NXms_crystal_set¶
Status:
base class, extends NXobject
Description:
Base class to describe data about observed crystals in microstructures.
Both experiments and computer simulations support that atoms organize into regions which are often separated and interconnected by different types of crystal defects. Crystalline regions show a long-range periodic arrangement (compared to the length scale of nearest neighbor distances). Although the group of atoms forming a crystal is virtually never static, due to diffusive in- and outflux of atoms and thermal fluctuations of the atoms about their equilibrium positions, crystals are relevant persistent features in microstructures. The size of crystals can span orders of magnitude from meters to nanometers.
There are two different and (somewhat extremal) views on crystals and how to describe their eventually very rich variety of internal defects. Either crystals can be coarse-grained into continuum objects or not. In the second case they need a more advanced internal description of defects inside the grains which convinces many that eventually a grain has to be viewed from its individual atoms, its material points, and the hierarchy of structural motifs local arrangements in the crystal.
Despite these details, identifying crystals is foremost a labeling task. Atoms are clustered into structural motifs and (noise) and these motifs are again clustered into crystals.
There are two main approaches how crystals are described in mesoscale microstructure evolution simulations. Assuming for now transformations in the solid state, precipitates, phase regions, sub-grains or grains are examples of crystals:
Objects are either tracked explicitly in the sense that their shape will be resolved through the crystal interfaces using e.g. a phase-field, level-set, grid, or finite element mesh based models and implementations. These simulations may keep track of explicit crystal/grain/object-related quantities. Such models can treat the interface network implicitly while still focusing their description on the explicit crystals. During such simulations the interface is often analyzed on-the-fly, because of technical demands (like in level-set implementations) or to trigger specific situations where it is relevant to resolve the position of the interfaces explicitly (like for placing seeds for phases, recrystallizing grains etc, or for visualization purposes), demand a description of interfaces between crystals. For explicitly tracking simulations this base class can be applied as is.
Objects are tracked implicitly in the sense that the computational domain is discretized into an ensemble of what one can call material points. Such models can be described at different length scale: On the one hand where atom dynamics are (whether the assumption is substantiated or not) homogenized/-able already or not. Each material point is assumed to have at least one associated constitutive phase. Such simulations usually do not/cannot resolve crystal-related quantities without executing an on-the-fly post-processing of snapshot data from which the spatial representation of the crystal is recovered. An important case are large-strain formalism crystal plasticity methods. Here the initial configuration represents most frequently material points on a regular grid. Within the course of the simulation this grid gets deformed implicitly. The code internally keeps no track of how the cells/ material points of what is essentially a Voronoi tessellation, are deformed. Only in the case when one would like to visualize the deformed configuration, a post-processing of the simulation snapshot data is executed which recovers the positions of the material points in the deformed configuration in the laboratory coordinate system from which one can then extract a representation of grains/precipitates, i.e. crystals. It is a signature of such simulations that quantities like orientation are defined as material point quantities. This means what constitutes the grain needs to be extracted by cluster analyses. In this regard, such simulation are essentially matching the representation and case of molecular dynamics simulations, with the important difference that these track atoms, from whose configuration in a snapshot a description has to be computed what are most likely the atoms that belong to the volume of the crystal or the interface/defect network.
Symbols:
The symbols used in the schema to specify e.g. dimensions of arrays.
d: The dimensionality of the description.
n_objects: The number of objects, which can be crystals, grains, phases or phase field regions.
- Groups cited:
none
Structure:
dimensionality: (optional) NX_POSINT {units=NX_UNITLESS}
number_of_objects: (optional) NX_POSINT {units=NX_UNITLESS}
identifier_offset: (optional) NX_INT {units=NX_UNITLESS}
Integer which specifies the first index to be used for distinguishing objects. Identifiers are defined either implicitly or explicitly. For implicit indexing the identifiers are defined on the interval [identifier_offset, identifier_offset+c-1]. For explicit indexing the identifier array has to be defined.
The identifier_offset field can for example be used to communicate if the identifiers are expected to start from 1 (referred to as Fortran-/Matlab-) or from 0 (referred to as C-, Python-style index notation) respectively.
identifier: (optional) NX_INT (Rank: 1, Dimensions: [n_objects]) {units=NX_UNITLESS}
Integer used to distinguish objects for explicit indexing.
object_size: (optional) NX_NUMBER (Rank: 1, Dimensions: [n_objects])
Depending on the value of dimensionality, the area or volume of each object.
Hypertext Anchors¶
List of hypertext anchors for all groups, fields, attributes, and links defined in this class.
