Abstract: Exposure to air pollution, such as particulate matter (PM), causes adverse health effects, particularly to the respiratory tract and cardiovascular system. PM is the collective name for all kinds of particles ranging from small particles and liquid droplets, which contain organic compounds, acids and metals, to soil or dust particles. One distinguishes PM10, PM2.5 and PM0.1, which have aerodynamic particle sizes smaller than 10, 2.5 and 0.1 µm, respectively. It is mainly the latter that is the most harmful, as PM0.1 penetrates deep into the respiratory system and carries relatively more toxic substances than the other PM fractions. Over a 15-year period, PM concentrations in European member states have fallen by about 30%. Nevertheless, the World Health Organisation (WHO) air quality guidelines, which became stricter in 2021, are exceeded in most places around the world. Particularly in cities, excessive levels of PM are measured and it is here that PM mitigation should be investigated. For this, the implementation of urban green infrastructure, including trees, shrubs, green roofs and green walls, is being looked at. Plants hinder airflow and remove PM from the air by deposition on their leaves and branches. This process is known as dry deposition. Plants can capture PM very efficiently, due to their complex structure of leaves and branches. Green walls offer significant advantages over other types of urban green infrastructure because they can grow on the huge available wall area and, because they do not hinder air circulation, as we sometimes see with trees. Green walls are believed to have a much greater, untapped potential to reduce PM pollution. However, a literature review showed that we do not know the quantitative impact of green walls and lack the tools and/or general methodology to do so. The objective of this thesis is therefore to develop a method for assessing PM removal by green walls, based on predictive models and based on relevant parameters that are experimentally determined. Computational fluid dynamics (CFD) is a numerical method to simulate airflow in complex environments such as cities. These models can also simulate the vegetation-wind interaction in detail and are interesting tools to assess the effect of green walls on PM concentrations in real environments. It is important to first study the aerodynamic effect of green walls and parameterise it correctly in CFD models. Plants decrease the wind speed and create turbulence through a combination of viscous and form drag, which are determined by the permeability (K) and drag coefficient (Cd), respectively. Wind tunnel experiments were conducted with three commonly found climbers (Hedera helix, Parthenocissus tricuspidata and Parthenocissus quinquefolia) and the variation of leaf area density was investigated for two of them. It was observed that the air resistance depended on plant species, leaf area density and wind speed. The difference between the plant species was assigned to the functional leaf size (FLS), the ratio of the largest circle within the boundaries of the leaf to the total leaf area. FLS is likely associated with other morphological characteristics of plants that, when considered collectively, provide a more comprehensive representation of leaf complexity. The pressure and velocity measurements obtained were used to optimise the permeability and drag coefficient in a CFD model. At wind speeds below 0.6 m s-1, the resistance was mainly determined by viscous drag and a larger leaf size resulted in a higher viscous drag. At wind speeds above 1.5 m s-1, form drag was dominant and the parameterised Cd decreased with increasing wind speed due to the sheltering effect of successive plant elements. The leaf area density had a significant effect on K and Cd and, is therefore an important plant parameters in CFD models. The main conclusion here is that the common practice of using a constant Cd to model the influence of plants on the air flow leads to deviations from reality. Wind tunnels are highly suitable to study the impact of green walls on PM concentration under controlled environmental conditions. For this purpose, a new wind tunnel setup was built and great attention was paid to obtaining a uniform air flow. Thus, based on CFD models, appropriate flow controllers were chosen, consisting of honeycombs and screens with different mesh sizes. New PM generation devices and measuring equipment were installed and set up appropriately. Devices were available for generating and measuring ultrafine dust (<0.1 µm, i.e. PM0.1) and fine dust (<0.3 µm, i.e. PM0.3) consisting of soot particles, and, on the other hand, fine dust with particle sizes smaller than 2.5 (PM2.5) and 10 µm (PM10) consisting of 'Arizona fine test dust'. With the new wind tunnel setup, it was possible to measure the influence of Hedera helix (common ivy), grown in a planter against a climbing aid, on the PM concentration and this was expressed by a collection efficiency, i.e. the difference in concentration in front and behind the plants normalised for the incoming concentration. The collection efficiency of H. helix depended on the particle size of the PM and wind speed. The collection efficiency decreased when the particle size increased from 0.02 to 0.2 µm and increased again for particle sizes above 0.3 µm. The collection efficiency also increased with increasing wind speed, especially for particle sizes > 0.03 µm. On the other hand, relative humidity and the type of PM (soot or dust) did not significantly affect the collection efficiency. The main objective of this study was to obtain an optimised size-resolved deposition model. Dry deposition occurs through several mechanisms, in particular gravity, diffusion, impaction and interception, and the subsequent resuspension of deposited PM back to the environment. The modelling of these mechanisms was described by \citet{Zhang2001} and \citet{Petroff2010}. The data obtained from the wind tunnel experiments allowed validating these deposition models. It was for the first time that deposition of real PM on green walls was studied. The different PM deposition mechanisms were found to be strongly dependent on particle size and wind speed. The models of \citet{Zhang2001} and \citet{Petroff2010} each matched PM concentration measurements for only certain particle sizes. Therefore, a combination of the two models was investigated and the root mean square error was lower by on average 3.5% (PM < 0.03 µm) and 46% (PM > 0.03 µm) compared to the original models at wind speeds greater than 1.5 m s-1. For wind speeds less than 1.5 m s-1, the optimised model did not differ from the original models. The optimised model was able to meet the imposed criteria for air quality models, where a correct model exhibits low deviation from measurements ('normalised mean square error' < 1.5), low bias ('fractional bias' between -0.3 and 0.3) and high R2. In comparison, the R$2$ of the optimised model was 0.57, while that of Zhang et al. (2001) and Petroff et al. (2010) was 0.23 and 0.31, respectively. The optimised model was however characterised by a high scatter, with the fraction of modeled results located within a factor of two of the measurements being lower than 50. A model study with a green façade oriented parallel to the incoming airflow showed that deposition by interception and impaction reduced remarkably, but that the orientation had no effect on deposition by Brownian diffusion. A promising green wall form for PM mitigation is the living wall system (LWS). LWS consist of supporting structures with substrate to grow plants in and can be planted with a variety of plant species. This allows to select plant species with optimal characteristics to achieve PM deposition. These characteristics refer to the macro- and microstructure of the leaves, and research has been conducted mainly on these. On the other hand, the influence of the supporting structure and substrate on PM concentrations has rarely been studied. With the new wind tunnel setup, LWS from different manufacturers were tested for their ability to capture PM. The setups were subjected for three hours to an air flow with a low PM concentration (resuspension phase) and then for three hours to an air flow to which additional PM was added (deposition phase). Some setups were able to decrease the PM concentration during both phases, while others just caused the concentration to increase. Some systems were able to reduce particulate matter concentration during both phases, namely LWS consisting of planters (-2% and -4% for PM0.1 and PM2.5, respectively) and textile cloths (-23% and -5% for PM0.1 and PM2.5, respectively). While other systems actually resulted in an increase in concentration especially LWS existing textile fabrics consisting of geotextiles (+11% for both PM fractions) and with moss as substrate (+2% and +5% for PM0.1 and PM2.5, respectively). This highlights the importance of careful selection of suspension systems to reduce particulate matter concentrations. Further research is therefore needed on the materials used in these systems in relation to their particulate content, as well as on plant development in these systems. In addition to air measurements, measurements were taken of the amount of PM deposited on the leaves and suspension system of LWS. This allowed the difference in PM resuspension and deposition between plant species to be investigated. The amount of deposited particulate matter was determined based on 'saturation isothermal remanent magnetisation' (SIRM), a measure of magnetisable particulate matter. This was possible because the added 'Arizona fine test dust' contained iron oxide. However, no significant difference was observed between the SIRM values measured before the wind tunnel experiment, after resuspension and after deposition. This suggested that the iron oxide content in the Arizona fine test dust was too low to measure a significant difference in the SIRM values on leaves after three hours. The plant species did give rise to different SIRM values ranging between 5 and 260 µ A. In particular, SIRM values above 26 µ A were observed for the plant species that were grouped due to their significantly higher accumulation of PM. 'Specific leaf area' (SLA), specifically the ratio of the one-sided 'fresh' leaf area to its dry mass, was the significant leaf characteristic. SLA correlated with leaf complexity. In particular, plant species with elongated leaves were characterized by low SLA, high FLS and high complexity and showed significantly higher SIRM values. Finally, the optimised size-resolved deposition model was also tested in an urban model to get an idea of the impact of a green wall on PM concentrations in a so-called 'street canyon'. These are narrow streets with high buildings on both sides, making air pollution more persistent. To this end, an ideal scenario was tested in which a green wall was introduced along both sides of the street over a length of about 270 m. The model result showed a decrease in PM2.5 and PM10 of 46 ± 12% and 52 ± 14%. This result is of course for a very optimal scenario where the green wall covers the entire building façades. Since this is not feasible in reality, other ways of promoting contact between green walls and polluted air can be explored. The insights obtained illustrate that the use of climbing plants can be a cost-effective and environmentally friendly solution to reduce PM concentrations. Moreover, the findings showed that models can be used to investigate the impact of green walls on PM levels. These findings fit within the broader context of designing healthy and sustainable urban environments and developing innovative solutions based on solid scientific knowledge.

Abstract: Nanoscience and nanotechnologies are of immense importance across many fields of science and for numerous practical applications. In this context, scanning transmission electron microscopy (STEM) and 4D-STEM are among the most powerful characterization methods at the atomic scale. Annular dark-field (ADF)-STEM can be used to quantify atomic structures in 3D by counting atoms based on a single projection image. In 4D-STEM a full diffraction pattern is recorded at each scan step, which enables more dose efficient imaging and the utilization of various advanced imaging modalities, which can however be complex and slow. Both, STEM and 4D-STEM suffer from noise and distortions. In the first section of this work the most important of these distortions are discussed and it is shown how image restoration with a dedicated convolutional neural network (CNN) can be beneficial for atomic structure quantifications in ADF-STEM. In the second part, a new 4D-STEM imaging method real-time-integrated-centre-of-mass (riCOM) is introduced, which is a very dose-efficient and fast algorithm that enables unprecedented live-imaging capabilities for 4D-STEM. It is based on the integrated centre-of-mass approach, but is reformulated with variable integration ranges and optional filters, which allows for a tunable contrast transfer function. This enables the imaging of light and heavy elements simultaneously at very low doses. In the third part another new 4D-STEM method, coined AIRPI (AI-assisted rapid phase imaging) is introduced, which uses a CNN to retrieve a patch of the specimen's phase image for each scan position, based on the diffraction patterns in the probe's immediate surroundings. This allows also live imaging in principle and surpasses comparable state-of-the-art algorithms in terms of resolution also at low doses. Different atomic columns can be reliably distinguished over a wide range of atomic numbers, enabling a very good image interpretability. Further, AIRPI can recover low frequency image components, which preserves thickness information. This is a unique and important feature which could make quantitative 4D-STEM feasible.

Abstract: This dissertation focuses on the materials and techniques used by the Brussels artists Michael Sweerts (1618-1664) and Michaelina Wautier (1614-1689). It tries to answer the following research questions: 1.Did 17th-century Brussels-based artists use idiosyncratic painting materials and techniques, and can their work be distinguished from paintings created in other artistic centers based on these materials and techniques? 2.How did Sweerts and Wautier influence each other’s work? In order to answer the research questions, a selection of relevant and representative paintings from Sweerts’ and Wautier’s oeuvre has been studied with MA-XRF scanning and microscopic examination. The results are discussed in volume 2 of this dissertation, and form the basis upon which the chapters in volume 1 are built. MA-XRF scanning is a non-destructive imaging technique based on the technique of X-ray fluorescence (XRF). The technique allows for the collection of elemental information about the materials present in a work of art. The work presented in this dissertation has shed more light on the working methods of Sweerts and Wautier. It has shown they both showed an openness and willingness to continue learning and developing as an artist. Wautier did not feel constrained to work only in one genre and apparently carried on developing her skills throughout her career, as evidenced by Flower Garland with Butterfly. She might have trained with a specialized artist to master their idiosyncratic working methods and this diverse genre. This suggests that Wautier had an interest in art education. Sweerts too had a passion for art education: he was involved in the drawing academy of Camillo Pamphilj in Rome, and later established his own academy for life drawing in Brussels. Sweerts’ work clearly shows the influence of his Italian contemporaries. Whilst living in Rome, he adapted his working methods to local customs such as working on brightly colored grounds. Wautier’s oeuvre shows that she was open to explore different genres and that she took inspiration from her contemporaries, including Theodoor van Loon and Daniël Seghers. In doing so, she furthered the development of different genres as she added original elements to her compositions. It is the conclusion of this dissertation that Michaelina Wautier can thus be considered one of the innovative artists of mid-17th-century Flemish painting, who inspired other artists in the Low Countries, including Michael Sweerts.

Abstract: Phenolic compounds can be found everywhere in our daily lives but exhibit high toxicity, low (bio)degradability and hormone-disrupting effects when they are released in the environment. It is for this reason imperative to develop detection strategies for these pollutants. A promising approach involves the use of a photoelectrochemical (PEC) sensor. In this sensor, a photosensitiser (PS) type II, which generates 1O2 under illumination, is used to oxidise phenolic compounds present in the sample. The oxidised phenols are reduced at the electrode surface leading to the generation of an electrocatalytic redox cycle. In this thesis, an in-depth understanding, through the identification of the reactive oxygen species (ROS) in the PEC sensing mechanism, is obtained. The detection strategy is optimised by choosing the PS with the highest 1O2 production and by optimising the detection parameters so that the PEC sensor can be successfully applied for the detection of phenols in industrial samples. First, it was determined that the use of highly fluorinated zinc phthalocyanine derivatives, F52PcZn and F64PcZn, as photocatalysts was optimal for the sensing of phenol due to their high 1O2 production and improved single-site isolation. However, next to 1O2, it was shown that the ROS O2•- and H2O2 were also generated in the PEC sensor. Their contribution to the photocurrent response was studied by rotating disk electrode measurements in function of the pH and applied potential. After this, the PEC detection strategy was optimised in terms of pH and applied potential for the detection of doxycycline, cefadroxil, and phenol. It was found that the use of alkaline pH-levels led to nmol L-1-level detection limits. The combination with square wave voltammetry (SWV) was, also, proposed to allow the quantification and identification of phenolic compounds in a specific sample. At last, the developed PEC and SWV sensors were applied for the measurement of phenolic compounds in industrial water samples. The PEC sensor could follow the decrease of the phenolic concentration throughout the wastewater treatment process while the SWV sensor provided the electrochemical fingerprints of these samples. The thesis concluded that the use of the PEC sensor was advantageous in the measurement of lower concentrated phenolic samples due to its high sensitivity and fast measurement time in comparison to commercial test kits.

Abstract: During the past decades, metallic nanoparticles (NPs) have attracted great attention in materials science due to their specific optical properties based on surface plasmon resonances. Because of these phenomena, plasmonic NPs (or nanoplasmonics) are very promising for application in biosensing, photocatalysts, medicine, data storage, solar energy conversion, etc. Currently, colloidal synthesis techniques enable scientists to routinely produce mono and bimetallic NPs of various shapes, sizes, composition, and elemental distribution, with superior properties for plasmonic applications. Two primary directions for further advancing nanoplasmonic-based technologies include synthesizing novel morphologies, such as highly asymmetric chiral NPs, and gaining deeper insights into the factors affecting the stability of produced nanoplasmonics. With the increasing complexity of nanoplasmonics morphologies and higher stability requirements, there is a pressing need for thorough investigations into their 3D structures and their evolution under different conditions, with high resolution. Electron tomography (ET) emerges as an ideal tool to retrieve shape and element-sensitive information about individual nanoparticles in 3D, achieving resolutions down to the atomic level. Moreover, ET techniques can be combined with in situ holders, enabling detailed studies of processes mimicking real applications of nanoplasmonic-based devices. The first part of this thesis will focus on detailed studies of chiral Au NPs, promising for spectroscopy techniques based on the differential absorption of left- and right-handed circularly polarized light. Specifically, I will discuss the primary strategies for wet-colloidal growth of the various types of intrinsically chiral Au NPs. Advanced ET methods will be demonstrated as powerful tools for characterizing the final helical morphologies of the produced Au NPs and for studying the chiral growth mechanisms by examining intermediate structures obtained during chiral growth. The second part will focus on the heat-induced stability of various Au@Ag core-shell NPs. Operating in real conditions, such as elevated temperatures, may cause particle reshaping and redistribution of metals between the core and shell, gradually altering nanoplasmonics properties. Hence, a thorough understanding of the influence of size, shape, and defects on these processes is crucial for further developments. Recently developed techniques, combining fast ET with in-situ heating holders, have allowed me to evaluate the influence of various parameters (size, shape, defect structure) on heat-induced elemental redistribution in Au@Ag core-shell nanoparticles qualitatively and quantitatively. Additionally, I will discuss the prospects of high-resolution ET for visualizing the diffusion of individual atoms within complex nanostructures.

Abstract: Carbon dioxide electrochemical reduction (CO2ER) has attracted considerable attention as a technology to recycle CO2 into raw materials for chemicals using renewable energies. We recently found that Zn-Al layered double hydroxides (Zn-Al LDH) have the CO-forming CO2ER activity. However, the activity was only evaluated by using the liquid-phase CO2ER. In this study, Ni-Al and Ni-Fe LDHs as well as Zn-Al LDH were synthesized using a facile coprecipitation process and the gas-phase CO2ER with the LDH-loaded gas-diffusion electrode (GDE) was examined. The products were characterized by XRD, STEM-EDX, BF-TEM and ATR-IR spectroscopy. In the ATR-IR results, the interaction of CO2 with Zn-Al LDH showed a different carbonates evolution with respect to other LDHs, suggesting a different electrocatalytic activity. The LDH-loaded GDE was prepared by simple drop-casting of a catalyst ink onto carbon paper. For gas-phase CO2ER, only Zn-Al LDH exhibited the CO2ER activity for carbon monoxide (CO) formation. By using different potassium salt electrolytes affording neutral to strongly basic conditions, such as KCl, KHCO3 and KOH, the gas-phase CO2ER with Zn-Al LDH-loaded GDE showed 1.3 to 2.1 times higher partial current density for CO formation than the liquid-phase CO2ER.

Abstract: We present a large dataset containing simulated core-loss electron energy loss spectroscopy (EELS) spectra with the elemental content as ground-truth labels. Additionally we present some neural networks trained on this data for element identification.  The simulated dataset contains zero padded core-loss spectra from 0 to 3072 eV, which represents 107 core-loss edges through all 80 elements from Be up to Bi. The core-loss edges are calculated from the generalised oscillator strength (GOS) database presented by Zhang et al.[1] Generic fine structures using lifetime broadened peaks are used to imitate fine structure due to solid-state effects in experimental spectra. Generic low-loss regions are used to imitate the effect of multiple scattering. Each spectrum contains at least one edge of a given query element and possibly additional edges depending on samples drawn from The Materials Project [2]. The dataset contains for each of the 80 elements: 7000 training spectra, 1500 test spectra, 600 validation spectra and 100 spectra representing only the query element. This results in a total 736 000 labeled spectra. Code on how to  – read the simulated data – transform HDF5 format to TFRecord format – train and evaluate neural networks using the simulated data – use the trained networks for automated element identification is available on GitHub at arnoannys/EELS_ID A full report on the simulation of the dataset and the training and evaluation of the neural networks can be found at:                    Annys, A., Jannis, D. & Verbeeck, J. Deep learning for automated materials characterisation in core-loss electron energy loss spectroscopy. Sci Rep 13, 13724 (2023). https://doi.org/10.1038/s41598-023-40943-7 [1] Zezhong Zhang, Ivan Lobato, Daen Jannis, Johan Verbeeck, Sandra Van Aert, & Peter Nellist. (2023). Generalised oscillator strength for core-shell electron excitation by fast electrons based on Dirac solutions (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7729585 [2] Anubhav Jain, Shyue Ping Ong, Geoffroy Hautier, Wei Chen, William Davidson Richards, Stephen Dacek, Shreyas Cholia, Dan Gunter, David Skinner, Gerbrand Ceder, Kristin A. Persson; Commentary: The Materials Project: A materials genome approach to accelerating materials innovation. APL Mater 1 July 2013; 1 (1): 011002. [https://doi.org/10.1063/1.4812323](https://doi.org/10.1063/1.4812323)

Abstract: The foodborne pathogens and microorganisms have played a prevalent role in the ebb and flow of the economy worldwide. The increasing population has strained the food processing industry to produce food in large quantity, which in turn has affected the quality of food. To curb this issue, there is immense pressure to produce and maintain quality food within a short time frame. Hence, high throughput technology is used to determine and timely assess the safety and hygiene of food. Further, the revolution of the food industry has also seen an upsurge of new pathogens and microorganisms, thereby increasing the risk of exposure towards rarest diseases to a larger population. This chapter sheds light on the different types of foodborne pathogens affecting the food industry and its social impact. It further emphasizes the safety measures to be taken on the prevention of the disease from the farm to the processing industries and in turn to the household.

Abstract: Generalized Möbius-Listing surfaces and bodies generalize Möbius bands, and this research was motivated originally by solutions of boundary value problems. Analogous to cutting of the original Möbius band, for this class of surfaces and bodies, results have been obtained when cutting such bodies or surfaces. The results can be applied in a wide range of fields in the natural science, and here we propose how they can serve as a model for the heart and the circulatory system.

Abstract: Illicit drugs are harmful substances, posing a threat to the health and safety of society. Policies, such as supply reduction and harm reduction, are in place to combat the illicit drug problem. Science can play a substantial role in this fight, by providing tools that enable these policies to be successfully enforced. One example are on-site detection tools, i.e. sensors that allow the on-site identification of an illicit drug in a sample of interest. Several technologies, such as color tests and portable spectroscopic techniques, are currently employed for this goal. Although these are valuable techniques, there is an opportunity for voltammetry, an electrochemical technique, to make an impactful addition to this repertoire of on-site detection tools. Despite its attractive features (low-cost, portable, short analysis time, indifference to color,...), voltammetric illicit drug sensor have failed to make an impact in real scenarios. The work outlined in this PhD thesis aims to change this by bringing the technology from the lab to the field. Strategic choices, fueled by feedback from end-users, were made to further develop those specific aspects of the technology that previously haltered the technology to fulfill its potential. A detection algorithm was introduced that converts the voltammetric output into a clear-cut interpretation thereof, opening up the technology to end-users without prior knowledge of the technology. A sensor that allows qualitative and quantitative detection of the psychoactive drug MDMA was introduced, and importantly, validated on a large set of 212 confiscated samples. A state-of-the art mobile application and adequate sampling methodology were developed, alongside other, often more practical studies and product developments, to evolve the technology into a product that truly creates value for end-users. Important steps towards multidrug detection were made with a festival sensor and a flowchart based on visual appearance that ties together a variety of voltammetric single sensors into a single multidrug sensing approach. Last but not least, multiple valorization aspects were researched, including a market study and an analysis to determine the optimal commercialization strategy. Overall, this PhD thesis has facilitated the transition of the voltammetric illicit drug sensing technology from lab to on-site application. The final application creates value for end-users, and is ready to make an impact in real on-site scenarios.

Abstract: It is standard practice in virtual reality applications to synthesize binaural audio based on a discrete set of directionally-dependent head-related impulse responses (HRIRs). This set of HRIRs is often time-aligned in a pre-processing step, to allow for high-fidelity interpolation between HRIRs corresponding with neighbouring directions. The fidelity of this operation depends on the similarity of neighbouring aligned HRIRs. The pairwise quality of similarity makes it a difficult criterion to optimize globally and consequently one often resorts to alignment methods based on a specific feature that can be extracted for each HRIR separately, e.g., the first-onset of the peak or the group delay. However, such proxies for similarity are very sensitive to noise and therefore require a high signal-to-noise ratio, which makes them less suitable for processing HRIRs acquired outside an anechoic room. In this paper, we advance a novel alignment method, which maximizes the similarity – defined as the correlation between the full-length HRIRs – between neighbouring aligned HRIRs for all directions at once. We show that this correlation-based alignment procedure outperforms the first-onset alignment with regards to the fidelity of the spherical harmonics representation of both the spectral and interaural time difference (ITD) information, when tested on the KEMAR HRIR and six human HRIRs. Finally, we show that the correlation-based alignment is more robust to noise.

Abstract: Delve into the diverse aspects of plant morphology, their responses to global climate change, and the spatiotemporal dynamics of forest productivity. Join us on a journey through the intricate web of plant characteristics and their impact on the environment.

Abstract: Nitrogen (N) is an indispensable building block for all living organisms as well as for pharmaceutical and chemical industry. In a nutshell, N is needed for plants to grow and beings to live and nitrogen fixation (NF) is the process that makes N available for plants as food by converting N2 into a reactive form, such as ammonia (NH3) or nitrogen oxides (NOx), upon reacting with O2 and H2. The aim of this thesis is to elucidate (wet) plasma-based nitrogen fixation with a focus on (1) the role of pulsing in achieving low energy consumption, (2) the role of H2O as a hydrogen source in nitrogen fixation and (3) elucidation of nitrogen fixation pathways in humid air and humid N2 plasma in a combined experimental and computational study. Furthermore, this thesis aims to take into account the knowledge-gaps and challenges identified in the discussion of the state of the art. Specifically, (1) we put our focus on branching out to another way of introducing water into the plasma system, i.e. H2O vapor, (2) we de-couple the problem for pathway elucidation by starting with characterization of the chosen plasma, next a simpler gas mixture and building up from there, (3) we include modelling, though not under wet conditions and (4) we focus on also analyzing species and performance outside liquid H2O. Firstly, based on the reaction analysis of a validated quasi-1D model, we can conclude that pulsing is indeed the key factor for energy-efficient NOx- formation, due to the strong temperature drop it causes. Secondly, the thesis shows that added H2O vapor, and not liquid H2O, is the main source of H for NH3 generation. Related to this, we discuss how the selectivity of plasma-based NF in humid air and humid N2 can be controlled by changing the humidity in the feed gas. Interestingly, NH3 production can be achieved in both N2 and air plasmas using H2O as a H source. Lastly, we identified a significant loss mechanism for NH3 and HNO2 that occurs in systems where these species are synthesized simultaneously, i.e. downstream from the plasma, HNO2 reacts with NH3 to form NH4NO2, which decomposes into N2 and H2O. This reduces the effective NF when not properly addressed, and should therefore be considered in future works aimed at optimizing plasma-based NF. In conclusion, this thesis adds further to the current state of the art of plasma-based NF both in the presence of H2O and in dry systems.