【作者】 杨朝霞；

【导师】 方健文；

【作者基本信息】 浙江师范大学 ， 光学， 2007， 硕士

【摘要】 调制光热反射(MPR)技术具有无损检测和高灵敏度的特点,已被广泛用于物理、材料工程等领域,并越来越受到人们的重视。作为检测手段,实验数据的拟合以及样品性质参数的定量获取一直没能得到很好地解决。新兴起的粒子群优化算法是一种智能进化计算技术,该算法的优点有望在测量技术的多参数拟合中得到应用,但为了保证其智能性和最优性,还应根据材料物理参数拟合的特点对算法加以改进。本文主要在总结分析粒子群算法的基础上,在MPR技术中引入了粒子群算法并相应开展了如下一些工作:1.对粒子群优化算法进行了研究。在总结分析粒子群算法的基础上,针对样品多参数拟合时的问题,对粒子群算法进行了改进,提出了以下改进策略:①当待拟合参数之间相关性高时,根据最优粒子的信息适时进行搜索范围的动态调整,缩小了搜索范围,减少了陷入局部极值的概率。②当待拟合参数取值范围广时,对最优粒子采用新的变异策略,增强了搜索过程的智能性,加快了搜索速度。2.对利用MPR径向扫描技术表征薄膜-衬底材料的热物性进行了研究。针对低热扩散率薄膜样品较难进行多参数拟合的难题,在讨论影响拟合效果的参数灵敏度和相关性问题的基础上,利用改进的粒子群算法对同时拟合薄膜热扩散率、衬底热扩散率和界面热阻等样品参数进行了数值模拟。结果表明,改进的粒子群算法比其它方法能更好地对高相关性的参数进行拟合。3.对利用MPR技术表征半导体材料的物理参数进行了理论和实验研究。根据理论模型并结合改进的粒子群算法,利用实验测得的MPR相位信号对硅材料的热扩散率、载流子寿命和面复合速率进行了拟合,获得了满意的拟合结果,并在粒子群优化算法中较好地解决了因待拟合参数取值范围广而引起的拟合运算问题。更多还原

【Abstract】 Modulated photothermal reflectance (MPR) technique, as a highly sensitive and nondestructive measurement method, has been widely applied to the fields of science and technology. The MPR technique has attracted much more attention. As a detection method, however, fitting experimental data and getting sample’s property parameters quantitatively haven’t been solved satisfactorily. The particle swarm optimization algorithm, which is a new kind of intelligent and evolutionary computation, is hopefully applied to multiparameter fitting in measurement technologies. In order to ensure the intelligence and the optimality, the algorithm should be improved according to characteristics of fitting material’s physical parameters. The particle swarm optimization algorithm is introduced in MPR technique based on the investigation of particle swarm optimization. The main contributions of this thesis are listed as follows:1. Particle swarm optimization algorithm is studied. In order to make up the deficits in fitting sample’s parameters, it is proposed an improved particle swarm optimization in the thesis on the basis of summarization and analysis of particle swarm optimization. Modified strategies are presented as following: Firstly, when there is strong correlation among the parameters which need to be fitted, a dynamic adjustment of searching regions based on information about optimal particle can reduce the searching region as well as the probalility of local convergence. Secondly, on the condition that the parameters’ searching region is very large, a new strategy of mutation is proposed, which enhances the searching intelligence and quickens the searching velocity.2. The theoretical study for characterizing thermal properties of thin films and substrates by MPR technique is presented. It is difficult to fit thermal parameters of film-substrate sample in which the film’s thermal diffusivity is lower than the substrate’s. On the basis of discussion about parameters’ sensitivity and correlation, three thermal parameters, i.e. the film’s thermal diffusivity, the substrate’s thermal diffusivity and the thermal resistance on the film-substrate boundary, are simultaneously fitted with the improved particle swarm optimization. The simulative results showed that the improved particle swarm optimization algorithm fit the strong correlation parameters better than other methods.3. The theoretical and experimental studies of measuring physical parameters of semiconductor by MPR technique are performed. According to the theoretical model and the improved particle swarm optimization algorithm, the silicon parameters including thermal diffusivity, charge carrier life time and surface recombination velocity are fitted with measurable phase signals. Using the improved particle swarm optimization algorithm, it not only obtains satisfying fitting results but also solves the fitting difficulty that is brought from large parameter values’ range.更多还原