NEAR-ULTRAVIOLET EXCESS IN SLOWLY ACCRETING T TAURI STARS: LIMITS IMPOSED BY CHROMOSPHERIC EMISSION^*

Our targets are the young stars RECX-1 and RECX-11 in the η Chamaeleontis group, the nearest open cluster at only 97 pc with an age of 5–9 Myr (Mamajek et al. 1999, 2000; Luhman & Steeghs 2004). With A_V ∼ 0, this region has little extinction (Luhman & Steeghs 2004), making it ideal for UV observations as corrections for reddening may introduce large uncertainties. The sources have similar spectral types, and were classified as K4 by Mamajek et al. (1999) and K5–K6 by Luhman & Steeghs (2004). Sources in this region have also been well characterized in the infrared with broadband J − L photometry (Lyo et al. 2004), Spitzer Infrared Array Camera (IRAC) and Multiband Imaging Photometer (MIPS) photometry, and Infrared Spectrograph (IRS) spectroscopy (Sicilia-Aguilar et al. 2009). The properties of each source are summarized in Table 1.

Observations of RECX-1 and RECX-11 were obtained between 2009 December and 2010 January with COS and Space Telescope Imaging Spectrograph (STIS) on Hubble Space Telescope (HST) in GO Program 11616 (PI: Herczeg). STIS NUV observations used the MAMA detector and the G230L grating providing spectral coverage from 1570 to 3180 Å with R ∼ 500–1000. Optical observations were completed during the same orbit as the NUV using the G430L grating which covers 2900–5700 Å with R ∼ 530–1040, resulting in almost simultaneous NUV to optical coverage with STIS.

The low-resolution STIS spectra were calibrated with custom-written IDL routines following the procedures described in the STIS data handbook. The wavelengths were calibrated from the location of identified emission lines within the spectrum, and fluxes were calibrated from spectra of WD 1337+705 in the NUV and HIP 45880 in the optical. The flux calibration also includes a wavelength-dependent aperture correction.

COS observations of RECX-11 were presented in France et al. (2011), and we discuss them here together with observations for RECX-1. The COS FUV observations were taken with the G160M and G130M gratings and were completed within 3–4 hr of the STIS observations. The combination of the two gratings provides FUV coverage from 1150 to 1775 Å with R ∼ 16, 000–18,000. The log of the observations is given in Table 2.

The COS spectra were processed by the standard CALCOS calibration pipeline. The individual spectral segments, after obtaining different wavelength settings to minimize instrumental fixed pattern effects and to provide full wavelength sampling, were combined using the IDL co-addition procedure described by Danforth et al. (2010). This procedure co-aligns the individual exposures and performs an exposure-weighted interpolation onto a common wavelength grid.

In our COS NUV acquisition image, we resolved RECX-1 into a binary with a separation of 0141 and a position angle of 21° (taking into account the fact that MIRRORB creates spurious images), indicating some relative motion between the 2004 April H-band images of Brandeker et al. (2006) and our 2010 January observations. Brandeker et al. (2006) found that the semi-major axis of the orbit was 042. The two stars are also marginally resolved in our STIS spectra, which were obtained with a slit position angle of −496. Separate spectral extractions for the components were obtained by fitting two Gaussians to the cross-dispersed profile at each wavelength position, subtracting off the emission from one component, and subsequently extracting the counts in the other component. From the blue spectrum, we estimate spectral types of K4 for the S component and K6 for the N component, consistent with the unresolved spectral type of K5–K6 (Luhman & Steeghs 2004). The near-UV spectra of the two stars are similar, except for a ∼1.5 times stronger Mg ii λ2800 flux from the S component. For our analysis, we use the combined spectra of the N and S components to increase the signal to noise. With both components having spectral types close to RECX-11 and very similar STIS spectra, little error is introduced by using the combined spectra.

Both sources were observed using Magellan Inamori Kyocera Echelle (MIKE) on the Magellan-Clay telescope at Las Campanas Observatory in Chile (Bernstein et al. 2003) on 2010 March 11, with a coverage of 4800–9000 Å and resolution R ∼ 35, 000. The data were reduced using the Image Reduction and Analysis Facility (IRAF) tasks CCDPROC, APFLATTEN, and DOECSLIT (Tody 1993).

We also obtained four low-dispersion spectra of RECX-11, covering the Hα line, between 2009 November 27 and December 22, using the Small and Medium Aperture Research Telescope System (SMARTS) 1.5 m telescope at CTIO. We used the RC spectrograph, which is a long-slit spectrograph with a Loral CCD detector; the slit subtends 5 arcmin on the sky and has a 1.5 arcsec slit width. All observations were made by SMARTS service observers and each night consisted of three observations with 5 minute integration time. The three observations are median filtered to minimize contamination by cosmic rays. The data were reduced with the SMARTS spectroscopic data reduction pipeline.¹⁷ We subtracted the bias and trimmed the overscan and flattened the image using dome flats. The spectra were extracted by fitting a Gaussian plus a linear background at each column. Uncertainties were based on counting statistics, including uncertainties in the fit background level. The wavelength calibration was based on an arc lamp spectrum obtained before each set of images.

Finally, we obtained R = 100, 000 spectra of the He i λ10830 line from RECX-11 with Phoenix on Gemini South on 2009 December 12 and from both RECX-1 and RECX-11 using CRIRES on VLT-UT1 on 2011 May 22. The observations were obtained in an ABBA nod pattern. The star η Cha, a B8 pre-main-sequence star with no detectable photospheric features at 10830 Å, was observed to provide the telluric correction. These spectra were reduced with custom-written routines in IDL. The wavelength solution is accurate to ∼1 km s⁻¹ and was calculated from a linear fit to telluric absorption lines obtained from spectra generated by ATRAN (Lord 1992) and available online.

3.1.1. Lines Produced by Accretion and Related Processes

Our spectroscopic observations, from the FUV to the IR, include a number of emission lines that trace phenomena such as accretion and outflows in T Tauri stars. Line widths of several hundred km s⁻¹ are used to identify accretors, but the most conspicuous indicators are redshifted and blueshifted absorption components on the emission line profiles.

Accessible in the optical, the Hα emission line is one of the most commonly used accretion tracers. Figure 2 shows MIKE Hα line profiles for RECX-11 and RECX-1. The Hα equivalent widths (EWs) of RECX-11 and RECX-1 are 4 Å and 1.3 Å, respectively, which would make RECX-1 a WTTS and RECX-11 a borderline CTTS/WTTS according to the standard criterion for a K5–K6 star (White & Basri 2003). The Hα profile of RECX-1 in Figure 2(a) agrees with the WTTS classification because it is symmetric and narrow. With a line width at 10% of the peak intensity of only ∼130 km s⁻¹, the line shows no indication of the high velocities that are characteristic of magnetospheric accretion. In contrast, the Hα profile of RECX-11 is typical of accreting sources, with a line width at 10% of ∼300 km s⁻¹ (White & Basri 2003). The blue emission wing is wide, extending to velocities of several hundred km s⁻¹, implying that material is accreting at nearly free-fall velocities. Inverse P Cygni absorption due to infalling material along the accretion streams is also seen (Lima et al. 2010; Muzerolle et al. 2003; Walter 1999). The line profile is consistent with model predictions for a high-inclination source (Muzerolle et al. 2001; Kurosawa et al. 2006), in agreement with the inclination of i ∼68° estimated from the values of the projected rotation velocity v sin i (Jayawardhana et al. 2006) and the rotational period (Lawson et al. 2001). Moreover, Lawson et al. (2004) compared their observed Hα profile of RECX-11 with magnetospheric accretion models and obtained a good fit assuming an inclination of i ∼70°.

Zoom In Zoom Out Reset image size

In Figure 2(b), we show our MIKE profile and the four SMARTS low-resolution Hα profiles of RECX-11 obtained within a few days before and after the HST observations. The Hα profile is highly variable, as already noted by Lawson et al. (2004) and Jayawardhana et al. (2006). The wide blue emission wing is present in all spectra, but variability is conspicuous in the red wing, sometimes showing narrow transient redshifted absorption components. This variability points to a complex geometry of the accretion streams very close to the star since redshifted absorption occurs when material in the accretion columns absorbs hot radiation produced in the accretion shock on the stellar surface (Bouvier et al. 2007; Muzerolle et al. 2001). The complex structure revealed by the line profiles of Hα, which in the case of RECX-11 is conclusive in identifying accretion, would have been missed when only considering the equivalent width of the line.

In Figure 3, we show two He i λ10830 profiles for RECX-11 obtained 18 months apart with Phoenix and CRIRES along with a CRIRES spectrum of RECX-1 for comparison. We also show the RECX-11 line profiles after subtraction of the RECX-1 line profile (with the Phoenix observation at the resolution of CRIRES) which removes any contribution by photospheric absorption lines. The He i λ10830 emission profiles show absorption components similar to those found by Edwards et al. (2006) and Fischer et al. (2008) in a large sample of CTTS, which were interpreted as arising in the wind and in the accretion columns. The earlier Phoenix profile shows a broad redshifted absorption component extending to ∼250 km s⁻¹ with several deep, narrow redshifted and blueshifted absorption lines at lower velocities, between −50 and 100 km s⁻¹. The second, CRIRES, profile also shows a complex absorption spectrum with a broad component extending from 200 to 400 km s⁻¹. It also shows a redshifted emission wing which was not observed in the Phoenix observation. The broad high-velocity absorption feature in the CRIRES spectrum occurs at velocities typical of material infalling along accretion streams. According to magnetospheric accretion models, material falling from 5R_* in the disk onto a star with the mass and radius of RECX-11 will reach velocities of ∼470 km s⁻¹, capable of producing the absorption features at the observed velocities (Calvet & Gullbring 1998).

Zoom In Zoom Out Reset image size

Fischer et al. (2008) found that sources with low accretion rates, observed to have low veiling at 1 μm (R_Y < 0.5), were more likely to have redshifted absorption in He i than sources with high accretion rates. Fischer et al. (2008) also found that in order to explain the He i λ10830 profiles absorption over a large range of velocities was necessary. They proposed that in low $\dot{M}$ sources accretion occurs in low-density, narrow accretion "streamlets" with small filling factors, covering a large velocity range, although none of their sources had low enough accretion rates to test this scenario. The narrow redshifted absorption features in the Phoenix spectrum of our very low accretor, RECX-11 (Section 4), may be observational evidence for these multiple, narrow accretion streams. Some of these narrow features, like the one at ∼100 km s⁻¹, may be related to the feature at comparable velocity seen in the MIKE Hα profile (Figure 2), although no definite conclusion can be stated since the lines were observed at different times. Simultaneous extensive coverage of Hα and He i λ10830 is necessary to understand the complexity of the accretion flows in RECX-11.

In addition to Hα and He i λ10830, the C iv doublet in the FUV spectrum can be used to separate accretors from non-accretors. The C iv λ1549 line emission has been shown to correlate with the mass accretion rate, indicating that its origin is, at least in part, in the accretion shock and flows (Johns-Krull et al. 2000; Calvet et al. 2004). In Figure 4, we compare the C iv line profiles of the accretor, RECX-11, to the non-accretor, RECX-1. The C iv line profile is much wider and stronger in the accreting source than in the WTTS in which the emission is expected to be chromospheric. The strength and width of C iv in RECX-11 are consistent with the accreting nature of RECX-11 (D. Ardila et al., in preparation).

Zoom In Zoom Out Reset image size

In the NUV range, the only significant difference between the spectra of RECX-11 and RECX-1 is in the [C ii] λ2325 line, which is much stronger in RECX-11 (Figure 1). This line has been observed in other accreting sources (Gómez de Castro & Ferro-Fontán 2005; Calvet et al. 2004), and its likely origin is the accretion shock.

3.1.2. The H₂ Lines

The FUV spectrum of RECX-11 contains a wealth of lines due to molecular species, including H₂ (France et al. 2011). Lines from H₂ in the UV are mainly due to two mechanisms. Lyα emission excites H₂, producing a fluorescent spectrum in the UV as the electrons cascade back to the ground electronic state (Herczeg et al. 2002, 2004; Yang et al. 2011). In addition, fast electrons can collisionally excite H₂, resulting in UV lines and continua (Bergin et al. 2004; Ingleby et al. 2009).

Using emission from collisionally excited H₂ measured in low-resolution FUV spectra of a large sample of both accreting and non-accreting young stars, Ingleby et al. (2009) showed that by the time accretion ends, the inner disks are depleted of gas. Our targets confirm this trend at high resolution. In Figure 5, we show the COS FUV spectrum of RECX-11 compared to RECX-1; the wavelengths at which H₂ pumped by Lyα may emit are labeled. H₂ emission lines at 1434.1 and 1445.3 Å may also be pumped by the C iii λ1174 multiplet (Herczeg et al. 2002). Figure 5 shows that the COS FUV spectrum of RECX-11 is rich in lines of molecular gas, similar to previously observed sources with confirmed accretion (Bergin et al. 2004; Herczeg et al. 2002, 2004; France et al. 2010). In contrast, molecular lines are not apparent in the FUV spectra of the WTTS RECX-1. If present, the flux of the H₂ lines in RECX-1 is <5% of the H₂ flux in RECX-11.

Zoom In Zoom Out Reset image size

Assuming that the molecular lines are formed in the atmosphere of the disk (Herczeg et al. 2004), we constructed synthetic models for an optically thin line formed between radii R₁ and R₂ in a Keplerian disk. We assumed that the emissivity depended on radius as R^γ and that the intrinsic line profile was a Gaussian with width corresponding to a temperature of 2500 K. The intensity at each radius was convolved with the rotational profile corresponding to the Keplerian velocity of the annulus, and the flux was then calculated by integrating the intensity over the range of radii. Finally, the resultant profile was convolved with the G130M line-spread function.

Figure 6 compares the predicted profile to a flux-weighted average of H₂ lines, selected from the spectral region (<1330 Å) in which all the lines were observed with the same grating, G130M. The continuum of the predicted profile has been scaled to the observed continuum. The shaded region in Figure 6 corresponds to predicted profiles for models with inner radius between R₁ = 1–20 R_* (0.01–0.1 AU), outer radius R₂ = 50–500 R_* (0.3–3.2 AU), and emissivity power-law exponent γ = −1.5. χ²_red = 3.1–3.6 for the range of models shown in Figure 6. We find that the inner radius is well constrained, within the uncertainties of the observations; small inner radii with high Keplerian velocities are needed to account for the width of the observed wings. The outer radius R₂, on the other hand, is not well constrained, since little emission comes from those regions. The power law of the emissivity is constrained by the width of the line near the peak, and we require it to fall rapidly with radius. This rapid falloff is likely due to the combination of a number of factors. For instance, the excitation will decrease as the flux of Lyα decreases with distance; similarly, the number of electrons in excited states that can absorb Lyα transitions decreases with radius, as the temperature of the upper disk layers drops (Meijerink et al. 2008). Our modeling indicates that the region of formation of the H₂ lines can be as close to the star as 0.01–0.1 AU, as was also found for the accreting brown dwarf 2M1207 (France et al. 2010). This indicates that gas likely extends close to the stellar surface, inside the dust truncation radius, and reaches the stellar magnetosphere (see Section 5.2).

Zoom In Zoom Out Reset image size

Additional tracers of accretion present in our data set, such as [O i] at 6300 and 6363 Å and the Ca ii infrared triplet, do not provide conclusive indication of accretion in RECX-11. [O i] emission, which is expected to originate in an outflow, is observed only in sources undergoing accretion (Hartigan et al. 1995). We show in Figure 7(a) the line profiles of [O i] λ6300 of RECX-11 and RECX-1. [O i] λ6300 appears in RECX-11 in excess over the WTTS spectrum of RECX-1, but the detection is <3σ, and therefore not reliable. Similarly, the Ca ii λ8542 is often used as an indicator of accretion, and even used to measure mass accretion rates (Muzerolle et al. 1998). However, the profile of this line in RECX-11, shown in Figure 7(b), is indistinguishable from the same line in RECX-1, indicating that the flux comes mostly from the chromosphere.

Zoom In Zoom Out Reset image size

The Mg ii λ2800 line luminosity also correlates with accretion luminosity in accreting stars (Calvet et al. 2004) and is seen in our STIS spectra (Figure 1). The line is observed to have the same strength in the weakly accreting star, the WTTS, and the dwarf star, indicating that it is mainly due to chromospheric activity. Moreover, its strength does not increase with the NUV emission level of the chromosphere. This agrees with the results of Cardini & Cassatella (2007), who show that at ages <0.3 Gyr, the Mg ii λ2800 fluxes are not observed to change due to saturation effects, and so the line strength does not reflect the level of chromospheric activity.

While dwarf or giant stars have been used to estimate the intrinsic stellar emission in CTTS, when measuring the UV excess WTTS are the best templates for accretion analysis because they have similar surface densities and are expected to have comparable levels of chromospheric emission. The presence of a UV excess in non-accreting pre-main-sequence stars and active stars (when compared to inactive dwarf stars) due to enhanced chromospheric activity is well established (Houdebine et al. 1996; Franchini et al. 1998). In fact, recent studies with Galaxy Evolution Explorer have relied on this excess to identify young stars in nearby star-forming regions (Rodriguez et al. 2011; Shkolnik et al. 2011; Findeisen & Hillenbrand 2010). This UV excess due to chromospheric emission affects measurements of the excess emission from which accretion luminosities are derived. Several studies have already taken these effects into account. For instance, Valenti et al. (2003) used the average spectrum of several WTTS to estimate the UV excess in low-resolution International Ultraviolet Explorer (IUE) spectra of accreting stars. Similarly, Herczeg & Hillenbrand (2008) used WTTS templates to measure flux excesses and estimate accretion rates in very low mass objects. However, many determinations of accretion luminosities and accretion rates based on UV fluxes have used dwarf standards as templates (Gullbring et al. 2000; Calvet et al. 2004). The omission of the chromospheric fluxes makes these determinations, and calibrations of line luminosities derived from them, uncertain at the low end.

In this work we have presented UV fluxes of RECX-1, which to date is the only WTTS spectrum with the signal and resolution needed for use as a chromospheric template. We compare the spectrum of the inactive K5 main-sequence star (HD 154363) with that of RECX-1, as shown in Figure 1. The flux between 2000 and 5000 Å from RECX-1 is 0.05 L_☉ brighter than that from HD 154363. If this luminosity difference were due to accretion, we would have estimated that RECX-1 is accreting at $\log \dot{M}> -8.5\,M_{\odot } \, {\rm yr^{-1}}$, which is typical of a CTTS. This is a lower limit because we have not accounted for any excess emission outside of the range of observations. This clearly demonstrates the effect of an active chromosphere on the UV spectrum and shows that inactive main-sequence standards should not be used to measure small NUV excesses. It can be argued that the chromospheric level of RECX-1 is not typical of WTTS because it is a binary. However, with a binary separation >15 AU (Köhler & Petr-Gotzens 2002; Brandeker et al. 2006) it is unlikely that the stars are interacting. In any event, a survey of NUV spectra of WTTS of different spectral types and luminosities is needed to determine the characteristic range of chromospheric NUV emission in these stars.

With evidence that RECX-11 is accreting, we use shock models to estimate upper limits to the accretion luminosity and mass accretion rate of RECX-11. A full description of the accretion shock models can be found in Calvet & Gullbring (1998); here we review the main characteristics. An accretion shock forms at the stellar surface when material from the disk falls onto the star along magnetic field lines (Section 1). The shock formed at the base of the accretion column slows the material to match the stationary photosphere, converting the kinetic energy into thermal energy, and causing the temperature to increase sharply. The shock emits soft X-rays into the pre-shock and post-shock regions and the photosphere below the shock. The heated gas emits the observed excess continuum emission which is strongest in the UV (Calvet & Gullbring 1998). The accretion shock model is used to fit the NUV excess emission and account for shock emission outside of the range of observations. The model yields the accretion luminosity, from which the mass accretion rate can be derived with knowledge of the stellar mass and radius, through $L_{{\rm acc}} = {\it GM \dot{M}/ R}$. The accretion shock model assumes that all the excess flux in the UV comes from accretion, neglecting any contribution from the outflow. Since the luminosity of the wind is <10% of the accretion luminosity for typical mass-loss rates (Cranmer 2009), this assumption is justified.

We model the spectrum of RECX-11 as the sum of the WTTS RECX-1, used as a proxy for the intrinsic stellar emission, and the accretion shock emission; our fit to the observations is shown in Figure 8 and has a χ²_red ∼ 11.5. Since the NUV fluxes of RECX-11 are essentially chromospheric, our aim is to determine the amount of shock emission that may be hidden in the spectrum of RECX-11. We find that $\dot{M}\sim 3\times 10^{-10}\,M_{\odot } \, {\rm yr^{-1}}$ is the highest value of the mass accretion rate that RECX-11 could have and still not show an excess over the chromospheric emission or veiling in the optical absorption lines.

Zoom In Zoom Out Reset image size

The upper limit on $\dot{M}$ is consistent with estimates of accretion from the emission lines shown in Figures 2 and 4. From the width of the Hα line at 10% of the maximum and the relation between the 10% width and $\dot{M}$ of Natta et al. (2004), we find $\dot{M}=2.3\times 10^{-10}\,M_{\odot } \, {\rm yr^{-1}}$. Also, from the flux in the C iv λ1549 line, we find $\dot{M}=1.6\times 10^{-10}\,M_{\odot } \, {\rm yr^{-1}}$ from the relation between C iv flux and $\dot{M}$ in Valenti et al. (2003). Lawson et al. (2004) fit their Hα line profile of RECX-11 with a magnetospheric accretion model, finding good agreement for a model with a mass accretion rate of $\dot{M}= 4 \times 10^{-11}\, M_{\odot } \, {\rm yr^{-1}}$, also consistent with our upper limit.

While it is ideal to measure L_acc from the peak of the excess emission, it is often not possible to obtain these observations in the UV. To overcome this difficulty, Gullbring et al. (1998) calibrated the excess luminosity in the U broadband, L_U, in terms of the accretion luminosity, L_acc, for a sample of stars for which they had medium resolution spectra from 3200 to 5400 Å. Accretion luminosities were measured for each star in this sample by extracting the excess flux through veiling measurements and using a slab model to account for energy outside the observed bands. The excess luminosity in U was measured by subtracting the flux in the U band of a main-sequence standard with the same V magnitude from the observed U-band fluxes. This correlation was later reproduced using the accretion shock model described in Section 4.2 (Calvet & Gullbring 1998). The L_acc versus L_U relation is commonly utilized in studies of large populations because it provides a convenient method for finding L_acc (Rigliaco et al. 2011; Ingleby et al. 2009; Grosso et al. 2007; Robberto et al. 2004; White & Ghez 2001; Rebull et al. 2000; Hartmann et al. 1998). However, as discussed above, the calibration was developed using dwarf stars as templates against which the U-band excess was measured and, as we have shown, T Tauri stars have active chromospheres which contribute to the U excess.

The UV excess observed in the WTTS RECX-1 significantly decreases the estimated excess from the accretion shock emission in the NUV. Assuming that the chromospheric emission in RECX-1 is typical for ∼K5 stars, then the intrinsic chromospheric NUV flux would be 0.05 L_☉ (Section 4.1). If this NUV excess were interpreted as an "accretion luminosity," using the Gullbring et al. (1998) relation between L_acc and U-band excess, the excess luminosity in the U band just due to chromospheric emission would be 0.01 L_☉. This suggests that for the ∼K5 spectral range, U-band excesses and derived accretion luminosities below these limits should be taken with caution. An extensive survey of the intrinsic chromospheric emission in WTTS covering a wide spectral range is necessary to quantify the expected chromospheric emission in CTTS. A re-calibration of the L_U versus L_acc relation using characteristic chromospheric contributions may extend its range of application to low values of L_acc and $\dot{M}$.

RECX-11 is in the η Chamaeleontis group, with an estimated age of 5–9 Myr. This age range is interesting because disk frequency studies indicate that ∼80% of the original disks have already dissipated by ∼5 Myr (Hernández et al. 2008). The disk around RECX-11 is then one of the few survivors and a study of its properties can give insight into the mechanisms of disk survival and dissipation.

In the age range of RECX-11, viscous disk evolution would predict a value for $\dot{M}$ between 10⁻¹⁰ M_☉ yr⁻¹ and 10⁻⁸ M_☉ yr⁻¹, extrapolating for the dispersion observed in other populations (Calvet et al. 2005). The $\dot{M}$ for RECX-11 is below this range, so additional disk dispersal mechanisms like photoevaporation may be at play. According to one type of photoevaporation model, EUV photons would ionize the disk surface; a wind is established outside a radius r_g where the gas escape velocity becomes higher than the sound speed. The predicted mass-loss rate for EUV photoevaporation is ∼10⁻¹⁰ M_☉ yr⁻¹ (Alexander & Armitage 2007). After the mass accretion rate becomes comparable to the mass-loss rate, mass does not reach the inner disk, which drains onto the star in timescales comparable to the local viscous time, ∼10⁵ years (Clarke et al. 2001). An inner disk hole is thus created, and EUV irradiation of its edge is expected to quickly erode the rest of the disk (Alexander et al. 2006). For the mass of RECX-11 (Table 1), the value of r_g (Clarke et al. 2001) would be ∼7 AU. With further refinements of the theory, the value of the critical radius can be decreased to ∼0.1–0.2 r_g (Font et al. 2004; Adams et al. 2004). This implies that if EUV photoevaporation were active with mass-loss rates greater than the low accretion rate of RECX-11, a hole of at least 0.7 AU should have developed in the RECX-11 disk.

A second type of photoevaporation model includes the effects of stellar X-ray and FUV emission and predicts higher mass-loss rates than the EUV models (Ercolano et al. 2008; Gorti & Hollenbach 2009). According to the predictions of the X-ray-driven photoevaporation models, the wind mass-loss rate scales with the X-ray luminosity. The X-ray luminosity of RECX-11 has been recently measured by XMM-Newton to be 3 × 10³⁰ erg s⁻¹ (López-Santiago et al. 2010), very similar to the value 1.6 × 10³⁰ erg s⁻¹ estimated from ROSAT observations (Mamajek et al. 2000). The disk of RECX-11 must have been subject to this influx of high-energy radiation for most of its lifetime, since the X-ray luminosity stays approximately constant in the 1–10 Myr age range (Ingleby et al. 2011). With the observed value of the X-ray luminosity, the mass-loss rate predicted by X-ray-driven photoevaporation is ∼10⁻⁸ M_☉ yr⁻¹, and the low value of the accretion rate would imply that the disk was quickly clearing its inner regions, with timescales ⩽10⁵ yr (Owen et al. 2011). So, for the X-ray photoevaporation models, the innermost disk regions should be cleared or in the process of being cleared.

To find evidence of the effects of photoevaporation on the disk of RECX-11, we have constructed the spectral energy distribution (SED) shown in Figure 9. We plot fluxes at the B, V, R, and I_C bands (Lyo et al. 2004; Lawson et al. 2001) and J, H, and K bands from the Two Micron All Sky Survey (Skrutskie et al. 2006) along with Spitzer/IRAC [3.6], [4.5], [5.8], and [8] and MIPS [24] and [70] bands from Sicilia-Aguilar et al. (2009). We also include the Spitzer IRS spectrum, observed in the SL1 and LL modes; the spectrum was downloaded from the Spitzer archive and reduced with version S18.7 of the Spitzer Science Center pipeline using the SMART data reduction package (Higdon et al. 2004) following the description in McClure et al. (2010). The SED of a K5 star, scaled at J, is shown for comparison; this SED was constructed with colors taken from Kenyon & Hartmann (1995). The median SED of Taurus (D'Alessio et al. 1999), also scaled at J, is also shown.

Zoom In Zoom Out Reset image size

The SED of RECX-11 shown in Figure 9 shows excess above the photosphere at all wavelengths >1.5 μm. Beyond the near-infrared, the fluxes are lower than the median SED of Taurus, a proxy for optically thick disks in which dust is present all the way into the dust sublimation radius. The slope of the SED between 13 and 31 μm of n_13–31 = −0.8 is steeper than the median value of −0.4 for Taurus, indicating a significant degree of dust settling in the RECX-11 disk (Furlan et al. 2006, 2009; McClure et al. 2010). In addition, the profiles of the 10 μm and 20 μm silicate features in the IRS spectrum are wider than those found in the interstellar medium (Sicilia-Aguilar et al. 2009) and contain substructure consistent with significant dust evolution, either in the form of grain growth or crystallization (Watson et al. 2009).

Despite the high degree of evolution of dust in the disk, the SED of RECX-11 does not show indications of an inner cleared region. In fact, the location of RECX-11 in the K-[8] versus K-[24] diagram corresponds to that of full disks according to the study of circumstellar disks at various stages of evolution by Ercolano et al. (2011). Moreover, the SED of RECX-11 shows a substantial near-IR excess, comparable to the median SED of Taurus, despite the low fluxes at longer wavelengths. The largest contributor to the near-IR excess in T Tauri star disks is the optically thick "wall" at the dust destruction radius, the sharp transition between the dust and gas disk (D'Alessio et al. 2006; Dullemond & Monnier 2010). For the parameters of RECX-11, the dust destruction radius is of the order of ∼0.06 AU, taking the simple expression for the equilibrium temperature, with T = 1400 K as the dust sublimation temperature and T_eff = 4350 K for a K5 star (Dullemond & Monnier 2010).

The evidence then seems to rule out inner clearing in the disk of RECX-11. First, it is accreting, so mass has to reach the magnetospheric radius, 3–5 stellar radii, ∼0.03 AU. Substantial H₂ emission is observed, coming from gas regions as close in as 0.01–0.1 AU to the central star, typical of all accretors. Moreover, dust extends inward to ∼0.06 AU. These inner radii, where disk material is present, are not consistent with those predicted for photoevaporation models, according to which the innermost <0.1 AU of the disk should be clear. It could be argued that the disk of RECX-11 has been caught in the short phase when the inner disk is still accreting its remaining mass onto the star. However, the recent analysis of the disk of another star in the η Cha group, ET Cha (ECHA J0843.3-7915 = RECX-15), casts some doubt on this suggestion (Woitke et al. 2011). The disk of ET Cha is also accreting (Lawson et al. 2004); it is very evolved with very low dust and gas masses, but still has a very substantial near-IR excess consistent with emission from the dust destruction radius (Woitke et al. 2011). For a 5–9 Myr population of less than 20 stars (Mamajek et al. 2000) the probability of finding two stars in a phase lasting only ∼10⁵ years is low. Overall, the evidence provided by low accretors in evolved populations suggests that current models of disk dissipation by photoevaporation with high mass-loss rates may need to be revisited. Our upper limit on $\dot{M}$ does not, however, rule out photoevaporation at the low mass-loss rates predicted by the EUV photoevaporation models.