SPEAKER 1: Welcome to this year's Bethe Lectures. I'm going to start by saying a few words about Hans Bethe and his connection to Cornell, and then I'll introduce the speaker, Francis Halzen.
Hans began his career in the '20s in Germany, where he was among the first people to study quantum mechanics in that very exciting time. He left Germany in 1935 with the rise of the Nazis and eventually ended up here at Cornell-- actually ended up at Cornell in 1935. And about four years later, he had his first major scientific contribution, where he explained how stars operate, how they burn hydrogen to helium, for which work he received the Nobel Prize in 1967.
Hans retired from Cornell in the mid '70s, but that was by no means the end of his scientific career. People have said that, if any of us had a career after he retired, we would retire after that second career very happily. Among the things that he did, in his research-- something that's connected to today's speaker-- he in 1986 had a big and important contribution to the solution of the solar neutrino problem that describes how the neutrinos from the sun change their flavor on their way to the earth.
And so that's just one of many ways that he contributed. In fact, I worked on that project as a student. And I remember when [? Snow, ?] the experiment, figured out how to resolve that problem experimentally, people came up here to Cornell. My thesis advisor came up here to Cornell to tell Hans Bethe how important he was. They wanted to personally convey the information to him. Hans was an active researcher until right before his death in 2005.
Hans also was very active in shaping the Physics Department. I think the collegial atmosphere we have here in our department, what this department is known for is really thanks to him. So his impact on the campus cannot be overstated, which is indicated also by now the house on West Campus, the Hans Bethe House, which is in his memory.
So let me switch now to introducing today's speaker, Francis Halzen. Professor Halzen is the Hilldale and Gregory Breit Distinguished Professor of Physics at the University of Wisconsin-Madison, and he's also the director of the Institute for Elementary Particle Physics Research, and the acting director of the Wisconsin IceCube Particle Astrophysics Center.
He's the principal investigator for the IceCube experiment, which we'll hear about today-- the kilometer-sized cube neutrino telescope buried in the Antarctic ice. Professor Halzen is also a fellow of the American Physical Society, and he received his Masters and PhD degrees from Belgium, from the University of Louvain. He's also received numerous other awards, including Helmholtz-Humboldt Research Award and honorary doctorates from a number of institutions. In addition, he's also an excellent writer. His essay "Antarctic Dreams" was featured in the best America science writing in the year 2000.
So before we welcome Professor Halzen, I'd like to point out that he will also be giving two additional talks, one talk tomorrow-- this a particle physics seminar at 4:00 PM in PSP 401 entitled "IceCube Neutrinos from Oscillations to PeV Dark Matter." And then, there will be a public lecture at 7:30 on Wednesday in this room, "Ice Fishing for Neutrinos." So I encourage all of you to come to these.
The title of today's colloquium is "IceCube-- The Discovery of High-Energy Cosmic Neutrinos." And before we introduce Professor Halzen, let me just remind you that for the graduate students, postdocs, and undergraduates there will be an opportunity to meet with Professor Halzen in 401 PSP?
Yes? OK. 401 PSP after the talk.
FRANCIS HALZEN: Well, thank you very much for the invitation. I realize this is a great honor for me. I am going to tell you about IceCube. And so the subject is really the detector, the discovery of cosmic neutrinos, and what we are going to do next. But to introduce the subject, I'm going to tell you a little bit of why we actually built this instrument. Why would you set out and spend 25 years of your life building a kilometer cube neutrino detector? There was actually a good reason for this.
So this is astronomy as I know it. I'm not an astronomer. But this picture tells you how many photons there are in the sky as a function of their wavelength or as a function of their energy. And so, you recognize this is the diffuse flux in the universe of radio waves, the microwave background-- 410 per cubic centimeter-- visible light, et cetera. And so, you see here the spectrum actually ends on GeV gamma rays in this picture, which is 1990 as you can see there.
And so, the idea already originated in the 1960s that maybe we had to do astronomy with neutrinos instead of photons because high-energy photons are very easy to detect. You can detect them with an air shower array, a cosmic ray array. But they were not seen. They were only upper limits, and some of these upper limits are very good actually. So the idea was maybe this part of the sky, we are going to do astronomy, with neutrinos.
And in fact, part of the next question is, is there anything there? And of course, there is something there, cosmic rays. And they are, of course, a serious background on this plot. The ratio of cosmic rays to gamma rays at TeV is a factor of 1,000. And so the question was, is this a background we cannot beat? But anyway, this is not a background for neutrinos.
Since 1990, people have seen TeV gamma rays. By the way, my unit will be 10 to the 12 electron volt, the TeV. And so it's the Fermilab accelerator beam that's my small unit in this talk.
So the subject of this talk I can summarize in one word here. This is what we discovered. We actually did discover. It was very surprising to us that we actually discovered what we were supposed to be doing. So that's the topic of this talk.
Now, neutrino astronomy, that idea is obvious. In fact Fred Reines once told me-- he discovered the neutrino in '56-- that everybody had the idea you could do astronomy once they realized there was a real particle. It's electrically neutral, just like a photon. So it doesn't get bent in magnetic fields. That's why we don't know where the cosmic rays come from, because they don't point back at their sources. They gyrate in the magnetic field of our galaxy before reaching the detector.
In this talk, they are massless. Tomorrow, they are not massless. But today, they are massless. They are essentially unabsorbed, also by the detector. And what's interesting, they track nuclear processes, so they are ideal to look for sources where there is actually hadronic physics happening. I'm a particle physicist. I cannot relate to normal astronomy. But as I said, they are not ideal because they are difficult to detect.
So in the '60s, in early '60s, the suggestion was made to do astronomy with neutrinos. There were papers written by Markov, by Reines himself, by Greisen here-- in 1963, very fundamental paper. Basically, if you take the IceCube proposal and this paper, it's the same except for the size of the detector.
And the size of the detector, it was '69 that [? Sobczak ?] and [? Barasinski ?] did a very simple calculation. And they calculated the following thing. Now, if you look here, these are the highest-energy cosmic rays. They have an energy of 100 million TeV, and so they don't feel the field of our galaxy, the magnetic field. They go in straight lines. So you could possibly do astronomy with them. And as they are not contained, they fill the universe. There may not be many of them, but they fill the whole universe. And they realized that, if they fill the whole universe, they live in the same place as microwave photons.
And so, if they live in the same place, they are going to meet, so the cosmic ray flux will interact with the microwave target. And we know everything. We know the beam. We know the density of the target, 410 per cubic centimeter. So you know how many pions you produce, and the pions will decay into neutrinos. And so that's 1950s particle physics. Also, we know the cross-section. So you can calculate the calculation I just did on the back of an envelope, and you will find, if you have a cubic-kilometer detector, you will see one event per year. I think for IceCube this number, with very sophisticated Monte Carlo's, I think it's 0.9.
Also, we have taken data for seven years now, but not with the full detector, and we have not seen an event, which is becoming very interesting. That's a different talk.
But these are great neutrinos because their energy is 10 to the 6 TeV. So it's like someone slams your detector with a hammer. You cannot miss it. And so I'll come back to them later.
So let me tell you a little bit about cosmic rays. I mean, there are sure people in the audience who know more about cosmic rays than I do, but this is kind of my three-slide lecture on cosmic rays.
This is a solar flare. And in a solar flare, you see, these are streams of relativistic particles. And when you have streams of relativistic particles, you make magnetic fields. And by shocks or reconnection, you can accelerate particles there. And this is the basic formula. If you want to accelerate a particle to energy E in a magnetic field B, you better contain it. So its gyro radius has to be smaller than the accelerator. In fact, if you apply this to Fermilab, you get the right answer.
So the gyro radius, E over B, has to be smaller than the size over which you have high magnetic field. And so that means that the maximum energy is B times R. And so this is a challenge. If you apply this formula to the picture I just showed, the answer is 10 GeV. And in fact, after a solar flare, you wait for a day, 10 GeV protons arrive at Earth. So nature does this very well. I don't quite understand why.
But it's very difficult now to go back to find magnetic fields over large enough scales to get particles at like 100 million TeV. And in fact, then, even if you have an idea, what you don't see on these plots-- experimentalists tell you it's 1 per kilometer squared per century or something like that-- but that's not the point. The actual luminosity of these beams in astrophysical terms is very high.
So there is basically only one idea, and that's the idea on this slide that the particles are accelerated when stars collapse. When they collapse to a neutron star, then you see they build up shocks over 1,000 years. That's a star that collapsed a few hundred years ago. And these filaments are exactly the same thing as the Sun, but bigger and higher magnetic field. And this is the favorite picture for explaining the cosmic rays produced in our galaxy.
The ones outside our galaxy that I was talking about, you saw the movie. It's the same thing, except the star collapsed to a black hole, and everything happens in seconds instead of 1,000 of years. And you can actually, dimensionally at least, get to 100 million. This is a great explanation. And so you have this gravitational collapse of which a little bit of energy-- maybe 1%, a few percent-- is transformed in acceleration of particles, and you have explained cosmic rays.
Problem is, there's no evidence for this. In fact, even before IceCube was complete, the simplest experiment, the simplest observation we can do is we can look for neutrinos. These gamma ray bursts are explosions. So astronomers see them happen. They tell you where to look in the sky and at what time. For a few seconds, there's no background in your detector. We have looked at over 1,000 of them. We have never seen a neutrino.
And in fact, the flux limit we have is less than 1% of the neutrinos we actually do see. that I will tell you about later. So it's kind of in trouble.
There's only one other good idea around-- there are other good ideas around, but not very credible-- and that is that this is an active galaxy. There is a supermassive black hole here. You get the idea. You need particle flows. So in the inflows or in the jets that are emitted by these black holes, you have also the opportunity to accelerate.
Now, you may say, what does this have to do with neutrinos? Well, it's very simple. These systems are beam dumps. So what's a beam dump? If you go to Fermilab or CERN, what people do is they accelerate a proton beam, shoot it in a dense target, and produce pions, kaons, everything-- the whole particle data book-- and neutrinos come out at the other end. You absorb everything in your target except the neutrinos.
Well, that's how these black hole or neutron star systems were. You have an accelerator, but the accelerator is surrounded by light, dust, molecular clouds occasionally. And so your accelerated particles move through a target-- for instance, the target is typically light in the extragalactic systems. You produce by pluses, but you also produce by 0s. So you have to remember, for every neutrino you potentially see, you have to see a gamma ray from the same energy. And remember, nobody has seen PeV gamma rays. I will come back to that.
So this was kind of the sexy sales pitch for building IceCube. Here is this slide. So let's just look here. The predicted flux falls with energy like 1 over E squared. That's the vanilla theoretical prediction. So we always plot E squared times the flux. And these are the predictions. Galactic supernova, gamma ray burst, this is the calculation that Sobczak and [? Barasinski ?] did. And you see this line up. They line up on a flux where you have about 10 to a 100 events per year. For on the detector, that's fully efficient over one kilometer cubed, and that's what we set out to build. That was the logic.
However, you see this. What's that? Well, that's the problem. These are measurements by our R&D project, which was called AMANDA and IceCube, of the flux of neutrinos in the atmosphere. Of course, you know these cosmic rays we are talking about, they interact in the atmosphere, produce pions that produce neutrinos and also muons, which will be a problem-- long-lived muons at high energy.
But you see, this was the idea, to somehow, if you come to this place-- this is a 5 in GeV, this is 100 TeV. Remember, that's the critical scale. At 100 TeV, this flux is too small to produce particles in your detector. So the simplest way to see an extragalactic or galactic flux is to find the fields that are much more energetic.
And we note we understand this flux very well. We can calculate it. We have measured in many different ways. And so we know how this flux behaves. Notice also it suggests we can measure energy. Come back to that.
So how do you build a kilometer cube detector? That idea also goes back to 1960. This is Markov who I mentioned before. This is Pontecorvo who had, essentially, every idea in neutrino physics except this one. This one was due to Markov, and it's shown on the next slide. Here is how you detect neutrinos. You look for a muon coming through the earth. Muons, of course, don't come through the earth. It means it was posted there by a neutrino that interacted under you detector somewhere.
And these muons at these very high energies, they are very efficient. They range from 50 meters to 50 kilometers in range. So all you have to do is measure the muon traveling through your detector. You map the Cherenkov cone, and you have the direction. And so every neutrino is like a pixel in the sky above Cornell if you looked at that the South Pole.
And so this idea was in Hawaii by putting photomultipliers in water. The twist we put on this idea was to try to put them in ice. And so this is the Geographic South Pole, somewhere here, where the National Science Foundation has a research station. This is the IceCube project. This is the runway where the planes land. And if you go one and a half kilometers deep, you find incredibly clear ice in which you can build this detector. And so that's what we did.
So you go essentially a mile deep. And you can imagine, this is a kilometer cube of ice. That's filled with 5,060 10-inch photomultipliers. They're about that big. So simple commercial photomultipliers. Here is a picture of one.
And they are, of course, in a glass pressure vessel because they're going to go two and a half kilometers deep some of them. But what you see here is basically data acquisition system in a PC that digitizes the data that come out of the photomultiplier and send it to the surface. So a typical signal looks like this. So it gives you a map of the light. Each time a photomultiplier detects takes light, it makes a map of what it detects, scans photons as a function of time, and you put the time stamp on it and send it to a computer at the surface.
Now, before I go on, everybody's now thinking, how do you put 10-inch photomultipliers a mile deep in ice? So that's easy. It's not easy, but it's easy to explain. I made a less-than-two-minute movie that answers that question. So the top 80 to 100 meters is snow, and you just melt it. And then, this is a hot water drill, and it just puts hot water under pressure and falls by gravity. And you wait two days, and it's two and a half kilometers deep, and then you have water instead of ice in the hole, and you pull it out.
And so all you need for this is a five-megawatt heating plant, which is about 40 car wash heaters. And so you put out 200 gallons per minute 90-degree water under pressure. And so the whole drilling system is built on sleds and put where you want to drill your hole to put photomultipliers in.
Here, this is a wonder of technology. It's a two-and-a-half kilometer-long hose, about this big, that can hang in hot water without collapsing. It was built in Italy. Here are the car wash heaters. They are just supplied by normal generators. There's nothing sophisticated about the system. And this is the moment the drill comes out. So ice is an insulator, so that water stays liquid for a while. You don't really have to worry that much. And so once the drill is out, you will see here are, of course, the circus on sleds moves to a different place.
And here are 60 light sensors. You can see them here. And so you actually build the string in real time as it goes in the hole, as you will see from the next picture. So this is the cable that brings the high voltage down and brings the signals up. And every 17 meters, you attach one of these. Here is the cable. And so this is the last one, and that's the last time you see. It sinks to the bottom, and gets frozen in forever.
So if you could see the detector, you would see strings with 60 optical modules. And if you look 125 meters away, you would see another string. You do that 86 times, you feel a kilometer cube. You have IceCube. And at the surface, these are the cables that bring the signals back. They go in this tower, into this two-story building. And this two-story building inside is just computers.
And so, in case you didn't get it yet, this is a cosmic ray muon through IceCube. Each of the white dots is a photomultiplier. You see the Cherenkov light, and you see it light up the sensors. In these displays, colors is time, and the size is the number of photons detected by that particular photomultiplier.
This is the Hollywood movie. This is reality, kind of a simulation. That's for the 93 TeV muon looks like. You see this hair growing out. That's our photons scattering through the ice until they die. But the interesting thing is you see the light comes in blobs. And of course, this muon is not a minimum ionizing particle. What you are looking at is this muon catastrophically emitting energy, producing radiating photons, producing electron-positron pairs, occasionally interacting with a nucleus. And so it then produces particles in each of these interactions, which emit Cherenkov light, and that's the blobs you see.
And if you look at an event-- this is a kilometer-long track, and you see the photons, how they come in catastrophic energy loss. And we made the big breakthrough when we actually started to fit in detail these wave forms. And you can actually reproduce energy to 1%. In practice, we do 10%, but that's all systematics.
So in the end, we have a detector that detects neutrinos from 10 GeV to infinity, and it can reproduce a muon track to 0.3 degrees at high energy, and the energy resolution is typically better than 15%.
So a few more pictures. So what does a detector like this see? At the South Pole, of course, if you look not through earth, whereas backwards, you get cosmic ray muons in large numbers. You're a mile deep with a kilometer square detector. And then, the other thing you see is you see neutrinos produced in the atmosphere all over the earth.
And so what that means is-- I'm going to show you as a picture-- that's the detector taking data. And you see there's no detector actually. There are 5,260 modules that send you these signals. And then, the computer at the surface puts them together in events. And you see fitting muon tracks like crazy. This is a muon bundle. This movie repeats. It's 10 milliseconds long.
And so what it amounts to, if you run this detector, you see a hundred billion cosmic ray muons per year. This is after trigger, which is the requirement that eight photomultipliers send you a signal. We see 100,000 atmospheric neutrinos per year. We actually have a lower threshold detector at the bottom. It's called the core, which sees another 100,000 per year. And remember, we were looking at 10 to 100 events. Fortunately, the answer is closer to 100.
And so this is what, by the time the detector was finished, we had seen a few events like this. This reconstructs as an 89 TeV muon, and you see red to green to blue. So it comes through the earth. And that's what a 100 TeV muon looks like. We didn't know then, but this is almost certainly a cosmic neutrino, not an atmospheric one.
And so now you run your detector, and the story should have been simple. So after two years with the completed detector-- in fact, it was only one year with the completed detector-- this is the flux we saw in atmospheric neutrinos. Remember, we see actually one atmospheric neutrino every six minutes. And so you remove them the muons by direction. You keep your atmospheric neutrinos. And here, you see the flux, and the blue is the calculation, which we tested on a previous slide I've shown.
And so what we were supposed to see is the green. What we saw instead is this. And I am not going to explain, but this is basically a 100 TeV. And so you see the excursion there, which, after two years, was 3.7 sigma. And so you are exactly seeing this deviation that I predicted before.
So if you actually analyze the data we have on this, here is the present as of recently. This is the calculation. This is what we see. And in fact, that is now a six sigma effect. And if you fit this flux, this was number of events. It's was not multiplied by the E squared. So if you fit this flux, you get index of 2.06 plus or minus 0.13-- so exactly what you are looking for.
And here are a few of these events. They mostly come from the horizon actually, because these neutrinos have such high cross-section that they don't come through the earth anymore from below. So you see, that's what the PeV neutrino looks like. Of course, there's a subtlety. You only measure the energy in the detector. You don't know how much energy it was before and how much it's carrying out. With standard model physics, you can't reproduce the neutrino energy.
But this event, which I cannot resist showing you a movie of, this event actually deposits 2.6 PeV, 2,600 TeV, in the detector. So if you calculate the neutrino energy, it's 9,000 TeV. And you see it go through the detector. It came 11 degrees below the horizon.
And so, as I said, this is where the action is because these we can reconstruct to 0.3 degree at high energy, and so that's our opportunity really to do astronomy. But for those of you who follow this field, you know this is not how we discovered cosmic neutrinos. Actually, this was the confirmation I just showed you. This is how we were supposed to discover them.
Except, the following thing happened. Remember this slide? This slide, of course, we were looking for these events that you couldn't miss. If you had a direct hit of one of these events, it would fill the whole detector with light. So in the same two years of data I just showed the analysis through the earth, through going muons, we were looking for these big events. And I am not going to describe the analysis. You basically look for something that produces a ridiculous amount of light in the detector.
And this is what we saw. This event, you see, it's like it started on a string then went up and down, and it's just like someone turns on a light bulb in your detector. And if you look at this event, we knew that's not a 1 million TeV neutrino, but it's something we had never seen before. It's too big. In fact, it's 1,000 TeV neutrino. You say, where is the muon track? There is no muon track. This is an electron neutrino that interacts in the detector.
So what does an electron neutrino do? It gives 80% of its energy to an electron. The electron makes a shower, and the shower particles emit Cherenkov light. But this shower is actually less than 10 meters long. In a kilometer cube detector, that's a point source of light-- so just like someone turns on a light bulb. And if you have any doubts, this is what a muon track of the same energy looks like.
Now, how do you know if these events are totally spherically symmetric? So how do you measure their direction? And that's a subtle thing. So I have to remind you how big these events are. This is the Madison campus along the lake, and this is a direct hit on the data center in Madison. And so, if you've never been there, you should come sometimes. And so this event is like, in any place you live, six city blocks. And we know each photon to two nanoseconds. And that's the key.
So I'm going to show you a simulation of the event, and you will see how it's totally spherically symmetric, but notice the color, which is the time the photons arrive at the photomultipliers. There we go. You see it's yellow, red here. Now it's green, yellow there. So what it means is, if the neutrino came from here, the shower is small, but it has a direction, and the photons arrived there before they arrived there. And from that, you can reconstruct the direction.
And remember, this was absolutely crucial. When we made this proposal, many people said we were wasting funds by digitizing these signals. And in fact, thank god we did. So not only is this tells you how many photons, we know exactly their arrival times, and so you fit these photomultiplier wave outputs. And that allows us to reconstruct a direction. Not to 0.3 degrees. There's no reason we cannot reconstruct them better, but we are not there yet. It's not easy. It's work.
So you see, for instance here, this is just one optical module, and this is the light signal it detects. And you see if this is the right fit. If you got it wrong by 180 degrees, you get the red. And so, in these events, there are more than 100,000 photoelectrons. So probably 300 something pictures like this that you fit. So there's information.
But what I didn't say is these are 1,000 TeV neutrinos, in fact, which you can tell. Just the size tells you the energy. You don't need a computer. but. It's not a million TeV neutrino, so we didn't know what this was. And so there were two events in the sample. So there were 1,000 TeV, so unlikely to be atmospheric. There were two events.
And then, the other thing is notice, if this neutrino had been produced in a air shower-- there was an air shower 20 kilometers above your head in the atmosphere-- it produces pions and muons, and these muons reach your detector because this is a high-energy air shower. In fact, the muon that came with the neutrino from the pi/K decay should have ended up in your detector. These neutrinos came by themselves. They are not accompanied.
And if we had really pushed-- you know, if you have a marginal signal, you can turn it into 5 sigma, you can turn it into nothing. It's just up to you, and we turned this into nothing. But what we did was, of course, everybody knows what you do now. These are golden events because you measure the whole energy. You're sure what it is, so you go and look for more.
And so you just run the data through a filter, which I won't explain, but the basic idea is that you forget collecting muon outside your instrumented volume. You divide your detector in a veto shield and an active volume. And the veto sheet, the veto is about nine photomultipliers at the top and one on the bottom and the side. So it's not an enormous sacrifice. And so you just look for-- this is a muon track, and the muon track starts in the detector. And you look back and make sure there is no light coming in.
And then, it's really, this was incredible because you have to reject these enormous backgrounds, which is a headache of through-going muons. Then, you find the background of neutrinos. This was easy. You just plot the data, and you see the signal. So in this plot, you see that's the number of events for one year. And this is the number of photoelectrons. So this is energy from low to high. And this is how many photons there are in the veto region, which is supposed to be zero. And you see there is the signal. So you see no light in the veto region and very high energy events.
And so we have, by now, analyzed four years of data. We have two more we have about to on blind. All of this is done in blind analysis, but so what? So every year looks the same. I mean, statistics is not an issue in this business. So at that moment, we had 26 more events, and we published.
So in fact, in the third year, it took us a year to have the courage to write this paper. So by then, we had seen a third year of data in which we discovered an event that had twice the energy of the ones we discovered originally. So if I make the classic plot of number of events versus energy-- here in photoelectrons-- you see here, and this deviation now, after four years, is almost seven sigma.
And you remember this plot. So these events, the high energy events, are totally consistent with the flux we actually saw later in muons going through the detector.
So here is the summary of the evidence. So you must note that these two analyses, they use totally different techniques, different software. It's done by different graduate students, which is the key. The only thing these two techniques have in common is that there are photomultipliers in the ice. All the energy determination, everything else, how you cut the back hand out of your data, that's all different. So this is like CMS and [? Matplus, ?] not but cheaper. It's the same detector-- not as good.
In fact, if you don't sleep at night-- which I know everything about, thinking about what could go wrong-- this is an event where the muons start inside the detector. You cannot see this. So it starts here, goes straight down. If you look backwards, I didn't mention this, but above IceCube there is an air shower array. And so there is no air shower. You see a few hits in the detectors of the air shower array, but they all come at the wrong time. There is absolutely nothing in the detector at the time this muon went through. So this really didn't come from the atmosphere. In fact, as it says on my slide-- I never bothered calculating-- but it must be close to five sigma if it's not over it.
So critical question, where do they come from? So I already told you that the real action is with the first sample, where we reconstructed neutrino directions to a fraction of a degree. But let's look at this sample, which came first, and triggered a lot of speculation.
This is four years of data, events starting inside the detector. And you look at this. And you look at it long enough, and then you begin to see things. In fact, just as a warning about doing small statistics, this event is the first one we measured, and it came with one degree of the center of the galaxy. And I went home very depressed because I said they all come from the center of the galaxy and that's it.
Actually now, the conclusion is that there's no evidence that any of them come from our galaxy. But you see the blue, actually, is an analysis that looks for clusters in this map. And this map, by the way, this is the universe, except it's in a projection where this is the plane of our own galaxy. So if they come from our galaxy, they should be in this plane. This is the center. And of course, clearly, many are not from one galaxy.
If you look at this long enough, you may see some correlation here. You may see some clustering in here. In fact, there's a famous statement of Greisen that is, if you look at maps for too long, you see things. It's like seeing pictures in the flames of a fire. That was a Greisen quote which I always liked.
In fact, the best way is to look at it that way. This is two years of data. So if you work out the probability that this is a cluster of events, it's at a few percent level. Nothing at a few percent level is ever true. That's something I learned in life. And this is also at a few percent level. But you see, if I go backwards, you double the data, and the evidence doesn't grow. So you know the answer. This is an extragalactic flux.
So what about the correlation to the galaxy? Well, if you say that galaxy seven and a half degrees wide, which is possible, then the correlation is 2 and 1/2%. That's not good enough to be evidence. And in this 2 and 1/2%, I've taken into account that I tried different widths of galaxies. So it's a real 2 and 1/2%.
The other thing you may ask is, if they come from far away, they must come in equal numbers because-- never mind oscillations-- if you look at a neutrino beam, it equilibrates between electrons, muons, and tau neutrinos. So we determined that, too. And so this plot I could spend a lot of time on. So it's consistent with equal numbers of the three flavors. So this looks like an extragalactic flux, far away sources.
So that's the preliminary conclusion. We observe a diffuse flux of neutrinos from extragalactic sources. A subdominant galactic component cannot be excluded. We try very hard. I actually feel that we, at some point, should see galactic sources. I didn't go-- I could have given a few slides on what a signal is from galactic sources. And if anything, it's a better case than for the extragalactic one I discussed.
But now, this is the critical question. I told you that, for every PeV neutrino, there is PeV gamma ray. I showed this slide. And I also told you in the beginning, one of the first slides I showed showed that nobody has ever seen a PeV gamma ray. In fact, air shower arrays set very stringent limits on that flux that are lower than the flux we actually see.
And so the answer is not totally enigmatic. So if you produce these blue gamma rays for a neutrino-- by the way, when I say equal, there can be factors of 2. It depends, but roughly equal-- cannot be factors of 10. So what happens to these gamma rays is that, remember the universe is full of light. This was my first slide. So it's filled with microwave photons, and these gamma rays do the same thing that the protons did on my third slide. They interact with CMB photons, [? reproduce, ?] and lose their energy.
And that, of course, is a simple QED calculation, which anyone can do. So you take one gamma ray for each neutrino and calculate what comes out. And it turns out that these gamma rays end up with the energies of typically 100 GeV, not 1 PeV.
And so this is one example of such a calculation. This is the flux I showed, events starting in the detector pushed to lower energy. I'll come back to this. The blue is the flux seen in muons going through the earth. You see again, they are consistent.
Now, this is not a fit. You just draw an E to the minus 2.15, which is consistent with what we see between these two fluxes. In fact, if you did a fit, it wouldn't be very different. And these are where the gamma rays come around. And this is the extragalactic flux measured by the Fermi satellite. So these are the highest energy gamma rays in the sky. These are the highest energy neutrinos in the sky. And they are totally consistent coming from the same objects, which is a very intriguing result in the sense that astronomers have made a non-thermal universe out of electrons and photons. This tells you that there are hadronic accelerators that are responsible for the gamma rays seem by astronomers.
In fact, if you calculate the energy in high-energy cosmic rays that I showed at some point early in the talk, it's also the same energy. So we are seeing some sources that deposit the same energy in gamma rays, neutrinos, and cosmic rays-- [? photons ?] or whatever. And of course, that was kind of the rationale for building a kilometer cube detector. Such a flux you have to see, and we see it.
Can you avoid this conclusion? Well, suppose I fit the flux like this. You can actually make models that have a flux like this if you work hard at it. Then, in fact, the photons coming out are this, but they're still at the 10% level, the same sources. And you must admit that this doesn't fit the neutrino data. In fact, look at this point, how many sigma it is away-- from both, actually.
So this is the important conclusion so far. That is, I just said this. And so it looks like the non-thermal universe runs on hadronic accelerators, not on electronic accelerators. And at some level, we may actually be seeing the same sources.
So what are they seeing? Well, they didn't know either, but this created a lot of incentive to go and look at what Fermi were seeing. And the present conclusion is-- I have a slide on this, but we'll flip through it-- you just want [INAUDIBLE] blazars. What are blazars? Blazars are these active galaxies. Remember the picture. You had inflows and an outflow of a jet. And it's one of these objects where the jets are pointing at you. And most of the photons seen by Fermi, like more than 85%, is in these sources.
And we have looked for the sources, too, and we don't see them yet, but this is consistent. But if you were a betting person, this is not a bad bet, but it's far from proven.
Of course, you can reverse the gain and take the sources-- I won't spend time on this slide-- take the sources that Fermi sees and translate them back in neutrinos. And so we are at the verge of having to see the sources. But you know, it could be that we see the same energy but we don't see the same sources. After all, in neutrinos, you see sources back to the beginning of time. Whereas, in gamma rays, you have a horizon, as I've shown my previous slides.
But, there is more. And you see, this was the first attempt we made to push this analysis to low energy. We found the green here is a power fit to the data. And notice, it's a rather steep power to make it go high at low energy. It actually doesn't fit the high-energy point here. And notice, there is an excess here.
In fact, this is another way of looking at this. And every analysis-- there are many analyses in IceCube now-- every analysis sees this bump. In fact, the best way of seeing it, you fit the power to the data. And depending on the energy range you fit, you get a different power-- way beyond statistics. And so there is this excess.
So instead of going from a cosmic flux into an atmospheric flux, there is some excess sitting there. Maybe a cosmic accelerator looks like this. This is the first time we see one directly in neutrinos. In gamma rays, they are always distorted-- either in the source by interaction of the photons or even the propagation to Earth. This is the first time we see an accelerator. So who knows?
OK I think I said all this. So what's next? Well, this is all very nice, this multi-wavelength stuff and looking at gamma rays and cosmic rays combined. What you really want to do is see sources of neutrinos, which means you want to see two neutrinos from the same place, and then, you know what the accelerators are.
So how many events do you need? And that calculation you can do on the back of an envelope, and I have it here. There is the calculation. It is a back of an envelope, but the answer is 740 events to get doublets. And this is a pessimistic calculation. Of course, it depends on how dense the sources are-- how many are close to you, for instance. But for blazars, this is the right number. There are 10 to the minus 5 per megaparsec cubed. And so we need not tens of events, but hundreds of events.
Now, I have already shown you in this talk something like maybe 50. But to do this game, you don't have to have cosmic neutrinos you bet your house on. You can afford the background. And so in 10 years, we should get there with a relatively pure sample that's hundreds of events. Also, if you compare with the map, you need less events than to get your own doublets. But of course, it would be nice to get your own doublets.
So instead of collecting we are doing now something like 30 golden events per year, you want 300 golden events per year. So you want a 10 times bigger detector, which you think we cannot afford. Well, you're wrong. Because, while we were building IceCube, we spent years of our lives figuring out the optics of ice. I passed this over.
But this is result, one of the results. What you see here is the absorption length of blue light in this deep ice. And you see here the critical numbers. On average, it's 100 meters. The surface is there. This is the top, this is the bottom of the detector. If you go another 450 meters, you are at the bedrock at the South Pole.
So you see, this absorption length is 100 meters. And at the bottom of the detector, it's 220 average. There are places where it's 300. The absorption length of tap water is 2 meters. If you distill, it's 8 meters. The clearest water in the Super-Kamiokande detector in Japan has an absorption length of 80 meters. And so we didn't know this. We had suspicions.
But how far your light travels is how far you can space you photomultipliers, basically. By the way, this is 100,000-year-old ice, typically. And so these variations, there's a small amount of dust in this ice that's picked up when-- this is snow that condensed. Each year, there is like this much snow in this detector. But when the climate is dry, then, for instance, the winds that go around Antarctica can pick up some dust in South America, and then your optics is more [? true, ?] but not as good.
And so, in fact, we measure the optics of the ice. And if you go 1,000 kilometers away to Vostok Station where they take calls, they see exactly the same structures as we do.
Now, if you collect photons that far away, they begin to scatter on this dust, and you have to take that into account in your reconstructions. And so this is how well we know the dust by now. Look, this is not a kilometer. This is 130 meters to 170. So this is 40 meters of detector. It's like from here to the back of the room. And this is the concentration of dust measure with a technique that was invented by Buford Price from Berkeley.
And so I like to show this picture because you see this peak, it's less than a centimeter wide. It's a volcanic eruption, and it's a famous volcanic corruption. It's the Toba eruption, which happened 74,000 years ago. And it was not seen in the ice cores, but it was seen in the geological record. And this was a problem. Well, here it is. We resolved it. Always want to be a geologist.
So now, you get the idea. You can space theses detectors not by under 25 meters, by 250 or more. And so the bottom line is that, with 5,000 photomultipliers, you can fill 10 kilometer cube instead of 1 kilometer cube. Also, this detector was designed-- the proposal was submitted in '99, so there are all kind of new ideas to build better and smaller light collectors, like light collectors that are made out of many small BMTs and wavelength shifter. And you make them smaller so that it's easier to drill holes. All this is in progress. But this is the basic idea of why this is not going to be prohibitively expensive. That's the propaganda.
Now, I did not talk about-- and I always show this slide-- but I will be talking about this tomorrow. It's not so much fun as this subject. But we have thresholds now with 10 GeV, and we compete for atmospheric neutrino physics. We can detect a supernova explosion, which, if it ever happened, would be much more interesting than what I talked about today-- I think.
We look for everything that people look for-- dark matter monopoles, you can put your quark nuggets, et cetera, et cetera, strange dark matter. We look for electronvolt-mass sterile neutrinos, and we don't see them. Paper to come soon. And that is another future project.
So let me conclude. So we have now figured out how this worked. We have found something, but the exciting thing IceCube is that many of the analyses, we have only and analyzed one or two years of data because we spend all our time improving the techniques. In blind analysis, you don't want to go on blind seven years of data before you have done your homework.
So none of these analyses is in the square root of time regime, and so there are a lot of results to come. So we do want to move to the next generation detector. This will be a detector that's better at high energy and at low energies. And there are two similar projects, one in the Mediterranean and one in Lake Baikal, that have been kind of restarted given that we found the flux. And of course, you really do these experiments not for what you know you're going to find, but to discover something. And as you know, neutrinos are never boring.
It's a good place to stop. Thank you.
SPEAKER 1: Questions?
AUDIENCE: I have a question. You mentioned that you can distinguish between tau neutrinos, muon neutrinos, and electron neutrinos. You didn't talk about the taus so much?
FRANCIS HALZEN: Actually, if you had looked carefully, we can't. So what we can tell is a shower from a track. So what you really measure well is new E plus new tau versus new mu. Now, a new tau, you saw all these pictures of taus and TeV neutrinos. The taus and TeV, the lifetime of a tau is 50 meters, and we can measure this. We've just been very unlucky. What it would mean is that you let the new tau come in. It interacts, makes a shower. But then, it makes a tau that travels 50 meters and makes another shower. And so you try to find these 50 meters in the light deposited.
And the problem we have, all the good events we have at high energy, they are all happening close to a string so that the photo tubes that have all the useful information are saturated. And so we haven't gone back to worry about this. And of course, the event that is a 10 PeV neutrino, that is a new mu. So we just have to get a good event, and we'll see new taus.
Also, you could separate them on a statistical basis, and we are at a level of having to do to see events. So somehow, the next analysis some graduate students do, we probably can make the separation, but we haven't done it.
AUDIENCE: Is there a maximum energy you could spend for the blocked sources. If you didn't have a chance to talk about it?
FRANCIS HALZEN: Well, cosmic ray physics is-- this is recorded.
Cosmic ray physics is very complicated. And so one of the things that I didn't touch on-- I have a slide which is blacked out-- is these PeV neutrinos are produced by cosmic rays that have an energy of 10 to the 16, 10 to the 17 electronvolt. Now, when I was young, that were galactic sources for sure. This would suggest they are not. But now, everybody's happy they are not.
So I think, in fact, around 10 to the 17, the array that's on top of IceCube discovered a very interesting structure in the cosmic ray spectrum, which could very well be the transition to extragalactic. But these spectrum people say, look at this beautiful power. Whenever you build an experiment that looks in detail in one energy region, you find all kind of structure. So this is infinitely complicated.
AUDIENCE: One of your sources was [? Emenga, ?] which is a neutron star?
FRANCIS HALZEN: Among these sources is anything. What this map was is you take the most prominent sources, and you take their gamma ray flux and translate it to neutrinos. If these gamma rays are made by electromagnetic processes, this is a meaningless calculation. So I wouldn't bet on any of these sources. I would bet on some of them. Some of them are molecular clouds emitting gamma rays, and that must be made by cosmic rays interacting.
And so there are some very close in the Cygnus region, and I have been spending a lot of my own time trying to find those. I think eventually we have to get them. But the Cygnus region has molecular clouds, but it doesn't have many supernovae. So maybe the accelerators haven't turned on yet. But we'll see. So you cannot overrate this math.
We at least know that there are a lot of sources that do emit neutrinos. We don't know in our galaxy. That's what I meant when I said I have the feeling we'll see our galaxy at some point.
SPEAKER 1: All right. If there are no more questions?
FRANCIS HALZEN: Thank you very much.
We've received your request
You will be notified by email when the transcript and captions are available. The process may take up to 5 business days. Please contact email@example.com if you have any questions about this request.
The IceCube project has transformed one cubic kilometer of natural Antarctic ice into a neutrino detector. The instrument detects more than 100,000 neutrinos per year in the GeV to PeV energy range, among these a recent flux of high-energy cosmic neutrinos. The high cosmic neutrino flux observed indicates that a significant fraction of the radiation in the non-thermal universe, powered by compact objects from neutron stars to supermassive black holes, is generated by proton accelerators.
Bethe lecturer Francis Halzen, Gregory Breit Professor and Hilldale Professor of Physics at UW-Madison and the principal investigator of IceCube, discusses the instrument, the analysis of the data, and the significance of the discovery of cosmic neutrinos, March 21, 2016 as part of the Department of Physics colloquium series.