How can you tell a defect from a harmless imperfection? In failure analysis, precision matters.
Discover how Scanning Acoustic Microscopy (SAM) can make or break your qualification process—and why selecting the right method is critical for ASIC reliability.
Failure analysis is a cornerstone of successful ASIC development and production. In today's increasingly complex semiconductor landscape, where devices continue to shrink in size while growing in functionality, identifying and resolving failures quickly is essential to maintaining product quality, reliability, and market competitiveness.
For ASIC projects specifically, failure analysis serves multiple critical functions:
When suppliers and customers disagree about observed failures, having robust, accurate analysis techniques becomes even more critical. This article explores how different types of Scanning Acoustic Microscopy (SAM) modes, settings, and equipment resolutions can yield different results when analyzing the same device, demonstrating the importance of proper technique selection and interpretation.
Scanning Acoustic Microscopy is a non-destructive imaging technique used in the electronics industry to inspect internal structures of components. It uses high-frequency sound waves to create detailed images of a sample's interior, allowing engineers to detect defects such as delaminations, voids, and cracks that are not visible from the surface.
SAM is a brilliant tool for analysis but be careful to avoid misinterpretation or miss out on important failures.
Have you ever tried being stuck in discussions with your supplier about failures you have observed? Being convinced by great pictures showing nothing is wrong? You are not alone.
These case studies explore how different types of SAM modes, settings and equipment resolution can get different results when analyzing the same device.
To illustrate SAM’s impact, let’s consider a case study about a BGA package with two dies stacked on top of each other. As a standard package, it should not raise any issues. The first engineering lot is always analyzed in different ways in addition to functional testing, to show if any general workmanship-related concerns must be raised, e.g. crossed wires, wire sweep and delaminations in the package. In this case, all was good.
Next step is qualification of device and package, first step in package qualification being MSL (Moisture Sensitivity Level) classification. This level determines how much time you have after breaking the sealing of a dry-pack, before the components must be through the soldering operation.
In this case, the package was to be classified as MSL3.
Post MSL3 stress, electrical testing and optical inspection showed no abnormalities. However, CSAM (C-Mode Scanning Acoustic Microscopy) performed from component topside detected some delamination at the edge of the bottom die. In some areas, the delamination was penetrating to the edge of the component. [See Fig 1]
Figure 1: CSAM image – delamination/crack in substrate observed at the image to the left (light color at substrate area).
Based on this knowledge, the failed devices were optically inspected with higher magnification at component edges, and cracks in the substrate were observed. [Fig 2]
Figure 2: Optical image showing a close-up of the edge of a device – a crack in the substrate can be observed.
To further analyze the cause of the delamination and cracking phenomenon, a cross-sectioning was performed. Results revealed that voids between the die-attach film and substrate led to cracking through the die-attach film during the soldering operation (pop-corn). These cracks propagated down in the substrate where the die-attach film ended. [Fig 3]
As soon as these results were shared with the supplier, they quickly accepted that the die-attach process had to be optimized.
After extensive Design of Experiments (DoE) , new parameters were found, and the package successfully passed MSL3 classification test.
The selection of appropriate transducer frequency is crucial for effective SAM analysis. Different frequencies provide varying levels of penetration and resolution. Higher frequency transducers offer better resolution but may not penetrate as deeply into the sample, while lower frequency transducers can penetrate deeper but with reduced resolution.
In this case study, we encountered a situation where a second equipment was needed for SAM analysis. The critical delamination could only be detected using a 110 MHz transducer on our primary equipment. When attempting to use this second equipment, we discovered that its 110 MHz transducer couldn't penetrate to the substrate surface, and the 50 MHz transducer couldn't detect the delamination either. [Fig 4]
This demonstrates the importance of selecting the optimal equipment for each specific analysis scenario. The transducer must be able to penetrate to the area of interest while providing sufficient resolution to detect the defect. When working with suppliers, it's advisable to inquire about which frequency they're using for their investigations to ensure appropriate analysis depth and resolution.
Figure 4: CSAM images showing analysis of the same component, same area with two different frequencies and two different sets of equipment (50 MHz to the left and 110 MHz to the right).
In through-scan technique, the sound is transmitted through the whole sample, and the obtained image appears black if any delamination is present that will reflect all signal. While this method confirms the presence of delamination, it does not indicate which layer is affected.
In Fig. 5, a through-scan analysis of a MSL3-failed device revealed a significantly larger delaminated region under the die—not visible using CSAM alone but later confirmed via cross-section analysis.
However, a very narrow delamination (materials not adhering together, but no real gap) may not be seen by through-scan, only by CSAM. For some packages, you will still be able to detect delamination under the die using CSAM.
Key Takeaway: Combining Through-Scan and CSAM methods provides a more comprehensive failure analysis, especially for complex packages.
Figure 5: Through-scan image of a device failed in MSL3 classification test. The whole black area represents delamination/cracking.
To strengthen supply robustness, Presto Engineering has more than one source for packaging of main products. Therefore, sometimes new suppliers must be qualified. This case focuses on a standard QFN (quad flat no-lead) package qualification.
The first product was analyzed and delamination observed between die corners and encapsulation material. It is not acceptable to have die surface delamination. However, if it were in the die-attach region it would be considered fully acceptable.
In the supplier setup, it had not been detected because they (like many others) used automated red coloring to highlight areas where the signal changes direction, indicating potential delamination. A general example is shown in Fig.7.
This red coloring result is, however, fully dependent on the settings and should be complemented by A-scan analysis (Fig 6) to confirm the interpretation.
Figure 6: CSAM picture combined with A-scans for interpretation.
In this case, this revealed that the issue was at the die surface, not the die-attach region, as the supplier initially claimed.
It was accepted that a delamination was present but there was no agreement on which interface. The A-scan was used for interpretation: the encapsulation material to the die surface is the first interface (from top), and therefore, when it is the first signal that changes, you know that it is in this interface you have the delamination.
After further discussions, the supplier implemented a plasma cleaning step prior to molding, resolving the issue.
We also had this discussion about red coloring recently with one of our old and qualified suppliers. It turned out after a little discussion that when they tuned the setting, getting the A-scan signal to be bigger, they experienced the same red coloring as we did and recognized the delamination.
Figure 7: Example of scanning using automatic red coloring of areas where the signal changes direction
These case studies demonstrate the strength of the SAM technique in failure analysis, but also the importance of choosing the right scanning mode(s) and parameters.
Most importantly, it shows that red coloring should not be blindly relied upon, even though it can be a strong tool when used with appropriate consideration.
It is crucial to understand that CSAM is a technique revealing delaminations that are not gaps and therefore may not be critical to the reliability of the device.
Whether observed delamination poses a reliability risk depends on the location and whether the delamination increases when exposed to reliability stress.
By applying rigorous SAM methodologies, semiconductor companies can proactively detect failures, improve supplier accountability, and enhance long-term device reliability.
As a full-service ASIC provider, Presto Engineering combines comprehensive failure analysis capabilities with extensive ASIC design and testing expertise. Our in-house failure analysis lab, staffed by experienced engineers, works closely with wafer fabs and assembly houses to ensure optimal quality throughout the production process. This integrated approach allows us to quickly identify and resolve potential issues, providing comprehensive failure analysis reports and implementing preventive measures based on historical data.
By leveraging advanced techniques like SAM, Presto Engineering supports customers throughout their ASIC development journey, ensuring robust and reliable solutions in today's demanding semiconductor industry.
Helle Rønsberg, Manager for Failure Analysis & Quality Control, Presto Engineering