University of Groningen

Optical spectroscopy of interstellar and circumstellar molecules Wehres, Nadine

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record

Publication date: 2011

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Wehres, N. (2011). Optical spectroscopy of interstellar and circumstellar molecules: a combined laboratory and observational study. Rijksuniversiteit Groningen.

Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

Download date: 29-09-2021 RIJKSUNIVERSITEIT GRONINGEN

Optical Spectroscopy of Interstellar and Circumstellar Molecules

A combined laboratory and observational study

Proefschrift

ter verkrijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus, dr. E. Sterken, in het openbaar te verdedigen op vrijdag 18 maart 2011 om 16.15 uur

door

Nadine Wehres geboren op 30 april 1979 te Viersen, Duitsland Promotores: Prof. dr. A. G. G. M. Tielens Prof. dr. H. V. J. Linnartz

Beoordelingscommissie: Prof. dr. P. Sarre Prof. dr. J. M. van der Hulst Prof. dr. E. F. van Dishoeck

ISBN: 978-90-367-4804-9 ISBN: 978-90-367-4805-6 (electronic version) Für meine Eltern. Front Cover: The Red Rectangle Proto-Planetary Nebula. Credits: Nasa – This image was taken with Hubble’s Wide Field Planetary Camera 2. “The secret of getting ahead is getting started. The secret of getting started is breaking your complex overwhelming tasks into small manageable tasks, and then starting on the first one.”

– Mark Twain

Table of Contents

1 Introduction 1 1.1 Introduction ...... 2 1.2 This Thesis ...... 2 1.3 Observational Spectroscopy of Molecules in the ISM ...... 2 1.3.1 The Molecular Inventory ...... 3 1.3.2 The Radio Regime of the ISM ...... 3 1.3.3 The Infrared Spectrum of the ISM ...... 6 1.3.4 The Optical Spectrum of the ISM ...... 8 1.4 Post-Asymptotic Giant Branch (AGB) Objects ...... 11 1.4.1 The Red Rectangle Proto-Planetary Nebula ...... 11 1.5 Thesis Outline ...... 12

2 Laboratory Astrophysics 17 2.1 Absorption Spectroscopy ...... 18 2.2 Production of Transient Species ...... 20 2.2.1 Pinhole Nozzle ...... 21 2.2.2 Slit Nozzle ...... 21 2.3 The Experimental Set-Up - CRDS ...... 23 2.4 The Experimental Set-Up - LIF ...... 25 2.5 Rotational Contour Simulations ...... 28 2.6 Optical Spectroscopy using the New Technology Telescope ...... 31 2.7 Optical Spectroscopy using the Mercator Telescope ...... 33

3 A Coincidence between a Hydrocarbon Plasma Absorption Spectrum and the λ5450 DIB 35 3.1 Introduction ...... 36 3.2 Cavity Ring-Down Spectroscopy ...... 38 3.3 Optical Observations ...... 39 3.3.1 HERMES @ Mercator Telescope ...... 39 3.3.2 McKellar @ DAO Telescope ...... 40 3.4 Results ...... 40 3.5 Discussion ...... 41

3Σ− 3Σ− 4 Rotationally Resolved A u –X g Spectrum of HC7H 45 4.1 Introduction ...... 46 4.2 Experimental ...... 47 4.3 Results and Discussion ...... 47

vii Table of Contents

5 Electronic Spectra and Molecular Geometry of the non-linear Carbon Chain C9H3 53 5.1 Introduction ...... 54 5.2 Experimental ...... 57 5.3 Results and Discussion ...... 59 5.3.1 Experimental Spectra and Analysis ...... 59 5.3.2 Consideration of the Molecular Geometry ...... 60

6 C2 Emission Features in the Red Rectangle – a combined observational/laboratory study 67 6.1 Introduction ...... 68 6.2 Astronomical Observations and Data Reduction ...... 70 6.2.1 Astronomical Results ...... 71 6.2.2 Laboratory Experiment ...... 72 6.2.3 Experimental Results ...... 73 6.2.4 The C2 Rotational Contour: Excitation Temperature and Velocity Shifts ...... 75 6.3 Fluorescent Emission in the Red Rectangle ...... 77 6.3.1 Model ...... 77 6.3.2 Results ...... 80 6.3.3 The Abundance of C2 ...... 86 6.4 Implications ...... 87 6.5 Conclusions ...... 88

7 The Spatial Distribution of the Optical Emission Features in the Red Rectan- gle Proto-planetary Nebula 91 7.1 Introduction ...... 92 7.2 Observations ...... 94 7.3 Results ...... 96 7.3.1 An Offset Dependent Catalogue ...... 96 7.4 Discussion ...... 106 7.4.1 Spatial Behaviour of the Emission Bands ...... 106 7.4.2 Constraints on the Carriers of the Emission Bands ...... 109 7.4.3 The Red Rectangle Emission Bands and the DIBs ...... 109 7.4.4 Summary ...... 111 7.5 Conclusions ...... 112

Nederlandse Samenvatting 115

Zusammenfassung der Arbeit 121

Bibliography 127

Publications 133 viii Table of Contents

Curriculum Vitae 135

Acknowledgements 137

ix

Introduction1

1 Chapter 1 Introduction

1.1 Introduction

Modern astronomy is still mainly based on the observations of the material that is sur- rounding the Earth. Most conclusions and theories about the main building blocks, the species, and even the formation and ongoing development of the Universe are based on the observations of light that reaches the Earth’s atmosphere in each moment. Light in this respect is a universal carrier of vast information about our past as well as our present. To understand the information that is encoded in form of electromagnetic wave packages is the task of spectroscopists and is the task I was facing when starting my thesis. Detecting light of different energies that reaches us from places far away from the Earth can give information about the ongoing chemistry, molecule formation, reaction channels and even about the physical conditions that occur in environments far away from Earth. Last but not least the information that we obtain from the spectroscopic investigation of light gives also hint about formation and creation of life. The origin of life provides a strong motivation to investigate the chemistry that is going on thousands of lightyears away and that took place millions of ago. The challenge that modern astrochemistry is facing is the limited number of building blocks that we know about, it is the harsh conditions in circumstellar shells, in proto- planetary disks, in photon-dominated regions (PDRs) or in general in the interstellar medium (ISM). These harsh conditions put constraints on possible reaction pathways. The restrictions and different conditions push researchers to a re-think of the rules of chemistry and physics prejudiced by the experiences we make on Earth. The field of observational astrophysics combined with experimental astrophysics or astrochemistry is a complex field and will be limited here mainly to reflect the ongoing search of the molecules in the ISM. The search for the “Molecular Universe”, its main building blocks and its understanding and the challenges that it provides will be described in the following chapters.

1.2 This Thesis

In this thesis the importance of the correlation between laboratory spectroscopy and ob- servational astronomy is discussed. It will be shown how the two different approaches correlate to each other and that mainly due to the combination of both techniques, the laboratory ∼ and the observational study, conclusions can be drawn on the molecular in- ventory of the interstellar and circumstellar medium. The outcome of the study on the molecular inventory can in turn put constraints on the chemistry and also on the physical conditions in specific environments.

1.3 Observational Spectroscopy of Molecules in the ISM

Information of the species that are abundant in space is important. The information assist our understanding of the mechanisms that drive the evolution of the universe. Atoms or

2 1.3 Observational Spectroscopy of Molecules in the ISM

molecules are not only abundant around , but they are also an important component in the medium in between the stars. As stars evolve, stellar winds can blow-off material from the stars that subsequently enriches the ISM. That way atoms like H, He, but also C, O and N become part of the ISM. Supernovae can also form heavier elements that also enrich the ISM. The molecular constituents of the ISM are known to be important for the heating and cooling mechanisms. At low densities for example, CO is an important coolant and dominates the process because of the high abundance of CO compared to other species. At higher densities, other species take over and dominate the cooling in molecular clouds, for example H2O and O2. In neutral regions, HI regions, heating can be traced mainly by observations of other ionized trace species, mainly larger molecules, but also dust grains. Dust grains can also be important in regions with a strong UV field. Here they absorb the photons and become excited. De-excitation can be through emission of photons in the IR, which lead to the so-called PAH (polycyclic aromatic hydrocarbons) emission bands at specific frequencies. The emitted photons can in turn excite other molecular species ro- vibrationally. In this way molecules have a strong influence of the mechanisms that drive the ISM, they influence the thermal balance and provide coupling between key processes that also drive and planet formation. The spectra of molecules also provide a sensitive tool to study the excitation and/or de-excitation mechanisms, which eventually reflect the radiation field, temperatures and densities of their environment (see Tielens (2005) for an overview on this topic). Eventually they are the key ingredients for the formation of life.

1.3.1 The Molecular Inventory

Table 1.1 gives an overview of the molecules that have been identified in the ISM and in circumstellar shells up to now. About 170 molecules proof that the molecular inventory is rich and provides a basis for a large chemical reaction network (see also (Herbst & van Dishoeck 2009, Tielens 2005)).

1.3.2 The Radio Regime of the ISM

A major fraction of the molecules displayed in table 1.1 is organic in nature, i.e. they contain carbon. Especially the larger species are dominated by carbon atoms forming often chain-like structures, as can be seen by the largest species identified so far: HC11N (Bell et al. 1997). Only recently the C60 and C70 molecules were detected (Cami et al. 2010). The molecules can be divided in saturated and unsaturated species, depending on their radical nature. Radical species contain double or triple bonds, that become saturated if sufficient hydrogen is available. The majority of these molecules have been identified mainly through the comparison of their rotational transitions with laboratory data. In order to be observable by rotational transitions the molecules require a dipole moment, which is the case for polar species. A spectrum showing the manifold of species that can be detected in the submillimeter regime is depicted in Fig. 1.1. The formation process of these unsaturated carbon chain species involves a simple chemistry which enhances formation in harsh environments and thus detection of the species.

3 Chapter 1 Introduction 2 N CHO CO 5 2 2 ) C OH) 3 3 2 CH 3 CH 10 atoms (CH 2 CN (CH OH CH O H 6 2 2 2 4 N ) H H 7 C H 3 8 8 3 3 CH CH C C C(O)NH C 3 3 HC 3 9 atoms (CH CH CH CH CH 3 N N 6 3 H H (?) C C 11 7 6 CN (2008) 2 3 (2010) (2010) COOH 2 CCHCN OHCHO C 3 H 2 2 12 atoms HC 60 70 CHCHO (?) 8 atoms CH 2 C C l-HC > CH CH HC(O)OCH CH NCH 2 CH H ? 3 2 O H 4 (?) 2 N H 5 6 H C NH CHO 6 2 3 3 OCH 3 CN (2009) H 5 C 7 6 CCHOH HC H (2008) 7 atoms 2 H 6 H C CH 12 atoms CH c-C CHCN (2008) CH 2 3 H 2 C C CH + O H 4 N 3 4 6 N 4 C SH NC CN N OH H H (?) 9 C C H NH 2 CHO CHO 3 3 3 5 3 5 4 2 3 2 3 2 2 C C C HC N (2008) l-H CCNH (?) CH 6 atoms l-HC CH CH CH 5 OCHO (2009) n-C CH 11 atoms c-H HC 2 HC NH 5 l-HC C H H 2 C + 2 2 3 O 4 N 4 H 2 H NC H Si 3 5 3 3 2 4 4 C CCN CNH NCN C COH 2 2 2 2 CH C C SiH 2 HNC HC H (2008) l-C H c-C 5 atoms HC H H H 4 HCOOH H C HC(O)CN (2008) + 3 2 + H 3 H 3 (2008) S O N 3 3 H CS O CN CO 3 3 3 + 2 2 3 2 2 ? (2008) C CH C C NH C H H l-C H H c-C N (2008) HNCS c-SiC HCCN 3 HNCO 4 atoms HCNH 3 C PH HSCN (2009) HCNO (2009) HOCN (2010) HOCO + 2 2 + + + + 2 2 2 2 S S O O O H 3 (2010) + 3 (2010) , HD H 2 2 2 2 2 2 2 C + H + + C SO H CO CH C C NH N H OCS HCP HNC HCO HCN HNO SiCN SiNC AlNC N HCS NaCN D HCO HOC c-SiC MgNC MgCN O 3 atoms Cl 2 2 2 CCP (2008) H AlOH (2010 – Molecules in the ISM or in circumstellar shells (December 2010, data taken from CDMS (Cologne Database for Molecular Spec- + + + + 2 2 2 (2010) H (2011) C O H CS CP HF SO PO NS PN SiS CO CN CH HD OH NO NH SiC SiN SiO AlF HCl KCl CF SO CO CH AlCl + NaCl + SiH ? FeO ? 2 atoms CN (2010) AlO (2009) H SH OH troscopy). Table 1.1 (?) Some detections that haveout been at reported the as moment secure or ones because are the indicated line by listThe (?), is because somewhat most (partial) small. relevant overlap to ofpast the lines two detection cannot to (including be three isotopic ruled years. species or vibrationally excited states) is given for recent results – the

4 1.3 Observational Spectroscopy of Molecules in the ISM

Figure 1.1 – The spectrum shows the rich variety of molecular species as they are detected in the microwave regime. The spectrum is adopted from Beuther et al. (2007).

Formation of molecules inside dense clouds involves photodissociation of CO to form C and O by the cosmic-ray induced photons following excitation of H2 into the Lyman + and Werner bands. Starting with carbon, reactions with H3 can take place to form smaller + species, i.e. CH . Upon reaction with another H2 hydrocarbon ions can be formed, + + for example: CH3 or CH5 . Upon proton transfer or dissociative electron recombination neutral species like CH3 and CH4 can form. + + → + + C H3 CH H2 + + → + + CH H2 CH2 H + + → + + CH2 H2 CH3 H + + → + CH3 H2 CH5 + + → + CH5 electron CH4 H Insertion reactions of C+ followed by dissociative electron attachment lead to longer + carbon chains or hydrocarbon-chains, for example C2H3 and C2H2 (acetylene) (Tielens 2005). + + → + + C CH4 C2H3 H + + → + C2H3 electron C2H2 H + + An important reaction to start interstellar chemistry is the formation of CH2 from C and H2 that proceeds rapidly under interstellar conditions: + + → + + ν C H2 CH2 h

5 Chapter 1 Introduction

That way cationic and neutral species, saturated ∼ and unsaturated ones, can be formed in the gas phase. The formation of saturated molecules mainly happens on sur- faces of dust grains. These reactions can also be simulated in the laboratory and it seems that especially pathways for the formation of H2O and larger species such as methanol or formic acid, i.e. CH3OH or HCOOH, are enhanced on grain surfaces or icy mantles (Ioppolo et al. 2008, Fuchs et al. 2009, Öberg et al. 2009, Ioppolo et al. 2010).

1.3.3 The Infrared Spectrum of the ISM

Shifting the wavelengths of observations from the radio regime to the mid-IR regime another plethora of spectral features is revealed. This wavelength regime mainly only opened up with space based or airborne observatories (Haas et al. 1995). In this regime molecules cannot only be traced due to their rotational transitions but also due to their vibrational transitions. In this regime many observations have been carried out and a complex inventory was revealed by the spectra that could be taken from the Infrared As- tronomical Satellite (IRAS), Infrared Space Observatory (ISO) or the Spitzer Space Tele- scope. IRAS discovered wide-spread emission in the ISM (the IR cirrus) around 12 µm. Many more of these bands were discovered subsequently and are located around: 3.3, 6.2, 7.7, 8.6, 11.2 and 12.7 µm. Nowadays, it is widely accepted that these emission bands originate from photon-pumping of larger molecules, mainly C20 -C100 (Allamandola et al. 1989, Puget & Leger 1989, van Dishoeck 2004, Tielens 2008). The emission features are plotted in Fig. 1.2 and can be detected in HII regions, reflection nebulae, young stellar objects and in the outflows of post-AGB stars (for example HD 441799 associated with the Red Rectangle proto-planetary nebula). These large molecules, so called PAHs (polycyclic aromatic hydrocarbons), show flu- orescent emission between 5 and 20 µm, illustrate the richness and dominance of molec- ular species in a wide variety of environments. It was shown that these species possess large UV cross-sections and that way they can absorb light in the UV. Due to large ro- vibrational overlap of the manifold of states, internal conversion and energetic dissipation of the absorbed photon energy is possible and the energy can efficiently be transferred to lower energy levels. Re-emission of the energy causes the emission bands in the IR. That way the species can survive harsh conditions of strong UV radiation fields in the ISM. Emission of radiation then occurs at very specific frequencies, corresponding to certain vibrational modes, i.e. bending and stretching modes of hydrocarbons (Duley & Williams 1979) as could be verified in laboratory studies. The molecules that may contribute to the PAH emission features are schematically depicted in Fig. 1.3. The chemistry of such molecules seems to be more complex than the ion-molecule chemistry. Most likely these species are reformed in a sooting process in ejecta of carbon- rich AGB stars (Latter 1991, Cherchneff et al. 1992) and very similar to soot formation on Earth (Frenklach & Feigelson 1989, Allamandola et al. 1989). CO as well as C2H2 (acetylene) molecules have been found to be very abundant in AGB objects. Mainly all carbon in these objects is locked up in form of these two molecules. The formation of larger PAHs mainly follows from the reaction of acetylene with hydrogen.

6 1.3 Observational Spectroscopy of Molecules in the ISM

Figure 1.2 – The spectrum shows the richness and dominance of the so-called PAH IR emission bands. These bands can be detected over a wide variety of objects in space. Figure adopted from Tielens (2008).

3 C2H2 → C6H5 + H

In short, two acetylene molecules react under hydrogen abstraction with a third acety- lene molecule. The resulting C6H5 radical forms upon ring closure the first aromatic ring species (Frenklach & Feigelson 1989). If for once these ring-bearing species could be formed, they enrich the ISM by winds from the central star (Speck & Barlow 1997, Boersma et al. 2006). Once the ring-bearing species have been formed, they can condense to form larger molecules of poly-condensed rings: polycyclic aromatic hydrocarbons. Some of the species that are formed that way are plotted in Fig. 1.3.

7 Chapter 1 Introduction

Figure 1.3 – Some examples of condensed ring-bearing species are shown - polycyclic aromatic hydrocarbons (PAHs). Figure adopted from Salama et al. (1996).

1.3.4 The Optical Spectrum of the ISM As we have seen the mid-IR regime, but also the radio regime of the ISM, shows a variety of species and is remarkably rich in its spectral features. But the longest standing history of observations of the interstellar material is the one of optical spectroscopy. Hartmann (1904) reported for the first time cases of a stationary ionized calcium line towards a spectroscopic binary. Two decades later the identification of both the interstellar calcium H & K lines, as well as the Na D1 and D2 lines were well accepted (Struve 1927, 1930). Mary Lea Heger in 1922 was the first person to report two new possible stationary diffuse absorption lines in the optical spectra of several B-type stars (Heger 1922). Systematic studies started with Merrill (1934), who studied binary systems and soon discovered four more interstellar absorption bands centred around ∼ 5780.4, 5796.9, 6283.9 and 6613.9 Å. In 1937 Merrill et al. (1937) discovered 5 more lines. The idea that interstellar molecules could cause some of the absorption bands came up by Swings & Rosenfeld (1937) and subsequently molecules like CH and CN could be identified (McKellar 1940) at ∼ 3874.6 Å and ∼ 4300.2 Å. After comparison with laboratory

8 1.3 Observational Spectroscopy of Molecules in the ISM

-1

Frequency [cm ]

24000 22000 20000 18000 16000 14000 12000

NormalizedIntensity

4500 5000 5500 6000 6500 7000 7500 8000 8500 9000

W avelength [Å]

Figure 1.4 – A simulated spectrum of the DIBs catalogue as available online by Hobbs et al. (2009).

data Douglas & Herzberg (1941) identified the CH+ cation in the spectra at ∼ 4232.6 Å, 3957.7 Å and 3745.3 Å. Especially the advances in laboratory spectroscopy lead to the identification of most molecular species in the ISM (see table 1.1). Up to now more than 400 absorption bands have been detected in the ISM in diffuse clouds that are seen towards reddened stars. These so-called diffuse interstellar bands (DIBs) are still elusive, even now, nearly 100 years after their discovery and despite of all the advances in radio and IR astronomy and technical developments in the laboratory. The DIBs seem to be easily reproduced by the different environments occurring in the interstellar medium, since the bands have been detected in different lines of sight. In 1975 Herbig (1975) published a first systematic survey of the DIBs as detected on photographic plates. Some 39 DIBs were recorded at this time. Jenniskens & Desert (1994) started a systematic search for the DIBs in the spectra of four reddened stars and his research group detected another 64 new bands and in addition another 69 “possible” DIBs, that still needed to be confirmed at this time. Only 6 years later, in 2000, Tuairisg et al. (2000) published a new survey of the DIBs and raised the number of DIBs to 226, along with ∼ 25 “possible” detections awaiting confirmation. Finally, Hobbs et al. (2009) published the most recent study up to now, using Echelle spectra of HD 183143 with a resolving power of R =38,000. This most recent catalogue has a total number of 414 bands and an additional list with 71 “possible” DIB detections is presented.

9 Chapter 1 Introduction

Fig. 1.4 shows the observed DIB spectrum, simulated here from the data as available online from Hobbs et al. (2009). Nowadays, nearly 100 years later, unambiguous identi- fication of the DIBs is still lacking. Solid state particles, such as silicates, were excluded due to lack of polarization (Adamson & Whittet 1992). The idea that the DIB carriers are in the form of gas-phase species more than in or on grains, is implied by the fact that some DIBs reveal substructures in their absorption bands. This substructure seems to indi- cate not fully resolved rotational substructure, as can be the case for molecular gas-phase species. Recent studies seem to indicate that the similar behaviour of the DIBs in different environments –and thus under different environmental conditions– reflect structurally re- lated carriers. The behaviour may reflect an interplay of ionization, electronic excitations, formation and destruction mechanisms of carbonaceous species (Herbig 1993, Cami et al. 1997, Sonnentrucker et al. 1997). Examination of the DIBs has revealed correlations that suggest that the DIBs or the family of DIBs respond differently due to ionizing radiation, shielding by interstellar grains and possibly a different behaviour because of different on- going chemistry. With this idea the characterization of DIBs into certain families of bands has been pushed forward. Krelowski & Walker (1987) have proposed three families of bands, based on the variations in strengths they found in different lines of sight. Among the most likely carriers – as it is thought nowadays – are pure carbon chains and hydro- carbons, cationic PAHs and fullerenes (Ehrenfreund & Foing 1995, Salama et al. 1996, Tulej et al. 1998, Foing & Ehrenfreund 1994, 1997, Zhou et al. 2006, Linnartz et al. 2010, Maier et al. 2010).

1.3.4.1 Polycyclic Aromatic Hydrocarbons (PAHs) Since neutral PAHs have been found to absorb in the UV and re-emit in the IR, it was supposed that the DIBs may be caused by cationic species of PAH molecules. The ab- sorption bands that way shift to smaller energies. PAHs seem to be among the abundant carbon species in the ISM and the fingerprints of emission bands in the IR can be seen in various environments (Puget & Leger 1989). About 1-10% of the galactic carbon could that way be locked up in form of PAHs.

1.3.4.2 Fullerenes Fullerens are a special form of pure carbon, like graphene or diamonds. Kroto et al. (1985) discovered the polyhedral structure and Krätschmer et al. (1990) synthesized the first fullerene C60 from condensed soot. The stability is remarkable and this suggests that fullerenes may be around in interstellar space (Ehrenfreund & Foing 1995). Indeed the first C60 and C70 molecules were just identified in the ISM and may strengthen the hypothesis further (Cami et al. 2010).

1.3.4.3 Pure Carbon and Hydrocarbon Chains In the laboratory carbon chains as well as hydrocarbons have been found to absorb and emit light strongly in the optical spectrum. Identification of many pure carbon chains and

10 1.4 Post-Asymptotic Giant Branch (AGB) Objects

hydrocarbons in different interstellar environments are often used to make the connection to carbon chains as possible DIB carriers (Bell et al. 1997, Gottlieb et al. 2000, McCarthy et al. 2006, Brünken et al. 2007, Linnartz et al. 2010, Maier et al. 2010). In 2003 the identification of C3 in the ISM was achieved by (Maier et al. 2001). In 2010 a broad absorption band of a hydrocarbon plasma showed overlap with the λ5450 DIB, one of the broad DIBs in HD 183143 (Linnartz et al. 2010) (see Chapter 3). This band has very recently been in the focus of another study that identifies the band as the linear propadienylidene molecule, l-H2C3.

1.4 Post-Asymptotic Giant Branch (AGB) Objects

The research of the molecular inventory goes to more and more complex species, as a recent review by Herbst & van Dishoeck (2009) shows. The five main types of sources in that especially the larger and more complex molecules have been detected (see Table 1.1) are: circumstellar envelopes around evolved stars, cold interstellar cores, hot cores and corinos, lukewarm corinos, outflows of post AGB stars. The latter object is interesting since especially during its late AGB evolution stellar pulsations and a strong external dusty mass loss is developed. Within timescales of 104 – 105 years this mass loss will remove essentially the whole envelope. Molecules like methanol (CH3OH), methyl for- mate (HCOOCH3), acetonitril (CH3CN) and ethanol (C2H5OH) have been detected in the outflows of post-AGB stars. Chemical reactions that further take place in the outflows enhance the formation of larger molecules or dust grains (Cami et al. 2010, van Winckel 2003). More than 60 different molecular species have been identified in the outflows of such objects as well as dusty minerals, silicates and carbonacious grains (Waters et al. 1996, Molster et al. 2002b,a,c, Cernicharo et al. 2001, Pardo et al. 2007, Kwok 2009). Just recently the largest molecules so far, C60 and C70, were identified in TC-1, a plane- tary nebula, formed from a post-AGB object. Also molecules like larger aromatic species, PAHs, are likely to be abundant in these environments. This argument is strengthened by the fact that benzene, polyynes and cyanopolyynes could already be identified in these environments (Cernicharo et al. 2001, Pardo et al. 2007).

1.4.1 The Red Rectangle Proto-Planetary Nebula A special post-AGB object in this respect is the Red Rectangle proto-planetary nebula. The post-AGB object HD 44179 that is associated with the Red Rectangle nebula has caught special attention since its identification in 1975 (Cohen et al. 1975). The figure on the cover shows a Hubble Space Telescope (HST) image of the Red Rectangle nebula. Most likely the central object is a binary (Waelkens et al. 1996), that cannot be seen directly, but is obscured by an optically and geometrically thick disc (Roddier et al. 1995, Osterbart et al. 1997, Bond 1997). The Red Rectangle itself is often cited as the proto- type of a carbon-rich post AGB star (Cohen et al. 2004). The nebula itself is carbon-rich and shows pronounced PAH emission features (Russell et al. 1978) in the infrared. In the optical the spectrum shows a broad and extended red emission (ERE) seen only in

11 Chapter 1 Introduction carbon-rich environments (Furton & Witt 1992), for example in reflection nebulae (Witt & Boroson 1990), in spectra of HII regions (Perrin & Sivan 1992, Darbon et al. 1999, 2000), in the galactic cirrus cloud (Guhathakurta & Tyson 1989, Szomoru & Guhathakurta 1998), the diffuse interstellar medium (DISM) (Gordon et al. 1998) and in other galaxies (Perrin et al. 1995). This broad emission band, spanning about 250 nm in the optical, was attributed to photo-luminescent processes of silicate particles or carbonacious grains (Witt et al. 1998, Ledoux et al. 1998, Wada et al. 2009). On top of the ERE, narrow emission features, were detected (Schmidt et al. 1980). While the ERE is widespread, the narrow emission bands superimposed on the ERE are unique. A spectrum of the Red Rectangle nebula is shown in Fig. 1.5. The emission bands show a lack of polarization (Schmidt et al. 1980) and exhibit molecular characteristics of gas-phase species. In several studies it was shown that these unique emission bands shift in wavelength to the blue and also become more narrow with distance to the central star (Sarre et al. 1995, Van Winckel et al. 2002) (see also Chapter 7 of this thesis). Some of these bands have caught special attention since with distance they shift very close to the band positions of some of the diffuse interstellar absorption bands. The narrow emission bands superimposed on the ERE will be described in more detail in Chapters 6 and 7. The nebula itself has been found to be carbon rich, including the blue luminescence (Vijh et al. 2004), the ERE and the PAH emission bands. Some smaller species have been identified in the UV and in the IR (Hall et al. 1992, Bakker et al. 1996, Hobbs et al. 2004) and only C2 could be identified in the optical (Sarre 2006, Glinski et al. 2009, Wehres et al. 2010a) (see also Chapter 6). The disk surrounding the central star instead is oxygen 12 13 rich, with molecules like OH, CO, CO, as well as CO2 in the gas phase and maybe also in the solid state (Leger & Puget 1984, Reese & Sitko 1996, Waters et al. 1998). The very weak microwave emission of CO (Jura et al. 1995) in the Red Rectangle, but the otherwise strong emission features of PAHs, as well as the ERE, seem to imply that further processing of the dusty outflows took place in a circumbinary rotating disk. Oxygen rich silicates were later found in the ISO spectrum of the Red Rectangle (Waters et al. 1998). This seems to indicate a spatially separated chemistry that is going on in this object and that the dusty disk and the nebula were formed at different evolutionary phases of the post-AGB phase. Most likely the oxygen-rich material in this carbon-rich environment has been formed earlier than the carbon-rich material.

1.5 Thesis Outline

The results described in this thesis address some of the key questions modern astrophysics is facing nowadays. The questions discussed here aim mainly in the direction towards the chemical compounds in the Universe. In this thesis we restrict ourselves to the molecules found in the ISM and moreover we will focus on the identification of molecules detected in the visible regime. The approach we are following is to compare the observational spectra of astronomical environments or specific objects in the ISM to laboratory spectra of specific molecules. The questions that arise from this approach are: • Is it possible to simulate astronomical environments in the laboratory in such a

12 1.5 Thesis Outline

Figure 1.5 – The figure shows the spectrum of the Red Rectangle in the optical regime. Figure adopted from Schmidt et al. (1980).

way, that we can create molecular species that can be detected in the ISM, more specifically, molecules which can be detected in diffuse interstellar clouds or in the Red Rectangle environment? The experimental approach will be described in Chapter 2 and the laboratory spectra will be discussed in Chapters 3 – 6.

• Is it possible to identify these species in the laboratory and subsequently in the ISM? Is it possible to put constraints on species that cause the DIBs or molecules that cause the emission bands that we detect in the Red Rectangle proto-planetary nebula? This question will be discussed in detail in Chapters 3 and 6.

• If we know about the molecules that are observed in the ISM, can we conclude on physical conditions that are necessary in order to form these species in a certain en- vironment? In Chapter 6 we discuss the identification of a molecular species (C2) in the environment of the Red Rectangle nebula. The implication of the identification is discussed and physical conditions are determined. Chemical implications will be discussed subsequently.

• If we know the physics in a specific environment, i.e. temperatures, radiation fields, densities of molecules, is it possible to determine on reaction pathways and estimate abundances of molecules in this environment?

• If we know the physics and chemistry, is it possible to determine on formation and destruction pathways and timescales in specific environments? These two questions will be explained in detail in Chapter 6, where likely reaction mechanisms of C2 in the Red Rectangle are studied that at the end also lead to an estimate of the abundance of C2.

• If we know about specific molecules, their chemistry, their network of reactions and the physics going on in certain environments in the ISM, is it possible to modify reaction schemes for other environments?

13 Chapter 1 Introduction

• Finally, if we understand the mechanisms that drive the formation and destruc- tion schemes of molecules, if we know about the physical conditions and ongoing chemistry, is it possible to also put constraints on the formation of life?

The last two bullet points are beyond the scope of this thesis but provide a strong motivation for studying the reactions of molecular species in the ISM.

In Chapter 2 an introduction to the laboratory experiments are given. The techniques are described that are used to form molecules of astrophysical interest and how their spec- trum is obtained. This addresses the first bullet point mentioned above and shows the tech- niques that are used in order to form molecular species that then are identified in the ISM. For the creation of molecular species two specific plasma sources are used. The spectrum of molecules are recorded using two different techniques: cavity ring-down spectroscopy and laser induced fluorescence spectroscopy. Both techniques will be explained and spec- tra will be shown and described. The spectra will be compared to observational spectra of diffuse interstellar clouds and to the Red Rectangle proto-planetary nebula. For this purpose it will briefly be summarized how it is possible to simulate the laboratory spectra under different conditions as we may face them in different regimes in the ISM. Observa- tional methods for obtaining optical long-slit spectra of the Red Rectangle nebula and an optical echelle spectrum of the diffuse cloud seen in the line of sight towards HD 183143 will be explained.

In Chapter 3 a laboratory study of a hydrocarbon plasma absorption spectrum as recorded in the laboratory is discussed and compared to spectra of the diffuse interstellar cloud towards HD 183143. The study shows a clear overlap of the laboratory ∼ and the observational study, but a clear assignment of the carrier could not be obtained. The ex- periment did, however, put constraints on the carrier of this specific absorption band in the ISM.

In Chapter 4 a specific molecule, HC7H, formed from a hydrocarbon plasma, as ex- plained in Chapter 2, is discussed. The laboratory absorption spectrum is compared to the diffuse interstellar cloud towards HD 183143, as online available (Hobbs et al. 2009). Several rotational transitions could be assigned in the laboratory study and information of the intrinsic molecular constants is obtained. That way simulations were run to model the spectrum for different temperatures and resolutions that allow for detailed comparison to observational spectra. No match between the simulations and the absorption spectrum of HD 183143 is obtained.

Chapter 5 discusses the C9H3 molecule recorded through an expanding hydrocarbon plasma. We describe the laboratory approach to determine intrinsic molecular parameters that allow for identification of the geometry of the carrier of the laboratory absorption band. Two transitions of the same carrier are identified in the laboratory study. Simula- tions were run to compare the bands to the diffuse interstellar absorption bands of diffuse interstellar cloud towards the star HD 183143. No overlap between the simulation and

14 1.5 Thesis Outline

the astronomical absorption spectrum could be obtained, although this species is said to be the seed of ring-bearing species in the ISM (Schmidt et al. 2003b).

In Chapter 6 the laboratory excitation spectra of the C2 Swan bands are recorded and compared to the observational study the Red Rectangle nebula, HD 44179. Identification of two narrow emission bands is discussed and conclusions on the ongoing chemistry and the physical conditions in this environment are drawn. The reaction pathways for the for- mation and destruction of C2 are described and the abundance with distance to the central star is derived.

In Chapter 7 we finally show an overview of all long-slit spectra of the Red Rectangle that were recorded with the New Technology Telescope in La Silla, Chile. An inventory of all optical emission bands superimposed on the extended red emission is presented. The emission bands are summarized in band position, band widths and intensities with respect to the position in the outflows of the post AGB object. A comparison of the narrow emission bands to the DIBs will be discussed and constraints on the carriers will be summarized.

15

Laboratory and Observational Astrophysics2

17 Chapter 2 Laboratory and Observational Astrophysics

Introduction

This chapter focuses on the experimental set-ups that are used in this study to obtain spec- tra of transient species of astrophysical interest. This chapter is split into four parts. The first part describes the principle of cavity ring-down (CRD) absorption spectroscopy. The CRD experiment is used to obtain electronic spectra of carbon chains and hydrocarbon species to compare to astronomical spectra obtained from diffuse interstellar clouds (see also Chapters 3-5). The second part of this chapter describes laser induced fluorescence (LIF) spectroscopy. Spectra of C2 are presented and compared to astronomical spectra of the Red Rectangle proto-planetary nebula (see Chapter 6). The CRD and LIF experiments are combined with special plasma expansions that are described as well. The third part of this chapter describes the approach to identify these species in different environments and how simulations using intrinsic molecular parameters can assist our understanding in obtaining further information about rotational temperatures of the molecules in the laboratory and in the ISM. That way simulations are an important tool in assisting unam- biguous assignments of the transitions detected in the laboratory and in the ISM. Finally, the fourth part of this chapter describes the observational techniques used to obtain the spectra of the Red Rectangle proto-planetary nebula and of the diffuse interstellar cloud located in the line of sight towards HD 183143.

2.1 Absorption Spectroscopy

Atoms and molecules absorb electromagnetic radiation at specific frequencies that are characteristic, i.e., each species possesses a unique “fingerprint” spectrum. The only chance of identifying molecules unambiguously in the interstellar or circumstellar medium (ISM or CSM) is to compare astronomical data with spectra recorded under laboratory controlled conditions. Absorption of electromagnetic radiation, under conditions where saturation does not occur is described by “Lambert-Beer’s” law: ( ) l = 0 × −σ I(ν) I(ν) exp (ν)nl (2.1)

0 In this equation I(ν) is the intensity of the incident radiation field at a certain frequency ν l ( ). I(ν) is the intensity after the radiation has passed a distance (l) through an absorbing medium. The transmitted light depends on the number of particles in the absorbing mate- rial (n) and the cross-section (σ(ν)) of the absorbing material. The product (σ(ν) ×n) is also known as the absorption coefficient, labelled (α(ν)). A crucial point and a limiting factor when recording absorption spectra in the laboratory is to measure the energy change of the transmitted light. Absorption spectroscopy is not a very sensitive technique, as gen- erally a rather small decrease in signal has to be measured against a large background signal. The number of species that can absorb (n) and the absorption pathlength (l) are the main parameters that can be increased in order to obtain higher sensitivity spectra. CRD spectroscopy is a special case of absorption spectroscopy that results in high detec- tion levels. In CRD spectroscopy the absorption length is extended by using two highly

18 2.1 Absorption Spectroscopy

Figure 2.1 – The basic principle of CRD spectroscopy is shown. A laser beam traverses the first mirror (R1). With each round trip in between the cavity mirrors (R1 and R2) the intensity becomes weaker due to absorption by gas and mirror losses. The light leaking out of the cavity is detected using a photomultiplier tube and the ring-down signal is displayed on an oscilloscope. Fitting the exponential decay gives the ring-down time. Recording the ring-down time as function of laser frequency yields the absorption spectrum. reflective cavity mirrors that form a high finesse optical cavity in which the light can be trapped (typically R ≥ 99.9%). The intensity of the light leaking out of the cavity de- creases by a fixed percentage during each round trip inside the cavity. This is due to the eventual absorption of the medium in the cell and the reflectivity losses of the mirrors. The intensity of the light leaking out of the cavity is then determined as an exponential function of time. In 1980, Herbelin et al. (1980) were the first to use this decay rate of light leaking out of an optical resonator to determine mirror reflectivities. The first paper on the CRD technique was published in 1988 (O’Keefe & Deacon 1988), applying the principle of reflectivity as a laser based direct absorption measurement technique. The CRDS principle is displayed in Fig. 2.1. A light beam traverses the first mirror

19 Chapter 2 Laboratory and Observational Astrophysics

(R1) and enters the cavity. The light bounces back and forth between the two cavity mirrors (R1 and R2). A small portion of the light is leaking out each time when hitting one of the highly reflective mirrors and reaches the photomultiplier tube (PMT) detector that is located on one side of the cavity. The resulting exponential decay is given by:

I(t) = I0 exp (−t/τ) (2.2) Here, (τ) is the decay time or the so-called “ring-down time” and is a measure for the time the light pulse resides in the cavity. (I(0)) is the signal intensity prior to its decay. The ring-down time is dependent on the medium inside the cavity and the mirror losses. At first the losses for an empty cavity are determined:

n l τ = · (2.3) 0 c 1 − R + X where (n) is the index of refraction within the cavity, (c) the speed of light in vacuum, (l) the cavity length, (R) mirror reflectivity, and (X) takes into account other optical losses. If upon gas injection the wavelength of the laser coincides with an allowed transition of molecules inside the cavity the losses will be higher and the ring-down time of the incident light pulse will be lower.

n l τ = · (2.4) c 1 − R + X + αl Here (α) is the absorption coefficient, mentioned before, which is specific for each sub- stance. The stronger the sample absorbs, the faster the light decays. Depending on the mirror reflectivity, the effective path length in a typical ring-down experiment can reach up to several kilometres. Detailed studies can be found in several review articles (Berden et al. 2000, Saykally & Casaes 2001, Paldus & Kachanov 2005) and several textbooks (Demtroeder 1996, Linnartz 2009) as well as in papers which discuss specific topics re- lated to CRDS theory (Zalicki & Zare 1995, Lehmann & Romanini 1996, Hodges et al. 1996, Lee et al. 1999, Parkes et al. 2003). The fast fitting algorithm that was adopted here for the fitting routine of the exponential decay curve is described in detail by Halmer et al. (2004). Different CRD schemes exist. The work described in chapters 3-5 is based on a pulsed CRD detection scheme. This is essentially the most straightforward way to use this method, as a cavity is fully transparent for rather broad laser pulses, i.e., it is not necessary, as in continuous wave (cw) CRD schemes to scan the cavity length in order to induce a ring-down event (see e.g. Birza et al. (2002)). This makes pulsed CRD spec- troscopy ideal to combine with pulsed plasma expansions that are described in the next section.

2.2 Production of Transient Species

The formation of transient species is a crucial step in this experiment. Discharge ∼ or plasma sources are needed in order to create conditions that enhance the formation of

20 2.2 Production of Transient Species molecules that are not stable under normal (terrestrial) conditions. Different kinds of discharge sources are used here, with specific advantages and disadvantages.

2.2.1 Pinhole Nozzle In Fig. 2.2 a pinhole discharge source is shown. The pinhole nozzle is used to form carbon bearing species, pure carbon chains, hydrocarbons, cationic species and radicals from a precursor gas. Here a mixture of 0.5% C2H2 in He is used. The discharge body consists of several plates in a multilayer geometry as described by Zhao et al. (2010) (see also Chapter 5). More specifically one metal plate is connected to a negative HV supply (C, cathode) and a second metal plate is connected to ground (A, anode). The plates are separated by a ceramic insulator (I1, insulator) plate. The bottom most ceramic plate (I2) is used to prevent the discharge from clogging. Formation of the transient species mainly takes place between the anode and cathode plate. The HV leads to fragmentation of the C2H2 molecules. High backing pressures of ∼ 10 bar lead to collisions between all the fragments and enhance formation of new species. The exact chemistry inside a plasma environment is still not very well understood, but similarly as in space, it is assumed that chain structures are formed from a series of barrierless + combination reactions starting from very small radicals, C2, CH , CH, C2H,..., and upon insertion into acetylene (C2H2) longer carbon chains or hydrocarbon chains are formed (Tielens 2005). For the experiments the pinhole nozzle has been optimized for obtaining high-resolution spectra of longer carbon chains. This is achieved due to the thickness of the bottom plate (I2) which decreases efficiently the jet opening angle of the gas expansion and enhances collisions. Collisional cooling therefore becomes more efficient and the species are rota- tionally and vibrationally cold, but can be in an electronically excited state. In addition, the smaller expansion angle decreases the Doppler broadening and allows for spectra with higher resolution.

2.2.2 Slit Nozzle The second type of discharge nozzle that has been used in these experiments, especially in the cavity ring-down experiment, is a so-called slit nozzle. A schematic overview is shown in Fig. 2.3. The figure shows the slit nozzle with the pulsed solenoid valve at the bottom (gas in- let). The right panel shows a cut through the nozzle body. Gas is injected and distributed evenly over the whole slit. The discharge body consists of an anode (A) and a cathode (C) plate separated by a ceramic insulator (I1) plate. The nozzle is used for the production of carbon chain species as well as hydrocarbons, similar to the pinhole nozzle described before. The main advantage of using the slit nozzle compared to the pinhole nozzle is the geometry which allows for a planar plasma expansion obtaining a nearly “Doppler- free” environment. Furthermore, the absorption pathlength is increased and as the density profile drops linearly with the distance to the nozzle orifice, the plasma environment is more efficient. The result is an increase in resolution compared to the pinhole nozzle

21 Chapter 2 Laboratory and Observational Astrophysics

Figure 2.2 – A schematic view of the discharge source used to generate transient species in a hydro- carbon plasma expansion. The discharge comprises two metal plates used as anode (A, grounded) and cathode (C, negative HV), separated by a ceramic plate (I1, insulator). The discharge is con- nected to a pulsed solenoid gas injection system (General valve series 9) that operates at high pres- sure, typically around ∼ 3-12 bar and separates the high pressure side from the vacuum side in the chamber (typically around ∼ 0.1 mbar for cavity ring-down spectroscopy and around ∼ 10−4 mbar for the laser induced fluorescence spectroscopy upon gas injection).

expansion. The disadvantage of the slit nozzle is that the cooling is less efficient which results in “hotter” spectra in which the molecules show more population in the excited ro- tational and vibrational levels. The main advantage of using both nozzles is the additional information that one can obtain by comparing the high resolution slit-nozzle spectrum (containing many congested bands) with the less “crowded” (colder) spectrum obtained

22 2.3 The Experimental Set-Up - CRDS

Figure 2.3 – A schematic diagram of the slit-nozzle discharge source. The system is essentially the 2D equivalent of the source shown in Fig. 2.2. The right panel shows a cut through the slit-nozzle in that the multi layer construction can be seen. As previously indicated for the pinhole nozzle I1 and I2 respresent the insulators, C represents the cathode and A is the anode. by the pinhole nozzle (with a loss of resolution).

2.3 The Experimental Set-Up - CRDS

Upon discharging the C2H2/He mixture, the carrier gas is expanded into a vacuum cham- ber. The supersonic plasma expansion is used to cool down the molecules adiabatically. Molecules in this environment have the characteristic of being rotationally and vibra- tionally (ro-vibrationally) rather cold, but can remain in an excited electronic state. This is an important aspect, since electronically excited species are also found in environments in the interstellar medium (see also Chapter 6). In Fig. 2.4 the experimental set-up is shown as used at the Laser Centre at Vrije Universiteit Amsterdam. The set-up has been described in detail in (Linnartz 2009) and (Wehres et al. 2010b) (see also chapters 3–5). The light of a frequency tripled Nd:YAG laser (∼ 355 nm) is coupled into a Sirah dye laser at a repetition rate of 10 Hz, with a pulse width of ∼ 6 ns and a bandwidth narrower than 0.04 cm−1. Absolute laser frequency cal- ibration is performed using an iodine or tellurium reference spectrum that is recorded simultaneously to the experiment. The light is guided into the spectroscopy chamber and intersects the discharge expansion perpendicularly. The light leaking out of the cavity

23 Chapter 2 Laboratory and Observational Astrophysics

Figure 2.4 – The figure depicts schematically the experimental CRDS set-up as it is used at the Laser Centre at the VU. mirrors (Research Electro-Optics, R ∼ 99.998% at 532 nm) traverses a narrow band pass filter and is then detected by a photomultiplier tube (PMT). The ring-down signal is anal- ysed by a LabVIEW computer program and about 6-10 ring-down events are normally averaged to determine one data point. A complete spectrum is plotted showing the sig- nal intensity with respect to the laser frequency. More details on this experiment can be found in detail in Naus et al. (1997), Witkowicz et al. (2004), Linnartz (2009), Wehres et al. (2010b). Fig. 2.5 shows an absorption overview spectrum of a hydrocarbon plasma as recorded 3 3 in the laboratory. The C2 Swan band system (d Πg ← a Πu) is visible and shows im- pressively the manifold of rotational states that are populated. Also absorption bands + + of previously assigned larger carbon species such as HC4H ,C6H and C5 can be seen (Prasad & Bernath 1994, Lloyd & Ewart 1999, Motylewski & Linnartz 1999, Raghunan- dan et al. 2009). The resolution of the laser is sufficient to resolve the rotational bands of C2. For larger carbon chains the spacing between the rotational levels is decreasing fast and rotational levels sometimes form an envelope covering multiple transitions. Also life- time broadening of states can lead to unresolved spectra which prevent clear identification of the carrier characteristics.

24 2.4 The Experimental Set-Up - LIF

W avelength [nm]

516 514 512 510 508

C (0,0)

2

C (1,1)

2

C (2,2)

2

Intensity [a.u.] Intensity

+ C +

5 C H HC H

6

4

19350 19400 19450 19500 19550 19600 19650 19700

-1

Frequency [cm ]

Figure 2.5 – The figure shows a typical absorption spectrum recorded through a supersonically expanding hydrocarbon plasma. Clearly visible are the many bands created from C2 Swan band 3 3 (d Πg ← a Πu) absorption.

Another spectrum illustrative of the high resolution that can be obtained using the slit-nozzle is presented in figure 2.6, which shows the A2Π ← X2Π origin band transition of C6H. The upper spectrum shows the spectrum as obtained in the laboratory and the lower (mirrored) spectrum shows the simulation using PGOPHER (Western 2007), a pro- gram that simulates the rotational contour and that will be explained in a bit more detail below. The laboratory spectrum shows that the P, Q and R branch can be recorded fully rotationally resolved for both spin-orbit components (Linnartz et al. 1999).

2.4 The Experimental Set-Up - LIF

The experimental set-up for laser induced fluorescence spectroscopy (LIF) is located at the S ackler Laboratory f or Astrophysics at Leiden Observatory. The experiment has been described in detail by Volkers et al. (2004) with the modifications mentioned in Wehres et al. (2010a). In Fig. 2.7 the experimental set-up is shown. A Scanmate dye laser is optically pumped by a frequency doubled Nd:YAG laser. The repetition rate is at 10 Hz and the laser pulse width is ∼ 7 ns. The resolution is ∼ 0.07 cm−1. Mirrors and

25 Chapter 2 Laboratory and Observational Astrophysics

W avelength [nm]

526.8 526.7 526.6 526.5

1.0

0.8

0.6

0.4

0.2

0.0

-0.2 Intensity [a.u] Intensity

Simulation

-0.4

T ~ 20 K

rot

-1

-0.6

R ~ 0.08 cm

-0.8

-1.0

18980 18982 18984 18986 18988 18990 18992 18994 18996

-1

Frequency [cm ]

Figure 2.6 – The figure shows the high-resolution spectrum of C6H recorded in a hydrocarbon plasma expansion using cavity ring-down (CRD) spectroscopy. The simulation is mirrored and yields a rotational temperature of 20 K for the laboratory experiment, which is comparable to tem- peratures in diffuse interstellar clouds. The resolution is about 0.08 cm−1. lenses are used to focus the light into the vacuum chamber. The laser intersects perpen- dicularly with a supersonic jet expansion about 15 cm downstream of the plasma nozzle (see Fig. 2.7). The species in the plasma expansion may be in different ro-vibronical states (ground and electronically excited states may be populated). An iodine calibration spectrum is recorded simultaneously to obtain absolute frequency calibration. Upon ab- sorption of laser light at a specific frequency by the molecular species in the expansion, the species will be (further) ro-vibronically excited. If de-excitation due to fluorescence takes place, the emitted light is detected by a photo-multiplier tube. The fluorescence signal is then recorded with respect to its absorption wavelength. The advantage of LIF spectroscopy over “regular” absorption spectroscopy is the en- hancement of the sensitivity. This is mainly achieved because the experiment is not based on a small decrease of laser intensity upon absorption of light of a certain frequency, but rather on the emission of light upon electronic excitation, i.e., fluorescence. Fluo- rescence is isotropic, which means that fluorescence takes place in all directions and can be monitored perpendicularly to the direction of the laser beam which in addition lowers background noise that can occur from scattering or reflection of the laser beam inside

26 2.4 The Experimental Set-Up - LIF

the experiment chamber. Fluorescence will occur if the molecule has no possibility to relax due to internal conversion, followed by ro-vibrational relaxation back to the ground state. The excess energy is then emitted in form of light after a short moment (typically after a couple of nanoseconds). The disadvantage of LIF spectroscopy with respect to direct absorption spectroscopy – like CRDS – is that not all molecules fluoresce and that only specific molecules can be detected that way. A rather good indication for whether a molecule is likely to fluoresce is given by looking at their absorption spectra. If the absorption spectrum of a certain species shows distinct sharp absorption bands this indi- cates that the lifetime of a species is not too short and internal conversion is not favoured over fluorescence. If the absorption spectra show broadened absorption bands this may indicate that the molecule may have a very short lived excited state. Lifetime broaden- ing occurs as the result of fast relaxation, mainly due to internal conversion, followed by ro-vibrational relaxation back to the ground state. An important issue in this experiment is the timing of the different pulsed parts that have to match in order to obtain spectra with high signal quality. The gas injection via the solenoid pulsed nozzle has been overlaid with a pulsed HV gate. Upon discharging the gas, the molecules are created rotationally and vibrationally excited (ro-vibrationally); in some cases the molecules can also be in an excited electronic state (ro-vibronically excited) in the plasma expansion. Expansion into the vacuum is followed by laser exci- tation of the still (electronically and) vibrationally excited species. Therefore the pulsed laser needs to match timely with the expansion in the vacuum chamber. The detector is time-gated and switches off for the moment when the laser pulse enters the spectroscopy chamber. This is mainly used to prevent the detector from saturating. In addition a broad bandpass filter is mounted in front of the detector which prevents the detector from stray light of the laser beam. Sophisticated light shields have been implemented to decrease the background noise which mainly is due to straylight from the plasma source when dis- charging the acetylene/helium mixture and from scattered photons from the laser. The plasma light was the mayor problem in this experiment since for high discharge powers it could still be detected up to 100 ms after the discharge pulse and thus was still detected by the PMT when recording the fluorescence signal. Active baseline subtraction of the noise from the signal did not yield any better results. Optimization of the system was performed with C2 and resulted in a multi-step trigger scheme which is plotted for completeness in Fig. 2.8. In this experiment the focus is on fluorescence of carbon bearing species that are cre- ated by discharging a mixture of 0.2% acetylene (C2H2) in helium. The pressures are lower than for the CRDS experiment and range typically between 10−3 and 10−4 mbar. In Fig. 2.9 the vibrational transitions of the Swan band system of C2 is presented. The spec- 3 3 trum shows the d Πg → a Πu (0,1) transition (Wehres et al. 2010a). The Swan bands are known to have very strong oscillator strengths and have been recorded in many astronom- ical environments as well as in combustion processes (Bakker et al. 1996, Klochkova et al. 1999, Gredel et al. 1989, Sarre 2006, Glinski et al. 2009, Wehres et al. 2010a). The spec- trum shows the assignments of the rotational transitions and a simulation is overplotted as a dotted line. The simulation yields a rotational temperature of ∼ 12 K. The main goal to look for fluorescence of carbon chains or hydrocarbon species in-

27 Chapter 2 Laboratory and Observational Astrophysics

Figure 2.7 – The figure depicts schematically the laser induced fluorescence (LIF) set-up. Excita- tion spectra of transient species are recorded in high-resolution and compared to spectra of the Red Rectangle proto-planetary nebula. stead of obtaining their spectrum is to compare the laboratory spectra to the astronomical spectra of the Red Rectangle proto-planetary nebula. This object has been described in Chapter 1 and will be described in more detail in Chapters 6 and 7. The importance of this object lies in the fact that it shows a very rich emission spectrum in the visible and since only specific molecules show fluorescence it is a very unique and also very well suited object to restrict the search for molecules of astronomical importance to molecules that indeed show fluorescence.

2.5 Rotational Contour Simulations

The spectra that are obtained in the laboratory, using CRDS and LIF spectroscopy, have the potential to identify species in the laboratory environment and subsequently also in the interstellar medium. The absorption experiment is used to compare the spectra imme- diately to spectra of diffuse interstellar clouds, especially towards HD 183143. In this line of sight many diffuse interstellar bands have been detected and an online catalogue allows

28 2.5 Rotational Contour Simulations

Figure 2.8 – The timing scheme for the LIF experiment is presented. The Master trigger is set to a repetition rate of 10 Hz and triggers the laser and the gas valve. All other instruments have to be matched in time using delay units to guarantee overlap of the gas pulse with the discharge, the laser and the detector gate. for comparison. The spectra recorded in the LIF experiment are used to compare to emis- sion spectra of the Red Rectangle proto-planetary nebula. Rotational contour simulations using PGOPHER (Western 2007) have a two-fold purpose here: First they assist in the identification of the intrinsic molecular parameters by simulating as precisely as possible the spectrum in the laboratory, followed by the identification of the specific transitions. Second, by simulating the spectrum for different resolutions and rotational temperatures we can identify the species and their transitions in the interstellar ∼ or circumstellar en- vironment. Furthermore, the simulation allows for a rough temperature estimate of the identified species in a specific environment and thus assists our understanding of the phys- ical conditions in the interstellar medium. To allow for comparison of the high-resolution laboratory data to the observational (telescope) data obtained at medium resolution, simu- lations are therefore essential to identify the molecule in the ISM. Temperature differences between the laboratory environment and in the interstellar medium can be taken into ac- count and can result in very different spectra. An example of the laboratory experiment and the corresponding simulation is given in Fig. 2.6 which models the rotational contour of the C6H molecule. The simulation at a rotational temperature of ∼ 20 K yields a nice fit to the laboratory spectrum.

29 Chapter 2 Laboratory and Observational Astrophysics

W avelength [Å]

5634 5632 5630 5628 5626 5624 5622

1.4

R P Q

1 1

1

1.2

e(4) e(3) f(4) f(2)e(3) f(2) e(5) f(4) e(5)

e(3)

1.0 R

P Q 2

2 2

f(2)

e(1) e(3)

e(3) e(1) f(2)

f(2)

0.8

R

3

0.6

e(1)

0.4 NormalizedIntensity

0.2

0.0

17745 17750 17755 17760 17765 17770 17775 17780 17785 17790

-1

W avenumber [cm ]

Figure 2.9 – Example of the rotationally resolved C2 spectrum as recorded using laser induced flu- orescence (LIF) spectroscopy is shown and the assignments are given. The experimentally obtained spectrum is overplotted by a simulation (dotted line). The spectrum will be discussed in detail in Chapter 6. The arrows indicate two additional rotational lines, which are not reproduced by the sim- ulation using a temperature of ∼ 12 K, but can be reproduced using higher rotational temperatures in the settings of the simulation.

The simulation for different rotational temperatures and resolutions is an important point, since spectra can vary to a large extent. This is shown in Fig. 2.10. The figure plots three molecules: C2 in the first panel, as a very simple diatomic species. The middle panel plots simulations for the linear hydrocarbon molecule C6H. The third panel shows the simulation for the non-linear carbon chain C9H3. The first trace plots the simulated rotational contour of all species as a stick diagram at 10 K. The second trace shows the same spectrum at a resolution of 0.07 cm−1, as obtained in the laboratory work. In the third trace the simulation is plotted at elevated temperatures of ∼ 30 K and the last trace −1 −1 shows the simulation for 30 K and a degraded resolution of 0.2 cm (for C2 ∼ 0.5 cm ), respectively. Note, for the observations on the Red Rectangle proto-planetary nebula (Chapter 6) the resolution was about 3.5 cm−1 only. The overall contour of the molecular spectra is strongly dependent on the intrinsic molecular characteristics, i.e. the moment of inertia along the principal axis. These mo- ments of inertia (or rotational constants that are the inverse of the moments of inertia) mainly determine the overall shape of the molecular spectra. It can be seen that the spec-

30 2.6 Optical Spectroscopy using the New Technology Telescope

C H C C H

9 3 2 6

stick diagrams at 10 K

-1

10 K R= 0.07 cm

-1

30 K R= 0.07 cm NormalizedIntensity

-1 -1

30 K R= 0.5 cm 30 K R= 0.2 cm

17740 17760 17780 18980 18985 18990 18995 18875 18880 18885

-1

Frequency [cm ]

Figure 2.10 – The simulated spectra for C2,C6H and C9H3 are shown for different resolutions and varying rotational temperatures. In the upper trace, the stick diagrams at 10 K are shown. The second trace gives the spectra at 10 K, with the resolution of the laser in the laboratory setting. The third trace shows the spectra at 30 K and the resolution of the laboratory experiment. The last −1 −1 panel degrades the resolution from 0.07 cm to 0.2 (0.5 cm for C2 respectively) and for rotational temperatures at 30 K.

tra can become very complicated for larger species, and that the rotational levels that are populated even at low temperatures increase rapidly for larger and more complex molecules like C6H and C9H3 (as can be seen in the stick diagrams, respectively).

2.6 Optical Spectroscopy using the New Technology Tele- scope

In February 2008 optical long-slit spectra using the EMMI (ESO multimode instrument) at the New Technology Telescope (NTT) in La Silla, Chile, were obtained. The goal of the observations was to obtain medium resolution data (R=5000 (∼ 3.5 cm−1) and R=2600 (∼ 7.3 cm−1)) of the Red Rectangle proto-planetary nebula at different off-sets from the central star. The wavelength coverage of all spectra is between 5500 and 6800 Å. The slit was positioned at the central star and at 3, 7, 16 and 2000 distance from the central star for

31 Chapter 2 Laboratory and Observational Astrophysics

Figure 2.11 – The slit positions for obtaining the long-slit spectra of the Red Rectangle proto- planetary nebula are depicted schematically. The slits were positioned at a fixed rotator off-set angle of 105◦ east from north, perpendicular to the nebula’s symmetry axis. The slit dimensions are 100 × 20000 and cover the nebula over the entire width. The image was taken from the ESO archive and taken by (Van Winckel et al. 2002). the different observations (Wehres et al. 2010a). The angle of the slit was fixed at 105◦ east from north, perpendicular to the nebula’s symmetry axis and is depicted in Fig. 2.11. The slit dimensions were chosen such that the slit lengths covered the entire width of the nebula to collect the light completely. The dimensions were 100 × 20000. The nebula itself can be traced up to about 6000 away from the central star and the distance of the Red Rectangle is estimated to be about ∼ 710 (Men’shchikov et al. 2002). Its visual magnitude is 9, but decreasing fast further away from the central object. The exposure times thus varied between 60 s at the central star going up to one hour for distances as far as 16 or 2000. The long-slit spectra that are obtained show a two-dimensional spectrum of the neb- ula in Fig. 2.12. On the x-axis the wavelength information is preserved and the y-axis

32 2.7 Optical Spectroscopy using the Mercator Telescope

Figure 2.12 – The two dimensional raw image of the Red Rectangle proto-planetary nebula is shown for a distance of 700 distance to the central star. The x-axis contains the wavelength information (here between ∼ 5500 and 6200 Å). The y-axis contains the spatial information. After collapsing the spectrum into one single point an increase of the S/N ratio can be achieved, resulting in a one- dimensional spectrum as can be seen in Chapter 1 (Fig. 1.5).

contains the spatial information. This results in the advantage of long-slit spectroscopy over single aperture spectroscopy, since the nebula is traced in small steps when posi- tioning the slit along the outflows. The advantage of long-slit spectroscopy over integral field spectroscopy, which also preserves spatial information, is that the S/N ratio can be increased for faint objects by collapsing the two dimensional slit spectrum into a single pixel row. That way the spatial information is still preserved, but the signal is amplified. A one-dimensional spectrum of the Red Rectangle nebula is plotted in Chapter 1 (Fig. 1.5) and the spectra will be discussed in more detail in Chapters 6 and 7. Data reduction was performed using the MIDAS software package and is described in more detail in Chapters 6 and 7 as well. The spectra that were recorded here, have been compared to and combined with spec- tra taken by Van Winckel et al. (2002) and resulted in a set of spectra at the central star as well as at 3, 6, 7, 11, 14, 16 and 2000 distance to the central star. The aim of this extensive data-set is to establish an archive of the spectral changes with distance to the central star. Identification of the species in the nebula and tracing the changes of the spectral shape with respect to the distance of the central source can then give information of the physical conditions and about the ongoing chemistry in that object at a given distance.

2.7 Optical Spectroscopy using the Mercator Telescope

In Summer 2009 a high resolution optical spectrum of the diffuse cloud towards HD 183143 has been obtained using the Mercator telescope on La Palma, Spain. Very recently the HERMES (high efficiency and resolution Mercator echelle spectrograph) had been in-

33 Chapter 2 Laboratory and Observational Astrophysics stalled (Raskin & Van Winckel 2008) which allowed for a resolution of R' 100,000 (0.02 cm−1). The 1.2 m telescope is operated by the Katholieke Universiteit Leuven, Belgium, and the Observatory in Geneva, Switzerland. The fiber-fed cross dispersed spectrograph has a fixed spectral format and the spectrum that was obtained covered the whole wavelength between ∼ 3800 and 9300 Å. The exposure times of the diffuse cloud spectra averaged to 1200 s each. A reference spectrum of the star HD 164353 was ob- tained to discriminate between stellar lines and absorption bands in the diffuse interstellar cloud. The spectrum was reduced using the Mercator pipeline as will be described in more detail in Chapter 3. Spectra of diffuse interstellar clouds are discussed in Chapters 3-5 and are used to compare to the laboratory absorption spectra using cavity ring-down spectroscopy. The aim is to identify the carriers causing the absorption bands as seen in these clouds and to constrain the chemistry and physics going on in these environments.

34 A Coincidence between a Hydrocarbon Plasma Absorption Spectrum and the3λ5450 DIB

H. Linnartz, N. Wehres, H. Van Winckel, G. A. H. Walker, D. A. Bohlender, A. G. G. M. Tielens, T. Motylewski and J. P. Maier

A&A, 2010, 511, L3

35 Chapter 3 A Coincidence between a Hydrocarbon Spectrum and the λ5450 DIB

Abstract

Aims. The aim of this work is to link the broad λ5450 diffuse interstellar band (DIB) to a laboratory spectrum recorded through an expanding acetylene plasma. Methods. Cavity ring-down direct absorption spectra and astronomical observations of HD 183143 with the HERMES spectrograph on the Mercator Telescope in La Palma and the McKellar spectrograph on the DAO 1.2 m Telescope are compared. Results. In the 543-547 nm region a broad band is measured with a band maximum at 545 nm and FWHM of 1.03(0.1) nm coinciding with a well-known diffuse interstellar band at λ5450 with an FWHM of 0.953 nm. Conclusions. A coincidence is found between the laboratory and the two independent observational studies obtained at higher spectral resolution. This result is important, as a match between a laboratory spectrum and a – potentially lifetime broadened – DIB is found. A series of additional experiments were performed in order to unambiguously identify the laboratory carrier of this band, but this was not successful. The laboratory results, however, restrict the carrier to a molecular transient, consisting of carbon and hydrogen.

3.1 Introduction

Diffuse interstellar bands are absorption features observed in starlight crossing diffuse interstellar clouds. Since their discovery in the beginning of the 20th century, scientists have been puzzled by the origin of these bands that appear both as relatively narrow and rather broad bands covering the UV/VIS and NIR (Tielens & Snow 1995). In the last de- cennia, the idea has been established that it is unlikely that all these bands originate from one or a very few carriers, and with the progress of optical laboratory techniques, sev- eral families of potential carriers have been investigated. It was shown that the electronic transitions of a series of PAH-cations do not match the listed DIBs (Salama et al. 1996, 1999, Bréchignac & Pino 1999, Ruiterkamp et al. 2002). Similarly, systematic laboratory studies of electronic spectra of carbon chain radicals have not resulted in positive identifi- cations either (Motylewski et al. 2000, Ball et al. 2000a, Jochnowitz & Maier 2008), even though it is known from combined radio-astronomical and Fourier transform microwave (FTMW) studies that many of such species are present in dense clouds (Thaddeus & Mc- Carthy 2001). Only C3 has been recorded unambiguously in diffuse interstellar clouds (Maier et al. 2001). Other studies, focusing on multi-photon excitation in molecular hydrogen (Sorokin et al. 1998), or spectra of fullerenes and nano-tubes (Kroto & Jura 1992, Foing & Ehren- freund 1994) have been unsuccessful as well. In the past years, several coincidences between laboratory and astronomical DIB studies have been reported in the literature. These have all turned out to be accidental, and from a statistical point of view, the chance of an overlap is also quite substantial, because DIBs cover a major part of the wavelength region between roughly 350 and 1000 nm. However, there are several conditions that have to be fulfilled before any coincidence of a laboratory and an astronomical DIB spectrum

36 3.1 Introduction

543.0 543.5 544.0 544.5 545.0 545.5 546.0 546.5 547.0

Simulation

HERMES spectra

Intensity [a.u.] Intensity

McKellar spectra

543.0 543.5 544.0 544.5 545.0 545.5 546.0 546.5 547.0

W avelength [nm]

Figure 3.1 – The λ5450 DIB. The top spectrum is a simulated spectrum available from DIB catalogues. The middle and bottom figures show observational spectra from the HERMES and McKellar spectrograph, respectively. The HERMES spectra show the λ5450 DIB recorded towards HD 183143 and towards a reference star (HD 164353). The McKellar spectra show a reference spectrum toward Rigel (top), the DIB spectrum also towards HD 183143 (bottom), and the corre- sponding spectrum (middle) in which the SII stellar line has been deblended. may be interpreted as a real match. These conditions have become stricter with the re- cent improvement in achievable spectral resolution, both in laboratory and astronomical studies. The two most important DIB matching criteria to link laboratory and astronomical data are the following:

1. The gas-phase laboratory and observational values of both peak position and band- width of the origin band transition should be identical, unless it can be argued that a spectral shift or band profile change may come from an isotope or temperature effect. An example of the latter is given by spectroscopic measurements on ben- zene plasma yielding an absorption feature coinciding with the strongest DIB at 442.9 nm (Ball et al. 2000b, Araki et al. 2004). The laboratory FWHM turned out to be narrower than in the astronomical spectrum. It was argued that the spectrum of a non-polar molecule cooled in a molecular expansion may be considerably colder

37 Chapter 3 A Coincidence between a Hydrocarbon Spectrum and the λ5450 DIB

than in space, where only radiative cooling applies. A similar discussion has been given by (Motylewski et al. 2000), who show that unresolved rotational profiles may change substantially for different temperatures, as has also been calculated and discussed by Cossart-Magos & Leach (1990).

2. Once the origin band overlaps with a DIB feature, gas-phase transitions to vibra- tionally excited levels in the electronically excited state of the same carrier molecule should match as well, and the resulting band profiles should behave in a similar way (i.e. with comparable equivalent width ratios) (Motylewski et al. 2000). A good ex- − ample for this is the electronic spectrum of C7 that has been regarded for several years as a potential carrier, because subsequent electronic bands fulfilled both con- ditions (Tulej et al. 1998, Kirkwood et al. 1998). Detailed follow-up studies show that the series of (near) matches was coincidental (McCall et al. 2001).

Despite much progress both from the observational and laboratory side, all efforts to as- sign DIBs have in the end resulted in a rather static situation, thereby triggering more and more exotic explanations for DIB carriers, and the origin of the DIBs is still as mysterious as it was nearly 100 years ago. In this letter we report a match of a laboratory spectrum with a diffuse interstellar band that is special, because the first condition is fulfilled for a rather broad and potentially life- time broadened DIB; i.e., the laboratory and astronomical spectra should be fully iden- tical, independent of temperature restrictions. New astronomical observations obtained with the Mercator telescope, using the HERMES spectrograph and the Dominion Astro- physical Observatory (DOA) 1.2 meter telescope, using the McKellar spectrograph, are presented in order to characterize the band profile of the λ5450 DIB with the best possible resolution. Even though we have not been able to unambiguously identify the laboratory carrier, which is most likely a smaller hydrocarbon bearing molecular transient, we think that this overlap is important to report, since it provides a new piece in the puzzle.

3.2 Cavity Ring-Down Spectroscopy

The experimental set-up is described in Linnartz et al. (1998) and Motylewski & Lin- nartz (1999), and has been extensively used to study many carbon chain radicals of astro- physical interest (Jochnowitz & Maier 2008). The monochromatic output, ∼ 0.1 cm−1 at 540 nm (∼ 18,500 cm−1), of a pulsed dye laser-based cavity ring-down set-up is focused into an optical cavity consisting of two highly reflective mirrors (R > 0.9999). A special, pulsed high-pressure slit-nozzle system capable of producing intense 300 µs long plasma pulses by discharging (-1 kV, 100 mA) an expanding gas mixture of 1 % acetylene (C2H2) in He is mounted inside the cavity with its slit parallel to the optical axis of the cavity. In the expansion, a wide variety of new species is formed, and as the technique is not mass selective, special care has to be taken when assigning bands to specific carriers.

38 3.3 Optical Observations

Mass selective matrix isolation spectra offer a good starting point for an assignment (Jochnowitz & Maier 2008). In the case of rotationally resolved spectra, unambiguous identifications are generally possible, either by combination differences of accurate spec- tral fits, or by isotopic studies using C2D2 instead of C2H2 (or a mixture of C2H2/C2D2). The source runs at 30 Hz, and special care is taken that the pressure inside the cavity remains constant during jet operation to reduce baseline fluctuations. Rotational tempera- tures are typically Trot ∼ 10–20 K. This low temperature results in a spectral simplification and simultaneously increases the detection sensitivity because of improved state density. In addition, the source offers a Doppler free environment with a relatively long effec- tive absorption path length. The laser beam intersects the 3 cm long planar expansion about 5-10 mm downstream using a sophisticated trigger scheme. Subsequent ring-down events (typically 20-30 µs for a 52 cm long cavity) are recorded as a function of the laser frequency by a photodiode and transferred to an averaged ring-down time by fitting 45 subsequent ring-down events. This value as function of the laser wavelength provides a sensitive way to record optical spectra. An absolute frequency calibration is obtained by recording an I2 reference spectrum simultaneously.

3.3 Optical Observations

The laboratory data are compared to observations from two different astronomical facili- ties.

3.3.1 HERMES @ Mercator Telescope The HERMES observations were carried out in service mode using the Mercator tele- scope at Roque de los Muchachos Observatory on La Palma. The 1.2 m telescope is operated by the Katholieke Universiteit in Leuven, Belgium, in collaboration with the Observatory in Geneva, Switzerland. The spectra were obtained in June 2009 with HER- MES (High Efficiency and Resolution Mercator Echelle Spectrograph) (Raskin & Van Winckel 2008), which is a fibre-fed-cross-dispersed spectrograph. The spectrograph has a fixed spectral format and samples the spectrum between 377 and 990 nm in 55 spectral orders on a 4.6 k x 2 k CCD. The spectral resolution is slightly variable over the field, but is 85,000 on average. We obtained 3 spectra of 1,200 s of HD 183143 (B7Ia, m(v)=6.92, B-V=+1.001), the DIB spectral standard with a reddening E(B-V) close to 1.0. The ref- erence star HD 164353 (B5Ib, m(v)=3.97, B-V=−0.002) was sampled in 3 exposures of 1 min. The spectral reduction was performed using the specifically coded HERMES pipeline, and it contains all the standard steps in spectral reduction. The wavelength cal- ibration is based on spectra of ThAr and Ne lamps. As we are mostly interested in the broad absorption feature that is centred around 545 nm, we focus further on this spectral region of HD 183143. The spectra are shown in Fig. 3.1 (middle rows) and compared to the λ5450 DIB profile as available from a series of digital DIB catalogues (Herbig 1975, Jenniskens & Desert 1994, Tuairisg et al. 2000, Galazutdinov et al. 2000) in the upper row.

39 Chapter 3 A Coincidence between a Hydrocarbon Spectrum and the λ5450 DIB

3.3.2 McKellar @ DAO Telescope

Fifty-five half-hour spectra were taken with the McKellar Spectrograph and SITe-4 CCD at the DAO 1.2 m telescope, operated by the National Research Council of Canada, over 6 nights between 16 and 23 July 2006 (UT) at a dispersion of 10.1 Å/mm giving 0.151 Å/pixel for a resolution ∼ 0.3 Å. The data was processed in a standard fashion using IRAF 1. The aggregate spectrum had a signal-to-noise ratio of about 1200/pixel before correction of telluric lines. Removal of the quite weak telluric features was per- formed conventionally with spectra (S/N ∼1600) of the A0 V star zeta Aql (HD 177724) as the template. Rigel, an unreddened comparison star with a B8 Ia spectral type very similar to the B7 Ia of HD 183143 was also observed to identify photospheric lines that contaminate the interstellar features observed in the latter star. The sharp line at approximately 5454 Å arises from SII and was removed from the spectrum of HD 183143 by simply fitting a Voigt profile to the line and subtracting this from the original spectrum. The final “de- blended” spectrum is plotted as a comparison in Fig. 3.1 (lower panel, middle spectrum).

3.4 Results

In Fig. 3.2 several spectra in the 543-547 nm region are compared. The top spectrum is the digital DIB spectrum of the λ5450 DIB (Herbig 1975, Jenniskens & Desert 1994, Tuairisg et al. 2000, Galazutdinov et al. 2000). The spectrum in the middle is a zoom- in on the deblended McKellar spectrum as shown in Fig. 3.1. The bottom spectrum is the laboratory spectrum recorded in direct absorption through an expanding 1% C2H2/He plasma. The similarity between the three spectra is striking. 1Π 1Σ+ This wavelength region was initially scanned to search for the u –X g electronic − origin band spectrum of the linear carbon chain radical C7 (following the C7 DIB discus- sion) that was located in matrix isolation experiments around 542.3 nm. The laboratory spectrum, shown in Fig. 3.2, consists of many narrow lines that come from small acety- lene fragments (typically C2 and CH) that get weaker when the distance from the nozzle orifice to the optical axis is increased, but there is clearly a broad feature lying under- neath. As this band shifts by 1.5 nm to the red upon C2D2 precursor substitution, it was initially neglected, because both C2H2 and C2D2 should result in an identical spectrum for C7. The shift is illustrated in Fig. 3.3. In addition, the deuterated spectrum appears to be somewhat stronger. Despite this negative result for C7, the profile hiding under the narrow lines in the C2H2 precursor experiment perfectly matches the λ5450 DIB available from the DIB databases, which is one reason additional observations were performed.

1 IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation.

40 3.5 Discussion

Figure 3.2 – The top spectrum shows the digital λ5450 DIB, the middle spectrum shows the de- blended McKellar data and the bottom spectrum shows the laboratory cavity ring-down absorption spectrum through a supersonically expanding acetylene plasma.

3.5 Discussion

There is little discussion possible about the coincidence between the recorded laboratory spectrum and the λ5450 DIB. Both bands have a central peak position of 545 nm and a FWHM of 1.03 (0.1) nm (laboratory spectrum) and 0.953 nm (observational spectrum) (Tuairisg et al. 2000). The uncertainty in the first value comes from the overlap of the many individual transitions, which prohibits a clear view of the broad feature. The ques- tion is more whether this actually represents a DIB match, and for this complementary information is needed. Additional laboratory work was performed, where it should be noted that the scans shown in Figs. 3.2 and 3.3 typically last 45 minutes to an hour, in order to achieve the required sensitivity and to cover a frequency domain broad enough to differenciate band profile and baseline; i.e., fast optimizations are impossible. The laboratory band does not show any structure that can be related to unresolved P, Q, and R-branches. With 1.03 (0.1) nm, the band is also much broader than the unresolved rota- tional profile of a larger carbon chain radical. For comparison, at 15 K, the band profile of the linear C6H radical (at 525 nm) is about five times narrower (Linnartz et al. 1999). It should also be noted that such a broad feature actually represents a large absorption

41 Chapter 3 A Coincidence between a Hydrocarbon Spectrum and the λ5450 DIB

543 544 545 546 547

C H /He

2 2

C D /He

2 2 Intensity [a.u] Intensity

543 544 545 546 547

W avelength [nm]

Figure 3.3 – Comparison between laboratory experiments sampling expanding plasma using regular acetylene (top) and deuterated acetylene (bottom) as a precursor gas. compared to many of the sharper DIBs. Changing the experimental settings to vary the final temperature in the expansion by measuring close (∼ 50 K “warm”) and far (∼ 10 K “cold”) downstream does not substantially change the FWHM of the spectral contour. As the narrow overlapping transitions have FWHMs close to the laser bandwidth, ex- perimental broadening artifacts, such as residual Doppler broadening in the expansion or amplified spontaneous emission, can be excluded. It is clear that the band profile is caused by a temperature-independent and carrier-specific broadening effect, presumably lifetime broadening. The observed bandwidth of 1.0 nm (∼ 35 cm−1 around 545 nm) corresponds to a lifetime of roughly 0.15 ps. The bandwidth profile does not allow conclusions on the nature of the laboratory carrier. The carrier must be a transient species (a molecular radical, a cation or anion, a weakly bound radical complex, possibly charged, or a vi- brationally or electronically excited species) because no comparable spectra are recorded without plasma (i.e. with a regular C2H2/He expansion). The use of a C2D2/He expan- sion results in a red–shifted spectrum (Fig. 3.3), and from this it can be concluded that the laboratory carrier must contain both carbon and hydrogen. To check whether there are equivalent H-atoms in this carrier, a C2H2/C2D2 1:1 mixture in He has been used as an expansion gas, but this only results in a very broad absorption feature covering the whole region where results are found for a pure C2H2 and a pure C2D2 expansion. It is impos-

42 3.5 Discussion sible to conclude anything about the actual number of equivalent H-atoms in the carrier + by determining the number of bands that show up, as could be demonstrated for HC6H or HC7H (Sinclair et al. 1999, Ball et al. 2000a, Khoroshev et al. 2004). Also the use of another precursor (e.g. allene) did not provide conclusive information. Additional experiments have been performed. The 543-545 nm region has been scanned using a two-photon REMPI-TOF experiment with the aim determining the mass of the carrier (Pino et al. 2001). No spectrum could be recorded, which may be related to the short lifetime of the excited state or with the fact that the carrier is an ion. Ions are in- deed formed in this planar plasma source (Witkowicz et al. 2004). Both smaller and larger species have been observed, with optimum production rates depending, among other things, on the backing pressure. The production of larger species is generally more crit- ical; e.g., higher backing pressures are needed, but this also may destabilize the plasma, which is unfortunate, particularly during long scan procedures. More complex species are generally found further downstream, but in this specific case, we did not observe large differences as a function of the distance from the laser beam to the nozzle orifice. This is the typical behaviour for a smaller constituent in the gas expansion. We tried to systemati- cally study the voltage dependence of the signal. For a positive ion, an increase in voltage should go along with a decrease in signal for distances further downstream, as the jaws carry a negative voltage. It is the opposite for anions, but 10 years of experience with this source have shown that negative ions are rather hard to produce. Again, the changes we recorded were small and did not allow drawing hard conclusions. Following condition 2 mentioned in the introduction, we also searched in other wavelength regions blue-shifted by values typical for an excited C–C, C=C, C≡C, or CH stretch in the upper electronic state. Such excited bands have not been observed here, but it should be noted that these bands can be intrinsically weak. In summary, we are left with a laboratory spectrum that coincides both in band maximum and band width with a known DIB band at 545 nm. Our measurements show that the ab- sorption spectrum of a transient molecule containing hydrogen and carbon reproduces the astronomical spectrum. The profile can be explained with life time broadening, and this is consistent with the observation that the laboratory and astronomical spectrum are identi- cal; i.e., with no temperature constraints. In addition, it explains why the large bandwidth of this DIB does not vary along different lines of sight. The large effective absorption may also be indicative of an abundant carrier. The exact carrier, as such, remains an open question. The present result, however, may be useful for stimulating upcoming DIB work.

43

3Σ− 3Σ− Rotationally Resolved A u –X g Spectrum of4 HC7H

N. Wehres, D. Zhao, W. Ubachs, H. Linnartz

Chem. Phys. Lett. 497 (2010) 30

45 3Σ− 3Σ− Chapter 4 Rotationally Resolved A u –X g Spectrum of HC7H

Abstract

3Σ− 3Σ− Context. The A u –X g electronic spectrum of the linear carbon chain radical HC7H has been recorded fully rotationally resolved. Cavity ring-down spectroscopy is used to record the origin band transition in direct absorption through a supersonically expanding planar plasma, discharging a diluted gas mixture of acetylene in helium. Rotational reso- lution is obtained by operating a commercial pulsed dye laser system in a second grating order configuration, resulting in a narrower bandwidth of about 0.035 cm−1. In total, 39 resolved P- and R-branch transitions are included in a standard fit, yielding for the first time accurate rotational constants for both electronic states.

4.1 Introduction

Highly unsaturated carbon chain radicals have attracted much interest in the last years. Following systematic Fourier transform microwave laboratory studies (McCarthy & Thad- deus 2001) several carbon chains have been identified in dense interstellar clouds, varying from cyanopolyynes HCnN as large as HC11N (Bell et al. 1997) to carbon chain cations + − − (e.g. HCCCNH (Gottlieb et al. 2000)) and anions (e.g. C6H and C8H (McCarthy et al. 2006, Brünken et al. 2007)). Matrix isolation studies (Maier 1998), in addition, have guided optical gas phase surveys to record electronic transitions of pure carbon chains Cn (+/−) and mixed XCnY species with X,Y = H, N, O and S containing chains, using sensitive spectroscopic detection techniques, such as REMPI-TOF, photo-detachment, cavity ring- down and plasma-frequency double modulation schemes (see Jochnowitz & Maier (2008) and Linnartz (2009) for a review). The resulting spectra have been used to compare optical spectra of carbon chains with diffuse interstellar band (DIB) positions, absorption features in the starlight crossing diffuse interstellar clouds, but so far without unambiguous identi- fications (Motylewski et al. 2000, Linnartz et al. 2010). One carbon chain radical that has been studied in the past is the linear triplet chain HC7H. This is a member of the HC2n+1H 3Σ− 3Σ− = series for which the electronic u – X g bands have been recorded for n 2-7 in a 5 K Ne matrix (Fulara et al. 1995). The matrix bands are broadened and their absorption fre- quencies are typically shifted by as much as 100 cm−1 compared to the gas phase – the reason why these data cannot be used to compare directly with DIB positions - but as such they are indicative for where to start searching in the gas phase. Using the available matrix data, electronic spectra of HC7H, HC9H, HC11H and HC13H have been recorded in the gas phase (Ball et al. 1999, 2000a, Vaizert et al. 2001). In the case of HC7H the matrix shift for the origin band turned out to be only 7 cm−1. A clear P- and R-branch structure as typical for a Σ–Σ transition was found in Refs. Ball et al. (1999, 2000a), but individual transitions could not be fully resolved. As no fluorescence signals were found, a radiationless lifetime broadening of the upper state of 100 ps or longer was proposed to explain the lack of rotational resolution. A conclusive identification of the spectrum, though, was possible by also recording spectra of HC7D and DC7D and comparing the isotopic shifts to those found in the matrix work. In Ref. Vaizert et al. (2001) a better resolved spectrum was obtained, but only a contour fit was presented based upon a theo-

46 4.2 Experimental

retically predicted ground state B-value (taking B0 ≡ Be). The present study extends the 3Σ− 3Σ− spectroscopic characterization of the A u –X g origin band transition of HC7H and pro- vides a full rotational analysis. The use of a planar plasma source (as in Ref. Vaizert et al. (2001)), providing an essentially Doppler-free environment, instead of a pinhole nozzle, and an improvement in laser bandwidth by operating a commercial pulsed dye laser sys- tem in a higher grating order, allows for recording nearly forty subsequent transitions in 3Σ− 3Σ− the A u – X g origin band spectrum around 504.5 nm.

4.2 Experimental

The experiment is based on a pulsed cavity ring-down detection scheme, monitoring ro- tationally cold hydrocarbon radicals in a supersonically expanding plasma by discharging a 1% C2H2 / He mixture. This method is well established and essentially identical to that described in detail in Linnartz (2009) and Motylewski & Linnartz (1999) and it is used in the present study in Amsterdam (see also Witkowicz et al. (2004)). More specifically, the present experiment comprises a 3 cm × 100 µm slit nozzle consisting of two metal jaws that act as cathode and a slotted metal anode, separated by a slotted ceramic plate. The nozzle body is mounted to a General Valve inside a vacuum chamber that is pumped by a roots blower system with 1000 m3/hr pumping capacity. The typical pressure in the chamber is 0.1 mbar during jet operation. Backing pressures are as high as 12 bar. The nozzle is mounted with its slit orifice parallel to the optical axis of a 50 cm long opti- cal cavity, off-set by a few mm. This cavity consists of two plano-concave mirrors with an average reflection coefficient R better than 0.9999 that are mounted on high precision alignment tools. The light leaking out of the cavity is detected with a photomultiplier tube and typical ring-down times amount to 12 – 15 µs. A multi trigger scheme is used to timely overlap ring-down event, discharge pulse (300 µs, ∼ 100 mA, - 750 V) and gas pulse (∼ 1 ms). The setup is operated at 10 Hz, which is determined by the maximum repetition rate of a tripled Nd:YAG laser that is used to pump a dye laser system (Sirah, Cobra-Stretch) with Coumarine 307 as laser dye, covering the 490 – 520 nm region. This commercial laser system is standard configured with a double grating resonator and has a typical best bandwidth in this wavelength range of 0.07 cm−1, when operated as described in the manual. A relatively simple trick allows substantial improvement of the achievable laser resolution. For this the 2nd order diffraction of the Littrow grating is used (instead of the first order), yielding a laser bandwidth improvement with a factor of two. I2 reference spectra are used to determine the laser frequency and an etalon with a free spectral range of 2.1 GHz is used for linearization of the wavelength scan. This yields an absolute laser frequency precision better than 0.02 cm−1.

4.3 Results and Discussion

3Σ− 3Σ− In Figure 4.1 (upper trace) the rotationally resolved A u – X g electronic origin band spectrum of HC7H is shown as measured here. As in the previous studies by Ball et al.

47 3Σ− 3Σ− Chapter 4 Rotationally Resolved A u –X g Spectrum of HC7H

3Σ− 3Σ− Table 4.1 – Observed rotationally resolved transitions of the A u – X g electronic origin band transition of HC7H.

N00 P-branch transitions R-branch transitions Obs. [cm−1] obs-calca [10−3] Obs. [cm−1] obs -calca [10−3] 2 19817.778 -1 19818.058 -2 4 19817.661 -3 10818.160 -12 6 19817.547 -2 19818.276 -7 8 19817.433 -1 19818.390 -2 10 19817.316 -2 19818.505 5 12 19817.203 3 19818.614 8 14 19817.09b 7 19818.712 -1 16 19816.97b 7 19818.818 3 18 19816.854 12 19818.925 9 20 19816.720 -1 19819.024 8 22 19816.593 -4 19819.121 9 24 19816.469 -4 19819.204 -2 26 19816.33b -15 19819.289 -8 28 19816.22b -3 19819.373 -12 30 19816.092 5 19819.457 -10 32 19815.94b -9 19819.539 -7 34 19815.82b -1 19819.636 4 36 19815.67b -8 19819.701 11 38 19815.53b -2 19819.755 -3 40 19819.809 -2

a The observed - calculated (obs - calc) values are derived using the constants listed in Table 4.2. b Blended transitions. A weight factor including a roughly three times lower accuracy for the frequency determination was used in the fitting routine.

48 4.3 Results and Discussion

P(N) R(N)

26 26 2 10 22 38 30 22 14 2 6 14 18 34 18 10 6

*

*

*

Experimental

*

*

Intensity [a.u.] Intensity

Simulated

19815 19816 19817 19818 19819 19820 19821

-1

Frequency [cm ]

3Σ+ 3Σ− Figure 4.1 – The rotationally resolved A u –X g electronic transition of HC7H recorded by cavity ring-down spectroscopy through expanding planar plasma by discharging a high pressure mixture of 1 % C2H2 in He (upper trace). Simulated spectra are shown in the lower trace for a linewidth of 0.04 cm−1 and a rotational temperature of 18 K. Irregularities, marked by an asterisk, are due to overlapping transitions, presumably C2.

(1999) and by Ball et al. (2000a), the P- and R-branch contour can be easily resolved. A Q-branch is not present. Irregularities, mainly in the P-branch, are due to overlapping transitions around 19816.2, 19816.7, 19817.0, and 19817.1 cm−1, that are most likely due to C2 Swan band transitions (Lloyd & Ewart 1999, Prasad & Bernath 1994). The com- bined application of a slit, i.e., Doppler free expansion and a narrower band width allows the identification of individual rotational transitions, also as every second transition is hard to observe. HC7H is a linear centro-symmetric molecule with nuclear statistical weights 1:3 for symmetric and asymmetric rotational levels, respectively, and as a consequence, mainly transitions starting from asymmetric levels are observable, resulting in an effective line splitting of the order of 4B, instead of 2B. In addition, each rotational level in both the ground and excited 3Σ state is split into three fine structure levels that are defined by the total angular momentum J with J = N+1, N and N−1 (except for N=0 where J=1). This spin structure is not resolved but determines, to some extent, the spectral appearance of individual transitions, particularly in the band origin region, as the spin-spin fine structure is most prominent in the lowest rotational states. In total 39 transitions, equally divided

49 3Σ− 3Σ− Chapter 4 Rotationally Resolved A u –X g Spectrum of HC7H

3Σ− 3Σ− Table 4.2 – Derived constants of the A u –X g electronic transition of HC7H and a comparison −1 with the iso-electronic species HC6N and NC5N. All values are in [cm ].

a b HC7H HC6N NC5N c ν00matrix 19824 21181 22737.3 ν00gas 19817.892(2) 21208.60(5) 22832.7(1) Matrix shift 6 28 95 B000 0.0283263(48) 0.02806299(2) 0.02799(4) −7 D000 (10 ) 2.217(39) – – B00 0.0282298(46) 0.02792(5) 0.02778(3) −7 D00 (10 ) 2.812(36) – – 00 B0 0 1.0034 1.0051 1.0076 B0 a (Vaizert et al. 2001) b (Linnartz 2001) c (Fulara et al. 1995) over P-branch (19) and R-branch (20) are resolved and the transition frequencies are listed in Table 1. In both, the figure and the table, the rotational assignment is indicated. All transitions were fitted using PGOPHER (Western 2007) and a standard Hamiltonian for a 3 3 Σ – Σ transition, with the band origin v00, rotational and centrifugal distortion constants, 00 00 0 0 B0 ,D0 and B0 and D0, in ground and excited state as parameters. Inclusion of distortion effects is essential to decrease the overall error of the fit that amounts to ∼ 0.01 cm−1, i.e., well below the improved bandwidth of the laser. The resulting constants are listed in Table 4.2 and the resulting observed-calculated values are given in Table 4.1. In Figure 4.1 the simulated spectrum is shown for these constants for a line width of 0.04 cm−1 (lower trace). The observed spectrum resembles that of the iso-electronic cyanopolyyne HC6N (Vaizert et al. 2001) and dicyanopolyyne, C5N2 (Linnartz 2001). Both these molecules have been studied previously in a similar way, also combining cav- ity ring-down spectroscopy and supersonic planar plasma expansions. The origin band transitions of these bands are close to that for HC7H, as one may expect for chains of comparable length. Also the other constants are of the same order and are listed for com- parison in Table 4.2. A good estimate for the ground state rotational constant is possible from theoretical 00 = −1 calculations (Aoki & Ikuta 1994), yielding Be 0.0279 cm . We performed a new calculation at the B3lyp/6-311++G∗∗ level using the Gaussian 98 software (Frisch et al. 1998), yielding an improved value of B00 = 0.02816 cm−1. This value is close to the e 00 00 −1 B0 ground state value B = 0.0283263 cm , as derived here. The 0 ratio amounts to 0 B0 1.0034 and reflects a small lengthening of the chain upon electronic excitation. This value is slightly lower than the values found for HC6N (1.0051) and NC5N (1.0076). The (fitted) band origin is located at 19817.892(1) cm−1, which is about 0.1 cm−1 (0.03 cm−1) red shifted from the value reported in Ball et al. (2000a) and Vaizert et al. (2001). A precise fit of the overall band contour, particularly of the band gap area, also re-

50 4.3 Results and Discussion

Intensity [a.u.] Intensity

19817.25 19817.50 19817.75 19818.00 19818.25 19818.50

-1

Frequency [cm ]

3Σ− 3Σ− Figure 4.2 – The band origin of the A u –X g electronic origin band spectrum of HC7H (upper trace) with simulated stick diagram (lower trace). The stick diagram, in the band gap region de- pending on the choice of the λ-constants, demonstrates the spectral fading when rotational lines and spin-spin splittings do not coincide.

quires knowledge of the spin-spin coupling constants in both ground and excited state. As shown in Fig. 4.2, each resolved peak consists of at least three close lying compo- nents with a quantum labeling that strongly depends on the spin-spin interaction. As the final resolution does not allow resolving irregularities because of interference of triplet splittings, this is hard to realize. From the fit a difference between the spin-spin constants for the ground and excited state – ∆λ = λ0 – λ00 – of about 0.27 cm−1 can be estimated. The actual values are more difficult to determine. In previous papers (see Gordon et al. (2000)), the ground state spin-spin constant was adapted to that of the iso-electronic HC6N (as available from FTMW work): λ00 = 0.36 cm−1. However, with this value it is not pos- sible to reproduce the present HC7H band profile; an empirical contour fit actually yields λ00 -values that are closer to 10 cm−1 than to 1 cm−1. Within this constraint, assuming a final rotational temperature of 18 K and a Gaussian linewidth of 0.04 cm−1, the band contour can be well reproduced (see Fig. 4.1). The latter value is interesting, as it puts an additional limit on the lifetime of the excited state. In Ball et al. (2000a) additional line broadening (0.04 cm−1) was found in excess of that (0.066 cm−1) expected from the laser linewidth and Doppler width in a pinhole expansion. A rapid radiationless transi-

51 3Σ− 3Σ− Chapter 4 Rotationally Resolved A u –X g Spectrum of HC7H tion from the upper state was taken to derive a lower boundary for the lifetime τ of the order of 0.1 ns. The absence of fluorescence signals yielded an upper limit, resulting in 0.1 ns < τ < 1 ns. The present study with FWHMs of the order of 0.035 cm−1 narrows down this range: 0.6 ns < τ < 1 ns. Finally, the band studied here has been compared with the newest (and substantially 3Σ− 3Σ− extended) list with DIBs (Hobbs et al. 2009). The region around the A u –X g elec- tronic origin band spectrum (504.5 nm) is remarkably empty in the astronomical spectrum and HC7H does not appear to be a good candidate for a carrier of a DIB, even not in the extended list.

52 Electronic Spectra and Molecular Geometry of the non-linear Carbon Chain C95H3

D. Zhao, N. Wehres, H. Linnartz, W. Ubachs

Chem. Phys. Lett. 501 (2011) 232

53 Chapter 5 Geometry of the non-linear Carbon Chain C9H3

Abstract

Context. Two electronic bands at ∼18881 and ∼18920 cm−1 – previously assigned to the carbon chain molecule C9H3 – have been recorded, resolving for the first time their K- stack structure. The C9H3 radicals are produced by discharging and expanding a diluted gas mixture of acetylene in helium employing a pulsed pinhole nozzle. Cavity ring-down spectroscopy is used to record spectra in direct absorption. The improved experimental data and spectrum simulations based on new theoretical structure predictions show that the HC4(CH)C4H isomer (with C2v symmetry) is a likely carrier of the two observed C9H3 bands.

5.1 Introduction

In the last three decades many studies have been dedicated to the formation, spectral (+/−) characterization and identification of carbon chain species of the form XmCnYy , with typically X,Y = H, N, O and S. These unsaturated species are important constituents in plasma environments, and have been found to be well represented in the interstellar medium (see Thaddeus et al. (1998), Tielens (2005), Wakelam et al. (2010) and refer- ences cited therein). Carbon chains as large as HC11N and carbon chain cations and anions have been identified in dense interstellar clouds by comparing radio astronomical and microwave laboratory spectra (Bell et al. 1997, Brünken et al. 2007). Rovibrationally resolved infrared spectra have been recorded in the gas phase particularly for pure and bare carbon chains, Cn, with C13 the largest system studied so far (Giesen et al. 1994). Optical gas phase spectra of carbon chains have been obtained using a range of experi- mental techniques, including REMPI-TOF (resonance enhanced multi-photon ionisation - time-of-flight), photo-detachment, degenerate 4-wave mixing, and ion trap schemes (Jochnowitz & Maier 2008) as well as cavity ring-down and plasma-frequency double modulation spectroscopy (see also Jochnowitz & Maier (2008), Linnartz (2009)). The resulting spectra have provided insight in the electronic nature of these species and have been used for comparisons with optical absorption spectra observed through translucent interstellar clouds to search for potential carriers of the so called diffuse interstellar bands (Linnartz 2009, Tielens & Snow 1995, Jochnowitz & Maier 2008). The majority of the carbon chain radicals that have been studied so far is linear, but several families of non-linear carbon chains have been reported (Schmidt et al. 2003a,b, Ding et al. 2003, Zhang 2004, Thaddeus & McCarthy 2001). In this letter the focus is on the optical spectroscopy of C9H3. This species is a member of the C2n+1H3 family that has been observed for n=3-6 in a mass selective REMPI-TOF experiment (Schmidt et al. 2003a,b, Ding et al. 2003). Schmidt and coworkers found that the origin bands of the C9H3,C11H3 and C13H3 all absorbed in a rather narrow wavelength range, between 520 and 530 nm. As absorption frequencies typically shift to the red with increasing chain length (Maier 1997) the authors suggested that this “un-chain-like” behaviour could be due to a ring-chain motif. The absence of clear rotational information, however, precluded a definitive determination of the geometrical structure.

54 5.1 Introduction

Figure 5.1 – Schematic view of the pinhole nozzle. The ceramic body is mounted to a pulsed valve system that runs at 10 Hz and provides a gas pulse that is discharged by applying a negative high voltage pulse (-1000 V) to the cathode (C), striking upstream towards the grounded anode (A). The two plates are isolated by ceramic insulators (I1,2). The plates have a thickness of (A/I1/C/I2 = 1.0, 0.8, 1.0 and 2.0 mm) and the diameter of the central holes is increasing downstream: (A/I1/C/I2 = 1.0, 1.5, 1.8 and 3.5 mm).

Density functional theory (DFT) calculations by Zhang (Zhang 2004) predicted the structure, stability and electronic transitions of C2n+1H3 for n=4-6. From this study it was concluded that the molecular geometry of the C9H3 carrier of the REMPI-TOF laboratory spectrum (Schmidt et al. 2003a) does not incorporate a carbon ring and that a non-linear bent structure is the more likely carrier of the experimental spectrum. The present study contributes to this discussion with improved experimental results providing for the first

55 Chapter 5 Geometry of the non-linear Carbon Chain C9H3

W avelength [nm]

529.70 529.65 529.60 529.55

(a)

K =0

(b) a

K =1

a

K =2

a

K =3

a

K =4

a

K =5

a

K =6

a NormalizedIntensity

K =7

a

K =8

a

K =9

a

K =10

a

K =11

a

K =12

a

18877 18878 18879 18880 18881 18882 18883 18884 18885

-1

Frequency [cm ]

Figure 5.2 – The experimentally obtained electronic origin band spectrum of C9H3. The upper trace of panel (a) shows the K-stack resolved spectrum. The arrows indicate the unresolved band heads of the subbands (Ka=0–12). The inset in panel (a) represents a zoom-in of the partially rotationally resolved R-branch (see also Fig. 5.5). The lower trace of panel (a) indicates an empirical contour fit (with a Gaussian linewidth of 0.07 cm−1) to derive approximate rotational constants. Panel (b) shows the simulated stick diagrams for the individual subbands. The rotational temperature derived from the simulation is ∼ 23 K.

time resolved K-stack spectra for two previously observed electronic bands.

56 5.2 Experimental

W avelength [nm]

528.65 528.60 528.55 528.50 528.45

*

* *

* Intensity [a.u.] Intensity

18916 18917 18918 18919 18920 18921 18922 18923 18924

-1

Frequency [cm ]

Figure 5.3 – The experimentally obtained C9H3 absorption spectrum (bold line) of an electronic transition involving a vibrationally excited state in the upper electronic state, about 39 cm−1 blue shifted with respect to the origin band transition. The asterisks indicate overlapping transitions of − smaller species, including C2 and C2 . A contour fit (regular line) of the spectrum is also plotted based on constants as used for the origin band transition.

5.2 Experimental

The experiment is based on a pulsed cavity ring-down detection scheme that has been described in detail previously (Linnartz 2009, Wehres et al. 2010b). The C9H3 radicals are produced in a supersonically expanding hydrocarbon plasma by discharging a 0.5 % C2H2/He mixture. The present experiment employs a pinhole discharge nozzle connected to a pulsed valve system (General Valve series 9), and is a modified version of the pinhole system described in Ref. (Bazalgette Courrèges-Lacoste et al. 2001) to generate transient species at low final rotational temperatures. The modifications are essential in order to realize the improvements necessary to record a K-stack resolved spectrum. More specif- ically, the present nozzle geometry comprises four separated layers – two metal plates and two ceramic plates – which act as electrodes and insulators, as shown in Fig. 5.1. In the modified system, the diameter of the central hole in each plate increases from 1.0 to 3.5 mm towards the orifice. The smaller distance between the anode and cathode effec- tively increases the current density in the discharge region. The final ceramic plate (I2)

57 Chapter 5 Geometry of the non-linear Carbon Chain C9H3

Figure 5.4 – The geometry of the A3 and A7 isomers as introduced in Zhang (2004) are shown. The bondlengths (in angstrom) and the angle of the bend C2v structure result from improved calculations at the DFT -B3LYP/6-311G∗∗ level (Table 5.1). The likely carrier of the spectra shown in Figs. 5.2 and 5.3 is isomer A3. has a hole with a diameter of ∼3.5 mm and is used to prevent the discharge from clog- ging. The ∼2 mm thickness of this plate narrows the jet expansion angle, enhancing the collisional rate in the jet and decreasing the Doppler broadening. The nozzle is mounted in a vacuum chamber, that is evacuated by a roots-blower pump system with 1000 m3/hr pumping capacity. The typical pressure is ∼0.03 mbar in the chamber during jet operation for a backing pressure of ∼7 bar. Tunable laser pulses are generated by a pulsed dye laser (Sirah, Cobra Stretch) pumped by a frequency tripled Nd:YAG laser operated at a 10 Hz repetition rate with a pulse width

58 5.3 Results and Discussion

of ∼ 6 ns and a bandwidth narrower than 0.04 cm−1. Convolving this value with an es- timated Doppler width of ∼ 0.05 cm−1 results in a spectral resolution of ∼ 0.07 cm−1. The absolute laser frequency is calibrated with a precision better than 0.02 cm−1 using an iodine absorption reference spectrum that is recorded simultaneously. Cavity ring-down events are obtained by injecting a fraction of the laser pulse into a high-finesse optical cavity, which consists of two plano-concave mirrors (Research Electro-Optics, reflectiv- ity ∼99.998 % at 532 nm) that are mounted on high precision alignment tools. The cavity length is ∼58 cm, yielding typical ring-down times of 60-80 µs. A multi trigger scheme is used to match the ring-down event to the gas pulse (∼ 1 ms) and the coinciding discharge pulse (∼ 300 µs and -1000 V). The optical axis of the cavity crosses the plasma expan- sion perpendicularly. The distance of the cavity axis to the nozzle orifice can be varied during jet operation from 0 to 20 mm. Spectra are recorded for three different distances at ∼ 2, 7 and 12 mm and the ongoing adiabatic cooling causes the rotational temperature to drop, yielding Trot ∼23, ∼14 and ∼7 K, respectively. The light leaking out of the cavity traverses a narrow bandpass filter before it is detected by a photomultiplier tube (PMT). The PMT signal is digitized by an oscilloscope and the obtained data is transferred to a computer. The ring-down signal is analysed in real time by a LabVIEW program. Typi- cally 6 to 10 ring-down events are averaged to determine one data point. A full spectrum is obtained by recording the averaged ring-down time as a function of laser frequency.

5.3 Results and Discussion

5.3.1 Experimental Spectra and Analysis

In panel (a) of Fig. 5.2 the electronic origin band spectrum of C9H3 is shown, recorded around 18881 cm−1. The unambiguous assignment of this rotationally unresolved band to this specific molecular formula follows from a previous mass selective REMPI-TOF study (Schmidt et al. 2003a,b). The spectrum consists of a P- and R-branch. The P- branch exhibits a clear K-stack-resolved structure. The individual K-stacks making up the spectrum are shown as individual stick diagrams (panel (b)) and are discussed in more detail later. Several partially resolved peaks in the R-branch have spacings of ∼0.11 cm−1 (zoom-in) and are the cumulative result of a series of coinciding rotational transitions starting from Ka=0 to 6. The clearly resolved K-stack progression (Ka=0-12) in the P-branch is typical for a non-linear molecule. The incremental spacing between neighbouring K-stacks towards the low frequency side indicates that the underlying electronic transition should be parallel and that the observed spectrum reflects an A-type transition. Indeed, using the PGOPHER software (Western 2007), a contour fitting for an A-type transition using an asymmet- ric top model and Cs symmetry reproduces the observed spectrum well. This empirical −1 contour fitting yields indicative values of the molecular constants: T0 ≈ 18881.4 cm , A" ≈ 0.215 cm−1, B" ≈ C" ≈ 0.0159 cm−1, and ∆(A-(B+C)/2) ' -0.0181 cm−1. Besides the origin band a transition to a vibronically excited state is observed around 18920 cm−1 (Fig. 5.3). The transition, only 39 cm−1 blue shifted, was also detected at

59 Chapter 5 Geometry of the non-linear Carbon Chain C9H3

Table 5.1 – Molecular parameters of the C9H3 A3 and A7 isomers compared to DFT calculations at the B3LYP/6-31G∗ level from Zhang (2004).

Isomer A3 [C2v] A7 [Cs] Basis-set 6-311G∗∗a 6-31G∗b 6-311G∗∗a 6-31G∗b Expt. ∆E [eV] 0 0 1.29 1.27 A”[cm−1] 0.22683 0.21892 0.17980 0.17988 B”[cm−1] 0.016377 0.01637 0.020562 0.02039 C”[cm−1] 0.015274 0.01523 0.018452 0.01831 −1 c ν39 [cm ] 43.4(a1) 45(a1) 56.5(a’) 60(a’) 38.8 IP [eV] 7.6 6.34 <8.2 eVd aThis work. b Ref. (Zhang 2004). c Value for the upper state, derived from the band origin values (T0) in Table 5.2. dThe total photon energy of the (1+10) REMPI experiment (Schmidt et al. 2003b,a). lower resolution in the previously mentioned REMPI-TOF experiment and assigned to an electronic transition of C9H3, involving excitation of a low lying bending mode in the upper electronic state. The spectrum recorded here has a similar contour as the origin band at 18881 cm−1, and exhibits a very similar K-stack resolved structure. It is likely that this transition is due to the excitation of the same isomeric C9H3 form and that the excited 0 state corresponds to a completely symmetric vibrational mode (a1 or a symmetry) with an energy of about 39 cm−1. In Fig. 5.3 a contour fit is shown based on similar rotational constants as used for the origin band transition in Fig. 5.2.

5.3.2 Consideration of the Molecular Geometry

DFT calculations at the B3LYP/6-31G∗ level have been reported in Zhang (2004) for a large set of possible C9H3 isomers, yielding their geometry, stability and vibrational frequencies. Our estimated values for the molecular constants, A” ≈ 0.215 cm−1 and B” ≈ C” ≈ 0.0159 cm−1, derived from the band contour fitting, are very close to the cal- culated rotational constants of the A3 and A7 isomers (Table II in Zhang (2004), see also Fig. 5.4) with C2v and Cs symmetry, respectively. The other geometries listed in Zhang (2004) are for transition states or have molecular constants that substantially deviate from the values estimated here, and are not further considered. A ring-chain motif (like A7) was suggested in Schmidt et al. (2003b) and in Schmidt et al. (2003a) based on REMPI spectra and electronic structure calculations (Ding et al. 2003) and the observation that the longer chains C11H3 and C13H3 also absorb in the 520-530 nm region, i.e. do not substantially redshift with respect to C9H3. In Schmidt

60 5.3 Results and Discussion

Figure 5.5 – The partially resolved rotational structure in the R-branch (inset Fig. 1) and simulated spectra for the A3 isomer with C2v symmetry and 1:3 spin statistical weights (left panel) and the A7 isomer with Cs symmetry and no spin statistical weights (right panel). The upper bold line gives the experimental spectrum (and is identical in both plots). The bottom spectrum shows the simulation for 0.02 cm−1 linewidth to visualize the underlying spin statistics. The middle trace is the resulting spectrum for a linewidth of 0.07 cm−1 (comparable to the experimental resolution).

et al. (2003a) the authors argue that C9H3 should have a low IP (≤6.4 eV), as the corre- sponding mass 111 amu was found to be the most abundant ion upon the irradiation of the products of a butadiyne/Ar discharge with a 193 nm (6.4 eV) laser. The DFT calcula- tions performed in the present work (Table 5.1) indicate that the ionization potential of the A3 isomer amounts to 7.6 eV, which is somewhat lower than the total photon energy of ≤8.2 eV used in the (1+10) REMPI measurement (Schmidt et al. 2003b), but larger than the appearance potential of the C9H3 mass determined from the irradiation experiment. The calculated IPs of the ring-bearing isomers (e.g., 6.34 eV for isomer A7) are smaller and agree with the experimental appearance potential for mass 111 amu. A consistent explanation for this discrepancy is that the butadiyne/Ar discharge actually does produce the isomer A7, but since the experiment only uses mass detection, this result does not necessarily link the IP value of the carrier to the actual spectra observed by Schmidt and co-workers. The DFT work (Zhang 2004), however, proposed the A3 isomer as the more likely carrier of the C9H3 bands, based on minimum energy considerations. The present work adds to this discussion with improved experimental spectra and additional calcula- tions. The latter have been performed only for the two selected C9H3 isomers A3 and A7, using the GAUSSIAN 98 software package (Frisch et al. 1998) and an extended basis set

61 Chapter 5 Geometry of the non-linear Carbon Chain C9H3

−1 Table 5.2 – Inferred molecular constants [cm ] for the two C9H3 bands discussed here.

∼18881 cm−1 ∼18920 cm−1 a a T0 18881.41(1) 18920.21(1) A”b 0.22683 0.22683 B”b 0.016377 0.016377 C”b 0.015274 0.015274 A’ 0.20925(3)c 0.20917(7)c B’ 0.01667(1)c 0.01666(3)c C’ 0.01555(1)c 0.01554(3)c ∆(A-(B+C)/2)d -0.01787 -0.01794 A”/A’ 1.082 1.084 B”/B’e 0.982 0.982 a The errors represent the statistical uncertainties as obtained in the least squares fit. b Fixed to the calculated values of isomer A3 (this work). c Errors are determined by a least squares fit of the resolved K-stacks at a rotational temperature of ∼ 23 K. d ∆(A-(B+C)/2)=(A’-(B’+C’)/2)-(A”-(B”+C”)/2). e Assuming that B”/B’' C”/C’.

B3LYP/6-311G∗∗. The calculated geometries and structural parameters are close to the values found in Zhang (2004) and indicated in Fig. 5.4. The resulting rotational constants for the ground state, vibrational frequencies and the ionization potentials (IPs) are listed in Table 5.1.

It is clear from the discussion above that the resolved K-stack structure alone is not sufficient to discriminate between the A3 and A7 isomeric forms, since both species result in a spectrum that is very similar to the one shown in Fig. 5.2. Additional arguments are needed to select one of the two isomers as the likely carrier of the bands observed here and reported in Ref. (Schmidt et al. 2003b). These arguments are systematically discussed below. I. The DFT calculations, both in Zhang (2004) and in the present study, find an A7 minimum energy that is higher (1.27 and 1.29 eV, respectively) than the A3 minimum energy value. This favors – from a pure theoretical point of view – the C2v bent iso- mer geometry as the more likely carrier of the observed spectra. In a reactive plasma environment with much excess energy, one should be careful to eliminate less stable con- figurations, on the basis of a minimum energy argument only. However, as no other obvious C9H3 features pertaining to another isomer have been observed in the REMPI

62 5.3 Results and Discussion

Figure 5.6 – The experimental spectra of the two C9H3 bands recorded for three different rotational temperatures: ∼ 23 K (2 mm), ∼14 K (7 mm) and ∼ 7 K (12 mm). The experimental spectra are represented by the middle black line in each panel and compared to the simulated spectrum for the A3 (A7) geometry and are plotted above (below) the experimental spectra.

work (Schmidt et al. 2003a), it is likely that the present bands arise from the minimum energy configuration. II. The DFT calculations show that both isomers are expected to have a completely symmetric vibrational mode with a low excitation energy in their electronic ground state: −1 −1 43 cm (a1) for the A3 isomer and 57 cm (a’) for the A7 isomer. For most larger carbon chains it is found that upon electronic excitation the molecular geometry barely changes (Araki et al. 2003, Dzhonson et al. 2007) and the vibrational spacings in the first ex- cited electronic state are therefore expected to be comparable to those in the ground state. The calculations of the isomer A3 at the B3LYP/6-31G∗ level in Zhang (2004) show that the bond lengths and bond angle(s) indeed do not show significant changes between the ground and excited states. Consequently, the experimentally observed blue shift of the band at 18920 cm−1 with respect to the origin band offers a selection criterion. The exper- imental value of 38.8 cm−1 for the excited state bending mode is closer to the predicted 43 cm−1 value calculated for the A3 ground state geometry than for the corresponding A7 geometry (57 cm−1). III. Although the simulations for both isomers reproduce the contour of the observed

63 Chapter 5 Geometry of the non-linear Carbon Chain C9H3 spectra reasonably well, only the simulation for the A3 isomer exhibits a similar profile of the partially resolved (and reproduceable) rotational structure in the R-branch (inset Fig. 5.2 and Fig. 5.5). The simulated high-resolution spectra (using a Gaussian linewidth of 0.02 cm−1 in the simulations) indicate that the R-branch of the spectrum of the A7 −1 isomer (Cs symmetry) will resolve lines with spacings of about 0.06 cm , whereas the −1 spacings are about 0.11 cm in the A3 isomer (C2v symmetry) spectrum, which is very close to the actually observed values. The two times larger spacing in the symmetric isomer A3 spectrum arises from the 1:3 intensity alternation of odd and even Kb values due to nuclear spin statistics. This is illustrated in Fig. 5.5, where also simulations are presented at the actual spectral resolution of 0.07 cm−1. The A3 simulation is clearly closer to the experimental spectrum than the A7 simulation, but more importantly, even though this rotational progression is the cumulative result of a series of individual and overlapping rotational transitions, a spacing of 0.11 cm−1 cannot be realized with individ- ual transitions being separated by 0.06 cm−1. I.e., this part of the spectrum conflicts with the A7 geometry. Arguments II and III hint for the bent C2v structure as the molecular carrier of the observed spectrum and are fully consistent with argument I. Additional experiments can be performed. We have tried to distinguish between both isomers by searching for spectral differences as a function of rotational temperature. For this measurements have been performed at different distances downstream in the expan- sion. Empirical contour fittings were performed, using the ground state rotational con- stants derived from the DFT calculations here. The simulated spectra for both isomers A3 and A7 are plotted and compared to the experimental spectra in Fig. 5.6. The plots show that the experimentally obtained spectra for the origin band transition (left panel) and the excited mode (right panel) can be reproduced both for the A3 (upper spectrum) and A7 (lower spectrum) constants, but unfortunately from this no conclusive information can be obtained on the actual carrier. The rotational structure of the recorded spectrum is not resolved, therefore a rotational contour fit (Western 2007), utilizing a least squares procedure is needed to fit the overall shape of the experimental spectrum. This method has been used previously to analyse + + the spectra of other non-linear carbon chains, like trans-C6H4 (Araki et al. 2003), C6H3 , + and C8H3 (Dzhonson et al. 2007). Here, a similar procedure is followed assuming that the spectrum of an asymmetric molecule is characterised by the rotational constants A, B and C in both states, the band origin, and the full-width-at-half-maximum (FWHM, ∼ 0.07 cm−1) of a Gaussian line profile. The rotational constants in the ground state are fixed using the results of isomer A3 from the new DFT calculations at the B3LYP/6- 311G** level. Furthermore, the rotational temperature was fixed using the estimated values from the empirical contour fit. That way the rotational contour fit yields the band origin and the spectroscopic constants for the upper state. The results are summarised in Table 5.2. In a similar way, the spectrum of the vibronic band around 18920 cm−1 is reproduced through a contour fit using the calculated ground state constants of isomer A3 and the same A-type transition. This is shown in Fig. 5.3. It is found that the resulting ratios of A”/A’ and B”/B’ for both the 18881 and 18920 cm−1 bands, are close to unity, in

64 5.3 Results and Discussion agreement with the assumption made under point II that the molecular structure will not substantially change upon electronic excitation.

All arguments given above, based on the new observations and calculations presented here, are consistent in that they point towards the A3 isomer as the more likely carrier of the observed laboratory spectra. Clearly, the pure rotational spectrum of the isomer A3 would settle the issue, and may be measured in a Fourier transform microwave experi- ment (Thaddeus et al. 1998) or by linking the pure rotational spectrum to the electronic spectrum presented here in a microwave-optical double resonance experiment (Nakajima et al. 2002).

65

C2 Emission Features in the Red Rectangle a combined observational/6laboratory study

N. Wehres, C. Romanzin, H. Linnartz, H. Van Winckel, A. G. G. M. Tielens

A&A 518 (2010) 36A

67 Chapter 6 C2 Emission Features in the Red Rectangle

Abstract

Context. The Red Rectangle proto-planetary nebula (HD 44179) is known for a number of rather narrow emission features superimposed on a broad extended red emission (ERE) covering the 5000–7500 Å regime. The origin of these emission features is unknown. Aims. The aim of the present work is to search for potential carriers by combining new observational and laboratory data. This also allows to interpret spectral emission features in terms of actual physical conditions like temperature and density constraints and to trace chemical processes in the outflows of the Red Rectangle. Methods. Observational spectra have been obtained with the EMMI-NTT at offsets of 300, 600, 700, 1100, 1600 and 2000 distance to the central star HD 44179. The spectra are com- pared to the outcome of a time-gated laser induced fluorescence laboratory study of an expanding acetylene plasma using a special supersonic pinhole discharge source. With this set-up the hydrocarbon chemistry in the Red Rectangle nebula is simulated under laboratory controlled conditions. The plasma source has the unique feature to generate electronically and vibrationally excited species at low rotational temperatures. The com- parison is facilitated by a simple model for fluorescent emission in the nebula. Results. Two of the astronomically observed narrow emission bands can be assigned 3 3 as originating from unresolved rovibronic progressions within the d Πg → a Πu Swan system of the C2 radical. The band appearance corresponds to a rotational temperature between 200 and 1000 K. The emission is driven by absorption in the C2 Phillips bands followed by intersystem crossing from the singlet to the triplet state and pumping in the Swan bands.

6.1 Introduction

The Red Rectangle bi-conical reflection nebula was first identified by Cohen et al. (1975) and is associated with the post AGB binary star HD 44179 (Van Winckel et al. 1995). The star itself is visually obscured by a nearly edge-on circumbinary disc (Roddier et al. 1995, Waelkens et al. 1996, Osterbart et al. 1997, Bond 1997) and can only be seen in scattered light. The disk leads outflows into the north-west and south-east direction from the star and creates the appearance of an X-shaped structure. This shape as well as the unique emission spectrum of the nebula have attracted special attention in the past. In the infrared regime emission features have been found at 3.3, 6.2, 7.7, 8.6, 11.3, 13.57 and 14.23 µm (Russell et al. 1978, Waters et al. 1998) that are typical for vibrational fundamentals of polycyclic aromatic hydrocarbons (PAHs). The spectrum at longer wavelengths is rich of narrow emission bands that can be assigned to crystalline silicates (Waters et al. 1998). In the optical regime the spectrum is dominated by an extended red emission (ERE), which is a strong and broad emission feature that covers the region between 5000 and 7500 Å (Schmidt et al. 1980). Different origins have been proposed for this ERE such as photo- luminescence processes within silicates or nanoparticles of carbonaceous grains (Duley 1985, Witt & Boroson 1990, Witt et al. 1998, Ledoux et al. 1998, 2001, Van Winckel et al. 2002, Witt & Vijh 2004). Blue luminescence was detected in the Red Rectangle (Vijh

68 6.1 Introduction et al. 2004, 2005, 2006) and it was suggested that the ERE and the blue luminescence may arise from fluorescence of PAHs with a size of 14–18 C atoms. Superimposed on the ERE, a peculiar set of narrow emission features has been observed for which identifications are lacking (Scarrott et al. 1992, Sarre et al. 1995, Glinski & Anderson 2002, Van Winckel et al. 2002, Sharp et al. 2006). The ERE is widespread in the interstellar medium, but the narrow emission features are very unique for the Red Rectangle (Witt et al. 2008). Studies have shown that such emission features shift to the blue and become narrower with distance from the central star. These characteristics are the telltale sign of (anharmonic) molecular emission in a cooling outflow. The blueshifts bring the peak positions closer to the wavelength positions of some of the well-known Diffuse Interstellar Bands (DIBs) (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002), but for an opposing view see Glinski & Anderson (2002). The DIBs are observed as absorption features through diffuse interstellar clouds and are attributed to electronic transitions of molecular transients (see also Linnartz et al. (2010)). Apart from the PAH and ERE emission, the carbon-rich nature of the nebula has been corroborated by additional molecular detections like CH, CN and CH+ (Hall et al. 1992, Bakker et al. 1996, Hobbs et al. 2004). The carbon nature of the star is also in agreement with the IR emission spectrum and the PAHs in the circumstellar gas (Leger & Puget 1984). Furthermore, the disk around the central star is oxygen-rich with molecules like 12 13 OH (Reese & Sitko 1996), CO and CO, gaseous and perhaps solid state CO2 and silicate dust (Waters et al. 1998). The presence of carbon-rich material in the outer part of the nebula and the oxygen-rich material in the central part of the nebula indicates a spatially separated chemistry (Waters et al. 1998). In all, the picture that emerges from these observations is that of an accretion disk surrounding the low mass companion. This accretion disk may have been (previously) fed by Roche lobe overflow from the post-AGB object, and part of the accretion disk ma- terial is accelerated outwards in a fast jet evident in the broad Hα and CO (UV) emission components (Sitko et al. 2008, Witt et al. 2009). The narrow Hα component has been attributed to a small HII region close to the binary stars, ionized by UV emission from the accretion disk (Jura et al. 1997, Witt et al. 2009). The submillimeter CO emission originates in a slowly expanding (' 0.8 km/s) and Keplerian rotating circumbinary disk (Jura et al. 1997, Bujarrabal et al. 2005). The cool and narrow component in the UV lines of CO presumably represents the PDR separating the HII region from the surrounding, circumbinary, molecular disk. The atomic lines of potassium as well as the CH, CH+, and CN lines may also originate in this PDR zone. All of these components are “seen” because of scattering into the beam and not directly. Indeed, given the polarization re- sults (Reese & Sitko 1996), the emission of these lines further out in the nebula seems to be largely due to scattering of photons from this PDR zone into the line of sight of the observer. In the present study, we combine observational data at different distances from the central star (i.e. for different physical conditions and at different chemical stages) and high resolution laboratory data. These are recorded in emission through a supersoni- cally expanding plasma. Such plasma expansions have become a standard to generate molecular transients of astrophysical interest and their applications in astronomy have

69 Chapter 6 C2 Emission Features in the Red Rectangle been reviewed recently by Linnartz (2009). The overall aim of this experimental study is to identify possible carriers of the narrow emission features as observed around the Red Rectangle and simultaneously to follow the chemical evolution in the outflows of this proto-planetary nebula. Here we focus on two narrow emission bands attributed to the C2 molecule (Sarre 2006, Glinski et al. 2009). The paper is organized as follows. Section 6.2 describes the new observations and experimental results and combines these results with existing ones. Section 6.3 describes the model we use in order to determine the abundance of C2 in the outflows of the Red Rectangle. Section 6.4 shows the implications and is followed by the conclusions in Section 6.5.

6.2 Astronomical Observations and Data Reduction

The astronomical spectra were obtained on February 4 and 5, 2008, at the 3.5 m NTT (New Technology Telescope) at ESO, La Silla in Chile, making use of the EMMI, ESO Multi Mode Instrument, in medium dispersion spectroscopy mode. All observations were carried out using long slit spectroscopy. Long slit settings of 100 by 200 00 were chosen. This has the advantage that light is sampled over only a narrow region at a given distance from the central star in the outflow of the nebula. The slit was oriented perpendicular to the polar direction of the nebula at a constant position angle of 105◦ east from north at the different offsets from the star. The slit spectrum provides the wavelength versus width (spatial information) and is later collapsed into one single point, which is justified by the constant offset angle perpendicular to the nebula’s symmetry axis. That way the S/N ratio increases which helps to obtain high quality spectra even in fainter regions of the nebula further away from the central source. The visual magnitude of the central source HD 44179 amounts to about 9 but is obscured by the nebula. For telescope settings further away from the central star the intensity of the flux is decreasing fast and thus integration times have to be varied with distance from the centre. The observation blocks are specified in Table 6.1 for three different settings using grating 6 in two different wavelength regimes in order to cover the whole ERE at a resolution of λ/∆λ = 5000. Grating 6 was used at central wavelengths of 5870 Å and 6500 Å. Spectra were obtained at the central star as well as at 300 and 700 distance. The third setting used grating 7 at larger offsets from the central star (1600 and 2000). Here we recorded spectra covering the whole ERE domain in order to keep integration times realistic. As a consequence the resolution decreased to λ/∆λ = 2600. The central wavelength position for this setting was at 6200 Å. For each spectral study three individual exposures were taken in order to correct for cosmic hits. The read-out mode for the spectra was chosen to be slow with a binning of 2 x 2 that lasted about 18 s. The data reduction was performed using MIDAS. A routine procedure included trim- ming of the spectra, flat-fielding, bias subtraction and cosmic hit subtraction for which appropriate observations were made. The wavelength correction was provided using a spectrum of a He-Ar discharge lamp and flux calibration was done using a standard star (Hiltner 600). The obtained wavelengths were then again corrected for the barycentric

70 6.2 Astronomical Observations and Data Reduction

Table 6.1 – NTT-EMMI Telescope Settings.

Distance to Grating Central Wavelength Resolution Integration HD 44179 Wavelength Coverage λ/∆λ time (arcsec) (Å) (Å) (s) 0 6 5870 640 5000 60 0 6 6500 640 5000 60 3 6 5870 640 5000 500 3 6 6500 640 5000 500 7 6 5870 640 5000 2000 7 6 6500 640 5000 2000 16 7 6200 1300 2600 3600 20 7 6200 1300 2600 3600

a Note: The slit orientation is 105◦ east from north, perpendicular to the nebula’s symmetry axis. radial velocity shift during exposure. The sky background was subtracted by extrapolat- ing a sky-spectrum outside the region of interest. This data-set is extended with spectra recorded at 600 and 1100 and is available from Van Winckel et al. (2002).

6.2.1 Astronomical Results Figure 6.1 presents an overview of the observational results. In all spectra the ERE is subtracted and spectra are stacked with an arbitrary offset, in order to allow for a better comparison. The emission features of the Na DII and DI lines at 5890 Å and 5896 Å are clearly visible in all spectra, as well as the Hα line at 6563 Å. The spectra further- more show a number of other emission features; stronger ones around 5800–5900 Å that have been topic of several previous studies (Scarrott et al. 1992, Sarre et al. 1995) and many weaker features, up to 6800 Å. These spectra will be described in detail elsewhere (Wehres et al., submitted, see also Chapter 7). Here, we draw the attention to the clear spectral variations with distance to the star; new bands arise, some bands become more intense, whereas others clearly decrease in intensity. Overall the emission features seem to be quite sensitive to the change in environmental conditions in the Red Rectangle neb- ula. In our spectra we can see that the most prominent emission features between 5800 and 5900 Å show up already at 300, whereas most of the other emission features become observable at larger distances from the star. In this paper, we focus on two weak features in the 5550–5650 Å range of the ob- servations at 300, 600 and 700 distance from the central star (Fig. 6.2). Besides the broad and unidentified feature at 5600 Å, this zoom-in reveals two weaker bands at 5585 Å and 3 3 0 00 5635 Å that can be assigned to vibronic bands in the d Πg → a Πu (v , v )=(0,1) and (1,2) Swan band system of C2, as will be discussed. C2 was identified in circumstellar material of post AGB stars (Bakker et al. 1996), in carbon rich outflows of proto-planetary nebulae (Klochkova et al. 1999), in cometary spectra (Gredel et al. 1989, Fink & Hicks

71 Chapter 6 C2 Emission Features in the Red Rectangle

4.5

Na DII + DI H

4.0

at 20"

3.5

at 16" 3.0

2.5

at 7"

2.0 Intensity [a.u.] Intensity

at 3"

1.5

central star

1.0

5600 5800 6000 6200 6400 6600 6800

W avelength [Å]

Figure 6.1 – Overview of all NTT-EMMI data taken towards HD 44179 on February 4 and 5, 2008. Complementary data for 600 and 1100 (not shown) are available from Van Winckel et al. (2002).

1996, Hobbs et al. 2004) and in interstellar clouds (Souza & Lutz 1977, Chaffee & Lutz 1978, Hobbs 1979, van Dishoeck & de Zeeuw 1984, Gredel 1999). In Sarre (2006) the 3 3 00 C2 (0,0) transition in the d Πg → a Πu at 5165 Å is mentioned for offsets up to 13 , but the corresponding spectrum was not shown. Very recently Glinski et al. (2009) showed a spectral overlap between a Red Rectangle emission band and a C2 emission spectrum of the comet Hale-Bopp.

6.2.2 Laboratory Experiment

An existing set-up (Volkers et al. 2004) designed for time-gated fluorescence spectroscopy of isotopologues of atmospheric interest (Volkers et al. 2006) has been modified by im- plementing a supersonic plasma source. This plasma source has been used previously to study vibrationally excited radicals at low rotational temperatures (Bazalgette Courrèges- Lacoste et al. 2001) and consists of a General Valve introducing a gas pulse in a multi- layer discharge geometry. Typically a 500 µs long negative high voltage discharge pulse (− 700 V) is offered to a synchronized 700 µs long gas pulse of a 0.2 % C2H2/He gas mix- ture that is expanded with a backing pressure of 4 bar into a vacuum chamber (10−5 mbar during jet operation) at a repetition rate of 10 Hz. The plasma expansion is intersected

72 6.2 Astronomical Observations and Data Reduction

1.0

(c) at 7"

0.8

0.6

(b) at 6"

0.4 x10 Intensity [a.u.] Intensity

(a) at 3"

x2.5

0.2

0.0

5560 5580 5600 5620 5640

W avelength [Å]

3 3 00 00 00 Figure 6.2 – Zoom-in on the Swan band d Πg → a Πu emission features at 3 , 6 and 7 offset from the central star.

about 10 cm downstream by the output of a second harmonic Nd:YAG pumped tunable dye laser. The latter is operated on Rhodamine 6G dye and covers 5550 Å to 5850 Å. A single stage amplification scheme is used and maximum power ranges from 4 mW/pulse at 5650 Å to 0.5 mW/pulse at 5850 Å. The fluorescence signal is recorded by a photo- multiplier tube (PMT) using a time-gated integration scheme. The variable gate width of the PMT is necessary to prevent for saturation effects, because of light emission from the plasma source or by scattered laser light. Special care has been taken to reduce the amount of interfering light by also using light skimmers and band pass filters. The recorded spec- tra correspond to the intensity of the fluorescence signal at the PMT as a function of the incident photon wavelength. Calibration of the excitation wavelength is obtained by simultaneously recording an iodine absorption spectrum.

6.2.3 Experimental Results

Although we were expecting a rather congested spectrum, reflecting the many compounds that may be formed in the plasma expansion (Witkowicz et al. 2004) and likely present in the outflows of the Red Rectangle, we only recorded a limited set of rotationally re- solved bands. In Figs. 6.3 and 6.4 two of these bands are shown in detail. These consist

73 Chapter 6 C2 Emission Features in the Red Rectangle of a set of subsequent rovibronic transitions that can be assigned to Swan band transi- 3 0 3 tions starting from the ground and vibrationally excited levels: d Πg (v = 0) → a Πu 00 3 0 3 00 (v = 1) and d Πg (v = 1) → a Πu (v = 2). Due to the degeneracy of the triplet state, the rotational ladder splits into three separate, slightly staggered, sets of PQR rotationally resolved branches that reflect the substates (Curtis & Sarre 1985). Although C2 is a very well studied molecule, high resolution laboratory emission studies are rare. Curtis & Sarre (1985) obtained high resolution C2 spectra of the Swan band (v=0,1) transitions using LIF spectroscopy with a mixture of Na and C2Cl4 reactants. Also Chen & Mazumder (1990) obtained emission spectra from a laser induced plasma during laser ablation of graphite in an Ar/He mixture. Their spectra are with a lower resolution. Most other studies comprise direct absorption spectroscopic techniques. As a consequence vibrationally excited levels are then only accessible in rather hot environments (i.e. discharge cells) with a number of disadvantages, such as spectral congestion and a lower detection sensitivity because of a lower state density. Comparable plasma sources have shown to generate a large sample of ± hydrocarbon radicals of the form CnHm (Motylewski & Linnartz 1999, Witkowicz et al. 2004). In direct absorption rotational spectra of molecules like C6H (Linnartz et al. 1999) + or HC6H (Pfluger et al. 1999) have been observed. No evidence for fluorescence signals of such species has been found in the present experiment. The two transitions reported in the laboratory spectra (Fig. 6.3 and Fig. 6.4) are the same transitions evident in the spectra of the Red Rectangle. The small shifts between ob- served and experimental spectra are due to different temperatures. We have simulated the spectrum of C2 using the PGOPHER program (Western 2007) with standard molecu- 3 00 00 lar parameters as listed by Prasad & Bernath (1994) for the a Πu v =1 and v =2 levels 3 0 and from Lloyd & Ewart (1999) and Prasad & Bernath (1994) for the d Πg v =0 and v0=1 states. In Fig. 6.3 and in Fig. 6.4, the experimental spectrum is plotted as a black solid line and shows the fluorescence intensity with respect to the excitation wavelength. Overplotted as a dotted line is a simulated spectrum. The linewidth of 0.65 cm−1 reflects residual Doppler broadening for an unskimmed beam expansion. The spectral pattern corresponds to a rotational temperature of about 15 K in the expansion. The plasma cre- ates electronically and vibrationally hot molecules whereas the expansion adiabatically cools rotational motion. The result is vibrational excitation due to the discharge but low rotational temperatures due to subsequent cooling by carrier gas molecules. The overall good agreement between experiments and calculations validates the use of the PGOPHER program to simulate the emission spectrum of C2. There are small dis- crepancies between experiment and simulation. Specifically, at 5625 Å and 5623 Å the experimental spectrum shows two bands not reproduced by the simulations at 15K (indi- cated with arrows; Fig. 6.3). Both bands do show up in the simulation at higher rotational temperatures. Also, Figure 6.4 reveals a peak at 5587.0 Å (indicated with an arrow), that is not reproduced in the simulation and that corresponds to the bandhead. This indi- cates that not only a low temperature component is present in the expansion, but that also species are formed at substantially higher temperatures.

74 6.2 Astronomical Observations and Data Reduction

W avelength [Å]

5634 5632 5630 5628 5626 5624 5622

1.4

R P Q

1 1

1

1.2

e(4) e(3) f(4) f(2)e(3) f(2) e(5) f(4) e(5)

e(3)

1.0 R

P Q 2

2 2

f(2)

e(1) e(3)

e(3) e(1) f(2)

f(2)

0.8

R

3

0.6

e(1)

0.4 NormalizedIntensity

0.2

0.0

17745 17750 17755 17760 17765 17770 17775 17780 17785 17790

-1

W avenumber [cm ]

Figure 6.3 – A comparison of the experimental time-gated and rotationally resolved fluorescence 3 0 3 00 spectrum of the C2 d Πg v =0 → a Πu v =1 Swan band transition (solid line) with a simulation calculated using the program PGOPHER (dotted line). The simulated spectrum has been arbitrarily normalized. Band assignments are shown at the top. The arrows indicate high temperature bands which are not modelled. See text for details.

6.2.4 The C2 Rotational Contour: Excitation Temperature and Ve- locity Shifts

The observed line profiles in the astronomical spectrum can in principal be used to esti- mate the rotational excitation temperature of C2 in the Red Rectangle. Simulations, using PGOPHER (Western 2007) are compared to the observational spectra at 700 in Figure 6.5, assuming a single excitation temperature and convolved to the instrumental resolution of the observations (R = 5000). Both band profiles reveal a rotational temperature depen- dence. At low temperatures (≤ 100 K), the PQR rotational structure is obvious, even when convolved to the low resolution of the observational data. As the temperature increases, the P-branch gains in intensity whereas the R-branch flattens. With the appearance of a more intense bandhead, the overall position also shows a shift to the red. This redshift is typical for molecules with a larger rotational constant in the upper state (v0) than in 00 3 3 the lower state (v ); the rotational constant in the d Πg → a Πu transition at 5585 Å −1 0 −1 00 3 3 is ' 1.7 cm in the v = 1 and ' 1.6 cm in the v = 2 state for the d Πg → a Πu transition at 5635 Å it is ' 1.7 cm−1 in the v0=0 state and ' 1.6 cm−1 in the v00=1 state.

75 Chapter 6 C2 Emission Features in the Red Rectangle

W avelength [Å]

5588 5586 5584 5582 5580 5578 5576 5574 5572

1.4

Q Q

1 2

1.2

P R

e(3) f(4) e(1) f(2)

1 1

1.0

e(5) e(3) f(4) f(4) e(3) e(5) f(2)

0.8

R P

2

2

0.6

e(1) f(2) e(3) e(3) f(2)

0.4 NormalizedIntensity

0.2

0.0

17890 17900 17910 17920 17930 17940 17950

-1

W avenumber [cm ]

Figure 6.4 – A comparison of the experimental time-gated and rotationally resolved fluorescence 3 0 3 00 spectrum of the C2 d Πg v =1 → a Πu v =2 Swan band transition (solid line) with a simulation calculated using the program PGOPHER (dotted line). The simulated spectrum has been arbitrarily normalized. Band assignments are shown at the top. The arrow indicates bandhead emission. See text for details.

The bandhead becomes dominant at temperatures of ∼ 150 K and develops a steep red edge. Starting with this temperature, the peak position of the unresolved bandhead also does not shift any further to the red. Only the line width shows then a slight temperature dependence, mainly due to increasing blending of the bandhead with R-branch transitions (Fig. 6.5). It should be noted that for the same number of C2 molecules, the bandhead intensity at 2000 K will be substantially lower than for 200 K. However, as this number is not known, we continue to work with normalized spectra and the resulting band width is used to obtain additional information. The fairly symmetric C2 bands in the astronomical spectra and their peak positions (5635 & 5585 Å) reveal that the rotational temperature has to exceed ' 200 K and cannot be higher than ' 1000 K (Fig. 6.5). A more recent spectrum of C2 (Glinski et al. 2009) was obtained at substantially higher spectral resolution (R ' 37,000) and barely resolves the bandhead as well. In Figure 6.6 we show the observational VLT spectrum overplotted by simulations for a ro- tational temperature of 300 K for two different resolutions. The simulation of this VLT data has to focus on the blended bandhead emission and this does not provide better con- straints on the temperature of the emitting C2 molecules. It does constrain however the

76 6.3 Fluorescent Emission in the Red Rectangle velocity of the gas; e.g., in order to obtain a decent match on the steep red-edge of the bandhead, the observational spectrum had to be shifted about 0.025 nm ± 0.005 nm to the blue. Also, the simulated width of the bandhead emission is somewhat less than observed (Fig. 6.6). Convolving the simulated profile with a Gaussian with a Doppler broadening parameter of 5.8 km/s provides a good fit to the observations. The measured wavelength shift corresponds to a velocity of 13.3 km/s ± 3.7 km/s. We note that better constraints – particularly on the temperature – would be possible with a somewhat higher spectral resolution and a higher S/N observation that would resolve the individual R branch tran- sitions. The observed C2 bandhead shift is somewhat less than the systemic velocity shift of the gas as probed by various optical emission lines (18 – 19 km/s) (Van Winckel et al. 1995, Hobbs et al. 2004, Witt et al. 2009). The derived Doppler broadening parameter corresponds to a FWHM of an unblended rotational line of ' 14 km/s. This width is comparable to the narrow component, prominent in the Hα profile and the atomic (e.g., KI, CaII) and molecular lines (e.g., CH, CH+, and CN) lines (Hobbs et al. 2004, Witt et al. 2009). Some atomic lines show a more complex velocity profile with two components, at 13 and 25 km/s (Sitko et al. 2008). Generally these components have comparable strength but not so for the Na D lines. A very broad Hα component as well as the fluorescent CO line emission in the UV reveal the presence of a hot (T ' 5500 K) fast molecular outflow (' 300 km/s). There is also evidence for a rotationally cold (T ' 50 K) but vibrationally excited CO component (T ' 2000 K) which requires no velocity shift and has a width of some 20 km/s, comparable to the value derived from the C2 lines (Sitko et al. 2008). In contrast, the observed broadening of the C2 bandhead is much larger than the width of the submillimeter, pure rotational, CO lines (' 5 km/s). Finally, the Hα and Na DI lines (as well as some of the UV lines of CO and OH) in the nebula are more highly polarized than the (ERE) continuum, indicating that they have a large or even dominant scattering contribution (Reese & Sitko 1996) and they may actually originate in the inner PDR zone that separates the ionized gas from the molecular torus. The Red Rectangle has been studied in depth in atomic and molecular lines as well as in the visible and infrared continuum emission (Witt et al. 2009, Waelkens et al. 1996, Bujarrabal et al. 2003, 2005, Hobbs et al. 2004, Cohen et al. 2004, Waters et al. 1998). In view of the width, the C2 emission may also originate in the PDR zone and not in the rotating, molecular disk. However, the small blue shift of the bandhead (relative to the atomic lines) indicates that the C2 emission in the lobes has a slightly different origin than these lines, and perhaps this molecular emission is due to material (partially) entrained in the jet while the atomic lines originate (by scattering) of the cavity walls.

6.3 Fluorescent Emission in the Red Rectangle

6.3.1 Model Now it is possible to combine these data to study the physical and chemical processes that are involved in the excitation mechanisms. For that reason, we present a model that

77 Chapter 6 C2 Emission Features in the Red Rectangle

11

(g)2000 K

10

(h)1500 K

9

(g)1000 K

8

(f)750 K

7

(e)550 K

6

(d)350 K

5

(c)250 K

4

(b)150 K

3 Normalized Intensity

(a)15 K

2

at 7"

1

0

5550 5560 5570 5580 5590 5600 5610 5620 5630 5640 5650

W avelength [Å]

3 3 Figure 6.5 – Simulated line profiles of the Swan Band d Πg → a Πu emission features as a function of the rotational excitation temperature convolved to a resolution of 1.1 Å. These are compared to the observational spectrum at 700. All spectra have been normalized and stacked with an arbitrary offset in order to allow for better comparison.

accounts for the observed C2 emissions in the Red Rectangle and that is also suited to infer quantitative information. Similar to comets, we attribute the Swan bands to fluorescent emission from C2 pumped by stellar photons. In this model, the from the central region is carbon-rich. Hence, all the oxygen is locked up in CO while the excess carbon is assumed to be in the form of C2H2. The acetylene is slowly photodissociated by UV photons from the star and/or from the putative accretion disk around the binary (Men’shchikov et al. 2002, Witt et al. 2009) to form sequentially C2H, C2, and eventually 00 00 C. As we observe the C2 signatures already at 3 and they extend out to 13 (Sarre 2006), this process has to start very quickly and be maintained for a long time compared to the outflow time scale (see also below). At any point, we will treat the C2 abundance as in steady state between photoformation from a reservoir of C2H2 and photodestruction. Following Gredel et al. (1989), we have examined all possible electronic transitions −1 of C2 up to the Mulliken band ∼ 44000 cm (Fig. 6.7). Besides the regular radiative 1Σ+ transitions, this also includes “intercombination transitions” between the singlet X g 3 and triplet a Πu levels. This analysis allows us to identify the energy levels that are im- portant in view of the energy distribution of the Red Rectangle nebula (Fig. 6.8). Given

78 6.3 Fluorescent Emission in the Red Rectangle

1.0

Observations VLT R=37,000

Simulation at R=18,800

Simulations at R=37,000

0.8

0.6

0.4 NormalizedIntensity

0.2

0.0

5631 5632 5633 5634 5635 5636

W avelength [Å]

Figure 6.6 – VLT spectrum of the transition at 5635 Å (Glinski et al. 2009). The resolution of the observation is ∼ 37,000. The observational spectrum is compared to simulations at a resolution of 37,000 and 18,800. The rotational temperature of the simulations is ∼ 300 K.

the relatively low energy flux, especially in the UV in the Red Rectangle (Men’shchikov et al. 2002), the main pump from the singlet states to the triplet states is through the 1Σ+ 1Π Phillips bands. Radiative excitation from the X g to the A u state is rapidly followed by fluorescence which can leave the molecule in a vibrationally excited state of the ground 1Σ+ ∗ electronic state, in the following defined as X g (v ). Because these states are close in en- 3 ergy to levels in the a Πu state, intercombination band transitions are rapid (van Dishoeck 3 & Black 1982). The intercombination band from the lowest vibrational state of the a Πu 1Σ+ state to the ground vibrational state of the X g state is actually relatively slow and this transition presents a bottleneck in the “flow” (Le Bourlot & Roueff 1986). Once popu- 3 3 lation has been transferred into the a Πu state, stellar photons can excite the d Πg state 3 3 3 3 ∗ (a Πu → d Πg) and its vibrationally excited levels (a Πg → d Πg(v )) from which the observed Swan bands become visible in fluorescence. Both, the lowest vibrational level 3 3 ∗ of the d Πg state and the excited vibrational levels of the d Πg(v ) state, show emission as can be seen in our observational and laboratory spectra. Swan band fluorescence that 3 leads to a vibrational excited level of the a Πu state most likely leads via the intercombi- nation transition back into the singlet system. Actually, Swan band emission that leads to 3 3 the lowest vibrational state of the a Πu state can be pumped several times into the d Πg

79 Chapter 6 C2 Emission Features in the Red Rectangle state before an intercombination transition transfers the population back into the singlet 1Σ+ state. How often the Swan band can be pumped in this way, before relaxing into the X g state, is strongly dependent on the distance to the exciting star. Since the excitation of the vibrational levels in the ground state occurs via the Phillips band, pumped by stel- lar photons, the rate is inversely proportional to the distance squared. Intercombination transitions occur radiationless and are not dependent on the stellar flux. The pump in the triplet state is again dependent on the photon flux of the star and hence also scales inversely with the distance to the star squared. Overall, Swan band fluorescence scales in this excitation scheme, thus, inversely with distance to the fourth power. Sophisticated models have been developed to determine the level populations of all rotational, vibrational, and electronic states involved in connection to the Swan band ex- citation in comets. However, the few levels observed in the Red Rectangle, the low res- olution and relatively poor quality of the observed transitions does not warrant such an extensive model. Rather, we have developed a straight forward model which does retain, though, the essential aspects of the energy level diagram and the various pumps involved (Figs. 6.7 and 6.8). With respect to the stellar radiation field we include four electronic energy levels which are relevant for the pump scheme in the Red Rectangle (Fig. 6.8) 1Σ+ 1Π 3Π 3Π – the singlet states, X g and A u, and the triplet states, a u and d g – and in order to mimic the intercombination and fluorescence transitions, we introduce three fiducial 1Σ+ ∗ 3Π ∗ 3Π ∗ vibrational levels representing excited states, the X g (v ), a u(v ) and d g(v ). Rota- tional levels are not taken into account.

6.3.2 Results Einstein A coefficients for spontaneous emission have been obtained for a series of transi- tions using the calculated oscillator strengths by Kokkin et al. (2007). The results for the Phillips and the Swan bands agree well with the coefficients calculated by van Dishoeck (1983) and Gredel et al. (1989). From these we have determined representative transition rates (Ki) for the relevant ground and fiducial vibrational states (Table 6.3.1). The rates for the intercombination transitions between the singlet and triplet states were calculated using the transition dipole moments given by Gredel et al. (1989). The results (Table 6.3.1) agree reasonably well with the outcome of the more detailed study by Le Bourlot & Roueff (1986) who provide intercombination transition rates split out to individual ro-vibrational levels. Einstein B coefficients for these transitions are calculated from the usual formulae linking the Einstein coefficients. Radiative excita- tion rates follow then by convolving these with the stellar spectrum for distances at 600 (Table 6.3.1). We adopted a Kurucz Te f f = 7750 K, log g = 1.5 model for the visual spectrum normalized to a luminosity of 6050 L (Men’shchikov et al. 2002). In our eval- uation of the radiative pumping rates, we ignore extinction within the nebula and hence these rates scale simply inversely with the distance to the star squared. We have checked that stimulated emission can be ignored at these wavelengths. Special care was taken to ensure that transitions involving the fiducial vibrational levels have representative rates, where we realize that given the Einstein coefficients involved and the (likely) low optical depth in the pumping transitions, only the lowest vibrational levels have to be taken into

80 6.3 Fluorescent Emission in the Red Rectangle ] 1 − [s 1.0(-1) 7.0(-2) 3.8(-4) Rate k ) 5.5(-3) ) 1.3(-2) ∗ ∗ v v ( ( u u + g + g Σ Σ Π Π 1 1 3 3 + g a a X X Σ 1 → → → → X ) ) ) ) ∗ ∗ ∗ ∗ v v v v → ( ( ( ( + g + g u u u Σ Σ Π Π Π 1 1 3 3 3 X X a a a Transition ] 1 − [s 5.4(-4) 17 7.1(-4) 16 1.7(-3) 15 Rate4.0(-4) Number1.9(-4) 13 Intercombination 14 k ) 1.3(-3) ∗ v ( u ) g g ∗ Π Π Π v 1 ( 3 3 u A g g d d Π Π Π 1 3 3 → → → A d d ) ) ) ∗ ∗ ∗ v → v v → → ( ( ( + g + g u u u u Σ Σ Π Π Π Π 1 1 3 3 3 3 X X a a a a Transition . 00 ] 1 − [s 2.8(6) 11 2.3(6) 10 7.0(6) 9 5.1(4) 7 Ratek Number Pumping ) 7.8(6) 12 ∗ v ( ) 2.0(4) 8 ) u u ∗ ∗ v Π Π v ( ( 3 3 in the Red Rectangle at 6 + g + g u u a a Σ Σ 2 Π Π 1 1 3 3 → → X X a a ) ) ∗ ∗ → → v v → → ( ( for C u u g g g g a Π Π Π Π Π Π 1 1 3 3 3 3 A A d d d d Emission cients ffi 6 5 4 3 2 Number Spontaneous 1 – Rate coe Note: Values in brackets give exponents to base 10. a Table 6.2

81 Chapter 6 C2 Emission Features in the Red Rectangle

4 4

5x10 5x10

1 +

D

3

u

e

g

4 4

4x10 4x10 ]

1 -1

C

g

Fox-Herzberg

4 4 [cm

3x10 3x10 e

Deslandres-

D'Azambuja 3

d

g

4 4

Mulliken 2x10 2x10

d-c Band

Swan Band

3 +

c

g Intercombination

4 4 RelativeEnergy T Transition I+II

1x10 1x10

1

3 - A

u b

g

3 Ballik-Ramsay

1 Phillips

a X

u

g

Triplet States Singlet States

Figure 6.7 – An overview of the electronic energy states in C2, summarizing the energy levels in C2 and the corresponding energetics. Arrows indicate the most common transitions in the singlet and triplet system. The two arrows connecting the singlet and triplet states are schematics for “intercombination transitions”, which are relevant in the excitation of C2 because levels of the vibrational states in the X1Σ state and the a3Π state are close in energy. account. Here excitation of the relevant triplet states is mainly assumed to result from optical pumping of the singlet states followed by intersystem crossing. Other mechanisms, such as collisional pumping through H/H2 collisions, electron impact excitation, or a non- thermal distribution following chemical reactions are considered to be less important. Given the low temperature and density of the gas, collisional excitation and de-excitation of the vibrational and electronic levels can be neglected as can be shown when adopting a hydrogen density as deduced from Men’shchikov et al. (2002) of ∼ 2.3 × 104 cm−3 00 −16 2 for a distance of 6 . We assume a cross–section of C2 given by σ ∼ 5 × 10 cm (van Dishoeck & Black 1982, Pouilly et al. 1983). Collisions will then occur with a rate of ∼ 2 × 10−6 s−1. This result is very small compared to spontaneous emission in either the singlet or the triplet excited electronic transitions (Table 6.3.1). The temperature in the outflow is low (200 − 1000 K) relative to the excitation energy of the relevant pumping levels (∼ 10,000–25,000 K) and hence collisional pumping of either the Phillips or the Swan band transitions is negligible. Electrons in the outflow will be provided by ionization of carbon and their abundance will be small. So, although their velocity is

82 6.3 Fluorescent Emission in the Red Rectangle

4 4

2.5x10 2.5x10

3

d (v*)

g

3

d

g

4 4

2.0x10 2.0x10 ] -1

4 4 [cm

1.5x10 1.5x10 e

Swan

1

4 4

A

1.0x10 1.0x10

u

Phillips

Intercombination

3 3 RelativeEnergy T

5.0x10 3 5.0x10

Transitions 1

a (v*)

X (v*)

u

g

3

a 1

u

X

g

Triplet States Singlet States

Figure 6.8 – Relevant energy levels of C2 in the environment of the Red Rectangle nebula. Also shown schematically are the fiducial excited vibrational levels within these electronic states.

intrinsically much larger than for H atoms, electrons are also not important for excitation or de-excitation of the levels involved.

The seven relevant energy levels in our fiducial C2 model system are coupled by the ra- diative transitions illustrated in Fig. 6.8 with the rates given in Table 6.3.1. We solved the statistical equilibrium equations connecting these levels, combined with the abundance equation, and the results are given in Fig. 6.9 as a function of distance from the exciting 1Σ+ star. Most of the C2 molecules are in the ground state of the singlet system (X g ). Ap- 3 preciable population is transferred to the a Πu state, but its excitation is lower by one to two orders of magnitude depending on the distance to the exciting star. The population of 3 ∗ 00 the a Πu(v ) state is down again by another one to two orders of magnitude at 6 . This 1Σ+ → 1Π just reflects the relatively slow pump in the Phillips band (X g A u) followed by the intercombination to the triplet state – that initiates the scheme – as compared to the inter- combination transition channeling the triplet excitation back to the singlet state. Perusal 1 of the rates in Table 6.3.1 shows that about 2/7 of the population in the A Πu is transfered 1Σ+ ∗ into the excited fiducial level of the X g (v ) state, whereas the rest is transferred back 1Σ+ = into the ground state X g (v 0). Since C2 is a homonuclear diatomic vibrational or ro- 1Σ+ tational radiative relaxation from the exited fiducial level in the X g to the ground level of the same electronic state is forbidden and hence intercombination transitions from the

83 Chapter 6 C2 Emission Features in the Red Rectangle

1

X

g

0

10

-1

10

3

a

u

-2

10

-3

10 PopulationDistribution

-4

10

3

a (v*)

u -5

10

0 5 10 15 20 25 30

Distance [arcsec]

Figure 6.9 – Level populations in our fiducial C2 molecule as a function of distance from the exciting star.

1Σ+ ∗ ff 3Π fiducial level of the X g (v ) state are e ective in populating the a u state. The inter- combination transition is also more efficient in transfering the population into the triplet system as excitation into another electronic level of the same multiplicity can take place. 1Σ+ ∗ Eventually all population that reaches the fiducial level (X g (v )) is transferred into the triplet system. Transitions back into the singlet state are slow and compete with excitation via stellar 3 photons into the d Πg state (Table 6.3.1). The timescales for both transitions, the inter- 3 combination transition back into the ground state and the pump into the d Πg state are of the same order of magnitude at 600 offset. The fast spontaneous emission in the Swan 3 3 3 bands d Πg → a Πu keeps the population very low in the d Πg state as compared to the 1Σ+ 3Π 00 X g and a u. For distances (r), as close as 3 to the exciting star, the pump rate into 3 the d Πg state is about one order of magnitude higher compared to the intercombination 3 transition back into the ground state. The population in the a Πg state is approximately given by,

k2 n(a3Π ) ( + )k7 u = k1 k2 + (6.1) 1Σ k3 n (X g ) ( )k9 + k13 k3+k4

84 6.3 Fluorescent Emission in the Red Rectangle ] 1 − sr 1 − s 2 − ] ] 2 2 − − [cm 0.4 (-11) 0.6 (10) [cm 0.1 0.6 (-7) [erg cm ± ± ± ± 1.0 1.1 (-11) 3.9 (-11) 0.2 (11) 6.3 (10) 0.6 (-7)0.8 (-1) 5.7 (-7) ± ± ± ± 2.8 (-7) 1.1 (11) 5.7 (-11) Transition at 5585 Å Transition at 5635 Å Unit state 4.0 (-1) g Π 3 a 00 – Observed fluorescent intensities at 6 ) 2 abundance 2 C N(C Intensity Column density of fluorescing molecules in the (excited) d Note: Values in brackets give exponents to base 10. a Table 6.3

85 Chapter 6 C2 Emission Features in the Red Rectangle

( ) 600 2 ' 6.7 × 10−2 (6.2) r00

3 where the rates are given in Table 6.3.1. The population in the d Πg state per C2 molecule 1Σ+ can then be expressed, taking into account that nearly all C2 molecules are in the X g ground state; as follows:

3 3 n(d Πg) k n(a Π ) = 9 u (6.3) n(C2) k3 n(C2) k2 k ( + )k7 ' 9 k1 k2 (6.4) k3 k3 ( + )k9 + k13 k3 k4 ( ) 600 4 ' 1.6 × 10−11 (6.5) r00

3 The population in the d Πg state scales thus with the fourth power, since k9 also scales inversely with distance squared. The last equation breaks down near the star when all of 3 the C2 is rapidly transferred into the triplet state (n(a Πu) ' n(C2)). Similarly, for the 3 ∗ excited state d Πg (v ), we have ( ) n(d3Π (v∗)) 00 4 g = . × −12 6 . 3 5 10 00 (6.6) n(C2) r

6.3.3 The Abundance of C2 The average surface brightness in the observed Swan band transitions corrected for the ERE is provided in Table 6.3. These observed intensities can be directly translated into the 3 number of fluorescing molecules in the relevant d Πg states using the Einstein A coeffi- cients of both transitions (∼ 2.52×106 s−1 for the (v0, v00)=(1,2) at 5585 Å; ∼ 1.95 × 106 s−1 for the (v0, v00)=(0,1) at 5635 Å). We have used our fiducial model to translate these into 00 the column density of C2 molecules present at 6 distance from the star in the Red Rectan- gle (eqns. 6.3 and 6.6). The two transitions yield values for the total C2 column density at this distance within 25%, surely coincidental, given the approximate nature of our model. The abundance of C2 in the Red Rectangle is calculated adopting the simple model for the Red Rectangle developed to explain the IR continuum emission (Men’shchikov et al. 2002). In this model, the cone has a number density of ∼ 5 × 105 molecules cm−3 00 at ro = 2.8 with a steep density gradient in the outer extended envelope of the nebula of ρ ∼ r−4. This yields a hydrogen density of 2.4× 103 cm−3 at 600. Assuming circu- lar symmetry for the outflow cone, the total hydrogen column density is 2 × 1021 cm−2. −11 The C2 abundance is then ∼ 5× 10 (Table 6.3). Note that this analysis assumes that the bandhead emission contains most of the emission. As figure 6.5 demonstrates this is reasonable for temperatures between 200 K and 1000 K. In addition, this abundance determination assumes the density distribution of Men’shchikov et al. (2002) and the un- certainty associated with this assumption is difficult to estimate.

86 6.4 Implications

6.4 Implications

−11 00 We have derived a total C2 abundance of ∼ 5×10 at a distance of 6 to the central star (Table 6.3). For comparison, typical column densities and abundances of C2 in diffuse clouds are 3×1013 cm−2 and ∼10−8, respectively, as measured through absorption in the Phillips bands. However, in this comparison it should be noted that absorption line studies are done along a pencil beam against a stellar continuum source (R∗ ' 30 R ) and hence 38 the total number of C2 molecules “observed” is ∼ 3× 10 . In contrast, the total number 00 44 of C2 molecules “observed” in the nebula of the Red Rectangle at 6 is ∼ 4× 10 . So, while only one out of every ∼ 15 C2 molecules is in the triplet system and only one out of every 1011 fluorescing photons is emitted into our beam, we are still sensitive enough to observe the fluorescence because there are six orders of magnitude more C2 molecules in our slit than in the astronomical absorption experiment. While the abundance of C2 is low, this still provides some further insight in the origin of this species. With a UV photodissociation cross section of 0.5 × 10−17 cm2 at 10.6 eV and 0.77 × 10−17 cm2 at 12.2 eV (Pouilly et al. 1983), the photodissociation rate at 600 of −7 −1 C2 is ∼ 4× 10 s if the UV light is dominated by the star (adopting the parameters of −5 Men’shchikov et al. (2002) (Te f f = 7750 K, log g = 1.5) and ' 2× 10 if the putative −4 accretion disk (with M˙ acc = 10 M /yr; Witt et al. (2009)) is responsible for the UV emission. Hence, irrespective of the origin of the dissociating UV photons, the lifetime of a C2 molecule is very short in the nebula (e.g., one day to one month). This is much shorter than the outflow timescale at this distance (R/v ' 3000 yr for an outflow velocity of 7 km/s) and hence, the C2 has to be rapidly replenished. This cannot come from C2H2 or C2H since the photodissociation rate of these species are some ten times faster than that of C2 (Pouilly et al. 1983, van Hemert & van Dishoeck 2008, Nee & Lee 1984). As- suming that the C2 is derived from a major reservoir of the carbon such as PAHs, which contains 10% of the elemental carbon, the (photo)destruction rate of this reservoir has to −6 3 4 be ' 10 that of C2, eg., a dissociation timescale of 10 – 10 yr. Quite reasonable given the outflow timescale. Conversely, the parent species of C2 (and by inference, C2H2) has to represent a major reservoir of the elemental carbon. We note that subtle variations in the profiles of the IR emission features with distance from the exciting star in the Red Rectangle have been interpreted in terms of chemical changes in the PAH family (Song et al. 2007). Quantitatively, adopting a UV absorption cross-section of 7× 10−18 cm2 (C- atom)−1, we estimate an absorption rate of UV photons of ' 4× 10−4 s−1 with hν >6 eV and about ' 10−5 s−1 with hν >10 eV, for a 50 C-atom PAH at a distance of 600 from the star. In order to explain the observed abundance of C2, the photo-destruction probability of PAHs should then be 10−9 and 3× 10−8, respectively. This corresponds to the estimated −6 photo-destruction rate of 10 compared to that of C2. In a recent analysis of PAH de- struction in the ISM, Micelotta et al. (2010) estimated a photo-destruction probability that is slightly higher (5× 10−7). Future laboratory experiments will be instrumental to deter- mine whether such photo–destruction probabilities are realistic for large PAHs. Finally, we note that PAHs are not the only conceivable source of C2H2 parents for the observed C2. Hydrogenated Amorphous Carbon grains, HACs, are also an important reservoir of carbon in the outflow of the Red Rectangle. Recent studies have suggested that cycling

87 Chapter 6 C2 Emission Features in the Red Rectangle of carbon in and out of dust is an important source of small hydrocarbon species (Jones 2009, Pety et al. 2005). Besides photo-destruction, in this case, these grains may also be broken down by shocks. Weak, extended Hα emission indicates the importance of shocks in the outflow itself (Cohen et al. 2004). Grain-grain collisions in the shock may lead to fragmentation and molecule formation (Jones et al. 1996). Finally, we note that the pumping rate in the Phillips and Swan bands – which is at the base of the fluorescence model – well exceeds the photo–destruction rate and therefore many Swan band photons are produced before C2 is destroyed. Hence, excitation of the Swan band system at formation is an ineffective pump relative to the visible fluorescence. The more since only 5% of C2 resulting from photo-dissociation of C2H is formed in the triplet state (Sorkhabi et al. 1997, Wodtke & Lee 1985).

6.5 Conclusions

3 3 0 00 0 00 1. The Swan band transitions of d Πg → a Πu type (v ,v )=(0,1) and (v ,v )=(1,2) type have unambiguously been identified in the outflow of the Red Rectangle proto- planetary nebula. We have simulated the spectra using the rotational contour pro- gram PGOPHER. The simulation was validated against a laboratory study of C2 obtained in a supersonic plasma expansion. These simulations constrain the rota- tional temperature of C2 in the Red Rectangle to be not lower than 200 K and not higher than 1000 K. 2. The mechanism for excitation and de-excitation has been established within the conditions of the Red Rectangle nebula. Our modelling of the excitation and de- excitation mechanism shows that the excitation of the C2 molecule in that environ- 1Σ+ ment is most likely due to transitions leading from the X g ground state of the 1 molecule via the Phillips Band to the A Πu state. From here the higher vibrational 1Σ+ ffi levels of the X g can be e ciently populated and a population transfer from the 1Σ+ → 3Π singlet to the triplet state is possible (X g a u). Once this level is populated, 3 the absorption into the d Πg state is followed very efficiently by spontaneous emis- sion in the Swan band.

3. We have developed a model for the C2 emission in the Red Rectangle. This model has been used to translate the observed strength of the Swan band transitions in the Red Rectangle into a local C2 abundance. The results reveal that the C2 abundance is very low ∼ 5× 10−11. 4. Visible fluorescence in a nebular setting is potentially a powerful probe of the pres- ence of molecular species if suitable electronic states can be populated and depop- ulated through electronic transitions.

5. The inferred abundance of C2 implies that C2 derives from a major reservoir of the elemental carbon in the outflow which is slowly destroyed – likely first into C2H2 and then C2 – on a timescale which is comparable to the outflow timescale of the nebula (3000 yr). Presumably, small PAH molecules represent this reservoir.

88 6.5 Conclusions

6. The experimental set-up shows the potential to simulate the physical conditions as constrained from observations and modelling in the Red Rectangle. In a discharge expansion molecules are created as observed in the emission bands from the Red Rectangle. Rotationally resolved spectra from vibronically excited species show potential for further comparison with the Red Rectangle emission features.

89

The Spatial Distribution of the Optical Emission Features in the Red Rectangle Proto-planetary7 Nebula

N. Wehres, H. Linnartz, H. Van Winckel, A. G. G. M. Tielens submitted to A&A

91 Chapter 7 The Optical Emission Features in the Red Rectangle

Abstract

Context. The Red Rectangle proto-planetary nebula provides a unique laboratory to mon- itor the physical conditions and chemical processes in stellar outflows. Snapshots of the ongoing chemical evolution can be obtained by monitoring spectra as function of the off- set from the central star. Aims. The focus in this study is on the characterization of the narrow optical emission features, that are superimposed on top of extended red emission (ERE). The primary aim is to provide a two-dimensional catalogue of the Red Rectangle spectral appearance for offsets varying from 300 to 2000 from the central star. Methods. The high resolution emission spectra for this catalogue have been obtained through optical long-slit measurements using the New Technology Telescope (EMMI- NTT) in La Silla, Chile. Results. The recorded spectra cover the range between 5550 and 6800 Å. A complete overview of the central band positions and bandwidths (FWHMs) of the stronger narrow emission features is provided. Only some bands are omnipresent in the nebula outflows and other bands only appear further away from the central star. Conclusions. The optical emission bands show strong variations over the nebula. We sug- gest that these variations reflect a spatially resolved active photochemistry where larger species – e.g. PAH molecules – are photolysed, producing daughter molecules which could be the carriers of the well-studied 5800 Å bands. Photolysis of these species may then lead to small hydrocarbon radicals which carry the weaker features that come and go within the nebula.

7.1 Introduction

The Red Rectangle bi-conical reflection nebula is associated with the post AGB binary star HD 44179 (Cohen et al. 1975, Van Winckel et al. 1995). The central object is an interacting binary with a ∼7500 K evolved primary feeding an accretion disk around the likely unevolved secondary. This accretion disk powers a small HII region in the very center of the system (Witt et al. 2009). The star itself is visually obscured by a nearly edge on circumbinary disc (see e.g. (Waelkens et al. 1996, Witt et al. 2009)) and can only be seen in scattered light. The disk leads outflows into different directions, creating the appearance of an X-shaped structure (Cohen et al. 2004). The outflows have attracted special attention as a unique laboratory providing snap- shots for spatially resolved chemical processes in space. In the IR regime a series of emission features has been found between 3 and 15 µm (see e.g. (Peeters et al. 2002) and refs. therein) that are typical for vibrational transitions of polycyclic aromatic hy- drocarbons (PAHs). Furthermore, spectra at longer wavelengths (∼ 20–45 µm) have been associated with crystalline silicates, specifically olivines (MgxFe1−x)2SiO4 and pyroxenes (MgxFe1−xSiO3) (Waters et al. 1998). In the optical regime the spectrum is dominated by extended red emission (ERE), a strong and broad emission feature that stretches over about 250 nm (Schmidt et al. 1980) between 500 and 750 nm and that may originate from

92 7.1 Introduction photo-luminescence of silicate nano particles or carbonaceous grains (see Duley (1985), Witt & Boroson (1990), Witt et al. (1998), Ledoux et al. (1998, 2001), Van Winckel et al. (2002), Witt & Vijh (2004)). Superimposed on the ERE a chaotic pattern of narrow emis- sion features has been found. The emission complex around 5800 Å has been described by Scarrott et al. (1992), Sarre et al. (1995), Van Winckel et al. (2002), Glinski & An- derson (2002) and Sharp et al. (2006). Recently, two of the weaker narrow emission bands have been attributed to C2 emission features starting from rovibronically excited energy levels (Glinski et al. 2009, Wehres et al. 2010a). The majority of the observed and rotationally unresolved transitions, however, has not been assigned. It is likely that these have a molecular origin, presumably transient in nature and it is possible that also rovibronically excited states play a role. Some of these narrow emission features have been proposed as the emission equivalent of some of the well known diffuse interstellar absorption bands (DIBs) (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002), but for an opposing view see also Glinski & Anderson (2002). The DIBs themselves are regularly linked to molecular hydrocarbons and recent studies by Linnartz et al. (2010) and by Maier et al. (2010) aim in this direction. Indeed, the Red Rectangle proto-planetary nebula itself has been found to be carbon rich (Waters et al. 1998) and apart from the spatially resolved PAH emission in the nebula, + identified molecules comprise besides C2 also CH, CH and CN (Balm & Jura 1992, Hobbs et al. 2004). Also oxygen rich molecules were identified like gas phase and solid state CO2 (Waters et al. 1998), OH (Reese & Sitko 1996) and CO isotopes (Waters et al. 1998). Crystalline silicates were detected (Waters et al. 1998) and these are likely stored in the resolved stable circumbinary disk which is in Keplerian rotation around the whole inner system (Bujarrabal et al. 2005). Variations in the spectral characteristics of the emission bands reflect different phys- ical conditions and ongoing chemical processes in the outflows of the Red Rectangle. The intensity of the radiation field, the temperature and the densities will decrease for in- creasing offsets. Higher temperatures and densities will be generally in favour of a more reactive chemistry, but close to the star the stronger radiation field may also efficiently photodissociate species. In parallel, ionization and/or molecular excitation may speed up chemical processes. Further out, in the outflows, this will be less important, but species already formed are a good starting point for the formation of more complex molecules and with the absence of an intense radiation field, these may survive the harsh conditions better. The main goal of the paper is to catalogue the emission bands detected in the Red Rectangle with different offsets from the central star. The second goal of this work is to derive qualitative trends that can be used as an analytical tool to probe ongoing processes in the outflows of the Red Rectangle. The focus is on a systematic and offset dependent overview of the spectral properties (band positions, widths and intensity ratios) for those narrow emission features that can be fitted, more than 40 in total, and is an extension of previous work by Van Winckel et al. (2002). This chapter is organized as follows. Section 7.2 provides details of the observations and the data reduction procedure. The results are described in detail in section 7.3. With this

93 Chapter 7 The Optical Emission Features in the Red Rectangle information in hand, we search in section 7.4 for systematics in spectral changes for the observed bands at different offsets. The observed features are compared with the on-line DIB data as available from the catalogue by Hobbs et al. (2009). The chapter finishes with the conclusions in section 7.5.

7.2 Observations

The 3.5 m NTT (New Technology Telescope) at ESO, La Silla in Chile has been used to record emission spectra of the nebula for different distances from the central star HD 44179. The spectra have been obtained in the nights from February 04 and 05, 2008, using the EMMI (ESO Multi Mode Instrument) spectrograph for medium disper- sion spectroscopy in long-slit spectroscopy mode. The visual magnitude of the central source HD 44179 amounts to 9.02, but this is scattered light only as the aspect angle on the system is such that the disk blocks the direct light from the central object. For telescope settings further away from the central star the flux is decreasing fast and an adaptation of the integration time with distance was necessary. The slit was chosen to be 100 in width and 20000 in length. These long-slit settings had a constant position angle of 105◦ east from north such that the slit length was positioned parallel to the disk (see also Van Winckel et al. (2002) Fig. 1). To increase the S/N, we collapsed the spatially resolved 2-dimensional spectra and these spectra were analysed as proxy of the nebu- lar characteristics at that given distance to the central source. This makes it possible to compare different spectra when converted from the 2-dimensional slit spectrum into a 1-dimensional collapsed spectrum. Different grating settings have been used (see Table 1). Grating 6 with a resolution of R = λ/∆λ= 5000 was used for spectra recorded at the central star and for 300 as well as 700 offset. For larger offsets (1600 and 2000) grating 7 was used that has twice the wavelength coverage and half the resolution. For each spectrum three exposures have been taken in order to correct for cosmic hits and to obtain better S/N ratios. The read-out mode for the spectra has been chosen to be slow with a binning of 2 x 2 that lasted about 18 s. The pixel size of the CCD was 15 x 15 µm. The mean bias level was 250 ADU, the gain 1.25 electrons per ADU with a read out noise of about 3.5 ADU. The data reduction was done using MIDAS in the same way as described by Wehres et al. (2010a). The standard procedure included trimming of the spectra, flat-fielding, bias subtrac- tion, and cosmic hit cleaning. The wavelength correction was provided using He-Ar lamp spectra and the flux calibration was done using a standard star (Hiltner 600). The sky background was subtracted by extrapolating a sky-spectrum outside the region of inter- est. The spectra were corrected for extinction, and different integration times in order to obtain absolute fluxes. Collapsing the 2-dimensional slit spectrum resulted in a higher S/N nebular spectrum, characteristic of that given distance from the central source. The absolute S/N in the 1-dimensional spectra is indicated in Table 7.1 as well. Furthermore, all spectra have been corrected for the barycentric radial velocity shift during exposure. The spectra that were recorded at the NTT are extended with a data-set taken by Van Winckel et al. (2002) at 6 and 1100. For the spectrum taken at 600, the ob-

94 7.2 Observations a N / 64 69 34 31 229 268 207 149 S and absorption lines. ∼ in 1-dimensional spectrum set from the central star. ff o 00 60 60 (s) 500 500 time 2000 3600 3600 2000 Integration λ ∆ 5000 2600 2600 5000 5000 5000 5000 5000 λ/ (Å) 640 640 640 640 640 640 1300 1300 east from north, perpendicular to the nebula’s symmetry axis. ◦ (Å) 6500 6200 6200 5870 6500 5870 6500 5870 N) was determined on a small portion of the spectrum avoiding emission / Wavelength Coverage 6 7 7 6 6 6 6 6 ) – NTT-EMMI telescope settings for spectra taken at the central star and at 3, 7, 16 and 20 7 0 0 3 3 7 00 16 20 ( The signal-to-noise (S a HD 44179 Distance to Grating Central Wavelength Resolution Note: The slit orientation is 105 Table 7.1

95 Chapter 7 The Optical Emission Features in the Red Rectangle servations are limited to the region between 5500 and 6200 Å. Also, a so-far unpublished spectrum by Van Winckel et al. at 1400 distance has been included to the present overview.

7.3 Results

7.3.1 An Offset Dependent Catalogue In Fig. 7.1 overview spectra are shown for offsets of 3, 7 and 2000, illustrating that both ERE as well as the narrow emission bands exhibit a clear distance dependent behaviour. At 300, the ERE is not detected in the spectrum at these wavelengths due to a wavelength shift from blue to the red near the central star. The most dominant part in the 300 spectrum is scattered light from the star. Only further away from the star, this scattering component vanishes and the nebular emission becomes prominent. Some spectral features are evident 00 at 3 , but weak compared to the Na D2 and D1 lines and the Hα line at 5890, 5896 and 6563 Å, respectively, and compared to the spectra at larger offsets where most of the emission bands start to appear. At 700, the ERE is very prominent and even at distances as far as 2000 the ERE is still visible in the spectrum, as is the case for many of the narrow emission features. More details on the evolving narrow emission features are given in Fig. 7.2. Here, the collapsed long-slit spectra are shown for all studied offsets after subtracting the ERE, following a fit with a cubic spline function, interpolating between the different baseline points by choosing the not-a-knot endcurve condition (Behforooz 1995). We selected about 20 baseline points avoiding known narrow emission bands (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002, Sharp et al. 2006) as well as absorption fea- tures. We expect that this procedure works well here, because the focus is on the narrow emission features superimposed on the featureless ERE. The uncertainty in this baseline is a possible source of systematic error. We have estimated the magnitude of this error by repeating this procedure many times. This systematic noise turns out to be small com- pared to the random noise of the observations and therefore only the latter is quoted in Table 7.2 – Table 7.5. The spectral parameters – central band positions and band widths (FWHM) – of these emission features have been derived after fitting the observed bands with Gaussians. The results are summarized in Table 7.2 – Table 7.5, where the uncertainties are given as 1σ error values. For ease of comparison, we have subdivided each individual spectrum shown in Fig. 7.2 into six parts. This is illustrated for an offset of 1100 in Fig. 7.3. For this offset most of the emission bands have become visible in the nebula. In this figure the Gaussian fits are included to illustrate the level of accuracy that can be obtained when fitting these features. In panels b, e and f this is clearly easier to realize than in panels a, c and d. The fitting procedure is not trivial. Several bands consist of overlapping components, complicating unambiguous band assignments. For asymmetric bands (e.g. bands with a blue steep edge and a red-degraded side) two Gaussians are used, one fitted to the steep blue side, which gives the band position, and one to the red-degraded side, which may

96 7.3 Results

20"

7"

Intensity [a.u.] Intensity

3"

Na D & D H

2 1

5600 5800 6000 6200 6400 6600 6800

W avelength [Å]

Figure 7.1 – The Red Rectangle nebula spectra at three different offsets, taken at 3, 7 and 2000. At 300 the scattered light from the star dominates the spectrum, whereas at larger offsets the nebular emission becomes prominent.

reflect rotational substructure, similar to the procedure described in Scarrott et al. (1992) and Sarre et al. (1995). Such unresolved bands are even more challenging to interpret for spectra recorded at offsets of 11, 14, 16 and 2000 where the selected grating has only half of the resolution. Other uncertainties that also need to be taken into account are the exact choice of the baseline and the interpretation of weaker signals, particularly in the faint parts of the nebula. It is therefore important to notice, that we specifically distinguish between bands that can be characterized unambiguously (marked with a (+) in Table 7.2 – Table 7.5) and bands that are visible, but with uncertain fit parameters (marked with a (?)), as well as bands (marked with a (-)) that can also be fitted choosing different sets of Gaussian parameters without substantially reducing the quality of the fit. As these bands have an increased uncertainty, we will not include these in the discussion. It should be noted that band intensities vary with offset and consequently this labelling changes as well. For 700 offset, about 20 bands can be characterized with a label (+). These bands will be discussed below and are indicated in bold face in Table 7.2 – Table 7.5. To illustrate the spectral variations with distance to the central star, three subsections (5775 – 5975 Å, 5980 – 6180 Å and 6450 – 6800 Å) are shown for different offsets in Figs. 7.4, 7.5 and 7.6 as examples of evolving emission bands. The most prominent bands

97 Chapter 7 The Optical Emission Features in the Red Rectangle

20"

16"

14"

11"

7"

Intensity [a.u.] Intensity

6"

3"

*

5550 5700 5850 6000 6150 6300 6450 6600 6750

W avelength [Å]

Figure 7.2 – An overview of all recorded spectra for different offsets to the central star as well as the spectra taken at the central star. The ERE, if visible in the spectra, has been subtracted. in Fig. 7.4 (at ∼ 5800, 5825 and 5852 Å and strong bands centred around ∼ 6380 Å (not shown) and 6615 Å in Fig. 7.6) have been described in detail before (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002, Sharp et al. 2006). Several of the narrow and weaker features have also been reported in former work for different offsets (Scarrott et al. 1992, Sarre et al. 1995, Glinski & Nuth 1997, Van Winckel et al. 2002, Sharp et al. 2006). Interesting in the present offset dependent study is the clear spectral variation of these and other bands with distance to the central star. Fig. 7.6 for example shows a clear intensity variation of the emission band that appears around 6615 Å relative to the Hα line emission at 6563 Å. A number of specific observations are reported below.

• The emission bands at ∼ 5585 and ∼ 5635 Å that recently were identified as orig- inating from unresolved rovibronic progressions (0-1 and 1-2) within the d3Π → a3Π Swan system (Sarre 2006, Glinski et al. 2009, Wehres et al. 2010a) do not re- veal clear band shifts for increasing offsets. The bands are only observed between 3 and 700 and are not detected for larger offsets (Wehres et al. 2010a). • The stronger emission bands in Fig. 7.3 (panel b, e and f), located at around ∼ 5800, 5825, 5852, 6380 and 6615 Å, have been reported previously to shift towards the

98 7.3 Results

(b) (a)

5575 5600 5625 5650 5675 5700 5725 5750 5775 5800 5825 5850 5875 5900 5925 5950 5975

(c) (d)

Intensity[a.u.]

6000 6020 6040 6060 6080 6100 6120 6140 6150 6175 6200 6225 6250 6275 6300

(e) (f)

6360 6375 6390 6405 6420 6435 6500 6550 6600 6650 6700 6750

W avelength [Å]

Figure 7.3 – An overview of Gaussian fits to the spectral features observed for 1100 offset. The frequency domain is divided in six nearly equally broad regions. The sum of all fits is given as a thick black line overlaid on the observational spectrum.

blue and to narrow with distance from the central star (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002, Sharp et al. 2006) or at least to vary with distance (Glinski & Anderson 2002). These trends are confirmed here: particularly when the bands already appear for an offset of 300, the bandshift becomes very obvious. Many of the other bands presented here show a similar effect, although less pronounced.

• The spectra at 300 are rather different from other offsets and only a few emission bands can unambiguously be characterized: the C2 lines, the Na D1 and D2 lines, the Hα line, as well as photospheric absorption lines (see Fig. 7.2 and Fig. 7.4). The photospheric absorption bands that are strongly visible at 300 between 5900 and 5975 Å are in general not observed for larger offsets (see e.g. Fig. 7.4).

• Three emission features between 5990 and 6170 Å are detected, a rather unstruc- tured band located between 5990 and 6075 Å, a pronounced emission band at around 6108 Å, and a narrow emission band at 6162 Å (Fig. 7.5). Most notice- able in this region is the clear relative intensity increase (1:2 to 2:1) of the 6162 Å

99 Chapter 7 The Optical Emission Features in the Red Rectangle

20"

11"

Intensity [a.u.] Intensity

3"

5775 5800 5825 5850 5875 5900 5925 5950

W avelength [Å]

Figure 7.4 – A comparison of the narrow emission bands in the 2nd spectral range (5775 – 5975 Å) at 3, 11 and 2000 offset. The Na doublet lines are visible at ∼ 5890 and 5896 Å. In the 300 spectrum photospheric absorption lines are still visible, but become weaker further away from the central star. Note: For displaying purposes the y-scale is different for each panel.

band with respect to the 6108 Å band for 6 and 1400 distance from the exciting star. Whereas the emission band at 6108 Å is not visible at 16 and 2000, the emission band at 6162 Å can still be detected (even though it is rather weak) for an offset of 2000. The broad and unstructured band starting at 5990 Å has a steep incline and a broad plateau-like appearance that most likely is the result of a series of at least three blended emission bands. Several separate Gaussians have been fitted to reproduce this generally rather weak pattern (see Fig. 7.3 (panel c) and Fig. 7.5). This emission band is only detected for offsets between 6 and 1400. • Four emission bands can be distinguished unambiguously around 6200 – 6420 Å (see Fig. 7.3 (panel d)) at 6196, 6204, 6222 and 6235 Å. The spectrum at 700 only shows a first onset of these features that progress further outwards in the nebula. These bands are special, since their appearance in the nebula is only clear at larger distances from the star. Especially the bands at 6196 and 6204 Å are detected starting from 1100 onwards. The spectral lines at 6196 and 6204 Å are absent in the

100 7.3 Results

14"

11"

Intensity [a.u.] Intensity

6"

5980 6000 6020 6040 6060 6080 6100 6120 6140 6160 6180

W avelength [Å]

Figure 7.5 – A comparison of the narrow emission bands in the 3rd spectral range (5980 – 6180 Å) at 6, 11 and 1400 offset towards the central star. Note: For displaying purposes the y-scale is different for each panel.

spectra reported by Scarrott et al. (1992), but were reported by Van Winckel et al. (2002) to start at 1100.

• Other stronger emission features are observed at ∼ 6379 and 6400 Å, and weaker bands at 6420 and 6446 Å. The stronger emission band at ∼ 6379 Å shows a very narrow upper part, whereas the base of the band seems to be broadened to both sides. There is another emission band located at 5913 Å that has a similar profile. To determine the band position here, three bands, a central one, and two bands on either side have been fitted to give the band position as precisely as possible (see Fig. 7.3 (panel e)).

• The 6450 to 6800 Å range is shown in Fig. 7.6. An emission band at 6553 Å is close in band position and partially overlaps with the Hα transition at 6563 Å. It is noteworthy that the Hα line is strong at 300, but then weakens fast for larger offsets. A stronger emission band at 6615 Å is accompanied by a series of weaker and equidistant bands (between 6633 and 6662 Å) with an approximate 60 cm−1 separation.

101 Chapter 7 The Optical Emission Features in the Red Rectangle 1.0 3.9 0.4 0.5 0.4 0.3 0.6 1.0 1.3 0.6 1.4 0.5 0.3 1.0 0.1 0.1 0.3 1.4 0.7 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Å FWHM 3.1 9.3 3.6 2.3 16.4 5.4 3.4 12.5 3.2 9.2 8.1 ) ) ) ) ) ) ) ) 2.3 ) 2.3 + + + + + + + + + 0.6 (?)1.5 (?) 9.8 0.6 (?) 13.4 5.7 0.2 ( 0.2 ( 1.0 0.2 ( 0.3 0.2 ( 0.2 0.1 ( 0.1 ( 0.2 ( 0.4 (?)0.5 (?) 9.8 4.5 0.5 (?) 13.1 0.2 ( 0.2 0.3 ( ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 Å 5798.5 5802.0 5650.0 5767.2 5825.4 5770.5 5852.2 5854.1 5880.6 5883.1 5912.2 0.6 5597.8 1.0 0.5 1.0 0.4 0.3 0.6 1.42.5 5717.6 5740.9 0.4 0.7 0.3 0.5 1.0 0.1 5890.0 0.1 5896.0 0.3 0.50.7 5936.2 5946.8 0.3 0.9 5752.0 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Å FWHM 11 2.8 8.9 3.8 19.3 2.0 8.9 3.2 12.3 2.6 9.5 8.7 ) ) ) ) ) ) ) )) 2.0 12.1 ) 2.4 ) 0.9 ) 1.0 + + + + + + + + + + + + 0.2 ( 0.3 ( 0.3 ( 0.3 ( 0.2 0.3 ( 0.3 ( 0.2 ( 0.6 (?)1.0 (?) 13.5 0.6 (?) 12.6 7.3 0.3 0.2 ( 0.3 0.3 ( 0.3 0.1 ( 0.1 ( 0.2 ( 0.3 (?)0.4 (?) 4.9 6.5 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 Å 5798.9 5651.5 5803.0 5826.1 5768.0 5771.6 5853.0 5854.9 5880.9 5885.9 5913.0 0.4 5598.6 0.5 1.34.0 5719.4 5741.3 0.6 0.4 0.2 0.5 5584.8 0.30.8 5634.9 0.2 1.30.4 5753.3 0.5 0.2 0.4 0.5 0.1 5890.0 0.1 5896.0 0.70.7 5936.9 5945.7 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Å FWHM 7 8.8 6.6 12.4 9.4 ) 2.7 ) 1.8 ) 13.9 )) 2.4 3.9 ) 19.9 ) 2.5 ) 3.2 ) 2.5 ) 0.9 ) 0.9 ) 10.9 + + + + + + + + + + + + 0.2 ( 0.3 ( 0.3 ( 0.3 ( 0.3 ( 0.2 0.2 ( 0.6 (?)1.7 (?) 10.5 0.4 (?) 17.4 0.2 ( 8.0 0.4 0.2 ( 0.2 0.2 ( 0.2 0.1 ( 0.1 ( 0.2 ( 0.4 (?)0.4 (?) 6.3 6.7 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 Å 5798.8 5651.5 5718.1 5740.0 5752.7 5767.7 5771.2 5855.0 5880.9 5885.9 5913.1 5936.8 5946.7 1.1 5598.6 0.4 5802.8 1.0 5826.0 0.4 5852.8 0.5 5584.9 0.3 5634.7 0.1 5889.9 0.1 5895.9 ± ± ± ± ± ± ± ± Å FWHM 6 13.1 ) 1.7 ) 10.4 ) 24.7 ) 11.0 ) 1.0 ) 1.0 ) indicates definite fit parameters, (?) indicates likely fit parameters, (-) indicates uncertain fit parameters. ) 1.5 + + + + + + + + – Summary of narrow emission features superimposed on the ERE 0.5 0.2 ( 0.2 ( 0.4 ( 0.4 ( 0.1 ( 0.1 ( 0.2 ( ± ± ± ± ± ± ± ± Note: ( 00 Å 3 First Range 5585.1 Second Range 5602.2 5634.7 5805.9 5827.4 5857.5 5890.0 5896.0 Table 7.2

102 7.3 Results 1.2 2.4 2.5 1.3 3.0 2.7 6.0 1.0 0.7 3.0 0.3 0.6 0.4 1.0 0.7 0.3 1.5 2.9 0.2 0.6 0.9 1.4 0.6 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 12.0 3.9 5.5 FWHM Å 6.0 5.7 4.0 8.9 8.4 3.4 5.1 ) ) ) ) ) ) ) ) 3.0 ) 2.4 ) 6.6 ) 3.1 ) 2.2 + + + + + + + + + + + + 1.9(-) 0.5 (-)1.1 (-)3.0 7.4 (-) 17.4 27.7 0.3( 0.3 ( 0.4 ( 0.3 ( 0.7 ( 0.3 ( 1.2 (?) 26.7 0.5 ( 0.2 ( 0.4 ( 3.0 (-) 0.2 ( 0.3 ( 0.7 0.7 (-) 6.7 0.4 (-)0.8 (-)1.0 16.4 (-) 16.1 16.3 0.4 ( ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 6196.9 63670.0 Å 6203.9 6552.0 6107.9 6377.8 6220.7 6161.6 6234.8 6383.8 6614.9 6398.0 6618.3 6446.1 3.55.2 6007.0 6043.6 0.8 3.8 6573.1 2.32.6 6634.4 3.1 6661.6 6708.4 2.6 0.9 5992.1 0.6 1.7 0.3 6562.8 1.0 0.7 3.7 0.3 1.4 1.2 1.8 6422.4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 6.8 4.5 FWHMÅ 11 15.3 5.5 8.5 3.9 7.3 3.5 5.3 8.0 ) ) ) ) ) ) ) ) 2.4 + + + + + + + + 2.3 (-) 0.4 ( 0.4 (-)0.5 (-)1.2 8.5 (-)0.3 18.3 ( 33.9 0.3 ( 0.3 ( 0.8 ( 0.9 (?) 35.8 0.2 ( 0.5 ( 1.9 (-) 1.0 ( 0.9 0.9 (-) 3.2 1.0 (-) 23.4 1.3 (-)0.7 (-) 21.6 12.0 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 6370.0 6553.9 Å 6107.9 6378.9 6562.8 6221.7 6575.6 6162.6 6615.1 6386.1 6398.2 6618.4 6421.3 6631.6 6663.8 6709.6 3.32.9 6009.1 1.5 6040.8 0.8 5994.4 1.0 ± ± ± ± ± FWHM 7 Å ) 15.0 ) 5.0 + + 0.5 (-) 7.8 1.0 (-)1.1 (-) 16.1 0.5 ( 48.0 0.3 ( ± ± ± ± ± 00 6 5994.9 Å 6010.0 6047.3 6107.5 6162.6 0.3 ± Fifth Range FWHM Å ) 1.2 ) indicates definite fit parameters, (?) indicates likely fit parameters, (-) indicates uncertain fit parameters. + + – Summary of narrow emission features superimposed on the ERE 0.3 ( ± Note: ( 00 3 Third Range Fourth Range Sixth Range Å 6563.6 Table 7.3

103 Chapter 7 The Optical Emission Features in the Red Rectangle 0.9 1.4 2.9 0.1 0.1 0.3 0.5 0.3 1.1 ± ± ± ± ± ± ± ± ± 9.7 14.3 Å FWHM )) 2.3 2.3 ) 17.7 ) 2.7 ) 3.7 + + + + + 0.0 ( 0.0 ( 0.3 0.3 ( 0.6 0.4 (?) 10.0 0.2 ( 0.2 ( 0.4 (?) 9.2 ± ± ± ± ± ± ± ± ± 00 Å 0.8 5853.5 1.0 5911.7 0.7 5824.8 0.10.1 5890.0 5896.0 0.3 5801.0 0.6 1.0 0.2 5798.2 0.5 5851.7 0.7 5881.0 1.2 ± ± ± ± ± ± ± ± ± ± ± ± 9.2 12.5 6.4 FWHMÅ 20 )) 2.3 2.3 ) 17.3 ) 2.8 ) 3.2 ) 2.7 + + + + + + 0.3 0.4 0.2 0.1 ( 0.1 ( 0.4 (?) 11.6 0.5 (?) 3.7 0.3 ( 0.5 (?) 5.7 0.2 ( 0.2 ( 0.3 ( ± ± ± ± ± ± ± ± ± ± ± ± 00 Å 1.0 5853.6 1.6 3.5 0.3 5824.9 1.0 5935.1 1.8 0.1 5896.0 1.7 0.4 1.0 0.5 5801.1 0.90.1 5881.0 5890.0 1.0 5946.2 0.5 5851.8 0.2 5872.9 1.0 0.6 5798.2 0.3 5912.2 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 8.6 1.5 5.5 12.0 3.3 16.6 3.5 4.7 9.3 FWHMÅ 16 4.0 3.0 ) ) ) ) ) ) ) ) 2.3 ) 2.3 ) indicates definite fit parameters, (?) indicates likely fit parameters, (-) indicates uncertain fit parameters. + + + + + + + + + + – Summary of narrow emission features superimposed on the ERE 0.1 ( 0.2 ( 0.2 ( 0.6 0.1 ( 1.7 (?)0.6 (?) 15.5 4.1 0.2 0.2 ( 0.2 ( 0.2 0.2 ( 0.3 0.5 (?)0.5 (?) 10.4 4.6 0.9 (?) 10.4 0.7 (?)0.3 ( 12.8 0.2 ( ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note: ( 00 5896.0 5912.0 5890.0 5766.9 5769.9 5947.1 5741.2 5751.6 5802.1 5825.4 5852.0 5854.3 5880.9 5886.1 5935.6 5716.2 14 First Range 5597.7 Second Range 5798.3 Å 5649.9 Table 7.4

104 7.3 Results 4.3 3.0 1.2 0.2 0.5 1.4 1.0 2.3 0.5 1.5 0.6 0.2 0.3 1.3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± Å FWHM 3.4 ) 2.0 ) 2.7 ) 3.1 ) 4.2 ) 4.8 ) 3.4 + + + + + + 0.1 (?) 1.3 0.2 (?) 1.9 0.5 ( 0.6 (?) 8.9 1.0 (-) 6.4 1.3 (?) 12.8 0.8 (?) 5.4 3.1 (-) 26.0 0.3 ( 0.2 ( 0.6 0.2 ( 0.5 ( 0.3 ( ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 6709.4 Å 3.6 6634.7 2.1 6370.5 1.5 6161.4 1.52.0 6204.1 6220.1 1.1 2.0 6236.1 3.8 6383.2 0.3 6563.1 0.42.2 6614.5 6619.6 1.5 6377.6 2.1 6397.6 1.1 6550.1 ± ± ± ± ± ± ± ± ± ± ± ± ± ± FWHMÅ 20 5.6 ) 4.0 ) 2.3 ) 3.2 ) 5.6 ) 6.3 ) 5.1 + + + + + + 1.0 (?) 3.3 1.0 ( 0.6 (?) 8.0 1.9 (-)0.5 (-) 10.8 4.1 1.2 (-) 7.0 1.3 (-) 17.3 1.4 (-) 2.3 0.3 ( 0.2 ( 2.5 0.2 ( 0.5 ( 0.6 ( ± ± ± ± ± ± ± ± ± ± ± ± ± ± 00 Å 5.5 1.0 1.1 3.2 2.3 2.01.5 6636.2 6660.4 0.6 6161.5 1.02.0 6204.1 6219.6 0.6 1.0 1.5 6235.6 2.0 6375.9 2.9 6386.5 2.3 0.5 6563.3 0.21.5 6614.7 6617.8 1.0 6377.9 0.6 6398.2 0.6 6550.8 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 15.5 3.9 9.0 FWHM 16 6.0 7.7 8.0 3.4 5.7 Å 4.4 3.7 6.9 ) ) ) ) ) ) ) ) 3.7 ) 3.0 ) 3.3 ) 5.3 ) 2.6 ) indicates definite fit parameters, (?) indicates likely fit parameters, (-) indicates uncertain fit parameters. + + + + + + + + + + + + + – Summary of narrow emission features superimposed on the ERE 0.4 ( 0.3 ( 0.5 ( 2.9 (-) 25.0 0.5 ( 0.3 ( 0.4 ( 1.5 (-) 11.1 1.0 (-) 20.6 0.7 ( 0.5 ( 1.0 (-) 0.3 ( 3.0 (-) 0.3 ( 1.1 0.3 ( 0.3 ( 1.8 (?) 26.6 1.3 0.8 (-)1.0 (-) 14.7 15.4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note: ( 00 6106.8 6161.6 6220.3 6708.4 6043.6 6204.0 6446.0 6614.9 14 Third Range 6002.2 Fourth Range 6196.9 Fifth Range 6369.8 Sixth Range 6552.0 Å 6234.9 6377.7 6384.3 6398.0 6421.3 6562.8 6571.2 6618.3 6633.7 6661.6 Table 7.5

105 Chapter 7 The Optical Emission Features in the Red Rectangle

16"

14"

11"

Intensity [a.u.] Intensity

7"

3"

6450 6500 6550 6600 6650 6700 6750 6800

W avelength [Å]

Figure 7.6 – A comparison of the narrow emission bands in the 6th spectral range (6450 – 6800 Å) at 3, 7, 11, 14 and 1600 offset towards the central star. Note: For displaying purposes the y-scale is different for each panel.

7.4 Discussion

7.4.1 Spatial Behaviour of the Emission Bands The shifts in band position, changes in FWHM and intensity variations of the selected (+) bands between 7 and 1400 are summarized in Fig. 7.7. More details on the band positions and FWHMs of the (?) and (-) bands are available from Table 7.2 – Table 7.5. In Fig. 7.7 the peak position and FWHM of the asymmetric bands fitted by multiple Gaussians are not the mean values of the fitted components. Instead, we determined the peak position and FWHM by fitting a single component. The intensity ratio shown in Fig. 7.7 is determined for the integrated intensities of each band at 1400 relative to that at 700. The spectral behaviour of the bands summarized in Fig. 7.7 only partially exhibits similar trends, but some general features can be concluded. Most of the bands become visible at larger offsets (i.e. beyond 300). As from 1400 onwards, several bands decrease in intensity. From the upper panel it can be concluded that for the larger offset (i.e. 1400) most band peak positions shift to the blue compared to 700. The shifts amount typically to a few cm−1. This is illustrated for the band at 5768 Å in Fig. 7.8. This is a less intense emission

106 7.4 Discussion

6

4 ] -1

2

[cm

0

Band shift

-2

80

60

40

20

0 [%]

-20 FWHM FWHM

-40

-60

-80

3.0

2.5

2.0

1.5

[%]

1.0

0.5

0.0 Intensity 14"/7"Intensity

5600 5800 6000 6200 6400 6600

W avelength [Å]

Figure 7.7 – The diagram shows the relative spectral changes for band peak position (upper panel, cm−1) and bandwidth (FWHM) (middle panel, in %). The lowest panel gives the intensity changes of the emission bands by comparing the values at 1400 and 700. The indicated uncertainties are 1σ- errors. The open squares include the band complex between 5800 and 5900 Å, the band at 6378 Å and the emission band at ∼ 6615 Å, that all have been reported previously.

band that exhibits the same trends as observed previously for the stronger emission bands. Three spectra are shown for offsets of 7, 11 and 1400, respectively. The bandshift goes together with a narrowing with distance to the central star. This narrowing is also observed for other bands (Fig. 7.7, middle panel). The bandshift is consistent with a scenario in which a temperature effect is at play resulting in a shift of the Boltzmann maximum. As the spectra concern emission lines, care should be taken to relate these trends directly to changes of the temperature in the ground state. Nevertheless, it is likely that the observed shifts are correlated with a decrease of the temperature in the outflows. Naively, one would expect that the bandshift always goes together with a narrowing, since both are indicative for the ro-vibronic excitation temperature. However, some bands in our survey (∼ 5651, 5913 and 6162 Å) only show a bandshift and no narrowing, which is puzzling in this respect. Fig. 7.9 shows the emission band at ∼ 5651 Å for distances at 7, 11 and 1400 examplewise. The lowest panel in Fig. 7.7 reflects the intensity variations of the emission bands between 7 and 1400. The absolute intensities at 1400 with respect to the absolute intensities

107 Chapter 7 The Optical Emission Features in the Red Rectangle

Table 7.6 – Intensity variations of the narrow emission bands seen towards HD 44179 between 7 and 1400. Intensities of the emission bands are given relative to the band at 5800 Å. For the latter band the absolute intensity is given.

Band position in Å 700 1100 1400 5598 0.07 0.07 0.07 5651 0.03 0.05 0.06 5768 0.05 0.05 0.06 5800 1.80E-13 8.70E-16 1.10E-15 5826 0.45 0.33 0.34 5852 0.70 0.54 0.55 5881 0.21 0.22 0.24 NaI D2 0.69 0.53 0.54 NaI D1 0.64 0.41 0.41 5913 0.20 0.19 0.19 5936 0.10 0.10 0.09 5990 0.07 0.07 0.06 6108 0.07 0.16 0.07 6162 0.04 0.11 0.10 6204 – 0.15 0.16 6220 0.05 0.11 0.14 6234 0.03 0.06 0.06 6379 0.10 0.26 0.26 6399 0.06 0.12 0.12 6446 – 0.06 0.04 6553 0.03 0.11 0.15 Hα 0.07 0.12 0.12 6575 0.03 0.08 0.09 6615 0.16 0.55 0.54 6633 0.03 0.09 0.08 6662 0.02 0.08 0.07 6710 0.03 0.06 0.08

a The absolute intensities of the 5800 Å emission band are given in: erg s−1 cm−2. b The uncertainties of the relative intensity variations are about 0.02.

108 7.4 Discussion of this band at 700 are presented. The intensities for all emission bands decrease between 7 and 1400 and only make up a fraction of the intensity at 700 (the intensity at 1400 is a few percent of the intensity at 700). Two emission bands at 6220 and 6550 Å show a less steep decrease in intensity compared to the other emission bands.

7.4.2 Constraints on the Carriers of the Emission Bands At this stage, it is good to ask the question, what constraints can be made from these measurements, particularly as only a very few of the narrow emission features have been assigned to specific carriers. First of all, a carrier has to be reasonably abundant. In addition, the carrier molecule must be able to emit light. The UV light present around the star is capable of exciting a molecule rovibronically after which fluorescence may take place, but only when other relaxation processes (e.g., internal conversion because of a short lifetime of the excited state, or intersystem crossing) or collisional de-excitation (unlikely in the Red Rectangle outflows (Wehres et al. 2010a)) do not take over. Fluorescence is also enhanced if large geometry changes in the potential energy surface occur between the ground state and the excited state, since this reduces the overlap of the electronic states and consequently, fast internal conversion is not possible. An increase or decrease of the band intensity correlates with an increasing or decreas- ing population density for a particular transition. In most cases intensity variations reflect a higher or lower abundance of the carrier in the nebula, but in the case of excited states, this may also be the consequence of a change in pumping efficiency. For this reason it may be helpful to look at the relative intensity variations of the emission bands. In Ta- ble 7.6, the relative intensities of all emission bands are given compared to the emission band at 5800 Å. The absolute intensities for the emission band at 5800 Å are indicated for the distances between 7 and 1400. Here it is possible to see similar trends of the emis- sion bands, i.e. bands that form at later stages in the nebula, decrease or increase with distance relative to the band at 5800 Å. Consequently, the normalized intensity behaviour of a single emission band may give additional information on the molecular origin of a transition. A non-varying emission intensity with respect to the band at 5800 Å indicates the presence of the carrier over a wide range of conditions; the carrier may be very sta- ble, or a pumping mechanism may be omnipresent, as can be seen for the bands between 5913, 5936 or 5990 Å, which hardly change their intensity relative to the emission band at 5800 Å and seem to develop together.

7.4.3 The Red Rectangle Emission Bands and the DIBs It has been proposed in the past that the spectral features observed in the Red Rectangle can be used to compare to some of the diffuse interstellar band (DIB) absorptions. This method has been questioned, as DIBs originate from a lower (most likely the ground) state, whereas emission features start from an excited state; i.e., the absorption pattern (both frequency position and FWHM) not necessarily has to coincide with the emission

109 Chapter 7 The Optical Emission Features in the Red Rectangle

14"

11"

Intensity [a.u.] Intensity

7"

5760 5762 5764 5766 5768 5770 5772 5774 5776 5778 5780 5782 5784 5786

W avelength [Å]

Figure 7.8 – The emission band at 5768 Å. The spectra and fits for three different offsets at 7, 11 and 1400 are shown. The band was fitted using two Gaussians, indicated by the dotted lines. The bandshift is indicated with the vertical dotted lines showing the band peak position and the band onset for the three different distances.

feature. Furthermore, the chance of an accidental overlap is far from non-zero, given the large number of DIBs and Red Rectangle emission features in this frequency do- main. This also becomes visible from Fig. 7.10, where the present emission data and the available DIB spectra are co-plotted. Only a few (of the stronger) bands show a partial spectral overlap, whereas FWHM-values generally differ considerably. Some of the emis- sion bands have been compared to the DIBs in previous studies; i.e., the emission bands at ∼ 5800, ∼ 5850, ∼ 6380 and ∼ 6615 Å that have been marked with an asterisk. All emission bands show a wider FWHM than observed for the corresponding DIBs. Nev- ertheless the onset of these emission bands is very close to the onset of the DIBs. With respect to Fig. 7.10 we can add several emission bands marked with an exclamation mark to this list: ∼ 5767, ∼ 5913, ∼ 6162, ∼ 6196, ∼ 6204, ∼ 6240 and at ∼ 6399 Å. All these emission bands have a comparable onset to the blue as the emission bands discussed pre- viously (see (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002, Sharp et al. 2006)).

110 7.4 Discussion

14"

11"

Intensity [a.u.] Intensity

7"

5644 5646 5648 5650 5652 5654 5656 5658 5660 5662 5664

W avelength [Å]

Figure 7.9 – The emission band seen at 5651 Å exhibits a blue shift with distance to the central star. A narrowing could not be determined.

7.4.4 Summary Table 7.7 summarizes all observed data. The table gives the peak position of the emission bands in the first column. The closest and furthest offset detection of the bands are men- tioned in column 2 and 3. The profiles given in column 4 reflect the average profile of the band over all spectra as asymmetric, symmetric or red-degraded, whereas two bands (at 5913 and 6380 Å) show broad wings on either side of the emission band. Column 5 gives an indication of the width of the emission band (narrow or broad). Columns 6-8 give a summary of the trends mentioned in Fig. 7.7 and the last two columns (9 and 10) give close DIB positions and their respective FWHM, as taken from Hobbs et al. (2009). The main results from this table and the previous discussion are summarized as follows: • Most of the bands show a symmetric profile. The most significant ones (at ∼ 5800 Å and ∼ 6615 Å) show red-degraded profiles. • Most of the emission bands show a blueshift with distance from the central star. • Most of the bands that show a blueshift also show a narrowing with distance from the central star, but not all of them. • All of the bands decrease in absolute intensity with distance to the central star.

111 Chapter 7 The Optical Emission Features in the Red Rectangle

• Many emission bands are close in band position to some of the DIBs, although the FWHM is quite different between emission bands and DIBs.

• Most of the emission bands are observed further away than 300 from the central star.

Overall, the emission bands show large spatial variations; in particular, bands ap- pear and disappear in the outflows of the nebula. In our opinion, this reflects an active (photo)chemistry rather than variations in the pumping mechanism. It may be instructive to recall here the behaviour of the C2 Swan-bands (Wehres et al. 2010a). Specifically, the Swan-bands at 5585 and 5635 Å were detected between 3 and 700 only. The cross-section for photo-destruction, however, indicated that the molecule is rather fragile and is destroyed very efficiently in the harsh radiation field of the central star. Wehres et al. (2010a) concluded that the C2 lifetime in the nebula is too short given the slow outflow velocity in the nebula and a replenishment (e.g., from a PAH reservoir) is needed in order to explain the appearance of the Swan-bands in several spectra. An active photo-chemistry is needed to account for this and to break down PAHs on a timescale of 3 4 10 until 10 years, producing C2H2 which is then quickly photolysed to C2 that also may play a role in other regions in the ISM (Jones 2009, Pety et al. 2005). In this respect it may be possible that the strong emission bands seen around 5800 Å are the photo-products of PAH molecules, which are photolysed throughout the nebula. Photolysis of the “daughter” molecules then may lead to the presence of smaller hydro- carbons (“granddaughter” molecules), which are detected at larger distances (≥ 300) and may be the carrier of the weaker emission bands that come and go in the outflows of the nebula. The possible correlation of the emission bands with the DIBs (Scarrott et al. 1992, Sarre et al. 1995, Van Winckel et al. 2002) suggests that the carriers of the DIBs in diffuse interstellar clouds are also due to smaller hydrocarbon species. This is in line with the recent possible identification of the λ5450 DIB with the C3H2 radical (Maier et al. 2010). We also note that the 60 cm−1 spacing in the 6615 Å band – if interpreted as a vi- brational progression – is indicative of a rather small hydrocarbon chain. C3 itself for −1 example has low lying bending modes of about 60 cm . While the evidence for C3 is not uncontroversial (Glinski & Nuth 1997) lower vibrational modes of larger carbon chains shift generally with the size of the chain and a 60 cm−1 shift would correspond to a size of 7 to 8 carbon atoms.

7.5 Conclusions

The offset dependent spectra for the Red Rectangle proto-planetary nebula are presented for 3, 6, 7, 11, 14, 16 and 2000 distance from the central star. The spectra show that spec- tral features change as function of distance:

• Only some of the emission bands are omnipresent and are found at 300 offset, and most bands become visible from 700 onwards.

112 7.5 Conclusions d Å – 0.73 0.6 1.17 0.88 0.94 1.03 0.57 – 1.08 – FWHM 1.3 4.50 0.91 0.65 0.93 – d 6376.21 Å – 5910.61 6108.19 6159.80 6221.02 6396.88 6553.95 – 6613.70 – 5652.22 5766.98 5797.20 5818.81 5849.88 – 00 4.8 1.2 8.9 [%] 7-14 00 blueshiftblueshiftblueshift 1.9 blueshift 1.5 3.0 7.6 stable 7–14 blueshift 3.4 00 FWHM bandshifts Intensity decrease DIB position stable stable stable narrowingbroadeningstable blueshift blueshiftnarrowing 2.2 blueshift 7.2 3.7 not resolved stable ∆ 7–14 stable not resolved stable narrowingnarrowingnarrowing blueshiftnarrowing blueshiftnarrowing blueshift 2.3 blueshift 2.2 blueshift 1.7 1.7 1.7 c b n n n n n n n Width n n n n b n n b broad. = broad wingss broad wingss s stable broad wingss broad wingss s narrowingrd blueshift 3.6 s profile as s rd rd s rd s a narrow or b = ) ) ) ) ) ) ) ) ) ) ) ) ) ] + + + + + + + + + + + + + 00 red-degraded profile showing a steep blue edge. = 20 (?) 14 ( 20 (?) 20 (?) 20 ( 20 ( 20 ( 20 ( 20 ( 20 ( in [ furthest detection 14 ( 20 ( 14 ( 20 ( 20 ( 20 ( 20 (?) a asymmetric, rd = ] ) ) ) ) ) ) ) ) ) ) ) ) ) ) 00 + + + + + + + + + + + + + + 6 ( 6 ( 6 ( 7 ( 7 ( 7 ( 7 ( central star 7 ( central star in [ 6 ( central star 6 ( 3 ( 3 ( 3 ( 6 ( symmetric, as – Summary of the narrow emission features superimposed on the ERE ) indicates definite detections, (?) indicates a likely detection. = 1 2 + (Hobbs et al. 2009). ( s Column 5 gives in indication whether a band is n a b c d α 5913 6108 6162 6220 6379 6399 6553 H 6615 NaI D Å Band position closest detection 5651 5767 5800 5826 5852 5881 NaI D Table 7.7

113 Chapter 7 The Optical Emission Features in the Red Rectangle

!

5570 5580 5590 5600 5610 5645 5650 5655 5660 5670 5700 5730 5760 5790

* *

5790 5795 5800 5810 5820 5830 5840 5850 5855 5860 5865

!

! Normalized Intensity Normalized

5900 5910 5920 6000 6050 6100 6155 6160 6165 6170 6175

! *

*

!

!

!

6200 6225 6250 6350 6400 6450 6500 6550 6600 6650 6700

W avelength [A]

Figure 7.10 – The Red Rectangle emission spectrum taken at 1100, subdivided into 12 separate plots for convenience. The emission bands of the Red Rectangle are plotted in black. Overplotted is the “inverted” DIB spectrum as thick black line (Hobbs et al. (2009)). Bands that show an asterisk have been compared to DIBs previously. Bands that show an exclamation mark show similar trends as the DIBs and may well be added to this list.

• The majority of the bands that can unambiguously be fitted, shifts a few cm−1 to the blue for larger offsets.

• Several of these bands simultaneously become narrower. As the spectra concern emission lines, care has to be taken to directly relate these ob- servations to e.g. temperature changes in the ground state, but it is likely that shifts and narrowing are due to a varying temperature in the outflows. Several bands are observed that are clearly indicative for ongoing chemical and physical processes in the outflows of the Red Rectangle: Some bands appear, others vanish. Fol- lowing the interpretation of the C2 Swan-bands, we interpret this as reflecting a chemical process (a molecule is formed or destroyed) and as such the emission bands are a good visualization of an ongoing active (photo)chemistry in the outflows of the Red Rectangle.

114 Nederlandse Samenvatting:

Optisch Onderzoek Aan Astrofysisch Relevante Radicalen

Inleiding

Optische sterrenkunde is werken met zichtbaar licht. De golflengte van dit licht strekt zich uit over een gebied van ruim 400 nm en volgt de kleuren van de regenboog, van ultraviolet/blauw, via groen, geel en oranje tot dieprood. Deze electromagnetische straling kan met ons oog gemakkelijk worden gedetecteerd en daarom is optische sterrenkunde van oudsher dé methode geweest om het firmanent met telescopen af te speuren. Een kleine honderd jaar geleden werd ontdekt, dat de golflengte van het licht omge- keerd evenredig is met de hoeveelheid energie die een lichtdeeltje draagt. Lichtdeeltjes (fotonen) met een golflengte in het optische bereik (ruwweg 400 – 800 nm) hebben een relatief hoge energie, in vergelijking tot bv. fotonen met infrarode of submillimeter golf- lengtes. Daardoor kan een optisch lichtdeeltje meer energie overdragen aan een molecuul dan bv. een submillimeter foton. Het is bv. mogelijk om met zichtbaar licht een electron in een molecuul naar een hogere baan te verplaatsen. Een infrarood foton met minder ener- gie is daartoe niet in staat en laat een molecuul hooguit trillen en draaien. Die excitatie is selectief. Niet iedere kleur optisch licht kan een electron van de ene naar de andere baan verplaatsen. Alleen kleuren die quantummechanisch zijn toegestaan, die precies passen op het energieverschil tussen twee toegestane electronenbanen, kunnen ook daadwerke- lijk worden geabsorbeerd door een molecuul. Op dezelfde wijze kan een molecuul niet iedere willekeurige kleur licht uitstralen, maar alleen die kleuren die corresponderen met een energieverschil tussen twee moleculaire toestanden. De absorptie en emissiespectra – dus de verzameling van alle kleuren die kunnen worden geabsorbeerd of geëmiteerd – zijn voor verschillende soorten moleculen uniek. Zo0n spectrum is a.h.w. een mole- culaire vingerafdruk, waarmee een molecuul overal kan worden geïdentificeerd; in het laboratorium, maar ook in een interstellaire wolk op lichtjaren afstand. Kan zo0n spec- trum in het laboratorium worden gemeten, en blijkt het identiek te zijn aan het spectrum van sterlicht dat door een interstellaire wolk wordt geabsorbeerd of uitgestraald, dan is het mogelijk om een molecuul in de ruimte eenduidig te bevestigen. De identificatie van zo0n molecuul biedt vervolgens inzicht in de mogelijke chemische processen die in de ruimte

115 Nederlandse Samenvatting plaatsvinden. En omdat een spectrum er voor verschillende dichtheden en voor verschil- lende temperaturen anders uitziet, kan ook informatie worden verkregen over fysische eigenschappen zoals het aantal voorhanden zijnde deeltjes en de lokale temperatuur.

Dit is het doel van dit proefschift: een gecombineerde experimentele en observatione- le studie om meer te weten te komen over materie in de ruimte. De nadruk ligt daarbij op niet stabiele moleculen, dus systemen die met een ongepaard electron in hun maag zitten, en daardoor zeer reactief zijn.

Moleculaire Spectroscopie – Op Jacht Naar Moleculen Tus- sen De Sterren

De ruimte tussen de sterren is ijl, maar niet leeg. In grote interstellaire wolken met dicht- heden van enkele duizenden tot miljoenen deeltjes per cm3 worden nieuwe moleculen gevormd. Het betreft zowel kleine moleculen (CO, CH3OH en NaCl) als grote complexe systemen (bv. dimethylether en ethyleenglycol). Een overzicht van alle moleculen die zijn nagewezen in de ruimte is weergegeven in Tabel 1.1. Op dit moment staat de teller op 168 (daarbij zijn isotopen niet meegerekend). In de laatste jaren is duidelijk geworden dat een groot aantal van deze stoffen wordt gevormd in gasfase reacties, andere moleculen ontstaan in de ijslagen op micrometer grote stofdeeltjes door inwerking van atomen, elec- tronen of UV straling en ook de wisselwerking tussen de gas∼ en de vaste stof fase is van belang om de veelvoud aan geïdentificeerde interstellaire moleculen te kunnen verklaren.

Moleculen In Het Heelal Het merendeel van de moleculen in Tabel 1.1 is gevonden m.b.v. radioastronomische waarnemingen door interstellaire spectra te vergelijken met laboratorium data. In dit laag frequente microgolfbereik met golflengtes van enkele mm worden emissielijnen zichtbaar van roterende moleculen. De rotatiebeweging is gequantiseerd en daardoor kunnen alleen bij welbepaalde microgolffrequenties emissielijnen worden waargenomen. Deze methode is uitermate precies en omdat de Aardse atmosfeer voor een belangrijk deel transparant is voor microgolfstraling, is het mogelijk gebleken met radiotelescopen veel moleculen in de ruimte na te wijzen. Een voorwaarde is wel, dat een molecuul een permanent dipoolmo- ment heeft, hetgeen er praktisch op neer komt, dat het molecuul geen puntsymmetrische structuur mag bezitten. Daardoor is een aantal moleculen, zoals het centro-symmetrische lineare HC7H, in dit frequentie bereik niet zichtbaar. Dit is de reden waarom in de jacht naar nieuwe moleculen in de ruimte ook bij andere golflengtes wordt gezocht. Infrarood licht biedt daarbij een goed alternatief, omdat de gecombineerde excitatie van een vibra- tie en een rotatie een zeer duidelijke moleculaire vingerafdruk oplevert. Dit deel van het electromagnetische spectrum is o.a. interessant voor het onderzoek naar polycyclische aromatische koolwaterstoffen (PAKs) in de ruimte, omdat deze grotere moleculen een aantal bewegingen ondergaan met energieverschillen die precies in het infrarood liggen

116 Moleculen In Het Optische Regime

(zoals bv. een C=C of C-H vibratie excitatie). De Aardse atmosfeer is echter weinig trans- parant voor dit licht met golflengtes van typisch enkele µm. Vooral water absorbeert sterk in het infrarood bereik. Daarom is in de afgelopen decennia veel werk verricht met infra- rood telescopen, die zich buiten de atmosfeer bevonden: IRAM, ISO en de tot recentelijk nog actieve Spitzer ruimte telescoop.

Moleculen In Het Optische Regime Ook in het optische regime zijn spectra uniek. In dit bereik liggen de electronische ex- citaties van moleculen. Figuur 1 laat het optische absorptie spectrum zien van sterlicht dat door een diffuse interstellaire wolk reist. Op de achtergrond is de kleur zichtbaar waarbij de absorptie plaatsvindt. Deze verzameling van absorptiebanden is bekend als de diffuse interstellaire banden. De eerste van deze banden werd reeds in 1922 ontdekt, en nog steeds is de moleculaire oorsprong van deze banden een groot raadsel. Een aantal hoofdstukken in dit proefschrift richt zich daarom op de identificatie van de moleculai- re dragers van deze banden. Daartoe wordt in het laboratorium een interstellaire wolk nagebootst met een speciale plasmabron en m.b.v. gevoelige spectroscopische absorptie technieken worden spectra gemeten van moleculen die mogelijk voorkomen in het diffuse interstellaire medium. Behalve optische absorptiespectra kunnen ook optische emissies worden waargeno- men in de ruimte. De Figuur op de cover van dit proefschrift laat de Red Rectangle zien. De Red Rectangle proto-planetaire nevel is een nevel ontstaan uit het uitstromende gas van de ster HD 44179. De figuur laat het uitstromende gas zien, dat naar twee kanten gaat. De ster is omgeven door een dikke schijf materie. In deze schijf zijn verschillende moleculen nagewezen (bv. OH, CO, en C2). De nevel zelf laat in het infrarood emissiebanden zien die typisch zijn voor PAKs. De centrale ster is daarbij de lichtbron die ervoor zorgt dat de moleculen in de omgeving worden geëxciteerd. Naast een brede ERE (extended red emission) laat de Red Rectangle ook een groot aantal smalle emissie banden zien. Deze emissies zijn typisch en karakteristiek en, wederom, bestaan er vrijwel geen toekenningen aan mogelijke moleculaire dragers. In dit proefschrift wordt onderzoek beschreven naar deze banden. Om de emissiespectra toe te kennen, wordt in het laboratorium een laser gebruikt om naar de fluorescentie te kijken van niet stabiele moleculen die in een expan- derend plasma zijn gevormd. Interessant is dat deze emissiebanden mogelijk gerelateerd zijn aan de diffuse interstellaire banden.

Dit Proefschrift

Het onderzoek in dit proefschrift is gebaseerd op het vergelijken van spectra gemeten in het laboratorium met interstellaire data. Het doel van het onderzoek is om astronomische spectra toe te kennen. Daartoe worden in het laboratorium nieuwe moleculen gemaakt. De nadruk ligt hier op “maken”, omdat de stoffen die zijn onderzocht – HC7H, C9H3, C2 – niet stabiel zijn. Ze moeten ter plekke worden gegenereerd. In de ruimte spelen dergelijk radicalen een rol op grond van de voorhanden zijnde stralingsvelden en de ge-

117 Nederlandse Samenvatting

Figuur 1 – Een gesimuleerd spectrum van DIBs. De online catalogus van de DIBs is bereikbaar door Hobbs et al. (2009). Deze figuur is eigendoom van: Jenniskens en Desert. ringe dichtheden (zie ook tabel 1.1). In het laboratorium worden speciale plasma bronnen gebruikt: gas expandeert door een smalle opening in een vacuum en wordt gelijktijdig ontladen (Fig. 2). Door de botsingen in de expansie ontstaan nieuwe radicalen en deze koelen daarbij gelijktijdig af tot temperaturen zoals die in het interstellaire medium gebruikelijk zijn. Zeer gevoelige detectietechnieken, zoals cavity ring-down spectroscopie en tijdsopgelos- te fluorescentie spectroscopie worden vervolgens gebruikt om de spectra te detecteren, in absorptie of emissie, respectievelijk. Naast laboratorium metingen, worden hier ook as- tronomische waarnemingen gepresenteerd. Deze zijn verkregen met de New Technology Telescope in Chili en laten een ruimtelijk opgeloste chemie in het uitstromende gas van de Red Rectangle zien.

Het proefschrift is op de volgende wijze opgebouwd. • Hoofdstuk 1 geeft een inleiding tot het vakgebied en beschrijft de relevantie van het hier gepresenteerde onderzoek. • Hoofdstuk 2 beschrijft in detail de gebruikte laboratorium experimenten; principe, opbouw en uitvoering. Details over de gebruikte cavity ring-down opstelling, het tijdsopgeloste fluorescentie experiment en de gebruikte plasma bronnen worden gegeven.

118 Dit Proefschrift

Figuur 2 – Plasma bronnen, als bv. deze jetplasma, worden gebruikt om nieuwe moleculen in het laboratorium te maaken (Birza et al. 2002).

• In Hoofdstuk 3 wordt een laboratorium spectrum van een expanderend acetyleen plasma vergeleken met een brede diffuse interstellaire band rond 5450 Å. Dit on- derzoek laat voor de eerste keer een perfecte overlap zien tussen een brede diffuse interstellaire band en een spectrum gemeten op Aarde. Zeer recentelijk is in een vervolgstudie de moleculaire drager als C3H2 voorgesteld. Mocht dit juist blijken, dan is dit de eerste keer dat een diffuse interstellaire band eenduidig is toegekend.

• In Hoofdstuk 4 wordt het electronische spectrum van HC7H gepresenteerd. Dit is een centrosymmetrisch lineair molecuul en derhalve niet radio-astronomisch waar- neembaar. De rovibronische analyse (de gelijktijdige excitatie van een electroni- sche vibratie en rotatie) resulteerd voor de eerste keer in gedetailleerde moleculaire constantes. Deze zijn nodig om HC7H in het diffuse interstellaire medium na te wijzen. Op basis van deze metingen kan worden geconcludeerd dat HC7H geen DIB drager is.

• Een ander koolstofwaterstof radicaal, C9H3, wordt behandeld in Hoofdstuk 5. Dit molecuul is ondanks de aanwezigheid van een permanent dipoolmoment, nog niet in het microgolfgebied bestudeerd. De spectra geven gedetailleerde informatie over de structuur van dit niet-lineaire molecuul. Veel van de koolstofketens in tabel 1.1 blijken lineair te zijn. Het is de vraag of niet lineaire moleculen met een verge- lijkbare abundantie voorkomen in de ruimte. Het C9H3 blijkt net als HC7H geen

119 Nederlandse Samenvatting

overtuigende overlap spectra te hebben met de bekende diffuse interstellaire ban- den.

• Hoofdstuk 6 beschrijft een gecombineerde waarneem en laboratorium studie. Een tweetal emissiebanden in de Red Rectangle kan eenduidig worden toegekend aan het C2 radicaal. Opvallend is, dat de overgangen behoren bij vibrationeel geëxci- teerde energieniveaus en op deze wijze inzicht bieden in het optisch pompmecha- nisme, waarmee deze toestanden worden bezet. Dit is in detail uitgewerkt en maakt het mogelijk om C2 dichtheden in het uitstromende gas af te schatten. • Uit Red Rectangle waarnemingen blijkt dat de smalle emissiebanden verschillende intensiteiten hebben voor verschillende afstanden tot de centrale ster. Voor de eer- ste keer wordt een gedetailleerde 2D spectroscopische catalogus gepresenteerd in Hoofdstuk 7: emissie spectra als functie van de afstand tot de ster. De astronomi- sche waarnemingen laten verder ”snapshots” zien van de chemische processen die plaatsvinden in het uitstromende gas. Ook al is het nog niet mogelijk gebleken om deze waarnemingen te vertalen in concrete processen, het is belangrijk dat op deze wijze kwalitatief de chemische evolutie rond een materie uitstotende ster in beeld is gebracht.

120 Zusammenfassung:

Optische Spektroskopie an Astronomisch Relevanten Molekülen

Einleitung

Beobachtende optische Astronomie arbeitet mit sichtbarem Licht. Die Wellenlänge von optischem Licht liegt zwischen 400 nm und 800 nm. Dabei kann es alle Regenbogenfar- ben einnehmen. Bei 400 nm erscheint das Licht violett/blau, es geht dann über in grün, gelb, orange und tiefrot bei ca. 800 nm. Diese elektromagnetische Strahlung kann von uns gesehen und detektiert werden und deswegen ist die optische beobachtende Astronomie auch die älteste Methode, um das Firmament mit Teleskopen abzusuchen. Noch keine 100 Jahre ist es her, dass man feststellte, dass die Wellenlänge des Lich- tes umgekehrt proportional zur Energie des Lichtes ist. Lichtteilchen (Photonen), die ei- ne Wellenlänge im optischen Bereich (also zwischen 400 und 800 nm) besitzen, haben eine relativ hohe Energie, im Vergleich zu Photonen im infraroten ∼ oder submillime- ter Bereich. Optische Photonen können daher mehr Energie an Moleküle übertragen als Photonen mit längerer Wellenlänge. Man sagt dann, dass ein Molekül in einen höheren Energiezustand angeregt wird. Die Energie der optischen Photonen reicht dabei aus, um Elektronen in den Molekülen auf höhere Energiebahnen, sogenannte Orbitale, anzuregen. Ein infrarotes Lichtteilchen hat dabei weniger Energie als ein optisches Photon. Ein infra- rotes Lichtteilchen kann ein Molekül “nur” zu Schwingungen oder Rotationen anregen; es finden hier keine elektronischen Anregungen statt. Die Anregung ist dabei immer selek- tiv. Im optischen Bereich bedeutet dies, dass ein bestimmtes Elektron auch nur von einem bestimmten Lichtteilchen angeregt werden kann und zwar genau dann, wenn das Elektron für einen Übergang zwischen dem unteren und dem oberen Orbital genau diese Energie benötigt. Besitzt das Lichtteilchen eine höhere oder niedrigere Energie als es für die An- regung des Elektrons benötigt, dann findet keine Reaktion zwischen dem Lichtteilchen und dem Elektron in diesem Molekül statt. Alle Energien, die dabei quantenmechanisch erlaubt sind, entsprechen also genau den Abständen zweier oder mehrerer Energieniveaus oder Orbitalen. Auf dieselbe Weise kann ein Molekül auch nur bestimmte Energiemengen oder Spektralfarben emittieren, nämlich genau diese Energien, die dem Abstand zweier Orbitalen entsprechen, wenn das Elektron von einem höheren auf ein niedrigeres Orbital

121 Zielsetzungen zurückkehrt. Die Absorptions ∼ und Emissionspektren eines Moleküls sind dabei die Ge- samtheit aller möglichen Übergänge innerhalb des Moleküls. Sie sind einzigartig für jedes einzelne Molekül. Wie ein Fingerabdruck bei einem Menschen können mit diesen Spek- tren Teilchen identifiziert werden. Dies ist sehr wichtig im Labor, um Reaktanden oder Produkte von Reaktionen zu bestimmen. Dies kann man aber auch nutzen um das Licht von Sternen, das durch interstellare Wolken auf die Erde fällt, zu beobachten und damit Teilchen im interstellaren Medium zu identifizieren. Wird ein Spektrum eines bekann- ten Stoffes in einem Labor gemessen, kann es mit Spektren von Teleskopen verglichen werden. Teleskope nehmen dabei das Sternenlicht auf, dass mehrere tausend Lichtjahre von uns entfernt ist und durch interstellare Wolken oder Nebelgebiete nach einer langen Reise auf die Erde trifft. Diese interstellaren Wolken setzen sich aus vielen verschiedenen Teilchen zusammen, die das Sternenlicht aufnehmen oder abgeben können. Ist das Spek- trum einer solchen interstellaren Wolke identisch mit dem Spektrum, das in einem Labor von einem bestimmten Stoff aufgenommen wurde, so kann man ein Molekül im interstel- laren Medium identifizieren. Weil die Absorptions∼ und Emissionsspektren von einem Molekül zudem bei verschiedenen Temperaturen anders aussehen, kann ein Spektrum au- ßerdem Aufschluss darüber geben, welche physikalischen Zustände dort herrschen und welche chemischen Prozesse dort ablaufen. Dies ist das Thema dieser Arbeit: Hier versuche ich aus einer Kombination von be- obachtender Astronomie und experimenteller Astrophysik Moleküle im interstellaren∼ und zirkumstellaren Medium zu identifizieren. Dies ist wichtig, um dann Rückschlüsse auf die Gegebenheiten im interstellaren Medium (Dichte und Temperatur) zu ziehen. Der Schwerpunkt dieser Arbeit liegt dabei auf nicht-beständige, sogenannte transiente, Mole- küle, die hochreaktiv und deswegen sehr instabil unter “normalen” Bedingungen sind.

Molekülspektroskopie – Auf der Suche nach den Molekü- len zwischen den Sternen

Der Raum zwischen den Sternen ist nicht komplett leer. In großen interstellaren Wol- ken mit Teilchendichten von einigen tausend bis millionen Molekülen pro cm3, werden stets neue Moleküle geformt. Dies trifft sowohl auf kleine Moleküle (wie z.B. Kohlen- stoffmonoxid (CO), Methanol (CH3OH) oder Natriumchlorid (NaCl)) zu, als auch auf größere Moleküle (wie z.B. Dimethylether (C2H6O) und Ethylenglykol (C2H6O2)). Ei- ne Übersicht, die alle bisher identifizierten Moleküle im interstellaren Medium (stand Dezember 2010) zeigt, findet sich in Tabelle 1.1 im ersten Kapitel. Momentan sind es rund 170 Moleküle, wobei Isotope nicht mitgerechnet sind. In den letzten Jahren zeigte sich, dass eine große Anzahl von Molekülen in der Gasphase entsteht. Andere Moleküle entstehen in den Eissichten von mikrometer großen Staubteilchen durch die Einwirkung von Atomen, Elektronen und UV Strahlung. Zudem entstehen viele Moleküle auch nur durch das Zusammenspiel und durch die Wechselwirkung zwischen den Gasteilchen und Feststoffen. Eis ∼ und Staubteilchen sind dabei wichtige Voraussetzungen, um die große Vielfalt der identifizierten Teilchen in interstellaren Wolken zu erklären.

122 Moleküle im Weltall

Moleküle im Weltall Die Mehrzahl der Moleküle in Tabelle 1.1 wurde durch radioastronomische Beobach- tungen von interstellaren Wolken und anschließendem Vergleich mit Spektren aus dem Labor, identifiziert. In diesem niedrigen Energiebereich, wo die Wellenlänge einige mm betragen, werden Emissionslinien von rotierenden Molekülen sichtbar. Die Rotationsbe- wegung ist gequantelt, d.h. auch hier sind nur bestimmte Energieniveaus erreichbar, und dadurch können auch hier nur ganz bestimmte Frequenzen der Emissionslinien detektiert werden. Diese Methode ist besonders präzise, und funktioniert nur, weil die Erdatmo- sphäre für einen Bereich der Strahlung transparent ist. Eine Voraussetzung ist allerdings, dass das entsprechende Molekül, das auf diese Weise untersucht werden soll, ein Dipol- moment besitzt. Mit anderen Worten, das Molekül muss mehr oder weniger asymme- trisch aufgebaut sein und darf keine punktsymmetrische Struktur besitzen. Unter dieser Voraussetzung können bestimmte Moleküle, wie z.B. das zentro-symmetrische und li- neare Molekül HC7H, in diesem Frequenzbereich nicht sichtbar gemacht werden. Dies ist der Grund, warum man auch in anderen Wellenlängenbereichen nach interstellaren Molekülen suchen muss. Infrarotlicht bietet dabei eine gute Alternative, weil die Kombi- nation aus Rotationen und Schwingungen eines Moleküls ebenfalls ein sehr individuel- les Spektrum erzeugt. Der Infrarotbereich des elektromagnetischen Spektrums ist dabei von großer Bedeutung für eine spezielle Klasse von Molekülen, den sogenannten poly- zyklischen aromatischen Kohlenwasserstoffen (PAKs). Dies sind größere Moleküle, die einige bestimmte Energieübergänge, Schwingungen, im Infraroten aufweisen (z.B. eine C=C oder C-H Schwingung). Die Erdatmosphäre ist dabei eher störend, wenn man genau diese Schwingungen untersuchen möchte. Vor allem, Wasser, das in unserer Atmosphä- re vorkommt, absorbiert viel Licht im infraroten Spektralbereich. Speziell in den letzten Jahrzehnten hat sich in diesem Spektralbereich viel getan und Teleskope, die im infra- roten Spektralbereich arbeiten und sich außerhalb der Erdatmosphäre befinden, wurden konzipiert: IRAM und ISO, sowie das relativ neue Spitzer Space Teleskop sind dabei nur einige Beispiele.

Moleküle im Optischen Spektralbereich Auch im optischen Spektralbereich sind die Spektren eindeutig und wie der Fingerab- druck eines jeden Moleküls. In diesem Spektralbereich werden die Elektronen von Mo- lekülen dazu angeregt ihre Energieniveaus zu verändern. Bei der Absorption eines Licht- teilchens findet eine Anregung in ein höheres Energieniveau statt. Bei der Emission von Lichtteilchen findet Relaxation von einem höheren Energieniveau zu einem niedrigeren Energiebereich statt. In Abbildung 1 (der niederländischen Zusammenfassung) kann man das optische Absorptionspektrum von Sternenlicht sehen, dass durch eine diffuse inter- stellare Wolke auf die Erde trifft. Im Hintergrund ist die Spektralfarbe sichtbar, bei der die Absorption stattfindet. Die Gesamtheit dieser Absorptionsbanden in den diffusen in- terstellaren Wolken, wird auch als diffuse interstellare Banden bezeichnet. Die erste Ab- sorptionsbande wurde im Jahre 1922 entdeckt und immer noch sucht man nach den Teil- chen, die diese Absoprtionsbanden verursachen.

123 Zielsetzungen

In den Kapiteln dieser Arbeit beschäftige ich mich vor allem mit der Identifikation von Molekülen, die als Träger dieser Absorptionsbanden in Frage kommen. Zu diesem Zweck werden im Labor spezielle Plasmaquellen eingesetzt, die es ermöglichen auch Molekü- le, die hier auf der Erde unter normalen Bedingungen nicht vorkommen, zu erzeugen. Sehr sensible Absorptionstechniken kommen dann zum Einsatz, mit denen Spektren von Molekülen gemessen werden, die im interstellaren Medium vorkommen. In der Abbildung auf dem Cover dieser Arbeit kann man ein Bild von dem proto- planetarischen Nebel, Red Rectangle, sehen. Der Nebel wird genährt aus dem ausströ- menden Gas eines alten Sternes, HD 44179. In der Abbildung kann man das ausströmen- de Gas gut erkennen und man sieht wie es hauptsächlich in zwei Richtungen strömt. Dies findet seine Ursache in einer dicken Scheibe, die aus Eis und Staubteilchen besteht und die den Stern umgibt. In dieser Scheibe befinden sich verschiedene Moleküle, wie z.B. OH (Hydroxid Radikal) und CO (Kohlenstoffmonoxid)). Der Nebel selbst zeigt im in- fraroten Spektralbereich deutliche Anzeichen von Emissionsbanden von polyzyklischen aromatischen Kohlenwasserstoffen (PAKs). Das Zentralgestirn ist dabei die Lichtquelle und das Licht des Sterns regt die Moleküle in seiner Umgebung an. Dadurch ergeben sich viele Emissionsbanden der PAKs im infraroten Bereich. Im optischen Bereich findet sich eine sogenannte ERE (extended red emission oder ausgedehnte Rotemission), die ebenfalls auf größere Moleküle wie Silikate oder Kohlenstoffteilchen schließen lässt. In dieser relativ ausgedehnten Rotemission, die sich über nahezu 200 nm erstreckt, finden sich übergelagerte schmale Emissionsbanden, die wiederum von kleineren gasförmigen Molekülen stammen. In dieser Arbeit beschäftige ich mich ebenfalls mit diesen schmal- bandigen Emissionsbanden im Red Rectangle. In verschiedenen Arbeiten wird dabei ein Zusammenhang zwischen diesen speziellen Emissionsbanden im Red Rectangle und den Absorptionsbanden in diffusen interstellaren Wolken hergestellt. Um nun diese Emissi- onsbanden zuordnen zu können, gebrauche ich im Labor einen Laser, der die nötige Ener- gie liefert um Moleküle elektronisch anzuregen. Auf diese Weise kann ich die Emission von Licht, die sogenannte Fluoreszenz, von Molekülen detektieren und untersuchen. Die Moleküle werden dabei in einer Plasmaexpansion erzeugt, wie es in Abbildung 2 (der niederländischen Zusammenfassung) zu erkennen ist.

Diese Arbeit

Diese Arbeit basiert auf dem Vergleich von den im Labor gemessenen Spektren mit den Daten, die wir vom interstellaren Medium haben. Die Idee ist dabei die Moleküle in den Spektren vom interstellaren∼ und zirkumstellaren Medium zu identifizieren. Zu diesem Zweck werden im Labor neue Moleküle hergestellt. Das Augenmerk liegt dabei auf der ”Herstellung” der Moleküle, weil Teilchen, wie z.B. HC7H, C9H3 und C2 sehr unstabil und kurzlebig unter normalen atmosphärischen Bedingungen sind. Sie müssen jedesmal aufs Neue vor der Messung erzeugt werden. Im Weltall spielen solche Radikale eine große Rolle, weil die dortigen Strahlungsverhältnisse und die geringe Dichte solche Teilchen be- vorzugt entstehen lässt (Vgl. Tabelle 1.1.). Im Labor werden dazu spezielle Plasmaquellen benutzt: Das Gas expandiert innerhalb einer schmalen Öffnung in ein Vakuum. Diese Öff-

124 Diese Arbeit nung wird unter eine starke Spannung gesetzt, wodurch eine Entladung stattfindet (Abbil- dung 2, niederländische Zusammenfassung). Durch die Expansion ins Vakuum entstehen viele neue Teilchen, die in Folge der Expansion gekühlt werden. Diese Temperaturen liegen nur wenige Grade über dem abso- luten Nullpunkt und sind damit vergleichbar mit den Temperaturen, die im interstellaren Medium herrschen. Sehr empfindliche Detektionstechniken kommen dann zum Einsatz, wie z.B. cavity ring-down Spektroskopie oder Laser induzierte Fluoreszenzspektroskopie, um die verschiedenen Teilchen zu messen. Dies geschieht durch Absorptionsspektren, wie auch durch Emissionsspektren. Im Anschluss an die Laboratoriumsmessung werden auch astronomische Spektren in dieser Arbeit behandelt. Diese Spektren wurden mit dem New Technology Telescope in Chile aufgenommen und zeigen räumlich aufgelöste Spektren, die auf die Chemie des ausströmenden Gases im Red Rectangle schließen lassen.

Diese Arbeit ist auf folgende Weise aufgebaut:

• Im ersten Kapitel wird eine Einleitung in das Fachgebiet gegeben. Es wird die Re- levanz der Arbeit näher erklärt und in ein übergeordnetes Thema eingeordnet. • Im zweiten Kapitel werden die im Labor gebrauchten Methoden beschrieben; das Prinzip, der Aufbau und die Ausführung. Dabei werden cavity ring-down Spektro- skopie und Laser induzierte Fluoreszenzspektroskopie im Detail erklärt und auch die Plasmaquellen, die hierbei zum Einsatz kommen näher erläutert. • Im dritten Kapitel wird das Laborspektrum eines expandierenden Acetylenplasmas mit dem Spektrum einer Absorptionsbande um 5450 Å einer diffusen interstella- ren Wolke verglichen. Das Resultat ist, das zum ersten Mal ein perfekter Überlapp zwischen einem breiten Absorptionsband und einem Laborspektrum erzielt werden konnte. Vor Kurzem wurde in einer weiterführenden Studie das Thema aufgegrif- fen und das Absorptionsband mit dem Molekül C3H2 in Zusammenhang gebracht. Trifft dies zu, ist das das erste Mal, dass ein diffuses interstellares Absorptionsband identifiziert werden konnte.

• Im vierten Kapitel wird das elektronische Spektrum des Moleküls HC7H vorge- stellt. Dies ist ein zentro-symmetrisches Molekül und kann deswegen nicht radio– spektroskopisch gemessen werden. Die ro-vibronische Analyse (also die gleichzei- tige Anregung von Elektronen, Schwingungen und Rotationen) resultiert in einer detaillierten Analyse der molekularen Konstanten, die so zum ersten Mal bestimmt werden konnten. Diese Konstanten sind wichtig, damit man u.U. HC7H auch im diffusen interstellaren Medium identifizieren kann. Auf dieser Grundlage wurde beschlossen, dass HC7H kein Träger der diffusen interstellaren Absorptionsbanden ist.

• Ein anderes Kohlenwasserstoffradikal, C9H3, wird in Kapitel 5 behandelt. Obwohl das Molekül ein permanentes Dipolmoment besitzt, wurde es noch nicht im Mikro-

125 Zielsetzungen

wellenbereich untersucht. Aus den Spektren resultiert eine detaillierte Analyse der Struktur dieses nicht-linearen Moleküls. Viele Kohlenstoffketten, die in Tabelle 1.1 zu sehen sind, sind tatsächlich linear aufgebaut. Nun stellt sich die Frage, ob auch dieses Molekül im interstellaren Medium vorkommt. Der Vergleich mit astronomi- schen Spektren ergab aber, ähnlich wie beim HC7H, keine Übereinstimmung mit den Absorptionsbanden im diffusen interstellaren Medium.

• In Kapitel 6 wird eine Kombination von astronomischen Spektren und Laborspek- tren vorgestellt. Zwei Emissionsbanden des Red Rectangle Nebels konnten ein- deutig mit dem C2 Radikal identifiziert werden. Auffallend dabei ist, dass auch schwingungsangeregte Zustände im Nebel vorkommen, die Einsicht in die Anre- gungsmechanismen innerhalb des Nebels bieten. Dies wird im Detail erklärt und resultiert in einer Angabe der Dichte von C2 innerhalb des ausströmenden Gases mit Abstand zum Zentralstern.

• Aus den Beobachtungen des Red Rectangle wird zudem ersichtlich, dass die Emis- sionsbanden verschiedene Intensitäten mit Abstand zum Zentralstern besitzen. Zum ersten Mal wird hier in Kapitel 7 eine detaillierte 2D Studie präsentiert: Emissions- spektren als Funktion des Abstandes zum Zentralstern. Die astronomischen Spek- tren lassen dabei Momentaufnahmen der stattfindenden Chemie im ausströmenden Gas des Nebels erkennen. Auch wenn es nicht möglich ist dies als konkrete chemi- sche Prozesse zu interpretieren, ist es wichtig eine qualitative Analyse zu formu- lieren, die sich zumindest mit der Evolution des ausströmenden Gases des Sterns beschäftigt, bevor weitere Schlüsse gezogen werden können.

126 Bibliography

Adamson, A. J. & Whittet, D. C. B. 1992, ApJ, 398, L69 Allamandola, L. J., Tielens, A. G. G. M., & Barker, J. R. 1989, ApJS, 71, 733 Aoki, K. & Ikuta, S. 1994, J. Mol. Struct.:theochem, 116, 220 Araki, M., Linnartz, H., Cias, P., et al. 2003, J. Chem. Phys., 118, 10561 Araki, M., Linnartz, H., Kolek, P., et al. 2004, ApJ, 616, 1301 Bakker, E. J., van Dishoeck, E. F., Waters, L. B. F. M., & Schoenmaker, T. 1996, VizieR Online Data Catalog, 332, 30469 Ball, C. D., McCarthy, M. C., & Thaddeus, P. 1999, ApJ, 523, L89 Ball, C. D., McCarthy, M. C., & Thaddeus, P. 2000a, J. Chem. Phys., 112, 10149 Ball, C. D., McCarthy, M. C., & Thaddeus, P. 2000b, ApJ, 529, L61 Balm, S. P. & Jura, M. 1992, A&A, 261, L25 Bazalgette Courrèges-Lacoste, G., Sprengers, J. P., Bulthuis, J., et al. 2001, Chem. Phys. Lett., 335, 209 Behforooz, G. H. 1995, Applied Mathematics and Computation, 72, 219 Bell, M. B., Feldman, P. A., Travers, M. J., et al. 1997, ApJ, 483, L61+ Berden, G., Peeters, R., & Meijer, G. 2000, Int. Rev. Phys. Chem., 19, 565 Beuther, H., Zhang, Q., Bergin, E. A., et al. 2007, A&A, 468, 1045 Birza, P., Motylewski, T., Khoroshev, D., et al. 2002, Chem. Phys., 283, 119 Boersma, C., Hony, S., & Tielens, A. G. G. M. 2006, A&A, 447, 213 Bond, H. E. 1997, in IAU Symposium, Vol. 180, Planetary Nebulae, ed. H. J. Habing & H. J. G. L. M. Lamers, 460 Bréchignac, P. & Pino, T. 1999, A&A, 343, L49 Brünken, S., Gupta, H., Gottlieb, C. A., McCarthy, M. C., & Thaddeus, P. 2007, ApJ, 664, L43 Bujarrabal, V., Castro-Carrizo, A., Alcolea, J., & Neri, R. 2005, A&A, 441, 1031 Bujarrabal, V., Neri, R., Alcolea, J., & Kahane, C. 2003, A&A, 409, 573 Cami, J., Bernard-Salas, J., Peeters, E., & Malek, S. E. 2010, Science, 329, 1180 Cami, J., Sonnentrucker, P., Ehrenfreund, P., & Foing, B. H. 1997, A&A, 326, 822 Cernicharo, J., Heras, A. M., Tielens, A. G. G. M., et al. 2001, ApJ, 546, L123 Chaffee, Jr., F. H. & Lutz, B. L. 1978, ApJ, 221, L91 Chen, X. & Mazumder, J. 1990, Appl. Phys. Lett., 57, 2178 Cherchneff, I., Barker, J. R., & Tielens, A. G. G. M. 1992, ApJ, 401, 269 Cohen, M., Anderson, C. M., Cowley, A., et al. 1975, ApJ, 196, 179 Cohen, M., Van Winckel, H., Bond, H. E., & Gull, T. R. 2004, AJ, 127, 2362 Cossart-Magos, C. & Leach, S. 1990, A&A, 233, 559 Curtis, M. C. & Sarre, P. J. 1985, J. Mol. Spectrosc., 114, 427 Darbon, S., Perrin, J., & Sivan, J. 1999, A&A, 348, 990 Darbon, S., Zavagno, A., Perrin, J., et al. 2000, A&A, 364, 723 Demtroeder, W. 1996, Laser Spectrosocpy, Springer Verlag, Berlin, 1996

127 Bibliography

Ding, H. B., Pino, T., Guthe, F., & Maier, J. P. 2003, J.Am. Chem. Soc., 125, 14626 Douglas, A. E. & Herzberg, G. 1941, ApJ, 94, 381 Duley, W. W. 1985, MNRAS, 215, 259 Duley, W. W. & Williams, D. A. 1979, Nature, 277, 40 Dzhonson, A., Jochnowitz, E. B., Kim, E., & Maier, J. P. 2007, J. Chem. Phys., 126, 044301 Ehrenfreund, P. & Foing, B. H. 1995, Planet. Space Sci., 43, 1183 Fink, U. & Hicks, M. D. 1996, ApJ, 459, 729 Foing, B. H. & Ehrenfreund, P. 1994, Nature, 369, 296 Foing, B. H. & Ehrenfreund, P. 1997, A&A, 317, L59 Frenklach, M. & Feigelson, E. D. 1989, ApJ, 341, 372 Frisch, M. J., Trucks, G., Schlegel, H., et al. 1998, Gaussian, Inc., Pittsburgh PA, 1998, 1, 1 Fuchs, G. W., Cuppen, H. M., Ioppolo, S., et al. 2009, A&A, 505, 629 Fulara, J., Freivogel, P., Forney, D., & Maier, J. P. 1995, J. Chem. Phys., 103, 8805 Furton, D. G. & Witt, A. N. 1992, ApJ, 386, 587 Galazutdinov, G. A., Musaev, F. A., Krełowski, J., & Walker, G. A. H. 2000, PASP, 112, 648 Giesen, T. F., van Orden, A., Hwang, H. J., et al. 1994, Science, 265, 756 Glinski, R. J. & Anderson, C. M. 2002, MNRAS, 332, L17 Glinski, R. J., Michaels, P. D., Anderson, C. M., et al. 2009, Ap&SS, 145 Glinski, R. J. & Nuth, III, J. A. 1997, Ap&SS, 249, 143 Gordon, K. D., Witt, A. N., & Friedmann, B. C. 1998, ApJ, 498, 522 Gordon, V. D., McCarthy, M. C., Apponi, A. J., & Thaddeus, P. 2000, ApJ, 540, 286 Gottlieb, C. A., Apponi, A. J., McCarthy, M. C., Thaddeus, P., & Linnartz, H. 2000, J. Chem. Phys., 113, 1910 Gredel, R. 1999, A&A, 351, 657 Gredel, R., van Dishoeck, E. F., & Black, J. H. 1989, ApJ, 338, 1047 Guhathakurta, P. & Tyson, J. A. 1989, ApJ, 346, 773 Haas, M. R., Davidson, J. A., & Erickson, E. F., eds. 1995, Astronomical Society of the Pacific Conference Series, Vol. 73, Airborne Astronomy Symposium on the Galactic Ecosystem: From Gas to Stars to Dust, volume 73 Hall, K. I., Miles, J. R., Sarre, P. J., & Fossey, S. J. 1992, Nature, 358, 629 Halmer, D., von Basum, G., Hering, P., & Murtz, M. 2004, Rev. Sci. Instrum. Hartmann, J. 1904, ApJ, 19, 268 Heger, M. L. 1922, Lick Observatory Bulletin, 10, 141 Herbelin, J. M., McKay, J. A., Kwok, M. A., et al. 1980, Appl. Optics, 19, 144 Herbig, G. H. 1975, ApJ, 196, 129 Herbig, G. H. 1993, ApJ, 407, 142 Herbst, E. & van Dishoeck, E. F. 2009, ARA&A, 47, 427 Hobbs, L. M. 1979, ApJ, 232, L175 Hobbs, L. M., Thorburn, J. A., Oka, T., et al. 2004, ApJ, 615, 947 Hobbs, L. M., York, D. G., Thorburn, J. A., et al. 2009, ApJ, 705, 32 Hodges, J. T., Looney, J. P., & van Zee, R. D. 1996, J. Chem. Phys., 105, 10278 Ioppolo, S., Cuppen, H. M., Romanzin, C., van Dishoeck, E. F., & Linnartz, H. 2008, ApJ, 686, 1474 Ioppolo, S., Cuppen, H. M., van Dishoeck, E. F., & Linnartz, H. 2010, MNRAS, 1765

128 Bibliography

Jenniskens, P. & Desert, F. 1994, A&AS, 106, 39 Jochnowitz, E. B. & Maier, J. P. 2008, Molecular Physics, 106, 2093 Jochnowitz, E. B. & Maier, J. P. 2008, Annual Review of Physical Chemistry, 59, 519 Jones, A. P. 2009, in Astronomical Society of the Pacific Conference Series, Vol. 414, Astronomical Society of the Pacific Conference Series, ed. T. Henning, E. Grün, & J. Steinacker, 473–+ Jones, A. P., Tielens, A. G. G. M., & Hollenbach, D. J. 1996, ApJ, 469, 740 Jura, M., Balm, S. P., & Kahane, C. 1995, ApJ, 453, 721 Jura, M., Turner, J., & Balm, S. P. 1997, ApJ, 474, 741 Khoroshev, D., Araki, M., Kolek, P., et al. 2004, J. Molec. Spectrosc., 227, 81 Kirkwood, D. A., Linnartz, H., Grutter, M., et al. 1998, in Chemistry and Physics of Molecules and Grains in Space. Faraday Discussions No. 109, 109–+ Klochkova, V. G., Szczerba, R., Panchuk, V. E., & Volk, K. 1999, A&A, 345, 905 Kokkin, D. L., Bacskay, G. B., & Schmidt, T. W. 2007, J. Chem. Phys., 126, 084302 Krätschmer, W., Lamb, L. D., Fostiropoulos, K., & Huffman, D. R. 1990, Nature, 347, 354 Krelowski, J. & Walker, G. A. H. 1987, ApJ, 312, 860 Kroto, H. W., Heath, J. R., Obrien, S. C., Curl, R. F., & Smalley, R. E. 1985, Nature, 318, 162 Kroto, H. W. & Jura, M. 1992, A&A, 263, 275 Kwok, S. 2009, International Journal of Astrobiology, 8, 161 Latter, W. B. 1991, ApJ, 377, 187 Le Bourlot, J. & Roueff, E. 1986, J. Mol. Spectrosc., 120, 157 Ledoux, G., Ehbrecht, M., Guillois, O., et al. 1998, A&A, 333, L39 Ledoux, G., Guillois, O., Huisken, F., et al. 2001, A&A, 377, 707 Lee, J., Lee, H., & Hahn, J. 1999, Japanese Journal of Applied Physics, 38, 6287 Leger, A. & Puget, J. L. 1984, A&A, 137, L5 Lehmann, K. K. & Romanini, D. 1996, J. Chem. Phys., 105, 10263 Linnartz, H. 2001, Chemical Physics Letters, 345, 89 Linnartz, H. 2009, Cavity ring down spectroscopy of molecular transients of astrophysical Interest; p.145-179, Blackwell publishers; Editors: G. Berden, R. Engeln Linnartz, H., Motylewski, T., & Maier, J. P. 1998, J. Chem. Phys., 109, 3819 Linnartz, H., Motylewski, T., Vaizert, O., et al. 1999, J. Mol. Spectrosc., 197, 1 Linnartz, H., Wehres, N., van Winckel, H., et al. 2010, A&A, 511, L3+ Lloyd, G. M. & Ewart, P. 1999, J. Chem. Phys., 110, 385 Maier, J. P. 1997, Chem. Soc. Rev., 26, 21 Maier, J. P. 1998, Journal of Physical Chemistry A, 102, 3462 Maier, J. P., Lakin, N. M., Walker, G. A. H., & Bohlender, D. A. 2001, ApJ, 553, 267 Maier, J. P., Walker, G. A. H., Bohlender, D. A., et al. 2010, ArXiv e-prints McCall, B. J., Thorburn, J., Hobbs, L. M., Oka, T., & York, D. G. 2001, ApJ, 559, L49 McCarthy, M. C., Gottlieb, C. A., Gupta, H., & Thaddeus, P. 2006, ApJ, 652, L141 McCarthy, M. C. & Thaddeus, P. 2001, Chem. Soc. Rev, 177 McKellar, A. 1940, PASP, 52, 187 Men’shchikov, A. B., Schertl, D., Tuthill, P. G., Weigelt, G., & Yungelson, L. R. 2002, A&A, 393, 867 Merrill, P. W. 1934, PASP, 46, 206

129 Bibliography

Merrill, P. W., Sanford, R. F., Wilson, O. C., & Burwell, C. G. 1937, ApJ, 86, 274 Micelotta, E. R., Jones, A. P., & Tielens, A. G. G. M. 2010, A&A, 510, A37+ Molster, F. J., Waters, L. B. F. M., & Tielens, A. G. G. M. 2002a, A&A, 382, 222 Molster, F. J., Waters, L. B. F. M., Tielens, A. G. G. M., & Barlow, M. J. 2002b, A&A, 382, 184 Molster, F. J., Waters, L. B. F. M., Tielens, A. G. G. M., Koike, C., & Chihara, H. 2002c, A&A, 382, 241 Motylewski, T. & Linnartz, H. 1999, Rev. Sci. Instr., 70, 1305 Motylewski, T., Linnartz, H., Vaizert, O., et al. 2000, ApJ, 531, 312 Nakajima, M., Sumiyoshi, Y., & Endo, Y. 2002, Rev. Sci. Instrum., 73, 165 Naus, H., de Lange, A., & Ubachs, W. 1997, Phys. Rev. A, 56, 4755 Nee, J. & Lee, L. 1984, JCP, 81, 31 Öberg, K. I., Garrod, R. T., van Dishoeck, E. F., & Linnartz, H. 2009, A&A, 504, 891 O’Keefe, A. & Deacon, D. A. G. 1988, Review of Scientific Instruments, 59, 2544 Osterbart, R., Langer, N., & Weigelt, G. 1997, A&A, 325, 609 Paldus, B. A. & Kachanov, A. A. 2005, Can. J. Phys., 83, 975 Pardo, J. R., Cernicharo, J., Goicoechea, J. R., Guélin, M., & Asensio Ramos, A. 2007, ApJ, 661, 250 Parkes, A. M., Fawcett, B. L., Austin, R. E., et al. 2003, Analyst, 128, 960 Peeters, E., Hony, S., Van Kerckhoven, C., et al. 2002, A&A, 390, 1089 Perrin, J., Darbon, S., & Sivan, J. 1995, A&A, 304, L21+ Perrin, J. & Sivan, J. 1992, A&A, 255, 271 Pety, J., Teyssier, D., Fossé, D., et al. 2005, A&A, 435, 885 Pfluger, D., Sinclair, W. E., Linnartz, H., & Maier, J. P. 1999, Chem. Phys. Lett, 313, 171 Pino, T., Ding, H., Güthe, F., & Maier, J. P. 2001, J. Chem. Phys., 114, 2208 Pouilly, B., Robbe, J. M., Schamps, J., & Roueff, E. 1983, J. Phys. B Atomic Molec. Phys., 16, 437 Prasad, C. V. V. & Bernath, P. F. 1994, ApJ, 426, 812 Puget, J. L. & Leger, A. 1989, ARA&A, 27, 161 Raghunandan, R., Mazzotti, F. J., Chauhan, R., Tulej, M., & Maier, J. P. 2009, Journal of Physical Chemistry A, 113, 13402 Raskin, G. & Van Winckel, H. 2008, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 7014, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series Reese, M. D. & Sitko, M. L. 1996, ApJ, 467, L105+ Roddier, F., Roddier, C., Graves, J. E., & Northcott, M. J. 1995, ApJ, 443, 249 Ruiterkamp, R., Halasinski, T., Salama, F., et al. 2002, A&A, 390, 1153 Russell, R. W., Soifer, B. T., & Willner, S. P. 1978, ApJ, 220, 568 Salama, F., Bakes, E. L. O., Allamandola, L. J., & Tielens, A. G. G. M. 1996, ApJ, 458, 621 Salama, F., Galazutdinov, G. A., Krełowski, J., Allamandola, L. J., & Musaev, F. A. 1999, ApJ, 526, 265 Sarre, P. J. 2006, J.Mol.Spectrosc., 238, 1 Sarre, P. J., Miles, J. R., & Scarrott, S. M. 1995, Science, 269, 674 Saykally, R. J. & Casaes, R. 2001, Laser Focus World, 37, 159 Scarrott, S. M., Watkin, S., Miles, J. R., & Sarre, P. J. 1992, MNRAS, 255, 11P

130 Bibliography

Schmidt, G. D., Cohen, M., & Margon, B. 1980, ApJ, 239, L133 Schmidt, T. W., Boguslavskiy, A. E., Pino, T., Ding, H., & Maier, J. P. 2003a, Int. J. Mass Spectrom., 228, 647 Schmidt, T. W., Ding, H., Boguslavskiy, A. E., Pino, T., & Maier, J. P. 2003b, J. Phys. Chem. A, 107, 6550 Sharp, R. G., Reilly, N. J., Kable, S. H., & Schmidt, T. W. 2006, ApJ, 639, 194 Sinclair, W. E., Pfluger, D., Linnartz, H., & Maier, J. P. 1999, J. Chem. Phys., 110, 296 Sitko, M. L., Bernstein, L. S., & Glinski, R. J. 2008, ApJ, 680, 1426 Song, I., McCombie, J., Kerr, T. H., & Sarre, P. J. 2007, MNRAS, 380, 979 Sonnentrucker, P., Cami, J., Ehrenfreund, P., & Foing, B. H. 1997, A&A, 327, 1215 Sorkhabi, O., Blunt, V., Lin, H., et al. 1997, JCP, 107, 9842 Sorokin, P. P., Glownia, J. H., & Ubachs, W. 1998, in Chemistry and Physics of Molecules and Grains in Space. Faraday Discussions No. 109, 137–+ Souza, S. P. & Lutz, B. L. 1977, in Bulletin of the American Astronomical Society, Vol. 9, 354 + Speck, A. K. & Barlow, M. J. 1997, Ap&SS, 251, 115 Struve, O. 1927, Popular Astronomy, 35, 212 Struve, O. 1930, ApJ, 72, 199 Swings, P. & Rosenfeld, L. 1937, ApJ, 86, 483 Szomoru, A. & Guhathakurta, P. 1998, ApJ, 494, L93+ Thaddeus, P. & McCarthy, M. C. 2001, Spectrochimica Acta, 57, 757 Thaddeus, P., McCarthy, M. C., Travers, M. J., Gottlieb, C. A., & Chen, W. 1998, Chemistry and Physics of Molecules and Grains in Space. Faraday Discussions No. 109. The Faraday Division of the Royal Society of Chemistry, London, 1998., p.121, 109, 121 Tielens, A. G. G. M. 2005, The Physics and Chemistry of the Interstellar Medium, ed. Tielens, A. G. G. M. Tielens, A. G. G. M. 2008, ARA&A, 46, 289 Tielens, A. G. G. M. & Snow, T. P., eds. 1995, Astrophysics and Space Science Library, Vol. 202, The diffuse interstellar bands Tuairisg, S. Ó., Cami, J., Foing, B. H., Sonnentrucker, P., & Ehrenfreund, P. 2000, A&AS, 142, 225 Tulej, M., Kirkwood, D. A., Pachkov, M., & Maier, J. P. 1998, ApJ, 506, L69 Vaizert, O., Motylewski, T., Wyss, M., et al. 2001, J. Chem. Phys., 114, 7918 van Dishoeck, E. F. 1983, Chem. Phys., 77, 277 van Dishoeck, E. F. 2004, ARA&A, 42, 119 van Dishoeck, E. F. & Black, J. H. 1982, ApJ, 258, 533 van Dishoeck, E. F. & de Zeeuw, T. 1984, MNRAS, 206, 383 van Hemert, M. C. & van Dishoeck, E. F. 2008, Chem. Phys., 343, 292 van Winckel, H. 2003, ARA&A, 41, 391 Van Winckel, H., Cohen, M., & Gull, T. R. 2002, A&A, 390, 147 Van Winckel, H., Waelkens, C., & Waters, L. B. F. M. 1995, A&A, 293, L25 Vijh, U. P., Witt, A. N., & Gordon, K. D. 2004, ApJ, 606, L65 Vijh, U. P., Witt, A. N., & Gordon, K. D. 2005, ApJ, 619, 368 Vijh, U. P., Witt, A. N., York, D. G., et al. 2006, ApJ, 653, 1336 Volkers, E. A., Bulthuis, J., Stolte, S., Jost, R., & Linnartz, H. 2006, J.Mol.Spectrosc., 237, 259

131 Bibliography

Volkers, E. A., Vredenborg, A., Linnartz, H., et al. 2004, Chem. Phys. Lett., 391, 106 Wada, S., Mizutani, Y., Narisawa, T., & Tokunaga, A. T. 2009, ApJ, 690, 111 Waelkens, C., Van Winckel, H., Waters, L. B. F. M., & Bakker, E. J. 1996, A&A, 314, L17 Wakelam, V., Smith, I. W. M., Herbst, E., et al. 2010, Space Science Reviews Waters, L. B. F. M., Cami, J., de Jong, T., et al. 1998, Nature, 391, 868 Waters, L. B. F. M., Molster, F. J., de Jong, T., et al. 1996, A&A, 315, L361 Wehres, N., Romanzin, C., Linnartz, H., van Winckel, H., & Tielens, A. G. G. M. 2010a, A&A, 518, A36+ Wehres, N., Zhao, D., Linnartz, H., & Ubachs, W. 2010b, Chem. Phys. Lett., 497, 30+ Western, C. M. 2007, University of Bristol, http://pgopher.chm.bris.ac.uk Witkowicz, E., Linnartz, H., de Lange, C., et al. 2004, Int. J. Mass Spec., 232, 25 Witt, A. N. & Boroson, T. A. 1990, ApJ, 355, 182 Witt, A. N., Gordon, K. D., & Furton, D. G. 1998, ApJ, 501, L111+ Witt, A. N., Mandel, S., Sell, P. H., Dixon, T., & Vijh, U. P. 2008, ApJ, 679, 497 Witt, A. N. & Vijh, U. P. 2004, in Astronomical Society of the Pacific Conference Series, Vol. 309, Astrophysics of Dust, ed. A. N. Witt, G. C. Clayton, & B. T. Draine, 115–+ Witt, A. N., Vijh, U. P., Hobbs, L. M., et al. 2009, ApJ, 693, 1946 Wodtke, A. & Lee, Y. 1985, J. Phys. Chem., 89, 4744 Zalicki, P. & Zare, R. N. 1995, J. Chem. Phys., 102, 2708 Zhang, C. J. 2004, J. Chem Phys., 121, 8212 Zhao, D., Wehres, N., Linnartz, H., & Ubachs, W. 2010, Chemical Physics Letters, In Press, Corrected Proof, Zhou, Z., Sfeir, M. Y., Zhang, L., et al. 2006, ApJ, 638, L105

132 Publications

Refereed papers

1. The spatial Distribution of the Optical Emission Features in the Red Rectangle Proto-planetary Nebula N. Wehres, H. Linnartz, H. Van Winckel and A. G. G. M. Tielens Submitted to Astronomy and Astrophysics (Chapter 7)

2. Electronic Spectra and Molecular Geometry of the non-linear carbon chain C9H3 D. Zhao, N. Wehres W. Ubachs and H. Linnartz Chemical Physics Letters, (2011) 501, 232 (Chapter 5) 3 3 3. Rotationally resolved A Σu –X Σg spectrum of HC7H N. Wehres, D. Zhao, W. Ubachs and H. Linnartz Chemical Physics Letters (2010) 497, 30 (Chapter 4)

4. C2 emission features in the Red Rectangle proto-planetary nebula - A combined observational and laboratory study N. Wehres, C. Romanzin, H. Linnartz , H. Van Winckel and A. G. G. M. Tielens Astrononmy & Astrophysics (2010) 518, A36 (Chapter 6) 5. A coincidence between a hydrocarbon plasma absorption spectrum and the λ5450 DIB H. Linnartz, N. Wehres, H. Van Winckel, G. A. H. Walker, D. A. Bohlender, A. G. G. M. Tielens, T. Motylewski and J. P. Maier Astronomy & Astrophysics (2010) 511L, L3 (Chapter 3) - Highlighted Paper 15 16 6. Vibronic spectrum of N O2 between 415 and 440 nm E. A. Volkers, J. Bulthuis, S. Stolte, R. Jost, N. Wehres, H. Linnartz Journal of Molecular Spectroscopy (2007) 245, 1 7. Anion photoelectron imaging of deprotonated Thymine and Cytosine B. F. Parsons, S. M. Sheehan, T. A. Yen and D. M. Neumark and N. Wehres and R. Weinkauf Physical Chemistry Chemical Physics (2007) 9, 3291 8. Spectroscopic characterization of the ground and low-lying electronic states of Ga2N via anion photoelectron spectroscopy S. M. Sheehan, G. Meloni, B. F. Parsons. N. Wehres and D. M. Neumark Journal of Chemical Physics (2006) 124, 064303

133

Curriculum Vitae

After finishing school and obtaining my Abitur I first started a traineeship at Bayer Corp. in Krefeld–Uerdingen under supervision of Mr. K. Mansfeld. The traineeship lasted 3 years and during this time I was working as a chemical lab assistant in different lab- oratories at Bayer. The work was fascinating and diverse and I gained first hands-on experience on research and development on solar cells and test wafers, production of different polyurethane pre-cursors, NMR-spectroscopy and research and development of polycarbonate and biochemical protection materials. After having finished the training at Bayer I decided to study chemistry at Hein- rich Heine Universität in Düsseldorf in 2001. During the course of my studies I was given the chance to pursue part of my studies at Université Louis Pasteur in Strasbourg, France. Here, I gathered my first experience abroad, which had a very deep influence on me. I enjoyed the time very much and I am really grateful to Prof. W. Kläui, Prof. D. Matt and Dr. D. Semeril for making this stay possible! Right after my Vordiplom examination I specialised deeper into the field of analytical and physical chemistry and I applied at the DESY (Deutsches Elektronen Synchrotron) re- search institute to participate on the summer student program 2004. During the stay I was working for two months on the HASYLAB facility under supervision of Prof E. Weckert. For the first time, I had the chance to work on a small but very own research project on the radiation damage of the enzyme trypsine exposed to synchrotron radiation (X-rays). The structural changes were followed in time using X-ray diffraction methods. Becoming more and more interested in physical chemistry, the compounds and char- acteristics of matter and the analysis of species, I specialised in spectroscopy and molec- ular dynamics. I followed my main Diplom studies at Heinrich Heine Universität in Düs- seldorf under supervision of Prof. R. Weinkauf. I am especially thankful to him for intro- ducing me to state-of-the-art research and providing me with a very interesting Diplom research project! I am also thankful to the group members Dirk Nolting, Manfred Hucke, Sascha Wiedemann und Shuokan Zhang for their help, guidance and all the discussions. Under the guidance of Prof. Weinkauf I obtained a research grant from the “Heinrich Hertz Stiftung”. The grant allowed for a 10 months stay at the University of California at Berkeley between 2005 and 2006. Here, I participated on experiments on anion pho- toelectron spectroscopy using the technique of velocity mapped imaging. My Diplom research project on the “acidity of the nucleobases thymine and cytosine” lead to a publi- cation. I would like to thank Prof. R. Weinkauf and Prof. D. Neumark for giving me the chance for spending a very beautiful and interesting time in California! Special thanks also to my colleagues Bradley Parsons, Sean Sheehan and Terry Yen. “Vielen, vielen Dank!” It was also in California, Berkeley, when I applied for a Ph.D. position. Here in

135 Curriculum Vitae

Berkeley, I also met for the first time one of my prospect Ph.D. supervisors: Xander Tielens. The “job interview” was interesting taking place in one of the cafes close to the Berkeley campus. I learned about the Ph.D. project as being part of a European FP6 Network “The Molecular Universe”. A second job interview took place in December 2005 in Leiden, where I met my prospect “day-to-day” supervisor Harold Linnartz. I got very much excited by his enthusiasm and started my Ph.D. in April 2006 under his supervision. Although my affiliation has been the Kapteyn Astronomical Institute in Groningen, I was always treated as a full member of his laboratory group! The project was foreseen to be split into a laboratory part taking place at the Sackler Laboratory for Astrophysics at Leiden Observatory and an observational/modelling part at the Kapteyn Astronomical Institute at Rijksuniversiteit Groningen. During the labo- ratory study I was working on laser induced fluorescence (LIF) spectroscopy of carbon bearing species produced in a supersonic jet expansion. I also did many experiments at the Vrije Universiteit in Amsterdam under the supervision of Wim Ubachs in my final year. Together with Dongfeng Zhao and Ali Haddad I was working on cavity ring-down spectroscopy of a hydrocarbon plasma expansion. The laboratory spectra were compared to astronomical spectra of diffuse interstellar clouds and to optical long-slit spectra of the Red Rectangle proto-planetary nebula. It was a very interesting time with many discus- sions that I enjoyed a lot. As one of the most exciting stays during my Ph.D. I was given the chance to be the principal investigator (PI) of an ESO proposal. The proposal involved optical long-slit spectra of the Red Rectangle that I was allowed to observe myself at La Silla, in Chile! That was a wonderful experience which would not have been possible without the help of Hans Van Winckel at Katholieke Universiteit Leuven who explained and helped and spend much time on explaining the data reduction to me. As a Marie Curie fellow I got the chance to also present the results of the research project in form of posters and talks on conferences and meetings in Germany, the Nether- lands, France, Scotland and Italy. I am happy and grateful that I got the chance to be part of the “Molecular Universe”. After my Ph.D. defence I will continue in Boulder, Colorado (USA), as a post-doctoral researcher at the JILA (Joint Institute for Laboratory Astrophysics) under supervision of Prof. Veronica Bierbaum and Prof. Ted Snow.

136 Acknowledgements

At first I would like to thank my promotors Harold Linnartz and Xander Tielens for pro- viding me with such an interesting and also challenging Ph.D. project. Without their help, guidance and patience over all these years this thesis would not be, what it is! Harold, I would like to thank you for your kind words whenever I thought that I was stuck in the work. Sharing your experience in the building-up of the experiment, the help and the dis- cussions on the set-up, the results and especially on how to write a good scientific paper was very much appreciated! Xander, during all our discussions I learned very much from you! I appreciate all the time that you tried to turn me from a chemist into “some sort of” astronomer. Your on-demand knowledge and your creativity are a true inspiration. I really learned a lot! Thank you very much! I would like to thank the “computer group” (Erik, Aart, David and Tycho) as well as the support staff (Anita, Kirsten. Liesbeth, Evelijn and Jeanne) for their kind and imme- diate help on all kind of issues. I am also grateful to the staff at the Kapteyn Astronomical Institute in Groningen (especially Lucia, Jackie and Wim). Thank you for your help and immediate responses on my mails. During all the time I spend in the laboratory, I had much help from the FMD (Ewie, Gijsbert and Martijn), especially the modifications on the light-shielding raised stimulat- ing discussions and a lot of headache. Here, I also would like to thank the members of the ELD, especially René and Ton for their patience in repairing and modifying the sensitive electronics on my set-up, especially the help on the HV feed-throughs was very much ap- preciated! When starting in Leiden, Erik Volkers, one of Harold’s former Ph.D. students, passed by regularly in the lab to share his experience with this set-up with me. “Hartelijk bedankt!” I am deeply grateful that I got the chance to take part on the cavity ring-down exper- iments at the Laser Center at Vrije Universiteit in Amsterdam during the last year of my thesis. Under the supervision of Harold and Wim Ubachs I was working together with Donfeng Zhao and Ali Haddad on the cavity ring-down experiment. It was a really inter- esting time and a lot of fun working together with you! I learned a lot and I very much enjoyed the stimulating discussions not only about science. For your projects at the VU I am wishing you all the best! A big part of my Ph.D project was concerned with the ESO observations on the Red Rectangle. I am deeply grateful to Hans Van Winckel whose help and experience in observations was making this project possible! When having first results on C2 in the observational and laboratory work, Ewine shared her knowledge on the excitation and de-excitation mechanisms with me. For the stimulating discussions and the deeper insights I am also very grateful! This paragraph is dedicated to the “core” of my colleagues in the laboratory. During

137 Acknowledgements the past years I got the chance to meet many exceptional people and I would like to thank all of you for that interesting time! Harald, Suzanne, Karin, Jordy and Sergio: It was a pleasure working with you and sharing the Ph.D. time at the observatory with you. To Edith, Karoliina, Gleb, Steven and Thanja: I am wishing you all the best for your Ph.D. “Succes!” Guido, Joseph and especially Claire who not only helped me with the science part, but also with the repair and maintenance of my many ”oil-leaking” vacuum pumps! Thank you for being there! To the new arrivals in the group: Jean-Baptiste, Emily, Junfeng and Anton, all the best for your projects! I also would like to thank my office mates and the astrochem group, who made my stay in Leiden a great pleasure and a great “international” experience! Nikta, I would like to thank you for sharing your observations on the Effelsberg (100 m dish) telescope with me! That was a fantastic experience and I enjoyed the observations very much! Andreas, I would like to thank you for introducing me to Squash. The lessons at the sports facility were very welcome after long office days. Herma, I would like to thank you for the great time I had going to “Efteling” and for explaining and introducing the dutch culture to me. Sharing a room at conferences was always a pleasure. Lars, thank you for the unforgettable experience of having a real danish christmas lunch (lasting until midnight) at your place! Ruud, Daniel, Pedro, Umut, Jeanette, Carina, Silvia, Olivier, Kalle, Christian, Isa, Irene and Nienke I always very much enjoyed your company in University and especially during all the dinners after work. An dieser Stelle möchte ich mich auch ganz herzlich bei meinen Freunden aus meiner Schulzeit bedanken: Katja, Simone, Isabel und Frank, Manuela und Martin, Conny und Silke. Danke für Eure Freundschaft, Eure vielen Besuche hier und die vielen aufmuntern- den Worte! Salvatore, auch Dir möchte ich herzlich für Alles danken. Danke, dass du da warst und zugehört hast, wann immer ich jemanden brauchte. Mein Dank gilt auch den Freunden, die mir aus meiner Unizeit in Düsseldorf geblieben sind: Fabian, Dirk, An- dreas, Christian, Qiangqiang und Kerstin. Fabian, die Reise durch Chile ist unvergessen! Helen, auch Dir danke ich für die vielen aufmunternden Worte, die vielen Besuche in Leiden und die Abwechslung, die Du jedes Mal in mein Leben bringst! Cristina, I wanted to thank you also for being a good friend and for importing the nice Granada teas year by year to the Netherlands! The tea was always very much appreciated! Letzendlich und vor Allem möchte ich aber meiner Familie danken: Mama, Papa, Janine und Michelle! Ihr seid grossartig und die beste Familie, die man sich wünschen kann. Ohne Eure Hilfe und Euren Zuspruch wäre ich niemals so weit gekommen! Ihr bedeutet mir Alles! Meinem Patenonkel Manni und Hilde einen dicken Kuss. Diese Arbeit ist auch für Euch!

138 “So eine Arbeit wird eigentlich nie fertig, man muss sie für fertig erklären, wenn man nach der Zeit und den Umständen das Möglichste getan hat.”

– Johann Wolfgang von Goethe