Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FU Orionis

The flaring of FU Orionis (FU Ori) in 1936 (Hoffleit 1939) marked the discovery of a new phase in low-mass star formation. FU Ori increased its brightness by over 5 mag within a year (Wachmann 1954). Herbig (1966) noted that the brightening could not be explained in terms of a nova event or by sudden changes in extinction and that it was most likely due to "some phenomenon of early stellar evolution." Hartmann & Kenyon (1985) argued that the increased luminosity could be explained by the rapid onset of accretion from a rotating accretion disk (i.e., an accretion outburst). In the accretion disk model, the energy released by the accreting gas heats the disk up to several thousand Kelvin, reproducing the increase in luminosity, the observed spectral energy distribution (SED), and peculiar spectral properties, such as the apparent wavelength-dependent stellar spectral type, which can be explained by absorption from different regions of the superheated disk atmosphere (Hartmann & Kenyon 1996; Hartmann et al. 2016).

For decades, FU Ori was the only known star of its class, but the progressive discovery of more outbursting sources has suggested that these are not isolated cases, but a common, yet short-lived, phase in star formation. Newly discovered outbursting stars have shown diverse outburst properties, forcing the definition of subclasses: FUors (named after FU Ori) have high-amplitude (> 5 mag), long-lived (years to decades) optical bursts (Herbig 1966), whereas EXors (misnamed after EX Lupi) have low-amplitude (2–4 mag) and shorter (days to months) periods of enhanced accretion (see, e.g., Audard et al. 2014 for a review). The advent of systematic photometric surveys has increased the discovery rate of new eruptive objects (e.g., Kun et al. 2014). Milky Way surveys indicate that 3%–6% of Class I protostars show eruptive behavior over a 4 yr span and that eruptive events are more common in Class I objects than Class II (Contreras Peña et al. 2017).

As new sources are discovered, the heterogeneous nature of their characteristics presents a significant challenge in terms of classification and understanding. Fischer et al. (2023) discuss the difficulties encountered in distinguishing between various types of outbursts, such as FU Ori–like events and EX Lupi–like events. These challenges arise due to the overlapping characteristics exhibited by these phenomena and the limitations imposed by the available observational data. Recent advancements in time-dependent, multiwavelength, and spectroscopic analyses suggest a continuum of diverse phenomena, rather than distinct and separable classes. This emphasizes the need for comprehensive and multifaceted approaches to understand the complexity of these outburst events fully.

Accretion outbursts are now believed to be central to the formation of low-mass stars. The well-known luminosity problem (Kenyon et al. 1990) arises from the fact that the measured luminosities of low-mass protostars are smaller than expected from the quiescent (nonoutbursting) mean accretion rates (10⁻⁶ M_⊙ yr⁻¹). The problem is mitigated if the accretion rates are variable, with stars accreting an important fraction of their mass during several bursts of enhanced accretion during the Class I phase (Evans et al. 2009; Dunham et al. 2010), while spending most of their time at lower quiescent accretion rates than predicted (10⁻⁸ M_⊙ yr⁻¹). Submillimeter monitoring of star-forming regions indicates that episodic accretion may occur even earlier, at the Class 0 stage (Safron et al. 2015; Johnstone et al. 2018).

The physical mechanisms responsible for the outburst remain a matter of debate. A funneling mechanism, in which matter from an envelope is accumulated in the accretion disk and released episodically, could explain the outbursts (Zhu et al. 2009). Thermal instabilities can also be responsible for the piling up and sudden release of inner disk material (Bell & Lin 1994). Additionally, the formation and fragmentation of spiral waves, driven by gravitational instabilities, have been suggested to trigger these accretion outbursts (see Kratter & Lodato 2016 for a review). Observations of spiral structures due to gravitational instabilities, such as those seen in the massive disk of Elias 2-27 (Pérez et al. 2016) and the recently observed fragmenting spirals around the V960 Mon system after its FUor-like outburst in 2014 (Weber et al. 2023), support this theory.

Despite their significance in our understanding of low-mass star formation, the exact causes of these outbursts remain elusive. The dense star-forming environments of these young circumstellar disks offer another possible explanation. It has been proposed that the sudden outbursts could be linked to interactions within binary systems or encounters with passing stars, which are quite common in these crowded regions (Bonnell & Bastien 1992; Reipurth & Aspin 2004). Additionally, the multiplicity of stars and the substantial cross-sectional area of these disks can foster interactions with passing stars (star–disk flybys; Pfalzner 2008; Cuello et al. 2019) as well as disk–disk interactions, which can excite the disks, turning them unstable (Muñoz et al. 2015).

Various FUor-like protostars have been identified as binary companions, including notable cases such as L1551 IRS5, RNO 1B/C, AR 6A/B, and HBC 494 (Pueyo et al. 2012; Nogueira et al. 2023 and references therein). The archetypal FU Ori system, FU Orionis itself, is a known binary composed of two stars separated by 05, where the northern component (hereafter FU Ori N) is the optically visible star driving the outburst. Wang et al. (2004) discovered the southern binary component, which, according to a spectroscopic analysis by Beck & Aspin (2012), is the more massive of the two (hence FU Ori S is the primary member). Early Atacama Large Millimeter/submillimeter Array (ALMA) Observations of FU Ori at 05 resolution resolved the binary system in continuum and revealed complex kinematical features (Hales et al. 2015). High-resolution ALMA observations of FU Ori at 004 resolution by Pérez et al. (2020) resolved each individual disk, reporting very compact sizes possibly truncated by binary interactions and kinematical signatures of Keplerian rotation around the outbursting source (FU Ori N). These discoveries have contributed to the ongoing debate as to whether the enhanced accretion can be attributed to close companion interactions. Another perspective posits perturbation by substellar companions (Clarke et al. 2005).

In terms of the evolutionary stage, FUor objects seem to resemble Class I protostars more than Class II objects. Many FUors have Class I–type SEDs (e.g., Gramajo et al. 2014) as well as dust silicate absorption features, and therefore they have been associated with Class I protostars (Quanz et al. 2006). Most FUors are surrounded by reflection nebulae, remnants from their parent cores, some of which are seen to brighten up during the outburst (Herbig 1966; Goodrich 1987; Aspin & Reipurth 2003). FU Ori itself, located near the Lynds Bright Nebula 878 (LBN 878), has a fan-shaped reflection nebula that appeared next to the star after the flare, which was not present before (Wachmann 1954). Figure 1 (left) shows large-scale view of the FU Ori environment, as seen in optical wavelengths. The optical composite image is courtesy of astrophotographer Jim Thommes, ¹⁴ also known for the discovery of the Thommes Nebula, a reflection nebula that appeared after the outburst of the FUor-type object V900 Mon (Thommes et al. 2011).

Zoom In Zoom Out Reset image size

The enormous energy released during the outburst can power winds and jets that, in turn, may help disperse the primordial core, a process that will eventually regulate the final stellar mass (Shu et al. 1987). FUors power prominent winds measured via the presence of P Cygni profiles in some specific lines (Connelley & Reipurth 2018), from which the inferred mass-loss rates correspond to 10% of the mass accretion rate, thus several orders of magnitude larger than those of classical T Tauri stars (Calvet 2004). The observed profiles can be well reproduced by the standard magnetocentrifugally driven wind model with the addition of an inner disk wind (Milliner et al. 2019).

Early single-dish searches for molecular outflows in FUor objects detected ¹²CO(3–2) in many targets, except FU Ori (Evans et al. 1994). The advent of ALMA enabled high-resolution interferometric spectral line observations, providing insights into the active circumstellar environments of FUors. These environments are characterized by conspicuous, wide-angle, and slow-moving outflows (Ruíz-Rodríguez et al. 2017a, 2017b; Zurlo et al. 2017).

On the other hand, EXors seem to show little outflowing activity. Outflows have been reported around the ambiguously classified FUor/EXor sources V1647 Ori and GM Cha (Cieza et al. 2018; Principe et al. 2018; Hales et al. 2020), and only tentatively around EX Lup (Hales et al. 2018; Rigliaco et al. 2020). Although the sample size is small, the differences in outflow activity between FUors and EXors suggest that the two types of objects represent different evolutionary stages, with EXors being more evolved than FUors.

There is also observational evidence that most FUor sources are surrounded by large envelopes that are still transferring material onto the disk (Fehér et al. 2017; Kóspál et al. 2017a). It has been proposed that the imbalance between the mass transferred from the envelope to the disk and the accretion of disk mass onto the star may trigger the disk instability responsible for the outburst (Bell & Lin 1994; Kóspál et al. 2017b). Studying the larger-scale structure of eruptive sources is thus crucial for understanding the nature of the FUor/EXor outbursts.

Asymmetric, non-Keplerian gaseous structures are now routinely discovered by ALMA. Kinematic modeling of these structures reveals that they can be interpreted as anisotropic infalls of material from the surrounding medium into the circumstellar disk. These structures end up following elongated patterns, for which they have been named 'accretion streamers' (see Pineda et al. 2023 for a review). Accretion streamers have now been detected around many Class 0 and Class I sources and are relevant for improving our understanding of low-mass star formation. For instance, their discovery directly challenges the traditional model of axisymmetric collapse of protostellar cores (Shu 1977). The role accretion streamers play in episodic accretion processes still needs to be demonstrated; to our knowledge, no accretion streamers have been discovered around eruptive stars yet.

In this work, we present ALMA observations of the surroundings of FU Ori at angular scales ranging from 160 to 25,000 au at 1.3 mm continuum and CO isotopologues. Section 2 provides details of the ALMA observations and data reduction, while Section 3 presents the primary observational findings and their analysis. These results are subsequently discussed in Section 4, and Section 5 offers a summary of the study's findings. Throughout the paper, we assume a distance to FU Ori of 408 ± 3 pc (Gaia DR3; Bailer-Jones et al. 2021).

Observations of FU Ori (project code 2017.1.00015.S, PI: J. Williams) were carried out between 2017 November and 2018 August using the ALMA 12-m, 7-m, and Total Power (TP) Arrays in order to obtain spatial coverage from 04 to up to a map size of 1' (160–25,000 au). The observations, calibrators used, weather conditions, and array characteristics are shown in Table 1. The full map required a 27-pointing mosaic for the 12-m Array and a seven-pointing mosaic for the 7-m Array. The total antenna numbers for each of the 12-m, 7-m, and TP Arrays were 45, 11, and 3, respectively.

The correlators were configured in frequency division mode (FDM) to observe ¹²CO(2–1), ¹³CO(2–1), and C¹⁸O(2–1) in a single Band 6 spectral setting, providing spectral resolutions of 141 kHz for ¹²CO(2–1) and 122.070 kHz for ¹³CO(2–1) and C¹⁸O(2–1), corresponding to velocity resolutions of 0.184, 0.166 and 0.167 km s⁻¹, respectively. The bandwidth coverage was 142.4, 79.7, and 80.0 km s⁻¹ for ¹²CO(2–1), ¹³CO(2–1), and C¹⁸O(2–1). A 1.875 GHz spectral window centered at 232.470 GHz was positioned to detect the continuum dust emission.

All data were calibrated by the ALMA staff using the ALMA Science Pipeline (version 40896 of Pipeline-CASA51-P2-B) in CASA 5.1.1 ¹⁵ (McMullin et al. 2007; CASA Team et al. 2022). The calibration process includes offline water vapor radiometer calibration and system temperature correction, as well as bandpass, phase, and amplitude calibrations. Online flagging and nominal flagging, such as shadowed antennas and band edges, were applied during calibration. Continuum subtraction in the visibility domain was performed on the interferometric data prior to imaging.

The imaging of the continuum and molecular emission lines was performed using the tCLEAN task in CASA 6.1. The TP observations were converted into visibilities using the task TP2VIS (Koda et al. 2019) and then input into tCLEAN along with the interferometric data.

The 12m array data were self-calibrated using the standalone version of the automated self-calibration module (J. Tobin et al., in preparation). The self-calibration process was performed separately for each observation, with one iteration of phase-only self-calibration. The improvements in the signal-to-noise ratio (S/N) after self-calibration ranged between 10% and 20%. The self-calibration tables derived from the continuum data were applied to the FDM spectral windows before imaging the CO lines. The continuum and spectral line data were imaged using the tCLEAN task interactively in CASA (using a Hogbom deconvolver, creating manually defined CLEAN masks, and using natural weighting to maximize sensitivity). Continuum subtraction was performed before imaging each molecular line using the task UVCONTSUB. During the cleaning of the spectral lines, the spectral axis was binned on the fly to 0.25 km s⁻¹ to increase the S/N. The resulting spatial resolution for the combined 12m+7m+TP data set corresponds to a synthesized beam size of 04×03 (PA = 416).

Integrated intensity (moment 0) and intensity-weighted mean velocity (moment 1) maps for the spectral line data were produced with the CASA task IMMOMENTS. The moment 0 images were produced by integrating the signal in channels with emission above 3σ. There is a strong cloud contamination close to the system's velocity (11.75 km s⁻¹; Hales et al. 2015), particularly strong in the 7m+TP data; thus, for display purposes, we created separate moment 0 maps for the redshifted and blueshifted portions of the observed large-scale emission.

We detect ¹²CO, ¹³CO, and C¹⁸O on a variety of angular scales. The source large-scale structure mapped by the 7-m+TP Arrays dominates the final image, overwhelming the structures traced by the 12-m Array. For this reason, we also produced continuum and spectral line cubes using the 12m data only. This approach allows for a more detailed examination of the specific characteristics and features contained in each data set.

Figure 1 (inset) shows the integrated intensity (moment 0) map for ¹²CO, and Figure 2 shows the integrated intensity maps for all three CO isotopologues at the spatial scales traced by the 7-m+TP Arrays. The angular resolution of the combined 12+7m+TP image is 04×03. However, the maps in Figure 2 have been smoothed to 25 to enhance the sensitivity to large-scale structures (the nominal resolution of only the 7m data is 49, whereas the angular resolution of only the 12m data is 02). Figure 3 displays the ¹²CO and ¹³CO moment 1 maps obtained by imaging the 12m array data only. C¹⁸O was not detected in the 12m array data.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.1. Wide, Slow-moving Outflows

3.2. Streamer in FU Ori

3.3. Outflow Properties

Swept-up cavities in the envelopes of young stellar objects play a significant role in understanding the complex processes occurring during the early stages of their formation (Arce et al. 2007 and references therein). These cavities are created by the powerful outflows and jets emitted from protostars, which interact with their surrounding molecular clouds. As the outflows propagate through the dense material, they sweep away and evacuate cavities, shaping the surrounding environment. These cavities serve as valuable indicators of the energetic feedback mechanisms associated with star formation. By studying the size, shape, and kinematics of these swept-out cavities, researchers can gain insights into the dynamics of protostellar outflows, the interactions between the outflows and their surroundings, and the impacts of these processes on the overall evolution of young stellar objects.

Assuming both the ¹²CO and ¹³CO lines trace the bipolar cavity, we use the ¹²CO and ¹³CO emissions to derive estimates of the mass of the outflow and its kinematic properties in the traditional manner (e.g., Cabrit & Bertout 1990; Dunham et al. 2014). Following the process described in Ruíz-Rodríguez et al. (2017a), we estimate the outflow quantities from the blue- and redshifted emissions separately. The optically thin tracer ¹³CO is used to correct for the optical depth effects in ¹²CO by computing the ratio of the brightness temperatures of each line from all the channels with detections above 5σ. To apply the correction factor to data with only ¹²CO detections, we extrapolate values from a parabola fitted to the weighted mean values. Figure 4 shows the best-fit parabola (solid green line). The derived outflow properties are shown in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Mass, Momentum, Luminosity, and Kinetic Energy of the Outflow

	20 (K)	50 (K)
Blue Lobe
Mass (M⊙)	1.37	2.08
Mass loss (M⊙ yr−1)	3.6 × 10−4	5.5 × 10−4
Momentum (M⊙ km s−1)	1.16	1.76
Energy (erg)	1.02 × 1043	1.55 × 1043
Luminosity (L⊙)	0.02	0.03
Red Lobe
Mass (M⊙)	0.019	0.029
Mass loss (M⊙ yr−1)	5.1 ×10−6	7.7 × 10−6
Momentum (M⊙ km s−1)	0.014	0.022
Energy (erg)	1.08 × 1041	1.64 × 1041
Luminosity (L⊙)	2.3 × 10−4	3.5 ×10−4

Download table as:  ASCIITypeset image

3.4. Streamer Model

To ascertain the infalling nature of the observed streamer, we fitted the ¹²CO(2–1) streamer emission (see Figure 3) with the analytical equations of infalling trajectories given by Mendoza et al. (2009). For this fitting, we use the Trajectory of Infalling Particles in Streamers around Young stars (TIPSY) code (Gupta et al. 2024) which simultaneously fits the morphology and velocity profile of the streamer observations. TIPSY requires prior estimations of the mass of the central object and the systemic line-of-sight velocity of the center of mass, which were taken to be 1.8 M_⊙ and 11.75 km s⁻¹, respectively, since the masses of FU Ori N and FU Ori S are 0.6 and 1.2 M_⊙, respectively. (Beck & Aspin 2012; Pérez et al. 2020). Pérez et al. (2020) inferred the disk dust masses for FU Ori N and FU Ori S to be 22 M_⊕ and 8 M_⊕, respectively. Assuming a typical gas-to-dust ratio of 100, the total masses of the disks are expected to be ∼7 M_Jupiter (or ∼6.6 × 10⁻³ M_⊙) and ∼3 M_Jupiter (or ∼2.8 × 10⁻³ M_⊙), respectively, which are negligible for the streamer model calculations.

The fitting results suggest that the morphology and the velocity profile of the observed streamer emission can be well represented as a trail of infalling gas, as shown in Figure 5. Using the distribution of trajectories that can fit the observed streamer, the infalling timescale of the streamer was estimated to be 8537 ± 3317 yr. Here, the infalling timescale is defined as the time taken for the infalling material to reach the closest point to the protostars on their trajectory, starting from the farthest point from the observed streamer. The uncertainties were computed as the standard deviation of the infalling timescales, as estimated for the trajectories that could fit the data reasonably well (see Section 2.3. in Gupta et al. 2024). The uncertainty on our trajectory solution is mainly due to the lack of curvature in the projected streamer morphology, which is useful in constraining gas velocities along the plane of the sky. Note that the theoretical trajectories used for the fitting assume that the only force acting on the particle is gravity from a point source. Although FU Ori is a binary system, this assumption is still valid at the length scale of the streamer, which is roughly six times the binary separation.

Zoom In Zoom Out Reset image size

The streamer was also observed in ¹²CO(3–2) emission (Hales et al. 2015), which allows us to simultaneously estimate the temperature and column density of the gas in the streamer. We estimate the brightness temperature of the infalling gas to be 11.3 K and 7.1 K in the ¹²CO(2–1) and (3–2) emissions, respectively. Using the radiative transfer code RADEX, brightness temperatures were found to correspond to the gas temperature of ∼25 K and gas column density of ∼2.5 × 10¹⁶ cm⁻². This is also in agreement with the nondetection of the streamer in the ¹³CO(2–1) emission. Assuming a typical CO to H₂ ratio of 10⁻⁴, the mass of gas in the streamer is then estimated to be ∼5.6 × 10⁻⁷ M_⊙. Although we use the same boundaries to select streamer emission in the 2–1 and 3–2 emission maps, the resulting ratios may be affected by the different angular scales traced by the two data sets (which differ by a factor of 2 in terms of angular resolution and maximum recoverable scale), thus adding extra uncertainty to the derived quantities.

Although these estimates do not include parts of streamers beyond the primary beams of the interferometric observations, they can still be used to estimate mass infall rates as ${\dot{M}}_{\inf }={M}_{{streamer}}/{T}_{\inf }$, where M_streamer is the observed streamer mass and T_inf is the infall time for the observed streamer. We find the streamer to be feeding the FU Ori system at a rate of ∼6.6 × 10⁻¹¹ M_⊙ yr⁻¹. Note that this value should be treated as an order-of-magnitude estimate, because of uncertainties on the streamer mass. A more reliable mass infall rate estimation will require the modeling of the total reservoir of bound gas and, ideally, in a more optically thin line emission like ¹³CO(2–1), which was not detected in our streamer observations.

4.1. Molecular Outflow

4.2. Accretion Streamer

The current study presents ALMA observations of the FU Ori outbursting system covering spatial scales from 160 to 25,000 au.

The remarkable sensitivity to a wide range of spatial scales unveils a wide, slow-moving molecular outflow. To obtain the first determination of the outflow properties, including mass, momentum, and kinetic energy, we have employed ¹³CO(2–1) data to correct for optical depth effects.

A thorough analysis of the higher-resolution interferometric data reveals intricate gas kinematics, likely resulting from a confluence of ongoing physical processes, including accretion, rotation, ejection, and the spreading of material due to binary interaction

The new observations corroborate the existence of an elongated ¹²CO feature that can be modeled consistently as an accretion streamer. The streamer analysis suggests that the stream cannot sustain the enhanced accretion rate. Nevertheless, the observed streamer may be feeding the disk, helping to produce conditions of instability that can give rise to the outburst. Follow-up hydrodynamical modeling is required to ascertain the plausibility of this scenario. Alternatively, the observed streamer might not be directly tied to the current outburst, but instead be a leftover from a prior, more massive streamer that previously contributed to the disk mass, rendering the disk unstable and triggering the FU Ori outburst. These results demonstrate the value of multiscale interferometric observations to enhance our understanding of the FU Ori outbursting system and provide new insights into the complex interplay of physical mechanisms governing the behavior of FUor-type and the many other kinds of outbursting stars.

We express our gratitude to Jim Thommes for generously providing the optical image of FU Orionis. This paper makes use of the following ALMA data: ADS/JAO.ALMA#2017.1.00015.S. ALMA is a partnership of ESO (representing its member states), NSF (USA), and NINS (Japan), together with NRC (Canada) and NSC and ASIAA (Taiwan), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO, and NAOJ. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under a cooperative agreement by Associated Universities, Inc. This work acknowledges support from ANID—Millennium Science Initiative Program—Center Code NCN2021_080. S.P. acknowledges support from FONDECYT grant 1231663. A.S.M. acknowledges support from ANID / Fondo 2022 ALMA / 31220025. P.W. acknowledges support from FONDECYT grant 3220399. J.P.W. acknowledges support from NSF grant AST-2107841.

Software: Common Astronomy Software Applications (McMullin et al. 2007; CASA Team et al. 2022), TIPSY (Gupta et al. 2024), and TP2VIS (Koda et al. 2019).

Figures 6, 7, and 8 show the 12m+7m+TP array channel maps for ¹²CO(2–1), ¹³CO(2–1), and C¹⁸O(2–1), respectively. At these spatial scales, the emission is dominated by the large-scale outflows traced by the 7m+TP array data. Figures 9 and 10 show the channel map data from the 12-m Array only for 12CO(2-1) and 13CO(2-1), respectively (C18O(2-1) was not detected).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size