Aerosols in Astrophysics stars complete their cycle of existence, there R.E.Stencel, C.A. Jurgenson & is a wholesale return of modified material T.A. Ostrowski-Fukuda back to the interstellar medium, affecting the The Observatories, University of Denver, next generation of star and planet formation. Department of Physics & Astronomy, Denver, Colorado 80208, USA Submitted: Evolved stars are characterized by surface 2002December30, Email: [email protected] temperatures low enough to allow molecules and solid-phase particles to exist in their ABSTRACT: atmospheres. In contrast to the nearly 6000K surface temperature of the Sun, the Solid phase material exists in low density kinetic temperatures of evolved stars may astrophysical environments, from stellar range between 2000K near their “surfaces” atmospheres to interstellar clouds, from to a few hundred K in their circumstellar current times to early phases of the universe. envelopes [CSE] at a few stellar radii We consider the origin, composition and distance. Time-dependent phase transitions evolution of solid phase materials in the from ionized plasma to cold, solid state are particular case of stellar outer atmospheres, observed, spectroscopically, to exist. as representative of many conditions. Also considered are the dynamics and interactions Some terminology may be helpful for the in circumstellar and galactic environments. reader not already acquainted with stellar Finally, new observational prospects for evolution. Solar mass stars are predicted to spectroscopy, polarimetry and leave the hydrogen-fusing “main sequence” interferometry are discussed. after approximately ten billion years, and progress through a rapid series of internal changes during what is known as the red 1.0 INTRODUCTION giant and asymptotic giant branch [RGB, AGB] phases, lasting one percent of the Solid phase material can be found nearly main sequence lifetime. These internal everywhere in space. In the astronomical changes are driven by interior literature, these tiny solids are referred to as thermodynamics, as helium fusion is “dust” and/or “grains” in discussions of the initiated and exhausted as a primary energy most important opacity source at source, past the main sequence. The wavelengths longer than the Lyman limit for important phenomenon during these latter ionizing hydrogen (91 nanometers). These phases is the high rate of evaporation (mass solids play key roles in molecular clouds loss) of the outer layers, which during peak and star formation – enabling cooling, times can rival the internal evolutionary shielding from radiation and energetic timescales. The outflows are estimated to particles, and forming the basis for alter the chemistry of the interstellar planetessimal formation in primitive solar medium and ultimately, the galaxy. The systems. The spectral energy distributions evaporation rate is modulated by the of all of these astronomical objects are quantity and composition of molecular and strongly modified by solid particles, which solid material in the outflow. Hence, Carl Sagan once characterized as having the understanding the role of solids in size and average composition of bacteria. In circumstellar environments is key to this review, we focus on the role that solids correctly predicting the evolutionary paths play in the most evolved stars. As these of stars of diverse initial masses and categories: thermal-driven, radiation-driven, compositions. wave-driven, and shock-driven (pulsation, radiation and dust). An excellent summary 1.1 Aerosols of the relevant physics was presented by Holzer and MacGregor (1985). Most studies treat solids and the surrounding gas as a two fluid system. While this has Two of the most important parameters been useful in advancing the subject, describing a stellar wind are derivable from modern computing power may allow us to observation: the mass loss rate [the amount consider the system in combination as an of mass lost by the star per unit time] and “aerosol”. An aerosol is defined in its the terminal velocity [the velocity of the simplest form as a collection of solid or stellar wind at a large distance from the liquid particles suspended in a gas (Hinds, star]. Stellar mass loss rates vary from 1999). Aerosols are two phase systems, present solar wind levels (10-14 solar masses consisting of the particles and gas in which per year) to AGB and post-AGB “super- they are suspended (cf . Seinfeld and Pandis wind” phases (up to 10-4 solar masses per 1998). This definition admits a range of year), deduced from spectroscopically sizes from molecular clusters (tens of observed density and velocity. The spectral nanometers) to solids (micron-sized), and lines from stellar winds can often be can include neutrals as well as ions – exactly distinguished from the photospheric lines the mix found in circumstellar environments due to their large width or Doppler shift (CSE). Terrestrial atmospheric densities and produced by the outflowing motion of the temperatures range from 10-3 g cm-3 (1019.4 gas in the wind. Wind lines can appear in cm-3, 288K) at the surface, to roughly six emission, in absorption, or as a combination orders of magnitude less density in the upper of the two [P Cygni profile], where stratosphere (~200K). While the observable absorption and emission lines temperatures resemble circumstellar ones, show a Doppler shift due to the outflow, terrestrial densities are orders of magnitude even with relatively small mass loss rates. larger. Micron sized terrestrial aerosols Stellar wind speeds range from 10-20 km float in a molecular nitrogen and oxygen per second for an AGB star, to ~500 km per gas, while circumstellar aerosols form and second for the solar wind near earth, to over move in a mostly hydrogen gas medium. 3,000 km per second (1% of the speed of Therefore, it is timely to review recent light!) for an early type star, such as an O6 progress in astrophysical descriptions of star. Although the velocity structure of the CSE, along with advances in terrestrial wind varies as it progresses out from the atmospheric aerosol chemistry and rarified photosphere, for a typical AGB wind gas dynamical theory and experiment, velocity, the crossing time for a wind applied to an astrophysical context. traveling out 1000 AU from the star would be about 50 years. 1.2 Mass Loss Mechanisms Stellar winds require heat input or It long been known that red giant and momentum input to become transonic, supergiant stars are losing mass at high rates varying from subsonic (typically, a few in the latter stages of life in an observable km/sec) to supersonic with height. The expansion called the stellar wind. Stellar energy per unit mass of an atmosphere, wind theories can be grouped into four main which is gravitationally bound, must be negative. The energy of an outflow, which photosphere, radiation force on the dust escapes the gravitational potential well of grains might drive massive outflows from the star, is positive at large distances. This the star. requires that energy must be added to the gas in order for it to escape. The energy can (c) Wave-driven winds involve the transport be added in the form of heat input or work of a mechanical energy flux through the done by an outward directed force stellar atmosphere and takes into account (momentum input). Four general wind only small amplitude waves for energy driving mechanisms have been discussed in transport, including weak shock waves. detail (cf. Holzer & MacGregor 1985): When a magnetic field is present the small thermally-driven, line-driven, wave-driven amplitude waves will either be compressive and shock-driven winds. or non-compressive, corresponding to acoustic or Alfven waves. However, the (a) The concept of thermally-driven winds compressive waves tend to steepen very was developed in the early 1960's mainly by rapidly into weak shocks and are dissipated Eugene Parker. In this model for the wind, within a few pressure scale-heights of the the only important outward force exerted on base of the atmosphere. Therefore, although the expanding atmosphere would be the acoustic waves can levitate material to a thermal pressure gradient. Thermally driven small height in the atmosphere they are not winds require high (coronal) temperatures, strong enough to lift a massive wind out of and exhibit much larger radiative fluxes, the stars' gravitational well. With sufficient especially in the ultraviolet and extreme- magnetic field strength relative to the ultraviolet. Unlike the sun, red giant and thermal energy, Alfven waves show promise supergiant stars do not have hot coronae, for driving outflows, particularly in the solar effectively ruling out this mechanism. case. Questions related to energy dissipation in these waves remain (b) For line-radiation driven winds, the unresolved. electromagnetic radiation field of a star provides a reservoir of momentum that can (d) Shock-driven winds (combination of be used to drive a stellar wind. This pulsational ‘levitation’ and outward mechanism works when there is a coupling radiation forces on dust) arise from the of the radiation field to the gas/dust in the coupling of the stellar radiation field to a atmosphere. Radiation-driven winds work sufficiently dusty atmosphere and CSE well for hot stars when the atmosphere already expanded by the effect of pulsations. exhibits many strong UV resonance lines Variable stars, such as the ones of interest in and the radiative flux is substantial in the this study, exhibit somewhat regular ultraviolet. The coupling comes from the pulsations that produce large-amplitude stellar radiation and the atmospheric opacity. shock waves propagating outward through Stellar radiation for cool stars peaks in the the photosphere and into the CSE. This red or near-infrared, and the strong allows for the stellar material to be lifted to resonance lines of the atoms occur in the sufficient height where dust will form and visible and ultraviolet. Therefore, this where the escape velocity is substantially mechanism acting alone can be ruled out for lowered. Often these winds are termed red giant and supergiant stars. Although, for “dust-driven”, but one should retain an open the most luminous and coolest AGB stars, mind to the idea of “wind-driven” dust in where the dust forms close to the these systems. form the initial molecules, and conditions 1.3 Stellar Winds in Evolved Stars supportive of a dust-driven stellar wind will not be generated. The dynamics of this are Among AGB stars, there is a subset known the subject of a later section of this review. as “Mira” variables, named after the prototype, omicon (Mira) Ceti. These are Dust condensation begins to occur at known to execute periodic, several temperatures as low as 1500K for aluminum magnitude variation due to large-amplitude oxides, and less than 1000K for silicon radial pulsations, on timescales of 100 to monoxide in oxygen-rich stellar 1000 days. Whether the usual oscillation atmospheres and amorphous carbon dust in mode is the fundamental or first overtone carbon-rich stellar atmospheres. Whether a remains unresolved, despite strong opinions stellar atmosphere is oxygen or carbon rich expressed on the subject. Dynamical is a strong function of stellar evolutionary structure models by Bowen (1988) found state. The interior production and dredge up that fundamental-mode pulsations better of carbon contaminates the atmosphere of matched actual observations, because the stars where oxygen may have been the models driven at their overtone periods slightly more abundant, locking up the free showed to have different properties, oxygen in carbon monoxide molecules. including shocks of smaller amplitude These temperatures are not significantly occurring at smaller radius and larger below the photospheric levels for cool stars, density than shocks in fundamental-mode. and the formation of molecules and post shock cooling offer means to reach Dust driven winds only occur in the very condensation temperatures while still near restricted portion of the HR diagram that the denser stellar photosphere (cf. contains the cool, high luminosity red Muchmore, Nuth and Stencel 1987). supergiant and AGB stars (including the Mira type). The effective (surface) The temperature of newly formed molecular temperatures of these stars range from 2000 clusters is mainly determined by the to 3000 K and the luminosities are 104 times equilibrium between the heating due to solar for the AGB and 105 times solar for the absorption of stellar photons, and the red supergiant stars. The low CSE cooling by thermal emission. Therefore, the temperatures allow for formation and particulate temperature depends on the growth of molecules and clusters of opacity of the cluster material, and on the molecules to occur. The temperature and ambient radiation field, and will roughly radiation field determine the altitude above decrease with distance from the star if the the photosphere where molecular formation CSE is optically thin. The innermost radius or “condensation” of particles begins, in where particulates can survive ranges from competition with the decreasing density of 1.1 to a few stellar radii for stars with the gas with increasing distance from the effective temperatures in the range 2200 to star. It is the density of the condensation 3000 K. Interestingly, highly variable SiO radius that determines how many solid phase masers are observed using radio astronomy materials form and whether there is methods, precisely in this domain. Once sufficient momentum coupling between the formed, these particles can be quickly gas and particles to sustain an outflow. If accelerated outward due to their large the value of the density is too low there will opacity (see following sections). In dust be too few collisions between gas atoms to driven winds, the sonic point and critical point both occur at about this condensation utility for diagnostic study of physical radius. Therefore, the mass loss rate is conditions using the following spectroscopic determined by the density at the features (cf. Spoon et al. 2002). Cold condensation radius. molecular gas component – CO and SiO bandheads at 2.29 and 4.67, and 4.0 and 8.0 microns respectively (cf. Winters et al. 1.4 Scope of this Review 2000; Arlinger 2000), plus rotational lines of H2 between 2 and 28 microns (cf. van den Whereas the topic is diverse and touches Ancker et al. 2000), 3.0 and 5.5-8 micron upon many fields in astrophysics, from water ice features, and the 4.26 micron cometary origins to galactic evolution, we carbon dioxide ice feature. The ice feature limit consideration to optically thin, O-rich profiles are highly temperature sensitive and dust shells, in the interest of providing a when irradiated exhibit signs of ice mantle focus. The related issues of carbon-rich processing in the “XCN ice feature at 4.62 circumstellar chemistry are discussed microns and the 4.67 micron CO ice feature elsewhere. The purpose of this review is to as well. The Poly-cyclic Aromatic provide access to the expanding discussion Hydrocarbon features [PAH] appearing at of silicate chemistry among evolved stars 3.3, 6.2, 7.7, 8.6, 11.3 and 12.8 microns, are during the current explosion of data taking highly sensitive to the irradiation from capabilities in the infrared and radio parts of nearby hot star sources (as in photo- the spectrum. The interested reader can dissociation regions, PDRs). But, it is the follow the numerous references provided in stretch mode of the Si-O bond that gives rise each of the papers we cite. In section Two, to the strongest feature, centered at 9.7 we consider the origin, composition and microns, but appearing over the 8 to 12 evolution of solid phase materials, and micron interval depending on degree of laboratory analogues. Section Three crystallinity, etc. Related silicate features discusses the dynamics and interactions in have been detected at 9.7 and 18.5 micron circumstellar and galactic environments. (amorphous silicates) and 11.2, 27.5, 33.5, Section Four describes some of the 35.8 and longer wavelengths (crystalline observational prospects for spectroscopy, silicates). In carbon stars, graphitic and polarimetry and interferometry. crystalline forms are indicated.

Grain composition: In a series of papers, 2. CIRCUMSTELLAR AEROSOLS Molster et al. (2002) report Infrared Space Observatory (ISO) short and long 2.1 Recent Infrared Spectroscopy wavelength spectrometer (SWS, LWS) observations of crystalline silicate dust in a Overview: In addition to ground-based mid- diverse sample of 17 evolved stars, infrared spectroscopy, hampered by including AGB, post-AGB, PN and RSG terrestrial atmospheric absorptions in the 1 and hot star prototypes, which contrast to 20 micron region, two important surveys quasi-spherical outflow versus disk-like from space have been completed: the circumstellar geometries. They defined Infrared Astronomical Satellite (IRAS Low seven wavelength regions with silicate Resolution Spectrometer, Olnon et al. 1983) “complexes” of subfeatures, near 10, 18, 23, and the Infrared Space Explorer (ISO, 28, 33, 40 and 60 microns. Most of the 1995). Key among the findings has been the bands were identified with forsterite (Mg2SiO4; the Mg-rich crystalline olivine) scenarios involves the question of the low and enstatite (MgSiO3; the Mg-rich incidence of elemental iron involvement crystalline pyroxene). For a general with silicates, such as fayalite (Fe2SiO4 ): discussion of the mineralogy, see, e.g., kinetic freeze-out; preferential condensation Jaeger et al. 1998, and Speck et al. (2000). as metallic iron, or, the preponderance of About 20 percent of the bands lacked amorphous forms of silicate in circumstellar identification. The broad feature in the outflows. The latter scenario would be 10 micron complex is thought to originate consistent with preferential loss of tiny from amorphous silicates, with sharper amorphous grains, as predicted by features from the crystalline forms, and dynamical models (see next section), and indicative of larger particle sizes or observed in the cases of silicate-rich shells composition differences. The outflow around some carbon stars. Interestingly, sources have a lower abundance of even the planet Mercury shows less crystalline silicates, but a higher abundance spectroscopic FeO than other terrestrial of crystalline H2O-ices (43 and 62 micron planets, suggesting the difficulty of binding features) which avoid UV processing into Fe into Mg-silicates. amorphous forms due to clumping. On average, they concluded, crystalline silicates Grain sizes: The consistent use of Mie are colder (<100K) than amorphous silicates theory for lack of a better approximation ( >100-300K). insures that well-developed dust shells will be evaluated to have grain sizes in the Tielens, et al. (1997) provide a discussion of domain of the wavelength of observation. these features in the context of mineralogical Molster et al. (1999) assume spherical condensation sequences. The narrowness of grains, an interstellar (MRN) power-law spectral peaks atop a broad feature ascribed size distribution and use Mie theory to to amorphous silicates, implies the existence calculate the optical properties of the grain of crystalline carriers (Koike et al. 1993). population. This approach allows for the Tielens et al. suggest initial formation of calculation of emission for arbitrarily larger corundum (Al2O3 ) at temperatues below grains, which are abundant in the shell of the 1700K, which reacts with SiO, Ca and Mg high luminosity source, AFGL 4106. They to form spinel (Mg Al2O4 ) and eventually conclude grain sizes of 0.4 to 6 microns are diopsides (CaMg Si2O6 ). Most of the the most common, with mass fractions of silicates nucleate and condense as forsterites 65% amorphous olivines, 15% corundum, (Mg2SiO4 ), the Mg-rich end member of the 10% enstatite, 5% each water ice and olivine family. Then, excess SiO converts forsterite and essentially no FeO. Sogawa & forsterite into enstaties (pyroxene, MgSiO3 Kozasa (1999) report that a homogeneous ), with the enstatite/forsterite ratio increasing condensation of corundum grains, the with decreasing temperature. The presence accretion of silicate, starting slightly inside of individual species in any given stellar the so-called sonic point, results in spectrum is indicative of “freeze out” heterogeneous grains consisting of a whereby the condensation sequence is corundum core and silicate mantle. In their interrupted by lack of density and model, homogenous silicate grains begin in temperature to sustain same, and thereby the 1.5-6 nanometer (nm) size range, and help interpret assorted spectral sequences heterogeneous grains grow from 0.15 to 0.4 (cf. Stencel et al. 1990 and others). An microns in size. Fabian et al. (2001) state important aspect of these condensation that observed ISO band positions (peaks at 9.4 and 11.4 microns) are better reproduced magnetic field strength in this key region. by a wide distribution of ellipsoidal grain Kemball and Diamond (1997) deduced a shapes, whereas the 9.8 micron peak is line of sight field strength in the Mira TX indicative of spherical grains. Difference in Cam of 5-10 gauss. Vlemmings et al. 2002 spectra may then reflect initial formation as have compiled this and related magnetic amorphous spheroids and subsequent field strength determinations for related stars annealing into elongated crystalline using SiO as well as H2O and OH masers at particles. On a still larger scale, in a series correspondingly greater distances to indicate of papers, Jura and colleagues have used an inverse square law like variation of infrared and submillimeter detections of magnetic field strength with distance around high luminosity, post AGB objects to infer evolved stars, and that the magnetic pressure grain sizes, based on the assumption of dominates the thermal pressure by a factor extrapolated extinction coefficients and of more than an order of magnitude. This other simplifying assumptions. The scaling clearly indicates how the dynamics can be laws (a = 3L*/16πGM*c ρ) and observed affected by the magnetic field. radio line widths argue for the existence of orbiting reservoirs of materials with 2.3 Recent Laboratory Work particles sizes as large as 5 millimeters, depending on whether the material obeys Whereas several groups have been pursuing opacity laws that vary with frequency to the laboratory studies of circumstellar aerosols, 0.6 power in the Rayleigh-Jeans limit. Ring recent work by Hallenbeck et al. (1998) and and disk masses as large as one solar mass Reitmeijer et al. (2002) appear to offer the were inferred (cf. Jura et al. 2000). Hence, most interesting prospects in connecting aerosol sizes from nanometer to millimeter astronomical and synthetic properties. In in circumstellar outflows have been both sets of experiments, amorphous estimated. magnesium silicate smokes were prepared by vapor phase condensation and annealed 2.2 Recent Microwave Maser Mapping in vacuum, then monitored by IR spectroscopy with respect to annealing time One of the key diagnostics of the dynamics and temperature. Hallenbeck et al. report of the CSE is the maser emission from SiO wavelength dependent shifts, moving from molecules (v=1, J=1-0 and J=2-1) detected 9.3 toward 11.2 microns, in silicates (Si-O at 43 and 86 GHz. Recently, improvements stretch modes) as the experiment progressed in polarization calibration at high from fully amorphous to increasingly frequencies and total intensity multi-dish crystalline, suggesting a natural pause interferometric observations have (“stall”) between the initially chaotic demonstrated the SiO masers spots are condensate and subsequently more ordered confined to a narrow ring like morphology crystalline forms. The stall represents a (tangentially amplified, cf. Diamond et al. stage in development of ordered material, 1994) around many late type stars, but chemistry (oxidation state) is suspended indicating orderly motions and systematic in favor of internal mobility of components velocity distributions in the 1 to few stellar (polymerization). Noteably, they report the radius domain. As the SiO molecule is appearance of a peak near 11.3 microns, nonparamagnetic, if exhibits significant which is cited as a crystalline feature in linear polarization in the presence of a cometary and astronomical spectra (cf. magnetic field, enabling determination of Molster et al., above), but appears in the lab well before significant degree of Once formed, the aerosol particles can crystallinity is developed. This suggests the behave independently from the gas in the prior interpretation of astronomical spectra stellar atmosphere, depending on the precise could be naïve. Hallenbeck et al. also conditions. There seems to be emerging studies iron silicate smokes and found these consensus that tiny grains will be decoupled to evolve orders of magnitude more slowly. from the gas and accelerated into the circumstellar envelope (MacGregor and The followup investigation reported by Stencel, 1992; Liberatore et al. 2001, Elitzer Rietmeijer et al. (2002) traces mineralogical and Ivezic 2001). Examining prior work by and chemical properties of the magnesium R. Gilman and by N. Berruyer, for under silicate lab samples, particularly an conditions typical of the circumstellar unanticipated size dependence of the envelopes of oxygen-rich red giant stars, petrological development of amorphous MacGregor and Stencel (1992) solved the phases as related to spectral changes. aerosol equation of motion and found that Specifically, grains smaller than 20 nm particles with radii smaller than 0.05 micron remained amorphous throughout the entire decouple from the ambient gas near the base thermal annealing experiment. Solid state of the outflow. Liberatore et al. (2001) changes only seemed to occur after smaller concur that electrostatic drag is many orders grains were fused and chemically of magnitude less important than collisional homogenized. Unstable nanocrystals of drag in outflows of this type, and describe independent SiO4 tetrahedra (neosilicates: two dynamical regimes: a boundary layer forsterite and tridymite) were present during where the particles are strongly accelerated stall phases until sufficient quantity of relative to the gas, and an outer regime magnesiosilicate sheet structures grew that where the two are strongly coupled, but do then supported development of chains of not compute details of the former. They tetrahedral nucleation (inosilicates, admit the decoupling will strongly influence including enstatite), in domains larger than grain growth. Simis et al. 2001 demonstrate 20 nm. They were able to fit the 9-11 that gas-particle drift is an essential feature micron spectral evolution with seven in time dependent hydrodyamical solutions subcomponent profiles, with narrowing and to explain the creation of quasi-periodic sharpening of peaks observed as annealing shells among highly evolved stars. progressed. Can evidence for analogous However, Elitzur and Ivezic (2001) proclaim behavior, including “stall” phases be seen in to have fully solved the dusty wind problem, CSE? Increasingly, astronomical at least from the layers outward after dust observations are beginning to sense the formation is completed. Once radiative presence in such environments, of a bimodal forces accelerate the particles relative to the distribution of particle sizes: small “hot” gas, the complete decoupling dominates and grains smaller than 50 nm, and larger sub- subsequent details are independent of the mm sized ones. This is a key point as we details of dust formation, depending only on next begin our discussion of the dynamics of final properties of the dust grains. these particles in aerosol states. Remarkably, they are able to predict from their general solution, that all optically thin 3.0 CIRCUMSTELLAR AEROSOL winds share universal velocity and aerosol DYNAMICS density profiles, including an interstellar, MRN power law distribution of emerging particles. In the late 1940’s, W.A. Hiltner and John Particle growth: Based on the foregoing, we Hall built the first astronomical optical are left with a picture of the outer polarimeters with the intent of measuring atmospheres of stars as turbulent regimes polarization signals emitted periodically where nucleation occurs, producing tiny from binary stars. Such an effect does exist, grains of corundum and small silicates although they failed to detect it. They did which accelerate quickly due to radiative however, observe that light from hundreds forces and decouple from the surrounding of stars, binary or not, is polarized, and that gas. Can we assemble a consistent picture this polarization increases in magnitude with that connects the gas-particle drift reddening, and that the position angles of phenomenon with the observed quantities of polarization tend to be parallel to the Milky amorphous and crystalline silicates? We Way (Hiltner 1949; Hall 1949). On a turn to calculations by Chokshi et al. (1993) historical note, notice that they published for insight regarding particle coagulation. separately in 1949 due to the fact that the Others have computed particle destruction in collaboration was aborted once they realized interstellar shocks, but Chokshi et al. deal the magnitude of their discovery. It was with the issues of particle adhesion as a during this time that Enrico Fermi result of collisions. They find that not only announced his theory on the magnetic do tiny particles accelerate better, but also acceleration of cosmic rays; leading to the show higher coagulation rates, neglecting postulation that magnetic alignment of dust charges. grains in the interstellar medium (ISM) is the cause of the anisotropic extinction of the starlight. In 1951, Hiltner then published 4.0 OBSERVATIONAL PROSPECTS polarization maps of the Galactic plane, from which Chandrasekhar & Fermi (1953) We have assembled a picture of aerosols in calculated a magnetic field strength on the circumstellar envelopes that includes the order of a few micro-Gauss, the formation of nanoparticles which are characteristics of which have been accelerated to hypersonic speeds relative to confirmed from observations of synchrotron the gas. Subsequent momentum coupling emission, Zeeman splitting, and Faraday and grain growth then account for the mass rotation. This was the beginning of the loss observed in stellar winds. However, study of polarimetry in astronomy to not exactly how do the nanoparticles form and only to probe the dust characteristics in the accelerate, and how do the subsequent galaxy, but in those from Young Stellar particle shapes and dynamics relate to the Objects (YSOs) to Planetary Nebulae (PNe). observed diversity of stellar winds and planetary nebula shapes? Two of the most There are two polarization mechanisms that potent tools available at the beginning of the arise from extinction of radiation: scattering 21st century involve spectropolarimetry in and dichroic absorption and emission from the mid-IR, and very long baseline aligned grains. Dichroism, as defined by interferometry [VLBI] of maser emission, Tinbergen (1996), is the differential which we now discuss. extinction of orthogonally polarized radiation components. Dichroism is 4.1 Spectropolarimetry generally considered to be the dominant mechanism in the mid-infrared (2-15 microns wavelength), indicative of a characteristic shift between peak absorption an extinction curve, and in the mid-IR, the and the peak polarization (Kobayashi et al. observed spectrum often has components 1980). Dust particles both in the ISM as due to absorption and emission. The well as circumstellar envelopes (CSEs), are physical structure and chemical mix of the non-spherical, and/or have grains also contribute to the shape and amorphous/crystalline structure. This position of the spectral feature, and results in a different extinction cross section spectroscopy only serves to identify the for the two transverse components of the chemical bonds, leaving more detailed incident electromagnetic radiation. If the information as to the physical and chemical wavelength of radiation is short compared to nature of the grains undeterminable. an effective radius of the grain, scattering will be the dominant polarization If one can assume that the underlying source mechanism. In the dipole approximation of radiation is unpolarized (an assumption used for calculating the extinction cross- that can be tested from the measurements), sections, (grain radius on the order of tenths than the polarization spectrum is the wavelength), the influence of scattering independent of the source spectrum, is diminished (α λ-4), and polarization due to presenting a less complicated view of the absorption becomes dominant as it is α λ-1 dust grains. Independent information (van de Hulst 1957; Bohren & Huffman regarding the chemistry and structure of the 1983). From Kirchhoff’s law, polarization grains is then available from due to absorption will be accompanied by spectropolarimetry since the polarization polarized emission (Hildebrand 1988), and if and extinction spectra are independent of scattering is negligible, the polarization one another (Aitken 1996). components due to absorption and emission As mentioned in the introduction, there are can be separated, as outlined by Aitken two competing polarization mechanisms (1996). At wavelengths much longer than brought on by the extinction of starlight grain radius, where the dipole approximation from dust grains: 1) scattering and 2) is no longer valid, polarization due to dichroic absorption/emission from aligned emission is the dominant process and can be grains. For pure scattering to occur, the thought of in terms of Babinet’s principle. effective grain radius must be much greater than the wavelength of radiation. Jura, et al. Infrared spectroscopy has long been used in (2000), have suggested 20 µm to 0.2 cm laboratories to identify vibrational grain sizes in CSE around select red giant transitions that give rise to the resonance stars, but the observations constrain the features characteristic of chemical bonds in dimensions of measureable particles, per the the solid state. Astronomical studies of dust dipole approximation: related spectral features should then provide a means of identifying the chemical 2π aeq <1 (3.6) properties of dust throughout the universe λ and to dissect the thermodynamics and magneto-hydro dynamics (MHD) associated Where aeq is the “equivalent” radius of the with their formation in stellar winds. grain, which is the radius of the equivalent Although in practice, one finds that the sphere having the same volume as the application is not so straightforward. The spheroid (Lee & Draine 1985). From the source of radiation often suffers from large polarization measurements, it will become visual extinction making it difficult to define apparent whether or not this assumption is correct, but it’s a good starting point as it use a single parameter. This parameter, first provides a reasonable test for determining derived by van de Hulst (1957) considered grain sizes in these environments the effect of incident radiation on particles (Hildebrand 1988). For wavelengths much in the dipole approximation, deviating from greater than aeq polarization due to dichroic spheres. This leads to the consideration of emission dominates, but this occurs out in spheroids, a class of ellipsoids with either the far-IR and sub-mm region of the oblate or prolate grain shapes. Oblate can electromagnetic spectrum. be thought of rotating an ellipse about its semi-minor axis, while prolate arises from In the Dipole approximation, the scattering rotation about the semi-major axis. and absorption efficiency “Q-factors” are Hildebrand & Dragovan (1995) compared found from Mie theory to be: measured versus predicted values of the polarizations using dielectric constants for 4 2 4 8  2πa   n2 −1   8  2πa   ε −1 astronomical silicate calculated by Draine Qsca = ℜ   = ℜ    2      (1985) for two interstellar clouds, the BN 3 λ   n + 2  3 λ  ε + 2    object in Orion and AFGL 2591. From their and studies they found oblate grains with ratio of short to long axis ~2:3 to best fit the 2  8πa   n2 −1    8πa   ε −1 observed polarizations. Qabs = ℑ   = ℑ    2       λ   n + 2    λ  ε + 2   The spectral region from 2-15 µm plays host to a variety of ice and carbonaceous/silicate Where n and ε are the complex index of dust species, appearing in both emission and refraction and complex dielectric function absorption. It attempting models to fit the respectively (Bohren & Huffman 1983). absorption and emission profiles, it would This result is important to show the λ-4 be desirable to put constraints based on the dependence of the scattering efficiency in arrangements of the dust species with the dipole approximation. In the analysis respect to one another. Some of the more that follows, scattering is assumed to be popular dust models have: 1) the ice and negligible in comparison to the absorption. carbonaceous/silicate species physically Then, there is solely polarization due to separated; 2) the core- mantle arrangement; absorption and emission, and a method for 3) fluffy aggregates. Spectropolarimetry can separating these two components can be place a constraint on the dust grain exploited. If scattering is important, than morphology, primarily in providing a the way in which the grain geometries and stringent examination of the core-mantle polarization mechanisms are calculated will arrangement. Correlations in the be inconsistent with what is observed. polarization magnitude and position angle should exist amongst those species that In the dipole approximation, it is possible to occupy the same region of space. If there is place constraints on the grain shape to first no correlation, then it can be said that the order, even though the polarization spectrum species cannot exist in a core-mantle is mostly dependent upon the grain and arrangement (Adamson et al. 1999; Aitken mantle chemistry. In order to characterize 1996; Holloway et al. 2002). So far, most of the shapes of the grains, since they are most the spectropolarimetric work done has been likely irregular and possibly aggregates of toward understanding dust composition on smaller particles, it makes the most sense to galactic center objects and young stellar objects (YSOs). Two most recent examples Just as valuable would be of this are the observations of Holloway et spectropolarimetric studies of late type stars, al. (2002) on YSOs and models proposed by in whose environments matter is continually Li & Mayo Greenberg (2002) for the ejected into dusty envelopes on the order of galactic center objects IRS3 and IRS7. a several stellar radii away from the stars themselves. The main difference between The spectropolarimetric observations of the study of late type stars and molecular Holloway et al. (2002) were concerned with clouds has to do with the formation and comparing polarization excess in the 3 µm evolution mechanism of the grains (Hagen, ice features of YSOs with the 9.7 µm Tielens & Greenberg 1983). In late type polarization of the silicate features observed stars it is expected that the grains condense by Smith et al. (2000). They found that in the outflows of matter from the star, good correlations exist between the whereas in the ISM (which includes YSOs polarization profiles of the two features, and galactic center objects), the grains are which would be indicative of ice mantles thought to condense in interstellar space, existing on silicate cores (Aitken 1996). possibly undergoing continual UV photo More specifically, similar polarization processing. The cyclic evolutionary dust position angles and specific polarization model (Mayo Greenberg 1989) predicts that (polarization per optical depth) should exist. interstellar grains will undergo It is inconclusive to say that if correlations approximately 50 cycles of processing in the exist, than ice mantled grains exist, but if ISM before they are consumed by star there is no correlation, then the core-mantle formation or become part of a comet (Mayo model can effectively be ruled out. Another Greenberg & Li 1999). Thus, in studying important result, is that the dominant dust in late type stars, one gets a glimpse of polarization mechanism in the near-IR was the raw products before they are ejected found to be due to dichroism. This was from the star and back into the ISM. Since found by measuring a characteristic shift in correlations in polarization have been found wavelength from the peak absorption between ice and silicate features, in support polarization values (Kobayashi et al. 1980). of a silicate core/ice mantle model, the Li & Mayo Greenberg (2002) used processing of grains might begin with ices, spectropolarimetric observations to being the precursor of the organic residue constrain a dust model based on near-IR found in the ISM. observations of the 3.4 µm carbonaceous feature in IRS7 (M2 I super giant) and the 4.2 Interferometry 9.7 µm silicate feature in IRS3. Their model predicts that a similar degree of polarization Recent decades have seen a flowering of should be measured for both spectral success with radio and infrared features in IRS7, where in this case, a interferometry, largely thanks to increased silicate core/hydrogenous amorphous carbon computing power. While the scope of this mantle is being tested. Although no review does not permit an exhaustive polarization was measured in the 3.4 µm summary, we will highlight current feature in IRS7, the core-mantle model representative results that indicate the power cannot be rejected due to the lack of 9.7 µm of the method and implications for the observations. studies of CSE aerosols.

The Infrared Spatial Interferometer (ISI) has 5. 0 SUMMARY a long series of successful mid-infrared measures, with high precision angular We have reviewed recent research on the diameters for alpha Ori (54.7 milliarcsec) origin and composition of solids in the and omicron Ceti (47.8 mas at phase 0.90) circumstellar envelopes of evolved stars, among the latest reports (see Weiner et al. including dynamics and interaction between 2000 and citations therein). gas and solid phases in outflows. New observational prospects for clarifying The Palomar Testbed Interferometer (van numerous issues regarding aerosols were Belle, et al. 2002 and references therein) has outlined. We thank Research Signposts for succeeded in resolving numerous AGB stars the invitation to create this review, Dr. at K band wavelengths (2.2 microns). With Shanhu Li for useful references to the milliarsecond resolution, in addition to aerosol literature, and acknowledge support providing evidence about fundamental mode of the estate of William Herschel Womble to pulsation, the work is beginning to separate the University of Denver for making this optical depth effects in CO and water ice work possible. components of the K band, and finding evidence for elliptical shapes in more than a minority of AGB stars examined. This REFERENCES offers one of the best ways to test proposed equatorially enhanced mass loss (cf. Stencel Adamson, A.J., Whittet, D.C.B., 2000 and references therein). Chrysostomou, A., Hough, J.H., Aitken, D.K., Wright, G.S. & Roche, P.F. 1999, Similarly, in the microwave regime, “Spectropolarimetric Constraints on the increasing numbers of stars have been Nature of the 3.4 Micron Absorber in the resolved in terms of their SiO, H2O and OH Interstellar Medium” ApJ, 512, 224 maser spot distributions, including the recent detection of collimated jet-like outflows in Aitken, D.K. 1996, in Polarimetry of the an AGB-like star (Imai, et al. 2002). These Interstellar Medium, eds. W.G. Roberge & observed distributions and maps are D.C.B. Whittet, ASP Conference Series, vol. challenging conventional interpretation, and 97. San Fransisco. Pgs. 225-242. open the way for novel observational efforts. Combined with new mid-infrared spectra Arlinger, B. 2000 Ph.D. Thesis, Univ. Wien, potentially forthcoming from NASA’s Space “The SiO Molecule in the Atmospheres and Infrared Telescope Facility (SIRTF) in 2003, Circumstellar Envelopes of AGB Stars” further tests of the silicate shape-microwave website: maser chronology are possible (Stencel et al. http://rigel.astro.univie.ac.at/~aringer/diss20 1990). 00/diss2000.html

One additional observational method that Bohren, C.F. and Huffman, D.R. 1983. deserves more attention is the production of Absorption and Scattering of Light by Small X-ray halos due to scattering by dust (cf. Particles (John Wiley & Sons). Smith and Dwek 1998). Red giant stars paired with x-ray sources in binary stars Bowen, G. 1988 “Dynamical modeling of should be examined for this effect. long-period variable star atmospheres” Astrophys. J. 329:299.

Chokshi, A., Tielens, A. and Hollenbach, D. Hiltner, W. 1949 “Polarization of Light from 1993, “Dust Coagulation”, Ap.J. 407: 806. Distant Stars by Interstellar Medium” Science vol. 109, p. 165. Diamond, P., Kemball, A., Junor, W., Zensus, A., Benson, J. and Dhawan, V. 1994 Hinds, W., 1999, Aerosol Technology, “Observation of a ring structure in SiO second ed., (New York, Wiley and Sons, maser emission from late-type stars” ISBN 0-471-19410-7). Astrophysical Journal v. 430, p. L61. Holloway, R.P., Chrysostomou, A., Aitken, Drain, B.T. 1985, “Tabulated optical D.K., Hough, J.H. & McCall, A. 2002, properties of graphite and silicate grains” “Spectropolarimetry of the 3-µm water-ice ApJ Suppl., 57: 587. feature towards young stellar objects” MNRAS, 336, 425. Elitzur, M. and Ivezic, Z. 2001 “Dusty Winds. I. Self-similar solutions” Monthly Holzer, T. and MacGregor, K. 1985 “Mass Notices R.A.S. 327: 403. loss mechanisms for cool stars” in Mass loss from red giants (proceedings), Dordrecht, D. Fabian, D., Henning, T., deJäger, C., Reidel Publishing Co., 1985, p. 229-255. Mutschke, H., Dorschner, J. and Wehrhan, O 2001, “Steps toward interstellar Imai, H.; Obara, K.; Diamond, P.; silicate mineralogy. VI. Dependence of Omodaka, T.; Sasao, T., 2002 “A collimated crystalline olivine IR spectra on iron content jet of molecular gas from a star on the and particle shape” Astron. & Astrophys. asymptotic giant branch” Nature vol. 417, 228: 238. p.829.

Hagen, W., Tielens, A.G.G.M. & Jaeger, C., Molster, F., Dorschner, J., Greenberg, J.M. 1983, “The three micron Henning, Th., Mutschke, H. and Waters, L., 'ice' band in grain mantles” A&A, 117, 132 “Steps toward interstellar silicate mineralogy. IV. The crystalline revolution”, Hall, J. 1949 “Observations of the Polarized 1998, Astronomy and Astrophysics, v.339, Light from Stars” Science vol. 109, p.166. p.904-916.

Hallenbeck, S.L., Nuth, J.A., Daukantas, Jura, M. , Turner, J., Van Dyk, S. and P.L. 1998, “Mid-Infrared Spectral Evolution Knapp, G. 2000, “What is Hatching in the of Amorphous Magnesium Silicate Smokes Egg (Nebula)?” Ap.J. 528: L105. Annealed in Vacuum: Comparison to Cometary Spectra” Icarus 131:198. Kemball, A. and Diamond, P. 1997 “Imaging the magnetic field in the Hildebrand, R.H. 1988, “Magnetic fields atmosphere of TX Camelopardalis” and stardust” Quarterly Journal of the Royal Astrophysical Journal v. 481, p.L111. Astronomical Society 29: 327. Kobayashi, Y., Kawara, K., Sato, S., Okuda, Hildebrand, R.H. & Dragovan, M. 1995, H. 1980, “Narrow-Band Polarimetry of the “The Shapes and Alignment Properties of Becklin/neugebauer Object and AFGL2591 Interstellar Dust Grains” ApJ, 450, 663 between 2 and 4 microns” PASJ, 32, 295. sample stars” Astronomy and Astrophysics, Jura, M. & Kahane, C. 1999, ApJ, 521, 302. v.382, p.184-221.

Lee, H.M. & Draine, B.T. 1985, “Infrared Muchmore, D., Nuth, J. and Stencel, R. extinction and polarization due to partially 1987 “The SiO Cooling Instability in aligned spheroidal grains - Models for the Envelopes of Cool, Giant Stars” Astrophys. dust toward the BN object” ApJ, 290, 211. Journal (Letters) 315: 141L.

Li, A. & Mayo Greenberg, J. 2002, “Mid- Nuth, J.A. and Donn, B. 1983 Laboratory Infrared Spectropolarimetric Constraints on studies of the condensation and properties of the Core-Mantle Interstellar Dust Model” amorphous silicate smokes”, Journal of ApJ, 577, 789. Geophysical Research -- February 15, 1983 -- Volume 88, Issue S1, pp. A847-A852. Liberatore, S., Lafon, J.P. and Berruyer, N., 2001, “Self-consistent Coupling of Olnon, F., Raimond, E. and the IRAS Radiative Transfer and Dyanmics in Dust Science Team, 1983 “The IRAS Low Driven Winds” Astronomy & Astrophysics Resolution Spectrometer Atlas” Astronomy 377: 522. and Astrophysics Supplement Series v. 65, pp.607-1065. MacGregor, K.B. and Stencel, R.E. 1992, “On the interaction between dust and gas in Rietmeijer, F.J.M., Hallenbeck, S.L., late-type stellar atmospheres and winds”, Nuth, J.A., Karner, J.M. 2002, “Amorphous Ap. J. 397: 644. Magnesiosilicate Smokes Annealed in Vacuum: The Evolution of Magnesium Mayo Greenberg, J. 1989, IAU Symp. 135, Silicates in Circumstellar and Cometary Interstellar Dust, ed. L.J. Allamandola & Dust” Icarus 156: 269. A.G.G.M. Tielens (Dordrecht:Kluwer), p.345. Seinfeld, J.H. and Pandis, S.N., 1998, Atmospheric Chemistry and Physics, (New Mayo Greenberg, J. & Li, A. 1999, York, Wiley and Sons, ISBN 0-471-17816- “Tracking the Organic Refractory 0). Component from Interstellar Dust to Comets” Adv. Space Research, vol. 24, 497. Simis, Y., Icke, V. and Dominik, C. 2001 “Origin of Quasi-periodic Shells in Dust Molster, F. J., Waters, L. B. F. M., forming AGB Winds” Astronomy & Trams, N. R., Van Winckel, H., Decin, L., Astrophysics 371: 205. van Loon, J. Th., Jäger, C., Henning, Th., Käufl, H.-U., de Koter, A. and Bouwman, J., Sogawa, H. and Kozasa, T. 1999, “On the 1999, “The composition and nature of the Origin of Crystalline Silicate in dust shell surrounding the binary AFGL Circumstellar Envelopes of Oxygen-rich 4106” Astronomy & Astrophysics 350: 163. Asymptotic Giant Branch Stars” Ap.J. 516: L33. Molster, F.J., Waters, L.B.F.M., Tielens, A.G.G.M., Barlow, M.J., 2002, “Crystalline Smith, R. and Dwek, E. 1998, “Soft X-ray silicate dust around evolved stars. I. The Scattering and Halos from Dust” Astrophys. Journal vol. 503, p.831. van den Ancker, M.; Tielens, A.; Smith, C.H., Wright, C.M., Aitken, D.K., Wesselius, P., 2000 “ISO Spectroscopy of Roche, P.F. & Hough, J.H. 2000, “Studies in the young bipolar nebulae S106 IR and Cep mid-infrared spectropolarimetry - II. An A East” Astronomy & Astrophys. Vol. 358, atlas of spectra” MNRAS, 312, 327. p.1035.

Speck, A.K., Barlow, M.J., Sylvester, R.J., Van de Hulst, H.C. 1957. Light Scattering Hofmeister, A.M., 2000, “Dust features in by Small Particles (Wiley, New York). the 10 micron infrared spectra of oxygen- Reprinted 1981 (Dover, New York). rich evolved stars” Astron. Astrophys. Suppl. Ser. 146, 437. Vlemmings, W., Diamond, P. and van Langevelde, H. 2002 “Circular polarization Spoon, H. Keane, V., Tielens, A., Lutz, D. of water masers in the circumstellar and Moorwood, A. 2002 “Ice features in the envelopes of late type stars” Astronomy & mid-IR spectra of galactic nuclei” Astrophysics v. 394, p.589. Astronomy & Astrophysics v. 385, p.1022. Weiner, J.; Danchi, W.; Hale, D.; Stencel, R.E., Nuth, J.A., Little-Marenin, I. McMahon, J.; Townes, C.; Monnier, J.; and Little, S. 1990, “The Formation and Tuthill, P., 2000, “Precision Measurements Annealing of Circumstellar Dust as Gauged of the Diameters of α Orionis and ο Ceti at by IRAS Low Resolution Spectra and the 11 Microns” Astrophys. Journal 544, 1097. Microwave Maser Chronology” Ap.J. 350: L45. Winters, J. M.; Keady, J. J.; Gauger, A.; Sada, P. V. 2000, “Circumstellar dust shells Stencel, R.E. 2000, “AGB Star Polar around long-period variables. VIII. CO Breakout and the Origin of Aspherical infrared line profiles from dynamical models Planetary Nebulae” in “Asymmetrical for C-stars” Astronomy & Astrophysics v. Planetary Nebulae II: From Origins to 359, p.651. Microstructures”, ASP Conference Series, Vol. 199. Edited by J. H. Kastner, N. Soker, and S. Rappaport. ISBN: 1-58381-026-9, 2000, p. 99 (END)

Tielens, A., Waters, L., Molster, F. and Justtanont, K. 1997, “Circumstellar Silicate Minerology” Astrophys. & Space Sci. v. 255, p.415.

Tinbergen, J. 1996, “Astronomical Polarimetry” Cambridge University Press. van Belle, G. T.; Thompson, R. R.; Creech- Eakman, M. J. 2002, “Angular Size Measurements of Mira Variable Stars at 2.2 Microns. II.” Astronomical Journal vol. 124, p. 1706. Addendum, 3/4/2004: and an isotropic wind acceleration mechanism. The wind couples to the field, Other than the pioneering, classic near-IR stretching it until the field lines become polarimetry of AGB stars by Johnson and mostly radial and oppositely directed above Jones (1990), one of the key diagnostics of and below the magnetic equator, as occurs in the dynamics of circumstellar envelopes is the solar wind. The interaction between the the maser emission from SiO molecules wind plasma and magnetic field near the star (v=1, J=1-0 and J=2-1) detected at 43 and 86 produces a steady outflow in which GHz. Recently, improvements in magnetic forces direct plasma toward the polarization calibration at high frequencies equator, constructing a disk. In the context -5 and total intensity multi-dish interferometric of a slow (10 km/s) outflow (10 Mo/yr) observations have demonstrated the SiO from an asymptotic giant branch star, MHD masers spots are confined to a narrow ring simulations demonstrate that a dense like morphology (tangentially amplified, cf. equatorial disk will be produced for dipole Diamond et al. 1994) around many late type field strengths of only a few Gauss on the stars, indicating orderly motions and surface of the star. A disk formed by this systematic velocity distributions in the 1 to model can be dynamically important for the few stellar radius domain. As the SiO shaping of molecule is nonparamagnetic, if exhibits planetary nebulae.” Similarly, we ask and significant linear polarization in the plan to investigate with IR spectro- presence of a magnetic field, enabling polarimetry whether these types of determination of magnetic field strength in conditions can shape the dust particle this key region. Kemball and Diamond geometry on micro and macro levels in (1997) deduced a line of sight field strength AGB star outflows. in the Mira TX Cam of 5-10 gauss. Vlemmings et al. (2002) have compiled this References: and related magnetic field strength Johnson, J. and Jones, T.J. 1991 A.J. determinations for related stars using SiO as 101:1735, “From red giant to planetary well as H2O and OH masers at nebula - Dust, asymmetry, and polarization” correspondingly greater distances to indicate Kemball, A. and Diamond, P. 1997 an inverse square law like variation of Astrophysical Journal 481: L111, “Imaging magnetic field strength with distance around the magnetic field in the atmosphere of TX evolved stars, and that the magnetic pressure Camelopardalis” dominates the thermal pressure by a factor Matt, S., Balick, B., Winglee, R. and of more than an order of magnitude. This Goodson, A. 2000 ApJ 545: 965, “Disk clearly indicates how the dynamics can be formation by asymptotic giant branch winds affected by the magnetic field, much as in dipole magnetic fields” proposed by Matt et al. (2000) who present a Vlemmings, W., Diamond, P. and van “simple, robust mechanism by which an Langevelde, H. 2002 A& A 394: 589, isolated star can produce an equatorial disk. “Circular polarization of water masers in the The mechanism requires that the star have a circumstellar envelopes of late type stars” simple dipole magnetic field on the surface

Figure 1 from Matt et al. 2000.