Robert C. Duncan, University of Texas at Austin

Soft gamma repeaters ("SGRs") are X-ray stars that emit bright, repeating flashes of soft (i.e. low-energy) gamma rays. The physical nature of these stars was a mystery for many years. In 1992, it was proposed that SGRs are magnetically-powered neutron stars, or magnetars. Subsequent observational studies lent support to this hypothesis. Astronomers now think that all emissions detected from SGRs and from a related class of stars known as anomalous X-ray pulsars ("AXPs") are powered by magnetic field decay. Here I will explain how these strange, physically-extreme stars form, and why they emit steadily pulsating X-rays with sporadic, bright outbursts. I will also tell the story of their discovery, and of the observational and theoretical efforts which helped to reveal their bizarre nature.

NOTE: this website was originally written in May 1998, in response to a surge of interest in magnetars. In early 2003 I updated the site, to answer questions raised by readers of a Scientific American cover story on magnetars. I made some improvements, and added a few new sections. However, I did not come close to covering all the recent progress in this rapidly-developing area of astrophysics.

- R.D., March 2003

Table of contents
Strong magnetic fields
Discovery of SGRs
SGR Bursts
The March 5th Event
The burster is located
March 5th Flare Theories
Other SGRs
Neutron star magnetic fields
A new kind of star?  
A brief history of magnetars
Trapped fireballs 
A year of magnetar breakthoughs 
The August 27th event 
Magnetar links 
Epilog: new horizons in magnetar science 

Strong Magnetic Fields

  Observations of SGRs give evidence that these stars have extrordinarily strong magnetic fields. One especially important and reliable measurement of SGR magnetism—indeed a crucial test of the theory—was first done by Chryssa Kouveliotou and her collaborators in 1998. This gave evidence for a field of 8 X 1014 Gauss, where "Gauss" is a unit of magnetic field strength. (1 Gauss = 10-4 Tesla)
Here I use scientific notation, so 1014 means 10 raised to the 14th power, which equals 1 with 14 zeros after it. Thus 8 X 1014 is 8 with 14 zeros after it, or 800,000,000,000,000.

To put this in perspective, here is a table of magnetic field strengths
The Earths magnetic field, which deflects compass needles measured at the N magnetic pole 0.6 Gauss
A common, hand-held magnet like those used to stick papers on a refrigerator 100 Gauss
The magnetic field in strong sunspots (within dark, magnetized areas on the solar surface) 4000 Gauss
The strongest, sustained (i.e., steady) magnetic fields achieved so far in the laboratory generated by huge electromagnets 4.5 X 105 Gauss
The strongest man-made fields ever achieved, if only briefly made using focussed explosive charges; lasted only 4 - 8 microseconds. 107 Gauss
The strongest fields ever detected on non-neutron stars found on a handful of strongly-magnetized, compact white dwarf stars.   (Such stars are rare. Only 3% of white dwarfs have Mega-gauss or stronger fields.) 108 Gauss
Typical surface, polar magnetic fields of radio pulsars the most familiar kind of neutron star; more than a thousand are known to astronomers 1012-1013 Gauss
Magnetars soft gamma repeaters and anomalous X-ray pulsars
(These are surface, polar fields. Magnetar interior fields may range up to 1016 Gauss, with some field lines probably wrapped in a toroidal, or donut geometry inside the star.)
1014-1015 Gauss

Physicists have not made steady fields stronger than 4.5 x 105 Gauss in the lab because the magnetic stresses of such fields exceed the tensile strength of terrestrial materials. If you try to make stronger fields, magnetic forces will blow apart your electromagnet.
Using chemical high explosives to drive implosions, it is possible to compress a magnetic field and reach higher field strengths, at least for a tiny fraction of a second. This has been done at Los Alamos Laboratory in the U.S., and at a nuclear weapons lab in Sarov, Russia, attaining fields of about 107 Gauss before the equipment was destroyed.

Atoms in very strong magnetic fields

The strongest magnetic field that you are ever likely to encounter personally is about 104 Gauss if you have Magnetic Resonance Imaging (MRI) scan for medical diagnosis. Such fields pose no threat to your health, hardly affecting the atoms in your body. Fields in excess of 109 Gauss, however, would be instantly lethal. Such fields strongly distort atoms, compressing atomic electron clouds into cigar shapes, with the long axis aligned with the field, thus rendering the chemistry of life impossible. A magnetar within 1000 kilometers would thus kill you via pure static magnetism -- if it didn't already get you with X-rays, gamma rays, high energy particles, extreme gravity, bursts and flares...
In fields much stronger than 109 Gauss, atoms are compressed into thin needles. At 1014 Gauss, atomic needles have widths of about 1% of their length, hundreds of times thinner than unmagnetized atoms. Such atoms can form polymer-like molecular chains or fibers. A carpet of such magnetized fibers probably exist at the surface of a magnetar, at least in places where the surface is cool enough to form atoms.

Grey boxes like this indicate material for advanced readers.

Ultra-strong magnetic fields

Many fascinating physical effects occur in magnetic fields with strength exceeding the "quantum electrodynamic field strength" of BQ = 4.4 X 1013 Gauss. (This field-strength given by a combination of fundamental constants: BQ = me2 c3 / h e, where me is the mass of the electron, c is the speed of light, h is Planck's constant divided by 2 π, and e is the charge on an electron.) In fields stronger than BQ, electrons gyrate at nearly the speed of light around magnetic field lines, even in their lowest quantum energy states. Consequently, the ultra-magnetized vacuum -- which, according to quantum mechanics, seethes with virtual electron-positron pairs and other particles -- becomes birefringent like a calcite crystal, capable of distorting and magnifying images ("magnetic lensing").   X-ray photons traveling through such strong fields readily split into two, or merge together; and many other novel physical effects come into play. Because BQ lies between the field strengths observed in magnetars and in ordinary radio pulsars, this new physics is important only in magnetars, making the theory of magnetars especially rich and interesting. Effects like magnetic photon splitting cannot be measured in terrestrial laboratories, but may be detected in emissions from these unique cosmic laboratores. For more information, see my review article, Physics in Ultra-strong Magnetic Fields.

Although magnetar fields are strong by most measures, they are weak compared to the strongest possible field that could theoretically exist in nature, which is 1049 - 1053 Gauss. A field stronger than this would literally break down the vacuum and decay, via the quantum mechanical process of magnetic monopole creation. However, there is no known way that such strong fields could be generated. Magnetars are the most strongly-magnetized objects yet known in the Universe.

TopBack to top

The Discovery of Soft Gamma Repeaters

The U.S. Dept. of Defense launched the "Vela" satellites in the late 1960s to search for gamma rays in space, in order to verify a treaty banning nuclear tests in outer space. To everyone's surprise, many brief bursts of gamma rays were found. They seemed to come randomly from all directions in the sky. Thus a new natural phenomenon, "gamma-ray bursts" (GRBs), was discovered and announced to astronomers in 1973.

Gamma rays are made up of very high-energy photons, more energetic than X-ray photons, which in turn have higher energy than ultraviolet, optical, infrared, microwave, and radio-frequency photons (listed in order of decreasing energy, or increasing "spectral softness"). Gamma rays can only be studied from outer space, because they are absorbed by the Earth's atmosphere as they ionize atoms. (That is, as they knock electrons out of atoms in the air, turning them into positively-charged ions.) Fortunately, gamma ray detectors are relatively cheap, compact and easy to put on spacecraft, so they were "piggybacked" on various space missions during the 1970's to give data on GRBs. By 1979, there were almost a dozen gamma ray detectors on Earth satellites and on space probes scattered around the solar system.

  1979 was a fantastic year for the study of soft gamma repeaters (SGRs), although nobody realized it at the time. The first SGR burst ever detected, from a source in the constellation Sagittarius, occurred on January 7, 1979. Then a truly powerful SGR outburst---indeed, by far the most intense blast of gamma rays that had ever been detected from outside our solar system (until another SGR outburst broke the record in 1998)---came just two months later, on March 5, 1979. This tremendous flare eventually allowed the SGR mystery to be unraveled. Only nine days after that a third SGR became active in a new part of the Galaxy, giving three bursts in a three day period. So during the first 3 months of 1979, three of the five known SGRs were discovered.

For years, astronomers didn't distinguish between SGR bursts and the much more frequently-observed "ordinary" or "classic" gamma-ray bursts (GRBs). (During the 1990's, with modern detectors, about one GRB per day was detected from some place in the sky. Fewer SGR bursts were seen, perhaps 10 to 50 per year, although they came in groups.) It wasn't until 1987 that SGRs were clearly recognized as a distinct set of objects. The name "soft gamma repeaters" focuses on properties which distinguish SGR bursts from GRBs. Unlike GRBs, which have never been verified to come more than once from the same spot in the sky, SGR bursts repeat sporadically from the same source. The gamma rays in SGR bursts are also "spectrally soft" compared to those in GRBs. This means that the average energy per gamma-ray photon is less. In fact, most SGR photons are really high-energy or "hard" X-rays, not true gamma rays.   A more descriptive name would thus be "hard X-ray flashers," but we are stuck with "soft gamma repeaters" because of the way these objects were historically distinguished from GRB sources.

TopBack to top

SGR Bursts

The "soft" in "soft gamma repeaters" does not mean "faint." Luminosity or brightness, the energy radiated away per second, is related to the number of photons being emitted, times the energy per photon---and the number of photons coming from SGR bursts is enormous. The term "soft" simply means that the energy per photon is less than in GRBs. Note that SGRs are spectrally soft only in comparison to GRBs---they are harder than all other known astronomical phenomena.
Normal SGR bursts can radiate away as much energy in a single second as the Sun does in a whole year. (By "normal" I mean to exclude the giant flares of March 5, 1979 and August 27, 1998 which were more than 1,000 times brighter.) SGR bursts commonly last for only a small fraction of a second, although some last for several seconds. All identified SGRs lie within our galaxy (4 of them) or in a clump of stars just outside our galaxy (1 of them).
[Our galaxy, the Milky Way, is a huge, flattened agglomerate of stars, shaped roughly like a disk about 100,000 light-years in diameter and about 1000 light-years thick, where a light-year is the distance that light travels in one year: about 6 trillion miles. If seen from the outside, the Milky Way would look like a luminous pinwheel, made of about 100 billion (1011) stars. The Sun and Earth lie about 3/4ths of the way out from the center.]
Although only five SGRs have been detected so far, many millions almost certainly exist in our galaxy, and a similar number probably exist in every other galaxy. But SGRs cease emitting bright bursts after only about 10,000 years---a brief instant of cosmic time---so only the youngest few have been detected.
It is interesting to compare SGRs with other repeating burst sources in the Galaxy. Astronomers have identified many such objects, and given them names like: "Type I and Type II X-ray Bursters, Black Hole X-ray Transients, Cataclysmic Variables, and Novae." These bursters are compact stars (white dwarfs, neutron stars, or black holes) into which material is falling from an orbiting companion star, in a double-star system. All of these other kinds of repeating bursts are fainter than the brightest normal SGR bursts by a factor of ~10,000 or more, except for the Black Hole X-ray Transients, which are fainter by only 1,000. However, the bursts from these other sources sometimes last much longer than SGR bursts, so the total energy in a burst can be comparable. (Because of their brevity, "flashers" would have been a good name for the SGRs.)
In summary, SGRs are, by far, the brightest known bursters which repeat. Supernovae and GRBs are much brighter still, but they are one-shot events, destroying the bursting star. Supernovae and GRBs are rare, occuring in our Galaxy only once in few hundred years (supernovae) or once in perhaps a million years (GRBs). This means that almost all detected supernovae and GRBs come from other, distant galaxies.
This brings us to the pivotal event in the history of SGRs: an outburst that was, for a brief time, brighter than a supernova.
TopBack to top

The March 5th Event

March 5th, 1979 ....
At 10:51 A.M. Eastern Standard Time, far out in space, two Soviet interplanetary space probes, Venera 11 and Venera 12, were drifting through the inner solar system when they were walloped by an unprecedented flux of gamma rays. Onboard gamma ray detectors jumped from 100 to 40,000 counts and then off-scale in a fraction of a millisecond---first on Venera 11, then 5 seconds later on Venera 12. The detectors had not been designed for such a flood of energy and they "saturated," losing count of the gamma rays pouring through them. Eleven seconds later, more gamma rays blasted an American space probe, Helios 2, in orbit around the Sun, also knocking its detectors off scale.
A plane wavefront of gamma rays was evidently sweeping through the Solar system at the speed of light. It soon reached Venus, where the Pioneer Venus Orbiter's gamma ray detector also went over the top. Then only 7 seconds later it reached Earth. Nobody noticed as it passed: life went on calmly beneath the protective atmosphere. It was a rainy, dreary, cold Monday morning on the U.S. East coast; chilly and clear elsewhere in the country. The lead story in U.S. newspapers concerned President Carter's attempts to advance an Israeli-Egyptian peace treaty. (I was 4000 miles across the Atlantic Ocean then, a young student at Cambridge University.)
  Meanwhile up in Earth orbit, three Vela satellites and a Soviet satellite named Prognoz 7 were swamped with a sudden flood of gamma rays. The Einstein X-ray Observatory, an orbiting X-ray telescope, also showed a strong signal. Gamma rays were diffusing copiously through the metal radiation shields surrounding its detectors.
  As the wavefront passed out of the solar system, it hit one more space probe: the International Sun-Earth Explorer (ISEE) in orbit around the gravitational null or Lagrangian point of the Sun-Earth system. (A few years later this probe left the Lagrangian point, and was sent drifing through the wider solar system in an effort to study comets, at which time it was renamed the International Cometary Explorer or ICE.) The gamma-ray detector on ISEE was pointed away from the oncoming gamma rays, but they passed through the solid body of the spacecraft, partially scattered and absorbed and were still able to kick the detector up to maximum. Sixteen years later a team of Los Alamos scientists would make elaborate computer simulations of gamma rays passing through the ISEE spacecraft in an attempt to extract more information about this intense burst.

Light curve of the March 5th event, as recorded by gamma-ray detectors aboard the Venera 12 space probe. (From E.P. Mazets et al., 1979, Nature 282, p. 587.)

All the detectors showed that the burst began with a "hard pulse" of gamma rays lasting 0.2 seconds. This pulse was about 100 times more intense than any burst of cosmic gamma rays that had been detected up to that time. Nineteen years later, it still held the record, by about a factor of 10. The hard pulse saturated the detectors. It was followed by a much fainter "soft tail" of soft gamma rays (or hard X-rays), lasting over 3 minutes, steadily fading. As it faded, the soft tail also varied in intensity in something like a sine wave, but with two peaks per cycle, and with a cycle period of 8.0 s. The 8-second modulations were clearly observed by many different detectors for more than 20 cycles. Nothing like this soft tail had ever been seen by astronomers, or was seen again for 19 years.
Fourteen and a half hours later, at 1:17 A.M. E.S.T. on March 6th, another, a fainter burst came from the same spot in the sky, lasting only 1.5 seconds. In retrospect we can see that it was a normal SGR burst in all its properties. Then, a month later on April 4th and again on April 24, more SGR-type bursts came, each lasting about 0.2 second. Over the next four years, 16 SGR-type bursts were seen from this source. Then in May 1983, the bursting ceased. No bursts have been detected from this source since.
Many people suggested that the SGR-like bursts were a residual effect of the huge March 5th event, perhaps a sign that the burster was "settling" into its post-burst state. Russian astrophysicists noted that spectrum of the soft tail of the March 5th event---that is, the distribution of energies of the detected hard X-ray photons---was almost identical to the spectrum of the SGR-like bursts which followed. Thus the soft tail could be considered a "super long-duration" SGR-type event, although the hard initial pulse was unique to March 5, 1979.
TopBack to top

The burster is located

In the months and years after March 5, 1979, scientists analyzed data from the different spacecraft. Each detector had a clock that tagged the time on when the gamma rays first hit, to the nearest millisecond. By comparing these times from spacecraft at different places in the solar system, astronomers were able to tell at what angle the plane wavefront of gamma rays had passed through the solar system. This in turn told them where in the sky the burst came from. It took more than a year to do this accurately. The result was a huge surprise.

An X-ray map of the supernova remnant N49 in the Large
Magellanic Cloud. This false-color image (with different colors
indicating different intensities of X-rays) was made using
data from the ROSAT X-ray telescope. The long, skinny,
white box in the upper center shows the position of the
1979 March 5th burster, as inferred from the arrival times
of the hard pulse at 7 different spacecraft. The point source
of X-rays near the center of the white box is the SGR (now
thought to be a magnetically-powered X-ray star). All the other
X-rays come from gas heated by the supernova explosion.
For an optical image of the same supernova remnant,
see this Hubble Space Telescope photo.
The source turned out to lie inside the tiny area of the sky which is covered by a "supernova remnant": the glowing cloud of hot gas left over from a massive stellar explosion. However, this particular supernova remnant (SNR), with the catalog name N49, is not in our own Milky Way galaxy. Instead, SNR N49 is in a "dwarf satellite galaxy" of the Milky Way called the Large Magellanic Cloud (LMC). The LMC is an irregular knot of stars that is prominent in the sky in the Southern Hemisphere. It is one of the nearest clumps of stars outside our galaxy, 180,000 light years from Earth. The LMC is called a "dwarf satellite galaxy" because it is a small galaxy that orbits the Milky Way.

Supernovae are common in the LMC; in fact, one was observed to go off there in February 1987---"Supernova 1987A."

Could the March 5th burster actually be much closer to us than the LMC? Almost certainly not. This would require that it just happens to have a position that overlaps with the tiny SNR in the LMC, which would be a tremendously improbable coincidence. Thus there is little doubt that the source was actually in the LMC, 180,000 light-years, or 1018 miles, away.

This was a shock. Everyone had expected that the source would be in the near galactic neighborhood, at most a few hundred light years away.
This meant that the burst actually occured 180,000 years ago, long before the dawn of history, but it took this long for the gamma rays to reach us. The "plane wavefront" passing through the solar system was actually part of an expanding sphere of radiation, 180,000 light years in radius; it only seemed flat locally because the sphere's radius was huge compared to the size of the solar system.
The fact that the source is so far away means that the burst was enormously bright, intrinsically. At its peak the burster was shining about 10 times brighter than all the stars in our galaxy put together, or about 10 times brighter than a supernova explosion at its peak photon brightness. (Note that galactic stars and supernovae both radiate mostly optical & UV photons, whereas the March 5th burster radiated mostly gamma rays; but the energy loss rates can be compared.)
In the first two-tenths of a second, the burster radiated away as much energy as the Sun radiates in 3000 years.
There was one more tantalizing clue... The position of burster, as precisely determined using data from 7 different spacecraft, did not lie at the center of the spherical SNR, but significantly displaced toward the edge. (See the figure above.) This displacement was verified in 1991 when a faint, steady "point source" of X-rays was found at the position of the burster, allowing its position to be accurately measured. (These X-rays are evidently emitted by the burster. Astronomers call it a "point source" because its size and shape are not measured: it is so small that it is indistinguishable from a point with present X-ray telescopes.)
TopBack to top

March 5th Burst Theories

What caused the March 5th event? Assuming that the March 5 burster formed in the supernova which gave rise to SNR N49, as seems likely, then we can infer that....
the March 5 burster has the following properties.
The burster is probably
a neutron star, because
neutron stars are known
to form inside supernovae
(e.g., the Crab pulsar inside
the Crab SNR).
A neutron star is an ultra-dense ball of neutrons, like a giant atomic nucleus floating and spinning in space. It is about 10 kilometers in radius, with a weight comparable to that of the Sun. Such a star forms when a massive, ordinary star depletes its supply of nuclear fuel. The core of the star collapses inward under its own gravity, becoming a neutron star, while the rest of the star is blown away in a supernova.
The burster is young, at least on astronomical time scales: than 10,000 years old. Astronomers can estimate the age of a SNR, that is, how long ago the supernova occured. This is done by measuring properties of the SNR, like its size and rate of expansion. The SNR containing the March 5th burster was roughly 5000 years old when the burst occurred.
The burster was born moving at a high velocity. Assuming that the burster formed in the center of the supernova explosion, as seems likely, then it must have acquired a recoil or kick velocity of about 1000 kilometers per second, in order to move it off-center during the roughly 5000 years that went by before the burst occurred. This is fast for a neutron star, but not unreasonably fast.
It is probably an isolated neutron star, i.e. not in a binary star system. Evidently, the neutron star acquired its high velocity when it formed. This means that it is probably not bound (by gravity) to orbit around another star. If it had been bound when it formed in the supernova, its high velocity would have broken it free.
The 8-second modulation seems to indicate that the star is rotating once every 8.0 seconds. This is quite slow for an isolated, young neutron star. For example, the Crab pulsar, a well-known neutron star in a young SNR, rotates once every 33 milliseconds, or 30 times per second. (When neutron stars form in stellar core collapse, shrinking suddenly by a factor of 300, they twirl up to high speeds for the same reason that an ice skater spins faster when pulling in her arms.)
The point source of X-rays indicates that the neutron star is steadily giving off energy, from some (unknown) energy source.
Nobody understood why a neutron star would have this strange set of properties, or what would cause it to burst so spectacularly.
Many theories were proposed in the 1980s, suggesting, for example, that the March 5th event was due to a small planet or a large asteroid slamming into a neutron star, or a "phase transition" in the core of a neutron star (i.e., the neutron star's core somehow abruptly changed its state as it cooled, like water does when it freezes, releasing energy in the process), or even more speculative suggestions involving hypothetical new objects, such as "a quark nugget falling onto a strange quark star." Most of these ideas accounted for only a limited subset of the known facts. Almost none of them attracted many believers, or were the subject of more than one research paper.
It was particularly difficult for theorists to account for the enormous gamma-ray brightness of the hard initial pulse of the March 5th event. If you try to power this from some material falling onto a neutron star (e.g., from a planet or asteroid) then the pressure associated with the outflowing gamma rays itself halts the inflow, and cuts off the energy supply. But if you try to power it from a source deep inside the neutron star, like a phase transition, then it is hard to get all the energy out quickly and completely enough in the form of gamma rays.
TopBack to top

The other SGRs

In the 1990's the SGR mystery stimulated many astronomers to point all kinds of telescopes at these objects. The X-ray point source at the location of the March 5th burster was discovered in 1991. Then in 1993 and 1994 the locations of two more SGRs were pinned down.

This was the first SGR ever seen to burst, in January 1979. It was very active in the mid-1980's, emitting over 100 detected bursts. Since 1990 it has been sporadically active. The numbers "1806-20" refer to its coordinates in the sky: 18 hours, 06 minutes right ascention, -20 degrees declination. Its location on the sky is near the position of the Galactic center, in the constellation Sagittarius. The center of the Galaxy is 25,000 light years away, but some researchers have argued that SGR 1806-20 lies well beyond that. Its true distance is uncertain.

In 1993, a team of (mostly) Japanese astronomers led by Toshio Murakhami detected SGR 1806-20 using the orbiting ASCA X-ray telescope. They showed that SGR 1806-20 is a bright X-ray star, even in quiescence (i.e., between bursting episodes). In 1998, Chryssa Kouveliotou et al. showed that the brightness of this X-ray star pulsates at a gradually-slowing rate, giving strong corroboration to the magnetar model. This was the first of several important developments during the "Magnetar Year" of 1998, which will be discussed below. (I recommend that you read the sections in order, but for the hypertext-obsessed, here's a jump forward to 1998.)

The region of the sky surrounding SGR 1806-20 is very crowded with galactic star-forming and star-dying activity, so it is not yet clear whether there exists a young supernova remnant associated with SGR 1806-20.

A Companion Star to SGR 1806-20 ?

(I will now explain how astronomers became confused about some properties of SGR 1806-20 during the mid-1990's. This is not scientifically important, but it will be of interest to advanced readers.)

Shortly after SGR 1806-20 was found to be an X-ray star by Murakhami et al., radio telescopes detected a splotch of radio-bright gas which seemed to surround it. This led astronomers to suppose that the SGR was blowing out a wind of charged particles, like a radio pulsar. (Note: radio pulsar winds are discussed in the next section). Optical/infrared telescopes then discovered a very massive, bright, hot, young, nuclear-powered "gas-ball" star positioned in the precise center of the radio emissions. Thus throughout the mid-1990's, most astronomers thought that SGR 1806-20 was in a binary system with this more ordinary star.

In 1999, Kevin Hurley of Berkeley and his collaborators used an Interplanetary Network (IPN) of gamma-ray detectors to pin down the location of SGR 1806-20 with very high precision. They found that the SGR was displaced 0.004 degrees (14 arcseconds) away from the center of the radio source, and from the massive, nuclear-powered star. This displacement was confirmed in 2001 by the orbiting Chandra X-ray telescope. We now know that the SGR 1806-20, like other SGRs, is a bright X-ray star, but it is not blowing out a steady, radio-bright wind of charged particles (at least, not at levels detectable so far), nor is it in a binary system with any known star.
This star, in the constellation Aquila the Eagle, emitted 3 bursts detected during two days in mid-March 1979. Then it remained quiet for nearly two decades, with the exception of 3 more bursts observed in 1992.

The location of SGR 1900+14 was found in 1994, and pinned down very precisely in 1998. It is a point source of continuous X-rays like other SGRs.

In the Summer of 1998, after decades of near-quiescence, all hell started breaking loose on SGR 1900+14. It soon emitted a giant flare, very similar to the March 5, 1979 event, as discussed below...

This source was only discovered in 1997, when it emitted 2 bursts observed by spaceborne observatories. Its position has not yet been determined accurately enough to see if it is associated with a point source of X-rays, or a SNR.

This fifth known SGR was discovered by NASA astronomers on June 15, 1998. It emitted about 100 bursts in June-July 1998 that were detected by 4 different X-ray and gamma-ray observatories in space. The new burster's position is consistent with a supernova remnant near the galactic plane. The rotation period of this SGR is not known with certainty, although there is some (ambiguous) evidence that it lies near 6.4 seconds.

While scores of astronomers were working to achieve these observational advances in the 1990's, the "magnetar" theory of SGRs was being developed. This theory is still being tested and debated. As more observations come in, it might be disproven. However, a wide variety of evidence now seems to favor it.
TopBack to top

Neutron Star Magnetic Fields

The "magnetar" theory of SGRs came about from an attempt to understand a completely different issue, namely: the origin of magnetic fields in radio pulsars.
Radio pulsars are "garden variety" neutron stars: over 1000 have been detected since they were discovered in 1968. They emit beams of radio waves which sweep through space as the stars rotate, like lighthouse beams, thus from afar pulsars seem to flicker or pulsate at their rotation periods. Careful measurements have shown that pulsar periods increase over time, implying that the stars are gradually spinning down. This is attributed to their magnetic fields. The field lines are anchored to the neutron star surface, because they are generated by circulating electric currents inside the star. Thus as the star turns the field also must turn. This drives magnetic waves outward, along with diffuse winds of charged particles (which emit the radio beams from just above the magnetic poles), carrying off energy and causing the star to gradually spin down. The measured rate of spin-down allows the magnetic field to be estimated. For almost all young pulsars it is a few times 1012 Gauss at the magnetic poles.
In 1986, Christopher Thompson (originally from Winnipeg, Canada; now at the Canadian Institute for Theoretical Astrophysics in Toronto) and I began to study a question that many astrophysicists had wondered about, namely: why are pulsar magnetic fields about 1012 Gauss?   At the time, we were both at Princeton University: Chris was a graduate student and I was a postdoctoral fellow. We were intrigued by computer simulations which had shown that neutron stars get all mixed up just after they form. We wondered how this would affect their magnetic fields.
Neutron stars are very hot when they first form. The computer models, by Adam Burrows of the University of Arizona and James Lattimer of the State University of New York at Stony Brook, showed that the dense fluid of neutrons inside a nascent neutron star roils and churns to help carry out heat, a little like water boiling in a pot. Such circulation is called "convection." It happens because hot parcels of fluid rise, while cool ones sink. (In a hot nuclear fluid, the density of electrons also affects fluid buoyancy and helps drive mixing.)
We found that the hot, ultra-dense neutron star fluid also can conduct electricity very well, because it contains some free electrons and protons along with the more abundant neutrons. These charged particles readily carry currents. This means that any magnetic field lines caught in the fluid initially are swept along by the convective motions; they cannot just "ignore" the moving fluid because the fluid is an electrical conductor.
If the star is born rotating fast enough, the combined effects of rotation and convection, which both drag field lines through the star, can build up the star's overall magnetic field, via a complicated process known as "dynamo action." Dynamos operate in the interior of the Earth and the Sun, giving them their magnetic fields.
We were amazed to find that, if a dynamo worked with ideal efficiency in a hot, newborn neutron star, it would generate a field of about 1016 Gauss: 10,000 times stronger than was actually found in pulsars! As the star cools, convection and dynamo action cease. This happens after only about 10 or 20 seconds in a neutron star, but 10 seconds is enough time for a very strong field to build up. After that, the field can remain trapped by the heavy, stratified liquid of neutrons and protons inside the neutron star.
This led us to conclude that the familiar radio pulsars were neutron stars in which large-scale dynamos had essentially failed to operate, probably because they were not born rotating fast enough. The spin period of the Crab pulsar at birth was about 20 milliseconds; we found that it needed to be considerably less than that for a dynamo to work.
The question of why a pulsar field was 1012 Gauss thus turned out to involve some subtle details of the residues of magnetism left over after a large-scale dynamo fails. We made some progress in estimating this, but we also couldn't help wondering: what happens if the dynamo succeeds?
TopBack to top

A new kind of Star?

For readers interested in the theory of magnetar formation:
I tried to give a concise summary (with just a few important equations)
at a March 1995 astrophysics conference in La Jolla, California.
The paper is posted
here. Incidentally, this was the first paper to use
the term "Anomalous X-ray Pulsar" or "AXP," and the first to suggest
that AXPs -- then a novel class of stars -- were magnetars.
We estimated that, at the pole of a dynamo-active young neutron star, the magnetic field could realistically reach 1014 - 1015 Gauss--- 100 to 1000 times stronger than in ordinary pulsars. What would such a strongly-magnetized neutron star, or `magnetar' look like?

Although it is born spinning somewhat faster than a pulsar, a magnetar spins down much more quickly, because the magnetic waves (and the related, diffuse winds of charged particles) which carry off the star's rotational energy are very efficient when the field is strong. This means that magnetars rarely send out widely-sweeping radio "lighthouse" beams as do radio pulsars. Except in a fleeting interval just after it is formed, a magnetar spins so slowly that its spindown-powered beams are exceedingly narrow or completely turned-off. (Recall that the radio beams in an ordinary pulsar come from a rotation-driven outflow of charged particles above the magnetic poles. When the rotation rate drops this ceases.)

To put it a different way: all of the observed emissions of ordinary radio pulsars (except for a faint X-ray glow from its tiny, hot surface) are powered by a slow loss of the rotational energy that the star is born with. A radio pulsar's magnetic field is essentially stable; its main role is to passively facilitate the loss of rotational energy. In a magnetar, on the other hand, the rotational energy quickly becomes negligible. However, we realized that the magnetic field itself can provide an energy source for potentially observable emissions. A magnetar's field is strong enough to push material around in the star's interior and crust, leading to the dissipation of a significant amount of magnetic energy during the first ten thousand years.

This has several consequences:

The star is kept hot by magnetic energy dissipation in its interior. Like any hot object, the star will glow. In the case a young magnetar (with age less than a few times 10,000 years), its surface is hot enough to glow brightly in X-rays. Moreover, the shifting magnetic field outside the star gives rise to streaming charged particles, that inevitably impart energy to X-ray photons by scattering against them. Some streaming charged particles also slam against the star when they reach the footpoints of magnetic field lines, heating patches on the surface, which glow brightly. In these ways, the evolving magnetic field of a magnetar can dissipate energy to X-rays.
As the tremendous magnetic field drifts through the solid crust of the magnetar, it stresses the crust with magnetic forces which get stronger than the solid can bear. This causes sudden shifts in the crust structure, leading to bright outbursts.

Why does the star have a crust?

A neutron star is mostly made of a dense liquid of neutrons, with a trace of protons and electrons. This is "nuclear fluid": the pure stuff of the atomic nucleus. It is denser than liquid water on Earth by a factor exceeding 1014. This great disparity is understandable since ordinary atoms, like those of water, are mostly made of empty space with a smattering of lightweight electrons that flitter around a tiny, heavy nucleus. The nucleus is what holds almost all the mass. In contrast, neutron star matter dispenses with the nearly-empty space: it is "wall-to-wall nucleus." A single tablespoon of nuclear fluid from deep inside a neutron star contains about 10 billion tons of material: as much matter as in a large mountain on Earth.

The ultra-dense nuclear fluid would explode like a nuclear bomb if you brought it to Earth, but inside a neutron star it is stable because it is held under tremendous pressure. However, in the outer layers of a neutron star (as in the outer layers of all stars) the pressure and the temperature both drop, although the force of gravity is still enormous. Here the fluid solidifies into a heavy crust, about a mile in depth. It is made of heavy atomic nuclei arranged in a quasi-cubic lattice, with electrons flowing between, somewhat like a terrestrial metallic alloy but much denser.

The properties of a neutron star's crust were worked out by Malvin Ruderman of Columbia University and other astrophysicists. The upper layers are made of iron, but the nuclei in the solid lattice get increasingly heavy and bloated with neutrons as you go deeper. Free neutrons are also present, along with free electrons, in the inner crust, at depths below about 1/2 mile. That is, a liquid of single neutrons and (negatively-charged) electrons permeates the (positively-charged) lattice. These free neutrons are continually getting stuck and unstuck to the neutron-bloated nuclei. At the base of the crust the bloated nuclei touch and merge; and the whole sub-crust star resembles one giant nucleus.

In a pulsar this outer, solid shell is essentially stable, but in a magnetar, it is stressed by unbearable magnetic forces as the field diffuses through it, and as the magnetic field in the liquid core drags on it from below. This deforms the crust and causes the magnetic field penetrating the crust to shift and move. Occasionally the crust and the field above it become catastrophically unstable.


Magnetar outbursts release a tremendous amount of magnetic energy very rapidly. They tend to come in bunches, at times when the crust is yielding to strong magnetic stresses. As the instability grows, the changing, shearing, twisting field drives strong dissipative currents above the star, energizing particles trapped in the exterior magnetic field. Simultaneously, the magnetic field rearranges itself to a state of lower energy. This produces a burst of hard X-rays (soft gamma rays) observed as ordinary, powerful SGR bursts.

Note that magnetic forces can also deform a radio pulsar's crust, but a typical pulsar's magnetic field is not strong enough to rapidly deform the crust and drive episodes of bright outbursts. The field must be roughly 1014 Gauss or more to cause the deep-crust solid to fail.

Giant Flares

Occasionally, the magnetic field becomes unstable on much larger scales, and rapidly rearranges itself to a state of lower energy. Giant flares inevitably involve significant shifts in the crust structure as well.

In a magnetar, the energy available for these magnetic flares is tremendous because the field is so strong. A magnetar's magnetic energy is easily sufficient to power the 1979 March 5th or 1998 August 27th giant flares; as well as all the common, short-duration bursts from SGRs.

Somewhat analogous outbursts often happen on the Sun, in explosive events called "solar flares." During a solar flare, magnetic field lines near the Sun's surface change the pattern by which they connect to each other, a process called "magnetic reconnection" which releases pure magnetic energy. This happens in magnetars too. Because it occurs quickly and cleanly (without much matter involved) it is capable of producing intense, short bursts of gamma rays. Indeed, solar flares often emit part of their energy in gamma rays.
TopBack to top

A brief history of magnetars

In 1992 we published a paper in The Astrophysical Journal proposing the magnetar theory of SGRs. We outlined most of the ideas described above, including the idea of reconnection-powered bursts (and magnetically-powered steady X-rays, buried in footnote 6 of the paper). We also noted that:
Because a magnetar's field is very strong, it spins down via magnetic waves and particle outflows very quickly, potentially accounting for the unusually long spin period of the March 5th burster. In fact, one can estimate how strong the field must be for the star to spin down to its observed period (8.0 s) in its observed age (less than 10,000 years, as inferred from observations of the supernova remnant N49 in which it presumably formed). We found roughly 6 X 1014 Gauss ---the first-ever estimate of a magnetar's field. This answer is not sensitive to the initial spin period, as long as the star was born rotating much faster than it is rotating now. (Two effects mentioned above -- magnetic vibrations and exterior field-twists -- can accelerate a magnetar's rate of spin-down during periods of magnetic activity and moderately reduce the required magnetic field, but calculations show that SGR fields in excess of 1014 Gauss are still required.)
Magnetars might naturally acquire a large recoil velocities at birth, via the "neutrino rocket effect" (or "neutrino magnetic starspots"). This effect is still being studied and debated, so I won't go into details here, except to note that this might explain the large observed displacements of some SGRs from the centers of their associated supernova remnants ("SNRs").
The theory was widely met with skepticism. This is a healthy part of the scientific process, causing us to work harder. In 1995 we published a long paper in the Monthly Notices of the Royal Astronomical Society (a British journal) with many more details. We outlined seven different ways to estimate the magnetic field of the March 5th burster, all of which seemed to indicate a field greater than 1014 Gauss.
In particular, we argued that if, and only if, the field exceeds 1014 Gauss, it could:
(1) Spin down the star to an 8.0 s period in the age of the SNR (as mentioned above).
(2) Provide enough energy for the March 5th event, a putative magnetic flare.
(3) Account for the short, 0.2-second duration of the March 5th event's hard spike. This is the time needed to make a large-scale magnetic readjustment, since magnetic disturbances must travel through the star.
(4) Drive magnetic dissipation quickly enough to explain SGR activity in a time of order 10,000 years, the ages of SGRs (as inferred from their associated SNRs).
(5) Provide enough energy to power the steady X-ray glow of SGRs. ("the X-ray point sources").
(6) Render the hot gas of particles which emitted the March 5th event's soft tail (and normal SGR bursts) nearly transparent to X-rays. This is necessary to explain why SGR bursts are so extraordinarily bright.

This important point was first noted by Bohdan Paczynski of Princeton University in 1992, shortly after our first magnetar paper.

(7) Hold down, with magnetic forces, the hot gas of particles which emitted the March 5th event's soft tail.
TopBack to top

Trapped Fireballs

This final point needs a bit of explanation. The initial pulse of the March 5th event was so hard and bright that it must have been emitted by a pure (nearly mass-free) explosion of energy, or a "fireball," blown out from the star at nearly the speed of light. This could have been powered by a magnetic flare.

The fireball evidently contained little matter, except for lightweight electron-positron pairs which are ubiquitous in tremendously hot gas. (Positrons are anti-electrons, a kind of antimatter. Particle-antiparticle pairs are spontaneously created from photons whenever sufficient energy is present.) If the fireball had been polluted by more than a trace of heavy particles (neutrons and protons) then it would have lost energy in blowing out the heavy matter, and it wouldn't have emitted such intense, hard gamma rays.

It is natural to expect that after the fireball dispersed, it left behind a residue. In this case, the expected residue is a hot cloud of electron-positron pairs, trapped near the star by the strong magnetic field.

Near a neutron star, magnetic field lines are anchored at both ends on the stellar surface, describing arches outside the star. (In this way, they resemble the field lines of an iron bar magnet, which arch from the north to the south poles.) Electrons and positrons are electrically charged, so they gyrate around field lines, but they cannot drift perpendicular to them. They can drift freely parallel to field lines, but in this direction they quickly run up against the stellar surface at the footpoints of a magnetic arch. Thus the magnetic field acts like a "bottle," holding charged particles. Now, there are also plentiful X-rays and gamma-rays inside the magnetic bottle, which can cross the field lines, but these photons bounce around between electrons and positrons and don't get far. (Besides bouncing, gamma-ray photons also continually make electron-positron pairs and get regenerated when pairs annihillate. This further impedes their motion.) Only at the surface of the bottle can the photons stream freely away.

Thompson and I realized that zones of such hot, magnetized photon-pair gas must have been left behind following the March 5th hard pulse, anchored onto the neutron star. We called this phenomenon the "trapped fireball." The hot gas loses energy as photons stream away from its outermost, exposed layers. Electrons and positrons in this outer shell steadily annihillate, so their energy is radiated away too. The trapped fireball inevitably shrinks over time. Nested sheaths of field-lines empty out, in succession.

This could explain the March 5th light curve. As the trapped fireball shrunk, its glowing surface area diminished and it got dimmer. Since the magnetic field lines were anchored to the rotating neutron star, the zones of glowing gas also turned every 8 seconds. We must have viewed the fireball from ever-changing angles, repeating on an 8-second cycle. Peaks in the brightness occured when the most luminous part faced toward us; and dips when the fireball was mostly occulted by the star. Meanwhile, it steadily shrunk and dimmed. After about 3 minutes, the trapped fireball evidently evaporated away entirely.

Similar zones of hot, trapped particles are probably made when magnetic energy is released in common SGR bursts. This can explain why the X-ray spectrum (the distribution of energies of X-ray photons) was essentially the same in the soft tail of the March 5th event as it was in the subsequent short bursts. But typical SGR bursts don't last long enough to show 8-second rotational dips. They simply don't have enough energy to persist that long. A low-energy trapped fireball evaporates quickly.
This brings us to point (7) above. The star's magnetic field had to be strong enough to confine the hot electron-positron-photon gas which emitted the soft tail of the March 5th event. X-ray measurements of the soft tail, summed over its whole 3-minute duration, tell us (roughly) what the total energy of the trapped gas was. A very simple calculation then shows that, in order to hold a particle gas with that much energy close to a neutron star (close enough to show such dramatic dips as it turns), using purely magnetic forces, the field must have been greater than about 4 X 1014 Gauss.
Now, because a huge amount of energy escaped at the beginning (as evinced by the hard initial spike) the magnetic trapping forces evidently were pushed to their limits. This suggests that 4 X 1014 Gauss is an estimate of the field strength, not just a lower limit on it.
The fact that this estimate agrees with other independent arguments (e.g., the spindown argument, etc.) is encouraging. Most of our estimates were based on the 1979 March 5th event, so that event might be called a "smoking gun" for extremely strong magnetic fields.
TopBack to top

1998: A Year of Magnetar Breakthroughs

  All of the above arguments for magnetars were proposed before the end of 1995. However, at that time few astronomers were interested in studying abstruse issues of neutron star magnetism, or considering reinterpretations of data which were 16 years old. This situation changed dramatically in 1998 when many new observational results came flooding in.

It began in May 1998, when Chryssa Kouveliotou of NASA Marshall Space Flight Center and an international team of 10 collaborators showed that the X-ray emissions from SGR 1806-20 pulsate on a regular 7.5 second cycle. These pulsations are almost certainly due to the rotation of a neutron star. As the star turns, bright and dim zones on its surface and in its surrounding magnetosphere ("hot and cold spots") evidently rotate in and out of our view. This means that SGR 1806-20 has a rotation period of 7.5 s, similar to that of the March 5th burster (8.0 s). Moreover, Kouveliotou and her collaborators measured the rate at which the pulsations (or rotations) were slowing down : namely 0.26 seconds per century. This might not sound like much, but it demonstrated that the braking on the star's spin is profound. It allowed Kouveliotou et al. to make a fairly direct estimate of the magnetic field, using the same method used in radio pulsar studies. If the braking was due to magnetic waves and associated particle flows carrying away energy and angular momentum, as seemed plausible, then the field strength was 8 X 1014 Gauss. This was, of course, precisely in the range that Thompson and I had favored for a number of different reasons (as outlined above ).

This dramatic result, published in the the 21 May 1998 issue of Nature raised much interest among astrophysicists. Moreover, it seemed as if the SGRs themselves decided to respond. During the last week of May 1998 SGR 1900+14 emitted over 50 detected bursts, some with unprecedented energy. It continued bursting into early June, when a completely new SGR in our galaxy, SGR 1627-41, showed itself for the first time. This new star emitted about 100 bursts over the next two months, as described above.

Then, in August, Kevin Hurley and his co-workers announced that they had detected 5.16 second pulsations in the continuous X-rays from SGR 1900+14 ( IAU Circular 7001 ).   Kouveliotou, Tod Strohmayer (NASA Goddard Space Flight Center) and Hurley, working with six other scientists, soon found that the X-ray pulsations of this star, like SGR 1806-20, were gradually slowing down. For the magnetic field to cause the star to spin down at the observed rate, it would need to be about 5 X 1014 Gauss.

But before the researchers had a chance to write a paper about this, they were scrambling to point all available X-ray telescopes toward SGR 1900+14 again...

TopBack to top

The August 27th Event

  On August 27, 1998 a giant flare from SGR 1900+14 set new records for the most intense flux of gamma-rays ever detected from a source outside our solar system. It blitzed gamma-ray and X-ray detectors on seven different spacecraft at locations throughout the solar system. Especially useful data were recorded by three experiments: the Russian Konus detector on the geo-space science Wind space probe which was orbiting near the Sun-Earth equilibrium point ("L1"), upstream of the Earth in the solar wind; the Italian-Dutch Beppo-SAX gamma-ray/X-ray observatory, in low Earth orbit; and a gamma-ray detector aboard the Ulysses spacecraft, a joint effort of the European Space Agency and NASA that was orbiting the Sun in a polar orbit at roughly the distance of Jupiter.

NASA's Rossi X-ray Timing Explorer (RXTE), another Earth-orbiting X-ray observatory, was pointed away from SGR 1900+14 when the burst occured, but it nevertheless recorded a strong signal. High-energy photons were diffusing through the metal shields surrounding its X-ray detectors. However, one proven workhorse for SGR studies, the Burst and Transient Source Experiment (BATSE) aboard NASA's orbiting Compton Gamma-ray Observatory, detected nothing. The BATSE team, led by mild-mannered Charles Meegan (who is BATSE-MAN) ran out of luck that day: the Compton Observatory was on the far side of the Earth at the time of the flare.

The flare hit the Earth on it's night side, in the zenith over the western Pacific Ocean, at 1:22 A.M. Hawaii time. It was intense enough to strongly ionize the Earth's outer atmosphere, affecting radio communications.

This requires some explanation. Radio waves, especially long-wavelength ones like those on the AM dial, bounce between the "ionosphere" and the Earth's surface as they propagate around our planet. The ionosphere is a layer of diffuse, ionized gas -- atoms of air which have lost electrons and become positively-charged ions -- in the upper reaches of the atmosphere. High-energy photons from the Sun keep the thin air up there well-ionized during the day, so the daytime ionosphere lies about 60 kilometers above the Earth's surface. At night, electrons recombine with ions, causing the inner edge of the ionosphere to recede upward, to 80 - 90 kilometers. This is why you can pick up very distant AM stations on your radio at night: radio signals generally travel farther if they must make fewer (power-sapping) bounces.

In the early morning of August 27th 1998, Stanford University engineers monitoring very-long-wavelength U.S. Navy radio transmissions (which carry coded messages for nuclear submarines) found that the altitude of the ionosphere plummeted for a five-minute period beginning at 3:22 A.M. PDT. Some mysterious source of ionization was apparently driving the ionized layer down to daytime altitudes (about 60 km). Curiously, the height of the ionosphere was also observed to vary cyclically over a period of 5.16 seconds... Of course, they had detected the rotation period of SGR 1900+14 in a remarkable new way, proving that you don't need a sensitive X-ray telescope to measure the spin period of a SGR. If you can wait for a giant flare, you can use a "Whole Earth Telescope"-- the bulk ionosphere of the whole planet -- to see the rotation period of a tiny neutron star, twenty thousand light years away.

As the wavefront of gamma-rays swept out of the solar system, the last spacecraft it reached was NASA's Near-Earth Asteroid Rendevous (NEAR) space probe, enroute to a rendevous with the asteroid Eros. The flare was bright enough to force NEAR's gamma-ray detectors into a protective shut-down mode.

Here is a graph of the intensity of the August 27 flare, as recorded by the Ulysses gamma-ray detectors (sensitive to gamma rays in the range of 25-150 "kilo-electron volts" or "keV," a unit of photon energy).

The two giant flares ( 1979 March 5. and 1998 August 27) were similar in many ways. Each began with a brief, hard spike of very intense gamma-rays, followed by a soft oscillating tail. In the August 27th event, the oscillations follow a regular 5.16-second cycle. (The cycles appear progressively shorter as you go to the right on the above figure, but that is only because the time-scale is not linear -- it gets more "squashed" as you move to the right, in a "logarithmic" way.)

The final stages of the August 27, 1998 flare, as recorded by
gamma-ray detectors on the Ulysses interplanetary spacecraft.
The vertical axis shows detector counts during successive 5-sec.
time intervals, which approximately averages over the rotation
period of the star. The dashed lines show expected emissions
from a trapped fireball on a magnetar, evaporating after about
380 seconds. (This figure, and the two figures which follow are
from M. Feroci, K. Hurley, R.C. Duncan & C. Thompson 2001,
Astrophysical Journal , 549, p.1021.

Although the August 27 event was intrinsically less powerful than March 5th (roughly by a factor ~10), it came from a source much closer to Earth, so it appeared brighter. Many improvements in detectors and data recording equipment had occured between 1979 and 1998, so we have much better data about the 1998 flare. For example, in 1979, almost no information about the terminal stages of the flare was recorded, due to the limited data storage capabilities of 1970's-era spacecraft computers. In 1998, the end of the flare was recorded by three different experiments ( Konus, Ulysses and Beppo-SAX). Remarkably, the bright flare emissions dropped quite abruptly, essentially to zero, at a time 380 seconds after the onset of the flare (see figure at right). This is as expected in the magnetar model, because a trapped fireball on the surface of a magnetar must "evaporate away" completely in a finite time, as electron-positron pairs in the fireball annihillate and their energy is carried away by the flare's hard X-rays. In contrast, emissions from a cooling "hot spot" on the surface of a star would gradually fade away as the hard X-rays "soften" -- i.e., the X-rays photon energies would gradually shift to lower values.

Strange Quark Stars and Afterglows
(some parenthetical remarks, for advanced readers)

Fading "hot spot" behavior is predicted by an alternative model for SGRs based upon the idea of "strange quark stars."   This model posits that SGRs are made of "strange quark matter," a hypothetical alternative state for very dense material (different from ordinary nuclear matter, as found in neutron stars). Theorists have noted that a glob or "nugget" of strange quark matter falling onto a strange quark star (if such an event somehow happened) and slamming onto its surface at high speed, would create a hot spot on the surface which could emit something resembling a SGR giant flare. This is because strong quantum "color" forces between quarks would hold down matter rather than gravity, allowing extremely bright gamma-ray and hard X-ray emissions to come from the star, without losing too much energy to the process of blowing out matter. However, the abrupt termination of the August 27th flare seems inconsistent with this scenario.

Note that even in the magnetar model, there should be a residual "hot spot" on the neutron star surface after the trapped fireball evaporates, which emits an X-ray "afterglow." (This was first noted by Thompson and I in our 1995 paper in Monthly Notices of the Royal Astronoical Society.) But this afterglow is orders of magnitude fainter that the bright, trapped fireball emissions in the flare, so it cannot be detected by all-sky gamma-ray detectors like the one on Ulysses . To detect SGR afterglow requires a true X-ray telescope, pointed at the source. Indeed a fading afterglow was found in the days following the 1998 August 27th event when RXTE and other X-ray telescopes were turned toward the SGR. Afterglows following three subsequent, bright bursts from SGR 1900+14 have been measured as well (on 1998 August 29; 2001 April 18 and 2001 April 28). These afterglows are consistent with cooling hot-spots on magnetars, but not on strange quark stars.

Perhaps the most striking feature of the August 27th event was the emergence of a strong four-peaked pattern in the light curve after about 40 seconds, as shown in these data from the Ulysses and Beppo-SAX gamma-ray detectors.

The four-peaked pattern persisted through much of the rest of the flare, as shown in the following graph. (Note: each box below represents one 5.16-second rotation cycle of the star. The Beppo-SAX spacecraft recorded the detailed pattern of flare emission for 7 1/2 more cycles than did Ulysses .)

These remarkable data indicate that the geometry of the trapped fireball became more complicated in regions close to the star, once the far-reaching, smoother emission zones cleared. This will be interesting to fit with detailed models, in order to learn more about the flare mechanism and the neutron star’s magnetic field.

TopBack to top

Magnetar Links

Here are links to some non-technical webpages and articles about SGRs and AXPs.
Scientific American article: "Magnetars"

This was the Scientific American cover story in February 2003. To see the pdf file for this 8-page article, click on the cover at right.

(Note: to read this file you need Adobe Acrobat Reader. If you don't have this on your machine, you can download it for free from Adobe's website.)

Magnetar Discovery Solves 19-year-old Mystery

This NASA Marshall Space Flight Center (MSFC) webpage from May 1998 describes the detection of gradually slowing X-ray pulsations from SGR 1806-20, a critical test of the theory, and a truly transformative result. Chryssa Kouveliotou led the NASA team which can rightly be credited with discovering magnetars.
This was also the first-ever webpage about magnetars. Links which follow are listed in rough chronological order.

The scientific reference is: C. Kouveliotou, S. Dieters, T. Strohmayer, J. van Paradijs, G.J. Fishman, C.A. Meegan, K. Hurley, J. Kommers, I. Smith, D. Frail and T. Murakhami, Nature , 393, p. 235 (1998).

A whole lot of shakin' going on

This NASA MSFC webpage describes how a new soft gamma repeater, SGR 1627-41, was discovered in June 1998.
Cosmic Flasher Reveals All!

This National Radio Astronomy Observatory webpage describes the detection of an afterglow following the August 27th flare, using the Very Large Array radio telescope in New Mexico. This gave direct evidence for an outflow of particles at nearly the speed of light in the flare, probably blown out during the initial hard spike.

Reference: D. Frail, S. Kulkarni and J. Bloom , Nature , 398, p. 127 (1999).
Astronomers try to catch runaway star

This University of California, Berkeley press release describes Kevin Hurley's ongoing effort to measure the velocity of SGR 1900+14, and solve one of the major outstanding questions about SGRs.
"Crushed by Magnetism" (pdf file) by Robert Irion

This discussion of current research on magnetars appeared in Science magazine in April 2004, as part of a special issue on neutron stars.

A companion piece about rotation-powered pulsars appeared in the same issue: "The Pulsar Menagerie" (pdf file) by Robert Irion.
TopBack to top

Epilog: New horizons in magnetar science

    Many astronomers are now studying SGRs and AXPs. These stars are most luminous in X-rays, but some have been detected in visible and infrared wavebands as well, where they show pulsating (and possibly semi-polarized) emissions. Astronomers have also detected changes in X-ray and infrared emissions, including changes in pulse shapes and in the X-ray photon energy distributions; and variations in the rates of spindown. These observed changes sometimes occur in concert and/or together with burst and flare activity.

Studies of these phenomena promise to tell us much about the physics of neutron stars and their surroundings. Magnetars seem especially well suited for such investigations. Radio pulsars are much more stable stars, so their relatively-unchanging emissions provide less interesting probes of the stars' structure.

This ongoing study of SGRs and AXPs is one growing area in magnetar astrophysics. Another possible growth area involves unexplained astronomical phenomena for which magnetars might prove culpable. Here are some intriguing possibilities.

This grey box is more speculative than the rest of this website, and can be skipped by readers who are not interested in unresolved questions on the frontiers of astrophysics.

Ultra-high Energy Cosmic Rays

Cosmic rays, discovered and named in 1912, are actually not rays at all. They are very fast-moving protons or atomic nuclei (i.e., self-bound knots of protons and neutrons). A diffuse flux of such particles rains down on the Earth all the time, from the depths of space.

Because cosmic rays travel at nearly the speed of light, they have very high energies. Most cosmic rays are believed to be accelerated in the shockwaves of supernova explosions, but the most energetic cosmic rays cannot be accounted for in this way. These rare "Ultra High-Energy Cosmic Rays" (UHECRs) have truly amazing energies. They probably are single protons or light nuclei (e.g., helium nuclei) each carrying as much kinetic energy as a bowling ball rolling toward the pins, or a speeding ball in baseball or cricket (choose your sport).

UHECRs are detected when they strike the upper layers of our atmosphere and dissipate their energy, creating "showers" of high-energy particles which can be seen as flashes of light. The study of these mysterious particles is a hot topic in astronomy right now.

UHECRs cannot travel truly vast distances (i.e., billions of light-years) across the Universe because they tend to lose energy by scattering against the ubiquitous low-energy photons that fill space: the "cosmic microwave background" left over from the Big Bang. This means that very distant quasars or active galactic nuclei (giant black holes surrounded by accretion disks and with outflowing jets) cannot be the sources of UHECRs. Neither can those cosmic gamma-ray burst (GRB) sources which are known to be billions of light years away. One must look instead for sources within about 100 million light- years of Earth, where there are no quasars.

Newborn magnetars may the sources, as suggested in a recent paper by Jonathan Arons of Berkeley. With their very rapid spins and high magnetic fields, young magnetars generate very strong electric fields that could accelerate particles to ultra-high energies. UHECR energies would then come ultimately from the rapid spin of a nascent neutron star. There are many galaxies within 100 million light-years of Earth in which magnetars presumably form, easily enough to account for the observed UHECRs.

For UHECRs to emerge from a fast-spinning magnetar within a supernova, the high-energy outflow from the magnetar must quickly "punch" out of the (relatively slow-moving) supernova gas, leaving open channels or jets. Otherwise the UHECRs would lose too much energy by interacting with the hot gas of the exploding star. Arons argues that the powerful wind of particles and radiation blown out from a young magnetar at nearly the speed of light is fully capable of "punching out."

How might this scenario be verified? Galaxies within 100 million lightyears of Earth tend to be concentrated in a vast structure called the "supergalactic plane" which manifests itself as a band across the sky, nearly perpendicular to the plane of our Galaxy (i.e., to the Milky Way). For example, the Virgo cluster of galaxies, about 60 Million light-years away, lies in the supergalactic plane, as do most nearby galaxies and galaxy clusters. Recent observations (now being checked) suggest that UHECRs tend to come from the supergalactic plane. Magnetars are arguably the most plausible UHECR sources known in this local zone of the Universe.

Gamma-Ray Bursts

Gamma-ray bursts (GRBs) are occasional flashes of very high-energy photons which are observed come from spots all over the sky. They tend to be much harder than SGR bursts, and show no evidence for repeating. Two distinct physical classes of GRBs are known: long-duration and short-duration GRBs. Long-duration GRBs tend to have softer (lower-energy) photons. The dividing line between the two classes is about 2 seconds; i.e., most GRBs which last for less than 2 seconds are in the short duration physical class (although a few intermediate bursts are difficult to classify). About two-thirds of the GRBs cataloged by the landmark BATSE experiment of the 1990's were long-duration ones. It is possible (although far from certain) that magnetars play some role in explaining both kinds of GRBs.

Long-duration GRBs

Many long-duration GRBs have been proven to come from very great, "cosmological" distances (i.e., their sources are billions of light years away, which means that the gamma rays have traveled across an appreciable fraction of the 14-billion-year-old observable Universe before reaching us). In some cases, X-ray, optical, and radio-wave afterglows from long-duration GRBs have been detected, allowing their positions to be precisely pinned-down and matched with a very distant galaxy which presumably contains the GRB source.

At present, the most popular model for long-duration GRBs invokes the core-collapse of a very massive, rotating star, as in familiar models for supernova, but under circumstances in which the core becomes too heavy to form a stable neutron star. The result is a rapidly-spinning black hole with an disk of hot, dense matter orbiting around it. This probably drives jets of fast-moving matter outward along its rotation axis, which could "punch" out of the surrounding gas and perhaps be observable as GRBs when the jet happens to point toward Earth. Because a black hole's gravity is so strong, such a scenario potentially produces more power than any model involving neutron stars, which is advantageous when trying to explain the tremendous power of cosmic GRBs.

Alternative models, in which the "central engine" powering the GRB is a rapidly-spinning, nascent magnetar, have also been suggested by many astrophysicists (beginning with our 1992 magnetar-proposal paper, section 3.3). With less available power, this is not widely favored. However, a subset of long-duration GRBs with low power was recently identified by Jay Norris of NASA's Goddard Space Flight Center. These GRBs are thought to be intrinsically faint because they show long "lag times" between the arrival of hard and soft gamma-ray photons. (This lag time is empirically anti-correlated with brightness in GRBs.) About 100 of these events were found in the BATSE catalog of GRBs. Their sky positions tend to be concentrated near the Supergalactic Plane, as expected if they come from stars in galaxies within a few hundred million lightyears of Earth. (Here is Norris’ scientific paper.)

It is tempting to speculate that these low-brightness GRBs are the magnetar-powered subset of long-duration GRBs. If so, these could be the same objects which produce UHECRs, as in Aron's model. These might be due to magnetar-forming supernovae in which the outer (hydrogen) envelope of the star has been lost prior to the explosion, allowing easier emergence of high-energy emissions flowing from the central engine.

Short-duration GRBs

About 1/3 of the events in the BATSE catalog are short-duration GRBs. These brief events have distinct properties (including harder gamma-ray photons) so they seem to come from a different kind of source than the long-duration GRBs. There is circumstantial evidence that short GRB sources are also distributed widely through the observable Universe (i.e., detectable out to many billions of light-years) which would imply that they are intrinsically very bright. However, as of March 2003, no X-ray, optical or radio afterglow has yet been detected following a (verified) short GRB, so their locations have not been accurately pinned-down and studied. As a consequence, the distances of short GRB sources from Earth remain uncertain.

Some short GRBs are probably magnetar flares like the March 5, 1979 event, coming from SGRs in galaxies beyond the Milky Way and Magellanic Clouds. BATSE could have detected the hard spike of the March 5th event out to a distance of about 40 million light-years, almost reaching the Virgo cluster of galaxies. However, BATSE was not sensitive enough to detect the ensuing soft, oscillating tail of the March 5th event if its source was more than 2 million light-years away (roughly the distance to our nearest-neighbor Andromeda Galaxy). Thus, for flaring SGRs between 2 and 40 million lightyears distant, the hard spike would have been recorded as a typical short-duration GRB by BATSE. If all magnetar flares are as bright as the March 5th event, there should be about a dozen such events listed in the BATSE catalog of GRBs, with considerable uncertainty. (For details, see here. I estimate a 16% chance that BATSE detected more than 20 magnetar flares, and a 5% chance for more than 30.)

In order to determine how many --and which -- short-duration GRBs are magnetar flares, we need to obtain precise positions of these events on the sky. (BATSE localized short-duration GRBs only to within a few degrees.) NASA's Swift satellite, launched in November 2004, is designed to do just that. Upon detecting a GRB, Swift swiftly (within 20-70 seconds) turns and points both an X-ray telescope and an optical-UV telescope toward it. If the event shows an afterglow, then the X-ray and UV-optical telescopes can pinpoint its location.

The Swift mission is now in its early stages. Exciting results may soon be forthcoming. In particular, magnetar flares might prove obvious to Swift, since they will be found to come from regions of active star formation in relatively nearby galaxies. Moreover, Swift's X-ray telescope might catch the oscillating, soft tail of the event (lasting 3-5 minutes) allowing the flaring magnetar's rotation period to be measured.

It is interesting to speculate that some magnetar flares might be much more powerful than the March 5th event. If this is true, then it is possible that all short-duration GRBs are extragalactic magnetar flares from sources within several hundred million lightyears of Earth, although this seems unlikely.

Even if only a minor fraction of BATSE short bursts prove to be extragalactic magnetar flares, as seems more plausible, the fraction of magnetar flares will likely increase in the future, as GRB detectors get more sensitive.

The study of magnetars will be invigorated by these new data. Even for events with no data on soft tails (i.e., no Swift-like rapid follow-up with a sensitive X-ray telescope), a large catalog of hard spikes from magnetar flares would hold a wealth of fascinating information about magnetic instabilities in neutron stars.

Magnetars in the Milky Way

We now turn from observed, mysterious astronomical phenomena which might be caused by magnetars, to some predicted manifestation of magnetars which has yet to be observed.

Twelve magnetar candidates have been found by astronomers, as of March 2003.
This map of the sky shows the locations of 10 of them. Most lie in the plane
of the Milky Way Galaxy, which extends horizontally across the center of this
map. In 2002 an AXP was found in the Small Magellanic Cloud in the lower
right part of this map, so a yellow diamond should be added, below and to the
left of the red dot representing the 1979 March 5th burster in the Large
Magellanic Cloud. A new burst source, SGR 1801-23, has also been detected,
at a (somewhat uncertain) position near the Galactic Center.
How many magnetars are there in our Milky Way galaxy?   A dozen SGRs and AXPs are now known in the Milky Way and in the neighboring Magellanic Clouds, all with rotation periods between 5 and 12 seconds. Because these stars are spinning down rapidly, they cannot be very old; otherwise they would have already reached much longer periods. (Here we assume that the stars’ dipole fields do not vary rapidly, and in particular cannot dramatically grow at late stages: a very likely true assumption.) Observed magnetar spins imply that they are all roughly 10,000 years old or younger, in agreement with the ages of the supernova remnants found surrounding some of them (and within which the stars evidently formed).

Now, if there exist 10 magnetars younger than 10,000 years in our Galaxy, then magnetars must be forming at a rate of about 1 per 1000 years. In reality, the birthrate is probably higher because we haven't yet found all the young magnetars in the Galaxy. (SGRs and AXPs are still being found.)

Another way to estimate the birthrate is to take a census of neutron stars associated with known, young supernova remnants (SNRs). Since the number of magnetar candidates associated with SNRs is comparable to the number of radio pulsars so associated, an appreciable fraction (perhaps half?) of all neutron stars may be magnetars. Now, the total neutron star formation rate is almost as large as the rate of core-collapse supernova in the Galaxy. (Only a minor fraction of core collapse events are thought to make black holes). Based on counts of supernovae within many other galaxies like the Milky Way, this rate is estimated in the range of 1 per 100 years. This might be closer to the true Galactic magnetar birthrate than 1 per 1000 years.

Let's suppose (as a first guess) that the birthrate is 1 per 300 years. Then in the 10-billion- year age of our Galaxy, about 30 million magnetars would have formed. If the supernova rate was higher in the past, due to vigorous star formation during the early history of the Milky Way (as seems likely) then the count would be higher.

Where are all these stars? The theory of neutron star magnetic evolution gives clues.

Magnetar evolution

(The next two paragraphs are intended for advanced readers. Please skip them if they seem confusing.)
    In the sub-crust fluid interior of a neutron star, magnetic field lines in flux tubes strongly interact with the charged particles in the nuclear fluid (protons and electrons, which gyrate around the field lines), but not with the abundant, electrically-neutral neutrons, which hold about 90% of the ultra-dense fluid's mass. Magnetic forces gradually drag the flux tubes and their entrained charged particles through the background neutron fluid, causing the field to evolve via a process know as "ambipolar diffusion."   This is limited by friction, due to collisions between charged particles and neutrons, and by the limited rate at which neutrino-producing processes create and destroy the charged particles in the fluid and so adjust their numbers. (The simplest such processes are proton + electron --> neutron + neutrino and neutron --> proton + electron + antineutrino. The (anti)neutrinos immediately escape from the star.) Neutrino emissions are very sensitive to temperature, so ambipolar diffusion is much more rapid when the star is hot.

    We studied the implications of this for magnetars in 1996 (in this paper ). The diffusing field in the liquid interior, which includes strongly-wound interior toroidal (donut-shaped) field component, pulls on the crust from below, stressing it and sometimes driving significant changes in its structure. (This picture is complicated by magnetic diffusion within the crust, perhaps accompanied by many small-scale shifts in the crust structure.) Because the friction of ambipolar diffusion dissipates magnetic energy, which heats the star, which in turn speeds the diffusion, there is a feedback effect which means that the dissipation of magnetic energy shuts down rapidly once the star cools below a (rough) threshold temperature. The progressive unwinding and diminishment of the field also reduces driving stresses. This could happen after several tens of thousands of years.

This may explain why magnetic activity ceases when magnetars age past the SGR/AXP phase. A magnetar's field probably does not dramatically decay away during this time, but diminishes significantly. Subsequently, the field is trapped within the cool star, and changes more slowly.

Dead magnetars

Thus, both observations and theory suggest the existence of dead magnetars: magnetars which are no longer significantly powered by ongoing magnetic dissipation. Many millions of such stars, perhaps hundreds of millions, probably exist in the Galaxy around us. These stars are X-ray dark and burst/flare quiet, and therefore quite difficult to detect.

How might these stars be found? One possibility is that their strong magnetic fields could sweep up diffuse gas from interstellar space, making dead magnetars glow. This gas-sweeping process is most effective when the star is moving quickly, since more gas is swept up faster. Unfortunately, a fast-moving star soon escapes from the galactic disk, and there is little gas outside it. Taking this factor into account, there is little hope that magnetic gas-sweeping greatly improves the prospects for finding dead magnetars.

There is, however, another hope. After the star turns cold and ambipolar diffusion (mostly) shuts down, slow diffusive processes might continue to operate, gradually driving the star toward new magnetic instabilities. Thus a long-dead magnetar might briefly come alive in an episode of magnetic activity. Even if such episodes are very rare, among many millions of dead magnetars they might be spotted.

For now this is just speculation, waiting for hard data. Millions of dark, slowly-reeling stars seem to be out there, drifting through space, a challenge for astronomers to find.

TopBack to top
Robert Duncan, May, 1998
Updated March, 2003