<%BANNER%>

Star Formation in Disk Galaxies: From Kiloparsec to Giant Molecular Cloud Scales

Journal of Undergraduate Research from the Center for Undergraduate Research
PRIVATE ITEM Digitization of this item is currently in progress.
  • STANDARD VIEW
  • MARC VIEW
MISSING IMAGE

Material Information

Title:
Star Formation in Disk Galaxies: From Kiloparsec to Giant Molecular Cloud Scales
Series Title:
Journal of Undergraduate Research
Physical Description:
Serial
Language:
English
Creator:
Shaske, Suzanne N.
Tan, Jonathan C.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Star formation
Pagiant molecular cloudsleoclimate
GMC
cloud collisions
collision rate
velocity dispersion
M33
star cluster
orbital dynamics
Genre:
serial   ( sobekcm )

Notes

Abstract:
Star formation is highly localized in space and time. It occurs mostly in a giant molecular clouds (GMCs), which occupy a very small fraction of galactic volume, and even within these clouds it is concentrated in parsec-scale dense clumps that produce star clusters. There are well-studied observational correlations between the large, kiloparsec-scale star formation activity of galaxies and their gas content, but no theoretical consensus to explain this behavior. In this paper we first test an analytic model for GMC collision rates by comparing the analytical predictions to those observed in a numerical simulation of GMCs suffer frequent mutual collisions every 10-20% of an orbital time. This process could be the link between the kpc scales of galactic orbital dynamics and the parsec scales of star-forming clumps within GMCs. Next, we use star cluster population synthesis models to investigate stochastic star formation models (e.g., ones that are regulated by turbulence) in GMCs. We apply this modeling to cloud populations from simulated disk galaxies for the Local Group spiral M33. For the later case, by comparing to the observed star formation activity from 80-parsec to kiloparsec scales, including its overall efficiency and dispersion, we are able to test the validity of stochastic star formation theories. In particular, the observed GMCs appear to have a larger dispersion in star formation activity than expected from the simple stochastic theory.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
sobekcm - UF00091523_00602
System ID:
UF00091523:00674

  • STANDARD VIEW
  • MARC VIEW
MISSING IMAGE

Material Information

Title:
Star Formation in Disk Galaxies: From Kiloparsec to Giant Molecular Cloud Scales
Series Title:
Journal of Undergraduate Research
Physical Description:
Serial
Language:
English
Creator:
Shaske, Suzanne N.
Tan, Jonathan C.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Star formation
Pagiant molecular cloudsleoclimate
GMC
cloud collisions
collision rate
velocity dispersion
M33
star cluster
orbital dynamics
Genre:
serial   ( sobekcm )

Notes

Abstract:
Star formation is highly localized in space and time. It occurs mostly in a giant molecular clouds (GMCs), which occupy a very small fraction of galactic volume, and even within these clouds it is concentrated in parsec-scale dense clumps that produce star clusters. There are well-studied observational correlations between the large, kiloparsec-scale star formation activity of galaxies and their gas content, but no theoretical consensus to explain this behavior. In this paper we first test an analytic model for GMC collision rates by comparing the analytical predictions to those observed in a numerical simulation of GMCs suffer frequent mutual collisions every 10-20% of an orbital time. This process could be the link between the kpc scales of galactic orbital dynamics and the parsec scales of star-forming clumps within GMCs. Next, we use star cluster population synthesis models to investigate stochastic star formation models (e.g., ones that are regulated by turbulence) in GMCs. We apply this modeling to cloud populations from simulated disk galaxies for the Local Group spiral M33. For the later case, by comparing to the observed star formation activity from 80-parsec to kiloparsec scales, including its overall efficiency and dispersion, we are able to test the validity of stochastic star formation theories. In particular, the observed GMCs appear to have a larger dispersion in star formation activity than expected from the simple stochastic theory.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
sobekcm - UF00091523_00602
System ID:
UF00091523:00674


This item is only available as the following downloads:


Full Text

PAGE 1

University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 1 Star formation is highly localized in space and time. It occurs mostly in giant molecular clouds (GMCs), which occupy a very small fraction of galactic volume, and even within these clouds it is concentrated in parsec scale dense clumps that produce star clusters. There are well studied observational correlations between the large, kiloparsec scale star formation activity of galaxies and their gas content, but no theoretical consensus to explain this behavior. In this paper we first test an analytic model for GMC collision rates by comparing the analytical predictions to those observed in a numerical simulation of GMCs orbiting in a galactic disk. We find the analytical and numerical mode l agree with each other at the ~10% level, confirming that GMCs suffer frequent mutual collisions every 10 20% of an orbital time. This process could be the link between the kpc scales of galactic orbital dynamics and the parsec scales of star forming clumps within GMCs. Next we use star cluster population synthesis models to investigate stochastic star formation models (e.g. ones that are regulated by turbulence) in GMCs. We a pply this modeling to cloud populations from simulated disk galaxies and for the Local Group spiral M33. For the latter case, by comparing to the observed star formation activity f rom 80 parsec to kiloparsec scales, including its overall efficiency and dis persion, we are able to test the validity of stochastic star formation theories. In particular, the observed GMCs appear to have a larger dispersion in star formation activity than expected from the simple stochastic theory. INTRODUCTION GMC collisions in a shearing disk may be an important mechanism for triggering star formation (Tan 2000), since they are a natural mechanism for creating parsec scale, gravitationally unstable dense gas clumps, which then form star clusters. At the same time, the rate of GMC collisions depends on the large scale properties of the galaxy, such as its gas content (especially the fraction of gas in molecular form, H 2 ) and its rotation curve (especially the local shear rate). Thus GMC collisions can help explain why star formation activity is so localized (e.g. Lada & Lada 2003), and yet also correlated with large scale global galactic properties (Kennicutt 1998; Bigiel et al. 2008). We would like to understand in more detail the physics of GMC c ollisions ( Sect. 2 ) and then begin to test whether alternative theories of turbulence regulated (i.e. stochastic, un triggered) star formation can explain observations of the nearby galaxy M33 ( Sect. 3 ). TESTING GMC COLLISION THEORY GMC Collision and Merger Timescales As GMC s travel around a galaxy they collide and merge with other GMCs. These collisio ns have the potential to drive star formation ; therefore it is important to develop and test an analytic theory for this process, which could then hel p us to understand the evolution of galaxies and starbursts. We begin by using data from a numerical simulation of a Milky Way l ike galaxy produced by Tasker & Tan (2009) at a time step of 250 Myr This data was then entered into analytic equations that p redict GMC properties. We first measure the orbital time of a GMC t orb : [1] w here R is the galactocentric radius, v circ Next, we investigated the time it takes for a G MC to collide with another GMC, t coll derived by Tan (2000) to be: [2a] [2b] w here r t is the tidal radius of the GMC N A is the number of GMC s per unit area of the disk, f G is a constant with a typical value of ~0.5 (Gammie et al. 1991), and is We evaluated the analytic estimate o f the collision rate of the GMC s in different a nnuli in the simulated galaxy.

PAGE 2

SUZANNE N SHASKE & DR JONAT HAN C TAN University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 2 These results were then compared to the values of t orb and t coll that were measured directly in th e numerical simulation show n in F igure 1 (Ta sker & Tan 2009). The comparison is show n below in F igure 2 where the blue lines are the collision times measured in the numerical simulation and t he red lines are the collision times estimated from analytic theory F igure 1. Tasker & Tan (2009) numerical simulation of a Milk Way Like disk galaxy at the time step of 200 Myr We find that the analytic theory predicts the GMC collision time to an accuracy of about 10%. Both the theory and the simulation indicate that the GMCs hav e frequent collisions with each other on timescales that are short and a relatively constant fraction of an orbital time. Therefore, this process may be important for injecting turbulent energy into GMCs and triggering star formation. GMC Velocity Dispersion The estimate of the GMC collision rate derived by Tan (2000) assumed a cloud velocity dispersion that is set by mutual gravitational interactions. In the model of Gammie et al. (1991), this velocity dispersion was found to have a form: [3] w here G is the gravitational constant, M cl is the mass of the This velocity is approximately equivalent to the shear velocity at the tidal radius of the GMC. We examined whe ther the GMCs in the simulation of Tasker & Tan (2009) followed this relation. For each GMC we measured its velocity relative to that of a circular orbi t at its radial location. The 3 D velocity dispersion of the cloud population in a given annulus was then calculated. These results are shown in Figure 3. We find that the simulated clouds have a velocity dispersion that scales with a galactocentric radius as R 1/3 consistent with being proportional to the shear velocity at a cloud tidal radius. However, the about 1.8 tim es larger than predicted by Gammie et al. (1991). This may be due to the simulated GMCs having smaller sizes (so smaller damping rates) than assumed in the calculation of Gammie et al., and/or excitation of random velocit ies in the simulation by density structures larger than the GMC scale. Figure 3 GMC Velocity Dispersion The dashed line (top) is obtained from the simulation by Tasker & Tan (2009), the solid red line is a best fit line with a slope of 0.39, and the dotted line (bottom) is the result from the analytic equation proposed by Gammie et al. (1991). Figure 2. Cloud Merger Timescale Comparison. The blue lines (joining triangles) are from the numerical simulation by Tasker and Tan (2009) and the red lines (joining squares) are the analytic estimates. The top set of data is for all GMCs, while the bottom set of data (dashed lines) is for large GMCs (>10 6 solar masses). The error bars indicate Poisson errors given the number of clouds in each annulus.

PAGE 3

S TAR F ORMATION IN D ISK GALAXIES : F ROM K ILOPARSEC TO G IANT M OLECULAR C LOUD S CALES University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 3 THE KENNICUT T SCHMIDT RELATION FROM KILOPARSEC TO GMC SCALES We used cluster population synthesis models to investigate theories for stochastic star cluster formation in GMCs (e.g. Krumholz & McKee 2005) We applied this modeling to cloud populations from the simulated galaxy produced by Tasker & Tan (2009) and to the Local Group spiral galaxy M33 shown in Figure 4 The goal is to understand the normalization and dispersion of the Kennicutt Schmidt relation linking the surface density of star formation rate (SFR) with the surface density of gas in galactic disks ( Kennicutt 1998; Bigiel et al. 2008). Stochastic Star Formation from GMCs Star fo rmation in GMCs appears to occur at a very low rate, converting just a few percent of the cloud mass into stars every local free fall time, t ff (Zuckerman & Evans 1973; Krumhol z & Tan 2007) providing an efficiency per free fall time ff =0.02 [4] At the same time star formation is known to be highly concentrated in star clusters, which appear to be born with an initial clu ster mass function (ICMF) that can be approximated as a power law in the mass range from ~100 to 10 6 solar masses (e.g. Dowell, Buckalew & Tan 2008 ; Larsen 2009 ): [5] The stars themselves are formed with an initial mass function (IMF ) that is also approximately a power law for m asses > ~1 solar mass, up to ~10 0 solar masses (Salpeter 1955). [6] Our approach for modeling star formation in GMCs is to first use equation [4] to predict the SFR. To do this we need to know the cloud mass and density. We then need to know how long a cloud has existed in a state where it is forming stars at this rate. We assume clouds have a lifetime, t GMC,max with fiducial value of 10 Myr. When we model star formation in a GM C population we will assume a uniform, random distribution of cloud ages, t GMC from 0 to t GMC,max Thus the total mass of stars formed in a given GMC is : [7] We create M *,GMC with a Monte Carlo sampling of the ICMF. In our fiducial method (A) we draw clusters one by one from the ICMF until the total stellar mass exceeds M *,GMC at which point we then consider whether the current or previous value is closest to the desired tota l before choosing the closest In a second method (B), we consider the mean cluster mass implied by the ICMF, which is = 920 M sun and then estimate the number of clusters expected in the GMC as equal to M *,GMC /. We then draw this number of clusters from the ICMF, accounting for fractional numbers with a simple probability estimate : e.g. if the expected number of clusters is 3.5, then 3 clusters will be drawn 50% of the time and 4 clusters drawn the remaining 50% of the time. We wrote our own code to carry out this sampling of the ICMF using IDL We also utilized the cluster population synthesis program SLUG (S tochastically Lighting Up Galaxies) developed by Fumagalli et al. (2011). This program constructs clusters by drawing randomly f rom the initial stellar mass function (IMF) and the initi al cluster mass function (ICMF), and tracks a number of observable properties of the star clusters, such as ionizing flux and FUV luminosity. To compare to observed galaxies, we then need to assess t he gas and star formation activity at different scales, dividing the galactic disk into a rectangular grid of various resolutions. St ar Formation in the Simulated Galaxy and Comparison to the Observed M33 Kennicutt Schmidt Relation We use the properties ( mass, size, mean density) and locations of the GMCs in the galaxy simulated by Tasker & Tan (2009) to pl ot the surface density of SFR ( SFR ) versus the surface density of molecular gas ( H2 ) (including Helium assuming n He =0.1n H ) at various sca les ranging from 80 to 1000 pc as seen in the Figure 4 orange points Here we also compare the results to the data f or M33 from Onodera et al. (2010 ) and a djust by 4/3 to utilize an X factor of 4x10 20 cm 2 /(K km/s) ( Gratier et al. 2011 ) and we also adjust by a factor of 1.4 to include He These results are shown by the black points in F igure 4 Making this comparison also requires adjusting the simulated data by cos i, where i=51 degrees is the inclination angle of M33, i.e. s are reduced by this factor. The simulated galaxy was designed to mimic the Milky Way, which causes our results to have much higher values of H2 Furthermore, the SFRs appear higher on the kpc scales in the region of overlap which may be due to ff being <0.02 in M33, given that GMC lifetimes are <10 Myr in M33, or the simulated GMCs are oo dense. Tasker & Tan (2009) favor this latter possibility since this simulation did not include internal feedback fro m star formation acting to disrupt GMCs.

PAGE 4

SUZANNE N SHASKE & DR JONAT HAN C TAN University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 4 Figure 4. Comparison of GMC Based Stochastic SFR From Simulated Galaxy to Measured SFR in M33 The orange dots are the SFR and molecular gas surface densities derived from the simulated galaxy and the dark blue line is the best fit line to this data. The black dots are observed values for M33 from Onodera et al. (2010 ) and the light blue line is the best fit line to the ir data Star Formation in the M33 GMC Population and Comparison to the Observed M33 Kennicutt Schmidt R elation To better compare to M33 we use the GMC catalog of Gratier et al. (2011) for the re gion mapped as shown in Figure 5 where the dotted lines are 1 kpc wide annuli. Each GMC mass, size, and density (and thus free fall time) is estimated and used with equation [4]. The galaxy gridding on scales from 80 to 1000 pc is repeated and the result is shown in Figure 6 as red p oints. These results are a much better match to those from Onodera et al. (2010 ) as show n by the black points. However, the mean SFR is still too high, which is displayed in Figure 7 This suggests again that a smaller star formation efficiency (<0.02) o r a shorter GMC lifetime (< 10Myr) is needed. Figure 5 M33 Galaxy from Gratier et al (2011 ) which show s locations of GMCs in the regions covered by their observations.

PAGE 5

S TAR F ORMATION IN D ISK GALAXIES : F ROM K ILOPARSEC TO G IANT M OLECULAR C LOUD S CALES University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 5 Figure 6 Comparison of GMC based Stochastic SFR to Measured SFR in M33 The red d ots are the SFR derived from the GMC masses measured by Gratier et al. (2011) and the dark blue line is the best fit line to the data. The black do ts are from Onodera et al. (2010 ) and the light blue line is the best fit line to the data

PAGE 6

SUZANNE N SHASKE & DR JONAT HAN C TAN University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 6 Figure 7 Onodera et al. (201 0 ), Tasker & Tan (2009) and Gratier et al. (2011) SFR Means At e ach r esolution the data were split into equal sized bins based on H2 and the logarithmic mean of the SFR in each bin with five or more data points was calculated. The x location of the points refers to the center of each bin The black lines (joining triangles) are the means of the Onodera et al. (2010 ) SFR data. The red lines (joining asterisks) are the means o f the SF R from the Gratier et al. (2011) The gold lines (joining diamonds) are the means of the SFR derived from the simulated galaxy produced by Tasker & Tan (2009) We can also consider the logarithmic dispersion of SFRs as an additional diagnostic. The dispersion is taken from the best fit lines shown in Figure s 4 and 6 We measure this in fixed intervals of H2 for each of the three data sets as shown in Figure 8 There is evidence for larger dispersions in the observed SFRs (Onodera et al. 2010 ) when compared to those with the Gratier GMCs especially on the GMC scale. This may indicate that the simple stochastic model for star formation may need to be revised. Mod els of star formation triggered by highly stochastic events, such as GMC collisions would increase the dispersion of SFRs. Our future direction is to investigate such models and also to study them on the individual cloud scales.

PAGE 7

S TAR F ORMATION IN D ISK GALAXIES : F ROM K ILOPARSEC TO G IANT M OLECULAR C LOUD S CALES University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 7 Figure 8 Onodera et al. (2010 ) Tasker & Tan (2009), & Gratier et al. (2011) SFR Dispersion At e ach r esolution the data were split into equal sized bins with five or more data points based on H2 an d the logarithmic dispersion of the SFR with respect to the best fit line was calculated The x location of the points refers to the cent er of each bin The black lines (joining triangles) are the disper sion of the Onodera et al. (2010 ) SFR data. The red lines (joining asterisks) are the dispersion of the SFR from the Gratier et al. (2011 ) The gold lines (joining diamonds) are the dispersion of the SFR derived from the simulation produced by Tasker & Tan (2009) CONCLUSIONS We have demonstrated that GMC collision rates can be predicted to an accuracy of about 10% using a simple analytic formula and by testing this for different locations and cloud masses in a hydrodynamic simulation of a disk galaxy. We predict that this p rocess occurs relatively frequently and thus could be the most important mechanism driving star formation activity in disk galaxies. Also, we have tested GMC scale models o f star formation that involve a fixed star formation efficiency per local free fall time and an assumed GMC lifetime in both a simulated galaxy and the observed M33 GMC population We found that relatively low efficiencies (<0.02) or shorter GMC lifetimes (< 10Myr) may be required. This i s the first time such joint constraints have been derived for GMCs. The observed dispersion in SFRs is larger than predicted by the simple stochastic model, suggesting additional triggering mechanisms for star formation activity may be needed. ACKNOWLEDGE MENTS We thank Elizabeth Tasker for providing data from her numerical simulation. We would like to acknowledge the support from the University Scholars Program at UF and an NSF REU Supplement grant.

PAGE 8

SUZANNE N SHASKE & DR JONAT HAN C TAN University of Florida | Journal of Undergraduate Resea rch | Volume 14, Issue 3 | S ummer 2013 8 REF E RENCES Bigiel F., Leroy, A., Walter, F., Brinks, E., De Blok, W. J. G., Madore, B., & Thornley, M. D. 2008, AJ, 136, 2846 Dowell, J. D., Buckalew, B. A., & Tan, J. C. 2008, AJ, 135, 823 Fumagalli, M., da Silva, R., Krumholz, M. R., & Bigiel, F. 2011, ASPC, 440, 155 Ga mmie, C. F., Ostriker, J. P., & Jog, C. J. 1991, ApJ, 378, 565 Gratier, P., Braine, J., Rodriguez Fernandez, N. J. et al. 2011, A&A, in press, arXiv1111.4320 Kennicutt, R. C., Jr. 1998, ApJ, 498, 541 Krumholz, M. R., & McKee, C. F. 2005, ApJ, 630, 250 Krum holz, M. R., & Tan, J. C. 2007, ApJ, 654, 304 Lada, C. J., & Lada, E. A. 2003, ARA&A, 41, 57 Larsen, S. S. 2009, A&A, 494, 539 Onodera, S., Kuno, N., Tosaki, T. et al. 2010, ApJ, 722, L127 Salpeter, E. E. 1955, ApJ, 121, 161 Tan, J. C. 2000, ApJ, 536, 173 Tasker, E. J., & Tan, J. C. 2009, ApJ, 700, 358 Zuckerman, B., & Evans, N. J., II. 1974, ApJ, 192, L149