Semiconductor Manufacturing Process

With the global push for carbon reduction, the growing market share of electric and hybrid vehicles has significantly increased the demand for compound semiconductors used in these electric transport systems. Leading international equipment manufacturers such as DISCO (Japan), Asahi Diamond, and Shin-Etsu Silicone are actively developing high-efficiency, high-yield processing equipment and components for materials like SiC, GaP, and GaAs, laying a solid foundation for compound semiconductor processing technologies.
Although Taiwan's semiconductor industry, led by major players like TSMC, UMC, and Unimicron, ranks among the world's top, these companies mainly focus on silicon wafer processes, advanced transistors, and PCB packaging technologies. They have invested relatively little in compound semiconductor processing and packaging. To address this gap and support the nation’s development goals in power semiconductors, automotive electronics, and green energy technologies, AIM-HI has prioritized the development of power semiconductor processing and packaging techniques.
In 2024, we focused on three major technological developments:

• a substrate processing signal feature analysis database
• a digital model for crack propagation in packaging materials
• a dynamic signal feature analysis system for machining processes

Details are outlined below:
In 2024, our team successfully developed and validated several key technologies:

• Built a machining signature analysis and interpretation model for wafer processing machines. By analyzing over six process parameter sets, we increased processing efficiency by 50%.
• Developed a numerical analysis model for single-crystal SiC crack propagation, achieving over 88% accuracy, which shortened process parameter integration time for partner companies by 30%.
• Implemented laser-assisted mechanical cutting for hard-brittle materials, improving the success rate of compound semiconductor material processing by 45% via laser pre-treatment.

1. Highlight Technology 1: Wafer Machining Signature Analysis and Interpretation Model
   Our team identified key machining signatures during wafer processing and, through combined mathematical and numerical analysis with multiple validation methods, built a high-efficiency model for estimating post-machining surface features. This provides a critical tool for future wafer process improvements. The results have been published in the International Journal of Advanced Manufacturing Technology.
2. Highlight Technology 2: Numerical Model for Crack Propagation in Single-Crystal SiC
   We analyzed the anisotropic properties of single-crystal SiC to identify key conditions affecting damage during cutting. Using TEM to measure microcrack geometries post-cutting and applying the Extended Finite Element Method (XFEM), we developed numerical models that simulate pre-crack geometries under different cutting parameters. These models help production engineers evaluate stress intensity factors based on cutting depth and feed rate variations.
3. Highlight Technology 3: Laser-Assisted Mechanical Cutting for Hard-Brittle Materials
   SiC wafer dicing is a critical step in compound semiconductor packaging. Since high precision and challenging material properties make processing difficult, we employed laser pre-treatment to soften hard-brittle materials, increasing both efficiency and yield. Our combined laser and mechanical cutting approach produced SiC wafers with no edge chipping defects. Cross-sectional analysis also revealed stress-induced amorphization of SiC, helping link process quality to microstructural changes. Results have been published in Materials Today Communications.