分析在類星體風中形成塵埃的成分比例

全文

(1)Department of Physics National Taiwan Normal University Master Thesis. Determining the Composition of Dust Formed in Quasar Winds. 彭渙婷 Huan-Ting Peng Supervisors: Francisca Kemper & Sébastien Foucaud & Wo-Lung Lee.

(2) 致謝「當你全心全意追求一個目標的時候，整個宇宙都會聯合起來幫助你!」。當我第一次看到這段話，內心中充滿鼓舞。因此，跟隨我的心，變成了我的信仰。每當夜幕低垂，滿天星斗灑滿天空，我渴望了解這個宇宙。然而，依著我的心，來到了物理所，卻發覺無法解決我的困惑。在研究生涯追尋知識的過程中，深感人類的渺小，自己知識的限制，總在困惑之中發現更多困惑。而我們僅能用人類僅知的些許知識，試圖解釋。這段學術學習旅程，並非單靠著我一人就可完成。感謝在研究初期，給我空間學習待人處事，以及磨練語言表達能力的 LeCosPA 團隊的老師及朋友們，謝謝你們的精神支持;也感謝 ASIAA 提供完善的環境、友善的團隊，讓我安心、放心的專注在研究上面，不用擔心經費、資源的問題。感謝 Sundar 細心指導我程式編寫，也謝謝圖書館的館員，提供良好舒適的圖書館環境，讓我靜心撰寫論文。感謝我的家人，包容著不常回家的我，更感激一路走來，幫我處理許多家事、心事的妹妹。更感謝三位願意給予無限耐心，並於脆弱時，提供心靈協助的指導老師:李沃龍、Sébastien Foucaud 和 Francisca Kemper 的細心帶領;也感謝一路走來，男友彥緯義無反顧的支持與陪伴。.

(3) Keyword: Quasar, Dust, Composition.

(4) Contents. 1 Introduction 1.1. 1.2. 1.3. 3. Physical processes in AGN . . . . . . . . . . . . . . . . . . . .. 3. 1.1.1. Supermassive Black holes. . . . . . . . . . . . . . . . .. 3. 1.1.2. Accretion disks . . . . . . . . . . . . . . . . . . . . . .. 5. 1.1.3. Radio Jets . . . . . . . . . . . . . . . . . . . . . . . . .. 6. 1.1.4. Dust Torus . . . . . . . . . . . . . . . . . . . . . . . .. 8. Active Galactic nuclei (AGN) . . . . . . . . . . . . . . . . . . 11 1.2.1. Seyfert galaxies . . . . . . . . . . . . . . . . . . . . . . 11. 1.2.2. Quasars . . . . . . . . . . . . . . . . . . . . . . . . . . 12. What is in the torus? . . . . . . . . . . . . . . . . . . . . . . . 13 1.3.1. Dust around the AGN . . . . . . . . . . . . . . . . . . 14. 1.3.2. Basic properties of Interstellar Dust Grain . . . . . . . 16. 2 Target and Sample. 22. 2.1. Target selection . . . . . . . . . . . . . . . . . . . . . . . . . . 22. 2.2. Spitzer IRS data . . . . . . . . . . . . . . . . . . . . . . . . . 25. 3 Methods. 29. 1.

(5) 3.1. Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29. 3.2. Q-values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.2.1. What is Q-value . . . . . . . . . . . . . . . . . . . . . . 33. 3.2.2. Relationship with opacity . . . . . . . . . . . . . . . . 34. 3.2.3. Two models to compute Q-values . . . . . . . . . . . . 35. 3.3. Selected Species . . . . . . . . . . . . . . . . . . . . . . . . . . 36. 3.4. Fitting method . . . . . . . . . . . . . . . . . . . . . . . . . . 40. 4 Results and Discussion. 44. 4.1. PG 2112+059 with 2 different methods . . . . . . . . . . . . . 44. 4.2. Common dust features of AGNs . . . . . . . . . . . . . . . . . 46. 4.3. Comparison between sources . . . . . . . . . . . . . . . . . . . 49 4.3.1. Board Absorption Line (BALs) . . . . . . . . . . . . . 49. 4.3.2. non-BAL and mini-BAL . . . . . . . . . . . . . . . . . 50. 5 Conclusion. 61. A Eddington limit. 63. Bibliography. 68. 2.

(6) Chapter 1 Introduction 1.1 Physical processes in AGN AGN (Active Galactic nuclei) are galaxies characterized by a very active and bright center. A lot of observational evidences lead us to suspect that an AGN is originated by a central supermassive black hole, and its accretion disk. Furthermore we have evidence that some AGN possess jets and a dusty torus (Antonucci, 1993).(Fig.1.1). 1.1.1 Supermassive Black holes Dynamical evidences indicate that AGN and galaxies in general, host supermassive black holes with masses in the range MBH ∼ 106 − 1010M (Elitzur and Shlosman, 2006; Krawczynski and Treister, 2013; Madejski, 2003 ). In fact, there are compelling evidence that our own galaxy hosts a central blackhole with a mass of  4 × 106M (Ferrarese and Ford, 2005; Schödel et al., 2003; Ghez et al., 2005; Volonteri, 2010). Some powerful quasars with Bolo3.

(7) Figure 1.1: The structure of an AGN. All classifications of AGN accord to the same physical phenomenon, such as radio galaxies, blazars, or Seyferts and so on. They all compose of a supermassive black hole (SMBH) in their center, and the large accretion disk of orbiting gas and dust with high energies around the SMBH. The high-energy matter and radiation are often ejected to form the jets. It also shows the size of central black hole, accretion disk and the thick dusty torus, and what type of object you see depends on the what viewing angle for observing. When the AGN is seen from some views, the torus will obscure the accretion disk, BLR and SMBH (Beckmann and Shrader, 2013). metric Luminosities > 1047 erg s−1 have been detected at z > 6 (Woo and Urry, 2002; Novak et al., 2012), about one billion years after the Big Bang. Through observations, it is confirmed that at least some of these quasars are powered by super-massive black hole with a mass of ' 109 M

(8) (Fan et al., 2001). This implies the first SMBHs (supermassive black holes) must have formed very early. Such a rapid formation of SMBHs is difficult to understand and little is 4.

(9) known about their origin. The current best guess is that they grow from gravitational collapse of small perturbations in a early quasi-homogeneous Universe. There are three main scenarios for the formation of SMBHs in the nearby universe, namely which include the accumulation of the remnants of the first generation of stars, gas-dynamical instabilities, and stellardynamical processes. For more detail, the reader is referred to Volonteri (2010).. 1.1.2. Accretion disks. The extremely strong gravitational field causes the material in the core of the galaxy to orbit the SMBH and form a disk. In principle, this disk is stable. However, when the amount of matter orbiting is high, friction can cause the matter to slow down and tighten its orbit, causing a contraction of the disk. Eventually, some material in the disk moves inside the innermost stable orbit, forcing the material in the accretion disk to fall onto the central compact object. Heat will be generated by the release of gravitational potential energy, and through dissipation the disk is heated to high temperatures, typically 105 K (Bonning et al., 2007), provoking high luminosity of the disk, which is responsible for the AGN phenomenon. The luminosity and temperature of accretion disks can be estimated, assuming a steady state and changes in the gravitational potential energy of M˙ (grams of matter per second) are being turned into radiated energy. There is an upper limit to the luminosity, which is known as the Eddington limit. The typical range of luminosity is from a few percent up to nearly the. 5.

(10) limiting value. (for details see Appendix) However, not all SMBH have an active accretion disk. For instance, the SMBH at the center of the Milky Way lacks an accretion disk. The mass of the central SMBH of the Milky Way is about 106 M

(11) . The accretion disk around the SMBH may have been active in the past, but if all the matter in the accretion disk has fallen onto the black hole and the accretion disk is not replenished, then the Milky Way stopped its activity nucleus. This kind of galaxies are called “dead quasars”. It means they may once have been quasar, but their center is not active anymore, because the black hole can not radiate at all without accretion disk. This means as well that if a accretion disk is formed again due to external trigger the galaxy can light up as an AGN again.. 1.1.3. Radio Jets. AGNs can be further classified into two categories; characterized by their luminosity in radio bands: radio-loud, and radio-quiet. 10% ∼ 20% of AGN are radio-loud, and display bright radio counterparts emission. They are about three times brighter in X-ray than the radio-quiet (Bradley W. Carroll, 2006). The radio-loud AGN usually have one or two detectable jets, and one or two radio lobes. Jets are constituted of a plasma, ejected from the nucleus in two opposite directions in a well-collimated outflow, and they are very narrow. They are detected easily because their length can reach to several kiloparsecs scale (Krawczynski and Treister, 2013). The charged particles emitted by the central engine feed the lobes, which could. 6.

(12) extend up to megaparsecs (see Fig.1.2) (Robson, 1995). The radio lobes are produced by interaction of the jets of charged particles and the inter galactic medium, and they originate at relativistic speeds from the central nucleus of the AGN.. Figure 1.2: This picture shows the AGN is at the center of a galaxy (right top), and the lobes is fed by the charged particle of outward jets. There are many gases and clouds within the torus around the nucleus. Illustration is from the (Schilling, 2001). Jets are narrow and collimated with direction, and they are produced by the magnetic field at the surface of accretion disk of the central object, which is twisted along the rotation axis of the central object, and eject particles outwards. It is hypothesized that fthe jets are electrically neutral since they 7.

(13) are composed of electrons, positrons, and protons (Bradley W. Carroll, 2006). Forming these collimated jets requires huge amounts of energy, hence some jets are considered to be generated by the SMBH directly via the Blandford-Znajek mechanism (Bradley W. Carroll, 2006). The BlandfordZnajek mechanism assumes the magnetic fields around black hole should be dragged, and the field lines will be gathered, and then shooting out to form the well-collimated jets (Bradley W. Carroll, 2006; Fanidakis et al., 2011). The material around the accretion disk escapes the accretion system, in a collimated way as jets, otherwise, in an uncollimated as winds or outflows (Krawczynski and Treister, 2013).. 1.1.4. Dust Torus. The central engine, is surrounded by an obscuring dusty cloud (Fig.1.2). According to recent observations using high-resolution IR, the torus size might be not more than a few parsecs (Elitzur, 2005; Elitzur and Shlosman, 2006; Krawczynski and Treister, 2013). For instance, the torus size in NGC1068 was observed with the VLTI at 12 µm, and the FWHM size of the emission is only ∼4 pc observation (Jaffe et al., 2004). The dusty cloud, also known as the torus, displays a toroidal, doughnut-like structure (Krolik and Begelman, 1988; Elitzur and Shlosman, 2006), presumably populated by dust and optically thick molecular clouds accreted from the galaxy. Therefore, these few parsec sized torus obscures the broad-line region (BLR) and accretion disk emission from view and the emission from central engine (ie:SMBH) in the infrared and optical (Krawczynski and Treister, 2013; Elitzur, 2007;. 8.

(14) Nenkova et al., 2008) (see Fig.1.1). However, the nucleus keeps heating the inner edge of the torus, and emits ionized radiations spherically from the central source, leading to be a surrounding spherical mist of cloudlets (Fig.1.3). There are high-ionization lines in the front, due to the presence of He II, He I, O VI, N V, and C IV; however, if the cloud is thick enough, the back of cloud will also emit low-ionization lines due to the presence of Ca II, Fe II, O I, and Mg II, which are well-known components of the AGN spectra (see Fig.1.3)(Gaskell, 2009).. Figure 1.3: left: The cartoon shows a common view of BLR. The clouds around nuclei are heated by its center. right: If the clouds is thick enough, the inner of clouds will be highly ionized and the back side will be nearly neutral. (Gaskell, 2009). Determining of the actual size of the torus must rely on their IR emission. Thanks to the current generation of high-resolution IR observations, we have 9.

(15) more information about the dust around AGN. The formation temperature of dust is very low, is only about 2000K. Thermal emissions from objects at these temperatures and below, are displayed as peaks in the infrared wavelength. This is the reason why we select the infrared telescopes for obtaining the spectrum of dust. The spectrum of Type 1 AGNs peaks at λ ≥ 1µm in IR (and the short wavelength: X-rays) with a power-law characteristic of accretion disks, and present both narrow and broad emission lines; in contract, the AGN of type 2 be obscured torus dust, so the spectrum of type 2 will display only narrow lines in the IR emission (Elitzur, 2007). The IR observations classify the AGN as type 1 and type 2, which may just be an angle-dependent view, and the broad-line region(BLR) and the dusty torus are the same. All the AGN have the same structure (supermassive black hole, accretion disk, and dusty torus), but we obtain different spectra by different viewing angle of observation. Because the torus provides anisotropic obscuration, type 1 AGN sources viewed face-on and type 2 edge-on. We know the narrow lines are emitted from the outside of nuclei, and the broad ones closer to the black hole in the center of AGN, Therefore an object observed edge-on (type 2) will have its center (near the black hole), which is hidden by the dust torus, and then will not present any broad line emissions. The theoretical analysis of the spectral energy distribution (SED) play an important role, Pier and Krolik (1992) computed a detailed dust radiative transfer model using for the first time a toroidal geometry. By comparing with the IR data from Granato et al. (1997), they estimated the torus size to be of several parsecs. Furthermore, the SEDs show the unification prediction 10.

(16) result: type 1= type 2 + AGN (Elitzur, 2007). Therefore, the clumpy dust torus play a fundamental role in the observation of AGN by: 1. Anisotropic obscuration of the central region led to the AGN are recognized as type 1 objects for sources viewed face-on and type 2 objects for sources viewed edge-on (Schmitt et al., 2001). 2. Thermal reemission by dust in the IR of the absorbed radiation from the AGN (Elitzur and Shlosman, 2006).. 1.2 1.2.1. Active Galactic nuclei (AGN) Seyfert galaxies. Seyfert galaxies are the most common type of galaxies with bright nuclei. They were first identified as an independent class by Seyfert (1943). He found some galaxies with some distinguishing characteristics: very bright emission lines of hydrogen, helium, nitrogen, and oxygen in their spectrum (Seyfert, 1943). The emission lines display strong Doppler broadening, which exhibit velocities up to 8500 km/s, and the hydrogen lines sometimes are broader than the other lines. These features indicates there is a bright nuclei in AGN. Woltjer (1959) studied the physics of Seyfert galaxies, and concluded that: 1. The size of the nucleus is less than 100pc (We know it is even much more compact than 100pc now.). 2. The nuclear emission must last more than 108 years. 3. The mass of the center should be very high, because the material in. 11.

(17) the nucleus is gravitationally bound. The velocity obtained from the widths of the emission lines are large up to 8500km/s. Assuming an upper limit r≤100pc (the above-mentioned) for the virial argument M ≈ ν 2 r/G, and that the AGN is spatially unresolved. Therefore, the upper limit need in place of a lower limit r≥1pc of the size of the emitted nucleus. Thus the mass of nucleus could deduce M ≈ 109±1 M

(18) (Peterson, 1997). These special galaxies are called Seyfert galaxies, and are a subclass of Active Galactic nuclei. The center of AGN is believed to a supermassive black holes with 106 ∼ 1010 Solar masses (Elitzur and Shlosman, 2006; Krawczynski and Treister, 2013; Madejski, 2003), and the emission lines should come from the accretion disk or clouds of gas illuminated by the central engine.. 1.2.2. Quasars. Quasars were first discovered as a result of a radio survey by Allan Sandage in late 1950s. He found that the galaxy 3C 48 (third Cambridge catalog) presents some broad and bright lines in a strange position of the spectrum. Schmidt (1963) found common properties in the spectrum of 3C 48 and 3C 273, and identified the emission lines as hydrogen emission lines but with very high redshift. A year later, Schmidt and Matthews (1964) studied a large amount of quasars, and identified their common properties (Schmidt, 1969b,a): 1. Those point-like sources are similar to stars in the optical which is the. 12.

(19) reason those objects are called Quasi Stellar Objects (quasars). But they can be identified with radio sources. 2. The continuum flux varies with time 3. They can be at very high redshift 4. They have a broad spectral energy distribution (SED) 5. They have a large UV flux In fact, QSOs are a class of AGN but with a nuclear source more than 100 times brighter than the host galaxy. Nowadays, we know the AGN spectrum is described in non-thermal processes and incoherent synchrotron radiation rather than blackbody emission at a single temperature. The broad band SED of a quasar continuum should display as a power law. Fν = Cν −α ,. (1.1). where C is a constant, α is the power-law index, and Fν means the specific flux (Peterson, 1997). Type I Quasars are the chosen target sources in this study since the spectrum of quasar is not much obscured by the torus (see Fig.1.1 the viewed of observation for quasar). We can study the dust components around the center or in the BLR via the radiative transfer.. 1.3. What is in the torus?. It is found that AGN are very dusty and that the dust concentrates in a geometrically and optically thick torus or dusty clouds surrounding the cen-. 13.

(20) tral black hole and accretion disk.The dust is composed of heavy (refractory) elements produced by previous generations of stars. So far, the processes of formation of such structure of AGNs have not been understood totally, and we are curious about what is the component for the torus. Where dose the dust come from?. 1.3.1. Dust around the AGN. The gas and dust around the hot object or black-body source absorbs the photons, coming from the nuclei of source. These photons with specific frequency resonate with some molecular or lattice in the gas and dust, which has greater amplitude at this specific frequencies than at others, for instance, the silicate with 10 and 18µm feature. Therefore, the spectrum has their characteristic, and we can analyze those characteristic for determining the compositions. It is generally believed the dust in the universe comes from the winds of late-type giant and supergiant stars since the environment could provide the low temperature (around 600 - 1400 K) and high density conditions (Frenklach and Feigelson, 1989; Frenklach et al., 1989). The reason is the temperature in the outer layers may be lower and temperatures below 2000K are good for forming dust when a star becomes old and expands. On the opposite, in the AGN, when the clouds are closer to the supermassive black holes, the temperature is very high, so it is difficult to form or maintain the dust grains around the AGN generally, the environment near the nucleus is even hostile to the dust. When away from the nucleus of AGN, the tem-. 14.

(21) perature could decrease low enough to allow dust formation. Therefore, it is considered the dust formation in the quasar does not happen normally (Elvis et al., 2002). However, the clouds of gas in the broad emission lines (BELs) region around the center of AGN (see Fig.1.1), and the cloud qof gas will be heated by the accretion disk with high temperature (see section 1.1.4), then the clouds of gas will be partially ionized. The high energy photons from the central engine will produce radiation pressure, which could cause a part of ion particle motion with high velocity and escape from AGN then form magnetocentrifugal quasar wind (Gaskell, 2009; Gallagher et al., 2013). The flow of outflowing winds of highly ionized material is similar to BEL clouds. According to Clavel et al. (1989), the outer radius of the region which cause strong BELs correspond with the dust sublimation radius, so Elvis et al. (2002) proposed a alternative model to form dust in quasar winds. Good condition for dust formation are gathered when the clouds in the broad emission lines (BELs) expand freely. The model assumes the BEL region to be in pressure equilibrium with surrounding warmer medium. But the outflowing warm winds could destroy this stable system of pressure, effect the BEL cloud (BELCs) to keep the system in balance. Under sufficiently high pressure, when the (BELCs) expand, the temperature of that region will be cooler, leading to suitable conditions for dust formation (see Fig.1.4). This process makes the quasars create the dust resembling soot; they are called ”smoking quasars” (Elvis et al., 2002). Typically, the temperature for forming dust is about 600-1400K or somewhat higher (Frenklach et al., 1989; Frenklach and Feigelson, 1989). This con15.

(22) dition is common in the late-type giants and supergiants (Sedlmayr, 1994). However, Elvis et al. (2002) found the conditions in expanding BELCs overlap the AGB ones (see Fig.1.4), so they supposed conditions in the quasar are similar to AGB ones and the quasar could form the dust by itself. The model by (Elvis et al., 2002) provides us with a new direction to think where the dust come from. However the model is theoretical, are we need the data to confirm it. Therefore, Markwick-Kemper et al. (2007) used the 5-40 µm Spitzer IRS mid-infrared spectrum of the broad absorbtion line (BAL) quasar PG 2112+059 to study and determine for the first time the dust components of an AGN. They reduce the overlapped data subtract the background continuum spectrum, and apply the flux calibration methods used in Bouwman et al. (2006). The result indicate the presence of different elements around this quasar, such as corundum (Al2 O3 ) and periclase (MgO), in addition to silicate. The coexistent highly refractory corundum, less refractory MgO, and crystalline or amorphous silicates, reflects an environment of PG 2112+059 with inhomogeneous temperature and density (Markwick-Kemper et al., 2007). We discuss these results in more details in section 4.1.. 1.3.2. Basic properties of Interstellar Dust Grain. Dust grains are solid and macroscopic particles, composed of dielectric and refractory materials. The composition of dust reflects the abundance of certain elements in the universe, which include hydrogen, oxygen, carbon, nitrogen, and silicon. Those common elements make up water (H2 O), methane. 16.

(23) (CH4 ), carbon dioxide (CO2 ), ammonia (NH3 ), and silicates (SiO−4 4 ), With a solid-state temperatures low (about 100K), that they are under form of ice. A model has been developed for describing the dust grain properties (see Fig.1.5). It assumes the core of the dust grain is very small (about 0.05µm) and may consist of silicates, iron, or graphite, silicate being in majority. Outside of the inner core, there is a about 0.5µm which mantle made of icy materials (Zeilik, 2002). If the grain enter a hot environment, the mantles will be evaporated, and then the grain consist of a bare core only. Infrared wavelengths can be used to investigate interstellar dust and get more information about the emissivity of dust grain, especially, from the silicate and ices (Zeilik, 2002). Because we know the temperature of formation dust (or non-evaporate) is below 2000K (more details at table.1.1 ), it is easy to estimate the wavelength emitted from dust via the Wien displacement law (Eq.1.2): Condensate H2 O Carbon Silicon Carbide Fe3 O4 Fe3 C Al2 O3 MgSiO3 MgSiO4 Iron. Condensation temperature (K) 150-200 1850 1500 400 1100 1760 1350 1440 1470. Table 1.1: The condensation temperature of some species (Evans, 1993).. λmax TW ien = 2890µmK, 17. (1.2).

(24) which λmax means the wavelength of maximum intensity and TW ien is temperature. Therefore, we could estimate the wavelength emitted from dust is ≥1.5µm, which is in the infrared. Although in the real case, the materials do not emit as perfect blackbodies, it is absorbed or emitted by some matters. Therefore, the infrared observation can give us the much useful information for the dust, and we could get the observed infrared flux distribution with different range of grain temperatures and sizes (Evans, 1993). There are a lot of gas and dust in the universe, and the gas-to-dust ratio could be estimated from the interstellar extinction (see Table.1.2). Deducing from the density of the interstellar gas and dust grains, we get: ρ(dust particles) ∼ 10−2 . ρ(gas). (1.3). The gas-to-dust ratio is about 100:1 in galaxies. The ISM, contains about 10% of the baryonic mass of the galaxy, including the 0.1% dust grain (Evans, 1993). This implies a large amount of heavy elements (heavier than hydrogen and helium), are linked up with dust grain, and the elemental abundances will decrease, with increasing atomic number. Ingredient Free electrons Molecules Small(a λ) particles Large(a λ) particles Particles with a ∼ λ. Extinction law λ0 λ−4 λ−4 λ0 λ−1. Cross-section σT ∼ πa2 πa2 πa2 Qext πa2. Density (kgm( − 3)) 4.1 × 10−22 4.4 × 10−26 1.5 × 10−27 1.5 × 10−27 1.2 × 10−23. Table 1.2: Interstellar extinction (Evans, 1993).. The dust grain is mostly composed of silicon atoms, which in fact is a 18.

(25) major component of interstellar medium in the galaxies (Hao et al., 2005). Since the atoms of silicon do not subsist alone, they always form networks with oxygen and other abundant elements, such as Fe and Mg, and sometimes link up with Ca and Al. In fact, oxygen and silicon are the most common non-gaseous and non-metallic elements in the dust composition. The silicon also tie with other abundant elements, In the laboratory, the spectrum of silicate dust has been measured. It has two spectral features: one is Si-O stretching mode, and the other is O-Si-O bending mode. The emission feature at the first one is 10 µm and the second one is 18µm (Knacke and Thomson, 1973). In the observation, we can indeed find the two spectral features are in the infrared region. The silicate emission was detected by Hao et al. (2005); Sturm et al. (2005); Siebenmorgen et al. (2005).. 19.

(26) Figure 1.4: This is dust formation window, which is given by the temperature gradient (T∝R−0.4 ) that is determined by the radiative transfer (Ivezic and Elitzur, 1997). It shows the range of the appropriate condition (the temperature and pressure) for forming dust. Below the thin solid lines, the dust precursor will be formed in the AGB case (adapted from Lodders and Fegley (1999)), and the thick solid lines show the range for forming the dust in the expanding BELCs. Top: Phase transition lines for O-rich; Bottom: Phase transition lines for C-rich environments.. 20.

(27) Figure 1.5: Simplified model for an interstellar dust grain.(Zeilik, 2002). 21.

(28) Chapter 2 Target and Sample 2.1. Target selection. Based on the model developed by Elvis et al. (2002) (see section 1.3.2), outflowing winds of quasars may be suitable environments for forming dust. We selected some quasars which display different degree signature of outflowing winds for analyzing their dust composition. We chose five radio-quiet quasars, including two BALs quasars with winds, two non BAL quasars (one with very strong outflows and the other without noticeable outflow), and a mini-BAL quasar with weak outflows. The spectrum of type 1 quasars show broad and blueshifted absorption resonance transitions in ultraviolet, indicating exposed gas from the center, and generating an observed spectrum with P Cygni-type features. Those features could be evidences for existence of outflowing winds in the type 1 quasar, and are more obvious in BAL quasars, which represent 20% (Hewett and Foltz, 2003) of optically selected type 1 quasars. 22.

(29) Silicate is a major element in the composition of galactic dust, so it can be used as a racer of presence of dust for selecting our sources. The silicate lines will be in emission for the type 1 AGN (Laor and Draine, 1993), therefore, we choose some spectra with high-luminosity, high signal-to-noise ratios, and obvious silicate emission feature both at 10 µm and 18 µm to be included in our sample. Two of them, PG 2112+059 (Gallagher et al., 2004; Barvainis et al., 1996) and PG 0043+039 (HTS archive), are two of the MIR-brightest BAL quasars. In addition, we also select two non-BAL quasars, PG 0050+124 and PG 1211+143, to be antithesis. Although PG 1211+143 (Pounds et al., 2003) is not a BAL, it presents very strong outflows and PG 0050+124 (I Zw 1) Hao et al. (2005); Barvainis et al. (1996); Dhanda et al. (2007) is a non-BAL without noticeable outflows. Finally, the last one, PG 1351+640 Hao et al. (2005); Zheng et al. (2001), is a mini-BAL with weak outflow. These quasars are all radio-quiet; it support us focusing on comparing their components and figuring out the relationship between components and outflows winds. (more details in table. 2.1) Although these quasars have dissimilar redshifts (three of them with local redshifts, PG 1351+640 is 0.0881, PG 0050+124 is 0.0808 and PG 1211+143 is 0.0611; two higher redshifts, PG 2112+059 is 0.466 and PG 0043+039 is 0.385), we assume the physical processes for forming dust owe to be the same in the galaxy at any redshift.. 23.

(30) 24. T ype shallow BAL, broad, blue smooth BAL, broad, red mini BAL, weak outflow non BAL, weak outflow non BAL, strong outflow. Reference Gallagher et al. (2004); Barvainis et al. (1996) HST archive Hao et al. (2005); Zheng et al. (2001) Hao et al. (2005); Barvainis et al. (1996); Dhanda et al. (2007) Pounds et al. (2003). Table 2.1: Sample Characteristics. We selection some different type quasars for analyzing, but they all are radio-quiet Type 1 QSOs. Name z PG 2112+059 0.466 PG 0043+039 0.385 PG 1351+640 0.0881 PG 0050+124 0.0808 PG 1211+143 0.0611.

(31) 2.2. Spitzer IRS data. The Spitzer Space Telescope is an infrared space observatory. The range of its observed wavelengths is 3-180µm, and it carries three instruments onboard: IRAC (Infrared Array Camera), IRS (Infrared Spectrograph), MIPS (Multiband Imaging Photometer for Spitzer). Here we use data from IRS (Infrared Spectrograph), which is composed of four sub-modules: two lowresolution modules (R ∼ 60-120), Short-Low (λ = 5.2 − 14.5µm) and LongLow (λ = 14.0 − 38.0µm) (abbreviate SL ans LL respectively) and two highresolution modules (R ∼ 600), Short-High (λ = 9.9−19.6µm) and Long-High (λ = 18.7 − 37.2µm)(abbreviate SH and LH) (Werner et al., 2004; Houck et al., 2004; Lebouteiller et al., 2011). (more details in the Table. 2.2) Our sources were all observed using InfraRed Spectrograph (IRS) and we download these publicly-available data from the CASSIS website1 . We included PG 2112+095 to be our first target, because it has already been analysed by Markwick-Kemper et al. (2007), and we can use it as consistency check for our new model. Once those data downloaded from CASSIS, we need to adjust the overlapping wavelength coverage at the edge of every module. The overlap regions have many redundant data, which come from the two different modules and can lead to the spectrum being badly calibrated (Fig.2.1). To remove the overlapping data, we need to determine precisely the boundary at the edge of modules. In the data files provided by CASSIS, the module sequence for the spectrum is ranked from small to large wave1. http://cassis.astro.cornell.edu/atlas/. 25.

(32) Module SL SL SL LL LL LL SH LH. Order(s) 1 2 3 1 2 3 11-20 11-20. λ(µm) 7.4-14.5 5.2-7.7 7.3-8.7 19.5-38.0 14.0-21.3 19.4-21.7 9.9-19.6 18.7-37.2. Aperture Size (”) 3.7 × 57 3.6 × 57 3.6 × 57 10.7 × 168 10.5 × 168 10.5 × 168 4.7 × 11.3 11.1 × 22.3. Pixel Size (”) 1.8 1.8 1.8 5.1 5.1 5.1 2.3 4.5. Table 2.2: Modules of the Spitzer /IRS (Lebouteiller et al., 2011).. length as SL2, SL3, SL1, LL2, LL3, LL1 (see the Table.2.2 and Fig.2.1). Only SL2, SL1, LL2, and LL1 data are really relevant to our analysises here, and the SL3 and LL3 data are just used for verifying whether the connection between SL2 to SL1 and LL2 to LL1 is smooth. The four modules have three overlap regions: between SL2 and SL1, between SL1 and LL2, and between LL2 and LL1, so we impose cuts at 7.5 µm, 14 µm, and 20.6µm. We use these boundaries to cutoff any unwanted data in each overlapping module. Furthermore, the observed data need to corrected for the cosmological redshift effect due to be universe expansion. The observed spectra are redshifted, and the laboratory data using in our model are in the rest frame, so we need to convert the observed spectrum to rest frame for fitting the model in the same coordinate system. Redshift may be characterized by the relative difference between the observed and emitted wavelengths of an object. In astronomy, the change is given by a dimensionless quantity, which is called redshift z. The z value is defined by the equation: 26.

(33) Figure 2.1: For instance, the data of PG 0043. Different module covers different wavelength range. z=. λobs − λemit , λemit. (2.1). where λobs is the wavelength we observed and the λemit represents the wavelength emitted from the object in the rest frame. Therefore, if we want to convert our observing data to the rest frame, we can use. λemit =. λobs , (1 + z). (2.2). which is also we use in our code for converting the wavelength to rest frame. After above procedure, we can get the spectrum of every sources (Fig.2.2).. 27.

(34) Figure 2.2: This plot shows the spectrum of five quasars we analyzed, including PG 2112+059, PG 0043+039, PG 1351+640, PG 0050+124, and PG 1211+143. They are all plotted by the data which go through the process of removing the overlapping and shifting the redshift.. 28.

(35) Chapter 3 Methods 3.1. Model. The direct environment of quasars is considered not suited for dust formation owing to its high temperature. However a large amount of dust is detected around quasars at somewhat larger distances, even at high redshift. Elvis et al. (2002) proposed a model for explaining the dust formation where dust can form in dense clouds entrained in high-velocity winds which lift of accretion disk powered by radiation pressure. In their model, they show that certain environments in the quasar wind are similar to the conditions in the outflows of asymptotic giant branch (AGB) star, which are well-known dustforming environments (see section 1.3.2). Based on the model by Elvis et al. (2002), we would like to know whether the composition of the dust in the quasar wind is the same as that in AGB or not. Markwick-Kemper et al. (2007) obtained a 5-40 µm spectrum of PG 2112+059 from the Spitzer Space Telescope, and analyze the composition 29.

(36) using some species commonly seen in AGB stars. The spectrum observed contains both continuum and dust emission feature spectrum, following:. Fν,obs = Fν,cont. × fsharp ,. (3.1). where Fν,cont. is continuum spectrum and fsharp represents sharp emission or absorption features. They used a power-law continuum model (Fν = Fν,0 ν α ), where α is the power-law index and Fν,0 is the continuum flux renormalisation, to fit the spectral energy distribution of PG2112+059; then they could get the best parameters value for Fν,0 and α respectively (Fig. 3.1). Their best parameter values of power-law index are α = −0.617 ± 0.004 and α = −0.674 ± 0.003 (the latter values are determined after adding photometric data to constrain the continuum model). After fitting the continuum model (Fig.3.1), they could fit the sharp features of species (fsharp ), and decompose the continuum-divided spectrum between different species (Fig.3.2). However, there is a degeneracy between the parameters for power-law continuum model, and the fitting of the dust species. We want to improve the method to avoid fitting inaccuracy. Therefore, in this study, we fit the publicly-available IRS spectra of AGNs with a power-law continuum model, and an optically thin dust component simultaneously. Our model in this study is:. Fmod [λi ] = (Aλαi )(1 +. X j. 30. aj qji ).. (3.2).

(37) Figure 3.1: The spectrum of PG2112+059. Top: Fitting the spectrum with two power-law continuum models. dashed curves: power-law model using the photometric data; thin-solid curves: the model using the spectroscopic data.).Bottom: The spectrum of PG2112+059 with two different continuum models, subtracted showing excess emission from 5µm to 25µm. (MarkwickKemper et al., 2007). 31.

(38) Figure 3.2: The flux of spectrum divided by the flux of suitable continuum model which is found at Fig.3.1 (Eq.3.1). It shows there are five main dust species in this spectrum: amorphous olivine (dotted line), forsterite (dashed line), corundum (Al2 O3 ; dash-dotted line), MgO (dash-triple-dotted line), and interstellar PAH (solid line in the panel 1 at the 5.5-8.5µm range). The two panel show the difference between using different continuum model (The solid and dashed line in the FIG.3.1). The thick solid line here is the total fit to the data. A difference between the two panels can be seen in the MgO contribution in particular. (Markwick-Kemper et al., 2007). 32.

(39) A is a normalization factor; α is the power-law index; aj represents the scale factors for dust opacities and the qji is the absorption coefficient for each individual species. The first part, Aλi , determine the power-law continuum P α , fit the model for the spectrum of sources, and the second part, 1 + j aj qji emission features of every considered species of dust for our sources. This model is based on the description by Markwick-Kemper et al. (2007), but fitting the continuum and emission features at the same time. The number of parameters depends on how many species we chose to fit. For instance, using six species in this study, Eq.(3.2) will become:. Fmod [λi ] = (Aλαi )(1 + a1 q1i + a2 q2i + a3 q3i + a4 q4i + a5 q5i + a6 q6i ).. (3.3). There are eight parameters in the above case, because the A and α stand for two parameters, and every species represent a parameter individually. Fitting all the parameters simultaneously will result in a more precise and systematic derivation of the dust composition and the continuum emission.. 3.2 3.2.1. Q-values What is Q-value. All of the species are measured in the laboratory, except for the PAHs, which are observed. Same as the method used in Markwick-Kemper et al. (2007), the Q value can reflect the different opacities for different components. The Q is extinction efficiency of the dust grain, and it represents the. 33.

(40) scattering and absorption efficiency as following:. Qext = Qscat + Qabs. (3.4). The Q-values characterize the spectrum of every dust species. That is the reason why we use the method for determining the components observed in the interstellar medium.. 3.2.2. Relationship with opacity. We need to determined Q-value for analysing the dust species. For instance, Forsterite, only κ (opacity coefficient)is provided. Therefore, unifying the unit is our first work. This relationship can deduced from Effective Cross-section extinction if the dust grain is assumed to be spherical with a radius as:. σ = πa2 Qext. (3.5). where σ is cross section. Then the Q can infer to. Qλ ≡. σλ . πa2. (3.6). Additionally, κ can be deduced by Optical Depth, and then we obtain the relationship between cross section per mass:. κ = σ/m. 34. (3.7).

(41) By Eq.3.6 and Eq.3.7, the relationship between Qλ and κλ can be showed: κλ =. πa2 m. Qλ ,. (3.8). where m is mass and a is radius of a dust grain, and it shows the relation between κλ and Qλ depends on the grain size and grain shape. Thus, we can unify the unit of all the species by Eq.3.8.. 3.2.3. Two models to compute Q-values. Mie model The interaction of light and particles could provide us a direction for analyzing the composition of dust. By solving Maxwell’s equation of electromagnetics, we can find out what the effect on the radiation is after it encounters single particle. In 1908, Gustav Mie worked out the rigorous solutions for the scattering of electromagnetic radiation by a sphere (homogeneous particles), therefore, this method is called Mie theory (Mie 1908). We applied Mie theory on a set of optical measured in the laboratory to calculate the opacity of spherical grains with a size of 1 µm. In this study, we use Q-values for three of species, determined assuming from Mie theory. Those include Fe-rich amorphous olivine (MgFeSiO4 ), Fepoor amorphous olivine (Mg2 SiO4 ), and alumina (Al2 O3 ) (Fig.3.4). CDE model If we assume the grain is ellipsoid, non-spherical, we need to consider both the size as well as the shape of the grain. In this case, we have one more 35.

(42) parameter, sharpness. The grain can be spherical, a little non-spherical, or strange ellipsoid, and assuming the number of grains of all types are all the same, we can average over the different shapes. This method for assumption of equal probability for all shapes grains is called ”Continuous Distribution of Ellipsoids” (CDE). We also use Q-values for the same three species assuming CDE model (Fig.3.5).. 3.3. Selected Species. We used the opacities for six different species (alumina (Al2 O3 ), periclase (MgO), Polycyclic Aromatic Hydrocarbons (PAH), Fe-rich amorphous Olivine, Fe-poor amorphous Olivine, and Forsterite), which are common minerals in terrestrial, determined from laboratory data using the Continuous Distribution of Ellipsoids (CDE) model or Mie Theory (Mie). Amorphous silicates are the most common component in interstellar space. Amorphous silicates, crystalline silicates and alumina (Al2 O3 ) are also common in AGB stars (Molster et al., 2010), therefore, we can compare the different dust species in the different astronomical environments. Fig.3.3, Fig.3.4, and Fig.3.5 show the laboratory data used in this study. To fit the broad silicate resonances at 10 and 18 µm, we use Fe-rich (MgFeSiO4 ) and Fe-poor amorphous olivine (Mg2 SiO4 ), which allows us to constrain the Fe-content of the amorphous silicates. Forsterite (Mg2 SiO4 ) is the crystalline form of olivine, and we use its abundance to constrain the crystallinity. Alumina (Al2 O3 ) is included to explain the broadening of the 36.

(43) Figure 3.3: The spectrum show the characteristic of interstellar PAHs top, periclase (MgO) middle, and forsterite bottom. 10 µm feature, and MgO adds opacity to the 18 µm feature. Polycyclic aromatic hydrocarbons (PAHs) are included, even though it is not a dust component but behaves more like a molecule, because AGN often show PAH features between 5-15µm. We used the six species defined in section 3.3 to fit the five sources described in section 2.1. Results are displayed in Fig.4.2 ∼ Fig.4.7. Two plots at the left-hand side show the spectrum fit by the CDE model, and the right-hand side ones by Mie model; the top plots show the spectrum with continuum, and the bottom plots show the continuum divided spectrum. The black line represents the Spitzer data, and other. 37.

(44) Figure 3.4: Three kinds of spectrum for species of dust, Fe-poor amorphous olivine (top) , alumina (Al2 O3 ) (middle), and Fe-rich amorphous olivine (bottom), are assumed by Mie theory. The Q-value of these sources with the continuum background is showed solid line, otherwise, the continuum-divided spectrum are dashed line. lines represent different species of dust, including Fe-rich amorphous olivine (green), Forsterite (blue), Corundum (orange), Periclase (magenta), PAHs (cyan), and Fe-poor amorphous olivine (violet). The red line shows the global fitting result. The presence of silicates in the AGNs is indicated by the absorption lines at 9.7µm and 18µm resulting of lattice resonance (Imanishi and Ueno, 2000). Recent studies concluded that silicates also generate an emission spectrum in some quasars and AGNs (Hao et al., 2005; Shi et al., 2006). The 9.7 µm 38.

(45) Figure 3.5: Three kinds of spectrum for species of dust, Fe-poor amorphous olivine (top) , alumina (Al2 O3 ) (middle), and Fe-rich amorphous olivine (bottom), are assumed by Continuous Distribution of Ellipsoids(CDE). The Qvalue of these sources with the continuum background is showed solid line, otherwise, the dashed line are continuum-divided spectrum. emission feature comes from the Si-O stretch resonance and the 18µm band is due to Si-O-Si stretch resonance. The amorphous silicates are the most abundant species in the interstellar space, hence observing the variation in the ratio of amorphous silicates with crystalline silicates provide important information on the observed source. In this study, we used two kinds of amorphous olivine, one Fe-rich: MgFeSiO4 , and one Fe-poor: Mg2 SiO4 , and one kind of crystalline olivine: forsterite. Therefore we could extract the value of crystalline fraction via the ratio between the fractions of crystalline 39.

(46) olivine (forsterite) and amorphous olivine (MgFeSiO4 and Mg2 SiO4 ). (See the table.4.3) Crystalline silicates are dominated by the Fe-poor form, therefore, the forsterite alone can probe crystalline silicates, and the ratio between forsterite and amorphous silicate provides information on the density of the environment where dust was formed. Indeed crystalline silicates form in higher densities and at lower temperature than amorphous silicate (Molster et al., 2002). The alumina and periclase produce strong and broad emission lines above 10µm (check with Fig.3.4 and Fig.3.5). In fact, these two species can fill the opacity between the two resonance peaks of silicates, 9.7µm and 18µm (J¨ager et al., 2003). Periclase has much lower condensing temperature (around 50K) than the Mg-rich olivine (about 1100K), therefore, if the temperature drops slowly, the Mg-rich olivine will be formed first and use up all the Mg, preventing the formation of MgO. The ratio of periclase is a good tracer of how fast the cooling happens (Ferrarotti and Gail, 2003).. 3.4. Fitting method. We used the MPFIT package in Interactive Data Language (IDL) to find the best-fit values of the above parameters in Eq.3.2. MPFIT uses a non-linear least squares fitting and iterative calculation method to get the minimal χ2 (Eq.3.9) and find the best fitting result (Markwardt, 2009). χ2 is defined by :. 40.

(47) 2. 4χ =. N X Fi − Aλαi (1 +. P. j. j. aj qji ). dFi2. i. The sum. P. 1 · . N. (3.9). aj qji refers to the sum of the species we use, fmod,i , and Fi. represents the spectrum of Spitzer-IRS; the N is degrees of freedom. MPFIT can be used to find the best-fit model parameters to fit any 1dimensional data with errors or uncertainties, including our study. We describe here the successive steps used in our fitting pro-. cedure: 1. Read the raw data which is from Spitzer-IRS 2. Shift the module data to match at band edge There is a jump at 14µm between SL and LL modules and it is so hard to fit the two channels. Therefore, we need to shift the data to match in the flux at the joining point before fitting. Source PG 1351+640 PG 0050+124 PG 1211+143 PG 0043+039 PG 2112+059. SL2 1 1 1 1 1. SL1 1 1 1 1 1. LL2 1 0.9 0.9 1 1. LL1 1 0.9 0.9 1 1. Table 3.1: This table shows the factors of every sources and modules for shifting in this study.. 3. Remove overlapping wavelength range:. 41.

(48) We set the boundaries at the edge of every modules redundant data. The boundaries are set at 7.5 µm, 14 µm, and 20.6µm. (see section 2.2) 4. Convert to rest wavelength: Using the redshift equation (see Eq.2.2), we convert the observed data to rest frame data. 5. Fitting (a) Read the Q-values from laboratory measurements: There are many different species for which Q value is measured in laboratory: Alumina(Al2 O3 ), Periclase(MgO), Fe-rich amorphous olivine, Fe-poor amorphous olivine, and Forsterite(Mg2 SiO4 ) (MarkwickKemper et al., 2007). We use both Q-values of spherical(mie) and non-spherical(CDE) in three of those components: Fe-rich amorphous olivine, Fe-poor amorphous olivine, and Alumina(Al2 O3 ), then we separate the Q-values of spherical and non-spherical and use all compositions for fitting; Polycyclic aromatic hydrocarbons (PAHs) are included. For PAHs, Forsteite, and Periclase(MgO),we use opacities directly because they do not have grain size or shape is assumed. The power-law continuum part of spectrum will be determined by our model, therefore, we need to use the continuum-divided Q-value for fitting with the quasars we select for analyzing. (b) Call our model function 42.

(49) Using the model we described at Sec.3.1 (Eq.3.2) (c) Choose a starting guess: We use the MPFITFUN function in the MPFIT package. The MPFITFUN need to set seed values as starting point. The starting guess value allows MPFITFUN to find a local minimum for chisquare (χ2 ) value. If we set a suitable starting value, we will find not only a local but also a global minimum for χ2 , and this minimal χ2 value will not have any change when we alter the starting guess. In this study, we set the starting values for all parameters to 1. for all the sources: PG 0043+039, PG 1351+640, PG 0050+124, PG 1211+143, and PG 2112+059. (d) Call MPFIT then plot the result The built-in function MPFITFUN solves for the best fit parameters and then optimizes for the 1-D curve, therefore, we used this function to fit and find the minimal χ2 .. 43.

(50) Chapter 4 Results and Discussion 4.1. PG 2112+059 with 2 different methods. Markwick-Kemper et al. (2007) selected PG 2112+059, therefore, we included it in our sample as consistency check for our new model. The Table.4.2 shows the results induced by the two methods. Although the χ2 value (about 5) in our method is larger than the value in the Markwick-Kemper et al. (2007), the χ2 values of other sources we selected are much small, therefore, we still can use this method to examine the difference between the two methods for PG 2112+059. The two different methods imply the similar power-law index, α. MarkwickKemper et al. (2007), used the photometric and spectroscopic data, and got α values of -0.617±0.004 and -0.674± 0.003 respectively, for the model Fν = Fν,0 ν α . Using our model, we get α=0.555±0.003 and α=0.559±0.004 (see the Table.4.1), for CDE and Mie models respectively . (As our model P Fmod [λi ] = (Aλαi )(1 + j aj qji ), is in the λ frame, the α values are positive in 44.

(51) our case). When comparing, our α values are slightly smaller than the values in Markwick-Kemper et al. (2007), our continuum model fitting being more slack and Markwick-Kemper et al.’s being more steep (Fig. 4.1). Fitting method determines the continuum model and an optically thin dust components for the different species simultaneously, then our continuum model are determined by dust specious we used, therefore, even it the results of the continuum model is not very close to the spectrum, they also provide a new direction for figuring out the contribution from various dust species in these sources. Source α-value PG 1351+640 ...CDE 1.297±0.004 ...Mie 1.408±0.003 PG 0050+124 ...CDE 1.016±0.002 ...Mie 1.054±0.002 PG 1211+143 ...CDE 0.596±0.003 ...Mie 0.642±0.003 PG 0043+039 ...CDE 0.364±0.023 ...Mie 0.386±0.022 PG 2112+059 ...CDE 0.555±0.003 ...Mie 0.559±0.004 PG 2112+059 (2007) photometric 0.674±0.003 spectroscopic 0.617 ±0.004. Table 4.1: This table shows the results for power-law index, α, of all the quasars we used in this study. The lowermost shows the power-law index in the results of Markwick-Kemper et al. (2007).. Fig.4.1 and Fig.4.2 show the spectrum fitted with our different models compared with Markwick-Kemper et al. (2007). Table. 4.2 summarizes the fraction of every dust species in PG 2112+059 for our two models and 45.

(52) Markwick-Kemper et al.’s two models. The fits of the spectrum with the different methods yield similar; results with large amount of amorphous olivine and alumina, as well as a small quantity of MgO. We determine the composition of PG 2112+059 to be 51.5%±3.5% amorphous olivine, 39.9%±1.27% alumina, and 8.3%±4.4% periclase (those values are average of CDE model and Mie); When Markwick-Kemper et al. (2007) measured values of alumina and periclase of 38 ± 3% alumina and 5.9 ± 2.6%, and 56.55 ± 1.4% amorphous olivine. Although slightly different the values for amorphous olivine are consistent at the 2σ level.. 4.2. Common dust features of AGNs. To investigate the realism of Elvis et al. (2002) model, we want to compare the difference between quasars and interstellar medium in normal galaxies. Dust in galaxies is thought to be primordial produced by AGB stars with as common species amorphous silicates, crystalline silicates, alumina, and periclase. Although the amorphous silicate are definitely common species in the interstellar medium, we do not have powerful evidence that crystalline silicates really exist in the ISM. In the winds of evolved stars, the abundance of crystalline silicates is of maximum 10% (Molster et al., 2010), even as low as an upper limit of 2% for the silicate mass to be crystalline in proto-stars (Demyk et al., 1999). Furthermore, Kemper et al. (2004) demonstrated the fraction of crystalline silicate is still small in the ISM of the Galaxy, with a fraction less than 2.2 %. 46.

(53) 47. PG 2112+059 PG 2112+059(2007) Average CDE Mie Continuum 1 Continuum 2 54.0%±3.6% 49.0%±4.8% 51.5%±3.5% 54±1 59±1 40.8%±0.9% 39.0%±0.9% 39.9%±1.27% 37±2 39±2 5.2%±0.5% 11.4%±3.8% 8.3%±4.4% 9±2 2.7±1.7 0% 1.2%±0.2% 0.6%±0.8% 4±2 6±2 35.5%±2.9% 28.3%±3.8% 31.9%±5.1% 70±44 49±40 18.5%±2.2% 20.7%±2.9% 19.6%±1.6% 30±44 51±40. 56.5±1.4 38±3 5.9±2.6 5±3 ... .... Average. Table 4.2: The left three columns are our work in this study, and the right three columns are work of MarkwickKemper et al. (2007). We compare the result of them.. Silcate/total Alumina/total MgO/total Crystalline/silicate Mg2 SiO/amorphous silicates MgFeSiO4 /amorphous silicates. Fraction.

(54) Table.4.3 presents the results for the composition of every sources. For those quasars, the dominant species is in all cases amorphous olivine, and on the contrary, forsterite (crystalline olivine) is very rare. It means the densities in those environment are much lower, because the crystalline silicates need higher density than amorphous ones. Although for PG 2112+059 Markwick-Kemper et al. (2007) inferred a 5±3% crystalline olivine hance a ratio superior to normal galaxy (2.2 %; Kemper et al. (2005)), our results imply a lower proportion of crystalline olivine as similar level than in normal galaxies. In particular the CDE model implies 0% of crystalline olivine. Based on Dust nucleation model, the alumina (Al2 O3 ) is believed to be one of the first condensates for oxygen-rich outflows via thermodynamic equilibrium (Gail and Sedlmayr, 1998; Molster et al., 2010). The condensation temperature of alumina is around 1500 K, and the alumina always will be covered by the layers of silicate when the temperature decrease too rapidly. Although the alumina (Al2 O3 ) is also a common species in the winds of AGB stars, AGB stars just contain momentous fractions in the low mass loss rate case (Blommaert et al., 2006). However, no observation data backup the existence of alumina in the diffuse interstellar medium (Molster et al., 2010). In fact, our result involve a high ratio of alumina, particularly for PG 2112+059, PG 0043+039, and PG 1211+143. The coexistence of alumina and olivine in those quasars imply the environments for forming dust is inhomogeneous for density structure.. 48.

(55) 4.3. Comparison between sources. Elvis et al. (2002) proposed the quasars winds could be the site of dust production, especially, in the outer radius strong broad emission lines (BEL). Because some outflowing winds from the center of quasars can bring the system around the BEL regions out of pressure balance, the clouds will expand and the temperature will cool down as the system expand, fostering good conditions for formation of dust. Analyzing the components of these BALs quasars can allow us to investigate the environment in such strong winds. We do not find any reason why the physical processes within the galaxy may change with redshifts, therefore, we analyze these quasars with different redshift. Our targets present both strong outflows ar weak outflows, enabling us to investigate such scenario. This discussion with be separated in two parts: BALs (PG 2112+059 and PG 0043+039), and non-BAL or mini-BAL (PG 1211+143, PG 0050+124 and PG 1351+640).. 4.3.1. Board Absorption Line (BALs). We select two BAL quasars at similar redshift, PG 2112+095 and PG 0043+039, as they are presenting strong winds. PG 2112+095 and PG 0043+039 show similar content of Amorphous olivine, 54.0%±3.6% and 54.9%±5.0% in CDE and 49.0%±4.8% and 44.8%±5.0% in Mie respectively, but of different nature: PG 0043+039 contain only Fe-rich amorphous olivine (MgFeSiO4 ); while PG 2112+095 contain large amount of Fe-poor amorphous olivine (31.9%±5.1%). In substance, the two sources, PG 2112+095 and PG 0043+039, 49.

(56) are composed of almost 50% (51.5%±3.5% and 49.9%±7.1% averagely respectively) amorphous olivine (MgFeSiO4 and Mg2 SiO4 ) and little crystalline olivine (forsterite) but in different amount. Therefore, their crystalline fraction are different: small in the case of PG 2112 (0.6%±0.8%) , and big in the case of PG 0043 (2.2%±1.1%). (the results at Fig.4.3, Fig.4.4 and Table.4.3) As a whole, the crystalline olivine in the two BALs quasars is as rare as in normal galaxies (for instance, it is in the Milk Way) (small than 2.2%; Kemper et al. (2005)), but on the contrary, the amorphous olivine amount in these quasars is much smaller than in normal interstellar medium. Alumina is much more abundant than in the AGB stars, and implying the environment of the BAL quasar with strong outflows might provide a highly inhomogeneous density structure for formation of dust. The periclase amount in PG 0043+039 is larger than in the PG 2112+059, which indicates that the variation in temperature of PG 0043+039 is more rapid than in PG 2112+059, because the stabile limits of periclase (∼50 K) is below the Mg-rich olivines (∼1100K). Therefore, if the temperature cool down slowly, the Mg2 SiO4 will be formed first and there is not enough Mg to form much MgO. In this case, the distribution of component in the results of these two quasars show the temperature should cool down more rapidly in the PG 0043+039.. 4.3.2. non-BAL and mini-BAL. PG 1211+143 is a non-BAL quasar with very strong outflow. According to Table.4.3, we find that the components in PG 1211+143 are very different from those in the others. The contribution of components (ampor-. 50.

(57) phous olivine, alumina, and periclase) is more uniform, with 38.4%±4.3% amorphous olivine, 39.2%±0.4% alumina and 20.3%±3.2% periclase. Fig.4.7 and Table.4.3 show the ratio of periclase is higher than in other sources, even though the amorphous olivine is still dominating. On the contrary the amount of the crystalline olivine, forsterite, is still small, but given the small amount of amorphous olivine, the crystalline fraction is higher. The amorphous olivine is completely dominated by Fe-rich amorphous olivine (MgFeSiO4 ). PG 1351+640 is a mini-BAL quasar and PG 0050+124 is a non-BAL quasar, but they both present weak or no outflows. They have similar redshift, 0.0881 and 0.0808 respectively, but as started already there is no physical reason that the redshift has an influence on dust formation. Amorphous olivine still represents the dominant species in these cases, and is even more dominant than the case of BALs. The proportion of amorphous olivine run up to 78.8%±9.5% and 82.8%±0.4% respectively. However, amorphous olivine is more Fe-poor for PG 0050+124, and Fe-rich for PG 1351+640. The amount of forsterite in these quasars with weak outflow is far lower than in the BALs; With ∼ 0.9% for PG 1351+640, and even absent totally for PG 0050+124. As a consequence, the crystalline fraction in the quasar with weak outflow is also lower than that in BALs. (See Fig.4.5, Fig.4.6 and Table.4.3) These three quasars are all non-BALs or mini-BAL, but the amount of alumina in PG 1211+143 with very strong outflow is much higher than other two quasars with weak outflow. Quasars with strong outflow produce warm winds, causing fluctuations in the temperature and density. This may provide a more inhomogeneous density environment for forming alumina. This result 51.

(58) is in agreement with the outflow model, proposed by Elvis et al. (2002). In addition, the high ratio of periclase implies that the temperature in the winds of PG 1211+143 drops rapidly when the BEL clouds expanded after the divergence of the warm outflowing winds taking this system out of pressure balance (Elvis et al., 2002). In the case of the other two quasars with weak outflow, the amount of alumina and periclase is far lower, implying that the weak outflow may not provide suitable conditions to form such species. In fact, the amorphous olivine in PG 1351+640 and PG 0050+124 is just as dominant as in normal galaxies. From our limited sample we can conclude that ratio of highly refractory alumina and less refractory periclase increase with stronger outflows, which provide therefore a good environment for both species to form and coexist. Consequently, our results could support the conclusion in Elvis et al. (2002), and we can say the winds of quasars are necessary for QSOs to form dust, but on the contrary, if the quasar without winds, the dust around quasar may come from AGB stars.. 52.

(59) CDE. MIE. Figure 4.1: The two plots on left hand side are spectrum of our fitting results, and the left hand side one is the result in Markwick-Kemper et al. (2007). The power-law index in the Markwick-Kemper et al. (2007) are -0.617±0.004 and -0.674± 0.003 (dashed line: the continuum curve is determined by photometric data, the diamonds. solid line: the continuum curve is determined by spectroscopic data.) The values of power-law index in our results are more slack, and they are 0.555 and 0.559 in the CDE model (left-up) and Mie(left-bottom) individually.. 53.

(60) Figure 4.2: Two kinds of method for determining the components of PG 2112+059. Left: We determine the components of PG 2112+095 using our model, including CDE model (top) and Mie (bottom). The different species show in different colors, for instance, Fe-rich amorphous olivine (green), Forsterite (blue), Alumina (orange), Periclase (magenta), PAHs (cyan), and Fe-poor amorphous olivine (violet). Right: Determing the components of PG 2112+095 by the method in Markwick-Kemper et al. (2007) (including two kinds of continuum model: photometric and spectroscopic).. 54.

(61) Figure 4.3: Best 5-25 µm fits for BAL quasar, PG 2112+059. The top plots show spectrum with power-law continuum model, and the bottom ones are continuum-divided spectrum. Left: The two plots at left hand side show the compositions determined using CDE (non-spherical grain) model, and we can see the dominant dust components here are corundum (Al2 O3 ; orange line) and Fe-poor amorphous olivine (Mg2 SiO4 , violet), Fe-rich amorphous olivine (MgFeSiO4 ; green). Right: The two plots at right hand side represent the component determined using Mie (spherical grain) modle. In this class corundum (Al2 O3 ) are still in the great majority, and the percentage of Fe-poor amorphous olivine (Mg2 SiO4 ) and Fe-rich amorphous olivine (MgFeSiO4 ) are a little similar. Comparing the results of spherical and non-spherical, the periclase (MgO; magenta) is much more abundant in the spherical case then in the non-spherical case.. 55.

(62) Figure 4.4: Best 5-25 µm fits for BAL quasar, PG 0043+039. The top plots show spectrum with power-law continuum model, and the bottom ones are continuum-divided spectrum. Left: The two plots at left hand side show the compositions are determined using CDE model (non-spherical grain), and dominate here are Fe-rich amorphous olivine (MgFeSiO4 ; green) and periclase (MgO, magenta), followed by corundum (Al2 O3 ; orange line). Right: the two plots at right hand side represent the component determined using Mie model (spherical grain). The results here are similar with the CDE (non-spherical) results: Fe-rich amorphous olivine (MgFeSiO4 ; green) and periclase (MgO, magenta) dominate the composition, dust components with the exception of forsterite being more abundant in the spherical grain case.. 56.

(63) Figure 4.5: Best 5-25 µm fits for a radio-quiet quasar, PG 1351+640. The top plots show spectrum with power-law continuum model, and the bottom ones are continuum-divided spectrum. Left: The two plots at left hand side show the compositions determined using CDE model (non-spherical grain), and there the dominant component is Fe-rich amorphous olivine (MgFeSiO4 ; green), followed by alumina (Al2 O3 ; orange line). Right: the two plots at right hand side represent the component determined using the Mie model (spherical grain). The result here are very similar with the CDE (nonspherical) results: Fe-rich amorphous olivine (MgFeSiO4 ; green) occupy most percentage of compositions.. 57.

(64) Figure 4.6: Best 5-25 µm fits for a radio-quiet quasar, PG 0050+124. The top plots show spectrum with power-law continuum model, and the bottom ones are continuum-divided spectrum. Left: The two plots at left hand side show the compositions are determined using CDE model (non-spherical grain). In this case, the dominant component is Fe-poor amorphous olivine (Mg2 SiO4 , violet). With in addition, some corundum (Al2 O3 ; orange line). Right: the two plots at right hand side represent the component determined using the Mie model (spherical grain).Results from CDE (non-spherical) and Mie(spherical) are very similar in this case: the Fe-poor amorphous olivine (Mg2 SiO4 ; violet) in spherical is near the percentage in the non-spherical case, and the Fe-rich amorphous olivine (MgFeSiO4 ; green) are none in the two kinds of cases.. 58.

(65) Figure 4.7: Best 5-25 µm fits for a radio-loud quasar, PG 1211+143. The top plots show spectrum with power-law continuum model, and the bottom ones are continuum-divided spectrum. Left: The two plots at left hand side show the compositions determined using Mie model (non-spherical grain). Right: the two plots at right hand side represent the component are determined using the modle (spherical grain). In this case, the non-spherical and spherical models bring very similar results: Fe-rich amorphous olivine (MgFeSiO4 ; green) and alumina (Al2 O3 ; orange line) dominate the spectrum, while Periclase (MgO) is less abundant.. 59.

(66) 60 .... 56.5 ± 1.4 %. PG 1351+640 ...CDE ...Mie Average PG 0050+124 ...CDE ...Mie Average PG 1211+143 ...CDE ...Mie Average PG 0043+039 ...CDE ...Mie Average PG 2112+059 ...CDE ...Mie Average. PG 2112+059 (2007). .... Fe-poor 14.8%±3.5% 4.8%±4.4% 9.8%±7.0% 83.1%±1.1% 82.5%±1.1% 82.8%±0.4% 0% 0% 0% 0% 0% 0% 35.5%±2.9% 28.3%±3.8% 31.9%±5.1% .... 0.38%±0.06% 1.43%±0.05% 0.91%±0.75% 0% 0% 0% 1.7%±0.1% 2.8%±0.1% 2.25%±0.78 0.2%±0.4% 1.8%±0.4% 1.0%±1.1% 0% 0.6%±0.06% 0.3%±0.4%. Forsterite. 5 ± 3%. 0.53%±0.09% 1.65%±7.11% 1.1%±0.8% 0% 0% 0% 3.9%±0.3% 7.4%±0.4% 5.7%±2.5% 0.4%±0.7% 3.9%±0.5% 2.2%±2.5% 0% 1.2%±0.2% 0.6%±0.8%. Crystalline fraction. 38 ± 3%. 13.9%±0.5% 11.09%±0.55% 12.5%±2.0% 16.9%±0.7% 17.5%±0.6% 17.2%±0.4% 38.9%±1.1% 39.4%±1.1% 39.2%±0.4 23.6%±4.1% 25.2%±4.2% 24.4%±1.1% 40.8%±0.9% 39.0%±0.9% 39.9%±1.27%. Al2 O3. Table 4.3: The best χ2 fitting results to every source. The first row shows the dust species and following rows are mass fractions for every dust species in the different source. The first column means different source. All the mass fractions of every source are observed within the ∆χ2 less than a factor 2.3, excepting for PG2112.. Fe-rich 67.2%±2.7% 80.7%±3.2% 74.0%±9.5% 0% 0% 0% 41.4%±1.7% 35.3%±1.5% 38.4%±4.3% 54.9%±5.0% 44.8%±5.0% 49.9%±7.1% 18.5%±2.2% 20.7%±2.9% 19.6%±1.6%. Amorphous olivine. Total of amorphous olivine 72.0%±4.4% 85.5%±5.4% 78.8%±9.5% 83.1%±1.1% 82.5%±1.1% 82.8%±0.4% 41.4%±1.7% 35.3%±1.5% 38.4%±4.3% 54.9%±5.0% 44.8%±5.0% 49.9%±7.1% 54.0%±3.6% 49.0%±4.8% 51.5%±3.5%. Source. 5.9 ± 2.6%. 3.8%±0.2% 2.0%±0.2% 2.9%±1.3% 0% 0% 0% 18.0%±0.5% 22.5%±0.5% 20.3%±3.2% 21.3%±2.5% 28.2%±3.0% 24.8%±4.9% 5.2%±0.5% 11.4%±3.8% 8.3%±4.4%. MgO. 2. 5.24 5.635. 0.60 0.76. 2.82 3.58. 3.58 2.57. 2.21 3.58. ∆χ2.

(67) Chapter 5 Conclusion We determined the dust composition of radio-quiet quasars using SpitzerIRS data: two BAL quasars: PG 2112+059 and PG 0043+039; non-BAL with strong outflow: PG 1211+143; two quasars with weak outflow: PG 1351+640 (mini-BAL) and PG 0050+124 (non-BAL). We developed a new method for deriving the fraction od carious dust species by SED fitting in which the SED consists of a features we adopted power-law continuum and optically thin dust. Six species seen commonly in dust composition, but also providing ideas on the physical processes in action: amorphous olivine, crystalline olivine, forsterite, alumina, periclase, and PAHs, are used in our model fitting, and the spectral features of three of them (amorphous olivine, crystalline olivine, and alumina) are both calculated by CDE model and Mie. First, we have confirmed that our result is consistent with the result obtained by Markwick-Kemper et al. (2007), who used a different method. The quasars with strong outflow winds could produce more various environment, for instance, different degree of temperature as well as density, and. 61.

(68) that can help to form more diverse species around quasars. Especially, there is a good condition for forming MgO, because the temperature can cool down rapidly in the quasars with strong winds. Therefore, in our cases, we find the three quasars with strong outflow winds present higher amount of refractory alumina (Al2 O3 ) and periclase (MgO) and smaller amount of crystalline olivine and forsterite than the other two quasars, with weak outflows. The component in those quasars with strong winds display more various species. On the contrary, in the case of other two quasars with weak outflow winds, amorphous olivine is largely dominant over other species. Their compositions are in fact similar to the composition observed in normal galaxies and AGB stars. Therefore, outflow winds might provide a viable environment for formation of diverse dust species coexisting, such as highly refractory alumina, less refractory periclase, amorphous olivine and small amount of crystalline olivine. Our results also imply that the environment of those quasars with winds present inhomogeneous temperature and density, thus can provide suitable environments for dust formation, conforming to the model developed by Elvis et al. (2002).. 62.

(69) Appendix A Eddington limit To deduce the Eddington limit, for simplicity case, we suppose it is a completely ionized hydrogen gas. In order to avoid disintegration, we assume the outward force of radiation pressure is equilibrated with the inward force of gravity. The flux of outward energy at the some distance r from the center is F = L/4πr2 .Because the momentum can deduce form the energy (p(momentum)=E(energy)/c(lightspeed)), the outward momentum flux (or pressure) also can know. Prad =. F L = . c 4πr2 c. (A.1). L is the luminosity in radiation. Then the Thomson cross section σT can be used for estimating how much the scattering of low-energy light from an electron 8π σT = 3. . e2 4πε0 me c2. 2. 63. = 6.65 × 10−25 cm2. (A.2).

(70) where e is the electron’s charge ; me is the mass of electron (Peterson, 1997). Therefore, the momentum transferred to a single electron is. Prad = σT. L , 4πr2 c. (A.3). and then the momentum per unit time is a force. Frad = σT. L rˆ. 4πr2 c. (A.4). According to above supposing, we assume the outward force of radiation pressure is equilibrated with the inward force of gravity, and then we can suppose that: |Frad | ≤ |Fgrav |.. (A.5). For the ionized hydrogen case, the gravitational force will act on an electronproton pair by a M central mass, and the mass of proton is much larger than mass of electron, hence the gravitational force will become Frad = − GM (mr2p +me ) rˆ ≈ − GMr2mp rˆ The equality |Frad | ≤ |Fgrav | will be GM mp σT L ≤ 2 4πcr r2 4πGcmp L≤ M σT. (A.6) (A.7). ≈ 6.31 × 104 M erg/s. (A.8). ≈ 1.26 × 1038 (M

(71) /M )erg/s. (A.9). M is the mass of the compact object (central mass) and L is the luminosity 64.

(72) in radiation of the central source. The inward gravitational force acting on the ionized hydrogen gas need taking balance or exceed the outward force, otherwise it will disintegrate. As the result, the L is known as the Eddington limit, and we define the Eddington luminosity from above equation. LE =. 4πGcmp M, σe. (A.10). Which can be estimate the maximum luminosity of a source of mass M,can then we could deduce the minimum of the mass (ME )from Eddington luminosity. If the case is AGNs,. ME = 8 × 105 L44 M

(73). (A.11). Where the L44 is the luminosity of central source in unit of 1044 ergss−1 ,which is also characteristic of the luminosity Seyfert galaxy. If the case is typical quasar, the luminosity of the central should in unit of 1046 ergss−1 ,therefore, the central mass should be about ∼ 108 M