硅化学机械抛光（CMP）技术广泛应用于新硅晶圆制造、芯片前段制程制造和晶圆再生制造中。随着技术节点的快速发展，硅CMP之后的清洗技术面临巨大挑战，愈发成为硅CMP制程发展的“卡脖子”难题。本文针对14 nm技术节点需求的硅CMP后清洗关键技术做了系统研究。 首先对硅CMP中污染物种类和产生机制进行了深入分析。经过对污染物分布、形貌和成分检测，阐明了颗粒、有机残留、水痕和细菌等主要缺陷的形态和主要特征，深入探讨了各类污染物缺陷的主要来源，为深入研究晶圆清洗关键技术提供了指导。 构建了喷淋式兆声清洗数学模型，结合实验研究确认了喷淋式兆声系统不能满足14 nm节点中污染物控制要求。因此，本文开发了浸没式的兆声系统，实现了清洗液的循环、过滤和加热；且硅晶圆在兆声槽体中旋转，实现了兆声能量在硅晶圆表面的均匀分布，进而提升兆声清洗效果。建立了浸没式兆声系统数值模型，从流体和声学角度对腔室结构和振板布置进行了分析与优化设计。开展了系统的实验研究，获得了化学品浓度、兆声功率、硅片转速和化学品温度等工艺参数对清洗效果的影响规律，进而降低了78%的直径40 nm以上污染物数量。 晶圆刷洗技术是实现超洁净清洗的关键，本文研究了PVA(聚乙烯醇)滚刷的特点，探究了硅晶圆夹角、清洗刷转速、硅片转速及清洗刷与硅晶圆距离等关键参数对污染物的去除效率的影响规律，提出了面向高端制程硅晶圆刷洗的工艺优化方法，在兆声优化后的基础上，降低了95%的直径40 nm以上污染物数量。 干燥是晶圆清洗的最后一道关口，本文针对竖直旋转马兰戈尼Marangoni干燥技术，进行了深入的仿真和实验研究。分析了腔体在不同硅晶圆转速下的气流分布；阐明了Marangoni效应与离心力联合作用的晶圆干燥机制。结合实验研究，对硅晶圆转速、异丙醇（IPA）、氮气流量、去离子水流量、摆臂运动速度、干燥时间等参数进行了系统研究，提出了最佳干燥工艺方法，在兆声、刷洗优化后的基础上，降低了98.5%的直径40 nm以上污染物数量。 基于对各关键清洗技术系统的研究开发和多种清洗技术联合作用，开发的硅晶圆成套清洗工艺最终满足了14 nm节点对污染物的控制要求。该研究成果已应用于国内两个十二英寸集成电路制造工厂，成功将40 nm以上污染颗粒稳定的控制在50颗以下，提升了工厂产品良率，给产业界创造了较大的价值。
Silicon chemical mechanical polishing (CMP) technology is widely used in silicon substrate manufacturing, chip front-end manufacturing, and wafer reclaim manufacturing. With the rapid development of technology nodes, the cleaning technology after silicon CMP is facing enormous challenges, becoming increasingly a "bottleneck" problem in the development of silicon CMP manufacturing processes. This paper systematically studies the key technologies for silicon CMP post cleaning required by the 14 nm technology node. The types and generation mechanism of pollutants in silicon CMP were analyzed in depth. Through analysizing the distribution, morphology, and composition of pollutants, the morphology and main characteristics of major defects such as particles, organic residues, water marks, and bacteria were clarified, and the main sources of various pollutant defects were discussed in depth, providing guidance for in-depth research on key wafer cleaning technologies. A mathematical model of spray type megasonic cleaning was constructed, and experimental research confirmed that the spray type megasonic system could not meet the pollutant control requirements in the 14 nm node. Therefore, an immersion type megasonic system has been developed to achieve circulation, filtration, and heating of the cleaning fluid; And the silicon wafer rotates in the megasonic tank, achieving a uniform distribution of megasonic energy on the surface of the silicon wafer, thereby improving the megasonic cleaning effect. A numerical model of the immersion type megaacoustic system was established, and the chamber structure and vibration plate arrangement were analyzed and optimized from the perspectives of fluid and acoustics. Systematic experimental studies have been conducted to obtain the laws of the influence of process parameters such as chemical concentration, megasonic power, silicon wafer rotation speed, and chemical temperature on cleaning effectiveness, thereby significantly reducing 78% pollutants with a diameter of more than 40 nm. Wafer brushing technology is the key to achieving ultra clean cleaning. This paper studies the characteristics of PVA (Polyvinyl alcohol) brushes, explores the impact of key parameters such as silicon wafer angle, cleaning brush rotation speed, silicon wafer rotation speed, and the distance between the cleaning brush and the silicon wafer on the removal efficiency of pollutants, and proposes a process optimization method for high level process silicon wafer brushing, which reduced 95% 40 nm pollutants combined with immersion megaacoustic best condition. Drying is the last module of wafer cleaning. This article has conducted in-depth simulation and experimental research on the vertical rotation Marangoni drying technology. The gas flow distribution in the cavity at different silicon wafer speeds was analyzed. The mechanism of wafer drying under the combined action of Marangoni effect and centrifugal force was clarified. Based on experimental study, the parameters such as silicon wafer rotation speed, IPA/N2 flow rate, deionized water flow rate, swing arm movement speed, and drying time were systematically studied, and the optimal drying process method was proposed, which reduced 98.5% 40 nm pollutants combined with immersion megaacoustic and brushes best condition. Based on the research and development of various key cleaning technology systems and their coupling effection, the silicon wafer cleaning process meets the pollutant control requirements of the 14 nm node. The research results have been applied to two 12-inch integrated circuit manufacturing plants in China, successfully controlling 40 nm pollution particles to less than 50, improving the yield of factory products, and creating greater value for the industry.
