Particle Detectors & Calorimetry
Introduction
Particle detector is a device that is used to quantify the momenta and identify the particle that passes through the detector after being produced in a collision or a decay, which also referred to as an event. The event might be a collision deliberately engineered to occur within the detector leading too many particles which in turn decay into many more, or it could be a decay the occurs naturally. To reconstruct the event and identify every particle produced by the collision, its necessary to know the mass and the momentum of the particles. The mass can be found by measuring the momentum and either the energy or velocity. The modern large scale particle detectors like LHCb or ATLAS, encompass many forms of tracking chambers and calorimeters that surround the interaction point also know as hermetic detectors.
Momentum Determination
Tracking Detectors
Tracking detector shows the track or the path taken by a particle. Momentum measurements can be made by applying magnetic field perpendicular to the direction of the travel in a tracking detector, this causes the particle to curve into a circular orbit with a radius proportional to the momentum of the particle. The radius is also proportional to the mass-charge ratio m/q, so if two particles have the same speed, and one is half as heavy as the other but has twice the charge, they will both orbit with the same radius. Since magnetic field exerts a force perpendicular to the direction of the particle’s direction, it can do no work on the particle, hence the particle does not lose energy as a result of this process. Thus determining the momentum, the energy can be measured by a calorimeter and the mass of the particle, then we can deduce the identity of the particle. Also, the direction of the particle curvature in the magnetic field reveals its charge, since negative charges curve in the opposite directions.
Cloud Chamber
Cloud chamber is the earliest tracking detector. In oversaturated vapour, primary ionization clusters left behind a charged particle will become centres of condensation. Droplets will follow the track of a particle, their number per unit of length being proportional to the density of ionization. Such conditions can be achieved by the fast expansion of the chamber volume, which in its turn can be triggered by an external particle detector.
Emulsions
Emulsion consist of crystals of AgBr and AgCl suspended in a body of gel. An ionizing particle passing through the emulsion breaks up these molecules. After developing the film, the released metallic Silver grains are locked in the main body of the emulsion film while the remaining part will be washed away. With the of a microscope, these grains usually of about half micron diameter, it can be observed including their exceptional high spatial resolution as a black dots. Emulsion films are still being used due to their many advantages
Bubble Chamber
A bubble chamber is a container with liquefied gas like hydrogen, the gas is close to the boiling point. When a charged particles going through the chamber it will leave behind clusters of ionization. If the pressure is dropped for a short time, the boiling temperature drops and the liquid becomes overheated. Now the boiling is about to begin, bubbles are formed on ionization clusters and start growing. In about 1 ms, when bubbles are about 0.1−1mm in size, pictures from different angles are taken and pressure is restored. The increased pressure brings the liquid back below the point of boiling and all bubbles collapse. Multiple pictures take from different directions allow for unambiguous 3D reconstruction.
Wire Chamber
The proportional chamber is a tube usually with a diameter of 1 cm with a wire typically 20 − 100 µm in diameter stretched along its axis. A positive high voltage potential is applied to the wire. When a charged particle goes through the tube it leaves about 100 ionization electron per cm. These electrons will drift toward the positively charged wire and result in a weak current in the external electric circuit, however, the signal is so weak for detection. If the applied voltage is large, the electric field near, but at some distance away from the wire may become strong enough to accelerate electrons drifting towards the wire to energies sufficient for producing secondary ionization, which will result in an avalanche-like process of electron number multiplication. The total charge collected on the wire (Q) can be easily G = 10⁵ times the initial ionization left behind by a charged particle. The factor G is commonly called gas gain. The total charge, which is now large and can be easily detected/measured. Also, the charge is proportional to the initial ionization, hence the name proportional.
Wire chamber includes:
- Multi-wire proportional chamber
- Drift chambers
- Time Projection chamber
Semiconductor Detector
Semiconductor detector operation principle is similar to gaseous detectors with electron-hole pairs being produced as a result of ionization losses. Large density of electrons and small average energy needed for ionization allow one to obtain a rather large and easily detectable signal in the thin substrate without additional amplification mechanism like the gas gain is needed. The main advantage of this type of detectors is its speed, compactness, and precision.
Energy Determination
Calorimeters
The name calorimetry in particle physics comes originally from thermodynamics which means measure the heat released by some reaction, A calorie is a unit for measuring energy and was introduced by Nicolas Clement. Moreover, it is defined as the energy required to increase the temperature of one gram of water, one degree Celsius at one atmospheric pressure. Calorimeters used for particle detection used different techniques for signal generation and can be classified: Calorimeters are either homogeneous or have a sampling structure. In the latter case, the functions of particle absorption and signal generation are exercised by different materials, called passive and active media. Homogeneous calorimeter, the entire calorimeter is active. This category includes detectors consisting of crystals which either produce scintillation or Cherenkov light. While calorimeters used in experiments at the highest energy particle collides are sampling type, mainly because of cost and compactness. The active material in sampling calorimeters generates a signal in term of scintillation or Cherenkov light or electric charge, by direct ionization or the production of electron-hole pairs. Although crystals are still popular at e+ e- rings, especially those operating at the J/ψ and ϵ resonances.
Scintillation
When a charged particle traverse through matter, they lose energy through the electromagnetic interaction with the Coulomb fields of the electrons. This energy may be used to ionize the atoms or molecules into an excited state, that is basically what a scintillation is. This excitation is unstable and the excited atom of the molecule will quickly return to the ground state. In this process, the excitation energy is released in the form of photons. The duration of this phenomenon is determined by the excitation energy, the quantum numbers of the states involved, and the number of available return paths. Some times the energy difference is such that the emitted photons are in the visible spectrum, this process is called fluorescence.
Cherenkov Radiation
When a charged particle travels faster than the speed of light in some medium v > c, it loses energy by emitting Cherenkov radiation. The emitted radiation has a characteristic angle called Cherenkov angle θ_c = arccos 1/β; β = v/c The spectrum of Cherenkov radiation exhibits a characteristic 1/λ² dependence and therefore the visible part of the spectrum is observed as blue light. This blue light can be observed in highly radioactive environments such as moderating liquids in the nuclear reactor. Because the Cherenkov phenomena are sensitive to the velocity of the particle it can be used to determine the mass of the particles of which the momentum has been determined by means of deflection in the magnetic field. One aspect of Cherenkov radiation is its instantaneous character (no delaying factors).
Ionization
When a charged particle traverse through matter, the particle ionize the atoms of which this matter consists. One or several electrons are released from their Coulomb field in this process, leaving behind an ionized atom. Collection of these liberated electrons is applied as the signal producing technique in a wide variety of particle detectors. The electrons produced along the trajectory of the ionized particle may be amplified during this process. Ionization chambers may use liquid-based media liked liquefied noble gases like Argon, Xenon, Krypton. One Important feature of noble liquids is very radiation hard, unlike scintillating crystals. The ionization chamber may also be based on gaseous media. In these devices, the electrons produced in the ionization process undergo considerable multiplication before being collected at the anode. Finally, the detector of charges can also be based on solid-state devices utilizing Silicon, gallium arsenide or germanium. An additional advantage of semiconductors crystals is their fast response time.
Cryogenic Phenomena
Cryogenic phenomena include:
- Elementary excitation, some excitations require a very small amount of energy. For example, cooper pairs in the µeV — meV and may be broken by photon absorption.
- Some material exhibit specific behaviour like change in magnetization, latent heat release, that may provide signals for detectors. Also under low temperatures, the thermal noise of the detectors and the associated electronics become very small.
- Specific heat of dielectric, The specific heat for dielectric crystals and for a super-conductors decrease to a very small value at these low temperatures.
Deep Learning in High Energy Physics
It is known that deep neural networks can easily model high dimensional distributions, which enables improved statical analysis and efficient and fast particle simulations. The deep learning approach is to train a neural network to learn the simulations pool from the traditional pool of the simulated events. This approach creates a generative model G which approximate the distributions of the sample generated by the simulator by mapping the space of the input number to space of the data. Another promising approach is by using GAN (Generative adversarial neural network); basically, the training happens by two neural network training against each other the generator generates data, and the discriminator tries to classify whether the data are fake or real, by doing so the neural network will generate events that are similar to the real data. GAN approach to simulation has been applied before to the electromagnetic showers in a multi-layer calorimeter, which are very expensive computationally, They reported larger speedups in term of computations. Another promising approach is the use of hyperbolic neural networks; it’s common in machine learning to represent data as being embedded in Euclidean space ℝⁿ The convenience of Euclidean space made it the norm for machine learning as this space has a vectorial structure, closed-form formulas of distance and inner product. As it been shown that many types of complex data from a multitude of fields exhibit a highly non-Euclidean latent anatomy. In such cases, the Euclidean space does not provide the most powerful and meaningful geometrical representations. For example, the decay of particles can be represented in directed graphs or non directed graphs; such a complex graph will not be fully representable in Euclidean space. It has been shown that arbitrary tree structure can not be embedded with arbitrary low distortions (Preserving their metric) in the Euclidean space with the unbounded number of dimensions. However, this task becomes surprisingly easy within hyperbolic space with only two dimensions where the exponential growth of distance matches the exponential growth of nodes with the tree depth. Utilizing these methods has not been done before in particle physics due to the fact that the hyperbolic neural network is currently being heavily researched. Hopefully, we will utilize these techniques to increase the accuracy and the speed of the simulations.
