Tutorial bits on galaxy spectra

[NGC3314]

Just so I'm giving back to the community, and egged on by some PM queries - here are some tidbits about galaxy spectra, addressing things that have come up in various posts.

First up - the break in galaxy spectra near 4000 Angstroms (400 nm) in the galaxy's own reference frame, otherwise known as the Ca II break or HK break. These names refer to part of this "step" in the spectrum of a group of old stars being produced by the two strong lines of Ca+, absorption  lines which were designated H and K by Fraunhofer nearly 200 years ago. [Historical note - in discussions of atomic physics, superscript 0 refers to spectral lines arises from neutral atoms of a given species, a single + to once-ionized, and so on. However, in spectroscopic discussions, we start with roman numerals - I for neutral, II for once-ionized, and so on, so Ca II absorption lines arise from calcium atoms once ionized.] Here's a piece of elliptical-galaxy spectrum showing the break region:



Notice the "step" in intensity, where the spectrum to the red side of the break is almost twice as bright as to the blue side. It's not just the calcium absorption lines that do it - there's a CN band on the blue side, and light from old stars is dropping off rapidly to the blue anyway. The strength of this break is conveniently measured by the ratio of mean intensity in the two shaded regions, and it changes a lot with age of a group of stars. Here's a set of models (from Guy Worthey's web service) showing how it strengthens as the stars age:



Thus the amplitude of the break tells at a glance something about the stellar population from a galaxy spevtrum. When an active galactic nucleus (in the post extreme cases, a quasar) is present, its spetrum is continuous right across 4000 Angstroms, so the break is diluted by whatever fraction of light the nucleus contributes. In full-fledged quasars, so little of the light comes from the surrounding galaxy that the 4000-A break can be very weak or unmeasurable.

And that leads me into the next post, spectra of quasars and other kinds of active galaxies...

[Hrundi]

Thank you :heart:!
I've been dying to get a comprehensive newbie guide.

I have this question about the Ca break though.

The CaII break is near the HeI line, right? But what of the designated CaII absorption lines near 9000 angstroms?

[NGC3314]

The CaII break is near the HeI line, right? But what of the designated CaII absorption lines near 9000 angstroms?

Yes - the Ca II break is around the two lines denoted H-epsilon and K on that plot to the blue end - it's happenstance that the Ca II "H" line almost coincides in wavelength with the H-epsilon line of hydrogen, both close to 3970 A in the emitted reference frame. There are three far-red Ca II lines as well, as you noted (the "infrared triplet") which are very strong in cool stars, but they don't have an associated spectral break. In fact, their use in unravelling the content of galaxies hasn't been as fully explored as the 4000-A break. One reason is that it doesn't take much of a redshift to slide them into wavelengths where we need different kinds of detectors to see them, and there start to be more and more pieces of spectrum eaten up by absorption from water vapor in the Earth's atmosphere as well.

[NGC3314]

Next up - spectra of active galactic nuclei. These can be broadly defined as galactic nuclei which show large amounts of energy release that we can't account for by the normal processes of stars, ther formation and death. They come in several varieties, originally depending on how they were discovered by subsequently somewhat unified as we learned more.

Quasars - started as an acronym for quasi-stellar radio sources (QSRS), although we know find that most such (also known as quasistellar objects, QSOs) are not very strong radio sources. A quasar is starlike through a typical groundbased telescope, has a significant redshift (say z>0.1, some as high as 6.4), and shows broad emission lines in its spectrum.

Seyfert galaxies - discussed as a class by Carl Seyfert in a 1943 paper. (By the way, he pronounced his name more like See-fert than Say-furt). These are distinguished from normal galaxy cores by their spectra. There are two subtypes. Type 1 show broad emission lines, with widths that imply internal Doppler shifts of several thousand km/s. Type 2 show narrower lines, a thousand km/s or so of Doppler shift, but the relative strengths of these lines are quire different from what we see in star-formng regions such as the Orion Nebula or starburst galaxies. For example, the [N II] line next to H-alpha can be nearly as strong as the hydrogen line. [Semantic digression - the sqare brackets indicate a so-called forbidden transition, one in which the atom takes so long to radiate that it will be left along that long only in a very good vacuum, such as interstellar space. It took a long time to work this out, so there are some old articles which still refer to some of these as unknown elements such as nebulium and coronium.] The conditions in these galaxies also mean that gas with a wide range of oinization states exists in close proximity - the best example, is oxygen, for which we often see strong emission lines from the neutral and one- and twice-ionized states (O I, [O II], and [O III] in spectroscopic parlance) Star-forming galaxies can make plenty ot [O II] and [O III] but not [O I] at the same time. Some very nearby galaxies have Seyfert nuclei if you look closely enough. M81 has a simmering low-level one. The most famous ones are probably NGC 4151 (type 1) and NGC 1068 or Messier 77 (type 2). The spectrum of a type 1 Seyfert is almost undistinguishable from a typical quasar. Both show not only large line widths (which suggests in itself matter anchored in a strong gravitational field), but the kinds of atoms radiating indicate dense gas exposed to radiation rich in far-ultraviolet and X-ray emission (both quite distinct from gas lit up by young stars, which aren't such strong sources of these high-energy bits of the spectrum). This fits with the widely-held idea of a supermassive black hole and surrounding gas which is heated by particle collisions at very high velocities giving rise (through more complex processes than we first expected...) to the observed radiation.

BL Lacertae objects or blazars - their optical spectra are almost competely lacking in features, so getting a redshift is no picnic and may be best done using the faint light from the surrounding galaxy. They seem to be quasars that we happen to see looking right down a jet emerging near the speed of light, so what we see is dominated by the glare of the hard radiation for the jet boosted by effects of relatively.

Radio galaxies - their defining features are obviously strong radio emission, usually in the form of twin blobs or lobes on either side of the galaxy. The optical spectra may look like Seyfert galaxies or either type, or like an ordinary run-of-the-mill elliptical. Some radio galaxies are pretty clearly quasars seen from such an angle that surrounding dust blocks our view of the bright core (from which careful measurements can sometimes pull out a reflected signal.) The same is true of some type 2 Seyferts - some at least are type 1 objects seen through absorbing material so we don't see the core region directly.

LINERs are recognized as sets of emission lines from otherwise normal galaxy cores which may indicate low-level activity, like a Seyfert with the volume turned down. These are seldom powerful enough to show up against the galaxy light in SDSS spectra, so you shouldn't see these often.

This graphic compares the visible-light spectra of various kinds of AGN, all shifted to zero redshift for comparison:


Lots more on AGN can be read here, with a text intro, glossary, and many images and plots. Points of note here -

- You may well see galaxies with a Seyfert nucleus where you see some of the strong emission lines and also see absorption lines from the stars in the surrounding galaxy.

- The SDSS spectroscopic pipeline is better at picking out type 2 (narrow-line) Seyferts than broad-line type 1 objects, simply because it doesn't have to then guess a line width to identify in what may be a noisy set of data. So you may well see a broad-line spectrum like the Seyfert 1 above in something not autoclassified as a quasar.

[Hrundi]

The H-alpha+[NII] emission line is merged into one due to their extreme proximity, right? So it makes classifying stuff by the length of the [NII] emission quite a bit harder?

[Hrundi]

What's this?

http://cas.sdss.org/astro/en/tools/explore/obj.asp?id=587732701783589053

[NGC3314]

What's this?

Cool! That's what we see from a post-starburst galaxy, seen in the aftermath of a massive burst of star formation. The strong absorption lines of hydrogen show that much of the visible light comes from A-type stars like Vega, which live a billion years or so. In a typical spiral, the overall star formation rate is pretty continuous, so that A stars never dominate the whole spectrum. One might see such a spectrum from a merger a couple of rotation periods after the nuclei meld, or from a galaxy which underwent a very dramnatic but non-merging interaction.

On the previous question - the Sloan spectra can split the H-alpha and [N II] 6583-Angstrom emisison lines when they are narrow enough. But since the optical fiber can include much of the light of distant galaxies (by design), even a rapidly rotating spiral can have them so smeared by the Doppler shifts as to blend. If there are other emission lines such as H-beta or the oxygen lines at 4959, 5007 Angstroms, you can make an estimate of the H-alpha/[N II] ratio from the mean wavelength of the blend (taking into account that there is a second [N II] line on the other side of H-alpha, at 6548 A; their intensity ratio is fixed at 2.93 by atomic physics).

Next up when I have a chance - emission and absorption lines from stellar populations...

[Hrundi]

Thanks I'm also surprised at how many H II region like spectra I find.

This is from a spiral. Am I missing something? As in, is this a type of AGN, or are H II core spirals like triangulum much more common than the ones with no H II regions in the middle.

[NGC3314]

This time we'll look at the connection between the spectra of individual stars and those of galaxies which you get when you pile a lot of them together. To a first approximation, the spectrum of a star looks like a blackbody, the ideal case of an opaque hot material. The wavelength of its peak intensity varies inversely with the temperature of a blackbody; the Sun peaks around 5800 A and it's at about 5500 Kelvin, so a star at half that temperature will peak longward of 10,000 Angstroms (1 micrometer), while the hottest massive stars at 30,000 K or more will peak deep in the ultraviolet. Thus, when you mix light from a bunch of stars, you get a mixture which is spread more broadly over the spectrum than any single kind of star. For two reasons, there are more cool, low-mass stars than hot, massive ones. First, in every star-forming region we can study in detail, there are many more low-mass stars formed than high-mass ones. On top of that, stars live progressively longer when their masses are lower, by a huge factor (very roughly inverse cube of the mass), so low-mass stars contribute for a much longer time. From calculations of the life cycles of stars, we can work out what mix to expect at various ages and for various histories of star formation. It's helpful to separate red giants of various temperatures. This figure, made from data obtained in the early 1980s, makes the point, by showing on the left what various ages of a star-forming region should look like (T9=age in billions of years, the bottom represents a constant level of star formation), and a comparison with red giants of various spectral type (=temperature) on the right (coolest at the bottom). (When you get to such cool stars, they start to depart from the blackbody ideal by quite a lot as their cooler atmospheres modify the radiation from the interior - you can see very wide bites taken out by absorption of such molecules as TiO and eventually even water vapor for the very coolest). Speaking of molecules, the huge dip near 7600 Angstroms is from oxygen in our own atmosphere, something that the SDSS pipeline processing did an approximate correction for in their data.



This shows that younger mixes of stars are bluer; as time goes by, the most massive hot stars finish their life cycles and vanish into neutron stars, black holes, or supernova debris, so their blue contribution goes away. When there is nearby gas, the hottest stars can ionize it. In this process, deep ultraviolet light is absorbed by the gas, stripping electrons free. We see the flip side, known as recombination - when these electrons rejoin atoms, they give off radiation in a cascade including distinct wavelengths associated with jumps between possible energy levels (we see this most oftem asthe hydrogen lines H-alpha, H-beta, and so on. The strength of these line tells you how much UV starlight there is, so that give insight into stars so hot that they don't contribute much to the visible light (although internal dust can swallow the UV light first). The stronger the emission lines are, the more ongoing star formation. Additional emission lines can be formed from collisions of the loose electrons with heavy-element atoms. Emission lines can also come from gas ionized by the hard radiation field of an active nucleus; we can usually distinguish these cases by the ratios and widths(very narrow lines from star-forming galaxies, and only AGN give strong [O I] lines, as specific examples). 

Elliptical galaxies are generally dominated by old stellar populations - more or less like a mix of 50% of the T9=15 spectrum above and 50% K-type red giants. Spirals can look like anything from this to a pure young population depending on their history of star formation. Because the massive stars in a star-forming region produce so much light per solar mass, the spectrum is dominated by whatever the last major star-forming event was; we have to look pretty hard to work out what happened before that.

[Hrundi]

Thanks a lot This answered a lot of questions I had.
I do wonder though, how to identify these galaxies at higher redshifts? Once the H-alpha lines are already past 9000 angstroms and even more extreme cases?

[Nightblizzard]

Well x) I don't really understand much of this. The spectrum mostly shows how much of what element is in a galaxy and thus you can conclude several things... that's what I got so far o.o. But one thing that confuses me... the spectrum is always from 4000 to 9000 nm, which is the infrared-area. Why is there no visisble-light area spectrum? Also, what does the y-axis say? (I didn't read all posts, just the first one... but it already confused me enough x). The SDSS tutorial only explains HII areas. At least I got that...)

[NGC3314]

Well x) I don't really understand much of this. The spectrum mostly shows how much of what element is in a galaxy and thus you can conclude several things... that's what I got so far o.o. But one thing that confuses me... the spectrum is always from 4000 to 9000 nm, which is the infrared-area. Why is there no visisble-light area spectrum? Also, what does the y-axis say? (I didn't read all posts, just the first one... but it already confused me enough x). The SDSS tutorial only explains HII areas. At least I got that...)

That's an optical spectrum (fairly similar to the SDSS ones) - although often unlabelled, they show the region from about 4000-9000 Angstroms or 400-900 nm. 4-digit numbers are so easy to remember and carry more information than 3 that astronomers have been loathe to abandon use of Angstroms, although this is depreceted in the official SI units. The y-axes in all these cases are intensity of the light received at each wavelength, usually per unit wavelength (so something like erg/cm^2 second reaching the top of the atmosphere per angstrom of spectral dispersion, for example).

[altymczuk]

A truckload of thanks to you for your ongoing tutorial(s) on spectra.  They certainly help a great deal.

[NGC3314]

Thanks a lot This answered a lot of questions I had.
I do wonder though, how to identify these galaxies at higher redshifts? Once the H-alpha lines are already past 9000 angstroms and even more extreme cases?

If it's really important to get H-alpha and neighbors, try to get time on a huge telescope with IR spectrograph... Other than that, for unravelling the stars, there are eventually additional spectral features in the UV that the redshift brings into the optical window (for example a Mg II break near 2800 Angstroms that acts somewhat like the 4000-Angstrom break), but you do lose information on the cooler/older stellar populations. Hot stars present strong lines both in absorption and emission from their winds, which are so prevalent in the UV that they dominate the spectra of star-forming galaxies. Among emission lines, you get the H-beta region to about z=0.6, and the violet lines like O+ at 3727 Angstroms even longer. There are no strong lines from H II regions blueward of that, so in the UV one gets diagnostics of the stars but not the gas from typical data.

[Hrundi]

What of AGN galaxies when their OI, H-beta OIII, NII and SII emissions are gone? And another type of thing is that I've seen things with pretty odd OIII and H-beta ratios in which the OIII is much stronger, along with strong SII emission lines, but no OI.

And thanks a lot. I think I'm actually getting the hang of all this.