The largest telescope that will ever be built*

There's an asterisk in the title of this video. Of course there's an asterisk. There has to be. And I'll be honest... I had a whole script ready to go here as the sun sets over the Atacama Desert in Chile. But, I've found so much more than I expected over the last day and a half. So many stories. I've been exhausted both from the altitude – we're three kilometres up here – and from dashing around with handheld cameras trying to capture everything. So, what's going to happen now is we're going to cut to a version of me in a couple weeks' time who is coherent and who's had time to process all this. [Caption+ by JS* caption. plus | @caption_plus] Here's the story. I got an invitation to visit the Extremely Large Telescope. Scientists love giving telescopes names like that. It's a telescope. It's extremely large. The invite was from the UK Science and Technology Facilities Council and the European Southern Observatory. They arranged everything, but they have no editorial control over what I'm saying. And I paid for my own travel. Those flights were expensive. So there'll be an advert for NordVPN later on, because it was actually useful to me while I was in Chile. Anyway, the Atacama Desert is the most desolate place I've ever been to. Rolling hills of stone and rock for hour after hour of driving. I'm sure there's life out there somewhere, but we didn't see any of it. Eventually, we arrive at the base camp at Paranal. Let me give you the lay of the land. At the bottom of the hill is the Residence, where visiting scientists and non-scientists stay. Not a hotel. Very definitely not a hotel. You can't pay for a room there if you try. It's more like university dorms, if university dorms ever won architectural awards. That's at the bottom of the hill, along with all the support buildings and structures that are needed to support life out there. Up at the top of the hill is the Very Large Telescope, built about 20 years ago. And a few kilometres away, a little dot on a flattened-off hill on the horizon: that's the Extremely Large Telescope, the new one. And that's likely going to be the largest optical telescope that will ever be built. But to understand why that is, I need to show you around. Our first stop is at the Very Large Telescope – the VLT – up on the hill about 2km from base camp. In theory, you can walk it. There's a path called the Star Track, but... I didn't feel like getting altitude sickness. The VLT has been operational for more than 20 years now. Results from it have won Nobel Prizes. It was the first telescope to take an actual picture of a planet around another star. It tested Einstein's general relativity by tracking a star around the supermassive black hole at the centre of the galaxy. And it's actually several telescopes. The four big ones are called the Unit Telescopes, or UTs. And we were visiting UT4. Are we good to just go in? Okay. Oh, wow! (laughs) So, at Paranal, we have four UT telescopes. These are the large telescopes we have here. Where the primary mirror, this thing that you see here, underneath us here, is 8. 2 metres in diameter. And that is a mirror... very, very scientifically precise, but the same kind of optical mirror that you would have just to do your makeup in the morning or something like that. Um, yes, yes. (laughs) Lesson one: Do not compare the extremely expensive scientific instrument to a makeup mirror. The important thing is that the telescopes here are optical telescopes. That's the first reason for the asterisk. There are already bigger radio telescopes. I was lucky enough to visit Arecibo before the collapse, and I recently visited Parkes in Australia. They're all much bigger. But they collect radio waves, not light. They pick up very different frequencies, and they're useful for observing different objects and doing different science. There's stuff you can do with radio that you can't do with optical, and vice versa. Radio telescopes have to be much larger because the wavelengths of the microwaves and radio waves they pick up range from millimetres to tens of metres. But they don't have to be as precise. As long as you build a big bowl that's the right shape, you don't need to make a perfect optical mirror finish. It will still work. Arecibo's surface was just kind of rough mesh, but that still worked. Optical astronomy looks for visible light or infrared. The stuff we can see with the naked eye, or close to that. The wavelengths are from about 100 micrometres to 100 nanometres. And so the optics have to be so, so precise, just like any other camera lens. Well actually, more precise, because mirrors this big have other problems. When you get to this large telescope, it's actually extremely difficult to keep the mirror stable and in the same position. It has 150 actuators underneath it. It's like poking the mirror ever so slightly. That must be microns. That must be a tiny, tiny amount it's moving. Yeah, and this is basically to take into account gravity of the mirror and the temperature difference we have within here, because when you move the mirror, then from gravity, then the shape of the mirror is also changing as well. That's called "active optics", deforming the mirror ever so slightly to deal with the effects of gravity as you tilt it, and from thermal expansion as the temperature changes. That's how accurate this has to be.

We'll come back to that later, 'cause I got to see the mirror and those actuators close up, but first... We've got to explain what that mirror actually does. They were getting it ready for the night, so they had to test it. Because the whole thing moves. Oh, oh! We're moving. That's way faster than I thought it was gonna be. We basically test the rotation. We test all of the movements of the telescopes. We check for the safety of the system, before we're going to start the nighttime operations. You can't hear the movement, right? This is so fast! Oh, we're going back the other way. I'm just giggling. This is several hundred tons of material. It's moving like this, and basically it's kind of floating on top of this very small amount of oil film that's underneath here. Because during the night, when we're doing the scientific observations, we may be staying on one target. So observing one galaxy for several hours. Because the Earth is rotating, so therefore we need to move the telescope so that we keep pointing at exactly the same position. So this movement needs to be extremely smooth, because we don't want to add any additional movement onto the observations when we're observing our science targets. Rotation test complete, now the tilt. We're going to open the dome... and we want to put the mirror in such a way that if anything falls down from the dome, that the probability of that actually falling on the mirror is extremely low. In the middle of the desert, constantly you're having dust that's accumulating onto the mirror. I mean, large things falling onto it is extremely rare. (chuckles) I just looked over there. Oh, wow! So the light would come in from a star. Yes. Hit the main mirror, which is the big reflective thing there... back to M2 just here. Exactly. And then down to M3 in the middle. It can rotate to send it to any two at the instrument, or it opens to send it to the instrument that's below the M1 mirror. Right. There's a lot of very bad animations out there, which show satellite transmissions and stuff like that beaming not into the bowl of the dish, but to the detector, the thing that sticks out. And that is the wrong way round. That's really important. Light bounces off M1, the bowl, and gets focused onto the focal point. And then here it gets bounced down again, either to the equipment at the sides or down into the basement. Light's going to come in and hit M1. Yes. But isn't M2 in the way of what you're trying to observe? No, it doesn't, because you have parallel light beams that are arriving to the M1. It's not going to come across on camera, I don't think... But there is quite a little white mark just there, and a lot of dust on the mirror. How much is that going to affect the observations? For any given object of a given brightness, as the mirror gets dirtier, then you'll have to observe for longer to get the same amount of light on your detector. We recoat the mirror once every two or three years, maintain the reflectivity. They do use dry ice to dust the mirrors every few weeks, but that can't get everything that builds up over time. Dust on the mirror doesn't change the optics. It doesn't spoil the image. It just means that some light isn't reflected at all. So over time, the image gets less and less bright. I did get to see the mirror recoating lab later. They'd just started to clean one of the other telescope mirrors, the one from UT1. And when I arrived at the lab, the first thing I saw was the structure they used to transport the whole thing down the hill. How on earth did you get that thing all the way down the hill from UT1? First, we have to disconnect the cell from the main structure of the telescope. And then we put this structure on the yellow part you can see, which is the carriage. And this carriage is put on top of a truck. And we drive three kilometres. This system is over air cushions, and we can move with the air cushions. Wait, it's a hovercraft? Yeah. Exactly. It's basically a hovercraft! It is exactly the same principle. Because the mirror is not here right now, maybe we can go up to it. Yeah, yeah. Let's go up. Do you want to go up? Yeah, absolutely. So what you can see from up top is the telescope without the mirror on it. The mirror, I thought, was kind of a bowl shape. But this looks flat to me. The sag is not so high. It's 30 cm. So it looks flat, but it's a sphere. Okay, it's just a section of a very big sphere. Exactly. You can imagine that if we make the full bowl out of this mirror, it will be a bowl of 60 m diameter. So let's go see the mirror being cleaned. They'd moved it into the clean room next door. And to start with, it was just being washed. Also, I realise I look stupid with the hoodie on under the clean room gown. I mean, first, it's not a medical or semiconductor-grade clean room. They assured me it's fine. And second, the reason I'm always in the hoodie is that we were always going out and in, so the temperature was always switching from cold, air conditioned interior to brutally hot sun that I needed to cover myself from. So it just seemed like the best uniform for the job. Oh, that is the mirror. Yeah, so we have a small window so you can... Oh, I'm scared to breathe now. What you can see is the washing unit. Basically, it's the place where we do the stripping of the coating and the washing before entering the vacuum chamber. Right now, we are in the first step of the process, which is a first cleaning to remove all the dust, also all the stains that was collected

during two years by the mirror. And then, we can start to do the stripping or etching of the coating, using this rotating arm to pour the acid and to have the layer going out. So, wash the mirror to remove the contaminants, then acid etch the old coating away, and then... When we have cleaned the mirror, and we will close this big chamber, do the vacuum, and deposit the aluminium. You have the heart of the machine. This is what we call the magnetron, and it's where we have the target of aluminium. 99. 9% of aluminum, very pure. That's way beyond what you'd normally get for industrial use. And by the way, there are some shutters. It is closed, so we cannot see the target. The rainbow colours are some deposition during the process. During years you have these thin layers, then it's nice. How much is being added here? The thickness of the coating is around 0. 1 micron, which is very, very small. And a micron is 1/1,000th of a millimetre. Exactly. 1/10,000th of a millimetre. You can basically count the number of atoms there. It's a 1,000 atoms' layer. One thousand. (heavy exhale) The weight is around seven grams. Over the entire eight metre mirror? They atomise aluminium to create the mirror layer. And when I say atomise, I don't mean like perfume sprayer atomiser. I mean literally, they use plasma to make individual atoms of aluminium float around in a vacuum, and then they just let it softly rain down onto the mirror until they have a coating about 1,000 atoms thick. Anyway, up at UT4, the dome was opening. So this is going to point at the sky. Yes. That's obviously a lot more light than it would normally get? You have the shutter in place. So this, you know, black curtain that you have here is there so no light's going to get through there. Oh, it's just a physical blackout curtain. Okay, yeah. I can see that just there. And I just noticed the guide star lasers. I know they've been there all the time. They've been really obvious. But now it's pointed this way, I can see one, two, three, four giant laser emitters! Those are enormous! It's giant laser time! But we can only see that after the sun sets. That's not the ocean, by the way. Those are clouds. We're above the clouds. Anyway, sun sets, stars come out. The Milky Way is just stretched out above Paranal, and it is so beautiful! I've never seen the stars this clearly! And they don't seem to twinkle, 'cause there's less atmosphere up here to cause that twinkling, that distortion. The atmosphere isn't one consistent thing. There are different temperatures and pressures of air, and the light gets refracted. It's not like wind blowing balloons around. The photons of light aren't being blown away. But it's the same reason that an object in a swimming pool looks distorted when you look down at it from above the surface. That's a difference in the medium that the light is travelling through. The different pockets of air that are constantly moving about in the upper atmosphere cause the same sort of effect. That's why the stars twinkle. The light is slightly bending its path as the pockets of air move around. That's the reason that the telescope is built high up on a desert mountain. There is less atmosphere to get in the way. But there is still some twinkling, even if I can't spot it myself. Enough that it'd still cause problems with observations. And that's what the lasers are for. The lasers didn't look quite that bright in person. Those are long exposure photos, but they looked cool enough. And down in the control room, the scientists there explained what they used them for. This is the laser control? In some way, it's like a status monitor, where we can see the status of the laser beacons, which are these four images we see up here, and also what commands are being sent to the deformable mirror to correct the wavefront. Four lasers beamed out, camera scanning exactly what's happening to the beams in terms of just position? Is it just finding the brightest dot in the sky? The lasers generate a point source that we can measure with the wavefront sensor. And the idea behind the adaptive optics is to make that point source as tight and as compact as possible. So we generate these artificial stars, the laser beacons. We measure them, we apply correction to the deformable mirror, and that hopefully improves the quality of the laser beacon, and thus improving science image. Every millisecond, we're making a measurement of the wavefront, and it's being commanded to the DSM. What's the lag on that? It's a few milliseconds. But the DSM is wibbling like this really quickly. Yeah! Every millisecond. That's absolutely amazing. Yeah. Why four lasers? So, multiple lasers allows us to correct the wavefront over a large area on the sky, providing uniform image quality across the science observation. If you only had one laser, you would get good correction in the middle. But then it would degrade as you go further out. And I figure they're at a specific frequency that you can then notch out in the rest of the observations? Exactly, yes. This is the sodium wavelength. On the instrument, we have a filter to filter out that light. Also on the telescopes. Yeah. On the guide star. So, in short, they shoot the laser up, see how it wobbles, and then do maths

to subtract that wobble from the observation. Look, I'm gonna say "they do maths" at a few points during this video, because it's the sort of calculation that people spend years learning about. Suffice it to say, they do maths, work out how the laser guide stars are moving, and then physically move and distort the M2 mirror to subtract that atmospheric distortion from the actual stuff they're looking at. So, now you know how a modern optical telescope works, and how precise and difficult the work is. That's the first thing you need to know to understand why the Extremely Large Telescope, the new one, is probably the biggest that will ever be built. The next day, we're able to go down into the basement underneath the four telescopes. Because when I said that the VLT was four Unit Telescopes... Yes it is, but they can all act as one, called the VLTI: Very Large Telescope Interferometer. There is a lot of competition for 'biggest telescope in the world'. There's things like the Event Horizon Telescope, which is a network of radio telescopes all around the globe. They all point at the same thing at the same time, and because they are so far apart on other sides of the planet, the astronomers can record the data that's received, analyse it with signal processing algorithms and supercomputers, and do maths, and create a virtual telescope that is the size of the planet. That's called interferometry, because the maths is all about how those signals interfere with each other. We basically completely change the way that we handle the data the minute we start combining these beams. We're thinking about interfering waves of light rather than an image. And the pattern we get from the interference of that light is where we extract all of the information that we want. Radio astronomy can do that by recording the radio waves, and then processing it all afterwards. But remember how I said the wavelength of radio is longer, and the telescopes don't have to be so precise? Well, turns out a lot of other things don't have to be so precise for that either. Atmospheric variations are a huge limitation for us in the optical because, you know, the atmosphere is full of water. It's full of things that vary very quickly at optical and infrared wavelengths. The atmosphere is not varying so crazily at radio and submillimetre wavelengths. So you have— you're able to maintain all that phase information. It doesn't get completely smeared out by the atmospheric variations. You know if there's a party with a really loud sound system a way away? Then you won't hear the treble, tss-tss-tss-tss. But you will hear the thump-thump-thump of the bass? Radio waves, and... Apologies to physicists watching this, but radio waves are the bass of the electromagnetic spectrum. They travel just fine through the atmosphere with minimal distortion. That's why nobody bothers sending radio telescopes into space. We don't need to. But at optical frequencies, it's different. All the information that you need to do interferometry, that information gets smudged much more easily as it passes through the atmosphere. And that's not even the worst problem. For light, there is not a computer in the world fast enough to deal with all that information. And there is not a clock in the world precise enough. Even radio interferometry requires very precise atomic clocks synchronised across the globe. But there just isn't a clock in the world accurate enough to do that for the micrometre wavelengths or nanometre wavelengths, and terahertz frequencies of light. A radio wave... You can just record it. Even before digital computers were a thing, you could just record a radio wave onto magnetic tape. Every single detail of it. Every frequency, every little nuance, everything. Heck, a boombox from the '80s can do that. When people used to tape songs off the radio, that is a very rough and imprecise version of radio astronomy. But for light, there is not a computer in the world fast enough to record that data. Not with the accuracy required for interferometry. So instead... they do it physically in real time. Remember when I said that the light from the Unit Telescopes could be sent off to the equipment at the side, or down into the basement? Let's go see the basement. So where are we headed now? We are going inside what we call the coudé room. Coudé? Yeah, this is the room containing the adaptive optics system for the interferometer. And coudé means... Elbow. Elbow, in French. Okay. (laughs) So I'm guessing from the name that this is a movable joint? A hinge that can change the direction of the light? Yes. Then this, the coudé, for us, the coudé path, is what we call the nine mirrors. That's a serious warning sign, that is. (chuckles nervously) Yeah. Oh! Okay, we are under the telescope, and we are under what we call the coudé path. And this is really where we receive the light, which will be, let's call it organised, in order to bring it towards the VLTI. The light comes into each of the telescopes, it gets bounced around the mirrors of the delay tunnel, gets funnelled into the VLTI lab, instrument, and that's where it gets combined. So the light beam isn't recorded anywhere.

It's just bounced into tiny tunnels from each telescope. And its next stop is a 150 metre long human accessible tunnel to synchronise all the beams. The delay lines. So these tunnels run all the way from all the telescopes to a central point? Oh? Sorry. Cannot cross to this side of the room. Okay, my entire body needs to stay... This is the boundary. Alright, thank you. (chuckles) Sorry. Okay, so I'm not going to go near their equipment if I'm told not to. But in my head, I'm like, that's a little strange. That looks like fairly heavy infrastructure. Surely I can't do any damage if I accidentally just... bump up against it? Well, the light arrives in the delay lines in that tunnel from the side. It hits a series of mirrors designed to synchronise all the beams, and each beam is bounced up the tunnel to one of the carriages and reflected back. So how precise does that have to be? I can show you. Okay. We use the delay lines for observation, and you can see here the error. 136 nanometres. I assumed that the carriage would be accurate to maybe a millimetre? And then there'd be some electronics and mirrors in there that would do the nanometre correction. But is that carriage accurate to 100 nanometres in its position? The error that I showed you there is the precision that the carriage has. I understand why I got told not to even get close to that line. That's incredible! The carriages can be moved with 100 nanometres of precision. 100 nanometres! That's why they didn't want me near them. They'll have to recalibrate each night anyway, of course, and it'll get cleaned from time to time, but they don't want some idiot potentially bumping into things that have to be 100 nanometres accurate. Because that's how they delay the light and do optical interferometry. That's how everything is synchronised. No recording, no playback, no computers analysing things. It's analogue. It's physical. Explaining what happens in the science part of that, inside the freezing cold vacuum of... science stuff, in the rest of the lab, is way beyond me. That's the stuff astronomy PhDs are made of. And it's still a lot of work to record the interference patterns and work back from that. But, if all those beams are synchronised enough... it means they can approximate a telescope the size of the entire VLT platform, all four Unit Telescopes, instead of just the individual telescope mirrors. But all those telescopes, for optical, have to be physically connected. So, we've covered interferometry. We've covered the mirrors and resurfacing. And we've covered the sheer scale of these telescopes. You now know everything you need to know to work out why the ELT – the Extremely Large Telescope – will probably be the largest optical telescope ever constructed. And that ELT... is our final stop. If you're wondering why I'm rocking the extremely fashionable combination of awful raggedy old baseball cap that I found in my checked baggage, and hoodie up... The altitude is about three kilometres here. The UV index is just "yes" 'cause that's one of the reasons they built here. We are as close to the top of the atmosphere as you can reasonably get. Let's go, guys. Oh, we've actually got named safety gear in our size. Okay, there we go. Required protective equipment up here: Helmet, gloves, high vis, steel toe cap boots, and factor 50 sunblock. The construction site is 20 kilometres away. Down the hill that the VLT is on, up the next one. It is colossal, and it looks so much closer than it is. My brain refused to admit that a telescope dome could be that big. The construction is 74 metres high, 86 metres in diameter. And that... That is an extremely large telescope. The main mirror, the M1, 39. 2 metres in diameter, comprised of 798 hexagonal panels. We will make the biggest optical mirror ever built. That is... colossal! We are expecting images five times sharper than the James Webb Telescope that is in outer space. Sorry, five times sharper than the James Webb? The one that's out there Yeah. beyond the atmosphere? Yeah, exactly. The first reason that there's likely to be nothing bigger than this is sheer cost. The cost line does not follow a linear relationship, but it goes exponential. So maybe you will add, you say, a couple of metres more, and the costs double. There have been plans for bigger telescopes. There was something called the Overwhelmingly Large Telescope, the OWLT, that was briefly planned. It could happen. But it'd cost tens of billions of euro, and frankly, the technology's unproven. Second reason: We're reaching the limits of what is physically possible with current construction technology. They're engineering something that is part telescope, part skyscraper. So it's lots of different sections all working together? Yeah, all the sections are working together, and its geometry will be readjusted

through three actuators per segment, for a total number of 2,394 actuators. The eight metre mirrors in the VLT, they sag under their own weight. And granted, the ELT is using mirror segments, not a single monolithic mirror, but keeping them all aligned isn't easy, and the entire structure will still stretch and warp as it's moved around. And with thermal expansion as the temperature changes. This wall is basically the upper foundation, where on top of it, the telescope will be mounted. Right. So this is fixed to the ground, and on top of it, there is a hydraulic bearing system that will be moving the telescope. How precise does that have to be, for those movements? We are talking about 100/1,000, depending on the observation, of arc seconds. Okay, so what does a thousandth of an arc second of precision mean? I did some maths, and it means that this colossal skyscraper telescope will move so smoothly and so precisely that if you told it to look at and track a spot on the moon, it would be accurate to within two metres! The actual imaging resolution's not quite that small. It could image buildings about 10 metres across at that distance, but again... That's a distance that can be measured in light seconds! We are at the limits of what it's possible to do with current technology. And the last reason – and I think the most important one – is that at some point, we're going to solve optical interferometry. People are working on it. Right now, it still has to be physical. Computers aren't fast enough. Clocks aren't precise enough. But they're only going to get better. At some point in the future, we are probably going to figure that out, and we'll be able to do optical interferometry the same way that we can do radio. And as soon as that happens, the Extremely Large Telescope will be... not obsolete. It'll still be useful for some types of observations, because there are advantages to the size. Big telescopes let you detect very small things and very faint things. Interferometry gets you the small stuff, but not the faint stuff, because there's less mirror surface for the light to bounce off. But odds are, at some point, there'll be much cheaper and easier virtual ways of making really big telescopes. The ELT is still going to do a hell of a lot of science before that can happen, if it ever does happen. But once that's done, maybe in a few decades, maybe a century, the case for building something this big kind of goes away, if instead, you can build a few small telescopes and piece them all together. So this, the Extremely Large Telescope, will probably be the last of the great optical telescopes. I could be wrong. Who knows? Maybe in a century's time, this will seem ridiculous, and there'll be a 100 metre mirror sitting high on a plateau somewhere in South America. But if I had to place a bet, I think there's a very strong chance the ELT will be the largest optical telescope to ever be built. And, as I said earlier, NordVPN was extremely useful to me. I have been using it a lot, because while I was in Chile, websites kept showing up in the wrong language. They're all like, 'No voy a renunciar a ti'. Plus, over the last year or so, I've saved hundreds of dollars on car rental, because some rental companies charge way less just depending on where they think your computer's located. Same car, same contract, same driving licence, same everything. I went all the way to the final checkout to make sure. I just clicked NordVPN's magic button to say I was back in the UK. And suddenly I got a rental for almost half price. And of course, if you want to watch shows from back home, you can. Just check the streaming service's terms first. You can use NordVPN on six devices at the same time, across Windows, Mac, Linux, iOS, and Android. And there's a 30-day money back guarantee if you want to test it out first. If you go to nordvpn. com/tomscott, click the link in the description or scan the QR code, then you'll get the best deal they're currently offering whenever you're watching this video.