Ripples in Cosmic Web Measured Using Rare Double Quasars
Astronomers believe that matter in intergalactic space is distributed in a vast network of interconnected filamentary structures known as the cosmic web. Nearly all the atoms in the Universe reside in this web, vestigial material left over from the Big Bang. A team led by researchers from the Max Planck Institute for Astronomy have made the first measurements of small-scale fluctuations in the cosmic web just 2 billion years after the Big Bang. These measurements were enabled by a novel technique using pairs of quasars to probe the cosmic web along adjacent, closely separated lines of sight. They promise to help astronomers reconstruct an early chapter of cosmic history known as the epoch of reionization. The results are being published in the April 28 edition of the journal Science.
In-depth description: Ripples in Cosmic Web Measured Using Rare Double Quasars
On the largest scales, hundreds of millions of light-years and beyond, the dominant structure in the universe is the cosmic web: a network of intersecting filaments that consist mostly of dark matter, some of them spanning billions of light years. Dotted along these filaments, each within its own little halo of dark matter, are galaxies, like our own Milky Way, from dwarf galaxies to the largest galaxies more than 100 000 light-years across. Typically, the most massive galaxy clusters are found at the nodes of the web, where cosmic filaments intersect.
Galaxies however account for only a tiny fraction of the web's volume. Most of the volume is made up of intergalactic space. In these vast expanses between the galaxies there are neither stars nor planets. The only matter present is in the form of solitary atoms: a diffuse haze of hydrogen gas left over from the Big Bang, on average one per cubic meter in total, and a few per cubic meter within the filaments of the cosmic web. If the galaxies are the decorative dots along the filaments, this hydrogen is a diffuse haze, tracing the dark matter distribution throughout the cosmic web.
An evolving network of filaments
The evolution of this cosmic web is one of the central issues of cosmic history, and is described by complex cosmological simulations. Current models assume that the this large-scale structure grew out of tiny fluctuations in the density of material just after the big bang, 13.8 billion years ago. Our earliest picture of the Universe comes from the cosmic background radiation: the thermal radiation that was set free after the universe became transparent, when the initial plasma had condensed to form (mostly hydrogen) atoms. The inhomogeneities within the cosmic background radiation are the seeds of large-scale cosmic structure, corresponding to regions of slightly more or less than average density.
Simulations like the Millennium simulation and its kin can reproduce how these initial inhomogeneities evolved into the large-scale filamentary structure we see today, 13.8 billion years later: Regions of higher density having slightly stronger gravitational attraction, and thus increasing their density ever further; asymmetries leading to the contraction of inhomogeneous regions into slender filaments, forming the cosmic web. This scenario forms the backbone of the current models of cosmic evolution; the search for an understanding of how galaxies formed and evolved in this framework is a central question of modern cosmological research.
All the more important are tests of the currently accepted models of the evolution of the cosmic web. But when it comes to observing the cosmic web, astronomers face a fundamental difficulty. Dark matter is, by definition, dark. But even the intergalactic gas is so rarified, its density so small, that it emits no measurable amounts of light.
Tracing the web: quasars as intergalactic beacons
Fortunately, there is a specific observational technique that can detect the elusive intergalactic medium. The observations in question target not light emitted by the hydrogen gas, but light emitted by a source that is farther away from us than the hydrogen, and whose light is then partially absorbed by the gas. The light sources in question are known as quasars: a brief hyper-luminous phase of the galactic life-cycle, powered by the infall of matter onto a galaxy's central supermassive black hole. Quasars are very bright indeed – a single quasar can be as bright as all the hundreds of billions of stars in a galaxy put together. This means that quasars are visible over extremely large distances. The most distant known quasars are so far away that their light has travelled nearly 13 billion years to reach us.
Compared with galaxies, the light-emitting regions of quasars are very small; as far as we are concerned, they might as well be point-like. By analyzing quasar spectra, it is possible to deduce both the amount and the distance of intergalactic hydrogen gas the quasar light has encountered on its way. Light is absorbed by hydrogen atoms at specific wavelengths. In the spectrum, the rainbow-like decomposition of light, this absorption is visible in the shape of sharply defined, dark lines, known as absorption lines. The more hydrogen the light encounters, the deeper and more pronounced the absorption lines. As a consequence of cosmic expansion – galaxies in the universe all moving away from each other on large scales –, light from distant objects such as our quasar is shifted towards longer wavelength in what is known as the cosmological redshift.
Depending on how far the quasar light has already travelled before it encounters intergalactic hydrogen gas, it has already been redshifted by a certain amount. The redshift is related directly to how far light has already travelled before encountering the gas. Depending on the redshift, absorption will happen in different regions of the spectrum of the quasar's light. In consequence, the quasar light examined by observers here on Earth can be decoded to show not only the amount of hydrogen the light has encountered, but the entire history of encounters: not only the fact that some of the light has been absorbed, but where along the line of sight the absorption took place.
In this way, quasars become potent tools that allow astronomers to study intergalactic atoms residing between the quasars and Earth. The drawback is that the hyper-luminous episodes that transform a galactic nucleus into a bright quasar last only a tiny fraction of a galaxy’s lifetime. In consequence, quasars are correspondingly rare on the sky, and are typically separated by hundreds of millions of light years from one another. In other words: we need to be very lucky to find a quasar in any one particular location on the sky. Quasars provide only very sparse opportunities to "X-ray" selected portions of the intergalactic medium.
A matter of different length scales
A spatial distribution can exhibit interesting structures on a range of different length scales. The filaments of the cosmic network can be hundreds of millions of light-years long or longer. But their density structure can vary on much shorter scales – for instance, each filament can have a certain density at this particular location, but a slightly different density a few thousand light-years on. If we want to test the cosmological simulations comprehensively, we will need to compare their predictions with observations on all the different length scales.
In the radial direction, comparing more and less distant regions, the technique using the cosmological redshift does not allow for the study of short length scales. The redshift of absorption lines is influenced by the motion of the gas clouds as a whole and by the motion of the atoms in the clouds (Doppler shift); on small scales, there is no clean separation between these effects and the effect of the cosmological redshift.
Thus in order to study the web's changing density on a scale of, say, ten thousand light years, we need to target areas on the sky that lie almost in the same direction, corresponding to regions of the cosmic web that are side by side, ten thousand light years apart, and compare their densities.
The problem is that each single quasar will only give us one glimpse of the cosmic web. To compare regions, we would need at least to glimpses, and in close proximity. This is where the scarcity of quasars becomes a problem.
The search for quasar pairs
A sizeable part of the work of Joseph Hennawi and his collaborators over the last few years was dedicated to solving this problem by finding ultra-rare pairs of quasars, that is, quasars that are right next to each other on the sky. Given that finding a quasar in a particular region is rare enough, the requirement of finding two quasars right next to each other makes these quasar pairs exceedingly rare. Normally one would have to observe 10,000 pairs of objects before finding a quasar pair, making the search for quasar pairs appear hopelessly inefficient. Hennawi pioneered the application of algorithms from ‘machine learning’, a branch of artificial intelligence, to efficiently locate quasar pairs in the massive amounts of data produced by digital imaging surveys of the night sky.
Hennawi led an independent research group at the Max Planck Institute for Astronomy (MPIA) from 2009-2016; during that time, he and his collaborators conducted a systematic search for quasar pairs in the data bases of large surveys, using databases like the Sloan Digital Sky Survey, the Baryonic Oscillation Spectroscopic Survey (BOSS), and the 2dF quasar redshift survey. Owing to these efforts, several hundreds of pairs of quasars are now known. But since the spectroscopy of these surveys is not suited to mapping the spectra of very close quasar pairs, the astronomers needed to take custom spectra of the pairs they had identified.
Quasar pairs, observations, and statistics
Once a quasar pair is found, it can be used to study subtle differences in the absorption of intergalactic atoms measured along the two sightlines. Even then, it is anything but straightforward to make deductions about the structure of the cosmic web from what is, after all, still a fairly limited sample of glimpses.
When Alberto Rorai started to work as a PhD student at the Max Planck Institute in 2011, the current project, of developing a new method to measure the small-scale structure of the cosmic web from quasar pairs, was conceived. Rorai's task was to develop the appropriate mathematical and statistical tools for the job. To this end, Rorai developed different analysis techniques that were tailored to the questions the researchers were most interested in answering. One such question is the way that gas pressure serves to disperse particularly dense regions, smoothing the structure of the cosmic web on small scales.
Rorai applied his tools to spectra of 25 quasar pairs obtained with Hennawi and other colleagues on the largest telescopes in the world, including the 10m diameter Keck telescopes at the W. M. Keck Observatory on Mauna Kea, Hawaii, as well as two telescopes located in the Chilean Atacama Desert: the 8m diameter Very Large Telescope (VLT) at the European Southern Observatory (ESO) and the 6.5m diameter Magellan telescope at Las Campanas Observatory. Light from these quasars has taken between 10.4 and 12.3 billion years to reach us (z=2.0 to 4.4). Absorption features in the light of these quasars can show us features of the cosmic web as it was between 1.4 billion years after the big. The quasar pairs are so close together that they allow for an exploration of the cosmic web on a scale of a few hundred thousand light-years – almost the length scale of individual large galaxies.
The aftermath of re-ionization
The astronomers compared the result of this analysis to supercomputer models that simulate the formation of cosmic structures from the Big Bang to the present. In these models, there are certain parameters that describe the temperature variations within the gas of the cosmic web. These parameters, in turn, contain information about cosmic history between about 650 million and 2.2 billion years after the big bang. In this time period, known as re-ionization, the intense radiation from the first stars, and from active nuclei from some of the first galaxies, heated up and ionized the intergalactic hydrogen and helium – an important phase transition in cosmic history.
Over the following billions of years, the gas in the cosmic web slowly cooled. But temperature variations are correlated with density fluctuations – wherever the temperature rises, the gas will expand, reducing its density. These density modifications are much more long-lived than the temperature differences, and preserve a record of the temperature structure: By measuring small-scale fluctuations using quasar pairs, it is possible to draw inferences about the temperature of gas in the cosmic web during the reionization period – the details of which, including the exact timing, are still one of the biggest open questions in the field of cosmology.
Comparison with simulations
Drawing these inferences from a comparison of simulations and observations, takes massive computational effort. The simulations in question were run by José Oñorbe, a post-doctoral researcher at the Max Planck Institute for Astronomy. They incorporate both how dark matter particles flocked together after the Big Bang phase, forming the backbones of the cosmic web and the evolution of the hydrogen and helium gas clinging to the filamentary structure. The researchers compared 13 alternative possibilities for this evolution, with different parameters for the temperature structure and the properties of pressure smoothing out any homogeneities, with their observations. On a single laptop, these complex calculations would have required almost a thousand years to complete, but modern supercomputers enabled the researchers to carry them out in just a few days.
In the end, the astronomers found a set of parameters that agreed with their observations – linking quasar pair observations to the small-scale structure of the universe around 11 billion years ago, to the thermal history after the reionization era. This is good news for the current models of cosmic evolution: there does exist a set of parameters for which these models can explain the quasar pair measurements. The parameter values themselves, determined by the comparison, fill in important details about the reionization phase. Specifically, the comparison shows significant smoothing on length scales of 200,000 to 300,000 light-years in the present universe – which is in line with the prediction of the best current predictions for structure formation and reionization.
Both Hennawi and Rorai have since moved on from MPIA: Hennawi to the University of California, Santa Barbara, in 2015, where he took up a professorship; Rorai moved on to a postdoc position at the University of Cambridge in 2014.