M dwarf spectral classification L dwarf and T dwarf spectra L dwarf characteristics - temperature and mass
M dwarf spectral classification
L dwarf and T dwarf spectra
L dwarf characteristics - temperature and mass
Somewhere between 60 and 75% of the total flux emitted by late-type M and L dwarfs is concentrated in the near-infrared, at wavelengths between 1 and 2.5 microns. Most of the remaining energy is emitted between 2.5 and 5 microns. In the case of T dwarfs, the methane bands at H and K eliminate approximately 50% of the flux from those bands, with much of the radiation redistributed to the 4.5-micron window. Thus, infrared photometry provides an effective means of both identifying and studying ultracool dwarfs.
General information on broadband colours of late-type (M, L, T) dwarfs and compilations of data for parallax stars are provided on the colour-magnitude diagrams page.
Figure L3.1: Atmospheric transmission at near-infrared wavelengths: the absorption features at 1.1, 1.4, 1.85 and 2.5 microns are all due to water; the dotted lines mark the transmission functions of the 2MASS J, H and KS filters.
The three primary near-infrared passbands, J (effective wavelength 1.256 microns), H (1.633 microns) and K (2.21 microns), were chosen to match the relatively transparent windows between the strongest H2O. The original infrared broadband photometry system was devised by Johnson (1964), and included only the J and K at shorter wavelengths, together with the L, M and N passbands (discussed further below). The H passband was not added until 1968 (Johnson et al, 1968). Johnson's original observations of K and M dwarfs includes JKL photometry for only 17 stars (Johnson, 1965).
Figure L3.2: Ultracool dwarfs at near-infrared wavelengths - photometric passbands. The lower panel plots the Cousins I-band and the standard CIT JHK bands; the upper panel plots Cousins I and 2MASS JHK passbands.
The original definition of the Johnson JHK system was followed by number of minor variations on that theme, with the CIT system being probably the best defined variant before 2MASS. Bessell & Brett (1988) summarise the main properties of that system and some of the alternatives, while Carpenter (2001) has compared most of the major systems against 2MASS. The major differences in the last-mentioned system lie in the J-band, which is broader, and the K band, where the shorter-wavelength red cutoff (reducing the contribution from re-radiated thermal emission) moves the effective wavelength to ~2.15 microns. The latter filter is known as the KS filter. Figure L3.2 superimposes the CIT and 2MASS filter transmission curves on spectrophotometric data for late-type L dwarfs. T major intrinsic absorption features lie outwith the broadband filter curves, and observations show little evidence for significant colour terms between the two systems.
Figure L3.3: T dwarfs at near-infrared wavelengths - photometric passbands. The upper panel superimposes the Cousins R- and I-band over far-red optical data for SDSS1624, while the lower panel plots the 2MASS JHK passbands and near-infrared data for Gl 229B.
Figure L3.3 matches the same far-red and near-infrared filters against T dwarf spectrophotometry. The dramatic effect produced by the methane bands at near-infrared wavelengths is obvious, but note also the broad potassium absorption overlapping with most of the I-band at far-red optical wavelengths.
Figure L3.4: near-infrared colour-magnitude and colour-colour diagrams for M, L and T dwarfs. Crosses are M dwarfs - magenta, Hipparcos parallaxes; yellow, 8-parsec sample; cyan, USNO observations. Solid green points are L dwarfs; red 5-point stars are T dwarfs.
The effect of these changing bandstrengths on the broadband colours is illustrated in Figure L3.4, which plots far-red and near-infrared dara for M, L and T dwarfs. In (MJ, (I-J)), the main-sequence shows several distinct changes in slope, notably at (I-J)~1.5, (I-J)~3.5 and, probably, (I-J)~4.5.
Figure L3.5: the upper panels plot (I-J) and (J-K) as a function of spectral type; the lowermost panel plots J-band bolometric as a function of spectral type.
Both broadband colours and spectral types are correlated primarily with effective temperature. Hence one expects a correlation between these empirical parameters. Figure L3.5 shows the results for (I-J) and (J-K); the general trend is towards redder colours at later types, but, particularly in (J-K), the dispersion at later types exceeds the photometric and spectral typing uncertainties. The origin of this dispersion remains unclear, but probably reflects a range of molecular absorption at a given optical spectral type - H2 in the K-band and water in the J-band are the most likely culprits (perhaps gravity differences?). More near-infrared spectra are required to further illuminate this issue.
Bolometric magnitudes provide the primary link between observations and theoretical models. Near-infrared magnitudes, measured close to the peak in the flux distribution, offer a relatively reliable means of estimating the total luminosities. The offset between the observed J magnitude and mbol, the bolometric correction, is defined here as
Figure L3.5 shows that ultracool dwarfs are typically 2 magnitudes brighter at J than in the bolometric frame. In absolute terms, the magnitude range spanned is as follows:
|Dwarf||Spectral type||MI||MJ||MK||Mbol||log (L/LSun)||Comments|
|2M0746+20||L0.5||15.22||11.84||10.60||13.85||-3.67||Known near-equal mag. binary; mags. adjusted to single dwarf|
|Denis1228||L5||17.44||13.59||11.96||15.25||-4.24||Known near-equal mag. binary; mags. adjusted to single dwarf|
|Denis0205||L7||17.71||13.86||12.33||15.60||-4.37||Known near-equal mag. binary; mags. adjusted to single dwarf|
Ground-based observations longward of the 2.2-micron K band become increasingly difficult due to both atmospheric absorption and the higher sky background produced by re-radiated thermal emission. Spaceborne observations avoid this problem, but neither IRAS nor ISO had sufficient sensitivity to detect ultracool dwarfs. The forthcoming SIRTF mission should provide sigificantly improved data in the 3-20 micron range; until then, we are restricted to a relatively small set of ground-based observations.
Figure L3.6: Atmospheric transmission between 3 and 5.5 microns. The green dotted line plots the typical transmission at Mauna Kea; the blue dotted lines mark the original Johnson L and M bands; the magenta solid lines are the transmission curves of filters in the more recently adopted L' and M' system.
Moving longward of the K-band, the next terrestrial windows are centred on 3.8 and 4.8 microns - the Johnson L and M bands (Figure L3.6). The original L-band observations were made using PbS detectors, and their decline in sensitivity longward of 3.5 microns limited observations to the shorter-wavelength section of the L wind. Once InSb detectors became available, that mismatch could be corrected, and the L' passband, effective wavelength = 3.80 microns, was defined. Slightly broader than the L band, the L' band is better matched to the atmospheric transmission curve. This passband has been the standard since the mid-1980s, albeit with subtle variations between individual systems.
The situation at 5 microns is more complicated, since terrestrial absorption is present to some extent over the full wavelength range. However, it is clear that the Johnson passband can be improved. At the IRTF, this goal has been achieved by defining a narrowband-M system, the M' system, with effective wavelength 4.6 microns and FWHM~0.24 microns (the MKO NIR system: Tokunaga \& Simons, 2001). As Figure L3.6 shows the passband is centred in the optimum region of the 5 microns window.
Figure L3.7: (K-L') and (L'-M) colours as a function of spectral type. The blue line plots the colours predicted by the Allard et al. DUSTY models.
While L-band observations can be obtained for most of the ultracool dwarfs discovered by 2MASS and DENIS, only the brightest dwarfs are accessible to M-band observations. The available observations are summarised by Reid & Cruz (2001): Berriman & Reid (1987) obtained data for a handful of early- and mid-type M dwarfs; Leggett et al (2002) publsihed photometry of two L and two T dwarfs; and Reid & Cruz themselves observed half a dozen mid- and late-type M dwarfs, and 2M0746. The results are plotted in Figure L3.7, which also shows a comparison with the predicted behaviour based on models computed using the Allard et al DUSTY prescription. Overall, the agreement is reasonable. The (L'-M') colours show relatively little dispersion with increasing spectral type. The dotted line at (L'-M')=0.0 marks the colour expected for an object with a Rayleigh-Jeans energy distribution; thus, the fact that most of the ultracool dwarfs fall below this line indicates that the 3.5/4.8 micron flux ratio is higher than the R-J prediction. This probably reflects CO absorption in the M' band (see Reid & Cruz for further details).
Figure L3.8 plots the full energy distribution of the M9 dwarf, LHS 2065, and the L0.5 binary, 2M0746. The preponderance of flux emitted at near-infrared waveelngths is obvious.
Figure L3.9: a comparison between the Chabrier et al (2000) theoretical isochrones for the DUSTY models and observations. The three isochrones plotted are for ages of 0.5- (green), 1- (cyan) and 5-Gyrs (red).
Finally, Figure L3.9 compares the empirical distribution in the (MM', (K-M')) plane against the theoretical isochrones predicted by Chabrier et al. (2000) using the DUSTY models. The observed datapoints lie blueward of the theoretical isochrones, and analysis of individual L dwarfs (specifically, 2M0746 and 2M0036) suggests that this arises not from errors in the theoretical 5-micron fluxes, but rather from the models underestimating the K-band flux (Reid & Cruz, 2001).
NStars home page
PMSU main page
INR home page