In the past two years several lines of evidence have begun to point towards the destruction of very small grains in regions of very high uv radiation intensity. The most direct evidence comes from infrared spectroscopy. Roche (1988) and Desert and Dennefeld (1988) have shown that the broad 3-12 µ features attributed to very small grains are absent in the spectra of many Seyfert galaxies (Fig 7a). Destruction of very small grains is also presumably the reason that Rowan-Robinson and Crawford (1989) found that the disc component was very weak or absent in many Seyferts (Fig 7b).
Figure 7. (b) Ratio of infrared luminosity in starburst component to optical luminosity, versus ratio of infrared luminosity in cirrus component to optical luminosity for IRAS galaxies (Rowan-Robinson & Crawford 1989). The Seyferts (filled circles) are deficient in the cirrus component.
Reasonably direct evidence for the destruction of very small grains in a high radiation intensity comes from the decline in the ratio of S(12) / S(100) near hot stars. Ryter et al (1987) showed this effect for Sco and Boulanger et al (1988) showed it for Per.
Telesco et al (1989) argue that a similar effect is seen in the center of M82. Fig 8a shows the increase in S(25) / S(12) with increasing uv intensity found by Telesco et al for M82 superposed on the curve derived from Boulanger et al's observations of Per. However the spectrum of the emission from outside the nucleus of M82 (and of the integrated emission from the galaxy) is very similar to that for the NGC1068 starburst, and for compact Galactic HII regions, shown in Fig 5b, and one would normally assume that the bulk of this emission arises in regions where the visible and ultraviolet optical depth is >> 1. The 10 µm emission from such a cloud does not arise from very small grains. Fig 8b shows the integrated spectrum of M82 compared to the optically thick starburst model of Rowan-Robinson and Crawford (1989): the agreement is good. Also shown is the shape of the spectrum of the central region of M82, derived from the colors measured by Telesco et al (1989). The change in spectrum towards the center of M82 is essentially a shift of the emission peak from 80 µm to 60 µm, presumably due to the increase in intensity of the radiation from the starburst towards to nucleus. It seems unlikely that we are seeing emission from optically thin dust (the ratio of Brackett-alpha to -gamma gives a value for AV of 14 for M82 (Kawara et al 1989) ) and hence the analogy with Per appears to be spurious.
Figure 8b. Top: Integrated spectrum of M82 (filled circles) compared with starburst model (data from Telesco 1988, Smith et al 1990a). The crosses (arbitrary vertical scale) show the relative shape of the spectrum of the core of M82. Bottom: Integrated spectrum of the SMC compared with cirrus model (X = 30) in which abundance of 5 Å grains has been reduced by 2/3rds.
Similarly unconvincing evidence comes from the far infrared colors of galaxies (Pajot et al 1986, Gosh & Drapatz 1987, Helou 1989). Here again the problem is confusion with the role of the optically thick starburst component, for which, in the model of Rowan-Robinson and Crawford (1989), S(12) / S(60) = 0.04, but radiative transfer effects in normal 0.01-0.1 µm dust rather than small grain depletion is the cause. Fig 4b above showed Helou's (1989) compilation of the IRAS colors of compact Galactic HII regions and of galaxies superposed on the range of colors seen in Per by Boulanger et al (1988). The agreement is good, but in my view this is fortuitous in the case of Galactic HII regions and galaxies dominated by starbursts since in most cases the optical depth in these sources is high and the analogy with Per therefore of doubtful significance. If the 60/25 µm color ratio, ignored by Helou, is also considered, the agreement with Per is less impressive. However the case of the Small Magellanic Cloud (Schwering 1988) is convincing because the spectrum of this galaxy does indeed look like cirrus in which the smallest grain component is depleted (see Fig 8b).
In an interesting development, Leene and Cox (1987) have found that the 0.22 µ feature is also suppressed in regions of high radiation intensity, which suggests that this feature is associated with the very small carbonaceous grains responsible for the broad features and diffuse emission at 2-20 µ.