Donald V. Reames
NASA, Goddard Space Flight Center
Greenbelt, MD 20771
For nearly 30 years it was thought that all solar energetic particles
(SEPs) were accelerated in solar flares and that they could somehow
diffuse across magnetic field lines to distant longitudes by a
mysterious mechanism known only as "coronal diffusion."
During recent years, a new class of observations has revealed
two distinct populations of SEPs, with completely different origins,
based upon the abundances, ionization states and time profiles
of the particles as well as the longitude distribution and the
radio, optical, X-ray and gamma-ray associations of the events (23,
25, 27, 28). It has become clear that the largest and most energetic
particle events at Earth are associated with shock waves driven
out into interplanetary space by coronal mass ejections (CMEs).
Energetic particles in these events have, on average, the same
element abundances and ionization states as those in the ambient
plasma of the corona or solar wind; these particles have not
come from the hot plasma of a flare or reconnection region where
they would have become highly ionized. The other class of SEP
events have strong associations with impulsive H-alpha and X-ray flares
and type III radio bursts. The energetic particles from these
events have distinctive 1000-fold enhancements in the 3He/4He
ratio and 10-fold enhancements in heavy element abundance such
as Fe/O. Elements up to Si are fully ionized and Fe has charge
20 indicating heating or other ionization of the plasma. The
strong evidence of electron beams in these events has led to the
suggestion that the enhancements come from resonant interactions
of the particles with waves generated by electron-beam-driven
instabilities in the flare plasma (33, 19, 20). The properties
of gradual and impulsive events are summarized in Table 1.
Historically, the terms impulsive and gradual referred
to the time duration of the soft X-rays in the event. However,
the X-ray duration gives only a poor, statistical distinction
of the underlying mechanisms, while the particle abundances, for
example, distinguish them cleanly. Therefore, we now use the
terms impulsive and gradual to refer to the underlying acceleration
mechanisms, independently of the actual X-ray duration in an event.
Of course, there are events in which both impulsive and gradual
phenomena occur (24, 3)
More recent observations have extended these conclusions to particles
of very high energy. Even in ground-level events (GLEs) particles
of ~20 GeV have a clear association with CME-driven shocks. These
new observations are discussed in the next section. One of the
important consequences of the new paradigm for SEP events is the
demise of the concept of "coronal diffusion" of particles
across magnetic field lines. The consequences of this demise
are not fully appreciated by the authors of some recent papers,
as discussed in the subsequent section. Finally, element abundances
in gradual and impulsive SEP events are considered.
Impulsive Gradual
Particles: Electron-rich Proton-rich
3He/4He ~1 ~0.0005
Fe/O ~1 ~0.1
H/He ~10 ~100
QFe ~20 ~14
Duration Hours Days
Longitude Cone <30 deg ~180 deg
Radio Type III, V (II) II, IV
X-rays Impulsive Gradual
Coronagraph - CME (96%)
Solar Wind - IP Shock
Events/year ~1000 ~10
Ionization states of Fe and lighter elements have now been measured
in large SEP events by 4 experiments on 3 different spacecraft,
spanning an energy range from 0.3 to 600 MeV/amu (14, 15, 13,
34). These measurements are summarized in Table 2. While there
are some differences, it is clear that the Fe has not been subjected
to electron temperatures above 2-3 MK. These ionization states
are consistent, in magnitude and variance, with those found for
Fe, and lighter elements, in the solar wind.
Neither has the Fe been ionized by passing through material at
high velocity, so the acceleration must have taken place in a
low-density region. The ionization states of high-energy Fe would
be measurably altered in less than 1 sec at a density of 10^10
atoms/cm^3 that is typical of the low corona where flare acceleration
occurs. It is extremely unlikely that Fe could be accelerated
to >200 MeV/amu in such a short time. Hence, the ionization
states suggest acceleration of Fe from low-density ambient plasma
in the high corona or in the solar wind.
Conversely, QFe ~ 20 in impulsive events suggests that the acceleration
must occur in a sufficiently dense plasma that collisions with
electrons can occur. It is not clear whether the Fe is ionized
by thermal electrons or by collisions with the electron beam in
these events. We will see in a later section that ionization
probably occurs during or after acceleration in impulsive flares.
In the meantime, Kahler (8, 9) studied the injection altitude
of protons up to 21 GeV in large gradual SEP events and found
that the proton intensities peak when the CME-driven shock is
at >5 solar radii, i.e., outside the corona, in the
solar wind. The event of 1989 September 29 was one event where
the 21 GeV protons reached maximum intensity when the shock at
the leading edge of the CME was at ~6 solar radii. Protons of ~1
GeV generally reached maximum when the shock was at 12 solar radii or more.
The 1989 September 29 event was also one of the events included in the
Tylka et al. (34) observation of Fe at 200-600 MeV/amu
with an average charge of 14.1. This event occurred behind the
west limb at W105°. These observations are quite consistent
with a shock wave from the CME propagating across the high corona
to accelerate high-energy protons and elements up through Fe from
the ambient plasma at ~6 solar radii near the base of the field line connected
to Earth (35).
MeV/amu QFe Events S/C Reference
0.3 - 2 14.1±0.2 12 ISEE 3 Luhn et al. 1987 (14)
0.5 - 5 11.0±0.2 2 SAMPEX Mason et al. 1995 (15)
15 - 70 15.2±0.7 2 SAMPEX Leske et al. 1996 (13)
200 - 600 14.1±1.4 3 LDEF Tylka et al. 1995 (34)
"Coronal diffusion" is an artifact of the "flare myth" (6, 29). If particles are only accelerated at a flare, i.e., a point source in space and time, a mechanism must be invented to transport them to longitudes of 90° or more from that source, where they are observed, as seen in Fig. 1(a). It is an irony of history that there was abundant radio evidence of type II bursts produced by shock waves and shock acceleration of protons was proposed by Wild, Smerd and Weiss (36) before the birth of "coronal diffusion" (28). Shock waves were also observed in the interplanetary plasma at that time. Shock waves easily cross magnetic field lines but charged particles do not.
Evidently there is still some fascination in fitting mathematical "diffusive" forms to time profiles that have a fast rise and slow decline, by varying several adjustable parameters, without the necessity of having to consider physical processes underlying acceleration and transport. The practice persists to this day. Those time profiles that rise slowly or remain constant (over half of the events) are simply ignored by the coronal-diffusion advocates as not being "diffusive."
The evidence against "coronal diffusion" is as follows:
In fact, the particles in the gradual proton events are accelerated by the CME-driven shock that easily crosses field lines to accelerate particles as it goes. There is a 96% correlation between CMEs and proton events (11). In large events the shock has been directly observed by spacecraft near 1 AU that are separated in longitude by 160°, if the shock were symmetric about the source in that event, it would subtend a 240° longitude interval (30). Multispacecraft observations of time profiles of the particles vs. longitude are well organized and understood in terms of the evolution of an observer's magnetic connection to the shock.
If we grant that gradual events involve acceleration at shocks, there is still a question about the longitude distribution of impulsive events and its origin. It is difficult to confidently correct the observed longitude distribution for variation in solar wind speed, however, some estimates of the longitude spread from multispacecraft observations and persistence of a single active region suggest a distribution of width 5° to 20°. Recently Dröge (5) found electron events that were visible over a large spacecraft separation, however, with a factor of 600 decrease in intensity, this is not inconsistent with a ~15° e-folding angle for these electrons, if they are indeed from an impulsive flare.
It seems likely that the main contribution to the longitude distribution in a single impulsive event is the random walk of magnetic field lines discussed by Parker (see 22 and references therein). This process results in a distribution of the source longitudes for the field lines sampled during several hours of particle measurements near 1 AU. While the random walk of the field lines is diffusive in nature, the particles simply follow the bundle of pre-existing field lines that fan out from the flare site; they can not cross to other field lines that pass more than a few gyroradii away unless scattering is intense. Actual scattering across field lines only occurs in isolated regions of high turbulence (e.g., in the immediate vicinity of flares or shocks) or after long periods of time (days).
From the foregoing, we would not expect large, poorly-connected events to be 3He-rich. Yet, Chen et al. (2) found 16 of 29 large events with 3He/4He > 0.5% at 50-110 MeV/amu. Most of these events have 3He/4He ~1%, a 10-fold enhancement over the nominal value, however, they show no enhancement of Fe/O and some have source longitudes like E78 and E85. Of course, the solar wind plasma itself does occasionally have values of 3He/4He as large as 1% (4). Are these events truly 3He-rich in our context? Are the particles accelerated in an impulsive flares?
No. These are gradual events in which shock acceleration has produced enhancements of 3He only at high energies. Evidence for this comes from the flatter spectra that Chen et al. (2) found for 3He than for 4He. The work of Mazur et al. (16) on the spectra of H, He, O and Fe in large gradual events showed a strong Q/A dependence in the spectral indices at high energies. The H/4He ratio often increases by a factor of 100 between 1 and 100 MeV/amu. In this context, a 10-fold increase in 3He/4He between 1 and 100 MeV/amu is not surprising in large gradual events. This enhancement is of different origin than the resonant wave enhancement of 3He and Fe in impulsive events.
Recently, Kiplinger (12) has reported a high correlation between the existence of 10 MeV protons at Earth and a characteristic pattern of X-ray spectral evolution for 18 associated flares. Again, many of these flares are far to the east. Do the protons come from the flare in these large events? What do X-rays produced by electrons in the flare have to do with protons at 1 AU?
In fact, these 18 events were among the 235 proton events that led Cane et al. (1) to propose proton acceleration at CME-driven shocks. The longitude span, time profiles and abundances show the 18 events to be typical large, gradual, proton events. It seems quite clear that the protons are accelerated at the CME-driven shock. The only possibility for a valid correlation seems to be that flares that accompany powerful CMEs occur in a unique configuration of the magnetic field, for example, that leads to a characteristic evolution of the electron spectrum. It is more likely that the correlation results from "big flare syndrome" (7), which states that, despite appearances, not all processes occurring in conjunction with big flares are causally related to the flare or to each other, even when well correlated.
It is instructive to revisit a similar type of correlation, mentioned earlier, namely, X-ray duration and proton events. The long duration of X-ray emission presumably occurs by electron acceleration and heating on "post-flare" loops that are created by magnetic reconnection beneath a rising CME. The protons we see are accelerated by the shock ahead of the CME. Late in the event the reconnection may stop, but the shock and proton acceleration may continue far out in the heliosphere. The apparent correlation between X-ray properties and proton events does not imply that protons are accelerated on post-flare loops where the X-rays originate. In general, X-rays may reveal the presence of energetic electrons, but they tell us little about ion acceleration.
Gradual Events
The event-averaged abundances of elements in gradual events, obtained from low-energy measurements, provide a direct measure of element abundances in the corona and solar wind. These abundances are almost entirely independent of the temperature and ionization state of the source plasma. It has been well known for many years (17) that the ratio of coronal and photospheric abundances of elements is a well-organized function of the first ionization potential (FIP) of the element. A recent summary of abundances is shown in Table 3 (26).
Energetic particles from impulsive flares show element abundances
that differ from those in the corona in that elements with Z>8
are strongly enhanced relative to coronal abundances while He,
C, N, or O. are not (31). Ne, Mg, and Si are enhanced by almost
the same factor, relative to O. This pattern of enhancement
is consistent with acceleration of the ions from a plasma in the
temperature range of 3-5 MK (31, 18). Elements with the same
charge-to-mass ratio, Q/A, have the same magnetic gyrofrequency,
thus they resonate with the same part of the wave spectrum and
are enhanced by the same amount. At 3-5 MK, He, C, N and O are
fully ionized, with Q/A=0.5, while Ne, Mg and Si all have Q/A
~ 0.42. These ion must become fully ionized later in the event;
if they were fully ionized before acceleration they would have
the same Q/A as He, C and O and could not be relatively enhanced.
The pattern of enhancements is discussed in terms of wave absorption
in the "He valley" by Meyer (18, see also 32) and is
modeled by a theory of cascading waves by Miller and Reames (20).
Z FIP Photosphere SEP Corona Flares
(Gradual Events) (Impulsive Events)
H 1 13.53 1.18x10^6 (1.57±0.22)x10^6 ~1x10^6
He 2 24.46 1.15x10^5 57000±3000 46000±4000
C 6 11.22 468 465±9 434±30
N 7 14.48 118 124±3 157±18
O 8 13.55 1000 1000±10 1000±45
F 9 17.34 0.0351 <0.1 <2
Ne 10 21.47 161 152±4 400±28
Na 11 5.12 2.39 10.4±1.1 34±8
Mg 12 7.61 44.6 196±4 408±29
Al 13 5.96 3.54 15.7±1.6 68±12
Si 14 8.12 41.7 152±4 352±27
P 15 10.9 0.433 0.65±0.17 4±3
S 16 10.3 20.4 31.8±0.7 117±15
Cl 17 12.95 0.218 0.24±0.1 <2
Ar 18 15.68 4.21 3.3±0.2 30±8
K 19 4.32 0.157 0.55±0.15 2±2
Ca 20 6.09 2.55 10.6±0.4 88±13
Ti 22 6.81 0.10 0.34±0.1 <2
Cr 24 6.74 0.563 2.1±0.3 12±5
Fe 26 7.83 37.9 134±4 1078±46
Ni 28 7.61 2.05 6.4±0.6 42±9
Zn 30 9.36 0.0525 0.11±0.04 6±4
Recent evidence shows that particles accelerated by CME-driven shocks, rather than by flares, produce most of the largest particle events seen at 1 AU, even the ground-level events with particles of energies as high as ~20 GeV (8). Protons of 1-10 MeV may be accelerated throughout a large fraction of the heliosphere in large events. During the first few days, the temporal evolution of these particles at a point in space is determined almost entirely by the changes in the locus of the observers magnetic connection to the shock (30, 27). This locus scans across the surface of the outbound shock, sampling different source intensities as a function of time. Particles of the highest energies are primarily accelerated nearer the sun, where the shock is strongest, yet chiefly outside the corona. For these particles, the radial variation of the source strength can dominate.
We can distinguish flare-accelerated particles by their unique abundances, enhancements in 3He/4He, Ne/O, Mg/O, Si/O and Fe/O. The resonant wave-particle processes that produce these enhancements seem to occur in all the flare-associated events we observe (now over 200 events). Of course, we can only observe ions from magnetically well-connected events. With the exception of one event where broad gamma-ray lines were observed (21), all our information on the nature of the ion acceleration process comes from the particle observations in interplanetary space. Radio and X-ray observations tell us of the electron beams that may be required for the 3He enhancements (33, 19), but the nature and even the existence of resonant ion physics in solar flares was revealed only by the particles at 1 AU. The gamma-ray and neutron observations tell us about the time scale for ion acceleration
Abundances and charge states of the elements in SEP events have provided information on the temperature of the source plasma, the composition of the solar corona and the existence and nature of resonant ion acceleration in flares. They have also corrected serious errors in our understanding of particle origin and transport. Those errors were caused by an excessive focus on only protons and electrons and their time profiles, and by a neglect of shocks and CMEs in favor of flares.
I would especially like to thank Prof. Y. Muraki and the Solar-Terrestrial Environment Laboratory at Nagoya University for their hospitality during my recent 3-month visit. A substantial part of this paper was written during that time. I would also like to thank C. K. Ng for many pleasant and productive hours of discussion on this subject.
1. Cane, H.V., Reames, D.V., and von Rosenvinge, T. T., J. Geophys. Res.
93, 9555 (1988).
2. Chen, J., Guzik, T.G., and Wefel, J.P., Astrophys. J. 442, 875 (1995).
3. Cliver, E.W., this volume (1996).
4. Coplan, M.A., Ogilvie, K.W., Bochsler, P., and Geiss, J., Solar
Physics 93, 415 (1984).
5. Dröge, W., this volume (1996)
6. Gosling, J.T., J. Geophys. Res. 98, 18949 (1993).
7. Kahler, S.W., J. Geophys. Res. 87, 3439 (1982).
8. Kahler, S.W., Astrophys. J. 428, 837 (1994).
9. Kahler, S.W., this volume (1996).
10. Kahler, S.W., Cliver, E.W., Cane, H.V., McGuire, R.E., Stone, R.G.,
and Sheeley, N.R., Jr., Astrophys. J. 302, 504 (1986).
11. Kahler, S.W., Sheeley, N.R., Jr., Howard, R.A., Koomen, M.J.,
Michels, D.J., McGuire, R.E., von Rosenvinge, T.T., and
Reames, D.V., J. Geophys. Res. 89, 9683 (1984).
12. Kiplinger, A.L., Astrophys. J., 453, 973 (1995).
13. Leske, R.A., Cummings, J.R., Mewaldt, R.A., Stone, E.C., and von
Rosenvinge, T.T., Astrophys. J., in press (1996)
14. Luhn, A., Klecker, B., Hovestadt, D., and Mbius, E.,
Astrophys. J. 317, 951 (1987).
15. Mason, G.M., Mazur, J.E., Looper, M.D., and Mewaldt, R.A.,
Astrophys. J. 452, 901 (1995)
16. Mazur, J.E., Mason, G.M., Klecker, B., and McGuire, R.E.,
Astrophys. J. 401, 398 (1992).
17. Meyer, J.P., Astrophys. J. Suppl. 57, 151 (1985).
18. Meyer, J.P., this volume (1996).
19. Miller, J.A. and Viñas, A.F., Astrophys. J. 412, 386, 1993.
20. Miller, J.A. and Reames, D.V., this volume (1996).
21. Murphy, R.J., Ramaty, R., Kozlovsky, B., and Reames, D.R.,
Astrophys. J. 371, 793 (1991).
22. Parker, E.N., Physics Today 40 (July) 36 (1987).
23. Reames, D.V., Astrophys. J. Suppl. 73, 235 (1990).
24. Reames, D. V., Astrophys. J. (Letters) 358, L63 (1990).
25. Reames, D.V., Adv. Space Res. 13 (No. 9), 331 (1993).
26. Reames, D.V., Adv. Space Res. 15 (No. 7 ), 41 (1994).
27. Reames, D.V., Third SOHO Workshop: Solar Dynamic Phenomena
and Solar Wind Consequences, Ed. A. Poland, Estes Park, CO, ESA,
p. 107 (1994).
28. Reames, D.V., Revs. Geophys (Suppl.) 33, 585 (1995).
29. Reames, D.V., Eos, 75, 405 (1995).
30. Reames, D.V., Barbier, L. M., and Ng, C. K., Astrophys. J.,
in press (1996).
31. Reames, D.V., Meyer, J.P., and von Rosenvinge, T. T., Astrophys. J.
Suppl. 90, 649 (1994).
32. Steinacker, J., Meyer, J.P., Steinacker, A., and Reames, D. V.,
Astrophys. J., in press (1996).
33. Temerin, M. and Roth, I., Astrophys. J. (Letters) 391, L105 (1992).
34. Tylka, A.J., Boberg, P.R., Adams, J.H., Jr., Beahm, L.P.,
Dietrich, W.F., Kleis, T., and Astrophys. J. (Letters) 444,
L109 (1995).
35. Vestrand, W.T., and Forrest, D.J., Astrophys. J. (Letters)
409, L69 (1993).
36. Wild, J.P., Smerd, S.F., and Weiss, A.A., Ann. Rev. Astron. Ap. 1,
291 (1963).