© The Authors 2023

1 Introduction

Similar to other accretion processes in the universe, such as on black holes at the centers of radio galaxies, the formation of stars and planetary systems is always accompanied by accretion disks and outflows. The theory of magnetohydrodynamic (MHD) winds launched from magnetized accretion disks was first outlined by Blandford & Payne (1982), and applied to circumstellar disks around young stars by Pudritz & Norman (1983). These winds are predicted to be launched from most of the disk surface, efficiently carry away angular momentum from the disk, and should therefore rotate. They enable accretion from the disk onto the forming star at a rate of Ṁ_of /Ṁ_* ≈ 0.1… 0.2, where Ṁ_of is the mass outflow rate, and Ṁ_* is the mass accretion rate onto the star (e.g., Pudritz & Norman 1986; Watson et al. 2016; Lee 2020). An alternative mechanism was proposed by Shu et al. (1994), in which high-velocity outflows are launched from the corotation radius of the star and the disk, where the star’s magnetosphere interacts directly with the inner edge of the disk, the so-called X-point (see also Shu et al. 2000). Similar to the disk wind models, this mechanism also predicts that jets and outflows carry away angular momentum and should thus rotate.

While rotation signatures in protoplanetary disks have been observed since the advent of radio (millimeter) interferometers (Sargent & Beckwith 1987), and since then they have been observed and spatially resolved more routinely (e.g., Simon et al. 2000; Launhardt & Sargent 2001; Qi et al. 2003; Panic et al. 2009; Öberg et al. 2010; Huang et al. 2021), rotation signatures in outflows seem to be notoriously more difficult to detect. Transverse velocity gradients have been observed across a number of protostellar high-velocity jets using high-resolution NASA Hubble Space Telescope (hSt) spectra (e.g., Bacciotti et al. 2002, 2003; Coffey et al. 2004, 2007; Woitas et al. 2005) and more recently by, for example, Lee et al. (2017). Rotation signatures have now also been found in a number of slow molecular winds and outflows (e.g., Launhardt et al. 2009; Lee et al. 2009; Klaassen et al. 2013; Chen et al. 2016; Bjerkeli et al. 2016; Hirota et al. 2017). Recent reviews on the theory and observations of protostellar jets, outflows, as well as disk winds can be found in Pudritz & Ray (2019), Lee (2020), and Pascucci et al. (2023).

In this paper, we present new data and analyze the physical properties of the disk and outflow related to the young T Tauri star (TTS) embedded in CB 26. CB 26 (L 1439) is a small cometary-shaped, double-core Bok globule located ≈10° north of the Taurus-Auriga dark cloud, at a distance of 140 ± 20 pc¹ (Launhardt et al. 2010, 2013; Loinard et al. 2011). While the eastern sub-core is starless, a second dense core with signatures of star formation is located at the south west rim of the globule (Stutz et al. 2009; Launhardt et al. 2013). OVRO observations of the millimeter dust continuum emission and of the¹³ CO (1–0) line have revealed a nearly edge-on circumstellar disk of radius ≈200au with Keplerian rotation, surrounding a very young (≤1Myr) low-mass (≈0.5 M_⊙) TTS (Launhardt & Sargent 2001). It is associated with a small bipolar near-infrared (NIR) nebula that is bisected by a dark extinction lane at the position and orientation of the edge-on disk (Stecklum et al. 2004). This disk and its associated NIR nebula are referred to as CB 26 - YSO 1² in the following. The source is surrounded by an optically thin asymmetric envelope with a well-ordered magnetic field directed along PA ≈ 25° (Henning et al. 2001). Furthermore, a Herbig-Haro (HH) object was identified by Hα and S[II] narrow-band imaging, 6.’15 north west of CB 26 at PA ≈ 145° (HH494; Stecklum et al. 2004). Figure 1 shows an overview of CB 26 and its immediate surroundings.

Using the PdBI, we have detected a small (2 × 2000 AU) collimated bipolar molecular outflow from this source, which shows peculiar kinematic signatures. We used an empirical steady-state outflow model combined with 2D line radiative transfer calculations and χ² -minimization to derive a best-fit model and constrain parameters of the outflow (Launhardt et al. 2009). We could show that the data are best reproduced by an outflow that is rotating about its polar axis. This hypothesis was supported by the fact that disk and outflow are corotating, fit together energetically, and by the presence of HH 494, which is located ≈0.25 pc away from CB 26 but is exactly aligned with the outflow axis (Stecklum et al. 2004). Although we could not ultimately exclude alternative scenarios such as jet precession or two misaligned jets from a hypothetical embedded binary system as possible origins of the observed peculiar velocity field, CB 26 was at this time the most promising source in which to study the dispersion of disk angular momentum by a rotating molecular outflow. More recently, López-Vázquez et al. (2022) have shown that the rotating outflow can indeed be well-explained by a disk wind.

To verify the hypotheses of outflow rotation and disk wind origin in CB 26 against the possible alternative scenarios mentioned above and in Launhardt et al. (2009), we obtained higherresolution ¹²CO(2–1) data, which we present and analyze in this paper. Since the CO emission from the disk atmosphere and the outflow are not separated spatially at the origin of the outflow, we establish a physical disk model based on radiative transfer models of the dust continuum emission, use the results to model the CO emission from the disk, and subtract it from the data.

This paper is structured as follows: in Sect. 2 we describe the observations and data reduction. The direct results on the disk and outflow are presented in Sect. 3. In Sect. 4, we present the disk model and analyze the disk-subtracted CO emission from the outflow. These results, their uncertainties, and the physical implications are discussed in Sect. 5. Finally, Sect. 6 summarizes the paper.

2 Observations and data reduction

CB 26 - YSO 1 has been observed at many wavelengths, ranging from 0.9 µm to 6.4 cm. The thermal dust continuum observations and data reduction have been described in detail and the photometry over the entire spectral range is listed in Zhang et al. (2021). Here we only describe the molecular line observations and the optical and NIR observations used in this paper.

2.1 OVRO millimeter observations

CB 26 was observed with the Owens Valley Radio Observatory (OVRO) between January 2000 and December 2001. Four configurations of the six 10.4 m antennas provided baselines in the range 6-180 kλ at 2.7mm (110GHz). Average SSB system temperatures of the SIS receivers were 300–400 K. The digital correlator was centered on the ¹³CO(1–0) line at 110.2GHz, adopting the systemic velocity³ of CB26, υ_LSR = 5.95 km s⁻¹. Spectral resolution and bandwidth were 0.15 and 4.8 km s⁻¹, respectively. The OVRO observations and data are described in more detail in Launhardt & Sargent (2001). The data presented here include additional ¹³CO(1–0) observations conducted in 2001. The OVRO raw data were calibrated and edited using the MMA software package (Scoville et al. 1993). After mapping and checking the results for all individual data sets, the calibrated continuum uv tables were combined and imaged together with the respective PdBI uv tables as described below. The ¹³CO line data were also imaged with GILDAS in the same way as the PdBI line data.

Fig. 1 Overview of the CB 26 region. The left panel shows a wide-field optical “true-color” image, which is based on Hα (blue), [SII] (green), and I-band (red) images (Stecklum et al. 2004). The globules CB 25 and CB 26 as well as Herbig-Haro object HH494 are marked in the image. The zoomed-in right panel shows a NIR J-band image of the bipolar reflection nebula (color, 07 6 resolution), overlaid with contours of the 1.3 mm dust continuum emission from the disk (white contours at 3, 7.5, and 15 mJy beam−1). Green contours show the integrated 12CO(2–1) emission (0.5–12 km s−1) from the bipolar molecular outflow as presented in Launhardt et al. (2009). The red and blue arrows indicate the large-scale outflow orientation. Beam sizes are shown in respective colors in the lower right corner. The white dashed square marks the image section shown in Fig. 2. The reference position is 04h59m50.74s, 52°Q4′43.80″ (J2000).

2.2 IRAM PdBI and 30 m single-dish millimeter observations

Observations of HCO⁺ (1–0) and ¹²CO(2–1) in CB26 were carried out with the IRAM PdBI in November 2005 (D configuration with 5 antennas) and December 2005 (C configuration with 6 antennas; project PD0D). Two receivers were used simultaneously and tuned single side-band (SSB) to the HCO⁺ (1–0) line at 89.188526 GHz, and the ¹²CO(2–1) line at 230.537984GHz, respectively, adopting the systemic velocity of CB 26. Further observations of H¹³CO⁺ (1–0) where carried out in November 2008 (C configuration) and in March 2009 (D configuration; project SC1C), with 6 antennas and using the new 3 mm receivers. Additional higher-resolution observations of ¹²CO(2–1) at 230.5 GHz were carried out in January 2009 (B configuration) and in February 2009 (A configuration with 6 antennas; project S078). Several nearby phase calibrators were observed during all tracks to determine the time-dependent complex antenna gains. The correlator bandpasses were calibrated on 3C454.3 and 3C273, and the absolute flux density scale was derived from observations of MWC 349. The flux calibration uncertainty is estimated to be ≤20% at both wavelengths. The phase center was at α₂₀₀₀ = 04^h59^m50.74^s, δ₂₀₀₀ = 52°04′43.80″. Observing parameters, including beam sizes, channel spacings, etc. are summarized in Table 1. At a distance of 140 pc, the ¹²CO (2–1) mean synthesized FWHM beam size of 0/5 corresponds to a linear resolution of 70 au.

Short-spacing data (maps) for all three lines were obtained in September 2006 with the IRAM 30 m telescope at Pico Veleta in Spain. They were combined with the interferometric uv data using the short-spacings processing tool of the mapping⁴ software to help the imaging and deconvolution. The continuum emission was derived from the line-free channels in the inter-ferometric data and subtracted before combining the uv tables, using the Gildas task uv_subtract. Imaging and deconvolution of the line data was done with natural uv-weighting and the Hogbom algorithm (Högbom 1974), resulting in the synthesized beam sizes and 1σ noise levels in the maps (measured in line-free channels) listed in Table 1. For the ¹²CO (2–1) data, we also produced an additional smoothed version at ≈1.″0 resolution by applying uv tapering with radius 200 m, as well as a version with robust uv weighting resulting in an effective angular resolution of ≈0.″45.

2.3 SMA millimeter observations

Observations with the Submillimeter Array⁵ (SMA; Ho et al. 2004) were made on 2006 December 6 (extended configuration) and December 31 (compact configuration), covering the frequency ranges around 267 and 277 GHz in the lower and upper sidebands, respectively. This setup includes the lines of HCO⁺(3–2) (267.558GHz) and HCN(3–2) (265.886GHz) which were observed with a channel spacing equivalent to 0.23 km s⁻¹. Typical system temperatures were 350-500 K. Only the HCO⁺(3–2) data are used for this paper. The quasar 3C279 was used for bandpass calibration, and the quasars B0355+508 and 3C111 for gain calibration. Uranus was used for absolute flux calibration, which is accurate to ≈20%. The data were calibrated and imaged using the Miriad toolbox (Sault et al. 1995). The 1.1 mm continuum map was constructed using line-free channels in both sidebands. Observing details are summarized in Table 1. In addition, Table 2 summarizes the purpose and use of all millimeter molecular lines used in this paper.

2.4 Optical/NIR observations observations

Optical narrow-band images in the Hα (656 nm) and S[II] (674 nm) filters, as well as in the R-band (641 nm) were taken with the 2048 × 2048 CCD prime focus camera (pixel scale 1.″23) of the 2m Alfred Jensch Telescope⁶ at the Thuringian state observatory in Tautenburg/Germany on 2013, August 2. Two exposures were obtained with each filter with integration times of 1200 s for the narrow-band filters and 180 s for R-band. After the standard image processing, the world coordinate system (WCS) was established using the SCAMP astrometry program (Bertin 2006), which yielded an astrometric RMS of 0.″15. The new images complement the earlier ones taken in 2001 with the same setup (Stecklum et al. 2004), thus providing a temporal baseline of about 12 yr for the proper motion analysis of HH 494 (Appendix C). For compiling the spectral energy distribution (SED; Sect. 3.2), we also use photometry from the Spitzer (Werner et al. 2004), IRAC (Fazio et al. 2004) and MIPS (Rieke et al. 2004) instruments, as well as from the ALLWISE (Cutri et al. 2013) and unWISE (Schlafly et al. 2019) catalogs.

3 Results

3.1 Overview

Figure 1 shows an overview of the CB 26 region, with CB 26 - YSO 1 located at the south eastern tip of the double-core globule CB 26 and the Herbig-Haro object HH494 located at 6.′15 toward the north west (cf. Stecklum et al. 2004). The optical “true-color” image (left panel) is based on wide-field Hα (blue), [SII] (green), and I-band (red) images obtained with the Schmidt CCD camera of the 2-m Alfred Jensch Telescope at the TLS Tautenburg (Stecklum et al. 2004).

CB 26 - YSO 1 is resolved into an edge-on circumstellar disk (Launhardt & Sargent 2001), a bipolar NIR reflection nebula with a dark lane at the position of the disk (Stecklum et al. 2004), and a collimated bipolar molecular outflow that was reported to show kinematic signatures of rotation around its axis with the same orientation as the disk (Launhardt et al. 2009). The deep J-band image of the reflection nebula shown in the right panel of Fig. 1 also shows a faint extended lobe toward the north west, but not toward the south east. Since the north western lobe of the bipolar CO outflow is blue-shifted, and thus slightly directed toward us, this faint extended NIR emission likely marks the region where the north western outflow lobe is breaking out of the dense globule core.

Figure 2 shows the central part of the NIR reflection nebula, again with the 1.3 mm thermal dust continuum emission overlaid on the central dark lane, but now with the integrated CO emission of the higher-resolution CO(2–1) data, which we present and analyze in this paper. The high-resolution CO(2–1) data show an X-shaped morphology at the location of the disk and the origin of the outflow. The kinematic structure indicates the same corotation of disk and outflow as concluded from the lower-resolution data presented in Launhardt et al. (2009), although disk and outflow emission cannot be separated unambiguously close to the disk based on the data only. To separate their contributions to the total emission and enable us to reveal the morphology and kinematics of the outflow as close as possible to its origin, we use the dust continuum emission (Sect. 3.2) to establish a physical model of the disk (Sect. 4.1), which we then use to model and subtract the CO emission from the disk.

Figure 3 shows an overview of the molecular line data used in this paper. The top row shows the spectra of ¹²CO(2–1), ¹³CO(1–0), HCO⁺(1–0), and HCO⁺(3–2) at the position of the disk center (integrated over about 1.5 beams, see Table1). The middle row shows the respective integrated intensity maps with contours of the 1.3 mm dust continuum emission from the disk overlaid. The bottom row shows the 1st moment maps of all four lines. The corresponding channel maps are shown in Figs. A.1 and A.4–A.6. The H¹³CO⁺(1–0) line was only marginally detected when integrating over the entire primary beam area; a channel map is therefore not shown here.

The integrated intensity maps of ¹³CO(1–0) and HCO⁺(3–2) show that we have mostly detected and spatially resolved the emission from the disk (Fig. 3). The HCO⁺(1–0) map is also dominated by emission from the position of the disk, but the disk is not resolved spatially and some emission from the north west lobe of the outflow (and the envelope) is also recovered. In contrast, the ¹²CO(2–1) map shows a very different morphology. It resembles an “X” or butterfly, centered on the disk, but with no emission detected from about the outer half of the disk (midplane). In Sects. 3.3 and 4, we discuss and model this peculiar morphology in terms of a flared disk with a resolved-out (and possibly self-absorbing) envelope and a collimated disk wind seen edge-on. The 1 st moment maps, also shown in Fig. 3 (bottom panels), depict clearly the Keplerian rotation of both the disk and the molecular disk wind, the latter one in ¹²CO(2–1) only. Since ¹²CO(2–1) is the only line that traces the outflow, and the main purpose of this study is to reveal the nature and morphology of the outflow, we analyze quantitatively only the ¹²CO(2–1) data, and use the other lines only for qualitative arguments on the disk rotation.

The position angle (P.A.) of the disk and outflow axis was derived from the three continuum images (Zhang et al. 2021) and the low-resolution ¹²CO(2–1) map (Sect. 3.3) to be PA= 148 ± 1° (E of N). This value agrees remarkably well with the relative PA = 147.5 ± 0.5° (with regard to CB26-YSO 1) of the Herbig-Haro object HH 494 at a projected separation of 6.15′ (corresponding to ≈5 × 10⁴ au or 0.25 pc at a distance of 140 pc) reported by Stecklum et al. (2004, see also Fig. 1 and Appendix C).

In Sect. 4, we analyze, model, and discuss these data in terms of morphology and kinematics of both the disk and the outflow, and their relation. For this purpose, we rotate all maps counterclockwise by 32° (such that PA′ = 180°) in order to align the disk and outflow axes with the z-axis. We furthermore subtract the systemic velocity υ₀ = 5.95 km s⁻¹ (Sect. 3.3) from the line spectra and refer in the following to ∆υ = υ_LSR – υ₀.

Fig. 2 NIR J-band image of the bipolar nebula (see Sect. 2.4) overlaid with contours (3, 8, 13, and 18 mJy beam−1) of the 1.3 mm dust continuum emission from the disk (Zhang et al. 2021). Overlaid as blue and red contours, representing the rotation-related blue- and red-shifted sides (cf. Fig. 3), is the integrated 12CO(2–1) emission (0.5–12 km s−1) from the bipolar molecular outflow as seen in the new, higher-resolution CO(2–1) data, which we present and analyze in this paper. The 1.3 mm continuum and 12CO(2–1) synthesized beam sizes as well as the angular and linear scales are indicated in the lower right corner (cf. Fig. 1).

3.2 Dust continuum emission

Early single-dish observations of CB 26 revealed a compact and basically unresolved 1.3 mm dust continuum source with a total flux density of (Launhardt & Henning 1997; Launhardt et al. 2010). Subsequent interferometric observations showed that ≈50% of this emission originates from an edge-on protoplanetary disk with an outer radius of r_out = 172 ± 22 au, an inner hole of size au, and a total (dust+gas) mass of M_disk ≈ 0.1 M_⊙ (Launhardt & Sargent 2001; Zhang et al. 2021). The other 50% of the 1.3 mm dust continuum emission was attributed to a remnant envelope (or remnant dense globule core) of poorly constrained size (≈ 10⁴ au) and total mass of Menv ≈ 0.1–0.2M_⊙ (Launhardt et al. 2010, 2013; Zhang et al. 2021). Figure 2 shows our highest-resolution dust continuum image of the CB 26 disk, obtained by combining interferometric 230 GHz continuum data from PdBI and OVRO (Zhang et al. 2021), overlaid on the NIR J-band image of the bipolar nebula.

The high-resolution dust continuum images of the CB 26 disk at λ = 1.1 mm (v = 268 GHz), 1.3 mm (230 GHz), and 2.9 mm (102 GHz), together with the SED (Fig. 4) were already used by Sauter et al. (2009), Akimkin et al. (2012), and Zhang et al. (2021) to derive a physical model of the disk by means of continuum radiative transfer modeling. All three studies, albeit using slightly different approaches and slightly different versions of the data, derive very similar basic parameters for the CB 26 disk, which we summarize in Table3. The values for may look very different at first glance, but their error bars fully overlap with all three estimates predicting an upper limit of ≈50au.

In Sect. 4, we use a physical model of the disk with these parameters, which are summarized in Table 4, to derive a model of the ¹²CO(2–1) emission from the disk with the two-fold aim of (i) further constraining certain parameters of the disk (e.g., its inclination) and the central star (e.g., its kinematic mass), and (ii) subtracting this model from the CO data in order to better reveal the morphology and kinematics of the disk wind at its launch region on the disk.

Fig. 3 Line spectra of CB 26, integrated over ≈1.5 beams around the disk center (top row), integrated intensity maps (middle row), and mean velocity fields (1st moment maps, bottom row). From left to right: 12CO (2–1) (0.5–4.8 and 7.0–12.0 km s−1), 11 CO (1–0) (3.4–4.9 and 6.2–7.7 km s−1), HCO+ (3–2) (1–10 km s−1), and HCO+ (1–0) and H13 CO+ (1–0) in blue (2.0–5.0 and 6.3–9.0 km s−1). Dashed vertical lines in the spectra (top) indicate the velocity range within which the spectra are most strongly affected by resolved-out emission and self-absorption from the extended envelope (see Figs. A.1–A.5). Total intensity contours in the maps start at 3σ. Overlaid as red contours in the middle row at 2.5, 7.5, and 15 mJy beam−1 is the 230 GHz dust continuum emission from the disk. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corners (dark gray: line, light gray: continuum). The bottom panels show the respective 1st moment maps with contours of the total intensity overlaid. A black triangle marks the location of the central star (and center of the disk). We note that, except for HCO+ (3–2), the envelope-dominated central velocity channels were masked out before the moment maps where generated.

3.3 ¹²CO(2–1)

The ¹²CO(2–1) channel maps are shown in Fig. A.1 and the spectrum at the position of the disk center is shown in Fig. 3. CO emission is detected in the entire velocity range ∆υ ≈ −5.2 to +5.9 km s⁻¹, translating into a Keplerian radius range of r < –18 au and >+14au, respectively. However, we do not infer the size of a possible “CO hole” directly from this lack of emission at larger relative velocities, but refer to the modeling of the CO emission in Sect. 4.1. The outer velocity channels in the range |∆υ| ≈ 4… 6 km s⁻¹ show relatively compact emission, slightly shifted toward the north west and south east, respectively, with regard to the center of the disk (continuum image). This emission is likely to represent the inner, fast rotating part of the Keplerian disk (cf. Fig. A.2). At lower relative velocities, in the range |∆υ| ≈ 2.4… 3.4 km s⁻¹, narrow streaks extending perpendicular to the plane of the disk become evident. This is likely to be the signature of a disk wind, the morphology and kinematics of which we analyze in more detail in Sect. 4. At even lower relative velocities, in the range |∆υ| ≈ 1.4…2.2 km s⁻¹, these streaks become wider and are complemented by a strong and asymmetric butterfly-like structure in the middle, with left (north east) and right (south west) parts strictly separated between lower and higher velocities. We show below (Sect. 4) that this butterfly structure originates from the warm atmosphere of the flared Keplerian disk, while the streaky extensions perpendicular to the disk originate from a rotating disk wind. The features described above are more clearly visualized in the binned channel maps of the observed CO emission (Sect. 4.2).

At the lowest relative velocities, in the range |∆υ| ≲ 1.2 km s⁻¹, the structure of the recovered ¹²CO(2–1) emission becomes very complex and fainter. Despite the complementation by short-spacing data from single-dish observations, these central velocity channels could not be restored well and suffer from resolved-out extended emission and possibly also from self-absorption. They were therefore masked for generating the moment maps (Figs. 3 and 5) and are not considered in the subsequent modeling and analysis.

At an effective angular resolution of <0.5″, the total intensity map (0th moment) of ¹²CO(2–1) (Figs. 3 and 5) shows an X-shaped morphology with emission coming from both the disk itself and an outflow or disk wind. The 1st moment map shown in the same figures indicates that the entire X-shaped CO structure, including the extended vertical streaks, is coherently rotating with the same orientation as the disk, as illustrated by the 1st moment maps of ¹³CO(2–1), HCO⁺(1–0) and (3–2), which are also shown in Fig. 3. In order to illustrate how this inner resolved X-shaped structure around the disk is related to the image of the smooth rotating molecular outflow presented in Launhardt et al. (2009), we also show in Fig. 5 the original 1 st moment map from 2009, as well as a smoothed image produced from the new data with uυ tapering (Gaussian radius 200 m) resulting in ≈1.″0 resolution. The new, high-resolution image, which reveals the inner X-shaped structure with the rotation signature, is fully compatible with the smooth image of the extended rotating outflow presented in Launhardt et al. (2009). Not surprisingly, the addition of the longer-baseline data did not significantly affect the flux and the new high-resolution CO map recovers the same total CO flux as the lower-resolution map from 2009 (35 ± 3 Jy km s⁻¹). We analyze the flux distribution in more detail in Sect. 4.2.

Figure 6 shows the position-velocity diagram (PVD) of¹² CO(2–1) along the mid-plane of the disk, along with PVDs of the other three lines. The Keplerian rotation velocity field can be clearly seen in the ¹²CO PVD, with the drawback that the emission from the outer parts of the disk at lower relative velocities might be affected the velocity overlap with the surrounding self-absorbing envelope, although the Keplerian velocity at the outer edge of the disk () would still be υ_kep ≈ 1.6 km s⁻¹, which is significantly larger than the affected ∆υ ≈ ±1.2 km s⁻¹. From the symmetry of the ¹²CO(2–1) (and HCO(3–2)) PVDs (mirrored and cross-correlated), we derive a systemic velocity for the CB26 disk of υ₀ = 5.95 ± 0.05 km s⁻¹. The ¹²CO(2–1) velocity field can be fit with a Keplerian disk around a central mass of M_* = 0.55 ± 0.1 M_⊙ (Sect. 4). To guide the eye in the PVDs (Fig. 6), we overplot a Keplerian rotation curve for M_* = 0.55 M_⊙ and υ₀ = 5.95 km s⁻¹; this is not a fit to the data. A more detailed analysis of the velocity field, including an attempt to separate the contribution from the disk and the outflow (disk wind) is undertaken in Sect. 4.

Fig. 4 Spectral energy distribution of CB 26-YSO 1, adopted from Zhang et al. (2021), where we also list the fluxes in Table 1. Red squares and green dots with error bars show the observational data. The dashed black line shows best-fit model SED of Zhang et al. (2021). The dashed-dotted blue line represents the adopted contribution from free-free emission. The dotted red line corresponds to the sum flux of dust and free-free emission between 2.9 mm and 4.0 cm. The break in the SED at ~103 µm is due to the transition from total flux measurements at shorter wavelengths to interferometric data at longer wavelengths, which do not recover the total flux from the extended envelope emission.

3.4 ¹³CO(1–0)

The¹³CO(1–0) channel maps are shown in Fig. A.4 and the spectrum at the position of the disk center is shown in Fig. 3. Due to the narrow spectral bandwidth of the correlator (4.8 km s⁻¹; Sect.2.1), only the relative velocity range between −2.5 km s⁻¹ and +2.1 km s⁻¹ is covered, translating into a Keplerian radius range of r < −80 au and > 110 au, respectively, which means that the observations do not trace the inner |r| ≈ ±100 au of the disk, even if there would be ¹³CO(1–0) emission from this range. Emission from the disk is detected in the relative velocity range from ∆υ ≈ −2.35 to 1.65 km s⁻¹, translating into a Keplerian radius range of r < −90 au and > 180 au, respectively. The central velocity channels between ∆υ ≈ −1.0 and 0.5 km s⁻¹ could not be restored well and also suffer from resolved-out extended emission (plus possibly self-absorption). Similar to ¹²CO(2–1), these channels were therefore masked for generating the moment maps (Fig. 3). The envelope and the outflow are not, or only marginally, detected at low relative velocities, and may also suffer from resolved-out structure and self-absorption.

The total intensity map (04th moment) of ¹³CO(1–0) (Fig. 3) traces the disk with most of the emission coming from the outer regions. The 1 st moment map, also shown in Fig. 3, indicates the same rotation signature as the other lines, indicative of a Keplerian disk. This can also be seen in the PVD shown in Fig. 6.

Fig. 5 12CO(2–1) mean velocity field (1st moment map, color) and integrated intensity (black contours) of CB 26 at successively higher angular resolution. Left: as presented in Launhardt et al. (2009) at 1.″47 resolution. Middle: with the complete data set presented in this paper, but mapped with a uυ taper of radius 200 m and natural weighting (to restore a maximum of extended emission), resulting in an effective angular resolution of ≈1.″0. Right: same as middle, but mapped without tapering and with robust uυ weighting, resulting in an effective angular resolution of ≈0.″45. This image shows a smaller (zoomed) area than the other two maps, as indicated by the gray dashed rectangle in the middle panel. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corners of each panel. Overlaid as gray contours at 2.5, 7.5, and 15 mJy/beam is the 230 GHz dust continuum emission from the disk.

Fig. 6 Position-velocity diagrams along the plane of the disk of 12CO(2–1), 13CO(1–0), HCO+ (1–0), and HCO+ (3–2). Contours start at 3σ. Except for HCO+ (3–2), the envelope-dominated central velocity channels were masked out for the analysis (see also Figs. A.1, A.4 and A.5). Overlaid as red contours for comparison (with the y-axis being ∆y instead of υLSR) is the 230 GHz dust continuum emission from the disk (2.5, 7.5, and 15 mJy beam−1), rotated by 32°. Green lines show the Keplerian rotation curve for a central mass of 0.55 M⊙ and systemic velocity of υ0 = 5.95 km s−1 (only to guide the eye; this is not the fit to the data).

3.5 HCO⁺

The channel maps of HCO⁺(1–0) and HCO⁺(3–2) are shown in Figs. A.5 and A.6 and the spectra at the position of the disk center are shown in Fig. 3. HCO⁺(1–0) emission from the disk is detected in the relative velocity range |∆υ| ≈ ±2.2 km s⁻¹, translating into a Keplerian radius range of |r| > 80 au. Some extended emission mainly from the north west outflow lobe (blue, pointing toward us) is also detected in the relative velocity range −2.2 to −1 km s⁻¹. Despite the complementation by shortspacing data from single-dish observations, the central velocity channels of HCO⁺(1–0) are corrupted between ∆v ≈ −0. 8 and 0.4 km s⁻¹ by resolved-out emission and self-absorption from the extended envelope. Similar to ¹²CO(2–1), these channels were therefore masked before generating the moment maps (Fig. 3). The total intensity map of HCO⁺(1–0) (Fig. 3) shows the disk basically unresolved plus some extended emission from the envelope and the north west outflow lobe. H¹³CO⁺ (1–0) emission is only marginally detected in the relative velocity range − 1 to 0.1 km s⁻¹ and only when integrating over the entire primary beam area, indicating that it most likely originates in the envelope.

HCO⁺(3–2) emission is also detected in the relative velocity range |∆υ| ≈ ±2.2 km s⁻¹, translating into a Keplerian radius range of |r| > 80 au. The central velocity channels between ∆v ≈ −0.7 and 0.3 km s⁻¹ might also be slightly affected by resolved-out emission and self-absorption from the extended envelope (see also Fig. 3). Since this effect is much smaller than in the lower-excited line, we did not mask these channels. In contrast to HCO⁺(1–0), the total intensity map of HCO⁺(3–2) shows the disk slightly resolved and no traces of the envelope or outflow. The 1st moment maps of both HCO⁺ lines (Fig. 3), as well as the PVDs along the plane of the disk (Fig. 6) indicate the same rotation signatures as the CO lines.

4 Modeling and analysis

4.1 Disk model

A physical model of the disk was generated with the DUSK code, which described in detail in Akimkin et al. (2012). The DUSK code calculates axially symmetric density and temperature distributions assuming that the disk is Keplerian and remains in vertical hydrostatic and thermal equilibrium. It is heated mostly by stellar radiation and is cooled by its own infrared emission. Dust and gas are assumed to be well-mixed and have the same temperature. The main parameters of the disk model are (i) the inner and outer radius of the disk, R_in and R_out, (ii) the gas surface density distribution of the disk, Σ^gas(r), (iii) the inclination of the disk, i_disk, and (iii) the mass, M_*, effective temperature T_{eff, *}, and luminosity, L_*, of the central star.

Parameters R_out, Σ^gas(r) are taken from Akimkin et al. (2012) from their best-fit model of CB 26 “230+270 GHz” (see their Table 2). The fraction of stellar radiation that is intercepted and absorbed by the disk is derived by Akimkin et al. (2012) to be 0.1 L_⊙. Under the assumption of a razor-thin infinitive disk, which intercepts one quarter of the stellar radiation (Armitage 2010), the bolometric luminosity of the system (stellar luminosity) should then be at least 0.4 L_⊙. However, in the model of Akimkin et al. (2012), the stellar luminosity degenerates with disk surface density (see their Fig. 7), so that the derived fraction 0.1 L_⊙ is uncertain within a factor of few. In our current modeling, we do not vary the stellar luminosity and disk surface density to keep the number of parameters manageable.

The total mass of the adopted disk model is ≈ 0.2 M_⊙ when assuming a gas-to-dust mass ratio of 100. The mass of the central star(s) will be constrained from the Keplerian velocity field of the disk and is therefore a free parameter in the disk modeling. The same applies to the disk inclination, which is only loosely constrained by the continuum data. The size of the dust emission-free inner hole of au derived by Akimkin et al. (2012; see also Sauter et al. 2009) is not compatible with the observed velocity structure of ¹²CO(2–1). In Sect. 3.3, we show that CO emission is clearly detected from rotational velocities up to ≈5.9 km s⁻¹, which corresponds to an inner Keplerian radius of au, given the range of possible stellar masses that fit the Keplerian velocity field. It is therefore safe to assume that the dust hole in the disk is not gasfree and the gas disk extends down to an inner radius of about 10 au or smaller. On the other hand, Zhang et al. (2021), in their recent analysis, show that the dust hole could actually also be as small as 10 au (). Therefore, we also treat as a free parameter in the disk modeling.

The consistency between modeled and observed CO(2–1) emission was measured with the following (reduced) χ²-criterion: (1)

where and are the observed and modeled intensities at a given spatial and spectral pixel, N is the number of adopted data points, and σ is the noise level in the observed spectra. The sum is calculated only over those spatial and spectral pixels which satisfy the following two conditions. To minimize contributions from the outflow in the observed spectra, we compare model and observations only at those spatial positions where the model intensity integrated over the line profile is higher than a certain threshold, which we chose to be 1 Kkm s⁻¹. We also exclude the velocity channels at ∆υ = ±1.2 km s⁻¹ around the central velocity since they are resolved-out by the interferometer and may be affected by self-absorption from the envelope (see Sect. 3.3). With these criteria, we effectively use ≈ 50 000 out of the 102 × 102 × 61 = 614 644 total data points in the observed data cube.

The star mass, disk inclination, and inner radius of the gas disk are the only free parameters for our initial grid of models. Parameter ranges of the model grid are listed in Table 4. The synthetic molecular channel maps are calculated with the URAN(IA) code described in Pavlyuchenkov et al. (2007). The abundance of CO is described as: X(CO) = 1 × 10⁻⁴ where T > 20 K and X(CO) = 1 × 10⁻⁷ where if T ≤ 20 K, to account for the CO snowline (e.g., Bosman et al. 2018). The synthetic maps are convolved with the respective Gaussian clean beam before they are compared to the observed ¹²CO(2–1) maps derived with robust deconvolution (see Table 1). Figure 7 shows the marginal two-parameter χ² maps for the three model grid parameters, from which we determine the best-fit model that we use to subtract from the data. For the disk inclination, we derive a value of i_disk = 87 ± 2°, consistent with earlier estimates (see Table 3). For the size of the inner hole in CO, we derive a value of au. For the mass of the central star, we derive M_* = 0.55 ± 0.1 M_⊙. Table 4 summarizes all parameters of the best-fit disk model that we use in this paper, including the parameters adopted from Akimkin et al. (2012) and the ranges and best-fit values of the parameters evaluated here in the CO modeling. Figure 8 shows the distributions of the hydrogen and CO volume density and kinetic gas temperature of the best-fit disk model that was used to calculate the synthetic CO(2–1) channel maps of the disk emission. Clearly visible in the middle panel is also the CO iceline at ~80 au, where the mid-plane temperature drops below 20 K.

The ¹²CO(2–1) channel maps of the best-fit disk model are shown in Fig. A.2. These channel maps are then subtracted from the observed channel maps (Fig. A.1). The residual ¹²CO(2–1) emission (Fig. A.3) is now considered a much better representation of the disk wind than the total observed emission, albeit we still have to consider possible residual disk contamination due to model imperfections and the simple subtraction procedure. It nevertheless allows us to trace the outflow closer to the disk than with the combined emission. The subsequent analysis of the outflow morphology is therefore based on these residual channel maps (Fig. A.3).

Fig. 7 Marginal two-parameter χ2 maps for model grid parameters M*, idisk, and derived by comparing the synthetic maps of the 12CO(2–1) disk models with the observations (robust deconvolution). Dashed lines indicate the parameter values for the chosen “best-fit” disk model that was then subtracted from the data.

Fig. 8 Hydrogen volume density, kinetic gas temperature, and CO volume density distribution of the best-fit disk model that was used to calculate the synthetic CO(2–1) channel maps of the disk emission (Fig. A.2). The white line in the middle panel marks the CO snow line at Tgas = 20 K.

Fig. 9 Mean velocity field (1st moment maps) of 12CO(2–1) in the velocity range ±6 km s−1 of the best-fit disk model (left; Table4 and Fig. A.2) and the residual CO emission after subtracting the disk model and masking the corrupted central velocity channels between ±1.2 km s−1 (right). Black solid contours show the total integrated intensity of the CO emission from the model disk. Magenta contours show the 230 GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Dashed contours show the total integrated intensity of the residual CO emission. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner (dark gray: CO, light gray: dust). The maps are rotated counterclockwise by 32° such that the disk and outflow axes are aligned with the z-axis.

4.2 Outflow analysis

Here we analyze the channel maps and PVDs of the residual ¹²CO(2–1) emission, that is, after subtracting the CO disk model, and derive the outflow parameters. As for the disk model, we use the rotated (by 32°, Sect. 3.1) maps such that the disk midplane is aligned with the x-axis, and the outflow and disk rotation axis is aligned with the z-axis. Following Hirota et al. (2017, their Fig. 4), we use cylindrical coordinates and denote the axis parallel to the disk midplane “r”, and the perpendicular rotation axis “z”. The high-resolution CO data used here to analyze the detailed structure of the outflow as close as possible to its origin reliably trace the outflow cone walls only out to ≈600 au below the disk. But, the integrated intensity maps, especially at lower angular resolution (Fig. 5), indicate a total length of the south eastern outflow lobe of ≈ 1100 au, which we adopt here as the total half-length of the CO outflow.

Figure 9 shows the mean velocity fields (1st moment maps) of ¹²CO of the best-fit disk model and of the residual (disk-subtracted) CO emission, overlaid with contours of the integrated line emission (0th moment). The butter fly-shaped appearance of both the total intensity map and the mean velocity field already indicate that the residual emission resembles a weakly opened double cone that connects to the disk at r ≈ 40 au and rotates with the same orientation as the disk.

This is even more clearly seen in the binned channel maps (Fig. 10). At the highest rotation velocities with ∆υ = ±4.3 km s⁻¹, we see emission from the innermost parts of the Keplerian disk (middle left panel), but also already disk wind residuals emerging nearly perpendicular off the disk surface at ∆r ≈ ±0″2 – 0.′25 (30–40 au). At intermediate rotation velocities with ∆υ = ±2.9 km s⁻¹, the disk wind residuals in the lower middle panel show long linear streaks extending nearly perpendicular from the disk with a very small opening angle in both directions and both up and downward at Δr ≈ ±40 au. At the lowest rotation velocities with Δυ = ±1.7 km s⁻¹, the disk wind residuals seem to suggest a slightly larger opening angle, in particular at the upward-facing blue-shifted left wing. But this impression may, at least in part, be caused by the not perfect separation of emission from the disk atmosphere from the disk wind emission. The lower side of the outflow appears to be slightly bent toward the left at heights z < −200 au below the disk. The upper red-shifted (right) cone wall is in all three velocity ranges the least prominent wing as it may be most affected by the combination of red-shift due to rotation and blue-shifted outflow direction (facing toward us), and thus be blended by the self-absorbing envelope. The same, but to a lesser degree, applies to the lower blue-shifted (left) cone wall.

Figure 11 shows the total intensity map of the residual ¹²CO (2−1) emission after subtracting the disk model and masking the corrupted central velocity channels between ±1.2 km s⁻¹, together with the total intensity contours of the CO emission from the disk atmosphere and the thermal dust emission from the disk. The X-shaped morphology of the outflow with pronounced upper left (blue-shifted) and lower-right (red-shifted) wings becomes again very evident. The PVD of the horizontal cut through the disk mid-plane (top) shows the excellent, albeit not perfect match between the total observed CO emission and the CO model of the disk. The PVDs of the horizontal cuts through the residual (disk-subtracted) CO emission at z = ±80 au show that there is clearly excess emission at Δr ≈ ±40 au from the disk rotation axis and at Δυ_LoS ≈ ±3 km s⁻¹. The PVDs of the vertical cuts through the residual CO emission at Δr = ±40 au show that the LoS velocity of the CO emission from the vertical streaks does not significantly change (decrease) with separation from the disk.

To derive more quantitative outflow parameters, such as the outflow width, opening angle, and specific angular momentum distribution, we generate PVDs, such as in Fig. 11, at different heights z = −480 to 320 au by 80 au below and above the disk and apply Gaussian fits to cuts through the wind cone emission along the X axis, and visual inspection of cuts along the velocity axis (to avoid confusion with the residual disk and envelope emission). This approach corresponds to the “double-peak separation method” described by Tabone et al. (2020), with the only difference that we determine the offsets and FWHM by fitting 1-D profile cuts through the PVD, instead of just locating the peak in the PVD image. At vertical heights |z| ≥ 320 au, the cone wall profiles are no longer approximately Gaussian and symmetric, but show a broad wing of emission toward the inner side of the cone. Here, we only fit the core of the emission profile. Figure 12 shows the resulting values for the X-location, r, and mean LoS velocity, υ_LoS, of the outflow cone walls, as well as the radius, R, the toroidal or rotation velocity, υ_rot, specific angular momentum, j, and half-opening angle, θ, as function of vertical distance from the disk midplane, Δz.

The right panel a of Fig. 12 shows that the outflow remains strongly collimated out to z ≈ −500 au below the disk. Above the disk, the emission from the right cone wall is only traced reliably out to z ≈ 100 au, possibly due to the reasons mentioned above. Therefore, it is unfortunately impossible to derive the outflow width, rotation velocity, etc., at larger z above the disk. However, the location and LoS velocity of the upper left cone wall suggests that the upper (blue-shifted) lobe behaves very similar as the lower (red-shifted) lobe, at least out to distances of z ≈ −500 au, until which the left cone wall is traced. Furthermore, it is evident that the entire outflow seems slightly bent toward the left at vertical distances z ≲ −200 au below the disk. We take a very conservative approach to estimating the uncertainties from the FWHM widths of the cone walls and their velocity, rather than using the formal uncertainties of the Gaussian profile fits, which would result in unrealistically small error bars. Although this may lead to an overestimation of the actual uncertainties, this approach should include the errors introduced by the self-absorption by the envelope, as well as the imperfect disk model subtraction from the data. We note that the cone wall FWHM widths are all smaller than the synthesized beam size of ≈70 au, which implies that the cone walls are spatially unresolved.

Panel b shows the LoS velocity of the left (blue-shifted, that is, negative velocities) and right (red-shifted) cone walls. The two velocities stay approximately symmetric with respect to the systemic velocity of the disk, while the absolute LoS velocity offsets slowly decrease with vertical distance from the disk. What is remarkable, though, is that there is no significant jump in the mean velocity between the upward, on large scales blue-shifted, and downward, on large scales red-shifted, outflow wings. Only at distances z ≤ −320 au downward, there is a slight shift of −0.8 km s⁻¹. If we use the mean velocity offset between the upward and downward-facing cone walls of ±0.12 km s⁻¹, we derive with i = 87 ± 2° a radial outflow velocity of . However, this analysis is hampered by the fact that the emission from the right upward-facing (blue-shifted) cone wall is only traced reliably out to z ≈ 100 au. We therefore consider this value a lower limit to the radial outflow velocity υ₀. On the other hand, we derive in Launhardt et al. (2009) from the lower-resolution CO maps a mean LoS velocity offset of ±0.65 km s⁻¹ by cross-correlating the mirrored PVDs of the two outflow lobes. Using i_disk = 87 ± 2°, we derive , which we consider an upper limit. In lack of more precise constraints, we therefore adopt a mean value of for the radial outflow velocity.

Panel c shows the radius of the outflow cone R = r^left – r^rıght, which increases slowly with separation from the disk from R ≈ 43 ± 5 au at z = ±40 au to R ≈ 102 au at z = −480 au. The upward-facing cone is only fully traced out to z = 80 au, where we determine a cone radius of R ≈ 57 au. Seeing that the cone radius increases approximately linearly with z, we extrapolate a launch radius of the cone walls at z ≈ 40 au, where the surface of the flared disk atmosphere at this radius is (see Fig. 8), of R_l = 40 ± 5au. The observed thickness of these wind cone walls is spatially unresolved (i.e., ≪70 au), and there is no direct indication in the channel maps that the wind is launched from a wider range of radii. However, given the angular resolution of our data, we cannot exclude that the main launching region extends over a few tens of au on the disk surface, that is, basically from the inner rim of the CO disk at au to the CO iceline at ≈80 au (Sect. 4.1).

Panel d shows the rotation velocity⁷, , of the outflow cone walls, which gradually decreases from υ_rot ≈ 3.0 ± 0.35 km s⁻¹ at z = ±40 au to ≈ 1.7 km s⁻¹ at z = −480 au. Considering a central mass of 0.55±0.1 M_⊙ and the increasing radius R of the outflow cone with vertical distance z from the disk, this decreasing rotation velocity is consistent with (specific) angular momentum conservation j = R × υ_rot ≈ 140 ± 33 au km s⁻¹ of a flow that is launched from the disk surface at the single radius R_l = 40 au with the Keplerian rotation velocity of υ_K ≈ 3.5 ± 0.4 km s⁻¹.

Panel e shows the specific angular momentum, j = R × υ_rot, and its 1σ uncertainty. Although it seems that j is slowly increasing with z from 110 au km s⁻¹ at z = −40 au to 175 au km s⁻¹ at −480 au, this is significant only at the 2σ level.

Panel f shows the half-opening angle, θ = arctan(). Below the disk, where we can trace the cone walls out to z ≈ 500 au, the opening angle appears to be approximately constant with θ ≈ 7°. Above the disk, where we can trace the cone walls out to only z ≈ 160 au, the opening angle appears to be slightly larger with θ ≈ 14 ± 2°.

As already mentioned in Sect. 3.3, the new high-resolution CO map recovers the same total flux as seen in the lower-resolution map of Launhardt et al. (2009, ≈35 Jy km s⁻¹). Of this, 10% (3.5 Jy km s⁻¹) originate from the disk atmosphere, ≈19% (6.5 Jy km s⁻¹) from the disk wind cone walls, and 71% (25 Jy km s⁻¹) from the diffuse component. To be more precise, what we call here the “wind cone walls” are only the edge-on-seen parts of the cone walls, while the “diffuse component” is mostly related to the other, in projection less-structured parts of the cone walls, such that the total flux from the disk wind amounts to 31.5 ± 3 Jy km s. However, due to the masked central velocity channels at Δυ = ±1.2 km s⁻¹ (Sect. 3.3), this value may slightly underestimate the actual total CO flux.

Making the simplifying assumption that most of the CO emission from the outflow is optically thin (which may not be strictly fulfilled for the cone walls), assuming local thermodynamic equilibrium, and following the formalism outlined in Bourke et al. (1997, their Eq. (A7)), here rewritten for a standard value of X = N(H₂)/I/(CO) = 2.3 × 10²⁰ cm⁻² (K km s⁻¹) (e.g. Rohlfs & Wilson 1996), standard ISM composition with mean molecular weight µ_m = 1.36, and substituting , the total gas mass, often referred as the CO mass, is given by (2)

where Ω is the source solid angle, D is the source distance in kpc, and λ is the observing wavelength. Using Ω ≈ 40 arcsec², D = 0.14 kpc, λ = 1.3 mm, and ∫I_vdv = 31.5 Jy km s⁻¹, we derive a total mass of the visible outflow of 8.4 × 10⁻⁴ M_⊙, which we round up to M_CO ≈ (1 ± 0.3) × 10⁻³ M_⊙ because of the CO flux under-estimation mentioned above. This is basically the same value as was derived by Launhardt et al. (2009) by integrating the CO model of the outflow. Table 5 summarizes this and other derived parameters of the CB 26 CO outflow. The schematic cartoon shown in Fig. 13 summarizes the geometry of the disk and disk wind derived here.

Fig. 10 Channel maps of the12CO (2−1) emission in binned, 0.8 km s−1 wide channels around the velocity offsets Δυ from υ0 = 5.95 km s−1 indicated on top. Top row: observed total emission (see Fig. A.1); middle row: disk model (see Sect. 4.1 and Fig. A.2); bottom row: residuals, that is, observed total emission with disk model subtracted. Blue contours indicate negative velocity offsets, red contours positive ones. CO contours start at 15 mJy beam−1 (left and middle panels), and 40 mJy beam−1 (right panels) and are spaced by 15 mJy beam−1. Blue and red arrows in the lower left panel show the inclination orientation of the disk and outflow. Black contours show the 230 GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner of the lower left panel (dark gray: CO, light gray: dust). The underlying gray-scale image shows the NIR K-band image of the bipolar reflection nebula (see Fig. 1).

Fig. 11 Bottom left: total intensity map (gray-scale and black solid contours) of the residual12CO (2−1) emission after subtracting the disk model and masking the corrupted central velocity channels between ±1.2 km s−1, integrated over the velocity range ±6 km s−1. Red contours show the observed 230GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Green contours show the total integrated intensity of the CO emission from the model disk. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner (dark gray: CO, light gray: dust). Top: position-velocity diagrams (PVD) of 12CO (2−1) at three cuts parallel to the plane of the disk (a, b, c, as indicated by dashed lines in the total intensity map; gray-scale and black contours). The middle panel shows the PVD of the unmasked total observed CO emission along the disk mid-plane, while the upper and lower panels show disk wind PVDs of the residual CO emission (after subtracting the disk model) and with the central corrupted velocity channels masked. Green contours show the PVDs of the disk CO model along the disk mid-plane. Bottom right: same as above, but for the three vertical cuts labeled d, e, and f. Green contours show the PVDs of the disk CO model along the same vertical cuts.

Fig. 12 Outflow parameters derived from PVDs of disk-subtracted CO residuals at different vertical heights below and above the disk midplane: (a) X location r and FWHM width of the outflow cone walls. The errorbars denote the FWHM widths of the cone walls, not the uncertainty of the mean (which is smaller than the symbols). The light gray data points at z = ±40 au might be affected by imperfections in the disk model that was subtracted from the data. (b) Line-of-sight (LoS) velocity (relative to systemic velocity υ0) and FWHM width of the outflow cone walls. Red squares show the center velocity between left and right outflow cone. (c) Radius R of the outflow cone, with uncertainties derived from the FWHM widths of the outflow cone walls. (d) Rotation velocity and 1σ uncertainty of the outflow cone walls. The red solid and dotted lines show the expected rotation velocity and uncertainty for angular momentum conservation of a wind that is launched from the disk at a single radius Rconn = 35 ± 10 au with a Keplerian rotation velocity of υK = 3.7 ± 0.3 km s−1. (e) Specific angular momentum j and 1σ uncertainty. (f) Half-opening angle 9 and 1σ uncertainty.

5 Discussion

5.1 The disk and central star

According to its spectral index can be classified as a Class I embedded YSO (Adams et al. 1987). Thanks to the relatively tight constraint on the mass of the central star, evolutionary track models provide reasonable constraints on the age of the system, despite the relatively large uncertainty of its luminosity. While a main-sequence star of M_* = 0.55 ± 0.1 M_⊙ (M0.5 V) would have a bolometric luminosity of L_* ≈ 0.05 L_⊙, a star with this mass with L_* = 1.0 ± 0.4 L_⊙ (Appendix B) should have an age of 1 ± 0.5 Myr, depending on the model used (e.g., Siess et al. 2000; Allard 2014; Baraffe et al. 2015). The effective temperature of such a star should be T_eff = 3700 ± 150 K. Even if we consider the unlikely case of the mass and luminosity split into two equal-mass stars, the mean (most likely) age would still be ≈ 1.5 Myr. On the other hand, Simon et al. (2019) have shown that, when using using magnetic stellar evolution isochrones from, for example, Feiden (2016), the inferred ages of young stars (1–10 Myr) can be up to three times larger than the ones derived from nonmagnetic models. Considering this effect, we adopt an age of Myr for CB 26-YSO1. We also tried to derive further constraints on the spectral type(s) of the central star(s) by employing visual long-slit spectroscopy on the bipolar reflection nebula, hoping that we could recover at least the strongest stellar lines, but this was not successful.

The size of central hole in the dust disk, which was independently derived by several authors (see Table 3), has a relatively large uncertainty with values (incl. uncertainties) ranging from au to ≈50 au. The value and uncertainty derived in the most recent and comprehensive study and based on the most extensive data set, au (Zhang et al. 2021) actually overlaps with au, which we derive from the CO emission (Sect. 4.1, Table 4).

An inner hole in the detected dust emission does not necessarily mean that there is a sharp physical cut-off at this radius, but it could also result from a combination of decreased dust opacity and column density. For the gas emission, the steep Keplerian velocity gradient at small radii spreads out the emission over a steeply increasing velocity range. Furthermore, spatial “beam dilution” can lead to a significant loss of sensitivity to emission from the inner disk, where the ratio of emitting area to beam size becomes increasingly smaller, in particular for HCO⁺ with the relatively large synthesized beams of ≈3″.5 for HCO⁺(1−0) (corresponding to 490 au linear resolution) and ≈1″.35 for HCO⁺(3−2) (190 au). Hence, the lack of detected emission from Keplerian velocities corresponding to radii <80 au (Sect. 3.5) does not necessarily mean that there is a HCO⁺ hole of that size in the disk. The same argument also holds for the CO emission, that is, the value of au, derived in Sect. 4.1, may as well be affected by this spectral and spatial beam dilution and the CO disk might have a much smaller or even no inner hole at all. Molecular line observations of CO and HCO⁺ at higher angular resolution and better sensitivity would be required to better constrain the existence and size of the inner whole in the gas disk. Despite these uncertainties, the CB 26 disk with such a significant inner hole in the dust emission can be considered as a transitional disk (Strom et al. 1989; Skrutskie et al. 1990; Espaillat et al. 2014), although the primary cause for this inner disk clearing remains undisclosed. We don’t know whether it is caused by a stellar companion, newly formed giant planets, or photoevaporation (e.g. Alexander et al. 2014; Espaillat et al. 2014; Pascucci et al. 2023).

The mass of the flared disk is also uncertain with estimates ranging from M_disk ≈ 0.08 M_⊙ (Zhang et al. 2021) to ≈0.2 M_⊙ (this paper, Sect. 4.1), and ≈0.3 M_⊙ (Sauter et al. 2009). These differences can, at least in part, be attributed to the use of different dust opacities. Considering that the latter estimate seems unrealistically high, we therefore adopt a value of M_disk ≈ 0.15 ± 0.05 M_⊙. With this, the disk mass amounts to ≈27 ± 10% the mass of the central star. The disk would thus be prone to gravitational instabilities and developing quasi-stable spiral arms that process infall from the surrounding cloud (e.g., Kratter & Lodato 2016, see also Sect. 5.2).

Fig. 13 Purely empirical schematic cartoon illustrating the main features and geomoetry of the disk and disk wind derived in this paper.

5.2 The envelope

We actually do not see any emission from the extended envelope in our interferometric molecular line and thermal dust continuum maps used in this paper. The reason could be that such extended emission is simply resolved out in the interferometric maps. But we clearly see the effects of self-absorption due to cool CO (and HCO⁺) gas in the envelope. While the fast-rotating inner parts of the Keplerian disk are not affected, the CO emission from the slower-rotating outer parts might still be affected by this self-absorption. The size and total mass of the envelope was estimated from single-dish observations to be r_env ≈ 10⁴ au and M_env ≈ 0.1–0.2 M_⊙ (Launhardt et al. 2010, 2013). The total gas mass of the globule CB 26, in which CB 26 - YSO 1 is embedded, was estimated to be ≈2.4 M_⊙ (Launhardt et al. 2013).

Since the CB 26 YSO and disk are still embedded in a denser envelope, which itself is embedded in the globule CB 26 (Fig. 1), it is very likely that the disk still accretes gas from the envelope, even though we have no direct observational evidence of such late accretion. As Küffmeier et al. (2023) have shown, such late accretion can occur well beyond an age of 1 Myr, and it can still transport significant amounts of mass and angular momentum onto the disk. This could explain the relatively large disk mass (≈27 ± 10% the mass of the central star; Sect. 5.1), which is expected to reach its maximum during this phase (Hogerheijde 2001). The disk wind then just carries away this excess angular momentum.

5.3 The outflow

The disk wind is launched from the surface of the flared disk with a nearly constant half-opening angle θ ≈ 7°, leading to a rotating cone that continuously opens from radius ≈ 45 au at |Δz| = 80 au to ≈ 90 au at |Δz| = 400 au and can be traced by our high-resolution CO maps until about 300 au above, and 600 au below the disk midplane. The reason that the cone walls cannot be traced further away from the star and disk by the high-resolution data analyzed here is probably related to a combination of self-absorption by the envelope, decreasing contrast between the widening cone walls and ambient material, as well as sensitivity.

The analysis of the residual CO emission (Sect. 4.2), after subtracting the modeled emission from the disk atmosphere, suggests a launch radius at the disk surface of R_l = 40 ± 5 au, which corresponds to the inner part of the optically thick disk (, ). Since this value was derived by linear extrapolation of the cone wall locations from |z| > 80 au inward and with no further observational constraints of possible collimation acting inside |z| < 80 au, we consider it to represent an upper limit to the mean launch radius. We compare this with the ejection radius predicted by Anderson et al. (2003, their Eq. (4)) for the launching radius of a cold, steady, axisymmetric MHD disk wind at large distances, where the gravitational potential of the star is negligible, in the form derived by Pety et al. (2006): (3)

where R is the distance from the outflow axis to the considered volume element, is the Keplerian velocity at this position, and V_rot and V₀ are the measured rotation and radial outflow velocities at that position. Using the values and uncertainties for R and V_rot from Fig. 12 for z = −240… − 480 au, and a constant (with z) radial outflow velocity , we derive consistently au. This value is only slightly smaller than, and overlaps in the error margin with the launch radius R_l derived directly from the data, and we consider it as a lower limit to the mean launch radius. Therefore, we finally adopt R_l ≈ 20–45 au as the most likely range for the mean wind launch radius.

Despite its large uncertainty, the radial lift-off wind velocity of (Sect.4.2) is still is significantly smaller than the tangential velocity of HH 494 (86 ± 17 km s⁻¹, Appendix C), suggesting that an as-yet undetected high-velocity jet, rather than the outer disk wind, is responsible for HH 494. In Appendix C, we derive a dynamical age for this HH object-exciting jet of ≈3000 yr, which could imply that a, possibly FU Orionis-like (Hartmann & Kenyon 1996) accretion event a few thousand years ago must have filled the inner disk with gas (if there was a gas hole already present at this time) and triggered an energetic inner disk wind that produced this jet. Unfortunately, we could not derive any constraints on the current mass accretion rate onto the star, if there still is any, due to the extreme edge-on morphology.

With the launch radius and the specific angular momentum of the disk wind at hand, we can derive the magnetic lever arm parameter, λ: (4)

where j is the mean specific angular momentum of the outflowing gas, and Ω₀ is the Keplerian angular rotation speed at the launch point R_l (Blandford & Payne 1982; Spruit 1996; Ferreira et al. 2006; Tabone et al. 2020). Using the values and uncertainty ranges listed in Table 5, we derive λ ≈ 1.3 ± 0.7. Such a small value of λ, which is an essential feature of MHD wind models that relates the mass outflow rate to the accretion onto the disk, would imply a very efficient extraction of angular momentum, but is in line with recent nonideal MHD simulations of magnetized winds from protoplanetary disks (see de Valon et al. 2020, and references therein), as well as with observational constraints derived for other rotating disk winds in recent years (see Sect. 5.4 and Table 6). Vice versa, this efficient extraction of angular momentum by the disk wind implies that the angular momentum flux by accretion onto the disk, for which we have no direct observational constraints, must be of the same order as the one carried away by the disk wind. that is, a few times 10⁴ M_⊙ km² s⁻¹ yr⁻¹ (Table 5).

An important parameter for gaining insights into the wind-launching mechanism is the observed ratio of the outflow momentum flux (thrust) to the maximal possible thrust that can theoretically be provided by the bolometric luminosity of the central star (Pudritz et al. 2007). In Sect. 4.2 (Table 5), we derive a total outflow momentum flux of km s⁻¹ yr⁻¹. The maximum thrust that can be provided by a central star with bolometric luminosity L_* = (1 ± 0.4)L_⊙ is L_*/c = (2 ± 0.8) × 10⁻⁸ M_⊙ km s⁻¹ yr⁻¹, which is nearly three orders of magnitude smaller than the CO outflow thrust. Hence we conclude that photoevaporation cannot be the main driving mechanism for this outflow, but it must be predominantly a MHD-driven disk wind, even if we have no observational constraints on the morphology and strength of the local magnetic field⁸. Indeed, the CB26 disk wind falls right on the relation F_CO/F_rad = 250 (L_bol/10³ L_⊙)^−0.3 found by Cabrit & Bertout (1992), and later verified and extended by Wu et al. (2004) for a larger sample of molecular outflows from both low and high-mass YSOs. This suggests that small-scale molecular disk winds are, at least energetically, indistinguishable from larger-scale molecular outflows, which are usually interpreted as swept-up material (Pascucci et al. 2023).

5.4 Comparison to other rotation detections of jets and disk winds

Rotation of jets and outflows has long been predicted theoretically, but observational evidence started accumulating only over the past 20 yr. Rotation signatures in optical jets have been observed in a few sources and could be interpreted as tracing the helicoidal structure of the magnetic field in accretion-ejection structures (e.g., Bacciotti et al. 2003; Coffey et al. 2004, 2007; Woitas et al. 2005).

The first reports of rotation signatures in slow molecular winds (Pascucci et al. 2023) started to emerge about 15 yr ago. Lee et al. (2009) report the discovery of rotation signatures in the low-velocity molecular outflow around the highly collimated jet from the ~50 M_Jup BD-mass, ~2 × 10⁴ yr old protostar HH 211 located in the IC 348 complex of Perseus. Using Anderson’s relation (Anderson et al. 2003), they deduce an upper limit of the launching radius of ~0.014 au, which is much smaller than what we find in CB 26, and conclude that the wind is consistent with being launched by magneto-centrifugal force. They estimate that the flow has a mean specific angular momentum of ~50 au km s⁻¹. At about the same time, we reported in Launhardt et al. (2009) the discovery of rotation in the molecular outflow from the low-mass (~0.5 M_⊙), ~1 Myr old Class I YSO CB 26 - YSO 1.

Klaassen et al. (2013) report the discovery of a rotating molecular disk wind from the young (4–7 Myr) Herbig Ae star HD 163296. They derive a mass loss rate that is twice as high as the accretion rate onto the star, but no constraints on the launch radius. Chen et al. (2016) report the discovery of rotating molecular bullets associated with the high-velocity jet from the young variable Class 0/I transition protostar SVS 13 A located in the NGC 1333 star-forming region. Using Anderson’s relation (Anderson et al. 2003), they deduce that the jet-launching footprint on the disk has a radius of ~ 7.5 au. Bjerkeli et al. (2016) report the discovery of a rotating molecular disk wind that is ejected from a region extending up to a radial distance of 25 au around the young (few 10⁵ yr) low-mass (~0.5 M_⊙) protostar TMC1A located in the Taurus molecular cloud. They derive a specific angular momentum of the outflowing gas of j ≈ 200 au km s⁻¹ that is slowly increasing with distance from star. This source is thus the one that is most similar, both in terms of launch radius and specific angular momentum, to the CB 26 disk wind we report in this paper. Hirota et al. (2017) report signatures of rotation in the bipolar outflow driven by Orion Source I, a high-mass (5–6 M_⊙) YSO candidate. They derive launching radii of > 10 au and an outward outflow velocity of ~10 km s⁻¹ and conclude that the rotating outflow must be directly driven by a magneto-centrifugal disk wind. Zhang et al. (2018) report the discovery of clear signatures of rotation from about 120 au up to about 1400 au above the disk midplane of the low-mass (~0.2 M_⊙) protostar NGC 1333 IRAS 4C in the Perseus Molecular Cloud. From the angular momentum distribution in the outflow, they deduce the most likely launching radii to be 5–15 au.

A break-through discovery was probably the detection of rotation in the high-velocity protostellar jet launched at R_L = 0.05 au from the edge-on 0.04 Myr old, ~0.25 M_⊙ Class 0 protostar HH 212 in the L 1630 cloud of Orion by Lee et al. (2017). In a series of succeeding papers, Lee et al. (2021) analyze the interaction between the outer disk wind and the jet, and Lee et al. (2022) find that the jet is the densest part of a wide-angle wind that flows radially outward at distances far from the (small, sub-au) launching region around the protostar HH 212.

Another key observation was recently published by Moscadelli et al. (2022), who used observations at 22 GHz with the VLBI network to obtain very high-spatial-resolution astrometry of water masers that are associated with the jet originating from a massive (≈5.6 M_⊙) YSO embedded in the IRAS 21078+5211 star-forming region at a distance of 1.63 ± 0.05 kpc. They could show that the masers trace the velocities of individual streamlines spiraling outward along a helical magnetic field, launched from locations on the disk at radii ≤6–17 au, and that their motion is consistent with a MHD disk wind.

Table 6 summarizes the main parameters of the rotating molecular winds described here, approximately ordered by their respective launch radius R_L. For comparison with CB 26, we also list the respective values for the specific angular momentum of the outflown gas, j, and the magnetic lever arm parameter, λ, where these were explicitly given. While the rotating jets associated with the youngest (few 10⁴ yrs) low-mass Class 0 protostars HH 211 and HH 212 are launched from within a few stellar radii with specific angular momenta of a few tens au km s⁻¹, the outflows from the somewhat older (~1 Myr) Class I sources are driven by more extended molecular disk winds launched at radii of a few tens of au and specific angular momenta of a few hundred au km s⁻¹. The disk wind from CB 26-YSO 1 appears to be the one with the largest launch radius, consequently also the best-resolved one, albeit quite comparable to, for example, TMC1A.

6 Summary and conclusions

We obtained with the IRAM PdBI and analyze high angular-resolution (0″.53 × 0″.47) ¹²CO(2−1) molecular line and thermal dust continuum data at 230 GHz with the goal of revealing the nature and origin of the rotating molecular outflow from the young () edge-on-seen (i = 87 ± 2°) low-mass (0.55 ± 0.1 M_⊙) YSO - protoplanetary disk system CB 26 - YSO 1. The flared disk has an outer radius of au, an inner hole in dust emission of au, and a total mass of M_disk ≈ 0.15 ± 0.05 M_⊙. The size of the inner hole in the CO emission of au, which we derive from a least-squares fit of our disk CO model to the channel maps, might actually be an over-estimation and the CO disk might have a much smaller or even no inner hole at all. The source is embedded in the Bok globule CB 26 (Launhardt et al. 2009). This relative isolation certainly favors both the undisturbed morphology and the good observability of the described disk wind. Furthermore, only the fact that the source is still embedded in its birth cloud enables the possibility of ongoing accretion onto the disk, which in turn could trigger such a powerful disk wind. Since the source is located too far north to be accessible to ALMA, the IRAM PdBI provided the best opportunity to obtain such high-resolution millimeter data.

Our observations confirm the disk-wind nature suggested by Launhardt et al. (2009) and reject the alternative scenarios such as jet precession or two misaligned jets from a hypothetical embedded binary system that were mentioned in the same paper. The new high-resolution data reveal an X-shaped morphology of the CO emission close to the disk, and vertical streaks extending from the disk surface out to vertical heights of ≈600 au below, and ≈300 au above. We interpret this emission as the combination of the disk atmosphere (the X-shaped part close to the disk) and a well-collimated disk wind, of which we mainly see the outer walls of the outflow cone. We decompose these two contributions by subtracting a chemo-dynamical model of the CO emission from the disk atmosphere combined with line radiative transfer calculations, fit to the data, from the channel maps of the observed CO emission. This allows us to trace the disk wind down to vertical heights of ≈40 au where it is launched from the surface of the flared disk at a mean radius of R_L =20–45 au. The full launch region may actually extend over the entire inner disk from the inner rim at 10-30 au out to the CO iceline at ≈80 au.

The disk wind is rotating with the same orientation and speed as the Keplerian disk and the velocity structure of the cone walls, which open slowly with height (from R ≈ 45 au at |z| = 40 au to R ≈ 100 au at |z| = 500 au, with a half-opening angle 6 = 7 ± 2°) is consistent with angular momentum conservation of a flow that is launched from the disk surface at the single radius R_L = 35 ± 10 au with the Keplerian rotation velocity of V_K = 3.7 ± 0.3 km s⁻¹. From the ¹²CO(2−1) total intensity maps, we derive a total half-length of the CO-outflow of r₀ ≈ 1100 au and a total gas mass of M_CO ≈ (1 ± 0.3) × 10⁻³ M_⊙. With a radial lift-off velocity of km s⁻¹, we derive a dynamical age of the observed CO outflow of yr, which does most likely not reflect the actual age of the outflow. A Herbig-Haro object (HH 494) with a tangential velocity of 86 ± 17 km s⁻¹, located at 6’15 toward the north west, and exactly aligned with extended axis of the CO outflow at PA = 148 ± 1°, is suggested to originate from an accretion event that created an energetic inner disk wind and launched a yet undetected high-velocity jet about 3000 yrs ago.

The currently observed outer disk wind is found to very efficiently (λ ≈ 1 – 2) carry away angular momentum at a rate of j_CO ≈ 3 × 10⁴ M_⊙ km² s⁻¹ yr⁻¹, which implies that the angular momentum flux by accretion onto the disk, for which we have no direct observational constraints, must be of the same order. The wind has a total outflow momentum flux (thrust) of km s⁻¹ yr⁻¹, which is nearly three orders of magnitude larger than the maximum thrust that can be provided by a central star with bolometric luminosity L_* = (1 ± 0.4) L_⊙. Therefore, photoevaporation cannot be the main driving mechanism for this outflow, but it must be predominantly a MHD-driven disk wind. Indeed, the CB 26 outflow falls right on the relation F_CO/F_rad = 250 (L_bol/10³L_⊙)^−0.3 found by Cabrit & Bertout (1992) for a large number of molecular outflows over a wide range of luminosities. It is thus far the best-resolved rotating disk wind observed to be launched from a circumstellar disk in Keplerian rotation around a low-mass YSO, albeit also the one with the largest launch radius. It confirms the observed trend that disk winds from Class I YSO’s with transitional disks have much larger launch radii, and also larger specific angular momenta, than the inner disk winds and jets ejected from Class 0 protostars.

Acknowledgements

We wish to thank Ralph Pudritz for inspiring discussions and very helpful comments on the manuscript. We also thank Christian Fendt and Ilaria Pascucci for helpful comments and discussions, and Uma Gorti for her contributions in the early stages of this project. Special thanks to Thomas Müller for his help in creating the cartoon of Fig. 13. We acknowledge the Plateau de Bure IRAM staff for their help during the observations. We also acknowledge the Grenoble IRAM staff for their support during the first data reduction session in January 2006. The French Program of Physico-Chemistry (PCMI) is thanked for providing fundings to this project. T.H. acknowledges support from the European Research Council under the Horizon 2020 Framework Program via the ERC Advanced Grant Origins 83 24,28. Y.P. and V.A. were supported by the RSF grant 22-72-10029. This publication makes use of data products from the Wide-field Infrared Survey Explorer (WISE), which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/ California Institute of Technology, funded by the National Aeronautics and Space Administration. IRAM is supported by INSU/CNRS (France), MPG (Germany) and IGN (Spain). The Owens Valley millimeter-wave array was supported by NSF grant AST 9981546. The SMA is a joint project between the Smithsonian Astrophysical Observatory and the Academia Sinica Institute of Astronomy and Astrophysics and is funded by the Smithsonian Institution and the Academia Sinica.

Appendix A Channel maps

Here we show the channel maps of ¹²CO (2−1) as observed, as modeled for the disk atmosphere, and observed minus disk model, as well as of ¹³CO (1−0), HCO⁺ (1−0), and HCO⁺ (3−2). All channel maps are overlaid with contours of the 1.3 mm dust continuum emission from the disk.

Fig. A.1 12CO(2−1) channel maps of CB 26, obtained with PdBI in 2005 and 2009, rotated counterclockwise by 32°. Contour levels start at 15 mJy/beam (2σ r.m.s., see Table 1). Red contours mark the 1.2mm dust continuum emission from the disk (3 and 15 mJy/beam. The reference position is α2000 = 04h59m50.74s, δ2000 = 52°04′43.80″. The 12CO synthesized beam size is indicated as the gray ellipse in the lower right corner of the first channel map. The channel number and mean Δυ = υLSR _ 5.95 (in km s−1) are indicated in the top left and right corners of the maps, respectively. Maps in the central velocity channels between Δυ ≈ ±1.2 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope.

Fig. A.2 Same as Fig. A.1, but for the modeled 12CO (2−1) emission of the best-fit disk model for CB 26 (see Table 4). Contour levels start at 10 mJy/beam.

Fig. A.3 Same as Fig. A.1, but for the residual 12CO(2−1) emission from CB 26 after subtracting the best-fit disk model (Fig. A.2). Contour levels start at 15 mJy/beam.

Fig. A.4 Same as Fig. A.1, but for 13CO(1−0), obtained with OVRO in 2001. Contour levels start at 70 mJy/beam (2 σ r.m.s., see Table 1). Maps in the central velocity channels between ~ ±1.2 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope. Maps in the central velocity channels between Δυ ≈ −1.0 and 0.5 km s−1are corrupted by resolved-out emission and self-absorption from the extended envelope.

Fig. A.5 Same as Fig. A.1, but for HCO+ (1−0), obtained with PdBI in 2005. Contour levels start at 15 mJy/beam (~ 2 σ r.m.s., see Table 1). Maps in the central velocity channels between Δυ ≈ −0.8 and 0.4 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope.

Fig. A.6 Same as Fig. A.1, but for HCO+ (3−2), obtained with the SMA in 2006. Contour levels start at 300 mJy/beam (2 σ r.m.s., see Table 1).

Appendix B The luminosity of the central star

Integrating the SED (Fig. 4) and assuming isotropic irradiation and a distance of 140 pc, yields a total luminosity of ≈0.43 L_⊙. However, this value may not correctly reflect the intrinsic bolometric luminosity of the central YSO since, in particular at the shorter wavelengths, anisotropic irradiation due to the nonspherical geometry becomes significant. We therefore also derive separately the thermal contribution to the luminosity at λ > 12 µm, which is less affected by the nonspherical geometry, and where most of the observed luminosity originates. In this wavelength range, we derive a total thermal luminosity of , of which (58%) can be attributed to the disk and (42%) to the envelope. The value of is probably a good approximation, albeit slight underestimation, of the fraction of stellar luminosity that is intercepted and reprocessed by the disk.

Due to the nonspherical geometry and extreme edge-on configuration, the shorter-wavelength contribution to the luminosity from scattered light at λ ≤ 12 µm is likely to be significantly underestimated (see, e.g., Whitney et al. 2003). Comparison with radiative transfer models from the grid of SED models of YSOs provided by Robitaille et al. (2006) for configurations similar to CB 26 (e.g., model-ID 3016525) suggests that the NIR to MIR luminosity could be underestimated by of up to a factor of ten and the total luminosity by up to a factor of two. Vice versa, for a similar system seen pole-on, the observed luminosity could overestimate L_bol by up to a factor of two. We therefore estimate the most likely value of the intrinsic bolometric luminosity of the central star(s) to be . This value is in good agreement with Launhardt et al. (2009) and Zhang et al. (2021), who both estimated L_* ≈ 0.9 L_⊙.

Appendix C Proper motion of HH 494

The precise alignment of HH 494 at P.A. = 147.5 ± 0.5° with regard to CB 26-YSO1 (Stecklum et al. 2004) with the axes of the CB 26 disk and outflow (P.A. = 148 ± 1°, Sect. 3.1, see also Fig. 3) suggests that this HH object may originate from a jet that was ejected from CB 26 - YSO 1. Although no such jet nor another HH object have been detected yet, part of the explanation for the location of HH 494 may come from the specific environment of CB 26. As indicated by the diffuse cloudshine⁹ (Fig. 3), the vicinity of the Bok globule is relatively void of extincting and light-scattering material. HH 494 is located where the connecting line with CB 26 - YSO 1 encounters the first layer of diffuse material. It may thus mark the interaction spot of the (currently unseen) jet with the first layer of material it encountered after leaving the boundaries of the globule.

Based on the data described in Sect. 2.4, we also derived the proper motion vector of HH 494. Since the position of the HH object is identical on the Hα and S[II] images of both epochs within the uncertainties, we co-added them to increase the signal-to-noise ratio (S/R). Because the HH object is located close to a star of comparable brightness, which might affect the proper motion analysis, the continuum emission was removed by subtracting the properly scaled R-band frame, after accounting for a fractional pixel shift. On the resulting images, object detection was performed using the SEXTRACTOR code (Bertin & Arnouts 1996) at the 3 σ detection level. Thereby, the positions of HH 494 for the two epochs were obtained with coordinate uncertainties of 0″.12 – 0″.17. From these measurements, coordinate differences of ΔRA=(0″.89±0″.17 and ΔDEC=1″.29±0″.22 were derived. At a distance of 140 pc and for the epoch difference of 12.1478 yrs, these translate into a tangential velocity of 86±17 km s⁻¹. Since the inclination of the disk and outflow axes is very close to 90° (see Sect. 4.1 and Table4), this tangential velocity should be very close (within 2%) to the actual space velocity. The position angle of the proper motion of 325°.3±4°.8 (E of N) is consistent (within the uncertainties) with the orientation of the bipolar nebula and the molecular outflow (148 + 180 ⇒ 328 ± 1°, Sect. 3.1). This excellent agreement supports our hypothesis that HH 494 originates from a jet that was ejected from CB 26 - YSO 1.

With a distance of 140±20 pc and a very close to edge-on orientation, the projected separation of HH 494 from CB 26-YSO 1 of 6’.15 translates into a physical separation of ≈ 5 × 10⁴ au or 0.25 pc If we now assume that (i) this jet travelled at constant speed and (ii) that HH 494 has preserved the jet velocity, the corresponding dynamical time scale would 2860 ± 360 yrs. Since neither of these two assumptions, in particular the second one, may be correct, this dynamical time scale is only a very crude estimate of the age of the corresponding ejection event and the actual uncertainty may be much larger than our formal error. We discuss the context and implications of this estimate in Sect. 5.3.

References

All Tables

All Figures

Fig. 1 Overview of the CB 26 region. The left panel shows a wide-field optical “true-color” image, which is based on Hα (blue), [SII] (green), and I-band (red) images (Stecklum et al. 2004). The globules CB 25 and CB 26 as well as Herbig-Haro object HH494 are marked in the image. The zoomed-in right panel shows a NIR J-band image of the bipolar reflection nebula (color, 07 6 resolution), overlaid with contours of the 1.3 mm dust continuum emission from the disk (white contours at 3, 7.5, and 15 mJy beam−1). Green contours show the integrated 12CO(2–1) emission (0.5–12 km s−1) from the bipolar molecular outflow as presented in Launhardt et al. (2009). The red and blue arrows indicate the large-scale outflow orientation. Beam sizes are shown in respective colors in the lower right corner. The white dashed square marks the image section shown in Fig. 2. The reference position is 04h59m50.74s, 52°Q4′43.80″ (J2000).

In the text

Fig. 2 NIR J-band image of the bipolar nebula (see Sect. 2.4) overlaid with contours (3, 8, 13, and 18 mJy beam−1) of the 1.3 mm dust continuum emission from the disk (Zhang et al. 2021). Overlaid as blue and red contours, representing the rotation-related blue- and red-shifted sides (cf. Fig. 3), is the integrated 12CO(2–1) emission (0.5–12 km s−1) from the bipolar molecular outflow as seen in the new, higher-resolution CO(2–1) data, which we present and analyze in this paper. The 1.3 mm continuum and 12CO(2–1) synthesized beam sizes as well as the angular and linear scales are indicated in the lower right corner (cf. Fig. 1).

In the text

Fig. 3 Line spectra of CB 26, integrated over ≈1.5 beams around the disk center (top row), integrated intensity maps (middle row), and mean velocity fields (1st moment maps, bottom row). From left to right: 12CO (2–1) (0.5–4.8 and 7.0–12.0 km s−1), 11 CO (1–0) (3.4–4.9 and 6.2–7.7 km s−1), HCO+ (3–2) (1–10 km s−1), and HCO+ (1–0) and H13 CO+ (1–0) in blue (2.0–5.0 and 6.3–9.0 km s−1). Dashed vertical lines in the spectra (top) indicate the velocity range within which the spectra are most strongly affected by resolved-out emission and self-absorption from the extended envelope (see Figs. A.1–A.5). Total intensity contours in the maps start at 3σ. Overlaid as red contours in the middle row at 2.5, 7.5, and 15 mJy beam−1 is the 230 GHz dust continuum emission from the disk. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corners (dark gray: line, light gray: continuum). The bottom panels show the respective 1st moment maps with contours of the total intensity overlaid. A black triangle marks the location of the central star (and center of the disk). We note that, except for HCO+ (3–2), the envelope-dominated central velocity channels were masked out before the moment maps where generated.

In the text

Fig. 4 Spectral energy distribution of CB 26-YSO 1, adopted from Zhang et al. (2021), where we also list the fluxes in Table 1. Red squares and green dots with error bars show the observational data. The dashed black line shows best-fit model SED of Zhang et al. (2021). The dashed-dotted blue line represents the adopted contribution from free-free emission. The dotted red line corresponds to the sum flux of dust and free-free emission between 2.9 mm and 4.0 cm. The break in the SED at ~103 µm is due to the transition from total flux measurements at shorter wavelengths to interferometric data at longer wavelengths, which do not recover the total flux from the extended envelope emission.

In the text

Fig. 5 12CO(2–1) mean velocity field (1st moment map, color) and integrated intensity (black contours) of CB 26 at successively higher angular resolution. Left: as presented in Launhardt et al. (2009) at 1.″47 resolution. Middle: with the complete data set presented in this paper, but mapped with a uυ taper of radius 200 m and natural weighting (to restore a maximum of extended emission), resulting in an effective angular resolution of ≈1.″0. Right: same as middle, but mapped without tapering and with robust uυ weighting, resulting in an effective angular resolution of ≈0.″45. This image shows a smaller (zoomed) area than the other two maps, as indicated by the gray dashed rectangle in the middle panel. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corners of each panel. Overlaid as gray contours at 2.5, 7.5, and 15 mJy/beam is the 230 GHz dust continuum emission from the disk.

In the text

Fig. 6 Position-velocity diagrams along the plane of the disk of 12CO(2–1), 13CO(1–0), HCO+ (1–0), and HCO+ (3–2). Contours start at 3σ. Except for HCO+ (3–2), the envelope-dominated central velocity channels were masked out for the analysis (see also Figs. A.1, A.4 and A.5). Overlaid as red contours for comparison (with the y-axis being ∆y instead of υLSR) is the 230 GHz dust continuum emission from the disk (2.5, 7.5, and 15 mJy beam−1), rotated by 32°. Green lines show the Keplerian rotation curve for a central mass of 0.55 M⊙ and systemic velocity of υ0 = 5.95 km s−1 (only to guide the eye; this is not the fit to the data).

In the text

	Fig. 7 Marginal two-parameter χ2 maps for model grid parameters M*, idisk, and derived by comparing the synthetic maps of the 12CO(2–1) disk models with the observations (robust deconvolution). Dashed lines indicate the parameter values for the chosen “best-fit” disk model that was then subtracted from the data.
In the text

	Fig. 8 Hydrogen volume density, kinetic gas temperature, and CO volume density distribution of the best-fit disk model that was used to calculate the synthetic CO(2–1) channel maps of the disk emission (Fig. A.2). The white line in the middle panel marks the CO snow line at Tgas = 20 K.
In the text

Fig. 9 Mean velocity field (1st moment maps) of 12CO(2–1) in the velocity range ±6 km s−1 of the best-fit disk model (left; Table4 and Fig. A.2) and the residual CO emission after subtracting the disk model and masking the corrupted central velocity channels between ±1.2 km s−1 (right). Black solid contours show the total integrated intensity of the CO emission from the model disk. Magenta contours show the 230 GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Dashed contours show the total integrated intensity of the residual CO emission. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner (dark gray: CO, light gray: dust). The maps are rotated counterclockwise by 32° such that the disk and outflow axes are aligned with the z-axis.

In the text

Fig. 10 Channel maps of the12CO (2−1) emission in binned, 0.8 km s−1 wide channels around the velocity offsets Δυ from υ0 = 5.95 km s−1 indicated on top. Top row: observed total emission (see Fig. A.1); middle row: disk model (see Sect. 4.1 and Fig. A.2); bottom row: residuals, that is, observed total emission with disk model subtracted. Blue contours indicate negative velocity offsets, red contours positive ones. CO contours start at 15 mJy beam−1 (left and middle panels), and 40 mJy beam−1 (right panels) and are spaced by 15 mJy beam−1. Blue and red arrows in the lower left panel show the inclination orientation of the disk and outflow. Black contours show the 230 GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner of the lower left panel (dark gray: CO, light gray: dust). The underlying gray-scale image shows the NIR K-band image of the bipolar reflection nebula (see Fig. 1).

In the text

Fig. 11 Bottom left: total intensity map (gray-scale and black solid contours) of the residual12CO (2−1) emission after subtracting the disk model and masking the corrupted central velocity channels between ±1.2 km s−1, integrated over the velocity range ±6 km s−1. Red contours show the observed 230GHz dust continuum emission from the disk at 2.5, 7.5, and 15 mJy beam−1. Green contours show the total integrated intensity of the CO emission from the model disk. Synthesized FWHM beam sizes are shown as gray ellipses in the lower right corner (dark gray: CO, light gray: dust). Top: position-velocity diagrams (PVD) of 12CO (2−1) at three cuts parallel to the plane of the disk (a, b, c, as indicated by dashed lines in the total intensity map; gray-scale and black contours). The middle panel shows the PVD of the unmasked total observed CO emission along the disk mid-plane, while the upper and lower panels show disk wind PVDs of the residual CO emission (after subtracting the disk model) and with the central corrupted velocity channels masked. Green contours show the PVDs of the disk CO model along the disk mid-plane. Bottom right: same as above, but for the three vertical cuts labeled d, e, and f. Green contours show the PVDs of the disk CO model along the same vertical cuts.

In the text

Fig. 12 Outflow parameters derived from PVDs of disk-subtracted CO residuals at different vertical heights below and above the disk midplane: (a) X location r and FWHM width of the outflow cone walls. The errorbars denote the FWHM widths of the cone walls, not the uncertainty of the mean (which is smaller than the symbols). The light gray data points at z = ±40 au might be affected by imperfections in the disk model that was subtracted from the data. (b) Line-of-sight (LoS) velocity (relative to systemic velocity υ0) and FWHM width of the outflow cone walls. Red squares show the center velocity between left and right outflow cone. (c) Radius R of the outflow cone, with uncertainties derived from the FWHM widths of the outflow cone walls. (d) Rotation velocity and 1σ uncertainty of the outflow cone walls. The red solid and dotted lines show the expected rotation velocity and uncertainty for angular momentum conservation of a wind that is launched from the disk at a single radius Rconn = 35 ± 10 au with a Keplerian rotation velocity of υK = 3.7 ± 0.3 km s−1. (e) Specific angular momentum j and 1σ uncertainty. (f) Half-opening angle 9 and 1σ uncertainty.

In the text

	Fig. 13 Purely empirical schematic cartoon illustrating the main features and geomoetry of the disk and disk wind derived in this paper.
In the text

Fig. A.1 12CO(2−1) channel maps of CB 26, obtained with PdBI in 2005 and 2009, rotated counterclockwise by 32°. Contour levels start at 15 mJy/beam (2σ r.m.s., see Table 1). Red contours mark the 1.2mm dust continuum emission from the disk (3 and 15 mJy/beam. The reference position is α2000 = 04h59m50.74s, δ2000 = 52°04′43.80″. The 12CO synthesized beam size is indicated as the gray ellipse in the lower right corner of the first channel map. The channel number and mean Δυ = υLSR _ 5.95 (in km s−1) are indicated in the top left and right corners of the maps, respectively. Maps in the central velocity channels between Δυ ≈ ±1.2 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope.

In the text

	Fig. A.2 Same as Fig. A.1, but for the modeled 12CO (2−1) emission of the best-fit disk model for CB 26 (see Table 4). Contour levels start at 10 mJy/beam.
In the text

	Fig. A.3 Same as Fig. A.1, but for the residual 12CO(2−1) emission from CB 26 after subtracting the best-fit disk model (Fig. A.2). Contour levels start at 15 mJy/beam.
In the text

Fig. A.4 Same as Fig. A.1, but for 13CO(1−0), obtained with OVRO in 2001. Contour levels start at 70 mJy/beam (2 σ r.m.s., see Table 1). Maps in the central velocity channels between ~ ±1.2 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope. Maps in the central velocity channels between Δυ ≈ −1.0 and 0.5 km s−1are corrupted by resolved-out emission and self-absorption from the extended envelope.

In the text

	Fig. A.5 Same as Fig. A.1, but for HCO+ (1−0), obtained with PdBI in 2005. Contour levels start at 15 mJy/beam (~ 2 σ r.m.s., see Table 1). Maps in the central velocity channels between Δυ ≈ −0.8 and 0.4 km s−1 are corrupted by resolved-out emission and self-absorption from the extended envelope.
In the text

	Fig. A.6 Same as Fig. A.1, but for HCO+ (3−2), obtained with the SMA in 2006. Contour levels start at 300 mJy/beam (2 σ r.m.s., see Table 1).
In the text