Electromechanical Properties · foundry-ml
piezoelectric_tensor_v1-1
A density functional theory dataset of piezoelectric tensors for 941 inorganic materials for machine learning.
What piezoelectric_tensor_v1-1 is
piezoelectric_tensor_v1-1 provides density functional theory (DFT) calculated electromechanical properties for 941 inorganic crystal structures. Developed by Maarten de Jong, Wei Chen, Henry Geerlings, Mark Asta, and Kristin A. Persson, the collection aids data-driven discovery and computational design of functional inorganic materials.
Piezoelectricity describes the linear coupling between mechanical strain and electrical charge. The dataset pairs crystallographic descriptors with complete piezoelectric tensors and related response metrics, providing structured targets and input features for training regression and classification algorithms aimed at sensor and actuator engineering.
How it was made
The dataset compiles first-principles electronic structure calculations for crystalline compounds indexed by Materials Project identifiers. For each material, atomic geometries are provided in multiple representations, including fractional coordinate structural representations, POSCAR files, Crystallographic Information Files (CIF), and Pymatgen structure objects, accompanied by unit cell volume, point groups, and space group numbers.
Target values detail the electromechanical response, featuring the full piezoelectric tensor formatted according to IEEE standards in units of coulombs per square metre (C/m2). Derived targets include the maximum longitudinal piezoelectric modulus, its base-10 logarithm, and the corresponding crystallographic direction vector.
Key facts
- Number of materials
- 941
- Calculation method
- Density functional theory (DFT)
- Target properties
- Piezoelectric tensor, maximum longitudinal modulus, crystallographic direction
- Units
- C/m2 for piezoelectric modulus and tensor
- Input structural formats
- CIF, POSCAR, Pymatgen structure
- Dataset splits
- train
- Release year
- 2022
- DOI
- 10.18126/p280-xrvg
What you can use it for
- Screening sensor and actuator materials
- Training regression models on crystallographic features to predict the maximum longitudinal piezoelectric response across candidate inorganic structures.
- Piezoelectric tensor prediction
- Developing multi-output machine learning architectures that map crystal symmetry and atomic coordinates directly to full IEEE-format piezoelectric tensors.
- Electromechanical response classification
- Categorising inorganic crystal structures into discrete classes based on predicted electromechanical modulus thresholds.
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/piezoelectric_tensor_v1-1 at the moment you asked. Nothing is cached or stored — every row above came from that request.
How the columns are distributed
Over 941 rows of default/train
cif
string_text- min
- 734
- median
- 1,022
- mean
- 1019.32
- max
- 1,314
eij_max
float- min
- 1.0e-5
- median
- 0.25
- mean
- 0.76
- max
- 46.19
formula
string_text- min
- 1
- median
- 6
- mean
- 5.99
- max
- 14
log(eij_max)
float- min
- -5.22
- median
- -0.61
- mean
- -0.64
- max
- 1.66
material_id
string_text- min
- 5
- median
- 8
- mean
- 8.20
- max
- 9
meta
string_text- min
- 1,134
- median
- 1,630
- mean
- 1623.95
- max
- 2,122
nsites
int- min
- 2
- median
- 8
- mean
- 7.98
- max
- 14
piezoelectric_tensor
string_text- min
- 135
- median
- 171
- mean
- 180.09
- max
- 267
point_group
string_label- mm219%
- -43m14%
- -42m11%
poscar
string_text- min
- 164
- median
- 356
- mean
- 353.47
- max
- 545
space_group
int- min
- 1
- median
- 121
- mean
- 108.74
- max
- 217
structure
string_text- min
- 241
- median
- 518
- mean
- 523.27
- max
- 794
v_max
string_text- min
- 15
- median
- 15
- mean
- 18.88
- max
- 49
volume
float- min
- 11.92
- median
- 132.55
- mean
- 143.01
- max
- 590.14
Strengths and limits
Good at
- Supplies full piezoelectric tensors in standardised IEEE format rather than only scalar approximations.
- Includes structural representations across multiple conventions, including POSCAR, CIF, and Pymatgen objects.
- Links calculated electromechanical tensors directly to Materials Project identifiers and crystallographic symmetry descriptors.
Watch out for
- Restricted in size to 941 inorganic materials.
- Contains density functional theory calculations rather than experimental measurements.
- Provided with a single default training partition without predefined validation or test sets.
Questions
Can I look inside piezoelectric_tensor_v1-1 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 piezoelectric_tensor_v1-1 under?
Its authors declare cc-by-4.0. Read the terms on the original page before reusing the data.
What target variables are included in the dataset?
The primary targets are the complete piezoelectric tensor in IEEE format (in C/m2), the maximum longitudinal piezoelectric modulus eij_max, its base-10 logarithm log(eij_max), and the crystallographic direction vector v_max corresponding to this maximum.
What structural file formats are provided for each compound?
Structures are provided as Pymatgen structure representations, Crystallographic Information Files (CIF), and VASP POSCAR files, alongside chemical formulas and unit cell volumes.
How are individual entries cross-referenced?
Each material entry includes a standard Materials Project identifier in the material_id field.
How to cite it
de Jong, Maarten and Chen, Wei and Geerlings, Henry and Asta, Mark and Persson, Kristin A., "A database to enable discovery and design of piezoelectric materials", Materials Data Facility, 2022. https://doi.org/10.18126/p280-xrvg
Written from piezoelectric_tensor_v1-1'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.