Richard Ellis (Caltech)

Searching for Cosmic Dawn: The First Stars & Galaxies in the Universe

W M Keck Observatory, April 20th 2006 Hubble Ultradeep Field

Time or redshift

today

Big Bang

`First light’

Witness the past! looking deep means looking back in time…

An Observational Adventure Starring..- the two Keck 10-meter reflectors & their spectrographs

NIRSPEC

… and three unique space telescopes

Hubble Spitzer

WMAPHubble: exquisite deep imaging

Spitzer: sensitive to older stars

WMAP: studies of microwave background radiation and its scattering by foreground material

Tracing the History of StarlightTracing the History of StarlightOur quest is to find and understand the earliest cosmic systems containing the first stars which formed barely 100-500 million years after the Big Bang - when the Universe was only 3% of its present age. Some of these stars long since died but many are still shining!

Galaxies represent the giant systems where these stars now reside.

To trace the history of starlight we must trace the history of galaxies

Unraveling Cosmic HistoryUnraveling Cosmic History

2-3 byr

Today (13.7 byr)5 byr

Keck and Palomar, aided by remarkable Hubble Space Telescope images have enabled us to explore the history of the rich variety of present-day galaxies. We have pieced together the story of galaxy formation and evolution back to 2 billion years after the Big Bang (85% of cosmic history)

How to find the most distant galaxies seen at early times?

• At large redshift, the signal from a remote galaxy declines due to hydrogen absorption at a particular frequency which enters the range of optical telescopes

• Search for tell-tale ‘drop’ in signal in ultraviolet signal:

• Palomar does the searching

• Keck verifies the distance via a spectrum

C. Steidel and co-workers at Caltech have found hundreds of distant galaxies successfully via this remarkable technique

Finding Distant Galaxies using `DropoutsFinding Distant Galaxies using `Dropouts’’

Spectroscopic Confirmation at Keck

NB: The technique selects only star-forming systems whose strong ultraviolet signal is redshifted into the optical region

Deep Palomar image Keck spectrum

Redshift z~3

Hubble images of these spectroscopically-confirmed galaxies with redshifts z ~ 3 reveal small physical scale-lengths and irregular morphologies: many appear to be merging or assembling from smaller units -immature systems

What do these early galaxies look like?What do these early galaxies look like?

Infrared cameras have revealed an important additional population of sources present at early times; these systems appear to have completed their star formation by redshift 2

Securing a complete inventory of distant galaxies through deep infrared imaging

Distant `quiescentDistant `quiescent’’ galaxiesgalaxies

2

Quiescent (non-starforming) galaxies selected by prominent infrared and weak optical signals: verifying their distances to these infrared-selected galaxies has been remarkably difficult yet it is suspected they lie alongside their star-forming cousins!

Optical Infrared

Z=2.43

Z=2.43

Z=2.43

Z=2.71

Z=3.52

Keck Spectroscopy of Distant Red GalaxiesKeck Spectroscopy of Distant Red Galaxies

P van Dokkum et al

Cole et al 2dF

Star formation history Mass assembly history

Cosmic history since Cosmic history since redshiftredshift 5 5 (1.2 billion years after the Big Bang)(1.2 billion years after the Big Bang)

Surveys by Keck, Hubble and other telescopes reveal a coherent picture of star formation and mass assembly over the past 12 billion years. Growth is gradual driven by mergers. Above a certain mass, galaxies become quiescent and stop forming stars.

redshift →

redshift →

Time or redshift

today

Big Bang

Earliest signal revealed by microwave experiments: cosmic microwave background

Progress in Microwave Background Studies

Microwave background corresponds to separation of matter & radiation at redshift z =1088 ±1 when age = 372,000 years

What happened next? What happened next?

Microwave background radiation is seen 372,000 yrs after creation representing the time when hydrogen atoms form for the first time

Universe then enters a period called the `dark ages’: cold hydrogen clouds clump and eventually collapse to form stars

Stars eventually energize hydrogen in deep space breaking it into electrons and protons (process called `reionization’)

DARK AGES

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

time

End of the Dark Ages: End of the Dark Ages: ReionizationReionization of Hydrogen by First of Hydrogen by First StarStar--forming Galaxiesforming Galaxies

Polarization in WMAP Data Polarization in WMAP Data

Polarization in microwave background probes electron scattering in the foreground i.e. electrons from the time of reionization

WMAP signal suggests reionization occurred at 6<z<15 corresponding to 300 -900 million years after Big Bang

Distant Quasars Provide Another ProbeDistant Quasars Provide Another Probe

Redshift

Observer

Quasars are intense beacons lighting up intervening clouds of hydrogen gas.

As we approach the end of reionization we would expect an abrupt change in the amount of hydrogen absorption

Is it seen?

X. Fan (U. Arizona)

Keck Spectra of the Most Distant QuasarsKeck Spectra of the Most Distant Quasars

Hydrogen absorption

Depth of Hydrogen Absorption in Quasar SpectraDepth of Hydrogen Absorption in Quasar Spectra

Beyond redshift 5.5, there is a tantalizing upturn in the amount of hydrogen absorption - we may be approaching the end of the dark ages!

Finding the Earliest GalaxiesFinding the Earliest Galaxies

Summary:

Indirect evidence from high redshift quasars and polarization of the microwave background suggests there was a sharp transition in deep space sometime between z=6 and 15, corresponding to 300-900 million years after the Big Bang

Most likely this was the blaze of light from the first luminous systems:

Can we detect them?

End of the dark ages

The Hubble Ultra Deep FieldThe Hubble Ultra Deep Field

GOODS field – 13 orbits HUDF – 400 orbits

The HUDF remains the deepest ever optical image

Very distant sources in deep Hubble imagingVery distant sources in deep Hubble imagingHubble data

Dropout

Wavelength

We can extend the `dropout’ technique used successfully at z=3 to find examples of star-forming galaxies beyond z=5 by looking for dropouts in redder bands

Keck DEIMOS spectroscopy of iKeck DEIMOS spectroscopy of i--band dropoutsband dropouts

z=5.83

z=5.78

Census of starCensus of star--forming galaxies at z=6forming galaxies at z=6

Combining wide field GOODS survey with narrower Ultra Deep Fieldreveals the luminosity distribution of z~6 galaxies

Counts are consistent with those at z=3 with a decline in abundance of ×6

We are approaching the beginning of the star-forming era!

÷ 6

UDFGOODS

Sky surface density

←Luminosity

Abundance of Star Formation Sources 3 < z < 10Abundance of Star Formation Sources 3 < z < 10

A sharp decline (×10) is seen in the number of star forming sources from z=3 to 10

Despite some uncertainty, the number of luminous sources seems to be insufficient for reionization

Required for reionization

redshift →

Census of Stars in Place at Redshift 5

D. Stark et al

z=5.554, 1.1 1011 M z=4.831, 1.6 1011 M

In a remarkable development, the 60cm infrared Spitzer Space Telescope has detected a large number of z~5-6 `dropouts’enabling us to do a census of the assembled mass in stars at this early time; this provides an `accounting record’ of all the past star formation, i.e. before z=5

Assume a Star Formation History…

luminous

all

1 + redshift →

Can it account for the stars seen at redshift 5?

D. Stark et al

Mass assembled at z=5 is greater than can be accounted for by earlier generations of luminous star-forming galaxies

luminous

all

1 + redshift →

Mass

in old

stars

Must be lots of earlier activity we can’t currently detect!

Lyman Lyman αα SurveysSurveysA more effective way to find early star-forming galaxies utilizes the fact they will contain hot gas emitting the Lyman alpha line of hydrogen redshifted from the far ultraviolet

As much as 6-7% of a young galaxy light may emerge in this single spectrum line!

Wide Field Imaging from SubaruWide Field Imaging from Subaru

Megacam

SuprimeCam

Mauna Kea `Ohana’:

Panoramic imaging with Subaru with Keck spectroscopic verification to ensure narrow line is high redshift Lyman alpha

E. Hu (U Hawaii)

How it works: LyHow it works: Lyαα emitters at z=6.5emitters at z=6.5

spectra

Emitter detected via contrast between narrow & broad filter

Narrow filters in atmospheric `windows’

Night sky spectrum

z(Lα) = 4.7 5.7 6.6

Night sky spectrum

The The RedshiftRedshift 7 Barrier7 Barrier

Finding and studying sources becomes harder beyond z=7:

• Sources become fainter against strong sky background

• Lyman alpha shifts to infrared, beyond reach of optical cameras

• Infrared technology has lagged behind that for the optical

Infrared region

Boosting the signal with gravitational lensesBoosting the signal with gravitational lenses

Gravitational Gravitational LensingLensing

EddingtonEddington (1919) (1919) confirms the sun confirms the sun deflects starlight deflects starlight via measurements via measurements at a total eclipseat a total eclipse

Fritz Zwicky(Caltech)

First to deduce the presence of copious amounts of dark matter.

Suggested use of gravitational lensingin clusters of galaxies would extend the power of a telescope ..a point overlooked by Einstein and which lay dormant for a further 50 years!

QuickTime™ and aSorenson Video decompressorare needed to see this picture.

Courtesy: Lars Christensen (ESA)

Simulated view through transparent dark lens

Combining Keck and a `GravitationalCombining Keck and a `Gravitational’’ TelescopeTelescope

Critical lines (points of high magnification) for background sources at

z=1

and for

z=5

Blind search and more detailed follow-up

Utilizing strong magnification (×10-30) of clusters, probe much fainter than other methods but in small areas

• Magnification = ×30 → 20 × fainter than unlensed searches

• Very low mass (10 million solar masses) and age < 1-2 million yrs

Very faint young galaxy at z=5.7

Extension to Extension to higher higher redshiftredshift

z=5.6

z=6.8

Spitzer detection of multiplySpitzer detection of multiply--imaged imaged redshiftredshift z~6.8 sourcez~6.8 source

Infrared eye of Spitzer sensitive to older stars (unlike Hubble) so can, for the first time, measure age and mass of this early system

Deciphering past history of z~6.8 systemDeciphering past history of z~6.8 system

Hubble Spitzer Hubble Spitzer

Spitzer → this is already a well-established system 800 Myrs after Big Bang

Star formation rate = 2.6 solar masses/yr; stellar mass ~ 0.5% Milky Way

Age at this epoch: 100 – 450 million yrs, so formed at 9 < zF < 12

If representative, this is an object in decline after `first light’!

Young stars

Old stars

Pushing Further Back - I: Keck, Hubble & Spitzer

Imaging of 8 clusters (in progress)Drop outs with z~8-12 being foundAim: confirm with Keck (v hard!)

MS1358 - H

critical linemag →∞

mag > × 6

HJz

candidate with z~11

Pushing Further Back - II: Keck Lyman α mapping

NIRSPEC (infrared)Slits arranged to lie on the critical lines of very high magnification

We have extended the successful optical survey for magnified Lyman alpha emitters into the near-infrared where we are sensitive to sources with redshift z~8-10

Using Keck’s NIRSPEC, the goal is to provide the first constraints on whether there are sufficient feeble sources to contribute to reionization and hence to end the `dark ages’

Significance 3.7σz = 9.0 f = (2.2±0.6)x10-17

cgsSFR(unlensed)=0.1 Myr-1

42x0.76 arcsec NIRSPEC slit

Abell 2219 NIRSPEC 2D Spectrum

Abell 2219 HST R-band imageNo coincidentoptical detection; await future deep NIR broadband image from HST NICMOS to search for coincident NIR continuum emission.

Candidate LyCandidate Lyαα Emitters: I z=9.0Emitters: I z=9.0

f =1.2±0.2 x 10-17 cgs mag ~×50; L= 2.5x1041 cgs

Extensive LRIS/NIRSPEC followup fails to reveal associated lines (e.g. if source is at low z or a companion of z=1.58 object):

→ a promising possibility with 0.25 M yr-1.

Abell 68: 1.255μm, z=9.3? No broad-band detection but close companion at z=1.58

Candidate LyCandidate Lyαα Emitters Emitters -- II: z=9.3II: z=9.3

Summary: What We Have LearnedSummary: What We Have Learned

• The accumulated stellar mass at z~5 (1.2 billion years after the Big Bang) points to a lot of early star formation which probably enough to have ended the dark ages.

• Currently we cannot see enough star formation at earlier times (to z~10, 500 million years) to be consistent with this - maybe it is mostly in feeble objects (Keck/Hubble/Spitzer data will hopefully tell us soon)

• If not, maybe `first light’ occurred even earlier than redshift 10

• We are making progress but it is getting more and more challenging

Where Next?Where Next?

Improved performance from our existing telescopes will extend present work

• adaptive optics on Keck: - detailed studies of selected z~5 systems

• Keck MoSFIRE: - multi-object infrared spectra for z > 7 sources

• new infrared camera WF3 on Hubble ( if the shuttle flies again): panoramic imaging for z > 7 sources

2014+: James Webb Space Telescope and a 30m ground-based telescope (TMT)

• a new partnership, similar to the successful one between Hubble and Keck

• more detailed surveys beyond z~10 and fainter sources z~7-10

Adaptive Optics Studies of z~5 arcsExample: internal rotation of z~1 arc

z > 4 arcs with OSIRIS: lensing + AO probe 150pc scales at z~5 !

Keck laser guide star

Thirty Meter Telescope Project

http://www.tmt.org

A partnership of Caltech, U California, US National Observatory & Canada

• In detailed design & costing phase:

• Builds on successful Keck technologies

• Operational ~2015

James Webb Space Telescope

6.5m infrared space telescope

Under construction & due for launch ~2013

"At the last dim horizon, we search among ghostly errors of observations for landmarks that are scarcely more substantial. The search will continue. The urge is older than history. It is not satisfied and it will not be oppressed.”

Edwin Hubble (Realm of the Nebulae)

Conclusions: Is it Worth it?Conclusions: Is it Worth it?

timeSlow merging of dark matter under gravity is proposed to explain the diversity of galaxy types

Hubble Deep FieldsHubble Deep Fields

NICP5

NICP12

ACSUDF

NICP34

400 orbits

200 orbits

200 orbits

100 orbits

100 orbits

Keck I + HIRES 20 hrs

log N(CIV) > 11.7 cm-2

MetallicityMetallicity of the Intergalactic Spaceof the Intergalactic Space

Keck II + ESI 12 hrs

C IV forest

CIV forest is still present at z=4.5Songaila & Cowie (2002, 2003)

Distribution of 515 LyDistribution of 515 Lyαα emitters at z=5.7emitters at z=5.7

Ouchi et al

Challenges of nb imaging in Infrared

• The 1.0 to 1.8 micron IR sky is very dark between the OH lines which contain 95% of broad band background

• The narrowest gaps are narrower than in the optical; filter widths of 0.1 per cent are needed c.f. 1% filters in optical

• Thus wide field imaging even more important!

optical infrared

Example: Example: AbellAbell 23902390Cluster critical line for zS > 7

NIRSPEC slit positions

Wavelength sensitivity (1.5hr)

10-17 cgs

10 clusters completed October 2005 (Stark et al, in prep):

• 1 convincing (6σ) Lyα emission candidate• 6 moderately-convincing candidates (3-4.5σ)• In process of attempting to confirm these

How it works: three broad regimes

What the observer sees, viewing through a lens, depends on the focusing power of the lens, the relative distances of lens and background source and the degree of alignment of both

• rings & arcs

•arclets

• weak distortion

LENS

Arcs & Multiple Images

Background sources are magnified and distorted depending on alignment with foreground `lens’ -also get multiple images!

Source Alignment Image

Courtesy: J-P Kneib