For catalogued sources detected in three or more colours we can produce colour-colour distributions. The two examples presented in the upper two panels of Figure 1 display the UVW2−UVM2 - UVM2−UVW1 plane and the U−B - B−V plane, both with relatively distinctive features. No attempt has been made to de-redden or k-correct the colours.
Fig 1: UV and optical colour-colour diagrams of catalogued sources taken from XMM-SUSS. Intensity scales are logarithmic. The red line represents synthesized stellar spectra from the ATLAS9 project (Castelli and Kurucz, 2004), ranging within photospheric temperatures 3,000 < T < 50,000 K. The three curves correspond to gravities logg = 0.0, 3.0 and 5.0 cm s-2. The red dots refer to the Pickles (1998) standard stars of spectral types ranging from M6 to O5 and luminosity types ranging from I to V. The various galaxy curves are evolutionary tracks for Starbursts, ellipticals, Spirals and Irregulars taken from the spectral models of Rocca-Volmerange and Guiderdoni (1988)which include stellar evolution, nebular emission and internal extinction. Characteristic AGN colours are taken from the STSDAS composite AGN library, constructed from IUE spectra.
A family of spectral models and standard sources, with no Galactic extinction or redshifts applied, have been folded through the XMM-OM filter bandpasses and displayed in the two lower panels of figure 1. The red tracks and pink dots correspond to various stellar models for the Galactic population and spectral standards stars. In optical colours, the distribution of this sample is easily understood with stars revealing ever-bluer colours, right-to-left and top-to-bottom as we travel from cool M to hot O stars. The UV colours reveal a different trend in the cool stars however, stars initially become redder as we travel from M stars to hotter types before turning a corner at approximately solar spectral types. This cool-star behaviour result from a combination of negligible UV flux and the wings of the UV transmission curves extending into the optical bands. Figure 2 reveals the extent of these wings and how more M star blue photons are collected through the UVW2 filter relative to UVM2.
Fig 2: The left panel plots both the effective area curves for the UV filters and an ATLAS9 synthetic M6 V spectrum. The right panel shows the wavelength distribution of photons, as a fraction of the total, detected through each filter. For cool stellar sources, the bluest UVW2 filter has a better optical response than the UVM2 filter, explaining the turnover in the UV colour-colour distribution.
Extragalactic models and standard spectra have been included in the lower panels of figure 1. Using these particular combinations of colours, the optical set provide a more sensitive diagnostic for discriminating Galactic from Extragalactic sources, in the absence of dust and cosmological expansion.
The observed colour-colour distributions are broader than those of the combined models and standard sources caused in part to photometric measurement uncertainty. Also we have yet to consider source reddening due to dust within our galaxy. In figure 3 we take the logg = 5.0 stellar sequence from the ATLAS9 models and extinguish them with increasing columns of Galactic dust content before folding the spectra through the XMM-OM effective area curves.
Fig 3: Colours from the ATLAS9 stellar spectra reddened by varying columns of Galactic dust. Spectral extinction coefficients are taken from Pei (1992), with Rv = 3.08.
UVM2−UVW1 colours naturally tend towards the red as dust extinction increases. UVW2−UVM2 colours however evolve towards the blue with increased redenning. This is a consequence of the broad 2175 graphite feature in the Galactic extinction curve. Due to the intrinsically red nature of cool stars there is degeneracy in the UV colours between M dwarfs and e.g. highly redenned G stars, a degeneracy which does not exist in optical colours, highlighting the importance of associated optical observations.
Figure 3 takes the family of cold elliptical galaxy models from Rocca-Volmerange and Guiderdoni (1988) and rebins the spectra to discrete redshifts out to z = 0.5. The absence of far-UV information for the oldest elliptical populations prevents us from extending the models accurately to higher redshifts.
For most sources along the track of cold ellipticals, UV colours between 0.00 < z < 0.15 indicate a general trend towards redder UVW2−UVM2 colours with redshift which, however, reverses towards bluer colours at > 0.15. There is a UV upturn in the spectral models due to horizontal branch populations which the XMM-OM bandpasses cannot detect in the restframe, but sources at z > 0.15 become increasingly sensitive to the feature as it shifts towards the red. It is the upturn which is causing the trend towards blue colours with higher redshift. While UVW2−UVM2 tracks the old, post-main sequence population at high redshift, UVM2−UVW1 colours remain insensitive to the UV upturn. Instead, they provide a tracer for star formation, albeit a tracer potentially confused by foreground dust (fig. 3).
The optical colours from the elliptical galaxy simulation evolve towards the top-left of the U−B - B−V diagram between 0.00 < z < 0.35. At redshifts > 0.35 the previous trend is reversed. By far the strongest spectral feature within ellipticals at z < 0.5, is the Balmer limit at a rest wavelength λ3646Å which passes from the XMM-OM B filter to the V filter z ~ 0.35. It is this transition across filter bandpasses which causes the switch in colour trend.