Page 15 - IJAMD-2-2

International Journal of AI for
Materials and Design AI-driven material development for AM

Table 3. (Continued)
Name Description Material Source Availability
NIST DATA Comprehensive repository offering experimental Metals, polymers DFT, CALPHAD, https://www.nist.gov/data/
and computational material property data, MD, Monte (License required for some
including phase diagrams, thermophysical Carlo, FEA, datasets)
properties, and structural information experiment
NOMAD repository 49 Platform for computational materials science Metals DFT https://nomad-lab.eu/(Free
providing FAIR (Findable, Accessible, access)
Interoperable, Reusable) data sharing; includes
the world’s largest repository of computational
raw data, normalized into a code-independent
format, enabling data mining and machine
learning for materials discovery
NREL MatDB Computational materials database with a Metals DFT https://materials.nrel.gov/
specific focus on materials for renewable (Free access)
energy applications, including, but not limited
to, photovoltaic materials, materials for
photo-electrochemical water splitting, and
thermoelectric
OpenKIM 50 Curated repository of interatomic potentials Metals DFT, experiment https://openkim.org/(Free
and analytics for making classical molecular access)
simulations of materials reliable, reproducible,
and accessible
Predictive integrated Repository combining materials science data, Metals DFT, CALPHAD, http://www.prisms-center.
structural materials models, and workflows to support advanced experiment org/(Free access)
science simulations
The materials project 51 Open-access platform for computational Metals DFT https://materialsproject.org/
materials science that uses high-throughput (Free access)
calculations to compute and disseminate
properties of materials; provides structural,
electronic, and thermodynamic properties for
accelerating materials discovery and design
Total material Comprehensive materials information platform Metals, polymers Experiment https://www. totalmateria.
providing access to data on over 540,000 metallic com/(License required)
and non-metallic materials; includes extensive
mechanical and physical property data, global
standards and equivalencies, stress-strain, and
fatigue properties
The open quantum High-throughput database containing over Metals DFT https://oqmd.org/(Free
materials database 52 200,000 DFT calculations of crystal structures access)
and formation energies; serves as a resource for
materials discovery, providing thermodynamic
stability analysis, chemical potential data, and
datasets for training machine learning models
Abbreviations: CALPHAD: Calculation of phase diagrams; DFT: Density functional theory; FEA: Finite element analysis; MD: Molecular dynamics.

particularly when predicting material performance under remains an essential step for advancing AI-driven material
AM-specific conditions. However, these traditional development in this field.
datasets can still serve as initial guidance for material
selection and composition screening. By integrating such 4.2. Data pre-processing
data with AM-specific experimental results and process Data pre-processing is for ensuring that raw data are
parameters (e.g., laser power, scan speed, and cooling systematically refined into a structured format suitable
rates), AI models can be adapted to better reflect the for computational analysis and model training. Given
unique processing-structure-property relationships in AM. the inherent imperfections in raw datasets, such as noise,
Developing specialized AM databases that incorporate inconsistencies, and incomplete records, pre-processing
both material composition and AM process conditions plays a crucial role in mitigating these issues. The presence

Volume 2 Issue 2 (2025) 9 doi: 10.36922/IJAMD025100007