2.4. Results and astrophysical implications

Table 1 compares the abundances of refractory elements in two ZDA and the Li & Draine [46] dust models to the constraints imposed from abundance determinations in stars and the ISM. Abundances are normalized to 10⁶ hydrogen atoms or parts per million (ppm). The abundances in the dust were derived by subtracting the observed gas phase abundances from the respective solar, F and G stars, and B star abundances. Note that the F and G star abundances have larger uncertainties in their O, Mg, and Si determinations than their solar and B stars counterparts. However, all abundances are consistent within the 1 uncertainties in their determinations.

Table 1. Inferred dust phase abundances in the diffuse ISM

	reference	C	O	Mg	Si	Fe
Total	Solar *	245 ± 30	457 ± 56	34 ± 8	32 ± 3	28 ± 3
	F & G stars †	358 ± 82	445 ± 156	43 ± 17	40 ± 13	28 ± 8
	B stars **	190 ± 77	350 ± 133	23 ± 7	19 ± 9	29 ± 18
Gas		75 ± 25 ‡	385 ± 12 §	0	0	0
Dust	Solar	170 ± 40	72 ± 57	34 ± 8	32 ± 3	28 ± 3
	F & G stars	283 ± 86	60 ± 156	43 ± 17	40 ± 13	28 ± 8
	B stars	115 ± 81	0 ± 134	23 ± 7	19 ± 9	29 ± 18
Models	ZDA (BARE-GR-S)	246	133	33	33	33
	Li & Draine [46]	254	192	48	48	48
	ZDA (COMP-NC-B )	196	154	28	28	28
* Asplund, Grevesse, & Sauval [4] † Sofia & Meyer [60] ** Sofia & Meyer [60] ‡ Dwek et al. [17] § André et al. [2], average between samples A and C

The first ZDA model (BARE-GR-S) consists of bare silicate and graphite grains and PAHs, and was derived using the Holweger [32] solar abundances constraint. The dust composition and optical properties are identical to those of the Li-Draine model. However, they differ significantly in their grain size distribution (see Figure 19 and Table 7 in ZDA for the comparison and the analytical fit to the derived ZDA size distribution). Table 1 shows that the ZDA BARE model reproduces the updated abundances constraints better than the standard Li-Draine model, with the strictest constraint provided by the Mg, Si, and Fe abundance. The Li-Draine model requires ~ 70% more Fe and ~ 50% more Mg or Si to be in the dust than is available from either set of stellar abundances. Figure 1 shows how the solar abundance determinations of Mg, Si, and Fe varied over time, and compares them with the three dust models listed in the table. The abundance determination of Mg and Si have remained quite constant over time and consistent with meteoritic abundance determinations. The formerly large discrepancy between the solar and meteoritic Fe abundances reported by Anders & Grevesse [1] has been resolved by the more recent measurements of Asplund, Grevesse, & Sauval [4], which has settled on the lower value of 28 ± 3. Also listed in the table is model COMP-NC-B from ZDA, which consists of silicates, composite grains, and PAHs. All the carbon in this model is in PAHs and in the organic refractory component of the composite grains. This model was constructed to fit the B star abundances and requires the least amount of carbon to be locked up in the solid phase of the ISM.

Figure 1. The relative steadiness of the solar (or meteoritic) abundance determinations of Mg, Si, and Fe: Cameron [6]-open diamond; Gehren [28]-open square; Anders & Grevesse [1]-open circle; Grevesse & Sauval [29]-filled star ; Asplund, Grevesse, & Sauval [4]-filled diamond. The Fe abundance of Anders & Grevesse [1] is represented by both, the meteoritic and solar abundance determination because of the large discrepancy between the two values. The horizontal lines represent the results of the models discussed in the text.