How Old Are Galaxies?

"How old are you?" A simple question, frequently asked of children, but rarely asked of an adult in polite company. When we ask the question, we are asking for the number of years that have elapsed since you were born. But how old are you, really? There were nine months of cellular development before birth, so for some purposes, maybe we should include that. Also, the average age of cells in your skin is less than two weeks, your stomach cells are typically less than 5 days old, and your blood cells are less than 4 months old. The average age of all the cells in your body is less than 10 years. So how old are you, really?

Okay, so for humans, if you want to get persnickety, maybe age isn't such a well-defined concept, but still, time since birth is generally a pretty useful definition.

When did galaxies form?

Two nearby galaxies. At the center is an elliptical galaxy,
which is basically a ball of old stars. A spiral galaxy, similar to
Milky Way,  appears in the upper right. Spiral galaxies are
still forming stars today. Do these galaxies have the same age?
This turns out to be a tricky question to answer.
Image from the Hubble Space Telescope.

When does the clock start for galaxies? When I was in graduate school, there was a concept of the epoch of galaxy formation, when big galaxies started forming. This idea was largely driven by the observation that nearby elliptical galaxies - and the central bulges of spiral galaxies - are filled with old stars that are not organized into thin, rotating disks.  It seemed likely that this was because galaxies formed their stars before the gas had a chance to settle into a thin disk.

If galaxies formed fast, then they should have been very bright when they were forming all those stars. So bright that once telescopes were equipped with modern CCD detectors in the late 1970's, it should have been possible to see them at large distances. No one could find them. Astronomers found lots of little faint blue galaxies, but these were smaller galaxies, that were much closer to us than expected.

At about the same time that observers weren't finding the epoch of galaxy formation, theorists were developing the idea that galaxies formed hierarchically, centered on the densest patches of dark matter in the early universe.  These early galaxies started forming their first stars when they had acquired only a tiny fraction of their present day material. They subsequently grew larger as more gas fell in at later times, and when they merged with other galaxies. This hierarchical theory simultaneously explains why we couldn't find the epoch of galaxy formation and why galaxies cluster together on large scales.

The implication is that galaxies started out with very few stars, and may have taken a long time to reach their peak star-formation rate. That's indeed what we infer today when we estimate the average number of stars formed per unit volume in the universe. This cosmic star-formation rate peaked when the universe was about 3.5 billion years old, even though we have now found galaxies that existed when the universe was less than a billion years old.

Inferring galaxy ages from colors

Even though the hierarchical models have been several decades and we have known for more than a decade that the globally-averaged cosmic star-formation rate started out much lower than the peak rate, it has taken a while for astronomers to realize that when they try to infer the age of an individual galaxy from its colors, the traditional assumptions are probabably incorrect. The traditional assumption has been that the star-formation starts off high and drops off exponentially. This was motivated by the idea that a galaxy starts with a fixed reservoir of gas and its star formation must keep dropping as that gas is used up. These are known as "tau models" in the jargon (not this kind of tau model!), because the Greek letter tau is usually used to designate the timescale for star-formation to reach roughly half of its initial peak.  Recent studies have shown that analyzing galaxy colors using tau models can give very misleading estimates of star-formation rates and ages, and that adopting different models can improve the estimates. See, for example, papers by my fellow bloggers Janine Pforr, Stijn Wuyts and my former student Joshua Lee.

Star-formation histories of galaxies from a hierarchical model (blue), compared to the best-fit tau models that were inferred from the galaxies' colors. You can see that the tau models are not at all representative of the true star-formation histories. But the problem is that the blue curves are models too. For real galaxies, we don't know what the true answer is. What we need is simple models that work well for a wider variety of possible star-forming histories than the tau models. From Lee et al. 2009.

 

A new definition of age?

 

Back when it seemed reasonable to use tau models, it seemed reasonable to designate the galaxy age as the elapsed time since it formed its first stars. This would be equivalent to galaxy birth. But now that we know that galaxies formed slowly, it's much less obvious what we should use as an age. On the one hand, the "chemical evolution" of a galaxy - the build-up of the heavy elements created in stars - is highly influenced by the first few stars. Once the gas in a galaxy contains heavy elements, it's ability to cool to form stars is dramatically altered. So we are interested in when galaxies formed their first stars. But on the other hand, most of the stars in a typical galaxy didn't form until billions of years later.

When using galaxy colors to estimate ages, the simplest approach is to estimate when the galaxy had formed half of its present stellar mass. This is probably the most reliable estimate of age that we can make if we have nothing but the galaxies' colors to work with, and no particular preconceptions about their star-formation histories.  But this is like starting the clock when the galaxy was a teenager. It's also a peculiar definition in the sense that a galaxy's age will not increase linearly with time. In fact, under this definition, if a galaxy doubles its stellar mass every 100 million years, then it will always be 100 million years old. If it then suddenly doubles its mass in only 10 million years, it will become only 10 million years old. This can be a bit confusing.

Another possibility is to adopt a different set of models that might be more representative of the true star-formation histories of galaxies than than the tau models. We can then define age as the time since t=0 in those models, as was done for the tau models. Unfortunately, there is currently no consensus on what to use as an alternative model. This is a topic of debate and discussion at galaxy evolution conferences. Maybe this will settle out and we will all agree on how to set our clocks. Or maybe the consensus will end up being that age is not a useful concept for galaxies. It's too soon to tell.

So the question "how old is this galaxy?" turns out to be more subtle than we might have thought. Scientific progress is a continual process of unveiling our ignorance. Each advance in our understanding leaves us still ignorant, but ignorant at a much deeper level.