Skip to main content

Advanced Simulation Design of Nanostructured Thermoelectric Materials with Enhanced Power Factors

Periodic Reporting for period 3 - NANOthermMA (Advanced Simulation Design of Nanostructured Thermoelectric Materials with Enhanced Power Factors)

Reporting period: 2019-07-01 to 2020-12-31

Roughly one-third of all energy consumption ends up as low-grade heat. Thermoelectric (TE) materials could potentially convert vast amounts of this waste heat into electricity and reduce the dependence on fossil fuels. State-of-the-art nanostructured materials with record-low thermal conductivities (κ~1-2W/mK) have recently demonstrated large improvements in conversion efficiencies, but not high enough to enable large scale implementation. Central to this low efficiency problem lies the fact that the Seebeck coefficient (S) and the electrical conductivity (σ), the parameters that determine the TE power factor (σS2), are inversely related. Relaxing this inverse interdependence has never been achieved, and TE efficiency remains low. The project is an integrated project, consisting of large scale simulations and experiments, and utilizes nanostructured designs to relax the adverse interdependence of the σ and S, and achieve unprecedentedly large power factors, thus, enabling the thermoelectric technology to achieve high conversion efficiencies. High thermoelectric efficiencies will allow waste heat energy harvesting, and contribute to the reduction of the use of fossil fuels and promote energy sustainability.
The project focuses around four ambitious objectives: i) Theoretically establishes and generalize the strategies that relax the adverse interdependence of σ and S in nanostructures and achieve power factors >5× compared to the state-of-the-art; ii) Experimentally validate the theoretical propositions through well-controlled material design examples; iii) Provide a predictive, state-of-the-art, high-performance, electro-thermal simulator to generalize the concept and guide the design of the entirely new nanostructured TE materials proposed. Appropriate theory and techniques are being developed so that the tool includes all relevant nanoscale transport physics to ensure accuracy in predictions. Simulation capabilities for a large selection of materials and structures are being included; iv) Develop robust, ‘inverse-design’ optimization capabilities within the simulator, targeting maximum performance. In the long run, the simulator could evolve as a core platform that impacts many different fields of nanoscience as well.
The work objectives revolve around three main categories, and the reporting follows achievements in those categories (1-high power factor and high thermoelectric efficiency strategies, 2-simulator development, 3-experimental validation of the strategies):

High power factor and high thermoelectric efficiency strategies

1. The project has already provided parts of the full design guidelines on how to nanostructure a material in a hierarchical manner to provide very high power factors. The method has been published in Physical Review B, Physica Status Solidi A, Journal of Applied Physics, and Journal of Electronic Materials. We were able to link the geometrical features with the doping densities, as well as internal material properties and parameters for maximizing the power factor.

2. The project has also demonstrated that hierarchically nanostructured materials, allow for much lower thermal conductivity reductions compared to what more simplified methods predict, provided insight for the reasons behind this, and developed much better compact models than existing ones, which take into account the effect of disorder in thermal conductivity. The results were published in Physical Review B, and the Journal of Electronic Materials

Simulator development:

1. A Non-Equilibrium Greens Function (NEGF) code has been developed specifically for nanostructured thermoelectric materials. Using this code a complete strategy for the factors that affect the power factor in nanostructured materials, as well as hierarchical nanostructured materials has been identified, resulted in several publications (Physical Review B, Journal Applied Physics, Journal Electronics Materials). 7 invited talks have been given on this topic by the PI in the best conferences and workshops in the field of TEs.

2. A fully functional Monte Carlo simulator for phonon transport has been developed, together with a user interface. This was used to investigate phonon transport in hierarchically nanostructured Si-based materials, with results published in Physical Review B, and the Journal of Electronic Materials, as well as presented in 4 conferences and workshops (in two cases received the best conference award).

3. A fully functional Boltzmann Transport simulator is developed, which can extract the thermoelectric coefficients of complex bandstructure materials, using arbitrary electronic structures, even ones provided by ab initio, atomistic calculations. The simulator was used to provide design and optimization strategies for band alignment in complex band materials, with the results currently under review in Physical Review B.

4. A Monte Carlo simulator for electron transport is currently being developed, and we are currently (Dec. 2018) in the testing phase. This will be able to describe electronic transport in hierarchically nanostructured materials, to complement the phonon transport simulator developed as described above.

5. A first version of a Markov Chain Monte Carlo Machine learning algorithm, implementing simulated annealing is constructed, and is currently tested on optimizing thermal rectification in SI-based nanomaterials.

Experimental validation:

The high power factor approach that the project develops is energy filtering through potential barriers, achieved by alternating doped/non-doped regions in the material, as in a superlattice manner. We have developed a process, fabricated a device, and carried out preliminary tests of its thermoelectric properties in view of an extensive validation of our structures. The fabricated thermoelectric device includes a material with alternating highly doped/undoped regions, which creates the potential barriers in the undoped regions. The device includes heaters, temperature sensors, electrical contacts, metal tracks, and metal pads. 9 devices were fabricated on a 1cm x 1 cm SOI wafer, in which 2 devices without periodic potential energy barriers were serving as reference devices. The fabrication process of the TE device includes 5 e-beam lithographic steps, wetting etching of silicon, oxidation, doping, metal deposition and lift-off. At this point (Dec. 2018), we have preliminary evidence at least one device doubling its power factor compared to the pristine device. Further measurements and assessment of reproducibility are under way to prove our theoretical predictions. Furthermore, additional measurements and an optimization of the device layout are needed, and will be performed in the remaining half of the project to fully optimize the benefits of nanostructuring.
Progress beyond the state-of-the-art

1. The project reported the first even study on the thermoelectric power factor of hierarchically designed nanostructures, and novel strategies on significant power factor improvements. As of now, hierarchical nanostructuring is used only for thermal conductivity reductions. The results created significant interest, with the PI given 5 invited talks on top conferences and workshops on the topic, and already having 2 more invitations (European Conference on Thermoelectrics, Enrico Fermi Lecture Series) to present these results.

2. Presented a detailed study on the Lorenz number in complex bandstructure thermoelectric materials, using more involved theory and simulations compared to what the literature uses to-date. The study was published in Physical Review B, and essentially corrected (or detailed on how to correct) the extraction of the thermal conductivity in complex bandstructure materials, indicating how the Lorenz number, thought to be a constant and used as one in extracting the thermal conductivity across the literature, can vary by a lot, and this needs to be taken into account when extracting the thermal conductivity.

3. New models were created, out of large-scale Monte Carlo simulations that describe the thermal conductivity in highly disordered materials, which can be used directly by experimentalists, and essentially can replace the models that are used to-date by the thermal conductivity experimental community, but fail in the presence of nanoscale disorder. The models are published in Physical Review B.

Expected results until the end of the project:

1. The project gradually, according to the action plan, constructs simulators that compute thermoelectric properties in advanced nanomaterials. These are building blocks of a much larger simulator infrastructure that will be developed until the end of the project. Several components are in place, once all components are in place, the integration will take place as well. This will be an infrastructure clearly beyond the state-of-the-art.

2. The complete strategy for obtaining very large thermoelectric power factor improvements will be constructed. As of now the project has verified theoretically (and experimentally) parts of the complete strategy. We still need to optimize the potential barriers for filtering, the dopant distribution, and the thermal conductivity variability in the channel. There are tasks that will be performed in the immediate future, in synergy between theory and experiment.

3. The project expects that the very high power factors predicted by theory, and the strategies that will be constructed by simulations, will be experimentally verified as well in Si nanowires structures, Si thin layer structures, SiGe nanowires, and SiGe thin layers.