In-Situ Nanoindentation of Epitaxial TiN/MgO (001) in a Transmission Electron Microscope

Transcription

1 Journal of ELECTRONIC MATERIALS, Vol. 32, No. 10, 2003 Special Issue Paper In-Situ Nanoindentation of Epitaxial TiN/MgO (001) in a Transmission Electron Microscope A.M. MINOR, 1,5 E.A. STACH, 2 J.W. MORRIS, JR., 1,3 and I. PETROV 4 1. Materials Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA National Center for Electron Microscopy, Lawrence Berkeley National Laboratory, Berkeley, CA Department of Materials Science and Engineering, University of California, Berkeley, CA Department of Materials Science and the Frederick Seitz Materials Research Laboratory, University of Illinois, Urbana, IL aminor@lbl.gov The deformation behavior of the epitaxial TiN/MgO (001) thin film/substrate system was studied through in-situ nanoindentation in a transmission electron microscope (TEM). The required sample geometry was prepared using Ga focused ion beam (FIB) etching. During room-temperature indentation, both the TiN film and the MgO substrate deformed through the motion of dislocations. The result was a localized hemispherical plastic zone in the TiN film directly under the indentation contact area, forming an 8º tilt boundary. These results show directly that small-scale plasticity in TiN can occur at room temperature through the motion of dislocations. Key words: Nanoindentation, in situ, transmission electron microscopy, epitaxial film, plasticity, dislocation INTRODUCTION The transition-metal carbonitrides are a technologically important class of materials whose mechanical properties remain poorly understood. 1 For instance, macroscopically TiN and TiC are brittle at room temperature, showing no evidence of plastic yielding in room-temperature bending, tension, or compression tests. 2 Indentation tests, however, have shown evidence for dislocation motion through slip steps surrounding indentations 3 and through direct analysis of the post-indentation microstructure by transmission electron microscope (TEM) analysis. 4 However, dislocations have been shown to have low mobility at room temperature in the transitionmetal carbonitrides, 5 and their role in the roomtemperature indentation behavior of these materials is not well known. The transitional-metal carbonitrides have very high melting temperatures; TiN for instance has a melting temperature of approximately 3,000ºC. 6 Because substrate temperatures in high-vacuum growth chambers are rarely greater than 1,000ºC, the growth of TiN in thin-film form usually occurs at a temperature of less than half the homologous (Received May 10, 2003; accepted June 12, 2003) temperature. As a result, transition-metal carbonitride thin films typically have a relatively high amount of intrinsic defects and very small grain sizes. 7 High dislocation densities and small grain sizes make the imaging of these materials difficult when using diffraction contrast in a TEM, and thus, detailed microstructural characterization of these materials has proven to be particularly challenging. 8 Additionally, because of the high defect densities in as-deposited films, microstructural changes related to any induced deformation are easily obscured by the intrinsic defects. However, it is possible to grow heteroepitaxial thin films of TiN on relatively large pieces of single-crystal MgO. Because epitaxial films are free of grain boundaries, they are ideal for studying the fundamental deformation mechanisms in these materials. In this study, we employed the unique mechanical testing technique of in-situ nanoindentation in a TEM 9 11 to directly observe the deformation behavior of the epitaxial TiN/MgO (001) thin film/substrate system. EXPERIMENTAL PROCEDURES The TiN films studied were grown epitaxially on single-crystal MgO (001). Approximately 350-nmthick films were deposited by ultrahigh vacuum, reactive magnetron sputtering of pure Ti on MgO 1023

2 1024 Minor, Stach, Morris, and Petrov Fig. 1. A schematic of the FIB sample geometry. A trench is milled out of the sample leaving behind a window thin enough to be electron transparent. The diamond indenter can approach the thin window in a direction normal to the electron beam, as shown. (001) at 850ºC in pure N 2 discharges maintained at a pressure of 5 mtorr ( 0.67 Pa). Further details of this film deposition have been reported elsewhere To prepare the samples for the in-situ nanoindentation experiments, an FEI (FEI Corporation, Hillsboro, OR, USA) Strata, 235 dual-beam focused ion beam (FIB) microscope was used to mill the samples to the proper geometry. Figure 1 is a schematic of the general sample geometry that the FIB sample-preparation method is designed to achieve. The basic method for FIB sample preparation is to mill out two opposing trenches on a flat piece of material, which leaves behind a thin window that is electron transparent ( 250 nm in thickness). An important aspect of the FIB sample-preparation process is the protection of the sample from damage caused by the incident 30-kV Ga ion beam. To decrease the amount of damage inflicted by the Ga beam, three techniques were employed during the preparation of the epitaxial TiN/MgO (001). The first technique was to take advantage of the electron imaging capabilities of the dual-beam FIB to decrease the total amount of time the sample was exposed to the Ga beam. The dualbeam FIB is equipped with a field-emission scanning electron microscope (FESEM), which allows the user to image the sample with electrons while blanking the sample from the gallium ion beam. The second technique employed to limit ion-beam damage was to slowly decrease the ion-beam cross section as the milling approached the area of the sample that would eventually be the final electron-transparent window. The FIB used has the capability of decreasing the ion-beam current from 20,000 pa to as little as 1 pa, which decreases milling rate but also substantially limits the ion-beam footprint. An ion-beam current of 3,000 20,000 pa was used for the large area milling of the beginning of the trenches, and as the milling approached the area of the final window, the current was steadily decreased to 30 pa. The third and most effective technique for limiting the damage to the sample caused by the ion beam during thinning was to deposit a protective material on the surface of the sample that could later be removed. In typical FIB sample preparation, a thin layer of Pt, Pd, or W is deposited in situ in the FIB for this purpose, but is not removed afterward. For the in-situ nanoindentation experiments, the protective layer had to be removed after the sample was thinned because the surface of the sample that the protective layer was attached to would eventually be the surface indented in the TEM. Thus, the material used as a protective coating was limited to materials that could be easily removed after thinning and where the removal method did not damage the underlying sample. The protective layer used for the TiN/MgO (001) samples was an evaporated Al thin film approximately 500-nm thick. The Al film was deposited via thermal evaporation in a separate chamber at room temperature and at a relatively low vacuum ( 10 3 torr). Because the Al film was to be used purely as a sacrificial layer to absorb incident Ga ions during FIB sample preparation, a film of poor quality was advantageous for removal after FIB processing. Deposition in a low vacuum and at room temperature presumably resulted in a film with high impurity content and poor adhesion. Following deposition of the protective Al layer, trenches were milled out to electron transparency, which was approximately 200 nm in thickness. Figure 2 is an FESEM image of one window taken at an angle of 52º from the surface normal. In this figure, the Al/TiN/MgO trilayer structure can be discerned, where the outermost layer of the window is the Al protective film, followed by the layer of TiN and then the MgO substrate. Figure 3 shows an image of the window taken with a TEM before removing the protective Al layer. As can be seen, the trilayer structure is still intact, implying that the top of the TiN was left undamaged by the Ga beam. The same cannot be said, however, for the sides of the window, because the Ga beam necessarily had to come in contact with the sides at the end of the milling process. However, by the use of a 30-pA ion beam during the final milling stage, the amount of damage is as minimal as can be effectively achieved with this specimen preparation technique. After the electron-transparent window was milled out with the Ga beam, the Al protective layer was removed by immersing the entire sample in a solution of dilute HCl for approximately 5 min. A final TiN/MgO window is shown in Fig.4 with the Al removed, where the TiN layer can be seen on top of the MgO, with very little residual Al. As can be

3 In-Situ Nanoindentation of Epitaxial TiN/MgO (001) in a Transmission Electron Microscope 1025 Fig. 4. A TEM image of an epitaxial TiN/MgO (001) sample prepared with an FIB. The protective Al layer has been removed. The diamond indenter can be seen at the top of the image. Fig. 2. An SEM image of an FIB-prepared Al/TiN/MgO sample at a 52 tilt. Part of one window can be seen on the left side of the image, and a full window can be seen across the middle. Fig. 3. A TEM image of an FIB-prepared Al/TiN/MgO sample before the Al protective layer was removed. seen from this image, the windows are accessible to the diamond indenter normal to the surface that had been protected with Al. The diamond is a threesided Berkovich-type indenter that is boron doped to be electrically conductive in the TEM. The diamond indenter is controlled in three dimensions by a piezoceramic actuator. Further details of the in-situ nanoindentation technique are discussed elsewhere RESULTS AND DISCUSSION Figure 5a and b shows the epitaxial TiN/MgO (001) sample prior to indentation. As can be seen by both the bright-field (Fig. 5a) and dark-field (Fig. 5b) micrographs, there is a very high density of pre-existing defects in the TiN microstructure prior Fig. 5. The (a) bright-field and (b) dark-field TEM micrographs of the epitaxial TiN/MgO (001) sample prior to indentation. Both images are taken in the g ( 220) condition. The arrows showing the [100] direction in both images also point to the location of the indentation shown in Fig. 6.

4 1026 Minor, Stach, Morris, and Petrov to indentation. It is not known whether these defects were created during the growth of the film or during the ion-beam milling. Both processes are known to create defects. The exact character of these defects is also not known because their size is too small to be individually resolved with the TEM (at the sample thickness required for an in-situ nanoindentation). Damage layers have been reported in numerous materials as a result of ion-beam thinning. 15 The implantation of the energetic Ga ions (typically 30 kev) is known to cause amorphization in a layer 30-nm thick in Si. 16 No studies have been conducted regarding Ga ion damage to TiN specifically, although the TRIM software predicts 25% less lateral straggling for 30 kev Ga ions in TiN as compared to Si. 17 Because TiN has a nominal hardness of 20 GPa 18 and MgO has a nominal hardness of 9 GPa, 19 it is to be expected that the softer MgO substrate would deform substantially during indentation. During the indentation of TiN/MgO, plastic flow in the MgO was extensive, and the TiN deformed by a more limited dislocation motion. Figure 6a and b shows bright-field and dark-field micrographs of the postindentation microstructure from an indentation taken to a peak depth of approximately 100 nm (the thickness of the TiN film was approximately 350 nm). Dislocations have arranged themselves in what closely resembles a hemispherical plastic zone surrounding the point of indentation in the TiN. The post-indentation surface of the TiN at the contact interface was so heavily damaged that it simply appears black in both bright-field and any attainable dark-field conditions. This indicates that the black region at the point of indentation in TiN has such a high defect density that the electrons entering this region are almost completely absorbed or scattered. During the in-situ indentation, a large amount of dislocations were generated in the MgO substrate at the film/substrate interface. As can be seen in Fig. 6, indentation-induced dislocations propagated approximately twice as far in the MgO substrate as in the TiN film. During this relatively large indentation (the indentation depth was approximately 1/3 of the film thickness), the TiN film did not de-adhere from the MgO substrate. Rather, the TiN film was forced to accommodate the plastic deformation of the MgO substrate by bending, i.e., the extensive nature of plastic deformation in the MgO led to a bending of the TiN film, as observed in Fig. 6. A hemispherical distribution of dislocations led to an 8º rotation of the crystal around the 001 zone axis underneath the indenter. The TiN crystal was bent 4º on each side of the indentation, as evidenced by comparing pre-indent and post-indent diffraction patterns. Selected area diffraction patterns of the MgO substrate taken before and after the indentation showed no significant changes (i.e., no rotation or splitting of the diffraction spots). Figure 7 shows the post-indent hemispherical plastic zone in the TiN along with selected area Fig. 6. The (a) bright-field and (b) dark-field TEM micrographs of the epitaxial TiN/MgO (001) sample taken after indentation. Both images are taken in the same g ( 220) condition as Fig. 5. The arrows showing the [100] direction in both images also point to the location of the indentation. The diffuse nature of the dislocation plasticity can be seen in the MgO substrate, while the hemispherical configuration of dislocations around the indentation can be seen in the TiN film. diffraction patterns of the TiN film taken before and after the indentation. The circle drawn on the TEM micrograph in Fig. 7a shows the region of the TiN from which the diffraction information was taken. Figure 7c shows the splitting of the 0 22 diffraction spot into approximately two spots with 8º of rotation between them (around the 001 zone axis). These two spots now represent each side of the TiN underneath the indenter, where the TiN film has formed what is effectively a diffuse grain boundary. Essentially, as the splitting of the 0 22 spots shows, one could now describe the TiN film as having a diffuse, 8º tilt boundary formed directly underneath the axis of the indentation direction. This boundary is comprised of a dislocation array formed during the indentation.

5 In-Situ Nanoindentation of Epitaxial TiN/MgO (001) in a Transmission Electron Microscope 1027 Fig. 7. (a) The post-indent hemispherical plastic zone in the TiN film. (b) Selected area diffraction pattern of the TiN film before the indentation. (c) Selected area diffraction pattern from the plastic zone of the indentation. The circle drawn over the image in (a) corresponds to the region from which the diffraction information is taken. In (c) the splitting of the [0 22] diffraction spots can be seen, indicating the bending of the TiN film 4 to each side. To our knowledge, the movement of dislocations in TiN at room temperature has never been directly observed before. Significantly, the defor-mation of TiN during room-temperature nanoindentation has been shown to occur through the motion of dislocations. The bending of the TiN film 8º and the creation of a diffuse tilt boundary comprised of dislocations is a clear indication that small-scale plasticity can occur in a material that is traditionally (at least macroscopically) classified as brittle. CONCLUSIONS The deformation behavior of the epitaxial TiN/MgO (001) thin film/substrate system was studied by indenting the TiN in situ in a TEM. The required sample geometry was milled with a FIB after protecting the surface with a sacrificial film of Al, which was later removed. The deformation of the harder TiN film was controlled by the deformation of the softer MgO substrate. The MgO substrate deformed by diffuse dislocation motion, while the TiN created a localized, hemispherical plastic zone directly under the indentation contact area. The plastic zone formed in the TiN accommodated an 8º rotation of the crystal centered on the axis of indentation. The indentation of the TiN film essentially created a diffuse 8º tilt boundary comprised of geometrically necessary dislocations. These results show that small-scale plasticity can occur in TiN at room temperature through the motion of dislocations. ACKNOWLEDGEMENTS This work was funded by the Office of Energy Research, Office of Basic Energy Sciences, Materials Sciences Division, the U.S. Department of Energy, under Contract Nos. DE-AC03-76SF00098 and DEFG02-91ER The continuing assistance of Velimir Radmilovic with the focused ion beam instrument is greatly appreciated. REFERENCES 1. J.E. Sundgren, A. Rockett, and J.E. Greene, J. Vac. Sci. Technol. A 6, 2770 (1986). 2. For example, W.S. Williams, Progress in Solid State Chemistry (New York: Pergamon Press, 1971), pp W.S. Williams, Proprietes Thermodynamiques, Physiques et Structurales des Derives Semi-metalliques (Paris, editions du Centre National de la Rechereche Scientifique, 1967), pp M. Oden, H. Ljungcrantz, and L. Hultman, J. Mater. Res. 12, 2134 (1997). 5. W.S. Williams, J. Appl. Phys. 35, 1329 (1964). 6. L.E. Toth, Transition Metal Carbides and Nitrides (New York, Academic Press, 1971), p J.S. Chun, I. Petrov, and J.E. Greene, J. Appl. Phys. 86, 3633 (1999). 8. H. Holleck, J. Vac. Sci. Technol. A 4, 2661 (1986). 9. E.A. Stach, Microsc. Microanal. 7, 507 (2001). 10. A.M. Minor, J.W. Morris, and E.A. Stach, Appl. Phys. Lett. 79, 1625 (2001). 11. A.M. Minor, E.T. Lilleodden, E.A. Stach, and J.W. Morris, Jr., J. Electron. Mater. 31, 958 (2002). 12. B.W. Karr, I. Petrov, P. Desjardins, D.G. Cahill, and J.E. Greene, Surf. Coating Technol , 403 (1997). 13. B.W. Karr, I. Petrov, D.G. Cahill, and J.E. Greene, Appl. Phys. Lett. 70, 1703 (1997). 14. B.W. Karr, D.G. Cahill, I. Petrov, and J.E. Greene, Phys. Rev. B 61, (2000). 15. R. Menzel, K. Gartner, W. Wesch, and H. Hobert, J. Appl. Phys. 88, 5658 (2000). 16. S. Rubanov and P.R. Munroe, J. Mater. Sci. Lett. 20, 1181 (2001). 17. J. Ziegler, TRIM, H. Ljungcrantz, M. Oden, L. Hultman, J.E. Greene, and J.-E. Sundgren, J. Appl. Phys. 80, 6725 (1996).