Observations of galaxies at higher redshift

• Observing at higher redshift allows us to look back in time.
  • Relation of look-back time to redshift (e.g., LCDM; at redshift z=1, lookback time is between 4.5 and 14.5 Gyr dependending on h and $\Omega$ (note Ned Wright's cosmology calculator).
  • Only bright objects can be seen at higher redshift. For LCDM
    • z=0.5, m-M=42.3
    • z=1.0, m-M=44.1
    • z=2.0, m-M=46.0
    • z=3.0, m-M=47.1

• look to see if implied star formation history from local galaxies is observed statistically in galaxies at earlier epoch.

• Also look for evolution of galaxy density: are there some types of galaxies which were present in past but not now or vice versa?

• Since it's difficult to study individual galaxies at a variety of redshifts, and in particular to know which galaxies at one redshift to associate with which galaxies at another redshift, people often study statistical properties of population; one main tool has been the galaxy luminosity function.
  • can in principle study more than just distribution of luminosities, e.g., one can study LFs of different morphological types and relative numbers of different morphological types, or using spectral type/color, etc.
  • since we've seen that galaxies are not just a one-parameter family (e.g. in luminosity), but a multidimensional one (e.g. in luminosity, size/SB, and/or velocity), one would like to study galaxy evolution in multiple parameters. Note, however, that studying size-related quantities requires high spatial resolution because typical images sizes of galaxies are a few arcsec at z=0.5-1, with scale lengths less than an arcsec (e.g. HST), and studying velocity related quantities requires large telescopes (and possibly also high spatial resolution).

• Recall also the potential importance of selection effects. Even just a simple flux limit on a survey can create the appearance of strong evolution; at large distances, only intrinsically most luminous objects will be seen, and it is even possible that these objects will be preferentially missed in nearby universe if they are intrinsically rare compared with less luminous objects.

Evolution of the luminosity function

• Complications:
  • selection effects (e.g. surface brightness, see summary LF figure from Cross et al. 2001),
  • large scale structure / cosmic variance
  • K-corrections (large especially if rest frame is shortward of $\sim 3600$,
  • cosmological model (from angular size-redshift relation for volume size and distance-redshift relation for distance modulus),
  • method of LF determination.

• Observations often characterized with respect to some simple models:
  • no evolution (local LF with k-correction only),
  • passive evolution (local LF with k-correction, evolution back in time of currently observed stars, no star formation),
  • luminosity evolution (SF occurs),
  • density evolution (objects appear/disappear).

• Historical: redshifts not available, so measurements were made of galaxy counts as a function of magnitude/color.
  • Euclidean expectation is $N\propto m^{0.6}$, less steep given expanding universe.
  • Large numbers of faint galaxies are in fact observed (Koo and Kron Fig 1), and these are found to be blue (K and K Fig 2) - the so-called faint blue galaxies.
  • Question: what are these galaxies? Are they local, low-luminosity galaxies or more luminous distant galaxies? If the latter, to what current type (if any) are they related, and what do they tell us about evolution in the corresponding type of galaxies?

• Significant recent progress with completion of several redshift surveys for field galaxies to intermediate redshift:
  • CFRS (I-selected with $17.5<I<22.5$, $\sim$ 600 galaxies, ApJ 455, 108 and refs therein),
  • Autofib (B-selected, combination of several different studies over wide range of $11.5<b<24$, $\sim$ 1700 galaxies, MNRAS 280, 235),
  • CNOC1, CNOC2 (r band selection, $\sim$ 400 galaxies, ApJ ),
  • Hawaii K-band ( K-selected $K<20$ with additional galaxies included based on I and B brightness, $\sim$ 400 objects, AJ 112, 839),
  • DEEP and DEEP2, $\sim$ 40,000 galaxies with $R_{AB}<24.1$
  • COMBO-17, $\sim$ 30,000 galaxies with $R<24$, 17-band photometric redshifts
  • VVDS
  • Larger local surveys:
    • 2dF
    • SDSS
  • Summary table (somewhat out-of-date!) (Tresse lecture)

• Redshift survey results
  • CFRS 1995. Luminosity function evolves, with different evolution when split by color. Little evolution in red galaxies, more in blue galaxies.

  • Similar split in behavior seen if separation made by spectral type (Autofib)

  • Similar conclusions reached with CNOC2
    • Galaxies with different color/type (corresponding to morphology, but only roughly) have different LFs (CNOC2 Schecter parameters), also show different degrees of evolution (CNOC2 LF evolution)..
    • Early and intermediate types show mostly luminosity evolution, late types mostly density evolution (CNOC2 evolution of parameters).
    • Results seem to be fairly robust (CNOC2 sensitivity to errors).
    • Note the evolution refers to LF changes only, and not necessarily to physical changes in galaxy population (though this may also be the case).
    • CNOC2 shows no evidence for faint-end slope changes, but other studies (e.g. AUTOFIB) do. Results are comparable between CNOC and CFRS and AutoFib.

  • K-band survey from Cowie et al suggests significant evolution in star forming, lower luminosity, galaxies, but mostly passive evolution in more massive galaxies, e.g. result for blue galaxies ((Cowie etal Fig 24)). Original suggestion of ``downsizing''.

  • However, more recent results with significantly larger number of galaxies and deeper (COMBO-17 and DEEP2) suggest the picture may be different.
    • These studies use bimodality, rather than (K-corrected) color splits to divide galaxy. They also extend to higher redshift and have larger samples
    • Luminosity function shows evolution, including red galaxies
    • All samples show luminosity evolution, but red galaxies now show number density evolution. Schecter parameters
    • Conclusion supported by lack of evolution of luminosity density of red galaxies (potentially measured more robustly); since older galaxies should have higher M/L, implies an increase in number density

• Interpretation
  • red galaxies:
    • Density evolution for red galaxies is somewhat debated, with some studies suggesting little or no density evolution, others some. Clearly, this is important for understanding the degree to which early type galaxies are formed hierarchically (and if so, when).

    • red galaxies cannot have recent star formation. If numbers have evolved, they must have come from mergers of a visible population, e.g. migration from the blue sequence to the red sequence

    • Migration could occur in multiple ways
      • Early merging to red sequence, then subsequent merging along red sequence
      • Late merging to red sequence
      • Intermediate case

    • Entirely early merging unlikely: not enough faint red galaxies at higher redshift. Also, can't get stellar population variations along red sequence as previously discussed.

    • Entirely late merging unlikely: can't get variation of structure along red sequence as previously discussed

    • Multiple paths to red sequence suggested? Downsizing of typical ``quenching'' timescale?

    • Still some question about Mg-$\sigma$ relation. Issues with IMF? (which would open all sorts of cans of worms!)

  • blue galaxies
    • Blue galaxies may substantial evolution, either in luminosity or number (or a combination). K-band results suggest evolution in luminosity, since the intrinsically brightest galaxies appear only at higher redshift and there's a smooth trend of brightness with redshift. However, it's possible that there is some additonal population of bright blue galaxies at higher redshift which has essentially disappeared by the present.

    • Theories for origin of blue galaxies:
      • luminosity evolution: number of galaxies same as present with evolving (declining) SF rate
      • bursting dwarfs: some galaxies have intense bursts of SF, then fade rapidly and become invisible; mechanism suggested by Babul and Rees in which ioniziation by background light (e.g., from quasars) keeps small clouds ionized until $z\sim 1$ then rapid burst in SF makes bright objects, but gas is ejected and SF ceases
      • mergers
      • some combination

    • Emission line measurements strongly suggest that luminosity evolution is ocurring, since higher redshift galaxies have much greater fraction of [OII] emitters (CFRS XIV (Fig 3).
      • also evidence, however, that metallicity of higher z galaxies is lower, and less dust could partly explain strong evolution of emission line properties.

Morphology of distant galaxies

When samples are taken and imaged at high spatial resolution, one can look to see what the galaxies actually look like; in particular, one would like to assess the question as to whether the faint blue galaxies are progenitors of current galaxies or a special population. Several observations:

• normal Hubble types are recognizable ( Schade et al ApJL, 451, 1 (1996) Fig 1 (F814W) and Fig 2 (F450W)).

• with good spatial resoution, one can measure the surface brightness of disk galaxies, and one finds that the SB is higher for disk galaxies at $z>0.5$ (Simard etal fig 5); this suggests that part of the LF evolution is due to luminosity evolution of disk galaxies, in which they were brighter in the past; size evolution is also a possibility. Interpreting this in terms of luminosity evolution implies higher SF rates in the past. However, note that some or all of this may be a selection effect, since higher SB galaxies preferentially make it into a sample which is defined from low-res observations with a magnitude cutoff (Simard et al fig 6); cannot separate hypothesis of systematic brightening of SB with increasing redshift from increased spread in SB at larger redshift.

• there is a population which does not appear to have a modern counterpart: the blue nucleated galaxies (Schade Fig 4). These comprise 15-30% of the high redshift sample, and are preferentially located in objects with peculiar morphology. In general, the frequency of peculiar morphology is higher at higher redshift. This is also supported by the HDF observations, to be discussed later. Note some degree of this might come from observing at shorter wavelengths, but NICMOS observations suggest that this is not the dominant portion of the story.

• There may also be a population of high SB galaxies that does not exist at low redshift (Simard et al Fig 12); these may be a new population, early-type spirals exhibiting passive evolution, or just represent galaxies missing from the local sample, or very possibly, a combination of all of these.

Evolution of the luminosity density

• Can integrate over LF to get total luminosity density of the universe, in hopes of picking out overall trend of SF averaged over all galaxies; also important to understand the mean metallicity trend with time in the IGM, etc.

• It appears that the luminosity density has decreased since z=1 (e.g. CFRS XIII (Fig 1), where a significant portion of the increase comes from blue galaxies (open circles). Interpreting this in terms of population synthesis models

  implies that a significant increase in SF rate appears to be required as one goes back to z=1, roughly proportional to $\tau^{-2.5}$, or $\tau^{-3}$ (CFRS Fig 2).

• If one computes SF rates directly from integrated [OII] density using relation from local galaxies, one finds much higher SF in the past, as well; but so much higher that it overproduces the number of stars seen today (CFRS XIV Fig 6)! Possible resolution: less dust at higher redshift and/or lower metallicity, giving different relation between [OII] and SFR, different IMF at higher z.

• To a large degree the evolution in the luminosity density comes from the morphologically peculiar galaxies Ellis 2001 (Fig 12).

Evolution of the stellar mass density

• Can also consider estimates of stellar mass, e.g. by looking at near-IR light which is less sensitive to details of the star formation history (Ellis Fig 13).)
• results suggests that stellar mass in large systems is roughly constant out to z=1, but strong evolving in interacting/peculiar systems (Ellis Fig 15), suggesting number evolution as important, because, while stellar mass density can grow through star formation, it cannot decline.

Evolution in clusters

Interesting to compare evolution of cluster galaxies, because these may be subject to an additional evolutionary effect, namely one induced by the cluster environment. From ground, it was noted that some high-z clusters ($z\sim 0.4$) have abnormally large fraction of blue galaxies, the Butcher-Oemler effect. HST helps in that one can get morphologies. One learns:

• most blue galaxies are ``normal'' spirals, though there is a subset of abnormal-looking galaxies which comprises a larger fraction of cluster galaxies than at present,

• ellipticals are nearly as abundant in high z clusters as a present, and increase of spirals appears to come at the expense of S0 galaxies (Dressler & Smail Fig 2,

• ellipticals show quite uniform colors, suggesting an early epoch of formation, and SB evolution consistent with passive evolution; fundamental plane correlations also support this point of view. In many respects, evolution in clusters is similar to that in the field, except spirals may be becoming transformed into S0 systems. Note that this requires a physical distinction between S0 and E galaxies - not necessarily as indicated by local studies (as discussed previously).

• Increasing blue fraction may just result from hierarchical clustering, in that groups and clusters form relatively late, so as one goes back in redshift, they've had less time to inhibit star formation (e.g. via processes discussed for local clusters).

• While the blue fraction increases with redshift, it is still smaller than the field blue fraction, at least to a $z\sim 1$.
  • at $z>1$, DEEP2 (Gerke et al MNRAS 376, 1425 (2007) suggests, that group blue fraction converges to field blue fraction, suggesting quenching doesn't happen earlier than $z\sim 2$