Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

The comminution of spent lithium-ion batteries (LIBs) produces a powder containing the active cell components, commonly referred to as “black mass.” Recently, froth flotation has been pro-posed to treat the fine fraction of black mass (< 100µm) as a method to separate anodic graphite particles from cathodic lithium metal oxides (LMOs). So far, pyrolysis has been considered as an effective treatment to remove organic binders in the black mass in preparation for flotation sepa-ration. In this work, the flotation performance of a pyrolyzed black mass obtained from an indus-trial recycling plant was improved by adding a pre-treatment step consisting of mechanical attri-tion. The LMOs recovery in the underflow product increased from 70% to 85% and the graphite recovery remained similar, around 86% recovery in the overflow product. To understand the flo-tation behaviour, the spent black mass was compared to a model black mass, comprising fully lib-erated LMOs and graphite particles. In addition, ultrafine hydrophilic particles were added to the flotation feed as an entrainment tracer, showing that the LMO recovery in overflow products is a combination of entrainment and true flotation mechanisms. This study highlights that adding kerosene during attrition enhance the emulsification of kerosene, simultaneously increasing its (partial) spreading on LMOs, graphite and residual binder, with a subsequent reduction of selectivity.

We demonstrate a straightforward and easily applicable technique for growing the highly improved quality of artificially BaZrO3 doped YBa2Cu3O6+x films on a commercially used buffered metallic template by pulsed laser deposition. Our method relies on reducing the grain size of the target material, which completely prevents the transfer of the harmful grain boundaries or weak links from the substrate through the buffer layers on the deposited film. We have also observed a great improvement in the self-assembly of BaZrO3 dopants and the critical current density is increased in the high temperature range up to 40%. As an extra benefit, our method allows increasing the growth rate of the film by 25%. We have discussed the results comprehensively with the help of the Ginzburg-Landau theory and provided an universal quantitative model of the grain boundary transfer from the substrate to the deposited film. The presented technique can be considered as a groundbreaking advancement for the vastly growing coated conductor industry.

In this work, the radiation responses of Zr/Nb nanostructured metallic multilayers (NMMs) are studied. The nanostructures with different layer thicknesses were deposited on Si (1 0 0)
substrate by using magnetron sputtering and are subjected to heavy-ion irradiation at room temperature with different fluences. Nanoindentation, XRD, SIMS, and Variable Energy
Positron Annihilation Spectroscopy (VEPAS) techniques were used to study the changes in the hardness of the structures as well as the type and distribution of defects, and strain within the material as a function of damage. Our results suggest that the irradiation hardening is independent of the type of implanted ions, and its magnitude decreases with decreasing
individual layer thickness indicating that the number of interfaces has a direct effect on the radiation tolerance enhancement. For thin layers with a periodicity of 27 nm (Zr/Nb27), a
transition from hardening to softening occurs at high fluence, and a saturation point is reached in thick layers with a periodicity of 96 nm (Zr/Nb96). The as-deposited multilayer with the
smaller periodicity experienced a significantly higher atomic-scale disorder which increases with ion irradiation compared to the multilayer with thicker individual layers. VEPAS reveals
the vacancy defects before and after irradiation that contribute to the presented strain. Based on the findings, thin nanostructured Zr/Nb multilayered structures possess excellent radiation
resistance due to the high density of interfaces that act as sinks for radiation-induced point defects.

The talk will give an overview on the numerical simulation of liquid metal batteries, polymer electrolyte fuel cells and electrolysers. After a brief introduction to liquid metal batteries, different exemplary simulations will be presented - with a focus on the interaction of fluid flow and electrochemistry. The presentation will be limited to the continuum mechanical open-source finite-volume library OpenFOAM.

The reaction between 2,3-dimercaptopropane-1-sulfonate (DMPS, unithiol) and four phenylarsonic(V) acids, i.e. phenylarsonic acid (PAA), 4-hydroxy-3-nitrophenylarsonic acid (HNPAA), 2-aminophenylarsonic acid (o-APAA) and 4-aminophenylarsonic acid (p-APAA), is investigated in aqueous solution. The pentavalent arsenic compounds are reduced by DMPS to their trivalent analogs and instantly chelated by the vicinal dithiol, forming covalent As–S bonds within a five-membered chelate ring. The different types and positions of polar substituents at the aromatic ring of the arsonic acids influence the reaction rates in the same way as observed for reaction with glutathione (GSH), as well as the syn : anti molar ratio of the diastereomeric products, which was analyzed using time- and temperature-dependent nuclear magnetic resonance (NMR) spectroscopy. Addition of DMPS to the conjugate formed by a phenylarsonic(V) acid and the biologically relevant tripeptide GSH showed the immediate replacement of GSH by chelating DMPS, underlining the importance of dithiols as detoxifying agent.

It is the intention of this study to elucidate the nested formation of calcium carbonate polymorphs or polyamorphs in the different nanosized compartments. With these observations it can be concluded how the bacteria can survive in a harsh environment with high calcium carbonate supersaturation. The mechanisms of calcium carbonate precipitation at the surface membrane and at the underlying cell wall membrane of the thermophilic soil bacterium Geobacillus stearothermophilus DSM 13240 have been revealed by high-resolution transmission electron microscopy and atomic force microscopy. In this Gram-positive bacterium nanopores in the surface layer (S-layer) and in the supporting cell wall polymers are nucleation sites for metastable calcium carbonate polymorphs and polyamorphs. In order to observe the different metastable forms, various reaction times and a low reaction temperature (4 °C) have been chosen. Calcium carbonate polymorphs nucleate in the confinement of nanosized pores (Ø3-5 nm) of the S-layer. The hydrous crystalline calcium carbonate (ikaite) is formed initially with [110] as the favored growth direction. It transforms into the anhydrous metastable vaterite by a solid-state transition. In a following reaction step calcite is precipitated caused by dissolution of vaterite in the aqueous solution. In the larger pores of the cell wall (Ø 20-50 nm) hydrated amorphous calcium carbonate is grown, which transforms into metastable monohydrocalcite, aragonite or calcite. Due to sequence of reaction steps via various metastable phases the bacteria gain time for chipping the partially mineralized S-layer, and forming a fresh S-layer (characteristic growth time about 20 min). Thus, the bacteria can survive in solutions with high calcium carbonate supersaturation under the conditions of forced biomineralization.

We present an improved method for in-vivo proton range verification by prompt gamma-ray timing based on multivariate statistical modelling.

To this end, prompt gamma-ray timing distributions acquired during pencil beam irradiation of an acrylic glass phantom with air cavities of different thicknesses were analysed. Relevant histogram features were chosen using forward variable selection and the Least Absolute Shrinkage and Selection Operator (LASSO) from a feature assortment based on recommendations of the Image Biomarker Standardisation Initiative. Candidate models were defined by multivariate linear regression and evaluated based on their coefficient of determination \(R^2\) and root mean square error \(RMSE\).

The newly developed models showed a clearly improved predictive power (\(R^2 > 0.7\)) compared to the previously used models (\(R^2 < 0.5\)) and allowed for the identification of introduced air cavities in a scanned treatment field. %The parameter selection showed better predictive power of the energy-specific models (RM SE < 1,8 mm) compared to the energy-independent models (RM SE > 3 mm).
%for counting statistics equivalent to a single spot measured with eight detector units.

These results demonstrate that elaborate statistical models can enhance prompt gamma ray based treatment verification and increase its potential for routine clinical application.

Y-stabilized zirconia (YSZ) phases were found to incorporate Th atoms and the corresponding solubility ranges depend on the relative Y content. For the tetragonal phases with lower Y concentration of 14 at.% a maximal possible Th intake on the Zr/Y metal site reached ca. 10.3 at.%. Cubic phases with higher Y content could dissolve 11 at.% Th in equilibrium state and up to ca. 12.3 at.% Th under non-equilibrium conditions. Larger Th-Zr/Y solubility range for phases with higher Y concentration was found to be related to the associated symmetry increase, as concluded from synchrotron radiation powder diffraction data. Specifically, introduction of Th into tetragonal YSZ induces evolution towards higher cubic symmetry via flattening of the Zr/YO8 polyhedra. In addition, tetragonal YSZ crystal lattice exhibits strongly anisotropic expansion with concomitant decrease in tetragonality upon the intake of Th. This results in easier accommodation of bigger Th atoms via structural stabilization of longer Zr/Y-O bonding distances which yields more symmetrical coordination of central Zr/Y metal ions by surrounding O atoms. Cubic symmetry is, therefore, more favorable to incorporation of large tetravalent Actinide elements.

Difficulties in dating Mississippi Valley-type (MVT) mineralization and the often closely-associated regional dolomitization events have led to considerable controversy regarding the typical tectonic contexts in which they form. This paper presents the first radiometric datings for the Gorno district in Lombardy, Italy, a classic example of the Alpine subclass of MVT deposits. U-Pb dating of hydrothermal carbonate minerals pre- and post-dating the main ore-forming event shows that both regional dolomitization and base-metal mineralization must have occurred within <10 Ma of the deposition of the Carnian host rocks. This implies that the ore deposits of the Gorno district, and probably other Alpine-type deposits, formed at shallow burial, prior to the western Tethys rifting phase. Concurrent Triassic magmatism and extensional tectonics likely contributed to the high geothermal heat fluxes necessary to drive the associated large-scale hydrothermal system. These findings contradict the general paradigm for MVT genesis, i.e., a preferred occurrence in deep-burial settings associated with compressional tectonics.

This paper provides an assessment of fluid transport and mixing processes inside the primary circuit of the test facility ROCOM through the numerical simulation of Test 2.1 with the system code ATHLET. The experiment represents an asymmetric injection of cold and non-borated water into the reactor coolant system (RCS) of a pressurized water reactor (PWR) to restore core cooling, an emergency procedure which may subsequently trigger a core re-criticality. The injection takes place at low velocity under single-phase subcooled conditions and presents a major challenge for the simulation in lumped parameter codes, due to multidimensional effects in horizontal piping and vessel arising from density gradients and gravity forces. Aiming at further validating ATHLET 3-D capabilities against horizontal geometries, the experiment conditions are applied to a ROCOM model, which includes a newly developed horizontal pipe object to enhance code prediction inside coolant loops. The obtained results show code strong simulation capabilities to represent multidimensional flows. Enhanced prediction is observed at the vessel inlet compared to traditional 1-D approach, whereas mixing overprediction from the descending denser plume is observed at the upper-half downcomer region, which leads to eventual deviations at the core inlet.

The exploitation of mine wastes materials as secondary resources requires in-depth mineralogical analyses, with metal deportment being of particular relevance for metal recovery. Using a combination of the Scanning Electron Microscope (SEM)-based Mineral Liberation Analyser (MLA) and Electron Probe Micro-Analyser (EPMA) methods, the deportment of Zn and Pb in the historic Plombières mine site (East Belgium) was investigated. The mine site comprises four different materials: soil, metallurgical waste, brown and yellow tailings. The integration of the MLA and EPMA data allowed the identification and quantification of Pb- and Zn-bearing phases, including minerals present in low abundances as well as slag phases. Slags and Pb oxide are the main source of Pb in different types of mine waste, followed by cerussite and/or anglesite. Zinc is mainly hosted by slags or by fraipontite, with minor amounts contributed by sphalerite, gahnite, willemite and bannisterite. Uncertainties on the metal deportments were estimated by bootstrap resampling and are typically low. The highest uncertainties are observed when the metal-bearing phases are present in low abundances, which is particularly noticeable in the yellow tailings. Mineral deportment studies should become a standard step in the process of maximising the potential of the mine waste deposits. The integrated approach of MLA-EPMA applied in this paper has proven to be effective for assessing the metal deportment between mineral and slag phases in mine wastes and could be successfully applied for other metals and waste deposits which originated from pyrometallurgical processing of the ore.

Since the earliest days of the Industrial Revolution, Europe has been discarding vast quantities of mining residues, commonly referred to as “extractive waste.” Even today, 900 Mton/year of extractive waste is being generated and stored in tailing facilities or ponds. This 900 Mton/year fgure corresponds to 26% of the EU’s current waste output . Without proper management, tailing ponds can lead to environmental problems, ranging from acid mine drainage (AMD) and water contamination to dam bursts and fooding, air pollution, and soil erosion and contamination.

Here we present in-situ U-Pb laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) ages of andradite-grossular garnet from four magmatichydrothermal polymetallic skarn prospects in the Schwarzenberg District, Erzgebirge (Germany), located in the internal zone of the Variscan Orogenic Belt. Within the geochronological framework of igneous rocks and hydrothermal mineralization in the Erzgebirge, the obtained garnet ages define three distinct episodes of Variscan skarn formation: (I) early late-collisional mineralization (338-331 Ma) recording the onset of magmatic-hydrothermal fluid flow shortly after the peak metamorphic event, (II) latecollisional mineralization (~327-310 Ma) related to the emplacement of large peraluminous granites following large-scale extension caused by orogenic collapse and (III) post-collisional mineralization (~310-295 Ma) contemporaneous with widespread volcanism associated with Permian crustal reorganization. Our results demonstrate that the formation of skarns in the Schwarzenberg District occurred episodically in all sub-stages of the Variscan orogenic cycle over a time range of at least 40 Ma. This observation is consistent with the age range of available geochronological data related to magmatic-hydrothermal ore deposits from other internal zones of the Variscan Orogenic Belt in central and western Europe. In analogy to the time-space relationship of major porphyry-Cu belts in South America, the congruent magmatic-hydrothermal evolution in the internal zones and the distinctly later (by ~30 Ma) occurrence of magmatic-hydrothermal ore deposits in the external zones of the Variscan Orogenic Belt may be interpreted as a function of their tectonic position relative to the Variscan collisional front.

We present extensive new ab initio path integral Monte Carlo results for the momentum distribution function n(k) of the uniform electron gas in the warm dense matter regime over a broad range of densities and temperatures. This allows us to study the nontrivial exchange-correlation-induced increase of low-momentum states around the Fermi temperature, and to investigate its connection to the related lowering of the kinetic energy compared to the ideal Fermi gas. In addition, we investigate the impact of quantum statistics on both n(k) and the off-diagonal density matrix in coordinate space, and find that it cannot be neglected even in the strongly coupled electron liquid regime. Our results were derived without any nodal constraints, and thus constitute a benchmark for other methods and approximations.

Nonlinear optical (NLO) properties of materials can be enhanced by assembling them as thin polymer composite films alternating with other polymers and forming dielectric mirrors, 1D photonic crystals (1DPCs), wherein the input light intensity is increased. Based on the poly(vinyl carbazole) (PVK) and poly(vinyl alcohol) (PVA) contrasting polymer pair, variants of such structures, with graphene and fullerene in their high-index layers, have been produced. Their optical switching characteristics have been studied with ns, cw, and quasi-cw fs laser sources in the IR and with a fs laser in the visible range. We have demonstrated slow optical bistability in the polymeric 1DPCs determined by the thermal expansion of polymer composites at intensities over 100W/cm² as well as fast and ultrafast optical switching due to thermo-optic and Kerr nonlinearities, respectively. A subpicosecond fast spectral shift of the 1DPC bandgap has been found. Our results and analysis provide a clear picture of the NLO behavior of 1DPCs at different time scales. The results stimulate the subsequent design of ultrafast switches and bistable memory cells based on polymeric 1DPCs whose micrometer thickness and flexibility offer promise for implementation into fiber and microchip configurations.

Roasting is often required to convert sulfidic minerals into their sulfate and/or oxide
forms with the aim to increase the extractability of targeted metals. In this study, sulfidic
tailings of Neves Corvo Cu-Zn-Pb-mine (Portugal) underwent conventional roasting
(CR) and microwave-assisted roasting (MR) to compare the effect of both heating
technologies on the roasting behavior. Upon roasting in air, transformations of mineral
phases in the tailings were studied by different techniques using quantitative X-ray
diffraction, mineral liberation analysis, Raman spectroscopy and
thermogravimetry−differential scanning calorimetry coupled to mass spectrometry.
Also, the leaching behavior of elements from the roasted tailings was assessed in
water. It was shown that CR and MR induce different reaction pathways for mineral
transformations during roasting. While CR leads to indirect pyrite oxidation through
intermediate sulfate formation, direct pyrite oxidation to oxides is the main
transformation pathway during MR. This change in reaction pathway can be attributed
to the mineral-selective heating induced by microwaves, which in particular accelerates
the oxidation rate of pyrite. In agreement with the proposed roasting mechanism, the
leaching behavior in water showed that the CR tailings resulted in a higher extraction
of the studied metals (Cu, Zn, Pb, As and Fe) – since sulfate phases are more
abundant – compared to the MR tailings at the same roasting temperature. Overall, the
optimal leaching efficiency of the metals was reached after one hour roasting at
500−550 °C for both CR and MR

A short report on experiences with UV photocathode lasers and laser beam line for ELBE SRF Gun in 2020-2021

Malignant melanoma has a steadily increasing incidence, but treatment options are still limited and prognosis for patients, especially for men, is poor. To investigate whether targeting estrogen receptor (ER) signaling is a valid therapeutic approach, we retrospectively analyzed ER gene expression profiles in 448 melanoma patients. High ERα gene expression was associated with improved overall survival (HR=0.881; 95% Cl=0.793-0.979, p=0.018) and increased with tumor stage, while ERβ gene expression did not change with tumor progression. This seemingly protective function of ERα led us to speculate that specific targeting of ERβ has a therapeutic benefit in malignant melanoma. A new type of ERβ-selective ER modulator with a nitric oxide (NO•) releasing moiety (NO-SERM 4d) significantly reduced the pro-metastatic behavior of two melanoma cell lines (A2058 and MelJuso). Epithelial-mesenchymal transition in melanoma is consistent with a switch from E- to N-cadherin expression, mediating the invasive phenotype. NO-SERM 4d reduced N-cadherin expression and impaired spheroid formation in A2058 cells. In addition, growth of A2058 spheroids was significantly reduced, confirming the anti-tumorigenic potential of NO-SERM 4d. Targeting ERβ signaling combined with targeted NO• release represents a promising therapeutic approach in malignant melanoma that has potential to prevent metastatic spread and reduce tumor growth.

Background and Aims
After the rediscovery of the normal tissue sparing FLASH effect of high dose rate radiation, research activities on this topic have been revived. But especially for protons, the portfolio of accelerators capable of performing studies at ultra-high dose rates is limited. Laser-plasma accelerators (LPA) can generate extremely intense proton bunches of many 10 MeV kinetic energy. In combination with dedicated dose delivery systems, LPA proton sources facilitate peak dose rates well above 10^8 Gy/s in a pulse structure regime complementary to conventional accelerators.

Methods
The reliable generation of proton spectra beyond 60 MeV at DRACO-PW [Ziegler et al, SciRep2021], combined with a dedicated energy selective pulsed magnet beam transport system [Brack et al, SciRep2020], allows tailored sample-specific dose distributions. Adapted on-shot dosimetry enables the required spectral monitoring of every proton bunch. Two irradiation series on volumetric biological samples were performed at DRACO-PW, accompanied by reference irradiations at the University Proton Therapy Dresden.

Results
The first small animal pilot study at a laser-driven proton source was conducted successfully. The mouse-ear tumor model’s requirements [Beyreuther et al, PLoS One2018] were fulfilled and verified at high precision (+/- 5%) concerning predefined dose value and conformity. Complementary, a study investigating dose-rate effects such as FLASH was performed irradiating zebrafish embryos with above 10^9 Gy/s.

Conclusions
We present a laser-based irradiation platform at the DRACO-PW facility that enables systematic radiobiological studies, laying the foundations for further studies at LPA sources exploring ultra-high dose-rate effects, such as FLASH, over previously unreachable parameter space.

Resource estimates are crucial to assess the economic potential of tailings storage facilities (TSF) for re-mining and the extraction of critical raw materials. However, a lack of consensus exists on best practices in sampling for this purpose. This study aims to address this gap by assessing different sampling schemes for the resource classification of TSFs. To do so, one layer of a TSF was sampled with regular and nested grids of varying sizes and additional random holes. Systematic spatial trends in the data were fitted with low-order polynomial functions of the coordinates and the grayscale values of a historical aerial photograph of the same layer of the TSF. Variography was performed on the trend residuals, and exponential or Gaussian variogram models were fitted. Universal Sequential Gaussian Simulation was then used to produce 1000 realisations of the potential ground truth. The optimum sampling strategy was investigated by re-sampling these realisations with varying sample densities and configurations and using geostatistical modelling of the re-sampled data to assess uncertainties on the estimated metal tonnages. Robust estimates of metal tonnages can be achieved at relatively low sampling densities, particularly with regular grids. When historical image information is used, spatial variability is better reproduced, and a lower number of samples is required to reach a certain confidence level. Furthermore, an approach to approximate expected errors on grades/tonnages estimated with a given sampling scheme was developed to assess whether further sampling is required. Overall, the findings of this study have clear and transferrable implications for the best-practice sampling and modelling of TSFs and their critical raw materials resource.

Tailings deposits pose a significant threat to the environment and/or contain residual metal contents that may be of economic interest. The mineralogical and textural properties of the tailings dictate both processing behaviour and potential for acid mine drainage. Automated Scanning Electron Microscope-based image analyses enable valuable quantitative mineralogical and textural data to be obtained. Such methods are, however, time and cost intensive. This study investigates the extent to which geochemical data can be related to mineralogical and textural data of tailings materials. Models based on the dominant components (PCs) from principal component analysis were tested for mineral abundance, particle and mineral grain size distributions (PSD, GSD), and degree of liberation. Sedimentary-style deposition of the tailings is represented by the main trend between PCs 1 and 2. This being the dominant process in the tailings, PC 1 is a robust estimator for mineral contents, PSD and mineral GSDs. Degree of liberation is less well constrained by the PCs. This study clearly demonstrates that prediction of mineralogical and textural parameters of tailings with geochemical data is possible. Considering that the dominant processes should be the same in similarly deposited tailings deposits, these findings are widely applicable.

Laser-driven proton pulse provide unique properties in terms of pulse structure (ns) and instantaneous dose rate (10^9 Gy/s) but - inherently broadband and highly divergent - pose a challenge to established beamline concepts on the path to application-adapted irradiation field formation, particularly for three-dimensional cases. We present the successful implementation and characterisation of a highly efficient and tuneable dual pulsed solenoid beamline at the Draco PW facility[1] to generate volumetric dose distribution tailored to specific applications[2].
The vast experimental scope and already successfully performed studies range from systematic volumetric in-vivo tumour irradiations in a dedicated mouse model (with a stable mean dose delivery of ±10 % and a spatial dose homogeneity of ±5 % over a cylindrical volume of 5 mm diameter and height) to high-dose-rate irradiations in the FLASH regime (using proton peak dose rates of up to 10^9 Gy/s with about 20 Gy/shot homogeneously over a cylindrical sample volume of 4.5 mm diameter and 3 mm height) as well as particle diagnostics commissioning (with a multitude of spatial and spectral dose distributions).
The beamline setup is complemented by a complex beam monitoring and dosimetry detector suite adapted to the ultra-high dose rate pulses and is in its unique synergy and redundancy capable of %-level precision dose delivery to samples as required for systematic irradiation studies. In addition to established radiochromic film dosimetry, the detector suite includes saturation-corrected (transmission) ionisation chambers [3] as well as screen and bulk scintillator setups, partly with tomographic reconstruction capabilities for 3D dose distribution retrieval. Moreover, non-invasive, single-shot-capable online time-of-flight-based spectral characterisation of filtered proton pulses has proven a powerful tool for beam monitoring as well as dosimetric purposes.
In this presentation the complex and versatile dose delivery system of laser-driven protons at the Draco PW using pulsed solenoids will be discussed. Its characterisation, technological development and improvement as well as the dosimetry suite as a vital part of the precise dose delivery will be addressed, while the presentation by U. Schramm covers recent experimental activities in detail.
[1] T. Ziegler, et al., Proton beam quality enhancement by spectral phase control of a PW-class laser system, https://arxiv.org/abs/2007.11499 (2020)
[2] Brack, et al., Spectral and spatial shaping of laser-driven proton beams using a pulsed high-field magnet beamline, SciRep, 10:9118, (2020)
[3] Gotz M, et al., A new model for volume recombination in plane‐parallel chambers in pulsed fields of high dose‐per‐pulse. Phys Med Biol., 62: 8634, (2017)

Negative electron affinity (NEA) GaAs- and GaN-based photocathodes are used in modern night vison detectors and light emitting diodes. GaAs semiconductors are already used as electron sources in particle accelerators and well- studied. Like GaAs, GaN belongs to the
III-V semiconductor group. It is assumed that GaN, like GaAs, shows enormous potential as a novel electron source for particle accelerators.

P-type GaN on a substrate material (sapphire, silicon, copper or SiC) is activated by a thin layer of caesium and illuminated by ultra-violet (UV) light to observe the gained photocurrent.
To produce an NEA surface and thus highest QE, it is necessary to remove any residual impurities on the GaN surface. Therefore, GaN is chemical cleaned and vacuum thermal cleaned before it is caesium-activated. The aim of the thermal treatment is to gain an atomically clean surface.
With a comparison of different substrate materials, chemical pre-cleaning, thermal treatment, and activation parameters (e.g. caesium-flux), the quantum efficiency, lifetime and the re-activation of the photocathode is studied. Additionally, the morphology and surface composition of the GaN is examined by XPS, AFM, SEM and EDX.

Non-invasive technologies are key to ensure the sustainable exploration and mining of critical materials such as lithium and rare earth elements (REE), which are needed to fuel battery, magnet, and photovoltaic technologies for the transition towards low-carbon economies (Simandl, 2014; Heredia et al., 2020). Innovative sensing methods such as hyperspectral imaging allow the remote identification and high spatial resolution mapping of characteristic light-material interactions (“spectral fingerprints”) indicative of specific minerals and raw materials. These maps can be produced rapidly to support geological investigations during both exploration and mining. Accurate positioning of the spectral information in 3D space allows creating digital twins of outcrops and mines and integrating the spectral mapping results with other data such as from drill-cores or geochemical sampling. However, the low concentration and generally subtle spectral features associated with Li-bearing minerals and REEs present challenges for spectral mapping at the outcrop scale and, as a result, past approaches tended to use indirect methods based on, for example, mineral associations (Cardoso-Fernandes et al., 2020).

Direct mapping of REE and Li-minerals abundances is possible using high-resolution, high-quality spectral data and appropriate correction techniques. Sensor miniaturisation and development of new platforms now allow rapid data collection across large areas. These advances, combined with novel radiometric and geometric correction workflows (e.g., Thiele et al., 2021), recently allowed the direct detection of REE- and Li-bearing minerals at outcrop scale, using close-range ground-based (Boesche et al., 2015; Booysen et al., 2021) and drone-borne platforms (Booysen et al., 2020).

Two case studies of open pit mines, one from Siilinjärvi, Finland, (Fig. 1), the second from Uis, Namibia (Fig. 2) demonstrate direct 3D mapping of REE and Li mineralization. First, we captured ground-based photographs that were used for structure-from-motion, multi-view-stereo (SfM-MVS) photogrammetry to create digital 3D models/point clouds. We also acquired terrestrial hyperspectral data using a Specim AisaFENIX hyperspectral line scanner (push-broom scanner). The sensor covers the electromagnetic spectrum from 350 to 2500 nm over 384 spatial pixels and creates a datacube by steady rotation on a tripod. The resulting panoramic scan covers the important spectral features for both REE (578 nm, 740 nm and 799 nm for neodymium) and Li-bearing minerals (1593 nm and 1839 nm for cookeite, and 1538 nm, 1766 nm and 1850 nm for montebrasite). However, the results are subject to immense geometric distortion as well as radiometric influences that complicate spectral mapping, spatial localization, and scaling. We applied the processing workflow of Thiele et al. (2021) to perform accurate re-projection in original 3D space, as well as important radiometric and topographic corrections by combining the hyperspectral scans with the photogrammetric 3D point clouds. We subsequently mapped the occurrence of REEs and Li-bearing minerals based on their subtle characteristic spectral absorptions, and validated the occurrences by field sampling and mineralogical and elemental analysis.

Innovative remote sensing approaches allow us to rapidly create accurate digital twins of mine faces and natural geological outcrops. The 3D point clouds provide information on geological structure and can be populated with an arbitrary number of point-specific, high-dimensional spectral attributes to create so-called “hyperclouds”. Accurate processing and correction of the contained spectral information enables us to directly map the subtle spectral features indicative of Li- and REE-bearing minerals. Efficient remote mapping, even in complex terrains, is highly beneficial not only for initial targeting, but also for monitoring ore extraction, thus facilitating both exploration and optimized extraction.

We thank AfriTin and Yara Oy for support during field work, access to the mine site, and geological information and geochemical data.

The alpaka library is a header-only C++14 abstraction library for accelerator development. Its aim is to provide performance portability across accelerators through the abstraction (not hiding!) of the underlying levels of parallelism. In this lecture we will cover the basics of how to program HPC applications with alpaka in a portable yet performant way and have a look at the software ecosystem surrounding alpaka.

The Niederschlag fluorite-barite vein deposit in the Western Erzgebirge, Germany, has been actively mined since 2013. We present the results of a first comprehensive study of the mineralogy, petrography, fluid inclusions, and trace element geochemistry of fluorite related to the Niederschlag deposit. Two different stages of fluorite mineralization are recognized. Stage I fluorite is older, fine-grained, associated with quartz, and forms complex breccia and replacement textures. Conversely, the younger Stage II fluorite is accompanied by barite and often occurs as banded and coarse crystalline open-space infill. Fluid inclusion and REY systematics are distinctly different for these two fluorite stages. Fluid inclusions in fluorite I reveal the presence of a low to medium saline (7–20% eq. w (NaCl+CaCl2)) fluid with homogenization temperatures of 140–180 °C, whereas fluorite II inclusions yield distinctly lower (80–120 °C) homogenization temperatures with at least two high salinity fluids involved (18–27% eq. w (NaCl+CaCl2)). In the absence of geochronological data, the genesis of the earlier generation of fluorite-quartz mineralization remains enigmatic but is tentatively related to Permian magmatism in the Erzgebirge. The younger fluorite-barite mineralization, on the other hand, has similarities to many fluorite-barite-Pb-Zn-Cu vein deposits in Europe that are widely accepted to be related to the Mesozoic opening of the northern Atlantic Ocean.

Deportment data is essential for the planning of mining and minerals processing operations. This need is particularly tangible for deposits of noble metals, such as gold and silver. Therefore, the current paucity of published gold and silver deportment data for individual ore deposits and districts – and the concomitant lack of understanding how this relates to salient geological and mineralogical features of these deposits – is surprising. In the present study, we apply a combination of bulk geochemistry, laser ablation-inductively coupled plasma-mass spectrometry, electron microprobe analysis, and mineral liberation analysis to gold- and silver-rich samples from the epithermal Freiberg Ag-Pb-Zn district, to investigate variability in gold and silver deportments, as well as the corresponding geological/mineralogical controls. The results show that the main carriers of gold are electrum and arsenopyrite, whereas silver is mostly hosted by Ag-sulfosalts (pyrargyrite, miargyrite, polybasite) and fahlore (freibergite). Deportments vary greatly between samples. These variations can be related to the relative abundances of different minerals within the samples, which in turn reflect their spatio-temporal position within the district. Comparisons with other epithermal Ag-Pb-Zn districts similar to Freiberg indicate that the results presented here are of general significance.

With the interest in radiometal-containing diagnostic and therapeutic pharmaceuticals burgeoning, appropriate ligands to coordinate completely and stably said radiometals is essential. Reported here are two novel, bis(amido)bis(oxinate)diamine ligands, H2amidohox and H2amidoC3hox, that combine two 8-hydroxyquinoline with two amide donor groups and differ by one carbon in their 1,2-ethylenediaminevs. 1,3-diaminopropane backbones, respectively. Both ligands have been thoroughly studied via metal complexation, solution thermodynamics and radiolabeling with three radiometal ions: [nat/64Cu]Cu2+,
[nat/111In]In3+, and [nat/203Pb]Pb2+. X-ray crystallography determined the structures of the hexacoordinated Cu2+-ligand complexes, indicating a better fit of Cu2+ to the H2amidohox binding pocket. Concentration dependent radiolabeling with [64Cu]Cu2+ was quantitative as low as 1 μM with H2amidohox and 10 μM
with H2amidoC3hox within 5 minutes at room temperature. However, [64Cu][Cu(amidohox)] maintained higher kinetic inertness against a superoxide dismutase enzyme-challenge assay and ligand challengescompared to the [64Cu][Cu(amidoC3hox)] counterpart. Similarly, H2amidohox had significantly higher
radiochemical conversion with both [111In]In3+ (97% at 1 μM) and [203Pb]Pb2+ (97% at 100 μM) under mild conditions compared to H2amidoC3hox (76% with [111In]In3+ at 1 μM and 0% with [203Pb]Pb2+). By studying non-radioactive and radioactive complexation with both ligands, a comprehensive understanding of the coordination differences between two- and three-carbon diamine backbones is discussed. Overall, the ethylenediamine backbone of H2amidohox proves to be superior with rapid radiolabeling and kinetic inertness towards competing ligands and proteins.

In this work, we introduce a multi-sensor subspace-based clustering algorithm that benefits from fine spectral-resolution hyperspectral images (HSIs) and fine spatial resolution RGB images. In order to extract spatial information, a hidden Markov random field (HMRF) is employed on the fine spatial-resolution RGB image, whereas, spectral information is derived from an HSI using an advanced sparse subspace clustering algorithm. The proposed algorithm is validated on two real geological data sets. The experimental results in this study show that the proposed algorithm outperforms the state-of-the-art clustering algorithms in terms of clustering accuracy.

Mineral dust is an important natural source of aerosols and significantly influences air quality (Querol et al.,
Environ. Int., 2019) and the global radiation budget (Schepanski, Geosci., 2018). Frequent dust intrusions are
observed in the Mediterranean region (Ealo et al., Atmos. Chem. Phys., 2016; Pikridas et al., Atmos. Environ.,
2018) and Central Europe (Collaud Coen et al., Atmos. Chem. Phys., 2004; Schauer et al., Aerosol Air Qual.
Res., 2016), with high potential to cause exceedances of daily PM10 levels. To separate the influence of
anthropogenic and natural contribution to the PM10 levels, the new method was developed within the DNAAP
project (Detection of non-anthropogenic air pollution – http://www.aerosol.si/dnaap/).
Dust weakly absorbs light in the near ultra-violet and short wavelengths of the visible range, while the light
absorption of dust in longer wavelengths from the visible and near infra-red range is negligible. We used
filter-based photometer Aethalometer AE33 (Drinovec et al., Atmos. Meas. Tech., 2015) to measure the light
absorption at seven wavelengths, from 370 to 950 nm. The mineral dust is not the only light-absorbing aerosol
in the air. Black carbon (BC), a unique primary tracer for combustion emissions, strongly absorbs light across
the entire visual, near infra-red and near ultra-violet spectral range. Since optical absorption of mineral dust is
weaker than the optical absorption of black carbon, the coarse mode mineral particles have to be concentrated
using the high-volume virtual impactor (VI). The method is based on the optical absorption measurements of
the two sample streams, sampling particle size below 1 µm and sample stream with the concentrated coarse
mode particles, where mineral dust contribution is substantial. Experimental configuration includes two
Aethalometers AE33 with different size selective inlets: VI inlet for sampling coarse aerosol mode (mostly
mineral dust) and PM1 inlet for sampling fine mode of aerosols (mainly BC). The optical absorption of mineral
dust can be determined by subtracting the absorption of fine aerosol fraction (PM1) from the absorption of
aerosol sampled by the VI, taking into account the enhancement factor of VI setup (Drinovec et al., Atmos.
Meas. Tech., 2020). The mineral dust mass concentration is then calculated using mass absorption cross-section
(MAC) for dust which could be site and source-region specific.
The results from the measurement campaigns performed at six locations in the Mediterranean region will be
presented. The measurements took place in NE Spain (Barcelona – BCN, Montseny – MSY, Montsec – MSA),
on Cyprus (Nicosia – NI, Agia Marina Xyliatou – AMX), and in Slovenia (Ljubljana – LJ). Two year-long
datasets will be presented, focusing on the analyses of aerosol optical properties of PM1 and VI fractions. The
results were validated using low time resolution chemical specification of offline filters and a statistical
approach where dust was extracted from PM10 measurements for dust intrusions periods determined by models
and back-trajectory studies. For better understanding, helium ion microscopy (HIM) was applied to study the
microscopic differences between mineral dust and black carbon captured on the AE33 filter tapes.
This work was supported by SPIRIT Slovenia – Public Agency for Entrepreneurship, Internationalization,
Foreign Investments and Technology, project DNAAP.

Hyperspectral image analysis has considerably evolved during the past decades. The conventional model-based image processing and machine learning techniques are not efficient for hyperspectral image analysis, therefore, other advanced models such as spatial- spectral models were proposed to boost the hyperspectral analysis. Recent advances in ma- chine learning i.e., deep learning (DL) confirm that if an adequate amount of training data is supplied then DL-based algorithms outperform the conventional (shallow) ones. However, the available training data is often limited in hyperspectral imaging, therefore, the advan- tage of DL-based algorithms compared with shallow ones remains an open question. In this paper, we address this issue for two vibrant fields in hyperspectral image analysis i.e., Unmixing and Feature Extraction for Classification.

In this article, we introduce a deep learning-based technique for the linear hyperspectral unmixing problem. The proposed method contains two main steps. First, the endmembers are extracted using a geometric endmember extraction method, i.e., a simplex volume maximization in the subspace of the data set. Then, the abundances are estimated using a deep image prior. The main motivation of this work is to boost the abundance estimation and make the unmixing problem robust to noise. The proposed deep image prior uses a convolutional neural network to estimate the fractional abundances, relying on the extracted endmembers and the observed hyperspectral data set. The proposed method is evaluated on simulated and three real remote sensing data for a range of SNR values (i.e., from 20 to 50 dB). The results show considerable improvements compared to state-of-the-art methods. The proposed method was implemented in Python (3.8) using PyTorch as the platform for the deep network and is available online: https://github.com/BehnoodRasti/UnDIP.

In this letter, we propose a sparse unmixing technique using a convolutional neural network (SUnCNN) for hyperspectral images. SUnCNN is the first deep learning-based technique proposed for sparse unmixing. It uses a deep convolutional encoder-decoder to generate the abundances relying on a spectral library. We reformulate the sparse unmixing into an optimization over the deep network's parameters. Therefore, the deep network learns in an unsupervised manner to map a fixed input into the sparse optimum abundances. Additionally, SUnCNN holds the sum-to-one constraint using a softmax activation layer. The proposed method is compared with the state-of-the-art using two synthetic datasets and one real hyperspectral dataset. The overall results confirm that the proposed method outperforms the other ones in terms of signal to reconstruction error (SRE). Additionally, SUnCNN shows visual superiority for both real and synthetic datasets compared with the competing techniques. The proposed method was implemented in Python (3.8) using PyTorch as the platform for the deep network and is available online: https://github.com/BehnoodRasti/SUnCNN.

We propose a new fusion-based classification technique for optical multi-source remote sensing images called OptFus. OptFus is developed to merge and process optical imagery having different spatial and spectral resolutions. The spatial features are extracted using morphological filters from the RGB data containing high spatial resolution. A feature fusion technique is developed to combine all the sensor data in a subspace using a common set of representative features. Finally, the fused features are classified using a Support Vector Machine to ensure a robust supervised spectral classification. The proposed method is designed to allocate varying weights to the data from various imaging sensors in the fusion process. OptFus is applied to two multi-source optical datasets captured from geological drill-core samples. The classification accuracy demonstrates considerable improvements compared to the state-of-the-art. A Matlab implementation of OptFus is available online: https://github.com/BehnoodRasti/OptFus.

I present some results for an average-atom model derived from first principles. In particular, I compare different choices of boundary condition and exchange-correlation functional and examine their impact on the equation-of-state data and mean ionization state.

In simulations of the warm dense matter regime, it is typical to use a combined finite-temperature Kohn-Sham density-functional theory (KS-DFT) and molecular dynamics approach. However, in KS-DFT, (i) scaling worsens with increasing temperature, and (ii) temperature dependence is usually neglected in the exchange-correlation (XC) functional. We present a derivation from first-principles which reduces the full many-body Hamiltonian to an average-atom model in the dilute gas limit, which significantly reduces the computational cost of the KS-DFT calculation. We also show preliminary results including a comparison of temperature-dependent and zero-temperature XC functionals.

I demonstrate the key steps in a first-principles derivation of a density-functional average-atom model, and show the importance of the boundary condition used to solve the Kohn-Sham equations on results.

In this talk, I present a first-principles derivation of an average-atom model, which is solved using Kohn-Sham density-functional theory. I also demonstrate the impact of various assumptions and approximations of the model, such as the boundary condition and exchange-correlation functional, on various properties such as mean ionization state. I consider some possible causes of error in the model, specifically self-interaction and density-driven errors. Finally, I introduce our new open-source average-atom code for matter under extreme conditions, atoMEC.

We introduce an Öffner stretcher for the diode-pumped Petawatt laser system PENELOPE developed at the HZDR. Spectral and temporal aspects of the stretcher system with large aperture spherical mirrors (600 mm and 250 mm) are analyzed with respect to pulse shape and pulse contrast ratio.

The warm dense matter regime is notoriously challenging from a computational modelling perspective, since a quantum mechanical description is required for large length and time scales, at temperatures well above the ground-state. Average-atom (AA) models reduce the many-body system of interacting electrons and nuclei to an effective ‘average’ atom, with relevant properties typically computed via Kohn–Sham density-functional theory (KS-DFT): this approach has clear computational advantages relative to full ab initio KS-DFT (or alternative) simulations. However, there is some uncertainty regarding the accuracy and predictive capabilities of AA models. In this talk, I will present a first principles derivation of a KS-AA model, carefully analysing the assumptions made and terms neglected in this approach. The results obtained from our model highlight the importance of the choice of boundary conditions and the significance of the self-interaction error. I will discuss the implications of these findings for future developments in AA models.

Remote sensing provides valuable information about objects or areas from a distance in either active (e.g., RADAR and LiDAR) or passive (e.g., multispectral and hyperspectral) modes. The quality of data acquired by remotely sensed imaging sensors (both active and passive) is often degraded by a variety of noise types and artifacts. Image restoration, which is a vibrant field of research in the remote sensing community, is the task of recovering the true unknown image from the degraded observed image. Each imaging sensor induces unique noise types and artifacts into the observed image. This fact has led to the expansion of restoration techniques in different paths according to each sensor type. This review paper brings together the advances of image restoration techniques with particular focuses on synthetic aperture radar and hyperspectral images as the most active sub-fields of image restoration in the remote sensing community. We, therefore, provide a comprehensive, discipline-specific starting point for researchers at different levels (i.e., students, researchers, and senior researchers) willing to investigate the vibrant topic of data restoration by supplying sufficient detail and references. Additionally, this review paper accompanies a toolbox to provide a platform to encourage interested students and researchers in the field to further explore the restoration techniques and fast-forward the community. The toolboxes are provided in https://github.com/ImageRestorationToolbox.

Due to powerful feature extraction capability, convolutional neural networks (CNNs) have been widely used for hyperspectral image (HSI) classification. However, because of a large number of parameters that need to be trained, sufficient training samples are usually required for deep CNN-based methods. Unfortunately, limited training samples are a common issue in the remote sensing community. In this study, a dual graph convolutional network (DGCN) is proposed for the supervised classification of HSI with limited training samples. The first GCN fully extracts features existing in and among HSI samples, while the second GCN utilizes label distribution learning, and thus, it potentially reduces the number of required training samples. The two GCNs are integrated through several iterations to decrease interclass distances, which leads to a more accurate classification step. Moreover, a new idea entitled multiscale feature cutout is proposed as a regularization technique for HSI classification (DGCN-M). Different from the regularization methods (e.g., dropout and DropBlock), the proposed multiscale feature cutout could randomly mask out multiscale region sizes in a feature map, which further reduces the overfitting problem and yields consistent improvement. Experimental results on the four popular hyperspectral data sets (i.e., Salinas, Indian Pines, Pavia, and Houston) indicate that the proposed method obtains good classification performance compared to state-of-the-art methods, which shows the potential of GCN for HSI classification.

Earthquakes and heavy rainfalls are the two leading causes of landslides around the world. Since they
often occur across large areas, landslide detection requires rapid and reliable automatic detection
approaches. Currently, deep learning (DL) approaches, especially different convolutional neural
network and fully convolutional network (FCN) algorithms, are reliably achieving cutting-edge
accuracies in automatic landslide detection. However, these successful applications of various DL
approaches have thus far been based on very high-resolution satellite images (e.g., GeoEye and
WorldView), making it easier to achieve such high detection performances. In this study, we use freely
available Sentinel-2 data and ALOS digital elevation model to investigate the application of two wellknown FCN algorithms, namely the U-Net and residual U-Net (or so-called ResU-Net), for landslide
detection. To our knowledge, this is the first application of FCN for landslide detection only from freely
available data. We adapt the algorithms to the specific aim of landslide detection, then train and test
with data from three different case study areas located in Western Taitung County (Taiwan), Shuzheng
Valley (China), and Eastern Iburi (Japan). We characterize three different window size sample patches
to train the algorithms. Our results also contain a comprehensive transferability assessment achieved
through different training and testing scenarios in the three case studies. The highest f1-score value of
73.32% was obtained by ResU-Net, trained with a dataset from Japan, and tested on China’s holdout
testing area using the sample patch size of 64 × 64 pixels.

atoMEC is a python-based average-atom code for simulations of high energy density phenomena such as in warm dense matter. It is designed as an open-source and modular python package. atoMEC uses Kohn-Sham density functional theory, in combination with an average-atom approximation, to solve the electronic structure problem for single-element materials at finite temperature.

The 2021 Data Fusion Contest, organized by the Image Analysis and Data Fusion Technical Committee (IADF TC) of the IEEE Geoscience and Remote Sensing Society (GRSS), promotes research on geospatial artificial intelligence (AI) for social good. The global objective is to build models for understanding the state of and changes in artificial and natural environments from remote sensing data toward sustainable development.

Electrons in zero external magnetic field can be studied with density functional theory (DFT) or with spin-DFT (SDFT). The latter is normally used for open shell systems because its approximations appear to model better the exchange and correlation (xc) functional, but also because so far the application of DFT implied a closed-shell-like approximation. Correcting this error for open shells allows the approximate DFT xc functionals to become as accurate as those in SDFT. In the limit of zero magnetic field, the Kohn-Sham equations of SDFT emerge as the generalised KS equations of DFT.

This article presents the scientific outcomes of the 2020 Data Fusion Contest (DFC2020) organized by the Image Analysis and Data Fusion Technical Committee of the IEEE Geoscience and Remote Sensing Society. The 2020 Contest addressed the problem of automatic global land-cover mapping with weak supervision, i.e., estimating high-resolution semantic maps while only low-resolution reference data are available during training. Two separate competitions were organized to assess two different scenarios: 1) high-resolution labels are not available at all; and 2) a small amount of high-resolution labels are available additionally to low-resolution reference data. In this article, we describe the DFC2020 dataset that remains available for further evaluation of corresponding approaches and report the results of the best-performing methods during the contest.

Finite-temperature Kohn--Sham density-functional theory (KS-DFT) is a widely-used method in warm dense matter (WDM) simulations and diagnostics. Unfortunately, full KS-DFT-molecular dynamics models scale unfavourably with temperature and there remains uncertainty regarding the performance of existing approximate exchange-correlation (XC) functionals under WDM conditions. Of particular concern is the expected explicit dependence of the XC functional on temperature, which is absent from most approximations. Average-atom (AA) models, which significantly reduce the computational cost of KS-DFT calculations, have therefore become an integral part of WDM modelling. In this paper, we present a derivation of a first-principles AA model from the fully-interacting many-body Hamiltonian, carefully analysing the assumptions made and terms neglected in this reduction. We explore the impact of different choices within this model -- such as boundary conditions and XC functionals -- on common properties in WDM, for example equation-of-state data, ionization degree and the behaviour of the frontier energy levels. Furthermore, drawing upon insights from ground-state KS-DFT, we discuss the likely sources of error in KS-AA models and possible strategies for mitigating such errors.

Recent progress on advanced photocathodes for superconducting RF guns. We will together view the achievements as well as open questions for the reliable gun operation.

Recognizing less salient features is the key for model compression. However, it has not been investigated in the revolutionary attention mechanisms. In this work, we propose a novel normalization-based attention module (NAM), which suppresses less salient weights. It applies a weight sparsity penalty to the attention modules, thus, making them more computational efficient while retaining similar performance. A comparison with three other attention mechanisms on both Resnet and Mobilenet indicates that our method results in higher accuracy. Code for this paper can be publicly accessed at https://github.com/Christian-lyc/NAM.

Kinetics and efficiency of mineral dissolution of multi-phase rocks depend on the properties of the mineral assemblage, the fluid composition and the environment (e.g. temperature, fluid velocity at the surface). Links between the different effects can be studied by reactive transport simulations, although the input of experimentally obtained parameters is not always intuitive and its accuracy depends on the type of experiment. Using gold ore leaching with deep eutectic solvents as an example, we present an in-situ experimental setup to image in 3D the changes of the solid phases during mineral dissolution using laboratory X-Ray computed tomography (CT). Our new approach allows measuring the volume and surface area at different dissolution times, which are used to calculate the dissolution rate spectra for different mineral phases and to study the kinetics of the dissolution process. Additionally, the 3D model of the dissolving solid imaged by CT allows direct surface extraction for fluid flow simulations, which are used to investigate the effect of fluid velocity on the dissolution rate spectra. Altogether, our method can be used to study complex geochemical mineral – fluid interactions and to directly capture the link between different factors that affect dissolution, e.g.rate spectra and flow, which traditionally can only be studied by an indirect combination of experimental methods and modelling.

In this work, we propose a modular multi-modal architecture to automatically detect Alzheimer’s disease using the dataset provided in the ADReSSo challenge. Both acoustic and text-based features are used in this architecture. Since the dataset provides only audio samples of controls and patients, we use Google cloud-based speech-to-text API to automatically transcribe the audio files to extract text-based features. Several kinds of audio features are extracted using standard packages. The proposed approach consists of 4 networks: C-attention-acoustic network (for acoustic features only), C-Attention-FT network (for linguistic features only), C-Attention-Embedding network (for language embeddings and acoustic embeddings), and a unified network (uses all of those features). The architecture combines attention networks and a convolutional neural network (CAttention network) in order to process these features. Experimental results show that the C-Attention-Unified network with Linguistic features and X-Vector embeddings achieves the best accuracy of 80.28% and F1 score of 0.825 on the test dataset.

The Witwatersrand Supergroup in South Africa is not only the best-preserved sequence of Archean sedimentary rocks but also hosts the largest gold deposit on earth, yet discovered. The gold is situated in quartz-pebble conglomerates and is generally associated with a wide variety of minerals, including pyrite (FeS2), uraninite (UO2), pyrobitumen, base metal sulfides and phyllosilicates. Despite extensive research over the past 100 years, the origin of the gold is still debated with two models receiving most of the attention: the modified paleoplacer model and the hydrothermal models. The modified paleoplacer model assumes that detrital gold was transported into the host rock by fluvial processes, followed by a short-range mobilization (micrometer- to meter scale) by hydrothermal fluids that infiltrated the host rock. In the hydrothermal model, gold was introduced into the host rock by postdepositional hydrothermal fluids from an external source.
To find new evidence for the origin of gold, we present new high-resolution 3-dimensional (3D) data based on the combination of X-ray computed micro tomography (micro-CT) and spectral X-ray computed micro tomography (Sp-CT). We combine both imaging techniques as micro-CT is an excellent tool for high resolution structural characterization and Sp-CT has the power to chemically identify a selection of minerals relevant in this study. Sp-CT is based on the analysis of X-ray absorption spectra, in particular specific K-edges, which assists in identification of chemical elements in a sample. Since most of the previous research of the Witwatersrand gold focus on 2-dimensional methods, the outcomes of this study will help to understand the 3D spatial and size distribution of the gold and provide morphological information about individual particles.
This research has received funding from the European Institute of Innovation and Technology (EIT). This body of the European Union receives support from the European Union's Horizon 2020 research and innovation program.

Image-based analytical tools in geosciences are indispensable for the characterization of minerals, but most of them are limited to the surface of a polished plane in a sample and lack 3D information. X-ray micro computed tomography (micro CT) provides the missing 3D information of the microstructures inside samples. However, a major drawback of micro CT in the characterization of minerals is the lack of chemical information that makes mineral classification challenging.
Spectral X-ray micro computed tomography (Sp-CT) is a new and evolving tool in different applications such as medicine, security, material science, and geology. This non-destructive method uses a multi-pixel photon-counting detector (PCD) such as cadmium telluride (CdTe) in combination with a conventional CT scanner (TESCAN CoreTOM) to image a sample and detect its transmitted polychromatic X-ray spectrum. Based on the spectrum, elements in a sample can be identified by an increase in attenuation at specific K-edge energies. Therefore, chemically different particles can be distinguished inside a sample from a single CT scan. The method is able to distinguish elements with K-edges in the range from 25 to 160 keV, which applies to elements with Z > 48.
We present results from various sample materials. Different pure elements and element oxides were measured to compare the position of theoretical and measured K-edge energies. All measured K-edge energies are slightly above the theoretical value, but based on the results a correction algorithm could be developed. Furthermore, different monazite grains were investigated, which can be divided into two groups with respect to the content of different RE elements on the basis of the spectrum: La-Ce-rich and La-Ce-poor. In addition, samples from the Au-U Witwatersrand Supergroup demonstrate the potential applications of Sp-CT for geological samples. We measured different drill core samples from the Kalkoenkrans Reef at the Welkom Gold field. Sp-CT can distinguish gold, uraninite and galena grains based on their K-edge energies in the drill core without preparation.

Most ore characterization studies are based on analytical tools that are limited to a 2-dimensional (2D) surface of a 3-dimensional (3D) sample. This not only limits the number of particles available for analysis but also results in a lack of 3D morphological information. Especially for the characterization of trace phases, 2D analysis is time consuming. X-ray computed micro-tomography (micro-CT) represents an established 3D technique that is used in numerous applications such as medicine, material science, biology, and geoscience. However, a major drawback of micro-CT in the characterization of ores is the absence of chemical information, which makes mineral classification challenging.
Therefore, we present Spectral X-ray computed micro-tomography (Sp-CT). It is an evolving technique in different research fields and is based on a semiconductor detector that provides chemical information of a sample (e.g., CdTe). This detector can be used with a conventional CT scanner (TESCAN CoreTOM in this study) to image a sample and detect its transmitted polychromatic X-ray spectrum. Based on the spectrum, elements in a sample can be identified by an increase in attenuation at specific absorption edge energies. Therefore, chemically different minerals can be distinguished inside a sample from a single CT scan in the micrometer range. The method is able to distinguish elements with absorption edges in the range from 25 to 160 keV, which applies to elements with Z > 48.
We present the workflow of an ore characterization study using a combination of Sp-CT and high-resolution micro-CT with an example of Au-U ore from the Witwatersrand Supergroup. Different drill core samples from the Kalkoenkrans Reef at the Welkom Gold field were investigated. With the chemical information from the Sp-CT, minerals such as gold, U-phases, and galena can be identified based on their K-edge energies in the spectrum. Several Sp-CT scans were used to train a machine learning segmentation model in the software Dragonfly (version 2021.1) to segment the high-resolution CT data into multiple segments, e.g., gold, U-minerals, sulfide minerals and matrix minerals. The segmented data was then used to extract 3D mineral properties of the segments such as 3D volume or 3D surface area. This new non-destructive approach provides 3D information on distribution of chemically different minerals without any sample preparation. This information can be used for mineral processing simulations but also for genetic mineralogical studies.

Encouraged by the need for ecologically and economically sustainable technologies for the recovery of metals from complex raw materials, ionometallurgical leaching using deep eutectic solvents is emerging as a promising alternative to conventional hydro- and pyrometallurgy for metal recovery. Current approaches of studying leaching processes do not provide a mineral-based understanding of the leaching process – thus limiting the opportunities for process optimization. This study addresses this shortcoming by combining laboratory-based X-ray computed tomography (CT) and scanning electron microscopy-based image analysis. The latter method provides robust information on the mineralogy and texture of the leach feed material, whereas CT is used to observe the progress of the leaching process through time. Leaching of a Au-Ag bearing sulfide flotation concentrate by deep eutectic solvent ethaline with iodine as oxidizing agent is used as a relevant case study. Results show that time lapsed CT provides an accurate estimation of the dissolution rate of pyrite, chalcopyrite, galena, telluride minerals and gold. Dissolution rates were used then to simulate metal recoveries from the mineral concentrate as a function of leaching time. Simulation results are within 5% variation of recoveries obtained by batch leaching experiments. The developed workflow can be easily transferred to other mineral concentrates or ore types; results may be used to optimize industrial leaching process.

COVID-19 serves as an opportunity to apply scientific methodologies from various fields to better understand the nature of the spread. In this paper, we are specifically looking into the effect of the presence/absence of national borders on the spread of the disease. We work with the dataset aggregated from the seven-day incidence in the municipalities in Saxony, and Czechia. We further apply the regression model, where a metric to estimate the potential causality of incidence growth in each of two municipalities is dependent on the distance between them, their population ratio, and the existence of the national border. Our preliminary results showed that the effect of the borders is very strong in most of the study areas.

We present a new approach to 3-dimensional (3D) chemical imaging based on X-ray computed micro tomography (micro-CT), which enables the analysis of the internal elemental chemistry. The method uses a conventional laboratory-based micro-CT scanner (Tescan CoreTOM) equipped with a cadmium telluride (CdTe) semiconductor line detector (Tescan PolyDet). Based on the X-ray absorption spectra, elements in a sample can be distinguished by their specific K-edge energy. The capabilities and performance of this approach are illustrated with different experiments. We present results from various sample materials (e.g., pure element reference samples, mineral mixtures and rocks). Different pure elements and element oxides were measured to compare positions of the theoretical K-edge energy with the measured one. Furthermore, we show the results of a particle mixture with quartz as a low-absorbing matrix. Finally, samples of the Au-U Witwatersrand Supergroup demonstrate the possibilities this approach for geological samples. All results show that the method can distinguish elements with K-edges in the range of 25–160 keV. This corresponds to elements with Z > 48 (Cd). Moreover, the spectral information allows a distinction between materials, which show little to no X-ray attenuation variation in the reconstructed CT image.

The aim of this study was to assess the usefulness of pretherapeutic primary tumor
metabolic tumor volume (MTV) in the prognosis of radically treated cervical cancer patients. Ret-
rospective, single-centre analysis was performed on a group of 508 cervical cancer patients. All
patients underwent a pretreatment [18 F]FDG PET/CT study for the assessment of the disease stage.
Several PET-derived parameters—namely, maximum standardized uptake value (SUVmax ), mean
standardized uptake value (SUVmean ), total lesion glycolysis (TLG) and MTV, as well as the clinical
parameters, were analysed in terms of the overall survival (OS), event-free survival (EFS), locore-
gional control (LRC) and freedom from distant metastases (FFDM). Hyperthermia and brachytherapy
were prognostic for EFS, OS, and LRC.FIGO stage > II showed a significant effect on EFS, OS, and
FFDM. Moreover, hysterectomy was prognostic for OS and histology was prognostic for FFDM. From
the PET-derived parameters only MTV of the primary tumour had a significant influence on OS
(cutoff point: >12.7 mL, HR: 2.8, 1.75–4.48 95% CI, p < 0.001), LRC (cutoff point: >13.7 mL, HR 2.82,
1.42–5.61 95% CI, p = 0.003), EFS (cutoff point: >10.4 mL, HR: 2.57, 1.67–3.97 95% CI, p < 0.001) and
FFDM (cutoff point: >10.4 mL, HR: 5.04, 1.82–13.99 95% CI, p = 0.002). The pretreatment of MTV
in primary tumour is the only independent prognostic parameter in OS, LRC, EFS, and FFDM in
radically treated cervical cancer patients and should be used in clinical practice in assessing prognosis

Vortrag bei der JEFF Nuclear Data Week November 2021, JEF-DOC 2071

We are going to analyse the Herten aquifer, which is situated in Southern Germany. Multiple outcrop faces where surveyed and interpolated to a 3D dataset.

GeoStatTools provides geostatistical tools for various purposes:

• random field generation
• simple, ordinary, universal and external drift kriging
• conditioned field generation
• incompressible random vector field generation
• (automated) variogram estimation and fitting
• directional variogram estimation and modelling
• data normalization and transformation
• many readily provided and even user-defined covariance models
• metric spatio-temporal modelling
• plotting and exporting routines

Hyperspectral image (HSI) classification is an important topic in the community of remote sensing, which has a wide range of applications in geoscience. Recently, deep learning-based methods have been widely used in HSI classification. However, due to the scarcity of labeled samples in HSI, the potential of deep learning-based methods has not been fully exploited. To solve this problem, a self-supervised learning (SSL) method with adaptive distillation is proposed to train the deep neural network with extensive unlabeled samples. The proposed method consists of two modules: adaptive knowledge distillation with spatial–spectral similarity and 3-D transformation on HSI cubes. The SSL with adaptive knowledge distillation uses the self-supervised information to train the network by knowledge distillation, where self-supervised knowledge is the adaptive soft label generated by spatial–spectral similarity measurement. The SSL with adaptive knowledge distillation mainly includes the following three steps. First, the similarity between unlabeled samples and object classes in HSI is generated based on the spatial–spectral joint distance (SSJD) between unlabeled samples and labeled samples. Second, the adaptive soft label of each unlabeled sample is generated to measure the probability that the unlabeled sample belongs to each object class. Third, a progressive convolutional network (PCN) is trained by minimizing the cross-entropy between the adaptive soft labels and the probabilities generated by the forward propagation of the PCN. The SSL with 3-D transformation rotates the HSI cube in both the spectral domain and the spatial domain to fully exploit the labeled samples. Experiments on three public HSI data sets have demonstrated that the proposed method can achieve better performance than existing state-of-the-art methods.

Large-scale energy storage plants based on power-to-gas-to-power technologies incorporating high temperature electrolysis, catalytic methanation of H₂ and CO₂ and novel, highly efficient methane-fired Allam reconversion cycles allow for a confined and circular use of CO₂ and thus an emission-free storage of intermittent renewable energy. The Allam power cycle is considered as a beneficial power plant concept, which employs supercritical CO₂ as working fluid as well as an oxy-combustion process to reach high efficiencies of up to 66%. The combination of said process chain could reach a maximum roundtrip efficiency of 54.2 % assuming the presence of sufficient storage capacities for all relevant technical gases. In a technically feasible scenario, paired with a separate air separation unit instead of stationary O₂ storages, roundtrip efficiencies of 49.0 % were determined..
The implementation of said energy storage systems into existing national energy systems will pose a major challenge, since they will require far-reaching infrastructural changes to the respective systems itself, such as extensive installations of renewable generation and electrolysis capacities as well as sufficient subsurface storage capacities for both CO₂ and CH₂. Furthermore, an exemplary energy system forecast for Germany for the year of 2050 is presented to show the viability of the energy storage concept. In case of a fully circular use of CO₂, when electricity is solely generated by renewable energy sources (RES), 736 GW of RES, 234 GW of electrolysis and 62 GW of gas-to-power capacities are required. The total storage volume on the national scale of Germany for both CO₂ and CH₄ was determined to be 7.8 billion Nm³, respectively, leading to a CH₄ storage capacity of 54.5 TWh. The present investigation illustrates the feasibility of large-scale energy storage systems for renewable electricity based on high temperature electrolysis, catalytic methanation and Allam power cycles paired with large subsurface storages for CO₂ and CH₄.

Invited plenary presentation on:

Unravelling the secrets of laser plasma particle (ion) acceleration with x-rays from the European XFEL

This paper proposes a new approach based on an unsupervised deep learning (DL) model for landslide detection. Recently, supervised DL models using convolutional neural networks (CNN) have been widely studied for landslide detection. Even though these models provide robust performance and reliable results, they depend highly on a large labeled dataset for their training step. As an alternative, in this paper, we developed an unsupervised learning model by employing a convolutional auto-encoder (CAE) to deal with the problem of limited labeled data for training. The CAE was used to learn and extract the abstract and high-level features without using training data. To assess the performance of the proposed approach, we used Sentinel-2 imagery and a digital elevation model (DEM) to map landslides in three different case studies in India, China, and Taiwan. Using minimum noise fraction (MNF) transformation, we reduced the multispectral dimension to three features containing more than 80% of scene information. Next, these features were stacked with slope data and NDVI as inputs to the CAE model. The Huber reconstruction loss was used to evaluate the inputs. We achieved reconstruction losses ranging from 0.10 to 0.147 for the MNF features, slope, and NDVI stack for all three study areas. The mini-batch K-means clustering method was used to cluster the features into two to five classes. To evaluate the impact of deep features on landslide detection, we first clustered a stack of MNF features, slope, and NDVI, then the same ones plus with the deep features. For all cases, clustering based on deep features provided the highest precision, recall, F1-score, and mean intersection over the union in landslide detection.

invited talk on "High dose-rate in-vivo proton irradiation at DRACO-PW, pilot study results and prerequisites"

In recent years, many convolutional neural network (CNN)-based methods have been proposed to address the scene classification tasks of remote sensing images. Since the number of training samples in RS datasets is generally small, data augmentation is often used to expand the training set. It is, however, not appropriate when original data augmentation methods keep the label and change the content of the image at the same time. In this study, label augmentation (LA) is presented to fully utilize the training set by assigning a joint label to each generated image, which considers the label and data augmentation at the same time. Moreover, the output of images obtained by different data augmentation is aggregated in the test process. However, the augmented samples increase the intra-class diversity of the training set, which is a challenge to complete the following classification process. To address the above issue and further improve classification accuracy, Kullback–Leibler divergence (KL) is used to constrain the output distribution of two training samples with the same scene category to generate a consistent output distribution. Extensive experiments were conducted on widely-used UCM, AID and NWPU datasets. The proposed method can surpass the other state-of-the-art methods in terms of classification accuracy. For example, on the challenging NWPU dataset, competitive overall accuracy (i.e., 91.05%) is obtained with a 10% training ratio.

The impact of high hydrostatic pressure release after high-pressure torsion on subsequent high hydrostatic pressure annealing was analyzed by performing experiments on nanostructured Ni. Ni was deformed by high-pressure torsion at a pressure of 6GPa in 5 turns. Directly after deformation, the pressure was reduced to 2GPa, and under 2GPa annealing at 400 °C was conducted for 5 min. For comparison, samples were also annealed under 2GPa after deformation without loading between processes. Microhardness measurements, detailed microscopy observations and positron annihilation spectroscopy investigations were performed to elucidate the changes in the microstructures obtained after different processing routes. It is demonstrated that the pressure applied between deformation and high hydrostatic pressure annealing caused an increase in microhardness by 20% in comparison with pressure realize. Moreover, the pressure applied had an impact on the vacancy concentration, and consequently on the microstructure, leading to a smaller average grain size and a more heterogenous microstructure in terms of grain size, leaving space for optimizing the strength-ductility balance.

Im Rahmen einer ressourceneffizienten Kreislaufwirtschaft (engl. Circular Economy) hat das Recycling von Produkten am Ende ihrer Lebenszyklen das Potenzial, den Ressourcenverbrauch und klimaschädliche Umweltauswirkungen von Produktsystemen zu verringern. Für eine nachhaltige Etablierung von Recycling-Lösungen sind neben der Recyclingindustrie auch Produkthersteller bereits in der Konstruktionsphase mit einzubeziehen. Hierfür fehlt es derzeit jedoch meist an Methoden, um die Auswirkungen von Designentscheidungen auf die Recyclingfähigkeit abzuschätzen und zu optimieren (engl. Design-for-Recycling). Deshalb arbeiten wir an einem digitalen Modell, das die Bewertung der Recyclingfähigkeit schon im Produktentstehungsprozess ermöglichen soll. Ein besonderer Fokus liegt dabei auf dem Prozessschritt der Aufschlusszerkleinerung, da eine zunehmende Anzahl von Produkten, von Fahrzeugen bis Haushaltsgeräten, aus Multi-Material Strukturen bestehen. Hier müssen Verbindungen zwischen unterschiedlichen Materialien im Recycling meist wieder gelöst werden, um hohe Recyclingraten für alle verbauten Materialien zu erzielen. Dies erfolgt typischerweise mechanisch durch Zerkleinerungsprozesse.
Vor diesem Hintergrund wollen wir herausfinden, welche Parameter bereits während der Bauteilentwicklung beeinflussbar sind, um den Materialaufschluss bei der mechanischen Aufbereitung zu optimieren, ohne dabei Funktion und Lebensdauer der Struktur in der Nutzungsphase zu beeinträchtigen. Dazu haben wir zum einen Zerkleinerungsexperimente von Prüfkörpern aus der Automobilbranche in einem Rotorreißer durchgeführt. Im Mittelpunkt der Untersuchungen stand dabei insbesondere der Einfluss verschiedener Verbindungscharakteristika auf das Aufschlussverhalten. Da die experimentelle Datenerhebung aufgrund hoher Parametervariabilität im Produktdesign und des Zerkleinerungsprozesses aufwändig ist, entwickeln wir zum anderen ein physikalisch basiertes, numerisches Modell der Aufschlusszerkleinerung mithilfe der Finiten Elemente Methode. Hierzu nutzen wir die Software LS-DYNA, verwenden darin Materialmodelle, welche die Plastizität und das Versagen der beteiligten Werkstoffe sowie deren Interfaces berücksichtigen. Zudem wird die Simulation für verschiedene Lastfälle parametrisiert, wie beispielsweise unterschiedliche Orientierungen des Prüfkörpers im Rotorreißer. Neben dem experimentellen zeigen wir erste numerische Ergebnisse der Berechnungen am Beispiel einer Metall-Kunststoff-Hybridstruktur. Damit leistet unsere Arbeit einen Beitrag dazu, den Einfluss von Konstruktionsentscheidungen auf das Aufschlussverhalten abzuschätzen, sowie Erfahrungen bei der Erstellung einer durchgängigen digitalen Kette vom Design über Fertigung bis hin zum Recycling zu sammeln.

We study the dissipative Fermi-Hubbard model in the limit of weak tunneling and strong repulsive interactions, where each lattice site is tunnel-coupled to a Markovian fermionic bath. For cold baths at intermediate chemical potentials, the Mott insulator property remains stable and we find a fast relaxation of the particle number towards half filling. On longer time scales, we find that the anti-ferromagnetic order of the Mott-Néel ground state on bi-partite lattices decays, even at zero temperature. For zero and non-zero temperatures, we quantify the different relaxation time scales by means of waiting time distributions which can be derived from an effective (non-Hermitian) Hamiltonian and obtain fully analytic expressions for the Fermi-Hubbard model on a tetramer ring.

Multi-material structures are usually designed into products with the aim of optimizing the operation phase, e.g. lightweight structures leading to higher energy efficiency of vehicles. However, in the recycling phase, most materials connected in multi-material structures should be liberated and separated again to enable high material recoveries. We want to contribute to estimating recyclability of multi-material structures already at the design stage by investigating material liberation through shredding. Therefore, we conducted an experimental shredding study, where we observed liberation behaviour and characterized in- and output particles. Furthermore, we are currently working on the numerical simulation of shredding processes using the Finite Element Method. In the future, we want to analyse the influence of design parameters on liberation behavior and, thereby, contribute to making more sustainable design decisions.

Most products in consumables, aerospace and automotive industry are multi-material structures, which consist of materials connected through different joining techniques. Multi-material design aims for optimizing performance during production and service phase but usually does not consider the recycling. During recycling, materials combined in multi-material structures need to be liberated, i.e. disconnected, again to enable high material recoveries in subsequent recycling processes.
The goal of this study is to investigate material liberation of multi-material structures through mechanical processing, namely shredding. An experimental study is conducted where automotive A-frame-based specimens are shredded in a rotary shear. Particle analysis and characterization is supported by computer tomography. Furthermore, a first approach to model material liberation using Finite Element Method is introduced. This study is a contribution to estimating product recyclability in terms of material liberation already at the design stage, thereby supporting the design-engineer in sustainable design and product optimization.

The accurate numerical treatment of matter under extreme conditions is crucial for the understanding of important physical phenomena such as radiation damage in fusion reactor walls, or planetary interiors. Yet, such simulations are unfeasible with state-of-art methods, e.g., density functional theory (DFT) if performed at large length and time scales, due to unfavorable scaling behavior. One possible route to mitigate these scaling issues are machine-learning based surrogate models; DFT data is used to calculate models that allow access to the same quantities of interest a DFT simulation would, at drastically reduced computational cost. CASUS (in cooperation with SNL and ORNL) develops a framework called "Materials Learning Algorithms" (MALA), drawing on which DFT surrogate models can easily be created and applied. Here we present an overview of MALA and recent results, such as size transferability and automated model construction.

The accurate modeling of materials is a fundamental task in material science. Advanced methods such as Density Functional Theory (DFT) provide quantum chemical accuracy through explicit calculation of the electronic structure of materials, but they come at high computational costs. These computational demands are especially prohibitive in the context of dynamic investigations. Increasingly efficient implementations of DFT can only alleviate this problem to a certain degree.
Here, we present a different approach to tackle this problem. Feed-forward neural networks are trained on electronic structure data in order to replace DFT calculations at a fraction of the computational cost. Such surrogate models can be used to model matter under extreme conditions as they occur in planetary interiors or fusion reactors across multiple length and time scales.
To facilitate the training, testing, and application of DFT surrogate models, the Center for Advanced Systems Understanding develops the Materials Learning Algorithm (MALA) package as an open-source software project in collaboration with the Sandia National Laboratories and Oak Ridge National Laboratory.

With the growth of computational resources, the scope of electronic structure simulations has increased greatly. Artificial intelligence and robust data analysis hold the promise to accelerate large-scale simulations and their analysis to hitherto unattainable scales. Machine learning is a rapidly growing field for the processing of such complex datasets. It has recently gained traction in the domain of electronic structure simulations, where density functional theory takes the prominent role of the most widely used electronic structure method. Thus, DFT calculations represent one of the largest loads on academic high-performance computing systems across the world. Accelerating these with machine learning can reduce the resources required and enables simulations of larger systems. Hence, the combination of density functional theory and machine learning has the potential to rapidly advance electronic structure applications such as in-silico materials discovery and the search for new chemical reaction pathways. We provide the theoretical background of both density functional theory and machine learning on a generally accessible level. This serves as the basis of our comprehensive review including research articles up to December 2020 in chemistry and materials science that employ machine-learning techniques. In our analysis, we categorize the body of research into main threads and extract impactful results. We conclude our review with an outlook on exciting research directions in terms of a citation analysis.

The crystal structure and magnetic properties of the multicomponent compounds (Tb_1−xY_x)_0.8Sm_0.2Fe₂H_z (x = 0, 0.2, 0.4, 0.6, 0.8, 1; z = 0 and 3.7) are investigated. The compounds crystallize in the MgCu₂ type of structure. While the parent compounds Tb_0.8Sm_0.2Fe₂ and Y_0.8Sm_0.2Fe₂ are single phase, we detect 5%–8% of a second phase with a crystal structure of the PuNi₃ type (space group R3m) in the alloys with 0.2 ≤ x < 0.8. Hydrogen absorption does not change the space group of the (Tb,Y,Sm)Fe₂ compounds but boosts significantly the lattice parameter a. A large volume change of ΔV/V ∼ 28% upon hydrogen absorption is observed. By applying high magnetic fields up to 58 T, we observed rotations of the magnetic sublattices and hence we were able to determine the critical transition fields, H, from the ferrimagnetic to the ferromagnetic state and the inter-sublattice exchange parameter λ. The magnetic compensation occurs at x ≈ 0.6 and 0.2 in (Tb_1−xY_x-)_0.8Sm_0.2Fe₂H_z at z = 0 and 3.7, respectively. While maintaining the collinear magnetic structure, the phenomenon of compensation in hydrides should be observed at x ≈ 0.4.

Polydisperse multiphase flows appear in a multitude of industrial processes. Depending on the application, the fluid or solid particles differ not only in size, but also with respect to other variables such as their velocity, shape, temperature, crystal structure or chemical composition. These secondary properties can significantly influence the corresponding process or the performance of the end product. An example are bubbly flows in vertical channels. The velocity vectors of the individual bubbles depend on their size, which ultimately determines the gas phase distribution in the pipe cross-section. Another example is the gas phase synthesis of ceramic powder in high temperature processes. The resulting particle aggregates are usually non-spherical due to the competition of aggregation and sintering of primary particles. The aggregate morphology affects the likelihood of collisions and the primary particle size determines the characteristics of the product powder. The evolution of property distributions within the dispersed phase can be described with the population balance equation. Its coupled solution with the governing equations of fluid flow allows to consider spatial dependencies. In the present thesis, a flexible population balance model has been developed and combined with a multifluid solver within the open source Computational Fluid Dynamics library OpenFOAM. The population balance equation is solved with the method of classes. The applied technique preserves the total mass and number of particles and allows for an arbitrary discretization of the distribution function. A new formulation that allows a direct implementation of binary breakup models with an implicitly given daughter size distribution is proposed which eliminates the need for an additional numerical integration. Further, a general approach for predicting the evolution of secondary size-conditioned properties is presented. The flexibility of the developed population balance model is demonstrated by applying it to two fundamentally different problems. First, the cocurrent flow of air and water in a vertical pipe is simulated. Predicting the development of the lateral void fraction profile is still a largely unsolved problem and requires proper modeling of several physical mechanisms. In dealing with the complexity a stepwise validation strategy is adopted, whereby the limits of each model layer are determined for a large matrix of measured superficial velocities. By employing an established model for the momentum exchange between the phases it is shown that, in the case of a nearly developed flow, especially the transition region between bubbly and slug flow can be simulated reliably. Next, using volume-averaged flow parameters, the performance of several coalescence and breakup model combinations is assessed. Promising results are obtained for some cases, albeit the models still require further development and calibration. Finally, the developing flow is simulated and it is shown that a complete model for predicting transitions between flow regimes must account for the size dependency of the bubble motion, as possible with the developed population balance model. The second application is the synthesis of titania in an aerosol reactor. The specific surface area of the aggregates is considered as a secondary property here. In combination with a constant fractal dimension their collision diameter can be modeled. The mean primary particle diameter can be inferred from it as well. The created model helps in explaining the trends observed in the experiment, which is not possible on the basis of considering merely the aggregate size distribution or using a simplified description of the reactor geometry.

Efcient magnetic control of electronic conduction is at the heart of spintronic functionality for memory and logic applications. Magnets with topological band crossings serve as a good material platform for such control, because their topological band degeneracy can be readily tuned by spin confgurations, dramatically modulating electronic conduction. Here we propose that the topological nodal-line degeneracy of spin-polarized bands in magnetic semiconductors induces an extremely large angular response of magnetotransport. Taking a layered ferrimagnet, Mn₃Si₂Te₆, and its derived compounds as a model system, we show that the topological band degeneracy, driven by chiral molecular orbital states, is lifted depending on spin orientation, which leads to a metal–insulator transition in the same ferrimagnetic phase. The resulting variation of angular magnetoresistance with rotating magnetization exceeds a trillion per cent per radian, which we call colossal angular magnetoresistance. Our fndings demonstrate that magnetic nodal-line semiconductors are a promising platform for realizing extremely sensitive spin- and orbital-dependent functionalities.

Magnetocaloric materials change their temperature when a magnetic field is applied or removed, which allows building a magnetic cooling device. We derive an analytical expression for the maximum heat that such a material can transfer in one cooling cycle by investigating the operation of a simplified active magnetic regenerator (AMR). The model largely only depends on the adiabatic temperature change, the specific entropy change, and the temperature span between the hot and cold reservoirs. While this expression overestimates the performance of a real AMR due to its simplification, it can predict an upper limit of any AMRs’ performance independent of the implementation details. Based on this, we calculate the upper limit of the cooling power of magnetic cooling devices at any temperature span, frequency, mass, and material. This upper limit is used to predict how the thermal span is scaling with the applied magnetic field, and it can be utilized for the optimization of the magnetic field source. Additionally, we confirm that the product of isothermal entropy and adiabatic temperature change, already used in the literature, is a suitable figure of merit for magnetocaloric materials.

Aim: Further development and improvement of disease diagnostic devices and protocols
So far: Very precise, but bulky, time-consuming and costly, trained personnel and laboratory conditions required

The family of 2D layered materials has gained enormous attention of materials scientists and researchers from other fields of science. This stems from the fact that 2D monolayers (1Ls) can exhibit remarkably different electronic properties than their bulk counterparts. Moreover, stacking different 1Ls, results in yet different electronic properties than these of the 1Ls. Recently, among others, van der Waals heterostructures (vdW HS) of transition-metal dichalcogenides (TMDC) have been extensively studied due to their type-II band alignment.
This thesis summarizes four different theoretical studies on layered 2D materials. The first study investigates the potential existence of a new family of bulk layered materials with chemical formula XY3 (where X = group 14; Y = group 15). The low cleavage energies indicate the potential exfoliation as mono- and bi- layers (2Ls), where most of the exfoliated layers are thermally and dynamically stable. Interestingly, many 1Ls and 2Ls show strong quantum confinement and turn into indirect semiconductors, unlike bulks which are all metals. Such metal to semiconductor transition was previously known for noble-metal dichalcogenides. Next study shows one of the potential applications of XY3, that is, single-material logical junction for gas sensing applications. A device that consists of metallic multilayers (3L) as electrodes and semiconducting 1L as scattering region. To do so, one of the exemplary materials (SnP3) was picked to construct a single-material device. The results combining density functional theory (DFT) and non-equilibrium Green’s function (NEGF) calculations revealed that SnP3 is an ideal material for gas sensing applications, especially for poisonous NO gas molecules. For NO molecules, this device showed a negative differential resistance (NDR) at small bias voltages.
Moreover, electronic properties of vdW TMDC HS were investigated for the HS having up to six layers. In this part, it was essentially studied how the electron and hole states in heterobilayer (HBL), namely MoS2/WSe2, will change as a function of additional electron/hole layers. We found that additional electron layers will result in equal delocalization of electron states among all layers forming conduction bands. However, additional hole layers do not alter the holes states distribution much, i.e., hole states stay localized at HBL+1L (WSe2).
The last study, in this thesis, focused on understanding the interplay between the so- called atomic collapse states and moiré potential in twisted bilayer graphene (tBLG) systems. It was found that individual graphene layers did host collapse states. However, moiré potential in tBLG may destroy the collapse states, although tBLG systems have similar band dispersion as in graphene.
The knowledge acquired from the findings presented in this thesis can provide new potential candidates as well as some helpful insights into electronic properties of existing layered materials, for their applications in the future nano(opto)electronic devices.

Introduction: Selective A₂AR antagonists have emerged as potential therapeutics for multiple diseases. With regard to Parkinson’s disease, adjunctive treatment of A₂AR antagonists potentially reduces adverse effects of long-term L-DOPA treatment. Therefore, imaging of receptor availability during the A2AR-tailored therapy is of utmost importance. We recently developed [¹⁸F]FLUDA as an novel A₂AR-specific PET radiotracer [1].
Methods: [¹⁸F]FLUDA was synthesized by an automated procedure Biological evaluation was performed in healthy mice and piglets. In vitro autoradiography was performed with brain cryosections. In vivo metabolism was analysed by radio-HPLC of plasma and brain homogenate. Pharamcokinetics and biodistribution was assessed by dynamic PET imaging under control and blocking conditions (2.5 mg/kg tozadenant and/or 1.0 mg/kg istradefylline). SUV ratio (SUVr) of striatum-over-cerebellum was used as a metric for specific uptake. A single dose acute toxicity study was performed in Wistar rats according to the ICH guideline M3(R2). Radiation dosimetry was investigated in piglets.
Results: In vitro autoradiography revealed an A₂AR affinity (KD;) of 4.3 and 0.7 nM and an A₂AR density (Bmax) of 556 and 218 fmol/mg in the striatum of mice and piglets. No radiometabolites were detected in the mouse brain at 15 min p.i., whereas radiometabolites were found in piglet plasma but are assumed to not cross the blood-brain barrier. PET demonstrated high specific binding of [¹⁸F]FLUDA in both species (Fig.1). Toxicity studies revealed no adverse effects up to a dose of 30 µg/kg (~4000-fold of expected human dose). The ED to humans is 16.4 µSv/MBq, which is in the range of other ¹⁸F-labeled radiotracers [2].
Conclusion: We have demonstrated that [¹⁸F]FLUDA is suitable for determination of the A₂AR availability in the striatum. No safety concerns are expected upon administration of [¹⁸F]FLUDA according to toxicity and dosimetry data. These results encourage the clinical translation of [¹⁸F]FLUDA.
Acknowledgement: This work (Project No. 100226753) has been funded by the European Regional Development Fund (ERDF) and Sächsische Aufbaubank (SAB).
References: [1] Lai, T.H., Toussaint, M., Teodoro, R., Dukić-Stefanović, S., Gündel, D., Ludwig, F.-A., Wenzel, B., Schröder, S., Sattler, B., Moldovan, R.-P., Falkenburger, B.H., Sabri, O., Winnie Deuther-Conrad, W., Peter Brust, P. Improved in vivo PET imaging of the adenosine A₂A receptor in the brain using [¹⁸F]FLUDA, a deuterated radiotracer with high metabolic stability. Eur J Nucl Med Mol Imaging 2021, 48, 2727–2736. [2] Sattler, B., Kranz, M., Lai, T.H., Gündel, D., Toussaint, M., Schröder, S., Moldovan, R.-P., Winnie Deuther-Conrad, W., Teodoro, R., Sabri, O., Peter Brust, P. Preclincal incorporation dosimetry of [¹⁸F]FLUDA - a novel ¹⁸F-labeled tracer for PET imaging of the expression of the adenosine A₂A receptor (A₂AR). J Nucl Med 2020, 61:1014.

²¹⁰Pb (𝑇1∕2=22.2 y) is an important source of background in rare event searches, such as neutrinoless double-𝛽 decay and dark matter direct detection experiments. In this paper, the capabilities of the A.E. Lalonde AMS Laboratory at the University of Ottawa for ²¹⁰Pb measurements are discussed.
For fluoride targets, the blank ²¹⁰Pb/²⁰⁶Pb ratio was in the 1e-14 to 1e-13 range, but (PbF₃)- current output was lower and less stable. For oxide targets, (PbO₂)- current output showed better stability, despite a significant difference in current output for commercial PbO and processed samples, and background studies suggested a background not much higher than that of the fluoride targets. Both target materials showed, therefore, good performance for ²¹⁰Pb Accelerator Mass Spectrometry assay.
Measurements of Kapton ultra-thin films were performed. 90% C.L. upper limits for the ²¹⁰Pb specific activity in the range of 0.74–2.8 Bq/kg were established for several Kapton HN films.

This document is intended as a supplement to the EANM “Guidelines on current Good Radiopharmacy Practice (cGRPP)” issued by the Radiopharmacy Committee of the EANM [1]. The aim of the EANM Radiopharmacy Committee is to provide a document that describes how to manage risks associated with small-scale “in-house” preparation of radiopharmaceuticals, not intended for commercial purposes or distribution.

Background: Cyclotrons form a central infrastructure and are a resource of medical radionu- clides for the development of new radiotracers as well as the production and supply of clini- cally established radiopharmaceuticals for patient care in nuclear medicine.
Aim: To provide an updated overview of the number and characteristics of cyclotrons that are currently in use within radiopharmaceutical sciences and for the development of radiopharma- ceuticals to be used for patient care in Nuclear Medicine in Germany (D), Austria (A) and Switzerland (CH).
Methods: Publicly available information on the cyclotron infrastructure was (i) consolidated and updated, (ii) supplemented by selective desktop research and, last but not least, (iii) vali- dated by members of the committee of the academic “Working Group Radiochemistry and Radiopharmacy” (AGRR), consisting of radiochemists and radiopharmacists of the D-A-CH countries and belonging to the German Society of Nuclear Medicine (DGN), as well as the Radiopharmaceuticals Committee of the DGN.
Results: In total, 39 cyclotrons were identified that are currently being operated for medical radionuclide production for imaging and therapy in Nuclear Medicine clinics, 29 of them in Germany, 4 in Austria and 6 in Switzerland. The majority of the cyclotrons reported (69%) are operated by universities, university hospitals or research institutions close to a university (clinic), less by/in cooperation with industrial partners (26%) or a (non-university) clinic/PET-center (5%). Most of the cyclotrons (82%) are running with up to 18 MeV proton beams, which is sufficient for the production of the currently most common cyclotron-based radionuclides for PET imaging.
Discussion: The data presented provide an academically-updated overview of the medical cy- clotrons operated for the production of radiopharmaceuticals and their use in Nuclear Medi- cine in the D-A-CH countries. In this context, we discuss current developments and trends with a view to the cyclotron infrastructure in these countries, with a specific focus on organi- sational aspects.

[18F]FLUDA is a selective radiotracer for in vivo imaging of the adenosine A2A receptor (A2AR) by positron emission tomography (PET). Promising preclinical results obtained by neuroimaging of mice and piglets suggest the translation of [18F]FLUDA to human PET studies. Thus, we report herein a remotely controlled automated radiosynthesis of [18F]FLUDA using a GE TRACERlab FX2 N radiosynthesizer. The radiotracer was obtained by a one-pot two-step radiofluorination procedure with a radiochemical yield of 9 ± 1%, a radiochemical purity of ≥ 99% and molar activities in the range of 69 333 GBq/µmol at the end of synthesis within a total synthesis time of approx. 95 min (n = 16). Altogether, we successfully established a reliable and reproducible procedure for the automated production of [18F]FLUDA.

The talk will cover some recent attempts to recreate and describe the states of matter as in brown dwarfs, ice giant planets, or the earth. In particular, state of the art methods will be described to model the x-ray scattering signal of such warm dense matter states. Density functional theory seems to be a very promising tool, although upon close inspection, several problems arise that strongly limit the predictive power of this method. This leads to investigations of the underlying model system, the electron gas, using quantum Monte Carlo methods. Finally, we show a promising new method to study higher order correlations in the system by nonlinear excitations.

Ultrafast magnetization dynamics are governed by energy flow between electronic, magnetic, and lattice degrees of freedom. A quantitative understanding of these dynamics must be based on a model that agrees with experimental results for all three subsystems. However, ultrafast dynamics
of the lattice remain largely unexplored experimentally. Here, we combine femtosecond electron diffraction experiments of the lattice dynamics with energy-conserving atomistic spin dynamics (ASD) simulations and ab-initio calculations to study the intrinsic energy flow in the 3d ferromagnets cobalt (Co) and iron (Fe). The simulations yield a good description of experimental data, in particular an excellent description of our experimental results for the lattice dynamics. We find that the lattice dynamics are influenced significantly by the magnetization dynamics due to the energy cost of demagnetization. Our results highlight the role of the spin system as the dominant heat sink in the first hundreds of femtoseconds. Together with previous findings for nickel, our work demonstrates that energy-conserving ASD simulations provide a general and consistent description of the laser-induced dynamics in all three elemental 3d ferromagnets.

Ion stopping in warm dense matter is a process of fundamental importance for the understanding of the properties of dense plasmas, the realization and the interpretation of experiments involving ion beam-induced warm dense matter samples, and for inertial confinement fusion research. The theoretical description of the ion stopping power in warm dense matter is difficult notably due to electron coupling and degeneracy, and measurements are still largely missing. In particular, the low-velocity stopping range around the Bragg peak, that features the largest modelling uncertainties, remains virtually unexplored. Here, we report proton energy-loss measurements in warm dense plasma at lower projectile velocities than previous experiments, coming significantly closer to the Bragg peak region. Our energy-loss data, combined with a precise target characterization based on plasma temperature measurements us-
ing two different diagnostics, demonstrate a significant deviation of the stopping power from classical models in this regime. In particular, we show
that our results are consistent with recent first-principles simulations based on time-dependent density functional theory.

The quasiparticle electronic structure and optical excitation of anatase TiO2 is determined within the framework of many-body perturbation theory (MBPT) by combining the G0W0 method and the Bethe-Salpeter Equation (BSE). A modified version of the HSE06 screened hybrid functional, that includes 20% exact Fock exchange (HSE06(20)) as opposed to 25% in the standard HSE06 functional, is used to set up the starting Hamiltonian for G0W0+BSE calculations. The HSE06(20) functional accurately predicts the ground state electronic band structure. BSE calculations based on data from G0W0+HSE06(20) yield direct optical excitation energies and oscillator strengths in excellent agreement with existing experiments and theoretical calculations characterizing direct excitation. In particular, an exciton binding energy of 229 +- 10 meV is obtained, in close agreement with experiments. The projections of excitonic states onto the quasiparticle band structure in a fatband representation shows that the lowest optical transition of anatase TiO2 consists of excitons originating from the mixing of direct transitions within band pairs running parallel to the Gamma-Z direction in the tetragonal Brillouin zone. This implies a strong spatial localization of excitons in the xy plane of the lattice. This investigation highlights the importance of a suitable non-interacting Hamiltonian for the MBPT based quasiparticle G0W0 and subsequent BSE calculations and suggests HSE06(20) as an optimal choice in the case of anatase TiO2.