Experimentally validated machine learning predictions of ultralow thermal conductivity for SnSe materials

Barua, Niloy; Golabek, Andrew; Oliynyk, Anton O.; Kleinke, Holger

doi:10.1039/d3tc01450a

Cited by 6 publications

(2 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides these studies, a recent ML study from our group looking at SnSe-based TE materials using highquality experimental data included elemental vector matrix, weighted multivariable, and synthesis features to predict ultralow κ (<1 W m −1 K −1 ) for new doped compositions of SnSe. 22 Furthermore, Li et al developed a model to predict the zT of TE materials using a dataset size of 5038 with a wide variety of TE materials and 57 weighted multivariable features. 23 The weighted multivariable features are statistical calculations using the maximum, minimum, average, standard deviation, etc., of elemental features.…”

Section: Introductionmentioning

confidence: 99%

“…This reference provides a nice overview of ML studies on thermal conductivity; those studies are either based on experimental data with less than 750 entries or on large crystallographic data from which the thermal properties were simply calculated by different means. Besides these studies, a recent ML study from our group looking at SnSe-based TE materials using high-quality experimental data included elemental vector matrix, weighted multivariable, and synthesis features to predict ultralow κ (<1 W m −1 K −1 ) for new doped compositions of SnSe …”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Interpretable Machine Learning Model on Thermal Conductivity Using Publicly Available Datasets and Our Internal Lab Dataset

Barua,

Hall,

Cheng

et al. 2024

Chem. Mater.

Self Cite

View full text Add to dashboard Cite

Machine learning (ML), a subdiscipline of artificial intelligence studies, has gained importance in predicting or suggesting efficient thermoelectric materials. Previous ML studies have used different literature sources or density functional theory calculations as input. In this work, we develop a ML pipeline trained with multivariable inputs on a massive public dataset of ∼200,000 data utilizing a high-performance computing cluster to predict the thermal conductivity (κ) using four test sets: three publicly available datasets and a dataset built using previously published data from our own group. By taking advantage of this massive dataset, our model presents an opportunity to further expand the understanding of the selection of features with various thermoelectric materials. Among the several supervised ML models implemented, the eXtreme Gradient Boosting algorithm (XGBoost) turned out to be the best method during the 5-fold cross-validation method, with their averaged evaluation coefficients of R 2 = 0.96, root mean squared error (RMSE) = 0.38 W m −1 K −1 , and mean absolute error (MAE) = 0.23 W m −1 K −1 . Additionally, with the aid of feature selection and importance analysis, useful chemical features were chosen that ultimately led to reasonably good accuracy in the series of test sets measured as per the evaluation coefficients of R 2 , RMSE, and MAE, with values ranging from 0.72 to 0.89, 0.52 to 1.08, and 0.40 to 0.66 W m −1 K −1 , respectively. Checking the worst outliers led to the discovery of some errors in the literature. Postmodel prediction, the SHapley Additive exPlanations (SHAP) algorithm was implemented on the XGBoost model to analyze the features that were the key drivers for the model's decisions. Overall, the developed interpretable methodology produces the prediction of κ of a large variety of materials through the influence of chemical and physical property features. The conclusions drawn apply to the research and applications of thermoelectric and heat insulation materials.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Interpretable Machine Learning Model on Thermal Conductivity Using Publicly Available Datasets and Our Internal Lab Dataset

Barua,

Hall,

Cheng

et al. 2024

Chem. Mater.

Self Cite

View full text Add to dashboard Cite

show abstract

Machine learning for next-generation thermoelectrics

Saglik,

Srinivasan,

Victor

et al. 2024

Materials Today Energy

View full text Add to dashboard Cite

Insights into Thermal Conductivity of Pnictogen Chalcogenides: Machine Learning Stereochemically Active Lone Pairs and Hybridization

Minhas,

Jena,

Sharma

et al. 2024

Chem. Mater.

View full text Add to dashboard Cite

Stereochemically active lone pairs (SCALPs) are pivotal in influencing the lattice thermal conductivity (κ L ), representing a critical aspect in formulating strategies for achieving high thermoelectric performance. Despite the transformative potential of the material genome paradigm for screening materials with tailored properties, accurately describing SCALPs in terms of performance indicators remains a challenge. In this machine learning (ML) study, we introduce specialized chemical bonding descriptors that capture the empirical hidden influence of SCALP and chemical bonding hierarchies in pnictogen chalcogenide materials. The ML model, trained with screened data sets from the Open Quantum Materials Database, the Materials Project, and experimental reports, achieved a significant reduction in test error scores by using chemical bonding descriptors over conventional features in predicting κ L values for pnictogen chalcogenides. We predict five materials, MnTl 2 As 2 S 5 , Ba 2 As 2 Se 5 , Bi 14 Te 13 S 8 , AgCu 2 PbBiS 4 , and Tl 2 SnAs 2 S 6 , exhibiting ultralow κ L values of ≤0.40 W m −1 K −1 at room temperature. Additionally, we specified the precise ranges for ionicity, hybridization, number mismatch, and polarizability required for ultralow κ L for 245 newly predicted materials. Our data-driven approach not only identifies promising candidates with ultralow κ L but also reveals new avenues for the design of pnictogen-based thermoelectric materials, emphasizing the crucial influence of lone pairs and hybridization.

show abstract

Experimentally validated machine learning predictions of ultralow thermal conductivity for SnSe materials

Abstract: Machine-learning (ML) models are used to predict optimal thermoelectric properties for efficient thermoelectric devices.

Cited by 6 publications

References 61 publications

Interpretable Machine Learning Model on Thermal Conductivity Using Publicly Available Datasets and Our Internal Lab Dataset

Interpretable Machine Learning Model on Thermal Conductivity Using Publicly Available Datasets and Our Internal Lab Dataset

Machine learning for next-generation thermoelectrics

Insights into Thermal Conductivity of Pnictogen Chalcogenides: Machine Learning Stereochemically Active Lone Pairs and Hybridization

Contact Info

Product

Resources

About