Zeitschriftenaufsätze

Analyzing time series data like EEG or MEG is challenging due to noisy, high-dimensional, and patient-specific signals. Deep learning methods have been demonstrated to be superior in analyzing time series data compared to shallow learning methods which utilize handcrafted and often subjective features. Especially, recurrent deep neural networks (RNN) are considered suitable to analyze such continuous data. However, previous studies show that they are computationally expensive and difficult to train. In contrast, feed-forward networks (FFN) have previously mostly been considered in combination with hand-crafted and problem-specific feature extractions, such as short time Fourier and discrete wavelet transform. A sought-after are easily applicable methods that efficiently analyze raw data to remove the need for problem-specific adaptations. In this work, we systematically compare RNN and FFN topologies as well as advanced architectural concepts on multiple datasets with the same data preprocessing pipeline. We examine the behavior of those approaches to provide an update and guideline for researchers who deal with automated analysis of EEG time series data. To ensure that the results are meaningful, it is important to compare the presented approaches while keeping the same experimental setup, which to our knowledge was never done before. This paper is a first step toward a fairer comparison of different methodologies with EEG time series data. Our results indicate that a recurrent LSTM architecture with attention performs best on less complex tasks, while the temporal convolutional network (TCN) outperforms all the recurrent architectures on the most complex dataset yielding a 8.61% accuracy improvement. In general, we found the attention mechanism to substantially improve classification results of RNNs. Toward a light-weight and online learning-ready approach, we found extreme learning machines (ELM) to yield comparable results for the less complex tasks.

Interfaces are key elements that define electronic properties of the final device. Inevitably, most of the active interfaces of III-V semiconductor devices are buried and it is therefore not straightforward to characterize them. The Tapered Cross Section Photoelectron Spectroscopy (TCS-PES) approach is promising to address such a challenge. That the TCS-PES can be used to study the relevant heterojunction in epitaxial III-V architectures prepared by metalorganic chemical vapor deposition is demonstrated here. A MULTIPREP polishing system that enables controlling the angle between the sample holder and the polishing plate has been employed to improve the reproducibility of the polishing procedure. With this procedure, that preparing the TCS of III-V semiconductor devices with tapering angles lower than 0.02˚ is possible is demonstrated. The PES provides then information about the buried interfaces of Ge|GaInP and GaAs|GaInP layer stacks. Both, chemical and electronic properties have been measured by PES. It evidences that the preparation of the TCSs under an uncontrolled atmosphere modifies the pristine properties of the critical buried heterointerfaces. Surface states and reaction layers are created on the TCS surface, which restrict unambiguous conclusions on buried interface energetics.

Brake-wear particle emissions are the result of the components of a friction brake being in tribological contact, and they are classified as non-exhaust emissions. Since most of the emitted particles belong to the size classes of particulate matter (≤10 μm) and differ significantly in terms of their physico-chemical properties from automotive exhaust emissions, this source is of particular relevance to human health and, therefore, the focus of scientific studies. Previous studies have shown that coated brake discs offer significant wear and emission reduction potential. Nevertheless, no studies are available that describe the specific particle formation process, the contact conditions, the structure of the friction layer and the differences compared to conventional grey cast iron discs. The aim of this study is to describe those differences. For this purpose, the tribological behaviour, the structure of the friction layer and the associated particle dynamics within the friction contact between a laser cladding coated disc and a conventional grey cast iron disc are compared. The required investigations are carried out both ex situ (stationary) and in situ (dynamic). Parallel to the tribological investigations, the particle emission behaviour is determined on an inertia dynamometer using a constant volume sampling system (CVS) and equipment for particle number and particle size distribution measurement. The results show that, for two different brake pads, the laser cladding brake disc has lower wear and less particulate emissions than the grey cast iron brake disc. The wear behaviour of the coating varies significantly for the two brake pads. By contrast, the grey cast iron brake disc shows a significantly lower influence.

Peralkaline Sm3+-doped aluminosilicate glasses with different network modifier ions (Mg2+, Ca2+, Sr2+, Ba2+, Zn2+) were investigated to clarify the effect of glass composition and glass structure on the optical properties of the doped Sm3+ ions. For this purpose, the Sm3+ luminescence emission spectra were correlated with the molecular structure of the glasses derived by molecular dynamics (MD) simulations. The different network modifier ions have a clear and systematic effect on the peak area ratio of the Sm3+ emission peaks which correlates with the average rare earth site symmetry in the glasses. The highest site symmetry is found for the calcium aluminosilicate glass. Glasses with network modifier ions of lower and higher ionic radii show a notably lower average site symmetry. The symmetry could be correlated to the rare earth coordination number with oxygen atoms derived by MD simulations. A coordination number of 6 seems to offer the highest average site symmetry. Higher rare earth coordination probabilities with non-bridging oxygen result in an increased splitting of the emission peaks and a notable broadening of the peaks. The zinc containing glass seems to play a special role. The Zn2+ ions notably modify the glass structure and especially the rare earth coordination in comparison to the other network modifier ions in the other investigated glasses. The knowledge on how glass structure affects the optical properties of doped rare earth ions can be used to tailor the rare earth absorption and emission spectra for specific applications.

Engineering core-shell materials with rationally designed architectures and components is an effective strategy to fulfill the high-performance requirements of supercapacitors. Herein, hierarchical candied-haws-like NiCo2S4NiCo(HCO3)2 core-shell heterostructure (NiCo2S4@HCs) is designed with NiCo(HCO3)2 polyhedrons being tightly strung by cross-linked NiCo2S4 nanowires. This rational design not only creates more electroactive sites but also suppresses the volume expansion during the charge-discharge processes. Meanwhile, density functional theory calculations ascertain that the formation of NiCo2S4@HCs heterostructure simultaneously facilitates OH− adsorption/desorption and accelerates electron transfer within the electrode, boosting fast and efficient redox reactions. Ex situ X-ray diffraction and Raman measurements reveal that gradual phase transformations from NiCo(HCO3)2 to NiCo(OH)2CO3 and then to highly-active NiCoOOH take place during the cycles. Therefore, NiCo2S4@HCs demonstrates an ultrahigh capacitance of 3178.2 F g−1 at 1 A g−1 and a remarkable rate capability of 2179.3 F g−1 at 30 A g−1. In addition, the asymmetric supercapacitor NiCo2S4@HCs//AC exhibits a high energy density of 69.6 W h kg−1 at the power density of 847 W kg−1 and excellent cycling stability with 90.2% retained capacitance after 10 000 cycles. Therefore, this novel structural design has effectively manipulated the interface charge states and guaranteed the structural integrity of electrode materials to achieve superior electrochemical performances.