The text below is part of a chapter of "Anomaly!" which I eventually removed from the book, mainly due to the strict page limit set by my publisher. It is a chapter that discusses the preparations for Run 2 of the Fermilab Tevatron, which started in 2002 and lasted almost 10 years. There were many, many stories connected to the construction of the CDF II detector, and it is a real pity that they did not get included in the book. So at least I can offer some of them here for your entertainment... [A disclaimer: the text has not been proofread and is in its initial, uncorrected state.]
The CDF Run 2 Upgrade
In preparation for Run 2 the CDF and DZERO experiments both underwent massive upgrades. These had mainly three purposes. First of all, modifications were needed to adapt the detectors to the different running conditions in the upcoming run, especially the shorter time interval between consecutive crossings of proton and antiproton bunches. Second, improvements were due to those systems which showed reduced performances due to radiation exposure, as well as those which could most profit from the technological advances that had taken place during the past twenty years. And third, upgrades were advisable to increase wherever possible to efficiency with which rare and interesting events were collected.
To meet those demands, CDF decided to totally change its tracking detectors inside the central solenoid, to increase the coverage of its muon detector system, and to replace most of its data acquisition electronics, which were not capable to handle the increased collision rate. The central tracking chamber was substituted with a new detector, named “central outer tracker” (COT, see picture above). It was based on a similar conceptual design but it would be more performant, and it allowed a much faster readout.
The silicon detector was replaced by a more complex system of three subdetectors in order to increase coverage and precision of the inner tracking. As for the muon detectors, they were augmented by the completion of angular coverage in the central region, and the addition of a new “intermediate muon system” of drift tubes and scintillators, providing coverage for muons traveling at smaller angle with respect to the beams.
Like CDF, the DZERO collaboration also had many important upgrades to perform to get ready for a competition in Run 2. But in addition, DZERO also needed to expiate its childhood mistakes. In Run 1 a solenoid capable of bending charged particles had been sorely missed, as due to its absence the experiment lacked the ability of measuring particle momenta (except muons, whose trajectories got bent traversing magnetized iron outside the calorimeters).
Moreover, the lack of a silicon detector in its central tracking system had meant that DZERO could not compete with CDF in precision B-physics measurements. In Run 2 the two detectors would become much more similar to one another than they had been in Run 1, as they were now both endowed with central solenoids and powerful inner silicon trackers.
The SVT: A Technological Marvel
Despite the increased similarity of the two Fermilab experiments, CDF demonstrated to be able to make further innovative steps forward. One example was in the field of data acquisition. A crucial task in Run 2 was indeed to keep the quality of data written to data storage as high as possible in the face of the one-order-of-magnitude increase in collision rate that the accelerator upgrade promised to deliver. Hence the trigger was subjected to significant improvements, in part allowed by the lowered cost of commercial processors and in part granted by studies that had started over a decade earlier. The latter was the case of SVT, the Silicon Vertex Trigger: a innovative marvel and a brainchild of Luciano Ristori and Aldo Menzione, which would later win them the prestigious Panofsky prize in 2009.
SVT (left, Ristori attending one of the system's racks) was a highly parallelized set of custom electronic modules, capable of reconstructing on the fly and with great precision the trajectories of charged particles crossing the silicon detector layers. Within 20 microseconds or less, hits in the silicon sensors were read in, their pattern was compared to pre-defined ones stored in associative memory banks, and a linearized track fit was performed on sets of hits that matched the pattern that real tracks could display.
The fit extracted the particle momenta, azimuthal angles, and impact parameters with respect to the beam position in the plane transverse to it. Using that information, the level-2 trigger could decide whether the event contained hints of production of particles with long lifetime, which produced high impact-parameter tracks in their decay. SVT ended up providing a huge boost to the B physics potential of CDF II.
The success of the CDF silicon tracker in Run 1 was a strong stimulus to increase the scope of that subdetector in Run 2. The higher fluence of hadrons expected in the still harsher environment of Run 2, due to higher rates and energy, could be handled by detectors kept at lower temperature, which could then withstand higher doses of radiation without undergoing a significant change in their operativity.
The prospects of a Higgs boson discovery, as studied originally in the TeV 2000 report by the group led by Amidei and Brock, critically depended on an increase of the angular coverage and efficiency of the silicon sensors, now called to intercept particle trajectories down to smaller angles with respect to the beams direction. This could be achieved either by increasing the length of the cylinder surrounding the beam spot or by adding suitable disks on both sides of it; DZERO chose the latter, more innovative option, while CDF opted for the former, going from a two-barrel geometry (two adjacent barrels of silicon layers) to a three-barrel one. In addition, the intermediate region left between the inner silicon tracker and the COT could be instrumented with additional silicon layers. These were called “Intermediate Silicon Layers” (ISL).
Shooting through a brick
Besides covering more solid angle around the interaction point, the other way to increase the b-tagging potential of CDF was to measure more precisely the impact parameter of tracks than the SVX and SVX’ detectors could do in Run 1. The simplest way to achieve this was to add layers of silicon as close to the beam as physically possible. Realizing the possibility of doing so, Joe Incandela took a leading role in proposing a “Layer 00” of small silicon sensors that could be mounted directly onto the 2.5”-wide beryllium beampipe, obtaining a precise measurement point of track coordinates at less than 2” from the interaction point.
Incandela’s project suffered from the start from significant internal opposition: many of his colleagues thought that the project was too daring due to mechanical and spatial constraints, and opposed to the idea of further additions of material in the particle paths. The shootout happened at a meeting on December 9th 1997, when Incandela’s colleague Rick Snider presented the proposal of L00 to the collaboration.
The motivation for L00 as Rick presented it was that while the new five-layer silicon detector was expected to do a great job in terms of tracking efficiency, as charged particle trajectories would be easy to reconstruct with five spatial measurements along their path, the considerable amount of material constituting it also produced a significant amount of multiple scattering, the collective effect of electromagnetic interactions with the atoms of the crossed material which little by little deflected the incident particle. Multiple scattering deteriorated the resolution on the impact parameter, the crucial quantity used to select tracks coming from the decay of long-lived particles. Adding yet one more layer would further increase the effect, but would on the other hand also provide a measurement point very close to the origin of particle paths, where particles had not withstood any scattering yet.
It was Larry Nodulman, who had spent the better part of his life building and attending to the central calorimeter, and whose studies of electroweak physics benefited of the precise measurement of electrons (whose energy resolution would be damaged by addition of material in the tracker), who put it most effectively:
“So let me get this straight: Because we have too much silicon, you want to add silicon?”
“Yes, exactly!”, was Snider’s candid reply.
Everybody laughed. “That is crazy!” said Larry. But it was only one of the two possible points of view: it made sense, and yet it didn’t, depending on how you looked at it. Bill Foster sarcastically commented the proposal with the following words:
“Silicon is like a drug: you can’t have enough of it.”
But silicon detectors had discovered the top, and the CDF collaborators were generally biased in its favour. A heated debate ensued, but in the end Incandela and Snider were given a green light: they could now attack the problem of finding funds to build L00. One result of the discussion was that from then on Nodulman would not lose any occasion to address the CDF II silicon detector in a half-mocking way, referring to the troubled path of particles coming out of the interaction point before reaching his beloved electromagnetic calorimeter:
“Things will be tough this time - we’ll be shooting through a brick!”
An extra layer of silicon in CDF also faced the initial opposition of the laboratory director. As Incandela presented his project to John Peoples, explaining that besides increasing the chances of precise b-tagging the added sensors could extend the operational life of the detector, the director shook his head. Fermilab was in bad waters with its budget, which had started to decrease since the mid nineties. Besides the fact that the CDF and DZERO upgrades were already draining most of the resources of the lab, Peoples was facing another challenge: a nasty structural problem of Wilson Hall, the Fermilab headquarters building, required large investments to be fixed.
The design of the beautiful 15-story Hirise was due to Robert Wilson, the first Fermilab director, who had a penchant for art and architecture but little trust in computer calculations of structural stresses, which in the sixties were also not straightforward to perform. It turned out that the stresses caused by temperature differences between the inside and outside of the building had in the long run produced significant structural damage to the concrete: the need for an invasive and costly intervention could not be ignored any further.
Pressed by his budgetary constraints, Peoples denied to Incandela any kind of financial support for the L00 project, claiming that overall the upgrade of CDF was already way over budget, and that it was also running late. Yet Incandela got the upper hand as he gave a presentation on the CDF upgrade to Fermilab’s Physics Advisory Committee in Aspen, in June 1998. He snuck in his presentation two slides about L00 which he had not agreed upon in advance with Peoples, and got his audience excited about the idea. The project eventually got approved, and with considerable technical difficulty and headaches for the L00 team the daring extra silicon layer was successfully installed on the beryllium beampipe in time for the start of Run 2.
Horse dung in the silicon detector
As for the ISL, this was another daring project. At a length of 80 inches and a diameter of 25 inches the barrel was going to be the largest silicon detector ever built: it had the same dimensions of the tracking detectors designed for the LHC experiments. The coincidence was a favourable one, as ISL could benefit from the availability of equipment and laboratory space that had been allocated in Pisa for the CMS experiment.
Incandela, who was now the head of the silicon detector center at Fermilab, broke a deal with Guido Tonelli in order to have access to the facilities in Pisa, especially the large coordinate-measuring machine needed to construct the structure; in exchange, the CDF member had to promise he would personally provide help in the construction of the CMS tracker. Years later, this agreement would lead Incandela to become a member of CMS, and eventually to succeed to Tonelli as spokesperson of that experiment, at the time of the discovery of the Higgs boson in 2012.
The construction of the ISL was a joint responsibility of Fermilab and Pisa university. Once the support structure was ready, a complex system of narrow, thin-walled aluminum cooling pipes was installed onto it by Italian technicians in June 2000 at Fermilab.
In those days the European Soccer Championship was going on in Belgium and the Netherlands, and the Italian team was among the most credited competitors for the title. The technicians intended to follow the soccer games on TV, despite the latter were broadcast during daytime working hours. As a result, most of the assembling work ended up being performed during the night. This was a suboptimal choice, which may have contributed to cause a very big problem: the 3M epoxy used to glue the aluminum tubes leaked inside in some places, obstructing the tubes.
Unfortunately, the shortcoming was realized only months later. The delayed realization occurred because the cooling pipes were connected by complex manifolds and there was no one-to-one correspondence between input and output, so the simple tests made after assembly by the technicians had shown no issue. The flow of water in some lines did look funny to them, but they concluded that everything was in good order.
After the cooling pipes were mounted, the installation of all the silicon ladders could take place. Time was running out to complete the system in time, so all that was done was a quick electrical check that did not require the system to be cooled. Finally, the detector was transported to the collision hall, ready to be installed in the core of CDF.
A thorough verification of the cooling could be made at that point, to check that the water flown in the system could indeed maintain the detector at a working temperature, removing the excess heat off the readout electronics. Only then was it discovered that the temperature got quickly too high in one third of the silicon ladders. The complexity of the manifolds however still made it unclear what exactly was going on, and detailed investigations were needed to ascertain what was causing the problem. A task force was formed, with Douglas Glenzinski directing the work.
Although the logical explanation was that some of the tubes were obstructed by gluing epoxy that had snuck inside the cooling tubes, this was such a nightmarish concept that alternative explanations were entertained for quite a while. Finally it was possible to visually inspect the interior of the tubing, using a very long endoscope bought for the occasion. Large clots of glue were then spotted in several places –mostly hard-to-reach spots next to tube junctures. With an approaching commissioning run of the Tevatron it looked unthinkable to take the detector apart and construct a new tubing system; hence all conceivable options were studied to get rid of the excess glue.
The simplest idea would be to use a solvent, but the 3M epoxy proved a tough nut to crack: nothing was found that could attack it while keeping the aluminum unaffected. Melting the glue with a powerful laser beam was also put forth early, but was initially thought the craziest idea; some of the task force members also spent some time getting experienced with sandblasting tools, with poor results.
It was finally Cigdem Issever, a post-doc from Santa Barbara University, who demonstrated that a laser was the way to go. With the help of the endoscope one could sneak into the narrow tubes a fiber optic laser, and direct the beam on the epoxy, melting it. While the endoscope allowed the silicon experts to see where the glue was, directing the laser onto it was by no means easy.
The laser they employed was a powerful Neodymium unit which could punch a hole in a penny in the matter of milliseconds; fortunately, its power and repetition rate were entirely adjustable, and detailed tests on mock-ups convinced the team members that they could melt the glue without harming the tube walls. In order to see whether they were directing the beam on the glue or to the aluminum walls, the Fermilab scientists used tiny mirrors that they could insert inside the pipes, and a photon detector which could see the light scattered back by the aluminum.
Eventually, the glue was melted in all the critical places, and the cooling system regained almost 100% of its functionality. A drawback of the otherwise successful operation is the scorn of some malignant souls, who to this day claim that the ISL is the first particle detector containing horse dung: the endoscope was a special 2-meter-long system, of the kind only built to diagnose colon cancer in horses – and to save money, the CDF managers had managed to buy a second-hand unit...