Improved Beam Dynamics Through Control of Magnetic Fringe Fields

点击翻译按钮获取中文摘要

Particle accelerators are essential tools for research in a wide variety of scientific fields. The Diamond synchrotron light source at Harwell, for example, played an important role in the UK's work to develop an effective COVID-19 vaccine, by providing researchers with a better understanding of the structural biology of the SARS-CoV-2 virus. Accelerators such as Diamond depend on magnets to control the beams of high-energy particles. Dipole magnets provide nearly uniform fields and are responsible for beam steering, and quadrupole magnets (consisting of four poles) create a more complex magnetic field that focuses the beam. Higher order multipoles serve other purposes such as chromatic correction and compensation for field errors in the lower order multipoles. In the main body of a magnet, the magnetic field is relatively simple, and can be assumed to be constant along the beam axis. At the entrance and exit of a magnet, however, the field is more complex as it tapers to zero. This intermediate region is known as the fringe field, and is characterised by its dependence on position along the beam axis, as well as a dependence on distance away from the axis. The effects of fringe fields in dipoles are relatively simple to describe, and can be implemented in the early stages of the design of a new accelerator. In quadrupoles and higher order multipoles, however, the fringe field effects can be more difficult to predict, and are often not described in detail until the design of the magnet itself is near completion. At that point in the accelerator design process, it is usually too late or too expensive to correct for adverse effects of the fringe fields.The goal of this project is to develop techniques for characterising fringe fields in multipoles, linking key features of the fringe field to the design of the magnet. Reverse engineering this process would (in principle) allow for the magnet design process to take account of the fringe fields from the beginning, avoiding adverse effects on the beam and potentially providing additional means to control the beam properties. An important part of the project will be to develop tools to allow rapid exploration of the effects of changes in specific details of the magnet design. At present, magnet design relies on the use of detailed modelling codes, which makes the design process computationally intensive and time-consuming. Alternative approaches that use recent advances in machine learning would allow more rapid characterisation of fringe fields. The success of this project would help to streamline the magnet design process, and allow more effective control of the beams in particle accelerators, improving accelerator performance and reducing design and construction costs.