Publications of the Astronomical Society of the Pacific 99:116-125, February 1987

THE RING-SHAPED NEBULAE AROUND FU ORIONIS STARS*

ROBERT W. GOODRICH Lick Observatory, Board of Studies in Astronomy and Astrophysics University of California, Santa Cruz, California 95064 Received 1986 October 27

ABSTRACT All five of the known FU Orionis stars are associated with ring-shaped nebulae. This type of nebula is not unique to the FU Orionis class, but it is rather unusual among other young stars. Deep CCD images and spectra are presented of three of the nebulae, those associated with V1057 Cyg, V1515 Cyg, and V1735 Cyg (= Elias 1-12). In the latter case the star is more highly reddened than the nebula, possibly indicating a localized concentration of dust near the star. The spectra of all three nebulae are clearly from reflection. A crude model for the dust shells is briefly discussed, and it is argued that the shells may indicate a relatively advanced evolutionary state for the FU Orionis stars. The shells may represent one lobe of a bipolar structure, with the opposite lobe hidden from view by dust in the equatorial "waist" of the structure. Alternatively, the elongated shape of the shells may be the result of the relative motion of the star through the parent molecular cloud, in which case there is only a single cavity. The FU Orionis outburst itself may then occur when the star reaches the edge of the cavity and fresh material is accreted. Some of the more recent models for the outburst itself are discussed and some observational tests of these models suggested. Key words: circumstellar matter-nebulae-stars: young

I. Introduction changes in FU Orionis stars have been put forth. Herbig The existence of the FU Orionis stars as a distinct class (1977) discusses several of these, and the reader is re- of variable stars has been known for some time now. ferred to that paper for further details. Subsequent to that Herbig (1977) gave a thorough review of both the observa- summary, however, some other theories have been pro- tional and the theoretical status of this phenomenon as it posed. On the basis of infrared spectra and excesses, both was known at that time. In 1977 there were three known Elias (1978) and Mould et al. (1978) proposed that the FU members: FU Orionis itself, V1057 Cygni, and V1515 Orionis stars were binaries, in which a large, late-type M Cygni. Herbigs paper helped to define the observational star was orbiting an F or G supergiant. Larson (1980) characteristics of the class—a large and usually rapid rise suggested that a single, rapidly rotating star could explain in brightness, an optical spectrum at maximum light re- the phenomenon and also provide a natural explanation sembling that of an F or G supergiant, a spectral class for the estimated frequency of occurrence of these types which is dependent on the wavelength region in which of outbursts. More recently, Hartmann and Kenyon the classification is made, association with a molecular (1985) have suggested that what we are seeing at maxi- cloud and a peculiar reflection nebula, a characteristic mum light is not the photosphere of a star at all, but rather infrared excess, etc. Subsequent to his work Elias (1978) an optically thick accretion disk undergoing an outburst. found what is regarded as the fourth member of the class, Little work has been presented on the reflection nebu- Elias 1-12 = V1735 Cygni, a star in the IC 5146 dark cloud lae associated with the FU Orionis stars. Herbig (1977) complex that had brightened by five or more magnitudes. noted that the ring-shaped nebulae associated with FU Recently Graham (1983; see also Graham and Frogel Orí, V1057 Cyg, and V1515 Cyg are rather unusual struc- 1985) has discovered yet another example, a star associ- tures, although not unique to these three stars. Similar ated with the Herbig-Haro (HH) object H H 57 in the nebulae have been found to be associated with the other Norma I association. This find was significant in being the two FU Orionis stars V1735 Cyg and HH 57 1RS 1. Her- first documented association of an FU Orionis star with an big also noted that while spectra of the nebulae had not HH object. been obtained, photographic plates taken with interfer- On the theoretical side, many explanations for the ence filters isolating the Ha emission line and the nearby sudden increase in brightness and the apparent structural continuum indicated that they are predominantly reflection. Duncan, Harlan, and Herbig (1981) performed area *Lick Observatory Bulletin No. 1055. photometry on the nebula associated with V1057 Cyg,

116 © Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System FU ORIONIS STARS 117

concluding that the nebula was fading in proportion to the TABLE I fading of the star, thus establishing conclusively the con- nection between nebula and star. Images This paper will begin by presenting some observational data on the morphology (Section III) and spectra (Section IV) of the nebulae surrounding three FU Orionis stars. Object Date Exposure The implications of these and other data for the evolution- (m) ary status of the stars are discussed in Section V, while Section VI reviews some of the outburst models currently being discussed in the literature. ■V1057 Cyg 24 Oct. 1984 60 25 Oct. 1984 60 25 Oct. 1984 60 II. Observations Images of the reflection nebulae around the stars V1057 V1515 Cyg 5 May 1984 10 5 May 1984 10 Cyg, V1515 Cyg, and V1735 Cyg were obtained with the 29 Sept. 1984 20 1-m Nickel telescope on Mount Hamilton, using a G.E.C. charge-coupled device (CCD) at the //17 V1735 Cyg 27 Sept. 1984 30 Cassegrain focus of the telescope. This detector has a 24 Oct. 1984 60 scale of 0''26 pixel-1 and a field of view of 1.'7 by 2.'5. A red 25 Oct. 1984 60 interference filter with a broad passband and high throughput (described by Djorgovski 1985) was used to Haro 2-249 24 Oct. 1984 30

avoid most of the strong emission lines from nearby San a Jose. To obtain lower noise and a wider areal coverage, RNO 54 3 Jan. 1986 1 3 Jan. 1986 3 several images of the fields were taken, often with different centers. These individual fields were then combined into the mosaics presented in Figure 1. Also shown in a 3-m images; all others from 1-m. Figure 1 are an image of the FU Ori star near H H 57 (from Schwartz, Jones, and Sirk 1984) and images of the two young stars RNO 54 and Haro 2-249. The H H 57 image used, with each pixel perpendicular to the slit represent- was taken on a photographic plate with the CTIO 4-m ing 0''71 on the sky. In the case of V1057 Cyg and V1515 reflector, while the RNO 54 image is a CCD montage Cyg, the slit was placed only across the nebulae to avoid from the 3-m Shane telescope on Mount Hamilton. The saturation of the CCD by the stars themselves. In the case Haro 2-249 was taken in the same manner and with the of V1735 Cyg, however, the star is heavily reddened and same equipment as the first three images. The relevant hence could be placed on the slit without worry. This has data for the individual exposures are given in Table I. the added advantage of obtaining a simultaneous long-ex- Another set of data was taken through a narrow-pas s- posure spectrum of the star itself. Even so, the exposure band interference filter centered on the red [S π] emis- of VI735 Cyg and its nebula was split in two to avoid sion lines at 6717 A and 6731 A. These observations were potential saturation of the CCD at the red end of the taken with the imaging mode of the CCD spectrograph on spectrum. Table II contains pertinent exposure informa- the 1-m telescope. A Hasselblad imaging lens of 110-mm tion and Figure 2 presents the spectra. focal length was used to give an effective focal ratio of about 2.4 and a scale of roughly Γ/Ι pixel-1. These images, III. Nebular Morphology covering almost 9' on the sky, were used to search for One rather striking characteristic of the nebulae of the [S ll] excesses in the vicinities of the stars. Light scattered FU Orionis stars is their ring-like morphology. This ex- into the optical path by infrared-emitting LEDs in the tends even to the cases of Η Η 57 1RS 1 and FU Ori spectrograph limit somewhat the usefulness of these data, (although the latter is also associated with a bright fan- but there is no evidence for any [S ll]-emitting nebulosity shaped nebula). As Herbig (1977) noted, there is a bright around these three stars. Hence, these images are not arc of nebulosity extending from FU Ori, similar to a part presented in this paper. of the rings seen around V1057 Cyg and V1515 Cyg. Of Spectra were obtained of these three nebulae using the course, one can find other examples of elliptical ring- Cassègrain spectrograph of the 3-m Shane telescope at shaped reflection nebulae among young stars. RNO 54 Mount Hamilton. This spectrograph employs a virtual- and Haro 2-249 (Fig. 1) are two well-defined examples, phase, 800 X 800 pixel T.I. CCD at an effective focal ratio and V1331 Cygni is seen almost at the center of a ring of of 1.2. A 420-line mm"1 grism gave a dispersion of 5.5 A nebulosity. Partial rings or "arcs" are seen in some other pixel-1 and a spectral resolution of 14 A. A 2'-long slit was stars such as SU Aurigae and Ζ Canis Majoris. However, a

F, (e) (f)

Fig. 1—Images of the nebulae near four FU Orionis stars: (a) V1057 Cyg, (b) V1515 Cyg, (c) V1735 Cyg, and (d) the HH 57 star. Also shown are the ring-shaped nebulae associated with two Ha-emission stars, (e) RNO 54 and (b) Haro 2-249. North is up and east to the left on all images, and the tick marks are separated by 30".

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System FU ORIONIS STARS 119 look through a catalog such as Parsamian and Petrossian axis of its nebula, but there is evidence for some fainter (1979) shows a predominance of/an- shaped nebulae asso- nebulosity lying even further away to the northeast, per- ciated with young stars—the fan-shaped nebulae near R haps indicating that if one could get a deeper image of the Monocerotis and LkHa 208 are two good examples. Rings entire nebula V1057 Cyg may appear to lie on the major or partial rings are the exception rather than the rule. axis.) Bechis and Lo (1975) have interpreted the reflection If we take the point of view that the nebulae surround- nebula of V1057 Cyg in terms of a more-or-less conical ing FU Orionis stars are indeed unusual among the nebu- surface which is seen tangentially to the southwest side. lae associated with young stars, we might ask what clues The ring morphology certainly suggests that the dust may they might give us about the FU Orionis phenomenon. be in a shell-like configuration, and with this in mind, I Before proposing a model, however, it is interesting to have constructed a very simple model to qualitatively test note that the stars themselves lie near the edges of the this idea. rings, and generally lie on the major axes of the ellipses Specifically, I will assume that the dust forms an ellip- defined by the rings, although in the case of V1057 Cyg soidal shell with the star at one end of the ellipsoid. this is not clear. (V1057 Cyg appears to lie on the minor Because, in general, the star appears to lie on the projected major axis of the ring, the ellipsoid must be pro- TABLE II late—a star along the short axis of an oblate spheroid will appear projected on the minor axis, which may actually be Spectra the case for V1057 Cyg. The dust shell is assumed to be both physically thin and optically thin to scattering. In both V1057 Cyg and RNO 54 the observed widths of the Object Exposure Slit p.a. Arcsec rings are 5% or less of the ring radii. V1515 Cyg, on the (m) along slit other hand, may have a relatively thicker shell. The assumption of low optical depth to scattering means that effects due to multiple scattering may be ignored. A given V1057 Cyg 60 157 63 model may be parameterized by two quantities, then: e is V1515 Cyg 20 90 25 the ellipticity of the shell and θ0 is the observation angle V1735 Cyg 40a 0 29 (defined as zero along the long axis of the spheroid). These two variables may in principle be constrained by observations of the shape of the projected ring and the position of two separate 20 minute exposures. the star within that ring, although in practice the bright-

7000 8000 5000 6000 Wavelength (A) Fig. 2-Spectra of the nebulae associated with the FU Orionis stars (a) V1057 Cyg, (b) V1515 Cyg, and (c) V1735 Cyg. The spectrum of the star V1735 Cyg is shown in panel (d), and is clearly reddened much more than its nebula.

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System 120 ROBERT F. GOODRICH ness of the superimposed star and the nonuniformity of where To is the scattering optical depth (<< 1) of the shell the nebula may prevent accurate determination of one or seen normal to the surface. Of course, when χ = 90° this both of these parameters. formula predicts an infinitely bright rim due to the as- The brightness of the observed reflection nebula at any sumption of an infinitely thin shell. However, there are given point is determined by the distance of that point some qualitative conclusions which can be made from from the star, r, the illumination angle, φ, and the angle these models. First, an intensity peak forms in the interi- at which the line of sight passes through the shell, χ. ors of some of the models. As the distance to the star, r, These variables are shown in Figure 3, and the formula decreases Σ can approach infinity unless the cosine of the relating the brightness is given by illumination angle φ approaches zero even faster than r2. This effect is very dependent on the precise shape of the Tq cos φ v — shell at these small radii, however. If the parts of the shell ^ r 2 cos χ ? closest to the star are better represented by a cone, for instance, then the cos φ term would always be zero and there would be no internal intensity peak. Note that this is also equivalent to removing all of the dust from the near vicinity of the star (so that To is zero within some distance fo)· Another qualitative effect seen in the models is that as the observation angle θο approaches zero (for a given ellipticity e ) the contrast between parts of the nebula near the rim and the interior of the ring decreases. (This contrast also decreases as e decreases.) This is in qualitative agreement with V1515 Cyg, which lies relatively closer to the center of its nebula (and hence has a smaller θο) than the other FU Orionis stars—the V1515 Cyg nebula also has the least contrast between rim and interior. On the other hand, the physical thickness of the V1515 Cyg dust shell is larger than that of the other stars, and this will also decrease the contrast between rim and interior. Another way to analyze these models is to calculate the relative rim brightness as a function of the azimuthal angle around the rim. The rim is defined by the condition cos χ = 0, and after "renormalizing" one can show that the relative rim brightness is given by

"in Βços4>r \C '/ J-Vß

where C is the curvature of the nebula at the rim and accounts for the relative path lengths of different lines of sight through parts of the nebula with different curva- tures. There is an overall normalization which is lacking in this picture of an infinitely thin shell, so it is most appro- priate to plot such rim brightnesses on a semilog scale. To compare with the theory, I have chosen RNO 54, which is not an FU Orionis star but has a well-defined elliptical ring that may be studied in some detail. The ring of V1057 Cyg is very nonuniform, while the V1735 Cyg ring may STAR suffer from differential extinction (discussed in Section IV). The nebulae associated with both HH 57 and V1515 Fig. 3-A schematic drawing of the simple ellipsoidal-dust-shell model Cyg are small and difficult to measure, especially close to discussed in the text. The star lies at one end of the long axis of a prolate the star. Figure 4 shows the predicted rim brightnesses ellipsoid. The angle between the axis and the line of sight is θο, while the angle between the line of sight and the normal to the shell at any given for two models and the measured brightness profile of point is χ. The variable φ is the angle between the normal and the radius RNO 54. The solid line in Figure 4 shows the brightness vector to the star. distribution predicted for the model which most closely

-100 0 100 Position angle

Fig. 4-A comparison beween the observed rim brightness of the RNO 54 nebula and that predicted by the dust-shell model described in the text. The solid line is the model assuming a uniform dust density throughout the shell, while the dashed line assumes a r 2 falloff with distance from the star. Neither model matches the observational points well. reproduces the observed ellipticity and position of the IV. Spectra of the Nebulae star within the RNO 54 nebula. This profile is clearly The spectra taken with the 3-m Shane reflector are much too shallow compared to the data. The dashed line 2 shown in Figure 2. It is clear from these spectra that these shows the same model with a r " falloff in the dust distri- nebulae are indeed all reflection nebulae. Ha can be seen bution imposed. This curve is steeper than the first but is in emission in the spectrum of the V1057 Cyg nebula and still not as steep as the observations. Of course, any in absorption in the spectrum of the V1515 Cyg nebula. number of reasons for suspecting the models may be Also included in Figure 2 is the spectrum of V1735 Cyg entertained, but apart from those assumptions mentioned itself. It is clear from this figure that the star V1735 Cyg is above it is also possible that the reason for the discrepancy more highly reddened than the visible nebula. The long- lies in the dust distribution around the star. If the dust is slit spectra of the nebula was used to determine whether more abundant closer to the star, then one would expect a this differential reddening was due to a gradient across steeper brightness profile, as is observed. We note that the nebula toward the star, or whether it might be due to the observed brightness profile of the V1057 Cyg nebula, something more localized. Three regions of the spectrum if it could be corrected for nonuniformities, is apparently were chosen to compute the reddening difference be- fairly flat, which is what is predicted by the models. tween star and nebula: 4810 Â-5350 A, 6000 Â-6550 A, Of course, these models were only calculated for illus- and 6610 Â-7170 A. Under the assumption that the red- trative purposes. More sophisticated models might in- dening law is similar to the "normal" interstellar law as clude different dust distributions as a function of radius given by Whitford (1958), the difference in E _ between and the effects of multiple scattering and a finite thickness B V shell. The observed shapes of some edge-on examples of star and nebula was computed as a function of distance reflection nebulae like that of LkHa 208 might be used along the slit. The three wavelength bands allow three instead of an ellipsoidal model. Better yet, a picture such combinations from which EB_V may be calculated, two of as that of Adams and Shu (1985) might be used to specify which are independent. All three of these combinations all of these parameters—dust distribution, shape, etc. gave consistent results, so that they were averaged and

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System 122 ROBERT F. GOODRICH this average is presented in Figure 5. This figure demon- still capable of a highly collimated mass outflow, a charac- strates that between 20" and 55" away from the star the teristic taken by many authors as indicating a relatively reddening is sensibly constant (to within ± 0.2 mag). young star with a great deal of circumstellar material still Between the star and the inner edge of the visible nebula present. In this regard it might be mentioned that the the reddening increases by 2 mag. This may indicate that Η Η 34 system appears to have associated with it a faint at least 2 mag of the extinction to V1735 Cyg is due to a outer emission shell which is similar in morphology to the very localized source of dust (consistent with a flattened reflection rings discussed in this paper. This shell is only distribution for the material, but by no means proof of visible on very deep images of the Η Η 34 region, but can such). be made out in the Ha image by Mündt (1986). The If V1735 Cyg is assumed to have the (V —R) color of a relationship of this feature to the reflection nebulae asso- normal supergiant between spectral types F5 and G5, ciated with FU Orionis stars is not clear. then it is possible to calculate the total extinction to the On the other hand, the existence of the shells of dust star (and hence to the nebula). The monochromatic mag- around the FU Orionis stars may indicate that there has nitudes measured from Figure 5 are V = 18.6 and R = been enough time for the molecular outflows associated 15.4. According to FitzGerald (1970) the {V—R ) colors of with such young stars to have swept up the gas and dust supergiants vary from 0.35 to 0.67 in this range, implying within a cavity and piled it at the cavity edges. In this a color excess between 2.6 and 2.9 in (V—R) for V1735 sense the FU Orionis stars might be regarded as relatively Cyg. Assuming the Whitford extinction law, this implies a more evolved than other, more active stars. In the value of EB_V = 3.4 ± 0.2, quite consistent with the scheme of Adams and Shu (1985), for example, as the determination by Elias (1978). Then the extinction to the circumstellar envelope evolves, progressively more gas nebula is EB_V = 1.4, a value which may indicate the and dust are removed from streamlines near the axis of interstellar contribution to the reddening due to that symmetry, implying that more evolved envelopes will portion of the dark cloud in front of V1735 Cyg. have undergone more clearing of their interiors. Note that much of this discussion has implied (by anal- V. Evolutionary Status ogy with molecular and Η Η outflows) that the dust shells The evolutionary state of the FU Orionis stars relative associated with FU Orionis stars are bipolar, with the lobe to other Τ Tauri stars is unknown. Various attempts to pointing away from us, being obscured by dust surround- explain the FU Orionis phenomenon as a stage very early ing the waist of the two opposing shells. This is the in the life of a star have not proved tenable (see Herbig impression one gets from images of the Η Η 57 star, which 1977). However, the association of Η Η 57 with the FU shows a small projection of nebulosity on the side of the Ori star found by Graham indicates that at least this star is star opposite to the main body of reflection. On the other hand. Section VII describes a mechanism in which a ^ ^ ^ ^ ^ I I ^ I ^ ^ single cavity may be formed by motion of the star relative to the molecular cloud. VI. Outburst Models Numerous models have been proposed to attempt to account for the FU Orionis phenomenon, several of which are discussed by Herbig (1977) and by Hartmann and Kenyon (1985). I will restrict this discussion only to cq four of the more recent theories. The binary models proposed by Elias (1978) and by Mould et al. (1978) have Λ - V1735 Cyg two major difficulties associated with them. The first is nebula the lack of any radial-velocity variations in the spectrum of V1057 Cyg and FU Ori (Herbig 1977). This implies that CQ either the cool M star needed to explain the infrared spectrum must be a star of very low mass or that its orbit must lie so far from the hot F supergiant that there is no chance of mass outflow from one star to the other. If there Q ι ι I I I I I I I 1 1 is no mass outflow, then the existence of the M star is an 0 20 40 60 ad hoc construction to explain the infrared spectrum of Arcseconds from star the star and cannot help to explain the outburst itself. The Fig. 5-The relative reddening of V1735 Cyg and its nebula. The star is other problem is that it is hard to see how the M star could more highly reddened than the visible nebula, perhaps indicating a have remained undetected in all five of these systems fairly localized dust concentration near the star. before the outburst. (Now, of course, the light of the F

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System FU ORIONIS STARS 123 star dominates in the optical and one would not expect to the total line width of FU Ori was the same when the see any trace of a fainter late-type star.) splitting was seen as when the lines appeared single One very peculiar characteristic of the FU Orionis stars (Herbig, private communication). As Hartmann and is the variation in spectral type (and hence effective tem- Kenyon point out, it is difficult to support a model in perature) as one looks in diiferent regions of the spec- which two stars of similar spectral type are orbiting each trum. Herbig (1977) has noted that in the red region of the other at the velocities indicated. spectrum (5800 Â-6700 A) in 1975 the spectral type of The apparent splitting of the lines seen in FU Ori and V1057 Cyg appeared to be early G, but on blue spectro- V1057 Cyg might in reality be the addition of an emission grams (3800 Â-4800 A) of this star taken at the same time, component to the absorption lines. This might be due to the spectral type was early F. Again, the infrared spectra the passage of a transient spot, flare, or plage region of Mould et al. indicate the presence of what appears to be across the face of the star, for example as is seen in the an M-type photosphere in both V1057 Cyg and FU Ori. spectral lines of HR 1099 (Vogt and Penrod 1983). In this This argues for a stratification of the photospheric temper- case, the detection of line splitting would be sporadic, as atures, and with this in mind, Larson (1980) proposed that indeed seems to be the case. The temperature stratifica- the FU Orionis stars were single stars which were rotating tion might then be explained by extreme gravity darken- close to their breakup velocities, and that the outburst ing as discussed above. was a manifestation of a secular instability, which sheds An alternate model which still relies on orbital motion some of the angular momentum of the star and allows it to was presented by Hartmann and Kenyon. They showed contract and cool further. The release of energy involved that a simple accretion-disk model could reproduce many in this instability is seen as the outburst. Mould et al. of the features of V1057 Cyg and also account for the (1978) also proposed that a rapidly rotating star might splitting of the lines by the orbital velocities within the show gravity darkening severe enough to explain the disk itself. In this model the star itself is essentially invisi- stratification of photospheric temperatures observed. In ble, and the temperature stratification is accomplished by this case the temperature structure is laid out across the distributing the effective temperature as a function of face of the star as a function of latitude, with the cooler radius within the disk. The outburst of the disk might be temperatures found near the equator and the hotter near caused by an instability similar to that which appears to the poles. explain the outbursts of dwarf novae (e.g., Lin, Pa- Another tantalizing clue is the apparent velocity-split- paloizou, and Faulkner 1985), or it might be due to a ting of the lines in FU Orionis discovered by Herbig and sudden increase of the mass infall onto the disk. This Petrov and reported by Hartmann and Kenyon (1985). model removes the problem of the photospheric stratifi- These latter authors also discovered a similar effect in cation from the stellar surface to the surface of an accre- V1057 Cyg. The line splittings were found to be ^ 60 km tion disk where it occurs naturally. However, the mecha- s-1 in FU Ori and 30 km s_1 in V1057 Cyg. The line nism for the outburst and the existence and stability of profiles appear to be consistent with binary stars in which such optically thick accretion disks have not yet been the two components are equally bright and of the same demonstrated. Still, this model does make some predic- spectral type. There is, however, an immediate problem tions which can be tested. with the total luminosity of the system; since the stars are Because the temperature decreases as a function of now equally luminous, they must both have undergone radius in the accretion disk model and the orbital veloc- an outburst at the same time. This would argue for some ities within the disk also decrease as a function of radius, external trigger for the outburst, such as sudden accretion then this model predicts that the linewidths should de- of material from an accretion disk. In a close binary sys- crease as the effective temperature at which the lines are tem still trying to accrete material from its parent molecu- formed decreases. Note that this is opposite to the predic- lar cloud, it is possible that an accretion disk will form tion made by the gravity-darkening model (which has the surrounding both stars which is truncated at an inner lowest temperatures at the equator and hence at higher radius defined by a tidal resonance with the orbital period velocities). As Hartmann and Kenyon point out, the effect of the binary. As matter flows through the disk and accu- in their accretion-disk model for V1057 Cyg is small be- mulates at this inner limit, the disk may well become cause the lines in the optical part of the spectrum come thicker until the matter can finally fall onto the stars in a from a relatively narrow range in radius, and hence a more-or-less spherical accretion process, or until some narrow range in velocity. However, the effect should be as-yet-undefined instability allows a portion of the gas to enhanced if one could look with high-enough resolution overcome the resonance barrier and fall onto the stars. in the infrared, and the effect in the case of gravity-dark- The sudden accretion of a great deal of material onto the ening might be large enough to measure readily in the stars may then cause the observed outburst. However, optical. Clever choice of lines to use is also clearly advis- there has been no evidence for the velocity-crossing of the able. A cautionary note should be added, however. In lines normally seen in spectroscopic binaries. Further, many Τ Tauri stars an effect called "blue veiling" is ob-

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System 124 ROBERT F. GOODRICH served (Herbig and Goodrich 1986) in which the absorp- maintained within the low-density region, then the cavity tion lines in the blue are partially filled in by a blue walls will tend to fall back into the cavity. Clearly proper continuum. If this phenomenon also occurs in FU Orionis motions of the FU Orionis stars would be of interest to stars, then the rotational velocity determined from lines determine whether they are traveling away from the in the blue may be systematically different from the veloc- apparent centers of their reflection nebulae. These ity determined from red lines because of this veiling. proper motions should really be compared to the motions of the parent molecular clouds, thus complicating the VII. Discussion task. This explanation for the cavities may be applicable to The presence of flattened disks of material around the any of the theoretical models discussed in the previous FU Orionis stars might be inferred from a number of lines section. of evidence. Of course, the outburst model of Hartmann The rather unusual nebular morphology of the FU and Kenyon (1985) requires an accretion disk extending Orionis stars demonstrated in Figure 1 brings up the almost to the stellar surface. Any binary-star model which possibility of finding other FU Orionis stars by looking for has both stars undergoing an outburst simultaneously similar nebulae. The stars associated with these nebulae may require an accretion disk as the external trigger for may be either post- or preoutburst FU Orionis stars. the outbursts, although these models do not seem to RNO 54, used as a test case in Section III, was reported reproduce very well the observations discussed in the by Cohen (1980) as having a spectral type of F5 II, similar previous section. A single star rotating near its breakup to the spectral types associated with FU Orionis stars. velocity does not require an accretion disk, but will shed This spectral type was determined from low-resolution an equatorial disk (Durisen and Tohline 1985). On a larger Cassegrain spectra in late 1976 or early 1977. Two more scale, some sort of equatorial dust distribution must be recent higher-resolution spectra of RNO 54 were ob- invoked if it is assumed that the reflection nebulae sur- tained by M. M. DeRobertis using the 20-in coudé cam- rounding the FU Orionis stars are actually some type of era on the 3-m Shane reflector at Mount Hamilton. The manifestation of the bipolar outflows commonly seen dispersion was 0.5 A pixel-1 on the CCD detector, and around other young stars in the form of molecular out- the two regions covered were centered on the Na ID lines flows and HH objects. Levreault (1983) has observed and Ha. G. Herbig kindly provided an approximate spec- V1735 in CO and finds evidence for a molecular outflow tral type from these spectra, indicating an early-G star of which he interprets as a wind-driven shell. Perhaps the luminosity class lb or II, again similar to the spectral types shell sweeps all of the dust within it to the shell edge, common in postoutburst FU Orionis stars. However, the creating the observed reflection ring. However, only in Ha line is in emission (with an absorption feature just the case of the H H 57 star is a second reflection lobe seen blueward of line center) and appears similar to many among the FU Orionis stars. other "normal" young stars. The resolution of the spectra An alternate model for the formation of a single cavity is not high enough to determine an accurate rotational was discussed in some detail by Duncan et al. (1981). velocity, nor are the infrared characteristics of this star They hypothesized that the cavity may be the remnant of well determined, so that it cannot be claimed that this star a previous outburst, which swept up the gas and dust is related to the FU Orionis stars. Other stars such as SU within the cavity and deposited it at the cavity walls. The Aur and LkHa 233 (G2 III and A7, respectively; Herbig cavity walls presumably represent the point at which the and Rao 1972) are in a similar position of possessing ambient molecular cloud has stopped the original out- reflection nebulae which look like those surrounding FU flow. The cavity may then be assumed to be at rest with Orionis stars, but have not been studied in enough detail respect to the molecular cloud, and the relative motion to allow one to say whether or not they appear to be between cloud and star will result in the apparent dis- post-FU Orionis stars. (In any case, one might expect placement of the star from the cavity center. As these them to be pre-FU Orionis stars, and it is not clear that authors have discussed, a cross velocity of the star of 5 km they should appear spectroscopically unusual in any way.) s_1 will be enough to shift the star to its present position in It would be interesting to obtain, for instance, high-reso- 104 yr. One might further speculate that the prolate shape lution infrared spectra of these stars to see whether they of the cavity is caused by repeated outbursts which are have the CO and H2O absorption features found in FU originally more-or-less spherically symmetric, but due to Ori and V1057 Cyg by Mould et al. (1978) and indicative the relative motion between star and cavity will have the of a cool, late-type photosphere. Higher-resolution coudé effect of elongating the cavity along the direction of mo- spectra may also show whether these stars possess large tion of the star. The outbursts themselves may occur as rotational broadenings of their absorption lines, another the star reaches the end of the cavity and encounters a characteristic of FU Orionis stars. Finally, as mentioned fresh supply of material which it can then accrete, causing in the previous section, it would be interesting to try to a new outburst in the cycle. However, the stability of the detect a systematic trend in the rotational broadening of cavity must be questioned. If a sufficient pressure is not lines in the FU Orionis stars themselves with wavelength,

© Astronomical Society of the Pacific · Provided by the NASA Astrophysics Data System FU ORIONIS STARS 125 as a successful observation of this would differentiate Djorgovski, G. 1985, Pub. A.S.F., 97, 1119. between the accretion disk model of Hartmann and Duncan, D. K., Harlan, Ε. Α., and Herbig, G. H. 1981, A.]., 86, 1520. Durisen, R. H., and Tohline, J. E. 1985, in Protostars and Planets Π, Kenyon and the gravity-darkened single-star model of ed. D. C. Black and M. S. Mathews (Tucson: University of Arizona Mould et al. Press), p. 534. Elias, J. H. 1978, Ap. J., 223, 859. The author would like to express his thanks to many FitzGerald, M. P. 1970, Astr. Αρ., 4, 234. people who have contributed to this work. Drs. G. Her- Graham, J. A. 1983, IAU Circular, No. 3785. big and J. S. Miller provided much advice, equipment, Graham, J. Α., and Frogel, J. A. 1985, Ap. J., 289, 331. observing time, and encouragement, as well as the com- Hartmann, L., and Kenyon, S. J. 1985, Ap. J., 299, 462. munication of unpublished data. Very useful discussions Herbig, G. H. 1977, Ap./., 217, 693. Herbig, G. H., and Goodrich, R. W. 1986, Ap. /., 309, 294. with A. Noriega-Crespo, S. Ruden, Dr. P. Bodenheimer, Herbig, G. H., and Rao, Ν. Κ. 1972, Αρ. /., 174, 401. Dr. D. Lin, and Dr. F. Shu are also acknowledged. Dr. Larson, R. Β. 1980, M.N.R.A.S., 190, 321. R. Schwartz kindly gave permission to use the plate of the Levreault, R. M. 1983, Ap. J., 265, 855. Lin, D. N. C., Papaloizou, J., and Faulkner, J. 1985, M.N.R.A.S., 212, HH 57 star shown in Figure 1(d). Dr. B. Jones provided 105. support and advice on this project, which was partially Mould, J. R., Hall, D. Ν. B., Ridgway, S. T., Hintzen, P., and Aaron- supported by Gal Space grant G S 63-85 and NSF grant son, M. 1978, Ap. /. (Letters), 222, L123. AST 84-06843. Mündt, R. 1986, Canadian J. Phys., 64, 407. Parsamian, E. S., and Petrossian, V. M. 1979, Catalogue of Cometary REFERENCES Nebulae and Related Objects (Yerevan: Armenian Academy of Sci- ences). Adams, F. C., and Shu, F. H. 1985, Ap. /., 296, 655. Schwartz, R., Jones, B. F., and Sirk, M. 1984, A./., 89, 1735. Bechis, K. P., and Lo, K. Y. 1975, Ap./., 201, 118. Vogt, S. S., and Penrod, G. D. 1983, Pub. A.S.P., 95, 571. Cohen, M. 1980, A./., 85, 29. Whitford, A. E. 1958, A./., 63, 201.