Researchers at the University of Osaka, working with collaborators across institutions in Japan, Taiwan, the UK, and France, have used an ultrathin, large-area suspended graphene target to accelerate protons to 132 MeV, nearly half the speed of light, in a laser-driven ion acceleration experiment.
Laser-driven ion acceleration is being explored as a more compact alternative to conventional radio-frequency particle accelerators, with potential applications ranging from medicine to laboratory astrophysics. Pushing protons to higher energies generally means using thinner targets, but ultrathin targets are easily destroyed by the weak "prepulse" that precedes a laser's main high-intensity pulse. The team addressed this using large-area suspended graphene (LSG) targets just a few nanometers thick (4, 8, and 16 atomic layers), exploiting graphene's unusual combination of extreme thinness and durability to withstand the prepulse and remain intact until the main pulse arrived.
Using a relatively long-pulse (1.5 picosecond), moderate-intensity laser at Osaka's Institute of Laser Engineering, rather than the ultra-short, ultra-intense pulses more commonly used for this kind of experiment, the researchers found protons were accelerated for several picoseconds, longer than in typical short-pulse setups, by a propagating electrostatic wave moving through the laser-generated plasma. Computer simulations describe the effect as similar to "surfing acceleration," with protons continuously gaining energy as they ride the moving wave rather than receiving a single short kick.
Because high-energy protons are rare and their detector signals are faint, identifying them required scanning millions of microscope images of the detector surface for signs of individual ion impacts. The team trained a convolutional neural network to pick out these signals, reaching 99.2% precision in distinguishing genuine high-energy proton signals from background noise, an approach they say will be essential for pushing the technique toward even higher energies.
