Previous (Working) Topic: Clusters as Population Tracers Previous Topic: Open Clusters Lecture Index Next Topic: Integrated Light Evolution

ASTR 5610, Majewski [SPRING 2018]. Lecture Notes

## CMDs, Population Synthesis and Dwarf Galaxies

Quantitative Analysis of the CMDs of Simple Composite Populations:

Star Formation Histories From Population Synthesis

Example: Study of the Leo I dwarf spheroidal by Gallart et al. (1999, AJ, 118, 2245)

Here is the CMD of the dSph Leo I.

• Note: Most distant known Milky Way satellite, at 270 kpc.
##### CMD of the Leo I dSph from Gallart et al. (1999, AJ, 118, 2245).

Immediately you should be able to see the unusual nature of this CMD compared to the globular cluster CMDs we have looked at so far.

• What makes the galaxy's CMD unusual?
• What can you infer about the SFH immediately?
To understand this diagram in detail requires a more quantitative approach... Population Synthesis.

First, create library of SSPs of given age, [Fe/H], and IMF:

• Select [Fe/H], IMF and create a ZAMS.
• Evolve the SSP to a given age.

##### Models for a CMD computed for a population with constant SFR from 15 Gyr ago to the present (with Z=0.0004) and no binary stars. Note the sequence of ages in both the MS and the subgiant branch and, although less definite, also in the RC and HB. (Note: The models correspond to [Fe/H]=-1.7.) From Gallart et al. (1999, AJ, 118, 2245).

To mimic real data, one then subjects the SSP models to errors of the type one would expect.

• "Fuzzes out" the model features a bit (see below).
Next, define CMD regions sensitive to age variations. For example:

##### Adapted from Gallart et al. (1999, AJ, 118, 2245).

The goal is to count and compare stars in the CMD regions between the real data and various combinations of superposed SSP models. For these combinations, the different SSPi are varied in their relative density normalization, a.

• In the simple case here, where the SSPi are only differentiated by age (not metallicity), the Star Formation Rate, SFR(t), is a description of the variation of the normalization parameter a as a function of time (i.e., as function of discrete model SSPi ).
Then, the number of stars in given region j :

where ai are the coefficients showing the relative SFR(ti ) in m different time intervals ti , and Nij is the number of stars contributed to box j in time interval ti for constant SFR(ti ).

For each SFR(ti ) --> family of ai , compare to data N0,j :

where N0,j is the number of stars observed in box j, and l is a normalization factor to ensure the model gives the correct total number of stars in the entire set of regions of the CMD being tested:

and ν is the NDOF = n - 1.

As may be seen, this mechanism produces reasonable results!

##### Comparison of observed Leo I CMD to a model version created via population synthesis by Gallart et al. (1999). The inset in the right panel shows the derived SFR(t) = star formation history.

But:

• Note that Leo I is somewhat unique in that there seems to be little chemical evolution in the galaxy -- i.e., can get away with no [Fe/H] variation.
In reality, models used by Gallart et al. and other groups are more complex than described, and incorporate as variables:

• SFR(ti)

• Z(ti) - Chemical enrichment law

• IMF - ξ() ∝ x (e.g. x = -3.2 above)

• β(f,q) - Binary star function: fraction f of stars with mass ratio distribution q

Gallart et al. 50 combinations of Z, IMF, β; total of 6x107 models of ai <--> SFR

• Under such circumstances, one always worries about uniqueness problems.

Other SFHs

Some examples of other various dwarf galaxy/composite stellar population systems with different SFHs including continuous and bursty versions.

##### Courtesy Carme Gallart.

As can be seen, many of the other Local Group galaxies also contain SFHs that are more complex:

##### (Left) HST CMD of Sextans A by Skillman and collaborators. Boxes used to isolate specific regions for analysis are shown. (Right) Hodge Population Box derived from this CMD by Dolphin et al. (2005).

Another, more complicated, example from Siegel et al.'s (2007, ApJL, 667, L57) analysis of the Sagittarius dSph and its core (M54):

Here is the resulting "Hodge Population Box":

Dwarf Spheroidals, Dwarf Ellipticals and Dwarf Irregulars

(with thanks to Evan Skillman for helpful images and notes).

As established early in this course, there are a variety of dwarf galaxy types in the Local Group.

• Why is this?? Is there some over-riding property that causes the differences?

Comparison of properties:

• Structure:

• Long been known that dIrr and dSph have similar underlying structures (Faber & Lin 1983, Kormendy 1985).

• Dynamics:

• Not as clear as we once thought?

• Latest evidence is that dIrrs and some dEs have similar kinematics.

E.g., some Virgo dEs are found to have significant rotational support and follow the Tully-Fisher relation.

Rotating dEs look similar to those without rotation (but more work needed).

##### From van Zee, Barton & Skillman (2004).
• Gas content:

• dIrr: M(HI)/L ~ 1

• Transition types: M(HI)/L ~ 0.1-1

• dSph: M(HI)/L < 0.01

##### Image courtesy of Evan Skillman.
• Star formation histories:

We have seen before the star formation histories of Local Group dwarfs, as derived from specific types of indicators:

##### From Grebel (1997, Reviews in Modern Astronomy, 10, 29).

Here are more updated versions of the SFHs as derived from population synthesis, in many cases from HST data:

-----------------------------------------------------------------------------------------------------------------------------------------------------

##### Hodge population boxes for Local Group galaxies derived from population synthesis by Dolphin et al. (2005). Dwarf spirals, dIrr and transitional types above the dashed line and dE and dSph galaxies below. Images courtesy of Evan Skillman.

Note that all dwarfs seem to show evidence of an old population (>10 Gyr old).

But it is immediately obvious that a major difference between the different types of dwarfs is their current star formation rate.

• dIrr show active star formation now.

• Transition types have no HII regions, modest recent SF.

• dSph have no current star formation.

Another difference seems to be in the overall trend of the SFH:

• dIrr tend to show more or less continuous SFH.

• Transition types show a declining SFH.

• dSph show a wide variety of SFH, but typically more rapidly declining, with perhaps some bursts at later times.

Comparing SFH can be difficult when visualizing in differential form:

• Different masses mean different normalizations for SFR.

• In addition, errors in derived SFR from time bin to time bin can be highly correlated.

For this reason, it can be more useful to compare SFH in cumulative form (the "Holtzman Diagram").

##### (Left) Holtzman diagram for different dIrrs. (Right) Same for different dSph. The greater similarity and more typically constant SFH of dIrr compared to dSph is obvious.
But is there a real difference in dSph and dIrr types?

Not always: Note the comparison of dIrr IC1613 with dSph Leo I we saw above?:

##### (Left) CMD for IC1613 from Skillman and collaborators. (Right) Comparison of derived SFH and metallicity evolution for IC1613 against that for the dSph Leo I. The SFH's appear identical, except for the very recent SF in IC1613.
Clearly one of the major differences between late and early type dwarf galaxies is the current SFR, which has a profound effect on the present appearance of the galaxy.

• Thus, there can be close similarity in SFH between two galaxies, but the morphology is dictated by just the very most recent SF.

• But we have seen that on average, dSphs have had a dramatic decrease in over SFR with time. What causes the differences in SFR?

• Reionization?:

• Recall our discussion of the "missing satellite" problem we discussed that perhaps only the largest CDM subhalos form stars and become luminous.

• This process requires something to suppress SF, and one way is to have heating of the material to keep it from collapsing.

• Early sources of intense ionizing background radiation are the first round of quasar and supernova formation (from Pop III) in the Universe, which create the epoch of reionization.

According to WMAP, this occurred at z = 11, or 600 million years after the Big Bang.

##### The age of reionization. S. G. Djorgovski et al. & Digital Media Center, Caltech.
• An already ionized universe is transparent to extreme UV radiation, and this energy goes into destroying cold molecular clouds. Only the densest clumps before reionization had time to collapse.

• It does look like some dIrr, like Sex A, had very delayed SF, as if it were suppressed.

• But this is not universal.

• Blow-Out?:

• Galaxies with particularly strong early SF might effectively blow-out residual cold gas, squelching subsequent star formation (MacLow & Ferrara).

• Many dSphs seem dominated by an early SF burst and not much later on.

• dIrr often have more even and continuous SF.

• Morphology-Density Relationship:

• dIrr: Occur in many environments.

• Transition types: Higher density regions.

• dSph: Near large spirals.

##### Image from Skillman, Cote & Miller (2003).
Clearly environment matters.

• More frequent tidal encounters ("tidal stirring") may accelerate SF in bursts and earlier exhaustion of fuel.

##### Face-on (top) and edge-on (bottom) views of the evolution of a high surface brightness, dwarf disk galaxy into a dwarf spheroidal due to tidal stirring induced while in an orbit around a MW-sized halo, from simulations by Mayer et al. (2001). The tidal field from the host galaxy induces the formation and subsequent buckling of a bar in the dwarf, which dynamically heats the system and, over the course of Gyr, transforms it into a pressure-supported galaxy. Ths heating induced by bar buckling is found to be more important than direct heating by the tidal field. In the figure the boxes are about 30 kpc on each side. Modified from Mayer et al. (2001).
• Tidal Stirring:

• Dwarf disk galaxy in tidal field develops a bar.

• The bar subsequently buckles, and this dynamically heats the system.

• Transforms rotating disk into a pressure-supported dSph type system.

• Can create dSph systems with significant ellipticities, which are actually observed in the MW system (Lokas et al. 2010).

• Or fuel may be stripped out by ram pressure interaction with hot coronal gas around large systems.

(Remember the Magellanic Stream.)

Aside: Simulations of dwarf satellites starting as a baryonic (luminous) component buried inside a larger CDM halo cocoon show that when undergoing significant tidal stripping, first the dark matter is peeled off and eventually these satellites actually evolve into systems that resemble mass follows light (MFL) (Mayer et al. 2001, Klimentowski et al. 2007), and after halos are stripped to the point that stars are also being lost, then the system behaves basically as MFL (Bullock & Johnston 2005):

##### A comparison between the final total mass ( solid lines) and the stellar mass profiles ( dot-dashed lines) for N-body, smoothed particle hydrodynamical simulations of initially rotationally supported, two-component (i.e., luminous matter embedded in an extended dark matter halo) dwarf galaxies that become tidally stirred and stripped while orbiting in a MW-like potential, by Mayer et al. (2001). The models include a low surface brightness (LSB) dwarf with disk scale length 4.4 kpc that eventually turns into a dSph, a high surface brightness (HSB) dwarf with a 2 kpc disk scale length that eventually turns into a dwarf elliptical, and a very faint dIrr, like the LG system GR8, with disk scale length 76 pc and a very high central dark matter density, that evolves into an extremely dark matter dominated dSph, like Ursa Minor and Draco. Note that only in the last case does the dark matter still greatly ``cocoon" the baryonic matter, whereas, particularly in the LSB/dSph case, the end state is essentially a mass-follows-light system.

Area-Resolved Population Synthesis and SFHs

A new level of CMD population synthesis, possible now with large area, deep surveys of nearby galaxies, is studying populations by region in a galaxy.

• Example 1: Population age variations shown by different classes of age indicator.

A study of the populations in the Fornax dSph by Grebel & Stetson (1999, The Stellar Content of Local Group Galaxies, IAU Symp. 192, eds. P. Whitelock & R. Cannon, p. 165).
• One of the most massive dSphs in the Local Group.
• One of few dSphs that has own globular cluster system (five globulars).
• The CMD for Fornax looks rather different in different parts of the dSph.
• ##### CMD of the Fornax dSph from Grebel & Stetson (1999).
• It is apparent that the center of the dSph has a CMD with younger populations (left panel above) than at larger radii (see right panel).
• ##### The distributions of different types of stars -- young MS, young red clump, older red clump and RR Lyrae -- in the field of the Fornax dSph. From Grebel & Stetson (1999).
• Fornax clearly has an age gradient, with younger populations concentrated in the core.
• Fornax contains the youngest population ever found in a dSph, 100-200 Myr.

Note how the younger population is more irregularly distributed than the older populations.
• What is amazing is that, despite having recently formed stars, Fornax does not seem to have any gas!

The same is true of almost every dSph that has been searched for gas -- even those with youngish pops in them (e.g., Sextans, Sculptor, Leo I, Sagittarius) have no gas.

How can there be such careful tuning for us to be at a "special epoch"??
• Why are there no dSph's forming stars now? Is this another "fine tuning" problem?

Or is something else going on??? -- Think of a possible selection effect that may explain this phenomenon.

• Example 2: Population metallicity and second parameter variations based on changes within classes of tracers.

Study of the HB variation, and the RGB variation, in the Sculptor dSph galaxy by Majewski et al. (1999, ApJL, 520, 33).

• Sculptor clearly has a wide RGB indicating a spread in metallicities.
• Close inspection of the CMD suggests it is more bimodal than continuous.
• ##### CMD of the Sculptor dSph by Majewski et al. (1999). The two globulars shown "bracket" the RGB of the dSph and give some idea of a two population, composite model for the system.
• That two populations of different metallicity are present is suggested by the different magnitudes of the RGB bump, which, as I pointed out in another class, is also an [Fe/H] indicator based on its magnitude relative to that of the HB.
##### HB of the Sculptor dSph by Majewski et al. (1999). The ZAHB models show [Fe/H]=-2.3, -1.9 and -1.5 for two different [O/Fe] assumptions for the different panels. The hash marks and red ends of the ZAHBS show 0.8 and 0.85 solar mass ZAHB stars corresponding to 15 and 12 Gyr population maximum possible ZAHB masses, respectively, but stars should never reach these points because they represent no mass loss.
• There are two HB populations evident, one rather red, and one rather blue.

These can be associated one each to the two different RGBs seen (can you guess how?).
• Even so, the red HB, associated with the more metal rich ([Fe/H] ~ -1.5) population is still too red for this metallicity and so has a second parameter effect.

Thus, Sculptor is unusual in having one population with no second parameter effect, and another one that does express a second parameter effect.
• The Sculptor system has an internal second parameter problem.

• We can count the numbers of red and blue HB stars, and red and blue RGB stars, as a function of radius.

The figure below shows the HB morphology (filled circles) and the RGB morphology (open circles) as a function of radius form the center of Scl.

Recall the definition of the HB morphology index:

HB = (B-R)/(B+V+R)

where B=number of BHB, R=number of RHB and V=number of RR Lyrae stars.
• ##### by Majewski et al. (1999).
• The existence of a radial gradient in the kinds of HB and RGB stars is evident.

--> Likely shows that the more metal rich, second parameter, and younger (by about 3 Gyr) population is more concentrated.

• Example 3: Population gradients like those seen in Fornax and Sculptor seem to be a common characteristic of dSph galaxies.

Study of the HBs and RGBs of nine examples by Harbeck et al. (2001, AJ, 122, 3092):

• CMDs of nine dSphs collected, then sliced up similarly to Majewski et al. example.
• ##### CMDs for nine dSph galaxies, showing the variety of HB types and how the HB and RGB samples are divided by Harbeck et al. (2001).
• Cumulative distributions (normalized) indicate the same population trend seen in Sculptor example:

##### Radial trends of the HBs for nine dSph galaxies by Harbeck et al. (2001). Click here to see similar results for the RGB stars.
RHB stars more centrally concentrated than BHB --> suggests more metal-rich or second P populations more concentrated.

Redder (more metal rich) RGB stars more centrally concentrated than bluer (more metal poor) RGB stars.
• Trend: Tendency is for more metal-rich, younger populations to be more centrally concentrated in dSph galaxies.
• Same also observed in dIrr galaxies:

##### Holtzman diagram of different fields in the dIrr NGC 6822, showing older mean populations in the outer parts.
• May be related to successive starbursts having progressively less available gas...

• Gas locked up in stellar remnants.
• Gas expelled by kinetic energy of supernovae, stellar winds past escape velocity.
... which must condense to the center of the gravitational well before concentrated enough to form a new starburst.

• Example 4: The Magellanic Clouds are a particular favorite of this area-resolved population synthesis work:

• Nikolaev & Weinberg (2000, ApJ, 542, 804) mapping of characteristic populations in the 2MASS infrared CMD of the LMC

The plots below show the 2MASS CMDs for a high latitude population and for the LMC. The color-color diagrams for each are also shown.
• ##### The LMC CMD from 2MASS and broken into separate populations by Nikolaev & Weinberg (2000).

• The distribution of the stars in the selected regions of the CMD by Nikolaev & Weinberg (2000):

##### Spatial distributions of the stars in the regions shown in the 2MASS LMC CMD by Nikolaev & Weinberg (2000). Note that some of the regions these authors selected correspond to regions of the CMD dominated by non-LMC stars.

• Again, note variations of distribution by stellar type.

Note that some of the parts of the CMD are dominated by the field stars of the Milky Way, and these show no special concentration as expected for LMC stars.
• When many stars are present, as in the huge catalogues now available for the Magellanic Clouds, one can do population synthesis region by region in the galaxy.

##### Slide from a presentation by Jason Harris at Ringberg Castle, Dec. 2004. Courtesy Jason Harris.

Click to http://www.noao.edu/staff/jharris/Viz/smcsfh.mpg and http://www.noao.edu/staff/jharris/Viz/smc_phot.mpg to see movies of the star formation history of the Small Magellanic Cloud by Harris & Zaritsky.
• ##### SFH of the SMC from a presentation by Jason Harris at Ringberg Castle, Dec. 2004. The SF gap apparent in the movie is obvious (note change in horizontal scale between left and right panels of the SFH. Courtesy Jason Harris.

• Example 5: The warning provided by the Sagittarius dSph:

Tidal disruption creates mass loss at largest radii -- so if there are metallicity gradients, bias toward stripping of more metal poor populations.

##### From Mei-Yin Chou et al.'s (2007) work. The top panel shows the metallicity distribution function (MDF) of the main Sgr body, and the next two panels show the MDFs at increasingly larger separations from the Sgr center.

All material copyright © 2003,2006,2008,2010,2012,2018 Steven R. Majewski. All rights reserved. These notes are intended for the private, noncommercial use of students enrolled in Astronomy 551 and Astronomy 5610 at the University of Virginia.
 Previous (Working) Topic: Clusters as Population Tracers Previous Topic: Open Clusters Lecture Index Next Topic: Global Properties and Spectral Evolution of Stellar Populations