Studying the Evolution of Warm Dust Encircling BD +20 307 Using SOFIA

TL;DR

This study monitors the warm dust around star BD +20 307 over nine years, revealing an unexpected increase in infrared emission that challenges steady-state models and suggests recent catastrophic events.

Contribution

First observational evidence of short-term variability in warm debris disks around stars with extreme dustiness, highlighting the need for dynamic models.

Findings

Infrared emission increased by 10% over nine years.

Shorter wavelength flux increased more than longer wavelengths.

Steady-state models cannot explain the observed variability.

Abstract

The small class of known stars with unusually warm, dusty debris disks is a key sample to probe in order to understand cascade models and extreme collisions that likely lead to the final configurations of planetary systems. Because of its extreme dustiness and small radius, the disk of BD +20 307 has a short predicted collision time and is therefore an interesting target in which to look for changes in dust quantity and composition over time. To compare with previous ground and Spitzer Space Telescope data, SOFIA photometry and spectroscopy were obtained. The system's 8.8-12.5 $\mu$m infrared emission increased by $10 \pm 2 \%$ over nine years between the SOFIA and earlier Spitzer measurements. In addition to an overall increase in infrared excess, there is a suggestion of a greater increase in flux at shorter wavelengths (less than 10.6 $\mu$m) compared to longer wavelengths (greater than 10.6 $\mu$m). Steady-state collisional cascade models cannot explain the increase in BD +20 307's disk flux over such short timescales. A catastrophic collision between planetary-scale bodies is still the most likely origin for the system's extreme dust; however, the cause for its recent variation requires further investigation.

Tables

Table 1. Table 1: Summary of previous observations of BD +20 307’s dust.

Observatory	Instrument	Wavelengths ($\mu$m)	Dates	Observation Type
Keck	LWS	3.9-24.5	08/29/2004	Photometry & Spectroscopy
Gemini-North	Michelle	7.7-18.1	09-10/2004	Photometry & Spectroscopy
Spitzer	IRAC	3.5-7.9	08/20/2005	Photometry
Spitzer	IRS	5.2-37.2	01/15/2006	Spectroscopy
Spitzer	MIPS	24, 70, 160	02/04/2007	Photometry
WISE		3.4, 4.6, 12, 22	01/19/2010-01/20/2011	Photometry
Herschel	PACS	70, 100, 160	07/2011, 01/2012	Photometry
Spitzer	IRAC	3.6, 4.5	2012-2013	Photometry

Table 2. Table 2: Spectroscopy observations with SOFIA’s FORCAST instrument using the ‘G111’ Grism in the Short Wavelength Channel.

Instrument	Date	Altitude Range (ft)	Integration Time (s)
FORCAST ‘G111’	2015 Feb 04	39,953-41,014	486
	2015 Feb 06	39,032-39,031	754

Table 3. Table 3: Photometry observations with SOFIA.

Instrument	Date	Altitude (ft)	Integration Time (s)
FORCAST (Imaging SWC, ‘F111’)	2015 Feb 04	39955	15.61
	2015 Feb 04	39953	15.61
	2015 Feb 04	41013	15.73
	2015 Feb 04	41010	15.73

Full text

1 Introduction

Typical debris disks include leftover planetesimals and are thought to evolve through collisions or evaporation of solid bodies, ranging from small planetesimal to protoplanet/planet-sized (Krivov, 2010; Wyatt, 2008). Just like the Solar System’s Kuiper Belt located beyond the orbit of Neptune, most debris disks contain low-temperature dust ($\leq$ 100 K) orbiting far from the host star. However, there exists a small class of known stars with unusually warm, dusty debris disks that serve as a key sample to probe in order to understand cascade models and extreme collisions that likely lead to the final configurations of planetary systems. Collisional cascade, a process in which larger planetesimals in the disk collide and are continually broken up into smaller objects can explain most debris disks. In a collisional cascade, small debris disks with warm dust do not last for very long because once the dust has reached a small enough size, removal mechanisms operate quickly, such as radiation pressure that blows the dust out of the system or Poynting-Robertson drag that causes dust particles to fall into the star (Wyatt, 2008).

Compositional changes in warm, dusty disks may be observable on extremely short timescales of years. As a case in point, observations of TYC 8241 2652 1, a young, Sun-like star with a warm, dusty disk, saw a decrease in the disk’s dust emission by a factor of 30 in less than two years, and there is currently no physical model that can explain what would cause such rapid dust depletion (Melis et al., 2012). Similarly, Meng et al. (2015) used Spitzer to study five debris disks with unusually high fractional luminosities and found major variations on timescales shorter than a year.

BD +20 307 can perhaps provide insight into extreme collisional events occurring beyond our Solar System because its warm dust makes it the dustiest star known for its age of $\geq$1 Gyr. The system is a tidally-locked spectroscopic binary composed of nearly identical late-F-type dwarf stars orbiting with a short 3.4-day period, located at 120.0 $\pm$ 0.7 pc from Earth (Weinberger, 2008; Fekel et al., 2012; Gaia Collaboration et al., 2018). Two previous instruments obtained infrared spectra for BD +20 307: the first in 2004-2005 using Keck and Gemini at 8–13 $\mu$m (Song et al., 2005); the second in 2005-2007 using Spitzer at 5–37 $\mu$m (Weinberger et al., 2011). In addition, long wavelength photometry was collected with Spitzer and in 2011-2012 using Herschel (Vican et al., 2016). About a decade later in 2015, we obtained new 8–13 $\mu$m infrared spectra using the Stratospheric Observatory for Infrared Astronomy (SOFIA), which is the focus of this paper.

For BD +20 307’s disk, assuming its $L_{IR}$/$L_{\star}$ = 0.032 represents the surface density of the dust grains, we expect a collision time of only 2.4 years. In addition, since small grains ($\leq$1 $\mu$m) are likely created by collisions, radiation pressure should blow them away on an orbital timescale. Given the old age of the system, collisional cascade cannot explain the large amount of observed dust over the lifetime of BD +20 307 (Wyatt et al., 2007). Dust at this flux level cannot last over the system’s age, so the dust must be transient. Therefore, another explanation is needed for the cause of the large amount of dust around such a mature system. In addition, it is important to note that, based on the spectral energy distribution (SED), Weinberger et al. (2011) and Vican et al. (2016) conclude that there is no cold dust emitting at far-infrared wavelengths and that the mean temperature of the dust is $\sim$420 K, which suggests an extreme planetary-scale collision within 1 AU to explain BD +20 307’s debris disk.

In this paper, we report on the changes in BD +20 307’s debris dust by analyzing its spectrum from SOFIA and comparing it to the previous spectra from Keck/Gemini and Spitzer. In Section 2, we summarize previous observations. In Section 3, we describe our SOFIA observations of BD +20 307 and the analysis of its new spectra. We present our results in Section 4, which suggest that we detect significant differences between the SOFIA spectrum and those earlier spectra from Spitzer and the ground-based instruments. In Section 5, we discuss possible mechanisms to explain the observed changes in this extremely warm and dusty debris disk over a decadal timescale. Finally, in Section 6 we summarize our work and propose additional avenues for further analysis of BD +20 307’s unusual debris disk system.

2 Previous Observations

The first detailed set of observations to study the debris surrounding BD +20 307 were obtained using two ground-based telescopes: the W.M. Keck Observatory and the Gemini-North Telescope, both located on Mauna Kea in Hawaii. Using the Keck Long Wavelength Spectrometer (LWS) and the Gemini Michelle instrument, Song et al. (2005) obtained spectra of BD +20 307 taken over the course of three months from August to October of 2004. The resolving powers of the Keck LWS and the Gemini Michelle instrument are R$\approx$150 and $\approx$900, respectively. Song et al. (2005) assumed that there was no change in flux over the time period of their observations.

A year later, the Spitzer Space Telescope gathered the second detailed set of spectroscopy and photometry observations on BD +20 307 (Weinberger et al., 2011). On August 20, 2005, Spitzer’s Infrared Array Camera (IRAC) gathered short wavelength photometry, and on January 15, 2006, the Infrared Spectrometer (IRS) obtained spectra of BD +20 307. About a year later, on January 21, 2007, the Multiband Imaging Photometer for SIRTF (MIPS) gathered long wavelength photometry of BD +20 307. The IRS spectral resolution from 8 to 13 $\mu$m was R$\approx$120-600. Weinberger et al. (2011) also made the assumption that there were no changes in the flux or spectrum despite the fact that the observations spanned August 2005 to January 2007.

We re-normalized the Spitzer IRS spectrum to the IRAC 5.7 $\mu$m filter instead of to the MIPS 24 $\mu$m filter as in Weinberger et al. (2011) because the IRAC photometry was gathered closer in time to the IRS spectrum. There are slight differences in the Spitzer IRS spectrum normalized to the IRAC flux compared to the MIPS flux (see Section 2.3 of Weinberger et al. (2011)). By averaging them over SOFIA’s 11.1 $\mu$m filter transmission curve, we found that they differ by 6.8%. In the $\sim$ 1.5 years between the Keck/Gemini and Spitzer observations, there is indication that the flux increased slightly over time, particularly at shorter wavelengths. The ratio of the weighted average flux values between the Spitzer and Keck/Gemini spectra for wavelengths less than 10.6 $\mu$m is $1.14\pm 0.002$. It is important to note that we have taken the as-published Keck/Gemini data, and there may be additional calibration uncertainties that were accounted for. The uncertainty we have quoted should be considered as a lower bound.

In January and July 2010, during the cryogenic mission, WISE measured BD+20 307 at all four of its bands (3.5 - 22 $\mu$m), and in January 2011, it did a scan in the two short wavelength bands. Meng et al. (2012) noted that BD +20 307’s disk may vary by $\sim 7\%$ between the two cryogenic scans taken $\sim$188 days apart but only in the W3 and W4 bands (12 and 22 $\mu$m, respectively). We show the WISE photometry at each epoch in Figure 1, and find, in agreement with Meng et al., that all four bands show the same trend, i.e. a small brightening over time, although W1 and W2 are constant within their uncertainties.

In July 2011 and January 2012, the Herschel Space Observatory used its Photodetector Array Camera and Spectrometer (PACS) to measure BD +20 307 at 70 and 100 $\mu$m. The reported 70 $\mu$m flux density of 47 $\pm$ 3 mJy (Vican et al., 2016) is substantially larger than that reported with MIPS from 2007 of 28.6 $\pm$ 1.9 mJy (Weinberger et al., 2011).

Most recently, from 2012 to 2013, Meng et al. (2015) made near-infrared time-series observations of five extreme debris disks including BD +20 307 using Spitzer’s IRAC at 3.6 and 4.5 $\mu$m. Although they did not detect a significant trend at 3.6 $\mu$m, at 4.5 $\mu$m they found that BD +20 307’s disk flux increased over their period of observations with an average increase rate of $2.5\pm 0.7$ mJy per year. Such an increase rate coincides within a few percent of the disk flux every year (Meng et al., 2015). Table 1 summarizes the main sets of previous infrared observations on BD +20 307’s dust.

3 Methods

3.1 SOFIA Observations

For the SOFIA Cycle 2 Program 02_0050 (PI: Weinberger), we used SOFIA’s Faint Object Infrared Camera (FORCAST), a dual-channel mid-infrared camera and spectrograph sensitive to 5-40 $\mu$m. Observations of BD +20 307 were obtained in 2015 flying at an average altitude of 40,000 feet. Tables 2 and 3.1 below summarize the SOFIA observations for gathering spectroscopy and imaging data of BD +20 307. Using the ‘Nod-Match-Chop’ grism observing mode with a chopper throw of 60*′′*, data were taken with the ‘G111’ grism that has a nominal resolving power of R=130 for the 4.7” x 191” slit and covers 8.4-13.7 $\mu$m.

A total of 55 individual spectra were taken over two nights, 23 on February 4 and 32 on February 6. We analyzed the products produced by the standard ‘FORCAST_REDUX’ pipeline version 1.2.0. The general pipeline process is as follows: load the data and calculate variance; clean bad pixels; correct for applied channel suppression (i.e., droop); correct for image non-linearity; perform background subtraction; remove jailbars; perform spectral extraction using FSpextool algorithm (for spectra only); combine multiple observations (i.e., stacking); calibrate flux. It is important to note that there is a large amount of noise in the data between 9.5-9.9 $\mu$m caused by telluric ozone absorption. The spectrum from each night is available from the SOFIA Data Cycle System as a FITS file that contains the combined spectrum corrected for atmospheric transmission and instrumental response.