Probing the epoch of reionization

1
Probing the epoch of reionization

• Jeremy Davey

2
The Big Bang
The Dark Ages
Epoch of Reionization
Today
3
Timing of the Epoch of Reionization (EoR)

• Still do not know the exact timing of EoR
• WMAP measured an electron scattering optical
  depth of 0.17 0.04 implying reionization at z
  17 5 (Kogut et al. 2003, Spergel et al. 2003)
• Spectra of quasars in Sloan Digital Sky Survey
  (SDSS) show presence of Gunn Peterson troughs ?
  end of EoR at z 6.2
• When did EoR actually occur?
  ? we need observations of sources
  between 6.5 lt z lt 17

4
The Epoch of Reionization

• Bubbles of ionized material around sources
• Expand and overlap
• Post-overlap stage reached all points in IGM
  exposed to a number of sources

from Barkana Loeb (2001)
5
Sources of reionization
Quasars
- not enough at z 6 (Fan et al. 2001)
- possible if stellar initial masses are
different from today (Ricotti et al 2004)
Galaxies
Pop III stars
- first objects formed, metal free stars, highly
skewed IMF with M gt 100 M?
- Intermediate Mass Black Holes (IMBHs) form from
Pop III stars and accrete material ? shine as
mini-quasars
Mini-quasars
6
Searches for reionization sources

• Steidel Hamilton (1992) designed a technique
  to detect galaxies by identifying a break in
  their continuum the Lyman break (blueward of
  91.2 nm)
• Comparison of transmission through bands indicate
  the presence of Lyman break
• Technique originally used at z 3 but has been
  applied to high redhsifts
• Lehnert Bremer (2003) detected 6 galaxies at z
  5
• Stanway et al. (2004) detected 11 candidates at z
  6

7
Emission line searches

• Narrow band imaging and spectroscopy
• Narrow band filters
• comparison of flux transmitted through narrow
  band with broad band can detect presence of Ly?
  emission lines
• Hu et al. (2004) detect 18 Ly? emitters at z
  5.7
• Willis Courbin (2005) find no candidates at z
  9
• Ly? emission line
• redshifted to optical and infrared at redshifts
  we are interested in ? high sensitivity and
  resolution observations possible
• Advantages of emission line searches
• avoid sky lines (in near infrared)
• confirms the redshift
• deep observations (low luminosity)

8
Lack of detection

• Despite 35 hours of observation by Willis
  Courbin (2005) - no detections at z 9
• Also, Bouwens et al (2005) use Lyman break
  technique but find no confirmed sources at z 10
• Neither go deep enough

need to use gravitational lensing to go deeper
9
Advantages of gravitational lensing
Surface brightness remains unchanged
Lower luminosity sources can be detected
But source appears M times larger on sky
10
Gravitational lensing

• Primary advantage probe deeper with similar
  observation times
• depth increased by M by increasing area of a
  source
• Ellis et al. (2001) observed galaxies magnified
  by cluster A2218 - confirmed
• Pello et al. (2004) claimed a detection of a z
  10 galaxy
• Weatherley, Warren Babbedge (2004) do not
  confirm after detailed reanalysis
• Bremer et al. (2004) also find no galaxy with
  deeper Gemini images

11
Motivation

• Despite spurious nature of Pello z 10 galaxy,
  it is high quality data
• a rigorous reduction may yield candidates and
  useful flux limits
• Uses gravitational lensing - a great method for
  probing the EoR
• Can effectively avoid the OH emission from the
  sky
• Possibly observe galaxies at high redshift
• estimate the Initial Mass Function (IMF)
• determine if they are source of cosmic
  reionization

12
Data

• Pello et al. (2004) spectra along critical
  lensing line for z 9

• Cover wavelength range 1.162 ?m lt ? lt 1.365 ?m
• corresponding to Ly? redshift range z 8.5
  10.5
• Taken with ISAAC on VLT on July 2003
• Raw frames retrieved from ESO archive
• Each frame consists of 1024x1024 pixels

13
Improvements over the Pello analysis

• Reduction is complicated and rigorous
• Pello z 10 galaxy not real
• possibly caused by bad pixel/s
• We improve limits by
• carefully removing bad pixels using masking and
  also by rejection during the final combination
  (Pello did not)
• using a 2d distortion correction

14
Typical frame

• Before and after reduction

15
Detection

• Match filters with 2d Gaussians
• 4 different filters to account for extension in
  spatial and dispersion axes
• Run up each column on array ? noise estimate as
  function of wavelength
• Leads to a signal-to-noise map
• Few possible candidates above 5 ? (greater than
  the noise at particular wavelength) under
  analysis
• Average 5 ? detection limit of 7 x 10-18 erg s-1
  cm-2

16
Detection limits
17
Discussion

• Surveyed a volume of (30.9 / M) h-3 Mpc-3
• Using Kennicutt (1998) relation and that Ly?/H?
  10 implies that any detections above 5? will have
  star formation rates of
• Detections of such a galaxy would tell us if
  galaxies were sources of reionization

SFR (6 / M) M? yr-1
18
(?M, ??, h) (0.3, 0.7, 0.7)
(2005)
density evolution
magnification
luminosity evolution
Santos et al. (2004) for z 5 low luminosity
galaxies
Comparison to other surveys success of lensing
survey
19
Future Work

• Finish analysis of Pello spectroscopy
• Retrieve additional lensed observations from ESO
  archive and analyse
• Examine Willis Courbin (2005) data for possible
  detections in narrow band only
• Submit observing proposal to VLT to study
  environments of known z 6 quasars
• Search UKIDSS data for z 6 quasars and follow
  up spectroscopy using Gemini (if any candidates
  found)

20
The end
Special thanks to My supervisor Dr Steve
Warren And Dr Steve Weatherley
21
Additional slides
22
Reduction

• Carried out along guidelines from ISAAC manual
• Bias removal
• Dark subtraction
• Flat fielding
• Bad pixel correction
• Wavelength calibration
• Flux calibration

23
Reduction

• Mainly along ISAAC guidelines
• Pickup noise removal
• Fourier transform mask method
• Sky subtraction
• Bright emission lines from OH in atmosphere
• Multiple telescope offsets allow for subtraction
  of sky lines
• Variable pixel scale correction
• Star trace data used to determine correction
  necessary
• Final combination and flux calibration

24
Star Trace Correction
25
Uncertain Initial Mass Function

• It is speculated that at high redshift the IMF
  might be considerably different from today
• Ricotti et al. (2004) looked at Pello et al.
  (2004) detection to show that galaxies at z 10
  could be reionization sources
• fit to decreasing SFR density using a top-heavy
  IMF
• So
  need to search
  for galaxies between 6.5 lt z lt 17