Data from the AS & E X-ray spectrographic telescope on Skylab were used for a preliminary study of two solar flares on 1973, August 9 and September 5. Photographic images taken through broad band filters during the early flare onsets and subsequent evolution of the flares were analyzed by microdensitometry and subsequent conversion to energy maps of the flare areas.

Recent flare studies have shown that soft X-ray data are not compatible with simple isothermal models of the source (Herring and Craig, 1973; Craig, 1973; Neupert et al., 1973). With this in mind, the emitting flare plasma has been represented by the temperature-emission measure distribution function, where ζ(T) is the differential emission measure (cm–3 per 106 K), T is the electron temperature in units of 106 K, T0 is a low temperature cut-off for the distribution, αi are real positive numbers, and Ai are positive coefficients determined from data (for appropriate values of T0 and αi) by a least squares fitting procedure. Such a distribution is suggested by results obtained by the present author using simple delta-function representations for ζ(T) (with n ≤ 4); these discreet multi-temperature models usually indicate that the emission measure decreases with increasing temperature. Also, as discussed by Brown (1974), a power law distribution for ζ(T) is consistent with the observed bremsstrahlung emission in the hard X-ray (> 10 keV) domain. In attempting to find a suitable form for the differential emission measure, a simple empirical function of the type assumed by Chambe (1971) for active regions was also tried, but the fit, as evidenced by the χ2 test was unsatisfactory.

Solar X-rays in the energy range 1–100 keV originate in hot plasmas and streams of energetic electrons in solar flares, and since these phenomena may represent a significant fraction of the energy in a flare, an understanding of them is important for any flare theory. This paper presents the results of the University of California, San Diego, solar X-ray instrument on the OSO-7 satellite. Study of the time evolution of the emission measure in a typical burst indicates that the growth of soft X-ray emission is due to the addition of new hot material to the flare plasma, and the study of the time evolution of the temperature of the plasma indicates that conduction is the dominant cooling mechanism. Comparison of the hard (10–100 keV) and soft (5–10 keV) data indicates that the main heat input to the flare plasma is not collisions by the electrons which make the hard X-rays. The fraction of soft X-ray bursts observed by the instrument which also have a detectable hard X-ray component is this result is the same for bursts which occured near the center of the disk (θ < 60°) and for those bursts believed to have been partly occulted by the limb, indicating that hard X-ray emission comes at least part from high in the corona. For a sample of 62 hard X-ray bursts which occurred near or beyond the limb, the spectral index of the hard X-ray power law was significantly larger, as compared with the spectra of a comparable number which occurred at solar longitudes less than 60°.

The UCSD experiment on the OSO-7 satellite has provided hard and soft X-ray observations of a large number of solar flares. Of these a sample of 123 had sufficiently large fluxes to permit analysis of their spectra (Datlowe et al., 1974). The locations of these flares upon the solar disk have been obtained by comparison with Hα flare listings. We find that the soft (5.1–6.6 keV) X-ray bursts, above a threshold of 1000 photons (cm2 s keV)–1, have a relatively flat distribution from center to limb. The frequency of occurrence of hard (20–30 keV) X-ray bursts, as normalized to the longitude distribution of the soft X-ray bursts, shows a statistically insignificant excess of 19±34% in the longitude range 80–90°. Furthermore, the limb flares exhibit a small but statistically significant spectral softening.

Various authors have presented arguments for either the thermal or the nonthermal interpretations of impulsive E > 20 KeV X-ray bursts and slowly varying E < 10 keV X-ray bursts. In this review the arguments for and against the prevailing opinion that the impulsive bursts are nonthermal and the slowly varying bursts are thermal are presented.

For the impulsive bursts we discuss the spectra, electron mean free paths, center-to-limb distributions of both the numbers of events and spectra of events, and polarization data as relevant criteria. For the slowly varying events we discuss electron self collision times, distribution of X-ray temporal parameters, associated gradual rise and fall radio bursts, spectral and time profiles of special events and center-to-limb distributions of numbers of events as the relevant criteria.

The Utrecht Hard Solar X-Ray Spectrometer on board the ESRO TD-1A satellite (launch March 1972) is permanently Sun-pointed and measures the solar radiation between 30 keV and 1000 keV, in 12 logarithmically spaced energy channels, with a continuous fine time resolution, viz. 1.2 s for the four lowest energy channels and 4.8 s for the rest. The detector has a 5 cm2 Cs I (Na) crystal; counts due to particles are rejected and even during the largest solar flares saturation effects (e.g. pulse pile-up) are absent. For further details see Van Beek (1973), and Van Beek and De Feiter (1973).

A study was made of the hard X-ray component in the impulsive phase of solar flares. In 36 randomly chosen events the value for the slope in the differential electron power spectrum E−δ electrons cm−2 s−1 keV−1, was related to the 20–32 keV spike rise time (e-folding) as trise = 0.56 exp (0.88δ) in the thin target model and tfrise = 0.10 exp (0.88δ) in the thick target picture. In the thin target model, the above empirical relation would imply that the acceleration of electrons can be longer when the acceleration rate is smaller. An alternative interpretation would be that an impulsive hard X-ray burst is a superposition of two components emitted from thin and thick targets; when the former predominates, the duration is longer and the photon spectral index is larger, while when the latter predominates, the duration is shorter and the photon spectral index is smaller; 3 ≲δ ≲4 is required (Figure 1). The uncertainty in δ is 0.5 while that in the rise time is 1 s.

The importance and difficulties of determining the height of hard X-ray sources in the solar atmosphere, in order to distinguish source models, have been discussed by Brown and McClymont (1974) and also in this Symposium (Brown, 1975; Datlowe, 1975). Theoretical predictions of this height, h, range between and 105 km above the photosphere for different models (Brown and McClymont, 1974; McClymont and Brown, 1974). Equally diverse values have been inferred from observations of synchronous chromospheric EUV bursts (Kane and Donnelly, 1971) on the one hand and from apparently behind-the-limb events (e.g. Datlowe, 1975) on the other.

Broadband photographs and spectra of the white light flare of 1972, August 7 have been compared with hard X-ray spectra from the same event. There is a very close temporal correspondence between the hard X-ray and white light emission curves, and these emissions come from layers that are separated by a height of less than 2000 km. The flare shows at least two distinct particle acceleration phases: the first, occurring at a stationary source, gave very bluish continuum emission from 4 bright stationary knots while the X-ray (E > 60 keV) spectrum hardened and reached peak intensity. This phase occurred between 1520 and 1523 UT. In the second phase (1524–1537 UT) the bright knots dissolved and a faint wave moved out from the flare center at 40 km s–1. The spectrum of the wave was nearly flat in the range 4950–5900 Å and analysis of the spectrum indicates that the emission was probably due to heating and ionization by 20–100 keV electrons. The X-ray spectrum, as derived from Interkosmos 7 and ESRO TD-1A satellite data, becomes softer during the wave phase. The close correspondence between the X-ray and continuum emission events shows that, in effect, the hard X-ray source has been resolved. It consists of several changing patches approximately 3″ × 5″ in area, consistent with the upper limit of 1′ from balloon observations (Takakura et al., 1971).

The importance of interpretation of hard X-ray burst spectra, polarisation and directivity to the flare process as a whole is emphasised. After critically reviewing observations of these and related burst characteristics, the problems of analytic and numerical inversion of the X-ray spectrum to give the flare electron spectrum are discussed and it is concluded that electron spectra cannot be accurately and unambiguously inferred from their bremsstrahlung emission. Consideration of directional, albedo, and model-dependent effects, on the other hand, shows that none of the X-ray data are at present inconsistent with a power-law electron acceleration spectrum.

Characteristics of thick-target, thin-target and electron-trap models of hard X-ray sources are discussed quantitatively and their ability to fit the observations is examined. Selection of a satisfactory model is precluded by lack of both sufficient observations and of adequate theoretical description of models. Nevertheless, it is suggested that redistribution of the flaring atmosphere and the effects of collective energy losses may reconcile even behind-the-limb burst observations and interplanetary electron spectra with a thick-target description (which fits other data well). This is attractive since a thick-target X-ray source makes the minimal demand on flare energy. Even a thick-target, however, requires an embarrassingly large number and energy of fast electrons. Therefore the review is completed by discussing how these requirements might be reduced if thermal emission extended to hard X-ray energies or if multiple reacceleration of electrons occurred.

This review discusses the available observational material of solar hard X-ray bursts, their interpretation in terms of a model of the source region and their relation with other flash-phase phenomena, in particular the impulsive microwave bursts.

Recent observations of impulsive microwave and hard X-ray emissions during the early phase of the flares are briefly reviewed in order to deduce the dynamics of energetic electrons consistently from two view points of the microwaves and X-rays. An emphasis is put on the necessity of distinction between temporal and spatial variations so far confused in the interpretation of the time histories of the X-ray and radio emissions. The role of plasma turbulence on the dynamics of the energetic electrons is shown to be important in deducing the model of X-ray and radio sources.

The solar X- and gamma-ray experiment aboard Prognoz 2 recorded solar gamma-radiation bursts during the 1972, August 2, 4 and 7 events. This work analyses the general evolution of the August 2 and 7 events and the possible evidence for γ-ray lines during these two events.

The present status of our knowledge concerning the production of gamma-ray lines and continuum during the impulsive phase of solar flares is reviewed. Our data in this field is based solely on the OSO-7 observations made in 1972, August 4 and August 7. The experimental data will be reviewed. These observations along with theoretical work of Ramaty and Lingenfelter (1973a, b) and the charged secondary observations of the Chicago group (Anglin et al., 1973) lead to the investigation of different hypothetical models to explain the production of neutral and charged secondaries in the solar atmosphere. At the present time it is not possible to rule out the preflare and postflare accumulation models if all the data is considered. We will discuss the outstanding experimental questions to be answered in future investigations.

Observations of a burst of gamma radiation, starting at 201247 UT, 1972, May 14 are reported. The measurements were made with a 0.5 m2 actively shielded scintillator which was the front element of a double Compton telescope, during a balloon flight at an altitude of 5 mb from Palestine, Texas. The maximum intensity of the burst was 0.10±0.02 cm-2 s-1 above 1 MeV, and the duration was 3.5±0.4 min. The burst intensity during the first 2 min is constant to within 10%. No known electronic or detector malfunction could have produced this effect, and we believe it is a real event.

We have treated in detail the theory of gamma-ray line production in solar flares. The strongest line, both predicted theoretically and detected observationally at 2.2 MeV, is due to neutron capture by protons in the photosphere. The neutrons are produced in nuclear reactions of flare accelerated particles which also produce positrons and prompt nuclear gamma rays. From the comparison of the observed and calculated intensities of the lines at 4.4 or 6.1 MeV to that of the 2.2 MeV line it is possible to deduce the spectrum of accelerated nuclei in the flare region; and from the absolute intensities of these lines it is possible to obtain the total number of accelerated nuclei at the Sun. The study of the 2.2 MeV line also gives information on the amount of He3 in the photosphere. The study of the line at 0.51 MeV resulting from positron annihilation complements the data obtained from the other lines; in addition it gives information on the temperature and density in the annihilation region and on the anisotropy of the accelerated electron beam which produces continuum gamma rays at energies greater than about 1 MeV.

Because ∼5–100 keV electrons are frequently accelerated and emitted by the Sun in small flares, it is possible to define a detailed characteristic physical picture of these events. This review summarizes both the direct spacecraft observations of non-relativistic solar electrons, and observations of the X-ray and radio emission generated by these particles at the Sun and in the interplanetary medium. These observations bear on the basic astrophysical process of particle acceleration in tenuous plasmas. We find that in many small solar flares the ∼5–100 keV electrons accelerated during flash phase constitute the bulk of the total flare energy. Thus the basic flare mechanism in these flares essentially converts the available flare energy into fast electrons. These electrons may produce the other flare electromagnetic emissions through their interactions with the solar atmosphere. In large proton flares these electrons may provide the energy to eject material from the Sun and to create a shock wave which could then accelerate nuclei and electrons to much higher energies.

This paper presents a comprehensive review of the up to date knowledge on nuclear species of Z>2 from solar flares. It covers the following five topics:

1. (I) Solar flare particles of energies >15 MeV n-1 and solar composition;

2. (II) solar flare particles of energies >15 MeV n-1, the enrichment of heavier elements;

3. (III) theoretical interpretation of the enrichment;

4. (IV) the charge states of solar particles; and

5. (V) isotopic abundances of solar flare particles.

During the Skylab mission, many solar flares were observed with the NRL XUV spectroheliogram in the wavelength region from 150 to 650 Å. Because of its high spatial resolution (∼2″) the three-dimensional structures of the flare emission regions characterized by temperatures from 104 K to 20 × 106 K can be resolved. Thus the spatial relationship between the relatively cool plasma and the hot plasma components of a flare, and the associated magnetic field structure can be inferred. For example, the Fe XXIV plasma (T≃20 × 106 K) observed near the soft X-ray maximum during the 2B (M2) disk flare of 1973, July 15 is elongated perpendicular to the neutral line shown on the photospheric magnetogram, while lines of lower ionization temperatures such as He II–Fe XVI show the familiar double ribbon structures on either side of the neutral line (Widing and Cheng, 1974). The disk flare of 1973, September 5 shows the same spatial structures. The flare was observed at 1831 UT at the X-ray maximum, and shows that the hot Fe XXIV cloud is located spatially between the ribbons of the He II, Fe XIV–Fe XVI emissions. As the flare cools at 1837 UT, the Fe XXIV cloud disappears while the gap is filled with emissions from ions of lower ionization potentials, and exhibit loop structure. Many other flares also show similar spatial distributions in the XUV emissions. The linear dimensions of the Fe XXIV plasmas as measured from the photospheric plates range from 7000 km to 14000 km, and the heights of associated loops range from 10000 to 30000 km. From the spatial distributions of the XUV emissions of the many flares, and the comparisons with the magnetograms, we concluded that the magnetic field configuration for the flares we observed are simple bipolar magnetic flux loops with the hot flare plasma located near the top and the cool plasma component on the footprints of the loop.

In many small solar flares the ∼10–100 keV electrons accelerated during the flash phase contain the bulk of the total flare energy output. In large flares, such as those in the period 1972, August 2–7, the flash phase electrons are present in substantially greater numbers. These electrons can explosively heat the chromosphere-lower corona and eject flare material. The ejected matter can produce a shock wave which will then accelerate nucleons and electrons to relativistic energies. We analyze energetic particle, radio, X-ray, gamma ray and interplanetary shock observations of the 1972 August flares to obtain quantitative estimates of the energy contained in each facet of these large flares. In general these observations are consistent with the above hypothesis. In particular:

1. (1) From the X-ray emission (van Beek et al., 1973) the energy contained in >25 keV electrons is calculated to be 2 × 1032 erg for the 1972, August 4 event. Since the lower energy cutoff to the electron spectrum is known to be below 25 keV and possibly below 10 keV, the electrons contain enough energy to produce the following interplanetary shock wave, which has by far the bulk of the energy dissipated in the flare. Similar numbers are obtained for the large August 7 flare event.

2. (2) From the γ-ray emission (Chupp et al., 1973) the energy in protons dumped at the same level of the atmosphere, assuming a thick target situation, is at least a factor of three smaller than the electrons. Moreover the γ-ray emission indicates that the bulk of the protons are accelerated at least several minutes after the electrons. Thus it is more likely that the electrons are responsible for the flare optical (Hα and white light) emissions which occur in the chromosphere.

3. (3) Approximately 5% of the electrons and 99% of the protons escape into the interplanetary medium to be observed by spacecraft. This situation is consistent with the hypothesis of shock acceleration of the protons high in the solar corona.

4. (4) The four most intense X-ray bursts observed during the period July 31–August 11 are the only bursts followed by an interplanetary shock wave and a new injection of energetic protons into the interplanetary medium.