A FIVE-YEAR SPECTROSCOPIC AND PHOTOMETRIC CAMPAIGN ON THE PROTOTYPICAL α CYGNI VARIABLE AND A-TYPE SUPERGIANT STAR DENEB

α Cygni variables are highly luminous OBA stars that exhibit low amplitude variations both in photometry (ΔV ≲ 0.15 mag; van Genderen et al. 1989) and velocity (e.g., Abt 1957). These variations are common for massive stars and are observed in most hot supergiants as well as the overluminous supergiants such as luminous blue variables. The variations have been observed in some main-sequence massive stars, and the mechanism responsible for the variations has eluded theory and observers. Photometric and spectroscopic campaigns on multiple stars that span the upper H–R diagram and exhibit these variations will be crucial to understanding these variations and their source.

Deneb (α Cygni, HR 7924, HD 197345) is an early-type A supergiant (spectral type A2 Iae) and is the prototype for the α Cygni type variable stars. With a great apparent brightness and because it can be observed for most of the year at northern latitudes, Deneb can be easily studied spectroscopically with small or moderate aperture telescopes. Deneb has been one of the most studied A-type supergiants. Fath (1935) and Paddock (1935) made the first variability studies of this star, using photometry and radial velocities, respectively. Abt (1957) observed nine supergiants, including Deneb, and showed that early- and intermediate-type supergiants exhibited oscillations of their atmospheres. Lucy (1976) presented a harmonic analysis of Deneb which demonstrated that multiple oscillations were present in the atmosphere of the star. He also discovered what appeared to be a binary period in the radial velocities of Deneb. Parthasarathy & Lambert (1987) made 123 spectroscopic observations of Deneb in the near-infrared between 1977 and 1982 at McDonald Observatory. They found radial velocity variations similar to those observed by Paddock and analyzed by Lucy, and claimed that the binary motion was recovered but presented no time-series analysis.

The underlying cause of the brightness and velocity variations of the α Cygni variables has been recently analyzed by a few authors. Saio et al. (2006) found that long-period oscillations in HD 163899 could be understood as g-mode pulsations that are trapped in a convection zone positioned above the H-burning shell of the star. An investigation of Deneb's variability by Gautschy (2009) gave evidence that Deneb also has a sub-photospheric convection zone driving the variability. Cantiello et al. (2009) showed how iron convection zones above the H-burning shell could be the cause of the microturbulence and clumping in the base of the wind of hot stars. The variability for the α Cygni variables decreases in amplitude and timescale toward higher effective temperatures, and therefore these processes are easier to observe in the cooler A-type supergiants such as Deneb (e.g., Lefever et al. 2007).

Kaufer et al. (1996, 1997) examined the radial velocity variations and the Hα profile variations of Deneb and five other supergiants. They found that the radial velocities were multiperiodic and showed different periods in different observing seasons. They also found that the equivalent width of Hα exhibited multiperiodic variations and evanescent periodicities. In contrast, Morrison & Mulliss (1998) reported finding a much more active, and possibly cyclic, variability in the Hα profile. Often there are "absorption events" in which a secondary absorption component appears on the blue wing of the absorption component of the Hα profile.

The literature on the photometry of Deneb is remarkably sparse, mostly due to its great apparent brightness (m_V = 1.25), combined with the fact that there are no comparison stars of similar brightness in the vicinity of Deneb. Adelman & Albayrak (1997) used HIPPARCOS observations of Deneb, as well as other early A-type supergiants, to show that these objects are small or moderate amplitude variables (a few hundredths or a tenth of a magnitude in amplitude). The light curve was not complete enough to state more than there was a period of the order of about 2 weeks, but there were most likely other effects to take into account.

It has long been known (e.g., Abt & Golson 1966) that the strength of the Hα emission in hot supergiants is correlated with the stellar luminosity. The potential of this fact for the use of these optically luminous stars as extragalactic distance indicators was greatly advanced when Kudritzki (1999) and Kudritzki et al. (1999) developed the wind momentum–luminosity relationship (WLR). However, only four A-type supergiants were in their sample, and the variability of most of these objects was neglected, with the exception of HD 92207 (A0Ia) which showed little variability in the momentum flow (${\dot{M}}v_\infty$) (Kudritzki 1999). The variability of these objects needs to be investigated further in order to quantify the likely error budget in the use of this relationship. With A-type supergiants being among the brightest stars in the optical, it is important to understand this relationship between momentum flow, as derived from an Hα P Cygni profile, and the luminosity.

The variability of the Hα profile is an important diagnostic for hot supergiants in understanding the amount of clumping in the wind of these stars. If variations observed in Hα profiles were found to be repeatable, they might be attributed to co-rotating interaction regions in the circumstellar matter (e.g., Cranmer & Owocki 1996), which would constrain the rotation period of the star.

In this paper, we present numerous spectroscopic and photometric observations obtained during the calendar years 1997 through 2001. Such a large data set is important to understanding the complex observational variations in Deneb and other hot supergiant stars, especially in the context of the WLR. Further, variability studies in stars such as this could reveal the underlying physical mechanisms that cause the observed changes (e.g., Maeder 1980). For example, Gautschy (2009) demonstrated that some pulsational modes would be consistent with a deep convection zone for α Cygni variables.

Three hundred and thirty-nine high-resolution (R = 26,000) red–yellow spectra of Deneb were obtained between 1997 March 20 and 2001 December 21 with the Ritter Observatory 1 m telescope and fiber-fed echelle spectrograph. Details of the instrument are available in Morrison et al. (1997).⁴ For the most part, one spectrum was obtained per night, although multiple exposures were obtained on some nights. Exposure times ranged from 60 to 1200 s, with most being about 300–400 s. The average signal-to-noise ratio was around 100 for the continuum near Hα. Portions of nine orders of echelle data were recorded, and the orders including the Hα profile and the Si ii doublet at λλ 6347, 6371 are presented here. They were reduced using standard techniques for echelle spectroscopy via an IRAF⁵ script.

Three hundred and seventy differential Strömgren photometric measurements were made with the Four College Automated Photoelectric Telescope (FCAPT) during the concurrent time period of 1997 October 29 to 2001 July 2. Non-variable check (ch) and comparison (c) stars were used. Each observation consisted of a measurement of the dark count, followed by the sequence sky-ch-c-v-c-v-c-v-c-ch-sky. The light from Deneb had to be passed through a 5 mag neutral density filter in order to avoid detector nonlinearities. The light from the comparison and check stars was not passed through the filter. The estimated error in the differential photometry is about 0.004 mag, as calculated from the standard deviation of the difference between check and comparison stars. After HD 197036 had been used as the comparison star for the first two years, with HD 198151 as the check star, it was discovered that HD 197036 is a small amplitude, long-period variable star. Thus, it was replaced with HD 198151 for the remainder of these observations, and a new check star, HD 199311, was used. HD 197036 was not used for this analysis.

Reductions beyond the observatory's pipeline script included removal of telluric lines, application of the heliocentric Doppler correction, and normalization to the continuum. Telluric lines were removed by means of the IRAF task telluric. The reference spectra were artificial templates created by mapping telluric lines in the spectra of rapidly rotating hot stars taken with the same instrumentation. Each telluric line in the template was represented by a Gaussian of the appropriate equivalent width and full width at half-maximum, superimposed on a unit continuum.

When multiple exposures were taken in one night, the spectra were co-added before normalization, as spectral variability is not present on these timescales (e.g., Lucy 1976). From the continuum normalized spectra, dynamical spectra were created for the Hα profile. When observations were not available on consecutive days, the spectra were linearly interpolated for the missing observations, provided the gap was less than 10 days. The net equivalent width of Hα was measured for each of the continuum normalized spectra. The integration window was 22 Å centered on the laboratory wavelength of Hα, and the estimated errors in the measurements are less than 5% based on the techniques described by Chalabaev & Maillard (1983).

Radial velocities were measured for the Si ii λλ 6347, 6371 doublet by means of a Gaussian fitted to each line. A weak Fe ii line was found to be blended into the blue wing of Si ii λ 6371 and was also fitted when determining the radial velocity of that line. When more than one observation was available for a single night, the radial velocity was measured for each observation; the variations were never more than 0.5 km s⁻¹ over the course of a night. Observation series of stable giant stars with this spectrograph and camera generally yield standard deviations of a single measurement of the order of 0.5 km s⁻¹. Typically, the radial velocities of the two Si ii lines agreed with each other to within 1 km s⁻¹, but usually to within less than 0.5 km s⁻¹.

The differential magnitudes were approximately converted to standard Strömgren uvby magnitudes using values for the comparison stars from Hauck & Mermilliod (1998). The colors u − b and b − y and the Strömgren indices c₁ and m₁ were calculated. Because no attempt was made to correct for any color dependence that may exist in the attenuation of the neutral density filter, and because multiple uvby standard stars were not used, the colors we derived for Deneb are not precisely on the standard system.

4.1. Dynamical Spectra of Hα

Dynamical spectra were created for each observing season in order to explore the temporal behavior of the Hα profile. Figure 1 shows the variations of Deneb during calendar year 1997. Several strong absorption events occur on the blue wing of the P Cygni profile at intervals of approximately 40 days. We define an absorption event for Deneb by the presence of a second absorption component in the blue wing of the principal absorption component. Usually this second absorption had a velocity blueshifted less than 100 km s⁻¹. The emission component remained fairly static, with the exception of two periods of strengthening. The absorption events we observed do not show a progression to the terminal velocity, as seen in discrete absorption components of other hot, massive stars or in the high-velocity absorption events described by Kaufer et al. (1996) in hotter supergiants such as Rigel (B8 Ia). Aufdenberg et al. (2002) found that the terminal wind speed for Deneb is 225 km s⁻¹. Neither this observing season nor subsequent observing seasons show evidence that the Hα absorption reaches the terminal wind speed. The line probably has a low optical depth at large distances from the star due to insufficient population in the n = 2 level.

Zoom In Zoom Out Reset image size

The cyclic-type behavior during the 1997 observing season was not repeated in the other observing seasons. The 1998 season (Figure 2) exhibits two absorption events, which are farther apart than the 40 day interval of the 1997 season. The emission feature is nearly static for that year of observations. The 1999 season (Figure 3) shows an exceptionally static Hα profile for this star, with a possible beginning of an absorption event at the end of the observing season. One absorption event is observed in 2000 (Figure 4), but showed a more rapid onset and disappearance than the other observed events in our data set. The 2001 season (Figure 5) began with a very strong absorption event, and included one smaller event around HJD 2452075–2452100. The time span of September and October of 2001 (HJD 2452160–2452240) provided an interesting event that is discussed in Section 4.2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

4.2. The High-velocity Absorption Event of 2001

The remarkable event of late 2001 began with a quick onset of a strong emission component. The absorption component reached a larger negative velocity than had previously been observed by us or by Kaufer et al. (1996, 1997). As the event subsided, we observed a quick and dramatic fading of the P Cygni emission. Such a small level of emission had not previously been observed in Deneb. Figure 6 illustrates the details of this event in line plots.

Zoom In Zoom Out Reset image size

The event also showed at least two additional absorption components at some epochs, and a dynamical spectrum created by subtracting the average line profile of the time period (Figure 7) reveals that this was actually two absorption events, spaced by the same period of 40 days observed with minor events in 1997. There is a sudden onset of a highly blueshifted (∼−100 km s⁻¹) absorption that increases in strength with time. About 40 days after the onset of this event, we see a second high-velocity absorption event occur. As this second event reached its maximum absorption strength, we observed an onset of absorption on the red wing of the profile. This event fits the definition of a high-velocity absorption event given in Kaufer et al. (1996). Their definition comprises a larger blue velocity and deeper absorption than normally observed, a similar morphology throughout the time development of the event, and no evidence of any spherical symmetry in the wind at these epochs. The high-velocity absorption events are much more extreme than the "normal" absorption events described in Section 4.1.

Zoom In Zoom Out Reset image size

4.3. Hα Equivalent Widths

4.4. Radial Velocities and Photometry

4.5. Time-series Analysis and the Binarity of Deneb

In order to verify the reliability of various time-series analysis algorithms, their output was compared to that obtained by Lucy (1976) for the data collected by Paddock (1935). A Scargle periodogram analysis (Scargle 1982) of the data from Paddock (1935) only reproduced one aspect of the analysis of Lucy (1976), the long-term variability that was suspected to be binary motion of Deneb.

The confirmation of the binary nature of this prototypical supergiant would be useful for understanding the star, as well as its early evolution (including mass transfer) and estimating the current mass. We used the Scargle periodogram and the CLEAN algorithm (Roberts et al. 1987) to search our data set for the binary period and found that the power spectrum reached a relative minimum in the period range that was suspected. As Lucy (1976) could not add other data to the data of Paddock to better constrain the orbit (Section III of his paper), and our data do not suggest any binary motion, the binary hypothesis for Deneb is not supported.

A method that is more robust for data sets containing multiple periodicities, the CLEAN algorithm, was used to search for periodicities in the data sets. An analysis of the radial velocity data from Paddock (1935) produced similar results to that of Lucy (1976). The frequencies found with CLEAN in our data set are shown in Figure 8, where we plot the frequencies and their relative strengths found for each observing season for net Hα equivalent width, Si ii radial velocities, and Strömgren y. The time-series analysis was performed on the entire data set, but because of the changing characteristics of the star, no global properties were found with a high probability of being real. If the same period were found for all three data sets, there would be some evidence for a photospheric-wind connection, but our analysis showed no evidence of this behavior. Similarly, if there were evidence for similar periods in just the photometry and radial velocities, that would provide information about pulsational behavior of the photosphere of Deneb.

Zoom In Zoom Out Reset image size

4.6. Radial Pulsations

Our analysis did find some periods that were the same for the radial velocities and for the photometry during 1998 and 1999. The phased (to HJD 2450000.0) data were fitted (minimum χ²) with a sine curve of the form

$$\begin{equation} N=A\times \sin (2\pi (p-x_c))+c \end{equation} \tag{ 1 }$$

where N is the parameter we are fitting (radial velocity or magnitude), p is the arbitrary phase we are solving for, A is the amplitude, x_c is a phasing offset, and c is an offset for the systemic velocity or magnitude. The period was not allowed to vary during these fits. The parameters for our photometric and radial velocity fits are given in Table 4 for the two epochs of radial pulsations and the variations are shown graphically in Figure 9. The offsets in phase between the photometry maximum and radial velocity maximum for the two periods with the highest probability of radial pulsations (epochs with strong signals that match with both the photometry and radial velocity data; the fall of 1998 (P = 17.8 days) and the fall of 1999 (P = 13.4  days)) are roughly 0.25P, which strongly supports the hypothesis that Deneb was experiencing radial pulsations at these epochs.

Zoom In Zoom Out Reset image size

Table 4. Coefficients for the Sine Wave Fits (Through χ² Minimization) to the Radial Pulsations Shown in Figure 9 as well as Other Variables we Fitted

Season	Type	Period	A	xc	c
		(days)	(km s−1) or (mag)	(phase)	(km s−1) or (mag)
Fall 1998	u	17.8	−0.016 ± 0.002	0.39 ± 0.02	2.208 ± 0.001
Fall 1998	v	17.8	−0.011 ± 0.002	0.42 ± 0.04	1.335 ± 0.002
Fall 1998	b	17.8	−0.011 ± 0.002	0.38 ± 0.04	1.162 ± 0.002
Fall 1998	y	17.8	−0.011 ± 0.002	0.42 ± 0.03	1.160 ± 0.001
Fall 1998	Si ii RV	17.8	2.27 ± 0.62	0.19 ± 0.03	−2.56 ± 0.43
Fall 1999	u	13.4	−0.023 ± 0.002	0.38 ± 0.02	2.211 ± 0.002
Fall 1999	v	13.4	−0.016 ± 0.002	0.39 ± 0.02	1.336 ± 0.001
Fall 1999	b	13.4	−0.017 ± 0.002	0.39 ± 0.02	1.155 ± 0.001
Fall 1999	y	13.4	−0.017 ± 0.002	0.39 ± 0.02	1.152 ± 0.001
Fall 1999	Si ii RV	13.4	1.21 ± 0.57	0.17 ± 0.08	−2.76 ± 0.42

Notes. The errors are the standard errors of the fit.

Download table as:  ASCIITypeset image

We have searched for high-velocity absorptions in a sample of B8 Ia–A2 Ia supergiants that have been monitored at Ritter Observatory since 1992. Our initial results, which are shown in Table 5, begin to constrain the range of effective temperatures in which supergiant stars exhibit high-velocity absorption events. With the five years of data we have analyzed combined with the data presented in Kaufer et al. (1996), only one high-velocity absorption event has been observed in Deneb. These events are thus very rare for this star. Other stars of similar properties (ν Cep⁶ and 6 Cas) have not shown any signs of high-velocity absorption in a long time series of observations. We can therefore conclude that Deneb is at the lower end of the range of effective temperature and luminosity (T_eff∼8500 ± 200 K, $\log ({L\over L_{\odot }})\sim$5.2; Aufdenberg et al. 2002) for which supergiants can exhibit high-velocity absorption events.

As discussed in Section 4.6, we found evidence for radial pulsations in the atmosphere of Deneb at two epochs during our five-year investigation on this star. These pulsations were present for a few cycles and then vanished. Kaufer et al. (1997) did not find radial pulsations for Deneb, although their data may have suggested such variability. Simultaneous photometry of the star is necessary for finding the radial pulsations because the atmosphere is also unstable to many non-radial modes simultaneously. The light curves at the epochs where we found radial pulsations have an amplitude such that they cannot be easily explained with non-radial modes (see, e.g., Unno et al. 1979 for a description of non-radial oscillations).

Evidence was found for radial pulsations at two epochs during this campaign. Radial pulsations are somewhat rare in the atmosphere of Deneb, where the atmosphere does not usually display these variations. With only two short epochs having radial pulsations, we find that Deneb displays radial pulsations ⩽20% of the time. While photometry was difficult to obtain for such a bright target, it is necessary to find such behavior. A complete look at all the available data collected for this star will potentially lead to a better understanding of its interior. Gautschy (2009) has demonstrated that the variability of Deneb may be indicative of an underlying convective region for this star. Detailed analyses of the variability of a larger population of supergiants could demonstrate if these intermittent radial pulsations are normal for the population.

A 40 day period was found with absorption events in 1997 as well as in the high-velocity absorption event of 2001. A similar period is reported by Kaufer et al. (1996) for the 1991 observations (Hα) of Deneb. Aufdenberg et al. (2002) derived a stellar radius of 180 R_☉ and vsin i = ≈25 km s⁻¹ for Deneb, which corresponds to a rotational period of 58 days. The vsin i was used in the spectral line fitting of the star, and no error bars were given, but a small increase of ≈10 km s⁻¹ to this parameter will yield a rotational period of ≈40 days. Similarly, the radius of the star could be reduced by ≈30% to yield this period. With a poorly determined parallax (π = 2.31 ± 0.32 mas; van Leeuwen 2007), the interferometric radius determined by Aufdenberg et al. (2002) cannot be used to constrain this parameter. Therefore, we conclude that the 40 day period is consistent with the rotation period of the star and these observations may indicate occasional coupling between the photosphere and the wind.

During each observing season, the behavior of the Hα profile was different. Therefore, long-term monitoring (over multiple seasons) is crucial to understanding the variability mechanisms of stars such as Deneb. With the extreme variability seen at the time of the high-velocity absorption event in 2001, caution should be used in assigning mass-loss rates and other parameters to a single-epoch observation for Deneb and similar stars. Further work will be needed for the wind–momentum luminosity relationship (Kudritzki 1999; Kudritzki et al. 1999) with the variability of these stars taken into account.

There is a need for observing campaigns at high spectral resolution for stars in the temperature and luminosity regime of Deneb. The high-velocity absorption event we observed in 2001 is the only observed event for Deneb. High-resolution spectroscopic monitoring of Deneb at Ritter Observatory spans nearly two decades (1993 through present), and analysis of the entire data set is planned and will provide additional insights into the frequency of occurrence of high-velocity absorption events, of episodes of radial pulsation, and perhaps of other phenomena. Similar stars (in temperature and luminosity) need monitoring to see if Deneb is unique and where the temperature and luminosity boundary for high-velocity absorption events is located.

We acknowledge the anonymous referee for making suggestions that improved this paper. Publication of this paper was supported in part by The Scott E. Smith Fund for Research at Ritter Observatory. We are grateful to the Ritter Observing team who performed the spectroscopic observations during the calendar years 1997 through 2001, especially Karen Bjorkman, Anatoly Miroshnichenko, David Knauth, Chris Mulliss, Howard Ritter, Tracy Smith, Will Fischer, Patrick Ouellette, John Wisniewski, and Josh Thomas. The spectra were reduced by Chris Mulliss, David Knauth, and John Wisniewski by means of a script written by R.C. Dempsey and later modified by J.P. Aufdenberg. Chris Mulliss created the first dynamical spectrum of Deneb from Ritter spectra and began this project. Amber Ferguson analyzed radial velocity standard stars which helped to understand the error associated with our measurements; she was supported by NSF-PREST. We thank Observatory Technician Robert Burmeister for keeping the telescope in excellent working condition. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. Support for E. Kryukova on this project came from the NSF–REU program. Support for Ritter Observatory during the time of the observations was provided by the Fund for Astrophysical Research and the American Astronomical Society Small Grants Program. Support for FCAPT during the time period came from the NSF Grant AST-0587381. Special thanks to Louis J. Boyd, George McCook, and Robert J. Dukes, Jr. for their help in keeping the FCAPT in good working order. We thank Douglas Gies (Georgia State University) for helpful discussions related to the pulsational behavior of Deneb and for some software used in Figure 7.

Facilities: Ritter Observatory, - Four College Automated Photoelectric Telescope -