Intergalactic Shock Fronts


When galaxy clusters collide, they generate shock fronts in the hot intracluster medium. Observations of these shocks can provide valuable information on the merger dynamics and physical conditions in the cluster plasma, and even help constrain the nature of dark matter. To study shock fronts, one needs an X-ray telescope with high angular resolution (such as Chandra), and be lucky to see the merger from the right angle and at the right moment. As of this writing, only a handful of merger shock fronts have been discovered and confirmed using both X-ray imaging and gas temperature data — those in 1E 0657–56, A520, A754, and two fronts in A2146. A few more are probable shocks awaiting temperature profile confirmation — those in A521, RXJ 1314–25, A3667, A2744, and Coma. The highest Mach number is 3 in 1E 0657–56, while the rest has . Interestingly, all these relatively weak X-ray shocks coincide with sharp edges in their host cluster’s synchrotron radio halos (except in A3667, where it coincides with the distinct radio relic, and A2146, which does not have radio data yet). This is contrary to the common wisdom that weak shocks are inefficient particle accelerators, and may shed light on the mechanisms of relativistic electron production in astrophysical plasmas.

Harvard-Smithsonian Center for Astrophysics,
60 Garden St., Cambridge, MA 02138, USA
Review talk at 12th Marcel Grossman Meeting, Paris, July 2009 (updated with 2010 results)

Keywords: Galaxy clusters; Intergalactic medium; Shock fronts; Dark matter; Cosmic rays

1 Introduction

Mergers of galaxy clusters provide a unique laboratory to study the cluster physics — they generate shocks, hydrodynamic instabilities and turbulence in the hot intracluster medium (ICM). They also cause “cold fronts”, or contact discontinuities, around the cores of infalling subclusters or in the disturbed cool core of the main cluster. In turn, these ICM disturbances distort and amplify magnetic fields (which are frozen into the ICM) and generate ultrarelativistic electrons that produce diffuse synchrotron emission of cluster-wide radio halos. Of these phenomena, with the current instruments, we can directly observe shock fronts and cold fronts as sharp brightness and temperature edges in high angular resolution X-ray images — Fig. 1 shows a poster child of the merger shocks, the “bullet cluster” 1E 0657–56, and Fig. 2 shows a cold front in the “cannonball galaxy” NGC 1404. In the radio, the synchrotron emission is a product of density of ultrarelativistic electrons and , where the magnetic field is deduced by indirect means. Disentangling the relativistic electron density and the magnetic field (e.g., by observing inverse Compton emission from the radio halos), as well as direct observation of turbulence in the ICM, are beyond the current observing capabilities.



Fig. 1: The Bullet cluster 1E 0657–56. (a) Chandra X-ray image (0.8–4 keV band, size), (b) gas temperature map (color; blue is keV and yellow is keV) and X-ray brightness (contours). A cool gas bullet is preceded by a hot shock front.



Fig. 2: Elliptical galaxy NGC 1404 falling toward the center of the Fornax cluster. (a) Chandra X-ray image (0.5–2 keV band, size), (b) gas temperature map (a zoom-in on NGC 1404). The sharp NW boundary of the galaxy is a cold front.
Fig. 3: Schematic gas density, pressure and specific entropy profiles in a sector of the Bullet cluster crossing the bullet nose and the shock front. The density jump around is a cold front (the front boundary of the gas bullet), and the jump at is a shock front.

Figure 3 schematically shows the ICM density, pressure and specific entropy profiles in a sector in front of the “bullet” in the Bullet cluster. The sector crosses a cold front (the boundary of the bullet) and the shock that propagates ahead of it (Fig. 1). While both these X-ray brightness edges exhibit similar density jumps, the pressure across the cold front is almost continuous, as expected for a contact discontinuity between the dense, low-entropy remnant of an infalling subcluster and the surrounding shock-heated gas. The shock front exhibits a large pressure jump and a modest entropy increase, corresponding to its relatively low Mach number, .

Cold fronts are more easily observable than shock fronts – they were first discovered with Chandra in a merging cluster A2142 and subsequently observed in most mergers. More surprisingly, they are also seen in the cool cores of at least half of all relaxed clusters. Merger shock fronts quickly move away from the central, bright cluster regions into the faint outskirts, where they are difficult to observe in X-rays. To be discernible in an X-ray image, they also require a merger to occur almost exactly in the plane of the sky. For these reasons, only a handful is currently known, and they are the subject of this review. There are other types of shock fronts expected in clusters — far in the outskirts, a strong shock should separate the virialized cluster region and the outside infalling matter, but the X-ray brightness at those distances is far too low for these shocks to be observable at present. Explosions of the central AGNs also generate shocks in the cluster cores, usually with , but sometimes much stronger. These shocks are outside the scope of this review.

2 Constraints on dark matter from the Bullet cluster

Fig. 4: X-ray image of 1E 0657–56 (same as in Fig. 1a) with contours showing total projected mass from weak lensing. The gas subcluster trails its mass peak because of ram pressure. This offset is direct proof of the existence of dark matter.

We will start with some physical results already derived using the most prominent cluster shock front, the one in the Bullet cluster 1E 0657–56. Fig. 1 shows an X-ray image and projected temperature map derived from a 500 ks Chandra exposure of 1E 0657–56, and Fig. 4 overlays a weak lensing total mass map on the Chandra image. It is clear from these figures that 1E 0657–56 is a merger of two subclusters flying apart just after a collision. The cooler gas bullet, with its characteristic shuttlecock shape, appears to be a remnant of the cool core of the smaller subcluster, pushed back from its host dark matter peak by ram pressure. The X-ray brightness and temperature profiles across the shock front give a Mach number () and a velocity of the shock front (4700 kms). The velocity of the subcluster is probably lower than that of the shock front, because the pre-shock gas flows in from the cluster outskirts toward the shock front under the gravitational pull of the subcluster. The two subclusters should have passed through each other just 200 million years ago.

It has been quickly realized that the offset between the peaks of the total mass and the gas mass peaks seen in Fig. 4 offers the first direct and model-independent evidence for the existence of dark matter, as opposed to some forms of modified gravity, where the visible matter is all there is and the laws of gravity are incorrect on large scales — an alternative possibility put forward to explain the longstanding problem of “missing mass” in galaxy clusters. Indeed, the X-ray emitting gas is by far the dominant visible mass component in clusters, and yet, as the lensing mass map shows, the peaks of the total mass density are clearly located elsewhere. The measured total masses in those peaks are as expected from the ratio of gas mass, stellar mass in the galaxies, and total mass normally observed in clusters. What makes this cluster unique is the moment at which it is caught by the observer — when gas and dark matter have been temporarily spatially separated by a violent merger, revealing that these are indeed two different kinds of matter.

The survival of two dark matter subclusters after a near-direct collision (indicated by the X-ray data) also provides an upper limit on self-interaction cross-section for the dark matter particles. The dark matter peaks coincide with peaks of the galaxy number density, and both dark matter and galaxy peaks are located ahead of the gas, as expected for collisionless dark matter and galaxies vs. fluid-like ICM. The observed mass-to-light ratio within the gas-depleted subcluster peaks is found to be similar to that in other clusters within similar radii. This excludes the possibility that a significant fraction of dark matter particles has escaped the subclusters as a result of particle collisions during the subclusters’ recent passage through each other. Taking into account the merger velocity derived from the X-ray, this observation places a limit of cm/g on the dark matter self-interaction cross-section., excluding most of the astrophysically interesting range of the cross-section..

3 Electron-proton equilibration timescale in the ICM



Fig. 5: Simulations of a cluster merger, with colors showing the temperature and contours showing the mass. There are two shock fronts propagating ahead of two subclusters that have just crossed each other. (a) Average plasma temperature, (b) electron temperature assuming Coulomb electron-proton equilibration timescale. The linear scale is in Mpc and the temperature scale is in keV. The X-ray measured may be a significant underestimate of thermodynamic temperature at shocks.

The passage of a merger shock in a fully ionized intracluster plasma should heat the protons dissipatively, while electrons (for shocks with , which is always true for cluster mergers) are compressed adiabatically and subsequently heated by other mechanisms, such as Coulomb collisions with hotter protons. The linear sizes and temperatures of galaxy clusters are such that on a timescale of collisional electron-proton equilibration, , the shock travels significant distances, creating observable regions where the electron temperature (the only one that we can currently measure in X-rays) is below the true thermodynamic temperature. This is illustrated in Fig. 5, which shows simulated maps of the average and electron temperatures for a merging cluster with two shocks propagating outwards, assuming collisional . The nonequilibrium is expected in may types of astrophysical objects, from solar wind to supernovae to WHIM filaments. However, it is usually impossible to measure and independently and map them on linear scales on which their equilibrium is expected to be achieved. Shock fronts in clusters provide a unique opportunity to do so and thus constrain . Because cluster shocks are relatively weak, the easily measurable plasma density jump at the shock is sufficiently far from its asymptotic value (factor of 4 for a ), which allows one to derive the Mach number of the shock from the density jump using the Rankine-Hugoniot jump conditions, and predict the equilibrium post-shock value of the plasma temperature independently of the measured jump in . The pre-shock gives the sound speed, which, together with the shock density jump, gives the speed of the post-shock gas flow relative to the shock front. This can be used to predict the post-shock profile for various values of . For the shock in 1E 0657–56 with , the expected rise in post-shock for the case of Coulomb collisions can be spatially resolved by Chandra, as shown in Fig. 6. The measured values of the post-shock exclude the Coulomb timescale at a 95% significance, favoring a much shorter .

Fig. 6: Predicted profiles for the shock in 1E 0657–56, for the Coulomb electron-proton equilibration timescale (blue band) and for instant equilibration (yellow band). Overlaid is the Chandra measurement (deprojected, with error bars), which indicate a shorter equilibration timescale.

4 Other clusters with known shock fronts

Until recently, only two merger shock fronts were known and confirmed by the X-ray temperature measurement — the 1E 0657–56 discussed above and A520 shown in Fig. 7a. The latter has a shock with ; its recent long Chandra observation is currently being analyzed to validate the above constraint on . A520 also exhibits a similar offset between gas and dark matter.

Recently, several more shock fronts and shock candidates were found in X-rays. A unique case of two fronts located on the opposite sides of the cluster has been reported in A2146, shown in Fig. 7b — a geometry expected for a symmetric merger with a small impact parameter. Both shocks have . Similar second shocks are not observed in either 1E 0657–56 or A520, most likely because those shocks have already moved out of the bright central regions in these mergers of subclusters of very different masses.

Now that we know how the merger shocks look like, it is possible to find them in archival X-ray data of lower quality. A possible shock front was detected in the ROSAT PSPC image of A754 and recently confirmed using the Chandra temperature measurement across the shock, as shown in Fig. 8. This is a relatively weak shock with .



Fig. 7: (a) Chandra image of A520; the shock front with is to the SW of the bright central structure. (b) Chandra image of A2146, showing two shocks about 300 kpc SE and NW of the center.



Fig. 8: (a) Chandra image of A754 with the sector showing a shock front with . (b) Chandra gas temperature profile showing a temperature jump across the front in a sector shown in panel (a) (radius is from the shock’s center of curvature), confirming that this is indeed a shock.



Fig. 9: (a) ROSAT PSPC mosaic of Coma (), showing a brightness edge in the eastern sector (shown by dashes) that might be a shock front. (b) X-ray brightness profile in that sector, showing this feature at around (red dash).



Fig. 10: (a) Chandra image of A2744 (), showing a subtle brightness edge in the SE sector (shown by dashes) that might be a shock front. (b) X-ray brightness profile in that sector, showing this feature at around (red dash).



Fig. 11: (a) Chandra image of A521, with GMRT 327 MHz synchrotron radio relic contours overlaid, delineating a possible shock front. (b) Chandra X-ray brightness profile in the sector covered by the relic; radius is from the relic’s center of curvature. Overlaid is a model shock profile with predicted from the radio spectrum of the relic, which is consistent with the X-ray imaging data. Gas temperature data with sufficient quality is not yet available to confirm this shock.

As of this writing (August 2010), these are the only known merger shocks with the X-ray temperature profiles confirming that an edge-like feature in the X-ray brightness is indeed a shock front (and not a cold front, for example). There are several other likely shock fronts seen in the X-ray images but awaiting temperature confirmation. One is in Coma cluster, seen in a ROSAT PSPC mosaic (Fig. 9). There is some indication of a temperature jump of the correct sign in the XMM-Newton temperature map, though a measurement more matched to the shock position is needed to verify this. Coma is big enough for this putative shock front to be seen by the Planck observatory that maps the Sunyaev-Zeldovich signal, which is proportional to gas pressure and thus should readily reveal shocks.

Another candidate front is found in the Chandra image of A2744, a spectacular merger at . An X-ray brightness profile in a sector containing the putative shock is shown in Fig. 10. The putative shock is a very low-contrast feature, likely corresponding to a low Mach number. We will see below that both Coma and A2744 exhibit edge features in their radio halo maps that coincide with these putative shocks.

It has long been suggested that peripheral radio relics are caused by electrons accelerated at shock fronts in the periphery of merging clusters.. However, most relics are located far from the X-ray bright cluster regions, so it is rarely possible to check for the presence of a shock front in an X-ray image, and even more difficult to get a confirming temperature measurement.

A likely shock in A521 was discovered by analyzing the Chandra X-ray image at the position of a prominent radio relic, see Fig. 11. There is a subtle X-ray brightness edge right where the “front” side of the relic is. The relic has a good radio spectrum, well represented by a power law. If one assumes that the radio-emitting relativistic electrons come from Fermi acceleration at the shock, the spectrum implies . The density jump for such a shock is in good agreement with the X-ray brightness profile, as shown by red model in Fig. 11b. At lower radio frequencies, this cluster reveals a giant radio halo that starts at the relic and spans the entire cluster (Fig. 12, discussed below).

An X-ray brightness edge that looks like a shock was discovered in an XMM-Newton observation of the famous NW radio relic in A3667; if this is indeed a shock, the gas density jump indicates . A possible shock front was discovered at the position of a radio relic in the cluster RXJ 1314–25. The curious M shape of this front indicates a velocity gradient in the pre-shock gas — possibly an extreme case of the effect seen in the hydrodynamic simulations of the Bullet cluster. Table 1 gives a summary of the currently known merger shocks.

  Cluster jump T jump M Radio edge? X-ray refs. 1E 0657–56 yes yes 3 yes 5, 6 A520 yes yes 2 yes 25 A754 yes yes 1.6 yes 28, 29, 30 A2146 N yes yes 2 no data 27 A2146 S yes yes 2 no data 27 A521 yes 2 yes 33 RXJ 1314–25 yes 2 yes 36 A3667 yes 2 yes 35 A2744 yes yes this work Coma yes yes this work  

Table 1: X-ray merger shock fronts and candidates (as of summer 2010)

5 Radio halos and X-ray shock fronts

(1E 0657–56)






Fig. 12: Contours of radio halos overlaid on X-ray images of clusters with shock fronts (1E 0657–56, A520, A754) or front candidates (A521, A2744, Coma). In all cases, there is coincidence of a shock front seen in the X-ray with an edge in the radio halo.

Merger shock fronts are expected and, as we saw above, observed to have relatively low Mach numbers. Indeed, a test particle falling into the gravitational potential of a typical cluster would acquire . An infalling subcluster would generate multiple shocks in front of it, preheating the ambient gas and further reducing its Mach number by the time it arrives into the cluster’s X-ray bright central region, where we observe it. Such weak shocks are believed to be inefficient accelerators of electrons from the thermal pool to Lorentz factors required to produce relics and radio halos. Yet, it has been noticed that synchrotron radio halos in 1E 0657–56 and A520 display brightness edges coincident with the X-ray shocks (Fig. 12), indicating that these shocks must have something to do with producing radio emission at least at those locations — either accelerating electrons, or compressing or re-accelerating a pre-existing relativistic electron population, and perhaps strengthening the magnetic fields beyond simple compression. Re-acceleration of pre-existing relativistic electrons is an attractive theoretical possibility.

All the clusters with newly discovered X-ray shocks and shock candidates exhibit radio halos or relics, except A2146 that doesn’t have sensitive radio data yet. It is interesting to overlay the radio images on the X-ray shocks, which we do in Fig. 12. In addition to 1E 0657–56 and A520 discussed above, it shows A521, which exhibits a relic coincident with the probable shock, plus a halo extending from that relic across the whole cluster — a radio morphology very similar to A520. A754, with the weakest shock (), exhibits a steep-spectrum relic at 74 MHz, coincident with the shock front, and a radio halo extending from the relic westward across the cluster. Finally, the two shock candidates in A2744 and Coma, which we report in this paper, both have edge-like features seen in recent high-quality radio images of their halos, coincident with putative X-ray shock fronts. In Coma, the X-ray edge spans the southern half of the radio edge (Fig. 12). The cluster RXJ 1314–25 (not shown) also has a relic at the shock front and and a small radio halo extending from it.

Thus, spatial coincidence of the merger shocks with edges of radio halos — or with radio relics that delineate an edge of the radio halo — is quite ubiquitous. It supports the scheme that we proposed for A520 and A521, in which a weak merger shock re-accelerates pre-existing relativistic electrons (which may remain from earlier mergers, or generated by collisions of the long-lived cosmic ray protons with thermal protons). As these electrons move downstream from the shock, they rapidly cool, creating a narrow arc-like “relic”, but at a certain distance are picked up and re-accelerated again by turbulence generated by the merger behind the shock, which produces the cluster-wide halo. In this picture, the edge of the radio halo and the bulk of the halo are distinct phenomena, though both caused by the same merger. Depending on the Mach number of the shock, the magnetic field behind the shock (that determines the rate of synchrotron cooling of the electrons and thus the width of the “relic” in the direction across the shock front), and the power spectrum of turbulence in the body of the ICM behind the shock, the relic may appear at some frequencies as a distinct radio source, while at other frequencies, merge with the halo. In A521, the relic is dominant at GHz, while in A754, the relic dominates at 74 MHz but merges with the halo at higher radio frequencies. The prediction of this picture is that the radio spectrum at the halo “edge” should be a power law determined by the shock’s Mach number (Fermi acceleration from thermal pool and re-acceleration from plausible initial electron distributions result in the same spectrum). This is again consistent with observations of A521 and A754. The radio spectrum of the rest of the halo (more exactly, the frequency of its exponential cutoff) is determined by the velocity of the turbulence and the strength of the magnetic field. Depending on that spectrum, it may or may not be possible to see the steepening of the radio spectrum of the front edge due to aging of electrons as one moves from the shock front inward (as was possible for A521).

We note that some more irregularly shaped cluster radio relics, such as a bright relic in A2744 (seen in the left edge of the image in Fig. 12), A2256 or A3667, may have different origin, perhaps involving a shock passage across a distinct region of fossil radio plasma.

It is clear that detailed, spatially resolved studies of cluster merger shocks in the X-ray and at multiple radio frequencies is a promising way to study the cosmic ray acceleration mechanisms in astrophysical plasmas. We may be starting to collect a sufficiently large sample for such studies.


  • 1. M. Markevitch & A. Vikhlinin, Phys. Reports, 443, 1 (2007).
  • 2. R. Cassano, in The Low-Frequency Radio Universe, ASP Conf. Series, 407, 223 (2009).
  • 3. C. Sarazin, JKAS, 37, 433 (2004).
  • 4. Brunetti, G., Cassano, R., Dolag, K., & Setti, G., A&A, 507, 661 (2009).
  • 5. Markevitch, M., et al., ApJ Lett., 567, L27 (2002).
  • 6. Markevitch, M., in The X-ray Universe 2005, 604, 723 (2006).
  • 7. Markevitch, M., et al., in preparation (2010).
  • 8. Machacek, M., et al. ApJ, 621, 663 (2005).
  • 9. Carilli, C. L. & Taylor, G. B. ARAA, 40, 319 (2002)
  • 10. Markevitch, M., et al., ApJ, 541, 542 (2000).
  • 11. Markevitch, M., Vikhlinin, A., Forman, W. R. in Matter and Energy in Clusters of Galaxies, ASP Conf. Series, 301, 37 (2003).
  • 12. Croston, J. H., et al. MNRAS, 395, 1999 (2009).
  • 13. Million, E. T., & Allen, S. W. MNRAS, 399, 1307 (2009).
  • 14. Owers, M. S., Nulsen, P. E. J., Couch, W. J., & Markevitch, M. ApJ, 704, 1349 (2009).
  • 15. Clowe, D., Gonzalez, A., & Markevitch, M. ApJ, 604, 596 (2004).
  • 16. Clowe, D., Bradač, M., Gonzalez, A. H., Markevitch, M., Randall, S. W., Jones, C., & Zaritsky, D. ApJ, 648, L109 (2006).
  • 17. Springel, V., & Farrar, G. R. MNRAS, 380, 911 (2007).
  • 18. Milgrom, M. ApJ, 270, 365 (1983).
  • 19. Bekenstein, J. D. Phys. Rev. D, 70, 083509 (2004).
  • 20. Zwicky, F. ApJ, 86, 217 (1937).
  • 21. Markevitch, M., et al. ApJ, 606, 819 (2004).
  • 22. Randall, S. W., et al. ApJ, 679, 1173 (2008).
  • 23. Spergel, D. N., & Steinhardt, P. J. Phys. Rev. Lett., 84, 3760 (2000).
  • 24. Takizawa, M. ApJ, 520, 514 (1999).
  • 25. Markevitch, M., Govoni, F., Brunetti, G., & Jerius, D. ApJ, 627, 733 (2005).
  • 26. Clowe, D., et al., in preparation (2010).
  • 27. Russell, H. R., et al. MNRAS, 406, 1721 (2010).
  • 28. Krivonos, R. A., Vikhlinin, A. A., Markevitch, M. L., & Pavlinsky, M. N. Astronomy Letters, 29, 425 (2003).
  • 29. Henry, J. P., Finoguenov, A., & Briel, U. G. ApJ, 615, 181 (2004).
  • 30. Macario, G., et al., ApJ, submitted (2010).
  • 31. Wik, D. R., et al. ApJ, 696, 1700 (2009).
  • 32. Ensslin, T. A., Biermann, P. L., Klein, U., & Kohle, S. A&A, 332, 395 (1998).
  • 33. Giacintucci, S., et al. A&A, 486, 347 (2008).
  • 34. Brunetti, G., et al. Nature, 455, 944 (2008).
  • 35. Finoguenov, A., Sarazin, C. L., Nakazawa, K., Wik, D. R., & Clarke, T. E. ApJ, 715, 1143 (2010).
  • 36. Mazzotta, P., et al. in preparation (2010).
  • 37. Kang, H., & Jones, T. W. JKAS, 35, 159 (2002).
  • 38. Brunetti, G., Setti, G., Feretti, L., & Giovannini, G. MNRAS, 320, 365 (2001).
  • 39. Kang, H., & Jones, T. W. ApJ, 620, 44 (2005).
  • 40. Kassim, N. E., Clarke, T. E., Enßlin, T. A., Cohen, A. S., & Neumann, D. M. ApJ, 559, 785 (2001).
  • 41. Bacchi, M., Feretti, L., Giovannini, G., & Govoni, F. A&A, 400, 465 (2003).
  • 42. Liang, H., Hunstead, R. W., Birkinshaw, M., & Andreani, P. ApJ, 544, 686 (2000).
  • 43. Govoni, F., Feretti, L., Giovannini, G., Böhringer, H., Reiprich, T. H., & Murgia, M. A&A, 376, 803 (2001).
  • 44. Venturi, T., et al. in preparation (2010).
  • 45. Brown, S., & Rudnick, L. MNRAS, in press (2010); arXiv:1009.4258
  • 46. Brunetti, G., & Lazarian, A. MNRAS, 378, 245 (2007).
  • 47. Miniati, F., Jones, T. W., Kang, H., & Ryu, D. ApJ, 562, 233 (2001).
  • 48. Enßlin, T. A. & Gopal-Krishna A&A, 366, 26 (2001).
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description