jump to navigation

CMS and extensive air showers: ideas for an experiment February 6, 2009

Posted by dorigo in astronomy, cosmology, physics, science.
Tags: , , , , , , ,

The paper by Thomas Gehrmann and collaborators I cited a few days ago has inspired me to have a closer look at the problem of understanding the features of extensive air showers – the phenomenon of a localized stream of high-energy cosmic rays originated by the incidence on the upper atmosphere of a very energetic proton or light nucleus.

Layman facts about cosmic rays

While the topic of cosmic rays, their sources, and their study is largely terra incognita to me -I only know the very basic facts, having learned them like most of you from popularization magazines-, I do know that a few of their features are not too well understood as of yet. Let me mention only a few issues below, with no fear of being shown how ignorant on the topic I am:

  • The highest-energy cosmic rays have no clear explanation in terms of their origin. A few events with energy exceeding $10^{20} eV$ have been recorded by at least a couple of experiments, and they are the subject of an extensive investigation by the Pierre Auger observatory.
  • There are a number of anomalies on their composition, their energy spectrum, the composition of the showers they develop. The data from PAMELA and ATIC are just two recent examples of things we do not understand well, and which might have an exotic explanation.
  • While models of their formation suppose that only light nuclei -iron at most- are composing the flux of primary hadrons, some data (for instance this study by the Delphi collaboration) seems to imply otherwise.

The paper by Gehrmann addresses in particular the latter point. There appears to be a failure in our ability to describe the development of air showers producing very large number of muons, and this failure might be due to modeling uncertainties, heavy nuclei as primaries, or the creation of exotic particles with muonic decay, in decreasing order of likelihood. For sure, if an exotic particle like the 300 GeV one hypothesized in the interpretation paper produced by the authors of the CDF study of multi-muon events (see the tag cloud on the right column for an extensive review of that result) existed, the Tevatron would not be the only place to find it: high-energy cosmic rays would produce it in sizable amounts, and the observed multi-muon signature from its decay in the atmosphere might end up showing in those air showers as well!

Mind you, large numbers of muons are by no means a surprising phenomenon in high-energy cosmic ray showers. What happens is that a hadronic collision between the primary hadron and a nucleus of nitrogen or oxygen in the upper atmosphere creates dozens of secondary light hadrons. These in turn hit other nuclei, and the developing hadronic shower progresses until the hadrons fall below the energy required to create more secondaries. The created hadrons then decay, and in particular K^+ \to \mu^+ \nu_{\mu}, \pi^+ \to \mu^+ \nu_{\mu} decays will create a lot of muons.

Muons have a lifetime of two microseconds, and if they are energetic enough, they can travel many kilometers, reaching the ground and whatever detector we set there. In addition, muons are very penetrating: a muon needs just 52 GeV of energy to make it 100 meters underground, through the rock lying on top of the CERN detectors. Of course, air  showers include not just muons, but electrons, neutrinos, and photons, plus protons and other hadronic particles. But none of these particles, except neutrinos, can make it deep underground. And neutrinos pass through unseen…

Now, if one reads the Delphi publication, as well as information from other experiments which have studied high-multiplicity cosmic-ray showers, one learns a few interesting facts. Delphi found a large number of events with so many muon tracks that they could not even count them! In a few cases, they could just quote a lower limit on the number of muons crossing the detector volume. One such event is shown on the picture on the right: they infer that an air shower passed through the detector by observing voids in the distribution of hits!

The number of muons seen underground is an excellent estimator of the energy of the primary cosmic ray, as the Kascade collaboration result shown on the left shows (on the abscissa is the logarithm of the energy of the primary cosmic ray, and on the y axis the number of muons per square meter measured by the detector). But to compute energy and composition of cosmic rays from the characteristics we observe on the ground, we need detailed simulations of the mechanisms creating the shower -and these simulations require an understanding of the physical processes at the basis of the productions of secondaries, which are known only to a certain degree. I will get back to this point, but here I just mean to point out that a detector measuring the number of muons gets an estimate of the energy of the primary nucleus. The energy, but not the species!

As I was mentioning, the Delphi data (and that of other experiments, too) showed that there are too many high-muon-multiplicity showers. The graph on the right shows the observed excess at very high muon multiplicities (the points on the very right of the graph). This is a 3-sigma effect, and it might be caused by modeling uncertainties, but it might also mean that we do not understand the composition of the primary cosmic rays: yes, because if a heavier nucleus has a given energy, it usually produces more muons than a lighter one.

The modeling uncertainties are due to the fact that the very forward production of hadrons in a nucleus-nucleus collision is governed by QCD at very small energy scales, where we cannot calculate the theory to a good approximation. So, we cannot really compute with the precision we would like how likely it is that a 1,000,000-TeV proton, say, produces a forward-going 1-TeV proton in the collision with a nucleus of the atmosphere. The energy distribution of secondaries produced forwards is not so well-known, that is. And this reflects in the uncertainty on the shower composition.

Enter CMS

Now, what does CMS have to do with all the above ? Well. For one thing, last summer the detector was turned on in the underground cavern at Point 5 of LHC, and it collected 300 million cosmic-ray events. This is a huge data sample, warranted by the large extension of the detector, and the beautiful working of its muon chambers (which, by the way, have been designed by physicists of Padova University!).  Such a large dataset already includes very high-multiplicity muon showers, and some of my collaborators are busy analyzing that gold mine. Measurements of the cosmic ray properties are ongoing.

One might hope that the collection of cosmic rays will continue even after the LHC  is turned on. I believe it will, but only during the short periods when there is no beam circulating in the machine. The cosmic-ray data thus collected is typically used to keep the system “warm” while waiting for more proton-proton collisions, but it will not be a orders-of-magnitude increase in statistics with respect to what has been already collected last summer.

The CMS cosmic-ray data can indeed provide an estimate of several characteristics of the air showers, but it will not be capable of providing results qualitatively different from the findings of Delphi -although, of course, it might provide a confirmation of simulations, disproving the excess observed by that experiment. The problem is that very energetic events are rare -so one must actively pursue them, rather than turning on the cosmic ray data collection when not in collider mode. But there is one further important point: since only muons are detected, one cannot really understand whether the simulation is tuned correctly, and one cannot achieve a critical additional information: the amount of energy that the shower produced in the form of electrons and photons.

The electron- and photon-component of the air shower is a good discriminant of the nucleus which produced the primary interaction, as the plot on the right shows. It in fact is a crucial information to rule out the presence of nuclei heavier than iron, or the composition of primaries in terms of light nuclei. Since the number of muons in high-multiplicity showers is connected to the nuclear species as well, by determining both quantities one would really be able to understand what is going on. [In the plot, the quantity Y is plotted as a function of the primary cosmic ray energy. Y is the ratio between the logarithm of the number of detected muons and electrons. You can observe that Y is higher for iron-induced showers (the full black squares)].

Idea for a new experiment

The idea is thus already there, if you can add one plus one. CMS is underground. We need a detector at ground level to be sensitive to the “soft” component of the air shower- the one due to electrons and photons, which cannot punch through more than a meter of rock. So we may take a certain number of scintillation counters, layered alternated with lead sheets, all sitting on top of a thicker set of lead bricks, underneath which we may set some drift tubes or, even better, resistive plate chambers.

We can build a 20- to 50-square meter detector this way with a relatively small amount of money, since the technology is really simple and we can even scavenge material here and there (for instance, we can use spare chambers for the CMS experiment!). Then, we just build a simple logic of coincidences between the resistive plate chambers, imposing that several parts of our array fires together at the passage of many muons, and send the triggering signal 100 meters down, where CMS may be receiving a “auto-accept” to read out the event regardless of the presence of a collision in the detector.

The latter is the most complicated thing to do of the whole idea: to modify existing things is always harder than to create new ones. But it should not be too hard to read out CMS parasitically, and collect at very low frequency those high-multiplicity showers. Then, the readout of the ground-based electromagnetic calorimeter should provide us with an estimate of the (local) electron-to-muon ratio, which is what we know to determine the weight of the primary nucleus.

If the above sounds confusing, it is entirely my fault: I have dumped here some loose ideas, with the aim of coming back here when I need them. After all, this is a log. a Web log, but always a log of my ideas… But I wish to investigate more on the feasibility of this project. Indeed, CMS will for sure pursue cosmic-ray measurements with the 300M events it has already collected. And CMS does have spare muon chambers. And CMS does have plans of storing them at Point 5… Why not just power them up and build a poor man’s trigger ? A calorimeter might come later…


1. Daniel de França MTd2 - February 6, 2009

If we had 1 million CMS on the moon, a really huge CMS, or something like that, could we use it like a super ultra LHC?

2. dorigo - February 6, 2009

How Daniel ? You mean as a cosmic ray telescope, by making a coincidence between earth and moon ? Well, that’d be hard, since we do not really know the magnetic field in the path.

But maybe I misunderstood you…

3. Daniel de França MTd2 - February 6, 2009

Not really. What I meant is something that I thought for so many years, but is increadibly naive. In the moon, the cosmic rays directly hits the ground, so, why not making some kind of target above the ground, made of an homogenous material, convering a great area. Put the detector right under the target. Wouldn’t that be like working with an accelarator 10 million times more powerful than LHC?

4. dorigo - February 7, 2009

I see. No, it would not. The energy of a 10^20 eV cosmic ray proton hitting another proton equates that of a collider with a c.m. energy of s = 2 m_p E, where m_p is the target mass and E is the incident energy. So \sqrt s = 4 \times 10^{14} eV, which is an energy just 30 times the LHC. However, such events would be very rare -one per year for a huge setup- and they would just show “simple” QCD processes.

The LHC is terribly powerful because its beams are intense. Only with intense beams can one probe phenomena whose cross section is tiny. Of course, a larger energy makes up for the intensity somewhat, because the higher the energy, the larger the cross section for rare processes: this is a rule of thumb which holds true for most things we care to measure. But between one event per year and a hundred billion events per year, I would choose the latter, even if the energy is 30 times smaller!


5. Luboš Motl - February 7, 2009

What if Daniel wanted to locate two high-energy cosmic quanta and direct them towards a collision? Or maybe ten of them? 😉

6. dorigo - February 7, 2009

Yeah Lubos, a gang-bang of cosmic quanta 🙂

7. Daniel de França MTd2 - February 7, 2009

But given that most of the energy-momentum is intact, wouldn’t it justify the experiment? Every such cosmic ray would give rise to ~30.000 above from 500 to 100TeV, if the detator were build with a lot of layers.

8. Daniel de França MTd2 - February 7, 2009

But given that most of the energy-momentum is intact, wouldn’t it justify the experiment? Every such cosmic ray would give rise to ~30.000 above from 500 to 100TeV, if there was a lot of layers.

9. dorigo - February 7, 2009

Daniel, collecting a handful of very high-energy cosmic rays -not knowing their composition, their angle of incidence, etcetera by the way- does not justify an experiment on the moon, no. An experiment must have a clear reason to be built. What would be the reason for studying very high-energy cosmic rays on the moon ? Of course cosmic rays are interesting in their own right, but we can study them on the Earth or in space, and we do, with several designs. Each and every design has a clear physics or astrophysics case behind its development. A scientific scope.
A large array on the moon may one day be built, but it seems very far-fetched, and in any case its interest for particle physics proper is close to zero, because most of those energetic interactions would be simple QCD, and we would not learn much we do not know yet. The interest of making CMS a cosmic muon detector is due to its scientific output/ cost ratio close to infinity, due to the fact that the cost is really small.


10. Daniel de França MTd2 - February 7, 2009

“An experiment must have a clear reason to be built..What would be the reason for studying very high-energy cosmic rays on the moon ?”
But you said: knowing their composition

And besides, I am not arguing against making a CMS a cosmic muon detector.

But is it really farfeched to build large detector on the moon, or floating on high atmosphhere, or orbiting earth?

dorigo - February 8, 2009

Well Daniel, knowing the composition is important, but it can be done from space too. The problem is a bit more complex though: the idea is to obtain the primary composition in correlation with the muon multiplicity underground, because the anomalous large number o muons in energetic air showers could be explained by nuclei heavier than iron, or by our incorrect modeling.

I believe we do not have the moneys needed for such a huge technological effort. Bringing stuff in space costs a fortune: even the most advanced spectrometers sent in orbit are lightweight. To go to the moon WRT in orbit would have the only advantage of providing support for large structures, which we cannot afford.


11. carlbrannen - February 8, 2009

The only reason for doing it on the moon, or in space, would be to get a better idea what is happening at the primary vertex. That is the position of maximum weirdness.

There are two places where this sort of high strangeness has been seen, on the earth’s surface, at high mountains, one puts sheets of x-ray photographic emulsion in between sheets of lead. Every now and then one gets a cosmic ray that makes it most of the way through the atmosphere and one is left looking at the early shower. The other place is in the balloon borne emulsion experiments. These are much lighter, are run for shorter time periods, but some seem to have caught the vertex of cosmic rays themselves. Very strange.

12. Daniel de França MTd2 - February 8, 2009

I know this is prohibitely expensive, but it is something to begin to think about. For example, when a particle accelator with up to 800TeV will be built? This seems to be so far into the future that it seems to be easier to think that a mass production of parts on earth and its self assemble of repetitive partsn space can be achieved, this is unlike the ISS, which is highly heterogenous.

I know of 2 examples, that combined, can make one think about such case. The OWL concept for large telescopes
A mini case in space of a simple case of self assembling is the james webb telescope http://www.jwst.nasa.gov/.

I don’t know if it is possible with any kind of detectors.

13. Daniel de França MTd2 - February 8, 2009

Hi Carl,
Indeed, I’ve been thinking about the primary vertex. Take a 1,0e20, cited above. In the CMS reference, considering particles with the same mass, it’s like colliding 2 particles at 150 TeV. By itself, it might give rise to a lot of particles or not. If a lot of particles appear, the penetration in the atmosphere will be low, because there will be a lot of particles sharinng 1,0e20 mass energy in the earth reference. If it is a low number,the penetration will be deeper, but would the detection signal be smaller?

14. dorigo - February 8, 2009

Daniel, the important thing you insist neglecting is the rate. We do not learn anything by collecting a few 10^20 eV events, because all of them are just good-old QCD scatterings. The interest of these cosmic rays is in their production processes, not in the collisions they have a chance to create.

To put it in a more concrete way: 10^20 eV means a center-of-mass energy of 400 TeV. Now, take ten such events (as many as one could think of collecting in several years with a huge detector), and imagine they are all protons (it only makes things worse if they are heavy nuclei, because the energy available is smaller in that case, since one has to rescale down by the number of nucleons): the chance that none of the parton collisions has a total energy above 4 TeV is very large!

In other words, a billion LHC collisions have a chance of reaching higher in energy than a handful of the highest-energy cosmic ray interactions.

Also note that, even if one had 40 TeV c.m. interactions aplenty, most of them would still consist in good-old QCD processes, from which we would not learn a darn thing.

You might argue that the energy of the primary does not get all lost in the first interaction: if one interaction makes 4 TeV, the other 396 all go into the kinetic energy of the remnants from the collision. However, to study in depth the shower, the detector would need to have a thickness of dozens of meters. This, for an area of a square kilometer, is beyond anything I wish to even start conceiving. There is no water on the moon, so a cheap detector volume is not feasible; there is no gas either, so cherenkov radiation is not conceivable.

It simply seems impossible to do this. Maybe a 100 square meter thing could be done, but it would not add much information to what we have in orbit today…


15. Daniel de França MTd2 - February 9, 2009

I am not ignoring the rate… I was thinking in something like about 100km^2…

16. Daniel de França MTd2 - February 9, 2009

“There is no water on the moon”-> What about Europa? A lot of ice in the vaccum.

I must confess that I am starting to feel bored about fundamental physics. There could be so much more funding in particle physics, and in the theoretical side, nearly nothing newer to be explained that couldnt be by the old SM.

17. dorigo - February 9, 2009

Ok, if your proposal is to build a 100 km^2 detector on the moon, well -I might just as well say, “oh yeah, sure, great idea, why not?”. But I guess it would mean to disrespect you. I rather prefer to say there is a zillion better ways to invest that amount of money, human effort, human lives.

If, one day a long time from now, computers will be capable of mass-producing robots and sending them to the moon, where they can build stuff with the material mined there, then a 100 km^2 detector will have a chance to see the light. I doubt there would be anybody left to care, though. And I insist, the particle physics output you would extract from it would be rather poor, whatever the design. With a fraction of that effort you could rather build an accelerator in orbit, as large as you want it…


18. Daniel de França MTd2 - February 9, 2009

I completely agree with you, Tommaso. @16, I merely said I wasn’t ignoring the the rate. In my second post, I confessed that this insane idea comes from my boredom with everything. I know it is impractical.

So bored… 😦 . I wish I could find some motivation right now.

19. msantander - February 10, 2009

The flux of cosmic rays near the GZK cutoff is of the order of 1 part km^-2 century^-1, so even with a 100 km^2 detector you would only get 1 particle/year…

dorigo - February 10, 2009

right santander, for the very highest-energy ones the flux is too small for any conceivable use other than counting and source identification.


20. Daniel de França MTd2 - February 10, 2009

Hmm, thank you msantander. I didn’t know it was that small. It’s not even fit for sci-fi stories.

Sorry comments are closed for this entry

%d bloggers like this: