Spectroscopy and kinematics of low-mass members of young moving groups

We performed a simple analysis to obtain the objects’ spectral types. We measured the PC3 index as defined in Martín et al. (1999) and applied the index-spectral type relation to derive a spectral type. We measured also TiO5 and VO-a indexes as defined in Cruz & Reid (2002). We derived spectral type from TiO5 following the double valued relation given in Cruz & Reid (2002), using the relation for Ti05$<$0.75, except for object with spectral type later than M7 were we used the relation given for Ti05$>$0.30. We also derived a spectral type from VO-a index following the relation gives in Cruz & Reid (2002) for the correspondent spectral type interval. We took as definitive spectral type the average of these three spectral types rounded to the closest 0.5 class. When one of the spectral areas covered by these indexes were no usable we make the average between the other two. The spectral type estimated error is $\approx$1.0 spectral class derived from these relations.

Due to the wavelength range of the Phoenix data, we can not measure PC3, so we measure the spectral types instead using the I-J colour index-spectral type relation from Leggett (1992).

We then calculated a more accurate spectroscopic distance using a PC3-absolute J-magnitude relation from Crifo et al. (2005). The error in distance following Crifo is 12%. For Phoenix data, distances were calculated from the absolute J magnitude-spectral type relation from Dahn et al. (2002). The spectral classification and distances are given in Table 2, where previous known data from the literature is also given. The values in both sets are in agreement except for two objects, 2MASS J0334-2130 and SIPS 1632-0631, that differ in more than one spectral type and SIPS2039-1126 that differ one spectral type. The value from the literature for 2MASS J0334-2130 and SIPS 2039-1126 classification comes from Cruz et al. (2003), where all spectral types for M dwarfs were determined via visual comparison with standard star spectra taken during the course of the program (the objects were typed by being normalised and plotted between spectra from a grid of eight standard M1-M9 dwarfs from Kirkpatrick, Henry & McCarthy 1991). In this case we think our index average measurements will give more accurate spectral types. For SISPS 1632-0631, we took the spectral type from Gizis (2002), where the final classification comes from the measurements of several spectral indexes (including TiO5, PC3, CrH, TiO-b and visual inspection). The PC3 and TiO5 in Gizis (2002) (1.99, 0.25) are very similar to the ones calculated by us (1.95, 0.26), but our value of VO-a (2.35) is the double of Gizis’s (1.14). Although Gizis (2002) takes an average of six indicators while we only take three, we rely on our classification. As we mentioned above, the spectral type-index relation of TiO5 and VO-a indexes in the Cruz & Reid (2002) is double value with a turning around M7 and M9 for TiO5 and VO-a respectively, that is that there is a different relation for objects with spectral type under or over around M7 for TiO5 and M9 for VO-a. SIPS 1632-0631 is just in the M7-M9 limit. We find that the best agreement in the spectra type with PC3 is when we used the relation for $>$M7 in TiO5 and for $<$M9 in VO-a, giving same spectral type, M8.5, for the three indexes.

Due to the spectroscopic binarity, as we will see in Sect. 5.3 or possible binarity without confirmation of some object in the sample we expect that the error in the indexes spectral types and distances are larger than the ones shown in Table 2.

To obtain spectral types for our FEROS objects we initially used the PC3 index from Martín et al. (1999), however this coincided often with a unusable region of spectra. We thus used the VO index from Kirkpatrick, Henry & Simons et al (1995), CaH from Kirkpatrick, Henry & McCarthy (1991), VO-a and TiO5 from Cruz & Reid (2002) and PC3, TiO1+TiO2 and VO1+VO2 from Martín et al. (1999) to create a mean spectral type (see Paper I for details).

The heliocentric radial velocities have been determined by using the cross-correlation technique. The spectra of the targets were cross-correlated order by order, by using the routine fxcor in IRAF, against spectra of radial velocity standards with similar spectral type.

The radial velocity was derived for each order from the position of peak of the cross-correlation function peak (CCF), and the uncertainties were calculated by fxcor based on the fitted peak height and the antisymmetric noise as described by Tonry & Davis (1979).

In the case of the FEROS data, we cross-correlated each individual order of the reduced spectra using an IDL routine that determines the cross-correlation of two arrays and finds the maximum cross-correlation function by fitting a Gaussian to the central peak. Our final radial velocities and associated uncertainties came from the mean and standard deviation of these measurements. A heliocentric correction was made to the measured shift between the object and the reference star so that the we could calculate radial velocities with respect to the Sun.

In Table 3 we give the information about the reference stars used in every run. The object LP 944-20 exhibits $\sim$3.5 km s^-1 radial velocity variability, as is shown in Martín et al. (2006), so when we had another radial velocity template in the run we checked the velocity of LP 944-20 with respect to this other template and found good agreement within the uncertainties.

In Table 4 we list, for each spectrum, the heliocentric radial velocities ($V_{\rm r}$) and their associated uncertainties ($\sigma_{V_{\rm r}}$) obtained as weighted means of the individual values deduced for each order used. Those orders which contain chromospheric features and prominent telluric lines have been excluded. In the case of Phoenix data, the signal-to-noise of the spectra is low and so we decided to add an uncertainty deduced from the variation found in the standard stars calibrations (LP 944-20 and GL 406 in this case).

For the binary case (2MASS0123-3610), we have to take into account an additional uncertainty due to the systematic distortion of the peak from the presence of a companion.

To constrain space motions, we obtained a number of distances depending of potential MG membership for each object with radial velocity measure.

As detailed in Paper I, to allow for younger objects appearing intrinsically brighter, we created isochrones corresponding to the ages of our MGs and that of a field dwarf using atmospheric models from Baraffe et al. (1998). According to each object’s candidature group membership from photometric and astrometric criteria, the difference in absolute J-magnitude of each object is used in Dahn et al. (2002) distance-spectral type relation to give a new corrected distance assuming group membership. This way we obtained a number of distances for each of the objects.

Using these distances or any parallaxes available from the literature, the proper motion data, and the radial velocities calculated in Sect. 4.2, we computed the Galactic space-velocity components ($U$, $V$, $W$) of the sample by using the transformation matrices of Johnson & Soderblom (1987). Positive U is towards the Galactic centre, positive V is in the direction of Galactic rotation, and positive W is in the direction of the Galactic north pole.

The resulting values of ($U$, $V$, $W$) according to potential group membership and associated errors are given in Table 4. In the table, the objects that appear to be kinematic members, that is, that are in agreement with the membership, are highlighted in bold (see Section 5).

To determine rotational velocities we made use of the cross-correlation technique in our high resolution echelle spectra by using the routine fxcor in IRAF.

When a stellar spectrum with rotationally broadened lines is cross-correlated against a narrow-lined spectrum, the width of the cross-correlation function (CCF) is sensitive to the amount of rotational broadening of the first spectrum. Thus, by measuring this width using Gaussian profiles, one can obtain a measurement of the rotational velocity of the star. The target spectra were cross-correlated against the spectrum of the template stars. We give the data of the templates used in each run in Table 3. The templates should be of equal or similar spectral types as the targets, but an M-type template can be used for all M type dwarfs (see e.g. Bailer-Jones 2004; Zapatero-Osorio 2006), therefore as our targets and references are all in M4-M9 range, we applied our templates to all stars.

The calibration of the width (FWHM) of the templates to yield an estimation of $v\sin{i}$ is determined by cross-correlating artificially broadened spectra of the template star with the original template star spectrum. We rotate them up to 100 km s^-1 in 1 km s^-1 steps. The broadened spectra were created for $v\sin{i}$ is spanning the expected range of values by convolution with a theoretical rotational profile (Gray 1992) using the program starmod (developed at Penn State University; Barden 1985) and the vsini routine of the SPECTRUM synthesis code (Richard O. Gray, 1992-2008). The resultant relationship between $v\sin{i}$ and FWHM of the CCF was fitted with a fourth-order polynomial. This method should not be used at rotational velocities greater than around 50 km s^-1 since the CCF profiles radically differ from Gaussianity. Therefore, in our rotational range of interest, 5$>$$v\sin{i}$$>$50, the fit is good. We used several spectral ranges for these measurements. As mentioned in Reiners & Basri (2008), absorption band of FeH around 1$\mu$m is a rich area in ultracool stars, free of telluric absorption and is not pressure-broadened, but the spectra that we had covering that area (UVES data) present an artifact that do not allow us to make any measurements with the reliability needed. So, we use that band only to check the values obtained in other areas.

The rotational velocities given in the 8th column of Table 4 are the average of the values measured when more than one template star was used.

The uncertainties on the $v\sin{i}$ values obtained by this method have been calculated using the parameter R defined by Tonry & Davis (1979) as the ratio of the CCF height to the rms antisymmetric component. This parameter is computed by the task fxcor and provides a measure of the signal-to-noise ratio of the CCF. Tonry & Davis (1979) showed that uncertainties in the FWHM of the CCF are proportional to 1/(1 + R) and Hartmann et al. (1986) and Rhode et al. (2001) found that the quantity $v\sin{i}$[1/(1 + R)] provides a good estimate for the 90% confidence level of any $v\sin{i}$ measurement using this methodology. This error, $v\sin{i}$[1/(1 + R)], should be a reasonable estimate of the uncertainties on our $v\sin{i}$ measurements. Nevertheless, we estimate the uncertainties to be at least 3 km s^-1 for all values, which is the level of variation presented by the reference stars. The reason is that the main source of uncertainty comes from the low signal-to-noise of the spectra and so is not reflected properly in the statistical results of the calculations.

Taking into account the resolution of our spectra, the rotation of our templates (see Table 3) and the signal-to-noise, we consider that the minimum detectability value of $v\sin{i}$ would be in the range $\approx$6-10 km s^-1. Thus, we have assumed a general limit of 10 km s^-1. All data below this limit would be marked as $v\sin{i}$ $\leq$ 10 km $s^{-1}$ in Table 4.

We apply a simple kinematic criterion to distinguish old and young disk objects in our sample of 68 objects with high resolution spectra.

In Figure 1, we plot the UV and WV planes for the whole sample including the boundaries (continuous line) that determine the young disk population as defined by Eggen (1984a,b, 1989). We then classify objects as young disk (YD) if they lie within the young disk region in the UV plane (or if they could, given U and V 1$\sigma$ errors). Objects whose location in the UV plane is inconsistent with YD membership are all consistent with an old disk (OD) classification.

If we follow Legget (1992) criterion, the sample should be divided into young disk (YD), old disk (OD), young-old disk (YO), old-disk halo (OH) and halo (H) objects: all stars with V $<$ -100, or with an eccentricity in the UV plane $>>$0.5, were defined as H; objects with an eccentricity in the UV plane of $\approx$0.5 were defined as OH; stars within the young disk region in the UV plane, approximately defined by -20 $<$ U $<$ 50, -30 $<$ V $<$ 0, and with -25 $<$ W $<$ 10, were defined as YD; objects with an eccentricity in the UV plane less than 0.5, that lie outside the young disk region, or that lie within the young disk region but have $|W|$ $>$ 50, were defined as OD. Finally, objects that lie around the edge of the young disk region or lie within the area with $|W|$ $<$ 50 but greater than that required for classification as YD, were defined as YO. However, in our sample all the objects would be classified by young disk or old-young disk membership, (see Fig. 1 and W value in Table 4), and will refer to them as YD and OD for simplicity.

In addition we identify possible members of the five MGs by their relative position in the UV and VW diagrams (Fig. 1) with respect to the boxes (dashed lines) that mark the velocity ranges of each MG. The centre of the five main MGs are marked (with large open symbols) and our kinematic members are shown as corresponding filled symbols. The boxes are defined by studies with samples of higher mass members from the literature (Barrado Y Navacués 1998; Montes et al. 2001). See Paper I for a detailed description. As mentioned above, a kinematic member was defined as either lying inside the relevant box in each diagram, or having uncertainties that overlapped appropriately. Objects inside YD boundaries should be young although it is not clear that they belong to an established MG.

According to their kinematic behaviour, we have 49 objects that belong to the young disk area, and 36 of them are candidates members of one of the five MGs. Some objects are candidates for more than one MG (see Table 7). We find seven possible candidates to Castor, 25 to Hyades, seven to IC 2391, five to Pleiades and none to Sirius. Eleven objects belong to young disk region without clear membership to any of these five MGs.

These new candidates represent a significant addition to the low-mass membership of several MGs that for the most part have few low-mass members confirmed to date, and provide a corresponding improvement in any determination of the low-mass initial mass function (IMF). Our seven candidate members of the Castor MG represent an increase of 10% in the total known membership (see e.g. Caballero 2010), and an increase of 50% in the very low-mass regime. The Hyades group presents a deficiency of very low-mass objects (15 of $\approx$500 stars). Bouvier et al (2008) suggest that the Hyades originally had 150-200 brown dwarfs (BDs) compared to its current 10-15, based on the assumption that the Hyades is a dynamically evolved version of the Pleiades and considering the two clusters present mass functions over the 0.005-3M_⊙ range. Hogan et al (2008) reported 12 new L dwarf Hyades members (estimating $\approx$20% contamination). Here we present 25 low-mass candidates which, if confirmed, will further aid our understanding of the evaporation of young clusters. IC 2391 as well as other young clusters present also a brown dwarf deficit, that might be explained by an onset of larger-size dust grain formation in the upper atmosphere of objects with spectral types M7-M8 or later (Dobbie et al. 2002; Jameson et al. 2003; Barrado Y Navacués et al. 2004; Spezzi et al. 2009). From 180 members around a dozen are substellar (Spezzi et al. 2009), a low number from which to draw a statistically significant conclusion. Although our candidates are M4.5-M7 spectral type, they increase the low-mass population and favour future statistics. Pleiades is a very well studied and populated cluster and many works have undertook the low-mass and BD population search and IMF study (e.g. Londieu et al. 2007), our five candidates are in M6-M8.5 spectral type and will contribute to populate this spectral type range.

We have to keep in mind that contamination of the young disk space velocity area by old field population occurs. It has been studied in several kinematic moving groups as we mentioned in the introduction, concluding that a high percentage of contamination happens (e.g. López-Santiago et al. 2009 and reference therein). So the next logical step is to further asses the kinematic candidates found with additional information, such as the study of several youth indicators, to confirm their membership.

The rotational evolution of low-mass object is complicated. Several studies have demonstrated that after becoming a fully convective object (spectral types later than $\approx$M3), when gravitational contraction is finished, rotational braking is present at least in all M type objects. The spin-down times are longer for Late M dwarfs than for earlier M types (e.g. Zapatero-Osorio et al. 2006; Reiners & Basri 2008, Jenkins et al. 2009). Jenkins et al. (2009) show a statistical relationship between mid/late type M dwarf rotation and age and hence taking all these works together we can use rotational velocity as a way to differentiate between young and older M type UCD populations. As demonstrated in Figs. 9 and 10 of Reiners & Basri (2008), such populations occupy distinct regions of the $v\sin{i}$-spectral type diagram depending on its age, with the older objects concentrated in an ”envelope of minimum rotation velocity”. Even for the latest M dwarfs this envelope corresponds to a fairly low rotation rate ($v\sin{i}$ $<$ 20 km s^-1). The model overlays that Reiners & Basri (2008) derive (based on a breaking law fit to the overall observed population) that the great majority of mid-late field M dwarfs (e.g. with ages $\sim$2-10 Gyr) have $v\sin{i}$ $<$ 30 km s^-1, and that younger objects of this type should be much more rapid rotators ($v\sin{i}$ = 30-100 km s^-1). Although the random inclination of rotation axis can reduce measured $v\sin{i}$, it seems clear that rotational velocity constraints can provide a very helpful complement to our kinematic membership.

One Reiner & Basri (2008) result is that the rotational evolution according to a wind-braking law that scales with temperature can reproduce better the observational data for L dwarfs. While for M dwarfs both rotational evolution according to a wind-braking law that scales with temperature or with mass are very similar. We choose the first one here although this is a moot point since both models give similar results over our spectral type range. In Fig. 2 we plot spectral type versus $v\sin{i}$. On the left, we plot our sample with Reiners & Basri (2008) Figure 10. Circles are from them and open triangles from Zapatero-Osorio et al. (2006). Following Reiners & Basri (2008), filled blue circles are probably young, filled red circles old and open circles have no age information. Our sample data is plotted as different symbols depending if they have been classified as YD or OD. As in Reiners & Basri (2008), we mark in dashed lines ages of 2, 5 and 10 Gyr (from upper left to lower right). To obtain a criterion to separate objects of differing age, we plot in blue a continuous line: the ”apparent” separation between a ”young” and ”old” object and so use it as an indicator of MG membership for our targets. We include LP 944-20 in the plot as it shows a strong candidature to be a member of Castor MG (see Ribas 2003b for details). It is plotted as an arrow with our measured value of $v\sin{i}$, in agreement with previous works (see Table 3) and with being young. On the right side we show a zoom version where we plot the targets in different symbols, which indicate for which MG they are candidate members. In this case we do not plot literature data for clarity. Note that the objects marked as $<$10 km $s^{-1}$ in Table 4 are plotted with this value in Fig. 2, so this should mark an upper limit of rotational rate.

We measure the rotational velocity of 54 of the 68 objects in the sample, and of 42 of the 49 YD objects. From these 42, we find that 31 present a projected rotational velocity in agreement with our criterion of youth and 12 that can not be dismissed as they have velocities in the limit between both populations or we have no rotational velocity information.

Despite being outside the young disk region, 10 objects have youthful rotational velocities. Uncertainties in the (U, V, W) measurements would be compounded by possible binarity among older objects which can allow them to keep high rotation rates longer that in case of singles.

Table 7 sums up these results. The first and second columns are name and kinematic classification. Third column classifies the objects as young (Y) if they lie in the upper part of the limiting line, old (O) if they lie under the line and unknown age (Y-O) if they are in the limit of both areas or the rotational velocity superior limit is not clear enough to classify them in one of the groups. Column four highlights to which MG the candidates are kinematic members. In this case, HY for Hyades, SI for Sirius, CA for Castor, PL for Pleiades and IC for IC 2391, and OYD for other young disk members without a membership of one of the five MG in consideration. The confirmed 31 YD members objects are highlighted in bold in Table 7, and the 12 more including as possible members are highlighted in bold with a question mark. Only two objects that presented candidature to any of the MGs have been dismissed due to the rotational velocity and have a (N) in Table 7.

As we have used the projected rotational velocity for our study, the real rotational velocity can only be higher and so the objects over our youth limit will stay there, but the dispersion introduced by the inclination and the fact that the rotation rate can be influenced by other external factors, like binarity, make us use this method only as a supporting criterion.

Five objects of our list, DENIS0041-5621, 2MASSJ0429-3123, SIPS0440-0530, 2MASSJ1507-2000 and DENIS2200-3038, overlap with Reiners & Basri (2009) and Seifahrt et al. (2010). The kinematic shows they are YD except for 2MASSJ1507-2000, while two of them show MG candidature, SIPS0440-0530 and DENIS2200-3038. DENIS0041-5621 shows evidence of accretion and presence of Li i 6708 Å, which implies that it is very young (Reiners & Basri 2009). We had classified this object as other young disk, as it is kinematically young and the rotational velocity is in agreement. 2MASSJ0429-3123 shows no Li i in Reiners & Basri (2009). We have this object classified as YD due to kinematics but the rotational velocity indicate it could be old, classified as an M7.5 with no lithium, it is probably an old object. SIPS0440-0530 is classified by us as a young object member of Hyades MG. Reiners & Basri (2009) found no Li i but this could be compatible with being a HY member. H$\alpha$ and other indicators of activity will give us more information as it is M7.0 spectral type. 2MASSJ1507-2000 shows quite different radial velocity measurement in Reiners & Basri (2009) (-2.5 km s^-1 instead of our -22.18 km s^-1), we can not explain this difference, the additional error maybe produced by low signal-to-noise spectra is not enough for producing this discrepancy. This object is kinematically classified as OD and also shows no Li i in Reiners & Basri (2009). DENIS2200-3038 is kinematically classified as Hyades member, which is consistent with the findings of Seifahrt et al. (2010), but with M9 spectral type the rotation criterion gives an old object. Further spectroscopic criteria is needed to confirm these results.

Multiplicity of stars can yield dynamical mass constraints, as well as being an important constraint of stellar formation and evolution. The properties of multiple stellar systems have long provided important empirical constraints for star formation theories. Binary properties of the low-mass stars of FGK and early M spectral types have been extensively studied (e.g., Duquennoy & Mayor 1991 and Fischer & Marcy 1992). But, despite the interest in the multiplicity of very low mass objects (stars at the bottom of the main sequence and brown dwarfs) in recent years (e.g. Martín et al. 2003; Close et al. 2003, 2007; Gizis et al. 2003; Pinfield et al. 2003; Basri & Reiners 2006; Reid et al. 2006; Ahmic et al. 2007; etc), and the subsequent progress in finding and characterising them, very few direct measurements of their physical properties have been made, in particular dynamical masses (e.g. Zapatero Osorio et al. 2004, Dupuy et al. 2009a and Dupuy et al. 2009b), and our understanding of their formation processes is not clear.

Double (or multiple) lined spectroscopic binaries (SBs) allow precise determination of dynamical properties of the components, including the mass ratio and, if the inclination can be determined, the individual component masses.

During the reduction process of our sample we find some objects that show possible spectroscopic binarity, since they show some contribution of the secondary in the spectra, but we can confirm only one of them. We present here an object that can be clearly defined as double line spectroscopic binary or SB2.

On the left-hand side of Fig. 3 we plot a example of the fxcor cross-correlation were the peak of each component can clearly be seen and so fitted separately. On right-hand side of the Fig. 3, we plot a piece of spectra of our target (below) and the reference single star GI 876 (above). The absorption lines from both components are also well distinguished.

Due to the SB2 nature we could measure the radial and rotational velocity of both components (see Table 4 and Sect. 4). We derived Galactic space-velocity components ($U$, $V$, $W$) in the same way we specify in Sect. 4.

Since we have the double lines in the spectra, when we applied the indexes criteria for obtaining the spectral classification, we have the lines of both components together that we could not disentangle. This produce an uncertainty in the spectral classification an so in the distances derived.

In order to obtain an estimate of the spectral type we use other objects observed in the same run to construct synthetic spectra by using the program starmod (see Sect. 4.4). starmod allows to rotationally broaden spectra and shift them in the velocity space. We ran it on the spectra of several spectral types objects and combined with the appropriate weights to create composite spectra that were used as template for comparison with our target. The best fit we find corresponds to a M7 primary and at least M7 or later secondary in a 50/50% contribution, but earlier primary is possible and more spectra would be necessary to improve the results with this technique. The two wavelength regions chosen for the fits are plotted in Fig. 4, centred in the first line of Na i doublet ($\lambda\lambda$ 8183, 8195 Å) and centred in $\approx$8385 Å where there are several Ca i and Fe i lines. The observed spectra of 2MASS0123-3610 is plotted as continuous line and the correspondent synthetic spectra is overplotted as dashed line.

From the different spectral types possible for the primary, we derive the corresponding distances and galactic velocity components. Fig. 5 represent the correspondent position in the YD area in the UV and VW planes. In any case the binary lies inside the YD area and also inside Hyades MG or IC 2391 MG boxes. Taking into account these possible memberships, we estimate the masses of 0.15M_⊙ and 0.09M_⊙ for primary and secondary components respectively if the system belongs to the Hyades MG and 0.06M_⊙ and 0.04M_⊙ if it belongs to the IC 2391 MG.

Rotational velocity measurements give a value at the limit of our sensitivity ($\approx$10 km s^-1) for both components. This fact, the binarity and the uncertainty of the spectral classification gives us no clear conclusion about its youth using rotational velocity criterion.

We will take high resolution spectra to follow up the target and measure radial velocities through the period in order to get the orbital parameters. Due to the SB2 nature we will be able to get the masses of both components. If the system is confirmed to belong to Hyades or IC 2391, we will have a presumably coeval UCD system with known age, metallicity and stellar parameters. Identifying such objects is crucial to increase the sample of benchmarks that can provide empirical constraints for star formation theories.

We studied the spectra of a 68 target sample of low-mass object previously selected via photometric and astrometric criteria, as possible members to five known young moving group. Using high resolution spectra we measure spectral types and derive distances to determine galactic space-velocity components. After applying kinematic criterion we find that 49 targets belong to the young disk area and that 36 of them possibly belong to one of our five moving groups. We measure when possible projected rotational velocities in order to use the rotation rate as a supporting criterion of youth and so confirm the kinematic members. We find that from the young disk targets, 31 have rotational velocities in agreement with their youth and 12 can not be dismissed as YD due to rotational velocities and further criteria should be applied.

These new candidates significantly increase the known population of low-mass young MG members, which gives rise to an important statistical improvement in calculations of each clusters IMF.

Within the sample, we find a SB2 binary possible member of Hyades or IC 2391 MGs. The spectral type classification gives as a $\approx$M5-M7 primary and a possible later than M7 secondary. The rotational velocity criterion in this case is not conclusive so further criteria should be applied. Follow up spectroscopy will be carried out in order to improve the classification and to obtain orbital solution and derive physical parameters. If confirmed as a MG member, this object can serve as a very valuable testbed for evolutionary models.

Further reliable spectroscopic criteria should be applied to confirm the youth of the candidates found here. The analysis of some age sensitive spectroscopic signatures such as Li i 6708 Å presence an equivalent width, H$\alpha$ emission (e.g. West et al. 2008) and gravity sensitive features (e.g. Gorlova et al. 2003; McGovern et al. 2004) will provide reliable age constraints for our sample. As we mentioned in Section 3, UVES observing runs include Li i but it is on a very low signal-to-noise region and so no conclusion could be reached. FEROS run included both H$\alpha$ and Li i region, but although we can see the emission in H$\alpha$ in nearly all FEROS targets, the signal-to-noise did not allow us to do any equivalent width measurement, and the Li i region is also unusable due to low signal. We are so currently carrying out new spectroscopic observations to study these two important age features. Detailed information of the criteria and how we will use them depending on the spectral type and MG candidature of our sample can be seen in Sect. 4.4.1 and Fig. 9 in Paper I.

To improve the space motion determination and so improve the kinematic criterion, we are also carrying out a parallax program that will allow us to measure more accurate distance.

Discovering planetary systems around young and low-mass objects would be very revealing for the understanding of planet and stellar formation. By studying these systems we will probe the host mass impact on the formation of different kind of planets and test actual theoretical formation models (Laughlin et al. 2004, Kornet & Wolf 2006, Kornet et al. 2006)

Our compiled sample of young MG members will be targeted in the search for lower mass companions by imaging techniques. Their proximity allows the exploration of the faint circumstellar environment at relative small distances from the star and, since substellar companions cool and fade with time, targeting young systems will give higher probability of detection compared to typical older field objects. From the complete sample, our youngest closest MG members (in the IC 2391 or Pleiades MGs) would allow AO to probe to 1M_J at separations of $\sim$1AU using for example NACO (Nasmyth Adaptive Optics System (NAOS), Near-Infrared Imager and Spectrograph (CONICA)) at the VLT.