Hard carbon is the most promising anode material for sodium‐ion batteries and potassium‐ion batteries owing to its high stability, widespread availability, low‐cost, and excellent performance. Understanding the carrier‐ion storage mechanism is a prerequisite for developing high‐performance electrode materials; however, the underlying ion storage mechanism in hard carbon has been a topic of debate because of its complex structure. Herein, it is demonstrated that the Li+‐, Na+‐, and K+‐ion storage mechanisms in hard carbon are based on the adsorption of ions on the surface of active sites (e.g., defects, edges, and residual heteroatoms) in the sloping voltage region, followed by intercalation into the graphitic layers in the low‐voltage plateau region. At a low current density of 3 mA g–1, the graphitic layers of hard carbon are unlocked to permit Li+‐ion intercalation, resulting in a plateau region in the lithium‐ion batteries. To gain insights into the ion storage mechanism, experimental observations including various ex situ techniques, a constant‐current constant‐voltage method, and diffusivity measurements are correlated with the theoretical estimation of changes in carbon structures and insertion voltages during ion insertion obtained using the density functional theory.

No abstract

Recent recommender systems have started to employ knowledge distillation, which is a model compression technique distilling knowledge from a cumbersome model (teacher) to a compact model (student), to reduce inference latency while maintaining performance. The state-of-the-art methods have only focused on making the student model to accurately imitate the predictions of the teacher model. They have a limitation in that the prediction results incompletely reveal the teacher's knowledge. In this paper, we propose a novel knowledge distillation framework for recommender system, called DE-RRD, which enables the student model to learn from the latent knowledge encoded in the teacher model as well as from the teacher's predictions. Concretely, DE-RRD consists of two methods: 1) Distillation Experts (DE) that directly transfers the latent knowledge from the teacher model. DE exploits "experts" and a novel expert selection strategy for effectively distilling the vast teacher's knowledge to the student with limited capacity. 2) Relaxed Ranking Distillation (RRD) that transfers the knowledge revealed from the teacher's prediction with consideration of the relaxed ranking orders among items. Our extensive experiments show that DE-RRD outperforms the state-of-the-art competitors and achieves comparable or even better performance to that of the teacher model with faster inference time. CCS CONCEPTS • Information systems → Learning to rank; Collaborative filtering; Retrieval efficiency.

Cancer cells, compared to normal cells, are under oxidative stress associated with an elevated level of reactive oxygen species (ROS) and are more vulnerable to oxidative stress induced by ROS generating agents. Thus, manipulation of the ROS level provides a logical approach to kill cancer cells preferentially, without significant toxicity to normal cells, and great efforts have been dedicated to the development of strategies to induce cytotoxic oxidative stress for cancer treatment. Fenton reaction is an important biological reaction in which irons convert hydrogen peroxide (H2O2) to highly toxic hydroxyl radicals that escalate ROS stress. Here, we report Fenton reaction-performing polymer (PolyCAFe) micelles as a new class of ROS-manipulating anticancer therapeutic agents. Amphiphilic PolyCAFe incorporates H2O2-generating benzoyloxycinnamaldehyde and iron-containing compounds in its backbone and self-assembles to form micelles that serve as Nano-Fenton reactors to generate cytotoxic hydroxyl radicals, killing cancer cells preferentially. When intravenously injected, PolyCAFe micelles could accumulate in tumors preferentially to remarkably suppress tumor growth, without toxicity to normal tissues. This study demonstrates the tremendous translatable potential of Nano-Fenton reactors as a new class of anticancer drugs.

We report an effective way to fabricate mechanically strong and multifunctional polyimide (PI) nanocomposites using aminophenyl functionalized graphene nanosheet (APGNS). APGNS was successfully obtained through a diazonium salt reaction. PI composites with different loading of APGNS were prepared by in situ polymerization. Both the mechanical and electrical properties of the APGNS/ PI composites were significantly improved compared with those of pure PI due to the homogeneous dispersion of APGNS and the strong interfacial covalent bonds between APGNS and the PI matrix. The electrical conductivity of APGNS/PI (3:97 w/w) was 6.6 × 10 −2 S/m which was about 10 11 times higher than that of pure PI. Furthermore, the modulus of APGNS/PI was increased up to 16.5 GPa, which is approximately a 610% enhancement compared to that of pure PI, and tensile strength was increased from 75 to 138 MPa. The water vapor transmission rate of APGNS/PI composites (3:97 w/w) was reduced by about 74% compared to that of pure PI. ■ INTRODUCTIONAromatic polyimide (PI) is a high-performance polymer with applications in the fields of microelectronics, optoelectronics, adhesives, and aerospace owing to its high thermal stability and favorable chemical and mechanical properties. 1,2 However, PI has a few limitations, such as electrostatic accumulation, poor heat dissipation, and low electrical conductivity for special applications. In recent years, much attention has been paid to PI composites with carbon nanomaterials because the incorporation of carbon nanofillers can effectively enhance the thermal, mechanical, and electrical properties of the nanocomposites. 2−6 Graphene, a one-atom-thick planar sheet of carbon atoms densely packed in a honeycomb crystal lattice, 7 has revolutionized the scientific frontiers of nanoscience and condensed matter physics due to its exceptional electrical, 8 physical, 9 and chemical properties. 10 The excellent properties of graphene have opened new pathways for developing a wide range of novel functional materials. In addition, graphene has a distinctive mechanical property with fracture strains of ∼25% and a Young's modulus of ∼1 TPa. 9 However, poor dispersion in organic solvents and weak interfacial interactions between graphene and the polymer matrix limit the widespread use of graphene. In contrast, graphene oxide (GO) produced by the oxidation of graphite can solve these issues. It is easily dispersed in water and polar solvents due to the functional groups, such as ketones, diols, epoxides, hydroxyls, and carbonyl, on its edges and basal planes. 11 Nevertheless, GO has limited compatibility with certain polymers and limited solubility in hydrophobic solvent owing to its hydrophilic nature, which can reduce the reinforcement effects of interfacial interaction in the polymer matrix. 12,13 Therefore, the key issue is to improve both the homogeneous dispersion and strong interfacial interaction between the polymer matrix and graphene for the development of high performance polymer/graphene nanocomposites. R...