The velocity structure of the intracluster medium of the Centaurus cluster

ABSTRACT

There are few direct measurements of the intracluster medium (ICM) velocity structure, despite its importance for understanding clusters. We present a detailed analysis of the velocity structure of the Centaurus cluster using XMM–Newton observations. Using a new European Photon Imaging Camera-pn energy scale calibration, which uses the Cu Kα instrumental line as reference, we are able to obtain velocity measurements with uncertainties down to Δ|$\mathit{ v}$| ∼ 79 km s⁻¹. We create 2D spectral maps for the velocity, metallicity, temperature, density, entropy, and pressure with a spatial resolution of 0.25 arcmin. We have found that the velocity structure of the ICM is similar to the velocity structure of the main galaxies, while the cold fronts are likely moving in a plane perpendicular to our line of sight with low velocity. Finally, we have found a contribution from the kinetic component of |$\lt 25{{\ \rm per\ cent}}$| to the total energetic budget for a radius >30 kpc.

1 INTRODUCTION

Theoretical simulations predict that the intracluster medium (ICM) should contain turbulent, or random, motions and bulk flows caused by the merger of other clusters and subcomponents (Lau, Kravtsov & Nagai 2009; Vazza et al. 2011; Schmidt et al. 2017; Ha, Ryu & Kang 2018; Li, Zhu & Zhao 2018; Vazza et al. 2021). In addition, there can be relative bulk motions of ∼500 km s⁻¹ due to sloshing of the ICM in the potential well, generated by merging substructures (Ascasibar & Markevitch 2006; Ang Liu et al. 2015, 2016; Vazza et al. 2018; ZuHone et al. 2018; Ichinohe et al. 2019). The action of the relativistic jets and inflation of bubbles by the central active galactic nucleus (AGN) also likely generate motions of a few hundred kilometres per second (Brüggen, Ruszkowski & Hallman 2005; Heinz, Brüggen & Morsony 2010; Randall et al. 2015; Yang & Reynolds 2016; Bambic & Reynolds 2019). This is also important for several other reasons. Turbulent motions affect calculations of hydrostatic equilibrium and cluster mass estimates given that they provide additional pressure support, particularly at large radii (e.g. Lau et al. 2009). Measuring the velocities would help to constrain AGN feedback models because the distribution of energy within the bulk of the cluster depends on the balance between turbulence and shocks or sound waves (see Fabian 2012 for a review). Velocity is an excellent probe of the microphysics of the ICM, such as viscosity, which acts to smoothen the velocity structure. Simulations predict that there is a close connection between the velocity power spectra and the overall ICM physical state (which is identified by its temperature, density, pressure, and entropy; see e.g. Gaspari et al. 2014). Motions will also cause transport of metals within the ICM, due to uplift and sloshing of metals by AGNs (e.g. Simionescu et al. 2008; Werner et al. 2010). In addition, measuring velocities should directly measure the sloshing of gas in cold fronts, which can remain for several Gyr (Roediger et al. 2012, 2013; Walker et al. 2018).

Despite its importance, the velocity structure of the ICM remains poorly observationally constrained. Suzaku observations were used to obtain velocities in several systems by measuring the Fe–K line. Tamura et al. (2011, 2014) placed upper limits on relative velocities of 300 km s⁻¹ over scales of 400 kpc in Perseus. Ota & Yoshida (2016) examined several clusters with Suzaku, although systematic errors from the Suzaku calibration were likely around 300 km s⁻¹ and its point spread function (PSF) was large. Other direct methods of measuring velocities include using XMM–Newton Reflection Grating Spectrometer (RGS) spectra to measure line broadening and resonant scattering, showing low-turbulence motion with velocities between 100 and 300 km s⁻¹ (Sanders & Fabian 2013; Pinto et al. 2015; Ogorzalek et al. 2017), but limited to the cluster core. Finally, indirect measurements of the velocity structure in clusters include looking at the power spectrum of density fluctuations and linking this via simulations to the velocity spectrum (e.g. Zhuravleva et al. 2014) or examining the magnitude of thermodynamic perturbations (Hofmann et al. 2016). However, these methods are model dependent.

Random and bulk motions in the ICM were directly measured by the microcalorimeter Soft X-ray Spectrometer (SXS) X-ray detector onboard the Hitomi observatory using the Fe–K emission lines. In the core of the Perseus cluster, it measured a gradient of 150 km s⁻¹ bulk flow across 60 kpc of the cluster core and a line-of-sight velocity dispersion of 164 ± 10 km s⁻¹ between radii of 30 and 60 kpc (Hitomi Collaboration 2016). They indicate that the measured level of turbulence may be sufficient to offset radiative cooling if driven on scales comparable with the size of the largest bubbles in the field (∼20–30 kpc). On the other hand, the low level of turbulence in the system indicates that it might not be fast enough to replenish heating in the short time necessary to balance cooling, although shocks or sound waves could plausibly do this (Fabian et al. 2017). Gravity waves, which could replenish turbulence, do not propagate efficiently radially (Fabian et al. 2017). Simulations analysed by Bambic, Morsony & Reynolds (2018) indicated that in the presence of large-scale magnetic fields the resulting production of turbulence is inefficient, even when able to preserve AGN-driven bubbles. In their analysis of XMM–Newton observations of the cool core clusters RXCJ1504.1−0248 and Abell 1664 done by Liu et al. (2021), they found that the turbulent energy density is less than 9 and 27 per cent, respectively, thus insufficient for AGN heating to propagate to the cool core via turbulence. Unfortunately, due to the loss of Hitomi we will not be able to make further measurements in other clusters or in different regions of Perseus. The next planned observatories with instruments capable of measuring velocities are likely the X-ray Imaging and Spectroscopy Mission (XRISM Science Team 2020), to be launched in 2023, and athena (Barret et al. 2018), in 2030.

Sanders et al. (2020) present a novel technique to improve velocity measurements calibrating the absolute energy scale of the detector to better than 150 km s⁻¹ at Fe–K by using background X-ray lines seen in the spectra of the XMM–Newton EPIC-pn detector. Using this technique, Sanders et al. (2020) mapped the bulk velocity distribution of the ICM over a large fraction of the central region of two nearby clusters of galaxies, Coma and Perseus. For the Coma cluster, they found that the velocity of the gas is close to the optical velocities of the two central galaxies, NGC 4874 and NGC 4889. In the case of the Perseus cluster, they detect evidence for sloshing associated with a cold front. Following the same methodology, Gatuzz et al. (2022) analysed the velocity structure of the Virgo cluster. They identified a complex high-velocity structure, most likely indicating the presence of an outflow near the AGN. Moreover, they found that the hot gas located within the radio flows moves in opposite directions (i.e. blueshifted and redshifted) possibly driven by the radio jets or bubbles and with velocities of ∼331 ∼ −258 km s⁻¹, with respect to M87.

The Centaurus cluster is an excellent target to apply such a technique due to it being one of the X-ray brightest in the sky, its proximity allowing measurements on small scales (z = 0.0104; Lucey, Currie & Dickens 1986), its high metallicity giving a strong Fe line, and it being a system with a merger and AGN feedback. Using optical observations, Lucey et al. (1986) identified two subgroups in the Centaurus cluster, the main one centred on the galaxy NGC 4696 (referred to as Cen 30) and a second one centred on the galaxy NGC 4709 (referred to as Cen 45) 5 arcmin to the east with a line of sight (see Fig. 1). The heliocentric velocities of the main galaxies are as follows: NGC 4696, 2958 ± 15 km s⁻¹ (de Vaucouleurs et al. 1991); NGC 4709, 4678 ± 4 km s⁻¹ (Smith et al. 2000); and NGC 4696B, 3111 ± 7 km s⁻¹ (Smith et al. 2000). Walker, Fabian & Sanders (2013) found evidence for a shock-heated region between the two systems, which could be explained by a simple shock-heating model. Despite this, Ota & Yoshida (2016) found that the velocity of Cen 45 was consistent with the main cluster velocity, with a velocity difference of <750 km s⁻¹ (90 per cent limit). However, Suzaku had a large PSF and the spatial regions examined were large. The cluster contains three or more cold fronts, likely caused by the ICM sloshing in the potential well. These are seen as discontinuities in surface brightness, temperature, and metallicity (Sanders et al. 2016). Centaurus has probably the clearest cold front morphology seen in any nearby galaxy cluster. There is a possible large-scale Kelvin–Helmholtz instability in the core (Walker et al. 2017). Moreover, by analysing Chandra observations, Sanders et al. (2016) have shown that the central AGN appears to have been repeatedly active over long time-scales with periods of tens of Myr. This is seen in edge-filtered Chandra X-ray images, which show multiple cavity-like structures and possible sound waves. There is also extended low-frequency radio emission associated with some of these cavities. Finally, in the core of the cluster there are also two weak shocks, surrounding the nucleus and inner cavities. Given all these highly energetic processes, we expect to find high velocities and, perhaps, velocity structure.

We present an analysis of the ICM velocity structure in the Centaurus galaxy cluster using XMM–Newton observations. The outline of this paper is as follows. We describe the data reduction process in Section 2. The fitting procedure is shown in Section 3, while a discussion of the results is included in Section 4. Finally, Section 5 presents the conclusions and summary. Throughout this paper, we assume a distance of Centaurus of z = 0.0104 (Lucey et al. 1986) and a concordance Lambda cold dark matter (ΛCDM) cosmology with Ω_m = 0.3, |$\Omega _\Lambda = 0.7$|⁠, and H₀ = 70 km s⁻¹ Mpc⁻¹.

2 DATA REDUCTION

Information on the Centaurus cluster observations analysed in this paper is provided in Table 1, including IDs, coordinates, dates, and clean exposure times. We reduced the XMM–Newton European Photon Imaging Camera (EPIC; Strüder et al. 2001) spectra with the Science Analysis System (sas,¹ version 18.0.0). Each observation was processed with the epchainsas tool, including only single-pixel events (PATTERN==0) and filtering the data with FLAG==0 to avoid regions close to CCD edges or bad pixels. Bad time intervals were filtered from flares applying a 1.0 cts s⁻¹ rate threshold.

To obtain velocity measurements down to 150 km s⁻¹ at Fe–K, we have applied the technique described in Sanders et al. (2020) to calibrate the absolute energy scale of the EPIC-pn detector using the background X-ray lines identified in the spectra of the detector. This calibration method includes corrections (1) to the average gain of the detector during the observation, (2) to the spatial gain variation across the detector with time, and (3) to the energy scale as a function of detector position and time. In particular, we used the calibration files computed by Gatuzz et al. (2022). However, we do not expect changes with new caldb as observations are older than 2020.

Fig. 2 shows the X-ray image obtained in the 0.5–9.25 keV energy band (left-hand panel). We have identified point sources using the sas task edetect_chain, with a likelihood parameter det_ml >10. In the following analysis, we have excluded the point sources. Fig. 2, top right-hand panel, shows the number of counts in each 1.59 arcsec pixel in the Fe–K complex (6.50–6.90 keV, in rest frame), after subtracting neighbouring scaled continuum images (6.06–6.43 and 6.94–7.16 keV, in rest frame). A Gaussian smoothing of σ = 4 pixels was applied for illustrative purposes. The image shows a large number of counts in multiple directions around the cluster centre, due to the spatial offset in the observations. Fig. 2, bottom right-hand panel, shows the total exposure time in the 4.0–9.25 keV energy range. Fig. 3 shows the fractional difference the 0.5–9.25 keV surface brightness from the average at each radius. The central region of the cluster exhibits a complex structure, no clear centre, and a very faint point source at the location of the AGN (Sanders et al. 2016), thus preventing us from performing a velocity analysis using RGS spectra.

3 DATA ANALYSIS

We included a tbabs component (Wilms, Allen & McCray 2000) to account for the Galactic X-ray absorption. For each spatial region analysed, we combined the spectra from different observations together and rebinned the spectra to have at least one count per channel. Then, we loaded the data twice in order to fit separately, but simultaneously the 1.5–4.0 and hard 4.0–10 keV energy bands. The free parameters in the model are the metallicity, log(σ) of the temperature distribution, temperature, and normalization. The redshift is a free parameter only for the 4.0–10 keV energy band given that for lower energies the new EPIC-pn energy calibration scale cannot be applied. We note that the inclusion of the lower energy band data, for which the redshift parameter is fixed to the cluster value, leads to a better constraint for the temperatures and metallicites. We fixed the column density to 8.10 × 10²⁰ cm⁻², although it is important to note that in the energy range analysed the absorbing component has a weak effect in the modelling. For the data analysis, we used the xspec spectral fitting package (version 12.10.1²). We assumed Cash statistics (Cash 1979). Errors are quoted at 1σ confidence level unless otherwise stated. Finally, abundances are given relative to Lodders & Palme (2009). As background components, we have included Cu Kα, Cu Kβ, Ni Kα, and Zn Kα instrumental emission lines, and a power-law component with its photon index fixed at 0.136 (the average value obtained from the archival observations analysed in Sanders et al. 2020).

3.1 Spectral maps

We created a velocity map of the cluster following the method shown in Sanders et al. (2020) and Gatuzz et al. (2022). We created elliptical regions with a 2:1 axial ratio, rotating them such that the longest axis lay tangentially to a vector pointing to the central core. In order to reduce uncertainties on the velocities, the radii of the ellipses changed adaptively to have ∼500 counts in the Fe–K complex after continuum subtraction. We moved in a grid with a spacing of 0.25 arcmin. For each region, we extracted the spectra for all observations and performed a combined fit. Then, we created weighted-average ancillary response files (arf) for each ellipse from the individual ones for each observation but used the same response matrix for all the pixels.

Fig. 5 shows the velocity maps obtained. In order to improve the clarity, we use a red–blue colour scale to indicate whether the gas is moving away from (red) or towards (blue) the observer. The map shows the complexity of the velocity structure in the system. Interestingly, there is no clear spiral pattern associated with gas sloshing (see e.g. fig. 18 in Sanders et al. 2020) or an opposite blueshifted–redshifted structure around the cluster centre (see e.g. fig. 11 in Gatuzz et al. 2022). Such features could point out that our line of sight is perpendicular to the sloshing plane of the system (see e.g. ZuHone et al. 2016). There is a clear blueshifted structure along the south-west direction from the cluster centre with velocities larger than 1000 km s⁻¹, which could be due to the impact of AGN outflows from the NGC 4696 system. This could be a hint of azimuthal structure in the velocity. However, such structure corresponds to a region with large velocity uncertainties, which could also be a feature of gas sloshing (see e.g. fig. 19 in Gatuzz et al. 2022). The velocity spectral map shows a large blueshifted region southward from the Galactic cluster core. Such velocity increase could be associated with gas-sloshing signatures (see figs 2 and 3 in ZuHone et al. 2018). The blueshifted gas displays an X shape, which we have not identified in magnetohydrodynamic simulations. It is important to note that the resolution obtained in these spectral maps (∼10 kpc for the smallest ellipse) is larger than the radio observations (see e.g. Taylor et al. 2006), therefore preventing us from performing a direct analysis of the velocity structure within the extended component of the central radio source. Moreover, Taylor et al. (2006) have shown that the radio emission seems initially directed north–south, with hints of interaction with the thermal gas, which leads to an east–west collimation. However, the main velocity field along the structure of the radio source is fairly close to the plane of the sky. Fig. 6 shows the temperature, log (σ), and metallicity spectral maps. Near the Centaurus cluster core, there is a high-metallicity region with low temperature. A similar structure was identified, with higher resolution due to the Chandra instruments, by Sanders et al. (2016; see fig. 3). Outside the central region, the temperature and metallicities change smoothly, with the temperature increasing and the metallicity decreasing as we move to the outskirts of the cluster. We note that the two opposite regions with large velocity uncertainties located NE and SW from the cluster core correspond to regions with large log (σ) and large temperature uncertainties where the Fe–K line is weak and broad.

Using these spectral maps, we calculated the projected pseudo-density, pressure, and entropy in each spatial bin and assuming a constant line-of-sight depth for all spectral regions. We calculated the pseudo-density as |$n\equiv \sqrt{\eta }$|⁠, where η is the normalization of the lognorm model. Then, we estimated pseudo-entropy as S ≡ n^−2/3 × kT and pseudo-pressure as P ≡ n × kT (see Hofmann et al. 2016). Fig. 7 shows the density, pressure, and entropy maps created from the best-fitting results. We noted that the cluster centre displays a high-density gas and low entropy, similar to the maps obtained by Sanders et al. (2016). However, the region with larger density seems to be displaced in the west direction with respect to the cluster centre.

3.2 Fitting spectra concentric rings

We studied the velocity structure by analysing manually extracted regions for two cases:

• case 1: complete concentric rings, square root spaced and with the centre located in the cluster centre.

• case 2: concentric regions divided in east–west zones (i.e. following the cold front location), square root spaced and with the centre located in the cluster centre.

The top panel in Fig. 8 shows the exact regions analysed for case 1. Table 2 shows the best-fitting results obtained per region. Figs 9 and 10 show the velocities, temperatures, log (σ), and metallicities obtained from the best fit per region. We have obtained velocities measurements with uncertainties down to Δv ∼ 89 km s⁻¹ (for ring 5). The largest blueshift/redshift, with respect to the main cluster redshift, corresponds to −254 ± 114 and 295 ± 197 km s⁻¹ for rings 2 and 14, respectively. Both the metallicity and temperature profiles display a discontinuity at ∼50 and ∼100 kpc. We noted a complex structure in the temperature profile below such discontinuity (similar to results from Walker et al. 2013). Previous analyses of the Centaurus cluster have shown a drop in metallicity near the cluster centre within <10 kpc (e.g. Panagoulia, Fabian & Sanders 2013; Lakhchaura, Mernier & Werner 2019). However our spectra analysis does not include the soft energy band; thus, it could be overestimated.

The bottom panel in Fig. 8 shows the exact regions analysed for case 2. Figs 11 and 12 show the velocities, temperatures, log (σ), and metallicities obtained from the best fit per region. In the plot, green points correspond to results obtained with the E direction, while blue points correspond to the W direction. We have obtained velocity measurements with uncertainties down to Δv ∼ 123 km s⁻¹ (for ring 7, east direction). The largest blueshift/redshift, with respect to the main cluster redshift, correspond to −434 ± 265 km s⁻¹ for ring 1 (east direction) and 457 ± 232 km s⁻¹ for ring 15 (west direction). Both the metallicity and temperature profiles display a discontinuity at ∼30 kpc. We noted a difference in the metallicity profiles for the outer region (i.e. >50 kpc), with larger values in the east direction compared with those in the west direction.

3.3 Cluster velocity substructures

We extracted spectra for spatial regions following the X-ray surface brightness and having roughly an equal number of counts in the Fe–K complex (∼2000). Fig. 13, top panel, shows the specific regions analysed. The best-fitting results are listed in Table 3. Fig. 13, bottom panel, shows the complex distribution of velocities obtained. We have obtained velocity measurements with uncertainties down to Δv ∼ 120 km s⁻¹ (for region 12). The largest blueshift/redshift, with respect to the Centaurus cluster, corresponds to −223 ± 250 and |$1760^{+514}_{-564}$| km s⁻¹ for regions 3 and 17, respectively. In the following, we compare the ICM X-ray velocities obtained with the optical spectroscopic redshifts measured for the individual galaxies NGC 4696, NGC 4709, and NGC 4696B.

We have found that for regions 1, 2, and 3, close to the main system NGC 4696, there is an excellent agreement between the obtained redshift and the Centaurus cluster redshift. For these points, we obtained an average redshift of 0.0102 ± 0.0007. That is ∼−61 km s⁻¹ with respect to the Centaurus cluster. Other regions surrounding the Cen 30 system display similar velocities to that of the system (indicated in blue colour in Fig. 13, top panel), except for region 7 (261 ± 200 km s⁻¹). Fig. 14 shows the temperatures, log (σ), and metallicity profile obtained. The overall metallicity and temperature distribution is in good agreement with the spectral map shown in Fig. 6. Regarding the metallicity and temperature, we have found that regions 4 and 5, located south-west of the cluster core, have lower metallicities and larger temperatures than those located north-east of the cluster core (e.g. regions 6 and 7). The largest metallicity and lowest temperature correspond to the cluster core (i.e. regions 1 and 2).

We note that region 22, containing NGC 4709 (i.e. the main galaxy of Cen 45), displays a velocity of 216 ± 258 km s⁻¹ (with respect to the Centaurus cluster), within the velocity of the Cen 30 system, but significantly smaller than the Cen 45 system velocity (∼1500 km s⁻¹ above Cen 30). Ota & Yoshida (2016) found an upper limit for the velocity around NGC 4709 of <750 km s⁻¹. They suggested that such results indicate that the Centaurus cluster has experienced a subcluster merger along the line of sight and gas near NGC 4969. The temperature obtained for that region is in good agreement with the temperature obtained by Walker et al. (2013), which could large be due to heating of gas near NGC 4696 by the strong shock along the merger process. On the other hand, for the region beyond NGC 4709 the velocity is 696 ± 310 km s⁻¹ (region 23). For region 17, we have found a velocity of |$1759^{+565}_{-515}$| km s⁻¹, in good agreement with the NGC 4709 system velocity. As suggested by Ota & Yoshida (2016), such offset between the galaxy distribution and the mass centroids of the ICM along the sightline could be a fingerprint of a previous subcluster merger. For region 15, which contains the NGC 4696B, we have found a velocity of |$642_{-249}^{+267}$| km s⁻¹, much larger than the system velocity, which is close to the Centaurus cluster velocity. We found that the most external region beyond NGC 4709 (region 23) displays the largest temperature (⁠|$4.48^{+0.11}_{-0.17}$| K) and lowest metallicity (0.47 ± 0.03). However, for regions with high kT and low Z the Fe–K lines become fainter, leading to larger uncertainties.

3.4 The cold fronts

We have created regions to measure the velocity structure in the cold fronts identified in the Chandra observations by Sanders et al. (2016). Fig. 15 shows the exact regions analysed (top panel), while the best-fitting results are listed in Table 4. Fig. 15 shows the velocity obtained for each region (bottom panel). For two cold fronts, there is a hint of larger velocities in the hottest region (i.e. region 2 compared with region 1); however, the uncertainties are too large to provide a good constraint on the velocities. These results suggest that the cold front motion is difficult to observe because the gas is moving in a plane perpendicular to our line of sight. Fig. 16 shows the temperatures, the log (σ), and the metallicities obtained. As expected, there are clear differences in metallicities and temperatures in the interface between the cold fronts. The lack of significant metal mixing in cold fronts is likely due to their properties as a transient wave phenomena (Roediger et al. 2011).

Following the fractional difference in the 0.5–9.25 keV surface brightness, we manually selected multiple regions in the innermost part of the cluster to study the velocity structure. The top panel in Fig. 17 shows the extraction regions analysed. Figs 17 and 18 show the best-fitting results, while Table 5 list the best-fitting parameters. We have obtained velocity measurements with uncertainties down to Δv ∼ 79 km s⁻¹ (for region 5). The average velocity for all regions is 52 ± 171 km s⁻¹. We note that regions in the east direction (i.e. 1 and 2) display lower velocity, larger temperature, and larger metallicity than regions located in the west direction (i.e. regions 5 and 6).

3.5 Energy budget

We estimated the ratio of turbulent to thermal energy by measuring velocities for non-overlapping elliptical regions obtained in the spectral maps for different radii. Fig. 19 shows the ellipse extraction regions analysed, where different radii are indicated by different colours. For each radius, we measured the velocity mean and σ-width, assuming a Gaussian distribution, as a function of the distance to the cluster centre. Fig. 20, top panel, shows that the σ-width decreases as we move away from the inner radius. As a comparison, we included the values obtained for the Virgo cluster by Gatuzz et al. (2022). The σ-width decreases faster for the former, due to the strong influence of the AGN outflows near the cluster centre. However, uncertainties are large in both cases.

We compute the sound speed for each region as |$c_{\mathrm{ s}}=\sqrt{\gamma kT/\mu m_{\mathrm{ p}}}$|⁠, where kT is the best-fitting temperature, γ is the adiabatic index, μ is the mean particle mass, and m_p is the proton mass. Then, we computed the Mach number as a function of the radius, by dividing the bulk velocity by the sound speed and assuming that velocities are isotropic. We have found that for the innermost radius the Mach number is ∼1.12, a value expected for AGN-driven outflow, while for a radius >30 kpc the Mach number is M ∼ 0.17–0.33, a range of values expected for gas sloshing (see middle panel in Fig. 20). The bottom panel in Fig. 20 shows upper limits for the ratio of turbulent (ϵ_turb) to thermal (ϵ_ther) energy computed as |$\epsilon _{\mathrm{ turb}}/\epsilon _{\mathrm{ ther}}=\frac{\gamma }{2}M^{2}$|⁠. For a radius >30 kpc, we found a contribution from the turbulent component |$\lt 25{{\ \rm per\ cent}}$|⁠. It is important to note that we are assuming that motions are isotropic. Simulations computed by ZuHone et al. (2018) have shown that the kinetic energy may be underestimated near the cluster core by ∼10–60 per cent depending on the line of sight, due to the conservative assumption of isotropy.

4 DISCUSSION

Sanders et al. (2016) identified multiple substructures in the Centaurus cluster by analysing Chandra observations, including a 1.9 kpc radius shell around the core, which could be a shock generated by an AGN outburst, small cavities, filaments, and a plume-like structure. Walker, Sanders & Fabian (2015), in their analysis of Chandra observations, found one-component velocities in the range of 100–150 km s⁻¹ on spatial scales of 4–10 kpc. Due to the lower spatial resolution in the XMM–Newton observations, we are not able to study the velocity distribution in such structures. We have found for most of regions velocities similar to the Centaurus cluster velocity, including for regions near the NGC 4709 subsystem, in good agreement with the upper limits obtained in the Suzaku data analysis done by Ota & Yoshida (2016). The velocities measured near the cluster centre are much lower than those found by Dupke & Bregman (2006) in their analysis of Chandra observations. Finally, the temperature and metallicity profiles obtained in our analysis in the >20 kpc scale (see Figs 10 and 12) are in good agreement with previous Chandra and XMM–Newton results (Takahashi et al. 2009; Panagoulia et al. 2013; Walker et al. 2013; Lakhchaura et al. 2019).

Idealized merger simulations indicate that sloshing motions can produce line shifts of the order of 100–200 km s⁻¹ near the cluster core (ZuHone et al. 2016, 2018). Idealized AGN simulations are also capable of producing low line-of-sight velocity dispersions of 150 km s⁻¹ consistent with those observed by Hitomi in the Perseus cluster (Bourne & Sijacki 2017; Ehlert et al. 2021). Velocities obtained from cosmological simulations, on the other hand, vary between 100 and 400 km s⁻¹ (Lau et al. 2017; Roncarelli et al. 2018). In this sense, the theoretical predictions are in good agreement with the velocity measurements.

5 CONCLUSIONS

We have analysed the velocity structure in the Centaurus cluster using the technique developed by Sanders et al. (2020) to calibrate the absolute energy scale of the XMM–Newton EPIC-pn detector. Our results indicate that the ICM is dominated by the low-velocity motions and, possibly, by the impact of the AGN in the NGC 4696 system. In this section, we briefly summarize our findings.

• We have obtained accurate velocity measurements with uncertainties down to Δv ∼ 79 km s⁻¹.

• We have created 2D projected maps for velocity, temperature, log(σ), metallicity, density, entropy, and pressure distribution for the Centaurus cluster. There is no clear indication of the spiral pattern in the velocity map associated with gas sloshing. While there is a hint for blueshifted motion near the cluster centre, most likely due to the influence of AGN outflows, the velocity uncertainties are large.

• We have studied the velocity distribution by creating non-overlapping circular regions. We found that the gas located at <20 kpc from the cluster core displays low-blueshifted velocities (with respect to the Centaurus cluster velocity), while gas located at 20–150 kpc displays a velocity close to the velocity of the cluster. Finally, gas located at large distance (i.e. >150 kpc) displays a redshift behaviour with low velocities (<500 km s⁻¹).

• We have analysed the velocity distribution along the east and west directions by creating non-overlapping circular regions. We have found that while the metallicity distribution shows larger values along the east direction for large distances, together with the lower temperatures in the same direction, the velocity distribution is similar along both directions (including the uncertainties). The velocities are <600 km s⁻¹.

• We have analysed spectra for spatial regions with a similar number of counts to study the substructures within the cluster. We have found that the gas located near the cluster centre displays velocities similar to the velocity from the main system NGC 4696. We have found a region that displays a large redshift similar to the velocity of the system NGC 4709, suggesting an offset between the galaxy and the ICM velocity distribution.

• We have analysed the velocity structure following the surface brightness near the cluster core. We have found that the velocities are similar to the velocity of the Centaurus cluster.

• We have analysed the velocity structure following the three cold fronts located towards the east and west directions from the cluster core. Our measurements indicate that despite lack of significant metal mixing, the velocities are close to the velocity of the NGC 4696 system. In that sense, the cold fronts are likely moving tangentially (i.e. in a plane perpendicular to our line of sight with low velocity).

• By fitting multiple non-overlapping regions, we have found that the width of the velocity distribution decreases as we move away from the cluster centre. We have found a Mach number of M ∼ 0.17–0.33 for a radius >30 kpc from the cluster core, a range of values expected for gas sloshing. We also have found a contribution from the turbulent component of |$\lt 25{{\ \rm per\ cent}}$| to the total energetic budget for a radius >30 kpc.

Future work will include the analysis of the soft band (<1.5 keV).

ACKNOWLEDGEMENTS

This work was supported by the Deutsche Zentrum für Luft- und Raumfahrt (DLR) under the Verbundforschung programme (Kartierung der Baryongeschwindigkeit in Galaxienhaufen). This work is based on observations obtained with XMM–Newton, an European Space Agency (ESA) science mission with instruments and contributions directly funded by ESA Member States and the National Aeronautics and Space Administration (NASA).

DATA AVAILABILITY

The observations analysed in this article are available in theXMM–Newton Science Archive (XSA³).

Footnotes

REFERENCES

APPENDIX A: XMM–NEWTON SPECTRA AND FITTING PROCEDURE

Fig. A1 shows the best-fitting spectra obtained for the inner region analysed in case 1 in Section 3.2, as an example of the EPIC-pn spectra analysed in this work. As described in Section 3, we load the data twice in order to fit separately, but simultaneously the 1.5–4.0 keV and hard 4.0–10 keV energy bands. The instrumental emission lines used as part of the background emission for the energy scale calibration are Ni Kα (7.47 keV), Cu Kα (8.04 keV), Zn Kα (8.63 keV), and Cu Kβ (8.90 keV).

Fig. A2 shows the instrumental Cu Kα redshift variation for each region in the case 1 analysis, after applying the energy calibration scale correction. This plot shows a systematic uncertainty ∼75 km s⁻¹ coming from the calibration uncertainty. This could be compared with the residual calibration uncertainties at the ∼150 km s⁻¹ level found by Sanders et al. (2020), which depend on the CCD spatial location. We also note that the Cu Kα redshift is obtained from the background fit, which also includes the Ni Kα, Zn Kα, and Cu Kβ lines, thus increasing the uncertainty.

Previous analysis of the Centaurus cluster have shown the presence of a multitemperature component near the Centaurus cluster centre (Sanders & Fabian 2008; Panagoulia et al. 2013; Lakhchaura et al. 2019). In that sense, we have tested a two-temperature model [tbabs*(apec+apec)] in our fitting procedure; however, we note that for multiple extraction regions the second temperature component was not well constrained. In order to avoid switching between 1T and 2T models in some arbitrary way, we decided to use the lognorm model. Fig. A3 shows a comparison of the best-fitting statistic obtained for all regions in the case 1 analysis when fitting the spectra with the lognorm model and with the 1-apec model (see Section 3.2). In all cases, the lognorm leads to a better fit, from a statistical point of view.