Detection of a High-velocity Prominence Eruption Leading to a CME Associated with a Superflare on the RS CVn-type Star V1355 Orionis

V1355 Orionis (=HD291095) is an RS CVn-type binary system discovered through the ROSAT WFC all-sky X-ray survey (Pounds et al. 1993; Pye et al. 1995). V1355 Orionis has been investigated by Cutispoto et al. (1995), Osten & Saar (1998), and Strassmeier (2000). These studies have shown that this binary comprises K0-2IV and G1V stars. Table 1 summarizes the basic physical parameters of the K-type subgiant star. Strassmeier (2000) reported a large flare on V1355 Orionis in 1998 April, which showed 70 times larger equivalent width of Hα than the pre-flare.

2.2.1. Photometric Observation: TESS

TESS observed V1355 Orionis in Sector 34 for 27 days with the 2 minutes time cadence. Figure 1(a) shows the TESS light curve for this period (BJD = 2459201.7-2459227.5). Figure 1(b) shows the enlarged TESS light curve around the flare described in this paper. The quiescent radiation component during the flare is estimated by fitting a polynomial to the light curve for −500 to −10 minutes and +150 to +380 minutes from the flare start (see the sky-blue dashed line in Figure 1(b)). Figure 2(a) shows the light curve for the flare component, created by subtracting the quiescent component from the light curve of TESS during the flare. We calculated the bolometric energy of the flare from this detrended light curve following the method of Shibayama et al. (2013). See Section 3.1 for more details about caluculating the flare energy.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

2.2.2. Spectroscopic Observation: The Seimei Telescope

As shown in Figure 2(a), the white-light flare lasted about 110 minutes. By setting the flux of Figure 2(a) to $C{{\prime} }_{\mathrm{flare}}$ (= luminosity of flare/luminosity of star) in the Equation (5) of Shibayama et al. (2013), we estimated the flare area ${A}_{\mathrm{flare}}$ as follows:

$$\begin{eqnarray}&&{A}_{\mathrm{flare}}(t)=\displaystyle \frac{\pi C{{\prime} }_{\mathrm{flare}}(t){\sum }_{i=\mathrm{1,2}}\left\{{R}_{i}^{2}\int {R}_{\lambda }{B}_{\lambda }({T}_{i})d\lambda \right\}}{\int {R}_{\lambda }{B}_{\lambda }({T}_{\mathrm{flare}})d\lambda }\end{eqnarray} \tag{ 1 }$$

where λ is the wavelength, B_λ is the Planck function, R_λ is the TESS response function (Ricker et al. 2015), and T_i is the effective temperature of the K- and G-type stars of the binary (4750 K and 5780 K, respectively; Strassmeier 2000). Further, ${T}_{\mathrm{flare}}$ represents a flare temperature of 10,000 K (Mochnacki & Zirin 1980; Hawley & Fisher 1992), and R_i is the radius of the K- and G-type stars of the binary (4.1R_⊙ and 1.0R_⊙, respectively; Strassmeier 2000). Assuming that the flare can be approximated by blackbody radiation with a temperature of ${T}_{\mathrm{flare}}=10,000\ {\rm{K}}$, flare luminosity ${L}_{\mathrm{flare}}$ is

$$\begin{eqnarray}&&{L}_{\mathrm{flare}}(t)={\sigma }_{\mathrm{SB}}{T}_{\mathrm{flare}}^{4}{A}_{\mathrm{flare}}(t)\end{eqnarray} \tag{ 2 }$$

where σ_SB is the Stefan–Boltzmann constant. Finally, by integrating ${L}_{\mathrm{flare}}$ with the duration of the white-light flare (∼110 minutes), the bolometric flare energy E_bol is

$$\begin{eqnarray}&&{E}_{\mathrm{bol}}=\int {L}_{\mathrm{flare}}(t)\ {dt}=7.0\times {10}^{35}\,\mathrm{erg}.\end{eqnarray} \tag{ 3 }$$

Given that the K-type star is more magnetically active and the bolometric flare energy is very large, we consider this flare to have occurred on the K-type star of the binary.

Our spectroscopic observation revealed that a remarkable blueshifted excess component exists in the Hα emission line for ∼30 minutes after the start of the flare. Figure 3(a) shows the spectra during the time when the blueshifted excess component was observed during the flare. We estimated the spectrum of the quiescent component, denoted by red spectrum in Figure 3(a), by combining the spectra from 15 minutes before the flare start ("pre-flare"). Figure 3(b) shows the spectrum of the flare emission component, which were obtained by obtaining the difference between the spectrum at each time and the pre-flare spectrum. A particularly large blueshifted excess component was observed in the pre-flare subtracted spectrum, especially 5–18 minutes, extending over −1000 km s⁻¹. Figure 4(d) presents a color map showing the time variation of the blueshifted excess component in the pre-flare subtracted spectrum. The blueshifted excess component was fast over the time period when white light and Hα reach their peak.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While the spectrum before the difference was not redshifted (Figure 3(a)), the Hα peak of the pre-flare subtracted spectrum was slightly redshifted (∼ + 50 km s⁻¹; Figures 3(b) and 4(d)). All spectra we present in this paper are not corrected for the radial velocity. According to Strassmeier (2000), the radial velocity of V1355 Orionis varies in the range of about +0 − + 70 km s⁻¹. That is, the velocity of the redshift is within the range of the radial velocity variability. The redshift might be caused by the downward chromospheric condensation (Ichimoto & Kurokawa 1984) or the post flare-loop (Claes & Keppens 2019).

We divided the Hα emission line into flare and blueshifted excess components and then calculated their equivalent widths. Figure 4(c) shows their light curves. The emission lines were separated into flare and blueshifted excess components by fitting only the long wavelength side of the lines with the Voigt function. See Section 4.1 for details about the fitting. From the light curve separated into flare and blueshifted excess components, we calculated the energy of the flare emitted in Hα (E_Hα ). Using the R-band magnitude (m_R ; Cutispoto et al. 1995), R-band Vega flux zero-point per unit wavelength (f_λ = 217.7 × 10⁻¹¹ erg cm⁻² s⁻¹ Å⁻¹; Bessell et al. 1998), and the distance between the Earth and V1355 Orionis (d; Gaia Collaboration et al. 2016), the equivalent width of the flare (${\mathrm{EW}}_{\mathrm{flare}}$) can be converted to luminosity L_Hα ,

$$\begin{eqnarray}&&{L}_{{\rm{H}}\alpha }(t)=4\pi {d}^{2}{f}_{\lambda }{10}^{-0.4{m}_{R}}\times {\mathrm{EW}}_{\mathrm{flare}}(t).\end{eqnarray} \tag{ 4 }$$

Table 1 lists the values of d and m_R . By integrating L_Hα with the duration of the Hα flare (∼210 minutes),

$$\begin{eqnarray}&&{E}_{{\rm{H}}\alpha }=\int {L}_{{\rm{H}}\alpha }(t)\ {dt}=1.1\times {10}^{34}\,\mathrm{erg},\end{eqnarray} \tag{ 5 }$$

is obtained.

As discussed in Section 3.2, a clear blueshifted excess component in the Hα emission line was identified during the flare. This blueshifted excess component must reflect a prominence eruption since its velocity extends to more than 1000 km s⁻¹. A cool upflow (Tei et al. 2018) cannot explain such a rapid blueshift. Therefore, we estimated the velocity and mass of the prominence that appears as the blueshifted excess component.

4.1.1. Velocity

We estimated the velocity of the prominence using a method similar to that used by Maehara et al. (2021). As shown in Figures 5(a) and (c), we first fit the pre-flare subtracted spectrum at each time with the Voigt function only at wavelengths longer than the line center. Then, we calculated the residuals between the Voigt function and the pre-flare subtracted spectrum. The residual is shown by the blue line in Figures 5(b) and (d). Finally, the residual was fitted with a Gaussian. The wavelength of the Gaussian peak was converted to the Doppler velocity to evaluate the prominence velocity.

Zoom In Zoom Out Reset image size

Spectra exist over multiple time periods when the residual appeared to have two peaks, as shown in Figures 5(b) and (d). Therefore, when fitting the residuals, we performed two types of fits: one- and two-component Gaussian fits (Figures 5(b) and (d), respectively). See Section 4.2 for the details of and interpretation on why the blueshifted excess component appears to have two peaks.

The time variation of the velocity of the prominence examined in this manner is shown in Figures 6(d) and (h). In the case of a one-component fit, the velocity of the prominence reached −990 ± 130 km s⁻¹ at the peak. In the case of a two-component fit, the velocity of the prominence reached −1690 ± 100 km s⁻¹ and −760 ± 90 km s⁻¹ at the peak. In the case of both fits, the prominence velocity was almost always much faster than the escape velocity (−347 km s⁻¹) at the surface of the K-type star of V1355 Orionis. Therefore, the prominence that erupted with this flare would certainly have flown outward from the star and developed into a CME.

Zoom In Zoom Out Reset image size

Moreover, in both cases, the velocity of the blueshifted excess component decelerated more rapidly than the gravitational acceleration of the K-type star after reaching its peak (see gray lines in Figures 6(d) and (h)). The two possible interpretations are as follows. The first interpretation is that what we observe in Figures 6(d) and (h) is not the actual deceleration. The fast part of the erupted prominence becomes invisible rapidly, and the slow part becomes gradually dominant. This is observed in the Sun-as-a-star analysis of a filament eruption analyzed by Namekata et al. (2022a; see the supplementary information therein). The second interpretation is that the magnetic field applies force to the prominence in addition to gravity. Simulations conducted by Alvarado-Gómez et al. (2018) have shown that the magnetic field of an active star could contribute significantly to the slowing of prominences.

4.1.2. Mass

We estimated the upper and lower limit prominence mass from the equivalent width of the blueshifted excess component. In the method used by Maehara et al. (2021), the upper limit of prominence mass is proportional to the 1.5 power of the area of the region emitting Hα. Therefore, for a prominence eruption as large in scale as this case, the method can significantly overestimate the upper limit of prominence mass because the prominence shape would be far from cubic-like solar prominences. So, we improved the method used by Maehara et al. (2021).

As shown in Figure 4(c), the maximum equivalent width of the blueshifted excess component is ∼1 Å. Converting the equivalent width to luminosity using Equation (4), the luminosity of the blueshifted excess component L_blue is obtained as

$$\begin{eqnarray}&&{L}_{\mathrm{blue}}\sim 1\times {10}^{30}\,\mathrm{erg}\,{{\rm{s}}}^{-1}.\end{eqnarray} \tag{ 6 }$$

We assume that the NLTE model of the solar prominence (Heinzel et al. 1994) can be adapted to the present case, and further assume that the optical thickness of Hα line center τ_p is 0.1–100.

• 1.  
  τ_p ∼ 0.1: The Hα flux of the prominence per unit time, unit area, and unit solid angle F_Hα is

  $$\begin{eqnarray}&&{F}_{{\rm{H}}\alpha }\sim {10}^{4}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\ {\mathrm{cm}}^{-2}\ {\mathrm{sr}}^{-1},\end{eqnarray} \tag{ 7 }$$

  (see Figure 5 in Heinzel et al. 1994). As the integral of F_Hα over the region emitting the Hα and the solid angle in the direction toward us is L_blue,

  $$\begin{eqnarray}&&{L}_{\mathrm{blue}}=\int \int {F}_{{\rm{H}}\alpha }\ {dAd}{\rm{\Omega }}=2\pi {{AF}}_{{\rm{H}}\alpha },\end{eqnarray} \tag{ 8 }$$

  where A is the area of the region emitting Hα. From Equations (6)−(8),

  $$\begin{eqnarray}&&A\sim 5\times {10}^{24.5}\,{\mathrm{cm}}^{2},\end{eqnarray} \tag{ 9 }$$

  is obtained. From Equation (7), the emission measure ${n}_{{\rm{e}}}^{2}D$ of the prominence is

  $$\begin{eqnarray}&&{n}_{{\rm{e}}}^{2}D\sim {10}^{28}\,{\mathrm{cm}}^{-5},\end{eqnarray} \tag{ 10 }$$

  (see Figure 15 in Heinzel et al. 1994) where D and n_e are the geometrical thickness and the electron density of the prominence, respectively. Heinzel et al. (1994) assumed a range of D. On the other hand, F_Hα , and ${n}_{{\rm{e}}}^{2}D$ are largely uniquely determined for a value of τ_p without much influence from the indefiniteness of D. On the other hand, we need to assume values of the electron density n_e and the hydrogen density n_H. The typical electron density of a solar prominence is

  $$\begin{eqnarray}&&{n}_{{\rm{e}}}\sim {10}^{10-11.5}\,{\mathrm{cm}}^{-3}.\end{eqnarray} \tag{ 11 }$$

  (Hirayama 1986). From Equations (10) and (11),

  $$\begin{eqnarray}&&D\sim {10}^{5-8}\,\mathrm{cm}.\end{eqnarray} \tag{ 12 }$$

  Labrosse et al. (2010) showed the ratio between the hydrogen density n_H and the electron density n_e of a solar prominence is

  $$\begin{eqnarray}&&{n}_{{\rm{e}}}/{n}_{{\rm{H}}}\sim 0.2-0.9.\end{eqnarray} \tag{ 13 }$$

  From Equations (9), (11), (12), and (13), the mass of the prominence

  $$\begin{eqnarray}&&M\sim {m}_{{\rm{H}}}{n}_{{\rm{H}}}{AD}\end{eqnarray} \tag{ 14 }$$

  is

  $$\begin{eqnarray}&&9.5\times {10}^{18}\,{\rm{g}}\lt M\lt 1.4\times {10}^{20}\,{\rm{g}},\end{eqnarray} \tag{ 15 }$$

  where m_H is the mass of hydrogen atom. The error range comes from the assumed range of electron density and degree of ionization.
• 2.  
  τ_p ∼ 100: When the value of τ_p is ∼100,

  $$\begin{eqnarray}&&{F}_{{\rm{H}}\alpha }\sim {10}^{6}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\ {\mathrm{cm}}^{-2}\ {\mathrm{sr}}^{-1},\end{eqnarray} \tag{ 16 }$$

  (see Figure 5 in Heinzel et al. 1994). Calculated as in case τ_p ∼ 0.1,

  $$\begin{eqnarray}&&A\sim 1.6\times {10}^{23}\,{\mathrm{cm}}^{2}.\end{eqnarray} \tag{ 17 }$$

  Assuming Equations (11) and (13) as in case τ_p ∼ 0.1,

  $$\begin{eqnarray}&&D\sim {10}^{8-11}\,\mathrm{cm}\end{eqnarray} \tag{ 18 }$$

  $$\begin{eqnarray}&&9.5\times {10}^{19}\,{\rm{g}}\lt M\lt 1.4\times {10}^{21}\,{\rm{g}}.\end{eqnarray} \tag{ 19 }$$

Combining the ranges of M in Equations (15) and (19),

$$\begin{eqnarray}&&9.5\times {10}^{18}\,{\rm{g}}\lt M\lt 1.4\times {10}^{21}\,{\rm{g}}.\end{eqnarray} \tag{ 20 }$$

As shown in Equations (15) and (19), the upper and lower limits of the prominence mass varied by only about an order of magnitude when τ_p varied significantly. Since we do not know the value of τ_p, we set an extreme value of τ_p ∼ 100 as the upper limit in this paper. The energy of this flare is ∼10⁶ times that of a typical solar flare (∼10³⁰ erg). The energy of a flare is proportional to the cube of the spatial scale (Shibata & Yokoyama 2002). The typical value of the optical thickness of the solar prominence is ∼1. Simply put, the optical thickness of the prominence is proportional to the geometric thickness of the prominence. Given the spatial scale of the prominence, the upper limit of τ_p can be roughly considered to be $\sim 1\times {({10}^{6})}^{1/3}=100$. On the other hand, the lower limit of τ_p was limited by the hemispheric area of the star. When τ_p is set to an extremely small value, A is significantly larger than the hemispherical area of the star (see Equations (6) and (7)). Such a situation is unrealistic. The value of A in Equation (9) corresponds to ∼10^1.5 π R². The filling factor effect may also affect the prominence mass estimation (Kucera et al. 1998). So, a more detailed study of the stellar prominence mass calculation is needed in the future.

Sometimes the blueshifted excess component appeared to have two clear peaks, as shown in Figures 5(b) and (d), whereas at other times the case was opposite. Figures 6(c) and (g) show the time variation of the equivalent width of the Gaussian fitted to the residual for the one- and two-component fits, respectively. As shown in Figure 6(g), the equivalent width of one of the two Gaussians was close to zero at the beginning and end of the time when the blueshifted excess component was visible. Both equivalent widths were above a certain value in the intervening. That is, the blueshifted excess component initially appeared to be one component, then became two components, and finally appeared to be one component again.

We assume that the prominence visibility for the Sun, where the prominence is considered the emission component of Hα and the filament is regarded as the absorption component of Hα (Parenti 2014), holds true here. Then, the fact that the blueshifted excess component appears to be composed of two components can be interpreted in two ways:

• (i)  
  Emission + Emission:As shown in Figure 7(a), two prominences are present above the limb and each prominence is visible as an emission line in Hα. This would be the situation when the prominence eruption occurred twice, or when the erupted prominence split in two while moving.
• (ii)  
  Absorption + Emission:As shown in Figure 7(b), the erupted prominence contains both the area above the limb and is visible as an emission line in Hα, and that inside the limb and is visible as an absorption line in Hα. When mixing of those emission and absorption components, two peaks seem to exist in the blueshifted excess component. In this case, the width of the emission component must be broader than that of the absorption component to reproduce the observed spectrum. However, the physical interpretation for this situation is not well understood.

Nevertheless, whether the premise of these interpretations can be applied to the present case remains unclear. That is, unlike the Sun, filaments may be visible as the Hα emission components on K-type stars. Leitzinger et al. (2022) showed that for dM stars, thermal radiation from filaments dominates the source function over the scattering of the star's incident radiation so that even filaments can be considered emission line components of Hα through 1D NLTE modeling and cloud model formulation. Given that our eruptive event occurred on a K-type star, which is cooler than the Sun, a filament may not necessarily be visible in the absorption component in Hα, similar to that claimed by Leitzinger et al. (2022) for M-type stars. Therefore, modeling and simulation of prominence/filament eruptions on K-type stars are required for a more advanced understanding of the two-peaked blueshifted excess components.

Zoom In Zoom Out Reset image size

We compared the fast prominence eruption observed on V1355 Orionis with other events regarding mass, velocity, and kinetic energy. Figure 8 shows the (a) mass, (b) velocity, and (c) kinetic energy of the prominence eruptions and CMEs as a function of flare energy emitted in the GOES wavelength band (1–8 Å) and bolometric flare energy. We used Equation (1) of Moschou et al. (2019) when converting the energy emitted in Hα to GOES X-ray flare energy: L_X = 16L_Hα . We also assumed the relationship L_bol = 100L_X, which is shown to hold on solar flares by Emslie et al. (2012). On the other hand, Osten & Wolk (2015) showed L_bol = L_X/0.06 for active stars. Therefore, the bolometric energy of the stellar flares in Figure 8 may be a bit smaller. Note that for stars, only examples estimated from blueshifted excess components of chromospheric lines are plotted in Figure 8. The lower and upper limits of the velocity of the prominence eruption on V1355 Orionis in Figure 8(b) are obtained by fitting with two components. The pink circle denotes the velocity obtained by fitting with one component.

Zoom In Zoom Out Reset image size

Figure 8(a) indicates that the erupted prominence on V1355 Orionis has roughly the mass that expected from the scaling law of solar CMEs. This suggests that the prominence eruption on V1355 Orionis was caused by the same physical mechanism as the solar prominence eruptions/CMEs (e.g., Kotani et al. 2023). Figure 8(a) also shows that it is the largest prominence eruption observed by blueshift. These facts may provide important clues as to how a large event can be caused by the physical mechanism of solar prominence eruptions. Our observations may provide an opportunity to understand extreme eruptive events on stars.

Figure 8(b) shows that the prominence velocity on V1355 Orionis is indeed fast, but overwhelmingly below the theoretical upper limit of the velocity (Takahashi et al. 2016) estimated from the flare energy. Therefore, the fast prominence eruption is physically possible to occur in association with a 10³⁵ erg class flare.

Figure 8(c) indicates that kinetic energy corresponds roughly to the value predicted from the scaling law of the solar CMEs. In our calculation, the kinetic energy of the prominence is 4.5 × 10³³ erg < K < 1.0 × 10³⁷ erg. As discussed in Moschou et al. (2019), the kinetic energy of the prominence on other stars is smaller than that expected from the scaling law of solar CMEs. The possible reason for this discrepancy is that prominence eruptions generally have a lower velocity than CMEs in case of the Sun (Maehara et al. 2021; Namekata et al. 2022a), although it is also proposed that the suppression by the overlying large-scale magnetic field can contribute to small kinetic energies (Alvarado-Gómez et al. 2018). For the blueshift events except on V1355 Orionis, the kinetic energy tends to be smaller than the scaling law. However, we are not sure if the prominence of V1355 Orionis is below the scaling law due to the large indefiniteness of the kinetic energy.

Given these considerations, the prominence eruption on V1355 Orionis and solar prominence eruptions may have a common physical mechanism. However, the large uncertainties in the mass estimate make it difficult to compare them with the solar CME scaling law. For the mass estimation, we made various assumptions, as discussed in Section 4.1.2, that may be incorrect. A more accurate derivation of the prominence mass requires a simulation as performed in Leitzinger et al. (2022), as well as concerning the two components in Section 4.2.