3.3.3.20. NXapm_paraprobe_config_nanochem

Status:

application definition, extends NXobject

Description:

Configuration of a paraprobe-nanochem tool run in atom probe microscopy.

Symbols:

The symbols used in the schema to specify e.g. dimensions of arrays.

n_ityp_deloc_cand: How many iontypes does the delocalization filter specify.

n_control_pts: How many disjoint control points are defined.

n_fct_filter_cand: How many iontypes does the interface meshing iontype filter specify.

n_fct_iterations: How many DCOM iterations.

n_ivec: Maximum number of atoms per molecular ion.

Groups cited:

NXapm_input_ranging, NXapm_input_reconstruction, NXcg_cylinder_set, NXcg_ellipsoid_set, NXcg_face_list_data_structure, NXcg_hexahedron_set, NXcs_filter_boolean_mask, NXcs_profiling, NXentry, NXmatch_filter, NXprocess, NXspatial_filter, NXsubsampling_filter

Structure:

ENTRY: (required) NXentry

@version: (required) NX_CHAR

Version specifier of this application definition.

definition: (required) NX_CHAR

Official NeXus NXDL schema with which this file was written. ...

Official NeXus NXDL schema with which this file was written.

Obligatory value: NXapm_paraprobe_config_nanochem

program: (required) NX_CHAR

Given name of the program/software/tool with which this NeXus ...

Given name of the program/software/tool with which this NeXus (configuration) file was generated.

@version: (required) NX_CHAR

Ideally program version plus build number, or commit hash or description ...

Ideally program version plus build number, or commit hash or description of ever persistent resources where the source code of the program and build instructions can be found so that the program can be configured ideally in such a manner that the result of this computational process is recreatable in the same deterministic manner.

analysis_identifier: (required) NX_CHAR

Ideally, a (globally persistent) unique identifier for referring ...

Ideally, a (globally persistent) unique identifier for referring to this analysis.

analysis_description: (optional) NX_CHAR

Possibility for leaving a free-text description about this analysis.

results_path: (optional) NX_CHAR

Path to the directory where the tool should store NeXus/HDF5 results ...

Path to the directory where the tool should store NeXus/HDF5 results of this analysis. If not specified results will be stored in the current working directory.

time_stamp: (required) NX_DATE_TIME

ISO 8601 formatted time code with local time zone offset to ...

ISO 8601 formatted time code with local time zone offset to UTC information included when this configuration file was created.

number_of_processes: (required) NX_UINT {units=NX_UNITLESS}

How many individual analyses should the tool execute as part of the analysis.

PROCESS: (required) NXprocess

dataset: (required) NXapm_input_reconstruction

filename: (required) NX_CHAR

@version: (required) NX_CHAR

dataset_name_reconstruction: (required) NX_CHAR

dataset_name_mass_to_charge: (required) NX_CHAR

iontypes: (required) NXapm_input_ranging

filename: (required) NX_CHAR

@version: (required) NX_CHAR

group_name_iontypes: (required) NX_CHAR

spatial_filter: (optional) NXspatial_filter

windowing_method: (required) NX_CHAR

CG_ELLIPSOID_SET: (optional) NXcg_ellipsoid_set

dimensionality: (required) NX_POSINT

cardinality: (required) NX_POSINT

identifier_offset: (required) NX_INT

center: (required) NX_NUMBER

half_axes_radii: (required) NX_NUMBER

orientation: (required) NX_NUMBER

CG_CYLINDER_SET: (optional) NXcg_cylinder_set

dimensionality: (required) NX_POSINT

cardinality: (required) NX_POSINT

identifier_offset: (required) NX_INT

center: (required) NX_NUMBER

height: (required) NX_NUMBER

radii: (required) NX_NUMBER

CG_HEXAHEDRON_SET: (optional) NXcg_hexahedron_set

dimensionality: (required) NX_POSINT

cardinality: (required) NX_POSINT

identifier_offset: (required) NX_INT

hexahedra: (required) NXcg_face_list_data_structure

CS_FILTER_BOOLEAN_MASK: (optional) NXcs_filter_boolean_mask

number_of_objects: (required) NX_UINT

bitdepth: (required) NX_UINT

mask: (required) NX_UINT

identifier: (required) NX_UINT

evaporation_id_filter: (optional) NXsubsampling_filter

iontype_filter: (optional) NXmatch_filter

method: (required) NX_CHAR

match: (required) NX_NUMBER

hit_multiplicity_filter: (optional) NXmatch_filter

edge_of_the_dataset: (required) NXprocess

The tool enables to inject a previously computed triangle soup or ...

The tool enables to inject a previously computed triangle soup or triangulated surface mesh representing a model (of the surface) of the edge of the dataset. This model can be used to detect and control various sources of bias in the analyses.

filename: (required) NX_CHAR

Name of the HDF5 file which contains vertex coordinates and facet ...

Name of the HDF5 file which contains vertex coordinates and facet indices to describe the desired set of triangles which represents the edge of the dataset.

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which documents ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

dataset_name_vertices: (required) NX_CHAR

Absolute path to the HDF5 dataset in the respectively specified HDF5 ...

Absolute path to the HDF5 dataset in the respectively specified HDF5 file under filename which details the array of vertex positions.

dataset_name_facet_indices: (required) NX_CHAR

Absolute path to the HDF5 dataset in the respective specified HDF5 ...

Absolute path to the HDF5 dataset in the respective specified HDF5 file under filename which details the array of facet indices.

ion_to_edge_distances: (optional) NXprocess

The tool enables to inject precomputed distance information for each ...

The tool enables to inject precomputed distance information for each point/ion which can be used for further post-processing and analysis.

filename: (required) NX_CHAR

Name of an HDF5 file which contains the ion distances.

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which documents ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

dataset_name: (required) NX_CHAR

Absolute HDF5 path to the dataset with distance values for each ion.

delocalization: (optional) NXprocess

Discretization of the ion point cloud on a three-dimensional grid.

input: (required) NX_CHAR

Delocalization in the field of atom probe microscopy is the process ...

Delocalization in the field of atom probe microscopy is the process of discretizing a point cloud. By default the tool computes a full kernel density estimation of decomposed ions to create one discretized field for each element.

Although, this uses an efficient multithreaded algorithm, the computation is costly. Therefore, it can be advantageous for users to load an already computed delocalization. This can be achieved with the load_existent option. When using this option the user is responsible to assure that the settings which were used for computing this already existent delocalization are specified in the same manner as they were.

Any of these values: default | load_existent

isotope_whitelist: (required) NX_UINT (Rank: 2, Dimensions: [n_ityp_deloc_cand, n_ivec]) {units=NX_UNITLESS}

Matrix of isotope vectors representing iontypes. ...

Matrix of isotope vectors representing iontypes. The filter specifies a matrix of isotope_vectors which is the most general approach to define if and how many times an ion is counted. Currently, paraprobe_nanochem performs a so-called atomic decomposition of all iontypes. Specifically, the tool interprets of how many elements/atoms a molecular ion is composed; and thus determines the atoms multiplicity with respect to the iontype.

Let’s take the hydroxonium H3O+ molecular ion as an example: It contains hydrogen and oxygen as atoms. The multiplicity of hydrogen is three whereas that of oxygen is one. Therefore in an atomic decomposition computation of the iso-surface each H3O+ ion adds three hydrogen counts. This is a practical solution which accepts the situation that during an atom probe experiment not each bond of each ion/a group of neighboring atoms is broken but molecular ions get detected. The exact ab-initio details depend on the local field conditions and thus also the detailed spatial arrangement of the atoms and their own electronic state and that of the neighbors before and upon launch. Being able to measure the information for such sites only as molecular ions causes an inherent information loss with respect to the detailed spatial arrangement. This information loss is more relevant for local electrode atom probe than for field ion microscopy setting how precisely the atomic positions can be reconstructed. Accounting for multiplicities assures that at least the compositional information is analyzed.

gridresolutions: (required) NX_FLOAT {units=NX_LENGTH}

List of individual grid resolutions to analyse. ...

List of individual grid resolutions to analyse. Paraprobe discretizes on a cuboidal 3D grid with cubic cells, with an edge length of values in gridresolutions.

kernel_size: (required) NX_UINT {units=NX_UNITLESS}

Half the width of a :math:`{(2 \cdot n + 1)}^3` cubic kernel of voxel ...

Half the width of a \({(2 \cdot n + 1)}^3\) cubic kernel of voxel beyond which the Gaussian Ansatz function will be truncated. Intensity beyond the kernel is refactored into the kernel via a normalization procedure.

kernel_variance: (required) NX_FLOAT {units=NX_LENGTH}

Variance of the Gaussian Ansatz kernel :math:`\sigma_x = \sigma_y = 2 \c ...

Variance of the Gaussian Ansatz kernel \(\sigma_x = \sigma_y = 2 \cdot \sigma_z\).

normalization: (required) NX_CHAR

How should the results of the kernel-density estimation be computed ...

How should the results of the kernel-density estimation be computed into quantities. By default the tool computes the total number (intensity) of ions or elements. Alternatively the tool can compute the total intensity, the composition, or the concentration of the ions/elements specified by the white list of elements in each voxel.

Any of these values:

  • total

  • candidates

  • composition

  • concentration

has_scalar_fields: (required) NX_BOOLEAN

Specifies if the tool should report the delocalization 3D field values.

isosurfacing: (optional) NXprocess

Optional computation of iso-surfaces after each computed delocalization ...

Optional computation of iso-surfaces after each computed delocalization to identify for instance objects in the microstructure (line features, interfaces, precipitates).

edge_handling_method: (required) NX_CHAR

As it is detailed in M. Kühbach et al. 2022 npj Comp. Mat., ...

As it is detailed in M. Kühbach et al. 2022 npj Comp. Mat., the handling of triangles at the edge of the dataset requires special attention. Especially for composition-normalized delocalization it is possible that the composition increases towards the edge of the dataset because the quotient of two numbers which are both smaller than one is larger instead of smaller than the counter. By default, the tool uses a modified marching cubes algorithm of Lewiner et al. which detects if voxels face such a situation. In this case, no triangles are generated for such voxels. Alternatively, (via setting keep_edge_triangles) the user can instruct the tool to not remove these triangles at the cost of bias.

Specifically, in this case the user should understand that all objects/microstructural features in contact with the edge of the dataset get usually artificial enlarged and their surface mesh often closed during the marching. This closure however is artificial! It can result in biased shape analyses for those objects. The reason why this should in general be avoided is a similar argument as when one analyzes grain shapes in orientation microscopy via e.g. SEM/EBSD. Namely, these grains, here the objects at the edge of the dataset, were not fully captured during e.g. limited field of view. Therefore, it is questionable if one would like to make substantiated quantitative statements about them.

Thanks to collaboration with the V. V. Rielli and S. Primig, though, paraprobe-nanochem implements a complete pipeline to process even these objects at the edge of the dataset. Specifically, the objects are replaced by so-called proxies, i.e. replacement objects whose holes on the surface mesh have been closed if possible via iterative mesh and hole-filling procedures with fairing operations. In the results of each paraprobe-nanochem run, these proxy objects are listed separately to allow users to quantify and analyze in detail the differences when accounting for these objects or not. Especially this is relevant in atom probe microscopy as objects can contain a few dozen atoms only. Users should be aware that results from fairing operations should be compared to results from analyses where all objects at the edge of the dataset have been removed.

Also users should be careful with overestimating the statistical significance of their dataset especially when using atom probe to compare multiple descriptors: Even though a dataset may give statistically significant results for compositions, this does not necessarily mean it will yield also statistically significant and unbiased results for three-dimensional object analyses. Being able to quantify these effects and making atom probers aware of these subtleties was one of the main reasons why the paraprobe-nanochem tool was implemented.

Any of these values: default | keep_edge_triangles

edge_threshold: (required) NX_FLOAT {units=NX_LENGTH}

The ion-to-edge-distance that is used in the analyses of objects ...

The ion-to-edge-distance that is used in the analyses of objects (and proxies) to identify whether these are inside the dataset or close to the edge of the dataset. If an object has at least one ion with an ion-to-edge-distance below this threshold, the object is considered as one which lies close to the edge of the dataset. This implements essentially a distance-based approach to solve the in general complicated and involved treatment of computing volumetric intersections between not-necessarily convex closed 2-manifolds. In fact, such computational geometry analyses can face numerical robustness issues as a consequence of which a mesh can be detected as lying completely inside a dataset although in reality it is epsilon-close only, i.e. almost touching only the edge (e.g. from inside). Practically, humans would state in such case that the object is close to the edge of the dataset; however mathematically the object is indeed completely inside. In short, a distance-based approach is rigorous and more flexible.

phi: (required) NX_FLOAT {units=NX_ANY}

Array of iso-contour values. For each value the tool computes ...

Array of iso-contour values. For each value the tool computes an iso-surface and performs subsequent analyses. The unit depends on the choice for the normalization of the accumulated ion intensity values per voxel:

  • total, total number of ions, irrespective their iontype

  • candidates, total number of ions with type in the isotope_whitelist.

  • composition, candidates but normalized by composition, i.e. at.-%

  • concentration, candidates but normalized by voxel volume, i.e. ions/nm^3

has_triangle_soup: (required) NX_BOOLEAN

Specifies if the tool should report the triangle soup which represents ...

Specifies if the tool should report the triangle soup which represents each triangle of the iso-surface complex. Each triangle is reported with an ID specifying to which triangle cluster (with IDs starting at zero) the triangle belongs. The clustering is performed with a modified DBScan algorithm.

has_object: (required) NX_BOOLEAN

Specifies if the tool should analyze for each cluster of triangles ...

Specifies if the tool should analyze for each cluster of triangles how they can be combinatorially processed to describe a closed polyhedron. Such a closed polyhedron (not-necessarily convex!) can be used to describe objects with relevance in the microstructure. Users should be aware that the resulting mesh does not necessarily represent the original precipitate. In fact, inaccuracies in the reconstructed positions cause inaccuracies in all downstream processing operations. Especially the effect on one-dimensional spatial statistics like nearest neighbor correlation functions these effects were discussed in the literature B. Gault et al. In continuation of these thoughts this applies also to reconstructed objects. A well-known example is the discussion of shape deviations of Al3Sc precipitates in aluminium alloys which in reconstructions can appear as ellipsoids although they should be almost spherical, depending on their size.

has_object_geometry: (required) NX_BOOLEAN

Specifies if the tool should report a triangulated surface mesh ...

Specifies if the tool should report a triangulated surface mesh for each identified closed polyhedron. It is common that a marching cubes algorithm creates iso-surfaces with a fraction of very small sub-complexes (e.g. small isolated tetrahedra).

These can be for instance be small tetrahedra/polyhedra about the center of a voxel of the support grid on which marching cubes operates. When these objects are small, it is possible that they contain no ion; especially when considering that delocalization procedures smoothen the positions of the ions. Although these small objects are interesting from a numerical point of view, scientists may argue they are not worth to be reported: Physically a microstructural feature should contain at least a few atoms to become relevant. Therefore, paraprobe-nanochem by default does not report closed objects which bound not at least one ion.

has_object_properties: (required) NX_BOOLEAN

Specifies if the tool should report properties of each closed ...

Specifies if the tool should report properties of each closed polyhedron, such as volume and other details.

has_object_obb: (required) NX_BOOLEAN

Specifies if the tool should report for each closed polyhedron an ...

Specifies if the tool should report for each closed polyhedron an approximately optimal bounding box fitted to all triangles of the surface mesh of the object and ion positions inside or on the surface of the mesh. This bounding box informs about the closed object’s shape (aspect ratios).

has_object_ions: (required) NX_BOOLEAN

Specifies if the tool should report for each closed polyhedron ...

Specifies if the tool should report for each closed polyhedron all evaporation IDs of those ions which lie inside or on the boundary of the polyhedron. This information can be used e.g. in the paraprobe-intersector tool to infer if two objects share common ions, which can be interpreted as an argument to assume that the two objects intersect.

Users should be aware that two arbitrarily closed polyhedra in three-dimensional space can intersect but not share a common ion. In fact, the volume bounded by the polyhedron has sharp edges. When taking two objects, an edge of one object may for instance pierce into the surface of another object. In this case the objects partially overlap / intersect volumetrically; however this piercing might be so small or happening in the volume between two ion positions and thus sharing ions is a sufficient but not a necessary condition for object intersections.

Paraprobe-intersector implements a rigorous alternative to handle such intersections using a tetrahedralization of closed objects. However, in many practical cases, we found through examples that there are polyhedra (especially when they are non-convex and have almost point-like) connected channels, where tetrahedralization libraries have challenges dealing with. In this case checking intersections via shared_ions is a more practical alternative.

has_object_edge_contact: (required) NX_BOOLEAN

Specifies if the tool should report if a (closed) object has ...

Specifies if the tool should report if a (closed) object has contact with the edge of the dataset. For this the tool currently inspects if the shortest distance between the set of triangles of the surface mesh and the triangles of the edge model is larger than the edge_threshold. If this is the case, the object is assumed to be deeply embedded in the interior of the dataset. Otherwise, the object is considered to have an edge contact, i.e. it is likely affected by the fact that the dataset is finite.

has_proxy: (required) NX_BOOLEAN

Specifies if the tool should analyze a doppelganger/proxy mesh for ...

Specifies if the tool should analyze a doppelganger/proxy mesh for each cluster of triangles whose combinatorial analysis according to has_object showed that the object is not a closed polyhedron. Such proxies are closed via iterative hole-filling, mesh refinement, and fairing operations. Users should be aware that the resulting mesh does not necessarily represent the original precipitate. In most cases objects, like precipitates in atom probe end up as open objects because they have been clipped by the edge of the dataset. Using a proxy is then a strategy to still be able to account for these objects. Nevertheless users should make themselves familiar with the potential consequences and biases which this can introduce into the analysis.

has_proxy_geometry: (required) NX_BOOLEAN

Like has_object_geometry but for the proxies.

has_proxy_properties: (required) NX_BOOLEAN

Like has_object_properties but for the proxies.

has_proxy_obb: (required) NX_BOOLEAN

Like has_object_obb but for the proxies.

has_proxy_ions: (required) NX_BOOLEAN

Like has_object_ions but for the proxies.

has_proxy_edge_contact: (required) NX_BOOLEAN

Like has_object_edge_contact but for the proxies.

has_object_auto_proxigram: (required) NX_BOOLEAN

Specifies if the tool should report for each closed object a ...

Specifies if the tool should report for each closed object a (cylindrical) region of interest placed, centered, and align with the local normal for each triangle of the object.

has_object_auto_proxigram_edge_contact: (required) NX_BOOLEAN

Specifies if the tool should report for each ROI that was placed ...

Specifies if the tool should report for each ROI that was placed at a triangle of each object if this ROI intersects the edge of the dataset. Currently paraprobe-nanochem supports cylindrical ROIs. A possible intersection of these with the edge of the dataset, i.e. the triangulated surface mesh model for the edge is performed. This test checks if the cylinder intersects with a triangle of the surface mesh. If this is the case, the ROI is assumed to make edge contact, else, the ROI is assumed to have no edge contact.

This approach does not work if the ROI would be completely outside the dataset. Also in this case there would be no intersection. For atom probe this case is practically irrelevant because for such a ROI there would also be no ion laying inside the ROI. Clearly it has thus to be assumed that the edge model culls the entire dataset. Instead, if one would cut a portion of the dataset, compute an edge model for this point cloud, it might make sense to place a ROI but in this case the edge contact detection is not expected to work properly.

interfacial_excess: (optional) NXprocess

Analyses of interfacial excess.

interface_model: (required) NX_CHAR

Interfacial excess computations are performed for local regions-of-int ...

Interfacial excess computations are performed for local regions-of-interests (ROIs) at selected facets or interface patch. For instance many scientist compute the interfacial excess for selected triangle facets of a created iso-surface. In this case, computed iso-surfaces of paraprobe could be used. An example are triangle facet sets about closed polyhedra, for instance to compute interfacial excess related to phase boundaries of second-phase precipitates.

Another example are free-standing triangle patches of the iso- surfaces which paraprobe creates. These could be characterized for interfacial excess. The sub-routines during iso-surface computations already include a procedure to automatically align local triangle normals based on the gradients of e.g. composition fields. In this case, these triangulated surface patches could also be used as a source for computing interfacial excess.

Often scientists face situations, though, in which there is no immediately evident composition gradient across the interface (grain or phase boundary) and orientation information about the adjoining crystal is neither available nor reliable enough.

In this case P. Felfer et al. proposed a method to manually place control points and run an automated tessellation-based algorithm to create a triangulated surface patch, i.e. a model of the location of the interface. In a post-processing step this triangle set can then be used to compute again interfacial excess in an automated manner by placing ROIs and aligning them with consistently precomputed triangle normals.

A similar use case is conceptually the one proposed by X. Zhou et al. They used first a deep-learning method to locate planar triangulated grain boundary patches. These are eventually processed further with manual editing of the mesh via tools like Blender. Once the user is satisfied with the mesh, the computations of interfacial excess reduce again to an automated placing of ROIs, computations of the distributing of ions to respective ROIs and reporting the findings via plotting.

Yet another approach for constructing an triangulated surface patch of an interface is to use point cloud processing methods which have been proposed in the laser-scanning, geoinformatics, and CAD community. Different computational geometry methods are available for fitting a parameterized surface to a set of points, using e.g. non-uniform rational B-splines (NURBS) and triangulating these according to prescribed mesh quality demands.

The advantage of these methods is that they can be automated and pick up curved interface segments. The disadvantage is their often strong sensitivity to parameterization. As a result also such methods can be post-processed to yield a triangulated surface patch, and thus enable to run again automated ROI placement methods. For example like these which were explored for the use case of iso-surfaces with closed objects and free-standing surface patches that delineate regions of the dataset with a pronounced composition gradient normal to the interface.

This summary of the situations which atom probers can face when requesting for interfacial excess computations, substantiates there exists a common set of settings which can describe all of these methods and, specifically, as here exemplified, the automated placing and alignment functionalities for ROIs that is an important step all these workflows.

Specifically, paraprobe-nanochem operates on an already existent triangle set.

Any of these values: isosurface | external

external: (optional) NXprocess

The interface model is the result of a previous (set of) processing ...

The interface model is the result of a previous (set of) processing steps as a result of which the user has created a triangulated surface mesh (or a set of, eventually connected such meshes). These interface models are useful, if not required, in cases when there is no other independent approach to locate an interface.

These are cases when insufficient crystallographic latent information is available and also no consistent concentration gradient detectable across the interface. It is then the users’ responsibility to deliver a triangle mesh of the interface model.

file_name: (required) NX_CHAR

Filename to HDF5 file which contain vertex coordinates, facet indice ...

Filename to HDF5 file which contain vertex coordinates, facet indices, facet unit normals. The user is responsible for the triangle and winding order to be consistent. Input is expected as a matrix of the coordinates for all disjoint vertices, a (Nvertices, 3)-shaped array of NX_FLOAT. Input is expected to include also a matrix of facet indices referring to these disjoint vertices. This matrix should be a (Nfacets, 3)-shaped array of NX_UINT. Further required input is a (Nfacets, 3)-shaped array of NX_FLOAT signed facet unit normals and a (Nvertices, 3)-shaped array of NX_FLOAT signed vertex unit normals. Vertex indices need to start at zero and must not exceed Nvertices - 1, i.e. the identifier_offset is 0 and facet indices are indexed implicitly, i.e. [0, Nvertices-1].

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

dataset_name_vertices: (required) NX_CHAR

Absolute HDF5 path to the dataset which specifies the ...

Absolute HDF5 path to the dataset which specifies the array of vertex positions.

dataset_name_facet_indices: (required) NX_CHAR

Absolute HDF5 path to the dataset which specifies the ...

Absolute HDF5 path to the dataset which specifies the array of facet indices.

dataset_name_facet_normals: (required) NX_CHAR

Absolute HDF5 path to the dataset which specifies the ...

Absolute HDF5 path to the dataset which specifies the array of facet signed unit normals.

dataset_name_facet_vertices: (required) NX_CHAR

Absolute HDF5 path to the dataset which specifies the ...

Absolute HDF5 path to the dataset which specifies the array of vertex signed unit normals.

Users should be aware that triangulated surface meshes are only approximations to a given complex, eventually curved shape. Consequently, computations of normals show differences between the vertex and facet normals. Vertex normals have to be interpolated from normals of neighboring facets. Consequently, these normals are affected by the underlying parameterization and curvature estimation algorithms, irrespective of how contributions from neighboring facets are weighted. By contrast, facet normals are clearly defined by the associated triangle. Their disadvantage is that they the normal field has discontinuities at the edges. In general the coarser an object is triangulated the more significant the difference becomes between computations based on facet or vertex normals. Paraprobe-nanochem works with facet normals as it can use parts of the numerical performance gained by using cutting edge libraries to work rather with finer meshes.

interface_meshing: (optional) NXprocess

Create a simple principle component analysis (PCA) to mesh a ...

Create a simple principle component analysis (PCA) to mesh a free-standing interface patch through a point cloud of decorating solutes. These models can be useful for quantification of Gibbsian interfacial excess for interfaces where iso-surface based methods may fail or closed objects from iso-surfaces are not desired or when e.g. there are no substantial or consistently oriented concentration gradients across the interface patch.

The interface_meshing functionality of paraprobe-nanochem can be useful when there is also insufficient latent crystallographic information available that could otherwise support modelling the interface, via e.g. ion density traces in field-desorption maps, as were used and discussed by Y. Wei et al. or are discussed by A. Breen et al.

It is noteworthy that the method here used is conceptually very similar in implementation to the work by Z. Peng et al. Noteworthy, her team uses the DCOM approach originally proposed by P. Felfer et al. However, both of these previous works neither discuss in detail nor implement inspection functionalities which enable a detection of potential geometric inconsistencies or self-interactions of the resulting DCOM mesh. This is what paraprobe-nanochem implements via the Computational Geometry Algorithms Library.

initialization: (required) NX_CHAR

Method how to initialize the PCA: ...

Method how to initialize the PCA:

  • default, means based on segregated solutes in the ROI

  • control_point_file, means based on reading an external list of control points, currently coming from the Leoben APT_Analyzer.

The control_point_file is currently expected with a specific format. The Leoben group lead by L. Romaner has developed a GUI tool A. Reichmann et al. to create a control_point_file which can be parsed by paraprobe-parmsetup to match the here required formatting in control_points.

Any of these values: default | control_point_file

filename: (required) NX_CHAR

The name of the control point file to use.

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

control_points: (required) NX_FLOAT (Rank: 2, Dimensions: [N, n_control_pts]) {units=NX_LENGTH}

X, Y, Z coordinates of disjoint control point read from ...

X, Y, Z coordinates of disjoint control point read from an HDF5 file named according to control_point_file.

method: (required) NX_CHAR

Method used for identifying and refining the location of the ...

Method used for identifying and refining the location of the interface. Currently, paraprobe-nanochem implements a PCA followed by an iterative loop of isotropic mesh refinement and DCOM step(s), paired with self-intersection detection in a more robust implementation.

Obligatory value: pca_plus_dcom

number_of_iterations: (required) NX_UINT {units=NX_UNITLESS}

How many times should the DCOM and mesh refinement be applied?

target_edge_length: (required) NX_FLOAT (Rank: 1, Dimensions: [n_fct_iterations]) {units=NX_LENGTH}

Array of decreasing positive not smaller than one nanometer real values ...

Array of decreasing positive not smaller than one nanometer real values which specify how the initial triangles of the mesh should be iteratively refined by edge splitting and related mesh refinement operations.

target_dcom_radius: (required) NX_FLOAT (Rank: 1, Dimensions: [n_fct_iterations]) {units=NX_LENGTH}

Array of decreasing positive not smaller than one nanometer real values ...

Array of decreasing positive not smaller than one nanometer real values which specify the radius of the spherical region of interest within which the DCOM algorithm decides for each vertex how the vertex will be eventually relocated. The larger the DCOM radius is relative to the target_edge_length the more likely it is that vertices will be relocated so substantially that eventually triangle self-intersections can occur. If the code detects these it warns and stops in a controlled manner so that the user can repeat the analyses with a smaller value.

target_smoothing_step: (required) NX_UINT (Rank: 1, Dimensions: [n_fct_iterations]) {units=NX_UNITLESS}

Array of integers which specify for each DCOM step how many times ...

Array of integers which specify for each DCOM step how many times the mesh should be iteratively smoothened.

Users should be aware the three array target_edge_length, target_dcom_radius, and target_smoothing_step are interpreted in the same sequence, i.e. the zeroth entry of each array specifies the values to be used in the first DCOM iteration. The first entry of each array those for the second DCOM iteration and so on and so forth.

decorating_iontypes_filter: (required) NXprocess

Specify the types of those ions which decorate the interface and ...

Specify the types of those ions which decorate the interface and can thus be assumed as markers for locating the interface and refining its local curvature.

candidates: (required) NX_UINT (Rank: 1, Dimensions: [n_fct_filter_cand]) {units=NX_UNITLESS}

Array of iontypes to filter. The list is interpreted as a whitelist, ...

Array of iontypes to filter. The list is interpreted as a whitelist, i.e. ions of these types are considered the decorating species (solutes).

composition_profiling: (optional) NXprocess

Functionalities for placing regions-of-interest (ROIs) in the dataset ...

Functionalities for placing regions-of-interest (ROIs) in the dataset or at specific microstructural features to characterize composition profiles and cumulated profiles for quantification of interfacial excess. Paraprobe-nanochem currently places cylindrical ROIs. ROIs are probed across the triangulated surface of a user-defined mesh. ROIs are placed at the barycenter of the triangular facet.

The tool can be instructed to orient the profile for each ROIs with the positive normal of the triangle facet normals. Profiles are computed for each ROI and facet triangle. The code will test which ROIs are completely embedded in the dataset. Specifically, in this test the tool evaluates if the ROI cuts at least one triangle of the triangulated surface mesh of the edge of the dataset. If this is the case the ROI will be considered close to the edge (of the dataset) and not analyzed further; else the ROI will be processed further. Users should be aware that the latter intersection analysis is strictly speaking not a volumetric intersection analysis as such one is much more involved because the edge model can be a closed non-convex polyhedron in which case one would have to test robustly if the cylinder pierces or is laying completely inside the polyhedron. For this the polyhedron has to be tessellated into convex polyhedra as otherwise tests like the Gilbert-Johnson-Keerthi algorithm would not be applicable.

Specifically, the tool computes atomically decomposed profiles. This means molecular ions are split into atoms/isotopes with respective multiplicity. As an example an H3O+ molecular ion contains three hydrogen and one oxygen atom respectively. The tool then evaluates how many ions are located inside the ROI or on the surface of the ROI respectively. All atom types and the unranged ions are distinguished. As a result, the analyses yield for each ROI a set of sorted lists of signed distance values. Currently, the distance is the projected distance of the ion position to the barycenter of the triangle and triangle plane.

This will return a one-dimensional profile. Post-processing the set of atom-type-specific profiles into cumulated profiles enable the classical Krakauer/Seidman-style interfacial excess analyses. Furthermore, the tool can be instructed to compute for each (or a selected sub-set of facet) a set of differently oriented profiles.

distancing_model: (required) NX_CHAR

Which type of distance should be reported for the profile. ...

Which type of distance should be reported for the profile.

Obligatory value: project_to_triangle_plane

direction_model: (required) NX_CHAR

In which directions should the tool probe for each ROI. ...

In which directions should the tool probe for each ROI.

Obligatory value: triangle_outer_unit_normal

roi_cylinder_height: (required) NX_FLOAT {units=NX_LENGTH}

For each ROI, how high (projected on the cylinder axis) ...

For each ROI, how high (projected on the cylinder axis) should the cylindrical ROI be.

roi_cylinder_radius: (required) NX_FLOAT {units=NX_LENGTH}

For each ROI, how wide (radius) should the cylindrical ROI be.

feature_mesh: (required) NXprocess

The feature mesh enables the injection of previously computed triangle ...

The feature mesh enables the injection of previously computed triangle soup or mesh data. Such a mesh can be the model for a grain- or phase boundary patch (from e.g. interface_meshing) jobs.

filename: (required) NX_CHAR

Name of the HDF5 file which contains vertex coordinates and facet ...

Name of the HDF5 file which contains vertex coordinates and facet indices to describe the desired set of triangles which represents the feature.

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which documents ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

dataset_name_vertices: (required) NX_CHAR

Absolute path to the HDF5 dataset in the respectively specified HDF5 ...

Absolute path to the HDF5 dataset in the respectively specified HDF5 file under filename which details the array of vertex positions.

dataset_name_facet_indices: (required) NX_CHAR

Absolute path to the HDF5 dataset in the respective specified HDF5 ...

Absolute path to the HDF5 dataset in the respective specified HDF5 file under filename which details the array of facet indices.

dataset_name_facet_normals: (required) NX_CHAR

Absolute path to the HDF5 dataset in the respective specified HDF5 ...

Absolute path to the HDF5 dataset in the respective specified HDF5 file under filename which details consistently oriented facet normals of the facets.

patch_identifier_filter: (optional) NXmatch_filter

method: (required) NX_CHAR

match: (required) NX_NUMBER

ion_to_feature_distances: (optional) NXprocess

The tool enables to inject precomputed distance information for each ...

The tool enables to inject precomputed distance information for each point which can be used for further post-processing and analysis.

filename: (required) NX_CHAR

Name of an HDF5 file which contains ion distances.

@version: (required) NX_CHAR

Version identifier of the file such as a secure hash which ...

Version identifier of the file such as a secure hash which documents the binary state of the file to add an additional layer of reproducibility from which file specifically contains these data.

dataset_name: (required) NX_CHAR

Absolute HDF5 path to the dataset with distance values for each ion.

performance: (required) NXcs_profiling

current_working_directory: (required) NX_CHAR

Hypertext Anchors

List of hypertext anchors for all groups, fields, attributes, and links defined in this class.

NXDL Source:

https://github.com/nexusformat/definitions/blob/main/contributed_definitions/NXapm_paraprobe_config_nanochem.nxdl.xml