Evolution of an Embedded Super Star Cluster
A full description of the models may be found in the paper.
The three-dimensional code used to run these models allow for arbitrary
geometries and multiple dust species. The models were run using a spherical
geometry with varying inner shell radius. This shape approximates how a
super star cluster evolves out of its embedding cocoon:
We run the models with large grains as well as polycyclic aromatic
hydrocarbon and very small grain templates (Draine & Li
2007).
Models were run with three different outer cocoon radii were used: 25,
50, and 100pc. Five different instantaneous star formation
efficiencies (SFEs; 5%, 10%, 15%, 25%, and 50%) were used, where star
formation efficiency is defined as the mass in stars divided by the
sum of the mass in stars and gas at formation. Models were run with
entirely smooth dust distributions, as well as a varying fraction of
clumpy dust. Clumpy dust distributions make interpretation of the data
difficult, because the observed SED varies with sightline.
The above plot shows the evolution of the embedded cluster's spectral
energy distribution (SED) as the inner radius of dusty envelope moves
outward.
The clumpy dust distribution's affect on a 3-D model's SED with
sightline is apparent in this figure. The gray lines are the SEDs
along all two-hundred computed sightlines through a clumpy embedding
envelope, while the thick black line is the average SED. Near-IR and
Spitzer IRAC/MIPS points are shown as diamonds.
In order to break the degeneracies introduced by a clumpy dust
distribution, diagnostics were developed to recover the input
parameters.Given that the value of the input parameters and the
viewing angle can both significantly change an embedded source's SED
and colors, there is a need to determine which, if any, photometric
measurements can be used to reliably constrain the physical
geometry. For a given input parameter (SFE, clumpiness, inner radius,
or outer radius), the following calculation was made: for each color,
the mean color and standard deviation was measured with the input
parameter fixed and everything else (i.e. the other input parameters
and all viewing angles) variable. This was done at each fixed input
parameter value (e.g. for each of the five different SFE
values). Finally the difference in the means is divided by the
greatest of the standard deviations - this is a measure of how much
the input parameter affects the color, compared to the other input
parameters and the viewing angle ambiguities.
This analysis is related to a Principal Component Analysis, but
adjusted to our goal of finding the minimal set of colors with maximal
physical diagnostic power. Rather than solving freely for eigenvectors
of the model set in color space, which may not directly correspond to
the physical variables, we hypothesize that a principal component
exists for which the eigenvalue would directly correspond to a
physical parameter, and then measure the projection of that
hypothetical component onto each of our color axes. This process
allows of course that there is no such `physically diagnostic
principal component,' in which case the variation of all colors with
the physical parameter would be small compared to the standard
deviation in the color due to other causes. Below are maps that locate
the best infrared colors (y-axis magnitude minus x-axis magnitude)
that can be used to recover a given input parameter; maximal physical
diagnostics are shown in white.
The best diagnostic for recovering the SFE is [70]-[160].
The best diagnostic for recovering the fraction of the envelope that
is clumpy is [3.6]-[5.8].
The best diagnostic for recovering the radii of the envelope is [8.0]-[24].
Last modified
May 8, 2011 at 22:22:05 EDT