Temperature Dependent Absorption Cross-Sections of O3 and NO2

” Temperature Dependent Absorption Cross-Sections of O3 and NO2 in the 240 - 790 nm range determined by using the GOME-2 Satellite Spectrometers for use in Remote Sensing Applications”

Dissertation zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften (Dr. rer. nat.) am Fachbereich Physik der Universit¨at Bremen

vorgelegt von Dipl.-Physiker Bilgehan G¨ur

Bremen, Februar 2006

1. Gutachter: Prof. Dr. rer. nat. John P. Burrows 2. Gutachter: Dr. rer. nat. habil. Johannes Orphal

Eingereicht am: 21.02.2006 Tag des Promotionskolloquims: 05.05.2006

Abstract

Absorption spectra of O3 and NO2 have been measured in three independent campaigns using the three highly stabilized and accurately characterized GOME-2 satellite spectrometers, ﬂight models FM2, FM2-1, and FM3. GOME-2 (Global Ozone Monitoring Experiment) is an enhanced follow-up project of GOME, which was launched on ESA’s second European Remote Sensing Satellite (ERS-2) in 1995. A new generation of satellites for earth observation will be available with the MetOp series, starting most likely in the second half of 2006. MetOp comprises three polar-orbiting satellites to be launched sequentially over 14 years. One of the operational instruments onboard these satellites will be GOME-2, a nadir-viewing spectrometer that observes solar radiation transmitted or scattered from the Earth atmosphere or from its surface.

Spectra were recorded at ﬁve temperatures (203 K, 223 K, 243 K, 273 K, and 293 K) for O3 and four temperatures (223 K, 243 K, 273 K and 293 K) for NO2 with a spectral coverage of 240 to 790 nm at a resolution of 0.24 to 0.53 nm full width at half maximum. The relative temperature dependences were determined.

The achieved set of laboratory measurements of O3 absorption spectra is in very good agreement (about 1-3 % at 293 K) with the recommended O3 absorption cross sections at 10 diﬀerent wavelengths. For the temperature dependence in the range of 400 to 450 nm an upper limit estimate of no more than 10 % decrease with falling temperature was found at 425 to 430 nm in disagreement with one previous publication. There is some evidence that continuous absorption as measured in the blue wing of the Chappuis band at 429.5 nm drops by a few percent when temperature is reduced from 293 K to 203 K. At the same time the peak of the band at 426.5 nm increases slightly by 1 %. This is supported by a clear increase of amplitude of diﬀerential absorption cross≈ section at 426 nm of 13 % with falling temperature. At 604.61nm at the top of the Chappuis band a slight increase of cross section with falling temperature of 1 % is found in agreement with three previous publications and in disagreement with on≈e most recent one.

The measurements of NO2 absorption spectra are in very good agreement with literature data and recommended NO2 absorption cross sections at ambient temperature (about 1-2 % in the main DOAS window between 400 and 500 nm). The determined temperature dependence is in agreement with previous observations, i.e. a linear increase of the peak- to-valley absorbance diﬀerence with decreasing temperature from 293 K to 223 K.

The purpose of this study is mainly focused on the clariﬁcation of the above mentioned open issues regarding the O3 absorption cross sections. Nevertheless both sets of newly achieved absorption spectra of O3 and NO2 are of high importance as reference data for remote sensing, and will be used to derive (with GOME-2) a detailed picture of the atmospheric content and proﬁle of O3, NO2 and other trace gases. It will furthermore help to validate and improve the spectroscopic database.

Contents

1 Introduction and Motivation 5

2 Atmospheric Chemistry 13 2.1 TheEarth’sAtmosphere ...... 13 2.2 StratosphericChemistry ...... 16

2.2.1 O3 in the Global Stratosphere and the Role of NO2 ...... 16 2.2.2 Ozone Depletion in the Polar Regions ...... 18 2.2.3 Ozone Depletion in Mid Latitudes ...... 20 2.3 TroposphericChemistry ...... 21

2.3.1 O3 and NO2 intheTroposphere ...... 21

3 Relevant Aspects of Molecular Spectroscopy 23 3.1 MolecularSpectroscopy...... 23 3.1.1 ElectronicSpectra ...... 26 3.1.2 VibrationalSpectra...... 28 3.1.3 RotationalSpectra ...... 29 3.1.4 Symmetry Properties of Polyatomic Molecules ...... 32

3.2 Absorption-Spectrum of O3 ...... 35 3.3 Absorption-Spectrum of NO2 ...... 40

4 Experimental Setup 41 4.1 TheGOME-2Spectrometer ...... 41 4.2 TheCATGASSetup ...... 44 4.2.1 Light-Sourceandoptics ...... 45 4.2.2 Gas-Vessel...... 46

4.2.3 Gas-System O3 ...... 46 4.2.4 Gas-System NO2 ...... 51 4.3 TheOpticalInterface...... 58

5 Data-Aquisition 61 5.1 Measurement-ProcedurewithGOME-2 ...... 61

5.1.1 Documentation of measured Intensities I0 and I ...... 62 5.2 Reference-Interpolation and Calculation of Optical Densities ...... 66 5.3 ConcatenationofaSpectrum ...... 68 5.3.1 Quality analysis of overlap region ...... 72

3 5.4 Origin-Projects ...... 73 5.4.1 Data...... 73 5.4.2 Spectra ...... 75 5.4.3 Baselines...... 78

6 Wavelength Calibration 79

7 O3-Absorption Measurements with GOME-2 83 7.1 Integrated Absorption Cross Sections ...... 84

7.2 Absolute Scaling of the GOME-2 O3-Spectra...... 87 7.2.1 Final GOME-2 O3-Spectra ...... 92 7.3 Results and Comparison with Literature Data Base ...... 96 7.3.1 Comparison of GOME-2 spectra at ambient temperature with literature values at 10 single wavelengths ...... 97 7.3.2 Comparison of relative temperature dependence ...... 99

7.3.3 Modelling the temperature dependence of O3 by a polynomial ﬁt . . 106 7.3.4 Comparison of available spectra in the Huggins bands ...... 108 7.3.5 Integrated cross sections at diﬀerent temperatures ...... 112 7.4 ErrorAnalysisandPropagation ...... 115

8 NO2-Absorption Measurements with GOME-2 119 8.1 Correction of the N2O4-Absorption ...... 120 8.2 Absolute Scaling of the NO2 Spectra obtained from GOME-2 Measurements122 8.2.1 Final GOME-2 NO2-Spectra...... 122 8.3 Results and Comparison with Literature Data Base ...... 124 8.3.1 Comparison of GOME-2 data with literature at ambient temperature124

8.3.2 Temperature dependence of NO2 spectra obtained from the GOME- 2study ...... 130

9 Outlook 135

10 Conclusion 137

A Quality-Analysis Overlap Region 139

B I0 - I Documentation 145 Chapter 1

Introduction and Motivation

Although minor constituents, ozone (O3) and nitrogen dioxide (NO2) are central species in atmospheric processes and therefore of major interest in research activities. O3 had already been discovered in 1840 by C.F. Schönbein, who suggested the presence of an atmospheric gas having a peculiar odor. He called this gas ”ozone”, coming from ”ozein”, the Greek word for scent and smell. Since then, many studies have been performed to describe different aspects of ozone, such as molecular properties, the chemical basis of its existence in the atmosphere and its interaction with other atmospheric components. Spectroscopic studies showed already in the late 19th century that ozone is present at higher concentrations in the upper atmospheric layers than close to the ground, leading to the term ”ozone layer”. Nowadays it is not only in the scientific community well known that this layer is of high importance for human life on earth because of one decisive optical property of ozone, i.e. its absorption of biologically harmful ultra violet radiation.

In the 1970’s a research group from the British Antarctic Survey, who was monitoring the atmosphere above Antarctica, noticed a dramatic loss of ozone in the lower stratosphere over Antarctica. It is reported that when the first measurements were taken in 1985, the drop in ozone levels in the stratosphere was so dramatic that at first the scientists thought their instruments were faulty. Replacement instruments were built and flown out, and it wasn’t until they confirmed the earlier measurements, several months later, that the ozone depletion observed was accepted as genuine. Satellite measurements were becoming available at the same time and showed indeed also a massive depletion of ozone on a large scale over most of the Antarctica continent (figure 1.1). However, these were initially rejected as unreasonable by data quality control algorithms (they were filtered out as errors since the values were unexpectedly low); the ozone hole was only detected in satellite data when the raw data was reprocessed following evidence of an ozone hole in the above mentioned in situ observations. From thereon the ozone hole attracted worldwide publicity and caused increased research activities.

5 6 CHAPTER 1. INTRODUCTION AND MOTIVATION

A further atmospheric trace gas of importance is nitrogen dioxide NO2. Nitrogen oxides (NOx) in general are emitted by combustion processes and play a key part in the catalytic production of tropospheric ozone by photochemical processes. This aﬀects the local air quality as well as the global tropospheric chemistry. NO2 can therefore be used as an indicator for air pollution. Here as well satellite based observations provide such information on a global scale. Just recently Richter et al. [1]published results from measurements with the satellite instruments GOME (Global Ozone Monitoring Experiment) and SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric CHar- tographY), indicating an increase of tropospheric NO2 over China and showing also air polluted regions over Europe (ﬁgures 1.2 and 1.3).

Figure 1.1: Ozone depletion at 1980 and 1991 over the south pole, recorded with the satellite instrument TOMS (Total Ozone Mapping Spectrometer) [With friendly permission from the Center for Atmospheric Sciences at the University of Cambride]. 7

VC NO2 [molec / cm 2 ] > 4.0 x 1016

2.0 x 1016

16 e 1.0 x 10

t 15

a 5.0 x 10

L 2.0 x 1015 1.0 x 1015 5.0 x 1014

Longitude

Figure 1.2: Tropospheric column amounts of NO2 over China retrieved from the satellite instruments GOME and SCIAMACHY [1].

VC NO2 [molec / cm 2 ] > 4.0 x 1016

2.0 x 1016

e 1.0 x 10

i 15 t 5.0 x 10

L 2.0 x 1015 1.0 x 1015 5.0 x 1014

Longitude

Figure 1.3: Tropospheric column amounts of NO2 over Europe retrieved from the satellite instruments GOME and SCIAMACHY [1]. 8 CHAPTER 1. INTRODUCTION AND MOTIVATION

The importance of satellite based global observations of O3 and NO2 is evident. However, in order to achieve precise results the availability of accurate reference data is required. Highly accurate knowledge of absorption cross sections of atmospheric species, such as O3 and NO2 in this case, provide an essential fundament for remote sensing activities and monitoring atmospheric constituents and their distribution from space as well as from ground. Furthermore it gives useful information for kinetic studies and photolysis rates, when investigating these species in interaction with other atmospheric constituents. The question arises, why the spectra cannot be calculated theoretically, as it is partly done for infrared spectra ? Is an appropriate model available that would generate spectra close to experimental spectra ? This may be illustrated for instance with the O3 spectrum. Baloitcha and Balint-Kurti recently published results from ab initio calculations regarding the photo dissociation of ozone in the Hartley continuum [2]. Figure 1.4 illustrates that the comparison between computed and experimental ozone data in the Hartley continuum shows high discrepancies. At the current state it is therefore not possible to generate absorption spectra of ozone within the required accuracies of about 2-3 %, which is needed for the desired accuracy of 2-3 % in the atmospheric retrieval. The only feasible approach is to perform accurate experimental measurements. Laboratory measurements are therefore profound milestones on the way to improved understanding of atmospheric processes.

(a) Computed and experimental (Ref. [77]) (b) Computed and experimental (Ref. [77]) total absorption cross section for ozone in total absorption cross section for ozone in the region of the Hartley continuum. Ini- the region of the Hartley continuum. The tially, ozone was taken to be in its lowest upper-B˜ -state diabatic surface has been vibrational-rotational state with zero total shifted down by 0.002 95 hartree to bring angular momentum. the peak position into line with the experiment

Figure 1.4: Comparison between computed and experimental ozone spectra in the Hartley continuum [2]

Due to the mentioned importance of O3 for the filtering of UV radiation, O3 has been in the focus of several laboratory measurements. Nowadays many different measurements are available, which have been performed with different techniques and parameters and in different wavelength ranges. This is necessary due to different reasons: Firstly, the absorption cross sections vary over seven orders of magnitude in the 220-790 nm range, it is therefore not feasible to cover the complete range with just one measurement. Secondly 9

the diﬃcult handling of O3, which is unstable, corrosive and under certain circumstances even explosive.

There still exist discrepancies in the literature regarding absorption cross sections of O3. With respect to remote sensing two wavelength regions are of particular importance, i.e. 325-345 nm in the Huggins bands and 400-500 nm in the blue wing of the Chappuis band (a more detailed description of the O3 absorption spectrum will be given in chapter 3.2). While the temperature dependence in the Huggins bands is well established, the eﬀects in the Chappuis band remain unclear. Burkholder and Talukdar observed a strong decrease of 40 % at 420 nm with decreasing temperature between 293 and 220 K [81]. Just re- ≈ cently Helou et al. published measurements of absolute absorption cross sections of O3 in the Chappuis band, observing an about 3 % decrease of the cross sections in the peak of the Chappuis band around 600 nm with decreasing temperature from 293 K to 203 K [4]. This is in contradiction to previous measurements, in which a decrease of temperature in the same range resulted in a slight increase of absorption cross sections of not more than 1 %. These topics will be discussed in detail in chapter 7.3. ≈

The current scientific state regarding the absorption cross sections of NO2 is less critical compared to O3. Nevertheless it is of particular advantage, when performing dedicated on-ground and pre-launch absorption measurements with the satellite spectrometer. During the development of GOME, a recommendation was made by the GOME Science Advisory Group (GSAG)/Characterisation and Calibration Sub-group, that the temperature dependent trace gas absorption spectra should be measured under representative in-flight conditions with the GOME Flight Model. Following this recommendation the Molecular Spectroscopy Group at the Institute of Environmental Physics in Bremen developed for this purpose a mobile gas phase absorption spectroscopy set-up including stabilized optics, reaction vessel, and stabilized gas handling system, called CATGAS (Calibration Apparatus for Trace Gas Absorption Spectroscopy). Since then CATGAS has been used for on-ground measurements with the satellite spectrometers GOME, SCIAMACHY and GOME-2, an enhanced follow-up project of GOME. Three independent measurement campaigns with three flight models of the GOME-2 series were performed, which are presented in this thesis and form the foundation for this study.

In parallel to the GOME project a study at the Institute of Environmental physics yielded high resolution absorption spectra of NO2 and O3 at a selected set of temperatures. One of the intentions of this study was to provide the data to investigate the accuracy of combining an accurate knowledge of the instrument response function coupled with high resolution gas phase absorption spectra to reduce the requirement for the measurement of trace gas spectra with the FM during calibration. In the case of unstable absorbers (e.g., free radicals BrO and OClO) it is diﬃcult to measure reference spectra using the Flight Models at all, as either time resolved measurement modes for the FM would be needed or impractically high absorber concentrations were needed.

In practice at least for the dominant absorbers in the spectral range to be measured by GOME, the error on the reference spectra has not been reduced by using the two step pro- 10 CHAPTER 1. INTRODUCTION AND MOTIVATION

cedure. The reason is that it is not feasible to measure the instrument response function of the spectrometers with suﬃcient accuracy. So in spite of their limitations the GOME spectra still provide the best reference spectra for the GOME analysis. The GOME spectra have therefore been used as the reference spectra in the generation of GOME Level 2 data products.

The above result has been conﬁrmed by a study commissioned by ESA and entitled ”A Comparison between Predicted and Measured GOME Spectra”. This investigation has shown explicitly that signiﬁcant errors can be introduced if the high resolution reference spectrum is convoluted with the instrument response function, rather than absorption spectra measured with the Flight Models themselves are used as reference spectra for retrieval algorithms. These errors are dependent on the knowledge of the characterization of the instrument response function.

In conclusion in order to maximize the information content and to guarantee the quality of Level 2 data products from GOME-2, it is necessary to make on-ground pre-ﬂight measurements of the trace gas absorption spectra with the GOME-2 Flight Models under operational temperature and pressure conditions (i.e. under vacuum, the instrument being at approximately 5◦C).

This is schematically illustrated in ﬁgure 1.5. The data are retrieved from the satellite, analyzed and columns of O3 and NO2 are determined. The analysis of the data requires reference spectra, the accuracy of the determined columns highly depend on the accuracy of the reference spectra. One way of providing these spectra is by performing laboratory measurements of high resolution spectra. This requires though a good knowledge of the instrumental line shape of the observing instrument, with which the highly resolved spectra are convoluted. The knowledge of this instrumental line shape though is often inadequate. The alternative is therefore provided by CATGAS. With CATGAS measurement of temperature dependent trace gas absorption spectra under representative in-ﬂight conditions with the Flight Model can be performed, resulting in reference absorption spectra, in which the information regarding the instrumental line shape is already included 11

Laboratory Measurements Instrumental CATGAS of High Resolution Spectra Line Shape (Calibration Apparatus for Trace (requires good knowledge (definition often about instrumental line shape) inadequate) Gas Absorption Spectroscopy)

Measurements of temperature dependent trace gas absorption spectra under representative in-flight conditions with the satellite instrument Æ Spectra contain ILS .

Convolution

O 3 - Reference-Spectra O 3 - Reference-Spectra

Data - Retrieval and Analysis (Reference Spectra required)

Determination of O3 - Columns in the Atmosphere

Figure 1.5: Principal scheme of satellite based remote sensing. Details are given in the text

Chapter 2

Atmospheric Chemistry

In the following sections an overview shall be given over the Earth’s atmosphere, atmospheric chemistry and the physical laws applied in this study. This is not only to provide a theoretical background of the present work but also to underline the motivation of this study and the importance of accurate knowledge of absorption cross sections of atmospheric trace gases. The focus will be on O3 and NO2, of which absorption cross sections were measured with three ﬂight models of the GOME-2 satellite spectrometer series.

2.1 The Earth’s Atmosphere

The Earth is surrounded by a blanket of air, which is called the atmosphere and which reaches over 560 kilometers from the surface of the Earth. Early attempts at studying the nature of the atmosphere used clues from the weather, the beautiful multi-colored sunsets and sunrises, and the twinkling of stars. Today, with the use of sensitive instruments from space, we are able to get a much better view of the functioning of our atmosphere. Life on Earth is supported by the atmosphere, solar energy, and our planet’s magnetic ﬁelds. The atmosphere regulates the energy budget, recycles water and other chemicals, and works with the electrical and magnetic forces to provide a moderate climate. It also protects us from high-energy radiation of the sun and the frigid vacuum of space.

The atmosphere can be described as a composition of different layers, i.e. (from the ground up) the troposphere, stratosphere, magnetosphere and the ionosphere. These layers differ in several parameters, such as thermal characteristics (temperature changes), chemical composition, movement and density. Consequently each layer attach a different importance to the provision of required conditions for human life on earth.

The layer structure is illustrated in ﬁgure 2.1, followed by a brief description of each layer. Section 2.2 and 2.3 will then focus on the chemistry of the stratosphere and troposphere, since the trace gases O3 and NO2 play a signiﬁcant role in these atmospheric layers with important consequences on the conditions of human life.

13 14 CHAPTER 2. ATMOSPHERIC CHEMISTRY

120 110 Thermosphere 100 90

80 Mesopause

) T 70 emp m era k tur ( e Mesosphere

e 60

t l 50 Stratopause

A 40 Stratosphere 30 ozone layer 20

10 Tropopause Troposphere

-100 -90 -80 -70 -60 -50 -40-30-20-10 0 10 20 30 Temperature (°C)

Figure 2.1: Layers of the Earth atmosphere

Troposphere The troposphere starts at the Earth’s surface and extends 8 to 14 • kilometers high. This part of the atmosphere is the most dense, the air pressure at the top of the troposphere is only 10 % of that at sea level (0.1 atmospheres). As shown in ﬁgure 2.1 the temperature in the troposphere falls (in general) with increasing altitude, on average from about 17 degrees Celcius at the surface (due to strong heating from absorption of radiation) down to -50 to -60 degrees at the top of the troposphere. As a consequence of this temperature distribution vertical mixing occurs, species emitted at the surface can be transported quickly to upper layers. The troposphere is therefore of rather instable and chaotic character, causing a fast spreading of trace gases and pollution and furthermore being responsible for surface weather conditions (greek: tropos: turning, easily changing). The thin buﬀer zone between the troposphere and the next layer is called the tropopause.

Stratosphere The stratosphere starts just above the troposphere and extends to • 50 kilometers high. Compared to the troposphere, this part of the atmosphere is dry and less dense. One major characteristic of the stratosphere is the ”inversion”, i.e. a profile of increasing temperature with increasing altitude (figure 2.1). This results in a stable layer structure (lat.: stratum: layer), since an air mass of cooler air and being more dense will not rise through an upper warmer layer. Therefore only limited vertical mixing occurs in the stratosphere. As a consequence pollutants emitted from different sources (airplanes, volcanoes, etc.) persist for a long period of time in the stratosphere. 2.1. THE EARTH’S ATMOSPHERE 15

The described temperature distribution is caused by a thin layer with high concentrations of ozone, which absorbs and scatters the solar ultraviolet radiation. This will be illustrated in more detail in the following chapter. The UV absorption by the ozone layer results in a gradual increase of the temperature in this region to -3 degrees Celsius. 99 % percent of ”air” is located in the troposphere and stratosphere. The upper limit of the stratosphere is called the stratopause.

Mesosphere The mesosphere starts just above the stratosphere and extends to 80 • kilometers high. In this region, the temperatures fall again as low as -90 degrees ≈ Celsius with increasing altitude. Due to the much higher energetic solar radiation that reaches the mesosphere, chemicals are in an excited state. The mesopause sep- arates the mesosphere from the thermosphere. The regions of the stratosphere and the mesosphere, along with the stratopause and mesopause, are called the middle atmosphere.

Thermosphere The thermosphere (or ionosphere) starts just above the mesosphere • and extends to 600 kilometers high. Thermospheric temperatures increase with altitude due to absorption of highly energetic solar radiation by the small amount of residual oxygen still present. Temperatures in this region can go as high as 1700 degrees Celsius, depending on solar activity. Chemical reactions occur much faster than on the surface of the Earth. This layer is known as the upper atmosphere. The Thermosphere is very thin, but it is where aurora take place. The structure of the ionosphere is strongly inﬂuenced by the charged particle wind from the sun (solar wind), which is in turn governed by the level of solar activity.

The atmosphere is primarily composed of Ni- trogen (N2, 78%), Oxygen (O2, 21%), and Ar- gon (Ar, 1%). A myriad of other very inﬂuen- tial components and trace gases are also present which include water (H2O, 0 - 7%), ”greenhouse” gases or Ozone (O3, 0 - 0.01%), Carbon Dioxide (CO2, 0.01-0.1%) and others. Beyond the At- mosphere the exosphere starts at the top to the thermosphere and continues until it merges with interplanetary gases, or space. In this region of the atmosphere, Hydrogen and Helium are Figure 2.2: Constituents of the Atmo- the prime components due to their low atomic sphere masses and are only present at extremely low densities. 16 CHAPTER 2. ATMOSPHERIC CHEMISTRY

2.2 Stratospheric Chemistry

This chapter gives an overview about O3 and NO2 related atmospheric chemistry. These trace gases play a central part in this ﬁeld despite of their relatively low concentrations. A more detailed and extensive lecture in atmospheric chemistry can be found in ref. [5].

2.2.1 O3 in the Global Stratosphere and the Role of NO2

O3 is a central species in stratospheric chemistry. This results from its interaction with other stratospheric components and its signiﬁcant absorption of biologically harmful ultraviolet radiation. The stratospheric ozone layer (ﬁgure 2.1) contains about 90 % of the atmospheric ozone. Ozone originates in the stratosphere through the photo dissociation of molecular oxygen by ultra violet radiation. This process generates two oxygen atoms, which react with molecular oxygen to ozone:

O + hν(λ 242nm) O + O (2.1) 2 ≤ → 2[O + O +M O + M] (2.2) 2 → 3 Net : 3O +hν 2O (2.3) 2 → 3 M is required as a reactant for the conservation of energy and momentum. The major part of oxygen atoms in the atmosphere is in the form of molecular O2. This indicates processes where O3 is formed back to O2. Sidney Chapman, atmospheric scientist at the university Oxford, suggested 1930 that the recovery occurs through the following reaction sequence [6]:

O +hν(λ 1140nm) O+O (2.4) 3 ≤ → 2 O+O 2O (2.5) 3 → 2 Net : 2O +hν 3O (2.6) 3 → 2 Equation 2.3 and 2.6 (with its sub-reactions) are known as the Chapman cycle, a sequence, which is suited to produce naturally a steady state concentration of O3 in the stratosphere. In the 1960’s it became more evident that the observed O3 columns in the stratosphere were smaller than predictions based on the Chapman cycle. The Chapman cycle therefore provides a qualitative explanation for the ozone layer in the stratosphere but is not sufficient to describe the efficiency of the observed loss mechanism. Bates and Nicolet [7] proposed catalytic cycles in which only small amounts of different components would play a significant role in the depletion of stratospheric O3. These cycles can be summarized as follows:

X+O XO+O (2.7) 3 → 2 O +hν O+O (2.8) 3 → 2 O+XO X+O (2.9) → 2 Net : 2O +hν 3O (2.10) 3 → 2 2.2. STRATOSPHERIC CHEMISTRY 17

Such a mechanism would remove ”odd oxygen” from the Chapman cycle leading to a reduction of the eﬃciency of ozone formation. In the following a brief overview is given regarding several diﬀerent components, which are involved in the above mentioned reaction pattern. Please note that at this point the discussion is made with respect to the global stratosphere. The mechanism leading to ozone depletion in the polar regions and mid latitudes in particular will be discussed more detailed in the following chapters.

Nitrogen Dioxides NOx • In 1970 Paul J. Crutzen proposed nitrogen oxides (NOx), such as NO and NO2, to be an important catalysts in the stratosphere [8]. NOx is a product of N2O in the stratosphere, which originates mainly from emissions close to the ground from microbiological processes. Being rather inactive in the troposphere (lifetime around 150 years) N2O is decomposed in the stratosphere by reaction with energy rich oxygen atoms. N O+O(1D) 2NO (2.11) 2 → This process leads to NOx productions, initiating diﬀerent catalytic reaction cycles in which O3 removal occurs.

The following cycle illustrates the rather complicated structure with regards to NOx chemistry.

O + NO O + NO (2.12) 3 → 2 2 NO +O NO + O (2.13) 2 → 2 Net : O + O 2O (2.14) 3 → 2

2 (O + NO O + NO ) (2.15) × 3 → 2 2 NO +O NO + O (2.16) 2 → 2 NO +hν NO + O (2.17) 2 → 2 Net : 2O +hν 3O (2.18) 3 → 2

O + NO O + NO (2.19) 3 → 2 2 NO +O NO +O (2.20) 2 3 → 3 2 NO +hν NO + O (2.21) 3 → 2 Net : 2O +hν 3O (2.22) 3 → 2

Hydrogen Oxides HOx • A second catalytic chain of O3 depletion runs trough hydrogen oxides HOx. These species appear in the stratosphere mainly as a result of reactions of water vapor and methane with oxygen atoms and through the photolysis of H2O:

CH + O(1D) OH + CH (2.23) 4 → 3 H O+O(1D) 2OH (2.24) 2 → H O+hν OH + H (2.25) 2 → 18 CHAPTER 2. ATMOSPHERIC CHEMISTRY

The produced hydroxyl radicals initiate also diﬀerent cycles providing a further contribution in the O3 loss on a global scale:

O + OH O + H (2.26) → 2 H+O +M HO + M (2.27) 2 → 2 HO +O OH+2O (2.28) 2 3 → 2 Net : O + O 2O (2.29) 3 → 2

O + OH O + H (2.30) → 2 H+O OH+O (2.31) 3 → 2 Net : O + O 2O (2.32) 3 → 2

O + OH O + HO (2.33) 3 → 2 2 HO +O OH+O (2.34) 2 → 2 Net : O + O 2O (2.35) 3 → 2 Chlorine (Cl) • The halogen chlorine is involved in a further O3 depleting cycle. This cycle was introduced by Stolarski and Cicerone in 1974 [9], and associated with Chlorofluoro- carbons (CFC) in the same year by Molina and Rowland [10]. CFC’s are chemically very stable and do not react in the troposphere with hydroxyl radicals or other oxi- dating components. Of particular importance are CFCl3 and CF2Cl2. Within only a few years these gases ascend from their origin on the surface into the stratosphere. Above 20-25 km the solar radiation is sufficiently energetic to decompose these and to release chlorine atoms and chlorine monoxide molecules, which are highly efficient catalysts for ozone depletion. The cycle is as follows:

Cl+O ClO + O (2.36) 3 → 2 O +hν(λ 1140nm) O+O (2.37) 3 ≤ → 2 ClO + O Cl+O (2.38) → 2 Net : 2O +hν 3O (2.39) 3 → 2

A natural source of chlorine in the stratosphere are volcanoes and CH3Cl from the troposphere, which is released by seaweed. Chlorine from CFC’s, which have been released from anthropogenic sources led, compared to the natural sources, in the 90’s to a ﬁvefold concentration of chlorine in the stratosphere.

2.2.2 Ozone Depletion in the Polar Regions In 1985 Farman et al. reported a dramatic loss of ozone in the antartic region [11]. Obser- vations since the 70’s had shown a clear tendency of decreasing ozone concentrations in the arctic spring. This was conﬁrmed by satellite measurements (ﬁgure 1.1) and furthermore an extension over the complete antartic region was shown [12]. 2.2. STRATOSPHERIC CHEMISTRY 19

Despite of an increase of the above mentioned compounds in the stratosphere the magnitude of the ”ozone hole” appeared larger than predicted by different atmospheric models. Nowadays it is well known that the ozone loss over polar regions occurs through an interaction of dynamic and chemical processes. These processes shall be briefly described at this point. For a detailed description reference [5] may be recommended. The chemistry in the antarctic region results from quite particular meteorological conditions. During the ”polar night” the air masses cool down. Due to the resulting temperature gradient the atmospheric pressure within the polar region increases, causing strong downward and outward meridional winds leaving the polar region. These radial winds are diverted by the Coriolis effect leading to a rotation of the air masses around the pole center. This appearance is referred to as the polar vortex. This vortex isolates very efficiently the air masses over the antarctic region from the rest of the stratosphere, which allows photochemically active products to build up and lay one basis for ozone destruction. A further foundation is given through the extreme low temperatures in the stratosphere during the polar winter. These cause the generation of ”polar stratospheric clouds (PSC’s)”, frozen solutions of H2SO4 and HNO3 (Typ I) and water (Typ II). These PSC’s provide a surface for heterogenous chemistry:

N O + H O 2HNO (2.40) 2 5 2 → 3 ClONO + H O HOCl + HNO (2.41) 2 2 → 3 ClONO + HCl Cl + HNO (2.42) 2 → 2 3 HOCl + HCl Cl + H O (2.43) → 2 2 N O + HCl ClNO + HNO (2.44) 2 5 → 2 3 This heterogenous chemistry has decisive consequences. Reactions with low rate constants in the gas phase proceed now quite rapidly on PSC surfaces. For example: ClONO2 and HCl act as a non reactive chlorine reservoir. But on surfaces provided by PSC’s they are converted over the winter into photochemically active Cl2 and ClNO2 (eqn. 2.42-2.44). When the sun comes up in the spring, these species are rapidly photolyzed, initiating the in the foregoing section discussed chain of ozone destruction.

An additional effect of equations 2.40 to 2.44 is the removal of oxides of nitrogen (”denitrification”) intensifying the lack of possible chlorine reservoir compounds. In general these described mechanisms lead to a distinct different composition of the stratosphere at the end of the polar night, which is indicated as ”disturbed chemistry” (higher concentration of reactive halogen compounds and lower concentrations of nitrogen oxides and water vapor)

Due to the unique conditions in the polar vortex two further cycles are of particular importance for the ozone depletion: The ClO-ClO cycle [13] and the BrO-ClO cycle [14]:

ClO + ClO + M (ClO) + M (2.45) → 2 (ClO) +hν Cl + ClOO (2.46) 2 → 20 CHAPTER 2. ATMOSPHERIC CHEMISTRY

ClOO + M Cl+O + M (2.47) → 2 2 Cl+O (ClO) + O ) (2.48) × 3 → 2 Net : 2O 3O (2.49) 3 → 2

ClO + BrO + M Br + OClO (2.50) → Br + ClOO Br+Cl+O (2.51) → → 2 BrCl + O Br+Cl+O (2.52) → 2 → 2 Br+O BrO+O (2.53) 3 → 2 ClO ClO + O (2.54) 3 → 2 Net : 2O 3O (2.55) 3 → 2

The interaction of the described mechanisms lead to a very eﬃcient depletion of ozone within the polar vortex. Due to increasing temperatures in the antarctic spring this vortex eventually breaks down, which then allows a ﬂow of ozone and nitrogen oxide compounds from the mid latitudes and the creation of halogen reservoirs. The catalytic depletion stops and regeneration of the ozone layer takes place.

The ozone depletion in the northern polar region is less pronounced, although significant differences in the involved gas concentrations do not exist. It results rather from differences in the sea-land contrast in the northern hemisphere causing dynamic processes that differ from the ones above the antarctic region. This leads to a vortex above the northern polar region with less stability, which on its parts has further consequences. The temperatures in the northern polar stratosphere are not as low as in the antarctic stratosphere, resulting in a formation of PSC’s in a lesser extent. Therefore the above described halogen activation, denitrification and dehydration in the stratosphere, which initiates the ozone depletion, is less intense.

2.2.3 Ozone Depletion in Mid Latitudes

The chemistry in the mid latitude stratosphere follows in general the procedure discussed in the previous chapters. However the meteorological conditions and dynamics in the mid latitudes diﬀer clearly from the ones described for the polar regions, resulting in a lesser magnitude of the ozone depletion. Nevertheless it is important to note that increased UV radiation in the mid latitudes would have higher impact on human life than in the polar regions. The analysis of available data sets revealed a decrease of 4 % in the northern and 3-6 % in the southern mid latitudes [15]. Regular variations≈ were observed on a seasonal basis, with the strongest depletion in winter and spring. Several further processes, which are irregular and therefore less predictable inﬂuence the total ozone column in the atmosphere, such as solar activity, changing global wind conditions or volcano eruptions. 2.3. TROPOSPHERIC CHEMISTRY 21

2.3 Tropospheric Chemistry

In this section a brief overview is given regarding O3 and NO2 related chemistry that occurs in the troposphere. As shown in figure 2.1 the troposphere is the first atmospheric layer reaching a height of 8-12 km. Compared to the properties of the stratosphere it differs mainly in temperature distribution (decreasing temperature with increasing height, resulting in more dynamic vertical mixing processes) and solar radiation budget reaching the surface.

2.3.1 O3 and NO2 in the Troposphere 10 % of atmospheric ozone is present in the troposphere. Part of the tropospheric ozone originates from the stratosphere through exchanging processes, another is formed in the troposphere itself through photochemical processes. As mentioned before the ozone forming source in the atmosphere is that between atomic and molecular oxygen. Atomic oxygen results from the photo dissociation of molecular O2 below 242 nm. The sunlight reaching the lower layer of the atmosphere though exceeds only a range above 290 nm. Therefore the source of atomic oxygen in the troposphere can not be O2. The most important oxygen generating reaction is the photolysis of NO2 to NO and atomic oxygen, which forms in a following reaction with molecular oxygen ozone. Oxides of nitrogen play a central role in the chemistry of the troposphere. When NO and NO2 are present in sunlight, ozone formation occurs as a result of the photolysis of NO2 at wavelengths below 410 nm:

NO +hν( 410nm) NO + O(3P) (2.56) 2 ≤ → O(3P)+O +M O +M∗ (2.57) 2 → 3 M is a third molecule that absorbs the excess vibrational energy and thereby stabilizes the O3 molecule formed. In principle the produced NO could react back with ozone to NO2 and O2, which would build a photo stationary state.

NO + O NO +O (2.58) 3 → 2 2

But since NO also reacts with HO2 and other peroxy radicals (RO2) to NO2 the cycle leads eﬀectively to an ozone production.

NO + HO NO + OH (2.59) 2 → 2 NO + RO NO + RO (2.60) 2 → 2 A significant source of nitrogen oxides in the atmosphere are as mentioned combustion of fossil fuels in industry, household and traffic. Measurements of nitrogen oxides in the atmosphere therefore give an important indication regarding air quality and air pollution (figures 1.2 and 1.3).

Chapter 3

Relevant Aspects of Molecular Spectroscopy

This chapter contains an overview regarding relevant aspects of molecular spectroscopy, which provide the theoretical background and foundation for this study. The reader may be aware that in this thesis the molecular spectroscopy can not be dealt with within its full complexity. For a detailed and extensive description the author would like to refer to standard literature, in particular though to ref. [16] and [17].

The ﬁrst section, chapter 3.1, describes the fundamentals of molecular spectroscopy and gives an overview about the origin of electronic, vibrational and rotational spectra, which are located in the UV, visible and near IR wavelength region (representing the observing window of the GOME-2 satellite spectrometer). A further subsection summarizes geometrical symmetry considerations of molecular structures, which will be needed for the description of the O3 and NO2 absorption spectra in chapter 3.2 and 3.3.

3.1 Molecular Spectroscopy

Spectroscopy with electromagnetic radiation provide in all wavelength regions from radio frequencies to gamma-rays a variety of powerful tools to investigate molecular properties. The method which has been applied in this study was absorption spectroscopy in the gas phase (spectroscopic analysis of the light transmitted by an absorbing medium which is placed between the light source and the spectrometer) and can be ranged in the field of optical spectroscopy. Optical molecular spectra are in general derived through the interaction of electromagnetic radiation in the UV, visible and IR wavelength range with molecules. They can be observed in absorption as well as in emission. In an absorption process energy from an electromagnetic field is transformed into internal molecular energy. The energy of a molecule can be described as a sum of different single energy distributions. Atoms in molecules undergo a variety of motions relative to each other. These can be separated into vibrational motions involving the various chemical bonds in the molecule and rotation of the molecule as a whole. The positions and movements of the atoms within the molecule can be influenced by an impact of radiation, causing a change of the molecular energy. If the absorbed energy is sufficiently high it can additionally lead to changes in the electron distribution in the molecule, called electronic transitions, and

23 24 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

eventually to breaking of chemical bonds. In a ﬁrst approach the energy of a molecule can therefore be described as a sum of the above mentioned single energy distributions, i.e. rotational, vibrational and electronic energy.

E = Eel + Evib + Erot

The diﬀerent types of spectra are listed below. A more detailed description about the origin of the according molecular transitions is given in the following sections.

Electronic Spectra • Transitions between diﬀerent electronic states of a molecule occur usually simultaneously with variation of vibrational and rotational quantum numbers and produce band spectra in the near infra red, visible and ultra violet region. Consequently these spectra are very complicated, since all quantum numbers can change at such transitions. If the energy is suﬃciently high, dissociation can occur leading to a continuous spectrum. From the analysis of electronic spectra information about the excited electronic states of a molecule can be derived.

Vibrational Spectra • Transitions between vibrational levels (quantum number υ) of one electronic level produce vibrational spectra, which appear in the infra red spectral range. These spectra consist of multitude bands, i.e. groups of closely spaced lines. From vibrational spectra molecular constants in the ground state can be derived.

Rotational Spectra • Rotational spectra are caused by transitions between rotational levels (quantum number J) of the same vibrational level in a certain electronic state. Thus only the quantum number J changes. Rotational spectra appear in the far infra red and microwave region and consist typically of a large amount of almost equidistant lines. These spectra allow predictions of rotational constants and moments of inertia from which conclusions regarding the molecular structures can be derived.

Figure 3.1 shows by example potential energy curves for the ground state E0 and an electronically excited state E1 of a hypothetical molecule. The quantum numbers υ indicate vibrational levels, the closely spaced lines above them represent rotational levels. Since rotational, vibrational and electronic energy levels diﬀer clearly from each other, changes in the molecular energy can be observed in diﬀerent spectral ranges of the electromagnetic spectrum (table 3.1). 3.1. MOLECULAR SPECTROSCOPY 25

E E1

‘ Q3

‘ 2 Q E0 Electronic Transition ‘ Q1 (UV / Visible) nergy

E ‘‘ Q5 Rotational ‘‘ Q4 Transition ‘‘ J5 ‘‘ (Microwave) Q3 ‘‘ J4 ‘‘ Q3 ‘‘ J3 Vibrational ‘‘ ‘‘ Transition Q2 J2 ‘‘ (Infrared) ‘‘ Q1 J1

Internuclear distance R

Figure 3.1: Potential energy curves for the ground state E0 and an electronically excited state E1 of a hypothetical molecule. Indicated are vibrational levels (quantum number υ) and rotational levels (quantum number J, closely spaced orange lines)

Spectral Frequency Energy Wavenumber Wavelength Transition Range (Hz) (eV) (cm−1) (mm) Microwaves 109-1011 4 10−6-4 10−4 3 10−2-3 3-300 Rotation of · · · heavy molecules Far Infrared 1011-1013 4 10−4 - 0.04 3 - 300 3 10−2 - 3 Rotation of · · light molecules vibrations of heavy molecules Infrared 1013-1014 0.04 - 0.4 300 - 3000 3 10−3 - 3 10−2 Vibration of · · light molecules Rotational and vibrational structure UV-Visible 1014-1016 0.4 - 40 3 103 - 3 105 3 10−5 - 3 10−3 Electronic · · · · Transitions

Table 3.1: Types of transitions and their according energies and spectral ranges 26 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

3.1.1 Electronic Spectra Quantum mechanics builds the basis for the theoretical study of the structure and transformation of atoms, molecules and other polyatomic systems. The main equation is the non-relativistic Schr¨odinger equation:

HΨ(−→r )= EΨ(−→r ) (−→r )=(r1, r2, ..., rN ) (3.1)

Ψ is the wave function of a system depending on the spin and space coordinates of all particles of the system and E the total internal energy of the system. The Hamiltonian H includes the operators of kinetic energy T of all particles in the molecule and also all Coulomb interactions between the charged particles involved, i.e. the potential energy V :

H = T + V (3.2)

1 2 1 1 2 T = pi + pk (3.3) 2m i 2 Mk X Xk The ﬁrst term of the latter equation is the kinetic energy of the electrons (mass m), the second that of the nuclei (mass Mk). The electronic energy levels are in general derived by solving the eigenvalue problem of the Schr¨odinger equation. Ψ(−→r ) is the overall state function and describes all N electrons in the molecule, E is the eigenvalue of the corresponding state. The solution is in general not exact, but an approximation.

Intensities of electronic transitions Transition intensities are inﬂuenced by two diﬀerent factors. The initial state of the system, i.e. the population density and the probability that a system in a certain state merges into another, i.e. the transition probability.

(1) Population density: The central statement here is that in a system with two energy levels with equal transition probabilities to a third state the higher line intensity will appear for the state with the higher population density. In an equilibrium the population in the states n and m is given by the Boltzmann distribution:

Nn = exp( ∆E/kBT ) (3.4) Nm − −23 ∆E is the energy gap between both states, kB the Boltzmann constant (1.38 10 J/K) and T the temperature in Kelvin. ·

(2) Transition probability: The probabilities of transitions under the inﬂuence of radiation are determined by the eigenfunctions of the states involved. The interaction of an electromagnetic wave having an electric vector E with an atomic or molecular system is in a ﬁrst approximation the interaction with the (variable) electric dipole moment M of the system, whose components are

Mx = ekxk, My = ekyk Mz = ekzk (3.5) X X X 3.1. MOLECULAR SPECTROSCOPY 27

ek are the charges on the N particles of coordinates xk, yk, zk. If the interaction M E is introduced into the wave equation it is found that the probability of a transition between· two states m and n produced by the interaction is proportional to the square of the magnitude of certain vector quantities Rmn-the matrix elements of the electric dipole moment- whose components depend in the following manner on the eigenfunctions ψm and ψn of the two states:

mn ∗ mn ∗ mn ∗ Rx = ψnMxψmdr, Ry = ψnMyψmdr, Rz = ψnMzψmdr (3.6) Z Z Z The matrix element Rmn is called the transition moment. If Rmn diﬀers from zero the two states combine with each other with a certain probability with emission or absorbtion of radiation; if it is zero the transition under consideration is forbidden as a dipole transition. Wave mechanics supplies the following relation between the transition moment and the transition probability in absorption:

3 8π mn 2 Bmn = R (3.7) 3h2c| |

Absorption Spectroscopy Experimentally the absolute intensity of electronic transitions is usually determined from the absorption spectrum. An absorption spectrum is obtained by the spectroscopic analysis of the light transmitted by an absorbing medium which is placed between the light source and the spectrometer. The basic principle used for absorption measurements is the Lambert-Beer-Law, which is deﬁned by the relation

−σ(˜ν)C∆l I(˜ν)= I0(˜ν)e (3.8)

I and I0 are the intensities before and after transmission through a column of length ∆l of the gas with concentration C. The absorption cross section σ(˜ν) varies with wave number and temperature. Two diﬀerent types of absorption measurements can be performed, i.e. relative or absolute measurements. Measurements of absolute absorption cross sections require accurate knowledge of the inﬂuencing parameters, such as temperature and in particular concentration. Relative measurements produce absorption spectra in terms of optical density: I (˜ν) OD(˜ν)=ln 0 = σ(˜ν) C l (3.9) IAbs(˜ν)! · · These obtained optical densities are then scaled to absolute and recommended values according to literature data. In this study the latter method was applied.

For small ∆l the light absorbed by the transition m n is given by → mn Iabs = [I0(˜ν) I(˜ν)]dν˜ = I0(˜ν)C∆l σ(˜ν)dν˜ (3.10) Z − Z Accordingly the integrated absorption cross section can be expressed by

3 8π ν˜mn mn 2 σ(˜ν)dν˜ = NmBmnhν˜mn = Nm Re (3.11) Z 3hc | | 28 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

This equation allows the comparison between the experimental quantity σ(˜ν)d˜ν and the theoretical quantity Rmn R Frequently the experimental data are expressed in terms of the oscillator strength f mn. It represents the ratio of the quantum theoretical and classical contribution of the transition m n to the refractivity N 1 (N=index of refraction) where the classical → − value is calculated on the assumption of a single oscillating electron of frequency c˜νmn. The oscillator strength f mn can be expressed as

2 3 mn µhc ν˜mn 8π νmn˜ mn 2 f = Bmn = Nm R (3.12) πe2 3hc | e |

3.1.2 Vibrational Spectra Harmonic Oscillation of a diatomic molecule • For the description of the most simple case of a diatomic molecule the following model is used. May two masses m1 and m2 be connected by an elastic spring, who oscillate around the distance of equilibrium x0. Following the assumption that the elastic force F is proportional to the deviation ∆x, the relation between both can be expressed as F = κ ∆x (3.13) − · This can easily be led back to the harmonic oscillator with the following solution for the frequency: κ ν = (3.14) sµ

The proportional constant κ is the suspension rate and µ the reduced mass with m m µ = 1 · 2 (3.15) m1 + m2

The solution of the related Schr¨odinger equation

d2Ψ 2µ 1 + (E κx2) = 0 (3.16) dx2 h¯2 − 2 provide the discrete energy values 1 E(υ)= hν (υ + ) (3.17) · 2

E(υ) ν 1 G(υ)= = (υ + ) [cm−1] (3.18) hc 2πc · 2 where ν is the classical oscillation frequency and υ the vibrational quantum number. Typical characteristics for the possible energy states of the harmonic oscillator are 1 equidistant gaps and that the ground state is unequal 0 (υ = 0, E0 = 2 hν¯ ). 3.1. MOLECULAR SPECTROSCOPY 29

Anharmonic oscillations of a diatomic molecule • The assumption regarding harmonic oscillations is connected with parabolic potential curves and represents therefore an approximation, which deviates from the real behavior. Firstly, it is more likely that the potential curve will increase more strongly when both nuclei approach each other. Secondly, the attractive force will decrease and reach eventually 0 with increasing internuclear distance, leading to a constant value of the potential energy. Both circumstances cannot be described adequately by a parabolic function. Therefore the expression of the vibrational motion must be based on a non-parabolic potential curve, which can be described by a model of an anharmonic oscillator. The Morse-potential, an empirically adapted potential form, has been proven useful for such descriptions:

V (R)= V + V 1 exp a δx 2 (3.19) 0 0 ·h − {− · }i The solution of the Schr¨odinger equation leads to the energy values

ν 1 1 2 1 1 2 G(υ)= (v + ) χe (v + ) ) =ν ˜ (v + ) χe (v + ) ) (3.20) 2πc · 2 − · 2 · 2 − · 2

−1 1 ν ν˜ is the wavenumber in cm (˜ν = λ = c ), ν the frequency and c the speed of light. The term χe is the anharmonicity constant:

hν¯ hcν˜2 χe =ν ˜ = (3.21) 4V0 4V0

3.1.3 Rotational Spectra The rotational spectra of molecules can be classified according to their ”principal moments of inertia”. Assume that the molecule rotates as a rigid body, i.e. the relative nuclear positions are fixed. The moment of inertia I of any molecule about any axis through the center-of-mass is then given by 2 I = miri (3.22) i X where mi and ri are the mass and perpendicular distance of atom i from the axis. One can always find one axis, called the c-axis, about which the moment of inertia has its maximum value, and another axis, labelled the a-axis, about which I has its minimum value. It can be shown that the a and c axes must be mutually perpendicular. Together with the b-axis, which is also perpendicular to the other axes, these form the ”principal axes” or ”mass centroid axes” of three dimensional body. The according principal moments of inertia are therefore denoted as Ia, Ib and Ic. Consequently, according to convention, the principal axes are ordered:

Ic Ib Ia (3.23) ≥ ≥ Based on this convention and their principal moment of inertia molecules can be classiﬁed into diﬀerent groups: 30 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

Linear Molecules with Ic = Ib >Ia = 0 •

Symmetric Top with Ic = Ib >Ia = 0 • A symmetric top can further be divided into6 two subgroups, i.e. a ”prolate” symmetric top with Ic = Ib >Ia and an ”oblate” symmetric top with Ic >Ib = Ia

Spherical top with Ic = Ib = Ia •

Asymmetric Top with Ic = Ib = Ia • 6 6

Diatomic Rigid Rotator, Ia = 0 As a ﬁrst approximation a diatomic rigid rotator is discussed. It can easily be seen that a diatomic molecule is linear. One principal axis is the connection line between the two atoms (a). Considering the atoms as point masses, the regarding moment of inertia Ia is 0. The remaining principal axes b and c go through the center of mass and are perpendicular to each other and to a. Every rotation of the molecule then takes place to one principal axis perpendicular to a. The classical relation between energy and momentum can then be expressed as:

−→J E = , with I = µ r2 (3.24) 2I ·

µ represents the ”reduced” mass of the system and is equal to m1m2/(m1 + m2). The solution of the Schr¨odinger equation leads to the following (quantized) expressions:

h¯2 E = BJ(J + 1) (3.25) 2I

E h F (J)= = J(J +1) = BJ(J + 1) (3.26) hc 8π2Ic

J ... Rotational quantum number of the angular momentum −→J , −→J = ¯h J(J + 1) • | | 2I q F ... Energy in cm−1 • h B...Rotational constant [B = 2 ] • 8π Ic

Diatomic Rigid Rotator, Ia > 0 The model in the foregoing section was based on the approximation of point masses of the atoms. If an extension of the atoms is considered (due to the distribution of electrons in shells around the nucleus), the moment of inertia Ia regarding the connection line between the atoms is unequal 0. The solution 3.24 remains in principal the same, but the angular momentum −→J of the diatomic molecule is now compiled by vector addition of the angular momentums of the nuclei −→N and electrons −→Λ . The angular momentum of the electronic shell is quantized according to −→Λ = Λ¯h, where Λ is the angular momentum quantum number, i.e. the sum of the single angular momentums of the electrons (Λ = λi ). | | The solution of the Schr¨odinger equation leads to the following quantized energyP levels: 3.1. MOLECULAR SPECTROSCOPY 31

F (J, Λ) = BJ(J +1)+(A B) Λ2 (3.27) − · with the rotational constant h A = 2 (3.28) 8π Iac

Symmetric Molecules The solution in the forgoing section can now be extended to polyatomic symmetric molecules. The nuclei inﬂuence the momentum of inertia Ia more signiﬁcantly, which is then comparable to Ib or Ic (”prolate” or ”oblate”). By usual convention the quantum number regarding the angular momentum with respect to the principal axis a is labelled with K.

F (J, K)= BJ(J +1)+(A B) K2 (3.29) − · Possible values for the quantum number K are

K = J, J + 1, ..., 0, ...J − − consequently the angular momentum −→J is in maximum perpendicular to a Every angular momentum with h¯ −→J = J(J + 1) (3.30) | | 2I q has 2J + 1 sub-levels. 2J of them are degenerated, since K lead to the same amount ± of energy. The angular momentum −→K with respect to the axis a is often labelled as the z-component of the angular momentum.

Asymmetric Molecules The great majority of polyatomic molecules can be referred to the class of asymmetric molecules with three different moments of inertia Ia, Ib and Ic. There is no longer a preferred direction, which carries out a simple rotation about −→J , which results in the removal of the before mentioned degeneracy for symmetric molecules. Spectra of asymmetric molecules have in general complicated energy level schemes, these energy levels cannot be represented by an explicit formula analogous to that of symmetric molecules. For each −→J there are 2J + 1 different energy levels with h¯ −→J = J(J + 1) (3.31) | | 2I q But there is no quantum number having a definite physical meaning that distinguishes these levels. Therefore in general the subscript τ is added to J for reference. τ takes the values τ = J, J + 1, J + 2, ... + J − − − Complex calculations lead to quantitative equations, which represent the different energy levels and which are then compared with experimental spectra: 1 1 F (Jτ )= (B + C) J(J + 1) + [A (B + C)] Wτ (3.32) 2 · − 2 · 32 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

Similar to the previous nomenclature C is the rotational constant with respect to Ic:

h C = 2 (3.33) 8π Icc

Wτ depends on A, B, C and J in a complicated manner and is calculated by a set of polynomial equations with 2J + 1 solutions corresponding to the 2J + 1 sub-levels mentioned above.

3.1.4 Symmetry Properties of Polyatomic Molecules Symmetry considerations of molecules play a constant part in spectroscopy. This section gives a brief introduction into symmetry elements and symmetry operations of polyatomic molecules, providing a basis for the description and understanding of the absorption spectra of O3 and NO2 in upcoming sections.

The expression ”symmetry” of a molecule is defined as the symmetry of the configuration of its nuclei or nuclear frame respectively. Each molecule, seen from a geometrical point of view, may have one or several ”symmetry elements” (plane, center, axis) with corresponding ”symmetry operations”, i.e. a coordinate transformation (e.g. rotation) that will produce a nuclei configuration indistinguishable from the original one. The symmetry elements of a molecule comprise together a ”point group” to which the molecule belongs (Point groups are so named because of the fact that the symmetry operations in the groups leave at least one point in space unchanged). The application of symmetry arguments to atoms and molecules has its origin in group theory developed by mathe- maticians in the 19th century. A rigorous mathematical formulation cannot be given at this point but a simple introduction is outlined below. If a set of elements A,B,C,D,... obeys the following conditions, they form a ”group”:

1. Closure. If A and B are any two members of the group, then their product A B must also be a member of the group. ∗

2. Associativity. The rule of combination must be such that the associative law holds. That is, if A, B, and C are any three elements of the group, then (A B) C = A (B C). ∗ ∗ ∗ ∗ 3. Identity. The group must contain a single element I such that for any element A of the group, A I = I A = A. I is called the identity element. ∗ ∗ 4. Inverse. Each element A of the group must have an inverse A−1 that is also a member of the group. By the term inverse we mean that A A−1 = A−1 A = I, where I is the identity element. ∗ ∗

Possible symmetry elements are

a plane of symmetry, usually designated by σ • a center of symmetry, usually designated by i • 3.1. MOLECULAR SPECTROSCOPY 33

a p-fold axis of symmetry, usually designated by Cp, where p = 1, 2, 3, ...; C means • 1 no rotational symmetry, C2 a two-fold axis of symmetry, etc.; C∞ means an axis of cylindrical symmetry, as in a linear or diatomic molecule

a p-fold rotation-reflection axis usually designated by Sp; a molecule having such an • element of symmetry will be transformed into itself by a rotation through an angle 360◦/p followed by a reflection at a plane perpendicular to the axis; the identity designated by I or E • In general a molecule has several of these symmetry elements. Possible combinations of these elements form as mentioned above a point group. The different point groups will not be discussed in detail at this point, for a more explicit description the author would like to refer to ref. [17] and [18]. The focus here will be on the C2v-symmetry, to which the molecules of interest in this study, i.e. O3 and NO2 belong. Figure 3.2 illustrates the various symmetry elements belonging to the C2v group.

C2 VQ (yz)

O y

O O

VQ (xz)

Figure 3.2: An illustration of the various symmetry elements belonging to the C2v group, with O3 as an illustrative case

As shown in ﬁgure 3.2 the following symmetry elements can be pointed out: The “identity” element I: the symmetry operation I consists of doing nothing to • the molecule, so that it may seem too trivial to be of importance. However, as noted above it is a necessary element required by the rules of group theory. All molecules have the identity element of symmetry.

A “two-fold axis of symmetry” C2: rotation of the molecule by 2π/n radians, with • n=2, about the z-axis produces a conﬁguration which is indistinguishable from the initial one.

A “symmetry plane” σv(xz) perpendicular to the plane of the molecule: that is, • reﬂection through the plane to an equal distance on the opposite side produces a conﬁguration indistinguishable from the initial one. The subscript ‘v’ stands for vertical and implies that the plane is vertical with respect to the highest-fold axis, which is C2 in this case (Planes that are perpendicular to the highest-fold symmetry axes are called horizontal, or σh, planes.). 34 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

A symmetry plane σv(yz) in the plane of the molecule. Any planar molecule has at • least one plane of symmetry. A useful device is what is called the multiplication table, which tabulates the products of various pairs of elements within the group. From these tables and a consideration of how the symmetry elements aﬀect various coordinate systems, it is possible to come up with a variety of matrices that multiply in the same way as the symmetry elements do. Any set of non-null square matrices that multiply in the same way as the elements of a group is said to form a representation of that group. If the matrices of a representation can be converted by the same similarity transformation into the same block diagonal form, the representation is said to be reducible, otherwise the representation is said to be irreducible. In spectroscopy irreducible representations of point groups build the clear majority. Valuable and often suﬃcient information can be provided by the traces of the matrices. If, in a certain representation, the matrix D(Rˆ) corresponds to the symmetry operation Rˆ, then the trace of D(Rˆ) is called the character of Rˆ for that representation. Tabulations of the characters of the various representations for a group are called character tables:

C2v I C2 σv(xz) συ(yz)

A1 +1 +1 +1 +1 Tz A2 +1 +1 -1 -1 Rz B1 +1 -1 +1 -1 Tz, Ry B2 +1 -1 -1 +1 Ty, Rz

Table 3.2: Diﬀerent ”symmetry types” in the character table of the C2v-group

The different irreducible representations, or symmetry types, are given labels: A1, A2, B1, and B2 in this case. Functions which are symmetric with respect to the principle symmetry axis Cn are denoted with the letter A, whereas those that are antisymmetric with respect to Cn are denoted with the letter B. The subscripts 1 or 2 then follow from the behavior under the other elements σv(xz) and σv(yz). The representation A1 is called the ”totally symmetric” representation and is always listed first in character tables. As they should be, the character sets for each of the representations are orthogonal to each other. Also listed in Table 3.2 are the symmetry types of the (x, y, z) coordinates (or translation operators). The utility of group theory in spectroscopy is far reaching. The central statement is now that the Schrödinger equation is invariant to symmetry operations and consequently the eigenfunctions can only be symmetric (+1) or antisymmetric (-1) with respect to these symmetry operations if the state is non-degenerate, while it can also transform into a linear combination of mutually degenerate eigenfunctions for a degenerate state. For example, if a plane symmetry is present the symmetry operation σ will not influence the wavefunction:

ψv(σv) (+1)ψv, symmetric with respect to σv −→ ψv(σv) ( 1)ψv, asymmetric with respect to σv (which can occur in a −→ − non-degenerated point group) This results in a significant simplification of calculation procedures. The wave functions don’t necessarily need to be derived directly by solving the Schrödinger equation but by considering the symmetry behavior of the eigenfunctions. 3.2. ABSORPTION-SPECTRUM OF O3 35

3.2 Absorption-Spectrum of O3

The observing window of GOME-2 covers the wavelength range between 220 and 790 nm. Figure 3.3 illustrates a typical absorption spectrum of O3 at ambient temperature in this range.

1E-17

(O ) @ T293K

Hartley-Band

3 ) 2

1E-18 (cm

1E-19 tions

Huggins-Bands

1E-20 Sec

Wulf-

1E-21 Cross Bands

Chappuis-Band

1E-22 sorption Ab 1E-23

1E-24

300 400 500 600 700 800

Wavelength (nm)

Figure 3.3: Absorption spectrum of O3 at ambient temperature in the 220-790 nm range

As one can see the absorption cross sections of O3 vary in the region of interest over seven orders of magnitude. This requires several measurements with diﬀerent experimental conditions in order to cover only a certain part of the spectrum and ﬁnally to concatenate the certain measurements to an overall spectrum. The approach regarding this issue will be addressed in detail in chapter 4.2.3.

The absorption spectrum of O3 in the UV, visible and near IR consists of several electronic transition systems, which are named by their discoverers Hartley-, Huggins-, Chappuis- and Wulf-bands. Due to the low dissociation energy mainly continual absorption and diﬀuse vibration bands are observed. Rotational structures were only observed in the Wulf bands. The following subsections describe more detailed the spectroscopic origin of the certain electronic transition systems in the O3 absorption spectrum.

The Hartley Band

The Hartley absorption spectrum is the strongest absorption band ofO3 in the UV region and covers approximately the 200-310 nm region. The strong absorption in this wavelength range prevents harmful solar radiation from reaching the Earth’s surface. Due to this importance the UV absorption spectrum of O3 has been the focus of several studies. 36 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

1 1 The Hartley bands correspond to the B2 X A1 transition in the C2ν point group of the ground state. It consists of a broad continuum← with a peak around 255 nm which is superimposed by small diffuse structures with irregular spacings between the peaks, varying from 150 to 300 cm−1. The continual absorption can be explained by a fast photo 1 dissociation after an excitation into the B2 state. The potential energy surface of the 1 B2 state is repulsive and lies about 1 eV above the dissociation limit [19]. The origin of the overlying diffuse structures remain more unclear. R.T. Pack provided already in 1976 a model regarding the appearance of diffuse vibrational structures in continuous UV spectra of polyatomic molecules [20]. Based on this model Joens referred 1 the diffuse structures in the Hartley band to transitions from the A1(0,0,0) state in 1 ′ ′ B2(ν1,ν2,0) states [21]. Analogous assignments were performed through the study of the 16 18 structures of the Hartley bands for two isotopes ( O3 and O3) by Parisse et al. [22]. Despite numerous theoretical investigations and the achieved improvements there is not yet a quantitative agreement between calculated and observed Hartley absorption oscillations [2, 23](see also figure 1.4).

The Huggins Bands The region between 310 and 370 nm is known as the Huggins bands and is mainly characterized by a series of peaks superimposed on the continuum of the UV range. The bands with wavelengths less than approximately 313.5 nm have irregular spacings which 1 are probably caused by a perturbation due to the proximity to the O2(1∆g)+O( D) dissociation limit ( 32240 cm−1). The cross section on the red end is very small but increases by more≈ than four orders of magnitude within the range of the Huggins band towards smaller wavelengths. In contrast to the Hartley band the Huggins structures show a strong temperature dependence, which increases progressively when approaching the longer wavelengths. The range between 325 and 345 nm represents an important window for remote sensing, one part of the analysis in chapter 7.3 will therefore focus on this issue. Although a large number of experimental and theoretical studies have been performed there are still several open questions and inconsistencies. One of the main disagreements concerns the electronic and vibrational assignment of the spectroscopic features in the Huggins band. An initial assignment by isotopic studies was performed by Katayama in 1979 and 1986 [24, 25]. However a weak point was that the upper electronic state’s symmetry was not determined rigorously. This was revived and discussed by Sinha et al. [26] in the same year. From the interpretation of their laser induced fluorescence spectrum obtained in a supersonic beam they assigned the Huggins absorption bands to weak 1 absorptions into one of the two Cs wells of the B2 excited state. That meant that the 1 1 Huggins and Hartley bands represented different regions of the same B2 X A1 transition. These results were supported and confirmed later on by Le Quéréand← C. Leforestier 1 [27], whose studies based on computed ab initio calculations of the B2 electronic state by Yamashita et al. [28]. Nevertheless this key question has not been answered with certainty, whether the Huggins band system is due to a transmission to the same upper electronic state as the Hartley band. Contradictive results were obtained by Banichevich et al. [29, 30], whose calculations differed in the precise location of the minima and height and second 1 1 derivatives of the saddle point of the B2 state. At the same time properties of the 2 A1 3.2. ABSORPTION-SPECTRUM OF O3 37

1 state were studied. The transition 2 A1 X is forbidden with regard to the dominant ← 1 electronic configurations. The geometry of the 2 A1 surface though showed a minimum 0.5 eV below the minimum of the 1B state, allowing a coupling process with the 1B ≈ 2 2 state in Cs symmetry. Based on these results and from symmetry considerations and requirements Joens reassigned the electronic transition responsible of the Huggins bands to the 21A X1A transition [31]. 1← 1 A comprehensive resume of the assignments suggested in the literature till 2001 is given in ref. [32], leading also to the conclusion that the Huggins bands almost certainly 1 1 terminate on the 2 A1 state rather than the 1 B2 state. Most recently though Qu et al. published results from a theoretical analysis of the Huggins bands of ozone by performing dynamic calculations using a new (diabatic) po- 1 tential energy surface for the B2 state [33, 34]. Their vibrational assignment mostly agrees with that of O’Keeffe et al. [32] but disagrees in the electronic assignment, leading consequently to the conclusion that the Huggins band is due to the two Cs potential wells 1 1 of the B2 state rather than the single C2v well of the 2 A1 state. That would mean again that the Huggins band is due to a transition that is the same state responsible for the Hartley band. This is also illustrated in figure 3.4, where the lowest potential energy surfaces are shown (1A symmetry).

Figure 3.4: Lowest potential energy surfaces with 1A electronic symmetry along the dissociation coordinate R1. The vertical arrows indicate the absorption in the Hartley and Huggins bands [33]. 38 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

The Chappuis Band

The Chappuis absorption bands cover the 400-690 nm range with a maximum absorption cross section around 600 nm. It shows mainly a continuum absorption feature, which is superposed by an irregular series of vibrational bands in the blue wing between 400 and 550 nm. The absorption cross section of the Chappuis bands ( 5 10−21cm2/molecule) are more than three orders of magnitude weaker than those of th≈e Hartley× bands maximum cross sections ( 1.15 10−17cm2/molecule ). ≈ × With regard to C2v symmetry the Chappuis system can be referred to transitions 1 1 1 1 1 from the X A1 ground state to the A2 and B1. While B1 X A1 is allowed within 1 1 ← C2v symmetry the A2 X A1 transition can only be explained by including vibronic excitations via the antisymmetric← stretch. According to ab initio calculations both states 1 1 ◦ A2 and B1 show the same bond length of 1.35 Abut˚ differ in the bond angle (100 for 1 ◦ ◦ 1 A2, 116 -125 for B1) [23]. Further states contributing to the Chappuis system are the 11A” and 21A” states 1 1 in Cs symmetry, resulting from an avoided crossing of the A2 and B1 states. The energetic lower and dissociative 11A” state (dissociation in the ground state products 3 − 3 1 O2( Σg ) and O( P)) is comparable to the A2 state in C2v symmetry near its saddle point and contributes to the underlying featureless absorption in the red wing of the 1 1 Chappuis system. In contrast the energetic higher 1 A” state corresponds to the B1 in C2v symmetry near its minimum and represents therefore a bound state. Absorption in this state cause the structure in the blue part of the Chappuis system. A dissociation 1 1 would lead to the excited products O2( ∆g) and O( D). These have not been observed though, but only the above mentioned ground state products. The reason is as follows: 1 1 ◦ A conical intersection between A2 and B1 occurs near 120 over a wide range of bond lengths [35, 36]. The conical intersection serves to enhance higher bending modes in the Chappuis bands. This causes a non-adiabatic coupling between the 21A” and 11A” state, i.e. the excited molecule leaks from the upper 21A” adiabatic surface to the lower 11A” state, which dissociates to mentioned ground state products. 1 1 The vibrational assignments, which also belong to the B1 X A1 transition, are not clarified yet. Several groups proposed different assignments [29,30,37-40],← which however don’t provide a perfect explanation regarding all experimental and theoretical results (electronic energies, positions of bands, isotope shifts, vibrational frequencies). The proposed assignments are recalled and discussed by Anderson and Mauersberger [41].

The Wulf Bands

Starting at the red wing of the Chappuis band the Wulf bands reach the NIR at about 1050 nm and are only covered partly by the GOME-2 instrument. Experimental study of the Wulf band has been diﬃcult because of its extremely low intensity. As a result, the electronic character of the excited state has been incorrectly referred as a singlet 1 1 A2 X A1 transitions [42, 43]. High level ab initio calculations [35, 37, 44] indicated already← between 1990 and 1995 that the correct assignment of the upper electronic states would be a triplet state. 3 3 3 The adiabatic electronic energy of the three lower triplet states A2, B2 and B1 were 3 3 3 characterized and a vibrational assignment for the transitions to A2, B2 and B1 proposed [39, 42]. Both the electronic and vibrational assignments were later experimentally 3.2. ABSORPTION-SPECTRUM OF O3 39

conﬁrmed by high-resolution spectroscopic analysis of the rotational structure of several low-lying vibrational features [41,45-49]. 3 1 The A2 X A1 transition is doubly forbidden (A2 A1 and ∆S=0). Its appearance ← 6↔ 6 3 was explained by Braunstein et al., who showed that the transition to the A2 state gains 1 intensity through spin-orbit mixing with the B2 state (responsible for the strong UV Hartley bands) and has furthermore calculated the related oscillator strengths [37, 44]. Minaev et al. performed studies regarding the source of intensity of the Wulf bands focussing in particular on the role of spin-orbit coupling [50]. From this study transition moments and oscillator strengths pertaining to the three lowest triplet states were derived. Most recently Xie et al. reported accurate ab initio potential energy surfaces of the 3 3 A2 and B1 state and their non-adiabatic coupling near the ground state equilibrium geometry, enabling the ﬁrst characterization of all three vibrational modes in the Wulf bands [51]. It can be summarized that nowadays it is well established that the three low-lying 3 3 3 triplet states A2, B2 and B1 are involved in the Wulf band absorption, although ab- 3 sorption to the B2 state can be neglected because of extremely small oscillator strength [44, 50]. 40 CHAPTER 3. RELEVANT ASPECTS OF MOLECULAR SPECTROSCOPY

3.3 Absorption-Spectrum of NO2

The second main target gas of GOME-2 is NO2. Figure 3.5 illustrates the absorption cross sections of NO2 at ambient temperature in the 220-790 nm wavelength range.

(NO ) @ T293K

2 ) 2

(cm 1E-19 tions Sec

1E-20 ross C

1E-21 Absorption

300 400 500 600 700 800

Wavelength (nm)

Figure 3.5: Absorption spectrum of NO2 at ambient temperature in the 220-790 nm range

The absorption spectrum of NO2 has a slightly asymmetric bell shape which spreads from 1000 nm to about 280 nm and is centered at about 400 nm. The spectrum is extremely complex, covered with a great number of lines densely packed, most of them with no apparent regular pattern and definite spectroscopic assignment. It is well established that the absorption spectrum of NO2 in this spectral range belongs to transitions from 2 2 the electronic ground state, X A1 , to the two first excited electronic states, A B2 and 2 −1 B B1, whose lowest vibrational levels are located at about 9735 cm ( 1027.22 nm) and 13 385 cm−1 ( 747.10 nm), respectively [52]. The mentioned complex≈ structure is explained by a conical≈ intersection between the potential energy surfaces of the two lowest 2 2 electronic states, i.e. X A1 and A B2 in C2v symmetry. This conical intersection produces strong vibronic interactions between these states, causing chaotic properties between the origin of the A2B up to the dissociation limit and above ( 25000 cm−1 or 400 nm). 2 ≈ This complexity has attracted high scientific interest and has made NO2 one of the most extensively studied triatomic molecules. A full description of the current scientific state though would be far beyond the scope of this thesis. For details the author would like to refer to the most recent publications and the references therein [52-58].

Similar to O3 the NO2 absorption cross sections vary over several orders of magnitude requiring the same approach mentioned above, where several measurements with diﬀerent experimental conditions have to be performed in order to cover one part of the spectrum. A detailed description about the settings and how these experimental conditions were achieved will be given in chapter 4.2.4. Chapter 4

Experimental Setup

This chapter contains the experimental setup used in this study. Firstly a short description of the GOME-2 instruments will be given, followed by the mobile Calibration Apparatus for Trace Gas Absorption Spectroscopy CATGAS. Finally the optical interface connecting the CATGAS setup and GOME-2 instrument will be introduced.

4.1 The GOME-2 Spectrometer

The GOME-2 spectrometer is a successor of GOME and very similar to the concept used therein. In principle the instrument collects light arriving from the Sun-illuminated Earth’s atmosphere and decomposes it into its spectral components. The GOME-2 instruments are four-channel grating spectrometers operating in the ultraviolet and visible wavelength regions between 240 and 790 nm at a spectral resolution of 0.24 to 0.53 nm. A double spectrometer design with a pre-disperser prism and holographic gratings is used to cover the entire wavelength region. All refractive optics are made of quartz (Suprasil 1) and are multilayer-coated for maximum eﬃciency and low stray light. The oﬀ-axis parabolas are made of aluminum, nickelcoated and machined with a single-point diamond turning technique. Polishing then achieves a surface quality compliant with the low-stray-light application in the ultraviolet. The light is detected by four cooled silicon diode array detectors each with 1024 pixels. The spectral width of pixels is about 0.1-0.2 nm per pixel. The pixel exposure times (PET) can independently be chosen for each channel. The GOME-2 instruments have a large dynamic range and can easily measure absorptions from less than 0.001 safely up to 1.5 to 2 in units of optical density.

The measurements described in this paper were performed in three campaigns in 2002- 2004 at a special facility at TPD/TNO in Delft (The Netherlands), where the GOME- 2 instruments were maintained at in-ﬂight conditions in a stabilized cryo-vacuum tank (vacuum, temperature of the optical bench) and calibrated and characterized under these conditions [59-61].

In the following the main characteristics of the GOME-2 instrument are listed, an illustration is shown in ﬁgure 4.1. A detailed description of the GOME-2 instruments are given in [62].

41 42 CHAPTER 4. EXPERIMENTAL SETUP

GOME-2 Main Characteristics 1

Spectrometer type Double monochromator with pre-disperser prism and four holographic gratings

Spectral range 240 - 790 nm

Field of view 0.286 deg (across-track) x 2.75 deg (along-track) 4 km x 40 km

Entrance slit 0.2 mm (across-track) x 9.6 mm (along-track)

Channels and 1: 240 - 315 nm 0.24 - 0.29 nm resolution 2: 311 - 403 nm 0.26 - 0.28 nm 3: 401 - 600 nm 0.44 - 0.53 nm 4: 590 - 790 nm 0.44 - 0.53 nm

Polarisation 200 detector pixels Monitoring Unit 312 - 790 nm in 12 programmable bands Spectral resolution: 2.8 nm @ 312 nm to 40 nm @ 790 nm

Viewing modes: Nadir across-track 1920, 960, 480, 360, 240, 120 km Solar± Fixed angle± once± per± day ± ± Lunar Fixed varying angle, 6 times per year ≈ Spectral calibration Fixed angle (once per day to once per month) White Light Source Fixed angle (once per day to once per month) Dark signal Fixed angle (night side of the orbit)

Spatial resolution 40 km x 40 km (960 km swath and integration time of 0.1875 s) 40 km x 5 km (for polarisation monitoring)

Data rate 400 kbit/s (GOME-1: 40 kbit/s)

Mass 73 kg (GOME-1: 55 kg)

Power 58 W (avg) (GOME-1: 32 W)

Dimensions Zenith nadir 656 mm, across-track 848 mm, velocity 468 mm

1Details are given in ref. [62] 4.1. THE GOME-2 SPECTROMETER 43

Figure 4.1: The GOME-2 Spectrometer: The upper graphics show a block diagram of the GOME-2 optics, the lower part an artist’s impression of the optical layout [62] 44 CHAPTER 4. EXPERIMENTAL SETUP

4.2 The CATGAS Setup

The concept is a modular compounded mobile absorption spectroscopy setup consisting of gas-vessel, optics and gas-system. This chapter describes the setup with the several sub-components. The following graphics gives a principle overall overview of the setup:

Glass fiber Off-Axis Parabolic Xenon-lamp 1

GOME-2

IUP- spectrometer light source monitoring (near-realtime-reference)

flip mirror under TTL pressure control head ethanol

thermostated absorption cell p

gas mixture IN

gas mixture OUT multipass (evacuation) White-optic

Figure 4.2: Principle absorption spectroscopy setup

The analysis light passes a vessel containing the gas of interest (red path in figure 4.2) and can then be analyzed by either the GOME-2 instrument or, for verifications of the experimental conditions, by the commercial IUP-spectrometer. Two major differences have been made compared to previous CATGAS setup’s:

An integration of a TTL-controlled ﬂip mirror for monitoring the light source during • the absorption measurements (blue path in ﬁgure 4.2)

Development of new gas mixing units for measurements with O3 and NO2, respec- • tively and integration of a pressure regulator in order to maintain a constant pressure in the vessel during gas exchange.

Both eﬀorts lead to a signiﬁcant improvement in baseline stability. In the following picture the complete set up on the optical bench is shown: 4.2. THE CATGAS SETUP 45

Flip mirror protected lamp housing

Optical bench

Stabilised gas exchange

Figure 4.3: CATGAS setup on the optical bench

A detailed description of the several sub components is given in the following sections.

4.2.1 Light-Source and optics As a light source a new xenon-lamp of the typ Hamamatsu was used. In contrast to previous setup’s a higher stability could be gained by a protected lamp housing. A lens focuses the light into the entrance window of the gas cell, which is equipped with a so- called White-optics. This optical arrangement, following the method of White [63], allows multiple reﬂections within the vessel and therefore optical pathlengths up to 33m. The analysis light coming from the CATGAS setup is directed either to

1. the actual optical CATGAS-GOME-2 connection consisting of a multi ﬁbre bundle and an optical interface developed by TPD (see section 4.3) or

2. a small glass ﬁbre leading to the commercial IUP-spectrometer (used for veriﬁcation of the experimental conditions or checking the stability of the system e.g. after temperature change)

Once the correct experiment conditions are set, the off-axis parabolic mirror No. 2 in figure 4.2, can be removed and the light can than be guided to the GOME-2 fibre by the off-axis parabolic mirror No. 1. 46 CHAPTER 4. EXPERIMENTAL SETUP

4.2.2 Gas-Vessel The gas vessel is a double jacketed fused silica gas cell. The inner cell is the gas reservoir and also contains the White-optics. Through the middle cell a ﬂow of ethanol can be used from a thermostated reservoir in order to cool down the temperature down to -70◦C. The outer cell can be evacuated for insulation. This is illustrated in ﬁgure 4.4.

Eva cuated Ethanol -Flow Gas

Figure 4.4: Double-jacketed gas-vessel

The setup consists of two such vessels, one for O3 one for NO2-measurements (ﬁgure 4.3).

4.2.3 Gas-System O3

In principle the O3 gas-system consists of a 50l oxygen-bottle (5.0 purity), a newly developed gas-exchange unit (O2/3-Flow switching unit), two 2000 sccm (standard cubic centimeter) flowcontrollers, ozonisers and an ozone-destroyer unit (figure 4.5). For measurements with O3 a setup was chosen for using a flow modus. It was intended to keep the pressure in the cell constant, even at gas exchange. The initial gas-flow coming from the gas bottle is separated into two lines: One line is connected directly with the gas-exchanging unit, the other flow passes the ozonizer in which O3 is produced by electrical discharge. Two types of ozonizer (one commercial one [typ OG5] and one built at IUP/IFE) were used depending on the amount of O3 needed. The OG5 produces higher O3-concentrations and is therefore used for ranges with low absorption cross section. The efficiency can be regulated by a power control. For measurements in regions with high absorption cross sections as in the Hartley-band (mixtures 6 to 8) the IUP/IFE ozonizer with a much lower production of O3 concentrations was used. 4.2. THE CATGAS SETUP 47

This branch with a O2 / O3-mixed flow is also connected with the gas exchanging unit. Each line is connected with a 3-way-valve, depending on its position the flow is either directed in the reaction vessel or evacuated. The two valves from each line are connected with each other through a hand gear in a way that they react anticorrelated. That means that, if one flow is connected with the reaction vessel the other one is automatically evacuated and vice versa. In this way the flow towards the vessel is hardly disturbed and ¡the pressure remains within +/- 0,1mbar.

FC 2000 Ozonizer Ozone / Oxygen Mixing−Unit FC 2000 Gas−Vessel

50l Ozone− PR Oxygen Evac. Destroyer− Unit

Figure 4.5: O3 gas system

Figure 4.6 illustrates more detailed the O3 mixing unit.

The above described method about changing the ﬂows to the vessel by using the gear is however only valid buffer volume pressure and pre-cooling head for measurements referred to mixtures valve

1 to 5. For measurements regarding pressure head O FC mixtures 6 to 8, covering the Hartley- 2 eva- absorption cell band in channel 1, a slight modiﬁca- ozonizer cuation tion is required: Interconnection O 22 and O / O 3 - line synchronized 3-way valves

Due to the high absorption cross pressure FC head sections of O3 in the 220 to 310 nm valve region, even with the lowest amount of produced O3 one reaches far to by-passing gas line high optical densities. As seen in ﬁg- ure 4.6 we have an interconnection between the two lines, which has no eﬀect as long as the valve in this con- Figure 4.6: O3 mixing unit I nection is closed.

Opening the valve however allows additional diluting of the O3-flow with O2. The rate of dilution can be controlled by the settings for the flowcontroller of the O2 branch. The change from a reference to an absorption measurement and back is simply achieved by switching the ozonizer on and off. Figure 4.7 shows the mixing unit with the hand gear (red circle), with which the flow can be changed and the interconnection of the two branches (blue circle). 48 CHAPTER 4. EXPERIMENTAL SETUP

Figure 4.7: O3 mixing unit II

As mentioned before it was an important requirement for the baseline stability to keep the pressure in the vessel constant during all measurements regarding one mixture, even and especially at gas-exchange. Beside the above described gas mixing unit for this purpose a pressure regulator was integrated at the output of the vessel. Once the desired pressure in the vessel is reached the regulator is activated and the ﬂows and pressure in the vessel remain constant(ﬁgure 4.8).

Pressure regulator downstream of vessel

Pressure stability during gas exchange better than 1%

Figure 4.8: Pressure regulator at the output of the gas vessel 4.2. THE CATGAS SETUP 49

Experimental Settings for O3-Measurements As described in chapter 3.2 several absorption measurements with different experimental conditions are required due to the variation of the O3 absorption cross sections over seven orders of magnitude in the 240-790 nm range. Different conditions can be achieved by different combinations of optical pathlength and O3 concentration, which we will call ’mixtures’ in the following. As for instance expressed in equation 3.4 high concentrations are required in regions with small absorption cross sections and vice versa. The following figure 4.9 illustrates an absorption spectrum of O3 at ambient temperature.

100000

2,24e13 8 10000 1,12e14 7

1000 5,60e14 6

2,80e15 5 100 1,40e16 4 [a.u.] 10 7,00e16 3 2 1 3,50e17 1,75e18 1 0,1

FPA- 1 FPA- 2 FPA- 3 FPA- 4 0,01 200 300 400 500 600 700 800 Wavelength (nm)

Figure 4.9: Absorption spectrum of O3 with diﬀerent regions of interest. The numbers on the left give roughly the achieved O3 concentrations in the certain experimental settings. Note that the channel borders have always been covered by one combination.

The spectrum is divided into eight areas, shown by the horizontal lines, which represent different experimental settings (named mix1, mix2, ..., mix8) of optical pathlength and O3-concentration. In the respective combination of these parameters the optical density will ideally be between 0.1 and 1 in a certain range and cover a certain part of the spectrum. For instance, the settings for ”mix1” will mainly cover the minimum range of the O3-spectrum between 350 and 450 nm. In table 4.1 the certain combinations of experimental conditions (optical pathlength, power of the O3 generator, total pressure and gas flow rate) are listed. Please note that after the first campaign with GOME-2 FM2 the procedure was slightly changed. Due to the importance of the DOAS window between 325 and 345 nm for remote sensing, it was decided to cover the Huggins bands with more than three measurements (originally declared as ”mix3, mix4, mix5”). Combination 3 in table 4.1 therefore represents seven different measurements each with different settings for the O3 generator. Combinations 1-3 and 6-8 have been measured with different O3 generators. 50 CHAPTER 4. EXPERIMENTAL SETUP

As explained in the preceding section the gas mixing and exchanging method differs for measurements regarding mixture 6 to 8. Beside the use of a different O3 generator the valve between both branches in the mixing module is open (see figure 4.6).

Combination Wavelength Optical Power OG Pressure in Flow in Flow in and Channel range Pathlength [m] (1-3, [a.u.]) reaction vessel O2-branch O3-branch [nm] (6-8, [V]) [mbar] [sccm] [sccm] 1 (FPA-2+3) 355-455 28.40 260 800 100 100 2 (FPA-2) 340-365 28.40 60 800 400 400 (FPA-3) 430-500 32 (FPA-2) 315-350 28.40 40-120 100 400 400 (FPA-3+4) 475-790 6 (FPA-1) 300-325 2.40 12 320 0 1000 7 (FPA-1) 280-310 2.40 8.5 65 0 1000 8 (FPA-1) 240-290 2.40 12 36 200 200

Table 4.1: Experimental conditions for O3 measurements. The units for the ﬂow controllers sccm stand for standard cubic centimeter per minute. Since relative measurements have been performed an accuracy statement for the ﬂow controllers is not necessary.

2 Combination 3 represents seven different measurements, each with different settings for the O3 generator and covers the range, which was originally declared as ”mix3, mix4, mix5” (fig. 4.9). More details are given in the text. 4.2. THE CATGAS SETUP 51

4.2.4 Gas-System NO2

The NO2 system differs from the O3 system, although the principle of a constant cell pressure remains. The principal item for the NO2 gas-system is, similar as in the O3 set up, the mixing unit, with which two flows of synthetic air and synthetic air with 2,5% NO2 can be combined to the right mixing ratio with the required NO2 concentration in the vessel. The principle NO2 gas-system is illustrated in figure 4.10.

FC 2000 FC 10

FC 10000 50l FC 2,5% NO2 1000 in 50l FC FC Synthetic Synthetic Air Air 100 5000 Pre− Gas−Vessel cooling unit PR Evac.

Figure 4.10: NO2 gas system

Two 50l gas bottles, containing synthetic air and NO2 (2,5% NO2 in synthetic air) respectively, are connected with the mixing unit; synthetic air with 3 flowcontrollers, allowing a maximum flow of 10000, 1000 or 100 sccm, NO2 with 2 flowcontrollers allowing a maximum flow of 2000 or 10 sccm respectively. After passing the fc’s both branches are united and then directed to the vessel. Between the mixing unit and the vessel an additional flowcontroller (5000sccm) is integrated, the flow of which is in general lower than the sum of the flows of the mixing unit. The resulting overpressure is deflated through another valve into the vacuum. A pressure regulator at the output of the vessel ensures, similar to the O3 system, a stable pressure in the cell. The switch from reference measurement to absorption measurement is simply done by activating the NO2 -flow. The required concentration of NO2 in the vessel is achieved by diluting the NO2-flow with the synthetic air flowcontrollers. This will be described more detailed in section 4.2.4

Calculation of NO2-Concentrations and Dimerisation

Figure 3.5 in chapter 3.3 shows an absorption spectrum of NO2, which varies, similar to an O3 spectrum, over several orders of magnitude. Here as well it is necessary to take several measurements for complete coverage in the 240-79 nm range, whereas an additional factor has to be respected: The dimerisation of NO2 to N2O4. This issue needs to be introduced before listing the experimental settings and conditions for NO2 measurements. 52 CHAPTER 4. EXPERIMENTAL SETUP

As already seen in figure 4.9 with the O3 absorption spectrum the NO2 spectrum as well can be divided into six sections representing different experimental conditions (see figure 4.15). The numbers given on the left are again indications for NO2-concentrations. The maximum possible concentration of NO2 in the vessel (based on a 2,5% NO2 in synthetic air mixture) can be determined as follows:

Avogadro’s number gives the amount of molecules in 1 mole of a gas under norm conditions, meaning at 273K and atmospheric pressure. Under these conditions the gas has a volume of 22,4 103cm2. This corresponds to a concentration of · 6, 022 1023molecules [c] = · 2, 7 1019 molecules/cm3 (4.1) 22, 4 103cm3 ≈ · ·

With 2,5% NO2 in synthetic air this would mean a concentration of NO2 in a vessel with 1000mbar:

25mbar 2, 7 1019molec/cm3 [NO ]= · · = 6, 75 1017molec/cm3 (4.2) 2 1000mbar ·

This value can then be used as a reference value. In figure 3.5 the highest concentration is however given one order of magnitude lower. This is due to the requirement of a second flow needed for dilution. Only with a second flow of pure synthetic air the flow of the NO2-branch can be stopped (as needed for reference measurements) without interrupting the flow through the vessel(as needed for baseline stability). Initiating from this point the other required concentrations can be estimated by the following procedure: The concentration of NO2 at any temperature can be expressed by

19 3 273.15K [NO ] , = PNO2a,b 2.7 10 molec/cm (4.3) 2 a b · · · T where P is the partial pressure and 2.7 1019 molec / cm3, as calculated before, the concentration under norm conditions. At this· point another critical issue to be thought of is the following: NO2 is constantly in an equilibrium with its dimer N2O4.

2NO2 ↽⇀ N2O4

This equilibrium is shifted towards N2O4 at lower temperature or higher NO2 partial pressure. This has to be respected in the calculations of further concentrations of NO2 at diﬀerent temperatures and partial pressures. The index a or b in the foregoing equation

is referred to a concentration without and with dimerisation respectively. PNO2b can be determined by the following equation

Kp 8Pvessel χ PNO2b = 1+ · 1 (4.4) 4 "s Kp − #

where Kp is the equilibrium constant of NO2 with N2O4 as given in [64] and χ the per- centage of NO2 in the ﬂow, in this case 2,5%. An additional ﬂow of synthetic air reduces 4.2. THE CATGAS SETUP 53

the partial pressure of NO2 in the vessel: FlowFC1 PNO2a = 0, 025 P (4.5) · vessel · FlowFC1 + FlowFC2

Based on these deliberations the interaction of both flowcontrollers at different temperatures could be determined, which was done with the scientific software origin. The following table shows such an origin-project for calculations at 293 K, which can also be used for future references.

The pressure in the vessel is 500 mbar, the ﬂow of the ﬂowcontroller FC1 (2.5% NO2 in synthetic air) is 10 sccm.

Worksheet: ”Dilution at 293 K and 500 mbar pressure in the vessel”:

T Kp Pvessel FlowFC2 PNO2a concNO2a ... Kelvin mbar mbar sccm mbar molec / ccm ... without Dimerisation without Dimerisation 292,99585 97,04215 500 0 12,5 3,13101E17 ... 292,99585 97,04215 500 20 4,16667 1,04367E17 ... 292,99585 97,04215 500 40 2,5 6,26203E16 ... : : : : : : ...

... PNO2b concNO2b PN2O4 concN2O4 ... mbar molec / ccm mbar molec / ccm ... with Dimerisation with Dimerisation with Dimerisation with Dimerisation ... 10,30949 2,58233E17 2,19051 5,4868E16 ... 3,85965 9,66769E16 0,30702 7,69024E15 ... 2,38297 5,96889E16 0,11703 2,93144E15 ... : : : :

The meaning of the abbreviations in the worksheet are as follows:

T ... Temperature in Kelvin Kp ... Equilibrium constant Pcell ... Overall pressure in cell FlowFC2 ... Flow of second ﬂowcontroller, with which the NO2-ﬂow is diluted PNO2a ... Partial pressure of NO2 without Dimerisation concNO2a ... Concentration of NO2 without Dimerisation PNO2b ... Partial pressure of NO2 with Dimerisation concNO2b ... Concentration of NO2 with Dimerisation PN2O4 ... Partial pressure of N2O4 concN2O4 ... Concentration of N2O4

With the created worksheets the desired NO2-concentration in the vessel can easily be read oﬀ in dependence of temperature and partial pressure. Figures 4.11 to 4.14 illustrate the NO2-concentration (10 sccm 2.5 % NO2 in s.a.) in dependence of the dilution, i.e. additional ﬂow of synthetic air coming from FC2. 54 CHAPTER 4. EXPERIMENTAL SETUP

Concentration NO and N O at 293 K NO

2 2 4 2

1E17

N O

2 4

1E16

1E15 3

1E14

lec / cm o

1E13 M

1E12

1E11

100 1000 10000

Additional flow FC2 (sccm)

Figure 4.11: NO2-concentration (10 sccm of 2.5 % NO2 in s.a.) as a function of diluting synthetic air ﬂow coming from FC2 for T 293 K

1E18

NO Concentration NO and N O at 273 K

2 2 4

N O

2 4

1E17

1E16 3

1E15

lec / cm 1E14 o M

1E13

1E12

1E11

100 1000 10000

Additional flow FC2 (sccm)

Figure 4.12: NO2-concentration (10 sccm of 2.5 % NO2 in s.a.) as a function of diluting synthetic air ﬂow coming from FC2 for T 273 K 4.2. THE CATGAS SETUP 55

1E17 Concentration NO and N O at 243 K NO

2 2 4 2

N O

2 4

1E16 3

1E15

lec / cm o M

1E14

1E13

100 1000 10000

Additional flow FC2 (sccm)

Figure 4.13: NO2-concentration (10 sccm of 2.5 % NO2 in s.a.) as a function of diluting synthetic air ﬂow coming from FC2 for T 243 K

Concentration NO and N O at 223 K NO

1E17 2 2 4 2

N O

2 4

1E16 3

1E15 lec / cm o M

1E14

100 1000 10000

Additional flow FC2 (sccm)

Figure 4.14: NO2-concentration (10 sccm of 2.5 % NO2 in s.a.) as a function of diluting synthetic air ﬂow coming from FC2 for T 223 K 56 CHAPTER 4. EXPERIMENTAL SETUP

Experimental Settings for NO2-Measurements

In this section the experimental settings and conditions for NO2 measurements are listed (table 3.2). As illustrated in ﬁgure 4.15 the complete NO2 absorption spectrum in the 240-790 nm range can be covered with six measurements under diﬀerent experimental conditions, whereas it has to be noted that during the measurements N2O4 is produced as well and has to be extracted. The required procedure to achieve N2O4 corrected NO2 absorption spectra will be described in chapter 8.1.

100 2,77e14 6

8,33e14 5 10 2,50e15 4 [a.u.] 7,50e15 3 1 2,25e16 2

6,75e16 1

0,1 FPA- 1 FPA- 2 FPA- 3 FPA- 4

100 200 300 400 500 600 700 800 Wavelength (nm)

Figure 4.15: Absorption spectrum of NO2 with diﬀerent regions of interest. The numbers on the left give roughly the achieved NO2 concentrations in the certain experimental settings. 4.2. THE CATGAS SETUP 57

Temperature Mixture FC[10] FC[2000] FC[10000] FC[5000] [Kelvin] NO2-Flow NO2-Flow Dilution Flow to Vessel [sccm] [sccm] [sccm] [sccm] 293 1 5,5 - 3300 1000 2 10 - 2800 1000 3 10 - 1050 1000 4 - 50 2200 1000 5 - 90 1800 1000 6 - 350 1800 1000 273 1 4 - 3600 1000 2 4 - 1400 1000 3 10 - 1300 1000 4 4 - 1200 1000 5 - 93 1250 1000 6 - 140 1250 1000 243 1 2,8 - 2000 1000 2 5 - 1300 1000 3 10 - 1200 1000 4 10 - 800 1000 5 - 50 1600 1000 6 - 250 1500 1000 223 1 2,8 - 1500 1000 2 5 - 1500 1000 3 8,5 - 2000 1000 4 - 37 1600 1000 5 - 170 1200 1000 6 - 220 1200 1000

Table 4.2: Experimental conditions for NO2 measurements 58 CHAPTER 4. EXPERIMENTAL SETUP

4.3 The Optical Interface

The light coming from the CATGAS setup is to be guided to the GOME-2 instrument. This is done by a multi fibre bundle and and optical interface. The fibre bundle has a circular (front) entrance and a rectangular exit (spatial), as seen in figure 4.16.

(a) Fibre-Entrance (b) Fibre-Exit

Figure 4.16: Fibre bundle connecting the CATGAS set-up and TPD optical interface

The detailed characteristics are listed in the following table:

Length 5 m Diameter single ﬁbre 70 µm Front aperture Ø 6.2 mm Exit aperture 19 x 1.6 mm NA ﬁbre 0.25

These characteristics, especially the exit aperture and NA of the ﬁbre need to be adapted to the GOME-2 instrument spatial entrance acceptance area and ﬁeld:

FOV 2.75◦ x 0.286◦ Aperture 25 x 5 mm

In order to correctly transfer the light coming from the ﬁbre into the spectrometer TPD developed an optical interface consisting of a 3 mirror system. Figure 4.17 illustrates the optical and mechanical layout of the interface3, the according characteristics are given below. Mirror 2 and 3 are coated with an aluminium layer, furthermore mirror 2 is optimized for the UV-range by having an additional UV enhanced layer. Due to the high optical surface quality of mirror 1 (manufactured in the TPD optical workshop on an optical precision diamond turning machine) an aluminium layer on this mirror is not needed. Further details regarding optical homogeneity and detailed calculations regarding the interface are given in [65]

3Kindly provided by TPD 4.3. THE OPTICAL INTERFACE 59

G F E Topview

TVC

CATGAS fibre D A B C

A: Fibre input (from the CATGAS setup) B: Optical rail C: Mirror 1 f = 700mm; D (min) = 210mm (parabolic and off-axis) D: Mirror 3 f = ; D (min) = 200mm (spherical) E: Mirror 2 f = 3175mm; D (min) = 317mm (fold mirror) F: Optical rail G: Focal plane (on the GOME scanning mirror)

Figure 4.17: Optical and mechanical layout of optical interface [65]

Characteristics of optical interface:

Input NA 0.13 ( total light cone 15◦ ) ApertureStop D=183mm ∼ Magniﬁcation 4.5 Image Dimensions (on 85.5 mm ( length ) GOMEScanningmirror) 7.2mm(width)

Chapter 5

Data-Aquisition

5.1 Measurement-Procedure with GOME-2

Operating CATGAS starts with turning on the system and letting it stabilize for about two to seven hours depending on the desired temperature. Whether the system is stable or not, after the desired settings (gas flow, pressure in the vessel, temperature) are set, can be checked with the commercial spectrometer. Once stable conditions are reached the off-axis parabolic mirror No.2 (see figure 4.2) is removed and measurements with GOME-2 can be done. For a measurement with GOME-2 the following sequence was performed:

1. Measurement of ﬁrst reference vessel contains no gas of interest (O , NO ) ⇒ 3 2 2. Measurement of absorption measurements with gas of interest in the vessel

3. Measurement of second reference vessel contains no gas of interest ⇒ Each measurement consisted of the following sub-steps: First the TTL-controlled flip- mirror was in a position, where the light going trough the vessel (signed as ”VESSEL”) was detected. In this position two measurements were taken by GOME-2. The flip- mirror switched then in a position, where the light coming directly from the Xe-lamp (signed as ”DIRECT”) could be monitored. Also two measurements were taken. Then the ratio VESSEL / DIRECT was build. This sequence was repeated three times. These ”ratio-files” contain the mean value of these three pairs. The label gives reference to each measurement, containing information about mixture, temperature, date, time and type of measurement(O3 or NO2, reference or absorption):

ratio−2003-02-03−14−08−47−O3b1m1t203r−01.FPA-1.dat

These steps are provided by TPD. The data is in ascii-format and contains pixel number, wavelength, ratio average, ratio standard deviation, ratio max and min.

61 62 CHAPTER 5. DATA-AQUISITION

5.1.1 Documentation of measured Intensities I0 and I A further required procedure in preparation for measurements were the settings of the pixel exposure times (PET). For a linear response of the photodiodes the signal should not exceed 50.000 bu and should not fall below 3 % of this maximum value. The documentation of the light intensities in binary units therefore provides an additional technical quality check.

Table 5.1 shows by example the light intensities in binary units for measurements with GOME-2 FM2-1 at 293 K. Column 1 gives the respective mixture, column 2 and 3 the regarding relevant channel and wavelength range. Column 4 and 5 contain the intensities in binary units for I0 (reference) and I (absorption), respectively. Regarding mixtures 6 to 8 there is additional information given. These particular wavelengths correspond to certain Hg-lines (253nm, 289nm, 296nm, 302nm) and are in- dicators for the quality of the spectra in comparison to the literature, especially at 253 nm. Therefore these information are included in the table.

The table is followed by graphical illustration of the intensities. The upper part of the respective ﬁgure shows the complete channel, the lower part a zoom of the actual corresponding, relevant wavelength range. Only the measurements at 293 K are given here as an example. Appendix B contains the complete documentation (tables and graphics) regarding the intensities at all temperatures and all FM’s.

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 15000 - 18000 5000 - 17000 1 3 400 - 455 5000 - 15000 4000 - 8000 2 2 340 - 360 20000 5000 - 15000 2 3 430 - 490 10000 - 20000 8000 - 15000 3 2 310 - 345 5000 - 18000 2500(low c) - 15000(high c) 3 3 475 - 600 20000 - 33000 15000 - 5000 3 4 590 - 790 5000 - 27000 2000 - 15000 6 1 298 - 310 17000 - 5000 7000 - 2500 Hg(302) 16000 7000 6 2 310 - 315 7000 - 22000 5000 - 13000 7 1 285 - 303 15000 - 28000 6000 - 25000 Hg(289) 20000 10000 Hg(296) 26000 20000 8 1 240 - 290 2000 - 20000 1800 - 19000 Hg(253) 3000 2000

Table 5.1: Intensities in binary units for measurements with GOME-2 FM2-1 @ 293 K

Please note that this documentation is only given for O3 measurements. One reason is the documentation in the UV region (especially at 253 nm, which is an indicator for the quality of the spectra), where the light intensity drops rapidly due to the properties of the Xe-lamp and the optics. In general, the handling, measurement and analysis of O3 is more diﬃcult and critical than NO2 measurements. 5.1. MEASUREMENT-PROCEDURE WITH GOME-2 63

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 3 16000 14000 2.5

12000 2 10000 1.5 8000 Intensity [b.u.] Intensity [b.u.] 6000 1 4000 0.5 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 16000 14000 1.5 12000 10000 1 8000 Intensity [b.u.] Intensity [b.u.] 6000 0.5 4000 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure 5.1: Intensities regarding mixture 1 measured with GOME-2 FM2-1 @ 293K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3

3 15000 2.5

2 10000 1.5 Intensity [b.u.] Intensity [b.u.] 1 5000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2

15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

5000 0.5

340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure 5.2: Intensities regarding mixture 2 measured with GOME-2 FM2-1 @ 293K. The lower graphics show the corresponding relevant wavelength range 64 CHAPTER 5. DATA-AQUISITION

4 4 x 10 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3

3 1.5 2.5

2 1 1.5 Intensity [b.u.] Intensity [b.u.] 1 0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 4 x 10 x 10

3 1.5 2.5

2 1 1.5 Intensity [b.u.] Intensity [b.u.] 1 0.5 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure 5.3: Intensities regarding mixture 3 measured with GOME-2 FM2-1 @ 293K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3

4 x 10 Mixture 3 FPA−4

2.5

1.5 Intensity [b.u.]

0.5

600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure 5.4: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM2-1 @ 293K. 5.1. MEASUREMENT-PROCEDURE WITH GOME-2 65

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 16000 2.5 14000 12000 2

10000 1.5 8000 Intensity [b.u.] 6000 Intensity [b.u.] 1

4000 0.5 2000 220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

4 x 10 16000 14000 2 12000 1.5 10000 8000 1 Intensity [b.u.] ← Hg−Line @ 302nm Intensity [b.u.] 6000 4000 0.5 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure 5.5: Intensities regarding mixture 6 measured with GOME-2 FM2-1 @ 293K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 7 FPA−1 x 10 Mixture 8 FPA−1

2.5 2.5

2 2

1.5 1.5

Intensity [b.u.] 1 Intensity [b.u.] 1

0.5 0.5

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

4 x 10

5000 2.5

4000 2 ← Hg−Line @ 296nm 1.5 3000

Intensity [b.u.] 1 ← Hg−Line @ 289nm Intensity [b.u.] 2000 ← Hg−Line @ 253nm 0.5 1000 285 290 295 300 250 260 270 wavelength [nm] wavelength [nm]

Figure 5.6: Intensities regarding mixture 7 and 8 measured with GOME-2 FM2-1 @ 293K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done. 66 CHAPTER 5. DATA-AQUISITION

5.2 Reference-Interpolation and Calculation of Opti- cal Densities

With the ratio-files in ascii-format the following procedure is required. As described above a sequence starts with a reference measurement. In case of O3-measurements this means that the cell is filled with pure O2. That is followed by an absorption measurement, the vessel is now filled with an O3 /O2 mixture. After the absorption measurements the vessel is filled again with pure O2 and another reference measurement is taken. The second reference measurement is required to characterize a baseline shift, meaning whether and how the optical property of the setup has changed. This can be expressed in terms of optical density by applying the natural logarithm to the ratio of both reference measurements:

I0,1 ODBaseline = ln (5.1) I0,2 !

Since a change is very likely the absorption measurements, or more precisely the reference measurements before and after the absorption measurements, need to be time- interpolated, before the optical density can be calculated by applying Lambert-Beer’s law. The assumption is made that the optical property changes linearly, therefore a linear time-interpolation is applied. May

I0,1 I0,2 ODRef1 = ln and ODRef2 = ln (5.2) ILM,0,1 ! ILM,0,2 !

be the reference-measurements at times t1 and t2 before and after an absorption measurement at time ti. The ﬁrst step is to ﬁnd the time-interpolated reference at time ti, on which together with the absorption measurement at time ti the Lambert-Beer law can be applied. The interpolated reference can be obtained by the following equation:

t2 ti ti t1 OD i = OD 1 − + OD 2 − (5.3) Ref Ref · t t Ref · t t 2 − 1 2 − 1 This leads to Lambert-Beer:

I0,i IA,i OD = ln = ODRefi + ODAbsi (5.4) ILM,0,i · ILM,A,i ! This sequence is also illustrated in figure 5.7 and consists of 12 single operations. We have developed a software package for automation of this procedure. The software tool is started in a prepared directory and goes then through the subdirectories and calculates automatically the baselines and optical densities (figure 5.8), with which the actual compounding of the spectra can be started, at least for O3. The NO2-measurements are still superimposed with an absorption spectrum of the dimer N2O4. This has to be separated first, which was described in chapter 8.1. 5.2. REFERENCE-INTERPOLATION AND CALCULATION OF OPTICAL DENSITIES67

Reference 1 Æ18:27 120 min Absorbtion Æ20:27 189 min 69 min Reference 2 Æ21:36

69 min OD(Reference 1) = ln(REF1) 189 min + Sum – OD(ABS) 120 min OD(Reference 2) = ln(REF2) Interpolated OD 189 min

12 Mathematical Operations

Figure 5.7: Calculation of time-interpolated optical density

293K RUN Output Block 1 (Ascii) Mix 1 Absorbtion OD@T293Kb1m1 OzoneIN OD@T293Kb1m1 OzoneOUT OD@T293Kb1m1 References BaselineT293Kb1m1

Mix 2 Absorbtion OD@T293Kb1m2 OzoneIN OD@T293Kb1m2 OzoneOUT OD@T293Kb1m2 References BaselineT293Kb1m2 Block 2 Mix 3 Absorbtion 15 - 17 Absorbtion 19 - 23 References 14 - 18 < 30 sec References 18 - 24

Figure 5.8: Directory-structure and application of software tool 68 CHAPTER 5. DATA-AQUISITION

5.3 Concatenation of a Spectrum

This section contains the applied method and procedure for the concatenation of the certain parts of the spectra to one overall spectrum. As an example we use the O3- measurements at 293 K, the basic principle is valid for the NO2 spectra as well. The gluing has been done manually using the scientiﬁc software origin, the description of the speciﬁc projects is given in section 5.4.

An optical density between 0.1 and 1 is assumed to be optimal in the O3 measurements, in this range the pixel response∼ is expected to be linear. Therefore the gluing of two measurements is to be done in an overlapping region, where both measurements fulﬁll this requirement. It can be stated though that optical densities of up to 2.5 showed also good results due to the high linearity of the GOME-2 instruments. The starting point is a measurement, where the optical density at 576.96nm (Hg-Line) in channel 3 is around 1, which is usually the case with experimental settings referring to mixture 3 (table 4.1). This measurement is then as a ﬁrst approximation scaled to 0.0477, representing the average absorption cross section (in units of 10−19cm2)1 at 298 +/- 5K, as given in the ”Critical Review of the Absorption Cross-Sections of O3 in the 240 - 790 nm Region” by Johannes Orphal [66]. This step is illustrated in Figure 5.9.

Mix3a

OD @ 576,96nm

(Hg-Line)

Mix3a / 16.60754717

0,1

caDe l n si ty

Opti 0,01

x.Mix3a

0,0477 @ 576,96nm

1E-3 (Hg-Line)

350 400 450 500 550 600

W avelength (nm)

Figure 5.9: Initial scaling @ 576.96nm (Hg-line)

One advantage of measurements with the GOME-2 spectrometer is the simultaneous coverage of the Chappuis band and parts of the Huggins band.

1In [66], page 43, Table 6-11 the unit for this point is erroneously given in 10−20cm2 5.3. CONCATENATION OF A SPECTRUM 69

The same scaling factor is therefore then transferred to channel 2 of the same measurement, covering a part of the Huggins bands (ﬁgure 5.10).

Mix3a.cut

x.Mix3a.cut

Mix3a / 16.60754717

Channel 2 Channel 3 ity

0,1 ns De al

0,01 Optic

1E-3

1E-4

300 350 400 450 500 550 600

W avelength (nm)

Figure 5.10: Equal scaling factor of the same measurement in channel 2

According to the current scientific state the temperature dependence in the peaks of the Chappuis band is not expected to be larger than 1 %. The initial scaling itself does not influence the compounding itself but will allow a first estimation about the quality of the spectra by, for instance, checking the temperature dependence in the Hartley band. It is therefore important to note that this doesn’t necessarily have to be the final scaling to absolute absorption cross sections.

In order to compound now another measurement to mixture 3, in an initial step a wavelength window is defined, in which both mixtures meet the requirement to have optical densities between 0.1 and 1. For this purpose the original optical densities are plotted in one plot. An example is given in 5.11. In the same plot also the scaled measurement 3a, as seen in figure 5.10, is given (labelled as x.Mix3x). One can see that an optimal overlapping region between mixture 2 and 3a is given mainly in the wavelength range 475 to 500 nm. Therefore the next step would then be to scale mixture 2 to the scaled mixture 3a in a way that in this range the congruence between both measurements is as good as possible. The congruence between both measurements in the overlap region can then be verified with the ratio between both measurements. Based on the ratio it can then be decided where to finally cut the measurement and concatenate to the other (figures 5.12 to 5.14). 70 CHAPTER 5. DATA-AQUISITION

1 ity

ns 0,1 De al

475 500

Mix2 0,01 Optic

Mix3a

x.Mix3a

1E-3

350 400 450 500 550 600

W avelength (nm)

Figure 5.11: Compounding of two mixtures

) 2

Mix2 cm

x.Mix2 -19

x.Mix3a 10 (

n 0,1 o ecti

-S ss 0,01 Cro

475 500

1E-3 scaled) (

1E-4

350 400 450 500 550 600

W avelength (nm)

Figure 5.12: Scaling of two measurements in optimal overlap region 5.3. CONCATENATION OF A SPECTRUM 71

1,06

ratio.m2.m3a

1,05

1,04

1,03

1,02

1,01 o

1,00 ati r

0,99

0,98

0,97

0,96

0,95

0,94

480 485 490 495 500

W avelength (nm)

Figure 5.13: Ratio between two measurements in the overlap

Figure 5.14: Concatenation of diﬀerent measurements in channel 3 72 CHAPTER 5. DATA-AQUISITION

5.3.1 Quality analysis of overlap region An overlap region between two measurements is, as mentioned above, in principle deﬁned by the wavelength range, where both measurements have an optical density between 0.1 and 1.5. Some ratios show however that even regions with higher OD can be used. A ﬁnal≈ decision regarding the gluing was therefore not only based on an optimal density but also on the ratio.

In an optimal case the ratio is horizontal around 1 in the corresponding wavelength range. Figure 5.15 illustrates such a ratio between mixture 6 and 7 in channel 1 at 243K. The left part around 295nm corresponds to an optical density of mixture 6 of about 2.5. Also included in the graphics is the mean value with the standard deviation in the certain wavelength window. Both the mean value and the standard deviation in the overlap region can therefore be used as a quality criteria.

In appendix A a table listing of gluing points, overlap region and quality criteria by the above mentioned parameters is given for every gluing point. In general the standard deviation is smaller than 1%, only for the measurements at 223K it amounts about 1.2%.

1,015

ratio.m7.m6

Mean +/- SD

1,010

1,005 tio

1,000 a

0,995

0,990

0,985

296 298 300 302 304

Wavelength (nm)

Figure 5.15: Ratio between two ”scaled” measurements corresponding to mixture 6 and 7 in Channel 1 (GOME-2 FM3 @ 243 K). 5.4. ORIGIN-PROJECTS 73

5.4 Origin-Projects

5.4.1 Data The data handling after the optical densities were calculated was done with the scientiﬁc soft- GOME-2_O3_FM3_T293K_Data ware origin. The description of the certain projects, the labels of the worksheets, graphics, etc. is Block 1 topic of this chapter. As an example still the O3- measurements (measured with FM3) are used. Mix 1 In a ﬁrst step a new project is started and named in the following way: Mix 2 ”GOME-2−FM3−O3−[Temperature]−Data”.

For the data at 293K measured with GOME-2 Block 2 FM3 this leads to:

”GOME-2−FM3−O3−T293K−Data”. Mix 3 In this project the relevant data, meaning the reference interpolated and calculated opti- Block 3 cal densities are imported in respective subfolders and origin-worksheets. Not every mixture Mix 6 is useful for every channel. For the O3 analysis usually the following mixtures are imported: Mix 7

Channel 1: Mixture 6 - 8 Mix 8 Channel 2: Mixture 1 - 6 Channel 3: Mixture 1 - 3 Channel 4: Mixture 3 Figure 5.16: Origin-Project ”Data”

The project itself contains several subfolders named Block1, Block2, Block3, which in itself consists of several folders as well, named by the mixtures. The directory structure in such a project is therefore as in ﬁgure 5.16. The NO2-projects diﬀer only slightly: The subdirectories are not divided into blocks but only into mixtures and the project has also (besides ”Data”) a further subdirectory labelled as ”[Temperature]−Eisinger−corrected−data”, which contains the worksheets from the N2O4-correction.

The certain subfolders contain worksheets which are labelled as follows: ”T293Km8c1” for instance contains the information:

T293K ... Temperature 293K • m8 ... Mixture 8 • c1 ... Channel 1 • 74 CHAPTER 5. DATA-AQUISITION

Additional information is given for worksheets of mixture 1 and 2 in the O3 projects:

O3IN ... Optical densities during the time period, • when O3 was ﬁlled in the vessel O3OUT ... Optical densities during the time period, • when O3 was evacuated ABS ... Actual absorption measurement, corresponds • usually to the last measurement of O3IN

The columns are named as follows:

LambdaCh1 ... Wavelength channel 1 • Mix8 ... Column contains the optical densities corresponding • to mixture 8 Mix8.cut ... A ﬁrst cut version of column mix8, where parts with • saturation or bad signal to noise have been cleared in the worksheet 5.4. ORIGIN-PROJECTS 75

5.4.2 Spectra

With the imported data the project is saved as ”GOME-2−FM3−O3−T293K−Spectrum”. GOME-2_O3_FM3_T293K_Spectrum

That means we have two separate projects Data containing the data. Not only is this a safety back up, but also simplifies the trouble shoot- Spectrum ing in case of need. It was part of our phi- Channel 1 losophy during the analysis that with every Overlap_m6_m7 new operation a new project, file, worksheet or column would be created. In this way a Overlap_m7_m8 permanent check up would be given (for in- Channel 2 stance after ”copy” and ”paste” actions) and, Overlap_m2_m1 in case of errors, it could be easily gone back one step in the procedure. Overlap_m3a_m2 In the new project a subfolder is created Overlap_m3a_m3b named ”Spectra”, in which the actual com- Overlap_3b_m3c pounding was done in separate files, which are named by the channel. Every ”Channel”- file itself contains also folders regarding the compounding of two measurements. These Channel 3 are named ”Overlap−[Mixture]−[Mixture]”. Channel 4 Figure 5.17 illustrates the directory structure in the ”Spectrum”-project. Figure 5.17: Origin-Project ”Spectrum” The content in the certain files are as follows:

”Spectrum” • – Channel 1: Worksheet ”T293c1” (Temperature 293 K Channel 1) ∗ This worksheet contains initially the columns ”LambdaCh1”, ”Mix8.cut”, ”Mix7.cut”, ”Mix6.cut”, which are imported from the ”Data”-ﬁle. During the analysis procedure this is then added by further columns, for example like

x.Mix8.cut ... scaled version of Mix8.cut x.Mix8.cut2 ... ﬁnal cut version x.Mix7.cut ... scaled version of Mix7.cut ... 76 CHAPTER 5. DATA-AQUISITION

The following table illustrates this form:

LambdaCh1 Mix8.cut x.Mix8.cut x.Mix8.cut2 ... : : : : ... 288,2141 0,20928 16,90148 16,90148 ... 288,3226 0,20701 16,69759 16,69759 ... 288,4312 0,20485 16,52362 - ... : : : : ...

Subfolder ”Overlap−m7−m8” This subfolder contains the information ∗ regarding the compounding of mixture 7 and mixture 8. It includes the following worksheet and graphics: Worksheet "Ratm8m7c1 - Ratio x.Mix8.cut / x.Mix7.cut - Channel · 1"

LambdaCh1 x.Mix8.cut x.Mix7.cut ratio.m8.m7 : : : : 288,2141 16,89944 16,90148 0,99988 288,3226 16,71632 16,69759 1,00112 288,4312 16,54128 16,52362 1,00107 : : : :

Graphics "T293Km7m8c1" · Shows the initial optical densities without scaling, important for validating the overlap in the proper wavelength range

Graphics "Overlapm7m8c1 - Scaled data x.Mix7.cut and x.Mix8.cut · - Channel 1" Shows the scaled optical densities of both mixtures, in this case mixture 7 and 8.

Graphics "GrRatm7m8c1 - Ratio x.Mix7.cut / x.Mix8.cut - Channel1" · Shows the ratio between mixture 7 and 8, mainly in the overlapping region.

It is likewise as described above for the certain ”Channel”-ﬁles and its subfolders. Once the individual gluing has been done, the certain parts have to be concatenated to the overall spectrum. The ﬁle ”Spectrum” consists therefore, beside the certain subdirectories, of the following worksheets: 5.4. ORIGIN-PROJECTS 77

– Worksheet ”cut2c1” ... ”cut2c4”: When the data have been scaled and cut to the ﬁnal state (labelled by ”cut2”), they have been copied in an extra worksheet. This also gives at the same time a chance for a validation-check:

The following table illustrates this form:

LambdaCh1 x.Mix8.cut2 x.Mix7.cut2 x.Mix6.cut2 : : : : 288,2141 16,89944 – – 288,3226 16,71632 – – 288,4312 – 16,52362 – 288,5397 – 16,37435 – : : : : 300,5292 – 3,60879 – 300,6368 – 3,56429 – 300,7444 – – 3,51657 300,8521 – – 3,456

– Worksheet ”FM3T293K” Contains ﬁnally the overall spectra in just one column. The column named "sigma.T293K.FM3" is the spectrum divided by 10−19, representing the absorption cross section in units of cm2.

wavelength T293K sigma.T293K.FM3 : : : 306,8737 1,57052 1,57052E-19 306,9811 1,53816 1,53816E-19 : : : 78 CHAPTER 5. DATA-AQUISITION

5.4.3 Baselines

The origin-project ”Baselines(T)−FM2-1.OPJ” contains the baselines at all temperatures and mixtures. The directory structure is similar to the one in the data- projects. The worksheets are named as ”[Temperature][Block][Mixture]” (only ”[Tem- perature][Mixture]” for NO2 respectively) and contains all 4 channels with the baselines:

LambdaCh1 BaselineCh1 LambdaCh2 BaselineCh2 ... : : : : ... 288,2141 0,00193 375,4274 0,0021 ... 288,3226 0,00268 375,544 0,0021 ... 288,4312 0,00238 375,6605 0,00221 ... : : : : ...

The graphics are labelled as ”Gr[Temperature][Block][Mixture]” and show the baseline shift in all 4 channels. Figure 5.18 shows an example of the baseline shift for ”GrT293Kb3m7” (measured with FM3):

0,010

BaselineCh1

0,008

BaselineCh2

BaselineCh3

BaselineCh4

0,006 ity

0,004 Dens

0,002 Optical

0,000

-0,002

-0,004

250 300 350 400 450 500 550 600 650 700 750 800

W avelength (nm)

Figure 5.18: Baseline for mixture 7 at 293K

The baseline stability is in general (60-70 % of all measurements) better than 0.5 %, especially for the mixtures 6 to 8. In measurements regarding mixture 1, where the references before and after the absorption measurements have been taken in a time distance of up to 4 hours, the baseline stability is still better than 2 %, usually around 1 %. Very rarely ( 10 % of all measurements the ) we could see a baseline shift larger than 2 %, however in≈ general not more than 3.5 %. Chapter 6

Wavelength Calibration

For the wavelength calibration1 a GOME-2 type external spectral line source (SLS), consisting of five individual lamps, is used which illuminates GOME via an integrating sphere to create an uniform field. The light from the external SLS passes through the TVC window, which has a flat transmission over the GOME- 2 spectral range. The dark signal is taken with the lamps of the set-up covered. Further, the internal SLS is used to do a regular monitoring. This means that for each measurement with the internal SLS a wavelength calibration is performed. The dark signal is measured with the scan mirror in dark position. Before the data is analyzed some general pre-processing is performed. In this pre- processing the following steps are performed:

1. The raw data is averaged for each kind of exposure. 2. The dark-signal is subtracted from the light signal. 3. The dark-signal subtracted light-signal is normalised to 1 second by dividing with the integration time.

After the general pre-processing the detailed wavelength calibration is performed. A pre- deﬁned list of selected lines serves as input for the procedure. This list contains for each line the following information:

1. Literature wavelength in nm 2. Usable pixel window for the line.

Within each pixel window a Gaussian fit is performed to the data. The Gaussian function to be fitted is given by: 2 1 x a1 y = a0 exp − + a3 (6.1) "−2 a2 # where: y is the predicted signal x the pixel number a0,a1,a2,a3 the parameters to be fitted

1Kindly provided by TPD

79 80 CHAPTER 6. WAVELENGTH CALIBRATION

The FWHM is given by:

FWHM = a2√8 ln 2 (6.2) The pixel position of the peak is now equal to a1. After all pixel positions are known for each selected peak the dispersion curve for each FPA channel can be determined. A fourth order polynomial ﬁt through the determined pixel positions is made for each FPA channel, giving the wavelength as a function of pixel. A detailed description regarding all calibration aspects are given in [59-61].

A comparison of the results obtained in independent campaigns with diﬀerent ﬂight models provides a further possibility to validate the consistency of the wavelength calibration.

Figure 6.1 shows the ratios of the O3 absorption spectra at ambient temperature between two FM’s in the 325 to 345 nm range. This region is, as outlined before, an important window for remote sensing applications and is part of the Huggins structures. The ratios show especially between FM2-1 and FM3 residuals of up to 6 to 8 %. This can be reduced by applying a (linear) wavelength shift of - 0.03 nm on the FM2-1 data in channel 2. Note though that the ratio of two data sets with 1-2 % accuracy each can produce uncertainties of 3-4 %.

1,06

Ratio FM2 / FM3

Ratio FM2-1 / FM3

1,04

Ratio FM2-1 (- 0.03nm) / FM3

1,02

1,00 o ti

a 0,98

0,96

0,94

0,92

0,90

320 325 330 335 340 345 350

Wavelength (nm)

Figure 6.1: Ratios of the O3 absorption spectra at ambient temperature between diﬀerent FM’s in the Huggins region

In order to evaluate the wavelength calibration of GOME-2 FM3 a comparison is performed with high resolution data of Voigt et al. [67] in the selected Huggins range. Com- parisons in this region is critical due to the strong influence of small wavelength shifts and spectral resolution. This issue is approached by convoluting the high resolution data with an instrumental line shape comparable to GOME-2 and then comparing with GOME-2 absorption cross sections by applying a non-linear least-squares fitting procedure. This procedure uses five parameters, in which one of them regards to a linear wavelength shift. 81

A detailed description of this procedure will be given in section 7.3.4, where an extensive comparison of the GOME-2 spectra with available literature data sets in the Huggins region has been performed. At this point it shall be sufficient to illustrate the results of the comparison of GOME-2 FM3 at ambient temperature with the data set of Voigt et al. [67]. The upper graph in figure 6.2 shows the two compared spectra (after the adjustments) together with a retrieved baseline difference. In the lower graph the residuals are shown. The interesting value at this point though is the applied wavelength shift, i.e. -0.005 nm for GOME-2 FM3. This underlines the good wavelength calibration of FM3 (in channel 2) and the necessity for a shift of the other FM’s with respect to FM3.

2,0

IUP FTS T293K ) 2

GOME-2 FM3 T293K m c 1,5 Baseline -20

Wavelength-shift GOM E-2 FM 3 : - 0.005 nm (10 s

Ratio (GOM E-2 FM 3 / IUP FTS) : 0.971

1,0

Section

0,5 ss o Cr

0,0

324 326 328 330 332 334 336 338 340 342 %

10 n

5 s l a

idu -5 s e -10 R

324 326 328 330 332 334 336 338 340 342

Wavelength (nm)

Figure 6.2: Comparison of O3 data from GOME-2 FM3 with high resolution data of Voigt et al. in the Huggins region

A similar validation has been done with the obtained NO2 spectra. For the NO2 data a different window is important for remote sensing applications, i.e. the main DOAS region between 400 and 500 nm (channel 3). Figure 6.3 shows, similar as figure 6.1, the ratios of the NO2 absorption spectra at ambient temperature between different FM’s between 400 and 425 nm. The largest residuals appear again in the ratio between FM2-1 and FM3. A linear wavelength shift of -0.04 nm, apllied to FM2-1, leads to a significant improvement (green line). The evaluation of the wavelength accuracy of FM3 has been performed by a comparison with the high resolution data of Vandaele et al. [68] (illustrated in figure 6.4), which has been recommended as a standard in the literature [69] (the data of Voigt et al. [70] show in this case a significant baseline error. This will be also clarified in chapter 8.3). The agreement with the data of Vandaele et al. [68] is in principle very good (as given by the ratio 1.001), but a wavelength shift of - 0.037 nm is to be applied for FM3. 82 CHAPTER 6. WAVELENGTH CALIBRATION

1,06

Ratio FM2 / FM3

1,05

Ratio FM2-1 / FM3

1,04 Ratio FM2-1 (- 0.04 nm) / FM3

1,03

1,02

1,01 o ti

a 1,00

0,99

0,98

0,97

0,96

0,95

0,94

400 410 420 430 440 450 460 470 480 490 500

Wavelength (nm)

Figure 6.3: Ratios of the NO2 absorption spectra at ambient temperature between diﬀer- ent FM’s

75,0

) 70,0 2

65,0 m

60,0 c

55,0 -20

50,0 (10

45,0 s

Wavelength-shift GOM E-2 FM 3 : - 0.037 nm

40,0

Ratio (GOM E-2 FM 3 / Vandaele) : 1.001

35,0

30,0

25,0 Vandaele T294K (* HWHM 0.25) Section

20,0

GOME-2 FM3 T293K ss

15,0 o

Baseline

10,0 Cr

5,0

0,0

-5,0

406 408 410 412 414 416 418 420 422 424 %

5 n

i s l a

idu s e -5 R

406 408 410 412 414 416 418 420 422 424

Wavelength (nm)

Figure 6.4: Comparison of NO2 data from GOME-2 FM3 with high resolution data of Vandaele et al. [68] Chapter 7

O3-Absorption Measurements with GOME-2

This chapter documents the O3 absorption measurements performed with the GOME-2 satellite spectrometers and is organized as follows: Integrated Cross Sections: The first section provides a theoretical background • regarding integrated absorption cross sections and their potential independence of temperature. This will be needed for describing the original motivation and procedure of scaling of relative measurements to absolute absorption cross sections. Scaling: Chapter 7.2 explains in detail the applied procedure for scaling the relative • O3 spectra from the GOME-2 study to absolute absorption cross sections. Final Spectra: Three relative O data sets (each at five different temperatures) • 3 have been obtained from three independent campaigns with different FM’s, i.e. FM2, FM2-1 (a refurbished version of FM2) and FM3. From these data sets one resulting over-all spectrum for each temperature has been derived. These ”final spectra” will be included in the comparisons with the literature data base.

Results and Comparison: In chapter 7.3 the results of the obtained O3 spectra • in this study will be presented and compared with previous measurements available in the literature data base. The discussion will include the following: – Comparison at ambient temperature at ten selected single wavelengths, which are located in the Hartley and Chappuis band – Temperature dependence in diﬀerent spectral regions and selected wavelengths and comparison with previous measurements. Of particular interest is the temperature dependence in the peak of the Chappuis band, where recent measurements proposed a decrease of the absorption cross section of more than 3% with decreasing temperature between 293 K and 203 K [4].

– Comparison of all available spectra in the Huggins bands, an important O3 window for remote sensing applications – Integrated absorption cross sections at diﬀerent temperatures Error Analysis: Finally, in chapter 7.4, an error analysis and propagation will be • given, which, although discussed here, will be equivalent to the NO2 measurements.

83 84 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

7.1 Integrated Absorption Cross Sections

Molecular absorption cross sections that are integrated over entire electronic transitions present two main advantages in comparison to measurements at single wavelengths: They are nearly independent of temperature (which has been observed for O3, NO2, BrO, OClO, H2CO), and they are much less sensitive to instrumental eﬀects such as spectral resolution or accuracy of wavelength calibration. We will now discuss the ﬁrst point, i.e. the temperature invariance, from a spectroscopic point of view.

In general the temperature behavior of the integrated absorption cross section is mainly inﬂuenced by changes of population densities of rotational, vibrational and electronic states with varying temperature. As described in chapter 3.1 the absorption cross section σ(˜ν) and the transition probability (Einstein coef.) for an electronic transition m n build the following relation: →

e σ(˜ν)dν˜ = hν˜nmBnm (7.1) Z The integration is made over the complete range, where the electronic transition takes place. The Einstein coeﬃcient B of absorption is related to the transition moment R of an absorption through the equation

3 8π 2 B = 2 R (7.2) (4πǫ0)3h | | This leads to 3 8π 2 σ(˜ν)dν˜ = hν˜nm 2 R (7.3) Z (4πǫ0)3h | | Equation 7.1 to 7.3 are also valid for ro-vibronic transitions, i.e. electronic transitions in which vibrational and rotational quantum numbers change as well. The integration is then to be made over the entire electronic band system. The extension of the equations above to a complete band system is premised on the Born-Oppenheimer approximation, allowing the separation of the eigenfunction of a ro-vibronic state into three wave functions, i.e. the electronic, vibrational and rotational wave function respectively. Consequently the transition moment of a ro-vibronic transition can be expressed as the product of the three certain transition moments:

e vib rot R = RnmRυ′υ′′ RJ′J′′ (7.4)

In order to obtain the total transition probability Bnm for an electronic band system of the transition m n, all transitions of the lower state (m, υ′′,J ′′) in the higher state (n, υ′,J ′) are to be→ added up. The following rules are to be applied for the summation of vib rot the matrix elements Rυ′υ′′ and RJ′J′′ [16]:

vib 2 Rυ′υ′′ = 1 (7.5) ′ | | Xυ rot 2 ′′ RJ′J′′ = 2J + 1 (7.6) ′ | | XJ This results in the total transition probability for the complete band system: 7.1. INTEGRATED ABSORPTION CROSS SECTIONS 85

3 ′ ′ ′′ ′′ 8π e 2 e Bmn = B(n, m, υ ,J , υ ,J )= 2 Rnm = Bnm (7.7) ′ ′ (4πǫ0)3h | | Xυ XJ e e Rnm and Bnm are electronic parts of the transition moment and the Einstein coeﬃcient e respectively. Rnm is also expressed by

e Rnm = ψ n µψmdτ (7.8) Z ∗ The summation of all transitions (eqn. 7.7) is in accordance with the integration of the absorption cross section over the entire band system (eqn. 7.1). According to 7.7 the integration of all vibrational and rotational transitions of the band system can be expressed by the electronic transition alone. Based on the validity of the Born-Oppenheimer approximation the sum 7.7 is independent of the quantum numbers υ′′ and J ′′ of the lower e state. As a consequence the same value Bnm results for the total transition probability for all thermic populated vibrational and rotational levels υ′′ and J ′′ of the lower state m. The oscillator strength is therefore not inﬂuenced by thermic populated vibrational and rotational states in m. From this it follows that the integrated absorption cross section over an complete band system of an electronic transition m n can be expressed by the e → pure electronic transition moment Rnm:

3 8π e 2 σ(˜ν)dν˜ = hνñm 2 Rnm (7.9) Z (4πǫ0)3h | | Following eqn. 7.9 the temperature dependence of the integral is mainly influenced by e the temperature dependence of the electronic transition moment Rnm. As expressed in eqn. 7.8 the transition moment depends on the wave functions of the related states. The essential question is now, whether and how the the internuclear distance reacts on temperature changes. If the internuclear distances vary with temperature, this would effect the wave functions and accordingly also the transition moment. As indicated in chapter 3.1.2 a change of the internuclear distance results from the anharmonic properties of the electronic potential. Higher temperatures lead to vibrational excitations. The three 16 −1 −1 −1 lowest transition energies for O3 are 1103.14 cm , 700.93 cm and 1042.14 cm for the symmetric stretch, the bending, and the antisymmetric stretch modes ν1, ν2 and ν3 respectively [71]. One can now calculate the population density in the ground state by applying the Boltzmann distribution, as expressed in eqn. 3.4:

Nn (T = 293K) = exp( ∆E/kBT ) Nm − 700 cm−1 = exp − −1.38 10 23J/K 293K! · × 700 cm−1 = exp − −4.04 10 21J! ·

Auxiliary calculation:

−21 E 4.04 10 J − ν˜ = = · = 203.24 cm 1 h c 6.626 10−34Js 3 1010 cm/s × · × · 86 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

−1 Nn 700 cm (T = 293K) = exp −1 = 0.032 Nm −203.24 cm !

′′ That means that at 293 K 96.8 % of the molecules are in the state υ2 = 0 and 3.2 % in ′′ the state υ2 = 1. The same calculation for 203 K leads to a population of more than 99 ′′ % for υ2 = 0. Similar calculations can also be applied to the NO2 molecule (ν1 = 1319.7, ν2 = 749.8, ν3 = 1617.75) [72]. The results are summarized in table 7.1.

O3 NO2 (υ1,υ2,υ3) T=293K T=203K T=293K T=203K (0,0,0) 96.8 % 99.3 % 97.5 % 99.5 % (0,1,0) 3.2 % 0.7 % 2.5 % 0.5 %

Table 7.1: Population of vibrational modes of O3 and NO2 in the 293 K to 203 K temperature range

It can therefore be concluded that the population of vibrational levels change only little with temperature and consequently a change of the internuclear distance is expected to be small. A similar argumentation is provided by the aforementioned Born-Oppenheimer approximation. It is based on the separation of the wave functions of nucleus and electrons, leading to a negligible variation of the electronic eigenfunction with varying internuclear distance. 7.2. ABSOLUTE SCALING OF THE GOME-2 O3-SPECTRA 87

7.2 Absolute Scaling of the GOME-2 O3-Spectra

Absorption measurements can be performed in two ways: Measurements of absolute cross sections require highly accurate knowledge of the influencing parameters, i.e. in particular temperature, optical pathlength and concentration. Relative measurements provide spectra in terms of (dimensionless) optical density, which then need to be scaled to absolute absorption cross section values according to the existing literature data base. In the GOME-2 study the latter method was applied. Based on the given argumentation in the foregoing section about the temperature invariance of integrated cross sections the strategy is now as follows: The relative spectra, which were recorded at different temperatures, are normalized to unity integrated optical density (integration across all bands of an electronic transition) and directly be placed on a common scale, which then provides correct relative temperature dependence among them. Dividing the series of spectra at different temperatures by one spectrum at an arbitrarily chosen reference temperature, yields the temperature dependence relative to the chosen one. The GOME-2 spectra were normalized in the described way based on the sum of two integration regions from 245 to 340 nm and from 410 to 690 nm. This covers the Hartley, Huggins and Chappuis bands and is the best available approximation to the requirement that full electronic bands have to be covered.

Following the normalization the relative temperature dependence of the O3 absorption spectrum was determined at selected wavelengths, where a large number of reference data for determination of absolute absorption cross section as a function of wavelength is available from literature. In the 400 to 450 nm region the differential structure of the relative absorption cross section spectra was determined by subtracting a suitable polynomial of 2nd or 3rd order following the procedure well known from the analysis of atmospheric remote sensing data as zenith sky measurements or Differential Optical Absorption Spectroscopy DOAS, see ref. [73, 74] and references therein. Thereby variations of any background absorption, which could possibly be present in the data, are suppressed and the relative temperature dependence of differential absorption cross section can be determined. Finally the relative temperature dependence for all wavelengths was determined by fitting a suitable polynomial to the found variation of cross section at each wavelength. To scale this series of relative spectra to absolute cross sections, a single scaling factor has to be determined, which scales the whole temperature dependent series at once, rather than determining individual scaling factors for each spectrum at a given temperature. This procedure preserves the correct temperature dependence between them. For the determination of a scaling factor data sets from the literature are required. In the past several measurements of absorption cross sections have been performed at single wavelength as well as over large spectral regions (e.g. [4,75-80]). From this data the relative temperature dependence was determined and compared to that obtained from the relative spectra of this study. However, in the determination of the scaling factor for the obtained relative GOME-2 spectra only absolute determinations of O3 absorption cross section were used. Any relative data, which itself had been scaled to absolute measurements, was rejected. This ensured that the scaling of the GOME-2 spectra is independent of any assumptions which had possibly been necessary in the scaling of the other relative data. With respect to absolute determinations both single wavelength absorption cross sections (from line source measurements) as well as continuous spectra of absorption cross section 88 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

were used. From the latter single wavelength absorption cross sections were determined by interpolation, wherever resolution and shape of the spectrum allowed this. Considered were all cross section data available at temperatures within or near the temperature range used in this study. Selected were the wavelengths 253.65 nm, 289.36 nm, 296.73 nm, 302.15 nm, 334.15 nm and 604.61 nm. At each wavelength a scaling factor was determined between the group of available absolute absorption cross section determinations at various temperatures and the group of the five optical densities from the GOME-2 spectra at that wavelength and at the five temperatures used in this study. The scaling factor was determined by a least squares approach which minimized the deviation of the available reference data points from the scaled polygon of our five data points. The uncertainties of the reference data points as stated in the corresponding references were used as weights in the least squares approach. From each fit a standard error for the determined scaling factor was calculated. The final scaling factor was determined as a weighted average of the factors obtained at the considered individual wavelengths. The individual results and the uncertainty estimations obtained from the fit are listed in table 7.2.

Scaling Scaling Scaling GOME-2FM2 GOME-2FM2-1 GOME-2FM3 [10−16 cm2/molec] [10−16 cm2/molec] [10−16 cm2/molec] 253.65 nm 3.542 ±1.61 % 3.528 ±1.61 % 3.545 ±1.61 % 289.36 nm 3.485 ±0.37 % 3.508 ±0.37 % 3.504 ±0.37 % 296.73 nm 3.454 ±0.41 % 3.454 ±0.41 % 3.484 ±0.40 % 302.15 nm 3.427 ±0.38 % 3.448 ±0.41 % 3.479 ±0.40 % 334.15 nm 3.225 ±0.34 % 3.105 ±0.35 % 3.156 ±0.35 % 604.61 nm 3.485 ±0.63 % 3.495 ±0.63 % 3.462 ±0.64 % Ø 3.510 ±0.11 % 3.508 ±0.11 % 3.522 ±0.11 %

Table 7.2: Scaling Factors at selected wavelengths. These were determined as one factor per wavelength and per set of spectra, i.e. per ﬂight model campaign FM2, FM2-1, and FM3.

The scaling factors agree to within 1 % and better between the three campaigns em- phasizing the consistency and reproducibility of the results. Especially the quality of concatenation of the numerous subsections of the spectrum obtained by the 12 different combinations (compare table 4.1) is clearly confirmed by this. The scaling factor obtained at 334.15 nm in the Huggins band falls low, which is most likely due to systematic effects of different resolution in the reference data and the comparatively low resolved GOME- 2 spectra. For the determination of the final scaling factor for the spectra the results obtained at that wavelength is ignored. Figures 7.1 to 7.6 illustrate the described procedure at the mentioned selected wavelengths. Filled circles represent available reference data. Data used for the determination of the scaling factor is marked by an additional open square. As outlined before, the advantage of this approach is the preservation of the relative temperature dependence between the relative spectra and the inclusion of all reference data, which are available at the selected wavelength and - this is important - at different temperatures. The comparison of the relative temperature dependence with literature data will be discussed later. 7.2. ABSOLUTE SCALING OF THE GOME-2 O3-SPECTRA 89

1,22E-017

Available reference data

Used reference data

1,20E-017

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1

Scaled GOME-2 FM3

1,18E-017

1,16E-017

Section @ 253.65 n m ss

1,14E-017 o Cr

1,12E-017 olute s Ab

1,10E-017

180 200 220 240 260 280 300 320

Temperature (K)

Figure 7.1: Scaling factor from data at 253.65 nm

Available reference data

1,55E-018

Used reference data

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1 1,53E-018

Scaled GOME-2 FM3

1,50E-018 ion @ 289.36 n m ect

1,48E-018

ss o

1,45E-018 Cr

1,43E-018 olute s Ab

1,40E-018

200 220 240 260 280 300

Temperature (K)

Figure 7.2: Scaling factor from data at 289.36 nm 90 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

6,60E-019

Available reference data

Used reference data

6,40E-019

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1

6,20E-019 Scaled GOME-2 FM3

6,00E-019

Section @ 296.73 n m

o 5,80E-019 Cr

5,60E-019 olute s Ab

5,40E-019

200 220 240 260 280 300

Temperature (K)

Figure 7.3: Scaling factor from data at 296.73 nm

3,20E-019

Available reference data

Used reference data

3,10E-019

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1

3,00E-019 Scaled GOME-2 FM3

2,90E-019

Section @ 302.15 n m

ss 2,80E-019 o Cr

2,70E-019 olute s

2,60E-019 Ab

200 220 240 260 280 300

Temperature (K)

Figure 7.4: Scaling factor from data at 302.15 nm 7.2. ABSOLUTE SCALING OF THE GOME-2 O3-SPECTRA 91

Available reference data

5,50E-021

Used reference data

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1

5,00E-021

Scaled GOME-2 FM3

4,50E-021

Section @ 334.15 n m

4,00E-021 ss o Cr

3,50E-021 olute s

Ab 3,00E-021

200 220 240 260 280 300

Temperature (K)

Figure 7.5: Scaling factor from data at 334.15 nm

Available reference data 5,60E-021

Used reference data

Scaled GOME-2 FM2

Scaled GOME-2 FM2-1

5,40E-021

Scaled GOME-2 FM3

5,20E-021

Section @ 604.61 n m

ss o Cr 5,00E-021 olute s

4,80E-021 Ab

200 220 240 260 280 300

Temperature (K)

Figure 7.6: Scaling factor from data at 604.61 nm 92 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

7.2.1 Final GOME-2 O3-Spectra Before comparing the obtained GOME-2 data with the literature data base in this section a ﬁrst comparison is done between the individual FM’s and furthermore one ﬁnal spectrum out of all three data sets (per temperature) is generated. Figure 7.7 illustrates the ratios between the FM’s at ambient temperature over the complete wavelength range covered by GOME-2.

1,10

Ratio FM2 / FM2-1

1,08

Ratio FM2 / FM3

Ratio FM2-1 / FM3

1,06

1,04

1,02 o ti a 1,00

0,98

0,96

0,94

0,92

0,90

250 300 350 500 600 700 800

Wavelength (nm)

Figure 7.7: Ratios between individual FM’s at 293K

Overall one can see a good agreement between the FM’s. The deviations do not extend 4%, except for the minimum range between 370 and 425 nm (for a better illustration the± axis contains a break in this region). Note that the ratio of two data sets with 1-2 % accuracy each can produce uncertainties in their ratios of 3-4 %. In the wavelength area between 370 nm and 425 nm the O3 absorption cross section is very small, i.e. around 10−24 cm2, which is about seven orders of magnitude smaller than in the Hartley band. Therefore small baseline drifts and other inconsistencies, such as possible additional layers on the mirrors due to high O3 concentrations in the vessel, cause large impacts on the measurements. This will be discussed more detailed in chapter 7.3. Figures 7.8 to 7.10 provide a closer look in different regions of the O3 absorption spectrum, i.e. the Hartley band, the Huggins region an the Chappuis band. The agreement in the Hartley band is, despite of a small baseline, very good and lies around 2%. Small structures are still visible, indicating small differences in the wavelength calib± ration. The uncertainties in the Huggins region are larger, showing also small baselines and jumps. Nevertheless the agreement is within 4%. Note that the Huggins region is, due to its sharp structures, sensitive to small differences± in wavelength calibration, which would explain the remaining structures in the ratios. The agreement in the Chappuis band again is very good, about 1.5% at the peak of the Chappuis band. The wings show slightly larger uncertainties. ± 7.2. ABSOLUTE SCALING OF THE GOME-2 O3-SPECTRA 93

1,04

Ratio FM2 / FM2-1

Ratio FM2 / FM3

1,03

Ratio FM2-1 / FM3

1,02

1,01 o ti a 1,00

0,99

0,98

0,97

0,96

230 240 250 260 270 280 290 300

Wavelength (nm)

Figure 7.8: Ratios between individual FM’s at 293K in the Hartley Band

1,10

Ratio FM2 / FM2-1

1,08

Ratio FM2 / FM3

Ratio FM2-1 / FM3

1,06

1,04

1,02 o ti a 1,00

0,98

0,96

0,94

0,92

0,90

310 320 330 340 350 360 370

Wavelength (nm)

Figure 7.9: Ratios between individual FM’s at 293K in the Huggins region 94 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

1,10

Ratio FM2 / FM2-1

1,08

Ratio FM2 / FM3

Ratio FM2-1 / FM3

1,06

1,04

1,02 o ti a 1,00

0,98

0,96

0,94

0,92

0,90

450 500 550 600 650 700 750

Wavelength (nm)

Figure 7.10: Ratios between individual FM’s at 293K in the Chappuis Band

For the use of the spectra in remote sensing applications it is recommended that the certain data sets are used for the corresponding flight model. Nevertheless it appears useful to generate one spectrum (per temperature) out of the three data sets. This has been done by simple averaging of the data as a function of wavelength, whereby the different axes were fitted and interpolated to the axis corresponding to FM3. Comparison of the obtained spectra with the literature data base, i.e. in particular with the data of Voigt et al. [67], has shown the best agreement with the FM3 axis in the main DOAS window between 325 and 345 nm (chapter 6). The detailed results will be discussed in chapter 7.3.4.

It is worth to note that a more detailed analysis regarding wavelength (e.g. non-linear wavelength shifts in every channel) and averaging (e.g. with diﬀerent weights) could be an appropriate following step. This would be though an time intens procedure, which could not be performed in the remaining time frame. Figure 7.11 shows as an example the obtained absorption spectra of O3 in the complete wavelength range at ambient temperature. The averaged spectrum (labelled as GOME-2) has been included in the comparisons with the literature data base, which will be topic in chapter 7.3. 7.2. ABSOLUTE SCALING OF THE GOME-2 O3-SPECTRA 95

1E-17

GOME-2 FM2 )

2 GOME-2 FM2-1

1E-18 m

GOME-2 FM3 c (

GOME-2 (mean FM's) n

1E-19 Sectio

1E-20 ss

o Cr

1E-21 n ptio r

1E-22 o s Ab

1E-23

250 300 350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

Figure 7.11: O3 spectra @ 293 K obtained with GOME-2 96 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

7.3 Results and Comparison with Literature Data Base

Due to its particular importance in protecting the Earth’s surface from harmful solar UV radiation, O3 has been of high scientiﬁc interest. Several laboratory measurements have been performed in the past. These data are of high importance for satellite and ground based remote sensing applications as well as for kinetic studies. As outlined in chapter 1 the theoretical generation of these data are of high complexity. A constant eﬀort in laboratory measurements is therefore required. Despite of the amount of available data sets in the literature data base, there exist still inconsistencies between them.

In this chapter the results of the obtained O3 absorption spectra from the GOME-2 study will be presented and compared with previous measurements available in the literature. The results and comparisons will be discussed for the individual ﬂight models (FM’s) FM2, FM2-1 (a refurbished version of FM2) and FM3 and for the from all FM’s generated spectra (labelled as GOME-2 in the graphical illustrations).

This chapter is organized as follows:

- As a ﬁrst step a comparison of the GOME-2 data with literature values at ambient temperature at ten selected single wavelengths, which are located in the Hartley and Chappuis band, will be given (chapter 7.3.1). This provides a ﬁrst quality evaluation of the GOME-2 data.

- The O3 absorption measurements with GOME-2 have been performed at five different temperatures, i.e. 293 K, 273 K, 243 K, 223 K and 203 K. One main focus is therefore on the temperature dependence of the absorption spectra in different spectral regions and selected wavelengths, in particular in the peak and the blue wing of the Chappuis band, where still significant inconsistencies exist. This will be topic in 7.3.2, followed by a section describing the modelling of the temperature dependence in the Huggins and Chappuis region (7.3.3).

- The 325-345 nm range in the Huggins region is, as mentioned before, an important window for remote sensing applications. In order to evaluate the quality of the obtained data sets from the GOME-2 study, comparisons with available measurements in the literature data base are done and discussed in chapter 7.3.4.

- Derivation of integrated absorption cross sections

The intention of this study is twofold: Clarifying existing inconsistencies, while validating and improving the accuracy of the literature data base and providing reference data for remote sensing applications with GOME-2 as underlined in chapter 1. 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 97

7.3.1 Comparison of GOME-2 spectra at ambient temperature with literature values at 10 single wavelengths In the literature exist many previous measurements at 10 selected wavelengths, corresponding to Hg lamp and He-Ne laser lines, which have been reviewed and summarized for room temperature in ref. [66]. Four of these wavelengths are located in the Hartley band, six in the Chappuis band (ﬁgure 7.12). Ref. [66] furthermore provides averages of absolute determinations of absolute cross sections at room temperature, which were obtained based on diﬀerent selections of available data. For the present comparison only those averages were selected, which were obtained without usage of scaled spectra. This was to avoid dependence of the described approach in this study on any assumptions possibly needed in the scaling of any previous relative spectra. Figure 7.13 shows as an example the O3 absorption cross sections at 293 K obtained with GOME-2 FM3 and the comparison with the recommended values of ref. [66] at the selected wavelengths. The individual values for all FM’s are given in table 7.3.

λ [nm] Ref. [66] RMS[%] FM2 ∆[%] FM2-1 ∆[%] FM3 ∆[%] GOME-2 [%] ∆[%] 253.65 1141 0.9 1116 -2.2 1124 -1.5 1117 -2.1 1128 -1.1 289.36 149 2.0 149 0.2 149 -0.2 150 0.5 151 1.3 296.73 59.8 1.6 60.0 0.4 60.2 0.6 60.0 0.4 60.5 1.2 302.15 29.2 1.8 29.4 0.6 29.2 0.1 29.2 ±0.0 29.5 1.0 543.52 0.311 1.3 0.307 -1.2 0.306 -1.5 0.312 0.2 0.311 ±0.0 576.96 0.472 0.8 0.472 ±0.0 0.471 -0.2 0.478 1.2 0.478 1.3 594.10 0.461 1.2 0.462 0.3 0.461 ±0.0 0.468 1.5 0.468 1.5 604.61 0.518 1.0 0.512 -1.2 0.511 -1.3 0.518 ±0.0 0.518 ±0.0 611.97 0.465 0.7 0.456 -2.0 0.455 -2.1 0.462 -0.7 0.462 -0.6 632.82 0.342 1.2 0.337 -1.5 0.337 -1.4 0.342 ±0.0 0.342 ±0.0 mean values 1.3 -0.7 -0.8 0.1 0.5

Table 7.3: Comparison of GOME-2 data with recommended cross sections at 293 K at 10 selected wavelengths. All cross section values are given in units of 10−20 cm2. The second column gives the average values obtained in ref. [66] (NOTE: Average values of ABSOLUTE measurements) together with the RMS in the third column indicating the 1-σ root-mean-squared diﬀerences of the literature values given in ref. [66]. ∆ indicates the relative diﬀerences between the values of this study and those from ref. [66].

The GOME-2 data show high consistency among the three FM’s as well as in comparison with previous measurements. The deviations of the absolutely scaled GOME-2 spectra from the reference points are all well centered around zero not exceeding the range of 2.2 %. ± The deviations of the GOME-2 spectra from the room temperature reference value at 253.65 nm appear to be slightly larger than the remaining ones. This could indicate either a slight systematic deformation of the GOME-2 spectra or a systematic deviation between the averaged reference data at diﬀerent wavelengths. A further reason could be the following: The reference values from [66] were obtained from room temperature only. Opposed to that the GOME-2 spectra were scaled in a way that considered data points not only at room temperature, but - at each selected wavelength - of all data points within the considered temperature range were used. This is in particular reasonable, when considering the scatter of the reference data in ﬁgure 7.1. 98 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

1E-17

Ozone @ 293K (GOME-2 FM2)

Ozone @ 293K (GOME-2 FM2-1) ) 2

1E-18 Ozone @ 293K (GOME-2 FM3) m c

Ozone @ 293K (GOME-2) ( n

Literature values

1E-19 Sectio -

1E-20 ss

o Cr

1E-21 n ptio

r 1E-22 o s Ab

1E-23

250 300 350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

Figure 7.12: O3 spectra @ 293 K obtained with GOME-2 and comparison at 10 single wavelengths with literature values

1E-16

) Ozone @ 293K (GOME-2 FM3) 2

Literature values m c

Deviation of GOME-2 data to n (

1E-17

literature values

253,65nm Sectio

289,36nm -

1E-18

- 2.1%

ss + 0.5%

296,73nm o

302,15nm

+ 0.4% Cr

+/- 0.0%

1E-19

240 250 260 270 280 290 300 310 320 330

Wavelength (nm)

1E-20 )

604,61nm 2 8E-21 576,96nm m

+/- 0.0%

c + 1.2%

6E-21

543,52nm n (

632,82nm

+ 0.2%

+/- 0.0%

4E-21 Sectio - ss

594,10nm 611,97nm o

+ 1.5% - 0.7% Cr

2E-21

530 540 550 560 570 580 590 600 610 620 630 640 650

Wavelength (nm)

Figure 7.13: O3 spectra @ 293 K obtained with GOME-2 FM3 and comparison at 10 single wavelengths with literature values 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 99

7.3.2 Comparison of relative temperature dependence In the past several measurements of temperature dependent absorption cross sections have been performed at single wavelength as well as over large spectral regions (e.g. [75-80]). These data were converted from absolute to relative temperature dependence and compared to that obtained from the relative spectra in this study, which - by the described normalization to integrated optical density (chapter 7.2)- nevertheless display the correct relative temperature dependence among them. Note that at this point only comparisons have been made with literature data, which was based on measurements of absolute absorption cross sections. The comparison is done at selected wavelength, i.e. at 253.65 nm, 289.36 nm, 296.73 nm, 302.15 nm, 334.15 nm and 604.61 nm,which were also used for the scaling procedure described in chapter 7.2.

Figure 7.14 shows the relative temperature dependence at 253.65 nm as observed in previous measurements [75-79] compared to those obtained from the GOME-2 campaigns.

1,040

GOME-2 FM2

1,035 GOME-2 FM2-1 5 nm GOME-2 FM3 6

1,030

GOME-2 (mean FM's)

Molina & Molina 1986

1,025

Barnes & Mauersb. 1987

Yoshino et al. 1988 1,020

Malicet et al. 1989 ence @ 253. d

1,015

Brion et al. 1993 epen 1,010 -d

1,005

1,000 Relative T

normalized

0,995

200 220 240 260 280 300

Temperature (K)

Figure 7.14: Relative temperature dependence of the O3 cross section at 253.65 nm

Even though by some authors the data by Molina & Molina [75] is considered as overes- timating the true absolute cross section, it is nevertheless considered for relative temperature dependence based on the notion that the relative scaling of the spectra is correct. This is supported by the ﬁgures 7.15 to 7.18. The reference data at 253.65 show an increase of about 1 % with temperature decreasing from 293 K to 203 K. The temperature dependencies obtained in this study agree very well with the observed behavior while yielding a slightly larger increase with temperature. The data from the GOME-2 FM3 100 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

campaign shows a larger deviation at 223 and 243 K, which is most likely due to the signal to noise ratio, which was signiﬁcantly poorer in these particular measurements (combination 8 in Table 4.1). Nevertheless it is worth to note that the deviation is still less than 1% and therefore within the experimental uncertainties of 3%. ± A previous comparison of literature data in ref. [66] showed the appearance of a crossing point around 260 nm, where the temperature dependence changes its direction. In agreement with that, the GOME-2 data at 289.36 nm, 296.73 nm, 302.15 nm, and 334.15 nm show a clear decrease of the O3 absorption cross sections with decreasing temperature, while the magnitude of the temperature dependence increases up to 10 %. This is illustrated in ﬁgures 7.15 to 7.18.

The relative temperature dependence obtained from all three GOME-2 campaigns shows an excellent consistency with previous measurements [75, 77, 79]. It is worth to underline at this point again that only comparisons have been made with literature data, which was based on measurements of absolute absorption cross sections. All of them were converted from absolute to relative temperature dependence. This high degree of consistency found between the available reference data and the GOME-2 data supports the assumption that the selected interval of integration and the quality of our observational data are suﬃcient to indeed preserve relative temperature dependence of spectra.

1,04

GOME-2 FM2

1,03

GOME-2 FM2-1 nm 6

GOME-2 FM3 1,02

GOME-2 (mean FM's)

1,01

Molina & Molina 1986

Yoshino et al. 1988

1,00

normalized Brion et al. 1993 ence @ 289.3

0,99 d

0,98 epen -d

0,97

0,96

0,95 Relative T

0,94

200 220 240 260 280 300

Temperature (K)

Figure 7.15: Relative temperature dependence at 289.36 nm 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 101

1,03

GOME-2 FM2

1,02 GOME-2 FM2-1

GOME-2 FM3 .73 nm

1,01 6

GOME-2 (mean FM's)

1,00 Molina & Molina 1986

Yoshino et al. 1988 normalized

0,99

Brion et al. 1993 ence @ 29

0,98 d

0,97 epen -d

0,96

0,95

0,94 Relative T

0,93

200 220 240 260 280 300

Temperature (K)

Figure 7.16: Relative temperature dependence at 296.73 nm

1,06

GOME-2 FM2

GOME-2 FM2-1 5 nm

1,04 1 GOME-2 FM3

GOME-2 (mean FM's)

1,02

Molina & Molina 1986

Yoshino et al. 1988

1,00

Brion et al. 1993

normalized ence @ 302. d

0,98 epen

0,96 -d

0,94

0,92 Relative T

0,90

200 220 240 260 280 300

Temperature (K)

Figure 7.17: Relative temperature dependence at 302.15 nm 102 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

1,20

GOME-2 FM2

1,15

GOME-2 FM2-1 5 nm 1

1,10 GOME-2 FM3

GOME-2 (mean FM's)

1,05 Molina & Molina 1986

Yoshino et al. 1988

1,00

Brion et al. 1993 ence @ 334.

normalized d

0,95 epen

0,90 -d

0,85

0,80 Relative T

0,75

200 220 240 260 280 300

Temperature (K)

Figure 7.18: Relative temperature dependence at 334.15 nm

The region between 400 and 500 nm, i.e. the blue wing of the Chappuis band, is a further important window for ground based remote sensing applications. This and existing inconsistencies in the temperature dependence of the absorption cross section in the peak of the Chappuis band around 600 nm require accurate attention. The results and the comparison of the GOME-2 data with previous measurements in this range will be sub- ject in the following. There are only a few measurements available in literature.

Part of the Huggins bands and the Chappuis band had been covered simultaneously by a large number of measurements (mainly corresponding to mixture 3 as stated in 4.2.3 and table 4.1). It is reasonable to assume that the systematic behavior of relative temperature dependence is also preserved in the range of 400 to 450 nm and at the maximum of the Chappuis band at 604.61 nm.

At 420 nm a temperature dependent decrease of 40 % when reducing temperature from ≈ 298 K to 220 K was reported [81]. The present data of this study do not support this observation. Relative temperature dependence at 426.1 nm and 429.5 nm as obtained from the three GOME-2 campaigns shows a decrease with falling temperature of at maximum 10 % for the scattered data of FM2 and FM3 and for the more systematic data of FM2-1 of no more than a few percent. This is illustrated in ﬁgure 7.19. 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 103

1,20

FM2 426.1nm (peak)

FM2 429.5nm (trough)

FM2-1 426.1nm (peak)

1,15 429.5 nm d

FM2-1 429.5nm (trough) an

FM3 426.1nm (peak) 1 .

1,10

6 FM3 429.5nm (trough)

1,05

ence @ 42 d

1,00 epen

normalized -d

0,95

0,90 Relative T

200 220 240 260 280 300

Temperature (K)

Figure 7.19: Relative temperature dependence at 426.1 and 429.5 nm

These two wavelengths have been chosen on peak (426.1 nm) and off peak (429.5 nm). As these data points were obtained at comparatively small optical densities - as was the case in the measurements of ref. [81] - the possible effects of instability of throughput and background absorption could have played a critical role in their determination. This can be seen from the behavior of data points, which varied significantly differently between the three series. But checks on stability of optical throughput of those measurements unequivocally proved that the observed behavior was not caused by instabilities of the optical system. Instead, comparison of the temporal behavior of differential structures belonging to the O3 absorption spectrum during the filling of the vessel to temporal behavior of the spectrally broad background clearly showed that both behaved significantly differently. This indicates that in those measurements, which focussed on that spectral 17 region a further absorption must be present. O3 concentrations had to be high at 10 3 ≈ molec/cm due to the O3 cross sections being very small in that region and with O2 as bath gas - both as in ref. [81]. With the available data it is not possible to decide, whether the observed additional absorption is caused by a gas phase absorber - which could possibly beanO3/O2 or O3/O3 dimer - or an interaction of O3 with the molecular layers of H2O on mirrors and windows. Such molecular layers of water are always present at the typical pressures and temperatures of such measurements. Changes in these molecular layers by interaction of O3 and H2O would slightly and spectrally broadly change the reflectance and transmittance of mirrors and windows respectively. The effect would be similar to that observed. Consid- ering this effect as a possible source for the scatter of FM2 and FM3 data in figure 7.19 one can consider the systematic behavior of data of FM2-1 as a criterion which indicates the absence of background effects. This is supported by the fact that the data points of 104 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

FM2-1 show a systematic decrease of cross section of 3 % with falling temperature at 429.5 nm, which is located in a minimum of the band≈ structure. Opposed to that at 426.1 nm - the maximum of the neighboring band - the cross section slightly increases by 1 % with falling temperature. Both observations indicate an increase of differential structure≈ and a decrease of the underlying continuum. But the decrease of the continuum absorption is significantly smaller than that observed in ref. [81]. As their measurements were performed at similarly high O3, with O2 as bath gas and with similarly low optical densities as in the present work, it appears possible that their observed strong decrease of continuum absorption could have been caused by changes of the same continuous background absorption as it was found in the GOME-2 measurements. To eliminate effects of continuous background absorption the differential structure of the O3 absorption spectrum was determined by subtracting a slowly varying polynomial. The amplitude of these differential structures at 426 nm clearly shows a systematic increase of 10 %, 15 % and 15 % (campaigns FM3, FM2, FM2-1 resp.), on average 13 % with falling temperature (293 K to 203 K). This is illustrated in figure 7.20.

4,00E-008

293K

273K

3,00E-008

243K

223K

2,00E-008

203K

1,00E-008

FM2-1

15% peak to peak 0,00E+000

relative to 293K

-1,00E-008

-2,00E-008 Differential absorption (optical density)

-3,00E-008

422,5 425,0 427,5 430,0

Wavelength (nm)

Figure 7.20: Relative temperature dependence of diﬀerential absorption cross section

This is in agreement with the observation stated qualitatively in the text of ref. [81]. Apart from these changes of diﬀerential cross section also slight shifts in wavelength are observed in the 400-500 nm region which is frequently used for O3 retrieval from ground based measurements. This is in agreement with previous publications (e.g. [81, 82]). 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 105

At 604.61 nm at the maximum of the Chappuis band, where larger optical densities are reached with clearly lower O3 concentrations and where no problems with continuous background absorptions were observed, the data by [79-81] show a slight increase of absorption cross section of less than 0.5%, 1% and 0.5% respectively, when temperature decreased from room temperature to about≈ 220 K.≈ Please note: In some publications, e.g. [83], the data of Amoruso et al. [80] is considered as underestimating the absolute cross section. In the present context of relative temperature dependence the data is nevertheless included based on the assumption that the relative magnitude of spectra at diﬀerent temperatures is preserved. Opposed to the increase reported in the aforementioned publications [80, 81, 83] a recent study claims the observation of a clear decrease of 3% over the same temperature range [4]. Figure 7.21 compares the available data to the relative temperature dependence obtained from the average of GOME-2 measurements.

1,05

Mean of FM2/FM2-1/FM3

1,04

Linear fit to mean m

Amurosu et al. 1990 n

1,03 1

Burkh. & Taluk. 1994 6

Brion 1998

04. 1,02 6 Helou et al. 2005 @

1,01

1,00

0,99 ependence

0,98 normalized -d

0,97

0,96 Relative T

0,95

180 200 220 240 260 280 300

Temperature (K)

Figure 7.21: Relative temperature dependence at 604.61 nm

Even though the data points of the present study(average from relative temperature dependence from all three campaigns FM2, FM2-1, and FM3) show some scatter, they clearly show an increase of cross section with falling temperature, which agrees with an increase of 0.5 to 1% as reported in ref. [80, 81, 83]. A linear ﬁt to the GOME-2 data points, which eliminates irregular scatter clariﬁes this even more. The result of the present study does not support a reduction of absorption cross section as reported in ref. [4]. 106 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

The statistical error of the GOME-2 spectra in the region of the Chappuis maximum varies from 0.4% (293 K) to 1 - 2.3% (203 K) which transforms to 0.6% (293 K) to 1.4 - 3.2% (203 K) for the ratio of two spectra, as considered for the relative temperature dependence. A measure for the systematic error is the standard deviation obtained from the average of results obtained in the three different campaigns and which resulted to 0.5 % (293 K) to 1.8 % (203K). As a conservative worst case error estimate the sum of statistical and systematic error is considered and plotted in figure 7.21. Even though the result in ref. [4] at 223 K falls within this conservative (and very likely too pessimistic) error interval of the GOME-2 data, the continuation of their data to lower temperatures lies clearly outside the error interval. One might argue that data which displays an error of the magnitude of our worst case error estimate can not be used to deduce a statement about a temperature dependence below this threshold of error. But the isolated consideration of error bars neglects the systematic trend, which is clearly visible in the GOME-2 data and which can be determined by a linear fit smoothing out irregular variations. The disagreement between on one hand [81, 80, 83] and the GOME-2 data and on the other hand that of [4] is not easily understood, especially as the latter used the original vessel of Brion et al. [83] and their own. The discussion of their result in comparison to previous measurements, especially that of reported in ref. [83] does not indicate possible reasons for the disagreement.

7.3.3 Modelling the temperature dependence of O3 by a polynomial fit In the past several empirical models have been proposed to reproduce the temperature dependence of O3 absorption cross sections (e.g. second-order polynomials, exponential functions etc. See [66] and references therein). Such models help to improve the accuracy of the data since instrumental effects (like straylight or baseline differences), which have no systematic temperature variation can be reduced. In the present study this approach was applied by using a quadratic polynomial fit with three parameters to model relative temperature dependence:

σ(λ, T ) 2 = c0(λ)+ c1(λ) T + c2(λ) T (7.10) σ(λ, T0) · ·

Temperature is expressed in degree Celsius, and T0 is set to 273.15 K. The coefficients obtained from these fits are shown in figures 7.22 and 7.23 for the Huggins and the Chap- puis bands respectively (with FM3 as an example). In the Huggins bands the sensitivity of the cross sections strongly increases towards longer wavelengths, while in the Chappuis band the spectrum is much more stable with respect to temperature changes. Only in the short- and long-wavelength wings of the Chappuis band, the temperature variation of the cross sections becomes more important. 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 107

1,04

1,03 )

1,02 ( 0

c 1,01

1,00

0,99

300 305 310 315 320 325 330 335 340 345 350

) 0,01 (

1 c

1E-3

300 305 310 315 320 325 330 335 340 345 350

1E-4 2 /K )

( 2

c 1E-5

300 305 310 315 320 325 330 335 340 345 350

Vac. Wavelength (nm)

Figure 7.22: Coeﬃcients obtained from a quadratic polynomial ﬁt with three parameters to model relative temperature dependence in the Huggins region

1,02 )

1,00 ( 0 c

0,98

0,96

450 500 550 600 650 700 750

1E-3 /K ) (

1E-4 c

1E-5

450 500 550 600 650 700 750

1E-5 /K )

( 2 1E-6 c

1E-7

450 500 550 600 650 700 750

Vac. Wavelength (nm)

Figure 7.23: Coeﬃcients obtained from a quadratic polynomial ﬁt with three parameters to model relative temperature dependence in the Chappuis region 108 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

7.3.4 Comparison of available spectra in the Huggins bands The Huggins bands consist of a series of individual peaks in the 310 to 380 nm region, showing a strong temperature dependence both due to the varying slope of the Hartley band and to the sharpening of the individual bands at lower temperatures. Comparison of O3 cross sections at single wavelengths in this region is critical because of the strong inﬂuence of small wavelength shifts and of spectral resolution. However, the range between 325 and 345 nm is an important spectral window for remote sensing of atmospheric O3. In order to make a useful comparison in this region a diﬀerent approach is required. Therefore, the following procedure was applied (as already used in ref. [66]):

Convolution of high resolution spectra (Bass and Paur, Brion et al., and Voigt et al.) • with a Gaussian of 0.3 nm FWHM (which is very close to the GOME-2 instrumental line shape as characterized by TPD/TNO)

Comparison of the O3 absorption cross sections with a non-linear least-squares ﬁtting • program using ﬁve parameters:

– a second order baseline polynomial (3 parameters) – a scaling coefficient to adjust the amplitude of the cross sections (1 parameter) – a linear wavelength shift coefficient (1 parameter) to take into account differences in spectral calibration.

The convolution minimizes the influence of spectral resolution when comparing high resolution data with O3 absorption cross sections recorded at lower resolution such as the GOME-2 data of the present study. The non-linear least-squares fitting program is applied to minimize the influence of wavelength uncertainties and of baseline drifts. The comparisons were made with the data sets of Bass and Paur [84, 85], Brion et al. [79, 86, 87] and Voigt et al. [67]. An example of such a fit is given in figure 7.24.

GOME-2 FM3 T293K ) 2

Bass & Paur T293K 2,0 m

c Baseline

0 -2 0

1,5

Wavelength-shift Bass & Paur : 0.028 nm 1

Ratio (Bass & Paur / GOM E-2 FM 3) : 1.003

1,0

0,5 ross Section ( C

0,0

324 326 328 330 332 334 336 338 340 342

-5

-10 Residuals in %

324 326 328 330 332 334 336 338 340 342

Wavelength (nm)

Figure 7.24: Comparison of GOME-2 data with high resolution data in the Huggins bands 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 109

The upper graph shows the two O3 spectra (after the adjustments) together with the retrieved baseline difference. In the lower graph the residuals (differences between the two spectra after the adjustment) are shown. The comparison was made in the 323- 343 nm region at five different temperatures between 203 and 293 K. A summary of all comparisons with the above mentioned high resolution spectra is given in table 7.4 and 7.5. Table 7.4 contains the comparisons for the individual FM’s, table 7.5 the results for the final spectra, which were generated from the individual spectra. Due to the importance of the Huggins region for remote sensing the comparisons have been done at this point for two types of spectra. The first column in table 7.5, labelled with GOME-2, are the comparisons with the spectra generated as described in 7.2.1 by averaging the complete spectrum. The second column, labelled as GOME-2 wls, are spectra, where the averaging procedure was performed in the Huggins region after a wavelength shift for FM2 and FM2-1 was applied, providing an optimized match of all FM’s in this region. This way possible influence of large deviations in other regions, which would influence the averaged results, was avoided.

GOME-2 FM2 GOME-2 FM2-1 GOME-2 FM3 λ-shift ratio λ-shift ratio λ-shift ratio Reference FM2 Ref/FM2 FM2-1 Ref/FM2-1 FM3 Ref/FM3 Voigt et al. [67] 293 K 0.010 0.973 -0.034 0.993 -0.005 0.972 273 K 0.010 1.064 -0.035 1.100 0.000 1.054 243 K 0.013 1.061 -0.030 1.095 0.000 1.054 223 K 0.022 1.084 -0.030 1.095 -0.002 1.029 203 K 0.038 1.186 -0.013 1.157 0.023 1.178 Bass & Paur [84, 85] 293 K -0.018 1.002 -0.037 0.976 -0.028 1.003 273 K -0.018 0.982 -0.029 0.933 -0.022 0.990 243 K -0.017 0.999 -0.026 0.957 -0.021 1.004 223 K -0.016 0.991 -0.033 0.964 -0.028 1.039 203 K -0.014 0.974 -0.019 0.975 -0.017 0.977 Brion et al. [79, 86, 87] 293 K 0.025 1.001 -0.019 0.980 0.010 1.002 273 K 0.008 0.979 -0.033 0.945 - 0.002 0.989 243 K 0.004 1.006 -0.038 0.974 - 0.009 1.013 223 K 0.010 1.002 -0.040 0.990 - 0.013 1.055

Table 7.4: Comparison of individual FM’s with high resolution spectra 110 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

GOME-2 GOME-2 wls λ-shift ratio λ-shift ratio Reference GOME-2 Ref/GOME-2 GOME-2 Ref/GOME-2 Voigt et al. [67] 293 K -0.010 0.955 -0.005 0.958 273 K -0.009 1.038 -0.004 1.041 243 K -0.006 1.060 -0.001 1.063 223 K -0.004 1.049 0.001 1.053 203 K 0.016 1.146 0.021 1.149 Bass & Paur [84, 85] 293 K -0.030 1.020 -0.028 1.018 273 K -0.025 1.002 -0.023 1.001 243 K -0.022 0.997 -0.021 0.995 223 K -0.028 1.019 -0.027 1.017 203 K -0.017 1.003 -0.017 1.002 Brion et al. [79, 86, 87] 293 K 0.005 1.020 0.010 1.017 273 K -0.011 1.005 -0.006 1.002 243 K -0.016 1.007 -0.010 1.004 223 K -0.016 1.036 -0.011 1.032

Table 7.5: Comparison of ﬁnal spectra with high resolution spectra

While the agreement of the data of this study with the spectra of Bass and Paur as well as Brion et al. is good to very good (scaling of generally better than 3 % and of opposite sign with respect to the two data sets), the situation with respect to the FTS spectrum by Voigt et al. is considerably poorer. There in the low temperature measurement at 203 K the deviations reach up to +19 %, +16 %, and +18 % with respect to the three FMs and are of the order of +5 to +10 % for all other but the room temperature spectrum. Based on the agreement of the GOME-2 spectra with the two former data sets, this indicates problems with lamp drift or similar eﬀects possibly being present in the spectra of [67]. Having been obtained with an FTS at high resolution, therefore long scanning times and without light source monitoring, the stability of the lamp could only have been monitored before and after the absorption measurements. Correction had to rely on linear interpolation to the time of the absorption measurement. This makes baseline drifts for these measurements a plausible possibility.

For the remaining reference data the ratios do not show a similarly clear systematic behavior. For FM2 and FM3 the deviations are with the exception of FM3, 223K of the order of 1-2 % and more or less centered around zero, i.e. scaling of 1.0. For FM2-1 the deviations are slightly larger of 3-4 % and all falling systematically low by approx. 1-2 %. The wavelength shifts are given in nm and are deﬁned with respect to the corresponding reference spectrum. As these are either FTS spectra or grating spectra obtained at higher resolution, it is reasonable to assume that their calibration is more precise than that of the GOME-2 instruments. 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 111

For the latter the width of pixels is approx. 0.1 - 0.2 nm, while the targeted wavelength accuracy was at 0.04 pixel [88]. Wavelength calibration of the GOME-2 spectra was provided by the GOME-2 instrument’s calibration measurements by TPD/TNO. Similar as for the ratios also the wavelength shifts for the Voigt et al. spectrum at 203 K indicate a very clear systematic deviation. This is approx. +0.02 to +0.025 nm (i.e. to the red) relative to the others. The surprising fact that this deviation occurs in the same way for all three FMs makes it unlikely that this deviation is caused by statistical errors in the three GOME-2 data sets, which were calibrated independently and obtained independently. Furthermore wavelength calibration for the GOME-2 spectra was identical within one campaign, i.e. for one individual FM at all temperatures. This indicates that the observed deviation is caused by wavelength calibration problems of the Voigt et al. spectrum. A similar but much weaker effect is found for the 203 K data of Bass and Paur displaying an apparently red-shifted 203K spectrum relative to those at the remaining temperatures. On the other hand according to [66] the 203 K spectrum by Voigt et al. shows strongest shifts relative to Bass and Paur while the latter itself shows stronger shifts at low temperature relative to [82]. For the Brion et al. data no such effect occurs in the available data. But it has to be noted that for that data set no 203 K measurement exists. Therefore the GOME-2 data supports the conclusion, which is also indicated by the analysis in ref. [66], that the wavelength calibration of the 203 K spectra by Voigt et al. as well as by Bass and Paur might require a correction to the blue. With the exception of the shifts between FM2-1 and Bass and Paur one can see an excellent agreement of the shifts between reference spectra as found in table 7.4 in [66] and those, which one obtains when comparing the shifts for one FM relative to the corresponding reference spectra. For example the difference of the shifts of FM2 relative to Bass and Paur at 293 K (-0.018 nm) and to Voigt et al. (+0.01 nm) agrees very well with the shift between Bass and Paur and Voigt et al. (-0.029 nm) as given in table ?? in [66]. In the same way the shifts between the three FMs and the Brion et al. data as well as for the Voigt et al. data agree with the shifts determined between the different FM’s data sets underlining the internal consistency of the analysis. In ref. [66] it was shown that with respect to wavelength calibration the spectrum by Voigt et al. at room temperature is likely to be closest to the correct wavelength calibration, i.e. zero shift. This is in agreement with the consistency found for the higher temperature data as discussed above. As a side effect of this analysis the wavelength shifts found between the three FMs can be corrected relative to the spectrum of Voigt et al. at 293 K in order to improve the wavelength calibration of the GOME-2 spectra as well.

Similar tendencies can be observed in comparisons with the ﬁnal spectra in table 7.5, whereby the agreement with the ”Huggins” spectra (averaging after applying a wavelength shift on FM2 and FM2-1 for optimal match in the Huggins region) is slightly better than the ones of the complete spectra (averaging of FM2, FM2-1 and FM3 spectra without a prior wavelength shift). 112 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

7.3.5 Integrated cross sections at diﬀerent temperatures

As mentioned above the integrated absorption cross sections are rather insensitive to differences in spectral resolution and wavelength calibration. Therefore they provide a good means for comparing cross sections in diﬀerent regions. Here especially a comparison between the region of the Huggins band and that of the Chappuis band is of interest. As these regions were in large parts covered simultaneously in a considerable number of our measurements, these measurements provide a well-suited set of data for an intercomparison of absorption cross sections between these two regions. At this point integrated absorption cross sections from the GOME-2 spectra were calculated after they had been scaled to absolute cross sections. This is no limitation to the validity of the intercomparison as this scaling was the same factor for all spectra within a selected campaign (FM2/FM2-1/FM3). The obtained integrated cross sections will all be on the same common scale with respect to their temperature dependence. Cross sections were integrated over the following regions, according to those given in [66]:

- in the Hartley band between 245 and 340 nm: label ”Hartley” - in the Chappuis band between 410 and 690 nm: label ”Chappuis” - in the Huggins bands between 325 and 340 nm: label ”Huggins” - in the blue wing of the Chappuis band between 410 and 520 nm: label ”Blue Wing”

The results are given in table 7.6. They show a high consistency between the three FMs, which is reflected in the small standard deviations of the averages, see bottom block of table 7.6. The results show a clear systematic in the integrated cross section of the Hug- gins band. In the other regions no clear systematic dependence on temperature is found. Considering the data for the Chappuis band it is found that the integral across this region is independent of temperature within less than 1 % corresponding to an average and standard deviation of (6.44 0.05) 10−20 cm2/molec.± ± × From these results the ratios between integrated cross sections from different regions were determined. It is important to note, that these ratios are independent of any scaling, which was performed to obtain the final spectra, as any scaling cancels from a ratio between two regions of the same spectrum. These ratios are the same, if calculated from the original unscaled and solely concatenated spectra. The ratios are listed in table 7.7. They show no significant change with temperature between the integrated cross sections of the full Hartley region and that of the full Chappuis band. The same is true for the integrated values of the ”Blue Wing” relative to the full Chappuis band. But the ratios between the integrated cross sections of the Huggins band relative to the ”Blue Wing” as well as relative to the Chappuis band display a clear dependence on temperature, which has its origin in the temperature dependence of the Huggins band. The values obtained here are a direct measure for intercomparison of absorption cross sections between the different regions. 7.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 113

Hartley Huggins Blue Wing Chappuis 10−16nm×cm2/molec 10−20nm×cm2/molec 10−19nm×cm2/molec 0−20nm×cm2/molec FM2 203K 3.5036 5.45 6.27 6.40 223K 3.5037 5.66 6.20 6.36 243K 3.5036 6.21 6.31 6.53 273K 3.5036 7.10 6.25 6.30 293K 3.5036 8.10 6.28 6.44 FM2-1 203K 3.5014 5.52 6.27 6.38 223K 3.5013 5.87 6.32 6.43 243K 3.5012 6.31 6.39 6.46 273K 3.5014 7.30 6.24 6.35 293K 3.5014 8.12 6.24 6.36 FM3 203K 3.5160 5.70 6.55 6.64 223K 3.5163 5.77 6.26 6.33 243K 3.5161 6.29 6.42 6.54 273K 3.5161 7.19 6.41 6.55 293K 3.5162 8.20 6.36 6.56 average 203K 3.5070 0.0079 5.56 0.13 6.36 0.16 6.47 0.14 223K 3.5071±0.0080 5.76±0.10 6.26±0.06 6.38±0.05 243K 3.5070±0.0080 6.27±0.05 6.37±0.06 6.51±0.04 273K 3.5070±0.0079 7.20±0.10 6.30±0.10 6.40±0.13 293K 3.5070±0.0080 8.14±0.06 6.30±0.06 6.45±0.10 ± ± ± ± Table 7.6: Comparison of integrated absorption cross sections at diﬀerent temperatures and in selected wavelength regions 114 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

Hartley/Chappuis Huggins/BlueWing Huggins/Chappuis BlueWing/Chappuis FM2 203 K 558.71 0.852 0.0869 0.1020 223 K 565.09 0.890 0.0913 0.1026 243 K 555.58 0.951 0.0985 0.1036 273 K 560.56 1.127 0.1136 0.1008 293 K 557.68 1.257 0.1290 0.1026 FM2-1 203 K 558.60 0.865 0.0881 0.1019 223 K 553.70 0.912 0.0928 0.1017 243 K 548.22 0.977 0.0989 0.1012 273 K 561.28 1.148 0.1170 0.1018 293 K 560.60 1.276 0.1300 0.1018 FM3 203 K 536.46 0.859 0.0870 0.1012 223 K 561.42 0.910 0.0921 0.1011 243 K 547.82 0.962 0.0980 0.1018 273 K 548.14 1.098 0.1121 0.1022 293 K 552.46 1.251 0.1289 0.1030 average 203 K 551.26 12.8 0.858 0.006 0.0873 0.0007 0.1017 0.0004 223 K 560.07± 5.8 0.904 ± 0.012 0.0920 ± 0.0007 0.1018 ± 0.0008 243 K 550.54 ± 4.4 0.963 ± 0.013 0.0984 ± 0.0005 0.1022 ± 0.0012 273 K 556.66 ± 7.4 1.124 ± 0.025 0.1142 ± 0.0025 0.1016 ± 0.0007 293 K 556.91 ± 4.1 1.262 ± 0.013 0.1293 ± 0.0006 0.1025 ± 0.0006 ± ± ± ± Table 7.7: Ratios of integrated cross sections from diﬀerent spectral regions 7.4. ERROR ANALYSIS AND PROPAGATION 115

7.4 Error Analysis and Propagation

In this section an overview is given regarding the error budget and error propagation. Although given in the chapter containing the O3 measurements it is worth to note, that the calculations in this section would be equivalent to the NO2 measurements. The systematic error is of the order of 0.001 to 0.005 (”50 % of measurements), around 0.01 to 0.02 (30-40% of measurements) and 0.02 to 0.04 (remaining 10-20% of measurements). All errors in units of optical density. By interpolation of references between the first and last reference measurement, the systematic error can be partly eliminated from the determined spectra of optical density. This elimination is the better, the closer the true temporal behavior of observed drift - i.e. the observed systematic error is to being linear in time. Based on this it is well justified to estimate this error for the majority of measurements (80-90%) to be of the order of 10−3 or even better. The remaining 10-20% of measurements could have larger systematic errors of some 10−2. The statistical error, i.e. the noise on the calculated optical densities, is much smaller. It is determined by firstly removing the systematic error by subtracting a polynomial and secondly determination of the standard deviation of the resulting noise within each channel. The statistical error varies from 2 10−3 in the left (”blue”) half of channel 1 to − · 4.9 10 4 in the right (red) half of channel 1. In the remaining channels 2,3 and 4 it takes on· values of 1.6 10−4, 1.4 10−4, and 1.8 10−4 respectively (all in units of optical density). · · · Both errors have to be put into perspective to the selected optimal range of optical density of 0.1 to 1.0 (O3) and 0.1 to 0.5 (NO2). The dominant contribution results from the systematic error limiting the error of the final product, the spectrum. With 80-90% of measurements having a systematic error of the order of a few 10−3 or even better and a minimum signal of 0.1 optical density the final error of these measurements is of the order of some 10−2 meeting the required accuracy of 3%. As pointed out in the proposal to this study, the error will be larger wherever the maximum optical density reached in the experiment is smaller than 0.1. In such regions the resulting error will be above 3%. Apart from this more general statement, the error propagation was also determined for each spectrum yielding a relative error, see next chapter. The initial error measure of peak to peak variation within three subsequently measured ratios is crude and therefore results in a number of cases into relative errors, which are larger than the demanded 3%. Due to the crude initial error measure the relative error determined in 5.1 rather defines an upper limit of error.

Error Propagation • As described in chapter 5.2 the initial point of calculating optical densities and absolute absorption cross sections are the ratio-ﬁles provided by TPD (see chapter 5.1). These ﬁles contain the following information: Pixel number, wavelength, ratio average, ratio standard deviation, ratio max and min. Figure 5.7 illustrates the mathematical operations required for the calculation of a time-interpolated optical density. The error can be estimated in the following way: 116 CHAPTER 7. O3-ABSORPTION MEASUREMENTS WITH GOME-2

Originally there are three ratio ﬁles from three measurements:

Reference R1, ∆R1 • Absorption A, ∆A • Reference R , ∆R • 2 2 with ∆ = ratio−max - ratio−min. May t1 be the time between R1 and A, t2 the time between A and R2 and tTotal the time between R1 and R2. As illustrated in Figure 5.7 the ﬁrst term of the sum is given by:

t2 OD(R1)=ln R1 (7.11) · tTotal

The uncertainty can be estimated via the standard propagation of errors:

2 2 2 t2 ∆R1 ln R1 t2 ln R1 ∆OD(R1)= v + ∆t + · 2 ∆t (7.12) u tTotal · R1 tTotal ! (tTotal) ! u t

The calculation for the second term is likewise, only t2 is replaced by t1 and R1 by R2. The ﬁnal time-interpolated optical density OD is given by:

OD = OD(R )+OD(R ) OD(A) (7.13) 1 2 −

Accordingly the error can be estimated by:

2 2 2 ∆OD = ∆OD(R1) + ∆OD(R2) + ∆OD(A) (7.14) q

The same way is applied when the optical densities are scaled to absolute absorption cross sections:

σ = OD c (7.15) ·

∆σ = (c∆OD)2 + (OD∆c)2 (7.16) q 7.4. ERROR ANALYSIS AND PROPAGATION 117

Figures 7.7 to 7.10 in chapter 7.2.1 illustrated the ratios between the individual FM’s, indicating the experimental accuracy of the recorded GOME-2 spectra. Figure 7.25 illustrates a further example and shows a similar illustration, i.e. ratios of the O3 absorption spectra at 243 K between diﬀerent FM’s, although here in comparison with the statistical error. As outlined before the agreement is very good in the Hartley and Chappuis region (green boxes) . The deviations in the Huggins region extend only slightly the statistical errors (orange box), while the minimum range shows overall the largest uncertainties (red box). In general it can be said that the ratios are in consistency with the estimated statistical errors.

1,10

Ratio FM2 / FM2-1

1,08

wav elength

Ratio FM2 / FM3

calibration

1,06 Ratio FM2-1 / FM3

Statistical Error

1,04

1,02 o

1,00

Rati

0,98

good

0,96

0,94

minimum range 0,92

low ab so rp tio n

0,90

0,88

250 300 350 500 600 700 800

Wavelength (nm)

Figure 7.25: Ratios of the O3 absorption spectra at 243 K between diﬀerent FM’s in comparison with the statistical error

Chapter 8

NO2-Absorption Measurements with GOME-2

The results and discussion regarding the obtained NO2 absorption spectra will be topic in this chapter. NO2 measurements have been taken at four diﬀerent temperatures, i.e. 293 K, 273 K, 243 K and 223 K. This chapter includes the following sections:

N O -Absorption As outlined in chapter 4.2.4 NO is constantly in an equilibrium • 2 4 2 with its dimer N2O4. The obtained optical densities contain therefore parts of N2O4, which reach, depending on the NO2 concentration and temperature, non negligible amounts. These need to be extracted from the optical densities. The corresponding procedure is given in 8.1

Scaling: Chapter 8.2 describes the scaling of the relative NO2 spectra from the • GOME-2 study to absolute absorption cross sections.

Final Spectra: Similar to the O3 measurements three relative NO2 data sets (each • at four diﬀerent temperatures) have been obtained from three independent campaigns with diﬀerent FM’s. From these data sets one resulting overall spectrum for each temperature has been derived, which will be included in the comparisons with previous measurements.

Results and Comparison: In chapter 8.3 the results of the obtained NO2 spectra • in this study will be presented and compared with previous measurements available in the literature. The discussion will include the following:

- Comparison of GOME-2 data with recommended literature data at ambient temperature (8.3.1)

- Temperature dependence of NO2 spectra obtained from the GOME-2 study (8.3.2)

Of particular interest will be the region between 400 and 500 nm, which is an important NO2 window for remote sensing applications.

119 120 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

8.1 Correction of the N2O4-Absorption

NO2 is constantly in an equilibrium with its dimer N2O4.

2NO2 ↽⇀ N2O4

This equilibrium is shifted towards N2O4 at lower temperature or higher NO2 partial pressure. N2O4 is absorbing below 400nm, therefore the measured NO2 absorption spectra in this spectral range are a superposition of a NO2- and N2O4-spectrum. In order to achieve a pure NO2-spectrum the superposed N2O4-spectrum needs to be separated. For this purpose a method is applied, which was described by Eisinger in 1994 [89]. For application the following requirements need to be fulﬁlled:

constant temperature in absorption cell • Measurement of three absorption spectra at diﬀerent NO partial pressures • 2

1. A spectrum with low NO2 partial pressure and therefore low N2O4-absorption. This will , according to the notation given in [89], in the following be notated as spectrum B and can be seen as the NO2-reference.

2. A spectrum with intermediate NO2 partial pressure (the actual measurement- spectrum). This will be notated as NO2-spectrum or spectrum A.

3. A spectrum with high NO2 partial pressure and therefore strong N2O4-absorption. This will be notated as N2O4-spectrum or spectrum F.

NO and N O in an equilibrium in the cell • 2 2 4

[NO ] and [N O ] const 2 2 4 ≈

The spectral separation is based on the assumption that N2O4 does not absorb at wavelength above 400nm. The spectra to be corrected must therefore content this spectral range. This is due to the broad spectral range of the GOME-2 spectrometer fulﬁlled. The knowledge of the temperature dependent equilibrium constant Kp is not required for the application of this method. In the following the several steps of the Eisinger-procedure are described:

1. Isolation of the N2O4-absorption in the NO2-spectrum and N2O4-reference with respect to the NO2-reference.

(a) Calculation of the relative NO2-absorption in the NO2-spectrum and N2O4- reference with respect to the NO2-reference.

[NO2]A [NO2]F rA = and rF = (8.1) [NO2]B [NO2]B

(b) Subtraction of the scaled NO2-reference from the NO2-spectrum and N2O4- reference respectively. 8.1. CORRECTION OF THE N2O4-ABSORPTION 121

C = A rA B (8.2) − · D = F rF B (8.3) − ·

C and D content only the N2O4-absorption

2. Calculation of the relative N2O4-absorption in spectrum C, in the NO2-reference and in the NO2-spectrum with respect to spectrum D. For spectrum C it is:

[N2O4]C sC = (8.4) [N2O4]D

sC is ascertained by division of the spectra C and D and averaging of the ratios in a pixel range close to the maximum of the N2O4-absorption in channel 2. According to the deﬁnition of C the following expression is valid:

sC = sA + rA sB (8.5) ·

sA and sB are the sought scaling factors. Due to the equilibrium condition, sA and sB can be found in the following relation:

sA rA = (8.6) sB

From this it follows that:

rA sA = sC (8.7) rA 1 · − and rA sB = sC (8.8) rA(rA 1) · −

rA is to be signiﬁcantly larger than 1, meaning that the NO2-concentration in the NO2-reference and the NO2-spectrum must be diﬀerent.

3. Removal of the N2O4-absorption from the NO2-spectrum and the NO2-reference.

E = A sA D (8.9) − · G = B rB D (8.10) − ·

E and G content only the NO2-absorption. 122 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

8.2 Absolute Scaling of the NO2 Spectra obtained from GOME-2 Measurements

The absolute scaling of the NO2-measurements has been done by using integrated absorption cross sections. The argumentation is as explained in chapter 7.1. The integration has been performed in the 400-500 nm range for two main reasons: Most of the available cross sections cover this region and it represents a very important window for remote sensing applications.

Integration of the optical density in the 400-500 nm range •

Dividing the whole spectrum by the obtained integration value •

Multiply the spectrum with 4.50 10−17 cm2 nm, representing the average of all • available data sets as summarized· in ref. [69]. This value is in agreement with the recommendation expressed in ref. [90].

In ref. [69] the appearance of a slight decrease of the integrated absorption cross section with decreasing temperature from 293 K to 223 K results from all comparisons of all data sets. This eﬀect could not be conﬁrmed though within the experimental uncertainties of 3-5 % in the available data sets. Davidson et. al observed that the integrated absorption cross section was essentially independent of temperature within the uncertainty in the data [72], Harwood and Jones concluded as well that any temperature trend lied with the uncertainties in the measurements and that the overall spectral shape is independent of temperature [91]. The above mentioned value has therefore been used for all temperatures.

8.2.1 Final GOME-2 NO2-Spectra

Similar to the O3 section, at this point a first comparison is done between the individual FM’s. Figure 8.1 illustrates the ratios of NO2 measurements at 293 K recorded with different FM’s in the complete wavelength range. Figure 8.2 provides a closer look in the 300 to 550 nm region. As mentioned the 400 to 500 nm range is of particular interest, since it represents an NO2 window for remote sensing activities. The deviations increase rapidly below 260 nm, which is most likely due to low intensities of the reference light and a less good signal to noise ratio. Above 600 nm the absorption cross sections are very small, which causes most likely the observed inconsistencies, similar to the minimum range of the O3 absorption cross section between 370 and 425 nm. In the range between 400 and 500 nm the FM’s show an excellent agreement within 2%, partly even within 1%. ± ± For each temperature a final spectrum has been generated out of the three data sets from different FM’s. These are not presented here though, but have been included in the comparison with previous measurements and will be discussed in the following section. 8.2. ABSOLUTE SCALING OF THE NO2 SPECTRA OBTAINED FROM GOME-2 MEASUREMENTS123

2,0

Ratio FM2 / FM2-1

1,9

Ratio FM2 / FM3

Ratio FM2-1 / FM3 1,8

1,7

1,6

1,5 o

1,4

Rati

1,3

1,2

1,1

1,0

0,9

0,8

250 300 350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

Figure 8.1: Ratios of NO2 measurements at 293 K temperature with diﬀerent FM’s

1,10

Ratio FM2 / FM2-1

1,08

Ratio FM2 / FM3

Ratio FM2-1 / FM3

1,06

1,04

1,02 o

1,00

Rati

0,98

0,96

0,94

0,92

0,90

300 320 340 360 380 400 420 440 460 480 500 520 540

Wavelength (nm)

Figure 8.2: Ratios of NO2 measurements at 293 K temperature with diﬀerent FM’s 124 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

8.3 Results and Comparison with Literature Data Base

In this section the results of the obtained NO2 spectra will be presented and discussed. Two points will be mainly focused:

- First the GOME-2 data will be compared with literature values at ambient temperature

- Following that the temperature dependence will be evaluated.

NO2 has also been topic of many measurements, due to its important role in atmospheric chemistry (chapter 2). Nevertheless, as outlined in chapter 1.5, it is highly desirable to provide reference data for satellite based remote sensing applications, as in the case of GOME-2, which already contain the information regarding the instrumental line shape of the satellite instrument. The second part of this study intends to provide these NO2 reference spectra. It is of high importance that the reference spectra are of a high quality, which will now be discussed in the following sections.

8.3.1 Comparison of GOME-2 data with literature at ambient temperature In the past many laboratory measurements have been performed. A useful summary and critical review is provided in ref. [69]. It is also recommended to define a standard of temperature dependent NO2 absorption cross sections for the entire region 240-790 nm, based on the available laboratory data. It is proposed to use the cross sections of Vandaele et al. (1998) [68] at 294 K and at 220 K together with a linear model to interpolate for intermediate temperatures. This data set is consistent with other recent laboratory measurements of NO2 absorption cross sections at both temperatures, and was recorded at a spectral resolution that is high enough for most remote sensing applications [69]. In the following the GOME-2 spectra are compared with the data of Vandaele et al. [68] and additionally with the data of Voigt et al. [70] at ambient temperature. Figures 8.3 to 8.5 illustrate the available data sets in one graphics and in different wavelength intervals (here only with the individual FM’s). Especially in figure 8.5 the differences in resolution are clearly visible. Nevertheless one can see in general a good agreement, in particular in the important 400 to 500 nm region. Larger deviations with the Voigt et al. data exist above 700 nm. 8.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 125

Vandaele et al. Voigt et al. 1E-18 GOME-2 FM2 GOME-2 FM2-1 ) 2 GOME-2 FM3 1E-19

1E-20 Cross-Section (cm

1E-21

1E-22 300 400 500 600 700 800 Wavelength (nm)

Figure 8.3: Comparison of GOME-2 data with Vandaele et al. and Voigt et al. at ambient temperature

1E-18

Vandaele et al. 9E-19

Voigt et al.

8E-19

GOME-2 FM2

7E-19

GOME-2 FM2-1 ) 2

GOME-2 FM3

6E-19

5E-19

4E-19 ross-Section (cm C

3E-19

2E-19

400 410 420 430 440 450 460 470 480 490 500

Wavelength (nm)

Figure 8.4: Comparison of GOME-2 data with Vandaele et al. and Voigt et al. at ambient temperature 126 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

1E-18

Vandaele et al.

9E-19 Voigt et al.

GOME-2 FM2

GOME-2 FM2-1

8E-19 ) 2

GOME-2 FM3

7E-19

6E-19 ross-Section (cm C

5E-19

4E-19

422 424 426 428 430 432 434 436

Wavelength (nm)

Figure 8.5: Comparison of GOME-2 data with Vandaele et al. and Voigt et al. at ambient temperature

A more detailed evaluation and quantification can be done by building the ratio between the GOME-2 and high resolution data. Figure 8.6 shows the ratio between the GOME-2 spectra and Vandaele et al. data. The ”final” GOME-2 spectrum that was generated out of the three individual FM’S is labelled as GOME-2. The ratio shows a mean deviation of about 2%, only below 280 nm both data sets drift apart significantly. This is however not crucial for remote sensing applications, since the main DOAS window is between 400 and 500 nm. The high noise results from building the ratio of two data sets with different resolution. In figure 8.7 an application of a smoothing procedure (adjacent averaging) is illustrated. The upper graph illustrates the procedure, while the lower graph shows the results for all FM’s and the final spectrum. This also confirms the mean deviation of about 2 %. A baseline can also be seen, which appears stronger for the ratio with GOME-2 FM2. Similar baseline features have been already observed in the Vandaele data and reported in [69]. The deviation of about 2 % is highly consistent with the comparison of integrated cross sections in the 400-500 nm range between several data sets in ref. [69]. The integrated cross section of the Vandaele data set in this wavelength region is 4.58 10−17 cm2 nm, therefore approximately 2% higher than the recommended value in [90] and· [69]. 8.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 127

1,30

Ratio GOME-2 FM2 / Vandaele et al.

1,25

Ratio GOME-2 FM2-1 / Vandaele et al.

Ratio GOME-2 FM3 / Vandaele et al.

1,20

Ratio GOME-2 / Vandaele et al.

1,15

1,10 o

1,05

Rati

1,00

0,95

0,90

0,85

0,80

300 350 400 450 500 550 600

Wavelength (nm)

Figure 8.6: Ratio between GOME-2 data and Vandaele et al.

1,20

Ratio GOME-2 FM3 / Vandaele et al.

1,15

"Smoothing" procedure (adjacent averaging)

1,10 o

1,05 ati

1,00

0,95

0,90

300 350 400 450 500 550 600 650

1,20

GOME-2 FM2

1,15

GOME-2 FM2-1

1,10 GOME-2 FM3 o

GOME-2 ati

1,05 R

1,00

0,95

0,90

300 350 400 450 500 550 600 650

W avelength (nm )

Figure 8.7: ”Smoothing” of the ratio between GOME-2 and Vandaele et al. 128 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

Figures 8.8 and 8.9 show the ratios between the GOME-2 data and Voigt et al. in different intervals. As outlined in figure 8.3 the deviations increase above 650 nm. In general the comparison with the Voigt et al. data seems more problematic. Applying the same smoothing procedure, as before to the ratio between GOME-2 and Vandaele, confirms this first impression. In figure 8.10 one can see remaining structures in the ratio together with a baseline. These results are as well in high consistency with investigations made in [69], in particular the jump and strong feature at 500 nm.

1,6

Ratio GOME-2 FM2 / Voigt et al.

1,5

Ratio GOME-2 FM2-1 / Voigt et al.

1,4

Ratio GOME-2 FM3 / Voigt et al.

1,3

Ratio GOME-2 / Voigt et al.

1,2

1,1

1,0

0,9

Ratio

0,8

0,7

0,6

0,5

0,4

0,3

0,2

300 350 400 450 500 550 600 650 700 750 800

Wavelength (nm)

Figure 8.8: Ratio between GOME-2 data and Voigt et al. 8.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 129

1,30

Ratio GOME-2 FM2 / Voigt et al.

1,25

Ratio GOME-2 FM2-1 / Voigt et al.

Ratio GOME-2 FM3 / Voigt et al.

1,20

Ratio GOME-2 / Voigt et al.

1,15

1,10

1,05

Ratio

1,00

0,95

0,90

0,85

0,80

300 350 400 450 500 550 600

Wavelength (nm)

Figure 8.9: Ratio between GOME-2 data and Voigt et al.

1,20 1,15 Ratio GOME-2 FM3 / Voigt et al. 1,10 "Smoothing" procedure (adjacent averaging) 1,05 1,00

Ratio 0,95 0,90 0,85 0,80 300 350 400 450 500 550 600 650 1,20 GOME-2 FM2 1,15 GOME-2 FM2-1 1,10 GOME-2 FM3 GOME-2 1,05

Ratio 1,00

0,95

0,90 300 350 400 450 500 550 600 650 Wavelength (nm)

Figure 8.10: ”Smoothing” of the ratio between GOME-2 and Voigt et al. 130 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

8.3.2 Temperature dependence of NO2 spectra obtained from the GOME-2 study This section contains the description of the results concerning the temperature dependence of the NO2 spectra obtained from the GOME-2 study. As mentioned above a linear temperature behavior between 293 and 223 K is well established in the literature. Figure 8.11 shows the NO2 absorption spectrum recorded with GOME-2 FM3 at four different temperatures. As described in 8.2 all spectra at all temperatures were scaled to absolute cross sections through integrated cross sections in the 400 to 500 nm region. Figures 8.12 and 8.13 show a more detailed picture in this range, which is the main DOAS window. In 8.13 we can see the curves at different temperatures at arbitrary progressions in the 400 to 500 nm range, which confirms a consistent temperature dependence, i.e. an increase of the differential structure with decreasing temperature.

The linearity of the temperature dependence is illustrated in ﬁgures 8.14 to 8.17. The absorption cross section at two arbitrary peaks ( 428 and 431 nm) in the DOAS range between 400 and 500 nm have been plotted at diﬀerent≈ temperatures.

(NO ) @ T293K.FM3

(NO ) @ T273K.FM3

1E-18

(NO ) @ T243K.FM3

2 ) 2

(NO ) @ T223K.FM3 m

2 c ( 1E-19

Main DOAS Window

1E-20 Cross-Section

1E-21

1E-22

300 400 500 600 700 800

Wavelength (nm)

Figure 8.11: NO2 spectra at diﬀerent temperatures 8.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 131

1E-18

(NO ) @ T293K.FM3 9E-19

8E-19 (NO ) @ T273K.FM3

(NO ) @ T243K.FM3 7E-19

2 ) 2

(NO ) @ T223K.FM3

2 6E-19 m c (

5E-19

4E-19 Cross-Section

3E-19

2E-19

400 410 420 430 440 450 460 470 480 490 500

Wavelength (nm)

Figure 8.12: NO2 spectra in the main DOAS window (Zoom I)

8E-19 (NO ) @ T293K.FM3

(NO ) @ T273K.FM3

(NO ) @ T243K.FM3

7E-19 2 ) 2

(NO ) @ T223K.FM3

m 2 c (

6E-19

5E-19 Cross-Section

4E-19

425 426 427 428 429 430 431 432 433 434 435

Wavelength (nm)

Figure 8.13: NO2 spectra in the main DOAS window (Zoom II) 132 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

6,40E-019

(T) @ 428nm (FM2)

(T) @ 431nm (FM2)

6,35E-019

Linear Fit of (T) @ 428nm

Linear Fit of (T) @ 431nm

6,30E-019 ) 2 m

6,25E-019 c (

6,20E-019

6,15E-019

Y = A + B * X

6,10E-019

428.0nm (T) 431.0nm Cross Section

A 7,24654E-19 7,37191E-19

6,05E-019

B -4,13151E-22 -4,53289E-22

SD 1,20095E-22 2,01106E-22 6,00E-019

223 233 243 253 263 273 283 293

Temperature (K)

Figure 8.14: Temperature dependence of the NO2 absorption cross section at two arbitrary wavelengths in the main DOAS window, recorded with GOME-2 FM2

6,40E-019 (T) @ 428nm (FM2-1)

(T) @ 431nm (FM2-1)

6,35E-019

Linear Fit of (T) @ 428nm

Linear Fit of (T) @ 431nm

6,30E-019 ) 2

6,25E-019

6,20E-019

6,15E-019

Y = A + B * X Cross Section (cm 428.0nm (T) 431.0nm 6,10E-019

A 7,15958E-19 7,27745E-19

6,05E-019 B -3,62500E-22 -4,09776E-22

SD 1,11453E-22 7,49376E-22

6,00E-019

223 233 243 253 263 273 283 293

Temperature (K)

Figure 8.15: Temperature dependence of the NO2 absorption cross section at two arbitrary wavelengths in the main DOAS window, recorded with GOME-2 FM2-1 8.3. RESULTS AND COMPARISON WITH LITERATURE DATA BASE 133

6,40E-019

(T) @ 428nm (FM3)

(T) @ 431nm (FM3)

6,35E-019

Linear Fit of (T) @ 428nm

Linear Fit of (T) @ 431nm

6,30E-019 ) 2

6,25E-019

6,20E-019

Y = A + B * X

6,15E-019

428.0nm (T) 431.0nm Cross Section (cm

A 7,08547E-19 7,27908E-19

6,10E-019

B -3,42353E-22 -4,05904E-22

SD 5,17591E-22 5,2018E-22

6,05E-019

223 243 253 263 273 283 293

Temperature (K)

Figure 8.16: Temperature dependence of the NO2 absorption cross section at two arbitrary wavelengths in the main DOAS window, recorded with GOME-2 FM3

6,40E-019 (T) @ 428nm (GOME-2)

(T) @ 431nm (GOME-2)

6,35E-019 Linear Fit of F(T) @ 428nm

Linear Fit of F(T) @ 431nm

6,30E-019 ) 2

6,25E-019

6,20E-019

6,15E-019

Y = A + B * X

428.0nm (T) 431.0nm Cross Section (cm

6,10E-019

A 7,14748E-19 7,28972E-19

B -3,70226E-22 -4,17193E-22

6,05E-019

SD 8,72859E-22 1,67417E-22

6,00E-019

223 243 253 263 273 283 293

Temperature (K)

Figure 8.17: Temperature dependence of the ”ﬁnal” GOME-2 NO2 absorption cross section at two arbitrary wavelengths in the main DOAS window 134 CHAPTER 8. NO2-ABSORPTION MEASUREMENTS WITH GOME-2

In all cases the behavior could be reproduced with a linear interpolation. It can therefore be concluded that the GOME-2 NO2 spectra are in high consistency with recommended spectra in the literature, in particular in the important region between 400 and 500 nm. The temperature dependence is in accordance with well established previous observations, i.e. a linear dependence between 293 K and 223 K. It is therefore highly recommendable to use the GOME-2 spectra (of the certain FM) for data retrieval and analysis of remote sensing applications with GOME-2. Chapter 9

Outlook

Before concluding the results of the GOME-2 study an outlook is given in this chapter with respect to possible improvements for future CATGAS measurements.

Baseline drift • Baseline drift was in general clearly smaller than 2 %. But in a number of cases, especially when long purging and measurement periods lay between the first and last reference measurement, it exceeded 2 %. This is likely to be caused by insufficient correction of long term drift by the near-real time light source monitoring. Usually three pairs of vessel and direct measurements were recorded for this correction. By increasing the number of pairs the quality of the correction could be improved as this would be a better approximation to the aimed at near real time measurement. As a draw back this would increase the overall measurement time because the minimum time per pair is determined by the longest PET across the four channels plus overhead time for setting the flip mirror.

Unidentiﬁed absorption in the O absorption minimum region • 3

The determination of the O3 absorption spectrum in the absorption minimum and especially in the neighboring red end of the Huggins band and the blue end of the Chappuis band is still problematic. The observed behavior of absorption was irregular and was clearly not caused by the O3 absorption. Technical effects like mechanical drift of the mirror mounts or the vessel windows can be excluded after the different efforts made during the study. Two causes appear possible for the observed effect: Firstly an interaction between O3 (the effect is strongest at high concentrations of O3) and the surface of the mirrors. And secondly a gas phase absorber, which could be an adduct of O3 and O2 or possibly an O3 dimer. But there is experimental evidence, that the observed effect occurs and disappears with a certain temporal lag relative to the filling and purging of the vessel. This makes an interaction with the mirror surface more likely. Especially an interaction between the at the used pressures and temperatures always present molecular layers of H2O with O3 could change the reflectivity of the mirrors slightly and spectrally broad banded. Different approaches can be thought of to improve this.

1. Placing the multipath optics outside of the vessel would inhibit interaction of the surface layer (MgF2 and H2O layers) with O3. But if the same eﬀect takes place

135 136 CHAPTER 9. OUTLOOK

between H2O layers on the quartz windows and ozone, the eﬀect would thereby not be avoided. In addition the multiple transition of the analysis beam through the vessel windows would signiﬁcantly reduce the optical throughput leading to strongly increased exposure times. This is critical with respect to the overall available measurement time per campaign.

2. Purging the mirror surfaces with an inert gas, which in the context of O3 measurements could be O2 or N2, could avoid interaction of O3 with the mirror surfaces. This necessitates a new design of the mirror mounts and reconsideration of the ﬂow conditions of the diﬀerent mixtures during the experiment to guarantee stable mixing conditions during measurements. Apart from the later this appears to be the most promising technical approach to overcome the observed problems. Here techniques established in the context of cavity ring down spectroscopy could prove helpful.

3. An analytical approach could use extraction and separation techniques developed in the context of iodine oxides spectroscopy. There concepts of independent component analysis and least squares techniques are being used to separate overlapping absorptions caused by diﬀerent absorbers [92], [93], [94]. By applying such methods to the already available data obtained in this study the observed unknown absorption could possibly be separated from the ozone spectrum. If the approach could be validated with the present data, this would avoid technical changes to the set-up in future application. Nevertheless a combination of this analytical method and the purging of mirrors appears advisable. Chapter 10

Conclusion

In this study absorption cross section spectra of O3 and NO2 have been measured in three independent campaigns using the three highly stabilized and accurately characterized GOME-2 (Global Ozone Monitoring Experiment 2) satellite spectrometers, ﬂight models FM2, FM2-1, and FM3. The spectra for O3 were recorded at ﬁve temperatures, i.e. 203 K, 223 K, 243 K, 273 K, and 293 K and with a spectral coverage of 240 to 790 nm at a resolution of 0.24 to 0.53 nm full width at half maximum. NO2 was measured at four temperatures between 223 and 293 K.

The Huggins bands and the Chappuis bands of the O3 absorption spectrum were covered simultaneously by these measurements. The relative temperature dependence of the O3 spectrum was determined. For all three campaigns and at the selected wavelengths of 253.65 nm, 289.36 nm, 296.73 nm, 302.15 nm, and 334.15 nm it agrees accurately with the available literature data. For the temperature dependence of the O3 cross section in the range of 400 to 450 nm an upper limit estimate of no more than 10 % decrease with falling temperature was found at 425 to 430 nm in disagreement with one previous publication. There is some evidence that continuous absorption as measured in the absorption minimum at 429.5 nm drops by a few percent when temperature is reduced from 293 K to 203 K. At the same time the peak of the band at 426.5 nm increases slightly by 1 %. This is supported by a clear increase of amplitude of diﬀerential absorption cross section≈ at 426 nm of 13 % with falling temperature. At 604.61 nm at the peak of the Chappuis band a slight increase of cross section with falling temperature of 1 % is found in agreement with three previous publications and in disagreement with one≈ most recent one. The integrated absorption cross section of the Chappuis region was found to be independent of temperature within less than 1 %. From simultaneously measured spectra covering the Huggins and Chappuis bands the± ratios of integrated absorption cross sections were determined between the diﬀerent regions providing a robust means for intercomparison of cross section in these regions.

The recorded NO2 spectra are highly consistent with previous measurements and show an agreement within 2 % in the main DOAS window between 400 and 500 nm for all FM’s. Also the linear temperature dependence is in high consistence with previous literature data.

137

Appendix A

Quality-Analysis Overlap Region

As described in chapter 5.3.1 the following tables give information about the ratio between two mixtures (i.e. two diﬀerent absorption measurements), the corresponding channel and wavelength range and the mean value with standard deviation.

GOME-2 FM2-1 T293K • Ratio Channel Wavelength Range Pixel Mean SD [nm] [%] ratio m8m7 FPA-1 285 - 292 [680:745] 1.00123 0.12514 ratio m7m6 FPA-1 297 - 304 [790:855] 1.00412 0.29201 ratio m6m3f FPA-2 309 - 321 [ 90:190] 1.00762 0.54939 ratio m3fm3e FPA-2 318 - 323 [170:205] 1.00242 0.22516 ratio m3em3d FPA-2 321 - 333 [190:285] 1.001 0.26363 ratio m3dm3b FPA-2 329 - 341 [255:355] 1.00143 0.31525 ratio m3bm2 FPA-2 338 - 345 [335:390] 0.99009 0.97547 ratio m2m1 FPA-2 348 - 354 [420:465] 0.99964 0.4582 ratio m1m2 FPA-3 459 - 479 [320:415] 1.00068 0.12555 ratio m2m3b FPA-3 476 - 501 [400:520] 0.99643 0.16781 Ø 1.00047 0.34976

139 140 APPENDIX A. QUALITY-ANALYSIS OVERLAP REGION

GOME-2 FM2-1 T273K • Ratio Channel Wavelength Range Pixel Mean SD [nm] [%] ratio m8m7 FPA-1 275 - 288 [590:700] 1.00387 0.12509 ratio m7m6 FPA-1 292 - 300 [740:815] 1.00565 0.26512 ratio m6m3f FPA-2 309 - 316 [ 95:145] 1.00361 0.86722 ratio m3fm3e FPA-2 317 - 324 [160:210] 0.99791 0.44255 ratio m3em3d FPA-2 321 - 328 [190:250] 0.99245 0.62326 ratio m3dm3c FPA-2 326 - 338 [230:335] 0.99637 0.4611 ratio m3cm3b FPA-2 329 - 340 [255:355] 1.00003 0.34033 ratio m3bm3a FPA-2 331 - 345 [275:390] 0.99458 0.75203 ratio m3am2 FPA-2 338 - 346 [335:395] 0.99671 1.06203 ratio m2m1 FPA-2 348 - 353 [415:460] 0.99821 0.79885 ratio m1m2 FPA-3 438 - 471 [220:375] 1.00078 0.09559 ratio m3am2 FPA-3 477 - 514 [405:585] 0.99435 0.37126 Ø 0.99871 0.51704

GOME-2 FM2-1 T243K • Ratio Channel Wavelength Range Pixel Mean SD [nm] [%] ratio m8m7 FPA-1 281nm - 293nm [635:745] 1.00552 0.28303 ratio m7m6 FPA-1 294nm - 299nm [760:800] 1.00221 0.241 ratio m6m3h FPA-2 309nm - 315nm [ 95:140] 1.00424 0.455 ratio m3hm3f FPA-2 318nm - 324nm [165:210] 1.00097 0.21032 ratio m3fm3g FPA-2 319nm - 332nm [170:280] 0.99983 0.22322 ratio m3gm3e FPA-2 324nm - 335nm [210:305] 1.00285 0.19802 ratio m3em3b FPA-2 328nm - 338nm [250:330] 0.99999 0.46095 ratio m3bm3c FPA-2 329nm - 341nm [255:355] 1.00301 0.55307 ratio m3cm2 FPA-2 338nm - 345nm [335:390] 0.99762 1.71431 ratio m2m1 FPA-2 348nm - 353nm [415:460] 1.01527 1.23934 ratio m1m2 FPA-3 459nm - 476nm [320:400] 1.00125 0.09167 ratio m2m3c FPA-3 493nm - 519nm [480:610] 1.00483 0.45083 Ø 1.00313 0.50032 141

GOME-2 FM2-1 T223K • Ratio Channel Wavelength Range Pixel Mean SD [nm] [%] ratio m8m7a FPA-1 278 - 290 [610:725] 1.00698 0.38611 ratio m7am6 FPA-1 294 - 300 [760:815] 1.00226 0.14066 ratio m6m3f FPA-2 309 - 315 [ 95:140] 1.00457 0.59446 ratio m3fm3h FPA-2 310 - 319 [100:170] 1.00272 0.1721 ratio m3hm3g FPA-2 313 - 324 [125:210] 0.99962 0.29843 ratio m3gm3e FPA-2 316 - 329 [150:255] 1.00368 0.12971 ratio m3em3d FPA-2 323 - 332 [205:280] 1.00583 0.32893 ratio m3dm3c FPA-2 326 - 335 [230:305] 1.00668 0.32032 ratio m3cm3a FPA-2 329 - 338 [250:335] 1.00394 0.2418 ratio m3am3b FPA-2 331 - 341 [275:355] 1.00466 0.42356 ratio m3bm2 FPA-2 338 - 345 [335:390] 0.98812 2.47101 ratio m2m1 FPA-2 348 - 353 [415:460] 1.01481 1.5381 ratio m1m2 FPA-3 432 - 471 [235:375] 1.00051 0.11704 ratio m2m3b FPA-3 482 - 514 [430:585] 1.00059 0.08188 Ø 1.00321 0.51744

GOME-2 FM2-1 T203K • Ratio Channel Wavelength Range Pixel Mean SD [nm] [%] ratio m8m7 FPA-1 287 - 292 [690:740] 1.00397 0.24017 ratio m7m6 FPA-1 294 - 303 [760:840] 0.99979 0.05617 ratio m6m3h FPA-2 311 - 317 [110:160] 1.00585 0.81213 ratio m3hm3f FPA-2 315 - 324 [140:215] 1.00496 0.23021 ratio m3fm3e FPA-2 319 - 332 [170:280] 1.00514 0.30995 ratio m3em3d FPA-2 326 - 335 [230:305] 1.00682 0.37984 ratio m3dm3c FPA-2 328 - 338 [250:335] 1.00341 0.37986 ratio m3cm3a FPA-2 329 - 341 [255:355] 1.00992 0.6793 ratio m3am1 FPA-2 345 - 348 [390:415] 1.0937 4.94736 ratio m1m3a FPA-3 459 - 473 [320:385] 1.00016 0.61048 Ø 1.01337 0.86455 142 APPENDIX A. QUALITY-ANALYSIS OVERLAP REGION

GOME-2 FM3 T293K •

Ratio Channel Wavelength Range Pixel Mean SD SD [nm] [%] ratio m8m7 FPA-1 285 - 294 [680:760] 1.0012 0.00222 0.22145 ratio m7m6 FPA-1 298 - 303 [795:850] 0.99997 0.00336 0.33614 ratio m6m3h FPA-2 311 - 315 [165:200] 0.99852 0.00302 0.30253 ratio m3gm3h FPA-2 315 - 326 [200:290] 0.99973 0.00236 0.23603 ratio m3fm3g FPA-2 318 - 329 [225:315] 1.00078 0.0019 0.19015 ratio m3em3f FPA-2 323 - 335 [265:360] 0.99993 0.00197 0.19665 ratio m3dm3e FPA-2 325 - 338 [285:390] 1.00118 0.00252 0.25142 ratio m3cm3d FPA-2 328 - 340 [310:410] 1.00087 0.00229 0.22907 ratio m3bm3c FPA-2 328 - 341 [310:415] 1.00095 0.00202 0.2021 ratio m3am3b FPA-2 331 - 344 [335:445] 0.99982 0.00208 0.20816 ratio m3am2 FPA-2 340 - 345 [410:450] 1.00114 0.00412 0.4118 ratio m2m1 FPA-2 353 - 354 [515:525] 0.97945 0.00307 0.31308 ratio m1m2 FPA-3 439 - 450 [225:280] 0.99938 0.00478 0.47811 ratio m2m3 FPA-3 477 - 502 [405:525] 1.00006 0.00348 0.34791 Ø 0.99878 0.0028 0.28033

GOME-2 FM3 T273K •

Ratio Channel Wavelength Range Pixel Mean SD SD [nm] [%] ratio m8.2m6 FPA-1 290 - 300 [715:815] 1.00155 0.00313 0.31288 ratio m6m3h FPA-2 312 - 320 [175:240] 0.99856 0.00501 0.50213 ratio m3hm3g FPA-2 318 - 329 [225:315] 0.99959 0.00201 0.2012 ratio m3gm3e FPA-2 325 - 334 [285:360] 0.99919 0.00269 0.26964 ratio m3em3b FPA-2 328 - 338 [310:395] 1.00136 0.00285 0.285 ratio m3bm2 FPA-2 338 - 345 [395:450] 0.9987 0.00633 0.6338 ratio m2m1 FPA-2 352 - 354 [510:525] 0.98692 0.00527 0.53354 ratio m1m2 FPA-3 437 - 453 [215:290] 1.00411 0.00128 0.12794 ratio m2m3b FPA-3 476 - 493 [400:485] 0.96841 0.00829 0.85632 Ø 0,99584 0,00374 0,37759 143

GOME-2 FM3 T243K •

Ratio Channel Wavelength Range Pixel Mean SD SD [nm] [%] ratio m8m7 FPA-1 287nm - 293nm [700:755] 0.99834 0.00241 0.24182 ratio m7m6 FPA-1 300nm - 304nm [815:860] 0.99909 0.00204 0.20462 ratio m6m3e FPA-2 312nm - 317nm [180:215] 1.00147 0.00206 0.20578 ratio m3em3c FPA-2 319nm - 329nm [230:315] 1.00095 0.00217 0.21657 ratio m3bm3c FPA-2 323nm - 334nm [265:360] 1.00045 0.00342 0.3414 ratio m3am3b FPA-2 325nm - 335nm [285:365] 0.99955 0.00222 0.22214 ratio m3am3d FPA-2 326nm - 338nm [290:390] 1.001 0.00255 0.25512 ratio m3dm2 FPA-2 337nm - 340nm [385:410] 0.99411 0.00963 0.96852 ratio m2m1 FPA-2 348nm - 352nm [475:510] 1.03082 0.01543 1.49683 ratio m1m2* FPA-3 459nm - 476nm [320:400] 1.00123 0.00132 0.13166 ratio m2m3d FPA-3 480nm - 521nm [420:620] 1.00002 0.00187 0.18656 Ø 1,00388 0,0052 0,5146

GOME-2 FM3 T223K •

Ratio Channel Wavelength Range Pixel Mean SD SD [nm] [%] ratio m8m7a FPA-1 287 - 293 [700:755] 1.00017 0.00241 0.24079 ratio m7am6 FPA-1 296 - 305 [780:865] 1.0004 0.00165 0.16523 ratio m6m3h FPA-2 313 - 317 [185:220] 1.00119 0.00354 0.35361 ratio m3hm3g FPA-2 320 - 331 [240:335] 1.00228 0.00214 0.21354 ratio m3gm3f FPA-2 323 - 334 [265:360] 1.00243 0.00242 0.24131 ratio m3fm3e FPA-2 325 - 338 [285:390] 1.0077 0.00566 0.56155 ratio m3em3b FPA-2 328 - 338 [310:390] 0.98317 0.01891 1.92294 ratio m3bm3a FPA-2 331 - 345 [335:450] 1.01031 0.00703 0.69573 ratio m3am2 FPA-2 337 - 345 [385:450] 1.16138 0.08974 7.72718 ratio m2m1 FPA-2 347 - 352 [470:510] 1.03299 0.0237 2.29461 ratio m1m2 FPA-3 439 - 457 [225:310] 0.99826 0.00387 0.38806 ratio m2m3a FPA-3 527 - 558 [650:795] 1.00039 4.49623E-4 0.04494 Ø 1,01672 0,01346 1,23746 144 APPENDIX A. QUALITY-ANALYSIS OVERLAP REGION

GOME-2 FM3 T203K •

Ratio Channel Wavelength Range Pixel Mean SD SD [nm] [%] ratio m8m7 FPA-1 287 - 293 [700:755] 1.00726 0.00365 0.36237 ratio m7m6 FPA-1 297 - 304 [790:860] 1.00135 0.00151 0.15089 ratio m3im6 FPA-2 311 - 315 [170:200] 0.99986 0.00388 0.38851 ratio m3gm3i FPA-2 320 - 326 [240:290] 0.9949 0.00509 0.51167 ratio m3bm3g FPA-2 325 - 334 [285:360] 0.98833 0.0103 1.04185 ratio m3am3b FPA-2 328 - 338 [310:390] 1.00674 0.00683 0.67834 ratio m3cm3a FPA-2 328 - 341 [310:415] 0.994 0.00535 0.53852 ratio m2m3c FPA-2 338 - 341 [390:415] 1.01644 0.03105 3.05509 ratio m1m2 FPA-2 347 - 352 [470:510] 0.97037 0.02312 2.383 ratio m1m2 FPA-3 449 - 457 [275:310] 0.99958 0.00124 0.12371 ratio m2m3c FPA-3 503 - 517 [535:600] 1.00099 0.00134 0.13392 Ø 0,99816 0,00849 0,85162 Appendix B

I0 - I Documentation

Intensities in b.u. for measurements with GOME-2 FM2-1 at 273K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 17000 - 20000 5000 - 19000 1 3 400 - 455 3000 - 15000 2500 - 6000 2 2 340 - 360 20000 5000 - 18000 2 3 430 - 490 8000 - 25000 7000 - 15000 3 2 310 - 345 6000 - 20000 2500 - 18000 3 3 475 - 600 15000 - 24000 14000 - 5000 3 4 590 - 790 19000 - 4000 10000 - 2000 6 1 298 - 310 50000 - 15000 25000 - 10000 Hg(302) 50000 25000 6 2 310 - 315 8000 - 25000 7000 - 22000 7 1 285 - 303 30000 - 50000 12000 - 42000 Hg(289) 32000 20000 Hg(296) 45000 35000 8 1 240 - 290 3000 - 35000 2000 - 30000 Hg(253) 5200 2700

145 146 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 2 2

1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 16000 14000

1.5 12000 10000 1 8000

Intensity [b.u.] Intensity [b.u.] 6000 0.5 4000 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.1: Intensities regarding mixture 1 measured with GOME-2 FM2-1 @ 273K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 2 2

1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 16000 14000

1.5 12000 10000

1 8000

Intensity [b.u.] Intensity [b.u.] 6000 0.5 4000 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.2: Intensities regarding mixture 2 measured with GOME-2 FM2-1 @ 273K. The lower graphics show the corresponding relevant wavelength range 147

4 4 x 10 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 2

2 1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 2

2 1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.3: Intensities regarding mixture 3 measured with GOME-2 FM2-1 @ 273K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3

4 x 10 Mixture 3 FPA−4

1.8

1.6

1.4

1.2

1 Intensity [b.u.]

0.8

0.6

0.4

0.2 600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.4: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM2-1 @ 273K. 148 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 5 3.5

3 4 2.5

3 2

2 1.5 Intensity [b.u.] Intensity [b.u.] 1 1 0.5

220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 5 2.5

4 2

3 1.5 ← Hg−Line @ 302nm 2 Intensity [b.u.] Intensity [b.u.] 1

1 0.5

300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.5: Intensities regarding mixture 6 measured with GOME-2 FM2-1 @ 273K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 7 FPA−1 x 10 Mixture 8 FPA−1 5 5

4 4

3 3

2 2 Intensity [b.u.] Intensity [b.u.]

1 1

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

4 x 10 5 10000 4 ← Hg−Line 8000 3 @ 296nm 6000 2 ← Hg−Line @ 289nm Intensity [b.u.] Intensity [b.u.] 4000 ← Hg−Line 1 2000 @ 253nm

285 290 295 300 250 260 270 280 wavelength [nm] wavelength [nm]

Figure B.6: Intensities regarding mixture 7 and 8 measured with GOME-2 FM2-1 @ 273K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 149

Intensities in b.u. for measurements with GOME-2 FM2-1 at 243K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 15000 - 18000 5000 - 17000 1 3 400 - 455 3000 - 15000 3000 - 6000 2 2 340 - 360 20000 5000 - 17000 2 3 430 - 490 9000 - 20000 8000 - 15000 3 2 310 - 345 7000 - 20000 6000 - 19000 3 3 475 - 600 16000 - 27000 15000 - 5000 3 4 590 - 790 22000 - 4000 11000 - 2500 6 1 298 - 310 50000 - 15000 10000 - 20000 Hg(302) 50000 20000 6 2 310 - 315 8000 - 24000 6000 - 21000 7 1 285 - 303 30000 - 49000 11000 - 42000 Hg(289) 32000 19000 Hg(296) 42000 32000 8 1 240 - 290 2500 - 33000 2000 - 30000 Hg(253) 5000 2500

4 4 x 10 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 2 2.5

1.5 2

1.5 1

Intensity [b.u.] Intensity [b.u.] 1

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2

15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

0.5 5000

350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.7: Intensities regarding mixture 1 measured with GOME-2 FM2-1 @ 243K. The lower graphics show the corresponding relevant wavelength range 150 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 2 2.5

1.5 2

1.5 1

Intensity [b.u.] Intensity [b.u.] 1

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2

15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

0.5 5000

340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.8: Intensities regarding mixture 2 measured with GOME-2 FM2-1 @ 243K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 2 2.5

1.5 2

1.5 1

Intensity [b.u.] Intensity [b.u.] 1

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 2 2.5

1.5 2

1.5 1

Intensity [b.u.] Intensity [b.u.] 1

0.5 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.9: Intensities regarding mixture 3 measured with GOME-2 FM2-1 @ 243K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 151

4 x 10 Mixture 3 FPA−4 2.2

1.8

1.6

1.4

1.2 Intensity [b.u.] 1

0.8

0.6

0.4

0.2 600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.10: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM2-1 @ 243K.

4 4 x 10 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 5

3 4 2.5

3 2

2 1.5 Intensity [b.u.] Intensity [b.u.] 1 1 0.5

220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 5 2.5

4 2

3 1.5

2 ← Hg−Line @ 302nm

Intensity [b.u.] Intensity [b.u.] 1

1 0.5

300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.11: Intensities regarding mixture 6 measured with GOME-2 FM2-1 @ 243K. The lower graphics show the corresponding relevant wavelength range 152 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 7 FPA−1 x 10 Mixture 8 FPA−1

4 4

3 3

2 2 Intensity [b.u.] Intensity [b.u.]

1 1

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

4 x 10

10000 4

← Hg−Line 8000 3 @ 296nm 6000 2 ← Intensity [b.u.] Hg−Line @ 289nm Intensity [b.u.] 4000 ← 1 2000 Hg−Line @ 253nm

285 290 295 300 240 250 260 270 wavelength [nm] wavelength [nm]

Figure B.12: Intensities regarding mixture 7 and 8 measured with GOME-2 FM2-1 @ 243K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 153

Intensities in b.u. for measurements with GOME-2 FM2-1 at 223K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 15000 - 18000 3000 - 17000 1 3 400 - 455 3000 - 12000 3000 - 5000 2 2 340 - 360 17000 4000 - 16000 2 3 430 - 490 8000 - 15000 7000 - 10000 3 2 310 - 345 6000 - 20000 5000 - 18000 3 3 475 - 600 14000 - 21000 12000 - 5000 3 4 590 - 790 16000 - 4000 8000 - 2000 6 1 298 - 310 45000 - 12000 25000 - 10000 Hg(302) 45000 25000 6 2 310 - 315 8000 - 23000 6000 - 21000 7 1 285 - 303 28000 - 45000 10000 - 40000 Hg(289) 30000 18000 Hg(296) 41000 31000 8 1 240 - 290 2500 - 31000 2000 - 28000 Hg(253) 5000 2500

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 2 15000

1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

5000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

14000

15000 12000

10000

10000 8000

Intensity [b.u.] Intensity [b.u.] 6000

5000 4000

2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.13: Intensities regarding mixture 1 measured with GOME-2 FM2-1 @ 223K. The lower graphics show the corresponding relevant wavelength range 154 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 2 15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

5000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

14000

15000 12000

10000

10000 8000

Intensity [b.u.] Intensity [b.u.] 6000

5000 4000

2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.14: Intensities regarding mixture 2 measured with GOME-2 FM2-1 @ 223K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 2 2

1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 2 2

1.5 1.5

1 1 Intensity [b.u.] Intensity [b.u.]

0.5 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.15: Intensities regarding mixture 3 measured with GOME-2 FM2-1 @ 223K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 155

Mixture 3 FPA−4

16000

14000

12000

10000

Intensity [b.u.] 8000

6000

4000

2000 600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.16: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM2-1 @ 223K.

4 4 x 10 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2

3 4 2.5

3 2

2 1.5 Intensity [b.u.] Intensity [b.u.] 1 1 0.5

220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 2.5

4 2

3 1.5 ← Hg−Line @ 302nm 2

Intensity [b.u.] Intensity [b.u.] 1

1 0.5

300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.17: Intensities regarding mixture 6 measured with GOME-2 FM2-1 @ 223K. The lower graphics show the corresponding relevant wavelength range 156 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 7 FPA−1 x 10 Mixture 8 FPA−1

4 4

3 3

2 2 Intensity [b.u.] Intensity [b.u.]

1 1

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

4 x 10 7000

4 6000

← 5000 3 Hg−Line @ 296nm 4000 2 ← 3000 Intensity [b.u.] Hg−Line @ 289nm Intensity [b.u.] ← Hg−Line 1 2000 @ 253nm

1000 285 290 295 300 245 250 255 260 wavelength [nm] wavelength [nm]

Figure B.18: Intensities regarding mixture 7 and 8 measured with GOME-2 FM2-1 @ 223K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 157

Intensities in b.u. for measurements with GOME-2 FM2-1 at 203K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 15000 - 18000 3000 - 17000 1 3 400 - 455 3000 - 12000 3000 - 5000 2 2 340 - 360 16000 10000 - 15000 2 3 430 - 490 7000 - 15000 7000 - 12000 3 2 310 - 345 5000 - 17000 2000 - 15000 3 3 475 - 600 12000 - 20000 10000 - 3000 3 4 590 - 790 15000 - 4000 2000 - 7000 6 1 298 - 310 31000 - 8000 15000 - 7000 Hg(302) 31000 15000 6 2 310 - 315 6000 - 18000 5000 - 16000 7 1 285 - 303 18000 - 31000 5000 - 25000 Hg(289) 20000 8000 Hg(296) 27000 19000 8 1 240 - 290 2000 - 22000 1800 - 18000 Hg(253) 3200 2000

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3

16000 14000 1.5 12000 10000 1 8000 Intensity [b.u.] Intensity [b.u.] 6000 0.5 4000 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

16000 12000 14000 10000 12000 10000 8000

8000 6000 Intensity [b.u.] Intensity [b.u.] 6000 4000 4000 2000 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.19: Intensities regarding mixture 1 measured with GOME-2 FM2-1 @ 203K. The lower graphics show the corresponding relevant wavelength range 158 APPENDIX B. I0 - I DOCUMENTATION

Mixture 2 FPA−2 Mixture 2 FPA−3

16000 14000 15000 12000 10000 10000 8000 Intensity [b.u.] Intensity [b.u.] 6000 5000 4000 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

16000 12000

14000 10000 12000 8000 10000 8000 6000 Intensity [b.u.] Intensity [b.u.] 6000 4000 4000 2000 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.20: Intensities regarding mixture 2 measured with GOME-2 FM2-1 @ 203K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3

15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

5000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10

15000 1.5

10000 1 Intensity [b.u.] Intensity [b.u.]

5000 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.21: Intensities regarding mixture 3 measured with GOME-2 FM2-1 @ 203K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 159

Mixture 3 FPA−4

14000

12000

10000

8000 Intensity [b.u.]

6000

4000

2000

600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.22: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM2-1 @ 203K.

4 4 x 10 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 3 2.5

2.5 2 2 1.5 1.5

Intensity [b.u.] Intensity [b.u.] 1 1

0.5 0.5

220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 3 2

2.5 1.5 2

1.5 ← Hg−Line @ 302nm 1 Intensity [b.u.] Intensity [b.u.] 1 0.5 0.5

300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.23: Intensities regarding mixture 6 measured with GOME-2 FM2-1 @ 203K. The lower graphics show the corresponding relevant wavelength range 160 APPENDIX B. I0 - I DOCUMENTATION

4 4 x 10 Mixture 7 FPA−1 x 10 Mixture 8 FPA−1

3 3

2.5 2.5

2 2

1.5 1.5 Intensity [b.u.] Intensity [b.u.] 1 1

0.5 0.5

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

4 x 10

3 6000

2.5 5000 2 4000 ← Hg−Line 1.5 @ 296nm 3000 Intensity [b.u.] Intensity [b.u.] 1 ← Hg−Line @ 289nm 2000 ← Hg−Line 0.5 @ 253nm 1000 285 290 295 300 250 260 270 280 wavelength [nm] wavelength [nm]

Figure B.24: Intensities regarding mixture 7 and 8 measured with GOME-2 FM2-1 @ 203K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 161

Intensities in b.u. for measurements with GOME-2 FM3 at 293K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 7000 - 9000 2500 - 9000 1 3 400 - 455 2500 - 20000 2500 - 8000 2 2 340 - 360 7000 3000 - 6000 2 3 430 - 490 10000 - 43000 10000 - 25000 3 2 310 - 345 2000 - 8000 1800(low c) - 7000(high c) 3 3 475 - 600 20000 - 38000 20000 3 4 590 - 790 22000 - 35000 10000 - 20000 6 1 298 - 310 13000 - 8000 11000 - 6000 Hg(302) 12770 10140 6 2 310 - 315 2400 - 15000 2200 - 10000 7 1 285 - 303 8000 - 14000 3500 - 12000 Hg(289) 10060 5550 Hg(296) 13070 10045 8 1 240 - 290 1800 - 12000 1600 - 9000 Hg(253) 2560 1780

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 9000 4 8000 3.5 7000 3 6000 2.5 5000 2

Intensity [b.u.] 4000 Intensity [b.u.] 1.5 3000 1 2000 0.5 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 9000 8000 2 7000 1.5 6000 5000 1

Intensity [b.u.] 4000 Intensity [b.u.] 3000 0.5 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.25: Intensities regarding mixture 1 measured with GOME-2 FM3 @ 293K. The lower graphics show the corresponding relevant wavelength range 162 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 9000 4 8000 3.5 7000 3 6000 2.5 5000 2

Intensity [b.u.] 4000 Intensity [b.u.] 1.5 3000 1 2000 0.5 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 9000 2.5 8000 2 7000 6000 1.5 5000

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 0.5 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.26: Intensities regarding mixture 2 measured with GOME-2 FM3 @ 293K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 10000 3.5

8000 3 2.5 6000 2 1.5 Intensity [b.u.] Intensity [b.u.] 4000 1

2000 0.5 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 10000 3.5

8000 3 2.5 6000 2 1.5 Intensity [b.u.] Intensity [b.u.] 4000 1

2000 0.5 310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.27: Intensities regarding mixture 3 measured with GOME-2 FM3 @ 293K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 163

4 x 10 Mixture 3 FPA−4

2.5

Intensity [b.u.] 1.5

0.5

600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.28: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM3 @ 293K.

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2

12000 4

10000 3 8000

6000 2 Intensity [b.u.] Intensity [b.u.]

4000 1

2000 220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

12000 14000 ← 12000 10000 Hg−Line @ 302nm 10000 8000 8000 6000 Intensity [b.u.] Intensity [b.u.] 6000

4000 4000

2000 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.29: Intensities regarding mixture 6 measured with GOME-2 FM3 @ 293K. The lower graphics show the corresponding relevant wavelength range 164 APPENDIX B. I0 - I DOCUMENTATION

Mixture 7 FPA−1 Mixture 8 FPA−1 14000 14000

12000 12000

10000 10000

8000 8000

6000 6000 Intensity [b.u.] Intensity [b.u.]

4000 4000

2000 2000 220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

14000

12000 3000

10000 ← Hg−Line @ 296nm 8000 2500

6000 Intensity [b.u.] ← Hg−Line @ 289nm Intensity [b.u.] 2000 4000 ← Hg−Line @ 253nm

2000 1500 285 290 295 300 255 260 265 wavelength [nm] wavelength [nm]

Figure B.30: Intensities regarding mixture 7 and 8 measured with GOME-2 FM3 @ 293K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done. 165

Intensities in b.u. for measurements with GOME-2 FM3 at 273K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 7000 - 9000 2500 - 9000 1 3 400 - 455 2000 - 17000 2000 - 7000 2 2 340 - 360 7000 2300 - 6500 2 3 430 - 490 8000 - 25000 7500 - 17000 3 2 310 - 345 1700 - 7000 1600 - 6000 3 3 475 - 600 17000 - 35000 15000 - 5000 3 4 590 - 790 30000 - 2500 16000 - 2500 6 1 298 - 310 11000 - 7000 4000 - 6000 Hg(302) 10500 5800 6 2 310 - 315 2500 - 12000 2200 - 10500 7 1 285 - 303 6300 - 11000 3000 - 9000 Hg(289) 7500 4400 Hg(296) 9700 7500 8 1 240 - 290 1700 - 8000 1550 - 7000 Hg(253) 2200 1670

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 9000 8000 3 7000 2.5

6000 2 5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 0.5 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 9000 8000 1.5 7000 6000 5000 1

Intensity [b.u.] 4000 Intensity [b.u.] 3000 0.5 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.31: Intensities regarding mixture 1 measured with GOME-2 FM3 @ 273K. The lower graphics show the corresponding relevant wavelength range 166 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 9000 3.5 8000 3

7000 2.5

6000 2 5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 2000 0.5 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 9000 2 8000 7000 1.5 6000 5000 1

Intensity [b.u.] 4000 Intensity [b.u.]

3000 0.5 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.32: Intensities regarding mixture 2 measured with GOME-2 FM3 @ 273K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3

8000 3

7000 2.5 6000 2 5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 2000 0.5 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10

8000 3

7000 2.5 6000 2 5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 2000 0.5 310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.33: Intensities regarding mixture 3 measured with GOME-2 FM3 @ 273K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 167

4 x 10 Mixture 3 FPA−4

2.5

1.5 Intensity [b.u.]

0.5

600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.34: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM3 @ 273K.

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2

10000 3.5 3 8000 2.5

6000 2 1.5 Intensity [b.u.] Intensity [b.u.] 4000 1

2000 0.5 220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

10000 12000

10000 8000 8000 6000 ← Hg−Line @ 302nm 6000 Intensity [b.u.] Intensity [b.u.] 4000 4000

2000 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.35: Intensities regarding mixture 6 measured with GOME-2 FM3 @ 273K. The lower graphics show the corresponding relevant wavelength range 168 APPENDIX B. I0 - I DOCUMENTATION

Mixture 7 FPA−1 Mixture 8 FPA−1

10000 10000

8000 8000

6000 6000 Intensity [b.u.] Intensity [b.u.] 4000 4000

2000 2000 220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

3000 10000

8000 ← Hg−Line 2500 @ 296nm 6000 2000

Intensity [b.u.] ← Hg−Line @ 289nm Intensity [b.u.] 4000 ← Hg−Line @ 253nm 1500 2000 285 290 295 300 255 260 265 270 wavelength [nm] wavelength [nm]

Figure B.36: Intensities regarding mixture 7 and 8 measured with GOME-2 FM3 @ 273K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 169

Intensities in b.u. for measurements with GOME-2 FM3 at 243K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 9000 - 1000 2500 - 10500 1 3 400 - 455 2000 - 18000 2000 - 7000 2 2 340 - 360 9000 4500 - 8500 2 3 430 - 490 10000 - 25000 9000 - 20000 3 2 310 - 345 1800 - 6300 1530 - 6000 3 3 475 - 600 17000 - 35000 15000 - 5000 3 4 590 - 790 22000 - 3000 10000 - 2500 6 1 298 - 310 10000 - 5500 3500 - 5500 Hg(302) 9000 4800 6 2 310 - 315 2300 - 10500 2100 - 9500 7 1 285 - 303 5700 - 9000 2500 - 7500 Hg(289) 6500 3500 Hg(296) 8200 6200 8 1 240 - 290 1700 - 7000 1590 - 6000 Hg(253) 2150 1690

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 3 10000 2.5 8000 2

6000 1.5 Intensity [b.u.] Intensity [b.u.] 4000 1

0.5 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 10000

1.5 8000

6000 1 Intensity [b.u.] Intensity [b.u.] 4000 0.5 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.37: Intensities regarding mixture 1 measured with GOME-2 FM3 @ 243K. The lower graphics show the corresponding relevant wavelength range 170 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3

10000 2.5

8000 2

6000 1.5

Intensity [b.u.] Intensity [b.u.] 1 4000

0.5 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 10000

8000 1.5

6000 1 Intensity [b.u.] Intensity [b.u.] 4000 0.5 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.38: Intensities regarding mixture 2 measured with GOME-2 FM3 @ 243K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 8000 3 7000

6000 2.5

5000 2

4000 1.5 Intensity [b.u.] Intensity [b.u.] 3000 1

2000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 8000 3 7000

6000 2.5

5000 2

4000 1.5 Intensity [b.u.] Intensity [b.u.] 3000 1

2000 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.39: Intensities regarding mixture 3 measured with GOME-2 FM3 @ 243K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 171

4 x 10 Mixture 3 FPA−4

1.8

1.6

1.4

1.2

Intensity [b.u.] 1

0.8

0.6

0.4

0.2 600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.40: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM3 @ 243K.

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 9000 3 8000 2.5 7000 6000 2

5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 0.5 2000 220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

9000 8000 10000 7000 8000 6000 6000 5000 ← Hg−Line @ 302nm

Intensity [b.u.] 4000 Intensity [b.u.] 4000 3000

2000 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.41: Intensities regarding mixture 6 measured with GOME-2 FM3 @ 243K. The lower graphics show the corresponding relevant wavelength range 172 APPENDIX B. I0 - I DOCUMENTATION

Mixture 7 FPA−1 Mixture 8 FPA−1 9000 9000 8000 8000 7000 7000 6000 6000 5000 5000

Intensity [b.u.] 4000 Intensity [b.u.] 4000 3000 3000 2000 2000 220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

9000 8000 3000 7000 ← 6000 Hg−Line 2500 @ 296nm 5000 2000

Intensity [b.u.] 4000 Intensity [b.u.] ← Hg−Line @ 289nm ← Hg−Line @ 253nm 3000 1500 2000 285 290 295 300 255 260 265 270 275 wavelength [nm] wavelength [nm]

Figure B.42: Intensities regarding mixture 7 and 8 measured with GOME-2 FM3 @ 243K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 173

Intensities in b.u. for measurements with GOME-2 FM3 at 223K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 5000 - 6500 2000 - 6000 1 3 400 - 455 2000 - 12000 2000 - 5000 2 2 340 - 360 8000 4000 - 7800 2 3 430 - 490 10000 - 25000 9000 - 19000 3 2 310 - 345 3200 - 28000 1900 - 26000 3 3 475 - 600 22000 - 35000 20000 - 5000 3 4 590 - 790 30000 - 3500 15000 - 3300 6 1 298 - 310 9000 - 2000 5000 - 1900 Hg(302) 8500 5500 6 2 310 - 315 2300 - 10000 2100 - 9500 7 1 285 - 303 5500 - 8500 2300 - 7000 Hg(289) 6300 3300 Hg(296) 8000 5800 8 1 240 - 290 1700 - 6500 1590 - 5800 Hg(253) 2100 1700

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3 2.5 6000

2 5000

1.5 4000

1 Intensity [b.u.] 3000 Intensity [b.u.]

2000 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

6000 14000 12000 5000 10000

4000 8000

Intensity [b.u.] 3000 Intensity [b.u.] 6000 4000 2000 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.43: Intensities regarding mixture 1 measured with GOME-2 FM3 @ 223K. The lower graphics show the corresponding relevant wavelength range 174 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 3 9000 2.5 8000 7000 2 6000 1.5 5000

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 0.5 2000 300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 2 9000 8000 7000 1.5 6000 5000 1

Intensity [b.u.] 4000 Intensity [b.u.] 3000 0.5 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.44: Intensities regarding mixture 2 measured with GOME-2 FM3 @ 223K. The lower graphics show the corresponding relevant wavelength range

4 4 x 10 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 3.5 3 3

2.5 2.5

2 2

1.5 1.5 Intensity [b.u.] Intensity [b.u.] 1 1

0.5 0.5

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 4 x 10 x 10 3.5 3 3

2.5 2.5

2 2

1.5 1.5 Intensity [b.u.] Intensity [b.u.] 1 1

0.5 0.5

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.45: Intensities regarding mixture 3 measured with GOME-2 FM3 @ 223K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 175

4 x 10 Mixture 3 FPA−4

2.5

1.5 Intensity [b.u.]

0.5

600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.46: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM3 @ 223K.

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 3 8000 2.5 7000 6000 2

5000 1.5

Intensity [b.u.] 4000 Intensity [b.u.] 1 3000 0.5 2000

220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

10000 8000 7000 8000 6000 ← Hg−Line @ 302nm 5000 6000

Intensity [b.u.] 4000 Intensity [b.u.] 4000 3000

2000 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.47: Intensities regarding mixture 6 measured with GOME-2 FM3 @ 223K. The lower graphics show the corresponding relevant wavelength range 176 APPENDIX B. I0 - I DOCUMENTATION

Mixture 7 FPA−1 Mixture 8 FPA−1

8000 8000

7000 7000

6000 6000

5000 5000

Intensity [b.u.] 4000 Intensity [b.u.] 4000

3000 3000

2000 2000

220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

8000

7000 3000

6000 ← Hg−Line 2500 5000 @ 296nm

Intensity [b.u.] 4000 Intensity [b.u.] 2000 ← Hg−Line @ 289nm 3000 ← Hg−Line @ 253nm 2000 1500

285 290 295 300 255 260 265 wavelength [nm] wavelength [nm]

Figure B.48: Intensities regarding mixture 7 and 8 measured with GOME-2 FM3 @ 223K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) 177

Intensities in b.u. for measurements with GOME-2 FM3 at 203K

Mixture FPA λ [nm] I0 [bu] I [bu] 1 2 353 - 400 5000 - 6500 1800 - 5000 1 3 400 - 455 2000 - 12000 2000 - 4000 2 2 340 - 360 4800 2000 - 4500 2 3 430 - 490 6000 - 15000 5000 - 12000 3 2 310 - 345 2000 - 6800 1600 - 6000 3 3 475 - 600 15000 - 25000 12000 - 2400 3 4 590 - 790 16000 - 2200 3000 - 9000 6 1 298 - 310 11000 - 7000 14200 - 6400 Hg(302) 12770 10140 6 2 310 - 315 2300 - 12000 2100 - 10000 7 1 285 - 303 6000 - 10000 2600 - 8700 Hg(289) 7200 3800 Hg(296) 9400 7000 8 1 240 - 290 1700 - 8000 1590 - 7000 Hg(253) 2200 1700

4 Mixture 1 FPA−2 x 10 Mixture 1 FPA−3

6000 2

5000 1.5 4000 1 Intensity [b.u.] 3000 Intensity [b.u.]

0.5 2000

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

6000 12000

5000 10000

8000 4000 6000 Intensity [b.u.] 3000 Intensity [b.u.] 4000 2000 2000 350 360 370 380 390 400 400 410 420 430 440 450 wavelength [nm] wavelength [nm]

Figure B.49: Intensities regarding mixture 1 measured with GOME-2 FM3 @ 203K. The lower graphics show the corresponding relevant wavelength range 178 APPENDIX B. I0 - I DOCUMENTATION

4 Mixture 2 FPA−2 x 10 Mixture 2 FPA−3 6000 2

5000 1.5 4000 1

Intensity [b.u.] 3000 Intensity [b.u.]

0.5 2000

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

6000 12000 5000 10000

4000 8000

6000 Intensity [b.u.] 3000 Intensity [b.u.] 4000 2000 2000 340 345 350 355 360 430 440 450 460 470 480 490 wavelength [nm] wavelength [nm]

Figure B.50: Intensities regarding mixture 2 measured with GOME-2 FM3 @ 203K. The lower graphics show the corresponding relevant wavelength range

4 Mixture 3 FPA−2 x 10 Mixture 3 FPA−3 7000 2 6000

5000 1.5

4000 1 Intensity [b.u.] Intensity [b.u.] 3000 0.5 2000

300 320 340 360 380 400 400 450 500 550 600 wavelength [nm] wavelength [nm]

4 x 10 7000 2 6000

5000 1.5

4000 1 Intensity [b.u.] Intensity [b.u.] 3000 0.5 2000

310 320 330 340 480 500 520 540 560 580 600 wavelength [nm] wavelength [nm]

Figure B.51: Intensities regarding mixture 3 measured with GOME-2 FM3 @ 203K. As described in the report mixture 3 corresponds to several measurements with slightly diﬀerent concentrations. The red line shows the intensity with the lowest and the green line the highest concentration regarding mixture 3 179

4 x 10 Mixture 3 FPA−4

1.8

1.6

1.4

1.2

1 Intensity [b.u.]

0.8

0.6

0.4

0.2 600 620 640 660 680 700 720 740 760 780 wavelength [nm]

Figure B.52: Intensities regarding mixture 3 in channel 4 measured with GOME-2 FM3 @ 203K.

4 Mixture 6 FPA−1 x 10 Mixture 6 FPA−2 3.5 10000 3

8000 2.5

2 6000 1.5 Intensity [b.u.] Intensity [b.u.] 4000 1

0.5 2000 220 240 260 280 300 320 300 320 340 360 380 400 wavelength [nm] wavelength [nm]

12000 10000

10000 8000 8000 ← 6000 Hg−Line @ 302nm 6000 Intensity [b.u.] Intensity [b.u.] 4000 4000

2000 2000 300 305 310 315 310 311 312 313 314 315 wavelength [nm] wavelength [nm]

Figure B.53: Intensities regarding mixture 6 measured with GOME-2 FM3 @ 203K. The lower graphics show the corresponding relevant wavelength range 180 APPENDIX B. I0 - I DOCUMENTATION

Mixture 7 FPA−1 Mixture 8 FPA−1 10000 10000

8000 8000

6000 6000 Intensity [b.u.] Intensity [b.u.] 4000 4000

2000 2000 220 240 260 280 300 320 220 240 260 280 300 320 wavelength [nm] wavelength [nm]

10000

8000 2500 ← Hg−Line 6000 @ 296nm 2000 Intensity [b.u.] Intensity [b.u.] 4000 ← Hg−Line @ 289nm ← Hg−Line @ 253nm 1500 2000 285 290 295 300 255 260 265 270 wavelength [nm] wavelength [nm]

Figure B.54: Intensities regarding mixture 7 and 8 measured with GOME-2 FM3 @ 203K. The lower graphics show the corresponding relevant wavelength range together with arrows indicating the location of Hg-Lines, where comparisons with literature data have been done at 293K (At this temperature only at 253nm) Bibliography

[1] Andreas Richter, John P. Burrows, Hendrik Nüß, Claire Granier & Ulrike Niemeier ”Increase in Tropospheric Nitrogen Dioxide over China Observed from Space”, Na- ture 437, 129-132 (1 September 2005) [2] E. Baloitcha and G.G. Balint-Kurti, ”Theory of the photodissociation of ozone in the Hartley continuum: Potential energy surfaces, conical intersections, and photodissociation dynamics” The Journal of Chemical Physics 123, 014306 (2005) [3] J. B. Burkholder and R. K. Talukdar, ”Temperature Dependence of the Ozone Ab- sorption Spectrum over the Wavelength Range 410 to 760 nm”, Geophysical Research Letters 21, No. 7, 581-584 (1994) [4] Z. El Helou, S. Churassy, G. Wannous, R. Bacis and E. Boursey, Journal of Chemical Physics 122, 244311 (2005) [5] B.J. Finlayson-Pitts and J.N. Pitts Jr., ”Chemistry of the Upper and Lower Atmo- sphere”, Academic Press New York (2000) [6] S. Chapman, ”A theory of upper-atmosphere ozone”, Memoirs of the Royal Meteo- rological Society, Vol. III no. 26 (1930) [7] D. Bates and M. Nicolet, ”The photochemistry of atmospheric water vapor”, Geo- physical Research Letters 20, 711 (1950) [8] P.J. Crutzen, ”The influence of nitrogen oxide on the atmospheric ozone content” Q.J.R. Meteorol. soc. 96, 320-327 (1970) [9] R.S. Stolarski and R.J. Cicerone, ”Stratospheric chlorine: A possible sink for ozone”, Canadian Journal of Chemistry 52, 1610-1615 (1974) [10] M.J. Molina and F.S. Rowland, ”Stratospheric sink for chlorofluoromethanes: Chlo- rine atom-catalyzed destruction of ozone”, Nature 249, 810-812 (1974) [11] J.C. Farman, B.G. Gardiner and J.D. Shanklin, ”Large losses of total ozone in An- tartica reveal seasonal ClOx/NOx interaction”, Nature 315, 207-210 (1985) [12] R.S. Stolarski, A.J. Krueger, M.R. Schueberl, R.D. McPeters P.A. Newman and J.C. Alpert, ”Nimbus 7 satellite measurements of the springtime Antarctic ozone decrease”, Nature 322, 808-811 (1986)

[13] L.T. Molina, M.J. Molina, ”Production of Cl2O2 from the self-reaction of the ClO radical”, Journal of Physical Chemistry 91, 433-436 (1987)

181 182 BIBLIOGRAPHY

[14] M.B. McElroy, R.J. Salawitch, S.C. Wofsy and J.A. Logan, ”Reduction of antarctic ozone due to synergistic interaction of chlorine and bromine”, Nature 321, 759-762 (1986)

[15] N.R.P. Harris, G. Ancellet, L. Bishop, D.J. Hofmann, J.B. Kerr, R.D. McPeters, M. Prendez, W.J. Randel, J. Staehelin, B.H. Subbaraya, A. Volz-Thomas, J. Zawodny, and C.S. Zerefos, ”Trends in stratospheric and free tropospheric ozone”, Journal of Geophysical Research 102 (D1), 1571-1590 (1997)

[16] G. Herzberg, ”Molecular Spectra and Molecular Structure Vol. I - Spectra of Di- atomic Molecules”, Krieger Publishing Company Malabar, Florida, USA, 2nd. Edi- tion (1989)

[17] G. Herzberg, ”Molecular Spectra and Molecular Structure Vol. III - Electronic Spec- tra and Electronic Structure of Polyatomic Molecules”, Krieger Publishing Company Malabar, Florida, USA, 2nd. Edition (1989)

[18] H. Haken and H.C. Wolf, ”Molek¨ulphysik und Quantenchemie”, Springer-Verlag Berlin Heidelberg New York (2002)

[19] G. Parlant, ”Classical survival probability for ozone photodissciation in the Hartley band, Journal of Chemical Physics 112 (16), 6956-6958 (2000)

[20] R.T. Pack, ”Simple theory of diﬀuse vibrational structure in continuous UV spectra of polyatomic molecules. I. Collinear photodissociation of symmetric triatomics”, Journal of Chemical Physics 65 (11), 4765-4770 (1976)

[21] J.A. Joens,”An assignment of the structured features in the Hartley band absorption spectrum of ozone”, Journal of Chemical Physics 100 (5), 3407-3414 (1994)

[22] C. Parisse, J. Brion and J. Malicet, ”UV absorption spectrum of ozone: structure analysis and study of the isotope eﬀect in the Hartley system”, Chemical Physics Letters 248, 31-36 (1996)

[23] R. Bacis, A. J. Bouvier and J. M. Flaud, ”The ozone molecule: electronic spectroscopy”, Spectrochimica Acta A 54, 17-34 (1998)

[24] D.H. Katayama, ”New vibrational quantum number assignments for the UV absorption bands of ozone based on the isotope eﬀect”, Journal of Chemical Physics 71 (2), 815-820 (1979)

[25] D.H. Katayama, ”The UV Huggins bands of ozone”, Journal of Chemical Physics 85 (11), 6809-6810 (1986)

[26] A. Sinha, D. Imre, J. H. Goble, Jr. and J. L. Kinsey, ”Excitation spectroscopy of jet- cooled ozone: The Huggins system”, Journal of Chemical Physics 84 (11), 6108-6114 (1986)

[27] F. Le Qu´er´eand C. Leforestier, ”Theoretical calculation of the Huggins band of ozone”, Chemical Physics Letters 189 (6), 537-541 (1992) BIBLIOGRAPHY 183

[28] K. Yamashita, K. Morokuma, F. Le Qu´er´eand C. Leforestier, ”New ab initio potential surfaces and three dimensional quantum dynamics for transition state spectroscopy in the ozone photodissociation”, Chemical Physics Letters 191 (6), 515-520 (1992)

[29] A. Banichevich and S. D. Peyerimhoﬀ, ”Theoretical study of the ground and excited states of ozone in its symmetric nuclear arrangement”, Chemical Physics 173, 93-109 (1993)

[30] A. Banichevich, S. D. Peyerimhoﬀ and F. Grein, ”Potential energy surfaces of ozone in its ground state and in the lowest-lying eight excited states”, Chemical Physics 178, 155-188 (1993)

[31] J.A. Joens, ”Reassignment of the vibrational structure of the Huggins absorption band of ozone”, Journal of Chemical Physics 101 (7), 5431-5437 (1994)

[32] P. O’Keeﬀe, T. Ridley, K. P. Lawley and R. J. Donovan, ”Re-analysis of the ultraviolet absorption spectrum of ozone”Journal of Chemical Physics 115, 9311-9319 (2001)

[33] Z.W. Qu, H. Zhu, M. Tashiro, R. Schinke and S.C. Farantos, ”The Huggins band of ozone: Unambiguous electronic and vibrational assignment”, Journal of Chemical Physics 120 (15), 6811-6814 (2004)

[34] Z.W. Qu, H. Zhu, S.Y. Grebenshchikov, R. Schinke and S.C. Farantos, ”The Huggins band of ozone: A theoretical analysis”, Journal of Chemical Physics 121 (23), 11731- 11745 (2004)

[35] M. Braunstein, P.J. Hay, R.L. Martin and R. T. Pack, ”The lowest excited 1A2 and 1B1 states of ozone: Two conical intersections and their impact on photodissociation”, Journal of Chemical Physics 95 (11), 8239-8247 (1991)

[36] M. Braunstein and R.T. Pack, ”Simple theory of diﬀuse structure in continuous ultraviolet spectra of polyatomic molecules. II. Photodissociation of bent symmetric triatomics”, Journal of Chemical Physics 96 (2), 891-897 (1992)

[37] M. Braunstein and R.T. Pack, ”Simple theory of diﬀuse structure in continuous ultraviolet spectra of polyatomic molecules. III. Application to the Wulf-Chappuis band system of ozone”, Journal of Chemical Physics 96 (9), 6378-6388 (1992)

[38] H.B. Levene, J.C. Nieh and J.J. Valentini, ”Ozone visible photodissociation dynamics”, Journal of Chemical Physics 87 (5), 2583-2593 (1987)

[39] S.M. Anderson, J. Maeder and K. Mauersberger, ”Eﬀect of isotopic substitution on the visible absorption spectrum of ozone” , Journal of Chemical Physics 94 (10), 6351-6357 (1991)

[40] A. Banichevich, S.D. Peyerimhoﬀ, J.A. Beswick and O. Atabek, ”Dynamics of ozone photoabsorption: A theoretical study of the Chappuis band”, Journal of Chemical Physics 96 (9), 6580-6590 (1992) 184 BIBLIOGRAPHY

[41] S.M. Anderson and K. Mauersberger, ”Ozone absorption spectroscopy in search of low-lying electronic states”, Journal of Geophysical Research 100 (D2), 3033-3048 (1995)

[42] S.M. Anderson, J. Morton and K. Mauersberger, ”Near-infrared absorption spectra 16 18 1 of O3 and O3: Adiabatic energy of the A2 state”, Journal of Chemical Physics 93 (6), 3826-3832 (1990)

[43] J.I. Steinfeld, S.M. Adler-Golden and J.W. Gallagher, ”Critical Survey of Data on the Spectroscopy and Kinetics of Ozone in the Mesosphere and Thermosphere”, Journal of Physical Chemistry Ref. Data 16 (4), 911-951 (1987)

[44] M. Braunstein, R.L. Martin, and P.J. Hay, ”Investigation of the role of triplet states in the Wulf bands of ozone”, Journal of Chemical Physics 102 (9), 3662-3666 (1995)

[45] S.M. Anderson, P. Hupalo and K. Mauersberger, ”Rotational structure in the near- infrared absorption spectrum of ozone”, Journal of Chemical Physics 99 (1), 737-739 (1993)

[46] A.J. Bouvier, R. Bacis, B. Bussery, S. Churassy, D. Inard, M. Nota, J. Brion, J. Malicet and S.M. Anderson, ”Characterization of a metastable state of ozone by high-resolution Fourier transform spectrometry”, Chemical Physics Letters 255, 263- 266 (1996)

[47] B. Abel, A. Charvat and S.F. Deppe, ”Lifetimes of the lowest triplet state of ozone by intracavity laser absorption spectroscopy”, Chemical Physics Letters 277, 347-355 (1997)

[48] A.J. Bouvier, D. Inard, V. Veyret, B. Bussery, R. Bacis, S. Churassy, J. Brion, J. Malicet and R.H. Judge, ”Contribution to the Analysis of the 3A X1A Wulf- 2← 1 Transition of Ozone by High-Resolution Fourier Transform Spectrometry”, Journal of Molecular Spectroscopy 190, 189-197 (1998)

[49] D. Inard, A.J. Bouvier, R. Bacis, S. Churassy, F. Bohr, J. Brion, J. Malicet and 3 M. Jacon, ”Absorption cross sections and lifetime of the A2 ’metastable’ state of ozone”, Chemical Physics Letters 287, 515-524 (1998)

[50] B. Minaev and H. A˚gren, ”The interpretation of the Wulf absorption band of ozone”, Chemical Physics Letters 217 (5,6), 531-538 (1994)

[51] D. Xie, H. Guo and K.A. Peterson, ”Ab initio characterization of low-lying triplet state potential-energy surfaces and vibrational frequencies in the Wulf band of ozone”, Journal of Chemical Physics 115(22), 10404-10408 (2001)

[52] J. Orphal, S. Dreher, S. Voigt, J. P. Burrows, R. Jost and A. Delon ”The near- infrared bands of NO2 observed by high-resolution Fourier-transform spectroscopy”, Journal of Chemical Physics 109 (23), 10217-10221 (1998)

[53] V.N. Serov, V.S. Ivanov and O. Atabek, ”Split operator method for the nonadiabatic (J=0) bound states and A X absorption spectrum of NO ”, Journal of Chemical ←− 2 Physics 115 (14), 6450-6458 (2001) BIBLIOGRAPHY 185

[54] A. Delon, R. Jost and M. Jacon, ”Laser induced dispersed ﬂuorescence spectroscopy −1 of 107 vibronic levels of NO2 ranging from 12000 to 17600 cm ”, Journal of Chemical Physics 114 (1), 331-344 (2001)

2 2 [55] A. Delon and R. Jost, ”The NO2 vibronic levels near the X A1-A B2 conical intersection: Jet cooled laser induced ﬂuorescence between 11680 and 13900 cm−1”, Journal of Chemical Physics 110 (9), 4300-4308 (1999)

2 2 [56] M. Joyeux, R. Jost and M. Lombardi, ”An eﬀective model for the X A1-A B2 conical intersection in NO2”, Journal of Chemical Physics 119 (12), 5923-5932 (2003)

[57] J. Lievin, A. Delon and R. Jost ”Absorption cross section of NO2 by the reﬂection method from ab initio calculations involving the three low lying electronic states”, Journal of Chemical Physics 108(21), 8931-8943 (1998)

2 2 [58] B. Kimse, A. Delon and R. Jost ”The NO2 vibronic levels near the X A1-A B2 conical intersection observed by laser induced dispersed ﬂuorescence”, Journal of Chemical Physics 108(16), 6638-6651 (1998) [59] TPD Document MO-TR-TPD-GO-0077i1 GOME-2 Calibration: FM2 Wavelength Calibration Update (2005) [60] TPD Document MO-TR-TPD-GO-0109i1, GOME-2 FM2-1 Calibration: Wavelength Calibration and Slit Function (2005) [61] TPD Document MO-TR-TPD-GO-0097i3, GOME-2 FM3 Calibration: Wavelength Calibration and Slit Function (2005) [62] Callies, J., Corpaccioli, E., Eisinger, M, Hahne, A, and Lefebvre, A., GOME-2 - Metop’s Second-Generation Sensor for Operational Ozone Monitoring”, ESA bulletin 102 (2000) [63] J.U. White, ””Long optical paths of large aperture”, Journal of the Optical Society of America 32, 285-288 (1942)

[64] H.K. Roscoe, A.K. Hind ”The equilibrium constant of NO2 with N2O4 and the Tem- perature Dependence of the Visible Spectrum of NO2: A Critical Review and the Implications for Measurements of NO2 in the Polar Stratosphere” Journal of Atmospheric Chemistry 16, 257-276 (1993) [65] TPD Document MO-TN-TPD-GO-0060i1, GOME-2 Calibration Report (2005)

[66] J. Orphal, ”Critical Review of the Absorption Cross Sections of Ozone and NO2in the 240 - 790 nm Region” Part 1: Ozone”, ESA Technical Note MO-TN-ESA-GO- 0302 (2002), see also ”A critical review of the absorption cross sections of O3 and NO2 in the ultraviolet and visible”, Journal of Photochemistry and Photobiology A: Chemistry 157, 185-209 (2003) [67] S. Voigt, J. Orphal, K. Bogumil and J.P. Burrows, ”The temperature dependence (203-293 K) of the absorption cross sections of O3 in the 230-850 nm region measured by Fourier-transform spectroscopy”, Journal of Photochemistry and Photobiology A: Chemistry 143, 1-9 (2001) 186 BIBLIOGRAPHY

[68] A.C. Vandaele, C. Hermans, P.C. Simon, M. Carleer, R. Colin, S. Fally, M.F. Me- rienne, A. Jenouvrier and B. Coquart ”Measurements of the NO2 Absorption Cross Section from 42000 cm−1 to 10000 cm−1 (238-1000 nm) at 220 K and 294 K”, Journal of Quantitative Spectroscopy and Radiative Transfer 59, No. 3-5, 171-184 (1998)

[69] Johannes Orphal, ”Critical Review of the Absorption Cross Sections of Ozone and NO2in the 240 - 790 nm Region” Part 2: Nitrogen Dioxide” ESA Technical Note MO-TN-ESA-GO-0302 (2002), see also ”A critical review of the absorption cross sections of O3 and NO2 in the ultraviolet and visible”, Journal of Photochemistry and Photobiology A: Chemistry 157, 185-209 (2003)

[70] S. Voigt, J. Orphal and J.P. Burrows, ”The temperature and pressure dependence of the absorption cross sections of NO2 in the 250-800 nm region measured by Fourier- transform spectroscopy”, Journal of Photochemistry and Photobiology A: Chemistry 149, 1-7 (2002)

[71] R. Siebert, P. Fleurat-Lessard, S.C. Farantos, M. Bittererova and R. Schinke, ”The vibrational energies of ozone up to the dissociation threshold: Dynamics calculations on an accurate potential energy surface”, Journal of Chemical Physics 116 (22), 9749-9767 (2002)

[72] J.A. Davidson, C.A. Cantrell, A.H. McDaniel, R.E. Shetter, S. Madronich and J.G. Calvert, ”Visible-Ultraviolet Absorption Cross Sections for NO2 as a Function of Temperature”, Journal of Geophysical Research 93, 7105-7112 (1988)

[73] S. Solomon, A.L. Schmeltekopf and W.R. Sanders, ”On the Interpretation of Zenith Sky Absorption Measurements”, Journal of Geophysical Research 92, 8311-8319 (1987)

[74] U. Platt, ”In Air Monitoring by Spectroscopic Techniques”, Chemical Analysis Series 127, Wiley, New York (1994)

[75] L.T. Molina and M.J. Molina, ”Absolute Absorption Cross Sections of Ozone in the 185 to 350 nm Wavelength Range”, Journal of Geophysical Research 91 14501-14508 (1986)

[76] J. Barnes and K. Mauersberger, ”Temperature Dependence of the Ozone Absorption Cross Section at the 253.7 nm Mercury Line”, Journal of Geophysical Research 92 D12, 14861-14864 (1987)

[77] K. Yoshino, D.E. Freeman, J.R. Esmond and W.H. Parkinson, ”Absolute Absorption Cross Section Measurements of Ozone in the Wavelength Region 238-335nm and the Temperature Dependence”, Planetary Space Science 36(4), 395-398 (1988)

[78] J. Malicet, J. Brion and D. Daumont, ”Temperature Dependence of the Absorption Cross Sections of Ozone at 254nm”, Chemical Physics Letters 158, 293-296 (1989)

[79] J. Brion, A. Chakir, D. Daumont, J. Malicet and C. Parisse, ”High resolution laboratory absorption cross section of O3. Temperature eﬀect”, Chemical Physics Letters 213, 610-612 (1993) BIBLIOGRAPHY 187

[80] A. Amoruso, M. Cacciani, A. Di Sarra and G. Fiocco, ”Absorption Cross Sections of Ozone in the 590 to 610 nm Region at T = 230K and T = 299K”, Journal of Geophysical Research 95 D12, 20565-20568 (1990) [81] J. B. Burkholder and R. K. Talukdar, ”Temperature dependence of the ozone absorption spectrum over the wavelength range 410 to 760 nm”, Geophysical Research Letters 21(7), 581-584 (1994) [82] J.P. Burrows, A. Richter, A. Dehn, B. Deters, S. Himmelmann, S. Voigt and J. Orphal, ”Atmospheric Remote-Sensing Reference Data from GOME: 2. Temperature- dependent Absorption Cross Section of O3 in the 231-794 nm Range”, Journal of Quantitative Spectroscopy and Radiative Transfer 61, 509-517 (1999) [83] J. Brion, A. Chakir, D. Daumont, C. Parisse and J. Malicet, ”Absorption Spectra Measurements for the Ozone Molecule in the 350-830nm Region”, Journal of Atmo- spheric Chemistry 30, 291-299 (1998) [84] A.M. Bass and R.J. Paur, ”The Ultraviolet Cross Sections of Ozone, Part I, The Mea- surements”, C.S. Zerefos, A. Ghazi, D. Reidel (Eds.), Proceedings of the Quadrennial Ozone Symposium on Atmospheric Ozone, Norwell, Mass. 606-610 (1985) [85] R.J. Paur and A.M. Bass, ”The Ultraviolet Cross Sections of Ozone, Part II, Result and Temperature Dependence”, C.S. Zerefos, A. Ghazi, D. Reidel (Eds.), Proceedings of the Quadrennial Ozone Symposium on Atmospheric Ozone, Norwell, Mass. 611- 616 (1985) [86] D. Daumont, J. Brion, J. Charbonnier and J. Malicet, ”Ozone UV Spectroscopy I: Absorption Cross-Sections at Room Temperature”, Journal of Atmospheric Chem- istry 15, 145-155 (1992) [87] J. Malicet, D. Daumont, J. Charbonnier, C. Chakir, A. Parisse, J. Brion, ”Ozone UV Spectroscopy II: Absorption Cross Sections and Temperature Dependence”, Journal of Atmospheric Chemistry 21, 263-273 (1995) [88] EUMETSAT Document EUM/OPS-EPS/MAN/05/2005, Issue: 1.0 Date: 28/02/05 GOME-2 Products Guide [89] Michael Eisinger, ”Nachweis von BrO über mittleren nördlichen Breiten mit differ- entieller Absorbtionsspektroskopie”, Diploma-Thesis, University Bremen (1994) [90] S. A. Nizkorodov, S. P. Sander and L. R. Brown ”Temperature and Pressure Depen- dence of High-Resolution Air-Broadened Absorption Cross Sections of NO2 (415-525 nm)” Journal of Physical Chemistry A 108(22), 4864-4872 (2004) [91] M.H. Harwood and R.L. Jones,”Temperature dependent ultraviolet-visible absorption cross sections of NO2 and N2O4: Low-temperature measurements of the equilibrium constant for 2NO2 ↽⇀ N2O4”, Journal of Geophysical Research 99 (D11), 22955-22964 (1994) [92] J.C. Gómez Mart´ın, P. Spietz, J. Orphal and J. P. Burrows, ”Principal and Inde- pendent Component Analysis of Overlapping Spectra in the Context of Mulichannel Time-resolved Absorption Spectroscopy” Spectro. Chim. Acta A 60, 2673-2693 (2004) 188 BIBLIOGRAPHY

[93] P. Spietz, ”Absorption cross sections for iodine species of relevance to the photolysis of mixtures of I2 and O3 and for the atmosphere”, Thesis, Doctoral Program ”Dr. rer. nat.”, Department of Physics, University of Bremen, March 2005

[94] P. Spietz, J.C. G´omez Mart´ın and J.P. Burrows, ”Spectroscopic Studies of the I2/O3 Photochemistry, Part 2: Improved Spectra of Iodine Oxides and analysis of the IO Absorption Spectrum”, Manuscript submitted for publication at the Journal of Photochemistry and Photobiology A, July 2005 Acknowledgement

I would like to express my deep gratitude to my wife Charlotte and my parents for outstanding support during my studies.

I thank Prof. Dr. John P. Burrows, Head of the Institute of Environmental Physics, for giving me the opportunity to perform my studies in his group and under his supervision. I express my gratitude for giving me the responsibility over a very interesting project and for sharing his advice and experience in many cases to ﬁnalize this project successfully.

I express deep gratitude to my direct supervisor Dr. Peter Spietz for his patience, his time and his valuable advice during my PhD project. I am especially grateful for his topical support and his precise and accurate working culture.

I am deeply grateful for Dr. Johannes Orphal, whose topical competence allowed me to learn valuable lessons in the ﬁelds of molecular spectroscopy and remote sensing. I thank him for excellent support and counsel during the complete project, in particular for his kind and friendly hospitality during three stays in Paris for data analysis.

I thank our student Christian Albers for his time and support in the lab and in the NO2 analysis during the CATGAS GOME-2 study.

I would like to express my gratitude to ESA and EUMETSAT for funding this project under contract No. 16007/02/NL/SF. I especially thank J¨org Callies and Alain Lefebvre from ESA and Abelardo Perez-Albinana and Rose Munro from EUMETSAT for fruitful discussions during several CATGAS-GOME-2 progress meetings.

I furthermore express high appreciation to Gerard Otter, Luud van Riel, Martin Eschen, Pepijn Kenter and further members of the TPD/TNO team for outstanding support and cooperation before, during and after the CATGAS campaigns in Delft, Holland.

Last but not least I express my appreciation to the many members and colleagues of the institute for their kind and patient help, whenever it was needed. In particular to my colleagues Juan-Carlos Gomez-Martin, Lars Reichert, Dr. Maria Dolores Andres Hernan- dez, Deniz Kartal and Petra Schumacher for their support, our system administrators Heiko Schr¨oter and Heiko Schellhorn for their constant help in IT related problems and our secretaries for their patient dealings with ”us” scientists.