A unique instrument for the biggest telescope



Cosmic web

Matter in the Universe is organized into a cosmic web, an intricate structure that consists of large void regions, sheet-like walls, and connected filaments. This majestic complex contains most of the matter in the Universe—be it the mysterious dark matter or baryons—and its properties, shaped by the processes in a very young Universe, could help to answer fundamental cosmological questions. The web can be portrayed as a medium of very low density that exists between galaxies—the intergalactic medium (IGM). Both the formation and evolution of galaxies are impacted by the web and a careful study of the interplay between galaxies and the IGM is necessary to understand how galaxies evolve.

The IGM at low redshifts (z < 1) has been mapped by galaxy redshift surveys; observing hundreds of thousands of galaxies has revealed a clear 3D structure in space. But at higher redshifts, the bulk of the galaxies gets progressively fainter and spectroscopic observations much more time-consuming. An alternative method has to be found.

IGM tomography – mapping the z > 2 Universe

How can we map the 3D structure at higher redshifts? A very efficient technique at z > 2 is to use the light from bright, distant sources—like star-forming galaxies or quasars—as a probe of the IGM in the sight line between us and the sources.

Credit : C. Stark & K. G. Lee

As the light travels toward Earth, it gets absorbed by atoms of (mostly) neutral hydrogen that reside in the IGM clouds. Each cloud produces a series of absorption lines at high energies: Lyman-alpha, Lyman-beta, Lyman-gamma, and so on. The wavelengths of the lines in the spectrum are governed by atomic physics and the expansion of the Universe: the farther the cloud is from Earth, the longer the wavelengths of the absorption lines in the spectrum of the background source. Many clouds lie along each sight line, resulting in a characteristic forest of lines. The most prominent part of this spectral feature is known as the Lyman-alpha forest (Lyman-beta, Lyman-gamma, etc. forest also exist, but are not relevant for this study). The principle is visualized in this video.

One background source provides information on the position and density of IGM clouds along one sight line. The next step is to observe many sources in a small area in the sky. Once the IGM clouds are mapped along all these sight lines, advanced interpolation techniques can be used to obtain a detailed map of gas density between the sight lines, in the transverse direction, as well. The technique, called IGM tomography, provides a 3D map of the IGM in the front of the background sources.

Practical considerations

A spectroscopic survey of galaxies at z > 3 is challenging. The requirement to observe thousands of relatively faint galaxies demands large telescopes and multi-object spectrographs. Only recently has the first attempt been made to carry out such a survey. It is not yet clear, what is the quality of the spectra of the background sources needed to carry out a successful IGM tomography. Furthermore, the technique itself is only an intermediate step: the maps will be used to tackle a science case. How good must the spectra be to successfully apply them to study voids? Or the interplay between the cosmic web and galaxy evolution?

We approached the problem head-on by simulating a real survey with the help of the Horizon-AGN cosmological simulation. From the simulation, we extracted a spectral catalog of galaxies at 3 < z < 4 in an area of 1 degree^2 and used the spectra to perform the tomography of the IGM in the 3 < z < 3.5 range. The workflow is schematically shown in the image.

We aimed to understand how the quality of the spectra, characterized by signal-to-noise (S/N) and resolution, affects the reconstruction of the density field, that is, the IGM tomography. Furthermore, we checked how the reconstructed field can be applied to the studies of galaxy evolution. The nominal resolution of the instrument in the blue will be R = 5000. However, the mapping of the IGM can be carried out at a lower resolution. The acquired spectra can be binned, resulting in a lower spectral resolution but higher S/N, which would save the precious observational time on the ELT. We investigated two different spectral resolutions after the binning has been applied: R = 1000 and 2000. A series of tests give consistent results. Briefly, the S/N (per resolution element) of ~ 6 (4) at resolution R = 1000 (2000) is necessary for a good reconstruction and to detect gradients of galaxy stellar mass and star-formation rate towards the filaments of the cosmic web. We find that a much higher S/N would not improve the results enough to justify a much higher demand on the telescope time.


Caption: The original (bottom) and reconstructed fields (top and middle) of the IGM taken from the Horizon-AGN simulation. Source

IGM tomography survey with MOSAIC

We simulated the performance of the MOSAIC spectrograph at blue wavelengths, based on the instrument design at the end of Phase A study. In particular, we obtained the relations between S/N and various other characteristics like resolution R and exposure time. The details of the performance of the instrument are provided in the paper and summarized here. With this information available we could link the requirements of the IGM tomography (S/N and R) to the minimal necessary exposure time per galaxy. The estimated numbers of nights on the ELT, required to carry out the survey, are provided in the table.

The estimates are given for the two observation modes (VIFU and HMM-VIS) and the two studied configurations (R = 1000 and 2000). The numbers in parentheses indicate the estimates if the observations are conducted under poor seeing conditions. The third line gives an independent estimate of the survey speed ratio of the two modes (see also "Simulating surveys for ELT-MOSAIC: status of the MOSAIC science case after phase A" ).



The mapping of the IGM is one of the primary scientific goals of several current and future surveys. Each survey uses a slightly different approach, though the underlying principle is always the same as outlined in the preceding text. Most of the surveys are focused on reconstructing the IGM in the 2 < z < 3 redshift range. MOSAIC, on the other hand, will enable the reconstruction at z > 3 down to Mpc scales. It will, for example, offer the possibility to probe the full redshift evolution of galaxy growth in the cosmic web throughout a cosmic period of intense star formation in galaxies. The survey with MOSAIC should also result in a robust reconstruction of voids and walls, allowing to put constraints on cosmological theories (note that the role of MOSAIC for this science case has not been studied in detail yet).

A spectroscopic survey of thousands of faint galaxies requires information on the galaxy positions and their preliminary redshifts. Not to mention their physical properties, like stellar mass and star-formation rate, that are needed to study galaxy evolution coupled to the large-scale environment. The MOSAIC project will therefore have to be coordinated with other surveys. In particular, the combined data set from space missions Euclid and JWST will be invaluable.