DrAdrianDC/DFT-and-ML-rocksalt-oxides

Dataset Card for Rocksalt Complex Oxides Hydrogen Adsorption (DFT & ML)

Dataset Summary

This dataset contains 336 density functional theory (DFT) calculated observations of hydrogen adsorption energies on rocksalt-structure complex oxides, accompanying electronic structure properties, and local chemical coordination descriptors.

The dataset was developed to train and evaluate supervised machine learning regression models capable of predicting surface reactivity and chemisorption behavior without invoking computationally expensive DFT electronic geometry relaxations for every local cation arrangement.

• Primary Repository: GitHub Repository
• Publication: Theoretical Chemistry Accounts (2024) 143:50
• DOI: 10.1007/s00214-024-03124-x
• Author / Point of Contact: Dr. Adrian Dominguez-Castro

---

Dataset Structure

The dataset consists of a single standardized table (Data.csv) containing 336 rows and 15 columns.

Detailed Feature Schema

Column Name	Data Type	Physical Unit	Domain Range	Description
Eads	float64	$\text{eV}$	Continuous	Target Variable. Calculated hydrogen adsorption energy defined as $E_{\text{ads}} = E_{\text{surf+H}} - E_{\text{surf}} - \frac{1}{2}E_{\text{H}_2}$.
Bader-charge	float64	$e$	Continuous	Partial charge on the active surface adsorption oxygen site, computed via Bader charge analysis.
Ave-O2p-up	float64	$\text{eV}$	Continuous	Average energy center of the spin-up Oxygen $2p$ valence states relative to the Fermi level.
Ave-O2p-down	float64	$\text{eV}$	Continuous	Average energy center of the spin-down Oxygen $2p$ valence states relative to the Fermi level.
Ave-diff-EN	float64	Dimensionless	Continuous	Average Pauling electronegativity difference ($\Delta\text{EN}$) between the active surface site and neighboring metal cations.
Ave-diff-IE	float64	$\text{eV}$	Continuous	Average first ionization energy difference ($\Delta\text{IE}$) among constituent cations in the local chemical environment.
BLVE Neighbor 1	float64	Arb. Units	Continuous	Bond-Length Valence-Electron descriptor for the 1st nearest neighbor coordination sphere.
BLVE Neighbor 2	float64	Arb. Units	Continuous	Bond-Length Valence-Electron descriptor for the 2nd nearest neighbor coordination sphere.
BLVE Neighbor 3	float64	Arb. Units	Continuous	Bond-Length Valence-Electron descriptor for the 3rd nearest neighbor coordination sphere.
BLVE Neighbor 4	float64	Arb. Units	Continuous	Bond-Length Valence-Electron descriptor for the 4th nearest neighbor coordination sphere.
BLVE Neighbor 5	float64	Arb. Units	Continuous	Bond-Length Valence-Electron descriptor for the 5th nearest neighbor coordination sphere.
Freq Ni	int64	Count	$[0, 5]$	Absolute frequency of Nickel ($\text{Ni}$) cations in the immediate local coordination environment.
Freq Mg	int64	Count	$[0, 5]$	Absolute frequency of Magnesium ($\text{Mg}$) cations in the immediate local coordination environment.
Freq Cu	int64	Count	$[0, 5]$	Absolute frequency of Copper ($\text{Cu}$) cations in the immediate local coordination environment.
Freq Zn	int64	Count	$[0, 5]$	Absolute frequency of Zinc ($\text{Zn}$) cations in the immediate local coordination environment.

---

Technical Specifications & Integrity

• File Encoding: UTF-8
• Delimiter: Comma (,)
• Decimal Mark: Dot (.)
• Missing Values: 0 null / NA entries across all 336 rows.
• Precision: IEEE 754 64-bit floating point preservation.

---

Usage in Python (datasets)

from datasets import load_dataset

# Load dataset directly from Hugging Face Hub
dataset = load_dataset("DrAdrianDC/DFT-and-ML-rocksalt-oxides")

# Convert to Pandas DataFrame for ML pipeline processing
df = dataset["train"].to_pandas()
print(f"Loaded dataset shape: {df.shape}")
print(df.head())