Submitted by Ana Hilda Fonseca (anahilda@ufba.br) on 2016-09-06T16:08:00Z No. of bitstreams: 1 Lopes W A Tese de Doutorado IQ-UFBA 2007.PDF: 2585828 bytes, checksum: 5e4ecce83a58663cda34dce3a68a966c (MD5) / Approved for entry into archive by Vanessa Reis (vanessa.jamile@ufba.br) on 2016-09-08T10:48:19Z (GMT) No. of bitstreams: 1 Lopes W A Tese de Doutorado IQ-UFBA 2007.PDF: 2585828 bytes, checksum: 5e4ecce83a58663cda34dce3a68a966c (MD5) / Made available in DSpace on 2016-09-08T10:48:19Z (GMT). No. of bitstreams: 1 Lopes W A Tese de Doutorado IQ-UFBA 2007.PDF: 2585828 bytes, checksum: 5e4ecce83a58663cda34dce3a68a966c (MD5) / CNPq, FINEP, FAPESB e ANEEL / Os Hidrocarbonetos policíclicos aromáticos (HPA) são poluentes formados geralmente durante a combustão incompleta de petróleo e derivados, carvão mineral, biomassa e outros materiais orgânicos. São emitidos por diversas fontes naturais ou antrópicas e estão presentes na atmosfera, hidrosfera e litosfera. A atmosfera é o seu principal meio de transporte onde se encontram em fase vapor ou adsorvidos em material particulado. Em centros urbanos, sua emissão está associada principalmente aos veículos automotores, movidos a diesel e gasolina. Os HPA e seus derivados, por exemplo, os nitro-HPA e oxi-HPA, são agentes cancerígenos e/ou mutagênicos, daí a importância de estudos sobre a sua presença na atmosfera e em outras matrizes ambientais, sobre a reatividade e atividades biológicas. Neste trabalho, foram determinadas as concentrações atmosféricas para 16 HPA considerados como poluentes prioritários pela agência de proteção ambiental dos Estados Unidos da América (U. S. EPA), em amostras de material particulado atmosférico (MPA), usando o método de CG-EM com monitoramento de íons selecionados. As amostras de MPA foram coletadas em amostradores de grande volume (Hi-vol) e, em seguida, foram extraídas usando a técnica de sonicação e uma solução de acetonitrila e diclorometano (3:1) como solvente. Os extratos foram pré- concentrados (10/1) e em seguida analisados por CG-EM/SIM. Com o objetivo de assegurar uma boa separação dos 16 HPA, no menor tempo de análise, foi utilizado o planejamento multivariado para estabelecer as condições de programação de temperatura do forno (aquecimento da coluna). O processo de otimização foi realizado por meio do planejamento fatorial fracionário e através do planejamento do tipo Box-Behnken, sendo avaliados os seguintes fatores: temperatura inicial (oC), velocidade de elevação da temperatura N° 1 (°C.min-1), temperatura intermediária (oC), velocidade de elevação da temperatura N° 2 (°C.min-1) e temperatura final (oC). As condições otimizadas para a separação dos 16 HPA foram estabelecidas em: 70 °C (2 min) ? 200 °C (30 °C/min, 5 min) ? 300 °C (5 °C/min, 1,67 min). O tempo total da análise, com uma boa separação dos 16 analitos, ficou em 33 minutos. A validação da técnica de CG-EM, no modo SIM, apresentou resultados considerados satisfatórios para a repetitividade da resposta do detector, repetitividade do tempo de retenção, e linearidade das curvas de calibração. Os limites de detecção foram estabelecidos entre 0,13 e 0,34 ng mL-1 (área) e 0,18 e 0,72 ng mL-1 (altura de pico) e os limites de quantificação entre 0,38 e 1,04 ng mL-1 (área) e entre 0,61 e 2,39 ng mL-1 (altura de pico). As amostras ambientais de MPA foram coletadas em quatro diferentes locais, apresentando os seguintes resultados: i) Estação da Lapa (Salvador, BA) - Os HPA presentes em concentrações mais altas foram o criseno (CRI), pireno (PIR) e benzo[b]fluoranteno (BbF), apresentando concentrações médias de respectivamente 2,62, 1,32 e 1,30 ng m-3, e perfil compatível com emissões de veículos movidos a diesel; ii) Porto de Aratu (Candeias, BA) - Os HPA presentes em maiores concentrações foram o benzo[b]fluoranteno (BbF), benzo[g,h,i]perileno (BgP) e Indeno[1,2,3- c,d]pireno (IND), apresentando concentrações médias de respectivamente 2,53, 1,22 e 1,12 ng m-3 (AGV-PTS) e 1,74, 0,82 e 0,73 (AGV MP10); iii) Ilha de Maré, (Salvador, BA) – os valores mais altos foram observados para o benzo[b]fluoranteno (BbF), benzo[b]fluoranteno (BkF) e benzo[g,h,i]perileno (BgP), apresentando concentrações médias de respectivamente 1,62, 0,73 e 0,72 ng m-3; iv) Cidade Universitária da USP (São Paulo, SP) - Os HPA que apresentaram maiores concentrações médias foram o benzo[b]fluoranteno (BbF), criseno (CRI) e o benzo[g,h,i]perileno, com os valores de, respectivamente, 0,93, 0,50 e 0,49 ng m-3. Os resultados foram avaliados em termos de perfil das concentrações, prováveis fontes de emissões e, em alguns casos, por comparação com análises anteriores. / Polycyclic aromatic hydrocarbons (PAH) are ubiquitous pollutants that are emitted from several natural or anthropic sources and may be present in the atmosphere, hydrosphere and lithosphere. They usually originate during the incomplete combustion of petroleum and its derivatives, mineral coal, biomass and other organic materials. They are mainly transported in the atmosphere, being present either in the vapor phase or adsorbed in particulate material. In urban centers, emission is mainly associated with diesel- and gasoline-powered motor vehicles. Since PAH and derivatives, e.g. nitro-PAH and oxy-PAH, are carcinogenic and/or mutagenic agents, to study the effect of their presence in the atmosphere and in other environmental matrices on the biological activity and reactivity are of key relevance. In this study, the atmospheric concentrations of the 16 PAHs that are considered priority pollutants by the U.S. Environmental Protection Agency were evaluated in samples of atmospheric particulate material (APM), using gas chromatography / mass spectrometry with selected-ion monitoring mode (GC-MS/SIM). APM samples were collected in high-volume air samplers (Hi-vol) and then extracted under sonication with a solution of acetonitrile and dichloromethane (3:1) as a solvent. Extracts were pre-concentrated (10/1) and analyzed by GC-MS/SIM. In order to assure an adequate separation of the 16 PAHs, in the shortest analysis time, a multivariate design was used to set the conditions of the oven temperature program (column warming). The optimization process was carried out using factorial fractional design and Box-Behnken design. The following factors were evaluated: initial temperature (ºC), temperature rate N°1 (ºC.min-1), intermediate temperature (ºC), temperature rate N°2 (ºC.min-1) and final temperature (ºC). The optimized conditions for the separation of the 16 PAH were set at: 70 ºC (2 min) → 200 ºC (30 ºC/min, 5 min) → 300 ºC (5 ºC/min, 1.67 min). Total analysis time, with an adequate separation of the 16 analytes was 33 minutes. Validation of GC-MS in the SIM mode rendered satisfactory results for repeatability of detector response, repeatability of retention time, and linearity of calibration curves. Detection limits were established between 0.13 and 0.34 ng mL- 1 (area) and 0.18 and 0.72 ng mL-1 (peak height). Environmental samples of PMA were collected at four different locations, with the following results: i) Lapa Bus Station (Salvador, BA) - The PAH with the highest concentrations were crysene (CRY), pyrene (PYR) and benzo[b]fluoranthene (BbF), with mean concentrations of 2.62, 1.32, and 1.30 ng m-3, respectively, suggesting a profile compatible with diesel vehicle emissions; ii) Port of Aratu (Candeias, BA) – The PAH with the highest concentrations were benzo[b]fluoranthene (BbF), benzo[g,h,i]perylene (BgP) and indeno[1,2,3-c,d]pyrene (IND), with mean concentrations of 2.53, 1.22, and 1.12 ng m-3 (TSP) and 1.74, 0.82 e 0.73 (PM10), respectively; iii) Ilha de Maré, (Salvador, BA) – The PAH with the highest concentrations were benzo[b]fluoranthene (BbF), benzo[k]fluoranthene (BkF) and benzo[g,h,i]perylene (BgP), with mean concentrations of 1.62, 0.73 e 0.72 ng m-3, respectively; iv) University of São Paulo (São Paulo, SP) - The PAH with the highest mean concentrations were benzo[b]fluoranthene (BbF), chrysene (CRI) and benzo[g,h,i]perylene (BgP) with values of 0.93, 0.50 and 0.49 ng m-3, respectively. The results were evaluated with respect to concentration profile, probable emission sources and in some cases, were compared to prior analyses.

La réglementation plus sévère sur les niveaux de matières particulaires (PM) en atmosphère ambiante fixée par l'Union européenne à l'horizon 2015 impose de pouvoir quantifier les contributions des différentes sources d'émission. Les sources d'émissions primaires de particules peuvent être anthropiques (chauffage au bois, au fioul, émissions véhiculaires ou industrielles, combustion de déchets verts, activités de cuisine,…) ou naturelles (végétaux, poussières crustales). Des processus secondaires de formation des particules (conversion gaz-particules) peuvent également contribuer aux taux de PM mesurés. L'identification et la quantification des sources peuvent être conduites notamment à travers l'étude de la composante organique des aérosols réalisée à partir d'analyses chimiques de prélèvements sur filtres en atmosphère ambiante ou à l'émission. Dans ce travail, des méthodologies de caractérisation des sources d'émission par l'analyse de la fraction organique des PM et la quantification de leur influence sur les concentrations ambiantes en particules fines ont été développées et/ou adaptées de méthodologies existantes sur de nombreux sites d'observation. Une attention particulière a été portée sur différents sites de la région Rhône-Alpes où de nombreux épisodes de dépassements des valeurs limites en PM sont régulièrement enregistrés. Les méthodologies ont été développés dans un premier temps sur un site urbain de référence (sites des Frênes à Grenoble) et s'appuient à la fois sur des approches qualitatives d'études des sources (étude d'empreinte de grandes familles chimiques, utilisation de ratios de composés traceurs) et quantitatives (ACP, mesures isotopiques du carbone, Molecular-Marker Chemical Mass Balance). Elles permettent une bonne estimation des sources de matière organique (OM) et des PM en hiver. Leurs applications à d'autres sites de topologies différentes (ruraux, fond de vallées, proximités de sources, marin) en France et en Suisse et à d'autres saisons a permis de montrer la complémentarité des différentes approches introduites et la nécessité de leur adaptation aux spécificités des sites pour permettre une bonne estimation des PM. L'estimation des contributions des sources à l'OM reste cependant encore à améliorer. De plus, dans le cadre de leur application à des sites de vallées alpines suisses, l'approche quantitative de type CMB basée sur l'étude de traceurs organiques a pu être confrontée à une méthode basée sur l'étude de la composante inorganique des aérosols. Les méthodologies développées et mises en œuvre au cours de ce travail constituent des outils de référence dans le cadre d'études des sources d'aérosol à portée scientifique ou réglementaire à l'échelle nationale ou internationale. / Tougher European Union regulations of ambient particulate matter (PM) will require apportionment of the different sources. Primary sources can be anthropological sources (biomass burning, fossil-fuel combustion, cooking, ….) and natural sources (vegetative detritus, crustal dust). PM can also be formed by gas-to-particle conversion processes (secondary sources). The identification and sources apportionment can be conducted by studying the organic fraction of aerosols using chemical analysis on off-line filters in ambient and in emissions of sources. In this work, methodologies of emission sources characterization by analyzing the organic fraction of PM and the sources apportionment on fine aerosol mass were developed and/or adapted from previous methodologies on many ambient sites. A special focus was placed on different sites of the Rhône-Alpes Region where many exceedances of authorized PM values are observed. Methodologies were developed primarily for an urban background site (Les Frênes site in Grenoble) and are based both on qualitative approaches (footprints study of large chemical family, use of tracer ratios) and quantitative approaches (PCA, isotopic carbon measurements, Molecular-Marker Chemical Mass Balance). They allow a good estimation of OM (organic matter) and PM sources in winter. Applications of those methodologies for other sites with different topologies (rural, valley bottoms, near sources, marine, in France and in Switzerland, for different seasons) showed the complementarity of the different methodologies proposed. It also pointed out the necessity of adapting them to the specificities of each sites in order to have a good estimation of PM. In addition, further work should still be done to improve sources apportionment of OM. Moreover, in the context of applying those methodologies to Swiss alpine valley sites, it got possible to compare the quantitative CMB approach based on the study of organic compounds with another approach based on the study of the inorganic fraction of aerosols. Methodologies developed and optimized during this work will be used like reference to scientific or regulation applications in other national or international studies.

Atmospheric particulate matter is known to have significant effects on human health, visibility, and global climate. The magnitudes of these effects, however, depend in complex ways on chemical composition, relative humidity, temperature, phase state, and other parameters. Current regional air quality models such as CMAQ (Community Multiscale Air Quality model) ignore many of these considerations, and consider that the formation of secondary organic aerosol (SOA) can be calculated by assuming thermodynamic ideality in the organic particulate matter (OPM) phase as well as negligible uptake of water into the OPM phase. Theoretical predictions and model simulations considering non-ideality and water uptake show that the standard model assumptions can lead to large errors in predicted SOA mass, and that the magnitude of these errors is sensitive to the composition of the OPM phase. The SOA module in CMAQ v4.7.1 has been revised in this work to allow consideration of the effects of both non-ideality and water uptake. First, a reasonable specific surrogate structure was assigned to each of the lumped products assumed to be produced by reaction of the different precursor hydrocarbons considered in CMAQ (e.g., isoprene, benzene, and toluene). Second, the CMAQ code was modified to allow iterative calculation (at each point in space and time) of the gas/particle partitioning coefficient for each of the SOA-forming products and for water. Third, model simulations were performed for the Eastern US at a resolution of 36-km x 36-km for late summer 2006, under a range of relative humidity conditions. When compared with an appropriate base case, the modified code produced increases in SOA ranging from 0.17 to 0.51 micrograms per cubic meter. The average change was 0.30 micrograms per cubic meter, corresponding to a 37% increase in SOA formation. Incorporation of phase separation effects would likely lead to further increases in predicted SOA levels.

<p>SO₂ and NO_x (NO+NO₂) are important trace gases in the atmosphere as they adversely affect air quality and are precursors to sulfate and nitrate aerosols in the atmosphere. However, there are significant uncertainties in the emission inventories and the atmospheric chemistry processes of both gases. Addressing these uncertainties will help us to 1) better regulate their emissions from anthropogenic activities, 2) understand the formation mechanism of aerosol pollution events, during which rapid accumulation of nitrate and sulfate aerosols are commonly observed, and 3) better constrain the impact of SO₂, NO_x, sulfate aerosols and nitrate aerosols to the global radiation balance. Stable isotopes of nitrogen and sulfur are useful tools in understanding both the origins and chemistry of SO₂ and NO_x since different emission sources usually display distinct sulfur and nitrogen isotopic compositions, and different SO₂ and NO_xoxidation pathways fractionate sulfur and nitrogen isotopes differently. In this dissertation, five studies are conducted to 1) use sulfur isotopes to investigate the sources and chemistry of atmospheric sulfur, and 2) improve our understanding of the isotopic fractionation processes associated with the atmospheric chemistry of reactive nitrogen. </p><p>Using stable sulfur isotopes, we first analyzed the sources of sulfate aerosols collected at Baring Head, New Zealand and atmospheric deposition at the Atacama Desert. At Baring Head, we found that the secondary sulfate, i.e., sulfate formed from atmospheric oxidation of SO₂, is mainly observed in fine aerosols (<1 µm) while the sulfate in coarse aerosols (>1 µm) is mostly sea salt sulfate. 73-77% of the secondary sulfate is sourced from biogenic emissions by ocean phytoplankton, and the rest is originated from anthropogenic activities. The sulfate deposition across the Atacama Desert, on the other hand, is a mixture of sea salt sulfate (only near the coast), anthropogenic SO₂ emissions, local soil, and lake salts. Then, sulfur isotopes were used to investigate the formation chemistry of sulfate aerosols collected during a strong winter haze episode in Nanjing, China, where the sources of SO₂ were well-understood. We found that, although the sources of sulfur remain unchanged during the haze episode, the sulfur isotopic compositions of sulfate vary significantly, suggesting isotopic fractionation occurred during the formation of sulfate aerosols. We interpreted the variation using a Rayleigh distillation model to evaluate the contribution of sulfate formation pathways. The model suggested that the Transition Metal Ion catalyzed O₂ oxidation pathway contributed 49±10% of the total sulfate production, while the O₃/H₂O₂ oxidations accounted for the rest. </p><p>Next, we conducted experiments in an atmospheric simulation chamber to determine the isotopic fractionations between NO and NO₂. This isotopic fractionation is controlled by a combination of two factors: 1) the equilibrium isotopic exchange between NO and NO₂ molecules, and 2) the kinetic isotope effects of the NO_x photochemical cycle, namely the Leighton Cycle Isotope Effect (LCIE). Our experiments showed that the fractionation factor during the isotopic exchange is 1.0289±0.0019, and the fractionation factor of LCIE is 0.990±0.005. A model was constructed to assess the relative importance of the two factors, showing the isotopic exchange should be the dominant factor when NO_x >20 ppb, while LCIE should be more important at low NO_x concentrations (<1 ppb) and high rates of NO₂photolysis. Last, we quantified the overall nitrogen isotopic fractionation during the formation of nitrate aerosols collected at Baring Head, New Zealand. Our results showed that significant and variable (0-15‰) isotopic fractionations occurred during the formation of nitrate aerosols. The isotopic fractionation factors are lower in the summer and higher in the winter, which is mainly caused by seasonal variations in nitrate formation pathways. </p><p>Overall, this dissertation first applied stable sulfur isotopes in aerosol samples collected in different environments, demonstrating that isotopes are excellent tools in identifying the origins and chemistry of atmospheric sulfur. Then, we investigated the isotopic fractionation processes during the atmospheric nitrogen chemistry, which can be useful for future studies aimed at understanding the origins and chemistry of atmospheric nitrogen using stable nitrogen isotopes.</p>

<p>  </p> <p>Atmospheric particles originate from all over the globe with wildly different sources such as sea spray aerosols of the ocean, mineral dust from deserts, biogenic emissions from forests, anthropogenic emissions of urban and industrial areas, volcanic eruptions, and many more. All of these particles can then be transported during which aging can occur where the external and internal chemical composition of particles and drastically be altered in which their physiochemical properties change or new particles as a whole are formed. Understanding what can cause this aging and correctly identifying how these particles change is vital for assessing climate in local areas. </p> <p>Chapter 3 focuses on dry intrusion (DI) and non-DI periods where vertical mixing of air occurs and allows for long range transport of particles. DI periods introduces populations of aged particles from far away sources into local regions. Identification and chemical characterization is performed for both of these periods to highlight the changes the DI period introduces in regards to particle morphology, chemical composition and lifetime. Analysis was performed via computer controlled scanning electron microscopy (CCSEM) for external information of the particles and scanning transmission x-ray microscopy (STXM) was used for internal information. The combination of these two techniques allows for a complete and thorough understanding of the particles during the two periods. </p> <p>Chapter 4 covers the first experiment done on a newly constructed cryogenic cooling cart which was created in the hopes to identify individual ice nucleating events of particles <em>in situ</em> when mimicking real world atmospheric conditions through temperature and humidity control. </p>

The estimates of global black carbon direct radiative forcing are still subject to substantial uncertainties ranging between 0.20 and 0.42 Wm-2 . In order to reduce these uncertainties, we must improve the representation of complex black carbon mixing state when simulating their optical properties and applying in global models. The main motivation behind this thesis is to gain a better understanding of the current state of the art when it comes to accurately modeling the radiative properties of BC-containing aerosols. We investigated how the optical properties of BC fractal aggregates are affected by the physiochemical properties that change during their ageing cycle, such as the fractal dimension, organic fraction, and particle size. Findings from the numerical study showed that the radiative forcings of black carbon particles can vary by as much as 61% depending on the particle's morphology and chemical composition. This influence of morphology and composition is particularly relevant to polluted urban environments with high mass concentrations of BC. Starting with soot from laboratory-based soot generators, representative for freshly emitted soot in the atmosphere, different morphological representations of BC and their modelling techniques were extensively tested. For BC particles with mobility diameters ³ 70nm and coating fractions £ 60%, the morphological representation as fractal aggregates performed well for modeling their optical properties. Our findings indicate that the use of the volume mean diameter of BC-containing particles, rather than the number mean diameter, is beneficial for calculating their optical properties. In addition, considering the particles to be fractal aggregates for the entire size distribution significantly reduced the uncertainty in optical simulations to nearly ten percent. Investigations conducted in Delhi, a polluted Asian urban environment, confirmed that fractal morphology plays a critical role in reducing the overestimation of aerosol light absorption simulations. When anthropogenic BC emissions are high, the particle light absorption can be overestimated by 50 to 200% by using assumptions of both external mixing and internal core-shell mixing. To address the computationally expensive nature of fractal simulations, we propose a metric known as the morphology index (MI). In order to obtain an accurate particle light absorption, this index distributes weightage between two extreme cases, the core-shell model and the fractal aggregate model. The measurements in Delhi showed that the aerosols transported over long distances had an MI of 0.78 ± 0.20, while the locally emitted aerosols had an MI of 0.48 ± 0.18. In order to identify the most relevant particle size mode for light absorption, a BC-based aerosol classification approach was developed. The method is based on correlations between BC mass concentrations and aerosol number concentrations at different particle sizes (BC-size correlation spectra). We introduce the concept of BC-size correlation spectra as a method for understanding the heterogeneous size-dependent distribution of aerosol light absorption as well as using it in conjunction with MI to better represent the optical properties of aerosols in different BC regimes. As a means of reducing the computational burden associated with optical simulations of fractal BC particles, a machine learning-based application was developed. In order to train the model, a comprehensive dataset was derived from the commonly used but computationally expensive multi- sphere T-matrix method (MTSM). This fast machine learning algorithm can predict the optical properties of coated BC fractal aggregates at different stages of atmospheric ageing in a fraction of the time compared to time-consuming numerical scattering calculations, with an average deviation of less than one per cent.:Bibliographische Beschreibung 1 Bibliographic Description 2 List of Figures 4 List of Symbols and Abbreviations 5 1. Introduction 7 2. Theoretical background 10 2.1 Aerosol properties 10 2.1.1 Physiochemical properties of aerosols 10 2.1.2 Optical properties of aerosols 12 2.2 Black carbon fractal aggregates 13 3. Methodology 14 3.1 Overview of the methods 14 3.1.1 Simulating BC fractal aggregates 14 3.1.2 Particle light scattering calculations 15 3.1.3 Radiative forcing model 16 3.1.4 BC size-correlation spectrum 17 3.1.5 Morphology Index 17 3.1.6 Machine learning methods 18 3.2 Summary of Methodology for each Paper 19 4. Motivation and Main Conclusions for Paper I-IV 21 4.1 Paper I 21 4.2 Paper II 22 4.3 Paper III 23 4.4 Paper IV 24 5. Summary 25 6. Future Work 27 Bibliography 28 Appended Papers 33 List of Publications included in this dissertation 162 Author’s Contribution to the Papers 163 Publications not included in this dissertation 164 Acknowledgements 165 / Die Abschätzungen des Strahlungsantriebs von Schwarzem Kohlenstoff (engl. Black Carbon, BC) sind immer noch mit erheblichen Unsicherheiten behaftet, die zwischen 0.20 und 0.42 Wm-2 liegen. Um diese Unsicherheiten zu verringern, müssen wir die Darstellung des komplexen Mischungszustands von schwarzem Kohlenstoff bei der Simulation seiner optischen Eigenschaften in globalen Modellen verbessern. Die Hauptmotivation für diese Arbeit ist es, ein besseres Verständnis des aktuellen Stands des Wissens zu erlangen, wenn es um die genaue Modellierung der Strahlungseigenschaften von BC-haltigen Aerosolen geht. Es wurde ein grundlegendes Verständnis dafür geschaffen, wie die optischen Eigenschaften fraktaler BC-Aggregate durch die physiochemischen Eigenschaften beeinflusst werden, die sich während ihres Alterungszyklus ändern, wie die fraktale Dimension, der organische Anteil und der Partikelgröße. Die Ergebnisse der numerischen Studie zeigen, dass die Strahlungsantriebe von schwarzem Kohlenstoff je nach Morphologie und chemischer Zusammensetzung der Partikel um bis zu 61 % variieren können. Dieser Einfluss von Morphologie und Zusammensetzung ist besonders relevant für verschmutzte städtische Umgebungen mit hohen Massenkonzentrationen von BC. Ausgehend von Labormessungen mit Ruß-Generatoren, deren Emissionen frisch emittierten Ruß in der Atmosphäre entsprechen, wurden verschiedene morphologische Darstellungen von BC und deren Modellierungstechniken in umfassend getestet. Für BC-Partikel mit Mobilitätsdurchmessern ³ 70 nm und Beschichtungsanteilen £ 60% erwies sich die morphologische Darstellung durch fraktale Aggregate für die Modellierung ihrer optischen Eigenschaften als gut geeignet. Unsere Ergebnisse deuten darauf hin, dass die Verwendung des mittleren Volumendurchmessers von BC- haltigen Partikeln anstelle des mittleren Größendurchmessers für die Berechnung optischer Eigenschaften von Vorteil ist. Wenn die Partikel für die gesamte Größenverteilung als fraktale Aggregate betrachtet werden, reduziert sich die Unsicherheit in den optischen Simulationen auf fast zehn Prozent. Untersuchungen in Delhi, einer verschmutzten asiatischen städtischen Umgebung, bestätigten, dass die fraktale Morphologie eine entscheidende Rolle bei der Verringerung der Überschätzung von Simulationen zur Lichtabsorption von Aerosolen spielt. Wenn die anthropogenen BC- Emissionen in Delhi hoch sind, kann die Lichtabsorption der Partikel um 50 bis 200 % überschätzt werden, wenn sowohl eine externe als auch eine interne Kern-Schale-Mischung angenommen wird. Um dem hohen Rechenaufwand fraktaler Simulationen zu begegnen, schlagen wir eine Metrik vor, die wir als Morphologieindex (MI) bezeichnen. Um eine genaue Lichtabsorption der Partikel zu erhalten, verteilt dieser Index die Wichtung zwischen zwei Extremfällen, dem Kern-Schale Modell und dem Modell der fraktalen Aggregate. Die Messungen in Delhi ergaben, dass die über weite Strecken transportierten Aerosole einen MI von 0.78 ± 0.20 aufwiesen, während die lokal emittierten Aerosole einen MI von 0.48 ± 0.18 hatten. Um den für die Lichtabsorption wichtigsten Partikelgrößenbereich zu ermitteln, wurde ein BC- basierter Aerosolklassifizierungsansatz entwickelt. Die Methode basiert auf Korrelationen zwischen BC-Massekonzentrationen und Aerosolanzahlkonzentrationen bei verschiedenen Partikelgrößen (BC-Größen-Korrelationsspektren). Wir führen das Konzept der BC-Größenkorrelationsspektren als Methode zum Verständnis der heterogenen größenabhängigen Verteilung der Lichtabsorption von Aerosolen ein und nutzen es in Verbindung mit MI zur verbesserten Darstellung der optischen Eigenschaften von Aerosolen in verschiedenen BC-Regimen. Um den mit optischen Simulationen fraktaler BC-Partikel verbundenen Rechenaufwand zu verringern, wurde ein auf maschinellem Lernen basierende Anwendung entwickelt. Dieser schnelle Algorithmus für maschinelles Lernen kann die optischen Eigenschaften beschichteter fraktaler BC- Aggregate in verschiedenen Stadien der atmosphärischen Alterung in einem Bruchteil der Zeit berechnen, die im Vergleich zu zeitaufwändigen numerischen Streuungsberechnungen erforderlich ist, und das mit einer durchschnittlichen Abweichung von weniger als einem Prozent.:Bibliographische Beschreibung 1 Bibliographic Description 2 List of Figures 4 List of Symbols and Abbreviations 5 1. Introduction 7 2. Theoretical background 10 2.1 Aerosol properties 10 2.1.1 Physiochemical properties of aerosols 10 2.1.2 Optical properties of aerosols 12 2.2 Black carbon fractal aggregates 13 3. Methodology 14 3.1 Overview of the methods 14 3.1.1 Simulating BC fractal aggregates 14 3.1.2 Particle light scattering calculations 15 3.1.3 Radiative forcing model 16 3.1.4 BC size-correlation spectrum 17 3.1.5 Morphology Index 17 3.1.6 Machine learning methods 18 3.2 Summary of Methodology for each Paper 19 4. Motivation and Main Conclusions for Paper I-IV 21 4.1 Paper I 21 4.2 Paper II 22 4.3 Paper III 23 4.4 Paper IV 24 5. Summary 25 6. Future Work 27 Bibliography 28 Appended Papers 33 List of Publications included in this dissertation 162 Author’s Contribution to the Papers 163 Publications not included in this dissertation 164 Acknowledgements 165

The Moderate-resolution Imaging Spectroradiometer (MODIS) onboard NASA’s Terra and Aqua satellites have provided a global distribution of aerosols. The space-based inversion of MODIS measurements requires assumption about the surface and aerosol properties, both are highly heterogeneous in space and time. This thesis has investigated the role of surface reflectance and aerosol properties on the retrieval of aerosols from MODIS over the Indian region. The aerosol properties retrieved by MODIS including total aerosol optical depth (AOD) and aerosol fine mode fraction (AFMF, fractional contribution of fine mode aerosols in the total AOD) were compared with that obtained from Aerosol Robotic Network (AERONET) at Kanpur (26.45◦N,80.35◦E), Indo-Gangetic Basin, northern India. This region is a special region for the study of aerosols as it offers strong aerosol seasonality, where the region is influenced by dust aerosols during pre-monsoon (March to June) and dominated by the fine mode particles in winter (November to February). The MODIS Collection 004 (C004) aerosol products systematically overestimated AOD in the presence of dust and underestimated when fine particles were dominant. The errors in the retrieval of dust AOD were correlated with the apparent reflectance at 2.1 µm, from which the surface reflectance in the visible channels (0.47 µm and 0.66 µm) were estimated using the “dark target” spectral correlation method. The error in the retrieval of AOD were also found to be large in the scattering angle range 120◦150◦, where the scattering properties of the non-spherical dust aerosols differ from that of the assumed spherical particles. AFMF of C004 was found to be highly biased to fine mode at Kanpur. The Collection 005 (C005) aerosol retrieval of the second-generation aerosol algorithm, however, showed improved retrieval of spectral AOD, which is likely to be attributed to the use of updated aerosol models and parameterized surface reflectance. In contrast to the C004 products, fine AOD and fine-model weighting (FMW) of C005 were biased very low at Kanpur and also over the greater Indian land region. This has indicated that the inversion of the space-based MODIS measurements is non-unique in which an improper combination of surface reflectance and aerosol model provide more accurate retrieval of the total aerosol optical depth. The surface reflectance relationships between the visible and shortwave-infrared 2.1 µm channels derived from the actual measurements of the surface reflectance using a spectroradiometer onboard an aircraft over Bangalore (12.95◦N,77.65◦E) in the southern India were found to have higher slope and intercept than that assumed by the MODIS algorithm over the same region. The high spectral correlations between the measured reflectance at longer wavelengths indicated some potential to estimate the surface reflectance at these wavelengths which needs further investigation. An experiment on the retrieval of aerosols carried out with several combinations of aerosol models and visible surface reflectance clearly shown that the surface reflectance in the visible channels assumed in the MODIS aerosol algorithm should be increased from its current parameterization in order to retrieve more accurate total as well as size-segregated aerosol optical properties at Kanpur and also over the greater Indian land region. In addition to the visible channels, inclusion of longer wavelengths in the aerosol inversion would likely improve the accuracy of retrieval over land by resolving the spectral dependence of aerosols. This in turn can help in separating the anthropogenic and natural aerosols in the total aerosol loading.

In order to reduce uncertainty of atmospheric particle emissions and to examine the mechanism of new particle formation from precursor gases, measurements were conducted in a megacity (Paris, France), an urban area (Patras, Greece) and a remote location (Finokalia, Greece). At Finokalia, the composition of particles with diameter smaller than 1 μm (PM1) depended on air mass origin. The highest concentrations, and most frequent, were observed when air masses were coming from Europe. Organic aerosol was found to be 80% water soluble and the increased organic to elemental carbon ratio correlated with ozone concentration. These findings indicate that particulate matter (PM) at Finokalia was not emitted near the site but was transported from source regions hunderd of kilometers away and thus the area can be considered as a background of Europe. At Finokalia, atmospheric nucleation was observed more frequently during winter when sunlight intensity was below average and favored by air masses that crossed land before reaching the site. This behavior was explained by ammonia involvement in the nucleation process. PM1 was mainly acidic during summer and consumed all available ammonia, contrary to winter when, due to the lower sunlight intensity, particles were neutral and ammonia was available. During both seasons nucleation would only occur if particles were neutral which resulted in higher frequency of events during winter. Air masses that crossed land before reaching the site were enriched with ammonia, thus it was more likely for nucleation to occur. Number size distributions were monitored in Paris, France at fixed and mobile ground stations along with airborne measurements. The Paris plume was identified at a distance of at least 200 km from the city center and the number concentration was found to increase even by a 3-fold when air masses crossed Paris. During summer nucleation was observed approximately half of the campaign days; when the condensational sink was lower than average contrary to winter when no event was identified due to higher sink. Increased number concentration was observed at an altitude outside of the Paris plume simultaneously with new particle formation observed on the ground and was attributed to that phenomenon. At Patras, the legislated by E.U. daily PM10 standards were found to be violated. Exceedances were more frequent (58 of a total of 75) during the colder months (October to March) of the year. The warmer months (April to September) 80% of the PM2.5 was transported from other areas. Contrary during the colder months the contribution of transported PM reduced to 70% during autumn and 50% during winter, when the highest concentrations were observed on average. Local traffic contributed approximately 15% during winter and the remaining 35% was primarily due to domestic heating. PM2.5 and PM1 concentrations were found to exceed 100 μg m-3 on several occasions during nighttime due to domestic heating, either diesel or biomass combustion. Potassium, a tracer of biomass combustion, correlated well (R2=0.79) with PM2.5 during winter indicating a biomass source. Potassium concentrations were higher within the urban premises than a rural area located 36 km away from the city, indicating that at least a portion of the biomass combustion related PM2.5 were emitted locally. / Με σκοπό την μείωση της αβεβαιότητας των εκπομπών ατμοσφαιρικών σωματιδίων (ΑΣ) καθώς και διευκρίνισης του μηχανισμού σχηματισμού ΑΣ από την οξείδωση πρόδρομων αερίων, μετρήσεις πεδίου έλαβαν χώρα σε μία μεγαλούπολη (Παρίσι, Γαλλία), μία αστική περιοχή (Πάτρα, Ελλάδα) και σε μία απομακρυσμένη τοποθεσία (Φινοκαλιά, Ελλάδα). Στην Φινοκαλιά, η σύσταση των σωματιδίων με διάμετρο μικρότερη από 1 μm (ΑΣ1) εξαρτιόταν από την προέλευση των αερίων μαζών. Τις υψηλότερες συγκεντρώσεις εμφάνιζαν οι αέριες μάζες από τη Ευρώπη, που ήταν και οι πιο συχνές. Οργανικές ενώσεις των ΑΣ, εμφάνιζαν, υψηλή διαλυτότητα στο νερό (80%) και αυξημένο λόγο οργανικού προς στοιχειακό άνθρακα που συσχετιζόταν θετικά με τις συγκεντρώσεις όζοντος. Όλα τα παραπάνω υποδεικνύουν πως τα ΑΣ στην περιοχή της Φινοκαλιάς μεταφέρονταν από γειτονικές περιοχές εκατοντάδες χιλιόμετρα μακριά και συνεπώς η περιοχή μπορεί να θεωρηθεί ως σταθμός υποβάθρου για την Ευρώπη. Στην Φινοκαλιά, το φαινόμενο της ατμοσφαιρικής πυρηνογένεσης ήταν συχνότερο τους χειμερινούς μήνες, όταν η ένταση φωτός ήταν χαμηλότερη, και σε αέριες μάζες που παρέμεναν σημαντικό χρόνο πάνω από την στεριά πριν φτάσουν στον σταθμό. Αυτή η συμπεριφορά εξηγήθηκε με την συμμετοχή της αμμωνίας στην διαδικασία της πυρηνογένεσης. Τα ΑΣ1 το καλοκαίρι ήταν κατά κανόνα όξινα και κατανάλωναν όλη την διαθέσιμη αμμωνία σε αντίθεση με τον χειμώνα, όπου εξαιτίας της χαμηλότερης έντασης φωτός, τα ΑΣ1 ήταν ουδέτερα και υπήρχε διαθέσιμη. Και στις δύο περιόδους η πυρηνογένεση λάμβανε χώρα μόνο όταν τα σωματίδια ήταν ουδέτερα, το οποίο είχε ως αποτέλεσμα υψηλότερη συχνότητα του φαινομένου τους χειμερινούς μήνες. Οι αέριες μάζες όταν παρέμεναν πάνω από στεριά εμπλουτίζονταν με αμμωνία, αυξάνοντας την πιθανότητα πυρηνογένεσης. Κατανομές μεγέθους αριθμού μετρήθηκαν στο Παρίσι, Γαλλίας σε επίγειους σταθμούς, σταθερούς και κινητούς, καθώς και σε υψόμετρο. Ο θύσανος του Παρισιού ταυτοποιήθηκε σε απόσταση τουλάχιστον 200 km από την πόλη και οι συγκεντρώσεις αριθμού ΑΣ αύξαναν ακόμα και κατά 300% όταν οι αέριες μάζες προέρχονταν από το Παρίσι. Το καλοκαίρι πυρηνογένεση έλαβε χώρα τις μισές μέρες της δειγματοληψίας, όταν η διαθέσιμη επιφάνειας συμπύκνωσης ήταν χαμηλή, ενώ το χειμώνα, επειδή η διαθέσιμη επιφάνεια ήταν υψηλότερη, δεν ταυτοποιήθηκε το φαινόμενο. Αυξημένες συγκεντρώσεις αριθμού ΑΣ ταυτοποιήθηκαν εκτός του θυσάνου του Παρισιού ταυτόχρονα με πυρηνογένεση στο έδαφος και αποδόθηκαν σε αυτό το φαινόμενο. Στην Πάτρα τα θεσμοθετημένα από την Ε.Ε. ημερήσια όρια ΑΣ10 βρέθηκαν να παραβιάζονται. Οι υπερβάσεις ήταν πιο συχνές (58 από τις 75) τους ψυχρούς μήνες (Οκτώβριο - Μάρτιο). Τους θερμούς μήνες (Απρίλιο - Σεπτέμβριο) το 80% των ΑΣ2.5 μεταφέρονταν από άλλες περιοχές. Αντίθετα τους ψυχρούς μήνες η συνεισφορά από μεταφερόμενα ΑΣ μειωνόταν στο 70% το φθινόπωρο και 50% το χειμώνα, όταν και οι συγκεντρώσεις ΑΣ2.5 ήταν κατά μέσο όρο οι υψηλότερες στην περιοχή. Η τοπική κυκλοφορία συνείσφερε περίπου 15% τον χειμώνα ενώ ένα σημαντικό κομμάτι από το υπόλοιπο 35% οφειλόταν στην οικιακή θέρμανση. Συγκέντρωση ΑΣ2.5 και ΑΣ1 ίση ή μεγαλύτερη των 100 μg m-3 μετρήθηκε κατ'επανάληψη τις νυχτερινές ώρες των χειμερινών μηνών εξαιτίας της οικιακής θέρμανσης, είτε με πετρέλαιο είτε με καύση βιομάζας Η καύση βιομάζας υποδεικνύεται από την συσχέτιση (R2=0.79) των συγκεντρώσεων ΑΣ2.5 με τις συγκεντρώσεις καλίου, ένα δείκτη καύσης βιομάζας. Οι συγκεντρώσεις αυτού του δείκτη βρέθηκαν υψηλότερες μέσα στον αστικό ιστό από μία αγροτική περιοχή 36 km μακριά από την Πάτρα, αποκλείοντας την αποκλειστική μεταφορά ΑΣ2.5 καύσης βιομάζας από γειτονικές περιοχές.

Les nuages sont essentiels dans le cycle de l'eau et dans le budget climatique mais certains aspects de leur formation sont encore incompris. La théorie de Köhler prédit que les surfactants devraient favoriser l'activation des particules en goutte de nuage alors que les modèles actuels les considèrent comme négligeables. Au début de ce travail de thèse, quelques études commençaient à démontrer le contraire mais des preuves du rôle de ces composés dans l'atmosphère étaient encore manquantes, d'où l'objectif de ce travail de thèse. Le développement d'une méthode pour déterminer la concentration en surfactants dans les aérosols a conduit aux premières courbes de tension de surface de surfactants atmosphériques dans des aérosols PM2.5 côtiers (Suède), et à l'identification du ratio C/CMC comme paramètre clé contrôlant l'efficacité des aérosols à former des nuages. Une seconde étude a révélé des corrélations fortes entre la présence de nuages et les propriétés intrinsèques des surfactants dans des aérosols PM1 boréaux (Finlande), démontrant pour la première fois le rôle des surfactants dans la formation des gouttes de nuage à partir d'observations directes dans l'atmosphère. Les résultats prédisent un nombre de noyaux de condensation en moyenne quatre fois plus important que lorsque les effets des surfactants étaient négligés, montrant l‘importance d'inclure l'effet des surfactants dans les modèles prédictifs. Cette importance a été confirmée en laboratoire par des expériences sur des gouttes individuelles microniques par l'augmentation de leur taille en présence de surfactants. Enfin, les observations à partir des différentes études indiquent une origine biologique des surfactants dans les aérosols atmosphériques / Clouds are essential components of the Earth’s hydrological system and climate but some aspects of their formation are still not completely understood. In particular, although Köhler theory predicts that surfactants should enhance cloud droplet activation, current models consider this role negligible. At the time of this PhD work, a few studies had started to demonstrate the contrary but atmospheric evidence for the role of these compounds was still missing and very little was known about their atmospheric concentrations, sources, and mechanism of action. The objective of this PhD work was to investigate these aspects. A method was developed to quantify surfactant concentrations in aerosols. Its application led to the first absolute atmospheric surfactants’ surface tension curves, in coastal PM2.5 aerosols in Sweden, and to the identification of the ratio C/CMC as the key parameter controlling the cloud-forming efficiency of aerosols. A second study revealed strong correlations between cloud occurrence and intrinsic surfactant properties in boreal PM1 aerosols in Finland, demonstrating for the first time the role of surfactants in cloud formation from direct atmospheric observations. The results predicted Cloud Condensation Nuclei numbers four times larger on average than when neglecting surfactant effects, showing the importance of including surfactant effects in cloud predictions models. The role of surfactants inferred from macroscopic measurements was confirmed by laboratory experiments on individual micron-sized droplets showing an increase of droplet growth in the presence of surfactants. Finally, observations from the different field studies indicated a biological origin for the surfactants present in atmospheric aerosols

L'atmosphère est un milieu hautement hétérogène contenant de la matière condensée : les aérosols. Ceux-ci sont des composants importants de l'atmosphère car ils impactent le bilan radiatif planétaire mais aussi la qualité de l'air. En particulier les aérosols organiques secondaires (AOS), produits par la transformation chimique dans l'air de nombreux composés organiques, plus ou moins volatils, représentent une fraction conséquente dans le budget global des aérosols atmosphériques pour laquelle de nombreuses incertitudes persistent. En particulier, leurs voies de formation et de transformation dans la troposphère restent très mal décrites. C'est pourquoi, cette thèse décrit principalement l'étude de trois aspects de la croissance et du vieillissement (transformation) des aérosols : caractérisation de la croissance des AOS par des processus photosensibilisés ; investigations mécanistiques du vieillissement des AOS et de la photochimie des photosensibilisateurs ; analyse chimique des composés issus du vieillissement des AOS / Aerosols are important constituents of the atmosphere and secondary organic aerosols (SOA) represent a main fraction of the organic aerosols in the total budget. This thesis mainly reports the investigation of three aspects of the growth and aging of SOA: the photosensitized SOA growth ; the mechanistic investigation of SOA aging and of the photochemistry of photosensitizers ; the analysis of the chemical composition of aged SOA. The photosensitized growth and aging processes of SOA were investigated using an aerosol flow tube coupled with various aerosol and gas sensing instruments. For further analysis of the aerosol composition and a better understanding of the formation and growth of SOA in these experiments the aerosols produced in the dark and in the light were sampled on filters at the exit of the flow tube