Search for Surviving Companions of Progenitors of Young LMC SN Ia Remnants

We have obtained new HST images of SNR 0519–69.0, DEM L71, and SNR 0548–70.4, using the UVIS channel of Wide Field Camera 3 (WFC3) in Program 13282 (PI: Chu). For DEM L71 and SNR 0548–70.4, the images were obtained with the F475W (B band), F555W (V band), F814W (I band), and F656N (Hα line) filters. For SNR 0519–69.0, new images were obtained with only the F475W and F814W filters because HST F550M and F658N images taken with the Advanced Camera for Surveys (ACS) are available in the Hubble Legacy Archive. The HST images used in this paper are listed in Table 1. The continuum-band images can be used to search for surviving companions and study the underlying stellar population, while the Hα images can be used to analyze physical structures of the remnants and to determine their explosion centers.

The UVIS channel of WFC3 has a 162'' × 162'' field of view (FOV). The 004 pixel size corresponds to 0.01 pc in the LMC, at a distance of ∼50 kpc. The observations of DEM L71 and SNR 0548–70.4 were dithered with the WFC3-UVIS-GAP-LINE pattern for 3 points and point spacings of 2414. Each observation had a minimum total exposure time of 1050 s. In order to mitigate the charge transfer efficiency (CTE) issues in WFC3, FLASH = 5, 11, and 4 options were used for the F475W, F656N, and F814W observations, respectively. The observations of SNR 0519–69.0 in the F475W and F814W passbands were dithered using POSTARG. The F475W observation had a total exposure time of 1070 s and FLASH = 5; the F814W observation had a total exposure time of 1174 s and FLASH = 4.

To examine the stars projected within or near the three Balmer-dominated SNRs, we have used the stellar photometry package DOLPHOT, adapted from HSTPHOT with HST-specific modules (Dolphin 2000), to perform point-spread function (PSF) photometry on the HST images. Following Williams et al. (2014), we adopt specific criteria on photometric parameters, such as signal-to-noise ratio (S/N) > 5, sharp² < 0.1, and crowd < 1.0, to filter out nonstellar sources (Dolphin 2000). All stars have been measured in the Vega magnitude system for the B, V, and I passbands. The HST photometry allows us to compare the locations of stars in the CMDs with the expected evolutionary tracks of the post-impact surviving companions.

SNR 0519–69.0, DEM L71, and SNR 0548–70.4 were observed with the Advanced CCD Imaging Spectrometer (ACIS) of the Chandra X-ray Observatory for Program 01500024 (PI: Holt; 39.2 ks), Program 01500900 (PI: Hughes; 45.6 ks), and Program 02500872 (PI: Borkowski; 59.3 ks), respectively. The event files of these observations are available in the Chandra Data Archive, and images in various energy bands are conveniently available from the Chandra SNR Catalog. ⁹ While these SNRs are detected from 0.3 keV to 4–7 keV, the 0.3–2.1 keV band contains the bulk of the X-ray emission, and the 0.3–2.1 keV band images are used to locate the boundaries of SNRs in order to assess their explosion centers.

MUSE observations of SNR 0519–69.0 and DEM L71 were obtained with the Very Large Telescope (VLT) UT4 of the European Southern Observatory (ESO) on 2015 November 16 and 2016 January 17 (see Table 2) for Program 096.D-0352(A) (PI: Leibundgut). VLT MUSE is an integral field unit (IFU), providing optical spectrum for every position in the FOV. The wide-field mode used for the observations has a 60'' × 60'' FOV. This FOV is large enough to cover the entire SNR 0519–69.0 and adequate surroundings for background, but only about 80% of DEM L71 and a small region for background. The spectral coverage is 4750–9350 Å, which includes nebular emission lines such as Hα, Hβ, [O iii] λλ4595, 5007, [N ii] λλ6548, 6583, and [S ii] λλ6716, 6731, as well as stellar absorption lines such as Hα, Hβ, He i λ5875, He ii λλ5411, 4686, and Na i D lines λλ5890, 5896. The spatial and spectral samplings are 02 spaxel⁻¹ and 1.25 Å pixel⁻¹, respectively. The VLT MUSE observations of SNR 0519–69.0 and DEM L71 each had an exposure time of 900 s.

We use the VLT MUSE data reduction pipeline (Weilbacher et al. 2014) to process bias subtraction, flat-fielding, and wavelength and geometrical calibrations. The fields are crowded, and large parts of the FOV are covered by either stars or nebular emission. To subtract the sky background, we use one pointing to blank sky and assume that the result would be viable for all pointings, and we make use of algorithms from the Zurich Atmospheric Package (Soto et al. 2016); however, artificial flux fluctuations are found in the wavelength range 7700–9000 Å, and they are likely caused by the imperfect subtraction of sky background. To perform the flux calibration, we use existing photometry of stars in the field and synthetic photometry from the VLT MUSE data cube to make sure that differences of magnitudes of stars are within 0.05 mag. The detailed procedures of data reduction are described in Krühler et al. (2017). The VLT MUSE observations can be used to carry out spectroscopic analyses to search for stars that have large peculiar radial velocities.

2.3.1. Extracting Spectra from VLT MUSE Data

2.3.2. Stellar Parameter Inference

Spectral template fitting with the PHOENIX grid was performed using the method described in Do et al. (2013) and Støstad et al. (2015) with the software package StarKit (Kerzendorf & Do 2015). StarKit is a modular spectral fitting framework using Bayesian inference to determine the best-fit parameters and their uncertainties. StarKit simultaneously fits the physical parameters of stars, the spectral continuum, the radial velocity, and the rotational velocity. The set of physical parameters available for fitting is determined by the parameters sampled by the spectral grid. For the PHOENIX grid, the parameters are stellar effective temperature T_eff, surface gravity $\mathrm{log}g$, metallicity [M/H], and α-element-to-iron abundance ratio [α/Fe]. We use a linear interpolator to interpolate between the synthetic spectral grid points. We then convolve the spectral resolution of the grid to R = 3000 in order to match the instrumental resolution.

Statistically, the fit is done by computing the posterior distribution in Bayes's theorem:

$$\begin{eqnarray}&&P(\theta | D)=\displaystyle \frac{P(D| \theta )P(\theta )}{P(D)},\end{eqnarray} \tag{ 1 }$$

where D is the observed spectrum and the model parameters $\theta =({T}_{\mathrm{eff}},\,\mathrm{log}g,\,[{\rm{M}}/{\rm{H}}],[\alpha /\mathrm{Fe}],{V}_{r},{V}_{\mathrm{rot}})$, where V_r is the radial velocity and V_rot is the rotational velocity. The priors on the model parameters are P(θ), and P(D) is the evidence, which acts as the normalization. The combined likelihood for an observed spectrum is

$$\begin{eqnarray}&&P(D| \theta )=\displaystyle \prod _{\lambda ={\lambda }_{0}}^{{\lambda }_{n}}\displaystyle \frac{1}{{\epsilon }_{\lambda ,\mathrm{obs}}\sqrt{2\pi }}\exp (-{\left({F}_{\lambda ,\mathrm{obs}}-{F}_{\lambda }(\theta )\right)}^{2}/2{\epsilon }_{\lambda ,\mathrm{obs}}^{2}),\end{eqnarray} \tag{ 2 }$$

where F_λ,obs is the observed spectrum, F_λ (θ) is the model spectrum evaluated with a given set of model parameters, and _λ,obs is the 1σ uncertainty for each observed flux point. This likelihood assumes that the uncertainty for each flux point is approximately Gaussian. For computational efficiency, we use the log-likelihood in place of the likelihood:

$$\begin{eqnarray}&&\mathrm{ln}P(D| \theta )\propto -\displaystyle \frac{1}{2}\displaystyle \sum _{\lambda ={\lambda }_{0}}^{{\lambda }_{n}}({\left({F}_{\lambda ,\mathrm{obs}}-{F}_{\lambda }(\theta )\right)}^{2}/{\epsilon }_{\lambda ,\mathrm{obs}}^{2}).\end{eqnarray} \tag{ 3 }$$

We use flat priors in all the fit parameters and sample the posterior using Nestle, a nested sampling implementation (Feroz et al. 2009). We use the peak posterior value to be the best-fit values and the marginalized 68% central confidence intervals for each fit parameter to be its uncertainty. Based on the tests against empirical references described in Feldmeier-Krause et al. (2017), we include a systematic uncertainty term added in quadrature to the statistical uncertainties of each fit of ${\sigma }_{{T}_{\mathrm{eff}}}=200$ K, σ_[M/H] = 0.2, σ_[α/Fe] = 0.2, and ${\sigma }_{\mathrm{log}g}=1.0$.

After the explosion of an SN Ia, the SN ejecta expands and interacts with the partially neutral ambient medium to produce the Balmer-dominated shell (Chevalier et al. 1980). If its progenitor had a nondegenerate stellar companion, as in the SD scenario, the stellar companion would survive the explosion and run away at a velocity comparable to its final orbital velocity. If the binary progenitor had a translational motion through the interstellar medium (ISM; e.g., N103B, Li et al. 2017), the Balmer-dominated shell would still be centered at the site of SN explosion, as the expansion velocity of the SN ejecta is much higher than the progenitor's translational velocity. The surviving companion will run away from the site of SN explosion at a velocity that is the sum of its orbital velocity and the progenitor's translational velocity.

With the above understandings, we first estimate the site of SN explosion using the Balmer-dominated shell. If an SNR's Balmer shell is regular and exhibits reflection symmetry about two orthogonal lines of symmetry, the shell can be fitted by an ellipse, and its geometric center can be easily identified and measured with small uncertainties. We then adopt the geometric center to be the SN explosion site (e.g., Litke et al. 2017).

If an SNR's Balmer shell is irregular, we still visually fit its general shape with an ellipse. Usually, over 75% of the shell periphery can be fitted well by an ellipse, and regions showing large deviations are faint and may be caused by a particularly low ambient density; thus, it is reasonable to adopt the center of the ellipse as the site of SN explosion. This is the case for SNR 0519–69.0 and SNR 0548–70.4.

If the ellipse fitted to an SNR's Balmer shell is visibly unsatisfactory, for example, no ellipse can describe more than 70% of DEM L71's Balmer shell periphery, then X-ray images of the reverse-shocked SN ejecta are used to make independent estimates of the site of SN explosion (e.g., Schaefer & Pagnotta 2012; EPS2012; EPS2012 PS2015;). The differences among these different estimates contribute to the uncertainty of the SN explosion site.

A surviving companion runs away from the site of SN explosion with a velocity that is the sum of its orbital velocity and the progenitor's translational velocity. Assuming ∼2–3 M_⊙ for an SN Ia progenitor's total mass and a period of ≤1.0 day, the orbital velocity of the companion can reach as high as 400 km s⁻¹ (Schaefer & Pagnotta 2012). The translational velocity of the SN progenitor is unknown, but it is most likely within the range of peculiar velocities of stars and much lower than velocities of runaway stars. In the post-impact models of surviving companions (Pan et al. 2014), the largest runaway velocity is 730 km s⁻¹ for a helium star companion. To be on the safe side, we adopt a translational velocity of <100 km s⁻¹ and a "runaway velocity" of <730 km s⁻¹ for the surviving companion. Note that in a thermonuclear SN model with a WD and a subdwarf helium star donor going through double detonation, the companion may be ejected at velocities up to 900 km s⁻¹ (Geier et al. 2013; Bauer et al. 2019); however, these surviving subdwarfs are too faint to be detected in the LMC.

The ages of the young Balmer-dominated SNRs have been determined from SN light echo observations (Rest et al. 2005) or sizes and shock velocities (Smith et al. 1991; Ghavamian et al. 2003). The "runaway distance" of a surviving companion from the site of SN explosion is equal to the runaway velocity times the SNR's age t. We thus adopt a search radius of 7.5 × 10⁻⁴ t_yr pc for the surviving companions. The younger the SNR is, the easier it is to search for a surviving companion within, as the surviving companion has not moved too far.

The impact of SN ejecta on a nondegenerate stellar companion has been studied with numerous hydrodynamic simulations (Marietta et al. 2000; Pakmor et al. 2008; Pan et al. 2010, 2012a; Liu et al. 2012, 2013; Bauer et al. 2019). Among these, Pan et al. (2014) have calculated the post-impact evolution of luminosity and effective temperature of the surviving companion and plotted them in the theoretical Hertzsprung–Russell (H-R) diagram, i.e., luminosity versus stellar effective temperature. For direct comparisons with observations, a blackbody model with the surviving companion's temperature and luminosity is used to calculate its magnitudes in different passbands. The post-impact evolution of a surviving companion can then be plotted in an observational H-R diagram, i.e., CMDs.

Schaefer & Pagnotta (2012) have considered various published SD models for SN Ia progenitors, such as recurrent novae, symbiotic stars, supersoft sources, helium donor companions, etc., and suggested that a surviving companion would have V < 22.7 mag at the distance of the LMC. Thus, in this study we consider only stars with V < 23.0 in the search for surviving companions.

We plot both stars in the search radius and stars within a large radius encompassing the entire SNR and some surrounding regions in V versus B–V and I versus V–I CMDs. Stars within the large radius allow us to establish a background stellar population for comparison. We then compare the stars in the search radius with the post-impact evolutionary tracks of surviving companions from Pan et al. (2012b, 2013, 2014). A star falling on an evolutionary track with consistent age would be a candidate for a surviving companion. Only a few stars appear double and too blended for reliable photometric measurements, and most of these are outside the search radii; therefore, we choose to ignore them.

We use VLT MUSE observations of SNR 0519–69.0 and DEM L71 to carry out spectral analyses of stars within the search radius from the site of SN explosion, and we use stellar atmosphere model fits to determine physical parameters and radial velocities of stars. We find that stellar atmosphere models show more reliable fits for stars with V < 21.6 mag. This turns out to be a safe limiting magnitude for our spectroscopic search for surviving MS companions, as the post-impact evolution of surviving MS companions modeled by Pan et al. (2014) shows V well brighter than ∼21.0 mag within the Balmer-dominated phase of Type Ia SNRs, less than ∼10,000 yr after the SN blast. We thus spectroscopically examine the photometric candidates of surviving companions that have V < 21.6 and are located within the search radii, using large radial velocities as diagnostics of surviving companions.

The distributions of radial velocities of these stars are plotted and examined. The standard deviation of the distribution, σ, is computed. Assuming a Gaussian distribution of the radial velocities, we expect only 0.3% of the population to have velocities deviating by more than 3σ from the mean ($\overline{{V}_{r}}$). For a population of <200 stars, fewer than one star is expected to be >3σ from the mean. Therefore, stars with radial velocities more than 3σ from the mean are carefully examined to determine whether they are viable candidates for surviving companions of the SN progenitors.

EPS2012 argued that the SNR shell of 0519–69.0 does not suggest highly asymmetric explosion and that its geometric center is close to the site of SN explosion. They used nine sets of perpendicular bisectors from edge to edge for the Hα shell and X-ray shell and adopted their average center, 05^h19^m3483, −69°02'0692 (J2000), to be the site of SN explosion in SNR 0519–69.0. In our study, we fit an ellipse to the bright rim of the Balmer-dominated shell and adopt its center as the site of SN explosion, 05^h19^m3472, −69°02'0757 (J2000). As shown in Figure 4, there is an offset of 045 between our center and that of EPS2012. This offset is caused by our different treatment of the faint northeast arc—we consider it an anomaly due to a lower density in this quadrant and ignore it when fitting an ellipse to the Hα shell rim, while EPS2012 included the faint arcs as the shell edge. Without knowing details of the explosion geometry, we can only treat the difference in these two determinations of SN explosion site as additional uncertainty. We adopt the average of centers determined by EPS2012 and our method as the site of SN explosion, 05^h19^m3477, −69°02'0725 (J2000); see Figure 4. The coordinates of these centers are listed in Table 3.

Zoom In Zoom Out Reset image size

SNR 0519–69.0 has an age of 600 ± 200 yr (Rest et al. 2005). Adopting the largest runaway velocity of a surviving MS companion, 270 km s⁻¹, and a surviving helium star companion, 730 km s⁻¹ (Pan et al. 2014), the runaway distance for a surviving MS companion can be up to 0.2 pc (09), and that of a surviving helium star companion can be up to 0.6 pc (25). The search radii for MS and helium star companions listed in Table 3 are determined by adding the uncertainty of explosion center, 02, to the runaway distance. All stars with V < 23.0 mag within the helium star search radius, 27, are examined for plausible candidates of a surviving companion.

Within the search radius 27 in SNR 0519–69.0, only eight stars are brighter than V = 23.0 mag. These eight stars, numbered in the order of increasing distance from the explosion center, are listed in Table 4 and marked in the close-up images in Figure 5 and the V versus B–V and I versus V–I CMDs in Figure 6. These stars are too few to show clearly the locations of the MS and the RG branch in the CMDs. We have thus plotted stars with V < 23.0 mag within 220 from the site of SN explosion in the CMDs in Figure 6. The greatest majority of these stars belong to the background stellar population, and the MS and RG branch are clearly visible in the CMDs, providing convenient references for candidate stars within the search radius.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In these CMDs we have also plotted the post-impact evolutionary tracks of MS (the curved tracks above the MS) and helium star (the vertical tracks to the left of the MS) surviving companions (Pan et al. 2014). None of the eight stars within 27 from the site of explosion fall on the two sets of tracks. Thus, we conclude that no viable candidates of surviving MS or helium star companions are present based on the V versus B–V and I versus V–I CMDs, consistent with but more stringent than the conclusion of EPS2012.

Only five stars are brighter than V = 21.6 mag within the search radius 27 of SNR 0519–69.0. The physical parameters and radial velocities of these stars are derived from model fits and are listed in Table 5. We have plotted the radial velocities of these five stars versus their distance to the SN explosion site in Figure 7. These five stars are too few to illustrate a statistically meaningful radial velocity distribution; therefore, we have added all stars with V < 21.6 mag within 60 from the SN explosion site in Figure 7 to show the radial velocity distribution in this neighborhood. It is noted that this larger radius, 60, encompasses candidate stars with V < 21.6 mag that are selected in EPS2012. From Figure 7, we find that star 5 has a radial velocity, 182 ± 0 km s⁻¹, that deviates more than 2.5σ from the mean velocity, 264 km s⁻¹; furthermore, its radial velocity is not well populated by the LMC or Galactic stars. To examine whether the radial velocity from the model fit is reliable, we display the spectrum of star 5 with its model fit in Figure 8. As shown in this figure, the Hα and Hβ lines of star 5 are contaminated by the nearby Balmer-dominated shocks and are not possible for the determination of radial velocity, but the calcium line at 8542.1 Å, with the model fit, looks viable. This peculiar radial velocity indicates that star 5 may be a promising candidate of a surviving companion. Furthermore, assuming that star 5 has a mass of ∼1 M_⊙, the stellar effective temperature and surface gravity from the spectral fit imply an M_V of 2.54 mag, and the observed V = 21.0 mag requires the distance to be ∼49 kpc, consistent with the LMC distance. We will further discuss this star in Section 7.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

PS2015 applied the perpendicular bisector method (Schaefer & Pagnotta 2012) to the GMOS Hα image and Chandra X-ray images (Hughes et al. 2003; Rakowski et al. 2003) of DEM L71 to assess an average geometric center of the remnant, and they adopted this center as the explosion site. For the X-ray images, the authors used four gas regions to locate the X-ray geometric center: the extreme faint outer edge, the bright rim of the outer shell, the inner region, and the central minimum. Their final SN explosion site is adopted to be 05^h05^m4271, −67°52'4350 (J2000). These centers are marked in Figure 10.

The physical origins of the above four gas regions are different and need to be noted. DEM L71 presents an ideal example of the double-shock structure, with forward shocks expanding into the ISM and reverse shocks moving into and heating the ejecta (Hughes et al. 2003; Rakowski et al. 2003). The outer shell in X-rays is thus associated with the forward (collisionless) shock into the partially neutral ISM. The bright X-ray shell rims correspond to sight lines parallel to the shock fronts and include the longest emitting path lengths, while the faint emission extending beyond the bright shell rim is associated with blowout-like structures into a lower-density medium. The inner X-ray emission region corresponds to SN ejecta that has been heated by the reverse shock, and the central minimum originates from the central cavity.

In our previous study of N103B (Li et al. 2017), the geometric center of its Balmer-dominated shell was identified as the site of SN explosion because the Balmer shell follows closely the shock fronts into the ISM. The Balmer-dominated shell of DEM L71 is somewhat irregular. We have visually fitted two ellipses to the shell (as shown in Figure 9) and mark their centers and their average in Figure 10 for comparison. Our centers are offset to the northeast of the Hα center of PS2015 because of our different treatment of the faint southwest protrusion of the Balmer shell. We assume that the faint protrusion is caused by a local anomalously low density in the ISM and fit ellipses to only the bright rims of the Balmer shell, while PS2015 include the outer edge of the protrusion in their derivation of Hα center.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 10 also shows that the centers PS2015 determined from the central minimum and the inner X-ray emission are both offset significantly to the northeast of the center we derived from the Balmer shell. As noted above, the inner X-ray emission is associated with the SN ejecta. This offset is expected because the SN ejecta carried the same orbital velocity of the WD while the Balmer shell interacts and expands in the ISM. The offset between the center of ejecta and the center of SNR is expected to increase with age.

The detailed analysis of the physical structure of DEM L71 will be reported in a future paper. Here we adopt our average Balmer center as the site of SN explosion and use the difference between our centers as the uncertainty. All centers are listed in Table 3.

DEM L71 has an age of 4360 ± 290 yr (Ghavamian et al. 2003). Adopting the largest runaway velocity, 270 and 730 km s⁻¹, for surviving MS and helium star companions, respectively, the runaway distance for a surviving MS companion can be up to 1.3 pc (51), and that for a surviving helium star companion can be up to 3.5 pc (139). We determine the search radii for surviving MS and helium star companions by adding the uncertainty of the SN explosion site, 06, to the runaway distance. These search radii are listed in Table 3. In this study, all stars with V < 23.0 mag within the helium star search radius, 145, are examined as potential candidates of a surviving companion.

A total of 89 stars are brighter than V = 23.0 mag within the search radius 145 in DEM L71. These 89 stars are numbered in the order of increasing distance to the site of explosion and marked in the close-up images (Figure 11) and V versus B–V and I versus V–I CMDs (Figure 12). Their photometric measurements are listed in Table 6. In the CMDs, these 89 stars do not show clearly the locations of the MS and RG branch. We have thus plotted all stars with V < 23.0 mag within 480 from the site of explosion in the CMDs (Figure 12) in a different symbol (gray cross) to illustrate the MS and RG branch in the CMDs.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the CMDs (Figure 12), we have overplotted the post-impact evolutionary tracks of the MS (the curved tracks above the MS) and helium star (the vertical tracks to the left of the MS) surviving companions (Pan et al. 2014). From locations of stars in the CMDs, we find no star matches the evolutionary tracks for MS and helium star surviving companion candidates within the search radius of 145, as shown in the close-up in Figures 13 and 14. Our study has a different search area and candidates from PS2015, but our photometric search has not found any obvious candidate of a surviving companion, either.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

There are 32 stars with V < 21.6 mag within the 145 search radius in DEM L71. The stellar parameters and radial velocities of these stars from model fits are listed in Table 7. We have plotted the distribution of radial velocities versus distances to the explosion site of these stars in Figure 15. While the overall distribution roughly reveals a Gaussian profile, we find that star 80 has a radial velocity, 213 ± 0 km s⁻¹, that is deviated more than 2.5σ from the mean, 270 km s⁻¹ (Figure 15); furthermore, its radial velocity is not well populated by the LMC or Galactic stars. The model fit of the spectrum of star 80, displayed in Figure 16, looks reasonable. The peculiar radial velocity of star 80 makes it an intriguing candidate for a surviving companion. Assuming that star 80 has a mass of ∼1 M_⊙, the stellar effective temperature and surface gravity from the spectral fit implies an M_V of 1.33 mag, and the star would be at a distance of ∼38 kpc, as it has an observed V = 19.25 mag; if a mass of ∼1.69 M_⊙ is assumed, the star would be at a distance of ∼50 kpc. Follow-up spectroscopic observations of star 80 in blue wavelengths are needed for better determination of its physical properties. We will further discuss this star in Section 7.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The Balmer shell of SNR 0548–70.4 shows a regular shape, similar to the case of SNR 0509–67.5 (Litke et al. 2017). We find that over 75% of the outer shell periphery can be well fitted by an ellipse. In the northwest and southeast quadrants, the faint extended emission deviates from the fitted ellipse and appears to be associated with blowout structures. Thus, we identify the center of the ellipse as the site of SN explosion (see Figure 17). We will use a larger search radius to take into account the uncertainty due to the irregularity in the Balmer shell periphery.

We choose the MS and helium star search radii of SNR 0548–70.4 that are sufficiently larger than the respective runaway distances of the MS and helium star companions from the explosion site. SNR 0548–70.4 has an age of 10,000 yr (Hendrick et al. 2003). Adopting the largest runaway velocity of 270 and 730 km s⁻¹ for surviving MS and helium star companions, we calculate their runaway distance up to 2.8 pc (110) and 7.5 pc (298), respectively. We then adopt 5.0 pc (200) and 10.0 pc (400) as the search radii for MS and helium star companions, respectively (Figure 17). These limits correspond to a kick velocity of 500 and 1000 km s⁻¹, respectively. All stars with V < 23.0 mag within the search radius of the helium star, 400, are compiled for surviving companion candidates.

Zoom In Zoom Out Reset image size

Within the search radius 400 in SNR 0548–70.4, 973 stars are brighter than V < 23.0 mag. These 973 stars are numbered in the order of increasing distance from the explosion site and listed in Table 9 with their photometric measurements. We have marked these stars in the Hα (Figure 18) and color-composite images (Figure 19), as well as in the V versus B–V and I versus V–I CMDs (Figure 20). In the CMDs, these stars show hints of the MS and the RG branch. To better illustrate the locations of the MS and RG branch in the CMDs, we have also plotted all stars with V < 23.0 mag within 600 from the explosion site.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In the CMDs (Figure 20), we have overplotted the post-impact evolutionary tracks of surviving MS (the tracks above the MS) and helium star (the vertical tracks to the left of the MS) companions (Pan et al. 2014). We find that star 98 lies on a track of a surviving MS star model in the V versus B–V CMD (Figure 21), but not in the I versus V–I CMD (Figure 22); furthermore, its location in the V versus B–V CMD indicates an MS surviving companion going through ∼110 yr after the SN explosion. It is not consistent with the age of SNR 0548–70.4, 10,000 yr (Hendrick et al. 2003). Assuming negligible extinction, star 98's B–V ∼ 0.34 suggests an early F spectral type, and its V–I ∼ 1.08 suggests an early K spectral type.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While photometric comparisons with post-impact evolution models in CMDs do not yield promising candidates for surviving companions, we notice puzzling photometric properties of star 11. In the CMDs, star 11 is located in regions not well populated by LMC or Galactic stars, as shown in Figure 20. It is far away from the post-impact evolution tracks of stars; furthermore, its colors are inconsistent with any spectral types. Assuming negligible extinction, star 11's B–V ∼ 0.74 suggests a spectral type of G, and its V–I ∼ 2.68 suggests a spectral type of M3. In the V band, star 11 is twice as bright as the Sun, while in the I band star 11 is 13 times as bright as the Sun. This particularly red V − I color suggests infrared excess. Interestingly, star 770 and star 934 appear to have puzzling photometric properties in the CMDs as well. These three stars will be discussed further in Section 7.

To probe the parameter space of a close nondegenerate companion of a WD that leads to an SN Ia of SD origin, detailed models of binary stellar evolution for a range of separations, mass ratios, and WD masses have been calculated (Hachisu et al. 1999; Han & Podsiadlowski 2004; Meng et al. 2007, 2009; Hachisu et al. 2008; Han 2008; Wang et al. 2009; Wang & Han 2009, 2010); however, most of these models end at the SN explosion without extending to the SN impact on the stellar companion and afterward. On the other hand, some hydrodynamic model calculations start from the SN Ia explosion and focus on the impact of SN ejecta on the surviving stellar companion and its subsequent evolution, taking into consideration explosion geometry and impact angle (Marietta et al. 2000; Pakmor et al. 2008; Liu et al. 2012; Pan et al. 2014). In addition, less sophisticated calculations assuming ad hoc energy input and mass stripping have been made for the post-impact evolution of a 1 M_⊙ subgiant or MS surviving companions (Podsiadlowski 2003; Shappee et al. 2013).

Among the above models, Pan et al. (2014) have carried out comprehensive calculations that detail the post-impact evolution of a surviving companion's effective temperature and luminosity. They considered only MS and helium star companions. For a given temperature and luminosity of a surviving companion, its stellar emission was approximated by a blackbody model and convolved with a filter response curve to compute the photometric magnitude in that filter. Thus, the post-impact evolution of a surviving companion in the temperature−luminosity domain could be easily plotted in CMDs for direct comparisons with observations.

Using HST photometric measurements of stars in SNRs 0519–69.0, DEM L71, and 0548–70.4, we have constructed CMDs in V versus B − V and I versus V − I and compared the stars with Pan et al. (2014) evolutionary tracks of surviving MS and helium star companions (Figures 6, 12, and 20). We do not find any MS or helium star candidates with V < 23.0 for surviving companion in these three Type Ia SNRs. As Pan et al. (2014) did not model cases with RG companions, we cannot assess the existence of surviving RG companions in these SNRs, although there are a number of stars within our search radii that lie on the RG branch in the CMDs. As shown by spectroscopic observations and discussed in the next subsection, some RG stars appear to show large peculiar radial velocities. Future models of surviving RG companions are needed for comparison.

Interestingly, in the SNR 0548–70.4, we find star 11 within the MS search radius in a strange location in the CMDs (Figure 20). Its colors cannot be reproduced by stars of any temperature with a normal extinction law. As there are additional sources with similar colors and location in the CMDs (e.g., star 770 and star 934), and the colors are similar to some cataloged galaxies, we think that star 11 is most likely a background galaxy, instead of a star, as well as source 934. Unfortunately, star 11 is below the detection limit of the Two Micron All Sky Survey (2MASS) catalog; future sensitive JHK photometric measurements are needed to extend its SED to confirm its nature as a background galaxy.

We have also used VLT MUSE observations to carry out spectroscopic analyses, and we have used stellar spectra to search for high radial velocities as diagnostics of surviving companions. In the SNRs 0519–69.0 and DEM L71, we find that each has a star with radial velocity that deviates by more than 2.5σ (75 and 50 km s⁻¹) from the mean of the underlying stellar population, 264 and 270 km s⁻¹, respectively (see Figures 7 and 15). These peculiar radial velocities are intriguing and may suggest that these stars are surviving companions. In the case of 0519−69.0, star 5 is halfway between the MS and helium star search radii, while in the case of DEM L71, star 80 is close to the helium star search radius; however, both stars with peculiar radial velocities are located in the RG branch in the CMDs. Because of the uncertainties in the exact site of SN explosion and our inadequate understanding of the SN Ia progenitors, we can neither confirm nor exclude these two stars as candidates for surviving companions in SNRs 0519–69.0 and DEM L71. Note that DEM L71 has two other stars with radial velocities at >2σ but <2.5σ from the mean radial velocity of the underlying population, and both their locations appear to be in the MS in the CMDs. Follow-up observations to measure chemical abundance and stellar rotation of these stars are needed to determine whether they are indeed surviving companions of SN Ia progenitors.

Besides the aforementioned methods, we have also looked into Gaia Data Release 2 (DR2) to investigate proper motions of stars near the explosion sites of the three SNRs, and we search for large proper motions to diagnose surviving companions. A transverse velocity of 300 km s⁻¹ at the LMC distance of ∼50 kpc corresponds to a proper motion of 1.26 mas yr⁻¹. Most stars observed in these Type Ia SNRs are fainter than V = 18–19. As the quoted uncertainty in proper motion in Gaia DR2 is 1.20 mas yr⁻¹ for a star with G = 20, it would be impossible to use Gaia DR2 to conduct a meaningful investigation of peculiar proper motions of intermediate- and low-mass stars. Stars near the tip of the RG branch are bright enough to have more accurate proper-motion measurements, but no such RGs are seen in these Type Ia SNRs.

In the SD scenario, the companion of the WD progenitor is expected to survive the SN explosion and be identifiable (Marietta et al. 2000; Pakmor et al. 2008; Liu et al. 2013; Pan et al. 2014); however, recent searches of a surviving companion in the Milky Way (MW; Ruiz-Lapuente et al. 2004; González Hernández et al. 2009, 2012; Kerzendorf et al. 2009, 2012, 2013a, 2013b, 2018a, 2018b; Ruiz-Lapuente et al. 2018, 2019) and the LMC (Schaefer & Pagnotta 2012, EPS2012, PS2015, Li et al. 2017, Litke et al. 2017) have not unambiguously identified and confirmed any surviving companion. In this study, we have used photometric and spectroscopic methods to search for candidates of a surviving companion within the three young LMC Type Ia SNRs, but we are still unable to make unambiguous identification and confirmation. See Appendix A for a detailed compilation of these searches and their results.

Nevertheless, the lack of an obvious surviving companion cannot exclude the SD scenario. In the SD scenario, the modeled post-impact properties of a surviving companion are usually calculated under reasonable assumptions for a simplified condition. It is conceivable that the assumed conditions are not appropriate and lead to discrepancies between the observations and model predictions of surviving companions. For instance, models of Pan et al. (2014) adopt a classical SD scenario, in which a normal Chandrasekhar mass model (Whelan & Iben 1973; Nomoto 1982) with an initial spherically symmetric explosion is assumed. If the WD in the SD case has a sub-Chandrasekhar mass instead and explodes through double detonation (Nomoto 1982), the explosion energy will be lower than that assumed by Pan et al. (2014). Furthermore, the SN explosion may not be spherically symmetric, as a result of the binary orbit and stellar rotation (Kashi & Soker 2011). These different explosion energies and geometries will certainly affect the post-impact temperature and luminosity of the surviving companions and change their locations in the CMDs. Future post-impact evolution models of surviving companions need to consider different types of SN explosions and, more importantly, include RG companions.

It should be noted that the conventional SD and DD classifications of SNe Ia may be an oversimplification. Many alternative models for SNe Ia have been proposed (see Wang & Han 2012; Maoz et al. 2014; Ruiz-Lapuente 2014, 2018; Wang 2018, for reviews), such as the sub-Chandrasekhar model (Nomoto 1982; Woosley et al. 1986; Bildsten et al. 2007; Shen & Bildsten 2009; Fink et al. 2010; Guillochon et al. 2010; Dan et al. 2011; Raskin et al. 2012; Pakmor et al. 2013; Shen et al. 2018), the super-Chandrasekhar model (Yoon & Langer 2004; Hachisu et al. 2012a, 2012b; Boshkayev et al. 2013; Wang et al. 2014), the spin-up/spin-down model (Di Stefano et al. 2011; Justham 2011), the single-star model (Iben & Renzini 1983; Tout et al. 2008), the violent DD merger model (Pakmor et al. 2010, 2011, 2012), the collisional DD model (Chomiuk et al. 2008; Raskin et al. 2009; Kushnir et al. 2013), the core-degenerate (CD) model (Kashi & Soker 2011), the M dwarf model (Wheeler 2012), etc. These models predict different outcomes for SNe Ia apart from the classical SD scenario, resulting in different geometric distribution and abundance of the SN ejecta, whose impact on a surviving companion would manifest differently and need further exploration.

Finally, from studies of the five young Balmer-dominated Type Ia SNRs in the LMC, we find the following: no surviving stellar companion exists in 0509–67.5 (Schaefer & Pagnotta 2012; Litke et al. 2017); star 1 in N103B is a promising candidate for a surviving companion (Li et al. 2017); 0548–70.4 has no obvious candidate for a surviving companion, while DEM L71 and 0519–69.0 each have a possible candidate for a surviving companion based on anomalous radial velocities (this paper). These results imply that 20%–60% of them may originate from SD SNe Ia. This is not inconsistent with the 20% suggested by González Hernández et al. (2012); however, these results are derived with small number statistics. Future surveys of young Balmer-dominated Type Ia SNRs in nearby galaxies, such as M31 and M33, should be made in order to expand the Type Ia SNR sample size, and the 30 m class telescopes under construction can be used to search for surviving companions of SNe Ia in the future.