6

GAMMA-RAY BURSTS AS PROBES OF THE UNIVERSE

The cosmos is full beyond measure of elegant truths; of exquisite interrelationships; of the awesome machinery of nature.

—Carl Sagan, “The Shores of the Cosmic Ocean"

episode of Cosmos: A Personal Voyage (1990)

 

Until now we have concentrated on the observations and theory of GRBs. The events have shown themselves as a complex, panchromatic phenomenon that can arise from a variety of different progenitors in both the nearby and distant Universe. The GRB community has succeeded in measuring the diversity of basic parameters of the explosions and clearly learned a great deal about the physics of the emission mechanisms. But despite a raft of certainties, we could fill a book with what we do not know. Indeed, filling in these sometimes gaping holes and searching for new insights into the phenomenon keep many of us awake a night.^*

In parallel with the ongoing study of these events and progenitors, new lines of inquiry have burgeoned: using GRBs as unique probes of the Universe in ways that are almost completely divorced from the nature of GRBs themselves. It may, at first, seem born of an ill-conceived crusade to admit that GRBs might tell us something interesting and deep about the Universe if we are willing to admit a large measure of consternation about our understanding of the probe itself. But the history of astronomy is littered with similar (apparently quixotic) endeavors. Pulsars are a prime example: we know that they come from spinning, magnetized neutron stars, but the precise emission physics responsible for the complex phenomenology is not well understood. Nevertheless, because pulsars are such stable clocks, they netted the discovery of the first extrasolar planets¹ and are the prime laboratories for the study of General Relativity in the weak and strong gravity regimes.² Quasars, also complex and diverse beasts, are used to probe the chemical evolution and large-scale structure of the Universe. So we may indeed take it as a sign of maturation that GRBs have arrived into the noble ranks of the Superlative (Albeit Enigmatic) Probes of the Universe. This chapter is devoted to the “killer apps”³ of GRBs.

The unassailable fact about GRBs that makes them such great probes is that they are fantastically bright and so can be seen to the farthest reaches of the observable Universe. As we shall see, from a practical standpoint, the uniqueness as powerful probes stems from the observation that afterglows are apparently featureless (or at least simply described) and that afterglows eventually disappear from view. For some applications, the inference that the events occur in and around where star formation happens is also critical. As beacons to interesting and far-off places in the darkness, GRBs are indeed marvelous lighthouses (figure 6.1).

6.1 Studies of Gas, Dust, and Galaxies

In the context of understanding the nature of GRBs, the importance of the inferences of gas and dust properties along the line of sight was discussed in §3.1.2 and §4.1.1. But what impact do such inferences have on our understanding of gas and dust in the Universe? We noted previously (§4.1.1) that metallicity can be inferred using the absorption fingerprints of different atomic species. Metallicity influences the efficiency of mass loss in stars and thus how (massive) stars end their life.^* Metallicity may also affect the distribution of masses of stars at birth (the so-called initial mass function). The measurement of metallicity serves as an instantaneous snapshot of the cumulative effects of chemical enrichment in a galaxy caused by generation after generation of stars expelling their nuclear waste as they die. More massive galaxies not only produce more stars over time but also tend to retain the metals expelled during supernova explosions. This leads to an expected “mass-metallicity” relationship and, since galaxy luminosities generally correlate with mass, a similar “luminosity-metallicity” relationship. Measuring these relations and how they evolve with cosmic time thus provides a fundamental view of star formation and galaxy evolution.

Figure 6.1. Bright gamma-ray bursts compared to other bright lighthouses of the Universe (quasars [QSOs] and supernovae). Plotted is the absolute optical magnitude (M_r) versus time since explosion. Recall that magnitudes are a logarithmic scale where more negative values imply a brighter source. A typical galaxy has M_r ≈ − 18 to −22. SN 2006gy was one of the brightest supernovae ever observed. The brightest QSOs found in the Sloan Digital Sky Survey (SDSS) are a factor of ten thousand times brighter than SN 2006gy at peak. The GRBs shown were about a factor of one thousand brighter at peak than the brightest quasar. From J. S. Bloom et al.,ApJ 691, 723 (2009).

Since the discovery and follow-up of GRBs depend little on the host galaxy properties, metallicity measurements using afterglows afford the study of a population of galaxies that were not selected because of their brightness. Thus, GRBs can provide a complementary picture⁴ at the faint end of the luminosity function of galaxies. Initial results are intriguing: GRB hosts at redshifts above z = 2 appear to be more metal poor than predicted by simple prescriptions for chemical enrichment in galaxies based on their brightness. This suggests either an enhanced efficiency of star formation in dwarf galaxies or that SNe winds are more efficient at carrying away metals from galaxies.⁵

Broadband model fits to UVOIR afterglow data reveal a broad distribution of absorption depths due to dust. Most well-studied afterglows exhibit little to no extinction.^*Those that do can be used to infer the properties of the dust in other galaxies. What is measured is the degree to which light at some wavelengths is absorbed when compared to the absorption of light at other wavelengths; this selective extinction is a direct manifestation of the properties of the dust grains themselves. This distribution, in turn, derives from the nature of the dust factories that pollute the GRB sightlines. In the local universe (perhaps even in the last ten billion years or so), the distribution of dust is thought to be dominated by material from the late stages of stars once like our Sun (called “asymptotic giant branch” [AGB] stars). We strive to understand the nature of dust because it affects all our UVOIR measurements of objects and events outside the galaxy.^* Despite the centrality of dust in extragalactic studies, it is important to note that measuring dust beyond the local universe is exceedingly difficult because it requires knowledge of the intrinsic spectrum of the background light source. Quasars and galaxies have complicated intrinsic spectra but GRBs do not, and in this sense, GRB afterglows are potentially more useful for measuring dust. Indeed, the measurement of extragalactic dust using GRB afterglows is becoming an important cottage industry.

In many sightlines, a selective extinction profile commonly seen in the Small Magellanic Cloud (SMC) is preferred over other locally observed profiles (see also §4.1.1). Since the SMC shares many of the characteristics (similar low mass, low luminosity, and active star formation) of GRB hosts, this is not too surprising. A common Milky Way ultraviolet absorption feature (called the “2175 Å bump”⁶) is not seen in the SMC and, likewise, only seldom observed in GRB afterglows. Interestingly, in the well-observed afterglow of GRB 071025 there is good statistical evidence that none of the local flavors of dust can account for the inferred extinction; instead, a (theoretical) dust profile originating from SNe provides a much better match to the data (see figure 6.2). This is the best-fit dust profile using a GRB beyond z = 3 and constitutes some of the best evidence that dust was different in the early universe.⁷ One explanation for this is straightforward: at the high redshift of some GRBs, there was not enough time in the Universe for an appreciable number of stars comparable to the mass of the Sun to evolve to the AGB phase. So at that time the only major contributors of dust were supernovae, not the evolved stars that are the dominant polluters of today’s universe.

GRB afterglows also penetrate galaxies between us and host galaxies. These “foreground” galaxies are completely unrelated to the GRB and any special ingredients (e.g., heavy-element metals) in the host galaxy cauldron that might enhance the rate of GRB production. Like the population of innocent bystanders caught in the crossfire of street fights, GRB-selected foreground galaxies offer up a more random sampling—and thus a more pristine view—of the distribution of galaxies. We first “see” these galaxies in absorption, typically in a common low-ionization metal species of magnesium (Mg II). This ion tends to be both blown to large distances from galaxies (upward of 50 kpc), and it is produced in such large quantities that it leaves an indelible absorption mark on the light that passes through it.⁸ After the afterglow fades we can image the region around the GRB and study the properties of intervening galaxies. Not surprisingly this has given us an unprecedented window on the properties of galaxies on the faint end of the “galaxy luminosity function”: these galaxies might have been detected in other imaging surveys, but they are systematically absent from spectroscopic redshift surveys (which generally target the brightest galaxies). This is a fairly new direction for the field. In the future, we can use several sightlines to infer generic properties about the structure and extent of the halos of gas around galaxies as a function of galaxy properties.⁹

Figure 6.2. Inference of an SN-dominated dust profile in GRB 071025. The top panel shows the observed data in the optical and near-infrared as well as the model afterglow spectrum attenuated by neutral hydrogen in the intergalactic medium (IGM; dot-dashed lines). That is, the dot-dashed line shows the GRB afterglow as it would appear without dust. The dotted line is the dust-attenuated afterglow using a model for SN-dominated dust (from R. Maiolino, R. Schneider, E. Oliva, S. Bianchi, A. Ferrara, F. Mannucci, M. Pedani, and M. Roca Sogorb,Nature 431, 533 [2004]); this line is an excellent fit to the data. The bottom plot shows the “residuals” (differences between the data and the model afterglow plus dust), including a result from a much poorer fit using Small Magellanic Cloud (SMC) dust extinction. From D. A. Perley et al., MNRAS 406, 2473 (2010).

6.2 The History of Star Formation

Understanding star formation (SF) in the Universe is a central pursuit of observational cosmology. At a fundamental level, it helps us understand how we got where we are today: where the elements on Earth came from, why our Sun has the composition it has, etc. But SF also puts into context the diversity of galaxy structures we see and plays an important role in how galaxies evolve with time. The cumulative number of stars in a given galaxy can be inferred by mass and light measurements, albeit with some uncertainties. The instantaneous¹⁰ rate of star formation is measured by a variety of techniques. At optical wavelengths, we measure the intensity of certain emission lines, which gives us some census of hot, big, and newly formed stars. This is not the cleanest picture of SF because we need to assume a ratio between the numbers of less massive stars we do not detect and with those massive stars we do. More importantly, like fireflies in a dense fog bank, heavily obscured star formation goes unseen (at visible wavebands) and uncounted. Since this obscured light will serve to warm the dust that blocks it, we might infer how much nascent starlight we are missing by studying the reemission of starlight energy by dust (at infrared and sub-millimeter wavelengths). Studies of such dust have suggested that upward of one-third of star formation in the distant universe may be enshrouded and thus be undercounted in optical studies. Last, studies at radio wavelengths witness the effects of star formation in the form of emission from supernova shocks. There are many model dependencies in turning a radio flux measurement into an instantaneous star-formation rate (SFR), but, since radio light effectively penetrates dust, the radio-measured SFR tends to be viewed as the most robust.

The gamma-rays and X-rays from GRBs also penetrate dust. And since we believe that long-duration GRBs mark the birthplace (and death site) of massive stars, we get to pinpoint the locations of SF irrespective of whatever dust might get in our way. If some fraction of star formation is truly obscured by dust, then we might expect a similar fraction of GRBs to show evidence for dust extinction in the UVOIR afterglow. In the extreme case of opaque clouds of dust, the UVOIR afterglow might be completely extinguished. In the less extreme case, large amounts of dust might be relatively transparent to infrared light but effectively block ultraviolet light. Therefore, by measuring the distribution of concealed/obscured and unobserved afterglows, we might hope to measure directly the obscuration of all star formation in the Universe.

Faint or missing UVOIR afterglows are numerous—recall that roughly half of all Swift afterglows have no UVOIR afterglow detection. Some of the deficit can be chalked up to observational obstacles, such as when a GRB is detected near the Sun (even in this day and age we UVOIR astronomers still need darkness to do our business!) or if inclement weather hampers quick follow-up by telescopes with enough sensitivity to discover afterglows. GRBs detected during major sporting events, such as the World Cup of Soccer, or during August vacation recess for most of the world’s GRB astronomers, tend to get less attention. But even some GRBs that get adequate rapid and deep follow-up have faint or undetected UVOIR afterglows. This population is referred to as “dark bursts.” By connecting the brightness of the UVOIR afterglow to the brightness of the observed X-ray afterglow at some fixed time, we can ensure that the overall faintness of a GRB afterglow does not admit the event into the dark-burst club. Páll Jakobsson (University of Iceland) and collaborators proposed an operational definition of dark bursts that required the optical afterglow to be fainter than the most liberal synchrotron extrapolation from the X-ray data.¹¹

In one homogeneous sample of thirty rapidly observed GRBs with a robotic 60-inch telescope at Palomar Mountain, fifteen events matched this dark-burst criterion. Of the entire sample, twenty-two (73 percent) had faint optical afterglows detected. In a follow-up study of the sample, my group at UC Berkeley showed that all but one event¹² had a plausible host galaxy detection (or optical afterglow detection). This, along with the observation that some of those hosts appear red (as if extinguished by dust throughout the galaxy), implies that ∼25 percent of (high-mass) star formation in the Universe may be strongly obscured by dust.^* This obscured fraction may indeed be an underestimate since, if the dust is near to the GRB explosion site, the light of the afterglow may quickly destroy the enshrouding dust; GRBs in such a configuration can thus claw themselves out of obscurity to join the UVOIR-detected group. The most extreme example of this is GRB 080607 which was observed to be an incredibly red afterglow, likely embedded in a thick blanket of dust (see also §4.1.1).

Dust issues aside, measuring the rate at which GRBs occur in the Universe as a function of redshift is a promising proxy for a measurement of the universal star-formation rate. There are some important caveats however. First, it appears that, as moss preferentially grows on the northern side of rocks and trees,¹³ GRB progenitors prefer low metallicity (§4.1.2), implying that they serve as biased tracers of SF. Second, our ability to detect the GRB itself depends strongly on its intrinsic brightness as well as its distance from us. Third, and perhaps most important, our ability to obtain a redshift of a GRB depends crucially on the ability to detect its optical afterglow. This is especially true beyond a redshift of 2–3 where emission-line redshifts from host galaxies become exceedingly “expensive” (viz., time expended on big telescopes). Despite these challenges, by modeling the observed properties of the GRB population, both those with redshift and those without, we can infer the rate of GRBs as a function of redshift and compare it to the rate of star formation as inferred by other means. Figure 6.3 shows this inferred rate. We see a rapid rise of the GRB rate from z = 0 (now) to z = 1, and then a sharp decline after z = 4. This is qualitatively similar to the inferred SFR except that the metallicity bias suppresses the GRB rate at low redshift relative to the universal SFR.

6.3 Cosmic Dawn: Measuring Reionization and the First Objects in the Universe

If we appeal only to the finite sensitivity of detectors, we might (on purely demographic grounds) expect that our detailed knowledge of the constituents of the Universe decreases monotonically as we move to higher and higher redshift. This is a reasonable first guess, but (unintuitively) it breaks down at the very highest redshift: we actually know a great deal about the Universe and its constituents around a redshift of z = 1100, about three hundred thousand years after the Big Bang. This is the redshift at which the light of the cosmic microwave background (CMB) originates, the epoch when the evolution of light and ordinary matter decoupled from their until-then-entwined existence; beyond this redshift is an opaque cloud through which we can never peer with electromagnetic eyes.¹⁴ After that epoch, as the Universe cooled, neutral atoms of hydrogen could persist for the first time. Decades of study of the CMB and clouds of primordial gas at lower redshift have given us precise measurements of the distribution of matter at that time. Relative to the diverse and extreme clumpiness we see around us today,¹⁵ the Universe then was a rather simple and placid soup of mostly hydrogen and helium atoms. Importantly, CMB measurements give us a precise (few percent uncertainty) measurement of the time of this epoch since the Big Bang and the distance to the “surface” of the CMB-emitting cloud.

Figure 6.3. The inferred rate of long-duration GRBs, (z), relative to the star-formation rate (SFR) in the Universe _SFR), normalized to the rates at z = 1. The light gray curve shows a smooth fit to the observed SFR data from various studies of galaxies. The data points with error bars (and associated gray region) show the inferred GRB rate from a fit to the Swift data. The dotted lines show what a metallicity-biased SFR might look like, created by imposing a metallicity threshold cut (Z_th/Z no more than 20 percent and 50 percent) on galaxies that can create stars. Remarkably, this metallicity-biased SFR looks similar to the inferred GRB rate, suggesting a strong correlation between metallicity and GRB production. Adapted from N. R. Butler, J. S. Bloom, and D. Poznanski, ApJ 711, 495 (2010).

Our current interests lie at redshifts and distances intermediate to the nearby and CMB bookends. At redshifts less than 1,100 and higher than about ∼10, we know that the first clouds of primordial gas began to collapse, that the first stars were formed, that the first heavy elements were synthesized, and that the first galaxies were assembled. We also know that the Universe underwent a fundamental phase transition, changing from 100 percent neutral hydrogen (at z < 1100) to where almost all hydrogen was a fully ionized plasma by z = 5 (and continues to be to this day).^* That this “epoch of reionization” happens to coincide with the formation of the first stars and structures might suggest a causal connection, but this is far from certain. One of the holy grails in observational cosmology today is to measure the progression of ionization as a function of time and to uncover the source (or sources) of reionization. This epoch in the Universe (about 100–800 million years after the Big Bang) is sometimes referred to as the Dark Ages, capturing both our lack of understanding of that time (figuratively) and the relative lack of stars that shine during it (literally).^*In situ measurements of objects and events beyond z = 7 must then be the true path to enlightenment (again both figuratively and literally). This ability to push observations beyond z = 7 is one of the great hopes for GRBs as probes.

The extreme energy release in GRBs also leads to the extreme brightness of afterglows across the electromagnetic spectrum. As figure 6.1 demonstrates, optical afterglows can be thousands of times brighter than the brightest quasar in the Universe and millions of times brighter than the brightest supernovae and galaxies. Importantly, GRB afterglows also stay brighter than most quasars for hours after the GRB trigger. Therefore, we not only expect to see GRBs from very far away, but we also might expect to see them farther away than other bright sources or events. Allowing ourselves to denigrate the cosmic competition for a moment, we might also note that the intrinsic brightness of the typical quasar and galaxy appears to be decreasing at higher redshift, whereas there is no good evidence to suggest that GRBs are also getting fainter at high redshift.^† Even more fundamental is the recognition that the Universe appears to create the largest (and/or most massive) gravitationally bound structures at later times.¹⁶ Therefore, the progenitors of GRBs and SNe (that is, stars), were likely formed before big galaxies (large collections of stars), which likely formed before the first massive black holes (perhaps a collection of stellar remnants) that power quasars. This uniqueness in space and time is probably only important when considering measurements at redshifts beyond z ∼ 15. Nevertheless, taking off our cheerleader hat, we must admit that GRBs at high redshift are exceedingly rare. As figure 4.2 implies, there is probably only 1–2 bright GRBs per year at the Swift detection threshold beyond a redshift of the highest redshift quasar (z = 6.5 in 2010). So while observations of high-redshift GRBs are the hottest and the most informative tickets in town, the performance is about as frequent as the annual Academy Awards. Even more unfortunate, show times vary, and because of earthly constraints (like rain and daytime hours) we often will not know that there is a spectacular event unfolding until it is too late.

How do we recognize high-redshift events if their redshift is not encoded in observations of the bursts themselves? The simplest answer is that we get a spectrum of the afterglow and look for telltale absorption lines. But the practical answer is more complicated: since we need to decide how to allocate precious spectroscopic resources on large-aperture telescopes, we not only have to discover the precise position of the afterglow quickly,¹⁷ but we also have to gain some confidence that it is a high-redshift candidate.

The good news is that Nature is rather kind on this front. Despite being ionized at some degree, the Universe at high redshift has enough neutral atoms of hydrogen to selectively block out light at a specific wavelength (1216 Å) corresponding to the electronic transition of the n = 1 → 2 state.^* As a photon with this specific wavelength passes by a hydrogen atom in the n = 1 state, it has a high probability of being absorbed, causing an excitation of the atom^† and the removal of that photon from the light train on its way to Earth. Now, since the Universe is expanding, atoms of neutral hydrogen existing at different places (and times) will see the photons more and more redshifted the farther away they are from the event. So while the atoms block the same wavelength of light as they see it, they are effectively blocking different wavelengths of light as viewed by us. Clouds of neutral hydrogen are so numerous in the z > 5 universe that we experience an almost complete blocking of light blueward of the Lyman-α transition shifted by the redshift of the GRB. More precisely, if we look for optical or ultraviolet afterglow at wavelengths

^λobserved <(1 + z) × 1216 Å,            (6.1)

we should not find it. At z = 7, equation 6.1 suggests that we should witness a dramatic fall off (the so-called “Lyman-α break”) blueward of 9728 Å. This wavelength corresponds to the approximate transition between optical and near-infrared wavebands. So in this case we “look” for a bright infrared afterglow without a detectable optical afterglow. Recall that a lack of (or very faint) optical afterglow essentially defines the dark-burst sample (§6.2), and since most dark bursts are dust-extinguished, infrared observations are essential for distinguishing dark bursts (which would appear red at in the infrared bands) from truly high-redshift events (which would appear relatively blue at infrared bands). Importantly, the infrared observations must be made around the same time as the optical observations to minimize the effect of a variable light curve on the inferred colors.

Our prowess at photometrically identifying high-redshift GRBs continues to grow. Access to rapid infrared imaging has improved on both small- and large-aperture telescopes, and new instruments¹⁸ provide simultaneous optical-through-infrared photometric imaging. By 2010 at least four events appeared to have photometric redshifts beyond z > 6, and two of these had spectroscopic confirmation of the inferred photometric redshift. GRB 050904 was at a redshift of z = 6.29, and GRB 090423 was at a redshift of z = 8.2, making it the most distant spectroscopically confirmed source/event in the Universe at the time (see figure 6.4). GRB 090423 occurred just ∼630 million years after the Big Bang, when the Universe was less than 5 percent its present age and eight hundred times more dense than now.

Obviously, the discovery of high-redshift GRBs con-firms the long-held suspicion that the progenitors of GRBs may be formed in the early universe when the typical metallicity of galaxies was lower than in today’s universe.^*This is an important statement about the progenitors of these events.

Figure 6.4. Spectrum of GRB 090423 about sixteen hours after the trigger. The dramatic drop-off of flux density blueward of about 1.1 µm is due to the collective opacity of neutral hydrogen clouds in the intergalactic medium along the line of sight. Letters (“z,” “H,” “SZ,” etc.) are names of specific optical and infrared filters where imaging data were obtained. Inset shows a fit to the inferred redshift and hydrogen column density in the host galaxy of the GRB. Adapted from N. R. Tanvir et al., Nature 461, 1254 (2009).

So what do we do with a high-redshift GRB? We saw in §6.2 that the incidence of high-redshift events helps us determine the rate of star formation in the Universe. But individual spectra are also potentially very powerful as diagnostics of the Universe itself. The “break” around redshifted Lyman α is not infinitely sharp: it progresses over tens of Ångströms, as dictated by the quantum mechanics of the electronic transition, effects of dust, the amount of neutral hydrogen surrounding the GRB in the host galaxy (N_H; see §4.1.2), and the ionization state of the Universe. By exploiting the powerlaw nature of the intrinsic afterglow spectrum, in principle, we should be able to determine the ionization state of the Universe at the time of the GRB. In practice, the two highest redshift GRBs with spectroscopic follow-up exhibited large N_H values, which washed out the measurement of _H. The hope is, based on low-redshift demographics of N_H values, that something like 10–20 percent of z > 6 GRBs will have low-enough N_H to yield a direct measurement of _H. One such measurement would be a great achievement and a set of such measurements at different redshifts would allow an unprecedented view of _H(z), giving direct insight into the timing and origin of reionization.

Each well-observed GRB afterglow also allows us to constrain the gas-phase metals that it has encountered on its long journey to us. Along with a measurement of N_H, then, we can constrain the metallicity in surrounding subgalactic and galactic scales of the GRB (see also §4.1.2). A single absorption-line metallicity measurement beyond z = 7 would be novel and would certainly inform our understanding of the growth of stellar structure in the early universe. Perhaps more enticingly, we should be able to study the pattern of abundances of different metals with respect to each other. Just as the characteristics of dust may be shaped by the types of environments in which it was created, so too might abundance patterns reveal the origin of metals in the early universe. Beyond some redshifts the abundance pattern will be dominated by the ways in which the first stars (so-called “Population III” stars)¹⁹ create and subsequently expel heavy elements. There is reasonably strong theoretical evidence to believe that Pop III stars are sufficiently different^* from later generations in their lifecycles and death that we might witness a unique imprint in GRB absorption lines.

Irrespective of whether we have the sensitivity in our spectrographs to measure the effects of unique nuclear chemistry from Pop III stars in individual GRB afterglow spectra, it should be clear that GRBs offer a special service to high-redshift enthusiasts: they boisterously announce themselves and tell us where to look. Like any good lighthouse, high-redshift GRBs tell us where a special place in the Universe is. Months or even years later we can return to the position of GRBs and study the galaxy and even galaxy-cluster scales around where we know a high-redshift structure must exist. Knowing the GRB redshift ahead of time lets us tune up our imagers and spectrographs to get the most detailed physics out of the subsequent study.²⁰

6.4 Neutrinos, Gravitational Waves, and Cosmic Rays

We have so far concerned ourselves with the detection of light from GRBs. But of course messages from the heavenly bodies are delivered not just with electromagnetic light but via other forms of energy. The study of these “alternative” forms of radiation—which are very likely the manifestation of important physical processes in GRBs—should give us a very different perspective on the objects and events involved. The hope too is that we can use these other forms of emitted energy to test basic questions in physics.

6.4.1 Gravitational Waves

If indeed short bursts are due to the merger of compact objects (see §5.2), then in most scenarios, we expect that the GRB event would also be accompanied by a telltale signature of gravitational waves. Russell Hulse and Joseph Taylor demonstrated that the slow leakage of gravitational radiation from an NS–NS binary system could explain the gradual decay of its orbit. In about one billion years, the Hulse/Taylor binary will deteriorate to the point where the orbit will rapidly decay and the NSs will coalesce. As we have discussed (§5.2.1), in the last few milliseconds before the death plunge, the system will release a tremendous amount of energy in the form of gravitational waves.

The gravitational wave (GW) signal is oscillatory with a frequency comparable to the orbital time of the binary; so as the orbit decays, we expect the frequency of the signal to rise rapidly, like the chirping sound of birds.^* And just as chirps from different species are distinctive, each GW merger event can be analyzed to reveal uniquely the properties of the merging objects (mass, size, perhaps even their internal structure).

Gravitational waves deform space as they propagate, and it is this deformation that gravitational-wave detectors measure. By 2010, there was no credible direct detection of gravitational waves from any astrophysical source despite the remarkably impressive sensitivities obtained by major facilities.²¹ The main impediment is that the most sensitive facilities can only “hear” gravitational waves from NS–NS mergers out to ∼10 Mpc, wherein the rates of NS–NS mergers are expected to be only about one event per ten years. Fortunately the Advanced LIGO project (starting science operation around 2015), which will improve the NS–NS sensitivity to 300 Mpc or so, should detect tens of events per year.²²

Clearly a detection of gravitational waves emanating from the same time and place as a short-duration GRB would be a smoking gun for the progenitors of such events (§5.2.2). If the merger of an NS–NS or NS–BH binary caused the event, the chirp signal should be an unassailable confirmation of such configurations. In a broader context, the detection of a chirp from a GRB would be the first direct measurement of a gravitational-wave event and open up several new vistas on extreme physics. If, for instance, a neutron star is involved in the merger, then analysis of the chirp signal would allow a sensitive probe of the internal structure of neutron stars and reveal the nature of matter in an important new regime of high pressure and high density.²³ It would serve not only as a confirmation of General Relativity but allow stringent tests of General Relativity in a new regime where gravity is strong.²⁴

6.4.2 Cosmic Rays and Neutrinos

Cosmic rays—heavy charged particles like protons and nuclei—bombard satellites in space and smash into the atoms in the Earth’s atmosphere. Most low-energy cosmic rays are produced in the outskirts of the Sun,²⁵ and observational evidence presented in the mid-1990s showed that high-energy cosmic rays come from the remnants of supernovae. Essentially, SNe act as gigantic particle accelerators: charged particles moving with the nonrelativistic flow of an SN get bounced around in a magnetic field and eventually achieve speeds near the speed of light.

The very highest energy cosmic rays inferred^* cannot have been produced in SN shocks, so some set of astrophysical entities must be responsible for the extreme particle acceleration. The origin of these cosmic rays were a complete mystery for decades. In 1995, Eli Waxman (then at the Institute for Advanced Study, Princeton) and Mario Vietri (Università di Roma) noted that the relativistic fireballs of GRBs would make ideal particle accelerators²⁶ and could easily produce cosmic rays with energies > 10²⁰ eV. Moreover, the rate of high-energy cosmic-ray detection was reasonably consistent with the approximate rate that GRBs could produce such particles. While GRBs remain one of the most viable and tantalizing production mechanisms for high-energy cosmic rays, a direct association of a given particle with a given GRB is very unlikely.^*

However, if the hypothesis about particle acceleration is correct, some of the high-energy charged particles produced in GRB fireballs will interact with the gamma rays of the event itself.²⁷ These interactions lead to the production of smaller-mass particles (like positrons) and high-energy neutrinos.^† Neutrinos produced in the fireball should have energies around 10¹⁴ eV. Neutrinos might also be produced from the decay of fast-moving protons in the external reverse shock, leading to a brief “neutrino afterglow” (typical neutrino energies of 10¹⁸ eV).²⁸

Unfortunately, neutrinos are exceedingly difficult to detect and identify with their source: though about ten billion Solar neutrinos pass through an area the size of a penny in one second, only twenty-four neutrinos from an astrophysical source other than the Sun have been identified (and all were from a nearby supernova called SN 1987A). The good news is that, unlike cosmic rays produced in GRBs, we expect neutrinos from a GRB event to arrive nearly simultaneously with the gamma rays. This simultaneity would allow GRB neutrinos to be distinguished from neutrinos that are unrelated to the GRB and thereby improves the sensitivity of detection. The South Pole experiment called IceCube (reaching full sensitivity by 2012) and the Japanese Experiment Module on the Extreme Universe Space Observatory (JEM-EUSO; to be launched in 2013) are both poised to find the first neutrinos from GRBs.

Just as the detection of gravitational waves would be a major vindication of some progenitor models, the discovery of GRB-produced neutrinos would also be a major triumph for the fireball and afterglow theory. And as with gravitational waves, GRB-neutrinos would open up new perspectives on basic physics. For example, since neutrinos are particles with mass and gamma rays are not, it might be possible to devise stringent tests of the notion that gravity acts the same on light and mass.^*

6.5 Quantum Gravity and the Expansion of the Universe

The extravagant reality of a universe accelerating away from itself by unknown means was a wonderfully fitting close to the twentieth century. Just as Hulse and Taylor were awarded the Nobel Prize for the astrophysical confirmation of gravitational waves (a basic prediction of General Relativity), inferences of a dark energy force counteracting the attraction of gravity pointed to a future of scientific inquiry beyond Einstein’s dreams.^* Coupled with technical advances in string theory, in some observational solidification of inflation theory,^† and in sizing up the contribution of dark matter, the twenty-first century stage was set to start grappling with physics decidedly outside our comfort zone.

The very heart of Relativity—the notion that the speed of light is constant—has been prodded by beyond-Einstein thinkers for decades. Some believe that on the smallest size scales (something like 10²⁵ times smaller than a hydrogen atom!) empty space becomes grainy enough to affect the propagation of light. On these scales, the deterministic tenets of gravity break down, and the unwieldy uncertainty of quantum mechanics reigns supreme. One ramification of this is that photons of different energy may travel at speeds slightly different than what we now think of as the universal speed c. Over very long distances, then, two photons with different energies beginning a cosmic race at exactly the same time will arrive at different times. Some have suggested that in distant GRBs, the arrival time of high-energy (GeV and TeV) photons relative to low-energy photons places the tightest limits today on this so-called Lorentz Invariance Violation (LIV) effect.

Some of the best-timed events, such as GRB 090510, do show a delay of photon-arrival times, but these are attributed to unmodeled intrinsic effects at the source (that is, reflecting when the photons started their race). Since we know so little about the emission mechanisms of GRBs compared to how much we know about General Relativity, it seems that all departures from equal-arrival time, even if tantalizingly consistent with a single LIV theory,²⁹ will always be chalked up to a more mundane explanation (such as a tauntingly evil framework of intrinsic GRB physics that mimics an LIV effect some hope to see; or, even worse, a conspiratorial evolution of GRB physics with redshift that gives a seemingly exciting LIV result with many GRBs).

On such grounds, most scientists, I believe, would agree that the extraordinary evidence^* for a positive detection of LIV might never be obtainable. But a more subtle concern, almost epistemological in nature, arises: if assumptions about the emission mechanism of GRB physics must also be invoked to constrain LIV, isn’t extraordinary evidence required as well?^† I claim yes, and I also claim that extraordinary evidence of this sort has not been presented thus far. Since my thoughts on this subject do not necessarily capture a widely held consensus, a positive spin on this LIV controversy is that, in just this one respect, GRBs are nevertheless pushing the envelope as probes of ideas beyond standard physical theory.

The idea of using GRBs to measure the evolution and fate of the Universe is also an exciting, albeit controversial, direction. Unfortunately, measuring and understanding the evolution of the Universe (“cosmography”) is like trying to decipher the ingredients in a cake by carefully watching how the dough rises in the oven. That is, the task is immensely challenging. Thankfully, however, working under the aegis of General Relativity, the parameters that encapsulate the contribution of these ingredients are both finite and, in principle, measurable. The main task is to determine observationally how the redshift of the source maps to its distance from us. This is typically done by using large sources/structures whose sizes we can infer by some basic physics (“standard rulers”) or whose intrinsic brightnesses we can infer by observations of local analogs (“standard candles”).

GRBs are clearly not standard candles, in that the luminosity and energy release from burst to burst can vary by orders of magnitude (e.g., figure 4.3). However, if other observable parameters, such as afterglow break times, E_peak, duration, and such, could be shown to correlate intrinsically with luminosity or energy, then the events could be “standardized,” implying that a GRB redshift measurement could also be tied to its true distance from us. Perhaps the closest analogy to “standardization” would be if you could tell the wattage of a newly lit light bulb by observing how fast it turned on or by its color. There are indeed many such correlations with GRBs, such as the so-called Amati relation connecting the isotropic-equivalent energy release in gamma rays (E_iso)to the observed peak energy (E_peak). Many claims have been made that by using a correlation (or combining many of them) the luminosity distance of a GRB may be inferred independently of the redshift. This remains a controversial subject, however, because many others (including those in my group) have suggested that the correlations come about largely as detector-threshold effects and, therefore, hold little cosmographic power. Regardless, given the statistical uncertainty in the relationships and those endemic to individual events measurements, it is difficult to see now how long-duration GRB energetics studies could ever be competitive with more-precise cosmographic tools.

One promising avenue for the cosmographic utility of GRBs comes not from long-duration bursts but from gravitational-wave detections of short-duration GRBs. If a GW event is indeed discovered, the impact on our understanding of the progenitors would be profound (§5.2, §6.4.1); the cosmographic connection is more subtle but no less sublime. GW chirps encode the distance to the source but not the redshift of the source^* (recall that both redshift and distance are required for a cosmographic measurement), yet GW sources are exceedingly difficult to pinpoint on the sky. This means that obtaining a redshift of the galaxy host is even more hopeless than for GRBs before the afterglow era. But if a GRB is also found coincident in time with a GW event and its position is consistent with a crude localization of the GW event, then the afterglow of the GRB can be used to identify the host galaxy and measure its redshift.

The best estimates are that with a dozen or so GW–GRB events we could measure the expansion rate of the Universe locally^* to an accuracy of just a few percent, at least a factor of two better than currently obtained by other means.³⁰ This, in turn, would improve our understanding of the precise contribution of dark matter and dark energy to the dynamics of the Universe. Importantly, this measurement would rely only on an appeal to General Relativity (to map the chirp signal to a distance) and so would be an independent check of cosmographic measurements made by other means.

6.6 The Future of Gamma-Ray Bursts: At the Nexus of Physical and Astrophysical Inquiry

The pace of transformative breakthroughs seen in the early days of the afterglow era has slowed. Frameworks for understanding the high-energy events, afterglows, and progenitors have settled in. And while the internal-external shock model still has some troubling disconnects with the data, it is difficult to see how the basic understanding of the events (compactness of the energy source, relativistic motion, association with stellar death) could be shown to be vastly wrong. To be sure, there will be new GRB-discovery satellites that will yield fainter events at a higher rate. Some events will confirm, clarify, and (perhaps) expand the notions we have about the diverse zoo of progenitors. Some will be oddities that cause us to question our basic inferences about the physics that gives rise to the events.

On many fronts, the equanimity of a maturing endeavor has replaced the precociousness and verve of basic discovery. And yet, why does the GRB field continue to feel so vivacious and so inviting? One answer might be that there are just so very many interconnected facets of the GRB puzzle that no one subfield may be considered truly solved until the entire picture is seen with greater clarity. Even once rock-solid inferences such as jet breaks appear more tenuous with the enlightening onslaught of more data. Another answer is that, for all the vagaries about the phenomenon itself, GRBs have proven themselves to be unique and useful probes in so many burning questions we have about the physical universe. Soon, events discovered beyond z = 10 will allow us to probe the first galaxies and stars, and more nearby GRBs might offer a powerful gateway to the dynamic gravitational-wave and neutrino universe. Indeed, as we savor the richness of GRBs themselves, it is difficult to point to any other phenomenon that touches not just on such a wide range of pursuits but on inquires that are so central to twenty-first-century astrophysics and physics.

 

^*This is both figuratively and literally true: GRB practitioners on the front lines of observation are often awoken at odd hours of the day and night by text messages from satellites announcing new GRB discoveries. GRBs are terrible strains on interpersonal relationships!

^*Higher metallicity generally makes for more efficient mass loss.

^*This reflects both intrinsic and extrinsic selection effects. Very extinguished afterglows—externally dimmed at UVOIR wavebands—are less likely to be detected than those afterglows that have not been absorbed as much; we will explore these so-called “dark bursts” in §6.2. Such high degrees of absorption in afterglows are fairly rare, not too surprising given that the majority of the light coming from ordinary stars in the Universe is also not heavily obscured by dust. In the cases where the immediate environment of the GRB is itself dusty, the afterglow may destroy enough dust in the first few seconds to erase the signs of the enshrouding dust.

^*Dust, for instance, is a major contributor to the uncertainty of cosmological parameters derived from standard-candle measurements. See §6.5 for an expanded discussion.

^*There are some alternative possibilities for the nondetection of optical afterglows that we considered in our study but did not favor, such as different afterglow spectra with intrinsically fainter optical emission. See D. A. Perley et al., AJ 138, 1690 (2009).

^*We seek to measure H(z), the average neutral fractional of hydrogen at a given redshift. The value H(z) is the ratio of the number density of neutral hydrogen atoms divided by the number density of neutral hydrogen atoms plus protons. At z > 1100, H(z) = 0. At redshifts just less than z = 1100, H(z) = 1 and likely remains near unity until around redshift 20–10 when it plummets to near zero during the epoch of reionization. At z < 5, H(z) is very small (< 10⁻⁵).

^*However, this is also the time when the Universe as we know it begins to blossom; so a more rosy view of this epoch is captured in the name “Cosmic Dawn.”

^†This makes intuitive sense if you believe that most long-duration GRBs are due to the death of massive stars (at a qualitative level and putting aside the metallicity issues, why should a GRB progenitor care substantially about what time it is since the Big Bang?).

^*This is the Lyman-α transition, from the ground state to the first excited state.

^†The atom will eventually reemit the energy, in the form of a photon, as it falls from the n = 2 to n = 1 state. Since it has a very low chance of reemitting the photon in the same direction it was heading before it was absorbed, the incoming beam of light at that wavelength is effectively extinguished.

^*The average metallicity at z = 7 was certainly lower than −2 (that is, one hundred times less metal abundance than our Sun) and probably closer to −3 (one thousand times less than the metal abundance of the Sun).

^*In particular, they are thought to be much more massive than the average stars formed today.

^*Indeed, the GW signal of merger events is often referred to as a chirp signal. Interestingly, the last moments of NS–NS mergers produce chirp signals in the audible frequency range. You can “listen” to such mergers at a website maintained by Scott Hughes at MIT: http://gmunu.mit.edu/sounds/comparable_sounds/comparable_sounds.html.

^*Around 10²⁰ eV, comparable to all the energy required to light at 10 W light bulb for one second! This is several orders of magnitude higher than the particle energies obtained in the Large Hadron Collider (LHC).

^*Magnetic fields in the space between galaxies are large enough to cause the charged particles to take a meandering path. There are two effects of this: (1) Most particles from a given GRB will be deflected away from us, and (2) the time delay due to the meandering between the GRB event and the particle event will be about ten million years. This is too long to wait!

^†Neutrinos, light-weight elementary particles, interact only weakly with matter and fly through the atmosphere, you, and the Earth almost unnoticed. Most low-energy neutrinos are produced in the Sun as a by-product of the nuclear fusion that powers the star.

^*This is the so-called weak equivalence principle, one of the tenets of General Relativity. See E. Waxman,Royal Society of London Philosophical Transactions Series A 365, 1323 (2007) for a review.

^*To be sure, as noted already in §5.1.3, Einstein did include a cosmological constant in some of his original General Relativistic (GR) solutions for the dynamics of the Universe. While a seemingly good parameterization of universal acceleration, the apparent existence of a nonzero cosmological constant does not tell us why it exists. The answer—the physical origin of acceleration—lies decidedly beyond GR.

^†Inflation posits another acceleration period in the Universe, just after the Big Bang.

^*Carl Sagan is famous for saying that extraordinary claims require extraordinary evidence.

^†That is (more formally), fail to disprove the null hypothesis of different arrival times as due to a breakdown in General Relativity.

^*This is basically the opposite of what most astronomers are used to: redshift of an astronomical source is relatively easy to obtain, but distance is not easily encoded in our observations.

^*Measuring an important cosmological parameter called “Hubble’s Constant.”