INTERSTELLAR EXTINCTION, GALACTIC DAVID A. WILLIAMS Interstellar extinction is the attenuation of starlight caused by the absorption and scattering of the light by interstellar dust along the line of sight between the star and the observer. Extinction has been recognized for about 50 years as a general phenomenon in the Galaxy, and one that is particularly marked in the galactic plane. The effects of extinction were already evident some 200 years ago in the surveys of stars by William Herschel. The apparent absence of stars in a localized region of the sky caused Herschel to remark that this was indeed a "hole in the heavens." The application of photography to astronomy (particularly by Edward Emerson Barnard), a century later, indicated that these "holes" were in fact obscuration due to dust clouds. The fact that there is a general obscuration in the galactic disk, not merely confined to prominent isolated dust clouds, was slowly recognized. Particularly convincing arguments concerned the observations of stellar clusters whose distances could be estimated both from their angular sizes and from the luminosities of stars within them. The luminosity-derived distances are affected by extinction whereas the distances based on angular sizes are independent of it. Comparisons of these two measures established that extinction was severe and ubiquitous in the plane of the Galaxy, and confined to a relatively thin layer. It was also evident that blue light is more heavily extinguished than red, so that an apparent reddening of starlight is observed. This interstellar reddening can also be used to map the presence of dust in the Galaxy. The general trend of increasing extinction at shorter wavelengths is, with some exceptions, now known to be broadly true not only over the visual wavelengths but also from infrared to ultraviolet. The accepted origin for the bulk of the extinction, at least in the visual, is the absorption and scattering of starlight by a population of dust particles whose sizes are comparable to the wavelength of light. A variety of models consistent with the cosmic constraints on mass and chemical composition have been developed and are described elsewhere. The concept of extinction by dust particles is supported by evidence of many kinds; in particular, the serendipitous discovery of interstellar polarization is interpreted as the differential extinction caused by the partial alignment of aspherical dust particles. Radiation in the visual and ultraviolet that is absorbed by dust particles heats the dust and is reradiated at longer wavelengths; this longer wavelength radiation is observed directly. Radiation in the visual and ultraviolet that is scattered by the dust creates a diffuse background of light, the diffuse galactic light that is observed both in the vicinity of bright stars near dust clouds and throughout the plane of the Galaxy, similar to the zodiacal light in the solar system observed in the ecliptic plane of the sky. INTERSTELLAR EXTINCTION CURVE The intensity *(*) of radiation received at wavelength * compared with that which would have been received, **(*), in the absence of extinction caused by an intervening slab of material, may be written ************************ Here, *(*) is called the optical depth. More extinguishing material in the slab leads to a larger value of *. In practice, astronomers use a logarithmic measure known as magnitudes to measure changes in brightness. On this scale, a difference in brightness of 100 is represented by a magnitude difference of 5. In these terms, extinction measured in magnitudes is very simply related to optical depth: ************************* Differences in extinction at different wavelengths, * and **, cause a change in color, leading to a color excess *****************************. A widely used method for studying interstellar extinction is the so-called pair method in which the intrinsic colors for unreddened stars are compared to the colors obtained for reddened, but otherwise similar, stars. The extinction *(*) measured to individual stars in this (and other) ways depends on the amount of extinguishing dust along the lines of sight to those stars. This amount depends in turn on the distances to the stars, and on the distribution and number density of the dust particles at every point along the lines of sight. To compare extinctions in a standard way it has become conventional to measure the colors *(*-**) relative to the color at a "visual" wavelength *=V (defined to be 5400 *) and to normalize this color excess E(*-V) for each star so that E(B-V)=1, where B represents a blue wavelength, defined to be 4200 *. In this way, a meaningful comparison of interstellar extinctions along different lines of sight can be made. It should, however, be emphasized that this conventional and convenient normalization of extinction may be misleading if there are wide variations in the character of extinction in the region of the B and V wavelengths. The normalized extinction E(*-V)/E(B-V) is found to vary with wavelength. As extinction *(*) tends to zero as wavelength increases, the normalized extinction approaches -R in this limit, where **************************** is the ratio of total to selective extinction. Observations indicate that R has about the same value (approximately 3) along many lines of sight. This implies that dust particles along the various lines of sight tend to have similar optical properties in the visual region and may, therefore, be chemically and structurally similar throughout the Galaxy. Assuming that R is known, then the relatively straightforward measurement of the color E(B-V) allows an immediate estimate of the total visual extinction A*. An average normalized interstellar extinction curve for the Galaxy is shown in Fig. 1. Whereas early work on interstellar extinction curves drew attention to the near uniformity of interstellar extinction curves (apart from a few localized anomalies), it is now recognized that quite significant variations exist between different lines of sight. These variations are particularly apparent in the far ultraviolet though this is perhaps, in part, due to the normalization process. In general, however, the behavior indicated for the average in Fig. 1 represents at least qualitatively the extinction observed on particular lines of sight. INFRARED EXTINCTION Extinction measurements in the infrared are subject to substantial systematic errors because of the difficulty of allowing correctly for infrared emission from warm dust grains in stellar or circumstellar environments. The infrared extinction curve between about ***= 0.2 and 0.8 *m** (*=5 and 1.25 *m) has recently been reevaluated for a population of reddened early-type stars within 2-3 kpc of the Sun. The stars are widely distributed about the galactic plane and obscured predominantly by dust in atomic hydrogen clouds. The extinction is fairly accurately represented by the formula *********************** where *=1.70*0.08 in the range 1.6-5*m. The data extrapolate to ***=0 to give R=3.05*0.05. There is little evidence of strong variation in infrared extinction for this sample of stars. Deviations from this mean extinction law occur in particular absorption bands, which are associated with vibrational transitions in the solid matrix of the dust material. These absorptions may substantially enhance the mean extinction. Along lines of sight through darker molecular clouds (say, A****) the extinction is modified by the appearance of a variety of absorption bands arising, in most cases, from molecular ices frozen onto the surfaces of dust grains. Table 1 lists the features observed and their usual identifications. The water-ice feature at 3.0 *m can be a particularly important absorption, significantly affecting total extinction in this wavelength region. The absorptions at 9.7 and 19 *m are observed even in diffuse clouds. It is accepted that these two bands arise in the silicate component of the dust-a view supported by the detection of the 9.7-*m feature in emission from hot grains near stars. The strength of the 9.7-*m feature in absorption is proportional to visual extinction according to ************************, showing that it is a true interstellar extinction feature. VISUAL EXTINCTION Visual extinction is characterized by a monotonic rise, that occurs approximately linearly with 1/*. Differences in normalized extinction from star to star are detectable but generally minor, and the normalized extinction curves show much uniformity. The 1/* dependence suggests that much of the visual extinction is contributed by dust particles of a size comparable with the wavelength of visual light. The ratio R of total to selective extinction is about 3.1 for many lines of sight. Recently, some departures from the average extinction have been detected at wavelengths longer than 5000 *. The departure can be interpreted as a broad emission extending over a width of about 2000 * and peaking near 6500 or 7000 *. It is likely to be due to dust luminescence from particles in the vicinity of hot stars that is excited by the stellar ultraviolet radiation. In addition, certain well-known, ubiquitous but as yet unidentified absorption bands have been detected in the visual. The band at shortest wavelength, and the most prominent, is centered at 4430 *. The origin of these bands is controversial, but may lie in the dust particles. ULTRAVIOLET EXTINCTION Ultraviolet extinction has been measured from the visual out to ***=10 *m**. The characteristic shape of the extinction curve in the ultraviolet is a broad peak of extinction centered near ***=4.6 *m**, and a rise from about ***=6 *m** into the far ultraviolet. The central wavelength of the peak shows little variation between the lines of sight to different stars though, when normalized to constant E(B-V), the strength and width of the peak may show substantial variation. (This feature is sometimes called the "2200 * extinction bump.") In a sample of 45 stars the mean central wavelength is 2174.4 * with no individual star deviating more than *17 * from this mean. It is interesting to note that two particular stars with extremely different far ultraviolet extinctions differ in the wavelength of the peak by only 1.4 *. The width of the peak is substantial: The full width of the peak at half maximum strength varies from 360 to 600*. Broader peaks are found in dark clouds and reflection nebulae, whereas in the diffuse medium and in star formation regions the peaks are narrower. Its origin in the solid state seems certain, but remains controversial. The strength of the absorption at 4.6 *m** (as measured by the area under the peak) varies by more than a factor of 3 and correlates well with E(B-V), with a few exceptions. The carrier of this feature must, therefore, either be in the same population of dust grains also responsible for visual extinction or in a population coexisting with them. Far ultraviolet extinction is observed to show considerable variation between different lines of sight. Extreme variations occur: One star in Orion has a far ultraviolet extinction that is almost flat, that is, wavelength independent, for **** 6 *m**, whereas other stars show a very steep rise. These are not exceptions to normal behavior: Statistical studies show that the strength of the far ultraviolet extinction is poorly correlated either with visual extinction or with the strength of the peak at ***=4.6 *m**. It appears that the dust must contain a component that contributes significantly neither to visual extinction nor to the 4.6 *m** peak, but that must be capable of substantial variation in the far ultraviolet. Additional Reading Allamandola, L.J. and Tielens, A.G.G.M., eds. (1989). Interstellar Dust. Kluwer Academic Publishers, Dordecht. Fitzpatrick, E.L. and Massa, D.(1986). An analysis of the shapes of ultraviolet extinction curves. I. The 2175 bump. Ap. J. 307 286. Fitzpatrick, E.L. and Massa, D.(1988). An analysis of the shapes of ultraviolet extinction curves. II. The far-UV extinction. Ap. J. 328 734. Martin, P.G.(1978). Cosmic Dust: Its Impact on Astronomy. Clarendon Press, Oxford. Mathis, J.S.(1990). Interstellar dust and extinction. Ann. Rev. Astron. Ap. 28 37. Savage, B.D. and Mathis, J.S.(1979). Observed properties of interstellar dust. Ann. Rev. Astron. Ap. 17 73. Whittet, D.C.B.(1988). The observed properties of interstellar dust. In Dust in the Universe, M.E. Bailey and D.A. Williams, eds. Cambridge University Press, Cambridge, p. 25. Witt, A.N.(1988). Visual and ultraviolet observations of interstellar dust. In Dust in the Universe, M.E. Bailey and D.A. Williams, eds. Cambridge University Press, Cambridge, p. 1.