1 In Pursuit of Abundances

The Sun is a microcosm for many of the high-energy physical processes that occur throughout the Universe. Magnetic reconnection generates violent explosive flares and it drives solar jets and liberates coronal mass ejections (CMEs); these CMEs drive collisionless shock waves that ply the solar corona; magnetic turbulence scatters ions differentially, and wave-particle resonances occur throughout. These processes modify the relative abundances of ions of the available elements in definable ways wherever they occur, including the coronae of distant stars, perhaps even those with habitable planets. However, in the case of the Sun, unlike distant sources, we have the rare privilege to actually collect direct samples of the energetic ions themselves. “Multi-messenger” comparisons may help us relate these ions to the limited photons they emit elsewhere.

Measurements of the abundances of the chemical elements has been a rich source of information about the nature of the solar energetic particles (SEPs), their source environment, and of the physical processes that fractionate the solar corona itself, in addition to particle acceleration and transport under a wide variety of circumstances. We have only to disentangle the individual processes that contribute – a task that has kept us quite busy for 60 years.

The purpose of this article is to acknowledge this milestone, to review the accomplishments, and to highlight remaining questions.

1.1 Early Observations, 1961-

Study of element abundances in SEP events began when Fichtel and Guss (1961) reported the first observation of elements with atomic numbers \(Z > 2\). Their Nike-Cajun sounding rocket flight from Ft. Churchill, Manitoba exposed a nuclear emulsion payload above the atmosphere for several minutes during the SEP event of 3 September, 1960. The emulsions recorded about two dozen ions of C, N, and O above about 40 MeV amu−1 and a few Ne, Mg, Si, and S ions, but an absence of Li, Be, and B. The author of this review joined that group before the next solar cycle when measurements would be extended up to the abundance peak at Fe, using the same technique (Bertsch et al. 1969).

Comparisons began early of these SEP abundances of elements with those of the solar photosphere, which were also evolving, and Biswas and Fichtel (1964) summarized the abundances of He, C, N, O, Ne, Mg, Si, and S. With hindsight we might recognize that the heavier elements are somewhat suppressed in these large SEP events, but we would also note their high energy and that the brief rocket shots occurred hours to days after the events began, providing no time histories of the events.

The next decade saw the flight of \(dE\)/\(dx\) vs. \(E\) telescopes using higher-resolution detectors on satellites. Early experiments spent most of their time measuring the dominant H and He, but soon, priority systems were invented to insure that ions with \(Z > 2\) received a reasonable share of the telemetry space despite high intensities of electrons, protons, and He (e.g. Teegarden et al. 1973). These measurements showed that abundance ratios, such as C/O, were energy independent from 8 to 40 MeV amu−1 and that the abundances of the elements from C through Fe were somehow comparable with photospheric or coronal abundances, as far as these were known. They also disagreed strongly with a previous event measured by Mogro-Campero and Simpson (1972) that showed a smooth systematic increase in the abundance enhancement relative to the corona by a factor of order ten vs. \(Z\) between C and Fe. These authors recognized the enhancement as a probable dependence on the mass-to-charge ratio \(A\)/\(Q\) of the ions. Did SEP abundances have a strong dependence on \(A\)/\(Q\) or not?

Figure 1 compares samples of the evolution in particle resolution and measurement. Fig. 1a shows the data of Fichtel and Guss (1961), Fig. 1b shows resolution up to Fe by Bertsch et al. (1969), and Fig. 1c shows resolution typical of Si, solid-state detectors flown on satellites beginning in the 1970’s. It is the accumulation of ions over 43 large SEP events for the average abundance study by Reames (1995a).

Fig. 1
figure 1

Panel (a) shows identification of ions by counts of \(\delta \) rays (scattered electrons) per unit length along particle tracks in nuclear emulsions flown on sounding rockets by Fichtel and Guss (1961), (b) shows similar resolution of elements up to Fe by Bertsch et al. (1969), and (c) shows resolution of Si, solid-state detectors with data accumulation during 43 large SEP events from 1978 – 1988 in a histogram of \(dE\)/\(dx\) vs. \(E\) by Reames (1995a). SEP abundances are now measured by simply counting ions of each element in a constant velocity interval

Full size image

1.2 3He-rich Events, 1970-

Nearly every scientist who considered SEPs had previous experience with galactic cosmic rays (GCRs) and the lessons learned from GCRs would all be tested against SEPs (often called SCRs for solar cosmic rays). GCRs are accelerated by shock waves from supernovae and during their ∼107-year sojourn in interstellar space they undergo nuclear collisions with interstellar H, producing significant abundances of secondary ions such as 2H and 3He and isotopes of Li, Be, and B. Early abundances of SEPs had shown a lack of Li, Be, and B. However, when Hsieh and Simpson (1970) observed an SEP event with 3He/4He = 0.021 ± 0.004, much enhanced compared with \(4\times 10^{-4}\) in the solar wind (e.g. Gloeckler and Geiss 1998), it seemed to be possible evidence of nuclear reactions in solar flares after all, especially since \(\gamma \)-ray lines had also been observed from flares (Chupp et al. 1973; Ramaty and Murphy 1987). However, subsequent observations were soon found of many events like that seen by Serlemitsos and Balasubrahmanyan (1975) where 3He/4He = 1.52 ± 0.10 but 3He/2H > 300. These huge enhancements of 3He with no 2H or 3H were incompatible with nuclear reaction products. The case closed further when limits on Be/O or B/O in large SEP events were found to be \(< 2 \times 10^{-4}\) (e.g. McGuire et al. 1979; Cook et al. 1984). Reaction products are indeed produced in flares, but flares occur on closed magnetic loops so ions cannot get out. SEPs in space do not come from flares, and we need a whole new resonance mechanism for these “impulsive” SEP events.

There followed many years of many proposals for the enhancement of 3He, first from preferential preheating by resonant wave absorption (e.g. Ibragimov and Kocharov 1977; Kocharov and Kocharov 1978, 1984; Fisk 1978; see Sect. 2.5.2 of Reames 2021b and references therein), which required additional acceleration. Generally, waves that resonate with the gyrofrequencies of the much-more-abundant H and 4He are absorbed by them during stochastic acceleration, but 3He with a gyrofrequency between these two dominant ions continues to absorb waves that it is too rare to damp. Essentially, while H and 4He assume nearly power-law spectra, nearly all of the 3He in the active volume gets accelerated to a peak in the region ∼0.1 – 1 MeV amu−1 (e.g. Liu et al. 2006). Temerin and Roth (1992) suggested preferential acceleration by absorption of electromagnetic ion cyclotron (EMIC) waves generated by streaming electrons. Impulsive SEP events had been shown to be accompanied by the intense beams of electrons (Reames et al. 1985) that produce type III radio bursts (Reames and Stone 1986). Temerin and Roth (1992) envisioned 3He ions absorbing incoming waves as they mirrored in magnetic fields, in analogy with ion conics seen in the Earth’s aurorae. Their model led to acceleration, not just preheating, and associated the 3He with the electrons.

It also became clear that the element abundances in 3He-rich events showed a systematic enhancement with \(Z\) that grew to a factor of ten for Fe/O (e.g. Mason et al. 1986; Reames et al. 1994; Mason 2007) when abundances in these small events were compared with average abundances in large “gradual” SEP events. A study of daily-averaged abundances over an 8.5-year period showed the existence of two distinct populations of particles (Reames 1988). On a plot of daily intensities of Fe vs. O at ≈2 MeV amu−1, the Fe-rich branch was also 3He-rich, electron-rich, and proton-poor. A modern version of this plot in Fig. 2 shows intensities of Fe vs. O with 3He/4He enhanced in color and point size in the upper panel and heavy element abundances (\(50 \leq Z \leq 56\))/O enhanced in the lower panel.

Fig. 2
figure 2

In each panel, particle intensities of Fe are plotted vs. intensities of O for all 8-hour intervals in an 8.5-year study period from 1994 through 2003 whenever Fe and O were measurable. The indicated intensity of anomalous-cosmic-ray (ACR) O forms a lower bound on the O intensity during solar minimum. The symbols used for points in the upper panel indicate the average value of 3He/4He during each interval, as indicated in the scale to the right of the panel. Symbols in the lower panel denote the (\(50\leq Z\leq 56\))/O abundance ratio shown in the right-hand scale. Lines drawn along Fe/O =1 and 0.1, approximately track the locus of impulsive and gradual SEP events, respectively (Reames and Ng 2004). This is an updated version of the bimodal abundance study of Reames (1988)

Full size image

Observations of \(\gamma \)-ray lines in large events also began to show evidence of enhancements in the Doppler-broadened lines of the accelerated Fe/O (Murphy et al. 1991) and 3He/4He (Mandzhavidze et al. 1999; Murphy et al. 2016), suggesting the same physics in flares as in the impulsive SEP events we see in space. Flares were understood to involve magnetic reconnection on closed field lines, but impulsive SEP events were not initially associated with jets, which involve reconnection on open field lines.

Thus, 3He-rich events were a major discovery that focused the mind on sources and physical mechanisms. Earlier theory of SEPs had been focused almost exclusively on diffusive transport with very little focus on acceleration (e.g. see Chap. 2 of Reames 2021b). If 3He-rich events were somehow related to solar flares, what was the origin of the “big proton events”? Why were they not 3He rich?

Following a brief earlier comparison by Reames (1999), it was found that as event sizes increased, 3He became depleted from its probable source volume and had a limited fluence (Ho et al. 2005).

1.3 Ionization States, 1984-

Luhn et al. (1984, 1987) provided the earliest direct measurements of ionization states, \(Q\), of the ions of elements up to Fe for energies 0.34 – 1.8 MeV amu−1. For gradual events they found average values of \(Q\)Si = 11.0 ± 0.3 and \(Q\)Fe = 14.1 ± 0.2 suggesting a temperature near ≈ 2 MK. However, for 3He-rich events they found \(Q\)Si ≈ 14 and \(Q\)Fe = 20.5 ± 1.2, which could suggest either a source temperature of ≈10 MK, or strip** of the ions after acceleration. Generally, the results were initially misinterpreted as the ions in impulsive events coming from hot flares. It was not initially recognized how difficult it would be to enhance abundances of any of the elements Ne, Mg, and Si relative to He, C, N, and O, if all of them were fully ionized with \(A\)/\(Q = 2\) at 10 MK, suggesting that they were actually accelerated at some lower temperature, then stripped by traversing a small amount of matter afterward.

Additional measurements of \(Q\) in gradual events, including those that used relative deflection of ions in the Earth’s magnetic field, measured ions with energies up to 200 - 600 MeV amu−1 and showed that \(Q\)Fe varied from 11 to 15 (Mason et al. 1995, Leske et al. 1995, 2001; Tylka et al. 1995; Klecker 2013) in different events.

The alternate interpretation of the ionization-state measurements in impulsive events became a requirement when DiFabio et al. (2008) found that the charges increased with particle energy, suggesting that the ions were increasingly stripped to a higher \(Q\) at higher speed, in fact, attaining an equilibrium \(Q\) appropriate for each speed. Thus these ions were clearly stripped after acceleration when they traversed a small amount of matter. This might occur for acceleration at a depth of about 1.5 RS in the corona, the authors suggested.

A different type of logic was applied by Reames et al. (1994) to determine ionization states. They noted that in impulsive events, on average, the elements He, C, N, and O had the relative abundances expected from a sample of the corona, Ne, Mg, and Si, were each enhanced by a factor of about 2.5, and Fe was enhanced by a factor of ≈7. They thought this suggested that He through O were all fully ionized with \(Q\)/\(A = 0.5\), suggesting a temperature of 3 – 5 MK; above ∼3 MK was required to keep O ionized. The similarity of the enhancements for Ne, Mg, and Si suggested that they were each in a highly stable state with two orbital electrons so that \(Q\)/\(A \approx 0.43\), as only occurs for \(T < 5\) MK, above which Ne loses electrons. This would later provide the basis for using abundances enhancements to estimate temperatures. At the time it was also another argument against the possibility of ∼10 MK solar flares as a source.

1.4 FIP and A/Q, 1975-

It was recognized early that SEP abundances, especially in large (gradual) events, had a dependence on the first ionization potential (FIP) of the elements, relative to the corresponding photospheric abundances. In the photosphere, elements with low FIP (<10 eV) were ionized while those with high FIP were neutral atoms and unaffected by electromagnetic fields. A similar FIP dependence had been suggested for the GCRs, compared with “universal” abundances, and this strong correspondence was noted early by Webber (1975).

Meyer (1985) reviewed the increasingly extensive SEP abundance measurements and characterized the events (excluding 3He-rich events) as having a common “un-biased baseline” that differed from the photosphere as a function of FIP, and a “mass bias” dependence that was actually a function of \(A\)/\(Q\) and varied from event to event. Presumably the baseline SEP abundances represented the abundances in the corona, determined long before acceleration, while the \(A\)/\(Q\) dependence might occur during acceleration. We do not need to decide between FIP- and \(A\)/\(Q\)-dependence of SEPs, we need both.

Meanwhile, Breneman and Stone (1985) used the new average \(Q\)-values from Luhn et al. (1984) to plot abundance enhancements vs. \(A\)/\(Q\) and actually showed a power-law dependence that could increase with \(A\)/\(Q\) in some events and decrease in others. Of course, one would not expect all SEP events to come from a single temperature environment and have the same pattern of \(Q\) values as the Luhn et al. (1984) average, but \(A\)/\(Q\) probably represents the residual affects of ion magnetic rigidity, and one could easily expect physical processes of SEP acceleration and transport that vary as power laws in magnetic rigidity. Thus, if we want to understand the abundance pattern in an SEP event, we must know the power-law \(A\)/\(Q\) pattern of the elements, i.e. we must know an effective temperature for the source plasma contributing the ions with appropriate \(Q\) values in that event. For SEPs from the corona, a temperature determines the \(Q\)-values and the power law in \(A\)/\(Q\) determines the abundances enhancements (e.g. Reames 2018b).

Breneman and Stone (1985) also believed that correction by a power-law in \(A\)/\(Q\) was needed between average SEP and coronal abundances. More-recent SEP abundances averaged over many large events are usually treated directly as coronal abundances (Reames 1995a, 2014). This is based partly on the idea that differences in ion transport will cause different spatial distribution of different elements, but these will average out when we integrate over many events observed from diverse perspectives. The elements Mg and Si have nearly the same FIP as Fe, but different values of \(A\)/\(Q\); yet, they have similar relative enhancements on a plot of FIP dependence.

Isotope resolution, for elements other than He, have been extended up to Fe (e.g. review by Leske et al. 1999, 2007). Basically, these measurements show variations in \(A\)/\(Q\) for single elements that are similar in character to those seen more broadly for different elements (e.g. Breneman and Stone 1985).

1.5 Self-generated Waves

Early work ascribed nearly all aspects of space-time variation in SEP events to scattering during transport from a point source with small scattering mean free paths (\(\lambda \sim 0.1\) AU), the “Palmer (1982) consensus,” which were treated as invariant, pre-existing properties of interplanetary space. Then Mason et al. (1989) found that 3He-rich events were nearly scatter free (\(\lambda \geq 1\) AU), even when they occur in the slowly declining phase of a large gradual SEP event (e.g. Chap. 2 in Reames 2021b). If the slow decline in intensity is caused by intense scattering, how can the newly-injected ions travel scatter free?

At the other extreme, the largest gradual events showed evidence of proton amplified waves that differentially affected ions, altering abundance ratios, limited intensities early in events, and flattened low-energy spectra (Reames 1990; Reames et al. 2000; Ng and Reames 1994; Reames and Ng 2010; Ng et al. 1999, 2003, 2012). When intense, streaming protons are present, scattering can become a strong function of rigidity, space, and time.

2 The Turn of the Century

As we approached the year 2000, we had fairly extensive observations of SEPs at locations throughout the heliosphere. Two major classes of SEP events had been resolved, called impulsive and gradual, where ion abundances suggested different physics dominated particle acceleration. The origin of gradual SEP events seemed most clear; particles in gradual events were accelerated at shock waves (e.g. Lee 1983, 2005) driven by fast, wide CMEs (Kahler et al. 1984) and they could rapidly reach GeV energies (Zank et al. 2000, 2007; Cliver et al. 2004; Sandroos and Vainio 2007; Ng and Reames 2008). Kahler et al. (1984) found a 96% correlation between large gradual SEP events and fast, wide CMEs; Mason et al. (1984) found consistent abundance ratios over broad longitude intervals that would not support ideas of cross-field transport (see many reviews e.g. by Reames 1999, 2013, 2021b; Lee et al. 2012; Desai and Giacalone 2016). Impulsive events were loosely associated with magnetic reconnection and solar flares involving stochastic acceleration (Ramaty 1979; Miller et al. 1997; Miller 1998). The extreme enhancement of 3He required resonant wave-particle interactions, as discussed above, possibly related to the associated streaming electrons that produced type III radio bursts (Reames 1985; Reames and Stone 1986). Actually, the first suggestion of two classes of SEP events had been based upon radio bursts (Wild et al. 1963), type III bursts were produced by “pure” streaming electrons while type II bursts took place at shock waves where proton acceleration was known to occur. These “pure” electron events (Lin 1970, 1974) turned out to actually be 3He-rich events (Reames et al. 1985).

The final decade of the 1990s had surfaced an extensive controversy between the SEP community and physicists who still assumed that all SEPs must come from flares, especially energetic SEPs that could threaten astronauts. Gosling’s (1993, 1994) paper entitled “The solar flare myth” reiterated the importance of shock waves driven by CMEs, not of flares, for the largest gradual SEP events, and reviewed evidence for two classes of SEP events. This paper was seen to “wage an assault on the last 30 years of solar-flare research” (Zirin 1994). Three perspectives on the controversy were invited by Eos: Hudson (1995) argued that the term “flare” should include the CME, shock, and any related physics; Miller (1995) argued that flares were more numerous and thus better subjects for acceleration studies; Reames (1995c) argued that distinguishing the physics of all sources, especially their spatial extent, was important for SEPs. The next decades, described below, would improve our estimates of SEP source properties, and would implicate solar jets, with magnetic reconnection involving open field lines, as the actual source of the impulsive SEP events seen in space (Kahler et al. 2001; Bučík 2020).

From a modern perspective, it turns out that as much as half of the energy of magnetic reconnection can directly produce energetic particles (Aschwanden et al. 2019), especially electrons. When trapped on closed loops, these SEPs soon scatter into the magnetic loss cone to deposit their energy in the denser corona below, where the heating causes hot, bright plasma to expand back up into the loops – a solar flare. Thus it may be more accurate to say that SEPs cause flares than the converse. Flares exist precisely because these SEPs and the heated, >10 MK plasma are well trapped.

Of course, this evolution of our understanding of SEPs was not only driven by abundance measurements. Spatial distributions of protons (e.g. Reames et al. 1996, 1997), correlations with CMEs (Kahler 2001; Kouloumvakos et al. 2019; Rouillard et al. 2011, 2012, 2016), onset timing (Kahler 1994; Tylka et al. 2003; Reames 2009a,b), electron/proton ratios (Cliver and Ling 2007; Cliver 2016) and intense events and spectral breaks (Tylka and Dietrich 2009; Mewaldt et al. 2012; Gopalswamy et al. 10.

The abundance of He shows event-to-event variations in both impulsive and gradual events which is often suggested to be related to the slow ionization of He because of its very high FIP = 24.6 eV, but Ne at 21.6 eV does not share these variations at all. Impulsive events with 3.2 MK source temperatures, where \(A\)/\(Q\) variations are precluded, have source He/O ≈ 90, as do SEP3 events that reaccelerate these ions. Despite this, a few rare small SEP1 events have strongly suppressed values as low as He/O ≈ 2 (Reames 2019a); we do not understand these events. Are there sometimes He-poor jets? Most SEP4 gradual events have He/O ≈ 50, not 90 (Reames 2017). Why? Where do they sample the ambient coronal plasma? It seems unlikely that these events have \(A\)/\(Q\)-enhanced O.

The ratio of C/O = 0.420 ± 0.010 in SEPs is uniquely below theoretical expectations as is the ratio in the solar wind (e.g. Reames 2020a). Could the photospheric C/O = 0.589 ± 0.054 be 40% too high? The SEP C/O is always below 0.5, in every event, well below the photospheric value. Is there some mechanism that selectively suppresses C in the corona, or in the SEPs? Older, e.g. Anders and Grevesse (1989), photospheric values of C/O = 0.489 are in better agreement with SEPs.