New Algorithm Can Speed Up Material Discovery By 1000 Times

  • A new algorithm can discovery thermoelectric materials by solving quantum mechanical equations. 
  • It works by predicting material’s electronic transport properties. 
  • Using this method, experimentalists discovered a material that was both stable and efficient.

Discovering new materials requires precise computation of basic physical properties and identification of techniques that control performance. This enables fast examination of material candidates.

The approach is even more difficult in the field of thermoelectric materials, where both predicting and measuring high temperature transport properties is necessary.

In almost all cars, more than half of total energy produced by gasoline is lost through heat in radiator and exhaust pipe. To deal with this, researchers at the Harvard University are trying to build a new thermoelectric material that can efficiently transform heat energy into electrical energy. Ultimately, this could enhance the fuel efficiency by 5%.

However, this is not as simple as it sounds. The existing thermoelectric materials for recovering waste heat takes too much time to develop and it’s also very expensive. In fact, one of the advanced material made of zirconium and hafnium –commonly used in nuclear reactors — took 15 years to reach its optimal performance-phase from the time it was first discovered.

To reduce this time, engineers have made an algorithm that takes months, instead of years, to find and optimize material for better energy conversion. The algorithm works by solving quantum mechanical equations and doesn’t require any sample or experimental inputs.

What Exactly They Have Done?

The algorithm they’ve developed is capable of predicting material’s electronic transport properties based on chemical elements of the crystalline crystal. The main objective was to speed up the process by 10,000 times by using first-principle computations of electron-phonon scattering.

Using this algorithm they screened several combinations of crystal structures that had never been synthesized in the past. They whitelisted some potential candidates and did further optimization to select top performer.

Using the data — provided by the algorithmic computations  — experimentalists synthesized these top performers and discovered a material that was both stable and efficient. In fact, it was 10 times cheaper than previous thermoelectric materials. The overall process took total 15 months — instead of 15 years — to get to the final optimized material.

Reference: Advanced Energy Material | doi:10.1002/aenm.201800246 | Harvard SEAS 


More specifically, they showed that the energy dependence of the electron relaxation time in thermoelectric materials can significantly affect their transport properties, including the Lorenz number and Seebeck coefficient, which are typically considered to be independent of relaxation time.

By directly measuring the electrical parts of the thermal conductivities, they detected deviations from Wiedemann– Franz law in these materials at low carrier concentrations and high temperatures. This indicates all possible risks involved in usual procedures used to interpret outcomes of thermal transport and electronic computations.

Image Credit: Second Bay Studios / Harvard SEAS

Researchers demonstrated that the whole complexity of electron-phonon coupling matrix isn’t required to measure the precise relaxation times and electronic transport coefficients in thermoelectric materials. Furthermore, they figured out that the effective mass of electron is a useful qualitative descriptor of thermoelectric performance. This is can be used to screen and prioritize materials.

Read: A New Material That Could Prevent Overheating of Phones And Laptops

What’s Next?

Currently, researchers are working to make this technique even cheaper and faster. It opens opportunities to understand intrinsic transport properties of complex semiconductor and enable different computational material design in a wide range of technological applications.

Leave a reply