The First Instalment – Click here to find out more about my first two weeks.
The Second Instalment – Click here to find out about my third week.
The Third Instalment – Click here to find out about my fourth and fifth weeks.
The Final Instalment – Click here to see my results.
Alternatively, read all the blog posts in one place here!
In this internship, I will be studying the SC4K sample of Lyman alpha emitters (LAEs) (Sobral et al 2018). The SC4K catalogue contains 3,908 LAEs across the COSMOS field (Capak et al 2007). The catalogue spans a redshift from z ~2 to z~6 across sixteen different filters corresponding to sixteen redshift slices. See the video we made at Lancaster about SC4K and how we discover and study super-distant galaxies!
Lyman-alpha is intrinsically the brightest emission line and can be used to trace the circumgalactic medium (CGM) surrounding a LAE. This line, seen in the rest-frame UV spectrum of a galaxy, is often used to study high redshift galaxies. This is because beyond redshift ~2 the Lyman-alpha line is redshifted into the optical spectrum and becomes easier to observe from the ground. Lyman alpha photons originate from high star formation regions where the young, highly ionising O and B type stars produce photons that ionise the neutral hydrogen gas in the star-forming regions, in so-called HII regions. As Hydrogen atoms recombine they release a variety of photons, with Lyman-alpha being the most abundant and energetic. Furthermore, as these photons are produced, they will likely encounter clouds of fully neutral Hydrogen which can trace the sizes of entire galaxies and producing what is known as Lyman alpha haloes (LAHs). Another contribution to the production of Lyman alpha photons is from active galactic nuclei (AGN) and recombination in the accretion disk or even in giant HII regions around powerful quasars. This also means that, in general, LAEs can be classed as star forming (if Lyman-alpha is being produced by O and B stars) or AGN. LAEs can be very efficiently discovered or identified by an excess emission in a given narrow band (NB) image that covers rest-frame 1216A, when compared with the broadband or continuum image at a similar wavelength. From a spectrum of a galaxy you can determine the redshift of a source by identifying the Lyman break, as it shifts into optical for higher redshifts and the Lyman break technique can then be used to identify the line in the spectrum and also confirm the redshift. Another way to determine the redshift, such that this can then be used to identify if a Lyman alpha line is present in a sources spectrum, is to use colour-colour ratios to determine the photometric redshift.
Lyman alpha photons are scattered by the neutral hydrogen gas surrounding the galaxy, due to the lines’ resonant nature. This can result in the formation of Lyman alpha halos which can be seen when subtracting the UV continuum flux from a NB image of the galaxy to obtain a Lyman alpha image. However, in principle, cooling radiation (inflows), and multiple minor mergers of faint LAEs could also lead to similar haloes. For powerful quasars, the Lyman-alpha haloes are typically from recombination at large scales, and not scattering, due to the high level of ionisation around quasars.
Studying these Lyman alpha haloes can aid the current understanding of the evolution of galaxies and unveil the powering sources and the physical origin of haloes, since the epoch of re-ionisation. This epoch is the time in the universe, following the formation of the first stars and galaxies, in which neutral atoms could be ionised. It is believed to have occurred between 20 ≥ z ≥ 6 (Barkana and Loeb 2001) and measuring the evolution of LAHs can also attempt to further constrain the time in which this epoch took place. Since Lyman alpha photons trace the amount of neutral hydrogen, there should be an increase in the size of the halos at redshifts after those ranging this epoch, when there was supposedly more neutral hydrogen available. This should in principle be accompanied by a significant reduction in the surface brightness profiles as a whole, as Lyman-alpha escapes a lower surface brightness if it encounters a significantly neutral CGM and IGM.
We can measure and quantify the size of our LAHs in the SC4K catalogue using a scale-length, rn, which is also used in other studies making comparison possible (Momose et al 2014, Leclercq et al 2017). We will measure the scale length by plotting a surface brightness profile of our images, where the functional form tells you about the structure of the galaxy.
The images I will be using are not simply of each source in the SC4K catalogue, but instead various stacks of a collection of images of each source taken with the Subaru telescope. Stacking the individual images of LAEs gives us an image that shows the general properties of these galaxies and is useful as it cancels out the noise and amplifies the signal. The background of the images, where there are no sources, will have a varying pixel count due to noise variations. This should have a Gaussian projection when looking across pixel values. Thus stacking all the images together should cancel out the noise, due to oppositely varying signals in corresponding pixels between the individual images. The images were stacked using a median combined stacking method which provides a representative view of the sample and is relatively unaffected by outliers when compared to the mean combined stacking method.
We can create stacks per property of a galaxy, so for each filter (and hence redshift slice) we can stack all sources that belong to this filter to get an image of all sources at that redshift. We can also do stacks of different properties within each redshift slice, such as per filter we will stack only the AGN sources or only the bright sources etc. The different groupings for each filter are; all sources, AGN only, no AGN, bright, faint, high equivalent width (EW) and low EW. Bright = Lyman alpha luminosity > 43.0 ergs-1, faint = Lyman alpha luminosity < 43.0 ergs-1, high EW = EW>80Å, low EW = EW<80Å.
Equivalent width is how a measure of how strong a line is, it is the amount of continuum you need to integrate to get the same area as the emission line, where the continuum is the light from the star. Need to measure in the rest frame (EW0) as the observed equivalent width (EWobs) gets larger with redshift. A LAE with high EW is usually a more metal poor galaxy. EW0 = EWobs/(1+Z).
We also stack the same number of non-Lyman alpha emitter galaxies, with the same properties and magnitude distribution, that are in the COSMOS field, in order to measure the low-level systematics of the data and infer the noise. We can repeat these stacks with different random galaxies that satisfy the conditions to produce a cube of different realisations of this stack and we can measure the noise variations from stack to stack.
The stacks of non-emitters, which have been handled the same way as the emitter stack, show negative values in the central regions due to over subtraction, followed by a ring-shaped bright region and then a dark halo. Since there is no extended halo here we know the Lyman alpha halo seen in the emitter stacks is real and not due to systematics. The bright region arises due to the seeing mismatch between the broadband and narrow band images. We can subtract this stack of non-emitters from our LAE stacks in order to correct for systematics.
Before making measurements on the stacked image, a process called smoothing is carried out to reduce effects of noise in the images by using a Gaussian filtering method.
Since when the LAE stack is corrected for systematics, if we still see a halo of Lyman alpha around the galaxy, we know this halo is real and we can measure the extent using surface brightness (SB) profiles as mentioned previously. We perform annuli photometry on the stacks which is similar to aperture photometry but the key difference is that as we go to larger radii we will not be measuring the central bright regions of the galaxy which tell us about the stars but not about the halo. The SB is a measure of flux density of a source per area of annuli with units of ergs-1cm-2arcsec-2. We can fit our exponential fit to the regions of the SB profile that tell us about the haloes to get a measure of the scale length. The regions that tell us about the halo are those beyond the point spread function (PSF) of the source and the central light of the galaxy. The PSF is a measure of how much a point source is already extended as a consequence of the atmospheric seeing when observing. Any extent seen is only real if it is past the extent of the PSF. The PSF can be quantified by measuring the extent of stars in the images, as these should appear point like under perfect conditions.
Using the measurements of the shapes and sizes of the LAHs will help gain a better understanding of the physics that forms these structures. Are they caused by scattering of Lyman alpha photons by neutral hydrogen or are there other important contributions from radiation? We can investigate if the scale lengths are evolving with redshift and if there are clear differences between the haloes of galaxies that host AGN and star forming only galaxies?
Stay tuned to find out more!
Emma Dodd
The First Instalment – Click here to find out more about my first two weeks.
The Second Instalment – Click here to find out about my third week!
The Third Instalment – Click here to find out about my fourth and fifth weeks.
The Final Instalment – Click here to see my results.
Alternatively, read all the blog posts in one place here!
XGAL: Time travelling through the Universe 