Deposition of TiN/NbN Superlattice Hard Coatings by Ionised Magnetron Sputter Deposition

Transcription

1 Deposition of TiN/NbN Superlattice Hard Coatings by Ionised Magnetron Sputter Deposition Yi Long, Robert J. Stearn, Zoe H. Barber, Stephen J. Lloyd, William J. Clegg Department of Materials Science and Metallurgy, University of Cambridge, Pembroke Street, CB2, 3QZ, UK. ABSTRACT We have investigated the effect of ion bombardment on the structure and hardness of thin coatings of TiN/NbN multilayered structures and monolithic films of both TiN and NbN. A radio frequency coil was used to generate an additional inductively coupled plasma between the substrate and the target enabling the sample to be bombarded by a high flux of relatively low energy ions under the appropriate conditions. It is shown that the effect of such bombardment in the case of the monolithic films is to reduce the porosity. This gave an increase in the hardness of both the TiN and the NbN films up to a power of 100 W (using a coil with a cross-sectional area of mm 2 ). Further increasing the power density led to a decrease in hardness. TiN/NbN multilayer coatings were made under the optimum deposition conditions for the monolithic materials and gave hardnesses greater than those observed in either TiN or NbN and approximately 50% greater than that predicted by a mixtures rule. INTRODUCTION It has been shown that substantial increases in hardness, sometimes up to a factor of 3, can be achieved by making layered structures in which the layers have a thickness of the order of 10 nm [1]. Some recent observations have also shown that layers of a TiN/NbN multilayer film embedded within a TiN matrix stay relatively intact during mechanical deformation with no large slip bands, whereas the deformation of the surrounding, monolithic TiN occurred by the movement of dislocations along persistent slip bands [2]. Theories proposed to explain this apparent strengthening include models based on differences in the elastic properties or in Burgers vectors of the layers [3]. However, there are many instances where no such hardening is observed, even though the structures appear to be the same, suggesting either that some other effect is leading to weakening or that the hardening effect does not arise as has been suggested [4,5]. Previous work using d.c. magnetron sputtering showed that substantial increases in hardness could be obtained by bombarding the growing film with more energetic ions, although there was no observation of any multilayer hardening [5]. In this paper we report the effect of using a radio frequency (r.f.) coil within the chamber to generate an additional plasma so that the surface of a biased substrate might be bombarded with relatively low energy ions, during the growth, and the resulting changes in structure and hardnesses of multilayer TiN/NbN and monolithic films has been studied. EXPERIMENTAL DETAILS Single TiN and NbN layers and epitaxial multilayer TiN/NbN films were deposited using ultrahigh-vacuum reactive ionised d.c. magnetron sputtering in an Ar 40% N 2 gas mixture at a gas pressure of 2.3 Pa onto (001)-oriented MgO single crystals heated to 800 C. A three turn r.f.

2 coil made of copper and having a square cross-section with an area of mm 2, with a coil spacing of 10 mm, was used to generate the additional plasma. The substrate holder was biased at 100 V. The distance between substrate and magnetron target was 45 mm, whilst that between the upper coil and the target was 20 mm. The effect of ion bombardment was studied by varying the power to the r.f. coil from 0 to 150 W. Having established the optimum deposition conditions for TiN and NbN, multilayer films were made by moving the substrates alternately under Ti and Nb targets to obtain multilayers with a wavelength varying between 2 and 15 nm and with a thickness fraction of TiN of 0.3. This range of wavelengths was chosen as work elsewhere had indicated that a maximum hardness occurred for a multilayer with wavelength of about 7 nm [1]. The hardness was determined by nanoindentation (Micro Materials Ltd, NanoTest 600), using the analysis of Oliver and Pharr [6]. The crystallographic texture of the film was determined using X-ray diffraction (XRD), whilst the structure was studied using cross-sectional transmission electron microscopy (TEM), using sample preparation techniques described elsewhere [7]. RESULTS AND DISCUSSION Deposition of monolithic layers Initial experiments were carried out to establish the most suitable conditions for deposition of monolithic TiN and NbN layers. Increasing the r.f. power to the coil from 0 to 150 W increased the rate of film deposition, the increase being approximately 20% for the TiN but 300% for the NbN. There was no effect of increasing the substrate bias on the deposition rate. The r.f. power also influences the crystallographic texture of the film. At zero r.f. power XRD shows that there is a small (111) peak as well as the dominating (002) peak. This (111) peak disappeared as the r.f. power was increased. It is thought that this occurred because under conditions of low adatom mobility the preferential orientation is determined by the crystallographic direction in which the fastest growth is possible, which is (111). When the adatom mobility is increased by ion bombardment, then the thermodynamically preferred orientation prevails. Observations of the shapes of voids in growing films suggest that the low energy surfaces are {100}, consistent with calculations in the rocksalt-structured TiC [8]. The in-plane shape of the columnar grains is also influenced by the ion bombardment. If the r.f. power is zero, then the columns of TiN have a lateral dimension of approximately 100 nm, (figure 1a), consistent with images of the surface using atomic force microscopy. Increasing the r.f. power to the coil reduces the column diameter (figure 1b), although at the highest power some larger diameter columns are again observed. In the NbN (figure 1c), the columnar grains are much smaller than those in the TiN. However, by far the most striking change induced by the ion bombardment is the very substantial reduction, although not complete removal, of the porosity. The structure of the TiN film grown under zero r.f. power (figure 1a), contains very porous regions between the TiN columns. However in the TiN film grown under ion bombardment at an r.f. power of 100 W the volume fraction of this intercolumnar porosity has been greatly reduced. The pore volume fraction is negligible in the lower part of the films, with cuboidal pores being observed in the

3 (b)tin 100W (a) TiN 0W column width (c)nbn 100W 50 nm porosity 50 nm MgO (001) Figure 1. Bright field TEM images of (a) TiN film deposited using an r.f. power = 0 W, bias = -100 V (b) TiN film with r.f. power = 100 W, bias = -100 V, and (c) NbN film with r.f. power = 100 W, bias = -100 V. These images were taken at a small defocus to enhance the visibility of the porosity by Fresnel contrast. 30 Hardness (GPa) TiN, bias = 0 V TiN, bias = -100 V NbN, bias = 0 NbN, bias = -100 V r.f. power Figure 2. The variation in hardness of the TiN and NbN monolithic layers with r.f. power to the coil.

4 middle. At the surface of the film the pores are more elongated and have a lateral width of about 10 nm. A similar effect is seen in the NbN film, where the elongated pores at the substrate/film interface are approximately 3 nm, increasing to about 6 nm at the film surface. These changes in structure are also reflected in the variations in the hardness observed for the two materials with different r.f. power densities. As the porosity is reduced by ion bombardment and the columnar width is reduced, the hardness increases (figure 2). Furthermore the greater changes of structure in the TiN, compared to NbN, are reflected in a greater increase in the hardness. However at the highest r.f. power the hardness decreases, with a maximum being observed at 100 W for both TiN and NbN. This is consistent with observations elsewhere that under more severe bombardment resputtering can cause the film to become porous again [9]. Structure and hardness of TiN/NbN superlattice structures From the experiments above it is clear that the optimum conditions for depositing both TiN and NbN occurs (under an applied substrate bias of 100 V) at an r.f. power of 100 W. NbN and TiN were alternately deposited under these conditions to give superlattice structures with wavelengths, Λ, varying from 2 to 15 nm. The structure of a multilayer with Λ of 11.6 nm is shown in figure 3. The alternate TiN and NbN layers are clearly visible, although the individual layers appear wavy rather than being flat. The growth is columnar with the column boundaries being relatively parallel and extending continuously through both layers. The width of the columns is approximately 40 nm, and similar to the values for TiN grown under the same conditions. XRD shows that, like the monolithic films, they have an (002) texture. 100 nm TiN buffer layer MgO (001) Figure 3. Bright field TEM image of a TiN /NbN multilayer with a wavelength of 11.6 nm deposited at a substrate bias of 100 V and an r.f. power to the coil of 100 W. The lighter layers are TiN.

5 Hardness (GPa) TiN NbN Λ, nm Figure 4. Hardness of TiN/NbN multilayers as a function of multilayer wavelength, Λ. Also shown are measured values of TiN and NbN monolithic films grown under the same conditions (substrate bias of 100 V and r.f. power to the coil of 100 W), and the hardness predicted from the rule of mixtures. The error in the hardness measurements is ±0.5 GPa. The hardness of the modulated structures is greater than that of either TiN or NbN alone, the value at Λ = 12 nm being approximately 50% greater than the hardness based on a rule of mixtures, (figure 4). However it is also clear from figure 4 that there is relatively little effect of the modulation wavelength over the range studied here. The hardnesses observed here are about 20% greater than those observed in earlier work in which conventional d.c. magnetron sputtering was used. There it was suggested that the hardness of the multilayer films was controlled by the porosity on the columnar boundaries [5]. While remaining columnar the films here are harder because there is much less porosity at the column boundaries. CONCLUSIONS The density and hardness of sputtered films of TiN and NbN can be improved by ion bombardment using ionised magnetron sputter deposition operating at intermediate power levels. TiN/NbN multilayers have been successfully deposited under conditions at which monolayers of both materials had their greatest hardness. The variation of the hardness with wavelength, however, appears to be relatively small, although a maximum value of 29 GPa was greater than that of either material and about 50% greater than that predicted by the rule of mixtures. Comparison with the structure and hardness of similar multilayers grown with conventional d.c. magnetron sputtering [5] suggests that it is porosity at column boundaries, rather than simply a columnar structure that weakens the multilayer.

6 ACKNOWLEDGEMENTS We thank Y. Kusano, now at the Risoe Laboratory, Denmark, for help in setting up the r.f. apparatus. We are grateful for financial support from the Overseas Research Scholarship and Cambridge Trust, the Royal Society and the DTI through the project Multilayer Engineering Materials. REFERENCES 1. M. Shinn, L. Hultman and S.A. Barnett, J. Mater. Res. 7, 901 (1992). 2. S.J. Lloyd and J.M. Molina-Aldareguia, Phil. Trans. Roy. Soc. Lond. A, in press (2003). 3. J.S. Koehler, Phys. Rev. B 2, 547 (1970). 4. H. Ljungcrantz, M. Odén, L. Hultman, J.E. Greene and J.-E. Sundgren, J. Appl. Phys. 80, 6725 (1996). 5. J.M. Molina-Aldareguia, S.J. Lloyd, M. Odén, T. Joelsson, L. Hultman and W.J. Clegg, Phil. Mag. A 82, 1983 (2002). 6. W.C. Oliver and G.M. Pharr, J. Mater. Res. 7, 1564 (1992). 7. S.J. Lloyd, J.M. Molina-Aldareguia and W.J. Clegg, Phil. Mag. A 82, 1963 (2002). 8. L. Hultman, L.R. Wallenberg, M. Shinn and S.A. Barnett, J. Vac. Sci. Technol. A 10, 1618 (1992). 9. K.-F. Chiu, M.G. Blamire and Z.H. Barber, J. Vac. Sci. Technol. A 17, 2891 (1999)