2013ApJ...766..121M

We have conducted an archival Spitzer study of 38 early-type galaxies in order to determine the origin of the dust in approximately half of this population. Our sample galaxies generally have good wavelength coverage from 3.6 µm to 160 µm, as well as visible-wavelength Hubble Space Telescope (HST) images. We use the Spitzer data to estimate dust masses, or establish upper limits, and find that all of the early-type galaxies with dust lanes in the HST data are detected in all of the Spitzer bands and have dust masses of ∼10⁵-10^6.5 M_☉, while galaxies without dust lanes are not detected at 70 µm and 160 µm and typically have <10⁵ M_☉ of dust. The apparently dust-free galaxies do have 24 µm emission that scales with the shorter-wavelength flux, yet substantially exceeds the expectations of photospheric emission by approximately a factor of three. We conclude this emission is dominated by hot, circumstellar dust around evolved stars that does not survive to form a substantial interstellar component. The order-of-magnitude variations in dust masses between galaxies with similar stellar populations rule out a substantial contribution from continual, internal production in spite of the clear evidence for circumstellar dust. We demonstrate that the interstellar dust is not due to purely external accretion, unless the product of the merger rate of dusty satellites and the dust lifetime is at least an order of magnitude higher than expected. We propose that dust in early-type galaxies is seeded by external accretion, yet the accreted dust is maintained by continued growth in externally accreted cold gas beyond the nominal lifetime of individual grains. The several Gyr depletion time of the cold gas is long enough to reconcile the fraction of dusty early-type galaxies with the merger rate of gas-rich satellites. As the majority of dusty early-type galaxies are also low-luminosity active galactic nuclei and likely fueled by this cold gas, their lifetime should similarly be several Gyr.