【作者】 邬融；

【导师】 李永平；

【作者基本信息】 中国科学技术大学 ， 光学， 2008， 博士

【摘要】 本论文是围绕国家重大项目,结合高功率激光装置需求的基础上,对光束控制元件的设计、应用和测试等方面进行了系统性的论述,主要研究内容及获得的成果如下:一、大规模点阵的并行计算方法随着装置对光束控制元件DOE的尺寸需求越来越大,也就意味着设计上需要更多的采样点阵以保证满足采样定理同时扩大合适的输出计算窗口。然而单机在点阵成倍扩大时,计算能力不足的劣势凸现,内存以及时间都成为瓶颈。本论文利用MPI(Message Passing Interface)并行程序语言,开发了适用于DOE设计的并行计算方法。使用课题组的小型并行计算平台,计算时间上相比于单机大大缩短,可以满足今后大尺寸DOE器件的设计时间要求。二、平面靶的差值反馈优化算法当前针对光束整形的衍射光学元件的设计中,均需要采取模拟退火算法(SA)做进一步优化已得到更好的输出光斑。然而由于SA的随机性,所需的搜索时间很长,相比于光斑均匀性的提高量,性价比很低。差值反馈优化算法克服了SA的随机性,它通过一定的数学近似,将实际输出与理想输出的差值反馈到输入的位相结构上,进行相应调整,从而可以在很短的时间内收敛到一个更优的位相结构,极大的缩短了元件的设计周期。同时,根据该算法的数学推导过程,对于任何线形系统,只要选取相应的差值,均可使用该算法对已有的较优解来做进一步优化。三、多目标面迭代算法实现焦点附近3维空间束匀滑传统DOE设计的目标输出面都只是单独的2维平面,偏离目标设计面的光斑均匀性明显变差;另一方面间接驱动的目标输出面并非垂直于光轴而是与之有一定夹角。因此这两方面都对焦点附近3维空间提出了束匀滑的要求。本论文通过离散化3维空间,设置不同的接收反馈因子,展开递归搜索等方式,开发实现了多平面迭代算法。该算法的设计结果与传统设计结果的对比表明,在焦点附近约0.2mm的空间里,光束的均匀性和衍射效率都得到了不错的保持,从而提高了现有的空间设计指标,增强了器件的适应性。四、正面到斜面的可逆传输实现间接ICF束匀滑在间接ICF的小焦斑进洞的工程模型中,其腔口平面相对于光轴是一个斜面,工程要求在腔口上实现圆型光斑的注入,并且尽可能压低其余旁瓣使得不烧蚀腔口端面以顺利进洞,反映到DOE设计中,就是一个斜面设计以及消旁瓣的问题。本文首先研究了几种正面到斜面的传输方法,并根据Fourier空间内波矢的旋转变换,综合出了可逆的谱旋转方法,解决了正面到斜面的可逆传输问题。然后利用该方法,开发了消旁瓣的设计方法。大点阵的设计结果表明,可以注入大部分激光能量到靶腔中,并且所需压制旁瓣区域的功率密度很低,设计上满足了工程需求。五、口径缩放以及中心漂移的设计算法提高DOE的口径宽容性无论是平面靶还是间接靶的DOE设计,都不得不考虑口径影响,这源于实际打靶中,照射到DOE器件上的光束口径有一定量的缩放,且还会偏离中心少许;而理论设计上均假定入射光束口径不变并且固定在中心,导致实测结果与设计偏差较大。本文对此作了相应研究,并将口径的变化直接纳入到DOE的迭代设计中,利用随机入射口径的方式来使得DOE元件对口径在给定范围内变化的入射光均具有较好的输出光斑。设计结果表明,对于任何在DOE有效位相单元内的入射光都能得到一个较好的平均结果,包括光斑德均匀性指标和所需压制旁瓣区域的功率密度。从而大幅提高了DOE的口径宽容性。六、320mm口径DOE器件的打靶测试,改进和应用前景由分步迭代算法设计制作的320mm有效口径的DOE器件,进行了实验打靶测试,实测结果与设计指标偏离较大:中心亮点较大,旁瓣较高,衍射效率太低。经计算分析,设计口径与实验口径相差较大以及DOE单元数太少很可能是出现问题的原因。因此,今后设计上应该要加入适于入射光口径变化的位相结构,同时扩大计算点阵以抑制旁瓣并尽量降低其余衍射级次的能量,就可以避免上述问题,使得DOE元件顺利应用到ICF光束控制系统中。更多还原

【Abstract】 The research devoted to this thesis focuses on the design and applications of large aperture diffractive optical elements (DOE) that will be used to realize uniform illumination on focal plane for inertial confinement fusion (ICF) system. The works are under the support of national project. The primary research and achievements are as follow:1. With the requirement of more large-size of beam control components DOE, we need much more sampling grid to satisfy sampling theorem and ensure appropriate calculation output window. However, with the expand of sampling matrix, the lack of calculation of PC is emerging. Using the Message Passing Interface (MPI) language, we have developed parallel computing methods for the design of DOE. The time of calculation is greatly shortened comparing to a single PC, which can meet the requirement of time-cost for the design of large-size DOE.2. Usually, we use simulate annealing (SA) algorithm to optimize the phase structure of DOE for more uniform output beam. However, due to the randomness of SA, the searching time is long, and the cost-effective is low compared to the improvement of uniformity. The difference feedback optimization (DFO) has overcome the randomness of SA. Through a mathematical approximation, It adjusts relative phase structure by the difference of real output and target output, which can quickly convergent to a better phase structure. Thus, this new algorithm has significantly reduced the time of design cycle.3. Traditionally, the target of beam shaping is a single plane, such as focal plane, and the uniformity of beam is degenerated when deviating from target plane. On the other hand, the target plane of indirect-driver ICF is not perpendicular to the optical axis but with a certain angle. So both of them propose 3D beam smoothing requirements. Through the dispersion of 3D space, setting of different feedback factor and launching recursive search, we have developed and realized a multi-plane iterative algorithm for 3D beam shaping. The results show that the uniformity and diffraction efficiency have been well maintained in the vicinity of 0.2mm around focal plane, which has enhanced the spatial specifications and adaptability of DOE.4. To solve the reversible propagation between vertical plane and incline plane, several propagation methods have been researched. According to the rotation of wave vector in Fourier space, we have found a reversible transform of spectrum rotation solving the hard problem. Then, using the reversible transform, we have developed the design method for eliminating sidelobe of output beam. Large-size design results show that it can inject most of laser energy to the target cavity with low sidelobe, meeting the needs of project.5. Whether the driver of ICF is direct or indirect, we have to consider the impact of caliber in DOE design because the beam diameter usually has a certain amount of zoom and a little deviation from center in actual target practice. Tradition design treat the input beam to be fixed, and it will inevitably lead to the measured results and design of large deviations. We take the aperture variation into the process of DOE design, and using the random incident way to ensure average output beam under a given variation of aperture. It has greatly increased the diameter tolerance of DOE.6. We have tested a DOE of 220mm diameter in the target experiment. The test results have some large difference to the design specifications, with higher center energy and sidelobe and lower diffraction efficiency. According to the computing analysis, the difference between design and experimental caliber and the lack of DOE unit will be the cause of these problems. Therefore, we should bring the diameter variation of incident beam into our design and enlarge sampling grid to make other diffraction orders as far as possible so that the loss of energy is as little as possible. Only in this way, DOE can be successfully applied to the beam control components of ICF system.更多还原