Extrapolation of Galactic Dust Emission at 100 Microns to CMBR Frequencies Using FIRAS

TL;DR

This work shows that the Galactic dust emission cannot be described by a single power-law emissivity across FIRAS’s 200–2100 GHz range. A physically motivated two-component dust model, with α1 ≈ 1.67 and α2 ≈ 2.70 and tightly coupled component temperatures via ISRF balance, provides an accurate fit to FIRAS data across high-latitude sky and enables extrapolation from 100 μm to microwave frequencies with quantifiable uncertainties. While this model accounts for thermal dust emission up to ν<3000 GHz, comparisons with DMR reveal excess dust-correlated emission at 31–90 GHz not explained by thermal emission alone, suggesting spinning dust or free-free mechanisms. The results furnish robust, publicly available full-sky dust templates for CMB foreground analyses and underscore the importance of multi-component emissivity and temperature structure in Galactic dust modeling.

Abstract

We present predicted full-sky maps of submillimeter and microwave emission from the diffuse interstellar dust in the Galaxy. These maps are extrapolated from the 100 micron emission and 100/240 micron flux ratio maps that Schlegel, Finkbeiner, & Davis (1998; SFD98) generated from IRAS and COBE/DIRBE data. Results are presented for a number of physically plausible emissivity models. We find that no power law emissivity function fits the FIRAS data from 200 - 2100 GHz. In this paper we provide a formalism for a multi-component model for the dust emission. A two-component model with a mixture of silicate and carbon-dominated grains (motivated by Pollack et al., 1994}) provides a fit to an accuracy of about 15% to all the FIRAS data over the entire high-latitude sky. Small systematic differences are found between the atomic and molecular phases of the ISM. Our predictions for the thermal (vibrational) emission from Galactic dust at ν< 3000 GHz are available for general use. These full-sky predictions can be made at the DIRBE resolution of 40' or at the higher resolution of 6.1 arcmin from the SFD98 DIRBE-corrected IRAS maps.

Figures (10)

• Figure 1: Difference spectra from COBE data, after CMBR monopole and dipole removal. The bright and faint regions of the sky are differenced for each channel in the DMR (diamonds) and FIRAS (solid lines) data sets, excluding the Galactic plane and Magellenic Clouds. The sky is divided into cold, warm and hot zones based upon DIRBE $I_{100}/I_{240}$ color ratios. The differences in each zone are renormalized to a $100\micron$ flux of $1.0{\rm ~MJy}/{\rm sr}$, which is a typical flux level for the high-latitude sky. Note the factor of two difference between the cold and hot zones at $\nu \mathrel{\hbox{$<$$\sim$}} 700 {\rm ~GHz}$, relative to the $100\micron$ normalization. For comparison, the dotted line represents $10^{-5}$ the level of the CMBR spectrum.
• Figure 2: FIRAS - DIRBE comparison. Comparison of FIRAS emission in a synthesized $500{\rm ~GHz}$ broad-band versus ( a) H I emission, ( b) DIRBE $100\micron$ ($3000{\rm ~GHz}$) emission with zodiacal contamination removed, ( c) prediction from SFD98 using single-component, $\nu^2$-emissivity model, and ( d) prediction from our best-fit two-component model. The comparisons are made over 71% of the sky. Straight lines are fit and overplotted using the statistical errors in the FIRAS data. The scatter about this line is $\sim 3.5$ times smaller in ( c) or ( d) as compared to ( a) or ( b). The slope in ( d) is almost unity, as expected for a good prediction.
• Figure 3: Dust-correlated emission, scaled by $\nu^2B_\nu(\bar{T})$ for ease of comparison. The FIRAS data (error bars) would be consistent with unity if the $\nu^2$ emissivity model were correct. Panel ( a) overplots broadened temperature models with $\Delta T_{FWHM} = 3{\rm ~K}$ (dashed line) and $\Delta T_{FWHM} = 6{\rm ~K}$ (dash-dot line). Panel ( b) overplots single-component models with $\nu^{1.5}$ (top), $\nu^{1.7}$, and $\nu^{2.2}$ emssivity laws. The horizontal dotted line corresponds to $\nu^{2}$. Panel ( c) overplots two-component models, with the best-fit model shown as a solid line. See Table \ref{['table_results']} for the specific model parameters. These results are not sensitive to an isotropic background in the FIRAS data. The DMR $90{\rm ~GHz}$ measurement is shown as a diamond. The DMR $30$ and $53{\rm ~GHz}$ measurements fall well above any model curves.
• Figure 4: Contour of $\chi^2$ for parameters $f_1$ and $q_1/q_2$, fixing the emissivity laws ($\alpha_1$, $\alpha_2$) to their best-fit values. Our best-fit two-component model is denoted by an X.
• Figure 5: FIRAS versus best-fit model correlation slopes. The sky is divided into three zones based upon temperature: ( a) cold regions (${\mathcal{R}} = I_{100\micron}/I_{240\micron} < 0.62$), ( b) warm regions ($0.62 < {\mathcal{R}} < 0.69$), and ( c) hot regions (${\mathcal{R}} > 0.69$). The systematic residuals between zones is not more than $\sim 5\%$. The vertical line is drawn at $240\micron$, where the models are constrained to fit the DIRBE measurements.