Complex systems and challenging mechanical structures for high energy physic experiments. Some examples from the Neutrino Platform. Diego Perini CERN, Europenn Center for Nuclear Research Geneva, Switzerland Diego.Perini@cern.ch Copyright © held by the author Abstract—High-energy physic experiments are complex As final example, I describe the design and construction of systems. The particle detectors, the electronics for data the ICARUS experiment aluminium cryostats. This experiment selection and acquisition, the services, and the mechanical is one of the milestones of the neutrino programme and it support structures are all integrated in a highly crowded and makes use of the largest liquid argon Time Projector Chamber optimized space. The size and sophistication of these systems built and operated so far with a bath of approximately 760 have been constantly growing during the last decades. tonnes of fluid. This note summarizes some basic common characteristics Keywords—Detectors, Neutrinos, Mechanical structures of these apparatuses and describes how these concepts are implemented in several experiments under design or I. INTRODUCTION construction to study the behaviour of neutrinos. Neutrinos are intriguing particles: they have no electrical A. Complexity in experiments – general description. charge, much smaller mass than the other particles and weakly interact with matter. The Standard Model of particle physics High-energy physic experiments are extremely complex as it is today cannot explain some of their measured properties. systems as shown in Fig. 1. They are designed to study the Therefore, the neutrino studies are gaining importance in the interactions of particles with other particles or with the field of high-energy physic. matter of a target. In both cases, the heart of the system is the detector. This latter is located around the region where One hundred and seventy-five research institutes all over particles collide and it is made of elements that react to the the world have established a common important programme of phenomena arising during the particle interactions (for experiments. It foresees the construction of a series of detectors example generation of photons or other particles). These from small prototypes to large elements operating in liquid ‘reactions’ produce signals that are collected, selected and argon cryogenic environment. The first prototypes have a size of a few cubic meters while the ultimate detector will be in four memorised by the data acquisition system of the experiment. elements 22.4 m x 14 m x 45.6 m each. One of these elements These data are then analysed and studied by the experimental will contain about 17’000 tonnes of liquid argon. They will be physicists. located in an old mine in South Dakota. The cavern is 1500 m Specific machines, the colliders produce and accelerate below the ground level, a challenge for the transport and the interacting particles. In other cases, the particles originate assembly of all the components. from natural sources like radioactive decays or cosmic rays. Fig. 1. An example of different systems integrated in a large experimental area. The CERN Neutrino Platform (in red the two containment structures of the ProtoDUNE detectors). To work properly, a detector needs a performing data employed in the construction must have relative magnetic acquisition system. It needs as well services like power permeability as close as possible to one in order not to impair supply elements, cooling fluids, ventilation of the the quality of the field and consequently the precision of the experimental area, control and regulation systems (pressure, particle tracking. Aluminium alloys, austenitic stainless steel temperature, etc.). All these services requires an impressing or plastic materials satisfy this requirement. amount of cabling and piping: large quantities of these elements are space consuming and heavy. Finally, adequate Finally, in some cases, the materials of the detectors and mechanical support structures must position and align the services must be radiation resistant. detectors and all the services must be integrated on board. Detectors have grown in size and complexity during the Certain detectors are conceived to be sensitive to a given last decades. For instance, the Large Hadron Collider (LHC) class of particles and are disturbed by others that generate detectors are located in caverns approximately 50 m long, 30 noise in the data acquisition system. Often, these detectors m wide and 50 m high. The weight of one of these detectors are ‘shielded’ by filters allowing only the right particles to is of the order of thousands of tonnes. Despite the large size, pass through. These filters can be made of less traditional the manufacturing and assembly precision of the most materials than those used in normal constructions. Lead, important elements is of the order of a few millimetres. As an example, the muon filter of ALICE experiment is made of tungsten, graphite, borated polyethylene, and cast iron are some example of materials for filters. three conical and cylindrical elements weighting 40, 18 and 11 tonnes respectively [1]. Thank to adjusting supports, the The detectors incorporate the sensitive elements into axis of each element could be located at less than + 2 mm structures that can be made of very ‘exotic’ materials: dense respect to its theoretical position. ALICE central detectors elements for certain applications or at the contrary, as light as are made in several 7-m-long elements with a total weight of possible structures for other applications. Sampling more than 80 tonnes and are supported by a cylindrical frame calorimeters, the detectors recording the energy associated to structure 8.5 m in diameter, 7 m long and 10 tonnes heavy. a given event, are made of sensitive layers alternated with The support and alignment structure has in this case a weight very dense layers (for example tungsten or lead). On the about eight times smaller than the supported detectors. The other hand, the inner tracking systems, the detectors closer to clearance between the structure and the detector elements is the interaction point in the colliders are made of silicon between 0 and 4 mm. The maximum deformation of the semiconductor elements positioned on extremely light support frame under the load of the detectors is about 5 mm. carbon fibre supports. To resume the order of magnitude of deformations and tolerances is a few millimetres while overall dimensions of In all cases, the ideal detector should be made of sensitive the structures are several meters and loads represent elements that cover all the solid angle around the particle hundreds of tonnes. interaction point. The support structures represent a discontinuity of the solid angle. It is a space filled by a The cost of high-energy physic experiments has grown in passive mass that do not detect interesting phenomena and parallel with complexity and size. Today it can be afforded on the contrary could generate noise in the detector. only by collaborations of many different institutes worldwide. Often, to save in costs, experiments are built re- Another strong requirement for the detectors is a precise using components or facilities: experimental caverns and alignment and position of their different components. The service buildings are re-adapted or detector elements are re- reconstruction programs of the data analysis systems use the furbished. This fact imposes some extra boundary conditions coordinates of the detected particle trajectories. Position to new projects and sometimes makes the technical and errors impair the precision of these computations. design choices more difficult than in an ideal ‘starting from In any mechanical structure, some clearance between zero’ configuration. parts under assembly facilitates the manufacturing work. The coordination of a large team of specialists from Clearance can allow an enlargement of the construction different institutes requires a considerable project tolerances. It can as well absorb without extra stresses the management effort during the design, construction, operation deformations of the structures under their weight, thermal and dismounting. This aspect, together with the life cycle of loads or other loads. In case of a detector, clearance is again the experiment and of its components will not be described a dead area, a place where there are no sensitive elements. in this note. For all these reasons, the structures of a detector must be In the following pages, I will focus on the structures of an as little invasive as possible and require precise construction, important programme of experiments, the neutrino baseline. small clearance, and little deformations under all load cases. I will show that many of the general aspects described above To summarize: are typical of this programme as well.  the design must maximize the rigidity of the structure and minimize its mass, B. Neutrino experiments.  the construction must be as precise as possible and cope with small clearance between parts. Neutrinos typically pass through normal matter All this with the basic requirement of keeping the industrial unimpeded and undetected since they interact weakly with costs at affordable and reasonable level. other particles [2]. They are the most abundant matter particles in the universe, and they are all around us, but we Several experiments are immersed in a magnetic field to know very little about them. bend and track charged particles. In this case, the material A Neutrino has no electric charge and the mass is much smaller than that of the other known elementary particles. For long time his mass was thought to be zero. An important phenomenon involving neutrinos is the neutrino oscillation. A neutrino created in a weak interaction and having a specific lepton flavour can later be measured with a different flavour [2]. The Neutrino oscillation is of great interest, as it implies that the neutrino has a non-zero mass. This fact requires a modification to the Standard Modell of particle physics. The experimental discovery of neutrino oscillation, and thus neutrino mass, by the Super- Kamiokande Observatory and the Sudbury Neutrino Observatories deserved the 2015 Nobel Prize for Physics. A way to measure the rate of flavour change is to generate a beam of neutrinos and then count how many neutrinos of a given flavour are present at the starting point, and how many at some distance away. Thus, neutrino oscillation experiments generally have more than one detector. The near detector is located a few hundred meters downstream of the neutrino source and it characterizes the neutrino beam in its initial state. The far detector is located at a distance that can be ‘short’ (less than one kilometre) or Fig. 2. The working principle of a liquid Argon neutrino detector. ‘long’ (several hundreds of kilometres). The density of liquid argon is 1392.8 kg/m3 II. LIQUID ARGON DETECTORS (approximately 1.4 times that of water). For large detectors Neutrinos can travel through dense matter without the hydrostatic pressure against the reservoir walls represents interacting with a single atom. To observe the rare a considerable load. For this reason, the containment interactions it is necessary to build and operate for years very structures must be carefully designed to limit stresses and special detectors. They are made of large masses of target deformations. materials and they record the track of the particles emerging from the rare interactions of neutrinos with the target atoms. Another important point is the operation temperature of Liquid-argon detectors represent one of the most promising liquid argon at atmospheric pressure, which is 78 K (-186 o technique to show what happens when a neutrino hits a C). The fluid must therefore be stored in a well-insulated nucleus of an atom. Tracks that the resultant particles leave structure and continuously cooled. For the large detectors behind are shown in high resolution, and it’s possible to under design the necessary cooling infrastructures will have distinguish the various particle types. They were first dimensions and complexity comparable to industrial plants proposed in the seventies [3] and further developed in the for chemical or petrochemical applications. last decades. III. THE NEUTRINO PROGRAM A liquid argon time projection chamber (LarTPC) is essentially a box of liquid argon as shown in Fig. 2. The The unknown properties of neutrinos could give answers Argon is both the neutrino interaction medium and the to several basic questions about the universe. tracking medium for charged particles produced in the One hundred and seventy-five research institutes from all interactions. During one of these events, ionization charges over the world have established an ambitious and long-term are produced along the tracks of the charged particles. These program to carry out experiments in the field of neutrino ionization charges drift at constant speed toward one side of physics. According to the different properties that each the detector under the influence of an electrical field ‘E’ experiment aims to measure, there are two possibilities. applied uniformly in the argon volume. A grid of sensor wires positioned on a plane finally collects the charges. For  The far detector is located at a few hundred meters each charge, the amplitude, wire position and arrival time are from the point in which the neutrino beam is recorded and used in the data analysis software to reconstruct generated (short baseline). the event topology [4]. Photosensitive detectors record as  The far detector is located at several hundred well the argon scintillation light emitted during the event. kilometres from the neutrino generation point (long This is used for event triggering and to determine the initial baseline). interaction time t0. In both cases, the large detectors can be used as well to One important parameter is the purity of liquid argon that study neutrinos coming from space (such as rare processes should be as high as possible. Argon is a noble gas and do like supernova neutrino detection). not intercept the drifting charges, while polluting elements absorb the charges and decreases the performance of the The two baselines are complementary in terms of both detector. For this reason, the surrounding structure must be physic studies and technology. The research and designed to avoid both fluid leakage and contaminations of development for the two baselines is carried out in parallel the argon bath. and the improvements are transferred from one to the other whenever possible. A. The short baseline at FNAL B. The long baseline DUNE experiment The Short-Baseline neutrino program (SNB) at Fermi In the long baseline, DUNE experiment, the neutrino National Accelerator Laboratory (FNAL) will measure the beam produced at FNAL in Illinois will be sent to a detector neutrino oscillation using the Booster Neutrino Beam. Three located at Sanford Underground Research Facility in South detectors are under construction: SBND (the near detector), Dakota (see Fig. 4). MicroBooNE (the intermediate detector) and ICARUS (the far detector). FNAL is upgrading its facilities in order to produce the intense neutrino beam required for this application. With this These three detectors use a liquid-argon time projection development, this will be the most particle-packed high- chamber (LArTPC). Each of them contributes as well to the energy neutrino beam in the world. development of this particle detection technology for the long-baseline Deep Underground Neutrino Experiment DUNE consists of two detectors, a smaller near detector (DUNE). at FNAL and a much larger far detector in a cavern 1500 m beneath the surface in an old South Dakota mine. The second The SBN far Detector is the ICARUS T600, the largest detector is 1300 km far away from the first. LArTPC built to date. This detector operated for some years in Italy as a far detector in a long-baseline experiment. When The far detector will be by far the largest ever built using that experiment completed, the ICARUS detector was taken liquid argon technologies. The ground breaking for this to CERN and refurbished. The modifications include newly project took place in July 2017 and the experiment is developed readout electronics and a new cryogenics system. expected to be operational in 2026. The far detector will be ICARUS will operate at ground level and not in a deep made of four cryogenic modules, each of which will contain cavern as it was during its first use. The need to separate the approximately 17’000 tons of liquid argon. A central service interesting neutrino events from the noise generated by cavern will house the cryogenics system, electrical power equipment, air-handling units, and other support equipment. cosmic events requires several modifications of the electronic and of the detecting elements [5]. About 875’000 tonnes of rocks will be excavated in the next years to create the experimental and service caverns. The ICARUS-T600 detector consists of two large identical modules with internal dimensions 3.6 x 3.9 x 19.6 The DUNE collaboration is as well constructing two 800- m3 each of them filled with ~385 tons of ultra-pure liquid ton prototype detectors, called ProtoDUNEs, at the CERN argon. These elements are surrounded by a common thermal Neutrino Platform. They will use a low-intensity particle insulation. Each module houses two TPCs separated by a beam provided by the CERN accelerator complex. The two common central cathode for an active volume of 3.2 x 2.96 x prototypes will assess the performances of two different 18.0 m3 (as shown in Fig. 3) [5]. configurations of detecting elements. The results of the tests will drive the final technological choices for the four far One TPC is made of three parallel wire planes. Globally, detector DUNE modules. there are 53’248 wires with length up to 9 m. A three- dimensional image of the ionizing event is reconstructed A smaller, 35-ton prototype for DUNE was tested at combining the wire coordinate on each plane at a given drift FNAL in early 2016. time with 1 mm3 resolution over the whole active volume. All the prototypes have a similar support structure. It The TPCs are installed inside two new aluminium consists of a box assembled from frames made of cryostats. The cryostats are self-supporting. In other words, construction steel beams welded together. The box support they withstand the load given by the liquid argon. The the inside thermal insulation material. An austenitic steel insulation and the other structures around the cryostats do not skin made of corrugated thin plates welded together in situ give any contribution to their stiffness and resistance. These assures the tightness and contain the argon. The skin lays cryostats are complicate objects in terms of design and against the insulation panels. Therefore, the hydrostatic construction. They are made of welded aluminium elements, pressure of the liquid argon loads the external box frames. require a construction precision within a few millimetres, are The large detectors in South Dakota will have a similar heavily stressed and must be leak tight. Chapter IV describes structure. The complication is given by the need to design in detail their design and construction. modular components that are small enough to pass through the access shaft that is a few meters wide (~ 5 m x 4 m). Fig. 4. The long baseline DUNE experiments. Neutrinos produced in Illinois are sent to the DUNE detector at 1300 km distance in South Dakota. Fig. 3. One of the ICARUS detectors during the refurbishment at CERN. IV. THE ICARUS CRYOSTATS Each ICARUS cryostat consists of an aluminium alloy structure with dimensions approximately 20 m x 4 m x 4 m. They were manufactured at CERN in the framework of this detector refurbishment programme. ICARUS will finally be used as far detector for the neutrino short baseline in FNAL as described in chapter III. At the end of the construction, the cryostats were cleaned, then the TPCs were inserted and the doors were definitely closed. Finally, the assembly was transported and installed at FNAL. At FNAL, the first phase of the operations will be the removal of the air from the cryostats. Then the system will be cooled and filled with liquid argon at 87 K. During their lifecycle, two loading configurations are critical for the two cryostats: Fig. 6. First pass weld of one of the panels.  the phase with vacuum inside and atmospheric pressure outside, The final assembly and welding took approximately five  the nominal working conditions with liquid argon months for the first cryostat and four for the second one. inside. Each cryostat required the execution of approximately For the structural design, the Eurocode 9 [6] and related 540 meters of welds. Each of these welds was in two passes; harmonized norms were used whenever possible despite the therefore, the total welding length was more than two fact that no specific norms exist for this kind of aluminium kilometres for both the cryostats. The first pass was the one construction. originating the majority of the contraction after solidification, approximately three quarters of the total. The manufacturing of the structure foresees a modular assembly starting from extruded profiles (grade EN AW The project had to adapt to the existing TPC dimensions 6082 T6). The profiles were welded together by the and to the size of the building in BNAL. Consequently, the extruding company to form sandwich panels. Then the clearance between the TPC and the cryostat inner walls was panels were machined to the design size on a large milling extremely reduced (locally about 10 mm). Since each TPC machine. In this way, all the dimensional changes caused by had to be inserted in its cryostat from one side, the shape the weld shrinkage were corrected. The panels are between tolerance of the cross section was + 5 mm and the 4 m and 6 m long and about 4 m wide. The thickness of the straightness + 10 mm on the total 20-m long structure. sandwich is 170 mm and the walls of the profiles are 8 mm thick. Once delivered to CERN the panels were pre- The shrinkage of each weld due to the solidification of assembled to form the cryostat structure and then welded as the melted aluminium was of the order of some millimetres; shown in Fig. 5 and Fig. 6. A mixed team of CERN the same order of magnitude of the assembly tolerances. The personnel and project associates from the Pakistan Agency challenge during the final assembly was to keep the for Atomic Energy (PAEC) carried out the work. shrinkage of the welds under control. This was the only way to assemble the different parts and obtain a result within tolerances. The aluminium welds were computed, executed and controlled according to the European standards for vacuum and pressure vessels (class B for the level of defects) [7] [8]. Class B for the structural welds means very high quality and an extremely low rate of defects. This was necessary since this application requires leak tightness at cryogenic temperature. The required quality of the welds can be achieved only when the welds are executed in the optimal, flat position. A few tests were carried out with welds in different position (vertical) but the results were not correct. Therefore, it was mandatory to develop a technique and the appropriate tools to allow several rotations of the structure during its manufacturing. The tool was designed to rotate smoothly an object approximately 20-m long and 30 tonnes heavy. The first rotation, shown in Fig. 7, was the most critical since only a few welds were already made. Hinges, tack Fig. 5. Different phases of the pre-assembly of the aluminium panels. In red, welds, a provisional inside support structure, and friction in the temporary support elements used to position the parts. the locking joints of the panels kept the structure together. Fig. 7. The first part of the rotation of one of the cryostats. In red the two supporting wheels rolling on rotators. The rotation of the structure was simulated by finite critical positions to compare the measured values of element computations to check the stress in the tack welds deformation with the computed ones. The agreement and the relative movements between adjacent panels. between the two values was within 20%, with measured numbers slightly below the computed ones. This level of The number of tack welds had to be as low as possible. precision is quite reasonable for this kind of measurements. Too many tack welds would have blocked the relative position of the panels during the welding. The weld shrinkage in this case would have been blocked originating V. CONCLUSIONS unacceptable cracks in the welds. The chosen welding sequence was to complete all the The neutrino baseline foresees the design, construction first passes of the whole structure and then make the second and operation of several experiments. This work is carried ones. out by an international collaboration involving many The number of necessary rotations was quite large, institutes worldwide. between 15 and 20 per cryostat. This because the welds had An experiment represents the integration of many high to be executed alternatively on both sides of the cryostat to technology components in a highly optimized volume. Large compensate and limit the deformations. After one or two structures constitute the structural backbone of these welds on a side, the cryostat had to be turned to make systems. another one or two welds on the opposite side. Making all the welds on one side and then turning by 180 degrees and Accurate design and construction allow the possibility of making the welds of the other side would have originated combining high precision, rigidity and minimum space large out of tolerances in the straightness. occupancy. Transducers were located in strategic positions to control These structures are quite unusual, but they are designed that during the rotations the adjacent panels were not moving and manufactured following the international norms and one respect to the other. Fig. 8 shows the displacements standards currently applied for all industrial applications. recorded during the first rotation, the most critical one. They were extremely low and under control during all time. In ACKNOWLEDGMENT other words the adjacent panels staid in position. At the end of the manufacturing, the cryostat structural A warm thank to M. Nessi and his team for the help and soundness and leak tightness were tested. The contained air support during the challenging experience of building the was pumped away to have vacuum inside the cryostat and ICARUS cryostats. atmospheric pressure outside. Strain gages were glued in Fig. 8. Relative displacement of adjacent panels during the first rotation of the first cryostat. The horizontal axis reports the time. The vertical axis gives the relative displacement (in millimeters) of one panel respect to the adjacent one. Each colored line corresponds to the reading of one transducer. The values are practically constant during the whole operation. This means that the parts stay safely in place. A great acknowledgment to all the engineering, design [2] F. Close, “Neutrinos” Oxford University Press. ISBN 978-0-199- and construction personnel of CERN EN-MME group, in 69599-7, 2010. particular N. Kuder, G. Favre, P. Freijedo Menendez, V. [3] C. Rubbia, “The liquid-Argon time projection chamber: a new Maire, M. Guinchard, J.-P. Brachet, D. Lombard, P. concept for neutrino detector ” CERN-EP/77-08. 1977. Mesenge and our colleagues and friends from PAEC: D. [4] G. S. Karagiorgi, “Current and future liquid argon neutrino Makrani, F. El-Abdioui, M. F. Iqbal, A. S. Butt, G. Murtaza, experiments” AIP Conference Proceedings 1663, 100001 (2015). M. W. K. Ghauri, J. Abbas. [5] M. Bonesini and ICARUS/WA104 collaboration, “The WA104 experiment at CERN ” J. Phys.: Conf. Ser. 650 012015, 2015. [6] Eurocode 9: Design of aluminium structures ( EN 1999 ) REFERENCES [7] EN 15614‐2. Specification and qualification of welding procedures for metallic materials. Welding procedure test Part 2: Arc welding of [1] ALICE: Technical proposal for a large ion collider experiment at the aluminium and its alloys. CERN LHC. Dec 1995 - 252 pages. CERN-LHCC-95-71, CERN- [8] EN 13458 - Cryogenic vessels - Static vacuum insulated vessels. LHCC-P-3.