Image processing for microprocessors TIM voids detection

Thermal interface material (TIM) is a substance designed to enhance thermal conductivity between two surfaces requiring efficient heat transfer. It is commonly applied between a microprocessor's silicon die and a heat sink or other cooling components, such as heat spreaders, to minimize thermal resistance caused by air gaps and surface irregularities. This ensures optimal operating temperatures, thereby improving the performance and longevity of the microprocessor.However, voids can form within the TIM layer during application to contact surfaces, significantly impairing thermal dissipation. To address this issue, non-destructive testing (NDT) imaging techniques are often employed to inspect and detect TIM voids. Common methods include Scanning Acoustic Microscopy (SAM) and X-ray Laminography, which provide detailed visualization of void distribution and structural defects.To automatically quantify void occupation within each rectangular shape die ROI (Region Of Interest), a traditional image processing-based void detection software has been developed. This software can automatically identify the rectangular chip die area and detect TIM voids within it. It then calculates the void occupancy percentage for each die area.The Niblack algorithm is employed in our void detection software. Histogram curve analysis is applied to automatically determine the threshold for the Niblack algorithm's test value, with manual adjustments available for fine-tuning. This approach allows for the recovery of small or weak voids that may have been initially missed. It also enhances or suppresses minor void information, leading to more balanced and consistent void detection outcomes.Currently, the software is utilized for quality assurance and failure analysis. It is the first commercially available software specifically for automatic TIM void analysis.