The bimodal initial mass function in the Orion Nebula Cloud
H. Drass, M. Haas, R. Chini, A. Bayo, M. Hackstein, V. Hoffmeister, N. Godoy, N. Vogt
MMNRAS , 1–11 (2016) Preprint 13 May 2016 Compiled using MNRAS L A TEX style file v3.0
The bimodal initial mass function in the Orion Nebula Cloud (cid:63)
H. Drass , † , M. Haas , R. Chini , , A. Bayo , ,M. Hackstein , V. Ho ff meister , N. Godoy , and N. Vogt Astronomisches Institut, Ruhr-Universität Bochum, Universitätsstraße 150, 44780 Bochum, Germany Department of Electrical Engineering and Center of Astro-Engineering UC,Pontificia Universidad Católica de Chile, Av. Vicuña Mackenna 4860, 7820436 Macul, Santiago, Chile Instituto de Astronomía, Universidad Católica del Norte, Avenida Angamos 0610, Casilla 1280 Antofagasta,Chile Instituto de Física y Astronomía, Universidad de Valparaíso, Av. Gran Bretaña 1111, Valparaíso, Chile Max-Planck Institut für Astronomie, Königstuhl, D-69117, Germany
Received May, 2014; Accepted 2016 May 05
ABSTRACT
Due to its youth, proximity and richness the Orion Nebula Cloud (ONC) is an ideal testbedto obtain a comprehensive view on the Initial Mass Function (IMF) down to the planetarymass regime. Using the HAWK-I camera at the VLT, we have obtained an unprecedenteddeep and wide near-infrared
JHK mosaic of the ONC (90% completeness at K ∼ . (cid:48) × (cid:48) ). Applying the most recent isochrones and accounting for the contamination of back-ground stars and galaxies, we find that ONC’s IMF is bimodal with distinct peaks at about0.25 and 0 .
025 M (cid:12) separated by a pronounced dip at the hydrogen burning limit (0.08 M (cid:12) ),with a depth of about a factor 2–3 below the log-normal distribution. Apart from ∼
920 low-mass stars ( M < . (cid:12) ) the IMF contains ∼
760 brown dwarf (BD) candidates and ∼ M > .
005 M (cid:12) , hence about ten timesmore substellar candidates than known before. The substellar IMF peak at 0.025 M (cid:12) could becaused by BDs and IPMOs which have been ejected from multiple systems during the earlystar-formation process or from circumstellar disks.
Key words:
Stars: brown dwarfs – Stars: formation – ISM: dust, extinction – Infrared: stars
The stellar Initial Mass Function (IMF) (Salpeter 1955) describesthe mass spectrum of a stellar population at birth or in young star-forming regions shortly after birth. The origin of the IMF is afundamental issue in the study of star formation. Basically, twocompeting theories try to explain the observations: the determin-istic view postulates that the IMF is essentially determined by theCore Mass Function (CMF) in the parental cloud (Alves et al. 2007;Nutter & Ward-Thompson 2007; André et al. 2010; Hennebelle &Chabrier 2013), while the stochastic view emphasizes the impor-tance of dynamical interactions and competing accretion (Bonnellet al. 1997; Reipurth & Clarke 2001; Vorobyov et al. 2013).The high-mass end of the IMF and the peak at intermedi-ate stellar masses (0.2 – 0.5 M (cid:12) ) for various star-forming regions(e.g. Bayo et al. (2011)) support a log-normal IMF shape (Miller &Scalo 1979; Kroupa 2001; Chabrier 2005). However, at the low-mass end the frequency of Brown Dwarfs (BDs) and isolated plane-tary mass objects (IPMOs) is poorly known with large uncertainties(Bastian et al. 2010; Scholz et al. 2013). (cid:63)
Based on observations made with ESO Telescopes at the Paranal Obser-vatory under programme ID 082.C-0032-1 † E-mail: [email protected]
The Orion Nebula Cloud (ONC) located at a distance of 414 pc(Menten et al. 2007) is a benchmark for studying the IMF of youngstar-forming regions and to peer into the substellar mass regime.Even so, in the central (5 (cid:48) × (cid:48) ) region around the Trapezium StarCluster the bright irregular emission of the nebula and the highextinction hampered the detection of faint objects even at near-infrared (NIR) wavelengths. The age of young stars in the ONCis in the range of 1 – 5 Myr; here we adopt 3 Myr (Da Rio et al.2012).While in the ONC a growing number of ∼
60 BDs has beenfound in recent years by means of spectroscopy (Slesnick et al.2004; Riddick et al. 2007; Weights et al. 2009), any evidence fora rich BD and IPMO population – exceeding the log normal ex-trapolation of the stellar IMF – is still a matter of a debate. On onehand, the IMF appears to be steeply declining towards the low-mass(BD) end (Hillenbrand & Carpenter 2000; Da Rio et al. 2012). Inthese cases, however, the reported IMF decline runs below the totalnumber of spectroscopically confirmed BDs, combined from theauthors mentioned above. This indicates a problem in these pho-tometric searches. On the other hand, Muench et al. (2002) haveclaimed an upturn of the IMF in the BD mass regime, albeit at theirobservational brightness limits and therefore quite speculative. Fi-nally, Lucas et al. (2005) reported on the so far most robust IMF c (cid:13) a r X i v : . [ a s t r o - ph . GA ] M a y H. Drass et al. D E C ( J ) [ deg ] Figure 1.
Surveyed areas covered by di ff erent studies, in black the areacovered by HAWK-I used in this work, in blue the area from Robberto etal. (2010), in orange the data from Muench et al. (2002), and in green thedata from Lucas et al. (2005). The circle of 10’ diameter marks the regionaround M43. study of the ONC, which reveals a dip with potential rise of thelow-mass IMF or simply a broad IMF plateau; as pointed out byLucas & Roche (2000), imperfections in the available isochronesprevent to easily distinguish between the two IMF shapes. Opticalwide field studies covering the outer (30 (cid:48) × (cid:48) ) ONC regions su ff erfrom extinction and failed to reveal a rich BD population (Da Rioet al. 2012). Still, the substellar IMF in Orion is controversial anda comprehensive investigation employing deeper data and state-of-the-art isochrones is needed. Using the HAWK-I camera at the VLT on November 8–11, 2008,we obtained a large 22 (cid:48) × (cid:48) and deep image mosaic of the ONC inthe JHK filters centred at 1 .
25, 1 .
65 and 2 . µ m (i.e. K s here forshort denoted K ) under good seeing conditions (FWHM < . (cid:48)(cid:48) ).The central mosaic position is at RA = .
68 and Dec = − ◦ (cid:48) . (cid:48)(cid:48) J , H , K and 1 s exposure time per frame. The basic data reduction was performed with IRAF. For calibra-tion we took regular dark and flat exposures close to the observa-tions and combined them into master dark and master flat. Subtrac-tion of the master dark and division by the master flat yielded amaster sky frame that was subtracted from the science frames. The Image Reduction and Analysis Facility (IRAF)(ver. 2.14.1) is dis-tributed by the National Optical Astronomy Observatories, which are op-erated by the Association of Universities for Research in Astronomy, Inc.,under cooperative agreement with the National Science Foundation.
At the time of the observations there was no suitable distor-tion map available. To achieve a solution for the distortion cor-rection the IRAF / MSCRED package was applied. After distortioncorrection, the comparison with 2 MASS (Skrutskie et al. 2006)showed that for objects with high quality detection according tothe 2 MASS point source catalog (those labelled ‘AAA’) the di ff er-ences are within 0.2" in 98% of the sources. All photometric operations were performed on the 144 quadrantimages separately using the IRAF / DAOPHOT package. A prelim-inary source list was created with IRAF / DAOFIND. Due to thestrong and variable nebular emission across the field the parame-ters were adjusted in a way to register only sources with a highsignal-to-noise ratio (S / N > K < . / DAOFIND interac-tively. Then, in a preparation step and for later comparison, aperturephotometry with IRAF / PHOT was done. To perform point spreadfunction (PSF) photometry of all sources we selected the most iso-lated, non saturated stars in each quadrant with IRAF / PSTSELECTand created a PSF with IRAF / PSF. The routine IRAF / ALLSTARcalculated the PSF magnitudes for all sources and subtracted theirPSF from the image. The resulting image was inspected by eyeand newly appearing objects, for instance faint close companionsto a bright star, were added to the source list. After repeating thePSF photometry and source subtraction process the images werereinvestigated. This process was repeated until all sources were re-moved from the frames. In order to receive sharpness and round-ness for the sources added by hand IRAF / DAOFIND was used witha very low threshold (S / N = . / xy2RaDec (Mink 2001) task. To find aphotometry for all sources the “handmade” source list was matchedwith the low threshold sources utilizing the ALADIN / xmatch task. The absolute calibration was done by comparison with 2 MASS.Each detector quadrant was calibrated separately and only the ob-jects with the best quality flags ‘AAA’ fainter than 12 mag whereused. First the average di ff erence and the mean deviation for the dif-ference between the magnitudes from 2 MASS and HAWK-I werecalculated. This procedure yielded a couple of sources which de-viated significantly from the mean standard deviation. Visual in-spection of these cases showed that the deviation is well explainedby the reduced spatial resolution of 2 MASS compared to HAWK-Iand the increased photometric errors in nebular regions. These out-liers have been excluded from the calibration. The remaining dif-ferences were averaged a second time and added to all instrumentalmagnitudes.Additionally, since 2 MASS and HAWK-I are using di ff erentfilter systems, we investigated the color term. The comparison be-tween the colors ( J − K ), ( J − H ), ( H − K ) for HAWK-I and 2 MASSare plotted in Fig. 2. MNRAS000
At the time of the observations there was no suitable distor-tion map available. To achieve a solution for the distortion cor-rection the IRAF / MSCRED package was applied. After distortioncorrection, the comparison with 2 MASS (Skrutskie et al. 2006)showed that for objects with high quality detection according tothe 2 MASS point source catalog (those labelled ‘AAA’) the di ff er-ences are within 0.2" in 98% of the sources. All photometric operations were performed on the 144 quadrantimages separately using the IRAF / DAOPHOT package. A prelim-inary source list was created with IRAF / DAOFIND. Due to thestrong and variable nebular emission across the field the parame-ters were adjusted in a way to register only sources with a highsignal-to-noise ratio (S / N > K < . / DAOFIND interac-tively. Then, in a preparation step and for later comparison, aperturephotometry with IRAF / PHOT was done. To perform point spreadfunction (PSF) photometry of all sources we selected the most iso-lated, non saturated stars in each quadrant with IRAF / PSTSELECTand created a PSF with IRAF / PSF. The routine IRAF / ALLSTARcalculated the PSF magnitudes for all sources and subtracted theirPSF from the image. The resulting image was inspected by eyeand newly appearing objects, for instance faint close companionsto a bright star, were added to the source list. After repeating thePSF photometry and source subtraction process the images werereinvestigated. This process was repeated until all sources were re-moved from the frames. In order to receive sharpness and round-ness for the sources added by hand IRAF / DAOFIND was used witha very low threshold (S / N = . / xy2RaDec (Mink 2001) task. To find aphotometry for all sources the “handmade” source list was matchedwith the low threshold sources utilizing the ALADIN / xmatch task. The absolute calibration was done by comparison with 2 MASS.Each detector quadrant was calibrated separately and only the ob-jects with the best quality flags ‘AAA’ fainter than 12 mag whereused. First the average di ff erence and the mean deviation for the dif-ference between the magnitudes from 2 MASS and HAWK-I werecalculated. This procedure yielded a couple of sources which de-viated significantly from the mean standard deviation. Visual in-spection of these cases showed that the deviation is well explainedby the reduced spatial resolution of 2 MASS compared to HAWK-Iand the increased photometric errors in nebular regions. These out-liers have been excluded from the calibration. The remaining dif-ferences were averaged a second time and added to all instrumentalmagnitudes.Additionally, since 2 MASS and HAWK-I are using di ff erentfilter systems, we investigated the color term. The comparison be-tween the colors ( J − K ), ( J − H ), ( H − K ) for HAWK-I and 2 MASSare plotted in Fig. 2. MNRAS000 , 1–11 (2016) he bimodal initial mass function in the Orion Nebula Cloud M ASS c o l ou r [ m ag ] Figure 2.
Investigation of a color term correction for the 2 MASS andHAWK-I filter combinations. The green line’s slope is 1 and intersects atzero. The red, blue, orange line marks the fitted solution for ( J − K ),( J − H ),( H − K ), respectively.The black, blue, yellow crosses show the colors for( J − K ), ( J − H ), ( H − K ), respectively. The results of a linear fits are given in the following equations.( J − Ks ) MAS S = (0 . ± .
01) ( J − K s) HAWK − I − (0 . ± . J − H ) MAS S = (0 . ± .
01) ( J − H ) HAWK − I − (0 . ± . H − Ks ) MAS S = (0 . ± .
03) ( H − K s) HAWK − I + (0 . ± . . < ( J − K ) < J − H ) and ( H − K ) are similar. In summary,we did not apply any color term correction. The resulting catalogs of all calibrated objects in J , H , and K con-tain 7975, 9630, and 9899 sources, respectively. To exclude anyfalse detections like artefacts or poorly determined sources, a re-fined selection based on the search and photometry parameters wasperformed. In a first step the sources were conservatively selectedby their sharpness and roundness (ratio of best fitting Gaussians)derived from aperture photometry and the sharpness calculated dur-ing the PSF photometry by having a deviation less than five timesthe mean absolute deviation from the corresponding average. Byeye inspection of the rejected sources showed that they were eithersaturated or extremely faint; other rejected sources were simplyobservational artefacts in the frame. For a second selection levelthe magnitude errors were fitted and quality flags (QF) for the fit(chi), s-roundness, g-roundness, PSF-sharpness and magnitude er-ror were created. The corresponding values for the quality flags areplotted in Fig. 3. Sources with the following properties were re-jected: • brighter than 12 . • magnitude errors at the object brightness larger than ten timesthe fitted average magnitude error • magnitude errors greater than 0 . • quality flag worse than 3 for the PSF- sharpnessBy analyzing the di ff erent quality flags it turned out that the combi-nation of several flags helps to find bona fide objects. Therefore thesum of the quality flags for chi, s-roundness, g-roundness, PSF-sharpness and magnitude error was calculated and only sourceswith an empirically determined value less than 13 were accepted. The mosaic was arranged with a su ffi cient overlap between the ad-jacent fields. Objects located in overlapping regions were thus mea-sured up to four times. In order to obtain only one measurementper source each identification in a radius of 0 . (cid:48)(cid:48) ff erence betweenthe two magnitudes is calculated. If the di ff erence is smaller than0.15 mag or smaller than ten times the fitted mean value of all ob-jects in the magnitude bin ∆ m = . ff erence be-tween all three identifications was first considered with the samecriteria as above. If all three identifications fulfil the conditions,their average is used as the best value. If not, the record with theworst magnitude error is rejected and the criteria are reconsidered.In case of acceptance, the mean is evaluated and used as the bright-ness of this object. The remaining identifications are then inves-tigated for their relative distance from the border as in the case oftwo records. Also acceptance and rejection is done in the same way.This solved the ambiguities for all objects.Finally, when an object was recorded four times the averageis calculated and only those two identifications with the smallestdi ff erence relative to the average were considered further on. Thenthe di ff erence between these remaining values was considered inthe same way as above. We found that no further selection criteriawere necessary. For these objects again the mean of the measure-ments was used as the final magnitude.The resulting master catalog contains all reliable sourcesfainter than 12 . K = . . J and H . The comparison presented in Fig. 4 demonstrates that theHAWK-I data reach 2 mag deeper than the previous data, whichstart to su ff er from incompleteness in the BD regime (vertical linesin Fig. 4). The total number of sources in the added catalog is 4340.All photometric errors are smaller than 0 . (cid:48) diametercentred on the nebula M43, because it is a distinct cluster with itsown, possibly di ff erent, mass function. The HAWK-I FoV is almostcovered by the data from Robberto et al. (2010). For a consistentliterature comparison only sources in the common field of view arefurther analysed. The resulting catalog for all sources measured inall three filter is given in Table A1. MNRAS , 1–11 (2016)
H. Drass et al. N u m be r o f O b j e c t s −10 −5 0 5 10PSF−sharpness101001000 N u m be r o f O b j e c t s −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5g−roundness101001000 N u m be r o f O b j e c t s −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5s−roundness101001000 N u m be r o f O b j e c t s Figure 3.
Quality flag plots. For the chi, PSF-sharpness, g-roundness the numbers give the flag value in shown range. For the s-roundness the limits are filterdependent. (Black, red, green are the average plus the mean standard deviation for J , H , and K , respectively.) J Filter
13 15 17 19 21 020406080100 C o m p l e t ene ss [ % ] inner areamiddle areaouter area H Filter
13 15 17 19 21 Brightness [mag]
K Filter
13 15 17 19 21
Figure 5.
Completeness function for
JHK in the inner (radius ( r ) < (cid:48) , solid curve), intermediate (4 (cid:48) < r < (cid:48) , dotted curve) and outer region (8 (cid:48) < r < (cid:48) ,dashed curve). Excluded is the area of 5 (cid:48) radius around the nebula M43. MNRAS000
JHK in the inner (radius ( r ) < (cid:48) , solid curve), intermediate (4 (cid:48) < r < (cid:48) , dotted curve) and outer region (8 (cid:48) < r < (cid:48) ,dashed curve). Excluded is the area of 5 (cid:48) radius around the nebula M43. MNRAS000 , 1–11 (2016) he bimodal initial mass function in the Orion Nebula Cloud N u m be r o f ob j e c t s Figure 4. K -band luminosity functions for the HAWK-I data (black) com-pared to Robberto et al. (2010) (blue) and Muench et al. (2002) (yellow)without completeness correction applied. The error bars correspond to √ N .The vertical dashed and dash dotted lines mark the completeness limits ofRobberto et al. (2010) at 90% and 70%, respectively.Filter inner area middle area outer area J H K Table 1.
90% completeness limits in mag.
To determine the completeness of our data set we used the routineIRAF / ADDSTAR to randomly add artificial stars down to 22 mag.Than we ran IRAF / DAOFIND with a 0 . σ threshold, which – ac-cording to our tests – corresponds to the by eye selection for the realsources. For the photometry IRAF / PHOT and IRAF / ALLSTARwere utilized. The x y − coordinates of the artificial stars and de-tected candidates were transformed to RA and Dec by using WC-STOOLS / xy2sky. In order to find only the artificial stars a crossmatch using ALADIN / XMATCH with the original list of artificialstars was performed. To ensure that only well measured objectswere re-discovered only objects with an error smaller than 0.1 magand a di ff erence between injected and re-discovered object smallerthan 0.5 mag were accepted. In Fig. 5 the completeness obtainedfrom the artificial star experiment is shown; the 90% completenesslimits are listed in Table 1. Fig. 6 displays the color-color-diagram for sources measured inthree filters, excluded are sources around M43 as mentioned before.The color-color-diagram shows that only a small fraction ( < K -band excess from cir-cumstellar material. Beside objects exhibiting K -band excess also ( J − H ) [ m ag ] Figure 6.
Color-Color Diagram for sources measured in three filters (greydots), excluded are sources around M43. The black dash-dotted curve isthe main sequence from Ducati et al. (2001) while the black solid curveindicates the 3 Myr isochrone from Bara ff e et al. (2015) and Allard et al.(2013). The straight line is the locus of the classical TTauri stars (Meyer etal. 1997); the reddening vector for A V =
10 is also shown (Rieke & Lebofsky1985). objects without excess can be members. Actually, most objects donot show a K -band excess. Figure 7a presents the K / ( J − K ) CMD of the observed sample. The3 Myr isochrone from Bara ff e et al. (2015) holds for a mass range of1.4 – 0.01 M (cid:12) . To cover the full substellar range as far as possible,the 3 Myr isochrone from Allard et al. (2013) was used to extendthe mass range to 0.003 M (cid:12) . Allard’s isochrone accounts for NIRspectral features of substellar objects (e.g. Canty et al. (2013)),which were not taken into account in former isochrones.We display the A V = / older isochrones virtually coincidewith the 3 Myr isochrone, except that for a given mass the ob-ject’s brightness declines with age; but the di ff erences of the 1 and5 Myr isochrones compared to the 3 Myr isochrone are small ( ∆ K < . . < K < . A preliminary analysis of Spitzer / IRAC photometry (3.6 µ m and 4.5 µ m)shows IR excess for 30% of the objects brighter than K =
13 mag. This isconsistent with an age of about 3 Myr. Because of the low irradiation powerno excess at shorter wavelength can be expected. For fainter objects theIRAC photometry becomes incomplete. Allard et al. (2013), CIFIST2011bc – Model, 2 MASS filter set, Vegamagnitudes, see appendix BMNRAS , 1–11 (2016)
H. Drass et al. K [ m ag ] (a) Color-magnitude diagram K vs J − K of the HAWK-I data (black dots).The thick blue, red, black and orange curves show the 1 Myr, 2 Myr, 3 Myrand 5 Myr isochrones (Bara ff e et al. 2015; Allard et al. 2013), respec-tively. All isochrones are shown for masses between 1.4 M (cid:12) and 0.003 M (cid:12) .The arrows are extinction vectors of length A V = (cid:12) ) and lower (0.012 M (cid:12) ) mass limit of browndwarfs. The dashed line at J − K < . A V < K [ m ag ] (b) Color-magnitude diagram K vs J − K of the known spectroscopicallyverified brown dwarfs (red dots) and background stars from the Besançonmodel (yellow dots) and background galaxies from UKIDSS ultra-deep field(blue crosses). The CMD position of the background objects is shown with-out reddening by the Orion dust screen. The rest of the symbols is the sameas in the left diagram. Figure 7.
Color-magnitude diagrams
While the isochrones by Bara ff e et al. (2015) is being exten-sively tested against observations in the brown dwarf regime, theisochrone from Allard et al. (2013) may be somewhat uncertain inthe mass range of IPMOs as indicated by the distinct CMD struc-tures at K > . J − K ∼ . A V vector starting atthe upper BD mass boundary of the isochrone (M = (cid:12) , at K = . J / ( J − H ) , J / ( J − K ) , J / ( H − K ), etc.). The large over-lap in magnitude and the good agreement between the HAWK-I andthe Robberto et al. (2010) catalogs (Fig. 4) support the idea that thedeficiency-stripe is not an artefact caused by the merge of the twocatalogs.At the distance of Orion (414 pc), contamination of our sam-ple by foreground objects unrelated to the Orion complex is neg-ligible; less than 150 objects are expected from the galactic modelof Robin et al. (2003). Regarding populations related to the Orioncomplex, Alves & Bouy (2012) and Bouy et al. (2014) found 2123foreground candidate members in about 10 square degrees towardsOrion including the clusters NGC 1980 and NGC 1981. After com-paring with the HAWK-I mosaic, we have identified and excluded45 sources belonging to the foreground cluster.Background stars and galaxies are expected to influence thesource counts significantly. For the following analysis we are us-ing an extinction and magnitude limited sample. The data are lim-ited to J − K < . A V < K = . J − K > .
6, i.e. IP- MOs below 0.006 M (cid:12) , it enables a proper estimate of the contam-ination by background stars from the Besançon model and back-ground galaxies from UKIDSS ultra-deep field in the Brown dwarfmass regime. This is already a conservative assumption, since inShimajiri et al. (2011) and Ripple et al. (2013) the extinction mapsshow that the extinction in Orion can be much larger.
Since the targeted sources are very faint, so far, there is no mem-bership criterion for each source available. Instead, we take advan-tage of the fact that the background sources su ff er heavily fromextinction by the Orion nebula cloud and use this fact to correct forcontamination of the IMF by background objects. To determine themass function, the sources are shifted (dereddened) in the CMD tothe isochrone along the direction of the AV vector. Note that thisdereddening changes the resulting luminosity functions (LFs), be-cause highly reddened source now become brighter.At each position of the HAWK-I map, the total extinction A V through the cloud can be determined from CO observations. Weused the CO map with a 7 . (cid:48)(cid:48) pixel size (Shimajiri et al. 2011)and the CO map with 20 (cid:48)(cid:48) pixel size (Ripple et al. 2013) andproduced two extinction maps with consistent results.Our HAWK-I data constitute an extinction limited samplewith A V < J − K < . A V < .
04 square degrees are transparent for background ob-jects. The background contamination consists of two components,stars and galaxies.We used the Besançon model of our Galaxy (Robin et al.2003) to create a sample of background stars in a cone of 0 . MNRAS000
04 square degrees are transparent for background ob-jects. The background contamination consists of two components,stars and galaxies.We used the Besançon model of our Galaxy (Robin et al.2003) to create a sample of background stars in a cone of 0 . MNRAS000 , 1–11 (2016) he bimodal initial mass function in the Orion Nebula Cloud Orion (at galactic latitude − . K / ( J − K ) CMD. Note that in Fig. 7a thetotal observed population is naturally reddened by the molecularcloud, but the model predictions for background stars (yellow dotsin Fig. 7b) do not yet include any reddening by the Orion nebulacloud. As a consequence the unreddened background stars are lo-cated in the CMD at their intrinsic colors. The few stars at J − K < . . < J − K < . J − K = .
89. In practice, due to reddening by theONC, they would be moderately reddened ( A V < A V vector. If they were reddenedby more than A V ∼ J − K = . H / ( H − K ) etc.).To reach our aim to find the LF of the ONC, the contaminatingbackground stars need to be subtracted from the total LF. There-fore, each star is moved along the reddening vector until it reachesthe isochrone and the magnitude of the intersection is assigned toit. To estimate the contribution from background galaxies, we usethe UKIDSS ultra-deep J and K data set of the ELAIS-north field,a region at high galactic latitude with very low extinction. Fig. 7bshows the location of the predicted background galaxies in the K / ( J − K ) CMD, without any reddening by the Orion nebula cloud.The intrinsic colors of the galaxies su ff er from a redshift-dependentK-correction, i.e. they display redder J − K colors compared to thebackground stars. The galaxies fainter than K = . ff ectonly the isolated planetary mass objects below 0 . (cid:12) , whereour J − K = . K = . J − K = . K = . J − K = .
6. About 6mag of extinctionshifts such a galaxy out of our sample. As a conservative estimatefor galaxies contributing as background contamination, we adopta reddening of the cloud with less than A V = A V < .
02 square degrees. We scaled the number of predicted back-ground galaxies to this area ( N g alaxies ∼ H -filter, we adopt intrinsic colors H − K = . · ( J − K ); we find thatthe resulting IMFs are similar for H − K = . · ( J − K ) and H − K = . · ( J − K ) as well, hence not very sensitive to the assumption onthe galaxies’ colors.As for the background stars, each galaxy is also moved alongthe reddening vector and the magnitude of the intersection is as-signed to it. In Sec. 5 we subtract the contamination caused by thegalaxies and the background stars from the ONC data in order tofind the LF of ONC members. Starting from the LF of all sources in the field we first subtractthe LF for reddened stars and then subtract the LF for reddenedgalaxies. Fig. 8 displays the LF of both total observed populationand predicted background sources shifted (dereddened) along thedirection of the extinction vector to the 3 Myr isochrone. Note thatthis dereddening changes the resulting LFs, because highly red-
10 12 14 16 18 200100200300 10 12 14 16 18 20K [mag]0100200300 N u m be r o f O b j e c t s Figure 8.
K-band luminosity function for the total HAWK-I data is shownin green, for background stars in yellow, and background galaxies in blue.Data and background objects are shifted to the isochrone along the A V vec-tor. The black line shows the total HAWK-I data minus background stars + galaxies. The vertical dash-dotted lines mark the luminosity boundaries ofbrown dwarfs. dened source now become brighter. The black curve shows the de-contaminated LF. It exhibits a remarkably clear dip at the BD limitand a peak in the BD range at K ∼
16 mag. Beyond the BD range( K > . A V vector).Beside stellar maximum as well known from Muench et al.(2002) and Robberto et al. (2010) (Fig. 4) there is a second rise intothe substellar regime peaking at K =
16 mag.The surveyed areas of Robberto’s, Muench’s and our analysisare very di ff erent (see, Fig. 1). To compare the relative di ff erencesbetween the KLF – features the LF normalized to the total num-ber of sources in each sample is presented in Fig. 9. The feature inthe substellar range is about 30% higher than the stellar maximumfor the HAWK-I data, while for Robberto and Muench the ratio isvice versa. The maximum in the stellar mass range has about 25%more sources for Robberto and even ∼ ff ering from. Therefore, both analysis can not presentthe same results as found here in this analysis. From the directcomparison between the data from Robberto et al. (2010) and theHAWK-I data set, the 90% completeness limit for the Robberto etal. (2010) is at ∼
17 mag (Fig. 4, dashed line). At the Brown Dwarflimit ( ∼
18 mag) Robberto presents a completeness 70% ( ∼ ∼
370 sources are detected. Hence the com-pleteness at K =
18 mag is only about 30%. Robberto’s estimationwas done by recovering artificially introduced stars. These mightbe easier to recover than the actual object and therefore yield a toohigh completeness.To test the consequences of a possible K -band excess on theKLF, in Fig. 10 the dashed green line shows the KLF of all oursources shifted by + . H − K ) (average excess from Muenchet al. (2002)). This probably overestimates the apparent K -band ex-cess, since also all potential background sources are shifted. Nev-ertheless, the KLF appears shifted by about 0.5 mag to fainter mag-nitudes but maintains essentially the same feature shape. MNRAS , 1–11 (2016)
H. Drass et al. N u m be r o f ob j e c t s Figure 9. K -band luminosity functions normalized to the total number ofsources in each sample. The HAWK-I data are marked in black and com-pared with Robberto et al. (2010) (blue) and Muench et al. (2002) (yel-low). For all samples the original values are shown. No de-reddening, i.e.no shifting to the isochrone, no completeness correction and no foregroundor background subtraction was done. The vertical dashed and dash dottedlines mark the completeness limits of Robberto et al. (2010) at 90% and70%, respectively.
10 12 14 16 18 200100200300 10 12 14 16 18 20K [mag]0100200300 N u m be r o f O b j e c t s Figure 10.
K-band luminosity function comparisons. The total HAWK-Idata, shifted to the isochrone, are shown in green. The black line shows thetotal, shifted HAWK-I data minus background stars + galaxies using thelatest model from Bara ff e et al. (2015) and Allard et al. (2013). The darkgreen dashed line presents the data shifted in K by 0 . K -band excess. The orange line shows the KLF using the Isochronefrom D’Antona & Mazzitelli (1998) and the purple line presents the KLFusing the Isochrone from Bara ff e et al. (1998). To analyse the influence of the isochrone used here, we com-pare older isochrones from Bara ff e et al. (1998) and D’Antona& Mazzitelli (1998) as used in Da Rio et al. (2012). The resultis presented in Fig. 10. These shape is very similar but the olderisochrones do not cover the complete BD range and are thereforenot considered in the following analysis. In order to compute the IMF of the ONC, we subtract the back-ground contamination in the mass space.Mathematically the subtraction of the “ mimicked ” back-ground mass function (MF) is equivalent to constructing the LFsand subtracting the background contamination from the LF andthen bin the mass function. However, the second procedure requirestwice binning steps, one for the LF and one for the MF. The advan-tage of subtraction of the mimicked background mass function isthat binning is required only once and that shot noise from the dou-ble binning is reduced.Together with the luminosity each potential (by potential wemean every detection) ONC member objects a mass was assignedaccording to the used isochrone. This procedure was applied to thenine CMD combinations, eventually deriving nine mass distribu-tions. These distributions have been converted into individual IMFsby counting the number of sources in bins of 0.56 dex in log(M)space. Finally we constructed an average IMF. Despite some re-dundancy of the CMDs, dispersion of the individual IMFs allowsto get a view of the uncertainties relative to deriving masses withthe same models but di ff erent data. Using the entire extinction lim-ited ( A V =
10 mag) HAWK-I data set yields the “ total IMF ” asshown in Fig. 11.The “ total IMF ” contains not only ONC members but alsobackground objects which are reddened by the ONC dust screen.While assigning luminosity to the background stars and galaxies asdescribed in Sec. 4 also a mass corresponding to the used isochronewas recorded for each object. When creating the “ total IMF ” fromthe data, such background objects are erroneously classified asONC members. To get rid of the background contamination in theIMF, we determine the “ mimicked ” background IMF from the pre-dicted background sources shown in Fig. 7b by binning them in thesame way as the data and then subtract them from the “ total IMF ”.Note that our aim is not to reconstruct the true type or mass of thebackground objects, rather we need to calculate their contributionto ONC’s IMF. We created the “ mimicked ” background IMF forthe background stars and galaxies, for each of the nine CMD com-binations as we did for the total IMF. For both the background starsand the galaxies the resulting “ mimicked ” IMFs shown in Fig. 11are steeply rising with decreasing mass and remarkably smooth. Inthe same way as for the member LF for the ONC, the subtractionof the “ mimicked ” background IMFs from the total IMF results inthe ONC member IMF that is shown in Fig. 12. Summarizing ourapproach to deal with contaminants, we shift the simulated popu-lations (both background stars and galaxies) along the A V vector tothe isochrone to treat them in the same way as the objects in thecloud.In order to take into account propagation of the photometricerrors in the mass estimations, we run a Monte-Carlo Simulationover a member candidate population determined by making ex-tended use of the extinction maps based on Shimajiri et al. (2011)and Ripple et al. (2013) (paper in preparation).The total extinction caused by the cloud can be used to dis-tinguish foreground stars from objects embedded in the star form-ing region or from background objects. For each object we com-pare the total extinction - as derived from CO measurements - withthe extinction along the line of sight to individual objects - as de-rived from the color-magnitude diagram. Objects behind the cloudsu ff er the same or even higher extinction than what is obtainedfrom CO measurements. Objects within the cloud and foregroundstars should display lower extinction values than the total extinction MNRAS000
10 mag) HAWK-I data set yields the “ total IMF ” asshown in Fig. 11.The “ total IMF ” contains not only ONC members but alsobackground objects which are reddened by the ONC dust screen.While assigning luminosity to the background stars and galaxies asdescribed in Sec. 4 also a mass corresponding to the used isochronewas recorded for each object. When creating the “ total IMF ” fromthe data, such background objects are erroneously classified asONC members. To get rid of the background contamination in theIMF, we determine the “ mimicked ” background IMF from the pre-dicted background sources shown in Fig. 7b by binning them in thesame way as the data and then subtract them from the “ total IMF ”.Note that our aim is not to reconstruct the true type or mass of thebackground objects, rather we need to calculate their contributionto ONC’s IMF. We created the “ mimicked ” background IMF forthe background stars and galaxies, for each of the nine CMD com-binations as we did for the total IMF. For both the background starsand the galaxies the resulting “ mimicked ” IMFs shown in Fig. 11are steeply rising with decreasing mass and remarkably smooth. Inthe same way as for the member LF for the ONC, the subtractionof the “ mimicked ” background IMFs from the total IMF results inthe ONC member IMF that is shown in Fig. 12. Summarizing ourapproach to deal with contaminants, we shift the simulated popu-lations (both background stars and galaxies) along the A V vector tothe isochrone to treat them in the same way as the objects in thecloud.In order to take into account propagation of the photometricerrors in the mass estimations, we run a Monte-Carlo Simulationover a member candidate population determined by making ex-tended use of the extinction maps based on Shimajiri et al. (2011)and Ripple et al. (2013) (paper in preparation).The total extinction caused by the cloud can be used to dis-tinguish foreground stars from objects embedded in the star form-ing region or from background objects. For each object we com-pare the total extinction - as derived from CO measurements - withthe extinction along the line of sight to individual objects - as de-rived from the color-magnitude diagram. Objects behind the cloudsu ff er the same or even higher extinction than what is obtainedfrom CO measurements. Objects within the cloud and foregroundstars should display lower extinction values than the total extinction MNRAS000 , 1–11 (2016) he bimodal initial mass function in the Orion Nebula Cloud Obj /M O • N u m be r o f O b j e c t s Obj /M Jup
Figure 11.
Mass functions. The symbols correspond to a single color-magnitude diagram, and the thick lines show the average. Total HAWK-Idata (green), background stars (yellow), and background galaxies (blue).The vertical dash-dotted lines mark the mass boundaries of brown dwarfs. throughout the cloud. This procedure assumes that there does notexist a significant contribution of circumstellar extinction due to anopaque disk which would probably remain undetected by the COmeasurements. Foreground objects are discarded using the analysisfrom Alves & Bouy (2012) and Bouy et al. (2014). This methodis limited by the lower resolution of the CO extinction maps andmight therefore result in slightly di ff erent number counts.For each data point we obtaining 100 realizations.This pro-vided us with individual masses and uncertainties for everyisochrone used. We used four di ff erent isochrones: 1 , , ff ect of the age in the derived massfunction, and finally, to be phase independent in the representationof such mass function, we estimated the Kernel Density Estimatorof each mass distribution (for each isochrone, where we are normal-izing to make them directly comparable and also with ONC mem-ber mass distribution of Fig. 12), with its 99% confidence level. Theresults using the Monte-Carlo Simulation together with the KernelDensity Estimator are shown in Fig. 13. In comparison with count-ing the number of objects in each mass bin the results are similar. The striking features of the ONC member IMF (Fig. 12) is the pres-ence of two peaks at about 0.25 and 0.025 M (cid:12) separated by a pro-nounced dip at the hydrogen burning limit (0.08 M (cid:12) ), which cor-responds to the zone of low object density already seen in Fig. 7a.The IMF extends into the mass regime of IPMOs ( < . (cid:12) ).The IMF contains 929 stars with M < . (cid:12) , 757 BD candidatesand 158 IPMO candidates with M > . (cid:12) , hence indicates ahigh fraction ( ∼ ff erent CMDs. The dispersion isdisplayed by the points in the IMF. For the Orion members it isminimal at 0 . (cid:12) with about ±
10% and maximal at 0 . (cid:12) withabout + / − Obj /M O • N u m be r o f O b j e c t s Obj /M Jup
Figure 12.
Initial mass functions. The symbols (open and filled) correspondto a single color-magnitude diagram, and the thick solid lines show the av-erage. Members of the entire Orion Nebula Cloud (black) were calculatedfrom Fig. 11 as the di ff erence between the total IMF and the IMF of back-ground stars and galaxies. The green solid line gives the mass function of asmall subarea containing the known BDs South and West of the Trapezium,where no background contamination is expected. The red solid line givesthe mass function of the known spectroscopically verified BDs located inthe subarea and shown as red dots in Fig. 7b. Vertical dash-dotted lines markthe mass boundaries of BDs. For comparison other IMFs are shown: blackdashed line: Chabrier (2005) standard IMF; Da Rio et al. (2012) red dashedline which – inconsistently – falls below the IMF of the known BDs (redsolid line); Muench et al. (2002) yellow solid line with thin black crossesand shaded area giving the error range, note the abrupt upturn in the lastbin; Lucas et al. (2005) blue solid line with thin black plus symbols. extinction of the ONC. To entirely remove the substellar IMF peakone needs to increase the number of background objects by fac-tor of two to three. It is not yet clear whether such a large in-crease could be present, for instance due to clumpiness of ONC’sdust screen. To test this Hypothesis, we performed an independentcheck, using the known spectroscopically verified BDs (see reddots in Fig. 7b). These BDs are located in the 3 (cid:48) − (cid:48) wide stripeSouth and West of the Trapezium which has been observed withthe Gemini-S telescope (Lucas et al. 2005). This subarea showsa high CO column density, corresponding to A V > A V < ff ected by background contamination. Thus with-out any background subtraction, we constructed the IMF of thatcomplete subarea using the HAWK-I data, which show about 50%more objects than the former Gemini-S data.Fig. 12 displays the new IMF of the subarea (green) and theIMF of the spectroscopically confirmed BDs therein (red). The sub-area IMF has also a bimodal shape and peaks at about 0.25 and0 . (cid:12) – exactly as the IMF of the entire ONC area; addition-ally, compared to the confirmed BDs (red), there are about 50%more BDs predicted (green). While for the IMF of the entire ONCarea the substellar peak is equally high as the stellar peak, for theIMF of the subarea the height of the substellar peak is about 30%lower than the stellar peak; this may be due to the di ffi culty to de-tect faint sources against the nebula emission in the subarea. Thecheck on the subarea gives further support for the bimodality of MNRAS , 1–11 (2016) H. Drass et al. the IMF for the entire 22 (cid:48) × (cid:48) ONC area mapped with HAWK-I .To make a further statistical test, we employ the widefield opticaland NIR data set provided by Bouy et al. (2014). Cross-correlatingboth catalogs gives 918 common objects within a 3” matching ra-dius. Using all bands, for 664 objects the membership probabilitygiven in Bouy et al. (2014) is less than 90% and for 649 objectsthe probability is < . < K < . , , , ≈ M Jup ). Nevertheless, the overall shape is essentially pre-served.For the Monte-Carlo Simulation it can be seen in Fig. 13,while the double peak structure is not present in the case of the1 Myr isochrone, it is clearly insensitive to the age use in the range2 − > ∼ (cid:48) × (cid:48) area mapped by Robberto et al. (2010) and used by Da Rio et al.(2012), their IMF appears puzzling. It may be a ff ected by incom-pleteness and overcorrection of background contamination, and the4-m telescope ISPI data may be too shallow and / or o ff ering toolow angular resolution for detecting the BDs in the Gemini-S sub-area. A similar decline of the inner ONC IMF across the Hydrogenburning limit has also been reported by (Hillenbrand & Carpen-ter (2000) their Fig. 16 ) based on their quite shallow Keck obser-vations of the central 5 (cid:48) × (cid:48) area, which shows very bright nebulaemission. Compared to the log-normal distribution parametrizationof Chabrier (2005) IMF (black dashed curve in Fig. 12), the depthof the dip at 0.08 M (cid:12) in our new ONC IMF is about a factor of 3(for the used mass bin sizes). Muench et al. (2002) have reporteda decline of the ONC IMF with a potential upturn at 0.02 M (cid:12) , but– in view of their small field (5 (cid:48) × (cid:48) ), limited depth of the data andthe large error bars – the claim of an IMF dip and low-mass up-turn based on the last mass bin only appears speculative (yellow Recently, we have obtained KMOS / VLT spectra of 20 new BD candi-dates, selected form the HAWK-I data, 16 of which turned out to be in factBDs. They are located towards a region with a CO-screen of A V ∼ Obj /M O • N u m be r o f O b j e c t s Obj /M Jup
Figure 13.
HAWK-I Initial mass functions using the 1 , , ff e et al. (2015) and Allard et al. (2013) are shown bythe solid blue, red, black, and orange lines, respectively. The dashed linesshow the results from the Monte-Carlo Simulation using a Kernel DensityEstimator. For easier identification, except for the 3 Myr isochrones, thepairs of isochrones are o ff set by factors of ten. Confidence intervals of 99%are displayed as blue shaded areas. The di ff erent colors represent the sameages as for the solid lines. The marks are the same as in Fig 12. curve in Fig. 12). The ONC subarea IMF by Lucas et al. (2005)(blue curve in Fig. 12) has been calculated with larger mass bins,which lead to higher numbers per bin and tend to smear out details.Nevertheless, accounting for the smaller number of sources in theGemini-S data ( ∼
400 vs ∼
600 in the VLT data), Lucas et al.’s IMFis nicely consistent with our more detailed bimodal IMF from theVLT HAWK-I data (green solid curve). While the previous IMFshave been derived using older isochrones, the isochrone di ff erencesat the mass range above 0.03 M (cid:12) are too small to explain any of thecoarse IMF di ff erences seen. The explanation for the bimodal shape of the IMF is not straightforward. The IMF dip occurs just at the Hydrogen burning limit,suggesting at a first glance that the burning mechanism of youngstars and BDs may play a role. However, the young BD objectsgain still most of their energy from mass accretion and gravita-tional contraction, so that the Hydrogen burning plays a minor role.Therefore other mechanisms might be needed to explain the originof the bimodal IMF.The predicted starless core mass function (CMF) for Orion,using an A V -dependent extrapolation method, steadily declines be-tween the peak at 1 M (cid:12) and 0.1 M (cid:12) , and at lower mass it stays ona constant plateau at a level of a factor 5 lower than the peak height(Sadavoy et al. 2010). This CMF shape is hardly consistent withthe bimodal IMF. It challenges deterministic theories which em-phasize the role of a direct mapping between CMF and IMF (Nutter& Ward-Thompson 2007; Hennebelle & Chabrier 2013; André etal. 2010). Therefore an attractive explanation for the bimodal IMFcould be ejection of low mass objects from small groups of proto-stars (Reipurth & Clarke 2001) or from fragmenting circumstellardisks (Vorobyov et al. 2013). MNRAS000
600 in the VLT data), Lucas et al.’s IMFis nicely consistent with our more detailed bimodal IMF from theVLT HAWK-I data (green solid curve). While the previous IMFshave been derived using older isochrones, the isochrone di ff erencesat the mass range above 0.03 M (cid:12) are too small to explain any of thecoarse IMF di ff erences seen. The explanation for the bimodal shape of the IMF is not straightforward. The IMF dip occurs just at the Hydrogen burning limit,suggesting at a first glance that the burning mechanism of youngstars and BDs may play a role. However, the young BD objectsgain still most of their energy from mass accretion and gravita-tional contraction, so that the Hydrogen burning plays a minor role.Therefore other mechanisms might be needed to explain the originof the bimodal IMF.The predicted starless core mass function (CMF) for Orion,using an A V -dependent extrapolation method, steadily declines be-tween the peak at 1 M (cid:12) and 0.1 M (cid:12) , and at lower mass it stays ona constant plateau at a level of a factor 5 lower than the peak height(Sadavoy et al. 2010). This CMF shape is hardly consistent withthe bimodal IMF. It challenges deterministic theories which em-phasize the role of a direct mapping between CMF and IMF (Nutter& Ward-Thompson 2007; Hennebelle & Chabrier 2013; André etal. 2010). Therefore an attractive explanation for the bimodal IMFcould be ejection of low mass objects from small groups of proto-stars (Reipurth & Clarke 2001) or from fragmenting circumstellardisks (Vorobyov et al. 2013). MNRAS000 , 1–11 (2016) he bimodal initial mass function in the Orion Nebula Cloud While simulations of fragmenting gas clouds indicate thatBDs can be produced by the ejection mechanism, the resultingstellar / substellar IMFs often show only a modest (factor <
2) andsmoothly running excess of substellar objects above the log normalextrapolation from the stellar IMF peak (e.g. Bate (2009), his Figs.7 & 8). So far, many simulations yield a continuous mass spectrumof the ejected objects, which results in a broad featureless IMF. Thiscontradicts the structured IMF and the distinct substellar peak thatwe observed for the ONC. On the other hand, recent simulations ofa fragmenting circumstellar disk of a 1.2 M (cid:12) pre-stellar core indi-cate that the ejected fragments have a mass spectrum of BDs andIPMOs (Vorobyov et al. (2013), their Fig. 4a). We conclude thatthe CMF might provide a basis of objects for the later IMF but theexcess of BDs and IPMOs is likely produced by ejected fragmentsfrom circumstellar disks. We suggest that the final explanation forOrion’s bimodal IMF has to be searched for along that direction.
ACKNOWLEDGMENTS
The author would like to thank their anonymous referee for her / hishelpful comments. HD likes to thank ESO, Chile for the gener-ous support during a two year studentship. HD acknowledges fi-nancial support from FONDECYT project 3150314. A. Bayo ac-knowledges financial support from the Proyecto Fondecyt Ini-ciación 11140572 . This work was supported by Deutsche For-schungsgemeinschaft, DFG project number Ts 17 / REFERENCES
Allard, F., 2013, http: // perso.ens-lyon.fr / france.allard / Bara ff e, I., Homeier, D., Allard, F., & Chabrier, G. 2015, A&A, 577, A42,arXiv:1503.04107Alves, J., Lombardi, M., & Lada, C. J. 2007, A&A, 462, L17, arXiv:astro-ph / ff e, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337,403, arXiv:astro-ph / ff er, J., et al. 2011, A&A, 536, A63,arXiv:1109.4917Canty, J. I. ,et al. MNRAS, 435, 2650, arXiv:1308.1296Chabrier, G. 2005, The IMF 50 Years Later, 327, 41, arXiv:astro-ph / / / / / / // / wcstools / Muench, A.A., et al. 2002, ApJ, 573, 366, arXiv:astro-ph / / / / APPENDIX A: CATALOG OF ONC MEMBERS
An example for the HAWK-I source catalog is shown in Table A1.The full catalog is available online.
APPENDIX B: ISOCHRONE
The 3 Myr isochrone from F. Allard (download on 25.05.2015) ispresented in Table B1.
This paper has been typeset from a TEX / L A TEX file prepared by the author.MNRAS , 1–11 (2016) H. Drass et al.
Table A1.
Catalog of the HAWK-I sourcesRA (J2000) Dec (J2000) J [mag] uncJ [mag] H [mag] uncH [mag] K [mag] K [mag]83.63357544 -5.17309999 15.299 0.009 14.813 0.013 14.456 0.01383.63535309 -5.34161282 14.177 0.010 13.263 0.010 12.857 0.02283.63700867 -5.22061920 17.710 0.010 16.581 0.015 16.398 0.02583.63703156 -5.17097521 16.234 0.008 15.524 0.014 14.911 0.00983.63822174 -5.16701651 18.278 0.009 17.425 0.019 16.673 0.01783.63856506 -5.12406111 14.863 0.001 14.177 0.001 13.930 0.00183.63997650 -5.13867235 17.839 0.004 16.955 0.007 16.624 0.01183.64015961 -5.20838356 20.423 0.039 19.270 0.023 18.884 0.050
Table B1. / Ms Te ff (K) L / Ls lg(g) R(Gcm) D Li J H K0.0005 501. -5.87 2.75 10.86 1.0000 1.0000 19.202 19.296 19.0370.0010 734. -5.26 3.10 10.23 1.0000 1.0000 15.897 15.831 15.6940.0020 1025. -4.65 3.38 10.49 1.0000 1.0000 14.498 13.517 13.3920.0030 1241. -4.29 3.52 10.93 1.0000 1.0000 13.873 12.602 12.0350.0040 1415. -4.03 3.62 11.34 1.0000 1.0000 13.418 12.186 11.3060.0050 1580. -3.80 3.68 11.79 1.0000 1.0000 13.180 11.836 10.7960.0060 1721. -3.62 3.72 12.27 1.0000 1.0000 12.187 11.123 10.3220.0070 1842. -3.46 3.76 12.77 1.0000 1.0000 11.599 10.686 10.0240.0080 1943. -3.34 3.78 13.29 1.0000 1.0000 11.075 10.320 9.7600.0090 2028. -3.23 3.79 13.86 1.0000 1.0000 10.674 10.035 9.538 MNRAS000