Red, gas-poor, and still spiral-shaped

The single most conspicuous fact about a rich cluster of galaxies is that its members are the wrong colour. In the field, a spiral galaxy is blue, forming stars, and full of cold gas. In a cluster, a galaxy of the same size and the same stellar mass, with the same spiral arms visible in its stellar light, is red, forming nothing, and has almost no gas at all.

Whatever did that acted on the gas and not on the stars. That is a strong constraint, because most of the things one can imagine doing to a galaxy — a tidal encounter, a merger, a passage through the cluster’s own gravitational field — act on both.

The condition, and how little is in it #

Gunn and Gott wrote the criterion down in 1972 in a paper about how clusters make ellipticals, and it is one line. Ram pressure is the momentum flux of the headwind, the intracluster density times the square of the relative velocity. The restoring force per unit area on a thin disc’s gas is the gravitational attraction of the disc’s own mass, which for a razor-thin sheet is two pi times the gravitational constant times the surface density — multiplied by the gas surface density to turn a force per unit mass into a pressure.

Set them equal:

$$\rho_{\rm ICM} v^2 \;=\; 2\pi G\, \Sigma_\star(R)\, \Sigma_{\rm gas}(R).$$

There is nothing else in it. No cooling, no magnetic field, no viscosity, no assumption about what happens to the stripped material. Both sides are measurable — the left from X-ray observations of the cluster and a redshift, the right from photometry and radio observations of the galaxy — so the criterion can be tested galaxy by galaxy rather than in the aggregate.

The wind, and where it comes from #

The intracluster medium is the majority of a cluster’s ordinary matter. It contains several times more mass than all the galaxies in the cluster put together, it is at ten million kelvin or more because that is the temperature corresponding to the cluster’s own velocity dispersion, and it radiates X-rays by thermal bremsstrahlung — the emission that lets a cluster be weighed from its gas alone.

The density at the centre of a rich cluster is about ten to the minus three particles per cubic centimetre and falls outward; the infall speed of a galaxy arriving from the outskirts is roughly the escape speed, which for a cluster with a thousand-kilometre-a-second dispersion is a couple of thousand. Putting those in gives a ram pressure of a few times ten to the minus eleven dynes per square centimetre.

That is a very small pressure by terrestrial standards — a hundred-trillionth of an atmosphere. It is enough because the thing it is competing against is also very small: the self-gravity of a sheet whose surface density is a few tens of solar masses per square parsec.

What survives, and what does not #

The crossing point moves inward as the infall speed rises and as the galaxy penetrates to denser gas, so a galaxy is stripped progressively as it falls. The outer disc goes first, then the inner, and the process stops when the galaxy passes pericentre and the wind weakens.

For a Milky-Way-like disc in a rich cluster the stripping radius comes out at a few kiloparsecs — well inside the optical disc. So the galaxy keeps its bulge and its inner gas, loses everything outside, and continues forming stars in a shrinking central region for as long as that gas lasts.

Why the stars do not care #

The asymmetry is the whole mechanism and it is worth being explicit about it.

Gas is a fluid. Two parcels of gas that meet collide, exchange momentum and heat up, so a disc of gas moving through a medium at two thousand kilometres a second experiences a genuine pressure at its leading face.

Stars are not. Two stars passing at two thousand kilometres a second do not collide; they do not even deflect each other measurably, because the relaxation time of a galaxy is far longer than the age of the universe. A disc of stars moving through a cluster’s gas feels nothing at all, because there is nothing for it to feel.

What it looks like #

The mechanism makes a prediction that is unusually easy to check by eye: the stripped gas should be visible, behind the galaxy, in a tail.

Those tails are observed. In several clusters there are galaxies with tails of ionised gas tens of kiloparsecs long, streaming away from the cluster centre, with young stars forming in them — stars forming outside a galaxy, in material that has left it — and the birth rate in such material is measured from light almost none of the young stars emit. The population is informally called jellyfish galaxies and there are now hundreds of them. The tails also settle a question the criterion cannot: whether the gas is removed cleanly or shredded. A clean removal would produce a smooth tail; the observed ones are filamentary and clumpy, which means hydrodynamic instabilities at the interface are doing a substantial part of the work.

What the criterion leaves out #

Three things, and they all make the stripping more efficient rather than less.

The first is that the wind is not steady. A galaxy on a radial orbit meets a rising density and a rising speed, and the peak ram pressure at pericentre can be ten times the average. The criterion applied at the average underestimates.

The second is Kelvin–Helmholtz instability at the leading face. Gas that is not stripped outright by the pressure can still be ablated by the shear, which removes material below the formal stripping radius over longer times.

The third is that stripping is not the only thing acting. Starvation — the removal of the galaxy’s hot gas halo, which would otherwise cool and resupply the disc — shuts off star formation over billions of years without removing anything from the disc at all. It is the same accounting as the gas running out before the galaxy does, with the supply cut off from outside rather than merely exhausted. Tidal interactions with other cluster members do the same job more violently, and a close passage between two galaxies draws a bridge and a tail out of both — which is a stellar signature as well as a gaseous one, and therefore distinguishable.

Separating the mechanisms is the actual research problem, and the observational discriminant is timing. Stripping is fast and produces galaxies with truncated gas discs and undisturbed stellar ones; starvation is slow and produces a smooth reddening with no truncation. Both signatures are seen, in different galaxies.

Doing the arithmetic once #

The abstraction hides how close the competition is, so it is worth putting the numbers in.

Take the Milky Way’s disc at the solar radius: a stellar surface density of about fifty solar masses per square parsec and a gas surface density of about ten. Converting both to grams per square centimetre and multiplying by two pi times the gravitational constant gives a restoring pressure of about two times ten to the minus twelve dynes per square centimetre.

Now the wind. A rich cluster at ten to the minus three particles per cubic centimetre gives a mass density of about two times ten to the minus twenty-seven grams per cubic centimetre, and a galaxy falling at two thousand kilometres a second meets a ram pressure of eight times ten to the minus eleven.

The wind wins by a factor of forty. At the solar radius of a galaxy like this one, in the core of a rich cluster, the gas would be removed outright — and the same calculation in a group, where the density is a hundred times lower and the speed three times smaller, gives a ram pressure a thousand times smaller and the disc keeps everything.

That factor of a thousand between clusters and groups, against a factor of forty of margin at the solar radius, is why the mechanism switches on so abruptly with environment. It is not a gradual process that happens faster in denser places; it is a threshold that rich clusters are far past and groups are far short of.

The truncation argument has one further merit worth stating: it works on galaxies that show no tail at all. A tail is visible only while the stripped gas is still bright, which is a short interval, so tails select the objects being stripped right now. A truncated disc persists for as long as the gas stays gone, so it selects everything that has ever been stripped — which is a far larger sample and a far less biased one.

What was actually measured #

Four quantities, and the chain is short.

The intracluster density comes from X-ray surface brightness. Thermal bremsstrahlung emissivity goes as the square of the density, so the surface brightness integrated along a line of sight gives the density profile once a geometry is assumed — usually spherical symmetry, which for a relaxed cluster is adequate.

The infall speed comes from the cluster’s velocity dispersion, measured from the redshifts of its member galaxies, combined with an assumption about the orbit. A galaxy’s own three-dimensional velocity is never known; only the line-of-sight component is, so the ram pressure on any individual galaxy carries a projection factor.

The galaxy’s gas surface density comes from radio observations of atomic hydrogen at 21 centimetres, and its stellar surface density from photometry plus a mass-to-light ratio — the same conversion that makes a rotation curve’s decomposition into disc and halo a matter of judgement.

The population-level consequence #

Zoom out, and the mechanism is one term in the largest correlation in extragalactic astronomy: galaxies in dense environments are red and galaxies in sparse ones are blue, across four decades in local density.

That emptiness is the argument for a fast mechanism, and it is why stripping remains the leading candidate despite the difficulty of separating it from starvation. Starvation alone reddens a galaxy over several billion years, which is slow enough to leave a visible population in transit.

There is one further consequence worth recording, because it runs the other way. A galaxy that has been stripped is a galaxy whose gas is now somewhere else, and the somewhere else is the cluster. Over the history of a rich cluster the accumulated stripped material is a substantial fraction of the intracluster medium’s metals — the medium is enriched to about a third of solar abundance, which is far more than primordial gas could supply, and stripping is one of the two ways the enrichment could have happened.

The other is that the gas was expelled by the galaxies before they arrived, driven out by supernovae and by their own nuclei. Distinguishing the two is a question about where in the cluster the metals sit: material stripped on infall should be deposited along the orbits of the galaxies that lost it, and material expelled beforehand should be smoothly mixed. The observed distributions are closer to the second, which is a point against stripping being the dominant enrichment channel and no argument at all against its being the dominant quenching one.

The galaxies that were dead on arrival #

The account so far treats a galaxy as arriving at a cluster with a full gas disc and losing it on the way in. A substantial fraction of them arrive with the job already done.

Galaxies do not fall into clusters one at a time. They fall in as groups, because the large-scale structure delivers them along filaments and the filaments are populated by groups rather than by isolated objects. So a galaxy’s environment before it reaches a cluster is already denser than the field, and whatever quenches galaxies in groups has already had time to act.

That is called pre-processing, and estimates put it at a quarter to a half of the quenched population in a rich cluster — meaning that a substantial part of the correlation between environment and colour was established before the galaxies concerned ever met the intracluster medium.

There is a second population that confuses the picture from the other direction. A galaxy on a radial orbit passes through the cluster core and comes out the other side, and it can be found at a large cluster-centric radius while having already been stripped. Such backsplash galaxies look like infalling objects by position and like processed ones by colour, and they inflate the apparent quenched fraction in a cluster’s outskirts.

Separating the three populations requires knowing each galaxy’s orbital history, which is not observable. What is done instead is statistical: the distribution of galaxies in the plane of projected radius against line-of-sight velocity is different for infalling, backsplash and virialised populations, so a sample can be divided probabilistically even though no individual member can be classified.

A correlation between environment and colour is a correlation with the environment a galaxy has experienced rather than the one it currently occupies, and the difference between those two is most of the difficulty in the subject.

Both of those populations dilute the signal a stripping model is trying to fit, and neither can be removed from a sample by any cut on position or colour alone.

The gas disc that stops too soon #

The cleanest observational signature of stripping is not a tail and not a colour. It is the size of the atomic hydrogen disc, and it is a signature because of what such discs normally do.

In an isolated spiral the neutral hydrogen extends well beyond the stars — typically to one and a half or twice the optical radius, sometimes further, because the gas is not consumed out there and nothing removes it. That is the ordinary state, and it is why radio observations of nearby galaxies routinely find discs larger than the photographs suggest.

In a cluster spiral the hydrogen disc is frequently smaller than the optical one. The gas has been removed from the outside inward, exactly as the pressure balance requires, and what is left is a truncated disc inside a stellar disc of undiminished extent.

That inversion is diagnostic in a way a colour is not. Starvation removes the halo and lets the disc’s gas be consumed, which shortens the disc slowly and from everywhere at once; a tidal encounter removes gas and stars together, so both discs shrink. Only ram pressure removes gas from the outside while leaving the stars untouched, and only ram pressure produces a hydrogen disc contained within its own stellar one.

The measurement is a deficiency parameter: the observed hydrogen mass compared with what a field galaxy of the same type and optical size would have. Cluster spirals are routinely deficient by factors of three to ten, the deficiency rises toward the cluster centre, and the most deficient objects are the ones with the most truncated discs.

A mechanism that acts on one component and not another is testable by comparing the two components’ extents, and that comparison is available for a few hundred galaxies rather than for the handful with visible tails.

The criterion is a comparison of two surface densities, and both of them are worth moving, because a galaxy’s fate is decided by their ratio rather than by either alone.

Where the ladder goes #

The next rung is what happens to the stripped gas: whether it joins the intracluster medium and enriches it, which would show up as a metallicity gradient in the cluster gas, or whether it forms stars in the tail and returns some of itself as an intracluster stellar population. Both are observed, and their relative importance is not settled.

The other direction is the same physics applied where it is weaker. Groups of galaxies have dispersions of a few hundred kilometres a second and gas densities two decades lower, so the ram pressure is three orders of magnitude smaller — and yet the red fraction in groups is already elevated. Whatever is quenching galaxies in groups is not this, and identifying it is the more consequential problem, because most galaxies live in groups rather than in clusters.