MULTI-WAVELENGTH PHOTOMETRY OF THE T TAURI BINARY V582 Mon (KH 15D): A NEW EPOCH OF OCCULTATIONS

This season's unfolded VRIJHK light curves (see Figure 1) include over four cycles and are characterized by: (1) a maximum that generally becomes progressively more bright and cusped, (2) a sharp drop-off, followed by a shallower, nearly linear decline in brightness during ingress, (3) a short minimum accompanied by a small brightness reversal, and (4) a linear increase, followed by a much steeper rise in brightness during egress. The shape of the last maximum is dramatically different from previous maxima (see Figure 1). The last full cycle contains a well-defined but asymmetric peak that occurs before phase = 0.5, followed by a nearly linear decline in magnitude across all bands. We discuss these aberrations further in Section 4. Except for the anomalous last cycle, the steady increase in the system's maximum brightness is consistent with trends predicted by the precessing ring model. As the screen reveals more of the stellar orbit, more of the stellar photosphere of star B is exposed.

Zoom In Zoom Out Reset image size

To better illuminate the periodic features, we phase-fold the light curves to the binary period of 48.37 d in Figure 2. Data from previous SMARTS seasons (2011/2012, red triangles; 2010/2011, blue squares) are added for comparison. The small brightness reversals reaching a few tenths of a magnitude are now more readily apparent, especially in V, I, and J; in the other filters, the minima do not have coherent shape but are dominated by large scatter. We note that the peak of central rebrightenings is offset from phase = 0 and that the rise to brightness reversal is shallower than the decline. As described earlier, ingress and egress are characterized by two regimes: steep, more curved profiles closer to maxima and shallow, linear slopes closer to minima. Discontinuities in slope, or points of inflection (POI) occur near I ≈ 17 mag, where the two regimes intersect (see Section 4). Since we first began the SMARTS campaign in 2010, the POI has shifted in phase by ∼0.05 yr⁻¹, consistent with the steady advancement of an occulting edge.

Zoom In Zoom Out Reset image size

We plot the long-term Cousins I-band behavior of KH 15D in Figure 3. The peak brightness of the system had been steadily decreasing from 2006 (epoch of star A) until it reached a "faint state" in late 2009 (epoch of complete obscuration). The system then brightened suddenly in late 2011 (epoch of star B), and the steady increase in maxima exhibits some symmetry with the decline in peak brightness seen in earlier data. For the 2012/2013 observing season, the system's maximum brightness never exceeded I = 14.8 mag, which is ∼0.3 mag dimmer than the maximum out-of-eclipse flux when the slightly later-type and presumably cooler, fainter K7 star was the source being eclipsed.

Zoom In Zoom Out Reset image size

From historical archives we know that the combined light of stars A and B amounted to I = 13.57 ± 0.03 mag (Winn et al. 2006). Modern (CCD) photometry informs us that star A has an unobscured brightness of I = 14.47 ± 0.04 mag, implying that star B has I = 14.19 ± 0.06 mag. This signifies that star B's photosphere is still significantly obscured by some (projected) portion of the CB ring. Based on the behavior of the upper envelope on the long-term I-band light curve, we expect star B to reach I = 14.4–14.5 mag next observing season and that it will not rise completely above the ring horizon until around 2015 February.

Figure 4 shows the color variations of KH 15D as a function of phase for the most recent observing season. Notice the flatness of color near maxima (phase = 0.5) when the star is (partially) unobscured near apastron. Blue "shoulders" are also present at phases 0.3 and 0.7 in the NIR colors, and the amplitude of bluing increases from V − J to V − K. Moreover, the system becomes redder by >0.5 mag in every color near mid-eclipse (phases 0.9–1.0 and 0.0–0.1). The difference in V − H color between in and out of eclipse reaches 1.0 mag. We discuss possible sources for the slight bluing and dramatic reddening in Section 4.

Zoom In Zoom Out Reset image size

Figures 5 and 6 show the system's colors as a function of brightness in magnitudes, where 2003–2005 JHK data are from Kusakabe et al. (2005) and 2002–2009 VI observations are from Hamilton et al. (2005) and Herbst et al. (2010). In general, the color–magnitude behavior of KH 15D may be divided into three regimes: (1) bright phase where continuum emission from the stellar photosphere dominates (I < 17 mag), (2) intermediate phase where the stellar disk is 100% covered (17 < I < 18 mag), and (3) faint phase where the system reddens sharply and ubiquitously (I > 18 mag). The colors in regime 1 are relatively constant, especially in V − I, where the standard deviation in bright phase color is a few hundredths of a magnitude. See Table 4 for a list of bright phase median colors for each epoch where data are available. The color behavior is more complex at intermediate and faint phases. The large scatter in color over the years, despite its chaotic appearance, has a simple explanation, which is presented in Section 4. In I − J and I − H, the colors seem to initially become slightly bluer near I ≈ 17 mag, consistent with the blue shoulders described earlier. Beyond I = 17.2 mag, the colors redden slightly and/or exhibit scatter, until finally, they redden significantly to levels beyond their bright phase values. The degree of reddening at the faintest phases suggests the presence of a cooler third body (see Section 4).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The long-term V − I color–magnitude diagram (see Figure 5) clearly demonstrates two distinct bright phase colors corresponding to star A (2002–2009) and star B (2011–2013). This confirms that star B is continuing to rise and set above the occulting horizon. Although star B remains partially covered at maxima, its bright phase V − I is quite stable. We measure a median V − I color of 1.15 ± 0.02 mag and 1.18 ± 0.03 for 2012/2013 and 2011/2012, respectively. Note that the previously published V − I value of star B (1.28 mag; Capelo et al. 2012) had not been transformed to the standard system. When the color term is taken into consideration, the system's bright phase V − I is closer to the expected color for the spectroscopically identified K1 star. The difference in color between star A (2002–2007) and star B (2012/2013) is Δ(V − I) = 1.55–1.15 = 0.4 mag (see Table 4), which most closely corresponds to the color difference between K7V and K4V (Bessell 1990; Briceño et al. 2007) or that between K7V and K1III (Bessell & Brett 1988).

It is evident from the long-term V − I diagram that during recent years (2011–2013), the system at intermediate and faint phases (I > 17 mag) tends to become redder than star B's median photospheric color. By contrast, when star A was undergoing regular occultations, the system exhibited significant bluing as it became fainter. The amount of bluing, Δ(V − I) = 0.3–0.4 mag, corresponds exactly to the color of star B's photosphere as revealed in recent data. We therefore interpret the bluing apparent in 2002–2009 observations as due to scattered light from (the then fully occulted) star B, the hotter and bluer of the pair. Conversely, we attribute the redness in V − I during 2011–2013 eclipses as due to contaminant light from the now fully occulted star A. The fact that the system color during 2009–2011, when both stars were completely obscured, is bound by the blueness and redness set by star B and star A, respectively, also supports our argument (note the upper and lower envelopes in Figure 5). Indeed, this indicates that forward scattering and selective reddening due to ring particles do not contribute significantly to the large-scale V − I behavior. Rather, the observed trends may be understood as a mixture of direct and neutrally scattered light from two stars of somewhat different color.

It is quite possible that during epochs when star A or star B is partially revealed, the system's bright phase color may also be somewhat affected by the other (fully occulted) star's scattered light. Indeed, the data (see Figure 5 and Table 4) demonstrate a slightly bluer bright phase color in 2007–2009 when star A was becoming increasingly more covered near apastron than 2002–2007 when star A was fully revealed. If the mixture of direct and scattered light plays a small but significant role even during bright phase, then the 2012/2013 bright phase V − I = 1.15 mag is only a lower limit on star B's photospheric color. That is, the star would be slightly bluer and would more closely resemble a K1V/K2V star.

The abrupt linear decline in magnitude from maximum during the last full cycle of this season's observations is intriguing and unprecedented. If the decrease in brightness were due to cold spots on star B, a common feature in weak-lined TTSs, then the spot (or collection of spots) would have to cover 50%–60% of the 50% revealed stellar disk. While this value is consistent with typical covering factors (10%–40%; Herbst et al. 1994), the effect of dark spots on the stellar continuum should decrease with increasing wavelength, given that the temperatures of spots are typically ∼750 K lower than photospheric values (Bouvier & Bertout 1989). The data, however, show remarkably similar linearity in the steady magnitude decline (6.1 ± 0.1 mag period⁻¹, $\chi _r^2$ = 1.7) across all filters V through K. Although this particular feature is inconsistent with rotational modulation of spots, dark blemishes on star B are not unexpected, given the spottedness of its (presumably coeval) companion (Hamilton et al. 2005; Herbst et al. 2010).

Perhaps the trailing edge of the screen is not as straight as the leading edge, although the saw-like top of the last cycle indicates an edge just as sharp. Based on the wavelength independence of the early onset of flux decline, a more favorable explanation may be the presence of a local protuberance of material of the occulting edge. This gray attenuation of flux is consistent with the behavior of the system immediately after complete photospheric obscuration, i.e., at the POI. The anomalies and the normal occultations have similar color behavior, suggesting a common cause.

Figure 7 shows the system's change in magnitude from median maxima values during 2003–2005 when star A is fully revealed, 2011/2012 when star B is ∼40% revealed, and 2012/2013 when star B is ∼50% revealed. The POIs (dotted lines) mark the transition between two distinct regimes of light behavior. We interpret the POIs as second and third contact when star B is at the cusp of 100% coverage for the following reasons. First, although the POIs steadily shift in phase, they occur at nearly the same brightness level, I ∼ 17 mag (see Figure 2 and Figure 4 of Herbst et al. 2010; note that for the latter plot, 2005–2010 SMARTS data were not transformed to Cousins I, and so a 0.1–0.2 mag offset may be expected). Second, the median maximum brightness between 2009–2011, when scattered light from the fully obscured binary dominated the system, is consistent with I = 17 mag. Finally, at light levels dimmer than I = 17 mag, system colors deviate significantly from photospheric values, which suggest intermediate and faint phase light is different from bright phase light (see Figure 5).

Zoom In Zoom Out Reset image size

Interestingly, the depth of eclipse from maximum to POI is approximately wavelength independent in the optical and near-infrared (see Figure 7), i.e., light attenuation is gray. While the larger Δmag value in K in 2012/2013 is enticing, it is not a significant detection since the POI is near the S/N limit at K ∼ 16 mag. Neither gray scattering nor gray transmission is compatible with the behavior of interstellar dust, which is typically sub-micron in size. That the attenuation of light for both star A and star B is gray out to ∼K indicates that the line of sight material is larger than the size of the wavelength of observation, i.e., grain radii >1 μm (Draine 2011). Additionally, the particles in the occulting ring must be similar in size at least on angular scales of the projected binary orbit (which is physically ∼0.5 AU across), since star A probes a different line of sight from that of star B.

The color data indicate that when viewing the limb of star B at egress or ingress, the system appears quite blue in many of the colors, especially in V − H and V − K (see Figure 4). The effect may be somewhat asymmetric, appearing stronger during egress. The bluing trend is opposite to what one would expect from limb darkening in an ordinary K star atmosphere. While we have no reason to doubt the integrity of the data at these phases, we also have no simple explanation for them. Perhaps the effects of external heating on star B's atmosphere (from accretion, chromospheric and coronal radiation, and star A) lead to a temperature inversion that manifests itself at short wavelengths as limb brightening. Perhaps forward scattered light plays a role. Or, perhaps the spotted surface of star B is such that we just happen to be seeing cooler (spot) material toward the center and hotter material toward the limb. Additional seasons of monitoring during which a full range of colors is observed should shed light on which of these explanations, if any, is correct.

The long-term O/NIR color plots (see Figure 5) compare the color behavior between the epochs of star A and star B. It is apparent that the colors redden dramatically at the faintest phases (beyond I ∼ 18.3 mag), especially during the most recent season (2012/2013) in I − J and I − H. We have established earlier that the reddening seen in 2011–2013 V − I at faint phases is likely due to the redness of the now invisible star A, complemented by the bluing seen in 2002–2009, when scattered light from the bluer but fully obscured star B dominated the system at faint phases. Here, we refer to the severe reddening—beyond the photospheric values of either star—seen at the faintest phases. At I ∼ 18.8 mag, I − J and I − H colors reach +1.5 and +2.3 mag, respectively, which correspond to color excesses of E(I − J) = 0.6 mag and E(I − H) = 0.7 mag. These reddening slopes for KH 15D are much steeper than one would expect for interstellar reddening—more than twice the values (Draine 2011). We also observe that the 2012/2013 data show a marginally redder (0.05 ± 0.03) bright phase J − H color (see Figure 6) than both 2011/2012 and 2003–05 data, when star B was ∼20% more covered and the slightly later-type star A was being eclipsed, respectively. The simplest explanation may be that we are detecting excess light, especially in J and H, when the overall system is very faint.

The existence of a shepherding body massive enough to constrain, but not disrupt, the ring (such as a young gas giant) is well motivated. Physical models of KH 15D necessitate the presence of a truncation mechanism at ∼5–10 AU to maintain the finite width, warp, and precession of the CB ring. A young super-Jupiter orbiting in (or near) the plane of the central binary should leave observable signatures such as excess near-infrared radiation possibly regulated by the orbital period of the planet.

According to "hot start" models, a ∼1 Myr old 10 M_J planet would have an absolute M_H = +8 mag (Fortney et al. 2008; Spiegel & Burrows 2012). Given the distance to NGC 2264, such an object will have H ∼ 17 mag and inject an amount of flux consistent with the potential H excess observed during bright phase. Thermal radiation from a young super-Jupiter is furthermore consistent with the observed degree of reddening at the faintest phases. Finally, the 2003–2005 data show little or no hint of reddening. While this is due to incomplete I-band coverage in Figure 5, the J − H versus J plot (see Figure 6) clearly demonstrates that, despite similar JH ranges, the system does not redden beyond the bright phase colors of star A in 2003–2005. This phenomenon may be explained by the fact that if there were a third body, then every few years or so the planet would be on the far side of the ring and its radiation signatures would be completely obscured. We emphasize, however, that testing this ring confinement theory requires spectroscopic, i.e., radial velocity, confirmation.

In recent years, a new mechanism has been proposed for ring formation in dusty disks with sufficient gas (Klahr & Lin 2005; Besla & Wu 2007; Lyra & Kuchner 2013), without requiring shepherding planets. Our data does not preclude this photoelectric instability as a possibility for ring confinement. This formation scenario occurs when dust heats gas in the circumstellar or CB disk via photoelectric heating, creating local pressure maxima which concentrate the solid particles. Lyra & Kuchner (2013) modeled this clumping instability and successfully produced narrow, eccentric ring structures similar to those observed in debris disks around Formalhaut and HD 61005. Although a gas-to-dust ratio of order unity is enough for this confinement mechanism to occur, the gas-to-dust ratio in the KH 15D disk is currently not well constrained (Lawler et al. 2010), and thus we do not know whether there is sufficient gas for this instability. While the near-IR excess we find during minimum light might be explained by radiation from grains composing the ring, the inferred temperatures of 1000–2000 K seem too high for a ring located 1–5 AU from the central stars. Near-infrared spectral data and longer wavelength flux measurements would be useful in evaluating this alternative model.