Accretion Variability of the Multiple T Tauri System VW Cha

A known multiple T Tauri system, VW Cha was identified as a binary by Brandner et al. (1996) with 072 separation and a ΔZ = 0.25 mag brightness difference between the companion and the primary at 1 μm. Brandeker et al. (2001) reported an additional close companion to the secondary star with a separation of 010 and suggested that VW Cha is a physical triple system. They found that the primary increasingly dominates the emission at increasingly longer wavelengths and reported a brightness difference of ΔJ = 0.73 and ΔK = 1.32 mag between the primary (A) and secondary (Ba+Bb) components. Daemgen et al. (2013) also detected this close companion of the secondary. Correia et al. (2006) found a wide companion (C) at 168. Melo (2003) suggested that the primary component (A) is a double-lined spectroscopic binary (SB2), and Nguyen et al. (2012) also reported that the primary is a suspected SB2. We will examine this possibility further in Section 4.4.

In order to investigate which component contributes to the observed flux and variability, we examined the Gaia EDR3 measurements. According to these data, the separation between the A and B components is 0664, and the position angle is 185778. The primary component is slightly brighter with a Gaia magnitude of G = 12.40 mag for A and G = 12.77 mag for B; i.e., 58.3% of the total flux comes from the primary component and 41.7% from the companion. In our photometric observations, we did not resolve the primary (A) and secondary (B) components.

Our spectroscopic observations were made with fiber-fed spectrographs. ESPRESSO has a 10 aperture, which means that when the fiber is centered on the primary, the secondary at 072 separation is excluded from the aperture; however, due to varying seeing conditions (see Table 1), we might observe some contamination from the companion. The FEROS instrument has a larger aperture (20), which means that the secondary contaminates our observations. Furthermore, if the primary (A) is an SB2, as suggested by Nguyen et al. (2012) and Melo (2003), both components would have been observed by the ESPRESSO and FEROS spectrographs. In addition, according to Daemgen et al. (2013), the probability of the A component hosting a disk is 1.00, whereas the same probability of the B component is only 0.01. For this reason, we attribute the accretion-related variability to the primary component.

The light curves of VW Cha carry important information on the nature of the variability and its origin. The TESS light curve, displayed with black dots in Figure 2, reveals variations on various timescales with a peak-to-peak amplitude of ∼0.8 mag. Small-amplitude fluctuations appear on timescales of hours, and a brightening event occurs toward the end of our observing period (JD ∼ 2,458,642, i.e., around 2019 June 7), which lasts for a few days. In order to look for any periodic signals in the TESS light curve, we carried out a period analysis. We computed a Lomb–Scargle periodogram (Lomb 1976; Scargle 1982) that reveals a period at P = 9.69 days (bottom panel of Figure 2). We also found a peak around 26 days (f = 0.037 day⁻¹), which suggests a longer-term oscillation in the light curve. No further peaks were found with higher frequencies outside of the indicated frequency range of Figure 2. The 2021 TESS light curve (Figure 3) shows a similar peak-to-peak amplitude (∼0.9 mag); however, the mean brightness of the system slightly increased. We calculated a Lomb–Scargle periodogram from the 2021 TESS light curve as well (Figure 3, bottom panel), which does not confirm the presence of the 9.69 day period but shows a small peak at 9.03 days instead. As the TESS photometry is not resolving the system, the results of the period analysis might be the superposition of the signals produced by all members of the system.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Although the ground-based SMARTS observations have a sparser cadence than the TESS measurements, they carry crucial information on the color variations during our observing period. The I_C -band measurements perfectly follow the variations seen in the TESS light curve with a similar peak-to-peak amplitude. This is expected, as TESS offers broadband photometry that is centered at the I_C band (see Figure 1 in Ricker et al. 2015). The shape of the near-infrared J-, H-, and K-band light curves also follow the shapes of the optical light curves with slightly increasing amplitudes toward the longer wavelengths, i.e., ∼0.6 mag peak-to-peak variations in the J band and ∼0.8 mag variations in the K band.

The ground-based SMARTS data allow us to examine the color variations during the observing period. We show the color–magnitude and near-infrared color–color diagrams in Figure 4 along with the extinction path by Cardelli et al. (1989), indicated with black arrows assuming R_V = 3.1. The near-infrared color–magnitude diagrams show that the pattern does not follow the extinction path in the near-infrared wavelength range, but VW Cha becomes redder as it brightens.

Zoom In Zoom Out Reset image size

We also show the publicly available ASAS-SN g-band measurements ⁸ in Figures 2 and 3. The g-band light curve also resembles the TESS observations. However, two brightening events at JD ∼ 2,458,605 and JD ∼ 2,458,642 stand out from the daily variations. Among the available optical light curves, the g-band observations display the largest peak-to-peak amplitude with ∼1.4 mag variations in the 2019 observing season and ∼1.2 mag in 2021.

The spectra of CTTSs encompass numerous absorption lines; however, these absorption lines are typically weaker than those of the nonaccreting stars. This so-called veiling is normally caused by an excess continuum emission originating from gas at the stellar surface heated by the accretion shock. This effect is detected in the absorption lines of VW Cha, and it is discussed further in Section 3.7.

The absorption lines of VW Cha also show some morphological variations. In order to investigate the origin of these changes, we calculated the average absorption profile using the least-squares deconvolution (LSD) method (Donati et al. 1997). The LSD is a cross-correlation technique for computing average profiles from thousands of spectral lines simultaneously. In order to achieve a better signal-to-noise ratio (S/N), we obtained the LSD profile from the red half of the spectra (λ = 584–788 nm) using a mask with absorption lines for which the normalized flux is less than 0.985 (Figure 5).

Zoom In Zoom Out Reset image size

The primary component of VW Cha is a suspected SB2, and the observed variations might originate from this close binary nature. In order to examine this possibility, we determined the radial velocity of the system by fitting Gaussians to the LSD profiles. The first ESPRESSO observation (JD = 2,458,635.5) shows a single Gaussian profile, whereas the second ESPRESSO measurement (JD = 2,458,642.6) hints at a split profile with two Gaussian components. The FEROS observations are noisier, but, as the first FEROS spectrum was taken on the same night as the second of the ESPRESSO spectra, we applied two Gaussians for this measurement, whereas we used one component for the last epoch. We show the fitted Gaussians in Figure 5 as well.

The results show that the average radial velocity of the system is 16.8 km s⁻¹. For the JD = 2,458,642 night, we found two radial velocity components with a separation of ∼19 km s⁻¹ in both the ESPRESSO and FEROS spectra. The radial velocity results are given in Table 4.

The spectra of VW Cha exhibit several emission lines (Figure 1). The Balmer lines, in particular the Hα and Hβ, are the most striking features of all of the spectra and show strong and variable line profiles. In addition, we identified several permitted lines, such as He i, Fe i and Fe ii, and the Na i doublet in emission; however, not all of them were clearly detected in all epochs. The wavelength region of the Ca ii infrared triplet was covered only by FEROS, and we detected two of the three lines on both FEROS epochs. The 854.2 nm line falls between two orders; therefore, it was not covered. Moreover, [O i] lines were also detected, but we did not find any other forbidden line.

The most conspicuous feature of the spectra of VW Cha is the Hα emission line, which shows significant variability over the observing period (Figure 6). The central peak exhibits amplitude variations that appear to be correlated with the photometric variations. When a brighter photometric state was observed, the Hα line presented a stronger peak, which implies that the Hα line became stronger by a larger factor than the continuum. Apart from the amplitude variations, we also found morphological changes in the Hα line. The first epoch shows a strong blueshifted absorption component around −150 km s⁻¹ that was mostly filled by the second epoch. This effect is also shown by the blue shaded area, which indicates the variance profile. The variance profile measures the variability in each velocity bin in the line profile and was obtained as described in Johns & Basri (1995), using the following relation:

$$\begin{eqnarray*}&&\sum _{v}={\left[\displaystyle \frac{\sum _{i=1}^{n}{({I}_{v,i}-{\overline{I}}_{v})}^{2}}{(n-1)}\right]}^{1/2},\end{eqnarray*}$$

where n is the number of observations, I_v,i is the intensity at a given velocity (v) in each observation, and ${\overline{I}}_{v}$ is the mean intensity of all of the observed profiles at a given velocity v. The variance profile of the Hα line suggests that besides the feature at −150 km s⁻¹, the red wing is also highly variable.

Zoom In Zoom Out Reset image size

The Hβ line exhibits similar amplitude variations as the Hα line; i.e., when the system becomes brighter, the emission gets stronger (see Figure 6). The line has a peculiar double-peaked shape, and the relative amplitudes and positions of the two peaks vary over time. Our first observation indicates one peak at the systemic velocity and one redshifted peak. By the second observation, the peak at the center had moved to the blueshifted side, whereas the redshifted peak remained around 130 km s⁻¹. Our last observation reveals similar peak positions as the first one; however, the redshifted peak shows a more pronounced amplitude decrease than the other.

The He i line at 587.7 nm exhibits a peak at the rest velocity and wide wings on both sides. Besides the amplitude variations of the central peak, the wings are also changing; when a brighter photometric state was observed, the line wings were more pronounced.

We also detected three [O i] lines in all of our spectra at 557.8, 630.2, and 636.3 nm. In Figure 6, we show the 630.2 nm line from the normalized spectra, as this was the strongest among the observed forbidden lines. The region of the [O i] line is contaminated by telluric lines; nonetheless, we observed some amplitude variations in the normalized line profiles. However, we note that these variations appear only in the continuum-normalized spectra; the line flux is almost constant during our observations. We discuss this further in Section 4.3. The Na i doublet shows high variability, with two observations in emission and two measurements below the continuum (Figure 6).

We note that two measurements, the first FEROS spectrum and the second ESPRESSO spectrum, were taken on the same night only a few hours apart. The two spectra look very similar, as expected, but some strong accretion tracer lines, such as the Hα and Hβ lines, show noticeable differences (Figure 6). In order to examine whether this discrepancy is an instrumental or a physical effect, we compared several emission and absorption lines in the two spectra. We found that most lines are in agreement within 10%. Only the strongest lines differ by ∼20%, but as these are typically accretion tracers, the change might have a physical origin.

The H i, He i, Ca ii, and Na i emission lines in the optical and near-infrared spectra of young stellar objects are tracing the accretion process (e.g., Alcalá et al. 2014; Hartmann et al. 2016). Therefore, we computed the accretion rates of VW Cha using the fluxes of these lines detected in our data.

While the FEROS spectra are Nyquist-binned, the ESPRESSO spectra are oversampled. To compute the line flux in a homogeneous way, we rebinned the ESPRESSO spectra in the Nyquist way. At first, we identified the emission lines automatically; we considered those emission lines that were above the 1σ level of the estimated continuum level, and the line width was defined by the difference between the two line edges, where the 1σ level intersects the line (i.e., λ_right − λ_left). However, due to the noise and occasionally occurring blends with neighboring lines, we needed to manually adjust the edges of the lines in some cases or discard the blended lines. We determined the line flux with a python routine that integrates the flux of the pixels contained between the local continuum, which is estimated with a linear fit, and the line. We computed the noise of the line by multiplying the standard deviation of the local continuum (rms) by the width of the line (Δλ_line) and by the square root of the number of pixels included in the line (N_pix). A line is considered to be detected when its S/N ≥ 3. For the nondetected lines, we computed the upper limit of the line flux (F^upp _line) by multiplying the noise by 3:

$$\begin{eqnarray}&&{F}_{{\rm{l}}{\rm{i}}{\rm{n}}{\rm{e}}}^{{\rm{u}}{\rm{p}}{\rm{p}}}=3\times \left({\rm{rms}}\times \displaystyle \frac{{\rm{\Delta }}{\lambda }_{{\rm{l}}{\rm{i}}{\rm{n}}{\rm{e}}}}{R}\times \sqrt{{N}_{{\rm{p}}{\rm{i}}{\rm{x}}}}\right).\end{eqnarray} \tag{ 1 }$$

The results are presented in Table 2.

Table 2. Line Fluxes for the Emission Lines of VW Cha

Species	λline	${F}_{\mathrm{obs}}^{{\rm{ep}}1}(E)$	${F}_{\mathrm{obs}}^{{\rm{ep}}2}(F)$	${F}_{\mathrm{obs}}^{{\rm{ep}}3}(E)$	${F}_{\mathrm{obs}}^{{\rm{ep}}4}(F)$
	[nm]	[10−14 erg s−1 cm−2]	[10−14 erg s−1 cm−2]	[10−14 erg s−1 cm−2]	[10−14 erg s−1 cm−2]
H3	656.32	382.788 ± 0.196	919.390 ± 0.525	726.983 ± 0.236	285.018 ± 0.517
H4	486.16	54.788 ± 0.088	122.292 ± 0.290	105.076 ± 0.080	24.481 ± 0.449
H5	434.07	25.390 ± 0.289	36.937 ± 0.540	41.370 ± 0.221	7.050 ± 0.583
H6	410.2	18.400 ± 0.344	25.994 ± 0.946	25.940 ± 0.280	94.672 ± 5.548
H7	397.03	21.356 ± 0.263	41.300 ± 1.145	40.656 ± 0.205	29.485 ± 2.256
H8	388.93	8.756 ± 0.348	<4.643	13.416 ± 0.286	7.110 ± 1.930
H9	383.56	10.940 ± 0.610	...	11.367 ± 0.465	...
H10	379.81	6.789 ± 1.057	...	13.272 ± 0.748	...
He i/Fe i	492.22	2.404 ± 0.130	14.947 ± 0.276	10.454 ± 0.096	2.681 ± 0.476
He i	402.64	<0.272	<1.687	<0.209	1.057 ± 0.315
He i	447.17	2.377 ± 0.059	3.225 ± 0.220	3.370 ± 0.050	3.524 ± 0.440
He i	471.34	0.776 ± 0.099	0.953 ± 0.195	0.957 ± 0.080	1.581 ± 0.296
He i	501.6	<0.342	19.588 ± 0.245	<0.800	4.593 ± 0.396
He i	587.6	9.293 ± 0.098	25.047 ± 0.206	19.603 ± 0.088	7.345 ± 0.275
He i	667.85	2.888 ± 0.044	8.839 ± 0.144	6.075 ± 0.049	2.546 ± 0.126
He i	706.56	2.729 ± 0.056	12.545 ± 0.150	6.786 ± 0.049	7.996 ± 0.190
He ii	468.61	1.101 ± 0.092	<0.998	1.127 ± 0.075	<2.063
Ca ii (K)	393.39	19.906 ± 0.324	40.339 ± 1.563	46.111 ± 0.258	21.368 ± 2.280
Ca ii (H)	396.87	20.968 ± 0.548	40.279 ± 1.172	40.079 ± 0.451	30.425 ± 2.240
Ca ii	849.85	...	252.089 ± 0.380	...	25.077 ± 0.456
Ca ii	866.26	...	232.491 ± 0.824	...	26.310 ± 0.642
Na i	589.03	<0.289	8.470 ± 0.445	4.650 ± 0.160	2.190 ± 0.284
Na i	589.63	<0.291	4.467 ± 0.398	2.105 ± 0.182	1.266 ± 0.213
O I	777.35	7.205 ± 0.143	29.358 ± 0.503	18.012 ± 0.150	5.970 ± 0.340
O I	844.68	...	54.033 ± 1.145	...	10.570 ± 0.543

Note. Fluxes shown here are not dereddened.

Download table as:  ASCIITypeset image

Pre-main-sequence stars are typically affected by extinction. In order to study the stellar parameters and accretion rates, we need to estimate the extinction, which can be computed in several ways. One of these is through the distance between the location of the star in the [J − H] versus [H − K] diagram and the CTTS locus, the location of the CTTSs if they are unextincted (A_V = 0). This color–color diagram is shown in Figure 7. Applying the extinction vector of Cardelli et al. (1989) and assuming R_V = 3.1, we found a mean value for the extinction of A_V = 1.70 ± 0.10 mag using all of our near-infrared photometry.

Zoom In Zoom Out Reset image size

The extinction of an accreting object can also be computed from the accretion luminosity estimated by accretion tracers at different wavelengths. We estimated the extinction from a grid of values from 0 to 10 mag, choosing the A_V that minimizes the difference between the accretion luminosity derived from different lines for the same epoch. We used only the detected lines (see Table 2). We obtained a mean value for the four epochs of A_V = 1.54 ± 0.27 mag.

Lastly, we performed the spectral typing of VW Cha by comparing our spectra with a grid of observed templates (Manara et al. 2013, 2017a) from G4 to M9.5 that have a typical step of one spectral subclass for spectral types G and K and 0.5 spectral subclass for M-type stars, a method similar to those performed previously in the literature for single CTTSs (e.g., Fiorellino et al. 2021). In order to compare the template with the spectra of VW Cha, we reddened the template and flux-calibrated it in a window of Δλ = 5 nm around λ = 570 nm. By varying the template and extinction and matching the shape and molecular features of our data with the ones of the templates, we find that the spectral type of VW Cha is in agreement with the K5−K7 templates. We find that the mean value of the extinction for the four epochs is A_V = 1.15 ± 0.53 mag. We should point out that, for this analysis, we treated VW Cha as if it was a single star.

We note that our A_V results are all in agreement within the errors. In the following, we use the value provided by the [J – H] versus [H – K] diagram, for which have the smallest error, and it is closer to the extinction values we found in the literature, for example, A_V = 1.9 mag in Manara et al. (2017b). We think that the slightly different result we get from our spectra can have two origins; one in that we have not subtracted a slab model from the spectra to eliminate the effect of the veiling while doing the spectral typing. The other is that we analyzed only the optical wavelength range, whereas Manara et al. (2017b) covered the range from the UVB to the near-infrared.

We used the detected accretion tracer lines from Table 2 to estimate the accretion luminosity (L_acc) of the VW Cha system. The accretion luminosity is related to the line luminosity of the accretion tracers (L_line) by empirical relations. We computed the line luminosity as L_line = 4π d² F_line, where F_line is the extinction-corrected line flux, and d = 194.5 pc is the distance (Bailer-Jones et al. 2021). Then, we estimated the accretion luminosity for each line by using the relations from Alcalá et al. (2017),

$$\begin{eqnarray}&&{\mathrm{log}}_{10}\left(\displaystyle \frac{{L}_{\mathrm{acc}}}{{L}_{\odot }}\right)=a{\mathrm{log}}_{10}\left(\displaystyle \frac{{L}_{\mathrm{line}}}{{L}_{\odot }}\right)+b,\end{eqnarray} \tag{ 2 }$$

where a and b vary from line to line. The relations in Alcalá et al. (2017) are based on spectroscopic observations of a sample of young stellar objects. They have some natural uncertainty originating from the measurement of the line flux and the linear fit between the line and accretion luminosities. The overall uncertainties for the relations differ from line to line, and the standard deviation of the liner fit ranges from σ = 0.26 to 0.45 among those relations that are suggested for deriving the accretion luminosity. The error on each line is obtained by propagating the error of F_line. As the best estimate of the accretion luminosity for each epoch, we used the mean value of all of the accretion luminosity provided by the detected lines of that epoch. The error on the mean accretion luminosity for each epoch was computed by dividing the standard deviation of the errors by the square root of the number of lines used. The results are shown in Table 3.

Table 3. Accretion Parameters for VW Cha

Date	$\mathrm{log}{L}_{\mathrm{acc}}$	$\mathrm{log}{\dot{M}}_{\mathrm{acc}}$
	[L⊙]	[M⊙ yr−1]
2019 May 31 (E)	0.022 ± 0.083	−6.785 ± 0.075
2019 Jun 8 (F)	0.360 ± 0.086	−6.447 ± 0.082
2019 Jun 8 (E)	0.225 ± 0.079	−6.582 ± 0.075
2019 Jun 11 (F)	−0.103 ± 0.082	−6.910 ± 0.082

Note. The values presented here are without taking into account the error on stellar parameters.

Download table as:  ASCIITypeset image

We computed the mass accretion rate with the following equation:

$$\begin{eqnarray}&&{\dot{M}}_{\mathrm{acc}}\sim {\left(1-\displaystyle \frac{{R}_{\star }}{{R}_{\mathrm{in}}}\right)}^{-1}\displaystyle \frac{{L}_{\mathrm{acc}}{R}_{\star }}{{{GM}}_{\star }},\end{eqnarray} \tag{ 3 }$$

where R_in is the inner disk radius, which we assume to be R_in ∼ 5 R_⋆ (Hartmann et al. 1998), ${M}_{\star }=0.75{\pm }_{0.35}^{0.5}\,{M}_{\odot }$, and R_⋆ = 2.55 ± 0.29 R_⊙ from Daemgen et al. (2013). We computed the error on the mass accretion rate in the same way as for the accretion luminosity and not taking into account the uncertainty on the stellar parameters. The results are shown in Table 3. Figure 8 displays the accretion luminosity and mass accretion rate as a function of time, which shows an increased accretion luminosity and rate in the second and third epochs and a decrease by the last epoch.

Zoom In Zoom Out Reset image size

The absorption spectra of CTTSs are often veiled by additional continuum emission due to accretion. As the veiling is also indicative of the accretion rate, we expect that when the accretion rate changes, the veiling and stellar brightness change accordingly. We calculated the veiling (V) for our four spectroscopic observations using the following relation:

$$\begin{eqnarray*}&&{W}_{0}^{\mathrm{eq}}={W}^{\mathrm{eq}}(V+1),\end{eqnarray*}$$

where W^eq and W^eq ₀ are the measured equivalent width of the absorption line of the target and a reference object, respectively.

Instead of using one absorption line, we used the LSD profiles previously obtained from the spectra. For reference, we obtained the LSD profile based on the same criteria for a nonaccreting K7-type weak-line T Tauri star, TWA 6, whose spectrum is available from the Polarbase database. ⁹ We used this reference in order to calculate the absolute veiling. The results are shown in Figure 9.

Zoom In Zoom Out Reset image size

It must be noted that the veiling is expected to be wavelength-dependent in a way that typically larger veiling is measured at shorter wavelengths (e.g., Dodin & Lamzin 2013). The use of the LSD profile ignores this wavelength dependence, as it serves as an average absorption line profile. In order to test the significance of its effect, we selected a few individual absorption lines above 500 nm to avoid the noisiest part of the spectra and calculated the veiling using them. Our results show that, within the uncertainties, the general temporal trend is similar to the bulk measurement provided by the use of the LSD profiles. However, the measurements using individual lines did not reproduce the expected trend (i.e., larger veiling at shorter wavelengths) but instead resulted in a scatter around the values determined by using the LSD profile. This might be due to the fact that in many cases, the individual absorption lines were still noisy and weak; furthermore, the stronger ones were often blended with neighboring lines.

Our ground-based near-infrared observations of VW Cha reveal an unusual trend: the source becomes redder as it brightens (Figure 4). In order to examine the physical origin of this behavior, we compared our results with three models described by Carpenter et al. (2001) that use the models by Meyer et al. (1997).

The first model involves starspots, which often appear on the surface of T Tauri stars and can be responsible for the variability observed in these systems. Carpenter et al. (2001) inspected the impact of starspots at near-infrared wavelengths and found that cool spots with small fractional coverage result in nearly colorless fluctuations. In the case of larger spot coverages and hot spots, the color–magnitude diagram shows that the system becomes bluer as it gets brighter.

Inhomogeneities in the inner circumstellar environment can cause changing extinction, which also results in photometric variability. The extinction model introduces a positive slope in the near-infrared color–magnitude diagrams; i.e., the object gets redder as it becomes fainter. In order to compare our observations with this model, we indicated the extinction arrow in Figure 4 corresponding to A_V = 1 mag, which shows that our data are not consistent with changing extinction.

The disk model discussed by Carpenter et al. (2001) and Meyer et al. (1997) attributes the near-infrared variation to changes in the accretion disk, such as variable inner disk structure or accretion rate. This model results in a negative slope in the near-infrared color–magnitude diagrams (i.e., the source gets redder as it brightens) and predicts a shallower slope in the [J − H] versus [H − K_S ] diagram than the abovementioned spot and extinction models. In order to compare the disk model with our observations, we indicated the color changes predicted by this model compared to our bluest data point with triangles in Figure 4. The open triangles represent a mass accretion rate of 10^−8.5 M_⊙ yr⁻¹, and the filled triangles show 10^−7.0 M_⊙ yr⁻¹ (Carpenter et al. 2001). The three triangles correspond to disk inner hole sizes of 1, 2, and 4 R_⊙. These models of Meyer et al. (1997) include a fiducial M0-type star with fixed M_⋆ = 0.5 M_⊙, R_⋆ = 1.8 R_⊙, and T_eff = 4000 K. These stellar parameters describe a typical T Tauri star but do not agree entirely with the parameters of VW Cha; therefore, our analysis gives a qualitative explanation for the observed trajectory. The comparison suggests that the near-infrared color variations seen in our data set are orthogonal to the trajectory of the spot and the extinction models, and they are consistent with those predicted by the disk model. This means that the observed color variations originate from the changes in the accretion disk.

We also considered another set of models by D'Alessio et al. (1998, 1999, 2001) that provides the spectral energy distributions (SEDs) for accretion disks with various disk parameters ¹⁰ (R_disk, R_hole, ${\dot{M}}_{\mathrm{acc}}$, α, i) for systems with different central stars. First, we selected a central star with stellar parameters closest to our target (T_eff = 4000 K, 1 Myr, R_⋆ = 2.64 R_⊙, M_⋆ = 0.7 M_⊙, L_⋆ = 1.6 L_⊙), and we studied the corresponding disk models. From the available models, we chose the ones with R_disk = 100 au, i = 30°, and α = 0.01, which parameters are expected for VW Cha. This left two parameters free: R_hole and ${\dot{M}}_{\mathrm{acc}}$. Unfortunately, the R_hole and ${\dot{M}}_{\mathrm{acc}}$ grids of these models are sparser than the ones presented by Meyer et al. (1997); therefore, these models are less representative of VW Cha. Nonetheless, we attempted to compare our data and the models by placing our observations on the SEDs of the models with the following pairs of parameters: R_hole = 9 R_⋆ and ${\dot{M}}_{\mathrm{acc}}={10}^{-9}\,{M}_{\odot }\mathrm{yr}$ ⁻¹, R_hole = 9 R_⋆ and ${\dot{M}}_{\mathrm{acc}}={10}^{-8}\,{M}_{\odot }{\mathrm{yr}}^{-1}$, R_hole = 11 R_⋆ and ${\dot{M}}_{\mathrm{acc}}={10}^{-7}\,{M}_{\odot }\mathrm{yr}$ ⁻¹, and R_hole = 22 R_⋆ and ${\dot{M}}_{\mathrm{acc}}={10}^{-6}\,{M}_{\odot }\mathrm{yr}$ ⁻¹. This comparison did not lead to any strong conclusions, which might arise from the fact that both the inner disk hole size and the accretion rate vary from model to model, and there is no option to change only one of these parameters. Additionally, we converted the fluxes close to the JHK bands from the SED models to the magnitudes, and we constructed the near-infrared color–magnitude and color–color diagrams. In general, we found that these models also reproduce the observed trend (i.e., the source becomes redder when it brightens). However, the trend appears in the opposite way compared to the Carpenter et al. (2001) models. In Carpenter et al. (2001), larger hole sizes correspond to smaller infrared excess, whereas the D'Alessio et al. (1998) models resulted in an opposite trend; i.e., larger hole sizes correspond to larger infrared excess. Again, this might arise from the fact that for the D'Alessio et al. (1998) models, both the inner disk hole size and the accretion rate change from model to model (larger inner hole size is paired with larger accretion rate). This does not contradict the Carpenter et al. (2001) models; their models with larger accretion rates (filled triangles) also predict redder colors.

Apart from changes in the inner disk hole size, other changes in the inner disk structure may contribute to the observed trend in the near-infrared color–magnitude and color–color diagrams. These involve, e.g., variations in the thickness of the inner disk edge of a warped disk (Mahdavi & Kenyon 1998; Lai 1999; Terquem & Papaloizou 2000), or the SB2 nature (see Section 4.4) might also disturb the inner disk edge and cause structural changes.

The light curves of VW Cha display significant photometric changes; we observed 0.6−0.9 mag peak-to-peak variability in the different photometric bands in 2019. This dynamical range is dominated by the brightening event at JD ∼ 2,458,642. We measured a spectrum before, during, and after this event, which allows us to study its nature and origin.

The spectra taken at the bright and faint states show noteworthy differences. The Balmer lines undergo the most striking evolution; however, other accretion tracers also vary in time. The significant amplitude variations of these lines in the normalized spectra suggest that accretion rate change might be the major contributor of the observed effect. The computed line fluxes also support this concept, since we measured higher line fluxes in the brighter state. The veiling measurements (Figure 9) are also consistent with the described picture; i.e., the veiling was higher during the brighter state, indicating a higher accretion rate. Indeed, we calculated the accretion rates for each epoch, which show an increase by a factor of 2.17 by the brighter state (Table 3).

The emission lines exhibit morphological evolution as well during the previously discussed brightening event. The He i lines show only mild changes; however, the wings become more pronounced in the brighter epoch. In addition, the variance profile highlights a slight asymmetry; the red wing varies more during our observations than the blue wing. This broad component extends to ∼250 km s⁻¹ and is thought to form in the magnetospheric flow (Hartmann et al. 2016). The more pronounced broad component in the brighter state might arise due to the increased accretion. The Hβ line exhibits an interesting double-peaked profile where both the strength and position of the two components are changing.

The Hα line has the most striking morphological differences between the different measurements. The first epoch shows an emission peak with a strong blueshifted absorption component. This suggests that the accretion is accompanied by an outflow, which is discussed in Section 4.3. As the accretion rate increases by the second observation, this blueshifted absorption component becomes filled.

Stauffer et al. (2014) examined the light curves and spectra for a sample of young stellar objects in NGC 2264, including several CTTSs. They studied light curves dominated by accretion bursts, similar to that of VW Cha, and found that these brief—several hours to a day long—brightenings have amplitudes in the range of 5%–50% of the quiescent level. The longer duration (several days) of the brightening event at JD ∼ 2,458,642 in the light curve of VW Cha and the fine structure, revealed by the high-cadence TESS data, suggest that this event might be a superposition of multiple smaller accretion bursts. The spectral features that we found are also consistent with the results of Stauffer et al. (2014); i.e., accretion burst stars have a modestly structured and centrally peaked Hα line during the accretion burst and the 667.8 nm He i line in emission.

In Figure 10, we compare VW Cha with other sources from the Chamaeleon I star-forming region using the results of the accretion rates and stellar parameters of Manara et al. (2019), and we use the fit from Fiorellino et al. (2021). We depict the accretion luminosity against the stellar luminosity in the top panel and the accretion rate against the stellar mass in the bottom panel. Our results show that, as we expected based on previous studies, VW Cha has a high accretion luminosity and accretion rate compared to the other members of this sample and places VW Cha at the high end of the distribution. While our new results match perfectly with the fitted relation for Cha I in the L_acc versus L_⋆ diagram, we note that the ${\dot{M}}_{\mathrm{acc}}$ is larger than the trend fitted for the overall region.

Zoom In Zoom Out Reset image size

Mass ejection is often observed in CTTSs. Strong winds and outflows are thought to be accretion-powered, and they lead to forbidden line emission and often blueshifted absorption components superimposed on background emission (Calvet 1997; Bally 2016).

Object VW Cha shows signs of outflow in the form of forbidden [O I] emission lines and a blueshifted absorption component of the Hα line. This absorption component, however, is not always present. The reason for this is the fact that it is superimposed on background emission arising due to the magnetospheric accretion. When the accretion rate increases, the accretion-powered emission component strengthens and fills the absorption component.

The [O i] emission lines are also detected in the spectra at 557.8, 630.2, and 636.3 nm. They appear to be variable based on the normalized line profile (Figure 6, bottom middle panel); however, this line is expected to be intrinsically stable on a timescale of days (Petrov et al. 2011), and the observed effect might be due to the fact that during the brighter state, the continuum level rose. Indeed, we calculated the line fluxes of the detected forbidden oxygen lines, which differ by only a few percent, suggesting that the normalized line profiles change due to continuum variations.

The observed line profile carries information on the kinematics of the emitting gas. In the case of VW Cha, the [O i] exhibits one low-velocity component (FWHM ∼ 30–60 km s⁻¹) that is slightly blueshifted compared to the systemic velocity. The low-velocity component with peak velocity v_p ≲ 30 km s⁻¹ is thought to be associated with extended disk winds (Kwan & Tademaru 1995; Pascucci et al. 2020). For a sample of T Tauri stars, Gangi et al. (2020) examined the component at the lowest peak velocity of the [O i] line, and our results for VW Cha are consistent with the typical line kinematic parameters that Gangi et al. (2020) found, suggesting that it originates from the innermost few au of the system.

The primary component of the VW Cha system is a suspected SB2. Melo (2003) observed line doubling based on their cross-correlation function and reported 15.31 km s⁻¹ as a mean radial velocity for the system. Nguyen et al. (2012) found evidence for two components of the primary separated by 20 km s⁻¹.

We determined the radial velocities by fitting the LSD profiles and report the results in Table 4 along with the radial velocities from Nguyen et al. (2012). In the cases where we were able to resolve the two components, our results are in agreement with Nguyen et al. (2012), where they found that their spectra of the primary show evidence for two components (Aa, Ab) separated by 20 km s⁻¹ (see their Figure 10.10). As in our observations, we found one component in the first epoch, two components in the second and third epochs, and one component in the fourth epoch 10 days later; we speculate an orbital period for the SB2 of ∼10 days. In case the primary is indeed an SB2, we need to account for the possibility that both components contribute to the observed spectroscopic variations in our new measurements. Frasca et al. (2021) examined a similar multiple system, CVSO 104, where they confirmed that the target is a double-lined SB2. They determined the mass accretion rates of the two components from the fluxes of the lines where the profiles of the two components could be deblended. In our case, the individual components of the accretion tracers, such as the Hα or He i lines, appear to be blended (see Figure 6), similar to the Hα line presented in Frasca et al. (2021); therefore, we were not able to determine the accretion rates for the individual components. The Hβ is the only line that shows a double-peaked line profile; however, the peak velocities do not correspond to the radial velocities that we determined for each epoch.