A Trio of Metal-Rich Dust and Gas Disks Found Orbiting Candidate White Dwarfs with K-Band Excess

Abstract

This paper reports follow-up photometric and spectroscopic observations,
including warm Spitzer IRAC photometry of seven white dwarfs from the SDSS with
apparent excess flux in UKIDSS K-band observations. Six of the science targets
were selected from 16,785 DA star candidates identified either
spectroscopically or photometrically within SDSS DR7, spatially
cross-correlated with HK detections in UKIDSS DR8. Thus the selection criteria
are completely independent of stellar mass, effective temperature above 8000 K,
and the presence (or absence) of atmospheric metals. The infrared fluxes of one
target are compatible with a spatially-unresolved late M or early L-type
companion, while three stars exhibit excess emissions consistent with warm
circumstellar dust. These latter targets have spectral energy distributions
similar to known dusty white dwarfs with high fractional infrared luminosities
(thus the K-band excesses). Optical spectroscopy reveals the stars with
disk-like excesses are polluted with heavy elements, denoting the ongoing
accretion of circumstellar material. One of the disks exhibits a gaseous
component - the fourth reported to date - and orbits a relatively cool star,
indicating the gas is produced via collisions as opposed to sublimation,
supporting the picture of a recent event. The resulting statistics yield a
lower limit of 0.8% for the fraction dust disks at DA-type white dwarfs with
cooling ages less than 1 Gyr. Two overall results are noteworthy: all stars
whose excess infrared emission is consistent with dust are metal-rich; and no
stars warmer than 25,000 K are found to have this type of excess, despite
sufficient sensitivity.

Full text

Page 1

arXiv:1112.5163v1 [astro-ph.SR] 21 Dec 2011
Mon. Not. R. Astron. Soc. 000, 000–000 (0000)Printed 23 December 2011(MN LATEX style file v2.2)
A Trio of Metal-Rich Dust and Gas Disks Found Orbiting
Candidate White Dwarfs with K-Band Excess
J. Farihi1⋆, B. T. G¨ ansicke2, P. R. Steele3, J. Girven2, M. R. Burleigh1,
E. Breedt2and D. Koester4
1Department of Physics Astronomy, University of Leicester, Leicester LE1 7RH, UK
2Department of Physics, University of Warwick, Coventry CV4 7AL, UK
3Max Planck Institut f¨ ur Astrophysik, D-85741 Garching, Germany
4Institut f¨ ur Theoretische Physik und Astrophysik, University of Kiel, 24098 Kiel, Germany
ABSTRACT
This paper reports follow-up photometric and spectroscopic observations, includ-
ing warm Spitzer IRAC photometry of seven white dwarfs from the SDSS with appar-
ent excess flux in UKIDSS K-band observations. Six of the science targets were selected
from 16785 DA star candidates identified either spectroscopically or photometrically
within SDSS DR7, spatially cross-correlated with HK detections in UKIDSS DR8.
Thus the selection criteria are completely independent of stellar mass, effective tem-
perature above 8000K, and the presence (or absence) of atmospheric metals. The in-
frared fluxes of one target are compatible with a spatially-unresolved late M or early
L-type companion, while three stars exhibit excess emissions consistent with warm
circumstellar dust. These latter targets have spectral energy distributions similar to
known dusty white dwarfs with high fractional infrared luminosities (thus the K-band
excesses). Optical spectroscopy reveals the stars with disk-like excesses are polluted
with heavy elements, denoting the ongoing accretion of circumstellar material. One of
the disks exhibits a gaseous component – the fourth reported to date – and orbits a
relatively cool star, indicating the gas is produced via collisions as opposed to subli-
mation, supporting the picture of a recent event. The resulting statistics yield a lower
limit of 0.8% for the fraction dust disks at DA-type white dwarfs with cooling ages
less than 1Gyr. Two overall results are noteworthy: all stars whose excess infrared
emission is consistent with dust are metal-rich; and no stars warmer than 25000K are
found to have this type of excess, despite sufficient sensitivity.
Key words: circumstellar matter— planetary systems— stars: abundances— stars:
low mass stars— brown dwarfs— stars: evolution— white dwarfs
1INTRODUCTION
Near-infrared photometry of white dwarfs examines a sci-
entific phase space that is difficult or impossible to probe
for main-sequence stars. The first to recognize this potential
was Probst (1981), who realized the compact nature of white
dwarfs allowed a straightforward photometric detection of
very cool stellar and substellar companions. Brown dwarfs,
which often require sophisticated detection techniques when
searched for at main-sequence stars, are up to 10 times larger
than a typical white dwarf and can be readily detected in
spatially-unresolved observations as excess near-infrared flux
(Probst 1983). Thus, white dwarfs are an excellent tool for
⋆E-mail: jf123@star.le.ac.uk
studies of the low-mass stellar and substellar mass function
via companions (Farihi et al. 2005; Zuckerman & Becklin
1992; Probst & O’Connell 1982). Photometry with Spitzer
has extended this potential to include closely orbiting brown
dwarfs and massive planets although no candidates are yet
apparent (Farihi et al. 2008b; Mullally et al. 2007).
Another way in which near-infrared photometry of
white dwarfs is advantageous is that circumstellar dust or-
biting within the Roche limit of the stellar remnant be-
comes heated sufficiently to emit in this wavelength range
(Kilic et al. 2006). This potential is precluded for main-
sequence stars because the analogous spatial region, where
km-size or larger solid bodies are tidally destroyed, does not
extend significantly above their surfaces (Davidsson 1999).
Furthermore, any material generated in that narrow region
c ? 0000 RAS

Page 2

2J. Farihi et al.
Table 1. Multi-Wavelength Photometry and Derived Parameters for SDSS White Dwarfs
SDSS Prefix09591159a
122112471320 1506 1557
Spectral Type
Spectroscopic Data
α (J2000; hms)
δ (J2000;◦ ′ ′′)
FUV (0.15µm; ABmag)
NUV (0.23µm; ABmag)
u (0.36µm; ABmag)
g (0.47µm; ABmag)
r (0.62µm; ABmag)
i (0.75µm; ABmag)
z (0.89µm; ABmag)
J (1.25µm; mag)
H (1.64µm; mag)
K (2.20µm; mag)
IRAC1 (3.55µm; µJy)
IRAC2 (4.49µm; µJy)
DAZ
WHT
09 59 04.69
−02 00 47.6
...
...
18.52 ± 0.02
18.09 ± 0.03
18.36 ± 0.02
18.54 ± 0.02
18.88 ± 0.06
18.32 ± 0.06
18.17 ± 0.09
17.62 ± 0.11
80 ± 4
76 ± 4
13280 ± 20
8.06 ± 0.03
0.64 ± 0.02
203 ± 4
Dust
DQp
SDSS
DAZ
WHT
12 21 50.81
+12 45 13.3
19.55 ± 0.06
18.78 ± 0.02
18.54 ± 0.02
18.19 ± 0.02
18.40 ± 0.02
18.56 ± 0.02
18.86 ± 0.05
18.43 ± 0.07
18.39 ± 0.10
18.01 ± 0.15
40 ± 5
40 ± 6
12250 ± 20
8.20 ± 0.03
0.73 ± 0.02
180 ± 4
Dust
DA
SDSS
DA
SDSS
DA
WHT
15 06 26.18
+06 38 45.9
18.40 ± 0.02
17.31 ± 0.01
16.96 ± 0.01
16.57 ± 0.01
16.63 ± 0.01
16.79 ± 0.01
16.93 ± 0.01
16.41 ± 0.01
16.37 ± 0.02
16.35 ± 0.03
75 ± 4
50 ± 3
10670 ± 10
8.21 ± 0.03
0.73 ± 0.02
71 ± 2
None
DAZ
SDSS+WHT
15 57 20.77
+09 16 24.6
18.03 ± 0.02
18.16 ± 0.01
18.50 ± 0.02
18.45 ± 0.01
18.81 ± 0.01
19.13 ± 0.01
19.34 ± 0.04
18.82 ± 0.06
19.03 ± 0.14
18.35 ± 0.15
35 ± 2
40 ± 2
22810 ± 40
7.54 ± 0.01
0.42 ± 0.01
566 ± 2
Dust
11 59 33.10
+13 00 31.6
> 20
19.62 ± 0.09
18.22 ± 0.03
18.14 ± 0.02
17.75 ± 0.01
17.70 ± 0.02
17.82 ± 0.02
17.40 ± 0.03
17.32 ± 0.06
17.08 ± 0.08
42 ± 2
28 ± 2
9500 ± 500
8.0
0.59
110 ± 00
Atm
12 47 40.93
+10 35 56.1
18.23 ± 0.02
18.33 ± 0.01
18.63 ± 0.02
18.50 ± 0.01
18.82 ± 0.02
19.13 ± 0.02
19.33 ± 0.06
18.95 ± 0.08
18.90 ± 0.11
18.47 ± 0.20
25 ± 2
17 ± 2
18540 ± 150
7.86 ± 0.03
0.54 ± 0.02
396 ± 9
dM/L
13 20 44.68
+00 18 54.9
16.52 ± 0.03
16.76 ± 0.01
17.08 ± 0.02
17.13 ± 0.01
17.47 ± 0.02
17.72 ± 0.02
18.05 ± 0.03
17.62 ± 0.04
17.71 ± 0.08
17.53 ± 0.11
35 ± 2
25 ± 2
20220 ± 180
8.37 ± 0.04
0.86 ± 0.02
150 ± 4
Bkgd
Teff(K)
log g (cms−2)
M (M⊙)
d (pc)
IR Excessb
Note. Ultraviolet, optical, and near-infrared photometry are from GALEX (Martin et al. 2005), SDSS (Abazajian et al. 2009), and
UKIDSS (Lawrence et al. 2007) respectively. For stars with multiple spectroscopic or photometric datasets, table values are the
weighted average of available measurements. The SDSS photometry is given in PSF magnitudes; the only catalog entries appropriate
for point sources. Stellar parameters are derived by fitting the Balmer line profiles as described in §3.
aThis star was selected independently from the six primary science targets (see §3.2).
bSee §3 for the various infrared excess descriptions.
above the star will rapidly dissipate due to radiation pres-
sure and drag forces. Thus, while asteroids and comets com-
monly pass sufficiently close to the Sun, and presumably
other main-sequence stars that host planetary systems, the
tidal disruption and subsequent accretion of this debris can
only be witnessed at white dwarfs (and possibly neutron
stars; Wang et al. 2006).
This paper reports warm Spitzer IRAC measurements
and optical spectroscopy of six white dwarf candidates se-
lected for apparent K-band excess fluxes based on the
ground-based photometric surveys the Sloan Digital Sky
Survey (SDSS) and the UKIRT Infrared Deep Sky Survey
(UKIDSS). Target selection criteria and survey data are
given in §2, along with a description of the Spitzer and
ground-based observations. The spectroscopic and photo-
metric data analysis is presented in §3, with a summary
of the overall results in §4.
2TARGET SELECTION AND OBSERVATIONS
The six primary science targets were selected as de-
scribed in detail by Girven et al. (2011). Briefly summa-
rizing, DA white dwarf candidates were selected both pho-
tometrically and spectroscopically from within SDSS DR7
(Abazajian et al. 2009), and spatially cross-correlated with
sources in the UKIDSS (Lawrence et al. 2007) DR4 (and
later DR8). A ugriz color selection was implemented based
on the Eisenstein et al. (2006) SDSS DR4 white dwarf cat-
alog, which was then used to select and download spectra
for 7444, and photometry (only) for 9341, DA star candi-
dates with g < 19ABmag. Cross-correlation with UKIDSS
DR8 then resulted in 1884 objects in common with both
H- and K-band detections; these were all fitted with white
dwarf atmospheric models to probe for K-band photometric
excess. Of these 1884 cross-correlated sources, 147 objects
were identified as having an infrared excess, including 12 ob-
jects (7 spectroscopic and 5 photometric only) in which the
excess was most pronounced in the K-band, and compatible
with either a brown dwarf later than L7 or a dust disk; the
UKIDSS data alone could not typically distinguish between
those two possibilities.
Six disk candidate white dwarfs found in this manner
were selected as targets for Spitzer Cycle 6 observations,
and are listed in Table 1. Two of these targets (1320 and
1557) were independently identified by Steele et al. (2011).
They performed a cross-correlation of the white dwarf cat-
alogs of Eisenstein et al. (2006) and McCook & Sion (1999)
with UKIDSS DR8 to identify stars with excess emission
via optical and infrared colors, as well as spectral fitting.
A seventh target, the DQ peculiar white dwarf LP494-12
(SDSS1159; Eisenstein et al. 2006) was selected on the ba-
sis of an independent measurement of K-band excess (Farihi
2009), corroborated by its UKIDSS photometry.
2.1 Optical Spectroscopy
Of the six primary science targets, SDSS spectra were avail-
able within SDSS DR7 for 1247, 1320, and 1557. In addition,
long slit spectroscopy was obtained for 0959, 1221, 1506, and
c ? 0000 RAS, MNRAS 000, 000–000

Page 3

White Dwarfs with K-Band Excess3
Figure 1. Normalized SDSS and WHT spectra of the six DA
white dwarfs in the study are shown in gray, along with the best
fit models plotted in black. The SDSS spectra stop at 3800˚ A.
The WHT spectra of 0959, 1221, and 1557 each contain a narrow
absorption line near 4481˚ A, that is due to Mgii.
1557 during two observing runs in 2010 April and May us-
ing using ISIS, the dual beam spectrograph mounted on the
4.2m William Herschel Telescope (WHT). The primary aim
of these observations was to corroborate the photometrically
selected DA candidates (0959, 1221, 1506) and to improve
on the SDSS spectroscopy of 1557. For this purpose, the
R600 grating was used in the blue arm with a 1′′slit and a
2 (spectral) by 3 (spatial) pixel binning. This setup covered
the wavelength range 3650 − 5120˚ A, i.e., the entire Balmer
series except Hα, with an average dispersion of 0.88˚ A per
binned pixel in the blue arm, and 7691−9184˚ A with 0.99˚ A
per binned pixel in the red arm. The red spectra had rela-
tively low signal-to-noise ratio (S/N).
The blue spectra were de-biased and flat-fielded using
the starlink1packages kappa and figaro and then opti-
mally extracted using the pamela2code (Marsh 1989). The
extracted spectra were wavelength calibrated using CuNe
and CuAr arc lamp exposures, and finally flux calibrated
using observations of appropriate standard stars obtained
with the same instrumental setup. The photometric classifi-
cation of 1221, 1506, and 1557 as hydrogen dominated white
dwarfs was confirmed by the ISIS spectroscopy. Normalized
WHT and SDSS spectra of all Spitzer DA targets are shown
in Figure 1.
Additional observations of 0959 and 1320 were obtained
with the VLT using X-Shooter (D’Odorico et al. 2006) in
service mode in 2010 June-July and 2011 February-March.
The raw frames were reduced using the X-Shooter pipeline
version 1.3.7 within gasgano3. The standard recipes were
used with default settings to extract and wavelength cali-
brate each spectrum. The final extraction of the science and
spectrophotometric standard spectra was carried out using
1Developed and maintained by the Joint Astronomy Centre and
available from http://starlink.jach.hawaii.edu/starlink
2Developed and maintained by T. R. Marsh and available from
http://www.warwick.ac.uk/go/trmarsh
3http://www.eso.org/sci/software/gasgano
Table 2. Independent JHK Photometry
StarJHKs
(mag)(mag)(mag)
1221a
1221b
18.49 ± 0.05
18.50 ± 0.05
18.21 ± 0.05
18.27 ± 0.05
18.05 ± 0.05
18.22 ± 0.10
1320a
1320b
17.64 ± 0.05
17.60 ± 0.05
17.44 ± 0.05
17.61 ± 0.07
17.49 ± 0.05
17.58 ± 0.11
1557 19.05 ± 0.0518.92 ± 0.05 18.56 ± 0.05
aAperture photometry
bPSF-fitting photometry
apall within IRAF. Finally, the instrumental response was
removed by dividing each observation by the response func-
tion, calculated by dividing the associated standard by its
corresponding flux table.
2.2Additional Near-Infrared Photometry
Independent near-infrared photometry for 1221, 1320, and
1557 was obtained on the 23 March 2011 with the WHT
using the Long-Slit Intermediate Resolution Infrared Spec-
trograph (LIRIS; Manchado et al. 1998). Images taken in
a 9-point dither pattern were obtained in the J-, H-, and
Ks-band filters (MKO system; R. Karjalainen 2011, private
communication) with typical total exposure times of 270s in
clear conditions. Three standard star (ARNICA; Hunt et al.
1998) fields were observed in a similar manner for photo-
metric zero-point calibration. The data were reduced in the
standard manner, by subtracting a median sky from each
image in the dithered stack, flat-fielding (using sky flats),
then averaging and recombining frames.
LIRIS suffers from what is known as a detector reset
anomaly, which appears in certain frames as a discontin-
uous jump (in dark current) between the upper and the
lower two quadrants. To remove this unwanted signal, af-
ter flat-fielding and sky subtraction, the detector rows were
collapsed into a median column (with real sources rejected),
and subsequently subtracted from the entire two dimen-
sional image. The resulting fully reduced frames exhibit
smooth backgrounds, free of the anomalous gradient.
Aperture photometry of standard stars was performed
using r = 3.′′75 aperture radii and sky annuli of 5′′− 7.′′5
in size. For the relatively faint science targets, smaller pho-
tometric apertures were employed with corrections derived
from several brighter stars within the same image field and
filter. For both 1221 and 1320, point-spread function (PSF)
fitting (i.e., daophot) was used in addition to photometry
with small apertures. All data taken in the Ks-band filter
were flux-calibrated using the ARNICA K-band standard
star photometry, and the error introduced by this should be
significantly smaller than the 5% absolute calibration un-
certainty. The independent JHKs photometry is listed in
Table 2.
c ? 0000 RAS, MNRAS 000, 000–000

Page 4

4J. Farihi et al.
Figure 2. Ultraviolet through infrared spectral energy distributions of the DA-type, infrared excess candidates. Stellar atmosphere
models are plotted as dotted lines, using parameters derived from model fits to hydrogen Balmer lines in the SDSS or WHT spectra,
and matched to the observed g-band fluxes. Table 1 photometry is shown as data points with error bars.
2.3
Spitzer IRAC Observations
Near-infrared imaging observations of the white dwarf tar-
gets were obtained with the warm Spitzer Space Telescope
(Werner et al. 2004) during Cycle 6 using the Infrared Ar-
ray Camera (IRAC; Fazio et al. 2004) at 3.6 and 4.5µm.
The total integration time in each channel was 1200s, where
the observations consisted of 40 frames taken in the cycling
(medium) dither pattern with 30s individual exposures. All
images were analyzed and fluxes measured as in Farihi et al.
(2010b) using 0.6pixel−1mosaics created using MOPEX. In
cases where the flux of a neighboring source was a potential
contaminant of the white dwarf photometry (i.e., 1221 and
1247, see §3.1), steps were taken to minimize or remove any
such external contributions, including small aperture pho-
tometry and PSF fitting with daophot and apex.
Of all IRAC targets, only 1506 is detected in the
WISE Preliminary Data Release. Reliable data exists only
c ? 0000 RAS, MNRAS 000, 000–000

Page 5

White Dwarfs with K-Band Excess5
Figure 3. Black lines are model fits to the Caii 3934˚ A and Mgii
4481˚ A line regions in the normalized ISIS and X-Shooter spectra,
plotted in gray, of all three white dwarfs found to have circum-
stellar dust. The stars are displayed top-to-bottom by decreasing
effective temperature: 22810K, 13280K, 12250K.
at 3.4µm, where the flux (76 ± 8µJy) is in agreement with
the IRAC photometry, albeit with larger errors. Generally,
the science targets are too faint for WISE, where confusion
can be a serious concern (see Melis et al. 2011), and hence
the choice of warm Spitzer.
3RESULTS AND ANALYSIS
The SDSS and WHT spectra of the six DA white dwarfs were
fitted using model atmosphere spectra computed with the
code described by Koester (2010), and following the method
described by Rebassa-Mansergas (2007). In brief, the grid
of model spectra was fitted to the normalized Balmer line
profiles, leading to a ‘hot’ and ‘cold’ solution of roughly equal
absorption line equivalent width. This degeneracy is broken
by fitting the models also to the slope of the flux-calibrated
spectra. The best-fit models are overplotted on the data in
Figures 1 and 2, and the parameters are reported in Table
1.
All targets stars but one reveal excess emission at IRAC
wavelengths when compared to the atmospheric models, de-
rived from their optical spectroscopy. However, there are
only four genuine cases where the excess radiation is asso-
ciated with an additional component physically associated
with the system: three dust disks and one (likely) low-mass
stellar companion. All individual stars are discussed below
in some detail.
3.1 New Disk-Polluted White Dwarfs
Three stars – 0959, 1221, 1557 – display infrared fluxes
consistent with dust disks orbiting within their respective
Roche limits for large asteroids. The excess emissions are
too strong, and the infrared colors are incorrect for sub-
stellar or planetary companions (Farihi et al. 2009, 2008a).
Specifically, even for the case of a substellar object with
anomalously red colors, the 4.5µ flux requires its size to be
Table 3. Relative Number Abundances in Disk-Polluted Stars
Objectlog(Mg/H)log(Ca/H)Mg/Ca
Sun
Chondrites
0959
1221
1557
−4.45
−0.72
−5.2
< −5.6
−4.5
−5.66
−1.96
−7.0
−7.5
< −5.7
16
17
59
< 85
> 15
Note. Measured abundance uncertainties are 0.2dex.
several to ten times that of Jupiter, in contrast with observa-
tions and models of brown dwarfs and planets Leggett et al.
(2010). Firmly corroborating the disk interpretation, each of
these DA white dwarfs was found to be metal-lined in fol-
low up WHT and VLT optical spectroscopy, consistent with
ongoing infall of circumstellar material. Specifically, their
ISIS spectra contain narrow absorption lines near 3934˚ A
and 4481˚ A, which are due to CaiiK and Mgii, respectively;
they are DAZ stars. Figure 3 displays model fits to the the
detected metal absorption features, using tlusty and syn-
spec (Hubeny & Lanz 1995; Lanz & Hubeny 1995) in order
to estimate the photospheric abundances.
Only Mg and Ca are detected, and Table 3 reports the
determined abundances and upper limits for these two el-
ements. The Mg/Ca ratio observed in 0959 is significantly
larger than that found in chondrites, but similar to that
found for the highly polluted white dwarf GALEX1931
(Vennes et al. 2011), where the relative poverty of Ca, Ti,
and Al is potentially consistent with the removal of crust
and mantle in a differentiated, terrestrial-like body during
the post-main sequence evolution of the star (Melis et al.
2011). The Mg abundances determined from the X-Shooter
and ISIS spectra of 0959 are consistent to within the errors,
while the low S/N around the CaiiK line prevents a mean-
ingful abundance estimate for that element based on the ISIS
data. For 1221 and 1557, only upper limits are derived for
Mg and Ca respectively, partly due to the inferior quality of
the ISIS spectra compared to the X-Shooter data. However,
another factor is that the undetected atomic transitions are
not easily excited at the respective effective temperature of
these stars (Zuckerman et al. 2003).
For all three stars, the excess infrared emissions are fit-
ted with optically thick, flat disk models (Jura 2003). As
discussed in previous white dwarf-disk studies, optically thin
dust models are inferior, primarily due to the Poynting-
Robertson timescales involved (von Hippel et al. 2007). Not
only would such dust be removed on the timescale of years,
but the particles orbit once every few hours, ensuring the
disk will relax rapidly into a flat configuration; a geometri-
cally thin, optically thick disk circumvents these issues. Im-
portantly, such a disk can harbor sufficient dust mass to ac-
count for the most highly polluted helium atmosphere stars
(Farihi et al. 2010a), whereas an optically thin disk cannot
(Jura et al. 2009).
The fitted disk parameters are listed in Table 4, and
the model fluxes are shown together with the observational
data in Figure 4. For each star, R/d is derived from the at-
mospheric models using the spectroscopically derived stellar
parameters and observed photometry: R is determined from
Teff and log g while d is specified by m − M for the same
c ? 0000 RAS, MNRAS 000, 000–000

Page 6

6J. Farihi et al.
Table 4. Disk Parameters
StarTeff
(K)
R/d
(10−12)
Tin
(K)
Ta
out
(K)
rin
(R∗)
ra
out
(R∗)
cosi
0959
1221
1557
13280
12250
22810
1.37
1.41
0.73
1600
1400
1400
800
800
800
10
11
25
25
23
52
1.0
0.7
0.5
aOwing to the wavelength coverage, the outer disk temperature
and radius is not as well-constrained by the data as the inner disk
temperature. See §3.1 for a detailed discussion of uncertainties in
the disk parameters.
(e.g., g−Mg, r−Mr). For fixed R/d, there are three free pa-
rameters for a flat disk; inner disk edge temperature, outer
disk edge temperature, and inclination. These models con-
tain a modest amount of degeneracy in their ability to fit
the data, even when longer wavelength data, such as 8 and
24µm photometry are available (Jura et al. 2007). However,
the models provide good, if broad constraints on the geom-
etry and temperature of the disks. The inner disk edge is
fairly well constrained by the excess emission at 2.2 and
3.6µm, but not uniquely so, while the 4.5µm flux can be
reproduced by a relatively wide temperature range and a
higher inclination, or a narrower temperature range and a
lower inclination. This is because the total disk emission is a
function of its temperature profile and solid angle, and the
latter is comprised of the inclination and the ring annuli.
Despite this uncertainty, the infrared excesses cannot be
satisfactorily fit with inner disk temperatures below 1200K,
nor with significant amounts of T < 500K dust. These new
discoveries essentially mimic the thermal emission profiles
observed at 16 dusty white dwarfs with longer wavelength
data in addition to the wavelengths covered by warm Spitzer
(Farihi et al. 2010b). The excess at 0959 is sufficiently strong
that it requires a nearly face-on disk, while the disk incli-
nations for 1221 and 1557 are less constrained. Importantly,
the excesses are consistent with inner disk temperatures ap-
proaching or exceeding that which causes rapid sublimation
of typical dust grains; gas thus produced gives rise to viscous
drag and enhances inflow of disk material onto the stellar
surface (Rafikov 2011). While the outer disk temperatures
are not more tightly bound, the range of acceptable values
places all the debris within 1.5R⊙, where large, solid bod-
ies will be shredded. Overall, the data are consistent with
circumstellar debris originating in large, star-grazing solid
bodies that is now being gradually falling onto the stellar
surface.
3.1.1 0959
This white dwarf displays Caii emission lines that are the
hallmark of closely-orbiting metallic gas. These emission fea-
tures are clearly detected in the X-shooter data, and, upon
close inspection, are marginally visible in the noisy red ISIS
spectrum. Both red spectra are shown in Figure 5, along
with SDSS1228, the prototypical white dwarf with emitting
metal gas and dust (Brinkworth et al. 2009; G¨ ansicke et al.
2006). Notably, the Caii line profiles for this star are sub-
stantially narrower and weaker compared to those detected
in the other three published cases. In 0959 the full width at
Figure 4. Infrared excesses fit by circumstellar dust disk models
whose parameters are listed in Table 4. The plot features are
the same as in Figure 2, but on a linear scale to emphasize the
infrared excesses. Dashed lines represent emission from optically
thick, flat disk models, and a the dashed-dotted lines represent
the sum of the stellar and circumstellar model fluxes.
zero intensity and equivalent width are close to 250kms−1
and 3˚ A, respectively, while the known cases have values in
the range 1000 − 1400kms−1and 10 − 60˚ A (Melis et al.
2010; G¨ ansicke et al. 2008, 2007, 2006). Assuming that the
structure of the gas disks in 0959 and 1228 are similar, the
velocity width of the Caii lines in the two systems can be
c ? 0000 RAS, MNRAS 000, 000–000

Page 7

White Dwarfs with K-Band Excess7
Figure 5. The Caii triplet region of 0959 in the normalized
WHT and X-Shooter spectra exhibit clear emission lines from
a gaseous disk component. For comparison, also plotted is the
normalized spectrum of the prototype gas disk around SDSS1228
(G¨ ansicke et al. 2006). The narrower emission features at 0959 in-
dicate a disk that is significantly inclined relative to that seen at
1228, and consistent with the high fractional infrared luminosity
measured with IRAC.
used to estimate the inclination of the disk orbiting 0959.
For 1228, two independent estimates yield an inclination
of 70◦(Melis et al. 2010; Brinkworth et al. 2009), and this
yields i ≈ 12◦for 0959. The independently estimated disk
inclination from the model fitted to the infrared emission
is 10◦, consistent with a near face-on disk of dust and gas.
G¨ ansicke et al. (2006) found that the Caii emission lines in
1228 are optically thick, and hence a low disk inclination
will naturally result also in a lower equivalent width.
SDSS0959 becomes the fourth published case of a
white dwarf with detectable, gaseous debris in addition
to solid dust particles in a circumstellar disk (Melis et al.
2010; Farihi et al. 2010b; Brinkworth et al. 2009). At Teff =
13300K it is by far the coolest star to host such dual-phase
debris, demonstrating conclusively that the detected gas in
dust disks is not related to grain sublimation in the presence
of higher temperature white dwarfs. In the three well-studied
cases, the gaseous and solid debris are essentially spa-
tially coincident (Melis et al. 2010; Brinkworth et al. 2009),
both spanning a region from around 20 to 100 stellar radii
(G¨ ansicke et al. 2006). Critically, there is a distinct lack of
emission from gas in the innermost regions where sublima-
tion is expected, and which may be due to lower disk surface
density, resulting from enhanced viscosity and inward drift
of gas (Melis et al. 2010). Corroborating this picture, sub-
limation of dust at 0959 is expected only within 15 stellar
radii, and hence cannot account for emitting gas out to 100
radii.
It is uncertain if the current disk at 0959 and other gas-
dust disk hosts are recently created or extant disks that have
been impacted by a additional body or material. Regardless,
the presence of detectable gas is almost certainly related to
the destruction of solid material.
3.1.21221
The IRAC images of this star reveal a point-like source
neighboring the white dwarf. Both sources were simultane-
ously modeled and deconvolved with PSF-fitting photome-
try. The neighbor is faintly but clearly visible in the UKIDSS
H-band image, and is identified as a separate source in that
survey. The LIRIS Ks-band images place it 1.′′8 distant dis-
tant at position angle 256◦. Together with its IRAC fluxes,
the JHK photometry of the neighbor yields colors inconsis-
tent with a (substellar) companion (Patten et al. 2006). Al-
though relatively faint, it appears to be diffuse and extended
in the ground-based images and is thus a background galaxy.
Both the UKIDSS and LIRIS photometry of the white dwarf
reveal strong infrared excess at K, with the latter dataset
confidently free of photometric contamination by the neigh-
bor, in agreement with the excess determined from IRAC
photometry. No Caii lines are detected in the low S/N red
ISIS spectrum.
3.1.31557
The mass of the white dwarf in 1557 is significantly below
the average mass of field white dwarfs (Liebert et al. 2005).
While the fit to the SDSS spectrum is consistent with a low-
mass, carbon-oxygen core, the WHT data are more sugges-
tive of a helium core. The SDSS data are of relatively poor
quality at S/N ≈ 10, while the WHT spectrum has S/N
≈ 30. Table 1 lists the weighted average of these two solu-
tions, favoring the low mass interpretation. If a low surface
gravity is corroborated in future observations, this may sig-
nify an innermost planet that was consumed during the first
ascent giant phase, ejecting sufficient envelope mass to pre-
vent the onset of helium ignition (Nelemans & Tauris 1998).
This picture is consistent with the evidence presented here
for a remnant planetary system. As in the case of 1221, no
Caii emission is seen in the poor quality ISIS data.
3.2Unusual Near-Infrared Colors for 1159
This DQ peculiar star (LP494-12, WD1156+132) was se-
lected on the basis of independent JHK photometry that
revealed a 0.3mag excess at K band (Farihi 2009) relative to
model expectations for a 10000K, helium atmosphere white
dwarf (Holberg & Bergeron 2006; Fontaine et al. 2001); its
UKIDSS photometry independently corroborates this result.
Due to the C2absorption features at blue-green wavelengths,
a pure helium atmospheric model was fitted only to the
izJH photometric data, as shown in Figure 6. The IRAC
fluxes reveal only a Rayleigh-Jeans tail to the spectral en-
ergy distribution, but at a shifted, lower temperature rela-
tive to the 0.6 − 1.5µm continuum. As found for another
DQp star (LHS2293; Farihi 2009), this apparent excess is
likely a re-distribution of emergent flux due to the highly
absorbed regions within the atmospheric C2 bands. Thus
the infrared excess only exists relative to a pure helium at-
mosphere, and is likely due to a distinct, absorption-induced
shape in the energy distribution of the star.
c ? 0000 RAS, MNRAS 000, 000–000

Page 8

8J. Farihi et al.
Figure 6. Spectral energy distribution of the DQ peculiar star
LP494-12, which has H − K > 0.25mag in both UKIDSS and
independently obtained photometry. Fitted to the izJH photom-
etry is a 9500K stellar model for a pure helium atmosphere white
dwarf (see §3.2).
3.3A Likely Ultracool Companion at 1247
The infrared excess observed at this star is distinct, as the
3.6µm flux excess is strong, but the 4.5µm flux is consis-
tent with a Rayleigh-Jeans type slope (see Figure 2). At first
glance this may seem puzzling, but most M dwarfs approach
this spectral behavior at IRAC wavelengths (Patten et al.
2006; Cushing et al. 2005). Neither the UKIDSS nor the
IRAC images indicate an additional source (i.e., a spatially
detectable companion), implying any secondary would have
a projected separation smaller than around 70AU, and es-
sentially ruling out a background object as the cause of the
observed emission. While the infrared excess is solid, and
possibly apparent at J band, the S/N of the UKIDSS pho-
tometry is below 10 at H and K, making it hard to esti-
mate a companion spectral type based on these data. As-
suming the observed excess emanates from a companion at
the correct distance, its 2.2, 3.6, and 4.5µm absolute bright-
nesses are 11.2, 10.0, and 9.8mag. Comparing these values
to the empirical absolute magnitudes of ultracool dwarfs at
these wavelengths yields an approximate spectral type of
L0 ± 1 (Leggett et al. 2010; Patten et al. 2006; Vrba et al.
2004; Dahn et al. 2002; Kirkpatrick & McCarthy 1994) for
a low mass companion. Superior JHK photometry would
tighten up this estimate, and possibly indicate a new L-type
companion to a white dwarf, which are rare (Steele et al.
2011; Farihi et al. 2005). Radial velocity monitoring of the
white dwarf has the potential to detect variability if the sys-
tem is a close binary, whereas high angular resolution imag-
ing may be required if the system has a moderately wide
separation.
3.4 A Likely Background Galaxy at 1320
The IRAC images of this star have full widths at half max-
imum that are inconsistent with a single point source; they
are enlarged by about 20% in both channels and there is
modest elongation in the 3.6µm image roughly along P.A.
45◦. Nothing is seen in the UKIDSS images, but a second
source appears faintly in the LIRIS H and Ks-band images
at a position consistent with the IRAC data. The image of
this additional source is diffuse and extended and almost
certainly a background galaxy. Moreover, there is an off-
set between the image centroids of the science target in the
LIRIS and IRAC images, indicating that the IRAC source
(and hence flux) is coincident with the additional source im-
aged with LIRIS, and not the white dwarf. The flux from
this likely galaxy falls within 1.′′0 of the white dwarf, and
contaminates all available infrared photometry. Even with
PSF-fitting photometry on the LIRIS images, the resulting
colors are slightly too red for a 20200K white dwarf. Given
the overlapping proximity of this source to the white dwarf,
the infrared excess is probably due entirely to this back-
ground object.
3.5No Near-Infrared Excess for 1506
Somewhat ironically, the only star without a corroborated
infrared excess via IRAC photometry is the brightest source
in the sample, and thus the star with the highest S/N
UKIDSS photometry. However, this star was selected for the
Spitzer program when the evaluation criteria were still tied
to UKIDSS DR4, and has since dropped out of the selec-
tion based on the release of DR8 and an improved analysis
(Girven et al. 2011). The model for 1506 shown in Figure 2
shows that an excess is neither observed with Spitzer, nor
in K-band. While this white dwarf was observed with ISIS,
confirming its nature as a DA star, it is hence dropped from
further analysis.
4 DISCUSSION AND CONCLUSIONS
4.1An Unbiased Dust Disk Frequency
What makes this study so far unique is that previous
IRAC programs have either targeted known metal-rich white
dwarfs, or near-infrared bright white dwarfs. While the first
approach has been highly successful and accounts for the
bulk of white dwarf disk discoveries to date (Farihi et al.
2010b, 2009; Jura et al. 2007), two drawbacks exist. First,
it requires 8m class telescope time using high-resolution
spectroscopy to efficiently identify (weak) metal lines in
white dwarfs (Koester et al. 2005; Zuckerman et al. 2003),
and second, only about 20% of metal-contaminated white
dwarfs have disks (Farihi et al. 2009). The second method,
employed by the Mullally et al. (2007) survey of more than
130 white dwarfs with Ks < 15mag, was unbiased by the
presence or absence of atmospheric metals, but only two
stars with dust were found (von Hippel et al. 2007), and is
hence only 1.5% efficient by number of targets. Here, five
candidate DA white dwarfs with K-band excess have been
selected with a success rate of 60% for circumstellar dust
and 80% for confirmed infrared excesses physically associ-
ated with the system. One of five targets has an infrared
excess probably due to a background galaxy.
Based on the candidate selection criteria and numbers
given in Table 6 of Girven et al. (2011), of 1884 candidate
and confirmed DA white dwarfs with the necessary UKIDSS
detections, there are 12 stars whose data are consistent with
disk-like, K-band excesses. Several years of Spitzer studies
c ? 0000 RAS, MNRAS 000, 000–000

Page 9

White Dwarfs with K-Band Excess9
have shown that only 50% of dusty white dwarfs have such
K-band excesses (Farihi et al. 2010b), implying that a de-
cent estimate of the disk fraction among DA white dwarfs
is at least 1.2%. However, as seen in this study, in some
cases the K-band data can be the result of neighboring back-
ground objects, as well as real companions such as low mass
stars or brown dwarfs. Thus a more realistic estimate is 3/5
of this number, implying at least 0.8% of DA (and presum-
ably all) white dwarfs have dust disks (and atmospheric met-
als), consistent with previous estimates (Farihi et al. 2009).
These results only apply to stars with cooling ages less than
1Gyr, as this corresponds to the approximate cutoff in ef-
fective temperature (8000K; Fontaine et al. 2001) for the
DA color selection. However, Spitzer studies have shown
a distinct lack of dust disks at older and cooler metal-
contaminated stars; only G166-58 has an (anomalous) in-
frared excess and a cooling age beyond 1Gyr (Farihi et al.
2008b).
4.2A Lack of K-Band Emitting Dust Disks at
Warm to Hot White Dwarfs
Interestingly, the DA selection criteria is quite sensitive to
warm to hot, hydrogen-rich white dwarfs, as the Balmer
decrement gives these stars unique colors in that wavelength
region (Girven et al. 2011). However, there are no candi-
dates for dust among thousands of stars(!). Either dust disks
at white dwarfs with Teff > 25000K exist yet rarely ex-
hibit K-band excess or dust at such stars is itself rare. The
latter possibility is somewhat contradictory based on the
paradigm of planetary systems that are dynamically rejuve-
nated at the onset of the post-main sequence (Bonsor et al.
2011; Bonsor & Wyatt 2010; Debes & Sigurdsson 2002). If
this apparent dearth of warmer systems at K-band is real,
it would likely require a physical mechanism to preclude
such emission (such as rapid removal of warm grains, or
long timescales for dust disk spreading toward the star).
Farihi (2011) plot the fractional infrared luminosity of
all known white dwarfs with dust circa mid-2010 as a func-
tion of cooling age, revealing a potential trend of increasing
dust emission towards cooler stars. This could be a natural
consequence of the potential for cooler white dwarfs to host
wider disks: as a star cools and its luminosity decreases, dust
grains may persist closer to the stellar surface prior to sub-
limation. Thus, any disk extending inward from the Roche
limit has the potential to be wider at cooler white dwarfs.
Conversely, at increasingly warmer stars, the inner dust disk
edge will be physically further from the star as the radius
at which grains are rapidly sublimated increases (see Table
4), eventually exceeding the Roche limit for large solid bod-
ies (von Hippel et al. 2007). While it is uncertain at which
precise effective temperature an evolving dust disk might be
completely sublimated, the results of this work clearly show
that disk emission analogous to G29-38 is rare among warm
and hot white dwarfs. More sensitive surveys with Spitzer
and WISE are needed to address this matter more compre-
hensively.
ACKNOWLEDGMENTS
The authors thank the referee Sandy Leggett for feedback
which improved the quality and clarity of the manuscript.
This work is based in part on observations made with the
Spitzer Space Telescope, which is operated by the Jet Propul-
sion Laboratory, California Institute of Technology under a
contract with NASA. Some data presented herein are part
of the Sloan Digital Sky Survey, which is managed by the
Astrophysical Research Consortium for the Participating In-
stitutions (http://www.sdss.org/), and the UKIRT Infrared
Deep Sky Survey.
REFERENCES
Abazajian, K. N., et al. 2009, ApJS, 182, 543
Bonsor, A., Wyatt, M. 2010, MNRAS, 409, 1631
Bonsor, A., Mustill, A. J. Wyatt, M. 2011, MNRAS, 414,
930
Brinkworth, C. S., G¨ ansicke, B. T., Marsh, T. R., Hoard,
D. W., Tappert, C. 2009, ApJ, 696, 1402
Cushing, M. C., Rayner, J. T., Vacca, W. D. 2005, ApJ,
623, 1115
Davidsson, B. J. R. 1999, Icarus, 142, 525
Dahn, C. C., et al. 2002, AJ, 124, 1170
Debes J. H., Sigurdsson S. 2002, ApJ, 572, 556
D’Odorico, S. et al. 2006, SPIE, 6269, 626933
Eisenstein D. J., et al. 2006, AJ, 132, 676
Farihi, J. 2009, MNRAS, 398, 2091
Farihi, J. 2011, in White Dwarf Atmospheres and Circum-
stellar Environments, ed. D. W. Hoard (Berlin: Wiley-
VCH)
Farihi, J., Becklin, E. E., Zuckerman, B. 2005, ApJS, 161,
394
Farihi, J., Becklin, E. E., Zuckerman, B. 2008a, ApJ, 681,
1470
Farihi J., Barstow M. A., Redfield S., Dufour P., Hambly
N. C. 2010a, MNRAS, 404, 2123
Farihi, J., Jura, M., Lee, J. E., Zuckerman, B. 2010b, ApJ,
714, 1386
Farihi, J., Jura, M., Zuckerman, B. 2009, ApJ, 694, 805
Farihi, J., Zuckerman, B., Becklin, E. E. 2008b, ApJ, 674,
431
Fazio, G. G., et al. 2004, ApJS, 154, 10
Fontaine, G., Brassard, P., Bergeron, P. 2001, PASP, 113,
409
G¨ ansicke, B. T., Koester, D., Marsh, T. R., Rebassa-
Mansergas, A., Southworth J. 2008, MNRAS, 391, L103
G¨ ansicke, B. T., Marsh, T. R., Southworth, J. 2007, MN-
RAS, 380, L35
G¨ ansicke, B. T., Marsh, T. R., Southworth, J., Rebassa-
Mansergas, A. 2006, Science, 314, 1908
Girven, J., G¨ ansicke, B. T., Steeghs, D., Koester, D. 2011,
MNRAS, 417, 1210
Holberg, J. B., Bergeron, P. 2006, AJ,132, 1221
Hubeny, I., Lanz, T. 1995, ApJ, 439, 905
Hunt, L. K., Mannucci, F., Testi, L., Migliorini, S., Stanga,
R. M., Baffa, C., Lisi, F., Vanzi, L. 1998, AJ, 115, 2594
Jura, M. 2003, ApJ, 584, L91
Jura, M., Farihi, J., Zuckerman, B. 2007, ApJ, 663, 1285
Jura, M., Farihi, J., Zuckerman, B. 2009, AJ, 137, 3191
c ? 0000 RAS, MNRAS 000, 000–000

Page 10

10J. Farihi et al.
Kilic, M., von Hippel, T., Leggett, S. K., Winget, D. E.
2006, ApJ, 646, 474
Kirkpatrick, J. D., McCarthy, D. W. 1994, AJ, 107, 333
Koester D. 2010, In Memorie della Societa Astronomica
Italiana, 81, 921
Koester, D., Rollenhagen, K., Napiwotzki, R., Voss, B.,
Christlieb, N., Homeier, D., Reimers, D. 2005a, A&A, 432,
1025
Lanz, T., Hubeny, I. 1995, ApJ, 439, 875
Lawrence, A., et al. 2007, MNRAS, 379, 1599
Leggett, S. K., et al. 2010, ApJ, 710, 1627
Liebert J., Bergeron P., Holberg J. B. 2005, ApJS, 156, 47
Manchado, A., et al. 1998, SPIE, 3354, 448
Marsh, T. R. 1989, PASP, 101, 1032
Martin, D. C., et al. 2005, ApJ, 619, L1
McCook G. P., Sion E. M. 1999, ApJS, 121, 1
Melis, C., Jura, M., Albert, L., Klein, B., Zuckerman, B.
2010, ApJ, 722, 1078
Melis, C., Farihi, J., Dufuour, P., Zuckerman, B., Bur-
gasser, A. J., Bergeron, P., Bochanski, J., Simcoe, R. 2011,
ApJ, 732, 90
Mullally, F., Kilic, M., Reach, W. T., Kuchner, M., von
Hippel, T., Burrows, A., Winget, D. E. 2007, ApJS, 171,
206
Nelemans, G.,Tauris, T. M. 1998, A&A, 335, L85
Patten, B. M., et al. 2006, ApJ, 651, 502
Probst, R. 1981, Ph.D. Thesis, University of Virginia
Probst, R. 1983, ApJS, 53, 335
Probst, R. G., O’Connell, R. W. 1982, ApJ, 252, L69
Rafikov, R. R. 2011, MNRAS, 416, L55
Rebassa-Mansergas, A., G¨ ansicke, B. T., Rodr´ ıguez-Gil, P.,
Schreiber, M. R., Koester, D. 2007, MNRAS, 382, 1377
Steele, P. R., Burleigh, M. R., Dobbie, P. D., Jameson, R.
F., Barstow, M. A., Satterthwaite, R. P. 2011, MNRAS,
416, 2768
Vennes, S., Kawka, A., N´ emeth, P. 2011, MNRAS, 413,
2545
von Hippel T., Kuchner M. J., Kilic M., Mullally F., Reach
W. T. 2007, ApJ, 662, 544
Vrba, F., et al. 2004, AJ, 127, 2948
Wang, Z., Chakrabarty, D., Kaplan, D. L. 2006, Nature,
440, 772
Werner, M. W., et al. 2004, ApJS, 154, 1
Zuckerman, B., Becklin, E. E. 1992, ApJ, 386, 260
Zuckerman, B., Koester, D., Reid, I. N., H¨ unsch, M. 2003,
ApJ, 596, 477
c ? 0000 RAS, MNRAS 000, 000–000