An Ongoing Mid-infrared Outburst in the White Dwarf 0145+234: Catching in Action the Tidal Disruption of an Exoasteroid?

A substantial fraction of white dwarfs (WDs) are known to show metal absorption lines in their spectra (Zuckerman et al. 2010; Koester et al. 2014). Because heavy elements diffuse out of the photosphere in a rather short time (days to hundred years, depending on T_eff and log g) in the strong surface gravity of a WD⁹ (Koester 2009; Bergeron et al. 2015), this indicates that these heavy elements were added to the surface very recently, presumably, by accretion of tidally disrupted asteroids or comets (Jura & Young 2014). Many WDs also display infrared excesses, which may be considered as evidence for dust disks (Zuckerman & Becklin 1987; Farihi et al. 2009; Rebassa-Mansergas et al. 2019). The presence of a warm disk is further supported by the detection of double-peaked infrared Ca ii emission lines (Jura 2003; Gänsicke et al. 2006; Veras et al. 2014). Dust in the disk in a WD is either transient or continuously replenished because grains can be dragged onto the star through the Poynting–Robertson force on timescales of several years (e.g., Reach et al. 2005; Rafikov 2011). Fragmentation of a planetary body has been proposed as a source of dust (Jura & Young 2014; Vanderburg et al. 2015; Manser et al. 2019). Variability in infrared flux on a timescale from less than a year to 10 years has been reported very recently in a number of WDs (Farihi et al. 2018; Xu et al. 2018; Swan et al. 2019a), but most of variations are generally of small amplitudes (e.g., 20%–40%, Xu & Jura 2014; Xu et al.2018; Swan et al. 2019b) on timescales of a year.

In this Letter, we report the discovery of a large MIR outburst in the WD 0145+234 (01:47:54.81 + 23:39:43.6; Wills & Wills 1974). This outburst has so far lasted one and a half years. This event was discovered during a blind search for large-amplitude MIR outbursts in the WISE/NEOWISE data archive (Wright et al. 2010; Mainzer et al. 2014). This could be a process of tidal disruption caught in action.

The Wide-field Infrared Survey Explorer (WISE) all-sky survey and NEOWISE Reactivation mission conducted repeated scans of the entire sky at 3.4 and 4.6 μm (hereafter W1 and W2) beginning in 2010 January through the present, except for the period 2011 February–2013 November. We retrieved the W1 and W2 point-spread function profile-fit photometry of 0145+234 from the AllWISE Multi-epoch Photometry Table and NEOWISE-R Single Exposure (L1b) Source Table,¹⁰ which contain all measurements from 2010 to 2019 July. The single-exposure data were first screened by the quality flag marked in the catalogs to remove measurements with poor quality or possible corruption: (qual_frame < 5), charged particle hits (saa_sep < 5), scattered moon light (moon_masked = 1), and artifacts (cc_flags > 0). The high-quality measurements in each six-month observation epoch were averaged to increase the signal-to-noise ratio of the photometry, resulting in 15 epochs (see Figure 1). No significant variations are present in the first 12 epochs and thus we take them as the quiescent state with average magnitudes of 13.87 ± 0.04 and 13.63 ± 0.06 in the W1 and W2 bands, respectively. The light curve displays a large outburst between the 2018 January and 2018 July epochs (ΔW1 = 0.93 ± 0.04, ΔW2 = 1.01 ± 0.06) and continues rising in the later epochs. Overall, the outburst appears to display a redder-when-brighter trend in which the variation amplitude in the W2 band is larger than that in the W1 band.

Zoom In Zoom Out Reset image size

The brightening of 0145+234 is not the result of object's motion carrying it close to a bright nearby source. The proper motion of the WD is measured by Gaia to be μ_R.A. = − 5.21 ±0.12 and μ_decl. = − 97.59 ± 0.08 mas year⁻¹ (Gaia Collaboration et al. 2018). Examination of the AllWISE Atlas Image (epoch 2010) shows that the closest MIR source is approximately 14'' to the southwest of the WD position. The nearby source was cleanly separated and measured in the AllWISE Catalog, and is 1.6 mag fainter in W1 and W2 than J0145+234 in 2010. Even if the WD moved directly onto that source, which it did not (see Figures 2 and 3), the apparent flux increase would be much too small to account for the brightening observed in the light curve.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The relative motion of 0145+234 is clearly detected from the AllWISE and NEOWISE astrometry that spans a time baseline of 9 yr and is fully consistent with proper motion measured by Gaia, as illustrated in Figure 4. The position of the white dwarf during outburst does not show any deviation from the expected motion larger than 01. Thus, a chance coincidence with a bright nearby source could not have shifted the photocenter of the WD, making this occurrence unlikely.

Zoom In Zoom Out Reset image size

W1 and W2 images comparing the region around 0145+234 in 2014 January, 2018 January, and 2019 January are shown in Figures 2 and 3. The 2014 January and 2018 images are pre-outburst, and the difference images between the 2018 and 2014 show only a very small positive/negative residual due to the slight motion of the source in the four years separating the observations. The differences between the 2019 and 2014 images show a bright, pointlike image at the position of the WD that is the source in outburst.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

0145+234 is bright in the optical band (V ∼ 13.93 mag) and thus has been well measured by various optical time-domain surveys. We retrieved the optical photometry data from the publicly released data archive from the Catalina Real-time Transient Survey ¹¹ (CRTS; Drake et al. 2009) and All-Sky Automated Survey for Supernovae (ASAS-SN; Shappee et al. 2014; Kochanek et al. 2017).¹² The CRTS survey has monitored 0145+234 since 2005 without filters, but the photometry is calibrated to a pseudo-V magnitude using a few dozen preselected standard stars in each field. The public CRTS data are available to 2013 October. Fortunately, the public ASAS-SN survey can serve the subsequent V-band photometry. Although the latest observations are performed in the g band, these g-band data largely overlap with V-band data. In contrast to the remarkable MIR variability, 0145+234 is quite stable and shows negligible variability in the long-term (more than one decade) optical light curves, including the MIR outburst period.

We made the quiescent spectral energy distribution (SED) of the WD by collecting data from GALEX, PANSTARRs (Chambers et al. 2016), Gaia (Gaia Collaboration et al. 2016), 2MASS (Skrutskie et al. 2006), and ALLWISE. We matched the UV to near-infrared photometry with the synthesized WD SED models (Figure 5).¹³ The models cover the range of T_eff from 2500 K to 90,000 K and log g from 7.0 to 9.0 with a step of 0.5 for DA WDs, and T_eff from 3250 K to 70,000 K and log g from 7.0 to 9.0 for DB WDs (Bergeron et al. 2011; Tremblay et al. 2011; Blouin et al. 2018). The interstellar reddening of the CCM law with R_V = 3.1 (Cardelli et al. 1989) was added with the E(B – V) as a free parameter. The best-fitted parameters converge to a model T_eff = 13,000 K, log g = 8.0, and E(B − V) = 0.014 ± 0.004. They are consistent with those derived from the spectroscopic model (T = 13,060 ± 217 K, log g = 8.13 ± 0.05, Gianninas et al. 2011; T = 12,910 ± 219 K, log g = 8.11 ± 0.05, Limoges et al. 2015). There is no evidence for a stellar companion for this WD either from broadband SEDs (Figure 5) or the optical spectrum. We examined their optical spectrum in the Montreal WD database,¹⁴ and found an absorption line of Ca ii 3934. Unfortunately, Ca ii 3969 is difficult to identify because it falls in the absorption trough of a Balmer absorption line. This suggests it is a DAZ WD. With these photospheric parameters, we also derived other parameters of the WD¹⁵ (Fontaine et al. 2001; Gianninas et al. 2011): the mass of M_* = 0.667 M_☉, the radius of ${R}_{\mathrm{WD}}=0.0116\,{R}_{\odot }$, the luminosity of L_WD = 0.00350 L_☉, and the age of 0.381 Gyr.

2.1. IR Excess in the Low State

2.2. Dust in the High State

In the high state, this source is more than one magnitude brighter in the W1 and W2 bands. Since we only have two data points, we chose to fit the MIR excess with a single-temperature blackbody model. We considered two different scenarios: (1) the quiescent dust disk has been transformed into the latter one; (2) the quiescent disk remains the same and there is a new dust component that contributes to the brightening of the source. In the first case, we fit the total observed excess in the W1 and W2 bands at high state with a blackbody model. In the second case, we only fit the varying W1 and W2 fluxes with a blackbody model to constrain the newly formed dust component. The varying fluxes were calculated by subtraction of the mean quiescent flux from the high state fluxes, and uncertainties were calculated through propagations of errors. These results are summarized in Table 1 (see also Figure 5). In the first case, the temperature was increased from 1163 K to 1769 K from quiescent epochs (1–12) to epoch 13, while the surface area might be increased but the large uncertainties in the quiescent state make this inconclusive. In the subsequent epochs, the dust temperature is decreased continuously to 1146 K at the last epoch, while the surface area increases by a factor of about four. In the second case, the newly formed dust had an initial temperature of about 1950 K, and decreased to 1115 K at the last epoch, while surface area increased by a factor of about six.

Table 1.  Blackbody Model Fit to the MIR Excess

Epoch	W1	W2	Scenario 1a			Scenario 2b
			T	A	R	T	A	R
	(mag)	(mag)	(K)	(1016 m2)	(R)☉	K	1016 m2	R☉
1–12	13.87 ± 0.04	13.63 ± 0.06	1163 ± 145	3.54 ± 1.50	1.46 ± 0.37	...	...	...
13	12.94 ± 0.01	12.62 ± 0.02	1769 ± 74	4.80 ± 0.51	0.63 ± 0.05	1949 ± 529	2.99 ± 1.88	0.52 ± 0.28
14	12.99 ± 0.01	12.60 ± 0.02	1515 ± 62	6.96 ± 0.84	0.88 ± 0.07	1570 ± 329	4.89 ± 2.77	0.80 ± 0.34
15	12.79 ± 0.02	12.18 ± 0.01	1146 ± 32	23.06 ± 2.08	1.52 ± 0.09	1115 ± 130	21.34 ± 8.80	1.59 ± 0.37

Notes.

^aFitting excesses to the WD model. ^bFitting quiescent state subtracted MIR flux.

Download table as:  ASCIITypeset image

With the above fit parameters, we can estimate physical quantities of the dust. The distance to the WD is 29.458 ±0.044 pc using the parallax from the Gaia DR2 data (Gaia Collaboration et al. 2016, 2018; Lindegren et al. 2018), which is consistent with that derived from the SED fitting. We adopted the bolometric luminosity of 0.0035 L_☉ and WD radius of 0.0116 R_☉ from the spectroscopic WD model¹⁶ (Fontaine et al. 2001; Gianninas et al. 2011). Assuming the dust is gray and in thermal equilibrium, we estimated the distance to the central star is on the order of

$$\begin{eqnarray}&&{R}_{\mathrm{dust}}={\left(\displaystyle \frac{{L}_{\mathrm{WD}}}{4\pi \sigma {T}_{d}^{4}}\right)}^{1/2}={\left({T}_{\mathrm{eff}}/{T}_{d}\right)}^{2}{R}_{\mathrm{WD}}.\end{eqnarray} \tag{ 1 }$$

The distances of the dust to the WD are also listed in Table 1. The closest distance to the WD is about 0.5 R_☉.

The covering factor of the dust can be estimated from the ratio of the blackbody luminosity to the WD luminosity assuming dust is optically thick to the UV to optical light of the WD. The covering factor in the high state sharply rises to around (1.5−2.0) × 10⁻² from (2.7 ± 0.3) × 10⁻³ in the quiescent state, and remains nearly constant in the last three epochs.

Real dust is likely not gray, so the derived temperature and dust mass depends on the size distribution of grains (N(a)) and their absorption coefficients (Q). In the following, we assume that the sizes of dust grains follow a power-law distribution with an index n in the range from a_min to a_max and none outside the range. The dust emission flux can be estimated

$$\begin{eqnarray}\begin{array}{rcl}{f}_{\nu } & = & {\int }_{{a}_{\min }}^{{a}_{\max }}N(a)4\pi {a}^{2}{Q}_{\nu }(a){da}\pi {B}_{\nu }(T)/(4\pi {d}_{L}^{2})\\ & = & \displaystyle \frac{3{M}_{d}}{4\rho {d}_{L}^{2}}\left\langle \displaystyle \frac{{Q}_{\nu }}{a}\right\rangle {B}_{\nu }(T),\end{array}\end{eqnarray} \tag{ 2 }$$

where

$$\begin{eqnarray}&&\left\langle \displaystyle \frac{{Q}_{\nu }}{a}\right\rangle =\displaystyle \frac{{\int }_{{a}_{\min }}^{{a}_{\max }}N(a){a}^{3}(Q/a){da}}{{\int }_{{a}_{\min }}^{{a}_{\max }}N(a){a}^{3}{da}},\end{eqnarray} \tag{ 3 }$$

and M_d, ρ are the mass and density of the dust, respectively. The d_L is the distance to the WD. In the blackbody case, the absorption coefficient is Q_ν = 1.

Using silicate coefficients¹⁷ in Laor & Draine (1993), we calculated a set of $\left\langle \tfrac{{Q}_{\nu }}{a}\right\rangle$ for a_min = 0.01 μm and ${a}_{\max }=0.1$, 1, 10, 100, 1000 μm. The excess infrared emission is then fitted with each of above models to derive mass and temperature of the dust. The lowest derived-temperatures are T_d = 770, 1031, 948, 800 K for epochs 1–12, 13, 14, 15, respectively, or 60%–70% of those from the blackbody model, when n = −4 and a_max = 0.1 μm. The highest temperatures are within 99% of those from the blackbody model when n = −1 and a_max =1000 μm. The mass of dust is in the range of 3 × 10¹⁵ ρ₁–$3\,\times {10}^{17}{\rho }_{1}\,\mathrm{kg}$ (ρ₁ = ρ/(1 g cm⁻³)), depending strongly on the size distribution (see Figure 6). A strict lower limit on the dust mass is a few times 10¹⁵ kg, obtained when most grains have a size of about 1 μm, which gives the maximum Q_ν/a in the W1 and W2 bands. This minimum dust mass is very close to these of disks in the dustiest WD, e.g., 10¹⁵ kg in G29-38 (Reach et al. 2005) and (0.3 – 1) × 10¹⁵ kg in GD 362 (Jura et al. 2007). For a given distribution, dust mass increases monotonically with time. The increase of the amount of dust can be produced by further breakup of large bodies due to collision.

The infrared light curve consists of a stable pre-burst quiescent phase, a short time rising outburst within half a year, and the MIR excess increasing phase with little variations in the optical light curve during the same period. This suggests that there is either a rapid inflow of an initial dust disk seen in the quiescent state into the inner region or tidal breakup of an asteroid into a clump of grains.

In order to understand various processes that may contribute to the dust changes, we examined the timescales of various processes, for the distance of dust to the host star of ∼1 R_☉, the WD mass of 0.667 M_☉, and the luminosity of 0.0035 L_☉. First, the dynamic time, or the orbit period, is about half an hour. Because we do not have the detailed light curve through the rising phase, we cannot reject the scenario that the newly formed dust is being brought by a giant comet. However, the late cooling (on a timescale of years) is too slow for a departing comet. It is highly unlikely that the event was caused by a passing comet.

Grains spiral onto the host star due to radiation drag (the Poynting–Robertson drag) on a timescale of (Burn 1979; Backman & Paresce 1993)

$$\begin{eqnarray}&&\begin{array}{l}{t}_{\mathrm{PR}}=4.73\left(\displaystyle \frac{b}{1\mu m}\right){\left(\displaystyle \frac{\rho }{{\rm{g}}{\mathrm{cm}}^{-3}}\right)}^{2}{\left(\displaystyle \frac{R}{{R}_{\odot }}\right)}^{2}\\ \times \left(\displaystyle \frac{0.0035{L}_{\odot }}{{L}_{{WD}}}\right)\displaystyle \frac{1}{1+\mathrm{albedo}}\mathrm{yr},\end{array}\end{eqnarray} \tag{ 4 }$$

i.e., a few years in this case. Destruction of a large grain into small ones through collisions takes place on a timescale of (Backman & Paresce 1993)

$$\begin{eqnarray}&&{t}_{\mathrm{col}}=3.5{\left(\displaystyle \frac{R}{{R}_{\odot }}\right)}^{3/2}{\left(\displaystyle \frac{0.667{M}_{\odot }}{{M}_{\mathrm{WD}}}\right)}^{1/2}\left(\displaystyle \frac{{10}^{-5}{L}_{\odot }}{{L}_{\mathrm{IR}}}\right)\mathrm{yr},\end{eqnarray} \tag{ 5 }$$

which is long, ∼10 yr for the quiescent dust disk. Accretion of dust due to friction in the dust disk is much longer than 10⁴ yr (Rafikov 2011; Girven et al. 2012). Given that this outburst event timescale appears to be consistent with the radiation drag and collision dust destruction timescales, it is most likely that the dust is produced by tidal breakup of an asteroid-like object.

The MIR observations are consistent with this picture. The temperature in the initial rising phase is as high as 1770 K in the blackbody fit. Assuming the dust is in radiative equilibrium, we estimate the distance of the dust to the WD is about (0.63 ± 0.05) R_☉, which is consistent within a tidal disruption radius, ${r}_{t}=0.69{\left(\tfrac{{M}_{\mathrm{WD}}}{0.667{M}_{\odot }}\tfrac{3{\rm{g}}{\mathrm{cm}}^{-3}}{\rho }\right)}^{1/3}{R}_{\odot }$ of a gravitationally bounded asteroid with a typical density of ρ = (1 − 7) g cm⁻³ (Carry 2012). The disrupted debris will have a range of specific energies and will be spread in space. A small fraction with the lowest energies may fall directly onto the WD, and pollute the surface of the WD. A large part of debris may form an eccentric disk. Collision in the clump of debris will further cause fragmentation, releasing smaller grains, and thus increases the surface area of dust as observed. Similar events were detected before, for instance, evidence for planetesimals in an orbit close to the tidal disruption radius was reported in WD 1145+017 through the transit signal (Vanderburg et al. 2015; Xu et al. 2016). More recently, Manser et al. (2019) reported detection of a planetesimal orbiting within the debris disk through its perturbation to the gas disk. Farihi et al. (2018) proposed that dust production and depletion through collisions (secular evolution) are common based on the detection of MIR variability of 20% peak-to-peak amplitude over 11 yr in GD 56. Their light curve looked rather smooth. Xu et al. (2018) explored three mechanisms, including tidal disruption events, external perturbation, and runaway accretion, but did not reach a conclusion. The abrupt change in our object, on the other hand, suggests a catastrophic event. Our discovery may be the first case of an asteroid breakup process caught in action.

We report a large MIR outburst of the WD 0145+234, which is still rising in the WISE W2 band. We interpret this event as tidal disruption of an asteroid by the WD. As the source MIR flux is still rising, further monitoring of this event will likely help provide insights to the fate of planet systems at their end of star lives. High resolution spectroscopic observations in the optical and UV can trace the composition of recent falling debris (asteroid; e.g., Swan et al. 2019b). According to the calculation given by Fontaine,¹⁸ the diffusion time for most metals are less than a month. The disruption may also release volatile gas into the interplanetary space, leaving absorption lines in the UV and optical spectra (Wilson et al. 2019). The debris may contain warm gas emitting infrared Ca ii emission lines, which can be used to trace the kinematic motion of the debris. Infrared photometric observations at long wavelengths and MIR spectroscopy can be also used to constrain dust composition.

We thank the referee for constructive comments. This work is supported by Chinese Science Foundation (NSFC-11833007,11421303). This research makes use of 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. This research also makes use of data products from NEOWISE-R, which is a project of the Jet Propulsion Laboratory/California Institute of Technology, funded by the Planetary Science Division of the National Aeronautics and Space Administration. This research has made use of the NASA/ IPAC Infrared Science Archive, which is funded by the National Aeronautics and Space Administration and operated by the California Institute of Technology.