AI improves controlling plasma accelerators for research and industrial applications

AI improves controlling plasma accelerators for research and industrial applications

AI improves controlling plasma accelerators for research and industrial applications

AI to monitor next-generation accelerators increases the capacity for applications in science, medicine, and business. Researchers used AI to power beams for the next generation of smaller cheaper accelerators for research, medical and industrial applications. Experiments performed by researchers at Imperial College London using the Central Laser Facility (CLF) of the Science and Technology Facilities Council have shown that the algorithm has been able to change the complicated parameters involved in operating the next generation of plasma-based particle accelerators.

An international team of accelerator experts with the participation of researchers from DESY has successfully demonstrated that an algorithm was able to tune the complex parameters involved in controlling the next generation of plasma accelerators.

The algorithm was able to optimize the accelerator even faster than a human user, and could also outperform tests on comparable laser devices. These accelerators concentrate the energy of the world’s most efficient lasers on a spot the size of a skin cell, generating electrons and x-rays for a fraction of the size of traditional accelerators. Electrons and x-rays can be used for scientific studies, such as measuring the atomic structure of materials; for manufacturing uses such as consumer electronics and vulcanized rubber for car tires; and for medical applications such as cancer therapies and medical imaging.

Several installations using these new accelerators are in different phases of preparation and development around the world, including the Extreme Photonics Applications Center (EPAC) of CLF in the UK, and a new discovery could allow them to function better in the future. The reports are published in Nature Communications today.

Plasma-based acceleration technology hopes to offer a new generation of accelerators that are more efficient, lightweight, and scalable than the existing ones. Accelerated electrons or x-rays emitted by them may be used for scientific studies, such as testing the atomic structure of materials; for industrial applications, such as consumer electronics; and may also be used in medical applications, such as cancer therapies and medical imaging.

First author Dr. Rob Shalloo, who finished his work at Imperial and is now at the accelerator center DESY, said The techniques we have developed will be instrumental in making the most of the current generation of advanced plasma accelerator installations under development in the United Kingdom and around the world.

“Plasma accelerator technology offers uniquely fast pulses of electrons and x-rays that are now being used in many fields of scientific research. With our inventions, we aim to improve the usability of these portable accelerators, enabling scientists in other fields and those wanting to use these devices for applications to benefit from the technology without becoming a specialist in plasma accelerators.”

The team was working with the laser wakefield accelerators. These integrate the world’s most efficient lasers with a plasma source (ionized gas) to produce intense beams of electrons and x-rays. Traditional accelerators require hundreds of meters to kilometers to accelerate electrons, but wakefield accelerators can handle the same acceleration within the range of millimeters, dramatically minimizing the size and expense of the devices.

However since wakefield accelerators work under intense conditions produced when lasers are mixed with plasma, they can be difficult to monitor and refine to achieve the best efficiency. In the wakefield acceleration, the ultrashort laser pulse is sent to the plasma, producing a ripple that is used to accelerate electrons. Both the laser and the plasma have many parameters that can be tweaked to monitor reactions, such as the form and strength of the laser beam, or the density and length of the plasma.

Although a human operator can change these parameters, it is impossible to know how to optimize so many parameters at once. Instead, the team turned to artificial intelligence, developing a machine-learning algorithm to maximize the efficiency of the accelerator. The algorithm set up six parameters to monitor the laser and plasma, activated the laser, evaluated the results, and reset the parameters, repeating this loop several times in succession until the optimum configuration of the parameter was reached.

Lead researcher Dr. Matthew Streeter, who finished his study at Imperial and is now at Queen’s University Belfast, said: “Our work has culminated in an autonomous plasma accelerator, the first of its kind. As well as allowing us to refine the accelerator effectively, it also simplifies their operation and allows us to dedicate more of our attention to studying the fundamental physics behind these extremes.

The team has demonstrated their technique using the Gemini laser method at the CLF and has also started to use it in further studies to test the atomic structure of materials under extreme conditions and to study antimatter and quantum physics. The data obtained during the optimization process also offered new insight into the nature of laser-plasma interaction within the accelerator, theoretically influencing future designs to further enhance accelerator performance.

Similar studies performed at the LUX accelerator, a collaborative project between DESY and Hamburg University, support the theory that the use of AI for steering accelerators has been raised to a new level these days (S. Jalas et al., submitted). “With AI and machine learning deployed on our current and next-generation accelerators at DESY and elsewhere, we expect unprecedented levels of performance to be truly exciting,” concludes Wim Leemans, Director of the Accelerator Division at DESY.