Materials and Engineering A359 (2003) 112 /118 www.elsevier.com/locate/msea Nanoindentation measurements of combustion CVD Al2O3 and YSZ films D.W. Stollberg a, J.M. Hampikian a,*, L. Riester b, W.B. Carter a a School of Materials Science and Engineering, Georgia Institute of Technology, 771 Ferst Drive N.W., Atlanta, GA 30332-0245, USA b High Temperature Materials Laboratory, Oak Ridge National Laboratories, Oak Ridge, TN 37831-6069, USA Received 2 May 2002; received in revised form 7 April 2003 Abstract Combustion chemical vapor deposition (combustion CVD) was used to deposit thin films of yttria stabilized zirconia (YSZ) and alumina (Al2O3). Nanoindentation analysis for hardness, modulus and fracture toughness of the films as well as several bulk materials, including single crystal Al2O3, MgO, SiO2, YSZ and polycrystalline Al2O3 was conducted. YSZ films that were produced with a total cation molarity of 0.005 M possessed significant surface roughness that was not conducive to nanoindentation measurements. The lower concentration investigated, 0.002 M provided a repeatable deposition condition that produced thin films with consistent hardness and modulus values. Load-displacement measurements at very low loads (up to total displacements of 200 nm) showed that the Al2O3 films tended to plastically deform, whereas the YSZ films showed less of a tendency to do so. For Al2O3 films that were 0.389/0.09 mm thick, the modulus, hardness and fracture toughness values were 28.69/1.6 GPa, 4799/15 GPa and 2.229/0.31 MPa m0.5, respectively; for YSZ films that were 0.659/0.05 mm thick, the values were 16.19/4.6 GPa, 3889/89 GPa and 1.679/0.46 MPa m0.5, respectively. These values were found to be consistent with the values measured via nanoindentation in this work for bulk samples of Al2O3 and YSZ, as well as to those reported in the literature. # 2003 Elsevier B.V. All rights reserved. Keywords: Nanoindentation; Combustion CVD; Combustion chemical vapor deposition; Alumina; Al2O3; Yttria stabilized zirconia; YSZ; Modulus; Hardness; Fracture toughness 1. Introduction The technique of measuring the mechanical properties of materials at very small scales is known as nanoindentation, or ultralow load indentation [1 /7]. This technique is used routinely to characterize the mechanical properties of thin films, and Pharr has conducted a recent review of the measurement of mechanical properties by ultralow load indentation [8]. From nanoindentation measurements, both the hardness and the elastic modulus may be extracted from load versus displacement curves; in this work the methodologies used are those described by Doerner and Nix [3] and Loubet et al. [2] but without the constraint of the flat punch indenter geometry [7]. This analysis can yield hardness * Corresponding author. Fax: /1-404-894-9140. E-mail address: janet.hampikian@mse.gatech.edu Hampikian). (J.M. 0921-5093/03/$ - see front matter # 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0921-5093(03)00339-3 (H) and elastic modulus (E ) values with accuracies better than 10% [7]. The fracture toughness, or amount of stress a cracked material can endure without failing, may also be extracted from ultralow load indentation testing [9]. The method is based on the radial cracking that occurs when brittle materials are indented by a sharp indenter. Lawn et al. [10] have shown that a simple relationship exists between the fracture toughness, Kc , and the lengths of the radial cracks, c : Kc a 1=2 E P H c3=2 (1) where P is the peak indentation load, a an empirical constant that depends on the geometry of the indenter (0.032 for the cube-corner shaped diamond stylus used in this work), H the hardness, and E the Young’s modulus. D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 The near-surface mechanical properties of hardness, Young’s modulus and fracture toughness for thin films of alumina (Al2O3) and yttria stabilized zirconia (YSZ) that were deposited via combustion chemical vapor deposition (combustion CVD) as well as bulk ceramics were examined using nanoindentation. The bulk ceramics examined included single crystal Al2O3, polycrystalline Al2O3, YSZ (9.5 mol%Y2O3 /ZrO2), amorphous silica and single crystal MgO. The objective of this research was to compare the properties of the combustion CVD films grown on Al2O3 with bulk properties of Al2O3 and YSZ, and to also evaluate the overall suitability of nanoindentation to assess films with a degree of surface inhomogeneity. 2. Experimental The NanoIndenterTM II (Nano Instruments Inc., Knoxville TN, USA) [1,3] located at the High Temperature Materials Laboratory of Oak Ridge National Laboratory in Oak Ridge, TN produced load versus indentation depth plots, using the trigonal Berkovich indenter with an initial tip radius of approximately 40 nm. The instrument was operated in the continuous stiffness mode. For H and E measurements, an array of 15 indentations in a 5 /3 matrix with a 15 mm spacing was selected. The samples were loaded to a total displacement of approximately 200 nm, with hold segments at 50 and 100 nm. For fracture toughness, indentations were made at various loads, in a 5 /5 matrix consisting of five indentations each at 400, 250, 100, 50 and 25 mN of load. The average crack length was determined from the three cracks formed from each of the twenty-five indents made. The experimental uncertainty in the data is the standard deviation of the calculated fracture toughness. In accordance with the expression developed by Lawn et al. [10] not all cracks formed by an indent were used; only well-developed radial cracks (lateral cracks and cracks shorter than approximately two times the indent size were excluded). A Hitachi S-800 scanning electron microscope was used to measure crack lengths. The bulk ceramic specimens that were tested are listed in Table 1. The single crystal alumina, MgO, silica and YSZ samples were 99.99% pure while the polycrystalline Al2O3 had a 99.6% purity level. All specimens were 0.5 mm thick except the YSZ which was 1 mm thick. All bulk specimens were 1 cm by 1 cm in area and polished by the manufacturer to a roughness of less than 2 nm Rq rms. Each sample was ultrasonically cleaned in acetone followed by an isopropanol rinse prior to nanoindentation. The thin film samples of Al2O3 and 4.5 mol% YSZ were prepared via liquid-fuel combustion CVD, using toluene as fuel and precursor solvent, and yttrium and 113 zirconium 2-ethylhexanoate as precursors for YSZ films, with aluminum acetylacetonate as the precursor for Al2O3 films. The substrate in both cases was a-plane Al2O3, and the deposition temperature 1550 8C. Details on the synthesis process conditions are provided elsewhere; in brief, the deposition technique involves burning the precursor-containing fuel in the open atmosphere, and depositing films on substrates positioned in or near the flame [11,12]. 3. Results and discussion YSZ samples were deposited with two different total cation molarities, 0.002 and 0.005 M, and characterized by plan view SEM, see Fig. 1a/b. Surface topography is strongly affected by solution precursor concentration, with smoother films observed at the lower concentration. Nanoindentation experiments that were conducted for both concentrations of YSZ material showed a large discrepancy between their load /displacement curves. The higher concentration samples displayed poor load / displacement curves in that after discounting the highest and lowest maximum loads, the maximum load to cause 200 nm of displacement varied between approximately 0.35 nm and 1.5 mN, corresponding to unrealistic values of hardness and modulus compared with bulk properties for this material. By contrast, the lower concentration samples required in excess of 8 mN of load for 12 of the 15 nanoindentations; and for these load displacement curves, the average maximum load at 200 nm displacement was 15 mN, with the average of the other three curves being only 0.7 mN. It was concluded that the surface roughness of the higher concentration samples was prohibitive to conducting meaningful nanoindentation work, and thus nanoindentation results from only the 0.002 M samples are presented below. Even at the lower concentration, it was necessary to scrutinize each load displacement curve, as the films produced by combustion CVD still had sufficient surface asperities as to cause approximately 20% of the load displacement curves to be erroneously low. Finally, as the film quality was higher for YSZ grown at 0.002 M, the Al2O3 films investigated in this work were grown from a precursor concentration of 0.002 M only. Plan view and cross-sectional SEM images of Al2O3 and YSZ deposited with 0.002 M precursor solutions are shown in Fig. 1. In both cases, thin films of Al2O3 and YSZ samples are produced, with thicknesses corresponding to 0.389/0.09 and 0.659/0.05 mm, respectively. The Al2O3 film is difficult to discern in cross-section in Fig. 1c, as the phase deposited is a-Al2O3, the same as the substrate. X-ray diffraction of the Al2O3-coated substrate (not shown) does not reveal the presence of any extra peaks from the Al2O3 coating (peaks from the Al2O3 substrate only were present), consistent with the 114 D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 Table 1 Mechanical properties of substrates and films, values obtained in this work compared with values obtained from literature Material Al2O3 SiO2 MgO YSZ (9.5mol% Y2O3) YSZ Film Al2O3 Film Hardness (GPa) Elastic Modulus (GPa) Fracture Toughness MPa m0.5 This Work Literature This Work Literature This Work Literature 30.99/1.23 8.79/0.3 8.49/1.7 199/0.2 16.19/4.6 28.69/1.6 25 [9] 8.3 [1] 7 [23] 13 [5] 4819/16 72.29/1.6 2909/38 3099/6 3889/89 4799/15 433 [1] 72.4 [22] 300 [16] 200 [5] 2.39/0.6 0.709/0.17 1.29/0.2 1.309/0.29 1.679/0.46 2.229/0.31 2.2 [15] 0.58 [1] 1.77 [24] 2.8 [17] smoothness of the Al2O3 films as compared with the YSZ films that were deposited, see Fig. 1c /f. A feature that was commonly noted on the Al2O3 films, see Fig. 2, was the presence of ‘‘pop-ins,’’ displacement discontinuities in the load /displacement response due to plasticity [13]. These features are commonly exhibited on hard, stiff materials such as Al2O3; of note is the absence of significant pop-ins on the uncoated Al2O3. As an artifact of the loading sequence, pop-ins were commonly seen immediately prior to the 50 and 100 nm hold segments, as seen in Fig. 2. They were also observed at other displacements. It should be noted that during the constant load hold segment corresponding to approximately 50 nm displacement, a very slow time-dependent strain rate was observed, 0.5% of the total penetration for the 16 s hold segment. The average load required for 50 nm displacement is 2.29/0.3 mN, which correlates approximately to a loading rate of / 0.14 mN s 1. Krell et al. [14] have also observed kinetic effects associated with their measurements on sub-mm polycrystalline Al2O3. In their work, nanoindentation experiments with loading to 200 mN on polycrystalline Al2O3 showed 0.3 /2% of the total penetration within 5 s after loading with 4.4 mN s 1 and less than 0.5% within 30 s after slow loading with 0.44 mN s 1. 3.1. Hardness and elastic modulus Fig. 1. Scanning electron microscope images of films deposited via combustion CVD: (a) plan view image obtained from 0.002 M total cation precursor concentration; (b) plan view image from 0.005 M concentration; (c /d) plan view and cross-section from 0.002 M Al2O3 film, (e /f) plan view and cross-section from 0.002 M YSZ film. formation of the a-Al2O3 phase. The presence of some surface roughness even at this lower concentration is noted, see for example Fig. 1f. Representative load /displacement curves (P /d) for the 0.002 M Al2O3 and YSZ coatings, as compared with a-plane Al2O3 substrate are shown in Fig. 2. Unlike the YSZ films, all 15 of the load-displacement curves for Al2O3 showed consistent load displacement curves with loads that ranged between approximately 22 to 24 mN at 200 nm displacement. This is likely due to the relative Fig. 3a shows the results of the hardness analysis versus the indentation depth for the materials tested. In total, the a-plane Al2O3 was tested six times, inserted at a lightly different ‘‘rotation’’ each time; the highest and lowest values obtained are shown in Fig. 3a. For clarity, the other four plots are not shown. The hardness of polycrystalline alumina and r-plane alumina, which were tested for comparison purposes, fell approximately within the range of the a-plane Al2O3 data, see Fig. 3a. The average hardness values for a-plane Al2O3 are presented in Table 1. Polycrystalline Al2O3 and r-plane Al2O3 were tested for comparison purposes, see Fig. 3. The hardness values for combustion CVD coatings of Al2O3 and YSZ were also evaluated; these results are presented in Fig. 3a and in Table 1. It is seen that the D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 115 Fig. 2. Load (mN) vs. displacement (nm) curves obtained from a-plane Al2O3, Al2O3 film and YSZ film. Note presence of ‘‘pop-ins’’ for Al2O3 film curves. alumina coating has hardness values that lie slightly below that of the bulk material. The hardness values for the combustion CVD-deposited YSZ coating, bulk YSZ, MgO and silica are also shown in Fig. 3a and also summarized in Table 1. The hardness values for MgO and silica are approximately the same; bulk YSZ has an average hardness value that is between that of alumina and silica/MgO, as expected Fig. 3. Hardness (a) and elastic modulus (b) plotted as a function of indentation depth for the Al2O3 and YSZ films as well as bulk materials including r-plane, a-plane and polycrystalline Al2O3, bulk ZrO2 (YSZ), SiO2 and MgO. D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 116 from literature values [1,5,9,23]. The hardness values for YSZ coatings are seen to be consistent with the values seen for bulk YSZ, see Fig. 3a. The YSZ films hardness values have error bars that are quite large, due to the large variation in load /displacement that is present with the combustion CVD deposited film. This is likely due to the surface roughness present on these films, although the coating may also not be fully dense. Since combustion CVD is an open air deposition process, with numerous variables in play at one time in deposition (e.g. nozzle conditions, oxidant flow rate, fuel flow rate, temperature, etc.), there was some concern about the repeatability of the process to yield films of similar properties. Therefore, five different films of YSZ were deposited on separate days and subsequently tested. Scanning electron microscopy of the surfaces of the films showed some differences in the surface topography for the materials, for example, see Fig. 1a /e. However, the hardness and elastic modulus values from the five YSZ coatings were found to be statistically the same, using the t-test, see Table 2. The error bars for each of the five different depositions are based on the different values obtained from each of the fifteen indents for that specimen, and are quite large relative to the substrates. The elastic moduli for the six orientations of a-plane Al2O3, the r-plane Al2O3 and the polycrystalline Al2O3 are not statistically different, see Fig. 3b. The average modulus for the six orientations of a-plane alumina was found to be 4819/16. The YSZ and MgO elastic moduli were approximately the same, with the value for bulk silica being somewhat lower, as summarized in Table 1. The modulus for the combustion CVD YSZ film had slightly higher values, but within experimental uncertainty of the bulk YSZ measurements, except at the highest load. The 200 nm displacement corresponds to approximately one third the YSZ film thickness (6509/ 50 nm), and thus the higher modulus seen at this load might be slightly influenced by the substrate. The hardness and elastic bulk modulus values measured via nanoindentation were close to their bulk material literature values, obtained from large scale testing as summarized in Table 2. For bulk hardness measurements, Vickers, Knoop, Brinell or Rockwell testing equipment was used, and for bulk elastic Table 2 Repeatability of nanoindentation results for five 100% YSZ films deposited via combustion CVD on different occasions Sample H (MPa) E (GPa) Kc (MPa m0.5) 1 2 3 4 5 15.29/2.5 15.89/1.4 15.19/2.3 15.39/1.8 13.89/5.0 3209/43 3409/35 3469/56 3289/38 3009/101 1.599/0.48 1.939/0.48 1.959/0.31 1.569/0.31 1.739/0.48 modulus values, tensile tests or three point bend tests were used [15 /17]. Results from other researchers performing nanoindentation on similar ceramics also agree with these hardness and elastic modulus values. For example, Oliver and Pharr [7] used a Berkovich indenter producing results on sapphire of /30 GPa for hardness and 450 GPa for elastic modulus and for silica, 9 /10 GPa for hardness and 70 /80 GPa for the elastic modulus. Boudoukha et al. [18] found hardnesses and elastic moduli in similar ranges as those reported here for polycrystalline alumina: hardness 20/40 GPa and elastic modulus 300 /500 GPa. For sapphire their results were somewhat higher: hardness 40/50 GPa and elastic modulus 500 /600 GPa. Using a Vickers indenter tip, Krell and Schadlich [14] measured the hardness of polycrystalline alumina and single-crystal sapphire to be 25/30 GPa. Yttria-stabilized zirconia was measured by Voevodin et al. [19] as 25 GPa for the hardness and 300 GPa for the elastic modulus, also agreeing with the results in this report. 3.2. Fracture toughness The results of the fracture toughness testing are shown in Table 1 and Fig. 4. All of the a-plane substrates produced very similar toughness values, / 2.3 MPa m0.5, within one standard deviation of each other, regardless of the orientation of the indenter. The r-plane Al2O3 had a significantly higher measured toughness than a-plane and polycrystalline Al2O3, 3.2 versus 2.3 MPa m0.5. The alumina film grown by combustion CVD had comparable fracture toughness values to bulk values, at 2.2 MPa m0.5. The error bars associated with fracture toughness, shown in Fig. 4, are based on uncertainty in crack length used to calculate fracture toughness (Eq. (1)). The YSZ film grown by combustion CVD also showed fracture toughness values that are comparable to values measured by nanoindentation on bulk YSZ in this work. Table 1 shows a comparison of the nanoindentation fracture toughness values for bulk materials (Al2O3, SiO2, MgO, YSZ) measured in this work to the fracture toughness from bulk materials from the literature. The nanoindentation results for bulk materials in this work are similar to the toughness values obtained by larger scale testing. All of the ceramics produced toughness values with experimental uncertainties on the order of 9/ 0.5 MPa m0.5, that were similar to fracture toughness values reported in the literature [9,20,21]. Silica’s nanoindentation fracture toughness, 0.70 MPa m0.5, was higher than the bulk value of 0.58 MPa m0.5. By contrast, the toughness for MgO was significantly lower for nanoindentation, 1.2 MPa m0.5, compared with the bulk value, 1.77 MPa m0.5. Similarly for bulk YSZ, the fracture toughness obtained via nanoindentation was 1.3 MPa m0.5 compared with reported literature value of D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 117 Fig. 4. Fracture toughness values obtained from this work for a-plane, r-plane, polycrystalline Al2O3, Al2O3 film, YSZ, YSZ film, MgO and SiO2. 2.8 MPa m0.5. The variations in fracture toughness measured via nanoindentation compared with fracture toughness reported in the literature are likely due to differences in the near surface regions of the material. For example, a surface with compressive residual stress would act to close cracks and result in higher fracture toughness (silica), whereas residual tensile stress in the surface would result in lower fracture toughness (YSZ and MgO). 4. Summary and conclusions Combustion-CVD thin films that are deposited at a precursor concentration of 0.002 M grow smoother films that are better suited for nanoindentation measurements than films grown at the higher concentration of 0.005 M. Repeatability tests on five different combustion CVD-deposited thin films show consistent values of hardness and modulus, although larger error bars are associated with combustion CVD film properties as compared with those found from bulk materials. Al2O3 films exhibited significant plastic deformation (pop-ins) in comparison with the YSZ films that were deposited. For Al2O3 films that were 0.389/0.09 mm thick, the modulus, hardness and fracture toughness values were 28.69/1.6 GPa, 4799/15 GPa and 2.229/0.31 MPa m0.5, respectively; for YSZ films that were 0.659/0.05 mm thick, the values were 16.19/4.6 GPa, 3889/89 GPa and 1.679/0.46 MPa m0.5, respectively. These values are consistent with the values measured via nanoindentation in this work for bulk samples of Al2O3 and YSZ, as well as to those reported in the literature. Acknowledgements This material is based upon work supported by the National Science Foundation under Grant No DMR9624927. The nanoindentation work was research sponsored by the Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Transportation Technologies, as part of the High Temperature Materials Laboratory User Program, Oak Ridge National Laboratory, managed by Lockheed Martin Energy Research Corp. for the US Department of Energy under contract number DE-AC05-96OR22464. This research was conducted with the support of the US Department of Energy, under Cooperative Agreement No. DEFC21-92MC29061. However, any opinions, findings, conclusions, or recommendations expressed herein are those of the authors and do not necessarily reflect the views of the DOE. References [1] J.B. Pethica, R. Hutchings, W.C. Oliver, Phil. Mag. A48 (1983) 593. [2] J.L. Loubet, J.M. Georges, J.M. Marchesini, G. Meille, J. Tribol. 106 (1984) 43 /48. [3] M.F. Doerner, W.D. Nix, J. Mater. Res. 1 (4) (1986) 601 /609. [4] W.C. Oliver, Mater. Res. Soc. Bull. 11 (1986) 15 /19. [5] W.D. Nix, Metall. Trans. 20A (11) (1989) 2217 /2245. [6] G.M. Pharr, W.C. Oliver, Mater. Res. Soc. Bull. 17 (1992) 28. [7] W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (6) (1992) 1564 /1583. [8] G.M. Pharr, Mat. Sci. Eng. A 253 (1998) 151 /159. [9] D.S. Harding, W.C. Oliver, G.M. Pharr, Mater. Res. Soc. Symp. Proc. Thin Films: Stresses Mech. Prop. V 356 (1995) 663 /668. [10] B.R. Lawn, A.G. Evans, D.B. Marshall, J. Am. Ceram. Soc. 63 (1980) 574. [11] J.M. Hampikian, W.B. Carter, Mat. Sci. Eng. A 267 (1999) 7. [12] D. Stollberg, W.B. Carter, J.M. Hampikian, et al. Paper Two. 118 D.W. Stollberg et al. / Materials and Engineering A359 (2003) 112 /118 [13] T.F. Page, W.C. Oliver, C.J. McHargue, J. Mater. Res. 7 (2) (1992) 450 /473. [14] A. Krell, S. Schadlich, Mater. Sci. Eng. A 307 (2001) 172 /181. [15] G.M. Pharr, D.S. Harding, W.C. Oliver, Mechanical Properties and Deformation Behavior of Materials Having Ultra-Fine Microstructures, in: M. Nastasi, D.M. Parkin, H. Gleiter, (Eds.), NATO ASI Series, Series E: Applied Sciences, vol 233, 1993, pp 449 /462. [16] Matroc, Morgan Advanced Ceramics, 582 Monastery Dr, Latrobe, PA 15650. [17] Guide to Engineered Materials, Advanced Materials & Processes, vol 156, sixth ed., in: M.W. Hunt (Ed.), Materials Park, OH, 1999, p. 143. [18] L. Boudoukha, F. Halitim, S. Paletto, G. Fantozzi, Ceram. Int. 24 (1998) 189 /198. [19] A.A. Voevodin, J.J. Hu, J.G. Jones, T.A. Fitz, J.S. Zabinski, Thin Solid Films 401 (2001) 187 /195. [20] A.G. Evans, A.H. Heuer, D.L. Porter, Fracture, I (TCF4), Waterloo, Canada, June 19 /24, 1977, pp 529 /556. [21] N. Claussen, J. Steeb, R.F. Pabst, Am. Ceram. Soc. Bull. 56 (6) (1977) 559 /562. [22] W.D. Kingery, H.K. Bowcn, D.R. Uhlmann, Introduction to Ceramics, Wiley, New York, NY, 1976. [23] Atlantic Equipment Engineers, 13 Foster St. Bergenfield, NJ, 07621. [24] D.G. Isaak, O.L. Anderson, T. Goto, Phys. Chem. Miner. 16 (1989) 704 /713.