It takes just 30kg of the liquid perfluorobutane (C4F10) to cool ALICE’s Silicon Pixel Detector (SPD). This non-toxic liquid, similar to that used in fridges, is cycled through the SPD cooling plant in a constant flux of evaporation and condensation to effectively bring down the temperature of the electronics. However, since the installation of the SPD and cooling plant in 2008 there has been a growing problem. Dust and pollutants have become clogged in the filters that periodically intersect the cooling pipes. This not only reduces the efficiency of the cooling system, but also decreases the performance of the entire detector.
Although cleaning the dust is a seemingly simple solution, the filters, tucked deep inside the ALICE experiment, are inaccessible beneath electronics, cables and modules of other detectors. It has been the task of Rosario Turrisi and a team of technicians, engineers, and physicists to create an innovative solution to the problem. The answer it seems is to drill.
The SPD is part of ALICE’s Inner Tracking System (ITS). This comprises three pairs of detector layers wrapped around the beam pipe. The SPD forms the innermost and smallest pair. It is also the detector closest to the collision point. In the grand scheme of the ALICE experiment its role is to identify the position of the collision course.
ALICE CollaborationThe Silicon Pixel Detector (in orange) is the smallest detector in ALICE and the one closest to the collision point
The SPD electronics, designed to record data from collisions, generate heat. Although the working temperature is only about 30oC, the 1.5 kW of power produced in the detector is done so by very light materials. Rosario explains, “With no cooling system in place even a small amount of heat is able to burn through these materials in around 1 minute. A lack of cooling would cause the entire detector temperature to increase by 1oC per second. In 1 minute it would be at 80oC and we could consider the detector destroyed.” Clogged filters also cause the pressure of the cooling liquid to drop. This, combined with premature heating of the liquid from thermal contact with the environment rather than detector, means the liquid bubbles and begins to evaporate before it reaches the electronics from which it is supposed to absorb heat.
In addition to danger to the electronics, inefficient cooling reduces the performance of the detector. Losing 20% of the detector performance reduces the number of tracks recorded on the electronics by half. This reduces the range of potentially interesting events in the sample and means fewer data to analyse.
Since 2008 Rosario and his team have been recreating the cooling problem in the lab. “Since the very beginning the cooling system had low efficiency so we built a replica of the hydraulics and played with the issue. We have tested several solutions such as increasing the pressure in the cooling system to push more liquid through, and cleaning with ultrasound, but these have been unsuccessful. The problem started a few years ago and so it is not easy to clean the compacted dust. I would say we have had to increase the violence of our intervention.”
The solution now proposed and being tested during this winter shutdown is to insert a cable with a sharp milling tool at one end into the cooling pipe. This can be fed down to the detector. “We can drill a hole in the filter to allow some liquid to pass. It only needs to be about 1 mm in diameter to recover the good performance.”
The cooling system functions much like a house refrigerator. C4F10 is pushed through pipes towards the detector via a pump in the cooling plant. The pipes come into physical contact with the electronics generating the heat. Here the liquid absorbs the heat and evaporates. This C4F10 gas is then recovered by a compressor in the plant where it is cooled and collected in a tank before being recycled through the system. The filters presumably causing problems are around 60 micron in porosity and therefore let only particles of under 60 micron in diameter pass through. Unlike a traditional grid filter the SPD filters are a kind of fuzzy mesh contained in a stainless steel gasket, as can be seen in scanning electron microscopy images.
CERN EN-MME-MMA scanning electron microscopy image of unclogged filters used in the SPD cooling system
The major problem with the drilling solution is that the cooling plant lies some 40 metres from the detector and blocked filters. The closest access point still leaves the team 5 metres from the offending filters. The cable must fit down the 4 mm wide pipes with a sharp bend before they reach the mesh.
“We must do this step by step. We cannot afford to be surprised and risk damage to the detector. So we will keep testing and try a few times in the cavern once we are as sure as we can be with what we are doing.”
Rosario and his team are currently awaiting their turn in the ALICE cavern to tackle the intervention. Watch this space for updates...