HERSCHEL OBSERVATIONS AND UPDATED SPECTRAL ENERGY DISTRIBUTIONS OF FIVE SUNLIKE STARS WITH DEBRIS DISKS

Observations from the Herschel Space Observatory have revealed a population of extrasolar debris disks with blackbody temperatures <60 K and peak flux densities at wavelengths near 100 μm. Photodetector Array Camera and Spectrograph (PACS; Pilbratt et al. 2010) 100 and 160 μm observations of the DUst around NEarby Stars (DUNES) Sunlike targets reveal a debris disk incidence of 20%, up from 12% at the shorter Spitzer MIPS wavelengths (Eiroa et al. 2013). Far-IR observations are, therefore, opening up a new discovery space of disks more than 100 times more luminous than the Edgeworth–Kuiper Belt, but too cold to detect at 70 μm.

We now know of over 30 disks with radii R > 100 AU (Eiroa et al. 2010; Morales et al. 2013; Duchêne et al. 2014; Pawellek et al. 2014). Since even wide disks require active collisional cascades to replenish 1–10 μm grains (Burns et al. 1979), planetesimal formation must be a robust process even at very large distances from the star. Planet formation at 125–250 AU may even have proceeded as far as super-Earth mass (Kenyon & Bromley 2015). The wide dust disks revealed by Herschel and earlier (sub)millimeter/scattered light observations (e.g., Greaves et al. 1998; Ardila et al. 2004; Krist et al. 2005, 2010; Wyatt et al. 2005; Hines et al. 2007; Kalas et al. 2007a, 2007b; Liseau et al. 2008; Golimowski et al. 2011) may be remnants of planetary systems, which orbit Sunlike stars but are nevertheless very different than our own: dynamical evidence indicates that the solar nebula was likely truncated between 80 and 100 AU due to photoevaporation or tidal interactions with nearby young stars (Kretke et al. 2012; Anderson et al. 2013). Here, we present Herschel observations of five Sunlike stars with debris disks, which transition from Kuiper Belt size to R > 150 AU. We use the Debris Disk Simulator (DDS) (Wolf & Hillenbrand 2005) to construct model spectral energy distributions (SEDs) and compare our disk radii, grain sizes, and luminosities with other debris disks orbiting Sunlike stars.

We begin with a discussion of target selection, observations, and data reduction (Section 2). Next, we discuss the extended structure observed in our images: our targets include one well resolved disk and three moderately extended sources (Section 3). We then present the new Herschel and Spitzer photometry in Section 4. Radiative transfer models of the disk SEDs follow in Section 5, with conclusions, and possible future work in Section 6.

We selected five Sunlike stars with debris disks discovered in Spitzer observations for Herschel followup. All targets have infrared excess emission seen in both Spitzer IRS observations at 32 μm (Dodson-Robinson et al. 2011; Chen et al. 2014) and in Spitzer MIPS observations at 70 μm (Beichman et al. 2006; Bryden et al. 2009). (HD 105211 also has an MIPS 24 μm excess, which was not realized at the time of publication (Beichman et al. 2006); see Section 5.1 for more detail.) The observing strategy was designed to ensure detection in all Herschel PACS bandpasses. Based on the two detections at different wavelengths, Dodson-Robinson et al. (2011) fit single-temperature blackbody models to the Spitzer excess emission, which we extended to the Herschel PACS bandpasses. Exposure times were selected such that each target would be detected at the 3σ level, where σ was the quadrature sum of the instrument noise and the predicted confusion noise reported in the Herschel Observers' Manual.⁶ (In reality, some of the sources have brighter 160 μm backgrounds than predicted, leading to detections with low significance or nondetections.)

Each observation consists of PACS medium-speed cross-scans separated in position angle by 40°. Program OT1_sdodsonr_1 includes simultaneous 100 μm/160 μm observations of all targets as well as simultaneous 70 μm/160 μm observations of HD 50554 and HD 105211. Observations are scan maps with eight scan legs of length 3' and a cross-scan step of 4''. We also obtained 70 μm/160 μm observations of HD 50554, HD 52265, and HD 202206 from the Herschel Science Archive (program OT1_amoromar_1, PI A. Moro-Martín). Data reduction, beginning with the Level 1 data cubes, was performed using the Herschel Interactive Processing Environment version 12.1.0 (HIPE; Ott 2010). We used the HIPE implementation of the JScanam map-making algorithm (Graciá-Carpio et al. 2015) to combine scans and cross-scans into oversampled mosaics of 1'' pixel⁻¹ at 70 and 100 μm and 2'' per pixel at 160 μm. The JScanam map-maker removes the 1/f signal-drift noise without subtracting extended emission, which was an important consideration, because some of our sources are not pointlike (see Section 3). However, JScanam also leaves point sources intact, and so can be used for any type of target. Table 1 lists the targets in our sample and their Herschel Science Archive observation IDs. The table also includes luminosity and age estimates from the literature (McDonald et al. 2012; Chen et al. 2014; Bonfanti et al. 2015), and luminosity estimates from this work.⁷ Figures 1–5 show the final mosaics in each band for all targets.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Determining the spatial extent of the disks is important for selecting the correct aperture sizes for photometry. Examining the 70 and 100 μm mosaics in Figures 1–5 suggests that the disks are not point sources, but instead have some extended structure. Figures 1, 2, 3, and 5 contain curves of growth for HD 33636, HD 50554, HD 52265, and HD 202206, showing encircled energy fraction (EEF) as a function of circular aperture radius for the 100 μm (green) and, where available, 70 μm (blue) mosaics. For comparison, the dashed lines show curves of growth from the point source Vesta (Lutz 2010). For HD 33636, HD 50554, and HD 52265, the curves of growth are well below those of the Vesta PSF, indicating that the sources have extended structure. HD 50554 is elongated at both 70 and 100 μm, but has a small apparent position angle shift between the two bands. Fitting an elliptical Gaussian function to the source, we find a position angle (east of north) of 54° at 70 μm and 51° at 100 μm. The 3° difference is too small to conclusively indicate morphology changes between dust emitting at different wavelengths, but should be kept in mind as a possible source of uncertainty in the photometry and SED fitting. HD 202206 is less well resolved than the other sources, but is still more extended than a point source. For some targets, the curves of growth turn downward instead of monotonically increasing with aperture size, indicating that the aperture has expanded to include pixels with negative flux densities. The photometric aperture must be kept smaller than the curve-of-growth turnover radius. HD 105211 (Figure 4) is clearly resolved in all PACS bands. The extended nature of our targets leads us use larger photometric apertures than the faint-source sizes used by the DUNES team (Eiroa et al. 2013). We use 12'' as our nominal aperture size at 70 and 100 μm (see Section 4), which captures the extended emission, yet avoids negative-valued pixels.

The upper-right panel of Figure 4 shows the deconvolved 70 μm image of HD 105211. Examination of a point source located in the 70° cross-scan, shown in the inset plots of the original image (upper left) and the deconvolved image, reveals the extent to which deconvolution has suppressed the trefoil PSF structure. Although our deconvolution procedure does not perfectly conserve flux, one can still see that most of the emission in the deconvolved image is concentrated in a narrow band, indicating a nearly edge-on disk. We performed an elliptical Gaussian source fit on the original (not deconvolved) 70 μm image and found ${\sigma }_{x}=6\buildrel{\prime\prime}\over{.} 8$ and σ_y = 39, where the x-direction is east–west and the y-direction is north–south. The position angle is 297 east of north. Assuming a circular debris ring, the radius of the 70 μm emitting ring is approximately $\sqrt{{\sigma }_{x}^{2}+{\sigma }_{y}^{2}}=7\buildrel{\prime\prime}\over{.} 9$, or 154 AU, in rough agreement with the SED fitting results (see Section 5). The inner edge of the ring is not resolved, possibly due to the high inclination of the disk.

For all targets except for HD 105211, our aperture photometry is based on the PACS ipipe script L3_pointSourceAperturePhotometry.py ⁸ , which uses the HIPE tasks annularSkyAperturePhotometry and photApertureCorrectionPointSource. For HD 33636, HD 50554, and HD 202206, the aperture sizes are 12'' at 70 and 100 μm, with background subtraction based on sky annuli of 25''–35''. For HD 52265, we use an 8'' aperture and an 18''–22'' sky annulus at both 70 and 100 μm to exclude a bright source west of the target. At 160 μm, where confusion noise is most severe, we use a 14'' aperture and decrease the sky annulus to 24''–28'' for HD 33636 and HD 202206. For HD 50554 at 160 μm we use an aperture of 10'' and a sky annulus of 20''–24'' in order to avoid a nearby source. HD 52265 is undetected at 160 μm due to confusion. For the extended source HD 105211, we use the Aperture Photometry Tool (Laher et al. 2012). We choose an elliptical aperture of semimajor axis 18'' and semiminor axis 12'' at an angle of 30° east of north. The sky annulus has major axis 30''–45'' and minor axis 20''–30'' at 70 and 100 μm, and major axis 30''–44'' and minor axis 20''–293 at 160 μm. The elliptical aperture corrections are based on Table 2 of Balog et al. (2014), which reports EEF as a function of circular aperture radius for each PACS filter. For an aperture with semimajor axis a and semiminor axis b, the multiplicative aperture correction A is

$$\begin{eqnarray}&&A=\displaystyle \frac{1}{\sqrt{\mathrm{EEF}(a)\times \mathrm{EEF}(b)}}.\end{eqnarray} \tag{ 1 }$$

For a = 18'' and b = 12'', we find aperture corrections of 1.201 at 70 μm, 1.241 at 100 μm, and 1.384 at 160 μm.

Uncertainties have several components: Poisson noise from the source, sky background, PACS absolute flux calibration, and residual 1/f noise, which may elude the map-making algorithm. Since the pixels in each final mosaic are not native, but instead are reconstructed from multiple scans of the source, they are not independent and the method of estimating random noise as ${\sigma }_{\mathrm{sky}}\sqrt{{N}_{\mathrm{pix}}}$ (where σ is the standard deviation of the sky flux and N_pix is the number of pixels in the aperture) does not work. Instead, the usual procedure for calculating the random component of the PACS-image error budget is to place apertures of the same size and shape used for the science target on the sky background and calculate the standard deviation of the aperture-corrected flux from each. Following Balog et al. (2014), we use six apertures placed evenly around the outer edge of the background sky annulus for all targets except HD 105211. The HD 105211 sky background is structured, with an extended bright source directly to the west of the disk (Figure 4; see especially the 160 μm image at the lower right). To estimate photometric error for HD 105211, we use 12 elliptical apertures placed randomly on the image, excluding the bright source. Finally, we include conservative estimates of the systematic errors—both from absolute flux calibration and repeatability—of 7% of the source flux (Balog et al. 2014). Quoted uncertainties are the quadrature sum of the random and systematic errors. Table 2 lists the measured PACS flux densities.

As a consistency check, we compare our PACS 70 μm flux densities with Spitzer MIPS 70 μm observations. Published values come from the catalog of Chen et al. (2014), which includes photometry from a variety of literature sources. We also present new photometry of both the 24 and 70 μm images based on the MIPS team in-house pipeline (Gordon et al. 2005; Su et al. 2010).⁹ MIPS did not observe HD 202206 at 24 μm. We use PSF extraction to find 24 μm flux densities for the remaining targets. All targets have MIPS 70 μm data. We use PSF extraction to measure 70 μm flux densities for all targets except HD 105211, for which we use an aperture with radius 35'' and a sky annulus of 39''–65''.

Table 2 lists the MIPS flux densities for each target. The two MIPS 24 μm analysis pipelines yield flux densities that are consistent at the 1–2σ level, although there is a clear discrepancy in how the error bars are computed for HD 105211. The MIPS 70 μm flux densities from the two pipelines are fully consistent for each source except HD 105211. The low HD 105211 70 μm flux density of 474 mJy reported by Beichman et al. (2006) is probably due to their choice to use a uniform aperture radius of 1.5 camera pixels (78) for the entire survey. Since the spatial extent of the HD 105211 disk in our PACS 70 μm images is $\gt 14^{\prime\prime}$ on the major axis, Beichman et al. (2006) may have underestimated the flux density. The remaining PACS and MIPS 70 μm fluxes are consistent at the 1σ level, indicating good agreement between the Spitzer and Herschel fluxes.

As a final test of our photometry pipeline, we measure fluxes from the PACS calibrator star α Ceti (HD 18884). We use scan/cross-scan AOR pairs 1342212853/1342212854 (100/160 μm; observing day 614) and 1342203030/1342203031 (70/160 μm; observing day 457). The scan/cross-scan pairs were processed with the same JScanam map-making script as the science observations. We then performed photometry with the nominal aperture and sky annuli used for HD 33636 and HD 202206 (our compact, uncrowded targets). Our results match those of the instrument team (Table 12, Balog et al. 2014) to within 1% at all wavelengths, demonstrating the high accuracy of our data pipeline for pointlike sources with clean backgrounds (see Section 5 for caveats about the HD 105211 observations).

In the next section, we combine Spitzer and Herschel observations to extend SED coverage to the peak wavelengths of dust emission, using a radiative transfer model to derive dust properties.

The disks in this study are cold enough so that observations at 70 μm and shorter wavelengths probe only the increasing side of the F_ν(λ) curve. The Herschel photometry extends coverage to the peak and the decreasing side of F_ν(λ). We use the DDS¹⁰ (Wolf 2003; Wolf & Hillenbrand 2005) to fit SED models to the Spitzer IRS/MIPS and Herschel PACS photometry. The DDS takes as input a stellar spectrum, an inner and outer radius for the dust ring, a dust mass, an analytical or numerical description of the dust number density n(r) (where r is the distance from the star), and a grain size distribution of the form $n(a)\propto {a}^{-q};{a}_{\min }\leqslant a\leqslant {a}_{\max }$ (a is the grain radius and we set q = 3.5). The assumption of a power-law grain size distribution is reasonable for the larger grains, but dynamical models that include loss processes such as Poynting–Robertson drag predict that for grains with a ≲ 10 μm, the size distribution turns over so that the grain abundance increases with radius (Wyatt et al. 2011; Kenyon & Bromley 2016). Since the far-infrared-emitting dust grains are most likely part of a collisional cascade that includes large grains, we set a_max = 1 mm for all models. Larger particles would be present, but the grain size distribution is so steep that the larger grains would contribute only ${({a}_{\max }/1\mu {\rm{m}})}^{-1/2}\,=3 \%$ of the total thermal emission. The DDS also allows the user to set the chemical composition of the grains; here we assume 100% astronomical silicate (Laor & Draine 1993). Since we do not explore the composition options, our model SEDs are plausible but not unique. In particular, the grains could have organic components or ice mantles, although ice would eventually succumb to photodesorption. However, the small number of spectral data points that cover the bulk of the disks' emission do not warrant extremely flexible fits.

The parameters we explore are minimum grain size a_min, the debris ring's distance from the star R, the width of the debris ring σ (assuming a Gaussian number-density profile centered at R), and the mass of grains smaller than a_max = 1 mm. Input star spectra are Atlas9 model photospheres (Kurucz 1992) originally fitted by Dodson-Robinson et al. (2011), based on Hipparcos/Tycho BT and VT magnitudes (Perryman et al. 1997; Høg et al. 2000) and 2MASS JHK magnitudes (Cutri et al. 2003). Where available, we added RI photometry from Bessell (1990). We also checked for consistency between the Dodson-Robinson et al. (2011) model photospheres and the allWISE photometry (Wright et al. 2010; Cutri et al. 2013) and found good agreement. The allWISE photometry was particularly important for HD 105211, which has conflicting photosphere models in the literature due to poor-quality 2MASS photometry; and for HD 52265, which has a poor-quality 2MASS J magnitude. Table 3 lists the fitted parameters for each SED, and Figures 6–10 show the model SEDs. We begin our discussion of SEDs with a detailed look at the photosphere model and mid-IR photometry of HD 105211, which requires extra care because of the imprecise 2MASS data, and then move on to the ensemble properties of our disks.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3.  Dust Ring Parameters from SED Fits and Images

Star	RSED (AU)a	σ (AU)b	amin (μm)c	${M}_{\mathrm{dust}}({M}_{\odot })$ d	TBB (K)e	RBB (AU)f	Γg	${L}_{\mathrm{dust}}/{L}_{* }$ h
HD 33636	85	12	3.0	10−8	46	37	2.3	5.4 × 10−5
HD 50554	45	4	4.5	3 × 10−9	59	27	1.7	4.1 × 10−5
HD 52265	70	10	3.0	3 × 10−9	58	32	2.2	2.0 × 10−5
HD 105211	175	20	4.1	6.6 × 10−8	49	86	2.0	5.4 × 10−5
HD 202206	105	20	3.0	4 × 10−8	45	40	2.7	1.3 × 10−4

Notes.

^aDistance of the center of the dust ring from the star, computed from SED. ^bHalf-width of debris ring, assuming Gaussian number-density profile for dust. ^cMinimum grain size. ^dDust mass. ^eTemperature of dust grains, assuming blackbody emission. ^fRadius of disk of blackbody grains. ^gR/R_BB. ^hRatio of dust luminosity to star luminosity.

Download table as:  ASCIITypeset image

5.1. HD 105211 Mid-IR Photometry and Photosphere

5.2. Dust Properties

We now turn our attention to the properties of the dust in our systems, including the disk radius, minimum grain size, location of dust relative to the parent planetesimal population, and possibility of planet sculpting. First, we find that all of the dust SEDs peak between 80 and 120 μm, which is typical of debris disks from the DUNES program (Krivov et al. 2013). We do not find any sources which are brighter at 160 μm than at 100 μm, a feature which could indicate either a background galaxy in the beam (Gáspár & Rieke 2014) or a disk of unstirred, primordial grains (Krivov et al. 2013). We diagnose the presence of superheated small grains by computing the parameter ${\rm{\Gamma }}=R/{R}_{\mathrm{BB}}$, the ratio of the disk radius from our SED models (in which dust temperature depends on grain size) to the disk radius if only blackbody grains were present:

$$\begin{eqnarray}&&{T}_{\mathrm{BB}}=5100\,{\rm{K}}(1\,\mu {\rm{m}}/{\lambda }_{\max })\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{R}_{\mathrm{BB}}={\left(\displaystyle \frac{278{\rm{K}}}{{T}_{\mathrm{BB}}}\right)}^{2}{\left(\displaystyle \frac{{L}_{* }}{{L}_{\odot }}\right)}^{1/2}\,\mathrm{au}.\end{eqnarray} \tag{ 3 }$$

In Equation (3), T_BB is the temperature of blackbody grains, λ_max is the wavelength of maximum emission, and R_BB is the radius of the ring of blackbody dust. The disks in our sample have 1.7 < Γ < 2.7 (Table 3). Although four out of five of our disks are not well enough resolved to measure disk sizes directly from the images, our SED-based values of Γ agree well with studies of resolved debris disks (e.g., Rodriguez & Zuckerman 2012; Booth et al. 2013). Each disk must contain a population of small grains that do not emit efficiently at $\sim 100\,\mu {\rm{m}}$ wavelengths, and so heat to above the local blackbody temperature. Given the presence of small grains, our sources are most likely self-stirred or planet-stirred planetesimal belts with grain-producing collisional cascades, which is consistent with our assumption of $q=3.5$ (Pan & Schlichting 2012; Matthews et al. 2014, p. 521).

Dodson-Robinson et al. (2011) estimated blackbody temperatures T_BB for our target disks based on Spitzer IRS and MIPS (spectro)photometry at 32 and 70 μm. For the warmer debris systems HD 50554, HD 52265, and 105211, their estimates agree well with ours. For HD 33636 and HD 202206, the peak flux density is much redder than 70 μm and the Herschel observations indicate colder blackbody temperatures than the Spitzer-based photometry. We emphasize that the SEDs modeled here are not based on blackbody grains—our comparison with single-temperature blackbody models simply highlights the importance of far-IR photometry in characterizing the debris rings.

The Spitzer IRS spectra provide the best constraint on minimum grain size. In the Rayleigh regime, where 2πa ≪ λ, the grain opacity κ declines with wavelength as $\kappa \propto {\lambda }^{-\beta }$, where β = 2 for simple conductors and insulators (Draine 2006). At 32 μm, the center of the spectral window used by Dodson-Robinson et al. (2011) to detect infrared excesses, 2πa = λ for λ = 5.1 μm. We find that the lack of Spitzer IRS fractional excesses at 20 μm (with the exception of HD 105211) and the small, though detectable, excesses at 32 μm exclude grains smaller than ∼3–4 μm, as such grains would emit efficiently from 20–30 μm. The minimum grain sizes predicted by our best-fit model SEDs are in good agreement with the Pawellek & Krivov (2015) SED-based measurements from 32 objects (see their Figure 4): all of their target stars with ${L}_{\odot }\lt {L}_{* }\leqslant 10{L}_{\odot }$ have ∼2 μm < a_min < ∼10 μm.

There is, of course a degeneracy between R and a_min. As an example, one model of the HD 50554 debris disk has R = 40 AU and ${a}_{\min }=5.5\,\mu {\rm{m}}$, while another has R = 45 AU and ${a}_{\min }=4.5\,\mu {\rm{m}};$ the difference between their reduced χ² statistics is only 0.2. Yet the value of R from the SED of the resolved source HD 105211 is in rough agreement with the measured size from the image (175 AU versus 154 AU), suggesting that we have broken the degeneracy for at least one source. Indeed, we find it difficult to reproduce the measured 32/70 μm flux ratios without setting a minimum grain size larger than the radiation-pressure blowout size of (Burns et al. 1979)

$$\begin{eqnarray}&&{a}_{\mathrm{rp}}\geqslant \displaystyle \frac{6{L}_{* }\langle {Q}_{\mathrm{pr}}(a)\rangle }{16\pi {{GM}}_{* }c\rho },\end{eqnarray} \tag{ 4 }$$

where L_* and M_* are the star luminosity and mass, $\langle {Q}_{\mathrm{pr}}(a)\rangle$ is the radiation-pressure coupling coefficient averaged over all frequencies in the stellar spectrum, and ρ is the particle density (2.2 g cm⁻³ for astronomical silicate). In the geometric optics limit where $\langle Q(a)\rangle =1$ for all values of a, we find blowout-size limits of 0.5–2.2 μm for the range of star luminosities and spectral types in our sample. Intriguingly, Pawellek et al. (2014) and Pawellek & Krivov (2015) show that ${a}_{\min }/{a}_{\mathrm{rp}}$, the ratio of the true minimum grain size to the minimum predicted value from Equation (4) (now with $\langle {Q}_{\mathrm{pr}}(a)\rangle$ calculated self-consistently from grain optical properties), is a decreasing function of stellar luminosity, approaching unity for A-type stars but with larger values for solar-type stars. Our results agree well with theirs: the most luminous star in our sample, HD 105211 with L_*/L_⊙ > 7, has ${a}_{\min }/{a}_{\mathrm{rp}}\sim 2$, while our least luminous star, HD 33636 with L_*/L_⊙ = 1.05, has a_min/a_rp ∼ 6.

We suggest that our Herschel observations most likely trace dust near the "birth ring" (Strubbe & Chiang 2006), where the source planetesimals reside. The largest grains that can emit like blackbodies at 160 μm have radii set by 2πa ≈ λ, such that a ≈ 25 μm. The lifetime τ of a grain against Poynting–Robertson drag is

$$\begin{eqnarray}&&\tau =\displaystyle \frac{16\pi {c}^{2}\rho {R}^{2}a}{3{L}_{* }},\end{eqnarray} \tag{ 5 }$$

where c is the speed of light. For 25 μm grains, we find $0.2\,\mathrm{Gyr}\leqslant \tau \leqslant 2\,\mathrm{Gyr}$ for all of the disks in our sample. The quantity star age/τ for 25 μm grains is between 0.7 and 6.5 for all systems except one; HD 50554 has age estimates which differ by a factor of 10 (3.3 Gyr from Bonfanti et al. (2015) versus 0.33 Gyr from Chen et al. 2014), and so star age/τ is either 1.6 or 16. Since the grains that can contribute significantly to the 160 μm emission have lifetimes against radiative drag of at least 1/4 the star age, they are likely to be destroyed by collisions before migrating significantly. Similarly, all systems have β < 0.05 (where $\beta ={F}_{\mathrm{pr}}/{F}_{g};$ F_pr is the radiation-pressure force given $\langle {Q}_{\mathrm{pr}}(a)\rangle =1$ and F_g is the gravitational force) for 25 μm grains, so their orbits are not substantially modified by radiation pressure. The dust ring widths from our SED fits are likely close to the true widths of the source planetesimal belts, with two caveats: (a) mutual grain collisions will increase the eccentricity dispersion and, therefore, the ring width (Thébault 2009), and (b) our models do not account for the possibility that the small grains have a wider spatial distribution than the large grains—we assume all grain sizes are well mixed throughout the dust annulus. In reality, radiation pressure might push the smallest grains to wide or eccentric orbits (Burns et al. 1979; Gáspár et al. 2012) and create structures similar to the β Pictoris, HR 8799, and HD 95086 "halos" seen in scattered light (Augereau et al. 2001; Su et al. 2009, 2015; Ballering et al. 2016). Pushing the small grains in the HD 105211 system to wider orbits might help resolve the conflict between the 32/70 μm flux ratio and the 160 μm flux density discussed in Section 5.1. The HD 105211 disk also has the smallest a_min/a_rp size ratio of any disk in the sample, and so is most likely to be affected by radiation pressure. Followup with ground-based adaptive optics or the Hubble Space Telescope could detect scattered light from blowout-size grains. If our Herschel photometry traces the birth rings, our systems can be added to the list of >30 debris disks with radii larger than the probable solar nebula radius of 80 AU (Kretke et al. 2012). Planetesimal formation seems to be a robust process even out to extremely large distances (Kenyon & Bromley 2013).

Finally, many lines of evidence suggest that at least some planetesimal belts are sculpted by planets (e.g., Quillen & Thorndike 2002; Chiang et al. 2009; Boley et al. 2012; Su & Rieke 2014, p. 318; Sai et al. 2015). Interestingly, the disk surrounding HD 50554 is a true Kuiper Belt analog in its spatial distribution, with a best-fit radius of 45 AU and ring half-width of 4 AU. Since HD 50554 is slightly younger than the Sun (3.3 Gyr; Bonfanti et al. 2015), its disk may be similar to the Kuiper Belt at an earlier stage of evolution and may have a Neptune-like planet at its inner edge. In N-body simulations of planetesimal belts with giant planets orbiting just inside their inner edges, Rodigas et al. (2014) find that the width of the planetesimal belt increases with both planet mass and planet/disk semimajor axes. Here, we find that the width of the best-fit dust ring roughly increases with distance from the star, from σ = 4 AU at R = 45 AU for HD 50554 to σ = 20 AU for both HD 202206 (R = 105 AU) and HD 105211 (R = 175 AU), although the width of the ring becomes less observationally constrained as R increases. However, we have no direct constraints on the presence or absence of planets on wide orbits, and we measure fractional dust luminosities of ${L}_{\mathrm{dust}}/{L}_{* }={10}^{-5}-{10}^{-4}$ (Table 3), which are consistent with dust production from icy planetesimals stirred by Pluto-size dwarf planets (Kenyon & Bromley 2010).

Here, we have presented new Herschel and updated Spitzer photometry of five Sunlike stars with debris disks of Kuiper Belt size and larger. Both extended structure in the images and SED fits from the DDS (Wolf & Hillenbrand 2005) indicate that the disks are not composed of blackbody grains; instead, small grains are present and disks are wider than their blackbody radii. As seen by Booth et al. (2013), the minimum grain sizes in the best-fit SEDs are in the 3.5–4.5 μm size range, which are larger than the radiation-pressure blowout size. All targets are younger than the Sun (although still on the main sequence), and so we may be seeing Kuiper Belt analogs at earlier phases of evolution. The disks have luminosities consistent with dust production from icy planetesimals stirred by Pluto-size bodies (Kenyon & Bromley 2010). We also find that the Herschel 160 μm emission likely emanates from the birth ring of planetesimals, and so we add our targets to the growing list of Sunlike stars with planetesimals beyond the radius of the solar nebula. One caveat is that our SED models may not be unique: we assumed 100% astronomical silicate grains and did not consider mixtures of silicate, ice, and/or organic grains (although icy grain mantles will likely photoevaporate). Another unexplored possibility is that different grain sizes have different radial distributions.

Although three of the stars host planets discovered by Doppler searches (HD 50554, HD 52265, and HD 202206), the planets are on short-period orbits, which should be dynamically decoupled from the debris disks. Any planets that sculpt the debris disks in this sample have not yet been detected. HD 33636 has an M6 companion in a ∼3 AU orbit (Bean et al. 2007), indicating that the debris disk at 85 AU is circumbinary. HD 105211 may also be a spectroscopic binary, although Eggleton & Tokovinin (2008) find a single star more probable. HD 33636 and (possibly) HD 105211 complement ∼30 previous detections of circumbinary debris disks, although there is some indication that the frequency of cold debris disks may be higher among single stars (Rodriguez et al. 2015). There is no conclusive evidence that any of our targets host "holey" debris disks, which have planets in the gaps (Kennedy & Wyatt 2014; Meshkat et al. 2015), although the IRS spectrum of HD 105211 may have a peak at 19 μm (see Section 5.1).

Based on target declination, ALMA followup would be possible for HD 33636, HD 52265, HD 105211, and HD 202206. The most compelling target is HD 105211 due to both its brightness—0.7 Jy at 100 μm—and its spatial extent of >14'', which is wider than the disk surrounding the T Tauri star TW Hydrae. HD 33636 and HD 52265 are also good ALMA targets based on their extended structure in the Herschel PACS images. We also suggest JWST followup of HD 105211 using the (a) the MIRI coronagraph with the 23 μm filter and (b) the MIRI low-resolution spectrometer. Based on the allWISE W1 photometry, we find that the the Dodson-Robinson et al. (2011) photosphere model is likely accurate, which indicates that HD 105211 has a strong infrared excess at 23 μm. Observations with JWST would allow us to precisely locate the inner edge of the cold debris ring and possibly detect a hot inner debris belt, if it exists.

Funding for this work was provided by NASA research support agreement 1524391. We thank Roberta Paladini of the HASA Herschel Science Center for guidance on data reduction and Sebastian Wolf for developing the publicly available Debris Disk Simulator tool. John Gizis and Neal Evans provided helpful input on data analysis methods. This research has made use of the following resources: (1) Herschel Interactive Processing Environment (HIPE) and the ipipe scripts for data reduction. (2) VizieR catalogue access tool, CDS, Strasbourg, France. The original description of the VizieR service was published in A&AS 143, 23. (3) NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the NASA. (4) SIMBAD database, operated at CDS, Strasbourg, France; Wenger et al. (2000). (5) Herschel Science Archive, www.cosmos.esa.int/web/herschel/science-archive. (6) Astropy python library for astronomy, www.astropy.org. (7) APLpy (Astronomical Plotting Library in Python), aplpy.github.io. (8) Data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration.