Contaminant Particle Formation in Radio Frequency Silane Plasmas

处理等离子体中超细粒子的形成多年来一直是一个令人担忧的问题。在微电子制造中，污染物颗粒会导致相当大的产品成品率损失，或者严重影响用于生产光伏电池或光学器件涂层的薄膜的质量。虽然对微米级颗粒在等离子体中的输运和充电的了解已经取得了很大的进展，但对颗粒的成核和纳米级颗粒的行为知之甚少。在等离子体增强化学气相沉积系统中，可以观察到小颗粒的形核和长大。硅烷等离子体中的粒子形成对于非晶硅的沉积是非常重要的，已经得到了广泛的研究，但粒子形成机制的真正性质还没有确定。不同的研究人员提出了在硅烷等离子体中形成粒子的不同甚至相互矛盾的场景。例如，硅烷等离子体中粒子成核的重要前体仍然存在争议。等离子体条件对成核过程的影响尚不清楚。此外，由于纳米颗粒的原位检测存在严重困难，因此关于颗粒成核阶段的实验结果很少。由于对颗粒成核的理解是一个非常复杂的问题，因此在拟议的项目中将采取实验和理论相结合的方法。提出的研究有三个主要目标：1.建立了硅烷等离子体中粒子成核模型，该模型考虑了所涉及的化学因素。2.成核模型将与硅烷等离子体化学气相沉积系统的自洽等离子体模型相耦合，以便能够预测粒子成核的发生。3.所建立的模型将通过对稀硅烷中电容耦合射频放电的大量实验研究来验证。所建立的成核模型将考虑氢化硅团簇的形成，包括阴离子和中性两种机制。它将与能够预测颗粒生长、凝结和传输的气溶胶模型相耦合。该模型将与电容式射频放电的动力学放电模型相结合。动力学排放模型将基于已有的代码，将用于计算成核过程中涉及的重要物种的空间分布。由于成核速率敏感地依赖于前驱体浓度，因此可以预测成核阈值和成核区的空间位置。与其他一些方法不同的是，非麦克斯韦电子分布函数的影响对于准确确定各种化学反应速率是至关重要的，它将通过求解Boltzmann方程来计算实际分布函数。实验研究将集中在小颗粒的检测以验证成核模型，以及使用朗缪尔探针测量电子分布函数以验证动力学放电模型。对于粒子探测，将使用高功率的YAG激光进行蒸发式粒子爆炸和电子分离。此外，将测试一种新的诊断方法，它基于耦合到等离子体中粒子群的声波的传播。这项诊断应该能够揭示关于颗粒大小和质量的现场信息，即使是亚纳米颗粒，这是极难用激光方法检测的。

9731568KortshagenThe formation of ultrafine particles in processing plasmas has been a concern for some years. Contaminant particles can cause considerable product yield loss in microelectronics fabrication or they can severely impair the quality of thin films used for the production of photovoltaic cells or the coating of optical devices.While considerable progress has been made in the understanding of the transport and charging of micrometer-sized particles in plasmas, little is known about the nucleation of particles and the behavior of nanometer-sized particles. The nucleation and growth of small particles is observed in particular in plasma enhanced chemical vapor deposition systems. The particle formation in silane plasmas, which are of great importance for the deposition of amorphous silicon, has been extensively studied, however, the true nature of the particle formation mechanism has not been determined. Different and even contradicting scenarios for particle formation in silane plasmas have been proposed by various researchers. For instance, the important precursor for the particle nucleation in silane plasmas is still controversially discussed. The influence of the plasma conditions on the nucleation process are not understood. Also, experimental results on the particle nucleation phase are scarce, due to the serious difficulties in in-situ detection of nanometer-sized particles.Since the understanding of particle nucleation is a very complex problem, a combined experimental and theoretical approach will be pursued in the proposed project. The proposed research has three main objectives: 1. A nucleation model for particles in silane plasmas will be developed which takes into account the chemical aspects involved. 2. The nucleation model will be coupled to a self-consistent plasma model for a silane plasma chemical vapor deposition system in order to enable predictions about the occurrence of particle nucleation. 3. The model to be developed will be verified by a number of experimental studies on a capacitively coupled RF discharge in diluted silane.The nucleation model to be developed will consider the formation of hydrogenated silicon clusters by both anionic and neutral mechanisms. It will be coupled to an aerosol model capable of predicting particle growth, coagulation and transport. This model will be combined with a kinetic discharge model for a capacitive RF discharge. The kinetic discharge model, which will be based on already existing codes, will be used to calculate the spatial profile of the important species involved in the nucleation process. Since the nucleation rate depends sensitively on the precursor concentrations it is expected that predictions about nucleation thresholds and the spatial location of nucleation regions can be obtained. Unlike in some other approaches, the effects of a non- Maxwellian electron distribution function, which are crucial for an accurate determination of various chemical reaction rates, will be taken into account by calculating the actual distribution function by solution of the Boltzmann equation.The experimental studies will focus on the detection of small particles in order to verify the nucleation model and on the measurement of the electron distribution function using Langmuir probes in order to test the kinetic discharge model. For the particle detection a high power YAG laser will be used for evaporative particle explosion and electron detachment. Furthermore, a new diagnostics will be tested which is based on the propagation of acoustic waves coupled to the particle population in the plasma. This diagnostic should be able to reveal in-situ information about the particle size and mass even for subnanometer particles, which are extremely difficult to detect with laser methods.