Below is a brief round-up of some of the key results from my academic research, with links to summary pages for the individual published papers where the full details can be found. Much of my work involves analysing data from the orbiting X-ray telescopes Chandra (run by NASA) and XMM-Newton (run by ESA). All of the plots shown on this page were created in R; see the R plot gallery for more examples of R plots and the code that produced them.



Cool cores in clusters of galaxies erased by mergers

Many galaxy clusters (around 50% to 75%) host so-called cool cores in the hot gas that fills the space in between the galaxies. A cool core is a compact central region where strong gas cooling is taking place. Ultimately this cooling gas provides material that can fuel star formation as well as feed the growth of supermassive black holes, which are found at the centres of all large galaxies.

Gas density cuspiness vs. X-ray/BCG offset

This plot shows a strong correlation between the inner steepness of the gas density profile and the projected offset between the centroid of this X-ray emissing gas and the brightest cluster galaxy (BCG) for a sample of 65 galaxy clusters from the LoCuSS survey.

This steepness parameter quantifies the strength of gas cooling in the cluster core, while the X-ray/BCG offset measures dynamical disturbance: in relaxed clusters, the brightest galaxy usually lies at the centre of the X-ray halo.

The strong correlation is consistent with the view that cluster merging (i.e. large scale disturbance) can weaken and even destroy cool cores, which have previously been thought to be able to resist such disruption. More information can be found in the published paper. To see the full size version of the plot, click here.




Cool core/non-cool core bimodality

The gas in the cores of some galaxy clusters shines so brightly in X-rays that it can lose enough energy to lower its temperature significantly over the lifetime of the cluster (billions of years). Without additional heating, the cluster quickly develops a so-called cool core, where the gas temperature drops sharply in the centre.

Gas entropy profiles for a sample of 20 galaxy clusters

Cool-core and non-cool core clusters are found in roughly equal numbers and it was previously thought that all clusters spanned a continuous range from one type to the other. However, this plot of the entropy profiles of 20 nearby galaxy clusters shows instead a clear separation into two distinct types, corresponding to clusters with cool-cores (steep profiles) and those without (flatter profiles).

The entropy of the gas is a powerful probe of cluster physics and provides great insight into the state of the gas.

The bimodal behaviour is revealed even more clearly in the plot below, which shows the smoothed distribution of logarithmic slopes for these entropy profiles (shown individually as jittered points). There are two widely-spaced humps, which are well-described by a pair of Gaussian curves (dashed lines), fitted using maximum likelihood with the task fitdistr in the MASS package in R.

Kernel-smoothed density distribution of entropy profile log slopes

The logarithmic slopes were measured by using quantile regression (with the quantreg package in R) to fit a power law to the entropy profiles (a straight line on a log-log plot). This is a robust form of regression in which the median of y (rather than the mean) is modelled as a linear function of x. This robustness discounts minority outliers and allows the fitted model to capture the steepness of the curves, without being influenced by any flattening off in the few points near the centre.

The two peaks in the distribution correspond to cool-core (steep entropy profiles, i.e. large slope) and non-cool core (shallow slope) clusters. The bimodality can be explained by the impact of thermal conduction redistributing heat within the gas. A entropy slope of 2/3 corresponds to a critical line, above which heat from the cluster outskirts can be transported fast enough to the core to offset the energy lost from cooling. However, entropy profiles steeper than 2/3 cool so quickly that even maximal heat conduction from the outskirts cannot prevent rapid temperature decline, leading to the formation of a cool core. A key factor in controlling this thermal conduction is the strength and orientation of magnetic fields, since the charged plasma particles in the hot gas preferentially move along magnetic field lines.

More information can be found in the published paper. To see the full size version of the two plots, click here and here.




Unexpected similarity breaking in cluster entropy scaling

Clusters of galaxies are the largest gravitationally bound structures in the Universe (comprising around 80-85% dark matter, by mass). Since gravity has no preferred scale, the properties of these massive objects are expected to vary in a self-similar way- a small cluster of galaxies ought to resemble a scaled-down version of a large cluster. This is because gravity pulls gas faster into more massive clusters, creating stronger shock waves that then heat it up to hotter final temperatures.

However, the radiation we observe from clusters (e.g. light from stars, X-rays from hot gas) comes from the small (less than 15-20%) fraction of mass in normal matter (baryons) in clusters, therefore cluster properties are also influenced by non-gravitational heating and cooling processes. Heating, in particular, is very important as it reveals the impact of supernova-driven galaxy winds and active galactic nuclei (AGN), powered by supermassive black holes. The hot gas in clusters ultimately cools to fuel the associated star formation and feed black holes, forming a crucial link in the cosmic feedback cycle.

Entropy vs. mean temperature

The entropy of the hot gas is an excellent probe with which to study galaxy feedback in clusters, since it would scale in direct proportion to the mean temperature of a self-similar cluster. However, real clusters do not behave in this simple way. The plot shows instead that gas entropy varies with mean temperature to the power 2/3 (solid line) for a sample of galaxy groups and clusters, compared to the self-similar expectation (dotted line).

This behaviour is caused by excess entropy in the coolest clusters and groups- either because the lowest entropy gas has cooled out to form stars, or because supernova or AGN have heated it up. Both cooling and heating from non-gravitational sources like these have relatively greater impact in smaller (cooler) systems. This leads to the flattening of the relationship between entropy and mean temperature, rather than a floor at low temperatures with a sharp transition to a self-similar slope of 1 in hotter clusters, as previously thought.

More information can be found in the published paper. To see the full size version of the plot, click here.




Systematic variation in gas content of galaxy groups and clusters

Groups and clusters of galaxies are filled with hot (millions of degrees) gas, containing up to five times as much mass as the stars in the galaxies themselves. At one time, all the normal material (so-called baryons) in the Universe existed in such a hot phase, but cooling has led to much of it condensing out to form stars, as well as feed supermassive black holes in the cores of massive galaxies.

In groups and clusters, both the gas and the galaxies are tightly bound by the gravity of the dark matter, which is around 5-10 times more massive than all the baryons put together. This ratio is similar but somewhat less than that found in the Universe as a whole. Dark matter accounts for roughly five times the mass of normal matter in the Universe, as deduced from the NASA WMAP mission, for example, which studies the Cosmic Microwave Background (CMB) relic radiation.

Gas mass fraction vs. mean temperature

You would expect that the fraction of mass in the form of hot gas would be similar in clusters of different masses, if only gravity was important in dictating its behaviour. However, this plot shows that the so-called gas fraction actually declines quite strongly in less massive clusters and groups of galaxies (which have cooler mean temperatures).

The blue shaded region shows the average fraction of mass in normal matter (baryons) in the Universe, measured by WMAP. This level represents an upper limit to the expected hot gas fraction in clusters, since some of the baryons in clusters exist in the form of stars.

The systematic trend seen in the plot means that cosmic feedback must be important in clusters, such as heating of gas via outbursts from supermassive black holes (so-called Active Galactic Nuclei, AGN) or supernova-driven galaxy winds. These processes heat and disrupt the hot gas, but have a disproportionate effect in smaller systems, which are less able to retain the gas as it is blown outwards. The result is a lower gas fraction for cooler clusters and groups.

More information can be found in the published paper. To see the full size version of the plot, click here.




Surprisingly short cooling times in non-cool core galaxy clusters

As objects emit light of any wavelength, they lose energy and will gradually cool unless this energy is replenished from some other source. In the cores of galaxy clusters the X-ray emitting hot gas shines so brightly that it can cool substantially over the lifetime of these objects (a few billions of years, compared to roughly 13.5 billion years for the age of the Universe, also known as the Hubble Time).

Not surprisingly, therefore, around half of all clusters have cool cores, where the temperature has dropped in the centre as energy has been radiated away. Nevertheless, many clusters show little or no sign of declining central temperatures. In some clusters this could be the result of recent disruption from merging, but this cannot be true in all cases as large cluster mergers are simply not frequent enough to head-off the development of a cool core persistently.

Profiles of hot gas cooling time for a sample of 20 galaxy clusters

The mystery of why so many clusters don't have a cool core deepens on investigation of the cooling time of the gas. The cooling time simply measures how long the gas can continue to emit X-rays at the present rate, before it runs out of energy: short cooling times imply rapid cooling. This plot shows how the cooling time varies with radius for a representative sample of 20 galaxy clusters.

As expected, clusters with cool cores have the shortest cooling times, but what is surprising is that many non-cool core clusters also have short cooling times (compared to the typical few billion-year lifetime of most clusters).

Could this be the result of cosmic feedback heating up the gas? Well, exploding supernova don't pack enough of a punch to do the job, but outbursts from supermassive black holes (in Active Galactic Nuclei, AGN) could be energetic enough, in theory. The trouble is, AGN activity is actually known to be much more common in clusters with cool cores than those without (most likely because the cooling gas actually feeds the black hole and powers the outbursts).

A more plausible explanation is that thermal conduction is able to transport heat from further outer in the cluster as soon as the temperature starts to drop in the core, thus keeping it warm. This can also explain why cool-core and non-cool core clusters appear to form bimodal populations. More information can be found in the published paper. To see the full size version of the plot, click here.




Modelling dark matter and hot gas in clusters of galaxies

Most of the mass in a galaxy cluster (80-85%) comprises dark matter, which cannot be directly observed. However, we know that this dark matter must be present in order for gravity to hold together all the fast-moving (around 1000 km/s) galaxies and the hot (millions of degrees), high-pressure X-ray emitting gas.

When hot gas settles down within the cluster as it forms, eventually the outward pressure of the gas (which depends on the density times the temperature) exactly balances the gravitational pull of the dark matter inwards, in a state called hydrostatic equilibrium. You can then figure out the distribution of the otherwise hidden dark matter by mapping the gas pressure profile entirely from its X-ray emission, assuming the overall shape is a sphere.

Profiles of residuals...

However, the traditional drawback with this approach is that it is only really suited to the brightest clusters, with the highest quality data. This is because of the highly flexible models fitted to the observations, which require lots of data points to help pin down their parameters accurately. However, a simplified cluster model developed recently by Ascasibar & Diego largely avoids this problem. Their phenomenological model has just 5 free parameters, yet does a good job of capturing the behaviour of real clusters.

Unlike in many previous studies, the gas temperature and density are not treated separately, but are jointly modelled: the assumption that the gas is in hydrostatic equilibrium implies that the density and temperature are coupled together. The above plot shows the residuals from the model for both the gas density and temperature, for a representative sample of 20 galaxy clusters. The lines and smoothed marginal distributions show no substantial trend and are reasonably consistent with the statistical noise expected, given the precision of the measurements. In other words, this relatively simple model is generally a good fit to the data for real clusters.

The success of this model in describing accurately a wide variety of real clusters means that the dark matter mass can now be determined for many more such objects, with relatively poorer quality data. The cluster model also allows many derived quantities to be calculated easily, such as the inner steepness of the gas density profile, which quantifies the strength of cooling in cluster cores.

More information can be found in the published paper. To see the full size version of the plot, click here.

Quick links


Jump to


Copyright © 2010-2013 Alastair Sanderson