Tian Tian

Miniaturization of light-emitting diodes below the diffraction limit of the emission wavelength can enable super-resolution imaging and on-chip light sources for ultrabroadband chiplet communication. Organic light-emitting diodes, although suitable for miniaturization due to their emission from localized excitons, suffer from the limited compatibility of organic materials with traditional photolithographic patterning. Here we develop a method for the scalable fabrication of nanoscale organic light-emitting diodes with pixel densities up to 100,000 pixels per inch, periodicity of 250 nm and the smallest pixel size in the order of 100 nm. We realize the direct nanoscale patterning of organic semiconductors by self-aligned nanostencil etching and lithography. The process is resist-free and involves etching and evaporation through nanoapertures in a free-standing film adhering to the substrate. A nanoscale organic light-emitting diode surface with over 1 megapixel exhibits an average external quantum efficiency of 13.1%. We also demonstrate electroluminescent metasurfaces with subwavelength-scale meta-atoms that can electrically modulate the emitted light. The diffractive coupling between nanopixels enables control over the far-field emission properties of light, including directionality and polarization. These results pave the way for hybrid integrated photonics technologies, including visible-light communication, lasing and high-resolution displays.

Understanding surface forces of two-dimensional (2D) materials is of fundamental importance as they govern molecular dynamics in nanoscale proximity. Despite recent understanding of substrate-supported 2D monolayers, the intrinsic surface properties of 2D materials remain vague. Here we report on a repulsive Lifshitz-van der Waals force generated in proximity to the surface of suspended graphene. In combination with our theoretical model taking into account the flexibility of graphene, we directly measured repulsive forces using atomic force microscopy. An average repulsive force of up to 1.4 kN/m² has been detected at separations of 8.8 nm between a gold-coated tip and a sheet of suspended graphene, more than two orders of magnitude greater than the long-range Casimir-Lifshitz repulsion demonstrated in fluids. Our findings imply that suspended 2D materials could exert repulsive forces on any approaching electroneutral object in close proximity, resulting in substantially lower wettability. This could offer technological opportunities such as molecular actuation and quantum levitation.

Density Functional Theory (DFT) is the de facto workhorse for large-scale electronic structure calculations in chemistry and materials science. While plane-wave DFT implementations remain the most widely used, real-space DFT provides advantages in handling complex boundary conditions and scaling to very large systems by allowing for the efficient use of large-scale supercomputers and linear-scaling methods that circumvent the cubic scaling bottleneck. The SPARC-X project (https://github.com/SPARC-X) provides highly efficient and portable real-space DFT codes for a wide range of first principle applications, available in both Matlab (M-SPARC (Xu et al., 2020; Zhang et al., 2023)) and C/C++ (SPARC (Xu et al., 2021; Zhang et al., 2024)). The rapid growth of SPARC's feature set has created the need for a fully functional interface to drive SPARC in high-throughput calculations. Here we introduce SPARC-X-API, a Python package designed to bridge the SPARC-X project with broader computational frameworks. Built on the Atomic Simulation Environment (ASE (Hjorth Larsen et al., 2017)) standard, the SPARC-X-API allows users to handle SPARC file formats and run SPARC calculations through the same interface as with other ASE-compatible DFT packages. Beyond standard ASE capabilities, SPARC-X-API provides additional features including 1) support of SPARC-specific setups, including complex boundary conditions and unit conversion, 2) a JSON schema parsed from SPARC's documentation for parameter validation and compatibility checks, and 3) a comprehensive socket communication layer derived from the i-PI protocol (Ceriotti et al., 2014; Kapil et al., 2019) facilitating message passing between low-level C code and the Python interface. The goal of the SPARC-X-API is to provide an easy-to-use interface for users with diverse needs and levels of expertise, allowing for minimal effort in adapting SPARC to existing computational workflows, while also supporting developers of advanced real-space methods.

The calculation of relative energy difference has significant practical applications, such as determining adsorption energy, screening for optimal catalysts with volcano plots, and calculating reaction energies. Although Density Functional Theory (DFT) is effective in calculating relative energies through systematic error cancellation, the accuracy of Graph Neural Networks (GNNs) in this regard remains uncertain. To address this, we analyzed ∼483 × 106 pairs of energy differences predicted by DFT and GNNs using the Open Catalyst 2020-Dense dataset. Our analysis revealed that GNNs exhibit a correlated error that can be reduced through subtraction, challenging the assumption of independent errors in GNN predictions and leading to more precise energy difference predictions. To assess the magnitude of error cancellation in chemically similar pairs, we introduced a new metric, the subgroup error cancellation ratio. Our findings suggest that state-of-the-art GNN models can achieve error reduction of up to 77% in these subgroups, which is comparable to the error cancellation observed with DFT. This significant error cancellation allows GNNs to achieve higher accuracy than individual energy predictions and distinguish subtle energy differences. We propose the marginal correct sign ratio as a metric to evaluate this performance. Additionally, our results show that the similarity in local embeddings is related to the magnitude of error cancellation, indicating the need for a proper training method that can augment the embedding similarity for chemically similar adsorbate–catalyst systems.

Modern machine learning (ML) models are becoming increasingly popular and are widely used in decision-making systems. However, studies have shown critical issues of ML discrimination and unfairness, which hinder their adoption on high-stake applications. Recent research on fair classifiers has drawn significant attention to developing effective algorithms to achieve fairness and good classification performance. Despite the great success of these fairness-aware machine learning models, most of the existing models require sensitive attributes to pre-process the data, regularize the model learning or post-process the prediction to have fair predictions. However, sensitive attributes are often incomplete or even unavailable due to privacy, legal or regulation restrictions. Though we lack the sensitive attribute for training a fair model in the target domain, there might exist a similar domain that has sensitive attributes. Thus, it is important to exploit auxiliary information from a similar domain to help improve fair classification in the target domain. Therefore, in this paper, we study a novel problem of exploring domain adaptation for fair classification. We propose a new framework that can learn to adapt the sensitive attributes from a source domain for fair classification in the target domain. Extensive experiments on real-world datasets illustrate the effectiveness of the proposed model for fair classification, even when no sensitive attributes are available in the target domain.

Aggregation-induced through-space charge transfer can be used to tune single-fluorophore emission in simple polymer systems.

One of the central aims of synthetic biology and metabolic engineering is to mimic the integrality of eukaryotic cells to construct a multifunctional compartment system to perform multistep incompatible cascade reactions in a one-pot, controlled, and selective fashion. The key challenge is how to address the coexistence of antagonistic reagents and to incorporate these functionalities into an integrated system in a smart and efficient way. A novel strategy called “iterative etching–grafting” is proposed here based on monodispersed photonic spheres (PSs) prepared by microfluidics, which constructs a universal platform for incompatible cascade reactions. As a proof of concept, we spatiotemporally regulated the degree of etching of PSs, then grafted precursory groups of acid and base onto PSs, and incorporated a photocleavage method, which were capable of compartmentalizing the acid and base inside PSs. Utilizing the band-gap offsets of PSs could track the progress of cascade reactions in situ, and grafting various charged polymers on the surface of the pores by surface-initiated atom transfer radical polymerization (SI-ATRP) achieved the selectivity of the substrates, which flexibly constructed a multifunctional and integrated acid–base photonic multicompartment system (PMCS). The created PMCS shows excellent catalytic performance, convenient monitoring, and efficient substrate selectivity in the deacetalization–Knoevenagel cascade reaction. Furthermore, two types of electrophile/nucleophile PMCSs have also been accessibly constructed, demonstrating the facile generation of other incompatible systems with the versatility as well as the advancement and extensibility of the developed strategy.

Reduced graphene oxide (rGO) is considered as one of the ideal sensing materials for high-performance room-temperature gas sensors owing to large specific surface areas, numerous active sites and high carrier mobility. However, the sensing performance cannot be maximized due to the inevitable sheet stacking and agglomeration. Herein, we firstly demonstrate multichannel room-temperature gas sensors using magnetic field induced alignment of three-dimensional (3D) [email protected]@rGO core-shell spheres. Moreover, the sensing channels composed of spheres can be tailored by changing the concentration of spheres and magnetic field. Experimental results suggest that the multichannel 3D [email protected]@rGO sensor exhibits an ultrahigh sensitivity of 34.41 with a good response stability and high selectivity toward 5 ppm NO2 at room temperature, which is ca. 7.96 times higher than that of the random 3D rGO gas sensor. The high performance can be mainly ascribed to a full utilization of their large specific surface area and active sites of rGO nanosheets. We believe that our results not only contribute to the development of high-performance rGO-based sensing devices, but also provide a general approach to maximize the sensing performance of other nanomaterials.