Radiation pressure is a bitch. Not only does it prevent black holes from achieving their full potential but the repercussions can also have a huge impact on the surrounding galaxies. The burps of material let rip by accretion disks around supermassive black holes can be incredibly energetic; enough to shoot out huge radio-emitting jets into intergalactic space that are longer than the galaxy is wide. One such burp that was found by astronomers in March 2020 was the biggest outburst ever seen. It blew out a cavity seventeen times bigger than the Milky Way in the gas between galaxies in a cluster. That’d be like if a human burped in the UK and blew out a cavity in the Earth’s atmosphere that went all the way from Newfoundland to the Middle East!

The fact that something so tiny can have such a huge impact is mind-boggling. Let’s talk size first: the Milky Way is 100,000 light years across, whereas its black hole is only 0.002 light years across. To put that into context, there’s a similar size ratio between a football and the entire Earth. Imagine if the kick of a football could impact the entire planet; that’s what we’re talking about here with a black hole affecting a galaxy. Sure, the black hole is also supermassive, but compared to the total mass of the galaxy it’s a drop in the ocean. The Milky Way’s total mass in stars is estimated to be somewhere around 64 billion times the mass of the Sun, but its central supermassive black hole is only 4 million times the mass of the Sun: just 0.006 per cent of the galaxy’s mass in stars. And that’s just the mass in stars; the total mass, taking into account all the things we can’t see like gas, planets, smaller black holes and dark matter, brings the Milky Way’s total mass to 1.5 trillion times the mass of the Sun, with the supermassive black hole just 0.0002 per cent of that.

This is why, if you were somehow able to remove the supermassive black hole from the centre of the galaxy, the galaxy wouldn’t fall apart. Considering that all the stars in the galaxy are orbiting around the black hole in the centre, that’s a little difficult to wrap your head around. If you removed the Sun from the centre of the Solar System then all hell would break loose; but that’s because, as we heard in the previous chapter, the Sun is 99.8 per cent of the mass in the Solar System. Lose it and there’s nothing holding the planets in orbit anymore and the whole thing would slump apart. But remove the supermassive black hole from the centre of the galaxy and there’s enough mass in the rest of the galaxy to hold everything together (something known as self-gravity).

Despite this, the supermassive black hole and galaxy are intrinsically linked: the ratio of their two masses is consistent across the Universe. This was first noticed in 1995 by American astronomers John Kormendy and Douglas Richstone. After collating observations of eight nearby galaxies with active supermassive black holes (including the likes of Andromeda and Messier 87), they noticed a correlation between the mass of the supermassive black hole and the mass of a galaxy’s central bulge of stars (you can think of galaxies like a fried egg: they have a beautiful spiral flat disk shape akin to the egg white, and a central blob of stars like the egg yolk). On average, the black holes were a 1,000 times less massive than their galaxies.

Now, eight galaxies aren’t exactly representative of the entire galaxy population of the Universe, which is likely in the trillions of galaxies,⁹⁵ and so there was a push to measure the supermassive black hole and bulge masses in yet more galaxies to confirm if this correlation was real. This requires being able to work out the Doppler shift from the light emitted by the accretion disk to get at the supermassive black hole mass, and then model how the light is distributed in a galaxy to get at the bulge mass. From how much light you can see, you can then assume a ‘mass-to-light’ ratio, i.e. if there’s this much light then how many stars must there be producing it? To do this, you also need to know what the typical distribution of stars of different masses is in a galaxy (how many massive stars vs. how many smaller stars, on average). It’s not a simple task getting all these measurements, but by 1998, thirty-two more galaxies had a bulge mass estimate. This was thanks to the work of Northern Irish astrophysicist John Magorrian, who at the time was working with a giant of the field, Canadian astrophysicist Scott Tremaine, at the University of Toronto.⁹⁶ Magorrian is now an Associate Professor of Theoretical Astrophysics at the University of Oxford.⁹⁷ They used observations from the recently launched (and fixed) Hubble Space Telescope to show that there was indeed a correlation (and a fairly tight one too, as astrophysics goes), with the supermassive black holes⁹⁸ around 166 times the mass of the bulges (the Milky Way is actually an outlier from this relationship, with a much smaller black hole than you’d predict for its size).

This correlation is now known as the ‘Magorrian relation’ and is akin to finding a fossil and learning something new about how life has evolved on Earth. The correlation shows how galaxies and black holes have evolved and grown over the 13.8 billion years of the Universe. It all comes back to a galaxy’s bulge; the egg yolk at the centre. Once the initial chaos of formation has settled down, most galaxies will start life as a flat disk of stars all orbiting on nice ordered orbits in the same direction and in the same plane. But if two galaxies get drawn together due to gravity they can merge together, doubling their mass, but disrupting those orbits and the beautiful spiral shape in the process. Through many gravitational interactions, some stars lose energy and sink towards the centre of the galaxy where they form a denser bulge of stars, with haphazard orbits in different directions and planes that resemble a swarm of bees.

The two supermassive black holes also merge as the galaxies merge,⁹⁹ growing in mass. But just as the stars interact to sink them to the centre, so do gas particles, which find themselves funnelled into the black hole’s accretion disk so that it can grow further. It’s this joint growth of the galaxy and its black hole in a galaxy merger that’s thought to be responsible for the Magorrian correlation between the two. This idea is known as ‘co-evolution’ of galaxies and black holes. My own recent work has challenged the idea that mergers are the only process that can drive this co-evolution. Along with my colleagues Brooke Simmons and Chris Lintott, we observed some galaxies with no bulge, and therefore no merger, and showed that they have supermassive black holes as massive as those that have had a merger. We then collaborated with some theorist friends¹⁰⁰ who simulated this non-merger growth and found that it could explain 65 per cent of all supermassive black hole growth in the Universe. It’s likely that mergers aren’t the dominant force driving this correlation between black holes and their galaxies, but leave it with me a bit longer, to work out what process is responsible instead!¹⁰¹

Regardless of what’s driving it, this correlation has now been seen across a huge population of galaxies thanks to observations from huge astronomical surveys. These are telescopes that are not at the beck and call of astronomers around the world to observe a few objects as part of their current niche project¹⁰², but telescopes that observe the entire sky night after night, slowly building up a mosaic image of the entire sky, detecting fainter and fainter objects with every pass. This allows you to build huge catalogues of the positions, images and spectra of all the stars and galaxies visible from that part of the world. One of the largest of these surveys (and one of the largest collaborations of astronomers in the world) is the Sloan Digital Sky Survey¹⁰³ (SDSS) which uses the 2.5-metre optical telescope at Apache Point Observatory in the middle of the Sacramento Mountains in New Mexico. In its first data release in 2003, it provided observations of 134,000 galaxies across the northern sky, including over 18,000 quasars. By 2009, those numbers had jumped to just under one million galaxies and over 100,000 quasars.

The realms of large-number statistics had been opened to astronomers by surveys like SDSS, and they allow us to study populations of growing black holes to understand what their effect on galaxies truly is. Observations with SDSS confirmed the Magorrian relation, but also showed that the mass of the supermassive black hole also correlates with the total mass in stars in a galaxy, not just in its central regions. One thing these big surveys noticed, though, is that there is a steep drop-off in the number of the most massive galaxies. These are the galaxies that are 100 per cent bulge. They’ve had so many mergers that their spiral shape has been completely destroyed and what’s been left is just one giant big blob of a galaxy.¹⁰⁴

This distribution of the different masses of galaxies is called the ‘luminosity function’ (since mass is intrinsically tied with how bright a galaxy is and it is brightness that we measure directly), and to figure out what shape it is you have to first know how galaxies form and at what masses, and how they evolve after that. The people who originally tried to have a crack at predicting what gave us these differing numbers of smaller and more massive galaxies were British astrophysicists Martin Rees and Simon White, along with American Jerry Ostriker¹⁰⁵ in the late 1970s. That trio is a dream dinner party right there. Rees is the current serving Astronomer Royal, and previously served as Master of Trinity College, Cambridge, and President of the Royal Society. White was a PhD student at Cambridge at the time and has since served as one of the directors of the Max Planck Institute in Garching, Germany. Ostriker completed his PhD at the University of Chicago in the late 1960s with none other than Subrahmanyan Chandrasekhar (of the maximum mass of a white dwarf fame) and has served as professor of astrophysics at Cambridge, Princeton and Columbia, along with a stint as provost of Princeton University. They definitely qualify as BNIPs. Together they came up with a model for how galaxies form in the early Universe as gas clouds start to cool down; if a gas is too hot it can resist the pull of gravity down and it won’t become dense enough to form stars.

Rees, Ostriker and White thought that the cut-off in the luminosity function at high masses could be explained if the most massive galaxies formed from the most massive gas clouds. They reasoned that the Universe hasn’t been around long enough for these most massive gas clouds to have had enough time to cool yet. Their basic model of cooling gas clouds would be continuously fine-tuned over the next few decades by a host of astrophysicists, to encompass mergers of gas clouds and the effect of newly formed stars (which put out more heat to stop the gas clouds from cooling). By the early 2000s, astronomers had a realistic model and, crucially, enough computing power to simulate galaxies forming and evolving in the Universe.

You can then directly compare a computer-simulated universe to the observed Universe to check if you got everything right; including the shape of the luminosity function by just counting how many galaxies of each mass you form. It quickly became clear that the two did not match in the slightest. There were far too many high-mass galaxies in the simulated luminosity function. That meant the simulations were missing something; either some law of physics coded into the simulation was wrong or a process affecting galaxies was unaccounted for.

At the forefront of the development of these simulations was a group of astrophysicists at Durham University’s Institute for Computational Cosmology, including Carlos Frenk, Cedric Lacey, Carlton Baugh, Shaun Cole, Richard Bower and Andrew Benson.¹⁰⁶ Together, they realised that the missing process in the simulations was the energy injected by outflows driven by radiation pressure from the accretion disks around supermassive black holes. By 2003, they had managed to incorporate this into their simulations and show how they recreated the steep drop-off of the luminosity function: their simulation no longer over-produced massive galaxies.

The idea goes that the outflow of radiation and material from accretion onto the black hole can either heat up the gas (so that it can’t cool and collapse down to make new stars), or eject it from the galaxy entirely. Either way, the effect would be to quickly shut off star formation in a galaxy, at least in galaxies with the most massive supermassive black holes which, as we’ve heard, are in the most massive galaxies. We call this a ‘feedback’ effect, because as the galaxy feeds the black hole, the black hole can in turn throw energy back out that has a negative effect on the galaxy; the galaxy essentially shoots itself in the foot. It’s this feedback that’s thought to regulate the co-evolution of both galaxies and their central black holes, stopping both from getting too big for their boots.

With many other simulation groups managing to recreate the Durham group’s result, this feedback hypothesis became accepted among the theoretical astrophysical community. The problem is that among us observational astrophysicists that use telescopes to take data on the real Universe, we haven’t found any evidence for this happening. There have been some individual cases in single galaxies where the negative effects of an outflow or jet from an accretion disk can be seen (sometimes even causing shocks that compress gas so that new stars form in a process called positive feedback), but not in the huge population-wide studies, for example using data from a big sky survey like SDSS, that would allow us to make conclusions about the Universe as a whole. I should know; this is exactly what I spend the other half of my research time on, trying to find statistical evidence for negative feedback, helping to add my own little nuggets of insight into the collective astrophysical knowledge like all those before me.

A cartoon ‘luminosity function’ showing the number of galaxies found at each brightness that we observe in the Universe (solid line), compared to the number originally found in simulations (dashed line). Initially, simulations over-predicted the amount of very bright and very faint galaxies, revealing that there was some physical processes that had been missed.

One way we can tell whether an outflow from an active supermassive black hole has had an impact on a galaxy is to look at its colour. As we learnt right at the beginning of this book, the most massive blue stars live much shorter lives than the smaller red stars. So if you look at the colour of a galaxy overall and it looks very blue, you know it must have formed new stars recently. Whereas if a galaxy appears red overall then you know that enough time must have passed that the more massive stars have died and gone supernova, leaving just the smaller, longer lived, redder stars; like the dying embers of a fire. We refer to galaxies that have stopped forming stars as ‘red and dead’, and interestingly we find that around 70 per cent of them are the big blob galaxies.¹⁰⁷ From the colour of a galaxy we can then infer its average star formation rate – how many new stars it is forming each year.

In 2016, as part of my PhD, I looked for correlations between the star formation rates of galaxies and the presence or absence of an active supermassive black hole across the galaxy population. I got very excited when I found there was a difference between those galaxies that had actively growing supermassive black holes and those that didn’t. I was about ready to shout from the rooftops that I’d found the evidence that astrophysicists had been searching for when I remembered something that’s drilled into us science students: correlation does not imply causation.

For example, the sales of ice creams and sunglasses are correlated. Do you put on your sunglasses and immediately want an ice cream because of it? Or eat an ice cream and then wish to look as cool as your frozen dessert? No. The two are correlated because they’re both caused by the fact that the weather is warm and sunny. Remembering this fact, I realised that what I’d found was evidence for a shutdown of star formation at the same time as the black hole was active – what if another process was what had actually caused both things? Something that had managed to both heat the gas to stop stars forming and at the same time funnel gas towards the centre to feed the black hole. Maybe a merger of two galaxies? Or something else entirely again?

So now in my research, I’m trying to find the smoking gun of supermassive black hole feedback, something that’s irrefutably caused by the outflow itself. To do this, I’ve joined a worldwide collaboration of astrophysicists working with the telescope that conducted the Sloan Digital Sky Survey. It recently finished a new survey, called MaNGA,¹⁰⁸ where instead of taking one single observation of a whole galaxy, it takes over a hundred, mosaic-ing a galaxy to observe each region individually in over 10,000 galaxies. No longer are we resigned to reducing a complex system of billions of stars to a single measurement; we can peer into the inner workings of a galaxy to answer the unanswered questions still plaguing those studying the evolution of galaxies.

My niche area of that collaboration is trying to trace the feedback effects that supermassive black holes have on galaxies. Is there a correlation between the star formation rate in a given area and the distance from the black hole in the centre? Does that drop in star formation trace the energy of the outflow as it moves through the galaxy? If this effect is there, is it more appreciable in galaxies with more massive black holes? These are the unanswered questions I spend my days pondering over. It’s complicated stuff and easy to get frustrated with a lack of progress. But breakthroughs don’t happen overnight; the history laid out in this book is a testament to how slow and steady wins the collective human knowledge race.

With time, my colleagues and I will all analyse our data and publish our results, which will collectively come together to give us the big jigsaw puzzle picture of what is going on. Are the outflows from accreting black holes responsible for shutting off star formation in their galaxies to make them ‘red and dead’? Either way, something has been killing galaxies: there’s been a murder. And us astrophysics detectives will crack the case.