zTSkut logo

zTSkut: Predict zT for skutterudite systems

A neural-network model for single and batch prediction of thermoelectric performance in skutterudite-based compositions. Empty optional fields are automatically treated as 0.

Composition note: Use measured or experimentally determined stoichiometric fractions, not nominal starting or target synthesis fractions. Nominal values may give unreliable predictions.

If you use this model, please cite the original paper:

Single-system prediction

Enter the composition, carrier concentration, temperature and carrier type. Leave unused composition/fraction boxes empty.

Fillers
Cations
Anions
Global parameters
⏳ Predicting zT... Please wait.
Prediction result will appear here.

Batch prediction

Download CSV template

Upload a CSV, JSON or Excel (.xlsx) file to predict zT for multiple skutterudite candidates in one run. Files should use measured or experimentally determined stoichiometric fractions, not nominal synthesis fractions.

Batch prediction results will appear here.

Required field format

a_comp_1,a_frac_1,a_comp_2,a_frac_2,a_comp_3,a_frac_3,
c_comp_1,c_frac_1,c_comp_2,c_frac_2,c_comp_3,c_frac_3,
f_comp_1,f_frac_1,f_comp_2,f_frac_2,f_comp_3,f_frac_3,
f_comp_4,f_frac_4,f_comp_5,f_frac_5,
n_300,T,p_n,ZT

p_n: 0 = p-type, 1 = n-type. ZT is a dummy field required for input formatting and is ignored by the model.