EMOTIONS are running high. Even though we are wearing hairnets, face masks and overalls to protect the delicate silicon surfaces from dust and moisture, I鈥檓 pretty sure I can see Tony Weidberg welling up. 鈥淭here were times when I thought we鈥檇 never get this far,鈥� he says. 鈥淎fter all these years, it is amazing to see it in the flesh.鈥�
It鈥檚 not exactly flesh that we are looking at, but it is Weidberg鈥檚 baby. Suspended from a frame in the middle of his laboratory at the University of Oxford is a monster cylinder, 1.5 metres long and covered in shiny scales made of silicon. Cables spew from both its ends, while a robot arm sits poised in front of it. To Weidberg, this is more than just another piece of equipment. For the past 12 years this has been his life.
Even now, the device is not quite ready to be packed up and shipped off to its final resting place at CERN, the European particle physics laboratory near Geneva. But when Weidberg tells me about the seemingly ridiculous design specifications he and his colleagues have been working to, I begin to appreciate why just standing in front of it gets him a little emotional. There have been meetings, for instance, about the properties of the nuclei inside the atoms in the nuts and bolts that hold this device together. Include the wrong nuclei, and you could ruin the experiment.
Advertisement
Everything in the device, even down to the glue, also has to withstand extraordinarily intense heat and radiation. It has to be light, strong and rigid: it can鈥檛 move so much as a hair鈥檚 breadth, even when bombarded by particles last seen shooting out of the big bang fireball. Oh, and one more thing: it absolutely has to work first time because, once set in place, it will be buried under 7000 tonnes of other equipment. 鈥淚f anything goes wrong after we have installed it, we won鈥檛 be able to fix it for 10 years. It might as well be in outer space,鈥� says Weidberg鈥檚 colleague, Steve McMahon of the Rutherford Appleton Laboratory in Oxfordshire, UK. It really is an extraordinary project.
A decade or so ago, the cylinder 鈥� known as the semiconductor tracker 鈥� existed only in physicists鈥� imaginations and in engineering blueprints. By the end of next year it will sit at the heart of ATLAS, the biggest of four experiments that will scoop up subatomic particles created in collisions at the Large Hadron Collider. The LHC is a 27-kilometre-circumference accelerator still being built under the Swiss-French border at CERN, and the tracker鈥檚 job will be to snare the types of particles that filled the universe just moments after the big bang. Among other things, it might see baby black holes, the notorious Higgs boson 鈥� predicted but not yet found 鈥� and a hotchpotch of exotica from another dimension (See 鈥淔ires of creation鈥�, bottom). But while the LHC鈥檚 vast energies and collision rates will be a boon to physicists hunting new particles, they create an engineering nightmare for Weidberg and his colleagues.
The tracker is made of more than 16,000 silicon wafers, amounting to some 60 square metres. They are arranged in four cylinders, nested like Russian dolls around the collision point, and capped at both ends with 18 discs. The design spec says that it has to pinpoint the position of particles rushing away from the collision to an accuracy of less than 20 micrometres, a fifth the diameter of a human hair. And that means it can鈥檛 judder about. At all. Build a detector that wobbles even ever so slightly and you ruin its ability to pin down particle trajectories. That could make all the difference to those hunting Higgs bosons and mini black holes.
Although theory says that these precious entities will decay into fragments almost the instant they are flung out from the collision point, all is not lost. Physicists expect that the Higgs and another hypothesised form of matter called supersymmetric particles will break down mostly into heavyweight neutral particles called B mesons, which are themselves unstable and decay just before they reach the tracker. The difference is that B mesons decay into charged particles. As they zip through the layers they will ionise the silicon, which is carved into strips just 16 micrometres wide. The electrons that are released in the silicon will be swept up to the semiconductor鈥檚 surface by an electric field. The original particle can be pinpointed from the electric current flowing through a narrow silicon strip.
Every charged particle will therefore leave the electronic equivalent of an aircraft contrail. Reconstructing each particle contrail is like a game of connect-the-dots, where physicists join up the signals in each layer of silicon. The resulting picture reveals starbursts of contrails lying just in front of the first layer of silicon, and the epicentre of each burst is the point where the B meson decayed into showers of other charged particles. Particle physicists have been using silicon detectors to take such snapshots of B mesons for years, most recently at the CDF and D0 experiments at Fermilab in Batavia, Illinois.
鈥淲hen he hands me some of the shell, I expect it to be much heavier. I nearly drop it鈥�
But if connect-the-dots sounds simple, the devil is most definitely in the detail, which is why rock-solid equipment is everything. With a wobbly semiconductor tracker the dots would show up in the wrong place, making it almost impossible to reconstruct the trails and the all-important starbursts: supersymmetric particles could be streaming through the tracker and the physicists would miss them. According to computer simulations by Sebastien Correard and others working on the ATLAS team, a judder of just 20 micrometres would mean missing a third of B mesons.
Stopping the wobbles isn鈥檛 as easy as you might hope. You can鈥檛 simply bolt the silicon wafers onto a sturdy frame of steel or titanium because the atomic nuclei of these metals create strong electric fields 鈥� strong enough to knock charged particles off course if they enter the material, making it impossible to work out where they originally came from.
This 鈥渕ultiple scattering鈥� is much worse in materials that contain elements with a high atomic number, such as steel, copper and lead, because their nuclei are stuffed with protons whose strong positive charge deflects any passing charged particles. As if that wasn鈥檛 bad enough, these strong fields also trigger photons to decay into pairs of electrons and positrons: a photon bouncing into a steel bolt can produce a spray of other particles that make it even harder to tease out the ones of real interest.
As a result, the ATLAS team has had to think carefully about the composition of every single component. The physicists started looking for materials that would minimise multiple scattering in 1991, and it took them four years to finalise the details. 鈥淒ay-long meetings agonising over the design of a single washer or bolt are not unusual,鈥� says McMahon.
Their solution was to mount the silicon wafers onto four 1.5-metre-long barrels made from lightweight carbon-fibre shells using glue and lightweight aluminium bolts. Each shell is made of a carbon skin just 200 micrometres thick and is filled with a reinforced plastic honeycomb that strengthens it even further. And it is incredibly light 鈥� when McMahon hands me a long length of the shiny, grey material, I nearly drop it because I鈥檓 expecting something much heavier.
Wired up
The support structure isn鈥檛 the only source of scattering that the ATLAS team has had to consider. Each silicon wafer has to be wired up to a voltage supply and the electronics that read out the signals and control the wafer. With 16,000 wafers to connect, that adds up to a lot of wiring. Most electrical devices use copper wires, but traditional wires are useless in ATLAS 鈥� the metal鈥檚 proton-rich nuclei would fling particles far and wide.
To get round the problem, Weidberg and his colleagues have replaced the copper wires that carry the myriad signals with much thinner fibre-optic cables.
But data-handling isn鈥檛 the only problem when you鈥檙e generating 40 million proton collisions every second: over the 10 years that it will operate, the tracker will receive a radiation dose equivalent to 2 billion chest X-rays. That鈥檚 a big problem for anything containing sensitive electronics. High-energy particles knock atoms out of silicon crystals, and that can stop them doing their job. Usually, a passing particle leaves behind a small amount of charge, which creates a current in the semiconductor. But a silicon atom knocked out of the lattice, or the hole it leaves behind, can capture a passing particle鈥檚 charge and so prevent it from being detected.
While you can never completely eliminate radiation damage in silicon, you can lessen its effects by cooling the semiconductor wafers. Tests carried out by McMahon鈥檚 colleagues, and others, have shown that cooling the tracker to -10 掳C stops the displaced silicon atoms from roaming through the crystal and picking up the ionisation electrons.
But cooling the detector is no mean feat. Almost all particle physics experiments have to be cooled 鈥� the semiconductor tracker, for example, pumps out a staggering 30 kilowatts of heat. However, the usual trick, pumping cold water through thin pipes running past the silicon wafers, wasn鈥檛 an option. 鈥淲e ruled out water because of the possibility of leaks,鈥� McMahon says. A burst water pipe would short the electronics and wreck the semiconductor tracker, as well as equipment surrounding it. And with no way of fixing the leak, the ATLAS experiment would be ruined.
A more promising option was to pump carbon dioxide gas through the pipes. But that was eventually ruled out because the gas pressure needed to take away the heat would be 80 times atmospheric pressure, and no one wants to risk a ruptured pipe. Although a leak of the inert gas wouldn鈥檛 damage the tracker, if the tracker went without cooling even for less than a minute, the detector would overheat and have to be switched off.
鈥淎 burst pipe would short the electronics. With no way to fix the leak, the experiment would be ruined鈥�
In the end, McMahon鈥檚 team plumped for a fluorocarbon called perfluoropropane (C3F8). They pump it as a mist into metal cooling pipes made from a copper nickel alloy 鈥� with walls just 70 micrometres thick 鈥� where it condenses on the inner walls. The liquid absorbs heat from the electronics and soon vaporises, just like water turning into steam. The advantage of this is that the latent heat of vaporisation required to boil off the liquid carries away far more energy than simply heating a liquid or gas, making perfluoropropane a very effective coolant.
Some of the engineering problems presented by the tracker had slightly easier solutions. Richard Apsimon and Martin Gibson of the Rutherford Appleton lab, for example, had to find a radiation-resistant adhesive that would stick the silicon wafers together. Tests showed that most adhesives crumble when placed in a beam of neutrons and protons: the particles rip through the glue鈥檚 long polymer bonds. Apsimon and Gibson soon realised that nothing widely available on the market would last more than a few months in ATLAS. So they turned to the chemical company Ciba in Basel, Switzerland, for help.
The pair discovered that the base substance for most of Ciba鈥檚 glues had much shorter molecules that would be less susceptible to radiation damage. The company usually mixes in glass beads, silver and other materials to thicken and modify their glues鈥� properties, but Apsimon and Gibson couldn鈥檛 do that because silver has a high atomic number, which of course means it scatters particles. Instead they mixed the glue with boron nitride powder to boost its ability to conduct away heat and to thicken it up for easy use. And when they tried their bespoke glue in a particle beam, it stayed the course.
The creative thinking of the ATLAS engineers has already produced spin-off benefits: while researchers at the University of California, Berkeley, were testing the silicon wafers, for example, they realised their technique for inspecting the silicon surfaces could restore old sound recordings (麻豆传媒, 5 June 2004, p 25). And a team looking into cooling techniques at Queen Mary, University of London discovered a way to make a type of graphite that can conduct heat four times as well as the best copper.
As we stare at the detector, I鈥檓 becoming aware that Weidberg has told me about only a few of the thousands of engineering solutions he and his colleagues have had to create. A total of 37 laboratories around the world 鈥� a team of more than 200 physicists and engineers 鈥� have each designed, built or tested some part of this device (see Diagram). Ingenuity has been an essential job requirement for all of them.
It seems a healthy dose of Eeyore-like gloom doesn鈥檛 go amiss either. Weidberg has a smile on his face now, but the LHC doesn鈥檛 power up for another two years, and he鈥檚 under no illusions that they are out of the woods. 鈥淭here are still a horrible and horrendous number of things that could go wrong,鈥� he says. And any one of those would bring a tear to his eye.
FIRES OF CREATION (from above)
The proton collisions at CERN鈥檚 Large Hadron Collider will be like mini fireballs that give birth to hundreds of different particles, each carrying off a portion of the collision energy. Each collision will release 14 teraelectronvolts of energy, seven times as much as today鈥檚 most powerful working accelerator, the Tevatron at Fermilab in Batavia, Illinois.
At these high energies, particle physicists hope to catch their first glimpse of the long-sought Higgs boson. This is the only missing ingredient in the standard model of particle physics, a theory developed over the past three decades that incorporates our best knowledge of the building blocks of matter and the forces that glue them together. According to the theory, the Higgs is what ultimately gives everything its mass (麻豆传媒, 17 July 2004, p 38).
But what physicists really crave from the LHC is something totally out of the blue. 鈥淭here are very strong indications that the exploration of this completely new energy regime will lead to the discovery of new physics,鈥� says Tony Weidberg of the University of Oxford. Among the strongest hints are the predictions of a theory called supersymmetry. Theorists believe it could plug some holes in the standard model, explain the origin of the four forces in nature and bridge the long-standing gap between gravity and quantum mechanics.
But this theory necessitates the existence of a host of 鈥渟upersymmetric鈥� particles that no one has ever seen. According to the theory, they would have filled the universe shortly after the big bang, but no particle accelerator has been powerful enough to recreate those conditions 鈥� until now, perhaps.
