Studying stars in M31 GCs using

NIRI and GNIRS

Ricardo Schiavon

Gemini Observatory

GSM 2012 – San Francisco

July 19, 2012

Collaborators

• Andy Stephens (Gemini)• Nelson Caldwell (SAO)

• Matthew Shetrone (HET)

• Katia Cunha (NOAO)• Katia Cunha (NOAO)

• Verne Smith (NOAO)

• Carlos Allende Prieto (IAC – Spain)

Goals

• Determination of CNO abundances of old

M31stars for the first time

• Detection of multiple stellar populations in

an M31 GC for the first time

• Pioneering ELT science, pushing AO-

assisted NIR spectroscopy to the limit

Ed

va

rdsso

ne

t a

l. 1

99

3

Burstein et al. (1984)

Chemical CompositionsE

dva

rdsso

n

Clusters in Andromeda

Spectroscopic sample

FOV of Hectospec

Young

Old

Focus on Old GCs

MW vs M31 MDFs

Metallicity Distribution Functions of M31 and the Galaxy are substantially different

Different histories of star formation (at least as far as GCs are concerned)

Caldwell et al. (2011)

MW vs M31 in index-index space

Indices sensitive to age, horizontal-branch morphology and abundances of carbon and magnesium

M31 and MW GCs occupy same locus in occupy same locus in index-index space. Probably same age and chemical compositions

At face value, this is inconsistent with Conclusion 1!

M31

Abundance Ratios vs [Fe/H]

Abundance patterns of M31

GCs suggest GCs suggest enrichment by SN

II

Schiavon et al. (2012, in prep)

Motivation: Abundance ratios

Abundance

analysis based on

high resolution

integrated

spectroscopy spectroscopy

yields promising

results.

But samples are

small. M31 GCs

are the green dots.

Colucci et al. (2009)

M31

MW

Abundances

vs mass

We find evidence for the presence

of multiple populations in populations in M31 clusters

No correlation between Fe, Mg, or Ca, suggests enrichment by ejecta of low mass stars

Schiavon et al. (2012, in prep)

C and N in GalacCluster Stars

Spreads in C and N

abundances in MW GC

stars have been known

for many decades now

They are currently

Ventura & D’Antona 2008

They are currently

understood as being

due to chemical

evolution of the GC

Sample Spectra

Metal-rich clusters

HδCN

CaCH

FeFe

C2Hβ Fe

Mg

Integrated Spectroscopy

M31 cluster B008 (H band)

1”

FWHM ~ 0.5”

Imaging with NIRI

1”

M31 cluster B008 (H band, Gemini/NIRI/ALTAIR-NGS)

GN-2009B-Q-109, GN-2010A-Q-39

FWHM ~ 0.1”

Abundances of M31 GC Stars

• Use AO-fed, medium resolution, NIR spectroscopy of

resolved stars in M31 GCs to determine their C, N, and

O abundances

• O from OH at H band, C from CO at K band, N from

CN at J band

• Comparison between these abundances and those

obtained for massive young stars provide a direct

assessment of the chemical evolution in M31 over the

Hubble time

• Detecting CN or CO band variations in stars of same

evolutionary state would provide evidence for chemical

evolution in M31 GCs

Target Selection

• Intermediate metallicity: -1.0 ≤ [Fe/H] ≤ -0.4

• Massive, not too compact

• Minimize field contamination --- avoid disk and bulge

Principles

• Minimize field contamination --- avoid disk and bulge

• Near a good star for AO correction (NGS)

• Availability of HST imaging a plus

• ~6 clusters, from which we selected:

=> B008: [Fe/H] ~ -0.7, MV=16.52, 4.4 × 105 MSun

Imaging with NIRI

1”

M31 cluster B008 (H band, Gemini/NIRI/ALTAIR-NGS)

GN-2009B-Q-109, GN-2010A-Q-39

FWHM ~ 0.1”

Color-Magnitude Diagram

• Well defined RGB

• Thickness dominated by

photometric error and crowdingphotometric error and crowding

• Allows for safe target selection

• Quality of correction in J band

too poor

• HST imaging available and will

be used.

GNIRS Configuration

• Long Camera, cross dispersed

• Pixel scale 0.05”

• 0.3” slit• 0.3” slit

• 10 l/mm grating

• R ~ 1,700

• Align slit to maximize number of objects

Target Selection

Projected slit width

of 0.3” and 7” slit

length

GN-2011B-Q-74, GN-2012B-Q-88

Include cluster center for an easy RV check for membership

Spectra

• CO bands detected in

spectrum of individual star

• Membership confirmed

• S/N ~ 15/pixel for H~18 in K

bandband

• 2.5 hr integration --- only

1/10 of the total requested

time

• Should be able to meet S/N

requirement with 2012B time

allocation – weather

permittingSpectra of MW GC stars from NIRI GN-2009A-Q-41

Proof of Concept: M71

Spectrum synthesis of CN-strong star in M71

Conclusions

• Preliminary results very promising!

• Should be able to obtain C and O abundances from H

and K band spectra

• Poor quality of J band AO correction makes N

abundances challenging

• Background subtraction and telluric correction need

work

Proof of Concept: M71

NIRI spectra of CN-strong and CN-weak stars in M71

Fundamentals of SP synthesis

Age Effects on Spectra

Balmer lines are

stronger in the spectra

of younger stellar

populations

Fundamentals of SP synthesis

Metallicity Effects on Spectra

Metal lines are

stronger in the spectra

of metal-rich stellar

populations

The ModelsModels

Integrated spectra are interpreted using stellar population synthesis models.

Models predict colors, spectra, and line indices of single stellar populations for various ages, IMFs, and the abundances of Fe, Mg, C, N, and Ca

For an extensive account of the method, see Schiavon (2007, ApJS, 171, 146)

Schiavon (2007) models widely used by the community

Both MW and M31

GCs have [Mg/Fe] > 0

Line Indices: Mg & <Fe>

This tells us that both

GC systems were

formed in less than ~ 1

Gyr

C2 4668 is sensitive

to total metalicity

and to the

abundance of C

Line Indices: C2 & <Fe>

[C/Fe] < 0 for old

MW and M31 GCs

This is probably

indicative of

presence of CNO-

enriched material

Line Indices: CN & <Fe>

CN1 is sensitive

to the abundances

of C and N, as

well as to the total

metallicitymetallicity

It seems like there

is a correlation

between [N/Fe]

and [Fe/H] in both

families of GCs

[Mg / F

e]

[Ca

/ F

e]Sample consists of ~ 180 M31 clusters

Abundance pattern of M31 clusters

[C /

Fe]

[Mg / F

e][N

/ Fe]

with ages older than ~ 4 Gyr and

-1.3 < [Fe/H] < +0.2

[Mg / F

e]

[Ca

/ F

e]

Abundance pattern of M31 clusters

Abundance pattern of M31 GCs different

from Sun [C

/ F

e]

[Mg / F

e][N

/ Fe]

[Fe/H] [Fe/H]

-1 -1 00

[Mg,Ca/Fe] > 0 suggests that the bulk of the GCs were formed in less than ~ 1 Gyr

That’s because SNe that contribute Fe only explode after ~ 1 Gyr, whereas those that contribute Mg explode

very quickly (10s of Myr)

from Sun

Impact of blue HB

stars on abundances A

ge

(Gyr)

[Mg/F

e]

Ages and Abundances

stars on abundances

is minor, and

virtually zero on

abundance ratios

[C/F

e]

[N/F

e]

Young metal-rich

clusters have low

[Mg/Fe] and [N/Fe],

and disk kinematics. A

ge

(Gyr)

[Mg/F

e]

Ages and Abundances

and disk kinematics.

Consistent with

longer star

formation history

[C/F

e]

[N/F

e]

MW vs M31

Mostly similar, but

note small differences

in [C/Fe] and [Ca/Fe]

Zero points are [C

a /

Fe] [M

g / F

e]

Zero points are

uncertain, though

CONCLUSION 2:

M31 and MW

clusters have similar

abundance patterns

(to first order)

[C /

Fe]

[Mg / F

e][N

/ Fe]

Abundances v

s

clu

ste

r m

ass

Abundances

clu

ste

r m

ass

Abundances v

s

clu

ste

r lu

min

osity

Abundances

clu

ste

r lu

min

osity

Abundances v

s

clu

ste

r lu

min

osity

Abundances

clu

ste

r lu

min

osity

CONCLUSION 3:

We find evidence

for the presence of for the presence of

multiple

populations in M31

clusters

C and N in Cluster stars

Models of self-

enrichment by

intermediate-mass AGB

stars fit abundance

ratios in metal rich and

poor GCs (but see

Ventura & D’Antona 2008

poor GCs (but see

Conroy 2011)

Hard to disentangle

original from enriched

composition on the

basis of integrated light

Conclusions

• Different MDFs suggest different SFHs for MW and

M31 GC systems, yet abundance ratios are similar,

suggesting similar SFHs.

• Mg and Ca tell us that M31 GCs were formed in less • Mg and Ca tell us that M31 GCs were formed in less

than 1 Gyr.

• C and N abundances suggest presence of self-

enrichment in M31 GCs.

Fundamentals of SP synthesis

Metallicity Effects

At fixed age, more

metal-rich single stellar

populations are redder

Age and Metallicity have similar effects on the colors of old (> 1 Gyr) stellar

populations, making it almost

Fundamentals of SP synthesis

The Age-Metallicity

Degeneracy

populations, making it almost impossible to disentangle the two effects on the basis of broad-band

colors alone

Spectroscopy is needed to estimate

both mean ages and metal abundances

An example: Carbon

Solar Scaled Models

Variable Abundance Ratios

Models

An example: Carbon

Variable Abundance Ratios

[X/Fe] = 0 [C/Fe] = +0.3

Models

Abundance Pattern

[Fe/H]

[C/Fe]

[Mg/Fe]

SDSS

Mr Mr

[C/Fe]

[Ca/Fe]

[N/Fe]

Mean

Age (Gyr)

Estimating cluster ages

and metallicities from

comparison of line index

Diagnostic Plots

Age

[Fe/H]

measurements with model

predictions for single stellar

populations

EZ_AgesModels

EZ_Ages matches ages and abundances of known Galactic globular clusters to within 0.1 – 0.2 dex (Graves & Schiavon 2008)

Line Indices

Hβ is mostly sensitive

to age and the to age and the

morphology of the

horizontal branch,

whereas <Fe> is mostly

sensitive to [Fe/H]

Line Indices

Hγ is also very

sensitive to age and sensitive to age and

horizontal branch

morphology

Line Indices

Focus on OLD clustersYoung

1 Gyr

Rich

Old

14 Gyr

Poor

[Fe/H]=-1.3

Rich

[Fe/H]=+0.2

Line Indices

No age differences

detected between MW detected between MW

and old M31 clusters

Line Indices

No age differences

detected between MW

Young

detected between MW

and M31 old clusters

Old

Line Indices: Mg & <Fe>

Mg b is sensitive to Mg b is sensitive to

total metallicity and

to the abundance of

Mg

Ages and Metallicities

Results from the

application of

EZ_Ages (Graves

& Schiavon 2008)

to index to index

measurements of

179 M31 clusters

Note population of

intermediate-age

metal-poor clusters

Issues with metal-poor clusters

Ages can’t be

trusted for clusters trusted for clusters

with [FeH] < -1 due

to the presence of

blue horizontal

branch stars

Issues with metal-poor clusters

Theoretical

isochrones employed

do not include blue

HB stars

Issues with metal-poor clusters

Theoretical

isochrones employed

do not include blue

HB stars

Ages and Metallicities

Systematic effect due

to blue HB stars

creates a population

of artificially younger

metal-poor clustersmetal-poor clusters

Differential effects

Blue HB stars

contribute more

light at lower λ,

thus affecting Hδ thus affecting Hδ

and Hγ more

strongly than Hβ

Ages and Metallicities

Due to HB stars, ages

inferred from Hδ are

younger than those based

on Hβ for clusters with on Hβ for clusters with

[Fe/H] < -1

Ages and Metallicities

Due to HB stars, ages

inferred from Hδ are

younger than those based

on Hβ for clusters with on Hβ for clusters with

[Fe/H] < -1

Ages and Metallicities

Old GCs have an age

of 11.8 ± 2 Gyr

But beware of zero-

point uncertainties

Ages can’t be trusted

for [FeH] < -1 due to

Issues with metal-poor clusters

for [FeH] < -1 due to

the presence of blue

horizontal branch stars

Ages can’t be trusted

for [FeH] < -1 due to

Issues with metal-poor clusters

for [FeH] < -1 due to

the presence of blue

horizontal branch stars

Ages and Abundances

Results from the

application of

EZ_Ages (Graves

& Schiavon 2008) & Schiavon 2008)

to the index

measurements

[Mg / F

e]

[Ca

/ F

e]

The Carbon-Nitrogen Conundrum

What do we make

of the [N/Fe]

correlation?[C

/ F

e]

[Mg / F

e][N

/ Fe]

correlation?

It is indicative of

secondary

enrichment

C and N in MW Cluster stars

Stars in Galactic GCs

have a range of C and

N abundances.

Also main-sequence

Carretta et al. 2005

Also main-sequence

stars, so not due to

internal mixing

Poorly understood,

but self-enrichment is

most likely scenario

[Mg / F

e]

[Ca

/ F

e]

Assumption: more massive clusters

should self-enrich

Self-enrichment in M31 clusters?

[C /

Fe]

[Mg / F

e][N

/ Fe]

should self-enrich more efficiently.

[Mg / F

e]

[Ca

/ F

e]

Ca, Mg, and C: same behavior as

whole sample

Low mass clusters4 < log (M/Mo) < 5

[C /

Fe]

[Mg / F

e][N

/ Fe]

whole sample (very scattered)

NO CORRELATION

in Nitrogen!

[Mg / F

e]

[Ca

/ F

e]

Ca, Mg, and C: same behavior as

whole sample

Intermediate-mass clusters5 < log (M/Mo) < 6

[C /

Fe]

[Mg / F

e][N

/ Fe]

whole sample

NOISY CORRELATION

in Nitrogen!

[Mg / F

e]

[Ca

/ F

e]Ca, Mg, and C:

same behavior as

6 < log (M/Mo)

Massive clusters

[C /

Fe]

[Mg / F

e][N

/ Fe]

same behavior as whole sample

(more scattered)

TIGHT CORRELATION

in Nitrogen!

Nitrogen in MW FieldF

igure

sto

len f

rom

a talk

by N

ikola

s P

rantz

os

Fig

ure

sto

len f

rom

a talk

by N

ikola

s P

rantz

os

Nitrogen in MW FieldF

igure

sto

len f

rom

a talk

by N

ikola

s P

rantz

os

[N/Fe]

[Fe/H]0-2

0

1

-1

-4

Fig

ure

sto

len f

rom

a talk

by N

ikola

s P

rantz

os

[Fe/H]

In the field, no correlation is found between [N/Fe] and [Fe/H], indicating a primary origin

• MMT/Hectospec• 1° FOV• 300 fibers, 1.5 arcsec diameter

Observations

MMT• 300 fibers, 1.5 arcsec diameter• 270 l/mm grating, resolution ~5Å over > 5500 Å, blue

Hectospec focal

surface

Motivation: MDFsConstraining the history of star

formation and chemical enrichment in

the nearest MW giant neighbor

Abundances and ages from

integrated spectra of globular clusters

Interesting results from early efforts

M31

-3 -2 -1 0

Brodie & Huchra (1991)

Interesting results from early efforts

(Brodie & Huchra 1990,1991), Puzia

et al. (2005), many others suggesting

that:

- M31 and MW clusters have bimodal

metallicity distributions

MW

-3 -2 -1 0

-3 -2 -1 0

[Fe/H]

• Spectroscopy of ~1,200 objects with MMT/Hectospec, plus many “background” (M31 field) spectra

• 25 pointings

• Resolution ~5Å, 3650-9200Å

• Median S/N ~ 75/Å at 5200Å

Observations

• Median S/N ~ 75/Å at 5200Å

• 450 clusters, plus stars, possible stars, background galaxies

• New catalog and details in Caldwell et al. 2009

• Study of young clusters in Caldwell et al. 2009, metallicitiesand ages of old clusters in Caldwell et al. 2011, kinematics of metal-rich clusters in Morrison et al. 2011., line strength comparisons with MW GCs by Schiavon et al. 2012

Imaging with NIRIM31 cluster B008 (H band, Gemini/NIRI/ALTAIR)

1”

CNO abundances of M31 cluster stars with ALTAIR/GNIRS (on-going multi-

semester program at Gemini-N (Schiavon, Stephens, Caldwell, Shetrone,

Allende-Prieto, Cunha, VV Smith)

MW vs M31 in index-index space

Indices sensitive to age, horizontal-branch morphology and abundances of carbon and nitrogen

MW vs M31 in index-index space

Indices sensitive to age, horizontal-branch morphology and abundances of calcium and carbon

MW vs M31 in index-index space

Indices sensitive to age, horizontal-branch morphology and abundances of carbon and magnesium