One of the goals of CANDELS is to study galaxies during Cosmic Dawn, when galaxies were just beginning to form. Other teams are also pursuing this kind of research, using the CANDELS data as well as deeper observations from the Hubble space telescope.
There has been some interesting progress in the past few weeks.
The Most Distant Galaxy in the Hubble Ultradeep Field
Redshift z>9 candidate galaxy from the
Hubble Ultra-deep field. From Rychard
On November 13, Rychard Bouwens posted a preprint that solidified the evidence that a previously-identified galaxy is really very distant. This galaxy is located in the Hubble Ultra-Deep Field, which is located within the CANDELS survey region. But this galaxy is too faint too see with the exposure times that we use for the rest of the CANDELS survey. It took four days of exposure time with the WFC3 infrared camera on Hubble to detect the galaxy. For comparison, the deepest CANDELS exposures are only about six hours. The best guess is that this galaxy is at a redshift of z>9. At redshift 9, the universe was 550 million years old. The light from that galaxy has taken at least 13.1 billion years to reach us. This particular candidate had been found previously by the same team using about three days of exposure time, so it's nice to see it confirmed with deeper data.
This galaxy, along with the other ones mentioned below, was found using the Lyman Break technique, which fellow blogger Russell Ryan describes in his blog post. Interestingly, the galaxy is invisible through all but one filter. The very fact that it is visible in the reddest filter, and invisible in all the bluer filters, is the strongest evidence that it is a very high-redshift galaxy. The interpretation is bolstered a bit by the fact that it is not detected in very deep observations with the Spitzer infrared observatory. It seems pretty likely that this object is not an old or dusty red galaxy at lower redshift. The source is just a little bit fuzzy in the Hubble images, so it doesn't appear to be a very faint red star in the Milky Way either.
So, it seems like a pretty good candidate. The new observations have made the detection through the one filter and the non-detection through the others more solid. But the evidence that it is truly distant is still quite tenuous, and based largely on ruling out other possibilities.
Finding a Distant Galaxy with a Little Help from a Gravitational Lens
On November 15, Dan Coe and the CLASH team posted a preprint identifying a very solid Lyman-break candidate with a redshift z~10.7. This galaxy turned up in infrared observations that used only four hours of exposure time compared to four days. Hubble got some help for this galaxy from the gravitational lensing effect of a giant cluster of galaxies located between us and the distant object. This boosted the light from the distant galaxy by about a factor of fifteen. It also resulted in three separate images of the galaxy. Each of the images is detected through the two reddest filters, but not detected at other wavelengths. So this is a really solid detection. There are no other objects in the entire CLASH survey (so far) that have similar colors.
Three separate images of the gravitationally-lensed galaxy at redshift z~10.7, from the CLASH preprint. The gravitational lens split the light into these three separate images, all of which have the color expected for a very distant galaxy. The galaxy is invisible at wavelength shorter than ~1.4 microns due to absorption by hydrogen clouds between us and the galaxy. This is known as the Lyman Break, and is the classic signature of a distant star-forming galaxy. The Lyman Break is particularly strong in this one.
The positions of the three images of the CLASH galaxy are marked as JD1, JD2 and JD3 on this image from their preprint. Many of the galaxies that you can see near the center of the image lie in the foreground cluster. This massive cluster acts as a gravitational lens, to magnify the images of the distant galaxy (and split it into several separate images). You can see this kind of lensing effect if you look at a candle through the bottom of a wine glass. As you move the glass around, you will see the image of the candle stretch and sometimes break into several separate images. The curves drawn in the image show the "critical lines" of maximum magnification for background sources at different redshifts. The fact that the JD images lie where they do (particularly JD1 and JD2) adds a lot of support to the interpretation that this is the most distant galaxy yet discovered.
Not only is it a solid detection -- the evidence that it is a high-redshift galaxy is about as solid as it can get without actually measuring the spectrum. The three images of the galaxy lie just about where they are expected to lie based on the models of the gravitational-lens that lies in the foreground. There is some uncertainty in those models, but those models will get better as the CLASH data analysis proceeds, so even without any further Hubble observations we might learn some more about this particular candidate.
Meanwhile, Back In the Ultra-Deep Field
Images of some of the new candidates identified in the Hubble
Ultra-deep Field, from the UDF12-team preprint. The galaxy at
the bottom is the same one that is shown at the top of this blog
post, which illustrates how different image processing can produce
a very different appearance. Nevertheless, both teams agree
that the galaxy is well detected at 1.6 microns and undetected
through all the filters at shorter wavelengths.
Just today the UDF12 team, led by Richard Ellis, have posted a preprint on their website about new candidate galaxies in the Hubble Ultra-Deep Field. These are the same observations used by Rychard Bouwens mentioned above, but analyzed by a different group. The paper concurs on the interpretation of the Bouwens object, estimating its redshift to be z=11.9. The paper also identifies a half-dozen other candidates at z>8.5. These other candidates are all detected in more than one filter (barely), but are so faint that it would not be at all surprising if a few of them turn out to be either spurious or at a different redshift. For statistical purposes, that's probably fine, because it's possible to estimate with reasonable confidence how many will turn out to be wrong.
The latest CANDELS Sample: A Bit Closer and a bit Brighter
Images of some of the brighter, closer, candidates
found in CANDELS, from Haojing Yan's preprint.
Their colors suggest that these galaxies are at
In this latter vein, Haojing Yan's CANDELS-team study of brighter candidates at z~8 (submitted last December) has now been accepted for publication. The CANDELS observations survey a wider area, but with shorter exposure times. So we can find the rarer, brighter sources, but not the fainter ones. There are seven decent candidates so far, all at z ~ 8 - so a bit closer than the galaxies discussed above. The evidence that these are distant galaxies is comparable to that presented in the Ellis and Bouwens papers, but not as strong as that in the Coe paper.
What Will it Take to Improve the Evidence that these are Really Distant Galaxies?
The quest to find and confirm the most distant galaxy will continue. The gold-standard for confirming that these are really distant galaxies will be to obtain a spectrum and measure the redshift precisely. The James Webb Space Telescope, slated for launch in 2018, can obtain a low-resolution spectrum for any galaxy that Hubble can see, with just about the same exposure time as the Hubble observations. We might get lucky and confirm some of these distant-galaxy candidates before then, but large-scale studies of galaxies at Cosmic Dawn will really take off when Webb flies.