Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~pgreen/Papers/irdas00.ps Äàòà èçìåíåíèÿ: Tue Feb 12 19:29:43 2002 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 00:33:11 2012 Êîäèðîâêà: koi8-r Ïîèñêîâûå ñëîâà: m 35

THE ASTROPHYSICAL JOURNAL, 540 : 992õ1004, 2000 September 10
2000. The American Astronomical Society. All rights reserved. Printed in U.S.A.
(
COOL COMPANIONS TO HOT WHITE DWARFS
PAUL J. GREEN1
Harvard­Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 ; pgreen=cfa.harvard.edu
BABAR ALI
Infrared Processing and Analysis Center, Caltech, Mail Code 100­22, 770 S. Wilson Avenue, Pasadena, CA 91125 ; babar=ipac.caltech.edu
AND
R. NAPIWOTZKI
Dr. Remeis­Sternwarte, Sternwartstr. 7, 96049 Bamberg, Germany ; ai23=sternwarte.uni­erlangen.de
Received 2000 February 5 ; accepted 2000 April 3
ABSTRACT
Low­mass companions to high­mass stars are difficult to detect, which is partly why the binary frac­
tion for high­mass stars is still poorly constrained. Low­mass companions can be detected more easily
however, once high­mass stars turn into white dwarfs. These systems are also interesting as the progeni­
tors of a variety of intensely studied interacting binary systems, like novae, CVs, symbiotics, Ba and CH
giants, Feige 24­type systems, and dwarf carbon stars. We describe a near­IR photometric search for cool
red dwarf companions to hot white dwarfs (WDs). IR photometry o+ers a sensitive test for low­mass
main­sequence (MS) companions. Our sample of EUV­detected WDs o+ers several advantages over pre­
vious (largely proper motionõselected) WD samples : (1) the high WD temperatures (24,000 \ T eff \
70,000 K) insure excellent IR ÿux contrast with cool dwarfs ; (2) the range of evolutionary parameter
space occupied by the WDs is considerably narrowed ; and (3) the random e+ects of the intervening ISM
provide a complete but reasonably sized sample. While some composite systems have been found opti­
cally among WDs detected in recent EUV All­Sky Surveys, we develop an IR technique that probes
farther down the main sequence, detecting yet more companions. We use detailed DA model atmosphere
ïts to optical spectra to predict K magnitudes and distances, against which we contrast our near­IR
observations. Our photometric survey reveals 10 DAs with a signiïcant excess in both J and K. Half are
newly discovered and are most likely previously unrecognized binary systems. Neither the frequency of
infrared excess nor the mass estimate of the red dwarf companion correlate with white dwarf mass, as
might be expected if either the EUV detectability or mass of the white dwarfs were signiïcantly a+ected
by a companion. Infrared spectra of these systems should help to determine the mass and spectral type
of the cool companions presumably causing the IR excess, leading to better estimates of the mass ratio
distribution in binaries. Counting previously known binaries, and resolved pairs, we ïnd the total binary
fraction of the sample is at least a third. Since most WD progenitors had initial masses º2 we thus
M _ ,
provide a photometric measure of the binary fraction of high­mass stars that would be difficult to
perform in high­mass main­sequence stars. We estimate that 90% of the companions are of type K or
later.
Subject headings : binaries : close õ stars : atmospheres õ stars : evolution õ white dwarfs
1. INTRODUCTION
EUV­detected stars have revealed in their UV spectra the
presence of about 15 hot white dwarf (WD) companions to
bright stars in noninteracting binary systems (e.g., Burleigh,
Barstow, & Fleming 1997). These WDs are hidden at
optical wavelengths, because of their close proximity to
much more luminous companionsõmain­sequence stars of
spectral type K or earlier, or evolved stars.
Many interacting binary systems where the WD is the
primary (i.e., optically brightest) star have also been found
among EUV­detected systems (e.g., six close, interacting
white dwarf/red dwarf binaries by Vennes & Thorstensen
1994). Optical or ultraviolet spectral observations are most
commonly used to detect companions to WD primaries, by
searching for (1) the presence of narrow Balmer line emis­
sion overlying the broad smooth Balmer absorption of the
WD, (2) a composite WD]main­sequence spectrum, or (3)
radial velocity (RV) variations. However, only WDs with
1 Visiting Astronomer, Kitt Peak National Observatory, National
Optical Astronomy Observatories.
very close, or intrinsically active, companions will be found
by method (1). For hot WD systems, composite spectra (2)
are only expected to be visible if the companionîs spectral
type is early. RV variations (3) require multiple obser­
vations at high spectral resolution, and detection strongly
favors close and/or massive companions.
Such discoveries have been strongly dominated by these
selection e+ects, with companions biased to earlier types
than predicted by the simulations of deKool & Ritter (1993)
and others. Scaling from the deKool & Ritter results,
Vennes & Thorstensen (1994) estimate that ```` at least twice
as many close binary systems remain to be identiïed from
EUV surveys, most of them with a low mass secondary.îî
The resulting sample of binaries known to date therefore
must diverge strongly from the intrinsic distribution, in
overall normalization, as well as in mass and spectral type
of the main­sequence companions.
Large proper motionõselected samples (including those
observed in the IR e.g., Probst 1983a, 1983b and Zucker­
man & Becklin 1992), are kinematically biased against lumi­
nous (statistically more distant) young, hot WDs and span a
bewildering range of temperatures, ages, and metallicities.
992

COOL COMPANIONS TO HOT WHITE DWARFS 993
Our EUV­selected sample is instead dominated by freshly
minted disk DAs, providing a narrower slice of parameter
space for stellar evolution (in the wide systems) or for
dynamical evolution (in any new close systems). Objects
known or proposed to contain white dwarf (WD) stars in
currently or previously interacting binary systems consti­
tute diverse laboratories for the study of stellar evolution. A
partial list includes novae, cataclysmic variables, symbiotic
stars, Ba and CH giants, Feige 24­type systems and dwarf
carbon stars (Green &Margon 1994).
We describe here a near­IR photometric survey for low­
mass companions to hot white dwarfs (WDs). By investigat­
ing only EUV­detected WDs, we obtain a very reasonably
sized but complete sample of young WDs, next to which
very late­type dwarf companions can be detected in the
near­infrared by searching for a K excess. Many hot white
dwarfs K ; Finley et al. 1993) have been
(T eff [ 24,000
detected in the recent EUV all­sky surveys. EUV detection
of these hot WDs depends primarily on their temperature,
distance, and the intervening Galactic ISM. Our sample of
EUV WDs (whose selection we deïne below) o+ers excel­
lent ÿux contrast in the IR relative to optical ; cool compan­
ions will almost always be brighter in the K band than the
hot WDs. IR photometry of hot white dwarfs is nearly non­
existent in the literature.
To know what K magnitude to expect for the WDs, we
beneït from constraints on log g, radius, and derivable
T eff
from optical spectra for the WDs in our sample, using
NLTE model atmosphere ïts (Napiwotzki, Green, & Sa+er
1999, hereafter NGS99). The resulting predictions for K
magnitudes allow a direct search for any IR excess from a
cool companion. In some cases, IR colors also provide a
preliminary spectral type for the companion.
2. SAMPLE SELECTION CRITERIA
We chose to limit our initial sample to DAs, for which
model atmospheres provide the best temperature and mass
constraints. We start with 73 known DA white dwarfs in the
EUV E bright source list (Malina et al. 1994). We exclude
sources at low Galactic latitudes (o b o \ 15) to avoid crowd­
ing problems, and in the South (d \[20), yielding a list of
27 DAs.2 A similar procedure for DAs listed in the ROSAT
Wide Field Camera survey Bright Source Catalogue
(Pounds et al. 1993) culls an additional 30 objects. Table 1
lists the full sample of 57, including published data used in
this paper. Model atmosphere ït parameters for DAs that
are not available in NGS99 are taken from Vennes et al.
(1997b) or other references listed in Table 1.
Of the 57 sample objects, we present new IR photometry
for 47. We did not obtain new measurements for 10 stars
with published sensitive optical spectrophotometry and IR
photometry and/or known binaries from optical studies
(Feige 24, HZ 43, GD 50, V471 Tau (Vennes, Christian, &
Thorstensen 1998), PG 0824]289 (Heber et al. 1993), HD
74389B (Liebert, Bergeron, & Sa+er 1990), REJ 1016[052,
REJ 1426]500 (Vennes et al. 1997a, 1997b), and IK Peg
(Wonnacott, Kellett, & Stickland 1993). Due to observing
constraints (a combination of weather, poor seeing and faint
objects, or celestial placement of objects) no photometry
was obtained for 0138]252 (PG 0136]251).
2 Emission from 40 Eri was resolved with the Einstein HRI, and likely
arise mostly from 40 Eri C, the dMe ÿare star (Cash et al. 1980), not the
white dwarf 40 Eri B.
On the other hand, we report new IR data for four
objects that fell outside the strict sample deïnitions just
outlined. These include the known binaries REJ 1036]460
(PG 1033]464), REJ 1629]780 as well as REJ 0148[253
(GD 1401, also a known binary), and REJ 0457[280
(MCT 0455[2812), which are outside the sample decli­
nation limits.
We note that since the initial sample selection, several
relevant discoveries pertaining to sample objects have been
made. REJ 0134[160 (GD984) has central Balmer emis­
sion components produced by a dMe companion. RE
1440]750 turns out to be a magnetic DA (Dreizler et al.
1994).
3. OBSERVATIONS AND DATA REDUCTION
We obtained J­ and K­band photometry of the sample
objects at the Kitt Peak National Observatory3 2.1 m tele­
scope using the Infrared Imager (IRIM ; Fowler et al. 1988).
The 256 ] 256 IRIM, completed in 1992, is a HgCdTe
NICMOS3 array, which at the f/15 focus of the 2.1 m yields
a 280@@ ] 280@@ ïeld of view (plate scale of pixel~1).
1A. 09
Tables 2 and 3 summarize our observing procedure.
Table 2 lists the individual observing runs dedicated to this
project. Column (1) identiïes the run. The universal time
(UT) date of the run and the photometric standard stars
used during each night are shown in columns (2) and (3).
Column (4) identiïes the references containing infrared (J­
and K­band) magnitudes for the standard stars, and
column (5) contains a short overall subjective description
for the night conditions. Table 3 summarizes our observing
strategy. Column (1) identiïes the program object (given as
an REJ nnnn]mmm). Column (2) lists the nights on which
the target was observed, and column (3) any exceptions to
our standard observing procedure of six dithered images
per target, each image consisting of six co­added 10 s expo­
sures. The exceptions are noted in column (3) of Table 3.
The number of exposures depended on the faintness of the
object. To minimize any hysteresis e+ects in pixel sensitivity,
during the observations, we varied the direction and dis­
tance between the position o+sets, and never placed a target
at the center of the array. At least once during each night,
10õ20 (co­added) dark frames of 10 s. duration were also
obtained. These are median­combined to form a single dark
frame free of cosmic ray hits. The atmospheric seeing (and
hence the FWHM of stars) varied from 1.4õ2.5 pixels during
the observations.
We reduced our image data using the Image Reduction
and Analysis Facility (IRAF) software. For each ïlter
separately, the reduction proceeds as follows. All images are
corrected for any detector nonlinearities (R. Joyce 1999,
private communication). Next, any bad pixels (as deïned by
the bad pixel mask ; see below) are replaced with the average
of the linearly interpolated values from adjacent pixels in
both dimensions of the image. Each image is then sky sub­
tracted and divided by the ÿat ïeld.
The sky frame is produced by median­combining all
exposures of the star. This also serves to subtract the dark
current, since the dark current is included in the sky frames.
The ÿat­ïeld image is created as a median­combined frame
from a large (º30) number of individual images. Individual
3 The Kitt Peak National Observatory is operated by the Association
of Universities for Research in Astronomy, Inc. (AURA), under coopera­
tive agreement with the National Science Foundation.

TABLE 1
SAMPLE OF EUV­SELECTED DA WHITE DWARFS
REJ Other Names V T eff log g M/M _ References Notes
0007]331a . . . . . . GD 2 13.85 45653 7.77 0.57 1, 2
0134[160a . . . . . . GD 984, PHL 1043 13.96 43722 7.70 0.54 1, 2, 9 DA ] dM
0138]252 . . . . . . PG 0136]251 15.87 38964 9.00 1.22 1, 1 Peculiar
0148[253a . . . . . . GD 1401 14.69 25707 7.70 0.50 1, 2, 6 NS, DA ] dM
0235]034 . . . . . . Feige 24 12.56 62947 7.53 0.54 1, 1, 13 DAe ] dM
0237[122a . . . . . . PHL 1400 14.92 31570 8.38 0.85 1, 2
0348[005 . . . . . . GD 50 14.04 38881 8.97 1.17 1, 2
0350]171 . . . . . . V471 Tau 13.65 34200 8.80 1.12 1, 1 DA ] K0
0427]740a . . . . . . 15.9 47549 7.84 0.61 3, 2 Visual pair, 8A
0443[034a . . . . . . 16.9 65140 9.12 1.29 3, 4
0457[280a . . . . . . MCT 0455[2812 13.95 50061 7.63 0.53 1, 2 NS
0512[004a . . . . . . 14.2 31333 7.34 0.45 3, 2
0521[102a . . . . . . 15.81 32775 8.53 0.94 1, 2
0827]288 . . . . . . PG 0824]289 14.22 51934 8.00 0.70 1, 1, 12 DA ] dC ] M3V
0841]032a . . . . . . 14.47 37687 7.68 0.52 1, 2
0845]488 . . . . . . HD74389B 15.5 : 39500 8.04 0.65 7, 5 DA ] A2V
0902[040a . . . . . . 13.19 22285 7.76 0.51 1, 2
0907]505a . . . . . . PG 0904]511 16.54 31791 8.04 0.66 1, 2
0916[194a . . . . . . 17.3 56400 9.12 1.29 3, 4 Visual pair, 6A. 4
0940]502a . . . . . . PG 0937]506 16.0 35511 7.62 0.51 3, 2
0957]852a . . . . . . 15.4 50205 8.29 0.82 3, 2
1016[052 . . . . . . 14.21 53827 8.08 0.65 1, 1, 4 DA ] dM
1019[140a . . . . . . 14.93 31100 7.85 0.57 1, 2
1029]450a . . . . . . PG 1026]454 16.13 35017 7.64 0.51 1, 2
1032]532a . . . . . . 14.45 42785 7.89 0.62 1, 2
1033[114 . . . . . . G 162[66, LTT 3870 13.01 23741 7.79 0.52 1, 2
1036]460a . . . . . . GD 123 14.34 28720 7.95 0.61 1, 2, 10 DA ] K
1043]490a . . . . . . 16.23 40461 7.87 0.60 1, 2, 14 Visual pair, 8A
1043]445a . . . . . . PG 1040]451 16.94 48000 7.98 0.63 3, 4
1044]574a . . . . . . PG 1041]580 14.64 29878 7.73 0.52 1, 2
1100]713a . . . . . . PG 1057]719 14.68 40328 7.75 0.55 1, 2
1112]240a . . . . . . Ton 61 15.77 39281 7.71 0.53 1, 2
1122]434a . . . . . . PG 1120]439 15.8 26152 8.23 0.75 3, 2
1126]183a . . . . . . PG 1123]189 14.13 52747 7.69 0.56 1, 2, 10 DA ] dM
1128[025a . . . . . . PG 1125[026 15.73 30227 8.15 0.71 1, 2
1148]183a . . . . . . PG 1145]188 14.33 24717 7.84 0.55 1, 2
1235]233a . . . . . . PG 1232]238 17.4 45639 7.77 0.57 3, 2
1236]475a . . . . . . PG 1234]482 14.38 55570 7.57 0.53 1, 1 Reclassiïed DB
1257]220a . . . . . . GD 153 13.38 38324 7.71 0.53 2
1316]295 . . . . . . HZ 43 12.99 49000 7.70 0.56 1, 8 DA ] dM
1336]694a . . . . . . PG 1335]701 15.4 29067 8.27 0.78 3, 2
1426]500 . . . . . . 13.6 29800 7.98 0.60 3, 4, 6 DA ] MVe
1431]370a . . . . . . GD 336 15.27 33988 7.86 0.58 1, 2
1440]750a . . . . . . 15.32 42400 8.54 0.96 3, 4
1446]632a . . . . . . 16.4 37340 7.72 0.53 3, 2
1629]780a . . . . . . 13.03 39882 7.88 0.61 1, 2, 11 DA ] M4V
1638]350a . . . . . . PG 1636]351 14.83 34879 7.90 0.61 1, 2
1643]411a . . . . . . PG 1642]414 16.2 28240 8.15 0.71 3, 2
1650]403a . . . . . . 15.83 37698 7.90 0.61 1, 2
1659]440a . . . . . . PG 1658]440 14.62 30510 9.36 1.31 3, 5 Magnetic DA
1711]664a . . . . . . 17.10 47994 8.78 1.09 1, 2 Visual pair, 2A. 5
1726]583a . . . . . . PG 1725]586 15.45 52413 8.14 0.75 1, 2 Visual pair, 6A. 7
1800]683a . . . . . . KUV 18004]6836 14.74 43868 7.74 0.56 1, 2
1820]580a . . . . . . 13.95 43296 7.70 0.54 1, 2
1845]682a . . . . . . KUV 18453]6819 15.5 35599 8.15 0.72 3, 2
2116]735a . . . . . . KUV 21168]7338 15.0 49777 7.65 0.54 3, 2
2126]192 . . . . . . IK Peg 6.07 . . . . . . . . . . . DA ] A8m
2207]252 . . . . . . 14.58 26129 8.20 0.74 1, 2 Visual pair, 8A. 9
2312]104a . . . . . . GD 246 13.09 51950 7.76 0.59 1, 2
NOTES.õNS means not in official sample, but may have optical ïts in NGS99 and/or IR data. Visual pairs within 10A and
*R\ 3 mag are noted with separations.
a New IR photometry presented in this work.
REFERENCES.õFirst reference is for V magnitude, the second for log g, and the third if listed is for companionîs
T eff , M/M _ ;
spectral type and/or mass of previously known companion. If a multiple system has no third reference listed, it has several
references available in SIMBAD. (1) Marsh et al. 1997 ; (2) NGS99 ; (3) V from McCook & Sion 1999 (4) Vennes et al. 1997b (5)
Schmidt et al. 1992 ; (6) Finley et al. 1997 (7) SIMBAD ; (8) Napiwotski et al. 1993 (9) Vennes et al. 1998 (10) Green et al. 1986 ;
(11) et al. 1995 ; (12) Green &Margon 1994 (13) Vennes & Thorstensen 1994 ; (14) Schwartz et al. 1995.
Catala
‘ n

COOL COMPANIONS TO HOT WHITE DWARFS 995
TABLE 2
SUMMARY OF OBSERVING RUNS
Standard
Standard Stars Stars
ID UT Date Useda References Conditions
(1) (2) (3) (4) (5)
KP01 . . . . . . 1996 Jan 04 FS 6, 10, 19, LHS 254, LHS 325 1, 2 Clear or light cirrus.
KP02 . . . . . . 1996 Jan 05 FS 10, 19, LHS 191, LHS 2347 1, 2 Cirrus, but eventually very clear.
KP03 . . . . . . 1996 Jan 06 FS 2, 10, 19, LHS 2502 1, 2 Heavy cirrus.
KP04 . . . . . . 1996 Jan 07 FS 15, 10, 19, LHS 191 1, 2 High thin cirrus.
KP05 . . . . . . 1996 Jul 02 . . . . . . Cloudy, no data.
KP06 . . . . . . 1996 Jul 03 . . . . . . Cloudy, no data.
KP07 . . . . . . 1996 Jul 04 . . . . . . Cloudy, no data.
KP08 . . . . . . 1996 Jul 05 . . . . . . Cloudy, no data.
KP09 . . . . . . 1996 Jul 06 . . . . . . Cloudy, no data.
KP10 . . . . . . 1996 Oct 29 FS 1, 6, 9, 12, 15, 29, 31 1 Mostly clear.
KP11 . . . . . . 1996 Oct 30 FS 1, 10, 14, 29, GJ 1057, GJ 1002 1, 2 Photometric.
KP12 . . . . . . 1998 Jan 20 . . . . . . Cloudy.
KP13 . . . . . . 1998 Jan 21 . . . . . . Fog. Telescope closed.
KP14 . . . . . . 1998 Jan 22 FS 2, 6, 21, 24, 18, 19 1 Clear.
KP15 . . . . . . 1998 May 09 FS 21, 23, 24, 27, 29, 33 1 Clear.
KP16 . . . . . . 1998 May 10 FS 16, 20, 21, 23, 35 1 Mostly clear with moderate haze.
KP17 . . . . . . 1998 May 11 FS 19, 21, 27 1 Hazy
a FS \UKIRT faint star.
REFERENCES.õ(1) Casali &Hawarden 1992 ; (2) Leggett 1992.
images are median­subtracted to account for sky/
background variations. The ÿat­ïeld image is dark sub­
tracted and normalized. The ÿat­ïeld image is also used to
produce the bad pixel mask, as deïned by the histogram
distribution of pixel intensities. All pixels with intensity
values near or beyond the extreme values of the histogram
distribution are deïned to be bad. The bad pixel mask was
produced once and used throughout.
Since our sample is deïned to be at high Galactic latitude
and hence uncrowded, simple aperture photometry is ade­
FIG. 1.õSample DA white dwarf temperature from best­ït model
atmospheres of NGS99 plotted against our predicted (J[K) color for the
star assuming it is solitary. DA colors are nearly degenerate for tem­
peratures above about 35,000 K.
quate. Following the arguments presented in Howell (1989)
we used a 4.0 pixel diameter aperture to maximize the
signal­to­noise ratio (S/N). Instrumental magnitudes are
measured for each object using the IRAF task QPHOT
with the radius of the inner sky annulus set to 10 pixels and
the width of sky annulus set to 5 pixels.
For target stars REJ 0443[034, REJ 0907]505, REJ
0940]502, and REJ 1442]632, visual inspection failed to
reveal the object on individual images. In such cases, we
shift (by integer pixels) and combine all the images of the
ïeld to form an average mosaicked image. Individual
frames are median­subtracted prior to averaging to remove
any zero point o+sets due to varying sky/background con­
ditions. The pixel shifts are calculated from other nearby
stars in the ïeld. The photometry procedure for these mosa­
icked images is described in the Appendix.
We derive the zero point calibrations for the instrumental
magnitudes from photometric standard stars observed at a
variety of air masses each night. These standards stars were
mostly selected from the UKIRT faint standard list (Casali
& Hawarden 1992). The photometric standards for each
night are listed in Table 2.
We eliminated the most extreme outlying measurement
for an object if its elimination reveals it to be more than 5
times the root mean square (rms) of the resulting average
magnitude. The resulting relative photometric error p is
taken to be the rms of all measurements from each individ­
ual frame. To obtain the ïnal magnitude error, we add to p
the rms dispersion in the standard star zero points taken
over the night. The ïnal magnitudes for targets are the
mean values produced by averaging the photometry from
all nights on which the target was observed. If the target
was observed on more than two nights, we use the algo­
rithm described above to eliminate outlying measurements.
Table 4 lists for each object with new IR observations
both the observed and predicted magnitudes. Column (1)
lists the RE name, (2) the observed J magnitude, and (3)
error. Columns (4) and (5) list the K magnitude and error,

996 GREEN, ALI, & NAPIWOTZKI Vol. 540
TABLE 3
SUMMARY OF TARGET OBSERVATIONS
Target Observing Runs Notesa
0007]331 . . . . . . KP02, KP04, KP10
0134[160 . . . . . . KP02, KP04,
KP10 jn3
0148[253 . . . . . . KP02, KP03, KP04
0237[122 . . . . . . KP04,
KP10 kn9
0348[005 . . . . . . KP01, KP02, KP03, KP04,
KP11 jn3, kn3
0427]740 . . . . . . KP01, KP03, KP04,
KP14 kn9
0443[034 . . . . . . KP03,
KP11 jn9, kn9
0457[280 . . . . . . KP03, KP11
0512[004 . . . . . . KP01, KP03, KP04, KP10,
KP14 jn9, kn9
0521[102 . . . . . . KP01, KP03,
KP10, jn9, kn9
KP14 kn9
0827]284 . . . . . . KP11
0841]032 . . . . . . KP01, KP10, jn9, kn9
KP14 kn9
0902[040 . . . . . . KP01, KP04
KP10 kn9
0907]505 . . . . . . KP03, KP04, KP11, KP14, jn9, kn9
KP17 kn18
0916[194 . . . . . . KP02 jn9, kn9
0940]502 . . . . . . KP01, jn9, kn9
KP15, jc1, kc1
KP16, jc2, jn9, kc2, kn9
KP17 kc9, kn18
0957]852 . . . . . . KP02,
KP14, kn9
KP16, jc2, jn9, kc2, kn9
KP17 kc9, kn9
1019[140 . . . . . . KP02
1029]450 . . . . . . KP02
1032]532 . . . . . . KP01, KP04,
KP14 kn9
1033[114 . . . . . . KP01, KP02, KP03, KP04
1036]460 . . . . . . KP02
1040]451 . . . . . . KP15, jc1, jn12, kc1, kn12
KP16, jc2, jn9, kc2, kn9
KP17 kc12
1043]490 . . . . . . KP02,
KP14, kn9
KP15, jc1, jn9, kc2, kn9
1044]574 . . . . . . KP02
1100]713 . . . . . . KP02,
KP17 kn15
1112]240 . . . . . . KP16 jc2, jn9, kc2, kn9
1122]434 . . . . . . KP02,
KP16 jc2, kc2, kn9
1126]183 . . . . . . KP01, KP04
1128[025 . . . . . . KP02,
KP16 jc1, kc1
1148]183 . . . . . . KP01, KP03
1235]233 . . . . . . KP14, kn9
KP16, jc2, jn9, kc3, kn9
KP17, kc12, kn9
1234]482 . . . . . . KP14 kn9
1257]220 . . . . . . KP01, KP02, KP03
1336]694 . . . . . . KP03, KP04,
KP14, kn9
KP16 jc1, jn9, kc2, kn9
1431]370 . . . . . . KP03, KP04
TABLE 3õContinued
Target Observing Runs Notesa
1431]370 . . . KP10, kn9
KP16, kn9,jn9,jc1,kc2
KP17 kn9,kc12
1440]750 . . . KP14,
KP16, jc1, jn9, kc2, kn9
KP17 kc12, kn9
1446]632 . . . KP14,
KP16, jc2, jn9, kc3, kn15
KP17 kc12, kn9
1629]780 . . . KP15 jc1, kc1
1638]350 . . . KP15 jc1, kc1, kn9
1643]411 . . . KP15, jc1, jn9, kc2, kn9
KP16, jc2, jn9, kc3, kn15
KP17 kc12, kn9
1650]403 . . . KP15, jc1, jn9, kc1, kn9
KP16 jc2, jn9, kc2, kn9
1658]441 . . . KP14,
KP15, jc1, jn9, kc1, kn9
KP16, jc2, kn9, kc2, kn9
KP17 kc12, kn9
1711]664 . . . KP15 jc1, jn18, kc1, kn9
1726]583 . . . KP15 jc1, jn9, kc1, kn9
1800]683 . . . KP15, jc1, jn9, kc1, kn9
KP16, jc2, jn9, kc2, kn9
KP17, kc12, kn9
1820]580 . . . KP11,
KP15 jc1, jn9, kc1, kn9
1845]682 . . . KP11,
KP15 jc1, jn9, kc1, kn9
2116]735 . . . KP10, KP11,
KP15, jc1, jn9, kc1, kn9
KP16, jc2, jn9, kc3, kn9
KP17 kc12, kn9
2127[221 . . . KP10 jn9, kn9
2207]252 . . . KP15, jc1, jn9, kc1, kn9
KP17
2312]104 . . . KP02, KP03, KP04
2353[243 . . . KP10 jn9, kn9
NOTES.õ```` XcY îî denotes that in the X bandpass the number of
co­adds performed per exposure is Y (e.g., jc1 means 1 co­add per
exposure in J band). ```` XnY îî denotes that in the X bandpass the
number of combined dithered exposures is Y (e.g., kn9 means 9
dithered exposures combined in the K band).
a The number of 10 s co­adds per exposure in both J and K
bands is nominally six, and the number of dithered exposures is
also six, except as noted in the last column.
respectively, with the (J[K) color and error in columns (6)
and (7). The absolute magnitude from the best­ït spec­
M V
tral model is shown in column (8). We list in column (9) the
predicted (V [K) color based on synthetic photometry of
the best­ït DA models. Predicted near­IR magnitudes for
the DA WDs are listed in columns (10)õ(11), and notes in
column (12). We note that our predictions of near­IR colors
for these systems have very little scatter over the full sample
range, with mean (J[K)\[0.24^ 0.02, as can be seen
from Figure 1. DA colorsõeven (V [K)õare nearly degen­
erate for DA temperatures higher than about 35,000 K.
4. ANALYSIS
4.1. Predicting Near­Infrared Magnitudes and Colors
We use a grid of 184 NLTE model spectra for DA WDs,
with from 20,000 to 100,000 (in 16 steps, coarser toward
T eff

No. 2, 2000 COOL COMPANIONS TO HOT WHITE DWARFS 997
TABLE 4
OBSERVED AND PREDICTED MAGNITUDES AND COLORS OF DA WHITE DWARF STARS
OBSERVED PREDICTED
RE NAME J p J K p K (J[K) p (J~K) M V (V [K) p J p K p NOTES
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
0007]331 14.62 0.02 14.82 0.07 [0.21 0.07 9.06 [0.72 14.33 14.57
0134ò160 12.95 0.01 12.21 0.01 0.74 0.01 9.00 [0.72 14.43 14.68
0148ò253 12.33 0.01 11.56 0.04 0.76 0.04 9.98 [0.60 15.08 15.29
0237[122 15.74 0.06 15.93 0.12 [0.18 0.14 10.68 [0.68 15.36 15.60
0348[005 14.80 0.06 14.66 0.52 0.14 0.52 11.46 [0.71 14.50 14.75 1
0427+ 740 15.64 0.03 15.62 0.12 0.02 0.12 9.13 [0.73 16.38 16.63
0443[034 17.38 0.05 17.99 0.07 [0.61 0.08 11.23 [0.75 16.49 16.75
0457[280 14.62 0.02 14.80 0.03 [0.18 0.03 8.69 [0.73 14.43 14.68
0512[004 14.43 0.03 14.67 0.01 0.15 0.40 8.81 [0.67 14.63 14.87
0521[102 16.76 0.20 17.13 0.13 [0.37 0.24 10.87 [0.69 16.26 16.50
0550[240 16.93 0.10 17.28 0.41 [0.34 0.42 9.03 [0.73 17.18 17.43
0841]032 15.17 0.02 15.49 0.01 [0.33 0.02 9.17 [0.70 14.93 15.18
0902[040 13.70 0.01 13.90 0.05 [0.20 0.05 10.37 [0.53 13.54 13.72
0907]505 17.16 0.10 17.54 0.64 [0.39 0.65 10.09 [0.68 16.98 17.22
0916ò194 13.94 0.02 13.64 0.04 0.36 0.10 11.36 [0.74 17.78 18.04
0940]502 16.73 0.07 16.97 0.02 [0.24 0.07 9.14 [0.70 16.45 16.70
0957]852 16.49 0.08 16.72 0.09 [0.23 0.12 9.87 [0.73 15.88 16.13
1019[140 15.17 0.04 15.45 0.14 [0.28 0.14 9.82 [0.67 15.36 15.60
1029]450 16.92 0.09 17.54 0.70 [0.62 0.71 9.20 [0.69 16.58 16.82
1032]532 15.38 0.01 15.45 0.11 [0.07 0.11 9.35 [0.72 14.92 15.17
1033[114 14.80 0.06 14.66 0.52 0.14 0.52 10.31 [0.56 13.38 13.57
1036+ 460 12.62 0.04 11.89 0.09 0.72 0.10 9.95 [0.65 14.76 14.99
1043]490 16.86 0.02 16.95 0.22 [0.10 0.22 8.40 [0.72 16.71 16.95
1043]445 18.00 0.05 17.48 0.09 0.52 0.11 9.43 [0.73 17.42 17.67
1044]574 15.32 0.04 15.73 0.17 [0.42 0.18 9.72 [0.66 15.06 15.30
1100]713 15.51 0.10 15.78 0.17 [0.27 0.20 9.19 [0.71 15.14 15.39
1112]240 16.37 0.21 16.51 0.55 [0.14 0.59 9.17 [0.71 16.24 16.48
1122]434 16.12 0.05 16.39 0.34 [0.27 0.34 10.82 [0.61 16.20 16.41
1126+ 183 12.74 0.09 12.04 0.10 0.70 0.14 8.47 [0.73 14.61 14.86
1128[025 16.18 0.06 16.48 0.17 [0.30 0.18 10.38 [0.67 16.16 16.40
1148]183 14.96 0.06 15.22 0.20 [0.26 0.20 10.31 [0.58 14.71 14.91
1235]233 18.31 0.11 . . . [2.36 0.28 9.06 [0.72 17.88 18.12 2
1236]475 14.91 0.06 14.90 0.16 0.01 0.17 8.37 [0.73 14.86 15.11
1257]220 14.17 0.06 14.43 0.02 [0.26 0.05 9.21 [0.71 13.84 14.09
1336]694 15.85 0.02 16.12 0.04 [0.28 0.04 10.66 [0.65 15.82 16.05
1431]370 15.97 0.04 16.37 0.16 [0.40 0.17 9.65 [0.69 15.72 15.97
1440]750 17.21 0.11 16.64 0.12 0.57 0.16 11.39 [0.70 15.77 16.02
1446]632 16.77 0.11 17.77 0.07 [1.00 0.12 9.26 [0.70 16.86 17.10
1629+ 780 10.96 0.03 11.19 0.04 [0.23 0.05 9.42 [0.71 13.49 13.74
1638]350 15.55 0.11 15.57 0.64 [0.01 0.64 9.67 [0.70 15.28 15.53
1643]411 16.86 0.10 17.46 0.05 [0.61 0.11 10.53 [0.64 16.62 16.84
1650]403 16.60 0.08 16.86 1.05 [0.26 1.05 9.54 [0.71 16.29 16.54
1658]441 15.47 0.10 15.68 0.02 [0.21 0.10 12.70 [0.67 15.06 15.29
1711+ 664 15.02 0.01 14.19 0.18 0.83 0.18 10.84 [0.73 17.58 17.83
1726]583 16.40 0.26 16.17 1.51 0.23 1.53 9.56 [0.73 15.93 16.18
1800]683 15.54 0.04 15.87 0.04 [0.33 0.06 9.06 [0.72 15.21 15.46
1820+ 580 14.07 0.04 13.42 0.01 0.65 0.04 9.01 [0.72 14.42 14.67
1845+ 682 14.85 0.10 14.37 0.19 0.47 0.22 10.06 [0.70 15.95 16.20
2116]735 15.54 0.01 15.93 0.06 [0.39 0.06 8.74 [0.73 15.48 15.73
2207]252 14.34 0.08 14.06 0.22 0.28 0.23 10.77 [0.61 14.98 15.19 3
2312]104 13.81 0.04 14.06 0.01 [0.14 0.75 8.88 [0.73 13.57 13.82
NOTES.õObjects with signiïcant IR excess, are shown in boldface. (1) Faint IR standard star photometry from Casali &Hawarden
1992. (2) Object is barely discernible on K band image ; K photometry unreliable. (3) IR photometry known to include nearby
optically resolved object at 2A.2.
the high end) and log g, in 12 equal steps from 7.0 to 9.75.
We predict colors or magnitudes for each grid model by
convolving the model spectrum over the observed ïlter
bandpasses. In the visual, we use Mathews & Sandage V
ïlter curves and typical CCD quantum efficiency (QE)
curves, while in the near­IR, we use the actual NOAO ïlter
transmission curves for the J and K ïlters, along with the
IRIM QE curve. Zero points for all magnitudes are cali­
brated by convolution with the Vega spectrum of Hayes &
Latham (1975), renormalized (by a factor of 1.028) to zero
magnitude. As a check on our synthetic photometry, we
tabulated synthetic colors derived from the spectro­
photometric atlas of Bruzual, Persson, Gunn, & Stryker
(BPGS ; Gunn & Stryker 1983) which includes infrared (IR)
distributions from Strecker, Erickson, & Witteborn (1979).
Our synthetic colors agree with published colors for the
corresponding spectral types to better than about 0.05 mag
for blue spectral types (Zombeck 1990).

998 GREEN, ALI, & NAPIWOTZKI Vol. 540
We predict colors for each program WD star from its
best­ït gravity and temperature using a two­dimensional
linear interpolation between the colors derived for the
nearest grid models. Predicted colors for each program star
are combined with observed visual DA magnitudes, to
predict their J and K magnitudes, assuming that they are
isolated. A table of and (predicted J and K
J p K p
magnitudes) for DA stars are listed in Table 4 along with
the observed magnitudes.
4.2. Deriving Distances and Searching for IR Excess
Because each of the model spectra are presented in stellar
surface ÿuxes (in cgs units ergs cm~2 s~1 cm~1), magni­
F j
tudes we derive for each model in the NLTE grid are e+ec­
tively unnormalized surface magnitudes V s \ 2.5 log F j .
The formula we derive for converting the best ït DA WD
surface ÿuxes (in ergs cm~2 s~1 cm~1), surface gravities
F j
log g, and masses into absolute magnitude is
M/M _
M V \[2.5 log F j ] 2.5 log g [ 2.5 log
A M
M _
B ] 31.042 .
Our own resulting values of di+er little from those for
M V
35 DAs in common with Vennes et al. 1997b (mean di+er­
ence 0.01 ^ 0.24 mag). Since correlates strongly with
M V
log g, this is consistent with the more detailed comparison
performed by NGS99 of internal, external, and systematic
errors in log g and between the two studies. For dis­
T eff
tances to the DA WD, we compare to tabulated V magni­
tudes, mostly from Table 3 of Marsh et al. (1997).
We derive the K magnitude of any cool companion, K c ,
from the di+erence between the predicted magnitude of the
DAWD, and the total observed magnitude using
K p , K o
K c \K p [ 2.5 log [dex 0.4(K p [K o ) [ 1] .
In Table 5 we list the properties of candidate cool compan­
ions detected in our IR­observed sample of DA WDs, esti­
mated using the combination of the best­ït NLTE models
of NGS99, our own infrared photometry, and published
optical magnitudes. Column (1) lists REJ name. The com­
panion J and K magnitudes calculated as detailed above
are listed in columns (2) and (3). Spectral type of the candi­
date companions can be crudely estimated from the derived
or from J[K. Since the combined errors on are in
M Kc M Kc
general lower, we list this in column (4) and the resulting
spectral type estimates in columns (5) and (6), assuming
young or old disk population for the red dwarf. The implied
V magnitude (from V [K for the old disk spectral type) is
shown in column (7), with notes in column (8).
4.3. Candidate Cool Companions
Our adopted criteria for claiming a signiïcant near­IR
excess over the predicted values for an isolated DA is simply
a 3 p excess in both J and K relative to the predicted IR
magnitudes of the DA white dwarf, and Errors in our
J p K p .
observed K magnitudes are generally 2õ3 times higher than
for J, but provide greater wavelength contrast to the
optical. We do not use (J[K) as a discriminant because its
combined errors are highest. The errors we adopt for the
excesses and include our photometric
(J p [J o ) (K p [K o )
errors in J and K, combined with assumed errors of 0.1 mag
each on the predicted values of and
K p J p .
As an example, for REJ 1845]682, we calculate from the
published V magnitude and model atmosphere ïts in Table
1 a predicted magnitude with its estimated error of K p \
16.2 ^ 0.1 (Table 4). The di+erence between this and the
observed magnitude K\ 14.37 ^ 0.19 (Table 4) is thus
mag, qualifying as an 8 p excess.
K excess \ 1.83 ^ 0.21
Of the 47 objects with reliable K photometry (which
excludes REJ 1235]233/PG 1232]238) we ïnd 10 systems
with a signiïcant IR excess (see Fig. 2). Of these, ïve are
known binary systems. Of the ïve that are newly recognized
systems with IR excess, four have spatially resolved objects
discernible either in the near­IR or optical images within
TABLE 5
DERIVED PROPERTIES OF CANDIDATE COOL COMPANIONS
M SUBTYPEa V c b
REJ NAME J c K c M Kc (YD) (OD) (OD) NOTES
(1) (2) (3) (4) (5) (6) (7) (8)
0134[160 13.3 12.3 7.4 4.5 3.5 17.4 1
0148[253 12.4 11.6 6.9 4 3 16.4 1
0427]740 16.4 16.2 9.4 6.5 6 23.5 3
0916[194 14.0 13.7 7.7 5 4 19.2
1036]460 12.8 12.0 7.6 5 4 17.3 1
1126]183 13.0 12.1 6.5 3.5 2.5 16.6 1
1629]780 11.1 11.3 7.7 5 4 16.6 1
1711]664 15.1 14.2 8.0 5.5 5 20.0 2
1820]580 15.5 13.8 8.9 6 5.5 21.0
1845]682 15.3 14.6 9.2 6.5 6 21.9
NOTES.õ(1) Known binary ; see Table 1. (2) IR photometry known to include
nearby marginally optically resolved object at separation indicated in Table 1. (3) This
star is spatially resolved at near­IR wavelengths, with a companion within less than 2A.
Some nebulosity (di+use emission) is also noted nearby. The more distant companion
noted in Table 1 is not included in the IR photometry.
a Approximate spectral type from and Leggett 1992, Table 6, for Young Disk
M Kc
and Old Disk stars.
b Approximate V magnitude of companion assuming Old Disk population, from
and (V [K) of Leggett 1992, Table 6.
M Kc

No. 2, 2000 COOL COMPANIONS TO HOT WHITE DWARFS 999
FIG. 2.õDi+erence between predicted and observed magnitudes plotted against observed magnitude for IR­observed hot DA white dwarfs. Objects for
which we determine a signiïcant IR excess are shown as ïlled circles.
10A. In only some cases, noted in Table 5, these objects are
included in the near­IR photometry.
The overall properties of DAs with IR excess reveal no
obvious bias within our observed sample. The mean dis­
tance of the DAs without excess is 144 ^ 98 pc, while those
11 with excess have 117 ^ 53 pc. Similarly, the mean values
and distributions of log g, V , and of the DAs
T eff , M V
without excess are indistinguishable from those 11 with
excess.
The mean value of for the systems with no IR
(K p [K o )
excess is [0.28 ^ 0.33. We ïnd a similar o+set for (J p [J o )
of [0.30 ^ 0.35.4 We consider several possibilities to
explain the zero point o+set. (1) Our predicted DA (V [K)
colors are too red. Since we derive our DA K magnitudes
from the predicted color and the published V mag­
(V [K) p
nitude, a di+erence in normalization in our synthetic V or K
photometry of the DA models could cause the o+set.
However, the synthetic V and K magnitude and colors we
derive for stars in the BPGS spectral library match their
normalizations and colors for each spectral type (see ° 4.1
above), and our (V [K) colors also match those in the spec­
tral library of Pickles (1998) to within a mean di+erence of
0.01 ^ 0.1 mag. (2) Our IR photometric zero point is too
faint. Comparing the two DAs also observed by Zuckerman
& Becklin (1992), our K magnitude are 0.1 mag fainter in
each case. While this di+erence is well within our errors,
we cannot rule out a photometric zero point shift of this
order as a contributor to the mean zero point o+set.
(3) Unmodeled, large equivalent width IR absorption features
in DAs. To account for a shift of 0.3 mag in K, spectral
absorption features with equivalent widths of nearly 1440 A#
would be required that are not present in the DA models.
Given that di+erent absorption features would be required
4 The mean of is zero ([0.03 ^ 0.29).
(J[K) p [(J[K) o
in the J band to yield a similar zero point shift, we disregard
this possibility.
Even with this unexplained zero point shift, our deter­
mination of an IR excess, based on the signiïcance of
are una+ected. However, our estimated compan­
(K p [K o ),
ion magnitudes and absolute magnitudes may be
K c M Kc ,
up to 0.3 mag too faint, leading in turn to spectral subclass
estimates approximately half a subclass too late.
The V magnitudes of the candidate companions listed in
Table 5 are in most cases signiïcantly fainter than the visual
magnitudes of the DAWDs. Since we elected not to observe
in the IR those systems known to be composite by their
optical spectra, this result is expected from our selection.
4.4. Spatially Resolved Companions
Finding charts for most of our sample have been
published by Shara et al. (Shara, Shara, & McLean 1993 ;
Shara et al. 1997). To account for visually resolved compan­
ions, we inspected images from the DSS (Shara et al. 1997)
and also searched the USNOA­1.0 (Monet 1996) catalog to
RD 19.5 for all stars within 15A, of the program star.
Resolved stars within a maximum distance of 10A, and
within 3 mag (in R) of the primary are noted in Table 1 as
visual pairs. For the closest WD distances in our sample of
D20 pc, the spatial resolution of D2A in both the archival
optical and new IR images, corresponds to 40 AU. On the
other hand, a search limited to 10A, corresponds to
maximum separations of 200 AU. For the farthest WDs in
our sample near 500 pc, imaging is instead sensitive to
resolved companions between 1000 and 5000 AU.
5. NOTES ON INDIVIDUAL OBJECTS
REJ 0134[160.õBased on radial velocity measure­
ments (Schultz, Zuckerman, & Becklin 1996) and its low­
mass (NGS99 ; Finley, Koester, & Basri 1997) mass transfer

1000 GREEN, ALI, & NAPIWOTZKI Vol. 540
is unlikely to have occurred in this binary DA ] M star
system.
REJ 0148[253.õFinley et al. (1997) claimed evidence
for a companion based on Hb emission and a slight red
excess. Our result of an IR excess conïrms the presence of
the companion.
REJ 0427]7406.õThis star appears to be a multiple at
near­IR wavelengths. Some nebulosity (di+use emission) is
also noted nearby. A ïnding chart is provided in Figure 3.
REJ 0443[034.õThis is the hottest DA in our sample of
IR­observed WDs. DA stars with such high temperatures
should cool extremely rapidly and thus be detected only
rarely. The suggestion by Finley et al. (1997) of a possible
red excess is not borne out by our measurements.
REJ 0512[004.õAs the least massive DA in the sample,
this DA might be expected to have a companion, since very
low­mass WDs are thought to form in binaries (Marsh,
Dhillon, &Duck 1995). The absence of IR excess indicates a
very cool and/or degenerate companion at best. Our IR
photometric method is insensitive to such objects, which are
best detected through radial velocity variations (Sa+er,
Livio, & Yungelson 1998). We discuss the overall e+ect on
our estimated binary fraction in ° 7.
REJ 0916[194.õWhile its IR excess suggests a late­type
companion, our estimate for its visual magnitude is only
V c
D2 mag fainter than the total V magnitude, so evidence for
the companion should have been seen in the red end of its
optical spectrum. However, the only spectrum (from Vennes
et al. 1997b) is of comparatively low S/N and extends only
to 6040 We expect that a higher S/N optical spectrum
A# .
extending farther redward, or our own planned IR spectros­
copy will readily reveal the companion.
REJ 1036]460 (PG 1033]464).õK photometry and K­
excess from Zuckerman & Becklin (1992) are within 0.1 mag
of the values we report here. However, their derived absol­
FIG. 3.õ5@ ] 5@ K­band image of the ïeld of REJ 0427]7406 reveals
several close neighbors to the DA. Di+use emission is also evident to the
SE, possible a background galaxy. North is up, east to the left.
ute red dwarf magnitude is fainter by 1.5 mag. Rather
M Kc
than using detailed spectral ïts, they had to use published
color­estimated distance moduli that assumed uniform
surface gravity for all white dwarfs, while in fact the absol­
ute magnitude depends strongly on log g. For the other
object in common between our samples, REJ 1126]183
(PG 1123]189), we note an even more striking discrepancy
(3.2 mag in They assume a distance of 30 pc, while the
M Kc ).
best estimate from detailed ïts of NGS is 135 pc. Their
resulting claim that the IMF is ÿat or increasing down to at
least 0.1 should be revisited with more accurate dis­
M _
tance determinations.
REJ 1043]490.õThis was suggested by Schwartz et al.
(1995) as a possible multiple system based on a slightly red
R[I color, but they suggested that further measurements
were needed to conïrm this. The residuals of the DA model
ït to the optical spectrum in NGS99 conïrmed a red excess
for j [ 4500 corresponding to a companion with
A#
V D 18.2 or at our distance of 370 pc. However,
M V D 10.4
our near­IR photometry is inconsistent with such a com­
panion, since a normal late M dwarf companion of this
absolute magnitude and distance would yield J D 14.5 and
KD 13.6. We note that the system is a visual binary, with a
red companion 8A West of the DA. Contamination of the
DA spectrum in NGS99 is thus the likely explanation.
REJ 1235]233 (PG1232]238).õThe object is barely
discernible in our K mosaic, so reported K photometry is
poor.
REJ 1711]664.õThis is a barely resolved visual pair,
with a late­type star only distant, which is included in
D2A.5
our IR photometry (Mason et al. 1995). Jomaron (1997) also
found the M­type secondary in residuals to the DA optical
spectrum, classifying the former as type M3õM4, with no
apparent H emission. Our IR photometry indicates a some­
what later type.
REJ 2207]252.õIR photometry for 2207]252 was
contaminated by a very bright, resolved object at 8A.9.
6. SELECTION BIASES
Strong biases are likely to be inherent in any sample of
EUV­detected WDs. The strongest is the selection against a
high column of intervening interstellar material. Columns
as low as 5 ] 1019 cm~2 easily quench ÿux in the EUV
below detection levels achieved in either the ROSAT WFC
or the EUV E.5 When compared to more distant areas in
the Galactic plane, the local interstellar medium lacks cool
dense material (n D 0.1 cm~3, T D 100 K). This ```` local
bubble îî has a mean radius of 70 pc (Diamond, Jewell, &
Ponman 1995). Accurate estimates of the intervening
column to each star are difficult, requiring high S/N EUV
spectral analysis well beyond the scope of this work. The
global e+ect amounts to a direction­dependent ÿux sensi­
tivity limit (and therefore volume) that is strongly depen­
dent on the ïne structure of the ISM.
As pointed out by NGS99, at a given temperature, WDs
with lower masses are larger, consequently more luminous,
and thus detectable to larger distances. For instance, at
30,000 K, a WD of 0.5 is detectable at distances 30%
M _
5 For example, one optical depth is reached at column densities of 30,
5, 1, and 0.5 ] 1018 cm~2 , respectively, for the EUV E 100, 200, 400, and
600 detection bandpasses (Vennes et al. 1996).
A#

No. 2, 2000 COOL COMPANIONS TO HOT WHITE DWARFS 1001
FIG. 4.õJ excess (di+erence between predicted and observed J
magnitudes) plotted against DA white dwarf mass in solar units from the
best­ït model atmospheres of NGS99. We see no evidence that IR excess
correlates with DAmass.
larger than a 0.7 WD. For many WDs at the maximum
M _
distance sampled by most optical surveys, interstellar
matter e+ectively absorbs all EUV radiation. Thus, the
dominant e+ect in EUV surveys is probably not a selection
for massive WDs as suggested by Vennes et al., but a selec­
tion against low­mass WDs due to a sample volume strong­
ly a+ected by interstellar absorption. However, there is no
direct bias induced by interstellar absorption on the
observed binary fraction.
There have been suggestions that detection of soft X­ray
ÿux from hot white dwarfs may be biased against close
binary systems. This could arise if heavy elements arising in
a wind from the MS companion have accreted onto the
surface of the white dwarf, enhancing atmospheric EUV
and soft X­ray opacity (Barstow et al. 1993). By including
only DA stars in our sample, we might thus be preferentially
excluding some binary systems. For instance, the hot white
dwarf in the close binary RE 1016[053 is a DAO star
whose mixed H/He composition is likely to be attributable
to slow accretion from a red dwarf companion (Vennes et al.
1997a). However, Barstow et al. (1994) found that hot
ROSAT (PSPC or EUV) detected WDs in noninteracting
binaries with early­type stars showed no signs of abundance
enhancements from accretion, their opacity sources follow­
ing the same temperature pattern as seen for isolated WDs.
Indeed, while those abundance enhancements may lead to
signiïcant X­ray opacity (e.g., Dupuis et al. 1995) in the
hottest WDs, they remain strong EUV sources, particularly
at j [ 200 A# .
Arguments for biases of opposite e+ect exist as wellõthat
increased EUV ÿux is more likely from a WD that su+ers
reheating due to accretion of material from its companion.
Since the DAs in our sample are not signiïcantly variable,
the interaction would have to be fairly weak or intermittent
(e.g., via weak wind accretion). Furthermore, Barstow et al.
(1992) ïnd for the DA ] K2V binary system V471 Tauri
that the temperature change in such cases (with accretion
rates of 0.4õ11 ] 10~13 yr~1) is small and occurs only
M _
in the region near the hot spot of the accretion column.
Measurement of these putative competing selection
e+ects would clearly beneït from larger samples. A large
number of EUV­detected DA white dwarfs with good con­
straints from atmospheric modeling now exist (Marsh et al.
1997 ; Vennes et al. 1997b ; Finley, Koester, & Basri 1997 ;
NGS99). The 2MASS will provide (less sensitive, K¹ 14)
IR photometry for brighter subsamples.
WDs with masses less than the canonical core­helium
ignition mass (0.49 Sweigart 1994) are likely to be in
M _ ;
short­period binary systems, where their evolution was cut
short by wind ejection during a common envelope phase
(Marsh et al. 1995 ; Marsh 1995). Only one of the WD stars
in our sample, REJ 0512[004, has a low mass (0.45 M _ ),
but it shows no signs of a companion. On the other hand,
lower mass companions may be preferentially removed by
merging, a process that is more e+ective the more massive
the WD.
It is thus of interest to check for any correlation between
DA mass and IR excess. Figure 4 shows that no such corre­
lation is detectable in our sample, as conïrmed by two
correlation tests. Similarly, there is no correlation between
the absolute K magnitudes of companions and those
M KC
of their corresponding DA white dwarfs.6 Two sample sta­
tistical tests also reveal no signiïcant di+erence between the
mass or absolute magnitude distributions of the DAs in the
no­excess and excess subsamples.7 Larger samples would
put better constraints on these hypotheses.
7. THE BINARY FRACTION
Our WD sample demands that the initial primary has
already passed through the AGB phase, imposing a lower
mass limit of about 2 on the WD progenitor. We thus
M _ ,
probe portions of the (initial) mass ratio distribution that
are quite difficult to reach in unevolved systems. The binary
frequency of high­mass stars has proven difficult to deter­
mine for several reasons. Photometric detection of a faint
low­mass star next to a bright early type primary requires
high precision photometry. The use of radial velocity varia­
tions can also be difficult due to the large rotational speeds
of many early­type starsõone reason why solar­type stars
are preferred in current planetary searches.
Out of our uniformly selected sample of 57 EUV­detected
DA white dwarfs, including our new detections of IR excess,
we ïnd strong evidence for multiplicity in 20 systems (35%)
within a 10A, radius. Simple Poisson statistics on the sub­
sample yields an error of 7%. We may also count multiple
systems by assuming that any resolved object within the
10A, radius is at the same distance as the WD (i.e., a pre­
sumed companion), and thereby tally objects within a
chosen limiting physical separation (projected onto the sky
6 Generalized Kendallîs Tau and Cox Proportional Hazard model tests
from the ASURV package (LaValley et al. 1992) show probabilities P of
18% and 72%, respectively, that a correlation is not present between
and DA mass. A signiïcant correlation requires P[ 99%. P of
(J p [J o )
26% and 36%, respectively, are obtained for vs.
M Kc M VDA .
7 Generalized Wilcoxon and logrank tests show that the two samples
are inconsistent with being drawn from the same parent population at only
the PD 75% level for DAmass and only the PD 93% level for M VDA1002 GREEN, ALI, & NAPIWOTZKI Vol. 540
plane). Since the largest estimated distance to any of the
WDs in our sample is 470 pc (for REJ 1235]233) we
choose for completeness8 a maximum projected physical
separation of 1000 AU. This removes only one system
(1726]583), leaving the binary fraction essentially
unchanged at 33 ^ 7%.
For comparison, among a complete sample of nearby
(d \ 5 pc) M dwarfs, Leinert et al. (1997) found 26 ^ 9% to
have companions. In a meta­analysis, Fischer & Marcy
(1992) found that 42 ^ 9% of M dwarfs have main­sequence
companions. This is lower than the best incompleteness­
corrected estimate for G dwarfs (D57% ; Duquennoy &
Mayor 1991), due in part to the smaller range of secondary
companion masses available.
As allowed by the faintness of WDs relative to main­
sequence companions, and the use of IR photometry, the
current study is sensitive to a wide range of companion
masses, from late M to early A type. The late­type compan­
ion masses corresponding to our estimates of absolute K
magnitudes range from about 0.5 down to 0.1
M KC M _ .9
Judging by known or estimated spectral types of unresolved
companions and by derived absolute magnitudes of the pre­
sumed companions discussed above, 90% of companions
from Tables 1 and 5 are of spectral type K or later.
We note that despite the signiïcant improvement in
sensitivity to low­mass companions that our IR photometry
a+ords, the binary frequency of our sample remains a lower
limit for several reasons. First, very low­mass companions
can fall below our IR detection limit of KD 16, with sensi­
tivity decreasing with distance. Also, while we assume only
a single degenerate in each system, double degenerates
would not likely be distinguished by our IR photometry.
Sa+er et al. (1998) derive a fraction of double degenerate
systems that may be as high as 20% when they include an
estimated correction factor of D1.4 for selection efficiency
in their sample.
Some of the hot DAs with masses higher than expected
from isolated evolution (D1.1 have been hypothesized
M _ )
to represent a population of coalesced double degenerate
systems (e.g., Marsh et al. 1997). Such mergers would slight­
ly reduce the observed binary fraction in our sample. While
they might also help constrain the Type Ia supernova for­
mation rate (e.g., Branch et al. 1995), the predicted peak of
such a merger distribution should be near 0.9 which is
M _ ,
not seen.
8 Assumes a spatial resolution of 2A, at a similar maximum distance of
500 pc. Including only unresolved systems is intuitively appealing, but it
introduces a bias against companions in nearby systems.
9 Below about 0.2 absolute magnitudes no longer correlate well
M _
with mass. There the relationship steepens considerably, in both V and K
bands, and di+ers little as a function of metallicity (Bara+e et al. 1998
Henry et al. 1999).
Among our sample of hot DAs may be some close
binaries that have passed through a phase of common
envelope evolution. While these are sometimes called preca­
taclysmic binaries, many will not become cataclysmic vari­
ables within a Hubble time (Ritter 1986). Based on
morphological classiïcation of planetary nebulae, and the
assumption that axisymmetrical mass loss is caused by
binary companions, Soker (1997) estimated that only 10%
of PN had no interaction with any companion, stellar or
substellar. Only 11% had interactions while avoiding a
common envelope phase. For the case of close binaries, the
observed distribution of companion masses may thus have
been modiïed by evolution. Some systems may have
coalesced, while others may have grown through wind acc­
retion, or by stable mass transfer. As the progenitors of
novae, CVs, symbiotics, Feige 24­type systems, and the
intriguing dwarf carbon (dC) stars (deKool & Green 1995),
these WD]MS systems warrant the improved constraints
on their masses and abundance enhancements that IR spec­
troscopy of our sample will provide. For instance, the dC
stars, which likely evolve into CH or Ba giants, are dwarfs
whose strong carbon bands betray episodes of mass­
transfer from the previously AGB companion, now the WD
(Green 2000). There is recent evidence (Heber et al. 1993 ;
Liebert et al. 1994) that dCs may show stronger abundance
enhancements soon after the AGB phase, i.e., while the WD
is still hot. IR spectra would test for this, and its potential
ramiïcations for the depths and timescales of mixing due to
chemical potential gradients (e.g., Proffit &Michaud 1989).
We are obtaining IR spectra of these systems at the
KPNO 2.1 m with CRSP to better constrain the mass and
spectral type of the cool companions presumably causing
the IR excess. We await these results before pursuing more
detailed discussions of the mass function of the companions,
the observed ïnal and implied initial mass ratios of the
systems, and detailed simulations of the e+ects of binary
evolution. The binary fraction may be a function of the
mass ratio, or of the primary mass ; these fundamental ques­
tions of stellar evolution remain open. If the observed
binary fraction for DA WDs proves to be inconsistent with
those of main­sequence primaries in a similar (initial) mass
range, it could mean that age­dependent formation condi­
tions (e.g., abundance, kinematics, or turbulence) strongly
inÿuence the binary fraction.
Many thanks to Dick Joyce and the sta+ at KPNO for all
their help during IRIM observing runs. Thanks to Fred
Ringwald for providing a catalog of unresolved, hot­high­
gravity with cool companion composites. The author grate­
fully acknowledges support provided by NASA through
grant NAG5­1253, and contract NAS8­39073 (CXC). This
research has made extensive use of SIMBAD.
APPENDIX
PHOTOMETRY OF MOSAICKED IMAGES
For some targets, visual inspection failed to reveal the source on the individual images. We derive photometry these targets
from a global mosaic of the ïeld produced by shifting and combining individual frames. We use integer pixel shifts and
average the overlap regions between frames, accounting for background/sky variations by subtracting the median value of the
frame from itself before averaging the ïelds. Integer pixel shifts do not necessarily conserve ÿux but are employed here for two

No. 2, 2000 COOL COMPANIONS TO HOT WHITE DWARFS 1003
reasons : (1) the stellar images on our frames are sub­Nyquist sampled. (2) The presence of signiïcant dead (insensitive) space
between pixels. The undersampled proïles of the star generally ensure that most of the stellar ÿux is contained within one
pixel, and the presence of dead space combined with undersampled images makes it difficult to produce ÿux conserved stellar
proïles in the ïrst place. Thus, we conclude that mosaicking images will not signiïcantly alter our photometry. These
assumptions are tested by comparing magnitudes derived from a mosaicked image to those from individual frames. Figure 5
illustrates this comparison. Figure 5 demonstrates that photometry from mosaicked images is consistent (within the errors)
with photometry obtained from individual frames.
The mosaicked images, however, only allow us to make an estimate of the formal (Poisson ] detector noise) errors and do
not account for any systematic errors or any additional possible degradation of the S/N created by the photometry software.
A comparison of the formal errors with those obtained by comparing the rms magnitude variation from individual frame
suggests that our errors are dominated by these systematic errors. The dashed line in Figure 6 shows the magnitude of the
formal errors. The solid dots are the observed rms variations of magnitudes from individual frames. Clearly, formal errors
underestimate the measured empirical estimates. The formal errors are derived using a modiïed form of the IRAF task
FIG. 5.õComparison of the derived magnitudes in the ïeld of PG 0904]511 using individual frames and a mosaic image. K­band magnitudes derived
from an image mosaic and those from individual frames are compared for ïve stars in the PG 0904]511 ïeld. The error bars show the 1 p deviation among
magnitudes from individual frames. The dashed line shows the resulting relationship if the two quantities were identical.
FIG. 6.õObserved error (estimated as the 1 p deviation among magnitudes derived from individual frames) vs. magnitude for the PG 0904]511 ïeld. The
dashed line shows the magnitude of the formal errors from S/N arguments. The solid line shows our error curve for the ïeld (see text for details). These data
suggest that we are dominated by systematic errors.

1004 GREEN, ALI, & NAPIWOTZKI
```` ccdtime îî (modiïed to run on the IDL package). To obtain a realistic estimate for the errors on mosaicked images, we
modeled the e+ects of systematic errors by artiïcially increasing the readout noise of the detector and minimizing the
relationship :
;
i/1
N (p obs [ p cal )2
p obs 2 ,
where are the observed errors, are errors calculated by increasing readout noise, and N is the number of stars in this
p obs p cal
experiment. The solid line in Figure 6 shows an example of this ït. We use the best­ït ```` readout noise îî to extrapolate the
formal S/N to stars for which only mosaicked image photometry is available.
REFERENCES
Bara+e, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337,
403
Barstow, M. A., et al. 1992, MNRAS, 255, 369
Barstow, M. A., et al. 1993, MNRAS, 264, 16
Barstow, M. A., et al. 1994, MNRAS, 270, 499
Burleigh, M. R., Barstow, M. A., & Fleming, T. A. 1997, MNRAS, 287,
381
Branch, D., Livio, M., Yungelson, L. R., Boffi, F. R., & Baron, E. 1995,
PASP, 107, 1019
Cash, W., Charles, P., & Johnson, H. M. 1980, ApJ, 239, L23
Casali, M. M., &Hawarden, T. G. 1992, JCMT­UKIRT Newsletter, 3, 33
M. S., Sarna, M. J., Jomaron, C. M., & Cannon, S. R. 1995,
Catala
‘ n,
MNRAS, 275, 153
deKool, M., &Green, P. J. 1995, ApJ, 449, 236
deKool, M., & Ritter, H. 1993, A&A, 267, 397
Diamond, C. J., Jewell, S. J., & Ponman, T. J. 1995, MNRAS, 274, 589
Dreizler, S., Heber, U., Jordan, S., & Engels, D. 1994, in Hot Stars in the
Galactic Halo, ed. S. J. Adelman, A. R. Upgren, & C. J. Adelman
(Cambridge : Cambridge Univ. Press), 228
Dupuis, J., Vennes, S., Bowyer, S., Pradhan, A. K., & Thejll, P. 1995, ApJ,
455, 574
Duquennoy, A., &Mayor, M. 1991, A&A, 248, 485
Finley, D. S., et al. 1993, ApJ, 417, 259
Finley, D. S., Koester, D., & Basri, G. 1997, ApJ, 488, 375
Fischer, D. A., &Marcy, G. W. 1992, ApJ, 396, 178
Fowler, A. M., Gatley, I., Stuart, F., Joyce, R. R., & Probst, R. G. 1988,
Proc. SPIE, 972, 107
Green, P. J. 2000, in Proc. IAU Symp. 177, The Carbon Star Phenomenon,
ed. R. F. Wing (Dordrecht Kluwer), in press
Green, P. J., &Margon, B. 1994, ApJ, 423, 723
Green, R. F., Schmidt, M., & Liebert, J. 1986, ApJS, 61, 305
Gunn, J. E., & Stryker, L. L. 1983, ApJS, 52, 121
Hayes, D. S., & Latham, D. W. 1975, ApJ, 197, 593
Heber, U., Bade, N., Jordan, S., & Voges, W. 1993, A&A, 267, L31
Henry, T. J., et al. 1999, ApJ, 512, 864
Howell, S. B. 1989, PASP, 101, 616
Jomaron, C. M. 1997, Ph.D. thesis, University College, London
LaValley, M., Isobe, T., & Feigelson, E. D. 1992, in Astronomical Data
Analysis Software & Systems, ed. D. Worrall et al. (San Francisco ASP)
Leggett, S. K. 1992, ApJS, 82, 351
Leinert, C., Henry, T. J., Glindemann, A., & McCarthy, D. W., Jr. 1997,
A&A, 325, 159
Liebert, J., et al. 1994, ApJ, 421, 733
Liebert, J., Bergeron, P., & Sa+er, R. A. 1990, PASP, 102, 1126
Malina, R., et al. 1994, AJ, 107, 751
Marsh, M. C., et al. 1997, MNRAS, 286, 369
Marsh, T. R. 1995, MNRAS, 275, 89
Marsh, T. R., Dhillon, V. S., &Duck, S. R. 1995, MNRAS, 275, 828
Mason, K. O., et al. 1995, MNRAS, 274, 1194
McCook, G. P., & Sion, E. M. 1999, ApJS, 121, 1
Monet, D. 1996, BAAS, 188, 5404
Napiwotzki, R., Green, P. J., & Sa+er, R. A. 1999, ApJ, 517, 399 (NGS99)
Pickles, A. J. 1998, PASP, 100, 863
Pounds, K., et al. 1993, MNRAS, 260, 77
Probst, R. 1983a, ApJS, 53, 335
õõõ. 1983b, ApJ, 274, 237
Proffitt, C. R., &Michaud, G. 1989, ApJ, 345, 998
Ritter, H. 1986, A&A, 169, 139
Sa+er, R. A., Livio, M., & Yungelson, L. R. 1998, ApJ, 502, 394
Schmidt, G. D., Bergeron, P., Liebert, J., & Sa+er, R. A. 1992, ApJ, 394, 603
Schultz, G., Zuckerman, B., & Becklin, E. E. 1996, ApJ, 460, 402
Schwartz, R. D., Dawkins, D., Findley, D., & Chen, D. 1995, PASP, 107,
667
Shara, M. M., Bergeron, I. E., Christian, C. A., Craig, N., & Bowyer, S.
1997, PASP, 109, 998
Shara, M. M., Shara, D. J., &McLean, B. 1993, PASP, 105, 387
Soker, N. 1997, ApJS, 112, 487
Strecker, D. W., Erickson, E. F., &Witteborn, F. C. 1979, ApJS, 41, 501
Sweigart, A. V. 1994, ApJ, 426, 612
Vennes, S., Christian, D. J., & Thorstensen, J. R. 1998, ApJ, 502, 763
Vennes, S., Dupuis, J., Bowyer, S., & Pradhan, A. K. 1997a, ApJ, 482, L73
Vennes, S., Thejll, P. A., Wickramasinghe, D. T., & Bessell, M. S. 1996, ApJ,
467, 782
Vennes, S., Thejll, P. A., Galvan, R. G., &Dupuis, J. 1997b, ApJ, 480, 714
Vennes, S., & Thorstensen, J. R. 1994, ApJ, 433, L29
Wonnacott, D., Kellett, B. J., & Stickland, D. J. 1993, MNRAS, 262, 277
Zombeck, M. V. 1990, Handbook of Space Astronomy and Astrophysics
(Cambridge : Cambridge Univ. Press)
Zuckerman, B., & Becklin, E. 1992, ApJ, 386, 260