3.1. Dust sources and relative contributions
It is clear that there is no universal dust model that can be applied to a galaxy as a whole, or to galaxies with different evolutionary histories. Consider the production of dust from an evolving single stellar population (SSP) with masses between ~ 0.8 and 40 M. The first dust that will be injected into the ISM will probably be formed in late-type Wolf-Rayet (WR) stars. Dust formation has only been observed to occur in the coolest C-rich stars of type WC8 and WC9. At least in a few cases, the formation of dust in these objects was induced by the interaction of the WR ejecta with the wind from a companion O-star . WR stars are however minor sources of dust that overall contribute less than 1% of the total mass of dust injected by supernovae and AGB stars into the ISM [35, 19]. After about 5 Myr the first stars of the SSP will undergo core collapse giving rise to Type II supernova (SN) events. SN ejecta contain layers that are C- and O-rich, which are not intermixed on a molecular level. Consequently, SNe can produce both carbon and silicate type dust particles . Stars with masses below ~ 8 M will undergo the AGB phase, lose mass, and evolve into white dwarfs. Figure 2 depicts the carbon and silicate yields from AGB stars with an initial solar metallicity. Stellar yields were taken from Marigo . Stars with a C/O > 1 ratio in their ejecta were assumed to condense only carbon dust, whereas stars with a C/O < 1 ratio were assumed to condense only silicate type dust. The yields were calculated assuming a condensation efficiency of unity in the ejecta. The mass range of carbon producing AGB stars is between ~ 1.6 and 4 M, a range that widens at lower stellar metallicities . So carbon dust from AGB stars will first be injected into the ISM after about 200 Myr, when ~ 4 M stars evolve off the main sequence.
Figure 2. The carbon and silicate yield from AGB stars, based on the Marigo  yields.
The AGB yields depicted in Figure 2 represent an idealized situation. In reality, AGB stars undergo thermal pulsations (the explosive ignition of the He-rich shell), that cause the convective mixing of C-rich gas with the outer stellar envelope. After repeated thermal pulses a star can evolve from an O-rich giant to a C-rich star. The changing composition of the stellar envelope will affect the chemistry of dust formation. Detailed kinetic nucleation calculations (Ferrarotti & Gail ) show that AGB stars of a given mass can indeed form both, carbonaceous and silicate, type dust particles. Including the effect of radiation pressure on the newly-formed dust on the dynamics of the envelope, they find AGB yields that are significantly lower, by factors between 3 and 10, from those presented in Figure 2. A similar conclusion was reached by Morgan & Edmunds  using a simpler model for dust formation in AGB stars.
Figure 3 shows the carbon and silicate yields from both, AGB stars and SN II, weighted by the stellar initial mass function (IMF), taken to be the Salpeter IMF between 0.7 and 40 M. SNe yields were taken from Woosley & Weaver , and a condensation efficiency of unity was adopted in calculating the dust yield. The figure shows that the main contributors to the interstellar carbon abundance are low mass AGB stars, whereas SN II are the main contributors to the silicate abundance in the ISM. We emphasize however, that the yields presented in the figure are ideal ones, and that the actual yield of dust in SNe and AGB stars is still highly uncertain (see Section 4 below).
Figure 3. IMF-weighted carbon and silicate yields from AGB stars and Type II supernovae. SN yields were taken from Woosley & Weaver .