arXiv subjects

Xue Xia

Publications and source records attributed to Xue Xia.

21 records · Page 2Linked to original sources

arXiv2019

Visualizing electrostatic gating effects in two-dimensional heterostructures

The ability to directly observe electronic band structure in modern nanoscale field-effect devices could transform understanding of their physics and function. One could, for example, visualize local changes in the electrical and chemical potentials as a gate voltage is applied. One could also study intriguing physical phenomena such as electrically induced topological transitions and many-body spectral reconstructions. Here we show that submicron angle-resolved photoemission (micro-ARPES) applied to two-dimensional (2D) van der Waals heterostructures affords this ability. In graphene devices, we observe a shift of the chemical potential by 0.6 eV across the Dirac point as a gate voltage is applied. In several 2D semiconductors we see the conduction band edge appear as electrons accumulate, establishing its energy and momentum, and observe significant band-gap renormalization at low densities. We also show that micro-ARPES and optical spectroscopy can be applied to a single device, allowing rigorous study of the relationship between gate-controlled electronic and excitonic properties.

cond-mat.mes-hall↗

arXiv2019

Indirect to direct gap crossover in two-dimensional InSe revealed by ARPES

Atomically thin films of III-VI post-transition metal chalcogenides (InSe and GaSe) form an interesting class of two-dimensional semiconductor that feature strong variations of their band gap as a function of the number of layers in the crystal [1-4] and, specifically for InSe, an earlier predicted crossover from a direct gap in the bulk [5,6] to a weakly indirect band gap in monolayers and bilayers [7-11]. Here, we apply angle resolved photoemission spectroscopy with submicrometer spatial resolution ($μ$ARPES) to visualise the layer-dependent valence band structure of mechanically exfoliated crystals of InSe. We show that for 1 layer and 2 layer InSe the valence band maxima are away from the $\mathbfΓ$-point, forming an indirect gap, with the conduction band edge known to be at the $\mathbfΓ$-point. In contrast, for six or more layers the bandgap becomes direct, in good agreement with theoretical predictions. The high-quality monolayer and bilayer samples enables us to resolve, in the photoluminescence spectra, the band-edge exciton (A) from the exciton (B) involving holes in a pair of deeper valence bands, degenerate at $\mathbfΓ$, with the splitting that agrees with both $μ$ARPES data and the results of DFT modelling. Due to the difference in symmetry between these two valence bands, light emitted by the A-exciton should be predominantly polarised perpendicular to the plane of the two-dimensional crystal, which we have verified for few-layer InSe crystals.

cond-mat.mtrl-sci↗

arXiv2018

Deep Smoke Segmentation

Inspired by the recent success of fully convolutional networks (FCN) in semantic segmentation, we propose a deep smoke segmentation network to infer high quality segmentation masks from blurry smoke images. To overcome large variations in texture, color and shape of smoke appearance, we divide the proposed network into a coarse path and a fine path. The first path is an encoder-decoder FCN with skip structures, which extracts global context information of smoke and accordingly generates a coarse segmentation mask. To retain fine spatial details of smoke, the second path is also designed as an encoder-decoder FCN with skip structures, but it is shallower than the first path network. Finally, we propose a very small network containing only add, convolution and activation layers to fuse the results of the two paths. Thus, we can easily train the proposed network end to end for simultaneous optimization of network parameters. To avoid the difficulty in manually labelling fuzzy smoke objects, we propose a method to generate synthetic smoke images. According to results of our deep segmentation method, we can easily and accurately perform smoke detection from videos. Experiments on three synthetic smoke datasets and a realistic smoke dataset show that our method achieves much better performance than state-of-the-art segmentation algorithms based on FCNs. Test results of our method on videos are also appealing.

cs.CV↗