8. Correlations between X-ray emission and other stellar properties
In order to investigate the X-ray emission of TTS more closely, we studied correlations between the X-ray emission and other stellar properties of the stars of our sample. For completeness, data from Hughes et al. (1994 ) on the optically selected TTS in Lupus were incorporated. All X-ray data are taken from Krautter et al. (1996 ).
We determined H -luminosities by scaling the H -equivalent width with the stellar luminosity in the band. In order to obtain the chromospheric fluxes, we corrected individually for the photospheric absorption by adding the mean H absorption of a main-sequence star of the respective spectral type to the measured equivalent width of our WTTS. This correction was determined from spectral standard stars observed during the runs for the optical identification of our new WTTS (Krautter et al. 1996 ). The results are plotted in Fig. 5. It can be seen that the RASS-selected 'off-cloud' WTTS show high X-ray surface fluxes independent of the H surface flux , while 'on-cloud' WTTS as well as CTTS exhibit a large spread of for any given , with the CTTS sytematically displaced towards higher .
For our sample of Lupus TTS, we observe a strong anticorrelation of vs. only, if all TTS are taken into account, while no correlation is found within either one of the three subsets defined above (see Table 5). As the H luminosity of CTTS, contrary to that of WTTS, contains a large non-chromospheric contribution from the accretion disk and the stellar wind, while their mean X-ray luminosity is significantly lower than that of WTTS, we conclude that the observed anticorrelation within the whole sample reflects different sources responsible for the H emission of CTTS and WTTS rather than a physical (anti-)correlation between and . In a study of the X-ray emission of TTS in Taurus-Auriga, Neuhäuser et al. (1995a ) also could not find a correlation of emission and .
Table 4. Mean values of HR1, HR2 and E for Lupus TTS. Also given are errors of the mean, i.e. (standard deviation)/ .
Table 5. Tests for correlations. For the test hypothesis of no correlation we give probabilities calculated from the test by Kendall' (), and the test by Spearman's (). Also given are the slopes whenever both probabilities are below 0.05. The test samples are (1) all Lupus TTS for which X-ray and optical data are available, (2) CTTS only, (3) 'on-cloud' WTTS only, and (4) 'off-cloud' WTTS only.
If we examine the correlation of the equivalent width vs. rather than vs. , the increase of photospheric absorption and continuum flux for early spectral type, which causes a decrease of , can produce correlations with the X-ray emission, especially for the 'on-cloud' WTTS, which span a large range of spectral types, as can be seen from Table 5.
For active late-type main-sequence stars Fleming et al. (1989 ) found an upper limit to the surface X-ray flux , such that the maximum for a given stellar radius scales as . Thus, for a sample of X-ray selected TTS, which preferentially contains X-ray bright stars, a correlation of with is expected, if the X-ray emission mechanism of TTS is the same as for late-type dwarfs, while for the surface flux no correlation with should be observed. These effects can be observed for our sample of RASS-detected 'off-cloud' WTTS.
It is usually assumed that the X-ray emission of TTS and other late-type stars is caused by magnetic heating of the corona (cf. Montmerle & André 1988 , Montmerle 1990 ). In this model, saturation is expected to occur when the stellar surface is completely covered by active regions, and therefore X-ray bright stars at the saturation limit show a constant . Fleming et al. (1989 ) have found a value of erg sec-1 cm2 for this saturation limit.
Thus for CTTS, most of which are well below this saturation limit (see Fig. 7), can increase during the contraction.
In recent studies of TTS X-ray emission a correlation of with the bolometric luminosity has been observed (cf. Neuhäuser et al. 1995a , Feigelson et al. 1993 , Strom & Strom 1994 , Casanova et al. 1995 ). This observation is complicated by the fact that the slope of the observed relation is different in different SFR. While Strom & Strom (1994 ) infer from their deep ROSAT pointed observation centered on V410 Tau, Neuhäuser et al. (1995a ) find for Taurus-Auriga, in agreement with Feigelson et al. (1993 ) and Casanova et al. (1995 ), who evaluated deep ROSAT pointed observations in Chamaeleon and Oph, respectively.
In Fig. 6 we plot vs. for Lupus TTS. Using statistical tests (see Table 5), we conclude that a correlation of with is indeed present for both the 'off-cloud' WTTS and the 'on-cloud' WTTS, but not for the CTTS.
Now, if all stars in a given sample had the same surface flux, and if within this sample there were some relation , then we would expect , i.e., , which, within the errors, is consistent with our results for the RASS-discovered WTTS in Lupus (Table 5).
However, if as well as show some trend to increase with (which might be due to selection effects, as will be discussed below), the situation is more complicated, and the combined effect of both will produce a stronger correlation of with rather than with .
As the magnetic activity of T Taur stars presumably is caused by a dynamo process, some correlation between and stellar rotation is expected. Such a correlation has in fact been observed (cf. Bouvier & Bertout 1989 , Neuhäuser et al. 1995a ).
It has also been observed that on the approach to the main sequence, the rotational velocities of young stars first increase (cf. Bouvier et al. 1993, Edwards et al. 1993 ), and then decrease again, as discussed in Soderblom et al. (1993 ). This observation is explained in the framework of a model which suggests that during the CTTS phase the star is magnetically coupled to its disk, thus preventing a spin-up during contraction (Camenzind 1990 , Cameron & Campbell 1993 ). After the dissipation of the disk, the star can spin up due to contraction and angular momentum conservation, while later on magnetic braking caused by the interaction of stellar magnetic field and stellar wind occurs. (cf. MacGregor & Brenner 1991 , Bouvier & Forestini 1995 ).
If X-ray activity in fact depends upon rotation, we expect it to rise with age first, and then fall again among young stars, like the rotational velocities. For young clusters it is known that the mean X-ray luminosity of their stars decreases with the age of the cluster (Stauffer et al. 1994 , Pye et al. 1994 ), while on the other hand for TTS the X-ray luminosity increases significantly from CTTS to WTTS, as expected from this model (see Table 5). Moreover, the CTTS in Lupus show an anticorrelation of and rather than a correlation of and , which might indicate that for these stars rather than is about constant with radius, in line with the idea of disk-locking of the rotation.
In Fig. 7, we plot the X-ray surface flux vs. stellar age for the Lupus TTS. It can be seen that the maximum X-ray surface flux increases with age, with none of the youngest stars displaying comparatively high values of . Statistical tests (see Table 5) give evidence of a correlation of with age for the WTTS, while no correlation shows up for the CTTS. However, this might be due to the large scatter of for the CTTS, as Fig. 7 shows that even for the CTTS alone the maximum values of increase with age. An increase of with age has also been observed by Neuhäuser et al. (1995a ) for Taurus-Auriga TTS.
As within our sample, due to selection effects, the oldest stars probably are also the most luminous, this might be the physical reason for the observed correlation of with (there seems to be no significant correlation of and age, probably due to the large scatter, but the mean values of of CTTS, 'on-cloud' WTTS and 'off-cloud' WTTS are 6.0, 6.4 and 7.0, respectively, while the mean luminosities are -0.56, -0.45 and -0.21). Thus, within our sample there seems to be at least some trend for to increase with age.
Fig. 8 shows the X-ray surface flux vs. stellar mass in our sample. It can be seen that nearly all of the 'off-cloud' WTTS have relatively high masses as well as strong X-ray emission, which presumably reflects the selection effect due to the flux limit of the RASS. On the other hand, for the optically selected CTTS there is a large spread of at any given mass. No correlation between mass and is observed for both samples.
'On-cloud' WTTS show the same large spread of as the CTTS at low masses. At high masses, only WTTS with high are detected. However, this might be explained by the fact that the 'on-cloud' WTTS with highest mass are also the oldest within this sample.
The Lupus CTTS are unbiased with respect to and consist of younger and less massive stars than the WTTS. The 'on-cloud' WTTS, which have been discovered by pointed ROSAT observations, are much less biased towards stars with large than the 'off-cloud' WTTS. Also on average they are younger, in line with their location near the dark clouds. The discussion above has shown the importance of both selection and evolutionary effects, which can explain the observed difference between these three subsamples with respect to the correlations noted in Table 5.
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998