium diboride thin films

Surface & Coatings Technology 209 (2012) 23–31

Contents lists available at SciVerse ScienceDirect

Surface & Coatings Technology journal homepage: www.elsevier.com/locate/surfcoat

Effect of substrate and film thickness on the growth, structure, mechanical and optical properties of chromium diboride thin films Sandeep Marka a, Menaka b, Ashok K. Ganguli b, M. Ghanashyam Krishna a,⁎ a b

School of Physics, University of Hyderabad, Hyderabad-500046, India Department of Chemistry, Indian Institute of Technology, Hauz Khas, New Delhi 110016, India

a r t i c l e

i n f o

Article history: Received 3 March 2012 Accepted in revised form 3 August 2012 Available online 10 August 2012 Keywords: Chromium diboride Thin film Substrate effect Optical and mechanical properties

a b s t r a c t The growth patterns of chromium diboride thin films deposited by thermal evaporation on borosilicate glass, fused silica, single crystal Si and Cu substrates are reported. It is shown that the adhesion of films is best on Cu, whereas on the other substrate films of thickness of >200 nm are not stable. Scanning electron micrographs reveal that the films on Cu are marginally dense whereas on the other three substrates there is evidence for microporosity, clustering and three dimensional cracks. The as-deposited films on borosilicate glass substrates were amorphous independent of thickness. The films on fused silica, in contrast, crystallized at 60 nm thickness and showed the formation of boron deficient CrB2. At higher thickness there was evidence for both CrB2 and the boron deficient Cr2B3 or Cr3B4. In the case of films on Si substrate, the presence of both CrB2 and the boron deficient Cr2B3 or Cr3B4 is evident. On Cu substrates, up to a thickness of 200 nm only reflections due to the CrB2 phase are observable. At higher thickness the films consist of both CrB2 and the boron deficient Cr2B3 or Cr3B4 phases. Nanoindentation studies reveal strong substrate dependent mechanical behavior. The hardness of the films is highest on fused silica with a value of 13.5 GPa. The highest hardness achieved on borosilicate glass and Cu substrates was 10 GPa. Interestingly, the Young's modulus value on all the substrates is less than 50% of the bulk value, ranging between 60 and 100 GPa. This has been correlated with the presence of microporosity and non-stoichiometry in the films. Spectral transmission studies on the films show that they become opaque on borosilicate glass and fused silica substrates at thicknesses > 150 nm. The reflectance of the Cu substrate is enhanced by 25% at 2500 nm due to the presence of the chromium diboride coatings. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Chromium diboride belongs to a class of transition metal diborides that exhibit superior hardness, corrosion resistance, and high electrical and thermal conductivities. Many first principle theoretical studies and experimental investigations indicate that these properties are from the unique nature of electronic structure, band structure and bonding in these compounds [1–11]. There is, hence, interest in chromium boride films for a variety of technological applications (in hard coating to make scratch proof and acid resistant materials). Interestingly, however, the reports on the growth and properties of CrB2 films are very limited [12–19]. Most of the work is on deposition of CrB2 films by sputtering, that requires large volume of material for target fabrication. The expensive processing of starting materials makes the thin film fabrication costs high and it is therefore important to develop physical vapor deposition (PVD) techniques that are cost-effective. Thermal evaporation is a cost-effective PVD technique that is used routinely for the deposition of thin films.

⁎ Corresponding author. Tel.: +91 40 23134255; fax: +91 40 23010227. E-mail address: [email protected] (M.G. Krishna). 0257-8972/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.surfcoat.2012.08.007

It was demonstrated by the current authors that thermal evaporation can be successfully used to deposit CrB2 films [20]. In the present a more detailed study of the role of the substrates and film thickness in determining the growth patterns of CrB2 films and the implications for physical properties is presented. CrB2 films were deposited at ambient temperature on borosilicate glass (BSG), fused silica, single crystal Si(311) and polycrystalline Cu substrates. The deposited films have been characterized for microstructure, structure, optical and mechanical properties. 2. Experimental High purity CrB2 powders (nanoparticles~25 nm) were synthesized separately via chemical route using chromium salt and boron as a starting material [20]. These CrB2 powders were used as the source material for thermal evaporation. Powders of predetermined weight were placed in a Mo boat and heated to the required temperature using a home built power supply. The substrates (Cu, borosilicate glass (BSG) fused silica and single crystal Si) were placed above the source and were not heated during film deposition. The rate of deposition was controlled by controlling the power to the source. The films were initially deposited at a source–substrate distance of 30 cm, but there was no

24

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

evidence for the formation of a film. The distance was then systematically decreased until there was visual evidence for the formation as inferred from the milky white layer of the film on a BSG substrate. The distance at which this occurred was between 2 and 3 cm from the source. This behavior is attributed to the high melting point of CrB2 and consequently the very low vapor pressure leading to low evaporation rates. It was also observed that the adhesive strength of CrB2 to different substrates was not the same, resulting in differences in the maximum thickness achieved on each of the substrates. The upper limit of thickness achieved on single crystal Si substrates was of the order of 150 nm, whereas on BSG and fused silica substrates the maximum film thickness obtained was 300 nm. In contrast, on Cu substrates, CrB2 films of thickness as high as 8000 nm could be deposited. Prior to deposition the BSG and fused silica substrates were extensively cleaned with distilled water, chromic acid, de-ionized water and iso-propyl alcohol in an ultra-sonicator for 5 min each. The Cu substrates were first cleaned with mild acid followed by ultrasonication for15 min. The Si substrates were cleaned in acetone and methanol followed by ultrasonication. The deposited films were characterized for crystal structure in a Bruker Discover D8 X-ray diffractometer system with Cu Kα radiation (λ =0.15406 nm). Microstructural information was obtained in a Field emission gun based Scanning electron microscope (FE-SEM) model Ultra55 of Carl Zeiss, Germany equipped with an Oxford instrument energy dispersive X‐ray (EDX) spectrometer. The thickness of the films was measured using a surface profilometer. Optical transmission and reflectance spectra have been measured in the wavelength range of 200–2400 nm by means of a dual-beam spectrophotometer (UV–vis–NIR, model, Jasco V-570) having a resolution limit of ±0.2 nm and a sampling interval of 2 nm. Specular reflectance was measured at a near-normal incidence of 6°. Nanomechanical properties were characterized in a nanoindenter [Triboindenter of Hysitron, Minneapolis, USA] [14–20]. Load applied ranged from 100 μN to 1000 μN. An atomic force microscope which enables imaging of the indented area is built into the apparatus.

Fig. 1. FE-SEM images of CrB2 films on Cu substrates at different thicknesses (a) 150 and (b) 8000 nm (higher magnification image of one of the areas is shown in the inset).

3. Results and discussion The FE-SEM images of the films deposited on Cu substrates are shown in Fig. 1(a) and (b) for different thicknesses. The 150 nm thick film (Fig. 1(a)) shows a slightly porous morphology with clustering of particles. The clusters are between 200 and 600 nm in diameter with non-spherical shapes. The SEM images of the thicker film in Fig. 1(b) shows a morphology consisting of more densely packed grains. It is evident from Fig. 1(b), corresponding to the 8000 nm thickness film, that the surface consists of very large areas of almost 10 μm in size separated by large three dimensional cracks. Closer examination of these areas revealed that they, in turn, comprised nanoparticles that are sub-100 nm in size with micropores. Delamination of the films on Cu, thus, occurs only for very high thicknesses. The morphology of the CrB2 films on Si, BSG and fused silica substrates presents very different pictures, as shown in Figs. 2, 3 and 4 respectively. The low magnification SEM image of the morphology of the 100 nm thickness films deposited on Si shown in Fig. 2(a) reveals a large agglomerated mass of film material with areas surrounding it remaining uncoated (indicated by arrows on the image). The agglomerated mass consists mainly of flakes of CrB2 thin film, as shown in the high magnification image in Fig. 2(b). The morphology of the films on BSG substrates displayed in Fig. 3(a) to (d) shows that the adhesion is very poor. This is evident from the presence of cracks and micropores in all the images, indicating that delamination occurs even at a thickness as low as 150 nm, in comparison to the 8000 nm for films on Cu. There is also evidence for clustering of particles that are loosely bound to the surface. The films on fused silica are smooth at a thickness of 60 nm as seen from Fig. 4(a), but develops cracks at the higher thickness of 150 nm (Fig. 4(b)).

Fig. 2. FE SEM images of the 100 nm thick CrB2 film on Si substrate showing (a) the lump of film mass and (b) the higher magnification image of the film portion.

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

25

There are generally four main steps in the evolution of thin films deposited on substrates by PVD [21]. These are nucleation, island growth, agglomeration and finally continuous films. The first two stages would lead to largely discontinuous thin films. The microstructure at these early stages of growth would be controlled by process parameters like rates of deposition and substrate temperature and most importantly the nature of the film–substrate interface. The morphology of the continuous film that forms eventually is, in turn, controlled by the processes at these early stages of growth. For constant temperature and rate of deposition, the growth would entirely be dependent on the nature of film–substrate interface and the thermodynamics in the early stages of thin film growth. The thermodynamics involves the minimization and control of the total free energy of the film–substrate interface. Many theoretical studies on the nucleation and growth of films especially those by Venables and Thompson [22,23] have shown that the growth of a film on a substrate is determined by the free energy balance involving the substrate, film and the interface between the film and substrate. An additional important factor is the contact angle, q between film material and substrate. This balance can be expressed as γs ¼ γ i þ γf cosq

ð1Þ

where γs is an excess free energy associated with the substrate surface, γi, is the free energy of the film–substrate interface, and γf is the free energy of the film surface. There will be a similar equilibrium force balance for the free surfaces and grain boundaries. The average grain size, ri, at impingement is defined as 1=2

ri ¼ ðA=π Þ

ð2Þ

Fig. 3. FE SEM images of CrB2 films on BSG substrates (a) 60 and (b) 150 nm thickness.

Fig. 4. FE SEM images of the films on fused silica substrates (a) 60 and (b) 150 nm thickness.

where A is the average in-plane area, and depends on the relative rates of island nucleation, N (number per unit area per unit time) and growth, L (length/time). The growth rate is also governed by the distance, δ, over which atoms adsorbed on the substrate surface can diffuse to a growing island, which depends on the adatom diffusivity on the substrate surface, D. The adatom diffusivity has an exponential temperature dependence. While the nucleation rate is not a direct function of δ, the fraction of the substrate available for nucleation does depend directly on δ. When D is low, δ is small, and ri is proportional to (L/N)1/3. This means that at ambient temperature both the island nucleation rates and the number of sites available for nucleation are very small leading to island formation. The shapes and sizes of these islands are then entirely driven by the interfacial energy. In addition to the kinetics and thermodynamics of the process conditions of the substrate also drive the morphology of the films. The effect of substrate roughness on film morphology has been studied over the years. Anders et al. [24] have investigated nucleation and coalescence of silver islands on glass by in situ measurements of the sheet resistance. Submonolayer amounts of niobium and other transition metals were deposited prior to the deposition of silver. It was found that in some cases, the transition metals lead to coalescence of silver at nominally thinner films with smoother topology. The smoothing or roughening effects by the presence of the transition metal was explained by kinetically limited transition metal island growth and oxidation, followed by a defect dominated nucleation of silver. Coluci and Cotta [25] have carried out a Monte Carlo simulation of the morphology evolution of films grown on rough substrates. The surfaces considered for the simulation are similar to those of substrates used for the growth of GaAs films by chemical and molecular beam epitaxies. Significantly, their study predicts a surface roughness dependent morphology but it has not been quantified. Mathur and Erlebacher [26] investigated the growth of thin (1–10 nm) films of Pt on Au(111). It was found that on flat Au(111), Pt grows in a layer-by-layer growth mode, but if the gold substrate is exposed to an acidic environment prior to Pt deposition, then the substrate becomes nanoscopically rough (islanded) and Pt growth follows a pseudo-Stranski–Krastanov (SK) growth mode in

26

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

which an initially thin wetting layer becomes rougher with increasing film thickness. The evolution of the interface width has been analyzed experimentally and numerically by Qi et al. [27,28] in their work on film growth on non-planar substrates. They report on the growth of ZrO2 films deposited on BK7 glass substrates of roughness of 0.4 nm by ion beam sputter deposition. The films were deposited for durations ranging from 10 to 150 min at a deposition rate of 1 nm/min. This work also shows clearly the role of surface roughness in determining morphology and adhesion of films to substrates. In addition there are studies which suggest that the interfacial thermal conductance also plays an important role in morphology evolution of films. Adhesion of thin films to substrates has been quantified using the work of adhesion and defined as [29] Wad ¼ γ f þ γ s −γfs

ð3Þ

where γf and γs are the surface energies of the film and substrate, respectively, and γfs is the surface energy between the two materials in contact after deposition. Even in this case, it is observed that the adhesion of film to a substrate is strongly dependent on the interfacial free energy, which in turn will change with the substrate for a given film. Benjamin and Weaver in a very early work [30] showed that the presence of an oxide layer can severely affect the adhesion of films to substrates. In some cases it enhances the adhesion while in other cases it can be detrimental to adhesion. In the present study two of the substrates, BSG and fused silica are based on SiO2 while Si may also have a thin native oxide layer on the surface. It is therefore possible that the low adhesion of CrB2 films to these surfaces is due to the higher interfacial energy required for adhesion. An additional factor that is known to affect both morphology and adhesion of films is the differences in thermal properties of the film and substrates [31]. As stated in the next section these are very different for CrB2 and the substrates used. Hence the differences in morphology and adhesion of the CrB2 films to the different substrates can be attributed to the variations in surface diffusion coefficients, substrate surface roughness and the thermal properties of the substrates. The results of the X-ray diffraction studies carried out on these films are presented in Figs. 5 and 6. The films as-deposited on BSG substrates did not show evidence of crystallization, independent of deposition conditions and thickness. However, on annealing at 500 °C for 1 h the

films crystallized and showed evidence for the presence of both CrB2 and the boron-deficient Cr2B3 or Cr3B4 phases (not shown here). The X-ray diffraction patterns of the CrB2 films deposited on fused silica and Si substrate are shown in Fig. 5(a) and (b) respectively. The films on fused silica substrates at a thickness of 60 nm are nanocrystalline and show a reflection from the boron deficient Cr2B3 or Cr3B4 phase. On increasing the thickness to 150 nm, there is a change in texture with the appearance of the (001) peak and suppression of the peak due to the boron deficient phase. The X-ray diffraction patterns of the film on Si substrate reveal a higher degree of crystallinity with more intense peaks which are indexed to the (002) and (101) planes of CrB2. In addition there are few reflections that can be attributed to either Cr2B3 or Cr3B4. In the case of CrB2 films on Cu, at lower thickness only the substrate reflections are observable. The onset of crystallinity is observed at 150 nm (Fig. 6(b)) as evidenced by the reflection from the (001) plane of CrB2. Similar to the films deposited on Si, at the highest thickness of 8000 nm crystallization is very evident with peaks from the (001), (100), (110), and (200) planes of CrB2. In addition, peaks attributable to the boron deficient Cr2B3 or Cr3B4 phases are also present. This is similar to the behavior on Si substrates, shown in Fig. 5(b). The appearance of diffraction peaks due to the boron deficient phases indicates dissociation of the source material resulting in non-stoichiometry. There were deviations between the observed and standard peak positions of the CrB2 phase in all cases, indicating the presence of large residual stresses in the films, which correlates well with the microstructural evidence presented in Figs. 1 to 4. In general, the stress was tensile for all the films, independent of the substrate, as indicated by the shift to lower 2θ values of all reflections as compared to the standard values. A first order estimate of the total residual stress using the formula [(d − do) / do] (d and do being the observed and standard d spacing, respectively) was between + 1 and + 5% in all cases. The origin of stresses can either be extrinsic or intrinsic [32–36]. The extrinsic stresses are due to factors such as differences in thermal expansion coefficients of the film material and substrate or their lattice parameters. In the current case the thermal expansion coefficients (TEC) of silica, Si and Cu are 0.55, 2.66 and 17 × 10 − 6/°C, respectively, whereas the value for CrB2 is 6.21–7.43 × 10 − 6/K. However, since the films were deposited at ambient temperature, thermal stresses due to the mismatch in the TEC can be neglected as a factor contributing to the total stress. Similarly,

Fig. 5. X-ray diffraction patterns of CrB2 films on (a) fused silica and (b) Si substrates.

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

27

Fig. 6. X-ray diffraction patterns of CrB2 films on Cu with different thicknesses (a) 100, (b) 150 and (c) 8000 nm.

lattice mismatch cannot cause stress on fused silica and BSG substrates, since they are amorphous. The lattice constants of CrB2 are a = 2.969 and c = 3.066 Å as compared to 3.615 Å for Cu and 5.430 Å for Si, which may be one of the causes for the large stress for films deposited on these substrates. The main cause for the stress, however, appears to be intrinsic, due to the accumulating effect of the crystallographic flaws that are built into the coating during deposition. It has been shown in a variety of metal films that for very low values of the ratio of the deposition temperature, Td, to that of the melting point of the film, Tm, intrinsic stresses are dominant. The melting point of CrB2 is >2000 °C, leading to a value of ~0.02 for Td/Tm. In a recent review by Koch [34], the role of intrinsic stresses has been illustrated in detail, especially in the case of thermally evaporated metal thin films. The evolution of stress as a function of film thickness has been evaluated in terms of a film force, Ff, which is defined as the force per unit width of film. It is further shown that different metals deposited on the same substrate exhibit very different stress behaviors. Evidently, for different film–substrate combinations the stresses will be different. It has been shown by many workers that interfacial stresses between the film and substrate can lead to debonding and delamination of the films [37–39]. The process of film formation on a substrate, by vapor deposition, involves the diffusion of adatoms on the surface until they find lowest energy sites, where they stick to the substrates. These form the initial nucleation sites for subsequent formation of a continuous film. Diffusion allows the number density of previously created defects to reduce, by grain growth, dislocation climb, annihilation, etc. resulting in a misfit as the film attempts to either contract or expand. However, since the in-plane misfit displacements are constrained by the substrate, elastic strains develop in the film. For films with a high defect density, these diffusional effects can be rapid, leading to large ‘intrinsic’ stresses. The residual stresses can also be redistributed by plastic deformation and creep. The processes of cracking, debonding and delamination in stressed thin films structures are subject to a steady

state energy release rate, Gss. When this energy release rate is equated to the relevant fracture energy, a lower bound condition for cracking is found. This condition is associated with a critical layer thickness (hc) having non-dimensional form [37]
2 T λ ¼ EΓ=hc σ

ð4Þ

where λ is a configuration dependent, nondimensional cracking number (of the order of unity) and Γ is the relevant fracture energy (ceramic or interface). Hence, when h b hc, there is insufficient strain energy to propagate a crack or debond to any appreciable distance. The critical thickness is dependent not only on the process but also on the various mechanisms at the interface between the film and substrate such as grain growth, point defects and voids at the interface, threading dislocations etc. Since these processes are inherently dependent on film–substrate combination, so is the critical film thickness. From these models it is evident that many mechanisms acting in conjunction cause the delamination to occur. A more quantitative treatment to isolate the dominant mechanism for delamination and debonding in the current study requires many detailed measurements, which are beyond the scope of this work. Only one of these techniques, nanoindentation, has been used to gain some insight into the reasons for delamination on some of the substrates, in addition to characterizing the mechanical properties of the films. In nanoindentation experiments, the basic assumptions are [40–43] a) Deformation upon unloading is purely elastic b) The compliance of the sample and of the indenter tip can be combined as springs in series, to give 2

2

1 1−νi 1−νs ¼ þ Er Ei Es

ð5Þ

28

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

where Er is the “reduced modulus”, E is the Young's modulus, ν is Poisson's ratio and subscripts i and s refer to the indenter and sample respectively. c) The contact can be modeled using an analytical model for contact between a rigid indenter of defined shape with a homogeneous isotropic elastic half space using pffiffiffiffi A S ¼ 2 pffiffi Er r

ð6Þ

where S is the contact stiffness and A the contact area. The contact depth, hc, is defined as the distance along the indenter axis that the indenter is in contact with the sample material. Extrapolation of the slope S of the load displacement curve to the displacement axis, at maximum load Pmax, yields the contact depth of the indent. To determine the hardness, a measure of the indentation area is needed. A convenient way to do this is to use the projected contact area at maximum load. If the tip shape is accurately known, a tip ‘contact area function’, Ac, can be generated where, Ac ¼ f ðhc Þ:

ð7Þ

The hardness is then obtained simply as H ¼ Pmax =Ac ðhc Þ:

ð8Þ

Note that this definition of the hardness is different from the conventional definition of hardness. In the nanoindentation analysis the hardness is calculated utilizing the contact area at maximum load whereas in conventional tests the area of the residual indent after unloading is used. Since an elastic contact analysis is used, the elastic modulus of the material can be obtained. The contact area is combined with the stiffness at maximum load to find the reduced modulus. If the Young's Modulus, E and ν for the indenter material are known, the ratio E / (1 − ν 2) for the sample material can be obtained. The contact solution predicts that the unloading data for an elastic contact for many simple indenter geometries (sphere, cone, flat punch and paraboloids of revolution) follows a power law that can be written as follows: P ¼ αh

m

ð9Þ

the unloading data for stiffness measurement is m

P ¼ α ðh  hf Þ

ð11Þ

where the constants α, m, and hf are all determined by a least squares fitting procedure. The initial unloading slope is then found by analytically differentiating this expression and evaluating the derivative at the peak load and displacement. The load limit is used to select different loading rates according to our choice in different segments. In this study 3 segments were used: loading, creep and unloading. For all experiments in this study, the loading and unloading rates were maintained at 100 μN/s. Creep test is used to determine the change in the indentation depth at maximum load for 10 s. There are 10 segments in the load function generator. By using 10 segments, loading function can be generated in several ways. These are used to find the effects of loading rates on the mechanical properties of films. In a typical sequence, segment 1 is for loading force from 0 to 1000 μN in 10 s, segment 2 is for holding the peak load for 10 s and finally segment 3 is for unloading the force from 1000 μN to 0 in 10 s. To determine the tip area function, we first carry out 36 or 60 indents on a standard sample (usually fused silica substrate, because the modulus of fused silica is known) at different loads starting from 100 μN to 8 mN and fit the area function for calibration. The obtained load–displacement curves have been used to calculate the tip area function from Eq. (8). A tip shape function, A (hc), is determined, given a sufficient number of measurements over a range of hc values, by fitting the A vs. hc data, typically using a multi-term polynomial fit of the form: 2

AðhÞ ¼ C0 h þ C1 hC2 h

1=2

þ C3 h

1=4

1=128

þ … þ C8 h

ð12Þ

where constants C0 = 24.5, C1 = − 2.2025 × 10 3, C2 = 2.158 × 10 5, C3 = − 1.7098 × 10 6, C4 = − 3.4482 × 10 6, C5 = − 1.9066 × 10 6 were obtained from the tip area function curve fit. The constant of 24.5 is used, as it is assumed that a Berkovich indenter has a perfect tip. The additional terms attempt to account for deviations from ideal geometry, such as blunting of the tip. The load–displacement curves for the ~200 nm films deposited on Cu, BSG and fused silica substrates are shown in Fig. 7. The same thickness is used to enable a reasonable comparison between films deposited on different substrates. The load–displacement curve for the film on Cu indicates that the maximum indentation depth on this film is ~150 nm. The width of the hysteresis curve is also very low indicating an elastic– plastic contact. The calculated hardness of the film on Cu was 10 GPa.

where P is the indenter load, h is the elastic displacement of the indenter, and α and m are constants. Values of the exponent m for some common punch geometries are m = 1 for flat cylinders, m = 2 for cones, m = 1.5 for spheres in the limit of small displacements, and m = 1.5 for paraboloids of revolution. This equation was further modified to account for the contact area changes during unloading, leading to the expression hc ¼ hmax −εðPmax =SÞ

ð10Þ

where ε = 0.72, 0.75 and 1, for cone, sphere and flat-punch geometries respectively. A power law function is then fitted to the unloading segment, which yields the contact stiffness as slope of this function at maximum load. This slope in addition to the appropriate value of ε is used in order to determine the actual contact depth so that it is finally possible to derive the indentation modulus (2) and the indentation hardness (3) from the measurement. Both values, indentation modulus as well as indentation hardness, depend strongly on the area function A (hc) and the accuracy with which it is determined. If hmax corresponds to the maximum depth, a to the half-diagonal projected on the surface, hf to the residual depth, hc to the contact depth, and hs to the deflection depth, then the unloading stiffness, S is defined for h = hmax. The actual relationship we use to describe

Fig. 7. Load–displacement curves for the CrB2 films deposited on Cu, BSG and quartz substrates.

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

The behavior of the films on BSG and fused silica shows that the indentation depth is of the order of 100 nm, which is much lower than that for Cu, indicating that these films are harder. The calculated hardness of the films was 10 and 13 GPa, respectively on BSG and fused silica. The calculated values of the Young's modulus from this data are 60, 110 and 85 GPa for the films on Cu, BSG and fused silica respectively. The values of hardness and the Young's modulus are both much lower than that observed experimentally by other workers and predictions from first principles calculations, which range from 30 to 40 GPa for hardness and 300–350 GPa for the Young's modulus. [9–19]. The values of hardness have been taken at a depth of 50–60 nm which is ~30% of the total thickness of the films. It can, therefore, be considered that the major contribution to the hardness is from the films only. The hardness at smaller displacements may not be accurate due to difficulties in tip calibration, surface roughness, and property inhomogeneities [44]. The hardness of a material exemplifies its resistance against permanent deformation. Hard materials can be divided into intrinsic and extrinsic types. It is accepted that the hardness of a perfect crystal is intrinsic while that of nanocrystalline and polycrystalline materials is extrinsic. The intrinsic hardness of strongly bonded covalent/ionic crystals is associated directly with the bond strength [45–47]. Extrinsic hard materials are based mainly on microstructural manipulation in the form of nanocomposites [48]. The hardness and elastic constants of materials are also influenced strongly by the presence of porosity. Many empirical relations that demonstrate the functional dependence of elastic moduli on porosity have been reported in literature [49] M ¼ Mo expð−bPÞ n

ð13Þ

M ¼ Mo ð1−fPÞ

ð14Þ

M ¼ Mo ð1−C1 P þ C2 PÞ

ð15Þ

M ¼ Mo ð1−PÞ=ð1 þ aPÞ

ð16Þ

where M and Mo are the Young's modulus for the porous and dense materials, respectively. P is the fraction of porosity (0b Pb 1); C1, C2, f and a are constants that describe the extent of porosity. A similar functional dependence of the Vicker's hardness on porosity has been reported [50]. H ¼ Ho expð−bPÞ:

ð17Þ

Here H and Ho are the hardness of the porous and dense materials respectively and b is a constant dependent on the applied load. Using these equations as well as experiments where porosity has been controlled in ceramics, it has been shown that porosity > 20% leads to a 50% decrease in both hardness and the Young's modulus. The SEM micrographs shown in Figs. 1 to 4 clearly suggest that the films, in the current case, are microporous and also have large cracks. It can therefore be concluded that the low values of hardness and the Young's modulus in the CrB2 films, in our case, is due to the high porosity (> 20%). An additional factor that contributes to the decrease in hardness is the presence of non-stoichiometric boron deficient phases in the films. The X-ray diffraction patterns suggest that there is a reasonably large volume fraction of Cr2B3 or Cr3B4 in the films, leading to a further lowering of the hardness. In a nanoindentation experiment, delamination manifests itself as a single or multiple pop-in event. A pop-in is defined as a displacement burst caused by local stress fields at the interface of the film and substrate. Pop-in generally occurs when maximum shear stress generated under the indenter is of the order of theoretical shear strength [41,42]. This high local stress causes homogeneous nucleation of threading dislocations beneath the indenter surface, producing a sudden displacement discontinuity. However, the threshold stress required for the occurrence of the first ‘pop-in’ event is more than that of the remaining

29

other ‘pop-ins’. The pop-in event shows up on the load–displacement graph as increase in the indentation depth at constant load, which is followed by the usual increase in depth with increasing load, i.e. like a step. Similar behavior was observed at some points in the CrB2 films in our study. The plots of load vs. indentation depth curves at these points on Cu, BSG and fused silica substrates are shown in Fig. 8. Thus, existence of microporosity leads to localized regions of high stress which act as initiators of cracks at the interface between the film and substrate leading to the delamination of the films due to crack propagation. The absence of the so called “pop-out” in the unloading part of the curve shows that the process of pop-in is irreversible and can therefore be related to delamination. The measured spectral transmittance curves of the films grown on BSG substrates are shown in Fig. 9(a) to (c) as a function of increasing thickness and the spectrum for the bare substrate is shown in Fig. 9(d). The 60 nm thick films are reasonably transparent showing 60% transmission at a wavelength of 750 nm that increases to 72% at 2500 nm. At the same wavelengths i.e. 750 and 2500 nm there are 10 and 18% decreases, respectively, in transmittance to 50% and 54%, respectively for the 120 nm thickness films. With further increase in thickness to 150 nm, there is a very sharp decrease in the transmission of the films at both the above wavelengths which make them nearly opaque at all wavelengths. The behavior on fused silica substrates is very similar to that of the films on BSG, as observed from Fig. 10(a) to(c) [the bare substrate spectrum is shown in Fig. 10(d)]. At a thickness of 120 nm the films exhibit increase in transmission with increase in wavelength, with values of 40 and 55% respectively at 750 and 2500 nm. The transition to opacity occurs at a thickness of 170 nm and the films remain opaque with further increase in thickness to 250 nm. The increase in transmission with wavelength for the thinner films is analogous to semiconductors that, generally, show increase in transmission, or decreased absorption, at wavelengths above the optical band gap. The minor differences in spectral transmission characteristics of the films can be attributed to the differences in morphology on these substrates. The measured specular reflectance spectra of the films on Cu substrates are shown in Fig. 11(a) along with that of the bare substrate spectrum in Fig. 11(b).In this case the transmission spectrum cannot be measured since Cu is opaque in the measured wavelength region of the spectrum between 190 and 2500 nm. There is a 25% increase in reflectance at 2500 nm due to the CrB2 coating of 200 nm thickness. There was no increase in reflectance with further increase in thickness. The shape of the reflectance curve, i.e. almost 35% increase in reflectance above 1000 nm suggests that these films can be used for solar selective absorber applications.

Fig. 8. Load vs indentation for CrB2 films on Cu, BSG and fused silica substrates showing evidence for pop-in behavior.

30

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31

Fig. 9. Measured spectral transmittance curves of CrB2 films on BSG substrate (a) 60 nm, (b) 120 nm and (c) 150 nm thickness and (d) is the bare substrate transmittance curve.

Fig. 11. Measured specular reflectance curves of (a) the bare Cu substrate and (b) 200 nm thickness CrB2 film on Cu substrate.

4. Conclusions

thanks the nanomission, DST for funding. Menaka thanks UGC for a fellowship.

Chromium diboride films have been grown by thermal evaporation on Cu, Si, BSG and fused silica substrates. The growth patterns are found to be strongly dependent on the substrates. It is shown that the films on Si, BSG and fused silica substrates delaminate at thickness of the order of 200 nm whereas on Cu much higher thicknesses can be achieved. X-ray diffraction studies show that the films comprise both the stoichiometric and non-stoichiometric boron deficient phases. The hardness and Young's modulus of the films are much lower than that of the bulk. This behavior is attributed to the presence of microporosity, interfacial stress and non-stoichiometry. The increased reflectivity of the films on Cu in the IR region shows promise for use in solar selective absorber applications.

Acknowledgments The authors acknowledge facilities provided by the School of Physics under the DST-FIST, UGC CAS and DST-ITPAR programmes. SM acknowledges a fellowship from the UGC-CAS scheme. AKG

Fig. 10. Measured spectral transmittance curves of CrB2 films on fused silica substrate (a) 120 nm, (b) 170 nm and (c) 200 nm thickness and (d) is the bare substrate transmittance curve.

References [1] R.B. Kaner, J.J. Gilman, S.H. Tolbert, Science 308 (2005) 1268. [2] H.Y. Chung, M.B. Weinberger, J.B. Levine, A. Kavner, J.M. Yang, S.H. Tolbert, R.B. Kaner, Science 316 (2007) 436. [3] Q. Gu, G. Krauss, W. Steurer, Adv. Mater. 20 (2008) 3620. [4] J.B. Levine, S.H. Tolbert, R.B. Kaner, Adv. Funct. Mater. 19 (2009) 3519. [5] R.W. Cumberland, M.B. Weinberger, J.J. Gilman, S.M. Clark, S.H. Tolbert, R.B. Kaner, J. Am. Chem. Soc. 127 (2005) 7264. [6] M. Nath, B.A. Parkinson, Adv. Mater. 18 (2006) 1865. [7] T.B. Massalski, H. Okamoto, P.R. Subramanian, L. Kacprazak, in: 2nd ed., Binary Alloy Phase Diagram, 1, ASM International, 1990, p. 473. [8] J.K. Sonber, T.S.R.Ch. Murthy, C. Subramanian, Sunil Kumar, R.K. Fotedar, A.K. Suri, Int. J. Refract. Met. Hard Mater. 27 (2009) 912. [9] P. Vajeeston, P. Ravindran, C. Ravi, R. Asokamani, Phys. Rev. B 63 (2001) 045115. [10] J. Matsushita, K. Shimao, Y. Machida, T. Takao, K. Iizumi, Y. Sawada, K. Shim, Mater. Sci. Forum 534–536 (2007) 1077. [11] N.L. Okamoto, M. Kusakari, K. Tanaka, H. Inui, S. Otanii, Acta Mater. 58 (2010) 76. [12] Ph.V. Kiryukhantsev-Korneev, J.F. Pierson, M.I. Petrzhik, M. Alnot, E.A. Levashov, D.V. Shtansky, Thin Solid Films 517 (2009) 2675. [13] M. Audronis, P.J. Kelly, A. Leyland, A. Matthews, Thin Solid Films 515 (2006) 1511. [14] M. Audronis, P.J. Kelly, R.D. Arnell, A.V. Valiulis, Surf. Coat. Technol. 200 (2006) 4166. [15] H.S. Choi, B. Park, J.J. Lee, Surf. Coat. Technol. 202 (2007) 982. [16] M. Zhou, M. Nose, Y. Makino, K. Nogi, Thin Solid Films 359 (2000) 165. [17] M. Audronis, A. Leyland, P.J. Kelly, A. Matthews, Appl. Phys. A 91 (2008) 77. [18] T.N. Oder, J.R. Williams, M.J. Bozack, V. Iyer, S.E. Mohney, J. Crofton, J. Elect. Mater. 27 (1998) 324. [19] C. Cheng, J. Lee, L. Ho, H.W. Chen, Y. Chan, J. Duh, Surf. Coat. Technol. 206 (2011) 1711. [20] M. Jha, S. Marka, M. Ghanashyam Krishna, A.K. Ganguli, Mater. Lett. 73 (2012) 220. [21] L. Maissel, R. Glang, Handbook of Thin Film Technology, McGrawHill, USA, 1970. [22] C. Ratsch, J.A. Venables, J. Vac. Sci. Technol. A 21 (2003) S96. [23] C.V. Thompson, Annu. Rev. Mater. Sci. 20 (1990) 245. [24] A. Anders, E. Byon, D.-H. Kim, K. Fukuda, S.H.N. Lim, Solid State Commun. 140 (2006) 225. [25] V.R. Coluci, M.A. Cotta, Phys. Rev. B 61 (2000) 13703. [26] A. Mathur, J. Erlebacher, Surf. Sci. 602 (2008) 2863. [27] H.J. Qi, J.D. Shao, D.P. Zhang, K. Yi, Z.X. Fan, Appl. Surf. Sci. 249 (2005) 85. [28] H.J. Qi, L.H. Huan, Z.S. Tang, C.F. Cheng, J.D. Shao, Z.X. Fan, Thin Solid Films 444 (2003) 146. [29] W.W. Gerberich, M.J. Cordill, Rep. Prog. Phys. 69 (2006) 2157. [30] P. Benjamin, C. Weaver Proc, R. Soc. Lond. Series A 261 (1961) 516. [31] R. Panat, K. Jimmy Hsia, D.G. Cahill, J. Appl. Phys. 97 (2005) 013521. [32] J.A. Thornton, D.W. Hoffman, Thin Solid Films 171 (1989) 5. [33] S.G. Mayr, K. Samwer, Phys. Rev. Lett. 87 (2001) 036105. [34] R. Koch, Surf. Coat. Technol. 204 (2010) 1973. [35] M.F. Doerner, W.D. Nix, CRC Crit Rev. Solid State Mater. Sci. 14 (1988) 225. [36] R. Abermann, Vacuum 41 (1990) 1279. [37] A.G. Evans, J.W. Hutchinson, Acta Metall. Mater. 43 (1995) 2507. [38] L.L. Mishnaevsky Jr., D. Gross, Inter. Appl. Mech. 40 (2004) 140. [39] W.D. Nix, Met. Trans.A 20 (1989) 2217. [40] W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564. [41] B. Bhushan, X. Li International, Mater. Rev. 48 (2003) 125.

S. Marka et al. / Surface & Coatings Technology 209 (2012) 23–31 [42] A.C. Fischer-Cripps, Nanoindentation, second ed. Springer, Mechanical Engineering Series, New York, 2004. [43] A. Gouldstone, H.J. Koh, K.Y. Zeng, A.E. Giannakopoulos, S. Suresh, Acta mater. 48 (2000) 2277. [44] H. Wen, X. Wang, L. Li, J. Appl. Phys. 100 (2006) 084315. [45] Q.M. Hu, K. Kádas, S. Hogmark, R. Yang, B. Johansson, L. Vitos, Appl. Phys. Lett. 91 (2007) 121918. [46] P. Lazar, X. Chen, R. Podloucky, Phys. Rev. B 80 (2009) 2.

31

[47] H.Y. Chung, M.B. Weinberger, J. Yang, S.H. Tolbert, R.B. Kaner, Appl. Phys. Lett. 92 (2008) 261904. [48] S. Veprek, M.G.J. Veprek-Heijman, P. Karvankova, J. Prochazka, Thin Solid Films 476 (2005) 1. [49] J. Luo, R. Stevens, Ceram. Int. 25 (1999) 281. [50] J. Musil, F. Kunc, H. Zeman, H. Polakova, Surf. Coat. Technol. 154 (2002) 304.