INFRARED SPECTROSCOPY OF SYMBIOTIC STARS. IX. D-TYPE SYMBIOTIC NOVAE

V407 Cyg was discovered as a nova-like variable when it underwent an outburst in 1936.⁴Ahnert et al. (1949) found brightenings with a period of ∼670 days. No spectroscopic observations were obtained of the 1930s outburst. However, V407 Cyg appears in the catalog of Merrill & Burwell (1950) as emission line object MHα 289-90 = AS 453 with a "combination" spectrum. Little further notice was taken of V407 Cyg until the late 1980s. Based on a review of the discovery and subsequent classification of V407 Cyg as a Mira, Meinunger (1966) remarked that this was no doubt a nova–Mira binary system.

Using broadband photometry extending as blue as the U band, Esipov et al. (1988) were not able to detect the hot companion of V407 Cyg. Following Esipov & Yudin (1986), Esipov et al. (1988) did detect emission from [N ii], which requires a nebula excited by a hot star. Nonetheless, as reviewed by Munari et al. (1990), at this time the status of V407 Cyg as a symbiotic was questioned. Munari et al. (1990) found the spectral type of the V407 Cyg Mira to be M6 III. At low resolution the absorption spectrum and hydrogen emission lines closely resembled o Ceti.

Munari et al. (1990) presented a summary of optical (B) photometry of V407 Cyg from the time of discovery to 1987. From this an ephemeris for the maxima of the Mira, B_max = JD 2,429,710 + 745E, was derived. From broadband optical and near-infrared photometry, Kolotilov et al. (1998) confirmed the period of 745 days and found a Mira-like K-band amplitude of ≲1 mag. Optical and near-IR light curves from 1970 to 2000 were presented by Munari & Jurdana-Šepić (2002). They found periods, which are a function of color, in the range 745 to 756 days. Munari et al. (2011) concluded that a 745 day pulsation was a good fit to near-IR photometry. However, Kolotilov et al. (1998) and Munari & Jurdana-Šepić (2002) determined that the pulsation of the Mira is highly variable from cycle to cycle. Shugarov et al. (2007) found a period of 763 days.

Munari et al. (1990) and Kolotilov et al. (2003) concluded that the pulsation period for V407 Cyg is the longest of any known symbiotic system. They furthermore noted that this period exceeds that of typical isolated Miras and is in the period range given by Engels et al. (1983) for OH/IR obscured Miras. They speculated that the outbursts of the V407 Cyg white dwarf companion prevent the Mira from becoming obscured. Munari et al. (1990) stated that V407 Cyg presents an opportunity to observe an unobscured Mira of exceptionally long period. Using J-H-K infrared colors and the Glass & Feast (1982) period–luminosity relation, Munari et al. (1990) calculated a distance of 2.7 kpc for V407 Cyg.

Tatarnikova et al. (2003a), Tatarnikova et al. (2003b), and Shugarov et al. (2007) reported the presence of a strong Li i λ6708 line in the spectrum of the Mira. Li is destroyed in the interior of main-sequence stars and material convected to the surface of red giants is devoid of Li (Brown et al. 1989). However, in agreement with the V407 Cyg observations a very small set of Li-rich AGB stars are known. These are massive AGB stars, >4 M_☉, where Li has been synthesized in hot-bottom burning during the third dredge-up (Boothroyd et al. 1993). Observations of LMC and SMC Li-rich AGB stars show that these stars cover a restricted range in absolute bolometric magnitude, M_bol ∼ −6 to −7 (Smith & Lambert 1990). This implies both a lower and upper bound for the mass, M ∼ 4–8 M_☉.

Mid-IR observations show that the V407 Cyg dust is a mix of silicate and graphite grains (Yudin 1999). This also suggests a fairly recent third dredge-up of hot-bottom-burned material that reduced previously carbon-rich material to C/O < 1. From a fit of the infrared spectral energy distribution (SED) Kolotilov et al. (1998) found an inner radius dust temperature of 600 K for silicate and 1000 K for graphite grains. Yudin (1999) suggested an inner radius of ∼5 R_*. Using literature photometry the Whitelock et al. (1994) calibration of the mass-loss rate for oxygen-rich objects indicates a mass-loss rate of 6 × 10⁻⁷ M_☉ yr⁻¹. Kolotilov et al. (1998) and Yudin (1999) found similar values. Kolotilov et al. (1998) remark that the mass-loss rate is at least an order of magnitude smaller than that expected from period–mass-loss rate relations (for example, Whitelock et al. 1994).

Munari et al. (1990) interpreted dust obscuration events as being orbitally related. Based on possible historic dust obscuration events in V407 Cyg Munari et al. (1990) suggest an orbital period of ∼43 yr with the last inferior conjunction occurring at JD 2,441,600. Munari et al. (1990) also conclude that the system is totally detached.

In large part, interest in V407 Cyg has been driven by the active phases. These occurred in the late 1930s, early 1990s (Munari et al. 1994), 1998–2002 (Kolotilov et al. 2003; Tatarnikova et al. 2003a), and most significantly in 2010. Munari et al. (2011) differentiated between the low-amplitude, long-lasting outbursts typical of symbiotic systems and larger amplitude nova outbursts. The latter are not associated with the rapid thermonuclear runaway events of the former. Low-level outbursts can be seen on historic light curves of V407 Cyg cited above. Shugarov et al. (2007) detected flickering in the 1998 outburst, confirming mass transfer to a white dwarf secondary. In spectra reported by Tatarnikova et al. (2003a) the Hα profiles show changing accretion by the hot star over the 1993–2002 period.

The 2010 event was a nova explosion resulting in an optical magnitude increase of ∼10 mag over quiescent values reported before the 1930 event (Nishiyama et al. 2010; Ahnert et al. 1949). Following the 2010 outburst of V407 Cyg, Shore et al. (2011) proposed reclassifying V407 Cyg as a member of the rare class of symbiotic recurrent novae (SyRNe). Other members of this class are S-type symbiotic systems that contain a very massive white dwarf. The continuum from the nova dominated the blue spectrum in 2010 March (Munari et al. 2011). Coronal emission lines of [Fe x] λ6375, [Fe xi] λ7890, [Ar x] λ5535, and [Ni xii] λ4233 were discovered (Munari et al. 2010) along with gamma-ray emission (Abdo et al. 2010), the first detection of gamma-ray emission from a symbiotic nova. Lü et al. (2011) discussed the long orbital period of D-symbiotics as a requirement for gamma-ray emission in symbiotic systems.

Munari et al. (2011) present results from optical and infrared observations of the 2010 nova episode. They found a very fast wind from the nova that is rapidly decelerating within the Mira wind. The extensive, massive Mira wind contains unaffected regions despite the high-energy processes occurring near the Mira. However, Munari et al. (1990) found that the radiation field and outburst from a white dwarf companion inhibited dust formation in the Mira wind.

Additional insight into the interaction between the nova and the Mira comes from the Cho & Kim (2010) detection of ²⁸SiO v = 1 and v = 2 masers in V407 Cyg. These masers have velocities of −28.1 and −27.9 km s⁻¹, respectively. They also reported an H₂O maser at a velocity of −31.5 km s⁻¹. Deguchi et al. (2011) concluded that the nova outburst of 2010 disrupted the SiO masers. The high velocity component of the maser disappeared two weeks after the nova outburst, roughly synchronized with the appearance of strong X-ray emission (Deguchi et al. 2011). Deguchi et al. (2011) proposed that the maser region was disturbed by the passing nova shock. Following from this, the nova-SiO maser and hence white dwarf–Mira separation is ∼20 AU, implying an orbital period of at least decades.

The spectroscopic evolution of the 2010 nova outburst of V407 Cyg is described by Shore et al. (2011). A spectrographic time series reported by Shore et al. (2011) supports the SyRN classification. Shore et al. (2011) found velocities in Hα approaching 3000 km s⁻¹. The highly ionized species had broad profiles, for instance [Fe x] λ6375 extended from −400 to +600 km s⁻¹. The Na i D line underwent systematic changes similar to those seen in SNe and related to the illumination of historic enhancements of mass-loss. The shock temperature of 2 × 10⁷ K for the first 30 to 60 days after the outburst implies a shock velocity of ∼800 km s⁻¹.

Shore et al. (2011) note that if the stellar mass is ∼4 M_☉ and the white dwarf mass is ∼1.2 M_☉, as suggested by the SyRN classification, the mass ratio is about 3. This sets constraints on the ratio of the Roche radius to the semimajor axis, with R_RL/a ∼ 0.3. Taking the Roche radius to be larger than the Mira radius, then the semi-major axis a > 6 AU, and the period is long, clearly at least the 43 yr suggested by Munari et al. (1990). A 43 yr period places the Roche radius at 6 AU, which is within the outer Mira atmosphere (Reid & Menten 1997). The quiescent accretion ratio of the white dwarf is ∼10⁻⁸ to 10⁻⁹ M_☉ yr⁻¹ (Munari et al. 1990), which is a few percent of the Mira mass-loss rate of ∼10⁻⁷ M_☉ yr⁻¹ Yudin (1999). The accretion rate, while highly uncertain, seems typical for detached Mira–white dwarf binaries (de Val-Borro et al. 2009) and suggests a period of >100 yr or a highly elliptical orbit.

A SED derived from literature photometry of V407 Cyg is shown in Figure 3. In Figure 4 the phased radial velocity data are presented. We find a best fit pulsation period of 770 days from the velocity measurements. There is a difference of 25 days, ∼3% of the period, between the photometric and spectroscopically derived periods.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

V1016 Cyg was first cataloged by Merrill & Burwell (1950, MHα 328-116 = AS 373) as a star with red magnitude ∼12 having strong Hα emission. Attention was focused on V1016 Cyg in 1964 due to a nova-like outburst (Hoffleit 1965). A light curve dating back to the 1920s can be found in FitzGerald et al. (1966). V1016 Cyg underwent brightening to a photographic magnitude near 15 in the 1950s prior to the nova-like event. The nova-like event resulted in a nearly 5 mag increase in brightness during 1964 (FitzGerald et al. 1966). Ultimately V1016 Cyg brightened to 10.66 at V (Philip 1969).

FitzGerald et al. (1966) reported a low-excitation emission line spectrum. The additional presence of late-M features on an archival near-IR spectrum dating from 1947 drove them to suggest that V1016 Cyg is a symbiotic variable. O'Dell (1967) subsequently noted continuum from a hot star with intense Balmer emission lines. Swings & Allen (1972) found that the near-IR (HKL) photometry of V1016 Cyg could be fit by 1000 K dust. Confirmation of the symbiotic nature followed when Harvey (1974) found that the K-band magnitude of V1016 Cyg underwent 1.0 mag changes with a ∼450 day period, suggesting that a Mira variable is present. Harvey (1974) was the first to suggest the combined presence of a Mira and a hot star in the system. Schild et al. (1992) show a medium-resolution K-band spectrum of V1016 Cyg taken in 1990. The spectrum displays the presence of dust emission with the signal increasing to the red. Broad H₂O absorption bands are present and the CO 2–0 band head can be seen. Brackett γ appears in emission. The signatures of H₂O and CO are characteristics of oxygen-rich cool M giants and especially Mira variables. Mürset & Schmid (1999) gave the spectral class of the giant as M7.

Spectroscopic observations in [N ii] and [O iii] by Solf (1983) revealed that the V1016 Cyg system contained a compact bipolar nebula with lobes separated by 040 spatially and 51 km s⁻¹ in velocity. The expansion velocity projected along the major axis is 120 km s⁻¹ with a nebular mass of 2 × 10⁻⁴ M_☉. Bipolar structure was reported by Hjellming & Bignell (1982) from Very Large Array (VLA) maps at 23.1 GHz. Girard & Willson (1987) provide an alternative conical geometry to model the optical spectra. However, deeper images and spectroscopy in [N ii] confirmed a nebula of about 20'' diameter, containing a bipolar kinematic feature extending 3'' from the center with projected velocities of ±30 km s⁻¹ (Corradi et al. 1999).

For V1016 Cyg Watson et al. (2000) provide citations to distance estimates, ranging from 2.1 to 10 kpc, that were derived from various techniques. An ephemeris of the times of maxima for the Mira is provided by Kenyon & Webbink (1984): Max(K) = JD 2,444,101 + 471E. Munari (1988) produced a revision: Max(K) = JD 2,444,852 + 478(±5)E. Taranova & Shenavrin (2000) presented 20 yr of near-IR photometry of V1016 Cyg. A pulsation period of 470 ± 5 days is found for the Mira. Using the Glass & Feast (1982) IR period–color relation for Miras, Taranova & Shenavrin (2000) found a distance of 2.8 ± 0.6 kpc and a luminosity for the Mira of 8600 L_☉. Parimucha (2003) determined a pulsation period of 474 ± 2 days and a spectral class of M7. Again, using the Glass & Feast (1982) period–color relation and standard relation for pulsating variables, Parimucha (2003) computed a distance of 2.9 ± 0.8 kpc. Assuming a distance of 3.9 kpc, Mürset et al. (1991) estimated that the hot component has a radiation temperature of 150,000–125,000 K, radius ∼0.3 R_☉, and luminosity ∼35,000 L_☉.

Taranova & Shenavrin (2000) found an optically thick dust envelope around V1016 Cyg, which had an optical depth of a few in the non-thermal near-IR. During the period of observation (1978–1999) the dust was dispersing. Munari (1988) proposed that the dust obscuration events are periodic with a period of ∼6 yr. IRAS photometry by Anandarao et al. (1988) previously had revealed two dust shells, one at ∼300 K with radius ∼60 AU and a second more massive shell at ∼60 K with radius ∼7500 AU. Taranova & Shenavrin (2000) found the dust to have a temperature of ∼600 K, a radius of 1400 R_☉, and a mass of ∼3 × 10⁻⁵ M_☉. Angeloni et al. (2010) also identified two dust shells by fitting the V1016 Cyg SED with a 3000 K Mira plus a 1000 K dust shell at 20 AU and a 400 K dust shell at 60 AU. Literature photometry and the Whitelock et al. (1994) calibration of the mass-loss rate for oxygen-rich objects indicate a mass-loss rate of 8 × 10⁻⁶ M_☉ yr⁻¹. The distance found by Parimucha (2003) from the K-period relation for Miras results in a peak dust shell flux that correlates well with other symbiotics. Infrared Space Observatory (ISO) SWS spectra of V1016 Cyg show strong silicate features (Angeloni et al. 2010).

From the Raman scattered He ii λ4850 line, Jung & Lee (2004) determined a mass-loss rate, which they said should be "taken with caution" due to high sensitivity to the system kinematics, of ≲ 4 × 10⁻⁷ M_☉ yr⁻¹ for V1016 Cyg. This is more than an order of magnitude less than the total (gas plus dust) mass-loss rate derived from infrared colors. From fitting the Raman O vi λ6825 line Lee & Kang (2007) computed a terminal wind velocity of 11 km s⁻¹. Watson et al. (2000) noted that V1016 Cygni is one of the brightest radio sources among the symbiotic systems. They suggested that this is related to the nova-like outburst in 1965. They concluded that at least the 6 cm radiation is the result of interacting winds associated with the Mira and white dwarf.

Schild & Schmid (1996) used spectropolarimetric observations of the Raman 6825 Å line to determine a change in position angle over 3 yr that is consistent with a binary period of 80 ± 25 yr. The inclination is moderate, 60° ± 20°, with the orbital plane oriented in agreement with the bipolar model of Solf (1983). Subsequent observations reported by Schmid & Schild (2002) suggest that the binary period is several times longer than originally proposed but confirm that the binary axis is near the plane of the sky. Using imaging data from the Hubble Space Telescope (HST), Brocksopp et al. (2002) measured an angular separation of 42.4 mas between the white dwarf and the Mira of V1016 Cyg. Assuming a distance of 2 kpc, this is a projected separation of 84 AU. Brocksopp et al. (2002) concluded that the orbital period is >540 yr, which differs significantly from the orbital period determined from Raman spectroscopy suggesting the orbit is highly eccentric.

An SED derived from literature photometry of V1016 Cyg is shown in Figure 5. In Figure 6 the phased radial velocity data are presented. From the velocities we find a best fit pulsation period of 492 days. The photometrically and spectroscopically determined periods differ by 14 days, ∼3% of the period.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

RX Pup is a CD star (CD −41 3911) and hence one of the longest studied, prototypical symbiotic novae. In her spectral survey of these objects Fleming (Pickering & Fleming 1897) found the Balmer lines in emission. Swings & Struve (1941) presented a summary of visual magnitude and spectrum variations from the turn of the 20th century through the early 1940s. Swings & Struve (1941) noted similarities of the visual RX Pup spectra to spectra of other stars that are now classified as symbiotics. RX Pup appears in the emission line object catalogs of Wray (1966, Wray 16-17), Sanduleak & Stephenson (1973, SS73 8), and Henize (1976, Hen 3-138). Investigators in the 1970s found that the optical spectrum has a low excitation phase, reminiscent of a Be star with a disk, and a high excitation phase, with Wolf–Rayet features similar to those of RR Tel (Allen & Wright 1988). Mikołajewska et al. (1999) reported on evidence that RX Pup underwent nova outbursts in the 1890s and 1970s. This implies a period of 80 yr, although due to a lack of data in the 1930s the period could be as short as 40 yr.

Feast et al. (1977) found a large amplitude infrared variation (ΔK = 1.4 mag) indicating that the RX Pup system contains a Mira variable. Feast et al. (1977) also concluded that the SED required a ∼900 K dust component in addition to the Mira. Barton et al. (1979) detected H₂O and CO bands in the 2 μm spectrum, again suggesting the presence of an oxygen-rich Mira variable. This, combined with the hot star characteristics mentioned above, led Barton et al. (1979) to propose that RX Pup be classified as a symbiotic star. H₂O and CO bands were confirmed by Whitelock et al. (1983). Based on near-IR spectroscopy, Schulte-Ladbeck (1988) found the spectral class of the Mira to be later than M5. Mürset & Schmid (1999) determined a spectral class of M5.5. The recurring nova outbursts and the symbiotic status clearly show that RX Pup is a member of the small class of symbiotic nova.

Whitelock et al. (1983) detected long-term trends at J and L suggesting a dust formation episode in the late 1970s. After correcting for the Mira pulsation variations Mikołajewska et al. (1999) found long-term changes in the light curve from 1974 through 1996, 2.5 mag in J and 0.8 mag in L. The L-band changes were not fully in phase with those in the J band but were not antiphased either. The changes in reddening in the Mira are not correlated with changes in reddening in the hot component and the emission line forming region. Thus, the obscuration only affects the Mira. Mikołajewska et al. (1999) also noted that the Mira pulsation amplitude is variable depending on the state of the hot component.

Ultraviolet spectra from IUE, reported on by Kafatos et al. (1985), show the presence of an accretion disk around the white dwarf. The emission lines have typical full widths of 200 km s⁻¹. The hot component ionizing the wind was found to have a color temperature in the range 50,000–100,000 K and a luminosity of ∼1000 L_☉. Seaquist & Taylor (1987) detected a variation of flux density and angular structure with frequency of the radio continuum. This is in agreement with thermal bremsstrahlung continuum originating from a stellar wind. Assuming a wind velocity of 60 km s⁻¹ and a stellar distance of 1 kpc they compute a mass-loss rate of ≳ 4 × 10⁻⁶ M_☉ yr⁻¹. Mürset et al. (1997) detected X-rays from RX Pup. The radiation from an optically thin region with a temperature of 7×10⁶ K suggests an origin in either colliding winds or an accretion disk.

The distance to RX Pup has been estimated with multiple techniques. Klutz et al. (1978) found a distance of 1 kpc based on interstellar Na i lines. From J-band photometry Whitelock et al. (1983) determined a period of 580 days with an amplitude of ∼1.8 mag. This period and the Mira J-band period–luminosity relation allowed Whitelock (1987) to derive a distance of 1.5 kpc. Allen & Wright (1988) determined two distances, one of 1.5–1.6 kpc, based on the X-band Mira period–luminosity relation, and a second of 0.7 kpc, based on mid-IR fluxes. They suggested a compromise distance of 1 kpc. From mid-IR colors Kenyon et al. (1988) derived a distance of 1.25 kpc for RX Pup and A_K = 0.8. Kenyon & Webbink (1984) provide an ephemeris of Max(K) = JD 2,442,810 + 580E. Using near-IR photometry, Mikołajewska et al. (1999) found a 578 day period for the Mira. From the Mira period–luminosity relation at K Mikołajewska et al. (1999) computed M_K = −8.7, for which the corresponding distance is 1.8 ± 0.5 kpc. Mikołajewska et al. (1999) warn that both the reddening for the Mira period–luminosity relation for variables of very long period and the circumstellar reddening of RX Pup are uncertain. During various epochs the RX Pup E(B − V) has been estimated to range from >1.0 to 3.3. Uncertainties in circumstellar reddening are a problem in the distances for all the D-symbiotics. Gromadzki et al. (2009) used SAAO K-band photometry to obtain a pulsation period of 575 ± 8 days and gave an ephemeris for maxima of 2,442,238 + 575E. With a K-band period–luminosity relation Gromadzki et al. (2009) estimated a distance of 1.6 kpc.

Kenyon et al. (1988) used mid-IR colors to determine a dust temperature of 325 K. Anandarao et al. (1988) found inner and outer dust shell radii of 43 and ∼3900 AU with temperatures of 364 and 45 K. These temperatures are much lower than the ∼900 K determined from near-IR photometry by, for example, Feast et al. (1977). From an analysis of JHKL magnitudes Kotnik-Karuza et al. (2007) report 900 and 700 K dust temperatures for RX Pup in different dust obscuration events. They determined a maximum grain size of 1.1 and 2.0 μm in these events. The visual optical depth was up to 7.5 mag. IRAS LRS spectra show strong silicate features (Anandarao et al. 1988). Literature photometry and the Whitelock et al. (1994) calibration of the mass-loss rate for oxygen-rich objects indicate a mass-loss rate of 8 × 10⁻⁶ M_☉ yr⁻¹. Similar photometric estimates are provided by Kotnik-Karuza et al. (2007) and Gromadzki et al. (2009).

A critical parameter of the RX Pup binary is the separation of the components. Allen & Wright (1988) noted that the system does not have the characteristics expected for Roche lobe overflow. Therefore, the minimum separation is on the order of twice the Mira radius or ∼7 AU. A better lower limit on the separation can be determined from the radio observations. Seaquist & Taylor (1992) reviewed radio and IR data and showed that the radio (cm through mm) observations are fitted by optically thick free-free. The infrared from about 100 μm to shorter wavelengths is dominated by blackbody radiation from the dust. From the radio data Seaquist & Taylor (1992) argued for a binary separation in the range ∼30–70 AU. The upper limit on the separation can be found from the structure surrounding the stars. Mikołajewska et al. (1999) concluded that the permanent dust shell around the Mira suggests a binary separation >50 AU, which corresponds to a P_orb > 200 yr.

Parts of the RX Pup system have been spatially resolved. Using a coronagraph and narrowband [N ii] filter Paresce (1990) found a one-sided jet-like feature co-aligned with the semimajor axis of the 6 cm radio nebula. Corradi & Schwarz (2000) presented high-resolution spatially resolved spectra of RX Pup. They confirmed the Paresce (1990) detection of extended [N ii] emission. However, they concluded that the [N ii] region has a velocity decreasing with distance from RX Pup, so it is not a jet. Outflow velocities in excess of 80 km s⁻¹ are measured. They suggested that the nebula is bipolar and has a size hundreds of times larger than the binary separation. Hollis et al. (1989) showed that RX Pup has a radio structure composed of at least three nearly co-linear components. They identified the strongest feature with the hot star. Under the assumption that RX Pup is 1.5 kpc distant, the other features are separated from the hot star by 230 and 590 AU. They suggested these other features are ejecta from activity at earlier epochs. Hollis et al. (1989) reviewed the radio detection history of RX Pup and noted that RX Pup undergoes episodes of radio flux variations.

Shore et al. (2011) highlighted similarities noted by Mikołajewska et al. (2002) between the S-type SyRN systems with massive white dwarfs and RX Pup. RX Pup appears to currently have similarities to the pre-nova observations of V407 Cyg, which underwent a nova outburst in 2010 and is now believed to be an SyRN system. However, the outflow velocities in RX Pup are significantly less than in V407 Cyg.

An SED of RX Pup derived from literature photometry is shown in Figure 7. In Figure 8 the phased radial velocity data are presented. From the velocities we find a spectroscopic pulsation period of 605 days. The difference between the photometric and spectroscopic periods is 30 days, ∼5% of the period.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Dokuchaeva (1976) reported an anonymous star, later named HM Sge, had brightened from m_ph ⩾ 17 mag to V = 11 mag in 1975. The star was a red variable with an emission line spectrum. Puetter et al. (1978) found a 950 K blackbody with strong 8–13 μm silicate emission to be a good match to the 2.5 to 8 μm spectrum. They also reported CO absorption at 2.3 μm and concluded that HM Sge contained a reddened cool star. Davidson et al. (1978) was the first to suggest that HM Sge was a symbiotic star and estimated a distance of 1–3 kpc. Allen (1980) argued that HM Sge was a slow nova and that slow novae were the same phenomena as symbiotic stars. Kenyon & Truran (1983) expanded on this point and placed HM Sge in the group of symbiotic stars that undergo eruptions. The properties of these "symbiotic novae" include outbursts extending over decades and long period binary systems containing an M giant.

Interest generated by the nova-like brightening resulted in observations of HM Sge at wavelengths from the radio through the X-ray. Allen (1981) reported that HM Sge is one of three symbiotics (RR Tel, V1016 Cyg, and HM Sge) detected in the X-ray region. Kwok & Leahy (1984) found that the X-ray flux exceeded that expected from thermonuclear reactions on the surface of the white dwarf during an eruption and proposed that the X-ray flux results in shocked gas when the fast wind from the white dwarf collides with the slow wind from the M giant. Purton et al. (1983) reviewed radio observations of HM Sge from 1977 to 1980. They found that the radio continuum emitting region in HM Sge had a diameter of ∼02.

A number of observers began time-series photometry in the late 1970s. From such narrowband near-IR observations Bregman (1982) suggested that the late-type star is a Mira with a period of 550 days. Using K-band photometry, Taranova & Yudin (1983) reported a period of ∼500 days. From near-IR photometry Lorenzetti et al. (1985) settled on a pulsation period of 540 days. Using a decade-long time series, Munari & Whitelock (1989) determined a pulsation period of 527 days for HM Sge with maximum light occurring at JD 2,440,310(±30) + 527(±2.0)E. Yudin et al. (1994) confirmed the period of 527 days from J-band data and provided an ephemeris for minima. From J-band photometric observations over the 1978–1999 interval, Taranova & Shenavrin (2000) found a pulsation period of 535 ± 5 days. These periods all support the classification of the M giant as a Mira. In addition, 1–5 μm narrowband photometric observations by Bregman (1982) showed that HM Sge has water vapor absorption bands characteristic of an oxygen-rich Mira. Schulte-Ladbeck (1988) estimated a spectral class for the Mira of later than M5, while Mürset & Schmid (1999) concluded that it has an M7 spectral class.

Stauffer (1984) found the effective temperature of the white dwarf to be varying, ranging from less than 70,000 K to 160,000 K with a timescale of months. The proposed mechanism was a hydrogen shell flash in the accretion envelope of the white dwarf, which removed the envelope exposing the hotter core of the white dwarf. Nussbaumer & Vogel (1990) reported a longer term variation with the radiation temperature of the white dwarf increasing from below 40,000 K in 1976 to 170,000 K in 1989, while the luminosity remained constant at ∼10⁴ L_☉. They suggested that increasing radiation temperature of the white dwarf is common following nova eruptions in symbiotic stars.

As with other D-symbiotics, the Mira period–luminosity relation can be applied to derive a distance but with the caveat that the circumstellar reddening is typically large and uncertain, leading to a large range of distances. For example, using Balmer line ratios of HM Sge, Blair et al. (1983) found a rather small value of A_V ∼ 1.2 mag, while also from Balmer line ratios Davidson et al. (1978) estimated A_V to be twice as large, ∼2.5. At the other extreme Bregman (1982) determined A_V = 12 from the infrared Brackett lines. Assuming a typical bolometric magnitude for the Mira, Lorenzetti et al. (1985) derived a distance of 4 kpc. Kenyon et al. (1986) noted that this produces an unreasonably large bolometric luminosity for the hot component in the binary. Kenyon et al. (1986) found a reddening of A_K = 2 for the Mira. Insight into this problem was provided by Kenyon et al. (1986), who concluded that consistency with optical observations of the hot star requires that the hot star lie outside the dust shroud surrounding the Mira. Based on a pulsation period of 540 days, Whitelock (1987) suggested a period–luminosity relation distance of 2.3 kpc. Taranova & Shenavrin (2000) derived a distance of 1.8 ± 0.4 kpc.

From 1–5 μm narrowband photometric observations Bregman (1982) found 1000 K dust present. During the mid-1980s HM Sge underwent infrared fading, which was attributed to dust obscuration. Munari & Whitelock (1989) showed that the infrared SED could be fit by 2500 K and 800 K blackbodies. During obscuration events the blackbody temperatures required to fit the SED remained constant, but the 2500 K Mira component decreased compared to the 800 K component. Schild et al. (1992) presented a medium-resolution K-band spectrum taken in 1990. The spectrum was dominated by dust emission increasing to the red with the only spectral line being weak Brackett γ emission. Schmid et al. (2000) reported that in 1998 the spectrum of the Mira was present in the red and near-IR. These observations agree with the report by Taranova & Shenavrin (2000) of long-term variations in the dust obscuration with maximum obscuration occurring near JD 2,447,500 (1988 December). Time-series near-IR photometry showing obscuration episodes contemporaneous with our near-IR spectroscopy is presented by Shenavrin et al. (2011). The emission line spectrum and SED of HM Sge are modeled in the context of interacting winds by Formiggini et al. (1995) and Angeloni et al. (2010). They find dust shells at 400 and 1000 K. ISO-SWS spectra of HM Sge show strong silicate features (Angeloni et al. 2010). Literature photometry and the Whitelock et al. (1994) calibration of the mass-loss rate for oxygen-rich objects indicate a mass-loss rate of 8 × 10⁻⁶ M_☉ yr⁻¹.

Solf (1984) reported on the first spatially resolved information about the HM Sge circumstellar shell. With spatial–spectral mapping a bipolar mass flow of ∼200 km s⁻¹ was found that collimate into two narrow lobes separated by 15. In addition, there are two low-velocity features of 02 separation and a shell of 05 diameter expanding at ∼60 km s⁻¹. Solf (1984) suggested that the expansion rate of the nebula implies a distance of ∼400 pc and a corresponding nebular mass of ∼10⁻⁴ M_☉. Corradi et al. (1999) presented narrowband images of HM Sge. There is a spatially extended circumbinary region of diameter 04, which is much larger than the expected spatial size of the binary orbit. In addition HM Sge has an extended structure of collimated knots out to about 9''. Corradi et al. (1999) suggested that these originate as a fast collimated wind from the white dwarf and accretion disk. With the Mira no longer obscured by dust in 1998 Schmid et al. (2000) observed the polarization in the Raman O vi line and derived the position angle of the binary axis. The Raman line polarization and the axis of the binary system are aligned parallel or perpendicular to the structure seen in the radio and narrowband optical images.

MERLIN maps of HM Sge at 6 and 18 cm, presented by Eyres et al. (1995), show that the inner nebula is bipolar with the peaks separated by ∼016. VLA measurements at 1.3 cm indicate that the peaks are moving apart at ∼9 mas yr⁻¹. Eyres et al. (1995) concluded that an interacting wind model agrees with the observations. Assuming a shock velocity of 57 km s⁻¹, derived from the brightness temperature, Eyres et al. (1995) found a distance to HM Sge of 3.2 kpc. Wallerstein et al. (1984) had previously demonstrated that the optical emission line profiles in HM Sge required an origin in a conical nebula resulting from interacting winds. Richards et al. (1999) continued monitoring the radio emission of HM Sge through the 1990s. They found hot spots that appear to be rotating in an inclined disk with a period of 90 yr. They revised the distance to ∼1 kpc and estimated the separation of the two stars to be 25 AU. The biconical outflow was attributed to winds from the nova outburst interacting with the pre-existing cool wind, while the disk was suggested to have an inclination of 60°.

Using HST and VLA observations, Eyres et al. (2001) were able to identify the binary components directly in images. The projected angular separation is 40 ± 9 mas with a binary axis position angle of 130° ± 10°. This agrees with the polarization position angle. Requiring consistency with a component separation of 50 AU (Richards et al. 1999) results in a distance of 1250 ± 280 pc. Eyres et al. (2001) found that the nebula shows two distinct regions. They associate these with the cool component wind or shielding of the nebula from the hot radiation field.

On a much smaller scale, the system contains an accretion disk around the white dwarf. Lee & Kang (2007) found that fits to high spectral resolution observations of the O vi λ6825 Raman line require a Keplerian thin disk around the white dwarf with an outer disk rim velocity of 26 km s⁻¹ and a terminal velocity for the Mira wind of 10 km s⁻¹.

Schild et al. (2001) presented mid-IR ISO observations of HM Sge. They reported that the mid-IR spectrum is dominated by silicate dust emission at a condensation temperature of 800–1600 K. The mid-IR spectrum also contains line emission from a number of high-excitation forbidden lines. Angeloni et al. (2007) interpreted the ∼500 km s⁻¹ FWHM of the mid-IR emission lines as originating between the two stars in a wind interaction shock. The ionization stages represented and the strength of the lines are in agreement with a 4 × 10⁶ K plasma temperature (Mürset et al. 1997). Angeloni et al. (2007) reported a best fit with a preshock density of 5 × 10⁵ cm⁻³ and a post-shock density of 10⁸ cm⁻³.

An SED derived from photometry of HM Sge taken from the literature is shown in Figure 9. In Figure 10 the phased radial velocity data are presented. From the velocities we find a best fit pulsation period of 521 days. The photometric and spectroscopic periods are in fairly good agreement for HM Sge, differing by 6 days, ∼1% of the period.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Fleming & Pickering (1908) first identified RR Tel as a peculiar variable star. Based on Harvard plates, Payne (1928) found a period of 384 days, which was revised to 387 days after another two decades of observation (Gaposchkin 1950). Mayall (1949) did a full search of the Harvard plate material for RR Tel historical variations. From 1889 to 1930 RR Tel had very low amplitude visual variability. The visual amplitude then grew until in the early 1940s it was ∼3 mag with a period of 387 days. RR Tel then underwent a slow nova outburst in 1944 (see light curve in Mayall 1949). Following the nova outburst, Mira variations in the blue (photographic) were not visible due to the brightness of the nebular emission line spectrum (Feast et al. 1983b). Infrared photometry confirmed the 387 day period of the Mira (Feast et al. 1983b). Kenyon & Bateson (1984) found that visual estimates of the brightness started showing periodic variability about 1976. From these data a period of 374 days was determined. Kenyon (1986) provided an ephemeris of Min(V) = JD 2,442,551.7 + 374.2E. Heck & Manfroid (1985) claimed that the period is not constant but varies between 350 and 410 days. Gromadzki et al. (2009), using SAAO data, computed a Mira period of 385 ± 4 days and an ephemeris for near-IR maxima of JD 2,442,207 + 385E.

RR Tel appears in the Henize (1976, Hen 3-1811) catalog of southern emission line stars. It was reported as having a nova-type spectrum by Thackeray (1950) with Payne-Gaposchkin (1955) confirming the spectral changes originally reported by Thackeray (1950). Following the 1944 outburst, an F supergiant spectrum evolved into an emission line spectrum with strong Fe ii and Ca ii lines. Velocities, measured by Thackeray (1953) in the early 1950s, ranged to −865 km s⁻¹, consistent with material ejected by a nova. Seaquist (1977) detected radio continuum at 5.0 GHz. Subsequent work has shown the existence of shocks in the circumstellar material and the nearby ISM (Contini & Formiggini 1999). RR Tel has a rich emission line spectrum in the visual with many forbidden and permitted lines present for a large range of ionization states. Thackeray (1977) presented an extensive survey of the nebular spectrum in the visual. Typical velocities are in the range −64 km s⁻¹ for hydrogen to −56 km s⁻¹ for helium. The nova outburst is reviewed by Nussbaumer & Dumm (1997).

Allen et al. (1978) detected strong H₂O and weak CO bands in the K-band spectrum of RR Tel, implying that the late-type star in this system is an oxygen-rich, late-type Mira variable. The oxygen-rich nature of RR Tel is confirmed by the presence of TiO bands in the red (Webster 1974). Schulte-Ladbeck (1988) estimated the RR Tel spectra class as ⩾M5. Mürset & Schmid (1999) classified the spectral class as M6.

Mürset et al. (1991) suggested a temperature for the hot star of 140,000 K with a radius of 0.15 R_☉ and a luminosity of ∼5000 L_☉. For the same temperature Nussbaumer & Dumm (1997) estimated a radius of 0.11 R_☉ and a luminosity of 3700 L_☉. RR Tel was cataloged as a luminous supersoft X-ray source (SSS) with a bolometric flux of 1.3 × 10³⁷ erg s⁻¹ by Greiner (2000). There are only 10 galactic SSSs known and all are believed to be massive white dwarfs in binaries (Greiner 2000; Di Stefano 2010).

Feast et al. (1983b) concluded that the dust is at least partly heated by the hot star in the system. Their near-IR photometry gave a dust temperature of ∼1000 K. Based on IRAS data, Anandarao et al. (1988) found a 248 K dust shell with a radius of 173 AU. Penston et al. (1983) computed the radius of the emitting area of the nebula to be between 47D^2/3 and 935D^2/3 AU, where D is the distance in kpc. The expansion velocity of the nova as computed by Thackeray (1977) gives a radius of ∼670 AU for the nebula in rough agreement with a distance of a few kpc. Eriksson et al. (2008) found that the Fe ii fluorescence nebula has a mean radius of 450 ± 50 AU. Munari & Whitelock (1989) noted that among all the well-studied D-symbiotics, RR Tel was the only one that had not been observed to undergo a dust obscuration event. However, Kotnik-Karuza et al. (2007) recently reported such an event. They also found a dust temperature derived from near-IR photometry of 750$^{+100}_{-50}$ K. Contini & Formiggini (1999) and Angeloni et al. (2010) model the emission line spectrum and SED of RR Tel. They find that 400 K and 1000 K dust is present. Both IRAS-LRS (Anandarao et al. 1988) and ISO-SWS (Angeloni et al. 2010) spectra show that silicate dust is present but the silicate signature is weak, suggesting graphite grains are also present. As for V407 Cyg this suggests that the current oxygen-rich outer layers result from third dredge-up of hot-bottom-burned material. Literature photometry and the Whitelock et al. (1994) calibration of the mass-loss rate for oxygen-rich objects indicate a mass-loss rate of 3 × 10⁻⁶ M_☉ yr⁻¹.

From the characteristics of the circumbinary nebula Thackeray (1977) estimated a lower limit to the distance to RR Tel of ∼2.5 kpc. Feast et al. (1983b), using a J-band period–luminosity relation, set a distance of 3.6 kpc. Whitelock (1988) revised the distance to 2.6 kpc based on a K-band period–luminosity relation. Using the same relation, Gromadzki et al. (2009) found a distance of 2.5 kpc. The K₀ − [12] color measured by Gromadzki et al. (2009) gives a mass-loss rate of 1.6 × 10⁻⁶M_☉ yr⁻¹.

From Raman line spectropolarimetry Schmid & Schild (2002) determined orbital motion of 13 yr⁻¹ for RR Tel. In 1999 the P.A. was 110°. In the case of a circular orbit the orbital period would be ∼300 yr. The semimajor axis of an orbit of period 300 yr is 56 AU or 22 mas at a distance of 2.5 kpc.

A SED derived from photometry of RR Tel taken from the literature is shown in Figure 11. In Figure 12 the phased radial velocity data are presented. From the velocities we find a best fit pulsation period of 375 days. The photometric and spectroscopic periods differ by 10 days, ∼3% of the period.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The spectra of the D-type SyN systems in the 1.6 μm region (Figures 1 and 2) are dominated by FeH, CO Δv = 2, and atomic lines, all of which are photospheric in origin. Line weakening from dust obscuration is a conspicuous feature in the 1.6 μm spectra of some objects. Figure 15 presents a set of spectra from three program stars, centered at 1.562 μm, that shows a progression of line weakening with HM Sge being the most obscured. A time series of 1.6 μm spectra of HM Sge is presented in Figure 16. HM Sge underwent a dust obscuration event during the time that the spectra were taken. Significant changes in the line depths are apparent including periods when the spectra are nearly featureless.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

A correlation does not exist between the Mira pulsation phase of HM Sge and the visibility of the lines. In isolated, unobscured Miras the 1.6 μm CO is observable throughout the pulsation cycle (Hinkle et al. 1982). The 1.6 μm second overtone CO lines require photospheric conditions and are not populated at the temperature of the ∼1000 K circumstellar dust. The spectrum has been obscured by either scattering or continuous emission by dust.

In Figure 17 the dates of our H-band spectroscopic observations of HM Sge are compared with the near-IR photometry of Taranova & Shenavrin (2000) and Shenavrin et al. (2011). The nearly featureless spectra, observed early in our investigation (JD ∼ 2,450,000), correspond to a prolonged dust obscuration event that is conspicuous by decreased flux in the J band. This suggests that scattering by dust produces the obscuration at 1.6 μm. The obscured spectra around JD ∼ 2,452,500 do not correspond to a long-term photometric obscuration event. Thus, we suggest that dust obscuration events of short duration also occur.

Zoom In Zoom Out Reset image size

A few observations of the program stars were made at wavelengths longer than 1.6 μm. The 2.31 μm region (Figure 18) contains CO Δv = 2 lines and H₂O lines. Some of the lines in this spectral region have low-excitation energies potentially allowing an origin in a circumstellar region. However, no features appear in our spectra at multiple or shifted velocities or with conspicuously larger depth suggesting an origin other than photospheric. Two additional spectra were taken of each of the objects in Figure 18 to explore the region around the 2–0 band origin. Similarly, no conspicuous circumstellar features are present. We also have spectra of other 2.3 μm regions in V1016 Cyg and RR Tel. These also have shallow lines and a photospheric spectrum without circumstellar features.

Zoom In Zoom Out Reset image size

Four HM Sge spectra centered at 2.31 μm, spanning over 30 yr (Table 8) from 1979 through 2012, are compared in Figure 19. The 1979 through 1997 spectra are identical except for observational uncertainties and telluric features. Photometric monitoring shows substantial changes in the dust obscuration during this time period (Schmid et al. 2000; Taranova & Shenavrin 2000). The 2012 spectrum has the same velocity as the others, but the features are much weaker. The 2.3 μm CO lines are shallow, with a maximum central depth of 28%, and relatively broad, with FWHM = 25 km s⁻¹, at an unchanging velocity (Tables 7 and 8) of −2 km s⁻¹, a few km s⁻¹ blue of the Mira center-of-mass velocity. The excitation temperature of these CO lines is 2200 ± 500 K, a close match to the Mira photometric temperature (Feast et al. 1983a).

Zoom In Zoom Out Reset image size

The 2.3 μm spectra of the D-symbiotic novae are unlike any other late-type spectra. While the D-symbiotics contain a Mira variable, D-symbiotics and Miras have very different 2.3 μm spectra. Figure 20 includes a spectrum of the 430 day period oxygen-rich Mira R Cas. R Cas has an M7e spectral type (Keenan et al. 1974) and strong water vapor lines (Hinkle et al. 2000), making it a good reference spectrum for HM Sge. The CO first overtone region in isolated Miras such as R Cas is characterized by very deep lines with blended velocity components. The velocities of the lowest excitation lines differ from those of the higher excitation lines. Both the line depth and velocities are the result of a contribution to lower excitation lines by a cool molecular region at a different velocity (Hinkle et al. 1982; Tsuji 1988, 2008). Figure 20 also includes a spectrum of the 635 day period M10 OH-IR Mira NV Aur. NV Aur is moderately obscured with I−K ∼ 6.6 (Lockwood 1985). As illustrated in Figure 20 the CO low-excitation lines are weaker in NV Aur than in R Cas and the water lines stronger. While the high-excitation (J'' ∼ 80) CO lines are not present in NV Aur, the low-excitation CO lines have multiple velocity components. Both the weaker CO low-excitation lines and the missing high-excitation lines can be attributed to the later spectral type of NV Aur. In summary, spectral features arising in the cool molecular region and characterizing the 2.3 μm spectra of Miras make no contribution to the 2.3 μm spectra of D-type SyNe.

Zoom In Zoom Out Reset image size

We postulate that the 2.3 μm spectrum in D-SyN symbiotics is formed by 1000 K dust continuum emission combined with the photospheric spectrum scattered by the 1000 K dust layer. Scattering in a stationary dust layer would allow the velocity to be fixed, which is observed in HM Sge. While scattering increases to the blue, the 1.6 μm velocities are photospheric. For instance, in the 1989 September 13 HM Sge observation, which covers both the 1.6 and 2.3 μm regions, the CO second overtone lines at 1.6 μm have a velocity of −7 km s⁻¹ matching the photospheric pulsation velocity curve of the Mira, while the CO first overtone lines at 2.3 μm have a velocity of −2.2 km s⁻¹. The SED of HM Sge (Figure 7) shows that the 1000 K dust makes a strong contribution to the spectrum red of 2 μm. The 1.6 μm region is well blue of the peak of the 1000 K dust emission in the region where photospheric radiation becomes more important.

Based on the absence of 2.3 μm CO lines attributable to the 1000 K gas, a limit for the CO column density can be set at ≲ 10¹⁹ molecules cm⁻². Assuming the solar C/H abundance this is ≲ 10²² H cm⁻². Smith & Gehrz (2005) modeled the dust in three bipolar symbiotics. They found that the mass in the 1000 K dust is negligible compared to the mass in the cooler dust. The reason for this is that the dust mass scales as the grain properties, density, and luminosity to the inverse forth power of the temperature. So a region with hotter dust can be prominent in the SED but contain relatively little mass compared to a cooler, equally prominent region. Thus, it is not surprising that 1000 K dust continuum makes a significant contribution to the spectrum but that the gas along the line of sight is insufficient to produce observable features in the 2.3 μm region.

Included in the broad wavelength coverage of the HM Sge FTS spectra is the Brackett γ line at 2.16 μm. This line is present in emission ∼25% above the continuum with an FWHM of ∼117 km s⁻¹. The heliocentric velocity of the peak of the Brackett γ emission is +47 km s⁻¹ on 1979 May 11. Emission lines of similar width and velocity have been seen in the optical and ultraviolet (Solf 1984; Mueller & Nussbaumer 1985) although Wallerstein et al. (1984) reported a Balmer emission line velocity of −2 km s⁻¹. Angeloni et al. (2007) found a broader, ∼500 km s⁻¹, FWHM for mid-IR emission lines. Wallerstein et al. (1984) connected the optical emission lines with the wind interaction shock. The velocity width of the Brackett γ profile suggests a relation to the wind interaction shocks but not to the 1000 K gas.

Figure 21 illustrates spectra in the 4.66 μm region for the same three program stars shown in Figures 15 and 18. The spectrum of HM Sge has no photospheric absorption, a continuum presumably from dust, and CO emission lines. The spectrum is similar to those seen from dust disks around post-AGB binary stars (Hinkle et al. 2007). The velocity of the emission lines, −0.7 ± 1.5 km s⁻¹ is, within the uncertainty, a match to the 2.3 μm velocity. The 4.6 μm emission line velocity is very different from the velocity of the hydrogen Brackett γ emission line. The spectrum of V1016 Cyg has absorption at the stellar velocity and slightly blueshifted emission. The V407 Cygni spectrum is dominated by stellar absorption. We note that none of the spectra of these three stars have cold circumstellar CO absorption features. Circumstellar CO absorption lines formed in an expanding circumstellar shell are ubiquitous 4.6 μm spectral features of isolated late-type giants (Bernat 1981).

Zoom In Zoom Out Reset image size

The flux of the low-J 2–1 ¹²CO emission lines in HM Sge increases with increasing rotational quantum number J, implying that the lines are not saturated. If the excitation temperature equals the temperature of the dominant dust component at this wavelength, 1000 K, the line flux should peak at J ∼ 13. The two ¹³CO lines in Figure 21, R12 and R13, would then be at the peak flux. For an excitation temperature of 1000 K the ¹²C/¹³C ratio would be about 10, which is a typical value for isolated Miras (Little et al. 1987). Assuming that the CO lines are optically thin and have a 1000 K excitation temperature, a 2 kpc distance for HM Sge results in a CO emission measure of n(CO)V_sh ∼ 4 × 10⁴⁶ molecules, where V_sh is the volume of the emitting region. Taking the carbon abundance as solar, the mass of the emitting region is ∼2 × 10²⁶ g, ∼9 × 10⁻⁸M_☉. For HM Sge the blackbody radius that is required to match the observed flux for 1000 K dust at a 2 kpc distance is 1.21 × 10¹⁴ cm or ∼3 Mira radii. From the relations of Smith & Gehrz (2005) the mass of the 1000 K dust is 7 × 10⁻⁹ M_☉. The gas-to-dust ratio is then ∼120. Smith & Gehrz (2005) suggested a typical gas-to-dust ratio of 230 for symbiotic objects. From the limit to the column density and the emission measure the size of the gas emitting area is >10²⁴ cm². If spherical, the emitting region is 1 × 10¹² cm ≈ 0.1 AU deep. The area found for the dust is ∼4 × 10²⁸ cm², suggesting that the gas is limited to an edge of the dust region.

The absence of detectable circumstellar lines in our spectra suggests that both the cool molecular region of the extended Mira atmosphere and the expanding circumstellar shell are absent. The weakness or absence of masers (Cho & Kim 2010), which are also formed in the molecular zone (Cotton et al. 2010), is additional proof of the disruption of the extended Mira atmosphere. Various authors (e.g., Whitelock 1987; Munari 1988; Munari et al. 1990) have invoked UV radiation from the hot companion to destroy dust or inhibit its formation in the circumstellar environment of D-symbiotics. However, in the case of UV disruption the hemisphere of the Mira shadowed from the white dwarf would have an inner molecular and cool expanding circumstellar shell. The interacting wind model includes wind penetration into the shadowed hemisphere of the Mira (Kenny & Taylor 2005) and explains the absence of spectral features from this region.

V407 Cyg, which underwent the most violent and recent eruption of any of the program stars, has the least dust contribution to the SED (Figure 3) and to the infrared spectra (Figures 15, 18, and 21). Both the white dwarf wind velocity and likely the white dwarf mass-loss rate are higher for this system than the others. The resulting mw ratio will be larger, resulting in a wind collision interface closer to the Mira. This appears to inhibit formation of dust perhaps through a narrowing of the wind interaction zone. The disruption of the weak D-symbiotic masers by nova outbursts (Deguchi et al. 2011) also suggests that the wind interaction zone is closer to the Mira during nova outbursts. In the contrasting case of HM Sge the wind interaction zone must be undergoing frequent dust formation events. The dust cannot have a lifetime of more than a few months in the wind interaction zone or the variations in the dust obscuration would have a much longer period than observed. Uneven flow or instabilities in the wind interaction zone between the wind shock fronts could result in enhanced dust formation (Folini & Walder 2000). As the white dwarf wind decreases, the apex of the wind interaction zone will move toward the white dwarf, and the interaction cone will open. This is also a possible source of instabilities.