README.md

metadata

license: cc-by-4.0
task_categories:
  - tabular-regression
  - tabular-classification
tags:
  - materials-science
  - chemistry
  - foundry-ml
  - scientific-data
size_categories:
  - 1K<n<10K

Machine learning for impurity charge-state transition levels in semiconductors from elemental properties using multi-fidelity datasets

Dataset containing DFT-calculated defect charge state transition levels of 2910 semiconductor-impurity pairs

Dataset Information

• Source: Foundry-ML
• DOI: 10.18126/ite0-3iah
• Year: 2022
• Authors: Polak, Maciej P., Jacobs, Ryan, Mannodi-Kanakkithodi, Arun, Chan, Maria K. Y., Morgan, Dane
• Data Type: tabular

Fields

Field	Role	Description	Units
host_material	input	Composition of host
host_element_a	input	Host element of A site
host_element_b	input	Host element of B site
impurity	input	Impurity element type
removed_element	input	Removed element type
site_type	input	Type of the lattice site the impurity resides
site	input	Name of the lattice site the impurity resides
is_a_latt	input	Whether impurity resides on the A sublattice
is_interstitial	input	Whether impurity is on an interstitial site
is_interstitial_a	input	Whether impurity is on a A-site interstitial site
is_interstitial_b	input	Whether impurity is on a B-site interstitial site
is_interstitial_n	input	Whether impurity is on a neutral interstitial site
M_Al	input	One-hot encoding of site type
M_As	input	One-hot encoding of site type
M_Cd	input	One-hot encoding of site type
M_Ga	input	One-hot encoding of site type
M_In	input	One-hot encoding of site type
M_P	input	One-hot encoding of site type
M_S	input	One-hot encoding of site type
M_Sb	input	One-hot encoding of site type
M_Se	input	One-hot encoding of site type
M_Te	input	One-hot encoding of site type
M_i_Cd_site	input	One-hot encoding of site type
M_i_S_site	input	One-hot encoding of site type
M_i_Se_site	input	One-hot encoding of site type
M_i_Te_site	input	One-hot encoding of site type
M_i_neut_site	input	One-hot encoding of site type
charge_from	input	Initial charge of defect	electrons
charge_to	input	Final charge of defect	electrons
host bandgap_[eV]	input	Experimental bandgap of the host material	eV
host lattice constant_[Ang.]	input	Lattice constant of the host material	Angstroms
host_epsilon	input	Dielectric constant of the host material
ba_shift	input	Band alignment correction shift	eV
pbe defect level (relative to VBM)_[eV]	target	DFT-PBE calculated defect charge state transition	eV
mba_pbe	input	Modified band alignment correction shift	eV
mba_pbe_gapfrac	input	Modified band alignment correction shift, given as
hse defect level (relative to VBM)_[eV]	target	DFT-HSE calculated defect charge state transition	eV

Splits

• train: train

Usage

With Foundry-ML (recommended for materials science workflows)

from foundry import Foundry

f = Foundry()
dataset = f.get_dataset("10.18126/ite0-3iah")
X, y = dataset.get_as_dict()['train']

With HuggingFace Datasets

from datasets import load_dataset

dataset = load_dataset("semiconductor_defectlevels_v1.1")

Citation

@misc{https://doi.org/10.18126/ite0-3iah
doi = {10.18126/ite0-3iah}
url = {https://doi.org/10.18126/ite0-3iah}
author = {Polak, Maciej P. and Jacobs, Ryan and Mannodi-Kanakkithodi, Arun and Chan, Maria K. Y. and Morgan, Dane}
title = {Machine learning for impurity charge-state transition levels in semiconductors from elemental properties using multi-fidelity datasets}
keywords = {machine learning, foundry}
publisher = {Materials Data Facility}
year = {root=2022}}

License

CC-BY 4.0

---

This dataset was exported from Foundry-ML, a platform for materials science datasets.