Department of Materials

Search

en

Materials Day 2003 – Materials Modeling

Introduction

Designing Materials: Computer vs. Laboratory?

The firm answer to this question is: Computer and Laboratory! It must be considered a major breakthrough in materials science and engineering that computers are now capable of providing valuable help in reducing experimental efforts –for example through the calculation of phase diagrams for multicomponent systems–, shortening product development cycles through full processing simulations, or even offering deeper fundamental understanding through theoretical estimates for many material properties. On the other hand, experimental validation of materials simulations and modeling will certainly remain a necessity in any foreseeable future and it should not be expected that computers would eventually make the development of materials in the laboratory obsolete. However, present benefits will not make us forget to lay foundations for the future. In this respect our emphasis must clearly be put on modeling concepts and ideas rather than on computers, which will continue to develop in parallel.

Today, material properties can be modeled on various length and time scales, ranging from nanometers to meters, as well as from femtoseconds to minutes or even hours. However, it is most likely in multiscale modeling that the key to the deepest possible understanding of materials is to be found: bridging scales currently is one of the major fundamental challenges in materials science. Indeed, "multiscale" can likely be expected to become a more than worthy successor to "micro" and "nano" as a buzzword for fund-raising and science policy.

All the above-mentioned aspects of modeling fall in the scope of the Materials Day 2003 and will be covered. The modeling activities pursued today within the Department of Materials at ETH Zürich share their aiming at serious progress through cross-fertilization with laboratory experiments. The Materials Day 2003 offers a unique overview of those activities on the one hand, but also provides an excellent opportunity of gaining insight into the role played by materials modeling in modern science and engineering.

Program

Abstracts

Atom and Continuum – Why and How

U.W. Suter; ETH Zurich, Switzerland
presentation not available

Guided Simulations for Bridging the Gap in Time Scales

Hans Christian Öttinger; Polymer Physics, ETH Zurich, Switzerland
Downloadcomplete presentation (PDF, 256 KB)

For this talk, I distinguish between two fundamentally different simulation approaches in materials science, "brute-force simulations" and "thermodynamically guided simulations." Brute-force simulations can be thought of as computer experiments mimicking the physical situation of interest directly on a computer; thermodynamically guided simulations rely on a nonequilibrium statistical ensemble containing the variables of some coarse-grained description of the system of interest. The availability of an appropriate coarse-grained level of description is thus crucial for thermodynamically guided simulations. Brute-force simulations require less insight, which may be considered good or bad. For reasons to be explained in the talk, I find a negative annotation going with brute-force simulations more appropriate.

The above remarks are elaborated in the context of polymer melts. It is shown how simulations based on nonequilibrium ensembles can help to bridge the wide range of time scales from monomer motions to polymer processing. The importance of coarse-grained models for specifying an ensemble and for identifying suitable quantities of interest is illustrated.

Computer simulation of complex fluids

Martin Kröger, Polymer Physics, ETH Zurich, Switzerland
Downloadcomplete presentation (PDF, 7.7 MB)

Complex fluids exhibit a wide range of interesting time and length scales. Computation of their material properties from first principles is far out of reach for modern supercomputers. In this talk I review some results for the structure and flow properties of non-atomistic, but microscopic models for complex fluids. We focus on polymeric liquids and ferrofluids subjected to external fields. Results are obtained via numerical integration of (stochastic) differential equations. The goal is to verify assumptions in the analytical treatment of these models, whose predictions are often in agreement with some experimental findings about diffusion coefficients, phase behavior, viscosities, scattering patterns, and to propose or identify variables for a unifying, more coarse-grained, and thermodynamically admissible description of the system of interest.

The Calphad Method or The Virtual Thermochemistry Lab

Bengt Hallstedt; Nonmetallic Inorganic Materials, ETH Zurich, Switzerland
Downloadcomplete presentation (PDF, 1.3 MB)

Phase diagrams are used throughout materials science to guide the development and processing of materials in order to reach the desired microstructure and, thus, the desired properties. Experimentally determined phase diagrams are usually only available for binary systems, to some extent for ternary systems and very rarely for higher order systems. In order to gain access to higher order systems thermodynamic modeling offers a possible path. In the Calphad (Calculation of phase diagrams) method each phase is modeled in terms of its Gibbs energy. These Gibbs energy functions contain parameters which can be optimized to fit experimental data on the phase diagram and thermochemical properties. The Calphad method offers a number of advantages:

  • By combining phase diagram and thermochemical data conflicts between different types of data can be identified and often resolved. The result is a high degree of consistency.
  • Extrapolations can be made into multicomponent systems to get an overview of phase relations where the number of degrees of freedom or complexity makes it impossible to achieve this experimentally.
  • Metastable states can be investigated in order to understand and predict the outcome of processes far from equilibrium such as CVD, PVD, rapid quenching etc.
  • In order to treat phase transformations quantitatively it is necessary to have data on driving forces for phase formation and chemical potentials. Calphad descriptions can deliver such data.
  • By studying simple model systems one can learn a lot on the interplay between thermochemical properties and phase equilibria. Here we will give some background on the Calphad method; its inner workings and we demonstarte what we can achieve with it.

We will also give a few examples, e.g. melt processing of Bi-2212 high temperature superconductors. We will try to illustrate each one of the five "advantages" mentioned above with a "real" example.

State-Space Modeling of Electrochemical Processes or Who uses up my battery power?

Michel Prestat, Nonmetallic Inorganic Materials, ETH-Zurich, Switzerland
Downloadcomplete presentation (PDF, 544 KB)

Electrochemistry governs many technical areas during production and use of materials. This is obvious in processes such as corrosion of metals or their aestivation but also during electrode position of protective coatings as well as when using batteries. And this is true particularly in new energy conversion systems such as fuel cells which convert chemical energy directly in electrical energy.

One of the most important aims of the characterization and modeling of an electrochemical device is to identify the rate determining step(s) among the consecutive and parallel elementary reactions of the electrodes processes. This allows efficiently to improve the system by new or modified materials.

Today the impedance of the system is characterized over a wide frequency range and the results are then compared with equivalent circuits consisting of ohmic resistances and capacitances in parallel and series. This strategy can be successful in simple cases but genreally does not make possible the identification of the reaction mechanism.

Recently, a new modeling approach (State-Space Modeling) was developed that provides a general method for the identification of heterogeneous electrode reaction mechanisms. This method allows the mathematical analysis of the basic chemical and electrochemical equations, the steady-state and dynamic simulation of the system as well as the parameter estimation (numerical optimization) within a single framework. The models are described by time-independent state-space equations from which the faradaic admittance transfer function is obtained. The general strategy of State Space Modeling will be presented and illustrated with the identification of the oxygen reduction mechanism at solid oxide fuel cell cathode/electrolyte interface.

Structural modeling – where are the atoms and why?

Walter Steurer; Laboratory of Crystallography, ETH Zurich
Downloadcomplete presentation (PDF, 5.3 MB)

What are the physical properties of carbon? Is it black or transparent, electrically conducting or an insulator, soft or extremely hard? It just depends on the modification of carbon. It can be graphite or diamond; it can have a layer structure or a framework structure. Generally, physical properties of a material depend on its chemical composition, type of chemical bonding and crystal structure. Structural order and symmetry also play an important role. Quartz crystals, for instance, show the piezoelectric effect. Quartz glass does not despite its same chemical composition.

The electrical and thermal conductivity of metals usually increase with decreasing temperature. Quasicrystals, a strange class of intermetallic phases with five-fold symmetry show just the opposite behavior. What is the origin of the extraordinary ordering of quasicrystals? Where are the atoms in quasicrystals and why?

To answer questions of this kind, to understand and predict crystal structure and physical properties for any material, this is the major goal of crystallography. The modeling of crystal structures, of their short- and long-range order as a function of temperature and pressure is shown on several examples.

Modelling of pore solution in the cement-water system

B. Lothenbach; external pageEMPA, Laboratory for concrete and construction chemistry, Dübendorf, Switzerland
Downloadcomplete presentation (PDF, 819 KB)

The interactions between clinkers, pore solution and hydrated minerals determine the setting and hardening of cements. The modelling of these interactions and the composition of pore solutions in cements using geochemical speciation codes forms the basis for the chemical understanding of these processes and of the factors influencing them.

Based on such models, the chemical aspect of the effect of different calcium sulphates and cement admixtures is investigated in order to identify possible further improvements. In addition, the modelling of weathering reactions such as carbonation or sulphate attack eases a systematic investigation of measures to prevent or to minimise these damages.

Modeling of surfaces: The third dimension in XPS analysis of multilayer structures

Antonella Rossi; Surface Science & Technology, ETH Zurich, Switzerland; external pageDipartimento di Chimica Inorganica ed Analitica dell’Università degli Studi di Cagliari, Italy
Downloadcomplete presentation (PDF, 5.7 MB)

Surface analysis using X-ray photoelectron spectroscopy (XPS) is today a well established and continuously expanding area. It is a non-destructive, "ex-situ" technique that allows the composition and the chemical state of the elements present in a surface layer to be established over a thickness of some nanometers based on the measurement of the energy of the photo-emitted electrons. The measured integrated intensities (peak areas) are related to the number of the different emitting species, thus XPS is also a quantitative analytical technique. The first aspect is the most widely applied: it is in fact, relatively easy to detect different chemical states that might play a key role in the surface reactivity of a material in contact with the environment. Quantitative XPS analysis has to take into account the attenuation of the electron in the material and thus is inherently more difficult because of the many spectroscopic and morphological parameters involved. The simplest model, most frequently applied, is based on the assumption that the material is homogeneous. Real surfaces are always covered by thin organic contamination layers and oxide films of some nanometer thickness. Thus they have to be modeled with a layered structure. A mathematical algorithm based on a first principles model and exponential attenuation of the electrons in the material has been developed, taking into account different material densities and different instrumental geometries. With this three-layer model the thicknesses of the layers, as well as the composition of the surface film and of the substrate can be simultaneously calculated from one single XPS measurement in a non-destructive way. This algorithm has been further developed to model also bi-layer or multilayer surface films with in-depth compositional variations. Examples from our research on functionalized surfaces with self-assembled monolayers, multi-component alloys, including ‘real systems’ such as steels, organically functionalized bio-surfaces, and polymers will be presented and critically evaluated for practical applications. The significance of the approximations used in the models will be discussed.

Simulation of Light Metals Processing

Arne Wahlen; external pageARC Light Metals Competence Center, Ranshofen, Austria
Peter J. Uggowitzer; Institute of Metallurgy, ETH Zurich, Switzerland
Downloadcomplete presentation (PDF, 6.2 MB)

The time and cost required to develop new materials and processing routes are a challenging problem in today’s industries. Product development times have been and will continue to be extremely shortened (e.g. car model cycle times in the automotive industry), but the development times of novel materials and process variants has not been reduced significantly. This opposite development creates a barrier to achieving optimal performance of components and assemblies. One possibility to solving these constraints provides the computational materials science by means of modeling and simulation of processing methods, alloy development and microstructural evolution as well as plastic deformation behavior. Furthermore, modeling tools permit the prediction of materials properties and component performance.

In this context, the joint efforts in modeling and simulation of the Light Metals Competence Center in Ranshofen, Austria and the Institute of Metallurgy at ETH Zurich are presented: From modeling evolving solidification in novel casting processes and finite element simulation of aluminum extrusion to evaluation of crashworthiness of magnesium front fender structures, the presentation covers the implementation of computational materials engineering along the manufacturing process chain.

Computer-aided design of structural parts from short fiber reinforced composites

Andrei A. Gusev; Polymer Physics, ETH Zurich, Switzerland
Downloadcomplete presentation (PDF, 2.2 MB)

In lightweight applications, large specific stiffness and strength are attractive properties for a material. By combining short fibers with an appropriate polymer matrix and by controlling the production process it is possible to make composite materials featuring large specific properties. In addition, the constituents, i.e. the polymer matrix and the fibers, are usually inexpensive and easily processed, for example by injection moulding. As a consequence of these technical and economical reasons, short fiber reinforced polymer composites are finding more and more industrial applications where high performance per weight at a reasonable price is required.

It is fairly common in practice that during injection moulding, the mold filling process results in non-uniform fiber orientation states across the final injection molded part. Consequently, one needs to be able to deal with spatially non-uniform elastic constants in order to describe the structural performance of the part. This is, in principle, no problem for the finite element method of structural analysis, provided the elastic constants for all the sections across the part are known.

Here, for the first time, we have developed and implemented a new finite-element-based numerical procedure to directly predict the stiffness and thermal expansion of periodic multi-fiber Monte Carlo computer models with predefined fiber orientation states. The numerical predictions were demonstrated to be in excellent agreement with direct measurements, thus allowing one to employ this new numerical procedure for the fast and reliable computer-aided design of advanced structural parts from short fiber reinforced composites.

JavaScript has been disabled in your browser