Katayun Barmak

Photoexcitation has been utilized to control quantum matter and to uncover metastable phases far from equilibrium. Among demonstrations to date, the most common is the photoinduced transition from correlated insulators to metallic states; however, the reverse process without initial orders has not been observed. Here, we show ultrafast metal-to-insulator transition in gate-doped WS_{2}/WSe_{2} and WSe_{2}/WSe_{2} moiré devices using photothermionic hole injection from graphite gates. The resulting correlated insulators are metastable, with lifetimes exceeding microseconds. These findings establish an effective mechanism for the ultrafast control of correlated electronic phases in van der Waals (vdW) heterostructures.

Recent observations of superconductivity in twisted bilayer WSe₂ (refs. ^1,2) have extended the family of moiré superconductors beyond twisted graphene^{3,4,5,6,7,8,9,10,11,12,13,14,15}. In WSe₂, two different twist angles were studied, 3.65° (ref. ¹) and 5.0° (ref. ²), and two seemingly distinct superconducting phase diagrams were reported, raising the question of whether the superconducting phases in the two devices share a similar origin. Here we address the question by experimentally mapping the evolution of the phase diagram across devices with twist angles spanning the range defined by the initial reports and comparing the results to twist angle-dependent theory. We find that the superconducting state evolves smoothly with twist angle and at all twist angles is proximal to a Fermi surface reconstruction with, presumably, antiferromagnetic ordering, but is neither necessarily tied to the Van Hove singularity nor to the half-band insulator. Our results connect the previously distinct phase diagrams at 3.65° and 5°, and offer new insight into the origin of the superconductivity in this system and its evolution as the correlation strength increases. More broadly, the smooth phase diagram evolution, repeatability between different devices and dynamic gate tunability within each device establish twisted transition metal dichalcogenides as a unique platform for the study of correlated phases as the ratio of interaction strength to bandwidth is varied.

Solid-state quantum emitters (QEs) are central to quantum photonic technologies, but existing plasmonic and dielectric cavities rarely combine nanoscale spatial control, high photon flux, and polarization stability. We present a triangular gap-plasmon cavity integrated with monolayer WSe₂ that achieves all three by coupling apex-concentrated strain fields with localized optical confinement, thereby realizing deterministic quantum light funneling to a sub-100 nm region. Finite-element simulations identify an optimal 66 nm geometry with a 20° apex angle, maximizing Purcell enhancement near 750 nm. Gold nanotriangles reproducibly activate three strain-induced QEs; the apex emitter exhibits g² (0) = 0.054 (0.141 when coupled), a lifetime reduction from 16.8 to 0.248 ns (63-fold on average), and saturation counts up to 126 MHz into the first lens. Base emitters show a lower ∼23-fold enhancement, yielding a tip-to-base ratio of 2.7. Polarization studies across 41 QEs confirm dipole alignment within ±5°. When further combining this triangular platform with resonant excitation and tuning schemes, a scalable route toward indistinguishable single-photon sources on chip could be achieved.

Metallic thin films offer a platform to experimentally study the dynamics of microstructural evolution, but the required transmission electron microscopy (TEM)-based imaging generates complex images that are challenging to segment and quantify. This work provides a comparative analysis of a new YOLOv8 model and an established U-Net model for bright-field TEM images of polycrystals, employing a framework leveraging physical observables to evaluate performance against two hand-traced benchmark datasets. This methodology obviates the comparison of large, diversely structured, and manually labeled datasets that are required to assess performance on a per-image/per-pixel basis. It is found that the YOLOv8 model, adapted for real-time instance segmentation, has up to 43× faster inferencing (NVIDIA GeForce RTX 4090) compared to U-Net and reconstructs hand-traced grain size distributions (GSDs) with excellent fidelity, finding mean diameter within 3% for grains near an optimal magnification; for grains that deviate from the optimal pixel-diameter, the size of small- (large)-diameter grains is systematically over- (under)-estimated. This is partially mitigated by including scale-aware augmentations during training. Moreover, when the bias is corrected post-inference by a rigid shift in distribution, the YOLOv8 model reproduces ground truth GSDs with exceptional fidelity, with statistical tests indicating <5% probability that the distributions are distinct. Based on ground truth data, calibration curves pertaining to this shift can be constructed for a given model. This issue is not present in the U-Net model’s results, indicating that for quantitative measurements where the true size of objects is of interest, special procedures must be implemented for YOLO-based models.

Machine learning strategies for the semantic segmentation of materials’ micrographs, such as U-Net, have been employed in recent years to enable the automated identification of grain-boundary networks in polycrystals. For example, most recently, this architecture has allowed researchers to address the long-standing problem of automated image segmentation of thin-film microstructures in bright-field TEM micrographs. Such approaches are typically based on the minimization of a binary cross-entropy loss function that compares constructed images to a ground truth at the pixel level over many epochs. In this work, we quantify the rate at which the underlying microstructural features embodied in the grain-boundary network, as described stereologically, are also learned in this process. In particular, we assess the rate of microstructural learning in terms of the moments of the k-th nearest-neighbor pixel distributions and associated metrics, including a microstructural cross-entropy, that embody the spatial correlations among the pixels through a hierarchy of n-point correlation functions. From the moments of these distributions, we obtain so-called learning functions that highlight the rate at which the important topological features of a grain-boundary network appear. It is found that the salient features of network structure emerge after relatively few epochs, suggesting that grain size, network topology, etc., are learned early (as measured in epochs) during the segmentation process.

Materials capable of light upconversion—transforming low-energy photons into higher-energy ones—are pivotal in advancing optoelectronics, energy solutions, and photocatalysis. However, the discovery in various materials pays little attention on few-layer transition metal dichalcogenides, primarily due to their indirect bandgaps and weaker light-matter interactions. Here, we report a pronounced light upconversion in few-layer transition metal dichalcogenides through upconversion photoluminescence spectroscopy. Our joint theory-experiment study attributes the upconversion photoluminescence to a resonant exciton-exciton annihilation involving a pair of dark excitons with opposite momenta, followed by the spontaneous emission of upconverted bright excitons, which can have a high upconversion efficiency. Additionally, the upconversion photoluminescence is generic in MoS₂, MoSe₂, WS₂, and WSe₂, showing a high tuneability from green to ultraviolet light (2.34–3.1 eV). The findings pave the way for further exploration of light upconversion regarding fundamental properties and device applications in two-dimensional semiconductors.