Surveys of Young Stars: Variability and Binarity
Marcelo Medeiros Guimara˜es
September 2008
MARCELO MEDEIROS GUIMARA˜ES
Surveys of Young Stars: Variability and Binarity
Tese apresentada a` UNIVERSIDADE FEDERAL DE MINAS GERAIS
como requisito parcial para obtenc¸a˜o do grau de DOUTOR EM FI´SICA
Orientador: Prof. Luiz Paulo Ribeiro Vaz
Co-orientador: Prof. Bo Reipurth
Departamento de F´ısica - ICEx - UFMG
Setembro de 2008
Agradecimentos
A cieˆncia exige dedicac¸a˜o exclusiva daquele que a estuda. Por vezes o cientista sai de seu
laborato´rio e leva consigo os problemas que ainda teimam em na˜o ser solucionados. Dorme
pensando neles, algumas vezes esses problemas sa˜o resolvidos em sonhos, outras vezes eles
requerem anos de intenso e a´rduo trabalho. Por isso e´ necessa´rio ter paixa˜o pelo que se faz.
Por isso fazemos cieˆncia, porque depois de longas horas, a`s vezes dias, semanas, o resultado nos
salta aos olhos e um sorriso nos vem ao rosto.
Eu tive a sorte de ter trabalhado, durante o meu doutorado, com dois exemplos claros de
dedicac¸a˜o a` cieˆncia. Luiz Paulo e Bo, cada um a` sua maneira, dedicam suas carreiras em
prol do avanc¸o da cieˆncia. Com Luiz Paulo aprendi a ter muito cuidado com os detalhes do
trabalho. Sempre ter certeza de como as coisas funcionam, nunca confiar em “caixas pretas”.
Bo me mostrou o lado apaixonante da cieˆncia, me mostrou que e´ necessa´rio ter forc¸a de vontade
para seguir em frente. Espero ter aprendido o suficiente com eles para poder tambe´m fazer
contribuic¸o˜es relevantes para o avanc¸o da astronomia. Sou muito grato a eles por tudo que me
ensinaram e por seus exemplos.
Uma vida dedicada a` cieˆncia requer sacrif´ıcios, que seriam muito piores na˜o fosse o apoio
constante da famı´lia, da namorada e dos amigos. Felizmente sempre recebi muito apoio das
pessoas a minha volta e sou muito grato a voceˆs por todo o amor, carinho, atenc¸a˜o, amizade
e respeito. Muitas pessoas passaram pela minha vida ao longo desses anos, algumas partiram,
outras ficaram, mas todas deixaram marcas que me fazem o que sou hoje. A todas elas, muito
obrigado.
So´ com sacrif´ıcio na˜o se faz cieˆncia, e´ necessa´rio financiamento. Agradec¸o a`s ageˆncias
financiadoras, CNPq e Capes, pelas bolsas de doutorado e doutorado sandu´ıche, que me per-
mitiram realizar essa tese. Agradec¸o novamente ao meu co-orientador, Bo, pelo financiamento
que me permitiu passar um ano trabalhando ao seu lado no Hava´ı e por ter me fornecido as
observac¸o˜es que sa˜o essenciais para essa tese. Tambe´m agradec¸o Andy Adamson por coordenar
as observac¸o˜es com o UKIRT e pelo seu apoio incansa´vel a esse projeto.
Um longo caminho foi trilhado ate´ aqui. Esse e´ um marco que separa duas etapas. Finalizo
a primeira etapa com essa tese, que apesar de na˜o ser capaz de conter todo o conhecimento
adquirido ao longo desses anos, certamente exemplifica o que sou capaz de fazer. A segunda
etapa se aproxima no horizonte, com boas expectativas.
Acknowledgements
Science demands exclusive dedication from those who study it. Many times the scientist
leaves his laboratory and takes with him the problems that still insist not to be solved. He
sleeps thinking about them, sometimes they are solved in his dreams, sometimes the problems
require years of intense, hard work. That is why it is necessary to have passion for what one
does. That is why we do science, because after long hours, sometimes days, weeks, the result
jumps before our eyes and a smile comes to our face.
I was lucky I had the chance, during my PhD, to work with two clear examples of dedication
to science. Luiz Paulo and Bo, each in his own way, dedicate their career to the development of
science. I learned from Luiz Paulo to be careful about the details, to always be sure how things
work and never trust “black boxes”. Bo showed me the passionate side of science and how it
is necessary to have will power to keep going. I hope a have learned enough with them to also
be able to make relevant contributions to the development of astronomy. I am very grateful to
them for all they have taught me and their examples.
A life dedicated to science requires sacrifices, which would be worst if it were not for the
constant support of the family, girlfriend and friends. Fortunately I have always received much
support from the people around me and I am very grateful to you all for all the love, care,
attention, friendship and respect. A lot of people were part of my life along these years, some
have gone by, others remained, but all of them left marks which made me what I am today. To
all of you, thank you very much.
Only sacrifice does not do science, financial support is necessary. I thank the supporting
agencies, CNPq and Capes, for the fellowships that allowed me to develop this thesis. I thank
again my co-advisor, Bo, for his financial support which allowed me to spent one year working
by his side in Hawaii and also for providing the observations which are essential for this thesis. I
also thank Andy Adamson for coordinating the UKIRT observations and for his tireless support
of this project.
I have come a long way until here. This is a mark that separates two periods. I end the first
one with this thesis, which although it is not able to contain all the knowledge acquired along
all these years, certainly examplifies what I am capable of doing. The second period approachs
in the horizon, with good expectations.
Content
RESUMO viii
ABSTRACT x
1 Introduction 1
2 Visual Binaries in the Orion Nebula Cluster 5
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Observation and Catalogs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Membership of the ONC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Completeness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.1 Binary Fraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.2 Separation Distribution Function . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.3 Wide versus Close Binaries . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.4 Flux Ratios and Substellar Companions . . . . . . . . . . . . . . . . . . . 21
2.6 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Individual Binaries of Interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 Variability 27
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Types of Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3 Statistical Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.4 Testing the Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4 The Orion Nebula Survey 41
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 WFCAM Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.2 Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.1 Workarounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
i
4.3.2 Superposition problems and calibration solutions . . . . . . . . . . . . . . 49
4.4 Variable Candidates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4.1 Eclipsing binary candidates . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4.2 Periodic variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.4.3 Interesting light curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.5 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5 The Cygnus OB2 Survey 67
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.3 Cometary Globules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.4 The Cygnus OB2 Center . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.5 Infrared Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.5.1 IRAS 20321+4112 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.5.2 IRAS 20327+4120 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.5.3 IRAS 20332+4124 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.5.4 IRAS 20333+4127 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5.5 DR18 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.5.6 IRAS 20304+4059 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.6 Summary of Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
6 Conclusions 87
7 Resumo da Tese em Portugueˆs 90
7.1 Introduc¸a˜o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
7.2 Bina´rias Visuais no Aglomerado da Nebulosa de O´rion . . . . . . . . . . . . . . . 93
7.2.1 Introduc¸a˜o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.2.2 Resultados . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7.3 I´ndices Estat´ısticos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
7.4 Levantamento Fotome´trico da Nebulosa de O´rion . . . . . . . . . . . . . . . . . . 96
7.4.1 Introduc¸a˜o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
7.4.2 Resultados . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.5 Levantamento Fotome´trico em Cygnus OB2 . . . . . . . . . . . . . . . . . . . . . 97
7.5.1 Introduc¸a˜o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.5.2 Resultados . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.6 Concluso˜es . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
REFERENCES 100
ii
List of Figures
2.1 Distribution of the visual binaries, with different separations, in the observed
region. The grey solid circle, with a radius of 60′′ and centered in θ1 OriC, is
called the exclusion zone. We did not consider, in our analysis, any star inside
this zone. Single stars are plotted as asterisks and binaries (triples included) are
plotted as black filled circles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Definition of the position angle (PA) and separation (S) between primary and
secondary in a binary system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 All binaries identified among the ONC members. Each stamp is 2′′ wide, with
north pointing up and east pointing left. . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Continued . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 Examples of the artificial binaries, created using the profiles of real stars of our
sample, in order to analyse our completeness. . . . . . . . . . . . . . . . . . . . . 15
2.5 These two plots show the ∆m versus binary separation in arcseconds. The
observed binaries are shown in the left panel, where the dashed line indicates
the 0.′′15 separation limit. The right panel shows the artificial binaries, created
from real stars present in our images, with different separations, position angles
and ∆m. In the right panel, filled circles indicate systems where we could detect
the companions, while open circles represent those where we were not able to
detect them. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.6 Left panel: surface density profile of the ONC members as a function of distance
to θ1 OriC. The exclusion zone is indicated by the dashed line. This profile was
used as a guide to understand and localize the subclustering in the region studied.
It was not used to obtain the probability of chance alignment. Right panel:
probability that a star would have a companion as a line-of-sight association,
obtained with Eq. (2.1). Presented here are all the 781 ONC members, squares
for single stars and circles for binaries. One can see a subclustering around 180′′.
We used the surface density obtained considering the number of stars inside an
area of 30′′ centered in each star. . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
iii
2.7 The left panel shows the histogram of binary angular separations in steps of 0.′′1.
The number of binaries in the innermost bin, with separations less than 0.′′1,
is incomplete. We filled this bin with vertical dotted lines. One can see the
abrupt decrease in the number of binaries when the separation increases beyond
0.′′5. The right panel shows the logarithmic separation distribution function of
the ONC binaries compared to the distribution of field binaries from Duquennoy
& Mayor (1991) across the separation range from 0.′′15 to 1.′′5. The dot-filled
part of the histogram bins represents the contamination by line-of-sight pairs,
obtained using Eq. (2.1). The bins representing large separations are the most
contaminated ones, as expected. The vertical solid lines present in each bin
represent the 1-σ errors calculated using a binomial distribution. The data from
DM91 within our separation range are shown as two dashed crosses, while the
Gaussian distribution they fit to their complete data set is represented by the
dash-dotted line, which can be seen above the histogram. The top axis indicates
separations in arcseconds at the distance of the ONC. . . . . . . . . . . . . . . . 19
2.8 The left panel shows the cumulative distributions of close (0.′′15 – 0.′′5) and wide
(0.′′5 – 1.′′5) binaries in the ONC as a function of distance from θ1 OriC. The
right panel shows the ratio of wide to close binaries in the ONC as a function of
distance to θ1 OriC. The dashed lines indicate the errors. The dotted horizontal
lines at the top of the panel are the same ratio from the DM91 field binary study.
The lower line comes from their Gaussian fit and the upper line comes from their
actual data points. The top axis scale shows the crossing time in years, assuming
a mean one-dimensional velocity dispersion of 2 km s−1. The vertical dot-dashed
line indicates the distance where the ratio becomes essentially flat, suggesting an
age for the ONC of about 1Myr. . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.9 Luminosity ratio of the binary population, based on the Hα fluxes of primaries
and secondaries. All binaries with separation between 0.′′1 and 1.′′5 are plotted,
except the saturated ones. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.10 R-band magnitudes as a function of spectral type for M dwarfs and very cool
objects (spectral type L). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.1 Examples of the light curves of the variables stars used to test the statistical
indices. The plots show the star’s V magnitude versus the phase. The period,
indicated in each panel in days, was obtained in the ASAS database. Each panel
has a label at the top indicating the variable type; starting in the first row we have
the α2 Canum Venaticorum (ACV), Cepheids (CEP), δ-Scuti (DSCT), eclipsing
binary (EB), RR Lyrae (RR) and unknown type variable (Unk). . . . . . . . . . 32
3.2 Statistical indices applied to an α2-Canum Venaticorum type star as a function
of time. In the first row we present, from left to right, the real light curve of
the star, the Mean Magnitude curve and the σ-curve. The second row shows
the σabs-curve, the skewness curve and the kurtosis curve. These last two curves
were obtained using the magnitude of the object. The third row presents the
skewness and kurtosis curves, obtained using the Mean Magnitude curve, and the
R-statistics curve. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
iv
3.3 Statistical indices applied to a Cepheid type star as a function of time. The
indices are the same as in Fig. (3.2). . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4 Statistical indices applied to a δ-Scuti type star as a function of time. The indices
are the same as in Fig. (3.2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.5 Statistical indices applied to an eclipsing binary as a function of time. The indices
are the same as in Fig. (3.2). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.6 Statistical indices applied to an RR Lyrae type star as a function of time. The
indices are the same as in Fig. (3.2). . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.7 Statistical indices applied to the light curve of a star with unknown variability as
a function of time. The indices are the same as in Fig. (3.2). . . . . . . . . . . . 39
4.1 Left panel: Details of the WFCAM focal plane. The four detectors are shown
in their respective position relative to each other. Detector 1 shows the channel
layouts. Right panel: The broad band filters used in WFCAM. We used only the
JHK filters in our survey. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2 Left panel: Field distortion. Right panel: Differential non-linear distortion. . . . 43
4.3 Left panel: Diagram showing how the crosstalks appear in different quadrants.
The crosstalk pattern rotates because the readout amplifier for each quadrant
is located on a different edge of the detector, as shown by the red lines. Right
panel: An example of crosstalk detected in one of our images. Whenever there is
a saturated star, a crosstalk pattern will be present. . . . . . . . . . . . . . . . . 44
4.4 This figure shows how the dark frames, for all detectors, are dominated by the
reset anomaly. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.5 Left panel: Difference between two dark frames taken in order to show the
decurtaining effect. Right panel: Dark frames with decurtaining solved. . . . . . 45
4.6 This figure shows an example of persistent images in the WFCAM. . . . . . . . . 45
4.7 This figure shows an example of the flatfield correction. . . . . . . . . . . . . . . 46
4.8 WFCAM layout and reading scheme used to observe the ONC. . . . . . . . . . . 47
4.9 DADJHK - layout of the table containing all the photometric information. . . . . 48
4.10 (Mean Magnitude - Magnitude) versus Magnitude for the filter J, fourteenth night,
for all the CCDs and areas. The panels are distributed in the same layout shown
in Fig. (4.8). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.11 Same as Fig. (4.10) but for the eighteenth night. . . . . . . . . . . . . . . . . . . 51
4.12 Light curves, Mean magnitude curves and phased light curves for three eclipsing
binary candidates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.13 Light curves, Mean magnitude curves and phased light curves for three eclipsing
binary candidates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.14 We present here the light curves, phased light curves, skewness and kurtosis curves
(of the mean magnitudes) for the stars that presented rotation or pulsation. . . . 58
4.15 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.16 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.17 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
v
4.18 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.19 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.20 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.21 Same as in Fig. (4.14). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.22 We show here examples of interesting light curves present in our sample. . . . . . 66
5.1 The Great Cygnus Rift in Hα, image taken from Schneider et al. (2006). . . . . . 68
5.2 The Great Cygnus Rift, extinction map taken from Schneider et al. (2006). The
white circles represent the radio continuum sources detected by Downes & Rinehart
(1966). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.3 Cygnus OB2 star density map. The left panel shows the 2MASS data while the
right panel shows the DSS red plates data. The image was taken from Kno¨dlseder
(2000). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4 DADJHK and COORDS, layout of the tables containing all the astrometric and
photometric information. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.5 WFCAM layout and reading scheme . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.6 The infrared sources revealed in our survey are displayed in this figure as filled
circles. The sources are labeled using IDs found in the literature. The arrows in
the bottom of the image represent the cometary globules we have found in our
survey, the size of the arrow bodies correspond to the size measured from the
head to the tail of the globule. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.7 Cometary globules CG203446+411446 (panel a), CG203453+405320 and CG203442-
+405115 (left and right on panel b) and CG203318+405911 (panel c). . . . . . . 76
5.8 Cometary globules CG203410+410700 (bottom) and CG203413+410813 (top).
Both globules are associated with red stars in their heads. . . . . . . . . . . . . . 77
5.9 Center region of the Cygnus OB2 association. The first cluster proposed by Bica,
Bonatto & Dutra (2003) corresponds to the concentration of blue stars in the
bottom-center of the figure. The second cluster corresponds to the top-center
concentration. The very bright blue stars in the center, which are members of
the association, were not taken into account by the authors. . . . . . . . . . . . . 78
5.10 IRAS 20321+4112 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.11 IRAS 20327+4120 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.12 IRAS 20332+4124 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.13 The IRAS 20333+4127 source. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.14 JHK composite color of DR18. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.15 Composite color image of the IRAS 20304+4059 infrared source. . . . . . . . . . 84
5.16 Central area close up of the previous image. The red double object is associated
to the IRAS source. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.17 Panels showing the same area around the IRAS source for different surveys. The
area of each panel is approximately the same. North is up and East is left, in all
panels. The arrow indicates the position of the IRAS 20304+4059 source. . . . . 85
vi
List of Tables
2.1 Pre-main sequence binaries with angular separation between 0.15′′ and 1.5′′,
outside the 60′′ exclusion zone around θ1 Ori C. . . . . . . . . . . . . . . . . . . . 14
2.2 Other binaries with angular separations < 1.5′′ toward the ONC. . . . . . . . . . 15
3.1 Statistical index values for all the variable stars in the test sample. . . . . . . . . 33
4.1 Observation log for the Orion Nebula. . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 Six candidates to be eclipsing binaries in M42 . . . . . . . . . . . . . . . . . . . . 53
4.3 Objects candidate to be variable, selected through the study of statistical indices 57
4.4 Some objects presenting interesting light curves, identified with the procedures
developed in the present work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.1 Cygnus OB2 properties, following Kno¨dlseder (2000) . . . . . . . . . . . . . . . . 70
5.2 Observation dates of the Cygnus OB2 survey. . . . . . . . . . . . . . . . . . . . . 71
5.3 Cometary globule ID, coordinates, position angle and assumed size. . . . . . . . . 76
5.4 Coordinates and magnitudes of IRAS 20304+4059. . . . . . . . . . . . . . . . . . 86
vii
Resumo
Com o intuito de aumentar nosso conhecimento sobre estrelas jovens, realizamos treˆs levanta-
mentos de larga escala, em duas regio˜es de formac¸a˜o estelar, a Nebulosa de O´rion e a associac¸a˜o
de estrelas OB conhecida como Cygnus OB2.
O primeiro levantamento foi feito com a “Advanced Camera for Surveys - ACS” a bordo
do telesco´pio espacial Hubble. Utilizamos um filtro estreito, centrado no comprimento de onda
da linha Hα, para mapear a regia˜o ao redor do centro da Nebulosa de O´rion. Nosso principal
objetivo e´ utilizar a alta resoluc¸a˜o da ACS (0.′′05 por pixel) para detectar novos sistemas bina´rios
visuais. Estabelecemos limites para a separac¸a˜o entre as componentes dos sistemas bina´rios (0.′′15
ate´ 1.′′5) utilizando testes de completeza da amostra e testes de probabilidade de projec¸a˜o na linha
de visada. Apo´s um minucioso estudo de cada sistema detectado em nossa amostra, conseguimos
atingir nosso objetivo e observamos ind´ıcios claros da evoluc¸a˜o dinaˆmica desses sistemas bina´rios
na Nebulosa de O´rion. Descobrimos 55 novos sistemas bina´rios visuais em uma amostra de 72
sistemas bina´rios e 3 sistemas triplos. Determinamos uma frequeˆncia de bina´rias de (8.8±1.1)%
na Nebulosa de O´rion, um valor 1.5 vezes menor do que para o campo estelar e 2.2 vezes
menor do que em associac¸o˜es T. Esse ja´ era um fato conhecido, mas baseado em estat´ısticas de
poucos nu´meros, ao contra´rio dos nossos resultados, que foram obtidos com a utilizac¸a˜o de uma
amostra estat´ıstica mais significante. A raza˜o entre o nu´mero de sistemas com separac¸a˜o grande
e pequena aumenta significativamente a partir de 460′′ do centro do aglomerado, que e´ tomado
como a estrela θ1 OriC. Esse fato, segundo nossa ana´lise, e´ um ind´ıcio da evoluc¸a˜o dinaˆmica do
aglomerado, que acontece atrave´s da destruic¸a˜o dos sistemas bina´rios com grandes separac¸o˜es,
apo´s algumas passagens pelo poc¸o de potencial gravitacional do centro do aglomerado. Um
sub-produto da nossa ana´lise e´ a detecc¸a˜o de candidatos a objetos sub-estelares (ana˜s marrons).
Determinamos a natureza dupla do objeto COUP1061, classificado na literatura como uma ana˜
marrom. Essa descoberta implica que o objeto COUP1061 e´ um sistema bina´rio composto por
duas ana˜s marrons, separados por aproximadamente 100 unidades astronoˆmicas.
Os demais levantamentos foram feitos com a “Wide Field Camera” do “United Kingdom
Infrared Telescope”. Observamos a Nebulosa de O´rion e Cygnus OB2, na regia˜o do infravermelho
pro´ximo, nas bandas J, H e K. O objetivo desses levantamentos fotome´tricos e´ determinar a
populac¸a˜o de estrelas varia´veis dessas regio˜es de formac¸a˜o estelar. Atrave´s da utilizac¸a˜o de
ı´ndices estat´ısticos, as estrelas varia´veis podem ser classificadas em seus diferentes grupos. Por
se tratarem de levantamentos fotome´tricos de a´reas grandes, desenvolvemos rotinas especiais
para lidar com os cata´logos, compostos por dezenas e algumas vezes centenas de milhares de
objetos.
A Nebulosa de O´rion foi observada por 101 noites, num intervalo de 178 dias. Foi feita
uma exposic¸a˜o de 2 segundos por noite. Um longo processo de catalogac¸a˜o e calibrac¸a˜o foi
viii
necessa´rio ate´ se obter um conjunto de dados que pode ser utilizado para a busca de estrelas
varia´veis. Atrave´s da aplicac¸a˜o de ı´ndices estat´ısticos conseguimos descobrir novas candidatas
a bina´rias eclipsantes e estrelas com variabilidade causada por pulsac¸o˜es ou por rotac¸a˜o. Um
longo processo de ana´lise e´ ainda necessa´rio para exaurir a enorme quantidade de informac¸a˜o
presente nesse levantamento.
A regia˜o de Cygnus OB2 foi observada por 112 noites, num intervalo de 217 dias. As
observac¸o˜es seguiram o mesmo esquema usado no levantamento da Nebulosa de O´rion, que
consiste em uma exposic¸a˜o (com durac¸a˜o de 2 segundos) por noite. A ana´lise preliminar dos
dados para Cygnus OB2 nos permitiu descobrir 6 glo´bulos cometa´rios de pequena escala. Essa
descoberta e´ uma forte evideˆncia do processo de formac¸a˜o estelar recente nessa regia˜o. Tambe´m
foi poss´ıvel detectar 4 fontes infravermelhas, aparentemente jovens e que possivelmente esta˜o
localizadas no Brac¸o Espiral de Perseu, muito mais distante do que a associac¸a˜o Cygnus OB2.
Determinamos a natureza dupla da fonte infravermelha IRAS 20304+4059 que e´ classificada na
literatura como poss´ıvel proto-estrela.
ix
Abstract
In the attempt of increasing our knowledge about young stars, we performed three large scale
surveys of two star forming regions, namely the Orion Nebula and the young OB association
known as Cygnus OB2.
The first survey was made with the Advanced Camera for Surveys (ACS) onboard the Hubble
Space Telescope. We used a narrow band filter, centered in the wavelength of the Hα line, to
map the region around the center of the Orion Nebula Cluster. Our main goal is to use the high
resolution of the ACS (0.′′05 per pixel) to detect new visual binaries. We set the separation limit
between the components of the system (0.′′15 to 1.′′5) using completeness tests and the probability
of chance alignment. After a minute analysis of each system detected in our sample, we reached
our goal and also observed clear evidence of the dynamical evolution of these systems in the
Orion Nebula Cluster. We discovered 55 new visual binary systems within a total of 72 binaries
and 3 triple systems. We determined a binary frequency of (8.8±1.1)% for the Orion Nebula
Cluster, which is 1.5 times smaller than the frequency for the field and 2.2 times smaller than
for the loose T associations. This was an already known fact but was based on small number
statistics, while our results are based on more significant numbers. The ratio between wide and
close binaries has a significant increase at 460′′ from the center of the cluster, assumed to be
the star θ1 OriC. Our analysis indicates that this is clear evidence of the dynamical evolution
of the Orion Nebula Cluster, which happens through the destruction of wide binaries, after few
passages through the cluster’s potential well. A byproduct of our survey is the detection of
substellar candidates. We determined the double nature of the object known as COUP1061,
which is classified in the literature as a brown dwarf. This finding implies that COUP1061 is a
binary system where both components are brown dwarfs, separated by ∼100 AU.
In order to perform the second and third surveys we made use of the Wide Field Camera
of the United Kingdom Infrared Telescope. We observed the Orion Nebula and Cygnus OB2
with near infrared filters, JHK. Our objective with these photometric surveys is to determine
the population of variable stars in these star forming regions. Using statistical indices we can
separate and classify the variable stars into different groups. Because these photometric surveys
cover large areas in the sky we developed special routines to deal with the catalogs, which are
composed of thousands and sometimes hundreds of thousands of objects.
The Orion Nebula was observed during 101 nights, covering 178 days. We performed short
exposures (2 seconds long) once each night. Long cataloging and calibration processes were
necessary until a satisfying dataset was ready to be used in the search for variable stars.
Using statistical indices we were able to discover new eclipsing binary candidates and star with
variability due to pulsation or rotation. A long analysis process is still necessary to exhaust the
huge amount of information obtained in this survey.
x
The Cygnus OB2 region was observed 112 nights, covering 217 days. The observations
followed the scheme used in the Orion Nebula survey, which consisted of one exposure (2 seconds
long) each night. A preliminary analysis of the Cygnus OB2 data allowed us to discover 6
small scale cometary globules. This finding is strong evidence of the recent star formation
process that occurred in this region. We also detected 4 infrared sources, apparently young and
possibly located in the Perseus Arm, far more distant than the Cygnus OB2 association. We
also determined the double nature of the IRAS 20304+4059 infrared source, which is classified
as a protostar.
xi
Chapter 1
Introduction
Star formation is nowadays one of the most important branches in astronomy. It all started
with the first observation of late-type low mass stars in the Taurus-Auriga region by Joy (1945).
Later, these stars were proposed to be young stars by Ambartsumian (1957) and they were
labeled T Tauri stars, after the object thought to be the prototype of young solar mass stars.
Herbig (1960) observed the intermediate mass relatives of T Tauri stars, the Herbig Ae/Be stars.
Since these pioneer works we have seen an incredible growth of observations, both in quantity
and quality, which supported the establishment of different models of formation, evolution and
structure of stars. We study these young stellar systems in the hope that they will give us hints
about the history of our own Solar System. We try to understand how stars and planets form
so that we can comprehend how the Earth, and other planets in the Solar System, formed when
the Sun was only a few million years old.
One of the things we have learned, during the study of young stars, is that most of them are
formed in multiple systems (e.g. Leinert et al., 1993; Mathieu, 1994; Ducheˆne, 1999; Bodenheimer
et al., 2000; Patience et al., 2002; Kroupa & Bouvier, 2003; Ducheˆne et al., 2004; Haisch et al.,
2004). If that is true then why is the Sun a single star? A multiple system can be disrupted
by different mechanisms such as the fast decay of small-N systems (Sterzik & Durisen, 1998;
Reipurth & Clarke, 2001; Durisen, Sterzik & Pickett, 2001; Hubber &Whitworth, 2005; Goodwin
& Kroupa, 2005; Umbreit et al., 2005) and the dynamical destruction caused by multiple interac-
tions in clustered environment (Kroupa, 1995a,b; Kroupa et al., 2003). The former mechanism
acts at the beginning of star formation, while the molecular cloud core ends its collapse.
Nonhierarchical systems with N > 3 decay into single objects, binaries and hierarchical multiples.
The latter acts during the whole life of the cluster. It consists of the interaction of multiple
systems with the cluster’s potential well and with the crowded environment that surrounds the
object. This interaction will disrupt multiple systems that are loosely bound. Some models
also predict that environmental influences, such as the molecular cores’ temperature, could be
important to define the initial binary fraction (Durisen & Sterzik, 1994; Sterzik, Durisen &
Zinnecker, 2003).
The implications of multiplicity for the planet formation scenario, and also for the evolution
1
of stars, are still unknown. Recently, Stassun et al. (2008) reported the discovery of a pair of
“stellar twins” in a binary system. Each star has a mass of 0.41 ± 0.01 M⊙, identical to 2%.
What makes the pair so special is the fact that although they are similar in mass, they have
temperatures differing in ∼10% and luminosities by ∼50%, with high confidence levels. This
binary system, with very similar components, though not identical as claimed by the authors,
can give us hints about the influence that multiplicity can exert during the formation of stars.
Multiple systems play an important role in astronomy. They are very important for the
calibration of evolutionary models and it is not much to say that the spectroscopic eclipsing
binaries act in astronomy as “weights and rulers”. These systems, in particular, can give us
very precise (. 1%) absolute parameters (mass, radius, ratio of temperatures, etc) which can be
used as constraints to theoretical models. Although multiple systems seem to be very common
at the early stages of star formation, only 22 pre-main sequence stars have parameters measured
through the use of dynamical techniques (Mathieu et al., 2007; Stassun et al., 2008). In order to
discover more eclipsing binaries it is necessary to perform accurate and comprehensive surveys
with long time-line bases.
Since the beginning of the study of young stars it became clear that they are extremely
variable objects. In fact, photometric variability was one of the criteria used to classify them.
Young stars present photometric and spectroscopic variability in all wavelengths, from radio
to X-rays. Some of the variability is caused by the circumstellar material in different ways.
Circumstellar material can cause variable extinction when it crosses our line of sight toward
the object, causing sudden drops of brightness. The process of accretion, through which the
star increases its mass, can still be active, causing variability in brightness. In young T Tauri
stars the convection zone in their interior creates magnetic fields strong enough to produce large
scale cool spots in the photosphere, which along with the star’s rotation produce a periodic
modulation in the light curve. Flares in young stars, similar to solar flares, can also produce
irregular photometric variability. The processes of accretion and ejection of matter can also
change the line profile of many atomic species present in the photosphere of the star, giving
birth to the spectroscopic variability in young stars (e.g. Guimara˜es et al., 2006). Pulsation was
also observed in young stars (e.g. Kurtz & Marang, 1995) situated at the instability strip of the
HR diagram.
After stating the above, we face here two important subjects, namely: multiple young stars
and variable young stars. Both subjects are important and they are directly associated, since
stars in multiple systems can be (and most of the time are) variable. We will only have a
coherent picture of the star formation scenario if we study individual systems in detail. High
precision in the determination of absolute parameters, of individual stars, is essential if we
want to build models that really represent processes occurring in nature. But in order to select
good candidates to study, we first need to discover them. That is the importance of large scale
comprehensive surveys. Because astronomers are stuck on Earth, and thus do not have access to
their experiments, it is necessary to map the whole sky in the search for the “good experiment”.
There are many difficulties in doing large scale surveys, for example, availability of telescope time
to perform observations in specific areas of the sky, correct instrumentation, storage capacity
for the huge amount of data, computational resources to process all the data collected, human
resource in order to analyse the data, etc. Many all sky surveys were done recently, covering
different wavelengths, such as: the Two Micron All Sky Survey (2MASS, Skrutskie et al., 2006)
in the JHK near infrared bands, the Sloan Digital Sky Survey (SDSS, Fukugita et al., 1996) in
2
the Sloan optical bands (u’, g’, r’, i’, z’), the All Sky Automated Survey (ASAS, Pojmanski,
1997) in the optical V and I bands, the UKIRT Infrared Deep Sky Survey (UKIDSS) considered
to be the successor to 2MASS, just to cite a few of them.
We are involved in the Variable Young Stars Optical Survey (VYSOS), which intends to
continuously monitor stellar objects in star forming regions located in the Galactic plane and
also to find new young objects in the same star forming regions. It was part of my project to
work with the data collected by these telescopes.
The project consists of 2 small telescopes (the diameter of the main mirror is 41 cm), one
installed in the northern hemisphere, at the Mauna Loa Observatory – Hawaii – USA, and the
other in the southern, at Cerro Murphy – Chile.
Each telescope will be equipped with a set of Sloan filters (Fukugita et al., 1996) and an SBIG
CCD, which has a dimension of 4096×2048 pixels and a pixel scale of 0.′′77. Such a combination
of pixel scale and dimension will produce a field of view of 27′×40′.
The northern telescope has been facing some mechanical and optical problems. In July 2006
I was at the Institute for Astronomy (IfA) in Hawaii, as part of my PhD program, when a major
problem with the optical system was detected. The first problem was detected during the process
of collimation. The optical system lacked one degree of freedom necessary to align the optical
beam. This problem was reported to the engineers who built the system and some solutions
were proposed by them. After trying different approaches we concluded that the missing degree
of freedom was really necessary and the problem could not be solved without making a new
base for the plain mirror, responsible for sending the optical beam straight to the camera. The
difficulties in the process of alignment and collimation made us check the whole optical design
and other problems were detected by Dr. Josh Walawender, a postdoc at IfA at that time. After
calculations and studies of the optical system the conclusion was that it did not work, except in
the optical axis. The set of optical lenses used to focus the beam leaving the tertiary mirror did
not posses the necessary quality. Thus, due to construction and optical problems, the telescope
was redesigned and many parts of it were rebuilt.
Formerly, the telescope was a modification of the Newtonian design. After being redesigned
it was built as a Newtonian telescope. Such a modification brought some advantages, for example
it decreased the number of mirrors and lenses used. The collimation and alignment were done
easily this time. However, some issues remained, dealing mainly with the tracking system. We
were not able to track during long exposures. Also, the telescope showed itself to be extremely
sensible to vibrations of the dome. This problem was caused by the fact that there is not a
separated pier for the telescope and all dome vibrations (caused by wind) were transmitted
to the telescope through the floor. The floor where the telescope is fixed also presented some
problems because it was not flat enough. The steel plates that should sustain all the weight of
the telescope showed some signs of bending. A new mechanical problem was also found, caused
by the 6.8 magnitude earthquake that stroke Hawaii on October 2006. The earthquake caused
the telescope to bounce and the aluminum wheel responsible for the right ascension movement
was slightly indented.
With so many issues dealing with the installation, the telescope operation was delayed. We
decided to focus our work on new projects, which also dealed with photometric variations of
young stars. Recently (August 2008) the telescope located in the Mauna Loa Observatory was
taken down and will be replaced by a new one. The southern telescope was installed in the same
3
month but it is not working yet.
Thus, because our main project was delayed, we decided to study binary stars and variable
stars from a different point of view. We will focus our attention in specific star forming regions,
in order to build a comprehensive database, covering long periods, usually hundreds of days.
Instead of looking at the whole sky, we will look at specific regions for a long time interval.
In our attempt to better understand multiple young stars, we report here the performance of
an imaging survey using the Advanced Camera for Surveys, onboard the Hubble Space Telescope,
of the well studied Orion Nebula. The main objective of this survey is to find new visual binaries
in a wide region around the Orion Nebula Cluster. Previous studies have shown that the binary
frequency of the Orion Nebula Cluster is lower than the binary frequency of the Galactic field.
Does this fact have something to do with the dynamical evolution of the cluster? Can we
distinguish between the two mechanisms responsible for the dynamical evolution of the binary
properties? We will try to answer these questions in Chapter 2.
Our second incursion in the star formation domain is related to two near infrared large scale
surveys of two different star forming regions. Our main goal is to unveil the young variable
population of the Orion Nebula and of the young OB association Cygnus OB2. For this task
we used the Wide Field Camera of the United Kingdom Infrared Telescope, the most capable
infrared imaging survey instrument in the world, currently installed at the summit of the Mauna
Kea Observatory. We will present in Chapter 3 some statistical techniques that can be used to
characterize the variable star population in these fields. Light curves from the ASAS were used
as tests in order to better understand the behavior of these statistical indices.
In Chapter 4 we will present the observations, procedures and results of the survey conducted
in the Orion Nebula. We have built a database using 98 nights, from 101 observed nights covering
178 days, which is so far the longest infrared survey of the Orion Nebula. Many interesting
variable objects were found after we applied the statistical indices to the database.
Cygnus OB2 will be analyzed in Chapter 5. It is a giant OB association, which was even
considered to be a young globular cluster, given the number of O type stars associated to it
(∼100 stars). Cygnus OB2 is situated in the Cygnus X star forming region, behind the Great
Cygnus Rift. When we look in that direction of the sky, we are looking through the Local Spiral
Arm and the Perseus Arm and, hence, we can see many star forming regions projected in the
same area of the sky. A better analysis of its low and intermediate mass content is necessary
in order to put constraints to its total mass and extent. We observed the center of this OB
association during 112 nights, covering an interval of 217 days, but given the huge amount of
data we were not able to construct the final database yet. However, four nights were used to
construct a preliminary catalog, which we used to study the properties of this interesting star
forming region. It is the richest infrared survey made for this association and given the attained
angular resolution we will be able to double the number of objects associated to it.
We will present our conclusions in Chapter 6.
4
Chapter 2
Visual Binaries in the Orion Nebula Cluster
2.1 Introduction
The Orion Nebula Cluster (ONC) is responsible for the ionization of the H II region know as
the Orion Nebula (M42 + M43 = NGC 1976), one of the most studied regions in the sky. The
ONC calls such attention because it is the nearest giant star forming region (∼450 pc). Its most
massive stars, know as the Trapezium stars, are located in its center. They were thought to be
a distinct entity (e.g. Herbig & Terndrup, 1986) but it has been argued that they are actually
the core of the ONC (Hillenbrand & Hartmann, 1998).
The ONC is not only the nearest site of high-mass star formation but is also very young,
with age ∼1 Myr. Its youth is supported by:
• the stars located above the ZAMS (Zero Age Main Sequence) (Herbig & Terndrup, 1986;
Prosser et al., 1994);
• the photometric variability of more than 50% of its stars (Jones & Walker, 1988; Choi &
Herbst, 1996);
• the existence of optical emission line spectra (Herbig & Bell, 1988, and references therein);
• the high polarization;
• the infrared excesses;
• the proplyds pointing toward θ1 OriC (the most massive of the Trapezium stars).
The wind and radiation from the massive stars ionize the surrounding medium, with two
important consequences. First, the ionization creates a low extinction (AV ≤ 2.0) line of sight
to the stars around the Trapezium that allows us to study in detail this population. Second,
the ionization of the molecular material behind the ONC creates a highly variable background
brightness. This molecular material is part of the Orion Molecular Cloud which has a large
column density (AV = 50−100). Such high extinction helps us in the study of the ONC because
it hides the background objects but also prevents us from achieving a complete census of the
ONC stellar population.
5
The ONC has been studied in all wavelengths, from radio to X-rays. Since it is one of the
most studied regions in the sky, a full review would be beyond the purpose of this work. We
will focus our attention on the orbital evolution of its multiple systems.
An intriguing result of the study of the formation and evolution of binary systems is that
binaries are more common (by a factor of 2) in T Tauri associations than in the field (e.g.
Reipurth & Zinnecker, 1993; Simon et al., 1995; Ducheˆne, 1999; Ratzka, Ko¨hler & Leinert,
2005). A more intriguing result is that the ONC has a binary frequency lower than T Tauri
associations and the field (e.g. Prosser et al., 1994; Padgett, Strom & Ghez, 1997; Petr et al.,
1998; Simon, Close & Beck, 1999; Ko¨hler et al., 2006).
The study of binary properties is marked by the detailed work with field binaries made by
Duquennoy & Mayor (1991, hereafter DM91). The properties of the binary field population
serve as a reference to the work on younger populations. But can we trust the binary field
population? How is it formed? Does the field population come from a mix of other cluster
populations (e.g. loose T associations, OB associations, open clusters and so forth)?
In the recent literature there are two main models for the evolution of binary properties:
1) fast dynamical decay of small-N systems within cluster cores (e.g. Reipurth & Clarke, 2001;
Sterzik & Durisen, 1998, 2003; Durisen, Sterzik & Pickett, 2001; Hubber & Whitworth, 2005;
Goodwin & Kroupa, 2005; Umbreit et al., 2005) and 2) dynamical destruction due to multiple
interactions in a clustered environment (e.g. Kroupa, 1995a,b; Kroupa et al., 2003).
The first mechanism acts in the initial stages of star formation, during the Class 0 phase,
while the second mechanism needs more time to act and does so for a longer time. We can thus
say that the fast decay mechanism acts on a small scale (the scale of the multiple system) while
the dynamical destruction mechanism acts on a larger scale (the scale of the whole cluster). We
will call them fast decay mechanism (FDM) and dynamical destruction mechanism (DDM).
Assume we have stars forming in a cloud. During the collapse of the cores many multiple
systems are formed. The FDM will act on these multiple Class 0 systems in a way that the less
massive components will be ejected and the resultant binaries will have their semimajor axis
decreased.
The DDM will act in all the systems during their whole lives within the cluster. In order to
help us understand how this works we will divide the binary systems into three classes.
1. the wide, or soft, binaries;
2. the dynamically active binaries;
3. the tight, or hard, binaries.
The wide binaries have orbital velocities much smaller than the velocity dispersion of the
cluster. The DDM will disrupt these systems and thus populate the field with single stars. The
hard binaries, on the contrary, have an orbital velocity much greater than the velocity dispersion.
These binaries will release energy to the cluster, becoming harder.
These facts lead Heggie (1975) and Hills (1975) to propose the following law: soft binaries
soften and hard binaries harden.
The dynamically active binaries have intermediate binding energies and are difficult to study
analytically. However, computations show that these binaries are efficient in exchanging partners
when interacting with the other binaries and single stars in a cluster.
6
An important point to note, coming from the computational simulations, is that dynamical
interactions cannot form a significant number of binaries from an initially single star population
(Goodwin et al., 2007).
Two points of view come into shock at this present time. Goodwin & Kroupa (2005) argue
that almost all stars are formed in multiple systems, meaning that the initial binary fraction
is close to unity. The later low binary fraction among M dwarfs would then be caused by the
preferential destruction of these low mass systems. They are the first ones to be disrupted by
both mechanisms. On the other hand, Lada (2006) argues that because most M dwarfs are
single and most stars are M dwarfs, then most stars form as single stars.
Ko¨hler et al. (2006) suggest that the initial binary frequency in the ONC was lower than in
Taurus-Auriga. This fact suggests that the binary formation rate is influenced by environmental
conditions. Their results do not agree with the mechanisms described by Goodwin & Kroupa
(2005).
The statistics of binarity for the ONC were poor and concentrated on regions around the
Trapezium stars. Such an important statement should be based on more solid grounds. We will
show here, using significant numbers, that the ONC indeed has a lower binary fraction than the
T Tauri associations and the field, as reported previously in literature.
We present here a study of the ONC using the Advanced Camera for Surveys (ACS) onboard
the Hubble Space Telescope (HST). Our major goal is to study the binary frequency of the ONC,
focusing on the outskirts of the cluster and excluding the inner region around θ1 OriC. The
results presented here were published in the Astronomical Journal, see Reipurth et al. (2007).
2.2 Observation and Catalogs
During program GO-9825, 26 fields were observed with the F658N filter (Hα+N[II]) and an
exposure time of 500 seconds per pointing. These 26 fields when combined yield a mosaic image
that covers an area of 415 arcmin2 with a very high angular resolution (0.′′05 per pixel). Further
details of the observations are described in Bally et al. (2006) and a schematic figure showing
the disposition of the observed areas is shown in Figure (2.1).
Due to the high stellar density in the Trapezium region we excluded a circular region with
radius 60′′ centered in θ1 Ori C, as can be seen in Fig. (2.1). We will not take this region into
account in our work. We focus our attention on a larger area around the Trapezium, where we
are able to probe the low mass stellar population.
We tested an automatic procedure to detect the stellar sources but it did not yield satisfactory
results. The routines were written in IDL (Iteractive Data Language) and made use of the
package IDLASTRO, which is a library with many routines written specifically to deal with
astronomical issues.
Each one of the 26 images was opened with the readfits.pro sub-routine and the data was
loaded into a matrix called data. The astrometrical information, contained in the header of
each image, was extracted with the extast.pro sub-routine to be used later in the recovery of
the coordinates. Then a sub-routine called find.pro was applied to the data matrix in order to
detect the astronomical sources. It requires some input parameters in order to control the level
of detection. The main input parameters for the find sub-routine are: roundness, sharpness,
hmin, FWHM (Full Width at Half Maximum).
7
Figure 2.1: Distribution of the visual binaries, with different separations, in the observed region. The grey solid
circle, with a radius of 60′′ and centered in θ1 OriC, is called the exclusion zone. We did not consider, in our
analysis, any star inside this zone. Single stars are plotted as asterisks and binaries (triples included) are plotted
as black filled circles.
The roundness parameter is related to the geometric form of the object being searched, it
is a 2-dimensional vector (between −1 and 1, the default values) and should only be changed if
the objects are elongated. Since we have many saturated objects in our images this parameter
was of not much use when applied to these cases and in fact introduced many false detections.
The sharpness parameter relates to the kind of statistical distribution followed by the objects.
It is also a 2-dimensional vector where the default values are 0.2 and 1.0. These values should
only be changed in case the objects’ point spread function is different from a Gaussian. As in
the case of the roundness limit, the sharpness limit was not of much help while dealing with
saturated objects and high background brightness.
The hmin parameter is the threshold intensity for the point source. Its value should be 3 or
4 σ above the background noise. We have very faint objects in our sample, thus we cannot put
the threshold limit too high. However, setting this limit to a low value would also cause a very
high number of detections, given the high background brightness variability. This parameter
became our biggest problem while trying to use an automatic detection procedure.
The FWHM is the value of the full width at half maximum to be used in the convolve filter.
Because we have so many different objects, with very different intensities, this parameter did
not help us discard the false detections.
8
We tested this automatic procedure many times, with different sets of values for the input
parameters, but the output was far from useful. The procedure always returned several thou-
sands of detections per image. Because of this we decided to search the image by eye, individually.
We had to take some care while checking the stellar sources by eye because some images
overlap, as can be seen in Fig (2.1). All the overlapping areas were carefully inspected and the
stellar sources detected in more than one image were catalogued and counted only once. In the
end we detected 1 051 stellar sources in the whole area.
The coordinates for each object were obtained using two routines, also written in IDL. The
first one was responsible for obtaining the pixel coordinates of the stellar sources, through the
use of a 2-dimensional Gaussian function. The center of the 2D-Gaussian was taken as the center
of the stellar source. The pixel coordinates were then transformed to celestial coordinates with
the IDLASTRO sub-routine called xy2ad.pro.
Once the stellar sources were catalogued we were able to go to the next step, selection of the
multiple systems. Visual binaries have two important parameters, the position angle (PA) and
the separation (S) between the pair. The position angle measures the angle between the primary
and secondary and it is counted from North to East (counterclockwise), using the primary as
reference. Figure (2.2) exemplifies the position angle and separation between two stars in a
binary system.
N
P.A.
E
S
Figure 2.2: Definition of the position angle (PA) and separation (S) between primary and secondary in a binary
system.
To calculate the position angle we used an IDLASTRO subroutine called posang.pro. The
separation between the stars was obtained in two distinct ways. The first method applies the
IDLASTRO sub-routine gcirc.pro, which calculates rigorous arc distances in spheres. In the
second method we used the pixel coordinates to obtain the distance between the objects in the
plane of the image, discarding any deformation caused by the sky projection onto the plane of
the CCD. Both methods yielded similar results, which can be explained by the fact that the
separations are too small to be influenced by the projection of the celestial sphere onto the
plane of the CCD, and also because the observed region is close to the equator, where the field
deformation is less pronounced. Because some binaries have a very small separation (<0.′′15)
we decided to apply the trigonometric method in order to obtain the angular separation. We
decided on this method because in our opinion, on these small scales, the pixel coordinate is
more accurate than the astronomical coordinate.
9
2.3 Membership of the ONC
We used 4 criteria to establish whether a star in our sample belongs to the ONC:
1. it has proper motion given by the Jones & Walker (1988, from now on, referred to as JW)
catalogue and the membership probability is ≥ 93%;
2. it is listed as an ONC member by the X-ray survey performed by the Chandra satellite
(COUP project – (Getman et al., 2005));
3. it presents irregular variability as listed by the GCVS (General Catalogue of Variable
Stars);
4. it has Hα emission (detected in our unpublished catalogue).
A star does not need to obey all the criteria at the same time, it is sufficient to present one
of them to be classified as a member of the ONC. From 1 051 stars in our sample, 655 are listed
in the JW catalogue and 596 have a membership probability ≥ 93%. We want to point out that
most of our binaries are not resolved in the JW survey, given it was made with photographic
I-band plates. This fact implies that the proper motions can be affected by changes in the
brightness of the companion stars. We observed such variations in our sample. One of our
binaries is located on the edge of two images which overlap themselves. Since the observations
were made in different epochs we were able to detect a brightness variation in the system. Such
variation can change the photocenter of the pair (in an unresolved image, like the photographic
plates used by JW) and thus lead to a wrong proper motion. We also would like to point out that
some stars with 0% membership probability showed signs of youth, which is good evidence of
its association with the ONC. Hence, we concluded that the proper motion information is useful
to support the membership of a system, but such information should not be used to exclude
membership. Information about the youth of the system can be gathered from other sources,
like X-ray emission, Hα emission and photometric variability. That is the reason why we used
the COUP catalogue (X-ray), the GCVS (variability) and the unpublished Hα catalogue. Not
all young stars present Hα emission, but we assume that a star in the Orion region that presents
such emission line is young. From the 1 373 ONC members in the COUP catalogue, we found
that 658 are present in our sample. We found 196 variable stars in our sample and 99 stars
presented Hα emission.
In the end, out of our 1 051 stars, 781 present at least one (and commonly several) of
these four characteristics. We found 75 visual binaries and 3 visual triples among these 781
ONC members. Table (2.1) presents these binaries, columns 1 and 2 list the JW number and
probability of the star being a member of ONC, respectively. Columns 3 and 4 present the name
in the GCVS catalogue and the COUP number, respectively. Right ascension and declination
are listed in columns 5 and 6, for equinox J2000.0. Columns 7 and 8 list the position angle and
the separation (in arcseconds) of the system, while column 9 lists the apparent magnitude in the
I band (Hillenbrand, 1997, hereafter called H97). The difference in magnitudes from the primary
to the secondary, in the Hα filter, is listed in column 10, where the symbol > indicates that the
primary is saturated (sometimes both stars), thus ∆m is only an aproximation. Columns 11 and
12 list the position angle and separation (in arcseconds) to θ1 OriC. Column 13 lists the spectral
type given by SIMBAD. Column 14 lists the membership criteria presented by the system: P
for proper motion; X for X-ray; H for Hα emission; V for irregular variability. Column 15 lists
10
the discovery paper in case the binary was already known and the letters correspond to the
references: a - Prosser et al. (1994); b - Padgett, Strom & Ghez (1997); c - Simon, Close & Beck
(1999); d - Ko¨hler et al. (2006); e - Getman et al. (2005); f - Lucas, Roche & Tamura (2005).
Table (2.2) lists binaries that were found in our sample but that were not used in our analysis.
The reasons for not using these binaries are: three binaries have separation in the range 0.′′11
and 0.′′15, other three binaries are within the exclusion zone and have separations less than 0.′′4
and seven binaries do not show any evidence of membership to ONC. The columns in Table
(2.2) are the same as in Table (2.1) with one more column that characterizes the binaries as:
1 - binaries with angular separation < 0.′′4 inside the exclusion zone; 2 - binaries with angular
separation <0.′′15 outside the exclusion zone and 3 - binaries with angular separation < 1.′′5
outside the exclusion zone but with no evidence of ONC membership .
With the exception of the object JW 945, which is a Herbig Ae/Be star, all of the binaries
in our sample consist of late-type stars, see Table (2.1). Figure (2.3 presents stamps for the 72
visual binaries, 3 visual triples and also the 3 visual binaries inside the exclusion zone.
2.4 Completeness
Since our major goal is to search for visual binaries in the ONC, we have to account for the
contamination due to line-of-sight alignment. We must set an inferior and a superior separation
limit for our search.
The inferior limit was determined based on the angular resolution obtained with the ACS
(0.′′05) and with a blind test. We selected 5 stars in our images that presented a good Gaussian
profile. An IDL routine was written to copy this Gaussian profile and create artificial stars,
which were then positioned beside real stars but with random separations, position angles and
flux ratios. Seven sets of 25 images each were created. One example of the artificial binaries is
shown in Fig (2.4).
The blind test consists of looking at each image in sequence during no more than 5 seconds.
If the pair is visible and easily detected, we call it a positive target. If it is not possible to
detect the pair within 5 seconds, we call it a negative target. Three of us (Bo Reipurth, Michael
Connelley and I) did this test for each set of artificial binaries and the positive and negative
targets were added. If one person counted a target as positive but the other two counted it as
negative, the target was then counted as negative. Likewise, two positives and one negative was
counted as one positive.
Figure (2.5) presents the results of this blind test. Our observed binaries are shown in the
left panel. The right panel shows the detections found during the blind test, where filled circles
indicate systems in which we could detect the companions and open circles represent those in
which we were not able to. We can detect binary systems down to 0.′′10, however, our sample
will not be complete in this limit, as can be seen in Fig. (2.5). We have set our inferior limit to
0.′′10, but we will use in our subsequent analysis only the binaries with separation ≥ 0.′′15.
The superior limit was determined based on the probability of a companion being a line-
of-sight contamination, as used by Correia et al. (2006). We used Eq. (2.1) to calculate this
probability:
11
Figure 2.3: All binaries identified among the ONC members. Each stamp is 2′′ wide, with north pointing up
and east pointing left.
12
Figure 2.3: Continued
13
Table 2.1: Pre-main sequence binaries with angular separation between 0.15′′ and 1.5′′, outside the 60′′ exclusion
zone around θ1 Ori C.
JW % GCVS COUP RA DEC PA(sys) S mI ∆mHα PA(cluster) OriC SpT Memb. Bin
39 99 V1438 ... 5:34:38.1 -5:27:41 21 0.38 14.29 2.6 246 628 M3 P00V
52 98 ... ... 5:34:40.8 -5:28:09 231 0.20 14.58 0.1 242 604 M5 P000
63 99 V1441 17 5:34:43.0 -5:20:07 0 0.27 12.75 1.5 291 537 K6 PX0V
71 99 ... 21 5:34:44.5 -5:24:38 105 0.20 15.61 1.4 261 483 M4 PX00
73 99 V1118 ... 5:34:44.7 -5:33:42 329 0.18 14.04 0.4 217 779 M1 P0HV
81 97 ... 28 5:34:46.4 -5:24:32 272 1.34 12.91 1.9 261 454 M0 PX0V
121 99 ... ... 5:34:51.2 -5:16:55 200 0.34 14.90 3.0 316 541 M3 P00V
124 99 ... 64 5:34:51.8 -5:21:39 200 0.48 13.94 0.0 286 382 M3.5 PX0V
127 99 ... 66 5:34:52.1 -5:24:43 254 0.48 14.28 0.6 258 373 M3.5 PXHV
128 97 V1458 67 5:34:52.2 -5:22:32 215 0.39 12.64 >0.6 278 366 M2.5 PXHV
135 99 V1460 72 5:34:52.7 -5:29:46 11 0.40 14.45 0.1 223 522 M3 PX0V
147 99 ... ... 5:34:54.4 -5:17:21 26 0.30 14.80 1.4 318 489 M5 P000
151 99 ... 95 5:34:54.8 -5:25:13 332 0.15 15.44 1.6 251 341 M3 PX00
152 99 ... 96 5:34:55.1 -5:25:30 51 0.18 14.12 2.5 248 343 M3 PX00
176 99 KQ 123 5:34:57.8 -5:23:53 336 1.28 11.58 >1.1 264 280 K8 PX0V e
190 99 ... 134 5:34:59.3 -5:23:33 355 0.56 15.33 0.3 268 256 M6 PX00 f
201 99 ... 150 5:35:01.0 -5:24:10 246 1.09 13.41 0.8 258 235 M2.5 PXH0
222 98 V1320 174 5:35:02.2 -5:29:10 89 1.07 14.14 1.4 212 407 M2 PX0V
223 99 ... 177 5:35:02.4 -5:20:47 130 0.34 13.64 1.3 307 261 ... PX00
224 99 ... 180 5:35:02.7 -5:19:45 164 0.27 15.38 2.4 317 299 M1 PX0V
235 0 ... 197 5:35:03.6 -5:29:27 346 0.54 13.69 0.4 208 411 ... PXHV d
248 99 V1274 214 5:35:04.4 -5:23:14 320 0.90 12.73 >3.7 273 180 M3 PX0V e
... ... ... 260 5:35:06.2 -5:22:13 284 0.40 16.20 0.3 295 169 M4 0X00 f
296 99 ... 275 5:35:06.6 -5:26:51 196 0.89 14.30 3.2 215 255 M4.5 PX00
296 99 ... 275 5:35:06.6 -5:26:52 158 0.27 14.30 0.4 215 256 M4.5 PX00
305 99 ... ... 5:35:07.6 -5:24:01 302 0.45 14.58 0.3 254 137 M3 P00V f
355 0 ... 402 5:35:10.9 -5:22:46 197 0.42 15.03 3.5 294 90 K: 0X00
370 99 ... 452 5:35:11.9 -5:19:26 122 0.73 13.86 3.5 344 246 K1.5 PX00
383 99 V1492 489 5:35:12.7 -5:16:14 280 0.19 14.57 2.1 353 433 M3 PX0V
392 99 ... 498 5:35:12.7 -5:27:11 185 0.23 15.21 0.2 194 234 M6 PX00
391 99 ... 501 5:35:12.8 -5:20:44 94 0.29 12.97 >3.3 341 168 M1 PX0V
399 99 ... 523 5:35:13.2 -5:22:21 345 0.22 16.77 0.1 322 78 ... PX00 a,c
410 99 ... ... 5:35:13.2 -5:36:18 167 1.19 16.00 1.3 184 777 ... P00V
406 99 V1327 543 5:35:13.5 -5:17:10 271 0.95 13.88 2.6 353 375 M1 PX0V d
... ... ... 562 5:35:13.6 -5:21:21 62 0.24 16.27 1.5 341 129 M3 0X00 a
422 99 V1495 566 5:35:13.7 -5:28:46 74 0.31 14.43 0.0 187 326 ... PX0V
436 0 ... 620 5:35:14.3 -5:22:04 238 0.32 16.97 1.0 338 85 ... 0X00 a,c
439 99 V1329 626 5:35:14.5 -5:17:25 150 0.30 14.97 0.1 355 359 M1 PX0V
439 99 V1329 626 5:35:14.5 -5:17:25 78 1.21 14.97 4.1 355 359 M1 PX0V
444 99 ... 651 5:35:14.6 -5:16:46 190 0.18 15.53 1.0 356 398 ... PX00
445 26 ... 645 5:35:14.7 -5:20:42 191 0.44 14.67 1.8 351 163 ... 0X00 a
... ... V1500 ... 5:35:14.9 -5:36:39 82 0.38 14.45 2.2 182 797 ... 000V
465 97 V409 ... 5:35:14.9 -5:38:06 271 0.43 14.71 4.1 182 883 ... P00V
498 99 V1504 ... 5:35:15.8 -5:32:59 122 0.70 13.81 0.2 181 576 ... P0HV
509 99 ... 789 5:35:16.2 -5:24:56 46 0.49 15.06 0.1 182 93 ... PX0V a,b,f
511 99 ... ... 5:35:16.3 -5:22:10 241 0.41 15.67 2.4 358 73 M1 PX00 a,c,e,f
... ... ... 822 5:35:16.8 -5:17:17 84 0.26 16.37 0.5 1 366 ... 0X00
551 99 ... 881 5:35:17.4 -5:25:45 265 0.14 13.87 0.0 174 142 M1 PX00 a
551 99 ... 881 5:35:17.4 -5:25:45 262 1.27 13.87 3.6 174 143 M1 PX00
552 99 V410 897 5:35:17.5 -5:21:46 156 0.46 14.50 >1.8 9 99 ... PX0V a,c
560 98 V1334 927 5:35:17.9 -5:15:33 235 0.60 14.17 3.2 3 471 ... PX0V
570 99 ... 937 5:35:17.9 -5:25:34 0 0.15 14.73 0.8 170 133 ... PX00 a,b
566 0 ... 939 5:35:18.0 -5:16:13 214 0.86 14.93 0.0 3 430 ... 0X00 d
... ... ... 967 5:35:18.4 -5:24:27 78 0.42 15.16 2.4 155 70 M2.5 0X00 a,b,f
592 99 ... 974 5:35:18.5 -5:18:21 285 0.61 14.37 4.0 6 304 M2.5 PX00
597 0 ... 994 5:35:18.8 -5:14:46 307 0.17 14.02 0.1 4 519 ... 0XHV
... ... ... 998 5:35:18.8 -5:22:23 314 0.20 17.54 0.3 31 70 ... 0X00 a
... ... ... 1061 5:35:20.0 -5:18:47 71 0.22 17.31 0.1 11 281 M9 0X00
638 99 ... 1077 5:35:20.0 -5:29:12 354 1.01 17.03 3.2 171 353 ... PX0V
681 99 V1524 ... 5:35:21.4 -5:23:45 232 1.09 16.39 0.9 107 77 K7 PX0V a
687 54 ... 1158 5:35:21.7 -5:21:47 231 0.49 14.66 2.1 39 124 ... 0X0V a,c
709 99 ... 1202 5:35:22.2 -5:26:37 309 0.22 12.92 0.5 156 213 M0.5 PX00
722 98 ... 1208 5:35:22.3 -5:33:56 216 0.36 14.51 0.2 172 639 M4.5 PX00
727 99 V1528 1233 5:35:22.8 -5:31:37 285 0.73 13.87 4.7 169 503 M2 PXHV
748 99 ... 1279 5:35:24.1 -5:21:33 277 0.35 14.77 2.0 46 159 M3 PX0V a
767 99 ... 1316 5:35:25.2 -5:15:36 75 1.13 13.89 >4.1 16 485 M2.5 PX0V d
776 99 V496 1328 5:35:25.4 -5:21:52 73 0.50 14.20 1.8 56 162 ... PX0V a
777 80 ... 1327 5:35:25.5 -5:21:36 347 1.19 15.15 1.2 52 173 K6 0X00
783 95 ... ... 5:35:25.5 -5:34:03 283 0.40 14.17 0.8 168 656 ... PX00
797 99 ... 1363 5:35:26.6 -5:17:53 323 1.42 16.30 2.8 25 363 ... PX00
... ... ... 1425 5:35:29.5 -5:18:46 329 0.23 16.06 2.6 35 338 ... 0X0V
841 99 ... ... 5:35:30.0 -5:12:28 16 0.41 14.41 0.1 17 686 M4 P0HV
867 99 ... 1463 5:35:31.3 -5:18:56 212 0.29 12.19 >0.3 40 347 K8 PXHV
884 99 ... ... 5:35:32.4 -5:14:25 287 1.32 14.91 2.8 24 589 ... P000
893 99 ... ... 5:35:33.2 -5:14:11 208 0.15 14.07 0.0 24 606 ... P00V
906 99 ... ... 5:35:34.7 -5:34:38 293 1.39 14.17 2.7 158 728 M3 P00V
924 99 ... ... 5:35:36.5 -5:34:19 73 1.04 16.63 1.3 156 722 ... P000
945 99 ... ... 5:35:40.2 -5:17:29 46 1.41 12.51 3.6 45 501 B6 P000
P (Σ,Θ) = 1− e−piΣΘ
2
(2.1)
where Σ is the surface density of stars and Θ is the separation limit, in our case it was set to
14
Table 2.2: Other binaries with angular separations < 1.5′′ toward the ONC.
JW % GCVS COUP RA DEC PA SEP mI ∆mHα PA OriC SpT M Bin C
553 99 V1510 899 5:35:17.6 -5:22:57 113 0.36 12.41 1.4 34 31 K3.5 PX0V a 1
596 99 AF 986 5:35:18.7 -5:23:14 78 0.30 13.06 2.2 75 34 K3.5 PX0V 1
... ... ... 1085 5:35:20.2 -5:23:09 104 0.21 ... 0.4 76 57 ... 0X00 1
182 48 ... 127 5:34:58.0 -5:29:41 77 0.11 16.56 0.2 216 467 ... 0X00 2
290 99 ... 266 5:35:06.4 -5:27:05 84 0.11 15.48 1.3 214 268 ... PX00 2
... ... ... 1195 5:35:22.3 -5:18:09 227 0.11 15.67 0.8 15 326 ... 0X00 2
... ... ... ... 5:34:19.5 -5:27:12 168 0.56 10.52 0.1 255 881 G5 3
58 0 ... ... 5:34:41.8 -5:34:30 282 1.43 16.13 2.2 218 844 ... 3
61 46 ... ... 5:34:42.7 -5:28:37 153 0.45 13.52 0.9 238 594 ... 3
... ... ... ... 5:34:45.8 -5:30:58 324 0.46 ... 0.1 225 645 ... 3
... ... ... ... 5:35:01.4 -5:24:13 292 0.28 ... 0.6 257 231 ... 3
... ... ... ... 5:35:18.3 -5:24:39 96 0.67 16.72 2.2 160 81 ... 3
... ... ... ... 5:35:30.0 -5:34:31 237 1.29 ... 0.3 163 699 ... 3
Figure 2.4: Examples of the artificial binaries, created using the profiles of real stars of our sample, in order to
analyse our completeness.
1.′′5. Because Correia et al. (2006) do not describe in detail the meaning of this equation, we
decided to give more details about it in the following.
First, we need to know what is the probability of finding a companion to a primary star given
a certain separation Θ. We can translate this to the following equation (P stands for primary
and C for companion):
prob(P in field AND C within Θ) = prob(P in field)prob(C within Θ) (2.2)
The probability of the primary being in the field is 1, because we are using it as the central
star. In order to calculate the probability of finding a companion within the given radius, we
will use the Poisson distribution, because these are rare events and we have lots of trials. Using
the Poisson distribution we have:
prob(C within Θ) = e−µ = e−piΘ
2Σ (2.3)
where µ is the mean value and it is equal to the given area (piΘ2) multiplied by the surface
density of stars (Σ).
15
Figure 2.5: These two plots show the ∆m versus binary separation in arcseconds. The observed binaries are
shown in the left panel, where the dashed line indicates the 0.′′15 separation limit. The right panel shows the
artificial binaries, created from real stars present in our images, with different separations, position angles and
∆m. In the right panel, filled circles indicate systems where we could detect the companions, while open circles
represent those where we were not able to detect them.
Hence, the probability of finding a companion to a star, given a certain radius is:
prob(P in field AND C within Θ) = e−piΘ
2Σ (2.4)
Now that we know the probability of finding a companion, we only have to subtract 1 to find the
probability of a companion being a line-of-sight contamination, which was given in Eq. (2.1).
To calculate the surface density we applied two distinct methods. The first method consists
of counting the number of stars inside a circle of 30′′ radius centered on each star of our sample
and dividing this number by the area of the circle. This way we obtain a local density of stars
which are influenced by the subclusters in that specific area. The second method consists of
defining large circular areas starting from the exclusion zone. Each annulus has width of 30′′
and the density was again obtained counting the stars inside each area and dividing by the
area of the annulus. One problem with this method is the shape of our mosaic, which can
be seen in Figure (2.1). Because we centered each annular area in θ1 OriC, when the radius
increases beyond 60′′ we start to have part of the annulus outside our surveyed area. We fixed
this problem by calculating the area outside the mosaic, which means that the area used to
calculate the density takes into account only the amount inside the mosaic. The first annulus
has inner radius starting at 60′′ and outer radius at 90′′. The second annulus starts at 90′′ and
ends at 120′′. The other areas continue with a step of 30′′ each with the last annulus having an
outer radius of 1 020′′. In the end we came to the conclusion that the correct procedure should
be the first one because it was the only way to account for the subclustering observed in the Hα
images.
The value of 1.′′5 was chosen because it allows us to study the wide visual binaries (at
16
Figure 2.6: Left panel: surface density profile of the ONC members as a function of distance to θ1 OriC. The
exclusion zone is indicated by the dashed line. This profile was used as a guide to understand and localize the
subclustering in the region studied. It was not used to obtain the probability of chance alignment. Right panel:
probability that a star would have a companion as a line-of-sight association, obtained with Eq. (2.1). Presented
here are all the 781 ONC members, squares for single stars and circles for binaries. One can see a subclustering
around 180′′. We used the surface density obtained considering the number of stars inside an area of 30′′ centered
in each star.
this separation these systems are still gravitationally bound) without compromising the overall
contamination of false pairs. To obtain the number of false pairs we added up the probability of
being a chance alignment of all the stars (781 members). We have 9 flase pairs in our sample,
which means we have a contamination of 11% (9 false pairs among 781 members). Thus, the
superior limit (1.′′5) was chosen as the separation that minimizes the contamination by false
pairs but still provides systems that are likely to be physically related.
The left panel of Fig. (2.6) shows the ONC density profile, calculated using the second
method described above, while the right panel shows the probability of the companion being
a line-of-sight association. In both panels it is possible to see a subclustering in the analyzed
area. Since we are using a narrow band filter, which is extremely sensitive to the background
brightness, we cannot be sure if the subclustering is real or not. It exists in our Hα images,
however, there are many stars which were not detected in these images. This is another
completeness problem in our sample and it is an unquantifiable one.
2.5 Results
2.5.1 Binary Fraction
With our range set to 0.′′15 – 1.′′5, we have 72 binaries and 3 triples, members of the ONC. If we
correct this number for the contamination described previously and count 1 triple as 2 binaries
we end up with 69 physical binaries. Such a number implies a binary fraction of (8.8±1.1)% in
the interval 67.5 to 675 AU (0.′′15 to 1.′′5), where the error was estimated as 1σ based on Poisson
statistics. If we count one triple as a single system we can calculate the multiplicity frequency,
which gives us a value of (8.5±1.1)%, with errors estimated in the same previous way.
Analyzing the inner 40′′×40′′ area of the Trapezium cluster, Petr et al. (1998) found four
binaries in the separation range 0.′′14 – 0.′′50, which corresponds to a binary fraction of (5.9±4.0)%.
17
In the same separation range we have 50 binaries leading us to a value of (6.4±0.9)%. The
numbers agree inside the error bars, however one must remember that we do not analyse the
inner zone of the Trapezium.
Reipurth & Zinnecker (1993) observed nearby T Tauri associations and found 38 binaries out
of 238 stars in the range 150 – 1800 AU. The common range between our work and Reipurth &
Zinnecker is 150 – 675 AU. We calculated the binary fraction (counting the three triples as six
binaries) in this common range for both samples. The numbers are: (11.8±2.2)% for the nearby
T Tauri associations and (5.3±0.8)% for ONC. These numbers lead us to conclude that the
binary fraction in T Tauri associations is higher than in ONC by a factor of 2.2, in qualitative
agreement with Petr et al. (1998).
2.5.2 Separation Distribution Function
The left panel of Figure (2.7) presents the separation distribution function in an angular scale,
with bins 0.′′1 wide. There is a clear decrease in the number of binaries with separations larger
than 0.′′5. The same distribution, but in a logarithmic scale, is presented in the right panel of
the same figure, where we also plotted the binary contamination and the distribution of the
field stars based on DM91. The contamination by line-of-sight pairs was calculated using Eq.
(2.1), but with the separation parameter set to the separation of each pair, the probabilities
were added up for each bin. This approach is meaningless for individual objects but it can be
used to distribute the false binaries in each bin. For example, the total of the probabilities for
all the systems in the last bin in the right panel of Fig. (2.7) is 68%. This means that out of
the 9 false binaries in our sample, 6 are located in this bin. The bins with the larger separations
have a larger number of false binaries, as expected. The errors are indicated by the straight
lines and were calculated assuming a binomial distribution.
Using our whole range of separations we have a binary fraction of (8.8±1.1)%, while using
data from DM91, in the same range of separations, we estimated a binary fraction of 13.7%
and 12.4%, using a trapezoidal approximation and a linear interpolation, respectively. It was
necessary to convert the periods given by DM91 to angular separations. Since the stars analyzed
by DM91 are solar-like stars (F9-G6 V) we assumed a typical G2V spectral class and transformed
the periods to separations. Hence, the field has a binary fraction 1.5 higher than ONC.
2.5.3 Wide versus Close Binaries
Is there a difference between the number of wide binaries and close binaries? Previous studies
(e.g. Ko¨hler et al., 2006) tried to find such difference without success. The dramatic change in
the separation distribution function seen in the left panel of Fig. (2.7) lead us to choose this
separation as a dividing point. Binaries with angular separation in the range 0.′′15 ≤ separation
≤ 0.′′5 are called “close” and binaries with angular separation in the range 0.′′5 < separation ≤
1.′′5 are called “wide”.
To study the distribution of these systems as a function of separation to θ1 OriC we build
cumulative distributions like the ones shown in Fig (2.8). The left panel shows a simple
cumulative distribution as a function of distance to θ1 OriC, created by adding up the close and
wide binaries as the distance to θ1 OriC increases, in steps of 1′′. We can see in this figure that
there are close and wide binaries distributed in the whole range of distances. This means that
18
Figure 2.7: The left panel shows the histogram of binary angular separations in steps of 0.′′1. The number
of binaries in the innermost bin, with separations less than 0.′′1, is incomplete. We filled this bin with vertical
dotted lines. One can see the abrupt decrease in the number of binaries when the separation increases beyond
0.′′5. The right panel shows the logarithmic separation distribution function of the ONC binaries compared to
the distribution of field binaries from Duquennoy & Mayor (1991) across the separation range from 0.′′15 to 1.′′5.
The dot-filled part of the histogram bins represents the contamination by line-of-sight pairs, obtained using Eq.
(2.1). The bins representing large separations are the most contaminated ones, as expected. The vertical solid
lines present in each bin represent the 1-σ errors calculated using a binomial distribution. The data from DM91
within our separation range are shown as two dashed crosses, while the Gaussian distribution they fit to their
complete data set is represented by the dash-dotted line, which can be seen above the histogram. The top axis
indicates separations in arcseconds at the distance of the ONC.
close binaries are not exclusive of the center regions and that we can detect wide binaries close
to the center as probably as far from it. However the distributions are not the same.
The right panel in Fig. (2.8) shows a different situation. Here we have the ratio of wide to
close binaries as a function of distance to θ1 OriC. The cumulative distribution function was
calculated in the following manner: the first point was calculated dividing the total number of
wide binaries by the total number of close binaries with distances to θ1 OriC, starting in the
exclusion zone and going up to 30′′ from it (the exclusion zone is at 60′′ from θ1 OriC, which
means that the first step used stars with distances from θ1 OriC in the range 60′′ to 90′′), the
next step increased the distance in 1′′ and made the same calculation. In this way, as the curve
moves away from θ1 OriC it accounts for more and more binaries until it reaches the end of
the distribution, where we have the mean ratio of wide-to-close binaries for the ONC. In the
figure, the dashed lines indicate the 1σ errors and the dotted lines indicate the same ratio for
the DM91 binaries, assuming a Gaussian fit (lower line), and the actual data points (upper line).
The DM91 lines represent the ratio for field stars.
It is evident that there is a very pronounced and almost monotonic change in the ratio of
wide-to-close binaries as one moves away from the core of the ONC until a distance of about
460′′, at which point the ratio becomes flat. It is also clear that the mean ratio of wide-to-close
binaries for the whole ONC is lower than DM91 values.
The separation distribution function in the ONC most probably has not yet found its final
shape. Two classic mechanisms operate that can affect the orbit of a young binary: rapid
dynamical decay in small-N clusters (e.g. Sterzik & Durisen, 1998; Reipurth & Clarke, 2001;
19
Figure 2.8: The left panel shows the cumulative distributions of close (0.′′15 – 0.′′5) and wide (0.′′5 – 1.′′5) binaries
in the ONC as a function of distance from θ1 OriC. The right panel shows the ratio of wide to close binaries in
the ONC as a function of distance to θ1 OriC. The dashed lines indicate the errors. The dotted horizontal lines
at the top of the panel are the same ratio from the DM91 field binary study. The lower line comes from their
Gaussian fit and the upper line comes from their actual data points. The top axis scale shows the crossing time
in years, assuming a mean one-dimensional velocity dispersion of 2 km s−1. The vertical dot-dashed line indicates
the distance where the ratio becomes essentially flat, suggesting an age for the ONC of about 1Myr.
Bate, Bonnell & Bromm, 2002) and the passage of a binary through a dense cluster (e.g. Kroupa,
1995a,b, 2000; Kroupa, Petr & McCaughrean, 1999). The former occurs primarily during the
Class 0 phase (Reipurth, 2000) and is no longer relevant for stars in the ONC. The latter,
however, may be essential for understanding the binary population of the ONC. A wide binary
falling through the potential well of a cluster will gain kinetic energy through encounters and
binaries with weak binding energies are eventually disrupted (Heggie, 1975). A prediction of
this scenario is that binaries at distances from the cluster center larger than the corresponding
crossing time should not show any dynamical alterations due to encounters with other cluster
members (Kroupa, Petr & McCaughrean, 1999; Kroupa, Aarseth & Hurley, 2001). The crossing
time of a star through a cluster is tcross = 2R/σ, where R is the cluster radius and σ is the mean
one-dimensional velocity dispersion in the cluster. Assuming that this velocity dispersion in the
ONC is of the order of 2 km s−1 (e.g. Jones & Walker, 1988), we have indicated in the right
panel of Fig. (2.8) the crossing time for different distances to the cluster center. The figure
suggests that for distances larger than roughly 460′′ there is no longer a measurable change in
the ratio of wide-to-close binaries. Given that the wide-to-close binary ratio changes by a factor
of 4–5 from the inner to the outer regions, this suggests that many, and perhaps most, of the
wide binaries are disrupted after only a few passages by the cluster center. The variation of the
ratio of wide-to-close binaries from the inner to the outer regions of the ONC offers the first
compelling observational evidence that dynamical interactions in the dense central region of the
ONC have taken place.
In principle, the right panel in Fig. (2.8) allows us to determine the age of the ONC. An
angular distance of ∼460′′ corresponds to a crossing time of about 1 million years. However,
an age determined in this manner is directly dependent on the velocity dispersion assumed.
20
Therefore, all we can say about the age of 1 million years we estimate for the ONC is that it is
consistent with other ONC age estimates (e.g. Hillenbrand, 1997).
The significantly lower numbers of binaries that we find in the ONC compared to associations
thus appears to be due, at least in part, to the dissolution of wide binaries. However, under
certain circumstances, an encounter could lead to hardening of the binary, making it closer than
our resolution limit, so it is not lost to the overall binary budget.
Even in its outermost regions, the ONC shows a smaller ratio of wide-to-close binaries than
seen in the field by DM91. Considering that the majority of field stars are likely to have been
formed in a cluster, it follows that many of the wide binaries in the field must have formed in
the gentler environment of a loose T association.
Durisen & Sterzik (1994) found, on theoretical grounds, that binaries are more likely to form
in clouds with lower temperature. Reipurth & Zinnecker (1993) found observational evidence
that clouds with more stars have relatively fewer binaries in the separation range under study
(mostly wide visual binaries). Both of these results indicate that loose T associations produce
or retain more wide binaries than clusters do. However, Brandeker et al. (2006) noted that the
young sparse η Chamaeleontis cluster has a deficit of wide binaries. Unless this small cluster
is the remnant of a much denser cluster, this result would seem to be in contradiction to the
notion that wide binaries are preferentially formed/preserved in loose associations. In any case,
there is no question that the field binary population is a mixture of binaries formed in clusters
and in associations.
2.5.4 Flux Ratios and Substellar Companions
We have determined the flux ratios of the unsaturated binaries and present a histogram as a
function of ∆mag in Fig. (2.9).
Figure 2.9: Luminosity ratio of the binary population, based on the Hα fluxes of primaries and secondaries. All
binaries with separation between 0.′′1 and 1.′′5 are plotted, except the saturated ones.
21
The majority of the binaries in the ONC have unequal components, although the bin with
equal components (∆m < 0.5 mag) is the most populated. Only very few of the companions
have a magnitude difference between the primary and the secondary of more than 3 mag.
In the absence of spectroscopic information, we have made a crude attempt to investigate
the nature of the companions based on their measured flux ratios. In order to do that we have
made some assumptions. First, we assume that the observed flux ratios reflect photospheric
fluxes, in other words, that Hα emission-line fluxes are not seriously affecting the ratios. This
may not always be a good assumption, since mid-to-late M dwarfs often are very active and
their photospheric fluxes are so low that Hα line emission could be a significant contribution. If
we mistake Hα line emission for photospheric flux from the primary, our estimate of a spectral
type for a companion will be earlier than it really is.
Second, most published photometry of late-type dwarfs is broadband, whereas we have
observed in the narrow band Hα filter, so we assume that our observed magnitude differences
can be compared to R-band photometry. Since we are dealing with the difference between two
stars, this is probably not a bad assumption, at least when the flux ratio is not large.
Third, we assume that the observed flux ratios are not affected by differences in extinction
between the components. Given the young age and occasional association with molecular clouds,
this may not hold true in all cases.
With these caveats spelled out, we have used the MI versus spectral type relation and the
R − I colors for M dwarfs (Dahn et al., 2002) to derive the difference in R-band magnitudes
as a function of spectral type for M dwarfs. The relation turns out to be essentially linear,
with a mean drop of 0.56 mag per spectral subtype throughout the M spectral range, as can
be seen in Fig. (2.10). We then used the spectral classifications for the primaries provided
Figure 2.10: R-band magnitudes as a function of spectral type for M dwarfs and very cool objects (spectral
type L).
by SIMBAD, of mixed provenance, with our flux ratios to estimate a spectral type for the
secondaries. For a cluster with an age between 0.5 and 3 million years, the substellar limit is
22
around spectral type M6.25, following the models of Baraffe et al. (1998) and Chabrier et al.
(2000) and using the temperature scale of Luhman et al. (2003, 2006). To our surprise, quite a
number of the secondaries appear to be brown dwarfs, in at least one case forming a wide brown
dwarf-brown dwarf binary. Given the assumptions involved and the simplistic nature of these
spectral type estimates, it is obvious that spectroscopy is required to establish the true nature
of the secondaries. These brown dwarf and very low mass systems are very important to the
study of the low-mass end of the initial mass function. The detection of such objects in binary
systems is also very important for evolutionary and star formation models.
2.6 Summary of Results
We have analyzed a large set of Hα images of the ONC acquired with the ACS onboard the
HST with the goal of detecting binaries among a sample of 781 ONC members. The following
results were obtained:
1. A total of 75 binaries and 3 triple systems were detected, of which 55 are new discoveries.
2. Within the limited angular range of 0.′′15 – 1.′′5, corresponding to projected separations
of 67.5 – 675 AU, we have found a binary fraction of (8.8±1.1)% after correcting for a
statistically determined contamination of nine line-of-sight binaries.
3. The field binary fraction for solar-type stars in the same separation range is 1.5 times
larger and for T Tauri associations it is 2.2 times larger than in the ONC, confirming with
more significant data the earlier results that the ONC is deficient in binaries.
4. The separation distribution function for young binaries in the ONC shows a dramatic
decrease in binaries for angular separations larger than 0.′′5, corresponding to projected
separations of 225 AU.
5. The ratio of cumulative distributions of wide to close binaries shows an increase out to a
distance of about 460′′ from the center of the ONC, after which it levels out. We interpret
this as clear observational evidence for dynamical evolution of the binary population as a
result of passages through the potential well of the ONC. These results are consistent with
an age of the ONC of about 1 million years.
6. It appears that, at least in part, the deficiency of binaries in the ONC compared to the field
star population, can be understood in terms of the destruction of wide binaries combined
with a secondary effect from orbital evolution of binaries toward closer separations that
are unobservable in direct imaging surveys.
7. Limited spectral information about the primary stars indicated that they are low-mass
T Tauri stars, except for one Herbig Ae/Be star. Assuming that most binaries are not
affected by differential extinction, we find that possibly as many as 50% of the binaries
have companions with spectral types later than M5 and from one-sixth to one-third of the
binaries may have substellar companions, all of which have separations of at least 50 AU.
This large number of wide substellar companions is of interest for theories of brown dwarf
formation.
The environment of the loose T associations seems to favor the survival of the wide binary
population, while the dense environment of the star forming regions like the ONC, seems to
be responsible for the disruption of part of these wide binaries. It is reasonable to suppose
23
that the wide binaries (mainly the late-type ones) that populate the field were born in loose T
associations and therefore survived, while the single late-type field stars were born in multiple
systems in dense star forming regions, which were disrupted after some close encounters in their
dense environment. Thus, regions like the ONC would contribute with single stars to the field
population, while T associations (like Taurus-Auriga, Lupus, Ophiuchus, etc) would contribute
with multiple systems, as well as with single stars.
The discovery of visual binaries, in a star forming region like the ONC, is very important
because we can try to determine precise stellar parameters of young stars. Although the ONC
is ∼450 pc away, astrometric observations of these binaries, with the ACS onboard the HST for
example, can be used to determine their projected orbit and in this way obtain mass, radius,
etc. Similar studies have already been conducted for other visual binaries (e.g. Chauvin et al.,
2005; Neuha¨user et al., 2008).
2.7 Individual Binaries of Interest
Below, we describe some of the interesting objects. We will assume that the difference in
magnitude between primary and seconday is not affected by extinction differences.
V1438 Ori (JW 39). The spectral type of M3 is from H97. The 2.6 mag difference implies
that the companion is an M8 brown dwarf at a projected separation of 170 AU. Stassun et al.
(1999) detected lithium in the system.
JW 71. The spectral type of M4 is from H97. The 1.4 mag difference implies that the
companion is an M7 brown dwarf with a projected separation of 90 AU.
V1118 Ori (JW 73). This is a member of the EXors class (Herbig, 1989). The star,
which is also know as Chanal’s Object, has had five outbursts since its discovery in 1984 and
typically varies within the range 14.5 < V < 18, with a rise time of less than a year and a
declining phase about twice as long. Optical and infrared spectroscopy of V1118 Ori is reported
by Parsamian, Mujica & Corral (2002) and Lorenzetti et al. (2006), and x-ray observations are
analyzed by Audard et al. (2005). The discovery that V1118 Ori has a companion only 0.4 mag
fainter in the Hα filter raises interesting questions about which of the components may drive
the variability (Herbig, 2008).
JW 121. The spectral type of M3 is from H97 and the 3.0 mag difference implies that the
companion is an M8 brown dwarf with a projected separation of 153 AU.
JW 147. The spectral type M5 is from H97 and the 1.4 mag difference implies that the
companion is an M8 brown dwarf with a projected separation of 135 AU.
JW 152. The spectral type of M3 is from H97 and the 2.5 mag difference implies that the
companion is an M7 brown dwarf with a projected separation of 81 AU.
JW 235. It is associated with object HH 504, discovered by Bally & Reipurth (2001); see
also O’Dell (2001). The binarity of the star was discovered by Ko¨hler et al. (2006). Its proper
motion in the JW catalogue implies that it has a very low probability of being a member of
the ONC. However, given similar brightness of the two components (0.4 mag), it is likely that
brightness variations of the stars shifted the photocenter used for astrometry. Evidence for ONC
membership is strong and is based on the presence of Hα emission, detection in X-rays, optical
variability and association with an HH object.
24
V1274 Ori (JW 248). The spectral type of the primary is somewhat in dispute. H97
suggests M0.5-M2 (with the Ca II lines in emission), Edwards et al. (1993) suggest M3 and
Meeus & McCaughrean (2005) suggest M0-M4. The binarity of the system was found by the
COUP survey. Our primary is the secondary in the COUP catalogue, indicating that our
secondary is highly X-ray active. Sicilia-Aguilar et al. (2005) detected lithium in the (optical)
primary and found an inverse P Cygni profile at Hα. If there is not an extinction difference
between the primary and the secondary, then the large flux difference (> 3.7 mag) suggests that
the companion could be an L0 brown dwarf. However, the system is only 3′ from θ1 OriC, so
there is a 1% chance of a line-of-sight association.
JW 296. It is the primary of a hierarchical triple system. The secondary and tertiary
are very faint stars, surrounded by a common proplyd-like envelope know as 066-652 (O’Dell
& Wong, 1996). The primary is of spectral type M4.5 according to H97 and if the brightness
difference is taken at face value, the secondary and tertiary should be L0 brown dwarfs. Althou-
gh, given the obvious association with nebulosity it is likely that their faintness is merely due
to extinction.
JW 355. The primary of this system is located at the edge of a proplyd known as 109-246
(O’Dell & Wong, 1996). The proper motion of the star suggests it is not an ONC member but it
is found to be a member by the COUP project. H97 suggests a mid-K spectral type. The system
is only 90′′ from θ1 OriC and the star density is so high that the possibility of a line-of-sight
association is several percent.
JW 370. There is a silhouette disk close (∼3.′′4) to this binary. Given the local density of
stars around this system, the probability that such a silhouette disk is just a chance alignment
is ∼7%. The chance of a line-of-sight alignment for the binary itself is 1.5%. H97 suggests a
spectral type of K0-K3.
V1492 Ori (JW 383). The primary is of spectral type M3 according to H97. The 2.1
mag brightness difference implies that the companion is an M7 brown dwarf with a projected
separation of 86 AU. Stassun et al. (1999) derived a rotation period of 7.11 days for the primary
and noted the presence of lithium in the spectrum.
Parenago 1806 (JW 391). The primary is of spectral type M1 according to Edwards
et al. (1993) and H97, who also noted the presence of Ca II emission. The primary is saturated
in our images, but with a magnitude difference of at least 3.3 mag, the companion will be an M7
brown dwarf. However, the pair is located in a region of high stellar density, so the probability
of a chance alignment is more than 5%.
JW 509. This binary was first detected by Prosser et al. (1994) and subsequently by
Padgett, Strom & Ghez (1997) and Lucas, Roche & Tamura (2005). Although it is a well known
binary it does not have a spectral type in the literature. Since this system is located in a crowded
region, the probability of a chance alignment is ∼ 2.7% but the reality of the binary is supported
by an interaction zone between the stars, visible in our Fig. (2.3), first row, third stamp.
Parenago 1914 (JW 551). The primary of this hierarchical triple system has a spectral
type M1 according to H97. The tertiary is not detectable either in the optical or in X-rays,
and since its separation from the primary is as large as 1.′′27 it could be a background object.
The system seems to be part of a small subcluster, which gives it a high probability of a chance
alignment (∼2.2%). Given the primary spectral type and the difference in magnitudes to the
tertiary (3.6 mag) and assuming that the tertiary is not a background object, the tertiary could
25
have a spectral type of M7 or M8 and thus would be a brown dwarf candidate.
COUP 967. This star was catalogued as a binary by Prosser et al. (1994), Padgett, Strom
& Ghez (1997) and Lucas, Roche & Tamura (2005). Our image shows that the primary is
associated with a proplyd (184-427 in O’Dell & Wong (1996)) and the secondary is a faint
object (∆mHα = 2.4). The spectral type for the primary is M2.5 according to H97. The system
is very close to θ1 OriC (∼ 70′′) and the probability of a chance alignment is 3.5%. If the
components have the same extinction, then the secondary would be of spectral type M6.5 and
thus at the hydrogen-burning limit.
JW 592. The spectral type is M2.5 according to H97 and M4 according to Edwards et al.
(1993). The 4 mag difference implies an M9.5 spectral type or later for the secondary, making
it a brown dwarf with a projected separation of ∼ 275 AU.
COUP 1061. This very close system (∼ 100 AU) has a combined spectral type M9-L0
according to Meeus & McCaughrean (2005), who did not resolve it, and since the components
have virtually the same brightness (0.1 mag), this is a bona fide brown dwarf-brown dwarf binary
with a projected separation of 100 AU.
V1524 Ori (JW 681). This binary was first catalogued by Prosser et al. (1994), according
to which it has spectral type K7. The secondary is associated with the proplyd 213-346 (O’Dell
& Wong, 1996). In the H97 catalogue this system has two entries, both with number 681 but
with different coordinates. SIMBAD identifies three different objects: H97b 681, H96b 681a and
H97b 681b. The first is JW 681, also known as V1524 Ori and is the primary in this pair. The
second is the unrelated object MLLA 312 and the third is the X-ray source COUP 1149, which
forms the secondary component, associated with the proplyd. Because this system is located
near θ1 OriC (∼77′′) the probability of a chance of alignment is 3.5%.
V1528 Ori (JW 727). The spectral type M2 is from H97 and the difference in brightness
suggests a spectral type of L0 for the companion if the components have the same extinction.
If so, the secondary is a brown dwarf with projected separation of 329 AU. The primary shows
Hα emission according to our unpublished survey.
JW 748. This binary was discovered by Prosser et al. (1994) and the spectral type of the
primary is M3 according to H97. Given the brightness difference the secondary could be a brown
dwarf with spectral type M7 and a projected separation of 158 AU.
JW 767. First discovered by Ko¨hler et al. (2006) this system has spectral type M2.5 from
H97. The secondary has a silhouette disk and is associated with the object HH 668 (Bally
et al., 2006). The extinction caused by the silhouette disk probably is responsible for the large
difference in brightness of more than 4 mag between the primary and secondary.
Parenago 2075 (JW 867). This object was observed by Ko¨hler et al. (2006) but despite
its brightness and angular separation (0.′′29) they did not resolve it, suggesting possible major
variability of the secondary. The system presents evidence of Hα emission, X-rays and optical
variability. A spectral type of M1 was suggested by Blanco (1963) and more recently Duncan
(1993) assigned a spectral type of K8 Ve. Sicilia-Aguilar et al. (2005) reported lithium in the
spectrum of the primary.
JW 906. The primary has spectral type M3 according to H97 and the 2.7 mag difference
implies that the secondary is a brown dwarf with spectral type M8 and a large projected
separation of 625 AU. Sicilia-Aguilar et al. (2005) reported lithium in the spectrum of the
primary.
26
Chapter 3
Variability
3.1 Introduction
Variability is an inherent characteristic of the stars. We can say that all stars are variable, in
different stages of their lives, with different amplitudes and time scales. During its lifetime, a
star travels along different paths in an HR diagram because it presents different characteristics.
However, it is impossible for us to observe a star as it evolves from its parental cloud, through
the main sequence towards its final moments as a white dwarf, a neutron star or even a black
hole. What we can do is study different stars in different evolutionary stages and try to put the
puzzle together in a coherent frame.
We will call “variable stars” those which present brightness variability in a time scale that
we can measure, be it seconds, days, years or decades, with amplitudes that we can measure
using CCDs, photometers or even photographic plates. The time scale, shape of the light curve
and amplitude of the brightness variations can be used to classify variable stars. Spectral type,
luminosity class and chemical composition are also important parameters to help us classify the
stars according to the origin of the variations.
Variable stars can be divided into two major groups: intrinsic variables and extrinsic varia-
bles. Intrinsic variable stars are those that vary due to physical processes in the star itself, while
extrinsic variable stars are those that vary due to processes external to the star.
Examples of intrinsic variables are the pulsating variables (there are many kinds of pulsating
variables, young ones, old ones, dying ones) and the eruptive (or explosive) variables (flare
stars, novae, supernovae). Extrinsic variables are, for example, eclipsing variables (the eclipse
may be due to a companion star, a planet or even the circumstellar material in young or very
old systems) and rotating variables (which include pulsars).
It must be clear that one kind of variability does not exclude the other. We can have, for
example, an eclipsing pre-main sequence star in which one of the components (or even both)
presents pulsations.
27
3.2 Types of Variables
The variable stars we are mainly interested in are the eclipsing binaries, irregular, pulsating pre-
main sequence and rotating variables. The light curve of some eclipsing binaries is very different
from other variables, because it presents itself constant except for its minima. The light curves
for irregular variables cannot be classified or grouped in any class, hence their name. The
pulsating and rotating variables can present very similar light curves and can be misclassified if
one uses only the shape of the light curve as a parameter for classification. Stellar properties
like the spectral type, luminosity class, etc, can help us distinguish between these two groups.
The four groups, listed above, have different characteristics and statistical indices can be used
to separate the sample.
Absolute parameters such as radius, mass and temperature with high precision (≤ 1%) can
only be obtained for spectroscopic eclipsing binary stars and are very important for putting
constraints on the models of star formation and evolution. There are only 22 pre-main sequence
stars in the literature, with absolute parameters that are well determined, through the use
of dynamical techniques (Mathieu et al., 2007; Stassun et al., 2008). The discovery of more
pre-main sequence eclipsing binaries is in demand.
Given that pre-main sequence stars (T Tauri and Herbig Ae/Be stars) still possess material
from their original environment and some of them are still accreting and/or ejecting matter, the
probability of finding a star with irregular photometric variability is quite high. Young stars
can also present flares which will cause an increase in brightness or present drops in brightness
caused by some obscuration in the line of sight. The physical cause of such events can be, for
example, the accreting material.
Spots on the photosphere of a star can also cause drops in the brightness. If we add to it
the rotation of the star, we end up with a modulation in the light curve that can be related to
the period of rotation. Many young stars in the Orion Nebula have their period determined in
that way (e.g. Rebull, 2001).
A few years ago pulsating pre-main sequence stars got the attention of the astronomical
community (Kurtz & Marang, 1995). These stars occupy a locus in the HR diagram known
as the instability strip, where many post-main sequence pulsating stars are also observed.
These variables are important for studying the interior of young stars and the determination of
pulsation modes and periods. They can be used as constraints to evolutionary models.
It is impossible to predict what kind of variable stars we will find in the Orion Nebula and
Cygnus OB2 surveys. We will probably have a mixture of different types.
3.3 Statistical Indices
In order to distinguish between different types of variability it is necessary to use some statistical
indices. Some of them can be applied directly to the light curve of a star, while others are suitable
to be applied to residual curves or even to mean values. In this work we will use the following
set of statistical indices:
1. mean magnitude;
2. standard deviation;
3. geometric deviation;
28
4. mean absolute deviation;
5. absolute value of the maximum amplitude variation;
6. maximum deviation from mean magnitude;
7. R-statistics;
8. skewness;
9. kurtosis.
1 - Mean magnitude:
m =
1
N
N∑
i=1
mi (3.1)
The mean magnitude of each object is obtained through a simple mean involving all detections.
It is used in other statistical indices, like the standard deviation, geometric deviation, mean
absolute deviation, maximum deviation from mean magnitude, R-statistics, skewness and kur-
tosis. We will see later that the study of the mean magnitude curve can help us in the
characterization of variable stars.
2 - Standard deviation:
σ =
√√√√ 1
N − 1
N∑
i=1
(mi −m)2 (3.2)
where N is the number of detections, mi is the differential magnitude in the i
th frame and m is
the mean magnitude. The standard deviation is useful in the detection of measurements that
clearly stand out of the normal behavior of the star. It is sensitive to large variations.
3 - Geometric deviation:
σg = exp
(
1
N
N∑
i=1
ln |mi −m|
)
if |mi −m| ≥ 0.001, (3.3)
as in Eq. (3.2), N is the number of detections, mi is the differential magnitude in the i
th frame
and m is the mean magnitude. The geometric deviation is the geometric mean of the absolute
deviations greater than 0.001. The geometric deviation is sensitive to small variations and it
is useful to detect constant stars. The smaller the deviation is from the mean, the smaller the
value of the geometric deviation is and that is the reason why we have set a limit of 0.001 to
the variation of the measurements. We can see in Eq. 3.3 that if the measurement is equal to
the mean magnitude we will have a null value in the argument of the neperian logarithm.
4 - Mean absolute deviation:
σabs =
1
N
N∑
i=1
|mi −m|. (3.4)
The mean absolute deviation is defined as the simple mean of the absolute deviations. In Eq.
(3.4), N is the number of detections, mi is the differential magnitude in the i
th frame and m is
the mean magnitude.
29
The standard deviation (σ) is very sensitive to very different values of the magnitude while
the geometric deviation (σg) behaves in the opposite sense, it is not sensitive to great differences
but it is sensitive to small differences. The geometric deviation is ideal to search for constant
stars. The mean absolute deviation (σabs) has a value between the standard deviation and the
geometric deviation. We can use it in stars that are not constants and also that do not present
great variations.
5 - Absolute value of the maximum amplitude variation is exactly what its name
indicates, it is the maximum variation among all the detections. It is useful to detect stars with
great variations.
6 - Maximum deviation from mean magnitude: yields the largest difference among
the detections of an object and its mean magnitude, it can be positive or negative. It is useful
to separate flare stars from eclipsing binaries.
7 - R-statistics:
The R-statistics (R from residuals) was introduced by Baptista & Steiner (1993) and im-
proved by Bruch (1999). It is an adimensional number that measures the systematicity of a set
of residuals, being zero if the data has a random distribution around its mean value, close to
1 if the data has a systematic behavior and close to -1 if the data has systematic deviations
alternating between positive and negative.
R =
1
N − 1
N∑
i=2
(mi −m)(mi+1 −m)
σ2res
, (3.5)
where N is the number of detections, mi is the differential magnitude in the i
th frame, m
is the mean magnitude and σres is the standard deviation of the deviations (mi − m) shown
by the object’s measurements (σ2res measures the variance of the residuals). If we have a
statistically significant number of measurements then R can be associated with the probability of
the deviations being systematic or random. If a star presents periodic variability, the R-statistics
value should be close to -1, indicating that the star’s brightness oscillates about a mean value.
8 - Skewness:
γ1 =
µ3
σ3
, (3.6)
where µ3 is the third moment about the mean and σ is the standard deviation. Skewness is
useful to indicate if deviations from the mean are positive or negative. The moment about the
mean (µk) for a random variable X is defined as:
µk = (X −X)k (3.7)
Distributions with negative skewness have left tails longer than right tails while distributions
with positive skewness have right tails longer than left tails. Usually the models assume that
data points have a normal distribution, meaning that they are symmetric about the mean and
the skewness is zero. In reality the data point distribution is not symmetric and the skewness can
help us determine how the data is distributed about the mean. If a star has negative skewness it
implies that the measurements showing high deviations are located in the left tail, which means
lower magnitude values (the star is brighter). A star that has a positive skewness presents the
30
measurements with high deviations in the right tail, which means higher magnitude values than
the mean (the star is fainter). We expect to see eclipsing binaries with high positive skewness
values.
9 - Kurtosis:
γ2 =
µ4
σ4
− 3, (3.8)
where µ4 is the fourth moment about the mean and σ is the standard deviation. The minus 3
is a correction to make the kurtosis of the normal distribution equal to zero. There is a lower
limit of -2 for the kurtosis, but there is no upper limit, it could in fact be infinite.
Kurtosis is a measure of whether the data is peaked or flat relative to a normal distribution,
which has kurtosis equal to zero. A positive kurtosis means that the distribution in question
has a sharper peak around the mean than the normal distribution. In other words, the values
around the mean have a higher probability than in a normal distribution. A negative kurtosis
means that the distribution has a less acute peak around the mean, the values around the mean
have a lower probability than in a normal distribution. The Laplace distribution, for example,
has a positive kurtosis while the discrete and continuous distributions have negative kurtosis.
Stars with irregular variability will present a kurtosis value lower than stars with well defined
variability, for example, eclipsing binaries. Because the detached eclipsing binaries spend most
of their time with a constant magnitude, their kurtosis tend to have high values.
3.4 Testing the Indices
In an attempt to see the effect of different types of variability in the statistical indices we will use
real light curves, obtained by the All Sky Automated Survey1 (ASAS) (Pojmanski, 1997). We
will use light curves of α2 Canum Venaticorum stars, Cepheids, δ-Scuti stars, detached eclipsing
binaries, RR Lyrae stars and unknown type variables.
Figure (3.1) shows an example of typical light curves for each of these six variable types.
The period (in days) for each star is indicated in the panels. The ASAS gives the V band
magnitudes and the periods, which we used to construct these phase diagrams. The last panel
shows a variable star of unknown type, with a trial period indicated by the ASAS database.
We selected five stars from each variable type and an IDL routine was written in order
to calculate the statistical indices for the whole sample. The statistical index values for all
the variable stars are presented in Table (3.1). The first column presents the object’s mean
magnitude, while the period in days is shown in the second column. The σ, σg and σabs are
shown in columns 3, 4 and 5. The R-statistics is shown in the sixth column while the seventh
column shows the empiric discovery probability (ω). The skewness and kurtosis are presented
in columns 8 and 9.
It is possible to see that the final values of skewness and kurtosis of the eclipsing binaries are
always positive and larger than 1.0, as we expected, while the final values of the same indices
for the RR Lyrae variables are always negative, except for the skewness value of the third star.
All the Cepheid stars present negative kurtosis, an indication that the measurements are more
scattered about the mean. The only other variable class that presents similar statistical indices
1http://www.astrouw.edu.pl/asas/
31
Figure 3.1: Examples of the light curves of the variables stars used to test the statistical indices. The plots show
the star’s V magnitude versus the phase. The period, indicated in each panel in days, was obtained in the ASAS
database. Each panel has a label at the top indicating the variable type; starting in the first row we have the
α
2 Canum Venaticorum (ACV), Cepheids (CEP), δ-Scuti (DSCT), eclipsing binary (EB), RR Lyrae (RR) and
unknown type variable (Unk).
to the Eclipsing binaries is the δ-Scuti class, although some δ-Scuti show negative indices. This
is an interesting feature since the light curves of these classes are completely different. The
kurtosis of eclipsing binaries is expected to have high positive values because most of the time
the star has an almost constant magnitude, only changing in the occurrence of eclipses. The
R-statistics index for the unknown type variables presents values close to unity, meaning that
these stars indeed present some systematic behavior. The skewness index, for the same class of
stars, presents values close to zero, which are expected for a normal distribution of data points.
A very interesting behavior can be seen for Cepheids and eclipsing binaries. Their period
seems to be correlated with the R-statistics value: the greater the period the higher the R-
statistics value. However, after a more detailed study using a larger sample (18 stars) we
concluded that there is no correlation between the period and the R-statistics.
All these values are obtained with the total light curve of each variable. In other words, they
are static numbers that do not measure the dynamics of the variability. Instead of calculating
these indices using the total light curve we will calculate them every time a new point is inserted
in the database. In the end we will have not a light curve but σ-curves, σg-curves, and so on.
32
Table 3.1: Statistical index values for all the variable stars in the test sample.
Mag P (d) σ σg σabs R Skewness Kurtosis
α2-Canum Venaticorum
9.135 4.1022 0.0388 0.0262 0.0335 −0.1227 −0.6851 −0.8437
11.323 2.4734 0.0549 0.0273 0.0423 0.1440 0.1017 0.5296
10.424 1.1288 0.0458 0.0317 0.0401 −0.1126 −0.0143 −1.1253
9.023 2.0787 0.0326 0.0169 0.0263 0.1442 −0.3499 −0.4929
10.963 1.9038 0.0490 0.0269 0.0409 0.2035 0.2660 −0.6990
Cepheid
12.444 65.70 0.3429 0.2322 0.2989 0.7903 0.3982 −0.9581
13.583 29.09 0.3192 0.1893 0.2659 0.3487 0.6048 −0.2670
13.154 33.36 0.3603 0.2207 0.3031 0.6327 0.3068 −0.7508
11.316 2.21 0.2060 0.1159 0.1739 0.0062 −0.1843 −1.1133
13.468 22.33 0.3072 0.1832 0.2596 0.4726 −0.0579 −0.8913
δ-Scuti
10.014 0.122072 0.1051 0.0477 0.0791 −0.0085 −0.7398 0.3976
13.727 0.130639 0.2393 0.1134 0.1756 0.0135 −1.0912 10.4336
11.446 0.062784 0.3925 0.0653 0.1270 0.0071 8.6554 80.5066
11.644 0.194607 0.1352 0.0880 0.1161 0.0929 0.4668 −0.9158
9.725 0.091002 0.1361 0.0616 0.0873 0.1712 8.2930 136.6497
Eclipsing Binary
11.843 1.462024 0.3385 0.1190 0.2025 0.0291 2.9088 8.5518
11.723 0.46322 0.1623 0.0549 0.0981 0.0451 2.6532 7.2993
10.963 8.715623 0.0584 0.0161 0.0310 0.1552 3.9536 19.3385
11.690 11.045 0.1216 0.0557 0.0869 0.2483 1.6225 2.2106
8.768 2.0846 0.2882 0.0822 0.1180 0.1174 9.7805 111.6151
RR Lyrae
10.311 0.49337 0.3515 0.2320 0.2985 0.44994 −0.84352 −0.5270
12.257 0.52974 0.3126 0.2007 0.2611 0.02951 −0.70139 −0.2916
13.846 0.74407 0.2626 0.1485 0.2141 0.11214 0.38185 −0.3857
13.557 0.51716 0.3468 0.2119 0.2921 0.17726 −0.31978 −0.7885
10.169 0.49545 0.3835 0.2576 0.3319 0.69940 −0.57875 −0.9648
Unknown Variable Type
9.921 25.67 0.0914 0.0453 0.0702 0.7556 0.1622 0.3296
11.307 59.5 0.2791 0.1787 0.2382 0.9032 0.1945 −0.9385
9.543 43.1 0.1025 0.0518 0.0807 0.7095 0.0836 0.0484
11.755 100.6 0.1792 0.1200 0.1556 0.8183 0.1697 −1.0165
11.718 41.6 0.1757 0.0913 0.1350 0.8875 1.2236 1.0639
.
Figure (3.2) shows the statistical indices for the first α2-Canum Venaticorum variable star.
The values for the statistical indices calculated using the whole light curve are listed in the first
row of the α2-Canum Venaticorum section in Table (3.1). Starting in the first row of Fig. (3.2)
we have, from left to right, the observed magnitude, the mean magnitude curve and the standard
deviation curve. In the second row we present the mean absolute deviation curve, the skewness
and the kurtosis curves applied to the magnitudes. The third row shows the skewness and
kurtosis curves applied to the mean magnitude and the R-statistics curve. We can see clearly in
these panels that the statistical indices are not static, they change along the observations. The
deviations are more visible at the beginning of the curves and this can be easily explained by
the fact that in the beginning we have few points to use in the statistic.
In this variable star we see a large variation in the mean magnitude at the beginning of
33
Figure 3.2: Statistical indices applied to an α2-Canum Venaticorum type star as a function of time. In the first
row we present, from left to right, the real light curve of the star, the Mean Magnitude curve and the σ-curve.
The second row shows the σabs-curve, the skewness curve and the kurtosis curve. These last two curves were
obtained using the magnitude of the object. The third row presents the skewness and kurtosis curves, obtained
using the Mean Magnitude curve, and the R-statistics curve.
the observations, which continues, but with a smaller amplitude, throughout the observations.
The σ and σabs decrease with time and they seem to reach a constant value at the end of the
observations. Skewness and kurtosis applied to the magnitudes oscillate around a negative value
without a clear tendency. On the other hand, the skewness and kurtosis applied to the mean
magnitudes show a tendency to increase without apparent limit. An interesting feature can be
seen in the beginning of the observations, where a decrease in the skewness curve correlates
with the increase in the kurtosis curve. The R-statistics curve also shows large variations at the
beginning but at the end seems to be oscillating around the final value.
Figure (3.3) shows the statistical indices applied to the light curve of the first Cepheid star in
34
Figure 3.3: Statistical indices applied to a Cepheid type star as a function of time. The indices are the same as
in Fig. (3.2).
Table (3.1). We can see large variations at the beginning of the observations but after some time
the mean magnitude, the σ and σabs tend to a constant value. The skewness and kurtosis applied
to the magnitudes still present small scale variations and oscillate around a final value. The
skewness and kurtosis applied to the mean magnitudes present opposite behavior, the skewness
decreases while the kurtosis increases, both without reaching a constant value. The R-statistics
curve starts at ∼−0.5 and quickly reaches the value around which it will oscillate until the end
of the observations (∼0.8). The Cepheid variables show on average high values of R-statistics.
Figure (3.4) shows how the statistical indices evolve along the observations of the first δ-Scuti
variable star in Table (3.1). The mean magnitude curve of the δ-Scuti star did not present a
constant plateau, even after more than 200 measurements. The σ and σabs seem to have reached
a constant value right at the beginning of the observations. The skewness curve began with a
35
Figure 3.4: Statistical indices applied to a δ-Scuti type star as a function of time. The indices are the same as
in Fig. (3.2).
high positive value, close to 1, and decreased to values around −0.5 after ∼50 observations, where
it stayed almost constant until the end. The kurtosis shows more variations at the beginning but
after some time it also stayed almost constant about a positive value. The skewness and kurtosis
applied to the mean magnitudes present opposite behavior, like the Cepheid. The R-statistics
curve presents a large variation at the beginning of the observations. It starts with value ∼−0.4
and after ∼30 measurements suffers a fast increase, reaching positive values. It presents some
oscillations but with smaller amplitudes than at the beginning.
Figure (3.5) shows the same statistical indices but for the first eclipsing binary in Table
(3.1). A peculiar fact about the statistical index curves of an eclipsing binary is their segmented
aspect. Only the skewness and kurtosis curves of the mean magnitudes did not present a
segmented aspect. The segmentation can be understood if we consider that the eclipsing binary
36
Figure 3.5: Statistical indices applied to an eclipsing binary as a function of time. The indices are the same as
in Fig. (3.2).
has a constant magnitude except during the eclipses. Because we do not observe the eclipse
continuously, we have sudden variations of brightness, as exemplified in the first panel of the
first row in Figure 3.5, where we can see that most of the observations have a similar value
except for those measured during the eclipses. These sudden brightness variations affect the
statistical indices in a very peculiar way, that can be used to classify the eclipsing binaries in a
large sample of light curves. All curves show a large variation around the observation number
50 and the most affected index is the R-statistics.
Figure (3.6) shows the statistical indices applied to the light curve of the first RR Lyrae star
in Table (3.1). It is interesting to note that although the amplitude of the brightness variation is
considerable (∆m=1.4) the mean magnitude oscillates with a small amplitude around a constant
value from the middle to the end of the observations. The σ and σabs curves present small
37
Figure 3.6: Statistical indices applied to an RR Lyrae type star as a function of time. The indices are the same
as in Fig. (3.2).
variations, and like the mean magnitude, reach an almost constant value from the middle to the
end of the observations. The skewness and kurtosis applied to the magnitudes present opposite
behavior in the beginning but also reach almost constant values after 100 observations. The
skewness and kurtosis applied to the mean magnitudes have opposite behavior.
Figure (3.7) presents the statistical indices for the first unknown type variable star in Table
(3.1). We can see large scale variations in all index curves. The mean magnitude seems to reach
an almost constant value from the middle to the end of the observations. The skewness and
kurtosis applied to the mean magnitude show a large variation with opposite behavior but from
the middle to the end of the observations they have the same behavior, a tendency to increase.
The R-statistics curve reached an almost constant value, ∼0.8.
Cepheids, δ-Scuti and RR-Lyrae stars have a similar behavior for the skewness and kurtosis
38
Figure 3.7: Statistical indices applied to the light curve of a star with unknown variability as a function of time.
The indices are the same as in Fig. (3.2).
applied to the mean magnitude. The skewness, in these variables, tends to have negative values,
meaning that the values with higher deviations are located in the left tail of the distribution. In
other words, higher variations happen when the star becomes brighter. The kurtosis tends to
have positive values, meaning that the distribution has a sharper peak around the mean than a
normal distribution. This can also be seen in the σ and σabs curves, which reach constant values
after some observations. These three are pulsating variables, hence, the opposite behavior of
these two statistical indices can be used to separate them from the other variables.
The α2-Canum Venaticorum variable star presents equal behavior for the skewness and
kurtosis curves applied to the mean magnitude, both increase with time. An increase in skewness
toward positive values implies that the deviations are higher when the star becomes fainter. This
variable presents rotational variability.
39
The peculiar aspect of the statistical index curves for the eclipsing binary can be used to
distinguish this class from the other variable stars in a large sample of light curves.
We still want to study in more detail the behavior presented by statistical indices of a
continuously growing database. However, we will try to apply some of them to our variability
surveys in order to test their reliability.
40
Chapter 4
The Orion Nebula Survey
4.1 Introduction
An overview of the Orion Nebula Cluster (ONC) was given in Chapter 2. We will discuss here
the variable stars in this star forming region.
Since the beginning of the study of young stars by Joy (1945) and Herbig (1960, 1962),
photometric variability served as one of the defining characteristics of pre-main sequence stars.
Variability was first detected and studied in the optical bands but later it became clear that
young stars are variable in all wavelengths. Rebull (2001) monitored ∼ 3600 young low mass
stars in the outer regions of the ONC using optical (UVIc) and near infrared filters (2MASS data)
and determined periods for 281 low mass stars. Carpenter, Hillenbrand & Skrutskie (2001) used
2MASS data to study near infrared photometric variability in the Orion A Molecular Cloud.
A large effort was done with the Chandra Orion Ultradeep Project (COUP) to understand the
emission of X-rays by young stars in the ONC (Getman et al., 2005). Flaccomio et al. (2005)
used the COUP data to study the modulation of X-ray emission in the ONC young stars and
they were able to find stars where the X-ray modulation is correlated with the rotation period.
Stassun et al. (2006) optically (BVRI) monitored the ONC while COUP was monitoring the
same region in X-rays and they report time-correlated optical and X-ray variations.
Our group has participated of a similar survey, using photographic films (Kodak Tech-Pan
4415 emulsion, sensitive from approximately 630 nm and 690 nm, roughly the R band) and
observations done at the ESO’ Schmidt 100/152 cm telescope (La Silla, Chile), scanned by the
SuperCOSMOS team1. The films covered a large area of 5◦ by 5◦ during a period of 2 years,
and some preliminary results were published by Lima (2002).
Our purpose is to observe the ONC and its surroundings, in near infrared wavelengths
(JHK), during a long time, in order to detect variable stars in timescales of a few days to a few
months. One of the most important byproducts of such a survey is the detection of eclipsing
binaries, whose stellar parameters are very important for putting constraints in the formation
and evolutionary stellar models.
1http://www-wfau.roe.ac.uk/sss/
41
yx
y
x
x
y
x
y
3 4
2
12.83’
13.65’
12.83’
13.65’
4.2
5’
8.7’
28
.0’
N
E
AG
26.48’
1
Figure 4.1: Left panel: Details of the WFCAM focal plane. The four detectors are shown in their respective
position relative to each other. Detector 1 shows the channel layouts. Right panel: The broad band filters used
in WFCAM. We used only the JHK filters in our survey.
We will try to attain our goals using observations made with the WFCAM, described in the
next section. Unfortunately we will not be able to use the final pipeline released data because
of problems detected in the subtraction of the sky background. We will instead use the summit
pipeline data, which will give our results the status of preliminary results.
4.2 WFCAM Survey
WFCAM stands for Wide Field Camera, which is the most capable IR imaging survey instrument
in the world, currently installed at the United Kingdom Infrared Telescope (UKIRT) at the
summit of the Mauna Kea Observatory. It is composed of four Rockwell Hawaii-II 2048×2048,
18 micron pixel array detectors. The pixel scale is 0.4′′ with a focal ratio of f/2.4. The detectors
are spaced apart from each other in a way that four exposures are combined to cover 0.8 degrees2.
The left panel in Fig. (4.1) shows the detectors’ arrangement. Each detector is divided into
quadrants and each quadrant is divided into eight channels of 128 × 1024 pixels, as exemplified
in detector #1, in the left panel of Fig. (4.1).
WFCAM has eight filter housings, five are used for the broadband set ZYJHK, two are used
for the narrow band H2 and Brγ and the last one is blanked for darks. In our survey we used
the broadband filters JHK. An illustration of the filter passbands is shown in the right panel of
Fig. (4.1).
4.2.1 Problems
There are many issues already known for the images acquired with the WFCAM. Some of these
issues are corrected during the pipeline reduction. We describe some of the most important
problems below. The description of these problems was taken from the instrument’s webpage2.
2http://casu.ast.cam.ac.uk/surveys-projects/wfcam
42
Figure 4.2: Left panel: Field distortion. Right panel: Differential non-linear distortion.
Field distortion can amount to 10′′ from the center to the edge of the field. Most of the
distortion can be accounted for by a cubic radial term, which is included in the WCS (World
Coordinate System) solution. Differential non-linear distortion can also be present (∼1%). Both
effects are presented in Figure (4.2).
Crosstalks are artificial structures created by saturated stars with either a doughnut appea-
rance from heavily saturated regions or half-moon-like from only weakly saturated stars. They
appear along the channels and have features at ∼1% of the differential flux of the source,
dropping to ∼0.2% and ∼0.05% further out. Figure (4.3) illustrates, in the left panel, where
crosstalk images are expected in the detector X-Y space relative to the position of saturated
objects for each quadrant. The crosstalk pattern rotates from quadrant to quadrant because the
readout amplifier for each quadrant is on a different edge of the detector (denoted by the red
lines). The right panel in Fig. (4.3) shows an example of crosstalk detected in our images.
Dark current and reset anomaly are 2D additive features that affect WFCAM. All IR
detectors suffer from reset anomaly to some degree and its main component is a large intensity
ramp at one edge of each quadrant (where the readout amplifier is located), reaching about 50
ADUs.
It has been shown that the reset anomaly ramp is stable throughout a night and even between
nights for a given set of exposure parameters. The ramp is remarkably similar for a wide range of
exposure times. This implies that master darks can be used to do the reset anomaly correction.
However, care must be taken to reject any dark frames that are exposed right after a filter
change to avoid any persistence effects.
Darks are routinely computed from daily observations, by combining as many darks taken
with the same readout parameters as the object frames of interest (that is exposure time, readout
mode and number of coadds). By combining many dark frames, artifacts such as cosmic rays
can be eliminated from the resulting master dark.
The average dark current is generally negligible and the dark frames are dominated by reset
anomaly variations, as shown in Figure (4.4).
Decurtaining is a pseudo-periodic ripple at the ±5 ADU level and can be seen at low level
in the darks. It is best illustrated by the difference between two dark frames of the same type,
43
Figure 4.3: Left panel: Diagram showing how the crosstalks appear in different quadrants. The crosstalk pattern
rotates because the readout amplifier for each quadrant is located on a different edge of the detector, as shown by
the red lines. Right panel: An example of crosstalk detected in one of our images. Whenever there is a saturated
star, a crosstalk pattern will be present.
Figure 4.4: This figure shows how the dark frames, for all detectors, are dominated by the reset anomaly.
as shown in Figure (4.5).
An algorithm is used to correct this problem and the results can be seen in the right panel
of Figure (4.5).
Persistence happens when a saturated image in a previous exposure leaves a persistent
image in following exposures. It has been observed in WFCAM data and is typical of infrared
44
Figure 4.5: Left panel: Difference between two dark frames taken in order to show the decurtaining effect. Right
panel: Dark frames with decurtaining solved.
Figure 4.6: This figure shows an example of persistent images in the WFCAM.
detectors. This phenomenon is not clearly understood and varies considerably between detectors.
Persistent images are very approximately top-hat shaped and cover the area that was saturated
in the contaminating frame. Persistence is not corrected by the pipeline reduction and an
example is shown in Figure (4.6).
4.2.2 Reduction
By the time we started the analysis of the M42 data we did not have the final reduction data.
We used the summit pipeline data, which was done using routines similar to the ones used in
the final pipeline. However, some of the problems described in the previous section were not
corrected during the summit reduction. We deal with them using our own routines in IDL.
The basic steps for the normal astronomical reduction were taken during the summit reduc-
tion and we describe them below.
Dithering is the procedure of splitting a long NIR exposure into several short exposures
at slightly different positions in order to minimize the contribution from bad pixels and cosmic
rays, and to avoid background saturation. Dithering shifts and combines these input frames into
a single output using the appropriate dither offsets, which are initially calculated from the WCS
45
Figure 4.7: This figure shows an example of the flatfield correction.
parameters in the FITS headers that have been generated from the astrometric calibration. The
offsets are then further refined by object position cross-correlation.
Flatfield in the WFCAM departs somewhat from the usual NIR processing strategies by
making extensive use of twilight flats, rather than dark-sky flats, which potentially can be
corrupted by thermal glow, fringing, large objects etc. The principle adopted is to attempt to
decouple sky estimation/correction from the science images. An example is shown in Figure
(4.7).
Ideally, sufficient (dawn) twilight flatfield frames to form master flats are taken once a week
using a series of 9-point jitter sequence observations in Z,J,K one night and Y,H the next. Note
that it is impossible to get good flats in all broadband filters in any one night. The suitable
individual 9-point jitter sequences (∼5000-20000 counts/pixel ie. ∼25000-100000 e−/pixel) are
dark-corrected and then scaled and robustly combined to form new master flats. If enough
are available, these may be further combined to reduce the random photon error in the master
flats even further. If available a pair of these flats at different count levels (∼factor 2) is then
used to automatically locate the bad pixels and create the generic confidence map, one for each
filter. Bad pixels are defined as those having properties significantly different from their local
neighborhood “average” in the individual master flats or in their ratio.
These flats give good dark sky correction (i.e. gradients are at the ∼1% level at most),
and show no fringing nor measurable thermal emission. The overall QE is good but the flats
show large spatial gradients across the detectors indicative of up to a factor of 2 sensitivity
variation. Since the general characteristics of these variations can be seen across all filters,
the simplest interpretation is that these variations in level reflect genuine sensitivity variations
across the detectors. This has been subsequently confirmed by examining the χ2 values of the
residuals from PSF-fitting. These sensitivity variations also cause problems for the generation
of the confidence maps, uniformity of surveys, and possibly also impacts on the calibration error
budget.
Photometric calibration is currently based on 2MASS, via color equations to convert to
the WFCAM instrumental system. 2MASS solutions for every catalogued frame are generated
and allow monitoring of effective ZPs (Zero Points) at the ∼few % level.
46
4.3 Observations
The observations began in October 2006 and ended in April 2007, giving us a total of 101 nights.
This is a short program that only uses some minutes per night, with 2 seconds per exposure.
Table (4.3) shows the observation dates for the ONC.
Table 4.1: Observation log for the Orion Nebula.
October - 2007 26 27 28 30 31
November - 2007 01 09 11 12 13 14 15 16 17 19 20
21 22 23 26 27 29 30
December - 2007 05 06 08 09 10 12 14 15 16 17 19
20 21 23 24 27 28 29 30 31
January - 2008 01 02 03 06 14 15 16 17 18 19 20
22 24 25 26 27 28 30
February - 2008 05 06 07 12 13 14 15 16 19 20 21
22 23 24 25
March - 2008 02 03 04 05 08 09 10 16 17 18 19
20 21 22 23 24 25 26 27 28 29
April - 2008 04 19 20 21
The observation method used by the WFCAM consists of performing four exposures with
a coordinate shift between each exposure, in a way that a mosaic is obtained in the end of the
process. We designate each area to be exposed as a letter: a, b, c, d. Each CCD in the camera
is also designated as a letter: w, x, y, z. Figure (4.8) shows the layout used to observe the ONC.
Each exposure took two seconds.













































































































































































































































































 
















































 




















 
















































 








































































































































































c
a a
a a
c
c
c
x
y
w
zy z
zzyy
x w
wxx w
a
c
wx
y z
b b
b b
d d
dd
d
b
Figure 4.8: WFCAM layout and reading scheme used to observe the ONC.
We had access to the summit pipeline reduced data during the observation period. The data
was kindly reduced by Dr. Andy Adamson after each night of observation. The summit pipeline,
47
however, does not have all the tools necessary to deal with deformations and field distortions.
We started working on the data at the end of 2006, while part of the observations was still
in progress. The catalogs have a simple structure in ASCII format, easy to be read by IDL
routines. Each CCD for each area has a separate catalog. An example of a catalog is given
below.
1 colours were created
K
Origin: IO::SExtractor
0 09 05 35 20.97 -05 31 21.69 1991.484 71.761 -14.8113 0.0010 0
0 10 05 35 21.03 -05 28 09.25 2949.217 69.721 -14.3697 0.0014 0
0 11 05 35 21.16 -05 25 56.98 3607.804 60.891 -14.8718 0.0009 0
0 13 05 35 21.30 -05 24 57.29 3905.070 51.429 -15.1335 0.0007 0
0 17 05 35 21.63 -05 26 57.76 3305.145 25.587 -13.0385 0.0049 0
0 19 05 35 21.56 -05 32 46.01 1572.265 27.137 -11.2798 0.0248 0
The first line indicates that there is only one color in the catalog, which is shown in the
second line. The third line indicates that the catalog was generated by the summit pipeline.
The parameters for each star start in the fourth line of the catalogs. The first column is an
empty flag and the second column gives an ID for each detection. The third, fourth and fifth
columns contain the right ascension (H:M:S.S), the sixth, seventh and eighth columns contain the
declination (D:M:S.S). The ninth column presents the number of counts and the tenth column
has the count error. The instrumental magnitude is presented in the eleventh column, with
the associated error in the twelfth column. The last column shows a flag that indicates if the
measurement is good (0) or bad (99).
An IDL routine was written to read the catalogs. We can see in the second column that
the running ID isn’t linear and in fact is not fixed for each star. It only denotes the order of
the detections. The same star has different IDs for different filters (JHK) and different nights,
meaning that we have to read each catalog and store the information about each star in a new
table.
DADJHK(3,14000,101,12)
Number of nights
Number of objects
Filters (J H K)
4X(JD, Mag,   Mag)∆
Figure 4.9: DADJHK - layout of the table containing all the photometric information.
Figure (4.9) presents the layout of the structure used in the new table that contains all the
photometric information, for each star. We created a matrix with dimensions shown in Fig.
(4.9), where the last dimension has twelve positions. We repeated the parameters; Julian data
(JD), magnitude (Mag) and error in the magnitude (∆Mag), four times because stars on the
borders of the CCDs can be detected more than once, since the CCDs overlap each other. Thus,
one star can be stored in different catalogs for the same night in the same filter.
The observations were performed using the filters in the KHJ order. We analyzed each object
48
individually in the first night because we wanted to be sure that only objects with stellar profiles
were being catalogued. Crosstalk, cosmic rays and persistence defects were rejected during this
procedure. We began facing problems when we started analyzing the second night. Due to the
fact that the field and differential distortions were not corrected by the pipeline, the same stars
had very different coordinates from one night to the other. It was necessary to create some sort
of correction for the coordinates.
4.3.1 Workarounds
In order to find the correct coordinates for stars in different nights, we wrote an IDL routine
that shifts the coordinates until the correct match (inside a given error limit) is achieved. We
searched inside an area of 0.′′6 radius trying to find matching objects. The separation from the
object being analyzed to the catalogued one is stored in a different matrix. All objects are
searched and the separations are stored. In a second step we perform a shift in the coordinates
using the values stored. The necessary shift is calculated using weights. We give more weight to
the objects with smaller separations, assuming that large separations are produced by defects
or previously undetected objects.
We realized that one shift was not enough to produce the desired effect and hence we started
to perform subsequent shifts. We were surprised to see that some nights required large shifts,
sometimes requiring initial shifts of 7′′. We ended up with a routine that uses six shifts, beginning
with large values (20′′) and performing the two last shifts with a search radius of 0.′′6. This
procedure corrected the field and differential distortions.
We were lucky to have few stars per CCD because a considerable computational time is
required to perform each shift. It took us approximately five months to have a preliminary
database with all the observed nights. Although the number of nights is 101, our database has
only 98 nights because 3 nights were not reduced by the summit pipeline.
4.3.2 Superposition problems and calibration solutions
When we began analyzing the stored data we faced a problem that we did not expect. Each
CCD has a different detection level for each area, and also for different nights. This means, for
example, that the CCD ‘w’ has a different detection level when used in area ‘a’ and area ‘b’.
Also the same CCD, ‘w’ for example, showed different detection levels in different nights even
when used in the same area. This problem would make the construction of the final database
much harder than we thought because stars located in the superposition areas had four different
magnitudes in the same night. We are not certain about the causes of these level variations.
Figure (4.10) shows plots of (Mean Magnitude - Magnitude) versus Magnitude, for all the
16 fields in the J band, that when put together comprehend the full mosaic of the fourteenth
night. We are using the instrumental magnitude and it is possible to see that the measurements
behave in the same manner. All the CCDs have the same level in all areas. We expect that, in
this kind of plot, most of the data points are situated about the zero value.
Figure (4.11) shows the same kind of plots but for the eighteenth night, where we can see
clearly that the CCDs present different levels for different areas. There is no measurement for
the CCD y in area d (first column, second row). We can see in the panels of Fig. (4.11) that
the data points have the behavior we expect, except that they are not situated about the zero
49
Figure 4.10: (Mean Magnitude - Magnitude) versus Magnitude for the filter J, fourteenth night, for all the
CCDs and areas. The panels are distributed in the same layout shown in Fig. (4.8).
value, but shifted by 2 and even 4 magnitudes.
Because we were dealing with data from the summit pipeline, we decided to analyze stars
located outside the superposition areas. In other words, we will only analyze stars that have
been detected once each night. This way we can perform the necessary level corrections for each
CCD individually.
The level correction for each CCD was done using constant shifts in the objects’ magnitude.
To obtain the shift values we first calculated the geometric deviation for each CCD, in each area,
for each filter, for each night. As explained in Chapter 3, the geometric deviation will help us
identify the most constant stars. We will use these stars to calculate the mean magnitude that
will be used to shift the data points toward the zero value.
After this step was taken one last procedure is necessary. Because we are dealing with data
50
Figure 4.11: Same as Fig. (4.10) but for the eighteenth night.
coming from the summit pipeline, and also due to the fact that Orion is a region with strong
background variation, we decided to do differential photometry for all the stars in the field. Such
procedure requires constant stars along the observations to be used as comparison stars. We
need a set of constant stars for each CCD in each area.
Using the geometric deviation, the absolute value of the maximum amplitude variation
(AMV) and the maximum deviation from the mean magnitude (MDM), we selected the six
most constant stars present in each CCD for each area. We considered geometric deviations
values lower than 0.01, AMV values lower than 0.30 magnitudes and MDM lower than 0.15
magnitudes. The limit value of the last two parameters changed for different CCDs in different
areas. The best values were achieved for the CCD w in area a, with maximum values for AMV
and MDM respectively 0.06 and 0.03 magnitudes. The worst values were found for the CCD y
in area d (0.30 and 0.15 magnitudes for AMV and MDM).
51
Once the constant stars were chosen the differential photometry could be obtained comparing
the magnitude of the star being analysed with the constant stars, as shown below.
∆M1 = M − C1
∆M2 = M − C2
∆M3 = M − C3
...
where ∆Mn is the differential magnitude relative to the n
th comparison star. However, we can
define the best comparison star, for each CCD in each area, as the primary comparison and
transform the other comparison stars to the primary.
∆C2 = C1 − C2
∆C3 = C1 − C3
...
where ∆Cn is the difference between the n
th comparison star and the primary, obtained using the
mean magnitudes (Cn). We use mean magnitudes because the comparison stars were considered
to be constant stars.
We can then transform the comparison stars into the primary using the following equations.
CP2 = C2 +∆C2
CP3 = C3 +∆C3
...
This procedure allows us to do n comparisons between the star in study and the primary.
∆M1 = M − C1
∆M2 = M − CP2 =M − C2 −∆C2
∆M3 = M − CP3 =M − C3 −∆C3
...
The mean value of the Mn differential magnitudes will be our differential magnitude. The
final equation to be applied to the data is:
〈∆M〉 =M − C1 +
1
n
∑
i=2,N
(Ci − Ci) (4.1)
where M is the instrumental magnitude of the star, C1 is the primary comparison in that CCD in
that area, n is the total number of comparison stars (six in our case) and Ci is the i
th comparison
star. This equation must be applied for each night of observation. M and Ci are the magnitudes
for the night being corrected.
After the level correction for each CCD and the procedure to obtain the differential pho-
tometry, we began to reorganize the photometric and astrometric data. We need to know the
position of a specific star inside the data matrix in order to retrieve the magnitudes for each
52
filter and each night. An IDL routine was written to reorganize the data and this step was easily
done.
With the data reorganized we were able to plot the light curves for all the stars, and also to
apply the statistical indices to the light curves.
4.4 Variable Candidates
We will show in this section some of the results achieved with the preliminary data for the Orion
Nebula. Although the results have a preliminary status, we want to emphasize that the kind of
variability shown here is real and in some cases corroborated by works of other authors.
We wrote IDL routines to calculate the statistical indices for the whole sample and after
that we analysed the results individually.
4.4.1 Eclipsing binary candidates
We will first present six eclipsing binary candidates. Table (4.2) presents the identification of
the candidates, their coordinates, trial periods and period errors. Due to the fact that we have
few points to define the minima of the light curves, we are not able to obtain precise periods. An
observational campaign is necessary in order to better constrain the period and also to obtain
the orbital solution for these binaries.
Table 4.2: Six candidates to be eclipsing binaries in M42
ID RA(J2000) DEC(J2000) Per(d) err(d)
JW 102 05:34:48.957 -05:28:16.70 10.6 0.1
V387 Ori 05:35:05.312 -05:34:28.43 8.59 0.01
JW323 05:35:08.531 -05:25:18.01 31 1
JW 363 05:35:11.080 -05:36:50.51 7.02 0.01
COUP 462 05:35:12.043 -05:28:08.32 7.28 0.01
OX Ori 05:35:54.622 -05:27:07.77 17.31 0.01
As described in Chapter 3, the Mean Magnitude curve of an eclipsing binary is very peculiar,
with descontinuities caused by sudden changes in brightness. We used these Mean Magnitude
curves to help us in the task of finding eclipsing binaries. There is, however, another kind of
variable star that also presents discontinuities in the Mean Magnitude curve, the flare stars. This
kind of variable presents sudden increases of brightness, which cause the same kind of peculiarity
in the curves we want to use. To help us distinguish between flare stars and eclipsing binaries
we used the maximum deviation from the mean, which is positive for flare stars and negative
for eclipsing binaries. We also used the continuous light curve as a visual guide when analyzing
individually the stars, since the light curve of these variables has an opposite behavior.
Figures (4.12) and (4.13) present the continuous light curve, the Mean Magnitude curve and
the phased light curve of the eclipsing binary candidates. Each plot presents the name of the
respective star. The phase is obtained using the following equation:
Φ = frac
(
τ − E0
P
)
, τ = t + δt (4.2)
53
where τ is the time of mid-observation with heliocentric correction, P is the period, E0 is the
epoch or instant of an adopted minimum and frac denotes the decimal part of that ratio.
The period of each star was obtained through the use of the phase dispersion method by
Lafler & Kinman (1965), which searchs for periodic signals. The test requires that the sum of the
squares of the magnitudes differences between observations of adjacent phase to be a minimum.
It uses the following equation:
θ =
Σi(mi −mi+1)
2
Σi(mi −M)2
(4.3)
where M is the mean magnitude and N is the number of observations. The denominator is
independet of the phase and thus only the numerator needs to be calculated for each trial
period. Hence, in principle, the period with the minimum value of θ is the nearest to the correct
one.
4.4.2 Periodic variables
We used the skewness and kurtosis of the mean magnitudes to construct skewness curves and
kurtosis curves, which were then used to identify stars that present pulsational or rotational
modulation. During the individual analysis of each star, we selected 23 stars that present clear
evidence of periodic modulation.
Table (4.3) presents the stars identified through this technique. The first column shows the
name of the stars, followed by their coordinates in the second and third columns. The period
determined in this work is presented in the fourth column, with the respective error in the
fifth column. Columns 6, 7 and 8 present the identification, period and error when a star is
present in the catalogue published by Stassun et al. (1999). Columns 9, 10 and 11 present the
identification, period and error listed in the catalogue published by Rebull (2001). We obtained
the periods of the stars using the same technique applied to the eclipsing binaries. The values
listed in Table (4.3) are mean values obtained using the periods for all filters (some stars do not
have JHK light curves, in such cases we used the filters available).
Except for V1289 Ori, V482 Ori and V1277 Ori, the periods obtained in our work and the
periods obtained by Stassun et al. (1999) and Rebull (2001) agree within the errors.
Figures (4.14) to (4.21) present the continuous light curve, the phased light curve, the
skewness curve and the kurtosis curve of the stars listed in Table (4.3) Each panel is identified
by the name of the star.
4.4.3 Interesting light curves
The catalogue is composed of approximatelly eight thousand stars and it is beyond the scope of
this work to present all the light curves. We present here examples of interesting light curves
found within our sample. A survey covering a large area of an important star forming region
like the Orion Nebula, with such a long timeline base, will certainly change our view about the
variability status of young stars.
Table (4.4) presents the names and coordinates of these interesting stars while Figure (4.22)
presents their light curves.
54
Figure 4.12: Light curves, Mean magnitude curves and phased light curves for three eclipsing binary candidates.
55
Figure 4.13: Light curves, Mean magnitude curves and phased light curves for three eclipsing binary candidates.
56
Table 4.3: Objects candidate to be variable, selected through the study of statistical indices
ID RA(J2000) DEC(J2000) Per err SMMV Per err R2001 Per err
MN Ori 05:35:15.774 -05:33:12.17 9.570 0.004
V1479 Ori 05:35:01.475 -05:28:20.45 5.467 0.004 1545 5.32 0.23
KO Ori 05:34:56.463 -05:31:35.81 5.740 0.003
JW 156 05:34:55.526 -05:36:05.82 7.87 0.01
V1313 Ori 05:34:43.394 -05:30:06.52 7.381 0.003 1099 7.04 0.40
Parenago 1327 05:34:00.308 -05:35:42.60 18.5 0.2
V385 Ori 05:33:33.865 -05:33:25.39 7.073 0.003
V386 Ori 05:33:44.904 -05:31:07.55 2.6276 0.0006
V1293 Ori 05:35:30.657 -05:27:16.75 6.336 0.004 3197 6.30 0.32
V1561 Ori 05:35:31.669 -05:30:04.10 5.549 0.006 3240 5.51 0.24
V1539 Ori 05:35:27.479 -05:35:19.41 4.688 0.008 3088 4.62 0.17
V947 Ori 05:35:49.466 -05:35:26.47 8.61 0.01 3710 9.27 0.69 2329 8.63 0.022
V1123 Ori 05:35:23.467 -05:10:51.78 7.64 0.01 2919 7.70 0.47 1989 7.64 0.009
TKK 891 05:35:29.062 -05:06:04.10 3.305 0.004
V1585 Ori 05:35:52.171 -05:39:24.89 6.226 0.005 3758 6.27 0.31 2356 6.24 0.012
V1558 Ori 05:35:31.232 -05:40:10.81 6.17 0.01 3220 6.15 0.30 2089 6.09 0.011
CHS2001 10147 05:35:26.118 -05:45:08.46 5.828 0.004
V1289 Ori 05:35:29.469 -05:16:33.40 10.65 0.03 3158 8.69 0.60
V1490 Ori 05:35:11.875 -05:45:37.76 4.041 0.001 1907 4.02 0.13 1791 4.04 0.004
V482 Ori 05:35:06.105 -05:45:30.97 8.326 0.004 1707 7.70 0.47 1710 4.18 0.008
COUP 172 05:35:02.006 -05:20:54.86 9.355 0.005
JW 131 05:34:52.618 -05:15:36.27 2.351 0.002
V1277 Ori 05:35:12.844 -05:15:23.80 9.04 0.02 1982 7.78 0.48
Table 4.4: Some objects presenting interesting light curves, identified with the procedures developed in the
present work
ID RA(J2000) DEC(J2000)
CHS2001 9298 05:35:19.562 -05:27:04.54
COUP 981 05:35:18.586 -05:26:24.96
V1500 Ori 05:35:14.895 -05:36:39.34
V982 Ori 05:35:06.429 -05:33:35.06
V381 Ori 05:37:02.354 -05:36:29.93
WY Ori 05:34:15.630 -05:32:23.90
R2001 759 05:33:45.406 -05:36:31.64
V1555 Ori 05:35:30.794 -05:30:36.40
4.5 Summary of Results
Surveys covering large areas and long timelines are very important for discovering new variable
stars. We conducted such a survey in the Orion Nebula and the completion of our catalog
of variable stars is near. The creation of this catalog presented many challenges, which were
studied and conquered step-by-step.
The use of statistical indices proved to be useful when trying to separate the stars of a
sample using their light curves. We were able to find variable stars with periodic modulation
(rotational and pulsational) and also eclipsing binary candidates. The discovery of eclipsing
binary candidates was, from the beginning, one of our main objectives. The next step is to plan
follow up observations, both photometric and spectroscopic, in order to obtain precise stellar
parameters, which are so important in our attempt to better comprehend how the stars are
formed and how they evolve.
57
Figure 4.14: We present here the light curves, phased light curves, skewness and kurtosis curves (of the mean
magnitudes) for the stars that presented rotation or pulsation.
58
Figure 4.15: Same as in Fig. (4.14).
59
Figure 4.16: Same as in Fig. (4.14).
60
Figure 4.17: Same as in Fig. (4.14).
61
Figure 4.18: Same as in Fig. (4.14).
62
Figure 4.19: Same as in Fig. (4.14).
63
Figure 4.20: Same as in Fig. (4.14).
64
Figure 4.21: Same as in Fig. (4.14).
65
Figure 4.22: We show here examples of interesting light curves present in our sample.
66
Chapter 5
The Cygnus OB2 Survey
5.1 Introduction
One of the richest known star forming regions in the Galaxy is located in the direction of the
Cygnus constellation and is known as Cygnus X. It is also one of the brightest areas of the sky
in all wavelengths. When we look in that direction we are looking through the Local Spiral Arm
and thus we are deluded by different star forming regions projected in the same area of the sky.
Some of these star forming regions are situated at several hundred parsecs away while others
are located in the other spiral arm, the Perseus Arm, a few thousand parsecs away. One of the
most important regions in this area is the young OB association called Cygnus OB2.
Figure (5.1) shows an Hα image, taken from Schneider et al. (2006), that shows part of the
Cygnus X region. The North America Nebula and the Pelican Nebula can be seen in the bottom
right corner. The dark feature in the center of the image is the Great Cygnus Rift. The Cygnus
OB2 association is located almost in the center of the rift. An extinction map toward this region
can be seen in Figure (5.2), also taken from Schneider et al. (2006). In the center of the image
we can see a circle delimiting the extent of Cygnus OB2, and also of other OB associations.
The white circles with numbers on them are the radio continuum sources reported by Downes
& Rinehart (1966). It is possible to see in Fig. (5.2) that the Cygnus OB2 association is located
in a region of low extinction compared to its surroundings, although the values for AV range
from 5 to 20 magnitudes.
Cygnus OB2 was discovered by Mu¨nch & Morgan (1953), who called attention to this
“possible association of blue giant stars”. Many studies followed this pioneer work, Johnson
& Morgan (1954) identified 14 association members as B1 stars or earlier, and Morgan, Meinel
& Johnson (1954) and Schulte (1956a) identified 80 possible members which were later confirmed
by Schulte (1956b, 1958) through UBV photometry. All studies pointed to the high extinction
towards Cygnus OB2 (∼5−20 magnitudes). A more comprehensive study was made by Reddish,
Lawrence & Pratt (1966) who identified hundreds of stars, possible members, through the use of
color-color diagrams. Torres-Dodgen, Tapia & Carroll (1991) and Massey & Thompson (1991)
analyzed in detail the massive population and determined a distance of 1.7 kpc to the Cygnus
67
Figure 5.1: The Great Cygnus Rift in Hα, image taken from Schneider et al. (2006).
OB2 association. An age of 1 to 3 Myr was suggested based on the presence of O3 stars in the
main sequence and evolved supergiants (Torres-Dodgen, Tapia & Carroll, 1991). Other studies
suggest younger ages (< 1 Myr) (e.g. Voelcker, 1974; Hutchings, 1981).
Cygnus X has been the target of many radio surveys: Westerhout (1958) at 1390 MHz,
Downes & Rinehart (1966) at 5 GHz, Wendker (1984) and Wendker, Higgs & Landecker (1991)
at 408, 1420 and 4 800 MHz, Taylor et al. (1996) at 327 MHz, Taylor et al. (2003) at 1420 MHz.
Some of the most massive stars in Cygnus OB2 were detected as non-thermal radio emitters.
Through the use of 2MASS infrared observations, Kno¨dlseder (2000) suggested that Cygnus
OB2 should be considered a young globular cluster, based on the number of OB stars (∼2600
±400 stars, among which ∼120±20 are O stars), its spherical geometry and its youth (1−3 Myr).
Kno¨dlseder (2000) also suggested that the geometric shape of the association was misinterpreted
by Reddish, Lawrence & Pratt (1966) because of the high extinction in the optical band.
Kno¨dlseder (2000) argues that CygOB2 is too compact to be an OB association and too massive
(4−10×104M⊙) to be an open cluster. Its spherical morphology and high central density indicate
that CygOB2 is a young globular cluster. Later, Kno¨dlseder et al. (2002) reclassified Cygnus
OB2 as a supercluster, similar to the Arches and Quintuplet clusters (Figer et al., 1999), NGC
3603 (Moffat, Drissen & Shara, 1994) and W94A (Conti & Blum, 2002). Hanson (2003) made a
study of the more massive Cygnus OB2 members and stated that many OB stars do not belong
to the association. For these authors the O population is smaller than the one proposed by
Kno¨dlseder (2000) and thus Cygnus OB2 would not be a supercluster.
We summarize the properties of this unusual association, following Kno¨dlseder (2000), in
Table (5.1). Figure (5.3) shows the stellar density distribution derived from the 2MASS point
source catalog (left panel) and from the DSS red plates (right panel) as shown by Kno¨dlseder
(2000). Darker regions correspond to higher densities. It is clear in Figure (5.3) why Reddish,
68
Figure 5.2: The Great Cygnus Rift, extinction map taken from Schneider et al. (2006). The white circles
represent the radio continuum sources detected by Downes & Rinehart (1966).
Figure 5.3: Cygnus OB2 star density map. The left panel shows the 2MASS data while the right panel shows
the DSS red plates data. The image was taken from Kno¨dlseder (2000).
Lawrence & Pratt (1966) misinterpreted the shape of the Cygnus OB2 region, extinction hides
most of the association in the optical bands.
Drew et al. (2008) identified the on-sky distribution of over 1000 A0-A5 stars towards,
around, within and beyond Cygnus OB2. They have found a peak in A-star density (∼ 200
stars) roughly centred on RA 20:32:00 DEC +41:00:00, ∼ 20 arcmin south of the Cygnus OB2
Trapezium. The Trapezium is composed of four stars named Schulte 8A, Schulte 8B, Schulte 8C
and Schulte 8D, there is also a fifth star, named [MT91] 893, which was not catalogued by Mu¨nch
69
Table 5.1: Cygnus OB2 properties, following Kno¨dlseder (2000)
center RA 20:33:10, DEC +41:12:00
diameter ∼ 2◦ (∼ 60 pc at a distance of 1.7 kpc)
half light radius 13′(6.4pc)
core radius 29′ ± 5′(14± 2pc)
tidal radius 93′ ± 20′(46± 10pc)
members earlier F3V 8600± 1300
OB star members 2600± 400
O star members 120± 20
total stellar mass (4− 10)× 104M⊙
central mass density 40− 150M⊙pc
−3
IMF slope −1.6± 0.1
& Morgan (1953); Morgan, Meinel & Johnson (1954); Schulte (1956a,b), but was catalogued by
Massey & Thompson (1991). Drew et al. (2008) also identified the on-sky distribution of the
reddenings for the same sample. Treating the A stars and OB stars as part of the same cluster,
Drew et al. (2008) found a distance of 1.4 kpc, if the A stars are 7 Myr old, or 1.7 kpc if they are
5 Myr old. Either way, these authors find a difference between the age of the A stars and OB
stars, the later ones supposedly 1−3 Myr old. Based on the findings of Drew et al. (2008) we
can think about the possibility of two bursts of star formation in Cygnus OB2 or a substantial
age spread in the association.
Kno¨dlseder et al. (2002) suggest that the gamma-ray and free-free emission in the Cygnus
region seems to be dominated by the Cygnus OB2 association. It would be responsible for 75%
of the ionization in Cygnus but would account for only 20% of the observed 1.809 Mev photons.
Kno¨dlseder et al. (2002) also rose some doubt about the suggested scenario used to explain the
anomalous stellar proper motions in the Cygnus area (Comero´n & Torra, 1994). In the suggested
scenario, the winds and supernovae of the Cygnus OB2 association would be responsible for the
formation of the surrounding associations: Cygnus OB1, OB3, OB7 and OB9. However, the
age estimated by Kno¨dlseder et al. (2002) is inferior to that of the surrounding associations.
Kno¨dlseder et al. (2002) suggested that the runaway stars could be a possible explanation for
the anomalous stellar proper motions. However, Kiminki et al. (2007) did not find any runaway
star in Cygnus OB2 in their radial velocity survey.
The optical counterparts for the first X-ray sources, found by the Einstein satellite, were
massive stars in the Cygnus OB2 association (Schulte 5, Schulte 8A, Schulte 9 and Schulte 12).
The same field was also observed by ROSAT, ASCA and Chandra (De Becker et al., 2006).
More recently Albacete Colombo et al. (2007) used data from the Chandra satellite to unveil
the low mass population. They found that Cygnus OB2 has a very rich population of low-mass
X-ray emitting stars, but the circumstellar disks are scarce. The X-ray activity is comparable
to the activity in the ONC.
Wendker, Higgs & Landecker (1991) did not detect any evidence of supernova remnants in
the area. This evidence implies that the association is still too young, maybe less than 1 Myr.
High mass evolved stars are present in the association and could be the first of this area to
become supernovas.
There is a lack of an HII region in the center of Cygnus OB2 and an explanation for this
fact is still missing. Wendker, Higgs, & Landecker (1996) have found what appears to be two
cometary HII regions close to the center of Cygnus OB2.
70
Table 5.2: Observation dates of the Cygnus OB2 survey.
April 01 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17
18 19 20 21 22 23 25 26 27 28 29 30
May 01 02 03 04 05 06 07 08 09 10 11 14 18 19 20 21
August 04 05 06 07 08 09 10 11 12 18 19 20 21 22 23 24
26 27 28 30 31
September 12 13 14 15 16 17 18 19 20 21 22 23 24 25 29 30
October 01 02 03 04 06 07 08 09 10 11 12 13 14 15 16 17
19 20 21 22 24 25 26 27
November 01 02 03
Schulte 8A was discovered as a binary by De Becker et al. (2006) and consists of an O6 If
star plus an O5.5 III(f) star. It is an eccentric system with e=0.24±0.04 and a period of
21.908±0.040 d. The semi-major axis is 142±6R⊙. The fact that Schulte 8A is a binary could
reconcile the high bolometric luminosity reported by Herrero, Puls & Najarro (2002) with its
spectral type (O5.5 I(f)) found at that time, when thought to be a single star.
Naze´ et al. (2008) reported the detection of a companion, in an eccentric orbit, to the star
Schulte 9, using radial velocity measurements. The period of 2.355 years is consistent with the
radio measurements reported by Van Loo et al. (2008), which used VLA data since 1980.
Kiminki et al. (2007) made a radial velocity survey, with a 6 year interval, in search for
massive contact binaries. A total of 146 OB stars were observed and 73 were identified as new
early-type stars. Using these stars they obtained a distance modulus of 11.3 and an IMF slope
∼-2.2±0.1. Few differences between the spectral classification of previously known members
were found, except for the stars Schulte 12 and Schulte 21. Schulte 12 is classified as a B3
Iae with a temperature class variation of B3 – B8 in an interval of one year. Schulte 21 was
previously classified as a B1 V and the spectral classification changed to B0 Ib. Kiminki et al.
(2007) obtained a massive binary frequency of 30% – 42% and detected no obvious OB runaways.
Most of the observations of Cygnus OB2 concern the high mass population and little attention
has been given to the low and intermediate mass population. High resolution deep surveys must
be done in order to better determine the total population of this incredible star forming region.
We have conducted such survey in the infrared band using the WFCAM. Our survey does not
intend to be too deep but has enough resolution to resolve many objects and help us to better
determine the real population of Cygnus OB2. The long duration of the survey will allow us to
unveil the variable population of this massive association.
5.2 Observations
The observations were performed with the WFCAM (see description in Chapter 4) from April
until November 2007. The procedure is identical to the one used in the Orion Nebula. This
project also uses only some minutes each night, with expositions lasting only 2 seconds. Table
(5.2) presents the dates of observation.
Access to the final pipeline reduction data was granted to us in January 2008. We retrieved
the images and catalogs from the CASU server and began analyzing its composition. It became
clear to us that the catalogs needed to be reorganized in order to build light curves for the stars.
The IDL routines developed to deal with the Orion Nebula data were adapted to deal with the
71
Cygnus OB2 data. The first problem was the difference in the catalog structure. Because the
Orion data was reduced using the summit pipeline, the catalogs have a different structure from
those created by the final pipeline. The Cygnus OB2 catalogs are stored in multi-extension
FITS files as FITS binary tables which were read using the IDLASTRO routine called ftab.pro.
Each FITS file has 4 extensions, one for each CCD. Each fits table consists of 80 columns with
the relevant parameters (described below) for each object.
N Name Description
1 Sequence number running number for ease of reference, in strict order of
image detections
2 Isophotal flux standard definition of summed flux within detection
isophote, apart from detection filter is used to define
pixel connectivity and hence which pixels to include.
This helps to reduce edge effects for all isophotally
derived parameters
3 X coordinate intensity-weighted isophotal centre-of-gravity in X
4 X coordinate err estimate of centroid error
5 Y coordinate intensity-weighted isophotal centre-of-gravity in Y
6 Y coordinate err estimate of centroid error
7 Gaussian sigma these are derived from the three general intensity-
weighted second moments
8 Ellipticity the equivalence between them and a generalised
elliptical Gaussian
9 Position angle distribution is used to derive Gaussian
sigma = $(\sigma_a^2 + \sigma_b^2)^{1/2}$
ellipcity = $1.0 - \sigma_a/\sigma_b$
position angle = angle of ellipse major axis wrt x axis
10 Areal 1 profile number of pixels above a series of threshold levels
relative to local sky.
11 Areal 2 profile levels are set as T, 2T, 4T, 8T ... 128T where T is the
threshold.
12 Areal 3 profile
13 Areal 4 profile
14 Areal 5 profile
15 Areal 6 profile
16 Areal 7 profile
17 Areal 8 profile
18 Peak height in counts relative to local value of sky - also zeroth
order aperture flux
19 Peak height err estimate error of peak height
20 Aper flux 1 Fitted flux within 1/2* core radius
21 Aper flux 1 err Error in fitted flux within 1/2* core radius
22 Aper flux 2 Fitted flux within 1/sqrt(2)* core radius
23 Aper flux 2 err Error in fitted flux within 1/sqrt(2)* core rad
24 Aper flux 3 Fitted flux within 1* core radiu
25 Aper flux 3 err Error in fitted flux within 1* core radiu
26 Aper flux 4 Fitted flux within sqrt(2)* core radius
27 Aper flux 4 err Error in fitted flux within sqrt(2)* core rad
28 Aper flux 5 Fitted flux within 2* core radius
29 Aper flux 5 err Error in fitted flux within 2* core radius
30 Aper flux 6 Fitted flux within 2*sqrt(2)* core radius
31 Aper flux 6 err Error in fitted flux within 2*sqrt(2)* core rad
32 Aper flux 7 Fitted flux within 4* core radius
33 Aper flux 7 err Error in fitted flux within 4* core radius
34 Aper flux 8 Fitted flux within 5* core radius
35 Aper flux 8 err Error in fitted flux within 5* core radius
36 Aper flux 9 Fitted flux within 6* core radius
37 Aper flux 9 err Error in fitted flux within 6* core radius
38 Aper flux 10 Fitted flux within 7* core radius
39 Aper flux 10 err Error in fitted flux within 7* core radius
40 Aper flux 11 Fitted flux within 8* core radius
41 Aper flux 11 err Error in fitted flux within 8* core radius
42 Aper flux 12 Fitted flux within 10* core radius
43 Aper flux 12 err Error in fitted flux within 10* core radius
44 Aper flux 13 Fitted flux within 12* core radius
45 Aper flux 13 err Error in fitted flux within 12* core radius
46 Petr radius
47 Kron radius
48 Hall radius
49 Petr flux
50 Petr flux err
51 Kron flux
52 Kron flux err
53 Hall flux
54 Hall flux err
55 Error bit flag bit pattern listing various processing error flags
56 Sky level local interpolated sky level from background tracker
57 Sky rms local estimate of variation in sky around image
58 Parent or child flag for parent or part of deblended deconstruct
59 RA Right ascension
60 DEC Declination
61 Classification flag indicating most probable morphological
classification: -1 stellar, +1 non-stellar, 0 noise,
-2 borderline stellar, -9 saturated
62 Statistic An equivalent N(0,1) measure of how stellar-like an
image is, used in deriving 61 in a necessary but not
72
DADJHK(3 , 4 , 4 , 20000 , 112 , 3 )
JD,  Mag,    Mag
Number of nights
Number of objects
CCD (w x y z)
Area ( a b c d)
Filters(J H K)
∆
COORDS(3 , 4 , 4 , 4 , 20000)
N M)(α δ
Figure 5.4: DADJHK and COORDS, layout of the tables containing all the astrometric and photometric
information.
sufficient sense.
From column 63 to 80 our fields are blank.
Right at the beginning of our work with the tables it became clear to us that we should deal
with the catalogs for the Cygnus OB2 region in a different way than the one used for Orion.
The main reason was the amount of data present in each CCD exposure. While we had ∼
500 detections per CCD in the Orion survey, for Cygnus OB2 the average number is ∼ 10 000.
Such a huge amount of data required an enormous amount of time, both computational and
personal, to be organized. Usually, one observation requires approximatelly 40 hours of work to
be processed.
The tables were organized in two different catalogs, one called coords which contains the
coordinates and another called dadjhk which contains the photometry. Figure (5.4) shows the
internal structure of the catalogs.
We decided to separate the tables into two catalogs because the process of detection in the
pipeline does not create a unique ID for each object. The pipeline procedure reads the mosaic
starting in filter J, Area ‘a’, and CCD ‘w’, the procedure ends in Area ‘d’, CCD ‘z’, after that
the next filter is H and the last is K. The first object detected in the first filter, first area, first
CCD, does not coincide with the first object detected in the second filter, first area, first CCD.
Actually, the first object detected in each of the filters, areas and CCDs does not coincide with
the first object of the same filters, areas and CCDs of the next night. To find each object detected
in one night in the subsequent night it is necessary to search through the whole respective table
of that filter, that area, that CCD. Figure (5.5) shows the layout of the WFCAM mosaic and
the reading steps through the CCDs and Areas.
We used the Classification Flag (ID 61 in the FITS header) to help us with the classification
of the objects. We noticed that they are not perfect, as one would expect. Some objects with
flag ‘0’ (noise) were actually stellar objects while some objects with flag ‘−1’ (stellar) were noise
or defects. At the beginning of the procedure we checked every point source and as we learned
the behavior of the objects we decided to restrain some parameters and make the procedure a
bit more automatic. Given a magnitude interval and a given error, if the object presents the
‘−1’ flag it was automatically accepted. Objects inside this interval but with different flags were
73



























































































































































































































































































































































































































































































































































c
d
a a
a a
b
b
bc
d
c
d
b
d
c
x
y
w
zy z
zzyy
x w
wxx w
a
b c
d
wx
y z
Figure 5.5: WFCAM layout and reading scheme
checked everytime.
The IDL procedure evolved during our work with the tables. Because of the huge amount of
data per CCD it was necessary to develop control loops for each CCD, each area, each filter and
each night. We ended with a procedure that allows us to stop and save the data almost anytime
we want. These cautions were necessary because of energy blackouts that would sometimes
occur in the middle of an unsaved 10 000 star loop.
The process became faster when we started the second night, because objects that matched
an already catalogued source were accepted automatically. New objects went through the same
process as in the first night.
The third night was the best night concerning seeing and sky transparency. We achieved
very good resolution and many new objects were catalogued, which slowed down the cataloging
process. Five days were necessary to finish the third night.
The fourth night catalogation was slower than the third night because of the increase in
the number of objects catalogued. Also, since the third night has better resolution it resolved
objects that were previously seen as single stars. Every time the procedure finds catalogued
object inside a circle of 1′′ from the object being analyzed, it asks which entry is the correct
one. Five days were also necessary to finish the fourth night. Our estimate is that this process
will take an average of five days of work per observed night.
One step used in the Orion survey was not necessary for Cygnus OB2 because of the
efficiency of the final pipeline. Because the Orion data was reduced using the summit pipeline,
it was necessary to shift the coordinates for each CCD in order to correct the distortions,
explained in Chapter 4. The final pipeline already corrected these distortions and no shifts
were necessary. This was a major evolution in the procedure because the time required to shift
∼10 000 coordinates was huge. We tested the shifts in the beginning of the second night and
they took almost half an hour on the fastest computer we have available. Assuming 30 minutes
for each shift and 48 shifts per night, we would spend 24 hours only calculating the shifts.
Since the process was slow, even without the shifts, we decided to postpone the variability
analysis to a near future. We will concentrate here on the analysis of the Cygnus OB2 region
which is poorly studied so far, compared to a region like the ONC. We have found evidences
74
Figure 5.6: The infrared sources revealed in our survey are displayed in this figure as filled circles. The sources
are labeled using IDs found in the literature. The arrows in the bottom of the image represent the cometary
globules we have found in our survey, the size of the arrow bodies correspond to the size measured from the head
to the tail of the globule.
that corroborate the youth status of the association. Our goal in this chapter is to show that
there are many regions in this area where star formation is still happening. Also, Cygnus OB2
harbors some of the most massive stars in the Galaxy. If we want to understand the upper part
of the initial mass function, we need to study in detail the stars in this region.
We would like to point out, however, that the results presented in this chapter are preliminary,
though they give us a glimpse of the importance of this survey. Figure (5.6) shows the distribu-
tion of the infrared sources and small scale cometary globules we have found during our study.
5.3 Cometary Globules
We have found six small scale cometary globules in our K-band images. The existence of such
objects is definitive evidence that star formation happened not so long ago in Cygnus OB2. The
cometary globules’ tails point radially away from the center of the OB association, as shown
in Figure (5.6). We named them using the coordinates of their heads in sexagesimal format:
CGhhmmss+ddmmss. Table (5.3) presents their names, the angle between their head and tail,
measured in the same way as the position angle (vide position angle in Chapter 2) and their
75
Figure 5.7: Cometary globules CG203446+411446 (panel a), CG203453+405320 and CG203442+405115 (left
and right on panel b) and CG203318+405911 (panel c).
extent in arcseconds and parsecs. To calculate the extent in parsecs we used a distance of 1.7
kpc, meaning that we associate the cometary globules with the Cygnus OB2 association.
Table 5.3: Cometary globule ID, coordinates, position angle and assumed size.
ID α(J2000) δ(J2000) PA Extent(′′) Extent(pc)
CG203318+405911 20:33:18.7 +40:59:11.8 177◦.6 46.14 0.38
CG203410+410700 20:34:10 +41:07:00 110◦.6 17.82 0.15
CG203413+410813 20:34:13.3 +41:08:13.9 95◦.8 44.83 0.37
CG203446+411446 20:34:46 +41:14:46.6 96◦.7 55.60 0.46
CG203442+405115 20:34:42.6 +40:51:15.9 124◦.4 44.20 0.36
CG203453+405320 20:34:53.3 +40:53:20.3 115◦.7 23.78 0.20
We show here the K-band images with a false color scale which helps us to see the cometary
globules. Figure (5.7) presents four cometary globules. The cometary globule shown in panel
(a) (CG203446+411446) was found close to the beautiful nebulosity associated to the DR18
radio source, which is presented in Figure (5.14). We will discuss this nebulosity in a specific
subsection.
Panel (b) presents two globules whose tails point in the same direction (CG203453+405320,
left; CG203442+405115, right). Both globules have a very reddened star in their heads, which
are probably associated with them.
76
Panel (c) presents a globule (CG203318+405911) with a very bright star in front of it. They
are likely not associated since we cannot detect any kind of ionization or reflection in the globule.
The bright star is probably associated with Cygnus OB2, but its alignment with the globule is
just a projection in the plane of the sky.
Figure (5.8) shows two globules, whose tails also point in the same direction, although the tail
of the bottom one (CG203410+410700) is less pronounced. The top one (CG203413+410813)
seems to be associated with some very reddened stars in its head and it is possible to see some
curvature in its tail.
Figure 5.8: Cometary globules CG203410+410700 (bottom) and CG203413+410813 (top). Both globules are
associated with red stars in their heads.
5.4 The Cygnus OB2 Center
We show in Figure (5.9) a composite color picture of the center of the Cygnus OB2 association,
where it is possible to see the large number of blue stars that characterize the region.
Bica, Bonatto & Dutra (2003) suggested that the center of the Cygnus OB2 association
harbors two new open clusters, one centred in coordinates RA 20:33:10 DEC +41:13:07 and the
other at RA 20:33:15 DEC +41:18:45. In Fig. (5.9) the first cluster corresponds to the dense
concentration of blue stars in the bottom-center part and the second cluster corresponds to the
less dense concentration of blue stars in the top-center. It is possible to see a line of very bright
saturated stars in the center of the picture. These stars are members of the association, but
they were not considered by Bica, Bonatto & Dutra (2003).
Bica, Bonatto & Dutra (2003) used DSS images and 2MASS photometry to characterize
these clusters. They also used the available spectroscopy and optical photometry of the stars
in the area to help constrain the reddening and distance to the clusters. They concluded that
both objects have the same distance and age.
77
It is possible to detect these concentrations of stars, considered by Bica, Bonatto & Dutra
as two open clusters, in our infrared images. We will use our photometry to study the stars in
these two regions and try to determine their association with Cygnus OB2. Our survey seems
to corroborate the work done by Bica, Bonatto & Dutra.
Figure 5.9: Center region of the Cygnus OB2 association. The first cluster proposed by Bica, Bonatto & Dutra
(2003) corresponds to the concentration of blue stars in the bottom-center of the figure. The second cluster
corresponds to the top-center concentration. The very bright blue stars in the center, which are members of the
association, were not taken into account by the authors.
5.5 Infrared Sources
The K band images allowed us to observe infrared (IR) sources within the Cygnus OB2 region.
As said previously, because of the projection of this region in the direction of the Local Arm,
some of these sources may not be associated with Cygnus OB2.
Figure (5.6) presents the distribution of these IR sources in the surveyed area. We determined
the center of these regions based on the extent of the bright nebulosities associated to them. We
will describe these IR sources in the next subsections, where we use their IRAS identification,
although we also list other IDs found in the literature.
78
5.5.1 IRAS 20321+4112
IRAS 20321+4112 is the IR source (radio source at 4 800MHz, Wendker, Higgs & Landecker,
1991) is the IR source closest to the center of the Cygnus OB2 association. In Figure (5.10) we
can see that it is divided into two structures; to the right we have a spherical glow with dark
stripes on it, while to the left we can see a myriad of faint reddened stars. Other identifications
in the literature are: [TGC96] 2032+4112 (radio source, Taylor et al., 1996), ECX6-27 (radio
source at 4 800MHz, Wendker, Higgs & Landecker, 1991), 18P 61 (radio source at 408MHz,
Wendker, Higgs & Landecker, 1991), 19P 31 (radio source at 1 420MHz, Wendker, Higgs &
Landecker, 1991), WSRTGP 2032+4112 (radio source at 327 MHz, Taylor et al., 1996), 40P
37 (radio source at 408MHz, Higgs et al., 1991), MOL 128 (radio source at NH3(1,1) and (2,2)
lines, Molinaris et al., 1996), [L89b] 80.346+00.721 (HII region, Lockman, 1989) and [KC97c]
G080.3+00.7 (radio source at 4.85GHz, Kuchar & Clark, 1997).
Palla et al. (1991) report the detection of this IR source in the HCO+ maser but not in
the H2O maser survey. Odenwald (1989) classified this IR source as type “O”, from “obscured
field with no stellar object apparent”. Parthasarathy, Jain & Bhatt (1992) suggested that the
absence of ultracompact HII regions in the Cygnus OB2 region could imply that IR sources
with far infrared luminosities of the order of 104 L⊙ are younger than ultracompact HII regions.
IRAS 20321+4112 has a far infrared luminosity ≃ 104 L⊙ and its flux distribution matches that
of massive young stellar objects.
Figure 5.10: IRAS 20321+4112
Comero´n & Torra (2001) determined a distance of 3.3 kpc for this region and suggest it is
located behind Cygnus OB2. They also suggest that their distance is underestimated given the
number of foreground stars in the studied sample. The radial velocities found by Lockman (1989)
and Bronfman, Nyman & May (1996), respectively -62.4 km s−1 and -66.4 km s−1, exclude the
possibility of a physical connection between this IR source and Cygnus OB2. Following Comero´n
79
& Torra (2001) and Bica, Bonatto & Dutra (2003) IRAS 20321+4112 lies at ∼ 10 kpc, at the
Outer Arm. Scalise, Rodriguez & Mendonza-Torres (1989) detected a maser counterpart for
this IR source and determined a kinematic distance of 6.5 kpc.
When finished, our database will be used to characterize the stars in this region. The
presence of variable stars like Cepheids, RR Lyrae and Luminous Blue Variables, can help us to
constrain the distance to this region. We defined its center as the coordinates RA 20:33:57 and
DEC +41:22:33.
5.5.2 IRAS 20327+4120
Figure (5.11) presents an IR source with a very bright and reddened star associated with it.
Patches of dark clouds can also be seen in this region. IDs in the literature include: WSRTGP
2032+4120 (radio source at 327 MHz, Taylor et al., 1996), 19P 35 (radio source at 1 420 MHz,
Wendker, Higgs & Landecker, 1991) and [TGC96] 2032+4120 (radio source, Taylor et al., 1996).
The detection of this region dates back to 1989 by Odenwald (1989), where it can be seen
that this IR source received an Optical Type “O”, which stands for “obscured field with no
stellar objects apparent”. The failure to detect any stellar object is understandable since the
observations were done in the optical domain.
Figure 5.11: IRAS 20327+4120
Palla et al. (1991) did not detect this source in their H2O maser survey. Parthasarathy, Jain
& Bhatt (1992) suggested that this IR source could have a mass of 2×104 M⊙ and a distance of 2
kpc, based on the CO survey conducted by Congo M., (PhD thesis, 1980, Columbia University).
Taylor et al. (1996) reported a ratio between the far infrared flux and the radio flux at 327 MHz
(R = S60µm/S327MHz) of R = 7 362, which means that this source is a radio thermal emitter.
80
Also, the value log(S60/S25) ≃ 1.13 is consistent with values of compact HII regions. Setia
Gunawan et al. (2003) detected a source (SBHW125, table 1 of the referred work) in their radio
survey at 1 400MHz, 30′′ away from the center of our IR source, with a flux density peak of
3.6 mJy, but did not detect it at 350MHz. In Fig. (5.11) the SBHW125 source coordinates
correspond to the bright blue star at the top. We defined the coordinates of the center of the
cluster as RA 20:34:31 and DEC +41:30:35.
The detection of IRAS 20327+4120 in radio surveys, its far infrared color and flux ratios
imply that it is a site where star formation is still happening. Deeper observations would be
necessary to detect, for example, outflows coming from the young stellar objects embedded in the
nebula. Our photometry allied to spectral observations would help us to determine the distance
and reddening for this region. The nature of IRAS 20327+4120 is still to be determined.
5.5.3 IRAS 20332+4124
Figure (5.12) presents the IR source called IRAS 20332+4124, whose coordinates correspond to
the region at the right with the brightest star of the region and dark stripes. IDs associated to
this object include: 19P 38 (radio source at 1 420 MHz, Wendker, Higgs & Landecker, 1991),
WFS 97 (submillimetric source, Williams, Fuller & Sridharan, 2004), [HLB98] Onsala 140 (maser
source, Harju et al., 1998), WSRTGP 2033+4124 (radio source, Taylor et al., 1996).
Figure 5.12: IRAS 20332+4124
IRAS 20332+4124 is similar to IRAS 20321+4112, both present two distinct regions, one
with dark stripes and the other with faint red stars. We assumed the center of IRAS 20332+4124
to be RA 20:35:01 and DEC +41:34:52, based on the extent of the bright nebulosity.
Harju et al. (1998) determined a distance of 3.2 kpc for the maser source associated to this
IR source. Scalise, Rodriguez & Mendonza-Torres (1989) and Palla et al. (1991) did not detect
it in their H2O maser survey and Te Lintel Hekkert (1991) also did not find it in his OH survey.
Schutte et al. (1993) did not detect any methanol maser associated with this IR source. Chan,
Henning & Schreyer (1996) summarized some information about this object, it has a distance of
81
2 kpc (Casoli et al., 1986), it is detected in CS (2-1 transition line, Bronfman, Nyman & May,
1996) and CO (1-0 transition at 110.201 MHz, Casoli et al., 1986). Sridharan et al. (2002);
Beuther et al. (2002) reported a detection of this source in 1.2 mm and CS emission and a
distance of 3.9 kpc. Setia Gunawan et al. (2003) detected this source in their 1 400 MHz and
350 MHz continuum survey. Edris, Fuller & Cohen (2007) reported this source as an IRAS
source with thermal absorption and/or thermal emission, based on the non-detection in their
OH maser observations.
The study of the population of this IR source can put constraints to its distance and age.
Our survey can be used, mainly in the K band, to search for eclipsing binaries and other variable
stars that can be used as distance markers.
5.5.4 IRAS 20333+4127
Figure (5.13) presents the IRAS 20333+4127 infrared source, which is permeated by a thin
nebulosity. The coordinates of the center of the source are RA 20:35:10 and DEC +41:38:18.
Figure 5.13: The IRAS 20333+4127 source.
IRAS 20333+4127 is a low luminosity source, Parthasarathy, Jain & Bhatt (1992) list this
source with a flux density of 3.22 Jy in 12 µm. It does not have any measurements in the other
IRAS bands.
From all the IR sources in our field it is the faintest one, but as can be seen in Figure (5.13),
there are many low luminosity stars associated to it. Whether these stars are located behind a
denser region or belong to a loose association, remains an open question, which can be solved
with our photometry and new spectroscopic observations.
5.5.5 DR18
Figure (5.14) presents the nebulosity associated with the radio source DR18. We can see a great
cavity illuminated by a bright star. It is also possible to see very reddened objects embedded
in, or behind, the cloud. Schneider et al. (2006) reported a velocity range of +7.7 to +10.3 km
s−1 based on the 13CO 2 - 1 emission line. This velocity range puts DR18 in the near edge of
82
the Cygnus OB2 region. Another characteristic that links DR18 and Cygnus OB2 is its globular
shape and filaments pointing away from the OB2 region, which were probably created by the
ionizing winds from the hot massive stars in the center of the OB2 association.
Other IDs found in the literature related to this object are: 18P 64, 19P 40 and ECX6-28
(radio sources respectively at 408 MHz, 1 420 MHz and 4 800 MHz, Wendker, Higgs & Landecker,
1991). We adopted the DR nomenclature (Downes & Rinehart, 1966) because it is the oldest
one and furthermore because it is the most cited one for this region.
Figure 5.14: JHK composite color of DR18.
Comero´n & Torra (1999) describe DR18 as an arc-shaped nebula with a central star of V
= 15.6 obscured by AV ≃8 magnitudes and spectral type around B0.5V. The arc nebula was
interpreted by these authors as the interface between a molecular cloud, that is being eroded
by the central star, and the resulting HII region. Usually the IRAS 20333+4102 infrared source
is considered to be associated to this structure but Comero´n & Torra (1999) suggest that this
IRAS source is not associated to any of the structures found in their study (which uses Hα and
JHK filters). They suggest that IRAS 20333+4102 could be an intermediate mass protostar
embedded deeper in the molecular cloud. The winds and UV photons from the central star were
not yet capable of uncovering this source. The fact that the central star is responsible for the
ionization of the molecular cloud puts it at the same distance as DR18. Comero´n & Torra (1999)
determined a distance of 1.6 ± 0.4 kpc to the central star, which means that we can assume a
distance of 1.6 kpc to the DR18 complex. At this distance we can also assume that DR18 is
associated with the Cygnus OB2 region. This assumption is also supported by Schneider et al.
(2006) using their 13CO emission line velocities.
Because our survey covers a large area centred in the Cygnus OB2 region, we have access to a
larger area around DR18, along with a better resolution and deeper photometry than Comero´n
& Torra (1999). We intend to use our data to study the population of this region.
83
Figure 5.15: Composite color image of the IRAS 20304+4059 infrared source.
5.5.6 IRAS 20304+4059
Another interesting object, close to the center of the Cygnus OB2 association, is the IRAS
20304+4059 infrared source. There is not much information about this source in the literature.
Odenwald (1989) reports this source in his IRAS survey of young stellar objects towards the
Cygnus X region. Parthasarathy, Jain & Bhatt (1992) report an α index of +1.48, meaning that
IRAS 20304+4059 has an energy distribution broader than a blackbody, with rising flux longward
of 2µm. Such index value is associated to Class I objects (protostars), in the classification scheme
of Lada & Wilking (1984) and Lada (1987).
Figure (5.15) shows a composite color picture, made with our JHK images, of the IRAS
20304+4059 source. The coordinates of the IRAS source point to the red double object in the
center of Fig. (5.15). A close up centered on the object can be seen in Figure (5.16). Our images
have enough resolution to resolve the IRAS source into two objects, clearly visible in the center
of Fig. (5.16). The high extinction of the nebulosity, where the objects are embedded, hides
most of the other stars in the field.
These are very embedded objects and in order to show the relevance of our survey, we show
in Figure (5.17) the same area around the IRAS source, but for different surveys. North is up
and East is left in all panels. The first panel shows our K-band image, where the nebulosity
responsible for the high extinction is easily seen. The second panel shows the Digital Sky Survey
(red plate) image, where no object is visible. The third and fourth panels show the 2MASS J
and K bands, respectively. It is possible to see that our survey goes deeper in the region, and
also has a better resolution.
The coordinates and magnitudes for both objects are shown in Table (5.4). The coordinates
are mean values, obtained using the coordinates for each filter and the magnitudes were taken
only from the third night. We call the left star ‘Source A’ and the right one ‘Source B’. Because
we do not have any clue about the spectral type of these objects, we cannot calculate their
reddening, although the J-H and H-K colors, obtained using the magnitudes listed in Table
(5.4), indicate that reddening is very high. As a matter of comparison we also list in Table (5.4)
84
Figure 5.16: Central area close up of the previous image. The red double object is associated to the IRAS
source.
Figure 5.17: Panels showing the same area around the IRAS source for different surveys. The area of each panel
is approximately the same. North is up and East is left, in all panels. The arrow indicates the position of the
IRAS 20304+4059 source.
the coordinates and magnitudes of the brightest blue star close to the IRAS source. In Figure
(5.16) we can see this bright blue star in the right part of the image. Its K magnitude is similar
to the value found for Source A, but the J-H color is much redder for the IRAS sources. While
the difference between J and K, for the comparison object, amounts to ∼0.7 magnitudes, the
same difference for Sources A and B amounts to ∼4 and 5 magnitudes, respectively.
The fact that IRAS 20304+4059 is classified as a Class I source (protostar) by Parthasarathy,
Jain & Bhatt (1992), along with the fact that we report it as a double source, and also the fact
that it could be possibly associated to Cygnus OB2, strengthens the youth status of the Cygnus
OB2 association. Spectroscopic observations of these objects are necessary to better characterize
85
Table 5.4: Coordinates and magnitudes of IRAS 20304+4059.
ID α(J2000) δ(J2000) J H K
Source A 20:32:16.34 +41:09:12.60 16.708 14.295 12.706
Source B 20:32:16.17 +41:09:13.30 16.606 13.716 11.648
Comparison 20:32:13.25 +41:09:11.7 12.073 11.628 11.352
their evolutionary status.
5.6 Summary of Results
Our preliminary results indicate that star formation took place recently (a few million years ago)
in Cygnus OB2, based on the existence of cometary globules, whose tails point radially away
from the Cygnus OB2 center. These objects are small scale counterparts of other cometary
globules found in other star forming regions. These globules must harbor massive protostars in
their centers in order to sustain such sizes and resist the winds coming from the O stars of the
Cygnus OB2 association.
Another important evidence of recent star formation is the detection of the embedded sources
associated to the IRAS 20304+4059 infrared source. These sources are deeply embedded in
nebulosity, detected in our K-band images, and spectroscopic studies are necessary to determine
their true nature.
Because we are looking through a spiral arm when we observe the Cygnus X region, we were
able to detect IR sources, in the same area of Cygnus OB2. Some of these IR sources have been
catalogued previously using IRAS and 2MASS observations. However, only with our better
resolution and sensitivity will a more detailed study be possible.
Our images of the center of the Cygnus OB2 region seems to corroborate the results found
by Bica, Bonatto & Dutra (2003). However, more study is necessary to proper establish the
membership of the stars.
We do not agree with Drew et al. (2008) on the position of the center of Cygnus OB2.
In their work it is not clear if the center of Cygnus OB2 has coordinates RA 20:33:06 DEC
+41:22:30 (J2000) or if in fact these are the coordinates of the Schulte 8 Trapezium. Either way
these coordinates are wrong. The authors themselves say that the center is located NE of the
Trapezium, but these coordinates place it NW of the Trapezium.
We observed only the central region of the association and still we were able to detect more
than 250 000 objects, some of them background objects and many are probably foreground
objects. We are still working on the final database, which will be used to analyse the variability
of the stars in this region.
86
Chapter 6
Conclusions
Our Hα imaging survey of the Orion Nebula Cluster allowed us to discover 55 new visual binaries,
among 75 binaries and 3 triples detected in the separation range of 0.′′15 – 1.′′5, corresponding to
67.5 – 675 AU. We have found a binary fraction of (8.8±1.1)% for the ONC, which corroborates
previous results showing that the field binary fraction is 1.5 times larger, and that the T
associations’ binary fraction is 2.2 times larger than the value for the ONC. The number of
stars used in the analysis is bigger than in other studies, meaning that our study has better
statistics.
We detected a dramatic decrease in the number of binaries with angular separations larger
than 0.′′5. We then used this value to distinguish between two populations, the close binaries
with separation ≤ 0.′′5 and the wide binaries with separation > 0.′′5. We detected an increase in
the ratio between wide to close binaries out to a distance of 460′′ (from the center of the ONC).
After this distance the ratio levels out. This effect was interpreted by us as clear observational
evidence of the dynamical evolution of multiple systems as a result of passages through the
potential well of the ONC.
At the beginning of this work we asked two questions: 1) Does the fact that the ONC’s
binary frequency is lower than the frequency of the Galactic field, have something to do with
the dynamical evolution of the binary properties? and 2) Can we distinguish between the fast
decay mechanism (FDM) and dynamical destruction mechanism (DDM), responsible for the
dynamical evolution of the binary properties? The answer to the first question is yes. We have
shown observational evidence that due to dynamical interactions between the binaries and the
cluster’s potential well, the wide binaries are destroyed after only a few passages through the
center of the cluster. This statement also answers the second question since the dynamical
interaction is explained by the DDM hypothesis.
A byproduct of our survey is the creation of a list of brown dwarf candidates in the ONC.
We used the limited spectral information of the primary stars, available in the literature, to try
to determine the spectral type of the secondaries, based on the flux ratios between primaries
and secondaries. Although our approach is simple and may present uncertainties, it seems that
as many as 50% of the binaries in our sample have companions with spectral types later than
87
M5 and from one-sixth to one-third may have substellar companions. Some of the companions
which we classified as brown dwarf candidates were confirmed as so in observations conducted
by other authors. The correct classification of these systems, as brown dwarfs or very low
mass stars, is important for the study of the initial mass function of the Orion Nebula Cluster.
These systems can also be used as constraints to the formation and evolution scenario of low
mass stars. Low resolution infrared spectroscopy, can be used to confirm the nature of these
substellar candidates.
The environment of loose T associations seems to favor the survival of the wide binary
population, while the dense environment of star forming regions, like the ONC, seems to be
responsible for the disruption of these wide binaries. It is reasonable to suppose that the wide
binaries (mainly the late-type ones, which are less massive and hence easier to disrupt) that
populate the field were born in loose T associations and therefore survived, while the single
late-type field stars were born in multiple systems in dense star forming regions, which were
disrupted after some close encounters in their dense environment. Thus the field population
would be a mixture of single stars coming from a dense environment such as the ONC and
binary stars (or multiple systems) coming from T associations such as Taurus-Auriga, Lupus,
Ophiuchus, etc.
The visual binaries we have discovered in the ONC are also important for astrometric studies,
which are able to determine precise stellar parameters based on the solution of the projected
orbit of these systems.
Our first infrared survey, towards the Orion Nebula, was able to detect many variable stars.
With the use of statistical indices we were able to separate some of the variable stars in different
classes. Most important however was the capability to detect eclipsing binary candidates, which
will be very important to be used as constraints to formation and evolutionary models. Stars
with rotational and pulsational modulation were also detected with the use of statistical indices.
The variable star population in the Orion Nebula seems to be huge. It was difficult to find
constant stars in our sample, most of the objects are variable.
Our second infrared survey allowed us to study the center of the Cygnus OB2 association.
We observed it during 112 nights in a time interval of 217 days. Although our final database
is not complete yet we were able to use our images to detect infrared sources and cometary
globules in the Cygnus OB2 region.
The cometary globules were probably created by the winds of the myriad of O-type stars in
the center of the association. The discovery of such objects in the Cygnus OB2 region is strong
evidence in favor of the youth of this association. Because there has to be a high concentration
of mass in order to produce a cometary globule, we assume that star formation is still happening
in the association.
We were able to detect four infrared sources in our surveyed area. Some of these sources were
previously reported in other works, which suggest that these sources are probably not associated
with Cygnus OB2. They are probably background objects situated in the Perseus Arm. The
indications come from radio surveys and maser emission.
We also observed the loose association DR18, which is illuminated and eroded by a bright
blue star. The IRAS source associated to DR18 seems to be a protostar still embedded in its
cloud.
Another observational evidence that corroborates the youth status of Cygnus OB2 is the
88
double source associated to the IRAS 20304+4059 source. The double source is deeply embedded
in (or hidden behind) a nebulosity that also hides many stars in the field. Spectroscopic
observations, along with our photometry, are necessary to determine the nature of these objects.
Our near infrared images suggest that the center of the Cygnus OB2 region is a single
structure with two separate open clusters as suggested by Bica, Bonatto & Dutra (2003).
89
Chapter 7
Resumo da Tese em Portugueˆs
7.1 Introduc¸a˜o
Um dos principais ramos da astronomia atual e´ aquele que estuda a formac¸a˜o estelar. Tudo
comec¸ou com a primeira observac¸a˜o de estrelas de baixa massa de tipo espectral tardio, na
regia˜o de Taurus-Auriga, por Joy (1945). Mais tarde foi proposto por Ambartsumian (1957)
que essas estrelas seriam estrelas jovens e foram denominadas estrelas T Tauri, tomando o
nome do objeto que se pensou ser o proto´tipo da classe de estrelas do tipo solar em formac¸a˜o.
Herbig (1960) observou as estrelas que correspoderiam a`s estrelas T Tauri, mas com massas
intermedia´rias, as estrelas Ae/Be de Herbig. Desde esses trabalhos pioneiros temos visto um
aumento impressionante de observac¸o˜es, tanto em quantidade quanto em qualidade, que da˜o
suporte ao estabelecimento de diferentes modelos de formac¸a˜o, evoluc¸a˜o e estrutura estelar.
Estudamos esses sistemas estelares jovens na esperanc¸a que eles nos deˆem pistas sobre a histo´ria
do nosso pro´prio Sistema Solar. Tentamos entender como as estrelas e planetas se formam para
que possamos compreender como a Terra e os outros planetas do Sistema Solar se formaram,
quando o Sol tinha apenas alguns milho˜es de anos de idade.
Uma das coisas que aprendemos, durante o estudo de estrelas jovens, e´ que a maioria se
forma em sistemas mu´ltiplos. Se isso e´ verdade, enta˜o porque o Sol e´ uma estrela singular?
Um sistema mu´ltiplo pode ser destru´ıdo por diferentes mecanismos, tais como: o decaimento
ra´pido de sistemas de poucos corpos (Sterzik & Durisen, 1998; Reipurth & Clarke, 2001; Sterzik
& Durisen, 2003; Durisen, Sterzik & Pickett, 2001; Hubber & Whitworth, 2005; Goodwin &
Kroupa, 2005; Umbreit et al., 2005) e a destruic¸a˜o dinaˆmica causada por mu´ltiplas interac¸o˜es em
ambientes densamente populados (Kroupa, 1995a,b; Kroupa et al., 2003). O primeiro mecanismo
age no in´ıcio da formac¸a˜o estelar, enquanto o nu´cleo da nuvem molecular termina seu colapso.
Sistemas na˜o-hiera´rquicos com poucos corpos (N<3) decaem em objetos singulares, bina´rias e
sistemas mu´ltiplos hiera´rquicos. O u´ltimo mecanismo atua durante toda a vida do aglomerado.
Ele consiste na interac¸a˜o de sistemas mu´ltiplos com o poc¸o de potencial do aglomerado e com
o ambiente densamente populado que o cerca. Essa interac¸a˜o vai destruir os sistemas mu´ltiplos
que sa˜o fracamente ligados. Alguns modelos tambe´m preveˆem que influeˆncias ambientais, tais
como a temperatura do nu´cleo da nuvem molecular, poderiam ser importantes para definir a
90
frac¸a˜o inicial de bina´rias (Durisen & Sterzik, 1994; Sterzik, Durisen & Zinnecker, 2003).
As implicac¸o˜es da multiplicidade para o cena´rio de formac¸a˜o de planetas e tambe´m para a
evoluc¸a˜o das estrelas sa˜o ainda desconhecidas. Recentemente, Stassun et al. (2008) notificaram a
descoberta de um par de “estrelas geˆmeas” em um sistema bina´rio. Cada estrela tem uma massa
de 0.41±0.01M⊙, ideˆntica em 2%. O que faz esse par ser ta˜o especial e´ o fato de que, apesar
de serem similares em massa, elas teˆm temperaturas que diferem em ∼10% e luminosidades que
diferem em ∼50%, com n´ıvel de confianc¸a alto. Esse sistema bina´rio, com componentes muito
similares, embora na˜o ideˆnticas como anunciado pelos autores, e´ um claro exemplo da influeˆncia
que a multiplicidade pode exercer durante a formac¸a˜o das estrelas.
Sistemas mu´ltiplos desempenham um papel importante na astronomia. Eles sa˜o importantes
para a calibrac¸a˜o dos modelos evolutivos e na˜o e´ exagero dizer que as bina´rias eclipsantes
espectrosco´picas atuam na astronomia como “balanc¸as e re´guas”. Esses sistemas em particular,
podem nos dar paraˆmetros absolutos (massa, raio, raza˜o de teperaturas, etc) muito precisos
(.1%), que podem ser usados como v´ınculos para os modelos teo´ricos. Apesar dos sistemas
mu´ltiplos parecerem ser muito comuns nos esta´gios iniciais da formac¸a˜o estelar, apenas 22
estrelas, na pre´-sequeˆncia principal, teˆm paraˆmetros absolutos medidos atrave´s de te´cnicas
dinaˆmicas (Mathieu et al., 2007; Stassun et al., 2008). Para descobrir mais bina´rias eclipsantes
e´ necessa´rio realizar levantamentos precisos e completos com uma base temporal longa.
Desde o in´ıcio do estudo de estrelas jovens ficou claro que elas sa˜o objetos extremamente
varia´veis. Na verdade, variabilidade fotome´trica era um dos crite´rios usados para classifica´-las.
Estrelas jovens apresentam variabilidade fotome´trica e espectrosco´pica em todos os comprimen-
tos de onda, desde o ra´dio ate´ os raios-x. A maior parte da variabilidade e´ causada, de formas
distintas, pelo material circunstelar. Esse material pode causar extinc¸a˜o varia´vel quando cruza a
nossa linha de visada em direc¸a˜o a` estrela, causando quedas no brilho. O processo de acresc¸a˜o,
atrave´s do qual a estrela aumenta sua massa, pode ainda estar ativo, causando variac¸o˜es no
brilho da estrela. Em estrelas T Tauri, a zona de convecc¸a˜o cria campos magne´ticos fortes o
suficiente para produzir manchas frias de larga escala em sua fotosfera, aliado com a rotac¸a˜o da
estrela isso produz modulac¸o˜es perio´dicas na curva de luz da estrela. ‘Flares’ em estrelas jovens,
similares aos ‘flares’ solares, podem produzir variabilidade fotome´trica irregular. Os processos
de acresc¸a˜o e ejec¸a˜o de mate´ria podem tambe´m mudar o perfil de linha de va´rias espe´cies
atoˆmicas presentes na fotosfera da estrela, dando origem a` variabilidade espectrosco´pica em
estrelas jovens. Pulsac¸a˜o tambe´m foi observada em estrelas jovens (e.g. Kurtz & Marang, 1995)
situadas na faixa de instabilidade do diagrama HR.
Dito tudo isso, nos deparamos com dois importantes to´picos: estrelas jovens mu´ltiplas e
estrelas jovens varia´veis. Ambos sa˜o importantes e esta˜o diretamente associados, ja´ que estrelas
em sistemas mu´ltiplos podem ser (e, na maior parte do tempo, sa˜o) varia´veis. So´ teremos um
quadro claro do cena´rio da formac¸a˜o estelar se estudarmos sistemas individuais em detalhe.
Alta precisa˜o na determinac¸a˜o de paraˆmetros absolutos, de estrelas individuais, e´ essencial se
queremos construir modelos que realmente reproduzem os processos que ocorrem na natureza.
Mas para selecionar bons candidatos para estudo, primeiro precisamos descobri-los. Essa e´ a
importaˆncia de levantamentos de larga escala. Uma vez que os astroˆnomos esta˜o presos a` Terra e
portanto na˜o teˆm acesso ao seus experimentos, e´ necessa´rio mapear todo o ce´u em busca do “bom
experimento”. Existem muitas dificuldades em se realizar levantamentos de larga escala, por
exemplo, disponibilidade de tempo em telesco´pios para fazer observac¸o˜es em a´reas espec´ıficas do
ce´u, instrumentac¸a˜o correta, capacidade de armazenamento de grandes quantidades de dados,
91
recursos computacionais para processar toda a informac¸a˜o coletada, recursos humanos para
analisar os dados, etc. Muitos levantamentos foram feitos recentemente com a intenc¸a˜o de
mapear todo o ce´u, em diferentes comprimentos de onda, tais como: o Two Micron All Sky
Survey (2MASS, Skrutskie et al., 2006) nas bandas JHK do infravermelho pro´ximo, o Sloan
Digital Sky Survey (SDSS, Fukugita et al., 1996) nas bandas o´ticas Sloan (u’, g’, r’, i’, z’), o
All Sky Automated Survey (ASAS, Pojmanski, 1997) nas bandas o´ticas V e I, o Variable Young
Stars Optical Survey (VYSOS), que quando estiver em funcionamento vai monitorar os ce´us do
norte e do sul nas bandas Sloan, o UKIRT Infrared Deep Sky Survey (UKIDSS) considerado o
sucessor do 2MASS, apenas para citar alguns deles. Nossa intenc¸a˜o e´ diferente, na˜o queremos
monitorar todo o ce´u, queremos focar nossa atenc¸a˜o em regio˜es espec´ıficas de formac¸a˜o estelar,
com o intuito de construir uma base de dados completa, cobrindo longos per´ıodos, usualmente
centenas de dias. Em vez de olhar para o ce´u todo, vamos olhar para regio˜es espec´ıficas por um
longo intervalo de tempo.
Na nossa tentativa de melhor entender as estrelas jovens mu´ltiplas, notificamos aqui a
realizac¸a˜o de um imageamento usando a Advanced Camera for Surveys, a bordo do telesco´pio
espacial Hubble, da Nebulosa de O´rion. O principal objetivo desse levantamento e´ descobrir
novas bina´rias visuais em uma ampla regia˜o ao redor do Aglomerado da Nebulosa de O´rion.
Estudos anteriores mostraram que a frequeˆncia de bina´rias nesse aglomerado e´ menor do que
a frequeˆncia de bina´rias do campo Gala´ctico. Esse fato tem alguma relac¸a˜o com a evoluc¸a˜o
dinaˆmica do aglomerado? Podemos distinguir entre os dois mecanismos responsa´veis pela
evoluc¸a˜o dinaˆmica das propriedades das bina´rias? Tentamos responder essas questo˜es no Cap´ı-
tulo 2.
Nossa segunda incursa˜o no domı´nio da formac¸a˜o estelar esta´ relacionada com dois levanta-
mentos de larga escala no infravermelho, de duas regio˜es de formac¸a˜o estelar. Nosso principal
objetivo e´ descobrir a populac¸a˜o de estrelas jovens varia´veis da Nebulosa de O´rion e da jovem
associac¸a˜o de estrelas OB Cygnus OB2. Para essa tarefa usamos a Wide Field Camera do
United Kingdom Infrared Telescope, o instrumento infravermelho destinado a levantamentos
mais eficiente no mundo atualmente, instalado no alto do Observato´rio Mauna Kea. Apresenta-
mos no Cap´ıtulo 3 algumas te´cnicas estat´ısticas que podem ser usadas para separar a populac¸a˜o
de estrelas varia´veis nesses campos. Curvas de luz do ASAS foram usadas como testes para
melhor entender o comportamento desses ı´ndices estat´ısticos.
No Cap´ıtulo 4 apresentamos as observac¸o˜es, procedimentos e resultados do levantamento
realizado na Nebulosa de O´rion. Constru´ımos uma base de dados usando 98 noites, das 101
noites observadas cobrindo 178 dias, que e´ ate´ esse momento o levantamento infravermelho
mais longo realizado para a Nebulosa de O´rion. Muitos objetos varia´veis interessantes foram
encontrados apo´s a aplicac¸a˜o dos ı´ndices estat´ısticos.
Cygnus OB2 e´ analisada no Cap´ıtulo 5. E´ uma associac¸a˜o OB gigante, que chegou a ser
considerado um aglomerado globular jovem, dado o grande nu´mero de estrelas do tipo O nessa
regia˜o (∼100 estrelas). Cygnus OB2 esta´ situada na regia˜o de formac¸a˜o estelar Cygnus X,
atra´s da regia˜o conhecida como “Great Cygnus Rift”. Quando olhamos nessa direc¸a˜o estamos
olhando atrave´s do Brac¸o Espiral Local e do Brac¸o Espiral de Perseu e portanto podemos
ver va´rias regio˜es de formac¸a˜o estelar projetadas na mesma a´rea do ce´u. Uma ana´lse mais
cuidadosa da sua populac¸a˜o de estrelas de massa baixa e intermedia´ria, e´ necessa´ria para se
colocar limites na sua massa total e extensa˜o. Observamos o centro dessa associac¸a˜o durante
112 noites, cobrindo um intervalo de 217 dias, mas dada a enorme quantidade de dados, ainda
92
na˜o foi poss´ıvel terminar a base de dados. Entretanto, quatro noites foram usadas para construir
um cata´logo preliminar, que foi usado para estudar as propriedades desse interessante regia˜o de
formac¸a˜o estelar. Esse e´ o levantamento infravermelho mais rico ja´ feito para essa regia˜o e dada
a resoluc¸a˜o angular obtida, seremos capazes de dobrar o nu´mero de objetos associados a ela.
7.2 Bina´rias Visuais no Aglomerado da Nebulosa de O´rion
7.2.1 Introduc¸a˜o
O Aglomerado da Nebulosa de O´rion e´ responsa´vel pela ionizac¸a˜o da regia˜o HII conhecida
como Nebulosa de O´rion (M42+M43 = NGC 1976), uma das mais estudadas regio˜es do ce´u. Ele
chama tanta atenc¸a˜o porque e´ a regia˜o de formac¸a˜o estelar gigante mais pro´xima ao Sol (situado
a ∼450 pc). Suas estrelas mais massivas, conhecidas como estrelas do Trape´zio, esta˜o localizadas
em seu centro. Pensava-se que elas eram uma entidade distinta (e.g. Herbig & Terndrup,
1986) mas tem se sugerido que elas sa˜o na verdade o nu´cleo desse aglomerado (Hillenbrand
& Hartmann, 1998).
O Aglomerado da Nebulosa de O´rion na˜o e´ somente a regia˜o de formac¸a˜o estelar de alta
massa mais pro´xima, mas tambe´m e´ muito jovem, com uma idade de ∼1 milha˜o de anos. Sua
juventude e´ sustentada por:
• estrelas localizadas acima da Sequeˆncia Principal de Idade Zero (Herbig & Terndrup, 1986;
Prosser et al., 1994);
• variabilidade fotome´trica de mais de 50% das suas estrelas (Jones & Walker, 1988; Choi
& Herbst, 1996);
• existeˆncia de linhas de emissa˜o no o´tico;
• alta polarizac¸a˜o;
• excesso no infravermelho;
• ‘proplyds’ apontando na direc¸a˜o de θ1 OriC (a estrela mais massiva das estrelas do
Trape´zio).
Um resultado intrigante, que vem do estudo da formac¸a˜o e evoluc¸a˜o de sistemas bina´rios, e´
que bina´rias sa˜o mais comuns em associac¸o˜es T (associac¸o˜es com estrelas de baixa massa e que
sa˜o fracamente ligadas) do que no campo (e.g. Reipurth & Zinnecker, 1993; Simon et al., 1995;
Ducheˆne, 1999; Ratzka, Ko¨hler & Leinert, 2005). Um resultado ainda mais intrigante e´ que o
Aglomerado da Nebulosa de O´rion tem uma frequeˆncia de bina´rias menor do que associac¸o˜es
T e do que o campo (e.g. Prosser et al., 1994; Padgett, Strom & Ghez, 1997; Petr et al., 1998;
Simon, Close & Beck, 1999; Ko¨hler et al., 2006).
O estudo das propriedades dos sistemas bina´rios e´ marcado pelo trabalho detalhado de
Duquennoy & Mayor (1991, daqui em diante DM91). As propriedades da populac¸a˜o de bina´rias
do campo serve como refereˆncia para o trabalho com populac¸o˜es mais jovens. Mas podemos
confiar na populac¸a˜o de bina´rias do campo? Como ela e´ formada? Ela vem de uma mistura de
populac¸o˜es de outros aglomerados?
Atualmente existem dois modelos principais para a evoluc¸a˜o das propriedades das bina´rias:
1) decaimento dinaˆmico ra´pido de sistemas de poucos corpos (e.g. Reipurth & Clarke, 2001;
Sterzik & Durisen, 1998, 2003; Durisen, Sterzik & Pickett, 2001; Hubber & Whitworth, 2005;
93
Goodwin & Kroupa, 2005; Umbreit et al., 2005) e 2) destruic¸a˜o dinaˆmica atrave´s de mu´ltiplas
interac¸o˜es em um ambiente densamente populado (e.g. Kroupa, 1995a,b; Kroupa et al., 2003).
O primeiro atua nos esta´gios iniciais da formac¸a˜o estelar, enquanto o segundo mecanismo precisa
de mais tempo para agir e o faz por mais tempo.
Ate´ esse momento, as estat´ısticas sobre a frequeˆncia de bina´rias no Aglomerado da Nebulosa
de O´rion eram pobres e concentradas em regio˜es ao redor das estrelas do Trape´zio. Mostramos
aqui, usando nu´meros significativos, que realmente esse aglomerado tem uma frac¸a˜o de bina´rias
menor do que as associac¸o˜es T e do que o campo.
7.2.2 Resultados
1. Um total de 75 sistemas bina´rios e 3 sistemas triplos foram detectados, dos quais 55 sa˜o
novas descobertas;
2. Dentro do nosso intervalo limitado de separac¸a˜o angular, 0.′′15 – 1.′′5, que corresponde a
67.5 – 675 UA, encontramos uma frac¸a˜o de bina´rias de (8.8±1.1)% apo´s a correc¸a˜o de
contaminc¸a˜o por pares falsos;
3. A frac¸a˜o de bina´rias do tipo solar no campo, no mesmo intervalo, e´ 1.5 vezes maior e
para associac¸o˜es T e´ 2.2 vezes maior do que no Aglomerado da Nebulosa de O´rion. O
que confirma, com dados estatisticamente significantes, resultados anteriores de que esse
aglomerado e´ deficiente em bina´rias;
4. A func¸a˜o distribuic¸a˜o de separac¸a˜o para as bina´rias jovens nesse aglomerado apresenta
uma diminuic¸a˜o drama´tica em bina´rias para separac¸o˜es angulares maiores que 0.′′5, corres-
pondendo a 225 UA;
5. A raza˜o entre as distribuic¸o˜es cumulativas de bina´rias com grande separac¸a˜o e bina´rias
cerradas aumenta ate´ uma distaˆncia de 460′′ do centro do aglomerado, depois da qual a
raza˜o fica constante. Interpretamos esse resultado como uma evideˆncia observacional clara
da evoluc¸a˜o dinaˆmica da populac¸a˜o de bina´rias, como resultado de passagens atrave´s do
poc¸o de potencial do aglomerado. Esses resultados sa˜o consistentes com uma idade de 1
milha˜o de anos para o Aglomerado da Nebulosa de O´rion;
6. Parece que, ao menos em parte, a deficieˆncia de bina´rias em O´rion, quando comparado
com a populac¸a˜o do campo, pode ser entendida em termos da destruic¸a˜o das bina´rias com
grande separac¸a˜o combinado com um efeito secunda´rio da evoluc¸a˜o orbital de bina´rias em
direc¸a˜o a separac¸o˜es menores, que na˜o podem ser observadas atrave´s de imageamentos
diretos;
7. Informac¸a˜o espectral limitada sobre as estrelas prima´rias da nossa amostra indicam que
elas sa˜o estrelas T Tauri de baixa massa, com excec¸a˜o de uma estrela que e´ uma Ae/Be
de Herbig. Supondo que a maioria das bina´rias na˜o sa˜o afetadas por extinc¸a˜o diferencial,
encontramos que possivelmente 50% das bina´rias teˆm companheiras com tipo espectral
mais frio que M5 e de um sexto ate´ um terc¸o das bina´rias podem ter companheiras
subestelares (ana˜s marrons). Esse grande nu´mero de companheiras subestelares e´ de
interesse para as teorias sobre formac¸a˜o de ana˜s marrons.
O ambiente mais calmo das associac¸o˜es de estrelas T Tauri parece favorecer a sobreviveˆncia de
bina´rias com grande separac¸a˜o, enquanto ambientes densos como O´rion parecem ser responsa´veis
pela destruic¸a˜o de parte dessas bina´rias. E´ razoa´vel supor que bina´rias separadas (principalmente
94
de tipo espectral frio) que populam o campo estelar, nasceram em associac¸o˜es pouco massivas
e portanto sobreviveram, enquanto as estrelas singulares do campo nasceram em sistemas
mu´ltiplos em regio˜es densas, que foram destru´ıdas apo´s alguns encontros nesse ambiente denso.
A populac¸a˜o do campo seria uma mistura de estrelas singulares vindas dessas regio˜es mais
densas, como O´rion, e sistemas bina´rios (ou mu´ltiplos) vindos de associac¸o˜es como Taurus-
Auriga, Lupus, Ophiuchus, etc.
A descoberta de bina´rias visuais em regio˜es de formac¸a˜o estelar, como O´rion, e´ muito
importante. Apesar de estar a uma distaˆncia de ∼450 pc, observac¸o˜es astrome´tricas dessas
bina´rias, com a Advanced Camera for Surveys, por exemplo, podem ser usadas para determinar
o´rbitas projetadas e dessa forma obter paraˆmetros estelares precisos. Alguns estudos nesse
sentido foram feitos em outra regio˜es (e.g. Chauvin et al., 2005; Neuha¨user et al., 2008).
7.3 I´ndices Estat´ısticos
Variabilidade e´ uma caracter´ıstica inerente das estrelas. Podemos dizer que todas as estrelas sa˜o
varia´veis, em diferentes esta´gios de suas vidas, com diferentes amplitudes e escalas de tempo.
Durante sua vida, uma estrela viaja ao longo de diferentes caminhos em um diagrama HR e
portanto apresenta diferentes caracter´ısticas. Entretanto, e´ imposs´ıvel observar uma estrela ao
longo de sua evoluc¸a˜o, desde sua nuvem natal, atrave´s da sequeˆncia principal, em direc¸a˜o a seus
momentos finais como uma ana˜ branca, uma estrela de neutroˆns ou um buraco negro. O que
podemos fazer e´ estudar diferentes estrelas, em diferentes esta´gios evolutivos e tentar colocar
esse quebra-cabec¸a junto em um quadro coerente.
Chamaremos de estrelas varia´veis aquelas que apresentam variabilidade no brilho em uma
escala de tempo que podemos medir, seja em segundos, dias, anos, ou de´cadas, com amplitudes
que podemos medir usando CCDs, fotoˆmetros ou ate´ mesmo placas fotogra´ficas. A escala de
tempo, a forma da curva de luz e a amplitude da variac¸a˜o de brilho podem ser usadas para
classificar as estrelas varia´veis. Tipo espectral, classe de luminosidade e composic¸a˜o qu´ımica
tambe´m sa˜o paraˆmetros importantes que nos ajudam a classificar as estrelas de acordo com a
origem das variac¸o˜es.
Estrelas varia´veis podem ser divididas em dois grandes grupos: varia´veis intr´ınsecas e ex-
tr´ınsecas. Varia´veis intr´ınsecas sa˜o aquelas que variam devido a um processo f´ısico na pro´pria
estrela, enquanto varia´veis extr´ınsecas sa˜o aquelas que variam devido a um processo externo a`
estrela.
Exemplos de varia´veis intr´ınsecas sa˜o as estrelas pulsantes e as estrelas eruptivas (ou explo-
sivas). Varia´veis extr´ınsecas sa˜o, por exemplo, sistemas eclipsantes e variac¸o˜es ocasionadas por
rotac¸a˜o. Deve ficar claro que um tipo de variabilidade na˜o exclui o outro. Podemos ter, por
exemplo, uma estrela pre´-sequeˆncia principal eclipsante em que uma das componentes (ou ate´
mesmo ambas) apresenta pulsac¸o˜es.
A fim de distinguir entre diversos tipos de variabilidade e´ necessa´rio usar alguns ı´ndices
estat´ısticos. Alguns podem ser aplicados diretamente a` curva de luz de uma estrelas, outros sa˜o
ideais se aplicados a curvas de res´ıduos ou mesmo a curvas de valores me´dios. Nesse trabalho
usamos os seguintes ı´ndices:
1. magnitude me´dia;
2. desvio padra˜o;
95
3. desvio geome´trico;
4. desvio absoluto me´dio;
5. valor absoluto da amplitude ma´xima de variac¸a˜o;
6. desvio ma´ximo da magnitude me´dia;
7. estat´ıstica-R;
8. obliquidade;
9. curtose.
Para testar esses ı´ndices estat´ısticos usamos curvas de luz reais, obtidas do banco de dados do
All Sky Automated Survey1. Utilizamos curvas de luz dos seguintes tipos de estrelas varia´veis:
α2 Canum Venaticorum, Cefe´idas, δ-Scuti, bina´rias eclipsantes, RR-Lirae e varia´veis de tipo
desconhecido.
Ale´m de aplicar os ı´ndices estat´ısticos descritos acima, calculamos tambe´m como esses ı´ndices
variam ao longo das observac¸o˜es. Isso serve para simular o comportamento dos ı´ndices a` medida
que mais medidas sa˜o adicionadas a um banco de dados. Isso sera´ de muita importaˆncia para
projetos que visam monitorar regio˜es de formac¸a˜o estelar durante um longo intervalo de tempo.
Dessa forma, constru´ımos curvas para esses ı´ndices e os nomeamos ı´ndices estat´ısticos dinaˆmicos.
O uso dos ı´ndices estat´ısticos finais e dinaˆmicos nos permitiu concluir que e´ poss´ıvel usar um
conjunto deles para separar diferentes tipos de estrelas varia´veis dentro de um grande conjunto
de dados. Isso sera´ de vital importaˆncia para projetos que monitoram grandes a´reas do ce´u,
como fizemos para a Nebulosa de O´rion e para Cygnus OB2.
As estrelas bina´rias eclipsantes podem ser separadas dos demais objetos se observarmos que
as curvas dos ı´ndices dinaˆmicos sa˜o segmentadas, devido a`s repentinas mudac¸as de brilho.
Estrelas do tipo Cefeida, δ-Scuti e RR-Lira, que apresentam pulsac¸o˜es, tambe´m apresentam
curvas para a obliquidade e curtose (calculadas a partir das magnitudes me´dias) que sa˜o bem
caracter´ısticas. Estrelas do tipo α2 Canis Venaticorum, que sa˜o estrelas que apresentam varia-
bilidade devido a` rotac¸a˜o, apresentam comportamento contra´rio ao das estrelas pulsantes.
De posse dessas informac¸o˜es, podemos aplicar nossos ı´ndices estat´ısticos a` nossa base de
dados em O´rion e em Cygnus OB2.
7.4 Levantamento Fotome´trico da Nebulosa de O´rion
7.4.1 Introduc¸a˜o
Desde o in´ıcio do estudo de estrelas jovens (e.g. Joy, 1945; Herbig, 1960, 1962) a variabilidade
fotome´trica serviu como uma das caracter´ısticas que definiam as estrelas na pre´-sequeˆncia
principal. Variabilidade foi primeiramente detectada e estudada na regia˜o de comprimentos de
onda vis´ıveis, mas logo ficou claro que as estrelas jovens sa˜o varia´veis em todos os comprimentos
de onda. Muito estudos fotome´tricos, em diferentes comprimentos de onda, ja´ foram feitos na
Nebulosa de O´rion (e.g. Rebull, 2001; Carpenter, Hillenbrand & Skrutskie, 2001; Getman et al.,
2005; Flaccomio et al., 2005; Stassun et al., 1999, 2006).
Nosso propo´sito e´ observar a Nebulosa de O´rion e seus arredores, na regia˜o do infravermelho
(JHK), durante um longo intervalo de tempo, para detectar estrelas varia´veis em escalas de
1http://www.astrouw.edu.pl/asas/
96
tempo de alguns dias ate´ alguns meses. Um dos produtos desse levantamento e´ a detecc¸a˜o de
bina´rias eclipsantes, cujos paraˆmetros estelares sa˜o muito importantes para se colocar v´ınculos
nos modelos de formac¸a˜o e evoluc¸a˜o estelar.
Para realizar esse levantamento utilizamos a Wide Field Camera (WFCAM), instalada no
United Kingdom Infrared Telescope. Infelizmente, devido a problemas na subtrac¸a˜o do ce´u,
na˜o poderemos usar os dados reduzidos pelo pipeline final. Usaremos entretanto os dados
reduzidos pelo pipeline do telesco´pio, o que dara´ aos nossos resultados um status de “resultados
preliminares”.
Foram observadas 101 noites, mas apenas 98 foram reduzidas pelo pipeline. Desenvolvemos
rotinas para lidar com diversos problemas que na˜o foram resolvidos durante a reduc¸a˜o do pipeline
e para organizar as observac¸o˜es. Conseguimos estabelecer um banco de dados com as 98 noites
observadas.
7.4.2 Resultados
Apo´s aplicarmos os ı´ndices estat´ısticos a` nossa amostra de estrelas, que consiste em aproxima-
damente 8 000 estrelas, conseguimos descobrir novas estrelas bina´rias eclipsantes, estrelas com
pulsac¸o˜es e com curvas de luz criadas devido a` combinac¸a˜o de manchas estelares e rotac¸a˜o.
Nossos ı´ndices estat´ısticos se mostraram u´teis para a separac¸a˜o de diferentes tipos de estrelas
varia´veis.
7.5 Levantamento Fotome´trico em Cygnus OB2
7.5.1 Introduc¸a˜o
Umas das regio˜es de formac¸a˜o estelar mais ricas da Gala´xia esta´ na direc¸a˜o da constelac¸a˜o do
Cisne e e´ conhecida como Cygnus X. Essa e´ tambe´m uma das regio˜es mais brilhantes do ce´u
em todos os comprimentos de onda. Quando olhamos nessa direc¸a˜o estamos olhando atrave´s do
Brac¸o Espiral Local e do Brac¸o de Perseu, de forma que somos iludidos por diferentes regio˜es de
formac¸a˜o estelar sobrepostas na mesma regia˜o do ce´u. Algumas dessas regio˜es esta˜o localizadas
a algumas centenas de parsecs, enquanto outras esta˜o a alguns milhares de parsecs de distaˆncia.
Uma das regio˜es mais importantes nessa a´rea e´ a jovem associac¸a˜o OB chamada Cygnus OB2.
Essa regia˜o foi descoberta por Mu¨nch & Morgan (1953), que chamaram a atenc¸a˜o para uma
“poss´ıvel associac¸a˜o de estrelas azuis gigantes”. Muitos estudos se seguiram a esse trabalho, com
o intuito de melhor definir a populac¸a˜o de estrelas massivas dessa associac¸a˜o. Diversos estudos
apontam para uma distaˆncia de 1.7 kpc ate´ Cygnus OB2 e tambe´m para um alta extinc¸a˜o (5 –
20 magnitudes). A idade dessa associac¸a˜o ainda e´ incerta, mas estima-se que esteja entre 1 e 3
milho˜es de anos.
Atrave´s de observac¸o˜es feitas com o 2MASS, Kno¨dlseder (2000) sugeriu que Cygnus OB2
fosse considerado um aglomerado globular jovem, dado o alto nu´mero de estrelas do tipo OB
(∼2600±400 estrelas, dentre as quais ∼120±20 sa˜o estrelas O), sua geometria esfe´rica e sua
juventude. Mais tarde, Kno¨dlseder et al. (2002) reclassificou Cygnus OB2 como um super-
aglomerado. A populac¸a˜o de estrelas de alta massa de Cygnus OB2 conte´m algumas das estrelas
mais brilhantes da nossa Gala´xia e o estudo desses objetos e´ muito importante para aumentar
nosso conhecimento sobre essa parte da func¸a˜o inicial de massa.
97
Observamos Cygnus OB2 durante 112 noites, cobrindo um intervalo de 217 dias. Ainda na˜o
foi poss´ıvel terminar a catalogac¸a˜o de todos os objetos e todas as noites observadas, devido ao
enorme volume de informac¸a˜o.
7.5.2 Resultados
Utilizando nossas imagens na banda K descobrimos 6 glo´bulos cometa´rios na regia˜o de Cygnus
OB2. Todos teˆm suas caudas apontando radialmente opostas ao centro da associac¸a˜o. Esses
glo´bulos, se associados a` Cygnus OB2, sa˜o mais uma prova da juventude dessa associac¸a˜o.
Muitos desses glo´bulos teˆm estrelas bem avermelhadas em suas cabec¸as, prova de que a formac¸a˜o
estelar ainda esta´ acontecendo nessa regia˜o.
Nossas imagens tambe´m nos permitiram observar 4 fontes infravermelhas, que grac¸as a outros
estudos, parecem estar localizadas a distaˆncias maiores do que Cygnus OB2. O estudo da
populac¸a˜o estelar dessas regio˜es e´ muito importante para melhorar nosso entendimento sobre
suas idades e melhorar suas determinac¸o˜es de distaˆncia.
Determinamos a natureza mu´ltipla da fonte infravermelha IRAS 20304+4059, que tem
caracter´ısticas de um objeto protoestelar.
7.6 Concluso˜es
Nosso imageamento em Hα do Aglomerado da Nebulosa de O´rion nos permitiu descobrir 55
novas bina´rias visuais, dentre 75 bina´rias e 3 sistemas triplos no intervalo de separac¸a˜o 0.′′15
– 1.′′5. Isso nos levou a obter uma frac¸a˜o de bina´rias de (8.8±1.1)% para esse aglomerado,
corroborando resultados anteriores que indicam que a frac¸a˜o de bina´rias do campo e´ 1.5 vezes
maior e das associac¸o˜es T e´ cerca de 2.2 vezes maior que a do Aglomerado da Nebulosa de O´rion.
O nu´mero de estrelas usadas nessa ana´lise e´ maior do que em outros estudos, o que implica que
nossa estat´ıstica e´ mais significativa.
Detectamos uma diminuic¸a˜o drama´tica no nu´mero de bina´rias com separac¸a˜o angular maior
que 0.′′5. Usamos esse valor para separar nossa amostra em dois grupos: bina´rias cerradas
com separac¸a˜o ≤0.′′5 e bina´rias com separac¸a˜o >0.′′5. Detectamos um aumento na raza˜o entre
bina´rias com grande separac¸a˜o e cerradas ate´ uma distaˆncia de 460′′ (do centro do aglomerado).
Apo´s essa distaˆncia essa raza˜o se nivela. Interpretamos esse efeito como uma clara evideˆncia
observacional da evoluc¸a˜o dinaˆmica dos sistemas mu´ltiplos, como um resultado de passagens
atrave´s do poc¸o de potencial do Aglomerado da Nebulosa de O´rion.
No in´ıcio desse trabalho fizemos duas perguntas: 1) O fato da frequeˆncia de bina´rias do
Aglomerado da Nebulosa de O´rion ser menor que a frequeˆncia do campo gala´tico, esta´ relaci-
onado com a evoluc¸a˜o dinaˆmica das propriedades das bina´rias? e 2) Podemos distinguir entre
o mecanismo de decaimento ra´pido e o mecanismo de destruic¸a˜o dinaˆmica, responsa´veis pela
evoluc¸a˜o dinaˆmica das propriedades das bina´rias? A resposta para a primeira pergunta e´: sim.
No´s mostramos evideˆncias observacionais que, grac¸as a interac¸o˜es dinaˆmicas entre as bina´rias
e o poc¸o de potencial do aglomerado, as bina´rias com grandes separac¸o˜es sa˜o destru´ıdas apo´s
algumas passagens pelo centro do aglomerado. Isso tambe´m responde a segunda pergunta, uma
vez que a interac¸a˜o dinaˆmica e´ explicada pela hipo´tese do mecanismo de destruic¸a˜o dinaˆmica.
Um subproduto do nosso imageamento e´ a criac¸a˜o de uma lista de candidatas a ana˜s marrons
98
no Aglomerado da Nebulosa de O´rion. Usamos a informc¸a˜o espectral dispon´ıvel sobre as estrelas
prima´rias para tentar determinar o tipo espectral das secunda´rias. Nos baseamos na raza˜o de
fluxos entre as prima´rias e secunda´rias. Apesar de nossa tentativa ser simples e apresentar
incertezas, parece que 50% das bina´rias em nossa amostra teˆm companheiras com tipo espectral
mais frio que M5 e entre um sexto e um terc¸o podem ter companheiras subestelares. Algumas
dessas companheiras, que classificamos como ana˜s marrons, foram confirmadas como tais por
observac¸o˜es feitas por outros autores. A classificac¸a˜o correta desses sistemas, como ana˜s marrons
ou estrelas com massa muito baixa, e´ importante para o estudo da func¸a˜o de massa inicial do
Aglomerado da Nebulosa de O´rion. Esses sistemas podem tambe´m ser usados como v´ınculos para
os modelos de formac¸a˜o e evoluc¸a˜o de estrelas de baixa massa. Espectroscopia de baixa resoluc¸a˜o
no infravermelho pode ser usada para confirmar a natureza dessas candidatas subestelares.
O ambiente das associac¸o˜es T parece favorecer a sobreviveˆncia de bina´rias com grandes
separac¸o˜es, enquanto ambientes mais densos e populosos, como O´rion, parecem ser responsa´veis
pela destruic¸a˜o dessas bina´rias. E´ razoa´vel supor que essas bina´rias com grandes separac¸o˜es
(principalmente as de tipo espectral mais tardio, que sa˜o menos massivas e portanto mais fa´ceis
de serem destru´ıdas) que populam o campo, nasceram em associac¸o˜es T e portanto sobreviveram,
enquanto as estrelas singulares, do mesmo tipo espectral, nasceram em regio˜es mais densas e
foram destru´ıdas apo´s alguns encontros nesse ambiente denso. Logo, a populac¸a˜o do campo
seria uma mistura de estrelas singulares vindas de ambientes densos, tais como O´rion, e bina´rias
(ou sistemas mu´ltiplos) vindas de associac¸o˜es T, como Taurus-Auriga, Lupus, Ophiuchus, etc.
As bina´rias que descobrimos no Aglomerado da Nebulosa de O´rion sa˜o tambe´m importantes
para estudos astrome´tricos, que sa˜o capazes de fornecer paraˆmetros estelares precisos, baseados
na soluc¸a˜o das o´rbitas projetadas desses sistemas.
Nosso primeiro levantamento infravermelho, na direc¸a˜o da Nebulosa de O´rion, foi capaz
de detectar muitas estrelas varia´veis. Atrave´s do uso de ı´ndices estat´ısticos fomos capazes de
separar algumas estrelas varia´veis em diferentes classes. O mais importante, contudo, foi a
capacidade de detectar candidatas a bina´rias eclipsantes, que sa˜o muito importantes quando
usadas como v´ınculos para os modelos de formac¸a˜o e evoluc¸a˜o estelar. Estrelas com pulsac¸o˜es
e com modulac¸a˜o causada por rotac¸a˜o tambe´m foram detectadas com os ı´ndices estat´ısticos. A
populac¸a˜o de estrelas varia´veis da Nebulosa de O´rion parece ser grande. Foi dif´ıcil encontrar
estrelas constantes em nossa amostra.
Nosso segundo levantamento infravermelho nos permitiu estudar o centro da associac¸a˜o
Cygnus OB2. Observamos essa associac¸a˜o durante 112 noites em um intervalo de 217 dias.
Apesar de nosso banco de dados na˜o ter sido finalizado, fomos capazes de usar nossas imagens
para detectar fontes infravermelhas e glo´bulos cometa´rios na regia˜o de Cygnus OB2.
Os glo´bulos cometa´rios foram provavelmente criados pelos ventos da mir´ıade de estrelas do
tipo O que se encontram no centro dessa associac¸a˜o. A descoberta de tais objetos na regia˜o de
Cygnus OB2 e´ uma forte evideˆncia em favor da juventude dessa associac¸a˜o. Devido ao fato de
que e´ necessa´ria uma alta concentrac¸a˜o de massa para produzir um glo´bulo cometa´rio, supomos
que a formac¸a˜o de estrelas pode ainda estar acontecendo nessas regio˜es.
Fomos capazes de detectar quatro fontes infravermelhas na regia˜o de Cygnus OB2. Algumas
dessas fontes ja´ foram observadas por estudos anteriores, os quais sugerem que essas fontes na˜o
esta˜o associadas a` Cygnus OB2. Elas sa˜o provavelmente objetos localizados no Brac¸o de Perseu,
segundo indicac¸o˜es vindas de levantamentos em ra´dio e emissa˜o maser.
99
Tambe´m observamos a fonte de ra´dio DR18, que consiste em uma nebulosidade iluminada
e erodida por uma estrela azul bem brilhante. Uma distaˆncia de 1.6 kpc foi determinada para
essa estrela por Comero´n & Torra (1999) e com isso ela foi associada com Cygnus OB2. A fonte
IRAS associada a essa fonte de ra´dio pode ser uma protoestrela ainda embebida na nuvem.
Outra evideˆncia observacional que corrobora o status de juventude de Cygnus OB2 e´ a fonte
dupla associada a fonte IRAS 20304+4059. Essa fonte dupla esta´ embebida (ou escondida)
em uma nebulosidade que tambe´m esconde muitas estrelas no mesmo campo. Observac¸o˜es
espectrosco´picas, em conjunto com nossa fotometria, sa˜o necessa´rias para determinar a verda-
deira natureza desses objetos.
Nossas imagens infravermelhas sugerem que o centro de Cygnus OB2 e´ uma estrutura u´nica
e na˜o dois aglomerados abertos distintos, como suposto por Bica, Bonatto & Dutra (2003).
100
References
Albacete Colombo, J.F., Flaccomio, E., Micela, G., Sciortino, S., & Damiani, F., 2007, A&A ,
464, 211
Ambartsumian, V. A., 1957, IAUS, 3, 177
Audard, M., Gu¨del, M., Skinner, S. L., Briggs, K. R., Walter, F. M., Stringfellow, G., Hamilton,
R. T., & Guinan, E. F., 2005, ApJ , 635, L81
Bally, J., Licht, d., Smith, N., & Walawender, J., 2006, AJ, 131, 473
Bally, J., & Reipurth, B., 2001, ApJ , 546, 299
Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H., 1998, A&A , 337, 403
Bate, M. R., Bonnell, I. A., & Bromm, V., 2002, MNRAS , 336, 705
Baptista R., Steiner J.E., 1993, A&A , 277, 331
Bessell, M., & Brett, J. M., 1988, PASP , 100, 1134
Beuther, H., Schilke, P., Menten, K. M., Motte, F., Sridharan, T. K., & Wyrowski, F., 2002,
ApJ , 566, 945
Bica, E., Bonatto, Ch., & Dutra, C. M., 2003, A&A , 405, 991
Blanco, V. M., 1963, ApJ , 137, 513
Bodenheimer, P., Burket, A., Klein, R. I., & Boss, A. P., 2000, Protostars and Planets IV, p.
675
Brandeker, A., Jayawardhana, R., Khavari, P., Haisch, K. E., & Mardones, D., 2006, ApJ , 652,
1572
Bronfman, L., Nyman, L.-A˚, & May, J., 1996, A&AS , 115, 81
Bruch A., 1999, AJ, 117, 3031
Carpenter, J. M., Hillenbrand, L. A., & Skrutskie, M. F., 2001, AJ, 121, 3160
Cash, W., Charles, P., Bowyer, S., Walter, F., Garmire, G., & Riegler, G., 1980, ApJ , 238, L71
Casoli, F., Dupraz, C., Gerin M., Combes, F., Boulanger, F., 1986, A&A , 169, 281
Chabrier, G., Baraffe, I., Allard, F., & Hauschildt, P. H., 2000, ApJ , 542, 464
Chan, S. J., Henning, T., & Schreyer, K., 1996, A&AS , 115, 285
Chauvin, G., Lagrange, A.-M., Lacombe, F., Dumas, C., Mouillet, D., Zuckerman, B., et al.,
2005, A&A , 430, 1027
Choi, P., & Herbst, W., 1996, AJ, 111, 283
Comero´n, F., Pasquali, A., Rodighiero, G., Stanishev,V., De Filippis, E., Lo´pez Mart´ı, B.,
Ga´lvez Ortiz, M.C., Stankov, A., & Gredel, R., 2002, A&A , 389, 874
Comero´n, F. & Torra, J., 1994, ApJ , 423, 652
Comero´n, F. & Torra, J., 1999, A&A , 349, 605
Comero´n, F. & Torra, J., 2001, A&A , 375, 539
Conti, P.S. & Blum, R.D., 2002, ApJ , 564, 827
Correia, S., Zinnecker, H., Ratzka, Th., & Sterzik, M. F., 2006, A&A , 459, 909
101
Dahn, C. C., Harris, H.C., Vrba, F.J., Guetter, H.H., Canzian,B., Henden, A.A., Levine, S.E.,
Luginbuhl, C.B., Monet, D.G., Pier, J.R., Stone, R.C., & Walker, R.L., 2002, AJ, 124, 1170
De Becker, M., Rauw, G., Sana, H., Pollock, A.M.T., Pittard, J.M., Blomme, R., Stevens, I.R.,
& Van Loo, S., 2006, MNRAS , 371, 1280
De Becker, M., Rauw, Pittard, J.M., Sana, H., Stevens, I.R., & Romero, G.E., 2007, A&A , 472,
905
Downes, D., & Rinehart, R., 1966, ApJ , 144, 937
Drew, J.E., Greimel, R., Irwin, M.J., & Sale, S.E., 2008, MNRAS , 386, 1761
Ducheˆne, G., 1999, A&A , 341, 547
Ducheˆne, G., Bouvier, J., Bontemps, S., Andre´, P., & Motte, F., 2004, A&A , 427, 651
Duncan, D. K., 1993, ApJ , 406, 172
Duquennoy, A., & Mayor, M., 1991, A&A , 248, 485
Durisen, R. H., & Sterzik, M .F., 1994, A&A , 286, 84
Durisen, R. H., Sterzik, M. F., & Pickett, B. K., 2001, A&A , 371, 952
Edris, K. A., Fuller, G. A., & Cohen, R. J., 2007, A&A , 465, 865
Edwards, S., et al., achar resto, 1993, AJ, 106, 372
Figer, D.F., Kim, S.S., Morris, M., Serabyn, E., Rich, R.M., McLean, I.S., 1999, ApJ , 525, 750
Flaccomio, E., Micela, G., Sciortino, S., Feigelson, E. D., Herbst, W., Favata, F., Harnden, F.
R., Jr., & Vrtilek, S. D., 2005, ApJS , 160, 450
Fukugita, M., Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K., Schneider, & D. P., 1996, AJ,
111, 1748
Getman, K. V., Feigelson, E. D., Grosso, N., McCaughrean, M. J., Micela, G., Broos, P.,
Garmire, G., & Townsley, L., 2005, ApJS , 160, 353
Goodwin, S. P., & Kroupa, P., 2005, A&A , 439, 565
Goodwin, S. P., Kroupa, P., Goodman, A., & Burkert, A., 2007, Protostars and Planets V, p.
133
Guimara˜es, M. M., Alencar, S. H. P., Vieira, S. L. A., & Corradi, W. J. B., 2006, A&A , 457,
581
Haisch, K. E., Jr., Greene, T. P., Barsony, M., & Stahler, S. W., 2004, AJ, 127, 1747
Hanson, M.M., 2003, ApJ , 597, 957
Harju, J., Lehtinen, K., Booth, R. S., & Zinchenko, I., 1998, A&AS , 132, 211
Heggie, D. C., 1975, MNRAS , 173, 729
Herbig, G. H., 1960, ApJS , 4, 337
Herbig, G. H., 1962, Adv. Astr. Astrophys., 1, 47
Herbig, G. H., 1989, in Proc. ESO Workshop on Low Mass Star Formation and Pre-Main
Sequence Objects, ed. B. Reipurth (Garching: ESO), 233
Herbig, G. H., 2008, AJ, 135, 637
Herbig, G. H., & Bell, K., R., 1988, Lick Obser. Bulletin, 1111
Herbig, G. H., & Terndrup, D. M., 1986, ApJ , 307, 609
Herrero, A., Corral, L.J., Villamariz, M.R., & Mart´ın, E.L., 1999, A&A , 348, 542
Herrero, A., Puls, J., & Najarro, F., 2002, A&A , 396, 949
Higgs, L. A., Landecker, T. L., Israel, F. P. & Bally, J., 1991, J. R. Astron. Soc. Can., 85, 24
Hillenbrand, L. A., 1997, AJ, 113, 1733
Hillenbrand, L. A., & Hartmann, L. W., 1998, ApJ , 492, 540
Hills, J. G., 1975, AJ, 80, 809
Hoffmeister C., Richter G., Wenzel W., 1985, Variable Stars, Springer-Verlag, p.253
102
Hubber, D. A., & Whitworth, A. P., 2005, A&A , 437, 113
Hutchings, J. B., 1981, PASP , 93, 50
Johnson H.L. & Morgan W.W., 1954, ApJ 119, 344
Jones, B. F., & Walker, M. F., 1988, AJ, 95, 1755
Joy, A.H., 1945, ApJ , 102, 168
Kiminki, D.C., Kobulnicky, A., Kinemuchi, K., Irwin, J.S., Fryer, C.L., Berrington, R.C., Uzpen,
B., Monson, A.J., Pierce, M.J., & Woosley, S.E., 2007, ApJ , 664, 1120
Kno¨dlseder, J., 2000, A&A , 360, 539
Kno¨dlseder, J., Cervin˜o, M., Le Duigou, J.-M., Meynet, G., Schaerer, D., & von Ballmoos, P.,
2002, A&A , 390, 945
Ko¨hler, R., Petr-Gotzens, M. G., McCaughrean, M. J., Bouvier, J., Ducheˆne, G., Quirrenbach,
A., & Zinnecker, H., 2006, A&A , 458, 461
Kroupa, P., 1995a, MNRAS , 277, 1491
Kroupa, P., 1995b, MNRAS , 277, 1507
Kroupa, P., 2000, NewA, 4, 615
Kroupa, P., Aarseth, S., & Hurley, J., 2001, MNRAS , 321, 699
Kroupa, P., Bouvier, J., Ducheˆne, G., Moraux, E., 2003, MNRAS , 346, 354
Kroupa, P., & Bouvier, J., 2003, MNRAS , 346, 369
Kroupa, P., Petr, M. G., & McCauhrean, M. J., 1999, NewA, 4, 495
Kuchar, T. A., & Clark, F. O., 1997, ApJ , 488, 224
Kurtz, D. W., & Marang, F., 1995, Delta Scuti Star Newsletter, 8, 2
Lada, C. J., 1987, Star Forming Regions, IAUS, No. 115, 1
Lada, C. J., 2006, ApJ , 640, L63
Lada, C. J. & Wilking, B.A., 1984, ApJ , 287, 610
Lafler, J., & Kinman T.D. 1965, ApJS , 11, 216
Leinert, C., Zinnecker, H., Weitzel, N., et al., 1993, A&A , 278, 129
Lima, G. H. R. A., 2002, Dissertac¸a˜o de Mestrado, DF-ICEx-UFMG
Lockman, F. J., 1989, ApJS , 71, 469
Lorenzetti, D., Giannini, T., Calzoletti, L., Puccetti, S., Antoniucci, S., Arkharov, A. A., Di
Paola, A., Larionov, V. M., & Nisini, B., 2006, A&A , 453, 579
Lucas, P. W., Roche, P. F., & Tamura, M., 2005, MNRAS , 361, 211
Luhman, K. L., Stauffer, J. R., Muench, A. A., Rieke, G. H., Lada, E. A., Bouvier, J., & Lada,
C., 2003, ApJ , 593, 1093
Luhman, K. L., Whitney, B. A., Meade, M. R., Babler, B. L., Indebetouw, R., Bracker, S., &
Churchwell, E. B., 2006, ApJ , 647, 1180
Massey, P., & Thompson, A.B., 1991, AJ, 101, 1408
Mathieu, R. D., 1994, ARA&A, 32, 465
Mathieu, R. D., Baraffe, I., Simon, M., Stassun, K. G., & White, R., 2007, Protostars and
Planets V, p. 411
Meeus, G., & McCaughrean, M. J., 2005, Astron. Nachr., 326, 977
Moffat, A.F.J., Drissen, L., Shara, M.M., 1994, ApJ , 436, 183
Molinaris, S., Brand, J., Cesaroni, R., & Palla, F., 1996, A&A , 308, 573
Morgan W.W., Meinel A.B. & Johnson H.M., 1954, ApJ, 120, 506
Mu¨nch L. & Morgan W.W., 1953, ApJ, 118, 161
Naze´, Y., De Becker, M., Rauw, G., & Barbieri, C., 2008, A&A , 483, 543
103
Neuha¨user, R., Mugrauer, M., Seifahrt, A., Schmidt, T. O. B., & Vogt, N., 2008, A&A , 484,
281
O’Dell, C. R., 2001, AJ, 122, 2662
O’Dell, C. R., & Wong, K., 1996, AJ, 111, 846
Odenwald, S. F., 1989, AJ, 97, 801
Padgett, D. L., Strom, S. E., & Ghez, A., 1997, ApJ , 477, 705
Palla, F., Brand, J., Cesaroni, R., Comoretto, G., & Felli, M., 1991, A&A , 246, 249
Parsamian, E. S., Mujica, R., & Corral, L., 2002, Astrophysics, 45, 393
Parthasarathy, M., Jain, S. K., & Bhatt, H. C., 1992, A&A , 266, 202
Patience, J., Ghez, A. M., Reid, I. N., & Matthews, K., 2002, AJ, 123, 1570
Petr, M. G., Coude´ du Foresto, V., Beckwithm S., Richichi, A., & McCaughrean, M. J., 1998,
ApJ , 500, 825
Pojmanski, G., 1997, Acta Astronomica, 47, 467
Prosser, C. F., Stauffer, J. R., Hartmann, L., Soderblom, D. R., Jones, B. F., Werner, M. W.,
& McCaughrean, M. J., 1994, ApJ , 421, 517
Ratzka, T., Ko¨hler, R., & Leinert, Ch., 2005, A&A , 437, 611
Rebull, L. M., 2001, AJ, 121, 1676
Reddish V.C., Lawrence L.C. & Pratt N.M., 1966, Publ. R. Obser. Edinburgh, 5, 111
Reipurth, B., 2000, AJ, 120, 3177
Reipurth, B., & Zinnecker, H., 1993, A&A , 278, 81
Reipurth, B., & Clarke, C. J., 2001, AJ, 122, 432
Reipurth, B., Guimara˜es, M. M., Connelley, M. S., & Bally, J., 2007, AJ, 134, 2272
Rutman J., 1978, Proc. IEEE, 66, 1048
Scalise, E., Jr., Rodriguez, L. F., & Mendonza-Torres, E., 1989, A&A , 221, 105
Schneider, N., Bontemps, S., Simon, R., Jakob, H., Motte, F., Miller, M., Kramer, C., & Stutzki,
J., 2006, A&A , 458, 855
Scholz, R.-D., Meusinger, H., Irwin, M. 1997, A&A, 325, 457
Schulte D.H., 1956a, ApJ, 123, 250
Schulte D.H., 1956b, ApJ, 124, 530
Schulte D.H., 1958, ApJ, 128, 41
Schutte, A. J., van der Walt, D. J., Gaylard, M. J., & MacLeod, G. C., 1993, MNRAS , 1993,
261, 783
Setia Gunawan, D. Y. A., De Bruyn, A. G., vann der Hucht, K. A., & Williams, P. M., 2003,
ApJS , 149, 123
Sicilia-Aguilar, A., et al., achar resto, 2005, AJ, 129, 363
Simon, M., Ghez, A. M., Leinert, Ch., Cassar, L., Chen, W. P., Howell, R. R., Jameson, R. F.,
Matthews, K., Neugebauer, G., & Richichi, A., 1995, ApJ , 443, 625
Simon, M., Close, L. M., & Beck, T. L., 1999, AJ, 117, 1375
Simonetti J.H., Cordes J.M., Heeschen D.S., 1985, ApJ , 296, 46
Skrutskie, M. F., Cutri, R. M., Stiening, R., Weinberg, M. D., Schenieder, S., Carpenter, J. M.,
Beuchman, C., Capps, R., Chester, T., et al., 2006, AJ, 131, 1163
Sridharan T. K., Beuther, H., Schilke, P., Menten, K. M., & Wyrowski, F., 2002, ApJ , 566, 931
Stassun, K. G., Mathieu, R. D., Mazeh, T., & Vrba, F. J., 1999, AJ, 117, 2941
Stassun, K. G., van den Berg, M., Feigelson, E., & Flaccomio, E., 2006, ApJ , 649, 914
Stassun, K. G., Mathieu, R. D., Cargile, P. A., Aarnio, A. N., Stemple, E., & Geller, A., 2008,
Nature, 453, 1079
104
Sterzik, M. F., & Durisen, R. H., 1998, A&A , 339, 95
Sterzik, M. F., & Durisen, R. H., 2003, A&A , 400, 1031
Sterzik, M. F., Durisen, R. H., & Zinnecker, H., 2003, A&A , 411, 91
Taylor, A. R., Gross, W. M., Coleman, P. H., van Leeuwen, J., & Wallace, B. J., 1996, ApJS ,
107, 239
Taylor, A. R., Gibson, S. J., Peracaula, M., Martin, P. G., Landecker, T. L., et al.., 2003, AJ,
125, 3145
Te Lintel Hekkert, P., 1991, A&A , 248, 209
Torres-Dodgen A.V., Tapia M. & Carroll M., 1991, MNRAS, 249, 1
Umbreit, S., Burkert, A., Henning, T., Mikkola, S., & Spurzem, R., 2005, ApJ , 623, 940
Van Loo, S., Blomme, R., Dougherty, S.M., & Runacres, M.C., 2008, A&A , 483, 585
Vink, J.S., Drew, J.E., Steeghs, D., Wright, N.J., Martin, E.L., Ga¨nsicke, B.T., Greimel,R., &
Drake, J., 2008, MNRAS , 387, 308
Voelcker, K. P. J., 1974, unpublished PhD thesis, Univ. Heidelberg
Wendker, H. J., 1984, A&AS , 58, 291
Wendker, H. J., Higgs, L. A., & Landecker, T. L., 1991, A&A , 241, 55
Wendker, H. J., Higgs, L. A., & Landecker, T. L., 1996, Astro. Nachr., 317, 35
Westerhout, G., 1958, Bull. Astron. Inst. Netherlands, 14, 215
Williams, S. J., Fuller, G. A., & Sridharan, T. K., 2004, A&A , 417, 115
105