Quantifying behavior is crucial for many applications in neuroscience. Videography provides easy methods for the observation and recording of animal behavior in diverse settings, yet extracting particular aspects of a behavior for further analysis can be highly time consuming. In motor control studies, humans or other animals are often marked with reflective markers to assist with computer-based tracking, but markers are intrusive, and the number and location of the markers must be determined a priori. Here we present an efficient method for markerless pose estimation based on transfer learning with deep neural networks that achieves excellent results with minimal training data. We demonstrate the versatility of this framework by tracking various body parts in multiple species across a broad collection of behaviors. Remarkably, even when only a small number of frames are labeled (~200), the algorithm achieves excellent tracking performance on test frames that is comparable to human accuracy.
Despite heavy research, blue perovskite nanocrystal LEDs have struggled to match the high efficiencies of their red and green cousins, particularly at wavelengths blue enough to meet the NTSC blue standard. One of the most critical problems is the low photoluminescence yield of the nanocrystals in thin films. Recently, manganese doping has been shown to increase the photoluminescence yield of the perovskite even as a decay pathway to a long-lived emissive state on the manganese ion is introduced. Here, we employ a two-step synthetic approach to carefully tune the manganese doping to increase the blue photoluminescence while preventing significant manganese emission, allowing for blue perovskite LEDs with quantum efficiencies over 2% that meet the NTSC standard. Manganese doping increases the photoluminescence yield and lifetime, reduces trap states, and makes the dots more monodisperse, reducing the emission bandwidth. Finally, we utilize perovskite nanocrystal downconverters to build an all-perovskite white LED.
Light‐emitting diodes utilizing perovskite nanocrystals have generated strong interest in the past several years, with green and red devices showing high efficiencies. Blue devices, however, have lagged significantly behind. Here, it is shown that the device architecture plays a key role in this lag and that NiOx, a transport layer in one of the highest efficiency devices to date, causes a significant reduction in perovskite luminescence lifetime. An alternate transport layer structure which maintains robust nanocrystal emission is proposed. Devices with this architecture show external quantum efficiencies of 0.50% at 469 nm, seven times higher than state‐of‐the‐art devices at that wavelength. Finally, it is demonstrated that this architecture enables efficient devices across the entire blue‐green portion of the spectrum. The improvements demonstrated here open the door to efficient blue perovskite light‐emitting diodes.
Single-atom catalysts have emerged as an exciting paradigm with intriguing properties different from their nanocrystal counterparts. Here we report Ni single atoms dispersed into graphene nanosheets, without Ni nanoparticles involved, as active sites for the electrocatalytic CO2 reduction reaction (CO2RR) to CO. While Ni metal catalyzes the hydrogen evolution reaction (HER) exclusively under CO2RR conditions, Ni single atomic sites present a high CO selectivity of 95% under an overpotential of 550 mV in water, and an excellent stability over 20 hours’ continuous electrolysis. The current density can be scaled up to more than 50 mA cm−2 with a CO evolution turnover frequency of 2.1 × 105 h−1 while maintaining 97% CO selectivity using an anion membrane electrode assembly. Different Ni sites in graphene vacancies, with or without neighboring N coordination, were identified by in situ X-ray absorption spectroscopy and density functional theory calculations. Theoretical analysis of Ni and Co sites suggests completely different reaction pathways towards the CO2RR or HER, in agreement with experimental observations.
Electrocatalytic CO2 reduction to higher-value hydrocarbons beyond C1products is desirable for applications in energy storage, transportation and the chemical industry. Cu catalysts have shown the potential to catalyse C–C coupling for C2+ products, but still suffer from low selectivity in water. Here, we use density functional theory to determine the energetics of the initial C–C coupling steps on different Cu facets in CO2 reduction, and suggest that the Cu(100) and stepped (211) facets favour C2+ product formation over Cu(111). To demonstrate this, we report the tuning of facet exposure on Cu foil through the metal ion battery cycling method. Compared with the polished Cu foil, our 100-cycled Cu nanocube catalyst with exposed (100) facets presents a sixfold improvement in C2+ to C1 product ratio, with a highest C2+Faradaic efficiency of over 60% and H2 below 20%, and a corresponding C2+ current of more than 40 mA cm–2.