Back to search

FRINATEK-Fri prosj.st. mat.,naturv.,tek

High-throughput alloy design of superior thermoelectric materials (Allotherm)

Alternative title: Systematisk søk etter bedre termoelektriske materialer ved legeringsutvikling

Awarded: NOK 12.1 mill.

One of the most pressing challenges of our society is to provide sustainable and secure access to energy resources for everyone. The largest unexploited source of energy is heat; more than 60% of global consumption is lost as heat, and huge primary sources like solar and geothermal heat are scarcely exploited. Thermoelectric (TE) generators have the potential to convert heat into electricity without problematic working fluids or moving parts. However, the current generation of TE generators is based on rare, expensive and toxic materials, and their efficiency is not high enough to be economically viable in many situations. The Allotherm project seeks to resolve this by identifying new, superior TE materials with increased efficiency, based on safe, available, and sustainable materials. This will be done by performing a high-throughput screening starting with all known and many hypothetical inorganic materials. So far, the theoretical part of the project has resulted in a wide range of predictions that could lead to new and better materials for thermoelectricity: 1. A screening study of more than 1,000 materials from the MaterialsProject database discovered eight new materials with a band gap, which is a prerequisite for a high thermoelectric effect. Of these, two materials were found to have promising thermoelectric properties. The study is published in Applied Physics Letters. 2. Another study developed a method to search for new materials with low thermal conductivity (which is beneficial for thermoelectricity) with an advanced machine learning algorithm. This was based on active sampling combined with principal component analysis, which makes it possible to use machine learning with a limited number of data points as input. The method was successfully demonstrated on 122 materials, all of which were identified with low thermal conductivity. This is published in Computational Materials Science. 3. An investigation of alloying materials at different crystallographic sites in the lattice was carried out on 122 half-Heusler alloys. These are cubic materials with three different sites with different elements; the general chemical formula is XYZ. Partial substitution of X, Y and Z was performed for all the materials and the thermal conductivity was predicted using quantum mechanical calculations. A set of general recommendations about such substitutions were formulated on the basis of the results, and new systems with potentially very low conductivity were identified. This is published in Electronic Materials. The experimental activity has started validating the theoretical predictions. Here, electrostatic doping of thin films has been used in addition to standard powder metallurgical production of polycrystalline materials and synthesis of single crystals. The first results (some of which are published in ACS Applied Electronic Materials) are positive, and efforts are underway to confirm the predictions experimentally.

One of the most pressing challenges of our society is to provide sustainable and secure access to energy resources for everyone. The largest unexploited source of energy is heat; more than 60% of the global consumption is lost as heat, and huge primary sources like solar and geothermal heat are not exploited. Thermoelectric (TE) generators have the potential to convert heat into electricity without harmful working fluids or moving parts. However, the current generation of TE generators are based on rare, expensive and toxic materials, and their efficiency is not high enough to be economically viable in many situations. The Allotherm project seeks to resolve this by identifying new, superior TE materials with increased efficiency, based on safe, available and sustainable materials. This will be done by performing a high-throughput screening starting with most known and many hypothetical inorganic materials. Pivotal in this development is alloy design, which in this context means substitution with isoelectronic elements (from the same group of elements). Thermoelectric properties will be predicted with first-principles calculations coupled with Boltzmann's transport equations. The screening will be step-wise, adding successively more advanced features to the selection criteria. Machine learning will be used to speed up the most computationally challenging part of the screening procedure, where the miscibility is predicted for all promising alloys. Novel use of online database resources and feedback loops will make this particularly efficient and precise. The final selection of materials systems will be based on accurate assessments of all relevant TE properties without any adjustable parameters. In addition, important features like maximum doping level and optimal dopant species will be predicted theoretically. The predictions will be tested experimentally with a range of techniques encompassing synthesis, processing, characterization and measurements.

Publications from Cristin

No publications found

Funding scheme:

FRINATEK-Fri prosj.st. mat.,naturv.,tek