Milky Way Mid-Infrared Spitzer Spectroscopic Extinction Curves: Continuum and Silicate Features

TL;DR

This study measures and analyzes mid-infrared extinction curves in the Milky Way, revealing differences in dust grain features and providing new observational insights into silicate and UV extinction properties.

Contribution

It presents the first direct observational confirmation that the 10 micron silicate feature and the 2175 Å bump originate from different grain populations.

Findings

Average diffuse extinction curve is lower than previous measurements.

10 micron silicate feature does not correlate with the 2175 Å bump.

Standard R(V) estimation method has a scatter of about 7%, improved to 3% using IRAC 5.6 micron band.

Abstract

We measured the mid-infrared (MIR) extinction using Spitzer photometry and spectroscopy (3.6--37 micron) for a sample of Milky Way sightlines (mostly) having measured ultraviolet extinction curves. We used the pair method to determine the MIR extinction that we then fit with a power law for the continuum and modified Drude profiles for the silicate features. We derived 16 extinction curves having a range of A(V) (1.8-5.5) and R(V) values (2.4-4.3). Our sample includes two dense sightlines that have 3 micron ice feature detections and weak 2175 A bumps. The average A(lambda)/A(V) diffuse sightline extinction curve we calculate is lower than most previous literature measurements. This agrees better with literature diffuse dust grain models, though it is somewhat higher. The 10 micron silicate feature does not correlate with the 2175 A bump, for the first time providing direct observational confirmation that these two features arise from different grain populations. The strength of the 10 micron silicate feature varies by $\sim$2.5 and is not correlated with A(V) or R(V). It is well fit by a modified Drude profile with strong correlations seen between the central wavelength, width, and asymmetry. We do not detect other features with limits in A(lambda)/A(V) units of 0.0026 (5--10 micron), 0.004 (10--20 micron), and 0.008 (20-40 micron). We find that the standard prescription of estimating R(V) from C times E(K_s-V)/E(B-V) has C = -1.14 and a scatter of $\sim$7%. Using the IRAC 5.6 micron band instead of K_s gives C = -1.03 and the least scatter of $\sim$3\%.