Automated multi-dimensional mapping of dynamic laser-liquid interactions

高强度激光与物质的相互作用产生具有非常高的温度和密度的极端环境，使得材料原子内的电子不再保持与原子核的结合，并且材料成为等离子体。这些相互作用可以为研究天体物理现象创造条件，包括超新星冲击和太阳耀斑，以及支持非常高的电场，可用于加速带电粒子的距离比射频加速器技术的极限短100到1000倍。这些紧凑型加速器已被证明可以产生具有非常理想的特性的离子束，用于材料测试，放射生物学和惯性聚变能的关键应用。到目前为止，由于难以在数值和计算模型中再现它们的复杂行为，以及由于用于产生等离子体和驱动粒子加速的高能脉冲激光的低重复率（通常<<0.002 Hz -每10分钟发射一次）造成的可用数据有限，因此对这些相互作用的充分探索和利用受到阻碍。这在易碎超薄不透明目标的研究中尤其如此，其中来自激光的能量的吸收导致目标加热和膨胀，导致目标随着密度福尔斯下降而变得透明。当这种情况发生时，激光可以传播通过靶，并且激光能量到等离子体的转移不再局限于靶表面。这种相互作用非常有趣，因为在这里记录了最高能量的激光加速质子。新一代的多赫兹高能激光技术使数据采集速率有了几个数量级的提高。为了开发这些新的激光器，还需要测试目标技术，该技术可以提供在几Hz重复频率下具有高位置稳定性的新超薄箔。此外，尽管在数据采集率的巨大增加的相互作用动力学的依赖性的大量变量（例如激光能量，激光的空间和时间能量分布，目标密度分布）意味着，“网格扫描”每个参数不是一个有效的方法来映射它们的相互依赖性。通过结合机器学习工具，激光器和目标实现的高数据速率可用于智能地对参数空间进行采样，以对这些新型加速器的相互作用进行建模并量化其稳定性。拟议中的合作将通过将美国SLAC国家加速器实验室（SLAC）开发的液体片靶与女王大学贝尔法斯特（QUB）研究人员开创的新的计算机引导激光等离子体实验方法相结合来应对这一挑战。这个新的实验平台的开发将使人们能够更深入地了解激光和等离子体之间的关键能量传递途径及其对实验变量的依赖性。该研究将直接影响等离子体建模、先进加速器研究、等离子体天体物理学、惯性约束聚变、材料测试和FLASH放射生物学。研究成果将通过以下研究主题纳入EPSRC 2022-2025战略优先事项：物理和数学科学强国，工程和人工智能前沿：通过利用人工智能进行实验科学，为工程，健康和政府提供人工智能和数据科学;通过惯性约束聚变获得能源;等离子体和激光，通过开发关键技术，促进对新辐射源的更深入理解和更广泛的利用;以及研究基础设施，通过提高高强度激光设施的能力。

High-intensity laser interactions with matter produce extreme environments with very high temperatures and densities such that the electrons within the atoms of the material no longer remain bound to the atomic nuclei and the material becomes a plasma. These interactions can create conditions for studying astrophysical phenomena, including supernova shocks and solar flares, as well as supporting very high electric fields that can be used to accelerate charged particles over distances 100s to 1000s times shorter than the limits of radio-frequency accelerator technology. These compact accelerators have been shown to generate ion beams with highly desirable properties for key applications in materials testing, radiobiology, and inertial fusion energy. So far, full exploration and exploitation of these interactions has been hampered by the difficulty in reproducing their complex behaviour in numerical and computational models and by the limited data available which is caused by the low repetition rate of the high-energy pulsed laser (typically <<0.002 Hz - a shot every 10 mins) used to create the plasma and drive particle acceleration. This is particularly the case in the study of fragile ultra-thin opaque targets where the absorption of energy from the laser causes the target to heat and expand leading to the target becoming transparent as the density falls. When this occurs the laser can propagate through the target and the transfer of laser energy to the plasma is no-longer localised at the target surface. This interaction is of significant interest as it is here that the highest energy laser-accelerated protons have been recorded. A new generation of multi-Hz high-energy laser-technology is facilitating orders of magnitude increase in data acquisition rate. In order to exploit these new lasers, it is also necessary to test target technology that can provide fresh ultra-thin foils with high positional stability at multi-Hz repetition rate. In addition, despite the enormous increase in data acquisition-rate the dependence of the interaction dynamics on a large number of variables (e.g. laser energy, laser spatial and temporal energy distribution, target density profile) means that `grid-scanning' each parameter is not an efficient method to map their interdependence. By incorporating machine learning tools the high data rate enabled by the lasers and target can be used to intelligently sample the parameter space to model the interaction and quantify the stability of these novel accelerators. The proposed collaboration will address this challenge by coupling a liquid sheet target, developed at the US SLAC National Accelerator Laboratory (SLAC), with a new computer-guided approach to laser-plasma experiments, pioneered by researchers at Queen's University Belfast (QUB). The development of this novel experimental platform will enable deeper understanding of the key energy transfer pathways between laser and plasma and their dependence on experimental variables. The research will directly impact on plasma modelling, advanced accelerator research, plasma astrophysics, inertial confinement fusion, materials testing and FLASH radiobiology. The research outputs will feed into EPSRC 2022-2025 strategic priorities on the physical and mathematical sciences powerhouse, frontiers in engineering and artificial intelligence up-skilling through the research themes: AI and Data Science for Engineering, Health and Government by exploiting AI for experimental science; Energy through inertial confinement fusion; Plasma and lasers by developing crucial technology to facilitate deeper understanding and broader exploitation of novel radiation sources; and research infrastructure by enhancing the capabilities of high-intensity laser facilities.