arXiv ScienceSearch

arXiv subjects

Steven R Spurgeon

Publications and source records attributed to Steven R Spurgeon.

2 recordsLinked to original sources

Scandium diboride: a semi-metallic, lattice, thermally matched substrate for vertical AlGaN power electronics

We report the properties of hexagonal (space group P6/mmm) scandium diboride ($\mathrm{ScB}_2$) single crystals grown by a laser diode floating zone method at growth rates of ~1mm/hr under B-rich conditions with (002) rocking curve widths $Δω$=38'' approaching the quality of commercial SiC/GaN substrates. Lattice expansion measurements reveal matching to $\mathrm{Al_{0.55}Ga_{0.45}N}$ with a coefficient of thermal expansion ~5ppm/K at typical AlGaN growth temperatures, enabling thick AlGaN layers for ultra-wide bandgap (UWBG) power electronics >1kV. We measure semi-metallic room temperature resistivity ~15$μΩ$ cm, climbing to ~93$μΩ$ cm at 773K with a $T^2$ dependence effectively eliminating substrate parasitic resistance, the limiting factor in exploiting the full potential of UWBG. The Debye temperature $θ_{D,ScB_2}$ from heat capacity and lattice expansion is ~850K well matched to $θ_{D,ScB_2}$, but lower than the 1100K measured for Sc-rich growth conditions. We discuss Debye matching as a key substrate codesign criterion providing significant overlap in phonon modes for heat removal and thermal matching during AlGaN growth. The competitive thermal conductivity at room temperature 53W/mK is half that from full first principles calculations, a discrepancy we attribute to the presence of Sc-vacancies generated by B-rich growth. while the resistivity is ~2x the theoretical value, indicating that both electrons and phonons play equal role in thermal transport. The smooth ~2.5nm rms roughness surface enables advanced heat removal modalities through engineered phonon bridges and phonon polaritons in $\mathrm{ScB}_2$/AlGaN interfacial heterostructures, potentially allowing ~10-100x increase in power handling over state-of-the-art GaN/SiC.

cond-mat.mtrl-sci

Unsupervised segmentation of irradiation$\unicode{x2010}$induced order$\unicode{x2010}$disorder phase transitions in electron microscopy

We present a method for the unsupervised segmentation of electron microscopy images, which are powerful descriptors of materials and chemical systems. Images are oversegmented into overlapping chips, and similarity graphs are generated from embeddings extracted from a domain$\unicode{x2010}$pretrained convolutional neural network (CNN). The Louvain method for community detection is then applied to perform segmentation. The graph representation provides an intuitive way of presenting the relationship between chips and communities. We demonstrate our method to track irradiation$\unicode{x2010}$induced amorphous fronts in thin films used for catalysis and electronics. This method has potential for "on$\unicode{x2010}$the$\unicode{x2010}$fly" segmentation to guide emerging automated electron microscopes.

cond-mat.mtrl-sci