1. X-ray Emission from Primordial Starbursts
Antara Basu-Zych (NASA/GSFC & UMBC)
Ann Hornschemeier
Tassos Fragos
Bret Lehmer
Andy Ptak
Panayiotis Tzanavaris
Mihoko Yukita
Andreas Zezas
First I want to thank the organizers for inviting me and to my collaborators for their help with this work. I want to talk to you today about my work on studying the X-ray emission from Primordial starbursts. As I was putting together my abstract, I was trying to come up with a clever title for my talk, and saw that the organizers magically had this one down in the program. I decided it was perfect. So in the next few slides I hope to introduce what this title means and why it properly conveys the bulk of the research I have done with Chandra so far.

2. X-ray emission in galaxies...
X-ray Emission from Primordial Starbursts
X-ray emission in galaxies...
Active galactic nuclei (AGN):
-- centrally located point source
-- accretion onto supermassive black hole
Chandra image of M82
keV
keV
2.2-6 keV
Hot Gas:
-- diffuse and spatially extended
-- contributes to the soft X-ray band (0.5-2 keV)
Most of this has been covered by now, but for completeness, the X-ray emission from galaxies originates from various sources. Here is a Chandra image of M82 to illustrae the X-ray emission from starburst galaxies. First, in the center of most galaxies, resides a supermassive black hole. X-rays are emitted due to accretion onto this supermassive black hole. However, we won’t worry about this (that’s for the next session, I think) since we are talking about “normal galaxies” where the X-ray emission is not dominated by AGN activity. Another source of X-ray emission is the hot, diffuse gas that typically contributes to the keV soft band flux. But for this talk, we won’t discuss this much. Instead I am mainly studying the emission from X-ray binaries which dominate in the hard band.
X-ray Binaries:
unresolved point sources
dominates in the hard X-ray band (2-8 keV)
- accretion onto compact objects

3. High Mass X-ray Binaries (HMXBs)
Low Mass X-ray Binaries (LMXBs)
astronomynow.com
High Mass X-ray Binaries (HMXBs)
Northwestern
As several of the speakers before me have discussed, there are two varieties of X-ray binaries. Those with low mass donor stars and those with high mass donor stars. The low mass X-ray binaries have relatively evolved stars fueling the compact object. Since these stars are older, LMXBs trace the past star formation or the stellar mass of the galaxy. Meanwhile, in high mass X-ray binaries, the donor is a young, hot, massive star that is short-lived so HMXBs trace recent SF.
• Massive O/B star (>8 M⊙) secondary feeding a compact object (neutron star or BH).
• Massive/short-lived (~10−30 Myr) - trace recent star-formation.
• Lower-mass star (<1.5 M⊙) evolves and swells to red giant feeding the compact object.
• Old stars (>1 Gyr) - trace star- formation history and stellar mass of the galaxy.

4. X-ray Binaries in Nearby Galaxies: Local Scaling Relations
α = (9.05 ± 0.37) × 1028 erg s−1 M⊙−1
β = (1.62 ± 0.22) × 1039 erg s−1 (M⊙ yr−1)−1
High Mass X-ray Binaries (HMXBs)
Northwestern
LX ∝ SFR0.9
Colbert et al. (2004)
Iwasawa et al. (2009)
Lehmer et al. (2010)
LX ∝ SFR0.7
Here is the local relation between X-ray emission and SFR. On the X-axis you have SFR, ranging over 5 orders of magnitude. On the y-axis, we show the 2-10 keV X-ray luminosity. The points on this plot are nearby galaxies that have contributions from both LMXBs and HMXBs. There is a pretty decent correlation, except at the high SFR-end, where is appears a shallower slope fits better. Well, remember that the y-axis, the X-ray luminosity, has contributions from both LMXBs and HMXBs, but only the HMXBs trace SFR. Another way to see this, is by this equation and manipulating it to give Lx/SFR we see a linear regime (inverse-linear actually) and a flat regime (constant) term. For galaxies with high specific SFRs, like starbursts with very recent star formation, this term is negligible and Lx/SFR is a constant. This is the where HMXBs dominate and the LMXB contribution is negligible. Therefore, the rest of this talk is really about X-ray emission from HMXBs.
Scatter ~ 0.5 dex
Scatter ~ 0.3 dex
• To first-order, a physically-motivated scaling of LX will include both SFR and M★ to account for HMXBs and LMXBs:
LX = LX(LMXBs)+LX(HMXBs) = α M★ + βSFR ⇒ LX/SFR = α (SFR/M★)-1 + β

5. z=0 Luminous IR galaxies Normal z=0 star-forming galaxies
SCREAMED THE DUST SPECK
X-ray Emission from Primordial Starbursts
Why do we care?
What does this mean?
Primordial Mode of Star Formation:
dominated by recent star formation: high SFR/M★
less chemically evolved: lower metallicities
and less dust attenuation
… compared to present-day (z=0) galaxies.
Lbol = LIR + LUV
Overzier et al, 2011
1000.0
100.0
10.0
1.0
0.1
LIR/LFUV
Finally, we come to the term ‘”primordial”. What does this mean and why is it important? In our case, we are not studying the first galaxies, we can’t reach back to the “cosmic dawn”. However, we can study galaxies that have a primordial mode of star formation. We expect that the first galaxies built up their stellar populations in similar conditions to these: first, they have active and recent star formation, as we just discussed. Second, these galaxies are less chemically evolved, that is, they are metal-poor. In this figure, known as the mass-metallicity relation, the stellar mass of the galaxy is on the x-axis, and y-axis shows the metallicity, measured by the oxygen abundance. The contours show the distribution for the SDSS local star-forming galaxy population. The mean is roughly given by this line. As you can see, the high-redshift, z=2, galaxies, blue line, have lower metallcities for a given stellar mass. We also expect less dust attenuation compared to the typical local galaxy. For example, this plot shows the bolometric luminosity on the x-axis over 4 orders of magnitude. The bolometric luminosity, ultraviolet + infrared emission, is a good proxy for the total SFR of the galaxy. The y-axis gives the ratio of IR to UV light, estimating the amount of dust attenuation, since dust reprocesses ultraviolet light into infrared. Most star-forming galaxies locally follow this black line, including the luminous IR galaxies. Again, the higher redshift z=2 galaxies follow a similar relation offset towards lower dust attenuations for a given luminosity.
Overzier et al, 2010
z=0 Luminous IR galaxies
z=0
SDSS local star-forming galaxies
(contours)
Normal z=0 star-forming galaxies
z=2 galaxies
(Reddy et al, 2010)
z=2
(Erb et al, 2006)
=SFR

6. z=0 Luminous IR galaxies Normal z=0 star-forming galaxies
X-ray Emission from Primordial Starbursts
Why do we care?
What does this mean?
X-ray binary formation and evolution
(White & Ghosh, 1998; Lehmer+2010; Cowie+2011, BZ+2013; Fragos+2013a, Kaaret2014)
Heating of the Intergalactic medium (IGM)
(Mesinger+2013; Fragos+2013b; Pacucci+2014)
Superwinds
(Strickland+2004,2009; Yukita+2012)
1000.0
100.0
10.0
1.0
0.1
LIR/LFUV
Why do we care? Primordial starbursts may have different X-ray emission properties that would give us insight into the formation and evolution of XRBs. Since the XRBs within the first galaxies were responsible for heating the IGM, we need to understand these differences. Finally, though we aren’t talking must about the hot gas component, it is worth mentioning that superwinds from starburst galaxies may be largely responsible for chemically enriching our Universe.
z=0 Luminous IR galaxies
z=0
SDSS local star-forming galaxies
(contours)
Normal z=0 star-forming galaxies
z=2 galaxies
(Reddy et al, 2010)
z=2
(Erb et al, 2006)
Lbol = LIR + LUV
=SFR
Overzier et al, 2010
Overzier et al, 2011

7. z=0 Luminous IR galaxies Normal z=0 star-forming galaxies
X-ray Emission from Primordial Starbursts
Our Focus:
X-ray binary evolution over cosmic time (driven by metallicity evolution?) using Chandra Deep Field-South data to study X-ray emission in galaxies between z=0—5
X-ray binary populations within individual, nearby (z < 0.1) analogs of high-z (z > 2) Lyman break galaxies (low metallicity, low dust attenuation)
Why do we care?
What does this mean?
Primordial: galaxies from
the early Universe (z > 2)
high SFR per stellar mass
lower dust attenuations,
and lower metallicities
… compared to present-day (z=0) galaxies.
Lyman break analogs, LBAs
Lbol = LIR + LUV
Overzier et al, 2011
1000.0
100.0
10.0
1.0
0.1
LIR/LFUV
So our focus is to study the HMXB evolution over cosmic time, which may be largely driven by metallicity evolution over the history of the Universe. Also, I will talk about a special population of local galaxies, the lyman break analogs,or LBAs, that exhibit this “primordial mode of star formation”. For example, the purple line shows how the LBAs appear similar to the z=2 galaxies in both metallicity and dust attenuation relations.
z=0 Luminous IR galaxies
z=0
SDSS local star-forming galaxies
(contours)
Normal z=0 star-forming galaxies
z=2 galaxies
(Reddy et al, 2010)
z=2
(Erb et al, 2006)
z~0.2
local analogs
z~0.2
local analogs
=SFR
Overzier et al, 2010

8. Chandra Deep Field-South: Deepest X-ray View of the Universe!
Chandra Deep Field-South (Xue et al. 2011)
Hubble Ultradeep Field (Beckwith et al. 2006)
4 Ms Chandra exposure
465 arcmin2
740 sources
So the first part of my work used the Chandra Deep Field South data. Speaking of the wonders of the Chandra telescope! This 4 Ms survey gave the deepest, most sensitive view of the X-ray Universe. Along with coverage from other multiwavelength surveys, we know there are 10s of thousands of galaxies. While most of the individually detected galaxies are AGN, we can use this rich dataset to study galaxies back to a few Gyrs since the Big Bang. Imagine what more we might do once the 7Ms data comes… which is perhaps even this second just finishing up! By the way, by studying the relative contributions from different sources as a function of flux, this paper by Bret Lehmer, finds that the most sensitive regions of this 7Ms data will likely be dominated by galaxies rather than AGN.
Chandra
We love you!
7 Ms CDF-S Observations (additional 3 Ms) are almost completed:
Expect to Be Galaxy-Dominated (vs. AGN) in Most Sensitive Regions.
(see Lehmer et al., 2012)
Within the Chandra Deep Field-South, the multiwavelength data (e.g., Hubble) have revealed there are many 10s of thousands of galaxies reaching back to when universe is <1 Gyr old!

9. Deep-Field Galaxy Selection from Multiwavelength Data:
Lyman break galaxies
Hubble
Chandra
VLA
Herschel
Spitzer
VLT
Significance of LBGs:
Efficient technique for discovering high redshift (z>3) galaxies
Trace cosmic star formation history
Highest z galaxies? (possibly first galaxies)
Excellent sample for studying the average X-ray emission properties of galaxies over cosmic time
Volume-averaged star formation rate for the Universe
To find distant galaxies, a color-selection technique capitalizes on looking for galaxies that dropout of the rest-frame UV filter because of the typical star-forming galaxy spectrum. These are the so-called Lyman break galaxies. The significance of LBGs includes these are the galaxies that best describe our Universe and our best chance at studying distant galaxies, since this technique is so effective in finding high redshift galaxies. In fact, LBGs are a key population in studying galaxy evolution. These galaxies trace the cosmic star formation history, including out to very high redshifts. We use published papers to draw our samples of galaxies at redshifts up to 8! This sample allows us to study the evolution of X-ray properties from X-ray binaries, keeping in mind a few caveats. We elimiate any source that is individually detected at these redshifts, since it is likely an ANG. Since we use stacking to study the X-ray emission, we will get the average property of the sample. z
instrument
NLBGS
referenceTgalaxies.
1.5
HST/WFC3 48
Oesch et al. (2010)
1.9 91
2.5
359
3.0
CTIO+HST/ACS
361
Lehmer et al. (2005)
3.8
HST/ACS
2098
Bouwens et al. (2007)
5.0
445
5.9
181
6.8
HST/ACS+WFC3 73
Bouwens et al. (2010)
8.0 60

10. Age of the Universe (Gyr)
Does the local relation hold at higher redshifts?
Age of the Universe (Gyr)
Redshift (z)
log LX/SFR (erg s-1 [M8 yr -1]-1)
Laird et al (2006)
Laird et al (2005)
Lehmer et al (2008)
Lehmer et al (2010)
Colbert et al (2004)
Iwasawa et al (2009)
Lehmer et al (2010)
High SFR
Stacked LBGS
Colbert et al (2004)
Iwasawa et al (2009)
Lehmer et al (2010)
High SFR
Stacked LBGS
Medium SFR
Colbert et al (2004)
Iwasawa et al (2009)
Lehmer et al (2010)
Remember the local relation I showed earlier to make the point that the X-ray emission from HMXBs correlates with SFR? The question now is whether the same relation holds at higher redshifts. By stacking in redshift bins and SFR bins we see that the lx/SFR relation is elevated in the higher-z samples. But is this significant and does this indicate some sort of evolution? A better way to see that might be to look at lx/sfr as a function of redshift. Our best fit gives us that there is mild evolution with redshift! Tassos before me, introduced his models and we can also add those to our observations and see that there is excellent aggrement with XRB population synthesis models, where metallicity evolution is driving the redshift evolution. So we conclude that the local relation does not hold at high redshifts, there is evolution observed that is consistent with X-ray binary population synthesis model predictions that are largely influenced by metallicity evolution of the Universe.
BZ+13
LBG samples binned by
Redshift and SFR:
z=1.5, 1.9, 2.5, 3.0, 4.0, 5.0, ...
(higher redshifts did not yield detections...)
SFR/[M8yr-1]= &
(<5 and >30 did not yield detections)
How does Lx/SFR evolve over cosmic time?
log LX =A log(1 + z ) + B log SFR + C
A = /- 0.07
B = /- 0.03
C = /- 0.03
(see also Cowie et al. 2011; Kaaret et al. (2014)
Models from Fragos et al. (2012)

11. We use optical emission lines to screen against selecting AGN!
Primordial Starbursts in our backyard
Studying the X-ray emission within individual low-metallicity, high SFR galaxies:
Sample of z~0.1 Lyman break analogs
UV-selected:
high SFRs, metal and dust poor
resemble LBGs:
metallicity, dust attenuations, SFRs, morphology, kinematics…
(Heckman+2005, Hoopes+2007, Basu-Zych+2007,2009a, 2009b, Overzier+2008,2010,2011, Goncalves+2001)
We use optical emission lines to screen against selecting AGN!
Another alternative is to study a sample of analogs that are nearby enough to detect individually so we can get more detailed information about their properties. Since the Lyman break galaxies are selected by their rest-frame ultraviolet properties, these Lyman break analogs are also selected based on their UV properties, thereby selecting high SFR galaxies that are relatively dust and metal poor. Based on various multiwavelength studies, we have found that they resemble the LBGs in every way possible, including the range of SFRs, metallicities, dust attenuation properties, amongst others. However, as they are closer, we can get better information, benefitting from better spatial resolution and increased sensitivity to fainter features. In the X-rays we also gain by being able to detect individual sources. Also important is that we have optical spectral information in order to screen against AGN.
z < 0.1 (LBAs) vs. z > 1.5 (LBGs)
study in better details -- better spatial resolution, fainter features
X-rays: study individual galaxies (vs. average properties)
Mrk 54 J
VV 114
Haro 11

12. Studying X-ray emission within individual z < 0.1 LBAs
High SFR
Stacked LBGS
Medium SFR
Colbert et al (2004)
Iwasawa et al (2009)
Lehmer et al (2010)
Haro 11
VV 114
LBAS
Basu-Zych et al. (2013)
Fragos et al. (2013b)
Once again we want to know where these lie compared to the local relation. And look! Just like all other observed properties, the X-ray emission of LBAs is similar to the LBGs, also offset towards higher X-ray luminosities. Two of these are well-studied galaxies, VV 114 and Haro 11 and we will come back to these later so remember them! Well, what causes the elevated X-ray to SFR in all these UV-selected galaxies? Perhaps it is that they are all metal-poor. So the theory is that HMXBs in low-metallicity enviroments are more numerous and more luminous per SFR since weaker stellar winds means that that the compact object and donor stars can be more massive and therefore have higher X-ray luminosities per SFR. Here is the prediction based on Tassos’ 2013 paper showing that LX/SFR decreases with increasing metallcity The observations seem to agree (albeit with a lot of scatter) with the theory.
What drives the elevated X-ray/SFR in UV-selected galaxies?
Lower Metallicities?
Theory: HMXBs in low metallicity environments are more luminous and UV-selected galaxies have lower metallicities compared to other higher SFR galaxies (LIRGs/ULIRGs).
Observation: Current constraints indicate a negative correlation between LX/SFR and metallicity at the 99.2% confidence level.

13. Studying X-ray emission within individual z < 0.1 LBAs
Prestwich et al. (2013)
Brorby et al. (2014)
NULX/SFR
Z/Z◉ < 10%
SINGS
VV 114
Haro 11
Basu-Zych et al. (2013)
Mapelli et al. (2010)
Fragos et al. (2013b)
Basu-Zych et al. (2013)
In fact, we are not alone in this observation. Mapelli, Prestwich, Brorby and others have noted that the numbers of ULXs per SFR increases with decreasing metallicity. So based on this study we highlight the points that LBAs and high-redshift LBGs have similarly elevated Lx/SFR values and metallicity is likely to be the cause for this. So we dig a little deeper…
12+log(O/H)=7.65
Observational evidence suggests that the X-ray binary populations per unit SFR are more luminous in low-metallicity galaxies (dwarfs and LBAs).

14. Distribution of X-ray binaries within spatially-resolved LBAs
Only the BRIGHT end of the luminosity distribution:
LX > 1040 erg/s (ULXs) J
Target
Exposure time
VV114
60 ks
Haro11
54 ks
J082355
9 ks
Contours:
Chandra X-ray data
2-10 keV
keV
Does this mean that metal-poor galalxies have a different distribution of HMXBs compared to typical sF galaxies? In three of the LBAs, we have the spatial resolution to identify individual XRBs. The contours show the Chandra data for these. Except, the top galaxy doesn’t have sufficient exposure to have detections in the hard band. So we use this case anectodally, and leave it out of our analysis. But in VV114 and Haro 11, the number of 2-10 keV detected sources can be compared to that from other star-forming galaxies. We use the study from Mineo 2012 where they looked at the HMXB X-ray luminosity functions from 29 nearby star-forming gaalxies. As you see we find more than expected.
5’’
VV 114
Haro 11
5’’

15. Distribution of X-ray binaries within spatially-resolved LBAs
Are the observed bright sources REALLY single ULXs?
OR the result of multiple blended sources?
Simulate the effects of source blending…
1. Draw random distributions ...
2. Marx ray tracing code
HST images:
Spatial distribution
Produce simulated keV Chandra images that match depth of actual observations
simulated
observed
VV114
Haro11
Now the question is whether we are truly seeing a different X-ray binary distributioin or whether source blending is artifically producing luminous sources. To test this, we simulate the effect of source blending. We draw random distributions using the HST images as guides for where to place our sources and the LF from Mineo as a guide for the typical luminosity distribution. Then we use the MARX ray tracing code to create simulated Chandra images for these typical sources. Here is the luminosity function result we see. The line is the input luminosity function, the data are these point. The simulated data gives you this line. So there is rough agreement, but it does seem that the points are always higher than the expected curve. So is there a better description of the luminosity function?
Mineo et al. (2012) XLF:
HMXB luminosity distribution +
5’’
VV 114
Haro 11
5’’

16. Distribution of X-ray binaries within spatially-resolved LBAs
Are the observed bright sources REALLY single ULXs?
OR the result of multiple blended sources?
Simulate the effects of source blending…
But include influence from metallicity this time!
Basu-Zych et al. (2013)
Renormalize the input luminosity function by the Lx/SFR enhancement due to low metallicity
Haro11
VV114
We redo this simulation for several models. Let me at least save you all from that tedious exercise and cut to the chase, and show you our best fit model! In that case, we included the effect of metallicity by using the XRB population synthesis prediction shown. Using the same LF as before, but renormalizing by the factor that accounts for metallicity, we find the following. The new input models are higher to account for metallicity, and the simulated output matches the data very well. This is still a work in progress that I am finishing up, but it is encouraging that the story is persisting– metallicity does account for elevated X-ray emission per SFR.
Mineo+ 2012a
Fragos+2013

17. Summary & the exciting future...
Metallicity is an important factor for driving the formation and evolution of HMXBs, based on three different investigations of “primordial starbursts”:
X-ray stacking analyses for z < 4 LBGs (covering ∼90% of the universe’s history) using the 4 Ms Chandra Deep Field South data
Studies of individually-detected LBAs, with similarly low metallicities and elevated LX/SFR as LBGs
… and characterizing the bright end of the X-ray luminosity function within spatially-resolved LBAs
So to conclude– you guess it. Metallicity metallicity metallicity in all the ways that we studied seems to answer the question about why the X-ray emission in primordial starbursts differs from the relation in local star-forming galaxies. In the future, we will be using the soon-to-be 7Ms CDFS data to refine the scaling relations with redshift even more.
Use upcoming Chandra Deep Field-South 7 Ms data
to take this study deeper!

18. Summary
Next steps...
Due to their uniquely low metallicities, low dust attenuations and high SFRs in the local Universe, z~0.1 LBAs represent an important population for studying X-ray emission within galaxies similar to those in the early Universe (in primordial mode of star formation) & offer some advantages over studying high-z samples:
individually detected (vs. studying average properties)
higher spatial resolution
Building up a larger sample of
low metallicity & high SFR galaxies
larger sample of LBAs
with deeper observations on other spatially resolved LBAs
Mrk 54 J
NGC 3310
Also, the LBAs have served as a valuable sample for studying an important population of galaxies– in the primordial mode of SF. They are individually detected, and can be spatially resolved to offer more insight into the XRB population within. Only possible with Chandra! Next we hope to expand this sample further by adding other metal-poor, high SFR galaxies.

19. Summary & the exciting future... Athena
Haro11
Ne IX He α
triplet
FeXX, XXI
Simulated spectrum for
25 ks with Athena XIFU
Due to their uniquely low metallicities, low dust attenuations and high SFRs in the local Universe, z~0.1 LBAs represent an important population for studying X-ray emission within galaxies similar to those in the early Universe & offer some advantages over studying high-z samples:
Individually detected
Less biased towards the brightest galaxies
Higher spatial resolution
A final nod to hot gas… the keV band is best for studying the hot gas component, and this is quickly redshifted out of view. So LBAs really offer the best chance to study the hot gas and superwinds in primordial types of galaxies. Here’s what we might see with 25 ks using Athena for Haro11– looks pretty good. And more detailed simulations can tell us also whether we might be able to get any valuable spectra from resolved regions within this galaxy… so much more to come.
Possibility of studying the hot gas contribution
Based on the average X-ray spectrum for 21 local star-forming galaxies, the hot gas component is well described by kT~ 0.3 keV (Mineo et al. 2012b)
NOT accessible for z > 1 galaxies.

20. Thanks!
I’ll end here… Thank you!

21. log LX = 0.93 log(1 + z) + 0.65 log SFR + 39.80.
Conclusions
& Next steps...
Based on X-ray stacking analyses for z < 4 LBGs (covering ∼90% of the universe’s history) using the 4Ms Chandra Deep Field South data, we find that the 2–10 keV X-ray luminosity evolves weakly with redshift (z) and SFR as
log LX = 0.93 log(1 + z) log SFR
Consistent with predictions from X-ray binary population synthesis models, the redshift evolution of LX/SFR appears to be largely driven by metallicity evolution in high mass X-ray binaries.
Based on X-ray emission studies of individually-detected Lyman break analogs, which have similarly low metallicities and elevated LX/SFR as high-z LBGs, we find that the relatively metal-poor, active mode of star formation in LBAs and distant z > 2 LBGs may yield higher total HMXB luminosity than found in typical galaxies in the local Universe.
Based on X-ray emission studies of individually detected Lyman break analogs, which have similarly low metallicities and elevated LX/SFR as high-z LBGs, the relatively metal-poor, active mode of star formation in LBAs and distant z > 2 LBGs may yield higher total HMXB luminosity than found in typical galaxies in the local Universe.

22. Lyman break analogs (LBAs) : IFUV ≥ 109 L kpc-2
UVLG : LFUV ≥ 2 x 1010 L
Lyman break analogs (LBAs) : IFUV ≥ 109 L kpc-2
LBGs/ LBAs
LBA Overview:
z ~
rare (at z<1 = 10-5/Mpc3) but dominate UV emission at z>3
compact : half light radii = 1-2 kpc)
high SFRs : M⊙/yr
high sSFR: SFR/M⊙~
Hoopes et al. (2007)

23. X-ray emission in galaxies...
X-ray Emission from Primordial Starbursts
X-ray emission in galaxies...
Average X-ray spectrum for 21 local star-forming galaxies (Mineo et al. 2012a,b)
is described well by models of:
bremsstrahlung (hot gas; kT ~ 0.3 keV) plus power-law due to X-ray binaries (Γ ~ 1.8)
Chandra image of M82
keV
keV
2.2-6 keV
2-10 keV
0.5-2 keV
Hot Gas
X-ray Binaries
Total Emission
Mineo et al. (2012a,b)
At higher redshifts, the X-ray binary contribution dominates.

24. Chandra Deep Field-South: Deepest X-ray View of the Universe!
Chandra Deep Field-South (Xuey et al. 2011)
Hubble Ultradeep Field (Beckwith et al. 2006)
4 Ms Chandra exposure
465 arcmin2
740 sources
• At 4 Ms depth, we estimate number counts to ~5 × 10−18 erg cm−2 s−1 (0.5−2 keV) and obtain source densities of ~28,000 deg−2.
Lehmer et al. (2012)
S−2.2
S−1.5
• At the 0.5−2 keV flux limit, AGNs and galaxies provide comparable contributions to number counts:
AGNs − 14,900 deg−2 (560 sources)
galaxies − 12,700 deg−2 (170 sources)
• Relatively sharp slope of normal galaxy counts (dN/dS ∝ S−2.2) indicate that we will soon be in a galaxy dominated regime.
Chandra
We love you!
Within the Chandra Deep Field-South, the multiwavelength data (e.g., Hubble) have revealed there are many 10s of thousands of galaxies, but only ~170 galaxies are individually detected by Chandra.
7 Ms CDF-S Observations (additional 3 Ms) are almost completed:
Expect to Be Galaxy-Dominated in Most Sensitive Regions.

25. Selecting Lyman break galaxiesUn G R
Shapley et al, 2003
This technique is very effective in finding high redshift galaxies. In fact, LBGs are a key population in studying galaxy evolution. These galaxies trace the cosmic star formation history. One might wonder if a rest-frame UV-selection of galaxies is adequate, since dust heavily absorbs ultraviolet light and reprocesses it into infrared. But this plot shows redshift versus the contribution of different populations to the SFR density. The infrared-selected sample, ULIRGs, contribute very little (less than 10 % by z=5) to the total SFR. And new telescopes (finger crossed about JWST), and the infrared instruments on HST/WFC3 now, are striving to use this dropout technique to push to the highest redshifts z =8 and beyond. This is certainly a crucial population of galaxies to study.
Color selection at
z=3:
(Un - G) ≥ 1+(G-R)
(Un -G) ≥ 1.6
(G-R) ≤ 1.2

26. HMXBs detected in spatially-resolved LBAs
Target
Exposure time
Nulx (observed)
Nulx(expected)
VV114
60 ks 5
1.3
Haro11
54 ks 2
0.5
J082355
9 ks
~3-4
0.6

27. Haro 11:Using VLT XShooter data, knot C appears to be associated with luminous blue variable stars Ages for Knots B and C are <10Myr (Guseva+2012)

28. -- Heating of the Intergalactic Medium --
Summary
Implications...
* First galaxies at z=10-20
* XRBs dominate over AGN (Fragos+2013)
* Lx/SFR does not follow the local relation, but evolves with metallicity evolution of the Universe, as predicted by XRB population synthesis models...
-- Heating of the Intergalactic Medium --
Based on three different investigations:
X-ray stacking analyses for z < 4 LBGs (covering ∼90% of the universe’s history) using the 4Ms Chandra Deep Field South data
Studies of individually-detected LBAs, with similarly low metallicities and elevated LX/SFR as LBGs
… and characterizing the bright end of the X-ray luminosity function within spatially-resolved LBAs
we find that metallicity is an important factor driving the formation and evolution of HMXBs within low-metallicity and high SFR galaxies (“primordial starbursts”).

29. is described well by models of:
Average X-ray spectrum for 21 local star-forming galaxies (Mineo et al. 2012a,b)
is described well by models of:
bremsstrahlung (hot gas; kT ~ 0.3 keV) plus power-law due to X-ray binaries (Γ ~ 1.8)
2-10 keV
0.5-2 keV
Hot Gas
X-ray Binaries
Total Emission
At higher redshifts, the X-ray binary contribution dominates, and hot gas component is not easy to study!
Mineo et al. (2012a,b)
At z=0, the hot gas component and XRB component are nearly equal at E=0.5-2keV

30. Next steps with Chandra 7 Ms Deep Field Data
Lehmer et al. in-prep
Based on three different investigations:
X-ray stacking analyses for z < 4 LBGs (covering ∼90% of the universe’s history) using the 4Ms Chandra Deep Field South data
Studies of individually-detected LBAs, with similarly low metallicities and elevated LX/SFR as LBGs
… and characterizing the bright end of the X-ray luminosity function within spatially-resolved LBAs
we find that metallicity is an important factor driving the formation and evolution of HMXBs within low-metallicity and high SFR galaxies (“primordial starbursts”).
SFRs from IR and UV.
SFRs from extinction- corrected UV only.