THE 7.1 HR X-RAY–ULTRAVIOLET–NEAR-INFRARED PERIOD OF THE γ-RAY CLASSICAL NOVA MONOCEROTIS 2012

Nova Monocerotis 2012 (Nova Mon 2012; V959 mon) is the third GeV nova discovered, with the distinction of being detected in γ-rays before being discovered at optical wavelengths. The two others detected by the Fermi-LAT (Atwood et al. 2009) are the symbiotic-like nova V407 Cyg (Abdo et al. 2010; Shore et al. 2011) and Nova Sco 2012 (Cheung et al. 2012a; Wagner et al. 2012). For V407 Cyg the γ-rays are believed to be formed either by proton–proton collisions following Fermi acceleration in shocks formed when the high velocity nova ejecta impact the dense wind from the red giant (RG) companion, or from inverse-Compton emission resulting from the shock-accelerated electrons interacting with photospheric photons from the RG (Abdo et al. 2010).

Novae are due to runaway nuclear burning on a white dwarf (WD); material is slowly accreted from a binary companion until a critical pressure is reached at the base of the accreted envelope. They occur in binaries with orbital periods of 85 minutes to several decades (e.g., Munari et al. 1990; Bianchini et al. 2003), the rare examples at wider separations containing an RG with a strong wind, which may underfill its Roche lobe.

Cheung et al. (2012b) first announced the discovery of a new γ-ray source by Fermi-LAT in the Galactic plane on 2012 June 22. At the time, this area of the sky was too close to the Sun for ground-based observations. However, on 2012 August 9, a new optical nova, Nova Mon 2012, was reported (Fujikawa 2012), which Cheung et al. (2012c) associated with the Fermi source. Nova Mon 2012 was also detected in the radio before the optical discovery; the radio flux increased in early 2012 September (Chomiuk et al. 2012). O'Brien et al. (2012) resolved the emission as a double radio source.

High spectral resolution X-ray grating observations were taken by Chandra in September and December (Ness et al. 2012b; Orio & Tofflemire 2012), 82 and 164 days after Fermi discovery. The first observation showed super-solar abundance, optically thin emission from outflowing plasma with kT ∼ 9 keV, while the second revealed that this plasma had cooled and that an additional hot WD component had emerged.

Munari (2013) confirmed Nova Mon 2012 as a neon nova on the basis of strong forbidden Ne iii–v emission lines, thus implying a high-mass WD.

A 7.1 hr period was first identified in the Swift (Gehrels et al. 2004) data (UV and X-ray; Osborne et al. 2013), and subsequently confirmed in the optical (Wagner et al. 2013). The decline of the super-soft source (SSS) was reported by Page et al. (2013). Following from and adding to these data, in this Letter we outline the X-ray/UV/optical/NIR evolution of the nova and show that the near-main-sequence secondary implied by an orbital period of 7.1 hr is not inconsistent with a recent model of γ-ray emission in novae.

Throughout this Letter, day zero has been taken as the time of the Fermi discovery: midnight on 2012 June 22 (JD 2456100.5). All magnitudes reported are on the Vega system.

2.1. Swift: X-Ray/UV Observations

2.2. MDM: I-band Photometry

2.3. AAVSO: I and V Photometry

V- and I-band photometry was obtained from the American Association of Variable Star Observers (AAVSO; see also Hambsch et al. 2013). The majority of the AAVSO observers provided consistent measurements; values from those that did not were excluded. These data, as well as our I-band photometry obtained after day 200, are plotted in Figure 1.

Zoom In Zoom Out Reset image size

3.1. Swift

3.2. MDM Photometry

4.1. Spectral Evolution

As demonstrated by the soft and hard X-ray light curves in Figure 1, there is clearly more than one emitting component after day 150. The spectrum was modeled with a combination of a plane-parallel, static, non-local thermal equilibrium atmosphere component⁶ (Rauch 2003; Rauch et al. 2010) to parameterize the soft, WD emission, and a single-temperature optically thin thermal plasma component at higher energies, both absorbed by a freely varying column (see Figure 2). The bolometric luminosity estimated from these WD atmosphere fits, for a distance of 3.6 kpc (Shore et al. 2013), was around 0.1–1 times the Eddington luminosity for a 1.2 M_☉ star. Although a full spectral analysis is beyond the scope of this Letter, no substantial spectral evolution in these two emission components, apart from the overall intensities, was seen.

Zoom In Zoom Out Reset image size

At longer wavelengths, the nova faded smoothly and at an increasing rate; over extended intervals magnitudes are proportional to log(time), i.e., f∝(t/1 day)^α. The light curves break at day ∼150 and more steeply at day ∼220. The uwm2 and I bands decline similarly after day 150, while the V-band decline is always markedly slower; thus, uvm2 − V gets redder, while V − I gets bluer. Between days 60–150, α ∼ 1.6, 1.1, 2.1 in the uvm2, V, and I bands, respectively. Then until day 220, the uvm2 and I bands showed α ∼ 2.8, with α_V ∼ 1.5. After this time, the decline indices doubled. We speculate that the slow decline in the V band may be due to persistent line emission (e.g., O and Ne; Shore et al. 2013).

4.2. Timing

A Lomb–Scargle periodogram of the Swift data after day 150 showed evidence for a 7.1 hr period in both the UVOT and XRT light curves. The top panel of Figure 3 shows the periodogram of the UVOT uvm2 data taken after the onset of super-soft emission (days 150–257), following detrending by subtraction of the best-fit third-order polynomial. (The earlier data are too sparse to help constrain a period of a few hours.)

Zoom In Zoom Out Reset image size

A folding search technique was used to measure the period (e.g., Leahy 1987; Larsson 1996), whereby χ² is calculated as a function of trial period when the heliocentrically corrected data are folded at each period (into eight phase bins). Fitting the resulting distribution of χ² near its maximum gave a period of 0.29571 ± 0.00010 day, where the error is at 90% confidence and was estimated by applying the same technique to Monte Carlo simulations of the data sampled at the same times as the observations, with appropriate Gaussian noise added.

The resulting UTC ephemeris derived from the UVOT uvm2 data is

$$\begin{equation} T_{{\rm min}} = {\rm HJD} 2456300.6684 (50) + 0.29571 (10) E, \end{equation} \tag{ 1 }$$

where E is the cycle count and the numbers in parentheses indicate the 90% errors, with the uncertainty on the epoch determined by estimating the error in the centroid of a Gaussian fit to the minimum seen in the phase-folded light curve. Consistent results were obtained from the other UVOT filters and NIR photometry.

Figure 3 (middle) plots the high-cadence Swift data obtained throughout 2013 January 16, folded according to Equation (1). The modulation is clearly seen in the soft X-rays (0.3–0.8 keV) and all three UV filters. Figure 3 (bottom) shows the folded I-band data from days 223–224. The 90% upper limit on the modulation in the hard (0.8–10 keV) X-ray data is 5%. Fitting the XRT spectrum from January 16 with a combination of an atmosphere model and optically thin thermal emission, we found that the optically thin component provides ∼10% of the counts in this soft band. Within this band there is no discernible spectral variation as a function of the 7.1 hr phase.

Figure 4 shows the data collected simultaneously by Swift-XRT, -UVOT and MDM. Short-term variations are seen across all three bands.

Zoom In Zoom Out Reset image size

The large amplitude variability of the super-soft X-ray emission (Figure 1) superficially appears random; however, it can be decomposed into a slower intensity evolution and the observational sampling of the 7.1 hr periodic modulation. Noting that the amplitude range, for much of the super-soft phase, is around a factor of 10, we constructed a count-rate model as a function of phase, of the form $C = f\cdot (\rm {A}+\rm {B \sin } \phi)$ count s⁻¹. The coefficients A and B were determined from an early section of the data (A/B = 1.71). The normalization parameter, f, was measured by fits to three-day sequences of observations, a compromise between desired time resolution and sufficient phase coverage (Figure 5). The point-to-point absolute variability in this plot averages ∼40%, much less than the ∼75% of Figure 1. Much of the fast variation in Figure 1 is due to the periodic modulation; removing this reveals the more steadily rising and declining SSS. The large-amplitude variability seen in the early SSS emission of RS Oph (Osborne et al. 2011) is not present in this system.

Zoom In Zoom Out Reset image size

Swift monitoring of the GeV ONe Nova Mon 2012 revealed a slowly evolving, hard X-ray-emitting, optically thin plasma and a much more variable super-soft X-ray component, which was first seen on day 150 and which faded to 2% of its peak rate by day 250. As shown by Schwarz et al. (2011), the late turn-on time of the SSS suggests an explosion that was unusually energetic within that sample, with an ejecta mass consistent with that reported by Shore et al. (2013). The time of the start of the decline of the SSS, around day 200, is consistent with the Hα emission-line velocity⁷ of 2400 km s⁻¹ FWHM (Shore et al. 2013), according to the relationship of Greiner et al. (2003). The turnoff time suggests a two magnitude decline time from maximum of t₂ ≲ 10 day (Schwarz et al. 2011), placing Nova Mon 2012 in the very fast class. The speed-class, together with the SSS turnoff time, are commensurate with the UV line-ratio-determined nuclear turnoff times measured by Vanlandingham et al. (2001) for ONeMg novae. The WD temperature and t₂ time are consistent with the correlation of Orio (2012), while the WD temperature requires a WD mass of ∼1.25 M_☉ (Starrfield et al. 2012), consistent with its ONe composition. In summary, as an ONe nova, Nova Mon 2012 appears to be a typical member of its class. We now consider the implications of the main result of this Letter.

We found a coherent 7.1 hr periodic modulation in the soft X-ray, UV, optical, and NIR photometry of Nova Mon 2012 during the epoch when the SSS is visible. This modulation is in phase over all wavebands, with the soft X-rays originating from the hot WD. The 7.1 hr period is almost certainly the orbital period of the system. We ascribe the modulation to the effects of an orbiting disk rim bulge viewed at moderately high inclination. Novae occur in binary systems with a wide range of orbital periods, with most below one day (Warner 2008). On-going accretion requires that the secondary star fills its Roche lobe which, for orbital periods ≲8 hr, dictates that it be close to the main sequence. In this case the secondary star mass is M₂ ≃ 0.8 M_☉, and it will have little appreciable wind. In contrast, symbiotic novae have orbital periods of the order of years, and giant secondaries with dense winds. The orbital period of Nova Mon 2012 makes it distinct from the GeV nova V407 Cyg which has an orbital period of 40–50 yr (Munari et al. 1990; Shore et al. 2011), a Mira RG secondary and thus a dense wind that acts as a target in the shock generation giving rise to the γ-ray emission.

Shock acceleration of particles to TeV energies can account for the lower shock velocity determined from X-ray temperature measurements than that measured by emission line widths in the IR, as was the case in the symbiotic nova RS Oph (Tatischeff & Hernanz 2007). Energy loss from the shock is due to very high energy particles dominating the photon luminosity by a factor ∼200, thus motivating the γ-ray emission model (Abdo et al. 2010). The inverse-Compton and proton–proton GeV γ-ray mission models of V407 Cyg were considered in more detail by Martin & Dubus (2013), who showed that a better fit to the Fermi-LAT data was achieved by a model in which a density enhancement around the WD gives rise to a dominant inverse-Compton component, as opposed to the original RG wind model.

Density enhancement in the orbital plane is a natural consequence of the presence of the WD, which gives rise to an accretion disk with a high angular momentum outflow in the plane, and to an accretion wake in the case of a significant secondary wind. Nova Mon 2012 does not have such a wind, so we must look to the accretion disk and related structures for the material against which the ejecta shock. Ejecta structure observed at high inclination is hinted at by the detection of a double radio source (O'Brien et al. 2012). The presence of such structure is also consistent with the UV and optical emission line analysis of Shore et al. (2013), who show that the horned line profiles are consistent with inclinations around 50–70°.⁸ Highly non-spherical nova ejecta have been derived by Walder et al. (2008) and Drake et al. (2009) for the symbiotic nova RS Oph, formed by an orbital plane density enhancement. The same mechanism may be at play in Nova Mon 2012.

We presume that the γ-ray emission of Nova Mon 2012 is due to shock-accelerated electrons, generated by the nova ejecta impacting the accretion disk or related structure, and boosting the nova light as in the Martin & Dubus (2013) model for V407 Cyg. In their favored model, the secondary star wind shock was not a significant source of γ-rays. Similarly, Shore et al. (2013) conclude that internal shocks in the ejecta, rather than a wind shock, are the cause of the GeV emission. The much smaller system dimensions of Nova Mon 2012 compared to V407 Cyg suggest a shorter GeV emission episode; however, a more complete test of this hypothesis, also making use of the harder X-ray emission, is beyond the scope of this Letter.

Our discovery of a 7.1 hr modulation shows that the accretion disk has reformed by day 150, and that the binary system is seen at a moderately high inclination. X-ray modulation, such as seen here, occurs due to the passage though the line of sight of a raised sector of the accretion disk rim where the accretion stream from the secondary impacts the disk, possibly augmented by reflection around phase 0.5. Such a disk was seen to be reforming between day 23–35 in the recurrent nova U Sco (Ness et al. 2012a). Treating the emitter as a point source, the lack of evident eclipses by the secondary star (R ≃ 0.8 R_☉) restricts the system inclination to i ≲ 70°. The fact that the modulation is in phase from the X-ray to the I band suggests that the modulated emission in all these bands originates interior to the disk rim. Residual nuclear burning on the WD, presumably also fed with fresh hydrogen-rich fuel by the accretion disk, is the cause of the observed super-soft X-rays, while the emission at longer wavelengths likely arises from the intensely illuminated disk itself (Suleimanov et al. 1999, 2003). The broad shape of the X-ray modulation, with the dip covering around half the orbital cycle, implies an extended absorbing or emission region. An optically thin electron-scattering cloud, potentially originating in gas stripped from the accretion disk surface, may be the cause of an extended emission region. Beardmore et al. (2012) made a similar inference for the super-soft nova HV Cet which also exhibited in-phase X-ray, UV, and optical orbital modulation.

We thank the Swift PI and MOC, and the staff of the MDM Observatory for their support of these observations. We acknowledge the observations from the AAVSO International Database contributed by observers worldwide. The Swift project at the University of Leicester is supported by the UK Space Agency. S.S. acknowledges partial support from NASA and NSF grants to ASU.