DLSIA: Deep Learning for Scientific Image Analysis.

DLSIA (Deep Learning for Scientific Image Analysis) is a Python-based machine learning library that empowers scientists and researchers across diverse scientific domains with a range of customizable convolutional neural network (CNN) architectures for a wide variety of tasks in image analysis to be used in downstream data processing. DLSIA features easy-to-use architectures, such as autoencoders, tunable U-Nets and parameter-lean mixed-scale dense networks (MSDNets). Additionally, this article introduces sparse mixed-scale networks (SMSNets), generated using random graphs, sparse connections and dilated convolutions connecting different length scales. For verification, several DLSIA-instantiated networks and training scripts are employed in multiple applications, including inpainting for X-ray scattering data using U-Nets and MSDNets, segmenting 3D fibers in X-ray tomographic reconstructions of concrete using an ensemble of SMSNets, and leveraging autoencoder latent spaces for data compression and clustering. As experimental data continue to grow in scale and complexity, DLSIA provides accessible CNN construction and abstracts CNN complexities, allowing scientists to tailor their machine learning approaches, accelerate discoveries, foster interdisciplinary collaboration and advance research in scientific image analysis.

Deploying Machine Learning Based Segmentation for Scientific Imaging Analysis at Synchrotron Facilities.

Scientific user facilities present a unique set of challenges for image processing due to the large volume of data generated from experiments and simulations. Furthermore, developing and implementing algorithms for real-time processing and analysis while correcting for any artifacts or distortions in images remains a complex task, given the computational requirements of the processing algorithms. In a collaborative effort across multiple Department of Energy national laboratories, the "MLExchange" project is focused on addressing these challenges. MLExchange is a Machine Learning framework deploying interactive web interfaces to enhance and accelerate data analysis. The platform allows users to easily upload, visualize, label, and train networks. The resulting models can be deployed on real data while both results and models could be shared with the scientists. The MLExchange web-based application for image segmentation allows for training, testing, and evaluating multiple machine learning models on hand-labeled tomography data. This environment provides users with an intuitive interface for segmenting images using a variety of machine learning algorithms and deep-learning neural networks. Additionally, these tools have the potential to overcome limitations in traditional image segmentation techniques, particularly for complex and low-contrast images.

Effects of the Grazing Incidence Geometry on X-ray Photon Correlation Spectroscopy Measurements.

X-ray photon correlation spectroscopy (XPCS) is a versatile tool to measure dynamics on the nanometer to micrometer scale in bulk samples. XPCS has also been applied in grazing incidence (GI) geometry to examine the dynamics of surface layers. However, considering GI scattering experiments more universally, the GI geometry leads to a superposition of signals due to reflection and refraction effects, also known from the distorted-wave Born approximation (DWBA). In this paper, the impact of these reflection and refraction effects on the correlation analysis is determined experimentally by measuring grazing incidence transmission XPCS (GT-XPCS) and grazing incidence XPCS (GI-XPCS) simultaneously for a thin film sample, showing non-equilibrium dynamics. The results of the GI and GT geometry comparisons are combined within the framework of the standardly applied, simplified DWBA. These calculations allow identifying the main contributions of the detected signal from the leading scattering terms along the out-of-plane direction qz, which dominate the measured intensity pattern on the detector. In combination with the calculation of the non-linear effect of refraction in GTSAXS and GISAXS, it is possible to identify experimental conditions that can be chosen to run experiments and data analysis as close as possible to transmission XPCS and to explain which limitations for data interpretations are observed. Consequently, the beam exposure can be significantly reduced by using GI geometry only. Calculations of experimental settings prior to experiments are detailed to determine suitable qz regions for a variety of material systems measured in bulk-sensitive GI-XPCS experiments, allowing us to determine the scaling behavior of typical decay times as a function of q that is comparable to the scaling behavior obtained in distortion-free GT-XPCS or transmission XPCS experiments.

A comparison of deep-learning-based inpainting techniques for experimental X-ray scattering.

The implementation is proposed of image inpainting techniques for the reconstruction of gaps in experimental X-ray scattering data. The proposed methods use deep learning neural network architectures, such as convolutional autoencoders, tunable U-Nets, partial convolution neural networks and mixed-scale dense networks, to reconstruct the missing information in experimental scattering images. In particular, the recovered pixel intensities are evaluated against their corresponding ground-truth values using the mean absolute error and the correlation coefficient metrics. The results demonstrate that the proposed methods achieve better performance than traditional inpainting algorithms such as biharmonic functions. Overall, tunable U-Net and mixed-scale dense network architectures achieved the best reconstruction performance among all the tested algorithms, with correlation coefficient scores greater than 0.9980.

Sizes of pure and doped helium droplets from single shot x-ray imaging.

Advancements in x-ray free-electron lasers on producing ultrashort, ultrabright, and coherent x-ray pulses enable single-shot imaging of fragile nanostructures, such as superfluid helium droplets. This imaging technique gives unique access to the sizes and shapes of individual droplets. In the past, such droplet characteristics have only been indirectly inferred by ensemble averaging techniques. Here, we report on the size distributions of both pure and doped droplets collected from single-shot x-ray imaging and produced from the free-jet expansion of helium through a 5 µm diameter nozzle at 20 bars and nozzle temperatures ranging from 4.2 to 9 K. This work extends the measurement of large helium nanodroplets containing 109-1011 atoms, which are shown to follow an exponential size distribution. Additionally, we demonstrate that the size distributions of the doped droplets follow those of the pure droplets at the same stagnation condition but with smaller average sizes.

MLExchange: A web-based platform enabling exchangeable machine learning workflows for scientific studies.

Machine learning (ML) algorithms are showing a growing trend in helping the scientific communities across different disciplines and institutions to address large and diverse data problems. However, many available ML tools are programmatically demanding and computationally costly. The MLExchange project aims to build a collaborative platform equipped with enabling tools that allow scientists and facility users who do not have a profound ML background to use ML and computational resources in scientific discovery. At the high level, we are targeting a full user experience where managing and exchanging ML algorithms, workflows, and data are readily available through web applications. Since each component is an independent container, the whole platform or its individual service(s) can be easily deployed at servers of different scales, ranging from a personal device (laptop, smart phone, etc.) to high performance clusters (HPC) accessed (simultaneously) by many users. Thus, MLExchange renders flexible using scenarios-users could either access the services and resources from a remote server or run the whole platform or its individual service(s) within their local network.

Preferred crystallographic orientation of cellulose in plant primary cell walls.

Cellulose, the most abundant biopolymer on earth, is a versatile, energy rich material found in the cell walls of plants, bacteria, algae, and tunicates. It is well established that cellulose is crystalline, although the orientational order of cellulose crystallites normal to the plane of the cell wall has not been characterized. A preferred orientational alignment of cellulose crystals could be an important determinant of the mechanical properties of the cell wall and of cellulose-cellulose and cellulose-matrix interactions. Here, the crystalline structures of cellulose in primary cell walls of onion (Allium cepa), the model eudicot Arabidopsis (Arabidopsis thaliana), and moss (Physcomitrella patens) were examined through grazing incidence wide angle X-ray scattering (GIWAXS). We find that GIWAXS can decouple diffraction from cellulose and epicuticular wax crystals in cell walls. Pole figures constructed from a combination of GIWAXS and X-ray rocking scans reveal that cellulose crystals have a preferred crystallographic orientation with the (200) and (110)/([Formula: see text]) planes preferentially stacked parallel to the cell wall. This orientational ordering of cellulose crystals, termed texturing in materials science, represents a previously unreported measure of cellulose organization and contradicts the predominant hypothesis of twisting of microfibrils in plant primary cell walls.

Electrophoresis Assisted Printing: A Method To Control the Morphology in Organic Thin Films.

A major advantage of organic solar cells (OSC) is the processability out of solution allowing for advanced printing methods toward large-scale production. Controlling the blend morphology of solution coated active layers is a key challenge to optimize their power conversion efficiency. We have derived a printing procedure from an industrial coating process that facilitates tuning the nanomorphology of a blend of poly(3-hexylthiophene) (P3HT) and [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) as model system for OSCs. Applying an electric field during printing and the film drying process modifies the vertical film composition of the photoactive layer and optimizes the polymer crystal orientation. The choice of chloroform as solvent allows us to obtain material transport within the wet film, due to an induced electrophoretic mobility. Tailoring the morphology improves the power conversion efficiency of the OSCs by up to 25%. Our findings indicate that electrophoresis assisted printing provides an efficient approach to optimize the active layer for various material and solvent combinations that exhibit an electrophoretic mobility.

Multiresolution 3D-DenseNet for Chemical Shift Prediction in NMR Crystallography.

We have developed a deep learning algorithm for chemical shift prediction for atoms in molecular crystals that utilizes an atom-centered Gaussian density model for the 3D data representation of a molecule. We define multiple channels that describe different spatial resolutions for each atom type that utilizes cropping, pooling, and concatenation to create a multiresolution 3D-DenseNet architecture (MR-3D-DenseNet). Because the training and testing time scale linearly with the number of samples, the MR-3D-DenseNet can exploit data augmentation that takes into account the property of rotational invariance of the chemical shifts, thereby also increasing the size of the training data set by an order of magnitude without additional cost. We obtain very good agreement for 13C, 15N, and 17O chemical shifts when compared to ab initio quantum chemistry methods, with the highest accuracy found for 1H chemical shifts that is comparable to the error between the ab initio results and experimental measurements. Principal component analysis (PCA) is used to both understand these greatly improved predictions for 1H , as well as indicating that chemical shift prediction for 13C, 15N, and 17O, which have far fewer training environments than the 1H atom type, will improve once more unique training samples are made available to exploit the deep network architecture.

Xi-cam: a versatile interface for data visualization and analysis.

Xi-cam is an extensible platform for data management, analysis and visualization. Xi-cam aims to provide a flexible and extensible approach to synchrotron data treatment as a solution to rising demands for high-volume/high-throughput processing pipelines. The core of Xi-cam is an extensible plugin-based graphical user interface platform which provides users with an interactive interface to processing algorithms. Plugins are available for SAXS/WAXS/GISAXS/GIWAXS, tomography and NEXAFS data. With Xi-cam's `advanced' mode, data processing steps are designed as a graph-based workflow, which can be executed live, locally or remotely. Remote execution utilizes high-performance computing or de-localized resources, allowing for the effective reduction of high-throughput data. Xi-cam's plugin-based architecture targets cross-facility and cross-technique collaborative development, in support of multi-modal analysis. Xi-cam is open-source and cross-platform, and available for download on GitHub.

Note: Setup for chemical atmospheric control during in situ grazing incidence X-ray scattering of printed thin films.

In order to tailor the assembling of polymers and organic molecules, a deeper understanding of the kinetics involved in thin film production is necessary. While post-production characterization only provides insight on the final film structure, more sophisticated experimental setups are needed to probe the structure formation processes in situ during deposition. The drying kinetics of a deposited organic thin film strongly influences the assembling process on the nanometer scale. This work presents an experimental setup that enables fine control of the atmosphere composition surrounding the sample during slot die coating, while simultaneously probing the film formation kinetics using in situ grazing incidence X-ray scattering and spectroscopy.

In situ dynamic observations of perovskite crystallisation and microstructure evolution intermediated from [PbI6]4- cage nanoparticles.

Hybrid lead halide perovskites have emerged as high-performance photovoltaic materials with their extraordinary optoelectronic properties. In particular, the remarkable device efficiency is strongly influenced by the perovskite crystallinity and the film morphology. Here, we investigate the perovskites crystallisation kinetics and growth mechanism in real time from liquid precursor continually to the final uniform film. We utilize some advanced in situ characterisation techniques including synchrotron-based grazing incident X-ray diffraction to observe crystal structure and chemical transition of perovskites. The nano-assemble model from perovskite intermediated [PbI6]4- cage nanoparticles to bulk polycrystals is proposed to understand perovskites formation at a molecular- or nano-level. A crystallisation-depletion mechanism is developed to elucidate the periodic crystallisation and the kinetically trapped morphology at a mesoscopic level. Based on these in situ dynamics studies, the whole process of the perovskites formation and transformation from the molecular to the microstructure over relevant temperature and time scales is successfully demonstrated.

On-the-Fly Data Assessment for High-Throughput X-ray Diffraction Measurements.

Investment in brighter sources and larger and faster detectors has accelerated the speed of data acquisition at national user facilities. The accelerated data acquisition offers many opportunities for the discovery of new materials, but it also presents a daunting challenge. The rate of data acquisition far exceeds the current speed of data quality assessment, resulting in less than optimal data and data coverage, which in extreme cases forces recollection of data. Herein, we show how this challenge can be addressed through the development of an approach that makes routine data assessment automatic and instantaneous. By extracting and visualizing customized attributes in real time, data quality and coverage, as well as other scientifically relevant information contained in large data sets, is highlighted. Deployment of such an approach not only improves the quality of data but also helps optimize the usage of expensive characterization resources by prioritizing measurements of the highest scientific impact. We anticipate our approach will become a starting point for a sophisticated decision-tree that optimizes data quality and maximizes scientific content in real time through automation. With these efforts to integrate more automation in data collection and analysis, we can truly take advantage of the accelerating speed of data acquisition.

Quantifying the Hierarchical Order in Self-Aligned Carbon Nanotubes from Atomic to Micrometer Scale.

Fundamental understanding of structure-property relationships in hierarchically organized nanostructures is crucial for the development of new functionality, yet quantifying structure across multiple length scales is challenging. In this work, we used nondestructive X-ray scattering to quantitatively map the multiscale structure of hierarchically self-organized carbon nanotube (CNT) "forests" across 4 orders of magnitude in length scale, from 2.0 Å to 1.5 µm. Fully resolved structural features include the graphitic honeycomb lattice and interlayer walls (atomic), CNT diameter (nano), as well as the greater CNT ensemble (meso) and large corrugations (micro). Correlating orientational order across hierarchical levels revealed a cascading decrease as we probed finer structural feature sizes with enhanced sensitivity to small-scale disorder. Furthermore, we established qualitative relationships for single-, few-, and multiwall CNT forest characteristics, showing that multiscale orientational order is directly correlated with number density spanning 109-1012 cm-2, yet order is inversely proportional to CNT diameter, number of walls, and atomic defects. Lastly, we captured and quantified ultralow-q meridional scattering features and built a phenomenological model of the large-scale CNT forest morphology, which predicted and confirmed that these features arise due to microscale corrugations along the vertical forest direction. Providing detailed structural information at multiple length scales is important for design and synthesis of CNT materials as well as other hierarchically organized nanostructures.

Printing Fabrication of Bulk Heterojunction Solar Cells and In Situ Morphology Characterization.

Polymer-based materials hold promise as low-cost, flexible efficient photovoltaic devices. Most laboratory efforts to achieve high performance devices have used devices prepared by spin coating, a process that is not amenable to large-scale fabrication. This mismatch in device fabrication makes it difficult to translate quantitative results obtained in the laboratory to the commercial level, making optimization difficult. Using a mini-slot die coater, this mismatch can be resolved by translating the commercial process to the laboratory and characterizing the structure formation in the active layer of the device in real time and in situ as films are coated onto a substrate. The evolution of the morphology was characterized under different conditions, allowing us to propose a mechanism by which the structures form and grow. This mini-slot die coater offers a simple, convenient, material efficient route by which the morphology in the active layer can be optimized under industrially relevant conditions. The goal of this protocol is to show experimental details of how a solar cell device is fabricated using a mini-slot die coater and technical details of running in situ structure characterization using the mini-slot die coater.

Fluorination of Donor-Acceptor Copolymer Active Layers Enhances Charge Mobilities in Thin-Film Transistors.

Several recent reports have demonstrated that fluorinated analogues of donor/acceptor copolymers surpass nonfluorinated counterparts in terms of performance in electronic devices. Using a copolymer series consisting of fluorinated, partially fluorinated, and nonfluorinated benzotriazole, we confirm that the addition of fluorine substituents beneficially impacts charge transport in polymer semiconductors. Transistor measurements demonstrated a factor of 5 increase in carrier mobilities with the degree of fluorination of the backbone. Furthermore, grazing-incidence X-ray diffraction data indicates progressively closer packing between the conjugated cores and an overall greater amount of π-stacking in the fluorinated materials. It is likely that attractive interactions between the electron-rich donor and fluorinated electron-deficient acceptor units induce very tightly stacking crystallites, which reduce the energetic barrier for charge hopping. In addition, a change in crystallite orientation was observed from primarily edge-on without fluorine substituents to mostly face-on with fluorinated benzotriazole.

Resonant Carbon K-Edge Soft X-Ray Scattering from Lattice-Free Heliconical Molecular Ordering: Soft Dilative Elasticity of the Twist-Bend Liquid Crystal Phase.

Resonant x-ray scattering shows that the bulk structure of the twist-bend liquid crystal phase, recently discovered in bent molecular dimers, has spatial periodicity without electron density modulation, indicating a lattice-free heliconical nematic precession of orientation that has helical glide symmetry. In situ study of the bulk helix texture of the dimer CB7CB shows an elastically confined temperature-dependent minimum helix pitch, but a remarkable elastic softness of pitch in response to dilative stresses. Scattering from the helix is not detectable in the higher temperature nematic phase.

Probing and controlling liquid crystal helical nanofilaments.

We report the first in situ measurement of the helical pitch of the helical nanofilament B4 phase of bent-core liquid crystals using linearly polarized, resonant soft X-ray scattering at the carbon K-edge. A strong, anisotropic scattering peak corresponding to the half-pitch of the twisted smectic layer structure was observed. The equilibrium helical half-pitch of NOBOW is found to be 120 nm, essentially independent of temperature. However, the helical pitch can be tuned by mixing guest organic molecules with the bent-core host, followed by thermal annealing.

Advanced grazing-incidence techniques for modern soft-matter materials analysis.

The complex nano-morphology of modern soft-matter materials is successfully probed with advanced grazing-incidence techniques. Based on grazing-incidence small- and wide-angle X-ray and neutron scattering (GISAXS, GIWAXS, GISANS and GIWANS), new possibilities arise which are discussed with selected examples. Due to instrumental progress, highly interesting possibilities for local structure analysis in this material class arise from the use of micro- and nanometer-sized X-ray beams in micro- or nanofocused GISAXS and GIWAXS experiments. The feasibility of very short data acquisition times down to milliseconds creates exciting possibilities for in situ and in operando GISAXS and GIWAXS studies. Tuning the energy of GISAXS and GIWAXS in the soft X-ray regime and in time-of flight GISANS allows the tailoring of contrast conditions and thereby the probing of more complex morphologies. In addition, recent progress in software packages, useful for data analysis for advanced grazing-incidence techniques, is discussed.