Virtualizing Materials to Create Real Ones

Gregory B. Olson

Design of the hierarchical structure in materials requires a hierarchy of models based on materials science, applied mechanics, and even quantum physics. Models to help design features on the coarsest structural level of solidification, such as the chemical banding visible in the patterned sword blades mentioned earlier, employ powerful thermodynamic codes such as THERMOCALC (1). These models enable designers to simulate the 10-μm scale of structure. In metal alloys, this is the level at which the chemical partitioning between liquid and solid phases evolves during solidification processing. Application of these models aids decision-making about thermal processing details, a practice we like to call “solidification design.”

At the 1.0-μm scale of structure, “transformation design” is the goal. This concerns the evolution of structural changes during quenching, whereby crystal grains present at high temperatures transform and subdivide as the hot alloy cools into a hierarchy of lower temperature crystalline units. The design goal here is to specify and control processing temperatures so that desirable microstructures will form, while hindering the formation of competing microstructures that are less beneficial.

The 0.1-μm scale represents the micromechanics design level. An example of the phenomena relevant at this scale is “grain refining,” in which the large grains formed at high temperatures that can embrittle alloys are made smaller by more precise thermal or compositional control. There's a trade-off here, because more smaller particles can also catalyze ductile fracture (breaking, that is), as there are more interfaces that can separate from one another. Micromechanics models typically are based on continuum descriptions of mechanical phenomena that can simulate the evolution of microstructures during material deformation and fracture.

In recent years, an even finer structural level—the nanoscopic level—has become better understood and more controllable. The control of 1-nm-scale particle dispersions in alloys created through solid state precipitation during “tempering” at intermediate temperatures, for example, provides efficient obstacles for resisting plastic deformation. Said differently, this nanometer-scale structuring strengthens the metal.

The development of design models for these diminutive scales builds on a half-century evolution of theory for both precipitation and strengthening in metals. A major materials science breakthrough of the 1950s was the identification of dislocations as the key defects that enable the sliding of crystal planes, which shows up as plastic deformation of many materials. While most structure-property relations, such as the Hall-Petch relation for grain refinement strengthening, are based on empirical correlations, a major triumph of the 1960s was the Orowan particle strengthening equation. Derived directly from dislocation theory, it relates strength to the inverse spacing of dislocation obstacles (such as nanoscale precipitates). Such developments in theory, which mathematically codify materials behavior, are cornerstones for the claim that materials can be designed largely in silico.

The new accuracy of theoretical predictions of structure at the nanometer scale is only possible because of recent advances in high- resolution instrumentation allowing precise calibration and validation of theory in this regime. This includes such techniques as x-ray and neutron diffraction, various electron microscopies, and atom-probe microanalysis. The latter is represented by the three-dimensional atomic reconstruction of a 3-nm strengthening carbide particle in an ultrahigh-strength steel shown below. Such new capabilities in structural and chemical analysis down to the atomic scale open a new era of quantitative materials nanotechnology.

The electronic level is the finest level relevant to real materials. This is the realm of quantum design. As acknowledged by the 1998 Nobel Prize in chemistry shared by John Pople and Walter Kohn, the development of computational quantum mechanics and its extension via density functional theory constitute a profound advance that already has had significant industrial impact. A collaboration of materials science, applied mechanics, and quantum physics has enabled some of us in the Steel Research Group (see main text) to apply computational quantum mechanics to engineer steel at the subatomic level.

Our approach was to recast models of the impurity-induced embrittlement of grain boundaries into thermodynamic terms (2) and to rely on precise models of the atomic structures at grain boundaries. As represented by the computed valence charge density contours for a phosphorus atom at the core of an iron grain boundary, total energy calculations are sufficiently precise and accurate to explain the known effects of interstitial components such as boron, carbon, phosphorus, sulfur, and hydrogen on the cohesion of iron grain boundaries. What's more, the calculations lead to new mechanistic insights at the electronic bonding level. Extension to elements that occupy substitutional Fe sites in the boundary has enabled us to predict new alloying elements for enhancing boundary cohesion in steels. This is creating a new generation of “quantum steels” that incorporate these properties derived directly from electronic-level predictions.