Korea Institute of Science and Technology

Transcription

1 Computational Materials Science and Engineering using MedeA a presentation to Korea Institute of Science and Technology June 25, 2010 Materials Design, Inc. 1

2 Outline Materials Design company profile Prediction and Understanding of Materials Properties with MedeA Illustrative Case Study Demonstration of the MedeA Interface Discussion

3 Company Mission To Provide Leading Simulation Software and Consulting Services for the Prediction and Understanding of Materials Properties

4 Company Profile Materials Design, Inc. started in 1998 in San Diego, California, USA 1999: Creation of European branch in France Customers in industry, government labs, and universities Automotive and Aerospace Metals and Alloys Electronics Energy (Nuclear) Chemical and Petrochemical Specialized sectors, e.g. Drilling Core competence: computational chemistry, physics, and materials science materials property databases scientific software engineering 20 employees + consultants and contractors Global network of technology partners with universities and research organizations Business partners in Japan, China, India, Korea, and Taiwan

5 Global Presence Over 100 software installations worldwide 5

6 Some Commercial Customers 6

7 Some Government and Not-For Profit Customers 7

8 Products and Services MedeA software Installation, training, online support, and maintenance Scientific/technological interactions Users group meetings Contract research Solution of specific industrial problems Leverages expertise and resources of MD s scientists Publicly funded programs Technology partnerships Development of customized modeling capabilities (e.g. Toyota)

9 MedeA M O D E L I N G & A N A L Y S I S Builders: solids, interfaces, surfaces, molecules Analysis: geometry, band structure, Fermi, Phonon Job Server D A T A B A S E S Experimental and Computed Structure and Property Data ICSD NIST Crystal Data Pauling Pearson Computed Task Servers Mechanical Thermal Chemical Kinetic Electric Optic Magnetic C O M P U T A T I O N O F P R O P E R T I E S Forcefield Monte Carlo - GIBBS ab initio QM - VASP Forcefield molecular dynamics - LAMMPS Semiempirical QM - MOPAC 9

10 Our Goal Predict and understand materials properties Approaches: Ab initio electronic structure calculations Forcefield simulations Statistical mechanics Analytical theory Empirical correlations Experimental data of existing materials (databases) as reference 10

11 Computed Materials Properties Structural properties Molecular structures Crystal structures Surface structures Structure around defects Adsorption geometries Structures of interfaces Liquids and amorphous systems Thermo-Mechanical properties Elastic moduli Speed of sound Vibrational properties Thermal expansion coefficients Thermodynamic properties U, H, S, G, heat capacity Binding energies Solubility Melting temperature Vapor pressure Miscibility Phase diagrams Surface tension Chemical properties Chemical reaction rates in gases and condensed phases Reactivity on surfaces Solid-solid reactions Pressure-induced reactions Photochemical reactions Transport properties Mass diffusion coefficient Permeability Thermal conductivity Viscosity Electronic, optical, and magnetic properties Electron density distribution - electrical moments Polarizabilities, hyperpolarizabilities Optical spectra Dielectric properties Piezoelectric properties Electrostatic potential Spin density distribution, magnetic moments Energy band structure - metal, semiconductor, insulator, superconductor Band gaps, band offsets at hetero-junctions Ionization energies and electron affinities Work function

12 Property Prediction 12

13 Material Properties A B Key Tasks Characterize pure systems: structure, properties Determine the effect of trace elements/defects on structure and stability (phase) and material properties as a function of concentration Identify lowest energy surfaces for interface models 13

14 Material Structure 14

15 Electronic Band Gaps 15

16 Interface Properties A B Key Tasks Analyze interface structure, properties and separation energy Determine the effect on interface structure and properties as a function of concentration 16

17 Al/Si 3 N 4 17

18 Fracture in Zr Intergranular cracking more prevalent as both temperature and iodine concentration increased Possible mechanism: Iodine in grain boundary diffuses to crack tip and lowers energy of separation 18

19 Effect of Impurities (Additives) Change in energy of separation of Zr grain boundary by impurities 19

20 Chemical Properties Al(CH 3 ) 3 CH 4 Al(CH3 ) 2 O A Cu Key Tasks Analyze O/Cu diffusion energetics for difference structures and composition Analyze surface chemistry energetics for layer deposition 20

21 O Diffusion in Ni Grain Boundary COMPUTATION OF DIFFUSION COEFFICIENTS: CURRENT CAPABILITIES AND PERSPECTIVES Erich Wimmer, Clive Freeman, Hannes Schweiger, Walter Wolf, and Paul Saxe Symposium: Diffusion in Materials for Energy Technologies TMS Annual Meeting, San Francisco, February 15-19, 2009

22 Computed Diffusion Coefficients Ab initio electronic energies Ab initio phonon dispersions including coupling with lattice vibrations Transition state theory Kinetic Monte Carlo Grain boundary diffusion dominates at temperatures below 600 C; bulk diffusion is faster at higher temperatures

23 ALD Growth Chemistry H 2 O Pulse Al(CH 3 ) 3 Pulse H 2 O Pulse Al(CH 3 ) 3 Pulse Purge Purge Purge ALD Nucleation Reactions Si X + ML 3 Si ML 2 + LX Si X + OR 2 Si OR + OR M 2 O 3 ALD Growth ½-Reactions Su* OR + ML 3 Su* O ML 2 + LR Su* O ML R 2 O R O M(OR) LR Effect of Ligand Structure? Z 1 Kinetics? Energy ZrCl 4 vs Zr(CH 3 ) 4 Thermodynamics? E a,0 R G H R P G Reaction Coordinate

24 ALD Growth Chemistry Journal of Chemical Physics 2003, 118, Nature Materials 2009, 8, 825. [112] B A C D E [112] Journal of Physical Chemistry B 2004, 108, E Rel (ev) ev R TS1 P1 TS2 P2 Reaction Coordinate 24

25 Illustrative Case Study: Workfunction Changes in HfO 2 /TiN Stack 25

26 Electronic Structure of Gate Stacks

27 Metal - Oxide Interface 27

28 Computation of Interface Structure 1. Find supercell and orientation with best match of HfO 2 (-111) and TiN(111) surfaces using MedeA-Interface builder 5. Add more layers of HfO 2 and TiN and relax all atoms in the system 2. Create thin slab of HfO 2 surface 3. Deposit thin layer of TiN 4. Perform simulated annealing keeping bottom layers of oxide frozen Interface remains abrupt with some relaxation of O towards Ti and N towards Hf

29 Annealing of TiN Film on HfO 2 Surface TiN HfO 2

30 Electrostatic Potential Across Interface macroscopic average 30

31 Electrostatic Potential Across Interface

32 Possible Effect of Anneal HfO 2 Metal TiN O 2 /N 2 anneal at 450 ºC HfO 2 Metal TiN O 2 /N 2 anneal at 450 ºC O 2 /N 2 anneal at 450 ºC as deposited after anneal Ab initio simulations reveal: 1. Substitution of N by O inside TiN is not the main cause of the increase in the EWF 2. Filling of oxygen vacancies at interface is important 3. Replacement of O by N directly at the HfO 2 /TiN interface has significant effect

33 Complete Gate Stack Model 33

34 Discovery of New Low Modulus Ti-based Alloys using First-principles Calculations Ikehata et al., Phys. Rev. B 70, (2004) 34

35 Mechanical Properties of Ti-Alloys Ti Ti.99 W.01 Ti.98 W.02 Ti.97 W.03 Ti.96 W.04 Ti.95 W.05 Ti.94 W.06 Ti.93 W.07 Ti.92 W.08 Ti.91 W.09 Ti.90 W.10 Ti.89 W.11 Ti.88 W.12 Ti.87 W.13 Ti.86 W.14 Ti.85 W.15 Ti.84 W.16 Ti.83 W.17 Ti.82 W.18 Ti.81 W.19 Ti.80 W.20 Ti.79 W.21 Ti.78 W.22 Ti.77 W.23 Ti.76 W.24 Ti.75 W.25 Ti.74 W.26 Ti.73 W.27 Ti.72 W.28 Ti.71 W.29 Ti.70 W.30 Ti.69 W.31 Ti.68 W.32 Ti.67 W.33 Ti.66 W.34 Ti.65 W.35 Ti.64 W.36 Ti.63 W.37 Ti.62 W.38 Ti.61 W.39 Ti.60 W.40 Ti.59 W.41 Ti.58 W.42 Ti.57 W.43 Ti.56 W.44 Ti.55 W.45 Ti.54 W.46 Ti.53 W.47 Ti.52 W.48 Ti.51 W.49 Ti.50 W.50 Ti.49 W.51 Ti.48 W.52 Ti.47 W.53 Ti.46 W.54 Ti.45 W.55 Ti.44 W.56 Ti.43 W.57 Ti.42 W.58 Ti.41 W.59 Ti.40 W.60 Ti.39 W.61 Ti.38 W.62 Ti.37 W.63 Ti.36 W.64 Ti.35 W.65 Ti.34 W.66 Ti.33 W.67 Ti.32 W.68 Ti.31 W.69 Ti.30 W.70 Ti.29 W.71 Ti.28 W.72 Ti.27 W.73 Ti.26 W.74 Ti.25 W.75 Ti.24 W.76 Ti.33 W.77 Ti.32 W.78 Ti.21 W.79 Ti.20 W.80 Ti.19 W.81 Ti.18 W.82 Ti.17 W.83 Ti.16 W.84 Ti.15 W.85 Ti.14 W.86 Ti.13 W.87 Ti.12 W.88 Ti.11 W.89 Ti.20 W.90 Ti.09 W.91 Ti.08 W.92 Ti.07 W.93 Ti.06 W.94 Ti.05 W.95 Ti.04 W.96 Ti.03 W.97 Ti.02 W.98 Ti.20 W.99 W Ti W Alloys derived from Ti, Ta, Mo, Nb, V and W are used in highperformance applications For key applications it is important to identify material solutions that combine low-density with low-young s modulus Experimental exploration of the design space for a given application for even a limited set of alloy formulations can be prohibitively expensive in terms of time, effort and cost Analyze binary Ti-based alloys: Ti 1-m X m (X = V, Nb, Ta, Mo, W)

36 Mechanical Properties of Ti-Alloys

37 Mechanical Properties of Ti-Alloys

38 Mechanical Properties of Ti-Alloys

39 Mechanical Properties of Ti-Alloys Simulation derived design rules!

40 Mechanical Properties of Ti-Alloys Ti-Nb-Ta-Zr-O alloys R&D Review of Toyota CRDL Vol.38 No.3. (2003)

41 Changes in Graphite Electrode Elastic Properties upon Li Intercalation using MedeA Qi et al., J. Electrochem. Soc. 157(5), A558 (2010) 41

42 Mechanical Properties of Electrodes Engineering scaling simulations require fundamental properties of the component materials in the device The properties of Li battery materials change with varying Li concentration Graphite is a common anode material; known to undergo a 10% volume increase on Li intercalation Accumulated internal stress, due to repeated Li diffusion in and out of the anode can lead to structural failure First-principles calculations were carried out to determine the Young s modulus and Poisson s ratio for graphite at Li loading levels representative of the phases formed during the cycling of a Li ion battery

43 Mechanical Properties of Electrodes LiC 18 LiC 12 LiC 6 The lithium graphite intercalation compounds (Li-GIC) are shown The Li filled interlayer spaces increase by ~10%, whereas at low loading levels the unfilled interlayer spaces undergo slight contraction, consistent with experimental measurements The Young s modulus (E) for the Li-GICs is predicted to increase by more than a factor of 3 in going from C LiC6.

44 Mechanical Properties of Electrodes Predictive model for use in continuum-scale models of electrode deformation The magnitude of the effect and correlation with loading levels is important for the development of engineering-scale models of electrode deformation and fracture Comparison with the E values for graphite and lithium indicate that this increase is not a linear combination of the reference data The Li-GIC Young s modulus is found to follow an approximate linear relationship in Li concentration.

45 MedeA Demo 45

46 MedeA M O D E L I N G & A N A L Y S I S Builders: solids, interfaces, surfaces, molecules Analysis: geometry, band structure, Fermi, Phonon Job Server D A T A B A S E S Experimental and Computed Structure and Property Data ICSD NIST Crystal Data Pauling Pearson Computed Task Servers Mechanical Thermal Chemical Kinetic Electric Optic Magnetic C O M P U T A T I O N O F P R O P E R T I E S Forcefield Monte Carlo - GIBBS ab initio QM - VASP Forcefield molecular dynamics - LAMMPS Semiempirical QM - MOPAC 46

47 MedeA: Three tiered architecture Key High Level Benefits Web browser based data retrieval on any client Sharing of results, e.g. URL for a given calculation Permanent record of all calculations Graphical Client/User Interface Job Server, Databases Independent net based update Clients and servers are not in lock step update servers without need to update clients Optimizes use of available compute resources Task Servers (data generation) 10/21/ Materials Design, Inc. Inc. 47

48 ZnO Band Structure Retrieve band structure: Analysis -> -> Band Structure, use the filter '*Zn*O*band*'. In In case of of the HSE calculations insert the record without the tag 'DFT'. BandStructure->check Measuring lines,vuse the zoom function (drag mouse cursor with right button clicked) to to see more details ev 2.61 ev 10/21/ Materials Design, Inc. Inc. 48

49 Ge Computed Optical Properties Retrieve optical properties: Analysis ->Optical properties, use the filter '*Ge*band*'. In In case of of the HSE06 calculations insert the record without the tag 'DFT'. Use the zoom function (drag mouse cursor with right button clicked) to to see more details. infrared refractive index n: HSE06: ~4.6 Exp.: 4.0 [2] 10/21/ Materials Design, Inc. Inc. 49

50 Mechanical Properties of Ti-Alloys

51 Surface Transport Transition State Pd trimer Pd dimer Pd tetramer 10/21/ Materials Design, Inc. Inc. 51

52 Local Potential - Metal/Oxide Interface Deformation density (SCF) - (atoms) 52

53 Local Potential - Effect of Adsorption Adsorption has long-range electrostatic effects

54 Discussion 54