The Massive Star Content of M31. Phil Massey Lowell Observatory The 2nd MMT Science Symposium

The Massive Star Content of M31 Phil Massey Lowell Observatory The 2nd MMT Science Symposium

The Team Maria Drout Knut Olsen Me Nelson Caldwell Kathryn Neugent Dave Silva Emily Levesque Georges Meynet Bertrand Plez Susan Tokarz

What We re Doing and Why

What We re Doing and Why Overall goal is to determine the number of massive stars in different evolutionary stages in galaxies of the Local Group covering a 20x factor in metallicity, providing a very exacting test of current stellar evolutionary theory.

What We re Doing and Why

What We re Doing and Why Having good massive star evolutionary models not only helps us understand massive stars better but also is crucial for:

What We re Doing and Why Having good massive star evolutionary models not only helps us understand massive stars better but also is crucial for: measuring the IMF in mixed-age populations where you need to know the relative stellar lifetime accurately any population synthesis studies of galaxies, where codes such as STARBURST99 are no better (and no worse) than the stellar evolution models that go into them.

M31 with 2x solar metallcity

NGC 206 the η and χ Per of M31! M31 with 2x solar metallcity

Massive Star Evolution Primer Massive stars spend 90% of their lives as OB stars. The other 10% is spent in various He burning phases, including yellow supergiants, red supergiants, and/or Wolf-Rayet stars

With the MMT we re working on M31: OB star content (NOAO time, data half taken) Yellow supergiants (Arizona time, completed, paper published) Red supergiants (NOAO time resulted in first paper, but more to do eventually) The Magellanic Clouds are providing the low-metallicity example (CTIO 4-m)

Yellow Supergiants in M31: A Critical Test of Evolutionary Models Drout et al 2009 ApJ, 703, 441-460

But not all of these stars are actually in M31! b=-22

Expected foreground contamination according to the Besancon model (Robin et al. 2003)

Blue and red supergiants not much contaminated but yellow supergiants are lost amidst a sea of foreground stars

That s too bad because [The yellow supergiant] phase is a sort of magnifying glass, revealing relentlessly the faults of calculations of earlier phases. --- Kippenhahn & Weigert (1990, Stellar Structure and Evolution)

Two color diagrams won t do it -1-1 -0.5-0.5 0 0 0.5 0.5 1 1 1.5 0 0.5 1 1.5 2 1.5 0 0.5 1 1.5 2

But M31 offers the chance to sort this out Systemic radial velocity of -300 km/sec Rotational radial velocity of 250 km/sec Places in M31 that are coming towards us at 550 km/sec! Use radial velocities to establish membership

How We Did It 1. Start with UBV photometry of M31 stellar population (NOAO Local Group Galaxies Survey, Massey et al. 2005), and select stars in right color and magnitude range to be yellow supergiants log L/L > 4.4 (>12 M ) U-B > -0.4, 0.4 < B-V < 1.4 V < 18.5

How We Did It 2. Select stars in locations such that the expected velocities are < -150 km/sec Left with a sample of 4,282 stars!!! Distribution within M31 suggests foreground contamination dominates the sample.

MMT and Hectospec Pretty obvious by now why we needed the MMT and Hectospec: Large aperture (V down to 18.5) Large FOV (M31 covers 3 ) Large multiplexing advantage (humongous number of stars)

MMT and Hectospec Program was carried out successfully in October- December 2007 using 2.5 nights of Hectospec queue and the 600 line/mm grating 15 faint configurations (3x15 min) [150-200 stars per shot] 5 bright configurations (3x5 min) with one repeated field to test velocities [100 stars per shot] 2901 radial velocities of 2889 unique stars!!!! (67.5% of the complete sample)

MMT and Hectospec No other telescope/instrument combination in the world has the large aperture, FOV, and # of fibers to make this practical. For instance, Hydra on the 3.5-m WIYN has ⅓rd the number of fibers, 3.5x less collecting area, and about half the through-put, or about 21x less powerful overall: 2.5 nights MMT vs 53 nights WIYN, not even counting the 3x longer configuring time.

Radial Velocities (km/s) from 23 Oct 2007 Radial Velocities (km/s) from 17 Oct 2007

So What Did We Find? Out of 2889 stars, there are 54 certain yellow supergiants, and another 66 probable yellow supergiants. Classification based upon difference in radial velocity between what was observed, and what was expected based upon location within M31.

Actually Excess shows the Observed Predicted by Milky Way model M31 members

Certain (blue) and probable (red) supergiants are indeed where you expect to find them spatially, along the CO ring where the OB associations are found!

So What Did We Find? Foreground contamination >96%! Don t expect to find too many yellow supergiants, as their predicted lifetimes are very short (<=80,000 years according to the models). But are their relative numbers in keeping with predictions?

So What Did We Find? Two critical tests we can make of the stellar evolutionary models: Is the relative number of yellow supergiants with different masses (e.g., 12-15M vs 15-25M ) in accord with the predictions of the relative lifetimes? Is the total number of yellow supergiants in M31 in accord with evolutionary theory given the number of unevolved massive stars (i.e., number of OB stars)?

So What Did We Find? The expected number N of stars with masses between m 1 and m 2 will depend upon the lifetimes τ and the IMF (assumed Salpeter Γ=-1.35):

Predicted yellow supergiant lifetimes (Geneva models) Mass Lifetimes 40M 25M 20M 15M 12M 50,800 yrs 18,500 yrs 78,300 yrs 3,200 yrs 5,300 yrs

So just count # stars between tracks

Relative numbers: observed vs models # # Ratio w.r.t. 12-15M Mass range All Certain All Certain Models 12-15M 41 20 1.0 1.0 1.0 15-20M 28 16 0.7 0.8 8.7 20-25M 8 4 0.2 0.2 5.7 25-40M 0 0 0.0 0.0 5.5 15-25M 36 20 1.0 1.0 3.6

Relative numbers: observed vs models Factor of MANY discrepancies: models predict many more high mass yellow supergiants than we see, relative to those of lower masses. Expect 110-150 yellow supergiants between the 25 and 40M tracks. Instead, we find none.

Relative numbers: observed vs models Could we have simply missed them? distribution of magnitudes in observed sample (vs what we wanted to observe) is, if anything, slightly skewed in FAVOR of the brighter stars (unsurprisingly...)!

Actually observed Parent sample

Relative numbers: observed vs models Built-in assumption that star-formation rate, averaged over all of M31, has not varied appreciably over the past 7-17Myr, as the lifetime of a 12M is 17 Myr, while that of a 25Mo is 7 Myr. If the star formation rate had decreased by 3,000-4,000% over that 10 Myr period that could explain it. But Williams (2003) finds that at most there has been a 25% increase (2σ result) in the star formation rate over that time period.

Absolute comparison Are the lifetimes of the lower mass stars (3,000 years) too short or are the lifetimes of the higher mass stars (100,000 years) too long? Can compare the number of yellow supergiants to OB stars.

Absolute comparison Massey (2009) uses the LGGS to estimate there are 24,800 OB stars in M31 with masses > 20M, with typical lifetimes of 5Myr. We observe 8 yellow supergiants with masses >20M. Correct that value by the fact we only observed 68% of our sample and that we only covered 73% of the same area. Thus true number of yellow supergiants with masses >20M is 16.

Absolute comparison 16/24800 x 5Myr = 3,200 years. Lower mass time scales are right, and higher mass time scales are wrong.

Conclusions So what s wrong with the evolutionary models? We don t know (yet)! Maybe the mass-loss rates during the red supergiant phases are wrong. Maybe there s some subtle issues wrong with main-sequence phase. Georges has urged us to look at the low metallicity case (SMC) and see what we find there.

Roger Smith/NOAO/AURA/NSF

M33 But there s still plenty of fun to be had in the north! NOAO/AURA/NSF

Kathryn Neugent