Episodic Accretion in Young Stars
Marc Audard, Péter Ábrahám, Michael M. Dunham, Joel D. Green, Nicolas Grosso, Kenji Hamaguchi, Joel H. Kastner, Ágnes Kóspál, Giuseppe Lodato, Marina Romanova, Stephen L. Skinner, Eduard I. Vorobyov, Zhaohuan Zhu
aa r X i v : . [ a s t r o - ph . S R ] J a n Episodic Accretion in Young Stars
Marc Audard
University of Geneva
P´eter ´Abrah´am
Konkoly Observatory
Michael M. Dunham
Yale University
Joel D. Green
University of Texas at Austin
Nicolas Grosso
Observatoire Astronomique de Strasbourg
Kenji Hamaguchi
National Aeronautics and Space Administration and University of Maryland, Baltimore County
Joel H. Kastner
Rochester Institute of Technology ´Agnes K ´osp´al
European Space Agency
Giuseppe Lodato
Universit`a Degli Studi di Milano
Marina M. Romanova
Cornell University
Stephen L. Skinner
University of Colorado at Boulder
Eduard I. Vorobyov
University of Vienna and Southern Federal University
Zhaohuan Zhu
Princeton University
In the last twenty years, the topic of episodic accretion has gained significant interest inthe star formation community. It is now viewed as a common, though still poorly understood,phenomenon in low-mass star formation. The FU Orionis objects (FUors) are long-studiedexamples of this phenomenon. FUors are believed to undergo accretion outbursts during whichthe accretion rate rapidly increases from typically − to a few − M ⊙ yr − , and remainselevated over several decades or more. EXors, a loosely defined class of pre-main sequencestars, exhibit shorter and repetitive outbursts, associated with lower accretion rates. Therelationship between the two classes, and their connection to the standard pre-main sequenceevolutionary sequence, is an open question: do they represent two distinct classes, are theytriggered by the same physical mechanism, and do they occur in the same evolutionary phases?Over the past couple of decades, many theoretical and numerical models have been developedto explain the origin of FUor and EXor outbursts. In parallel, such accretion bursts have beendetected at an increasing rate, and as observing techniques improve each individual outburstis studied in increasing detail. We summarize key observations of pre-main sequence staroutbursts, and review the latest thinking on outburst triggering mechanisms, the propagation ofoutbursts from star/disk to disk/jet systems, the relation between classical EXors and FUors, andnewly discovered outbursting sources – all of which shed new light on episodic accretion. Wefinally highlight some of the most promising directions for this field in the near- and long-term. . INTRODUCTION Episodic accretion has become a recent focus of atten-tion in the star and planet formation community, turninginto a central topic to understand the evolution of proto-stars and accreting young stars. Initially identified as youngstellar objects (YSOs) with strong, long-lived optical out-bursts (
Herbig , 1966, 1977), FU Orionis objects (hereafterFUors) have triggered many investigations to understandthe eruptive phenomenon. Several reviews have been pub-lished (
Herbig , 1977;
Reipurth , 1990;
Hartmann et al. ,1993;
Hartmann , 1991;
Kenyon , 1995ab;
Bell et al. , 2000;
Hartmann and Kenyon , 1996; the specific chapter on the FUOri phenomenon in
Hartmann , 2008; and the recent reviewby
Reipurth and Aspin , 2010). The field has exploded in thelast fifteen years thanks to new ground and space facilities.Observationally FUor candidates — and their possibleshort timescale counterparts, EX Lupi objects, dubbed EX-ors by
Herbig , (1989) — have been studied across the elec-tromagnetic spectrum, while theoretical studies have fur-ther explored the origin of the outburst mechanism. Erupt-ing young stars are no longer oddities but are now placedprominently on the grand scheme of star formation andtime evolution of mass accretion rates, from embedded pro-tostars to classical T Tauri stars (CTTS), and eventuallyweak-line T Tauri stars (WTTS). In parallel, recent stud-ies have led to doubt as to the need for separate observa-tional classification of FUors and EXors, as discoveries ofnew outbursting sources have resulted in a less definitiveseparation. Episodic accretion has also possibly resolved,amongst other issues, the so-called luminosity problem inlow-mass embedded protostars.In this review, we provide a summary of the literaturepublished on the “historical” FUor and EXor classes and onthe theoretical and numerical studies relevant to episodicaccretion. We start from the review by
Hartmann andKenyon , (1996), although we refer to older studies when-ever needed. Finally, we emphasize that this review focuseson episodic accretion, i.e., strong variability due to accre-tion events: we do not address small-scale variability orvariability caused by geometrical effects, clearing of dust,etc., although some of these aspects will be mentioned whenobserved in parallel with episodic accretion.
2. OBSERVATIONS2.1. Episodic Accretion During Star Formation
The general picture of the evolution of pre-main se-quence accretion (
Hartmann and Kenyon , 1996) suggeststhat much of the material added to the central star, and thematerial available for planet-forming disks, is influenced bythe frequency and intensity of eruptive bursts followed bylong periods of relative quiescence. It is suspected that thisprocess occurs at all early stages of star formation after theprestellar core, but becomes observable only as the circum-stellar envelope thins. In one picture, FUors and EXors are part of this continuum. In this picture, the FUor bursts arelonger and stronger compared to the bursts of EXors. Thebursts would occur in repeated cycles and are fueled by ad-ditional material falling from the circumstellar envelope tothe disk in between bursts, halted by some mechanism, andreleased in a dramatic flood quasi-periodically. In an al-ternative picture, EXors would be a separate phenomenonassociated with instabilities in the disks of T Tauri stars(TTS), while FUors span the divide between protostars withdisks and envelopes and TTS with disks. Observations re-veal a more complicated picture in which strong, long out-bursts can also occur in previously identified CTTS, andEXor-type short outbursts in relatively embedded youngstars with envelopes.In the next sections, we have kept this historical, ob-servational separation between FUors and EXors with theaim to draw commonalities within the classes. We aim tobuild on the observational and theoretical results to addressthe validity of the separation and to propose future steps tohelp determine how and if FUors and EXors are related.
The initial class of FUors was comprised of FU Ori,V1057 Cyg, and V1515 Cyg, all showing strong outburstswith amplitudes of several magnitudes, albeit over signif-icantly different timescales and durations (
Herbig , 1977).V1735 Cyg was added to the list shortly thereafter (
Elias ,1978). Although FU Ori has been slowly fading since its1936 outburst (
Kenyon et al. , 2000), it is still in a highstate at present. Notice that V1331 Cyg has often been in-cluded among lists of FUors (following
Welin,
Graham and Frogel , 1985;
Brand et al. , 1986;
Eisl¨offel et al. , 1990;
Staude and Neckel , 1991, 1992;
Stromand Strom , 1993;
McMuldroch et al. , 1995;
Shevchenkoet al. , 1997;
Sandell and Aspin , 1998;
Aspin and Sandell ,2001;
Aspin and Reipurth , 2003;
Movsessian et al. , 2003,2006;
Quanz et al. , 2007ab;
Tapia et al. , 2006;
K´osp´al et al. ,2008;
Magakian et al. , 2010, 2013;
Reipurth et al. , 2012).Many objects that are spectroscopically similar to classicalFUors but have never been seen to erupt are instead clas-sified as FUor-like objects (in analogy with nova-like ob-jects, see
Reipurth et al.,
Kravtsova et al. , 2007). They are associatedwith reflection nebulae and are associated with a moder-2
ABLE ON - EXHAUSTIVE LIST OF ERUPTIVE YOUNG STARS
Name Type Distance Onset Duration A V L bol ˙ M acc Companion References(pc) (yr) (yr) (mag) ( L ⊙ ) ( M ⊙ yr − )RNO 1B FUor-like 850 · · · >
12 9.2 · · · · · ·
Y (RNO 1C, 4”) 44,72,78RNO 1C FUor-like 850 · · · · · · · · · · · ·
Y (RNO 1B, 4”) 44,72V1180 Cas EXor? 600 2000, 2004 2.5, 7 4.3 0.07 (L) > > < > · · ·
66 (L) · · ·
Y (0.3”) 5,12,22PP13S FUor-like 350 · · · · · · ∼
40 30 · · ·
N? 19,73XZ Tau EXor? 140 1998 > − − · · · Y (0.66”) 23,36,56LDN 1415 IRS EXor? 170 > < · · · · · · > · · · · · · · · · > < > · · · · · · · · · · · · ∼ − −
25 (H) 2.5e-7 (L), 1e-6 (H) Y (0.18”) 14,36,39,55,59,68Haro 5a IRS FUor-like 450 · · · · · ·
22 50 · · · · · · > · · · · · · N 14,36,39,45,59V1143 Ori EXor 500 many ∼ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · > < > −
12 2-4.5 (L), 22-28 (H) 2e-6 (L), 1e-5 (H) Y? (11”) 24FU Ori FUor 450 1936 ∼
100 1.5 − − · · · Y (0.5”) 1,34,77V1647 Ori EXor? 400 1966,2003,2008 0.4 − > −
19 3.5 − −
44 6e-7,4e-6 − · · · · · · >
13 18 450 · · ·
Y (AR 6B, 2.8”) 13AR 6B FUor-like 800 · · · · · · > · · · · · · Y (AR 6A, 2.8”) 13V900 Mon FUor-like 1100 > < >
16 13 106 (H) · · ·
N 70Z CMa FUor 930 − −
10 1.8 − −
600 1e-3 Y (0.1”) 34,35,46,53,72,85BBW 76 FUor-like 1700 < ∼
40 2.2 287 7.2e-5 N 1,27,34V723 Car EXor · · · · · · · · · · · · · · · · · · · · · > & > < ∼ > >
12 135 · · ·
N 1,26,27,34,67OO Ser FUor-like 311 1995 >
16 42 4.5 (L), 26-36 (H) · · ·
N 32,40,41,42,48Parsamian 21 FUor-like 400 · · · · · ·
8? 3.4, 10 · · ·
N? 20,47,79V1515 Cyg FUor 1000, 1050 ∼ ∼
30 2.8 − · · · ∼ − · · · > < > −
12, 10 −
20 14 (L), 43 (H) 2.5e-7 (H) · · · > − −
12 (H) 1e-6 (H) · · · − > < >
20 5.8 14 − · · · · · · ∼
10 3.0 − −
800 4.5e-5 N 1,27,34,77V2495 Cyg FUor 800 1999 > · · · · · · · · · · · · > · · · · · · · · · · · · > < >
20 8.0 − · · · Y? (20-24”) 1,34,77HH 354 IRS FUor-like 750 · · · · · · · · · · · · · · · > < >
38 8 135 (H) · · · · · ·
OTE .— L: quiescent or low state, H: outburst or high state, Companion: The parentheses show the angular separation of the companion from FUor/EXor. References: 1: ´Abrah´am et al., (2004b); 2: ´Abrah´am et al., (2009); 3: ´Abrah´am et al., (2004a);4:
Andrews et al., (2004); 5:
Anglada et al., (2004); 6:
Aspin et al., (2010); 7:
Aspin et al., (2009b); 8:
Aspin et al., (2006); 9:
Aspin, (2011b); 10:
Aspin et al., (2008); 11:
Aspin et al., (2009c); 12:
Aspin and Sandell, (1994); 13:
Aspin and Reipurth, (2003); 14:
Audard et al., (2010); 15:
Brice˜no et al., (2004); 16:
Casali, (1991); 17:
Chavarria, (1981); 18:
Coffey et al., (2004); 19:
Cohen et al., (1983); 20:
Connelley and Greene, (2010); 21:
Covey et al., (2011); 22:
Eisl¨offel et al., (1991); 23:
Elias, (1978); 24:
Fischer et al., (2012); 25:
Gibb, (2008); 26:
Graham and Frogel, (1985); 27:
Green et al., (2006); 28:
Green et al., (2013b); 29:
Green et al., (2011); 30:
Grosso et al., (2010); 31:
Haas et al., (1990); 32:
Haisch et al., (2004); 33:
Hartiganand Kenyon, (2003); 34:
Hartmann and Kenyon, (1996); 35:
Hartmann et al., (1989); 36:
Herbig, (1989); 37:
Herbig, (1977); 38:
Herbig, (2007); 39:
Herbig, (2008); 40:
Hodapp, (1996); 41:
Hodapp, (1999); 42:
Hodapp et al., (2012); 43:
Jensen et al. (2007); 44:
Kenyon et al., (1993); 45:
K¨ohler et al., (2006); 46:
Koresko et al., (1991); 47:
K´osp´al et al., (2008); 48:
K´osp´al et al., (2007); 49:
K´osp´al et al., (2011a); 50:
K´osp´al et al., (2013); 51:
Kun et al., (2011a); 52:
Kun et al., (2011b); 53:
Leinertand Haas, (1987); 54:
Lombardi et al., (2008); 55:
Lorenzetti et al., (2006); 56:
Lorenzetti et al., (2007); 57:
Magakian et al., (2013); 58:
McMuldroch et al., (1993); 59:
Menten et al., (2007); 60:
Miller et al., (2011); 61:
Movsessian et al., (2003); 62:
Movsessian et al., (2006); 63:
Muzerolle et al., (2005); 64:
Parsamian et al., (1987); 65:
Peneva et al., (2010); 66:
Persi et al., (2007); 67:
Prusti et al., (1993); 68:
Reipurth et al., (2007b); 69:
Reipurth and Aspin, (1997); 70:
Reipurth et al., (2012); 71:
Reipurth et al., (2007a); 72:
Sandell and Weintraub, (2001); 73:
Sandell and Aspin, (1998); 74:
Semkov et al., (2011); 75:
Shevchenko et al., (1991); 76:
Sipos et al., (2009); 77:
Skinner et al., (2009); 78:
Staude and Neckel, (1991); 79:
Staude and Neckel, (1992); 80:
Stecklum et al., (2007); 81:
Strom et al., (1993); 82:
Tapia et al., (2006); 83:
Teets et al., (2011); 84:
Venkata Raman et al., (2013); 85:
Whelan et al., (2010)3 te level of extinction (A V ∼ µ m. The Fe I , Li I , and Ca I optical lines, as well as the near-infrared CO lines, are typically double-peaked and showline broadening that is kinematically consistent with a ro-tating disk origin ( Hartmann and Kenyon , 1996), althougha wind may also be required to explain these profiles (
Eis-ner and Hillenbrand , 2011). A different origin for the op-tical line splitting was put forward by
Petrov and Herbig ,(1992) and
Petrov et al. , (1998) who argued that the profilecan be explained simply by the presence of central emissioncores in the broad absorption lines (see also
Herbig et al. ,2003). Similarly, a model invoking a large dark polar spotwas put forward to explain the line profiles in the optical forFU Ori (
Petrov and Herbig , 2008). Despite the controversyat optical wavelengths (which may in fact relate to differentaspects of the same underlying phenomenon, see
Kravtsovaet al. , 2007), the accretion disk model generally works well(e.g.,
Hartmann et al. , 2004).In the infrared, the similarity between classical FUorsbegins to waver, as FU Ori exhibits pristine silicate emis-sion features and a relatively blue SED consistent with thatof a mildly flared disk (
Green et al. , 2006;
Quanz et al. ,2007c), while V1515 Cyg and V1057 Cyg are closer toflat-spectrum sources with weaker silicate emission. Theyare further distinguished in the far-infrared/submillimeterwhere envelopes often dominate SEDs: FU Ori and V1515Cyg have weak continuum beyond 100 µ m and very lit-tle CO line emission, while V1057 Cyg shows warm COgas and substantially stronger continuum ( Green et al. ,2013b). Nevertheless, near-infrared interferometry showsthat classical FUors all show significant contributions fromenvelopes (
Millan-Gabet et al. , 2006). Such envelopes havemasses of a few tenths of a solar mass and are comparableto Class I protostars rather than Class II stars (
Sandell andWeintraub , 2001;
P´erez et al. , 2010).During their outbursts, classical FUors have bolometricluminosities of − L ⊙ , with accretion rates between10 − and 10 − M ⊙ yr − (Sect. 4.4). One of the challengesis identifying candidate FUors before they occur: V1057Cyg is one of only two identified FUors (with HBC 722,discussed below) with a pre-outburst optical spectrum ( Her-big , 1977). It was found to have optical emission lines typ-ical of a TTS. Modern large spectral surveys of active star-forming regions should increase the likelihood of serendip-itous pre-outburst observations (Sect. 5).Since few FUor outbursts have been detected, the ques-tion arises as to whether some existing YSOs may have pre-viously undergone undetected FUor eruptions. For instancesome driving sources of Herbig-Haro (HH) objects might beFUor-like, because their near-infrared spectra are substan-tially similar to those of FUors (
Reipurth and Aspin , 1997;
Greene et al. , 2008). These authors noted that young starswith FUor-like characteristics might be more common thanprojected from the relatively few known classical FUors.
In the optical, the initialbrightening of ∼ mag is followed by a longer plateauphase, during which classical FUors display a relativelylong decay timescale: FU Ori has faded by about 1 magsince its 1936 outburst, at a rate of 14 mmag yr − ( Kenyonet al. , 2000). BBW 76 (also known as Bran 76) showed adecay of about 23 mmag yr − in V between 1983 and 1994,with a slow-down in the infrared around 1989, although ahistorical search indicated that BBW 76 has remained at asimilar brightness level since at least 1900 ( Reipurth et al. ,2002). To document the long-term flux evolution of FUors, ´Abrah´am et al. , (2004b) compared 1–100 µ m photometricobservations, obtained by ISO in 1995-98, with earlier datataken from the
IRAS catalog (from 1983). Both satellitedata sets were supplemented by contemporaneous ground-based infrared observations from the literature. In two cases(Z CMa, Parsamian 21) no significant difference betweenthe two epochs was seen. V1057 Cyg, V1515 Cyg andV1735 Cyg became fainter at near-infrared wavelengthswhile V346 Nor had become slightly brighter. At λ ≥ µ m most of the sources remained constant; only V346 Norseemed to fade. A detailed case study of V1057 Cyg re-vealed that the long-term evolution of the system at near-and mid-infrared wavelengths was consistent with modelpredictions of Kenyon and Hartmann , (1991) and
Turner etal. , (1997), except at wavelengths longward of 25 µ m. In addition tobroad lightcurve evolution, FUors exhibit a variety of short-timescale (hours to days) variability, both semi-periodic andstochastic.
Kenyon et al. , (2000) present rapid cadence pho-tometry of classical FUors. They find evidence for low am-plitude “flickering” ( ∼ ≤ d which they attribute to the inner accretion disk. Sim-ilar results were found by Clarke et al. , (2005). From highspectral resolution optical observations of FU Ori,
Pow-ell et al. , (2012) confirmed the modulation of wind lines( P ∼
14 days) and photospheric lines ( P ∼ Herbig et al. , (2003), while
Siwak et al. , (2013)used
MOST photometry and found the possible existenceof 2–9 d quasi-periodic features in FU Ori, which theyinterpreted as plasma condensations or localized disk het-erogeneities. The flickering and presence of periodicitywas also found in other erupting stars, such as V1647 Ori(
Bastien et al. , 2011).
Green et al. , (2013a) found a ro-tational period for HBC 722 of 5.8 days, and a variety ofsub-periods between 0.9 and 1.3 days, with a peak in the pe-riodogram at 1.28 days. This shorter period was attributedto instability near the inner disk edge (see also Sect. 3.4.3).In any case, episodic accretion can be distinguished from“normal” TTS variability that results from either small-scale accretion events, geometrical effects, or magnetic ac-tivity (
Costigan et al. , 2012;
Chou et al. , 2013;
Scholz etal. , 2013). Indeed, accreting TTS tend to vary in mass ac-cretion rates by 0.37-0.83 dex or less, depending on the ac-cretion diagnostic. Such variability occurs over time scalesof ≤ months, with a major part of variability occurringover shorter time scales, 8–25 days, i.e., comparable to stel-4ar rotational periods. In contrast, episodic accretion is char-acterized by stronger variations in mass accretion rate andoccurs on longer time scales (i.e., months to years). (Sub-)millimeter observations originally suggested thatclassical FUors are extremely young, more similar to Class Iprotostars than to Class II stars ( Sandell and Weintraub ,2001). These single-dish continuum maps showed resolvedemission and typically elongated disk-like morphology,with sizes of several thousand AU (e.g.,
Dent et al. , 1998;
Henning et al. , 1998). The mass in these structures was esti-mated at a few tenths of M ⊙ . Such reservoirs are sufficientto replenish circumstellar disks after episodes of rapid ac-cretion. Molecular outflows were generally detected, exceptfor the most evolved FUors ( Evans et al. , 1994;
Moriarty-Schieven et al. , 2008), with mass loss rates between a fewtimes 10 − and 10 − M ⊙ yr − , without correcting for op-tical depth; mass loss rates likely peak at 10 − M ⊙ yr − .Dense gas traced by HCO + was also detected ( McMuldrochet al. , 1993, 1995;
Evans et al. , 1994). Newly obtained lineprofiles obtained by
Green et al. , (2013b) with
Herschel are consistent with the above results. However, with single-dish observations, it is often difficult to determine whetherthe progenitor of the outflow is the FUor or in fact nearbyyoung protostars, such as in the case of HBC 722, whichremained undetected in the continuum, with an upper limitof 0.02 M ⊙ for the mass of the disk ( Dunham et al. , 2012),low enough to rule out gravitational instabilities in the diskas the driving mechanism for the outburst.Further evidence of gas emission associated with FUorswas obtained with
ISO by Lorenzetti et al. , (2000) whofound [OI], [CII], and [NII], and some cold molecular lines,both on and off-source. All FUors observed with
Herschel by Green et al. , (2013b) exhibited strong [O I] 63 µ m emis-sion, indicative of high mass loss rates or UV excitation. Onthe other hand, the warm CO emission lines, generally ob-served in Class 0/I protostars, were not detected, with onlyV1057 Cyg showing compact high- J CO emission.Interferometric observations are, however, required todistinguish between the circumstellar envelope and cloudemission.
K´osp´al , (2011) detected strong, narrow COJ=1–0 line emission in Cygnus FUors. The emission wasspatially resolved in all cases, with deconvolved sizes of afew thousand AU. For V1057 Cyg, the emitting area wasrather compact, suggesting that the origin of the emissionis a circumstellar envelope surrounding the central star. ForV1735 Cyg, the CO emission was offset from the stellarposition, indicating that the source of this emission mightbe a small foreground cloud, also responsible for the highreddening of the central star. The CO emission towardsV1515 Cyg was the most extended in the sample, and it ap-parently coincided with the ring-like optical reflection neb-ula associated with V1515 Cyg.Future interferometric observations in the (sub-)millimeterwill better disentangle the envelope and disk emissions, and constrain the mass reservoir around FUors.
There are a number of methods used to classify youngstars, such as the infrared spectral index α ( Andr´e et al. ,1993;
Greene et al. , 1994) or the bolometric temperature T bol ( Chen et al. , 1995;
Robitaille et al. , 2006). SomeFUors exhibit Class II SEDs, while others exhibit flat SEDs(characterized as having 350 K < T bol <
950 K,
Evanset al. , 2009;
Fischer et al. , 2013), near the Class I/IIboundary (
Green et al. , 2013b). The envelope mass de-rived from (sub-)millimeter dust continuum can also beused to discriminate between evolutionary Stages. A thor-ough discussion on the distinction between Stages and theobservationally-defined infrared Class is provided in thechapter on protostellar evolution by
Dunham et al.
Several studies have focused on the silicate featurearound 10 µ m and on ice properties ( Polomski et al.,
Sch¨utz et al. , 2005;
Green et al. , 2006;
Quanz et al. , 2007c).The silicate feature can be found either in absorption or inemission. The FUors with silicate absorption generallyshow an amorphous silicate structure, similar to the inter-stellar medium, although some extra emission or large sil-icates can be found. In addition, H O, methanol, and CO ice absorption is detected. In contrast, FUors with silicatein emission can show evidence for absorption in the diskphotosphere from blended H O vapor absorption (5-8 µ m; Green et al. , 2006). The silicate feature in emission doesnot show evidence of crystals, with some FUors showingweak, pristine silicate features (e.g., V1515 Cyg and V1057Cyg). This led
Quanz et al. , (2007c) to propose that FUorscan be classified in two categories, defined by silicate ab-sorption vs. emission. The emission features arise from thedisk surface layers, and represent evidence for grain growthbut no processing (Fig. 1). Objects with silicate in absorp-tion are likely young FUors, still embedded in an envelope,whereas objects with silicates in emission are likely moreevolved FUors with (partially) depleted envelopes.
Green et al. , (2013b) found that the mid-infrared se-quence proposed by
Quanz et al. , (2007c) broadly describedthe continuum of FUors for which there was sufficient(sub-)millimeter data. However, the line observations pro-vide a different picture from the continuum, as FUors maynot be well characterized by the Stage I/II sequence. WhileFU Ori and V1515 Cyg, two of the most evolved FUors,were found to have little warm ( ∼
100 K) gas, V1057 Cygexhibited CO rotational lines typical of Class 0/I protostars.If this CO originates in the shocked layer of the envelopesurrounding the outflow, FU Ori’s lack of emission can beexplained by a tenuous envelope; however the mid-infraredsimilarity between V1515 Cyg and V1057 Cyg (in contrastwith FU Ori) would not predict the difference in their sub-millimeter emission lines. The foregoing makes apparentthe difficulty of classifying FUors in the Stage I/II sequence.
Sketch of the two categories of FUors (
Quanz et al. , 2007c). The first category is related to embedded sources with silicatefeatures in absorption and younger ages than FUors of the second category, which show silicate features in emission. Reproduced bypermission of the AAS.
Classical EXors (
Herbig , 1989, 2008) were initially de-fined as a list of variable stars showing large-scale outburstsand TTS spectra. The prototype, EX Lup, largely influ-enced the definition of the class, with its repetitive, short-lived outbursts, M-type dwarf spectrum in quiescence, andabsence of features typical of FUors (see Sect. 2.2). Theoriginal list of EXors (
Herbig , 1989) has changed littlesince that time. New potential EXors were found, althoughtheir classifications as EXor or FUor are sometimes unclear(e.g.,
Eisl¨offel et al. , 1991;
Aspin and Sandell , 1994;
Steck-lum et al. , 2007;
Persi et al. , 2007;
Kun et al. , 2011a).Abundant photometric observations during and after out-bursts indicated brightness increases of several magnitudesover durations of several months (e.g.,
Coffey et al. , 2004;
Audard et al. , 2010). Bolometric luminosities are about 1–2 L ⊙ in quiescence (including about 0.3–0.5 L ⊙ of stellarphotospheric luminosity) and peak around several L ⊙ to afew tens of L ⊙ ( Lorenzetti et al. , 2006;
Audard et al. , 2010;
Sipos et al. , 2009;
Aspin et al. , 2010). Coverage tends to be-come less frequent after outburst, preventing precise deter-minations of EXor outburst durations. Nevertheless, erup-tions occur frequently in the same objects, with separationsof a few years between outbursts, and durations of about 1–2 years (e.g.,
Herbig , 2008). Oscillations ( ≈ Acosta-Pulidoet al. , 2007;
D’Angelo and Spruit , 2012).EXors show absorption spectra typical of K or M dwarfswith T Tauri-like emission spectral features during theirminimum light (
Parsamian and Mujica , 2004;
Herbig ,2008). Optical and near-infrared SEDs during outburstscan be well fitted with an extra thermal component, e.g.,a single blackbody component with temperatures varyingfrom 1000 K to 4500 K and emitting radii of 0.01 to 0.1AU, typical of inner disk regions (
Lorenzetti et al.,
Audard et al. , 2010;
Lorenzetti et al.,
Lorenzetti et al., (2012) argue that it does not exceed about 10 % of the stel-lar surface. In any case, color variations observed duringoutbursts cannot be explained by extinction effects alone.Signatures of infall and outflow are detected in Na I D , ,in a similar fashion as in CTTS. Near-infrared spectra showline emission from hydrogen recombination lines, with COovertone features and weaker atomic features commonlydetected both in emission and absorption and variable onshort timescales ( Lorenzetti et al. , 2009).The SEDs of EXors span the Class I/II divide symmetri-cally, with a peak in the “flat spectrum” category (
Gianniniet al. , 2009). In the mid-infrared, EXors show the 10 µ msilicate feature in emission ( Audard et al. , 2010;
K´osp´al etal. , 2012). Some of them show wavelength-independentflux changes, probably due to varying accretion. Othersare more variable in the 10 µ m silicate feature than in theadjacent continuum, which can be interpreted as possiblestructural changes in the inner disk. Several newly discovered eruptive young stars have beenfound, many in the last decade. Some of them often showspectral characteristics typical of FUors, but smaller lumi-nosities (see Tab. 1). HBC 722 (also known as V2493 Cyg)started with a . − . L ⊙ luminosity and a SED typi-cal of CTTS, then its luminosity rose to − L ⊙ in out-burst ( Semkov et al. , 2010;
Miller et al. , 2011;
K´osp´al etal. , 2011a). Similarly, V2775 Ori changed its bolometricluminosity from . L ⊙ to ≈ L ⊙ ( Caratti o Garatti etal. , 2011;
Fischer et al. , 2012). LDN 1415 IRS could alsobe a low-luminosity erupting star ( . L ⊙ in low state; Stecklum et al. , 2007), although its status as FUor or EXoris yet unclear. Moderate luminosities are also observed inOO Ser (
Hodapp et al. , 1996, 2012), with − L ⊙ inoutburst, whereas V733 Cep displayed stronger luminosity( L ⊙ ) and has slowly faded since 2007 ( Peneva et al. ,6
200 400 600 800 1000 1200 1400JD (2,455,000 + )25201510 M a g n i t u d e V1647 OriHBC 722 V1057 CygEX Lup V2492 Cyg
Fig. 2.—
Comparison light curves for the FUor V1057 Cyg, theintermediate case V1647 Ori, the new sources HBC 722 and VSXJ2025126.1+440523 (also known as PTF 10nvg or V2492 Cyg),and the classical EXor EX Lup, showing the continuum of outburstdurations. Adapted from
K´osp´al et al., (2011a). − L ⊙ ) in light of the uncer-tainty in its distance ( Magakian et al. , 2013). The recentlydiscovered V1647 Ori also showed moderate luminosity atpeak (about − L ⊙ ; Andrews et al. , 2004;
Muzerolle etal. , 2005;
Aspin , 2011b), although it may more closely re-semble an EXor than a FUor. Some new sources are young(typically Class I) and deeply embedded, hidden behindthick envelopes (e.g., OO Ser:
K´osp´al et al. , 2007; V900Mon:
Reipurth at al. , 2012). Others have no detectable en-velopes, suggesting a relatively evolved state (e.g., V733Cep:
Reipurth et al. , 2007a; HBC 722:
Green et al. , 2011,2013c;
Dunham et al. , 2012), despite sometimes displayingheavy extinction due to surrounding clouds.Even more interestingly, time scales for outbursts werefound in between those of classical EXors and FUors(Fig. 2). OO Ser was in outburst for about 8 yrs (
Hodappet al. , 1996, 2012). Together with its moderate luminosity,it bears resemblance to V1647 Ori, which has been twicein outburst since 2004 (see Sect. 2.9.3). HBC 722 also dis-played quite peculiar behavior: it started fading after peakbrightness with a rate much faster than typical for FUors,but the fading stopped, and the object started brighteningagain, with no clear signs that its outburst will end anytime soon (
Semkov et al. , 2012; Fig. 2). These discoveriesdemonstrate that eruptive phenomena span a considerablerange in evolutionary state, envelope mass, stellar mass,time behavior, and accretion rates.
Several unusual eruptive stars were recently identified:sources where the brightening may be due to the combina-tion of two effects (
Hillenbrand et al. , 2013). One effectis an intrinsic brightening related to enhanced accretion, asin all eruptive stars (e.g.,
Sicilia-Aguilar et al. , 2008). Theother effect is a dust-clearing event that reduces the extinc-tion along the line of sight, such as in UX Ori-type stars(e.g.,
Xiao et al. , 2010;
Chen et al. , 2012;
Semkov andPeneva , 2012). Tab. 1 does not include sources for whichextinction effects dominate the time variability (e.g., GMCep, V1184 Tau). Accretion and extinction changes of-ten happen synchronously, suggesting that they are phys-ically linked. The extinction can reach high values (e.g.,
Covey et al. , 2011;
K´osp´al et al. , 2011a). The variations arelikely due to obscuring structures lying close to the stars(i.e., within a few tenths of an AU,
K´osp´al et al. , 2013).Interestingly, the objects in question can also be highly em-bedded Class I objects, such as V2492 Cyg, and they candisplay rich and variable emission-line spectra like EXors(
Aspin , 2011a). The intermediate-mass star PV Cep, origi-nally classified as an EXor by
Herbig , (1989) on the basisof its large outburst in 1977–1979 (
Cohen et al. , 1981), de-spite its mass, embeddedness and higher accretion rate thantypical EXors, also shows variability resulting from an in-terplay between variable accretion and circumstellar extinc-tion, hinting at a rapidly changing dust distribution close tothe star (
Kun et al. , 2011b;
Lorenzetti et al. , 2011; see also
Caratti o Garatti et al. , 2013).These new observations indicate that structural changesoften happen in the innermost part of the disk or circumstel-lar matter, typically within 1 AU, supporting the conclusionthat these changes are related to the outburst. It is remark-able that all of the objects described above show EXor-like,repetitive outbursts. This implies that whatever structuralchange happens in the disk, it should be reversible on arelatively short timescale, setting strong constraints on thephysical processes involved.
Recent outbursts in classical EXors and in new eruptiveyoung stars triggered strong interest in obtaining detailedfollow-up observations at all possible wavelengths. We fo-cus on three well-studied objects — the classical EXors EXLup and V1118 Ori, and the recently discovered V1647 Ori— that reflect the diversity of outburst properties.
EX Lup is a young low-mass ( . M ⊙ ) M0V star ( Gras-Vel´azquez and Ray , 2005),with a quiescent L bol = 0 . L ⊙ . Its disk has a mass of0.025 M ⊙ , with an inner hole within 0.2–0.3 AU, and anouter radius of 150 AU ( Sipos et al. , 2009). There is no hintof any envelope. This prototype of the EXor class showssmall amplitude variations in quiescence, but has displayedseveral outbursts (
Lehmann et al. , 1995;
Herbig et al. ,2001;
Herbig , 2007) during which the photospheric spec-trum is veiled by a hot continuum, the equivalent widthsof the optical emission lines decrease, inverse P Cygni ab-7ig. 3.—
Silicate emission spectra for the interstellar medium (a), EX Lup in quiescence (b), during its 2008 outburst (c), and fortwo comets (d). The bottom curves in panels (c) and (d) show the emissivity curve of pure forsterite, for grain temperatures of 1250 Kand 300 K, respectively. The outburst spectrum of EX Lup shows evidence of forsterite, not observed during quiescence, and producedthrough thermal annealing in the surface layer of the inner disk by heat from the outburst. Adapted from ´Abrah´am et al., (2009). sorption components appear at the higher Balmer lines,the emission-line structures undergo striking variations, andmany emission lines exhibit both a narrow and a broad lineprofile component. All these signatures indicate mass infallin a magnetospheric accretion event. In quiescence, permit-ted emission lines and numerous metallic lines are detected(
K´osp´al et al. , 2011b;
Sicilia-Aguilar et al. , 2012), likelyoriginating from a hot (6500 K), dense, non-axisymmetric,and non-uniform accretion column that displays velocityvariations along the line of sight on timescales of days. Fur-ther evidence supporting a magnetospheric accretion modelis given by the spectro-astrometry of near-infrared hydro-gen lines (
K´osp´al et al. , 2011b), which suggest a funnelflow or disk wind origin rather than an equatorial boundarylayer.A strong outburst ( ∆ V ∼ Sicilia-Aguilar et al. , 2012).During the outburst, the SED indicates a hot single-temperature blackbody component which emitted 80%–100% of the total accretion luminosity (
Juh´asz et al. , 2012).Strong correlation is also found between the decreasing op-tical and X-ray fluxes, while UV and soft X-rays are as-sociated with accretion shocks (see Sect. 4.5). From COfundamental band emission lines,
Goto et al. , (2011) iden-tified a broad-line component that was highly excited, anddecreased as the source faded. This gas component likelyorbited the star at 0.04–0.4 AU, implying that it is the inner0.4 AU of the disk that became involved in the outburst, aregion coinciding with the optically thin, but gas-rich, innerhole. Furthermore,
Juh´asz et al. , (2012) concluded, mainly on the basis of accretion timescales, that thermal instabilitywas likely not the triggering mechanism of the 2008 out-burst. It remains an open question whether the inner diskhole, whose radius is larger than the dust sublimation ra-dius, is related to the eruptive phenomenon.
Spitzer spectra obtained at peak light displayed featuresof crystalline forsterite, while such crystals were not de-tected in quiescence (Fig. 3; ´Abrah´am et al. , 2009). Thecrystals were likely produced through thermal annealing inthe surface layer of the inner disk by heat from the outburst,a scenario that could produce raw material for primitivecomets.
Juh´asz et al. , (2012) presented additional multi-epoch
Spitzer spectra, and showed that the strength of thecrystalline bands increased after the end of the outburst, butsix months later the crystallinity decreased. Although verti-cal mixing in the disk would be a potential explanation, fastradial transport of crystals (e.g., by stellar/disk winds) waspreferred. The outburst also influenced the gas chemistry ofthe disk:
Banzatti et al. , (2012ab) found that the H O andOH line fluxes increased, new OH, H , and H I transitionswere detected, and organics were no longer seen.In summary, the EX Lup outburst indicated that itsenhanced accretion probably proceeded through magneto-spheric accretion channels which were present also in qui-escence but delivered less material onto the star. The classicalEXor V1118 Ori has displayed several outbursts (
Herbig etal. , 2008 and references therein).
Parsamian et al. , (2002)showed that optical spectra taken during the 1989 and 1992-1994 outburst were similar to those of TTS with strong Hand Ca II lines. The star is a close binary ( . ′′ ) with similarfluxes in H α ( Reipurth et al. , 2007b). V1118 Ori was fol-lowed during its 2005-2006 outburst in the optical, infrared,and X-rays (Fig. 4;
Audard et al. , 2005, 2010;
Lorenzetti etal. , 2006, 2007) until it returned to quiescence in 2008. TheX-ray results are described in Sect. 2.10. The optical andnear-infrared emission at the peak of the outburst was dom-8 −10 −5 l og F l u x ( a r b i t r a r y un i t s ) U IRVBJHKX−rayI1I2I3I4I1I2I3I4I1I2I3I4I1I2I3I4I1I2I3I4I1I2I3I4 F l u x den s i t y ( e r g / c m / s e c ) Julian Date - 2,450,000The Erupting Young Star V1647 Ori2000.0 2002.0 2004.0 2006.0 2008.0 2010.0 2012.0I-bandK-bandX-ray
Fig. 4.—
Optical, infrared, and X-ray light curves of V1118 Ori (left; from
Audard et al. , 2010; reproduced with permission fromAstronomy & Astrophysics, c (cid:13)
ESO) and V1647 Ori (right; M. Richmond, priv. comm.; adapted from
Teets et al. , 2011). The lightcurves show that X-rays follow the accretion event, albeit with a weak flux increase in V1118 Ori but a strong increase in V1647 Ori. inated by a hotspot (
Audard et al. , 2010; see also
Loren-zetti et al. , 2012), whereas disk emission dominated in themid-infrared. Star+disk+hotspot models suggested that themass accretion rate increased from quiescence to peak ofoutburst from . × − to . × − M ⊙ yr − ( Audard etal. , 2010), together with a significant increase in fractionalarea of the hotspot.
Lorenzetti et al. , (2012) used a differentapproach and fitted the difference of the outburst and qui-escent SEDs with a black body, obtaining a temperature of ∼ K with a radius for the emitting region (assumedto be a uniform disk) of 0.01 AU. I -band polarimetry in-dicated that V1118 Ori is polarized at the level of 2.5%,with higher and more variable values observed in quies-cence, suggesting that the spotted and magnetized photo-sphere can be seen once the envelope partially disappears( Lorenzetti et al. , 2007). Color-color diagrams showed vari-ations but no signature of reddening caused by circumstel-lar matter (
Audard et al. , 2010), consistent with the lowextinction ( A V ≤ ) found during all activity phases by Lorenzetti et al. , (2006).
Spitzer spectra showed a slight in-crease in flux of the crystalline feature at the peak of the out-burst compared to post-outburst, but no variation in shapethat would indicate a change in grain population. Fromthe CO bandhead emission observed during the decliningphase,
Lorenzetti et al. , (2006) derived a mass loss rate of (3 − × − M ⊙ yr − . A spectrum taken after the out-burst detected no emission lines and 2.3 µ m CO in absorp-tion . Herbig , (2008) also reported a switch from emissionduring outburst to absorption in quiescence for Li I and K I lines, and wind emission in H α in the decay phase but asymmetric profile after the outburst. Strong wind losseshave, thus, likely occurred only transiently. The out- burst of V1647 Ori was discovered in January 2004, illu-minating a surrounding nebula. Archival data showed thatthe source was previously in outburst for 5-20 months in1966–1967 (
Aspin et al. , 2006). The start of the st cen-tury outburst occurred in late 2003, leading to an increaseof ∼ I -band in 4 months ( Brice˜no et al. , 2004).Pre-outburst data indicated a flat-spectrum source with anestimated pre-outburst bolometric luminosity of ≈ . − . L ⊙ and age of 0.4 Myr, typical of CTTS ( ´Abrah´am et al. ,2004a, see also Andrews et al.,
Andrewset al. , 2004;
Muzerolle et al. , 2005;
Aspin , 2011b). The pe-culiarity of V1647 Ori is that its SED differs from the clas-sical EXors, and more closely resembles those of FUors,with strong extinction ( A V ∼ Ojha et al. , 2005). Thecircumstellar mass, however, is typical of TTS (
Andrews etal. , 2004; ´Abrah´am et al. , 2004a;
Tsukagoshi et al. , 2005).During the outburst, the near-infrared color indices ofV1647 Ori shifted to bluer colors along the reddening vec-tor, mainly due to intrinsic brightening and partly to de-creasing extinction (
Reipurth and Aspin , 2004a;
McGeheeet al. , 2004;
Acosta-Pulido et al. , 2007). The mass accre-tion rate during the outburst was estimated at a few − to − M ⊙ yr − ( Muzerolle et al. , 2005;
Aspin , 2011b;
Mosoni et al. , 2013). CO bandhead emission was detectedalong with strong H and He emission lines with P Cyg pro-files, indicating a strong wind with velocities up to 600 kms − , together with ice absorption from H O and CO and inNa I D (
Reipurth and Aspin , 2004a;
Vacca et al. , 2004;
Wal-ter et al. , 2004).
Rettig et al. , (2005) showed that the COemission was hot (2,500 K) and attributed the emission tothe accretion zone of the inner disk. They also detected nar-row CO absorption lines superimposed on the low- J emis-9ion lines that originated in a foreground cold (18 K) cloud.The ices are unprocessed with a temperature of ∼
20 K, alsoindicating a cloud origin. In later spectra taken in 2004–2005, evidence for a slow decline in brightness was mea-sured, including a decline of the hot, fast wind (measuredfrom the He I µ m absorption strength), and of thetemperature of the inner disk ( Gibb et al. , 2006;
Ojha etal. , 2006). From radio data taken in early 2005,
Vig et al. ,(2006) proposed a homogeneous H II region to explain theradio, H α , and X-ray measurements. Further X-ray resultsare described in Sect. 2.10.In late 2005, V1647 Ori suddenly returned to quiescenceover a period of a few months ( K´osp´al et al. , 2005;
Cho-chol et al. , 2006;
Acosta-Pulido et al. , 2007;
Aspin et al. ,2008). The He I µ m line was then observed in emis-sion , in contrast to the outburst phase ( Acosta-Pulido et al. ,2007). Blueshifted CO absorption lines (30 km s − , 700 K)appeared in 2006 Feb superimposed on the previously ob-served CO emission lines, but were not reobserved in thespectra obtained in 2006 Dec and 2007 Feb ( Brittain et al. ,2007;
Aspin et al. , 2009b). This transient post-outburst out-flow was possibly launched by the outward motion of themagnetospheric radius during the rapid decrease of the ac-cretion rate. In 2006 Jan,
Fedele et al. , (2007) also detectedforbidden emission lines indicative of shocked gas likelyproduced by a HH-like object driven by V1647 Ori.Mid-infrared interferometric data obtained during out-burst revealed that the emitting region was extended (7 AUat 10 µ m), no close companion could be seen, and the8–13 µ m spectrum exhibited no strong spectral features( ´Abrah´am et al. , 2006). Using a disk+envelope geometryand varying the mass accretion rate from (1.6– × − M ⊙ yr − , Mosoni et al. , (2013) reproduced the SEDs takenat different epochs during the outburst. The models sug-gested an increase in the inner radius of the disk/envelopeat the beginning of the eruption. However, the system wasmore resolved at the later epoch when the outbursting re-gion was already shrinking, indicating either that the enve-lope inner radius suddenly increased at late stages of theoutburst, or that the fading of the central source was notimmediately followed by the fading of the outer regions.V1647 Ori returned to the spotlight when a second out-burst was reported in 2008 Aug–Sep (
Aspin et al. , 2009c).The source photometry, accretion rate, luminosity, and mor-phology were similar to those seen after the onset of the pre-vious outburst. However, CO overtone emission was not de-tected, despite being seen shortly after the onset of the 2003outburst.
Aspin et al. , (2009c) concluded that the quiescentperiod between these two outbursts was due to a partial de-cline in the accretion rate (factor of 2–3) and reformation ofdust in the immediate circumstellar environment. They ar-gued that the 2008 outburst was caused by the combinationof enhanced accretion and sublimation of dust due to thisenergetic event. In the high state of the second outburst,the H α and Ca II lines did not change remarkably ( Aspin ,2011b). The CO overtone bandhead was still not detected,while the water vapor absorption remained strong.
Brittain et al. , (2010) further found that the temperature of the COemission had varied with the accretion rate, and showeda direct relation between the Br γ luminosity and line fullwidth at half maximum and the source brightness, indicat-ing that the accretion rate had driven the variability.In view of the duration of the outburst, its recurrence,and the various spectroscopic features observed before, dur-ing, and after the outburst, V1647 Ori does not fit well ei-ther with classical EXors or FUors; it may instead be anintermediate case. In fact, Aspin et al. , (2009b) presented aspectrum of V1647 Ori, taken in quiescence, that does notresemble those of late-type standards, Class I protostars, orEXors observed in quiescence, but does show considerablecorrespondence with several classical FUors observed dur-ing their outbursts. Given that V1647 Ori shares some char-acteristics of both FUors and EXors, they proposed the ex-istence of a “continuum” of eruption characteristics ratherthan two distinct classes.
Strong X-ray emission is characteristic of young stars(
Feigelson and Montmerle , 1999), but our knowledge of thehigh-energy behavior of eruptive young stars is limited tojust a few examples observed over the past decade or so. X-rays probe high-energy processes, such as coronal activityand accretion shocks, and they are an important source ofionization and heating of accretion disk atmospheres, andthus they may influence disk chemistry and strengthen thecoupling of disk gas to the stellar magnetic field.
The first X-rayspectrum of FU Ori obtained with
XMM-Newton appearedquite unusual. The emission consisted of a hot plasmaviewed through very high absorption — much higher thananticipated based on A V — and a cooler component seenthrough ten times lower absorption ( Skinner et al. , 2006).Subarcsecond X-ray imaging by
Chandra subsequentlyprovided an explanation for the sharply contrasting absorb-ing columns (
Skinner et al. , 2010): the high-temperaturecomponent was positionally coincident with FU Ori, whilethe centroid of the cooler component was offset from FUOri and displaced toward the infrared companion FU OriS. The “excess” absorption toward FU Ori is likely a resultof accreting gas, FU Ori’s powerful wind, or both. Time-variability in the hot component — implying a magneticorigin — was detected, but no such variability was seen inthe fainter, less-absorbed, cooler component. The classicalFUor V1735 Cyg was bright ( log L X ≈ . ergs s − )and displayed very hot ( T > MK), heavily-absorbedplasma but no cool plasma, possibly because the latter ismore susceptible to absorption by intervening cool gas. Incontrast, V1057 Cyg and V1515 Cyg both escaped detec-tion (
Skinner et al. , 2009). Interestingly a faint soft X-rayjet was detected with
Chandra in Z CMa after the outburst(
Stelzer et al. , 2009), with a position angle consistent withthat of the micro-jet launched by the FUor component ofthis close binary (
Whelan et al. , 2010). The jet was not10etected in a pre-outburst observation, suggesting that massejection occurred. Clearly more observations are neededto characterize the X-ray properties of the FUor class as awhole.
The eruptions ofV1118 Ori in 2005-2006 and EX Lup in 2008 were bothmonitored in X-rays.
Teets et al. , (2012) followed the EXLup outburst beginning two months after the outburst onsetuntil just after its conclusion, while
Audard et al. , (2005,2010) and
Lorenzetti et al. , (2006) followed V1118 Ori untilit returned in quiescence in 2008 (Fig. 4). In both outbursts,there were strong correlations between the decreasing opti-cal and X-ray fluxes following the peak of the optical out-burst, suggesting a relation with the accretion rate. How-ever — in contrast to V1647 Ori (see below) — the X-rayflux increased only mildly in both cases, with a decrease influx after outburst, and relatively cool plasma temperatures( − MK) were observed. The temperature in V1118 Oricontrasted with a serendipitous pre-outburst observation in2002 that showed a dominant 25 MK plasma (
Audard et al. ,2005). The plasma temperature then gradually returned tohigher values in later phases of the outburst (
Audard et al. ,2010), in similar fashion to EX Lup (
Teets et al. , 2012) —although the cool plasma was still detected before the endof the 2008 EX Lup outburst (
Grosso et al. , 2010).One possible explanation for the anticorrelation betweenoptical/infrared flux and X-ray temperature observed forboth V1118 Ori and EX Lup is that their soft X-ray outputswere enhanced during eruption, due to emission arising inaccretion shocks (along with possible changes in magneto-spheric configurations). In the case of EX Lup, this hypoth-esis is consistent with the strong variability in ultravioletemission detected by
XMM-Newton , which was likely dueto accretion spots; in addition, the X-ray emission at ener-gies above 1.5 keV showed far stronger photoelectric ab-sorption than the cool plasma, suggesting the coronal emis-sion was subject to absorption by overlying, high-densitygas in accretion streams (
Grosso et al. , 2010). , (2004,2006) followed the entire 2003–2005 duration of V1647Ori’s outburst with
Chandra , including a serendipitous pre-outburst observation. The X-ray flux from V1647 Oriclosely tracked the rise and fall of its optical-infrared flux(Fig. 4), thereby providing definitive evidence for the pro-duction of high-energy emission via star-disk interactions.The monitoring data also indicated hardening of the X-rayemission during outburst. Further monitoring of V1647 Oriin 2008–2009 — i.e., soon after the onset of its second out-burst — established the striking similarity between the twooutbursts in terms of X-ray/near-infrared flux ratio and X-ray spectral hardness (
Teets et al. , 2011).Deep
XMM-Newton and
Suzaku exposures were ob-tained during V1647 Ori’s first and second outbursts, re-spectively (
Grosso et al. , 2005;
Hamaguchi et al. , 2010).The observations showed warm (10 MK) and hot (50 MK)plasmas, the former indicative of high-density plasma as-sociated with small magnetic loops around coronally ac- Fig. 5.—
Observed SED of FU Ori (solid line) fitted with a steadyaccretion disk model with contributions from the inner hot disk(light dotted curve) and the flared outer disk (light dashed curve).Adapted from
Zhu et al. , (2008). tive stars (
Preibisch et al. , 2005), the latter consistent withmagnetic reconnection events. Strong fluorescent Fe K α emission was detected, providing evidence for irradiationof neutral material residing either in the circumstellar diskor at the stellar surface. The comparison of A V ( ≈ mag)with the hydrogen column density ( N H ∼ × cm − )points out a significant excess of X-ray absorption fromrelatively dust-free material.Via time-series analysis, Hamaguchi et al. , (2012) es-tablished that the puzzling short-term (hours-timescale) X-ray variability of V1647 Ori was due to rotational modula-tion. The ∼ Hamaguchi et al. , (2012) fur-ther demonstrated that a model consisting of two pancake-shaped hot spots of high plasma density ( ≥ × cm − ),located at or near the stellar surface, reproduced well theX-ray rotational modulation signature. Under the assump-tion that the hot spots are generated via magnetic reconnec-tion activity, the stability of the X-ray light curve during thecourse of two accretion-related outbursts further suggeststhat the star-disk magnetic configuration of V1647 Ori hasremained relatively unchanged over timescales of years.
3. THEORY3.1. SED Fits and Accretion Disk Origin
SED fitting based on theoretical models is a powerfultoo in understanding the origin of these outbursting ob-jects. The most successful model to explain the peculiaritiesof FUors was proposed by
Hartmann and Kenyon , (1985,11987ab) and
Kenyon et al. , (1988), who suggested that theSED was dominated by a protostellar disk accreting at ahigh rate ( ˙ M ≈ − − − M ⊙ yr − ). Within this pic-ture, it is easy to account for the main peculiarities observedin FUor spectra, such as the changing supergiant spectraltypes from optical to infrared, and the double-peaked lineprofiles, with peak separation decreasing with increasingwavelength ( Hartmann and Kenyon , 1996).Self-consistent disk atmospheric models including theaccretion process and the radiative transfer in the disk at-mosphere are needed to constrain the SED and, in partic-ular, disk properties (e.g.,
Calvet et al. , 1991ab;
Pophamet al. , 1996).
Zhu et al. , (2007, 2008, 2009c) extended thefirst models with a more complete opacity database (
Ku-rucz et al. , 1974). This simple steady accretion disk modelcould fit FU Ori’s SED (Fig. 5;
Zhu et al. , 2007, 2008) andconfirmed the Keplerian rotation of FU Ori’s disk (
Zhu etal. , 2009a). The derived size of the high accretion rate in-ner disk was from 5 R ⊙ to ∼ Eisner and Hillenbrand ,(2011). Incidentally, such a size leads to a viscosity param-eter α ∼ . − . in outburst ( Zhu et al. , 2007).Different from this traditional steady viscous disk ap-proach,
Lodato and Bertin , (2001) argued that FUor disksmight be gravitationally unstable. The radial structure ofsuch a gravitationally unstable disk should be significantlydifferent from a non-self-gravitating one. They contendedthat the observed flat SED in the infrared might be relatedto a combination of extra heating induced by non-local dis-sipation of large-scale spiral structures and by the possibledeparture from a purely Keplerian rotation, if the disk massis a sizeable fraction of the central object (see also
Adamset al. , 1988;
Bertin and Lodato , 1999).
Lodato and Bertin ,(2003) then investigated how such effects would modify theshape of the double-peaked line profiles.
A closely related question for the disk accretion modelpertains to the origin of the outburst. Two main schoolsof thought have been proposed to explain the triggering ofFUor outbursts: 1) disk instability and 2) perturbation of thedisk by an external body.Thermal instability is due to the thermal runaway pro-cess when the disk temperature reaches the hydrogen ion-ization temperature. In detail, a hydrostatic viscous diskcan be thermally stable only if the opacity changes slowlywith the temperature. When the disk temperature reachesthe hydrogen ionization temperature, the opacity increasesdramatically with the temperature ( κ ∼ T ). A slight tem-perature increase will cause a significant amount of energytrapping in the disk and the disk temperature continues torise, which leads to the thermal runaway. The latter endswhen the disk is fully ionized at ∼ K and the opacitychanges slowly with temperature again. After the thermalrunaway, a high disk temperature can lead to a high disk accretion rate since the viscosity ν is proportional to thedisk temperature ( ν = αc s / Ω ∝ αT / Ω , with c s the soundspeed, viscosity parameter α , and Ω the angular velocity).The thermal instability model can naturally explain thefast rise time of FU Ori. However, since thermal instabil-ity needs to be triggered at T ∼ ∼ R ⊙ . In order to maintain theoutburst for hundreds of years, the α parameter needs tobe as low as 10 − during the outburst. Furthermore, in or-der to produce enough accretion rate contrast before andduring the outburst, α needs to be 10 − in the quiescentstate ( Bell and Lin , 1994;
Bell et al. , 1995). 2-D radiationhydrodynamic simulations have been carried out to studythe triggering of thermal instability in disks and the thermalstructure of the boundary layer (
Kley and Lin , 1996, 1999;
Okuda et al. , 1997). The SED fitting for FU Ori does not re-quire an additional hot boundary layer (Fig. 5), which maysuggest that the boundary layer can be different betweenFUors and TTS (
Popham , 1996).Several models have been put forward to cope with thelimitations of the standard thermal instability model. Theseinclude variations on the thermal instability model itself,or an altogether new trigger mechanism, for example as-sociated with the onset of the magneto-rotational or gravi-tational instability. They are discussed in detail below. Acomparison of the light curves produced by these differentmodels can be found in Figs. 6 and 7, which show theirlong-term evolution (over a period of thousands of years)and the detailed evolution of a single outburst, respectively.
In ther-mal instability models, it has been noted that, in the absenceof a trigger, the instability would first occur in the inner-most parts of the disk and then propagate inside-out, in a“snowplough” fashion (
Lin et al. , 1985), decelerating as itpropagates out and thus leading to a slow rise in the lightcurve (as observed in the case of V1515 Cyg). A fast rise(as observed in FU Ori and V1057 Cyg) is produced only ifthe outburst is “triggered” somehow far from the disk inneredge, so that the instability can propagate outside-in, in an“avalanche” fashion.
Clarke et al. , (1990) have confirmedthis behavior by including an ad hoc density perturbation toproduce triggered outbursts. The latter approach has beendeveloped initially by
Clarke and Syer , (1996) and then by
Lodato and Clarke , (2004), who considered outbursts trig-gered by a massive planet. During quiescence the planetopens up a gap in the disk. For sufficiently massive planets,the resulting Type II migration is slow and the inner disk israpidly emptied out (in what would resemble a transitionaldisk), leading to a steepening of the density profile in theouter disk, that can trigger the thermal-viscous instabilityat the outer gap edge. Observationally, there are a num-ber of clues that indicate the presence of a planet in FUordisks.
Clarke and Armitage , (2003) have suggested that aplanet embedded in the disk of a FUor would lead to a clearspectroscopic signature in the form of a periodic modula-tion of the double-peaked line profiles on the orbital periodof the planet. Such periodic modulations of the line pro-12 ime (kyr)
245 250 255 260
Log ( L / L (cid:1) ) -10123 Time (kyr)
Log ( L / L (cid:1) ) -10123 disk-fragmentation-induced burstsplanet-induced bursts Log ( L / L (cid:1) )
0 5 10 15
Time (kyr) embedded phase early T Tauri phase
Time (kyr)
200 250 300 350
Log ( L / L (cid:1) ) -10123 MRI activation (IDZE trigger) MRI activation (GI trigger)
Fig. 6.—
Top:
Total (accretion plus photospheric) luminosityin the disk fragmentation model showing typical outbursts in theembedded phase of star formation (left) and in the early T Tauriphase (right).
Middle:
Bolometric lightcurves for the planet in-duced thermal instability model. Within this model the recurrencetime between outbursts is reduced as time increases.
Bottom:
To-tal luminosity for the MRI thermally activation models. The solidcurves are from models where MRI is triggered by GI, while thedotted curves are from models where MRI is activated at the innerdead zone edge (IDZE) due to a non-zero dead zone viscosity. files have been observed only for the fast rise FUors (FUOri and V1057 Cyg), with a period of ∼ Herbig etal. , 2003), corresponding to a semi-major axis of 7-10 R ⊙ (see also Powell et al. , 2012).The results of
Lodato and Clarke , (2004) confirm thata planet with a mass of a few Jupiter masses can producea fast rise outburst (with a rise time of the order of oneyear, as observed for FU Ori and V1057 Cyg). The issuewith this class of models is the duration of the outburst, pre-dicted to be of the order of ∼ years by such models, witha relatively shallow dependence on the planet mass. Thistimescale is set by the viscous time on the ionized branch ofthe stability curve at the outermost location reached by theinstability front, that is generally confined within ∼ R ⊙ ,thus making this timescale too small, unless α is assumedto be unrealistically low in the upper branch ( α ∼ − ). Clarke et al. , (2005) compared planet-triggered outburstmodels to a long-term monitoring campaign of FU Ori andV1057 Cyg optical light curves. The time-dependent mod-els were able to match the color evolution of the light curvesmuch better than a simple series of steady-state models withvarying ˙ M . Clarke et al. , (2005) also discussed the sud-den luminosity dips observed for V1057 Cyg at the end ofthe outburst and for V1515 Cyg, while FU Ori does not
Time (yr)
20 30 50 200 300 50010 100 1000
Log ( L / L (cid:1) ) -0.50.00.51.01.52.02.53.03.5 planet-indiced burstdisk-fragmentation burst disk-fragmentation burstMRI burst (GI trigger)MRI burst (IDZE trigger) Fig. 7.—
Time evolution of individual luminosity outbursts indifferent burst-triggering models. The zero-time is chosen ar-bitrarily to highlight distinct models. The two distinct outburstcurves in the disk fragmentation model stem from the state of thefragment when accreted onto the star. Tidally smeared fragmentsgive rise to a slowly rising curve (predominantly, in the embed-ded phase), while a sharply rising curve is produced by weaklyperturbed fragments in the early T Tauri phase. appear to show any such behavior. This non-periodic vari-ability was interpreted as a consequence of the interactionof a disk wind with the infalling envelope. Numerical sim-ulations show that powerful winds are able to push awaythe infalling envelope to large radii clearing up our view tothe inner disk, while for less powerful ones the envelope“crushes down” the wind occulting the inner disk. If thewind strength is proportional to the accretion rate in the in-ner disk, such models can explain the observed behavior.The interaction between the disk and a planet can bemuch more complex if there is mass transfer between theplanet and the star.
Nayakshin and Lodato , (2013) consid-ered the case of a few Jupiter masses inflated planet, migrat-ing within a disk. When the planet opens up a gap, masstransfer between the planet and the star leads to the planetlosing mass but retaining angular momentum, thus migrat-ing out and switching off the mass transfer. Conversely,if the gap is only partially open, a runaway configurationcan occur where the planet migrates further in, enhancingthe mass loss rate and giving rise to a powerful flare. Thismechanism gives rise to variability on several timescalesand of different amplitudes depending on parameters.
Both magnetorotational in-stability (MRI,
Balbus and Hawley , 1998 and referencestherein; see also the accompanying chapter by
Turner et al. )and gravitational instability (GI, see
Durisen et al. , 2007and references therein) are promising mechanisms to trans-fer angular momentum in disks. However, MRI only op-erates when the disk reaches enough ionization (that is, inthe hotter parts of the disk), and GI only operates when thedisk is sufficiently cold to become gravitationally unstable,13ccording to the classical criterion (
Toomre , 1964): Q = c s κπG Σ ≈ c s Ω πG Σ . , (1)where Q is the Toomre stability criterion, Σ is the surfacedensity and the epicyclic frequency κ can be approximatedwith the angular frequency Ω for a quasi-Keplerian disk.A realistic protoplanetary disk is likely to have a compli-cated accretion structure: the inner disk is MRI active dueto the thermal ionization, the region between ∼ AU to tensof AU has a layered accretion structure with a MRI stabledead zone, and the outer region can be MRI active due tocosmic/X-ray ionization. The outer disk can also be gravi-tationally unstable when the disk is at the infall phase witha significant amount of mass loading.This complicated structure is unlikely to maintain asteady accretion (
Gammie , 1996;
Armitage et al. , 2001;
Zhuet al. , 2009b, 2010a;
Martin et al. , 2012ab). The outer disktransfers mass to the inner disk either due to GI or due to thelayered accretion. With more and more mass shoveled tothe inner disk, the inner disk becomes gravitationally unsta-ble and continues to transfer mass to small radii. GI can alsoheat up the disk. Since the electron fraction in protoplane-tary disks depends nearly exponentially on temperature dueto the collisional ionization of potassium (
Gammie , 1996;
Umebayashi , 1983), the gaseous disk will be well coupledto the magnetic field when T ∼ K and MRI starts tooperate. This sudden activation of MRI leads to disk out-bursts. During the outburst, a thermal instability should alsobe triggered at the inner disk as a by-product. Two dimen-sional radiation hydrodynamic simulations have been car-ried out to verify such mechanisms (
Zhu et al. , 2009b).This MRI + GI mechanism shares similarities with thetraditional thermal instability mechanism. For a given disksurface density, the disk has two possible structures: 1) MRIinactive, and 2) MRI active. The switch from one to anotheroccurs at T ∼ K and leads to outbursts (
Martin andLubow , 2011, 2013). In the MRI+GI picture, GI triggersthe switch. However, if there are other mechanisms to heatup the disk and trigger the switch, the disk can also ex-perience outbursts. For example,
Bae et al. , (2013) foundthat the energy diffusion radially from the inner MRI activedisk to the dead zone can trigger the switch and lead to out-bursts, although these outbursts are weaker and shorter thanthe MRI+GI case. This may have implications for weakerFUors or EXors.After the envelope infall stage, the layered accretion canstill pile up mass in the dead zone and leads to disk outbursts(
Zhu et al. , 2010b), which is consistent with the fact thatsome FUors are in the T Tau phase. However, this massivedead zone would persist throughout the whole T Tauri phaseand should be easily observed by ALMA.
Another accretion burstmechanism is related to the phenomenon of disk gravita-tional fragmentation and subsequent inward migration ofthe fragments onto the protostar. Observations of jets andoutflows suggest that the formation of protostellar disks of- ten occurs in the very early stage of star formation, when aprotostar is deeply embedded in a parental cloud core.The forming disk becomes gravitationally unstable, ifthe Q -parameter drops below unity. A higher initial massand angular momentum of the parental core both favor theformation of disks with stronger gravitational instability( Kratter et al. , 2008;
Vorobyov , 2011). For relatively longcooling timescales (above a few dynamical timescales) theoutcome of the instability is to produce a persistent spiralstructure (
Gammie , 2001;
Lodato and Rice , 2004, 2005;
Mej´ıa et al. , 2005;
Boley et al. , 2006), able to transport an-gular momentum efficiently through the disk (
Cossins et al. ,2009, 2010) and trigger the MRI burst (Sect. 3.2.2).In the opposite regime, where the cooling timescale iscomparable to or shorter than the dynamical timescale,these disks can experience gravitational fragmentation andform bound fragments with mass ranging from giant planetsto very-low-mass stars (
Gammie , 2001;
Johnson and Gam-mie , 2003;
Rice et al. , 2005;
Stamatellos and Whitworth ,2009ab;
Vorobyov and Basu , 2010;
Vorobyov , 2013).The likelihood of survival of the fragments is, however,low. GI in the embedded phase is strong, fueled with acontinuing infall of gas from a parent cloud core, and re-sultant gravitational and tidal torques are rampant. As aconsequence, the fragments tend to be driven into the innerdisk and likely onto the protostar due to the loss of angu-lar momentum caused by the gravitational interaction withthe trailing spiral arms (
Vorobyov and Basu , 2006, 2010;
Cha and Nayakshin , 2011;
Machida et al. , 2011). The in-fall of the fragments can trigger mass accretion bursts andassociated luminosity outbursts that are similar in magni-tude to FUor or EXor events (
Vorobyov and Basu , 2005,2006, 2010), during which a notable fraction of the proto-stellar mass can be accumulated (
Dunham and Vorobyov ,2012). The protostellar accretion pattern in the disk frag-mentation models is highly variable (
Vorobyov , 2009) andis characterized by short-duration ( . − yr) burstswith ˙ M & a few × − M ⊙ yr − alternated with longer( – yr) quiescent periods with ˙ M . − M ⊙ yr − .The most intense accretion bursts can reach − M ⊙ yr − .This burst mechanism is most effective in the embeddedstage of disk evolution and diminishes once the parentalcore has accreted most of its mass reservoir onto the pro-tostar + disk system. Only occasional bursts associatedwith the fragments that happened to survive through theembedded stage can occur in the T Tauri stage. Moreover,the initial conditions in the parental core set the likelihoodof disk fragmentation and the burst occurrence; the rele-vant rotational energy vs. initial core mass diagram is pro-vided in Basu and Vorobyov , (2012) and
Vorobyov , (2013).Magnetic fields and strong background irradiation both actto suppress disk fragmentation and associated accretionbursts, but are unlikely to quench this phenomenon com-pletely (
Vorobyov and Basu , 2006, 2010).
The accretionand luminosity burst mechanisms considered in the previ-ous sections are induced by “internal” triggers (
Tassis and ouschovias , 2005). These triggers depend on the physicalconditions in the system and if they are suppressed, the pro-tostar is likely to accrete in a quiescent manner. Neverthe-less, accretion bursts can still be induced in quiescent sys-tems by the so-called “external” triggers, of which a closeencounter in a binary system was the first proposed can-didate ( Bonnell and Bastien , 1992).
Reipurth and Aspin ,(2004b) considered the case where dynamical interactionsin small-N systems might lead to close encounters. Startingfrom the fact that a couple of FUors are found to be in abinary system where both stars are FUor, they argued thatwhatever triggered the FUor eruption in one component islikely to also have triggered the eruption in the other com-ponent, and identified the breakup of a small multiple sys-tem as a natural common precursor event. Indeed, dynam-ical interaction in small-N systems can easily result in theformation of a close binary.
Reipurth and Aspin , (2004b)then argued that the decay of the binary due to interactionwith a circumbinary disk may lead to triggering of insta-bilities in the individual circumstellar disks. A quantitativemodeling of such processes, however, is still lacking.The idea of external triggering received further develop-ment by
Pfalzner et al. , (2008) and
Pfalzner , (2008), whoconsidered accretion processes in young and dense stellarclusters, choosing the Orion nebular cluster as representa-tive. They combined cluster simulations performed with theN-body code and particle simulations that described the ef-fect of a close passage on the disk around a young star todetermine the induced mass accretion. They concluded thatthe close encounters can reproduce the accretion bursts typ-ical for the FUor events, albeit with certain limitations.Close encounters with nearby stars have also been con-sidered as a way to induce fragmentation in a disk thatis gravitationally stable, but not fragmenting in isolation(
Mayer et al. , 2005), although the effectiveness of such pro-cess is probably limited (
Lodato et al. , 2007).However, the short rise times are difficult to achieve un-less the matter is stored somewhere close to the protostarand the inferred duration of FU Ori events requires a ratherhigh mass ratio between the intruder and the target.
Episodic accretion has several key implications for starand planet formation and evolution.
In one of the first statistically significant studiesof the luminosities of embedded protostars,
Kenyon et al. ,(1990, 1994) and
Kenyon and Hartmann , (1995) found thatthe observed luminosities of 23 protostars in Taurus weremuch lower than expected from simple theoretical predic-tions. Current samples of protostars with accurately deter-mined luminosities measure in the hundreds (e.g.,
Evanset al. , 2009;
Enoch et al. , 2009;
Kryukova et al. , 2012;
Dunham et al. , 2013; also see the accompanying chapterby
Dunham et al. ). While the details of these studies differ,they all confirm the existence of the luminosity problem and aggravate it by showing that the luminosity distribution ofprotostars extends to even lower luminosities than found bythe
Kenyon et al. surveys.As originally proposed by
Kenyon and Hartmann ,(1995), one possible resolution to the “luminosity prob-lem” is episodic accretion. If a significant fraction of thetotal accretion onto a star occurs in short-lived bursts, mostprotostars will be observed during periods of accretion be-low the mean accretion rate, thus most protostars will emitless accretion luminosity than expected assuming constantaccretion at the mean rate. To test the ability and necessityof episodic accretion for resolving the luminosity prob-lem,
Dunham et al. , (2010) modified existing evolutionarymodels of collapsing cores first published by
Young andEvans , (2005) and showed that a very simple implemen-tation of episodic accretion into the accretion model wascapable of resolving the luminosity problem and matchingthe observed protostellar luminosity distribution.
Offnerand McKee , (2011) also found that episodic accretion cancontribute a significant fraction of the stellar mass.
Dun-ham and Vorobyov , (2012) revisited this topic with moresophisticated evolutionary models incorporating the exacttime evolution of the accretion process predicted by hy-drodynamical simulations, and confirmed these findings.However, these results only demonstrate that episodic ac-cretion is a possible solution to the luminosity problem.
Stellar irradiationis known to have a profound impact on the gravitationalstability of protostellar disks (
Stamatellos and Whitworth ,2009ab;
Offner et al. , 2009,
Rice et al. , 2011). Flareddisks can intercept a notable fraction of the stellar UV andX-ray flux, which first heats dust and then gas via colli-sions with dust grains. The net effect is an increase in thegas/dust temperature leading to disk stabilization. X-rayscan also ionize gas. SPH simulations by
Stamatellos et al. ,(2011, 2012) demonstrate that quiescent periods betweenthe luminosity bursts can be sufficiently long for the disk tocool and fragment. Thus, episodic accretion can enable theformation of low-mass stars, brown dwarfs, and planetary-mass objects by disk fragmentation, as confirmed by therecent numerical hydrodynamical simulations presented by
Vorobyov , (2013).
The lumi-nosity spread observed in Hertzsprung-Russell (HR) dia-grams of young star-forming regions is a well known fea-ture (
Hillenbrand , 2009), but its origin is uncertain. It canbe attributed to observational uncertainties, significant agespread, or yet unknown physical processes. As
Baraffe etal. , (2009) demonstrated, protostellar evolutionary modelsassuming episodic accretion are able to reproduce the ob-served luminosity spread for objects with final ages of afew Myr. In contrast, non-accreting models require an agespread of at least 10 Myr.
Baraffe et al. , (2009), thus, con-cluded that the observed HR luminosity spread does notstem entirely from an age spread, but rather from the im-pact of episodic accretion on the physical properties of theprotostar. This idea was recently called into question by15 osokawa et al. , (2011), who argued that accretion variabil-ity had little effect on the evolution of low-mass protostarswith effective temperature below 4000 K.
Baraffe et al. , (2012) noted that
Hosokawa et al. , (2011)considered models with the initial protostellar “seed” massof 10 M Jup . Having provided justification for a smaller ini-tial mass of protostars on the order of 1.0 M Jup , Baraffe etal. , (2012) showed that the luminosity spread in the HR di-agram and the inferred properties of FU Ori events (stellarradius, accretion rate) both can be explained by the “hybrid”accretion scenario with variable accretion histories derivedfrom disk fragmentation models (Sect. 3.2.2). In this accre-tion scenario, a protostar absorbs no accretion energy belowa threshold accretion rate of − M ⊙ yr − and 20% of theaccretion energy above this value.Several important implications of protostellar evolution-ary models with variable accretion were emphasized by Baraffe et al. , (2012). First, each protostar/brown dwarfexperiences its own accretion history and ends up randomlyin the HR diagram at the end of the accretion phase. Itis likely that the concept of a birthline does not apply tolow-mass ( < . M ⊙ ) objects. Moreover, age determina-tions from “standard” non-accreting isochrones may over-estimate the age of young protostars by a factor of several ormore. Finally, inferring masses from the HR diagram usingisochrones of non-accreting protostars can yield severelyincorrect determinations, possibly overestimating the massby as much as 40% or more. Large changes in the ac-cretion luminosity of young stars due to episodic accretioncan drive significant chemical changes in the surroundingcore and disk. Several authors have published chemicalmodels exploring these effects and have shown that the COice evaporates into the gas-phase in the surrounding enve-lope during episodes of increased luminosity, affecting theabundances of many other species through chemical reac-tions (e.g.,
Lee , 2007;
Visser and Bergin , 2012;
Vorobyovet al. , 2013). Many of these effects can endure long afteran accretion burst has subsided, leading to non-equilibriumchemistry compared to that expected from the currently ob-served protostellar luminosity. In particular, the abundanceof gas-phase CO in the envelope can be used as an indi-cator of past accretion bursts and perhaps even a means ofmeasuring the time since the last burst (Fig. 8;
Vorobyovet al. , 2013). Episodic accretion bursts can also affect theabundances and chemical compositions of various molecu-lar ices frozen onto dust grains.
Kim et al. , (2012) showedthat a chemical evolutionary model including episodic ac-cretion could provide the necessary thermal processing andmatch the 15.2 µ m CO ice absorption features of low-luminosity protostars. If a protostar has a strong magnetic field, then the disk-magnetosphere interaction can lead to episodic accretion,variabilities at different time-scales, and outflows. In most Lu m i no s i t y ( L (cid:1) ) stellar luminosityTime (kyr)
80 85 90 95 100 C O ga s - pha s e f r a c t i on CO gas-phase fraction in the envelope
Fig. 8.—
Predicted CO gas-phase fraction ξ CO (dashed lines) andtotal stellar luminosity L ∗ (solid lines) vs. time elapsed since theformation of the protostar. The correlation between ξ CO and L ∗ isevident. In particular, ξ CO steeply rises during the burst to a max-imum value and gradually declines to a minimum value after theburst. The relaxation time to the pre-burst stage is notably longerthan the burst duration. Adapted from Vorobyov et al. , (2013). cases, the magnetic fields of protostars are not known.However, the measurements of the field in several CTTSindicate the presence of a magnetic field of a few kG (e.g.,
Johns-Krull , 2007;
Donati et. al. , 2008). The possible pres-ence of a 1 kG magnetic field in 0.05 AU of FU Ori wasreported by
Donati et al. , (2005) and
Green et al. , (2013a)estimated the magnetic field of HBC 722 to be 2.2–2.7 kG.There is also indirect evidence of a magnetic field from theX-ray emissions of FUors and EXors. If a star has a mag-netic field of a few kG, then the disk will be truncated by themagnetosphere at distance r m , where the magnetic stress isequal to the matter stress in the disk ( K¨onigl , 1991). InEXors, this radius may be as large as a few stellar radii,while in FUors it can be smaller than one stellar radius, orthe field may be buried. Numerical simulations predict thateven in the case of a tiny magnetosphere, the magnetic fluxof the star is not buried, but rather partially inflated into thecorona, and may drive strong outflows (
Lii et al. , 2012).
A model of cyclic accretion was proposed for stars accret-ing in a weak propeller regime. In the propeller regime,the magnetosphere rotates more rapidly than the inner disk,and matter of the disk can be ejected to outflows by arapidly-rotating magnetosphere (
Illarionov and Sunyaev ,1975;
Lovelace et al. , 1999). However, in a weak propellerregime, the magnetospheric radius, r m , is only slightlylarger than the corotation radius, r c (where the angular ve-locity of the Keplerian disk is equal to the angular velocityof the star), and such a weak propeller cannot drive out-flows. Instead, the star transfers its excess angular mo-mentum to the disk, matter accumulates in the disk for along period of time, and a “dead disk” is formed; then, partof this disk matter accretes to the star, the magnetosphereexpands and the process repeats in a cyclic fashion ( Baan ,1977;
Sunyaev and Shakura , 1977;
Spruit and Taam , 1993).16he time-scale of accretion episodes is determined by theaccretion rate in the disk and other parameters.
D’Angeloand Spruit , (2010, 2012) investigated this model for a widerange of parameters at which accretion is either cyclic orsteady and showed that the model can explain EXor out-bursts.
Different models have been proposed to explain high-velocity winds of FUors and EXors (see Sect. 4.4), in-cluding: the disk winds model, where matter is driven bycentrifugal force along the inclined field lines of the disk(e.g.,
Blandford and Payne , 1982;
Zanni et al. , 2007);accretion-powered stellar winds (e.g.,
Matt and Pudritz ,2005); and the X-winds, which are launched centrifugallyfrom the disk-magnetosphere boundary (
Shu et al. , 1994).The reader is encouraged to read the accompanying chapterby
Frank et al. . Conical Winds.
Recent numerical simulations show anew type of wind which can be important in the casesof EXors and FUors. These winds form at the disk-magnetosphere boundary during the episodes of high accre-tion rate. The newly-incoming matter compresses the mag-netosphere of the star, the field lines inflate due to differ-ential rotation between the disk and the star, and conically-shaped winds flow from the inner disk (
Romanova et al. ,2009;
Kurosawa and Romanova , 2012). These winds aredriven by the magnetic force, F M ∼ −∇ ( rB φ ) , whicharises due to the wrapping of the field lines above the disk( Lovelace et al. , 1991). They are also gradually collimatedby the magnetic hoop-stress, and can be strongly colli-mated for high accretion rates (
Lii et al. , 2012). Moreover,the star can rotate much more slowly than the inner disk (at r m ≪ r c ). This is different from the X-winds, which re-quire the condition of r m ≈ r c . Conical winds appear dur-ing a burst of accretion and continue for the entire durationof the burst. A magnetic field of a few kG is required forFUors, while in EXors the field can be weaker. The conicalwind model was compared with the empirical model basedon the spectral analysis of the winds in FU Ori ( Calvet etal. , 1993;
Hartmann and Calvet , 1995). A reasonably goodagreement was achieved between these models (
K¨onigl etal. , 2011).
Propeller-driven Winds.
If a protostar rotates much morerapidly than the inner disk (strong propeller regime) andthe accretion rate in the disk is relatively high, then a sig-nificant part of the disk can be redirected to the outflows bythe rapidly-rotating magnetosphere (
Romanova et al. , 2005,2009;
Ustyugova et al. , 2006). In this regime, accretion andoutflows also occur in cycles, where matter accumulates inthe inner disk, diffuses through the field lines of the rapidly-rotating magnetosphere, and is ejected to the outflows; then,the magnetosphere expands and the cycle repeats (
Goodsonet al. , 1997;
Lii et al. , 2013). The time-scale of the cyclevaries from a few weeks to a few months, and depends on anumber of parameters, such as the accretion rate in the diskand the diffusivity at the disk-magnetosphere boundary.In the case of a rapidly-rotating star, outflows have a second component: a magnetically-dominated, low-densityand high-velocity jet, where matter is accelerated rapidlyby the magnetic force that appears due to the winding ofthe stellar field lines (a “magnetic tower”). The jet car-ries significant energy and angular momentum from the starto the corona, causing the star to spin down rapidly (e.g.,
Romanova et al. , 2005). If protostars accrete most of themass during episodes of enhanced accretion, then powerfuloutflows observed can be associated with the episodes ofstrongest accretion (e.g.,
Reipurth , 1989). These outflowscarry matter and angular momentum into the cloud and mayinfluence the overall dynamics of star formation.
During aburst of accretion, different processes are expected to oc-cur at the disk-magnetosphere boundary. Matter may flowto the star above the magnetosphere in two ordered fun-nel streams and form two hot spots on the stellar surface(
Bertout et al. , 1988;
Romanova et al. , 2004). Alterna-tively, matter may accrete through the magnetic Rayleigh-Taylor instability, where several unstable “tongues” pene-trate the field lines and form spots of chaotic shape andposition (
Kulkarni and Romanova , 2008;
Romanova et al. ,2008). In these cases, the light curve and spectral changesappear chaotic, with a few accretion events per period ofthe inner disk (
Kurosawa and Romanova , 2013). Observa-tions of young stars often show variability at this time scale(e.g.,
Alencar et al. , 2010;
Cody et al. , 2013). In the unsta-ble regime, the tongues rotate with the angular velocity ofthe inner disk, and the frequency of the inner disk may bepresent in the Fourier spectrum of the light-curves. Varia-tions of the accretion rate will lead to variations of the innerdisk radius and this frequency will vary in time. In cases ofsmall magnetospheres, r m . (1 − R ⋆ , one or two reg-ular tongues rotate orderly with the frequency of the innerdisk ( Romanova and Kulkarni , 2009;
Bachetti et al. , 2010).This phenomenon can potentially explain the quasi-periodicvariability of 2-9 days, which has been recently observed inFU Ori and Z CMa by
Siwak et al., (2013). Alternatively,it can be connected with the waves excited in the disk bythe tilted dipole (e.g.,
Bouvier et al. , 1999;
Romanova etal. , 2013). The modulation of the blueshifted spectral linesin FU Ori with a period of 14 days (
Powell et al. , 2012)may be a sign of modulation of the wind by the waves inthe disk. Different longer-period variabilities (e.g.,
Hillen-brand et al. , 2012) can also be connected with the waves inthe disk, excited at different distances from the star.Another type of variability may be connected with episodic inflation and reconnection of the field lines con-necting a star with the inner disk (
Aly and Kuijpers , 1990).The signs of such variability on a time-scale of a few stel-lar rotations have been observed by
Bouvier et al. , (2003)in AA Tau and in young bursting stars by
Findeisen et al. ,(2013). This process may also lead to the phenomenon ofepisodic X-ray flares during the burst.
4. CURRENT VIEW OF EPISODIC ACCRETION
The recently discovered examples of outbursts have be-gun to fill in the notional gap between long-duration (clas-sical FUor) and short-duration (classical EXor) eruptions(Fig. 2). Several objects (e.g., OO Ser, V1647 Ori) appearto display outburst decay times of a few years, i.e., inter-mediate between the two classes. Such a conceptual pro-gression from a bimodal distribution to a continuum of out-burst timescales is hence perhaps merely the natural conse-quence of the discovery of additional outbursting YSOs andpre-main-sequence stars, combined with the longer base-line and more comprehensive arsenal of observational dataavailable to measure outburst decay times for both previ-ously and newly identified eruptive objects.On the other hand, the question of a clear distinction be-tween FUor and EXor classes in terms of outburst repeti-tion and duty cycle remains open. The repetitive nature ofEXor outbursts might be considered a defining characteris-tic of such objects — one that potentially links strong, short-timescale EXor eruptions to the lesser variability of TTSmore generally. In contrast, we simply have not had time toobserve the cessation of any of the classical FUor outburstsand, hence, we are unable to assess the potential for repe-tition of these long-duration eruptions. But essentially alltheoretical models predict that FUor outbursts should alsooccur more than once during early stellar evolution, with anaverage time span of thousands of years between outbursts(Sect. 3.2; Fig. 6);
Scholz et al. , (2013) estimated time in-tervals between outbursts of 5–50 kyr.FUor and EXor systems are often surrounded by ring- orcometary-shaped nebulae that evidently reside in the out-flow and/or the parent cloud (e.g.,
Goodrich , 1987). Thesebright optical/infrared reflection nebulae become illumi-nated by the central stars as their activity levels increase(sometimes revealing light-travel-time effects that facilitatedistance and luminosity determinations; e.g.,
Brice˜no et al. ,2004). Furthermore, HH objects and jets tend to be asso-ciated with sources that have recently entered elevated ac-tivity states (e.g.,
Reipurth and Aspin , 1997;
Takami et al. , 2006). Thus, outflow structures might be used to infer pro-tostellar and pre-main-sequence outburst duty cycles. How-ever, any direct connection between outflow and outburstactivity is difficult to infer; e.g., the time intervals betweenthe appearance of large working surfaces in HH flows, typi-cally 500 − Reipurth and Bally , 2001).
Just as the outburst duration gap between FUors andEXors has closed, the long-held view that longer durationeruptions involve higher protostellar accretion rates and oc-cur at earlier stages of pre-main-sequence stellar evolution(e.g.,
Hartmann and Kenyon , 1996;
Sandell and Weintraub ,2001) has become subject to question. This increasing am-biguity is in large part the result of the recent scrutiny ofFUors via higher-resolution infrared and submm imagingand spectroscopy. It is now apparent that classical FUorspresent a highly heterogeneous set of near- to mid-infraredspectral features and SEDs that could possibly relate to anevolutionary sequence within the FUor class (e.g.,
Quanzet al. , 2007c). In this scheme, the presence or (partial)depletion of an envelope may significantly modify the ob-servational properties (e.g., silicate in absorption or emis-sion, strong far-infrared excess). However, new outburstingsources have demonstrated that the case may not be as sim-ple: there are often more deeply embedded protostars lyingin close proximity to optically-identified eruptive objectsthat show FUor characteristics (e.g.,
Green et al.,
Dunham et al. , 2012). While some FUors appear relativelydevoid of significant circumstellar envelopes, it also appearsthat some FUors do possess massive, molecule-rich circum-stellar envelopes (e.g.,
K´osp´al et al. , 2011a). The behav-ior and strength of outbursts may also be influenced by thepresence of a wind/outflow and its interaction with a sur-rounding envelope (
Clarke et al. , 2005).Theoretically, it is difficult to provide a single evolution-ary framework to explain both EXors and FUors. The diffi-culty with providing such a unified scenario is probably dueto different causes. First, it is extremely difficult to self-consistently model the evolution of the entire disk, startingfrom sub-AU to hundred-AU scales, and as a result mosttheoretical models focus on either the inner or outer disk.Second, only few attempts have been made up to now todirectly compare theoretical models to actual observations,e.g., in terms of detailed light curve or color evolution mod-eling for specific events. As the models become more so-phisticated and observations richer, this is certainly a direc-tion to pursue in the future. Third, very little has been donefrom the theoretical standpoint in order to describe EXors.The models of
D’Angelo and Spruit , (2010, 2012) appearto be promising but still lack a detailed time-dependent cal-culation. The explanation of the origin of EXor outburstsas related to the innermost parts of the disk might establishthem as a separate class with respect to FUors.18hen focusing on individual theoretical models, thosepresented in
Nayakshin and Lodato , (2012; presented inSect. 3.2.1) do show variability on a variety of timescalesand amplitude that might reproduce, within a single sce-nario, both EXors and FUors. The models of
Vorobyov andBasu , (2005, 2006, 2010; presented in Sect. 3.2.3) produceluminosity outbursts with amplitudes typical for both FUorsand EXors but fail to reproduce the short rising times ofEXors, possibly due to limitations of the numerical code.Certainly, theoretical models do require significant devel-opment before a proper comparison with data can be made.Finally, radiative transfer models should attempt tomodel the behavior of disk interiors and their atmospheresduring outbursts, e.g., as a function of mass accretion rate ordisk and envelope masses. Such models should determinewhether such parameters can explain the CO bandhead inabsorption in FUors but in emission in EXors, and pursuewhether such CO features can change depending on theconditions, possibly explaining, e.g., the reversal of COobserved early in the outburst of V1647 Ori.
The search for tight binaries in FUors and EXors islinked with the quest to identify the origin of eruptingevents.
Bonnell and Bastien , (1992) proposed that FUoroutbursts may be due to a perturbation induced by a com-panion at periastron passage. Noting that there is already alist of known FUor binaries,
Reipurth and Aspin , (2004b)proposed that FUors may be newborn binaries that have be-come bound when a small nonhierarchical multiple systembreaks up and the two components spiral in toward eachother, perturbing their disks. This model derives particularmotivation from the observation that FU Ori is the northerncomponent in a close ( . ′′ ), pre-main-sequence binary sys-tem ( Reipurth and Aspin , 2004b;
Wang et al. , 2004) whosenon-outbursting component, FU Ori S, is likely more mas-sive than the FUor namesake (
Beck and Aspin , 2012;
Pueyoet al. , 2012). The scenario advanced by
Reipurth and Aspin ,(2004b) implies that FU Ori must be a close binary ( < Malbet et al. , 1998, 2005;
Quanz et al. , 2006)and, if so, the newly discovered companion is the outlyingmember in a triple system. Z CMa, a . ′′ FUor/Herbig Bebinary surrounded by a circumbinary disk (
Alonso-Albi etal. , 2009) — and with jets emanating from both components(
Whelan et al. , 2010;
Benisty et al. , 2010;
Canovas et al. ,2012) — is another potential example of a system that hasundergone binary-disk interactions, although the observedoutbursts in this binary do not always stem from FUor vari-ability (
Teodorani et al. , 1997; van den Ancker et al. , 2004;
Szeifert et al. , 2010;
Hinkley et al. , 2013).If such binary-disk interaction is the dominant mecha-nism to trigger FUor outbursts, then FUor eruptions shouldpreferentially occur in close binaries, i.e., in about 20% ofall stars. However, V1057 Cyg and V1515 Cyg, both “clas-sical” FUors, are not known to harbor close companions, al-though it is difficult to probe the 1–10 AU separation range, the most relevant for the binarity model, due to their dis-tances. In the case of EXors, some stars are known visualor spectroscopic binaries (e.g., V1118 Ori at 72 AU separa-tion,
Reipurth et al. , 2007b; UZ Tau E with a ∼ . AU,
Prato et al. , 2002; EX Lup might also harbor a brown dwarfcompanion,
K´osp´al , priv. comm.), while others show no ev-idence of binarity (e.g.,
Melo , 2003;
Herbig , 2007, 2008).In summary, the jury is still out on whether binarity playsany role in triggering eruption events in EXors or FUors;further studies aiming at discovering faint companions orpossibly planets in disks of erupting stars are clearly neededin the coming years. The reader is encouraged to read theaccompanying chapter by
Reipurth et al.
Pre-main sequence accretion rates are difficult to deter-mine and, for a given class of object, are determined via avariety of means and, thus, highly inhomogeneous. Erup-tive stars are no exception to this rule. For instance, thephotometric data and flux-calibrated spectra of the outburstphase can be compared with SED models to constrain theaccretion disk parameters. Correlations between mass ac-cretion rates and the emission line fluxes, obtained in theframework of magnetospheric accretion, can also be used.Nevertheless, some general statements can be madeabout accretion rates. The most luminous FUors have massaccretion rates that can reach − − − M ⊙ yr − (seeTab. 1). However, eruptive FUor-like objects with lowerluminosities imply lower mass accretion rates, as low as10 − M ⊙ yr − . Hence, there is a clear overlap between theranges of mass accretion rate observed during outburst inFUors and classical EXors (10 − –10 − M ⊙ yr − ) or inter-mediate objects ( ∼ − M ⊙ yr − ). Quiescent mass ac-cretion rates can start as low − M ⊙ yr − (e.g., Sipos etal. , 2009) for EXors with little or no envelope, i.e., probingepisodic accretion in later evolutionary stages. In summary,the peak mass accretion rate is likely not the only physicalparameter that determines the nature of the eruption.In optical/infrared spectra, EXors ubiquitously show in-verse P Cyg profiles due to infall (e.g.,
Herbig , 2007). Inaddition, blueshifted absorption and/or P-Cyg profiles areobserved in spectral lines such as H α and Na I D, indicat-ing strong winds. The wind velocities are typically − to − km s − , with maximum values up to − km s − in FUors (e.g., Croswell et al. , 1987;
Vacca et al. , 2004;
Reipurth and Aspin , 2004a). The maximum velocities ap-pear generally lower in EXors, although some fast wind canbe detected as well (e.g., V1647 Ori:
Aspin and Reipurth ,2009). In addition, blueshifted absorption is stronger inFUors than in EXors (e.g.,
Herbig , 2007, 2008, 2009).Many FUors also show millimeter signatures of CO out-flows. The typical velocity and mass loss rate of the out-flows are 10 −
40 km s − and 10 − − − M ⊙ yr − , respec-tively ( Evans et al. , 1994). However, some objects do notshow significant CO emission associated with outflows. Asdiscussed in Sect. 4.1, the connection between HH objects19nd outbursts is also difficult to infer. However, the strengthof the disk wind increases during EXor outbursts (e.g., EXLup;
Sicilia-Aguilar et al. , 2012) and decays as the outburstdecays (e.g.,
Aspin et al. , 2010). This relation indicates thatthe wind strength is, indeed, related to the mass accretionrate, as generally observed in CTTS (
Calvet , 1997).
The scant existing high-energy data make apparent thatepisodes of high accretion rate, whether of sustained ormore transient nature, are usually accompanied by en-hanced X-ray emission (see Sect. 2.10). In the case of sus-tained outbursts (i.e., for “classical” FUors and V1647 Ori),the emission (when detected) is overall relatively hard, be-traying an origin in magnetic activity. For shorter-durationoutbursts (as in the “classical” EXors), at least some of theenhanced X-ray flux appears to arise in accretion shocks.V1647 Ori appears to represent something of a hybrid case,in that its emission during high-accretion states is domi-nated by plasma that is too hot to be due to accretion shocks,yet is confined to “hot spots” very near the stellar surface(
Hamaguchi et al. , 2012). Further X-ray observations areneeded to more firmly establish whether and how the X-ray flux levels and plasma temperatures of eruptive youngstars correlate with both long- and short-term variations inoptical/infrared fluxes and other (e.g., emission-line-based)accretion and outflow signatures.
5. FUTURE DIRECTIONS
We identify here a number of potentially interesting di-rections that will or should be explored in the field ofepisodic accretion: • Continuum and molecular line images with ALMAwill provide new opportunities to firmly establish the enve-lope vs. disk masses of FUors and EXors, so as to comparewith each other and with those of deeply embedded proto-stars and CTTS, and to study the chemistry taking place indisks and being modified due to episodic accretion events. • The output of present and forthcoming generations oflarge-field optical monitoring facilities (e.g., Digitized SkySurvey, Palomar Transient Factory, Large Synoptic SurveyTelescope) will continue to enlarge the sample of eruptivepre-main-sequence objects. We can potentially take advan-tage of these data to deduce the frequency of eruptive ob-jects, and determine accretion burst duty cycles, as func-tions of mass and class. However, these identifications willnot include protostars at very early (cloud- and/or envelope-embedded) pre-main-sequence evolutionary stages. Norwill they allow continuous time monitoring to study timevariability over long timescales. • Our present understanding of episodic accretion is po-tentially heavily biased, due to our “traditional” reliance onoptically-detected eruptions in identifying FUors and EX-ors. The identification of eruptions in the near-infrared has begun to mitigate this bias somewhat, mainly thanksto the 2MASS survey.
WISE has now provided an all-skymid-infrared snapshot against which future wide-field mid-infrared imaging surveys can be compared, so as to identifyeruptions associated with much more deeply embedded pro-tostars (see
Antoniucci et al. , 2013;
Johnstone et al. , 2013;
Scholz et al. , 2013). Armed with such identifications, wecan begin to more accurately pinpoint the epoch of onset ofepisodic accretion during protostellar evolution, and obtainfollow-up observations from the ground or in space. • It is worth considering the extent to which dramaticaccretion-driven outbursts effectively cause young stars to“revert” to earlier stages of protostellar evolution. In otherwords, if we observe a Class II object or flat-spectrumsource enter a FUor outburst (e.g., HBC 722), are we ineffect seeing a born-again Class I protostar? • The understanding of outburst feedback on the innerdisk structure (crystallization, chemistry) would profit fromfurther investigation, especially in the region of the forma-tion of Earth-like planets. Such outbursts may, indeed, havetaken place in the history of our Solar system. High angu-lar resolution observations will, thus, help discern structureand physical conditions in the inner disk and search for veryclose companions. • Further modeling of the effect of episodic accretion onthe disk structure should be considered. The CO spectrumin absorption observed continuously in FUors but rarelyor transiently in EXors has traditionally been explained byheating of the disk interior during the accretion event, as-suming built-up material falling from the envelope. How-ever, some FUors, including FU Ori, do not show evidenceof massive envelopes. Thus, it remains unclear why theyshow these typical FUor spectral characteristics. • Finally, it could be worthwhile to investigate numer-ically the link between EXors and FUors by treating theinner and outer disk simultaneously, although this may beout of reach in the near (and mid) term.These and other future efforts should continue to fo-cus on the fundamental physical processes underlying out-bursts, such as narrowing down the possible mechanismsthat can lead to accretion bursts, identifying the key sys-tem parameters that control burst energetics (amplitude, du-ration, repetition), and constraining the range of key sys-tem parameters such as accretion rates, outflow rates, andstar/disk/outflow geometries.
Acknowledgments.
We dedicate this review to the lateGeorge H. Herbig, who passed away on October 12, 2013,for his pioneering and long-lasting work on eruptive youngstars. We acknowledge the fantastic works of a long listof authors mentioned in the reference list that have con-tributed to our knowledge of episodic accretion in star andplanet formation, among them Bo Reipurth whom we wishto thank for carefully reading the manuscript and provid-ing detailed comments as referee of this chapter. Hen-rik Beuther is also thanked for providing further edito-20ial comments to improve the manuscript. We are grate-ful to Michael Richmond for providing the V1647 Ori lightcurves in Fig. 4 and M´aria Kun for reading the manuscriptand providing corrections. Finally we thank several PPVIparticipants for coming forward and providing useful com-ments and feedback to improve this review, among themWilliam Fischer, Chris McKee, and Stella Offner.We have attempted to include all refereed publicationsfrom 1996 until 2013 that are relevant to the study ofepisodic accretion. We apologize if we have unintentionallymissed publications. MMR was supported by NSF grantAST-1211318, GL by PRIN MIUR 2010-2011, project2010LY5N2T, ZZ by NASA HST-HF-51333.01-A, and P ´Aand ´AK partly by OTKA 101393.
REFERENCES ´Abrah´am P., et al. (2004a)
Astron. Astrophys., 419 , L39.´Abrah´am P., et al. (2004b)
Astron. Astrophys., 428 , 89.´Abrah´am P., et al. (2006)
Astron. Astrophys., 449 , L13.´Abrah´am P., et al. (2009)
Nature, 459 , 224.Acosta-Pulido J. A., et al. (2007)
Astron. J. 133 , 2020.Adams F. C., et al. (1988)
Astrophys. J., 326 , 865.Alencar S. H. P., et al. (2010)
Astron. Astrophys., 519 , A88.Alonso-Albi T., et al. (2009)
Astron. Astrophys., 497 , 117.Aly J. J. and Kuijpers J. (1990)
Astron. Astrophys., 227 , 473.Andrews S. M. and Williams J. P. (2005)
Astrophys. J., 631 , 1134.Andrews S. M., et al. (2004)
Astrophys. J., 610 , L45.Antoniucci S., et al. (2013)
New A, 23,
Mon. Not. R. Astron. Soc., 324 , 705.Aspin C. (2011a)
Astron. J., 141 , 196.Aspin C. (2011b)
Astron. J., 142 , 135.Aspin C. and Reipurth B. (2003)
Astron. J., 126 , 2936.Aspin C. and Reipurth B. (2009)
Astron. J., 138 , 1137.Aspin C. and Sandell G. (1994)
Astron. Astrophys., 288 , 803.Aspin C. and Sandell G. (2001)
Mon. Not. R. Astron. Soc., 328 ,751.Aspin C., et al. (2006)
Astron. J., 132,
Astron. J., 135,
Astron. J., 137,
Astron. J., 137 , 2968.Aspin C., et al. (2009c)
Astrophys. J., 692,
L67.Aspin C., et al. (2010)
Astrophys. J., 719 , L50.Audard M., et al. (2005)
Astrophys. J., 635 , L81.Audard M., et al. (2010)
Astron. Astrophys., 511 , A63.Baan W. (1977)
Astrophys. J., 214 , 245.Bachetti M., et al. (2010)
Mon. Not. R. Astron. Soc., 403 , 1193.Bae J., et al. (2013)
Astrophys. J., 764 , 141.Balbus S. A. and Hawley J. F. (1998)
Rev. Modern Phys., 70,
Astrophys. J., 745 , 90.Banzatti A., et al. (2012b)
Astrophys. J., 751 , 160.Baraffe I., et al. (2009)
Astrophys. J., 702 , L27.Baraffe I., et al. (2012)
Astrophys. J., 756,
Astron. J., 142 , 141.Basu S. and Vorobyov E. I. (2012)
Astrophys. J., 750 , 30.Beck T. L. and Aspin C. (2012)
Astron. J., 143 , 55.Bell K. R. and Lin D. N. C. (1994)
Astrophys. J., 427 , 987.Bell K. R., et al. (1995)
Astrophys. J., 444 , 376.Bell K. R., et al. (2000) In
Protostars and Planets IV (V. Manningset al., eds.), pp. 897-926. Univ. of Arizona, Tucson.Benisty M., et al. (2010)
Astron. Astrophys., 517 , L3. Bertin G. and Lodato G. (1999)
Astron. Astrophys., 350 , 694.Bertout C., et al. (1988)
Astrophys. J., 330 , 350.Blandford R. D. and Payne D. G. (1982)
Mon. Not. R. Astron. Soc.,199 , 883.Boley A.C., et al. (2006)
Astrophys. J., 651 , 517.Bonnell I. and Bastien P. (1992)
Astrophys. J., 401 , L31.Bouvier J., et al. (1999)
Astron. Astrophys., 349 , 619.Bouvier J., et al. (2003)
Astron. Astrophys., 409,
Astrophys. J., 606,
L123.Brittain S. D., et al. (2007)
Astrophys. J., 670 , L29.Brittain S. D., et al. (2010)
Astrophys. J., 708,
IAU Symp. 182, Herbig-Haro Flows and theBirth of Stars (B. Reipurth and C. Bertout, eds.), pp. 417-432.Kluwer, Dordrecht.Calvet N., et al. (1991a)
Astrophys. J., 383 , 752.Calvet N., et al. (1991b)
Astrophys. J., 380 , 617.Calvet N., et al. (1993)
Astrophys. J., 402 , 623.Canovas H., et al. (2012)
Astron. Astrophys., 543 , A70.Caratti o Garatti A., et al. (2011)
Astron. Astrophys., 526 , L1.Caratti o Garatti A., et al. (2013)
Astron. Astrophys., 554 , A66.Casali M. M. (1991)
Mon. Not. R. Astron. Soc., 248,
Mon. Not. R. Astron. Soc.,415 , 3319.Chavarria C. (1981)
Astron. Astrophys., 101,
Astrophys. J., 445,
Astrophys. J., 751 , 118.Chochol D., et al. (2006)
Contrib. Astron. Obs. Skalnate Pleso,36 , 149.Chou M.-Y., et al. (2013)
Astron. J., 145 , 108.Clarke C. J. and Armitage P. J. (2003)
Mon. Not. R. Astron. Soc.,345
Mon. Not. R. Astron. Soc., 278,
L23.Clarke C. J., et al. (1990)
Mon. Not. R. Astron. Soc., 242 , 439.Clarke C. J., et al. (2005)
Mon. Not. R. Astron. Soc., 361 , 942.Cody A.-M., et al. (2013),
Astron. J., 145 , 79.Coffey D., et al. (2004)
Astron. Astrophys., 419 , 593.Cohen M., et al. (1981)
Astrophys. J., 245 , 920.Cohen M., et al. (1983)
Astrophys. J., 273 , 624.Connelley M. S. and Greene T. P. (2010)
Astron. J., 140,
Mon. Not. R. Astron. Soc., 393 , 1157.Cossins P., et al. (2010)
Mon. Not. R. Astron. Soc., 401 , 2587.Costigan G., et al. (2012)
Mon. Not. R. Astron. Soc., 427 , 1344.Covey K. R., et al. (2011)
Astron. J., 141 , 40.Croswell K., et al. (1987)
Astrophys. J., 312 , 227.D’Angelo C. R. and Spruit H. C. (2010)
Mon. Not. R. Astron. Soc.,406 , 1208.D’Angelo C. R. and Spruit H. C. (2012)
Mon. Not. R. Astron. Soc.,420 , 416.Dent W. R. F., et al. (1998)
Mon. Not. R. Astron. Soc., 301,
Nature, 438 , 466.Donati J.-F., et al. (2008)
Mon. Not. R. Astron. Soc., 386 , 1234.Dunham M. M. and Vorobyov E. I. (2012)
Astrophys. J., 747 , 52.Dunham M. M., et al. (2010)
Astrophys. J., 710 , 470.Dunham M. M., et al. (2012)
Astrophys. J., 755 , 157.Dunham M. M., et al. (2013)
Astron. J., 145,
Protostars and Planets V (B.Reipurth, D. Jewitt, and K. Keil, eds.), pp. 607-622. Univ. ofArizona, Tucson.Eisl¨offel J., et al. (1990)
Astron. Astrophys., 232,
Astrophys. J., 383 , L19.Eisner J. A. and Hillenbrand L. A. (2011)
Astrophys. J., 738 , 9. lias J. H. (1978) Astrophys. J., 224,
Astrophys. J., 692 , 973.Evans N. J. II, et al. (1994)
Astrophys. J., 424 , 793.Evans N. J. II, et al. (2009)
Astrophys. J. Suppl., 181 , 321.Fedele D., et al. (2007)
Astron. Astrophys., 472 , 207.Feigelson E. D. and Montmerle T. (1999)
Annu. Rev. Astron. As-trophys., 37,
Astrophys. J., 768,
Astrophys. J., 756 , 99.Gammie C. (1996)
Astrophys. J., 457 , 355.Gammie C. (2001)
Astrophys. J., 553 , 174.Giannini T., et al. (2009)
Astrophys. J., 704 , 606.Gibb A. G. (2008) In
Handbook of Star Forming Regions, Vol-ume I: The Northern Sky (ed. B. Reipurth), pp. 693-731. ASPMonograph Publications, San Francisco.Gibb E. L., et al. (2006)
Astrophys. J., 641,
PASP, 99 , 116.Goodson A. P., et al. (1997),
Astrophys. J., 489 , 199.Goto M., et al. (2011)
Astrophys. J., 728 , 5.Graham J. A. and Frogel J. A. (1985)
Astrophys. J., 289,
Astron. Astrophys., 443 ,541.Green J. D., et al. (2006)
Astrophys. J., 648 , 1099.Green J. D., et al. (2011)
Astrophys. J., 731 , L25.Green J. D., et al. (2013a)
Astrophys. J., 764 , 22.Green J. D., et al. (2013b)
Astrophys. J., 772,
Astrophys. J., 434,
Astron. J., 135 , 1421.Grosso N., et al. (2005)
Astron. Astrophys., 438 , 159.Grosso N., et al. (2010)
Astron. Astrophys., 522 , A56.Haas M., et al. (1990)
Astron. Astrophys., 230,
L1.Haisch K. E. Jr., et al. (2004)
Astron. J., 127,
Astrophys. J., 714,
L16.Hamaguchi K., et al. (2012)
Astrophys. J., 741 , 32.Hartigan P. and Kenyon S. J. (2003)
Astrophys. J., 583,
Physics of Star Formation and Early Stel-lar Evolution. NATO Adv. Study Inst. (C. J. Lada and N. D.Kylafis, eds.), pp. 623-648. Kluwer, Dordrecht.Hartmann L. (2008) In
Accretion Processes in Star Formation .Cambridge University Press, Cambridge.Hartmann L. and Calvet N. (1995)
Astron. J., 109 , 1846.Hartmann L. and Kenyon S. J. (1985)
Astrophys. J., 299 , 462.Hartmann L. and Kenyon S. J. (1987a)
Astrophys. J., 322 , 393.Hartmann L. and Kenyon S. J. (1987b)
Astrophys. J., 312 , 243.Hartmann L. and Kenyon S. J. (1996)
Ann. Rev. Astron. Astro-phys., 34 , 207.Hartmann L., et al. (1989)
Astrophys. J., 338,
Protostars and Planets III (E. H.Levy and J. I. Lunine, eds.), pp. 497–520. Univ. of Arizona,Tucson.Hartmann L., et al. (2004)
Astrophys. J., 609 , 906.Henning, Th., et al. (1998)
Astron. Astrophys. 336,
Vistas Astron., 8 , 109.Herbig G. H. (1977)
Astrophys. J., 217 , 693.Herbig G. H. (1989) In
ESO Workshop on Low Mass Star For-mation and Pre-Main Sequence Objects (B. Reipurth, ed.),pp. 233-246. ESO, Garching.Herbig G. H. (2007)
Astron. J., 133 , 2679.Herbig G. H. (2008)
Astron. J., 135 , 637.Herbig G. H. (2009)
Astron. J., 138 , 448.Herbig G. H., et al. (2001)
Publ. Astron. Soc. Pac., 133 , 1547.Herbig G. H., et al. (2003)
Astrophys. J., 595 , 384. Hillenbrand L. (2009), In
IAU Symp. 258, The Age of Stars , (E.Mamajek, et al., eds.), pp. 81-88. Cambridge University Press,Cambridge.Hillenbrand L. A., et al. (2013)
Astron. J., 145 , 59.Hinkley, S., et al. (2013)
Astrophys. J., 763 , L9.Hodapp K. W. (1999)
Astron. J., 118 , 1338.Hodapp K. W., et al. (1996)
Astrophys. J., 468 , 861.Hodapp K. W., et al. (2012)
Astrophys. J., 744 , 56.Hosokawa T., et al. (2011)
Astrophys. J. 738,
Astron. Astrophys., 39,
Astron. J., 134 , 241.Johns-Krull C. M. (2007)
Astrophys. J., 664 , 975.Johnson B. M. and Gammie C. F. (2003)
Astrophys. J., 597 , 131.Johnstone D., et al. (2013)
Astrophys. J., 765,
Astrophys. J., 744 , 118.Kastner J. H., et al. (2004)
Nature, 430 , 429.Kastner J. H., et al. (2006)
Astrophys. J., 648 , 43.Kenyon S. J. (1995a)
Astrophys. Space Sci., 233 , 3.Kenyon S. J. (1995b)
Rev. Mex. Astron. Astrophys. (Conf. Ser.),1 , 237.Kenyon S. J. and Hartmann L. W. (1991)
Astrophys. J., 383 , 664.Kenyon S. J. and Hartmann L. W. (1995)
Astrophys. J. Suppl., 101 ,117.Kenyon S. J., et al. (1988)
Astrophys. J., 325 , 231.Kenyon S. J., et al. (1990)
Astron. J., 99 , 869.Kenyon S. J., et al. (1993)
Astron. J., 105 , 1505.Kenyon S. J., et al. (1994)
Astron. J., 108 , 251.Kenyon S. J., et al. (2000)
Astrophys. J., 531 , 1028.Kim H. J., et al. (2012),
Astrophys. J., 758 , 38.Kley W. and Lin D. N. C. (1999)
Astrophys. J., 518 , 833.K¨ohler R., et al. (2006)
Astron. Astrophys., 458 , 461.K¨onigl A. (1991)
Astrophys. J., 370 , L39.K¨onigl A., et al. (2011)
Mon. Not. R. Astron. Soc., 416 , 757.Koresko C. D., et al. (1991)
Astron. J., 102 , 2073.K´osp´al ´A. (2011)
Astron. Astrophys., 535 , 125.K´osp´al ´A., et al. (2007)
Astron. Astrophys., 470 , 211.K´osp´al ´A., et al. (2008)
Mon. Not. R. Astron. Soc., 383 , 1015.K´osp´al ´A., et al. (2011a)
Astron. Astrophys., 527 , 133.K´osp´al ´A., et al. (2011b)
Astrophys. J., 736 , 72.K´osp´al ´A., et al. (2012)
Astrophys. J. Supp., 201 , 11.K´osp´al ´A., et al. (2013)
Astron. Astrophys., 551 , 62.Kratter K. M., et al. (2008)
Astrophys. J., 681 , 375.Kravtsova A. S., et al. (2007)
Astron. Letters 33 , 755.Kryukova E., et al. (2012)
Astron. J., 144 , 31.Kulkarni A. K. and Romanova M. M. (2008)
Astrophys. J., 386,
Astrophys. J., 733 , L8.Kun M., et al. (2011b)
Mon. Not. R. Astron. Soc., 413 , 2689.Kurosawa R. and Romanova M. M. (2012)
Mon. Not. R. Astron.Soc., 426 , 2901.Kurosawa R. and Romanova M. M. (2013)
Mon. Not. R. Astron.Soc., 431 , 2673.Kurucz R. L., et al. (1974). Smithsonian Institution, Washington.Lee J.-E. (2007)
J. Korean Astron. Soc., 40 , 83.Lehmann T., et al. (1995),
Astron. Astrophys., 300 , L9.Leinert Ch. and Haas M. (1987)
Astron. Astrophys., 182 , L47.Lii P., et al. (2012)
Mon. Not. R. Astron. Soc., 420 , 2020.Lii P. S, et al. (2013)
Mon. Not. R. Astron. Soc. , arXiv:1304.2703.Lin D. N. C., et al. (1985)
Mon. Not. R. Astron. Soc, 212 , 105.Lodato G. and Bertin G. (2001)
Astron. Astrophys., 375 , 455.Lodato G. and Bertin G. (2003)
Astron. Astrophys., 408 , 1015. odato G. and Clarke C. J. (2004) Mon. Not. R. Astron. Soc., 353 ,841.Lodato G. and Rice W. K. M. (2004)
Mon. Not. R. Astron. Soc.,351 , 630.Lodato G. and Rice W. K. M. (2005)
Mon. Not. R. Astron. Soc.,358 , 1489.Lodato G., et al. (2007)
Mon. Not. R. Astron. Soc., 374 , 590.Lombardi M., et al. (2008)
Astron. Astrophys., 480 , 785.Lorenzetti D., et al. (2000),
Astron. Astrophys., 357 , 1035.Lorenzetti D., et al. (2006)
Astron. Astrophys., 453 , 579.Lorenzetti D., et al. (2007)
Astrophys. J., 665 , 1182.Lorenzetti D., et al. (2009)
Astrophys. J., 693 , 1056.Lorenzetti D., et al. (2011)
Astrophys. J., 732 , 69.Lorenzetti D., et al. (2012)
Astrophys. J., 749 , 188.Lovelace R. V. E., et al. (1991)
Astrophys. J., 379 , 696.Lovelace R. V. E., et al. (1999)
Astrophys. J., 514 , 368.Machida M. N., et al. (2011)
Astrophys. J., 729 , 42.Magakian T. Y., et al. (2013)
Mon. Not. R. Astron. Soc., 432,
Astron. J., 139,
Astrophys. J., 507 , L149.Malbet F., et al. (2005)
Astron. Astrophys., 437,
Astrophys. J., 740 , L6.Martin R. G. and Lubow S. H. (2013)
Mon. Not. R. Astron. Soc.,432,
Mon. Not. R. Astron. Soc., 420 , 3139.Martin R. G., et al. (2012b)
Mon. Not. R. Astron. Soc., 423 , 2718.Matt S. and Pudritz R. (2005)
Astrophys. J., 632 , L135.Mayer L., et al. (2005)
Mon. Not. R. Astron. Soc.363 , 641.McGehee P. M., et al. (2004)
Astrophys. J., 616 , 1058.McMuldroch S., et al. (1993)
Astron. J., 106 , 2477.McMuldroch S., et al. (1995)
Astron. J., 110 , 354.Mej´ıa A. C., et al. (2005),
Astrophys. J., 619 , 1098.Melo C. H. F. (2003),
Astron. Astrophys., 410 , 269.Menten K. M., et al. (2007)
Astron. Astrophys., 474 , 515.Millan-Gabet R., et al. (2006)
Astrophys. J., 641 , 547.Miller A. A., et al. (2011),
Astrophys. J., 730 , 80.Moriarty-Schieven G. H., et al. (2008)
Astron. J., 136 , 1658.Mosoni L., et al. (2013)
Astron. Astrophys., 552 , A62.Movsessian T. A., et al. (2003)
Astron. Astrophys., 412 , 147.Movsessian T. A., et al. (2006)
Astron. Astrophys., 455 , 1001.Muzerolle J., et al. (2005)
Astrophys. J., 620 , L107.Nayakshin S. and Lodato G. (2012)
Mon. Not. R. Astron. Soc.,426 , 70.Offner S. S. R. and McKee C. F. (2011),
Astrophys. J., 736 , 53.Offner S. S. R., et al. (2009)
Astrophys. J., 703 , 131.Ojha D., et al. (2005)
Publ. Astron. Soc. Jap., 57,
Mon. Not. R. Astron. Soc., 368,
Publ. Astron. Soc. Jap., 49,
Astrophys. J. 439 , 288.Parsamian E. S. and Mujica R. (2004)
Astrophysics, 47 , 433.Parsamian E. S., et al. (2002)
Astrophysics, 45 , 393.Peneva S. P., et al.. (2010)
Astron. Astrophys. 515 , 24.P´erez L. M., et al. (2010)
Astrophys. J., 724 , 493.Persi P., et al. (2007)
Astron. J., 133,
Astrophys. J., 392,
Astron. J., 136 , 676.Petrov P., et al. (1998)
Astron. Astrophys., 331 , L53.Pfalzner S. (2008)
Astron. Astrophys., 492 , 735.Pfalzner S., et al. (2008)
Astron. Astrophys., 487 , L45.Polomski E. F., et al. (2005)
Astron. J., 129 , 1035.Popham R. (1996)
Astrophys. J., 467,
Astrophys. J., 473,
Mon. Not. R. Astron. Soc., 426 , 3315.Prato L., et al. (2002)
Astrophys. J., 579 , L99.Preibisch T., et al. (2005)
Astrophys. J. Supp., 160 , 401.Prusti T., et al. (1993)
Astron. Astrophys., 279 , 163.Pueyo L., et al. (2012)
Astrophys. J., 757 , 57.Quanz S. P., et al. (2006),
Astrophys. J., 648 , 472.Quanz S. P., et al. (2007a)
Astrophys. J., 656 , 287.Quanz S. P., et al. (2007b)
Astrophys. J., 658 , 487.Quanz S. P., et al. (2007c)
Astrophys. J., 668 , 359.Reipurth B. (1989)
Nature, 340,
IAU Symp. 137, Flare stars in star clusters,associations and the solar vicinity (L. V. Mirzoyan, et al., eds.),pp. 229-251. Kluwer, Dordrecht.Reipurth B. and Aspin C. (1997)
Astron. J., 114 , 2700.Reipurth B. and Aspin C. (2004a)
Astrophys. J., 606 , L119.Reipurth B. and Aspin C. (2004b)
Astrophys. J., 608 , L65.Reipurth B. and Aspin C. (2010) In
Victor Ambartsumian Centen-nial Volume, Evolution of Cosmic Objects through their Phys-ical Activity , (H. Harutyunyan, A. Mickaelian, and Y. Terzian,eds.), pp. 19-38. Gitutyun Publishing House, Yerevan.Reipurth B. and Bally J. (2001)
Annu. Rev. Astron. Astrophys., 39 ,403.Reipurth B., et al. (2002)
Astron. J., 124 , 2194.Reipurth B., et al. (2007a)
Astron. J., 133 , 1000.Reipurth B., et al. (2007b)
Astron. J., 134 , 2272.Reipurth B., et al. (2012)
Astrophys. J., 748 , L5.Rettig T. W., et al. (2005)
Astrophys. J., 626 , 245.Rice W. K. M, et al. (2005)
Mon. Not. R. Astron. Soc., 364 , L56.Rice W. K. M., et al. (2011)
Mon. Not. R. Astron. Soc., 418 , 1356.Romanova M.M. and Kulkarni A.K. (2009)
Mon. Not. R. Astron.Soc., 398,
Astrophys. J., 610 , 920.Romanova M. M., et al. (2005)
Astrophys. J., 635 , L165.Romanova M. M., et al. (2008)
Astrophys. J., 673 , L171.Romanova M. M., et al. (2009)
Mon. Not. R. Astron. Soc., 399 ,1802.Romanova M. M., et al. (2013)
Mon. Not. R. Astron. Soc., 430 ,699.Sandell G. and Aspin C. (1998)
Astron. Astrophys., 333 , 1016.Sandell G. and Weintraub D. A. (2001)
Astrophys. J. Supp., 134 ,115.Scholz A., et al. (2013)
Mon. Not. R. Astron. Soc., 430 , 2910.Sch¨utz O., et al. (2005)
Astron. Astrophys., 431 , 165.Semkov, E. H. and Peneva, S. P. (2012)
Astron. Sp. Sci., 338 , 95.Semkov E. H., et al. (2010)
Astron. Astrophys., 523,
L3.Semkov E. H., et al. (2011)
Bulg. Astron. J., 15 , 49.Semkov E. H, et al. (2012)
Astron. Astrophys., 542 , 43.Semkov, E. H., et al. (2013)
Astron. Astrophys., 556,
A60.Shevchenko V. S.,et al. (1991)
Soviet Ast., 35,
Astron. Astrophys. Suppl., 124,
Astrophys. J., 429 , 781.Sicilia-Aguilar A., et al. (2008)
Astrophys. J., 673 , 382.Sicilia-Aguilar A., et al. (2012)
Astron. Astrophys. 544 , 93.Sipos N., et al. (2009)
Astron. Astrophys., 507 , 881.Siwak M., et al. (2013)
Mon. Not. R. Astron. Soc., 432 , 194.Skinner S. L., et al. (2006)
Astrophys. J., 643 , 995.Skinner S. L., et al. (2009)
Astrophys. J., 696 , 766.Skinner S. L., et al. (2010)
Astrophys. J., 722 , 1654.Spruit H. C. and Taam R. E. (1993)
Astrophys. J., 402 , 593.Stamatellos D. and Whitworth A. P. (2009a)
Mon. Not. R. Astron.Soc., 392 , 413.Stamatellos D. and Whitworth A. P. (2009b)
Mon. Not. R. Astron. oc., 400 , 1563.Stamatellos D., et al. (2011) Astrophys. J., 730 , 32.Stamatellos D., et al. (2012)
Mon. Not. R. Astron. Soc., 427 ,1182.Staude H. J. and Neckel Th. (1991)
Astron. Astrophys., 244,
L13.Staude H. J. and Neckel Th. (1992)
Astrophys. J., 400 , 556.Stecklum B., et al. (2007)
Astron. Astrophys., 463 , 621.Stelzer B., et al. (2009)
Astron. Astrophys., 499 , 529.Strom K. M. and Strom S. E. (1993)
Astrophys. J., 412 , L63.Sunyaev R. A. and Shakura N. I. (1977)
Soviet Astron. Lett., 3 ,138.Szeifert T., et al. (2010)
Astron. Astrophys., 509 , L7.Takami M., et al. (2006)
Astrophys. J., 641 , 357.Tapia M., et al. (2006)
Mon. Not. R. Astron. Soc., 367,
Astrophys. J., 618,
Astrophys. J., 741 , 83.Teets W. K., et al. (2012)
Astrophys. J., 760 , 89.Teodorani M., et al. (1997)
Astron. Astrophys. Suppl., 126 , 91.Toomre A. (1964)
Astrophys. J. , 139, 1217.Tsukagoshi T., et al. (2005)
Publ. Astron. Soc. Jap., 57 , L21.Turner N. J. J., et al. (1997)
Astrophys. J., 480 , 754.Umebayashi T. (1983)
Prog. Theor. Phys., 69,
Astrophys. J., 646 , 304.Vacca W. D., et al. (2004)
Astrophys. J., 609 , L29.van den Ancker M. E., et al. (2004)
Mon. Not. R. Astron. Soc.,349 , 1516.Venkata Raman V., et al. (2013)
Research in Astron. Astrophys.,13,
Astron. Astrophys., 446 , 1021.Visser R. and Bergin E. A. (2012)
Astrophys. J., 754 , 18.Vorobyov E. I. (2009)
Astrophys. J., 704 , 715.Vorobyov E. I. (2010)
Astrophys. J., 729 , 146.Vorobyov E. I. (2013)
Astron. Astrophys., 552 , 129.Vorobyov E. I. and Basu S. (2005)
Astrophys. J., 633 , L137.Vorobyov E. I. and Basu S. (2006)
Astrophys. J., 650 , 956.Vorobyov E. I. and Basu S. (2010)
Astrophys. J., 719 , 1896.Vorobyov E. I., et al. (2013)
Astron. Astrophys., 557,
A35.Walter F. M., et al. (2004)
Astron. J., 128 , 1872.Wang H., et al. (2004)
Astrophys. J., 601 , L83.Welin G. (1976)
Astron. Astrophys., 49,
Astrophys. J., 720 , L119.Xiao L., et al. (2010)
Astron. J., 139 , 1527.Young C. H. and Evans N. J. II (2005)
Astrophys. J., 627 , 293.Zanni C., et al.. (2007)
Astron. Astrophys., 469 , 811-828Zhu Z., et al. (2007)
Astrophys. J., 669 , 483.Zhu Z., et al. (2008)
Astrophys. J., 684 , 1281.Zhu Z., et al. (2009a)
Astrophys. J., 694 , L64.Zhu Z., et al. (2009b)
Astrophys. J., 694 , 1045.Zhu Z., et al. (2009c)
Astrophys. J., 701 , 620.Zhu Z., et al. (2010a)
Astrophys. J., 713 , 1134.Zhu Z., et al. (2010b)
Astrophys. J., 713 , 1143., 1143.