CERN's Discerning Detectors

Novel pixel sensors will be key to capturing quarry

Photo:Paul Scherrer Institute

GETTING READY FOR THE HUNT: Roland Horisberger demonstrates the bonding machine developed at the Paul Scherrer Institute. It fits 16 CMOS readout chips to each high-purity silicon pixel sensor with an accuracy of 1 to 2 micrometers.

Toward the end of this year, scientists at the European particle physics laboratory CERN will power up the Large Hadron Collider (LHC), a circular, 27- kilometer-long crash-test course for protons, which straddles the French-Swiss border outside Geneva. Their main quarry will be a tantalizing subatomic particle called the Higgs boson, considered pivotal to our understanding of mass and predicted by the so-called Standard Model, an integrated explanation of all elementary particles and forces except for gravity.

Two of the LHC's detectors are designed specifically to find the Higgs: CMS (for Compact Muon Solenoid) and Atlas (A Toroidal LHC Apparatus). Both CMS and Atlas focus on proton-proton collisions, but their designs employ somewhat different detection techniques so as to hedge bets. The scale and scope of the rival experiments, not to mention the equipment itself, are almost unfathomable. Atlas pits 1800 physicists at 150 universities in 35 countries against a similar number of scientists developing the CMS experiment at 181 institutions in 38 countries.

To get what's involved in the historic CMS-Atlas matchup down to a human scale, it helps to visit Roland Horisberger, who heads the CMS pixel detector project. Horisberger, a lean, genial physicist with a ready manner, works at the Paul Scherrer Institute, a Swiss national research laboratory located on the bucolic shores of the Aare River in Villigen, near Zurich.

Searching for the Higgs boson is basically an exercise in forensics, explains Horisberger. The Higgs is so incredibly short-lived that the only way of catching it is to examine the pieces it decays into and reassemble them. Because the patterns of decay are characteristic, "the chance that something [other than the Higgs] is doing it is almost zero," says Horisberger. But the Higgs is only one of the decay products expected from proton-proton collisions, and a very rare one at that. Its telltale signatures have to be extracted from a vast number of events--on the order of 600 million collisions per second.

Detecting and processing all that activity has required scientists and engineers to develop silicon pixel sensors for a new kind of detector, versions of which will be used in both Atlas and CMS. The new device is the latest in several generations of electronic particle detectors introduced since the late 1960s.

Until then, the detectors in particle accelerators still consisted of cloud chambers, like those in which the trails of cosmic rays can be seen. But in 1968 the French Nobel laureate Georges Charpak invented the first all-electronic detector, the proportional wire chamber, a derivative of the Geiger counter. While the classic Geiger counter consists of a single wire in a chamber containing an ionizing gas, Charpak's device consisted of wire grids layered, with their wires running orthogonally, so that particles could be tracked in three dimensions.

Photo: AIP Emilio Segre Visual Archives

ENGINEERING NOBELISTS: Georges Charpak [right] invented the first all-electronic particle detector, and Simon van der Meer created a method of beam focusing that was crucial to the discovery of the W and Z bosons.

The next big step was the introduction of silicon strip detectors in the 1980s: a particle going through one of the silicon strips, each roughly the width of a human hair, left a trace that could be read out by electronics neatly wire-bonded along the side. Horisberger was among those who developed strip detectors for CERN's LEP, the Large Electron-Positron collider, the LHC's immediate predecessor. It was in the LEP's detectors that the W and Z bosons were found, confirming the Standard Model.

Silicon strip detectors worked well for the lower-energy LEP and in other notable experiments. But when considered for high intensity proton-proton collisions, their limitations were evident. For example, they measure only in a single direction. You could tell that a strip had been hit but not exactly where along the strip.

So from the earliest days of its planning, in 1994, the solution for the LHC clearly lay in creating a grid of square pixels, which would give two coordinates--and thus more detailed, precise information--on the particle strikes and paths. Now, however, the problem was how to accommodate all the needed electronics.

Proton-proton collisions are notoriously trashy, producing an enormous number of events and particles, of which only a tiny number will be of real interest. The LHC was expected to generate half a terabyte of information every second--yet at the time a good PC hard disk drive could store less than a thousandth of that. Moreover, the detector was going to require 1000 amperes of current to operate. Cooling it would take more than hand waving. Says Horisberger, "It seemed to all of us to be a ridiculous, ruinous business."

Nevertheless, working with colleagues at the University of Zurich and at the Swiss Federal Institute of Technology at Zurich, and in the United States, Horisberger and his team met these and other challenges. The information problem was solved by developing "smart" pixels, each equipped with a 250 transistor circuit that does a preliminary sifting of events. The pixels are organized into modules, 66 560 to a module, and connected vertically to the readout electronics. The pixels communicate with the outside world only if they have something relevant to say.

To visualize the CMS detector, think of Russian nested dolls. Running through the very center of the CMS is the 6-centimeter-wide accelerator beam pipe, where the collisions take place. The pixel detector, a 60-cm-long barrel-shaped instrument, wraps right around the beam. Its 65 million pixels face the beam pipe and transmit their data via 1500 optical-fiber links.

The pixel detector is further nested within several layers of silicon strip detectors, which in turn are nested within other sorts of detectors. The complete CMS structure that includes all these subsystems is the size of a barn and weighs a third as much as the Titanic.

Horisberger's in-house production factory has been hard at work fabricating the pixel modules, and his team has completed about half the number needed. During a walk through the lab, he points out the tiny screws supplied by Swiss watchmakers and the "orgies of plugs" that will seat the cabling. Once the detector is complete, in August, it will be trucked to CERN for installation.

These are intense days for Horisberger and his team. The final fabrication phase is going very well, he says. Moreover, he adds, "We are quite sure that the detector is going to work." And though the understanding of the Higgs boson is, as he puts it, "a purely cultural undertaking," the technology underpinning the pixel detector could have commercial use. It has been spun off as a company called Dectris, in Villigen, which makes pixel detector sensors for scientific X-ray investigation of protein structures.

An update to this article was posted on 2 May 2007. Click to read it here.

Related Stories