Metrosemi

Menu

Candela® 8720 Surface Defect Inspection System

Unpatented Wafer Inspection Solution for Power Device Applications

The Candela® 8720 advanced surface inspection system captures a variety of mission-critical substrate and epitaxial defects for the LED, photonics, communications and other compound semiconductor markets. Implementation of automated wafer inspection with statistical process control (SPC) methodology can significantly cut yield loss due to epi defects, minimize metal-organic chemical vapor deposition (MOCVD) reactor process excursions, and increase MOCVD reactor uptime.
Automated defect inspection for LED materials, enabling enhanced quality control of substrates, fast time-to-root cause determination, and improved MOCVD process control
A single-tool solution that combines multiple optical inspection techniques in a single scan for maximum efficiency in automated defect detection and classification
High sensitivity to yield-impacting defects across multiple compound semiconductor substrates
The Candela 8720 system can be operated in three modes to meet the needs of various applications: high-throughput, standard resolution, and high resolution. In the high-throughput mode, the Candela 8720 system can be used as a simple particle counter for process tool monitoring and qualification applications. In the advanced classification / high-sensitivity mode, multiple detection channels enable accurate detection and classification of various defect types to characterize process-related issues and identify yield-impacting defects. The inspection method achieves full-surface coverage in minutes to produce high-resolution images and wafer maps with automatically classified defects.

Example defect images

Comparison to scatterometry systems

The Candela 8720 inspection system provides the option to save raw data collected from multiple detectors. Defect signatures seen by different detectors can vary by defect type, which can help process engineers to accurately classify defects. The Candela 8720 system can detect and classify both macro and micro defects. Micro defect classification is done based on comparison of the optical signature from the normal- and oblique-incidence illumination. Macro defect classification uses optical signatures and defect attributes. MOCVD processes produce a variety of defects when GaN epi is grown on different substrate materials (silicon, sapphire, PSS, SiC and GaN). The Candela 8720 system is sensitive to common yield-impacting defects including micro-pits, cracks, hexagonal bumps, showerhead droplets, crescents, scratches and other topographic defects. Accurate classification is critical to driving key corrective actions for process control. The image gallery (Fig. 4) shows examples of different types of detected defects.