Annu. Rev. Astron. Astrophys. 1990. 28: 37-70 Copyright © 1990 by Annual Reviews. All rights reserved

2.1.2 THE 2175-Å ``BUMP'' The strongest spectroscopic feature in the entire observed spectrum, in terms of equivalent width in frequency units, is the "bump" situated at 2175 Å, or 4.6 µm^-1 (see Figure 2). The bump is present in all values of R_V. Its origin is not well understood (see ref. 38 for an excellent review), although there is general (but not unanimous) agreement that it is probably caused by graphite or a slightly less well-ordered form of carbon.

Let A_bump (^-1) be the extinction at a wavenumber ^-1 between 3.3 and 6 µm^-1 in excess of a linear extinction interpolated between the end points. The important properties of A_bump(^-1) are as follows (176; FM86; CCM89):

1. The bump is extremely strong, and must be produced by a very abundant material (which is why most theories attribute it to carbon). The equivalent width of the bump per A(V), known from CCM89 or Seaton (149), can be expressed in terms of the oscillator strength, f_bump, times the number of absorbing atoms, N_bump. Bohlin et al. (9) determined the mean N(H) / E(B - V), from which N(H) / A(V) follows. Dividing the expressions yields N_bump f_bump = 9.3 x 10^-6 N(H). Only the elements C, N, O, Ne, Mg, Si, and Fe (excluding noble gases) can provide enough absorption strength even if the transition is exceedingly strong, f_bump 1. Each of the elements Fe, Si, and Mg require f_bump = 0.3 even if the entire cosmic abundance is responsible for producing the bump, while 8% of the carbon would be required for the same f_bump.

2. The central wavenumber, ₀^-1, is surprisingly constant. For the stars in the FM86 sample, ₀^-1 = 4.599 ± 0.019 µm^-1, corresponding to ₀ = 2174 ± 9 Å. This amounts to a mean deviation of 0.4% in ₀^-1, while other properties of the extinctions vary considerably. However, there are real variations in ₀. The spreads of ₀^-1 for stars within a given cluster (101) are significantly less than among the field stars, and serve to establish an upper limit to the observational errors. The stars HD 29647 and HD 62542 have especially peculiar extinctions (18), with ₀ = 2128 Å and 2110 Å, respectively, that are smaller than the mean ₀ by many standard deviations. These stars have the broadest bumps known (see below) but a rather small central absorption A_bump(₀), so that the integrated strengths of their bumps are about average. Their environments are very different, one being in a quiescent region in Taurus and the other in the Gum Nebula.

It is remarkable that graphite, a completely ordered and stable form of carbon, has a resonance very close to 2175 Å of about the right width and strength to produce the bump. Small graphite particles (radii < 0.005 µm) of various sizes and similar shapes would have the resonance at a common wavelength independent of size, but larger ones would have ₀ shifted to longer wavelengths. Almost all theories suggest that the bump is produced by graphitic carbon in this way, along the lines suggested by Hecht (66).

3. The width of the bump, expressed as the full-width at half-maximum (FWHM), varies widely, currently with extremes of 0.768 µm^-1 (HD 93028) and 1.62 µm^-1 (HD 29647). The only significant correlation of the FWHM is with the mean gas density along the line of sight (FM86). The lack of correlation between the FWHM and ₀^-1, suggests that that the variations in width are not caused by coatings of varying thicknesses upon a common carrier particle. Such coatings should produce a shift in ₀^-1 which is related to the change in width. This variation in width, unrelated to the position of the resonance, is difficult to explain with graphite of any size. The total area of the bump relative to A(V), A_bump(^-1) / A(V) d^-1, varies by over a factor of two among the FM86 stars.

4. The albedo of the bump has probably not been determined reliably. Lillie and Witt (98), using the observations of the diffuse galactic light from Orbiting Astronomical Observatory 2, suggested that the albedo drops across the bump. This is expected if the particles producing the bump are small (as suggested by the constancy of ₀^-1, which can most easily be explained by absorption from small particles). However, plane-parallel axisymmetric geometry was used for interpreting the data, while the actual sky brightness in the UV is now considered to be uncertain and almost surely quite asymmetrical (see Section 6).

5. Two reflection nebulae show evidence of scattering in the bump, with a different profile in each (177), suggesting that some carriers of the bump can be large. (Small grains do not scatter). This scattering is not observed in other nebulae. The expected shift in ₀ to longer wavelengths (because of the relatively large particles causing the scattering) is not observed in the extinction of the exciting stars of these nebulae.

6. Observations of carbon stars (see Section 6.2) suggest that amorphous carbon, not graphite, is injected into the ISM. Perhaps small graphite particles can be produced later by annealing, but it is difficult to see how large graphite flakes can be made.