A detailed understanding of the connections of fabrication and processing to structural and thermal properties of low-dimensional nanostructures is essential to design materials and devices for phononics, nanoscale thermal management, and thermoelectric applications. Silicon provides an ideal platform to study the relations between structure and heat transport since its thermal conductivity can be tuned over 2 orders of magnitude by nanostructuring. Combining realistic atomistic modeling and experiments, we unravel the origin of the thermal conductivity reduction in ultrathin suspended silicon membranes, down to a thickness of 4 nm. Heat transport is mostly controlled by surface scattering: rough layers of native oxide at surfaces limit the mean free path of thermal phonons below 100 nm. Removing the oxide layers by chemical processing allows us to tune the thermal conductivity over 1 order of magnitude. Our results guide materials design for future phononic applications, setting the length scale at which nanostructuring affects thermal phonons most effectively.
8 July 2009 in The Journal of Physical Chemistry C
A detailed understanding of the relation between microscopic structure and phonon propagation at the nanoscale is essential to design materials with desired phononic and thermal properties. Here we uncover a new mechanism of phonon interaction in surface oxidized membranes, i.e., native oxide layers interact with phonons in ultrathin silicon membranes through local resonances. The local resonances reduce the low frequency phonon group velocities and shorten their mean free path. This effect opens up a new strategy for ultralow thermal conductivity design as it complements the scattering mechanism which scatters higher frequency modes effectively. The combination of native oxide layer and alloying with germanium in concentration as small as 5% reduces the thermal conductivity of silicon membranes to 100 times lower than the bulk. In addition, the resonance mechanism produced by native oxide surface layers is particularly effective for thermal conductivity reduction even at very low temperatures, at which only low frequency modes are populated.
First-principles techniques for electronic transport property prediction have seen rapid progress in recent years. However, it remains a challenge to predict properties of heterostructures incorporating fabrication-dependent variability. Machine-learning (ML) approaches are increasingly being used to accelerate design and discovery of new materials with targeted properties, and extend the applicability of first-principles techniques to larger systems. However, few studies exploited ML techniques to characterize relationships between local atomic structures and global electronic transport coefficients. In this work, we propose an electronic-transport-informatics (ETI) framework that trains on ab initio models of small systems and predicts thermopower of fabricated silicon/germanium heterostructures, matching measured data. We demonstrate application of ML approaches to extract important physics that determines electronic transport in semiconductor heterostructures, and bridge the gap between ab initio accessible models and fabricated systems. We anticipate that ETI framework would have broad applicability to diverse materials classes.