Elucidating proximity magnetism through polarized neutron reflectometry and machine learning

Abstract

Polarized neutron reflectometry is a powerful technique to interrogate the structures of multilayered magnetic materials with depth sensitivity and nanometer resolution. However, reflectometry profiles often inhabit a complicated objective function landscape using traditional fitting methods, posing a significant challenge for parameter retrieval. In this work, we develop a data-driven framework to recover the sample parameters from polarized neutron reflectometry data with minimal user intervention. We train a variational autoencoder to map reflectometry profiles with moderate experimental noise to an interpretable, low-dimensional space from which sample parameters can be extracted with high resolution. We apply our method to recover the scattering length density profiles of the topological insulator–ferromagnetic insulator heterostructure Bi2Se3/EuS exhibiting proximity magnetism in good agreement with the results of conventional fitting. We further analyze a more challenging reflectometry profile of the topological insulator–antiferromagnet heterostructure (Bi,Sb)2Te3/Cr2O3 and identify possible interfacial proximity magnetism in this material. We anticipate that the framework developed here can be applied to resolve hidden interfacial phenomena in a broad range of layered systems.

Progress and prospects in the quantum anomalous Hall effect

“…Extraordinary care is warranted in the TI growth, and the magnetic interfacial heterostructuring thereafter, to properly mitigate the adverse influence from residual chalcogen atoms since optimal TI demands a Se/Te-rich growth condition. This intriguing discovery has opened a vibrant arena for proximitized MI/TI heterostructures [110][111][112][113][114][115][116][117][118][119][120][121].…”

Neutron Reflectometry Studies of the Hydrated Structure of Polymer Thin Films

“…The lack of phase information also entails the abundance of local minima in the parameter space hindering the fitting process and increasing the possibility of overfitting 120 . To facilitate the search for the global minimum, fitting programs implement stochastic optimization methods 121,122 such as differential evolution, simulated annealing, or stochastic tunnelling, and novel machine learning methods are being investigated for evaluating reflectograms 123,124 . The recorded reflectivity spans several orders of magnitude in value.…”

Neural network analysis of neutron and X-ray reflectivity data: automated analysis using mlreflect, experimental errors and feature engineering

“…However, all of these approaches are iterative in nature and thus usually computationally slow. Recently, machine-learning-based methods have been proposed that could avoid a lengthy search of the MSE surface by providing an immediate guess for the thin-film parameters that is already very close to the ground truth (Greco et al, 2019;Mironov et al, 2021;Doucet et al, 2021;Carmona Loaiza & Raza, 2021;Greco et al, 2021) or by encoding the reflectometry data into a latent space where the error surface does not have as many local minima (Andrejevic et al, 2021).…”