SPECTROSCOPIC VARIABILITY OF IRAS 22272+5435

The optical spectrum of IRAS 22272+5435 is more complicated than is generally seen in stars of approximately solar temperature. A number of lines of heavy elements from weak transitions are visible, which are due to the strong enhancement of neutron-capture elements, and these s-process lines are significantly blended with the lines of another species throughout the spectrum. In addition, numerous weak and sharp C₂ and CN lines are present in the spectrum. These sharp lines probably originated in the AGB shell. Moreover, broader C₂ and CN lines in comparison with the shell lines are visible in some of the observed spectra (see Figures 2 and 3). These lines are variable. We inspected a number of lines from the blue to near-infrared wavelength region and concluded that most of the absorption features are variable on a timescale of a month to a few years. A smaller number of spectral features seem to be variable on a timescale of a week. A correlation of the intensity of the C₂ bandheads with the phase of the light variation is clearly visible in the spectra—the Swan system bandheads are weak near light maximums (e.g., on 2010 September and 2008 February) and strong near light minimums (e.g., on 2011 November and 2006 October). A shift of wavelengths is clearly seen for the C₂ bandheads relative to the photospheric radial velocity, the latter of which was measured on the basis of weak and symmetrical atomic lines. The intensity variation of CN Red system lines in general follows the intensity variation of C₂ Swan system lines (see Figure 3). In addition, the CN (5,1) Red system lines seem to display weak emission near the observed light maximums. The profiles of strong atomic lines are broad and split into two or three components that are partly resolved at this resolution (Figure 4). Comparison of seven spectra observed during the set of observations on 2010 September revealed only minor spectroscopic variability, which is seen only in some strong lines (see Section 3.4). Therefore, we combined these spectra, observed in a time span of 12 days, with the goal of increasing the S/N (especially in the blue wavelength region). A similar procedure was performed for the six spectra gathered during the set of observations on 2011 November. According to the photometry (see Figure 1), IRAS 22272+5435 is at its light maximum on 2010 September and at its light minimum on 2011 November. A quantitative analysis of the spectroscopic variability was performed using the combined spectra observed at these two light extrema.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The value of effective temperature for the central star of IRAS 22272+5435 was determined from the constraint that the relation between iron abundances (from individual Fe i lines) and the excitation potential has a zero slope, resulting in T_eff = 5600 ± 250 K (Začs et al. 1995). Reddy et al. (2002), using the same method, obtained a similar value within the uncertainty of T_eff = 5750 K. The best-fit model of spectral energy distribution for IRAS 22272+5435 gives T_eff ∼ 5800 K for the central star (Ueta et al. 2001). The interstellar extinction was found to be A_V = 2.5 mag for this best-fit model, which is larger than the observed value along the line of sight toward IRAS 22272+5435 for the adopted distance of 1.6 kpc (Neckel & Klare 1980). The determination of the effective temperature on the basis of color indices entails many complicated problems for the post-AGB stars. Chemical peculiarity and the presence of dust are responsible for the continuum depression in the blue and excess in the infrared wavelength regions, both affecting the observed colors. The central star of IRAS 22272+5435 is a pulsating star with time-varying color indices. This pulsation leads to temperature variation in the photosphere of the star where most of the absorption lines are formed. Thus, these changes of T_eff during the pulsation cycle should contribute to the observed spectroscopic variability. A crude estimation of the temperature variation for the central star was performed using the color index $(V-{R}_{C}),$ which is less affected by the continuum anomalies than is $(B-V).$ A maximal variation of the color temperature was estimated using the $(V-{R}_{C})$ observed at the light maximum on 2010 September and that at the light minimum on 2011 November, when two sets of high-resolution spectroscopic observations were gathered. Assuming a color–temperature relationship for IRAS 22272+5435 similar to that for a typical supergiant, the color variation of $\delta (V-{R}_{C})\;\simeq$ 0.12 mag translates into the temperature variation of about 500 K for the star with T_eff = 5750 K (Johnson 1966). Bearing in mind the published values of effective temperature for IRAS 22272+5435 and the probable temperature variation in the atmosphere due to pulsation, two final temperatures were adopted for the modeling of the spectrum for the central star, T_eff = 5750 and 5250 K (the hotter and cooler atmospheric models, respectively).

The surface gravity was adopted to be $\mathrm{log}g$ = 0.5 (cgs) for the final models, which is a typical value for the cool post-AGB stars (Pereira et al. 2012) and agrees well with the gravity determined for IRAS 22272+5435 using an ionization balance for iron. In addition, we examined some atmospheric models with lower T_eff and the gravity in a range from $\mathrm{log}g$ = 0.0 to 1.5 (cgs). According to Pereira et al. (2012), $\mathrm{log}g$ = 0.5 and 0.2 (cgs) on average for post-AGB stars with T_eff = 5750 and 5250 K, respectively (see Figure 2 in the cited paper). The microturbulent velocity was adopted to be ξ_t = 4.5 km s⁻¹, which is close to the mean published value (Začs et al. 1995, 1999; Reddy et al. 2002) found by forcing the calculated abundances from individual iron lines to be independent of the EWs. Začs et al. (1999) found evidence that the value of the estimated microturbulent velocity for IRAS 22272+5435 depends on the upper limit of the used EWs, with a tendency to give larger ξ_t for stronger lines.

We adopted the abundances of CNO elements calculated by Reddy et al. (2002), [C/H] = 0.17 dex, [N/H] = −0.24 dex, and [O/H] = −0.30 dex, which lead to the carbon-to-oxygen ratio C/O = 1.6. The standard notations are adopted everywhere.⁹ A medium iron deficiency, [Fe/H] = −0.7 dex, and an enhanced abundance for all neutron-capture elements were adopted, [n/H] = +1.5 dex. We adopted the scaled solar abundances, [X/Fe] = 0.0, for the rest elements, despite some overabundances reported for IRAS 22272+5435 in the literature. The fits between the synthesized atomic spectra and the observed ones confirm in general the correctness of the adopted atmospheric parameters and chemistry (see Sections 3.4–3.6), although for some species we detected the difference over a typical value of 0.2 dex in comparison with the abundances calculated by Reddy et al. (2002). The atmospheric parameters and abundances collected from literature are given in Table 3. Updates of the published abundances are foreseen using the time series of spectra and the new models in a future paper. On the other hand, the optical spectrum of IRAS 22272+5435 originates at various depths in the atmosphere of the central star, and these are affected differently by pulsations. As a result of this, a hydrostatic atmospheric model is not able to describe correctly the formation of all lines. According to the recent studies, the dynamical models fit the observed spectra of long-period variables (LPVs) much better than any hydrostatic model. However, for some spectral features, the variations in the line intensities predicted by dynamical models over a pulsation cycle give values similar to that of a sequence of hydrostatic models with varying temperature and constant surface gravity (Lebzelter et al. 2014). It is expected that absorption lines that are formed deep in the atmosphere of low-amplitude pulsating stars are calculated correctly using hydrostatic models. Loidl et al. (2001) demonstrated that hydrostatic atmospheric models are able to reproduce quite well the observed spectra of cool carbon-rich semiregular variables in the large wavelength region from 0.7 to 2.5 μm.

The standard atmospheric models prepared by Kurucz (1993) were used for the first iteration of the spectrum synthesis to examine the published abundances. Then a grid of new self-consistent atmospheric models was calculated in the range of effective temperature and surface gravity suitable for the central star of IRAS 22272+5435 (Table 4) to check the impact of models on the calculated spectrum. The new atmospheric models were calculated using the code SAM12 (Pavlenko 2003), a modification of the Kurucz code ATLAS12 (Kurucz 2005), designed to calculate atmospheric models of red giants of a given chemical composition. The modifications are as follows. The bound-free absorption caused by the C i, N i, and O i atoms was added (Pavlenko & Zhukovska 2003) to the continuum absorption sources included in ATLAS12. We also take into account collision-induced absorption—absorption of the molecular complexes He–H₂ and H₂–H₂ induced by collisions—which becomes an important source of opacity in the atmospheres of cool metal-poor stars (Borysow et al. 1997). Molecular and atomic absorption in the SAM12 code are taken into account by using the opacity sampling technique (Sneden et al. 1976). A compiled list of spectral lines includes atomic lines from the VALD3 database (Piskunov et al. 1995; Kupka et al. 1999) and molecular lines of CN, C₂, CO, SiH, MgH, NH, and OH from the Kurucz database (Kurucz 1993). In addition, the absorption by the HCN bands was taken into account according to Harris et al. (2003, 2006). We also took into account the absorption of isomers HCN and NHC. SAM12 was used successfuly for calculations of the atmospheric models for chemically peculiar stars like R CrB stars, Sakurai's object, and other evolved stars.

Table 3.  The Atmospheric Parameters and Abundances Collected from Literature

Teff	$\mathrm{log}g$	ξt	[Fe/H]	[C/H]	[N/H]	[O/H]	[n/H]a	References
(K)	(cgs)	(km s−1)	(dex)	(dex)	(dex)	(dex)	(dex)
5600	0.5	3.7	−0.49	+1.19	⋯	−0.10	+1.9 (6)	(1)
5600	0.5	3.7, 7.0	−0.5	+0.8	⋯	−0.2	+1.5 (17)	(2)
5750	0.5	4.5	−0.82	+0.17	−0.24	−0.30	+1.2 (7)	(3)
5800	⋯	⋯	⋯	⋯	⋯	⋯	⋯	(4)

Notes.

^aThe average enhancement of neutron-capture elements. The number of species is given in parentheses.

References. (1) Začs et al. 1995; (2) Začs et al. 1999; (3) Reddy et al. 2002; (4) Ueta et al. 2001.

Download table as:  ASCIITypeset image

The structure of the calculated atmospheric models depends on the adopted chemical composition; therefore, we used the actual abundances for IRAS 22272+5435 retrieved from the literature (see Section 3.2). The published metallicities for IRAS 22272+5435 are in the range of 0.3 dex, and the mean weighted metallicity was adopted for calculation, [Fe/H] = −0.7 dex. We adopted the carbon and oxygen abundances estimated by Reddy et al. (2002) because of higher resolution of their spectrum (lower blending effects) and larger number of employed lines in comparison with the previous papers. The published abundances for neutron-capture elements are contradictory—the differences between estimations for some elements are much larger than the predicted uncertainties (Začs et al. 1995, 1999; Reddy et al. 2002). The mean enhancement calculated on the basis of 17 neutron-capture elements, [n/H] = +1.5 dex, was adopted for all the n-capture elements in the SAM12 models. A comparison of the temperature structure of two SAM12 atmospheric models and the corresponding Kurucz models is given in Figure 5. As can be seen, the models are similar; however, the outer layers for the SAM12 models are cooler in comparison with the Kurucz models. The synthetic spectra were calculated in selected wavelength regions with the code WITA (Pavlenko 1997) and were appropriately convolved with a Gausian profile with FWHM ranging from about 0.4 (blue) to 0.8 (near-infrared) Å depending on the wavelength region. The VALD3 database was used as a primary source of atomic and molecular data for synthesis of the spectra. The list of molecular lines was adopted from the Kurucz database (Kurucz 1993) and SCAN tape (Jørgensen & Larsson 1990).

Zoom In Zoom Out Reset image size

Table 4.  The Grid of SAM12 Atmospheric Models Calculated for Modeling the Spectra

Teff	$\mathrm{log}$ g	[M/H]	C/O	[n/H]
(K)	(cgs)	(dex)		(dex)
5750	1.5	−0.7	1.6	+1.5
5750	0.5	−0.7	1.6	+1.5
5750	0.1	−0.7	1.6	+1.5
5250	0.5	−0.7	1.6	+1.5
5250	0.0	−0.7	1.6	+1.5
4750	0.5	−0.7	1.6	+1.5
4750	0.0	−0.7	1.6	+1.5

Note. The adopted atmospheric parameters and abundances are given.

Download table as:  ASCIITypeset image

Comparison of the observed spectra shows that most of the absorption lines are variable. However, the largest variations are seen mainly in the molecular lines (see Sections 3.5 and 3.6). Numerous lines of carbon-bearing molecules of different strengths are blended with atomic lines throughout the spectrum, especially near the light minima, mimicking a variability in the blended atomic lines. On the other hand, the atomic lines could be intrinsically variable as well because of variations of the atmospheric parameters for the central star during the pulsation cycles. The intrinsic variability of the atomic lines was examined in the blue wavelength region, where contamination from molecular lines is lower in the observed spectra during all phases of the light cycle. An example of the comparison of the spectra observed at light maximum on 2010 September and that observed at light minimum on 2011 November is shown in Figure 6. The color curve is also at its extreme values on these two dates, being bluer (hotter) when brighter. As can be seen, most of the spectral features reveal some minor variability. However, some of them display large intensity variations between the light and color extrema. The most variable atomic lines are marked by tick marks in Figure 6 and mainly result from neutral low-excitation transitions (Table 5). The variability of the spectrum due to temperature variation was modeled using the hot and cool atmospheric models calculated for IRAS 22272+5435. Synthesized spectra for the final atmospheric models with T_eff = 5750 and 5250 K are displayed in Figure 6 (middle panel). The sensitivity to temperature variation is clearly seen in a number of atomic lines, e.g., Ce ii at 4473.77 Å, Sm ii and Ce ii at 4476.5 Å, Fe i at 4489.74 Å, Pr ii at 4492.42 Å, and Dy ii at 4503.23. The sensitivity of the selected lines to the changes of surface gravity was found to be relatively low (see Figure 6; bottom panel). We conclude that most of the intrinsic intensity variations in the weak and medium strong atomic lines throughout the spectrum of IRAS 22272+5435 are because of temperature variations rather than changes in surface gravity in the line-forming region during the pulsation cycle.

Zoom In Zoom Out Reset image size

Table 5.  Lines in Blue Wavelength Region Displaying a Significant Intensity Variation between Spectra Observed at Light Maximum and Minimum

Species	Wavelength	LEP	$\mathrm{log}{gf}$	References
	( Å )	(eV)
Ce ii	4473.77	1.48	−0.489	PQWB
V i	4474.71	1.89	0.037	K09
Nd ii	4475.58	0.06	−1.827	MC
Fe i	4476.02	2.84	−0.819	K07
Fe i	4476.07	3.69	−0.175	K07
Sm ii	4476.48	0.38	−1.759	XSQG
Ce ii	4476.51	0.74	−1.860	PQWB
Fe i	4480.14	3.05	−1.932	BWL
Ce ii	4487.88	1.49	−0.789	PQWB
Fe i	4489.74	0.12	−3.966	K07
Pr ii	4492.42	0.42	−1.153	MC
Ce ii	4498.92	1.32	−1.349	PQWB
Ce ii	4502.57	1.25	−1.250	PQWB
Dy ii	4503.23	0.93	−1.487	MC

Note. Species identifications, rest wavelengths, lower excitation potentials, and oscillator strengths as adopted from the VALD3 database are given.

References. References according to the VALD database: http://www.astro.uu.se/valdwiki/VALD3linelists.

Download table as:  ASCIITypeset image

Inspection of the observed spectra gives evidence that some of the lines are extremely broad and split. The profiles of four Ba ii lines at 4934, 5853, 6141, and 6496 Å, the K i line at 7698, and the Na i D doublet in radial velocity scale relative to the systemic velocity, RV_⊙ = −40.8 km s⁻¹, are displayed in Figure 4. All of the selected lines are resonance or low-excitation lines; those cores are formed mainly in the outer layers of the stellar atmosphere. A medium strong Fe i line at 6411.65 Å with a Gaussian profile is shown for comparison purposes. As can be seen, the profiles of selected lines have a few components in the range of Doppler velocities from about −50 to +50 km s⁻¹ relative to the systemic velocity. The intensity and velocity of these components depend on pulsation phase. Redshifted components of variable intensity are seen at about 25 km s⁻¹ in the Na i D12 and K i profiles and at about 60 km s⁻¹ in the Na i D12 profiles. Blue wings up to about 50 km s⁻¹ are clearly visible in the spectra observed on 2006 October and November. One such broad and split absorption feature at 6390 Å was analyzed in detail using quantitative methods (Figure 7). Some weaker contributing atomic lines are identified and included in the list of lines for synthesis of the absorption profile to understand better the nature of splitting. Their positions are marked by vertical ticks according to the VALD3 database: Sm ii line at 6389.83 Å, Nd ii line at 6389.97 Å, La ii line at 6390.48 Å, and Ce ii line at 6390.61 Å. The profile at 6390.48 Å was calculated with two final atmospheric models for T_eff = 5750 and 5250 K; however, the synthesized profiles are not able to reproduce the observed blue/redshifted components. Thus, the sequence of profiles indicates probably that expanding and infalling layers are simultaneously present in the outer atmosphere of the central star of IRAS 22272+5435. We examined short-term variability in the profiles of split lines employing the time series of spectra observed near the light maximum (2010 September) and minimum (2011 October). An example of such an analysis is shown in Figure 7 (right panel) for an absorption feature formed mainly by the La ii line at 6390.48 Å. We detected a significant variation of intensity in the absorption component shifted to the red relative to the stellar photospheric velocity by about 10 km s⁻¹. A monotonic decrease of the intensity for this component was observed over 12 days from September 17 to 28. It should be noted that short-term variability was detected in most of the split low-excitation profiles examined here. The splitting and variability on the timescale of a week seem to be observational evidence of shock waves in the outer atmosphere of the star.

Zoom In Zoom Out Reset image size

The profile of the Balmer Hα line is complicated and variable in the spectrum of IRAS 22272+5435. The broad wings of the absorption profile are approximately in agreement with those calculated for a supergiant with the atmospheric model for T_eff = 5250 K. However, the central part of the profile displays a shell-like emission with a central absorption (see Figure 8). Such profiles with a central absorption within the emission were found to be common for PPNs of spectral type F (Hrivnak & Reddy 2003). Emission features of about equal height for IRAS 22272+5435 are located on both sides of the central absorption feature. A time dependence in the profiles is seen in Figure 8, along with the profiles calculated for the two final models for T_eff = 5750 and 5250 K. The intensity and position of the central absorption feature are variable relative to the calculated profile. As can be seen, in the spectrum observed on 2006 November the position and intensity of the central absorption agree well with those calculated for the photosphere. In the other observed spectra the central absorption is redshifted relative to the photospheric velocity. Radial velocities measured for the central absorption in the spectra observed on 2002 November, 2006 October, 2006 November, 2008 February, 2010 September, and 2011 November are, relative to the systemic velocity, δRV = +7.5, +9.7, +3.8, +9.7, +8.9, and +4.8 km s⁻¹, respectively. Thus, the central absorption is redshifted relative to the systemic velocity in all the observed spectra. The Balmer Hβ line is in absorption at approximately the photospheric velocity.

Zoom In Zoom Out Reset image size

Inspection of the observed high-resolution spectra revealed a significant intensity variation in the wavelength regions where the bandheads of the C₂ Swan system are located. The spectra around the Swan system bandhead (0, 0) at 5165 Å and the bandhead (0, 1) at 5635 Å were examined in detail. We varied the effective temperature of atmospheric models and measured the radial velocity of selected lines to understand the reason for variation.

Seven high-resolution spectra in the wavelength region from 5125 to 5187 Å gathered for IRAS 22272+5435 during the set of observations from 2010 September 17 to 28 at the light maximum (color index minimum) are displayed in Figure 9. The spectra are shifted in the intensity scale for clarity. We concluded that there are no variations in the intensity and wavelength of the absorption lines among the individual spectra that are above the limit of detection in this wavelength region. Thus, the resulting combined spectrum has a very high S/N. This combined spectrum is given on the top of Figure 9. The photospheric spectrum for the central star of IRAS 22272+5435 was calculated with the standard Kurucz (1993) model for the atmospheric parameters (T_eff = 5750 K, $\mathrm{log}$ g = 0.5, [M] = −0.5) determined by Reddy et al. (2002) and for the atmospheric chemistry given in Section 3.2. The agreement between the observed and synthesized spectra for the atomic lines is quite good, confirming in general the correctness of the adopted atmospheric parameters and abundances. However, the intensity of absorption features around 5165 Å was found to be larger in the observed spectrum in comparison with the calculated one. A weak C₂ bandhead in the spectrum observed at the light maximum was suspected. Redward of the C₂ (0, 0) bandhead we detected weak and narrow lines that were identified with single C₂ lines (Table 6). Unfortunately, the intensity of the C₂ (0, 0) bandhead in the calculated spectrum was found to be below the limit of detection, rejecting the photospheric origin of the observed bandhead. However, the bandhead could be stronger for the chemically peculiar atmospheric model. Therefore, the photospheric spectrum of IRAS 22272+5435 was calculated with new self-consistent SAM12 atmospheric models for two effective temperatures, T_eff = 5750 (estimated temperature at maximum light) and 5250 K (estimated temperature at minimum light), with the goal of checking the impact of the adopted models on the synthesized spectrum and of helping to clarify the formation site of the detected C₂ features. For the hot SAM12 atmospheric model the intensity of the bandhead in the calculated spectrum is still below the limit of detection. The fit between observed and calculated line intensities for the SAM12 model with T_eff = 5250 K is quite good near the C₂ bandhead, except for the position of the bandhead (Figure 10). However, the calculated intensities of the C₂ photospheric lines blueward of the bandhead are too strong in the cool model, and the profiles are broader in comparison with the observed ones. For these reasons we conclude that the site of formation of the weak C₂ (0, 0) bandhead and narrow single lines lies outside the photosphere of the central star of IRAS 22272+5435.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The profiles of seven narrow C₂ lines shortward of the (0, 0) bandhead observed at the light maximum on 2010 September are displayed in Figure 11 (panel (a)) in radial velocity scaled relative to the systemic velocity (see Table 2). The lines are shifted by δRV = −8.0 ± 0.4 km s⁻¹ relative to the systemic velocity. The blueshift and the width of these lines agree well with those observed for the circumstellar C₂ Phillips system lines in the spectrum of IRAS 22272+5435, δRV = −8.4 ±0.5 km s⁻¹ (Začs et al. 2009). This would lead one to conclude that the site of formation of the narrow Swan system lines is the AGB shell. However, in the spectra observed on 2011 November at the light minimum (color index maximum), the C₂ (0, 0) Swan system lines are much stronger. Because most of them are blends at light minimum, we selected for quantitative comparison only three less blended C₂ lines from the eight lines measured in the spectrum at the light maximum. The measured radial velocities, FWHMs, and EWs are given in Table 6, and the profiles are compared in Figure 11 (panel (a)). As can be seen, the radial velocity of the blueshift C₂ lines relative to the systemic velocity is the same to within 2σ at the light maximum on 2010 September and at the light minimum on 2011 November, δRV = −7.5 ± 0.6 km s⁻¹, despite the increased EWs and FWHMs at minimum light.

Zoom In Zoom Out Reset image size

The spectra of IRAS 22272+5435 obtained in different epochs and phases display large changes around the position of the C₂ Swan System (0, 1) bandhead (Figure 2). The spectra observed near the light maximum on 2010 September, on 2008 February, and on 2002 November are similar and with a low molecular contribution. The bandhead of the C₂ (0, 1) Swan system is clearly seen, with an equivalent width of about 60 mÅ on average. However, other lines of the Swan system are too weak to be identified with certainty. The bandhead is blueshifted by about 17 km s⁻¹ relative to the systemic velocity (Figure 11; panel (b)) in the spectrum observed on 2010 September, and its equivalent width changes from 87 to 60 and then to 40 mÅ in the spectra observed near the light maximums on 2002 November/2008 February/2010 September. In the spectrum observed at the light minimum on 2011 November the bandhead and the lines of the C₂ (0, 1) system are extremely strong. The measured blueshift of the bandhead is about 10 km s⁻¹ relative to the systemic velocity (lowest curve in Figure 11, panel (b)). Two less blended C₂ lines at 5629 and 5631 Å give a blueshift of 6 km s⁻¹ on average relative to the systemic velocity for the line-forming region.

The spectrum around the C₂ Swan system (0, 1) bandhead was synthesized using the Kurucz (1993) and SAM12 models, and the comparison is given in Figure 12. A fit between the spectrum observed at the light maximum and the spectra calculated with the hot atmospheric model for T_eff = 5750 K (estimated temperature at maximum light) is quite good for most of the identified s-process lines, indicating a correctness of the adopted s-process abundances. However, the synthesized spectra have not reproduced the C₂ bandhead at 5635 Å and a number of weak lines of unknown nature. In addition, the synthesis of strong atomic lines is problematic, e.g., a poor fit for the feature at 5641 Å identified with the Sc ii line. The spectra calculated with the Kurucz and the SAM12 models are similar, with the lines calculated with the SAM12 model being just a bit deeper. The spectrum observed at the light minimum on 2011 November was modeled with the cool SAM12 model for T_eff = 5250 K. In addition, the atmospheric model with T_eff = 4750 K was employed to reproduce formally the observed intensity of the C₂ bandhead. As can be seen, the calculated photospheric spectra with T_eff = 5250 (estimated temperature at minimum light) and 4750 K have not reproduced the observed molecular spectrum at minimum light. The observed C₂ spectrum is blueshifted relative to the calculated ones, and the molecular lines (e.g., at 5631.05 Å) are narrower in comparison with those in the modeled photospheric spectrum. This argues that the site of formation of the C₂ Swan system (0, 1) bandhead is outside the photosphere. At the same time, one does see from the atmospheric models calculated for different temperatures in the current region that the variability observed in the atomic lines (e.g., Ce ii at 5637.36 and 5638.11 Å) is partly because of temperature variation in the atmosphere of the star.

Zoom In Zoom Out Reset image size

A comparison of the observed spectra revealed a significant intensity variation of lines in the wavelength region around 6400 Å (Figure 3). Most of the variability was found to be in the lines of the CN Red system. We varied atmospheric models and measured radial velocity of selected lines again to understand the reason for variation. The spectrum observed at the light maximum on 2010 September was synthesized with the hot atmospheric model for T_eff = 5750 K (estimated temperature at maximum light). The fit among the calculated and the observed spectra is given in Figure 13. We conclude that the intensities of CN Red system lines as calculated with the hot Kurucz (1993) and SAM12 models are too weak to be identified in the spectra. In the spectrum observed on 2010 September, weak emissions are suspected at the positions of CN (5,1) lines. We examined the presence of such emissions in the rest spectra and confirmed it in the spectra observed at light maximum on 2008 February and 2002 November. On the other hand, a good fit between observed and calculated spectra was found for the s-process elements (Nd, La, Pr, Sm) and for the neutral and ionized iron lines (see list in Table 7), confirming the correctness of the adopted gravity, metallicity, and abundances of the selected s-process elements. The fit between the spectrum observed at the light minimum on 2011 November and the synthesized spectra is shown in Figure 13 (bottom panel). We concluded that the synthesized spectrum calculated with the cool atmospheric model for T_eff = 5250 K (estimated temperature at minimum light) reproduces correctly the atomic lines; however, the modeled molecular lines are too weak in comparison with the observed ones. At the light minimum, CN Red system lines are strong, and there is significant blending with the atomic lines in this region. We concluded that the synthesized spectrum calculated with the cool atmospheric model for T_eff = 5250 K reproduces correctly the atomic lines, but the modeled molecular lines are too weak in comparison with the observed ones. Even the spectrum calculated with the extremely cool model for T_eff = 4750 K does not produce strong molecular lines. In addition, the observed molecular lines are narrower than the calculated photospheric lines, and they are blueshifted relative to the systemic velocity (Table 6). The radial velocity measurements for selected CN Red system lines confirmed a blueshift of the site where the CN (5, 1) lines are formed at light minimum, δRV = −5.5 ± 0.7 km s⁻¹ (Figure 11, panel (c)).

Zoom In Zoom Out Reset image size

Table 7.  Selected Atomic Lines Synthesized in the Spectrum of HD 235858

Species	Wavelength	LEP	$\mathrm{log}{gf}$	References
	( Å )	(eV)
Fe i	5166.28	0.00	−4.194	K07
Nd ii	5170.94	0.55	−1.686	MC
Fe i	5171.60	1.48	−1.792	K07
Pr ii	5175.27	0.22	−1.241	MC
Pr ii	5175.84	0.42	−1.282	MC
Gd ii	5176.29	1.06	−0.739	DLSC
Nd ii	5618.99	1.77	−0.649	HLSC
Nd ii	5620.59	1.54	−0.309	XSCL
Ce ii	5623.00	0.96	−1.150	PQWB
Ce ii	5623.03	0.56	−3.260	PQWB
Fe i	5624.54	3.42	−0.754	K07
Nd ii	5625.73	0.93	−1.119	HLSC
Ce ii	5630.38	1.64	−0.940	PQWB
Pr ii	5636.47	1.05	−1.062	BLQS
Ce ii	5637.36	1.4	−0.501	PQWB
Ce ii	5638.11	0.6	−1.720	PQWB
Pr ii	5638.79	0.63	−1.070	MC
Sc ii	5641.00	1.5	−1.131	LD
Nd ii	6382.06	1.44	−0.750	HLSC
Fe ii	6383.72	5.55	−2.070	BSScorr
Nd ii	6385.15	1.16	−0.770	MC
Nd ii	6385.19	1.6	−0.360	XSCL
Sm ii	6389.83	1.17	−1.959	MC
Nd ii	6389.97	1.5	−0.770	HLSC
La ii	6390.48	0.32	−1.410	LBS
Ce ii	6390.61	0.61	−2.41	PQWB
Fe i	6393.60	2.43	−1.432	K07
Pr ii	6397.97	1.05	−0.940	BLQS
Fe i	6400.00	3.6	−0.289	K07
Sm ii	6406.25	1.36	−1.339	LD-HS
Fe i	6411.65	3.65	−0.594	K07
Fe ii	6416.92	3.89	−1.877	BSScorr
Fe i	6419.95	4.73	−0.240	K07
Fe i	6421.35	2.28	−2.027	K07
Ti ii	6559.56	2.05	−2.175	K10
Pr ii	6566.77	0.22	−1.721	MC
Ce ii	7850.03	0.74	−1.289	PQWB
Ce ii	7857.55	0.9	−1.289	PQWB
Ce ii	7860.98	0.7	−2.560	PQWB
Y ii	7881.88	1.84	−1.569	K11
Ce ii	7898.97	0.9	−1.119	PQWB
Nd ii	7900.39	1.35	−1.500	MC

Note. Species identifications, the rest wavelengths, lower excitation potentials, and oscilator strengths adopted from the VALD3 database are listed.

References. References according to the VALD database: http://www.astro.uu.se/valdwiki/VALD3linelists.

Download table as:  ASCIITypeset image

A comparison of the observed spectra for IRAS 22272+5435 in the near-infrared region around 7900 Å revealed a dramatical intensity variation. The spectra observed at the light maximum on 2010 September and at the light minimum on 2011 November are displayed in Figure 14, along with a telluric spectrum observed simultaneously (upper panel). The atomic lines (see Table 7) and the narrow CN Red System (2, 0) lines (see Table 6) were identified in the spectrum observed at light maximum. A fit between the observed spectrum and that calculated for the hot model with T_eff = 5750 K (estimated temperature at maximum light) confirms in general the accepted s-process abundances and some previously noted problems in fitting the profiles for strong atomic lines (Figure 14). We see that photospheric CN lines are too weak in the synthetic spectrum calculated with the hot atmospheric model to expect them to be visible in the photosphere of the star. Thus, their presence points to a nonphotospheric formation region. The width and shift of the identified CN (2, 0) lines confirm their circumstellar origin, with δRV = −8.6 ± 0.4 km s⁻¹ (Figure 11; panel (d)). The CN (2, 0) Red system lines are extremely strong in the spectrum observed at the light minimum on 2011 November, and the atomic lines are significantly blended in that wavelength region. The spectra calculated with the cool atmospheric model for T_eff = 5250 K (estimated temperature at minimum light) and with the extra cool atmospheric model for T_eff = 4750 K have not reproduced the observed spectrum—the observed CN lines are much stronger and narrower in comparison with the modeled photospheric lines. The radial velocity measured in the spectrum observed on 2011 November, using the three less blended CN lines, confirms a blueshift of the line-forming site, with δRV = −8.4 ± 0.6 km s⁻¹, relative to the systemic velocity (Table 6). Thus, the blueshift of the CN lines relative to the systemic velocity is the same within 1σ at the light maximum and at the light minimum, despite the large changes in their EWs and FWHMs.

Zoom In Zoom Out Reset image size