Quick viewing(Text Mode)

Globular Clusters

Globular Clusters

The Dynamic Lives Globularof Clusters

These ancient stellar systems continue to surprise astronomers with a host of fascinating phenomena.

By S. George Djorgovski

38 October 1998 Sky & ©1998 Sky Publishing Corp. All rights reserved. lobular clusters are the delight of amateur and professional astronomers alike. The oldest star systems known, globulars witnessed the birth of our These models were spectacularly . They now serve as laboratories successful. They predicted cluster for theories of and structures in excellent agreement with evolution. Extensively studied since the the observations available at the time, G and they still provide good fits with the birth of modern , globulars roughly like those between molecules in continue to yield surprising results and greatly superior data we have today. a uniform-temperature gas. The However, cluster models developed by challenge our understanding. are not physically colliding in this celes- The simple beauty of globular clusters Hénon at about the same time did tial pinball game, just deflecting each something entirely different: his simulat- is that they combine an elegant spherical other gravitationally. This energy ex- symmetry with the glittering richness of ed clusters didn’t stay in a slowly chang- change eventually leads to thermal equi- ing equilibrium. Instead, the models pre- thousands of individual stars. This ap- librium. In a typical cluster this takes parent simplicity has enticed astrono- dicted that globulars would fall in upon about a hundred million years, during themselves in a process called the gravo- mers to construct mathematical mod- which an individual star may cross the els of them. Early attempts arrived at a thermal catastrophe, or core collapse. cluster a hundred times. Thus there has Hénon reasoned like this: at any basically correct picture: collections of been plenty of time for globular clusters stars held together by their aggregate given moment the random motions of to reach relaxed equilibrium states, since stars in a cluster’s core can be consid- , with members that move ran- they formed at least 10 billion years ago. domly like molecules of gas, interacting ered to balance the gravitational poten- The King-Michie simulations pro- tial of the stars’ mutual attractions. with Newtonian gravity. In simple sys- duced globular clusters with cores of tems with a limited number of essen- When a fast-moving star escapes the nearly uniform stellar density. The con- gravitational pull of the cluster core, tially pointlike masses (such as our centration of stars then gradually de- or a widely separated bi- taking with it some kinetic energy, the clines out to an invisible boundary core must shrink a little. The remain- nary) the dynamics are mostly regular, called the cluster’s tidal radius. Here the predictable, and even boring. However, ing stars have to move a bit faster to gravitational attraction of the Milky compensate for the increase in the putamillionorsostars together in sev- Way is stronger than that of the cluster eral cubic light-years of space, let them gravitational binding energy of a itself, and stars become unbound and denser core. More of them can then es- orbit a sizable disk galaxy, and many join the galaxy’s stellar halo. Globular interesting new phenomena result. clusters were thus predicted to evapo- rate slowly, with galactic tides constant- Competing Theories for ly stripping away their stars. Collapsing Clusters The first successful dynamical models of globular clusters were developed in the Facing page: One of the southern sky’s showpieces, (NGC 104) glimmers from early 1960s by Richard Michie, Ivan 15,000 light-years away with the combined light of perhaps a million stars. Courtesy the Eu- King, and Michel Hénon, each working ropean Southern Observatory. Above: An easily located jewel in , M15 is a prime ex- independently. The starting point for ample of a highly evolved cluster that has undergone core collapse. This color-composite Michie and King was the ongoing ran- image was taken with the Canada-France-Hawaii Telescope. Courtesy Peter B. Stetson, Do- dom encounters between cluster stars, minion Astrophysical Observatory. Below: Many amateurs’ first acquaintance with globular clusters came courtesy the Messier catalog, named for its 18th-century author, .The Messier globulars depicted here and on the following page were photographed 1 by Evered Kreimer with a 12 ⁄2-inch reflector. North is up with east to the left in each photo. M M M

©1998 Sky Publishing Corp. All rights reserved. cape, and so the process repeats itself of a collapsing globular cluster? of the cluster as a whole! Here was the and accelerates. This eventually leads to Such an energy source does exist, in resolution of Hénon’s paradox: binaries a rapid shrinking of the cluster core. the form of binary stars. Binaries can can serve as energy sources that stabilize The density profile of such a cluster is exchange energy during close encoun- globular-cluster cores against collapse. much steeper than the uniform cores ters with single stars, thereby becoming And thus both King and Michie on of the King-Michie models, declining either more or less tightly bound. one side and Hénon on the other were approximately as the inverse of the ra- These interactions can be very com- right: globular cluster cores do col- dius squared. plex, resulting in the formation of a lapse, but the collapse is arrested by At first glance, there doesn’t seem to temporary triple-, the dis- the presence of one or more hard bi- be any way to stop the collapse and ruption of the binary, or the exchange naries. Precollapse clusters can look stabilize the cluster. But since globular of partners. When all is said and done, just like the King-Michie models, and clusters abound, and none seem to however, in most cases the interaction some stabilized postcollapse ones can have little quasars within them, ends with a binary flying off in one di- too. But clusters that have undergone a Hénon’s models were all but disregard- rection and a single star flying off in recent collapse and recovery still retain ed for more than 20 years. Yet subse- another. And this is where the miracle the characteristic Hénon density pro- quent theoretical investigations in the happens. As it turns out, so-called hard file, with its very small core. In the 1980s by Donald Lynden-Bell, Haldan binaries — those with binding energies mid-1980s, in the first systematic sur- Cohn, Jeremy Goodman, and many larger than the average among stars in veys of globular-cluster structure, others have essentially confirmed and the cluster — tend to lose energy in about one-fifth of all galactic globulars extended Hénon’s ideas. such encounters. They become ever showed the characteristic post-core- Virtually the same process of core more tightly bound and more likely to collapse morphology — a cusplike collapse happens in , but provide energy to the next star that density distribution near their centers. with a crucial difference: in protostars passes by. Examples include M15 and M30. On the core becomes hot and dense Extensive numerical simulations of the other hand, M5 and M13 are enough for thermonuclear reactions to binary-star/single-star interactions per- King-Michie type clusters. We now be- start. These reactions provide energy formed in the 1980s have confirmed lieve that most globulars evolve only that compensates for thermal losses, this prediction. Binaries provide a huge gradually toward core collapse, unless preventing collapse. But what can sub- reservoir of energy; a single hard binary they evaporate first. stitute for nuclear reactions in the core can have a binding energy equal to that Cluster cores may never settle down

Left: This simulation shows the interaction of a single star with a “hard” (tightly bound) binary in the core of a globular cluster. Such interactions typically provide kinetic energy to passing stars at the expense of the binary’s orbit, which tightens. Eventually, the binding energy of a frequent- ly star-crossed binary can rival that of the entire cluster. Below: From the relatively diffuse exem- plars on the bottom of the previous page to the more condensed ones shown here, globulars present a wide range of morphologies to eyepiece observers. However, quantitative bright- ness measurements are required to tell which clusters’ cores have actually collapsed.

Opposite page, bottom: The origin of a globular cluster’s core collapse. As more stars are lost the process accelerates, until it is halted by energy pro- vided by “hard” binary stars. Opposite page, top: At any given time, the core of a collapsing globular cluster has a uniform density, while just outside the core the density sharply decreases (left side of diagram). As time passes the core shrinks and the density increases, but the

E. STERL PHINNEY, CALTECH / STEIN SIGURDSSON, UNIVERSITY OF CAMBRIDGE overall shape of the density profile remains roughly constant. M M M

40 October 1998 Sky & Telescope ©1998 Sky Publishing Corp. All rights reserved. completely: core collapses and bounces integrated star clusters along with ters undergo core collapse every billion can recur in a cycle called gravothermal some dwarf satellite . The 150 years. In that same period, several clus- oscillation. Obviously, a collapse-and- or so globulars surviving today are ters evaporate under the influence of recovery sequence cannot be observed probably just a small fraction of those tidal shocks. Eventually there will be directly, as it may take tens or hundreds that once populated the . none left! of millions of years to complete, but Tidal shocks can also accelerate the Everything said so far presumes that the cycle’s existence seems convincingly evolution of clusters toward core col- stars can be treated as point masses in- established in numerical experiments. lapse. Whether a cluster will evaporate teracting gravitationally. But stars are before it has a chance to undergo col- extended objects and they can get very Cluster Environments: lapse depends on its initial mass and close to one another, especially during a External and Internal concentration (being high in both cluster-core collapse. In our neck of the So far we have neglected an important makes for sturdier survivors), as well as galaxy, there is about one star per 3 aspect of the story. Far from being iso- on the cluster’s position in the galaxy. cubic light-years, and direct stellar col- lated systems, globular clusters evolve Shocks are stronger for clusters whose lisions are extremely unlikely. But in within the tidal field of the — orbits pass closer to the some globular-cluster cores, many mil- that is, within its uneven and ever-chang- and more often through the plane of lions of stars can occupy the same ing gravitational influence. As clusters the disk. On the basis of such an analy- amount of space. Under such condi- pass through the or the sis, David Chernoff (Cornell Universi- tions physical interactions and colli- disk, they are subject to so-called tidal ty) and his collaborators predicted that sions of stars can and do occur. In shocks, in which tides suddenly increase most clusters near the galactic center some cases two stars merge, producing or decrease in strength. These shocks, should have collapsed cores. Indeed, originally investigated by Lyman Spitzer, there is a larger fraction of post-core- 1012 Jeremiah Ostriker, and their collaborators collapse clusters found in the inner Evolution of in the 1970s, can be very effective in has- Milky Way. Their more diffuse siblings 1010 Core Collapse tening the dissolution of clusters. (Think have been obliterated. Relatively sparse about driving fast over speed bumps and globulars can be found only in the 108 you will understand why.) It is now halo’s outer reaches. Models of the generally believed that our galaxy’s en- Milky Way’s entire globular-cluster sys- 106 tire stellar halo was produced from dis- tem suggest that nowadays a few clus-


Relative core density time 100

1 1 0.1 0.01 10 –3 10 –4 10 –5 10 –6 Relative distance from center

Cluster core shrinks a Fastest stars escape from little; its potential the core, taking away energy increases some kinetic energy

Stars move faster in order to balance the increase in potential energy

©1998 Sky Publishing Corp. All rights reserved. Sky & Telescope October 1998 41 a single, more massive object. In weaker glers. Discovered first in M3 more than clusters most astronomers now agree encounters the outer envelopes of red 40 years ago by Allan R. Sandage (as a that they are the product of direct stel- giants are stripped away. Close binaries part of his doctoral thesis at Caltech), lar collisions that result in a single star can also form, allowing mass exchange these stars seemed to be lagging behind with a mass higher than that for the to take place. In any case, the subse- the rest of the cluster’s members in their main-sequence turnoff. Some blue quent evolution of the affected stars is evolution. That is, all stars above a cer- stragglers have been shown to be in modified, sometimes profoundly. tain mass and in the globu- close, eclipsing binaries. They are ex- Such processes have spurred one of lar have evolved off the , pected to eventually dissipate enough the more active topics in globular-clus- as shown by its Hertzsprung-Russell of their orbital energy to merge. ter research today. The details of these (H-R) diagram. Yet the blue stragglers, More subtle evidence for stellar in- interactions are not yet known, and which fall above this cutoff, remain on teractions that can affect a star’s evolu- observed phenomena continue to baf- the main sequence (see below). tion is seen on the so-called horizontal fle both dynamicists and stellar-evolu- Blue stragglers have now been seen branch on the H-R diagrams of some tion theorists. However, evidence for in many globulars, and in all cases they globular clusters. In the course of nor- unusual stellar populations in many seem more concentrated toward the mal evolution, a star like the clusters is strong and abundant. cluster center than the rest of the visi- spends most of its lifetime on the main ble stars. This suggests that they have sequence, burning hydrogen in its core. Blue-Straggler Stars higher masses than either red giants or Once the hydrogen fuel is spent in the Among the long-standing puzzles in normal main-sequence stars, since center, the star leaves the main se- globular-cluster research is one such heavier stars are expected on average to quence and becomes a . Even- , so-called blue strag- sink farther toward the cluster center. tually the star’s helium core ignites. This is consistent with The giant loses much of its extended the plotted positions of envelope as it descends on the horizon- blue stragglers on the tal branch. Where exactly a star ends Asymptotic H-R diagram. Although this step in its evolution depends on 1000 giant branch there are many proposed the amount of mass lost in earlier AGB Red RGB mechanisms for the ori- phases. Blue horizontal Red giant branch gin of blue stragglers, in As it turns out, the color of horizon- branch RHB the context of globular tal-branch stars depends mainly on the 100 BHB

uminosity () Top left: A star spends most of its lifetime on the main sequence, burning hy- L 10 BSS Blue- drogen into helium. When the core hydrogen is exhausted the star has TO straggler reached the main-sequence turnoff, evolving into a red giant while burning stars Main-sequence turn off hydrogen in a shell around an inert helium core. (Blue-straggler stars form an extension of the main sequence past the cluster’s .) Soon the he- 1 MS Main sequence lium core ignites, and the star sheds some of its envelope and enters the hori- zontal branch.Those that have lost only a moderate amount of mass as red gi- ants end up on the red horizontal branch, while those that have lost more O BAFG KM Spectral type enter the blue horizontal branch. The next phase is the , with helium burning in a shell around an inert carbon core, and hy- 12 drogen burning in another shell around the first one. Lastly, a star may then 13 NGC 2808 shed its envelope and become the nucleus of a .

14 Below left: The H-R diagram for globular cluster NGC 2808 shows 15 all of the sequences mentioned in the generic dia-

(V) gram above. The placement of stars is scat- 16 tered by measurement errors, which are 17 larger for the fainter stars. NGC 2808 shows peculiar gaps on 18 its blue horizontal branch;

19 their origin is still not understood. Apparent Apparent 20


22 DATA: CRAIG SOSIN AND COLLABORATORS 0 0.5 1 1.5 Color index (B–V)

42 October 1998 Sky & Telescope ©1998 Sky Publishing Corp. All rights reserved. amount of material they possess other the more mass a star loses, the bluer it not yet been found. Explaining the ob- than hydrogen or helium, also known will appear. These astronomers have served phenomena remains a challenge as their . Stars in more found that among the clusters of a for theorists, just as their measure- metal-rich clusters are red, while those given intermediate metallicity, the ments are a challenge for the observers. in more metal-poor clusters are blue. morphology of the horizontal branch Superb data on globular-cluster cores Since the chemical composition of depends on the cluster density and are essential for this task, and the story stars within a single globular cluster is concentration, in the sense that denser is still unfolding. highly uniform, there should be a sin- and more concentrated clusters tend to All this is just a small part of the on- gle, well-defined bunch of horizontal- have more pronounced blue tails of the going saga of trying to understand branch stars in any given cluster. In horizontal branch. Pecci and colleagues these fascinating objects and their evo- most clusters this is indeed the case, proposed that this is a consequence of lution. Globular clusters are fun to ob- but there are many exceptions to this stellar interactions, such as grazing col- serve from Earth. But consider some metallicity rule; obviously, some other lisions or an enhanced formation of bi- lucky creatures living on a in factor must be at work. This is the no- naries in denser clusters, which would the core of a rich globular cluster — torious “second parameter” effect lead to some additional stripping of now that would be a starry sky! (metallicity being the first parameter red-giant envelopes. Giampaolo Piotto determining the morphology of the (University of Padua) and his collabo- S. George Djorgovski is a professor of as- horizontal branch). There have been rators found that red giants seem to be tronomy at Caltech. His scientific interests many candidates proposed for the depleted near the centers of some post- include observational cosmology, digital sky mythical second parameter, including core-collapse globulars. Again, stellar surveys, globular clusters, and gamma-ray small variations in age between globu- dynamics and interactions in globular bursts, among other topics. He can recognize lar clusters. clusters can affect the evolution of maybe three or four and However, as pointed out by Flavio their stellar populations. couldn’t find a globular cluster using binoc- Fusi Pecci (University of Bologna), ulars if his life depended on it. Roberto Buonanno (Astronomical Ob- The Effort Continues servatory of Rome), and their collabo- The exact mechanisms rators, many other factors can influ- responsible for these NGC 2808 ence the appearance of a globular’s puzzling effects within horizontal-branch stars. For example, globular clusters have NGC 6864 NGC 6558

Right: The surface-brightness profiles on the left are for two relatively un- evolved clusters with large, flat cores, while those on the right are for two clus- ters with sharply dropping densities characteristic of core collapse. Courtesy NGC 6624 the author. Below left: This color-composite image re- veals the central half arcminute of M15. Astronomers are unable to resolve its ultracompact core fully, even with Hubble’s resolution. Courtesy STScI. Below Relative right: Blue stragglers (circled) deep in the core of 47 Tucanae (NGC 104), also imaged by Hubble. These stars may be products of stellar mergers that oc- curred during an earlier stage of this massive, highly concentrated cluster’s dy- 110100 110100 namical evolution. Courtesy AURA/STScI. Radius (arcseconds)

©1998 Sky Publishing Corp. All rights reserved. Sky & Telescope October 1998 43