SN 2022vqz: A Peculiar Subluminous Type Ia Supernova with Prominent Early Excess Emission

We performed follow-up optical photometric observations of SN 2022vqz with multiple facilities, including the Tsinghua-NAOC 80 cm telescope at Xinglong Observatory of NAOC (TNT; Wang et al., 2008; Huang et al., 2012), the AZT-22 1.5 m telescope at Maidanak Observatory (Ehgamberdiev, 2018), and the Schmidt 67/92 cm and the 1.82 m Copernico Telescope at Cima Ekar Observing Station (Padova). Our multiband photometric monitoring began on 2022 Sep. 27.80 (MJD 59849.80) and has good coverage over a period of $\sim 2.5$ months. Standard IRAF¹¹1IRAF (Image Reduction and Analysis Facility) is distributed by the National Optical Astronomy Observatories (NOAO), which are operated by the Association of Universities for Research in Astronomy (AURA), Inc., under cooperative agreement with the U.S. National Science Foundation. routines were adopted to reduce the CCD images, including bias and flat-field corrections. The magnitudes are calibrated using a set of nearby reference stars from the Sloan Digital Sky Survey (SDSS) catalogue (York et al., 2000; Gunn et al., 2006). The number of reference stars ranges from 30 to 40 for the different facilities. For flux calibration of the $UBVRI$-band photometry from AZT and $BV$-band photometry from TNT, the SDSS $ugriz$ magnitudes of the reference stars are converted to standard $UBVRI$ magnitudes²²2http://classic.sdss.org/dr4/algorithms/sdssUBVRITransform.html#Lupton2005.

Although SN 2022vqz is offset from the centre of its host galaxy, background-light contamination is still prominent, especially for its late-time light curves. Usually, a host-galaxy template can be taken when the SN fades away, but this will take a long time for SN 2022vqz. As an alternative, we construct the host-galaxy templates of SN 2022vqz using two different methods. The first is smoothly interpolating the host-galaxy flux at the location of the SN by solving a diffusive equation bounded by surrounding starlight (Zhang et al., 2004). The second is using the flux information from the opposite side as a template since the host galaxy looks symmetric. The subtraction process follows Zrutyphot (Mo et al., in prep.). With the templates constructed by the above two methods, the final results of the photometry are consistent. Figure 1 shows an example of an original image, constructed template, and their difference image.

In our analysis, we also included the publicly available $gr$-band forced photometry from ZTF (Masci et al., 2019)³³3https://lasair-ztf.lsst.ac.uk/object/ZTF22abhrjld/ and $co$-band (cyan, orange) forced photometry from ATLAS. These photomeric results were obtained after subtraction of the corresponding template images taken before the SN explosion, and hence are more reliable. The first observation of ATLAS was on 2022 Sep. 23.471 (MJD 59845.471), less than 1 day after the latest nondetection. Combined with the early ZTF observations which started $\sim 1$ day later, the early-time photometric evolution of SN 2022vqz can be well constrained. Figure 2 shows the multiband light curves of SN 2022vqz.

Our spectroscopy of SN 2022vqz covered phases from $-7.1$ to $+97.4$ days relative to its peak brightness. We obtained 9 spectra with BFOSC mounted on the 2.16 m telescope of the Xinglong Observatory (XLT), 5 spectra with YFOSC mounted on the Lijiang 2.4 m telescope (LJT; Fan et al., 2015) of Yunnan Observatories (YNAO), 3 spectra with the Kast spectrograph(Miller & Stone, 1994) on the 3 m Shane reflector at Lick Observatory, and 1 spectrum with AFOSC mounted on the 1.82 m Copernico Telescope of Padova. In addition, 1 spectrum was taken of the host galaxy MCG+05-03-011 with XLT+BFOSC. The Lick/Kast spectra were taken with the slit oriented along the parallactic angle to minimise the effects of atmospheric dispersion (Filippenko, 1982). The journal of spectroscopic observations is shown in Table 1.

Standard IRAF routines were adopted for reduction of all the optical spectra taken by LJT, XLT, Lick 3 m Shane, and Padova 1.82 m Copernico telescopes, including bias, flat-field corrections, and removal of cosmic rays. The Copernico spectrum is reduced using a dedicated pipeline foscgui⁴⁴4foscgui is a graphic user interface aimed at extracting SN spectroscopy and photometry obtained with FOSC-like instruments. It was developed by E. Cappellaro. A package description can be found at http://sngroup.oapd.inaf.it/foscgui.html .. The wavelength scale was calibrated with comparison-lamp spectra, and the fluxes were calibrated with standard stars observed on the same night at similar airmasses. All spectra were further corrected for atmospheric extinction using the extinction curves obtained at each observatory, and telluric lines were also removed from the spectra.

We also included two classification spectra from the Spectroscopic Classification of Astronomical Transients (SCAT; Tucker et al. 2022; Tucker 2022) and ZTF (Maguire et al., 2022), and one spectrum taken by Las Cumbres Observatory (LCOGT; Brown et al., 2013) in our analysis. The spectral evolution of SN 2022vqz is shown in Figure 7.

Figure 2 shows the multiband light curves of SN 2022vqz. The data reveal an early-time peak that is clearly separated from the main peak. The SN brightness increased by $\sim 0.6$ mag in the $o$ band in the first day after discovery, reaching an early peak and then dropping to a minimum in another few days, followed by a slow rise to maximum light. No secondary peak or significant shoulder can be seen in the $i$ and $R$ bands, consistent with the light curves of subluminous SNe Ia such as the 02es-like or 91bg-like objects.

Applying a polynomial fit to the $B$-band light curve around maximum light indicates that SN 2022vqz reaches the peak on MJD $59863.43\pm 0.48$, with $m_{B,\rm max}=16.205\pm 0.053$ mag. We use the time of $B$ maximum as the reference epoch throughout this paper. The decline in the first 15 days after peak is measured as $\Delta m_{15,B}=1.33\pm 0.11$ mag, slightly fast among normal SNe Ia.

The distance modulus of SN 2022vqz can be estimated based on the redshift of its host galaxy, which is in the Hubble flow: $z_{\rm CMB}=0.01594$ after correcting for the solar peculiar velocity. Adopting a cosmological model with $\Omega_{m}=0.27$, $\Omega_{\Lambda}=0.73$, and $H_{0}=73.04~{}{\rm km~{}s^{-1}~{}Mpc^{-1}}$ (Riess et al., 2022), the corresponding distance modulus is $m-M=34.11\pm 0.15$ mag. With this value, the absolute peak magnitude of SN 2022vqz is $M_{B,\rm max}=-18.11\pm 0.16$ mag, after accounting for a Galactic extinction of $A_{B}=0.21$ mag (Schlafly & Finkbeiner, 2011). The extinction of the host galaxy is assumed to be negligible, considering that the SN is offset from the centre of the host and no Na I D absorption can be recognised in the SN spectra.

Figure 3 shows the location of SN 2022vqz in the $\Delta m_{15,B}$–$M_{B,\rm max}$ diagram, together with those of some comparison objects including normal SNe Ia. Its low peak luminosity and moderate decline rate fit into the category of 02es-like objects. One may note that with the discovery of some events in recent years, especially SN 2019yvq, the gap between 02es-like objects and normal SNe Ia now seems to be filled by a continuous distribution. However, the absolute magnitude of SN 2019yvq is highly uncertain owing to large discrepancies (over 2 mag) of its host-galaxy extinction estimates from different methods (Burke et al., 2021). In Figure 3 we adopt the value reported by Burke et al. (2021).

We further compare in Figure 4 the absolute $UBgV$ and $rRiI$ light curves of SN 2022vqz with those of some well-observed SNe Ia, including 02es-like SN 2002es (Ganeshalingam et al., 2012), SN 2016ije (Li et al., 2023), SN 2019yvq (Burke et al., 2021), iPTF14atg (Cao et al., 2015), 91bg-like SN 1999by (Garnavich et al., 2004), and the normal SN 2004eo (Pastorello et al., 2007) but with a similar $\Delta m_{15,B}$ (1.46 mag) as SN 2022vqz.

The light-curve morphology of SN 2022vqz is overall similar to that of iPTF14atg and SN 2019yvq. We notice that SN 2022vqz and other 02es-like objects do not exhibit a rapid decline after $t\approx 1$ month when compared to SN 2002es. This indicates an additional peculiarity intrinsic to SN 2002es that is still unexplained (Taubenberger, 2017). The absolute magnitudes of SN 2022vqz are slightly brighter than all objects in the 02es-like comparison sample, while they are all bounded by subluminous 91bg-like SN 1999by and normal SN 2004eo. Note that in Figure 4 we assumed zero host extinction for all 02es-like SNe. SN 2019yvq appears not to be brighter than other 02es-like objects without applying a significant host extinction correction. Some evolution features seen in SN 2016ije, like slower $B$ and $V$ decline rates after $t\approx+20$ days and strong line blanketing in the $U$ band, are absent in SN 2022vqz. These features can be interpreted as the result of a broader line-forming region and a smaller total amount of iron-group elements (IGEs) distributed in the line of sight. A more extended line-forming region could result in stronger line blanketing at early phases, while less IGEs can reduce their absorptions in the spectra, especially at short wavelengths, thereby providing more-luminous late-time fluxes and flatter light curves in $BV$ (Li et al., 2023). In contrast, SN 2022vqz should have a less extended line-forming region for IGEs, suggesting a more stratified and less mixed IGE distribution.

Figure 5 shows the comparison of the colour evolution, with all data corrected for Galactic extinction. Overall, the colour curves of SN 2022vqz show a close resemblance to those of other 02es-like SNe, while it seems to have a more extended and rapid evolution in $g-i$ relative to iPTF14atg. In $B-V$, SN 2022vqz shows slower evolution than SN 1999by, and it is $\sim 0.2$–0.3 mag redder than SN 2004eo before $t\approx 15$ days. Immediately after the explosion, SN 2022vqz appears exceptionally blue, $g-r\approx-0.15\pm 0.09$ mag at $t\approx-15$ days from maximum brightness, and it then evolves rapidly to a red peak two days later. This trend is coincident with the declining part of the early peak of SN 2022vqz, suggesting a blue colour of the early peak luminosity.

An accurate explosion time estimation is crucial for further analysis of the explosion physics of SN 2022vqz. A nondetection was reported $\sim 1$ day before the discovery, indicating that this SN exploded not long before the first observations. The time of explosion can be estimated from the early phase of the multiband light curves by adopting a simple “fireball” model (Riess et al., 1999; Nugent et al., 2011). Assuming the early-time SN can be described by a photosphere expanding at a constant velocity, its luminosity $L$ should be proportional to $(t-t_{\rm exp})^{2}$, where $t_{\rm exp}$ is the time of explosion and should be same for all bands.

Note that SN 2022vqz shows an early peak which lasts for $\sim 3$ days since time of discovery and cannot be fitted by the fireball model. To avoid its influence, photometric data before MJD 59848 are excluded from the fit. The fitting method follows that described by Xi et al. (2022), and data earlier than 5 days before $B$ maximum are used. The fitting procedure gives $t_{\rm exp}={\rm MJD}~{}59844.48\pm 0.20$, which coincides with the last nondetection time, MJD 59844.484. Given the uncertainty of $t_{\rm exp}$, the SN may have either not yet exploded or be too dim for detection at the latest time of nondetection. The rise time is thereby given as $t_{\rm rise}=18.63$ days. Figure 6 shows the fitting result.

Based on the $UBgVrRiI$-band photometry, we adopted SNooPy2 (Burns et al., 2011, 2014) to construct the bolometric light curve of SN 2022vqz. Since SN 2022vqz is a subluminous object, we use spectral energy distribution (SED) templates of subluminous 91bg-like SNe Ia (Nugent et al., 2002) for flux integration.

The bolometric light curve is found to reach its peak at $L\approx 5.1\times 10^{42}~{}\rm{erg~{}s}^{-1}$ on MJD 59863.3. According to Arnett’s rule (Arnett, 1982; Stritzinger & Leibundgut, 2005), the mass of synthesised radioactive ⁵⁶Ni can be inferred as $M_{\rm Ni}\approx 0.25{\rm M_{\odot}}$. A more robust way for Ni mass estimation is to fit the full bolometric light curve using the radiation diffusion model of Arnett (1982) (see also Chatzopoulos et al., 2012, 2013). The best fit yields the explosion time $t_{0}$, the initial mass of radioactive nickel $M_{\rm Ni}$, the light-curve timescale $t_{\rm lc}$, and the gamma-ray leaking timescale $t_{\gamma}$ as (respectively) MJD $59848.6\pm 2.2$, $0.20\pm 0.04$ M_⊙, $12.20\pm 3.87$ days, and $36.26\pm 4.73$ days. The explosion time inferred from the bolometric light-curve fit is about 4 days later than that from the early-time light curve, and well positioned at the minimum light between the early and main peaks, indicating a delay between the SN explosion and the emergence of the radioactive nickel-powered light curve. Such a delay was predicted as a “dark phase” (Piro & Nakar, 2013; Piro & Morozova, 2016) caused by the location of the radioactive ⁵⁶Ni deep within the ejecta. This delay will also result in a shorter rise time, hence a lower ⁵⁶Ni mass estimation.

The early peak bolometric luminosity was estimated as $4.5\times 10^{41}~{}\rm{erg~{}s}^{-1}$. However, this value may be inaccurate since it only uses the ZTF $gr$-band photometry, and the rest of the flux was extrapolated from the template SED.

Figure 7 shows the spectral evolution of SN 2022vqz, spanning from $t\approx-15.7$ days to $+97.4$ days relative to $B$ maximum, along with a spectrum of the host galaxy. The spectral evolution generally exhibits some common properties of subluminous SNe Ia. The Si II $\lambda$5972 and O I $\lambda$7774 absorptions are relatively prominent around maximum light, implying a low ejecta temperature and a significant amount of unburnt materials. We measured the equivalent width ($W$) of the Si II $\lambda$6355 and $\lambda$5972 features in the $-0.7$ day spectrum: $W(5972)=59\pm 12$ Å and $W(6355)=137\pm 12$ Å. These values make SN 2022vqz an extreme object in the “cool” group proposed by Branch et al. (2006). As explained by Branch et al. (2006), there is a temperature threshold of $\sim 7000$–8000 K (Hoeflich et al., 1993; Höflich et al., 2002), below which the line optical depth will abruptly change owing to variations in key ionisation ratios, so the “cool” objects can be well separated from normal ones in the $W(6355)$–$W(5972)$ plane. The equivalent width of O I $\lambda$7774 absorption in the same spectrum is measured to be $W(7774)=116\pm 14$ Å, significantly higher than those in normal SNe Ia ($\lesssim 70$ Å; Zhao et al. 2016). In Figure 8, the spectroscopic properties of SN 2022vqz are compared with those of other subluminous SNe Ia, including 02es-like objects iPTF14atg (Cao et al., 2015), PTF10ops (Maguire et al., 2011), SN 2002es (Ganeshalingam et al., 2012), and SN 2016ije (Li et al., 2023). A 91bg-like object SN 1999by (Garnavich et al., 2004) and the normal SN Ia 2004eo (Pastorello et al., 2007) are also included for comparison.

At $t\approx-7$ days (Figure 8a), the spectrum of SN 2022vqz is characterised by broad P Cygni lines of IMEs typical among SNe Ia: the Si II $\lambda$6355 and $\lambda$5972 absorptions, the “W"-shaped S II feature around 5400 Å, and deep absorptions of O I $\lambda$7774 and the Ca II near-infrared (NIR) triplet. The overall morphology is quite similar to that of iPTF14atg. However, no C II $\lambda$6580 is detectable in SN 2022vqz, while it is strong in iPTF14atg. The weaker features of Si II and Fe II around 5000 Å are narrower and hence less blended compared to PTF10ops or SN 2004eo.

Near maximum light (Figure 8b), all 02es-like objects show an absorption feature of C II $\lambda$6580, except SN 2022vqz. The IGE lines around 5000 Å are deeper and narrower than in iPTF14atg or SN 2016ije, indicating a more concentrated density distribution of the ejecta. The Ti II feature near 4200 Å, which is characteristic of subluminous SNe Ia, is less developed than those of SN 2002es and 91bg-like SN 1999by at this epoch, and is more similar to iPTF14atg.

By $t\approx$ 14 days (Figure 8c), the spectra of SN 2022vqz resemble those of iPTF14atg and SN 2002es. The Ti II lines have developed in the spectrum of SN 2022vqz, notably the deep trough around 4200 Å and the W-shaped feature around 6800 Å. These absorptions are less prominent or absent in SN 2016ije or SN 2004eo. The O I and Ca II lines remain strong, but narrower than those seen near maximum light.

At $t\approx 40$ days after maximum light (Figure 8d), the spectra are dominated by IGE lines. Si II $\lambda$6355 is still pronounced in SN 2022vqz, while it is contaminated by the Fe II lines and hardly noticeable in other objects of the comparison sample. The Si II $\lambda$5972 absorption is also replaced by Na I absorption, while the O I and Ca II lines are still persistent in the spectra.

The expansion velocity of SN ejecta can be measured from the blueshifted absorption of P Cygni profiles. Differences in line velocities can reveal the element distribution within the ejecta. The velocity evolution measured from absorptions of some IMEs, including S II $\lambda\lambda$5460, 5640, O I $\lambda$7774, Ca II $\lambda$8542, and Si II $\lambda$6355, appear to be exceptionally flat as shown in Figure 9(a). Despite having relatively large uncertainties, the Si II and S II velocities even show a global rise at early phases, conflicting with the trend that the line velocities should decrease rapidly after SN explosion as a result of the receding photosphere. This complex velocity evolution might be related to the double-peaked light curves of SN 2022vqz. Considering energy components, each results in a declining line velocity. When the first peak fades away and the second takes control, the apparent line velocity, as a weighted average of the two components, may show a temporary increase. The Si II velocity measured around the time of maximum light is $v_{\rm Si,max}=6900\pm 500~{}\mathrm{km~{}s}^{-1}$, which is significantly lower than the typical value of $\sim 10,500$ km s^-1 for normal SNe Ia, and it is among the lowest values even for subluminous 91bg-like and 02es-like objects. The post-peak velocity decline of SN 2022vqz is $\dot{v}=-55\pm 10~{}\mathrm{km~{}s^{-1}~{}d^{-1}}$, making it a low-velocity-gradient (LVG) object according to the classification scheme proposed by Benetti et al. (2005). We do not find signatures of high-velocity features (HVFs) in SN 2022vqz, while the absence of HVFs seems to be a common property of subluminous SNe.

The velocity evolution of Si II $\lambda$6355 is further compared to that of some other SNe Ia, including 02es-like iPTF14atg, SN 2019yvq, SN 2016ije, subluminous SN 1991bg (Turatto et al., 1996), normal SN 2004eo (Pastorello et al., 2007), and SN 2011fe (Pereira et al., 2013; Zhang et al., 2016). The latter two are overplotted to visualise the typical velocity evolution of normal SNe Ia. The flat evolution and low velocity make SN 2022vqz stand out in Figure 9(b). We also notice that, although SN 2019yvq was classified as a member of the 02es-like subclass according to its underluminous signatures (Burke et al., 2021), its velocity is extremely high for an 02es-like object, especially when compared with SN 2022vqz. This large diversity in ejecta velocity will be very challenging if a single mechanism is adopted to explain all 02es-like objects.

SN 2022vqz is hosted by the face-on lenticular galaxy MCG+05-03-011 with a distance modulus of $34.11\pm 0.15$ mag. The resulting $g$-band absolute magnitude is $-19.73\pm 0.15$ mag and the $g-i$ colour is 1.1 mag (SDSS; Almeida et al. 2023), which are typical for hosts of 02es-like SNe (White et al., 2015). Note that most 02es-like SNe Ia tend to explode in massive, early-type galaxies with little or no star-forming activity, such as SNe 2002es (Ganeshalingam et al., 2012), PTF10bvr, 10acdh, 10ujn (White et al., 2015), and iPTF14atg (Cao et al., 2015). For SN 2022vqz, however, the host spectrum exhibits prominent H$\alpha$ emission, indicating relatively active star formation at least near the centre of the host galaxy.

To determine the metallicity of the host galaxy, we measured the intensity ratio of [N II] $\lambda$6583 and H$\alpha$ as $\log(\mathrm{[N~{}II]/H}\alpha)=-0.4$, and that of [O III] $\lambda$5007 and H$\beta$ as $\log(\mathrm{[O~{}III]/H}\beta)=-0.1$. These values can be converted to a metallicity estimate of $12+\log(\mathrm{O/H})=8.64$ using an empirical relationship (Kewley & Ellison 2008, Eq. A9). We also used Firefly (Wilkinson et al., 2017) to fit the host spectrum with combinations of single-burst stellar population models, and obtained a stellar mass of $\log({M/\mathrm{M}_{\odot}})=9.79$, an age of 4.05 Gyr, and a metallicity $[\mathrm{Z/H}]=0.14$ dex. The metallicities inferred from the two methods are roughly consistent and comparable to the solar metallicity. Thus, we conclude that the host of SN 2022vqz is a medium-mass, solar-metallicity, star-forming galaxy.

Early-time photometry provides a distinct perspective and is crucial for discriminating different explosion mechanisms of SNe Ia (Maeda et al., 2018). Recent statistical studies based on ZTF samples found that $\sim 20$–30% of normal SNe Ia show some early-time flux excesses (Deckers et al., 2022; Magee et al., 2022). There is also large diversity in brightness, colours, timescales, and light-curve morphologies of these excesses. Different physical models have been proposed to explain these features, such as interaction with a nondegenerate companion star (Kasen, 2010), outward mixing of radioactive ⁵⁶Ni (Piro & Nakar, 2013; Piro & Morozova, 2016), detonation of a surface He layer (Polin et al., 2019), and interaction with CSM (Piro & Morozova, 2016; Piro et al., 2021). The rate of early excess in normal SNe Ia is related to the actual population of different physical channels. Even if all the normal SNe Ia are from the SD scenario, an early excess from companion interaction could only be detected in $\sim 10$% of the events owing to the viewing-angle effect (Kasen, 2010; Burke et al., 2022).

Obtaining high-quality early-time photometry necessitates early discovery, rapid classification, and high-cadence follow-up photometry of young SNe Ia, which require both extensive efforts and good luck. Early excess emission has been observed in detail only for a handful of normal SNe Ia, such as SN 2017cbv (Hosseinzadeh et al., 2017), SN 2018oh (Dimitriadis et al., 2019; Li et al., 2019; Shappee et al., 2019), SN 2019np (Sai et al., 2022), and SN 2023bee (Wang et al., 2023; Hosseinzadeh et al., 2023). In the future, with specially designed robotic facilities dedicated to high-cadence, multiband observations of young, nearby SNe Ia, we expect more such events to be observed and analysed in detail.

Although early-time flux excesses are relatively rare among normal SNe Ia, a series of recent studies reveals that the 02es-like SNe Ia tend to show them: see iPTF14atg (Cao et al., 2015), iPTF14dpk (Cao et al., 2016; Jiang et al., 2018), SN 2016jhr (Jiang et al., 2017), SN 2019yvq (Miller et al., 2020; Burke et al., 2021), SN 2022ywc (Srivastav et al., 2023), and SN 2022vqz (this work). This generic feature supports the idea that the 02es-like subclass likely originates from a single channel. However, the diversity in other aspects, including ejecta velocity, peak luminosity, and existence of early line blanketing, challenges the potential progenitor scenario and explosion mechanism. One possible scenario is the double-detonation model, in which the ignition of a surface He layer detonates the underlying C-O WD (Polin et al., 2019). The burning of the He layer will leave radioactive IGEs in the outermost ejecta and produce an early-time excess. This model was adopted to explain the early excesses of SNe 2016jhr and 2019yvq.

We compared the observations of SN 2022vqz with a grid of double-detonation models (Polin et al., 2019). The grid contains models with WD masses ranging from 0.6 to 1.3 ${\rm M}_{\odot}$ and He-shell masses varying between 0.01 and 0.1 ${\rm M}_{\odot}$. The explosion epoch is taken as the value inferred in Section 3.3. The best-fit model has a WD mass of 1.0 ${\rm M}_{\odot}$ and a He shell of 0.04 ${\rm M}_{\odot}$. Comparisons between this model and the observed $gr$-band light curves, the $g-r$ colour, and the spectra are shown in Figure 10.

The early $gr$ peaks of SN 2022vqz appear to be well reproduced, and the relative strengths of the early peak to the main peak is also satisfactory. In particular, the “red bump” in $g-r$ colour $\sim 4$ days after the explosion is overpredicted by the model. The general trend of the subsequent light-curve evolution is also reproduced, with minor differences. The model light curves are faster, bluer at early epochs and redder at later times, and overluminous by a factor of $\sim 2$ (0.8 mag). All of this makes the model light curves more consistent with those of fast decliners among normal SNe Ia rather than 02es-like events.

Most spectral features can be reproduced by the double-detonation model, but the predicted velocities are higher than the observed ones by 3000–5000 km s^-1. This is not unexpected given that SN 2022vqz is a peculiar object with an extremely low velocity, as described in Section 4.2. No carbon can be detected in the model and the observed spectra. O I $\lambda$7774 is underpredicted, especially in the $+14.6$ day spectrum. Moreover, the model predicts a strong high-velocity Ca II feature, but this is not observed in the $+14.6$ day spectrum.

Although the double-detonation model is capable of reproducing the early-time peak, all of the other discrepancies, especially those in spectral velocities, are difficult to reconcile. Reducing the WD mass to 0.9 ${\rm M}_{\odot}$ yields lower luminosity and ejecta velocities, providing a better match with the observations. On the other hand, changing the above parameters would give a worse match with the overall light-curve and colour-curve evolution, and the velocities of Si II and Ca II features would still be unreasonably high.

It is commonly accepted that SNe Ia are explosions of C-O WDs. Assuming the WD is composed of equal parts of C and O, and is fully burnt into iron-peak elements, the released specific energy is $1.55\times 10^{51}$ erg ${\rm M}_{\odot}^{-1}$ (Branch, 1992). This energy will be turned into kinetic energy of the ejecta, after compensating the binding energy of the WD. Since heavier WDs are more compact and have a larger specific binding energy, the thermonuclear explosion of a sub-$M_{\rm Ch}$ WD is expected to result in an SN with higher ejecta velocity than that of a $M_{\rm Ch}$ WD, assuming the same progenitor abundance and burning efficiency. This trend is beneficial in explaining high-velocity SNe like SN 2019ein (Xi et al., 2022) and SN 2019yvq (Burke et al., 2021), but causes problems for the low-velocity SN 2022vqz.

Reducing the burning efficiency will result in lower energy yields and slower ejecta velocities, leaving signatures of unburnt elements in the spectra. Most of the subluminous 02es-like SNe show a noticeable C II $\lambda$6580 feature around maximum light, indicating low-efficiency burning of C-O WDs. However, the spectra of SN 2022vqz do not reveal any trace of carbon. Thus, we tentatively consider a possibility that the progenitor WD is made of oxygen and neon (O-Ne). By replacing ¹²C with ²⁰Ne, the specific fusion energy reduces to $1.21\times 10^{51}$ erg ${\rm M}_{\odot}^{-1}$ and hence causes a decrease in the ejecta velocity by $\sim 2000$ km s^-1, which could potentially explain the low velocity observed in SN 2022vqz.

We further use SYNAPPS (Thomas et al., 2011) to identify the potential signatures of unburnt Ne in the spectra of SN 2022vqz. Figure 11 shows the decomposition of the spectrum synthesised by SYNAPPS compared to the $t=-7.1$ day spectrum of SN 2022vqz. We noticed that the shallow absorption features around 6800 Å and 7100 Å could be well fit by Ne I. These features are relatively common in subluminous SNe (see Figure 8). Since burning of a C-O WD may also produce Ne, this cannot be treated as an exclusive signature of an O-Ne WD progenitor. The absorption around 6800 Å might be also attributed to Ti II. However, the 6800 Å Ti II feature usually emerges at later epochs and is accompanied by another Ti II feature at $\sim 6600$ Å which is invisible in this spectrum. Thus, we still identify the 6800 Å absorption feature in the $t=-7.1$ day spectrum as Ne I rather than Ti II.

The electron-degenerate cores made of O-Ne are believed to result from carbon burning in “heavy-weight” intermediate-mass stars (i.e., $8{\rm M}_{\odot}\lesssim M\lesssim 11{\rm M}_{\odot}$), and they are expected to collapse into neutron stars through electron capture (Miyaji et al., 1980). However, O-Ne WDs can be ignited at lower central densities if a tiny amount of carbon ($\sim 1.5$%) exists in their cores (Gutiérrez et al., 2005), leading to thermonuclear SNe Ia. This tiny amount of carbon can be explained as the remnant of an off-centre carbon-burning process which does not propagate all the way to the centre. Bravo et al. (2016) and Willcox et al. (2016) examined the delayed detonation of such hybrid C-O-Ne cores. They found that the produced ejecta are characterised by slightly lower masses of ⁵⁶Ni and substantially less kinetic energy than those of normal C-O WDs, making this kind of event an interesting candidate for the subluminous class of SNe Ia.

While the C-O-Ne model can simultaneously explain the deficient of C and extremely low velocity of SN 2022vqz, it does not naturally reproduce the early excess emission, unless combined with the double-detonation model (see Section 5.1). Most of the observables can be well reproduced or explained by detonation of a C-O-Ne core triggered by a thick He shell. The companion star could either be a helium star or another degenerate He WD. Marquardt et al. (2015) studied a set of double-detonation models of $\sim 1.2$ M_⊙ O-Ne WDs. The ejecta structure and spectral morphology of O-Ne models are similar to those of the C-O models of similar WD masses, but the resultant explosions are less energetic with peak luminosity being fainter by $\sim 0.2$ mag in B and ejecta velocity slower by $\sim 3000$ km s^-1, respectively.

However, it should be noted that the minimal mass of a C-O-Ne core is $\sim 1.08$ M_⊙, which is the required mass for a C-O core to start off-centre carbon burning to Ne (Hurley et al., 2000; Marquardt et al., 2015). This value is slightly larger than the 1.0 M_⊙ inferred from model fitting in Section 5.1 and will predict a brighter peak magnitude. It is not clear whether this mass excess can be compensated by the subluminous nature of O-Ne core explosions compared with C-O cores. Viewing angle may also affect the observed luminosity since double detonation is intrinsically asymmetric. Further analysis requires a more thorough numerical examination of double detonation of C-O-Ne WDs, exploring the full mass range and viewing angle effects of such events.

The early excesses of some 02es-like events are attributed to interaction between ejecta and CSM. Kromer et al. (2016) suggest that interaction with a nonspherical CSM may be able to account for the early UV emission in iPTF14atg. Srivastav et al. (2023) also attribute the strong early-time peak observed in SN 2022ywc to CSM interaction, and rule out other scenarios like surface helium detonation, surface Ni distribution, or companion interaction.

Following the methodology of Srivastav et al. (2023), we use the CSMNI model of Modular Open Source Fitter for Transients (MOSFiT; Guillochon et al. 2018; Nicholl et al. 2017) to fit the $gr$-band photometry of SN 2022vqz. The best-fit parameters are $E_{\rm k}=0.51^{+0.15}_{-0.07}\times 10^{51}$ erg, $M_{\rm CSM}=0.0056^{+0.0042}_{-0.0015}~{}{\rm M}_{\odot}$, $M_{\rm ej}=0.87^{+0.11}_{-0.06}~{}{\rm M}_{\odot}$, $R_{0}=2.6^{+2.1}_{-1.6}\times 10^{14}$ cm, and $t_{\rm exp}={\rm MJD}~{}59844.75^{+0.08}_{-0.11}$, where $E_{\rm k}$ is the ejecta kinetic energy, $M_{\rm CSM}$ is the CSM mass, $M_{\rm ej}$ is the ejecta mass, $R_{0}$ is the inner radius of the CSM shell, and $t_{\rm exp}$ is the explosion time. Figure 12 shows the $gr$-band light curves of SN 2022vqz compared with those from the CSMNI model.

The CSMNI model reproduces the timescale and shape of the early peak very well, and it also provides a good fit to the main bulk of the light curves, with a slightly fainter predicted $g$-band peak. The estimated explosion time is $\sim 0.3$ days later than that inferred from the fireball model in Section 3.3, but within the quoted uncertainty. The ejecta mass is significantly less than $M_{\rm Ch}$, making near-$M_{\rm Ch}$ models like the delayed detonation unfavourable. This model requires a CSM distribution at $\sim 3\times 10^{14}$ cm, similar to that of SN 2022ywc, but with a mass of about one order of magnitude lower (i.e., $\sim 0.006~{}{\rm M}_{\odot}$).

The CSM could originate from interactions between two merging WDs, specifically the violent merger scenario. This scenario was favoured for multiple 02es-like events, including SNe 2002es, 2010lp, PTF10ops, and iPTF14atg. The secondary WD can be disrupted by the primary, forming a centrifugally supported disc (Yoon et al., 2007; Shen et al., 2012), or eject disc-originated material (DOM; e.g., disk winds or jets) along the polar directions (Levanon et al., 2015). Simulations of WD mergers indicate that the mass of CSM could reach $\sim 10^{-4}$–$10^{-2}~{}{\rm M}_{\odot}$ for discs formed by tidal tails (Raskin & Kasen, 2013; Dan et al., 2014), or $\sim 10^{-2}$–$10^{-1}~{}{\rm M}_{\odot}$ for disc wind-driven DOM (Levanon & Soker, 2019). The nonspherical configuration of CSM may lead to a few 02es-like events that have very bright early excesses, as SN 2022ywc, if the viewing angle aligns with the polar axis (Srivastav et al., 2023). The large parameter space and asymmetry of the violent merger scenario may account for the observed diversity in luminosity and ejecta velocity of 02es-like objects.

Some other mechanisms could also produce early excess emission in SNe Ia. One is the collision between SN ejecta and a nondegenerate companion star (Kasen, 2010; Dimitriadis et al., 2019); a bright UV/optical excess will be produced when the SN ejecta reach the surface of the companion. The separation between the SN and its companion should be comparable to the Roche-lobe radius of the companion star, $a/R_{\rm companion}=2$–3, where $a$ is the separation. For typical velocities of SN and radius of main-sequence stars, the interaction flux should appear a few minutes after the SN explosion, or a couple of hours in the case of a red-giant companion (Kasen, 2010).

Assuming the density profile of the outer SN ejecta follows a power law, $\rho\propto r^{-10}$, the isotropic equivalent collision-powered luminosity ($L_{\mathrm{c}}$) and effective temperature ($T_{\mathrm{eff}}$) can be analytically estimated as:

where $a$ is the distance between the WD and its companion, $M_{\mathrm{ej}}$ is the ejecta mass, $v_{\mathrm{ej}}$ is the ejecta velocity, $\kappa_{e}$ is the effective opacity in optical, and $t$ is the time after explosion, respectively. Assuming the early-time SED can be treated as a black-body radiation, multiband model photometry and colour evolution can be constructed from $L_{\mathrm{c}}$ and $T_{\mathrm{eff}}$.

We fit the observed bolometric light curve and $g-r$ colour evolution to the analytic model. As the early-time bolometric light curve is not sufficiently sampled, to reduce the free parameters in the fitting, the opacity $\kappa_{e}$ is set to $0.2$ cm² g^-1 according to the assumption made by Kasen (2010), whereas $M_{\mathrm{ej}}$ and $v_{\mathrm{ej}}$ are fixed to $M_{\mathrm{ch}}$ and 7000 km s^-1, respectively. Figure 13 shows the best-fit companion-collision model with separation $a=4.3$ R_⊙. While the decreasing part of the early bolometric peak is well fitted, the model curve does not show an initial increase before the early peak as observed in ATLAS $o$ and ZTF $r$ bands. Moreover, the model $g-r$ colour curve is too blue, and lacks a red bump as seen in the observed data of SN 2022vqz.

Another possible explanation for the early flux excesses is ⁵⁶Ni mixing to the outermost layers of the SN ejecta (Piro & Nakar, 2013; Sai et al., 2022). The emission from the outer ⁵⁶Ni can escape earlier than that from the deeper one, and will cause an early excess in first few days after the SN explosion. Piro & Morozova (2016) examined a series of theoretical models with different levels of Ni mixing, measured by the width of a boxcar smoothing routine on the Ni distribution curve (0.05 $\leq$ boxcar $\leq$ 0.25). Their model bolometric and colour curves are also overplotted in Figure 13 for comparison. When the level of mixing increases, the model bolometric light curve shows a more rapid increase at early times, and the colour curve becomes flatter and bluer. However, neither the early bolometric peak nor the red-bump feature of SN 2022vqz can be reproduced by the Ni-mixing models. The absence of a separate early peak is not surprising, since the outer Ni in these models is from mixing of a concentrated distribution, and the mass fraction of Ni still decreases monotonically from centre to the surface. In contrast, the outermost Ni in the double-detonation model is produced by surface He, thus ⁵⁶Ni can be more congregated in the outermost region of ejecta and produce a clear early peak.

The main strengths and weaknesses of the considered models in this paper are summarised in Table 2.