Final Pieces of the CMS Puzzle

By John Conway | July 31, 2008 5:51 pm

The last and arguably highest-tech detector elements are, this week, being installed into the giant CMS experiment at CERN: the pixel detectors. After these detectors are installed, there remains only the beam conditions monitor, a small device, and then the experiment can be buttoned up in anticipation of the first circulating proton beams, hopefully in August. Nearly the entire LHC machine is cold – superconducting cold – and so at long last it seems that we may soon see the first data. Rumor has it that there can be first protons circulating by August 9 (a week from Saturday!) but I bet it will take a bit longer. There will be a many-week shakedown process before “ramping” the beams to high energy. This year, if all goes well, it it foreseen to ramp to 10 TeV total collision energy; the design energy is 14 TeV and that will happen next year. (A TeV is a trillion electron volts, an energy equivalent to about a thousand proton masses.)

The pixel detectors are the innermost devices in CMS, and are the first to record the passage of high energy charged particles which emerge from the proton-proton collisions. The central driving idea of these devices is to record tiny three-dimensional space points along charge particle paths, allowing us to measure to within 10 microns (10 millionths of a meter; a human hair is 50 microns in diameter) the trajectory of the charged particles, and thereby infer where in space they may have emanated from.

This is particularly important information. The LHC machine has many “bunches” of protons in each counter-rotating beam, and each bunch is spread out over a length of around 8 cm. Every time bunches collide (and that will eventually be every 25 nanoseconds) we will get many proton-proton collisions. In all likelihood only one of these will be of interest for later analysis; we need to identify which particles come from that collision. The pixel detectors will help us pinpoint that location in space.

But perhaps of even more importance is to know when some of the particles appear to come from somewhere other than the “primary vertex” where the collision actually happened. The presence of these “secondary vertices” tell us that some particle travelled a distance and then decayed. In the case of a high energy bottom (b) quark, it can travel several millimeters or even centimeters and then decay into several charged particles. The presence of a b quark “jet” is often a good indicator of whether there were top (t) quarks, the heaviest of them all, produced in the event. There is a ton of physics, including searches for new physics beyond the Standard Model, that relies on these abilities of the experiment.

If we could strip away all the support frames, cooling, electronics, etc. from the pixel detectors, leaving only the detectors themselves, they would have an arrangement sort of like the diagram at right.

As you can see, there is a central “barrel” portion, and two “forward disks”. The detectors themselves are rectangular, and, as the name implies, segmented into very small pixels about a tenth of a millimeter in size. That’s a lot larger than the pixel size in your digital camera. But this detector can take 40 million pictures a second, keeping the interesting ones and discarding the vast majority.

The heart of the pixel detector is the readout chip, a silicon microchip specifically designed and fabricated for this detector, in this experiment. The effort to develop the readout chip was led by Roland Horisberger of the Paul Sherer Institut in Villigen, Switzerland. Each chip has over 4000 input channels arranged in a grid; each channel is bump-bonded to a sensor channel. The sensors are also very thin silicon wafers with one surface segmented into pixels. Each pixel channel can sense when a certain minimum amount of charge has been deposited by a passing charged particle, digitizes and time-stamps it, and sends it out onto the readout bus when a trigger signal matching the time stamp is received. All the thousands of readout chips in the detector do this in parallel, ultimately sending the torrent of data out on optical fibers to data acquisition electronics modules in the service cavern adjacent to the main detector cavern.

The PSI group built the central barrel portion of the CMS pixel detector, and the forward disks, which are somewhat more complicated mechanically, were built by a consortium of US universities and Fermilab. The forward disk detectors were assembled at Fermilab and then transported to CERN for final assembly, testing, and now installation.

My own involvement in the project has been varied, but most recently focused on getting the detectors to CERN last year, and then working with engineers at Fermilab and UC Davis to design and build the fixtures and procedures to get the forward detector installed.

A postdoc in the Davis group, Ricardo Vasquez Sierra, and I hand-carried the assembled half-disks aboard commercial aircraft from Chicago to Zurich to Geneva in four separate trips last year. These incredibly delicate devices were housed in special acrylic cases so as to facilitate security inspection. (We had made special arrangements with the TSA in Chicago…Zurich was more dificult.) The acrylic cases were in turn carried inside foam-lined hard shell cases. Needless to say, we carried each one, valued at about $500k, very carefully. People thought we were crazy – there is a certain history in our field of detectors arriving damaged when shipped – but we made it there with no problems at all. My biggest fear, I think, was some idiot tearing through the terminal and hitting one of our detectors with a luggage cart.

Meanwhile we needed to design a system to perform a sort of ship-in-a-bottle feat with the forward detectors. The detectors are deep inside the CMS tracker, the central bore of which is about seven meters long. The detector half-disks are mounted on two-meter-long carbon fiber service cylinders which also support the cables and tubes feeding power and cooling to the detector, plus some of the electronics. The two service cylinders sit vertically and slide into their final position along grooves in carbon fiber beds on the top and bottom of the bore. So as to have no uninstrumented regions in the vertical plane, at the end of travel the grooves are curved so as to make the half disks mesh. Thus, the two half cylinders need to be pushed in simultaneously with millimeter precision. Later the detectors need to be removed, at which point they will have become radioactive from exposure to the intense radiation environment in the center of the CMS detector. So the system had to be simple, easy and fast to use, so as to minimize radiation exposure to personnel.

Here is a remarkable photo of one of the forward pixel half cylinders half way into position. Note the converging tracks in which the half cylinder feet ride and the vertical beam pipe support that the detector has to clear on the way into position.

I have always been mechanically minded and enjoy problems like this. It was not the most sexy part of the pixel project, but an essential piece of making the whole thing work. We did a test of the inserttion a year ago when the tracker was still in a surface building at CERN. From the lessons we learned from a that test we built the final system for installation and tested it in May, before the beam pipe installation was completed.

So, just a few hours ago the CMS pixel detector was successfully installed. I was unable to be there, due to the recent birth of my son Ian. (Gotta have your priorities straight…) My able colleagues filled in seamlessly for me. Soon, though, the LHC and the ATLAS and CMS experiments will be up and running, and this great human adventure into inner space will commence.

  • Boltzmann’s Reptilian Brain

    So would “this time next year” be a reasonable current estimate for concrete results?

    [Sorry, this is a bore, but everyone wants to ask so I thought I would be the dope who actually does it…..]

    Amazing post by the way…. reassuring to know that there are people over there ingenious enough to deal with the almost infinite number of things that could go wrong…..

  • Pingback: Interesting Look into the Guts of the LHC « The First Excited State()

  • Moody834

    Thank you for this wonderful post. I have been chewing my lip to shreds in anticipation of the LHC’s first run. It is truly a grand machine, and the people (like you) behind it are, in my esteem, among the best people in the world. I can’t wait to see what you find.

    OK — just out of curiosity — I think I’ve seen this addressed elsewhere, but I can’t seem to find the info: What is the life expectancy of the LHC; how long before something even bigger needs to be made, and is it even a feasible consideration at this point?

    Thanks again.

  • Eugene

    That’s a fun post!

  • John

    Thanks for the nice comments and pingback.

    Moody: my bet is that the LHC could run at least 20 years. The first 6-8 years will definitively test the origin of electroweak symmetry breaking up to the TeV scale. I bet we’ll have a whole raft of new phenomena to unravel. Depending on the nature of those new phenomena, we could benefit greatly from “Super LHC”, plans for which are already well advanced. The super part is a big increase in luminosity, ten times more than we will enjoy in the first phase.

    Look at the Tevatron at Fermilab: it’s been going 20 years…and don’t count it out yet for a major discovery.

    But it is also possible that the nature of teh new physics will require a machine where we can precisely control the center of mass energy: an electron positron collider. Major discoveries at the LHC could revive flagging hopes of building the ILC, the International Linear Collider, possibly even in the US.

    BRB: we’ll have physics by this time next year from the LHC. In an incredibly lucky world, by Christmas this year we get a glimpse…

  • Christopher M

    This was awesome! There are so many books out there about the theoretical side of high-energy physics. Is there anything out there about the experimental side? As an amateur with no engineering training, it’s not at all clear to me how we could possibly take measurements with the precision necessary to do theoretical work at the scale of particle physics; and how we collect and successfully analyze the data. If lots of things collide at once, how do you know which sensed charges come from the collision you’re trying to watch? How do we map out these intense, specific decays?

    It just seems, on an intuitive level, like there should be an overwhelming number of unknowns — how many millimicrons did the Earth rattle during that run; are the electric sensors out of alignment by even 0.001%; and so on — and I don’t understand (even the basics of) how we probe the subatomic realm with any kind of confidence.

  • Claire C Smith

    Some people just have the best jobs!


  • collin237

    Do the charged particles stop when they hit a pixel detector? If not, how does the path traced out by the detectors get presented for viewing without decohering into only one hit?

    (Actually, I could ask the same thing about just an ordinary cloud chamber.)

  • Chris W.

    Somewhat off-topic: Physics World has just released an interesting article on the debate over DAMA’s alleged detection of dark matter. The intricacies of particle detection and interpretation of results plays a prominent role, of course. The article begins with the provocative Rudyard Kipling quote on DAMA’s home page.

    (Note: Juan Collar’s recent comments on CV are mentioned.)

  • hero

    “My biggest fear, I think, was some idiot tearing through the terminal and hitting one of our detectors with a luggage cart.”

    Actually it may not be a bad thing for the detector, the detector probably will think he finds the biggest particle in the world 😉

  • Pingback: What Will the LHC Find? | Cosmic Variance()

  • Sili

    Phil Plait just linked to these awesome pictures of the beast.

    If we put on out prognosticatory goggles and look twenty years into the future, will it possible to run anti-protons through the same pipes? Assuming of course that through the power of Science(tm), we’ll have found away to produce enough of them to get the necessary luminousity.

  • Richard

    This is (yet more) wonderful stuff.
    But we demand pictures of our intrepid voyagers cradling The Devices,
    or at least the undoubtedly-interesting engineering that went into just creating the transportation cradles for the devices.
    (A quick google image search for “cms forward disk” turns up nothing of human-interest.)
    Physics is of of course of highest interest, but logistics isn’t without its attractions, and the thousands of hours that skilled mechanical engineers and machinists must have put into making the detectors arrive in one piece oughtn’t go unrecorded.

  • Aristotle Pagaltzis

    Editorial note: the link to the post about the birth of your son points to the edit page for that post (which is, as it well should be, inaccessible to regular visitors) rather than to the permalink of the post on the publicly visible site.

  • Ellipsis

    Someone should answer the excellent questions from Christopher M, collin237, Sili, and Richard. I’m
    procrastinating from writing a paper, so I’ll do it:

    Christopher M: There are several books on the experimental side of high-energy physics, from technical books,
    to general textbooks, to popular accounts written by particle experimentalists. Some examples are: (general
    textbooks) _Detectors for Particle Radiation_ by K. Kleinknecht, _Experimental Techniques in High Energy
    Physics_ by T. Ferbel, and the excellent chapter in the classic _Introduction to High Energy Physics_ by D.
    Perkins. Physicists who already know the topic look things up at . Popular accounts include L. Lederman’s horribly-titled _The
    God Particle_, and _The Rise of the Standard Model_ by L. Hoddeson et al., amongst many others. As for the
    precision of the detectors, that is indeed very carefully measured and calibrated using a variety of means —
    primarily tracks of particles themselves. As for alignment, the typical scale that experiments need to
    calibrate the position of the innermost detector elements to is O(20 microns) or so for the very innermost
    pixels. The further out in the detector you go, the less sensitive we need to be. Now how, you may ask, do
    we use the very same particle tracks we are measuring to calibrate the detector? The answer is that the vast
    majority of events don’t get used for physics, so these aren’t typically the same events. But one thing you
    can be reasonably sure of with almost any particle track: _it doesn’t zigzag_. Particles travel in a smooth
    curve: in a constant magnetic field as within a detector, their tracks always form a helix. So if you see
    that the tracks your particles are making are zigzagging, and always in the same way when they cross a
    certain detector element, it is likely that you have the position of that element wrong. This is the
    technique by which the detector is aligned. We write computer programs to take a bunch of test events from
    the detector, determine zigzag patters, and thus determine an alignment pattern for the detector to remove
    them. We can align using tracks from cosmic rays or real beam tracks. A lot of effort is put into designing
    the detector elements such that they don’t vibrate, so we don’t need to do too much for effects that are
    rapidly time-dependent in-between accesses of the detector (but there are some other techniques for those, in
    fact ATLAS also uses a laser interferometry alignment system to check those types of movements, but they tend
    to be small). Hope that answers your question, ask if you have more.

    collin237: Particles almost never stop when they hit a pixel detector — they pass through the layers of
    pixels, as well as the outer tracking elements, and we detect a track if the particle is charged. How does
    one display particle tracks in 3-dimensional space, especially on a flat computer screen? This is a
    challenge, similar to many computer 3-D graphics challenges. Typically event displays allow one to rotate
    one’s view of the event and see it from different angles. There are many recent developments in computer
    graphics that may allow this to be improved in the future, see e.g.

    Sili: With the current LHC magnets, there is no way that antiprotons could be used — the antiprotons would
    bend in the wrong direction and hit the wall of the beam pipe immediately. Certainly one could conceive of
    an entirely new accelerator in the ring that could be p-pbar rather than p-p. But, upgrade-wise, under the
    hopefully quite safe assumption that the present LHC is successful, the focus is more on a luminosity upgrade
    of the present collider (in 7-8 years or so), and then probably later on (in 15-20 years or so) an energy
    upgrade, rather than changing it to p-pbar. We think we can get more physics reach by moving in those
    directions (luminosity and energy) rather than pushing hard on antiproton production and storage and
    switching to p-pbar — it would be hard to envision as significant gains in, say, Higgs or supersymmetry
    cross-sections by moving to p-pbar as by increasing the luminosity and (later on) the energy.

    Richard: As for the boxes that detectors are put in for shipping, I’m not sure where you can find a good set
    of pictures, maybe others can help. But there is a huge set of pictures and videos in

  • Ellipsis

    > calibrate the position of the innermost detector elements to is O(20 microns) or
    I meant O(2 microns)

  • Christopher M

    Thanks for the interesting explanation, and for the pointers to further reading!


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

See More

Collapse bottom bar