Understanding the Physics and Observational Effects of Collisions between Galaxy Clusters Daniel R. Wik1 Advisor: Craig L. Sarazin1 1 Department of Astronomy, University of Virginia, Charlottesville, VA, 22904 [email protected] ABSTRACT Galaxy clusters, the largest gravitationally bound objects in the universe, form over time from collisions and mergers between clusters and between clusters and smaller galaxy groups. The results of these collisions can be detected in a variety of ways. In one case, I have found that clusters which have recently undergone a merger will have an enhanced Sunyaev-Zel’dovich (SZ) Effect. The SZ signal from thousands of clusters will soon be measured in order to constrain the dark matter and dark energy content of the universe. However, I have shown that while observations of the central SZ effect, which is more sensitive to merger enhancements, will hopelessly bias these estimates, the integrated SZ signal will not and should be the quantity extracted from SZ surveys. Another result of cluster mergers is the acceleration of relativistic particles, which produce non-thermal emission that should be observable in X-rays. In the Coma cluster, the brightest merging cluster in the sky, detections of this emission have been claimed, though these findings are controversial. Using observations of Coma with the Hard X-ray Detector onboard Suzaku, I find no conclusive non-thermal emission and derive an upper limit on the emission that excludes the most recent previous detection. Introduction Over the past decade, the history, age, and likely future evolution of the universe have been reliably uncovered for the first time, prompting some to call our time the golden age of cosmology. While past advances have come from observations of the afterglow of the Big Bang and of supernovae in distant galaxies, it will soon be possible to study the largest scales of the cosmos with the largest and most massive objects in the universe, clusters of galaxies. These tightly bound collections of thousands of galaxies grow by pulling in galaxies, groups of galaxies, and even other clusters, which they attract with their immense gravity. When 2 galaxy clusters merge, they release enormous amounts of energy, up to 1057 Joules, making cluster mergers the most energetic events since the Big Bang itself. Part of this energy is used to heat up intracluster gas – the atmospheres of clusters filling up the space in between the galaxies – through shock waves and turbulence. In addition to heating up the gas, shocks and turbulence act to accelerate particles traveling near the speed of Wik light, which can emit light in the form of radio waves and X-rays. The existence of this light indicates that a cluster has recently or is currently undergoing a merger, an often difficult conclusion to draw from the optical light alone, and it can tell us about the dynamics and history of the merger, along with the specific physics necessary to produce that light. As well as studying individual clusters undergoing mergers, we can apply what we learn from these observations to understand how mergers affect large samples of galaxy clusters. Over the history of the universe, the nearby galaxy clusters we see now formed from many merger events, so if we observed one of these clusters in the past, it would hold fewer galaxies, smaller amounts of colder gas, and would overall contain less mass. The number of clusters at a given mass changes over time, and how the number evolves can tell us how much “stuff” is in the universe, namely the total amount of mass and dark energy – the mysterious substance that appears to be accelerating the expansion of the universe. In a cluster that 1 has reached equilibrium, the intracluster gas is in balance with the cluster’s gravity, so the energy in the particles of this gas directly corresponds to the mass of the cluster, which is the source of its gravity. This gas is very energetic and hot because the gravity in clusters is so strong, so we can observe its emission at X-rays energies. Also, the hot intracluster gas will distort the Cosmic Microwave Background (CMB), the relic radiation left over from the Big Bang, via the Sunyaev-Zel’dovich (SZ) effect. The amount of change in the CMB spectrum is also directly related to the energy in the cluster gas, and so can be used to measure the total mass of the cluster. From measurements of the X-ray temperature or SZ signal of many galaxy clusters, we can use their inferred masses to constrain the dark matter and dark energy content of the universe. However, the mass you derive is only valid if the cluster gas is in equilibrium and is energetically the dominant component. During violent cluster mergers, the gas is not in equilibrium but is transiently heated, which temporarily “boosts” the SZ signal. If a cluster is observed while it is undergoing a boost, the inferred mass will be incorrect. Also, the mass that is inferred assumes nearly all the energy in the cluster is in the thermal, hot gas particles. While this is a reasonable assumption, it is known that clusters contain nonthermal particles as well, in the form of relativistic electrons traveling near the speed of light. The precise amount of energy in these particles and another relativistic quantity, the magnetic field, has so far yet to be decisively measured, though both are required to explain the presence cluster-wide structures seen at radio frequencies, called radio halos and radio relics, that happen to be found only in merging clusters. The existence of this emission directly implies the presence of Inverse Compton (IC) emission at high X-ray energies, which if observed would uniquely determine the energetic contribution of relativistic electrons and magnetic fields to the intracluster gas. Thus, if clusters are to be used as cosmological probes, the physics taking place during cluster mergers must be well understood. Cluster Mergers and the SZ Effect To assess the effect of mergers on measurements of the SZ effect in galaxy clusters, we used a small suite of binary cluster merger simulations from Ricker & Sarazin (2001). These simulations include mergers between clusters with various mass Wik Fig. 1.— Images of the SZ parameter y from 3D snapshots of the 1:3 mass ratio, 2 rs impact parameter merger simulation. The image are viewed from a line-of-sight which is rotated 45◦ from the merger axis and 45◦ azimuthally from the merger plane 386 Myr before first core crossing. Fig. 2.— Same as Figure 1, but at 114 Myr after first core crossing. ratios (M< : M> = 1 : 1, 1 : 3, 1 : 6.5) and impact parameters (head-on to glancing impacts), totaling 8 unique simulations. From these we calculated the SZ observable y at each time step for 2 which data was saved. Sample images of y from the 1:3 mass ratio merger are provided in Figures 1 and 2. In practice, only one characteristic value for the SZ signal will be extracted for an individual cluster in a large sample. Generally in the literature, this value is taken to be either the central (or equivalently the maximum) value of y, ymax , or y is integrated over the entire cluster, usually denoted by Y . Though it is not immediately clear from the figures, note that the scale of Figure 2 is twice that of Figure 1, indicating that after the cluster centers have collided for the first time, the value of y increases or experiences a “boost.” These boosts, which appear in both Y and ymax , show up as a peak when the SZ parameter is plotted as a funciton of time. We present the evolution of Y and ymax over time during a merger event for the 1:1 mass ratio case in Figures 3 and 4, respectively. In Figure 3, the peak is always less than a factor of two of the initial value of Y , and appears to be less than the final value of Y , once the clusters have merged completely and have reached equilibrium. This is an excellent feature of Y regarding its use as a mass estimate for clusters, as it means that if a cluster is observed during a merger event, the inferred mass will not be overestimated since Y ∝ M 2 . In fact, catching the merger at its peak would give a more accurate mass estimate than observing it 1 Gyr later, when the mass would be underestimated slightly. However, for ymax the transient boost is much larger, though shorter lived. The ymax parameter is boosted by nearly a factor of 10 (see Figure 4), and though it is less likely a cluster will be observed during this shorter boost phase, if it is observed the inferred mass will be significantly larger than its actual mass. From these results, we need to determine whether enough clusters in a survey are affected by these transient boosts to bias the cosmological studies that typically would not take merger events into account. Because of the computational power required in simulating a large portion of the universe over billions of years, we instead use a semi-analytic method to approximate the growth of galaxy clusters, called the extended PressSchechter formalism (Press & Schechter 1974; Lacey & Cole 1993). This method allows us to follow the merger history of a present-day cluster back in time, so we know when mergers happen and what the mass ratio of the merger was. The only other parameter we need is the impact parameter, which can be taken from the probability Wik Fig. 3.— Evolution of Y during the 1:1 mass ratio mergers. The different curves are for different values of the merger impact parameter: b = 0 (solid, black line), 2 rs (dashed,red line), and 5 rs (dotted, blue line). The b = 2rs simulation run is offset downward by 0.15 and the 5rs run is offset downward by 0.3 for clarity. Fig. 4.— Same as Figure 3, but for ymax and with no artifical offset between the curves. distribution of likely values of b (Sarazin 2002; Bullock et al. 2001). By characterizing the nature of the boosts as a function of mass ratio and impact parameter, the observed value of Y and ymax can be found for clusters today or at any time in the past. This is the type of information surveys will get, so we can now find whether the surveys would be biased by neglecting merger effects since 3 we know the true masses of the clusters along with the inferred masses derived from either Y or ymax . A qualitative understanding of our findings can be inferred directly from Figures 5 and 6. In these figures, the value of Y or ymax is shown as a function of mass for 10,000 present-day clusters. The majority of clusters are not currently undergoing a boost, and so they lie along the equilibrium SZ– M relation, which shows up as a solid line. In the case of Y , almost all of the clusters lie along this relation, and only a few lie below it, indicating that the mass would be underestimated as Figure 3 suggested. In fact, the effect of merger boosts on Y is so slight that cosmological parameters such as the amount of mass in the universe and the nature of dark energy are determined to within ∼ 1%. This accuracy is well-within the typical uncertainties of the surveys themselves, making mergers an unimportant factor if Y is used to estimate cluster masses. In the case of ymax , on the other hand, there is a large number of clusters that are boosted from the equilibrium relation, as illustrated in Figure 6. The parameter ymax is preferentially boosted to larger values, which leads to a larger implied mass for those clusters. Even though the boost period for a given cluster is short, making it more difficult to catch a cluster during one of these periods, it is still long and strong enough to affect quite a few clusters. The implied masses lead to significantly biased cosmological parameter estimates, with the total mass of the universe being underestimated by 15-45% and the value of the dark energy equation of state parameter w to be similarly misestimated. In conclusion, it is clear that when these surveys get going in earnest, the SZ parameter that best returns the cluster mass, and that should be used to derive cosmological parameters, is the integrated parameter Y . Fig. 5.— Integrated Comptonization parameter Y versus total mass. The apparent solid line is the result of many individual clusters at or near their equilibrium values of Y . Fig. 6.— Same as Figure 5, but for ymax . The Search for Non-thermal Emission in the Coma Cluster While cluster mergers themselves can confuse cosmological studies that depend on the galaxy cluster mass distribution, mergers can also be exploited to census another component coincident with the gas – relativistic particles and magnetic fields. If these relativistic components contain a significant amount of energy relative to the thermal gas, then they would modify the equilibrium state of the gas upon which conversions between observables, such as Y , ymax , and X-ray tempera- Wik ture, and cluster mass are based. It is thought that relativistic particles are accelerated across shock fronts and turbulence induced by mergers (Sarazin 1999, 2002), which may explain why radio halos and relics are observed exclusively in clusters currently or recently undergoing a merger event (Feretti & Giovannini 2007). The radio emission is produced by relativistic electrons spiralling around magnetic field lines, so we know both of these relativistic components exist. However, the strength of this emission depends on both the number of electrons and the strength of the 4 magnetic field, so from the radio data alone these components cannot be disentangled. Certain assumptions, arrived at through equipartition arguments, lead to estimates of the number of relativistic particles and magnetic field strengths that do not contribute significant amounts of energy relative to the thermal gas. This would not remain true if the value of the average magnetic field in clusters was about 10 times larger than its equipartition value, as suggested by Faraday RM measurements from which the magnetic field can be derived along particular lines of sight. The relativistic electrons that produce the radio halos and relics also produce IC emission through interactions with the CMB, and this emission should be observable at hard X-ray energies. If emission is seen at both radio and X-ray frequencies, then the magnetic field can be directly determined. However, the IC X-ray emission has been difficult to observe with past instruments, generally leading to weak or non-detections (Nevalainen et al. 2004). The most recent observatory with hard X-ray capabilities, the Suzaku X-ray Observatory, should be ∼ 3× more sensitive where IC emission is expected to be detected. We report here an observation with the Suzaku Hard X-ray Detector (HXD) of the Coma cluster, which hosts the brightest radio halo in the sky. Coma was observed for over 160,000 seconds, or for about 2 days. The greatest difficulty in detecting IC emission from clusters like Coma is that they tend to be hot because they are so massive – massive clusters undergo more mergers typically and are thus more likely to host radio halos or relics. The hotter the gas in a cluster, the more thermal emission there is at higher energies, which is where we are looking for the IC emission. So, in order to confidently detect an IC signal, the thermal emission must be well characterized. To accomplish this goal, we extract complimentary data at lower energies, where the thermal emission is completely dominant, from a mosaic observation of Coma with the XMMNewton Observatory (Schuecker et al. 2004). For this data, we spatially weight the X-rays according to the spatial sensitivity of the HXD (since it is a non-imaging instrument while XMM is) to create a spectrum that exactly corresponds to the spectrum obtained from the Suzaku HXD observation. With these two spectra, we now have a long enough lever arm to accurately characterize the thermal emission and see whether there is evidence for IC emission as well. Our best model fit for these spectra is given in Figure 7. Wik Considering only statistical errors, we detect a non-thermal component to the total emission from Coma at the 3.4σ level. When systematic errors are also included, however, this non-thermal signal is no longer robustly detected. The main uncertainty lies in the precise determination of the non-X-ray background in the instrument, which is known to an accuracy of 3%. Since the flux of the IC emission is very similar to 3% of the background flux, if the background is actually 3% higher than we have assumed, then the nonthermal component is not needed. We determine an upper limit for IC emission instead, by setting quantities with systematic errors to the limit that would favor an IC detection. For the background level, we set it to be 3% lower than originally set, for example. We find an upper limit on the IC emission of 7.8 × 10−15 W/m2 . This value is 2× lower than the most recent detections of IC emission in Coma, with the RXTE observatory (Rephaeli & Gruber 2002) and the BeppoSAX observatory (Fusco-Femiano et al. 2004). This result is not entirely surprising, especially given that the BeppoSAX detection was controversial. Rossetti & Molendi (2004) analyzed the same data as Fusco-Femiano et al. (2004) and due to a different background determination, they only found an upper limit consistent with our result. The lack of a non-thermal component is also bolstered by a more detailed analysis of the lower energy X-ray data from XMM. In reality, the thermal component in Coma consists of a multiple temperature gas, and even a small amount of higher-than-average temperature gas can mimic a non-thermal signal at high energies. We divided the XMM data into spatial regions and found the temperature of the gas in each region, and then we created a multi-temperature model which was extrapolated to higher energies. In other words, if all the emission is due to thermal sources, then the fits we derive at low energies should also work at higher energies. If not, and there is an excess of emission at higher energies, then the existence of IC emission is supported. We found that our multi-temperature model more successfully fit the data than the model that included a non-thermal component. This result is consistent with recent findings from the INTEGRAL observatory that also detected hard X-ray photons from Coma that was entirely consistent with only thermal emission (Renaud et al. 2006; Eckert et al. 2007). 5 Fig. 7.— The photon detection rate as a function of energy in kilo-electron volts. The black data points are from XMM and the red points are from the Suzaku HXD. The bottom panel reports the residuals compared to the best-fit model, relative to the size of the error on the data points. The various histograms represent the contribution of thermal emission (the line almost overlapping with the data, deviating more at higher energies), the Cosmic X-ray background (line only in red), the Inverse Compton emission (line second from the bottom, close to the CXB line) and emission from point sources (line closest to bottom). Wik 6 Future Work Upper limits on the flux of IC emission in clusters with radio halos or relics provide a lower limit on the magnetic field strength, which sets a lower limit on the amount of energy in relativistic components in galaxy clusters. Even if no IC emission is detected, the derived lower limit could still indicate that a non-negligible amount of energy resides in relativistic components, which would modify observable quantity-mass relations necessary to use clusters to study cosmology. Because cluster surveys, both in X-rays and in the SZ effect, are achieving senstitivity to cosmological parameters, it is critical that merger effects are allowed for and that we fully understand the energetics of the intracluster gas. To that end, I plan to repeat my analysis of Suzaku HXD data on the Coma cluster for all the clusters so far observed with Suzaku, in order to produce a uniform analysis of their hard X-ray emission. This work will allow a better determination of the relativistic energy contribution in clusters, which will address the usefulness of current observable quantity-mass relations as well as probe the physical processes that occur during cluster mergers. Sarazin, C. L. 2002, The Physics of Cluster Mergers (ASSL Vol. 272: Merging Processes in Galaxy Clusters), 1–38 Schuecker, P., Finoguenov, A., Miniati, F., Böhringer, H., & Briel, U. G. 2004, A&A, 426, 387 REFERENCES Bullock, J. S., Dekel, A., Kolatt, T. S., et al. 2001, ApJ, 555, 240 Eckert, D., Neronov, A., Courvoisier, T. J.-L., & Produit, N. 2007, A&A, 470, 835 Feretti, L. & Giovannini, G. 2007, ArXiv Astrophysics e-prints Fusco-Femiano, R., Orlandini, M., Brunetti, G., et al. 2004, ApJ, 602, L73 Lacey, C. & Cole, S. 1993, MNRAS, 262, 627 Nevalainen, J., Oosterbroek, T., Bonamente, M., & Colafrancesco, S. 2004, ApJ, 608, 166 Press, W. H. & Schechter, P. 1974, ApJ, 187, 425 Renaud, M., Bélanger, G., Paul, J., Lebrun, F., & Terrier, R. 2006, A&A, 453, L5 Rephaeli, Y. & Gruber, D. 2002, ApJ, 579, 587 Ricker, P. M. & Sarazin, C. L. 2001, ApJ, 561, 621, (RS) Rossetti, M. & Molendi, S. 2004, A&A, 414, L41 Sarazin, C. L. 1999, ApJ, 520, 529 Wik This 2-column preprint was prepared with the AAS LATEX macros v5.2. 7
© Copyright 2026 Paperzz