NEW FEATURES OF THE SYMBIOTIC RECURRENT NOVA V407 CYGNI FOUND IN THE OUTBURST IN 2010

Figure 2 shows a tracing of our first spectrum obtained on 2010 March 14: day 3.4. The correction for the interstellar extinction is not applied to this and all successive tracings. The prominent emission lines were of H i, He i, Fe ii, [Fe ii], [O i], [N ii], and Si ii, while those of He ii were weak. The emission lines of H i, He i, and Si ii were broad, while those of Fe ii, [Fe ii], [O i], and [N ii] were narrow, e.g., the FWHMs of Hα and Hβ were 2300 and 1770 km s⁻¹ respectively, and those of [O i] 6300 and Fe ii 6456 were 146 and 155 km s⁻¹. Some Fe ii emission lines had broad emission tails as seen on Fe ii 5169 and Fe ii 5317 (Figure 2). The coexistence of narrow and broad emission components was observed also in the spectra of RS Oph on the outbursts (e.g., Iijima 2009). The emission lines of [Fe ii] were usually prominent in the spectra of classical slow novae and were not found in those of fast novae, with the single exception of CP Pup (McLaughlin & Greenstein 1960). In contrast to classical novae, the combination of fast fading of luminosity and the emission lines of [Fe ii] in the spectra is not rare among the symbiotic recurrent novae, because it was observed also in RS Oph (Iijima 2009) and V745 Sco (Williams et al. 1991).

Zoom In Zoom Out Reset image size

Spectra covering wider spectral ranges were obtained on March 15 and 16. The spectral features were effectively the same during these three days. The intensities of selected emission lines are given in Table 2.

A high-resolution spectrum covering the spectral range from 4330 to 6120 Å, with some gaps caused by the discontinuity between the orders, was obtained on March 24: day 13.4. A tracing is shown in Figure 3. The emission lines detected first in this spectrum were of [O iii] 4363, 4959, and 5007; N iii 4634 and 4641; [Fe vii] 5721 and 6086; O iv 6106; and an unidentified line at 6028 Å (see Section 5.5). The emission line of [Fe vii] 6086 was blended with [Ca v] 6086 (Shore et al. 2011), but the contribution from the latter seems to have been small, because [Fe vii] 5721 was well seen while [Ca v] 5309 was not detected during the outburst stage, and appeared in 2011 (Table 2). The emission lines of He ii 4686 and [N ii] 5755 largely strengthened relative to Hβ, while those of He i weakened (see Section 5.4).

Zoom In Zoom Out Reset image size

The narrow components of the emission lines of [O iii] and [N ii] were broader than those of [Fe ii] lines, that is the FWHM of [O iii] 5007 was 94 km s⁻¹ while the mean of the prominent [Fe ii] lines was 37.4 ± 1 km s⁻¹. As reported by Shore et al. (2011), the emission lines of [O iii] 4959, 5007, and [N ii] 5755 showed double peak profiles, but [O iii] 4363 did not. Widely spread emission tails were seen at the bases of [N ii] and Fe ii lines, namely the FWZI of [N ii] 5755 was about 2100 km s⁻¹ and that of Fe ii 5317 was 1600 km s⁻¹. The FWHMs of Hβ and He ii 4686 were 910 and 680 km s⁻¹, respectively.

Figure 4 shows the profile of Hβ in the same spectrum, where narrow absorption lines of Ti ii, Cr ii, and Y ii were seen on the broad emission component. Similar narrow absorption lines were seen also on Hγ. Their mean radial velocity was −39.8 ± 0.8 km s⁻¹, that is −24 km s⁻¹ in the frame of the LSR. This velocity agrees with that of SiO maser emission J = 1–0 v = 2 (Deguchi et al. 2011), because if we interpolate their results of −21.8 km s⁻¹ on March 22 and −26.8 km s⁻¹ on March 28, we have −23.6 km s⁻¹ at March 24.2 UT. The SiO maser emission was formed in the inner part of the circumstellar envelope (Deguchi et al. 2011). The narrow absorption lines were likely formed in the same region. The mean radial velocity of the narrow emission components of Fe ii and [Fe ii] lines was −32 ± 1 km s⁻¹, i.e., the narrow absorption lines were blueshifted by about 8 km s⁻¹. Probably, the blueshifts were due to an expansion of the circumstellar envelope.

Zoom In Zoom Out Reset image size

Medium-resolution spectra covering the spectral range from 5340 to 7754 Å were obtained on March 28: day 17.4. A tracing of one of the spectra is shown in Figure 5, where the coronal emission line [Fe x] 6374 was prominent. The emission lines [Ar x] 5535 and [Ar xi] 6919 were also detected, but the other coronal lines that appeared on the outbursts of RS Oph, e.g., [Fe xiv] 5303 or [Ni xiii] 5116 (Iijima 2009), were not detected. The coronal emission lines grew rapidly in intensity between March 28 and April 6, that is day 17.4–26.4, as will be discussed in Section 5.2. The intensities of selected emission lines are given in Table 2. The emission line [Ar x] 5535 was blended with Fe ii 5534.9. It was difficult to estimate the intensity of Fe ii 5534.9, because no other line in the same multiplet (Moore 1959) was detected. We experimentally evaluated the intensity ratios of Fe ii 5534.9 with other Fe ii lines assuming the absence of [Ar x] 5535 in the spectra obtained on 2010 March 14 and 15. The derived intensity ratios were Fe ii (5196/5535) = 1.75, Fe ii (5235/5535) = 2.26, Fe ii (5317/5535) = 5.63, Fe ii (5425/5535) = 0.74, Fe ii (6149/5535) = 1.35, and Fe ii (6248/5535) = 0.99. The contribution from Fe ii 5534.9 in the blend was estimated with these ratios.

Zoom In Zoom Out Reset image size

A tracing of a high-resolution spectrum obtained on April 22: day 42.3 is shown in Figure 6, where the emission lines of He ii strengthened. When this spectrum was obtained, the luminosity was about 3.5 mag below maximum (Figure 1), which means that the object should have been in the 4640 stage in the spectral evolution of novae (McLaughlin & Greenstein 1960). The emission lines of N iii around 4640 Å, however, were not intense (Figure 6); also, no emission line of N ii, e.g., that at 5680 Å, was detected in our observations. The weakness of the N iii and N ii emission lines seems to be a characteristic of the symbiotic recurrent novae, because those were weak also in RS Oph (Iijima 2009), V745 Sco (Williams et al. 1991), and V3890 Sgr (Williams et al. 1991), while fairly intense N iii and N ii emission lines were detected in the other types of recurrent novae, e.g., U Sco (Iijima 2002; Diaz et al. 2010), T Pyx (Catchpole 1969), CI Aql (Iijima 2012b), and V394 CrA (Williams et al. 1991). The mean radial velocity of the narrow absorption lines of Ti ii and Cr ii on the broad emission components of the H i lines was −43 ± 1 km s⁻¹, and that of the narrow emission components of the Fe ii and [Fe ii] lines was −39 ± 1 km s⁻¹.

Zoom In Zoom Out Reset image size

Shore et al. (2012) observed many metallic absorption lines including Ti ii, Cr ii, and Y ii on days 20 and 22. The mean radial velocity of the absorption lines of these three species was −55.4 ± 0.7 km s⁻¹. This blueshift was higher than both −39.8 ± 0.8 km s⁻¹ on day 13.4 and −43 ± 1 km s⁻¹ on day 42.3. Since the radial velocities of the absorption lines did not vary with in a linear way, there might have been an oscillation or something else in the circumstellar envelope. The metallic absorption lines were not seen in the spectrum obtained on day 48 (Shore et al. 2012), that is they disappeared between days 42.3 and 48.

A bend was seen on the light curve about 50 days after maximum light: JD2455316 (Figure 1). Similar features were seen also on the other optical (Munari et al. 2011; Shore et al. 2011) and UV (Shore et al. 2011; Nelson et al. 2012) light curves. When the light curve bent, the emission lines of [O i], [N ii], [O iii], and [Fe ii] started to strengthen suddenly, which suggests a rapid decreasing of the electron density in the nebulosity. This phenomenon will be discussed in Section 5.3.

The emission lines of Fe ii usually disappear in the early stages of outbursts of classical novae (McLaughlin & Greenstein 1960; Williams et al. 1991), while they were prominent even in the nebular stage in the present case (Figure 6). The coexistence of the emission lines of Fe ii and [O iii] seems to be a characteristic of a type of recurrent novae, because it was observed also in RS Oph (Iijima 2009) and V3890 Sgr (Williams et al. 1991). As a matter of fact, this is not a general characteristic of recurrent novae, because the emission lines of Fe ii disappeared in the early stages of the outbursts of U Sco (Iijima 2002; Diaz et al. 2010), T Pyx (Catchpole 1969), CI Aql (Iijima 2012b), V745 Sco (Williams et al. 1991), and so on.

Our last medium-resolution spectra in 2010 were obtained on May 18: day 73.3. A tracing of one of them is shown in Figure 7, where the prominent emission lines were due to H i, He i, He ii, Si ii, Fe ii, O iv, [O i], [O iii], [N ii], [S ii], [Fe ii], [Ni II], [Fe vii] etc. The coronal line [Fe x] 6374 had weakened relative to the other forbidden lines. The intensities of selected emission lines are given in Table 2.

Zoom In Zoom Out Reset image size

The distances to this object in the previous works, 2.7 kpc by Munari et al. (1990) or 1.9 kpc by Kolotilov et al. (1998), were based on the absolute K mag of the secondary component derived from the period–luminosity relations of Mira variables (Glass & Feast 1982; Feast et al. 1989). The relations, however, were derived using Mira variables with periods shorter than 420 days. Feast et al. (1989) reported that there was a bend in the period–luminosity relation of Mira variables around the period 420 days. Figure 8 shows periods and absolute mean K magnitudes of Mira variables with periods of 400 days and longer (Feast et al. 1989) and those of long-period variables given by Wood & Sebo (1996). The latter data are plotted as open circles. The straight line shows a result of the least-squares fitting to them, where the point at log(P) = 2.89 is excluded from the fitting, because it is isolated from the others. The relation is

$$\begin{eqnarray}{M}_{K} & = & -7.9\;\mathrm{log}(P)+12.4.\\ & & \pm 0.2\qquad \pm 0.6\end{eqnarray} \tag{ 4 }$$

Zoom In Zoom Out Reset image size

We have a mean absolute K mag of the secondary Mira variable ${M}_{K}=-10.4\pm 0.8$ mag with this relation at P = 762.9 days (Kolotilov et al. 2003), which is about 1.3 mag brighter than that expected in the relation of Feast et al. (1989).

There was serious ambiguity in the observed mean K mag of V407 Cyg in the quiescent stage, because the amplitude of its variation was unusually high, ${\rm{\Delta }}K\cong 1$ mag, and sometimes changed (Kolotilov et al. 1998, 2003). There might have been an unknown mechanism to amplify the variation. For the purpose of avoiding the effect of this mechanism, we did not use the raw mean of the observed K mag, but we used the observed K mag at minimum to estimate the mean of the variable K mag owing to the pulsation of the Mira variable. We found the K mag at minimum in the quiescent stage to have been ${m}_{K}=3.5\pm 0.2$ mag from the photometric data of Kolotilov et al. (2003). Since the expected amplitude of the variations in K mag of Mira variables owing to their pulsations is ΔK = 0.14 mag (van Belle et al. 1996), the mean K mag is likely 0.07 mag brighter than the minimum. Therefore, the mean K mag is expected to have been ${m}_{K}=3.43$ mag.

The interstellar extinction in the K band is ${A}_{K}=0.11\times {A}_{V}$ for the standard extinction curve of ${A}_{V}=3.1$ × $E(B-V)$ (Mathis 1990), that is we have ${A}_{K}=0.20$ mag. Using these results, we obtained a distance $5.3\pm 1$ kpc for V407 Cyg.

The time for 2 mag decline from maximum light was 6.5 ± 0.5 day (Section 3). If we apply the relation between the rate of decline and the maximum luminosity of classical novae (Della Valle & Livio 1995) to this result, we have ${M}_{V}=-8.85\pm 0.2$ mag as the absolute magnitude at maximum light. The distance to V407 Cyg would be 7.1 ± 0.6 kpc as reported in our previous works (Iijima 2012a, 2014) if we adopted this absolute magnitude, which, however, is inconsistent with that derived above. The relation of Della Valle & Livio (1995) gives the mean of absolute magnitudes at maximum light, but there is large dispersion in the observed data. The lower limit of the absolute V mag at maximum light in the 3σ dispersion of their formula is ${M}_{V}=-8.32$ mag, with which we have a distance 5.5 ± 0.5 kpc. This distance agrees with that derived above in the limits of the errors. The absolute V mag at maximum light of the outburst of V407 Cyg seems to have been about 0.6 mag fainter than that given by the relation of Della Valle & Livio (1995).

Chomiuk et al. (2012) estimated a lower limit for the distance to V407 Cyg to have been 3 kpc in an analysis of the interstellar absorption components of Na i D1 and D2 lines. Their result is consistent with ours.

Intensities of the coronal emission lines of [Fe x] 6374, [Ar x] 5535, and [Ar xi] 6919 relative to Hβ are plotted in Figure 9, where the broken line shows the soft X-ray flux (0.3–10 keV) given by Nelson et al. (2012) on a normalized scale. The intensities of the coronal emission lines and the soft X-ray flux rapidly increased around day 20. A similar phenomenon was observed during the outburst of RS Oph in 2006, where the variations in intensity of the coronal emission lines (Iijima 2009) coincided with that of the super-soft X-ray flux (0.3–0.55 keV) given by Ness et al. (2007).

Zoom In Zoom Out Reset image size

Nelson et al. (2012) proposed a model for the interaction between the ejecta and the circumstellar envelope around V407 Cyg, which well reproduced the variation of the soft X-ray flux during the outburst. The coronal lines were likely emitted in the same process in the early stage.

Nelson et al. (2012) themselves suggested that a certain fraction of the soft X-rays was due to the thermal radiation of the hot component and its fading around day 50 was related to the cessation of hydrogen burning. As discussed in the next subsection, however, the intensity ratio of He ii 4686/Hβ, which is an indicator of the surface temperature of the exciting star, did not decrease but increased when the soft X-rays faded. Our observations suggest that the fading of the soft X-rays was not related to a decreasing of the thermal emission of the hot component.

The flux of radio continuum emissions of higher frequency, those at 30 GHz (Peel et al. 2010) and 45 GHz (Chomiuk et al. 2012), slowly decreased in the period after day 40 and did not significantly increase or decrease around day 50. Their variations in intensity were rather similar to those of the coronal emission lines (Figure 9). Unfortunately, we are unable to make a further detailed argument for the similarity, because we have only a few data with low accuracy of the coronal lines in that period.

The soft X-ray flux faded and the non-coronal forbidden lines strengthened in the period between days 40 and 60: JD2455316±10 (Figure 10). The bends on the light curves were seen in the same period (Figure 1, see also Munari et al. 2011; Shore et al. 2011; Nelson et al. 2012). The growth in intensity of the forbidden lines suggests that the electron density in the emitting nebulosity suddenly decreased in that period.

Zoom In Zoom Out Reset image size

Hachisu & Kato (2012) proposed a model for the outburst of V407 Cyg, where the bends on the light curves were related to the cessation of hydrogen burning on the hot component. If this had been the case, however, the temperature of the hot component, which was related to the intensity ratio of the emission lines of He ii 4686/Hβ, should have decreased when the bends appeared, but such a phenomenon was not observed (Figure 14). It seems that the bends on the light curves were more likely related to a decline of the emission measure of the nebulosity owing to the decreasing of the electron density (Section 4.3).

Figure 11 shows ${m}_{V}$ until day 90. The broken line is the best-fitted third-order polynomial curve for ${m}_{V}$, where the first point near maximum light and those in the period from day 14 to day 60 were excluded in the fitting. The lower panel shows the excess of the flux in the V band from the fitted curve, and the dotted line shows the soft X-ray flux given by Nelson et al. (2012) on a normalized logarithmic scale. The same process was applied to the UV mag (Figure 12) given by Nelson et al. (2012) and the absolute intensity of the Hβ emission line given in Table 2 on a logarithmic scale (Figure 13). When only the intensity of Hα was available, that of Hβ was estimated using the intensity ratio of Hα/Hβ in the spectrum obtained on the nearest date. Because of the small number of data, a second-order polynomial curve was fitted for Hβ. The figure for ${m}_{B}$ is not presented, because it is nearly the same as that for ${m}_{V}$.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The excess of the intensity of the Hβ emission line varied contemporaneously with the soft X-ray flux until day 40 (Figure 13). The equivalent width of the emission line of O i 8446 Å reported by Shore et al. (2011) showed a similar variation. The excess of the fluxes in UV and V bands was large when the soft X-rays were intense, but their peaks of excess were found later than the peak of the soft X-ray flux (Figures 11 and 12). The soft X-ray flux is related to the emission measure of the nebulosity in the model of Nelson et al. (2012). The intensity of the Hβ emission line and the free–free continuum emission in the UV and optical regions are related to the emission measure too (e.g., Osterbrock 1989). Increasing of the free–free emission is a probable cause of the excess of the UV and optical continuum. The time lag of the peaks of excess in UV and V bands from that of soft X-rays might have been due to the time needed to decrease the electron temperature in the nebulosity.

The variations of the quantities argued here could be explained with the following scenario. The ejecta may have been surrounded by the circumstellar envelope at the beginning of the expansion, because the narrow absorption lines of the ionized metals were seen on the broad emission components of H i lines (Figure 4). The effective collision between the ejecta and the circumstellar envelope started around day 20, which means that the mass of the swept-up matter became comparable to that of the ejecta (Nelson et al. 2012). The emission measure of the nebulosity increased greatly, because additional ionized gas was supplied by the collisional ionization. The coronal emission lines and the rather hard X-rays (Nelson et al. 2012) were emitted in the hottest region, and the recombination lines H i and O i were emitted in a cooler region. Then the free electrons contributed to the free–free emission in the soft X-ray, UV, and optical regions, losing their energy. The top of the ejecta reached the outer boundary of the circumstellar envelope around day 40. The collision effectively ceased at that moment and the free expansion of the ejecta started. The major part of the ejecta possibly escaped from the circumstellar envelope until day 48, because the narrow absorption lines disappeared (Section 4.3). No more collisional ionization occurs, and the emission measure decreased suddenly due to the free expansion, which caused the sudden drops of the quantities mentioned here.

The model of Nelson et al. (2012) for the interaction between the ejecta and the circumstellar envelope well reproduced the variation of the soft X-ray flux during the outburst of V407 Cyg. They adopted, however, the very long orbital period, 43 years, proposed by Munari et al. (1990) for the binary system. Since the separation between the primary and secondary components should be much larger than the dimension of the envelope in their model, the ejecta collide with the circumstellar envelope from outside and penetrate it. It seems to be difficult to explain the sudden decreasing of the electron density between days 40 and 60 with such a model, because the major part of the ejecta, which was not directed to the secondary component, was able to expand freely from the first. A more realistic model is that the ejecta started to expand within the circumstellar envelope and its expansion was limited by the collision with it until the top of the ejecta reached the outer boundary of the envelope. The binary system may be more compact.

Chomiuk et al. (2012) also assumed the very long orbital period in their models and hypothesized that the secondary Mira variable was located in front of the hot component in our line of sight when the outburst occurred. Such a hypothesis, however, is inconsistent with the rapid disappearance of the metallic absorption lines between days 42.3 and 48 (Section 4.3).

Figure 14 shows the intensities of the emission lines of He i and He ii relative to Hβ. The intensities of the emission lines of He i relative to Hβ decreased in the first two weeks of the outburst, then began to increase later. A similar phenomenon was observed during the outburst of RS Oph in 2006, where the intensities of the He i emission lines suddenly increased between days 21 and 22 of the outburst, then decreased in the successive 30 days and began to increase later (Iijima 2009). The sudden increasing in intensity of the He i lines of V407 Cyg seems to have occurred prior to our first observation at day 3.4. The variations of the He i lines in intensity were faster in V407 Cyg than in RS Oph.

Zoom In Zoom Out Reset image size

The γ-ray flux of V407 Cyg rapidly increased in the first 3 or 4 days of the outburst, then decreased in the following 10 days (Abdo et al. 2010). The variations of the He i emission lines in intensity coincided with that of the γ-ray flux in the first two weeks of the outburst, while no other emission lines showed a similar variation. At the present time we do not know whether there was a relation between the emission mechanism of the He i emission lines and that of the γ-rays. Possible mechanisms for the γ-ray emission were discussed by several authors (Abdo et al. 2010; Nelson et al. 2012; Martin & Dubus 2013), but no relation was suggested for the He i emission lines.

The helium abundance is estimated from the intensities of the He i and He ii emission lines relative to Hβ. The basic idea is presented in the text of Osterbrock (1989) and the formulae are given by Iijima (2006). The effects of the collisional excitation of the He i lines are corrected with the formulae of Peimbert & Torres-Peimbert (1987). The emission line of [O iii] 5007 was weaker than Hβ throughout our observations in 2010 (Table 2), which means that the electron density in the nebulosity was high, namely ${N}_{{\rm{e}}}\geqslant {10}^{7}\;{\mathrm{cm}}^{-3}$. It is possible to estimate the electron temperature in such a nebulosity from the intensity ratio of the [O iii] emission lines ${I}(5007+4959)/{I}(4363)$ in the high density limit. We used the command temden in stsdas.analysis.nebular of IRAF, and obtained a mean electron temperature $10,800\pm 300$ K. The helium abundance, which is independent of the electron density in such a high density condition, was derived assuming T_c = 10,800 K. The helium abundance outside the outburst stage, on 2001 November 4 and 2011 September 21, is also derived. The electron temperature 10,800 K is assumed for the spectra in 2001 and 2011. The electron density on 2011 September 21 was estimated to have been ${N}_{{\rm{e}}}=3.6\times {10}^{6}\;{\mathrm{cm}}^{-3}$ from the intensity ratio of the [O iii] lines using the formula of Seaton (1975). The results are given in Table 3 and are plotted in Figure 15.

Zoom In Zoom Out Reset image size

The helium abundance (Figure 15) appears to have varied in intensity according to the variation of the He i emission lines (Figure 14). However, we cannot simply conclude that the helium abundance really changed, because the abundances around day 20 were unusually low with respect to those obtained outside the outburst stage: N(He) = 0.20 or 0.19 (Table 3). The low abundances were owing to the temporal fading of the He i emission lines at that moment (Figure 14), the cause of which is not known yet. A similar phenomenon might have occurred on the outburst of another recurrent nova, U Sco, because fairly different helium abundances were reported: those were 0.16 at 16 hr after maximum light (Iijima 2002), 0.4 on day 11 (Anupama & Dewangan 2000), and 2.0 on day 12 (Barlow et al. 1981). Further work is needed on the behavior of the He i lines in the outbursts of recurrent novae.

In contrast to the helium abundance, the mass of emitting nebulosity strongly depends on the electron density. Therefore, we are able to make only a rough estimate. The formula of Seaton (1975) gives an electron density ${N}_{{\rm{e}}}=1.0\times {10}^{7}\;{\mathrm{cm}}^{-3}$ for the intensity ratio of the [O iii] emission lines ${I}(5007+4959)/{I}(4363)=3.65$ on 2010 April 30: day 50.3, where the electron temperature is assumed to have been 10,800 K. The mass of the emitting nebulosity, the sum of the ejecta and the ionized circumstellar envelope, is estimated to have been $1.4\times {10}^{-4}\;{M}_{\odot }$, where we adopt the distance 5.3 kpc. Our result supports a model of Chomiuk et al. (2012) with a massive ejection (${M}_{\mathrm{ej}}\approx 1\times {10}^{-4}\;{M}_{\odot }$). The upper limit of the surface temperature of the hot component derived from the intensities of the He ii and He i emission lines relative to Hβ (Iijima 2006) was roughly 130,000 K on the same date.

The Raman emission band at 6830 Å (Schmid 1989) was prominent in the spectra of RS Oph on the outburst in 2006 (Iijima 2009), while no positive result for detection of this band was reported on V407 Cyg (Shore et al. 2011). Figure 16 shows traces of the red region of our spectra obtained on 2010 March 14 and April 9. The ordinate is an arbitrary intensity scale. The Raman band was not seen on March 14, but its weak trace was seen on April 9 (Figure 16). The intensities of the Raman band are given in Table 2 and are plotted in Figure 17.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

An emission line at 6028 Å was seen in some spectra (Figures 3 and 6). Its laboratory wavelength is estimated to be 6028.3 ± 0.2 Å. This line is presented as an unidentified one at 6028.3 Å in the catalog of Meinel et al. (1968). We examined again the spectra of RS Oph on the outburst in 2006 and found weak traces of this line in some spectra, but its intensity did not exceed 0.6% of Hβ. In the case of V407 Cyg, the intensity of this line reached about 4% of Hβ in 2010 May (Table 2). As seen in Figure 17, its intensity varied contemporaneously with that of O iv 6106, which is also given as an unidentified one at 6105.5 Å in the catalog of Meinel et al. (1968), and an identification as O iv 6105.98 Å was proposed by Iijima (2009). The intensities of the emission lines of Fe ii also showed similar variations, but they were prominent in the first spectra (Figure 17). The emission line at 6028 Å probably depends on a highly ionized ion such as O iv, but we are unable to find any reasonable candidate for identification in the NIST Atomic Spectral Database.²

The Li i 6708 absorption line was seen on March 14 (Figure 16). Its equivalent width was 0.14 Å at that time then weakened rapidly, to 0.14 Å on March 15, 0.06 Å on March 16, and 0.03 Å on March 28. It was hard to notice this absorption line in April and later because of its weakness and the low signal-to-noise ratios of the spectra.

Tatarnicova et al. (2003) detected the Li i 6708 absorption line in high-resolution spectra of V407 Cyg in the quiescent stage. The absorption line in their spectra was formed in the atmosphere of the secondary Mira variable (Tatarnicova et al. 2003). On the other hand, that in our spectra may have arisen from a different origin, because its depth decreased according to the luminosity of the outburst. Probably, it was formed in the circumstellar envelope like the other metallic absorption lines (Section 4.2).