THE ⁷Be ii RESONANCE LINES IN TWO CLASSICAL NOVAE V5668 SGR AND V2944 OPH

Lithium (Li) is a key element in the study of the chemical evolution of the universe because it has likely been produced in various sites and events—the Big Bang nucleosynthesis, interactions between energetic cosmic rays and interstellar matter, evolved low-mass stars, novae, and supernova explosions. The observed Li evolutionary curve has a plateau for young Galactic ages (<2.5 Gyr) followed by a steep rise. This indicates that relatively low-mass stellar components with long lifetimes are major sources of Li in the recent universe (Romano et al. 1999, 2001; Prantzos 2012). Some low-mass evolved stars have indeed been found to have Li-enriched surfaces (e.g., Melo et al. 2005). However, there has been no observational confirmation that such stars supply Li to interstellar medium. Li contained in stellar surfaces could easily be depleted by convection because Li should be destroyed inside stars where the temperature is higher than 2.5 million K.

Novae, which evolve from low-mass binaries, are expected to be one of the candidates for Li suppliers in the recent universe because they experience explosive nucleosynthesis and mass-loss simultaneously. Cameron & Fowler (1971) made the first theoretical study of Li production in novae. They predicted that the radioactive isotope of beryllium, ⁷Be, was produced via the reaction ³He(α, γ)⁷Be during the thermonuclear runaway (TNR) in the accumulated thin gas layer on the surface of a white dwarf (WD). The ⁷Be that is produced will be blown away from the surface of a WD by the outburst wind and then decay to form ⁷Li in the cooler interstellar environment within a short period of time (a half-life of 53.22 days). In the 1990s–70s, details of this process, which is called the Cameron-Fowler process, were studied by theoretical analyses (Arnoud & Nörgaard 1975; Starrfield et al. 1978; Boffin et al. 1993; Hernanz et al. 1996; José & Hernanz 1998).

Recently we have reported the detection of ⁷Be in the post-outburst UV spectra of the classical nova V339 Del (Tajitsu et al. 2015). Because the resonance doublet lines of Be ii are located at ∼3130 Å where the telluric absorption that is mainly caused by ozone severely obstructs observations from ground-based telescopes, quantitative studies of stellar Be abundances were enabled in 1990s by the advent of 8–10 m class telescopes on high mountains. Such telescopes are equipped with high-resolution spectrographs with high sensitivities in the near-UV region (e.g., Boesgaard et al. 1999 (Keck i telescope); Primas et al. 2000 (VLT)). Extensive analyses of stellar Be abundances were conducted in large samples of metal-poor stars (Boesgaard et al. 2011) and solar analogs (Takeda et al. 2011). In the case of V339 Del we found that the absorption lines of Be ii were purely originating from ⁷Be instead of the stable ⁹Be. Therefore it was the first observational confirmation of the Cameron-Fowler process in classical novae. The ⁷Be was found in highly blueshifted (radial velocities, ${v}_{{\rm{rad}}}$ ∼ 1000 km s⁻¹) absorption lines, which are assumed to correspond to nova ejecta blown off the surface of a WD by the outburst wind. Through the direct comparison of absorption strengths between ⁷Be ii and Ca ii we estimated that the ⁷Be abundance in the ejecta of V339 Del could be higher than the pre-existing theoretical prediction (José & Hernanz 1998). Izzo et al. (2015) reported the presence of the Li i λ6708 line in the blueshifted absorption line system in the early phase (7–13 days after the outburst) spectra of V1369 Cen (Nova Centauri 2013). The overabundance of Li in this nova amounts to the order of 10⁴ with respect to the solar photospheric composition.

Now it is quite interesting to know how commonly this ⁷Be (=⁷Li) production occurs among various types of classical novae and to evaluate their contribution to the Li evolution in the Galaxy. Prompted by these interests we have performed high-resolution spectroscopic observations of two Galactic novae discovered in March 2015—V5668 Sgr and V2944 Oph—during their early declining phases.

V5668 Sgr (Nova Sagittarii 2015 No. 2 = PNV J18365700−2855420) was discovered as a bright 6.0 mag (unfiltered) source by John Seach on 2015 March 15.634 UT and announced in the American Association of Variable Star Observers (AAVSO) Alert Notice (Waagen 2015a). The nova recorded its optical maximum (V = 4.3) on March 21.67 UT (MJD = 57102.67). Its optical magnitudes stayed close to the maximum ($V\sim 4.5$–6.5) for about 80 days then showed a rapid decline by dust formation (Banerjee et al. 2015a, 2015b; Gehrz et al. 2015). The possible candidates of the progenitor were found in the USNO-B1.0 catalog (Monet et al. 2003) within 6'' from the nova (USNO-B1.0 0610–078494.3, B1 = 17.17, R1 = 16.82, R2 = 16.42, and I = 15.55 mag; USNO-B1.0 0610–0784923, R1 = 14.36, B2 = 13.62, R2 = 13.65, and I = 15.08 mag) and in the Galaxy Evolution Explorer (GALEX; Martin et al. 2005) catalog of NUV (1171–2831 Å) sources 36 from the nova (GALEX J183656.8–285539, NUV = 17.323 mag) (Waagen 2015a). From a spectroscopic observation obtained immediately after the discovery, Williams et al. (2015) reported that it is an Fe ii type nova according to the classification introduced by Williams (1992).

V2944 Oph (Nova Ophiuchi 2015 = PNV J17291350−1846120) was discovered as a 12.2 mag (unfiltered) source by Yukio Sakurai on 2015 March 29.766 UT (Waagen 2015b). After its discovery the nova showed a gradual rise to its optical maximum at V = 9.0 on April 14.75 UT (MJD = 57126.75). Then it showed a quick decline to $V\quad \sim$ 11.5 within the following few days and stayed at $V\sim 12$ for >80 days. In the USNO-B1.0 catalog a possible candidate for the progenitor was found as a ∼18 mag star at the position of the nova (Waagen 2015b). Its spectrum during the pre-maximum phase showed He/N type characteristics (Danilet et al. 2015). However, Munari et al. (2015) concluded that it is an Fe ii type nova by spectroscopic observations carried out during the maximum and early decline phase. Munari & Walter (2016) presented observations of the time evolution in line profiles during the pre-maximum phase and pointed out the presence of pre-existing circumstellar materials around the nova. They concluded that the companion in the nova system is not a main-sequence dwarf as in most novae but rather an evolved subgiant.

In this paper we report the detection of strong ⁷Be absorption lines in these two novae. In Section 2 we describe our high-resolution spectroscopic observations and data analysis. The results from our observations are described in Section 3. The ⁷Be abundances estimated from our observed data are presented in Section 4. A brief discussion and conclusions are given in Section 5.

The post-outburst spectra of the two classical novae were obtained using the High Dispersion Spectrograph (HDS) (Noguchi et al. 2002) of the 8.2 m Subaru Telescope on 2015 May 29 (V5668 Sgr; +69 days after the optical maximum) and July 3 (V2944 Oph; +80 days), respectively. Figure 1 shows the AAVSO light curves of two novae with the epochs of our HDS observations. In the case of V5668 Sgr our observation was just before the start of the rapid decline in optical magnitudes by dust formation. For each nova we obtained spectra under three configurations of the spectrograph, which cover the wavelength regions from 3030 to 4630 Å, from 4110 to 6860 Å, and from 6670 to 9360 Å. Spectral resolving power was set to R ≃ 60,000 and 45,000 with 06 (0.3 mm) and 08 (0.4 mm) slit widths, respectively. The details of observations are reported in Table 1. Data reduction was carried out using the IRAF software in a standard manner. The nonlinearity in the detectors are corrected by the method described in Tajitsu et al. (2010). The typical residual of wavelength calibration performed using Th–Ar comparison spectrum is smaller than 10⁻³ Å for each spectrograph configuration. Spectrophotometric calibration was performed using the spectrum of σ Sgr (V = 2.058, B2V) and was obtained nearly at the same altitude of the nova on the same nights. All spectra were converted to the heliocentric scale. Correction for interstellar extinction has not been applied. The line identifications were carried out using the atomic line database of Kramida et al. (2014). We also used the line list of Kurucz & Bell (1995) for some weak transitions of Fe-peak elements.

Zoom In Zoom Out Reset image size

The post-outburst spectra of V5668 Sgr (+69 days) and V2944 Oph (+80 days) displayed in Figure 2 and Figure 3, respectively, show very similar spectral characteristics. They contain a series of strong broad emission lines originating from H i Balmer series and some other permitted transitions originating from Fe ii. As in the case of V339 Del described in Tajitsu et al. (2015), most of these broad emission lines are accompanied by blueshifted multiple absorption lines whose radial velocities correspond to ∼−2000 to −800 km s⁻¹. Many blueshifted absorption lines of Ti ii, Cr ii, and Fe ii are found in the near-UV region.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 displays the enlarged views of the spectrum of V5668 Sgr in the vicinities of the Hγ, Ca ii K, ⁷Be ii λ3131.228, Na i D₂, Fe ii (the multiplet number 42; Moore 1959) λ5169.03, and Li i lines on the velocity scale. The blueshifted absorption lines of the Hγ line is clearly found to be divided into two components: the low-velocity component (LVC) is a single sharp absorption line at a relatively low ${v}_{{\rm{rad}}}$ $\sim \quad -786$ km s⁻¹ and the high-velocity component (HVC) contains a series of subcomponents spreading between $-2200\quad \lt$ ${v}_{{\rm{rad}}}$ $\lt \quad -1350$ km s⁻¹. The radial velocities of each subcomponent are measured at the absorption dips in the Fe ii (42) and Hγ line and indicated with vertical lines in the figure. The sharp absorption of the LVC can be easily identified in all lines in the figure. These LVC lines originating from singly ionized Fe-peak elements (Fe ii, Ti ii, Cr ii, Mn ii, and Ni ii) dominate the spectrum in the UV range as in the case of V339 Del from +38 to +48 days. All of the identified transitions originate from levels of low excitation potentials (the lower energy level of the transition, ${E}_{{\rm{lower}}}\quad \lesssim$ 3.1 eV). The absorption strengths of the HVC vary among different transitions. All of six subcomponents in the HVC can be identified in the Hγ and Fe ii (42) lines, although some of them are weaker or missing in Ca ii and Na i lines. In the UV range only the subcomponents at ${v}_{{\rm{rad}}}$ = −1385 and −1444 km s⁻¹ of the HVC can be identified in the strong lines originating from Fe-peak elements.

Zoom In Zoom Out Reset image size

Figure 5 presents the spectrum of V2944 Oph displayed in the same way as in Figure 4. The spectrum of this nova shows quite similar characteristics to that of V5668 Sgr. It also shows the blueshifted absorption systems divided into two velocity components: the LVC at ${v}_{{\rm{rad}}}$ = −878 km s⁻¹ and the HVC between $-2000\quad \lt$ ${v}_{{\rm{rad}}}$ $\lt \quad -1300$ km s⁻¹. At least six subcomponents can be identified in the HVC as shown by vertical lines in the figure.

Zoom In Zoom Out Reset image size

As shown in the bottom panels of Figures 4 and 5, we cannot find any counterparts of blueshifted absorption line systems of the ⁷Li i λ6708 in spite of the high signal-to-noise ratios (S/N) of the spectra (∼700 and ∼160 at 6690 Å in V5668 Sgr and V2944 Oph, respectively). There is also no counterpart of the Ca i (32) λ6718 in the spectra of both novae, although this line has been detected together with the Li i λ6708 line in the first three weeks spectra of V1369 Cen (Izzo et al. 2015). The Ca i λ4227 and K i λ7699 lines do not have any counterparts in the LVC and the HVC in both observed novae. These observations imply that the ejected gas had been heated into a state so hot that almost all Li and Ca atoms are ionized in the circumstance during the epochs of our HDS observations.

Figure 6 displays the enlarged radial velocity profiles of Fe ii, Hγ, and the ⁷Be ii "doublet" lines in both novae. Although the ⁷Be ii $\lambda \lambda$ 3130.583 and 3131.228 doublet lines coalesce, the dips of each line can be partially identified. In V5668 Sgr, two absorption dips (A and B) are clearly found in all lines. In V2944 Oph, two dips (B and C) of the ⁷Be ii doublet coincide with those of Hγ. The dip A of ⁷Be ii λ3131.228 (the weaker component of the doublet) in the LVC in this nova is unclear due to the poor S/N, which is partly disturbed by cosmic ray hits on the spectrum. A subcomponent in the LVC at ${v}_{{\rm{rad}}}=-849$ km s⁻¹ (the dip A+) is found in the Fe ii and both ⁷Be ii lines. We can find no alternative candidates of identification for these dips from line lists of Fe-peak elements. Because no dips corresponding to the ⁹Be ii doublet [the isotopic shift; ${\rm{\Delta }}\lambda =-0.161$ Å (Yan et al. 2008)] can be found, we conclude that the absorption systems at ∼3110–3125 Å in both novae purely originate from ⁷Be ii as in the case of V339 Del. The dips of the ⁷Be ii doublet become hard to be identified toward the flat bottoms of the HVC in both novae. This indicates that the absorption of ⁷Be ii is saturated while the intensities at the flat bottoms remain ∼25 and ∼15% of the continua in V5668 Sgr and V2944 Oph, respectively. In V5668 Sgr the HVCs of the Balmer lines show wavy bottom shapes. This difference indicates that the ⁷Be ii absorption is stronger than the Balmer lines at this velocity range.

Zoom In Zoom Out Reset image size

It is interesting to see the similarity in the velocity profiles of the blueshifted absorption systems in both novae. Some classical novae also have similar blueshifted absorption systems with two distinct velocity components—the narrow LVC and the broad HVC (e.g., Nova LMC 2005, V2574 Oph in Williams et al. 2008), which are known as the "principal" and the "diffuse enhanced" systems in post-maximum nova spectra (McLaughlin 1960; Munari 2014). V1369 Cen at +13 days also has two velocity components at ${v}_{{\rm{rad}}}$ $=\quad -550$ and $-1300\quad \lt$ ${v}_{{\rm{rad}}}$ $\lt \quad -900$ (Izzo et al. 2015). One remarkable point found in this study is that ⁷Be is detected in both the LVC and the HVC. This means that both of the absorption systems consist of nova ejecta which have experienced TNR. The characteristics of the LVCs in V5668 Sgr and V2944 Oph closely coincide with those of the Transient Heavy Element Absorption (THEA) found by Williams et al. (2008) in post-outburst spectra of classical novae. They assumed that these THEAs are produced by pre-outburst ejecta coming from the secondary star. Considering our results on the ⁷Be detection, however, it is natural to assume that the THEAs are produced by nova ejecta that have experienced TNR.

The ⁷Be ii absorption may be contaminated by LVCs originating from other Fe-peak elements. We set requirements for major contaminants as ${E}_{{\rm{lower}}}\lt 3.1\;{\rm{eV}}$ and $\mathrm{log}({gf})\gt -1.0$ and select candidates for contaminants as listed in Table 2. Only some Cr ii (5) lines are found to be the main contaminants to the LVC and the HVC of the ⁷Be ii doublet. In V5668 Sgr, the dip of the Cr ii (5)λ3132.053 is clearly identified within the ⁷Be ii LVC (Figure 6 upper right panel). In the ⁷Be ii HVC, no Cr ii (5) lines can be identified because of the strong saturation effect. However, their equivalent widths (W) can be estimated using measured equivalent widths of nearby unblended Cr ii lines belonging to the same multiplets. We estimate that the combined contributions of these contaminants are less than 5% of the total equivalent widths of the ⁷Be ii absorption in both the LVC and the HVC in V5668 Sgr. In the case of V2944 Oph we find that only the Cr ii (5)λ3132.053 is present besides the LVC of the ⁷Be ii doublet. We estimate that the combined contribution of contaminants in the HVC in this nova is similar to or less than that in V5668 Sgr.

Table 2.  Lines Originating from Fe-peak Elements in the Vicinity of the ⁷Be ii Doublet

			V5668 Sgr			V2944 Oph
Lines	${\lambda }_{{\rm{lab}}}$	$\mathrm{log}({gf})$	${\lambda }_{{\rm{LVC}}}$ a	7Be ii b	W (Å)c	${\lambda }_{{\rm{LVC}}}$ a	7Be ii b	W (Å)c
Ti ii (67)	3106.246	−0.170	3098.11	⋯	<0.003	3097.16	⋯	<0.006
Ti ii (67)	3117.676	−0.500	(3109.51)	HVC	⋯	(3109.51)	HVC	⋯
Cr ii (5)	3118.646	0.000	(3110.48)	HVC	(0.054 ± 0.026)	(3109.52)	HVC	(0.037 ± 0.020)
Ti ii (67)	3119.825	−0.460	(3111.65)	HVC	⋯	(3110.69)	HVC	⋯
Cr ii (5)	3120.359	$+0.120$	(3112.18)	HVC	(0.071 ± 0.034)	(3111.23)	HVC	(0.049 ± 0.030)
Cr ii (5)	3120.497	−0.018	(3112.32)	HVC	(0.052 ± 0.025)	(3111.36)	HVC	(0.036 ± 0.020)
Cr ii (5)	3128.692	−0.320	3120.49	⋯	0.013 ± 0.003	3120.49	⋯	<0.006
Cr ii (5)	3132.053	$+0.079$	3123.85	LVC	0.067 ± 0.015	3123.85	⋯	0.045 ± 0.015
Cr ii (5)	3136.681	−0.250	3128.46	⋯	0.045 ± 0.015	3128.46	⋯	<0.006

Notes.

^aObserved and expected wavelengths of the LVCs (${v}_{{\rm{rad}}}$ $=\quad -786$ and −878 km s⁻¹ in V5668 Sgr and V2944 Oph, respectively) of each transition. The undetected lines in the ⁷Be ii HVC are listed in brackets. ^bVelocity component of the ⁷Be ii doublet expected to be contaminated by the line. ^cObserved or expected equivalent widths of each line. The values estimated using nearby same multiplets are listed in brackets.

Download table as:  ASCIITypeset image

We placed "best effort" local continua around each line ($-3000\lt {v}_{{\rm{rad}}}\lt 1000$ km s⁻¹) on the complex, undulating spectra of two novae by the fitting with a high-order (15–25) spline function to measure the equivalent widths of the absorption lines. Although the ⁷Be ii doublet cannot be resolved, its total equivalent widths, W(⁷Be ii)s, in the LVC and the HVC are measured. After subtracting the combined effects of contaminations discussed above we tabulate the resulting W(⁷Be ii) in Table 3. The errors of W are estimated from the S/N of the neighboring continuum. They do not include the uncertainties in the continuum placements.

Table 3.  Equivalent Widths of Blue-shifted Absorption Lines in Observed Novae

		HVC	LVC
Object	Day	$\begin{array}{c}W{(}^{7}\mathrm{Be}\;{\rm{ii}})\\ ({\rm{\mathring{\rm A} }})\end{array}$	$\begin{array}{c}W{(}^{7}\mathrm{Be}\;{\rm{ii}})\\ ({\rm{\mathring{\rm A} }})\end{array}$	$\begin{array}{c}W(\mathrm{Ca}\;{\rm{ii}}\;{\rm{K}})\\ (\mathring{\rm A} )\end{array}$	$\displaystyle \frac{N{(}^{7}\mathrm{Be}\;{\rm{ii}})}{N(\mathrm{Ca}\;{\rm{ii}})}$	$\displaystyle \frac{X{(}^{7}\;\mathrm{Be})}{X({\rm{Ca}})}$
V5668 Sgr	+69 days	6.40 ± 0.98	1.30 ± 0.22	0.35 ± 0.02	8.1 ± 2.0	1.4 ± 0.30
V2944 Oph	+80 days	3.64 ± 0.30	0.20 ± 0.10	0.18 ± 0.03	2.5 ± 1.5	0.44 ± 0.26
V339 Del (−1103 km s−1)	+47 days	...	0.16 ± 0.010	0.089 ± 0.003	3.9 ± 0.4	0.68 ± 0.07
V339 Del (−1268 km s−1)	+47 days	...	0.064 ± 0.015	0.023 ± 0.004	6.5 ± 2.5	1.1 ± 0.44

Download table as:  ASCIITypeset image

The abundances of ⁷Be in nova ejecta is estimated following the same procedure as used in Tajitsu et al. (2015). We compare the equivalent widths of Ca ii K ($\mathrm{log}({gf})\quad =\quad +0.135$) and ⁷Be ii $\lambda 3130.583+\lambda 3131.228$ ($\mathrm{log}({gf})=-0.178,-0.479$, respectively) in the LVC, which show a simpler structure than the HVC. Here we need the following assumptions: (1) both the ⁷Be ii and Ca ii lines are not saturated and (2) the covering factor of the absorbing gas cloud to the background illuminating source has no wavelength dependence. We notice that the LVC feature of ⁷Be ii at −786 km s⁻¹ of V5668 Sgr might be saturated. The bottom of the ⁷Be ii lines in the LVC almost reaches to that in the HVC where its flat profile suggests that the absorption is saturated. The corresponding LVC component of the Ca ii K line in V5668 Sgr is shallow and the line appears to be unsaturated. The abundance of ⁷Be in V5668 Sgr obtained under such a condition should be taken as a lower limit.

Following Spitzer (1998, Equations (3)–(48)), the ratio of column number densities, N, can be written as

$$\begin{eqnarray}\displaystyle \frac{N{(}^{7}\mathrm{Be}\;{\rm{ii}})}{N(\mathrm{Ca}\;{\rm{ii}})} & = & \displaystyle \frac{W{(}^{7}\mathrm{Be}\;{\rm{ii}})}{\lambda {({}^{7}\mathrm{Be}{\rm{ii}})}^{2}}\displaystyle \frac{\lambda {(\mathrm{Ca}{\rm{ii}}K)}^{2}}{W(\mathrm{Ca}\;{\rm{ii}}\;K)}\\ & & \times \displaystyle \frac{{gf}(\mathrm{Ca}\;{\rm{ii}}\;K)}{{gf}{(}^{7}\mathrm{Be}\;{\rm{ii}}\lambda 3130.583)+{gf}{(}^{7}\mathrm{Be}\;{\rm{ii}}\lambda 3131.228)}\\ & = & \displaystyle \frac{W{(}^{7}\mathrm{Be}\;{\rm{ii}})}{{3131}^{2}}\times \displaystyle \frac{{3934}^{2}}{W(\mathrm{Ca}\;{\rm{ii}}\;K)}\\ & & \times \displaystyle \frac{{10}^{+0.135}}{{10}^{-0.178}+{10}^{-0.479}}.\end{eqnarray} \tag{ 1 }$$

Using data of the W(⁷Be ii) and the $W(\mathrm{Ca}\;{\rm{ii}}\;K)$ listed in Table 3 the column density ratios N(⁷Be ii)/N(Ca ii) derived in the LVC of V5668 Sgr and V2944 Oph are 8.1 ± 2.0 and 2.5 ± 1.5, respectively. These ratios are close to that found in V339 Del at +47 days.

The ionization states of each species are keys to convert the above ratios into atomic abundances. We assume that the most of ⁷Be is in the singly ionized state because it still stays in hot ejecta. We also assume that Ca is in the singly ionized state. The absence of Ca i lines in the blueshifted absorption line systems supports this assumption. The difference in the second ionization potential (Be: 18.21 eV, Ca: 11.87 eV) may lead to different balances between the second and the third ionized ions and this may be a source of an error in determining the abundances. We tried but failed to detect absorption lines of doubly ionized ions of Fe-peak elements, which have intermediate second ionization potentials, consulting Kramida et al. (2014). Absorption lines of Fe iii are observed in optical range spectra of B-type stars (Thompson et al. 2008). This may support the assumption that all of ⁷Be and Ca contained in the LVC are in the singly ionized state. Adopting this assumption, the mass fraction of ⁷Be to the sum of all constituent mass components, $X$(⁷Be ii), can be presented as: $X$(⁷Be) ∼ N(⁷Be ii)/N(Ca ii) × 7/40 × X(Ca). We have obtained $X$(⁷Be)/$X$ (Ca) = 1.4 ± 0.3 for V5668 Sgr and 0.44 ± 0.26 for V2944 Oph. These $X$(⁷Be)/$X$(Ca) values correspond to logarithmic overabundances of Li by +6.2 dex and +5.7 dex with respect to the solar photospheric abundance (Asplund et al. 2009). As in the case of V339 Del, the Li abundances in ejecta of both nova explosions are comparable to or even higher than those of Ca. Because the results presented here are based on the observations ∼75–100 days after their outbursts, the amount of ⁷Be freshly produced in the TNR could be three to four times larger than our measurements.

We note that these abundances are obtained only using a limited fraction of the nova ejecta giving rise to the LVC absorption. They do not necessarily represent the abundances in the whole materials erupted from these novae. Furthermore, when the ⁷Be ii lines are saturated the above $X$(⁷Be)/$X$(Ca) values should be treated carefully. Therefore, the derived ⁷Be abundances in this study might involve large uncertainties. However, the above result implies that classical novae are indeed playing an important role in the process of the Galactic Li enrichment.

To produce ⁷Be via the reaction ³He(α, γ)⁷Be the total number of ³He atoms in the accreted gas before TNR should be larger than or equal to that of freshly produced ⁷Be atoms in the nova ejecta. This means that the number ratios, $N$(³He)/$N$(Ca), in the accreted gases of V5668 Sgr and V2944 Oph are $\gtrsim 8.1\pm 2.0$ and $\gtrsim 2.5\pm 1.5$, respectively. If we adopt the solar abundance ratio between He and Ca ($\mathrm{log}(N({\rm{He}})/N({\rm{Ca}}))\ =+\ 4.59;\;$Asplund et al. 2009), the lower limits of the number ratios between ³He and ⁴He, $N$(³He)/$N$(⁴He), in the accreted gases of V5668 Sgr and V2944 Oph should amount to 0.020 ± 0.005 and 0.006 ± 0.003%, respectively. These isotopic ratios are close to that found in the solar system (=0.0166%; Asplund et al. 2009). Hernanz et al. (1996) pointed out that ³He destruction via the reaction ³He(³He, 2p)⁴He occurs concurrently with ³He(α, γ)⁷Be during TNR. They noted that the ³He destruction via ³He(³He, 2p)⁴He is less pronounced in CO novae than in ONe novae. Because both V5668 Sgr and V2944 Oph are found to be CO novae (see Section 5) we guess that the process of ³He destruction should be ineffective and that most of  the ³He in the accreted gases has been converted into ⁷Be in both novae.

The detection of ⁷Be in the post-outburst spectra of V5668 Sgr and V2944 Oph strongly suggests that the explosive production of ⁷Be is a popular phenomenon among classical novae. The light curves of both novae show slower evolution than that of V339 Del in which the rapid decline was observed in ∼+40 days. Adopting the relation between the timescaling factor of the light curve and the WD mass (e.g., Hachisu & Kato 2006, 2015), the WD masses of both novae are lower than that of V339 Del. It is natural to assume that both novae have CO WDs as in the case of V339 Del. Therefore we infer that the explosive Li production via the Cameron-Fowler process occurs frequently at least among CO novae.

It is interesting to notice that we have found LVC and HVC absorptions of the Na i D lines in V5668 Sgr and V2944 Oph while no trace of blueshifted absorptions of the Na i D lines were detected in V339 Del (Tajitsu et al. 2015). We find no absorption features corresponding to Li i λ6708, K i λ7699, and Ca i λ4227 in both V5668 Sgr and V2944 Oph as well as in V339 Del. Izzo et al. (2015) found blueshifted absorptions of Hβ and Na i D lines in V1369 Cen at two velocities (−1100 and −550 km s⁻¹) on spectra obtained during an early phase after the outburst (7–13 days). They identified Li i λ6708, K i λ7699, Ca i λ6718, and Ca i λ4227 in the lower-velocity component. These apparent spectral differences such as detection or no detection of absorptions of the Na i D or Ca i lines should reflect differences in physical conditions (most probably the ionization state) in the absorbing clouds. On the other hand, the presence of ⁷Be both in the LVC and the HVC strongly indicates that the absorbing clouds in both velocity components consist of materials that have experienced TNR.

The abundance of Li obtained by Izzo et al. (2015) from the Li i λ6708 is lower than the ⁷Be (=Li) abundances obtained in the present study. There had been no report in the literature of detection of the Li i λ6708 line among post-outburst novae before Izzo et al. (2015). Because the Li i λ6708 is located in the easily accessible red spectral region, the paucity of observational information may imply that ejecta of post-outburst novae rarely permit the physical condition in which neutral Li can survive. However, it will be interesting to carry out long-term monitoring observations of post-outburst spectra of novae including both ⁷Be and Li lines and examine the consistency of derived Li abundances. Further observations of the ⁷Be ii line in various types of classical novae will help to quantify roles of novae as the source of Li in the current universe.

This work is based on data observed at the Subaru Telescope, which is operated by the National Astronomical Observatory of Japan (NAOJ). We acknowledge with thanks the variable star observations from the AAVSO International Database contributed by observers worldwide and used in this research.

Facility: Subaru(HDS) - Subaru Telescope.