Publications Repository - Helmholtz-Zentrum Dresden-Rossendorf

The gas flow modulation technique is a recently proposed approach for measuring the axial gas dispersion coefficient in bubble columns. This study presents a quantitative analysis of the experimental uncertainty associated with gamma-ray densitometry and ensemble-averaging of the data. The considered uncertainty sources are the statistics of the photon counting process, a mismatch between the modelled and the real radiation propagation due to the spatial extent of the detector, and a potential mismatch between modulation and sampling frequencies. The analysis is based on a numerical gamma-ray propagation model and a Monte Carlo approach to account for statistical uncertainty. The proposed algorithm supports the selection of an optimal total scanning time based on detector size, modulation parameters, involved fluids and column and source parameters. The analysis reveals that a mismatch between the modulation and sampling frequencies is most critical while the impact of the other considered uncertainty sources is rather marginal.

Mine waste is a large waste stream and typically contains significant amounts of metal(loid)s, which can pose environmental risks, especially when poorly managed. Reprocessing of mine waste can offer both economic and environmental benefits by contributing to the ever-growing global demands for valuable metals, as well as reducing the environmental risks associated with mine waste. Bioleaching is a global biotechnology that exploits the abilities of some microorganisms to catalyze the oxidative dissolution of sulfidic minerals, thereby expediting the extraction of metal(loid)s. Chemolithoautotrophic acidophilic microorganisms have been the focus of bioleaching studies for many decades and can effectively catalyze the solubilization of metals from ores or waste materials. However, bioleaching with acidophilic organisms is performed at low pH (pH ≤ 2), which could lead to the acidification of the environment. In addition, the tolerance of many acidophilic microorganisms to high chloride concentrations is limited, therefore freshwater is mainly used. There is a growing interest in the use of seawater for leaching purposes, especially in regions with less access to fresh water. Hence, this study investigated the bioleaching potentials of four halophilic (marine), moderately-halophilic sulfur-oxidizing bacteria: Thiomicrospira cyclica, Thiohalobacter thiocyanaticus, Thioclava electrotropha and Thioclava pacifica in shake flasks at room temperature. Results revealed T. electrotropha and T. pacifica as the most promising for bioleaching. In comparison to an acidophilic consortium which leached 95% Co, 0% Pb, 85% Zn, 80% As, 100% Cd, and 55% Mn from a sulfidic mine waste rock sample from the Neves Corvo mine Portugal, a pure cultures of T. electrotropha and T. pacifica solubilized 30-40% Co, and 10-20% Cu, Zn, K, Cd, Mn and Ag at a higher pH (pH ≥ 4) and high chloride concentration. Though still requiring process optimization, this new biotechnology seems promising and offers remarkable benefits such as preventing extreme acidification of the environment while also being applicable in seawater.

For many years, research on microbial-dissolution of metals from ores or waste materials mainly focused on the study of acidophilic organisms. However, most acidophilic bioleaching microorganisms have limited tolerance to high chloride concentrations, thereby requiring fresh water for bioleaching operations. There is a growing interest in the use of seawater for leaching purposes, especially in regions with less access to fresh water. Consequently, there is a need to find halophilic organisms with bioleaching potentials. This study investigated the bioleaching potentials of four moderately halophilic sulfur-oxidising bacteria: Thiomicrospira cyclica, Thiohalobacter thiocyanaticus, Thioclava electrotropha and Thioclava pacifica. Results revealed T. electrotropha and T. pacifica as the most promising for bioleaching. Pure cultures of the two Thioclava strains liberated about 30% Co, and between 8-17% Cu, Pb, Zn, K, Cd, and Mn from a mine waste rock sample from the Neves Corvo mine, Portugal. Microwave roasting of the waste rock at 400 and 500 °C improved the bioleaching efficiency of T. electrotropha for Pb (13.7 to 45.7%), Ag (5.3 to 36%) and In (0 to 27.4%). Mineralogical analysis of the bioleached residues using SEM/MLA-GXMAP showed no major difference in the mineral compositions before and after bioleaching by the Thioclava spp. Generally, the bioleaching rates of the Thioclava spp. are quite low compared to that of the conventional acidophilic bioleaching bacteria. Nevertheless, their ability to liberate potential pollutants (metal(loid)s) into solution from mine waste raises environmental concerns. This is due to their relevance in the biogeochemistry of mine waste dumps, as similar neutrophile halophilic sulfur-oxidising organisms (e.g. Halothiobacillus spp.) have been isolated from mine wastes. On the other hand, the use of competent halophilic microorganisms could be the future of bioleaching due to their high tolerance to Cl- ions and their potential to catalyse mineral dissolution in seawater media, instead of fresh water.

Recent advances in Machine Learning (ML) and Deep Learning (DL) are revolutionizing our abilities to analyze biomedical images and deepen our understanding of infection and disease. Among other host-pathogen interactions may be readily deciphered from microscopy data using convolutional neural networks (CNN). ML/DL algorithms may allow unambiguous scoring of virus-infected and uninfected cells in absence of specific labeling. Furthermore, accompanied by interpretability approaches, the ability of CNNs to learn representations, without explicit feature engineering, may allow uncovering yet unknown phenotypes in microscopy. One such example is our recent tandem segmentation-classification algorithm aimed to uncover morphological markers of Caenorhabditis elegans lifespan and motility. Taken together these novel approaches may facilitate novel discoveries in Infection and Disease Biology.

Coinciding with the global pandemic of SARS-CoV-2 and the resulting global public health crisis caused by COVID-19, artificial intelligence methods started playing an ever more important role in Infectious Medicine. On one hand this was a result of a continuous digital transformation of Infectious Medicine—a trend started decades ago. On the other hand, the pandemic catalyzed the adoption of artificial intelligence and other digital and quantitative techniques by Infectious Medicine. In this chapter we review recent works touching upon aspects of COVID-19 patient journey and how it interconnects with big data and artificial intelligence. These include early and clinical research, epidemiology and detection, diagnostics, clinical care and decision support, as well as long-term care and prevention. We cross-compare the published works and assess their maturity. Finally, we provide a conclusion on the state of artificial intelligence in the Infectious Medicine of COVID-19 and attempt a future perspective.

Particle-based process models offer a promising avenue towards greater predictability in geometallurgy, i.e., the ability to predict the outcomes of specific mineral processing routes from the mineralogical and microstructural ore characteristics. While the particle-based prediction of separation processes is already possible with acceptable levels of accuracy, the ability to predict the outcomes of comminution processes is currently limited to particle size distributions. Expanding comminution modelling tools to include particle microstructures would enable the full particle-based modelling of mineral processing flowsheets. As a step towards the inclusion of microstructure in comminution modelling, Laguerre tessellations are proposed to represent both the microstructure and the successive comminution steps. In contrast to the PARGEN library of simulated particles, our goal is to provide a low-parametric, dynamic, and efficient generator of parent and progeny material to inform forward and backward modelling efforts.

The idea is to follow a hierarchical decision structure in the simulation procedure. We first define an intensity field in 3D for the occurrence cell nuclei, which are then realised by a marked Poisson process. The first mark corresponds to realisations of a multinomial variable, and defines the mineral of each potential cell. Conditional on the mineral, the second mark follows a normal distribution, defining the weight of each cell, related to its size. A communition step is defined by a Voronoi mosaic, with a (t+1)-step exhibiting a higher intensity of its Poisson process than the previous t-step. To model preferential breakage, we inhibit some of the potential breakage surfaces with a probability depending on the weighted average hardness and the cleavage quality of the minerals that each surface cuts. Two consecutive comminution steps generate the corresponding parent and progeny particles, Each independently cut by a random plane to generate the equivalent of a 2D SEM-based automated mineralogy dataset.

Cross-validation and performance measures are standard components in the evaluation of a geostatistical model. These are well established in the univariate case, but measures for multivariate geostatistical modeling have not received as much attention. In the case of a single target variable, the univariate approaches remain valid, but in the fully multivariate case where a vector of variables needs to be estimated the evaluation needs to be based on all estimates simultaneously. An extension of cross-validation and associated performance measures to the fully multivariate case is presented and discussed for the case of regionalized compositions. The method is demonstrated by validating geostatistical models for two case studies: a sample drawn from a geochemical survey data set estimated with cokriging, and an application of direct sampling multiple point simulation.

Three-dimensional information is crucial to our understanding of biological phenomena. The vast majority of biological microscopy specimens are inherently three-dimensional. However, conventional light microscopy is largely geared towards 2D images, while 3D microscopy and image reconstruction remain feasible only with specialized equipment and techniques. Inspired by the working principles of one such technique - confocal microscopy, we propose a novel approach to 3D widefield microscopy reconstruction through semantic segmentation of in-focus and out-of-focus pixels. For this, we explore a number of rule-based algorithms commonly used for software-based autofocusing and apply them to a dataset of widefield focal stacks. We propose a computation scheme allowing the calculation of lateral focus score maps of the slices of each stack using these algorithms. Furthermore, we identify algorithms preferable for obtaining such maps. Finally, to ensure the practicality of our approach, we propose a surrogate model based on a deep neural network, capable of segmenting in-focus pixels from the out-of-focus background in a fast and reliable fashion. The deep-neural-network-based approach allows a major speedup for data processing making it usable for online data processing.

Understanding the 3D structure of biological entities is crucial for gaining mechanistic biomedical knowledge. A confocal light microscope is a well-established tool used to obtain 3D data from biological specimens. Yet, it comes with the drawbacks of high equipment prices and heavy human labor. In this project, we introduce a 3D focal stacking solution using deep neural networks (DNN). Instead of restoring 3D models from confocal microscopes, our model produces in-focus images by inputting widefield microscope images, which may be obtained with significantly simpler equipment. This enables the translation from widefield microscope images into the 3D model by segmenting the in-focus pixels, allowing the image of 3D biological specimens in vivo.

Two versions of a machine learning (ML1 & ML2) based modeling framework have been successfully used to provide operational forecasts of O3 at Kennewick, WA. This paper shows the ML system performance when applied to all available observation locations in the Pacific Northwest to predict O3 and PM2.5 concentrations. We used historical O3 and PM2.5 concentrations, Weather Research and Forecasting (WRF) meteorological forecast data (including temperature, surface pressure, relative humidity, wind speed, wind direction, and planetary boundary layer height) and time information (including hour, weekday, and month) to train the model. A 10-time, 10-fold cross-validation method was used to evaluate the model performance. Similar to our previous study, ML1 correctly captures more high-O3 events, but also generates more false alarms, and ML2 performs better overall (R2 = 0.79), especially for low-O3 events. Our ML modeling framework utilizes both ML1 and ML2 results to achieve the best forecast performance. Compared to the WRF-CMAQ based forecast (i.e., AIRPACT), our final ML forecasts reduce the normalized mean bias (NMB) from 7.6% to 2.6% when evaluating against the observed mixing ratios. Our ML-based forecasts also show clear improvements on Air Quality Index (AQI) forecasts; more accurate O3 AQI index predictions for each AQI index including high-O3 AQI events. For PM2.5, ML1 and ML2 demonstrate similar capabilities to predict high-PM2.5 events and ML2 keeps its accuracy for low-PM2.5 predictions, so ML2 is used to provide the final forecast values, instead of combining the two ML models that we are using for O3. During wildfire seasons (May to September) and cold, winter seasons (November to February) from 2017 to 2020, our ML model clearly performs better than AIRPACT. AIRPACT under-predicts the wildfire season PM2.5 concentrations in the PNW (NMB = -27%) and over-predicts at some sites in the cold season up to 200%, while ML2 has a lower NMB in both seasons (NMB = 7.9% in the wildfire season and 2.2% in the cold season) and correctly captures more high-PM2.5 events.

The size and shape of the primary dendrite tips determine the principal length scale of the microstructure evolving during solidification of alloys. In-situ X-ray measurements of the tip shape in metals have been unsuccessful so far due to insufficient spatial resolution or high image noise. To overcome these limitations, high-resolution synchrotron radiography and advanced image processing techniques are applied to a thin sample of a solidifying Ga-35wt.%In alloy. Quantitative in-situ measurements are performed of the growth of dendrite tips during the fast initial transient and the subsequent steady growth period, with tip velocities ranging over almost two orders of magnitude. The value of the dendrite tip shape selection parameter is found to be σ^*=0.0768, which suggests an interface energy anisotropy of ε_4=0.015 for the present Ga-In alloy. The non-axisymmetric dendrite tip shape amplitude coefficient is measured to be A_4≈0.004, which is in excellent agreement with the universal value previously established for dendrites.

Although uranium-cerium dioxides are frequently used as surrogate material for (U,Pu)O2-δ, there is currently no reliable data regarding the oxygen stoichiometry and the redox speciation of the cations in such samples. In order to fill this gap, this manuscript proposes a synchrotron study of highly homogeneous (U,Ce)O2±δ sintered samples prepared by wet-chemistry route. HERFD-XANES spectroscopy led to determine accurately the O/M ratios (with M = U + Ce). Under reducing atmosphere (PO2 ~ 610-29 atm), the oxides were found close to O/M = 2.00 while the O/M ratio varied with the sintering conditions under argon (PO2 ~ 210-6 atm). They globally appear to be hyper-stoichiometric (i.e. O/M > 2.00), the departure from the dioxide stoichiometry decreasing with both the cerium content in the sample, and the sintering temperature. Nevertheless, such deviation from the ideal O/M = 2.00 ratio was found to generate only moderate structural disorder from EXAFS data at the U-L3 edge. Indeed, all the samples retained the fluorite-type structure of the UO2 and CeO2 parent compounds. The determination of accurate lattice parameters thanks to SPXRD measurements led to complete the data already reported in the literature, and to propose a mathematic expression linking the unit cell parameter, the chemical composition and the deviation from the stoichiometry. Such relation can now be used as a first approximation to estimate the O/M stoichiometry of uranium-cerium mixed oxides on a large composition range.

The growing use of geographic information systems (GIS) and geographical analyses in different areas of the digital humanities highlights the need for geocoding, i.e. assigning geographic coordinates to records in a dataset. Such spatially-referenced datasets are a precondition for any spatial analysis and visualization. While GIS in general is a dynamically evolving branch of software development, there is a need for specialized applications which would assist researchers in geocoding datasets in history, archaeology, and the digital humanities. Therefore, we developed the “Historical Geocoding Assistant”, an open-source web tool that meets the specific needs of historical research and brings a solution to geocoding historical data in a convenient, fast, and reliable way.

The properties of rotating turbulence driven by precession are studied using direct numerical simulations and analysis of the underlying dynamical processes in Fourier space. The study is carried out in the local rotating coordinate frame, where precession gives rise to a background shear flow, which becomes linearly unstable and breaks down into turbulence. We observe that this precession-driven turbulence is in general characterized by coexisting two dimensional (2D) columnar vortices and three dimensional (3D) inertial waves, whose relative energies depend on the precession parameter Po. The vortices resemble the typical condensates of geostrophic turbulence, are aligned along the rotation axis (with zero wavenumber in this direction, kz = 0) and are fed by the 3D waves through nonlinear transfer of energy, while the waves (with kz ≠ 0) in turn are directly fed by the precessional instability of the background flow. The vortices themselves undergo inverse cascade of energy and exhibit anisotropy in Fourier space. For small Po < 0.1 and sufficiently high Reynolds numbers, the typical regime for most geo-and astrophysical applications, the flow exhibits strongly oscillatory (bursty) evolution due to the alternation of vortices and small-scale waves. On the other hand, at larger Po > 0.1 turbulence is quasi-steady with only mild fluctuations, the coexisting columnar vortices and waves in this state give rise to a split (simultaneous inverse and forward) cascade. Increasing the precession magnitude causes a reinforcement of waves relative to vortices with the energy spectra approaching the Kolmogorov scaling and, therefore, the precession mechanism counteracts the effects of the rotation.

The recomine-concept deals with a variable flowsheet that comprises three major modules: (I) processing of tailings, (II) processing of solid residues and (III) water recovery. The first module deals with the output of BHP’s processing route. A bulk sulfide flotation produces sulfide concentrates and a silicate residue. Two potential processes represent the concentrate’s subsequent treatment: (a) leaching and (b) roasting of the concentrate.
The major aim of treating the concentrate by leaching is to produce schwertmannite (SMT) as a product after precipitation. Alternatively, roasting the sulphide concentrate may result in other economic products: sulphuric acid and ferric sulphate. Silicate residues from module one will appear in a sequence of processing steps in module two. This module has a twofold aim: (1) separating ferro- and paramagnetic fractions and (2) dewatering the residues. Distinct steps will accomplish a separation of concentrated, iron-rich garnet. Novel membrane technologies will treat wastewater streams from modules one and two and represent module three. The major purpose of water treatment is to maximize the amount of clean water for nearby and downstream communities.
The flotation campaign applied an overall flotation time of 20 min. Tailings contain only 0.4 % of pyrite at a recovery of 6.3 %. Bioleaching experiments were able to turnover 75 % of the pyrite maximum. SMT precipitation was successful. Infrared spectroscopy and XRD proofed a pure and fine crystalline schwertmannite. Its’ ability to adsorb arsenates has been proven in according tests and schwertmannite from Antamina Tailings is potentially suitable for decontamination of mine water from arsenates, phosphates, and vanadates. Interest from chemical industry exist in designing schwertmannite as a pigment in colour chemistry (see section C in appendix). A crude product of garnet sand was separated magnetically. The recomine-team speculates that a potential annual production of crude garnet could provide the largest deposit for industrial garnet on the planet.

BHP with the support of Fundacion Chile, through its open innovation program EXPANDE, has launched “BHP Tailings Challenge”, an initiative that seeks to promote the development of innovative solutions for repurposing copper tailings. The BHP Tailings Challenge is a global competition aimed at identifying the most innovative companies, startups, consortia, research centers and universities to help transform fresh tailings and create innovative business models. The recomine-proposal, which was one of the 10 selected proposals out of 153, suggests a variable flowsheet that comprises three major modules: (I) processing of tailings, (II) processing of solid residues and (III) water recovery. The following paragraphs describe the modules individually but outline the material flow from one module to the next. The first module deals with the output of BHP’s processing route. The recomine-proposal starts with a bulk sulfide flotation that produces a sulfide concentrate and a silicate residue as an output. Two potential processes represent the concentrate’s subsequent treatment: (a) leaching and (b) roasting of the concentrate. The major aim of treating the concentrate by leaching is to produce schwertmannite as a product after precipitation. Alternatively, roasting the sulfide concentrate may result in other economic products: sulfuric acid, ferric sulfate and residual calcine. The exothermal roasting may furthermore provide heat emissions as an energy source for usage in BHP’s processing routine. The silicate residue from module one will appear in a sequence of processing steps in module two. This module has a twofold aim: (1) separating ferro- and paramagnetic fractions and (2) dewatering the residues. Several steps will accomplish a separation of concentrated, iron-rich garnet. Novel membrane technologies will treat wastewater streams from modules one and two and represent the key-technology in module three. The major purpose of water treatment is to maximize the amount of clean water for nearby and downstream communities.

The recomine alliance works on several concepts for remediation of mining waste. With 5 different fiel laboratories they work on remediation technologies for mining legacies. However, these technologies work as well in active mining and may improve current mining activities. The constant interaction of companies from the recomine alliance with leading research institutions allowed to upscale several technologies for remediating mining waste and gaining raw materials at the same time. The talk provides an overview of the recomine activities and their overarching fourfold strategy in significantly reducing the volume of mining waste by (1) analyzing, avoiding (2), re-mine, (3) and (4) transforming mining waste.

BHP with the support of Fundacion Chile, through its open innovation program EXPANDE, has launched “BHP Tailings Challenge”, an initiative that seeks to promote the development of innovative solutions for repurposing copper tailings. The BHP Tailings Challenge is a global competition aimed at identifying the most innovative companies, startups, consortia, research centers and universities to help transform fresh tailings and create innovative business models. The recomine-proposal, which was one of the 10 selected proposals out of 153, suggests a variable flowsheet that comprises three major modules: (I) processing of tailings, (II) processing of solid residues and (III) water recovery. The following paragraphs describe the modules individually but outline the material flow from one module to the next. The first module deals with the output of BHP’s processing route. The recomine-proposal starts with a bulk sulfide flotation that produces a sulfide concentrate and a silicate residue as an output. Two potential processes represent the concentrate’s subsequent treatment: (a) leaching and (b) roasting of the concentrate. The major aim of treating the concentrate by leaching is to produce schwertmannite as a product after precipitation. Alternatively, roasting the sulfide concentrate may result in other economic products: sulfuric acid, ferric sulfate and residual calcine. The exothermal roasting may furthermore provide heat emissions as an energy source for usage in BHP’s processing routine. The silicate residue from module one will appear in a sequence of processing steps in module two. This module has a twofold aim: (1) separating ferro- and paramagnetic fractions and (2) dewatering the residues. Several steps will accomplish a separation of concentrated, iron-rich garnet. Novel membrane technologies will treat wastewater streams from modules one and two and represent the key-technology in module three. The major purpose of water treatment is to maximize the amount of clean water for nearby and downstream communities.

The recent waves of COVID-19 highlighted the importance of understanding and quantifying spatiotemporal interactions to infer, model, and predict disease spread in real time. In this demonstration paper, we present a robust infrastructure for interactive exploration of interregional and international spatiotemporal interactions via time-lagged correlations of increases in COVID-19 incidence. This infrastructure consists of: (i) an operational data store (ODS) coupled with automated scripts for downloading, cleaning, and processing data from heterogeneous sources; (ii) a server application handling on-demand analyses of the database data through a RESTful API; and (iii) a web application providing the interactive dashboard to explore various correlation and geostatistical metrics of the integrated data in spacetime. The environment allows users to study focal spatiotemporal trends and the potential of regions to export and import the virus. Moreover, the application has the potential to reveal the effect of the national border to mitigate the interaction, particularly the spread of the virus. The infrastructure serves COVID-19 data from Germany, Poland, and Czechia, with the possibility of extension to other regions and topics. The dashboard is under active development and accessible on www.where2test.de/correlation.

In this paper, we present Computer-Assisted Semantic Text Modelling (CASTEMO), a novel approach to transformation of textual resources into deeply structured data stored in JSON-based document databases. We also present the InkVisitor application which assists this data collection workflow and helps validate the data. Both the workflow and the application were developed within the Dissident Networks Project (DISSINET, https://dissinet.cz).

CASTEMO is based on widespread ideas, such as the idea of semantic data (e.g. Semantic Web) and the syntactic structure of natural language sentences (in our case, subject-verb-object1-object2 quadruples), and we acknowledge convergent developments (mainly Roberto Franzosi’s Quantitative Narrative Analysis). Nevertheless, we follow our own path towards deeply structured and deeply semantic data drawn from texts which allow us to preserve, and thus quantitatively analyse, e.g.:

• the order and syntactic embeddedness of information;

• the textual embeddedness of information (i.e. who is speaking, to whom, and in what context);

• the original language, expression, and discourse;

• the distinction between epistemic levels.

CASTEMO thus offers a time-intensive but extremely powerful alternative to (1) text mining, which often fails to answer fine-grained questions, and (2) Computer-Assisted Qualitative Data Analysis Software (CAQDAS), where opportunities of quantification are too incidental and severely limited by the original hypothesis. CASTEMO should be of interest to projects interested in quantitatively analysing information strictly in the context of its production (“source criticism 2.0”), and looking at the discourse of texts.

In this paper, we present the foundations of this data collection workflow, its selling points, as well as caveats for potential users. We also provide a first public presentation of InkVisitor, an open-source browser-based application implementing the CASTEMO workflow.

Ziel der Masterarbeit soll es sein, eine Vorstufe (Präkursor) eines Radiotracers zu synthetisieren, welcher es ermöglichen soll, ^197(m)Hg zu binden und so als metallorganisches Radiopharmakon vorzuliegen. Diese Substanz soll letztlich dazu dienen, Gammastrahlung für die Diagnostik und Konversionselektronen für die Therapie auszusenden, sodass der fertige Radiotracer für theranostische Zwecke Anwendung finden kann. Dabei soll als Grundkörper ein Bispidin dienen, wobei die bisher verwendeten Phenylgruppen an Position C-1 und C-5 durch Methylgruppen ausgetauscht werden sollen, um die Lipophilie der Verbindung zu senken.
Dazu sollen mittels nucleophiler Substitution zwei neue Funktionalitäten an die sekundären Amine angebracht werden, mit denen es letztlich ermöglicht wird, eine Di-Aryl-quecksilberverbindung auszubilden. Die Synthese und Untersuchung unterschiedlicher Seitenarme sollen dabei ebenfalls genauer betrachtet werden.
Ein weiter Punkt soll sein, die C-9-Position am Bispidingerüst zu modifizieren, um ein entsprechendes Vektormolekül anbringen zu können. Dadurch wird ermöglicht, dass sich der resultierende Radiotracer selektiv an Zellen anlagert. Dabei kann das Vektormolekül auch vor der Anbringung der Seitenarme an der C-9-Position des Bispidins gebunden werden.

The cause of the rapid and geographically impressive spread of Mithraism in the Roman empire is still only partially understood. Scholars had speculated about the influence of the Roman army and the popularity of Mithraism among Roman soldiers. However, a meticulously conducted demographical study of Mithraic epigraphy problematized this view. To demonstrate the possible influence of the Roman military infrastructure on the spread of Mithraism in the Roman empire, we coded all sites of documented Mithraic presence and locations of the major Roman legionary fortresses, positioned them on the transportation network and used statistical analysis to detect a possible relationship between these datasets, both at the level of the whole Roman empire and regionally. Although we did not find, at the level of the Roman empire, a statistically significant overlap between the locations of the Roman legionary fortresses and Mithraic sites, we discovered the statistically significant presence of Mithraic evidence in nodes important on thresholded military subnetworks connecting the Roman legionary fortresses. These results support the view that the Roman army infrastructure contributed to the spread of Mithraism and can partially explain the geographical distribution of archaeologically attested Mithraic evidence in the Roman Empire but cannot be seen as a single factor playing a role in the transmission of Mithraism.

Isonicotinic acid (INA), as a prototypical N, O-donor bifunctional ligand, has demonstrated its ability to differentiate Th4+ from representative ions for products in spent nuclear fuels (Cs+, Ba2+, Mn2+, Fe2+, Fe3+, Co2+, Ni2+, Cu2+, Pd2+, ReO4-, La3+, Ce3+, Ce4+, UO22+), yielding an actinide metal-organic framework (An-MOF), Th-INA-1, by selective crystallization. This unprecedented motif with the highest ligand-binding number (i.e., 16) shows promise as a primary waste form due to its structural integrity, especially upon irradiation up to 6 MGy γ-or β-irradiation.

This paper presents an approach to the collection of structured relational data which we have developed over the last 2.5 years in the Dissident Networks Project (DISSINET, https://www.dissinet.cz/). Our goal has been to devise a data model and environment capable of capturing the detail of inquisitorial records: the persons, groups, events, attitudes and physical objects they describe, the reported social, spatial and temporal relations between them, but also the modality of speech (negation, question, possibility etc.), the chain of information flow in inquisitorial records (e.g. who is reporting what and when, who is inculpating whom), and the different modes of trial interaction and recording. We thus preserve the semantic structure and detail of our sources. The data thus collected then allows us to analyze the social, spatial, and discursive patterns of inquisitorial records, heresy trials, and medieval religious dissent using a variety of computational and quantitative methods, such as social and spatial network analysis, geographic information science, and natural language processing. In addition, our data model and the experience gained from devising it will be of interest even beyond heresy and inquisition research, above all to historians keen to explore the possibilities of analysis of structured data while preserving the detail and the discursive patterns of their sources.

Small actinium-225-labeled prostate-specific membrane antigen (PSMA)-targeted radioconjugates have been described for targeted alpha therapy of metastatic castration-resistant prostate cancer. Transient binding to serum albumin as a highly abundant, inherent transport protein represents a commonly applied strategy to modulate the tissue distribution profile of such low-molecular-weight radiotherapeutics and to enhance radioactivity uptake into tumor lesions with the ultimate objective of improved therapeutic outcome. Two ligands mcp-M-alb-PSMA and mcp-D-alb-PSMA were synthesized by combining a macropa-derived chelator with either one or two lysine-ureido-glutamate-based PSMA- and 4-(p-iodophenyl)butyrate albumin-binding entities using multistep peptide-coupling chemistry. Both compounds were labeled with [²²⁵Ac]Ac³⁺ under mild conditions and their reversible binding to serum albumin was analyzed by an ultrafiltration assay as well as microscale thermophoresis measurements. Saturation binding studies and clonogenic survival assays using PSMA-expressing LNCaP cells were performed to evaluate PSMA-mediated cell binding and to assess the cytotoxic potency of the novel radioconjugates [²²⁵Ac]Ac-mcp-M-alb-PSMA and [²²⁵Ac]Ac-mcp-D-alb-PSMA, respectively. Biodistributions of both 225Ac-radioconjugates were investigated using LNCaP tumor-bearing SCID mice.

The coupled DYN3D/ATHLET code system was recently adapted for Sodium cooled Fast Reactors (SFRs) applications. The main objective of this study is to validate further the DYN3D/ATHLET code system by performing a coupled 3D neutron kinetics/thermal-hydraulics analysis of six transient start-up tests conducted at the French Superphenix (SPX) reactor. The tests were a part of the startup test program aiming at evaluation of the core reactivity feedback characteristics. Peculiarity of these transients is the necessity of accounting for the thermal expansions of the primary system structural elements influencing the position of control rods in the core.
The paper includes a brief summary on the benchmark specification, description of the neutronics and thermal-hydraulics models, and comparison of the simulation results to the available experimental data. For all six transients, a good agreement between simulations and experiments was observed confirming a reasonable performance of DYN3D/ATHLET.

Adoptive therapy using CAR-T cells as "living drugs" is an emerging field in cancer immunotherapy. Our group particularly focusses on the development and clinical translation of novel adaptor CAR-T platforms (UniCAR & RevCAR), in which CAR T-cell activity is controlled by tumor-specific adaptor molecules. Apart from their immunotherapeutic potential demonstrated by numerous preclinical and an initial clinical proof-of-concept study, our adaptor CAR technologies represent also an ideal starting point for the development of cancer theranostics. Preclinical studies have shown that adaptor molecules can be easily radiolabeled for tumor therapy and diagnostics. Conversely, clinically used PET tracers have been successfully converted into highly efficient adaptor molecules for CAR T-cell immunotherapy. Overall, adaptor molecules are versatile tools for (i) CAR-T cell therapy, (ii) noninvasive diagnostic imaging, and (iii) targeted radioimmunotherapy, underscoring that combinatorial theranostic adaptor CAR-T approaches may open new avenues for effective cancer therapy.

Modification of autologous T cells with tumor-specific chimeric antigen receptors (CARs) has been shown to be an effective tool for treatment of hematologic malignancies. However, clinical translation towards solid tumors is still limited as demonstrated by the suboptimal performance of CAR T cells in first in-human studies. Combining advances in CAR-T cell therapy and cancer theranostics may provide a promising multimodal approach to increase therapeutic efficacy and achieve durable responses in cancer patients.
Here we present the development and functional characterization of different peptide- or antibody constructs targeting tumor-associated antigens or different structures of the tumor microenvironment. Equipment of these molecules with the E5B9 peptide epitope, enabled their use in the well-established UniCAR system. Accordingly, the constructs were able to specifically activate UniCAR T cells for efficient killing of both hematological and solid tumors in vitro and in vivo. Upon conjugation of chelators like DOTAGA and NODAGA, the peptide- or antibody derivates were further successfully applied for cancer theranostics, using e.g. 68Ga, 64Cu as diagnostic radionuclides and e.g. 67Cu, 225Ac as therapeutic radionuclides. PET imaging in xenotransplanted mice has demonstrated high contrast tumor accumulation. Consistent with the stable tumor accumulation, 225Ac-labeled molecules demonstrated therapeutic efficacy in a xenograft mouse model.
In summary, E5B9-tagged peptide- or antibody constructs open a new era of cancer immunotheranostics as they can be used multimodally for (i) UniCAR-T cell therapy, (ii) non-invasive diagnostic imaging and (iii) targeted radioimmunotherapy. Prospectively, they could thus bridge the gap between the fields of CAR-T cells and cancer theranostics.

CAR T-cell therapy achieved unparalleled clinical success rates for treatment of patients with hematological
malignancies. However, progress and clinical translation towards solid tumors is slow and hampered by many
factors e.g. increased complexity, high heterogeneity and an immunosuppressive tumor microenvironment.
Thus, CAR T-cell therapy alone might not result in durable antitumor responses. Combinatorial approaches are
promising strategies to improve CAR T-cell efficacy in solid tumor treatment. In this regard, we here aim to
combine conventional cancer theranostics with CAR T-cell immunotherapy in one single approach.

By using the well-established UniCAR system, a novel, multifunctional tool termed PSCA-IgG4 target module
(TM) was developed for dual prostate cancer theranostics. It comprises a human PSCA-specific binding
domain, the hinge and Fc-domain of human IgG4 molecules as well as the UniCAR epitope E5B9. As shown by
in vitro assays with PSCA-positive and PSCA-negative prostate cancer cells, the novel TM redirected UniCAR T
cells for efficient tumor cell lysis in a strictly antigen- and TM-dependent manner. After radiolabeling with
copper-64 or actinium-225, the novel PSCA-IgG4 TM was successfully applied for diagnostic imaging and
targeted radioimmunotherapy. The 64Cu-labeled PSCA-IgG4 TM showed maximal tumor accumulation with
optimal tumor-to-background ratios after 1.5 days. Furthermore, targeted alpha-therapy with the 225Aclabeled
TM significantly delayed the outgrowth of established tumors in mice.

In summary, the here presented, novel PSCA-IgG4 TM is a promising candidate for dual theranostics of
prostate cancer that may help to overcome present hurdles in solid tumor therapy. After radiolabeling it
facilitates not only targeted alpha-therapy and diagnostic imaging of PSCA, but can be also repurposed as a
TM for UniCAR T-cell immunotherapy.

Hyperspectral imaging is an important technology in the field of geosciences and remote sensing. However, the high-dimensional nature of hyperspectral images (HSIs) together with the limited availability of training/labeled samples challenge an efficient processing of HSIs. To alleviate these challenges, we propose a deep multi-resolution clustering network (DMC-Net) to analyze HSIs. DMC-Net, without requiring training/labeled samples for the training process, captures the non-linear intrinsic relation within data points in an HSI and analyzes the image at various resolutions by applying atrous convolutions. Furthermore, DMC-Net preserves the spectral information by directly incorporating extracted features from the original HSI into the reconstruction phase. In terms of clustering accuracy, experimental results on two real HSIs demonstrate the superior performance of DMC-Net compared to the state-of-the-art deep learning-based clustering approaches.

In the era of deep learning, annotated datasets have become a crucial asset to the remote sensing community. In the last decade, a plethora of different datasets was published, each designed for a specific data type and with a specific task or application in mind. In the jungle of remote sensing datasets, it can be hard to keep track of what is available already. With this paper, we introduce EOD - the IEEE GRSS Earth Observation Database (EOD) - an interactive online platform for cataloguing different types of datasets leveraging remote sensing imagery.

Antiferromagnetic insulators are a prospective material science platform for magnonics, spin superfluidity, THz spintronics and non-volatile data storage. Due to linear magnetoelectric effect at room temperature, Cr2O3 is a convenient playground to test fundamental predictions and new device ideas. In this presentation, we will cover our activities on physics and applications of Cr2O3. In particular, we provide insight into the nanoscale mechanics of antiferromagnetic domain walls in single crystals of Cr2O3 [1]. Furthermore, we will discuss the family of flexomagnetic effects in Cr2O3 thin films [2]. It is demonstrated that in addition to the conventional magnetic moment induced by the strain gradient, there is another flexomagnetic effect which impacts the magnetic phase transition resulting in the distribution of the Néel temperature along the film thickness. The details on the study of the defect nanostructure in Cr2O3 thin films [3,4] and bulks [5] and their impact on magnetism and magnetoelectricity of Cr2O3 will be discussed as well. We identified spin Hall magnetoresistance as a possible mechanism to explain all-electric readout of the magnetic state of Cr2O3 interfaced with a heavy metal like Pt [6,7]. The possibility to read-out the antiferromagnetic order parameter of magnetoelectric Cr2O3 all-electrically enabled realisation of antiferromagnetic magnetoelectric random access memory (AF-MERAM) [8].

[1] N. Hedrich et al., “Nanoscale mechanics of antiferromagnetic domain walls”. Nature Physics 17, 574 (2021).
[2] P. Makushko et al., “Flexomagnetism and vertically graded Néel temperature of antiferromagnetic Cr2O3 thin films”. Nature Comm. 13, 6745 (2022).
[3] I. Veremchuk et al., “Defect Nanostructure and its Impact on Magnetism of α-Cr2O3 Thin Films”. Small 18, 2201228 (2022).
[4] P. Appel et al., “Nanomagnetism of Magnetoelectric Granular Thin-Film Antiferromagnets”. Nano Lett. 19, 1682 (2019).
[5] I. Veremchuk et al., “Magnetism and Magnetoelectricity of Textured Polycrystalline Bulk Cr2O3 Sintered in Conditions Far out of Equilibrium”. ACS Appl. Electron. Mater. 4, 2943 (2022).
[6] R. Schlitz et al., “Evolution of the spin hall magnetoresistance in Cr2O3/Pt bilayers close to the Neel temperature”. Appl. Phys. Lett. 112, 132401 (2018).
[7] T. Kosub et al., “All-Electric Access to the Magnetic-Field-Invariant Magnetization of Antiferromagnets”. Phys. Rev. Lett. 115, 097201 (2015).
[8] T. Kosub et al., “Purely antiferromagnetic magnetoelectric random access memory”. Nature Comm. 8, 13985 (2017).

There is a wide range of flow phenomena to be expected in liquid metal battery systems. However, most of them have only been investigated numerically or in low-temperature model systems. To understand what actually takes place during operation and which flow phenomena prevail in a realistic liquid metal battery, an investigative study of a relatively new type of molten salt battery, Na||ZnCl2, is planned. This poster will present the objectives, challenges, and current status of a recently started x-ray and neutron imaging campaign of flow phenomena in high-temperature batteries at 600 oC.

Recent developments of heterojunction-based devices through bottom-up approach such as sensors, light-emitters, energy generation and storage have emerged with great interest due to their wide range of operation and application related flexibilities. This work demonstrates how as-grown hydrogen titanate nanotubes (HTNT) bend and wrap on pristine curpous oxide nanowires (CONW) when mixed together. The unique architecture of wrapping followed by junction formation enhances the active surface area and reduces the contact resistance between the adjacent CONW and HTNT. Such a film upon further ion beam irradiation produces a large-scale network of hetero- and homo-junctions. This newly formed thin film surface upon irradiation shows strong water repelling properties and higher electrical conductivity. The wrapping mechanism, bond formation and the change of conductivity are explained using first principles calculations. The ion beam modifications and large-scale joining are predicted by state-of-the-art TRI3DYN simulation, which is based on binary collision approximation and simulated in a Monte Carlo approach. The observed wrapping and heterojunction are expected to provide excellent mechanical strength and flexibility, which are suitable for fabrication of flexible electronic devices.

As next generation for CAR T cells, adaptor CAR platforms have been developed, which are designed to improve safety, but at the same time maintain the high efficiency of the CAR T cell approach. In our lab, we developed the UniCAR system, which consists of universal (Uni)CAR T cells and tumor-specific target modules (TMs), which work as bridging molecules between the UniCAR T cells and the target cells. Until now, type 1 and type 2 UniCARs were developed. Both UniCAR types consist of an extracellular binding domain derived from a monoclonal antibody (mAb) directed to the nuclear La/SS-B protein. Type 1 UniCARs are derived from the anti-La mAb 5B9. Type 2 UniCARs from the anti-La mAb 7B6. As both anti-La mAbs are not able to precipitate native La protein, both anti-La mAbs are directed to a specific cryptic epitope which is not accessible on the cell surface. Thus, the UniCAR T cell is per se inert. To activate the UniCAR T cell for tumor cell killing a TM is needed as a second component. Typically, a TM is composed of a tumor-specific binding domain and the respective La epitope.
Consequently, the affinity of the TM towards the target antigen but also towards the UniCAR T cell via the E5B9-tag plays an important role for functionality of the respective UniCAR system.
In this study, we representatively aimed to elucidate if and how the affinity of the type 1 UniCAR domain to the E5B9 epitope impacts the functionality of the UniCAR system. To alter the interaction of UniCAR and TM, we designed different mutated E5B9 peptides (M1-M3) carrying one or two amino acid (aa) changes. In detail, aspartic acid and/or glutamic acid were mutated to glycine residues as they most probably are involved in epitope/paratope interactions. We subsequently fused these mutated peptides to an scFv domain, resulting in three different mutated TM versions.
By conducting ELISA and flow-cytometry based binding studies, we showed that a single aa exchange (D3>G3) in M1 did not alter the affinity towards the mAb 5B9. However, replacing two aa resulted in a 4-fold (M2: E2>G2, D3>G3) or even 50-fold reduced affinity (M3: E2>G2, E6>G6) of the mAb 5B9 towards the mutated E5B9 epitopes. By chromium release assay, we could show that only the TM with M1 was able to induce efficient lysis with EC50 values comparable to the original TM containing the non-mutated E5B9-tag. The TMs with the mutated peptides M2 or M3, showing a lower affinity, were not able to redirect UniCAR T cells for tumor cell killing.
In summary, our data revealed that lowering the affinity between E5B9 peptide and mAb 5B9/ UniCAR T cell by a factor of four already impedes the functionality of the UniCAR system and that affinity of around 0.1 nM is required for proper functionality.

We report on the intrinsic magnetic properties of a GdCo4B single crystal as derived from magnetization measurements. The occurrence of a first-order magnetization process (FOMP) in a magnetic field applied perpendicularly to the easy magnetization direction provides a unique opportunity to determine the anisotropy parameters K1 and K2. To this end, the theoretical approach proposed previously for easy-plane magnets has been adapted for the case of easy-axis compounds exhibiting a FOMP. The obtained anisotropy parameters of GdCo4B are successfully compared with the values deduced by other classical techniques. The presence of a compensation point in the thermal dependence of the spontaneous magnetization has enabled the determination of the exchange field on Gd, Bex = 129 T, which is in good agreement with inelastic neutron scattering results published earlier. Influence of the applied pressure on the anisotropy parameters is quantified using the pressure dependence of the FOMP, revealing a significant sensitivity of the anisotropy parameters to pressure.

The COVID-19 pandemic has entered a new endemic phase with the recent, highly transmissible omicron variant. To navigate this transition, the Where2Test group develops an integrated suite of models, datasets, optimization algorithms, and user-friendly webapps to help users understand and manage SARS-CoV-2 infection risk in a range of different organizational settings. All of these applications are available on www.where2test.de.

The selective inhibition of enzymes that catalyze the conversion of arachidonic acid to inflammatory eicosanoids represents a promising approach for cancer therapy. We, therefore, focus on the incorporation of metabolically stable, sterically demanding and hydrophobic carboranes into existing dual cycloxygenase-2 (COX-2)/5-lipoxygenase (5-LO) inhibitors that are key enzymes in the biosynthesis of eicosanoids. Here, we present the first carborane-containing dual COX-2/5-LO inhibitors derived from RWJ-63556. The replacement of the fluorophenyl moiety by meta- or para-carborane resulted in five carborane-containing derivatives 3, 6, 9, 13 and 17 that show high inhibitory activities toward COX-2 and 5-LO in vitro. Cell viability studies on the A375 melanoma cell line revealed that meta-carborane derivative 3 shows higher anticancer activity compared to RWJ-63556 based on accumulation of lipid droplets in the cells due to blockage of the COX-2 and 5-LO pathways, indicating a promising approach for the design of potent dual COX-2/5-LO inhibitors.

Efforts towards digitalization in the material science and technology community have enhanced in
the last years. In 2019 the German digitalization initiative platform „MaterialDigital“ (MD) has been
started. Numerous projects concerning digitalization, including the copper related project
„KupferDigital“ (copper digital) have been initiated under the umbrella of MD. The initiative strives to
address numerous issues concerning data access, exchange, security, provenance and sovereignty.

Heterogeneous data origin, storage and evaluation often result in problems concerning comparability
and reproducibility of scientific and technological results. In many cases material data are recorded,
but the methods of testing are insufficiently described, or such information is not communicated
along with the raw data. The material data can also have numerous different formats such as paper
printouts, pdfs, excel sheets or csv-files. Hence, gathering and integrating material data from
different sources is challenging for potential users like materials scientists and engineers, especially if
there are contradictory data where the reasons for contradictions is not clear due their vague
description.

In order to address these problems, data should comply to the so called „FAIR“ principle which calls
for data to be findable, accessible, interoperable, and reusable (FAIR) and hence be accessible via
so-called decentralized but interconnected data spaces. By using knowledge representation with
ontologies, data can be enriched with meaning and the methods of the testing procedures can be
accurately provided.

In this presentation we want to introduce our approach to such knowledge representation based on
a high-throughput alloy development process for Cu-based alloys along with characterization
techniques such as hardness testing and microstructural characterization (e.g. EBSD – Electron
Backscattered Diffraction).

In the paper, the reactivity characteristics of the core of the large sodium fast reactor Superphenix (SPX) were evaluated and compared with available experimental data. The analysis was performed using the TRACE system code modified for the fast reactor applications. The simplified core model was developed aiming to overcome the lack of detailed information on design and realistic core conditions. Point kinetics neutronic model with all relevant reactivity feedbacks was used to calculate transient power. The paper focuses on challenging issue of modeling of the transient thermal responses of primary system structural elements resulting in reactivity feedback specific to such large fast reactor, which cannot be neglected. For these effects, the model was equipped with dedicated heat structures to reproduce important feedback due to vessel wall, diagrid, strongback, control rod drive lines thermal expansion. Peculiarly, application of the model was considered for a whole range of core conditions from zero power to 100% nominal. The developed core model allowed reproducing satisfactorily the core reactivity balance between zero power at 180C and full power conditions. Additionally, the reactivity coefficients k, g, and h at three power levels (about 20, 50, and 80% of the nominal power) were calculated and satisfactory agreement with experimental measurements was also observed. The study demonstrated feasibility of application of relatively simple model with adjusted parameters for analysis of different conditions of very complex system. Reducing some differences with experimentally observed behavior of feedback coefficients,
would require more sophisticated approaches on fuel pin model, more detailed information on management of control rods during power rise, more complicated models of primary system, its structural elements, and flow paths.

In recent years, wastewater-based epidemiology has proven its potential in predicting the pandemic outbreak and helping to understand the spread of the virus. This talk introduces the Saxonian Wastewater dashboard that explores the spatio-temporal relations between indicator values derived from sewage systems and COVID-19 incidence, as measured by conventional testing, in the neighborhood of the wastewater plants. The presented dashboard infrastructure is a collaboration between the projects Wastewater-CoV-2-Tracking (UFZ Leipzig) and Where2test (CASUS/HZDR).

In this paper, an Unprotected Loss of Flow (ULOF) assessment has been performed on the European Sodium Fast Reactor
developed in the ESFR-SMART EU project. To conduct the analysis, a simplified 42 channel thermal–hydraulic model has
been established in the TRACE system code, using a point kinetics model accounting for various reactivity feedback effects. The
assessment reveals the core behavior of a commercial size, 3600 MWth, sodium fast reactor using a state-of-the-art low
void effect reactor core design. The study focuses on the sodium boiling phenomenon and sodium reactivity feedback effect
evolution during the accident with the reference subassembly (SA) design. Following this analysis, a study has been
performed with a modified SA design. The boiling progression and phenomenology within the reference and the modified
core have been compared, and the impact of the SA modification was described.

Understanding exotic forms of magnetism in quantum spin systems is an emergent topic of modern condensed matter physics. Quantum dynamics can be described by particle-like carriers of information, known-as quasiparticles that appear from the collective behaviour of the underlying system. Spinon excitations, governing the excitations of quantum spin-systems, have been accurately calculated and precisely verified experimentally for the antiferromagnetic chain model. However, identification and characterization of novel quasiparticles emerging from the topological excitations of the spin system having periodic exchange interactions are yet to be obtained. Here, we report the identification of emergent composite excitations of the novel quasiparticles doublons and quartons in spin-1/2 trimer-chain antiferromagnet Na₂Cu₃Ge₄O₁₂ (having periodic intrachain exchange interactions J₁-J₁-J₂) and its topologically protected quantum 1/3 magnetization-plateau state. The characteristic energies, dispersion relations, and dynamical structure factor of neutron scattering as well as macroscopic quantum 1/3 magnetization-plateau state are in good agreement with the state-of-the-art dynamical density matrix renormalization group calculations.

The focus of this project is to accelerate and transform the workflow of multiscale materials modeling by developing an integrated toolchain seamlessly combining DFT, SNAP, LAMMPS, (shown in Figure 1-1) and a machine-learning (ML) model that will more efficiently extract information from a smaller set of first-principles calculations. Our ML model enables us to accelerate first-principles data generation by interpolating existing high fidelity data, and extend the simulation scale by extrapolating high fidelity data ( 10 2 atoms) to the mesoscale ( 10 4 atoms). It encodes the underlying physics of atomic interactions on the microscopic scale by adapting a variety of ML techniques such as deep neural networks (DNNs), and graph neural networks (GNNs). We developed a new surrogate model for density functional theory using deep neural networks. The developed ML surrogate is demonstrated in a workflow to generate accurate band energies, total energies, and density of the 298K and 933K Aluminum systems. Furthermore, the models can be used to predict the quantities of interest for systems with more number of atoms than the training data set. We have demonstrated that the ML model can be used to compute the quantities of interest for systems with 100,000 Al atoms. When compared with 2000 Al system the new surrogate model is as accurate as DFT, but three orders of magnitude faster. We also explored optimal experimental design techniques to choose the training data and novel Graph Neural Networks to train on smaller data sets. These are promising methods that need to be explored in the future.

In many energy and process engineering systems where fluids are processed, droplet-laden gas flows may occur. As droplets are often detrimental to the system’s operation, they need to be removed. Compact engineering solutions for the removal of entrained droplets are difficult to achieve with conventional flow control and heat transfer approaches and thus droplet removal devices are hence often costly and bulky. In this study, we analyzed the potential of a compact technology based on droplet capture and in situ evaporation by microwave heating. For that, we designed a microwave applicator containing a porous droplet separator for capturing and evaporating droplets. The application of open-cell ceramic foams as filter medium reduced 99.9% of the volumetric flow of droplets, while additional microwave exposure increases reduction to 99.99%. In addition, microwave-heated foams prevent droplet re-entrainment and structure-borne liquid accumulation within foams, thus avoiding water clogging and flooding.

We study electron-positron pair production within two counter-propagating, circularly polarized electromagnetic fields through the Wigner formalism. We numerically generate high-resolution momentum maps to perform a detailed spectroscopic analysis. We identify signatures of polarization and kinematics of the incident fields in the final positron distribution and, on this basis, provide an intuitive picture of helicity transfer in multiphoton pair production.

Predictive design of REHEDS experiments with radiation-hydrodynamic simulations requires knowledge of material properties (e.g. equations of state (EOS), transport coefficients, and radiation physics). Interpreting experimental results requires accurate models of diagnostic observables (e.g. detailed emission, absorption, and scattering spectra). In conditions of Local Thermodynamic Equilibrium (LTE), these material properties and observables can be pre-computed with relatively high accuracy and subsequently tabulated on simple temperature-density grids for fast look-up by simulations. When radiation and electron temperatures fall out of equilibrium, however, non-LTE effects can profoundly change material properties and diagnostic signatures. Accurately and efficiently incorporating these non-LTE effects has been a longstanding challenge for simulations. At present, most simulations include non-LTE effects by invoking highly simplified inline models. These inline non-LTE models are both much slower than table look-up and significantly less accurate than the detailed models used to populate LTE tables and diagnose experimental data through post-processing or inversion. Because inline non-LTE models are slow, designers avoid them whenever possible, which leads to known inaccuracies from using tabular LTE. Because inline models are simple, they are inconsistent with tabular data from detailed models, leading to ill-known inaccuracies, and they cannot generate detailed synthetic diagnostics suitable for direct comparisons with experimental data. This project addresses the challenge of generating and utilizing efficient, accurate, and consistent non-equilibrium material data along three complementary but relatively independent research lines. First, we have developed a relatively fast and accurate non-LTE average-atom model based on density functional theory (DFT) that provides a complete set of EOS, transport, and radiative data, and have rigorously tested it against more sophisticated first-principles multi-atom DFT models, including time-dependent DFT. Next, we have developed a tabular scheme and interpolation methods that compactly capture non-LTE effects for use in simulations and have implemented these tables in the GORGON magneto-hydrodynamic (MHD) code. Finally, we have developed post-processing tools that use detailed tabulated non-LTE data to directly predict experimental observables from simulation output.

Novel separation processes for end of life fluorescent lamp phosphors could greatly contribute towards a more sustainable rare-earth element market. Furthermore, surface-binding peptides bound to magnetic carriers are a promising biotechnological tool for selective particle separation processes. In this work, we investigate the magnetic carrier separation of the three most common rare-earth fluorescent lamp phosphors, facilitated by Dynabeads functionalized with previously identified selectively surface-binding peptides. We present an active ester-mediated coupling to chemically immobilize the peptides on amine and carboxylic acid coated beads. We report on the impact of the peptide functionalization on the surface properties of the beads, based on zeta potential investigations of variously functionalized beads and a Raman spectroscopic structural study of the investigated peptides. Fluorometrically, we show that the phosphor removal strongly depends on the medium and the surface coating of the beads. Furthermore, the Raman spectroscopic evidence of various simultaneously present disulfide bond conformations indicates an equilibrium of multiple peptide conformations and/or the presence of intermolecular disulfide bonds. Moreover, we found that carboxylic acid coated Dynabeads have a high affinity for the red phosphor Y2O3:Eu3+ and based on the determined isoelectric points we hypothesize that this is driven by electrostatic surface interactions. This work can contribute towards novel rare-earth phosphor separation processes and towards a better understanding of magnetic carrier separation processes facilitated by surface-binding peptides.

Due to their highly complex chemistry and structure, peptide molecules can have the ability to bind to certain inorganic surfaces with a high affinity and selectivity. This property can be utilized for the functionalization of biohybrid materials, in order to create multifunctional, biocompatible materials. In this context, the functionalization of magnetic carrier particles with selectively surface-binding peptides has a potential to play a key-role in innovative particle separation processes aimed at resource recovery and wastewater treatment.
In the junior research group BioKollekt, we have identified selectively surface-binding peptides with a high affinity to certain Rare Earth Element containing fluorescent lamp phosphors by Phage Surface Display (Lederer et al., 2017, [1]). To complete the potential of these peptides for a separation procedure, we have chemically immobilized these peptides on the surfaces of various magnetic carriers. The magnetic properties of such biocollectors enable an efficient and high-throughput separation process. We investigate the peptide immobilization via different coupling procedures and the resulting biocollectors are characterised with regard to their surface and binding properties. Finally, to optimize the technical application of the biocollectors, we investigate the separation capacity of fluorescent lamp phosphors in a rotary permanent magnet separator that was specifically designed for biotechnological purposes (Boelens et al., 2021, [2]).
In this work, the holistic view of a peptide-assisted biomagnetic particle separation process and its upscalibility for resource recovery processes is discussed.

Reference list
1. Lederer et al.; Identification of lanthanum-specific peptides for future recycling of rare earth elements from compact fluorescent lamps: Peptides for Rare Earth Recycling. Biotechnol. Bioeng. 2016, 114, doi:10.1002/bit.26240.
2. Boelens et al.; High-Gradient Magnetic Separation of Compact Fluorescent Lamp Phosphors: Elucidation of the Removal Dynamics in a Rotary Permanent Magnet Separator. Minerals 2021, 11, doi:10.3390/min11101116.

In a global effort towards a low-carbon and green economy, the ongoing search for novel techniques to recycle rare-earth elements (REEs) will play a crucial role, due to their increasing demand, high supply risk and incompatibility with conventional separation methods [1,2]. In this context, surface-binding peptides immobilized on magnetic carriers can facilitate highly specific interactions with their target materials to promote highly innovative and selective particle separation processes [3].
Previously, the Junior Research Group BioKollekt has applied Phage Surface Display to succesfully identify selectively surface-binding peptides that interact with the commercially applied green phosphor LaPO4:Ce,Tb [4]. Subsequently, we have thoroughly characterized a range of REE phosphors, commercially applied in fluorescent lamps, and have proven their compatibility with an upscalable biotechnological high-gradient magnetic separator [5]. Furthermore, we have investigated the interaction of the identified peptides with the target phosphors, in dissolved as well as chemically immobilized conformations [6].
In this work, we present the use of REE phosphor-binding peptides for the functionalization of magnetic nanoparticles by chemical immobilization. We give a comprehensive overview of the peptides’ roles in the nanoparticle functionalization, interaction with the target phosphors and an upscalable biomagnetic separation, as summarized in Fig. 1. Amongst others, we present the surface load of the immobilized peptides on the nanoparticles, as well as surface zeta potential measurement.
Finally, this work can shine a light on the future perspectives of peptides for their role in selective particle separation processes for environmental applications.

REFERENCES
[1] European Commission, Study on the EU’s list of Critical Raw Materials – Final Report (2020).
[2] Binnemans, K.; Jones, P.; Blanpain, B.; Van Gerven, T.; Yang, Y.; Walton, A.; Buchert, M. Recycling of Rare Earths a Critical Review. Journal of Cleaner Production 2013, 51, 1-22, doi:10.1016/j.jclepro.2012.12.037.
[3] Pollmann, K.; Kutschke, S.; Matys, S.; Raff, J.; Hlawacek, G.; Lederer, F. Bio-recycling of metals: Recycling of technical products using biological applications. Biotechnol. Adv. 2018, 36, doi:10.1016/j.biotechadv.2018.03.006.
[4] Lederer, F.; Curtis, S.; Bachmann, S.; Dunbar, S.; MacGillivray, R. Identification of lanthanum-specific peptides for future recycling of rare earth elements from compact fluorescent lamps: Peptides for Rare Earth Recycling. Biotechnol. Bioeng. 2016, 114, doi:10.1002/bit.26240.
[5] Boelens, P.; Lei, Z.; Drobot, B.; Rudolph, M.; Li, Z.; Franzreb, M.; Eckert, K.; Lederer, F. High-Gradient Magnetic Separation of Compact Fluorescent Lamp Phosphors: Elucidation of the Removal Dynamics in a Rotary Permanent Magnet Separator. Minerals 2021, 11, doi:10.3390/min11101116.
[6] Schrader, M.; Bobeth, C.; Lederer, F. Quantification of Peptide-Bound Particles: A Phage Mimicking Approach via Site-Selective Immobilization on Glass. ACS Omega 2021, XXXX, doi:10.1021/acsomega.1c04343.

Light reflected or emitted from a natural surface contains material-characteristic, spectral signatures like fingerprints. Hyperspectral imaging sensors can capture this information in image rasters with hundreds of discrete spectral channels. The resulting spectral data cube can be analysed to create detailed maps of the surfaces’ material composition. Hyperspectral imaging is experiencing rapid transformation, mostly due to the ongoing miniaturization of sensors, the boost in computer processing power and the need for fast and non-invasive characterization technologies. The technology is versatile in regards to application type and scale, and can be applied using passive illumination, e.g., sun- or skylight. Hyperspectral imaging already supports a variety of application fields in earth observation, such as agriculture, geoscience, urban planning and environmental monitoring, ranging from global (satellite-borne) down to sample scale (lab).

Mountainous environments pose a specific challenge for hyperspectral imaging, as topographic complexity often requires oblique (non-nadir) acquisition, while illumination conditions strongly vary in time and space, and entrenched 2-D data analysis techniques (e.g. using 2-D gridded data such as DEMs and orthomosaics) are of limited applicability. It is important to move beyond the current usage of hyperspectral data as 2D rasters and go towards a more complex, but also more realistic 3D representation. This avoids occlusion and false-neighbourhood effects and allows us to accurately correct illumination effects induced by the geometry of the target with respect to the illumination source and the sensor positions. It enables the deployment of hyperspectral sensors from innovative, yet challenging platforms and non-nadir observation angles, occurring with tripod- and drone-based acquisitions (Fig. 1). The required transfer of hyperspectral data to a 3D “hypercloud”, i.e., a geometrically and spectrally accurate combination of a photogrammetric point cloud and the hyperspectral datacube (Fig. 2), ultimately allows the fusion of multi-scale and multi-platform scenes as well as the integration of sample data or subsurface information. With careful correction, the resulting dataset can provide tremendous value in mountain research, e.g., for the estimation of variation in mineralogical composition for better understanding of rock-forming processes, the detection of plant species and lichen coverage or monitoring of environmental changes.

With this contribution, we give an overview of the challenges and opportunities of spectral imaging in the context of mountain research. We showcase best practices and trends in data acquisition, platforms and data correction workflows, and highlight the advantages of the hypercloud approach for the mapping of steep and complex targets. We give examples from geoscience and mineral exploration perspective, covering natural alpine outcrops and cliffs of different scale and geological setting, as well as artificial outcrops in mining and exploration context.

Observing and understanding Earth’s processes are more important than ever as the demand for resources and the human impact on the planet are skyrocketing. Hyperspectral mineral mapping is a crucial Earth observation (EO) tool with manifold applications, including understanding geological processes (e.g., for waste storage, geothermal energy), green- and brownfield mineral exploration, monitoring mining activities (e.g., grade control, monitoring of tailing dams), characterizing human-made mineral deposits (e.g. tailings, contaminations and mine drainage precipitates) and monitoring the local and global impact of mining on the environment.
EO enables digital archiving of mineralogic information and drives the transition from traditional maps to digital twins of the Earth’s surface. However, it also implies specific requirements that future spaceborne missions will need to meet in order to support the above applications:
Scale and orientation: Geological features of interest span a wide range and may require the simultaneous interpretation of cm-scale features (e.g. veins, fractures) and regional scale variations in mineral composition (e.g. alteration halos). At the same time, geological outcrops are often obliquely oriented and may be obscured if only nadir data are collected. The integration of data collected from different vantage points and at different scales (space-, air-, drone-borne, terrestrial) is critical for meaningful analysis. For a great part of applications (e.g. monitoring mining activities), additional temporal coverage is crucial. Coordinated acquisition of multi-mission data, established processing platforms, and careful corrections are required to enable this framework.
Spectral range: Mineralogically relevant information is contained (often exclusively) in confined spectral ranges, which are in the visible and shortwave, but also mid- and longwave infrared range. Especially the latter must not be forgotten if the full portfolio of hyperspectral mineral mapping is to be achieved. A careful selection of relevant spectral regions and adapted spectral resolution could help to reduce data load and improve spatial resolution.
Processing and validation: Tremendous progress has been made towards machine learning assisted processing of hyperspectral datasets. Nevertheless, developments too often rely on simple and small benchmark datasets. Large scale, mineralogically relevant datasets struggle with heterogeneous, scale dependent classes and often subjective geological interpretation. We recently established three reference sites in Europe that integrate reference data from different scales and technologies as part of the EU-funded INFACT project. We need to continue this effort and engage the community to provide both large-scale benchmarked datasets as well as architectures suitable for scalable machine learning approaches.

Weeping is known to significantly reduce tray and point efficiencies in distillation tray columns. Recently, an analytical method was developed to simultaneously determine tray and point efficiencies in case of weeping. The approach, however, is based on the assumption of uniformly distributed weeping, while experimental studies generally reveal non-uniform weeping. In this study, the effect of several non-homogeneous weeping distributions on tray and point efficiencies is evaluated. In particular, state-of-the-art non-uniform weeping distributions, which have low spatial resolution, have been compared with arbitrarily assumed high resolution weeping distributions. The results illustrate that accurately quantifying the weeping distribution on the tray, allows to significantly improve the accuracy of calculated tray and point efficiencies.

The absence of stray fields, their insensitivity to external magnetic fields, and ultrafast dynamics make antiferromagnets promising candidates for active elements in spintronic devices. Here, we demonstrate manipulation of the Néel vector in the metallic collinear antiferromagnet Mn₂Au by combining strain and femtosecond laser excitation. Applying tensile strain along either of the two in-plane easy axes and locally exciting the sample by a train of femtosecond pulses, we align the Néel vector along the direction controlled by the applied strain. The dependence on the laser fluence and strain suggests the alignment is a result of optically triggered depinning of 90° domain walls and their motion in the direction of the free energy gradient, governed by the magneto-elastic coupling. The resulting, switchable state is stable at room temperature and insensitive to magnetic fields. Such an approach may provide ways to realize robust highdensity memory device with switching time scales in the picosecond range.

We report a saturation magnetic field of 41 T at 4.2 K in the antiferromagnet EuIrSi₃ (T_N = 51.8 K), much larger than the values in typical S-state (net orbital state, L = 0) magnetic systems but consistent with the mean-field theory. We interpret this anomalous behaviour in conjunction with a higher density of states of conduction electrons in EuIrSi₃ compared to other isostructural members in the EuTX₃ (T = Ni, Pt, Rh, Ir; X = Si, Ge) homologous series. Moreover, low-temperature isothermal magnetization indicates spin orientation occurring at fields 13 T and 33 T, respectively. A magnetic phase diagram by aid of pulsed-field magnetization and magnetoresistivity measurements of EuIrSi₃ is constructed.

ASL-BIDS and OSIPI Pipeline Inventory

Superparamagnetic composite beads are widely used as magnetic carriers in biotechnological processes, including the purification of biomolecules, organelles and cells [1,2]. Their wide range of applications include diagnostic, as well as industrial purposes. Furthermore, the immobilization of surface-binding peptides can render highly specific surface properties to composite beads and facilitate their selective interaction with target particles. In this context, peptide functionalized composite beads have been shown to be promising tools for environmental applications, including biomining and wastewater treatment [3-5]. Nevertheless, to the best of our knowledge, their use has so far only been investigated in low magnetic field gradients, on a milliliter scale.

The waste of fluorescent lamps contains several valuable Rare Earth phosphors in the form of fine particles that are hard to separate and therefore lack efficient recycling schemes [6]. Our junior researchgroup, BioKollekt, has previously identified selectively surface-binding peptides that interact with the Rare Earth phosphor LaPO4:Ce,Tb [7]. Recently, we have chemically immobilized the identified peptides and have tested their interaction with several target phosphors [8], we have thoroughly characterized a range of Rare Earth phosphors and we have shown their compatibility with an upscalable High-Gradient Magnetic Separator [9], which was specifically designed for biotechnological separations with superparamagnetic carriers [10].

In this work, we investigate the use of Dynabeads® M-270, functionalized with previously identified peptides, for the separation of Rare Earth phosphors. First, we characterize the physical properties of functionalized and unfunctionalized beads. Subsequently, we examine the beads’ selectivities towards various Rare Earth phosphors in an LGMS setup. Finally, we compare the carrier behaviour of the beads in low and high magnetic field gradients by the use of an optical microscopic setup. A special focus is placed on the magnetically induced chain formation by sets of beads. Finally, this work can shine a light on the future perspectives of peptide functionalized superparamagnetic composite beads for a selective and upscalable separation process of fine particles.

REFERENCES
1. Leong, S.; Yeap, S.P.; Lim, J.K. Working principle and application of magnetic separation for biomedical diagnostic at high- and low-field gradients. Interface focus 2016, 6, doi:10.1098/rsfs.2016.0048.
2. Berensmeier, S. Magnetic Particles for the Separation and Purification of Nucleic Acids. Appl. Microbiol. Biotechnol. 2007, 73, 495-504, doi:10.1007/s00253-006-0675-0.
3. Cetinel, S.; Shen, W.-Z.; Aminpour, M.; Bhomkar, P.; Wang, F.; Borujeny, E.; Sharma, K.; Nayebi, N.; Montemagno, C. Biomining of MoS2 with Peptide-based Smart Biomaterials. Scientific Reports 2018, 8, doi:10.1038/s41598-018-21692-4.
4. Vreuls, C.; Genin, A.; Zocchi, G.; Boschini, F.; Cloots, R.; Gilbert, B.; Martial, J.; Weerdt, C. Genetically engineered polypeptides as a new tool for inorganic nano-particles separation in water based media. J. Mater. Chem. 2011, 21, 13841-13846, doi:10.1039/C1JM12440D.
5. Pollmann, K.; Kutschke, S.; Matys, S.; Raff, J.; Hlawacek, G.; Lederer, F. Bio-recycling of metals: Recycling of technical products using biological applications. Biotechnol. Adv. 2018, 36, doi:10.1016/j.biotechadv.2018.03.006.
6. Binnemans, K.; Jones, P. Perspectives for the recovery of rare earths from end-of-life fluorescent lamps. Journal of Rare Earths 2014, 32, 195–200, doi:10.1016/S1002-0721(14)60051-X.
7. Lederer, F.; Curtis, S.; Bachmann, S.; Dunbar, S.; MacGillivray, R. Identification of lanthanum-specific peptides for future recycling of rare earth elements from compact fluorescent lamps: Peptides for Rare Earth Recycling. Biotechnol. Bioeng. 2016, 114, doi:10.1002/bit.26240.
8. Schrader, M.; Bobeth, C.; Lederer, F. Quantification of Peptide-Bound Particles: A Phage Mimicking Approach via Site-Selective Immobilization on Glass. ACS Omega 2021, XXXX, doi:10.1021/acsomega.1c04343.
9. Boelens, P.; Lei, Z.; Drobot, B.; Rudolph, M.; Li, Z.; Franzreb, M.; Eckert, K.; Lederer, F. High-Gradient Magnetic Separation of Compact Fluorescent Lamp Phosphors: Elucidation of the Removal Dynamics in a Rotary Permanent Magnet Separator. Minerals 2021, 11, doi:10.3390/min11101116.
10. Hoffmann, C.; Franzreb, M.; Holl, W.H. A novel high-gradient magnetic separator (HGMS) design for biotech applications. IEEE Transactions on Applied Superconductivity 2002, 12, 963-966, doi:10.1109/TASC.2002.1018560.

The retention of actinides in different oxidation states (An(X), X = III, IV, VI) by a calcium-silicate-hydrate (C-S-H) phase with a Ca/Si (C/S) ratio of 0.8 was investigated in the presence of gluconate (GLU). The actinides considered were Am(III), Th(IV), Pu(IV), and U(VI). Eu(III) was investigated as chemical analogue for Am(III) and Cm(III). In addition to the ternary systems An(X)/GLU/C-S-H, also binary systems An(X)/C-S-H, GLU/C-S-H, and An(X)/GLU were studied. Complementary analytical techniques were applied to address the different specific aspects of the binary and ternary systems. Time-resolved laser-induced luminescence spectroscopy (TRLFS) was applied in combination with parallel factor analysis (PARAFAC) to identify retained species and to monitor species-selective sorption kinetics. 13C and 29Si magic-angle-spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy and X-ray photoelectron spectroscopy (XPS) were applied to determine the bulk structure and the composition of the C-S-H surface, respectively, in the absence and presence of GLU. The interaction of Th(IV) with GLU in different electrolytes was studied by capillary electrophoresis-inductively coupled plasma mass spectrometry (CE-ICP-MS). The influence of GLU on An(X) retention was investigated for a large concentration range up to 10−2 M. The results showed that GLU had little to no effect on the overall An(X) retention by C-S-H with C/S of 0.8, regardless of the oxidation state of the actinides. For Eu(III), the TRLFS investigations additionally implied the formation of a Eu(III)-bearing precipitate with dissolved constituents of the C-S-H phase, which becomes structurally altered by the presence of GLU. For U(VI) sorption on the C-S-H phase, only a small influence of GLU could be established in the luminescence spectroscopic investigations, and no precipitation of U(VI)-containing secondary phases could be identified.

Waste from electronic equipment (WEEE) is a fast-growing complex waste stream, and plastics represent around 25% of its total. The proper recycling of plastics from WEEE depends on the identification of polymers prior to entering the recycling chain. Technologies aiming for this identification must be compatible with conveyor belt operations and fast data acquisition. Therefore, we selected three promising sensor types to investigate the potential of optical spectroscopy-based methods for identification of plastic constituents in WEEE. Reflectance information is obtained using Hyperspectral cameras (HSI) in the short-wave infrared (SWIR) and mid-wave infrared (MWIR). Raman point acquisitions are well-suited for specific plastic identification (532 nm excitation). Integration times varied according to the capabilities of each sensor, never exceeding 2 seconds. We have selected 23 polymers commonly found in WEEE (PE, PP, PVC ABS, PC, PS, PTFE, PMMA), recognising spectral fingerprints for each material according to literature reports. Spectral fingerprint identification was possible for 60% of the samples using SWIR-HSI; however, it failed to produce positive results for black plastics. Additional information from MWIR-HSI was used to identify two black samples (70% identified using SWIR + MWIR). Fingerprint assignment in short- time Raman acquisition (1 -2 seconds) was successful for all samples. Combined with the efficient mapping capabilities of HSI at time scales of milliseconds, further developments promise great potential for fast-paced recycling environments. Furthermore, integrated solutions enable increased accuracy (cross-validations) and hence, we recommend a combination of at least 2 sensors (SWIR + Raman or MWIR + Raman) for recycling activities.

Hybrid organic-inorganic halide perovskites represent a promising next-generation photovoltaic material with drawbacks on structure stability and composition concerns. Demonstrations of ion migration and molecular dynamics suggest room for structural contraction and subsequent property adjustments. Here, we have deployed dielectric and infrared spectroscopy under external pressure to probe the full structural phase diagram and dielectric response of methylammonium lead halide perovskites CH₃NH₃$PbX₃ (X = I, Br, or Cl). Ion migration can be fully suppressed by pressure beyond 4 GPa. The low-temperature orthorhombic phase transition can be gradually enhanced and stabilized at ambient conditions with increasing pressure. A slow relaxation mode, presumably the motion of the CH₃NH₃+ cation, is observed at lower pressure and is absent in the orthorhombic phase for every halide.

Chimeric antigen receptor (CAR) T-cell therapies are undoubtedly a promising approach in cancer immunotherapy and have revolutionized the treatment options for hematologic malignancies. Yet, their effectiveness is more limited and challenging when it comes to tackling solid tumours. Antigen escape, on-target off-tumour effects, and the immunosuppressive tumour microenvironment (TME) of solid tumours are among the main hurdles. Aiming for increased clinical safety and effectiveness, our group works with the adaptor CAR T-cell technologies named UniCAR and RevCAR. The versatility of the adaptor molecules used in these types of approaches allows not only the targeting of antigens expressed on the surface of tumour cells but also of key molecules expressed on immunosuppressive cells as well as immune checkpoint inhibitors. Given the features of such approaches and the ease in engineering the adaptor molecules, combinatorial targeting can be exploited to pave the way for improved and personalized immunotherapies.

Chimeric antigen receptor (CAR) T-cells are unquestionably considered one of the most promising approaches in cancer immunotherapy. Nonetheless, mild to severe toxicities are associated with this approach which include e.g. on-target/off-tumour toxicities and cytokine release syndrome. Aiming for increased clinical safety, a modular universal CAR (UniCAR) platform was developed by our group in which UniCAR T-cells are exclusively activated in the presence of a target module (TM) that specifically establishes the cross-link between target cells and UniCAR T-cells. Fibroblast activation protein (FAP) is highly expressed on cancer-associated fibroblasts (CAFs) present in the tumour stroma and also found to be overexpressed in tumour cells. This protein plays an important role in promoting tumour growth, metastasis, and immunosuppression and has therefore been studied as a target for cancer diagnosis and treatment. Given this and the demonstrated efficacy, flexibility and switchability of the UniCAR system, currently demonstrated in phase I clinical trials, we hereby aimed to develop TMs targeting FAP that can be used for both immunotherapeutic and theranostic approaches. For that, the single-chain variable fragment (scFv) of an anti-human FAP mAb was fused to the peptide epitope E5B9 that is recognized by the UniCAR T-cells, creating low molecular weight TMs that are rapidly eliminated allowing a specific and recurrent on/off switch of UniCAR T-cell activity via TM dosing. Additionally, extended half-life anti-FAP TMs based on the human IgG4 Fc-domain, including a mutated version, were created intending to strengthen anti-tumour responses and to ease the clinical TM administration at later stages of tumour therapy. All TMs were tested in vitro based on naturally and artificially overexpressing 2D and 3D models, and proven to specifically redirect UniCAR T-cells to FAP-expressing target cells. Positron emission tomography (PET) using 64Cu radiolabelled anti-FAP IgG4 TMs demonstrated a FAP specific enrichment of these TMs at the tumour site of FAP overexpressing HT1080 xenografts resulting in a tumour SUV of 50 at 48h p.i. with almost no background. Single-photon emission tomography (SPECT) using 177Lu radiolabelled anti-FAP IgG4 TMs furthermore confirmed a high FAP-dependent tumour uptake. In conclusion, we hereby designed novel TMs targeting FAP with different formats that can be used for both endoradionuclide therapy and immunotherapeutic approaches using UniCAR T-cells, proving to be promising and innovative immunotheranostic tools to foster cancer treatment allowing a more convenient, individualized, and safe treatment of cancer patients.

Prediction and control of the behavior of asphaltenes have always been one of the most challenging topics in the oil and gas industry. To this aim, comprehensive knowledge of the bulk and surface behavior of asphaltenes and the possible interrelationship are required. Many attempts have been done so far to scrutinize the behavior of asphaltenes at the oil/water interface. In this study, the changes in surface behavior of asphaltenes upon aromaticity variation in the bulk are investigated. For this purpose, dynamic light scattering (DLS) experiments are accompanied with dynamic interfacial tension (IFT), and interfacial rheology measurements to analyze the asphaltene aggregation size, mobility and surface activity upon a change in the bulk aromaticity. A wide range of aromatic conditions was used to cover possible changes in asphaltene behavior before and after the onset point. The results revealed that the interfacial activity and the structure of the interfacial film of the asphaltenes are significantly influenced by the bulk properties. Low aromaticity conditions led to an increase in asphaltene surface activity and changed the dilational rheological parameters considerably although the general trend of the asphaltene adsorption remained similar. Moreover, the results demonstrated the importance of the aging time on the measured parameters, particularly for the samples beyond the onset point. Our findings also showed the importance of dynamic interfacial data over static ones. Finally, by applying the concept of diffusion coefficient it was suggested that only a small portion of asphaltenes remain surface-active practically, although their surface activity is directly correlated with the system aromaticity.

We present here the scientific outcomes of the 2021 Data Fusion Contest (DFC2021) organized by the Image Analysis and Data Fusion Technical Committee of the IEEE Geoscience and Remote Sensing Society. DFC2021 was dedicated to research on geospatial artificial intelligence (AI) for social good with a global objective of modeling the state and changes of artificial and natural environments from multimodal and multitemporal remotely sensed data toward sustainable developments. DFC2021 included two challenge tracks: “Detection of settlements without electricity” and “Multitemporal semantic change detection.” This article mainly focuses on the outcome of the multitemporal semantic change detection track. We describe in this article the DFC2021 dataset that remains available for further evaluation of corresponding approaches and report the results of the best-performing methods during the contest.

The effects of the spatial resolution of remote sensing (RS) data on wildfire susceptibility prediction are not fully understood. In this study, we evaluate the effects of coarse (Landsat 8 and SRTM) and medium (Sentinel-2 and ALOS) spatial resolution data on wildfire susceptibility prediction using random forest (RF) and support vector machine (SVM) models. In addition, we investigate the fusion of the predictions from the different spatial resolutions using the Dempster–Shafer theory (DST) and 14 wildfire conditioning factors. Seven factors are derived separately from the coarse and medium spatial resolution datasets for the whole forest area of the Guilan Province, Iran. All conditional factors are used to train and test the SVM and RF models in the Google Earth Engine (GEE) software environment, along with an inventory dataset from comprehensive global positioning system (GPS)-based field survey points of wildfire locations. These locations are evaluated and combined with coarse resolution satellite data, namely the thermal anomalies product of the moderate resolution imaging spectroradiometer (MODIS) for the period 2009 to 2019. We assess the performance of the models using four-fold cross-validation by the receiver operating characteristic (ROC) curve method. The area under the curve (AUC) achieved from the ROC curve yields 92.15% and 91.98% accuracy for the respective SVM and RF models for the coarse RS data. In comparison, the AUC for the medium RS data is 92.5% and 93.37%, respectively. Remarkably, the highest AUC value of 94.71% is achieved for the RF model where coarse and medium resolution datasets are combined through DST.

This study investigates a pixel-based image analysis methodology built on unsupervised Deep Learning (DL) for rapid landslide detection. The utilized data includes the Minimum Noise Fraction (MNF) and Normalized Difference Vegetation Index (NDVI) derived from Sentinel-2 images and the topographic slope factor derived from the ALOS PALSAR sensor. We used a Convolutional auto-encoder (CAE) for extracting deep features from our input data. The Mini Batch K-means is then used for clustering the resulting deep features. The resulting landslide detection maps were then compared with a landslide inventory dataset for accuracy assessment. The proposed approach achieved the highest values of 76%, 91%, 83%, and 70% in terms of precision, recall, f1-score, and mIOU, respectively. This is the first study investigating unsupervised DL for landslide detection using Sentinel-2 images to the best of our knowledge.

Recently, many collaborative representation-based (CR) algorithms have been proposed for hyperspectral anomaly detection. CR-based detectors approximate the image by a linear combination of background dictionaries and the coefficient matrix, and derive the detection map by utilizing recovery residuals. However, these CR-based detectors are often established on the premise of precise background features and strong image representation, which are very difficult to obtain. In addition, pursuing the coefficient matrix reinforced by the general -min is very time consuming. To address these issues, a nonnegative-constrained joint collaborative representation model is proposed in this paper for the hyperspectral anomaly detection task. To extract reliable samples, a union dictionary consisting of background and anomaly sub-dictionaries is designed, where the background sub-dictionary is obtained at the superpixel level and the anomaly sub-dictionary is extracted by the pre-detection process. And the coefficient matrix is jointly optimized by the Frobenius norm regularization with a nonnegative constraint and a sum-to-one constraint. After the optimization process, the abnormal information is finally derived by calculating the residuals that exclude the assumed background information. To conduct comparable experiments, the proposed nonnegative-constrained joint collaborative representation (NJCR) model and its kernel version (KNJCR) are tested in four HSI data sets and achieve superior results compared with other state-of-the-art detectors.

Recent landslide detection studies have focused on pixel-based deep learning (DL) approaches. In contrast, intuitive annotation of landslides from satellite imagery is based on distinct features rather than individual pixels. This study examines the feasibility of the integration framework of a DL model with rule-based object-based image analysis (OBIA) to detect landslides. First, we designed a ResU-Net model and then trained and tested it in the Sentinel-2 imagery. Then we developed a simple rule-based OBIA with only four rulesets, applying it first to the original image dataset and then to the same dataset plus the resulting ResU-Net heatmap. The value of each pixel in the heatmap refers to the probability that the pixel belongs to either landslide or non-landslide classes. Thus, we evaluate three scenarios: ResU-Net, OBIA, and ResU-Net-OBIA. The landslide detection maps from three different classification scenarios were compared against a manual landslide inventory map using thematic accuracy assessment metrics: precision, recall, and f1-score. Our experiments in the testing area showed that the proposed integration framework yields f1-score values 8 and 22 percentage points higher than those of the ResU-Net and OBIA approaches, respectively.

Humans rely on clean water for their health, well-being, and various socio-economic activities. During the past few years, the COVID-19 pandemic has been a constant reminder of about the importance of hygiene and sanitation for public health. The most common approach to securing clean water supplies for this purpose is via wastewater treatment. To date, an effective method of detecting wastewater treatment plants (WWTP) accurately and automatically via remote sensing is unavailable. In this paper, we provide a solution to this task by proposing a novel joint deep learning (JDL) method that consists of a fine-tuned object detection network and a multi-task residual attention network (RAN). By leveraging OpenStreetMap (OSM) and multimodal remote sensing (RS) data, our JDL method is able to simultaneously tackle two different tasks: land use land cover (LULC) and WWTP classification. Moreover, JDL exploits the complementary effects between these tasks for a performance gain. We train JDL using 4,187 WWTP features and 4,200 LULC samples and validate the performance of the proposed method over a selected area around Stuttgart with 723 WWTP features and 1,200 LULC samples to generate an LULC classification map and a WWTP detection map. Extensive experiments conducted with different comparative methods demonstrate the effectiveness and efficiency of our JDL method in automatic WWTP detection in comparison with single-modality/single-task or traditional survey methods. Moreover, lessons learned pave the way for future works to simultaneously and effectively address multiple large-scale mapping tasks (e.g., both mapping LULC and detecting WWTP) from multimodal RS data via deep learning.

Remote sensing image retrieval (RSIR), aiming at searching for a set of similar items to a given query image, is a very important task in remote sensing applications. Deep hashing learning as the current mainstream method has achieved satisfactory retrieval performance. On one hand, various deep neural networks are used to extract semantic features of remote sensing images. On the other hand, the hashing techniques are subsequently adopted to map the high-dimensional deep features to the low-dimensional binary codes. This kind of method attempts to learn one hash function for both the query and database samples in a symmetric way. However, with the number of database samples increasing, it is typically time-consuming to generate the hash codes of large-scale database images. In this article, we propose a novel deep hashing method, named asymmetric hash code learning (AHCL), for RSIR. The proposed AHCL generates the hash codes of query and database images in an asymmetric way. In more detail, the hash codes of query images are obtained by binarizing the output of the network, while the hash codes of database images are directly learned by solving the designed objective function. In addition, we combine the semantic information of each image and the similarity information of pairs of images as supervised information to train a deep hashing network, which improves the representation ability of deep features and hash codes. The experimental results on three public datasets demonstrate that the proposed method outperforms symmetric methods in terms of retrieval accuracy and efficiency. The source code is available at https://github.com/weiweisong415/Demo_AHCL_for_TGRS2022 .

Remote sensing hyperspectral cameras acquire high spectral-resolution data that reveal valuable composition information on the targets (e.g., for Earth observation and environmental applications). The intrinsic high dimensionality and the lack of sufficient numbers of labeled/training samples prevent efficient processing of hyperspectral images (HSIs). HSI clustering can alleviate these limitations. In this study, we propose a multiscale spectral–spatial association network (MS 2 A-Net) to cluster HSIs. The backbone of MS 2 A-Net is an autoencoder architecture that allows the network to capture the nonlinear relation between data points in an unsupervised manner. The network applies a multistream approach. One stream extracts spectral information by deploying a spectral association unit. The other stream derives multiscale contextual and spatial information by employing dilated (atrous) convolutional kernels. The obtained feature representation generated by MS 2 A-Net is fed into a standard k-means clustering algorithm to produce the final clustering result. Extensive experiments on four HSIs for different types of applications (i.e., geological-, rural-, and urban-mapping) demonstrate the superior performance of MS 2 A-Net over the state-of-the-art shallow/deep learning-based clustering approaches in terms of clustering accuracy.

This work evaluates the performance of three machine learning (ML) techniques, namely logistic regression (LGR), linear regression (LR), and support vector machines (SVM), and two multi-criteria decision-making (MCDM) techniques, namely analytical hierarchy process (AHP) and the technique for order of preference by similarity to ideal solution (TOPSIS), for mapping landslide susceptibility in the Chitral district, northern Pakistan. Moreover, we create landslide inventory maps from LANDSAT-8 satellite images through the change vector analysis (CVA) change detection method. The change detection yields more than 500 landslide spots. After some manual post-processing correction, the landslide inventory spots are randomly split into two sets with a 70/30 ratio for training and validating the performance of the ML techniques. Sixteen topographical, hydrological, and geological landslide-related factors of the study area are prepared as GIS layers. They are used to produce landslide susceptibility maps (LSMs) with weighted overlay techniques using different weights of landslide-related factors. The accuracy assessment shows that the ML techniques outperform the MCDM methods, while SVM yields the highest accuracy of 88% for the resulting LSM.

Motivated by the lack of research on land cover and dust activity in the Middle East, this study seeks to increase the understanding of the sensitivity of dust centers to climatic and surface conditions in this specific region. In this regard, we explore vegetation cover and dust emission interactions using 16-day long-term Normalized Difference Vegetation Index (NDVI) data and daily Aerosol Optical Depth (AOD) data from Moderate Resolution Imaging Spectroradiometer (MODIS) and conduct spatiotemporal and statistical analyses. Eight major dust hotspots were identified based on long-term AOD data (2000–2019). Despite the relatively uniform climate conditions prevailing throughout the region during the study period, there is considerable spatial variability in interannual relationships between AOD and NDVI. Three subsets of periods (2000–2006, 2007–2013, 2014–2019) were examined to assess periodic spatiotemporal changes. In the second period (2007–2013), AOD increased significantly (6% to 32%) across the studied hotspots, simultaneously with a decrease in NDVI (−0.9% to −14.3%) except in Yemen−Oman. Interannual changes over 20 years showed a strong relationship between reduced vegetation cover and increased dust intensity. The correlation between NDVI and AOD (−0.63) for the cumulative region confirms the significant effect of vegetation canopy on annual dust fluctuations. According to the results, changes in vegetation cover have an essential role in dust storm fluctuations. Therefore, this factor must be regarded along with wind speed and other climate factors in Middle East dust hotspots related to research and management efforts.

Recently, deep learning methods, especially convolutional neural networks (CNNs), have achieved good performance for hyperspectral image (HSI) classification. However, due to limited training samples of HSIs and the high volume of trainable parameters in deep models, training deep CNN-based models is still a challenge. To address this issue, this study investigates contrastive learning (CL) as a pre-training strategy for HSI classification. Specifically, a supervised contrastive learning (SCL) framework, which pre-trains a feature encoder using an arbitrary number of positive and negative samples in a pair-wise optimization perspective, is proposed. Additionally, three techniques for better generalization in the case of limited training samples are explored in the proposed SCL framework. First, a spatial–spectral HSI data augmentation method, which is composed of multiscale and 3D random occlusion, is designed to generate diverse views for each HSI sample. Second, the features of the augmented views are stored in a queue during training, which enriches the positives and negatives in a mini-batch and thus leads to better convergence. Third, a multi-level similarity regularization method (MSR) combined with SCL (SCL–MSR) is proposed to regularize the similarities of the data pairs. After pre-training, a fully connected layer is combined with the pre-trained encoder to form a new network, which is then fine-tuned for final classification. The proposed methods (SCL and SCL–MSR) are evaluated on four widely used hyperspectral datasets: Indian Pines, Pavia University, Houston, and Chikusei. The experiment results show that the proposed SCL-based methods provide competitive classification accuracy compared to the state-of-the-art methods.

The synergistic combination of deep learning (DL) models and Earth observation (EO) promises significant advances to support the Sustainable Development Goals (SDGs). New developments and a plethora of applications are already changing the way humanity will face the challenges of our planet. This article reviews current DL approaches for EO data, along with their applications toward monitoring and achieving the SDGs most impacted by the rapid development of DL in EO. We systematically review case studies to achieve zero hunger, create sustainable cities, deliver tenure security, mitigate and adapt to climate change, and preserve biodiversity. Important societal, economic, and environmental implications are covered. Exciting times are coming when algorithms and Earth data can help in our endeavor to address the climate crisis and support more sustainable development.

Deep neural networks have achieved great success in many important remote sensing tasks. Nevertheless, their vulnerability to adversarial examples should not be neglected. In this study, we systematically analyze the Universal Adversarial Examples in Remote Sensing (UAE-RS) data for the first time, without any knowledge from the victim model. Specifically, we propose a novel black-box adversarial attack method, namely, Mixup-Attack, and its simple variant Mixcut-Attack, for remote sensing data. The key idea of the proposed methods is to find common vulnerabilities among different networks by attacking the features in the shallow layer of a given surrogate model. Despite their simplicity, the proposed methods can generate transferable adversarial examples that deceive most of the state-of-the-art deep neural networks in both scene classification and semantic segmentation tasks with high success rates. We further provide the generated universal adversarial examples in the dataset named UAE-RS, which is the first dataset that provides black-box adversarial samples in the remote sensing field. We hope UAE-RS may serve as a benchmark that helps researchers design deep neural networks with strong resistance toward adversarial attacks in the remote sensing field. Codes and the UAE-RS dataset are available online ( https://github.com/YonghaoXu/UAE-RS ).

Complex oxides host a multitude of novel phenomena in condensed matter physics, such as various forms of multiferroicity, colossal magnetoresistance, quantum magnetism and superconductivity. Defect engineering via ion irradiation can be a useful knob to control these physical properties for future practical applications. Two prominent effects are disorder and uniaxial strain. Particularly, the uniaxial strain, manifesting as the elongation of the out-of-plane lattice spacing, is not limited to available substrates. In this contribution, we will take SrRuO3 thin films as an example to show the emerging properties upon defect engineering by ion irradiation. The irradiated SRO films exhibit a pronounced topological Hall effect in a wide temperature range from 5 to 80 K. It can be attributed to the emergence of Dzyaloshinskii–Moriya interaction as a result of artificial inversion symmetry breaking associated with the lattice defect engineering. Ion irradiation has been well developed for semiconductor-chip technology and is readily applicable for all kinds of oxide quantum materials.

Doping allows us to modify semiconductor materials for desired electrical and optical properties. The solubility limit is a fundamental barrier for dopants incorporated into a specific semiconductor. Ultra-doping or hyper-doping refers to doping a semiconductor much beyond the corresponding solid solubility limit and often results in exotic properties. In this talk, we show that ion implantation combined with flash lamp annealing in millisecond and pulsed laser melting in nanosecond can realize ultra-doping in widely used semiconductors, including Si [1-5], Ge [6-8] and GaAs [9]. Various dopants, from conventional shallow-level impurities to deep-level ones, can be substitutionally incorporated up to a few atomic percent. This leads to the insulator-to-metal transition and the large modification to the semiconductor bandgap. The ultra-doped semiconductors can be used as photodetectors and plasmonic elements [10]. Ion implantation followed by annealing is a well-established method to dope Si, being maturely integrated with the IC industry production line. Therefore, ultra-doped semiconductors can be a wafer-scale platform for photonics.

[1] M. Wang, et al., Phys. Rev. Applied. 10, 024054 (2018)
[2] M. Wang, et al., Phys. Rev. Applied. 11, 054039 (2019)
[3] M. Wang, et al., Phys. Rev. B 102, 085204 (2020)
[4] M. Wang, et al., Adv. Optical Mater. 9, 2001546 (2021)
[5] M. Wang, et al., Nanoscale, 14, 2826-2836 (2022)
[6] S. Prucnal, et al., Scientific Reports 6, 27643 (2016)
[7] S. Prucnal, et al., Phys. Rev. Materials 3, 054802 (2019)
[8] S. Prucnal, et al., New J. Phys. 22, 123036 (2020)
[9] J. Duan, et al., New J. Phys. 23, 083034 (2021)
[10] G. V. Naik, V.M. Shalaev, A. Boltasseva, Alternative Plasmonic Materials: Beyond Gold and Silver, Adv. Mater. 25, 3264 (2013).

Hyperspectral images provide rich spectral information corresponding to visible and near-infrared imaging regions, facilitating accurate classification, object identification, and target detection. However, the high volume of data creates a computational challenge in processing. The band selection process identifies specific informative and discriminative spectral bands from the data to speed up the processing without impeding the performance. This article presents an application-independent band selection framework that utilizes improved sparse deep subspace clustering and introduces an efficient multicriteria-based representative band selection (BS). The proposed sparse deep subspace clustering approach efficiently identifies the underlying nonlinear subspace structure of the data and organizes the data accordingly. The work introduces a novel, robust sparsity measure to obtain a powerful self-representation and ameliorated performance compared to the prevalent subspace clustering methods. The work subsequently selects the representative bands from each cluster by combining structural information of the band images with the statistical similarity measure. We evaluate the BS performance on standard real images using information-theoretic criterion, classification, and unmixing performance. The comparative performance assessment demonstrates that our proposed method identifies the informative bands and outperforms the other approaches in terms of the subsequent tasks.

Tellurium is one of the deep-level impurities in Si, leading to states of 200-400 meV below the conduction band. Non-equilibrium methods allow for doping deep-level impurities in Si well above the solubility limit, referred as hyperdoping, that can result in exotic properties, such as extrinsic photo-absorption well below the Si bandgap [1]. In this contribution, we will present an overview about Te hyperdoped Si. The hyperdoping is realized by ion implantation and pulsed laser melting. We will present the resulting optical, electrical properties and the perspective applications as infrared photodetectors. With increasing the Te concentration, the samples undergo an insulator to metal transition [2,3]. The metallic phase is governed by a power law dependence of the conductivity at temperatures below 25 K, whereas the conductivity in the insulating phase is well described by a variable-range hopping mechanism with a Coulomb gap. Surprisingly, the electron concentration obtained in Te-hyperdoped Si is approaching 10 21 cm-3 and does not show saturation [4]. It is even high than that of P or As doped Si and potentially meets the criteria of source/drain applications in future nanoelectronics. The infrared optical absorptance is found to increase with increasing dopant concentration [2]. We demonstrate the room-temperature operation of a mid-infrared photodetector based on Te-hyperdoped Si. The key parameters, such as the detectivity, the bandwidth and the rise/fall time, show competitiveness with the commercial products [5]. To
understand the microscopic picture, we have performed Rutherford backscattering/channeling angular scans and first-princiles calcluations [4]. The Te-dimer complex sitting on adjacent Si lattice sites has the smallest formation energy and is thus the preferred configuration at high doping concentration. Those substitutional Te-dimers are effective donors, leading to the insulator-to-metal transition, the non-saturating carrier concentration as well as the sub-band photoresponse. Moreover, the Te-hyperdoped Si layers exhibit thermal stability up to 400 °C with a duration of at least 10 minutes [6]. Therefore, Te-hyperdoped Si presents a test-bed for electrical and optical applications utilizing deep-level impurities.
[1] J. M. Warrender, Laser hyperdoping silicon for enhanced infrared optoelectronic properties, Appl. Phys.Rev. 3, 031104 (2016).
[2] M. Wang, ..., S. Zhou, Extended Infrared Photoresponse in Te-Hyperdoped Si at Room Temperature, Phys.Rev. Appl. 10, 024054 (2018).
[3] M. Wang, ..., S. Zhou, Critical behavior of the insulator-to-metal transition in Te-hyperdoped Si, Phys. Rev.B 102, 085204 (2020).
[4] M. Wang, ... Breaking the doping limit in silicon by deep impurities, Phys. Rev. Appl. 11, 054039 (2019).
[5] M. Wang, ..., S. Zhou, Silicon-Based Intermediate-Band Infrared Photodetector Realized by Te Hyperdoping, Adv. Opt. Mater. 9, 2001546, (2020).
[6] M. Wang, ..., S. Zhou, Thermal stability of Te-hyperdoped Si: Atomic-scale correlation of the structural, electrical, and optical properties, Phys. Rev. Mater. 3, 044606 (2019).

Deep subspace clustering (DSC) has achieved remarkable performances in the unsupervised classification of hyperspectral images. However, previous models based on pixel-level self-expressiveness of data suffer from the exponential growth of computational complexity and access memory requirements with an increasing number of samples, thus leading to poor applicability to large hyperspectral images. This article presents a neighborhood contrastive subspace clustering (NCSC) network, a scalable and robust DSC approach, for unsupervised classification of large hyperspectral images. Instead of using a conventional autoencoder, we devise a novel superpixel pooling autoencoder to learn the superpixel-level latent representation and subspace, allowing a contracted self-expressive layer. To encourage a robust subspace representation, we propose a novel neighborhood contrastive regularization to maximize the agreement between positive samples in subspace. We jointly train the resulting model in an end-to-end fashion by optimizing an adaptively weighted multitask loss. Extensive experiments on three hyperspectral benchmarks demonstrate the effectiveness of the proposed approach and its substantial advancement of state-of-the-art approaches.

Deep neural networks (DNNs) show impressive performance for hyperspectral image (HSI) classification when abundant labeled samples are available. The problem is that HSI sample annotation is extremely costly and the budget for this task is usually limited. To reduce the reliance on labeled samples, deep semisupervised learning (SSL), which jointly learns from labeled and unlabeled samples, has been introduced in the literature. However, learning robust and discriminative features from unlabeled data is a challenging task due to various noise effects and ambiguity of unlabeled samples. As a result, recent advances are constrained, mainly in the pretraining or warm-up stage. In this article, we propose a deep probabilistic framework to generate reliable pseudo-labels to explicitly learn discriminative features from unlabeled samples. The generated pseudo-labels of our proposed framework can be fed to various DNNs to improve their generalization capacity. Our proposed framework takes only ten labeled samples per class to represent the label set as an uncertainty-aware distribution (We use the Gaussian distribution to represent the uncertainty of the label set in the latent space.) in the latent space. The pseudo-labels are then generated for those unlabeled samples whose feature values match the distribution with high probability. By performing extensive experiments on four publicly available datasets, we show that our framework can generate reliable pseudo-labels to significantly improve the generalization capacity of several state-of-the-art DNNs. In addition, we introduce a new DNN for HSI classification that demonstrates outstanding accuracy results in comparison with its rivals.

Doping allows us to modify semiconductor materials for desired electrical, optical and magnetic properties. The solubility limit is a fundamental barrier for dopants incorporated into a specific semiconductor. Hyperdoping refers to doping a semiconductor much beyond the corresponding solid solubility limit and often results in exotic properties. In this talk, we show that ion implantation combined with flash lamp annealing in millisecond and pulsed laser melting in nanosecond can be a versatile approach to fabricate hyperdoped semiconductors. Mn hyperdoped III-V compound semiconductors become ferromagnetic, where Mn impurities are in 2+ valence and providing local moments of 5 μB and free holes [1-5]. Their Curie temperatures can be varied either by the Mn or free hole concentration, and also depend on the host semiconductors, which reveal different pd exchange strength. On the other hand, Ga and Al hyperdoped Ge exhibits superconductivity with controllable critical temperature [6, 7]. In combination with first-principles calculation, phonon-mediated superconductivity is counted for the mechanism. The critical-field reveals significant difference when the field is in-plane or out-of-plane. This remarkable anisotropy may be considered as proof that Ga is incorporated in the Ge matrix homogeneously in a thin layer. Ion implantation followed by annealing is a well-established method to dope Si and Ge, being maturely integrated with the IC industry production line. We propose ferromagnetic and superconducting semiconductors prepared by ion implantation can be a scalable platform for quantum technology.

[1] M. Khalid, et al., Phys. Rev. B 89, 121301(R) (2014).
[2] S. Zhou, J. Phys. D: Appl. Phys. 48, 263001(2015).
[3] S. Prucnal, et al., Phys. Rev. B 92, 222407 (2015).
[4] Y. Yuan, et al., ACS Appl. Mater. Interfaces, 8, 3912 (2016).
[5] Y. Yuan, et al., Phys. Rev. Mater. 1, 054401 (2017).
[6] T. Herrmannsdörfer, et al., Phys. Rev. Lett. 102, 217003 (2009).
[7] S. Prucnal, et al., Phys. Rev. Materials 3 054802 (2019).

Generative adversarial networks (GANs) have achieved many excellent results in hyperspectral image (HSI) classification in recent years, as GANs can effectively solve the dilemma of limited training samples in HSI classification. However, due to the class imbalance problem of HSI data, GANs always associate minority-class samples with fake label. To address this issue, we first propose a semisupervised generative adversarial network incorporating a transformer, called HyperViTGAN. The proposed HyperViTGAN is designed with an external semisupervised classifier to avoid self-contradiction when the discriminator performs both classification and discrimination tasks. The generator and discriminator with skip connection are utilized to generate HSI patches by adversarial learning. The proposed HyperViTGAN captures semantic context and low-level textures to reduce the loss of critical information. In addition, the generalization ability of the HyperViTGAN is improved through the use of data augmentation. Experimental results on three well-known HSI datasets, Houston 2013, Indian Pines 2010, and Xuzhou, show that the proposed model achieves competitive HSI classification performance in comparison with the current state-of-the-art classification models.

Doping allows us to modify semiconductor materials for desired electrical, optical and magnetic properties. The solubility limit is a fundamental barrier for dopants incorporated into a specific semiconductor. Hyperdoping refers to doping a semiconductor much beyond the corresponding solid solubility limit and often results in exotic properties. For example, Ga hyperdoped Ge reveals superconductivity and Mn hyperdoped GaAs represents a typical ferromagnetic semiconductor. Ion implantation followed by annealing is a well-established method to dope Si and Ge. This approach has been maturely integrated with the IC industry production line. However, being applied to hyperdoping, the annealing duration has to be shortenedto millisecond or even nanosecond. The intrinsic physical parameters related to dopants and semiconductors (e.g. Solubility, diffusivity, melting point and thermal conductivity) have to be considered to choose the right annealing time regime. In this talk, we propose that ion implantation combined with flash lamp annealing in millisecond and pulsed laser melting in nanosecond can be a versatile approach to fabricate hyperdoped semiconductors. The examples include magnetic semiconductors [1-5], superconducting Ge [6] and chalcogen doped Si [10-12].
[1] M. Khalid, et al., Phys. Rev. B 89, 121301(R) (2014).
[2] S. Zhou, J. Phys. D: Appl. Phys. 48, 263001(2015).
[3] S. Prucnal, et al., Phys. Rev. B 92, 222407 (2015).
[4] Y. Yuan, et al., ACS Appl. Mater. Interfaces, 8, 3912 (2016).
[5] Y. Yuan, et al., Phys. Rev. Mater. 1, 054401 (2017).
[6] S. Prucnal, et al., Phys. Rev. Materials 3, 054802 (2019).
[7] S. Prucnal, et al., New J. Phys., 22 123036 (2020).
[8] M. Wang, et al., Phys. Rev. Applied. 10, 024054 (2018).
[9] M. Wang, et al., Phys. Rev. Applied. 11, 054039 (2019).
[10] M. Wang, et al., Nanoscale 14, 2826 (2022).

Accurate classification of remote sensing (RS) images is a perennial topic of interest in the RS community. Recently, transfer learning, especially for fine-tuning pretrained convolutional neural networks (CNNs), has been proposed as a feasible strategy for RS scene classification. However, because the target domain (i.e., the RS images) and the source domain (e.g., ImageNet) are quite different, simply using the model pretrained on an ImageNet dataset presents some difficulties. The RS images and the pretrained models need to be properly adjusted to build a better classification system. In this study, an adaptive learning strategy for transferring a CNN-based model is proposed. First, an adaptive transform is used to adjust the original size of the RS image to a certain size, which is tailored to the input of the subsequent pretrained model. Then, an adaptive transferring model is proposed to automatically learn what knowledge from the pretrained model should be transferred to the RS scene classification model. Finally, in combination with a label smoothing approach, an adaptive label is presented to generate soft labels based on the statistics of the classification model predictions for each category, which is beneficial for learning the relationships between the target and nontarget categories of scenes. In general, the proposed methods adaptively manage the input, model, and label simultaneously, which leads to better classification performance for RS scene classification. The proposed methods are tested on three widely used datasets, and the obtained results show that the proposed methods provide competitive classification accuracy compared to the state-of-the-art methods.

Hyperspectral images (HSIs) are usually corrupted by various types of mixed noises, which degrades the qualities of acquired images and limits the subsequent applications. In this article, we propose a novel denoised method based on a hybrid spatial–spectral total variation (SSTV) regularization, which we refer to as l0 - l1−2 SSTV. Specifically, l0 - l1−2 SSTV can be treated as an integrated regularization embedding spatial–spectral l0 gradient model into l1−2 SSTV. l1−2 SSTV regularization exploits the sparse structures in both spatial and spectral domains. Hence, the correlations within HSIs are fully considered. Due to the good performance of l1−2 -norm in image restoration, l1−2 SSTV gives a tighter approximation for the gradient domains of HSIs. It can effectively avoid artifacts and oversmoothing caused by the limitation of the SSTV regularization based on l1 -norm ( l1 SSTV). Meanwhile, the l0 gradient regularization controls the number of nonzero gradients to promote the local piecewise smoothness, making denoised images preserve clear edges. With the effective combination of l1−2 SSTV and l0 gradient regularization, l0 - l1−2 SSTV produces high-quality restoration results in the denoising process: better detail preservation and sharper edges. The augmented Lagrangian method and the difference of convex algorithm are exploited to optimize the proposed model. The results for simulated and real experiments demonstrate the effectiveness and superiority of the proposed method compared with state-of-the-art methods.

In recent years, supervised learning has been widely used in various tasks of optical remote sensing image (RSI) understanding, including RSI classification, pixel-wise segmentation, change detection, and object detection. The methods based on supervised learning need a large amount of high-quality training data, and their performance highly depends on the quality of the labels. However, in practical remote sensing applications, it is often expensive and time consuming to obtain large-scale data sets with high-quality labels, which leads to a lack of sufficient supervised information. In some cases, only coarse-grained labels can be obtained, resulting in the lack of exact supervision. In addition, the supervised information obtained manually may be wrong, resulting in a lack of accurate supervision. Therefore, RSI understanding often faces the problems of incomplete, inexact, and inaccurate supervised information, which will affect the breadth and depth of remote sensing applications.

Although chronic inflammation inhibits bone healing, the healing process is initiated by an inflammatory phase. In a well-tuned sequence of molecular events, pro-inflammatory cytokines
are secreted to orchestrate the inflammation response to injury and the recruitment of progenitor cells. These events in turn activate the secretion of anti-inflammatory signaling molecules and attract cells and mediators that antagonize the inflammation and initiate the repair phase. Sulfated glycosaminoglycanes (sGAG) are known to interact with cytokines, chemokines and growth factors and, thus, alter the availability, duration and impact of those mediators on the local molecular level. sGAG-coated polycaprolactone-co-lactide (PCL) scaffolds were inserted into critical-size femur defects in adult male Wistar rats. The femur was stabilized with a plate, and the defect was filled with either sGAG-containing PCL scaffolds or autologous bone (positive control). Wound fluid samples obtained by microdialysis were characterized regarding alterations of cytokine concentrations over the first 24 h after surgery. The analyses revealed the inhibition of the pro-inflammatory cytokines IL-1β and MIP-2 in the sGAG-treated groups compared to the positive control. A simultaneous increase of IL-6 and TNF-α indicated advanced regenerative capacity of sGAG, suggesting their potential to improve bone healing.

The use of deep learning for water extraction requires precise pixel-level labels. However, it is very difficult to label high-resolution remote-sensing images at the pixel level. Therefore, we study how to utilize point labels to extract water bodies and propose a novel method called the neighbor feature aggregation network (NFANet). Compared with pixel-level labels, point labels are much easier to obtain, but they will lose much information. In this article, we take advantage of the similarity between the adjacent pixels of a local water body, and propose a neighbor sampler to resample remote-sensing images. Then, the sampled images are sent to the network for feature aggregation. In addition, we use an improved recursive training algorithm to further improve the extraction accuracy, making the water boundary more natural. Furthermore, our method utilizes neighboring features instead of global or local features to learn more representative features. The experimental results show that the proposed NFANet method not only outperforms other studied weakly supervised approaches, but also obtains similar results as the state-of-the-art ones.

Forests can be efficiently monitored by automatic semantic segmentation of trees using satellite and/or aerial images. Still, several challenges can make the problem difficult, including the varying spectral signature of different trees, lack of sufficient labelled data, and geometrical occlusions. In this article, we address the tree segmentation problem using multispectral imagery. While we carry out large-scale experiments on several deep learning architectures using various spectral input combinations, we also attempt to explore whether hand-crafted spectral vegetation indices can improve the performance of deep learning models in the segmentation of trees. Our experiments include benchmarking a variety of multispectral remote sensing image sets, deep semantic segmentation architectures, and various spectral bands as inputs, including a number of hand-crafted spectral vegetation indices. From our large-scale experiments, we draw several useful conclusions. One particularly important conclusion is that, with no additional computation burden, combining different categories of multispectral vegetation indices, such as NVDI, atmospherically resistant vegetation index, and soil-adjusted vegetation index, within a single three-channel input, and using the state-of-the-art semantic segmentation architectures, tree segmentation accuracy can be improved under certain conditions, compared to using high-resolution visible and/or near-infrared input.

In recent years, multiview subspace learning has been garnering increasing attention. It aims to capture the inner relationships of the data that are collected from multiple sources by learning a unified representation. In this way, comprehensive information from multiple views is shared and preserved for the generalization processes. As a special branch of temporal series hyperspectral image (HSI) processing, the anomalous change detection (ACD) task focuses on detecting very small changes among different temporal images. However, when the volume of datasets is very large or the classes are relatively comprehensive, the existing methods may fail to find those changes between the scenes, and end up with terrible detection results. In this article, inspired by the sketched representation and multiview subspace learning, a sketched multiview subspace learning (SMSL) model is proposed for HSI ACD. The proposed model preserves major information from the image pairs and improves the computational complexity using a sketched representation matrix. Furthermore, the differences between scenes are extracted using the specific regularizer of the self-representation matrices. To evaluate the detection effectiveness of the proposed SMSL model, experiments are conducted on a benchmark hyperspectral remote sensing dataset and a natural hyperspectral dataset and compared with other state-of-the-art approaches. The code of the proposed method will be available at https://github.com/ShizhenChang/SMSL .

Semantic segmentation methods based on deep neural networks have achieved great success in recent years. However, training such deep neural networks relies heavily on a large number of images with accurate pixel-level labels, which requires a huge amount of human effort, especially for large-scale remote sensing images. In this paper, we propose a point-based weakly supervised learning framework called the deep bilateral filtering network (DBFNet) for the semantic segmentation of remote sensing images. Compared with pixel-level labels, point annotations are usually sparse and cannot reveal the complete structure of the objects; they also lack boundary information, thus resulting in incomplete prediction within the object and the loss of object boundaries. To address these problems, we incorporate the bilateral filtering technique into deeply learned representations in two respects. First, since a target object contains smooth regions that always belong to the same category, we perform deep bilateral filtering (DBF) to filter the deep features by a nonlinear combination of nearby feature values, which encourages the nearby and similar features to become closer, thus achieving a consistent prediction in the smooth region. In addition, the DBF can distinguish the boundary by enlarging the distance between the features on different sides of the edge, thus preserving the boundary information well. Experimental results on two widely used datasets, the ISPRS 2-D semantic labeling Potsdam and Vaihingen datasets, demonstrate that our proposed DBFNet can achieve a highly competitive performance compared with state-of-the-art fully-supervised methods. Code is available at https://github.com/Luffy03/DBFNet .

The scientific outcomes of the 2022 Landslide4Sense (L4S) competition organized by the Institute of Advanced Research in Artificial Intelligence are presented here. The objective of the competition is to automatically detect landslides based on large-scale multiple sources of satellite imagery collected globally. The 2022 L4S aims to foster interdisciplinary research on recent developments in deep learning (DL) models for the semantic segmentation task using satellite imagery. Over the past few years, DL-based models have achieved performance that meets expectations on image interpretation due to the development of convolutional neural networks. The main objective of this article is to present the details and the best-performing algorithms featured in this competition. The winning solutions are elaborated with state-of-the-art models, such as the Swin Transformer, SegFormer, and U-Net. Advanced machine learning techniques and strategies, such as hard example mining, self-training, and mix-up data augmentation, are also considered. Moreover, we describe the L4S benchmark dataset in order to facilitate further comparisons and report the results of the accuracy assessment online. The data are accessible on Future Development Leaderboard for future evaluation at https://www.iarai.ac.at/landslide4sense/challenge/ , and researchers are invited to submit more prediction results, evaluate the accuracy of their methods, compare them with those of other users, and, ideally, improve the landslide detection results reported in this article.

Deep learning algorithms have obtained great success in semantic segmentation of very high-resolution (VHR) remote sensing images. Nevertheless, training these models generally requires a large amount of accurate pixel-wise annotations, which is very laborious and time-consuming to collect. To reduce the annotation burden, this paper proposes a consistency-regularized region-growing network (CRGNet) to achieve semantic segmentation of VHR remote sensing images with point-level annotations. The key idea of CRGNet is to iteratively select unlabeled pixels with high confidence to expand the annotated area from the original sparse points. However, since there may exist some errors and noises in the expanded annotations, directly learning from them may mislead the training of the network. To this end, we further propose the consistency regularization strategy, where a base classifier and an expanded classifier are employed. Specifically, the base classifier is supervised by the original sparse annotations, while the expanded classifier aims to learn from the expanded annotations generated by the base classifier with the region-growing mechanism. The consistency regularization is thereby achieved by minimizing the discrepancy between the predictions from both the base and the expanded classifiers. We find such a simple regularization strategy is yet very useful to control the quality of the region-growing mechanism. Extensive experiments on two benchmark datasets demonstrate that the proposed CRGNet significantly outperforms the existing state-of-the-art methods. Codes and pre-trained models are available online ( https://github.com/YonghaoXu/CRGNet ).