Near-infrared studies of nova V1674 Herculis: A shocking record-breaker

We present near infrared spectroscopy of Nova Herculis 2021 (V1674 Her), obtained over the first 70 days of its evolution. This fastest nova on record displays a rich emission line spectrum, including strong coronal line emission with complex structures. The hydrogen line fluxes, combined with a distance of 4.7 (+1.3 / -1.0) kpc, give an upper limit to the hydrogen ejected mass of 1.4 (+0.8 / -1.2) 10^{-3} solar masses. The coronal lines appeared at day 11.5, the earliest onset yet observed for any classical nova, before there was an obvious source of ionizing radiation. We argue that the gas cannot be photoionized, at least in the earliest phase, and must be shocked. Its temperature is estimated to be 10^{5.57 +/- 0.05} K on day 11.5. Tentative analysis indicates a solar abundance of aluminum and an underabundance of calcium, relative to silicon, with respect to solar values in the ejecta. Further, we show that the vexing problem of whether collisional or photoionization is responsible for coronal emission in classical novae can be resolved by correlating the temporal sequence in which the X-ray supersoft phase and the near-infrared coronal line emission appear.


INTRODUCTION
Classical nova (CN) systems consist of a semi-detached binary containing a white dwarf (WD) and a Roche-lobe-filling secondary star, usually a late-type dwarf. Material from the secondary spills on to the surface of the WD through the inner Lagrangian point, via an accretion disk. The material at the base of the accreted envelope becomes degenerate, triggering a thermonuclear runaway (TNR; for a review see Starrfield et al. 2020). Consequently, some 10 −5 − 10 −4 M of material, enriched in C,N,O,Al,Mg following the TNR, is ejected explosively, at several 100 to several 1000 km s −1 .
Corresponding author: C.E. Woodward chickw024@gmail.com CNe are characterized by the "speed class", the time t 2 (t 3 ) taken for the visual light curve to decline by 2 (3) magnitudes from maximum brightness (Warner 2012). There are empirical relationships between speed class and the energetics of the CN eruption: faster novae have the highest maximum bolometric luminosities, largest outburst amplitudes, and fastest ejecta. Indeed there is a relation between the absolute magnitude at maximum and the rate of the light curve decline, the "MMRD" relation (see Della Valle & Izzo 2020, and references therein). These relationships are likely a consequence of the mass of the WD on which the TNR occurred (Starrfield et al. 2020 Quimby et al. 2021). V1674 Her has been intensely observed since discovery, with optical spectroscopy at high cadence (Woodward et al. 2021c, and references therein), near infrared (NIR) spectroscopy (Woodward et al. 2021a, and references therein), and ultraviolet photometry and spectroscopy (Kuin et al. 2021). First detected in X-rays on June 14.41 UT, V1674 Her began to show supersoft (SS) X-ray emission near July 1 (Page et al. 2021). Drake et al. (2021) give a detailed discussion of its early Xray evolution. Radio observations reveal non-thermal emission (Paragi et al. 2021;Sokolovsky et al. 2021). V1674 Her is also a γ-ray source , and references therein). NICER and Chandra X-ray observations (Pei et al. 2021;Maccarone et al. 2021), as well as pre-outburst archival rband photometry Mroz et al. (2021), reveal a ∼ 501 s oscillation period that is interpreted to be the spin period of a WD in an intermediate polar system.
V1674 Her is distinctive in many ways. Classification based optical spectra obtained within days of outburst (Woodward et al. 2021c, and references therein) indicate that V1674 Her is of the FeII class (Williams 1992), with several FeII lines appearing, most prominent of which was the FeII(42) multiplet (however, the possibility of a hydrid FeIIb classification based on a FeII to He/N transition around ∼day 5.5 may be noted, Woodward et al. (2021b)). At this early epoch the peak of the Hα profile was corrugated, comprising around 9 to 10 distinct sub-peaks separated from each other in velocity by amounts ranging from 180 to 770 km s −1 , while exhibiting an overall FWHM 5850 km s −1 The OI 8446Å line, dominantly excited by Lyman-β fluorescence, was also prominent, having a FWHM of 5650 km s −1 . V1674 Her also showed coronal line emission remarkably early. The detection of coronal lines at 11 days past maximum light is possibly the earliest reported detection of such lines in a nova (see Benjamin & Dinerstein 1990).
On day 28, Wagner et al. (2021) report the detection of broad [Ne V] 3426Å, as well as [Ne IV] at 4714Å, and 4724Å arguing that the presence of these strong neon emission lines is likely attributable to overabundances of neon. If substantiated, V1674 Her would be a member of the "neon nova" class that includes QU Vul, V838 Her, and V1974 Cyg among others.
Here we discuss the exceptionally rapid near-infrared (NIR) spectral evolution of V1674 Her during the first ∼ 70 days post eruption, focusing our analyses on the very strong coronal emission which provides direct measures of gas temperatures, ejecta abundances, and insight into the TNR process in novae.

Light curve
From the AAVSO database (Kafka 2021), peak brightness was reached on MJD 59377.9605 (UTC 2021 June 12.96) at V = 6.14, in agreement with Munari et al. (2021) who find V = 6.18 on Jun 12.90 UT. We adopt Jun 12.96 UT (MJD 59377.96) as our time origin, t 0 , and V = 6.14 at maximum. Based on this t 0 , and the AAVSO light curve in Fig. 1, t 2 , t 3 and t 6 are 1 day, 2.2 days and 14 days respectively.  Table 1) of the NIR spectroscopy marked. (b): Temperature of the SS X-ray source as determined by fitting a blackbody to Swift data (Drake et al. 2021 When the t 2 and t 3 times are considered together, V1674 Her is again remarkable. It is the fastest Galactic nova known to date, of all the compiled, detailed light curves (Strope et al. 2010) surpassing V838 Her (t 2 = 1.0, t 3 = 3.0) and the recurrent nova U Sco (t 2 = 1.0, t 3 = 4.0). Optical (Woodward et al. 2021c;Wagner et al. 2021) and NIR (Woodward et al. 2021a) spectroscopy bear this out, with emission line widths and P Cygni profiles indicating ejection velocities ∼ 6000 km s −1 . Such extreme eruptions are characteristic of a TNR that occurred on a WD with a mass very close to the Chandrasekhar limit (Starrfield et al. 2020).

Reddening and distance estimate
The reddening, E B−V , is in the range 0.52-0.58 based on the 6613Å Diffuse Interstellar Band feature, clearly seen in the high-resolution (R = 220K) optical spectra (Woodward et al. 2021c). From these high-resolution spectra, the interstellar resonance K I line at 7699Å is measured to have an equivalent width of 0.1349Å, which gives a reddening E B−V = 0.52 using the calibration by Munari & Zwitter (1997). Munari et al. (2021), from the K I 7699Å line, measure E B−V = 0.55. The observed color of B-V = 0.72 and 0.55 at t 0 and t 2 respectively, when compared with the intrinsic colors of novae expected at these early evolutionary times (van den Bergh & Younger 1987), yields E B−V values in the range 0.49 to 0.57. We adopt E B−V = 0.55.
Adopting the method described in Banerjee et al. (2018, and references therein), an independent estimate of the reddening, together with a distance estimate, can simultaneously be obtained from the intersection of the extinction versus distance curve in the direction to the nova (Marshall et al. 2006), and the curve defined by where m v = 6.14 at maximum and M V = −9.05 ± 0.5 (for t 2 = 1 d derived from the MMRD relation of Della Valle & Livio 1995). The intersection of these two curves yields a distance of 4.7 +1.3 −1.0 kpc and visual extinction A V = 1.82 ± 0.33, or E B−V = 0.59±0.11, assuming that the totalto-selective extinction is 3.10. As a cross-check, applying this method to a similarly fast nova V838 Her (t 2 = 2 d), gives distance and A V estimates of 3.83 kpc and 1.40 magnitudes respectively, consistent with other compiled measurements (Harrison & Stringfellow 1994). Gaia estimates for the distance to V1674 Her (Bailer-Jones et al. 2021 have large uncertainties (the parallax measurements are at the limit of usefulness), but the central values are consistent with the latter 4.7 +1.3 −1.0 kpc estimate. The spectral energy distribution (SED) at maximum, constructed using BVRI values ) dereddened by E B−V = 0.55, yield a blackbody fit with T = 9900 K. The outburst luminosity, assuming d = 4.7 kpc, is L out = 3.7 × 10 5 L , which is ∼ 9 times the Eddington luminosity of a 1.3 M WD.

OBSERVATIONS AND DATA REDUCTION
Spectra of V1674 Her were obtained with SpeX (Rayner et al. 2003) on the 3.2 m NASA IRTF. These data were reduced with corrections for telluric absorption(s) using the SpexTool pipeline (Cushing et al. 2004). The spectra were flux calibrated using an A0 standard star observed at a comparable airmass ∆ AM < ∼ 0.14. The IRTF flux calibrations are accurate to ±10%. The observational log is presented in Table 1 and the spectra are shown in Fig. 2. The times of the observations in relation to the visual light curve are shown in the top panel of Fig. 1.
A single 0.82-2.52 µm spectrum of V1674 Her was obtained on UT 2021 August 10 at the Frederick C. Gillett Gemini North 8 m Telescope, with the facility spectrometer GNIRS (Elias et al. 2006) for program GN-2021B-FT-  (Table 1). The spectral points are clipped with a SNR threshold of > ∼ 2.5. Coronal lines (see Table 2), and representative He and O lines are identified. The small ticks in the bottom of each panel indicate the wavelength of various hydrogen emission line series (Brackett, Paschen, Pfund). (a) 0.7 to 2.5 µm range; data for day 58.39 and 68.32 have been divided by factors of 2 and 8 respectively, for clarity. (b) 2.7 to 4.3 µm range (SpeX LXD mode thermal wavelengths). The spectrum on day 68.32 in panel (a) and spectra on days 24.39 and 50.42 in panel (b) have been smoothed using a Savitzky-Golay filter using a SG window=7.5 and degree=2 (for a description see Cushing et al. 2004). (The observed spectra are available as data behind the figures). Table 1). Data reduction utilized IRAF 1 and Figaro commands and produced the spectrum shown in Fig. 2. The GNIRS observations were obtained in poor seeing and through thin but variable clouds; consequently the flux calibration is uncertain by ∼ 30%.

RESULTS AND DISCUSSION
The spectra are shown in Fig. 2, the two panels covering the wavelength ranges 0.7-2.5 µm and 1.8-4.1 µm, corresponding to the SpeX Short-Cross Dispersed (SXD) and Long-Cross Dispersed (LXD) wavelength ranges, respectively. Here, the discussion focuses mainly on the coronal lines; a more detailed discussion of the SEDs is deferred to a later paper.

Ejecta mass estimate
Using the dereddened fluxes of Pa-β and Br-γ, we ascertain that their relative strengths follow Case B predictions on day 11.51, within the parameter space expected to represent the ejecta, viz. T = 5000 K to 20000 K, and electron density n e ∼ 10 6 to 10 9 cm −3 . Other H lines could not be used because of blending with neighboring lines, leading to inaccurate line flux estimates. We assume the ejecta density on day 11.51 to be n e = 10 6.6 cm −3 since the [Al VI] 3.66 µm line, which has a critical density of n e = 10 6.6 cm −3 at T = 10000 K (Greenhouse et al. 1993), is detected on this day.
The ejecta mass is given by where f is the dereddened flux in an H line in W m −2 , is the emissivity in the line in erg s −1 cm 3 (Storey & Hummer 1995), and n e is the electron density per cm 3 . Using the Pa-β and Br-γ lines, and assuming d = 4.7 +1.3 −1.0 kpc and a pure H composition, the ejecta mass is M ej = 1.4 +0.8 −1.2 × 10 −3 ×(10 6.6 /n e ) M , wherein the upper limit on the mass is obtained by choosing n e = 10 6.6 cm −3 .

The coronal lines
The profiles of four nebular lines (two H I and two He I) on day 5.64 are shown in Fig. 3a to illustrate the complexities of the line profiles. The similarities of the details in each of these profiles are remarkable. In particular, there are identifiable features at radial velocities -2480 km s −1 , -1820 km s −1 , -60 km s −1 , 730 km s −1 , 1140 km s −1 , 1530 km s −1 and 2310 km s −1 , each with an uncertainty of ±100 km s −1 . The velocity structure seen at an epoch (day 50.43) when the coronal lines are strong, Fig. 3b, is similar.
The coronal line fluxes are listed in Table 2. Uncertainties in the flux determinations are estimated to be < ∼ ± 20% for the stronger lines, and as much as ±40% for the weaker. The table also lists the ionization potential (IP) of the lower ion (thus the entry for [S IX] gives the IP of S VIII). The parameter kT * denotes the temperature T * at which half the photons emitted by the corresponding blackbody can ionize the lower ion; we take this to imply the temperature of the ionizing source that is capable of producing the requisite ion.
The time-dependence of the silicon coronal lines, among the strongest and best recorded coronal lines in the spectrum, is shown in Fig. 1 (Page et al. 2021;Drake et al. 2021). In this respect V1674 Her is again a record breaker. Both V1500 Cyg and V838 Her, two of the fastest novae with t 2 times of 2 and 1 day(s) respectively, were first detected to show coronal emission at 29 and 17 days, respectively, past maximum light (Chandrasekhar et al. 1993;Benjamin & Dinerstein 1990).
The presence of coronal lines so early in the evolution implies that, despite the presence of a hot central WD remnant, the initial coronal emission could not have been caused by photoionization. For example, [Si VII] 2.48 µm is present on day 11.51 even though the ionization of Si VI requires 205 eV, far exceeding the temperature of the ionizing source (kT * = 86 eV) that would emit sufficient photons, and is well in excess of the source temperature as measured by The Neil Gehrels Swift Observatory (see Fig. 1).
We therefore conclude that the coronal emission arose from collisional ionization by particles that gained sufficient energy from shock heating. This contradicts the commonly assumed view that, when the central source is sufficiently hot, photoionization is the principal mechanism for coronal line generation (see Benjamin & Dinerstein 1990). Plausible support for shocks comes from the multi-peaked line profiles shown in Fig. 3, which were even more clearly resolved into a large number of sub-structures in very high resolution  Greenhouse et al. (1990). The effective collision strengths (Ψ), are the collisional strengths Ω averaged over a thermal electron distribution and are taken from the IRON Project (Badnell et al. 2006;Hummer et al. 1993) on-line database (http://cdsweb.u-strasbg.fr/tipbase/home.html) interpolated to a temperature of 10 5.57±0.05 .
(R = 220, 000) spectra (Woodward et al. 2021c). This points to parcels of gas in the ejecta having widely different velocities. Collisions between such parcels would not only lead to shock heating, but might also account for the gamma ray emission seen in the early stages of V1674 Her .
To make estimates, we suppose that the ejecta initially are fully ionized and have a temperature ∼ 10 4 K, typical for CNe. For a gas of solar composition, the root-mean-square ion speed is V rms ∼ 20 km s −1 . Fig. 3 suggests that there must exist in the ejecta parcels of gas with relative speed V rel ∼ 1000 km s −1 . Thus V rel /V rms ∼ 50, implying that the gas is shocked, and that the shock is strong.
Under these circumstances, the temperature of the shocked gas could be as high as which is ∼ 14 × 10 6 K, or kT shock ∼ 1200 eV. This is more than sufficient to account for the early presence of coronal lines. However, photoionization may still have a role in the coronal line emission. As the SS phase commences, an additional hard ionizing source of radiation becomes available (in addition to shock heating). The impact of this source is con-siderable, especially in the case of V1674 Her since it has been shown to have the hottest galactic SS source of all well documented SS sources (see Page & Osborne 2014). As the bottom panel of Fig. 1 shows, the degree of ionization for the Si ion slowly increases with time as the SS source becomes hotter. The strength of the [Si X] 1.43 µm line (IP = 351 eV), the strongest coronal line in the spectrum, likely reflects the extreme high temperature of the central WD surface. The overall picture that emerges is that both collisional and photoionization are operational, but the key result from this work is that collisional ionization preceded photoionization, since coronal lines were seen about a week before the SS phase was detected. Thus, a problem that is hard to resolve, whether collisional or photoionization is responsible for coronal emission in CNe, can be resolved as has been done here by correlating the temporal sequence in which the X-ray SS phase and the NIR coronal emission appear.

Temperature of the coronal gas
The coronal lines were first detected on day 11.51, and likely the coronal emission was entirely the result of collisions. We can, therefore, estimate the gas temperature using the fluxes of the various silicon lines at this time. We estimate the relative proportions of Si ions as a function of tempera-ture using the prescription of Greenhouse et al. (1990), the ionization equilibrium data in Arnaud & Rothenglug (1985) and the effective collisional strength values Ψ listed in Table 2. We find the gas temperature T gas 10 5.57±0.05 K, (kT gas ∼ 27 keV). This value is consistent with the upper limit on the flux in the [Si X] 1.43 µm line, which requires that T gas be less than 10 5.7 K. This value of T gas is similar to that found in the shock-excited gas around the recurrent nova RS Oph, in which the ejected material runs into, and shocks, the secondary wind (Evans et al. 2007;Banerjee et al. 2009). We assume T gas 10 5.5 K in what follows. Greenhouse et al. (1990Greenhouse et al. ( , 1993) describe a method for estimating the abundance ratios for selected species. For the present we use dereddened fluxes from Table 2 for lines where there are no blends. Using the [Al V] 2.90 µm line, we find that n(Al)/n(Si) = 0.17 using the [Si VII] line, and = 0.14 derived from the [Si VI] line. A value of n(Al)/n(Si) = 4.5 × 10 −2 is derived from the [Al VI] and [Si VI] lines. Hence the mean n(Al)/n(Si) is 0.11. Using the [Ca VIII] line, we get a more consistent result, viz., n(Ca)/n(Si) = 2.6 × 10 −3 from the [Si VI] line, and = 2.2 × 10 −3 from the [Si VII] line. The latter values represent lower limits as collision strengths were not explicitly computed at the assumed temperature for the [Ca VIII] line. The solar values are 0.081 (Al/Si) and 0.062 (Ca/Si) respectively (Asplund et al. 2021). We tentatively conclude that aluminum is near solar abundance, and calcium is under abundant relative to silicon, with respect to solar values. A more detailed abundance determination awaits future modeling efforts.

Abundance estimate
Synoptic multi-wavelength studies of this remarkable nova continue as the system returns to quiescence.