Data-driven simulations for training AI-based segmentation of neutron images.

Neutron interferometry uniquely combines neutron imaging and scattering methods to enable characterization of multiple length scales from 1 nm to 10 µm. However, building, operating, and using such neutron imaging instruments poses constraints on the acquisition time and on the number of measured images per sample. Experiment time-constraints yield small quantities of measured images that are insufficient for automating image analyses using supervised artificial intelligence (AI) models. One approach alleviates this problem by supplementing annotated measured images with synthetic images. To this end, we create a data-driven simulation framework that supplements training data beyond typical data-driven augmentations by leveraging statistical intensity models, such as the Johnson family of probability density functions (PDFs). We follow the simulation framework steps for an image segmentation task including Estimate PDFs → Validate PDFs → Design Image Masks → Generate Intensities → Train AI Model for Segmentation. Our goal is to minimize the manual labor needed to execute the steps and maximize our confidence in simulations and segmentation accuracy. We report results for a set of nine known materials (calibration phantoms) that were imaged using a neutron interferometer acquiring four-dimensional images and segmented by AI models trained with synthetic and measured images and their masks.

Relevance of hydrogen bonded associates to the transport properties and nanoscale dynamics of liquid and supercooled 2-propanol.

2-Propanol was investigated, in both the liquid and supercooled states, as a model system to study how hydrogen bonds affect the structural relaxation and the dynamics of mesoscale structures, of approximately several Ångstroms, employing static and quasi-elastic neutron scattering and molecular dynamics simulation. Dynamic neutron scattering measurements were performed over an exchanged wave-vector range encompassing the pre-peak, indicative of the presence of H-bonding associates, and the main peak. The dynamics observed at the pre-peak is associated with the formation and disaggregation of the H-bonded associates and is measured to be at least one order of magnitude slower than the dynamics at the main peak, which is identified as the structural relaxation. The measurements indicate that the macroscopic shear viscosity has a similar temperature dependence as the dynamics of the H-bonded associates, which highlights the important role played by these structures, together with the structural relaxation, in defining the macroscopic rheological properties of the system. Importantly, the characteristic relaxation time at the pre-peak follows an Arrhenius temperature dependence whereas at the main peak it exhibits a non-Arrhenius behavior on approaching the supercooled state. The origin of this differing behavior is attributed to an increased structuring of the hydrophobic domains of 2-propanol accommodating a more and more encompassing H-bond network, and a consequent set in of dynamic cooperativity.

Salting Up and Salting Down of Bovine Serum Albumin Layers at the Air-Water Interface.

The surface adsorption of bovine serum albumin in pure water and salted aqueous solutions was studied by neutron reflection. With the contrast match technique, the surface excess in null reflecting water as a function of the protein concentration was revealed. It is found that, in a concentration range from 1 ppm (parts per million, mg/L) to 1000 ppm, without salts, the surface excess shows a profound peak at around 20 ppm; with salts, the surface excess increases steadily with the protein concentration. When the surface excess at a specific protein concentration is viewed, the introduction of sodium chloride causes either a salting down effect (surface adsorption decline) or a salting up effect (surface adsorption increase), depending upon the protein concentration. The salting up effect is observed at the low (∼1 ppm) and high (∼1000 ppm) concentrations, and the salting down effect dominates the intermediate concentration range. The change in solution pH relative to the isoelectric point (PI) can act as a simple indicator for the salting up or salting down behavior. When the solution pH is shifted toward the PI by adding salts, surface adsorption enhances; when the solution pH is shifted away from the PI by adding salts, surface adsorption declines.

Information gain from isotopic contrast variation in neutron reflectometry on protein-membrane complex structures.

A framework is applied to quantify information gain from neutron or X-ray reflectometry experiments [Treece, Kienzle, Hoogerheide, Majkrzak, Lösche & Heinrich (2019). J. Appl. Cryst. 52, 47-59], in an in-depth investigation into the design of scattering contrast in biological and soft-matter surface architectures. To focus the experimental design on regions of interest, the marginalization of the information gain with respect to a subset of model parameters describing the structure is implemented. Surface architectures of increasing complexity from a simple model system to a protein-lipid membrane complex are simulated. The information gain from virtual surface scattering experiments is quantified as a function of the scattering length density of molecular components of the architecture and the surrounding aqueous bulk solvent. It is concluded that the information gain is mostly determined by the local scattering contrast of a feature of interest with its immediate molecular environment, and experimental design should primarily focus on this region. The overall signal-to-noise ratio of the measured reflectivity modulates the information gain globally and is a second factor to be taken into consideration.

Optimization of reflectometry experiments using information theory.

A framework based on Bayesian statistics and information theory is developed to optimize the design of surface-sensitive reflectometry experiments. The method applies to model-based reflectivity data analysis, uses simulated reflectivity data and is capable of optimizing experiments that probe a sample under more than one condition. After presentation of the underlying theory and its implementation, the framework is applied to exemplary test problems for which the information gain ΔH is determined. Reflectivity data are simulated for the current generation of neutron reflectometers at the NIST Center for Neutron Research. However, the simulation can be easily modified for X-ray or neutron instruments at any source. With application to structural biology in mind, this work explores the dependence of ΔH on the scattering length density of aqueous solutions in which the sample structure is bathed, on the counting time and on the maximum momentum transfer of the measurement. Finally, the impact of a buried magnetic reference layer on ΔH is investigated.

Measurement and modeling of polarized specular neutron reflectivity in large magnetic fields.

The presence of a large applied magnetic field removes the degeneracy of the vacuum energy states for spin-up and spin-down neutrons. For polarized neutron reflectometry, this must be included in the reference potential energy of the Schrödinger equation that is used to calculate the expected scattering from a magnetic layered structure. For samples with magnetization that is purely parallel or antiparallel to the applied field which defines the quantization axis, there is no mixing of the spin states (no spin-flip scattering) and so this additional potential is constant throughout the scattering region. When there is non-collinear magnetization in the sample, however, there will be significant scattering from one spin state into the other, and the reference potentials will differ between the incoming and outgoing wavefunctions, changing the angle and intensities of the scattering. The theory of the scattering and recommended experimental practices for this type of measurement are presented, as well as an example measurement.

Phase segregation of sulfonate groups in Nafion interface lamellae, quantified via neutron reflectometry fitting techniques for multi-layered structures.

Neutron reflectometry analysis methods for under-determined, multi-layered structures are developed and used to determine the composition depth profile in cases where the structure is not known a priori. These methods, including statistical methods, sophisticated fitting routines, and coupling multiple data sets, are applied to hydrated and dehydrated Nafion nano-scaled films with thicknesses comparable to those found coating electrode particles in fuel cell catalyst layers. These results confirm the lamellar structure previously observed on hydrophilic substrates, and demonstrate that for hydrated films they can accurately be described as layers rich in both water and sulfonate groups alternating with water-poor layers containing an excess of fluorocarbon groups. The thickness of these layers increases slightly and the amplitude of the water volume fraction oscillation exponentially decreases away from the hydrophilic interface. For dehydrated films, the composition oscillations die out more rapidly. The Nafion-SiO2 substrate interface contains a partial monolayer of sulfonate groups bonded to the substrate and a large excess of water compared to that expected by the water-to-sulfonate ratio, λ, observed throughout the rest of the film. Films that were made thin enough to truncate this lamellar region showed a depth profile nearly identical to thicker films, indicating that there are no confinement or surface effects altering the structure. Comparing the SLD profile measured for films dried at 60 °C to modeled composition profiles derived by removing water from the hydrated lamellae suggests incomplete re-mixing of the polymer groups upon dehydration, indicated limited polymer mobility in these Nafion thin films.