Ashes of FIRE: Modeling Dust Grain Size Evolution in the Local Group with FIRE

Abstract

We introduce a new, discretized grain size evolution model, incorporated into the GIZMO code and coupled with FIRE-3 stellar feedback and ISM physics, to investigate variations in dust abundance, chemical composition, and grain sizes observed in the Local Group. This model tracks the size evolution of specific dust species, and includes stellar production of dust, dust growth through gas-phase metal accretion, dust destruction by sputtering, SNe shocks, and astration, grain-grain collisional shattering and coagulation, and turbulent dust diffusion. Using idealized galaxy simulations, we test the dependence of MW dust properties on variations in each dust process and find that our model uniquely predicts a bimodal grain size distribution. This bimodality is due to our simulation's ability to resolve each dust process and where they occur in the ISM, unlike other works. We find that Local Group dust abundances are determined by dust growth and destruction, with little dependence on coagulation or shattering, explaining why models that do not include these processes can match abundance observations. We also find that variations in Local Group extinction curve slopes are determined by coagulation, with inefficient coagulation leading to steeper slopes. However, inefficient coagulation also results in stronger extinction curve bumps, which are not observed. We also do not predict a population of very small (${<}1$ nm) carbonaceous grains, required for MIR emission features, due to their rapid growth by accretion. These results highlight the possible necessity of ``top-down'' PAH formation from preexisting grains as a means to inhibit carbonaceous dust growth.

Figures (17)

• Figure 1: Pictorial representation of each dust process included in our dust evolution model and the ISM phase where they typically dominate. For each process, we include diagrams for how an initial MRN grain size distribution ( grey solid) would change assuming typical gas properties and timescales for each ISM phase. We also show this change for silicates ( blue dashed), carbonaceous ( red dash-dotted), and metallic iron ( green dotted) species since each process depends on dust physical properties as summarized in Table \ref{['tab:dust_quantities']}. Dust Creation: Dust is initially created in the metal-rich ejecta of SNe and AGB winds, with gas-phase metals directly condensing into solid dust grains. Stardust is believed to be large $(a\gtrsim0.1\micron)$ in size but may shatter in SNe. These dust grains are injected into the ISM where they are exposed to mass and number altering processes. Thermal Sputtering: Dust grains residing in hot, halo gas are eroded by collisions with energetic protons and electrons, shrinking and ultimately destroying them. SNe Shocks: As SNe shocks propagate through the ISM, they shatter and sputter dust grains residing in said gas. Shattering: In hot, turbulent gas, dust grains have large relative velocities which cause colliding grains to shatter into smaller grains, shifting grain mass from large to small sizes. Gas-Dust Accretion: In cool, dense gas, gas-phase metals have sufficiently low energy that when they collide with a dust grain, they stick and grow the grain. Coagulation: In molecular clouds, dust grains have low relative velocities allowing colliding grains to stick together, becoming one larger aggregate, shifting grain mass from large to small sizes. Astration: As gas cools and collapses to form stars, the dust residing in the gas is destroyed and contributes to the stellar metallicity.
• Figure 2: Gas phase diagram for all gas in our m12_lowres simulation at simulation end. (left) Phase-space diagram using volume-averaged density and temperature tracked in the simulation. (middle) Median sonic Mach number tracked in the simulation. (right) Phase-space diagram using rms density and effective temperature, factoring in sub-resolved clumping of gas following Appendix \ref{['app:gas_clumping']}. $\mathcal{M}$ rapidly increases in cool, dense gas, increasing the rms density and decreasing the effective temperatures beyond those tracked by the volume-averaged quantities in our simulation.
• Figure 3: Predicted coagulation velocity thresholds from chokshi_1993:DustCoagulationdominik_1997:PhysicsDustCoagulationyan_2004:DustDynamicsCompressible for interactions between grains of size $a_k$ and $a_j$ of the same dust species. We show the thresholds for silicate ( blue line), carbon ( red dashed), and metallic iron ( green dotted) species with corresponding shattering velocity thresholds ( grey). $a_k$ is set to 0.1 ( thick) and 0.001 $\micron$ ( thin). Larger grains have lower coagulation thresholds, and carbon and metallic iron have higher coagulation thresholds than silicate.
• Figure 4: Image and projections of our idealized MW-mass spiral galaxy. ( left) Mock SDSS u/g/r image following the lupton_2004:PreparingRedGreenBlueImages color algorithm created with SKIRT utilizing the local D/Z produced by our model, assuming MW dust opacities. ( left middle) Stellar surface density projection. ( right middle) Dust surface density projection. ( right) Stellar surface density projection for stars formed in the last 10 Myr.
• Figure 5: Resulting median and 16/84th percentiles of D/Z and element depletion into dust trends with neutral gas density for our Fiducial model. We include fits to sight line depletions from jenkins_2009:UnifiedRepresentationGasPhase ( thick black), estimates for the WNM depletions ( diamond), and an interpolation to the Jenkins' relation ( black-dotted). These trends are aggregated for D/Z. Due to the scarcity of C depletion observations, we show individual sight lines from jenkins_2009:UnifiedRepresentationGasPhase and parvathi_2012:ProbingRoleCarbon as an upper limit and a range of expected C depletions in dense gas from observations of C in CO irvine_1987:ChemicalAbundancesMolecularvandishoeck_1993:ChemicalEvolutionProtostellarvandishoeck_1998:ChemicalEvolutionStarForminglacy_1994:DetectionAbsorptionH2. Overall the Fiducial model reproduces observed element depletions and aggregated D/Z trends, but deviates from observed O and Fe depletions due to only considering silicates as O-bearing dust and possibly underpredicting Fe-bearing dust accretion rates respectively.