How auto-differentiation can improve CT workflows: classical algorithms in a modern framework.

Many of the recent successes of deep learning-based approaches have been enabled by a framework of flexible, composable computational blocks with their parameters adjusted through an automatic differentiation mechanism to implement various data processing tasks. In this work, we explore how the same philosophy can be applied to existing "classical" (i.e., non-learning) algorithms, focusing on computed tomography (CT) as application field. We apply four key design principles of this approach for CT workflow design: end-to-end optimization, explicit quality criteria, declarative algorithm construction by building the forward model, and use of existing classical algorithms as computational blocks. Through four case studies, we demonstrate that auto-differentiation is remarkably effective beyond the boundaries of neural-network training, extending to CT workflows containing varied combinations of classical and machine learning algorithms.

Interfaceless Exchange Bias in CoFe2O4 Nanocrystals.

Oxidized cobalt ferrite nanocrystals with a modified distribution of the magnetic cations in their spinel structure give place to an unusual exchange-coupled system with a double reversal of the magnetization, exchange bias, and increased coercivity, but without the presence of a clear physical interface that delimits two well-differentiated magnetic phases. More specifically, the partial oxidation of cobalt cations and the formation of Fe vacancies at the surface region entail the formation of a cobalt-rich mixed ferrite spinel, which is strongly pinned by the ferrimagnetic background from the cobalt ferrite lattice. This particular configuration of exchange-biased magnetic behavior, involving two different magnetic phases but without the occurrence of a crystallographically coherent interface, revolutionizes the established concept of exchange bias phenomenology.

Analysis of 3D elemental distribution in nanomaterials: Towards higher throughput and dose efficiency.

Many advanced nanomaterials rely on carefully designed morphology and elemental distribution to achieve their functionalities. Among the few experimental techniques that can directly visualise the 3D elemental distribution on the nanoscale are approaches based on electron tomography in combination with energy-dispersive X-ray spectroscopy (EDXS) and electron energy loss spectroscopy (EELS). Unfortunately, these highly informative methods are severely limited by the fundamentally low signal-to-noise ratio, which makes long experimental times and high electron irradiation doses necessary to obtain reliable 3D reconstructions. Addressing these limitations has been the major research question for the development of these techniques in recent years. This short review outlines the latest progress on the methods to reduce experimental time and electron irradiation dose requirements for 3D elemental distribution analysis and gives an outlook on the development of this field in the near future.

The Influence of Size, Shape, and Twin Boundaries on Heat-Induced Alloying in Individual Au@Ag Core-Shell Nanoparticles.

Environmental conditions during real-world application of bimetallic core-shell nanoparticles (NPs) often include the use of elevated temperatures, which are known to cause elemental redistribution, in turn significantly altering the properties of these nanomaterials. Therefore, a thorough understanding of such processes is of great importance. The recently developed combination of fast electron tomography with in situ heating holders is a powerful approach to investigate heat-induced processes at the single NP level, with high spatial resolution in 3D. In combination with 3D finite-difference diffusion simulations, this method can be used to disclose the influence of various NP parameters on the diffusion dynamics in Au@Ag core-shell systems. A detailed study of the influence of heating on atomic diffusion and alloying for Au@Ag NPs with varying core morphology and crystallographic details is carried out. Whereas the core shape and aspect ratio of the NPs play a minor role, twin boundaries are found to have a strong influence on the elemental diffusion.

3D Atomic-Scale Dynamics of Laser-Light-Induced Restructuring of Nanoparticles Unraveled by Electron Tomography.

Understanding light-matter interactions in nanomaterials is crucial for optoelectronic, photonic, and plasmonic applications. Specifically, metal nanoparticles (NPs) strongly interact with light and can undergo shape transformations, fragmentation and ablation upon (pulsed) laser excitation. Despite being vital for technological applications, experimental insight into the underlying atomistic processes is still lacking due to the complexity of such measurements. Herein, atomic resolution electron tomography is performed on the same mesoporous-silica-coated gold nanorod, before and after femtosecond laser irradiation, to assess the missing information. Combined with molecular dynamics (MD) simulations based on the experimentally determined 3D atomic-scale morphology, the complex atomistic rearrangements, causing shape deformations and defect generation, are unraveled. These rearrangements are simultaneously driven by surface diffusion, facet restructuring, and strain formation, and are influenced by subtleties in the atomic distribution at the surface.

Deep learning-based denoising for improved dose efficiency in EDX tomography of nanoparticles.

The combination of energy-dispersive X-ray spectroscopy (EDX) and electron tomography is a powerful approach to retrieve the 3D elemental distribution in nanomaterials, providing an unprecedented level of information for complex, multi-component systems, such as semiconductor devices, as well as catalytic and plasmonic nanoparticles. Unfortunately, the applicability of EDX tomography is severely limited because of extremely long acquisition times and high electron irradiation doses required to obtain 3D EDX reconstructions with an adequate signal-to-noise ratio. One possibility to address this limitation is intelligent denoising of experimental data using prior expectations about the objects of interest. Herein, this approach is followed using the deep learning methodology, which currently demonstrates state-of-the-art performance for an increasing number of data processing problems. Design choices for the denoising approach and training data are discussed with a focus on nanoparticle-like objects and extremely noisy signals typical for EDX experiments. Quantitative analysis of the proposed method demonstrates its significantly enhanced performance in comparison to classical denoising approaches. This allows for improving the tradeoff between the reconstruction quality, acquisition time and radiation dose for EDX tomography. The proposed method is therefore especially beneficial for the 3D EDX investigation of electron beam-sensitive materials and studies of nanoparticle transformations.

Spherical core-shell alumina support particles for model platinum catalysts.

γ- and δ-alumina are popular catalyst support materials. Using a hydrothermal synthesis method starting from aluminum nitrate and urea in diluted solution, spherical core-shell particles with a uniform particle size of about 1 µm were synthesized. Upon calcination at 1000 °C, the particles adopted a core-shell structure with a γ-alumina core and δ-alumina shell as evidenced by 2D and 3D electron microscopy and 27Al magic angle spinning nuclear magnetic resonance spectroscopy. The spherical alumina particles were loaded with Pt nanoparticles with an average size below 1 nm using the strong electrostatic adsorption method. Electron microscopy and energy dispersive X-ray spectroscopy revealed a homogeneous platinum dispersion over the alumina surface. These platinum loaded alumina spheres were used as a model catalyst for bifunctional catalysis. Physical mixtures of Pt/alumina spheres and spherical zeolite particles are equivalent to catalysts with platinum deposited on the zeolite itself facilitating the investigation of the catalyst components individually. The spherical alumina particles are very convenient supports for obtaining a homogeneous distribution of highly dispersed platinum nanoparticles. Obtaining such a small Pt particle size is challenging on other support materials such as zeolites. The here reported and well-characterized Pt/alumina spheres can be combined with any zeolite and used as a bifunctional model catalyst. This is an interesting strategy for the examination of the acid catalytic function without the interference of the supported platinum metal on the investigated acid material.

Fast versus conventional HAADF-STEM tomography of nanoparticles: advantages and challenges.

HAADF-STEM tomography is a widely used experimental technique for analyzing nanometer-scale structures of a large variety of materials in three dimensions. It is especially useful for studying crystalline nanoparticles, where conventional TEM tomography suffers from diffraction-related artefacts. Unfortunately, the acquisition of a HAADF-STEM tilt series can easily take up one hour or more, depending on the complexity of the experiment. It is therefore challenging to investigate samples that do not withstand long electron beam illumination or to acquire a large number of tilt series during a single TEM experiment. The latter would facilitate obtaining more statistically representative 3D data, and enable performing dynamic in situ 3D characterizations with a finer time resolution. Various HAADF-STEM acquisition strategies have been proposed to accelerate the tomographic acquisition and reduce the required electron dose. These methods include tilting the holder continuously while acquiring a projection "movie" and a hybrid, incremental, methodology which combines the benefits of the conventional and continuous technique. However, until now an experimental evaluation of these techniques has been lacking. In this paper, the different acquisition strategies will be experimentally compared in terms of speed, resolution and electron dose. This evaluation will be performed based on experimental tilt series, acquired for various metallic nanoparticles with different shapes and sizes. We discuss the necessary data processing and provide a general guideline that can be used to determine the most optimal acquisition strategy for specific electron tomography experiments.

An Expanded Surface-Enhanced Raman Scattering Tags Library by Combinatorial Encapsulation of Reporter Molecules in Metal Nanoshells.

Raman-encoded gold nanoparticles (NPs) have been widely employed as photostable multifunctional probes for sensing, bioimaging, multiplex diagnostics, and surface-enhanced Raman scattering (SERS)-guided tumor therapy. We report a strategy toward obtaining a particularly large library of Au nanocapsules encoded with Raman codes defined by the combination of different thiol-free Raman reporters, encapsulated at defined molar ratios. The fabrication of SERS tags with tailored size and predefined codes is based on the in situ incorporation of Raman reporter molecules inside Au nanocapsules during their formation via galvanic replacement coupled to seeded growth on Ag NPs. The hole-free closed-shell structure of the nanocapsules is confirmed by electron tomography. The unusually wide encoding possibilities of the obtained SERS tags are investigated by means of either wavenumber-based encoding or Raman frequency combined with signal intensity, leading to an outstanding performance as exemplified by 26 and 54 different codes, respectively. We additionally demonstrate that encoded nanocapsules can be readily bioconjugated with antibodies for applications such as SERS-based targeted cell imaging and phenotyping.

Unravelling the nucleation, growth, and faceting of magnetite-gold nanohybrids.

The chemical synthesis of nanoparticles with a preassigned size and shape is important for an optimized performance in any application. Therefore, systematic monitoring of the synthesis is required for the control and detailed understanding of the nucleation and growth of the nanoparticles. Here, we study Fe3O4-Au hybrid nanoparticles in detail using probes of the reaction mixture during synthesis and their thorough characterization. The proposed approach eliminates the problem of repeatability and reproducibility of the chemical synthesis and was carried out using laboratory equipment (standard transmission electron microscopy, X-ray diffraction, and magnetometry) for typically 10 µL samples instead of, for example, a dedicated synthesis and inspection at a synchrotron radiation facility. From the three independent experimental techniques we extract the nanoparticle size at 12 stages of the synthesis. These diameters show identical trends and good quantitative agreement. Two consecutive processes occur during the synthesis of Fe3O4-Au nanoparticles, the nucleation and the growth of spherical Fe3O4 nanoparticles on the surface of Au seeds during the heating stage and their faceting towards octahedral shape during reflux. The final nanoparticles with sizes of 15 nm Fe3O4 and 4 nm Au exhibit superparamagnetic behavior at ambient temperature. These are high-quality, close to stoichiometric Fe3O4 nanocrystals with nearly volumetric magnetic behavior as confirmed by the presence of the Verwey transition. Understanding the processes occurring during the synthesis allows the nanoparticle size and shape to be adjusted, improving their capabilities in biomedical applications.

Quantitative 3D Characterization of Elemental Diffusion Dynamics in Individual Ag@Au Nanoparticles with Different Shapes.

Anisotropic bimetallic nanoparticles are promising candidates for plasmonic and catalytic applications. Their catalytic performance and plasmonic properties are closely linked to the distribution of the two metals, which can change during applications in which the particles are exposed to heat. Due to this fact, correlating the thermal stability of complex heterogeneous nanoparticles to their microstructural properties is of high interest for the practical applications of such materials. Here, we employ quantitative electron tomography in high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mode to measure the 3D elemental diffusion dynamics in individual anisotropic Au-Ag nanoparticles upon heating in situ. This approach allows us to study the elemental redistribution in complex, asymmetric nanoparticles on a single particle level, which has been inaccessible to other techniques so far. In this work, we apply the proposed method to compare the alloying dynamics of Au-Ag nanoparticles with different shapes and compositions and find that the shape of the nanoparticle does not exhibit a significant effect on the alloying speed whereas the composition does. Finally, comparing the experimental results to diffusion simulations allows us to estimate the diffusion coefficients of the metals for individual nanoparticles.

Toxicity of iron oxide nanoparticles: Size and coating effects.

Toxicological research of novel nanomaterials is a major developmental step of their clinical approval. Since iron oxide magnetic nanoparticles have a great potential in cancer treatment and diagnostics, the investigation of their toxic properties is very topical. In this paper we synthesized bovine serum albumin-coated iron oxide nanoparticles with different sizes and their polyethylene glycol derivative. To prove high biocompatibility of obtained nanoparticles the number of in vitro toxicological tests on human fibroblasts and U251 glioblastoma cells was performed. It was shown that albumin nanoparticles' coating provides a stable and biocompatible shell and prevents cytotoxicity of magnetite core. On long exposure times (48 hours), cytotoxicity of iron oxide nanoparticles takes place due to free radical production, but this toxic effect may be neutralized by using polyethylene glycol modification.

Multimodal doxorubicin loaded magnetic nanoparticles for VEGF targeted theranostics of breast cancer.

In presented paper we have developed new system for cancer theranostics based on vascular endothelial growth factor (VEGF) targeted magnetic nanoparticles. Conjugation of anti-VEGF antibodies with bovine serum albumin coated PEGylated magnetic nanoparticles allows for improved binding with murine breast adenocarcinoma 4T1 cell line and facilitates doxorubicin delivery to tumor cells. It was shown that intravenous injection of doxorubicin loaded VEGF targeted nanoparticles increases median survival rate of mice bearing 4T1 tumors up to 50%. On the other hand magnetic resonance imaging (MRI) of 4T1 tumors 24 h after intravenous injection showed accumulation of nanoparticles in tumors, thus allowing simultaneous cancer therapy and diagnostics.