The evidence concerning primordial abundances of D, He, and Li appear to be consistent with D/H 28 ± 4 ppm produced by nucleosynthesis in the early universe (O'Meara et al. 2001, Kirkman et al. 2003). Observations of D/H in the Milky Way have been reviewed by Linsky (2003). The local ISM has a weighted mean D/H = 15.2 ± 0.8 ppm (1- uncertainties) within ~ 180 pc of the Sun (Moos et al. 2002). This value could be consistent with "astration" if ~ 50 % of the H atoms now in the interstellar medium have previously been in stars which burnt D to 3He. Observations appear to find spatial variations in the D/H ratio in the interstellar medium within ~ 500 pc, with values ranging from 7.4-1.3+1.9 ppm toward Orionis (Jenkins et al. 1999) to 21.8-3.1+3.6 ppm toward 2 Vel (Sonneborn et al. 2000). On longer sightlines in the Galactic disk, Hoopes et al. (2003) find D/H = 7.8-1.3+2.6 ppm toward HD 191877 (d = 2200 ± 550 pc) and 8.5-1.2+1.7 ppm toward HD 195965 (d = 800 ± 200 pc). These variations in D/H are usually interpreted as indicating variations in "astration", with as much as ~ 75% of the D on the sightline to Ori having been "burnt", vs. only ~ 20% of the D toward 2 Vel. Such large variations in astration between regions situated just a few hundred pc apart would be surprising, since the ISM appears to be sufficiently well-mixed that large local variations in the abundances of elements like N or O are not seen outside of recognizable stellar ejecta such as planetary nebulae or supernova remnants.
Jura (1982) pointed out that interstellar grains could conceivably sequester a significant amount of D. Could the missing deuterium conceivably be in dust grains?
Let us suppose that dust grains contain 200 ppm C relative to total H (as in the dust model of Weingartner & Draine 2001) with ~ 60 ppm in PAHs containing NC 104 atoms.
The solid carbon will be hydrogenated to some degree. The most highly pericondensed PAH molecules (coronene C24 H12, circumcoronene C54 H18, dicircumcoronene C96 H24) have H/C = (6 / NC)1/2, where NC is the number of C atoms; other PAHs have higher H/C ratios for a given NC. Let us suppose that the overall carbon grain material - including small PAHs and larger carbonaceous grains - has H/C = 0.25.
The carbonaceous grain population would then contain ~ 50 ppm of hydrogen. If ~ 20 % of the hydrogen in the carbonaceous grains was deterium, the deuterium in the grains would then be (D)grain / (H)total 10 ppm. If the total D/H = 20 ppm, this would reduce the gas phase D/H to 10 ppm, comparable to the value observed toward Ori.
Is it conceivable the D/H ratio in dust grains could be ~ 104 times higher than the overall D/H ratio? Some interplanetary dust particles have D/H as high as .0017 (Messenger & Walker 1997, Keller et al. 2000), although this factor ~ 85 enrichment (relative to D/H = 20 ppm) is still two orders of magnitude less than what is required to significantly affect the gas phase D/H value. Extreme D enrichments are seen in some interstellar molecules - D2CO / H2CO ratios in the range of .01 - 0.1 are seen (Ceccarelli et al 2001; Bacmann et al 2003), and attributed to chemistry on cold grain surfaces in dense clouds.
Could such extreme enrichments occur in the diffuse interstellar medium? The thermodynamics is favorable. The H or D would be bound to the carbon via a C-H bond. The C-H bond - with a bond strength ~ 3.5 eV - has a stretching mode at CH = 3.3 µm, while the C-D bond, with a larger reduced mass, has its stretching mode at CD 21/2 CH 4.67 µm. Because of the difference in zero-point energy, the difference in binding energies is
where the sum is over the stretching, in-plane bending, and out-of-plane bending modes, with CH = 3.3, 8.6, and 11.3 µm. This exceeds the difference EHD-H2 = .035 eV in binding energy between HD and H2. It is therefore energetically favored for impinging D atoms to displace bound H atoms via reactions of the form (Bauschlicher 1998)
The branching ratio f1 / f2 exp[( ECD-CH - EHD-H2) / k Td] > 104 if Td 70 K. If no other reactions affect the grain hydrogenation, then the grains would gradually become D-enriched.
However, the interstellar medium is far from LTE - the hydrogen is atomic (rather than molecular) and, indeed, partially ionized because of the presence of ultraviolet photons, X-rays, and cosmic rays. It is not yet clear whether the mixture of non-LTE reactions will allow the grains to become deuterated to a level approaching D/H 1/4, but it seems possible that this may occur. Deuterated PAHs would radiate in the C-D stretching and bending modes at ~ 4.67, 12.2, and 16.0 µm. The 12.2 µm emission will be confused with C-H out-of-plane bending emission (see Figure 3), but the other two modes should be searched for.
Even if extreme D-enrichment of carbonaceous grains is possible, it will take time to develop. Meanwhile, the gas in which the grain is found may undergo a high velocity shock, with grain destruction by a combination of sputtering and ion field emission in the high temperature postshock gas. D incorporated into dust grains would be released and returned to the gas phase if those dust grains are destroyed. PAHs, in particular, would be expected to be easily sputtered in shock-heated gas; destruction by ion field emission would be expected to be even more rapid. Thus if D is depleted into dust grains, we would expect to see the gas-phase D/H to be larger in recently-shocked regions. This could explain the large D/H value observed by Sonneborn et al. (2000) toward 2 Vel.
It should be noted that significant depletion of D from the gas can only occur if there is sufficient carbonaceous grain material to retain the D. This can occur if gas-phase abundances are approximately solar (with ~ 200 ppm C in dust) but would not be possible for abundances significantly below solar - e.g., the abundances in the LMC and SMC. The factor-of-two variations in D/H seen in the local interstellar medium would not be possible in gas with metallicities characteristic of the LMC, SMC, or high-velocity clouds such as "complex C" (see Jenkins 2003).