Highly enriched ⁷Be in the ejecta of Nova Sagittarii 2015 No. 2 (V5668 Sgr) and the Galactic ⁷Li origin

Abstract

We report on the evidence of highly blueshifted resonance lines of the singly ionized isotope of ⁷Be ii in high resolution UVES spectra of Nova Sagittarii 2015 No. 2 (V5668 Sgr). The resonance doublet lines ⁷Be ii at λλ313.0583, 313.1228 nm are clearly detected in several non-saturated and partially resolved high velocity components during the evolution of the outburst. The total absorption identified with Be ii has an equivalent width much larger than all other elements and comparable to hydrogen. We estimate an atomic fraction N(⁷Be)/N(Ca) ≈ 53–69 from unsaturated and resolved absorption components. The detection of ⁷Be in several high velocity components shows that ⁷Be has been freshly created in a thermonuclear runaway via the reaction ³He(α, γ)⁷Be during the Nova explosion, as postulated by Arnould & Norgaar, however in much larger amounts than predicted by current models. ⁷Be ii decays to ⁷Li ii with a half-life of 53.22 d, comparable to the temporal span covered by the observations. The non-detection of ⁷Li i requires that ⁷Li remains ionized throughout our observations. The massive Be ii ejecta result into a ⁷Li production that is ≈ 4.7–4.9 dex above the meteoritic abundance. If such a high production is common even in a small fraction (≈5 per cent) of Novae, they can make all the stellar⁷Li of the Milky Way.

nuclear reactions, nucleosynthesis, abundances, stars: individual: V5668 Sgr, novae, cataclysmic variables, Galaxy: evolution

INTRODUCTION

⁷Li is a unique element that shows a large variety of production processes. These include primordial nucleosynthesis, spallation processes by high energy cosmic rays in the interstellar medium, stellar flares in low mass stars, Cameron–Fowler mechanism in Asymptotic Giant Branch (AGB) stars and Novae, and neutrino induced nucleosynthesis in SNe explosions. Observations show that ⁷Li has a constant abundance among metal-poor stars and begins to rise at [Fe/H] ≈−1 to reach the meteoritic value at solar metallicities (Rebolo, Beckman, & Molaro 1988) requiring a net ⁷Li production (Romano et al. 1999). The rate of the Li increase favours AGB stars and Novae as the most significant |$\it stellar$| sources. Although ⁷Li has been observed in AGB stars the observational evidence for Novae has only recently been found by Izzo et al. (2015) with the first detection of the ⁷Li iλλ6708 line in the spectra of Nova Centauri 2013 (V1369 Cen) and by Tajitsu et al. (2015) with the first detection of ⁷Be in the post-outburst spectra of the classical Nova Delphini 2013 (V339 Del).

Here, we report a study of the Be ii by means of UVES observations of Nova Sagitarii 2015 No. 2 (V5668 Sgr). A spectrum from the High Dispersion Spectrograph of the Subaru Telescope taken at day 63 after maximum has been discussed by Tajitsu et al. (2016) who reported the presence of ⁷Be ii in this Nova, and also in V2944 Oph.

OBSERVATIONS

Evidence for ⁷Be ii

Nova Sagittarii 2015 No. 2 was discovered by Seach (2015) on 2015 March 15 and reached the first maximum on March 21 at 04h 04m ut with a magnitude of V = 4.3. The Nova re-brightened several times and remained bright for about 80 d before declining due to dust formation. Soon after the discovery we started a DDT programme with the UVES spectrograph at the ESO-VLT. Several UVES spectra were obtained at +58, 63, 69, 73, 82 and 89 d from maximum as reported in Table 1. The settings with central wavelength of 346 nm (range 305–388 nm), 437 nm (375–499 nm), 564 nm (460–665 nm) and 760 nm (570–946 nm), were used, thus covering the full optical range from the atmospheric cutoff to the red edge of 946.0 nm with small gaps of ≈10 nm around the red central wavelengths. The resolving power was R = λ/δλ ≈ 80 000 for the blue arm and ≈120 000 for the red arm. Overlapping spectra have been combined for each epoch to maximize the signal to noise.

Table 1.

Journal of the observations.

Date	ut	MJD	Day	346	437	564	760
2015	h m	57000	a.m.	s	s	s	s
05-19	08 49	161.36	58.7	400	2 × 150	2 × 100	2 × 150
05-24	05 49	166.24	63.6	2 × 400	2 × 150	4 × 100	2 × 150
05-30	03 20	172.14	69.5	400	100	9 × 7	3 × 10
06-03	05 18	176.22	73.6	400	100	9 × 7	3 × 7
06-12	05 25	185.23	82.6	400	60	9 × 7	2 × 10
06-19	01 29	192.06	89.4	2 × 602	2 × 60	22 × 15	3 × 15

At early epochs the spectra of the Nova show several broad emission lines of neutral hydrogen and other permitted transitions of neutral or singly ionized species often accompanied by sharp and blueshifted multiple absorption components reaching blue edge velocities of ≈−2300 km s⁻¹. Fig. 1 displays portions of the Nova spectrum of day 58 in the proximity of H γ, Ca ii K, Fe ii λλ 519.0 and Be ii λλ313.0 nm lines. These lines show several absorption components at heliocentric v_rad∼ −730, −1175, −1350, −1580, −1780 and −2200 km s⁻¹ with the more prominent ones marked with vertical black dotted lines in the figure. Ca ii K shows also narrow absorption components at −4.3 and −62.9 km s⁻¹ velocities caused by intervening Galactic interstellar medium. At the wavelength of the Be ii λλ313.1 nm doublet there is P-Cygni profile with a huge blueshifted absorption that Tajitsu et al. (2016) identified as ⁷Be. In Fig. 2 we zoomed the component at −1175 km s⁻¹ which is seen only at this epoch and provides a robust identification. The two sharp absorption components (FWHM ≈ 0.19 Å) are separated by 0.654 Å which corresponds precisely to the separation of the ⁷Be ii resonance doublet of λλ313.0583, 313.1228 nm. Moreover, the component is perfectly aligned in radial velocity with the other species providing evidence that it is ⁷Be ii and not ⁹Be ii which has an isotopic shift of −15.4 km s⁻¹. The dips of each line of the ⁷Be ii doublet can be identified also at −730 km s⁻¹ in Fig. 1, but become hard to see in the flat bottoms of the other components. At velocities ≤ −1600 km s⁻¹ the absorption profile ascribable to Be follows mainly the hydrogen lines rather than the metallic lines which are very weak or absent. If based only on this spectrum the identification of this part of absorption with ⁷Be ii would therefore be controversial.

Figure 1.

Nova spectrum at day 58. The figure displays the spectrum in the vicinity of Fe i λλ519.6 (green line) H γ (red), Ca ii K (blue) and ⁷Be ii (black) lines plotted on the velocity scale. The fluxes are scaled to provide roughly the same intensity shortwards the absorption. The velocity scale is adjusted to the ⁷Be ii λλ313.0583 nm line. The position of the main absorption components is shown with vertical dotted lines. The corresponding position of the expected ⁷Be ii λλ313.1228 nm line is shown by vertical dashed line.

Figure 2.

Zoom of Fig. 1 around the −1175 km s⁻¹ component. The ⁷Be ii λλ313.0583, 313.1228 nm are shown with a dashed line. The positions of the expected ⁹Be ii λλ313.0442, 313.1067 nm and ¹⁰Be ii λλ313.0484, 313.1129 nm are also shown.

The bottoms of the strong lines in Fig. 1 are totally flat suggesting that the absorption is saturated but with the absorbing material only partially covering the background light source. The Balmer lines also show flat bottoms. At this epoch in correspondence of the ⁷Be ii the intensities are ∼50 per cent but the intensity value varies with the day and the geometry of the outburst.

The ⁷Be ii absorption may be contaminated by the presence of other Fe-peak elements. Evidence is found for the presence of Cr ii (5) multiplet. The Cr ii λλ 313.2056 nm line is observed resolved both in the ∼−730, ∼−1175 km s⁻¹ components. The Cr ii (5)λλ 313.6680 and 312.8699 nm and the Fe ii (82) λλ313.5360 nm lines are now seen at −730 km s⁻¹. The Cr ii λλ 312.4978 nm line of the same multiplet (5) and with comparable intensity should therefore be present and contribute to the main absorption. Other lines show up on day 82 and are listed in Table 2.

Table 2.

Contaminants from Fe-peak elements in the range of the ⁷Be ii doublet measured in the spectrum of day 82.

Lines	λlab(Å)	log gf	W (mÅ)	W (mÅ)
			−820	−1900
Cr ii (5)	3120.3691	0.120	16: ± 8	–
Cr ii (5)	3124.973	−0.018	b	b
Cr ii (5)	3128.700	−0.320	25 ± 7	b
Cr ii (5)	3132.053	0.079	b	b
Fe ii (82)	3135.360	−1.130	40 ± 7	104 ± 10
Cr ii (5)	3136.681	−0.250	45 ± 8	51 ± 15
Fe ii (82)	3144.751	−1.740	b	83 ± 10

Fig. 3 displays portions of the Nova spectrum obtained on day 63. The components at velocities ≈−1380 and −1560 km s⁻¹ become sharper and it is possible to identify the partially resolved component of the ⁷Be ii λλ313.1228 nm line shown in the figure by vertical blue dashed lines. In addition, some absorption consistent with its presence can be observed in the −760 km s⁻¹ component as well as in all other lines. It is quite remarkable that while Hγ shows some structure, ⁷Be does not, and this is likely due to the presence of the doublet lines filling the inter-component velocity-space. Fig. 4 displays the spectrum obtained on day 69 which is very close to the spectrum analysed by Tajitsu et al. (2016). As it can be seen the narrow component identified by Tajitsu et al. (2016) as ⁷Be ii λλ 313.1228 nm is blended with the −1450 km s⁻¹ component of the Cr ii λλ313.2058 nm. Fig. 5 displays portions of the spectrum obtained on day 73. At this epoch the high velocity components at ≈−2000 km s⁻¹ weaken considerably in ⁷Be ii revealing the presence of the ⁷Be ii λλ 313.1228 nm line in the components marked with a dashed blue line in the figure. To note that at this day the ⁷Be ii absorption spans wider velocities than Hγ. Fig. 6 displays portions of the spectrum obtained on day 82. At this epoch the high velocity components of ⁷Be ii weaken considerably revealing few resolved components previously hidden inside the absorption. The ⁷Be ii λλ313.1228 nm line is partially resolved in the two components which break up at velocities ≈ −820 km s⁻¹ and is fully resolved in the components at velocities between ≈−1600 and −1725 km s⁻¹ which were previously strongly saturated. After few days from this observation all the metallic components disappeared and therefore later observations are not considered here. Due to the weakening of the ⁷Be ii on day 82 also the metallic contaminants due to iron-peak elements appear very clearly. The Cr ii (5) λλ313.6680 nm and the Fe ii (82) λλ313.5360 and 314.4751 nm lines are now seen at −1900 km s⁻¹ and should have been present also in the previous epochs but completely obscured by the strong ⁷Be ii absorption. The Cr ii (5) λλ313.6680 and 312.8699 nm and the Fe ii (82) λλ313.5360 nm lines are now seen at −820 km s⁻¹. Scaling with the relative strengths of the other iron-peak elements we estimate that the combined contributions of these contaminants are ≈3.5 per cent of the total equivalent widths of the ⁷Be ii absorption.

Figure 3.

Same as Fig. 1 with the Nova spectrum of day 63. Legend as in Fig. 1. Note the presence of the partially resolved components of the ⁷Be ii λλ313.1228 nm line for the components at velocities ≈−1380 and −1560 km s⁻¹.

Figure 4.

Same as Fig. 1 with the Nova spectrum at day 69. This observation is close to the one discussed in Tajitsu et al. (2016).

Figure 5.

Same as Fig. 1 with the Nova spectrum at day 73. The ⁷Be ii 313.1228 nm line is partially resolved in the component at velocities ≈−2050 km s⁻¹.

Figure 6.

Same as Fig. 1 with the Nova spectrum at day 82. The ⁷Be ii 313.1228 nm line is partially resolved in the two components which break up at velocities ≈−820 km s⁻¹ and fully resolved in the components at velocities at ≈−1600 and −1725 km s⁻¹ which were previously strongly saturated. Note the components at −1900 km s⁻¹ of the CrII and FeII which are now visible due to the ⁷Be weakening.

⁷Be abundance

The abundance of ⁷Be can be estimated by comparing the equivalent widths |$W({^{7}Be\,\small {II}})$| and |$W({Ca\,{\small II}\,K})$| for unsaturated and resolved lines, assuming that the covering factor of the absorbing expanding shell is constant with wavelength. Ca is not a Nova product and can be taken as a reference element. The more suitable components are the features of ⁷Be ii and the Ca ii K lines observed at −1500 km s⁻¹ on day 82. Though, we are aware that these abundances do not necessarily represent the abundances in the whole materials ejected.

Following Spitzer (1998) and Tajitsu et al. (2015) the ratio of column densities, N, can be written as

\begin{eqnarray} \frac{N({^{7}Be\,{\small II}})}{N({Ca\,{\small II}})} & = & 2.164 {\times} \frac{W({^{7}Be\,{\small II}\,, Doublet})}{W({Ca\,{\small II}\,, K})} \end{eqnarray}

(1)

For the components at −1500 km s⁻¹ on day 82 we measured |$W({^{7}Be\,{\small II}})$| = (0.095 +0.060) = 0.155 Å and the |$W({Ca\,{\small II}\,K})$| = 0.019 Å which provide a column density ratio of |$N({^{7}Be\,{\small II}})/N({Ca\,{\small II}})$| = 17.7. Assuming that most of ⁷Be and Ca are in the singly ionized state as discussed in Tajitsu et al. (2016) these are also the relative elemental abundances. The presence of Na i in the blueshifted absorption line systems along with the presence of Ca i on day 58, as shown below, and the absence of doubly ionized iron-peak elements support this assumption. Since our measurement refers to day 82 after maximum and ⁷Be decays to ⁷Li via K-electron capture with a half-life of 53.22 d, the amount of ⁷Be freshly produced by the Nova should have been ≈3 times larger which gives an atomic fraction N(⁷Be)/N(Ca) of ≈53. We can determine the ⁷Be abundance also for the component at −1175 km s⁻¹ on day 58, which is fully resolved but slightly saturated. In this case we have |$W({^{7}Be\,{\small II}})$| = (0.089 +0.073) = 0.162 Å and the |$W({Ca\,{\small II}\,K})$| = 0.011 Å which provides a |$N({^{7}Be\,{\small II}})/N({Ca\,{\small II}})$| = 31.9. Considering that on day 58 the original value should have been a factor of 2.15 larger, we obtain an original atomic fraction N(⁷Be)/N(Ca) of ≈ 69, which is quite consistent with the former value considering the uncertainties involved. Tajitsu et al. (2016) derived a N(⁷Be ii)/N(Ca ii) = 8.1 ± 2.0 in the component at −786 km s⁻¹ on day 63 without considering ⁷Be ii decay. Since this component is saturated, the derived abundance is a lower limit and therefore the two measurements are consistent with each other.

Nova ⁷Be production

Thermonuclear production of ⁷Be during the Nova explosions of hydrogen-rich layers containing some ³He has been proposed by Arnould & Norgaard (1975) and Starrfield et al. (1978). Peak temperatures of 150 million K are reached in the burning regions and ⁷Be is readily formed from the ³He coming from the companion star via the reaction ³He (α, γ)⁷Be (Hernanz et al. 1996). In this hot environment ⁷Be can be also destroyed and it needs to be carried to cooler regions by convection on a time-scale shorter than the destruction time-scale as in the Cameron–Fowler mechanism (Cameron & Fowler 1971). The cooler regions are subsequently ejected and observed in absorption in the Nova outburst. Carbon and oxygen (CO) Novae destroy less ³He with respect to oxygen and neon (ONe) Novae, and therefore CO Novae have higher ⁷Be yields (José & Hernanz 1998).

The detection of ⁷Be in the post-outburst spectra of Nova Sagittarii 2015 shows that thermonuclear production of ⁷Be is effectively taking place. The fact that ⁷Be is detected at all velocities implies that all the absorption components are made of ejecta which have experienced thermonuclear runaway nucleosynthesis. However, the observed yields are larger by about one order of magnitude than predicted by the models of José & Hernanz (1998) and even more if compared with the models of Boffin et al. (1993). The number of freshly produced ⁷Be atoms in Nova ejecta is necessarily lower than that of ³He atoms in the accreted gas from the donor star or produced in situ as a result of the so called ³He bump (Denissenkov et al. 2013). This implies that the fraction of (³He/H) should be greater than 10⁻⁴.

Nova ⁷Li production

The ⁷Be decays to ⁷Li with a half-life of 53.22 d which is comparable with our temporal span. However, we do not detect counterparts of blueshifted absorption line systems of the ⁷Li i λλ670.8 nm in spite of the high signal-to-noise ratios of our spectra. Fig. 7 displays the spectrum on day 58 in the vicinity of ⁷Li i λλ670.8 nm, Na i D doublet and Ca i λλ422.7 nm lines. The other epochs are similar with the only difference that the trace Ca i disappears. We note however that while Ca i is non-present the Na i D lines are relatively strong and the complete absence of ⁷Li is rather puzzling. It is interesting to report that these lines have been detected in the first three weeks spectra of Nova Centauri 2013 (Izzo et al. 2015) and that Izzo et al. (private communication) detected ⁷Li i λλ670.8 nm in spectra of V5668 Sgr taken on day 7. The non-detection of ⁷Li in our observations implies that the ejected gas has a temperature high enough that almost all Li and Ca atoms are ionized. Normally Li is not detected in Novae outburst spectra and the unique Li i detection by Izzo et al. (2015) implies that the physical conditions in the ejecta permit the survival of neutral ⁷Li i only in the very early stages. Since ⁷Be ii decays to ⁷Li ii the non-detection of ⁷Li i in our epochs requires that ⁷Li remains singly ionized while some Na i survives.

Figure 7.

Nova spectrum at day 58 in the vicinity of  ⁷Li iλλ670.8 nm (black line), Na i D doublet (red) and Ca i λλ 422.7 nm (blue).

Since ⁷Be = ⁷Li the X(⁷Be)/X(Ca) fraction derived here corresponds to a ⁷Li logarithmic overabundances of +4.7 dex with respect to the meteoritic value (Lodders, Palme, & Gail 2009), which is even higher than the overabundance of 4 dex obtained by Izzo et al. (2015) in Nova Centauri 2013. Theoretically the amount of ⁷Li is a sensitive function of the conditions achieved in the outburst and from the initial ³He of the companion star, and are expected to vary. The Novae in which ⁷Li or ⁷Be have been detected to date are all slow Novae characterized by t₂ > 60^d. However, it is quite remarkable that large ⁷Be yields are observed in all three Novae where ⁷Be ii has been searched for. For a total ejected mass of ≈10⁻⁵M⊙ the observed overproduction factor of V5668 Sgr implies a production of M_Li ≈ 7 × 10⁻⁹M⊙. The global Nova rate in the Galaxy is known within a factor of 2 (25–50 yr⁻¹) (della Valle & Livio 1994; Shafter 2016) and the slow Novae account for ≈10 per cent of the whole population (della Valle & Duerbeck 1993). However, a rate of 2 yr⁻¹ of slow Nova events with the observed ⁷Li overproduction in a Galaxy lifetime of ≈10¹⁰ yr is enough to produce M_Li ≈ 140 M⊙. This is comparable with the M_Li ≈ 150 M⊙ estimated to be present in the Milky Way inclusive of the M_Li ≈ 40 M⊙ produced in the big bang (Fields, Molaro, & Sarkar 2014). Thus, the slow Novae could indeed be the main factories of ⁷Li in the Galaxy.

SUMMARY AND CONCLUSIONS

We have analysed UVES high resolution observations of V5668 covering six outburst phases from day 58 to day 89 from maximum. The evolution of the absorption offers clear evidence in support of the identification of ⁷Be ii by Tajitsu et al. (2016). In particular, the weakening of the ⁷Be ii absorptions at a late epoch shows that the iron-peak species are a minor contaminant. By means of unsaturated Be ii components we derived an abundance of N(⁷Be)/N(Ca) ≈ 53–69 when the ⁷Be decay is taken into account. Assuming all the ⁷Be goes into ⁷Li this corresponds to a ⁷Li overproduction of 4.7–4.9 dex over the solar-meteoritic value. We then argue that a rate of 2 yr⁻¹ of such events in a Galaxy lifetime, i.e. only a small fraction of all Novae, could be responsible for the production of the whole ⁷Li required from stellar sources.

We also notice that such a high ⁷Be production should increase the probability of detecting the 478-keV γ-ray photons emitted in the ⁷Be to ⁷Li reaction which have been so far elusive despite several γ-ray searches.

This Letter is based on observations collected at the European Souther Observatory, Chile. Programme ESO DDT 294.D-5051. This is an ESO DDT programme and we acknowledge the ESO director for this opportunity and the ESO staff for care and competence in making the observations. LI acknowledges support from the Spanish research project AYA 2014-58381-P.