Metallic Glass · foundry-ml
dataset_metallicglass_dmax
Experimental maximum casting diameters and thermal properties for 998 metallic glasses to assess casting feasibility.
What dataset_metallicglass_dmax is
This dataset provides experimental property measurements for 998 metallic glasses, compiled by Paul M. Voyles, Lane E. Schultz, Dane Morgan, Carter Francis, Benjamin Afflerbach, and Abdulrhman Hakeem in 2021. Metallic glasses—amorphous metal alloys lacking long-range crystalline order—require rapid cooling during solidification to avoid crystallization.
The primary challenge in manufacturing bulk metallic glasses is achieving the critical cooling rate required for specific geometries and thicknesses. This collection records critical thermodynamic temperatures alongside experimental size metrics to support the screening and evaluation of casting feasibility via machine learning methods.
For materials scientists and computational researchers, the dataset serves as a benchmark for tabular regression and classification tasks, facilitating the predictive modelling of glass-forming ability directly from alloy composition and key transition temperatures.
How it was made
The data is structured as a tabular collection containing 998 entries, organised into a single training split. Each record documents an alloy's material composition alongside citation references pointing to the original measurement source.
The dataset tabulates three fundamental thermal parameters measured in kelvin: glass transition temperature (Tg), crystallization temperature (Tx), and liquidus temperature (Tl). These thermal properties, together with maximum casting size (Zmax in millimetres), serve as inputs to evaluate or predict the target property, the maximum casting diameter (Dmax in millimetres).
Key facts
- Dataset size
- 998 records
- Data format
- Tabular (Parquet)
- Target variable
- Dmax_[mm] (maximum casting diameter in mm)
- Input features
- Composition, Reference, Tg_[K], Tx_[K], Tl_[K], Zmax_[mm]
- Task types
- Tabular regression and tabular classification
- Curators
- Paul M. Voyles, Lane E. Schultz, Dane Morgan, Carter Francis, Benjamin Afflerbach, Abdulrhman Hakeem
- Release year
- 2021
- Digital Object Identifier
- 10.18126/fs5e-kr15
What you can use it for
- Casting Feasibility Screening
- Screening potential bulk metallic glass compositions using machine learning models to determine whether they can achieve critical amorphous casting diameters.
- Continuous Maximum Diameter Regression
- Training tabular regression models to predict the continuous numerical value of the maximum casting diameter (Dmax) from composition and transition temperatures.
- Glass-Forming Ability Classification
- Classifying metallic glass candidates into discrete categories based on critical dimension thresholds or glass-forming ability metrics.
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/dataset_metallicglass_dmax at the moment you asked. Nothing is cached or stored — every row above came from that request.
How the columns are distributed
Over 998 rows of default/train
Composition
string_text- min
- 7
- median
- 16
- mean
- 17.14
- max
- 38
Dmax_[mm]
float- min
- 0.03
- median
- 3.25
- mean
- 5.47
- max
- 150
Reference
string_label- 10.1016/j.commatsci.2016.02.03712%
- 10.1016/j.jallcom.2008.07.0877%
- 10.1016/j.msea.2006.06.0534%
Tg_[K]
float- min
- 293
- median
- 666
- mean
- 623.76
- max
- 952
11.5% empty
Tl_[K]
float- min
- 97
- median
- 1,141
- mean
- 1071.76
- max
- 1,559
17.5% empty
Tx_[K]
float- min
- 75
- median
- 723
- mean
- 672.05
- max
- 1,012
14.2% empty
Zmax_[mm]
float- min
- 0.50
- median
- 2
- mean
- 2.23
- max
- 6
96.7% empty
Strengths and limits
Good at
- Provides curated experimental measurements of maximum casting diameters across nearly a thousand distinct metallic glass formulations.
- Includes key characteristic temperatures (Tg, Tx, and Tl) directly tied to underlying crystallization and vitrification thermodynamics.
- Retains literature citations for the original experimental measurements across the sample records.
Watch out for
- Contains a relatively modest sample size of 998 records, requiring care to avoid overfitting during model training.
- The dataset is provided solely with a single pre-defined training split rather than standardised validation or test splits.
- Feature coverage is restricted to thermodynamic temperatures and casting size metrics, omitting detailed processing conditions such as cooling rates.
Questions
Can I look inside dataset_metallicglass_dmax 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 dataset_metallicglass_dmax under?
Its authors declare other. Read the terms on the original page before reusing the data.
What is the primary target variable in this dataset?
The target variable is Dmax_[mm], representing the experimental maximum casting diameter of the metallic glass in millimetres.
Which physical properties are included as input features?
The inputs include the alloy composition, original literature reference, glass transition temperature (Tg), crystallization temperature (Tx), liquidus temperature (Tl), and maximum casting size (Zmax).
What machine learning tasks is the data designed for?
It is formatted for both tabular regression, such as predicting continuous Dmax values, and tabular classification tasks related to casting feasibility.
How to cite it
Voyles, Paul M and Schultz, Lane E. and Morgan, Dane and Francis, Carter and Afflerbach, Benjamin and Hakeem, Abdulrhman, Metallic Glasses and their Properties, Materials Data Facility, 2021. https://doi.org/10.18126/fs5e-kr15
Written from dataset_metallicglass_dmax'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.