Mechanical properties of aluminum, zirconium ...

Surface & Coatings Technology 282 (2015) 36–42

Contents lists available at ScienceDirect

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

Mechanical properties of aluminum, zirconium, hafnium and tantalum oxides and their nanolaminates grown by atomic layer deposition Taivo Jõgiaas a,⁎, Roberts Zabels b, Aile Tamm a, Maido Merisalu a, Irina Hussainova c, Mikko Heikkilä d, Hugo Mändar a, Kaupo Kukli a,d, Mikko Ritala d, Markku Leskelä d a

University of Tartu, Institute of Physics, Department of Materials Science, Ravila 14C, EE-50411 Tartu, Estonia University of Latvia, Institute of Solid State Physics, Kengaraga str. 8, LV-1063 Riga, Latvia Tallinn University of Technology, Department of Materials Engineering, Ehitajate tee 5, EE-19086 Tallinn, Estonia d University of Helsinki, Department of Chemistry, P.O. Box 55, FI-00014, Finland b c

a r t i c l e

i n f o

Article history: Received 20 May 2015 Revised 11 September 2015 Accepted in revised form 6 October 2015 Keywords: Instrumented nano-indentation Metal oxide nano-laminates Mechanical properties Thin film Atomic layer deposition

a b s t r a c t The mechanical properties of two different metal oxide nanolaminates comprised of Ta2O5 and Al2O3, HfO2 or ZrO2, grown on soda–lime glass substrate by atomic layer deposition, were investigated. Ta2O5 and Al2O3 layers were amorphous, whereas ZrO2 and HfO2 possessed crystalline structure. Thickness of single oxide layers was varied between 2.5 and 15 nm. The total thickness of the laminate structures was in the range of 160–170 nm. The hardness values of single layer oxides on glass ranged from 6.7 GPa (Ta2O5) to 9.5 GPa (Al2O3). Corresponding elastic moduli were 96 GPa and 101 GPa. The hardnesses of laminates were in the range of 6.8–7.8 GPa and elastic moduli were between 93 and 118 GPa. The results implied a correlation between mechanical properties and the relative content of constituent single oxides. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The microstructural design has attracted a certain interest in the development of hard coatings consisting of a variety of materials ranging from single metal oxides [1–4] to composites or nanolaminates of chemically distinctive compounds [5–8]. Mechanically resistive nanolaminates of different oxides have been deposited by a variety of physical and chemical deposition methods, for example, laser ablation [9], electrophoretic deposition [9], or sputtering [10,11,8] of Al2O3– ZrO2, laser ablation of CeO2–Gd2O3 [12], and atomic layer deposition (ALD) of TiO2–Al2O3 [12–14], Al2O3–Ta2O5 [15], Al2O3–ZnO [17], ZnO/ Al2O3/ZrO2 [18], or aluminum oxide–aluminum alkoxide [3] systems. Moreover, fabrication of metal oxide nanolaminates by chemical vapor deposition (CVD) and atomic layer deposition (ALD) methods allows achieving conformal growth on rough surfaces [14]. Metal oxide laminates have been studied as protective coatings on surfaces of demanding metal alloys, e.g. Al2O3–TiO2 [15] or Al2O3–Ta2O5 laminates on steel [16]. In nanolaminates, different useful physical properties of constituent material layers can be tailored e.g. crystallinity and higher local

Abbreviations: CVD, chemical vapor deposition; ALD, atomic layer deposition; XRR, Xray reflection; SEM, scanning electron microscopy; FIB, focused ion beam; XRD, X-ray diffraction. ⁎ Corresponding author. E-mail address: [email protected] (T. Jõgiaas).

http://dx.doi.org/10.1016/j.surfcoat.2015.10.008 0257-8972/© 2015 Elsevier B.V. All rights reserved.

density of one component with amorphous and laterally homogeneous, uniformly disordered structure of another component material, accompanied by the repeatedly formed interfaces between constituent layers. Among other potential applications, wear-resistant oxide–metal multilayers deposited on glass have been of interest as hard and stable optical coatings [19,20]. Laminates previously have been mechanically evaluated by three point bending for the flexural strength determination [9], Vickers [13, 14] or Berkovich [9–11] indentation, and steel ball wear tests [18]. In the present study, HfO2–Ta2O5 [20], ZrO2–Ta2O5 [21], and Al2O3–Ta2O5 [22] nanolaminates grown by ALD on glass substrates were investigated by nanoindentation technique. Indentation tests were conducted using continuous stiffness method [23,24]. In parallel with laminates, single layer oxide samples grown to thicknesses comparable to that of laminates as well as bare glass substrates were tested. 2. Experimental details Nanolaminates were deposited in a hot-wall flow-type ALD reactor F120 (ASM Microchemistry Ltd.) [25] at 325 °C and at reactor pressure of 10 mbar using water vapor as oxygen source and Ta(OC2H5)5, Al(CH3)3, HfCl4 and ZrCl4 as metal precursors. Precursor pulse lengths were 0.2 s for metal precursors (0.4 s for HfCl4) and 0.5–2.0 s for H2O. The metal precursors were evaporated from open glass boats inside the reactor. Evaporation temperatures for HfCl4 and ZrCl4 were 140 °C

T. Jõgiaas et al. / Surface & Coatings Technology 282 (2015) 36–42

and 150 °C, respectively. Nitrogen was used as the carrier and purging gas with purge lengths of 0.5 s. The substrates used were soda–lime glass with the surface area of 5 × 5 cm × cm, cleaned prior to deposition in ethanol and dried in nitrogen flow. The laminates were deposited by alternate layering of two different oxides, varying thickness periods within HfO2–Ta2O5, ZrO2–Ta2O5, and Al2O3–Ta2O5 double-layers. The number of ALD cycles applied for the growth of a single 2.5–15 nm thick layer constituting a nanolaminate was varied in the range of 50– 300. Growth rates of single-oxide interlayers were evaluated from single oxide films grown to the thickness of about 160–170 nm. These growth rates were used in calculating the appropriate number of deposition cycles required to achieve the target thicknesses of the thin single layers in the multi-layer structure. The nominal thicknesses for the constituent layers were estimated, prior to the growth, considering average growth rates of 0.055 nm/cycle for HfO2 and ZrO2, 0.06–0.07 nm/cycle for Al2O3, and 0.04 nm/cycle for Ta2O5 [21,22]. In the following text nanolaminates will be designated via their growth recipes expressed by the nominal thicknesses of the intermittent layers. For example, the sequence written as 10 × (15 ZrO2 + 5 Ta 2O5) nm denotes 10 double layers of ZrO2–Ta 2O5 with each ZrO 2 and Ta 2O 5 single layer grown to thicknesses of 15 and 5 nm, respectively. The structure and phase compositions of nanolaminates were determined using either Panalytical X'Pert PRO or Rigaku Smartlab X-ray diffractometer in grazing incidence mode with incidence angles between 0.34 and 1°. X-ray reflection patterns (XRR, detector scanning angle 0– 3.5°) were also obtained using the same diffractometers. Scanning electron microscopy (SEM) using Helios™ NanoLab 600 (FEI) instrument was employed for imaging the traces of nanoindentation and sample cutting with focused ion beam (FIB). Prior to FIB cutting, magnetron sputtering was used to deposit a protective and conductive layer of platinum. Instrumented nanoindentation was performed by Agilent Nanoindenter G200 using continuous stiffness measurement technique. The area function of the Berkovich diamond tip was calibrated using fused silica as reference. Indentations were performed up to a depth of 1.8 μm. The results were averaged between data obtained from eight locations indented on the surface area of 1 × 1 cm × cm, approximately (Fig. 1). 3. Results and discussion 3.1. Film structure X-ray diffraction (XRD) patterns of single Al2O3 and Ta2O5 films did not reveal any peaks, verifying the absence of long-range order in these oxides. The Al2O3, Ta2O5 and also thinner HfO2 and ZrO2 films could be considered as X-ray amorphous (the possible crystal size remained below detection limit). The diffraction patterns of ZrO2–Ta2O5 and HfO2–Ta2O5 nanolaminates are depicted in Fig. 2. ZrO2 and HfO2 films were nanocrystalline, revealing the formation of structural mixtures of stable monoclinic and metastable cubic or tetragonal phases. Films containing HfO2 showed clear characteristic peaks of monoclinic phase but some reflections attributable to the metastable tetragonal phase also appeared. In general, the intensity of the reflections increased with the thickness of ZrO2 and HfO2 layers. The peaks in the HfO2 patterns started to form at layer thicknesses around 10 nm. ZrO2–Ta2O5 nanolaminates showed a presence of the tetragonal/cubic phase of the ZrO2–Ta2O5 nanolaminates, whereas the stable monoclinic phase was not detected. The reflections started to appear at ZrO2 layer thicknesses of 5 nm. One can also note, that 5 nm may occur at the possible limit to the convenient recognition of crystal growth. Below this value, approximately, the X-ray reflections may remain too wide for their reliable distinction from the background signal and the material, even when a short range order is present, it could then be regarded as X-ray amorphous phase. Nevertheless, such kind of laminates consists of structurally and chemically distinct metal oxide, e.g. ZrO2 and

37

Fig. 1. Optical microscopy (upper panel, the scale bar is 10 μm) and scanning electron microscopy (lower panel, scale bar is 4 μm) images of a triangular Berkovich indenter mark on the surface of the nanolaminate grown with layer sequence of 5 nm Ta2O5 + 33 × (2.5 nm ZrO2 + 2.5 nm Ta2O5).

Ta2O5, layers alternately deposited on planar substrates, forming sandwiches with microscopically distinguishable constituent layers. Visualizing further the growth of nanolaminate structures, a representative X-ray reflectivity pattern, measured at angles from 0 to 3.5°, is depicted in Fig. 3. Similar curves verifying the multilayer structure were recorded for all the nanolaminates. The fitted curve in Fig. 3 is in a good agreement with the nominal thickness sequence of 10 nm Ta2O5 + 8 × (10 nm HfO2 + 10 nm Ta2O5). Stable thicknesses of Ta2O5 layers have been fitted between 9.7 and 9.9 nm throughout the laminate. Also the thickness of the HfO2 layers remained, in practice, constant at 9.0 nm. The roughness of both Ta2O5 and HfO2 layers had a tendency to slightly decrease, looking from the substrate to the outermost surface, from 0.7 to 0.5 nm, and from 0.8 to 0.5 nm, respectively. For the comparison, XRR analysis of the laminate grown with a nominal thickness sequence of 5 nm Ta2O5 + 8 × (15 nm HfO2 + 5 nm Ta2O5), demonstrating a considerably higher degree of crystallinity (Fig. 2), revealed that the thickness of the Ta2O5 layers remained in the range of 6.4–6.9 nm, without a clear systematic change throughout the total laminate thickness. However, a slight change in the HfO2 thickness occurred, from 14.6 to 15.6 nm counted from the substrate to the surface. Throughout the thickness of the laminate the roughness of HfO2 layers increased from 0.3 to 1.2 nm, and for Ta2O5 layers from 0.4 to 1.4 nm. It could be seen, that the thickness of the constituent layers remained almost constant, i.e. varied in a rather narrow range throughout the thickness of the nanolaminate. A certain development in the roughness of the constituting layers may occur, i.e. roughness can increase towards the top layers of the laminate if the thickness of the crystalline HfO 2 interlayers is sufficient to allow stronger crystallization.

38


Fig. 4. A typical load–unload curve of glass substrate coated with metal oxide nanolaminate.

substrate, the nanolaminate with 10 nm Ta2O 5 + 8 × (15 nm HfO2 + 5 nm Ta 2 O 5) sub-structure and the magnetron sputtered protective platinum layer on top of all. Fig. 6 represents a close-up of the cross-section near the nanolaminate showing glass substrate below and platinum coating on top of the laminate. The measurements showed the thickness of the laminate to be 165 nm, on average.

Fig. 2. The grazing incidence X-ray diffraction patterns of ZrO2–Ta2O5 (upper panel) and HfO2–Ta2O5 (bottom panel) nanolaminates. Miller indexes of the most characteristic XRD reflections are indicated. Subscripts M, T, and C denote monoclinic, tetragonal, and cubic polymorphs, respectively. The depositions of interlayers with their nominal thicknesses are indicated as labels.

3.2. Nano-mechanical characterization of nanolaminates 3.2.1. Integrity and adhesion after indentation A typical load–unload curve obtained during indentation is shown on Fig. 4. The indentations were smooth, suggesting, for instance, no sudden pop-ins due to crack formation. No delamination was observed in samples after deposition and during storage. However, some flexural cracks have developed during indentation (Fig. 5). Focused ion beam cutting was used to explore the indentation imprint and crack cross-sections in more detail. The part of the imprint below the cutting line (Fig. 5, upper panel) was removed. The cutting revealed three distinctive structural regions — the glass

Fig. 3. Representative X-ray reflection pattern from an HfO2–Ta2O5 nanolaminate, obtained from a film grown with the complete sequence of constituent layers with nominal thicknesses 10 nm Ta2O5 + 8 × (10 nm HfO2 + 10 nm Ta2O5). The upper pattern shows the best result of the fitting.

Fig. 5. An indentation mark coated with 50 nm of platinum using magnetron sputtering prepared to be cut with focused ion beam (upper panel). The part of the indentation print below the white line was removed. The ion beam cutting (lower panel) revealed cross-section of glass substrate, deposited nanolaminate and platinum top layer.


39

Fig. 6. The microstructure of an HfO2–Ta2O5 nanolaminate cut revealing alternating layers of 15 nm of HfO2 and 5 nm of Ta2O5. The overall thickness of the nanolaminate was about 165 nm.

In Fig. 7, the cross-sections of the flexural cracks (can be seen on Figs. 5 and 6 also) are shown. 3.2.2. Evaluation of hardness and elastic modulus The comparison of hardness of single oxides and glass (Fig. 8) shows that coated samples exhibit higher hardness in comparison to bare glass and with increasing indentation depth the hardness values decrease and approach that of the substrate. The decrease is rather rapid and no plateau regions in hardness can be distinguished. This was expected due to a comparatively thin (170 nm) coated layer, which causes an early onset of the influence from the substrate. Such behavior has been observed for all coated samples. The indentation results of the films, however, are not to be treated independently of their substrates. The maximum hardness value of glass substrate was 6.70 GPa, as calculated from the indentation data. It has been recommended that indentation depth should not exceed 10% of the film thickness [18,19], but this criterion might not be sufficient in the case of harder film on a softer substrate. Thus the indentation depth that would yield a hardness value that characterizes coating with no influence from the substrate should be preferably

Fig. 7. Enlarged micrographs of the cut indentation imprint shown on Figs. 5 and 6. Top left panel shows a cross-section of the left flexural crack; top right panel shows two cracks seen on the right side on Figs. 5 and 6. The region of indentation tip is shown on the bottom panel.

Fig. 8. The hardnesses (upper panel) and elastic moduli (lower panel) of bare glass substrate, representative 160–170 nm thick (X-ray) amorphous Al2O3 and Ta2O5 films, polycrystalline ZrO2 and HfO2 films.

even considerably lower than 10% of the film thickness. Therefore, the contribution from a softer glass substrate to the hardness values should be noticeable already at 50 nm tip displacement (indentation depth) and the actual value of the film hardness is likely higher. Fig. 8 demonstrates that for Al2O3 the maximum hardness value is around 9.5 GPa, which has been recorded at the displacement of about 50 nm. This depth already is at about 30% of the total film thickness. The behaviors of elastic moduli for the single oxides (Fig. 8, lower panel) were similar to the hardness curves. Starting from about 25 nm displacement into the film the values started to drop and leveled off after penetrating few hundred nanometers in depth. The reason this happens more rapidly than in the case of hardness is because the elastic deformation reaches much further than the zone of plastic deformation meaning that the influence from the substrate is evident even sooner. It is worth noting that Al2O3 and HfO2 behaved similarly in terms of the exhibited maximum hardness, while the hardness of both ZrO2 and Ta2O5 remained clearly inferior compared to that of Al2O3 and HfO2. Also the hardness and elastic modulus of ZrO2 occurred noticeably lower compared to the other oxides. Hereinafter, the indentations are presented to show the first 300 nm of indentation depth. The rising part of all curves should be omitted due to calibration errors at very low indentation depths and therefore the measuring cannot be considered reliable. The elastic modulus for glass was around 68 GPa. In the case of metal oxide coatings on the glass, the recorded moduli were from 85 to 110 GPa. Similarly to the hardness values, also the elastic modulus of an oxide in the glass–oxide system would decrease due to the lower modulus of the glass substrate. The results of hardness and moduli and the change in respective property of the pure glass substrate compared to glass and pure metal oxide systems are summarized in Table 1.

40


Table 1 The measured hardnesses and elastic moduli maximum values (with standard deviations) of glass and glass with an oxide coating. Respective changes have been calculated to show enhancement of properties. Sample

Hardness H (st dev), GPa

Modulus of elasticity E (st dev), GPa

Change in H/E, %

Glass Al2O3/glass HfO2/glass ZrO2/glass Ta2O5/glass

6.7(0.2) 9.5(0.5) 9.1(0.7) 7.0(0.4) 6.7(0.2)

68(5) 101(4) 111(11) 86(4) 96(3)

+42/+48 +36/+63 +4/+26 0/+41

As it can be seen from Table 1, the surface properties of the substrate could be tailored to have hardness changed from 0 to 42% and modulus of elasticity could be enhanced up to 63%. The results show that the hardness, but not necessarily the moduli, decreased from aluminum oxide to tantalum oxide.. The percentages are given as pure mathematical estimation. If the measurement errors are taken into account, rather at least a 10% change in a value would be reasonably large to be considered to have some effect. The hardness values exhibited by nanolaminates of Al2O3 and Ta2O5 were higher than that of the bare glass substrates and pure Ta2O5 but lower than those of single Al2O3 but higher than that of pure Ta2O5 (Fig. 6). The results obtained from Al2O3–Ta2O5 laminates did not significantly depend on the relative thicknesses and sequence of the constituent layers. In Fig. 9, the hardness and elastic modulus of HfO2–Ta2O5 nanolaminates are depicted. From the XRD results (Fig. 2), it can be seen that the laminates were nanocrystalline, possessing multiphase composition. Regarding the

hardness, the hardest structure was evidently achieved with the thickest HfO2 interlayers (15 nm), demonstrating the highest degree of crystallization and, thus, probably containing sufficiently large and internally well-ordered dense nanocrystallites. In the nearly amorphous laminates with HfO2 layers grown to thickness of 10 nm and also in completely disordered (in accord with XRD) laminates with 5 nm thick HfO2 layers, considerably decreased hardness was measured, remaining similar to that of amorphous Ta2O5 or Ta2O5–Al2O3 described above, and eventually rapidly reaching the hardness of the substrate. The results of hardness and moduli and the change in respective property of the pure glass substrate compared to glass and Al2O3– Ta2O5 nanolaminate systems are summarized in Table 2. As it can be seen from Table 2, the surface properties of the substrate could be tailored to have hardness increased up to 10% and modulus of elasticity could be enhanced up to 44%. Also, it can be seen that consistency of relatively harder oxide in the nanolaminate could give a rise in hardness but not necessarily to the modulus of elasticity which stayed on the same level. All the HfO2–Ta2O5 nanolaminates exhibited the values of elastic moduli values in the range of 95–115 GPa, i.e. remarkably higher than that of the bare glass substrate. The curves representing elastic moduli curves of the different metal oxide laminates became quite similar to each other (Fig. 9, lower panel) as well as to those demonstrated by amorphous Al2O3 and Ta2O5. However, again the laminates with the thickest and the most intensely crystallized HfO 2 layers distinctively demonstrated the highest rise to the elasticity of the laminate-substrate stack. The results of hardness and moduli and the change in respective property of the glass substrate compared to glass and HfO2–Ta2O5 nanolaminate systems are summarized in Table 3 (Fig. 10). As it can be seen from Table 3, the surface properties of the substrate could be tailored to have hardness increased up to 10% and modulus of elasticity could be enhanced up to 62%. Similarly to Table 2, it seems that relatively higher consistency of harder oxide (here HfO2) in the nanolaminate increases the overall hardness. Somewhat similar results were obtained with the nanocrystalline ZrO2–Ta2O5 laminates (Fig. 11). The softest sample contained the thickest (15 nm) layers of Ta2O5 in the multilayered structure, i.e. the sample with the largest contribution from an amorphous oxide. The hardest sample was, however, built up using very thin layers, 2.5 nm, of both constituent oxides. On the other hand, a similar structure with the same thickness ratio of ZrO2 and Ta2O5 appeared to be about 0.5 GPa softer (Table 4). The main difference of the structures was the single oxide layer thickness of 2.5 nm versus 10 nm, respectively. Therefore, it might be concluded that thinner single layers could give rise in hardness. The comparison of moduli of the same samples does not reveal similar correlation, which suggests that, although the hardness was increased, the elastic modulus was not influence in the same way. As it can be seen from Table 4, the surface properties of the substrate could be tailored to have hardness increased up to 16% and modulus of elasticity could be enhanced up to 69%. There was no significant difference in properties if the order of oxide deposition was altered. Both, hardness and moduli remain very similar. Table 2 The measured hardnesses and elastic moduli maximum values (with standard deviations) of glass and glass with Al2O3–Ta2O5 nanolaminate coatings. Respective changes have been calculated to show enhancement of properties (the number in front of the oxide chemical formula indicates the respective single layer thickness of the oxide in nanometers).

Fig. 9. The hardnesses (upper panel) and elastic moduli (lower panel) results of Al2O3– Ta2O5 nanolaminates. For comparison, elastic moduli of single Al2O3 and Ta2O5 films and glass substrate are also depicted. The number in front of the chemical formula of respective oxide designates the nominal thickness of the oxide in nanometers.

Sample

Hardness H, GPa

Modulus of elasticity E, GPa

Change in H/E, %

Glass 10 nm Ta2O5 + 8 × (10 nm Al2O3 + 10 nm Ta2O5) 10 nm Ta2O5 + 8 × (15 nm Al2O3 + 5 nm Ta2O5)

6.7(0.2) 7.0(0.4)

68(5) 98(5)

+4/+44

7.4(0.5)

97(7)

+10/+44


41

Table 3 The measured hardnesses and elastic moduli maximum values (with standard deviations) of glass and glass with HfO2–Ta2O5 coatings. Respective changes have been calculated to show enhancement of properties. Sample



Change in H/E, %

Glass 10 nm Ta2O5 + 8 × (5 nm Al2O3 + 15 nm Ta2O5) 10 nm Ta2O5 + 8 × (10 nm Al2O3 + 10 nm Ta2O5) 10 nm Ta2O5 + 8 × (15 nm Al2O3 + 5 nm Ta2O5) 10HfO2 + 8(10Ta2O5 + 10HfO2)

6.7(0.2) 6.7(0.2) 6.8(0.2) 7.4(0.3) 7.2(0.3)

68(5) 102(4) 99(4) 110(5) 103(3)

0/+50 +1/+45 +10/+62 +7/+51

The hardness of the multilayer with the thickest ZrO2 layers (15 nm) was well comparable with the rest of the samples in the series, but did not quite match with that of the amorphous multilayers (ZrO2 thickness 2.5 nm). One can suppose that the loss in the hardness in materials consisting of nanocrystalline tetragonal/cubic ZrO2 grains in the films with 10–15 nm thick ZrO2 layers could be due to the build-up of a matrix of tiny crystallites with their defective grain boundaries sliding during indentation procedure. It has recently been observed that the hardness of the ALD-grown ZrO2 may actually decrease with the increasing thickness of the crystalline oxide [26]. For the comparison to thick “bulk” material layers, i.e. the relatively thick films, also evaluated by Berkovich indentation, one can mention that about 1 μm thick Ta2O5 films deposited by electron beam evaporation have been considered as rather soft material by Martin et al. [27]. These films exhibited hardness values of 5.3 and 6.5 GPa on glass and silicon substrates, respectively. These values could further be increased to 10 GPa by ion beam assisted deposition. Regarding thick aluminum oxide coatings, Figueiredo et al. [28] have measured ca. 3 μm thick Al2O3 films obtained by chemical vapor deposited on cemented carbide

Fig. 10. The hardnesses (upper panel) and elastic moduli (lower panel) results of HfO2– Ta2O5 nanolaminates. For comparison, elastic moduli of single HfO2 and Ta2O5 films and glass substrate are also depicted. The number in front of the chemical formula of respective oxide designates the nominal thickness of the oxide in nanometers.

(WC/Co) and achieved hardness values in the range of 20–22 GPa within the indentation depth down to 50 nm. Sartori et al. [29] have measured 1.9 μm thick ZrO2 layers grown by metal organic chemical vapor deposition from a cyclopentadienyl-based precursor and recorded certain development on steel substrates with measured hardness of 10 GPa with ZrO2 on steel and 4.5 GPa of bare steel. Venkatachalam et al. [30] have measured the hardness values 10–20 GPa of HfO2 films sputtered to thicknesses in the range of 150–200 nm on silicon substrates, and nanoindented into the depth reaching 50 nm. From Tables 2–4, several common features of coatings appeared. The first was that the harder the constituent oxides the harder the composition, which could be intuitively considered as a logical result. Secondly, the higher content of harder oxide in a coating would result in a harder structure, which again could be considered as a logical result. Thirdly, it did not matter which metal oxide was deposited first — the results were very similar if not identical.

Fig. 11. The hardnesses (upper panel) and elastic moduli (lower panel) results of ZrO2– Ta2O5 nanolaminates. The sequences of the constituent layers are described by labels. The number in front of the chemical formula of respective oxide designates the nominal thickness of the oxide in nanometers. The error bars have been removed to increase the readability.

42


Table 4 The measured hardnesses and elastic moduli maximum values (with standard deviations) of glass and glass with ZrO2–Ta2O5 coatings. Respective changes have been calculated to show enhancement of properties. Sample



Change in H/E, %

Glass 10 nm Ta2O5 + 8 × (5 nm Al2O3 + 15 nm Ta2O5) 10 nm Ta2O5 + 8 × (10 nm Al2O3 + 10 nm Ta2O5) 10 nm Ta2O5 + 8 × (15 nm Al2O3 + 5 nm Ta2O5) 10ZrO2 + 8(10Ta2O5 + 10HfO2) 7.5Ta2O5 + 16(5ZrO2 + 5Ta2O5) + 2.5Ta2O5 5Ta2O5 + 33(2.5ZrO2 + 2.5Ta2O5)

6.7(0.2) 6.7(0.3) 7.3(0.4) 7.3(0.7) 7.2(0.3) 7.2(0.3) 7.8(0.4)

68(5) 108(6) 115(4) 106(11) 113(9) 111(6) 112(4)

0/+58 +9/+69 +9/+55 +7/+66 +7/+63 +16/+63

4. Summary Al2O3–Ta2O5, HfO2–Ta2O5, and ZrO2–Ta2O5 nanolaminates grown by atomic layer deposition were analyzed by instrumented indentation. The periods in the nanolaminates comprised of double metal oxide layers with variable single layer thicknesses between 2.5 and 15 nm. The Al2O3–Ta2O5 films remained amorphous, whereas HfO2–Ta2O5 and ZrO2–Ta2O5 became nanocrystalline with HfO2 and ZrO2 layer thicknesses exceeding 5 and 2.5 nm, respectively. HfO2 layers were mixtures of the stable and metastable polymorphs. ZrO2 layers were single phase metastable structures. The hardest (9.5 GPa) single oxide film deposited in this study was Al2O3 layer on glass, while the highest modulus (111 GPa) was possessed by HfO 2 . The highest hardnesses (7.8 GPa) and moduli (115 GPa) of the laminates were achieved with the ZrO2 –Ta 2 O 5 multilayers. The nanolaminating increased considerably the surface elasticity of coated glass. The elastic moduli of nanolaminates containing Al2O3 were near 100 GPa and for nanolaminates with HfO2, the moduli were from 100 GPa to 110 GPa. The laminates with ZrO2 possessed moduli close to 110 GPa. The results showed that the substrate surface hardness could be changed from 0 to 9.5 GPa and moduli in-between 85 and 120 GPa. Considering measurement errors, the moduli of laminates with varying chemical compositions could be similar. Acknowledgments The work was partially supported by the Estonian Research Council (PUT170 and IUT2-24) and Finnish Centre of Excellence in Atomic Layer Deposition (Academy of Finland, project 251220). Mrs. Jekaterina Kozlova (MSc) is acknowledged for scanning electron microscopy imaging. Mr. Peeter Ritslaid (PhD) is acknowledged for magnetron sputtering. References [1] B.A. Latella, G. Triani, P.J. Evans, Toughness and adhesion of atomic layer deposited alumina films on polycarbonate substrates, Scr. Mater. 56 (2007) 493–496. [2] K. Tapily, J.E. Jakes, D.S. Stone, P. Shrestha, D. Gu, H. Baumgart, A.A. Elmustafa, Nanoindentation investigation of HfO2 and Al2O3 films grown by atomic layer deposition, J. Electrochem. Soc. 155 (2008) H545–H551. [3] D.C. Miller, R.R. Foster, Y. Zhang, S.-H. Jen, J.A. Bertrand, Z. Lu, D. Seghete, J.L. O'Patchen, R. Yang, Y.-C. Lee, S.M. George, M.L. Dunn, The mechanical robustness of atomic-layer- and molecular-layer-deposited coatings on polymer substrates, J. Appl. Phys. 105 (2009) 093527. [4] R. Saha, W.D. Nix, Effects of the substrate on the determination of thin film mechanical properties by nanoindentation, Acta Mater. 50 (2002) 23–38. [5] P.H. Mayrhofer, C. Mitterer, L. Hultman, H. Clemens, Microstructural design of hard coatings, Prog. Mater. Sci. 51 (2006) 1032–1114. [6] D.V. Shtansky, P.V. Kiryukhantsev-Korneev, I.A. Bashkova, A.N. Sheveiko, E.A. Levashov, Multicomponent nanostructured films for various tribological applications, Int. J. Refract. Met. Hard Mater. 28 (2010) 32–39. [7] G. Balakrishnan, A. Wasy, H.S. Ho, P. Sudhakara, S.I. Bae, J.I. Song, Study of Al2O3/ZrO2 (5 nm/20 nm) nanolaminate composite, Compos. Res. 26 (2013) 60–65. [8] W.D. Sproul, New routes for the preparation of mechanically hard films, Science 273 (1996) 889–892.

[9] Z. Chlup, H. Hadraba, L. Slabáková, D. Drdlík, I. Dlouh'y, Fracture behaviour of alumina and zirconia thin layered laminate, J. Eur. Ceram. Soc. 32 (2012) 2057–2061. [10] J.O. Carneiro, V. Teixeira, A. Portinha, S.N. Dub, R. Shmegera, Hardness evaluation of nanolayered PVD coatings using nanoindentation. Hardness evaluation of nanolayered PVD coatings using nanoindentation, Rev. Adv. Mater. Sci. 7 (2004) 83–90. [11] A. Portinha, V. Teixeira, J.O. Carneiro, S.N. Dub, R. Shmegera, C.J. Tavares, Hard ZrO2/ Al2O3 nanolaminated PVD coatings evaluated by nanoindentation, Surf. Coat. Technol. 200 (2005) 765–768. [12] M. Mishra, P. Kuppusami, V.R. Reddy, C. Ghosh, R. Divakar, R. Ramaseshan, A. Singh, R. Thirumurugesan, E. Mohandas, Influence of CeO2 layer thickness on the properties of CeO2/Gd2O3 multilayers prepared by pulsed laser deposition, Vacuum 113 (2015) 64–74. [13] E. Marin, L. Guzman, A. Lanzutti, W. Ensinger, L. Fedrizzi, Multilayer Al2O3/TiO2 atomic layer deposition coatings for the corrosion protection of stainless steel, Thin Solid Films 522 (2012) 283–288. [14] E. Marin, A. Lanzutti, L. Fedrizzi, Tribological properties of nanometric atomic layer depositions applied on AISI 420 stainless steel, Tribol. Ind. 35 (2013) 208–2016. [15] R. Matero, M. Ritala, M. Leskelä, T. Salo, J. Aromaa, O. Forsén, Atomic layer deposited thin films for corrosion protection, J. Phys. IV France 9 (1999) (Pr8-493–Pr8-499). [16] E. Härkönen, B. Díaz, J. Światowska, V. Maurice, A. Seyeux, M. Vehkamäki, T. Sajavaara, M. Fenker, P. Marcus, M. Ritala, Corrosion protection of steel with oxide nanolaminates grown by atomic layer deposition, J. Electrochem. Soc. 158 (2011) C369–C378. [17] R. Raghavan, M. Bechelany, M. Parlinska, D. Frey, W.M. Mook, A. Beyer, J. Michler, I. Utke, Nanocrystalline-to-amorphous transition in nanolaminates grown by low temperature atomic layer deposition and related mechanical properties, Appl. Phys. Lett. 100 (2012) 191,912. [18] H. Mohseni, T.W. Scharf, Atomic layer deposition of ZnO/Al2O3/ZrO2 nanolaminates for improved thermal and wear resistance in carbon–carbon composites, J. Vac. Sci. Technol. A 30 (2012) 01A149. [19] S.J. Bull, Size effects in the mechanical response of nanoscale multilayer coatings on glass, Thin Solid Films 571 (2014) 290–295. [20] S.J. Bull, Elastic properties of multilayer oxide coatings on float glass, Vacuum 114 (2014) 150–157. [21] K. Kukli, J. Ihanus, M. Ritala, M. Leskelä, Tailoring the dielectric properties of HfO2– Ta2O5 nanolaminates, Appl. Phys. Lett. 68 (1996) 3737–3739. [22] K. Kukli, J. Ihanus, M. Ritala, M. Leskelä, Properties of Ta2O5-based dielectric nanolaminates deposited by atomic layer epitaxy, J. Electrochem. Soc. 144 (1997) 300–306. [23] J. Hay, Introduction to instrumented indentation testing, Exp. Tech. (November/ December 2009) 66–71. [24] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented indentation: advances in understanding and refinements to methodology, J. Mater. Res. 19 (2004) 3–20. [25] T. Suntola, Atomic layer epitaxy, Thin Solid Films 216 (1992) 84–89. [26] M.A. Mamun, H. Baumgart, A.A. Elmustafa, ALD zirconium oxide (ZrO2) thin films mechanical/structural properties for high-tech applications, ECS J. Solid State Sci. Technol. 4 (2015) Q35–Q37. [27] P.J. Martin, A. Bendavid, M.V. Swain, R.P. Netterfield, T.J. Kinder, W.G. Sainty, D. Drage, Mechanical and optical properties of thin films of tantalum oxide deposited by ion-beam assisted deposition, Mater. Res. Soc. Symp. Proc. 308 (1993) 583–588. [28] M.R. de Figueiredo, M.D. Abad, A.J. Harris, C. Czettl, C. Mitterer, P. Hosemann, Nanoindentation of chemical-vapor deposited Al2O3 hard coatings at elevated temperatures, Thin Solid Films 578 (2015) 20–24. [29] A. Sartori, N. El Habra, C. De Zorzi, S. Sitran, M. Casarin, G. Cavinato, C. Sada, R. Gerbasi, G. Rossetto, Zirconocene alkoxides, promising precursors for MOCVD of zirconium dioxide thin films, Chem. Vap. Depos. 18 (2012) 151–158. [30] D.K. Venkatachalam, J.E. Bradby, M.N. Saleh, S. Ruffell, R.G. Elliman, Nanomechanical properties of sputter-deposited HfO2 and HfxSi1 − xO2 thin films, J. Appl. Phys. 110 (2011) 043527.