The Perseus OB2 Superbubble and the Taurus-Auriga-California-Perseus Molecular Cloud Loop


 Located at a distance of about 300 pc, Perseus OB2 (or Per~OB2 for short) is one of the major OB associations in the solar vicinity\cite{Zeeuw99,Belikov2002}, which has blown a supershell with a diameter of about 15 degree seen in the atomic hydrogen line surveys\cite{Sancisi1974,Heiles1984,Hartmann1997}. It was 
long considered that stellar feedback from the Per~OB2 association had formed a superbubble that swept up the surrounding interstellar medium into the observed 
supershell\cite{Bally2008}. Here we report the three-dimensional structure of the Per~OB2 superbubble, based on wide-field atomic hydrogen and molecular gas (traced by CO) surveys. The measured diameter of the superbubble is roughly 330 pc. Multiple atomic hydrogen shells/loops with expansion velocities of about 10 km/s are revealed in the superbubble, suggesting a complicated evolution history of the superbubble. Furthermore, the inspections of the morphology, kinematics and timescale of the Taurus-Auriga, California, and Perseus molecular clouds shows that the cloud complex is a super molecular cloud loop circling around and co-expanding with the Per~OB2 superbubble. We conclude that the Taurus-Auriga-California-Perseus loop, the largest star-forming molecular cloud complex in the solar neighborhood, is formed from the feedback of the Per~OB2 superbubble.

two large curved filaments are detected to the south of the cavity, which emerge from eastern cavity wall and extend to the west. In the last velocity range (red range; see Fig. 1c), these two large filaments appear to merge into one shell to the southeast of the cavity. In the mean velocity field shown in Fig. 1d, these different velocity components together delineate a superbubble, which is extending in the north-south direction. Figure 2 shows the HI position-velocity (PV) diagrams along different routings across the superbubble (see Fig. 1 for the PV routings). The HI emission of the superbubble is mainly within the velocity range from -6 to +11 km s −1 (i.e., the velocity range shown in Figure 1). Cavity-like PV patterns are detected in all the routings, indicating that the superbubble is expanding. Furthermore, spur-like protrusions, spatially coincident with the dense walls/filaments seen in the HI intensity images (see Figure 1), are also detected along the PV diagrams (see Figure 2). As seen in the diagrams, the cavity-like patterns, as well as the protrusions, could be well fitted by ellipses, with the position and velocity radii representing the projected radii and expansion velocities, respectively. Based on the PV diagrams and intensity images, two large HI loops, surrounding the central cavity, can be outlined in the velocity range of [-0.5, +4.5] km s −1 (the green range; see  with each other (∼ 10 ± 1 km s −1 ). We note that much fainter protrusions are also seen in the PV diagrams (see Figure 2), which are spatially coincident with the faint concentric shells detected in the intensity images (see HI velocity channel maps and Figure 1). Further high sensitivity and angular resolution HI observations are needed to study these diffuse structures.
Giant molecular clouds (GMCs), generally with size > 100 pc and mass > 10 5 M , are the primary reservoirs of cold, dense, and star-forming molecular gas in the Milky Way 10 and nearby galaxies 11 . However, the formation mechanisms of GMCs are not well understood yet 12,13 . The feedback from OB associations can drive converging flows to accumulate atomic hydrogen gas (i.e., super-shells and -bubbles) to form molecular clouds [14][15][16] , which has been observed in a few distant supershells in the Galaxy [17][18][19][20] . For solar neighborhood conditions, nevertheless, it was generally considered that the maximum mass of molecular clouds, formed by stellar feedback, is limited to be a few times of 10 4 M (ref. 12 ).
The Taurus-Auriga, California, and Perseus Clouds, with a total gas mass of ∼ 2 × 10 5 M , is the largest star-forming molecular cloud complex in the solar vicinity 8,21 . Surrounding the Per OB2 association, these well-known clouds are located at different distances along the line-of-sight 22 .
The Taurus Cloud, at a distance of ∼140 pc, is in the foreground of Per OB2, while the California Cloud, at a distance of ∼ 470 pc, is in the background of Per OB2. The distance of the Perseus Cloud is estimated to be ∼ 290 pc, and it is thus immediately at the front edge of the association.
It is of importance to study the relationship between these molecular clouds and the Per OB2 superbubble.
As seen in the CO velocity channel maps (Supplementary Figure 2), the molecular gas observed in the Taurus-Auriga, California, and Perseus Clouds is spatially co-moving with the HI shells of the Per OB2 superbubble -either lying along the dense walls of the cavity (for the Cal-ifornia and Perseus Clouds; see Supplementary Figure 3) or well embedded within them (for the Taurus-Auriga Cloud). This kind of CO-HI co-moving morphology is generally regarded as a key indicator for the molecular clouds forming as coherent parts of atomic shells 20 .
In the CO velocity-integrated intensity images, the Taurus-Auriga, California, and Perseus Clouds consist of a super molecular cloud loop (see Figure 3). We extract an PV diagram along the ridge of the loop to inspect its kinematics (see Figure 4). The PV diagram shows that the molecular cloud loop is expanding with a velocity of ∼ 10 km s −1 , with respect to a systemic velocity of ∼ 1 km s −1 . In this expanding loop, the California Cloud is blue-shifted, the Taurus Cloud is redshifted, while the Perseus Cloud appears at the tangential portion of the loop. This expansion picture is consistent with the distances of these clouds regarding to the Per OB2 association. A similar PV diagram is extracted along the HI Loop I (see Figure 4). The comparison between the CO and HI PV diagrams indicates that the Taurus-Auriga-California-Perseus loop is expanding together with the Per OB2 superbubble.
The CO velocity fields of the molecular clouds provide further evidence for physical relation between the molecular cloud loop and the Per OB2 superbubble (see Supplementary Figure 5).
As seen in the velocity field images, systematic velocity gradient is observed along the Perseus Cloud (see also the CO PV diagram in Figure 4). Furthermore, systematic velocity gradients are The kinetic age of an expanding bubble can be derived by where R bubble is the radius of the bubble in pc and V exp is the expansion velocity in km s −1 (ref. 14 ).
Adopting a radius of 165 pc and an expansion velocity of 10 km s −1 , the estimated kinetic age of the Per OB2 superbubble is roughly 10 Myr. This is consistent with the age estimated to the Per OB2 association (≤ 15 Myr; refs. 2, 6, 23 ), comparable with the average lifetime of the GMCs in nearby Galaxies (∼ 10 Myr; ref. 24 ), but much longer than the typical lifetime of molecular clouds in the solar neighborhood (a few Myr; ref. 15 ). It indicates that the Taurus-Auriga-California-Perseus loop is not a pre-existing molecular cloud complex around the Per OB2 superbubble.
Using the standard method 25 , the energy E E needed to open an interstellar bubble could be derived by where n gas is the density of ambient gas in cm −3 , R bubble is the radius of the bubble in pc, and V exp is the expansion velocity in km s −1 . Based on the HI line observations, the density n gas of the interstellar gas surrounding Per OB2 is estimated to be ∼ 1 cm −3 (see Methods). Adopting R bubble = 165 pc and V exp = 10 km s −1 , the estimated E E is ∼ 1.1 × 10 52 erg for opening the Per OB2 superbubble, which is much larger than the typical energy provided by a single supernova (SN) explosion (E SN ∼ 10 51 erg). This is not surprised because superbubbles could be produced by combined stellar feedbacks from a group of massive stars 26 and/or repeated stellar feedbacks from multiple star formation episodes in the OB associations 14 . Numerical studies show that superbubbles retain only ≤ 10% of the energy of an SN explosion 28 . On the other hand, ionizing radiation and stellar wind from a massive star can each provide a few 10 50 erg during the H II and mainsequence phases, respectively 14,27 . Assuming that a massive star releases in total 10 51 erg energy into a bubble during its whole lifetime (including ionizing radiation, stellar wind and eventually an SN), the formation of the Per OB2 superbubble should require roughly 11 OB stars. There are several known massive stars in the Per OB2 association (refs. 1, 2 ; see Figure 1a), such as 38 Per (B1III), ζ Per (B1Ib), 40 Per (B0.5V), and ξ Per (O7IIIe). We note that evidences of SN explosions were also found in Per OB2 (see Methods).
Based on the evidences from morphology, kinematics and timescale, we conclude that the Taurus-Auriga-California-Perseus molecular cloud loop is formed by the feedback of the Per OB2 superbubble. Compared with other well-studied superbubbles in the solar vicinity (see Methods), this super molecular cloud loop is unique, providing us a crucial sample to study the connection between HI superbubbles and molecular clouds. Furthermore, the multiple shells observed in the Per OB2 superbubble implies a complicated evolution history, which could be produced by repeated stellar feedback in the OB association. Suggested by the structures of the central cavity, HI Loops I and II, we consider that there are three star formation epochs in the Per OB2 association.
Given a kinetic age of 10 Myr estimated above, the mean star formation epoch is roughly 3 Myr, which favors the scenario of rapid star formation 15,29 . This is also consistent with the observations toward Per OB, in which three stellar groups with different ages (∼ 1 Myr, 1-5 Myr, and 5-10 Myr, respectively) are found 23 . The well-studied young stellar objects in the Taurus Cloud (with an age of ∼ 1-2 Myr; ref. 15 ), as well as young clusters found in the California Cloud (the LkHα 101 cluster with an age of ∼ 1 Myr; ref. 30 ) and Perseus Cloud (e.g., the IC 348 cluster with an age of 1-3 Myr; ref. 6,15 ) are likely formed in the last star formation epoch.

Methods
Multi-wavelength observational data. We retrieved the HI line data from the HI4PI (ref. 7  The gas density of the Per OB2 superbubble. By integrating HI spectroscopic data in velocity, one can infer the N HI column densities by where T B (v) is the brightness temperature profile of the HI gas 40 . And then, the density n gas can be estimated from the radius and column density of the bubble by where N shell is the column density measured at the shell of the bubble 41 . The measured column density of the shells toward the Per OB2 superbubble is about 100 K km s −1 or 1.823 × 10 20 cm −2 , and the estimated volume density is ∼ 1 cm −3 (adopting a radius of 165 pc), which is consistent with the mean volume density observed in the solar neighborhood 42 .
Supernovae in the Per OB2 association. The O7 III star ξ Per, the illuminating source of the California Nebula, is a high-velocity star running-away from the Per OB2 association (see Figure 1a). It was long suggested that ξ Per was expelled by an SN explosion of an even more massive companion 43 . The high-mass X-ray binary, X Per, provides the evidence for the other SN explosion in the Per OB2 association. More recently, a middle-age (∼ 0.54 Myr) fast-moving pulsar PSR J0357+3205 was discovered in the Per OB2 region 44 (see Figure 1a). According to its proper IRAS survey showed a super (∼ 7.5 • in radius or ∼ 60 pc at 450 pc) ring-like structure around the Vela OB2 association 54 , which is named as IRAS Vela Shell (IVS). The IVS is mainly seen in the dust and radio continuum, as well as neutral hydrogen 55 . A large molecular cloud complex along the Galactic plane, the so-called Vela Molecular Ridge (VMR), was found to the (Galactic) north of the IVS 56,57 . Further observations suggested that parts of the VMR, at a distance of ∼ 700 ± 200 pc, are associated with the IVS 58 .
The Cygnus superbubble is an extended 18 • × 13 • strong X-ray emission region, approximately centered on the Cygnus OB2 association 59,60 . However, it is still unclear that the superbubble is a real shell 61 or a structure resulted from a projection effect due to the emission from several separate features along the line of sight 60,62 . In the wide-field HI and CO observations, no bubble-like structure was found to be spatially coincident with the Cygnus X-ray superbubble 62