Researchers at the University of Osaka achieved a record 132 MeV proton acceleration using ultrathin graphene targets and long-pulse lasers, overcoming challenges in detecting rare high-energy particles through AI analysis. The experiment, published in Progress of Theoretical and Experimental Physics, demonstrated a new method for laser-driven ion acceleration.
The Experiment and Methodology
Scientists at the University of Osaka used large-area suspended graphene (LSG) targets—4-, 8-, and 16-layer configurations—to accelerate protons to 132 MeV, nearly half the speed of light. The process involved a long-pulse laser (1.5 picoseconds) and a moderate intensity of 1×10¹⁹ W cm⁻², which generated a propagating electrostatic wave through the plasma. This wave acted like a “surfing” mechanism, allowing protons to gain energy over an extended period. The findings were detailed in the journal Progress of Theoretical and Experimental Physics.
The experiment utilized the LFEX laser at Osaka’s Institute of Laser Engineering, with the LSG targets irradiated at normal incidence. Accelerated ions were diagnosed using a CR-39 stack detector, a Thomson parabola spectrometer (TPS), and an electron and ion spectrometer (EISM). The CR-39 sheet in the TPS was analyzed with a convolutional neural network (CNN) after 2 hours of chemical etching, revealing pit-size distributions with two peaks corresponding to proton (Q/M= 1) and heavier-ion (Q/M ≤ 1/2) tracks. Blue and red squares in the analysis indicated proton and carbon 6+ ions, respectively.
Challenges in Detection
High-energy protons are rare, making their detection difficult. Researchers scanned millions of detector images to identify signals from individual ions, distinguishing 132 MeV protons from background noise. A convolutional neural network (CNN) achieved 99.2% precision in this task, streamlining the analysis of CR-39 stack detectors and Thomson parabola spectrometers. The challenge is not only to produce these rare high-energy protons, but also to reliably identify them,
said senior author Yasuhiro Kuramitsu, as noted in Nanowerk.
The study’s 2D PIC simulations for 16-layer LSG showed spatial profiles of the acceleration field (x− Ex) stacked vertically, with n× 0.15 TVm⁻¹ increments and a time step of 0.08 ps. Blue and orange shades indicated proton and carbon 6+ ion presence, respectively. The time axis began at t = 0, when the laser started penetrating the target. The CNN’s ability to isolate high-energy proton signals was critical, as We need to search millions of detector images for signals left by individual ions and distinguish the highest-energy protons from background noise,
the researchers stated.
Implications for Future Research
The study highlights the potential of graphene’s durability and thinness to withstand laser prepulses, enabling prolonged proton acceleration. Researchers emphasized that long-duration acceleration could push energies beyond those from shorter pulses. By using ultrathin graphene layers and a relatively long laser pulse, we are able to accelerate protons for an extended period and reach a record energy of 132 MeV,
said lead author Takumi Minami. The method, outlined in Nanowerk, could advance compact particle accelerators for medical or astrophysical applications.
The collaboration involved institutions in Japan, Taiwan, the UK, and France, highlighting the global interest in laser-driven ion acceleration. The team’s approach, combining graphene’s unique properties with AI-driven analysis, sets a precedent for future experiments aiming to harness high-energy proton beams more efficiently. The experiment was reported on October 02, 2026, by Nanowerk News, which described the process as analogous to a proton “surfing” on a moving electric field.