arXiv subjects

Philipp Wagner

Publications and source records attributed to Philipp Wagner.

15 recordsLinked to original sources

arXiv2025

CsCl seed layer homogenizes co-evaporated perovskite growth for high-efficiency fully textured perovskite-silicon tandem solar cells

Monolithic perovskite-silicon tandem solar cells experienced a significant increase in efficiency, making them viable for industrial applications. Among the various scalable and industry-compatible metal halide perovskite deposition techniques, co-evaporation stands out as particularly well-suited for perovskite-silicon tandem solar cells due to its ability to conformally cover textured silicon bottom cells. Solution-processed [2-(3,6-Dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid (MeO-2PACz) is commonly used as a hole-transporting material for co-evaporated metal halide perovskites. However, we show that it covers the textured surface of silicon bottom cells unevenly, impacting the film growth and leading to the formation of residual PbI2 at the buried interface. The present study reveals via X-ray photoemission electron microscopy (XPEEM) and infrared scattering-type scanning near-field optical microscope (IR s-SNOM) that a CsCl seed layer fosters organic precursor incorporation across the MeO-2PACz/perovskite interface, even in the areas with a thin MeO-2PACz layer, thereby preventing the formation of interfacial PbI2 and leading to larger apparent grains. The improvement of the metal halide perovskite film quality on the MeO-2PACz/perovskite interface and the bulk perovskite film led to 30.3% (29.7% certified) efficient perovskite-silicon tandem solar cell. The present work highlights the importance of a seed layer for a robust growth of co-evaporated metal halide perovskite and represents an important milestone for the transfer of perovskite-silicon tandem solar cells from laboratory to industry.

cond-mat.mtrl-sci↗

arXiv2025

Sample-Efficient Bayesian Transfer Learning for Online Machine Parameter Optimization

Correctly setting the parameters of a production machine is essential to improve product quality, increase efficiency, and reduce production costs while also supporting sustainability goals. Identifying optimal parameters involves an iterative process of producing an object and evaluating its quality. Minimizing the number of iterations is, therefore, desirable to reduce the costs associated with unsuccessful attempts. This work introduces a method to optimize the machine parameters in the system itself using a Bayesian optimization algorithm. By leveraging existing machine data, we use a transfer learning approach in order to identify an optimum with minimal iterations, resulting in a cost-effective transfer learning algorithm. We validate our approach on a laser machine for cutting sheet metal in the real world.

cs.LG↗

arXiv2024

Audio-based Step-count Estimation for Running -- Windowing and Neural Network Baselines

In recent decades, running has become an increasingly popular pastime activity due to its accessibility, ease of practice, and anticipated health benefits. However, the risk of running-related injuries is substantial for runners of different experience levels. Several common forms of injuries result from overuse -- extending beyond the recommended running time and intensity. Recently, audio-based tracking has emerged as yet another modality for monitoring running behaviour and performance, with previous studies largely concentrating on predicting runner fatigue. In this work, we investigate audio-based step count estimation during outdoor running, achieving a mean absolute error of 1.098 in window-based step-count differences and a Pearson correlation coefficient of 0.479 when predicting the number of steps in a 5-second window of audio. Our work thus showcases the feasibility of audio-based monitoring for estimating important physiological variables and lays the foundations for further utilising audio sensors for a more thorough characterisation of runner behaviour.

cs.SD↗

arXiv2023

Model Reporting for Certifiable AI: A Proposal from Merging EU Regulation into AI Development

Despite large progress in Explainable and Safe AI, practitioners suffer from a lack of regulation and standards for AI safety. In this work we merge recent regulation efforts by the European Union and first proposals for AI guidelines with recent trends in research: data and model cards. We propose the use of standardized cards to document AI applications throughout the development process. Our main contribution is the introduction of use-case and operation cards, along with updates for data and model cards to cope with regulatory requirements. We reference both recent research as well as the source of the regulation in our cards and provide references to additional support material and toolboxes whenever possible. The goal is to design cards that help practitioners develop safe AI systems throughout the development process, while enabling efficient third-party auditing of AI applications, being easy to understand, and building trust in the system. Our work incorporates insights from interviews with certification experts as well as developers and individuals working with the developed AI applications.

cs.AI↗

arXiv2021

Kalman Bayesian Neural Networks for Closed-form Online Learning

Compared to point estimates calculated by standard neural networks, Bayesian neural networks (BNN) provide probability distributions over the output predictions and model parameters, i.e., the weights. Training the weight distribution of a BNN, however, is more involved due to the intractability of the underlying Bayesian inference problem and thus, requires efficient approximations. In this paper, we propose a novel approach for BNN learning via closed-form Bayesian inference. For this purpose, the calculation of the predictive distribution of the output and the update of the weight distribution are treated as Bayesian filtering and smoothing problems, where the weights are modeled as Gaussian random variables. This allows closed-form expressions for training the network's parameters in a sequential/online fashion without gradient descent. We demonstrate our method on several UCI datasets and compare it to the state of the art.

cs.LG↗

arXiv2017

Probing Convolutional Neural Networks for Event Reconstruction in γ-Ray Astronomy with Cherenkov Telescopes

A dramatic progress in the field of computer vision has been made in recent years by applying deep learning techniques. State-of-the-art performance in image recognition is thereby reached with Convolutional Neural Networks (CNNs). CNNs are a powerful class of artificial neural networks, characterized by requiring fewer connections and free parameters than traditional neural networks and exploiting spatial symmetries in the input data. Moreover, CNNs have the ability to automatically extract general characteristic features from data sets and create abstract data representations which can perform very robust predictions. This suggests that experiments using Cherenkov telescopes could harness these powerful machine learning algorithms to improve the analysis of particle-induced air-showers, where the properties of primary shower particles are reconstructed from shower images recorded by the telescopes. In this work, we present initial results of a CNN-based analysis for background rejection and shower reconstruction, utilizing simulation data from the H.E.S.S. experiment. We concentrate on supervised training methods and outline the influence of image sampling on the performance of the CNN-model predictions.

astro-ph.IM↗

arXiv2017

DiaSys: Improving SoC Insight Through On-Chip Diagnosis

To find the cause of a functional or non-functional defect (bug) in software running on a multi-processor System-on-Chip (MPSoC), developers need insight into the chip. Tracing systems provide this insight non-intrusively, at the cost of high off-chip bandwidth requirements. This I/O bottleneck limits the observability, a problem becoming more severe as more functionality is integrated on-chip. In this paper, we present DiaSys, an MPSoC diagnosis system with the potential to replace today's tracing systems. Its main idea is to partially execute the analysis of observation data on the chip; in consequence, more information and less data is sent to the attached host PC. With DiaSys, the data analysis is performed by the diagnosis application. Its input are events, which are generated by observation hardware at interesting points in the program execution (like a function call). Its outputs are events with higher information density. The event transformation is modeled as dataflow application. For execution, it is mapped in part to dedicated and distributed on-chip components, and in part to the host PC; the off-chip boundary is transparent to the developer of the diagnosis application. We implement DiaSys as extension to an existing SoC with four tiles and a mesh network running on an FPGA platform. Two usage examples confirm that DiaSys is flexible enough to replace a tracing system, while significantly lowering the off-chip bandwidth requirements. In our examples, the debugging of a race-condition bug, and the creation of a lock contention profile, we see a reduction of trace bandwidth of more than three orders of magnitude, compared to a full trace created by a common tracing system.

cs.DC↗

arXiv2013

Stable hydrogenated graphene edge types: Normal and reconstructed Klein edges

Hydrogenated graphene edges are assumed to be either armchair, zigzag or a combination of the two. We show that the zigzag is not the most stable fully hydrogenated structure along the <2-1-10> direction. Instead hydrogenated Klein and reconstructed Klein based edges are found to be energetically more favourable, with stabilities approaching that of armchair edges. These new structures "unify" graphene edge topology, the most stable flat hydrogenated graphene edges always consisting of pairwise bonded C2H4 edge groups, irrespective the edge orientation. When edge rippling is included, CH3 edge groups are most stable. These new fundamental hydrogen terminated edges have important implications for graphene edge imaging and spectroscopy, as well as mechanisms for graphene growth, nanotube cutting, and nanoribbon formation and behaviour.

cond-mat.mes-hall↗

arXiv2013

Open Tiled Manycore System-on-Chip

Manycore System-on-Chip include an increasing amount of processing elements and have become an important research topic for improvements of both hardware and software. While research can be conducted using system simulators, prototyping requires a variety of components and is very time consuming. With the Open Tiled Manycore System-on-Chip (OpTiMSoC) we aim at building such an environment for use in our and other research projects as prototyping platform. This paper describes the project goals and aspects of OpTiMSoC and summarizes the current status and ideas.

cs.AR↗

arXiv2013

Mechanical properties of nanosheets and nanotubes investigated using a new geometry independent volume definition

Cross-sectional area and volume become difficult to define as material dimensions approach the atomic scale. This limits the transferability of macroscopic concepts such as Young's modulus. We propose a new volume definition where the enclosed nanosheet or nanotube average electron density matches that of the parent layered bulk material. We calculate the Young's moduli for various nanosheets (including graphene, BN and MoS2) and nanotubes. Further implications of this new volume definition such as a Fermi level dependent Young's modulus and out-of-plane Poisson's ratio are shown.

cond-mat.mes-hall↗

arXiv2012

Graphene edge structures: Folding, scrolling, tubing, rippling and twisting

Conventional three-dimensional crystal lattices are terminated by surfaces, which can demonstrate complex rebonding and rehybridisation, localised strain and dislocation formation. Two dimensional crystal lattices, of which graphene is the archetype, are terminated by lines. The additional available dimension at such interfaces opens up a range of new topological interface possibilities. We show that graphene sheet edges can adopt a range of topological distortions depending on their nature. Rehybridisation, local bond reordering, chemical functionalisation with bulky, charged, or multi-functional groups can lead to edge buckling to relieve strain, folding, rolling and even tube formation. We discuss the topological possibilities at a 2D graphene edge, and under what circumstances we expect different edge topologies to occur. Density functional calculations are used to explore in more depth different graphene edge types.

cond-mat.mes-hall↗

arXiv2012

A comparative study of density functional and density functional tight binding calculations of defects in graphene

The density functional tight binding approach (DFTB) is well adapted for the study of point and line defects in graphene based systems. After briefly reviewing the use of DFTB in this area, we present a comparative study of defect structures, energies and dynamics between DFTB results obtained using the dftb+ code, and density functional results using the localised Gaussian orbital code, AIMPRO. DFTB accurately reproduces structures and energies for a range of point defect structures such as vacancies and Stone-Wales defects in graphene, as well as various unfunctionalised and hydroxylated graphene sheet edges. Migration barriers for the vacancy and Stone-Wales defect formation barriers are accurately reproduced using a nudged elastic band approach. Finally we explore the potential for dynamic defect simulations using DFTB, taking as an example electron irradiation damage in graphene.

cond-mat.mtrl-sci↗

arXiv2011

Ripple edge engineering of graphene nanoribbons

It is now possible to produce graphene nanoribbons (GNRs) with atomically defined widths. GNRs offer many opportunities for electronic devices and composites, if it is possible to establish the link between edge structure and functionalisation, and resultant GNR properties. Switching hydrogen edge termination to larger more complex functional groups such as hydroxyls or thiols induces strain at the ribbon edge. However we show that this strain is then relieved via the formation of static out-of-plane ripples. The resultant ribbons have a significantly reduced Young's Modulus which varies as a function of ribbon width, modified band gaps, as well as heterogeneous chemical reactivity along the edge. Rather than being the exception, such static edge ripples are likely on the majority of functionalized graphene ribbon edges.

cond-mat.mtrl-sci↗

arXiv2011

Low energy graphene edge termination via small diameter nanotube formation

We demonstrate that free graphene sheet edges can curl back on themselves,reconstructing as nanotubes. This results in lower formation energies than any other non-functionalised edge structure reported to date in the literature. We determine the critical tube size and formation barrier and compare with density functional simulations of other edge terminations including a new reconstructed Klein edge. Simulated high resolution electron microscopy images show why such rolled edges may be difficult to detect. Rolled zigzag edges serve as metallic conduction channels, separated from the neighbouring bulk graphene by a chain of insulating sp$^3$-carbon atoms, and introduce Van Hove singularities into the graphene density of states.

cond-mat.mtrl-sci↗

arXiv2011

Behavior of hydrogen ions, atoms, and molecules in alpha-boron studied using density functional calculations

We examine the behaviour of hydrogen ions, atoms and molecules in alpha-boron using density functional calculations. Hydrogen behaves as a negative-U centre, with positive H ions preferring to sit off-center on inter-layer bonds and negative H ions sitting preferably at in-plane sites between three B12 icosahedra. Hydrogen atoms inside B12 icosahedral cages are unstable, drifting off-center and leaving the cage with only a 0.09 eV barrier. While H0 is extremely mobile (diffusion barrier 0.25 eV), H+ and H- have higher diffusion barriers of 0.9 eV. Once mobile these defects will combine, forming H2 in the interstitial void space, which will remain trapped in the lattice until high temperatures. Based on these results we discuss potential differences for hydrogen behaviour in beta-boron, and compare with experimental muon-implantation data.

cond-mat.mtrl-sci↗