In this work, we present an atomic-scale investigation of L1₀-FePt particles that are partly or fully embedded in a single-crystalline MgO matrix. Hundreds of particles in different orientations, of different sizes, and with different side contact facets are statistically and quantitatively analyzed. It is found that the presence of side contact facets does not introduce any misorientation in particles with the c axis out-of-plane (OP), even in those of small sizes. In addition, a markedly higher proportion of in-plane (IP) variants is found in smaller particles of sizes less than 10 nm, and most IP variants and the IP part in multi-variants have a large area of side contact facets or are even fully embedded. Those results can be explained by the fact that the transformation strain and interface strain jointly affect the particle orientation, and the competition between total strain energy in the film plane and in lateral planes plays a key role in determining particle orientation. Thus, a suggestion on the L1₀-FePt nanoparticle film production can be proposed that a moderate area of side contact facets may help keep a perfect OP orientation in the OP particles without increasing the proportion of IP variants. Additionally, our work can provide reference information on the variant orientation preference during a post-annealing process in nanoparticle films on a rough substrate or embedded in a matrix as well as in core–shell material systems.

Electron magnetic chiral dichroism (EMCD) technique enables quantitative measurements of spin and orbital moments at a sub-nanometer resolution in principle, while it is practically challenging at high resolutions due to noise problems. Our work proposed a mathematical model that can be used to synthesize EMCD spectra with controllable noise levels and settable true values of magnetic moments. Using the model, the achievable quantification accuracy at different noise levels has been systematically assessed. The assessment has revealed the chain noise effect in routine post-processing steps and provided essential experimental instructions to improve quantification precision and accuracy at given noise levels. The ineffectiveness of several denoising methods in improving accuracy and precision has also been verified. Our work may stimulate the development of noise-dependent post-processing algorithms applicable to EMCD and advance the technique application, especially in research with a high demand for accuracy, such as mapping interfacial magnetic structures in nanomaterials.