A novel cellulose-derived hierarchical g-C3N4/TiO2-nanotube heterostructured nanocomposite was fabricated by in situ coating thin g-C3N4 layers onto the surfaces of the TiO2 nanotubes, which were synthesized by utilizing the natural cellulose substance (e.g., commercial ordinary filter paper) as the structural template. These g-C3N4/TiO2-nanotube composites with varied thicknesses (ca. 3–30 nm) of the outer g-C3N4 layers displayed improved visible-light (λ > 420 nm)-driven photocatalytic degradation performances toward methylene blue. The optimal nanocomposite with an outer g-C3N4 layer of ca. 7.5 nm composed of 46 wt % g-C3N4 displayed an apparent rate constant of 0.0035 min–1, which was 8.5- and 4-fold larger than those of the referential TiO2-nanotube and g-C3N4 powder. The excellent and durable photocatalytic activities of these cellulose-derived g-C3N4/TiO2-nanotube composites were ascribed to their hierarchically network porous structures replicated from the cellulose template, as well as the formation of close heterojunctions in-between the g-C3N4 and TiO2 phases. Moreover, it was demonstrated that the photocatalytic mechanism matched with the type-II heterostructured model, while the main effective species during the photocatalytic processes of the nanocomposite were proved to be superoxide radicals.

No abstract

Nanofibrous CoS–nanoparticle/carbon composite derived from a cellulose substance was fabricated, showing enhanced electrochemical performances as an anodic material for lithium-ion batteries.

No abstract

How to incorporate external knowledge into a neural dialogue model is critically important for dialogue systems to behave like real humans. To handle this problem, memory networks are usually a great choice and a promising way. However, existing memory networks do not perform well when leveraging heterogeneous information from different sources. In this paper, we propose a novel and versatile external memory networks called Heterogeneous Memory Networks (HMNs), to simultaneously utilize user utterances, dialogue history and background knowledge tuples. In our method, historical sequential dialogues are encoded and stored into the context-aware memory enhanced by gating mechanism while grounding knowledge tuples are encoded and stored into the context-free memory. During decoding, the decoder augmented with HMNs recurrently selects each word in one response utterance from these two memories and a general vocabulary. Experimental results on multiple real-world datasets show that HMNs significantly outperform the state-of-the-art datadriven task-oriented dialogue models in most domains.