IMAGING THE COOL HYPERGIANT NML CYGNI'S DUSTY CIRCUMSTELLAR ENVELOPE WITH ADAPTIVE OPTICS

The powerful OH/IR source and cool hypergiant NML Cyg is one of the most massive and luminous M stars in the Galaxy (∼40M_☉, M_bol ≃ −9.5; Hyland et al. 1972; Humphreys & Davidson 1979, 1994; Morris & Jura 1983; Schuster et al. 2006a; Schuster 2007). Originally discovered by Neugebauer, Martz & Leighton (1965), NML Cyg was easily identified as an extremely bright infrared source (I ∼ 8, K ∼ 1 mag). The star's relatively close proximity to the Cyg OB2 stellar association confirms its high luminosity (Morris & Jura 1983; Schuster et al. 2006a). Even though NML Cyg is luminous, it is also heavily obscured at visual wavelengths due to high interstellar (IS) and circumstellar (CS) extinction, with V fainter than 16.6 mag. Its visual/near-IR spectrum indicates a M6 spectral type (Wing et al. 1967) and its spectrum peaks in the 2–20 μm range (Gillett et al. 1968; Stein et al. 1969). Thus, the spectrum indicates a substantial, optically thick dusty CS envelope obscuring the central star. NML Cyg's thick envelope is a result of its strong post-main-sequence (post-MS) wind (with a velocity of 23 km s⁻¹) and high continuous mass-loss rate of 6.4 × 10⁻⁵ M_☉ yr⁻¹ (Hyland et al. 1972; Bowers et al. 1981; Morris & Jura 1983). Habing et al. (1982) mapped an unusual oblong-shaped H ii region⁴ at 21 cm nearly 1' in size around NML Cyg, revealing another unique feature of this enigmatic M supergiant. In addition, more recent high-angular resolution, high-contrast images from the Hubble Space Telescope (HST) Wide Field Planetary Camera 2 (WFPC2) camera reveal an asymmetric CS nebula of dust scattered stellar light (Schuster et al. 2006a). The CS nebula has a shape similar to the H ii region, but is about 300 times smaller in radius. It is necessary to consider NML Cyg's local environment and its effects on the star in order to understand this object's true nature.

Although NML Cyg is a very luminous post-MS star which is losing mass at a prodigious rate, it does not currently dominate its local IS environment because of its unique location within the Milky Way. It lies in relatively close proximity to Cyg OB2, which is possibly the largest OB stellar association—in size, mass, and number of OB stars—in the Galaxy (see Knödlseder 2003, and references therein for a multiwavelength review). The Cyg OB2 association spans nearly 2° on the sky, or ∼30 pc in radius at the distance of 1.74 ± 0.2 kpc (Massey & Thompson 1991), making it one of the closest massive associations to the Sun. Cyg OB2 has ∼120 O star members, including five of the 10 most luminous O stars in the Galaxy (Humphreys 1978), and approximately 2500 B stars. The stellar mass within the association is possibly as high as 10⁵ M_☉. The O stars provide the UV radiation responsible for a Strömgren sphere that extends to a radius of at least 274 on the sky, or ≳80 pc (the projected separation between Cyg OB2 and NML Cyg). The Lyman continuum photon flux of the whole association is estimated to be 10⁵¹ ph s⁻¹, or ≳10⁹ ph cm⁻²s⁻¹ at the location of NML Cyg. Mid-IR images from the Midcourse Space Experiment (MSX) satellite show NML Cyg is located in a relative void of gas/dust inside the Cygnus X super bubble (Schuster 2007). This configuration allows the Lyman continuum photons and near-UV radiation from the hot stars within Cyg OB2 to travel the large distance to NML Cyg relatively unimpeded.

Morris & Jura (1983) demonstrated that the Lyman continuum radiation from Cyg OB2 is responsible for the asymmetric H ii region observed around NML Cyg. The H ii region is the result of a photoionization interaction between the spherically symmetric, expanding hydrogen wind from NML Cyg balanced against an incident plane parallel Lyman continuum photon flux from Cyg OB2 (see Figures 1 and 2 in Morris & Jura 1983). The photoionization is inverted in the sense that usually the nearest massive star is the ionization source and the ionization occurs outward from the central source. Morris & Jura also demonstrated that the strength of the Lyman continuum flux from Cyg OB2 and the density of atomic hydrogen around NML Cyg are sufficient to produce the observed 21 cm emission, thus confirming NML Cyg's high continuous mass-loss rate. In a similar way, Schuster et al. (2006a) described how near-UV photons with energies between a few and 13.6 eV photodissociate the molecular gas in NML Cyg's wind. Schuster et al.'s model explains why the asymmetric shape of the CS nebula resolved in HST WFPC2 images is a consequence of this physical interaction with Cyg OB2. Recent observations of excited-state OH maser emission in NML Cyg's wind by Sjouwerman et al. (2007) are consistent with an increased OH column density, presumably arising from dissociated H₂O molecules at the water photodissociation region (PDR). Since the atomic and molecular gas around NML Cyg is disrupted, the dust facing Cyg OB2 is unprotected from the near-UV photons. It is thus likely that the CS dust grains are heated and destroyed by the radiation from Cyg OB2 and should emit a detectable infrared signature.

In Section 2, we present subarcsecond angular resolution, high Strehl ratio MIRAC3/BLINC mid-IR images of NML Cyg obtained at the 6.5 m MMT observatory with the adaptive optics (AO) system. These observations are among the highest resolution images of this star at mid-IR wavelengths and reveal the complexity in the asymmetric dust envelope surrounding NML Cyg. New photometry from the Spitzer Space Telescope's Infrared Array Camera (IRAC), the HST WFPC2 camera, and ground-based infrared observations are presented in Section 2. Thespectral energy distribution (SED) for NML Cyg's CS envelope presented in Section 3 shows an optically thick silicate dust absorption feature. In Section 4, we characterize the resolved structures in the dusty CS envelope. In Section 5, we show that the CS dust component facing Cyg OB2 exhibits a silicate feature in emission. We suggest that UV radiation from the hot, massive stars within Cyg OB2 likely causes the inversion in the silicate feature through external heating and dust destruction. We conclude by discussing these results in relation to previous works on NML Cyg and comparing NML Cyg to the extremely luminous M-type hypergiant VY CMa.

2.1. MMT/AO Observations

The inverse PDR model for NML Cyg's CS envelope described in Schuster et al. (2006a) implies the presence of warm dust near the photodissociation front(s). According to this model, Cyg OB2's UV radiation is heating/destroying the CS dust on the side facing Cyg OB2. In order to corroborate this model through direct observation, we obtained high-angular resolution, high-contrast mid-IR images of NML Cyg's dusty CS nebula. The observations were made using the University of Arizona's AO system (Wildi et al. 2003) and the MIRAC3/BLINC camera (Hoffmann et al. 1998; Hinz et al. 2000) on the 6.5 m MMT telescope. Table 1 summarizes the observations made in 2006 July and Figure 1 shows the NML Cyg images at 8.8, 9.8, and 11.7 μm. The BLINC module was used in `chop' mode with a frequency of ∼1 Hz, which is useful for background (sky) subtraction when combined with "nod" dithers. Using natural guide star mode, the MMT/AO system produced stable, high-Strehl ratio (∼96%–98%), high signal-to-noise, nearly diffraction limited point-spread functions (PSF) at 10 μm. We have used the source itself as a guide star for the AO system (while extended in the infrared, NML Cyg is a point source at optical wavelengths where the AO system operates). With a magnitude R ∼ 14, NML Cyg is at the sensitivity limit for the MMT/AO wavefront sensor, but was successfully used thanks to the excellent sky conditions.

Table 1. Observations

MMT AO MIRAC3 2006 Jul 23 UT	λ (μm)	Δλ (μm)	Airmass	Exposures no. × s
NML Cyg	8.80	0.88	1.10	13 × 20
.	9.80	0.98	1.18	11 × 20
.	9.80	0.98	1.02	2 × 20
.	11.70	1.12	1.14	11 × 20
γ Dra (calibrator)	8.80	0.88	1.08	15 × 30
.	8.80	0.88	1.18	15 × 30
.	9.80	0.98	1.07	20 × 30
.	9.80	0.98	1.23	36 × 30
.	11.70	1.12	1.09	20 × 30
.	11.70	1.12	1.28	20 × 30
Spitzer/IRAC	λ	Δλ	AOR	Exposures
2004 Jul 27 UT	(μm)	(μm)		no. × s
NML Cyg	3.550	0.750	6588416	3 × 10.4a
.	4.493	1.015	.	3 × 10.4a
.	5.731	1.425	.	3 × 10.4a
.	7.872	2.905	.	3 × 10.4a
OBO MN Bolometer	λ	Δλ	Filter	Exposure
2000 Aug 10 UT	(μm)	(μm)
NML Cyg	1.250	0.200	J	n/a
.	1.653	0.297	H	.
.	2.340	0.500	K	.
.	3.647	1.152	L	.
.	4.900	0.700	M	.
.	10.925	6.730	N	.
.	7.908	0.755		.
.	8.808	0.871		.
.	9.803	0.953		.
.	10.273	1.013		.
.	11.696	1.110		.
.	12.492	1.157		.
HST/WFPC2	λ	Δλ	Filter	Exposures
1999 Sep 16 UT	(μm)	(μm)		no. × s
NML Cyg	0.4293	0.0473	F439W (B)	6 × 500b
.	0.5337	0.1228	F555W (V)	20, 100, 4 × 400
.	0.6564	0.0022	F656N (Hα)	20, 2 × 260
.	0.6677	0.0867	F675W (R)	0.5, 10

Notes. Exposure time not applicable to UMN bolometer since the instrument measures changes in voltage to obtain instrumental magnitudes and signal-to-noise. ^aSaturated. ^bNo detection.

Download table as:  ASCIITypeset image

Zoom In Zoom Out Reset image size

Figure 1 also shows images of the IR standard γ Dra, which was observed at two separate epochs and sky orientations for calibration and PSF characterization (see Table 1, only one epoch is shown in Figure 1). The PSF stability was excellent throughout the night and the sky conditions were especially good, with seeing ≲025. In these PSF images the minima/maxima in the diffraction pattern are closely matched to the Airy pattern for a 6.35 m aperture,⁵ with a FWHM of 033 at 8.8 μm, 036 at 9.8 μm, and 044 at 11.7 μm.

The nearly diffraction limited images have similar benefits in resolution and stability to images from space-based observatories, and at mid-IR wavelengths we can detect and separate the externally heated dust from the rest of NML Cyg's CS nebula. Relatively long exposure times and a five point dither pattern, repeated multiple times with sub-pixel shifts, provided very high dynamic range (HDR) images of NML Cyg and the calibration/PSF standards, in the range ∼2 × 10³to10⁴ (PSF peak relative to background noise). The images were processed with the Drizzle algorithm to maximize the quality of the data and recover sampling resolution (Hook & Fruchter 1997; Koekemoer et al. 2000; Fruchter & Hook 2002). The processed MIRAC3 images were resampled by 2 × 2 from 00954 pixel⁻¹ to a scale of 00477 pixel⁻¹.

The 8.8, 9.8, and 11.7 μm filters were chosen to provide good spectral coverage of the 10 μm silicate dust feature observed for this luminous cool star (Monnier et al. 1997; Blöcker et al. 2001). The photometry of the source in the three MIRAC3 bands is listed in Table 2. The flux in each band was measured in an aperture of 2'' radius, with the sky residual background flux (after sky subtraction for each chop/nod sequence) determined outside this aperture. Photometric calibration was obtained using the PSF reference star γ Dra as photometric standard. The photometry has been corrected to a common airmass. The uncertainty in the photometry quoted in Table 2 do not include 5%–10% systematic uncertainties typical for ground based mid-IR observations.

Table 2. NML Cyg Photometry

Instrument		Filter	λ (μm)	Flux (Jy)	σF (Jy)
HST/WFPC2
1999 Sep 16 UT
		V	0.5337	2.377e-04	0.003e-04
		Hα	0.6564	2.395e-03	0.014e-03
		R	0.6677	6.124e-03	0.029e-03
OBO MN Bolometera
2000 Aug 10 UT
		J	1.250	21.5	1.4
		H	1.653	131	0.7
		K	2.340	333	1.0
		L	3.647	1287	4.2
		M	4.900	2121	10
		N	10.925	4325	58
			7.908	4522	90
			8.808	4124	60
			9.803	3881	43
			10.273	3905	71
			11.696	5020	120
			12.492	5366	84
Spitzer/IRAC PSF fittingb
2004 Jul 27 UT
		3.6	3.550	1150	160
		4.5	4.493	1670	240
		8.0	7.872	4160	830
MMT AO MIRAC3/BLINCa
2006 Jul 23 UT
			8.80	3735	63
			9.80	3780	160
			11.70	5280	130
IRAS PSCc
1983
			12	5580	560
			25	3990	400
			60	1030	100
			100	335	34

Notes. Photometry not color corrected. ^aObservational errors exclude typical 5%–10% systematic uncertainties. ^bNML Cyg's mag is near the limit for IRAC PSF fitting (see Schuster et al. 2006b). 5.8 μm PSF fitting was not reliable. ^cFluxes from the IRAS Point Source Catalog rejects—from the InfraRed Science Archive: http://irsa.ipac.caltach.edu, data tag: ADS/IRSA.Gator#2007/1009/122926_27251. 10% flux uncertainty assumed.

Download table as:  ASCIITypeset image

2.2. Spitzer/IRAC Observations

2.3. Ground-Based IR Photometry

2.4. HST/WFPC2 PC Photometry

Table 2 summarizes our photometric observations of NML Cyg along with archival IRAS point source fluxes. These observations are plotted on λF_λ versus λ SED in Figure 2, along with reprocessed archival Infrared Space Observatory (ISO) Short Wavelength Spectrometer spectra (Justtanont et al. 1996; Kraemer et al. 2002).

Zoom In Zoom Out Reset image size

NML Cyg's spectrum rises rapidly from optical wavelengths through the IR to peak in the 2–20 μm range, with a broad far-IR tail. Visible light images presented by Schuster et al. (2006a) show that NML Cyg is almost completely obscured by the CS envelope enshrouding the star, thus confirming significant CS extinction. This is evident from the broad 10 μm silicate dust absorption feature. Due to the optically thick envelope, this feature is a striking characteristic of the NML Cyg spectrum, setting this star apart from other cool hypergiants such as μ Cep (Gehrz et al. 1970), S Per (Humphreys 1974, and unpublished data), VY CMa (Smith et al. 2001), VX Sgr (Humphreys 1974), HR 5171 A (Humphreys et al. 1971), IRC +10420 (Jones et al. 1993), and M33 Var A (Humphreys et al. 2006), where the 10 μm silicate feature is seen in emission from optically thin CS nebula shells.

NML Cyg is also a semiregular variable with period ∼940 days (Monnier et al. 1997, and references therein). The vertical spread between data sets in Figure 2 is well within the ∼50% amplitude variation reported by Monnier et al. (1997) from 8 to 13 μm. Even with this large variability, Monnier et al.'s results show that the 10 μm silicate feature's shape is roughly constant over a span of nearly 19 years. In addition, the vertical offsets between observations at other wavelengths in Figure 2 are at least in part due to variability. However, the degree to which NML Cyg's spectrum varies, in either amplitude or shape, at other wavelengths has not been well established. IS extinction to NML Cyg is also high, with an estimated range of A_V ∼ 3.7–4.6 mag (Lee 1970; Gregory & Seaquist 1976).

We calculate NML Cyg's minimum bolometric luminosity by integrating⁷ the SED in Figure 2, obtaining L_bol = (1.04 ± 0.05) × 10⁵ · (d/kpc)² L_☉, or (3.15 ± 0.74) × 10⁵L_☉ at 1.74 ± 0.2 kpc (Massey & Thompson 1991, Cyg OB2 distance modulus 11.2). This luminosity does not include a correction for IS extinction, which would add at most a few percent to the total since most of the light is in the mid-IR where extinction is low. The uncertainty in this estimate does not consider the changes in the source brightness due to its variability. This luminosity value is in good agreement with the estimate of L_bol = 1.13 × 10⁵·(d/kpc)² L_☉ by Blöcker et al. (2001). Thus, NML Cyg's intrinsic luminosity, a few × 10⁵ L_☉ (M_bol ≃ −9.0) is similar to that of other extremely luminous M-type hypergiants.

Previously, Morris & Jura (1983) reported a luminosity of ∼5 × 10⁵ L_☉ for NML Cyg at a distance of 2.0 kpc. This luminosity was equal to the most luminous known M supergiants in the Milky Way and LMC, M_bol ≃ −9.5, ∼5 × 10⁵ L_☉ (Humphreys 1978; Humphreys & Davidson 1979). Our value is 37% lower, with 24% due to the downwardly revised distance, and 13% from our integration of the SED from 0.5 to 100 μm (including new photometry). With this revision, NML Cyg appears to be less luminous than the hypergiant VY CMa (M4-5e Ia, 4.3 × 10⁵L_☉ at 1.5 kpc, also without correcting for IS extinction), which has a comparable SED (Smith et al. 2001; Humphreys et al. 2007). The uncertainty in our estimate and the variability of the source makes it difficult to determine unambiguously which of these stars is more luminous. Even with this reduction in luminosity, however, NML Cyg remains the most luminous star with spectral type M6 or later known in the Milky Way.

Figure 1 shows that NML Cyg's CS envelope is clearly extended at mid-IR wavelengths, compared to the point-source (PSF) images of γ Dra, as the diffraction pattern appears "filled-in." As further evidence for the extension and degree of asymmetry, Figure 1 also shows the NML Cyg images after oversubtracting the PSF (epoch 1, scaled to give zero intensity flux at the center pixel in the subtracted images). The asymmetric CS envelope residuals are apparent, having a northwest/southeast elongation axis, and position angle (PA) in the range ∼120°–150°. Also visible in the PSF-subtracted images is a lopsided tail extending to the northwest (across the first diffraction ring). The MMT is an elevation/azimuth mount telescope, so the background sky predictably rotates in the image plane while tracking an object. Thus, it was possible to directly confirm the asymmetric extension by observing NML Cyg twice in the same night, at two sky rotation angles separated by ∼122° (at 9.8 μm only). The extended components rotated by the same number of degrees (not shown here), eliminating the possibility that the PSF or other instrumental effects are the cause of the observed extension.

NML Cyg's mid-IR envelope orientation is similar to the 130°–150° PAs observed in the OH (Masheder et al. 1974; Benson & Mutel 1979; Diamond et al. 1984), H₂O (Richards et al. 1996), and SiO (Boboltz & Claussen 2004, D. A. Boboltz & M. J. Claussen 2005, private communication) masers. Richards et al. (1996) have suggested that the northwest/southeast spatial distribution of the H₂O vapor masers may indicate a bipolar outflow, supported by recent SiO ground-state observations by Boboltz & Claussen. It is also possible that the maser emission is tracing an asymmetric, episodic outflow that may be reminiscent of the arcs and other structures seen in the CS nebula surrounding VY CMa (Smith et al. 2001; Smith 2004; Humphreys et al. 2005, 2007; Jones et al. 2007).

Earlier observations by Hinz (2001) also show NML Cyg's extended envelope using MIRAC3 with the BLINC module in nulling interferometer mode on the MMT. These observations were made prior to the installation of the AO system. Hinz's observations indicated a CS envelope extended beyond 028 at 10.3 μm, with nearly equal flux in the extended envelope relative to the unresolved structure (see his Equation (7.5) and Table 7.2, pp. 89–95). The better resolution and Fourier u, v coverage, of the AO images in Figure 1 reveal at least two structures in the CS nebula, the main envelope aligned with maser observations and an asymmetric component aligned more closely with the direction toward Cyg OB2.

4.1. Parametrizing the CS Envelope Image Intensity

To take full advantage of the exceptional stability and contrast in the MMT/AO MIRAC3 high-angular resolution images, we used deconvolution to more clearly reveal details seen in Figure 1 images of NML Cyg's CS envelope. Figure 3 shows our NML Cyg images after deconvolution with the PSF calibrator (using IRAF task LUCY; Richardson 1972; Lucy 1974; Snyder 1991). The deconvolution reveals the complex CS envelope structure highlighting the elongated envelope extended toward the northwestern (NW) direction. The two contours in the figure are set at the source 1/2 and 1/10 maximum flux level. The inner contour shows that the core of the emission is elongated along a southeast–northwest axis, while the external part of the CS emission is lopsided towards the northwestern (NW) quadrant. The main axis of the core emission has a PA of ∼140° (counterclockwise from north) in the 9.8 and 11.7 μm deconvolved images. This angle is similar to the alignment of the NML Cyg OH, H₂O, and SiO masers, also at ∼130°–150°. The PA is slightly smaller (∼120°) in the 8.8 μm image, possibly due to an image artifact in the direction of the MIRAC3/BLINC chop.

Zoom In Zoom Out Reset image size

To better characterize the shape and spatial scales of the NML Cyg CS envelope, we fit the images in Figure 3 with two two-dimensional Gaussians, one fitting the core emission (which we call the "core envelope" Gaussian) and the other fitting the NW extended low-level structures (the "outer envelope" Gaussian). Note that this fitting process is aimed only to find a practical way of describing the changing colors of the CS envelope at different distances from the center. The fitting process or results do not imply that the individual Gaussians represent individual physical structures in the CS envelope. The core Gaussian has a width and orientation analogous to the width and PA of the 1/2 height contour and is centered on the star. The second Gaussian fits the outer envelope emission and is offset in the NW direction as shown by the 1/10 contour level asymmetry. The best-fit parameters were determined by minimizing the residuals after subtraction of the two Gaussians from the deconvolved images. We start by fitting the first Gaussian to the image core, then the second Gaussian is fit to the core-subtracted image. The process was then iterated to minimize the residuals after subtraction of both Gaussians. Table 3 lists the best-fit parameters for each wavelength, including the size parameter σ and relative flux (in % of the total flux) of each component, the aspect ratio a/b between the major and minor semi-axis and the PA of the core component and the offset of the outer envelope with respect to the core. Figure 4 shows the radial profiles of the deconvolved image and best-fit Gaussians along the cut at PA = 140°. Note how the profiles show the large asymmetry between the southeastern (SE) and NW directions. While the SE profile is fit well by the two Gaussians, some residual flux above the fit is still present in the NW direction.

Table 3. CS Envelope Component Parameters

λ (μm)	Component	ΔRA (mas) a,b	ΔDec (mas)a,b	Flux (%)c	σ (mas)d,e	a/b	PA (°)f
8.8 μm
	Core envelope	0	0	74.1	91.0	1.09	120
	Outer envelope	14	87	25.9	171.7	1.00	⋅⋅⋅
9.8 μm
	Core envelope	0	0	70.4	124.0	1.20	138
	Outer envelope	44	105	29.6	176.5	1.00	⋅⋅⋅
11.7 μm
	Core envelope	0	0	65.6	113.1	1.02	139
	Outer envelope	45	98	34.4	171.7	1.00	⋅⋅⋅

Notes. ^aPosition offsets are relative to core component, +ΔR.A. is to the west. ^bThe errors, including fit uncertainty and field rotation during the observation, is ∼5 mas. ^cPercentages are relative to total flux. Uncertainties better than 2%. ^dFWHM = 2$\sqrt{2\ln {2}}$ × σ. ^eUncertainty of 10 mas or better. ^fMajor axis (a) PA is measured in degrees counterclockwise from north.

Download table as:  ASCIITypeset image

Zoom In Zoom Out Reset image size

The agreement in PA and size between the core envelope emission and the NML Cyg maser observations suggests that this emission arises from the NML Cyg's wind at, or just interior to, the photodissociation front(s) described in Schuster et al. (2006a). The relative flux contribution from the outer envelope component increases from 8.8 to 9.8 to 11.7 μm as we are resolving the optically thin warm dust (T_dust ∼ 200–500 K) farther out in the envelope.

4.2. The Asymmetric NW Excess

To test the accuracy of the deconvolution parametrization in image space, we have subtracted the Gaussian fit, convolved with the PSF reference (γ Dra), from the NML Cyg mid-IR images at each wavelength. The residuals are shown in Figure 5. The grid-like pattern is due to the spatial sampling and linear interpolation of the image on the detector grid, and is the limiting factor in the accuracy of the fit. This pattern may be reduced by increasing the number of unique offsets in the dithered observations and by applying higher order interpolation in the reconstruction of the image on a finer pixel grid. The uniform distribution of this residual pattern and the relatively low flux level in the center is indicative of the good quality of the deconvolutions and fits, and also the stability of the MMT/AO PSF.

Zoom In Zoom Out Reset image size

The figure also shows a clear excess residual in the NW quadrant. The excess peak brightness occurs around ∼03 to 05 (∼520–870 AU) from the central star, with a broad tail extending toward Cyg OB2. This excess cannot be reduced by adjusting the position and shape of the outer envelope Gaussian, constrained by the fit in the other three quadrants. The shape and orientation of this excess are reminiscent to the lopsided emission seen in the HST images from Schuster et al. (2006a, their Figures 2 and 6). The asymmetric excess is separated from the star in Figure 5 at the spatial resolution limit, i.e., it is resolved according to the Rayleigh criterion. It lies just outside the asymmetric dusty reflection nebula seen in the HST images (Schuster et al. 2006a). We suggest that the excess emission is likely from warm dust that is externally heated/destroyed by UV radiation from Cyg OB2. In the next section, this hypothesis will be tested by directly measuring the flux from this excess.

In Table 4, we list the total flux of the source, the fraction of the total flux in the core and outer envelope, the total flux of all the fit residuals shown in Figure 5, and the excess residual flux in the NW quadrant (residual flux in the NW quadrant minus the residual flux in the opposite SE quadrant). In Figure 6, we plot the [8.8]−[9.8] and [9.8]−[11.7] colors for each CS nebula component. Both the core envelope colors are bluer than the total for the source. This core component likely represents the collective flux from where the silicate absorption is deepest and thus characterizes the geometry of the dust responsible for the optically thick feature seen in the star's spectrum. NML Cyg's stellar radius is probably of the order of ∼10 AU, typical for the largest red supergiants, and therefore the material comprising the inner CS envelope (σ ∼ 012, 200 AU) has a size scale of approximately 20 stellar radii. The outer CS envelope colors are redder than those of the core envelope and the total source. This indicates dust with decreasing temperature and density.

Zoom In Zoom Out Reset image size

In contrast, the colors for the asymmetric excess are significantly different, being blue in [9.8]−[11.7] and red in [8.8]−[9.8]. These colors can be explained by the presence of optically thin hot dust with the 9.8 μm silicate feature in emission. The fact that the asymmetric dust emission excess appears only on the side facing Cyg OB2 is consistent with an increase in temperature resulting from the external UV radiation. The density is also likely reduced due to dust destruction. Since the asymmetric excess is shielded from the central star by the optically thick inner CS envelope, Cyg OB2's external energy is required to account for changes in the grain temperature, size, and/or spatial distribution ultimately responsible for the silicate feature's inversion.

In Schuster et al. (2006a), we showed that the UV flux from the Cyg OB2 association is strong enough to significantly heat the dust grains in the NML Cyg envelope, possibly destroying the smallest grains. The excess mid-IR flux we observe in the NW quadrant may be the direct effect of this heating and destruction. Its presence strongly suggests that the CS envelope of NML Cyg is indeed shaped by the interaction of its wind with the radiation field coming from the Cyg OB2 association. The direct detection of this physical interaction between NML Cyg and the Cyg OB2 association is further confirmation that the source is indeed in the neighborhood of Cyg OB2. This supports the distance determination of 1.74 ± 0.2 kpc by Massey & Thompson (1991) for this star as one of the most reliable distances for an evolved massive star near the upper luminosity boundary.

Figure 7 shows the NW excess in our 9.8 μm deconvolved image compared to the scattered light HST/WFPC2 F555W image from Schuster et al. (2006a). As discussed in Section 5, our interpretation for this mid-IR excess is another strong indicator that the UV radiation from the hundreds of massive young OB stars within Cyg OB2 is disrupting NML Cyg's post-MS wind through photoionization, photodissociation, and grain destruction. The models for these processes explain the inverted H ii region's existence and shape, the asymmetric nebula seen with HST, as well as the externally heated dust visible in our mid-IR images. The inverse photoionization (Morris & Jura 1983) and photodissociation (Schuster et al. 2006a) models for this interaction have assumed a radially symmetric, constant velocity hydrogen gas wind, i.e., density ρ ∝ r⁻². The agreement between theory and observation suggests that this steady-state, uniform outflow is an overall good approximation of NML Cyg's stellar wind.

Zoom In Zoom Out Reset image size

However, the asymmetric mid-IR image reveals that NML Cyg's dusty CS envelope does have a more complex underlying structure. One may speculate that the identified structures may represent the integrated signal from many arcs, loops, bipolar outflows, and other three-dimensional structures below our angular resolution that may be the result of episodic and asymmetric mass-loss, much like VY CMa (see Figure 3, Smith et al. 2001 as well as Humphreys et al. 2007; Jones et al. 2007). Asymmetric structures arising from episodic mass-loss may also modulate the maser and IR emission as they pass through the photodissociation and grain destruction regions. With even better angular resolution and sensitivity (perhaps with James Webb Space Telescope (JWST) and/or the next generation of ground-based nulling interferometers like the Large Binocular Telescope Interferometer and Keck Telescope Interferometer) it should be possible to observe these fainter, compact structures within the larger nebulosity, if they exist.

Previous 11 μm results from Monnier et al. (1997), Danchi et al. (2001), and Blöcker et al. (2001) have suggested the presence of multiple, concentric density enhancements (shells) surrounding the central star superposed on a r⁻² wind, characterized by an azimuthally symmetric, but non-steady-state dust outflow, i.e., $\rho _{\rm dust} \propto\kern-8pt{/}\,\, r^{-2}$. This, in turn, implied a time-dependent mass-loss, possibly in episodic/periodic events such as a "superwind" phase. The images presented in Section 4 show evidence for a complex distribution in the CS dust, but also reveal an asymmetric excess aligned to the northwest (toward Cyg OB2). The differences with these earlier results are not surprising given the better resolution and Fourier u, v coverage in the MMT/AO images. However, it should be noted that Monnier et al. acknowledge in their conclusions that deviations from spherical symmetry, particularly emission from dense or clumped material 350 mas from the central star, would necessarily change the meaning of their fits (and likewise for Danchi et al. 2001). In fact, Figures 3 and 4 show, through direct imaging, the presence of warm dust emitting throughout mid-IR wavelengths precisely at this distance from NML Cyg. Moreover, this emission is concentrated at PAs ranging from ∼−30° (CS envelope) to ∼−60° (dust facing Cyg OB2), or equivalently at alignments of ∼120°–150°. Monnier et al.'s and Danchi et al.'s journals of observations show that they made eight observations with resolution substantially (at least 20%) better than ∼035, and these observations were made with the interferometer aligned along PAs between 102° and 140° (see each author's Table 1). Thus, their observations just happened to be oriented with the NW excess emission resolved in our images. It is possible that the most external CS shell inferred from these interferometric observations may in fact be better explained by the asymmetric excess in our 9.8 μm image.

Using a combination of Keck aperture masking and IOTA interferometry, Monnier et al. (2004) found at 2.2 μm an elongated structure with a diameter of ∼50 mas. The orientation of this structure is orthogonal to the PA of the inner component of our images, represented by our "core" Gaussian fits. Monnier et al. (2004) interpreted this structure as an equatorial enhancement in the dust envelope, orthogonal to NML Cyg's maser outflow.

Our mid-IR images, when analyzed in combination with the previous observations discussed above, clearly demonstrate the complexity of NML Cyg CS environment. Any model of NML Cyg that attempts to explain the observations must thus take into account the observed asymmetries as well as the physical interaction between NML Cyg's wind and the UV radiation from Cyg OB2. In particular for NML Cyg, high-angular resolution (≲03) combined with more complete coverage of the Fourier u, v plane is crucial to further investigate asymmetries in its complex CS environment.

NML Cyg is part of our larger program to study the CS environments of cool stars near the empirical upper luminosity boundary in the HR diagram with evidence for high-mass loss and observed instabilities. As such, NML Cyg was perhaps the best candidate to have an extensive CS nebula, possibly like VY CMa. However, NML Cyg's CS nebulosity is more concentrated around the star and appears quite different compared to VY CMa, as imaged by Smith et al. (2001). Even though NML Cyg is heavily obscured, the extent of the CS nebula is much less than the almost 10'' nebula surrounding VY CMa (≳10⁴ AU at VY CMa's distance of 1.5 kpc). It is possible that the more distant material in NML Cyg's CS nebula has been largely dissipated by the winds and radiation pressure inside the Cygnus-X super bubble and is below our detection limits. If NML Cyg were not located in such close proximity to Cyg OB2, it might show a much more extended nebula comparable to VY CMa. However, VY CMa's mass-loss rate, (2–4) × 10⁻⁴M_☉ yr⁻¹ (Danchi et al. 1994), is 3–6 times higher than for NML Cyg and this difference may also have significant bearing on the more extensive CS nebulosity.

Our subarcsecond angular resolution AO mid-IR images of NML Cyg provide a new understanding of the complex geometry and physics of the CS environment of this high luminosity cool hypergiant. By spatially resolving the optically thick dusty envelope, we directly image the structures responsible for the creation of a deep 10 μm silicate absorption feature. This structure, which follows the same orientation of NML Cyg's maser outflow, is orthogonal to a near-IR equatorial enhancement found by Monnier et al. (2004).

By analyzing the mid-IR colors of structures located at increasing distance from the star, we observe a trend in which the optical depth of the dust decreases in the outer parts of the CS envelope. For the first time we isolate an asymmetric excess, at a distance of ∼520–870 AU, NW from the star, with colors consistent with the emission of hot, optically thin dust. We interpret this emission as the signature of the interaction of the NML Cyg CS envelope with the strong UV flux generated in the nearby Cyg OB2 association. This interaction was predicted in our previous paper (Schuster et al. 2006a) to explain the shape of the inverted PDR we imaged with the HST at optical wavelength. Our new mid-IR observations strongly support the validity of our previous results and the model we proposed for their explanation.

We are pleased to acknowledge interesting conversations with David Boboltz regarding his VLA observations. NASA provided support for this work. Observations reported here were obtained at the MMT Observatory, a joint facility of the Smithsonian Institution and the University of Arizona. This work is based in part on observations made with Spitzer, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. Support for R.D.G. was provided by NASA through grants 1215746 and 1256406 issued by JPL–Caltech to the University of Minnesota. This work is based in part on observations made with the NASA/ESA HST, obtained at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555.

Facilities: MMT, Spitzer, HST, UMN OBO