7 'ьььњ~؊؊؊ ؖv R^^|xr **Tь interstellar dust and extinction John S. Mathis Washburn Observatory, University of Wisconsin- Madison 475 N. Charter St., Madison, Wisconsin 53706, U.S.A. This paper is based largely upon the report in Annual Reviews of Astronomy and Astrophysics, 28, 37, 1990, except that an Appendix has been added describing the physical conditions in various phases of the interstellar medium. 1 Introduction Interstellar dust is an important constituent of the Galaxy. It obscures all but the relatively nearby regions in visual and ultraviolet wavelengths, and reradiates the absorbed energy in the far-infrared (FIR; l 60 m) part of the spectrum, thereby providing a major part ( 30%) of the total luminosity of the Galaxy. The FIR radiation from dust removes the gravitational energy of collapsing clouds, allowing star formation to occur. Dust is crucial for interstellar chemistry by reducing the ultraviolet (UV) radiation which causes molecular dissociations and providing the site of the formation of the most abundant interstellar molecule, H2. Probably grain surfaces are responsible for other chemistry as well. Dust controls the temperature of the interstellar medium (ISM) by accounting for most of the elements which provide cooling, but also providing heating through electrons ejected photoelectrically from grains. The past decade has seen an increase in interest in interstellar dust because of the discovery of spectroscopic features in both emission and absorption, along with laboratory studies of candidate materials. There have been good observations of the extinction law of dust in many directions. Probably the most important feature to emerge from these studies is that "interstellar dust" refers to a variety of materials of widely varying properties. Many studies of interstellar dust have involved lines of sight through the diffuse, low-density ISM, including some clouds of densities of up to several hundred H atoms cm3. In this review this material will be called "diffuse dust". In the literature, most references to "inters12tellar dust" apply to diffuse dust. Dust in the outer parts of molecular clouds which can be studied by optical and UV observations will be called "outer-cloud dust". Finally, there have been many studies of sources embedded so deeply within molecular clouds that only the near-infrared or perhaps optical part of the spectrum can be studied. This type of dust will be called "inner-cloud dust". There is, of course, a continuous gradation of properties from diffuse dust to inner-cloud dust, but these three designations will allow us to emphasize the rather different properties of interstellar dust in the various regions.This review will confine itself to diffuse- and outer-cloud dust. There are excellent reviews of inner-cloud dust (156,157, 165). Whittet (1987) Tielens and Allamandola (1987), and Tielens (1989). Recent general references regarding interstellar dust are: (a) the proceedings of a 1985 workshop held at Wye, Maryland, U.S.A. (Nuth and Stencel 1986)(123); (b) the 1987 conference "Dust in the Universe" in Manchester, UK (Bailey and Williams 1988)(7), and (c) IAU Symposium 135, "Interstellar Dust", held at Santa Clara University, U.S.A., July, 1988 (Allamandola and Tielens 1989)(3). In general, reviews in these volumes on specialized aspects will not be referenced specifically. In general, recent references are given in preference to older but important papers. Extinction (absorption plus scattering) is by far the best-studied property of diffuse dust and outer-cloud dust because it can be determined accurately over a wide range of wavelengths and for lines of sight sampling different physical conditions in the ISM. Both continuous extinction and certain spectral features (rather narrow wavelength regions over which the extinction varies appreciably) will be discussed in 2. Another very important diagnostic is the emission from dust (3), both in spectral features (which provide clues as to specific materials) and in the FIR, representing the emission from grains warmed by incident radiation or particles. Scattering (4), polarization (5), and other diagnostics (6) also provide information . Theories are outlined in 7 and the evolution of dust in 8. A summary is given in 9. In general, references are given to recent papers without explicit credit being given to important older work. In this article I refer to the wavelength region (0.9 m < l < 10 m) as the near-infrared (NIR), 10 m l < 30 m as the mid-infrared (MIR), and l > 30 m as the far-infrared (FIR). 2 Interstellar extinction 2.1. Continuous extinction Each line of sight has its own "extinction law", or variation of extinction with wavelength, usually expressed by A(l)/A(V) in this article. This means of expressing the extinction law is not unique; it has been common practice to use instead the ratios of two colors, E(l -V)/E(B-V), where E(l -V) = A(l) - A(V). The use of A(V) as the reference extinction is arbitrary, and it might be preferable to use a longer wavelength, such as the J bandpass ( 1.25 m), because the extinction law would then be virtually independent of line of sight (see 2.1.3). 2.1.1. Optical/Ultraviolet extinction of diffuse and outer-cloud dust. We can only observe the UV extinction in diffuse dust and outer-cloud dust. Since a great deal of information is gained by considering the UV part of the spectrum, it is just this difference which makes the division into types of dust convenient. There have been several studies of the spatial distribution of extinction in the optical part of the spectrum (FitzGerald 1968; Neckel and Klare 1970; Lucke 1978)(e.g., 51, 101, 122) which might be useful in estimating the amount of extinction in a particular direction, but this review will concentrate upon the form of the extinction law and the physical nature of dust. Ardeberg and Virdefors (1982)(5) discuss the optical extinction law, with references. Many authors have utilized the International Ultraviolet Explorer (IUE) to make detailed studies of the UV extinction law of diffuse- and outer-cloud dust. There are considerable differences among the various lines of sight. Cardelli, Clayton, and Mathis [198819, 1989 (20, hereafter CCM)] have used the UV observations of Fitzpatrick and Massa 1986(54, hereafter FM86; 198855; Massa and Fitzpatrick 1986109), with optical/NIR observations of the same stars, to explore the relationships between various extinction laws over the entire available interval of wavelengths. These observations were spread over the sky, and included both diffuse dust and lines of sight to the Ophiuchus, Orion, and other molecular clouds, as well as H II regions. CCM used the optical total-to-selective extinction ratio, RV = A(V)/E(BV), as a parameter. Figure 1 shows the observed extinction laws of many lines of sight, plotted against 1/RV, for several values of l ranging from the red to almost the limit of the IUE spacecraft (1200 ). There are fairly tight linear relationships between A(l)/A(V) and RV1 in each case, including the UV. The value of RV depends upon the environment along the line of sight. A direction through low-density ISM usually has a rather low value of RV, about 3.1. Lines of sight penetrating into a dense cloud, such as the Ophiucus or Taurus molecular clouds, usually show 4 300 eV, the absorption law of the dust is approximately the same as if all of its atoms were neutral in the gas phase. At lower energies, especially just above the thresholds of the abundant elements like carbon, the large grains are opaque and the effective cross section per H atom is reduced. At 24 eV, the reduction amounts to a factor of 4. The major effects of the dust as regards high-energy radiation are (a) to keep its constituents absorbing as neutral atoms, rather than possibly being ionized, and (b) to scatter the radiation, with a cross-section about equal to the absorption. This scattering can be observed as an X-ray halo around point sources (Mauche and Gorenstein 1986)(117). Dust does not affect the ionization equilibrium of H II regions very significantly (Mathis 1986b)(112) because its absorption, peaking at 17 eV, resembles hydrogen absorption too closely. 2.1.5. EXTRAGALACTIC EXTINCTION There are reasonably reliable measurements for the extinction laws and dust/gas ratios only for the Magellanic Clouds. In the Large Magellanic Cloud (LMC), RV, 3.2 0.2, virtually Galactic (Clayton and Martin 198524, Morgan and Nandy 1982119, Koornneef 198286). For the UV, the stars near the giant HII region 30 Doradus have weak bumps and extinctions rising steeply at the shortest IUE wavelengths, a behavior unfortunately known as "the LMC extinction law". However, the stars well away from 30 Doradus (> 500 pc projected distance), spread throughout the galaxy, have approximately galactic extinction laws (Clayton and Martin 198524, Fitzpatrick 198553). The N(H)/E(BV) is 2 x 1022 atoms mag-1 (Koornneef 198286, Fitzpatrick 198553), about four times the Galactic value (Bohlin, Savage, and Drake 19789) and about proportional to the gaseous carbon abundance in the LMC. In the Small Magellanic Cloud (SMC), there are almost no suitably reddened stars. In general there seems to be a low value of RV, almost no bump, and a very steep far-UV rise (Bouchet et al 198511, Nandy et al 1981 121), as might be expected from a small RV. One star, though, shows an extinction law similar to the Galaxy (Lequeux et al 198296). The N(H)/E(BV) is 4.5 x 1022 atoms mag-1, about 10 times galactic and consistent with the gaseous C abundance in the SMC. 2.2. The 9.7 and 18 m silicate features Spectral absorption features can provide a great deal of information regarding the compositions and nature of the grains. Such features have not been detected in the UV (York et al. 1973, Seab and Snow 1985, Fitzpatrick and Massa 1988)(55, 148, 184), except, of course, for the bump. However, many have been found in the NIR, especially from the icy mantles within molecular clouds. This review will concentrate on diffuse- and outer-cloud dust, and will not discuss the bands of molecular ices in general. There is a broad, smooth absorption feature peaking at 9.7 m, attributed to the SiO stretch in silicates, which is always seen in interstellar dust if A(V) is suitably large. The 9.7 m band is found in emission from warm circumstellar dust surrounding oxygen-rich stars. In these objects the heavy elements (Fe, Mg, etc.) are in silicates in the expanding envelope, while the carbon combines almost completely with oxygen to form CO. The 9.7 m feature is not seen in circumstellar dust surrounding carbon-rich objects, except for some dusty planetary nebulae in which the grains were expelled by an earlier, and possibly oxygen-rich, phase of stellar evolution. A somewhat weaker and even broader feature peaking at 18 m, the SiOSi bending mode, has been detected in circumstellar dust, in stars near the galactic center, and in molecular clouds (McCarthy et al. 1980; Pegourie and Papoular 1985; Volk and Kwok 1988; Aitken et al 1988)(2, 118, 128, 160). The 18 m band is much less well studied, but is found with the strength relative to that of the 9.7 m expected for silicates (about 0.4). The bands are polarized at the same angle (Knacke and Capps 1979; Aitken et al. 1988)(2, 84) with the amplitudes expected for silicates. The strength and profile of the 9.7 m band, relative to A(V), has been determined Roche and Aitken (1984) (137) from several Wolf-Rayet type WC stars. These objects have the advantage of being luminous and of having no intrinsic spectral features in the 10 m region (because they are carbon-rich and therefore contain no circumstellar silicate dust). If t9.7 is the optical depth of the silicate feature above the underlying continuum, the mean value of A(V)/t9.7 in diffuse dust is (18.5 1.0). However, there are substantial variations of A(V)/t9.7 within the Taurus molecular cloud (Whittet et al. 1988)(169). The line of sight towards the galactic center has A(V)/t9.7 9, about half the local value (Roche and Aitken 1985)(138), although the strong emission band at 7.7 m confuses the determination of the continuum underlying the band (Cohen, Tielens, and Bregman 1989)(28). The dust seems diffuse in the sense that there are only weak radio lines from molecules commonly seen in molecular clouds, but conditions in the inner galaxy (chemical composition, for instance) certainly differ from our neighborhood. The derived shape of the 9.7 m band is uncertain because the band is only about half as strong at maximum as the continuous extinction at the K-band (2.2 m). In diffuse dust towards WC stars, the shape of the band is similar to the emission seen in dusty oxygen-rich stars such as Cep (Roche and Aitken 1984; Little-Marenin 1986)(99, 137). The emission profile near the Trapezium in the Orion Nebula is 40% broader (Gillett et al. (1975)(60), and the profile in dense clouds appears to be 10% narrower (Pegourie and Papoular 1985)(128). It is not surprising that the shape of the band is different in various types of objects; very probably the silicates are in different states of order, with different degrees of impurities. I recommend the Roche and Aitken profile (1984)(137) as being typical of diffuse dust. The 9.7 m profile shows that interstellar silicate is not crystalline (Butchart and Whittet 198316). Crystalline silicate absorption peaks at about 10.5 m, rather than 9.7 m (see spectra in Sandford and Walker 1985143). Laboratory measurements of amorphous or radiation-damaged silicates (Day 1979; Krtchmer and Huffman 197933, 88) show a satisfactory, but not perfect, fit to the observed profile. Stars embedded in molecular clouds warm and partially anneal nearby grains (Aitken et al. 19882). The silicate band is weak for observational purposes, but it is difficult to account for its strength even if all of the silicon is in silicates and if a rather large opacity (3000 cm2 gm-1) is assumed for the maximum absorptivity (DL; Tielens and Allamandola 1987157). This value is larger then most laboratory measurements of amorphous or lunar silicates. The total equivalent width of the band is rather constant from one type of silicate to another, so the broad profile of the interstellar band limits the maximum strength. The fundamental Kramers-Krnig relations limit the strength of the band (DL) if astronomical silicates are similar to the minerals studied in the laboratory, but heavy contamination with other materials, and perhaps a porous nature, greatly complicate the issue. There are suggestions that interstellar silicates are hydrated (Knacke and Krtchmer 1980; Hecht et al. 198668, 85), based on a 6.00 m H-O-H bending band. However, in general the 6.00 m feature is not correlated with the 9.7 m band (Willner et al. 1982172). 2.3. mean extinction laws Table 1 is an estimate of the extinction law for the observable range of wavelengths, normalized to J (1.25 m) because the extinction law is assumed to be independent of environment for l > 0.9 m. The reader can convert to A(l)/A(V) by means of the tabulated A(V)/A(J). There are two columns for l < 0.9 m, representing the mean for diffuse dust (RV = 3.1) and outer-cloud dust (RV = 5), both calculated from CCM. The difference between the two laws is striking. The 9.7 and 18 m feature profiles, as given by "astronomical silicate" (DL), have been added to a power-law interpolation of an underlying continuum fitted between 250 and 7 m. The profile of the silicate band was truncated at 35 m, as appropriate for circumstellar dust (Pegourie and Papoular 1985128), but perhaps not for interstellar dust. The FIR opacity should be extrapolated to longer wavelengths with a l-2 dependence (see 3.2.3); the value in the table is determined by the estimate of Hildebrand (198370). The ionizing-UV cross sections are from Martin and Rouleau (1988107), adjusted by a factor of 1.15 to make them continuous with the extinction of CCM at 0.12 m. Both the "astronomical silicate" and ionizing-UV opacities are based upon the bare-silicate/graphite grain model (see 8) and depend (through the Kramers-Krnig relations) upon the assumption that interstellar grains have densities of 3 gm cm3. The entries in the table for l > 15 m are uncertain by at least a factor of two, and probably more. There are as yet very few observational constraints upon the extinction law between the longer wavelengths of the silicate feature (which probably varies from one line of sight to another) through l > 100 m, at which wavelengths the energy is produced by steady-state emission from large grains. The l > 100 m opacity is somewhat constrained by the emission from isolated clouds warmed by the interstellar radiation field which can be estimated. 3 Emission from Dust 3.1. The Unidentified Infrared Bands The realization (Sellgren 1984; Sellgren et al. 1985150, 151) that diffuse dust produces strong unidentified infrared emission bands (UIBs) in the 3.3 - 11.3 m range, and an associated continuum, has stimulated a great deal of research within the last five years. The carriers of the UIBs are surely important components of the ISM. The UIBs have been discussed extensively (Tielens and Allamandola 1987; Lger, d'Hendecourt, and Dfourneau (1989); Puget and Lger 1989; Allamandola, Tielens, and Barker 19893, 94, 133, 157). Some of the main features of the UIBs are: 1. The strongest bands are at 3.3, 6.2, 7.7, 8.6, and 11.3 m. These wavelengths all closely correspond to the CH or CC bond vibrations in aromatic (benzene-ring) structures. The simplest substances which can produce these bands are simple, planar molecules called polycyclic aromatic hydrocarbons (PAHs), but other, less-well-ordered configurations of carbon and hydrogen can also produce them (Sakata et al. 1984; Borghesi, Bussoletti, and Colangi 198710, 142). A suggestive fit to the bands is provided by absorption from vitrinite (Papoular et al. 1989125), partially ordered graphite from coal. A mixture of PAHs can reproduce all of the UIBs, including the weak ones and the profiles of the strong ones (Witteborn et al. 1989; Geballe et al. 198957, 180). 2. Diffuse UIB emission, found throughout the Galaxy (Giard et al. 198959), is responsible for 10 - 20% of the total radiation from dust. UIBs and the associated continuum dominate the IRAS filter responses at 12 and 25 m (Ryter et al 1987141), and are presumably responsible for the galactic "cirrus" emission in these filters (Boulanger and Prault 198812). 3. The bands are also found in planetary nebulae, "reflection" nebulae, H II regions, and extragalactic objects (Willner 1984; Cohen et al. 1986, 1989,27, 29, 171, and refs. therein), in carbon-rich or interstellar-dust environments, but not in dust produced by oxygen-rich objects. There is a direct relationship between the C/O ratio in planetary nebulae and the strength of the UIBs (Cohen et al. 198929). 4. The wavelength of the 11.3 m UIB shows that the hydrocarbons are not saturated with H. This band is due to the out-of-plane C-H bending, and occurs at 11.6 - 12.5 m if there are two C-H bonds on the same aromatic ring, and 12.4 - 13.3 m for three (Lger, d'Hendecourt, and Dfourneau 198994). The indicated amount of H coverage on the outer rings is 20 - 30%. Observational selection of relatively intense emission regions has meant that rather high radiation fields, and subsequent dehydrogenation, are favored; perhaps the 11-13 m emission from low-radiation environments will indicate more than one C-H bond on the same ring. 5. The bands are excited by the absorption of a single UV photon by the carrier. This is easy to understand if the carriers (planar PAHs or three-dimensional carbon structures no larger than about 5 ) float freely in space, so that a single photon can provide the energy required to emit the UIBs. The degree of excitation suggests that roughly 50 C atoms are required, with an upwards size range. If the carriers are attached to larger grains, the absorbed energy must be localized within a 5 region for the time required for the emission (of the order of a second). This process requires an exceedingly small thermal coupling. 6. The carriers of the UIBs are modified significantly by environment and history. The IRAS 12 m response shows that the UIBs are not present in regions of very high radiation fields (Ryter et al. 1987, Boulanger et al. 198813, 141), showing that the carrier can be modified or destroyed by intense radiation. The wavelength of at least the 7.7 m UIB is significantly different in planetary nebulae (where the carriers are newly produced by the carbon-rich material from the star) than in HII regions and reflection nebulae (where the carrier was presumably in the ISM before any interactions with the star presently causing the excitation). 7. PAHs would be mostly ionized in the diffuse ISM, since their first ionization potential is < 13.6 eV. Up to now, many laboratory studies of PAHs have, necessarily, involved only neutral molecules. 8. An individual PAH has strong discrete absorption bands in the visual through the UV, and there are no such features observed in interstellar extinction. A mixture of PAHs of varying sizes and structural arrangements produces continuous absorption. 3.2. Continuum emission Continuum radiation from dust arises from two mechanisms: (a) fluorescence, giving rise to a red continuum, and (b) thermal radiation, either in the 1-60 m range following the transient heating of a small grain by a single UV photon, or steady-state emission of larger grains for l > 100 m, or in the intermediate wavelength range where the effects compete. 3.2.1. Red continuum. Grains in the "reflection" nebula NGC 2023 were found to produce a red continuum peaking about 6800 (Witt, Schild, and Kraiman 1984178). Subsequent studies of many such nebulae confirm that there is an extended red emission in many of them (Witt and Schild 1988(175) and ref. therein), typically contributing 30 - 50% of the flux in the I-band (0.88 m). In NGC 2023, the red emission is found in filamentary structures which spatially coincide with patches of H2 infrared fluorescence and the 3.3 m UIB emission (Gatley et al. 198756), but not necessarily with the intensity of reflected light which reveals total dust density. IC 435, in the same molecular cloud, shows no red luminescence (Witt 1988174), nor does the Merope reflection nebula. A small patch of red fluorescent emission near the star g Cas also shows a strong H2 fluorescence in the UV (Witt et al. 1989179). The "Red Rectangle" (HD44179), a bipolar outflow excited by a central star, shows both strong UIBs and the red emission (Russell, Soifer, and Willner 1978; Schmidt, Cohen, and Margon 1980140, 146). The simplest interpretation of the red fluorescent emission is that it is excited by a strong UV flux incident upon hydrogenated carbon particles, either amorphous (Duley 198546) or PAHs (d'Hendecourt et al 1986b32), thereby producing both the fluorescent red emission and also H2 fluorescence. A strong enough radiation field will alter the carbon particles and dissociate the H2, as might be the case in the Merope nebula. 3.2.2. 3 - 30 m continuum. In addition to the UIBs, there is a continuous emission in the NIR/MIR range which accounts for a large part of the radiation from reflection nebulae (Sellgren et al 1985151) and, presumably, from the Galaxy as a whole. A survey at 11 and 20 m (Price 1981132) found unexpectedly high diffuse galactic emission. IRAS mapped almost the entire sky at 12, 25, 60, and 100 m and found the 12 and 25 m intensities vary considerably relative to the 100 m (Laureijs, Mattila, and Schnur 1987; Laureijs, Chlewicki, and Clark 198890, 91), which is roughly proportional to N(HI). The NIR/MIR emission arises from warm grains; simple considerations of the peak wavelength of the Planck function (and hence the emissivity of a grain) as a function of temperature show that this radiation must come from grains with temperatures of hundreds of Kelvins. By contrast, the mean local interstellar radiation field and optical constants for likely grain materials (carbon, silicates, organic refractory mantles) all suggest equilibrium temperatures of 20 K for the grains of the sizes (> 0.01 m) needed to account for the optical extinction. Grains of this size have large enough heat capacities so that their temperatures do not fluctuate appreciably after absorbing a single UV photon. The NIR/MIR emission must be produced by grains in the size range 5 - 50 , which are large enough to have an almost continuous density of energy states. In this case they radiate a continuum rather than emission bands. A single UV photon heats the grain to a high peak temperature which depends upon the size of the grain and the energy of the photon. The grain emits the NIR/MIR radiation and cools to very low temperatures between photon absorptions. Calculations of thermal fluctuations (Draine and Anderson 1985); Dsert et al 198634, 41) have shown that very small grains can account for the spectrum of the emission. The properties of small grains have been discussed clearly (Allamandola, Tielens,a nd Barker 1990, Puget and Lger (19894, 133). 3.2.3. FIR emission Grains with sizes of 0.01 m or more are cold ( 20 K) and reradiate most of the energy they absorb into the FIR part of the spectrum. Hildebrand (198370) has given a clear explanation of the process of determining grain properties and cloud masses from FIR observations. Cox and Mezger (198930), in a recent review of the galactic FIR/submillimeter radiation from dust, estimate that about 1010 LOAC(Sdo2(o),.), or 30% of the total luminosity of the Galaxy, is radiated in the FIR, mostly from dust heated in H I regions by the interstellar radiation field of early-type stars (Boulanger and Prault 198812). The FIR provides important information regarding the spatial variations of the interstellar radiation field throughout the Galaxy, and how molecular clouds form stars and evolve into H II regions, but is of limited use as a diagnostic of dust because each line of sight samples grains with temperatures which depend upon their local radiation fields. However, for wavelengths much longer than about 150 m, the peak of the emitted radiation, the FIR can be used to determine the relative opacities of the emitting grains. The Galaxy is optically thin at submillimeter wavelengths, in which case the intensity of emission, Il, is proportional to the opacity, kl, times the Planck function, Bl(T). For long wavelengths Bl(T) varies linearly with the temperature, so the wavelength dependence of Il provides relative values of kl. The results for diffuse radiation from the Galaxy (Matsumoto et al. 1988115) and from other galaxies (Chini, Krgel, and Kreysa 198922) show that for l > 100 m, kl is proportional to l-2, which is predicted by theory (e.g., DL). The constant in the proportionality is difficult to determine because it depends on estimating the column density of grains, or of hydrogen, along the line of sight of the observation. The theoretical opacities of DL, based on a graphite-silicate model for grains, are approximately half of Hildebrand's estimate (198370) based on a calibration in local dense globules (cf.Pajot et al 1989124). The uncertainties are probably at least a factor of two, if not more. Table 1 uses the Hildebrand value. The mass of interstellar dust , and thereby an estimate for the mass of the ISM, is often determined from the FIR intensity, together with the opacity of grains per mass and an estimated grain temperature. It is important to realize (Draine 199040) that this procedure is reasonably safe only if the observations are all on the long-wavelength side of the Planck function of the coldest grains: normally, in the submillimeter range. For instance, the temperature obtained from the ratio of the 60 and 100 m intensities from IRAS is biased by a few warm grains which can provide almost all of the 60 m emission. One can easily be off a substantial factor (>3) in the resulting mass estimate! In dust surrounding very young objects, bipolar outflows, or the cores of giant molecular clouds, the opacity even in the submillimeter range (400 - 1300 m) varies as l-1 or l-1.5 (Schloerb, Snell, and Schwarz 1987; Woody et al. 1989; Weintraub, Sandell, and Duncan 1989145, 162, 181) instead of l-2. Since the nebulae can hardly be so optically thick that radiative transfer effects are important at submillimeter wavelengths, these observations imply that the grains in these extremely dense objects are not extremely small in comparison to 1 mm. Cometary grains also extend up to this size range. The growth of fractal grains (Wright 1987182; see 8) would explain the observations. 4 scattering from dust Scattering from grains provides another diagnostic for their nature and composition, since cross sections for scattering at any angle (the "phase function") can be computed for a given material, much like they can be for extinction. However, the relative locations of the illuminating sources, scattering grains, and the observer are very important in determining the actual intensity of scattered radiation. The presence of fluorescent emission (3.2.1 above) complicates the interpretation of red and NIR scattering, but there seems to be little fluorescence in the range l < 0.5 m (Rush and Witt 1975139). In practice, the geometry is so uncertain that one attempts to determine only two quantities characterizing the scattering process: the albedo, or fraction of the extinction which is scattering, and g, the mean value of the cosine of the angle of scattering. For isotropic scattering, g = 0; for completely forward-throwing scattering, g = 1. Scattering can be observed in three general situations: (a) The "diffuse galactic light" (DGL), or scattering by the diffuse dust of the general incident interstellar radiation field. The DGL is strongly concentrated to the galactic plane, since the dust has a relatively small scale height. (b) Reflection nebulae, with a known source of illumination (usually an B or A star, because of their favorable luminosities). (c) Scattering of the general interstellar radiation field by a dark cloud, seen at high enough latitudes so that it contrasts with a relatively dark sky background. The DGL in the optical part of the spectrum has been analyzed (Witt 1968; Mathis 1973110, 173). It is quite faint and asymmetric in its angular distribution, and requires careful correction for the contribution of faint stars, airglow, and (especially) zodiacal light. Its advantage is that the geometry of the sources and scatterers is well known, in contrast to reflection nebulae. Witt (1988174) quotes a study (Toller 1981158) from Pioneer 10 at 3 AU (so that both the airglow and zodiacal light are negligible) which finds the albedo at 0.44 m is 0.61 0.07, and g = 0.60 0.22. Grains are very forward-throwing in the optical. In the UV, there are surprising spatial fluctuations in the diffuse brightness of the sky, with background intensities ranging over an order of magnitude at the same wavelength (cf. Paresce, McKee, and Bowyer 1980; Jakobsen et al 1984; Tennyson et al 1988, Murthy et al. 1988; Martin and Bowyer 198977, 103, 120, 127, 155). The minimum N(HI) in the sky corresponds to A(V) 0.05 mag (Lockman, Jahoda, and McCammon 1986100), or a scattering optical depth of 0.06 at 0.16 m. This translates into a sky brightness which is comparable to that observed. There may be an extragalactic component (which, in fact, is a common interpretation of the observed brightness). Clearly, the properties of grains, or the intensity of any extragalactic component, cannot be analyzed until the scattered intensity is known. Reflection nebulae, in which a bright star illuminates a nearby scattering cloud, have been analyzed in some detail by various authors. Their advantages over the DGL and scattering by clouds are that they are very much brighter than either and that they are reflecting light from well-studied early-type stars, while the others are illuminated by the much less well-known interstellar radiation field (especially at UV wavelengths). Reflection nebulae suffer from three disadvantages: (a) The geometry of the star and scattering particles is not well known, and the placement of a given grain relative to the star and observer is crucial for determining how much light it scatters into the observer's line of sight. Simple geometries used in modelling the scattering, such as plane-parallel or spherical, are not so appropriate as one would like. (b) The patchiness of the dust is more serious for the reflection nebula than for the DGL or the clouds. For clumpy objects, mean values of the albedo and g determined for the clump may not represent the values for a single scattering from an individual grain. (c) Some of the scattering is at large angles, in which case the assumed form of the phase function becomes important in the analysis. Astronomical observations of scattering have been interpreted by theoretical analyses employing the simple analytical Henyey-Greenstein (HG) phase function (Henyey and Greenstein 194169) which is computationally convenient but has no physical basis. The HG function is suitable for analyzing scattering at small to modest angles, in which case the important quantity is the fraction of the light thrown almost in the forward direction, but the HG function is not realistic for large-angle scattering. Whether the advantages of reflection nebulae outweigh the disadvantages is a matter of opinion. I feel the geometrical uncertainties are such that results should be taken with considerable caution. Prof. A. N. Witt (1988174, and ref. therein) disagrees, arguing that the brightest reflection nebulae must have a common geometry in which the intensity of scattered light is maximized. He also points out that in the UV, there might be three wavelengths at which the extinction optical depth is the same (two on each side of the bump and one more in the steeply rising part of the extinction law at very short wavelengths; see Figure 2). In this case, differences in the scattered light, relative to illuminating star's luminosity, directly reflect changes in the grains' albedo. His conclusion that g is smaller (more isotropic scattering) at l < 2000 than for longer wavelengths seems correct. Very possibly small, relatively isolated interstellar clouds ("globules") contain dust which is more similar to outer-cloud dust than to diffuse dust, but the determination of their optical properties is still of considerable interest. The geometry of scattering from globules at high enough latitude to be seen against a dark background (free from the DGL in the plane) is better known than for other reflection nebulae. At optical wavelengths, a cloud is seen strongly limb-brightened from scattered radiation from behind (FitzGerald, Stephens, and Witt 197652), which is a direct indication that grains are highly forward-throwing in the optical. If the scattering were isotropic, the center of the cloud (with the greatest optical depth) would be brightest. Mattila (1979116) determined albedo 0.6 and g 0.75 by comparing the brightnesses of two globules at different latitudes and assuming that their dust is intrinsically similar. 5 Polarization Interstellar polarization arises from the propagation of radiation through the ISM containing aligned elongated interstellar grains. The galactic magnetic field is responsible for aligning the grains, the grains spinning with their long axes perpendicular to the field. Under these conditions, radiation is subjected to extinction, to linear dichroism (differential linear extinction for the two waves polarized along and perpendicular to the direction of alignment), and to linear birefringence (differential phase shift between the two waves). The birefrigence produces circular polarization from the linearly polarized light, but to date the quantitative interpretation of the circular polarization has not been very fruitful because it depends upon the unknown geometry of the change of the direction of grain alignment along the line of sight. In principle, polarization is a diagnostic which provides another integral over the grain properties over the size distribution, similar to extinction and scattering. However, it involves an additional function which is poorly understood: the alignment of grains of various sizes (reviewed in Hildebrand 198871). Even so, polarization is important because it provides information regarding the optical properties of grains, and the conditions under which grains can be aligned. 5.1. continuum polarization 5.1.1. Optical and NIR The empirical wavelength dependence of optical/NIR polarization follows "Serkowski's Law", p(l)/p(lmax) = exp[Kln2(l/lmax)], with lmax the wavelength of maximum and p(max) the maximum polarization. The quantity K was originally taken to be 1.15, but an improved fit (Wilking et al 1980170) is K = 0.10 + 1.86lmax. This law is entirely empirical, and it would be important to determine deviations at large |ln(l/lmax)|. Salient features of the optical polarization law are: 1. The averaged value of lmax is 0.55 m, with extremes from about 0.34 m to about 1 m. The values of lmax, determined by a least-squares fitting to Serkowski's law rather than a direct search for the maximum, depend mainly upon the observations at extreme wavelengths. 2. The polarization law is broad, typically rising with wavelength through the ground-based UV to a maximum in the optical, and then falling slowly through the NIR. Such behavior bears little resemblance to the extinction law, which keeps rising monotonically, except for the bump, towards shorter wavelengths throughout the observable UV. The grains responsible for the extinction in the ground-based UV do not participate in polarization because they are not elongated, or not aligned, or both. 3. The value of lmax is almost proportional to RV (Whittet and van Breda 1978, 1980; Clayton and Mathis 198825, 167, 168), although there is more scatter in the relationship than for the extinction laws. To a large extent, optical polarization measurements can substitute for NIR extinction in obtaining RV. 4. The polarization law, p(l), varies as p(l) l-1.8 for both diffuse dust and outer-cloud dust for 0.9 m < l < 5 m (Martin and Whittet 1990108). The exponent is less well determined than for extinction, varying between 1.5 and 2.0 for various samples, but it is certainly similar to the value for extinction (1.7 - 1.8; see 2.1.3). Note that this relation involves the absolute polarization, not relative to p(lmax). The optical p(l) does vary strongly with RV (see point 2 above), and the silicate feature has strong polarization which dominates for l > 5 m. The independence of p(l) from RV again suggests that the size distribution of large grains is similar for clouds and the diffuse ISM. 5. The maximum value of p(lmax)/A(lmax) is about 0.03 mag-1, far less than from perfectly aligned spinning cylinders (0.22 mag1). This is interesting because the polarization direction closely follows the contours of the edges of several molecular clouds, presumably in regions where hydrogen changes its state from molecular to atomic relatively abruptly in space and perhaps in time. If the alignment mechanism keeps grains aligned under these adverse conditions, one would expect almost complete alignment when conditions are favorable, and a larger value of p(lmax)/A(lmax) than observed, in directions where the line of sight is perpendicular to the field. Perhaps there are two or more separate types of grains, only some of which are aligned. Alternatively, all grains might be well aligned but have shapes which are less efficient for producing polarization than a spinning cylinder. A third possibility is that there is always a randomly oriented component to the field. 6. Polarization in the UV is unknown except for two stars (Gehrels 197458). These limited data suggest that the bump is unpolarized. Upcoming observations from space (the WUPPE experiment on ASTRO missions) should provide a great deal of data. The polarization of the l2175 bump has been predicted if the bump is produced by aligned graphite (Draine 198836). An explanation for the form of the polarization law (Mathis 1986a111) assumes that grains can be aligned only if they contain one or more "superparamagnetic" particles (magnetite or other magnetic materials) which dissipate rotational energy as heat. Large grains are preferentially aligned because they are relatively likely to contain inclusions. Polarization is not specific to any particular grain model; if the large grains are aligned, and a model predicts the extinction correctly, it will do well for the polarization also. 5.1.2. fir polarization Polarization is observed in the emission from grains deep within the Orion molecular cloud and Sgr A, near the galactic center (Hildebrand, Dragovan, and Novak 1984, Dragovan 1986, Werner et al. 198835, 72, 163). The direction of the FIR polarization is perpendicular to the optical, exactly as expected: light transmitted in the optical is polarized in the direction of smaller absorption. Emission, on the other hand, is largest in the direction of largest absorption. Grain alignment is even more difficult for dense clouds than for the diffuse ISM (Hildebrand 198871). Alignment depends upon the grain being far from equilibrium with its surroundings, in which case there is no preferred axis by the equipartition of energy. Deep inside a cloud, a grain should come to thermal equilibrium with the dense surrounding gas. The observed polarization shows that the rotation of aligned grains within clouds is not thermalized; they are presumably kept spinning in a particular direction because of the ejection of particles (H2 after formation, or electrons) from particular sites (Purcell 1979134), so the momentum of the ejected particles is not random. However, deep inside a cloud one would expect the gas impinging upon the grain to already be overwhelmingly H2. Probably there also needs to be an enhanced dissipation of energy by superparamagnetic inclusions in the grains. 5.2 polarization in spectral features Inner-cloud dust provides information about diffuse dust through the polarization in the 3.08 m ice band and the 9.7 and 18 m silicate bands. If the polarization arises from aligned grains, the profiles of polarization, as compared to the corresponding extinction, are potentially important diagnostics for both the shapes and types of grains (Martin 1975DL; 105). If the sizes and optical properties of interstellar grains are such that the light wave acts as a uniform electromagnetic field across the particle (the "Rayleigh approximation"), the extinction and polarization cross sections are both simple functions of the optical constants of the material. The fundamental Kramers-Krnig relations force a relationship between the optical constants and, therefore, the extinction and polarization. When there is a strong frequency variation of the optical constants, as across a spectral band, the polarization peaks at longer wavelengths than the extinction. The silicate polarization indeed peaks at about 10.5 m and the extinction at 9.7 m (Aitken et al 1985, 19881, 2). The amount of shift depends upon the shapes of the grains as well as the optical constants. If the band is strong enough, there is a polarization reversal at wavelengths on the short side of the maximum polarization. The presence of coatings also affects the shape of the polarization relative to the extinction, even if the coatings have a weak wavelength dependence of optical constants across the band. Polarization can be produced by scattering in the NIR as well as by extinction by aligned grains. Such scattering, commonly observed in reflection nebulae around sources in dense, star-forming regions (Lenzen et al 1984; Pendleton et al 1986; Pendleton, Tielens, and Werner 199095, 129, 130), shows that grains in very dense regions have grown to sizes of at least the order of one micron, partly by acquiring the ice coatings producing the 3.08 m band. In the Beckin-Neugebauer object (BN) object in Orion (see Lee and Draine 198593 for references), the linear polarization is strong (16%) in the 3.08 m ice band, as opposed to 10% in the neighboring continuum. The polarization is 15% in the 9.7 m silicate feature while 1% in the continuum. The position angle is constant across the bands, and is the same as in the nebula in general, as expected from grain alignment but not from scattering. However, there is a reflection nebula around BN (see references in Pendleton, Tielens, and Werner 1990129) which complicates the interpretation of the polarization. By assuming that the polarization in BN is entirely from aligned grains, Lee and Draine (198593) found grains are oblate (disk-shaped) rather than prolate, with modest (2:1) axial ratios. One consequence of oblate grains is that the degree of alignment required is dropped by a factor of two, which is the ratio of the mean polarizing power of oblate and prolate grains. A potentially powerful diagnostic for grains is the comparison of the polarizations in the 9.7 m and 18 m bands, for which many of the geometrical uncertainties, such as angles of the magnetic field relative to the line of sight to the source, cancel. 6 other diagnostics of interstellar dust 6.1. Depletions in interstellar gas The strengths of the interstellar absorption lines of various ions in the spectra of background stars indicate the ionic column densities in the gas phase, and by inference the amounts of the elements in grains. Jenkins (198778) reviews the method, results, and several pitfalls. The principal results of depletions as regards grains are: 1. The elements O, N, and Zn are slightly depleted with respect to solar abundances, and their depletion does not vary measurably between dense and diffuse gas. The errors are such that they could be undepleted, or depleted by a factor of two. 2. Depletions of other elements increase significantly with average gas density along the line of sight. The elements P, Mg, and Cl are almost undepleted in diffuse gas ( 0.1 cm3), and are depleted by about an order of magnitude when 10 cm3. 3. The elements Fe, Cr, and Si are depleted about one order of magnitude in the diffuse ISM, and the depletions go roughly as the square root of the mean gas density, so that their depletion is about a factor of 10 more than P, Mg, and Cl, over a wide range in mean gas density (80, Joseph 1988). Thus, these depletions are two orders of magnitude along lines of sight through dense gas. This difference in depletion between outer-cloud dust and diffuse dust implies that grains evolve as they go from one environment to another. Ca, Ti, and Al have similar depletions in low-density gas but have a steeper dependence of depletion on mean gas density. 4. The depletion of C is, unfortunately, not reliably determined except for one line of sight, towards d Sco with the Copernicus satellite (Hobbs, York, and Oegerle 198273). For observations with IUE all lines of C+, the dominant ionization stage in H I regions, lie on the "flat" (insensitive) portion of the curve of growth. For d Sco, about 25% of the C is in the gas phase, which is about the amount found in CO in molecular clouds. The gas-phase carbon in the diffuse ISM is apparently simply converted to CO in molecular clouds, while the solid fraction remains approximately fixed. There is some confusion in the literature because the C depletion can be crudely estimated from the well-determined gas-phase abundance of neutral C by estimating C0/C+. Unfortunately, the ionization corrections are factors of hundreds to thousands! 5. At a given mean gas density, there is a surprisingly small dispersion of the depletions (about 0.3 dex), while some depletions vary one or two orders of magnitude. Much of this dispersion must arise from averaging various local conditions along the line of sight. Again we see that the state of grains must be quite well described by only one parameter (perhaps local gas density). 6. Depletions are a function of z, the height above the plane of the Galaxy (Edgar and Savage 198950). Probably the extinction law also depends on z (Kiszkurno-Koziej and Lequeux 198783). 6.2. Sites of dust formation Probably most dust is injected into the ISM from stars on the asymptotic giant branch, either C-rich or O-rich, although supernovae, in spite of a low rate of mass injection, might be important because of their large heavy-element composition. Carbon stars show an 11.15 m feature of SiC with a quite uniform profile (Cohen 1984, Little-Marenin 198626, 99). There is also a featureless optical/NIR continuum which can be modelled in terms of amorphous carbon but not graphite (Martin and Rogers 1987; Le Bertre 198792, 106), unless the graphite is in small particles with a loose fractal structure (Wright 1989183). Oxygen-rich stars show the silicate feature with a profile which varies appreciably from star to star (Papoular and Pegourie 1983, Little-Marenin 198699, 126). Differing physical conditions in the atmosphere affect the nature of the grains, and the silicate band in one star (Aitken et al. 19882) shows clear signs of annealing. A few circumstellar shells show the 3.08 m ice band as well. The UV bump is found in C-stars at 0.24 m, not 0.2175 m (Hecht et al 198467), and in one H-poor, C-rich planetary nebula (Abell 30) at 0.25 m (Greenstein 198163). These shifted wavelengths are consistent with amorphous C or fractal graphite grains. There is no bump in the circumstellar dust of oxygen-rich a Sco (Snow et al 1987153). The bump is seen in circumstellar dust surrounding a few hot stars (Sitko, Savage, and Meade 1981152) at the normal lmax, but the dust might be interstellar, remaining from the epoch of the star's formation. Novae and Wolf-Rayet stars (late-type WCs) are minor sources of dust because of their low mass input into the ISM. Both presumably inject carbon-rich dust. Planetary nebulae represent a considerable source of the return of gas to the ISM (Maciel 1981102), but are not a large source of dust because they have a low dust/gas ratio (Pottasch et al. 1984131). Isotopic anomalies in meteorites (see below) prove that some dust forms in expanding supernova shells. Dust is also found within hot supernova remnants (Dwek et al 1987a,b; 48,49), but it might have been produced by a pre-supernova red supergiant. How much dust is actually formed in supernovae is not known. 6.3. Solar system dust Primitive meteorites, interplanetary dust particles (IDPs), and cometary dust provide some information regarding interstellar grains, although all solar-system dust has been significantly processed, both chemically and physically, since having been in the pre-solar system molecular cloud. The various types of objects have been reviewed (Kerridge and Mathews 1988; Brownlee (198914, 15, 82). Meteorites are almost entirely asteroidal in origin, since cometary meteoroids cannot survive the entry into the earth's atmosphere. Most meteorites have undergone obvious metamorphism, with carbonaceous chondrites being most primitive. One of this class, Murray, has tiny SiC inclusions showing isotopic anomalies (Bernatowicz et al. 1987, Zinner et al. 19878, 186) proving an interstellar origin. The small amount of carbon in primitive meteorites is mostly poorly ordered, not graphitic, but tiny (ca. 25 ) diamonds are present (Lewis et al 198797). It is difficult to imagine a solar-system origin for these, so diamond bonding in interstellar carbon is strongly indicated. Meteoritic silicates are much more crystalline than the interstellar varieties, showing that meteorites must have been much warmer than interstellar temperatures. Graphite, if originally present, could have been lost by chemical reactions with water and hydrogen. IDPs are both cometary and asteroidal in origin, with the cometary being more primitive and, therefore, relevant to interstellar dust. Some cometary IDPs, collected from high-flying aircraft, are a few microns in size, with a very fragile, open structure consisting of submicron mineral grains, stuck together into an open matrix (see e.g. Brownlee 1985, 198914, 15). Their silicates have been annealed to crystalline forms (Sandford and Walker 1985143), with types of minerals similar to those observed in cometary dust. Some of the carbon, poorly ordered but aromatic in nature, occurs as submicron grains. There is also carbon coated onto the silicate materials. Some IDPs have relatively large D/H ratios (Zinner, McKeegan, and Walker 1983185), suggestive of molecular clouds. The crystalline structure of cometary silicates shows that even cometary dust is fairly heavily processed in comparison to interstellar. There are whole particles consisting of volatile material containing C, H, O, and N ("CHON"), with little refractory material within; there are also low-density silicate and carbonaceous grains. Meteoroids, mostly cometary, entering the upper atmosphere have densities of 0.01 - 1 gm cm3, so the fluffy refractory particles are common in comets and are probably present in interstellar dust at least deep within molecular clouds. In summary, solar system material shows that (a) small particles can survive all of the rigors of the ISM after formation in a supernova; (b) grains form large structures deep inside molecular clouds, with voids possibly packed with ices; and (c) there was heating and associated chemical processing which took place before the formation of comets, possibly in the molecular cloud material. 7 evolution of dust Relevant timescales show that the present form of interstellar dust must be more a reflection of the processing it has received within the ISM than of the conditions at its origin. A typical parcel of gas and dust is cycled back and forth through molecular clouds several times during its lifetime, changing its grain properties significantly each time. The lifetime of a grain against incorporation into stars can be estimated by dividing the surface density of the ISM by the rate of star formation. The local H I surface density is 107 MOAC(Sdo3(o),.) kpc-2 (Kulkarni and Heiles 198789), and 3 x 106 MOAC(Sdo3(o),.) kpc-2 of H2 (Scoville and Sanders 1987147). A mean rate of star formation of 3.4 x 10-3 MOAC(Sdo3(o),.) kpc-2 yr-1 would account for the present surface density of low-mass (M MOAC(Sdo3(o),.)) stars over 1010 yr (Bahcall and Soneira 19806); presumably the present rate is lower. High-mass stars contribute about 1.1 MOAC(Sdo3(o),.) kpc2 yr-1 (Ratnatunga and van den Bergh 1989135). The corresponding mean lifetime of a parcel of gas/dust in the ISM is > 3 Gyr. On the other hand, about 30% of the local ISM is in molecular clouds, each with a lifetime of perhaps 108 yr (the time for the gas to proceed from one spiral arm to the next) or less. These numbers imply that a given parcel of gas has been into and out of a molecular cloud at least every 3 x 108 yr, or > 10 times during its mean lifetime Each time, the differences in extinction laws between diffuse- and inner-cloud dust require that the grains be heavily modified. Let us follow the state of a typical parcel of gas/dust near the Sun (see also Draine 199039). Since most of the mass of the local ISM is in H I, the parcel must spend the bulk of its time outside of molecular clouds. During this phase, about 10% of its mass is returned to the ISM from stars. Perhaps 10 - 20% of the returned gas is from hot stellar winds with no dust, or from planetary nebulae with a dust/gas ratio lower than the ISM, providing substantial amounts of gaseous Fe, Al, and the other strongly depleted elements (Jura 198781). Refractory elements must encounter grains frequently and stick to them very efficiently in order to achieve the observed strong depletions in the denser parts of the diffuse ISM. Each time the gas/dust mixture is incorporated into the outer regions of a molecular cloud, many things happen to the grains: (a) the extinction law changes from diffuse dust to outer-cloud dust in such a way that the relative numbers of small, medium, and large grains depend primarily upon only one parameter (local gas density?), regardless of the local environment or past history. (b) The refractory elements are more strongly depleted onto the grains than in the diffuse phase. (c) The atomic H becomes molecular. Gaseous carbon recombines from C+ to C0, and finally to CO. (d) The grains almost certainly coagulate in the outer parts of molecular clouds before there is much coating of icy mantles. Much deeper in the cloud, the fluffy micron-sized cometary and interplanetary dust particles can be produced by further coagulation of the silicate and carbonaceous parts of the grains, with ices probably filling the voids and producing the observed molecular absorption features. It is difficult for me to see how fluffy cometary mineral grains can form within the cloud if icy or organic refractory mantles envelop the minerals before the coagulation. Coagulation seems to dominate the change in the size distribution in going from the diffuse ISM to outer-cloud dust. At least two well-observed stars in outer-cloud dust have A(V)/N(H) smaller than in diffuse dust (CCM). Since [A(l-1)/A(V)] dl-1 is much lower in the outer-cloud dust than in diffuse dust (Figure 2), in these stars the integrated extinction per H atom is substantially smaller than in diffuse dust. This reduction in grain cross-section per H, in spite of the accretion of small amounts of the refractory elements, can only be achieved by coagulation, which prevents the inner parts of the larger grains from participating efficiently in the extinction. Adding substantial coatings would increase, rather than reduce, the extinction cross-section per H for the dust in the cloud. There are real deviations of the various extinction laws about the mean. These differences must reflect somewhat different histories and present local environments (radiation field, shocks, magnetic fields, etc.) of the grains along each line of sight. Studying these deviations should lead to a better understanding of the factors which influence extinction laws. 8 Some theories and problems with each Space does not allow even a superficial discussion of many individual theories interpreting the observational evidence summarized above. We summarize a few. 1. Bare silicate/graphite grains. DL, in a very careful discussion of the optical constants of both graphite and silicates, greatly extended an older theory of Mathis, Rumpl, and Nordsieck 1979((114, often referred to as "MRN") in wavelength. The features of the theory are: (a) Individual grains are bare and homogeneous, composed of either silicate or graphite. (b) The size distribution is a power-law, where n(a), the number of grains of radius a, is proportional to a-3.5. (c) The size distribution is truncated at the upper end at 0.25 m. The lower end of sizes must be extended downward to a few ngstroms to fit the IRAS data (Draine and Anderson 1985; Puget and Lger 198941, 133), and very likely all the way to PAHs in order to produce the UIBs. DL subjected their theory to a much more exacting comparison with observations than did most other authors. The fit to the extinction law for diffuse dust, over the entire observed wavelength range from 0.1 m to 1000 m, is impressively good. On the other had, DL requires that large graphite particles be produced from the amorphous carbon late-type stars inject into the ISM. It is very difficult to understand how the necessary annealing can take place under interstellar conditions. Furthermore, the materials in the two types of grains (silicates and graphite) must be kept separate, in spite of the many cycles of coagulation and rearrangement of the size distributions which take place as the grains cycle into and out of clouds. 2. Core/mantle grains. Greenberg and his collaborators (see refs. in Greenberg 198962) believe that the bulk of interstellar grains have refractory silicate cores covered with an organic refractory mantle. This mantle is produced from the processing by both UV photons and cosmic rays deep inside molecular clouds, after the icy mantles observed in such clouds are deposited upon the grain surfaces. Laboratory studies show that molecules in such mantles can be partially converted into free radicals which are chemically active enough to react violently when warmed, producing complex molecules (d'Hendecourt et al. 1986a31). Such runaway reactions could be triggered on inner-cloud grains by cosmic rays. After such an event, a organic polymeric substance known as "organic refractory" material, stable at room temperature, remains. In this theory, organic refractory mantles on silicate cores produce the optical/NIR extinction, while the bump is produced by small graphite particles, PAHs produce the UIBs, and the shortest-wavelength extinction is from small particles, probably silicates. Chlewicki and Laureijs (198823) suggest that an additional component of iron will produce most of the 60 m emission observed by IRAS. These ideas have a great deal of appeal as regards events deep within clouds. The Greenberg scenario explains why interstellar molecules are found in the gas phase within dense clouds, where they should freeze onto grains in a short time the runaway reactions drive off the molecules, and gas-phase chemistry takes place before re-freezing onto grain surfaces. An alternative explanation of gas-phase molecules inside dense clouds is a very rapid circulation of grains between the surface and the center of the cloud (e.g., Chize and Pineau des Forts 198921). There are four problems as regards extending these ideas into diffuse dust: (a) Grains are larger in outer-cloud dust because of coagulation, not accretion of mantles, as shown by the reduced extinction per H atom in some cases. (b) Organic refractory mantles, which are less refractory than silicates and solid carbon, would be more readily destroyed by shocks in the diffuse ISM. The destruction rate of materials depends sensitively upon the binding energy (Draine and Salpeter 1979a, 1979b43, 44). (c) Solar system dust particles suggest that the silicate and carbon materials coagulate into large structures before icy mantles envelop them. (d) The 3.4 m C-H absorption band, seen in organic refractory material in the laboratory along with the 3.08 m ice band, is locally weak or absent. The object with the strongest 3.4 m band, IRS 7 near the galactic center (Butchart et al. 198617), is not typical of local dust. The 3.08 and 3.4 m bands are not yet seen towards the local star VI Cyg 12, A(V) 10 mag (Gillett et al. 197561), limiting these bands to < 0.3 the strength, per A(V), as IRS 7. The 3.4 m band is seen in Lynga 8/IRS 3 (Tapia et al. 1989154, A(V) = 17), about 0.4 times as strong as IRS 7. The 3.4 m band is always accompanied by a stronger 3.08 m ice feature, and there are some lines of sight (Harris, Woolf, and Rieke 197864) with no ice band for A(V) < 20 mag. It is possible (Tielens 1989156) that the organic refractory mantles are so heavily processed that they lose almost all of their N and O, becoming essentially amorphous carbon. This material would be difficult to distinguish from amorphous C injected directly into the ISM from carbon stars. In this case, the Greenberg theory is very similar to composite-grain theories (see below). 3. Silicate cores with amorphous carbon mantles: Duley, Jones, and Williams (198947) suggest that grains are silicates with mantles of hydrogenated amorphous carbon (HAC). One population is very small and produces the bump by (OH)- ion absorption in the presence of Si atoms (all other theories produce the bump from well-ordered carbon). The UIBs are caused by absorption of UV photons by "islands" of HAC on the silicate core surfaces, so thermally isolated that they can radiate like free particles (for about 1 second!). The rapid increase of extinction with wavenumber for l-1 > 6 m-1 is produced by diamond-like bonding in the "amorphous" C. This theory makes several predictions which can be tested. It explains the differences between diffuse dust and outer-cloud dust rather naturally, as arising from different depletions of carbon onto the silicate cores. However, it requires a very large fraction of the Si atoms to have OH- ions nearby, near the surfaces of small grains, even if the bump transition in OH- has an oscillator strength of unity. The thermal isolation of the "islands" of HAC is difficult to achieve. 4. Composite grains: Mathis and Whiffen (1989113) and Tielens (1989156) suggest that interstellar grains consist of an assembly of small particles of carbon and silicates, jumbled together loosely. These grains are the natural result of coagulation and disruption of grains as they cycle into clouds. The particles inside the porous structure are protected from shocks, and might well be covered with highly processed organic material. The bump is provided by small graphitic particles; PAHs can produce the UIBs. The rise in extinction for l < 0.16 m is provided by the diamond bonding in "amorphous" C. The composite grains are mostly open, in analogy with interplanetary dust particles. However, too much porosity provides too large an opacity in the FIR, making the grains too cold because they radiate efficiently. It is difficult to calculate the extinction of composite grains, so the calculated fit should be taken as provisional. 5. Fractal grains: Wright (1987182) suggested that interstellar grains are the product of coagulation into very large fractal structures resembling twisted branches (Hawkins and Wright 198865). If one defines the fractal dimension, a, by mass Ra, then a depends upon the sequence of coagulation (and the probability of the fractal grains' breaking up, neglected in the calculations). In general a < 3, and in some cases a < 2. One of the major features of fractal grains is an FIR absorption per unit mass larger by an order of magnitude or more over solid grains. Fractal grains can explain very large radar backscattering in comets (Wright 1989183) without large masses of dust. They also explain the very shallow (l-1) dependence of the opacity submillimeter opacity observed in some very dense nebulae (3.2.3 above). However, the FIR opacity of fractal grains is so large that the grains would be too cold to explain the observed FIR spectrum of galactic dust (T 20 K). 6. Biological grains. Hoyle, N.C. Wickramasinghe, and others (Jabir et al 1986, Wallis et al. 198976, 161, and references therein) have suggested that the grains producing visual extinction have a biological origin, with the bump provided by graphite. The extinction and polarization laws are fitted reasonably well. There are two problems with the model: (a) There is not enough cosmic phosphorous to accommodate the amount found in organisms (Whittet 1984; Duley 198445, 164, but see Hoyle and Wickramasinghe 198474). The cosmic abundance of P is low, and most of it is in the gas phase for low-density lines of sight, so this criticism seems valid. (b) Organisms, even when dried, show strong O-H and C-H stretch absorptions (Hoyle et al. 198275) which are not seen except deep within molecular clouds. 9 Summary "Interstellar dust" refers to materials with rather different properties, and the "mean extinction law" of Seaton (1979149) or Savage and Mathis (1979144) should be replaced by the expression given in Cardelli, Clayton, and Mathis (198920), using the appropriate value of total-to-selective extinction, RV. The older laws were appropriate for the diffuse ISM, but dust in clouds differs dramatically in its extinction law (Figure 2). However, there are certainly real deviations from the mean extinction law (see error bars in the inset in Figure 2). The extinction law for l > 0.9 m seems to be independent of environment, to within the present observational errors. Other diagnostics of dust, especially the depletions from the gas phase, confirm that properties of the grains vary along various lines of sight, but only one parameter, probably related to the local gas density, determines the grain properties surprisingly well. Dust is heavily processed while in the ISM by being included within clouds and cycled back into the diffuse ISM many times during their lifetimes. Consequently, grains probably reflect only a trace of their origin, although meteoritic inclusions with isotopic anomalies prove that some tiny particles survive intact from a supernova origin to the present. Grains apparently grow by coagulation while in clouds. Cometary and interplanetary dust suggest that very large sized grains are produced before extensive icy mantles are formed. Within the dark clouds, there is likely processing of the icy mantles by cosmic rays or the UV radiation produced by cosmic rays, and heavy molecules are released by runaway reactions. If there is an organic refractory mantle remaining after this processing, it is probably converted to almost pure amorphous carbon by the continued processing to which the grains are subjected. There are several theories which explain the extinction law for diffuse dust, but a much more challenging problem is to understand the relation between dust of all types. The evolution of dust is probably the next theoretical challenge. This review has been partially supported by contract 957996 with the Jet Propulsion Lab and grant NAGW-1768 with NASA. Comments and assistance from L.J. Allamandola, J.A. Cardelli, G.C. Clayton, B.T. Draine, P.G. Martin, and B.D. Savage are appreciated. appendix: The phases of the interstellar medium The density, velocity, magnetic field strength, and temperature of gas within the interstellar medium vary continuously with position and fluctuate over all scale lengths of interest. However, various regimes of temperature and density differ markedly in the nature of their basic physical processes, especially heating and cooling. It is convenient to refer to these regimes as specific phases of the interstellar medium. A summary of the physical conditions and physical processes in thise phases is given in the table below. The primary discriminent in phases is the condition of the hydrogen. Another important property is the temperature. The gas pressure follows from the density and temperature. Name State of HTypical n (cm3)T (K) (gas)HeatingCoolingHow observedRemarksMolecular cloudsH2 >100010 - 80Cosmic raysCO, Far-IR from dust CODust has icy mantlesH I cloudsH30100Photo-electrons from dust [C II] (158 m)21-cm (emission, absorp)Diffuse ISMWarm H I H0.1800021-cm emission Warm H IIH+0.03104Photo-ionization of H [O II], [S II]Ha,[Sdfo3()II], nebular linesVery faint but ubiquitousHot ISMH+10-3106.5Shocks from SNe X-raysX-raysLittle massH II regionsH+>100104H photo-ionization[O III] Ha, radio, other linesExpanding, transient Super-nova remantsH+(varies)107ShocksX-rays, IR from dustOptical, X-rays, IRASDynamic! Literature cited 1. Aitken, D.K., Bailey, J.E., Roche, P.F. and Hough, J.H. 1985. MNRAS, 215: 815. 2. Aitken, D.K., Roche, P. F., Smith, C. H., James, S. D., and Hough, J. H. 1988. MNRAS, 230: 629. 3. Allamandola, L.J., and A.G.G.M. Tielens, eds.. 1989. IAU Symposium 135, Interstellar Dust. Dordrecht: Reidel Publishing Company. 530 pp. 4. Allamandola, L.J., Tielens, A.G.G.M., and Barker, J.R. 1989. Ap.J. Suppl., 71, in press. 5. Ardeberg, A., and Virdefors, B. 1982. Astron. Astrophys., 115: 347. 6. Bahcall, J.N., and Soneira, R.M. 1980. Ap. J. Suppl . 44: 73. 7. Bailey, M.E., and Williams, D.A., eds. 1988. Dust in the Universe , Cambridge: Cambridge University Press, 573 pp 8. Bernatowicz, T., Fraundorf, G., Tang, M., Anders, E., Wopenka, B., Zinner, E., and Fraundorf, P. 1987. Nature, 330: 728. 9. Bohlin, R. C., Savage, B. D., and Drake, J. F. 1978. Ap. J., 224: 132. 10. Borghesi, A., Bussoletti, E., and Colangeli, L. 1987. Ap. J., 314: 422. 11. Bouchet, P., Lequeux, J., Maurice, E., Prvot, L., and Prvot-Burnichon, M.L. 1985. Astron. Astrophys., 149: 330. 12. Boulanger, F., and Prault, M. 1988. Ap. J., 330: 964. 13. Boulanger, F., Beichman, C., Dsert. F.X., Helou, G., Prault, M., and Ryter, C. 1988. Ap. J., 332: 328. 14. Brownlee, D.E. 1985. Ann. Rev. Earth Planet. Sci., 13: 147. 15. Brownlee, D.E. 1989. Highlights of Astronomy, 8: 281. 16. Butchart, I., and Whittet, D. C. B. 1983. MNRAS, 202: 971. 17. Butchart, I., McFadzean, A.D., Whittet, D.C.B., Geballe, T.R., and Greenberg, J.M. 1986. Astron. Astrophys., 154: L5. 18. Cardelli, J. A., and Savage, B. D. 1988. Ap. J., 325: 864. 19. Cardelli, J. A., Clayton, G. C., and Mathis, J. S. 1988. Ap. J. Lett., 329: L33 . 20. Cardelli, J. A., Clayton, G. C., and Mathis, J. S. 1989. Ap. J., 345: 245. (CCM) 21. Chize, J.P., and Pineau des Forts, G. 1989. Astron. Astrophys., 221: 89. 22. Chini, R., Krgel, E., and Kreysa, E. 1989. Astron. Astrophys., 216: L5. 23. Chlewicki, G., and Laureijs, R.J. 1988. Astron, Astrophys., 207: L11. 24. Clayton, G. C., and Martin, P. G. 1985. Ap. J., 288: 558. 25. Clayton, G. C., and Mathis, J. S. 1988. Ap. J., 327,: 911. 26. Cohen, M. 1984. MNRAS, 206: 137. 27. Cohen, M., Allamandola, L. J., Tielens, A. G. G. M., Bregman, J., Simpson, J., Witteborn, F. C., Wooden, D., and Rank, D. 1986. Ap. J., 302: 737. 28. Cohen, M., Tielens, A. G G. M., and Bregman, J. D. 1989. Ap. J. Lett., 344: L13. 29. Cohen, M., Tielens, A. G. G. M., Bregman, J., Witteborn, F. C., Rank, David M., Allamandola, L. J., Wooden, D. H., and de Muizon, M. 1989. Ap. J., 341: 246. 30. Cox, P., and Mezger, P.G. 1989. The Astron. Astrophys. Rev., 1: 49. 31. d'Hendecourt, L.B., Allamandola, L.J., Grim, R.J.A., and Greenberg, J.M. 1986. Astron. Astrophys., 158: 119. 32. d'Hendecourt,L.B., Lger, A., Olofsson, G., and Schmidt, W. 1986. Astron. Astrophys., 170: 91. 33. Day, K.L. 1979. Ap. J., 234: 158. 34. Dsert, F.X., Boulanger, F., and Shore, S.N. 1986. Astron. Astrophys., 160: 295. 35. Dragovan, M. 1986. Ap. J., 308: 270. 36. Draine, B.T. 1988. Ap. J., 333: 848. 37. Draine, B.T. 1989, in Infrared Spectroscopy in Astronomy, Proc. of 22nd ESLAB Symposium. 38. Draine, B.T. 1989, in proceedings of IAU Symposium 135, Interstellar Dust , ed. L.J. Allamandola and A.G.G.M. Tielens, eds., Dordrecht: Reidel, p. 313. 39. Draine, B.T. 1990, in The Evolution of the Interstellar Medium, ed. L. Blitz, San Francisco: Publ. Astr. Pacific Press, in press. 40. Draine, B. T. 1990, in The Interstellar Medium in Galaxies, ed. H.A. Thronson and J.M. Shull, Dordrecht: Kluver, in press. 41. Draine, B.T., and Anderson, N. 1985. Ap. J., 292: 494. 42. Draine, B.T., and Lee, H.M. 1984. Ap. J., 285: 89. (DL) 43. Draine, B.T., and Salpeter, E.E. 1979. Ap. J., 231: 77. 44. Draine, B.T., and Salpeter, E.E. 1979. Ap. J., 231: 438. 45. Duley, W. W. 1984. Q. J. Roy. Astr. Soc., 25: 109. 46. Duley, W.W. 1985. MNRAS, 215, 259. 47. Duley, W.W., Jones, A.P., and Williams, D.A. 1989. MNRAS, 236: 709. 48. Dwek, E., Dinerstein, H.L.,Gillett, F.C., Hauser, M.G, and Rice, W.L. 1987.Ap. J., 315: 571. 49. Dwek, E.,Petre, R., Szymkowiak, A., and Rice, W.L. 1987. Ap. J. Lett., 320: L27. 50. Edgar, R.J., and Savage, B.D. 1989. Ap. J., 340: 762. 51. FitzGerald, M.P. 1968. Astron. J., 73: 983. 52. FitzGerald, M.P., Stephens, T.C., and Witt, A.N. 1976. Ap. J., 208: 709. 53. Fitzpatrick, E. L. 1985. Ap. J., 299: 219. 54. Fitzpatrick, E. L., and Massa, D. 1986. Ap. J., 307: 286 (FM86). 55. Fitzpatrick, E. L., and Massa, D. 1988. Ap. J., 328: 734. 56. Gatley, I., Hasegawa, T., Suzuki H., Garden, R., Brand, P.W.J.L., Lightfoot, J., Glencross, W., Okuda, H., and Nagata, T. 1987. Ap. J. Lett., 318: L73. 57. Geballe, T.R., Tielens, A.G.G.M., Allamandola. L.J., Moorhouse, A., and Brand, P.W.J.L. 1989. Ap. J., 341: 278. 58. Gehrels, T. 1974. Astron. J., 79: 590. 59. Giard, M., Pajot, F., Lamarre, J.M., Serra, G., and Caux, E. 1989. Astron. Astrophys., 215: 92. 60. Gillett, F.C., Forrest, W.J., Merrill, K.M., Capps, R.W., and Soifer, B.T. 1975. Ap. J., 200: 609. 61. Gillett, F.C., Jones, T.W., Merrill, K.M., and Stein, W.A. 1975. Astron. Astrophys., 45: 77. 62. Greenberg, J.M. 1989. Highlights of Astronomy, 8: 241. 63. Greenstein, J.L. 1981.Ap. J., 245: 124. 64. Harris, D.H., Woolf, N.J., and Rieke, G.H. 1978. Ap. J., 226: 829. 65. Hawkins, I., and Wright, E.L. 1988. Ap. J., 324: 46. 66. Hecht, J. H. 1987. Ap. J., 314: 429. 67. Hecht, J.H., Holm, A.V., Donn, B., and Wu, C.-C. 1984. Ap. J., , 280: 228. 68. Hecht, J.H., Russell, R.W., Stephens, J.R., and Grieve, P.R. 1986. Ap. J., 309: 90. 69. Henyey, L.G., and Greenstein, J.L. 1941. Ap. J., 93: 70. 70. Hildebrand, R.H. 1983. Q. J. Roy. Astr. Soc., 24: 267. 71. Hildebrand, R.H. 1988. Q. J. Roy. Astr. Soc., 29: 327. 72. Hildebrand, R.H., Dragovan, M., and Novak, G. 1984. Ap. J. Lett., 284: L51. 73. Hobbs, L. M., York, D. G., and Oegerle, W. 1982. Ap. J. Lett., 252: L21. 74. Hoyle, F., and Wickramasinghe, N.C. 1984. Astrophys. Sp. Sci., 103: 189. 75. Hoyle, F., Wickramasinghe, N.C., Al-Mufti, S., Olavesen, A.H., and Wickramasinghe, D.T. 1982. Astrophys. Sp. Sci., 83: 405. 76. Jabbir, N.L., Jabbar, S.R., Salih, A.H., and Majeed, Q.S. 1986. Astrophys. Sp. Sci., 123: 351. 77. Jakobsen, P., Bowyer, S., Kimble, R., Jelinsky, P., Grewing, M., Krmer, G., and Wulf-Mathies,C. 1984. Astron. Astrophys., 139: 481. 78. Jenkins, E. B.1987, in Interstellar Processes (D. J. Hollenbach and H. A. Thronson, Jr., eds.), Reidel, Dordrecht, p. 533. 79. Jones, T. J., and Hyland, H. 1980. M N R A S , 192: 354. 80. Joseph, C.L. 1988, Ap. J., 335: 157. 81. Jura, M. 1987, in Interstellar Processes (D. J. Hollenbach and H. A. Thronson, Jr., eds.), Reidel, Dordrecht, p. 3. 82. Kerridge, J.F., and Mathews, M. 1988. Meteorites and the Early Solar System, Univ. of Arizona Press: Tucson. 83. Kiszkurno-Koziej, E., and Lequeux, J. 1987. Astron. Astrophys., 185: 291. 84. Knacke, R.F., and Capps, R.W. 1979. Astron. J.,, 84: 1705. 85. Knacke, R.F., and Krtchmer, W. 1980. Astron. Astrophys., 80: 281. 86. Koornneef, J. 1982. Astron. Astrophys., 107: 247. 87. Koornneef, J. 1983. Astron. Astrophys., 128: 84. 88. Krtchmer, W., and Huffman, D.R. 1979. Ap. Sp. Sci., 61: 195. 89. Kulkarni, S.R., and Heiles, C. 1987, in Interstellar Processes (D. J. Hollenbach and H. A. Thronson, Jr., eds.), Reidel, Dordrecht, p. 87. 90. Laureijs,R.J., Mattila, K., and Schnur, G. 1987. Astron. Astrophys., 184: 269. 91. Laureijs, R.J.,Chlewicki, G., and Clark, F.O. 1988. Astron. Astrophys., 192: L13. 92. Le Bertre, T. 1987. Astron. Astrophys., 176: 107. 93. Lee, H.M., and Draine, B.T. 1985. Ap. J., 290: 211. 94. Lger, A.,d'Hendecourt, L.,and Dfourneau, D. 1989. Astron. Astrophys.,216: 148. 95. Lenzen, R., Hodapp, K.-W., and Solf, J. 1984. Astron. Astrophys., 137: 202. 96. Lequeux, J., Maurice, E., Prvot-Burnichon, M.L., Prvot,L., and Rocca-Volmerange, B. 1982. Astron. Astrophys., 113: L15. 97. Lewis, R.S., Tang, M., Wacker, J.F., Anders, E., and Steel, E. 1987. Nature, 326: 160. 98. Lillie, C.F., and Witt, A.N. 1976. Ap. J., 208: 64. 99. Little-Marenin, I.R. 1986. Ap. J. Lett., 307: L15. 100. Lockman, F.J., Jahoda, K., and McCammon, D. 1986. Ap. J., 302: 432. 101. Lucke, P.B. 1978. Astron. Astrophys., 64: 367. 102. Maciel, W.J. 1981. Astron. Astrophys., 98: 406. 103. Martin, C., and Bowyer, S. 1989. Ap. J., 338: 677. 104. Martin, N., Maurice, E., and Lequeux, J. 1989. Astron. Astrophys., 215: 219. 105. Martin, P.G. 1975. Ap. J., 202: 393. 106. Martin, P.G., and Rogers, C. 1987. Ap. J., 322: 374. 107. Martin, P.G., and Rouleau, F. 1989. Berkeley EUV Colloquium, January, 1989. 108. Martin, P.G., and Whittet, C.C.B. 1990. Ap. J., submitted. 109. Massa, D., and Fitzpatrick, E.L. 1986. Ap.J. Suppl., 60: 305. 110. Mathis, J.S. 1973. Ap. J., 173: 815. 111. Mathis, J.S. 1986. Ap. J., 308: 281. 112. Mathis, J.S. 1986. Pub. Astr. Soc. Pacific, 98: 995. 113. Mathis, J.S., and Whiffen, G. 1989. Ap. J., 341: 808. 114. Mathis, J.S., Rumpl. W., and Nordsieck, K.H. 1977. Ap. J., 217: 425. 115. Matsumoto, T., Hayakawa, S., Matsuo, H., Murakami, H., Sato, S., Lange, A.E., Richards, P.L. 1988. Ap. J., 329: 567. 116. Mattila, K. 1979. Astron. Astrophys., 78, 253. 117. Mauche, C.W., and Gorenstein, P. 1986. Ap. J., 302: 371. 118. McCarthy, J.F., Forrest, W. J., Briotta, D. A., and Houck, J. R. 1980. Ap. J., 242: 965. 119. Morgan, D. H., and Nandy, K. 1982. M N R A S, 199: 979. 120. Murthy, J., Henry, R.C., Feldman, P.D., and Tennyson, P.D. 1988. Ap. J., 336: 954. 121. Nandy, K., Morgan, D.H., Willis, A. J., Wilson, R., and Gondhalekar, P.M. 1981. M N R A S, 196: 955. 122. Neckel, Th., and Klare, G. 1980. Astron. Astrophys. Suppl., 42: 251. 123. Nuth, J.A. III, and Stencel, R.E, eds. 1986. Interrelationships Among Circumstellar, Interstellar, and Interplanetary Dust (N.A.S.A. Conf. Publ. 2403), U.S. Government Printing Office. 124. Pajot, F., Gispert, R., Lamarre, J.M., Pomerantz, M.A., Puget, J.L., and Serra, G. 1989. Astron. Astrophys., in press. 125. Papoular,R., Conard, J.,Giuliano, M., Kister, J., and Mille, G. 1989. Astron. Astrophys., 217: 204. 126. Papoular, R., and Pegourie, B. 1983. Astron. Astrophys., 128: 335. 127. Paresce, F., McKee, C., and Bowyer, S. 1980. Ap. J., 240: 387. 128. Pegourie, B., and Papoular, R. 1985. Astron. Astrophys., 142: 451. 129. Pendleton, Y., Tielens, A.G.G.M., and Werner, M. 1990. Ap. J., in press. 130. Pendleton, Y., Werner, M., Capps, R., and Lester, D. 1986.Ap. J., 311: 360. 131. Pottasch, S.R., Baud, B., Beitema, D., Emerson, J., Habing, H.J., Harris, S., Houck, J., Jennings, R., and Marsden, P. 1984. Astron. Astrophys., 138: 10. 132. Price, S.D. 1981. Astron. J.., 86: 193. 133. Puget, J.L., and Lger, A. 1989. Ann. Rev. Astron. Astrophys., 27: 161. 134. Purcell, E.M. 1979. Ap. J., 231: 404. 135. Ratnatunga, K.U., and van den Bergh, S. 1989. Ap. J., 343: 713. 136. Rieke, G.H., and Lebofsky, M.J. 1985. Ap. J., 288: 618. 137. Roche, P.F., and Aitken, D.K. 1984. MNRAS, 208: 481. 138. Roche, P.F., and Aitken, D.K. 1985. MNRAS, 215: 425. 139. Rush,W.F., and Witt, A.N. 1975. Astron. J., 80: 31. 140. Russell,R.W., Soifer, B.T., and Willner, S.P. 1978. Ap. J., 220: 568. 141. Ryter, C., Puget, J.L., and Prault, M. 1987. Astron. Astrophys., 186: 312. 142. Sakata, A., Wada, S., Tanabe, T., and Onaka, T. 1984. Ap. J. Lett., 287: L51. 143. Sandford, S.A., and Walker, R.M. 1985. Ap. J., 291: 838. 144. Savage, B.D., and Mathis, J.S. 1979. Ann. Rev. Astron. Astrophys., 17: 73. 145. Schloerb, F.P., Snell, R.L., and Schwartz, P.R. 1987. Ap. J., 319: 426. 146. Schmidt, G.D., Cohen, M., and Margon, B. 1980. Ap. J. Lett., 239: L133. 147. Scoville, N.Z., and Sanders, D.B. 1987, in Interstellar Processes (D. J. Hollenbach and H. A. Thronson, Jr., eds.), Reidel, Dordrecht, p. 21. 148. Seab, C.G., and Snow, T.P. 1985. Ap. J., 295: 485. 149. Seaton, M. J. 1979. M N R A S, 187: 73p. 150. Sellgren, K. 1984. Ap. J., 277: 623. 151. Sellgren, K., Allamandola, L. J., Bregman, J. D., Werner, M. W., and Wooden, D. H. 1985. Ap. J., 299: 416. 152. Sitko, M.L., Savage, B.D., and Meade, M.R. 1981. Ap. J., 246: 161. 153. Snow, T.P., Buss, R.H., Gilra, D.P, and Swing, J.-P. 1987. Ap. J., 321: 921. 154. Tapia, M., Persi, P., Roth, M., and Ferrari-Toniolo, M. 1989. Astron. Astrophys., in press. 155. Tennyson, P.D., Henry, R.C., Feldman, P.D., and Hartig, G.F. 1988. Ap. J., 330: 435. 156. Tielens, A.G.G.M. 1989, in proceedings of IAU Symposium 135, Interstellar Dust (L.J. Allamandola and A.G.G.M. Tielens, eds.), D. Reidel and Co. (Dordrecht), p. 239. 157. Tielens, A.G.G.M., and Allamandola, L.J. 1987, in Interstellar Processes (D. J. Hollenbach and H. A. Thronson, Jr., eds.), Reidel, Dordrecht, p. 397. 158. Toller, G.N. 1981. Unpub PhD thesis, State Univ. of New York at Stony Brook. 159. Vrba, F.J., and Rydgren, A.E. 1985. Astron. J., 90: 1490. 160. Volk, K., and Kwok, S. 1988. Ap. J., 331: 435. 161. Wallis, M.K., Wickramasinghe, N.C., Hoyle, F., Rabilizirov, R. 1989. MNRAS, 238: 1165. 162. Weintraub, D., Sandell, G., and Duncan, W.D. 1989. Ap. J. Lett., 340: L69. 163. Werner, M.W., Davidson, J.A., Hildebrand, R.H., Morris, M.R., Novak, G., and Platt, S.R. 1988. Ap. J., 333: 729. 164. Whittet, D.C.B. 1984. M N R A S, 210: 479. 165. Whittet, D.C.B. 1987. Q.J.Roy.Astron.Soc., 28: 303. 166. Whittet, D.C.B. 1988, in Dust in the Universe (M. E. Bailey and D. A. Williams, eds.), Cambridge University Press, p. 25. 167. Whittet, D.C.B., and van Breda, I.G. 1978. Astron. Astrophys., 66: 57. 168. Whittet, D.C.B., and van Breda, I.G. 1980. MNRAS, 192: 467. 169. Whittet, D.C.B., Bode, M.F., Longmore, A.J., Adamson, A.J., McFadzean, A.D., Aitken, D.K., and Roche, P.F. 1988. MNRAS, 233: 321. 170. Wilking, B. A., Lebofsky, M. J., and Rieke, G. H. 1982. Astron. J., 87: 695. 171. Willner, S.P. 1984, in Galactic and Extragalactic Infrared Spectroscopy, ed. M.F.Kessler and J.P. Phillips (Reidel: Dordrecht), p. 37. 172. Willner, S.P., Gillett, F.C., Herter, T.L., Jones, B., Krassner, J., Merrill, K.M., Pipher, J.L., Puetter, R. C., and Stecher, T.P. 1982. Ap. J., 253: 174. 173. Witt, A.N. 1968. Ap. J., 152: 59. 174. Witt, A.N. 1988, in Dust in the Universe (M. E. Bailey and D. A. Williams, eds.), Cambridge University Press, p. 1. 175. Witt, A.N., and Schild, R.E. 1988. Ap. J., 325: 837. 176. Witt, A.N., Bohlin, R.C., and Stecher, T.P. 1984. Ap. J., 279: 698. 177. Witt, A.N., Bohlin, R.C., and Stecher, T.P. 1986. Ap. J. Lett., 305: L23. 178. Witt, A.N., Schild, R.E., and Kraiman, J.B. 1984. Ap. J., 262: 708. 179. Witt, A.N., Stecher, T. P., Boroson, T. A., and Bohlin, R. C. 1989. Ap. J. Lett., 336: L21. 180. Witteborn, F.C., Sandford, S.A., Bregman, J.D., Allamandola, L.J., Cohen, M., Wooden, D.H., and Graps, A.L. 1989. Ap. J., 341: 270. 181. Woody, D.P., Scott, S.L., Scoville, N.Z., Mundy, L.G., Sargent, A.I., Padin, S., Tinney, C.G., and Wilson, C.D. 1989. Ap. J. Lett., 337: L41. 182. Wright, E.L. 1987. Ap. J., 320: 818. 183. Wright, E.L. 1989. Ap. J. Lett., submitted. 184. York, D., Drake, J., Jenkins, E., Morton, D., Rogerson, J., and Spitzer, L. 1973. Ap. J. Lett., 182: L1. 185. Zinner, E., McKeegan, K.D., and Walker, R.M. 1983. Nature, 305: 730. 186. Zinner, E., Tang, M., and Anders, E. 1987. Nature, 330: 730. Table 1 Interstellar Extinction, A(l)/A(J), J 1.25 ma _________________________________________________________________________________l(m)A(l)/A(J)l(m)A(l)/A(J) RV = 3.1A(l)/A(J) RV = 5.0l(m)A(l)/A(J) RV = 3.1A(l)/A(J) RV = 5.0_____________________________________________________________________________________250b0.001550.0950.0950.249.035.131000.00413.40.1820.1820.21811.296.03600.00712.20.3820.3820.2010.085.32350.0131.650.6240.6240.188.934.66250.0481.251.001.000.159.444.57200.0750.91.701.700.1311.094.89180.0830.72.662.430.1212.715.32150.0530.553.553.060.091c17.2120.0980.444.703.670.07319.1100.1920.3655.534.070.0419.159.70.2080.335.874.120.0237.319.00.1570.286.904.340.0043.39 70.0700.267.634.590.0021.35a A(l)/A(1.25 m) is the same for l > 0.9 m for all lines of sight, to within present errors. To estimate A(l)/N(H), multiply tabulated entry for RV = 3.1 by 1.51 x 10-22 cm2 (H atom)-1. Except as noted below, entries are calculated from Cardelli, Clayton, and Mathis (1989(20). Other values of RV can be determined from that paper. b For l > 250 m, multiply entry for 250 m by (250 m/l)2. c For l < 0.1 m, entries are from Martin and Rouleau (1989)(107), increased by 1.15 for continuity at 0.12 m. figure captions Figure 1. The observations of A(l)/A(V) plotted against 1/RV, where RV = A(V)/E(BV) (from Cardelli, Clayton, and Mathis 198920). A12 refers to 1200 , A22 to 2175 , A28 to 2800 , and A70 to 7000 (the standard R filter). The observational values of black dots are from Fitzpatrick and Massa (198853). The regularity of the observations, and the scatter about the mean relationship, is shown. Figure 2. Three cases of a mean extinction law. Solid lines are obtained by fitting the slopes of the A(l)/A(V) - RV1 relationship (Figure 1) by an analytic formula (CCM20), and dashed lines are actual extinction laws of stars with the appropriate values of RV. The error bars in the lower panel show the standard deviations of the observations of the entire sample (54) from the extinction law obtained from the formula. The open circles in the panel show the deviations from the "standard" mean extinction law (Seaton 1979149) for the value of RV =3.2, appropriate for the diffuse ISM. Figure 3. The mean intensity of radiation at the center of a cloud with a radial extinction of A(V) = 5 magnitudes and no internal sources, expressed in terms of the mean intensity of the incident interstellar radiation field. Two values of RV [= A(V)/E(BV)] are shown: one corresponding to the mean value for diffuse dust (RV = 3.1), and one for a typical observed value for lines of sight penetrating clouds (RV = 5). The use of a mean extinction law corresponding to diffuse dust for predicting the radiation field within clouds leads to a gross error in the predicted radiation field within the cloud. vy{|ڀ@!no$%  ] > X WXmnwx)]HN*+CDux890_`jkm g  @] g h ""###{#|##$$$$%W%X%x%y%%& &&\&]'"'#'Y'Z''* * *E*F**,,-7-9-q-r--.].a.b.c.e.......// / ///"/?00#0`0d00001111"1122 2222222223333(3)  ^3)3*3f3g3h3j333333344444555557778Q8W8888929394969999::::::::;U;[;;j?jrj|jjjjk7kAktk@akkllllllllm!m3m4m5mjm{m|m}m~mmmmn n n nnUnZn[n\n]nnnnno>oOoPoQoRosotooooooooooop9p>p?p@pApwp}p~pppppppq1q:q;qqqqqqqqqqr!r"r.rhrnrorprrssssPs\ss@bsst%t*t+t,t-tRt[ttttttu>uEuFuuuuuuuuuvPvWvXvvwFw\w]x)x3x4x>x?xbxhxixjxxxyyyyyyyyyyzz-zBzzzzz{ {{{{{{{|!|R}!}&}'}(})}K}R}S}T}x}~~~L~S~T~U~~~~~@@`~~~~~FQRSTfr5BF kwx`a0367@Bx  @Z7P35BDUW232=TUst !ϘГЕЗИисыь@  H"#23j? ]U+Fyz7$&'*,{.W/P26R8]:=o>?@CEŻ٩ysm% % % % % % % % % %  %   %  %  % %  % % *EF5KCNPPPRW\%__aYdue|e}ekm?m@mUmzort/uxOz}V~! 9;ZFs,"ZĺĘ %  % % %  %  % %  %  % % :g~t 1ݳ޲UNV<9$]_v Af]%&)#ĺΰغؔ~% %  %  % % % %  %  %  %  %  % %  % 1)#+.02559~=>>??:?;AQBBB BB&B,B2B:BBBOBWBXBYBZB[B\B]B^ʽʯjc\  hP ,l X% L     8 % % % %  %  %   %  B^B_B`BaBrBvB|BBBBBBBBBBBBBBBCCCC%C&C/C0C2C6C;C=C?CECNCOCQCRC\C_CdChC~CCCCCCCCCCCº²ڢºڢºڢºڢº L  8      8 hP ,l X% L 6CCCCD D DDD!D%D8DADXDmDnDoDDDDDDDDDDDDDE=EF5FFGGHHbHI+IiIJJWJKKYKL ҚҌzzzzzzz% % % 8  L    8 hP ,l X% L  0L LYLLM7MxMN;NO4OOPWPPPQ)QR%RS-SkSST&T^TTU<UUVVPVVWWXXIXYY{YYZ-ZhZZ[=[|[[\H\\]g]^W^__@_`+`z`aa;aqabDbbc&c_cdddeeVeef fEffggTgghh4hphhiti% % `iijIjjkMklXlm?mmnnhno[ooppKppq qFqqr;r{rssjst8thtuuQuvvbwwwx?xuxy(yyzzz{#{||}4}^}~~`~~_M  %  %  % % %  % % G<=>?@ABCDEF򬗐xxp        X  X EhPs @@@@@@@@@ X ) lmnopqrstyŋsfYRF:  h` 4 h`   xh` L h`  h`  h`9hP[!!!!!!!!@@@@@@@!p hP ,8!   t  #).38=BGHKQUZ_djopsy}ƹƹƹƹƹƹ  xh`  h` 4 h`   L h`  h`  h`hP ,T!p  h`4    %+0237=BGLRWYZaƺƺƺƺƺƀ  h   xh` L h`  h`  h`hP ,T!p  h`  h` 4 h`2aglqv|)*}~тьzszzzmzzc^YYcQYL % % % % hP!!p hP ,T!p  h`  h` 4 h`   xh` L h` MBodyRefs Times doubTimes body doubBostonRef Underline- xhh h hrhhrhhr hh t  Reprinted with permission of Annual Reviews, Inc., P = nkT Note that the pressure, not tabulated below, is approximately constant for most of the phases, suggesting that a given parcel of gas might spend most of its time in the condition nT 3000 cm3 K. Of course, there are motions within the diffuse ISM, so the pressure equilibrium is only approximate. also the contribution of the magnetic field and cosmic rays to the total pressure is not negligible, and may even be dominant. (115 Ghz) @D"4F,Ymk |R֧o />:ER^ZjxxaJ+k  N $  73@_e@P"#23j>m>n@BBBBCCCC#C+C8CACBCCCDCECFCGCHCJC[C_CeCmCyCCCCCCCCCCCCCDD DD"D#D%D)D.D0D2D8DADBDEDODRDWD[DqDrDDDDDDDDDDDDDE EEEE+E4EKE`EbEuExEEEEEEEEEF0FG(GGHHIIUIJJ\JKKJKL LLLLMLMMN*NkNO.OP'PrPQJQrQQR% % % % % % % % % % % % % % 8       L  8      L  8       L  8        L L L  8 8      L L L  8       L L  8       L  8      L   8      L  % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % @%  g3)Fh]HWb--FPZczks~ьE)#B^CL iaь"HH(F(HH(d'=/RH -:LaserWriter (K(CCCA4A=A>AIABB"BCCϘBГИиBсBы