Quasars in the very early universe

Quasars are powered by the gravitational (potential) energy of their central supermassive black holes. However, their distinctive features – their extremely high luminosity in particular – are very dependent on characteristics of matter close to the black hole.

Most supermassive black holes (SMBH), including those at the centers of the Milky Way and our close neighbor M31 (Andromeda), are responsible for fairly small amounts of radiation in any part of the electromagnetic spectrum. This is generally because the radiation of a quasar is produced mainly by the infall of matter during a relatively brief period of the object's life – a few percent of the total, i. e. a few hundred million years. Once the nearby matter is used up, the lights go out.

(For earlier articles on quasars, see here.)

In a quasar, where matter in significant quantities is still being accreted, most of the radiation originates in a central "accretion disk" close to the black hole. The radiation is thermal ("black body") produced by very hot gas consisting mostly of hydrogen and helium. This radiation covers the spectrum from infrared to X-rays. Since quasars are so bright, they can be seen individually at high redshifts – z≥6, which is not true of ordinary galaxies. That corresponds to times within a billion years of the big bang. At z≈6 photon wavelengths are stretched by a factor of 7, so what we actually see is not the rest-frame spectrum, but a considerably red-shifted version of it.

If r_s is the Schwarzschild radius, then the accretion disk extends from a radius of about 3 times r_s outward to a few hundred times r_s. To give a sense of the scale, a largish SMBH has a mass of a billion solar masses. So for such an object r_s is about the radius of the orbit of Uranus, by a simple calculation given here.

Quasars and active galaxies (i. e. just smaller versions of the same thing) have been intensively studied for several decades. In that time, a fairly clear picture has emerged of how matter is distributed further out from the accretion disk. The most prominent feature of this region is a thick (compared to the accretion disk) torus-shaped ring of cooler gas and dust. The "dust" consists of very small particles composed of various elements heavier than helium. Electromagnetic radiation from the torus is mostly in the infrared part of the spectrum (rest frame), and is produced by re-emission (at lower energies) of higher energy photons from the accretion disk.

The shape of this region is not strictly a torus, since it's considerably flattened, especially at higher distances from the center, but it's referred to as a torus for simplicity. The outer limits of quasar tori are hard to determine, but probably extend hundreds of light years from the center. However, the inner parts are thick enough that unless we are seeing the quasar almost face-on (i. e., along the symmetry axis of the accretion disk and torus), we cannot clearly see the accretion disk itself, because it's obscured by the dust in the torus.

One interesting thing about quasars is that as far as we can tell (until quite recently), their characteristics are very similar no matter how distant they are. Although the present-day universe is quite different in many respects from what it was a billion years after the big bang, quasars seem hardly different at all.

Research published just this year is starting to change this story:

Dust-free quasars in the early Universe

The most distant quasars known, at redshifts z ≈ 6, generally have properties indistinguishable from those of lower-redshift quasars in the rest-frame ultraviolet/optical and X-ray bands. This puzzling result suggests that these distant quasars are evolved objects even though the Universe was only seven per cent of its current age at these redshifts. Recently one z ≈ 6 quasar was shown not to have any detectable emission from hot dust, but it was unclear whether that indicated different hot-dust properties at high redshift or if it is simply an outlier. Here we report the discovery of a second quasar without hot-dust emission in a sample of 21 z ≈ 6 quasars. Such apparently hot-dust-free quasars have no counterparts at low redshift. Moreover, we demonstrate that the hot-dust abundance in the 21 quasars builds up in tandem with the growth of the central black hole, whereas at low redshift it is almost independent of the black hole mass. Thus z ≈ 6 quasars are indeed at an early evolutionary stage, with rapid mass accretion and dust formation. The two hot-dust-free quasars are likely to be first-generation quasars born in dust-free environments and are too young to have formed a detectable amount of hot dust around them.

Some things in these results are actually more interesting than that a few very early quasars are different from all other quasars. In the very early universe at z>6, less intergalactic dust is to be expected. This is because the dust – which is composed of elements heavier than helium – is (by conventional accounts) produced mostly in stars. It is expelled from stars only either gradually, as the star evolves, or suddenly in the rare case of supernova explosions.

We still don't know very precisely when the very first stars formed (see here), but that probably happened only a few hundred million years after the big bang. Since the very first stars must have consisted almost entirely of hydrogen and helium, star-formation models indicate they should have been much more massive than typical later stars, and they should have expoded as supernovae after only a few tens of millions of years. As heavier elements gradually accumulated in the universe, stars of a more modern sort, initially containing small amounts of heavier elements, began to form. These later stars were small enough so that most never ended as supernovae, but they continued to manufacture and expel heavier elements – and hence dust.

Observations of very early quasars therefore tell us a little about the pace of this process. We now know, for example, of at least two quasars that formed so early that they do not have any dust around them that we can observe. The evidence for this is that emissions of these two quasars at rest-frame wavelengths from ultraviolet to very near infrared appear normal. Dust, however, should also produce rest-frame emissions in farther infrared – and that's not seen in these two examples, although it is in all other of the sampled z≈6 quasars.

One other feature of z≈6 quasars is particularly interesting. The mass of a quasar can be estimated from total luminosity and certain spectral features. In all z≈6 quasars that do have evidence of dust, the amount of dust is roughly proportional to the SMBH mass. However, in low-redshift quasars, dust abundance is almost uncorrelated with mass. The implication, then, is that in their youngest stages, but not later, quasars accumulate dust at about the same rate as SMBH mass. And indeed, the two quasars without apparent dust also have the smallest SMBH mass of any in the sample, about 2 to 3×10⁸ M_⊙.

This raises another interesting, unanswered question. Just where does this dust come from? It's generally thought that in the present universe most of the existing dust originated from ordinary stars. However, at z≈6 most stars would be less than 500 million years old, and it's not at all clear this would have allowed enough time for the production of sufficient dust. Supernovae are another possibility, but even in the early universe we don't know whether they would have been common enough.

Further, the existence of elements heavier than helium is necessary but not sufficient to produce dust. Most models assume also a proper combination of relatively low temperature (<2000 K) and high density is also required for dust to form. But a theoretical study (astro-ph/0202002) suggests that dust could actually form in the vicinity of a quasar itself. Some combination of all these possibilities may be the answer, but a lot more research will probably be needed to clear this up.

Every time we learn something new, it seems, we also find new questions.

Jiang, L., Fan, X., Brandt, W., Carilli, C., Egami, E., Hines, D., Kurk, J., Richards, G., Shen, Y., Strauss, M., Vestergaard, M., & Walter, F. (2010). Dust-free quasars in the early Universe Nature, 464 (7287), 380-383 DOI: 10.1038/nature08877

Further reading:

NASA's Spitzer Unearths Primitive Black Holes (3/17/10)

Quasar Dust in the Early Universe (3/26/10)

Primordial Black Holes Formed Just After Big Bang (3/17/10)

First generation of quasars (3/17/10)


Related articles:

Where the action is in black hole jets (5/12/10)

Active galaxies and supermassive black hole jets (4/25/10)

Galaxies are slowly running out of gas

Galaxies are made of stars, and stars are made of... gas. So a large part of understanding how galaxies evolve and grow is understanding how much "gas" (literally, not "gasoline") is present in galaxies – but has not yet been incorporated in stars – at different periods in the history of the universe.

What periods of the universe are most interesting in this regard? The answer is: periods somewhat less than the first half of the universe's existence since the time of the big bang, roughly the first 5.5 billion years, 40% of the total. That's because astronomers have good reason to believe that is the time when star formation, and hence galaxy growth, occurred most vigorously.

Assuming the best current estimate, that is has been about 13.7 billion years since the big bang, this means we're interested in observing the universe as it was more than 8.2 billion years ago. That's quite a long time ago, and until fairly recently observation of objects that far back in time has been infeasible. Technology is only now becoming available to study the details of such a remote time.

Astronomers find it convenient to represent distance (in either space or time) in terms of redshift. Because it takes light a finite amount of time to travel, any observable object is seen not as it looks "today", 13.7 billion years after the big bang, but instead as it looked a some time T<13.7, and so we see the object as it looked 13.7-T billion years ago. The light from such an object has taken 13.7-T billion years to reach us.

Due to the expansion of the universe, the wavelength of any photon of light has been increased by a factor of (z+1), where z is the observed redshift – z=0 corresponding to nearby objects for which the shift is negligible. z increases as a complicated function of the distance of the object, but it increases in a regular way as the distance increases. For objects at an age T=5.5 billion years, corresponding to a distance of 8.2 billion light years, the redshift would be about 1.1.

Astronomers have now done surveys of galaxies around z≈1.1. It's not easy, but there is plenty of data, even though only very large, bright galaxies can be observed in detail at that distance. It's even more difficult, though still feasible, to survey galaxies that are even more remote, say at z≈2.3, which corresponds to T≈2.9 billion years.

For the research under discussion here, the investigators relied on existing surveys to sample from, because of the difficulty of doing new surveys from scratch. There was a trade-off to be made. In order to be able to study a selected sample of galaxies in sufficient detail, it's desirable to pick the largest, brightest galaxies. On the other hand, it's also important to study galaxies that are representative of "typical" mature galaxies today, such as our Milky Way. Unfortunately, the most luminous objects at large z tend to be atypical things like quasars and merging galaxies. Those are "freaks", quite unlike typical nearby galaxies, and whatever we might learn about them might not tell us much about the typical case.

So the investigators had to select galaxies for study that were as large and bright as possible, but still "normal". In this case, they included only galaxies of estimated stellar mass (excluding dark matter) of ≥ 3×10¹⁰ M_⊙. (1 M_⊙ is our Sun's mass.) Since we are inside the Milky Way and can't see all of it (because of thick dusty regions), it's hard to be sure of our galaxy's total stellar mass, but it's estimated to be about 5×10¹⁰ M_⊙. (Ref: here.)

An important objective of the research was to get a better understanding of galaxies in which new stars are actively being formed – unlike the Milky Way and other nearby large spirals, which are currently forming stars at the rate of about 5 M_⊙ per year. Star formation rate is something else that's easier to determine from outside the galaxy, by measuring light flux in various parts of the spectrum (especially infrared and ultraviolet). For the present research, only galaxies with a star formation rate ≥ 40 M_⊙ per year were selected.

For reasons we're coming to, the required observations are difficult and time-consuming, so for this kind of preliminary study it was necessary to work with small numbers. The net result is that the study was done with 11 galaxies selected from one survey, with z≈1.2, and 12 galaxies selected from another survey with z≈2.3.

Remember that the ultimate objective of the research is to determine how much gas is available for star formation in typical galaxies at the given values of z. That's what is so difficult that it had not been done before (for such distant galaxies).

It's relatively straightforward to determine how much of a galaxy's mass is in the form of stars. This "stellar mass" is proportional to the intrinsic luminosity of the galaxy (which is known since the galaxy distance is known), because most galaxies consist of stars with a predictable distribution of stars of given mass and luminosity ("initial mass function").

However, the total mass of a galaxy also includes non-baryonic dark matter, whose mass in the universe as a whole is known to be about 5 times as large as the mass of "ordinary" baryonic matter. The total mass of a galaxy can sometimes be inferred from measuring galaxy rotation curves. The baryonic matter of a galaxy consists of stars, gas, and (perhaps) massive nonluminous objects such as black holes. Even if one knew reliably the total mass of a galaxy, including dark matter, and one could neglect the contribution of nonluminous objects, one still could not estimate the mass of gas as the difference between the mass of a galaxy's stars and the roughly 17% of total mass that baryonic matter represents in the universe as a whole.

That's because there's no a priori reason to expect that a 1:5 ratio of baryonic matter to non-baryonic matter is present in any particular galaxy – it might well be either more or less. So there needs to be some way to measure fairly directly the amount of matter a galaxy contains in the form of gas that could form stars.

It is known that stars form only out of gas that's rather cold – with temperature less than 100 K. This is simply because hotter gas has a higher internal pressure that prevents the gravitational collapse that's necessary to form a star. Gas that cold is very hard to detect. Black body radiation at 100 K peaks at 29 μm, in the far infrared, and cooler gas emits at even longer wavelengths. Redshift stretches the wavelengths even more (by factors of 2 or 3 for z=1 or 2). Most of the radiation at such wavelengths is blocked by our atmosphere, so is observable only from space – and the necessary instruments don't exist yet.

Fortunately, black body radiation is not the only type of electromagnetic emission from cold gas. Vibrations and rotations of gas molecules also radiate at certain frequencies. Now, most of a galaxy's cold gas is in the form of atomic helium and molecular hydrogen. It would be convenient if hydrogen molecules, especially, had emissions at convenient wavelengths for ground-based observation, but no such luck. It turns out, however, that there is one molecule present in small amounts in interstellar gas which does have a convenient emission: CO (carbon monoxide). CO has a rotational emission at 870 μm (346 GHz). At the values of z of interest here (1.2 and 2.3) these fall into the 2 mm and 3 mm bands – which can be observed.

In a nutshell, then, what the research under discussion did was to measure the total flux from the selected galaxies at the appropriate wavelengths. This indicates that amount of cold CO gas present in the galaxies. Studies of nearby galaxies show that this accurately indicates the total amount of cold gas present. From this, and the estimated mass in the form of stars, one has fraction of total mass (stars + gas) represented by the cold gas available to form stars.

For the galaxies in the sample, this fraction was found to be 34% at z≈1.2 and 44% at z≈2.3. By contrast, contemporary large spiral galaxies have fractions in the 3% to 12% range – quite a difference.

For a summary of the results, here's the abstract:

High molecular gas fractions in normal massive star-forming galaxies in the young Universe

Stars form from cold molecular interstellar gas. As this is relatively rare in the local Universe, galaxies like the Milky Way form only a few new stars per year. Typical massive galaxies in the distant Universe formed stars an order of magnitude more rapidly. Unless star formation was significantly more efficient, this difference suggests that young galaxies were much more molecular-gas rich. Molecular gas observations in the distant Universe have so far largely been restricted to very luminous, rare objects, including mergers and quasars, and accordingly we do not yet have a clear idea about the gas content of more normal (albeit massive) galaxies. Here we report the results of a survey of molecular gas in samples of typical massive-star-forming galaxies at mean redshifts of about 1.2 and 2.3, when the Universe was respectively 40% and 24% of its current age. Our measurements reveal that distant star forming galaxies were indeed gas rich, and that the star formation efficiency is not strongly dependent on cosmic epoch. The average fraction of cold gas relative to total galaxy baryonic mass at z = 2.3 and z = 1.2 is respectively about 44% and 34%, three to ten times higher than in today’s massive spiral galaxies. The slow decrease between z ≈ 2 and z ≈ 1 probably requires a mechanism of semi-continuous replenishment of fresh gas to the young galaxies.

The results from this research about star formation rates (SFR) are especially interesting. From other research involving much larger samples, it's known that when SFR is plotted against galaxy stellar mass, the distribution can be fit by a power law:

SFR (M_⊙/year) = 150 (M_*/10¹¹M_⊙)^0.8×([1+z]/3.2)^2.7

In this equation, M_* is the galactic stellar mass. Thus SFR depends on total stellar mass of a galaxy, which makes sense, because the larger the galaxy, the more cold molecular gas is available to make stars. Further, because of the factor involving 1+z (where z is redshift), the SFR curve is shifted upwards at larger z – the rate of star formation is greater, in a regular way, at earlier times in the universe.

This equation is pretty close even in the nearby universe, where z=0. For a galaxy the size of the Milky Way (which is not among the largest of spirals), M_* is estimated as 5×10¹⁰ M_⊙, predicting SFR of about 3.7 M_⊙/year – which is surprisingly accurate. So SFR has continued to decline in a fairly regular way.

Interestingly enough, however, in the galaxies sampled in the present research, the percentage of cold gas in a galaxy does not appear to have any clear relationship to either the SFR or the total stellar mass of a galaxy. So almost all of the variation in SFR is related to the total stellar mass. This is what it means to say that the "efficiency" of star formation is not very dependent on percentage of cold gas or cosmic epoch. Instead, SFR is probably largely dependent on total available cold gas, which is proportional (at a given z) to a galaxy's stellar mass.

One additional interesting conclusion can be drawn from the research. Namely, given the SFR in sampled galaxies at z≈2.3, there ought to be much less cold gas in equivalent galaxies at the later time (about 2.3 billion years later) corresponding to z≈1.2 than is actually observed. Much of that cold gas should have been incorporated into stars. Yet the amount of cold gas actually observed at the later time is more than the original amount less what was converted to stars. And so there is apparently more cold gas added over time, even though, as a whole, galaxies really are "runnng out of gas".

Tacconi, L., Genzel, R., Neri, R., Cox, P., Cooper, M., Shapiro, K., Bolatto, A., Bouché, N., Bournaud, F., Burkert, A., Combes, F., Comerford, J., Davis, M., Schreiber, N., Garcia-Burillo, S., Gracia-Carpio, J., Lutz, D., Naab, T., Omont, A., Shapley, A., Sternberg, A., & Weiner, B. (2010). High molecular gas fractions in normal massive star-forming galaxies in the young Universe Nature, 463 (7282), 781-784 DOI: 10.1038/nature08773

Further reading:

Young galaxies gorge on gas (2/10/10)

Why Today's Galaxies Don't Make As Many Stars As They Once Did (2/11/10)

Early Galaxies Formed Stars Fast Because They Had More Gas (2/10/10)

Stellar Baby Boom of Early Universe Explained (2/11/10)

Ancient Galaxies Packed More Raw Material for Stellar Formation (2/10/10)

In the News this month: the molecular content of early galaxies (3/4/10)

Astrophysics: Less greedy galaxies gulp gas (2/11/10)

Far out!

If you're interested in something out of the ordinary, astronomically speaking, the best place to look for the exotic may be as far away (in both space and time) as possible.

Perhaps that's why I like to consider really far out stuff, like the most distant gamma-ray burst seen yet. Or maybe I just like to get away from the depressing chaos and confusion of "modern" life.

In any case, there's always something new, just beyond the farthest thing we've seen yet. That far-out gamma-ray burst (GRB 090423) discussed in the post just linked – which resides at z~8.2, about 13 billion light-years away – has already been superseded in remoteness by 3 galaxies around z~10, 13.2 billion light-years away. z~10 objects are seen as they were only about 480 million years after the big bang. (See here for a refresher on how redshift works.)

How are high-z objects actually detected? It's surprisingly easy, in principle, even though astronomers are working at the outer limits of their instruments. The term of art for the technique used is "Lyman-break" detection, which we've discussed in detail here.

Here's the short summary of that technique. We assume that the objects of interest are galaxies composed of stars, instead of something really strange. (GRBs can be ruled out, since they appearance is very brief, and gamma-rays are far outside the range of emissions produced by stars, even given substantial redshifts.) The thing is that even the hottest stars produce relatively little output (such as x-rays) on the high-energy side of the part of the ultraviolet spectrum known as the "Lyman limit", which has a wavelength of 91.1 nm. And almost all higher energy photons will be absorbed by the interstellar medium anyhow.

At z=10, the shift factor (z+1) is 11, yielding a wavelength of 1002 nm – i. e. about 1 micron, which is in the infrared. So an object that's really at z~10 will not have any observable light to the blue side (shorter wavelength) of 1002 nm.

The main instrument used in the recently refurbished Hubble telescope to detect the high-redshift galaxies is known as the Wide Field Camera 3 (WFC3). It was installed in May 2009, and in August it did the infrared imaging on which the research described here is based. As noted, the research team identified 3 objects at z~10, using WFC3. Two of those objects were also captured in an image of the same area, using Hubble's Near Infrared Camera and Multi-Object Spectrometer (NICMOS).

WFC3 has three filters covering near-infrared wavelengths. The light from a galaxy at z~10 will be visible through the filter that passes only longer infrared photons, but not through the other two filters that pass only shorter (bluer) photons. So z~10 galaxies will stand out on account of their absence from images made using the shorter-wave filters.

The neat thing about this technique is that it does not require collecting a somewhat complete spectrum, which is a much more difficult feat. In principle, all sufficiently luminous z~10 galaxies should be caught. The problem is false positives – things that aren't really z~10 galaxies, such as dim reddish galaxies that are actually much closer (z<3), or even dim nearby stars.

The research that reports on this has used careful statistical tests to rule out false positives. The analysis in fact suggests that a slightly earlier study that claimed to find 20 z~10 galaxies could be all false positives.

This is the study we're discussing now:

Constraints on the First Galaxies: z~10 Galaxy Candidates from HST WFC3/IR

The first galaxies likely formed a few hundred million years after the Big Bang. Until recently, it has not been possible to detect galaxies earlier than ~750 million years after the Big Bang. The new HST WFC3/IR camera changed this when the deepest-ever, near-IR image of the universe was obtained with the HUDF09 program. Here we use this image to identify three redshift z~10 galaxy candidates in the heart of the reionization epoch when the universe was just 500 million years old. These would be the highest redshift galaxies yet detected, higher than the recent detection of a GRB at z~8.2. The HUDF09 data previously revealed galaxies at z~7 and z~8. Galaxy stellar population models predict substantial star formation at z>9-10. Verification by direct observation of the existence of galaxies at z~10 is the next step. ... Our z~10 sample suggests that the luminosity function and star formation rate density evolution found at lower redshifts continues to z~10, and pushes back the timescale for early galaxy buildup to z>10, increasing the likely role of galaxies in providing the UV flux needed to reionize the universe.

There was a more serious purpose behind this effort than simply bagging a few more objects at record-breaking distances. The ultimate goal is to understand the sequence of events when the earliest stars and galaxies formed in the early universe, and what characteristics those first stars and galaxies had.

As we discussed here and here, at the time of "recombination", about 380,000 years after the big bang, most of the hydrogen and helium gas in the universe was in the form of neutral (un-ionized) atoms. Much later (relatively speaking), in what is known as the "reionization" period, this gas became ionized again by the intense light from young, very hot stars and galaxies. This probably happened over a period of several hundred million years and was essentially complete by 900 million years after the big bang.

What is not so clear is when the reionization (and hence the first stars) began, or how rapidly it proceeded. The best way to understand this process is to determine the numbers of galaxies per unit volume of space throughout this period, and the intrinsic brightness of these galaxies.

Astronomers have gradually been able to make reliable counts of the number of bright galaxies for z<7, corresponding to ~780 million years after the big bang and later. Of course, many galaxies for even much smaller z are too dim to be visible, but it is possible to count the number of galaxies, having at least a minimum intrinsic brightness, per unit volume of space. There is a relatively simple function that fits the data and describes the number of galaxies we can see in a given patch of sky at different redshift values.

We can see more galaxies at smaller values of z not only because the necessary intrinsic brightness decreases with z, but also because there are actually increasing numbers of galaxies having any given intrinsic brightness as the universe grows older – up to a point. This is simply because galaxies themselves became larger and brighter over time as new stars formed.

Another way to view the accumulated data is in terms of the rate of star formation. The more rapidly stars are forming, the more rapidly galaxies grow to reach any particular intrinsic brightness. When read in this way, the data show that the number of stars being formed per year increased monotonically from as far back as we can tell up to a peak with z between 2 and 3, and that the rate of star formation drops after z~2, 3.3 billion years after the big bang.

However, the actual data, so far, become very sparse for z>6. If stars were forming at z~10 at the same rate as they do at z~6, astronomers should find 20±5 galaxies at z~10. Or, using z~7 as a baseline, there should be 9±3 galaxies at z~10. Instead, there were only 3. (Keep in mind this is all for a patch of sky of the same size.)

What this means is that for z>7, stars were forming at even slower rates. And in fact, if one extrapolates the function for the number of observable galaxies back to z~10, 2 or 3 fits very nicely. Further, if that is the case, then the energy emitted by galaxies at z~10 is only about 13% of what would be needed to completely ionize the interstellar gas. Which means, finally, that the largest part of reionization occurred more than 500 million years after the big bang.

Astronomers would, however, like to know much more than that. Ideally one would like more precisely the rates of star formation during the whole time period beginning with the first stars, perhaps 200 or 300 million years after the big bang. As well as other things, such as the distribution of shapes and sizes of galaxies in that period.

Such information isn't important just for its own sake, either. It will also help astronomers answer other questions, such as: What were the characteristics of the very first stars and galaxies, and how did they form? How was dark matter distributed at that time, and how did that affect the formation of stars and galaxies? What role did black holes play in the formation and evolution of galaxies?

As impressive as the recent upgrades to Hubble's instruments have been, they are still at the limits of their capability. So astronomers are eagerly awaiting the James Webb Space Telescope – a much bigger piece of equipment (with a 6.5 m diameter mirror) than Hubble – now scheduled for launch in 2014.

R. J. Bouwens, G. D. Illingworth, I. Labbe, P. A. Oesch, M. Carollo, M. Trenti, P. G. van Dokkum, M. Franx, M. Stiavelli, V. Gonzalez, & D. Magee (2009). Constraints on the First Galaxies: z~10 Galaxy Candidates from HST WFC3/IR Nature arXiv: 0912.4263v2

Update: The published paper may be found here: A candidate redshift z ≈ 10 galaxy and rapid changes in that population at an age of 500 Myr – arXiv.org

Further reading:

New-found Galaxies May Be Farthest Back In Time And Space Yet (1/3/09)

Hubble Spots Oldest Galaxies Yet (1/5/10)

Earliest Known Galaxies Spied in Deep Hubble Picture (1/5/10)

Hubble Reaches the 'Undiscovered Country' of Primeval Galaxies (1/5/10)

Hubble Ultra Deep Field 2009 detects earliest galaxies (1/6/10)

Oldest Galaxies Show Stars Came Together in a Hurry (1/15/10)

Most Distant Known Object In The Universe

On April 23 of this year the Swift Gamma-Ray Burst Telescope detected, just as it was designed to do, a gamma-ray burst (GRB). Within less than a day two of the most powerful Earth-based telescopes had begun studying the quickly fading light of the object, known as GRB 090423. (There were other observatories investigating it as well.)

Because the light from GRBs fades so rapidly, most such objects are detected from space-based instruments that are especially designed for the purpose, like Swift. On average about 1 GRB is detected every 2 or 3 days. But GRB 090423 turned out to be, perhaps, the most interesting one yet observed.

I've written about GRBs a couple of times before, such as here and here.

You can refer to those articles for more details, but the most widely accepted hypothesis concerning the nature of GRBs is that they result from either the core collapse supernova explosion of a massive star or the merger of neutron stars in a binary system. Both processes most likely occur. In the former case, the emission of light lasts somewhat longer than in the latter, so the corresponding types of GRBs are called either "long" or "short".

The long type lasts for more than 2 seconds and tends to be brighter than the short type (whose emissions last less than 2 seconds). Since short GRBs typically appear in locations where new star formation is not occurring, they must result from some cause other than the explosion of a very massive star, which is necessarily quite young. In both cases, there is a prolonged, but much dimmer "afterglow", which is the only part of the event that can be observed except from satellites such as Swift.

Not just any supernova or neutron star collision will result in a detectable GRB. Most of the energy from the event must also be directed in two very narrow and oppositely directed beams, and the Earth must be in the path of one of those. It is because the energy is so narrowly concentrated that a GRB can appear to be far brighter than an entire galaxy or quasar. For such narrowly focused beams to be produced, the progenitor star (or binary system) must have a great deal of angular momentum, due to rapid rotation.

The afterglow is thought to result from the interaction between the matter and energy contained in the beam and the interstellar medium at the location of the GRB. Although it's a secondary effect, there is a wealth of information that can be deduced from this afterglow.

However, the information concerns more than just details about the nature of GRBs – events that occur much closer to us are much more useful for that. There are two other things of far greater interest about which we can learn from very distant GRBs, such as 090423 and subsequent examples we are likely to observe.

Since this event had a redshift of z ≈ 8.2 it actually occurred very long ago, a comparatively short time after the big bang. About 630 million years after, to be more exact. GRB 090423 is farther and earlier than any other object or event we've actually observed, except for the cosmic microwave background (CMB). It is more distant in time and space than even any galaxy or quasar we've ever seen. (See here if you want to review how redshift works.)

GRBs, directly or indirectly, tell us about the nature of the largest stars at that time, which is one thing that's presently very difficult to study with other observations. There's simply no way to observe individual stars 13 billion light-years away. The other thing astronomers are very curious about regarding that time period is the nature of the intergalactic medium (IGM) then.

Let's start with the second of these things. 630 million years after the big bang is somewhere in the middle of what astronomers call the "cosmic dark ages". Precisely where it falls is what we don't know. I wrote about that subject here, last March.

We do know pretty closely when the "dark ages" began – about 380 thousand years after the big bang. That's when the CMB appeared. The time is equivalent to a redshift of z ≈ 1100. The CMB was, and still is, "blackbody" radiation, so wavelengths are distributed around a peak. At the peak of the distribution now, CMB photons have a wavelength of 1.9 mm, in the short microwave part of the spectrum. (That's a lot shorter, so more energetic, than in your microwave oven, where microwaves are about 12 cm.) But when the CMB actually appeared, its photons had a wavelength 1100 times shorter, around 2200 nm – in the infrared part of the spectrum. That's still much cooler and "darker" than human eyes can see.

Effectively, then, there wasn't much light in the universe, certainly not human-visible light, until the first stars showed up. So the time after the CMB appeared but before the first stars is known as the "dark ages". However, we don't know very closely when those first stars appeared – they're far too faint to observe, and even the first galaxies and quasars are too dim to be detected by our present instruments. GRBs seem to be our best observational hope, and GRB 090423 is the earliest we've seen yet.

Since existing GRB models presuppose an origin that involves massive stars, at least indirectly, we now know there were such stars 630 million years after the big bang. There is evidence in the GRB 090423 observations that the progenitor of this object was not one of the first population of stars to form. (Astronomers call such stars "Population III".) Also, GRB 090423 doesn't seem to have been one of the most powerful GRBs ever, so astronomers now figure that with current technology we may be able to find GRBs back to z ≈ 20. That would correspond to 250 million years after the big bang.

Theoretical models suggest that the first stars may have appeared even earlier than that, at perhaps 150 million years after the big bang. When they formed, these first stars did not contain elements heavier than helium, as such elements were first created (in other than trace quantities) only in the earliest generation of stars. (See here for a discussion of the earliest stars.)

The observations of GRB 090423 give some evidence that the galaxy in which it occurred did have small amounts of elements heavier than helium. Further, the afterglow that occurs following the peak brightness of any GRB is generally similar between both GRB 090423 and much closer GRBs. These facts are hints that the progenitor of GRB wasn't one of the Population III stars.

How does a GRB give us information about the intergalactic medium at z ≈ 8.2? That turns out to be closely related to the way that the redshift was estimated in the first place. Normally this is done by identifying known emission or absorption lines in a spectrum and simply calculating the amount of shift directly. However, GRB 090423 is so faint, and declined in brightness so quickly, that it wasn't possible in the time available to obtain a detailed spectrum.

Instead, an important feature of the IGM at that time comes to the rescue. The feature is that there is at that time a substantial, though not precisely known, quantity of un-ionized hydrogen atoms in the IGM. This hydrogen is left over from the period of "recombination" in which the CMB appeared. CMB photons are far too weak to "scatter" from hydrogen atoms, because quantum mechanics requires that a photon must have a certain minimal amount of energy to raise an electron bound in a hydrogen atom out of its "ground state" of lowest energy.

The minimum amount of energy required is that possessed by a 10.2 eV photon of wavelength 121.6 nm, which is in the ultraviolet part of the spectrum. This spectral point is known as "Lyman-α". (We covered this concept in detail here.) Ordinary stars like the Sun in the universe at present emit little light at this wavelength, but much larger, hotter, and brighter stars emit quite a bit of this ultraviolet light.

Consequently, as soon as such hot stars began to shine, the atomic hydrogen in the IGM began to reionize. There was a price to be paid for this, of course: all the photons with enough energy lost some of their energy in the process of reionizing the hydrogen. And so, most of the light whose photons had enough energy was absorbed or "scattered" in this reionization period.

Eventually most of the hydrogen in the IGM of the early universe would become fully ionized, so that very energetic photons could again pass unimpeded. That stage marks the end of the reionization period. The whole period may have lasted from 150 million years to a 900 million years after the big bang, but we are quite unsure of the actual endpoints.

GRBs may be able to tell us. To begin with, it's easy to derive the redshift of a GRB that flared during the reionization period. If the redshift is z, then in the "rest frame" of the GRB, the light will start to be absorbed strongly at 121.6 nm. But we will observe this drop-off to occur at a wavelength of (z+1)×121.6 nm. For example, if z = 8.2, we see the drop-off at 1120 nm, which is in the infrared. Reasoning backwards, we can compute z from the observed drop off.

If we have enough GRB observations with sufficiently good spectra available, we can infer the amount of reionization that has occurred from the rate of drop-off. Further, if we find GRBs where there is no drop-off, we can infer that reionization is complete. Unfortunately, GRB 090423 tells us little by itself, and only two other GRBs have been observed with z > 6, corresponding to somewhat less than 950 million years after the big bang.

Why do we care about the rate of reionization? Because it tells us indirectly about the rate of formation of very large, hot stars in the early universe. So it turns out that the two things that astronomers want to know about the early universe – the nature of the IGM and the rates at which stars were forming – are pretty closely related. On top of that, the distribution in time of early GRBs gives us an independent estimate of the rate of star formation, since the progenitors of GRBs are exactly the sort of large, hot stars that cause reionization.

Tanvir, N., Fox, D., Levan, A., Berger, E., Wiersema, K., Fynbo, J., Cucchiara, A., Krühler, T., Gehrels, N., Bloom, J., Greiner, J., Evans, P., Rol, E., Olivares, F., Hjorth, J., Jakobsson, P., Farihi, J., Willingale, R., Starling, R., Cenko, S., Perley, D., Maund, J., Duke, J., Wijers, R., Adamson, A., Allan, A., Bremer, M., Burrows, D., Castro-Tirado, A., Cavanagh, B., de Ugarte Postigo, A., Dopita, M., Fatkhullin, T., Fruchter, A., Foley, R., Gorosabel, J., Kennea, J., Kerr, T., Klose, S., Krimm, H., Komarova, V., Kulkarni, S., Moskvitin, A., Mundell, C., Naylor, T., Page, K., Penprase, B., Perri, M., Podsiadlowski, P., Roth, K., Rutledge, R., Sakamoto, T., Schady, P., Schmidt, B., Soderberg, A., Sollerman, J., Stephens, A., Stratta, G., Ukwatta, T., Watson, D., Westra, E., Wold, T., & Wolf, C. (2009). A γ-ray burst at a redshift of z ≈ 8.2 Nature, 461 (7268), 1254-1257 DOI: 10.1038/nature08459

Salvaterra, R., Valle, M., Campana, S., Chincarini, G., Covino, S., D’Avanzo, P., Fernández-Soto, A., Guidorzi, C., Mannucci, F., Margutti, R., Thöne, C., Antonelli, L., Barthelmy, S., De Pasquale, M., D’Elia, V., Fiore, F., Fugazza, D., Hunt, L., Maiorano, E., Marinoni, S., Marshall, F., Molinari, E., Nousek, J., Pian, E., Racusin, J., Stella, L., Amati, L., Andreuzzi, G., Cusumano, G., Fenimore, E., Ferrero, P., Giommi, P., Guetta, D., Holland, S., Hurley, K., Israel, G., Mao, J., Markwardt, C., Masetti, N., Pagani, C., Palazzi, E., Palmer, D., Piranomonte, S., Tagliaferri, G., & Testa, V. (2009). GRB 090423 at a redshift of z ≈ 8.1 Nature, 461 (7268), 1258-1260 DOI: 10.1038/nature08445

Other blog posts

Beyond the Farthest Star (12/14/09)

Reports on the newly-published research on GRB 090423

Most distant gamma-ray burst spotted (10/28/09)

A Blast From the Deep, Dark Past (10/28/09)

Astronomers explore 'last blank space' on map of the Universe (10/28/09)

Blast from the Past Gives Clues About Early Universe (10/28/09)

Blast from the Very Far Past (10/28/09)

Astrophysics: Most distant cosmic blast seen (10/29/09)

A big gamma-ray burst at a redshift of z ≈ 8.2 (10/29/09)

GRB 090423 at a redshift of z ≈ 8.1 (10/29/09)

Discovery of Radio Afterglow from the Most Distant Cosmic Explosion


Reports on the initial observation of GRB 090423

Scrambling to Read the Meaning Of the Sky's Most Ancient Flare (9/18/09)

Earliest astrophysical object yet seen (7/2/09)

Most Distant Known Object In The Universe / Science News (4/28/09)

The Farthest Thing Ever Seen (4/28/09)

Most distant object in the universe spotted (4/27/09)

The Most Distant Object Yet Discovered in the Universe (4/28/09)

Farthest Known Object: New Gamma-Ray Burst Smashes Cosmic Distance Record (4/28/09)

New Gamma-Ray Burst Smashes Cosmic Distance Record (4/28/09)

Ancient gamma-ray burst is most distant object ever seen (4/29/09)

Space Explosion Is Farthest Thing Ever Seen (4/28/09)

Telescope snaps most distant object (4/28/09)

Exploding star is oldest object seen in universe (4/29/09)

Cosmic blast sets distance mark (4/28/09)

Astronomers see oldest object in universe yet (4/28/09)

A glimpse of the end of the dark ages: the gamma-ray burst of 23 April 2009 at redshift 8.3

Tags: gamma-ray bursts, early universe

A surprisingly compact early galaxy

Astronomers are beginning to learn significant details of the structure of galaxies in the early universe. And what they're learning is rather surprising: at least some early galaxies are almost as massive as otherwise similar galaxies in the present universe, yet they are much smaller in linear size, by a factor of five, thus much more compact.

What time period are we talking about here? It's not actually the time that the earliest galaxies formed, which was less than a billion years after the big bang. Instead, the time in question was around 3 billion years after the big bang.

Although that's roughly 10.7 billion years ago, many galaxies at that time were actually fairly mature, even old. This is because they had been around for more than 2 billion years, which is more than time enough for all their massive, hot, bright stars to have burned out long before. If these galaxies had depleted most of their star-forming material, not many new, hot, young stars could form. The rate of star formation might be as small as it is now in the Milky Way, only two to four solar masses worth per year, compared to thousands per year at the peak.

Young stars include proportionately more massive stars, because it's the massive ones that burn out quickly. Stars that are more than 2 billion years old have to be smaller. Smaller stars are also dimmer, cooler, and redder in color. (Recall the Hertzsprung–Russell diagram, which displays the relationship between luminosity and color.) Consequently, older galaxies that are no longer forming many new stars are also redder, and that's how astronomers estimate roughly galaxy age, or at least the length of time since rapid star formation ceased.

Certain events, such as collisions and mergers between galaxies can fire up rapid star formation again. So the correlation between color and age is not at all exact, but it's still there.

Another fact about galactic appearance is that the central part of any galaxy, even a spiral, is much brighter than the outer reaches, like the spiral arms (if any), simply because the central part of a galaxy contains most of the stars. So at the distances we're concerned with here – over 10 billion light-years – all we can really observe with existing optical telescopes is the central part of a galaxy. To a first approximation, then, very distant galaxies look ellipsoidal in shape, even if they're really spirals.

For the time period we're interested in, 3 billion years after the big bang, the redshift of light we see from objects at that time (denoted by z) is about 2.2. (See here for a fuller explanation.) The definition of redshift means that the wavelength of light emitted at z~2.2 is stretched by a factor of z+1~3.2. So visible light, with a wavelength of about 400 to 700 nanometers is shifted to 1.3 to 2.2 microns, in the near infrared. Although this kind of infrared light can be studied by ground-based spectroscopy, the best optical imagery has to be done from space-based instruments, which makes the job a lot harder.

With all that as background, the newly published result we're concerned with is really pretty simple. It has confirmed that a certain galaxy, named 1255-0, at z=2.186 is about a third as massive (~2×10¹¹M_⊙) as the Milky Way and similar galaxies in our neighborhood at the present time, even though its central region is much smaller, with a radius of about 2500 light years. Consequently, stars in the central region of 1255-0 are packed much more closely together.

This observation raises two distinct problems: First, most existing models of galaxy formation do not predict that typical galaxies of that age will be so compact. Second, no galaxies of that sort seem to exist in our general neighborhood, so at least some of them presumably evolved from galaxies like 1255-0 – and it's not clear how that could happen.

Here's the research abstract:

A high stellar velocity dispersion for a compact massive galaxy at redshift z = 2.186

Recent studies have found that the oldest and most luminous galaxies in the early Universe are surprisingly compact, having stellar masses similar to present-day elliptical galaxies but much smaller sizes. This finding has attracted considerable attention, as it suggests that massive galaxies have grown in size by a factor of about five over the past ten billion years (10 Gyr). A key test of these results is a determination of the stellar kinematics of one of the compact galaxies: if the sizes of these objects are as extreme as has been claimed, their stars are expected to have much higher velocities than those in present-day galaxies of the same mass. Here we report a measurement of the stellar velocity dispersion of a massive compact galaxy at redshift z = 2.186, corresponding to a look-back time of 10.7 Gyr.

Note that the research isn't the first to identify very massive but compact galaxies at z~2. Rather, it's new in that it has confirmed the estimate of mass by a new method, and that's what's significant.

You see, there are basically two different ways, at present, to estimate the mass of a very distant galaxy. One method relies on a plausible assumption, that stars less than 2 billion years old, except for the very youngest, have fairly well-known distributions of mass and luminosity. And so, from the total luminosity of the galaxy that we can observe, we can form a good estimate of the total mass of stars. This is sometimes called the "photometric" mass.

This sort of measurement is what has been used to infer that a number of galaxies at z~2 may have been very massive in spite of being small in extent. As suggested above, this observation raises at least two problems, so astronomers would like to measure mass in a different way, just to be sure. Besides, perhaps the assumptions about the mass and luminosity distributions of the stars in such galaxies could be wrong.

Fortunately, there is another type of observation that can lead to good mass estimates, but it is much more difficult to make. This involves measuring the "dispersion" of velocities of stars in the galaxy. That is related to the distribution of stellar velocities. But since we can't distinguish individual stars at that distance we certainly can't measure their velocities (by very slight differences in stellar redshifts from the redshift of the galaxy as a whole).

Even though the measurement is difficult to make in practice, it's simple to describe. One simply looks at the width of a few absorption lines in the galaxy's spectrum. If the lines are wide, it means that individual stars have substantially different velocities, including a certain proportion which are quite large. This is, basically, the meaning of "dispersion".

From the relative number stars with high velocities one can infer the total mass. This yields what is called the "dynamical" mass of the galaxy. What the present research found is simply that the dynamical mass of 1255-0 is pretty close to the known photometric mass.

Why is it that lots of high-velocity stars indicates a substantial mass? Just fairly basic physics, based on two of Newton's laws. (If you're a physicist, you learned this a long time ago, so it's "obvious".) The first is Newton's law of gravitation, which is F=G×M×m/r². This describes the gravitational force (F) between two objects having masses M and m separated by a distance r. G is a certain constant called, of course, the gravitational constant. This can be applied to a galaxy with mass M and one of its stars, with mass m, where r is the distance from the star to the center of mass of the galaxy.

The other law is Newton's second law of motion, which says F=ma. F is, again, the gravitational force of the galaxy, m is the mass of a particular star, and a is the acceleration of the star due to the force. (F and a are actually "vector" quantities, of couse, meaning they have a direction in 3-space.) You can think of the acceleration of an object as a way of measuring the force acting on it.

Putting the two laws together, we find that for any particular star its accleration will satisfy a = G×M/r², so the star's mass doesn't matter at all, only the mass of the galaxy. Just as in our own galaxy, almost all stars are in orbit around the center of mass of the galaxy, so a star's velocity, as seen from far away, varies periodically in a predictable way, deducible from acceleration (which is the rate of change of velocity).

From calculations that are routine (at least for a physicist) one thus obtains a good estimate of the mass of a galaxy from the distribution of velocities of its stars, which in turn is deducible from the dispersion of spectral lines.

There is one additional complication: matter in any form other than what makes up stars, most especially dark matter. But the present research shows that the mass as estimated photometrically (where any nonluminous matter plays no part) and the mass as estimated dynamically (where dark matter could be important) are pretty close.

Consequently, there isn't much nonluminous matter (including nonbaryonic dark matter) in the central part of the galaxy where most of the stars are. This is as expected, since most galaxy models as well as observations have the dark matter distributed over a much larger volume than the central part of the galaxy. (Other elementary physics shows that matter outside the orbit of a star does not affect the star's motion, as long as that matter is evenly distributed.)

The net of all this is that the two problems mentioned above are real and pose questions that need to be answered.

How could massive galaxies as compact as 1255-0 have formed in the first place? It is not the case that massive compact galaxies like 1255-0 are exceptional anomalies at z~2. Instead, they seem to make up as much as 30 to 40% of galaxies whose masses have been estimated (photometrically) at that distance.

Existing models involving cold dark matter mostly do not predict such a thing. But this doesn't mean that the models can't be refined. In particular, the whole theory of cold dark matter as a driver of galaxy formation need not be discarded. It isn't necessary to invoke some exotic new physics or variations of Einstein's general relativity. The most natural approach is to find a suitable refinement of the galaxy formation model. There is research that was reported in January 2009 and offers one sort of model. It involves filaments of dark matter that conduct streams of cold gas into a central region around which a galaxy grows. Research paper: here. Additional stories: here, here, here, here.

The other problem has received less attention. The difficulty is in explaining how a galaxy (or rather, its central region) grows by a linear factor of five or so over a period of ~10 billion years, even though the mass contained in that region doesn't grow much at all. It just seems to "puff up".

Galaxies have long been presumed to grow through mergers of less massive galaxies, the most important of which are of roughly equal mass. One possibility is that few such mergers actually occur, and instead colliding galaxies mostly pass through each other without merging, but with some expansion of linear size each time. Another possibility is that there are many mergers involving mostly low-mass galaxies captured by much larger ones. That would also help explain another major puzzle: why many fewer low-mass galaxies are observed than current models predict.

Questions of this sort are not easy to resolve. They're a lot like questions about the evolution of life. All we can actually observe consists of snapshots from different points of time. Events unfold too slowly to actually see what happens. And moreover, the very small galaxies that might play a role are currently too faint to observe over most of the past 12 billion or so years of cosmic history.

van Dokkum, P., Kriek, M., & Franx, M. (2009). A high stellar velocity dispersion for a compact massive galaxy at redshift z = 2.186 Nature, 460 (7256), 717-719 DOI: 10.1038/nature08220

Further reading:

Astronomers Find Hyperactive Galaxies in the Early Universe (8/5/09) – press release

Speeding Stars Confirm Bizarre Nature of Faraway Galaxies (8/5/09) – article at space.com

Galactic evolution: more data, no more answers (8/12/09) – article at arstechnica.com

Galaxy formation: Too small to ignore (8/6/09) – Nature news article

Puffing up elliptical galaxies (10/3/09) – blog post (SarahAskew)

Tags: astrophysics, galaxy evolution

The cosmic "dark ages"

In brief, "dark ages" refers to the period after recombination occurred, about 380,000 years after the big bang, creating the cosmic microwave background (CMB), up to and partly including the time that the first stars had formed, perhaps as early as four hundred million years later, and caused the reionization of much of the neutral hydrogen in the universe.

That's a mouthful, but it's important to understand in order to have a useful discussion of the conditions that existed when the first stars and first galaxies in the universe formed. Many important open questions in astrophysics right now have to do with the nature of these events. Since I expect to discuss some of these questions, it's necessary to say some things about the "dark ages". That's what this note is about.

Let's start with the CMB. I'm going to give just a sketch. You might want to consult other references if you need more detail.

The period of time in which the CMB emerged is also known as the period of recombination. This was not an instantaneous process, but it did proceed relatively quickly, and is often thought of as a single event.

Basically, before recombination matter (mostly hydrogen and helium) and energy (light, i. e. photons) existed in thermal equilibrium. That is, the following reaction could occur with equal probability in either direction:

H + γ ⇄ p + e^-

Here, H stands for a hydrogen atom (with one electron), γ is a photon, p is a proton, and e^- is an electron. (We'll ignore helium, for simplicity.) What this formula says is that a photon of sufficient energy could dislodge an electron from a hydrogen atom to form a proton and a free electron, and with equal probability protons and free electrons could combine to form a hydrogen atom and a photon.

As the universe cooled after the big bang, the average temperature of the matter-energy plasma steadily dropped. Before the period of recombination, hydrogen atoms could exist, but not for very long, because most photons had enough energy to completely dislodge an electron, so the reaction went from left to right as often as from right to left.

Over a period of time lasting a few tens of thousands of years the situation changed so that most photons no longer had enough energy to dislodge an electron. (A low energy photon could still interact or "scatter" with a hydrogen atom by raising an electron to a higher energy level, but we can gloss over that detail.) Note that the average or "typical" photon energy may be a lot lower than what's needed to dislodge an electron, as long as there are still enough higher-energy photons around. This is a statistical situation governed by the Maxwell-Boltzmann equation, but all that really matters is that eventually photons and neutral hydrogen atoms "decouple" statistically.

So at some point you have recombination, when electrons combine with protons to make neutral hydrogen, and the process (mostly) doesn't reverse. Occasionally, a photon and a hydrogen atom may still interact, but as the universe expands, it's increasingly less likely for a photon and an atom to come close enough to interact, until the probability of interaction is essentially zero. This second stage is sometimes referred to as "decoupling" of matter and photons.

Thus the period of "dark ages" began right after the relatively brief process of recombination and decoupling. For simplicity, we date this point to a single time when the process was about half complete, roughly 380,000 years after the big bang.

This period is termed "dark", even though there were plenty of photons around, because there were as yet no stars or other compact sources of illumination. (There's another reasons for calling it "dark", which we'll get to in a moment.)

CMB photons have a nearly perfect black-body distribution. There is a clear peak of maximum energy in this distribution. What we observe is that this peak occurs at about 2 mm, in the microwave part of the spectrum. But the time of decoupling, 380,000 years after the big bang, corresponds to a redshift of z≈1100, so at the time of decoupling the peak photon energy was around a wavelength of just 2.2 μm (2200 nm) in the infrared part of the electromagnetic spectrum. (If you need to refresh your memory about how redshift works, check here.) So "dark" is not exactly the right term to use, but compared to abundant light from stars, it's not unreasonable.

After the time of recombination/decoupling, most hydrogen and helium atoms were neutral and un-ionized. This went on for several hundred million years. One of the most interesting open questions is about determining more exactly how long this lasted. We can reasonably guess what probably brought the dark ages to an end: formation of the first stars in the universe. But we don't have a good idea of just when this started, or how long the process took.

In this period, matter was beginning slowly to come together in higher-density clumps under the force of gravity. Dark matter, which outweighed ordinary matter then (as now), by a ratio of about 5.5:1 speeded up this process.

It was precisely this formation of regions of higher matter density that enabled the first stars to form. But because of this same higher density of matter around newborn stars, the abundant high-energy photons produced by these stars were again likely to interact with the nearby un-ionized matter, which scattered them and reduced their energy – dimming the light of these first stars.

One of the large uncertainties concerns the characteristics of this first generation of stars. We have no direct evidence about them. What we think we know about them is based on theoretical models rather than direct observation. (We discussed formation of the first stars back here.) However, it's widely believed that these stars were unlike stars formed later, right up to the present time. The first stars were probably quite large (maybe as much as 200 solar masses), very hot, and very bright. Because they burned their fuel so rapidly, their lifetimes would be very short, perhaps less than a million years.

Some of the light from extremely hot, massive stars such as those of the first generation is well into the ultraviolet part of the spectrum, around 90 nm. Such photons have an energy above 13.6 eV (91.2 nm wavelength), enough to completely ionize hydrogen. If these stars formed, say, 400 million years after the big bang, at a redshift of z≈11, the wavelength of their light would be shifted to the area beyond 1100 nm, which is in the infrared. That's beyond the range of human eyes, or most astronomical instruments. Consequently, we would have a very hard time detecting light from the first stars, even if they weren't so far away (over 13 billion light-years) and obscured by clouds of atomic hydrogen. That time period would still seem "dark" to us – there would be very little we could "see".

However, the first stars radiated so much energy, especially at ultraviolet wavelengths, that over time they effectively reionized all of the hydrogen in their vicinity. And this is (probably) what brought about the epoch of reionization in the universe, effectively ending the "dark ages".

The first stars were probably not part of galaxies, though we don't know for sure. If that's the case, they would be even harder to observe. The first objects we will be able to detect from this period almost certainly will be galaxies or quasars (which are galaxies with a very active central black hole). So the question of when galaxies began to form is separate, but equally puzzling. There is evidence that the first galaxies in fact did form before the end of the dark ages – because we can actually observe a few that show evidence of un-ionized hydrogen.

What we can say for sure is that the first stars must have consisted only of the primordial elements hydrogen and helium, since heavier elements (except for a small trace of lithium) formed and dispersed only when the first stars exploded as supernovae. But that did happen rather quickly, since the first stars were very luminous, and consequently had very brief lives.

As noted, we aren't very sure about when this first generation of stars appeared, because we can't yet observe them directly. But we do have some evidence concerning the epoch of reionization, and hence we have some idea of when it ended.

One kind of evidence involves studying the spectra of some of the most distant objects we are currently able to observe – quasars. We have detected a number of quasars at redshifts between 6 and 7. This range represents a time period from about 780 to 950 million years after the big bang.

There are absorption lines in the spectra of these quasars, and they tell us not only about the redshift, but give other information as well. Among the most important lines are those due to hydrogen that's not fully ionized, such as lines of the Lyman series. For the most part these lines are due to hydrogen in the vicinity of the source, in which case the lines are quite sharp and distinct.

But suppose there is a substantial amount of incompletely ionized hydrogen between us and the source. If this gas is at a distance sufficiently less (in terms of redshift), the absorption lines will be fuzzy instead of sharp. This effect is called a Gunn-Peterson trough.

In 2001 a quasar was identified at z=6.28, which showed a Gunn-Peterson trough, while other quasars with z≤6 did not. This suggests that reionization was mostly complete by 950 million years after the big bang, but not by 900 million years. [1][2]

Quasars, by themselves, are a possible contributor to reionization, in addition to the earliest stars. Quasars certainly produce enough high-energy photons. However, the question is whether there were enough quasars in existence during the epoch of reionization to account for the effect. Since only the very brightest quasars can currently be observed at that distance (z≥6), it's not possible to reliably estimate how many quasars altogether were around then. Rough estimates suggest there weren't enough.

There is another source of evidence for reionization, one very different from the Gunn-Peterson trough. This involves a very detailed study of anisotropies (irregularities at small angular scales) in the CMB. The CMB has many anisotropies due to conditions existing from the earliest moments after the big bang. However, if reionization occurred, certain kinds of additional characteristic anisotropies will also be present. These result from polarization of CMB light due to Thomson scattering of photons by free electrons (if such exist in sufficient numbers). Since free electrons are a by-product of reionization, they provide a very good marker, if they can be detected.

Unfortunately, the earliest data analysis (in 2003) from the Wilkinson Microwave Anisotropy Probe suggested that reionization occurred in the range 11<z<30, which corresponds to a mere 100 million to 420 million years after the big bang. This is not compatible with the quasar evidence. It's also rather implausible at the high z end, if reionization was caused by the first stars.

Fortunately, however, later data analysis (released in 2008) restated the range for reionization to 7≤z≤11. [3] z=7 still doesn't quite mesh with the quasar data, but it's pretty close. z=11 corresponds to 420 million years after the big bang, which is quite plausible for appearance of the first stars. Note that if this range is correct, then the reionization process took a lot of time, maybe 400 to 500 million years. First generation stars almost certainly weren't around that long. Such stars have very short lifetimes, and new stars of this kind can't form, because the gas from which they could form would contain considerable amounts of elements heavier than helium, precluding the formation of more stars like those of the first generation.

A third possible source of evidence comes from surveys looking for very faint, high-redshift galaxies (not quasars). Some objects have been found up to z=7.5 – about 710 million years after the big bang. [4] What isn't clear is whether these objects were either abundant enough or had hot enough stars to contribute significantly to reionization. But when the James Webb Space Telescope goes to work, sometime after mid-2013, we should be able to find many more early galaxies. The planned upgrade to the Hubble Space Telescope this year would also help – if it occurs.

Further reading:

[1] Evidence for Reionization at z ~ 6: Detection of a Gunn-Peterson Trough in a z = 6.28 Quasar – 2001 research article on first evidence for reionization (open access)

[2] First Light: Astronomers Use Distant Quasar To Probe Cosmic "Dark Age," Universe Origins (8/8/01) – press release describing the preceding research

[3] A New Day in Precision Cosmology (3/11/08) – news article describing analysis of WMAP data, including information on reionization

[4] Largest Sample Of Very Distant Galaxies Ever Seen Provide New Insights Into Early Universe (7/24/08) – press release

Tags: cosmology, cosmic dark ages

Which came first - the galaxy or the black hole?

This news isn't exactly fresh, having been announced several weeks ago, and it's been reported in many places. However, it's quite important (if correct), because it suggests that the supermassive black holes that seem to exist at the centers of most galaxies formed much faster, initially, than the rest of the galaxy.

We've already discussed a little about galaxy formation: here.

The study for present discussion was done using radio telescopes, which could detect spectral sequences due to carbon monoxide gas in order to make estimates of mass in galaxies' central regions. The galaxies studied were all observed as they existed less than two billion years after the big bang (redshift ≳ 3). At that age, all galaxies that can be observed in enough detail with optical telescopes are quasars, in which so much light comes from the hyperactive central black hole that almost nothing can be inferred about the rest of the galaxy. Use of radio telescopes circumvented that problem.

In most sufficiently regular galaxies closer to us, most of the mass of stars and gas is concentrated in a central bulge. In such galaxies, representing the large majority of the universe's age of 13.7 billion years, the ratio of mass in the central bulges to the mass of the central black hole is remarkably predictable and averages about 700:1. But in the four early galaxies studied, the ratio was only about 30:1. Although the mass of the central black holes studied was nearly as large as that of galaxies closer to us, the galaxies themselves appeared to be much less massive.

This anomaly in the ratios does not necessarily imply the black holes formed before the rest of the galaxy, but it does suggest they at least grew more quickly in the early stages. There are not yet good theories for why this might happen, or at what point growth of the rest of the galaxy accelerates compared to growth of the black hole.

The internal workings of quasars are not well understood. It may be that quasars can form only when the ratio of central mass to black hole mass is low, as in the observed examples. Or maybe gas in the neighborhood of a very active quasar is dispersed by the intense radiation, so that it can collapse further only when the activity dies down somewhat. Conversely, perhaps, in cases where the mass ratio is higher, radiation from the black hole has diminished, or never was as strong, so that we have not been able to detect those cases.

Good reasons exist to think that very active black holes should generally disrupt formation of stars in the surrounding gas. But there are also reasons they might at some point assist in star formation. We'll need much more observational evidence, and theoretical calculations, before the picture gets clearer.

There are other problems, too. For instance, the mass of the galactic bulges were estimated from the motion of gas rather than stars (which could not be observed). Magnetic fields could have affected gas flow, resulting in an underestimate of total mass. It's also not clear how supermassive black holes could have formed by themselves, without the gas they presumably grew from also collapsing to form individual stars.

While it's interesting to have a little evidence that the black holes formed first, a claim based on observations of only four early galaxies is less than fully convincing. Many more data points are needed.

And assuming the relationship holds up, even tougher questions remain. How and why did the black holes themselves form? How did they grow so big so quickly? What is the relationship over time between the size of the black hole and the galaxy that forms around it? How do they affect each other's growth?

Black Holes Lead Galaxy Growth (1/6/09)

Earlier studies of galaxies and their central black holes in the nearby Universe revealed an intriguing linkage between the masses of the black holes and of the central "bulges" of stars and gas in the galaxies. The ratio of the black hole and the bulge mass is nearly the same for a wide range of galactic sizes and ages. For central black holes from a few million to many billions of times the mass of our Sun, the black hole's mass is about one one-thousandth of the mass of the surrounding galactic bulge.

"This constant ratio indicates that the black hole and the bulge affect each others' growth in some sort of interactive relationship," said Dominik Riechers, of Caltech. "The big question has been whether one grows before the other or if they grow together, maintaining their mass ratio throughout the entire process." ...

"We finally have been able to measure black-hole and bulge masses in several galaxies seen as they were in the first billion years after the Big Bang, and the evidence suggests that the constant ratio seen nearby may not hold in the early Universe. The black holes in these young galaxies are much more massive compared to the bulges than those seen in the nearby Universe," said Fabian Walter of the Max-Planck Institute for Radioastronomy (MPIfR) in Germany.

"The implication is that the black holes started growing first."

The next challenge is to figure out how the black hole and the bulge affect each others' growth. "We don't know what mechanism is at work here, and why, at some point in the process, the 'standard' ratio between the masses is established," Riechers said.

However, another report quoting one of the researchers points out that only four very early galaxies were included in the radio telescope study:

In the Young Universe, Black Holes May Have Formed First (1/7/09)

“These very distant black holes are already about as massive as they will ever get — about 1 billion solar masses — so the only thing left is for the galaxy to form around them,” says Carilli. One implication, he says, is that the turbulent activity associated with accretion onto these black holes “may have a profound effect on the formation of the host galaxy very early in the universe.” But Carilli emphasizes that his team has examined only four galaxies from these early times. It’s possible, he says, that this handful of galaxies may have unusually heavy supermassive black holes.

“We really need to generalize to more galaxies that are less extreme,” he adds.

Martin Rees lays out the case for skepticism of the results:

“The results are interesting, and an important clue to the growth and evolution of galaxies,” comments Martin Rees of the University of Cambridge in England. However, he adds, “I think it is over-interpreting the data to say that ‘black holes come first.’ Even at [early times] the bulge mass could typically be about 100 times larger than the mass of the hole.”

Rees suggests that the bulges and holes form concurrently throughout cosmic history, but that in the early universe “it may be easier for infalling gas to go all the way to the center of a galaxy, forming a black hole rather than condensing into stars on the way in.”

A key question that’s still unanswered, he adds, is whether galaxies must have a minimum mass in order to possess a central black hole. “This is relevant to the issue of how the 'seed' black holes form, and to the role of mergers … in building up galaxies,” says Rees.

Other news reports:

• Black Holes Preceded Galaxies, Discovery Suggests (1/6/09) – Space.com
  
• Black holes may precede galaxies, astronomers say (1/7/09) – Reuters
  
• Did black holes form before galaxies? (1/7/09) – Nature
  
• The Pit in a Galaxy's Peach (1/7/09) – ScienceNOW
  
• Which came first: galaxies or black holes? (1/7/09) – Cosmos
  
• Black holes 'preceded galaxies' (1/7/09) – BBC
  
• Black holes grew up before galaxies (1/7/09) – New Scientist
  
• Study: Black holes seem to form before galaxies (1/7/09) – AP
  
• Which came first--galaxies or black holes? (1/7/09) – Scientific American (blog)
  
• Yo Galaxy's Mama Is a Black Hole (1/7/09) – Wired (blog)
  
• Black Holes First, Galaxies Second (1/8/09) – Sky & Telescope
  
• Monster Black Holes Spawned Early Galaxies (1/8/09) – National Geographic
  
• Black holes sprout galaxies, not vice versa (1/8/09) – Los Angeles Times
  
• How did early black holes get so big? (1/19/08) – New Scientist
  
• Chicken-or-Egg Problem: Did Black Holes Form Before the Galaxies That Surround Them? (1/8/09) – 80beats
  

Tags: galaxies, black holes