1: \documentclass[12pt,preprint]{emulateapj}
2: \usepackage{graphics}
3:
4: \def\gtorder{\mathrel{\raise.3ex\hbox{$>$}\mkern-14mu
5: \lower0.6ex\hbox{$\sim$}}}
6: \def\ltorder{\mathrel{\raise.3ex\hbox{$<$}\mkern-14mu
7: \lower0.6ex\hbox{$\sim$}}}
8:
9: \def\Msun{\>{\rm M_{\odot}}}
10:
11:
12: \shorttitle{White Dwarf Planet Detection}
13: \shortauthors{Debes et al.}
14:
15: \begin{document}
16: \title{Cool Customers in the Stellar Graveyard II: Limits to Substellar Objects
17: around nearby DAZ White Dwarfs}
18: \author{John H. Debes\altaffilmark{1}, Steinn Sigurdsson\altaffilmark{1},
19: Bruce E. Woodgate\altaffilmark{2}}
20:
21: \altaffiltext{1}{Department of Astronomy \& Astrophysics, Pennsylvania State
22: University, University Park, PA 16802}
23: \altaffiltext{2}{NASA Goddard Space Flight Center, Greenbelt, MD 27710}
24:
25: \begin{abstract}
26: Results from a concerted Hubble Space Telescope (HST) survey of nearby white
27: dwarfs for substellar objects is presented. A total of 7 DAZ
28: white dwarfs with distances of $<$ 50 pc had high contrast and high spatial
29: resolution NICMOS coronagraphic images taken
30: to search for candidate substellar objects at separations $\ltorder$10\arcsec\
31: away. Limits to unresolved companions are derived through analysis of 2MASS
32: photometry of the white dwarfs compared to expected fluxes based on the
33: WDs effective temperature, distance, and gravity. Our HST survey of seven DAZ white dwarfs identified candidate companions for four of the white dwarfs. For three of these four, HST and ground-based second epoch observations showed the candidates to be background stars. The fourth white dwarf, which is close to the galactic plane, has seven candidate companions at distances of 2\arcsec\ to 4\arcsec, which remain to be followed up. We find that for four of the white dwarfs
34: we are sensitive to planetary companions $\gtorder$10 M$_{Jup}$. For all of the targets, we are sensitive
35: to companions $>$ 18 M$_{Jup}$.
36: The lack of
37: significant near infrared excesses for our targets
38: limits any kind of unresolved companions
39: present
40: to be substellar. In light of these results we make several
41: comments on the possibility of determining the origin of metals in the
42: atmospheres of these white dwarfs.
43: \end{abstract}
44:
45: \keywords{circumstellar matter --- planetary systems --- white dwarfs --- stars: low-mass, brown dwarfs --- infrared:stars}
46:
47: \section{Introduction}
48:
49: The last ten years have shown a surge of new discoveries about objects of
50: substellar mass. Radial velocity surveys of main sequence K-F stars have found
51: few brown dwarf companions at separations of $<$3 AU, but a profusion
52: of planetary mass companions \citep{marcy00}.
53: Large all sky-surveys, such as 2MASS and
54: SDSS have found large numbers of free floating brown dwarfs \citep{burgasser03,hawley02}. Low mass
55: substellar objects down to planetary mass have been discovered in young
56: clusters such as $\sigma$ Orionis \citep[][and references therein]{lada03}.
57: At the same time, imaging
58: surveys of nearby main sequence stars have found several substellar companions
59: thanks to high contrast imaging \citep[e.g.][]{forveille04}.
60: One population of stars which still has little data are
61: intermediate mass stars with masses between 1.5-8 M$_{\odot}$.
62:
63: The reason
64: for the dearth of information around intermediate mass stars
65: is two-fold. First, the majority of searches for
66: planetary systems focus on Solar System analogues. Secondly, there are
67: technical limitations to searching for planets and brown dwarfs around
68: main sequence
69: F-B stars. Radial velocity surveys rely on a large number of narrow
70: absorption lines
71: in the stellar spectrum to achieve high precision velocity measurements \citep{
72: delfosse98,griffin00}.
73: As the effective temperature of a star increases,
74: metal line strengths decrease and
75: there are fewer lines for measurement. Stars with higher masses have a
76: correspondingly smaller reflex motion due to a low mass companion.
77: Radial velocity surveys of G giant stars can probe higher mass stars, but there
78: are only two planetary candidates currently reported \citep{sato03,setiawan05}. More massive stars have higher luminosities
79: as well, making high contrast imaging more limited
80: in its effectiveness if one is looking for the thermal emission from a
81: companion. Reflected light from substellar companions is most useful
82: within a few AU of a star and is negligible at larger distances \citep{burrows04}. In most cases, searches focus on detecting the thermal radiation from a
83: substellar companion.
84:
85: \begin{deluxetable*}{llccccccc}
86: %table of targets with age, mass, effective temperature
87: \tablecolumns{9}
88: \tablewidth{0pc}
89: \tablecaption{\label{tab:targs} Properties of the Target White Dwarfs}
90: \tablehead{
91: \colhead{WD} & \colhead{Name} & \colhead{M$_f$\tablenotemark{a}} & \colhead{T$_{eff}$} & \colhead{t$_{cool}$} & \colhead{D} & \colhead{M$_i$} & \colhead{t$_{cool}$+t$_{MS}$} & \colhead{References} \\
92: & & \colhead{($\Msun$)} & \colhead{(K)} & \colhead{(Gyr)} & \colhead{(pc)} & \colhead{($\Msun$)} & \colhead{(Gyr)} & }
93: \startdata
94: 0208+396 & G 74-7 & 0.60 & 7310 & 1.4 & 17 & 2.1 & 3.2 & 1 \\
95: 0243-026 & G 75-39 & 0.70 & 6820 & 2.3 & 21 & 3.2 & 2.8 & 1 \\
96: 0245+541 & G 174-14 & 0.76 & 5280 & 6.9 & 10 & 4.6 & 7.2 & 1 \\
97: 1257+278 & G 149-28 & 0.58 & 8540 & 0.9 & 34 & 1.7 & 3.3 & 1 \\
98: 1337+701 & EG 102 & 0.57 & 20435 & 0.1 & 25 & 1.6 & 3.3 & 2,3 \\
99: 1620-391 & CD-38$^\circ$10980 & 0.66 & 24406 & 0.1 & 12 & 3.1 & 0.7 & 4 \\
100: 2326+049 & G 29-38 & 0.70 & 11820 & 0.6 & 14 & 3.7 & 1 & 2,5 \\
101: \enddata
102: \tablenotetext{a}{Values for M$_{f}$, T$_{eff}$, and
103: t$_{cool}$ were determined from listed references. Distances derived from parallax measurements
104: compiled in (1). If not available,
105: (3) and (5) were used. See Section \ref{s4.2} for the calculation of
106: M$_i$ and the WDs' total ages.}
107: \tablerefs{(1) \citet{bergeron01} (2) \citet{liebert04} (3) \citet{hip}
108: (4) \citet{bragaglia95} (5) \citet{vanaltena95}}
109: \end{deluxetable*}
110:
111: The effects of higher intrinsic luminosity are illustrated by noting the
112: sensitivity to substellar companions of the NICMOS instrument on HST.
113: High contrast imaging can achieve $\Delta H
114: \sim$10 at 1\arcsec\ on the NICMOS coronagraph with PSF subtraction, allowing
115: 45 M$_{Jup}$ mass companions to be detected around a 1 Gyr solar mass star. For an
116: A star with a mass of 2 M$_\odot$ at 1 Gyr, a 90 M$_{Jup}$ companion can be detected. Finally, more massive stars are rarer in local space, forcing
117: observations of young star forming regions at larger distances.
118:
119: Recent images of young HAe/Be stars with circumstellar disks such as HD 141569,
120: HR 4796A, and AB Aurigae
121: motivate a search for planets
122: around higher mass stars \citep{weinberger99,rayjay99,grady99}. Sub-mm
123: observations of warped and clumpy disks around stars
124: such as Vega and Formalhut show
125: that planet formation may be vigorous for higher mass central stars \citep{holland98}.
126: What is still unclear is how planet formation efficiency varies with
127: stellar mass and whether the brown dwarf desert is present over the same
128: orbital separations for higher mass stars.
129:
130: The 145 or so discoveries of planets by radial velocity surveys have told
131: us much about planet formation, but discoveries of planets in orbit around
132: post main sequence objects have the opportunity to challenge many accepted
133: assumptions developed
134: on the basis of our current knowledge. For example, the first terrestrial
135: planets ever discovered were around a pulsar \citep{wolszczan92}. The oldest Jovian planet
136: discovered in the M4 globular cluster in orbit around a white dwarf
137: demonstrates that planet formation can occur in metal poor systems \citep{sigurdsson03}.
138: This discovery must be explained in the context of planet formation mechanisms
139: that favor stars with higher metallicity.
140:
141: Detecting substellar companions in orbit around white dwarfs have several
142: advantages compared to searching main sequence stars. Given their intrinsic dimness, white dwarfs allow high
143: contrast searches to probe interesting orbital separations \citep{burleigh02}.
144: In
145: addition, their higher effective temperature allows searches for unresolved
146: excesses at longer wavelengths \citep{ignace01}. Given the range of
147: progenitor masses, white dwarfs probe a large
148: range of stellar mass. Finally, they
149: complement radial velocity and transit searches that are biased towards close
150: companions. High spatial resolution and high contrast imaging in the near
151: infrared with the NICMOS camera on HST allows the best chance for detecting
152: faint cool companions to nearby white dwarfs. Planetary mass objects that are
153: less than 3~Gyr can be observed in the near-IR, specifically in the F110W ($\sim$J) and
154: F160W ($\sim$H) filters. For example, a 3~Gyr old 10~M$_{Jup}$ planet can be observed
155: out to 20~pc with an HST observation of $\sim$1200s.
156:
157: Searching a subset of white dwarfs that harbor markers for
158: substellar objects can maximize the return of a survey. Nearby hydrogen
159: white dwarfs with metal line absorption (DAZs) may fit this criterion. Three
160: hypotheses have been put forth to explain the presence of DAZs: interstellar
161: matter (ISM) accretion \citep{dupuis92,dupuis93a,dupuis93b},
162: unseen companion wind accretion \citep{zuckerman03},
163: and accretion of
164: volatile poor planetesimals \citep{alcock86,debes02,jura03}.
165:
166: ISM accretion
167: has a wealth of problems in predicting many aspects of DAZs such as the large
168: accretion rates required for some objects and the distribution of these objects
169: with respect to known clouds of dense material \citep{aannestad93,
170: zuckerman98,zuckerman03}. The quick atmospheric
171: settling times of hydrogen atmospheres imply that the white dwarfs are
172: in close proximity with accretionary material.
173:
174: Of the $\sim$34 DAZs known, seven of them have dM companions, supporting the argument that
175: DAZs could have unseen companions that place material onto the
176: WD surface through winds \citep{zuckerman03}. In order to accrete enough material,
177: companions must be in extremely
178: close orbits, bringing into question why these objects have yet to be discovered
179: through transits, radial velocity surveys of compact objects, or
180: observable excesses in near-IR flux. In most cases
181: the reflex motion from such objects would be easily detectable
182: \citep{zuckerman92}.
183: The idea of the presence of unseen companions
184: also cannot explain objects like WD 2326+049 (G~29-38) which
185: has an infrared
186: excess due to a dust disk at roughly the tidal disruption radius \citep{graham90, patterson91}.
187:
188: The invocation of cometary or asteroidal material as a method of polluting WD
189: atmospheres was developed to explain photospheric absorption lines due to metals in
190: the DAZ WD 0208+395 (G 74-7) \citep{alcock86}. However, the rates
191: predicted by these original studies could not
192: satisfactorily explain the highest accretion rates inferred for some objects
193: and could not easily reproduce the distribution of DAZs based on their
194: effective temperatures \citep{zuckerman03}.
195: However,
196: mixing length theory predicts a drop-off of observability for
197: accretion as a function of effective temperature which may swamp out
198: the earlier prediction of \citet{alcock86} \citep{althaus98}.
199: Also unclear is the
200: effect non-axisymmetric mass
201: loss could have on the fraction of comet clouds lost by their hosts during
202: post main sequence evolution
203: \citep{parriott98}.
204: By hypothesis, cometary clouds are the result of
205: planet formation, so the long term evolution of planetary systems and their
206: interaction with these comet clouds needs to be investigated
207: \citep{tremaine92}.
208:
209: The problems of the \citet{alcock86} model
210: can be overcome by studying the stability of
211: planetary systems during evolution of the central
212: star as it undergoes mass loss, leaving the main sequence and evolving
213: into a white
214: dwarf. Most planetary systems are stable on timescales
215: comparable to their current age.
216: During adiabatic mass loss, companions expand their orbits
217: in
218: a homologous way, increasing their orbital semi-major axes
219: by a factor M$_i$/M$_f$ \citep{jeans24}. This change in the central
220: stellar mass affects the dynamics of the planetary system.
221:
222: The change in stellar mass specifically affects the stability planetary
223: systems, typified by the
224: Hill stability criterion against close approaches for two comparable mass
225: planets. The stability criterion is
226: roughly described as $\Delta_c=(a_1-a_2)/a_1=3\mu^{1/3}$, where
227: $a$ is the semi-major axis, $\mu$ is the mass ratio of the planets to
228: the host star, and $\Delta_c$ represents the critical separation at which
229: the two planets become unstable to close approaches \citep{hill86,
230: gladman95}. The critical
231: separation grows as the relative separation of the two planets stays the
232: same, resulting in marginally stable systems being tipped over the edge of
233: stability. This instability can lead to orbital rearrangements,
234: the ejection of one planet, and collisions \citep{ford01}.
235: These three events dramatically change the dynamical state
236: of the planetary system, leading to a fraction of systems that perturb
237: the surviving comet cloud and sending a shower of comets into the inner system
238: where they tidally disrupt, cause dust disks, and slowly settle onto the
239: WD surface. This modification of the comet impact model can explain the
240: accretion rates needed for the highest abundances of Ca observed and
241: the presence of infrared excesses around WDs \citep{debes02}.
242:
243: For two of the three above explanations, unseen planetary or substellar objects
244: lurk in the glare of nearby white dwarfs with metal lines in their atmospheres.
245: DAZs represent a promising population for a search for cool objects in orbit around WDs. If such
246: companions can be detected, this will open an exciting chapter in the study of
247: extra-solar planets by presenting several objects that can be directly detected
248: and characterized, constraining a host of theoretical issues, such as
249: extra-solar planetary atmospheres and the long term evolution of
250: Jovian planets. Such observations in the stellar graveyard can support future
251: missions dedicated to the detection and characterization of terrestrial planets. White dwarfs represent an intermediate step between our current
252: technology and what is needed for observations made with the James Webb Space
253: Telescope (JWST) and the Terrestrial Planet Finder (TPF). Coupled with the
254: possible marker of metal absorption, a sample of nearby stars easier to study
255: than main sequence stars guaranteed to have some sort of planetary system
256: could enhance the efficiency of such long term searches and may provide
257: extra clues to the nature of planet formation.
258:
259: To that end, we were motivated to search the seven brightest and closest DAZ
260: white dwarfs with the
261: NIC-2 coronograph on the NICMOS instrument of the Hubble Space Telescope (HST).
262: This search was part of the the Cycle 12 program 9834, completed over the course of 2003 and 2004 with 14 orbits. The first results from this survey
263: focused on WD 2326+049 (G 29-38), a DAZ with an infared excess \citep[][hereafter DSW05]{debes05a}. We present the observations we made in
264: Section \ref{s1} and detail our data analysis in Section \ref{s2}.
265: We present
266: candidate planetary and brown dwarf companions in Section \ref{s3} as well
267: as place limits on the types of candidates we could have detected in
268: Section \ref{s4}. Finally, we discuss the implications of our work and
269: lay out future possibilities in Section \ref{s5}.
270: \begin{deluxetable*}{lcccc}
271: %table of observations, date, time,filter
272: \tablecolumns{5}
273: \tablewidth{0pc}
274: \tablecaption{\label{tab:obs} Table of HST Observations}
275: \tablehead{
276: \colhead{WD} & \colhead{Observation Group} & \colhead{Date \& Time (UT)} &
277: \colhead{Integration Time} & \colhead{Filter}}
278: \startdata
279: 0208+396 & N8Q320010 & 2003-09-15 19:42:00 & 575.877 & F110W \\
280: & N8Q322010 & 2003-09-15 20:10:00 & 575.877 & F110W \\
281: 0243-026 & N8Q322010 & 2003-09-18 18:11:00 & 575.877 & F110W \\
282: & N8Q323010 & 2003-09-18 18:39:00 & 575.877 & F110W \\
283: 0245+541 & N8Q318010 & 2003-08-26 21:11:00 & 575.877 & F110W \\
284: & N8Q319010 & 2003-08-26 21:39:00 & 575.877 & F110W \\
285: & N8Q368010 & 2004-10-24 07:45:00 & 575.877 & F110W \\
286: & N8Q369010 & 2004-10-24 09:11:00 & 575.877 & F110W \\
287: 1257+278 & N8Q316010 & 2004-02-18 11:12:00 & 575.877 & F110W \\
288: & N8Q318010 & 2004-02-18 11:41:00 & 575.877 & F110W \\
289: 1337+701 & N8Q302010 & 2003-12-01 17:09:00 & 25.918 & F205W \\
290: & N8Q302011 & 2003-12-01 17:11:00 & 25.918 & F205W \\
291: & N8Q302020 & 2003-12-01 17:20:00 & 21.930 & F160W \\
292: & N8Q302030 & 2003-12-01 17:24:00 & 19.936 & F110W \\
293: & N8Q308010 & 2004-02-05 21:40:00 & 575.877 & F110W \\
294: & N8Q309010 & 2004-02-05 22:45:00 & 575.877 & F110W \\
295: & N8Q310010 & 2004-02-05 23:13:00 & 575.877 & F160W \\
296: & N8Q311010 & 2004-02-06 00:27:00 & 575.877 & F160W \\
297: 1620-391 & N8Q303010 & 2003-09-07 06:12:00 & 23.924 & F205W \\
298: & N8Q303011 & 2003-09-07 06:13:00 & 23.924 & F205W \\
299: & N8Q303020 & 2003-09-07 06:22:00 & 17.942 & F160W \\
300: & N8Q303030 & 2003-09-07 06:25:00 & 15.948 & F110W \\
301: & N8Q312010 & 2004-03-08 03:22:00 & 575.877 & F110W \\
302: & N8Q313010 & 2004-03-08 03:52:00 & 575.877 & F110W \\
303: & N8Q314010 & 2004-03-08 05:00:00 & 575.877 & F160W \\
304: & N8Q315010 & 2004-03-08 05:27:00 & 575.877 & F160W \\
305: 2326+049 & N8Q301010 & 2003-10-20 10:07:00 & 17.942 & F205W \\
306: & N8Q301011 & 2003-10-20 10:08:00 & 17.942 & F205W \\
307: & N8Q301020 & 2003-10-20 10:15:00 & 11.960 & F160W \\
308: & N8Q301030 & 2003-10-20 10:20:00 & 11.960 & F110W \\
309: & N8Q304010 & 2003-09-14 19:31:00 & 575.877 & F110W \\
310: & N8Q305010 & 2003-09-14 19:59:00 & 575.877 & F110W \\
311: & N8Q306010 & 2003-09-14 21:07:00 & 575.877 & F160W \\
312: & N8Q307010 & 2003-09-14 21:35:00 & 575.877 & F160W \\
313: \enddata
314: \end{deluxetable*}
315:
316: \section{Observations}
317: \label{s1}
318: Only $\sim$34 DAZs are currently known to exist, since the detection of their weak
319: metal lines are difficult without a high signal-to-noise, high resolution
320: spectrograph \citep{zuckerman03}.
321: Six of the most promising DAZ white dwarfs discovered or confirmed in the
322: \citet{zuckerman03} survey
323: were targeted for observation
324: with NICMOS and are listed in Table \ref{tab:targs}. Our seventh target,
325: WD~1620-391, was chosen for the presence of circumstellar gas absorption
326: features as well as photospheric absorption due to Si and C \citep{holberg95}.
327: We chose these targets
328: based on the fact that these were the brightest and closest DAZs known.
329: Each target was observed with
330: the NIC-2 coronagraph in the F110W
331: filter. The most promising targets, WD~2326+049, WD~1337+701, and WD~1620-391
332: were
333: imaged in the F160W filter as well. With the exception of the newly
334: discovered DAZ GD~362, both WD 2326+049 and WD~1337+701 have
335: the highest [Ca/H] abundances measured \citep{gianninas04}. WD~1620-391 was
336: chosen for extra observations
337: due to the presence of its circumstellar material.
338: These three targets were also observed
339: without the coronagraph for shorter exposures in the F110W, F160W, and F205W
340: filters in an attempt to resolve
341: any smaller structure or companions
342: at separations $<$ 0.8\arcsec. Acquisition images were used for the other
343: targets. Following the
344: prescription of \citet{fraquelli04}, two coronagraphic
345: exposures of $\sim$600 s were taken
346: at two different spacecraft roll angles. Each exposure was separated by
347: a differential roll angle of 10$^{\circ}$. The differential roll angle between images
348: limits the angular separation at which one can
349: detect a point source, requiring at least a two pixel
350: separation between the centroids of the
351: positive and negative conjugates to avoid the self
352: subtraction of any point source companions. This requirement is tempered by
353: the need to spend most of the HST orbit observing the target and not rolling
354: the spacecraft. For our observations, we concentrated on integration time and
355: chose a roll angle of 10$^{\circ}$, leading to an inner radius limit to extreme
356: high contrast imaging with self subtraction of 0.86\arcsec.
357:
358: Table \ref{tab:obs} shows a log for all of the observations taken along with
359: the total exposure times and the filter used. Each F110W observation was
360: designed to be sensitive enough to detect an object with m$_{F110W}\sim$23 with
361: a S/N of 10,
362: which for a 1 Gyr substellar object at 10 pc would correspond to a
363: $\sim$5~M$_{Jup}$ planet. For our seven targets, which range in age from 1 Gyr
364: to 7 Gyr and 10 to 34 pc, we are sensitive to 7-18 M$_{Jup}$ objects.
365:
366: \begin{figure}
367: \plottwo{f1a.eps}{f1b.eps}
368: \caption{\label{fig:g29fig} Image of WD 2326+049 in the F160W filter
369: before (left) and after (right) PSF subtraction. The right panel has been
370: Gaussian smoothed to show a candidate and two extragalactic objects. Other
371: dark features not marked are detector defects.}
372: \end{figure}
373:
374: In addition to the seven targets, three reference stars were imaged with
375: the three WDs observed without the coronagraph. The goal was to use
376: these to subtract out the point spread function (PSF) that can obscure fainter
377: objects or dust disks. These targets were chosen to be close to the original
378: target and have similar near-IR colors to aid in PSF subtraction. No close
379: companions or structure were detected using the reference stars.
380:
381: One group of observations taken of WD 0245+541 failed due to an incorrect
382: calibration onboard the telescope, with the flight software (FSW).
383: As a result, WD 0245+541 was not placed behind the coronagraphic hole. The problem
384: was identified by the HST staff and further observations did not show the
385: same problem. A repeat
386: observation was taken in October 2004, but the original failed
387: observations were also used for our data analysis.
388:
389: Due to the detection of a candiate planetary candidate around WD 2326+049 (G 29-38), second epoch observations were taken with the Gemini North Telescope
390: using the Altair adaptive optics system in conjunction with the NIRI camera.
391: Altair can successfully guide on stars with
392: R$\sim$13, such
393: as WD 2326+049. By concentrating a diffraction limited
394: fraction of the total flux of a dim object, the background can be overcome for
395: extremely faint near infrared point sources. In addition, under decent
396: observing conditions, the full width at half-maximum (FWHM) of
397: the core on Altair images is $\sim$60-90 mas,
398: providing the possibility to resolve structures better than HST \citep{hutchings04}.
399:
400: The Gemini observations were taken on August 5, 2004.
401: A total of 4 $\times$ 15s frames were co-added at 10 dither points to
402: subtract the background and to remove pixel to pixel defects, for an effective
403: integration on source of forty minutes. Our total integration returned an
404: average FWHM of 75 mas, significantly smaller than the diffraction limit of
405: our F110W images with HST.
406:
407: \section{Data Analysis}
408: \label{s2}
409: Data was reduced by the calibration pipeline provided for NICMOS. In addition
410: to the pipeline, certain steps were taken in an effort to improve the quality
411: of the final images, roughly following the
412: procedure set out by \citet{fraquelli04}.
413: Each 600 s exposure was broken up into two or
414: three exposures for ease in rejecting cosmic rays. Each calibrated subexposure
415: had pedestal subtraction by the PEDSUB routine in IRAF through the STSDAS
416: package. Each subexposure was registered and median combined with sigma
417: clipping to create a final exposure at a particular roll angle. The two
418: images at different roll angles were subtracted one from the other and vice
419: versa to create two difference images: a ROLL1-ROLL2 image and a
420: ROLL2-ROLL1 image. One difference image was rotationally registered and median
421: combined to produce the final total image. Figure \ref{fig:g29fig} shows
422: the before and after pictures of a subtraction shown at the same image stretch.
423: The residual light due to the coronagraphic PSF is dominated by systematic
424: errors, but in general is a factor of 20-50 times dimmer after subtraction.
425:
426: In the case of WD 0245+541, several other steps had to be taken
427: for the failed observation since at each roll
428: angle the star was at a different position and not behind the coronagraphic
429: hole. To combat the poor positions, the two images were registered and
430: difference images
431: were produced. The final result was of sufficient quality to determine
432: the presence of several candidate objects in the field.
433:
434: \section{Candidate Companions and Extragalactic Objects}
435: Of the seven targets, only four showed candidate companions in their fields.
436: The rest did not show anything with the exception of WD 1257+278, which had
437: a resolved galaxy in the background. Any extended objects were interpreted to
438: be background objects and all point sources were flagged as potential companions. Where second epoch images were available with 2MASS or the POSS survey,
439: they were used or second observations were taken. Each candidate with second
440: epoch images was checked for common proper motion with the target WD by measuring the relative radius and pointing angle in degrees East of North of the companion. Extragalactic objects could potentially be of interest due to their
441: proximity to a bright object that could be used for guiding in a laser AO system
442: or multi-conjugate AO system.
443:
444: The second epoch Gemini data was processed using several IRAF tasks designed by
445: the Gemini Observatory and based upon the sample scripts
446: given to observers. Each
447: frame was flatfielded and sky subtracted. In addition, due to the on-sky rotation from the Cassegrain
448: Rotator being fixed, each frame was rotationally registered
449: and combined.
450:
451: To determine if an object had common proper motion with a target WD,
452: we calculated the predicted motion of the WD on the sky based on its proper
453: motion. When comparing possible companions with 2MASS or POSS data, proper
454: motion alone was sufficient to determine objects that were in the background.
455: For WD~2326+049, WD~1620-391, and WD~0245+541, the annual parallactic
456: motion of the star was also calculated for
457: an added means of determining background point
458: sources. Any object in orbit around a WD would also have to share both
459: proper motion and annual parallactic motion.
460:
461: It is also important to
462: adequately understand the errors in order to detect any possible proper
463: motion of the background object or to determine how significant a measure of
464: common proper motion is. The greatest sources of error are due to
465: uncertainties in the parallax of the WD, proper motion, and centroiding
466: errors in the PSF of the candidate. Centroiding errors for faint sources
467: can be determined by looking at images in two filters for one of our fields
468: that has a lot of background sources. The field of WD 1620-391 has several
469: background point sources that can be compared between filters and two
470: epochs. Comparing the difference of $\sim$30 sources between the F110W and
471: F160W filters of the observation sets of N8Q312010 and N8Q314010
472: yields a standard deviation between sources of $\sim$10 mas,
473: which we will adopt as our general centroiding error.
474:
475: \label{s3}
476: \subsection{WD 2326+049}
477:
478: Figure \ref{fig:g29fig} shows the NICMOS field of view around this WD, which included a
479: candidate planetary companion, that we designated C1. In addition,
480: there were two faint extended galaxies in the field. C1
481: is discussed in more detail in DSW05 and
482: has been confirmed to be a background object with a second epoch observation
483: with the Gemini North Altair+NIRI instrument. If
484: this object had been associated, its F110W and F160W magnitudes
485: were consistent with
486: a 7 M$_{Jup}$ object.
487: Table \ref{tab:gals}
488: presents all the extragalactic objects discovered in this survey along
489: with their positions and apparent
490: Vega magnitudes in the F110W and F160W filters
491: \begin{figure}
492: \plotone{f2.eps}
493: \caption{\label{fig:g747} Field of WD 0208+395 with its candidates. Candidates
494: are circled and the WD is masked to hide the systematic subtraction errors.
495: A galaxy is detected in the lower right of the image.}
496: \end{figure}
497:
498: \subsection{WD 0208+395}
499:
500: Figure \ref{fig:g747} shows two candidate objects, C1 and C2
501: and a galaxy in the field of WD~208+395.
502: Since the separation between these objects and WD~208+395 were greater than a few arcseconds, we pursued a
503: second observation with the Canada France Hawaii Telescope with the PUEO+KIR
504: instruments.
505: A second epoch image shows that both C1 and C2
506: are in the background. This result is discussed in detail in \citet{debes05c}.
507: If they had been associated, C1 was consistent with
508: a 3 Gyr old 15 M$_{Jup}$ brown dwarf and C2 consistent with
509: a 10 M$_{Jup}$ planet.
510:
511: \begin{figure}
512: \plotone{f3.eps}
513: \caption{\label{fig:g174} Field of WD 0245+541 with its candidates circled.}
514: \end{figure}
515:
516: \subsection{WD 0245+541}
517: This object, due to its failed first observation, was re-imaged $\sim$ one year
518: later (see Table \ref{tab:targs}) which provided an ample baseline to test candidates for common proper
519: motion.
520: Figure \ref{fig:g174} shows the surrounding area of WD 0245+541, along with three
521: candidates in the field. C1 appears to be a binary object at a distance
522: of $\sim$3\arcsec\, which in the second epoch image is clearly not co-moving.
523: C2 is at a separation of $\sim$6\arcsec\
524: and 270$^\circ$ PA.
525: Inspection of the POSS2 red image of this field clearly shows a point source
526: at a separation consistent with this object being a background source.
527: Finally, C3 has a separation of 2.5$\pm0.02$\arcsec\ and 348$\pm$1$^\circ$ PA. WD 0245+541 has a predicted motion between the two epochs of -667 mas
528: and -475 mas, leading to a predicted $\Delta\alpha$=0.17$\pm$0.12\arcsec\ and
529: $\Delta\delta$=2.94$\pm$0.12\arcsec\ if C3 is non co-moving, compared to the
530: observed $\Delta\alpha$=0.07 and
531: $\Delta\delta$=2.86. The candidate does not have common proper motion, and is
532: therefore a background object.
533: The main source of error was in the reported proper motion,
534: which had quoted errors of 0.1\arcsec yr$^{-1}$ \citep{bakos02}.
535: If C3 had been associated, its F110W magnitude would have been
536: consistent with an 18 M$_{Jup}$ brown dwarf
537: companion.
538:
539: \begin{deluxetable*}{cccccc}
540: %table of extragalactic objects
541: \tablecolumns{6}
542: \tablewidth{0pc}
543: \tablecaption{\label{tab:gals} Table of Extragalactic Objects}
544: \tablehead{
545: \colhead{DSW \#} & \colhead{RA} & \colhead{Dec} & \colhead{F110W} & \colhead{F160W} & \colhead{Notes}
546: }
547: \startdata
548: 1 & 02 11 20.51 & +39 55 14 & 21.36$\pm$0.04 & & \\
549: 2 & 12 59 45.63 & +27 34 01 & 22.8$\pm$0.1 & & $\sim$1.4\arcsec\ extent \\
550: 3 &23 28 47.96 & +05 14 38 & 23.7$\pm$0.2 & 22.1$\pm$0.1 & 0.23\arcsec\ aperture
551: \\
552: 4 & 23 28 47.67 & +05 14 40 & 24.0$\pm$0.2 & 22.8$\pm$0.2 & 0.23\arcsec\ aperture \\
553: \enddata
554: \end{deluxetable*}
555:
556: \subsection{WD 1620-391}
557:
558: WD 1620-391 resides quite near the galactic plane and as such
559: has an extremely crowded field with $\sim$36 sources of varying brightnesses,
560: which can be seen in Figure \ref{fig:wd1620}.
561: Any possible companion must
562: be separated from background objects. A viable candidate
563: in this field would have to be selected by an F110W-F160W color
564: being consistent with a substellar object. Since most of these objects are
565: background objects we must first see if there is any evidence to suspect that
566: there would be a candidate in this field rather than assuming that all
567: sources were background objects. The number of objects as a function of
568: distance should be $\propto\ r^2$ if the background distribution is truly
569: random. A different distribution would be caused either by the presence of
570: objects physically associated to the central white dwarf or due to physical
571: associations among background stars, such as binaries or clustering. To
572: look for a departure from the expected distribution,
573: we plotted the number of sources in the WD1620-391 field as a function of
574: radial distance from the WD as shown in Figure \ref{fig:dist}.
575: We compared this distribution to a pure $r^2$
576: distribution through means of a K-S test. We find that there is a
577: 97\% probability that the distribution is not based on the $r^2$ distribution
578: mainly due to the hump of sources present close to the WD. We believe that
579: those objects are viable candidates and that in a statistically
580: significant way the distribution of sources $<$ 4\arcsec\ is fundamentally
581: different than what would be expected. A caveat, however, is that since
582: the WD is at a low galactic latitude the statistical test may merely be
583: detecting some fundamental structure in the background sources rather than
584: the presence of a candidate. Additionally, the scenario of \citet{debes02}
585: would predict more than one planet in the system to efficiently slingshot
586: comets or asteroids to the surface of the white dwarf, so the potential exists
587: that two planetary candidates could be present in this ``hump'' of sources
588: $<$ 4\arcsec.
589:
590: \begin{figure}
591: \plotone{f4.eps}
592: \caption{\label{fig:wd1620} Field of WD 1620-391 with its candidates. Each
593: candidate that is circled is within 4\arcsec\ and has colors consistent within
594: the photometric errors to a candidate planetary object.}
595: \end{figure}
596:
597: \begin{deluxetable}{ccccc}
598: %table of candidates for WD1620
599: \tablecolumns{5}
600: \tablewidth{0pc}
601: \tablecaption{\label{tab:wd1620com} Candidates around WD1620-391}
602: \tablehead{
603: \colhead{Candidate} & \colhead{R} & \colhead{PA} & \colhead{F110W} & \colhead{F160W}
604: }
605: \startdata
606: C1 & 2.22\arcsec$\pm$0.09 & 328.5$^\circ\pm$0.7 & 22.9 & 21.6 \\
607: C2 & 2.56\arcsec$\pm$0.13 & 262$^\circ\pm$5 & 22.9 & 21.8 \\
608: C3 & 3.10\arcsec$\pm$0.10 & 265$^\circ\pm$3 & 22.4 & 21.0 \\
609: C4 & 3.13\arcsec$\pm$0.14 & 141$^\circ\pm$1 & 23.9 & 23.0 \\
610: C5 & 3.24\arcsec$\pm$0.12 & 129.6$^\circ\pm$0.8 & 22.7 & 21.5 \\
611: C6 & 3.63\arcsec$\pm$0.17 & 27$^\circ\pm$2 & 22.5 & 21.2 \\
612: C7 & 3.91\arcsec$\pm$0.11 & 279$^\circ\pm$2 & 22.9 & 21.8 \\
613: \enddata
614: \end{deluxetable}
615:
616: Regardless, we have plotted all the detected sources in a CMD and compared them
617: to a predicted isochrone of substellar objects in Figure \ref{fig:cmd}.
618: The WDs age is $\sim$1 Gyr so we used the 1 Gyr
619: models of \citet{bsl03} convolved with the HST filters. There
620: are some candidates that are within 4\arcsec\ and who have colors consistent
621: within the errors to be a planetary candidate. Table \ref{tab:wd1620com} lists the candidates, their magnitudes in F110W and
622: F160W. Every one of the candidates would be $\sim$5-6 M$_{Jup}$ in mass if
623: associated. This WDs proper motion is
624: $\sim$75 mas/yr in RA and $\sim$0 mas/yr in Dec \citep{hip},
625: so a second epoch image will be necessary in ruling out
626: any of these sources.
627:
628:
629: \begin{figure}
630: \plotone{f5.eps}
631: \caption{\label{fig:dist} Distribution of point sources as a function of
632: distance around the white dwarf WD 1620-391. }
633: \end{figure}
634:
635: While we have two sets of observations for WD~1620-391 separated by six months,
636: our first image is not sensitive enough to conclusively detect
637: any of our candidate companions. Six stars were bright enough to use as a
638: background grid of reference stars compared to WD 1620-391's position.
639: Of these six, five were distinct
640: point sources. The sixth appears to be extended,
641: either because it has a disk or because
642: it is a binary. When comparing the relative
643: position between these presumably stationary objects in six months and WD 1620-391, we measured a change in RA of 204$\pm$10 mas and in Dec of 16$\pm$10 mas.
644: We derived the error based on the standard deviation of the individual measurements from the mean. Taking into account WD 1620-391's parallax motion during
645: this period, one would expect a motion of 230 mas in RA and 28 mas in Dec
646: assuming WD 1620-391's reported parallax of 78.85 mas \citep{hip}. Subtracting
647: this motion leaves 26$\pm$10 mas and 12$\pm$10 mas from the measured
648: motion with our reference stars, suggesting that we can
649: detect common proper motion and common parallactic motion in a future epoch
650: with HST and these reference stars. Since we have successfully proposed for HST time in Cycle 14
651: to follow up these candidates, we expect to have a long enough baseline to
652: definitively determine if any of the candidates are physically associated.
653:
654: \section{Limits to Companions}
655: \label{s4}
656: The main goal of this search was to detect candidate companions, but upper
657: limits to the detection of such companions is also important for
658: understanding the true nature of DAZ WDs, as well as the process of planet and
659: brown dwarf formation around intermediate mass stars. To this end, in this
660: Section we quantify our sensitivity to companions that could have
661: been detected, in order
662: to determine the frequency of high mass planets and brown dwarfs.
663:
664: \begin{figure}
665: \plotone{f6.eps}
666: \caption{\label{fig:cmd} Color magnitude diagram of sources near the white dwarf WD 1620-391. Overplotted is an isochrone of 1 Gyr substellar models from
667: \citet{bsl03} convolved with HST filters at 12 pc. Thick crosses are sources $<$4\arcsec\ away.}
668: \end{figure}
669:
670: \subsection{Near-IR Photometry}
671: While direct imaging is most sensitive to companions $>$0.9\arcsec\,
672: unresolved companions could still be present for some of these targets. In
673: order to rule out companions at separations where imaging or PSF subtraction
674: could not resolve them, we turn to the near-infrared fluxes of these objects
675: provided by near-IR photometry, such as from 2MASS \citep{cutri03}. Looking
676: in the near-IR can
677: facilitate the discovery of cool objects around WDs \citep{probst82,zuckerman92,green00}.
678:
679: Our strategy was to take model values reported in the literature,
680: generate predicted 2MASS J, H, and Ks magnitudes by using the models
681: of \citet{bergeron95} and comparing
682: J$_{th}$, H$_{th}$, and K$_{th}$ with the observed magnitudes of the WDs. For our sample of
683: white dwarfs we took model values of T$_{eff}$, $\log{g}$, and the mass from
684: \citet{liebert04}, \citet{bergeron01}, and \citet{bragaglia95}.
685:
686: To compare the predicted magnitudes to those observed we took the difference
687: of the predicted magnitdues in the 2MASS filter system J$_{th}$, H$_{th}$, and K$_{s(th)}$ and the observed magnitudes J, H, and K$_{s}$.
688: A significant positive value would indicate an excess due to either an unseen
689: companion or a dust disk, while a significant negative value would indicate an
690: anomalous paucity of flux. While
691: we used the results of DSW05 for two of our white dwarfs for the rest of our targets we used the \citet{bergeron01} and \citet{bragaglia95} samples since they provide atmospheric parameters for the remaining five white
692: dwarfs. In general, we compared J magnitudes since WD 2326+049 has an
693: infrared excess due to a dust disk at wavelengths longer than $\sim$1.6~\micron. Excesses in J tend to be more sensitive because J band photometric errors
694: are smaller in 2MASS. For the rest of the targets we also checked to see if
695: there were excesses in any of the other bands or for other targets in the
696: sample. An excess was considered significant if it was greater than three
697: times the measured scatter of a sample and if it was present in more than one
698: filter.
699:
700: We tested the accuracy of the three samples of WD parameters to reliably report a 3$\sigma$
701: excess limit. We first examined the \citet{bergeron01} sample, which includes
702: WD 0208+395, WD 0245+541, WD 0243-025, and WD 1257+278.
703: Of the 150
704: white dwarfs we chose 146 of the sample that had reliable
705: photometry from \citet{bergeron01}
706: and converted their MKO magnitudes to 2MASS magnitudes
707: \footnote{http://www.ipac.caltech.edu/2mass/releases/allsky/doc/sec6\_4b.html}
708: to compare with our predicted magnitudes.
709:
710: We neglected any object
711: with an excess $>$ 3 $\sigma$ and recalculated the scatter in expected minus
712: observed magnitudes, repeating the process three times. We ensured that the
713: median values of the differences were consistent with zero.
714: From the 146 WDs we find that the 1~$\sigma$ error in total
715: of J, H, and Ks are 0.04, 0.04, and 0.05 mag. One important note is that \citet{bergeron01} used their JHK photometry to help fit several of the parameters
716: that we used to generate our theoretical magnitudes, namely $\log{g}$ and
717: T$_{eff}$. For this reason we had to be more careful intrepeting these limits
718: because it is possible the presence of a companion was ``fitted out''.
719: In this
720: case we are
721: placing limits to what kind of excess would have been detected by the
722: models, rather than extrapolating from the models and looking for excesses.
723: No objects in this sample showed a significant excess.
724:
725: For WD 2326+049 and WD 1337+705 we took the sample
726: of \citet{liebert04}
727: which is a study of DA WDs from the Palomar-Green survey
728: of UV excess sources. Of the 374
729: white dwarfs we chose the brightest 72 of the sample that had
730: a J $<$ 15, had unambiguous sources in 2MASS, and had reliable
731: photometry, i.e those objects that had quality flags of A or B in the 2MASS
732: point source catalogue for their J magnitudes. After determining the standard deviation of the sample, we found that 1$\sigma$\
733: errors for the sample in the J, H, and K$_s$ filters were 0.07, 0.10, and 0.15 mag, respectively. Further details of the \citet{liebert04} sample are presented in DSW05.
734:
735: For WD 1620-391, we needed to use
736: the sample in \citet{bragaglia95}, using $\sim$35 of the 50 WDs modeled in that
737: work. We again picked WDs with V$<$15, reliable 2MASS positions, and
738: reliable photometry in the three bands.
739: Six white dwarfs had poor photometry or incorrect distance moduli,
740: but these errors were corrected.
741: The final errors were
742: calculated, resulting in 1$\sigma$ errors of 0.09, 0.08, 0.15 mag
743: for J, H, and Ks
744: respectively. Two WDs remained with significant excess, WD 1042-690, and
745: WD 1845+019. WD 1042-690 is a known binary system with a dM companion, and
746: WD 1845+019 does not currently seem to be a candidate for an excess.
747: However, its
748: position in both the POSS and 2MASS plates based on the position given
749: by \citet{lanning00} shows that it is
750: blended with another point source.
751: Inspection of the POSS and 2MASS plates leaves it ambiguous whether this
752: barely resolved object (separcdation $\sim$3\arcsec) is co-moving or not, so we
753: mark this as a potential common proper motion WD/dM pair.
754:
755: Table \ref{tab:phot} shows the expected 2MASS magnitudes
756: based on the model values,
757: and the observed magnitudes of our target white dwarfs. All of our targets
758: fall within 1-2$\sigma$ of our expected values for all three filters, with the
759: exception of WD 2326+049, as mentioned above.
760:
761: Since none of our targets have significant excesses, we can use the
762: 3$\sigma$ limits in J to place upper limits to
763: unresolved sources. We took the predicted J magnitudes
764: from substellar atmosphere models, corrected for distance modulus,
765: calculated the excess, and compared it to
766: our sensitivity limit \citep{baraffe98,baraffe03}.
767: Table \ref{tab:sens} shows the unresolved companion
768: upper limits for each target. Any companion with a mass beyond the hydrogen
769: burning limit would have been detected for all of the target WDs.
770: \begin{deluxetable}{ccccccc}
771: \tablecolumns{7}
772: \tablewidth{0pc}
773: \tablecaption{\label{tab:phot} Comparison of Predicted vs. 2MASS Photometry}
774: \tablehead{
775: \colhead{WD} & \colhead{J$_{th}$} & \colhead{H$_{th}$} & \colhead{K$_{s(th)}$}
776: & \colhead{J} & \colhead{H} & \colhead{K$_{s}$} }
777: \startdata
778: 0208+396 & 13.74 & 13.61 & 13.57 & 13.76 & 13.66 & 13.61 \\
779: 0243-026 & 14.65 & 14.49 & 14.43 & 14.67 & 14.50 & 14.49 \\
780: 0245+541 & 13.86 & 13.61 & 13.47 & 13.86 & 13.67 & 13.58 \\
781: 1257+278 & 14.95 & 14.89 & 14.88 & 14.95 & 14.92 & 14.89 \\
782: 1337+701 & 13.23 & 13.36 & 13.41 & 13.25 & 13.36 & 13.45 \\
783: 1620-391 & 11.53 & 11.66 & 11.74 & 11.58 & 11.71 & 11.77 \\
784: 2326+049 & 13.13 & 13.19 & 13.22 & 13.13 & 13.08 & 12.69 \\
785: \enddata
786: \end{deluxetable}
787:
788: \subsection{Imaging}
789: \label{s4.2}
790:
791: \citet{schneider03} showed
792: a reliable way to determine sensitivity of an observation with
793: NICMOS, given the stability of the instrument. Artificial ``companions'' are
794: generated with the HST PSF simulation software TINYTIM \footnote{http://www.stsci.edu/software/tinytim/tinytim.html} and scaled until they are recovered.
795: These companions are
796: inserted into the observations and used to gauge sensitivity. We adopted
797: this strategy for our data as well. An implant was placed in the images.
798: Two difference
799: images were created following our procedure of PSF subtraction
800: and then rotated and combined for maximum signal to noise.
801: Sample images were looked
802: at by eye as a second check that the dimmest implants could be recovered.
803: The implants were normalized so that their total flux was equal to
804: 1 DN/s. The normalized value was converted to a flux in Jy or a Vega magnitude by
805: multiplying by the correct photometry constants given by the NICMOS Data Handbook.
806: We considered an implant recovered if its scaled
807: flux in a given aperture had a S/N of 5.
808:
809: \begin{figure}
810: \plotone{f7.eps}
811: \caption{\label{fig:wdsens} Sensitivity at 5$\sigma$ to point sources in F110W
812: around the WD WD 0208+395. The WDs F110W magnitude is $\sim$13.8, giving a contrast
813: of 10 magnitudes at 1\arcsec. Overplotted is the magnitude of a 10 M$_{Jup}$
814: planet 3.2 Gyr old at the distance of WD 0208+395 from the models of \citet{bsl03}.}
815: \end{figure}
816:
817: For our Gemini data, we
818: used the PSF of WD 2326+049 as a reference for the implant. The implant was normalized to a peak pixel value of one. Scaled versions of the implants were then
819: used to determine the final image's
820: sensitivity to objects at a S/N of 10, since siginificant
821: flux from the PSF remained at separations $<$ 1\arcsec. The relative flux of the implant with respect
822: to the host star was measured and a corresponding MKO
823: H magnitude was derived from the 2MASS H magnitude to give a final apparent
824: magnitude sensitivity. For our Gemini images we checked sensitivity starting
825: at a distance of $\sim$3 times the FWHM of WD 2326+049, or 0.22\arcsec,
826: out to 1\arcsec.
827:
828: Figure \ref{fig:wdsens} shows an example of the azimuthally averaged 5 $\sigma$
829: sensitivity for WD 0208+395. In order to determine the total age of WD 0208+396 as well as each other system, we took each inferred mass and
830: derived an initial mass by the relation $10.4\ln[(M_{WD}/\Msun)/0.49] \Msun$, the results of which are given
831: in Table \ref{tab:targs} \citep{wood92}. The mass then gave a main sequence lifetime given by 10 t$_{MS}^{-2.5}$~Gyr, which gave a total age when coupled with the inferred
832: cooling time from the same models used for our 2MASS photometry.
833: With a total estimated system age of $\sim$3
834: Gyr for WD 0208+395, we overplotted the lowest companion mass detectable, using
835: the models of \citet{bsl03}. These models differ slightly from the
836: models of \citet{baraffe03}, used for our 2MASS excess limits. The \citet{bsl03} models tend to predict dimmer near-IR magnitudes for
837: the planetary mass objects, but converge with the \citet{baraffe03} models
838: for higher masses. It is therefore possible that we are sensitive to objects
839: $\sim$1-2 M$_{Jup}$ less massive if the \citet{baraffe03} models are
840: correct. Table \ref{tab:sens}
841: has the mass limits for each WD for separations $>$ 0.9\arcsec.
842:
843: \section{Discussion}
844: \label{s5}
845:
846: We can use Table \ref{tab:sens} and the results of our excess limits to draw
847: some broad conclusions from this search. The combination of the 2MASS excess
848: determinations and the HST imaging create the most sensitive search for
849: planets around WDs to date. The sensitivity achieved could easily have detected
850: an object $>$ 10 M$_{Jup}$ at separations $>$ 30.6 AU, with the closest
851: detection possible at 9.3 AU. Taking into account that any primordial
852: companions' semi-major axis would have expanded by a factor of $M_i/M_f$, we
853: can infer the closest primordial separation these objects would have had if
854: they had been
855: detected. Taking the values of Table \ref{tab:targs} for M$_i$ and M$_f$ and
856: using our minimum projected angular separation, we
857: find that any object that formed at $>$10 AU could have been
858: detected, assuming that there were no forces that retarded expansion. Forces
859: that could retard expansion would be
860: due to tidal interactions with the giant star. However, this effect should be minimal at initial distances of 10 AU \citep{rasio96}.
861:
862: We can also make some initial comments about the origin of DAZ white dwarfs.
863: Given the upper limits on unresolved companions, we can infer the plausibility
864: of one of the possible explanations for the DAZ phenomenon. The problems with
865: ISM accretion have been documented extensively in the work of \citet{zuckerman03} and \citet{aannestad93}. In \citet{zuckerman03} they noted that a large fraction of
866: DA/dM objects had metal absorption lines in their atmospheres, and inferred
867: that other DAZs may be the result of unseen companions. If this scenario is
868: true, then for each of these objects, the maximum companion mass plausible
869: is $<$70 M$_{Jup}$.
870:
871: If the explanation
872: for DAZs is due to close brown dwarf companions, the frequency of DAZs is at
873: odds with the frequency of DAZs one would predict based on radial velocity
874: surveys. These surveys find that $\sim$0.5\% of stars have brown dwarfs with
875: semi-major axes $<$3~AU \citep{marcy00}. One would expect 0.5\% or less of field DAs to be
876: DAZs based on the radial velocity result. The only possible counter explanation is that brown dwarf
877: formation at these radii is $\sim$40 times more efficient for
878: higher mass main sequence stars.
879: Radial velocity surveys of G giants are too young
880: to reliably estimate the fraction of brown dwarf companions in orbits wider than $\sim$1 AU, but none have yet been found in $\sim$100 stars
881: \citep{sato03}.
882:
883: \begin{deluxetable*}{lcccc}
884: %table of resolved and unresolved sensitivities
885: \tablecolumns{5}
886: \tablewidth{0pc}
887: \tablecaption{\label{tab:sens} Upper limits to Companions}
888: \tablehead{
889: \colhead{WD} & \colhead{Excess Limit (m$_{J}$)} & \colhead{Mass} & \colhead{Sensitivity $>$ 0.9\arcsec} & \colhead{Mass} \\
890: & \colhead{(J)} & \colhead{(M$_{Jup}$)} & \colhead{(F110W)} & \colhead{(M$_{Jup}$)}}
891: \startdata
892: 0208+396 & 16.2 & 48 & 23.9 & 10 \\
893: 0243-026 & 17.5 & 51 & 24.1 & 10 \\
894: 0245+541 & 16.2 & 53 & 23.5 & 18 \\
895: 1257+278 & 17.0 & 40 & 23.8 & 14 \\
896: 1337+701 & 14.9 & 70 & 23.4 & 14 \\
897: 1620-391 & 12.9 & 61 & 22.9 & 7 \\
898: 2326+049 & 14.8 & 39 & 23.3 & 6 \\
899: \enddata
900: \end{deluxetable*}
901:
902: We can compare our results with those of radial velocity surveys. By comparing
903: both results we can look at predictions for the frequency of massive
904: planets around
905: a random sample of stars and around stars that possess planetary systems.
906: Since the numbers are small, we will merely look at percentages and assume that
907: they are constant as a function of distance and central stellar mass,
908: clearly naive assumptions. Since 5\% of field stars have
909: planetary systems, we need to estimate how many would have planets
910: massive enough to be detectable by our observations. Of the
911: 118 known planetary systems in orbit around solar type stars, $\sim$6
912: have companions with M$\sin i$ $>$ 10 M$_{Jup}$\footnote{http://www.obspm.fr/encycl/encycl.html}. The frequency of such planets amongst stars already
913: bearing one or more planets is then $\sim$5\%, leading to an overall
914: probability of 0.25\% of all field stars possessing a planet that we could
915: have detected. Assuming Poisson statistics,
916: to have a 50\% chance
917: at detecting one or two
918: planets would require a sample of 400 WDs with ages $\sim$3 Gyr.
919: The limit sensitive radial velocity studies have on
920: A stars can be partially circumvented
921: by searching G giants for radial velocity variations \citep{sato03}. G giants
922: are typically intermediate mass stars, although field giants tend to have
923: larger uncertainties in their mass compared to the main sequence stars in
924: other radial velocity surveys. As of the results published in \citet{sato03},
925: one planetary object with M$\sin{i}$=6-10 M$_{Jup}$ and semi-major axis
926: $\sim$1~AU had been detected in a
927: sample of $\sim$100 targets.
928: The implied frequency of $\sim$1\% would mean a slightly more favorable chance
929: to find one planet in a sample of $\sim$100 WDs.
930: If DAZs do not preferentially harbor planetary systems, it will be a long search if we only focus on them. Any search should include DAZs, but also focus
931: on a larger sample.
932:
933: Let us now consider the possibility that DAZs do preferentially harbor planetary systems, and based on our detection limits estimate
934: how many DAZs would need to be
935: observed. Since we could detect $>$ 10~M$_{Jup}$ objects and $\sim$5\% of
936: field stars with planetary systems have objects that massive, we can infer that
937: 5\% of DAZs could have planets that could have been detected.
938: If DAZs (and also
939: DZs or helium white dwarfs with metal absorption) are indeed
940: good markers for planetary systems, one would need a sample of
941: 20 WDs to have a 50\% chance to
942: detect a massive planet. To date $\sim$34 DAZs are known.
943: Currently the estimated fraction of
944: apparent single WDs that are DAZs is $\sim$20\%. If they all harbor planets, this estimated fraction implies a much higher
945: frequency of planets than that measured by radial velocity surveys. However,
946: radial velocity
947: surveys are starting to detect longer period systems, which may have a
948: higher frequency of formation and better represent the type of population
949: that would cause a DAZ \citep{jones02}.
950:
951: There are then two approaches to continuing the search--increasing the sample
952: size and increasing the sensitivity of a search. In the short term a large sample of WDs must be observed, since the
953: probable frequency of massive planets among WDs that harbor a planetary system
954: is small.
955: Future observatories such as the James Webb
956: Space Telescope should have an easier time detecting Jovian and sub-Jovian
957: planets, which will hopefully resolve the origin of DAZs. Such future observations will determine whether DAZs are
958: ultimately useful for planetary studies, including spectroscopy.
959:
960: The discovery of candidate planetary mass companions demonstrates that this
961: limited survey was sensitive to planets. These results
962: show that if massive planets were present
963: around these WDs we would have detected them. Even with a small sample, limits
964: can be placed on the frequency of massive planets in orbit around stars more
965: massive than the Sun, and begin to observationally address the question of
966: planet formation efficiency vs. spectral type. Ideally, the next step would be
967: to be to expand the sample of WDs studied and to probe to lower masses, where the planetary mass function peaks ($\sim$1 M$_{Jup}$). High spatial resolution and sensitivity missions like JWST would most
968: likely be able to detect such objects around nearby WDs.
969:
970: \acknowledgements
971: We would like to gratefully acknowledge Al Shultz and Glenn Schneider for
972: helpful conversations about coronagraphy with NICMOS, and Chad Trujillo and Joe
973: Jensen for critical help with the inner workings of Altair and the reduction
974: of Altair imaging data.
975:
976: Based on observations made with the NASA/ESA Hubble Space Telescope, obtained
977: at the Space Telescope Science Institute, which is operated by the
978: Association of Universities for Research in Astronomy, Inc., under NASA
979: contract NASĘ5-26555. These observations are associated with program \#9834.
980: Also based on observations obtained at the Gemini Observatory,
981: which is operated by the
982: Association of Universities for Research in Astronomy, Inc.,
983: under a cooperative agreement
984: with the NSF on behalf of the Gemini partnership: the National
985: Science Foundation (United
986: States), the Particle Physics and Astronomy Research Council
987: (United Kingdom), the
988: National Research Council (Canada), CONICYT (Chile), the Australian
989: Research Council
990: (Australia), CNPq (Brazil) and CONICET (Argentina). Near-IR Photometry obtained as part of the Two Micron All Sky Survey (2MASS), a joint project of the University
991: of Massachusetts and the Infrared Processing and Analysis Center/California
992: Institute of Technology, funded by the National Aeronautics and Space
993: Administration and the National Science Foundation. S.S. also acknowledges
994: funding under the Pennsylvania State University Astrobiology Research Consortium (PSARC).
995:
996:
997: \bibliography{g29bib}
998: \bibliographystyle{apj}
999:
1000:
1001: \end{document}
1002:
1003:
1004:
1005:
1006: