THE DISCOVERY OF NEW WARM DEBRIS DISKS AROUND F-TYPE STARS

Mid- and far-IR images were obtained using the MIPS instrument (Rieke et al. 2004) in small-field photometry mode. All four objects were measured at 24 and 70 μm, while two of them were also observed at 160 μm. In addition, low-resolution spectra (R = 70–120) were obtained for each source using the Infrared Spectrograph (IRS; Houck et al. 2004). All these observations belong to our program PID 20707 except the IRS spectrum of HD 13246, which was taken from the Spitzer Archive (PID: 241, PI: C. Chen). In order to improve the signal-to-noise ratio and search for possible variability, we performed follow-up IRS measurements for HD 13246 and HD 169666, obtained 41 and 35 months after the first epochs, respectively (PID: 50538, PI: Á. Kóspál).

MIPS: We started the data processing with the BCD files (pipeline version S16.1). Additional corrections, including flat field and a background matching at 24 μm, were performed using MOPEX (Makovoz & Marleau 2005). At 70 μm column mean subtraction and time filtering were applied following Gordon et al. (2007). BCD data were co-added with the MOPEX software. Output mosaics had pixels with sizes of 25, 4'', and 8'' at 24, 70, and 160 μm, respectively. Modified version of the IDLPHOT routines was used to detect sources on the final maps and to perform aperture photometry (including aperture correction). All of our sources were detected at 24 μm, and all but HD 53842 at 70 μm. Both 160 μm observations provided upper limits. In the case of non-detections, the positions derived from the 24 μm images were used at 70 and 160 μm for the position of the photometry. Photometric results are summarized in Table 2.

IRS: The observations were performed in standard staring mode using high-accuracy blue peak-up. The data reduction began with intermediate droopres products of the Spitzer Science Center (SSC) pipeline (version S15.3 for the earlier four spectra; and S18.02 and S18.1 for the follow-up observations of HD 169666 and HD 13246, respectively). We then used the SMART reduction package (Higdon et al. 2004), in combination with IDL routines developed for the FEPS Spitzer Science Legacy program (Meyer et al. 2006). The data reduction steps (including error budget calculation) are outlined in detail in Bouwman et al. (2008) and have been successfully used in a series of publications (e.g., Apai et al. 2005; Bouwman et al. 2006; Pascucci et al. 2008, 2009).

We compiled the spectral energy distribution of each star by combining IR fluxes from the IRS and MIPS observations with published optical and near-IR photometric data (Hipparcos, Tycho2, 2MASS catalogs). In Figure 1 we present the IR data, as well as stellar photospheric predictions obtained by fitting an ATLAS9 atmosphere model (Castelli & Kurucz 2003) to the optical and near-IR measurements. Metallicity data from Holmberg et al. (2007) and computed surface gravity values, used as inputs to the fits, are listed in Table 1. Since from distance estimates based on Hipparcos parallax data (van Leeuwen 2007) all our stars lie in the Local Bubble (Lallement et al. 2003), we assumed the extinction to be negligible. The fitted effective temperatures are included in Table 1. Predicted photospheric fluxes for the MIPS bands are listed in Table 2. All our sources exhibit significant (>3σ) excess over the predicted photosphere at 24 μm. At 70 μm HD 53842 was not detected, while the measured fluxes of the three other sources were found to be in excess.

Zoom In Zoom Out Reset image size

The IRS spectra show that the excesses are present even at shorter wavelength as they begin to depart from the expected photosphere at ∼9, 14, 14, and 8 μm in the case of HD 13246, HD 53842, HD 152598, and HD 169666, respectively.

Figure 2 presents the IRS spectrum of HD 169666. Several broad spectral features at around 11.3, 16.4, 23.7, 27.5, and 33.8 μm can be recognized. For comparison we plotted the spectrum of HD 69830 and comet Hale-Bopp, which also exhibit features at similar wavelengths (Beichman et al. 2005; Crovisier et al. 1997). These peaks, especially the 11.3 μm one, can be attributed to forsterite (Koike et al. 2003; Jäger et al. 1998), indicating the presence of micron-sized silicate grains in the disk of HD 169666. The spectra of HD 13246, HD 53842, and HD 152598 do not show any spectral features exceeding the noise. The lack of features in these disks may imply the depletion of small grains.

Zoom In Zoom Out Reset image size

We compared the two spectra of HD 169666 obtained at different epochs. Both the continuum level and the spectral features were found to be remarkably unchanged. The flux differences were typically in the order of 1%–2% over the whole spectral range. A similar comparison for HD 13246 showed more pronounced differences especially between 10 and 20 μm. However, these two measurements were processed with different pipeline versions, namely with S15.3 and S18.1. We performed an analysis of four calibration measurements of η¹ Dor processed with the same pipeline versions as HD 13246. The results indicated that the differences observed at HD 13246 could be explained by systematic calibrational effects related to the two pipeline versions. Thus, we conclude that no significant spectral change can be detected in our measurements.

We assume that the detected IR excess originates from optically thin dust confined into a circumstellar ring. The excess above the predicted photosphere was fitted by a single-temperature blackbody (as, e.g., Wyatt et al. 2005; Rhee et al. 2008). For the fitting process, the IRS spectra were robustly averaged in five bins centered at 8, 13, 18, 24, and 32 μm (the last bin was not used in the case of HD 169666 because of the presence of a relatively strong feature). For HD 13246 and HD 169666, where two IRS spectra are available, we used the earlier ones because they were obtained closer in time to the MIPS observations. An iterative method was used to compute and apply color corrections for the MIPS data (see Moór et al. 2006). The derived dust temperatures (T_dust) are listed in Table 1 and the best-fitting models are overplotted in Figure 1.

Assuming that the emitting grains act like a blackbody, we estimated the ring radius using the following formula (Backman & Paresce 1993):

$$\begin{equation} \frac{R_{\rm dust}}{{\rm AU}} = \left(\frac{L_{\rm star}}{L_{\rm sun}}\right)^{0.5} \left(\frac{278 \;K}{T_{\rm dust}}\right)^2. \end{equation} \tag{ 1 }$$

The fractional dust luminosities of the four systems were calculated as f_d = L_dust/L_bol and are in the range of 0.3–1.8 × 10⁻⁴. The derived disk properties are listed in Table 1. Note that using the second epoch spectra for HD 13246 and HD 169666, the fitted disk parameters are not altered significantly.

In all four systems the inferred radii of dust rings correspond to the region of the asteroid belt and the gas giants in our solar system. A comparison of the PR and the collision timescales for our dust rings (Equations (14) and (15) in Backman & Paresce 1993) implies that the grain evolution is dominated by collisions. The derived short lifetimes with respect to the ages of the stars demonstrate the second generation ("debris") nature of the dust grains, and indicates the presence of planetesimals co-located with the dust rings. Wyatt et al. (2007) used an analytical steady-state model to study the collisional evolution of planetesimal belts and argue that at any given age there is a maximum disk mass and fractional luminosity, since initially more massive disks consume their mass faster. We computed the maximum fractional luminosity value (f_max) for each system taking into account its properties as listed in Table 1 and using Equation (20) in Wyatt et al. (2007), also adopting their fixed model parameters (belt width: dr/r = 0.5; planetesimal strength: Q*_D = 200 J kg⁻¹; planetesimal eccentricity: e = 0.05; diameter of largest planetesimal in cascade: D_c = 2000 km). Comparing these values with the calculated fractional luminosities, we find that the disk properties of HD 13246, HD 53842, and HD 152598 are consistent with a steady-state evolutionary scenario within the uncertainties of the model. In contrast, in the case of HD 169666, a ratio of f/f_max>100 implies a high dust content around this old star, inconsistent with a steady-state model.

HD 13246 and HD 53842 belong to the Tuc-Hor association, implying an approximate age of 30 Myr. In such young stars, the disks are thought to be the sites of intense planet formation. In regions where the growing size of the largest planetesimals reach ∼2000 km, these protoplanets stir their surroundings leading to catastrophic collisions among the left-over planetesimals. A collisional cascade begins and produces large amounts of debris dust. Since the formation of giant planetesimals takes longer at larger radii, the site of debris production propagates outwards with time (Kenyon & Bromley 2002, 2004).

Using the evolutionary model of Kenyon & Bromley (2004) we calculated the radius of such a ring for an age of 30 Myr, as a function of initial disk surface density. We found that the 3.5–5 AU radii derived for our disks can be explained by disk surface density of Σ < 0.3Σ_MMSN, where Σ_MMSN is the surface density of the minimum solar mass nebula. This parameter implies a relatively low initial disk mass provided that the disk is really in this evolutionary phase. After this peak the average dust production drops, but there may be episodic spikes due to individual collisions between left-over rocky protoplanets. If the dust rings around HD 13246 and HD 53842 represent these later phases of the evolution, the initial surface density could have been higher.

HD 152598 is older than the previous two stars, and at its age of 0.21 Gyr most likely represent a later evolutionary phase, when dust grains are already generated by colliding minor bodies in a steady planetesimal ring.

HD 169666 is a new member of a distinguished, small group of FGK-type stars harboring warm debris disks of unusually high fractional luminosity. Only a handful of such systems have been identified so far: BD+20 307, HD 23514, HD 69830, HD 72905, and η Corvi (Song et al. 2005; Rhee et al. 2008; Beichman et al. 2005, 2006; Wyatt et al. 2005). The peculiar brightness of these disks is usually explained by an episodic increase in dust production (Wyatt et al. 2007). It is also a common property of HD 169666 and most other members of this group that they are old (>300 Myr) and their mid-IR spectra exhibit spectral features attributed to small silicate dust grains. Such small grains move and evolve primarily under the influence of radiation pressure and collisions.

In order to draw a picture on the dust evolution, we computed the maximum size of dust grains blown out of the HD 169666 system due to the radiation pressure. Approximating the absorption+scattering cross section by the geometric cross section and adopting a density of 2.5 g cm⁻³ and an albedo of 0.1, we found a "blowout-size" of 1.6 μm (Burns et al. 1979). Under the influence of radiation pressure alone these small grains would be removed very fast: first-order estimates of the grain removal from the mid-IR emitting region (T>130 K) suggest typical timescales of a few months to a year. With the most prominent opacity source removed so rapidly, significant changes in the mid-IR spectral features are to be expected on similarly short timescales. However, the mid-IR spectrum of HD 169666 showed no significant variations in the course of three years. The constant flux levels of the spectral features indicate that the small grains are continuously replenished on timescales of a few years or less. The existence of warm—small or large—dust in this system covers an even longer interval. Convolving our IRS spectrum with 25 μm filter profile of IRAS (Beichman et al. 1988) and comparing this synthetic flux value of 112 mJy with the IRAS measurement in 1983 (101 ± 13 mJy; Moshir et al. 1989), we could not see any variation within the measurement uncertainties. This finding suggests that the transient phase of HD 169666 lasts at least for a quarter of a century. It is an open question whether the small grain production is active also on this longer timescale. Nevertheless, the fact that most old transient systems (see above) exhibit spectral features might suggest that small grains are present, and their production mechanism is active over a significant fraction of the transient phase. Because no multi-epoch measurements exist for the other five warm disks, possible variations of the spectral features cannot be excluded. Details of the variability depend on the actual mechanism of small grain replenishment in the system.

Three possible mechanisms have been put forward to explain the origin of dust in systems like HD 169666. First, breakup of a large planetesimal similar to the event leading to the birth of the Veritas asteroid family (Nesvorný et al. 2003). The probability of such an event, however, is relatively low at an age of ∼2 Gyr (Wyatt et al. 2007). Second, dust may originate from the continuous evaporation of a super-comet as proposed for the case of HD 69830 (Beichman et al. 2005). The dust temperature in the disk of HD 169666 exceeds 110 K, the sublimation temperature of comets, consistently with this scenario. The third mechanism would be that the observed high fractional luminosity of these disks may be the result of an event similar to the Late Heavy Bombardment (Wyatt et al. 2007). Modeling the expected brightness and spectral evolution of transient disks around older stars in the three scenarios and comparing them with multi-epoch observations may help to identify the processes responsible for replenishing the observed massive dust belts in these extrasolar planetary systems.