Life after eruption – VI. Recovery of the old novae EL Aql, V606 Aql, V908 Oph, V1149 Sgr, V1583 Sgr and V3964 Sgr

A classical nova eruption plays an important role in the evolution of cataclysmic variable stars (CVs; see the books by Warner, 1995; Hellier, 2001). It represents the primary mechanism for mass-loss in these systems, preventing (most of) them to end up as supernovae Ia. Irradiation of the donor star by the eruption-heated white dwarf is furthermore suspected to cause an increase in the mass-transfer rate, which would have a strong effect on the evolutionary time-scale (Prialnik & Shara, 1986; Shara et al., 1986; Kovetz, Prialnik, & Shara, 1988; Patterson et al., 2013). Still, the connection between the nova eruption and the physical parameters of CVs is not very well known, mainly due to the fact that a large fraction of the post-nova systems remain unidentified (Tappert et al., 2012).

We are carrying out a systematic observing campaign in order to recover and study classical novae that erupted at least 30 yr ago (earlier than 1986). This implies that sufficient time has passed since the nova eruption for the optical signal to become dominated by the underlying CV instead of representing mainly the ejected material. The targets presented here were selected from Downes et al. (2005) and refer to fields with hitherto ambiguous identification of the post-nova (EL Aql, V606 Aql, V908 Oph, V1149 Sgr and V1583 Sgr). They were investigated by means of $U\!BV\!R$ photometry to identify potential candidates for the post-nova, based on their colours. The selected candidates were subsequently observed spectroscopically to confirm the post-nova via their spectroscopic characteristics (mainly the presence of characteristic emission lines). In addition, we also obtained spectrograms of candidates for the post-novae V1301 Aql, V1151 Sgr and V3964 Sgr that were included in Tappert et al. (2015). Other collections of post-nova spectra can be found in Ringwald, Naylor, & Mukai (1996), Tomov et al. (2015) and parts of our own series (Tappert et al., 2012, 2014, 2015, the latter hereafter is referred to as Paper V).

The $U\!BV\!R$ photometric data for the objects with ambiguous identification were taken in two service mode runs in 2012 (EL Aql, V606 Aql, V1149 Sgr and V1583 Sgr) and in 2014 (V908 Oph) at the Antu 8 m telescope that is part of the Very Large Telescope of the European Southern Observatory (ESO), at Cerro Paranal, Chile, using the FOcal Reducer/low dispersion Spectrograph (FORS2; Appenzeller et al., 1998) system with the high-throughput broad-band filters u_High, b_High, v_High and R_Special. Series of four and three frames per filter were obtained in the 2012 and the 2014 observations, respectively. In service mode, FORS2 usually operates with a mosaic of two 2k$\times$4k MIT CCDs. For the data obtained on 2012 August 18, exceptionally, the blue-sensitive E2V CCDs were used. The field of view of FORS2 is 6.8$\times$6.8 arcmin².

After combining the two mosaic CCD frames into one with the fsmosaic routine from the FORS Instrument Mask Simulator package, the further reduction was performed using the ccdred package of iraf. The bias-subtracted and flat-fielded data were then corrected for the telescope offsets between them and averaged to a single frame per filter. A 3$\sigma$ clipping algorithm was applied to exclude bad pixels and cosmic rays.

Subsequent photometry was performed using iraf’s daophot package and the stand-alone daomatch and daomaster routines (Stetson, 1992). The aperture radius for the photometry was chosen to be slightly smaller than the full width at half-maximum (FWHM) of the point spread function (PSF). The conversion to calibrated magnitudes was obtained using observations of standard fields (Landolt, 1983, 1992; Stetson, 2000). The $U$ passband is not a part of ESO’s standard calibration plan, and in most nights, the standard field observations in this filter covered only a very limited range of airmasses that did not allow us to derive a value for the extinction from them. Thus, a standard value for the extinction was used that was taken from the La Silla observatory website¹¹1http://www.eso.org/sci/facilities/lasilla/instruments/efosc/inst/zp/.html. While this observatory is located roughly 480 km south of Cerro Paranal, comparison of the $BV\!R$ passbands and also the $U$ data for the few nights, where a sufficiently large airmass range was covered, showed identical values within the uncertainties. Similarly, for some nights, also the $BV\!R$ standard data proved insufficient. For those filters, median values for a 60 d range are available at the observatory’s Health Check Monitor²²2http://www.eso.org/observing/dfo/quality/ALL/daily_qc1.html.

The spectroscopic data were obtained using the Gemini Multi-Object Spectrograph (GMOS; Hook et al., 2004) mounted at the Gemini-South 8 m telescope situated on Cerro Pachón, Chile. Three spectra were taken per object, in most cases subsequently. Since 2014, GMOS is equipped with a row of three 2048$\times$4176 Hamamatsu CCDs. Grating B600 was employed with a 1.5 arcsec slit and a two and four pixels binning in spatial and dispersion direction, respectively. This combination yielded a typical wavelength range of 4100 – 7100 Å and a resolving power of $\sim$800. The grating was operated at a central wavelength of 5600 Å to put the two 0.915 mm wide gaps between the chips at regions of comparatively low scientific interest. The typically affected wavelength ranges are 5030 – 5080 Å and, in principle, 6100 – 6050 Å. However, all observations suffered from a bright large-scale distortion in spatial direction that severely affected the signal at a much broader wavelength range of roughly 5900 – 6200 Å and could not be properly corrected for.

The data were reduced using the gemini – gmos extension for iraf as contained in the ureka software package³³3http://ssb.stsci.edu/ureka/. After subtraction of an averaged bias frame and subsequent division by a flat-field that was normalized by fitting a high-order cubic spline, a two-dimensional wavelength correction was applied that was obtained from spectra taken with a CuAr comparison lamp. After this transformation, the individual spectra were combined employing a 2$\sigma$ clipping algorithm to correct for cosmic rays. The sky was subtracted by fitting a Chebychev function of typically third order in spatial direction. Finally, the spectra were extracted and corrected for the instrumental response function by comparison with a spectrum of the standard star LTT 3864 taken on 2015 February 25. Since all calibrations are based on this single observation of a standard, we emphasize that, while the spectral energy distribution (SED) of our objects should be sufficiently well reproduced, the zero-point is not well determined, and thus the data cannot be regarded as flux calibrated. Above steps were performed both for the combined and the individual spectra to examine them for variability.

The details of the observations of the confirmed post-novae are summarized in Table 1. The coordinates in columns two and three were determined by calculating an astrometric correction on the photometric $R$ FORS2 images or the $r^{\prime}$ GMOS acquisition frames using Starlink’s gaia⁴⁴4http://astro.dur.ac.uk/~pdraper/gaia/gaia.html tool (version 4.4.6) with the US Naval Observatory CCD Astrograph Catalogue version 4 (Zacharias et al., 2013). The associated root mean square (rms) is given in column four. The fifth column states the date of the observations referring to the start of the night, while the sixth column gives details on the instrumental setup, i.e. filter or grating and slit, and column seven contains the total exposure time per object. Finally, the brightness of the object at the time of the observations is presented in column eight, with the corresponding passband indicated by a letter. For the spectroscopic data, the magnitudes were determined via differential photometry on the acquisition frames with respect to five comparison stars in the vicinity of the target. These values were calibrated with respect to previous photometric data, where available, or with the Guide Star Catalogue, version 2.2 (Lasker et al., 2008).

Colour–colour diagrams for the fields of reported novae with ambiguous identification for the post-nova are presented in Fig. 1. We have excluded objects with photometric errors $>$0.1 mag or $<$0.005 mag to minimize contamination by very faint sources and artefacts, respectively. Candidates for subsequent spectroscopy were selected based on their position in the colour–colour diagram with respect to the other stars in the field. An additional selection criterion was the distance to the reported coordinates of the nova, taking into account their detection history. For example, the search area for novae that were identified on photographic plates was limited to the central 1 arcmin². Subsequent long-slit spectroscopic data were used to confirm the post-nova. The spectra of the confirmed novae are presented in Fig. 2, their photometric data are summarized in Table 2.

We briefly describe the methods used to analyse the spectroscopic properties that apply to all targets. The emission line strength is characterized via the equivalent width $W_{\lambda}$, which is calculated with respect to three manually selected wavelength ranges. Two of them are used to define the continuum and the noise on either side of the line, while the third one contains the line itself. We then integrate over the range of the latter divided by a linear fit to the continuum. The uncertainty of the resulting $W_{\lambda}$ is estimated via a Monte Carlo simulation by repeating the measurement a thousand times to a variation of the data with respect to the addition of a random value within the noise interval. The resulting equivalent widths are presented in Table 3.

We make an attempt at recovering the intrinsic SED by correcting for the interstellar extinction in the line of sight. We use the interface on NASA’s Infrared Science Archive (IRSA) web pages⁵⁵5http://irsa.ipac.caltech.edu/applications/DUST/ to obtain values $E(B\!-\!V)$ from Schlafly & Finkbeiner (2011). These serve as input for the dereddening routine implemented in iraf and based on the relations from Cardelli, Clayton, & Mathis (1989). A standard value for the ratio of the total to the selective extinction $R(V)=A(V)/E(B\!-\!V)=3.1$ is employed. In order to characterize the corrected continuum slope, we fit the wavelength range between 5000 and 7000 Å with a power law $F\propto\lambda^{-\alpha}$. For the intrinsic SED, the exponent $\alpha$ then depends on the mass-transfer rate and the system inclination. Larger values for $\alpha$ correspond to higher mass-transfer rates (hotter and thus bluer discs) and lower inclinations (a larger fraction of the hotter inner disc is visible). However, there are large uncertainties involved in the reddening correction, because these are large-scale extinction maps with a resolution of 2$\times$2 deg and local fluctuations may be present. Furthermore, the exact depth position of the nova with respect to the absorbing material is also unknown. These uncertainties propagate to the derived $\alpha$, which should thus be taken only as an approximate indicator of the intrinsic SED.

In Table 4 we have collected several distinguished properties of the here presented post-novae. These are the maximum brightness $m_{\mathrm{max}}$, the brightness of the post-nova $m_{\mathrm{min}}$ at the time of our observations and the resulting eruption amplitude defined as $\Delta m=m_{\mathrm{min}}-m_{\mathrm{max}}$. In the next two columns, we give the time $\Delta t$ that has passed between the nova eruption and our spectroscopic data and the time $t_{3}$ by which the nova had declined to 3 mag below maximum brightness. The parameters in the subsequent two columns are associated with the determination of the continuum slope, representing the catalogued value for the interstellar reddening $E(B\!-\!V)$ and the negative exponent $\alpha$ of the fitted power law (Section 3). The next three columns give selected properties of the H $\alpha$ emission line: the equivalent width $W_{\lambda}(\mathrm{H}\alpha)$ and the FWHM of a Gaussian fit to the line, the latter both in wavelength and in velocity units. The final column states the deviation of the position of an object from the originally recorded coordinates of the nova eruption as given by Downes et al. (2005).

Since the absolute maximum brightness of novae shows a scatter of typically only 2–3 mag (della Valle & Livio, 1995; Kasliwal et al., 2011), but the absolute brightness of CVs in quiescence is distributed over a range of $\sim$9 mag (e.g., Patterson, 1984), differences in $\Delta m$ will mostly reflect the difference in the brightness of the post-nova. The latter principally depends on the brightness of the accretion disc (or, more generally, on the emission from the accretion process), and thus on the mass-transfer rate $\dot{M}$ and on the system inclination $i$ (Warner, 1986). Because $m_{\mathrm{max}}$ can be assumed to be largely independent of these two quantities, $\Delta m$ can be taken as an indicator of the absolute brightness of the post-nova and of the involved parameters. Apart from the spread in the absolute maximum brightness, additional uncertainties are introduced by the possibility that the recorded $m_{\mathrm{max}}$ does not correspond to the real maximum brightness, but that the nova was observed at some later point, and by the different passbands used for $m_{\mathrm{max}}$ and for $m_{\mathrm{min}}$. While in general, the colour differences in the optical range for CVs will amount to less than 1 mag (e.g., Szkody, 1994), these differences will be enhanced in regions with strong interstellar reddening. A thorough analysis of $\Delta m$ will thus have to correct for these effects. However, here we only aim at a rough classification of our post-novae within the CV subtypes, and we will do so by also considering other parameters, thus reducing the weight of above uncertainties.

Further quantities of Table 4 that are indicative of $\dot{M}$ are $\alpha$ and $W_{\lambda}(\mathrm{H}\alpha)$. The significance of the former and the involved uncertainties have already been presented in Section 3. The indicative power of the latter stems from low-$\dot{M}$, optically thin discs producing stronger emission lines than high-$\dot{M}$, optically thick ones (Tylenda, 1981; Patterson, 1984). However, in CVs, typically the accretion disc is not the sole contributor to the emission line profile (Tappert & Hanuschik, 2001). Because post-novae contain an eruption heated white dwarf, irradiation of the secondary star is likely to produce an additional emission component. Furthermore, there is also the possibility that the ejected material still contributes to the line profile especially in ‘younger’, i.e. more recent post-novae. Finally, because the continuum emission depends on $i$, so does $W_{\lambda}(\mathrm{H}\alpha)$ (Warner, 1987), and thus in our interpretation, we will also use the width of the line FWHM, which is indicative of $i$ due to the larger projected velocity values $v\sin i$ sampled in the line profile.

EL Aql and V606 Aql are the novae with the largest eruption amplitudes in the present sample. Indeed, a comparison with fig. 7 in Paper V shows that there are only $\sim$10 novae in our sample of pre-1986 eruptions with similar large $\Delta m$. The idea of this parameter being an indicator for the disc brightness in these two systems fits well with the comparatively strong emission lines. On the other hand, the lines are also broad, and thus additionally, the inclination will have contributed to the high value of $\Delta m$. While, for V606 Aql, also $\alpha$ is in agreement with a comparatively low-$\dot{M}$ disc, this is not the case for EL Aql. However, we note that the steepest continuum slopes are found for the systems with the largest interstellar reddening. This appears suspicious and suggests a tendency to overcorrect for this effect.

V908 Oph and V1149 Sgr have the smallest $\Delta m$ and are also spectroscopically consistent with an optically thick disc and thus high $\dot{M}$ in that they present only weak emission lines. In V908 Oph, the bluer hydrogen lines are even embedded in absorption troughs. Still, also here the line width suggests that the inclination is such that it amplifies that effect. In conclusion, because of the strong dependence of $\Delta m$ and $W_{\lambda}$ on $i$, those four post-novae could have very similar intrinsic properties with respect to $\dot{M}$ in spite of their different spectroscopic appearance.

The picture is somewhat different for the remaining two systems. V1583 Sgr presents narrow, but comparatively strong emission lines with a variable line profile and medium $\Delta m$. We interpret this as the combination of a broad but weak disc line profile and a narrow component that likely originates in the secondary star. In this way, all parameters would agree with V1583 Sgr having a medium to high $\Delta m$ and a medium to high inclination. Assuming that the narrow component is produced by irradiation, we would also expect that either the white dwarf in this system is still very hot or that the orbital period is comparatively short. Because of the ‘age’ of the post-nova ($\Delta t$ = 87 yr), the latter appears more likely. Finally, V3964 Sgr has broad and weak lines and shows a possible obscuration effect in one spectrum. This points to a high-$\dot{M}$ system seen at high inclination. The recorded medium $\Delta m$ represents only a lower limit that, if the speculations by Lundström & Stenholm (1976, 1977) are correct, could be up to 3 mag larger (see also Paper V, ). Thus, it would also be in agreement with a high $i$. Last, not least, the continuum slope appears rather flat for the suspected high $\dot{M}$, but again might reflect a high system inclination.

In Paper V, we have argued that there is a high probability that the CVs detected by us do, in fact, represent the post-novae corresponding to the respective nova eruption and not different CVs that coincidentally have similar coordinates. Our first point was that the low space density of CVs (Pretorius, 2014, and references therein) makes it unlikely that more than one CV could be present in a field of a few square arcmin and a limiting magnitude of $\leq$24 mag, although a proper quantitative treatment would have to take into account a detailed model of the Galactic structure. As a second point of support to our argumentation, we estimated our detection probability of a specific post-nova system. This is based on the limiting magnitude $m_{5\sigma}$ of our photometry, calculated as the faintest object that is detected in $U\!BV\!R$ and has S/N = 5 in $V$, corresponding to an uncertainty in the $V$ brightness of 0.198 mag. This gives us the maximum amplitude $\Delta m_{5\sigma}=m_{5\sigma}-m_{\mathrm{max}}$ that a nova with a brightness at maximum $m_{\mathrm{max}}$ can have to still be detected in that specific photometry. Finally, the fraction of novae with $\Delta m\leq\Delta m_{5\sigma}$ translates into the detection probability. This probability is not fixed, but depends on our current knowledge of the $\Delta m$ distribution. As of 2016 May 15, we count 92 pre-1986 post-novae with known $\Delta m$, with the shape of the distribution suggesting that it is biased to bright, small-$\Delta m$ post-novae (Paper V). Our estimate furthermore ignores the possibility that a nova could have very unusual colours and thus is not recognized as such.

Table 4 shows that four of our objects were detected within 5 arcsec of the reported eruption and thus in its immediate vicinity. We estimate the possibility of a different CV being present within this radius to have a very low probability, making it unnecessary to additionally calculate the detection probability for them. V1149 Sgr is slightly farther off, but only V908 Oph shows a very large discrepancy, due to the low precision of the coordinates of the nova eruption. For the photometries of V1149 Sgr and of V908 Oph we find $m_{5\sigma}$ = 23.0 and 20.5 mag, and thus $\Delta m_{5\sigma}$ = 15.6 and 11.5 mag, respectively. The much lower brightness and $\Delta m_{5\sigma}$ limit for V908 Oph is not due a lower quality of the photometry but a result of the strong interstellar reddening causing less than 1 per cent of the objects detected in $BV\!R$ being also detected in $U$. The corresponding detection probabilities then calculate to 99 and 55 per cent. Additionally taking into account the small positional deviation to the reported coordinates, we can thus be confident that V1149 Sgr is indeed the post-nova. For V908 Oph, the case is less clear, and both parameters leave room for speculation. Here, basically the only argument supporting the identification of this system with the post-nova of Nova Oph 1954 is the absence of other candidates in the field.

1. 1.

   We have used $U\!BV\!R$ photometry and optical spectroscopy to recover the post-novae EL Aql, V606 Aql, V908 Oph, V1149 Sgr, V1583 Sgr and V3964 Sgr. We furthermore report on spectroscopy of candidates for the post-novae V1301 Aql and V1151 Sgr,

2. 2.

   Historic observations of EL Aql were collected to build a complete eruption light curve that allows it to be classified as a class P nova (Strope et al., 2010). Its eruption amplitude and colours indicate that the previously suspected possibility that EL Aql is a recurrent nova, is based on source confusion. Furthermore, its spectroscopic features suggest that the object belongs to the group of intermediate polars.

3. 3.

   Comparison of several properties of the novae indicates that the differences in the spectroscopic appearances of EL Aql, V606 Aql, V908 Oph and V1149 Sgr regarding the aspect of the mass-transfer rate are likely to be an effect of the system inclination and thus do not allow for a conclusive classification.

4. 4.

   The shape of the H $\alpha$ emission line profile in V1583 Sgr indicates that it is dominated by a non-disc emission component, likely originating in the irradiated donor star.

5. 5.

   Inspection of the emission line profiles of the individual spectra proposes EL Aql, V606 Aql, V1583 Sgr and V3964 Sgr as good candidates for time series observations in order to determine the orbital period.

6. 6.

   We have calculated the probability that the detected objects are, in fact, the post-novae corresponding to the reported eruptions. We have found that the only potentially ambiguous case is that of V908 Oph, where the low precision of the reported nova coordinates and the strong interstellar reddening strongly reducing the number of objects detected in $U$ leave some room for doubt. Still, in the absence of other candidates, the here presented object should be regarded as the post-nova.

This paper is based on observations with Gemini-South, program IDs GS-2015A-Q-54 and GS-2015A-Q-75, and ESO telescopes, proposal numbers 089.D-0505(A) and 092.D-0225(A). We are indebted to the Gemini and ESO astronomers who performed the service observations. ESO furthermore provided office space and infrastructure for the two week stay of AE in Chile, during which large parts of the work were discussed and prepared. Many thanks for the kind hospitality!

This research was supported by FONDECYT Regular grant 1120338 (CT and NV). AE acknowledges support by the Spanish Plan Nacional de Astrononomía y Astrofísica under grant AYA2015-66211-C2-1. CT, NV and IFM acknowledge support by the Centro de Astrofísica de Valparaíso.

We gratefully acknowledge ample use of the SIMBAD data base, operated at CDS, Strasbourg, France, and of NASA’s Astrophysics Data System Bibliographic Services. iraf is distributed by the National Optical Astronomy Observatories.

The Guide Star Catalogue-II is a joint project of the Space Telescope Science Institute and the Osservatorio Astronomico di Torino. Space Telescope Science Institute is operated by the Association of Universities for Research in Astronomy, for the National Aeronautics and Space Administration under contract NAS5-26555. The participation of the Osservatorio Astronomico di Torino is supported by the Italian Council for Research in Astronomy. Additional support is provided by European Southern Observatory, Space Telescope European Coordinating Facility, the International GEMINI project and the European Space Agency Astrophysics Division.

Gemini Observatory is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), the Australian Research Council (Australia), Ministério da Ciência, Tecnologia e Inovação (Brazil) and Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina).