THE NEW ECLIPSING CV MASTER OTJ192328.22+612413.5—A POSSIBLE SW SEXTANTIS STAR

Photometry of J1923 was taken in the SDSS g band using the 1.8 m Vatican Advanced Technology Telescope (VATT) located at Mount Graham International Observatory over four consecutive nights, starting 2014 April 29 (UT). The images were taken using the VATT4K CCD with a typical exposure time of 30 s. A total of 880 images were taken over these four nights. There are deep eclipses visible on the third and fourth nights of the run. On the first night, there is an eclipse that is not well determined due to the onset of twilight.

A single night of V-band photometry was acquired at the VATT on the night of 2014 June 28 (UT). A total of 518 images were taken with a typical exposure time of 15 s. There is a clear eclipse visible in these data.

Data were also acquired from the dual 1 m telescopes C2PU facility at the Calern Observatory (Observatoire de la Côte d'Azur, France) in the second half of 2014 August. A total of 245 40 s exposures with no filter, 100 60 s exposures with no filter, and 83 simultaneous 60 s exposures using V and R filters were obtained. There are eclipses visible in all of these data sets.

Spectroscopy of J1923 was taken using the Dual Imaging Spectrograph on the 3.5 m ARC telescope at Apache Point Observatory on the night of 2014 June 21 (UT). The blue channel used the B1200 grating, which has a pixel scale of 0.62 Å pix⁻¹, while the red channel used the R1200 grating, which has a pixel scale of 0.58 Å pix⁻¹. A total of five 900 s long exposures were obtained on the blue and red channels. The first three spectra were obtained at an airmass greater than 1.4, while the final two spectra were taken later in the night, at an airmass of 1.2.

Figure 2 shows the average spectrum for J1923, with variations due to the orbital motion of the system unaccounted for. The average spectrum shows distinct double-peaked emission lines, corresponding to Hα, Hβ and Hγ lines, along with a rise toward the red end of the spectrum. The flux of J1923 in the V-band section of the spectrum in Figure 2 is ≈0.45 × 10⁻¹⁶ erg cm⁻² s⁻¹ Å⁻¹. This corresponds to a V-band magnitude of 19.8, which is the magnitude of J1923 in the low state.

Zoom In Zoom Out Reset image size

The period of the system was found by applying a phase dispersion minimization (PDM; Stellingwerf 1978) to the data, which were first normalized to have a common non-eclipse magnitude. The period range was set to between 0.15 and 0.2 day. The resulting theta plot can be seen in Figure 4. The best-fit period is 0.16765 ± 0.00004 day, with the error being the width of this peak. The periods on either side of the deepest trough were tested, and were found to give poor fits.

Zoom In Zoom Out Reset image size

The time of the midpoint of each observed eclipse was found by fitting a Gaussian to the eclipse profiles (see Figure 5 for an example), and using the center of the Gaussian as the time of mid-eclipse, with the error in the time being the error in fitting the center of the Gaussian. These times and the associated eclipse numbers can be seen in Table 1, where the eclipse number was found by taking the third night's eclipse as 0, and using the newly discovered period to find the remaining eclipse numbers relative to this night. The first night's eclipse has a much higher error than other nights as the eclipse was only partially observed. For this night, a model eclipse was constructed using a Gaussian, and the midpoint adjusted manually to see where the Gaussian matched the beginning of the eclipse.

Zoom In Zoom Out Reset image size

A linear fit was applied to the data, and the resulting ephemeris was

$$\begin{eqnarray}&&{T}_{\mathrm{mid}}(\mathrm{HJD})=2456778.83690(3)+0.16764612(5)E.\end{eqnarray} \tag{ 1 }$$

The phased light curve using the linear ephemeris can be seen in Figure 6.

Zoom In Zoom Out Reset image size

The first two nights of observations show a light curve with a pre-eclipse hump starting at phase 0.7, along with quasi-periodic oscillations (QPOs) just after the eclipse. The data were masked to remove the eclipses, and then each night of data was subjected to a Lomb–Scargle periodogram (Lomb 1976; Scargle 1982). The power spectra for each night were then multiplied together to reduce the power of the noise peaks while maintaining power in the peaks common to each data set. The resulting power spectrum can be seen in the lower half of Figure 7. There are several strong peaks around a frequency of ∼3.3 hr⁻¹ (P_QPO ∼ 20 minute). The top panel of Figure 7 shows a sine wave with a period of 21 minute plotted on top of the data from 2014 April 29, when the system was in its low state and the QPOs were most prominent. The broad power spectrum confirms these oscillations are not coherent and cannot be solely due to the spin of the central WD. Kilosecond QPOs have been seen in SW Sextantis stars before, and are thought to arise from magnetic WDs that are drowned by a high accretion rate (Patterson et al. 2002).

Zoom In Zoom Out Reset image size

The light curve also shows two distinct states. The quiescent state has a non-eclipse magnitude of 19.5 in the g band from the first two nights (frames (a) and (b) of Figure 3), and the pre-eclipse hump and QPOs are distinct. The system had an out-of-eclipse magnitude that gradually decreased from night to night as the system went into outburst. On the first night of outburst (Figure 3(c))), the average magnitude was 18.2, while on the second night of outburst, the system had brightened again to m < 18. The depth of the eclipse in outburst, which was well covered by observations, was Δm = 3.2 ± 0.1 mag as seen in the g band and Δm = 2.9 ± 0.1 in the V band.

The eclipse width was found by fitting a Gaussian to the V-band data, after conversion from magnitude to flux. The resulting fit gave an eclipse width (FWHM) of 0.081 ± 0.001 in phase.

By using the calculated ephemeris, we found the phase of each of the spectra obtained: 0.81, 0.87, 0.21, 0.29, and 0.35. The rise toward the red end of the spectrum is thought to be molecular bands visible from the companion star which match that of a late M-type star, similar to the companion of Lanning 386 (Brady et al. 2008). This is further discussed in Section 4.2. The equivalent width (EW) of the Hα, Hβ and Hγ lines can be seen in Table 2, along with the Balmer decrement.

The full width at zero intensity (FWZI) of Hα, Hβ, and Hγ was measured within IRAF, as were the peak-to-peak separations of the Hα, Hβ, and Hγ lines, found by fitting a double Lorentzian to the line region, and can be seen in Table 2. The fits gave separations of 16 Å, 16 Å, and 15 Å, respectively. The asymmetry between the blue and red peaks could be due to the presence of the hotspot.

The depth of the eclipse is 3.2 ± 0.1 mag seen in the g band on nights three and four of the data and 2.9 ± 0.1 mag in the V band on night five. The majority of CVs with an eclipse depth close to or larger than 2.5 mag have an inclination angle greater than 85° (Ritter & Kolb 2003). Hence, the inclination of J1923 is expected to be >85°, suggesting that J1923 is a nearly edge on disk system.

The Hα peak-to-peak separation of 16 Å corresponds to a velocity separation of 800 km s⁻¹. This would normally limit the inclination of the system to be <30° (Horne & Marsh 1986). However, such a deep eclipse is difficult to explain with such a low inclination. Instead, it is possible that this low peak-to-peak separation is due to a truncation of the accretion disk far from the surface of the primary, which might be due to the presence of a strong magnetic field on the WD, or a low velocity component (spot or wind perpendicular to the disk or magnetic curtain) could fill-in the line center (Hoard et al. 2003).

The lower bound on the FWZI of J1923 is smaller than the FWZI of some high inclination SW Sextantis systems (Dhillon et al. 2013), which are thought to contain magnetically truncated disks (Hoard et al. 2003). This supports the idea that the accretion disk in J1923 may be truncated due to the presence of a magnetic field.

4.1.1. Modeling the Quiescent Light Curve

The light curve in quiescence was modeled using the Eclipsing Light Curve Code (ELC; Orosz & Hauschildt 2000). The basic model consisted of a WD with a temperature of 32,000 K (see Section 4.2 for more detail), an inner accretion disk radius in the range of 1–13R_WD, a temperature gradient in the disk $\left(T(r)={T}_{i}{\left(\tfrac{{r}_{i}}{r}\right)}^{\xi }\right)$ of 0.75 (which assumes a "steady-state" disk), a hotspot with angular size in the range of 2°–4° centered on orbital phase 0.86, and an inclination between 72° and 90°. We computed best-fit models based on the χ² statistic using ELC for mass ratios $\left(q,=,\tfrac{{M}_{\mathrm{sec}}}{{M}_{\mathrm{WD}}}\right)$ of 0.5, 0.4, 0.33, 0.285, and 0.25 and for inner disk temperatures (T_d) of 15,000 K, 16,000 K, and 17,000 K. Table 3 shows the best-fit inclination and χ² for each model.

Table 3.  The Best-fit Models Computed Using a χ² Grid Search with ELC for Various Mass Ratios (q) and Inner Disk Temperatures (T_d)

q	Td(K)	i°	χ2	${\chi }_{R}^{2}$
0.5	15000	81.4–83.2	2140	4.89
	16000	81.4–83.2	2118	4.85
	17000	81.8–83.6	2097	4.80
0.4	15000	81.4–82.8	2191	5.01
	16000	81.7–83.0	2165	4.95
	17000	81.9–83.6	2147	4.91
0.33	15000	81.8–82.8	2243	5.13
	16000	82.0–83.1	2213	5.06
	17000	82.3–83.4	2187	5.00
0.285	15000	81.3–83.0	2293	5.25
	16000	81.8–83.1	2259	5.17
	17000	82.1–83.5	2230	5.10
0.25	15000	81.5–82.9	2323	5.32
	16000	81.8–83.1	2290	5.24
	17000	82.2–83.6	2260	5.17

Download table as:  ASCIITypeset image

The resulting V-band fit to the data can be seen in Figure 8, which shows the best and worst fits of our modeling. The best-fit inclination range over all models was very narrow, between 813 and 836. A mass ratio of q = 0.5 provides the best overall fit.

Zoom In Zoom Out Reset image size

The ingress of the eclipse is never well fit by any of our models, with the observed ingress always steeper than in any of our models. This is a result of the simple model ELC assumes for the hotspot. ELC simply increases the temperature of the accretion disk cells which correspond to the position of the hotspot, instead of making the hotspot an independent structure. As such, ELC is not able to include eclipsing of the accretion disk by a physical hotspot for high inclination systems, which, in reality, would steepen the ingress.

We next modeled the GALEX FUV and NUV fluxes, along with the optical spectrum, to constrain the temperature of the WD and the contribution of the secondary star. WD model atmospheres were generated using Hubeny's TLUSTY, SYNSPEC, and ROTIN programs (Hubeny 1988; Hubeny & Lanz 1995) for WD temperatures between 24–34,000 K and with a logg of 8.

Schlafly & Finkbeiner (2011) estimated the reddening in the direction of J1923 to be E(B − V) = 0.0471 ± 0.0015 mag. In the following, the spectrum has been de-reddened using this value and the Cardelli extinction function (Cardelli et al. 1989), implemented in the PYTHON module ASTROPYSICS.

Table 4 shows the results of our spectral fitting. The lowest temperature WD which provided an acceptable fit to the UV and the optical was a WD with a temperature of 24,000 K and a minimum amount of disk contribution, along with a companion of spectral type M5V (the stellar spectrum was obtained from the MILES catalog (Sanchez-Blazquez et al. 2006)). This fit can be seen in Figure 9.

Zoom In Zoom Out Reset image size

Table 4.  The Reduced χ² Values for Various Fits to the Optical Spectrum and GALEX Fluxes of J1923

WD Temp (K)	Td(K)	${\chi }_{R}^{2}$
24000	15000	1.15
	16000	1.15
	17000	1.16
26000	15000	1.13
	16000	1.14
	17000	1.14
28000	15000	1.12
	16000	1.12
	17000	1.13
30000	15000	1.10
	16000	1.11
	17000	1.12
32000	15000	1.09
	16000	1.10
	17000	1.11
34000	15000	1.09
	16000	1.10
	17000	1.10

Download table as:  ASCIITypeset image

The best-fit model from Table 4 gives a WD temperature of 34,000 K and an accretion disk with an inner temperature of 15,000 K. However, there is a degeneracy between all of our fits, which can be seen in the small change in the reduced χ² between all of the models. The only limit our modeling reveals is that the lowest temperature WD which fits our data is 24,000 K, as cooler WDs cannot be fit with an accretion disk contribution.

The contribution of the secondary is similar throughout our models, with a flux between (2–2.5) × 10⁻¹⁶ erg s⁻¹ cm⁻² Å⁻¹ at 5500 Å.

The V-band magnitude of our M5V template is 20.7 ± 0.3. This gives a V–W1 of 4.1 ± 0.3, assuming that the WISE magnitudes were taken when J1923 was in quiescence. This is much higher than the V–W1 of 6.2 for an M5V star, but is closer to the V–W1 of 5.2 for an M4V star (Pecaut & Mamajek 2013). Hence, we conclude the companion spectral type is M4–5. This spectral type is also in keeping with the results from Knigge et al. (2011), which shows that the expected spectral type for a CV with a period of 4 hr is a M4–M5. Furthermore, this spectral type is similar to the M5.5 ± 0.5 companion of HS 0220+0603, another eclipsing SW Sextantis system with an orbital period between 3 and 4 hr (Rodríguez-Gil et al. 2015).

We use the V-band magnitude of the companion in the model spectrum of 20.9 ± 0.3, along with an absolute magnitude of 12.3 for an M5 star, to compute a distance to J1923 is d = 520 ± 80 parsec. If the companion is a M4 star, then using an absolute magnitude of 11.2, the distance to J1923 is d = 900 ± 100 parsec.

Figure 10 shows a previously unpublished light curve of Lanning 386 in quiescence, taken in 2009 using the Galway Ultra Fast Imager (GUFI) on the VATT. The calibrated V-band magnitude of Lanning 386 outside eclipse is 17.5 in quiescence. The similarities between the light curve for Lanning 386 and those presented here for J1923 are beyond coincidence. Both objects show QPOs in the quiescent state, and have visible humps at phase 0.75. Both objects lose both of these features while in outburst, and brighten by similar magnitudes.

Zoom In Zoom Out Reset image size

The spectra of both objects, taken when both were in quiescence, are also very similar. The Balmer decrements of Lanning 386 are D₃₄ = 1.15 and D₅₄ = 0.85 (Brady et al. 2008), which are within the measurement errors of the decrement for J1923. Both spectra also show the same M-type molecular band past 7200 Å. Lanning 386 displays two very different spectra, depending on its state. When in quiescence, its spectrum is identical to that observed here for J1923. However, in outburst, Lanning 386 displays prominent ionized helium emission lines, along with the Bowen Blend structure and C iv(λ = 5806 Å). The spectrum shown in Figure 2 was taken when J1923 was in the low state, which might explain the absence of the features seen in the excited state in Lanning 386.

The peak-to-peak separation of the emission lines in J1923 is small compared to the expected value for its high inclination, and the FWZI is smaller than the value typically found in SW Sextantis stars. We propose that this is possibly due to a truncation of the accretion disk due to the WDs magnetic field.

The single-peak structure seen in SW Sextantis stars is thought to arise due to a strong magnetic accretion curtain (Hoard et al. 2003). A possible explanation for the slightly double-peaked emission lines in J1923 could also be that there is weaker line emission from the accretion curtain in J1923, as opposed to the strong emission line core seen from the magnetic accretion curtain in SW Sextantis stars.