The Mysterious Dimmings of the T Tauri Star V1334 Tau

With the primary goal of discovering transiting planets around bright host stars ($8\lt V\lt 11$), the KELT project is a ground-based photometric survey covering >70% of the entire sky on a 10–20 minute cadence (Pepper et al. 2007, 2012). The project consists of two telescopes, KELT-North in Sonoita, AZ, and KELT-South in Sutherland, South Africa. Each telescope has a Mamiya 645-series wide-angle lens with a 42 mm aperture and a 80 mm focal length (f/1.9), and observes in a broad R-band filter. The telescopes have a 26° × 26° field-of-view (FOV), and a 23'' pixel scale. V1334 Tau is located in KELT-North field 03 (field center at α = 3^h58^m12^s, δ = 31° 39'5616). The KELT-North telescope observed V1334 Tau from UT 2006 October 26 to UT 2013 March 13, obtaining 9186 observations after processing. For a detailed description of the KELT data acquisition and reduction process, see Siverd et al. (2012). The median per-point error is 0.005 mag.

Designed to discover nearby supernovae, the All-Sky Automated Survey for SuperNovae (ASAS-SN) is photometrically monitoring the observable sky every two days down to V ∼ 17 (Shappee et al. 2014). The survey has two sites, Cerro Tololo InterAmerican Observatory (CTIO) in Chile and Mount Haleakala in Hawaii. Each site is hosted by the Las Cumbres Observatory Global Telescope Network (Brown et al. 2013) and has two units with four 14 cm Nikon telephoto lenses and a 2K × 2K thinned CCD, with a $4.5\times 4.5$ degree FOV and a $7\buildrel{\prime\prime}\over{.} 8$ pixel scale. V1334 Tau was observed 532 times from UT 2012 January 20 until UT 2016 October 14. The median per-point error is 0.004 mag.

The Wide Angle Search for Planets (SuperWASP) is a high-cadence photometric survey designed to detect transiting exoplanets. Observing in a broad filter (centered at 550 nm), the survey has two observing locations, one at the Roque de los Muchachos Observatory on La Palma (WASP-North) and the other located at the South African Astronomical Observatory (WASP-South). With eight cameras at each location, each using a 200 mm f/1.8 lens and a 2048 × 2048 pixel CCD, SuperWASP is able to monitor large portions of the sky at a very high cadence (minutes). Each camera has a 78 × 78 FOV, and a $13\buildrel{\prime\prime}\over{.} 7$ pixel scale. V1334 Tau was observed in two separate seasons in the publicly released observations: UT 2004 august 15 to UT 2004 September 30 and UT 2006 September 11 until UT 2007 February 15.²⁵ All observations from WASP prior to 2006 are unfiltered, and after 2006 the observations are in a broad V-band filter. A 3σ clipping around the median was performed to remove large outliers. V1334 Tau was also observed from UT 2009 February 08 until UT 2012 January 29. Since the data reduction of the public SuperWASP data was performed differently than the more recent WASP observations, so the two data sets were kept separate throughout the analysis described here. Additionally, we binned the second SuperWASP data set in 1 hr bins and calculated the standard deviation of each bin. All data points within bins which had a standard deviation greater than 2% were then discarded. After processing and cleaning, 14,339 observations remained and were used in this work. The median per-point error is 0.009 mag for the public observations and 0.003 mag for the recent observations.

Designed to discover and catalog variable stars over the entire sky, the All-Sky Automated Survey (ASAS) observes all stars brighter than V = 14. For a detailed description of the observing strategy and data reduction, see Pojmanski (1997). ASAS uses two locations, Las Campanas, Chile and Haleakala, Maui. Each location observed in the V and I bands simultaneously using two wide-field Minolta 200/2.8 APOG telephoto lenses with a 2K × 2K Apogee CCD. Each telescope has a 88 × 88 FOV. ASAS observed V1334 Tau in the V-band from UT 2002 December 13 until UT 2009 February 25. 93 observations were obtained over this period. The median per-point error is 0.034 mag.

The MINERVA Project has four 0.7 m PlaneWave CDK-700 telescopes located on Mt. Hopkins, AZ, at the Fred L. Whipple Observatory, capable of both millimag photometry and precision radial velocity measurements (Swift et al. 2015). On the night of UT 2016 October 21, two telescopes simultaneously observed V1334 Tau using an Andor iKON-L 2048 × 2048 detector with a 209 × 209 FOV and a plate scale of $0\buildrel{\prime\prime}\over{.} 6$ pixel⁻¹. One telescope observed in the V-band with 200 exposures and achieved a per-point rms scatter of 0.0009 mag. The other observed in the B band with 140 exposures and achieved a per-point rms scatter of 0.0015 mag.

Using the 0.61 m Ritchey–Chrétien optical telescope at the KUO in Kutztown, Pennsylvania, V1334 Tau was observed in the B and V-bands on UT 2016 October 26, UT 2016 November 05 and 07. The observing setup uses a 3072 × 2048 CCD that has a $19\buildrel{\,\prime}\over{.} 5\times 13\buildrel{\,\prime}\over{.} 0$ FOV with a 038 pixel⁻¹. Standard aperture photometry was used, and the instrumental flux was color-corrected using several Landolt standard fields. In total, KUO obtained 160 B-band and 186 V-band observations. The median per point a per-point rms is 0.0007 mag in the B-band and 0.0006 V-band.

We obtained spectroscopic observations of V1334 Tau with the Tillinghast Reflector Echelle Spectrograph (TRES) on the 1.5 m telescope at the Fred Lawrence Whipple Observatory, on Mount Hopkins, Arizona, USA from UT 2016 October to December. We obtained nineteen 10 minute exposures over the span of the two months, with fourteen observations occurring on UT 2016 December 09 and 10. The spectra were obtained at a resolution of $\lambda /{\rm{\Delta }}\lambda \equiv R=44000$ over the wavelength range of 3900–9100 Å. The spectra are consistent with a ∼5500 K dwarf and our estimate from Section 4, rotationally broadened by $77.6,\mathrm{km}\,{{\rm{s}}}^{-1}$. We estimate the absolute radial velocity of the V1334 Tau system to be ∼19 km s⁻¹ which is in agreement for the known absoluted radial velocity of the Taurus–Auriga association (15 km s⁻¹, Bertout & Genova 2006). The TRES observations show no large (>1 km s⁻¹) radial velocity changes.

We searched for periodic variability in the high-cadence KELT data prior to and during the 2009 dimming using the Lomb–Scargle (LS) periodicity search algorithm (Lomb 1976; Scargle 1982) in the VARTOOLS analysis package (Hartman & Bakos 2016). For the each season of KELT data, we find a 0.3214 day period with an average peak-to-peak amplitude of ∼2% (see Figure 4). This periodic signal is clearly changing in amplitude and phase over the 9 years of observations. We find the same ∼0.32 day periodicity in every season except the 2008–2009 observing season where the source faded. Additionally, the targeted observations by MINERVA and KUO are consistent with this periodicity. There is no color evolution over the short duration of these targeted observations. During the 2009 dimming, the ∼0.32 day periodicity is recovered but at a lower signal-to-noise ratio. The periodicity is likely caused by processes near the star. If the 0.32 day periodicity is the rotation period of the star, then this would imply a rotational velocity of $\sim 347\pm 57$ km s⁻¹ for a stellar radius of ${2.619}_{-0.304}^{+0.597}$ $\,{R}_{\odot }$, compared to a break up period of 0.39 ± 0.13 days. Given our uncertainty in the stellar radius, V1334 Tau A could be rotating near break up velocity. We explore the possibility that V1334 Tau is a rapidly rotating star close to a pole-on orientation in Section 6. Another possibility is that the ∼0.32 day period is the orbital period of some circumstellar material that periodically obscures the star at a semimajor axis of ∼2.3R${}_{\star }$.

Zoom In Zoom Out Reset image size

However, the observed periodicity is persistent over the entire KELT data set, making it unlikely to be related to accretion processes. Additionally, if there is a significant amount of orbiting disk material right at the surface of the star, we would expect to see an IR excess in the SED analysis. The 14 spectra of V1334 Tau on UT 2016 December 09 and 10 show no significant large radial velocity variability ruling out the possibility of a close stellar binary. The orbital semi-amplitude from the TRES observations is 338 ± 56 m s⁻¹, when a photometric ephemeris with a period 0.3124 days is imposed. The velocity residuals from a circular orbit have an rms of 433 m s⁻¹. The rms of all the RV measurements is 1.3 km s⁻¹. Using our determined stellar mass of 1.67 $\,{M}_{\odot }$ and a period of 0.3214 days, we get an upper mass limit of a companion to be <2 $\,{M}_{{\rm{J}}}$ (rms = 433 m s⁻¹) or <6.2 $\,{M}_{{\rm{J}}}$ (rms = 1.3 km s⁻¹). The upcoming K2 campaign 13 observations of V1334 Tau should provide additional information on the nature of this short-period variability.

In UT 2002 December, V1334 Tau was in its "minimum" state, at a V ∼ 9.7 mag. The system took ∼280 days to return to its quiescent brightness of ∼9.6 mag. Unfortunately, there are no available photometric observations of V1334 Tau prior to UT 2002 December, so there is no constraint on the dimming duration. We only know that the dimming must be longer than the time between the start of the ASAS observations and the end of the observed brightening—i.e., greater than ∼320 days.

During late 2009 February, the V1334 Tau system began to fade from its median magnitude of V ∼ 9.59 mag. We estimate that the ingress of the dimming takes ∼273 days and that the event has a maximum depth of ∼0.12 mag (V ∼ 9.71, the maximum depth occurred during the 2012 and 2016 observing seasons). This ingress timescale is very similar to the estimated timescale to brighten in 2002–2003. The dimming lasts through the end of the data analyzed here on UT 2016 November 07, leading to a lower limit on the dimming duration of >2802 days. If we assume that the two dimming events are similar, the time between events is >13 years. Multi-band observations from MINERVA and KUO show no B − V color trend over the course of an ∼6 hr observing night.

With the combined high-cadence observations from KELT and WASP during the 2009 dimming, we are able to study the photometric variability during the event. If this is indeed an occultation, the additional brightening and dimming events seen during the large dimming (on a similar timescale as the initial dimming) suggests that we are observing sub-structure in the occulting body. After the ingress, the system stays at a constant brightness of ∼9.66 mag through the remainder of the 2009–2010 observing season and the first half of the 2010–2011 season. At JD ∼ 2455502, V1334 Tau begins to dim again, down to a brightness of ∼9.69 mag. At the start of the 2011–2012 observing season, V1334 Tau is near its minimum brightness of ∼9.71 mag. Since the end of this additional dimming occurred during the 2011 seasonal observing gap, we can only place an upper limit of ∼325 days on its duration. After V1334 Tau brightens to ∼9.65 in 2012, it remains at a depth of ∼9.69 mag for the following season. Following the 2013/2014 season, V1334 Tau brightens over the course of the entire season from ∼9.69 mag to ∼9.64 mag. Following this peak in brightness, the following two seasons of ASAS-SN, KUO, and MINERVA data suggest that V1334 Tau faded back to the maximum depth of ∼9.71 mag. Interestingly, the timescales of these variations are similar to the estimated ingress timescale in 2009. However, our ability to draw any conclusions is limited by the seasonal observing gaps. We discuss the possibility that the observed structure of the dimming may be caused by a change in the opacity of the occulting feature in Section 7.

To check for photospheric features on the stellar disk, we derived line broadening profiles using a least-squares deconvolution (LSD) of the TRES spectra (following Zhou et al. 2016). Indentations and protrusions in the broadening profile map the presence of spots, faculae, and pulsation on the stellar surface. The broadening profiles show large-scale variations over the course of hours. Figure 5 shows the time series of the spectra gathered over 2016 December 09 and 10, phased to the 0.3214 day photometric modulation period (see Section 5.1). We find that the broadening profiles do not phase well with the photometric period, suggesting that there is no persistent spot crossing the star, and that the variability is possibly due to non-radial pulsations on the star. In addition, we find no evidence of line blending indicative of a binary stellar companion orbiting the primary star.

Zoom In Zoom Out Reset image size

We also find a double-peaked Hα emission line, which may be an indication of an accretion disk (See Figure 5). The Hα feature has an equivalent width of ∼5 Å, which could be taken to mean that the star still possesses an inner accretion disk (Mohanty et al. 2005). The relatively high equivalent width also suggests that V1334 Tau should be reclassified as a Classical T Tauri Star (CTTS) (>3 Å, White & Basri 2003). Given our mass estimates, the peak Hα emission corresponds to a distance of >0.16 au from the star, assuming we are observing a Keplerian velocity for a possible inner disk. However, it is also possible that the Hα feature could be from cooler, neutral H i that is falling onto the star or disk winds. No other Balmer lines are found in emission, nor do we find other common accretion signatures, such as the He I 5876 Å and He i 6670 Å lines in emission. We see the Ca ii H and K lines with core emission, and equivalent widths of 1.5 Å and 2.6 Å respectively, perhaps indicative of the strong chromospheric activity expected for T Tauri stars. Lithium absorption at 6708 Å is also present, with an equivalent width of 0.25 Å, consistent with the absence of depletion expected for a solar-type star with an age of 2 Myr (Soderblom et al. 1993). The sodium Na I D feature has a depth and width consistent with a stellar absorption line, along with a superimposed narrow feature due to the interstellar medium. Unlike RW Aur (Facchini et al. 2016), we do not see a time varying P Cygni profile in the Na lines, which would suggest the existence of a strong disk wind. However, we do not have a spectroscopic observation of V1334 Tau prior to the large dimming, limiting our interpretation.

To determine the key properties of the occulting feature, we model the observed dimmings as an occultation of the host star by a large body in Keplerian motion. The leading edge of the occulting body can be perpendicular to its direction of motion (θ = 90, sharp edge) or can be inclined ("wedge-shaped," see Figure 9 from Rodriguez et al. 2015). The estimated ingress time of the 2009 dimming, where we have the best photometric coverage, is ∼273 days. The ingress timescale gives us an estimate of the occulting body's transverse velocity, ${V}_{T}\sim 2{R}_{\star }/({T}_{\mathrm{ingress}}\sin \theta )$. For ${{\rm{R}}}_{\star }$ = $2.62\,{R}_{\odot }$, we get a minimum velocity of ∼155.1 m s⁻¹. The full duration of the 2009 dimming is >2803 days. Using our estimated transverse velocity and the duration timescale, we estimate that the occulting body must be >0.25 au in width. If we assume that the occulting body is in Keplerian motion, the ingress timescale also provides an estimate of the orbital semimajor axis (see Equation (1) from Rodriguez et al. 2016c). However, using these estimated parameters, the occulting body would be unrealistically far from the host star (at ∼62,000 au), ruling this interpretation out.

Unfortunately, there is no measurement of the size of the circumstellar disk around V1334 Tau A. We know from Kohler & Leinert (1998) and Daemgen et al. (2015) that there are two nearby companions to V1334 Tau A: V1334 Tau B (1631) and C (0106). Therefore, we can constrain the occulting feature to be within the projected separation of the B component, $1\buildrel{\prime\prime}\over{.} 631$ (215 au for d = 132 pc). Assuming the occulting feature is at the projected separation of V1334 Tau B, it would require a transverse velocity of ∼2.6 km s⁻¹ and be ∼4.2 au in width. Also, constraining the occulter to 219 au would suggest a wedge-angle of <34. From our analysis of the TRES spectra, we estimate that the peak emission of the double-peaked Hα emission corresponds to a distance of >0.16 au. Assuming the occulting feature is associated with an inner disk and at a semimajor axis of 0.16 au would imply that it is moving at ∼98 km s⁻¹ and have a leading edge angle of 006. In this scenario, the duration of the occultation requires multiple orbits (≥160) of the same material across our line-of-sight to cause the dimming event that began in 2009. This could be explained if the entire inner disk has precessed into our line-of-sight, but this does not seem plausible. We can also place the occulting body at the distance of the C component (14 au for d = 132 pc). This would suggest a transverse velocity of ∼10.3 km s⁻¹, a ∼16.7 au wide occulter, and a leading edge angle of 086. The width would be ∼19% of the entire orbit at 14 au, assuming Keplerian motion. At 14 au, an orbital period would be ∼40 years. It is possible that the close-in nearby companion is associated with the observed dimming events.

Fortunately, we observed two separate dimmings, one pre-2004 and one beginning in 2009. Since the egress from the pre-2004 dimming and the ingress of the 2009 dimming are similar in duration, it is plausible that we are seeing a periodic occultation with a period of >13 years. For a stellar mass of 1.67 $\,{M}_{\odot }$, this period would suggest that the occulting feature is located $\geqslant 6.6\,\mathrm{au}$ from the host star. Using the same wedge-shaped model, the leading edge of the occulting body would have a highly inclined leading edge angle of ∼06, almost parallel to its direction of motion. This would also suggest that the occulter is moving at ∼15 km s⁻¹ and be $\geqslant 24\,\mathrm{au}$ in width, which is >59% of the entire orbit.

Observations from the Atacama Large Millimeter/submillimeter Array (ALMA) and the Hubble Space Telescope (HST) have shown that transitional disks can have large asymmetries (Pérez et al. 2014; Stark et al. 2014). Specifically, HST observations of MP Mus have shown that the eastern side of its disk is three times brighter than its western side (Schneider et al. 2014). The young star Oph IRS 48 has a large asymmetric dust structure that extends across almost one third of the entire disk ring (van der Marel et al. 2013). Therefore, if V1334 Tau has an edge-on, asymmetric disk, the large dimming could be caused by a dust trap in the disk crossing our line-of-sight, leading to a periodic dimming. Future millimeter mapping of the V1334 Tau system should determine the plausibility of this interpretation. However, no detectable IR excess out to 20 μm would suggest a very low mass, nearly transparent circumstellar disk, making this interpretation unlikely.

An alternate explanation for the UXor phenomena is hydrodynamical fluctuations of the inner disk rim. Dullemond et al. (2003) determined that the inner disk rim would have a scale height of ∼0.2 R_rim and fluctuate by ±∼0.1 R_rim. Using 0.16 au as an estimate of the inner disk radius, this would suggest a maximum fluctuation of the inner rim height to be ∼0.048 au. In addition to the hydrodynamical fluctuation, the inner disk of a YSO can be warped and perturbed by a misaligned close companion (Facchini et al. 2013, 2014). Without a constraint on the inner disk inclination, we are unable to test the feasibility of this interpretation. However, this scenario has been proposed as the cause of the observed dimming events of RW Aur (Facchini et al. 2016).

In any case, the variability during the dimming suggests an opacity change in the occulting feature. Another interpretation for the extended dimming is that a screen of hot dust can be created in the inner disk and pushed across our line-of-sight by high disk winds (Petrov et al. 2015). However, we find no evidence for high or enhanced disk winds in the TRES spectra, making this scenario unlikely. Future millimeter mapping of the disk around V1334 Tau and radial velocity monitoring for a short-period companion will better constrain these scenarios.

Another possible explanation of the dimming events seen in V1334 Tau is they are a result of significant increase in the magnetic field flux at the surface of the star. According to the MIST stellar evolution models, the interior structure of V1334 Tau is predicted to be fully convective. Current magneto-hydrodynamic stellar structure models of fully convective dwarfs suggest that these stars are able to support stable (>1000 days), large-scale dipolar fields (>1 kilogauss) despite rotating rapidly, resulting in the formation of large groupings of dark polar starspots blanketing a significant fraction of the stellar photosphere (Yadav et al. 2015). The formation of long-lived polar spots may result in multi-year dimming events similar to the ones seen here, particularly if we are viewing this star nearly pole-on. The high amount of Hα emission and the excess NUV and FUV GALEX flux observed in the star's SED suggest that V1334 Tau is magnetically active. Future observations should clarify if the dimming we observe is connected to magnetic activity. This includes stellar color information in and out of dimming events, mapping the stellar magnetic fields using Zeeman–Doppler imaging, and modeling the upcoming K2 high-precision light curves.

There are only a dozen or so objects known to show this type of phenomenon. The duration of the V1334 Tau dimming events strongly resembles the long-term dimming of the archetype AA Tau (Bouvier et al. 2013). Although similar in duration, the dimming event in AA Tau shows a much larger depth of ∼2 mag. Other systems like V409 Tau, DM Ori, and RW Aur have shown similar dimmings (again at much larger depths), and are interpreted as either a periodically obscuring feature in the circumstellar disk, high disk winds, or inner disk instabilities (Rodriguez et al. 2013, 2015, 2016b, 2016c; Petrov et al. 2015; Facchini et al. 2016). It is possible that the dimming mechanisms for all these systems are related.