The solar energetic particle (SEP) events observed first (Forbush 1946) were the largest and most energetic examples we know, where GeV protons produce a shower of particles cascading through the atmosphere to ground level in excess of that from the galactic cosmic rays (GCRs). However, these ground-level events (GLEs) provided little information on the composition of the incoming beam.

Measurement of multiple element abundances in SEPs began when Fichtel and Guss (1961) used nuclear emulsion detectors on a 4-min sounding-rocket flight from Ft. Churchill, Manitoba to observe a sampling of elements with 6 ≤ Z ≤ 16. The principle elements up to the Fe abundance peak were observed in the next solar cycle by Bertsch et al. (1969) using the same technique. Resolution of He isotopes and the continuous time coverage needed to observe smaller events would only begin when detector telescopes were flown on satellites.

The first clearly-enhanced abundance ratio of ³He/⁴He = (2.1 ± 0.4) × 10⁻² was reported by Hsieh and Simpson (1970), perhaps 50 times the value seen in the corona or solar wind, although differences in the spectra of the He isotopes were noted here.

Nearly every scientist involved in the early study of SEPs had previous experience with GCRs. After acceleration by shock waves at supernovae, GCRs spend ∼10⁷ years colliding with interstellar H to produce secondary ²H, ³He, and isotopes of Li, Be, and B. Thus the observation of ³He/⁴He of 2% by Hsieh and Simpson (1970) was immediately misinterpreted as evidence that the SEPs had traversed enough material to fragment some ⁴He into ³He, just like the GCRs. However, it was soon found that there were many events like that seen by Serlemitsos and Balasubrahmanyan (1975) with ³He/⁴He = 1.52 ± 0.10 but ³He/²H > 300. How could there be more ³He than ⁴He and yet no ²H? Such abundances were definitely not compatible with fragmentation. Subsequently, there were limits established on Be/O and B/O in SEP events that were found to be <2 × 10⁻⁴ (e.g., McGuire et al., 1979; Cook et al., 1984). Thus these ³He-rich events were not just an accident of fragmentation; they must involve a completely new resonance phenomenon.

Figure 1 shows a sample of ³He-rich events as we see them above 2 MeV amu⁻¹. As event intensities increase it becomes possible to see rarer ion species. Event 1 has only pre-event background levels of H, O, and Fe and ⁴He barely appears. ³He/⁴He varies greatly in events and vs energy, often peaking in the region 1–10 MeV amu⁻¹ (Mason, 2007), while Fe/O is much more stable once events are large enough to provide a measurable sample, so Fe/O is often used to define “impulsive” events (e.g., Reames et al., 2014a). As discussed below, very high-Z elements begin to appear in the larger impulsive events and ³He fluence may be limited by depletion of ³He ions in the source volume.

Reconnection of the solar magnetic field drives nearly all solar activity, and flares provide a ubiquitous visible marker of magnetically trapped heating from that reconnection. Perhaps it is not surprising that the earliest evidence of GLEs was associated with obvious, bright flares, but this association was taken much too literally.

Radio data were the first to distinguish two acceleration mechanisms for SEPs in space (Wild et al., 1963), type III bursts that associate with electrons streaming from open magnetic reconnection in impulsive SEP events, and type II bursts that associate with shock acceleration and large gradual SEP events, but, unfortunately, the radio evidence was largely ignored in early SEP history.

After CMEs were identified and their observation became common, Kahler et al. (1984) found a 96% correlation between large SEP events and fast, wide CMEs. Shock waves driven by the CMEs could explain the extremely broad longitude span of the events. Mason et al. (1984) found that the minimal rigidity dependences of abundance variations across longitude were inconsistent with a point source origin and they discussed large-scale shock acceleration that they labelled LSSA. The importance of CMEs and shocks was not taken seriously by some flare enthusiasts until Gosling (1993, 1994) review article entitled “The solar flare myth” was published. This article was then found to “wage an assault on the last 30 years of solar-flare research” (Zirin, 1994) even though most published “solar-flare research” ignored SEPs entirely. However, the importance of CMEs and the existence of two mechanisms of SEP acceleration began to be recognized (Reames, 1988, 1995b, 1999, 2013, 2015, 2020, 2021a, 2021b; Zank et al., 2000, 2007; Kahler, 2001; Cliver et al., 2004; Lee, 2005; Gopalswamy et al., 2012; Lee et al., 2012; Mewaldt et al., 2012; Desai and Giacalone, 2016; Kouloumvakos et al., 2019).

For a brief interval, distinguishing impulsive and gradual SEP events seemed simple. Impulsive events were ³He-rich and gradual events were not. Then Mason et al. (1999) found a small but significant enhancement of ³He in a large SEP event that would be called gradual in all other respects. It became clear that shock waves in gradual SEP events could reaccelerate residual suprathermal ions left over from previous impulsive SEP events. In fact these pre-accelerated ions might be preferred in some cases, as in quasi-perpendicular shocks when ions needed to overtake the shock from downstream (Tylka et al., 2001, 2005; Tylka and Lee, 2006). In fact, it became evident that large pools of ³He-rich, Fe-rich suprathermal ions were extremely common, and available for shocks to traverse (Richardson et al., 1990; Desai et al., 2003; Wiedenbeck et al., 2008; Bučík et al., 2014, 2015; Chen et al., 2015). Whenever there was no large SEP event in progress the default suprathermal ion abundances below ∼1 MeV amu⁻¹ seemed to be ³He-rich and Fe-rich, suggesting a large number of small unresolved jets (nanojets?) could generate SEPs faster than the solar wind could sweep them away. Thus, for 24% of gradual events, the Z ≥ 2 elements are dominated by reaccelerated ions from impulsive events, called SEP3 events (Reames, 2020), and in 69% of gradual events the shock predominantly accelerates ions, even Z > 2 ions, from the ambient coronal plasma (SEP4 events; Reames, 2020a; Reames, 2021a; Reames, 2021b).

Impulsive SEP events were small and difficult to associate with coronal features, but Kahler et al. (2001) were able to associate several of the larger impulsive SEP events with narrow CMEs. Figure 4 shows a narrow (54⁰), fast (1,360 km s⁻¹) CME from the impulsive event of May 1, 2000 SEP, seen by the Solar and Heliospheric Observatory (SOHO; https://sohowww.nascom.nasa.gov/), for which SEPs were shown in the central panel of Figure 2. Such CMEs had been associated with type III bursts and solar jets (Shimojo and Shibata, 2000). Unlike flares, jets involve magnetic reconnection on open field lines so the SEPs (and the CMEs) easily escape. Tracking of impulsive SEPs back to the Sun led to jets, often on the boundary between active regions and coronal holes (Wang et al., 2006; Nitta et al., 2006, 2015). However, onset times are poorly defined in small SEP events, making associations with type III bursts, etc. more difficult. Reconnection is often triggered by large-scale waves moving across the corona (Bučík et al., 2016). It is now possible to consider the nature of the associated jets directly for many ³He-rich events (Bučík et al., 2018a, 2018b, 2021; see review; Bučík, 2020). It has even been possible to compare the temperatures derived from the extreme ultraviolet (EUV) images of coronal source regions of 24 solar jets associated with ³He-rich events (Bučík et al., 2021) with the abundance-derived temperatures from SEPs (Reames et al., 2014b).

However, impulsive SEP events are not necessarily derived purely from magnetic reconnection; the CMEs from these jets are often fast enough to drive shock waves which can reaccelerate SEPs from the reconnection as is certainly the case for the 1,360 km s⁻¹ CME in the May 1, 2000 event shown in Figure 4. It may be possible to distinguish the pure (SEP1) events from those reaccelerated by a shock (SEP2) from the proton excess produced in the latter events as shown in Figure 3.

A sketch showing a jet formed by newly emerging magnetic field is shown in Figure 5. When the emerging field has opposite polarity from that of the preexisting field, reconnection takes place, not at a single point but in a series of “islands.” In realistic jets, the opposing fields rarely exactly cancel, but rather leave an out-of-plane, residual “guide field.” Particle acceleration occurs as particles, mainly electrons (Arnold et al., 2021), are Fermi-accelerated as they pitch-angle scatter in the evolving fields. As the SEPs and CME plasma are ejected on open field lines at the upper right in Figure 5, newly closing loops capture some SEPs in the lower left region labeled “flare” where they deposit their energy as heat. While much of the jet, including the SEPs, retain temperatures of 2–3 MK and emit EUV (Reames et al., 2014b; Bučík et al., 2021), the “flare” region with its captured SEP energy heats to 10–20 MK and emits X-rays. These X-rays often provide the source location. It is important to realize that the presence of the X-rays does not define the nature of the SEPs in space; both environments coexist since the reconnection that opens some field lines closes others. Some day someone will be able to provide us with an overlay of X-ray and EUV images that will map details of these source regions. As reconnection events become larger and more complex, both X-ray and EUV regions increase, contributing to a “big flare syndrome” (Kahler 1982), i.e., a misleading correlation between X-rays and the SEPs in space.

What about solar flares? There is evidence from measurements of Doppler-broadened γ-ray lines that the ions accelerated in large solar flares are ³He-rich (Mandzhavidze et al., 1999; Murphy et al., 2016) and Fe-rich (Murphy et al., 1991), just like the impulsive SEPs from jets that we see in space. The SEPs accelerated on closed loops, dominated by electrons, soon scatter into the loss cone and plunge into the denser corona below, scattering against ions to produce X-ray bremstrahlung, and heating the plasma which expands back up into the loops creating a hot (>10 MK), bright flare. SEPs from jets are not hot (∼2–3 MK; Reames et al., 2014b; Bučík et al., 2021) because the SEPs and the CME plasma escape. Flares exist precisely because their energy is magnetically trapped and can only escape as heat, light, or neutral particles. Reconnection of closed fields with other closed fields cannot produce open fields, except when those fields are eventually carried outward by a CME.

Shock waves, driven by fast, wide CMEs are the basis of gradual SEP events. However, the narrow (<60⁰) CMEs emitted from solar jets are also fast enough to drive significant shocks, such as the 1,360 km s⁻¹, 54⁰ width CME in the May 1, 2000 events discussed above. Yet, often, these fast, narrow CMEs from jets lack type II radio bursts that can signify shock acceleration, and the CME’s contribution to SEP acceleration may be questioned (e.g., Bronarska et al., 2018; Kahler et al., 2019). At quasi-parallel shock waves, ions are accelerated by scattering back and forth across the shock against Alfvén waves (often self-generated), but non-relativistic electrons cannot resonate with Alfvén waves, so they can only be accelerated in the V _S x B electric field in more quasi-perpendicular regions of the shock. A big wide hemispherical shock surely has some regions where type II electrons get accelerated, while a narrow CME may produce only quasi-parallel regions—accelerating few electrons but plenty of ions. The shock from the May 1, 2000 event may not accelerate enough electrons, but it surely accelerates more ions than any of the shocks we have been able to observe directly in situ with shock speeds as low as 300 km s⁻¹ (Reames 2012). The one feature that seems to show the presence or absence of shock acceleration in impulsive SEP events is the proton excess shown in Figures 2, 3 above (Reames 2019b), but the response of electrons and protons can be completely different.

Some people continue to defined “flare” to include every energetic phenomenon on the Sun, including flares, jets, CMEs, shocks, and even the SEPs at 1 AU. That, of course, makes flare research the most important discipline of all, studying the cause of everything solar, by definition. This deliberately blurs the physics, elevating the importance of flares and diminishing that of CMEs; this was exactly Gosling (1993) objection to “The Solar Flare Myth.” Applying “flare” to everything is not only meaningless but helps no one understand any physics. Beginning with Carrington (1860), many of us still think of “the flare” as that localized sudden bright flash of white light, Hα, or X-ray emission driven because reconnection energy is trapped in closed magnetic loops on the Sun—an event recorded by the Solar Flare Patrol and documented by its timing, latitude, longitude, and C-M-X-scale soft X-ray intensity. These flares are limited in spatial extent. In contrast, the shock wave that accelerates SEPs in a large gradual event is a nominally hemispherical structure initially active from 2 to over 3 R_S (Reames 2009a, b; Cliver et al., 2004), thus enclosing an accelerating volume perhaps ∼10 times the volume of the Sun itself—this is not a flare. Thus the source of SEPs has a major effect on their spatial distributions, among many other things, and lumping all possible sources together, to inflate the egos of a few flare researchers, is no help. please distinguish CMEs from flares in your publications.

This field began with a frequent assumption that all SEPs were somehow actually accelerated in flares. First we found that SEPs in gradual events were accelerated by CME-driven shocks instead. Then we found that the remaining impulsive SEPs in space came from jets. Now we find that it may be more correct to say that flares are caused by SEPs trapped on loops than the converse.