Faster than light neutrinos : What’s going on? Part II

The neutrino baseline between CERN and Gran Sasso

In the first part of this post, I explained why it is such a big deal to discover that the speed of light is not an unbreakable barrier. Now, I go into the details of the measurement conducted by the OPERA collaboration. I should say first that I was wrong when I said that they measured the production time of the neutrinos by estimating their decay positions by looking at the incoming meson and outgoing muon. I will clarify in an instant.

How did they do it?

Let’s start with the measurement of the distance between the place where the neutrino beam is generated and the OPERA detector, simply because it is the easiest to explain in (mostly) non-technical terms. I have to say first that when particle physicists design particle physics apparatus such as detectors and accelerators, the precision with which they position the different parts of these systems has to be very, very high. An impressive figure is about the 27 km tunnel that houses the Large Hadron Collider, which has been dug within millimeters of its design dimensions.

The neutrinos are generated at the CERN Neutrino beam to Gran Sasso (CNGS). The dimensions of the CNGS are known to millimeter accuracy. Couple this with an ultra-precise GPS antenna on-site, and you can fix every part of the CNGS to 2cm accuracy on a global geodesy reference frame called ETRF2000.

You can do something similar at Gran Sasso where the OPERA detector is but there is a complication: the detector is very deep underground. The GPS signal just doesn’t get through to exactly where the detector is. They determine the position of external landmarks like the entrance and exit of the nearby Gran Sasso highway tunnel. Using some knowledge of the position of the OPERA detector with respect to the tunnel, they can pinpoint the detector’s position on ETRF2000 within 20cm accuracy.

These GPS antennas that they use are maintained in place to monitor the Earth’s crust movements, since the measured distance changes due to continental drift. This is truly spectacular. On the next graph, you can see the 3-dimensional displacement of the GPS antenna at the Gran Sasso laboratory over the last three years. The jump you see in the middle corresponds to an earthquake that happened in the L’Aquila region in 2009.

From arXiv:1109.4897v1

All in all, they measure the distance between the start point at CERN and the end point at Gran Sasso to be  (731278.0 ± 0.2) m.

But which start point do they take? For the end point, it is rather easy, it is where and when you see neutrinos in the OPERA detector. But for the starting point, it is not so easy. It turned out that the most precise way of measuring the start time was to monitor the protons even before they hit the target to create the mesons (which would later decay into neutrinos). They cannot identify precisely which proton made which neutrino, so they cannot use the technique I proposed in part I.

Protons are charged and they are produced in very dense bunches by the Super Proton Synchrotron (SPS) at CERN. When such a tight bunch of charges travels around at the speed of light, it create magnetic disturbances. These disturbances can be picked up by a Beam Current Transformer (BCT), which can measure the shape and timing of the proton bunch train to a very, very high resolution. The next plot shows the kind of signature one of these bunch trains has:

From ArXiv:1109.4897v1

This is what is used as the start time. Now, you may say that the protons – and the mesons produced after the protons hit the target – don’t have the same mass as the neutrinos, so this should be taken into account. The stuff that is travelling between the measured start time and the end time are at some point protons, at some point mesons, and at some point neutrinos. This should be taken into account because the speed changes with the mass (to conserve momentum and see comment by suvayu below). Indeed, it is taken into account, and it has been found to introduce an error of only 0.2 nanoseconds so it can be safely ignored.

A more complicated issue is how do you synchronize your clock at CERN and your other clock at Gran Sasso? This is where the super-precise GPS antennas that we have been talking about before come to the rescue (again). The two antennas at CERN and at Gran Sasso are identical, and coupled to ultra-precise atomic clocks. They have been synchronized independently by two national (Swiss and German) metrology institutes, and the difference in the clocks at CERN and at Gran Sasso has been found to be about 3 nanoseconds.

Of course, you need to link the start time measured by the BCT and the stop time measured by OPERA to these precisely calibrated clocks. You still have a lot of stuff to synchronize, and many systematic delays to take into account. These details are a little too much to go into here, but they are fascinating. If you aren’t afraid of all the technicalities of controlling the timing of complex systems, have a look at the paper. I am of the opinion that it is actually a bit complicated (although I wouldn’t know how to make it simpler), and it could easily be where an error slipped by. A new experiment dedicated to do the same neutrino speed measurement should be careful to make the timing measurement as simple and as easy to understand as possible.

Now, let’s get into the more subtle aspects of the statistical analysis. Remember when I said that neutrinos are extremely hard to detect? Well, it can obviously happen that you send a complete stream of proton bunches to OPERA without seeing a single neutrino. So you need to look at it the other way around. Every time that you do detect a neutrino, you can associate it with a BCT signal from CERN. They managed to gather 16000 neutrino signals in a period of three years, yielding a very small statistical uncertainty.

Now, if you detect a neutrino, you can start by assuming that the neutrino got into your detector by travelling at the speed of light. Given the accuracy of your timing system, you can use this assumption to find the position of the proton in the original bunch train at CERN that made the neutrino you detected. If you do this for all the neutrinos you detect, you will find that some neutrinos appear to have originated before the entirety of the bunch train even passed through the BCT, and you will find that none of them appear to come from the last part of the bunch train. Given that they have 16000 signal neutrinos, they have lot of neutrinos that come from anywhere in the bunch train, and the effect is very easy to see. The interpretation is that your initial assumption – that the neutrinos go at the speed of light – is wrong.

Why is this serious science?

The speed of light barrier explains so much in physics that at this point, Occam’s Razor dictates that we treat the experimental result with large amounts of skepticism. The first people to apply this rule have been the members of the OPERA collaboration themselves. They have been going over their experimental procedures for many months before we were even aware of the result. 200 very brilliant scientists couldn’t find the systematic effect that would be responsible for their disturbing result.

The next step was to call for outside help, and this is what they did. Of course, calling for outside help comes at a price. It requires you to announce your results to the world, which is how the media fire surrounding this story was sparked. The day after the OPERA collaboration announced their result, they held a seminar at CERN with the express purpose of inviting other experts to scrutinize their work.

I attended the seminar (via the CERN webcast), and I was floored by the discussion at the end. I was expecting people to ask very difficult experimental questions that Antonio Ereditato – the spokesperson of the OPERA collaboration – would have trouble answering, but no. All questions of the type “Did you consider X?” were answered by “It was found to be a negligible effect” or “It has been corrected for”. Remember that the people asking these questions were world-class experts with a thirst for blood. They genuinely wanted to bring down this very disturbing measurement. Nobody could find the culprit.

I am still filled with doubt concerning the result. Whatever it is that the OPERA people forgot in their measurement is extremely subtle. At this point, our best hope to catch the mistake is to design an experiment with the express purpose of measuring the neutrino time of flight. This new experiment should try to keep all of its aspects as simple and as easy to control as possible. It should also try to beat OPERA in precision. Only such an experiment will be able to shed light on this mystery.

What does it mean for the future of physics?

You may have heard of the SN1987A supernova result, that confirmed that neutrinos do travel at the speed of light. Neutrinos from the explosion were observed just as the supernova itself was noticed. The extremely large distance should have amplified greatly any difference between the time-of-flight of the light and of the neutrinos. No difference was seen. There are a number of differences in the circumstances that could explain the disagreement between this observation and the OPERA result, but we need to enter the realm of speculation. I will indulge myself, but it is still way too early to attempt any kind of interpretation. For example, the neutrinos from SN1987A had an energy about 1000 smaller than the neutrinos at OPERA. But OPERA didn’t detect any sign of an energy dependence to the effect within their uncertainty. Although detecting an energy dependence was really stretching the precision OPERA was capable of. Another factor could be the nature of the neutrinos (were they electron neutrinos, muon neutrinos, anti-neutrinos?).

If the measurement is confirmed, this does not mean that all physics will fall apart. However, we will have to rethink an awful lot of it. The situation will be similar to when quantum mechanics superseded classical mechanics. Classical mechanics became an approximation of quantum mechanics for really large systems. We live in a world of very large systems, so most of the time, we don’t notice quantum effects. Similarly, our current quantum field theory and general relativity will become approximations to a greater theory, in which the light-speed barrier can be broken. However, we are far from being there yet. Pace yourselves, young neutrinos!

This entry was posted in Science and tagged , , , , . Bookmark the permalink.

5 Responses to Faster than light neutrinos : What’s going on? Part II

  1. Gerhard Blab says:

    Colour me doubtful. I still think there is something wrong with their baseline…
    Until proven otherwise, we better keep obeying the c-speedlimit 😉

  2. suvayu says:


    The stuff that is travelling between the measured start time and the end
    time are at some point protons, at some point mesons, and at some point
    neutrinos. This should be taken into account because the speed changes
    with the mass (to conserve momentum).

    I don’t think you can call that conservation of momentum since there are
    two decays involved in the chain. It might be better to call it a correction for loss
    of energy/mass, also the protons are travelling below c to begin with.

    • suvayu says:

      Oh forgot mention, its a damn good post. 🙂

    • Michel says:

      You’re absolutely right! Ah, the things you can overlook sometimes… Anyway, the OPERA people did this one right: I just didn’t describe it correctly. They did the full simulation of the decay chain and looked at the delay introduced opposed to the “neutrinos all the way” assumption.

  3. Michel says:

    Here’s an idea: install two dedicated optical fibers that goes from CERN to Gran Sasso, with a difference in length. When a neutrino is detected at Gran Sasso, send a signal back to CERN so that the clock at CERN can associate a time stamp to the neutrino detection. Measure the delay in the optical fiber that you use to send back this signal by using the other optical fiber, using this sum-difference technique they have been using. Then, you get completely rid of the need to synchronize two clocks, since you only use the one at CERN. You only need to control for the delays in the electronics at Gran Sasso, which is being currently done anyway. That would probably bust a few of their systematics.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s