Literature review of databases with materials and .
See literature review on ML models which utilize these datasets:
Literature review of existing studies done on predicting with machine learning.
https://supercon.nims.go.jp/ (website down?)
https://doi.org/10.48505/nims.3739 (also down?)
Advantages:
Largest experimental Tc dataset
Includes preparation conditions
Regular updates
Free access
Disadvantages:
No crystal structures
Only chemical compositions
Some incomplete entries
Data format can be inconsistent
https://github.com/aimat-lab/3DSC
Data-driven methods, in particular machine learning, can help to speed up the discovery of new materials by finding hidden patterns in existing data and using them to identify promising candidate materials. In the case of superconductors, the use of data science tools is to date slowed down by a lack of accessible data. In this work, we present a new and publicly available superconductivity dataset (‘3DSC’), featuring the critical temperature Tc of superconducting materials additionally to tested non-superconductors.
In this paper, they introduce and analyze two different 3DSC databases. Both are based on the SuperCon database, but one uses structures from the Materials Project (3DSCMP) and one uses structures from the ICSD (3DSCICSD). Using their matching and adaptation algorithm, they are able to match 5,759 (3DSCMP) and 9,150 (3DSCICSD) superconducting and non-superconducting materials from the SuperCon.
In addition to matching only exact chemical compositions (as in Stanev et al.7), they employ a systematic adaptation algorithm that approximates the three-dimensional crystal structures of materials without perfect match by artificial doping of similar crystal structures.
Published: 21 November 2023
Does a good literature review of other datasets and attempts at predicting Tc
3DSC is augmented by approximate three-dimensional crystal structures
Builds on Stanev et al. Machine learning modeling of superconducting critical temperature
Disadvantages:
Approximate structures (computationally generated)
Smaller than SuperCon
Limited to specific material classes
https://github.com/Gashmard/DataG_13022_superconducting_materials
This study employs the SuperCon dataset as the largest superconducting materials dataset. Then, we perform various data pre-processing steps to derive the clean DataG dataset, containing 13,022 compounds. In another stage of the study, we apply the novel CatBoost algorithm to predict the transition temperatures of novel superconducting materials. In addition, we developed a package called Jabir, which generates 322 atomic descriptors. We also designed an innovative hybrid method called the Soraya package to select the most critical features from the feature space. These yield R2 and RMSE values (0.952 and 6.45 K, respectively) superior to those previously reported in the literature. Finally, as a novel contribution to the field, a web application was designed for predicting and determining the Tc values of superconducting materials.
The dataset (DataG) is prepared after various steps of data pre-processing on the SuperCon dataset.
Published 2024-02-17
Contains 13,022 superconducting compounds
Chemical composition to element-based atomic features, but did not attempt crystal structure to electric properties like 3DSC
I don't think this was a very good study. Not very rigorous and not very innovative. Read it for yourself, but I don't think this paper should hold as much weight or reverence as some of the others.
Advantages:
Comprehensive computed properties
Well-documented API
Regular updates
Includes electronic structure
Disadvantages:
Primarily computational data
May miss experimental nuances
DFT limitations
Can be computationally expensive to regenerate results
https://icsd.fiz-karlsruhe.de/
Advantages:
Verified crystal structures
High-quality experimental data
Comprehensive structural info
Disadvantages:
Expensive subscription required
Not superconductor-specific
Limited computational properties
No direct Tc data
Advantages:
Standardized calculations
Robust API
High-throughput ready
Includes many properties
Disadvantages:
Primarily computational
May miss experimental features
Limited to specific property types
Can be overwhelming for beginners
This post will focus on the methods available to predict/derive of a material. We want to be able to build a pipeline where we can go beyond the available (and experimental) Tc data and train a model