4.1. Continuum detections
4.1.1 The Nevalainen et al. (2003) sample
Nevalainen et al. (2003) found evidence for soft X-ray excess in clusters of galaxies using XMM-Newton EPIC data. They derived a 20-40% soft excess in the central 500 kpc regions of the clusters Coma, Abell 1795 and Abell 3112 (z = 0.0703) in channels below 2 keV, consistently in both PN and MOS instruments, and in ROSAT PSPC data.
A thermal model fits the data better than a non-thermal one, but at the level of calibration accuracy at the time, the non-thermal nature of the soft excess could not be ruled out.
Thermal modelling yielded temperatures in the range of 0.6-1.3 keV and metal abundances consistent with zero. Assuming that this warm gas occupies the same volume as the hot gas, the electron densities are of the order of 10-4 - 10-3 cm-3. These values lead to a cooling time scale larger than the Hubble time, i.e. the structures are self-consistent.
4.1.2 The Kaastra et al. (2003) sample
Kaastra et al. (2003) examined a sample of 14 clusters of galaxies observed with XMM-Newton, and found significant evidence for soft X-ray excess continuum emission in 5 of them: Coma (centre studied only, since it is closer than the other 4), Abell 1795, Sérsic 159-03, Abell 2052 and MKW 3s. Different modelling of the hot gas temperature and metal abundance yielded lower temperatures for the soft component (~ 0.2 keV), compared to Nevalainen et al. (2003). The surface brightness of the warm gas is rather constant with radius, while that of the hot gas decreases with radius, falling below the warm gas surface brightness between 0.5 and 1 Mpc from the cluster centre. Note that in Coma, the central pointing does not cover this radius, so the hot gas brightness remains above the cool gas brightness, consistent with the above number. Such a behaviour is consistent with the Warm Hot Intergalactic Medium (hereafter WHIM) filament scenario, whereby the projected external filamentary structure is more extended than the cluster. Later, Kaastra et al. (2004) extended this sample to 21 clusters, and found 7 objects with soft excesses, the new cases being Abell 3112 (also in Nevalainen's paper) and Abell 2199.
Based on the quality of the spectral fits, in most cases the non-thermal model was also acceptable. Fitting the soft excess with a power-law model yielded rather constant photon indices (~ 2) with radius in a given cluster. At the cluster centres the luminosity of the non-thermal component is ~ 10% of that of the hot gas, while the percentage increases towards 100% at the largest radii.
Bregman & Lloyd-Davies (2006) challenged the results of Kaastra et al. (2003), arguing that their soft excess detection was due to incorrect background subtraction. However, in a rebuttal paper, Nevalainen et al. (2007) showed that, especially in the central regions, the cluster emission is so bright compared to the background, that the details of the background modelling are insignificant. Thus, Bregman & Lloyd-Davies's claim appears to be unjustified.
Nevalainen et al. (2007) also found that the changes in the EPIC calibration between the years 2002 and 2005 resulted in a decrease of the soft excess signal. Using the Kaastra et al. (2003) modelling, the PN soft excess even disappeared in some clusters. However, the MOS instrument still detects a soft excess in all the clusters of Kaastra's sample. A less conservative hot gas modelling (Nevalainen et al., in preparation) with the current calibration information obtains consistent soft excesses in both PN and MOS data.
4.1.3 Sérsic 159-03
One of the first clusters to be observed by XMM-Newton was Sérsic 159-03. This observation with a net exposure time of 30 ks was taken in 2000 (Kaastra et al. 2001). The same observation was used for a search for soft excess emission (Kaastra et al. 2003, see Sect. 4.1.2 above). A 60 ks observation taken two years later was analysed by Bonamente et al. (2005a) and de Plaa et al. (2006).
Bonamente et al. (2005a) confirmed the existence of the soft X-ray excess emission in Sérsic 159-03 out to a distance of 1 Mpc from the cluster centre. The soft excess in the 0.3-1.0 keV band increases from 10% at the centre to 80% at the largest radii. The properties of the soft excess differ from those derived for Sérsic 159-03 using PSPC data (Bonamente et al. 2001d), likely due to different modelling of the data: XMM-Newton allows to determine the hot gas component unambiguously in the 2.0-7.0 keV band where the soft component has a negligible contribution.
The same dataset of Sérsic 159-03 was also analysed by de Plaa et al. (2006). However, they focused their attention on the chemical evolution of the cluster and they determined the background for the spectral analysis in the 9'-12' region around the cluster, subtracting the spatially extended soft excess emission together with the soft foreground emission. For the remaining soft excess in the cluster core they propose a non-thermal emission mechanism arising from IC scattering between CMB photons and relativistic electrons accelerated in bow shocks associated with ram pressure stripping of infalling galaxies.
Werner et al. (2007) studied Suzaku data together with the two XMM-Newton data sets of Sérsic 159-03 obtained two years apart mentioned before. They found consistent soft excess fluxes with all instruments in all observations. From the XMM-Newton data they derived radial profiles and 2D maps that show that the soft excess emission has a strong peak at the position of the central cD galaxy and does not show any significant azimuthal variations. They concluded that the spatial distribution of the soft excess is neither consistent with the models of intercluster warm-hot filaments, nor with models of clumpy warm intracluster gas associated with infalling groups as proposed by Bonamente et al. (2005a). Using the data obtained with the XMM-Newton RGS, Werner et al. could not confirm the presence of warm gas in the cluster centre with the expected properties assuming the soft excess was of thermal origin. They therefore concluded that the soft excess in Sérsic 159-03 is most probably of non-thermal origin.
4.2. Line emission detection
A crucial piece of evidence for the thermal nature of the soft X-ray excess would be the detection of emission lines. At the indicated temperatures of 0.1-0.5 keV the most prominent emission lines are from O VII and O VIII. The resonance, intercombination and forbidden lines of O VII have energies of 574, 569 and 561 eV (see also Kaastra et al. 2008 - Chapter 9, this volume). For a low density plasma at 0.2 keV temperature, the centroid of the triplet has an energy of 568.7 eV (see Kaastra et al. 2003).
Geocoronal and heliospheric solar wind charge exchange (Wargelin et al. 2004, Fujimoto et al. 2007) also produces soft X-ray emission lines. These lines can vary by a factor of 3 on time scales of hours, which makes them difficult to model and subtract. The maximum observed brightness of these lines in the 0.5-0.9 keV range can reach the level of the cosmic X-ray background. Therefore, when the background level is important for the analysis, care should be taken when estimating the contaminating effects of this emission.
Finoguenov et al. (2003) observed a spatial variation of the soft excess emission in the Coma cluster. In the Coma 11 field they found an excess emission which is particularly strong. Recent observations with the Suzaku satellite (Takei et al., private communication) do not confirm the level of O VII and O VIII line emission reported by Finoguenov et al. (2003) for the Coma 11 field, but they are consistent with the lowest reported values for the other fields in Coma observed with XMM-Newton. They suggest that a large fraction of the reported excess soft X-ray emission and the line emission observed in the Coma 11 field with XMM-Newton was due to Solar wind charge exchange emission.
4.2.1 The Kaastra et al. (2003) sample
Kaastra et al. (2003) found evidence for O VII line emission in the clusters Sérsic 159-03, MKW 3s (z = 0.0450) and Abell 2052 in form of line-like residuals on top of the hot gas model at wavelengths consistent with the cluster redshifts. Note that the line emission is significant only in the outer regions (4' - 12') of the clusters.
The uncertainties also allow a Galactic origin (z = 0) for the emission. Also, Nevalainen et al. (2007) pointed out that a simple cluster-to-background flux level comparison does not exclude that the reported O VII line emission in the 0.5-0.65 keV band contains contributions from the geocoronal and heliospheric Solar wind charge exchange (see above). This possibility remains to be studied in detail.
Therefore this study does not give conclusive evidence for O VII line emission in these three clusters.
Finoguenov et al. (2003) analysed XMM-Newton data in the outskirts of the Coma cluster. The spectrum at ~ 1 Mpc distance from the centre exhibits a very strong, ~ 100% fractional soft excess continuum in channels below 0.8 keV, and the authors detect two emission lines in the 0.5-0.6 keV band. Both the excess continuum and the line features are well fit with a thermal model of kT = 0.22 keV (see Fig. 3). The temperature, as well as the derived baryon overdensity for the warm gas (~ 200) are consistent with the WHIM filament properties in numerical simulations (e.g. Davé et al. 2001, Yoshikawa & Sasaki 2006). Let us note however that the soft excess in Finoguenov's model consists mostly of line emission, but due to the low spectral resolution of EPIC at low energy, it looks more like a continuum.
The redshift of the O VII line (0.0-0.026) is consistent with that of Coma (0.023) and also with being Galactic (0.0), implying that the emitter is located between us and Coma. Optical data show that there is a significant galaxy concentration in the direction of the warm gas in front of Coma. This suggests that the warm gas and the galaxy structure are connected, consistent with the WHIM filament scenario.
However, since the soft excess flux in Coma is comparable to the background level, the charge transfer flux in an active state (i.e. at the maximum observed level) can reach the soft excess level detected in Coma (Finoguenov et al. 2003). Bowyer & Vikhlinin (2004) use this coincidence as a proof that the soft excess detection in Coma is due to charge transfer. However, they did not model the soft excess spectrum with a charge transfer emission model to show that the model is consistent with the observed spectral features in Coma. Thus, their claim of Coma soft excess being due to charge transfer mechanism is not proven.
An indication for the presence of WHIM in Coma was also recently reported by Takei et al. (2007b) at 2.3 level in absorption and in emission, based on RGS spectra of the Active Galactic Nucleus X-Comae, which is located behind the Coma cluster.