Selected Research Projects

Intracluster Gas Dynamics

Direct Methods - ICM Bulk Motions

Clusters of galaxies are the largest gravitationally bound systems in the universe, containing 102-3 galaxies and also large amounts of hot (107-8 K) X-ray emitting intergalactic gas confined by the gravitational potential of the system. The X-ray emitting gas makes up the largest baryonic component in these systems (>80%). Currently favored cosmological models involve the “bottom-up” scenario for large-scale structure formation. The formation of galaxy clusters is then due to the continuous accretion of smaller systems. One of the consequences of this formation process is the production of intracluster gas bulk flows, transitory or rotational.

I am leading a project to measure directly these bulk flows using the Doppler shift of the lines in the X-ray spectra of clusters. The initial results with ASCA revealed a strong velocity component (1-2 × 103 km/s) in a relatively high fraction (~5%-10%) of the sample, including the nearby Perseus and Centaurus clusters.




 Perseus cluster. X-ray contours are overlaid. The radius of each circular region is 20¢. 1¢ ~ 30 h50-1 kpc.


X-ray Spectra near the FeK lines for P1 and P5 (of the left figure) using the ASCA GIS 2 (TOP) & GIS 3 (BOTTOM) spectrometers.




Azimuthal Gas Velocity Distribution for the Centaurus cluster. Regions are analogous to those used in Perseus, but are only 5′ away from the center and with a radius of 3′. 1′ ~ 19 h50-1 kpc. The horizontal lines show the 1σ errors for the central region.


My research currently has expanded to include archival and new Chandra and XMM-Newton satellites. The results imply that a significant fraction of the total merger energy in some cases may still be kinetic more than 0.4 Gyr after the last merging event and also that even strong mergers may not show strong signs of turmoil in the projected temperature or density structure despite the energetics involved (1063-64 ergs). One of the clusters Abell 576 analyzed recently with both Chandra and XMM-Newton has the characteristics of a high velocity impact (>3300 km/s) in the line-of-sight, similar to the “bullet” cluster turned 90˚ towards the observer. For a press release click here. Using the typical masses and bulk velocities values currently observed, the ratio of bulk kinetic to gravitational energies in cluster’s central regions can be more than 5 times higher the typical currently desired/achievable level for cosmological use. 


Indirect Methods, Cold Fronts, Dark Matter Detection - Cluster-cluster/group mergers are the most energetic events in the universe. When they happen near the plane of the sky, they produce several observable features that allow us to infer the dynamics of the intracluster gas, such as shock fronts, and also to determine the properties of dark matter (e.g. the bullet cluster). A puzzling intracluster gas feature discovered after the launch of Chandra is the so-called “cold front”, which is a sharp surface brightness discontinuities characterized by a jump in temperature, maintaining the gas pressure continuous across the front. The most popular explanation for cold fronts is associated with subsonic or transonic motions of accreted substructures (Markevitch et al. 2000, 2001) such as gas clumps or small galaxy groups (figure below-left). This mechanism does well for the more typical cases of cold fronts where the clusters do not exhibit signs of strong mergers such as Abell 496.



Cold front model. During cluster mergers the remnant cores pass through each other without mixing. From Markevitch et al. (2000).


Deep Chandra image of the Bullet cluster. Shown in green are mass contours from weak lensing - reconstruction. From Clowe et al. 2006.

 One of the most extreme cases of clusters mergers is the Pandora cluster, discovered very recently (Merten, Coe,
Dupke, et al. 2011 - see also for a video about it and  press release
at The cluster seems to be formed by a near simultaneous 4-fold collision of similar
mass clusters. The complexity of this merger has produced a series of unexpected and unseen effects, such as a dark cluster (clump of dark
matter and galaxies but without X-ray gas) and an apparent Ghost cluster (a clump of gas without dark matter or bright galaxies)

HST image of the core of the Pandora cluster. Many strong  lensed gravitational arcs are easily seen. (courtesy of Dan Coe)


Composite of  the optical (background) X-ray (red) and mass (DM) map (blue). The multlple segregation of barionic and dark components is easily seen in the central region as well as the Ghost cluster (top-right) and the Dark cluster (middle-right).

My research has shown that alternative mechanisms such as gas sloshing due to off-center passages of pure dark matter clumps as modeled by Ascasibar & Markevitch (2006) match well the observations, and we have verified specific model predictions such as cold spiral X-ray arms in Abell 496 (below).





(a) – X-ray image of Abell 496. North is up. The locations of the cold fronts are indicated by the small blue arrows. The larger black arrows show the overall direction of the cold spiral arm (to the North) and also a cold tail (in the South) for comparison with the temperature map below.


(b) - Adaptively smoothed temperature map of Abell 496. We also overlay the X-ray contours. The outermost square contour corresponds to the CCD border, with dimensions ~320 kpc.


(c) - Zoom in of the core of simulated cluster 0.5 Gyr after the flyby of massive pure dark matter halo. From Ascasibar & Markevitch (2006). Yellow is ~8 keV and blue 2 keV. Dark matter density contours are overlaid and arrows indicate gas velocity (the longest corresponding to 500 km/s. The size of the panel is 250 kpc. The figure has been flipped vertically to match the configuration of the cold front in A496 shown in (a) and (b).







Virtual Cluster Exploratory - I am Co-investigator (PI: Gus Evrard) on a project to develop a Virtual Observatory (VCE) to compare virtual galaxy clusters, extracted from cosmological simulations to real ones, helping to zero in on the correct ICM models and test different cosmological recipes. A recently completed VCE project was focused on determining the expected distribution of clusters with velocity gradients (Pawl, Evrard & Dupke 2005). One of the results from this work was that the maximum velocity differences ΔV/cs in clusters fall as (ΔV/cs)-4 (Figure below). This indicates that the detections that we found with ASCA, Chandra and XMM-Newton maybe just the “tip of the iceberg”, with many more clusters expected to have intermediate (~700-1000) km/s) residual velocity gradients.  


Missing Baryons in the Local Universe: One of the current mysteries that has drawn a significant amount of attention is the fact that stars and gas in galaxies in the local universe accounts for only for a small fraction (<20%) of the baryonic mass (Fukugita & Peebles 2004; Danforth & Schull 2005; Nicastro et al. 2005). This is typically denoted as the “missing” baryon problem, which has drawn much attention, and the current leading solution is that most of the gas has remained in the gas phase (Fukugita, Hogan, and Peebles 1998; Cen and Ostriker 1999). However, since the Lyα forest in the local universe is small, accounting for ~29% of the baryon content (Penton et al. 2004; Sembach et al. 2004), a significant fraction of the gas must be in a warm-hot (> 105 K) phase (Cen and Ostriker 1999; Dave et al. 1999, Cen et al. 2001), which has evaded direct detection so far.

Although these filaments are too diffuse and cold to be directly observed by current X-ray telescopes, they can be detected indirectly using absorption lines of background AGNs. Cosmological simulations indicate that these cosmic filaments would be most prominent in superclusters of galaxies. In an initial pilot study of archived HST and FUSE AGN data, we found three AGNs located behind the lines connecting clusters in superclusters that could be used to detect the Cosmic Web  (Bregman, Dupke & Miller 2004), and one of them shown below. We are now carrying out a larger follow-up project to establish the detection of the cosmic filaments with better statistics in order to determine the characteristics of the filaments and explain the local baryon budget deficit using HST (for Lya) and FUSE (to get Lyb and OVI lines) and Sloan Digital Sky Survey, Las Campanas redshift survey and ROSAT North Ecliptic Pole Survey to locate superclusters. The latter (NEP Supercluster) has recently been approved to us for observation with the FUSE satellite.






FUSE spectrum of PHL1811 in the frequency range corresponding to Aquarius B. The Lyb lines are easily seen. The region encompassing ~1300 km/s from the closest cluster of galaxies is shown on the top of the plot. A Lya system is also seen at the redshift of Aquarius-Cetus with HST STIS.


Fossil Groups - Fossil groups are galaxy systems that present an unusual lack of bright galaxies in the inner regions, except for of a giant central E galaxy. Recent measurements of galaxy velocity dispersion in fossil groups are consistent with the dynamical state of the system as determined from X-ray observations. This indicates that they have relatively deep gravitational potential wells, more typical of clusters. The popular mechanism proposed to "wipe out" the big galaxies surrounding the central dominant galaxy is still cannibalism. If so, and if the galaxy population is heterogeneous, strong galactic winds resulting from galaxy merging might be trapped by their deep potential wells destroying the central enhancement of SN Ia/SN II ejecta ratio typically seen in other galaxy groups. I have started testing this prediction by looking at the distribution of metal abundance ratios near the core of a sample of fossil groups. Initial results shows an enhancement of SN II ejecta near the core of some fossil groups, where the available data allows for such analysis.


Discrimination between SN Explosion Mechanisms: The presence of gradients of the ratio of individual elemental abundances in the ICM allows the opportunity to test the theoretical explosion models for SN Ia. I pioneered this technique and tested it using the Ni/Fe ratio to compare the two most favored models for SN Ia explosion: the W7 “deflagration” model and the “delayed detonation” models (Dupke & White 2000, Dupke & Arnaud 2001).  These models differ in the characteristics of the propagation front and, consequently, in the nucleosynthetic yields for different elements. The results so far do not favor the delayed detonation explosion models. The improvement of the spectral resolution of X-ray spectrometers will soon allow the same technique to be applied with other abundance ratios and it is expected that it will provide a diagnostic for testing further SN explosion models.