Characterization and optimization of the CERN Secondary Emission Monitors (SEM) used for beam diagnostics

CONSEIL EUROP´ EEN POUR LA RECHERCHE NUCL´ EAIRE UNIVERSITAT POLIT` ECNICA DE CATALUNYA Characterization and optimization of CERN Secondary Emission Monitors (SEM) used for beam diagnostics. BACHELOR DEGREE IN PHYSICS ENGINEERING FINAL DEGREE PROJECT by Araceli Navarro Fern´andez Supervisors: Dr. Federico Roncarolo CERN, Gen`eve Dr. Jordi Llorca UPC, Barcelona May 2017 Abstract The European Organization for Nuclear Research, more commonly known as CERN, is one of the world’s most influential particle physics center. The organization is based in a northwest suburb of Geneva, on the Franco-Swiss border and has 22 member states. At CERN, people with different backgrounds from all around the world work together in order to understand, among other things, the fundamental structure of the universe. To do that, particles have to be accelerated up to extremely high energies in a controlled and safe way. This work has been carried out within the BE-BI-PM section, responsible of the beam diagnostics instruments that allow the observation of the transverse profiles of particle beams. Different instruments and techniques are used for this purpose but in the following pages only the monitors based on Secondary Electron Emission (SEE) phenomena will be tackled. After introducing the physics behind Secondary Emission Monitors (SEM), their performance and limitations will be discussed. This will be based on simulations and beam experiments performed during the thesis work, mostly related to the LINAC4 , a new linear accelerator just completed at CERN. The first part of this document contains an introduction to the CERN accelerator chain, with particular emphasis on LINAC4 . Chapter 2 focuses on various aspects of the particles interaction with matter, which are used to define the theory behind SEE, the main process behind the SEM’s signal generation. Chapter 3 starts presenting some examples of SEM detectors, focusing on SEM grids and Wire Scanners. This is followed with a practical example in which SEM grids were used in order to understand the flawed Beam Current Transformers (BCT) measurement of the beam transmission along the linac. Chapter 4 explains the theory defining the thermal impact of particles interacting with matter. Such a theory was used to simulate the heating of SEMs during their operation. Chapter 5 introduces the concept of beam emittance and how SE detectors can be used for its estimation. This explanation is again presented with a practical example of emittance measurement in the LINAC4 3 MeV Test Stand. Chapter 6 gives some insights on the physics behind electron stripping and CFI injection, followed with the description of the stripping system that will be installed at the exit of LINAC4 in order to inject protons into the Proton Synchrotron Booster (PSB). The main part of this chapter will present the experimental results of a series of experiments performed at the LINAC4 Half Sector Test (HST) facility by means of the H0/H−monitors. 1 Contents Abstract.......................................... 1 Introduction 3 CERNAcceleratorComplex............................... 3 LINAC4Project ..................................... 4 Interaction of particles with matter 5 Energy loss by interaction with electrons . . . . . . . . . . . . . . . . . . . . . . . . 5 The Bethe-Bloch theory and Range . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 SecondaryElectronTheory ............................... 8 SEY Dependence with angle of incidence . . . . . . . . . . . . . . . . . . . . . . . . 10 DeltaRays ........................................ 10 Backscattering ...................................... 11 Multiple Coulomb Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Secondary Emission Monitors (SEM) 13 WireGrids ........................................ 13 WireScanners ...................................... 14 Other Secondary Emission Monitors . . . . . . . . . . . . . . . . . . . . . . . . . . 15 SignalGenerationinSEM. ............................... 15 Profile monitoring with SEM grids . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Emittance growth due to Multiple Coulomb Scattering . . . . . . . . . . . . . . . . 18 Study of the beam induced wire heating 20 WireHeating....................................... 20 RadiativeCooling .................................... 20 ThermoionicCooling................................... 21 ThermalConduction................................... 22 Temperaturesimulations................................. 23 Transverse beam emittance and Slit-Grid system measurements 25 Basics of Beam Dynamics and Phase Space . . . . . . . . . . . . . . . . . . . . . . 25 Emittance measurement with Slit-Grid system . . . . . . . . . . . . . . . . . . . . 27 Emittance measurements at 3MeV Test Stand . . . . . . . . . . . . . . . . . . . . . 27 Emittance measurements accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 LINAC4 ’s H0H−Monitors 33 Charge Exchange Injection (CEI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 StrippingFoil....................................... 34 H0H−MonitorsinPSBHST.............................. 35 Electronics conceptual design and available signals. . . . . . . . . . . . . . . . . . . 37 Measurementsprocedure. ................................ 38 Oasis Signals and first conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Digitized signals and calibration factor. . . . . . . . . . . . . . . . . . . . . . . . . 41 Reliability of k factor for H0signals........................... 42 Conclusions 46 2 Chapter 1 Introduction CERN Accelerator Complex The CERN accelerator complex consists in a succession of many different types of linear and circular accelerators and interconnecting transfer lines that allow particles to achieve high energies. Each machine boosts the energy of the bean of particles before injecting them to the next machine in the sequence. In Fig. 1 we can see a scheme of the complex that culminates with the Large Hadron Collider (LHC), a 27 km tunnel where particle beams are accelerated up to the record energy of 6.5 TeV per beam. Figure 1: CERN Accelerator Complex At the beginning of the chain the protons are generated by a source, which consists in a bottle of hydrogen. After electron stripping by high electric fields, the resulting protons are extracted towards the first accelerator, LINAC2. This accelerator, about 80 m long, brings the particles up to an energy of 50 MeV. The protons are then transferred to the Proton Synchroton Booster (PSB) where they are accelerated up to 1.4 GeV and afterwards they are sent to the Proton Synchrotron (PS). The PS is a 628 m in circumference and the particles reach a 99.9 of the speed of light. After reaching 26 GeV in the PS, the protons are injected in the Super Proton Synchrotron (SPS) and accelerated up to 450 GeV. Finally, they are transferred to the two LHC rings. In one of the rings particles circulate clockwise whereas in the other they circulate anticlockwise. The counter rotating rings cross over in the four detector cabins (ALICE, ATLAS, CMS, LHCb) where particles collide. 3 The complex of the CERN accelerator is very versatile and far from being just the injectors of the LHC. Some of the machines have their own dedicated experimental areas to explore a wide range of physics phenomena. The complexity of these machines and experimental areas makes it impossible to explain all of them. In the following section we are going to give some insights of LINAC4 , which has been the context of all our simulations and measurements. LINAC4 was conceived to replace the old LINAC2. Its construction and commissioning up to its to energy of 160 MeV were completed at the beginning of 2017 and its connection to the PSB is foreseen for 2019. LINAC4 Project The source and the linac are essential for determining the LHC beams quality, since they determine the maximum beam brilliance, which can be seen as the maximum density of particles that can be brought into collision in a collider. Such a brilliance, that after the linac can only be deteriorated and not increased, is achieved by find the right compromise between maximum beam intensity and minimum transverse/longitudinal beam dimensions. More than 10 years ago, in order to address the LINAC2 ageing and to improve the beam brilliance, it was decided to design and build L4, an H−linear accelerator, able to accelerate pulses of up to 1014 particles to 160 MeV. Figure 2: LINAC4 schematic layout The LINAC4 layout is shown in Fig. 2. The chosen sequence of accelerating sections is quite standard for modern pulsed linac designs. The particles are produced in the ions source which is followed by a Low Energy Beam Transport (LEBT), needed to match the beam with the Radio Frequency Quadruple (RFQ) cavitiy. The RFQ is followed by a chopping line and a sequence of three accelerating structures. These three structures bring the energy up to 160 MeV and they are a Drift Tube Linac (DTL), a Cell-Coupled Drift Tube Linac (CCDTL) and a Pi-Mode Structure (PIMS). They increase the particles’ energy up to 50 MeV, 100 MeV and 160 MeV respectively. More details can be found in [1]. Along the 86 m of the new linear accelerator, a considerable amount of detectors are installed in order to monitor and control the beam. In the following pages we will focus on the transverse beam profile and emittance monitors, giving special attention to the SEM grids and Wire scanners (Chapter 3) and slit-grid system (Chapter 5). More details about all the other LINAC4 beam diagnostics can be found in [1]. 4 Chapter 2 Interaction of particles with matter When a particle propagates through matter, it will have certain probability to interact with the nuclei or with the electrons present in that material. The probability that the particles have of interacting with the medium, either with the electrons or the nucleus, is represented by what is called the cross sections. The total interaction cross section will be the sum of the cross sections of the individual possible interactions. As we can see in Fig. 3, when a charged particle travels through matter it predominantly interacts with the electrons of the medium, either by Coulomb force interactions or collision interactions. Interactions with the core of the nucleus are also possible (i.e. Rutherford scattering) but much less frequent. In the following chapter we shall focus our analysis on non-relativistic or near relativistic ions interacting with solid targets. Due to the range of energy we will be working on, we are going to focus in the interactions that an incident ion has with the electrons in the medium. Figure 3: Nuclear and electronic stopping power for protons in aluminium versus particle energy per nucleon. From [2] Energy loss by interaction with electrons The incident particle or ion will transfer energy to the electrons in the solid when traveling through it. These electrons can be either excited to higher energy levels or gain enough energy to escape from the solid. As this happens, the incident particle will lose energy, most frequently the energy losses are small and this is usually described by the mean differential energy loss dE/dx (or by the stopping power S=-dE/dx). For a given incident particle and target, this energy loss has been found to be very dependent with the velocity of the incident particle. Unfortunately there is no single theorem that can describe the energy loss at all range of energy. Depending on the energy range of the incident particle, some properties from the particle and the target become relevant. The energy loss of a muon in copper is 5 illustrated in Fig. 4. The pattern is rather complicated but allows defining three major energy ranges [3]: Energy [MeV/amu] Gamma factor γ Normalised momentum βγ Framework <10 <1.01 <0.15 Lindhard 10 −1061.01 - 1000 0.15 - 1000 Bethe - Bloch >106>1000 >1000 Radiative Losses Table 1: Energy range of the incident particle and its corresponding framework In the mid-energy range the electrostatic stopping power is well defined by the BetheBloch theory, which treats the exchange of energy between the incident particle and the atoms as the scattering of a charged particle from an isolated atom. At lower energies the velocity of the incident particle becomes comparable or even lower than the target electrons’ and it is required to take into account interactions between atoms as it is done with Lindhard theory. In ultra-relativistic energy range, radiative processes such as bremsstrahlung become the dominant contribution to the energy loss. Figure 4: Electronic stopping power for positive muons on copper as function of the particle kinetic energy. The solid curve represents the total stopping power From [4] The Bethe-Bloch theory and Range The Bethe-Boch theory describes the mean rate of energy loss (stopping power of the material) for energies of the incident particle from 10 −106MeV/amu and can be described as [5]: −dE dx =Kz2Z A 1 β21 2ln2mec2β2γ2Tmax I2−β2−δ(βγ) 2(1) Where Z and A are the atomic and mass number of the material. Zeis the charge of the particle, K/A is 0.307 MeV g−1cm2and the δ(βγ) is a predetermined density correction necessary for high energetic particles, units are MeV cm2g−1. I is the mean excitation energy for the given material, and varies from a few eV for materials with low Z to hundreds of eV for materials with high Z and can be calculated with the empirical formula: 6 I= 16Z0.9(2) Tmax represents the maximum energy that can be transferred to a particle in an elastic collision. It can be calculated as: Tmax =2mec2β2γ2 1 + γmec2 Mc2+mec2 Mc22(3) Where γand βare the relativistic parameters, M is the mass of the projectile and meis the electron mass. Figure 5: Bragg peak of 100 MeV protons in Tungsten. Energy deposition increases with distance, reaching a maximum at around 0.8 cm. The energy deposition is not constant, it has a peak because interaction cross section increases as the charged particle energy decreases. In Fig. 5, we can see an example of the Bragg curve that plots the energy loss of ionizing radiation during is travel through matter. If we can just take into account the particle loses because of ionization and atomic excitation we can define the range as the distance traveled by particles in matter until their energy is (almost) zero. The range depends on the type of particle, on its initial energy and on the material it goes through. For heavy particles it can be calculated integrating the Bethe-Bloch formula [6]. After integration, the relation between the range and energy for particles above few Mev is: R(E)∝E2(4) Therefore, the range will be greater as the energy of the particle increases. For electrons, the effect of multiple scattering induces chaotic trajectories and the range is very difficult to estimate. An excellent review of all electron range-energy was recollected by Katz and Penfold who proposed the following empirical relationship for energies from 0.01 MeV to 3 MeV [6] : R(E) = 41213 27 A ZEn(5) With n= 1.265 −0.0954ln (E), E the energy of the electron, A and Z, respectively the number of nucleons and the atomic number of the material. R is given in mgcm2. 7 Secondary Electron Theory When a particle passes through the interface of a material, it will transfer energy to the electrons in the medium. Depending on the energy these electrons get, they can be excited to a higher energy level or gain enough energy to be emitted from the material, this emission process is known as Secondary Electron Emission (SEE). This is a surface effect so we can find this effect when the particle either enters or exits the material. Usually the electrons in the outer shells are the ones to be ionized but ionization of the inner shell is also possible although less probable. The SEE process can be provided in three steps [7]: Figure 6: Ion induced secondary electron spectra for a variety of materials. In this case incident ions are protons at 500 keV. The integral of each curve in energy gives the total secondary electron yield. From [8] The first step is the creation of the secondary electron in the bulk of the material. The minimum energy you need for creating a SE is the one required to excite the electrons above the Fermi level to the conduction band. If the projectile is an ion containing electrons these electrons can be stripped off and also produce further ionization, but if the electrons from the ion are scattered off the material, they cannot be counted as secondary electrons. The second step in the SEE process could be the diffusion in the material. When the secondary electrons travel through the material, they lose their energy. The greater the energy lose the smaller the distance they will travel. Finally, the last step would be the emission process. In order to be emitted the secondary electrons have to have enough energy to overcome the surface potential, which is characterized with the work function eφ, so the barrier height would be W=EF+φe, with Ef the fermi level and φthe surface potential. The emission process is a directive phenomenon, as a consequence apart from having enough energy the electrons must have a velocity vector laying inside a escape cone with a maximum angle θmax normal to the surface: 8 pulse of 100 µs pulse length and 40 mA peak intensity. Above this limit the wires heating can end up in wire melting or sublimation. For the same reason, these devices cannot be normally used in accelerator rings where the particles circulated for multiturns consecutively. Another class of scanners, normally labeled as Fast Wire Scanners are used in the CERN synchrotrons (form the PSB to the LHC). They consists of fast moving mechanisms (up to 20 m/s) sweeping into the beam in few milliseconds. The beam profile amplitude signal is in this case normally extracted not by secondary emission on the wire, but from the monitoring of the secondary showers of high energy particles generated by the beam-wire interaction. Other Secondary Emission Monitors Apart from wire grids and scanners, other SEMs are used at CERN. In particular, the Super Proton Synchrotron (SPS), the second largest ring after the LHC, is equipped with a number of detectors dedicated to quantify and characterize the beams that are extracted to the SPS North Area facility for fixed target esxperiments: - BSPV/BSPH or BSMV/BSMH, used to calculate the beam position - BSPV/BSPH, which are plates split in two halves, used to measure the beam position - BSI, which are plates used to measure the beam intensity. - BSH or BSHS/BSVS, which are plates with a circular or rectangular hole int he middle, used to measure the beam halo. Signal Generation in SEM. All the phenomena discussed in the previous chapter, can be used to estimate the electric signal that is generated by a particle beam on a wire, which is part of detectors like wire grids or wire scanners, designed to sample and reconstruct the transverse beam profile. A general expression for the charge generation in the material per incident ion projectile is [18]: Q(e/Proj) = −Nelec ·(1 −BS)·µ+2·SEYp·η+2Nelec ·SEYe·µ+Nelec ·SEYBS +YD(17) with Nelec number of electrons, ηthe proportion of incident projectiles exiting the material, µthe number of incident electrons that do not cross the material and BS the fraction of back scattered electrons. The first term quantifies the negative charge left on the material due to the incident projectile’s electrons, that is the number of electrons from the projectile deposited in the material. The second and third terms correspond to the charge generated due to the SE phenomena due to the electrons and nucleus of the incident projectile. These two terms come along with a factor two because SEE is a surface phenomenon, happening at both the incident and exiting surfaces. The third term accounts for the SE generated due to the back scattered electrons. In this case there is no factor two because only the incident surface contributes to the charge generation. Finally, YDis the charge created by secondary electrons generated due to delta rays. In most of the cases we have considered BS,SEYBS and YDzero because of the smallness of these terms in comparison with the remaining ones. To give some numbers, consider the LINAC4 H−ions as incident projectiles, for which Eq. 17 becomes: Qe/H−=−2·µ+ 2 ·SEYP·η+ 4 ·SEYeµ(18) 15 The expected net charge generated per incident ion is showed in Table 5 for different beam energies and wire materials. The values of ηand µare material dependent and their values have been calculated using FLUKA. Considering, as an example, the case of graphite wires, the range of protons at 160 MeV is 11.6 cm and 2.5 cm at 60 MeV. It is safe to assume in this case that almost all the incident protons go through the wire (η≈1 in all cases). Beam Energy [MeV] Electron Energy [keV] Carbon (33 um) Tungsten (40 um) Titanium (40 um) 60 32.69 -1.93596 -1.78510 -1.87795 100 54.49 -1.95707 -1.84995 -1.91700 120 65.38 0.03743 -1.86760 -1.92729 160 87.18 0.030411 -1.89055 0.05946 Table 5: Values of the charge [eV] generated in wires of different materials for each H−incident particle. The electrons energies are from around 30 keV to 90 keV and, in this case, the electron range in graphite varies from 11 µm up to 80 µm respectively. This implies that for 60 MeV H−beams most of the electrons will remain in the material, generating a -2 charge, while for 160 MeV H−ions some of the electrons will pass through the material and the charge on the wire will be only due to SEE. As a last remark on the use of wire grids for monitoring H−beams, it must be noted that in some cases (namely when the beam energy and the wire material/diameter are such that the impinging particles are stopped into the wire) it is useful to try suppressing secondary emission in order to have the net charge dominated by the protons and/or electrons deposited charge. Since the secondary electrons energy is normally only few eV, this can be achieved applying a relative low bias (e.g. +30V) on the wire. Another common method is to install negativly biased repeller frames before and/or after the wire detector (see Fig. 7 and related text). It has also been experimentally proved that secondary electron suppression can be achieved using a magnetic field [19], which is re-directing the secondary electrons into the material. Profile monitoring with SEM grids In this section we will present the simulation of beam profile measurement with a SEM grid consisting of 40 wires 500 µm pitch and two wire types, 40 µm tungsten wire and 33 µm diameter carbon wire. We will consider an incident LINAC4 H−beam with the parameters listed in Table 6. In the previous section, we saw the effect one H−ion has when going through a wire. Now, instead of one particle, we consider a beam pulse populated by 1014 particles. By multiplying the results found for one particle, presented in the previous section, by the number of particles that reach each wire we can re-construct the beam profile via the wire signals. Figure 9 shows the expected signal for both tungsten and carbon wires (Left 60 MeV, Right 160 MeV). 16 Parameter Value Units Top Beam energy 160 MeV Beam Pulse length 400 us Repetition rate 1.2 Hz Average pulse current 40 mA Sigma x * 2 mm Sigma y * 1 mm Angular dispersion 0.49 mrad Table 6: Main LINAC4 beam parameters. *These two values have been selected for the purpose of this example For 60 MeV H−ions we can see that the signal produced by a carbon wire, in absolute value, is bigger than the signal on the tungsten wire. In both cases the H−electrons are deposited on the wire and SEE occurs. In the case of tungsten more secondary emission is expected and therefore the positive charge left behind due to this phenomenon is expected to be bigger than in the case of carbon. On the other hand, for 160 MeV the signal in carbon wires is positive because most of the H−electrons are able to exit the material. That leads to a positive signal which in absolute value is smaller than the one in the tungsten wire, which is still generated mainly by the incident H−electrons. Figure 9: Expected current for Tungsten and Graphite SEM grids. Left, 60 MeV incident H−ions. Right, 160 MeV incident H−ions. Both cases calculated with the parameters of Table 6. 17 Emittance growth due to SEM The aforementioned detectors are intercepting devices which means that have an effect on the beam. As already discussed in Chapter 2, each time a particle traverses the detector’s material, it experiences Multiple Coulomb Scattering. Considering the particle ensamble, this results in an emittance increase at the monitor exit and, in case of large scattering angles, to beam losses. Such an effect is well know in literature, here we just present the result of FLUKA simulations characterizing the scattering angle. As an example, we considered a SEM grid which consists of 11 tungsten wires of 40 µm diameter spaced 500µm, used to monitor a beam with initial σx= 2 mm and σy= 1 mm. FLUKA does not include the possibility of easily simulating H−ions so we have studied independently the effects of the detector on electrons and protons. Figure 10: Histogram of protons’ angular distribution with respect to the beam direction 10 cm after a SEM grid (11 tungsten wires, 40 µm thick, 500 µm pitch). Red line being the Gaussian fit of the angular distribution. For 60 MeV protons. Figure 11: Histogram of protons’ angular distribution with respect to the beam direction 10 cm after a SEM grid (11 tungsten wires, 40 µm thick, 500 µm pitch). Red line being the Gaussian fit of the angular distribution. For 160 MeV protons. 18 Figure 10 and 11 are a representation of the angular distribution with respect to the beam direction 10 cm after the SEM grid. We can clearly observe how the angular spread is bigger for 60 MeV protons (10) than for 160 MeV protons (11). This is not surprising as the probability of interactions for low energetic particles is higher than the one for high energetic particles. In this range of energies the electrons of an H−ion would be stopped in the tungsten wires, so an study of the scattering angle is not easy in this case. If we consider more energetic ions it’s observed that changes in electrons’ trajectories are more drastic than changes in protons’ trajectories. 19 Chapter 4 Study of the beam induced wire heating During the operation of the intercepting devices as wire grids and scanners, the energy deposited by the beam into the wire material translated into a temperature increase. This can generate the loss of electrons by thermionic emission (thus affecting the wire signal) and permanently damage the wires. Accounting for the various cooling mechanisms occurring during and after the beam passage, this results in thermal cycles that are important to address in order to design the detectors and set limits for their operation. In other words, the selection of the wire material and the the beam range of work (beam size, intensity, duration) are limited, by thermal reasons. A good understanding of the variation of the temperature and its consequences is therefore necessary [20]. The wire thermal behavior can be written as: ∆TTotal = ∆THeating −∆TRadadiative + ∆TThermoionic + ∆TConductive (19) The beam pulse causes the temperature increase in the material of the detector and it is followed by three cooling effects, radiative cooling, thermionic cooling and conduction cooling. The equations used to describe these effects are nonlinear due to the dependence of the specific heat with temperature and due to the nature thermionic cooling. As a result, in order to follow the thermal evolution, such equations have to be solved numerically. Wire Heating Given a beam pulse populated with NT ot paritcles distributed in the horizontal and vertical coordinates according to a Gaussian shape of width σxσy, the temperature variations of a material sample with surface ∆Sand volume ∆Vcan be written as: ∆THeating =Ni ∆V·Cp (T)·ρ·dE dx (20) where Cp(T) is the specific heat capacity of the material, ρis the density, dE/dx the stopping power of the particles in the material in units of hJ m2giand Ni= ∆S·NTot 2π·σx·σy ·e−1 2·x σx2+y σy2 (21) refers to the fraction of NT ot particles hitting the volume ∆V. Radiative Cooling In first approximation black body radiation is the dominant cooling effect and it is described by Stephan Boltzmann’s law. The heat radiated from the material surface is proportional to the forth power of the temperature and it is differentiated with the black body radiation by a factor called emissivity , that in most cases depend on the temperature. Discretized to our interests the model can be represented as: 20 ∆TRadiative =∆S·σSB ·(T)·T4−T4 0 Cp (T)·∆V·ρ·∆t(22) where, in addition to the variables defined above, (T) is the emissivity, and σSB StefanBoltzmann’s constant (5.6704 ·10−8J/sm2K4). Figure 12: Left, maximum temperature evolution on a carbon wire for a SEM grid. Right, maximum temperature evolution on a Tungsten wire for a wire scanner at three different velocities. All cases for a 3 MeV, 40 mA, 100 us pulse. Thermoionic Cooling With the increase of temperature, thermal energy is transferred to the electrons of the material. When the energy transferred to the charge carriers overcomes the work function they escape the material, leaving a charge equal in magnitude and opposite in sign. The current density emitted by the material is described by Richardson-Dushman [21]: Jth =A·T2·e−φ(T) KT (23) where φis the binding potential (or work function) and A is Richardson’s constant, theoretically equal to: A=4πmek2qe h3≈1.20173 A m2K2(24) with methe electron mass, k Boltzmann’s constant, qethe electron charge and h Plank’s constant. This effect, called Thermionic emission, thus generates a net positive charge that adds to SEE. In order to see how this effect can affect our SEE measurements, the thermionic current for different materials and temperatures was calculated as shown in Table 7). In most of the cases the estimated maximum temperature is below 1500 K and thermionic emission 21 Temperature Carbon Tungsten Titanium Alumina 300 0 0 0 0 1000 0 0 0 2.24e-14 1700 1.99e-9 1.59e-9 4.21e-9 4.64e-8 2600 1.94e-4 1.54e-4 2.73e-4 4.02e-3 3100 7.02e-3 5.62e-3 8.76e-3 1.38e-2 3500 6.14e-2 4.91e-2 7.11e-2 1.08e-1 4000 5.17e-1 4.14e-1 5.59e-1 8.31e-1 Table 7: Current due to Thermoionic emission in several materials. The numbers in red correspond to cases where the material is already melted. Melting points being 3773 K, 3695 K, 1941 K and 2345 K for Carbon, Tungsten, Titanium and Alumina respectively. can be neglected. Above this temperature the current density increases quickly but the current due to secondary emission is still several orders of magnitude higher (for Tungsten with a LINAC4 beam it was around 1 mA). For temperatures of the order of 2500-3000 K the thermionic emission and SEE magnitudes become comparable. Since these temperature levels are dangerously close to the material melting point, both Tungsten and Carbon wires are not suitable for such high beam powers and thermionic emission will not be considered in the signal calculations discussed later. On the other hand, we will consider the material cooling due to the thermionic emission, which can be expressed as: ∆TThermionic = ∆S·(φ+ 2σBT)·Jth Cp (T)·∆V·ρ·∆t(25) Thermal Conduction Due to microscopic collisions of particles and movement of electrons within a body, heat flows from hotter to colder parts of the material. In our case the part of the material that is heated by the beam core reaches a higher temperature than the rest of the material and therefore thermal conduction helps reducing the temperature in the hottest spots by transferring the heat to the colder ones. This phenomenon can be described by the Fourier formulation detailed in [20]. Nevertheless, the contribution of conduction in the cooling effects is very small compared with the radiative cooling. Because of this small contribution and the considerable increase on the simulation time, all the examples presented in this document do not consider conduction effects as a cooling effect. 22 Temperature simulations To estimate the heating and cooling of the detectors we assumed that the energy deposition is constant in the direction of the beam propagation. This approximation is not accurate for cases in which the Bragg peak occurs within the material thickness, because of low particle energy and/or particular dense (or thick) materials. On the other end, the approximation holds for the cases considered below. Table 8 shows the maximum temperatures reached in SEM grid detectors due to several incident beam types. The results show that materials with higher density are not well suited for some beam characteristics. Gold has a relatively low melting point (1437 K), which is reached in the first two cases. The rest of the materials do not reach their melting point temperature but Titanium and Tungsten are close to it in the second case. Even though Graphite has a lower SEY compared with the rest of the materials, its thermal properties make it one of the most common materials for this type of detectors. As we can see in Fig. 12 right and Fig. 13, the equilibrium is reached after few pulses. Obviously, this kind of calculations are of primary importance at the moment of designing and specifying the functional specifications of a wire detector, which must feature an equilibrium temperature safely lower than any damage threshold. Protons 60 MeV, I = 40 mA, Pulse length = 400 µs Material Tmax Bragg peak depth [cm] Energy Deposition MeV cm2/g Density gcm−3 Graphite 755 2.03 9.642 1.7 Titanium 956 0.94 7.510 5.506 Gold 2051 0.35 5.185 19.30 Tungsten 2155 0.35 5.275 19.35 Protons 60 MeV, I = 65 mA, Pulse length = 600 µs Material Tmax Bragg peak depth [cm] Energy Deposition MeV cm2/g Density gcm−3 Graphite 1070 2.03 9.642 1.7 Titanium 1652 0.94 7.510 5.506 Gold 3943 0.35 5.185 19.30 Tungsten 3374 0.35 5.275 19.35 Protons 160 MeV, I = 40 mA, Pulse length = 400 µs Material Tmax Bragg peak depth [cm] Energy Deposition MeV cm2/g Density gcm−3 Graphite 598 11.6 4.655 1.7 Titanium 689 5.58 3.705 5.506 Gold 1379 1.85 2.657 19.30 Tungsten 1567 1.85 2.706 19.35 Table 8: Maximum temperature reached by SEM grids of different materials and for different beam types. Energy deposition and Bragg peak depth calculated using PSTAR Even though we only discussed thermal issues, there are other processes that can lead to a wire damage, such as brittle failure, plastic failure or thermal fatigue. More details can be found in [22]. 23 Figure 13: Temperature evolution of a tungsten wire due to 160 MeV proton beam, 100 µs 24 Figure 20: CST Particle Studio simulation of the effect of polarized rings on secondary emission surpression. Left, no polarization on the rings. Right, Ring 1 = -600 V, Ring 2: -1200V Now lets go back to our second objective, that is, trying to suppress secondary electrons emitted at angles around 90 and 270 degrees with respect to the angle of the particle’s incidence. These electrons are going to collaborate with the SE problem and also generate some cross talk with the neighbor wires [32]. In this case both positive and negative bias can be used to suppress SEE. If we use a positive bias in a wire we are actually increasing the work function and making it harder for secondary electrons to escape from the material. The problem with this method is that wires can also attract some undesired background electrons. The other option is to use negative bias in nonconsecutive wires. That allows us to repel background electrons and at the same time surprises secondary emission on the non polarized wire. Figure 21: CST Particle Studio simulation of the effect of secondary emission supression with polarized wires. In this case the wires on the extremes are polarized at -10 V. The scale is aplicable in all the previous cases. 31 In Fig. 21 we can see an example of 40µm Tungsten wires with the central wire as an electron source and the neighboring wires polarized at -10 V. The main problem with this method is that negative polarized wires cannot be used afterwards as a source of information to reconstruct the beam profile, so the accuracy of the measurement is affected. It is necessary to mention that all the results aforementioned have not been experimentally tested yet, they are only simulation results. So before concluding in their possible utility, further tests are necessary. 32 Chapter 6 LINAC4 ’s H0H−Monitors At the moment, Linac2 is the first in the proton acceleration chain. The protons are accelerated form the few keV source to the 50 MeV LINAC2 top energy. As already discussed above, the LINAC2 will be replaced in 2019 by the new LINAC4 , which will bring H−particles up to 160 MeV. The injection into the PSB (4 rings, one on top of each other) will be performed by means of a H−charge exchange injection system, through a Carbon stripping foil (one per ring), converting ≈99% of the beam to protons. In this chapter we will talk about the charge exchange injection, what a stripping foil is and we will explain the arrangement of the H0/H−dump in the injection system. The core part of the chapter will focus on the new H0/H−monitors, their setup and the detailed explanation of their calibration procedure. Charge exchange injection (CEI) Among other upgrades, it is foreseen to increase the LHC luminosity by providing the collider with more bright (high density) beams form the injectors. At the moment, the first of the brightness limitations is represented by space charge effects in the PSB, which will be strongly reduced increasing the PSB injection energy from 50 MeV to 160 MeV. A further brightness improvement is expected from the charge exchange injection (CEI) [33]. The process consists in the stripping of electrons from H−ions, followed by the capture of the new generated protons by the circular accelerator. Having the injected and circulating beam with opposite charge polarity, will allow increasing the density of particles in the beam. The 160 MeV H−beam from the LINAC4 needs to be distributed to the 4 superposed synchrotron rings of the PSB. After the beam is deflected to the four appropriate apertures, it will be injected into the PSB by means of an CEI system. Figure 22 shows the layout of Figure 22: LINAC4 PSBooster injection Layout 33 the injection system. At each ring a set of 4 dipole magnets (BSW) will create the required injection bump and a stripping foil will convert the H−beam to H+. Four internal H0H− beam dumps will be installed downstream each stripping foil in order to absorb any residual partially stripped H0and unstripped H−. As already mentioned, the fundamental adventage of this technique is that due to the opposite charge of the injected and circulating beams the CEI can be designed for injection of succesive turns into the same phase space, increasing in this way the brightness of the beam [34]. Stripping Foil In between the two BSWs we can find a foil. This foil is in charge of stripping the electrons from the H−ions, thus generating the protons to be injected into the PSB. Commonly, Carbon stripping foils are used. Carbon foils have the advantage of being the material with the lowest Z that can be fabricated into a very thin foil, it is stable in vacuum at high temperatures and has good electrical and thermal properties [35]. After the H−beam goes through the stripping foil we can find three types of particles remaining, H+(protons), if two electrons were stripped, H0if only one electron was stripped and H−if neither of the electrons were stripped. There could also remain other types of particles due to the possibility of electron pick up. However, for energies above 100 keV the cross sections for electron pick up can be neglected. The ratio of H+after the stripping foil is given by [36]: fH+= 1 −1 σ−1,0+σ−1,1−σ0,1hσ−1,0e−σ0,1x−(σ0,1−σ−1,1)e−(σ−1,0+σ−1,1)xi(29) where σ−10, σ01, σ−11 are the cross sections of the reactions H−→H0+e−,H0→H++e− and H−→H++e−+e−, respectively. x=N0τ/A, where A is the atomic number of the Carbon foil and τis the area density. The stripping inefficiency can be expressed as: fH−=e−σ−1,0x(30) Therefore, the yielding of H0can be expressed as: fH0= 1 −fH+−fH−(31) Table 9 shows a summary of the cross sections at different energies. For a given foil thickness the stripping efficiency is higher if the energy is lower. Also, a relation between stripped electrons and foil thickness can be found. The proportion of H+and therefore the proportion of stripped electrons increases as the foil thickness increases, whereas the proportion of H0 has a maximum for a given thickness. The proportion of H−decreases with the foil thickness. Even if the efficiency is bigger for smaller energies and thicker foils, we have to pay attention to the temperature evolution of the foil and the effect it has on the beam emittance. In the charge exchange injection, both H+and H−cross the foil and deposits energy on it. The thicker the foil and the lower the energy, the greater the energy deposition and the beam 34 80 MeV 250 MeV σ−1,03.17 1.35 σ0,11.24 0.53 σ−1,10.056 0.024 Table 9: Cross sections of the reactions H−→H0+e−,H0→H++e−and H−→ H++e−+e−. Units 10−18cm2. From [36] emittance blow up. In the case of LINAC4 , several foil types are being tested. The foil will need to be about 20 mm high and 20 mm wide, with a thickness between 100 and 200 µm/cm2. During injection, it is expected to find a current of H0around 2% of the LINAC4 beam pulse current corresponding to an stripping efficiency of 98%. The level of H−coming from the stripping foil would be even lower, around 10−6%, although the current of H−can be higher because of the halo particles of the beam that miss the stripping foil. H0H−Monitors in PSB HST As discussed above, the future injection of LINAC4 ’s beam into the PSB will be based on the CEI technique. The PSB comprises 4 rings and one of this injection structures will be installed in each ring. The magnetic field of the BSWs is such that the protons reach the circulating beam, see Fig. 22. H0and H−are not going to be affected by the magnetic field on the same way as protons. H0are neutral and will not be affected, whereas H− are going to be bent in the opposite direction. These unwanted H0/H−particles will be stopped by a dump, see Fig. 23. H0/H−monitors will be installed in front of each dump in order to monitor the stripping efficiendy and so to assure a good injection. The system was tested for the first time in the PSB Half Sector Test (HST) where all the measurements explained in the following sections were taken. Other measurements results can be found in [37]. Figure 23: Mechanical representation of the H0H−dump and monitors on the beam pipe. These H0H−monitors consists in two Titanium plates of 1 mm of thickness. Each plate is divided in two halves with a cut placed on the center. This additional feature adds beam position information to the intensity measurements. The monitors were placed inside the BSW4’s magnetic field at a distance of 4 cm from the dump’s face. This distance is enough 35 to avoid the secondary electrons coming from the dump to affect the monitors’ signal. These monitors are connected to an interlock circuit, that is, if the intensity measured surpasses a safety limit, now set to 10% of the LINAC4 ’s peack intensity, the interlock circuit will stop the beam. So far this feature is not yet implemented as the monitors are still being tested. Before reaching that point, the monitors and the electronics behind them have to be calibrated and cover all their required features. Figure 24: Production drawing of the H0H−plates with dimensions and tolerances. During normal operation, the current in the H0plates is expected to be up to 2% of the nominal LINAC4 ’s current, which corresponds to around 0.5mA. Stripping degradation can be tolerated until 10% of the nominal intensity which would correspond to a signal of ≈4mA. This should be the threshold at which the interlock circuit stops the beam. In a very first approximation the expected signal per impinging H−would be 2 electron charges and 1 electron charge for H0plates. A more precise determination of the net charge left per incident particle can be found in Table 10. BS SEY BS δH[e/H−]He/H0 w S.E w.o. SE w SE w.o SE Titanium 0.23 0.0114 0.025 -1.42 -1.49 -0.61 -0.7336 Table 10: Expected net charges per impinging 160 MeV H0and H−particle in 1mm Titanium plate. The electronics readout is designed for integrating the charge in each plate between 50 ns and 1 µs. So the total number of charges read at each plate will depend on the beam intensity, beam pulse leangth and integration time. The injection time can vary between 50 ns to 150 ns. The 50 ns lower limit has been specified to be consistent with the minimum detectable signal. 36 Electronics conceptual design and available signals. The electronics after the monitors are designed to ensure a continuous measurement of the H0H−particles, to function as an interlock system to protect the dump and therefore to assure an efficient injection. Only a conceptual explanation is going to be given in this section, for more detailed description the reader may check [18]. Several versions of the electronics where tested during the measurements but all of them shared the same basic structure. Figure 25: Schematic layout for the continuous monitoring system. The system is divided in two main subsystems, the interlock circuit and the continuous monitoring system. One of the requirements is that the signal does not saturate even in the case of foil breakage. For this reason the continuous monitoring system has been split into two sub-circuits, high gain electronics and low gain electronics. With the high gain electronics the pulse length can go down to 50 ns but requires a long time for discharging the integrators. The low gain electronics allows the pulse to be longer than 70 ns, it includes two alternating fast integrators connected to a fast ADC that converts the integrated charge into digital samples spaced 1 µs. In this continuous monitoring circuit, shown in Fig. 25, also an unprocessed signal is at our disposal. This signal is connected to an oscilloscope channel (named OASIS) after passing through a fast amplifier. The interlock circuits on the board, see Fig. 26, comprises a slow and low gain amplifier fed with the outputs of each plate. The integrator will be connected to a comparator (or two consecutive comparators depending on the version of the circuit) that has as a second input a predefined reference. In the following sections we are going to talk about the analysis of the signals coming 37 from the OASIS scope, the ADC converter. Not a lot of details concerning the interlock signal will be given. Figure 26: Schematics of the interlock circuit. Measurements procedure. The main goal of these measurements was to obtain a calibration factor which, independently of the intensity or type of beam, was able to relate the signal on the plates with the number of particles reaching the dump. For that, several measurements with different beam lengths and intensities were taken. The procedure of the measurements was always the same. The beam was scanned along the four plates from left to right giving special attention to the points where the beam was between two plates and when it was at the center of one plate. The sequence of the measurements was the one indicated in Fig. 27. In the following pages we will refer to those points as H0L, H0LH0R, H0R, H0RHML, HML, HMLHMR, HMR respectively. Figure 27: Schema of the order followed during the measurements. In order to find the calibration factor the measurements were done with an H−beam impinging directly to the plates rather than first going through the stripping foil. This ap38 proach was chosen in order to reach all the plates with a beam of the same characteristics. If the beam goes through the stripping foil, the particles reaching the monitors are going to be mostly H0, in a quantity dependent on the foil’s efficiency. At the same time, if the incident particles were H−it was possible to reach all the plates by choosing an appropriate magnetic field value for the BSW3 and BSW4 magnets. On the contrary, if the particles were H0, only the first two plates were reachable. Oasis Signals and first conclusions When referring to OASIS signals, we are actually referring to the analytical signals coming from the plates. Only a fast amplification is applied to these signals before going to the scope. Image 28 shows an example of the type of signal read out by the plates. Two different procedures were followed in order to analyze these signals. The first consisted in calculating the signal integral in time. This allow us to obtain the total number of charges detected by the plate in one beam pulse. Afterwards, this value was normalized by the total number of charges detected by the closest Beam Current Transformer (BCT). The second method consisted in calculating the mean value of the signal along the pulse and normalizing it by the current measured by the BCT. Both types of analysis seemed to give pretty similar results. The integral method was finally chosen for practical reasons. Figure 28: Example of OASIS scope traces of a 4 mA,5 µs H−beam. In this case the beam was placed on the center of the HMR plate. Figure 29 summarizes the response of the plates to different intensity beams. In this particular set of measurements, we were interested in understanding, qualitatively, how the plates responded to different beam intensities and also to see, if a different response was appreciated between the different plates. In order to appreciate this relative difference, Fig. 29 is also normalized by the value of signal measured in H0L plate. From this graph we can see the effect of the plate’s gaps (1 mm). In the majority of the measurements it is also evident that the H−plates provided a slightly higher signal than the H0ones. As mentioned in Chapter 2, it has been experimentally demonstrated that secondary emission can be suppressed by using magnetic fields. In our scenario, we scan the beam along the plates thanks to BSW3 and BSW4’s magnetic fields. Particularly, for the beam to reach 39 Figure 29: OASIS measurements for different beam intensities. Beam scanned using both BSW3 and BSW4. X axis, position of the beam at the moment of the data extraction. Y axis, each point represents the sum of all the signals of the plates for that particular intensity and position normalized by the number of charges measured by an upstream BCT and also divided by the signal on the H0L plate. H0R plate, no magnetic field was needed. In order to reach H0L a very small magnetic field was necessary whereas if we wanted to reach the H−plates it had to be higher. In the plates the signals are mostly negative due to the charge deposition of the incident H−ions. Due to SE, electrons are emitted from the surface making the signal to become less negative. If SE is suppressed by magnetic fields it would explain why signals on the H−plates are higher than the H0plates’ signals. In order to check this theory, several measurements were done scanning the beam only with BSW3 and keeping BSW4 to a constant value. The monitors are placed inside BSW4’s magnetic field so keeping BSW4 at a constant value of around 0.12 Tm would make SE the same for all the measurements. Figure 30 shows the summary of these measurements. In this figure it is clearly appreciated how the signal difference between the H−and H0plates is no longer existent, thus confirming the SE suppression by the BSW4 field theory. Figure 30: OASIS measurements for different beam intensities. Beam scanned with BSW4’s magnetic field constant. 40 References [1] L. Arnaudon, et al. Linac4 Technical Design Report. CARE-Note-6006-22-HIPPI, CERN, Switzerland. [2] NIST, National Institute of Standars and Technology. Physical Meas. Laboratory. [3] A. Missimiliano Schiavi, Study of Laser produced Plasmas by X-Ray and Proton Radiography, Plasma Physics Group, PhD, London,2003. [4] D.E. Groom, et al., Muon Stopping Power and range Tables 10 MeV100TeV, Atomic Data and Nuclear Data Tables, Vol 76., No, 2, July 2001. [5] K. Nakamura et al. (PDG), Passage of particles through matter, February 16, 2012, JP G 37 075021. [6] W. Frass,Passage of particles through matter, Oxford physics 2009. [7] D. Krammer, Design and Implementation of a Detector for High Flux Mixed Radiation Fields, Department of physics, University of Livec, Phd, Czech Republic, 2008. [8] D. Hasselkamps, et al. Ion induced secondary electron spectra from clean metal surfaces, Nuc. Instrum. and Methods Phys. Res. B 18 (1987) 561. [9] http://www.fluka.org/fluka.php [10] G.Lakits, et al., Statistics of ion-induced electron emission from a clean metal surface, American Institute of physics, 22 June 1989. [11] E. Borokovsky, et Al. The secondary-electron yield measured for 5-24 MeV protons on aluminium-oxide and gold targets, Nuclear Instr. and Methods in Physics Research, 1988. [12] P.Thieberger, et al. Secondary-electron yield and their dependence on the angle of incidence on stainless-steel surfaces for three energetic beams, Physicak Review, Volume 61, 04291 [13] http://www.cloudylabs.fr/wp/nombre-de-delta-ray/ [14] Allan J. et al. Backscattering and secondary electron emission from metal targets of various thiknesses, Washinton, D.C., April 1965. [15] V.L. Highland. Some practical remarks in multiple scattering, Nucl. instr. and meth, Vol.129 p.497, 1975. [16] G.R. Lynch and O.I. Dahl. Approximation to multiple Coulom scattering, Nucl. instr. and meth, Vol. B58, p.6, 1990. [17] U.Raich, Beam Diagnostics, CERN, Geneva, Switzerland. [18] F.Roncarolo, et al. H−and H0beam current monitor electronics., PSB-BC-ES-0001, CERN, Geneva, Switzerland. [19] L.Wang, et al. , Suppression of secondary emission in a magnetc field using a swatooth and isosceles triangle surface, SLAC, Menlo Park, CA, USA. 47 [20] M. Sapinski, Model of Carbon Wire Heating in Accelerator Beam, CERN, Geneva, Switzerland, 2008. [21] Gorazd Planinˇsiˇc, Thermionic Emission,University of Ljubljana, Department of physics, April, 2008. [22] M. M¨oser, Fractography with the SEM (Failure Analysis), 2007. [23] F. Roncarolo, Accuracy of Transverse Emittance Measurements of the CERN Large Hadron Collider, Ecole Polytechnique federale de Lausanne, PhD Lausanne, 2005 [24] J. Buon, Beam Phase space and Emittance, LAL, 91405 Orsay, France [25] F.Roncarolo, et al., Accuracy determination of the CERN LINAC4 emittance measurements at the test bench for 3 and 12 MeV. CERN Geneva Switzerland. 2010. [26] J. Lettry, et al. , LINAC4 H* ion sources, Review of scientific instruments 87,02B139, 2016. [27] M. Lutz, CERN LINAC4 - The Space Charge Challenge, University of Berlin, PhD, 2013. [28] U. Raich, et al., Commisioning of the CERN LINAC4 wire scanner, wire grid and slitgrid monitors at 3 and 12 MeV, Proceedings of Linac, 2014, Geneva, Switzerland. [29] B. Cheymol, et al., Scattering of H−Stripped Electrons from sem grids and wire scaners at CERN LINAC4 , Proceedings of LINAX2012, Tel-Aviv, Israel. [30] https://www.cst.com/products/cstps [31] B. Cheymol, et al. , Comissioning of the LINAC4 ion source transverse emittance meter, CERN-BE-2010-014 BI. [32] F. Roncarolo, et al., Effect of H−stripped electrons on the LINAC4 profile measurements, CERN-ACC-NOTE-2014-008 TECH [33] C.Hojvat, et al., The multiturn charge exchange injection system for the fermilab booster accelerator, IEE Transactions on Nuclear Science, Vol. NS-26, No3, June 1979. [34] T. Kawakubo, et al., The H−exchange injection system in the booster of the KEK 12 GeV proton Synchrootron, National laboratory for high energy Physics, Japan. [35] V. Lienchtenstein, Carbon Stripper FoilsPreparation and quality, Kurchatov Institute, 123182 Moscow, Russia. [36] Huang Ming-Yang, et al., The effects of injection beam parameters and foil scattering for CSNS/RCS, institute of high energy physics, China, 2012. [37] B. Mikulec, et al. Commissioning and results of the half-sector test installation with 160 MeV H−Beam from LINAC4 , CERN, Geneva Switzerland. . [38] C. Bracco, et al. Commissioning of the stripping foil units for the upgrade of the PSB H−injection system, CERN, Geneva, Switzerland. [39] F. Roncarolo, et al. ,Beam instrumentation for the CERN LINAC4 and PSB Half Sector Test, CERN, Geneva, Switzerland 48