Accepted by PASP, 5 September 2012 The Stony Brook / SMARTS Atlas of (mostly) Southern Novae 1 2 Frederick M. Walter, Andrew Battisti & Sarah E. Towers Department of Physics and Astronomy, Stony Brook University, Stony Brook, NY 11794-3800 Howard E. Bond 3 Space Telescope Science Institute, Baltimore MD 21218 Guy S. Stringfellow Center for Astrophysics and Space Astronomy, University of Colorado, Boulder CO 80309 ABSTRACT We introduce the Stony Brook / SMARTS Atlas of (mostly) Southern Novae. This atlas contains both spectra and photometry obtained since 2003. The data archived in this atlas will facilitate systematic studies of the nova phenomenon and correlative studies with other comprehensive data sets. It will also enable detailed investigations of individual objects. In making the data public we hope to engender more inter- est on the part of the community in the physics of novae. The atlas is on-line at http://www.astro.sunysb.edu/fwalter/SMARTS/NovaAtlas/. Subject headings: novae, cataclysmic variables, accretion disks 1. Introduction Explodingstarshavebeennotedformillennia,andobserved(inascientificsense)forsomewhat th over a century. It wasn’t until the middle of the 20 century that a distinction could be made between the supernovae, the novae, and the eruptive phenomena seen in cataclysmic variables (the dwarfnovae). Kraft(1963)wasthefirsttosuggestthatthenovaeweretheconsequenceofexplosive hydrogen burning on the surface of a degenerate dwarf. It is now well accepted that the novae are manifestations of runaway thermonuclear reactions on the surface of a white dwarf (WD) accreting 1now at Dept of Astronomy, University of Massachusetts, Amherst MA 01003 2now at Dept of Physics, Western Michigan University, Kalamazoo MI, 49008 3current address: 9615 Labrador Ln., Cockeysville, MD 21030 – 2 – hydrogeninaclosebinarysystem(e.g.,Starrfield1971). Thenovaearehighlydynamicphenomena, withtimescalesrangingfromsecondstomillennia,occurringincomplexsystemsinvolvingtwostars and mass transfer. Theprimarydriveroftheevolutionoftheobservationalcharacteristicsofanovaisthetemporal decrease of the optical depth in an expanding atmosphere. The novae are marked by an extraor- dinary spectral evolution (Williams 1991, 1992). In the initial phases one often sees an optically thick, expanding pseudo-photosphere. In some cases one sees the growth and then disappearance of inverse P Cygni absorption lines from the cool, high velocity ejecta. As the pseudo-photosphere becomes optically thin, emission lines of the Hydrogen Balmer series strengthen, accompanied by either a spectrum dominated by permitted lines of Fe II, or of helium and nitrogen. The emis- sion line profiles and line ratios evolve as the optical depth of the ejecta decreases, and the nova transitions from the permitted to the nebular phase (Williams 1991). Beyond this template, in detail the novae exhibit a panoply of individual behaviors. Payne- Gaposchkin (1957) and McLaughlin (1960) described the evolution of novae as they were known at the time. Williams (1992) discussed the formation of the lines, and divided novae into the FeII and He-N classes, based on which emission lines dominated (aside from the ubiquitous Balmer lines of hydrogen). Novae are also categorized as recurrent and classical novae, with the former having more than one recorded outburst. Over a long enough baseline, it is likely that all novae are recurrent (e.g., Ford 1978). Bode & Evans (2008) present a recent set of reviews of the nova phenomenon. There exist well-sampled photometric records for many novae, such as those presented by Strope et al. (2010). They classify the photometric light curves, from plates amassed over the past century, into 7 distinct photometric classes. On the other hand, spectroscopic observations of novae have rarely been pursued far past maximum because most novae fade rapidly, and time on the large telescopes required for spectroscopy is precious. The most comprehensive past work was the Tololo Nova atlas (Williams et al. 1994) of 13 novae followed spectroscopically over a 4 5 year interval. The availability of the SMARTS telescope facilities (Subasavage et al. 2010) makespossibleroutinesynopticmonitoringprograms,bothphotometricandspectroscopic,oftime- variable sources. It is timely, therefore, to undertake a comprehensive, systematic, high-cadence study of the spectrophotometric evolution of the galactic novae. This atlas collects photometry and spectra of the novae we have observed with SMARTS. Most of the novae in the atlas are recent novae, discovered since 2003. Most are in the southern hemisphere. The observing cadences are irregular; we have concentrated on He-N and recurrent novae, novae in the LMC, and novae that otherwise show unusual characteristics. 4The Small and Medium Aperture Telescope System, directed by Charles Bailyn, is an ever-evolving partnership that has overseen operations of 4 small telescopes at Cerro Tololo Interamerican Observatory since 2003. – 3 – Our purpose here is to introduce this atlas. Our scientific aim is to facilitate a detailed com- parison of the various characteristics of the novae. These data can be used by and of themselves to study individual objects, for systematic studies to further define the phenomenon, and for correl- ative studies with other comprehensive data sets, such as the Swift Nova Working Group’s survey of X-ray and UV observations of recent novae (Schwarz et al. 2011). Our aim in making the atlas public is to make the data accessible to the community. We are focusing on certain novae, and on particular aspects of the nova phenomenon (§5), but simply cannot do justice to the full dataset. 2. Observations and Data Analysis 2.1. Low Dispersion Spectroscopy 5 The spectra reported here have been obtained with the venerable RC spectrograph on the SMARTS/CTIO 1.5m telescope. Observations are queue-scheduled, and are taken by dedicated service observers. The detector is a Loral 1K CCD. We use a variety of spectroscopic modes, with most of the spectra having been obtained with one of the standard modes shown in Table 1. We use slit widths of 1 and 0.8 arcsec in the low and the higher resolution 47/II modes, respectively. The slit is oriented E-W and is not rotated during the night. We routinely obtain 3 spectra of each target in order to filter for cosmic rays. We combine the 3 images and extract the spectrum by fitting a Gaussian in the spatial direction at each pixel. rd th Wavelength calibration is accomplished by fitting a 3 to 6 order polynomial to the Th-Ar or Ne calibrationlamplinepositions. Weobserveaspectrophotometricstandardstar,generallyLTT4364 (Hamuy et al. 1992, 1994) or Feige 110 (Oke 1990; Hamuy et al. 1992, 1994), on most nights to determinethecounts-to-fluxconversion. Becauseofslitlosses,possiblechangesintransparencyand seeing during the night, and parallactic losses due to the fixed slit orientation, the flux calibration is imprecise. We generally recover the correct spectral shape, except at the shortest wavelengths (<3800˚A) where apparent changes in the slope of the continuum are likely attributable to airmass- dependent parallactic slit losses. We have the capability to use simultaneous or contemporaneous photometry to recalibrate the spectra. There are some quality control issues that have not been fully dealt with, especially when we are near the sensitivity limits of the telescope. These include observations of an incorrect star, obviously incorrect flux calibrations, or spectra indistinguishable from noise. We are going through the data as time permits to address these issues. 5http://www.ctio.noao.edu/spectrographs/60spec/60spec.html – 4 – 2.2. High Dispersion Spectroscopy Wehaveasmallnumberofhighresolutionspectraofsomeofthebrighternovaenearmaximum. 6 These were obtained with the Bench-Mounted Echelle , and currently with the Chiron echelle 7 spectrograph . These data will be incorporated into the atlas at a later time. 2.3. Photometry 8 Most of the photometry was obtained using the ANDICAM dual-channel imager on the 1.3m telescope. Observations are queue-scheduled, with dedicated service observers. 2 The ANDICAM optical channel is a 2048 pixel Fairchild 447 CCD. It is read out with 2x2 binning, which yields a 0.369 arcsec/pixel plate scale. The field of view is roughly 6x6 arcmin, but until recently there has been significant unusable area on the east and south sides on the chip. The finding charts in the atlas show examples of ANDICAM images. We normally obtain single images, since the fraction of pixels marred by cosmic rays and other events is small. Exposure times range from 1 second to about 2 minutes. We use the standard Johnson-Kron-Cousins B, V, RC, and IC filters (the U filter has been unavailable since 2005, but we have extensive U band for some of the earlier novae, particularly V475 Sct and V5114 Sgr). 2 The ANDICAM IR channel is a Rockwell 1024 HgCdTe “Hawaii” Array. It is read out in 4 quadrants with 2x2 binning, which yields a 0.274 arcsec/pixel plate scale and a 2.4 arcmin field of view. The observations are dithered using an internal mirror. In most cases we use 3 dither positions, with integration times from 4 seconds (the minimum integration time) to about 45 seconds. We use the CIT/CTIO J, H, and Ks filter set. The optical and IR channels are observed simultaneously using a dichroic beam splitter. The observing cadence varies from nightly for new novae to ∼annual monitoring for the oldest novae in our list. We perform aperture photometry on the target and between 1 and 25 comparison stars in the field. The aperture radius R is either 5 or 7 pixels, depending on field crowding and sky brightness. Thebackgroundisthemedianvalueinanannulusofinnerradius2Randouterradius2R+20pixels centeredontheextractionaperture. Instrumentalmagnitudesarerecordedforeachstar. Thereare cases where the fading remnant becomes blended with nearby stars (within ∼1.5 arcsec). To date we have not accounted for such blending. Eventually we plan to employ PSF-fitting techniques in these crowded regions. 6http://www.ctio.noao.edu/noao/content/fiber-echelle-spectrograph 7http://www.ctio.noao.edu/noao/content/chiron 8http://www.astronomy.ohio-state.edu/ANDICAM/detectors.html – 5 – On most photometric nights an observation of a Landolt (1992) standard field is taken. On those nights we determine the zero-point correction and determine the magnitudes of the com- parison stars. We adopt the mean magnitudes for each comparison star. These are generally reproducible to better than 0.02 mag; variable stars are identified through their scatter around the mean, and are not used in the differential photometry. With only a single observation of a standard star field each night, we assume the nominal atmospheric extinction law and zero color correction. Using differential photometry, we can recover the apparent magnitude of a target with a typical th uncertainty of <0.03 mag at 20 magnitude. WhilewecoulddothesamewiththeIRchannelimages,wefinditsimplertousethecatalogued 2MASS magnitudes of the standard stars. We implicitly assume that the 2MASS comparison stars are non-variable, and that the color terms in the photometric solution are negligible. In addition, some higher cadence data have been obtained with the SMARTS 0.9m and 1.0m telescopes. The 0.9m detector is a 2048x2046 CCD with a 0.401 arcsec/pixel plate scale. On the 1.0m, we used the 512x512 Apogee camera that was employed prior to installation of the 4K camera. We perform the differential photometry in a manner identical to that for the ANDICAM, and merge the data sets. These data are not yet fully incorporated into the atlas. 3. Setup of the Atlas The atlas is on-line at http://www.astro.sunysb.edu/fwalter/SMARTS/NovaAtlas/. The atlas consists of a main page for each nova, giving finding charts (in both V and K bands), coordinates, and links to the spectra and photometry. The spectra are available as images, and may be downloaded in ascii (text) format. The photometry page shows plots of the light curve and colors, and permits one to download the data in ascii format. Note that the plots on the photometry page only show data with formal uncertainties <0.5 mag, while all measured magnitudesanduncertaintiesareincludedintheasciilistings. Thereisalinktoapageofreferences for other observations of the novae. 4. The Novae As of 1 July 2012 the atlas includes data on 64 novae. Of these, 29 are still bright enough (V<18) to reach spectroscopically with the 1.5m/RC spectrograph. Most are still detectable pho- tometrically with the 1.3m/ANDICAM imager. Only 5 are no longer on our photometric target list because they are too faint or too confused with brighter companions. The spatial distribution of these novae is shown in Figure 1 in both celestial and galactic coordinates. – 6 – Lists of our targets and particulars on the number and observing date distribution of the observations are in Tables 2 (novae from before 2012); 3 (novae discovered in 2012), and 4 (novae in the LMC). The reference time is ideally the time of peak brightness, but this is often not well known. In general, T0 is the time is discovery. In the case of T Pyx, which rises very slowly, To is the time of peak brightness as estimated from our photometry. For novae that were discovered well past peak, including N Sgr 2012b and XMMU J115113.3-623730, T0 is a guess. All the dates in the Tables are referenced to T0. The tabulated V is the last observed V magnitude; in most cases this is the brightness in June 2012. The Tables are current as of 1 July 2012. 4.1. Observing Statistics As of 1 July 2012 the full atlas contains 64 novae. We have between 1 and 368 spectra for the novae, with a median of 28 spectra per nova. The number of photometric points varies between 1 and 265, with a median of 35, for 53 novae. Since some of the observations were taken through thick clouds, not all observations have the best possible S/N. The photometric and spectral coverage is generally non-uniform in time. In addition to annual gaps due to the Sun, there is spotty coverage during the austral winter when the weather becomes worse. We do not have unlimited observing time, so we concentrate on those novae that tickle our astronomical fancy - the He-N novae, and those showing unusual characteristics. We do not attempt spectroscopy of targets fainter than V ∼18, because the 1.5m telescope has limited grasp. We generally do not make great efforts to obtain photometry from day 0, because amateur astronomers do such a good job. In many cases data available from the AAVSO can fill in the first few weeks, while the nova is bright (we do have bright limits near V = 8 and K = 6). Our forte is the ability to a) follow the evolution to quiescence, and b) to do so in the 7 photometric bands from B through Ks. In one case we were on the nova 1.1 days after discovery, but the median delay is 15 days. We try to start the spectroscopic monitoring sooner, because this is a unique capability of SMARTS. The first spectrum is obtained with a median delay of 8.0 days from discovery, but we have observed 1 nova within 0.6 days of discovery, 9 within 2 days, and 14 within 3 days. We have multi-epoch photometry of 52 novae over timespans of up to 3173 days (8.7 years), and multi-epoch spectroscopy of 63 novae over timespans of up to 3156 days (8.6 years). These durations will increase with time so long as SMARTS continues operating, and the targets are sufficiently bright. The median observation durations of 1317 and 360 days, respectively, for the photometry and spectroscopy, are limited mostly by target brightness. The median time between observations is skewed by the growing number of old, faint targets that are now observed with a cadence of 1-2 observations per year, so that the median time between spectra is 8.4 days, and is 27 days for photometric observations. – 7 – 4.2. The Example of V574 Pup We illustrate possible uses of the atlas with the example of V574 Pup, an Fe II nova for which we have good coverage. Aside from near-IR observations (Naik et al. 2010) and analysis of the super-soft X-ray source (Schwarz et al. 2011), there has been little discussion of this bright nova. Themainatlaspage(Figure2)presentsfindingchartsinV andK, alongwiththecoordinates, time of discovery, and links to the spectral and photometric data and references. The photometry consists of observations taken on 100 days with the 1.3m ANDICAM imager, starting on day 32 and running through day 2723 (5 May 2012). Most of these sets include all 7 ANDICAM bands, BVRIJHKs. This light curve is shown in Figure 3. It is possible to fill in the first 30 days with data from other sources, such as Silviero et al. (2005), or by using data from the AAVSO (www.aavso.org). We supplemented these with data taken on 20 days using a temporary small CCD on the SMARTS 1.0m telescope. These were opportunistic observations enabled by the unavailability of the wide-field 4k camera. We use the ∼2 hour long sequences to search for short periodicities. (Similar data exist for very few of the novae in the atlas.) Three long sequences in the B band, on days 87, 195, and 196, showed sinusoidal-like modulations. Removing a linear trend from the data on day 87 and normalizing to the mean magnitudes, we find a likely period of 0.0472 days (68 minutes; see Figure 4) from a shortest-string analysis (Dworetsky 1983). However, we cannot exclude some aliases. This is shorter than the minimum orbital period for CVs, and may be half an orbital period (ellipsoidal variability is a possible explanation). The amplitude of the best-fit sinusoid decreased from 0.02 to 0.007 mag from day 87 to days 195-196. Weobtained107spectrabeforethetargetbecametoofaintforthe1.5mtelescope. Weillustrate two types of investigations that can be supported by high cadence spectral observations. 1. Figure 5 shows the evolution of the P Cygni line profiles as the wind evolves against the backdrop of the optically-thick pseudo-photosphere. With daily spectra, it is clear that the absorption velocities are not constant, but rather are accelerating. Through day 14 the velocities can be described as a quadratic function of time. Hence the acceleration is linear in time. It is hard to see how this can result from decreasing optical depth effects in an envelope with a monotonic velocity law increasing outwards. Shore et al. (2011) show how similar structures seen in T Pyx can be explained as an outward-moving recombination front in an envelope with a linear velocity law. 2. Figure 6 shows the time-evolution of a series of lines of differing temperatures as the nova evolves through the nebular and coronal phases. The [FeX] λ6375˚A line requires high excita- tion, and its presence correlates well with the super-soft (SSS) X-ray emitting phase (Schwarz et al. 2011). V574 Pup was in its SSS phase from before day 180 through day 1118; it ended beforeday1312. Wecanusedatasuchasthesetoexplorehowwellopticallinesarediagnostic – 8 – of the SSS phase. 4.3. Notes on the Novae Notesherearenotmeanttobecompleteordefinitiveinanysense. Theyaremeanttohighlight past or ongoing work on select novae, or to note some particularly interesting cases. We have made no attempt to provide complete references here; they are in the on-line atlas. For the convenience of the reader, we have collected in Table 5 various basic measurements. These are: • Spectroscopic class. This is a phenomenological classification based on the appearance of the spectrum in the first few spectra after the emission lines appear. Physically, this is likely an indicator of the optical depth of the envelope. Of the 63 classifiable targets, most (47/63, or 75%) are Fe II type; 15 (24%) are or may be He-N, and one is a possible symbiotic nova. In one case we cannot tell because our first spectrum was obtained nearly 2 years after peak. We append a “w” in those cases where there is a clear P Cygni absorption in the Balmer lines (and sometimes in Fe II) indicative of an optically-thick wind. Half (32 of the 64 novae) show such P Cyg absorption. We caution that the absence of P Cyg absorption may be caused by the cadence of the observations. • Photometricclass. WeexaminedtheV bandlightcurvesforthefirst500daysandcategorized them by eye into one or more of the 7 classes defined by Strope et al. (2010). In many cases we have very little data during the first 3 months, and do not attempt to categorize these. In some cases we had a hard time shoe-horning the lightcurve into one class, and have given multiple classes. For example, N LMC 2005 maintained a fairly flat light curve for about 50 days (class F), then exhibited a cusp (class C). It also formed dust (class D), though the dip is not particularly pronounced. The presence of dust is indicated by the increase in the H and K fluxes as the optical fades. InmanycasesasignificantbrighteninginK,suggestiveofdustformation,isnotaccompanied by an optical dip, suggesting an asphericity in the dust. In some cases there is significant color evolution between the optical and near-IR. We will quantify this later. • The FWHM of the Hα emission line. We measure the first grating 47 spectrum (3.1˚A resolu- tion) that does not show P Cyg wind absorption, and report the day on which that spectrum was obtained. Uncertainties are of order 2%. Note that the FWHM can change significantly with time in the Fe II novae. For the He-N novae we measure the FWHM of the broad base, ignoring the narrow central emission component. In some cases there is a faint but broader component visible early on. The measurement of the FWZI of this component would be more representative of the maximum expansion velocity. We do not tabulate this because of incompleteness, and because of the difficulty defining the continuum level in some cases. – 9 – We have not estimated the times for the light to decay by 2 and 3 magnitudes at V (t2 and t3, respectively) in any systematic manner, because it is only in rare cases that we have sufficiently dense photometric sampling early enough to make a good estimate. We discuss these in the notes on individual novae. We note that the estimates of t2 and t3 can be highly uncertain, especially for fast novae. The reported discovery times are often past the peak. The discovery magnitudes are often visual estimates, or unfiltered CCD magnitudes, necessitating a color correction to V. A full analysis of the light curves, incorporating other published literature, AAVSO data (which are much denser near peak), and data from other sources, is beyond the scope of this paper. 4.3.1. N Aql 2005 = V1663 Aql This is a standard Fe II nova. On day 50 there was prominent λ4640˚A Bowen blend emission. The auroral [O III] lines were strong by day 85. Our last spectrum, on day 414, is dominated by Hα, [O III] 4959/5007, [N II] 5755, [Fe VII] 6087, [O I] 6300, and [Ar III] 7136. 4.3.2. N Car 2008 = V679 Car This Fe II nova never seemed to develop a coronal phase. We have limited photometric cover- age. 4.3.3. N Car 2012 = V834 Car This recent Fe II nova exhibited a strong wind through day 36. Evolution of the light curve has been uneventful. There was some jitter of ±0.5 mag from a smooth trend from days 12-40. We estimate t2 and t3 to be 20 and 38 days, respectively, with uncertainties of order ±3 days for t2 and ±1 day for t3. 4.3.4. N Cen 2005 = V1047 Cen We have no photometry, and only two spectra, of this FeII nova. 4.3.5. N Cen 2007 = V1065 Cen This dusty Fe II nova was analyzed by Helton et al. (2010), using SMARTS spectra through day 719. The atlas includes additional photometry, from days 944 though 1850. – 10 – 4.3.6. N Cen 2009 = V1213 Cen This Fe II nova became a bright super-soft X-ray source. The coronal phase extended from about days 300 to 1000, roughly coinciding with the SSS phase (Schwarz et al. 2011), with strong lines of [Fe X], [Fe XI], and [Fe XIV]. In quiescence the remnant is blended with two other objects of comparable brightness. 4.3.7. PNV J13410800-5815470 = N Cen 2012 This recent Fe II nova exhibited wind absorption through day 25. t2 is about 16±1 days; t3 occurs about day 34. The 2 mag brightening in K starting about day 35, with a contemporaneous drop in the B and V band brightness, suggests dust formation. The strong emission in the Ca II near-IR triplet on day 11 had disappeared by day 74. 4.3.8. PNV J14250600-5845360 = N Cen 2012b The K band brightness increased by 2 magnitudes between days 18 and 32, suggesting dust formation, but no drop is seen at optical magnitudes. The smooth V light curve yields t2 and t3 of 12.3 and 19.8 days, with uncertainties <1 day. The spectral development is similar to N Cen 2012. The strong emission in the Ca II near-IR triplet on day 16 had disappeared by day 61. 4.3.9. N Cir 2003 = DE Cir This fast nova was discovered by Liller (2003) in the glare of the setting Sun. Spectra obtained ondays11and12, athighairmass, showthiswasaHe-Nnova. Wedidnotobtainanyphotometry until after it reappeared from behind the Sun. Since then it has been in quiescence at V∼17, with a variance of ±0.4 mag. The strongest line in the quiescent spectrum is He II λ4686˚A. 4.3.10. N Cru 2003 = DZ Cru Thisisanothernovathatwasdiscoveredinthewestinthedusktwilight. Despitethediscussion about the “peculiar” early spectrum (Bond et al. 2003), our spectra show this was an Fe II nova discovered before maximum, as concluded by Rushton et al. (2008).