Dynamics and energetics of the thermal and nonthermal components in the solar flare of January 20, 2005, based on data from hard electromagnetic radiation detectors onboard the CORONAS-F satellite

Based on data from the SONG and SPR-N multichannel hard electromagnetic radiation detectors onboard the CORONAS-F space observatory and the X-ray monitors onboard GOES satellites, we have distinguished the thermal and nonthermal components in the X-ray spectrum of an extreme solar flare on January 20, 2005. In the impulsive flare phase determined from the time of the most efficient electron and proton acceleration, we have obtained parameters of the spectra for both components and their variations in the time interval 06:43–06:54 UT. The spectral index in the energy range 0.2–2 MeV for a single-power-law spectrum of accelerated electrons is shown to have been close to 3.4 for most of the time interval under consideration. We have determined the time dependence of the lower energy cutoff in the energy spectrum of nonthermal photons E _γ0(t) at which the spectral flux densities of the thermal and nonthermal components become equal. The power deposited by accelerated electrons into the flare volume has been estimated using the thick-target model under two assumptions about the boundary energy E ₀ of the electron spectrum: (i) E ₀ is determined by E _γ0(t) and (ii) E ₀ is determined by the characteristic heated plasma energy (≈5kT (t)). The reality of the first assumption is proven by the fact that plasma cooling sets in at a time when the radiative losses begin to prevail over the power deposited by electrons only in this case. Comparison of the total energy deposited by electrons with a boundary energy E _γ0(t) with the thermal energy of the emitting plasma in the time interval under consideration has shown that the total energy deposited by accelerated electrons at the beginning of the impulsive flare phase before 06:47 UT exceeds the thermal plasma energy by a factor of 1.5–2; subsequently, these energies become approximately equal and are ～(4–5) × 10³⁰ erg under the assumption that the filling factor is 0.5–0.6.     

Energetics of a double flare on November 8, 1980

Here we complete an energy balance analysis of a double impulsive hard X-ray flare. From spatial observations, we deduce both flares probably occur in the same loop within the resolution of the data. For the first flare, the energy in the fast electrons (assuming a thick-target model) is comparable to the convective up-flow energy, suggesting that these are related successive modes of energy storage and transfer. The total energy lost through radiation and conduction, 2.0 × 10²⁸ erg, is comparable to the energy in fast electrons 2.5 × 10²⁸ erg. For the second flare, the energy in the fast electrons is more than one order of magnitude greater than the energy of the convective up-flow. Total energy losses are within a factor of two lower than the calculated fast electron energy. We interpret the observations as showing that the first flare occurred in a small loop with fast electrons heating the chromosphere and resulting in chromospheric evaporation increasing the density in the loop. For the second flare most of the heating occurred at the electron acceleration site. The two symmetrical components of the Ca xix resonance line and a high velocity down-flow of 115 km s ^–1 observed at the end of the second hard X-ray burst are consistent with the flare eruption (reconnection) region being high in the flare loop. The estimated altitude of the acceleration site is 5500 km above the photosphere.     

On the threshold effect of proton acceleraton in solar flares

Based on the reconnection theory of a flare and on recent observational and statistical findings, the problem of the initial acceleration of solar cosmic rays (SCR) is discussed. Simple estimates of the electric fields required to start the electron acceleration are obtained and the problem of proton ionization losses for overcoming the Coulomb barrier is considered. We take into account also the possible differences between proton and electron spectra from the very beginning of the acceleration process. Special attention is paid to the distribution functions of solar flare events in various parameters (peak fluxes and/or energy fluences in X-ray and radio wave bursts, in proton and electron emissions, etc.). It is shown that the distribution functions allow the interpretation of some scale and time flare parameters in terms of expected threshold effects. However, these functions are still insufficient to evaluate the relative share of different emissions in the global energy budget of a flare. In this context, a more promising approach is to derive the direct ratio between the number of accelerated protons,N _p, and total flare energy,W _f, within the frame of a certain acceleration model. It is argued that an absolute threshold for proton production (in Hudson's formulation) does not exist. Meanwhile, the flux and threshold energy of accelerated protons overcoming the Coulomb loss maximum, in fact, may depend heavily on the global output of flare energy.     

X-ray imaging of three flares during the impulsive phase

The impulsive phases of three flares that occurred on April 10, May 21, and November 5, 1980 are discussed. Observations were obtained with the Hard X-ray Imaging Spectrometer (HXIS) and other instruments aboard SMM, and have been supplemented with Hα data and magnetograms. The flares show hard X-ray brightenings (16–30 keV) at widely separated locations that spatially coincide with bright Hα patches. The bulk of the soft X-ray emission (3.5–5.5 keV) originates from in between the hard X-ray brightenings. The latter are located at different sides of the neutral line and start to brighten simultaneously to within the time resolution of HXIS. Concluded is that:

1. The bright hard X-ray patches coincide with the footpoints of loops.
2. The hard X-ray emission from the footpoints is most likely thick target emission from fast electrons moving downward into the dense chromosphere.
3. The density of the loops along which the beam electrons propagate to the footpoints is restricted to a narrow range (10⁹ < n < 2 × 10¹⁰ cm^-3), determined by the instability threshold of the return current and the condition that the mean free path of the fast electrons should be larger than the length of the loop.
4. For the November 5 flare it seems likely that the acceleration source is located at the merging point of two loops near one of the footpoints.

It is found that the total flare energy is always larger than the total energy residing in the beam electrons. However, it is also estimated that at the time of the peak of the impulsive hard X-ray emission a large fraction (at least 20%) of the dissipated flare power has to go into electron acceleration. The explanation of such a high acceleration efficiency remains a major theoretical problem.     

First phase acceleration mechanisms and implications for hard X-ray burst models in solar flares

Requirements for the number of nonthermal electrons which must be accelerated in the impulsive phase of a flare are reviewed. These are uncertain by two orders of magnitude depending on whether hard X-rays above 25 keV are produced primarily by hot thermal electrons which contain a small fraction of the flare energy or by nonthermal streaming electrons which contain > 50% of the flare energy. Possible acceleration mechanisms are considered to see to what extent either X-ray production scenario can be considered viable. Direct electric field acceleration is shown to involve significant heating. In addition, candidate primary energy release mechanisms to convert stored magnetic energy into flare energy, steady reconnection and the tearing mode instability, transfer at least half of the stored energy into heat and most of the remaining energy to ions. Acceleration by electron plasma waves requires that the waves be driven to large amplitude by electrons with large streaming velocities or by anisotropic ion-acoustic waves which also require streaming electrons for their production. These in turn can only come from direct electric field acceleration since it is shown that ion-acoustic waves excited by the primary current cannot amplify electron plasma waves. Thus, wave acceleration is subject to the same limitations as direct electric field acceleration. It is concluded that at most 0.1% of the flare energy can be deposited into nonthermal streaming electrons with the energy conversion mechanisms as they have been proposed and known acceleration mechanisms. Thus, hard X-ray production above 10 keV primarily by hot thermal electrons is the only choice compatible with models for the primary energy release as they presently exist.     

Spatial and temporal evolution of soft and hard X-ray emission in a solar flare

We study the spatial and temporal characteristics of the 3.5 to 30.0 keV emission in a solar flare on April 10, 1980. The data were obtained by the Hard X-ray Imaging Spectrometer aboard the Solar Maximum Mission Satellite. It is complemented in our analysis with data from other instruments on the same spacecraft, in particular that of the Hard X-ray Burst Spectrometer.Key results of our investigation are: (a) Continuous energy release is needed to substain the increase of the emission through the rising phase of the flare, before and after the impulsive phase in hard X-rays. The energy release is characterized by the production of hot (5 × 10⁷ T 1.5 × 10⁸ K) thermal regions within the flare loop structures. (b) The observational parameters characterizing the impulsive burst show that it is most likely associated with non-thermal processes (particle acceleration). (c) The continuous energy release is associated with strong chromospheric evaporation, as evidenced in the spectral line behavior determined from the Bent Crystal Spectrometer data. Both processes seem to stop just before flare maximum, and the subsequent evolution is most likely governed by the radiative cooling of the flare plasma.     

Energies of Electrons Accelerated in Turbulent Reconnecting Current Sheets in Solar Flares

Yohkoh observations strongly suggest that electron acceleration in solar flares occurs in magnetic reconnection regions in the corona above the soft X-ray flare loops. Unfortunately, models for particle acceleration in reconnecting current sheets predict electron energy gains in terms of the reconnection electric field and the thickness of the sheet, both of which are extremely difficult to measure. It can be shown, however, that application of Ohm's law in a turbulent current sheet, combined with energy and Maxwell's equations, leads to a formula for the electron energy gain in terms of the flare power output, the magnetic field strength, the plasma density and temperature in the sheet, and its area. Typical flare parameters correspond to electron energies between a few tens of keV and a few MeV. The calculation supports the viewpoint that electrons that generate the continuum gamma-ray and hard X-ray emissions in impulsive solar flares are accelerated in a large-scale turbulent current sheet above the soft X-ray flare loops.     

Pulsations in solar hard X-ray bursts

We have analyzed the hard X-ray emission from 28 large solar events, searching for pulsations in intensity profiles. Periodicity occurred in 26 events, usually soon after the onset, with periods in the range 10–100 s. Pulsations occurring at common frequencies in different energy bands are observed to be closely in phase. Periodic behavior in hard X-ray emission is related to that at microwave and decametric wavelength. We discuss our observations briefly in terms of two models: that of McClean et al. (1971), applied to X-ray emission, and that of Brown and Hoyng (1975). As periodicity is normal in extended hard X-ray bursts and occurs through a broad energy band, it is probably directly related to a principal flare acceleration mechanism. Our observations constrain possible mechanisms of flare acceleration and physical properties of the acceleration region.This work began when the author was at the Institute for Plasma Research, Stanford University.     

Multi-Wavelength Analysis of the Flare on 2 October 1993

By use of Yohkoh hard X-ray flux and soft X-ray images, and of vector magnetograms and 2D spectral observations, a 1N/C6.5 flare observed on 2 October 1993 is analysed in detail. Evidence is provided not only morphologically but also quantitatively that the dynamics at kernels A and C of the flare in the impulsive phase were controlled mainly by electron beam bombardment, while the heating of kernel B is mainly due to heat conduction. By plotting the energy gradient of the electron energy flux as a function of energy for the various spectral indexes observed during the flare, the acceleration mechanism is found to be such that there is a constant energy E₀, close to 20 keV, for which the electron flux d F₁/dE is constant. It is shown that such a conclusion can be reached more directly by using the photon flux, which in that case must be constant for E=E₀, whatever the value of the power index. This result implies also that the electron spectrum is represented by a power law and that the X-ray photons are produced in a thick target. Instantaneous momentum balance is shown to exist between the upflowing soft X-ray-emitting and the downflowing Hα- emitting plasma at the kernels of the flare. The observed Hα red asymmetry is well reproduced by the non-LTE computation, with the down-moving condensation included. The observation of the magnetic field suggests that the flare was triggered probably by magnetic flux emergence.     

Correlation between CME and Flare Parameters (with and without Type II Bursts)

CMEs and flares are the two energetic phenomena on the Sun responsible for generating shocks. Our main aim is to study the relation between the physical properties of CMEs and flares associated with and without type II radio bursts. We considered a set of 290 SOHO/LASCO CMEs associated with GOES X-ray flares observed during the period from January 1997 to December 2000. The relationship between the flares and CMEs is examined for the two sets i) with metric-type IIs and ii) without metric-type IIs. Physical properties such as rise time, duration, and strength of the flares and width, speed, and acceleration of CMEs are considered. We examined the energy relationship and temporal relationship between the CMEs and flares. First, all the events in each group were considered, and then the limb events in each group were considered separately. While there is a relationship between the temporal characteristics of flares and CME properties in the case of with-type IIs, it is absent in the case of all without-type IIs. Among all the relations studied, the correlation between flare duration and CME properties is found to be highly significant compared to the other relations. Also, the relationship between flare strength and CME speed found in the with-type II events is absent in the case of all without-type II events. However, when the limb without-type II events (with reduced time window between flare and CME) are studied separately, we found the energy relationship and the temporal relationship.     

Coronal Mass Ejection of 15 May 2001: II. Coupling of the Cme Acceleration and the Flare Energy Release

We analyze the relationship between the dynamics of the coronal mass ejection (CME) of 15 May 2001 and the energy release in the associated flare. The flare took place behind the east limb and was disclosed by a growing system of hot soft X-ray (SXR) loops that appeared from behind the limb around the onset of the rapid acceleration of the CME. The highly correlated behavior of the SXR light-curve derivative and the time profile of the CME acceleration reveals an intrinsic relationship between the CME dynamics and the flare energy release. Furthermore, we found that the CME acceleration peak occurs simultaneously with the fastest growth (100 km s^-1) of X-ray loops, indicating that the reconnection plays an essential role in the eruption. Inspecting the CME/flare morphology we recognized in the Yohkoh-SXT images an oval feature that formed within the rising structure at the onset of the rapid acceleration phase, simultaneously with the appearance of the X-ray loops. The eruptive prominence was imbedded within the lower half of the oval, suggestive of a flux-rope/prominence magnetic configuration. We interpret the observed morphological evolution in terms of a reconnection process in the current sheet that presumably formed below the erupting flux-rope at the onset of the CME acceleration. Measurements of the tip-height of the cusped X-ray loop system and the height of the lower edge of the oval, enable us to trace the stretching of the current sheet. The initial distance between the oval and the loops amounted to 35 – 40 Mm. In about 1 h the inferred length of the current sheet increased to 150 – 200 Mm, which corresponds to a mean elongation speed of 35 – 45 km s^-1. The results are discussed in the framework of CME models that include the magnetic reconnection below the erupting flux-rope.     

Radiative and conductive cooling in a solar flare

We investigate the radiative and conductive cooling in the solar flare observed by RHESSI on 2005 September 13. The radiative and conductive loss energies are estimated from the observations after the flare onset. Consistent with previous findings, the cooling is increased with time, especially the radiation becomes remarkable on the later phase of flare. According our method, about half of thermal energy is traced by RHESSI soft X-rays, while the other half is lost by the radiative (∼38%) and conductive (∼9%) cooling at end of the hard X-rays in this event. The nonthermal energy input of P _nth (inferred from RHESSI hard X-ray spectrum) is not well correlated with the derivative of thermal energy of \fracdE_thdt\frac{\mathrm{d}E_{\mathrm{th}}}{\mathrm{d}t} (required to radiate the RHESSI soft X-ray flux and spectrum) alone. However, after consideration the radiation and conduction, a high correlation is obtained between the derivative of total thermal energy ( \fracdE_th+E_rad+E_conddt\frac{\mathrm{d}E_{\mathrm{th}}+E_{\mathrm{rad}}+E_{\mathrm{cond}}}{\mathrm{d}t}) and nonthermal energy input (P _nth) from the flare start to end, indicating the relative importance of conductive and direct radiative losses during the solar flare development. Ignoring the uncertainties to estimate the energy from the observations, we find that about ∼12% fraction of the known energy is transferred into the thermal energy for the 2005 September 13 flare.     

The emission and propagation of ∼ 40keV solar flare electrons

Observations of prompt 40 keV solar flare electron events by the IMP series of satellites in the period August, 1966 to December, 1967 are tabulated along with prompt energetic solar proton events in the period 1964–1967. The interrelationship of the various types of energetic particle emission by the sun, including relativistic energy electrons reported by Cline and McDonald (1968) are investigated. Relativistic energy electron emission is found to occur only during proton events. The solar optical, radio and X-ray emission associated with these various energetic particle emissions as well as the propagation characteristics of each particle species are examined in order to study the particle acceleration and emission mechanisms in a solar flare. Evidence is presented for two separate particle acceleration and/or emission mechanisms, one of which produces 40 keV electrons and the other of which produces solar proton and possibly relativistic energy electrons. It is found that solar flares can be divided into three categories depending on their energetic particle emission: (1) small flares with no accompanying energetic phenomena either in particles, radio or X-ray emission; (2) small flares which produce low energy electrons and which are accompanied by type III and microwave radio bursts and energetic ( 20 keV) X-ray bursts; and (3) major solar flare eruptions characterized by energetic solar proton production and type II and IV radio bursts and accompanied by intense microwave and X-ray emission and relativistic energy electrons.     

Extreme Ultraviolet Late-Phase Flares: Before and During the Solar Dynamics Observatory Mission

The solar extreme-ultraviolet (EUV) observations from the Solar Dynamics Observatory (SDO) have revealed interesting characteristics of warm coronal emissions, such as Fe xvi 335 Å emission, which peak soon after the hot coronal X-ray emissions peak during a flare and then sometimes peak for a second time hours after the X-ray flare peak. This flare type, with two warm coronal emission peaks but only one X-ray peak, has been named the EUV late phase (Woods et al., Astrophys. J. 739, 59, 2011). These flares have the distinct properties of i) having a complex magnetic-field structure with two initial sets of coronal loops, with one upper set overlaying a lower set, ii) having an eruptive flare initiated in the lower set and disturbing both loop sets, iii) having the hot coronal emissions emitted only from the lower set in conjunction with the X-ray peak, and iv) having the first peak of the warm coronal emissions associated with the lower set and its second peak emitted from the upper set many minutes to hours after the first peak and without a second X-ray enhancement. The disturbance of the coronal loops by the eruption is at about the same time, but the relaxation and cooling down of the heated coronal loops during the post-flare reconnections have different time scales with the longer, upper loops being significantly delayed from the lower loops. The difference in these cooling time scales is related to the difference between the two peak times of the warm coronal emission and is also apparent in the decay profile of the X-ray emissions having two distinct decays, with the first decay slope being steeper (faster) and the delayed decay slope being smaller (slower) during the time of the warm-coronal-emission second peak. The frequency and relationship of the EUV late-phase decay times between the Fe xvi 335 Å two flare peaks and X-ray decay slopes are examined using three years of SDO/EUV Variability Experiment (EVE) data, and the X-ray dual-decay character is then exploited to estimate the frequency of EUV late-phase flares during the past four solar cycles. This study indicates that the frequency of EUV late-phase flares peaks before and after each solar-cycle minimum.     

X-Ray bursts from solar flares behind the limb

From the UCSD OSO-7 X-ray experiment data, we have identified 54 X-ray bursts with 5.1–6.6 keV flux greater than 10³ photon cm^?2 keV^?1 which were not accompanied by visible Hα flare on the solar disk. By studying OSO-5 X-ray spectroheliograms, Hα activity at the limb and the emergence and disappearance of sunspot groups at the limb, we found 17 active centers as likely seats of the X-ray bursts beyond the limb. We present the analysis of 37 X-ray bursts and their physical parameters. We compare our results with those published by Datlowe et al. (1974a, b) for disk events. The distributions of maximum temperature, maximum emission measure, and characteristic cooling time of the over-the-limb events do not significantly differ from those of disk events. We show that of conduction and radiation, the former is the dominant cooling mechanism for the hot flare plasma. Since the disk and over-the-limb bursts are similar, we conclude that the scale height for X-ray emission in the 5–10 keV range is large and is consistent with that of Catalano and Van Allen (1973), 11000 km, for primarily 1–3 keV emission. Twenty-five or about 2/3 of the over-the-limb events had a non-thermal component. The distribution of peak 20 keV flux is not significantly different from that of disk events. However, the spectral index at the time of maximum flux is significantly different for events over the limb and for events near the center of the disk; the spectral index for over-the-limb events is larger by about δγ = 3/4. If hard X-ray emission came only from localized sources low in the chromosphere we would expect that hard X-ray emission, would be occulted over the limb; on the contrary, the observation show that the fraction of soft X-ray bursts which have a nonthermal component is the same on and off of the disk. Thus hard X-ray emission over extended regions is indicated.     

10–100 keV electron acceleration and emission from solar flares

We present an analysis of spacecraft observations of non-thermal X-rays and escaping electrons for 5 selected small solar flares in 1967. OSO-3 multi-channel energetic X-ray measurements during the non-thermal component of the solar flare X-ray bursts are used to derive the parent electron spectrum and emission measure. IMP-4 and Explorer-35 observations of > 22 keV and > 45 keV electrons in the interplanetary medium after the flares provide a measure of the total number and spectrum of the escaping particles. The ratio of electron energy loss due to collisions with the ambient solar flare gas to the energy loss due to bremsstrahlung is derived. The total energy loss due to collisions is then computed from the integrated bremsstrahlung energy loss during the non-thermal X-ray burst. For > 22 keV flare electrons the total energy loss due to collisions is found to be 10⁴ times greater than the bremsstrahlung energy loss and 10² times greater than the energy loss due to escaping electrons. Therefore the escape of electrons into the interplanetary medium is a negligible energetic electron loss mechanism and cannot be a substantial factor in the observed decay of the non-thermal X-ray burst for these solar flares.We present a picture of electron acceleration, energy loss and escape consistent with previous observations of an inverse relationship between rise and decay times of the non-thermal X-ray burst and X-ray energy. In this picture the acceleration of electrons occurs throughout the 10–100 sec duration of the non-thermal X-ray burst and determines the time profile of the burst. The average energy of the accelerated electrons first rises and then falls through the burst. Collisions with the ambient gas provide the dominant energetic electron loss mechanism with a loss time of 1 sec. This picture is consistent with the ratio of the total number of energetic electrons accelerated in the flare to the maximum instantaneous number of electrons in the flare region. Typical values for the parameters derived from the X-ray and electron observations are: total energy in > 22 keV electrons total energy lost by collisions = 10^28–29 erg, total number of electrons accelerated above 22 keV = 10³⁶, total energy lost by non-thermal bremsstrahlung = 10²⁴erg, total energy lost in escaping > 22 keV electrons = 10²⁶erg, total number of > 22 keV electrons escaping = 10^33–34.The total energy in electrons accelerated above 22 keV is comparable to the energy in the optical or quasi-thermal flare, implying a flare mechanism with particle acceleration as one of the dominant modes of energy dissipation.The overall efficiency for electron escape into the interplanetary medium is 0.1–1% for these flares, and the spectrum of escaping electrons is found to be substantially harder than the X-ray producing electrons.Currently at Tokyo Astronomical Observatory, Mitaka, Tokyo, Japan.     

Thermal Evolution and Radiative Output of Solar Flares Observed by the EUV Variability Experiment (EVE)

This paper describes the methods used to obtain the thermal evolution and radiative output during solar flares as observed by the Extreme ultraviolet Variability Experiment (EVE) onboard the Solar Dynamics Observatory (SDO). How EVE measurements, due to the temporal cadence, spectral resolution and spectral range, can be used to determine how the thermal plasma radiates at various temperatures throughout the impulsive and gradual phase of flares is presented and discussed in detail. EVE can very accurately determine the radiative output of flares due to pre- and in-flight calibrations. Events are presented that show that the total radiated output of flares depends more on the flare duration than the typical GOES X-ray peak magnitude classification. With SDO observing every flare throughout its entire duration and over a large temperature range, new insights into flare heating and cooling as well as the radiative energy release in EUV wavelengths support existing research into understanding the evolution of solar flares.     

X-ray emission characteristics of flares associated with CMEs

We present the study of 20 solar flares observed by “Solar X-ray Spectrometer (SOXS)” mission during November 2003 to December 2006 and found associated with coronal mass ejections (CMEs) seen by LASCO/SOHO mission. In this investigation, X-ray emission characteristics of solar flares and their relationship with the dynamics of CMEs have been presented. We found that the fast moving CMEs, i.e., positive acceleration are better associated with short rise time (< 150 s) flares. However, the velocity of CMEs increases as a function of duration of the flares in both 4.1–10 and 10–20 keV bands. This indicates that the possibility of association of CMEs with larger speeds exists with long duration flare events. We observed that CMEs decelerate with increasing rise time, decay time and duration of the associated X-ray flares. A total 10 out of 20 CMEs under current investigation showed positive acceleration, and 5 of them whose speed did not exceed 589 km/s were associated with short rise time (< 150 s) and short duration (< 1300 s) flares. The other 5 CMEs were associated with long duration or large rise time flare events. The unusual feature of all these positive accelerating CMEs was their low linear speed ranging between 176 and 775 km/s. We do not find any significant correlation between X-ray peak intensity of the flares with linear speed as well as acceleration of the associated CMEs. Based on the onset time of flares and associated CMEs within the observing cadence of CMEs by LASCO, we found that in 16 cases CME preceded the flare by 23 to 1786 s, while in 4 cases flare occurred before the CME by 47 to 685 s. We argue that both events are closely associated with each other and are integral parts of one energy release system.     

Hard X-ray bursts from flares behind the solar limb

The determination of the location of the region of origin of hard X-rays is important in evaluating the importance of 10–100 keV electrons in solar flares and in understanding flare particle acceleration. At present only limb-occulted events are available to give some information on the height of X-ray emission. In fifteen months of OSO-7 operation, nine major soft X-ray events had no reported correlated Hα flare. We examine the hard X-ray spectra of eight of these events with good candidate X-ray flare producing active regions making limb transit at the time of the soft X-ray bursts. All eight bursts had significant X-ray emission in the 30–44 keV range, but only one had flux at the 3σ level above 44 keV. The data are consistent with most X-ray emission occurring in the lower chromosphere, but some electron trapping at high altitudes is necessary to explain the small nonthermal fluxes observed.