A Serendipitous Observation of the Recently Discovered Cataclysmic Variable TUVO-21acq

The recently discovered non-magnetic cataclysmic variable TUVO-21acq was serendipitously observed in a quiescent state during an XMM-Newton observing campaign for the starburst galaxy NGC 4945. Data from this campaign was combined with archival serendipitous XMM observations to examine its X-ray and UV/optical characteristics. TUVO-21acq was found to have X-ray flux F X(0.4–3.0 keV) ∼ 1.3 × 10−14 erg s−1 cm−2 and features similar to other quiescent dwarf novae, with X-ray luminosity L X(0.4–3.0 keV) ∼ 8 × 1030 erg s−1 and a white dwarf mass MWD=0.78−0.27+0.37M⊙ . Its UV/optical spectrum was variable between observations, possibly due to changes in the accretion disk or the visibility of the bright spot.


Introduction
A cataclysmic variable (CV) is a binary system composed of a white dwarf (WD) and a main sequence, or near-main sequence, late-type star that has filled its Roche lobe and donates material to the WD.There are many ways to classify CVs; perhaps one of the most common is by the process through which matter is transferred to the WD, which is affected by the strength of the WD's magnetic field.If the magnetic field is strong enough to affect the accretion flow, the system is designated as magnetic.It can shape the flow in the following ways: if the field is very strong (B  10 7 G), material from the companion flows in a collimated stream along the magnetic field lines down to the WD's surface at or near the magnetic poles (Zhilkin et al. 2012).Magnetic systems can be divided into two subgroups: polars, in which the magnetic field is strong enough to prevent an accretion disk from forming, and intermediate polars, in which a partial accretion disk is able to form.If the field is moderately strong (10 6  B  10 7 G), an accretion disk forms but is truncated at the Alfvén surface and, again, the accretion flow follows the field lines down to the surface (Aizu 1973;Zhilkin et al. 2012;Inight et al. 2021).If the field is weaker than this (B  10 6 G), the system is considered non-magnetic, and the donor material flows into an accretion disk, which streams into a boundary layer at the disk/WD interface (Patterson & Raymond 1985).The disk radiates in UV and optical wavelengths, while the boundary layer and WD atmosphere radiate in the extreme UV and X-rays (Mukai et al. 2003).It is estimated that about 80%-90% of CVs are non-magnetic (Pretorius et al. 2013).
CVs are common; the Open CV Catalog (Jackim et al. 2020) lists nearly 15,000 candidate and confirmed systems.Even so, there is much about them that is not understood.For instance, the UV spectrum of many CVs cannot be fit with the standard disk models (Puebla et al. 2007;Balman et al. 2014;Godon et al. 2017); it has been suggested that disk winds or coronae may play a role (Matthews et al. 2015), as may a truncated inner disk (Linnell et al. 2005).Godon & Sion (2011) found that they could improve their fit to the far-UV spectrum of MV Lyr by adding a component to account for the boundary layer.Further, the X-ray spectra of some CVs also have challenged models, particularly during periods of high accretion rates.It was thought that such systems would have optically thick, soft X-ray emission (Narayan & Popham 1993;Popham & Narayan 1995).However, observations have shown that many systems undergoing high accretion rates have optically thin, hard X-ray emission-which may be evidence of a truncated accretion disk, in an intriguing echo of what was seen in the UV spectra (Balman et al. 2014;Modiano et al. 2022b).Simultaneous observations in the UV and X-rays are needed to allow the modeling of the accretion disk and the boundary layer together, and thus to characterize accretion in these systems in a more holistic way (Modiano et al. 2022a).
Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
In a recent observing campaign of NGC 4945 with XMM-Newton (Weaver et al. 2024), serendipitous detections of the newly discovered CV TUVO-21acq (α(J2000.0),δ(J2000.0)= 13 h 05 m 44 7, −49°32′58 2) (Modiano et al. 2022a)) were made in both soft X-rays, with the EPIC camera, and in the UV/optical bands, with the Optical Monitor (OM).Modiano et al. (2022a) identified it as part of their Transient UV Objects project (Modiano et al. 2022b).Their analysis of stacked archival Swift images and an optical spectrum from the South African Large Telescope indicated that that TUVO-21acq is a non-magnetic CV that had undergone a dwarf nova (DN) outburst.TUVO-21acq has galactic longitude and latitude (l, b) = (305°.31524, +13°.25602).Its distance was calculated by Bailer-Jones et al. (2021), who used Gaia parallax measurements to determine the posterior probability distribution over distance to individual stars; for TUVO-21acq, they found a median geometric distance d ∼ 2300 pc.Bailer-Jones et al. (2021) also calculated the photogeometric distance to stars, which took a star's colors into account to increase the accuracy of the geometric distance measurement.However, as TUVO-21acq's colors are variable, the geometric distance is preferred.TUVO-21acqʼs height above the disk is then estimated to be h ∼ 530 pc.For comparison, Ak et al. (2008) found that DNe are typically located within 1 kpc above or below the plane of the disk.
After noticing this object in the XMM data set, a search of the HEASARC archive showed two other XMM data sets of NGC 4945, as well as some Chandra observations, with detections of TUVO-21acq.However, the XMM data were preferred, as they had simultaneous observations in OM.Examination of these data sets showed that many could not be used (Section 2) but are listed in Table 1 for completeness.The MAST Archive was also queried for Hubble Space Telescope and Galaxy Evolution Explorer data, but none were found.

EPIC-PN
All data sets were reprocessed with SAS v 205 according to standard procedures. 6The latest calibrations were applied with the epproc task, and light curves were made and examined for times of high soft proton contamination.Unfortunately, three of the most recent observations-0903540201, 0903540301, and 0903540401-were subjected to long contamination periods and could not be salvaged.The 0903540101 observation had a short flare at the end which was removed; 0112310301 did not exhibit any contamination, and 0204870101 had flares for the majority of the observation, resulting in the loss of about 2/3 of the exposure.The effective exposure times are listed in Table 1.
The resulting event files were filtered conservatively, selecting only high quality single-and double-pixel events for E > 0.2 keV (PATTERN <= 4 and FLAG==0).The source and background regions were determined, and spectra extracted.The choices of the extraction region sizes were made after examining the image by eye.For the CV, an extraction circle with a radius of 20″ was used.For the background, an annulus with radius between 20″ and 110″ was used, as it was judged that this would provide a reasonable number of counts without including any neighboring sources on the sky.After extracting the spectra, the response files were made, with the bad columns and chip gaps corrected for.The regions are shown in Figure 1.
Examination of the two short observations showed that between 0.3 and 4 keV, 0112310301 had only 76 photons, while 0204870101 had only 85, including the background.For Notes. a X-ray good time after periods of high soft proton contamination were removed.b X-ray count rate between 0.4 and 3.0 keV in units of 10 −3 c s −1 , in a circle with radius = 20″ centered on the source.c "n/a" indicates there was no data in a filter.Numbers in square brackets are upper limits.Flux densities are in the AB system and have units 10 comparison, 0903540101 had 460 in the same energy range.
We therefore confined our analysis to 0903540101.Spectra were fit using cstat (an implementation of Cash statistics (Cash 1979)) and unbinned data in CIAO/Sherpa v 4.14. 7The abundances of Wilms et al. (2000) and atomic cross sections of Verner et al. (1996) were assumed throughout.
The standard method of background removal from a source spectrum is to extract a background spectrum from an annulus around the source and subtract it from the source spectrum.However, the source was very faint, and doing this produced a spectrum with negative counts.To avoid this unphysical result, the background spectrum was fitted by following procedures described by the X-ray Astronomy Handbook (Arnaud et al. 2011).To summarize, the background has many contributors: unresolved active galactic nucleus (AGN) (treated as an absorbed power law with photon index frozen at 1.45), the hot Galactic halo (an absorbed, collisionally ionized diffuse gas, modeled with apec), the Local Hot Bubble (LHB, an unabsorbed collisionally ionized diffuse gas, also modeled with apec), and residual soft proton contamination (SPC, modeled as a shallow power law).There also are instrumental features and solar wind charge exchange (SWCX) emission lines, which were fitted with gaussians.When fitting SWCX, the line width was held at 0. The instrumental contributions were not folded through the instrumental response files.
We began by using the HEASARC's X-ray Background Tool8 to find good initial values for the models' parameters used for the AGN, hot halo, and LHB contributions.We obtained ROSAT All-Sky Survey measurements of the soft X-ray background in a circle around the source with radius = 1°a nd fitted it in Xspec.The soft X-ray emission visible in the ROSAT image does not cover a broad range, varying only between ∼(5-8) × 10 −4 c s −1 arcmin −2 .
The XMM background spectrum was then fitted between 0.2 and 15 keV with these parameters and a power law for the residual SPC.After the underlying emission was reasonably well fitted, the instrumental and SWCX components were added as gaussians, using the line tables in the XMM-Newton ESAS Cook Book9 and Carter & Sembay (2008) as guides.AtomDB10 was used to identify SWCX lines that could not be found in these resources.Interstellar absorption was accounted for with the tbabs model and frozen at the average value for the Galactic absorbing column of N H (=1.34 × 10 21 cm −2 ) in a 0°. 1 circle centered on TUVO-21acq's position using the HI4PI map (HI4PI Collaboration et al. 2016) with the HEASARC's H I calculator. 11The hydrogen column density does not vary much in this section of the sky.An examination of a HI4PI map centered on NGC 4945 that was the same size as the EPIC camera's field of view (0°.5) showed that N H only varied between (1.28-1.44)× 10 21 cm −2 .
While fitting, it was noted that the hot Galactic halo component's normalization was consistent with zero, so it was removed.The best-fit parameters for each background component and cstat/DoF of the fit are in Table 2.
With the background spectrum parameterized, we then focused on the CV's spectrum.This was fitted over the range 0.4-3.0keV with a variety of commonly used singletemperature and multi-temperature models: collisionally ionized plasmas (such as mekal and apec), thermal bremsstrahlung (bremss), Comptonization (comptt), and a cooling flow model (cemekl).Multi-temperature models are expected to be appropriate because the X-ray emission may arise from optically thin plasmas with a range of temperatures as the accreted material cools down from the temperature of the shock to that of the WD's photosphere (Done et al. 1995;Mukai et al. 2003).Finally, we also fit it with a power law.All the models are listed in Table 3.
Interstellar absorption was accounted for in same way as the background, i.e., using the tbabs model with N H held at 1.34 × 10 21 cm −2 , and the background was frozen.In the mekal, cemekl, and bremss models, the abundances were initially allowed to float, but in each case, they trended to 0. Abundances were then frozen at interstellar (Wilms et al. 2000) for the rest of the fits.All models were tested to see if the addition of gaussians and/or in-system absorption (with a phabs term (Mukai 2017)) improved the fit.The in-system absorption term improved the fits of all models, as did a gaussian at 0.53 ± 0.02 keV.It should be noted that the line widths tended to 0, so the values in Table 3 are upper limits.
The parameters, fluxes, and cstat/DoF of each are in Table 3.For comparison, Modiano et al. (2022a) found that the quiescent DN X-ray spectrum from stacked Swift data could be fit with a collisionally ionized plasma model (mekal), with F X (0.3-10 keV) = 1.11 × 10 −14 erg s −1 cm −2 .The quality of the fits in Table 3 were extremely similar for all models, but the power law fit was very slightly better than the others.It is shown in Figure 2.

OM
The OM was not turned on during 0903540301 and 0903540401.For the other observations, the data were reprocessed according to standard procedures using the SAS task omichain.
A by-eye examination of the processed images from 0903540101 and 0204870101 showed a possible detection in the B band which was not picked up by the source detection algorithm's default settings, so the data were reprocessed again using a lower source detection threshold (1σ above the background, as opposed to the default 2σ) and the source was found.In these same observations, the source was not detected in the UVW2, UVM2, and V bands, even with the lower detection threshold, and nothing was visible by eye.In 0112310301 and 0903540201, the source was not detected in any filter.For observations where the source was not detected at the 3σ level, the flux upper limit was found.The flux densities in the AB system (Oke & Gunn 1983) in each filter are listed in Table 1, and the fluxes are plotted in Figure 2. Unfortunately, the data were taken in image mode, so the light curves could not be analyzed.

X-Rays
The X-ray luminosities were estimated from the unabsorbed flux and the distance to the source.These are listed in Table 3.For comparison, Verbunt et al. (1997) found typical values of L X (0.5-2.5 keV) between ∼10 29 and 10 32 erg s −1 .
The results of the spectrum fits were compared to those found by other studies of non-magnetic CVs.The photon index in the power law fit (Γ = 2.20 - + 0.43 0.42 ) is in agreement with Maiolino et al. (2020)ʼs finding that non-magnetic CVs are well-fit with power laws having Γ ∼ 1.8.(Fitting without the gaussian led to a shallower power law, Γ = 1.69 - + 0.37 0.40 , and a cstat/DoF = 1145.7/1047.While this fit is very slightly worse than what was found without the gaussian, it should be noted that the cstat/DoF for all the fits are very similar, and the uncertainties on all the fit parameters are too large, to be able to distinguish between models.) The WD mass could be assessed from the fit to the cooling flow model (cemekl) which yielded the shock temperature in the boundary layer.In the case of a magnetic WD, T max = 3GM WD μm H /8k B R WD , where G is the gravitational constant, M WD is the mass of the WD, μ is the mean molecular weight, m H is the mass of a hydrogen atom, k B is the Boltzmann constant, and R WD is the radius of the WD (Frank et al. 2002).For a non-magnetic WD such as TUVO-21acq, half of the gravitational potential energy of the infalling material is dissipated by the accretion disk's viscosity, so the maximum temperature is halved (Mukai 2017;Yu et al. 2018).The mass can then be solved for by using the WD mass-radius relation (Nauenberg 1972).Doing this, the WD mass was found to be . For comparison, the masses of WDs in CV systems seem to range from 0.6 to 1.2 M e (Warner 1973;Ritter 1987), with a more recent study finding an average mass of 0.83 ± 0.23 M e (Zorotovic et al. 2011).

UV
There are two main contributors to the UV brightness of a CV: the WD and the accretion disk.When the system is in outburst, the accretion rate is high and the disk dominates the UV flux.When the system is in quiescence, the accretion rate slows and the disk brightness diminishes; it may still contribute to the UV flux of the system, but it is also possible for the accretion rate to be sufficiently low that the WD can be detected.
The source could not be detected in the bluest filters (UVW2: 2120 Å, UVM2: 2310 Å) which describe the accretion process closer in to the WD, and in the V (5430 Å) band, which describe the system further from the WD.However, it was detected in the UVW1 (2910 Å), U (3440 Å), and B (4500 Å) bands in ObsIDs 0204870101 and 0903540101.It was not detected in these bands in ObsID 0112310301, so only the upper limits were found.Looking at TUVO-21acq's photometry across the different observations in Table 1 and Figure 2, as well as in Modiano et al. (2022a)ʼs Figure 3, it can be seen that TUVO-21acq is variable in quiescence.Variability in CVs can be due to several things; some potential causes are discussed below.

Flickers?
Flickers are a continuous series of propagating flares, due to disturbances in the accretion flow (Bruch 1992).Balman et al. (2011) and Balman & Revnivtsev (2012) used simultaneous X-ray and UVW1 light curves of dwarf novae in quiescence to show that flickers in these bands are correlated with each other, with changes in X-ray brightness lagging behind changes in UV by several minutes; this lag has been interpreted as the time needed for matter to travel from the innermost regions of the accretion disk to the WD.While there are no UVW1 light curves in these data, it is expected that the brightnesses will trend together over the course of the full exposure.To investigate this, the X-ray count rates between 0.4 and 3 keV were considered, and may suggest a small flux increase (see Table 1).However, this may have been due to the SPC and SWCX components of the background, which are variable.To try to get a better grip on the source brightness, the background and source spectra were extracted and fit for ObsIDs 0112310301 and 0204870101 following the procedure described in Section 2.1, despite the relatively low number of counts.The parameters from the ObsID 0903540101 background were used as a starting point, freezing the AGN and LHB components and letting the other background contributors float.The source spectra were then fitted with an absorbed power law, accounting for interstellar and in-system absorption, while holding the background frozen.This led to F X (0.4-3.0 keV) = (1.01 ± 0.28) × 10 −14 erg s −1 cm −2 and F X (0.4-3.0 keV) = (1.08 ± 0.30) × 10 −14 erg s −1 cm −2 for ObsIDs 0112310301 and 0204870101, respectively.These are slightly lower than what was found for the long observation, F X (0.4-3.0 keV) ∼ 1.3 × 10 −14 erg s −1 cm −2 , though the errors bars are large enough for them to be consistent with each other.The Pearson correlation coefficient between these and the UVW1 Notes.
a All fits accounted for in-system absorption with phabs.
b kT, T 0 , T max , LE, and LW in units of keV.N H is in units of 10 21 cm −2 .The 1σ errors are shown.c Flux between 0.4 and 3 keV in units of 10 −14 erg s −1 cm −2 .d Luminosity between 0.4 and 3.0 keV in units of 10 30 erg s −1 .fluxes was found to be 0.8, indicating a strong relationship, although it should be remembered that there only are three observations and these have large uncertainties.Further, as noted earlier, TUVO-21acq cannot be detected in the UVW2 and UVM2 bands in these observations.If matter were being transferred to the WD as expected in a flicker event, TUVO-21acq would be similarly bright in all the UV bands, as they represent the innermost parts of the accretion disk.
It is possible that these observations caught TUVO-21acq in the post-outburst phase where relatively little material was left in the disk, and it was building up matter from the donor star.No information is available about its outburst phases before 2010, but ObsID 0903540101 was taken ∼200 days after an outburst (Modiano et al. 2022a).Modiano et al. (2022a) found an upper limit for TUVO-21acq's recurrence time to be 316 days, so the system would still have been rebuilding the accretion disk.
Studies of the UV spectra of CVs in quiescence show sources that exhibit a flat continuum, or a continuum with a mild slope, which is interpreted as a sign of accretion (e.g., La Dous 1991; Puebla et al. 2007;Godon et al. 2017).There can also be line emission from the H Balmer series, He I, and lightly ionized elements such as He II and Ca II (Williams 1983;Marsh & Horne 1990;Hou et al. 2020).The slope of the continuum, obtained either from spectroscopy or broadband photometry, can provide insight into the disk's temperature profile and viscosity during outburst and in quiescence, thereby placing constraints on CV accretion disk models (Puebla et al. 2007;Godon et al. 2017).
Considering only the bands in which the CV was detected, linear fits to the flux densities from ObsIDs 0204870101 and 0903540101 showed that both observations tended toward being fainter at longer wavelengths, with slopes of (−4.0 ± 2.1) × 10 −21 erg s −1 cm −2 Å −2 and (−1.1 ± 0.3) × 10 −20 erg s −1 cm −2 Å −2 , respectively.The slope from ObsID 0204870101 was consistent with a flat continuum, and while the slope from ObsID 0903540101ʼs flux densities was non-zero, it was barely beyond 3σ away from also being consistent with a flat spectrum.However, the lack of detection in the V band suggests that the flux may be dominated by line emission, rather than continuum.
It also may be that these data contain a contribution from the bright spot, where material from the donor star meets the disk.Bright spots can range in temperatures that correspond to the bluer bands (e.g., Groot et al. 2001;Kára et al. 2023).If this is the case, the lack of flux in UVW2 and UVM2 suggest it has a temperature 11,000 K.

Conclusion
The recently identified non-magnetic CV TUVO-21acq (Modiano et al. 2022a) was observed serendipitously in XMM-Newton images of the NGC 4945 field in a quiescent state.The X-ray spectrum could be fit reasonably well with many commonly used models, and had F X (0.4-3.0 keV) ∼ 1.3 × 10 −14 erg s −1 cm −2 and L X (0.4-3.0 keV) ∼ 8 × 10 30 erg s −1 .Using the results of the cooling flow model fit, the WD mass was found to be -+  M 0.78 0.27 0.37 .Two archival XMM data sets were considered in addition to the serendipitous observation to try to characterize the system.TUVO-21acq was variable in the UV/optical bands between the observations.These may be due to changes in the accretion disk between outbursts, and/or the bright spot's visibility.The uncertainties on the X-ray fluxes were too large to discern if these were also variable.In each observation where it was detected, the UV/optical broadband spectrum showed a diminishing slope toward the red or a flat continuum (albeit with large uncertainties), which is what is expected from an accreting source and is appropriate for a CV.However, it is worth noting that the lack of detection in the V band and the relative brightness in bluer bands suggests a spectrum that is dominated by line emission.

Figure 2 .
Figure 2. The spectrum from the EPIC-PN (black) and OM (orange: ObsID 0112310301, light blue: ObsID 0204870101, dark blue: ObsID 0903540101, magenta: ObsID 0903540201).The effective energy of each OM band is indicated.In the inset, the X-ray spectrum's power law fit (red) to the source and its residuals are shown.The X-ray data have been grouped to 15 counts/ bin for clarity.

Table 2
The Fit to the Background a kT, LE, and LW are in units of keV.The 1σ errors are shown.b Γ was held at 1.45.c σ was held at 0.

Table 3
Fitted Models to the Source in ObsID 0903540101