FIRST RESULTS FROM Pan-STARRS1: FAINT, HIGH PROPER MOTION WHITE DWARFS IN THE MEDIUM-DEEP FIELDS

The Pan-STARRS1 system is a high-etendue wide-field imaging system, designed for dedicated survey observations. The system is installed on the peak of Haleakala on the island of Maui in the Hawaiian island chain. Routine observations are conducted remotely from the Maikalani Advanced Technology Center. We provide below a terse summary of the Pan-STARRS1 survey instrumentation. A more complete description of the Pan-STARRS1 system, both hardware and software, is provided by Kaiser et al. (2010). The survey philosophy and execution strategy are described in K. C. Chambers et al. (2011, in preparation).

The Pan-STARRS1 optical train (Hodapp et al. 2004) uses a 1.8 m diameter f/4.4 primary mirror and a 0.9 m secondary. The resulting converging beam then passes through two refractive correctors, an interference filter with a clear aperture diameter of 496 mm and a final refractive corrector that is the dewar window.

The Pan-STARRS1 imager (Tonry et al. 2008) comprises a total of sixty 4800 × 4800 pixel detectors, with 10 μm pixels that subtend 0258. The diameter of the field illuminated by the optical system is 33. The detectors are back-illuminated CCDs manufactured by Lincoln Laboratory. The detectors are read out using a StarGrasp CCD controller (Onaka et al. 2008), with a readout time of 7 s for a full unbinned image.

The Pan-STARRS1 passbands are designated as g_P1, r_P1, i_P1, z_P1, and y_P1 in order to clearly distinguish PS1 from other photometric systems. Photometry in the PS1 system is on the AB magnitude system (Fukugita et al. 1996).

Images obtained by the Pan-STARRS1 system are processed through the Image Processing Pipeline (IPP), on a computer cluster at the Maui High Performance Computer Center. The pipeline runs the images through a succession of stages, including flat fielding ("de-trending"), a flux-conserving warping to a sky-based image plane, masking and artifact removal, and object detection and photometry (Magnier 2006). The IPP also performs image subtraction to allow for the prompt detection of variables and transient phenomena. For the results presented here, the flat-fielded and warped Medium-Deep images were processed through a custom stacking and calibration process, as described in Section 3. Each night's data are collected with eight dithers, permitting an outlier rejection strategy that removes artifacts such as cosmic rays or satellite streaks.

The Pan-STARRS1 observations are obtained through a set of five broadband filters, which we have designated as g_P1, r_P1, i_P1, z_P1, and y_P1. Under certain circumstances Pan-STARRS1 observations are obtained with a sixth, "wide" filter designated as w_P1 that essentially spans g_P1, r_P1, and i_P1. Although the filter system for Pan-STARRS1 has much in common with that used in previous surveys, such as SDSS (York et al. 2000), there are important differences. The g_P1 filter extends 20 nm redward of g_SDSS, paying the price of 5577 Å sky emission for greater sensitivity and lower systematics for photometric redshifts, and the z_P1 filter is cut off at 930 nm, giving it a different response than the detector response defined z_SDSS. SDSS has no corresponding y_P1 filter. We stress that, like SDSS, Pan-STARRS1 uses the AB photometric system and there is no arbitrariness in the definition, only in how accurately we know the bandpasses.

The details of the photometric calibration and the Pan-STARRS1 zero-point scale will be presented in a subsequent publication (J. L. Tonry et al. 2011, in preparation), and E. Magnier et al. (2011, in preparation) will provide the application to a consistent photometric catalog over the 3/4 sky observed by Pan-STARRS1. We briefly describe the methodology used for the photometry presented in this paper.

We have carried out extensive, in situ measurements of the full transmission of the Pan-STARRS1 system (Stubbs et al. 2010), as well as knowing the filter, lens, and mirror characteristics from the vendor's measurements, and we have found these to be consistent. Provisional response functions (including 1.2 airmasses of atmosphere) are available at the project's Web site.⁶

Integrating the Pan-STARRS1, SDSS, and Johnson/Kron-Cousins bandpasses against spectrophotometry of 273 stars with a wide range of temperature and surface gravity gives us a means of transforming SDSS or J/K − C colors to the Pan-STARRS1 system for the stellar locus. Table 1 relates the Pan-STARRS1 color to a corresponding SDSS color: C_P1 = A + B × C_SDSS, valid over a certain range in C_SDSS. Since both systems are AB the constant A is negligible, and the relative redness of g_P1 and blueness of z_P1 relative to SDSS are reflected in B < 1. The term for y_P1 is purely a stellar locus extrapolation off of the SDSS system and should be used with caution, particularly for hot stars with Paschen absorption.

While the catalog of E. Magnier et al. (2011, in preparation) will use PS1 as a photometric instrument to link observations to fundamental standards such as helium WD observed by Hubble Space Telescope (supplanting standards such as Vega and BD+17 4708), the photometry presented here is based on SDSS DR7. Each Medium-Deep field observation in a filter is brought into relative calibration with every other observation, and then overlap with SDSS stars whose magnitudes have been transformed into the Pan-STARRS1 system provides a single zero point for all. We finally use the stellar locus of Covey et al. (2007) transformed to the Pan-STARRS1 system to provide cross-filter zero-point tweaks and create the most consistent colors possible. The procedure is described in detail below, but it is essential to emphasize that the photometry here is on the Pan-STARRS1 system, not on the SDSS system, although SDSS is the basis for zero points and colors. Like SDSS, the Pan-STARRS1 photometry includes 1.2 airmasses of atmospheric attenuation as a factor in the bandpasses, although the magnitudes are corrected to the top of the atmosphere. No correction is made for galactic extinction.

The Pan-STARRS1 IPP system performs flat fielding on each of the images, using white light flat-field images from a dome screen, in combination with an illumination correction obtained by rastering sources across the field of view. Bad pixel masks are applied and carried forward for use in the stacking stage. After determining an initial astrometric solution, the flat-fielded images are then warped onto a tangent plane of the sky, using a flux conserving algorithm. The plate scale for the warped images is 0.200 arcsec pixel⁻¹. The IPP removes the sky level from the images, but at this point makes no attempt to provide a consistent flux scale. Images obtained through clouds are included in the processing chain. We adjust the flux scale of each image at the level of a Pan-STARRS1 "skycell," which subtends 20 arcmin on the sky. There is no evidence for residual spatial structure in the attenuation from clouds in the resulting Medium-Deep field nightly stacks, each combining eight images of integration time over 100 s.

The images obtained from a single night in each band were typically obtained with small dithers in boresight and at a diversity of rotator angles. The stack of eight images per band per night were then assembled into a "nightly stack," using a variance-weighted combination of individual frames, with outlier rejection. These nightly stacked images are considered to be that night's image in the appropriate band. This stacking process successfully suppresses cosmic ray artifacts, masking losses, and any sources that move a distance larger than the PSF during the ∼30 minute observation interval.

Nightly stacks are combined into "stack–stacks" by weighting each by inverse variance times inverse PSF area. We find that this is nearly optimal for point-source detection. In principle we might do slightly better by convolution with PSF, but in practice the covariance between warped pixels is such as to make the improvement negligible. The resulting PSF of this stack–stack has a relatively sharp core and substantial skirt, of course, but we do not attempt to deconvolve the skirt into a more compact PSF, even though it would be a relatively robust operation. Instead, we always strive to match PSF models to the data or include skirt deconvolution as part of the normal kernel convolution required when image subtracting the stack–stack from a nightly stack.

In the interim, before the Pan-STARRS1 photometric catalog of the sky is released, we must resort to a somewhat involved procedure to bring all observations to a common and accurate zero point. In brief, these steps are applied to each set of observations of each Medium-Deep field, every 20 arcmin skycell of each field (about 70 per field) and each filter which are the following.

• 1.  
  Obtain instrumental magnitudes (fluxes) and positions for all stars in all nightly stacks using DoPhot (Schechter et al. 1993).
• 2.  
  By comparing the stellar attributes between all N(N−1)/2 pairs of observations, obtain photometric and astrometric offsets for each skycell of each nightly stack (with indeterminate zero point). All instrumental magnitudes are corrected to an aperture magnitude within a 6 arcsec box.
• 3.  
  Combine all nightly stacks into a stack–stack using inverse variance times inverse PSF area weighting.
• 4.  
  Obtain instrumental magnitudes for all stars in each stack–stack.
• 5.  
  Assemble a weighted combination of three estimates of photometric zero point by comparing (1) SDSS stars converted to the Pan-STARRS1 system (or its extension to MD01 and MD02 using relative photometry from Pan-STARRS), (2) z_P1 derived from Two Micron All Sky Survey (2MASS) magnitudes and stellar locus conversions, and (3) skycell-to-skycell relative zero points founded on the flattening performed by the IPP with the instrumental magnitudes of the stack–stack.
• 6.  
  Create an "object catalog" from the union over all filters for each skycell; perform forced-position photometry on each stack–stack for each object in the catalog.
• 7.  
  Compare four colors for all stellar objects in five filters with the standard colors of the stellar locus. This is derived by removing (small for MDS) Schlegel–Finkbeiner–Davis (SFD; Schlegel et al. 1998) extinction from each object and matching to the stellar locus (Covey et al. 2007), shifted to the Pan-STARRS1 system. Apply these (very small) offsets to the zero points of each filter, assuming zero median. This corrects (High et al. 2009) for small residual effects in color–color space, such as atmospheric extinction variations or small color terms between skycells.

This has the following features.

• 1.  
  All nightly stacks and stack–stacks should have very consistent photometry, limited by DoPhot's ability to match the PSF and the size of the aperture box.
• 2.  
  Use of the flattening constraint should prevent skycell to skycell jumps in zero point.
• 3.  
  The skycell by skycell tie to a standard stellar locus should make colors more accurate than absolute zero points.
• 4.  
  The absolute astrometry is founded on the 2MASS coordinate system that IPP currently uses to create the warps, but does remove small offsets (typically less than ∼50 mas) between nightly stacks that have occurred as IPP evolves. This ensures that the relative astrometry for proper motions is as accurate as we can currently make it.

Although this will eventually be superseded by the Pan-STARRS1 photometric and astrometric system, we believe that this procedure has satisfactory accuracy for our purposes.

Figure 1 shows the stellar locus of 500,000 stars from the ten Medium-Deep fields. The width of the locus in (r − i)_P1 at (g − r)_P1 = 0.6 is 0.04 mag, consistent with the estimated uncertainties in the photometry added in quadrature to 0.02 mag and constraining any failure to bring skycells and Medium-Deep field photometry into color agreement to ∼0.02 mag.

Zoom In Zoom Out Reset image size

There are some ten million objects in the ten MD fields for which we have 50–200 observations, spread over two years and the five filters. We have searched these for evidence of proper motion. The rms astrometric residuals for sufficiently bright objects (photometric precision of <0.05 mag) is 25–40 mas, per observation. For each object we assemble all the detections and fit the positions for a proper motion. Objects with r_P1 < 22 typically yield an answer with uncertainty of ∼10 mas yr⁻¹; fainter objects can be measured with correspondingly greater uncertainty. For brighter objects we can find SDSS or even USNO-B counterparts, but many are too faint.

There are some 60,000 objects that appear to have a 3σ proper motion in these two year data, and we further cut this by demanding a 6σ proper motion significance. We found that even this list was contaminated by occasional patches where the MD images apparently have astrometric errors, so we also augment the skycell alignment described above by a local astrometric bias correction. This consists of finding all objects within 003 of each candidate proper motion whose astrometric precision is greater than 1.5σ (typically ∼40 objects), assembling a median motion and quartile-derived rms, subtracting this median motion from the candidate proper motion, and adding the uncertainty in quadrature to the uncertainty of the candidate's. This has the effect of reducing the systematic error although increasing the uncertainty and decreasing the number that still meet the 6σ criterion. There are about 1000 objects that fulfill this bias-corrected, 6σ cut.

We regard this "bias correction" as a temporary expedient that permits us to begin to search for WDs, but we recognize that our proper motions are relative to stars that have non-zero motion themselves, as well as significantly diminishing the astrometric accuracy. There are a number of improvements we anticipate in the near future.

• 1.  
  Once Pan-STARRS1 moves to its own astrometric system instead of 2MASS, the need for this skycell adjustment and local bias correction should disappear, and we should see significantly tighter residuals for bright enough objects.
• 2.  
  To date we have worked only with nightly stacks, but of course it is possible to combine observations to achieve better astrometric precision for faint objects at the expense of fewer epochs.
• 3.  
  The present Pan-STARRS1 mission is slated to continue for at least two more years, doubling or tripling the temporal baseline (there were far fewer observations in 2009 than in 2010).

Although the colors of the majority of the ten million objects lie off of the stellar locus, these 1000 objects (and 60,000 3σ objects) lie gratifyingly close to the stellar locus; extremely few galaxies leak past the proper motion filter.

Comparisons with the SIMBAD database of known proper motions show excellent agreement for those objects faint enough not to be saturated for Pan-STARRS1 and bright enough to appear in SIMBAD. At r_P1 ∼ 15 only the poor seeing Pan-STARRS1 observations are unsaturated. Pan-STARRS1 achieves its best accuracy for objects in the range of 17 < r_P1 < 21, and without nightly stack combination Pan-STARRS1 loses accuracy to photon statistics fainter than r_P1 > 22.

Joint fits for parallax and proper motion over the baseline of Pan-STARRS1 observations did not turn up any highly significant parallax: the median uncertainty is 10 mas and the median parallax is 4 mas. There were three detections of parallax at 2.5σ at 13, 27, and 29 mas, but the covariance with proper motion makes these very insecure measurements. (Of course there are many stars in the MD fields closer than 30 pc, but they are all too bright and saturated.) We anticipate much better performance when we have the Pan-STARRS1 astrometric grid, perhaps uncertainties as small as 3 mas for bright objects with 17 < r_P1 < 21. Of course, the extended time baseline over the course of the Pan-STARRS1 project will also help improve the proper motion and parallax measurements.

The "reduced proper motion" (RPM), defined as H = m + 5log μ + 5 = M + 5log v_tan − 3.379, where μ is measured in arcsec yr⁻¹ and v_tan in km s⁻¹, is therefore a proxy for absolute magnitude for a given transverse velocity. With accurate photometry and astrometry, these diagrams can reveal a very clean separation of stars into main sequence, subdwarfs, and WDs. For example, the contamination rate of the WD locus by subdwarfs is only 1%–2% for the SDSS + USNO-B RPM diagram (Kilic et al. 2006). The $H_{g_{P1}}$ versus (g − r)_P1 and $H_{r_{P1}}$ versus (r − i)_P1 RPM diagrams for our 1000 proper motions are presented in Figure 2. That the gap between subdwarfs and WDs is so clearly visible is a testament to the high accuracy of the Pan-STARRS1 photometry and proper motions at this relatively stringent selection criterion.

Zoom In Zoom Out Reset image size

We use these RPM diagrams to identify 47 WD candidates. Astrometric and photometric data for these objects are given in Tables 5 and 6, respectively. About half of our sample has proper motion measurements available in the literature (Munn et al. 2004; Lépine & Shara 2005). Figure 3 shows a comparison of our proper motion measurements and the literature values. Open circles mark the objects with unreliable proper motions from the SDSS + USNO-B. Ignoring those, there is good agreement between our proper motion measurements and the literature values.

Zoom In Zoom Out Reset image size

Table 6. Pan-STARRS1 MDF White Dwarf Candidates, Photometric Data

OBJID	gP1	$\sigma _{g_{\rm P1}}$	rP1	$\sigma _{r_{\rm P1}}$	iP1	$\sigma _{i_{\rm P1}}$	zP1	$\sigma _{z_{\rm P1}}$	yP1	$\sigma _{y_{\rm P1}}$
PSO J035.1219−04.4538	22.21	0.04	21.26	0.02	20.82	0.02	20.66	0.02	20.64	0.04
PSO J036.1807−02.7150	19.60	0.04	19.01	0.05	18.89	0.05	18.80	0.05	18.70	0.05
PSO J036.1820−02.7157	19.60	0.04	18.96	0.05	18.78	0.05	18.69	0.05	18.62	0.05
PSO J053.8940−27.1420	22.77	0.06	21.89	0.04	21.48	0.03	21.33	0.03	21.28	0.12
PSO J053.9285−27.4034	19.83	0.02	19.89	0.02	19.98	0.03	20.16	0.03	20.35	0.07
PSO J128.8195+44.7108	22.32	0.04	21.52	0.02	21.17	0.02	21.05	0.02	20.98	0.05
PSO J128.9494+44.4259	18.43	0.02	18.66	0.02	18.88	0.02	19.11	0.02	19.24	0.02
PSO J130.1413+43.5130	20.27	0.02	19.99	0.03	19.89	0.02	19.88	0.02	19.91	0.04
PSO J130.3780+44.1562	20.61	0.02	20.28	0.02	20.12	0.02	20.09	0.02	20.04	0.03
PSO J130.5828+43.8370	20.72	0.02	20.45	0.02	20.33	0.03	20.31	0.02	20.33	0.03
PSO J131.0677+45.8815	21.48	0.04	21.24	0.05	21.11	0.05	21.17	0.04	21.20	0.15
PSO J131.2406+45.6086	15.96	0.05	15.98	0.03	16.02	0.03	16.14	0.02	16.27	0.02
PSO J132.0255+43.9883	21.24	0.02	20.92	0.02	20.67	0.02	20.78	0.02	20.90	0.07
PSO J132.2560+44.6595	17.95	0.02	17.62	0.02	17.50	0.02	17.49	0.02	17.52	0.02
PSO J148.6936+01.4568	19.21	0.04	19.12	0.05	19.19	0.05	19.27	0.06	19.24	0.15
PSO J148.9818+03.1285	22.79	0.06	21.91	0.00	21.75	0.00	21.44	0.00	22.30	0.30
PSO J149.3101+02.6967	20.27	0.02	19.47	0.02	19.10	0.02	19.00	0.02	18.95	0.02
PSO J149.3311+03.1446	19.34	0.02	19.10	0.02	18.98	0.02	19.03	0.02	19.01	0.02
PSO J149.7106+01.7900	15.56	0.07	15.87	0.05	16.12	0.02	16.38	0.02	16.52	0.02
PSO J149.7616+01.6465	17.95	0.03	18.13	0.02	18.34	0.02	18.52	0.02	18.68	0.02
PSO J149.8925+02.9603	18.24	0.02	18.11	0.02	18.07	0.02	18.12	0.02	18.21	0.02
PSO J150.1696+03.6030	20.90	0.02	20.56	0.02	20.40	0.02	20.36	0.02	20.42	0.04
PSO J150.8329+02.1407	21.53	0.02	20.96	0.02	20.74	0.02	20.71	0.02	20.81	0.03
PSO J151.3094+02.9046	20.55	0.00	19.76	0.00	19.36	0.00	19.42	0.06	19.18	0.05
PSO J159.5163+57.6086	21.92	0.03	21.04	0.02	20.61	0.02	20.47	0.02	20.41	0.03
PSO J160.5173+58.5631	19.05	0.02	18.96	0.02	18.97	0.02	19.07	0.02	19.16	0.02
PSO J160.6208+59.0860	22.09	0.06	21.27	0.00	21.09	0.07	21.07	0.06	20.70	0.20
PSO J161.4917+59.0766	18.33	0.00	18.19	0.00	18.16	0.00	17.98	0.00	18.14	0.03
PSO J161.8956+59.2131	17.80	0.02	17.83	0.02	17.93	0.02	18.06	0.02	18.19	0.02
PSO J162.2873+57.7484	23.79	0.13	22.86	0.07	21.86	0.04	21.59	0.07	21.43	0.12
PSO J164.1738+57.2466	18.40	0.03	18.27	0.02	18.27	0.02	18.35	0.02	18.51	0.02
PSO J164.1744+58.4375	18.3	0.15	18.24	0.03	18.15	0.04	17.9	0.15	18.09	0.04
PSO J183.8810+46.5039	17.30	0.02	17.08	0.02	17.00	0.02	16.99	0.02	17.05	0.03
PSO J185.6294+46.1184	20.96	0.02	20.39	0.02	20.11	0.02	20.07	0.02	20.09	0.03
PSO J186.4406+47.1036	19.61	0.02	19.04	0.02	18.86	0.02	18.95	0.02	19.05	0.02
PSO J213.0478+52.6587	20.62	0.02	20.54	0.02	20.57	0.02	20.62	0.02	20.70	0.03
PSO J214.3502+52.8743	19.32	0.02	19.25	0.02	19.26	0.02	19.32	0.02	19.41	0.02
PSO J215.0267+53.3759	16.97	0.02	16.84	0.02	16.85	0.02	16.90	0.02	16.94	0.02
PSO J241.3352+55.9465	17.80	0.02	17.57	0.02	17.50	0.02	17.52	0.02	17.56	0.02
PSO J242.8803+54.6275	22.14	0.03	21.24	0.04	20.82	0.02	20.71	0.03	20.60	0.03
PSO J333.8526−00.8187	24.07	0.12	23.57	0.06	22.87	0.05	22.63	0.07	22.56	0.17
PSO J334.4434−00.3065	21.38	0.02	20.73	0.02	20.45	0.02	20.39	0.02	20.48	0.03
PSO J334.5518−00.0227	20.89	0.02	20.32	0.02	20.05	0.02	19.97	0.02	19.95	0.03
PSO J334.7932+00.8453	19.82	0.02	19.63	0.02	19.59	0.02	19.64	0.02	19.68	0.02
PSO J334.8792+01.0115	19.43	0.02	18.96	0.02	18.75	0.02	18.69	0.02	18.73	0.02
PSO J352.7302+00.4814	19.71	0.02	18.93	0.02	18.60	0.02	18.46	0.02	18.43	0.02
PSO J353.2230−00.5556	20.90	0.02	20.37	0.02	20.17	0.02	20.08	0.02	20.08	0.03

Note. The columns present Object ID designations, g_P1, r_P1, i_P1, z_P1, y_P1, magnitudes, and uncertainties.

Download table as:  ASCIITypeset image

Twelve of our candidates have optical spectroscopy available in the literature: all of them are previously known WDs, including 6 DAs, 4 DQs, 1 DC, and 1 DZ. PSO J161.4917+59.0766 (LHS 291) is perhaps the most interesting, with a proper motion of 17 yr⁻¹. The identification of this object in our data demonstrates that the Pan-STARRS1 Medium-Deep field data are able to identify even the fastest moving halo objects.

We obtained spectroscopic observations of a dozen candidates, as summarized in Table 7, with the Hectospec instrument (Fabricant et al. 2005) on the MMT. The objects were selected on the basis of field access and were part of a broader program of Pan-STARRS1 follow-up spectroscopy. The Hectospec fibers are 15 in diameter. We operate the spectrograph with the 270 lines mm⁻¹ grating, providing wavelength coverage 3700–9200 Å and a spectral resolution of 5 Å with a dispersion of 1.2 Å pixel⁻¹. The spectra were reduced, including flat fielding and wavelength calibration, by the Telescope Data Center pipeline (Mink et al. 2007), at the Center for Astrophysics. Sky subtraction was performed using the spectra from fibers placed on blank sky regions. Flux calibration was performed using standard star observations from an earlier observing run. Therefore, the absolute flux calibration, as well as the relative continuum shape, of our spectra cannot be trusted. However, this level of resolution is sufficient to identify and separate WDs from metal-poor subdwarfs.

The resulting spectra are presented in Figure 4. Sky subtraction problems are noticeable in the 2011 September/October data due to non-photometric conditions for these observations. Three of the objects with Hectospec spectra were previously known WDs. PSO J164.1738+57.2466 is a known DZ WD with only Ca H and K absorption visible in its optical spectrum. PSO J215.0267+53.3759 is a known DA WD with Balmer absorption lines. PSO J352.7302+00.4814 is a known DC WD with no obvious absorption lines. Two of the objects observed with Hectospec, PSO J035.1219−04.4538 and PSO J162.2873+57.7484, are too faint for the instrument and conditions, and therefore not included in this figure. The remaining seven objects are newly confirmed WDs. PSO J036.1807−02.7150 and PSO J183.8810+46.5039 are DA WDs with hydrogen atmospheres. The lack of detection of Balmer lines in the relatively warm DC WDs PSO J213.0478+52.6587 and PSO J334.7932+00.8453 indicates that they have helium-dominated atmospheres. The most interesting of the bunch, PSO J242.8803+54.6275, is a cool DC WD with a featureless spectrum. Even though our spectrum is somewhat noisy, the lack of detection of strong MgH and Na absorption lines in the spectrum of such a red object ((g − i)_P1 = 1.32 mag) confirms the WD classification (see Figure 5 in Kilic et al. 2006). The spectral classification for these WDs, as well as the previously known WDs, is listed in Table 8. The fact 19 out of 19 RPM candidates for which optical spectroscopy was available turned out to be WD is evidence that the remainder of the RPM selected sample is highly likely to be WD as well.

Zoom In Zoom Out Reset image size

Figure 5 shows the optical color–color diagrams for our sample of WDs compared to model predictions from P. Bergeron (2010, private communication). There are two objects, PSO J162.2873+57.7484 and PSO J333.8526−00.8187, with very red (⩾0.7 mag) (r − i)_PS colors. We classify these objects as WD + M dwarf binaries. There are also two other objects, PSO J131.0677+45.8815 and J132.0255+43.9883, with (u − g)_SDSS versus (g − r)_SDSS colors similar to the known DQ WDs (bottom right panel). These objects may be DQ WDs as well, and optical spectroscopy is needed to see if they show the C₂ swan bands. PSO J148.9818+03.1285 is the only object with a relatively blue (z − y)_PS color. Hydrogen-rich cool WDs are expected to show strong flux deficits in the infrared from collision induced absorption due to molecular hydrogen (Bergeron et al. 1995; Hansen 1998). PSO J148.9818+03.1285 may be one such WD. However, the y_P1-band measurement for this object has a large error, and therefore, the absorption may not be real. Other than these outliers, the remaining 42 candidates overlap with WD model predictions in various color–color diagrams. Hence, these models can be used to derive temperatures for our targets.

Zoom In Zoom Out Reset image size

We use all five Pan-STARRS1 magnitudes to fit WD models with hydrogen atmospheres to determine the temperature, absolute magnitude, and distance of each target. Since the Pan-STARRS1 parallax measurements for our targets have limited accuracy, we assume a surface gravity of log g = 8 which determines the radius of the star for a given value of temperature. The mass distribution of WDs shows strong peaks at 0.61 ± 0.1 M_☉ for DA and 0.67 ± 0.1 M_☉ for DB WDs (Tremblay et al. 2011; Bergeron et al. 2011); these ranges imply log g = 8  ±  0.3. Hence, our assumption of log g = 8 is reasonable.

Cool WD spectral energy distributions are clearly affected by the Lyα red wing opacity in the blue (Kowalski & Saumon 2006). Since these models do not include the Lyα opacity, we use them to analyze only the WDs hotter than 4600 K. For cooler targets, we use the observed colors of cool WDs from Kilic et al. (2009, 2010, analyzed using Kowalski & Saumon 2006 models) as templates to estimate the temperatures of our targets. These templates cover the temperature range 3730–6290 K. Based on the Pan-STARRS1 and the SDSS colors, the temperature estimates from both the models and the templates agree to within 200 K for stars in the temperature range 4600–6000 K.

The photometric spectral energy distributions of six of the coolest objects in our sample are shown in Figure 6. The points with error bars show Pan-STARRS1 photometry and the solid lines show the best-fitting templates. For example, the energy distribution of PSO J035.1219−04.4538 (top left panel) is most similar to the WD SDSS J2222+1221 (Kilic et al. 2009), which has a temperature of 4170 K based on the Kowalski & Saumon (2006) models. Hence, we assign a temperature of 4170 K for PSO J035.1219−04.4538. Given this temperature and assuming log g = 8, we then estimate an absolute magnitude of M_g = 16.5 and a distance of 127 pc.

Zoom In Zoom Out Reset image size

One of the objects in Figure 6, PSO J148.9818+03.1285, may be a WD with a strong y_P1 flux deficit. If this absorption is real, then the best-fit pure hydrogen atmosphere WD model would have T_eff = 3080 K and a cooling age of 11.1 Gyr. If so, PSO J148.9818+03.1285 would be a very old thick disk or halo WD. However, near-infrared photometry is required to confirm the flux deficit in the infrared and to perform a detailed model atmosphere analysis. Many of the infrared-faint WDs have mixed H/He atmospheres (Kilic et al. 2010), where the effects of the collision induced absorption due to molecular hydrogen become significant at hotter temperatures (>4000 K) compared to the pure hydrogen atmosphere counterparts.

Table 8 lists the effective temperatures, distances, and tangential velocities for those stars whose position in the RPM diagram and color–color space indicates that they are WDs. The estimated distances and WD cooling ages for our targets range from about 50 pc to 300 pc and 0.2 Gyr to 8.6 Gyr, respectively. Two of the WD candidates, PSO J036.1807−02.7150 and PSO J036.1820−02.7157, are adjacent and have identical proper motions. Our temperature (4900 K and 5100 K) and distance (64 pc and 71 pc) estimates for these two stars agree to within the uncertainties, adding support to the hypothesis that these two stars are a physically associated binary system. The small differences between the temperatures and distances of the two stars can be explained by a small difference in mass.

Figure 7 shows the distribution of inferred ages and tangential velocities, with the assumption that the distance error is 20%. The coolest WDs in our sample have temperatures around 4200 K. Even though the individual ages for our targets cannot be trusted due to the unknown distances and masses, the average mass for our sample should be about 0.6 M_☉ and the average age for the oldest stars in our sample should be reliable. Adding 1.4 Gyr for the main-sequence lifetime of the 2 M_☉ solar-metallicity progenitor stars (Marigo et al. 2008) brings the total age to about 10 Gyr, entirely consistent with the oldest disk WDs known (e.g., Table 2 of Leggett et al. 1998), and the Galactic disk age of 8 ± 1.5 Gyr. A few of the oldest targets have large tangential velocities, and therefore they may be halo WDs. The best halo WD candidate is the 8.3 Gyr old (cooling age) PSO J053.8940−27.1420. Assuming a radial velocity of 0 ± 50 km s⁻¹ and a distance error of 20%, PSO J053.8940−27.1420 has U = 213 ± 52, V = −207 ± 55, and W = 2 ± 44 km s⁻¹ with respect to the local standard of rest (Hogg et al. 2005). Given the relatively large velocity errors, the remaining targets are consistent with thin or thick disk membership.

Zoom In Zoom Out Reset image size

Figure 8 presents finding charts for those new objects that do not reside in the SDSS fields.

Zoom In Zoom Out Reset image size