The peculiarity of the metal-ceramic interface

Abstract Important properties of materials are strongly influenced or controlled by the presence of solid interfaces, i.e. from the atomic arrangement in a region which is a few atomic spacing wide. Using the quantitative analysis of atom column positions enabled by CS-corrected transmission electron microscopy and theoretical calculations, atom behaviors at and adjacent to the interface was carefully explored. A regular variation of Cu interplanar spacing at a representative metal-ceramic interface was experimentally revealed, i.e. Cu-MgO (001). We also found the periodic fluctuations of the Cu and Mg atomic positions triggered by the interfacial geometrical misfit dislocations, which are partially verified by theoretical calculations using empirical potential approach. Direct measurements of the bond length of Cu-O at the coherent regions of the interface showed close correspondence with theoretical results. By successively imaging of geometrical misfit dislocations at different crystallographic directions, the strain fields around the interfacial geometrical misfit dislocation are quantitatively demonstrated at a nearly three-dimensional view. A quantitative evaluation between the measured and calculated strain fields using simplified model around the geometrical misfit dislocation is shown. Introduction The electrical properties in electronic industry are controlled by various interfaces, such as metal-ceramic and metal-semiconductor interfaces. In the fields of semiconductor technology and surface engineering, the metal-oxide interfaces are encountered frequently and playing a decisive role as they control the properties of metal–ceramic composites, protective coatings and thin metal–ceramic films in electronic devices etc. Although this obvious practical importance, our basic understanding of interfaces is still in its infancy, and lacks a fundamental correlation of interface structure and materials properties. The importance of interfaces essentially lies in the fact that physical and chemical properties may change dramatically at or near the interface itself. The significance of metal-ceramic interfaces in so many technological relevant composite materials and thin film electronic devices is strongly reflected in the continuing and extensive studies for many decades1,2,3,4,5,6,7,8,9,10,11. It is apparent that atoms at or near the interface do not all possess the same local environment. Thus, characterizing the local atomic structure at or adjacent to the interface becomes of vital importance as it controls the resulting properties. Although massive investigations of dislocation core structure of interfacial misfit dislocations have been performed using high–resolution transmission electron microscopy (HRTEM)3,7, the accurate and quantitative description on the local atomic structure is so far less advanced due to the limitations of lens aberration of the microscopes. This is especially true for semi-coherent interfaces, i.e. metal-oxide interfaces. The unambiguous knowledge of the exact atomic configurations and behaviors within a few layers at or adjacent to the interface is still scarce. This limitation can now be overcome with spherical aberration (CS)-corrected HRTEM. Now an abundance of knowledge on atomic scale structure from very local position can be gained. The intrinsic physics of materials is possible to be directly read out of the HRTEM images12,13. This largely prompts a new view on interface controlled materials, such as metal-ceramic composites and their interface structures. It is reasonably anticipated that a novel understanding can be achieved, and the physics behind metal-ceramic interfaces can be atomically unveiled when using the picometer-scale precision CS-corrected HRTEM12, and combined with the theoretical calculations. Due to the mismatch of the two lattices, a dislocation network exists at metal-ceramic interfaces. To discriminate between subtle differences in the interfacial structure it would be beneficial to observe the interfaces edge-on along two different directions at the atomic scale, which is hardly possible in normal transmission electron microscopy (TEM). However, the image-side CS-corrected microscopes with their wide pole-piece gap enabling large tilt angles and an ultrahigh-resolution that are fortunately available in our microscope provide a possibility to study the interface in two projections by successively atom-resolved imaging in e.g. [100] and [110] zone axes. This will help achieving a better understanding of misfit dislocation structures and local strain variations at a nearly three-dimensional view14. The Cu-MgO interface is a model system for analyzing metal-ceramic interfaces1,2,3,4,5,6,7,8,9,10,11. Both Cu and MgO are fcc lattices, and due to their different lattice parameter (aMgO = 0.42105 nm and aCu = 0.36148 nm), a large mismatch of 14.1% exists between the two lattices. The ratio of the metal and oxide lattice constants can be approximated by a simple ratio: 7aCu ≈ 6aMgO for Cu-MgO (001). For epitaxial Cu films grown on (001) MgO substrate, the dislocation network was found to lie along 〈100〉 directions with a Burgers vector of ½aCu 〈100〉 deduced from HRTEM images in an earlier study15, which is in agreement with other report4. On the other hand, as the interface possesses a large misfit, it could not be semi-coherent anymore, and the interface should be categorized into incoherent, forming a geometrical misfit dislocation (GMD) network. In this work, combining quantitative atom position measurements with theoretical calculations, and successively imaging of the same interface position along two different crystallographic orientations, we present a picometer-scale understanding of atom behaviors at and adjacent to the Cu-MgO interface. Experimental sections The Cu-MgO interface is a model system for metal-ceramic interfaces. Both Cu and MgO are fcc lattices, and a large mismatch (14%) exists. The ratio of the metal and oxide lattice constants can be approximated by a simple ratio: 7aCu ≈ 6aMgO for Cu-MgO (001). The epitaxial Cu film was deposited by using an unbalanced direct current magnetron sputtering system at a substrate temperature of 350 °C. Prior to deposition, a polished MgO (001) substrate was cleaned with successive rinses in ultrasonic baths of trichloroethylene, acetone, isopropyl alcohol, and deionized water and thermally degassed at 800 °C in vacuum. The 2 μm thick Cu layers were then sputtered in Argon atmosphere at 1 Pa total pressure, using a 99.99% pure Cu target, a power of 2 kW and at floating potential. Further details are reported in15. Possible point defects that form in the sputter-deposited Cu layer are expected to anneal out within seconds by Cu self-annealing at 350 °C16. The TEM sample was specially cut at an angle of 22.5° between [001] and [011] orientations (supplementary, Fig. S1), enabling to reach both zone axes by tilting in the microscope for imaging t (...truncated)