移动通信技术的快速发展，推进了射频前端滤波器的多样化需求，5G新增N77等频段要求滤波器具有宽带宽、低损耗的新特点，传统声波型滤波器聚焦低频窄带宽滤波，不满足需求。虽然LTCC（Low Temperature Co-fired Ceramic，低温共烧陶瓷）滤波器聚焦宽带宽，但是加工难度大、尺寸不易减小，因此，IPD（Integrated Passive Device，集成无源器件）滤波器应运而生。相比硅衬底，玻璃材料有着天然的低损耗、高阻抗等优势，采用TGV（Through Glass Via，玻璃通孔）技术的3D螺旋电感具有更高的品质因素，可大幅提升IPD滤波器的性能。但8inch及以上大尺寸TGV、深孔填Cu、RDL（Re-Distribution Layer，重布线层）等关键技术国内整体仍处于开发阶段，亟需快速突破。基于以上分析，本论文研究内容如下：（1）高质量金属化TGV基板是玻璃基IPD滤波器产出的前提，明确了制备金属化TGV基板的关键影响因素。系统分析了激光诱导技术中焦点位置和激光功率对TGV形貌的影响，验证了刻蚀液对玻璃形貌的影响，实现了孔径50μm、腰部25μm、玻璃厚度250μm的TGV基板。采用双面磁控溅射，TGV深孔种子层台阶覆盖率提升到12.5%，250nm Cu种子层厚度匹配分段电镀，实现了TGV无孔洞金属填充。（2）优化RDL关键工艺技术，产出满足要求的电容电感和滤波器件。筛选了3种电感方案，确定了减成法为电感RDL工艺路线。采用10%稀H2SO4及Ar等离子体溅射预处理去除TGV-Cu（TGV孔Cu）表面氧化层，利用Ti、Cu金属互扩散原理，溅射Ti、Cu结构，有效减少界面孔洞，降低电感阻抗。建立仿真模型，分析300℃高温环境下TGV-Cu膨胀的影响因素，建立了孔径、深宽比、玻璃厚度与应力的对应关系，便于优化设计。全因子SiNx成膜方案研究，SiNx均一性从7%提升至3%，最后制作Solder bump，形成完整玻璃基IPD滤波器，测试结果满足设计规格。（3）采用工程管理的工具方法，改善工艺技术，提出管理建议。针对改性玻璃的湿法腐蚀进行2因子多水平的DOE实验设计，优选实验条件，提升验证效率。使用SPC方法针对电容数据进行过程能力分析，指出相关不足，后续产业化落地需要重点关注。
The rapid development of mobile communication technology has put forward the diversified requirement for the filters in radio frequency front end. The addition bands such as N77 in 5G requires filters to have new characteristics such as wide bandwidth and low loss. Traditional acoustic wave filters focus on low-frequency narrow bandwidth filtering, which does not meet the requirements. The LTCC (Low Temperature Co fired Ceramic) filters focus on a wide frequency band, but they are difficult to process and the size is not easy to reduce. Therefore, the integrated passive device (IPD) filters come into being. Compared to the silicon substrate, glass materials have natural advantages, such as low loss, high impedance, etc. Three-dimension (3D) spiral inductors using through glass vias (TGV) technology have superior quality factors, which can greatly improve the performance of IPD filters. However, due to the fact that key technologies such as TGV, filling Cu in deep hole, and RDL (Re Distribution Layer) based on 8-inch and larger wafer are still in the development stage domestically, there is an urgent need for rapid breakthroughs. Based on the above analysis, the research content of this paper is as follows: (1) High quality metallized TGV substrate is a prerequisite for the production of glass based IPD filters. The key influencing factors for the preparation of metallized TGV substrates are identified. The influence of focal position and laser power on TGV morphology in laser induced technology is systematically analyzed. TGV glass with an aperture of 50μm, waist diameter of 25μm, and glass thickness of 250μm is achieved. Using double-sided magnetron sputtering, the step coverage of the TGV deep-hole seed layer is increased to 12.5%, and the thickness of the 250 nm Cu seed layer is matched by the segmental electroplating, realizing the Cu filling of TGV without holes. (2) Capacitive, inductive and filter devices that meet the requirements are output by improving the key process technology of RDL. Three inductance schemes are compared and the subtractive method is determined as the inductance RDL process route. 10% dilute H2SO4 and Ar plasma sputtering are used to remove the oxide layer on the surface of TGV-Cu. Based on the principle of mutual diffusion of Ti and Cu metals, the Ti and Cu structures are sputtered to effectively reduce the interface holes and reduce the inductance impedance. A simulation model is established to analyze the factors affecting the expansion of TGV-Cu at 300°C, and the relationship between the aperture, aspect ratio, glass thickness and stress is established to optimize design. The uniformity of SiNx films is increased from 7% to 3%, through the full-factor researches. Finally, the glass-based IPD filter is produced after solder bump process, and the test results meet the design specifications。（3）Engineering management tools and methods are used to improve the process technology and propose management suggestions. Conduct a 2-factor multi-level DOE experimental design for wet etch of TGV glass, optimize experimental conditions, and improve validation efficiency. Statistical process control (SPC) method in engineering management is adopted to analyze capacitance data, point out relevant shortcomings, which the follow-up industrialization needs to be focused on.
