The direct use of an advanced binder‐free additive manufacturing technique, namely laser powder bed fusion (L‐PBF), does not easily allow obtaining variously shaped, fully dense Nd–Fe–B magnets with high coercivity. The process inherently leads to the re‐melting of the powder and appearance/disappearance of undesired/desired microstructural features responsible for low and large coercivity. In this work, the development of a useful microstructure responsible for high coercivity in Pr21Fe73.5Cu2B3.5 and Nd21Fe73.5Cu2B3.5 alloys and a possible way to produce fully dense permanent magnets via additive manufacturing processes is demonstrated using: (i) suction casting technique, which provides a high cooling rate and thus similar microstructures as in L‐PBF but requires only very small amounts of powder; (ii) conventional L‐PBF processing using kg of powder, and (iii) a subsequent annealing treatment that is similar to a conventional sintering treatment. The subsequent heat treatment is necessary to develop high coercivity by forming a novel microstructure: hard magnetic (Nd,Pr)2Fe14B grains embedded in a matrix of intermetallic (Nd,Pr)6Fe13Cu phase. Furthermore, it is demonstrated that Pr21Fe73.5Cu2B3.5 exhibits a higher coercivity than Nd21Fe73.5Cu2B3.5 because of a finer and more homogeneous grain size distribution of the Pr2Fe14B phase. The final L‐PBF printed Pr21Fe73.5Cu2B3.5 samples provide a coercivity of 0.75 T.

We combined experimental investigations and theoretical calculations to unveil an abnormal magnetic behavior caused by addition of the nonmagnetic element Cu in face-centered-cubic FeNiCoMn-based high-entropy alloys (HEAs). Upon Cu addition, the probed HEAs show an increase of both Curie temperature and saturation magnetization in as-cast and homogenized states. Specifically, the saturation magnetization of the as-cast HEAs at room temperature increases by 77% and 177% at a Cu content of 11 and 20 at. %, respectively, compared to the as-cast equiatomic FeNiCoMn HEA without Cu. The increase in saturation magnetization of the as-cast HEAs is associated with the formation of an Fe-Co rich phase in the dendritic regions. For the homogenized HEAs, the magnetic state at room temperature transforms from paramagnetism to ferromagnetism after 20 at. % Cu addition. The increase of the saturation magnetization and Curie temperature cannot be adequately explained by the formation of Cu enriched zones according to atom probe tomography analysis. Ab initio calculations suggest Cu plays a pivotal role in the stabilization of a ferromagnetic ordering of Fe, and reveal an increase of the Curie temperature caused by Cu addition which agrees well with the experimental results. The underlying mechanism behind this phenomenon lies in a combined change in unit-cell volume and chemical composition and the related energetic stabilization of the magnetic ordering upon Cu alloying as revealed by theoretical calculations. Thus, the work unveils the mechanisms responsible for the Cu effect on the magnetic properties of FeNiCoMn HEAs, and suggests that nonmagnetic elements are also crucial to tune and improve magnetic properties of HEAs.

No abstract

Multi-agent deep reinforcement learning (MARL) suffers from a lack of commonlyused evaluation tasks and criteria, making comparisons between approaches difficult. In this work, we evaluate and compare three different classes of MARL algorithms (independent learners, centralised training with decentralised execution, and value decomposition) in a diverse range of multi-agent learning tasks. Our results show that (1) algorithm performance depends strongly on environment properties and no algorithm learns efficiently across all learning tasks; (2) independent learners often achieve equal or better performance than more complex algorithms;(3) tested algorithms struggle to solve multi-agent tasks with sparse rewards. We report detailed empirical data, including a reliability analysis, and provide insights into the limitations of the tested algorithms. * Authors contributed equally 1 We provide open-source implementations of two newly developed environments here: www.github.com/uoe-agents/lb-foraging, www.github.com/uoe-agents/robotic-warehouse

No abstract