There's a class of very distant galaxies (like, around 12 billion light-years distant) that have been the subject of a fair amount of research in the past decade or so – because the technology to even find them in the first place is not much older than that.
The class is called "Lyman-break galaxies". The reason for the name will emerge from the following, so just accept it for now.
The first thing we need to consider is how such very distant galaxies can be detected out of everything that's out there, most of which is considerably closer. It isn't actually so hard to get an image of a galaxy that far away. The so-called Hubble Deep Field survey, conducted in 1995, captured some objects as far away as 12.7 billion light years – as they existed only about a billion years after the big bang.
How do we know, roughly, what the distance is to these objects? That, too, isn't so hard. Astronomers just measure the redshift in the light spectrum from the object. This is, roughly, the amount by which well-known emission/absorption spectral lines have been shifted to longer wavelengths.
Because of the redshift-distance relationship discovered by Edwin Hubble, the redshift is a very good indication of the distance of the object. An accurate calculation of distance for a given redshift depends on the values of certain additional parameters. But these values have now been pretty well determined independently. You can experiment with the relationship yourself at this page.
The Hubble Deep Field survey was able to identify objects with a redshift as high as 6. That corresponds to a distance of about 12.7 billion light-years. Of course, objects at that distance are quite faint, so the only ones we can detect were extremely bright when the light we see was emitted. Essentially all we can see at that distance are quasars, which are galaxies with an extremely active, bright nucleus. Ordinary galaxies, especially smaller ones, are generally undetectable at that distance.
However, the Hubble telescope, and a few newer ground-based telescopes, can detect fairly small galaxies at a distance corresponding to z=3 or a little more. These are the Lyman-break galaxies. z=3 corresponds to an object that is about 11.5 billion light-years away. So we are now seeing the object as it was 11.5 billion years ago, about 2.2 billion years after the big bang – about 16% of the present age of the universe.
Now, even though it is possible and relatively straightforward to measure spectra of objects at z=3, it's still tedious and time-consuming. One would like to have an easier way to identify such objects in a survey of thousands of objects in something like the Hubble Deep Field.
As it turns out, there is a clever trick for finding objects with z=3 without even measuring spectra. The trick depends on the fact that even the brightest stars do not emit much light beyond a certain point in the extreme ultraviolet region of the spectrum. The cutoff is around a wavelength of 91 nm (billions of a meter). This is the so-called Lyman limit, which corresponds to the most energetic photons that can be emitted by very hot hydrogen.
The reason for this cutoff lies in the details of the Lyman series of lines in the emission spectrum of atomic hydrogen. Each line in that series corresponds to the energy of a photon which can be emitted when an electron moves to the lowest possible energy level from a higher level. There is a line at the same place in the absorption spectrum, due to an electron being boosted by a photon of the right energy from a lower to a higher level.
Because energy is quantized, the spectrum is not continuous and consists of discrete lines. However, at shorter wavelengths the lines come closer and closer together, until they reach a limit at 91.1267 nm. This is the Lyman limit. It represents the energy required to remove an electron completely from a hydrogen atom, starting at the lowest energy level. A hydrogen atom cannot emit a photon of higher energy, no matter how hot the gas is.
Since stars consist mostly of hydrogen gas, even the hottest stars cannot emit light with photons much more energetic (shorter wavelengths) than the Lyman limit. Since stars do contain other elements, especially helium, very hot stars are capable of radiating photons that are somewhat more energetic, but in general very little of the total energy output is from even farther into the ultraviolet. (Remember that shorter wavelength means higher energy.)
Even for stars that do radiate more energetic photons, such photons can ionize neutral hydrogen atoms, hence they will be absorbed by interstellar or intergalactic clouds of hydrogen. In fact, any photon more energetic than the Lyman limit is likely to be absorbed quickly by a hydrogen atom, because it can completely eject the only electron, with perhaps some energy left over (emitted as a longer wavelength photon).
The result is that there is a rather sharp cutoff (or "break") in a stellar emission spectrum at 91 nm. So normal galaxies whose light comes mostly from stars have the same cutoff. (In a quasar, where much of the radiation is due to matter falling into a supermassive black hole, it's possible for a large proportion of the energy output to consist of photons more energetic than the Lyman limit. Such photons are created, for example, in particle collisions between massive particles accelerated to near the speed of light.)
Bottom line: light from normal galaxies has a sharp cutoff at the 91 nm wavelength, when viewed from a great distance.
Now recall from our redshift discussion that if the redshift is z, then the factor by which wavelength is lengthened is z+1. Therefore, at a redshift of 3, 91 nm photons are shifted to a wavelength of 364 nm, in the near ultraviolet part of the spectrum. A normal star like the sun radiates a lot at this wavelength, in the range often called UV-A. So does a common "black light".
Simple filters are easily made that allow light to pass only in a narrow range of wavelengths. A blue filter, for example, passes photons only around 440-490 nm. If you have a filter that passes only ultraviolet light in a band of wavelengths less than 364 nm, no light at all will get through from objects with z≥3. Objects with z<3 may still be visible through such a filter, since there will still be some shifted photons with wavelengths somewhat less than 364 nm. But some light from a z=3 object could pass through a filter that admits light with wavelengths greater than 364 nm, including a filter for light in the visible range.
Objects with z>3 will have their Lyman limit shifted all the way into the visible part of the spectrum, so they might not be visible even through a blue filter. In fact, if the criterion is objects that don't make it through a blue filter, one can estimate that 3.5≲z≲4.5. At z=6, Lyman limit photons are shifted to 637 nm – red light. A red filter would therefore pass visible light from a z=6 object, but blue or green filters would not.
Astronomers take advantage of this situation by using an ultraviolet filter, together with filters in the visible range (e. g. red, green, and blue). Any object which is visible through all the filters must have z<3, but an object visible through red and blue filters, and not the ultraviolet filter, should have z≈3. By adjusting the short wavelength filter, it's possible to select for objects at higher z in the same way.
The technique isn't foolproof. For example, a galaxy that has few hot stars with strong emission near 91 nm will have most of its light shifted so it doesn't pass through the UV filter, even if z≈2. Indeed, even an ordinary star in our galaxy could pass the test if it has little emission at UV wavelengths. Of course, such a star would be a lot cooler than our sun, and have a negligible redshift besides. So the actual redshift has to be confirmed by spectrometer measurement. However, the procedure is pretty efficient – around 75% of candidate objects actually have z≈3.
Thus it's relatively simple to do a survey for objects around specific redshift values, without going to the trouble of doing a spectroscopic measurement. Galaxies discovered in this way are, naturally enough, called Lyman-break galaxies (LBG for short) .
OK, that's all well and good, but what have we actually discovered about LBGs? There are some things we can learn just by counting them. However, ideally we could image them through telescopes in enough detail to learn something about their size, morphology, rate of star formation, evolution, and so on.
Such questions are easiest to deal with for the galaxies near z=3, because we can detect more of them than at higher redshifts (the more distant ones may be too small or dim to see at all), and because we can see more detail in them than more distant ones.
The first thing to note is that most LBGs will have fairly active star formation going on, because only young, newly-formed stars are hot enough to emit light near 91 nm.
The distribution of such galaxies is somewhat interesting. When the filters are arranged to be able to identify galaxies with 2.4≤z≤3.6, surveys find upwards of 400 candidates. The redshifts of these can then be measured with a spectrometer. There are several peaks in the distribution, around z=2.95, 3.15, and 3.35. By combining the redshift data with spatial direction of each object, actual clusters of galaxies can be identified.
The existence of distinct clustering at less than 2 billion years after the big bang is an indication that the spatial distribution of matter, including dark matter, at that time was already fairly lumpy.
Now, about the galaxies themselves. What are we able to say about them? Well, in terms of the spectra we can measure, the LBGs seem to be much like galaxies near us that have a lot of star-forming activity. (Such galaxies are called starburst galaxies.) The spectra indicate the presence of many very large, hot stars, of spectral class O and B. Such stars are normally rare, because they have very short lives. So if we can detect that there are many such stars, relatively speaking, it means that the rate of forming new stars of this type (and, presumably other types) is fairly high – much higher (in terms of stars per unit volume) than in our galaxy or most nearby galaxies today.
Another example of what can be inferred from the spectra is that the LBGs seem to have a lot of interstellar gas and dust, compared to nearby galaxies with rapid star formation. Uncertainties about just how much dust is there complicates estimation of how many stars might be present. The spectra also suggest that there are large outflows of gas and dust from the LBGs. These outflows are typically driven by very hot, massive stars' stellar winds and supernovae (which are the terminal stages of massive stars). Such characteristics indicate that rapid star formation in LBGs and nearby star-forming galaxies may have different causes.
The most interesting and most fundamental question about galaxy evolution is that of whether a typical galaxy grows in a "monolithic" way, in which most news stars are formed as the mass of baryonic matter gradually contracts gravitationally. Or, on the other hand, whether some stars form in small proto-galaxies, which go on to merge with each other, assembling large galaxies like our own, in the process undergoing bursts of new star formation. It's also possible that both processes occur to significant degrees.
This question is difficult to answer with spectroscopic data. We really need to have direct optical information about the shape ("morphology") of early and evolving galaxies. The problem is that, in most cases, galaxies with z≳2 are too small and faint to image optically with adequate resolution, even with the current generation of sophisticated ground-based telescopes using adaptive optics. That is, we can't see directly whether the early galaxies are amorphous blobs, regular ellipsoids, or picturesque spirals. However, there seem to be relatively few examples with highly elongated shapes, corresponding to a spiral galaxy seen mostly edge-on.
But it's hard to discern morphology reliably even using a space telescope like Hubble. The resolution is such that a single pixel may cover a large part of the object – a few percent of the area, or much more. In a subsequent article I'll discuss an interesting case where we've gotten lucky.
In general, the Hubble finds that LBGs at z≈3 seem to be fairly compact and regular. We still can't necessarily distinguish ellipsoids from spirals. But with z≳4, objects appear to be more diffuse and irregular.
The size of the objects, at least in terms of the parts not too faint to be detectable, seem to be on the order of 50,000 light-years or less – smaller than half the size of our galaxy. But this represents emissions in the ultraviolet part of the spectrum (which we see shifted to the visible part), so the extent of the objects in terms of emitted visible light could well be larger.
Much of our knowledge of LBGs at z≈3 was established over 10 years ago. Since then we have learned somewhat more about the issues of clustering, stellar types and gas/dust content in early galaxies, and galaxy morphology, by a variety of means. We will look at some recent studies about such topics before too long (hopefully).
It's worth noting that the Lyman-break technique can be also be used at smaller and larger values of z. At values of z>3 the Lyman-break will be shifted all the way into the visible part of the spectrum, while for z<3, the Lyman-break will occur at shorter UV wavelengths. All these can be handled by proper choice of filters.
For example, for z up to about 6, around 1 billion years after the big bang, we have, from 2003, a study that shows relatively few bright galaxies. Since this was in the period of reionization, which required many hot, bright stars, there must have been many more galaxies around that were just too small to be detected.
New insight into the cosmic renaissance epoch (8/21/03)
In particular, the astronomers conclude on the basis of their unique data that there were considerably fewer luminous galaxies in the Universe at this early stage than 500 million years later.
There must therefore be many less luminous galaxies in the region of space that they studied, too faint to be detected in this study. It must be those still unidentified galaxies that emit the majority of the energetic photons needed to ionise the hydrogen in the Universe at that particularly epoch.
Ironically, it was a little later that the Lyman-break technique was applied to closer objects, z≈1. In 2006 we have:
Ubiquitous Galaxies Discovered In The Early Universe (3/9/06)
For the first time, Denis Burgarella and his team have been able to detect less distant galaxies via the Lyman-break technique. The team collected data from various origins: UV data from the NASA GALEX satellite, infrared data from the SPITZER satellite, and data in the visible range at ESO telescopes. From these data, they selected about 300 galaxies showing a far-UV disappearance. These galaxies have a redshift ranging from 0.9 to 1.3, that is, they are observed at a moment when the Universe had less than half of its current age. ...
From their observations of this sample, the team also inferred various information about these galaxies. Combining UV and infrared measurements makes it possible to determine the formation rate for stars in these distant galaxies for the first time. Stars form there very actively, at a rate of a few hundred to one thousand stars per year (only a few stars currently form in our Galaxy each year). The team also studied their morphology, and show that most of them are spiral galaxies. Up to now, distant galaxies were believed to be mainly interacting galaxies, with irregular and complex shapes. Denis Burgarella and his colleagues have now shown that the galaxies in their sample, seen when the Universe had about 40% of its current age, have regular shapes, similar to present-day galaxies like ours.
Mapping the Distant Universe
The Properties of Lyman-Break Galazies at z∼3 – excellent article
Searches for high-redshift galaxies
Links to this post: