Solar flares and coronal mass ejections (CMEs) are the most violent phenomena in the solar atmosphere. They both have the capacity to cause hazardous space weather in the near-earth environment. Therefore, studying their origin is of fundamental importance in space weather forecasting. Except the ones associated with quiescent filament eruptions, most of the flares and CMEs originate from solar active regions (ARs). ARs appear as strong magnetic flux concentrations on the photosphere, which are believed to be formed through the magnetic flux emergence from the solar interior. The emergence of a flux tube may form the simplest bipolar region (Schmieder et al., 2014). The strong magnetic flux concentrations are manifested as regions containing dark sunspots in white-light emission. The ability of the ARs to produce large eruptions varies greatly as suggested by the observations (Chen et al., 2011; Wang et al., 2011). To distinguish the productive and inert ARs, great efforts have been carried out in characterizing the magnetic properties of the ARs from the routine observations of the photospheric magnetic field, since the direct measurement of coronal magnetic field is not available until now (Falconer et al., 2002; Leka and Barnes, 2003; Falconer et al., 2006; Georgoulis and Rust, 2007; Leka and Barnes, 2007; Falconer et al., 2008; Bobra and Couvidat, 2015). It is found that the ARs capable of producing large eruptions usually hold a more complex configuration (Toriumi et al., 2016, and references therein), such as the δ sunspot, where two umbrae of opposite-sign polarities share a common penumbra (Künzel, 1960). This kind of complex configuration indicates a higher degree of non-potentiality; therefore, the parameters characterizing the non-potentiality of the regions, such as magnetic shear, electric current density, and current helicity, tend to be higher in the region (Sun et al., 2015; Liu et al., 2016). It should be noted that the higher degree of non-potentiality, i.e., enough magnetic free energy, is only a necessary condition to power large eruptions. The exact triggering of the eruptions might involve a more specific mechanism such as magnetohydrodynamics (MHD) instabilities (Török et al., 2004; Kliem and Török, 2006) or magnetic reconnection (Antiochos et al., 1999; Moore et al., 2001).

When looking into the eruption-producing ARs, the strong-field, high-gradient sheared polarity inversion line (PIL), which is formed between opposite-sign polarities that are located in close proximity, is found to be a common source of large eruptions (Schrijver, 2007; Schrijver, 2009). Shearing motion, sunspot rotation, and converging motions are frequently observed near the sheared PIL (Zhang et al., 2007; Li and Zhang, 2009; Green et al., 2011; Yan et al., 2015). The former two are thought to be able to shear and twist the field lines, injecting magnetic free energy and magnetic helicity to the coronal field. The latter is believed to be able to bring the opposite-sign magnetic field together, leading to flux cancellation (mild magnetic reconnection near the photosphere) which may form the magnetic flux rope and drive the subsequent eruptions as suggested by the flux cancellation model (van Ballegooijen and Martens, 1989).

Considering the entire evolution of an AR from its emergence to decay, the eruptions may occur at any stage (van Driel-Gesztelyi and Green, 2015, and references therein). Nevertheless, observations suggest that compared to the ARs entering the decaying phase, the ARs which are still emerging and evolving tend to produce more violent eruptions (Schrijver, 2009).

In general, it is suggested by the observations that the sheared PIL accompanied by the shearing motion and flux cancellation in emerging ARs is a place prone to major solar eruptions. The physics behind the phenomena is not entirely clear. Recently, a scenario named “collisional shearing,” is proposed to explain the origin of large solar eruptions from emerging ARs (Chintzoglou et al., 2019). Through studying the evolution of two well-observed highly productive ARs, the authors suggest that these are the relative motions between non-conjugated opposite-sign polarities leading to the formation of the sheared PIL and continuous shearing and flux cancellation around the PIL which are responsible for the series of eruptions. The relative motion between non-conjugated polarities resulted from the proper separation of conjugated polarities when a flux tube emerges from the dense interior of the Sun into the tenuous corona. Moreover, a statistical research study on 19 ARs which emerged and produced at least one major eruption on the visible solar disk also suggests that the collisional shearing between non-conjugated polarities may be a common process at the source locations of large eruptions in emerging ARs (Liu et al., 2021). Those works stressed the role of interaction between non-conjugated polarities of different bipoles in driving eruptions in emerging ARs.

In observations, the emerging ARs evolve rapidly, so the original conjugation of polarities may be blurred by multiple episodes of flux cancellation and magnetic reconnection later. Therefore, the most reliable way to determine the conjugation of polarities is to track the AR from the very beginning of its emergence to inspect the typical features between emerging conjugated polarities such as small moving dipoles on the photosphere (Strous and Zwaan, 1999) and arch filaments in chromospheric observation. However, the limited observations of a photospheric magnetic field, which are mainly for the visible solar disk at present, restrict larger statistical research on the “collisional shearing” process. On the other hand, Toriumi et al. (2016) performed a statistical research study on 29 ARs that produced 51 flares larger than GOES M5.0 class in solar cycle 24, mainly focusing on the pre-flare magnetic properties of the ARs rather than their evolution. In addition to the classification of magnetic patterns of δ sunspots proposed by Zirin and Margaret (1987), the authors classified the productive ARs into four groups, including the “spot-spot” group, “spot-satellite” group, “quadrupole” group, and the “inter-AR” group. For the latter three groups, different bipoles are involved in forming the sheared PIL where the eruptions occur, so that the bipole–bipole interaction plays a role in driving the eruption. For the former group which accounts for about one-third of the sample, the sheared PIL is formed between two major polarities or two groups of opposite-sign sunspots; it is not clear if the bipole–bipole interaction and thus “collisional shearing” exist and play a role in driving eruptions. Further study is needed.

In this work, we searched the sample in Toriumi et al. (2016) and found an AR (NOAA AR 12036) of spot–spot configuration which emerged and produced an M7.3-class eruptive flare on the visible solar disk. We tracked the evolution of the AR to see what caused it to produce the large eruption.

We used the photospheric vector magnetograms provided by the Helioseismic and Magnetic Imager (HMI, Scherrer et al. (2012)) onboard the Solar Dynamics Observatory [SDO, Pesnell et al. (2012)] to investigate the evolution of the magnetic properties of the AR. The data have a plate scale of 0.′′5 and a temporal cadence of 720 s. Here, a data segment which is de-projected from the native helioprojective Cartesian coordinate to a cylindrical equal area (CEA) coordinate is used (Bobra et al., 2014). HMI also provides the photospheric continuum intensity maps, which is used here to examine the evolution of the sunspots in the AR.

We used the ultraviolet (UV) and extreme ultraviolet (EUV) images provided by the Atmospheric Imaging Assembly [AIA, Lemen et al. (2012)] onboard the SDO to examine the coronal evolution of the AR and the eruption details of the M7.3-class flare. The data have a plate scale of 0.′′6 and a temporal cadence of up to 12 s.

To determine the connectivity between conjugated polarities in the early emergence phase more accurately, we performed a potential field extrapolation (Alissandrakis, 1981) to the normal component of the photospheric magnetic field (B _z ) and analyzed the extrapolated field lines.

We analyzed the motions and flows in the ARs in two ways. The first way is to track the macroscopic motion of the large polarities which formed the main PIL where the eruption occurred. The second way is to analyze the microscopic flow fields of the region. For the first way, we first detected the local peaks in each polarity on each magnetogram of B _z , and then, for each polarity, we calculated a flux-weighted centroid in a circle which took the detected peak as the center and a radius of 5 mm (Chintzoglou et al., 2019; Liu et al., 2021). For the second way, we used a Fourier-based local correlation tracking (FLCT) method (Welsch et al., 2004) to calculate the horizontal velocity field of the intensity features on the B _z magnetograms. The velocity field is suggested to be able to affect the photospheric footpoints of the coronal field.

We extracted the source PIL of the eruption by drawing a contour line of B _z = 0 on the smoothed B _z magnetograms. To further assess the degree of collision along the PIL, we calculated the gradient of B _z across the PIL through ▽ B z = ∂ B z ∂ x 2 + ∂ B z ∂ y 2 .

In this work, we tracked the evolution of the NOAA AR 12036 which emerged and produced an M7.3-class flare on the visible solar disk. It held a spot–spot-type configuration before the flare, i.e., a monolithic PIL was formed between the group of positive polarities and the group of negative polarities, from which the M7.3-class flare occurred. The observations suggested that the AR was formed by three bipoles that emerged sequentially. The two bipoles emerging earlier were located parallel. The third bipole emerged in between the two polarities of the second bipole. Its negative polarity was mixed with the negative polarity of the second bipole. In the early phase when the flux emergence was fast, the B _z gradient across the PIL remained small, and no collision or large eruption occurred.

Later, although the slow flux emergence of the entire AR continued, the positive polarity P3 of the third bipole started to show signs of decay; the sunspot in P3 lost its coherence gradually; and small magnetic features were found stripped away from the polarity, heading toward the negative polarity N1 of the bipole emerging first. The increasing area and decreasing mean B _z of the polarity P3 further supported that P3 was undergoing a decaying process. The small moving magnetic features brought the polarities P3 and N1 into contact about 7.5 h prior to the flare, i.e., the collision occurred between the polarities since then. The B _z gradient at the interface between P3 and N1 increased with the converging and collision processes.

Accompanied by the collision between P3 and N1, two precursor flarings occurred in 5 h before the flare, both of which ignited from around the interface region between P3 and N1. They all formed post-flare loops above the interface region and ignited a continuous S-shaped hot channel above the loops. In the very early phase of the M7.3-class flare, another small-scale flaring occurred. It was a jet process occurring at a possible null point located lower than the hot channel and may disturb the hot channel to erupt later. The hot channel eruption induced the main phase flaring of the flare and a CME.

Combining the aforementioned results, we found that it was the collision between the non-conjugated polarities in the decay phase of one polarity which created a location favorable of cancellation and flaring. The collision brought the footpoints of different sets of sheared loops belonging to different bipoles into close proximity, enabling the tether-cutting-type reconnection to occur between them. The resultant pre-flare flarings and possible cancellation helped to form and enhance the flux rope which was manifested as an S-shaped hot channel. The eruption of the hot channel was directly triggered by a low-lying jet process, which is similar to the case reported by Liu et al. (2015), which showed a jet process triggering a CME. It is important to note that accompanied by the collision, the shearing process is also needed to create the helical field lines of the flux rope. Similar to the rapidly emerging non-bipolar ARs (Chintzoglou et al., 2019), the shearing process in the spot–spot-type AR here was also induced by the proper motions between non-conjugated polarities. The proper motions of the two non-conjugated polarities before the collision had similar direction but different magnitudes, so the shearing gradually occurred between them.

The “collisional shearing” reported by Chintzoglou et al. (2019) and Liu et al. (2021) mainly occurred between the non-conjugated polarities in emerging non-bipolar ARs and resulted from the relative motions of non-conjugated polarities. Those motions were indeed caused by the proper separation of conjugated polarities. In the NOAA AR 12036 studied here, the arrangement of the bipoles, which was an overall bipolar configuration, did not create a condition favorable for contact and collision between non-conjugated polarities in the fast emergence phase; therefore, no large eruption occurred in this phase. Later on, although slow flux emergence continued, one of the main positive polarities started to decay, leading to collision between the non-conjugated polarities eventually. The two precursor flarings which helped to form the hot channel that erupted later were the result of collision.

Although the photospheric collision plays an important role in most ARs which produce the large solar eruption, it is not an absolute necessary condition for eruption. In a few events like the “inter-AR”-type cases by Toriumi et al. (2016), who account about 7% of all their samples, the eruption occurred above a PIL without close contact or strong collision between opposite-sign polarities. They may result from the process occurring in the corona, the details behind which need further study.

The results in this work may explain why there is no single bipolar AR in the statistical research of “collisional shearing” (Liu et al., 2021), though one should still be aware that their sample size is limited. For the AR holding a non-bipolar configuration, it contains more than one bipole when emerging. The arrangement of the bipoles may tend toward collisional shearing occurring between non-conjugated opposite-sign polarities. The flux emergence drives the collisional shearing to continue, eventually leading to eruptions. For the “spot–spot”-type AR we studied here, the bipoles were roughly parallel to each other in the fast emergence phase, the arrangement of which did not create a condition favorable for collision. Although the shearing gradually occurred between the non-conjugated polarities due to their proper motions, no large eruption occurred until the collision occurred as a result of the polarity decay. It is important to note that we have only one case here; it is possible that in some other “spot–spot”-type ARs, the “collisional shearing” process may not even occur. The rotation and shearing of the sunspots driven by another process may play a major role in producing large eruptions in those ARs as suggested by the previous research (Yan and Qu, 2007; Zhang et al., 2007; Yan et al., 2008; Li and Zhang, 2009; Yan et al., 2009). Statistical research is needed to check if “collisional shearing” is common in the “spot–spot”-type ARs. On the other hand, even in the “spot–spot”-type AR here, the collision occurred between non-conjugated polarities at the source PIL of the eruption and not at the self PIL between conjugated polarities. Statistical research considering the different evolution stages of the ARs is also needed to check whether this is common.