YSOVAR: THE FIRST SENSITIVE, WIDE-AREA, MID-INFRARED PHOTOMETRIC MONITORING OF THE ORION NEBULA CLUSTER

The Orion star-forming region has been the subject of more low-mass young star variability studies than any other region. While Haro (1969), Herbig & Kameswara Rao (1972), and Walker (1978) all conducted pioneering studies of the variability of young stars in Orion, the modern era using two-dimensional imaging cameras was initiated by Herbst and his team (Attridge & Herbst 1992; Choi & Herbst 1996). They obtained multi-year, optical time-series photometry of several regions of the Orion Nebula Cluster (ONC), eventually obtaining light curves for hundreds of pre-main-sequence (PMS) stars, often spanning several years. Those data were primarily used to derive rotation periods—made possible by the fact that the photospheres of young stars are often heavily spotted (cold and/or hot spots), resulting in flux variations modulated at the star's rotation period. Many other groups subsequently obtained time-series photometry of other portions of the ONC, using optical (Stassun et al. 1999; Rebull 2001; Herbst et al. 2002; Irwin et al. 2007) and near-IR (Carpenter et al. 2001) imaging data. In many cases, the light curve shapes are well fit by hot or cold spots, normally at moderately high latitudes since the light curves are seldom "flat-bottomed." However, for a significant fraction of the light curves, particularly those in the near-IR, spots do not seem to provide a good explanation of the observed variability (Carpenter et al. 2001).

Time-series photometry of young stellar objects (YSOs) can also address issues other than rotational velocities such as flares and flare frequency (Parsamian et al. 1993), disk instabilities (Fedele et al. 2007; Herbst et al. 2010; Plavchan et al. 2008a), and eclipsing binaries as tests of PMS isochrones (Stassun et al. 2004, 2006; Cargile et al. 2008). Mid-IR time-series photometry of YSOs could, in principle, detect variations in the temperature or spatial distribution of warm dust in the inner circumstellar disks of these stars, thereby informing models of disk evolution and perhaps of planet formation. We recently completed the first sensitive, wide-area, mid-IR time-series photometric monitoring program for the ONC using the Infrared Array Camera (IRAC; Fazio et al. 2004) on the Spitzer Space Telescope (Werner et al. 2004). We report here some early results from that program, with more detailed analysis of specific classes of variability to follow.

2.1. Spitzer Observations

2.2. Ground-based Data

Light curves were extracted in both IRAC bands for 1249 Class I and II previously known Orion YSOs with IR excess (T. Megeath et al. 2011, in preparation) and inferred masses between 0.03 and ∼2 M_☉, plus 820 other likely Orion members. This latter set of objects is formed by sources that fulfill at least one of the following conditions: (1) has been labeled as a member by proper motion studies such as Parenago (1954), Jones & Walker (1988), McNamara (1976), and Tian et al. (1996), (2) has been labeled as a member according to Hillenbrand et al. (1998), (3) has been labeled as a member by X-ray studies (Getman et al. 2005), (4) has H_α profiles typical of weak-lined T Tauris (WTTs; Tobin et al. 2009; Fűrész et al. 2008), (5) has been labeled as variable by Carpenter et al. (2001), or (6) finally, we have included all the sources that have been claimed to be periodic variables. Most of these stars are WTTs according to the Megeath et al. photometry, though some may be sources with excesses that escaped previous detection. In addition to these 2069 Orion stars, another 147 Class I and Class II sources plus 209 Orion members without IR excess have light curves in just one of the two IRAC bands.

We estimated the noise in the IRAC light curves by measuring the typical scatter for all the objects in the field. In that way, we estimate a relative per channel photometric accuracy better than 3% down to [3.6] = 14 and better than 10% for objects as faint as [3.6] = 16. For the J band, we derive accuracies of ∼2% down to J = 16.5 and of ∼10% to J = 18.5 for the ensemble. Note that the accuracy varies significantly over the surveyed field due to nebular emission.

Approximately 70% of the YSOs with IR excess are variable based primarily on the Stetson index calculated using both IRAC channels together (Welch & Stetson 1993; see Table 3 for statistics). The variable sources were visually inspected and, in addition to the Stetson variables, we considered as variables as well a few YSOs for which a period could be found even when the Stetson index alone would not reflect the variability. About 3% of the variable YSOs have been selected in this way. Usually these sources have very low amplitude; however, most of them have previously detected periods that match the ones derived here and thus we believe the variability is real. The variable stars exhibit diverse behavior with time, including slow drifts in brightness over one month or longer that may be caused by a slow change in accretion rate or by self-shadowing in a warped disk (Figure 1(a)), rapid chaotic flux variations probably caused by changes in the accretion geometry (Figure 1(b)), flares (Figure 1(c)), stars with light curves that look "periodic" for part but not all of the 40 day period or for which the periodic variations are mixed with other longer term changes (Figure 1(d)), periodic or pseudo-periodic variations arising from photospheric spots and/or disk warps (Figure 1(e)), changes in color that may be produced by variable extinction (Figure 1(f)), and stars that show narrow or broad dips in the observed flux superimposed on an otherwise nearly constant flux level (see the following section).

Zoom In Zoom Out Reset image size

Figure 2 illustrates the value of having contemporaneous optical/NIR/mid-IR monitoring (12%, 85%, and 37% of the Orion stars have K_s-, J-, and I_c-band light curves, respectively). When compared to the near-IR and optical data, the mid-IR light curves usually have similar shapes. We have derived the amplitude of the light curves at each band as the difference between the maximum and minimum values of the magnitude at that bandpass avoiding single deviant data points. There are generally larger amplitudes at shorter wavelengths (∼60% of the variables with shorter wavelength data—Figure 2(a)), though in some cases, the amplitudes of variation are similar at all bands (∼30% of the variables with shorter wavelength data—Figure 2(b)). In a small number of cases, the IRAC variations are smaller or non-apparent at shorter wavelengths, indicating that the source of variability is probably in the disk (∼10% of the variables with shorter wavelength data—Figures 2(c) and (d)). In only two cases do we find a phase difference between the optical and near-IR data relative to IRAC data (Figure 2(e)). Some of these effects can be explained by the presence of one or more hot spots on the surface of the star and/or the existence of warps in its disk.

Zoom In Zoom Out Reset image size

Several of the objects deserve particular mention. Figure 1(d) shows the well-known intermediate-mass protostar OMC-2 IRS 1, for which outflows have been detected (Yu et al. 1997; Reipurth et al. 1999; Takahashi et al. 2008; Anathpindika & Whitworth 2008); a similar light curve was previously reported for Orion BN/KL (J053514.12−052222.9; Hillenbrand et al. 2001), though the physical mechanism for such variability in these B-type (Johnson et al. 1990; Hillenbrand et al. 2001) YSOs is not obvious. In Figure 2(b), we show the protostar OMC-2 IRS 2, which has also been associated with a jet (Yu et al. 1997; Anathpindika & Whitworth 2008) but for which only one scattering lobe is seen, suggesting that its disk is significantly less inclined to the line of sight (Rayner et al. 1989). Figure 2(f) shows the light curve of a star identified by Stassun et al. (1999) as the fastest rotator in a study of the period distribution in the ONC (P = 0.27 days, Star 1161 of Stassun et al. 1999; R.A. 05 34 45.93, decl. −04 49 22.0, from his Table 1). Given its spectral type (K5; Rebull 2001), at 1–2 Myr, this period is shorter than breakup. However, the VI_cJHK photometry for the star suggests an age older than 30 Myr if it is at Orion distance. Because the light curve in Stassun et al. appears to have identical shape and amplitude to our 2009 light curve, we suspect this object is a tidally distorted binary, but it could also be a single near zero-age main-sequence rapid rotator (similar to H ii 1883 in the Pleiades).

In general, our variable YSOs show colorless changes at IRAC bands (∼88% of our sample shows IRAC [3.6]−[4.5] rms amplitudes smaller than 0.05 mag). The objects that do show IRAC color variations usually become redder as they fade. The typical peak-to-peak amplitudes in each IRAC band are ∼0.2 mag. The most extreme stars in our sample have changes in amplitude as large as ∼1.8 mag and the largest change in color is ∼0.3 mag.

We have used the NStED Periodogram Service³⁵ to derive periods for our sample. This service uses the Lomb–Scargle (Scargle 1982), Box-fitting Least Squares (Kovács et al. 2002), and Plavchan (Plavchan et al. 2008b) algorithms for computing periodograms from time-series data. Given the cadence of our data, the periodograms are formally reliable for periods between 1 and 40 days; however, we have found 11 sources with periods shorter than a day. Those sources all have periods from the literature that match our determinations and the light curves are so well phased that we believe the periods are real. We have not retained any period longer than 25 days. In this way we can determine a period for 23% of the variable YSOs with IR excess; 7% of these are Class Is and 93% are Class IIs. Among the non-excess sources, ∼44% are variables and the predominant variability is periodic, probably due to photospheric spots; 33% of the WTTs have periods discernible from the Spitzer data. In the total ONC sample (1249 Spitzer excess sources plus 820 Orion members not known to have IR excess), we find 150 new periods (see Table 4). About 30% of our total sample had previously published periods. Of these, we recover the published period for 57% of the WTTs and 47% for the YSOs with IR excess. This difference is likely due to the thermal emission from dust at IRAC wavelengths significantly degrading the sensitivity to photospheric phenomena but also because the stars themselves may change (Rebull 2001).

Finally, we have identified nine eclipsing binaries in our sample. Four of them, 2MASS 05352184−0546085 (Stassun et al. 2006), V1174 Ori (Stassun et al. 2004), Par1802 (Cargile et al. 2008), and JW 380 (Irwin et al. 2007) are already known. A fifth one, Theta 1 Ori E, is a known spectroscopic binary (Herbig & Griffin 2006; Costero et al. 2008) which was flagged as potentially eclipsing, but no confirmatory photometry had been obtained until now. We are working to obtain spectroscopic confirmation of the four new eclipsing binary candidates and will report on their status in a future paper. Their coordinates and tentative periods can be found in Table 5.

4.1. Identification and Characterization

One of the most surprising and interesting classes of variability we find is narrow flux dips. This subset of 41 variables shows brief, sharp drops in their flux density with durations of one to a few days. These objects are members of the "Type III" variable class identified by Herbst et al. (1994) based on an optical synoptic data set; their most famous prototype is AA Tau (Bouvier et al. 2003). A few examples of such "dipper" light curves are shown in Figure 3. Among the identified cases, some exhibit only one dip, but typically >1 dip is seen over the 40 days. Periodic dips are also detected in about a third of these objects, with periods from ∼2 to 14 days. The dip amplitude is typically several tenths of a magnitude in the IRAC bands and ranges from 0.16 to 1.5 mag at J.

Zoom In Zoom Out Reset image size

The criteria imposed for selecting dipper stars were (1) a distinctive dip extends over several IRAC epochs unless a single epoch IRAC dip is corroborated by the independent but contemporaneous shorter wavelength data. Implicitly, this means the IRAC-only dips will be broader in time, on average, than dips where shorter wavelength data confirm the dip. (2) If based on just an apparent dip in one IRAC epoch, both IRAC channels must agree and the I_c- or J-band data must show a larger amplitude dip by at least 50% to avoid misidentification of eclipsing binary stars as AA-Tau analogs. (3) The "continuum" must be flat enough so that the IRAC dip "stands out." Although their identification by eye is generally straightforward, it is not as straightforward to determine the frequency of such dips in the overall data set. To do so, one must account for the likely superposition of such events with other types of variability that are occurring in the same stars. The optical and near-IR data were obtained at intervals of one to a few days as compared with the IRAC data which were obtained at half-day intervals, thus preventing confirmation of some IRAC dips. To detect these events requires good signal-to-noise ratio (S/N) data in all the relevant bandpasses, which eliminates many of the fainter stars and/or stars in regions where the nebular background is bright or very structured.

Based on the above criteria, we identify 38 of the Class I or II YSOs as having one or more AA-Tau-like flux dips. Given the coverage, cadence, and S/N that we have in different filters, we estimate that we would be able to identify ∼70% of broad dips (Figures 3(a) and (d)) and about 40% of the narrow dips (those present in just one IRAC epoch, Figures 3(b) and (e)). Thus, at present, we estimate the overall frequency of dipper stars at ∼5% or higher. We note that this fraction is much lower than the 28% estimated by Alencar et al. (2010) for periodic "AA-Tau-like" behavior in optical wavelength CoRoT data among a smaller set of young stars in NGC 2264. There are, in fact, a variety of reasons that CoRoT detects a larger fraction of "dipper" sources in NGC 2264 than the Spitzer YSOVAR survey of the ONC: CoRoT's much better relative photometric accuracy (<1 mmag) and higher cadence (every 512 s during 23 days of uninterrupted observations), the larger amplitude of these dips at shorter wavelengths, the selection criteria itself, and the greater complexity of the mid-IR light curves due to the disk contribution. However, there may also be some real difference between the YSOs in the two clusters. Examination of the Carpenter et al. (2001) monitoring NIR data reveals that ∼50% of our dippers displayed similar behavior 10 years ago, suggesting some stability in physical mechanism inducing the variations. We have identified 13 objects, or 34% of the "dipper" sample, as being periodic with P < 40 days (the observing window). Carpenter et al. (2001) derived similar periods (less than 0.6 days of difference and usually less than 0.1 days) for five of these objects. The dependence of depth on wavelength is roughly consistent with that expected for the standard extinction law, as previously reported for AA Tau (Bouvier et al. 2003). Coordinates, periods, and amplitudes for the dippers can be found in Table 6.

Table 6. ONC AA-Tau Analogs

Sourcea	Mean I	Mean J	Mean [3.6]	Mean [4.5]	Dip Depth Ib	Dip Depth Jb	Dip Depth [3.6]b	Dip Depth [4.5]b	Dip Period
	(mag)	(mag)	(mag)	(mag)	(mag)	(mag)	(mag)	(mag)	(days)
ISOY_053432.02−052742.7	...	13.255	10.437	10.015	...	0.7	0.2	0.17	...
ISOY_053435.98−045218.0	...	13.972	12.131	11.754	...	0.6	0.2	0.15	2.34 ± 0.12
ISOY_053440.64−050658.7	...	13.989	10.509	10.136	...	1.1	0.5	0.38	3.85 ± 0.39
ISOY_053446.79−052129.2	...	13.424	11.213	10.862	...	0.42	0.25	0.20	...
ISOY_053450.72−045836.8	...	12.147	10.193	9.968	...	0.34	0.20	0.26	6.32 ± 0.38
ISOY_053450.87−053929.2	14.319	12.623	10.688	10.35	...	0.20	0.12	0.10	...
ISOY_053453.74−051715.0	...	16.316	13.158	12.675	...	0.7	0.45	0.34	...
ISOY_053455.10−044941.5	...	13.253	11.09	10.749	...	0.85	0.27	0.18	...
ISOY_053502.18−051845.3	...	16.589	12.772	12.347	...	0.45	0.16	0.13	...
ISOY_053502.76−053202.8	...	14.451	12.566	12.217	...	1.1	0.35	0.25	3.24 ± 0.78
ISOY_053503.16−051830.0	14.424	12.904	11.038	10.736	1.35	1.25	0.58	0.45	7.57 ± 0.72
ISOY_053503.23−051753.3	15.098	13.002	10.696	10.252	1.40	1.15	0.35	0.25	...
ISOY_053503.96−051859.8	15.105	13.069	10.841	10.423	0.95	0.60	0.40	0.35	...
ISOY_053504.68−044621.9	...	15.488	11.697	11.148	...	1.55	0.6	0.5	...
ISOY_053504.86−054426.7	18.038	15.385	11.701	11.305	...	0.6	0.5	0.4	...
ISOY_053504.87−052057.5	16.334	13.455	10.665	10.271	0.95	0.65	0.22		8.22 ± 1.23
ISOY_053504.90−053949.7	...	15.952	13.007	12.658	...	0.37	0.25	0.2	...
ISOY_053507.53−051114.4	...	12.409	9.991	9.783	...	0.32	0.16	0.13	...
ISOY_053508.07−054853.8	15.701	13.16	9.927	9.45	1.25	1.2	0.18	0.14	7.97 ± 1.59
ISOY_053508.53−052518.0	14.210	12.376	10.231	9.973	1.3	1.07	0.40	0.28	...
ISOY_053508.60−052619.4	11.956	13.274	11.03	10.68	0.25	0.18	0.1	0.1	...
ISOY_053510.07−051706.8	13.722	14.587	10.752	10.068	1.1	0.8	0.25	0.2	...
ISOY_053510.19−052021.0	11.58	13.571	10.938	10.642	0.6	0.45	0.15	0.11	...
ISOY_053510.94−043957.6	...	13.636	11.342	11.075	...	1.25	0.3	0.3	...
ISOY_053510.99−051521.8	...	16.255	9.17	8.561	...	1.7	0.6	0.35	11.29 ± 0.56
ISOY_053516.33−051538.0	...	11.961	9.617	9.354	...	0.6	0.23	0.20	...
ISOY_053516.74−052020.0	...	13.275	9.621	9.34	...	0.8	0.2	0.18	...
ISOY_053518.42−044000.2	...	13.522	12.018	11.67	...	0.28	0.12	...	...
ISOY_053518.70−051801.8	12.457	13.932	10.153	9.85	0.7	1.1	0.4	0.4	7.37 ± 0.80
ISOY_053523.98−052509.8	10.991	13.227	10.6	9.774	0.35	0.35	0.17	0.11	...
ISOY_053524.14−044930.3	...	13.941	11.904	11.515	...	0.79	0.46	0.44	...
ISOY_053525.51−054544.8	14.306	12.887	9.816	9.38	1.6	1.25	0.30	0.25	7.57 ± 0.31
ISOY_053526.73−051645.1	11.22	13.185	11.006	10.806	1.1	0.6	0.26	0.24	2.96 ± 0.62
ISOY_053529.03−050604.1	...	11.675	10.797	10.802	...	0.24	0.06	0.06	...
ISOY_053530.45−052811.3	12.139	13.72	11.575	11.179	0.46	0.22	0.12	0.10	...
ISOY_053534.41−051838.6	...	15.817	11.332	10.781	...	1.15	0.50	0.37	4.75 ± 0.33
ISOY_053535.61−053908.1	15.154	13.793	11.142	10.653	1.25	1.15	0.6	0.5	...
ISOY_053536.42−050115.5	...	10.665	8.632	8.355	...	0.69	0.15	0.14	...
ISOY_053547.65−053738.8	...	12.808	10.413	10.068	...	0.95	0.3	0.25	...
ISOY_053556.10−052533.0	...	15.613	12.045	11.727	...	0.8	0.35	0.26	5.17 ± 0.15
ISOY_053558.94−053253.9	...	14.063	11.466	10.99	...	1.05	0.45	0.42	...

Notes. ^aISOY stands for Initial Spitzer Orion YSO. ^bDip depths are mean values for sources with more than one dip.

Download table as:  ASCIITypeset image

Considering other existing data for our ONC stars, specifically the Spitzer photometry including all IRAC bands and MIPS 24 μm, Two Micron All Sky Survey (2MASS) near-infrared, and literature optical photometry and spectral types, we find that there is no clear distinction of the AA-Tau analogs in either Spitzer/IRAC or 2MASS magnitudes and colors relative to the rest of the ONC population, other than these objects are most often associated with Class II type YSOs. Only one of the 129 Class Is versus 37 of the 1175 Class IIs are identified as AA-Tau analogs.

4.2. Interpretation

The YSOVAR Orion data provide the first large-scale survey of the photometric variability of YSOs in the mid-IR. In Fall 2009, Spitzer/IRAC observed at 3.6 and 4.5 μm a 0.9 deg² area centered on the Trapezium cluster twice a day for 40 consecutive days, producing light curves for 1249 known disked YSOs and ∼820 young but generally diskless WTTs, with typical accuracies ∼3%. These data are publicly accessible through the YSOVAR Web site (http://ysovar.ipac.caltech.edu/) or in Table 2.

We find that 70% of the disked YSOs and 44% of the WTTs are variable. The WTT variations are mostly spot-like, while there is a wide range of variability types for the YSOs. Accordingly, 79% of the variable WTTs are periodic, while periods are detected for only 24% of the YSOs. In Table 4 we list 150 new periods found based on these data. We identify five new PMS eclipsing binary candidates, including one already known spectroscopic binary, Theta Ori E, and another one that is fainter than all previously known ONC PMS eclipsing binaries. Coordinates and preliminary periods for the four new eclipsing binary candidates are given in Table 5.

We have discussed in detail a sample of the Orion variables that show short duration flux dips (AA-Tau events). The mean magnitudes of these stars plus the depth of the dips and periods are given in Table 6. We interpret these events as the star being extincted by either clouds of relatively higher opacity in the disk atmosphere or geometric disk warps of relatively higher latitude which pass through the line of sight of the star. This hypothesis is consistent with both the duration and the wavelength-dependent behavior of the events.

We thank the anonymous referee for her helpful comments. This work is based on observations made with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. This work is also based (in part) on observations obtained with WIRCam, a joint project of CFHT, Taiwan, Korea, Canada, France, and the Canada–France–Hawaii Telescope (CFHT) which is operated by the National Research Council (NRC) of Canada, the Institute National des Sciences de l'Univers of the Centre National de la Recherche Scientifique of France, and the University of Hawaii.

Facility: Spitzer (IRAC) - Spitzer Space Telescope satellite