To explore the potential of atomic clusters in device applications, we illustrate the thermal rectification (TR) effect and underlying mechanisms in the cluster-assembled Sierpiński tetrahedron (ST) fractals. Using the synthesized supertetrahedral $T_2$-type ${\mathrm{Sn}}_4{\mathrm{Se}}_{10}$ clusters as building blocks and nonequilibrium molecular dynamics with a machine-learning neuroevolution potential trained on a multidimensional ab initio dataset, we show that the TR coefficient of the assembled ${\mathrm{ST}}_4$ fractal reaches 17.4% at an average temperature of 300 K with a temperature bias of 20 K. This further increases to 39.4% under the same temperature condition for the higher-level fractal ${\mathrm{ST}}_5$ as the void ratio and surface roughness increase significantly with the fractal level. Moreover, unlike in conventional atom-based nanosystems, the dependence of the TR coefficient on temperature bias in the ST fractals is nonmonotonic, where a high TR efficiency can be achieved just with a small temperature bias. Our work demonstrates the merits of cluster-assembled fractals for high-performance thermal rectifiers and the advantages of cluster-assembled materials for thermal energy management.

Close

PrevNext

View figure in article

• Energy applications
• Nanoclusters
• Nanostructures
• Deep learning
• Density functional theory
• Materials modeling
• Molecular dynamics