Measurement and Prediction of Local Structure in Electronic Ceramics

Our goal is to provide analytical tools that allow measurement and prediction of local structure to enable the development of ceramic materials for electronic applications. Under this project, we develop data analysis methods for quantitative determination of local structure from multiple experimental techniques, and theoretical methods for prediction of local atomic configurations from first principles.

Electronic ceramics enable applications in the computer, data storage, and wireless communication sectors. The properties of many of these ceramics are critically dependent on deviations of local atomic arrangements from the global average of the crystal. No single measurement technique can provide an accurate description of the local structure. Therefore, our efforts focus on developing analytical tools (methods and computer software) that allow integration of measurements from multiple experimental methods and theory to enable accurate and comprehensive determination and prediction of local atomic configurations. Specifically, we develop methods and computer software for structural refinements using simultaneous fitting of diffraction and x-ray absorption data, and modeling tools for ab initio simulations of diffuse scattering in systems with nanoscale order, Raman spectra in systems with local disorder, and point defects and their complexes in oxide dielectrics. All of this input is needed for the determination of local structure in electronic ceramics.

Impact and Customers:

• The availability of methods and analytical tools for measuring local structure will accelerate the formulation of structure-property relations for advanced electronic ceramics and allow more rapid selection and optimization of materials, thereby decreasing the development time of new electronics products.Pic2 measurement and prediction

• Computer software for simultaneous refinements of atomic arrangements using diffraction and x-ray absorption fine structure data was developedand transferred to our partners at ISIS in the United Kingdom, for web posting.

• An invited review article in Science, which described the problem of solving structure at the nanoscale, followed by input from scientists from industry (e.g., GE, ExxonMobil, Ford, IBM, Sematech), government agencies, and universities, inspired a NIST workshop, "Measurement Needs for Local Structure Determination", that took place in February, 2008.

Electronic ceramics with the Perovskite structure are widely used in functional device applications. Most of the technologically relevant perovskites are solid solutions and therefore their local atomic arrangements necessarily deviate from those described by the average crystallographic positions. These local deviations control the functional properties of such perovskites; therefore, accurate knowledge of the local structure is needed to determine structure-property relationships. We addressed this problem by developing Reverse Monte Carlo (RMC) modelling methods that utilize a combined real space fit of the neutron and x-ray total scattering pair-distribution function (PDF), with extended X-ray absorption fine structure (EXAFS) data. The method was implemented as an extension to the public domain computer software RMCProfile. We tested the effectiveness of the combined PDF-EXAFS RMC refinements using the prototype perovskite system Ca(Zr,Ti)O3.

Simultaneous RMS fits of multiple types of data for Ca(Zr0.5Ti0.5)O3 solid solutions

The analyses revealed that combining these two types of data yields correct distributions of the Ti-O and Zr-O bond lengths. The use of either technique alone would not have allowed this outcome because of the substantial overlap between the Ti-O and Zr-O partial PDF's. The combined refinements enabled reasonably accurate reproduction of most of the local structure characteristics, including the dependence of Ca displacements on the local Ti/Zr ratio around Ca. Additionally, we developed a new approach for extracting cation short-range order parameters in perovskite solid solutions.

We have further used our first-principles calculations to develop models to correlate local structure with measurable properties for a variety of systems with chemical disorder. For example, toward the ultimate goal of making Raman spectroscopy a quantitative probe of local structure, estimated Raman spectra were calculated for a variety of supercells of the perovskite-like material, SrAl1/2Nb1/2O3-SrTiO3. Even for the small supercells studied,significant differences are seen in the predicted Raman spectra.

Models for defect complexes were also created for dielectric materials that will be used as gate dielectrics in silicon integrated circuit devices. We found that Al substitutes for Hf in HfO2 rather than occupying an interstitial site, and that Al substitutions congregate near an O vacancy. The structure of such defects and their effects on the properties of HfO2 as a gate dielectric material are of great interest to the microelectronics industry, since diffusion of Al into HfO2 occurs during processing when Al2O3 is used as an additional dielectric layer with HfO2 . We have also modeled defect structures in gate dielectrics using a cluster expansion (CE) effective Hamiltonian, and have successfully established this method as a viable theoretical technique for calculating Schottky pair (complex) distributions in ionic crystals.

Energies of Schottky complexes in HfO2