Luminous and Variable Stars in NGC 2403 and M81^*

Most of our targets were selected from the Zickgraf & Humphreys (1991) catalogs for NGC 2403 and M81 based on their apparent magnitude and color. We emphasized those most likely to be luminous early-type stars and suspected variables. To select additional candidates, we used aperture photometry measured from the Hubble Legacy Archive (HLA) Advanced Camera for Survey (ACS) images of NGC 2403 and M81 in the ANGRRR and ANGST programs.¹ Known variables from Tammann & Sandage (1968) and several stars with previously observed spectra (Humphreys & Aaronson 1987b; Zickgraf et al. 1996; Sholukhova et al. 1998b, 1998a) were also included. For the fiber assignment with the MMT/Hectospec, targets were ranked based on their apparent magnitudes, colors, previous spectra suggesting that they may be supergiants, and on their variability. Since one of our goals is to identify stars that may be candidates for high mass loss, we used the LBT nearby galaxy survey (Kochanek et al. 2008; Gerke et al. 2014) to initially identify candidates for variability as described in Grammer et al. (2015). We identified targets as potentially variable if their rms variability is greater than their median photometric error. The brightest stars, based on their apparent V magnitude with clear indications of variability, received the highest priority, rank 1, for spectroscopy with the Hectospec. The light curves for several of these stars are shown and discussed in Section 3.

Following these criteria, we selected 124 stars in NGC 2403 with V magnitudes between 18.0 and 20.1, and 91 in M81 with V magnitudes between 18.5 and 20.1. Eighty-six in NGC 2403 were assigned fibers in two separate pointings, fields F1 and F2, and 61 stars in M81 were assigned fibers.

The spectra were observed with the Hectospec, a multi-object spectrometer mounted on the MMT (Fabricant et al. 1998, 2005). The Hectospec² is a fiber-fed MOS with a 1° FOV and 300 fibers; each fiber subtends 15 on the sky. We used the 600 mm⁻¹ grating with the blue tilt centered on 4800 Å and the red tilt centered on 7300 Å. The red tilt was chosen to include Hα plus the Ca ii triplet at ∼8500 Å. The 600 mm⁻¹ grating gives a spectral coverage of ∼2500 Å with 0.54 Å pixel⁻¹ resolution. The journal of observations is given in Table 1.

The NGC 2403 targets were observed in 2012 October and December in two fields, F1 and F2. The total exposures times in the blue were 4H and 4H 15M for F1 and F2, respectively, and for the red spectra, 3H for F1 and 2.5H for F2. The two pointings overlapped so that 27 stars were in common yielding total integration times of 8.25H in the blue and 5.5H in the red for these stars. Unfortunately, the M81 observations were plagued by poor observing conditions. Consequently, the spectra were acquired over two observing seasons in 2012 and again in 2014. The three best sets of blue spectra were combined to give a total integration time of 6.75H. Since two of the data sets were separated in time, the spectra provide an opportunity to check for spectroscopic variability before being combined. The red spectra were likewise observed in the two seasons for a total integration time of 4.5H.

The spectra were reduced using an exported version of the CfA/SAO SPECROAD package for Hectospec data E-SPECROAD.³ The spectra were bias subtracted, flat-fielded, wavelength calibrated, and sky subtracted. The reduced spectra are available at http://etacar.umn.edu/LuminousStars/NGC2403M81/.

Although our selection criteria favored the visually brightest stars of spectral types A and F, we identified 10 hot or emission-line stars in NGC 2403 and six in M81. Three B[e] supergiant candidates in M81 and two WN stars in N2403 are described here.

The B[e] supergiants share many spectral characteristics with LBVs (Humphreys et al. 2017c) including prominent Fe ii and [Fe ii] emission. In a recent spectroscopic survey of emission-line stars, Aret et al. (2016) designated [O i] λλ6300,6364 emission as one of the characteristics of the sgB[e] class. [O i] emission is not observed in confirmed LBVs and can be used to separate the two types (Humphreys et al. 2017c). But these lines are also present in the night sky spectrum and in H ii regions. For faint stars in N2403 and M81, this can be a problem, with contaminating nebulosity in the aperture and residual or poor sky subtraction. For that reason, we rely on the velocity of the [O i] lines to identify them with the star, i.e., if they have the same velocity as the He i and the Fe ii lines presumably formed in the circumstellar ejecta, although they may be nebular in origin.

With this criterion, we identified three possible sgB[e] stars in M81: 10584-4.1, 10584-8.4, and 10584-9.1. Their blue and red spectra are shown in Figure 1. All three stars have prominent Balmer emission with P Cyg profiles and Fe ii and Fe[ii] lines in the blue plus the λ6300,6363 [O i] lines in the red. He i is also in emission with P Cyg profiles in all three stars. Thus, they all show evidence for stellar winds and mass loss. Their outflow velocities, measured from the absorption minima in their P Cygni profiles relative to the emission-line peak indicate moderate wind speeds of 170–180 km s⁻¹ for 10584-8.4 and 10584-9.1 and 250 km s⁻¹ for 10584-4.1. These velocities, measured in the same way, are typical of other sgB[e] stars as well as LBVs (Humphreys et al. 2016). We note that 10584-4.1 and 10584-9.1 are spectroscopically very similar. Both also have broad Thomson scattering wings on their Hα and Hβ emission profiles.

Zoom In Zoom Out Reset image size

The spectrum of 10584-8.4 shows several absorption lines including some He i lines, which, together with other lines such as Mg ii, λ4481, permit us to estimate a late B/early A spectral type for this star. 10584-8.4 also has the [Ca ii] doublet at λλ7291,7324 in emission, another characteristic of some of the sgB[e] stars, shared with the warm hypergiants (Humphreys et al. 2017c). The Ca ii triplet line near λ8500 can also be seen in emission at the red edge of the spectrum. It shows a split profile, which could be due to a bipolar outflow or rotation, a characteristic also observed in some sgB[e]'s. The peaks are separated by 108 km s⁻¹. 10584-4.1 and 10584-9.1 are apparently much warmer stars. There are no obvious absorption lines in either spectrum. In addition to strong He i, the OI lines at λ7774 and λ8446 are also in emission in 10584-4.1.

Light curves for 10584-4.1 and 10584-9.1 are shown in Figure 2, but there is insufficient data for 10584-8.4. 10584-4.1 shows significant variability over five years by 0.2 to 0.4 mag in the U, V, and R bands. 10584-9.1 declined by about 0.5 mag from 2011 to 2013. Thus, both stars are variable.

Zoom In Zoom Out Reset image size

We also identify two late WN stars in N2403: 10182-pr-9 and ZH 2016. The blue spectra of both show prominent broad nitrogen emission features from 4630 to 4700 Å and from 5670 to 5700 Å with strong H and He i emission lines. He ii λ 4686 emission is present in 10182-pr-9.

The A- and F-type supergiants are the visually brightest stars. They define the upper luminosity boundary in the HR diagram for evolved post-main-sequence massive stars with initial masses typically less than 40–50 M_⊙ (Humphreys & Davidson 1979, 1994). Many stars that lie near this boundary show evidence for high and episodic mass loss in their spectra and spectral energy distributions (Humphreys et al. 2013). Blue and red spectra of two luminous intermediate-type supergiants are shown in Figure 3.

Zoom In Zoom Out Reset image size

ZH 553 in N2403 is a late A-type supergiant of high luminosity. It is known as star IVa28 in previous publications (Humphreys & Aaronson 1987a, 1987b). Its red spectrum shows Hα plus the [N ii] and [S ii] nebular lines in emission. The Hα profile is asymmetric to the red with broad wings characteristic of Thomson scattering in its wind plus P Cygni absorption. The star is thus experiencing mass loss with a moderately slow outflow velocity of 98 km s⁻¹ measured from the P Cyg absorption minimum relative to the emission peak. The nebular lines appear to be double peaked. We noticed this in several stars in our similar survey of stars in M101 (Grammer et al. 2015), which in M101 we suspected may be due to emission from two sides of large H ii regions or from more than one emission region along the line of sight. Since ZH 553 does not have strong [O iii] nebular emission lines in the blue, the double peaks may rise either from contaminating emission in the aperture plus nebular emission from its circumstellar ejecta, or from the two sides of its expanding ejecta. The velocity difference between the blue and red components average 56 km s⁻¹. There are no other emission lines in the blue or red spectra.

S94 was originally listed by Sandage (1984a) as one of the brightest resolved objects in N2403. It has the spectrum of a high-luminosity F-type supergiant. The red spectrum shows Hα plus the [N ii] and S[ii] nebular lines in emission, and [O iii] λ5007 is visible in emission in the blue spectrum. Hα has a P Cygni absorption feature. The absorption minimum relative to the peak emission has an expansion velocity of 143 km s⁻¹ indicating a slow wind and mass loss typical of luminous intermediate-type supergiants.

Both ZH 553 and S94 have marginal variability of 0.1 mag or less. Their Balmer emission lines with P Cyg profiles and evidence for mass loss are typical of luminous intermediate-type supergiants, but neither show evidence for circumstellar dust (Section 4.1). We do not consider either to be an LBV/S Dor candidate.

The Tammann & Sandage (1968) survey identified several irregular blue variables in N2403. The most famous is V12, also known as SN 1954J, which had a non-terminal giant eruption. Another, V37 also received a supernova designation as SN 2002 kg due to what was soon recognized as a typical LBV/S Dor high mass loss or maximum light event. In a recent paper on these two "impostors", we discussed the spectrum and light curve of V37 in some detail and showed that its progenitor was an evolved massive star of ∼60 M_⊙ (Humphreys et al. 2017b). V12 or SN 1954J survived its giant eruption and is now obscured by circumstellar dust from that eruption. Our spectral analysis revealed that V12 is actually two stars: a ∼20000 K star, which is the probable progenitor and survivor of the giant outburst, plus a G-type supergiant close neighbor or companion. Interestingly, we find that the hot star was initially only about 20 M_⊙ and the G supergiant of slightly lower mass. The HR Diagram for V12 and V37 and their stellar environments is discussed in Section 5.

Here we include spectra of two additional blue variables in N2403: V38 and V52. Based on the low-resolution spectrum of V38 shown in Humphreys & Aaronson (1987b), we suggested that it was an LBV or LBV candidate. Our higher-resolution blue and red spectra and their variability now confirm that designation. The blue spectrum (Figure 4) shows strong nebular and Balmer emission with He i and Fe ii emission lines. N ii absorption lines are present at λ5660–5680 Å. Hα has broad wings, and a P Cyg absorption feature is present at Hβ. The [O i] emission lines in the red spectrum at λλ6300,6363 are also present, at the same velocities as the nebular lines. Thus, we suspect that they are nebular in origin, and the star is not an sgB[e]. Its light curve in Figure 5 shows short-term variability of a few tenths of a magnitude.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

V52, however, is a foreground late F-type dwarf (Figure 4). Its variability reported by Tammann & Sandage (1968) was marginal, and the LBT survey did not show any variability.

Sandage (1984b) identified six irregular blue variables in M81. We observed four of them: I1 (ZH 244), I2(ZH 364), I8(ZH 1406), and I3. I3 is a foreground F5 dwarf. I2 was observed in our earlier study (Humphreys & Aaronson 1987b), and based on its low-resolution spectrum, we considered it a candidate LBV(LBVc). Our new blue and red spectra (Figure 6) reveal a complex spectrum with emission lines of H, He i, and Fe ii and [Fe ii]. Strong nebular emission lines are also present. The nebular lines have a somewhat different average velocity of ≈−130 km s⁻¹ compared to the He i, H and Fe ii emission with velocities of −80 to −100 km s⁻¹. The [O i] lines at 6300 Å have velocities of −125 km s⁻¹, so we assume that they are nebular in origin and are not from the star.

Zoom In Zoom Out Reset image size

I8 is an F5 supergiant. Its blue spectrum does not show any emission lines. The red spectrum has strong nebular emission but no other emission lines. I8's variability is marginal. Thus, we do not consider it an LBV or candidate. It is worth noting that during their high mass loss or dense wind state, LBV/S Dor variables have absorption line spectra that resemble A or F-type supergiants due to their optically thick cool winds. Thus the "less luminous" LBV/S Dor variables (Humphreys et al. 2016), when in "eruption", overlap the position of the normal luminous intermediate temperature supergiants on the HR Diagram, but LBVs in "eruption" also have strong H emission with prominent P Cyg profiles and Fe ii emission.

The blue spectrum of I1 has a low signal-to-noise ratio (S/N), limiting the accuracy of the classification, but its lines are consistent with a late B- or early A-type supergiant. The only emission line is Hα with wings that are asymmetric to red. I1 apparently has a stellar wind, but there are no P Cyg features or other emission lines. Its light curve (Figure 7) however shows a pattern of smooth variability over five years. It may be similar to M33C-4640 (Humphreys et al. 2016, 2017c), a candidate for post-RSG evolution.

Zoom In Zoom Out Reset image size

ZH 354 in M81 is another emission-line star that we include here as a candidate LBV (Figure 6). The [O i] λλ6300,6363 lines are present, but their near-zero velocities compared with −220 to −250 km s⁻¹ velocities of the other emission lines confirm that they are residual night sky lines. ZH 354 also has features in common with the Of/WN stars. Its spectrum shows strong Balmer emission, nitrogen emission in the λ4600 to 4700 Å region with He ii λ4686 in emission. There is no obvious He i emission in the rather low-S/N spectrum. Hα has very broad wings but with no P Cygni absorption. Its light curve shows only marginal variability of ±0.1 mag over five years. Many LBVs in their quiescent state have Of/late WN features, so it is possible that ZH 354 is an LBV in quiescence.

The light curves of I2 (ZH 364), I1 (ZH 244) and ZH 354 are shown in Figure 7.

Based on our spectra and the available light curves, V37 and V38 in N2403 are confirmed LBV/S Dor variables, and I2(ZH 364) and ZH 354 in M81 should be considered candidate LBVs.

In addition to our previous papers (Humphreys & Aaronson 1987a, 1987b), Sholukhova et al. (1998b, 1998a) observed several of the stars listed by Sandage (1984a, 1984b) plus others from Zickgraf & Humphreys (1991) and Zickgraf et al. (1996). Based on their classification of several as LBV candidates, we included them on our program but found that most are not LBVs. As already described above, S94 is a very luminous F-type supergiant but not an LBV candidate while V52 is a foreground dwarf. For convenience and ease of comparison, we list the stars in common with previous work in Table 4, together with our classifications in this paper.

Despite the limitations of the multi-wavelength photometry and especially the lack of infrared data for many of the stars, we show a selected sample of SEDs in Figure 8, specifically of stars of interest such as the LBV candidates, the sgB[e]'s, and and others with possible circumstellar dust.

Zoom In Zoom Out Reset image size

The SEDs for the two LBV candidates in M81 are shown in the top panel. Since these are strong emission-line stars, their photometry is corrected for interstellar extinction using the mean A_v. ZH 364(I2) has a near-infrared excess that we attribute to free–free emission. Its spectrum shows a strong Hα emission line with broad wings. We also note the raised photometric points in its SED in the R-band due to Hα and in the U-band possibly due to continuum emission. Thus, its near-infrared excess is most likely due to free–free emission and not warm dust. We show Planck curve fits to their corrected broadband data to estimate the temperature shown and integrated to derive a luminosity (M_Bol) in Table 6. However, the temperature for ZH 354 from the fit to the LBT/LBC photometry is inconsistent with the much higher temperature implied by its emission lines such as He ii in its spectrum (Figure 6). It is possible that the extinction correction is much larger than the adopted mean, or the LBC photometry is identified with the wrong star. Its luminosity derived from the SED is not used for this reason.

The SEDs for two sgB[e] stars with infrared data are shown in the middle panel. Their mid-infrared fluxes demonstrate the presence of significant circumstellar dust as found for many sgB[e]'s in other galaxies (Kraus et al. 2014; Humphreys et al.2017c). Although the optical photometry for 10584-8.4 in M81 is limited, its SED exhibits a large circumstellar excess due to dust plus extensive circumstellar gas revealed by the [Ca ii] and Ca ii emission lines in its red spectrum, Section 3.1. The mid-infrared fluxes may seem high or elevated with respect to the visual photometry; however, this strong infrared signature for circumstellar dust is not uncommon for sgB[e]'s (Humphreys et al. 2017a, 2017c). Although the absorption lines in its blue spectrum suggest a late B/early A classification, the Planck fit to the nonuniform optical photometry yields a temperature of 21700 K. We consider this result doubtful, though, because the fit is dominated by the uncertain U band point from the LBT/LBC imaging. Using only the HST magnitudes, the best fit yields 10900 K. 10584-9.1 is a much hotter star as indicated both by is spectrum and SED. The Planck fit to the optical photometry suggests a temperature of 18000 K. Its mid-infrared excess may be a combination of free–free emission and dust.

We also show the SEDs for two luminous intermediate-type supergiants in the bottom panel. 10584-8.1 may have circumstellar dust, although its spectrum did not show any stellar wind emission lines. Planck curve fits to their optical photometry are shown.

The temperatures and derived luminosities estimated in this way are used to place the emission-line stars on the HR Diagram discussed in the next section. We note, as per usual, that Planck curves are only rudimentary approximations.