Diffusion Coefficients · foundry-ml
diffusion_v1-4
Tabular dataset of DFT-calculated solute migration barriers for 408 host-impurity pairs in dilute alloys.
What diffusion_v1-4 is
diffusion_v1-4 provides density functional theory (DFT) calculated dilute alloy impurity diffusion barriers across 408 host-impurity pairs. Developed by Haijin Lu, Nan Zou, Ryan Jacobs, Ben Afflerbach, Xiao-Gang Lu, and Dane Morgan, it supports machine learning modelling of solid-state ion transport and inorganic systems.
Understanding diffusion behaviour is critical for alloy development and electrolyte design. This dataset was constructed to investigate error assessment and evaluate optimal cross-validation approaches when machine learning methods are applied to predict impurity diffusion parameters.
How it was made
The dataset tabulates dilute alloy systems where a solute element migrates through a host element. For each of the 408 pairs, the solute migration barrier was calculated using density functional theory (DFT) and recorded as a relative energy shift measured in electronvolts (eV).
Data records are organised into tabular features comprising host element compositions, solute element compositions, and the calculated migration energy, prepared for tabular regression or classification tasks.
Key facts
- DOI
- 10.18126/uppe-p8p1
- Dataset size
- 408 host-impurity pairs
- Input fields
- Material compositions 1 (host) and Material compositions 2 (solute)
- Target field
- E_regression_shift (DFT-calculated solute migration barrier in eV)
- Format
- Tabular / Parquet
- Data splits
- train
What you can use it for
- Impurity migration prediction
- Train tabular regression models to predict solute migration energy barriers in dilute alloys directly from host and solute elemental pairs.
- Cross-validation benchmarking
- Assess error estimation techniques and evaluate different cross-validation strategies tailored for materials property datasets.
- Solid-state transport screening
- Identify elemental pairs with favourable diffusion characteristics to inform the design of solid electrolytes and structural alloys.
Explore the data
The schema, real rows, full-text search and a SQL filter — read in your browser, nothing downloaded.
Explore
Read the real rows without downloading anything
Reading rows…
LiveRead from foundry-ml/diffusion_v1-4 at the moment you asked. Nothing is cached or stored — every row above came from that request.
How the columns are distributed
Over 408 rows of default/train
E_regression_shift
float- min
- -0.94
- median
- -0.06
- mean
- 0.09
- max
- 1.94
Material compositions 1
string_label- Al13%
- Mg12%
- Ni10%
Material compositions 2
string_label- Ag4%
- Zr3%
- Cd3%
Strengths and limits
Good at
- Features consistent DFT-calculated migration barrier benchmarks across 408 distinct host-impurity elemental combinations.
- Formatted in a straightforward tabular structure compatible with common machine learning and data science libraries.
- Provides continuous target values in standard physical units (eV) suitable for regression and classification.
Watch out for
- Coverage is limited to 408 host-solute pairs in dilute alloy regimes.
- The dataset provides only a single train partition without pre-defined validation or test splits.
- Restricted to solute migration barriers rather than complete diffusion profiles across varying temperatures and concentrations.
Questions
Can I look inside diffusion_v1-4 without downloading it?
Yes. This page reads its schema, real rows and per-column statistics in the browser, with full-text search and a SQL filter. Nothing is downloaded to your machine.
What licence is diffusion_v1-4 under?
Its authors declare cc-by-4.0. Read the terms on the original page before reusing the data.
What target property is measured in this dataset?
The target property is E_regression_shift, representing the DFT-calculated solute migration barrier relative energy given in electronvolts (eV).
What input features are included?
The inputs are Material compositions 1, representing the host element, and Material compositions 2, representing the solute element.
How many records are available?
The dataset contains 408 dilute alloy host-impurity pairs.
What formats can the data be accessed in?
The dataset is provided as tabular data formatted in Parquet and can be loaded via data frameworks such as pandas, polars, and datasets.
How to cite it
Lu, Haijin and Zou, Nan and Jacobs, Ryan and Afflerbach, Ben and Lu, Xiao-Gang and Morgan, Dane (2022). Error assessment and optimal cross-validation approaches in machine learning applied to impurity diffusion. Materials Data Facility. https://doi.org/10.18126/uppe-p8p1
Written from diffusion_v1-4's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.