Machine Learning for Stable Synchronization of Accelerator Beams

2019-12-01

Scientists in the United States have used machine learning to reduce unnecessary fluctuations in photon beams from synchrotron light sources. This technology achieves this by stabilizing the electron beam of the synchrotron accelerator, providing important insights for the development of next-generation facilities.


This work was carried out by Simon Leemann and his colleagues at the Lawrence Berkeley National Laboratory (LBNL) in California, and it enables emerging analytical techniques that require high beam stability, such as X-ray Photon Correlation Spectroscopy (XPCS), to be implemented on synchrotrons.


Synchrotron light sources are extremely useful scientific instruments as they emit bright, high-quality electromagnetic radiation beams. Light is generated by accelerating electrons in a storage ring using powerful magnets, taking advantage of the fact that accelerated electrons emit electromagnetic radiation. In modern synchrotrons, the use of wigglers and undulators enhances the acceleration.Now, Leemann's team has demonstrated that higher stability can be achieved by using machine learning algorithms to adjust existing beam control instruments. They developed a neural network designed to prevent width fluctuations in the electron beam of the Advanced Light Source (ALS) at LBNL. The algorithm was trained using two datasets: one consisting of the magnetic "excitation" distribution used to control the shape of the electron beam, and the other consisting of the width of the photon beam obtained from the ALS.


Using this information, the neural network constructed a table that illustrates how the magnetic excitation affects the final width of the photon beam. By continuously updating this table and referencing it three times per second, the synchrotron can automatically adjust its electron beam. This results in vertical electron beam fluctuations of only 200 nm, which is only 0.4% of the vertical range of the electron beam, suitable for fourth-generation synchrotrons.


Future synchrotrons will be able to generate beams three times brighter than current ones. These facilities should enable the study of physical and chemical processes at the nanometer and micrometer length scales. One potential application is obtaining 3D maps of complex chemical reactions, which could lead to a deeper understanding of electrochemical systems, including batteries and fuel cells.A potential limitation of the performance of fourth-generation devices is the relatively large instability in the vertical profile of the synchrotron electron beam. This instability leads to fluctuations in the position and size of the beam, making it challenging to implement techniques such as XPCS.


Current stability techniques combine the use of predetermined models and tedious manual calibration to achieve a stability of about 10% in the electron beam width. This is sufficient for third-generation devices, but fourth-generation synchrotrons will require stricter control.

RECENT AI NEWS

RECENT AI TOOLS