On autumn nights, when one looks to the sky from a dark location such as the Wise Observatory in the Negev or other modern astronomical observatories, one sees not only the individual stars, planets and the Moon of the night sky but also a luminous band stretched across the sky. This is the Milky Way (MW), known as "our Galaxy". The autumn segment of the Milky Way looks "ragged"; it is like someone tore out a patch from the center of the luminous band. This phenomenon was known to astronomers through the ages and the region was called "the rift in Cygnus" after the constellation that harbors this part of the MW; for centuries it was thought that through this region one actually sees the space outside the MW.

Figure 2: The constellation Cygnus seen on a mosaic of sky images from Google Earth shows a conspicuous lack of stars in a diagonal band running from upper left to lower right; this is the "Rift in Cygnus", interpreted as a concentration of dust clouds along the equatorial plane of the Milky Way. (Source GEarth, original here).

In the early 1930 the Swiss-American astronomer Robert Julius Trumpler compared distances to star clusters determined from their brightness, with distances determined from their angular size. The basic assumption was that the clusters had identical dimensions and brightness. Trumpler found that the "brightness" distance was consistently larger than the "size" distance; this could be either because the clusters were becoming intrinsically fainter the further they were from us, which would put us at the center of the Universe in a difficult non-Copernican position, or that some material in the space between us and the clusters could be causing the dimming of light. Trumpler found essentially that interstellar dust dims the light of objects behind it; this is now called "light extinction".

Since the early-1900s the study of the interstellar extinction advanced significantly with observational extensions in the near-infrared (IR) and the ultraviolet (UV). Much of the IR and the UV never reach ground-based observatories and require observations from space. Earth-bound locations are strongly affected by the atmospheric extinction due to light scattering by atoms and molecules in gas phase; this is what makes the daytime sky blue and the Sun red at sunrise or sunset. The extension of the astronomical reach into wider spectral regions allowed the formulation of the interstellar extinction in terms of physical parameters. The simplest expression of this is through the "extinction curve", a plot showing the fraction of light lost from the beam because of scattering and absorption, as a function of inverse wavelength, and shown in Figure 5 below. The plot shows that the light loss is approximately proportional to the inverse wavelength: bluer light is lost much more than red light, thus an object affected by dust extinction appears redder.

The basis for understanding the extinction is a fundamental piece of work from 1957 by the Dutch astrophysicist Hendrik van der Hulst (HvdH) from Leiden, where he calculated the properties of the light extinction by dust. HvdH found that the principal parameter affecting the light extinction is the typical size of dust grain particles. His work, and the follow-up work of the Dutch-American Jerome Mayo Greenberg and others, painted a coherent picture of the interstellar dust and its effects on the light of background sources.

Figure 3: The late Professor Jerome Mayo Greenberg holds a life-size model of a dust grain composed of individual smaller grains. (Original here)

Greenberg showed that dust grains evolve. They form in atmospheres of cool, late-type giant stars as tiny iron needles or as amorphous silicate or carbon particles, and grow by accreting mantles of astrophysical ices (water, methane, carbon dioxide and monoxide, etc.) when they are incorporated in dense gas clouds. The mantles are "processed" by ultraviolet radiation when the grains are exposed to the light of young and massive stars; simple molecules are ripped up by energetic UV photons and the free radicals recombine in new and sometimes unexpected ways. Laboratory simulations showed that such photo-processing can produce very heavy organic molecules in the cold deep of space.

The grains travel for tens or hundreds of millions of years together with gas atoms and molecules in interstellar matter clouds. Due to various internal or external processes some clouds compress; with increased density, parts of a cloud may become self-gravitating. Their fate, at this moment, is sealed: a new star will form from the collapsing mass of gas and dust, and with it an associated planetary system. The new planets will receive, soon after their formation, a blessing rain of debris from the system construction. These bits and pieces, in the form of comets and asteroids of various sizes, incorporate the interstellar dust grains and their evolved mantles; this is how young Earths can get their starting load of life-forming material.

This "short history of dust" is only one facet of this fascinating cosmic ingredient. The other is related to its property of dimming the starlight. While this has been fairly well established for our Milky Way, the knowledge of the interstellar dust properties in other galaxies and in the space between galaxies is very limited. Apart from our galaxy, only two other galaxies have had their dust content analyzed reasonably well: these are two satellite galaxies of our Milky Way that are 180,000 (the Large Magellanic Cloud=LMC) and 210,000 light-years (the Small Magellanic Cloud=SMC) away from us. The reason for this amazing lack of information is that the method used to derive the dust extinction requires the intimate comparison of identical stars, one whose light is extinguished by the dust and another that is unaffected. The more distant an object is, the harder it is to separate its stars and identify one that exactly matches a local, unextinguished star.

Figure 4: The Large Magellanic Cloud galaxy. The pink region to the left and above the main stellar body exhibits vigorous star formation. (Original image from NASA)

The findings from comparing Milky Way and Magellanic dust showed important differences in the extinction, mostly at the ultraviolet end of the spectrum, as Figure 5 shows. The SMC dust dimmed UV light much more efficiently than LMC dust, and that was more efficient than MW dust in extinguishing light. A special spectral feature at 217.4 nm (inverse wavelength 4.6 μm^-1), now attributed to some form of carbon, perhaps large polycyclic aromatic molecules (see Figure 6), was missing in the SMC. Not much more was known about extragalactic dust until a new method was invented to tackle the problem.

Figure 5: Wavelength dependence of the interstellar extinction in the Milky Way, Large and Small Magellanic Clouds. The horizontal axis is the inverse wavelength of the light (in microns) and the vertical is the relative extinction between any wavelength λ and the wavelength of yellow-green light called "V". (Original plot here)

Instead of using individual stars to determine the extinction, the method uses the integrated light of an assembly of stars. The targets of this study are elliptical (E) and lenticular (S0) galaxies. These objects are known from extensive studies to have smooth light distributions, becoming fainter per unit area the further one observes away from their centers. The assumption is that these galaxies formed their stellar population in a single and relatively short episode shortly after the Big Bang, about ten billion years ago. The construction material for forming stars and planetary systems was used up then and, at present, these galaxies do not show traces of interstellar matter dust.

Figure 6: Model of the interstellar extinction law as a sum of three grain components: large (~0.2 micron cylinders-solid line), tiny ~100Å absorbing particles producing the 217.4 nm bump (dashed line), and ~10Å PAHs (polycyclic aromatic hydrocarbon molecules-dotted line). From Li and Greenberg 1997 (available on-line)

However, studies in the last three decades showed that there is a class of E and S0 galaxies that do contain dust, in non-negligible quantities. This dust is probably not "endemic" to the galaxy, but was accreted from a donor galaxy or was accreted along with a donor galaxy that is still being incorporated in the galaxy. Galactic cannibalism is well-known in the Universe; even our own Milky Way shows signs of swallowing at least two much smaller stellar systems. In fact, one model explaining the assembly of galaxies starts from smaller units that are dwarf galaxies, and these are accreted during the ages to form a much larger, major galaxy.

Figure 7: One galaxy in front, the left spiral, casts a dust shadow on the other galaxy, an elliptical in the back, to the right. It is possible to deduce the extinction law by the dust in the spiral from modeling the background elliptical. The object is AM1316-241, 400 million light-years away in the Hydra constellation (original image from NASA)

Our method of measuring the properties of extragalactic dust relies, therefore, on observations of E and S0 galaxies with dark lanes. We model the light distribution of the underlying galaxy, which should follow the canonical profile of an unperturbed galaxy of the same morphological class. We compare the model with the actual light distribution measured at a number of spectral bands. The comparison yields immediately the amount of light missing; this is light that was extinguished by the dust in the dark lane and the measurement yields the extinction at one specific wavelength. Repeating the process at a number of discrete wavelengths yields the wavelength-dependence of the extinction, and this is compared with the similar relation from MW dust. The consensus is that part of the missing light, which was absorbed by the grains, is ultimately re-emitted as long-wavelength infrared radiation.

Figure 8: Southern S0 galaxy with multiple thin dust lanes in the constellation Leo. Image obtained for Finkelman's research project at the Southern African Large Telescope.

This method has been applied by now to about three dozen galaxies, with the latest result from an analysis of southern sky objects by Ido Finkelman, a PhD student at Tel Aviv University. The results are amazing; the dust in very distant galaxies seems to be very similar to our own Milky Way dust and to have a similar grain size. Why should this happen? Is there a cosmic conspiracy to produce similar dust grains in all galaxies, or is this a bias of our research method? Why does the total amount of dust calculated from the missing light not match that calculated from the infrared emission? One testable possibility is that a galaxy may contain different dust grain types, some sufficiently small to affect light in the visible domain and extinguish preferentially bluer light than redder, and others much larger that will only block the light without "reddening" it.

An intriguing result that forms part of an MSc thesis at Tel Aviv University seems to indicate that this might well be the case. Evgeny Gorbikov studied the distribution of more than five million stars measured by one of the largest sky surveys in existence, the Sloan Digital Sky Survey (SDSS), in about 900 square degrees of the sky high above the Milky Way plane where one does not expect to encounter much interstellar dust. He found that at least in three locations it is possible to infer the existence of obscuring dust, since stars are missing there in comparison with nearby locations, but that this dust does not cause reddening. The implication, as he showed in his thesis, is that these are large dust grains. Such dust has been identified in circumstellar disks where large grains form by the coagulation of smaller grains as a stage in the process of planetary system formation, and has been hypothesized to exist in supernova remnants where small dust grains are preferably destroyed and in gamma-ray-burst host galaxies where a similar destruction mechanism may take place.

Figure 8: A circumstellar ring of dust surrounds the star Fomalhaut (in the constellation "the Southern Fish") 25 light-years away. Deep within the dust ring orbits a planet, seen within the square frame. The enlargement at lower right shows the planet's motion between two years. (Source: NASA and ESA; original here)

The dust clouds identified by Gorbikov are only about 3000 light-years from us. The existence of much more distant clouds was inferred from studies of the distribution of galaxies and from maps of far-infrared emission. Such clouds, if real, could be extragalactic. If they are composed of significant amounts of large dust grains and if there is a large population of such clouds, this could dim background light sources without significantly reddening them. Such an effect, if proven to be real, may modify our present view of the Universe by reducing the cosmological dimming of distant supernovae without requiring an accelerating Universe.