A POSSIBLE DETECTION OF OCCULTATION BY A PROTO-PLANETARY CLUMP IN GM Cephei

The current paradigm suggests that stars are formed in dense molecular cores, and planets are formed, almost contemporaneously with the star, in circumstellar disks. The grain growth process already initiated in the parental molecular cloud continues to produce progressively larger solid bodies. Details are still lacking on how grain coagulation proceeds to eventual planet formation in a turbulent disk. Competing theories include gravitational instability (Safronov 1972; Goldreich & Ward 1973; Johansen et al. 2007) and planetesimal accretion (Weidenschilling 2000). In any case, density inhomogeneities in the young stellar disk mark the critical first step in the process. Measurements of the fraction of stars with infrared excess—arising from thermal emission by circumstellar dust—indicate a clearing timescale of optically thick disks in less than ∼10 Myr (Mamajek et al. 2004; Briceño et al. 2007; Hillenbrand 2008). Observationally, this epoch corresponds to the pre-main sequence (PMS) stellar evolution from disk-bearing classical T Tauri stars (CTTSs) to weak-lined T Tauri stars with no optically thick disks.

The open cluster Trumpler 37 (Tr 37), at a heliocentric distance of 870 pc (Contreras et al. 2002), is associated with the prominent H ii region IC 1396 and is a part of the Cepheus OB2 association. With a disk frequency of ∼39% (Mercer et al. 2009) and an age of 1–4 Myr (Marschall et al. 1990; Patel et al. 1995; Sicilia-Aguilar et al. 2005), Tr 37 serves as a good target to search for and to characterize exoplanets in formation and early evolutionary stages (see Neuhäuser et al. 2011 and references therein on Tr 37).

GM Cephei (GM Cep; R.A. = 21:38:17.3, Decl. = +57:31:23, J2000) is a solar-type variable in Tr 37. The star has a spectral type of G7 to K0, an estimated mass of 2.1 M_☉, and a radius of 3–6 R_☉ (Sicilia-Aguilar et al. 2008). The youth of GM Cep is exemplified by its emission-line spectrum, prominent infrared excess (Sicilia-Aguilar et al. 2008), and X-ray emission (Mercer et al. 2009), all characteristics of a CTTS. The star has a circumstellar disk (Mercer et al. 2009) with an accretion rate up to 10⁻⁶ M_☉ yr⁻¹, which is two to three orders higher than the median value of the CTTSs in Tr 37 (Sicilia-Aguilar et al. 2006). It is also one of the fastest rotators in the cluster, with vsin i ∼ 43.2 km s⁻¹ (Sicilia-Aguilar et al. 2008).

Most PMS objects are variables. Herbst et al. (1994) classified such variability into three categories. One class of variation is caused by the rotational modulation of cool star spots. Another class of variation arises because of the unsteady accretion onto a hot spot on the stellar surface; stars of this type are called EXors, with EX Lupi being the most extreme case. Stars with the third kind of variation, called UX Orionis-type variables or UXors, are those that experience variable obscuration by circumstellar dust clumps. About a dozen UXors have been identified so far, with some showing cyclic variability with periods ranging from 8.2 days (Bouvier et al. 2003) to 11.2 years (Grinin et al. 1998).

GM Cep is known to be an abrupt variable, but interpretations about its variability have been controversial. Sicilia-Aguilar et al. (2008) collected photometry of the star from 1952 to 2007 in the literature, supplementing with their own intensive multi-wavelength observations, and suggested GM Cep to be an EXor-type variable, i.e., with outbursts and accretion flares. Xiao et al. (2010) measured archival plates taken between 1895 and 1993, and concluded otherwise—that the variability in the century-long light curve is dominated by dips (possibly from extinction) superposed on quiescent states. If this is the case, GM Cep should be a UXor-type variable, as also claimed by Semkov & Peneva (2011).

GM Cep has been observed by the Young Exoplanet Transit Initiative (YETI) collaboration, a network of small telescopes in different longitude zones (Neuhäuser et al. 2011). In addition to the YETI data, the observations reported here also included those collected during non-YETI campaign time, by the SLT 40 cm telescope at Lulin in Taiwan, the Tenagra II 81 cm telescope in Arizona, USA, the Jena University Observatory 25 cm and 90/60 cm telescopes in Germany (Mugrauer & Berhold 2010), and the 1.5 m telescope of Moleitai Observatory in Lithuania. For the list of the YETI telescope and instrument parameters, please refer to Neuhäuser et al. (2011). While the primary goal of the YETI campaigns, each with uninterrupted monitoring of a target cluster for 7–10 days, is to search for exoplanet transit events in young open clusters—hence possibly finding the youngest exoplanets—the continuous and high-cadence observations produce data sets also valuable for a young stellar variability study that is very relevant to planet formation (Bouvier et al. 2003). Here we present the light curve of GM Cep from 2009 to 2011 which reveals T-Tauri-type flares and UXor-type variability, with the possible detection of cyclic occultation events by a dust clump in the circumstellar disk.

All the CCD images were processed by the standard procedure of bias, dark, and flat-field correction. The photometry of GM Cep was calibrated by a linear regression with the seven comparison stars listed by Xiao et al. (2010). Images taken under inferior sky conditions were excluded in the analysis. Figure 1 shows the light curves of GM Cep and one of the comparison stars observed from mid-2009 to mid-2011. The variability of GM Cep is obvious. The star experienced a sharp brightening soon after our observations started in mid-2009, prompting us to follow this star closely beyond the YETI campaigns. Our intense monitoring started in 2010. A brightness dip with a depth of ΔR ∼ 0.82 mag lasting for 39 days occurred, followed by a gradual brightening (by ∼1 mag) and fading. The falling and rising parts of the dip are roughly symmetric. In 2011, a dip also happened, but with rapid fluctuations. The star fluctuated some ΔR ∼ 1.7 mag in 2010 and also in 2009. We conclude that the sharp brightening in 2009 corresponded to the rising part of the dip seen in 2010. If so, the recurrent timescale of the dip would be 346 days, and the minimum of the dip brightened from 2009 (R ∼ 14.2 mag), to 2010 (R ∼ 13.9 mag), to 2011 (R ∼ 13.2 mag). When this trend is taken out, the gradual brightening and fading is more or less symmetric in time with the peak luminosity happening between two consecutive dips, much like the round-topped light curves seen in contact binaries. Such repeated dips plus a slow brightening and fading can be seen in the long-term light curve reported by Xiao et al. (2010), who claimed no periodicity in the data perhaps because of the sparse sampling.

Zoom In Zoom Out Reset image size

Figure 2 shows the light curves of GM Cep in the B, V, and R bands since late 2006, with additional data taken from Sicilia-Aguilar et al. (2008) and AAVSO. Analysis by the NStED (NASA/IPAC/NExScI Star and Exoplanet Database) Periodogram Service, based on the Lomb–Scargle algorithm, shows the first-ranked period to be 311 days with a broad peak in the power spectrum suggesting a quasi-periodicity as shown in Figure 3. Such a recurrence timescale of 310–320 days indeed seems to coincide with the minima in the light curve (see Figure 2), at least for the last five cycles for which sampling has been sufficiently dense (Hu et al. 2012). In addition, superimposed on the above light variations, there are sporadic flaring-like episodes with an amplitude less than 0.5 mag, each lasting for about 10 days, characteristic of T Tauri activity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While the YETI campaigns are carried out in the R band, our intensified observations of GM Cep since 2010 also included those taken in the V band. The color changes during the dip, as well as during the brightening and fading episodes, are particularly revealing. Figure 4 shows the R-band light curve and V − R color variations in 2010/2011. The dip in the beginning has a depth of about ΔV ∼ 0.68 mag, so while the star became fainter (depth in R was 0.82 mag), the V − R value decreased, i.e., its color turned bluer. During the general brightening, the star also became bluer.

Zoom In Zoom Out Reset image size

To summarize, the light curve of GM Cep is characterized by (1) a brightness dip of about 1 mag lasting for a month, with a recurrence timescale of about a year, (2) in between the dips, a gradual brightening of about 1 mag, followed by a roughly symmetric fading, and, superimposed on both (1) and (2), (3) intermittent flares ≲ 0.5 mag, each lasting for several days.

The abrupt behavior in GM Cep's light curve is not uncommon among Herbig Ae/Be stars with modulations of various timescales, i.e., "cyclic, but not exactly periodic" (Herbst & Shevchenko 1999, p. 7), superimposed on the deep minima. A flare with a blue color can be accounted for by the enhanced accretion of clumpy material. Semkov & Peneva (2011) published the B, V, R, and I light curves of GM Cep from mid-2008, i.e., one year earlier but in lower cadence than our data. Their data showed R ∼ 12 mag in 2008 with no obvious dips, an obscuration event in 2009, and another one in 2010. These authors proposed that GM Cep is a UXor variable. At the end of their observations, in early 2011, the star again reached R ∼ 12 mag, also shown in our data.

The most striking feature of the light curve of GM Cep is the month-long dips. There are various possible mechanisms for producing such a phenomenon, e.g., by star spots or a rotating accretion column, which has a typical timescale of a few hours to days. A notable case, the T Tauri star AA Tau, is known to show deep fading (∼1.4 mag) lasting for about a week and believed to be caused by occultation by a warp in the magnetospheric accretion disk (Bouvier et al. 1999) with a quasi-cyclic timescale of 8.2 days (Bouvier et al. 2003, 2007). The dip phenomenon appears to be common among young stars with inner dusty disks (Herbst & Shevchenko 1999). In a study by the CoRoT satellite of the young star cluster NGC 2264, Alencar et al. (2010) found a fraction of 30%–40% young stars exhibiting obscuration variations.

We propose that the month-long dip seen in GM Cep is a manifestation of obscuration by an orbiting dust concentration in the circumstellar disk, i.e., GM Cep is a UXor-type variable, as reported by Xiao et al. (2010) and by Semkov & Peneva (2011). If so, the orbital period of the dip gives information on the distance of the clump from the star, whereas the duration of the obscuration and amount of starlight extinction give, respectively, the size and the column density of the clump. The mass of the star is uncertain for this PMS star, but assuming 2.1 M_☉ (Sicilia-Aguilar et al. 2008), a Keplerian motion, and a period of P = 320 days, the orbital distance of the clump would be r ∼ 1.2 AU. The duration of the obscuration, t ∼ 39 days, is related to the half-size of the clump R_c by t/P = (2R_c)/(2πr); hence R_c ∼ 0.4 AU, or about 15–30 stellar radii (Sicilia-Aguilar et al. 2008).

The extinction A_λ at wavelength λ is related to the amount of obscuring dust along the line of sight, i.e., A_λ = 1.086 N_d σ_d Q_ext, where N_d is the column density of the dust grains, σ_d = πa² is the geometric cross-section of a grain of a radius of a, and Q_ext is the dimensionless extinction efficiency factor. Stars as young as GM Cep should have large grains settled into the midplane, but because the disk is inclined (Sicilia-Aguilar et al. 2008), we assume that the obscuration is caused mostly by small dust grains with an average radius of a ∼ 0.1 μm. Thus Q_ext ∼ 1, and we cautiously note the possibility of abnormal dust sizes in the disk (Sicilia-Aguilar et al. 2008, or along the line of sight, Clayton & Fitzpatrick 1987). It follows from the observed obscuration of 0.68 mag in the V band that N_d = 2.0 × 10⁹ cm⁻². This amount of intervening dust is hardly excessive. The flux drop during the dip phase, ∼1 mag, is comparable to the extinction of the star A_V ∼ 1.5 (Contreras et al. 2002; Sicilia-Aguilar et al. 2004), a value commonly seen among CTTSs. The moderate extinction also indicates a line of sight out of the disk plane. What is intriguing in GM Cep, of course, is the distinct on–off behavior of the obscuration. The column mass density is, given the same amount of extinction, proportional to the dust size a and in this case is Σ ∼ 2.9 × 10⁻⁵ g cm⁻². Even for a = 10 μm grains, the column mass density would still be several orders less than the minimum solar nebula, for which Σ is a few thousands g cm⁻² at 1 AU (Weidenschilling 1977).

It is not clear whether the clump has a line-of-sight (radial) dimension comparable to its transverse size (2R_c) or is merely a ringlet. Even if it is spherical, thus yielding the maximum mass, the mean volume density would be n_d = N_d/2R_c = 1.7 × 10⁻⁴ cm⁻³ at the clump's center. Given the proximity of the clump to the star (r = 1.2 AU), we assume the dust composition to be mostly silicates, having an average density of ρ = 3.5 g cm⁻³. This leads to an estimated mass of M_d = 2.3 × 10²¹ g for the clump, which is about that of an asteroid, if the mass is uniformly spread. For a clump this substantial in size, our line of sight does not need to line up to the orbital plane in order to detect the occultation. From the fast rotation, the infrared spectral energy distribution, and the Hα profile, an intermediate inclination angle was inferred (Sicilia-Aguilar et al. 2008). A clump extending in a radial direction would have been tidally unstable. The clump is thus extended along the orbit, but short radially.

The blueing phenomenon during the obscuration is most puzzling. It has been seen in UX Ori itself (Herbst & Shevchenko 1999) and other UXors (Grinin et al. 2001). Semkov & Peneva (2011) also reported the "color reversal" or the blueing effect in GM Cep, and attributed it to possible anomalous dust properties, or disk geometry such as self-shadowing or a piled up wall in the inner disk (Dullemond et al. 2003). One appealing proposal by Grinin et al. (1994) is that blueing happens when dust along the line of sight completely dims the star, and dust particles near the line of sight scatter preferentially blue light into view, a mechanism supported by increased polarization during maximum extinction. In GM Cep when the clump blocks out the star, either the hot boundary layer—a region between the star and the active accretion disk—or the magnetospheric accretion column must have contributed much to the emission during the dip phase.

It is interesting to note that, except for the flare events, the light curve of GM Cep, namely the repeated occultation modulated by gradual, symmetric brightening and fading, bears resemblance to that of an eclipsing binary or an exoplanet transit with phase variations (Borucki et al. 2009), though the time and flux change scales are vastly different. In GM Cep the flares are caused by enhanced accretion activity and the dip, as we propose here, by the occultation of the central star by a patch of dust in the circumstellar disk. The gradual brightening and fading, then, is the result of the orbital modulation of reflected starlight, as witnessed in high-precision light curves of eclipsing binaries or transiting exoplanets (Borucki et al. 2009). Without the shape information of the clump, it is difficult to quantify this effect, but the amount of reflected light allows us to estimate the height of the clump. If the yearly brightening trend in 2009–2011 is removed, the gradual brightening in 2010 amounted to ∼0.7 mag, meaning approximately an equal contribution between the reflected light and the direct starlight. Without knowledge of the density distribution and optical properties of dust, we made a simple analogy of dust grains as a translucent mirror, made up of a total number of N_tot particles. Assuming the Bond albedo a_B, the reflected light is (L_*/4πr²) πa²a_BN_tot, and an ensemble of dust on the back side of, and 1.2 AU away from, the star would yield N_tot ∼ 3 × 10³⁶/a_B. A rudimentary estimate, assuming an albedo of 4% (cometary nuclei), thus gives a height not much less than the perimetric dimension.

If our hypothesis that the same clump has been responsible for the yearly dip holds, the clump must be dynamically stable. The mass we derived is only for the dust, and there is no evidence, even with a sufficient amount of associated gas, that the clump is on the verge of gravitational instability (Chang & Oishi 2010). In any case, the density of the clump is not likely to have a high contrast relative to the rest of the disk. In other words, it may be just a density inhomogeneity, such as a local dust concentration in a warped, spiral-armed disk or density enhancement by a companion star (Grinin et al. 1998), that gives rise to the characteristic light curve seen in GM Cep.

In conclusion, our photometric monitoring of GM Cep confirms its UXor nature. Moreover, the light curves and color variations suggest a density inhomogeneity of dust in the young stellar disk. Such enhanced density contrast may be a signpost of the transition phase from grain growth to the onset of planetesimal formation. GM Cep may not be an isolated example, and intense monitoring should be carried out for young stars known to exhibit abrupt light variations. Further characterization of the clumpy disk of GM Cep, e.g., by polarization, infrared spectroscopy, and high angular resolution submillimeter imaging, at epochs in and out of the occultation, should shed light on our hypothesis of this interesting young star.

We thank the anonymous referee for very constructive suggestions on an earlier version of the paper. The work at NCU is financially supported in part by the grant NSC99-2119-M-008-021. Lulin Observatory is operated by the National Central University of Taiwan and is funded partially by the National Science Council and Ministry of Education of Taiwan. The Jena co-authors thank DFG, the Thuringian government, and the EU (MTKD-CT-2006-042514) for support. The Jena and Toruń co-authors thank DAAD PPP-MNiSW (50724260-2010/2011) for support. Part of the observations reported here have been obtained with telescopes at the University Observatory Jena, which is operated by the Astrophysical Institute of the Friedrich-Schiller-University.