In the 1996 South African "Cold Dust" Symposium (Block & Greenberg 1996), the possible existence of a population of very cold dust (with equilibrium temperatures T < 10 K ) in interstellar space was a subject received much attention. In his invited paper titled "How Cold Could Galaxies Be?" published in the proceedings for that symposium, Mike Disney wrote "... An eminent cosmologist once advised me to forget all about very cold dust because the T4 law ensures that it cannot emit, and therefore by implication cannot absorb, much radiation. He sounded plausible, as Cosmologists are apt to sound, but he was in fact totally wrong, as Cosmologists are apt to be."
Disney (1996) argued that for grains with higher IR and far-IR emissivities, they can achieve rather low temperatures. At first glance, this appears plausible as can be seen in Eq.(1): for a given interstellar radiation field, grains with fixed UV and optical absorption properties would obtain lower temperatures if their long wavelength emissivities are enhanced. Therefore, Disney (1996) wrote "... [Since] we are not confident about their [interstellar grains] size distribution and their emissivities, particularly at long wavelength, ... we have to keep our minds open to the possible existence of a significant amounts of very cold (< 10 K) material in spiral galaxies."
So far, the detection of very cold dust (T < 10 K ) in interstellar space on galactic scales has been reported for various objects: NGC4631, a low metallicity (~ 1/2 of solar) interacting galaxy with T ~ 4-6 K (Dumke, Krause, & Wielebinski 2004); NGC1569, a low metallicity (~ 1/4 of solar) starbursting dwarf galaxy with T ~ 5-7 K (Galliano et al. 2003); inactive spiral galaxies UGC3490 with T ~ 9 K (Chini et al. 1995), NGC6156 with T ~ 8.6 K , and NGC6918 with T ~ 9.4 K (Krügel et al. 1998); and several irregular and blue compact dwarf galaxies with T < 10 K in the Virgo Cluster (Popescu et al. 2002). Very cold dust with T ~ 4-7 K has also been detected in the Milky Way Galaxy (Reach et al. 1995; also see Boulanger et al. 2002). This component is widespread and spatially correlated with the warm component (16-21 K ). By comparing the dust mass calculated from the IRAS data with the molecular and atomic gas masses of 58 spiral galaxies, Devereux & Young (1990) argued that the bulk of the dust in spiral galaxies is <15 K regardless of the phase of the ISM.
How can dust get so cold? In literature, suggested solutions include (1) the dust is deeply embedded in clumpy clouds and heated by the far-IR emission from "classical grains" (Galliano et al. 2003; Dumke et al. 2004); 18 (2) the dust has unusual optical properties (e.g. fractal or porous grains with enhanced submm and mm emissivity; Reach et al. 1995; Dumke et al. 2004). However, as discussed in detail by Li (2004b), while the former solution appears to be inconsistent with the fact that the very cold dust is observed on galactic scales, the latter violates the Kramers-Kronig dispersion relation (Purcell 1969; Draine 2003b; Li 2003b), except for extremely elongated conducting dust (Li 2003a). Perhaps the submm and mm excess emission (usually attributed to very cold dust) is from something else? To avoid the "temperature" problem, one can adopt the Block direct method to identify and characterize the presence and distribution of the cold and very cold dust: using near-IR camera arrays and subtracting these from optical CCD images. This method measures the dust extinction cross section and does not require the knowledge of dust temperature (see Block 1996; Block et al. 1994a, b, 1999). Otherwise, a detailed radiative transfer treatment of the interaction of the dust with starlight (e.g. Popescu et al. 2000, Tuffs et al. 2004) together with a physical interstellar dust model (e.g. Li & Draine 2001b, 2002c) is required.