edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 27 NO. 2 6 was poor. Compared to a known good sample image, the scattering signature is markedly different. This allowed the failure analysis team to focus their efforts to look for package failures, driving deeper to root cause. For cases like these, it is important to have a known good sample to be used for comparisons. CASE STUDY 2: DETECTING IN-OPERATION FAILURES Failure in modules during operation is first detected by a drop in insertion loss greater than what is expected. In such cases, one of the best uses of the infrared scope is narrowing down the list of suspects in an optical path. One of the more recent failure modes the GlobalFoundries team has reported in literature is the permanent damage that occurs in our PICs when the input power goes beyond a certain threshold.[8] Using the infrared microscope, we were able to isolate the failure in the optical path to be at the silicon-based spot size converters, which is used to bring in light from an external fiber to the silicon chip (Fig. 5). The primary mechanism is due to the two-photon absorption that occurs in silicon waveguides. Siliconnitride based spot size converters which do not exhibit the two-photon absorption problem are better suited for high optical power applications. CASE STUDY 3: DETECTING YIELD DETRACTORS Some failure modes occur due to processing yield issues. In such cases, the first step toward driving to root cause is narrowing down the area on the chip. In this case, the IR images between a good and bad module clearly distinguish the failure mode location(s) in the optical path (Fig. 6). This information was later fed back to the FA team to dig deeper into looking for physical faults. CONCLUSION The near IR microscope is a powerful failure analysis tool to assist in silicon photonics applications. Its primary function is to provide the first signals and narrow down the scope of area where more detailed analysis needs to be performed. Localization of faults is the main strength in terms of failure analysis. With its ability to scale up to any microscope optics, the technique lends itself to a wide flexibility depending on the application. It is expected that the usefulness of this technique will continue as the photonics ecosystem evolves. REFERENCES 1. V. Gupta: “GLOBALFOUNDRIES Silicon Photonics,” Consortium for On-Board Optics (COBO), September 1, 2020. Fig. 5 (a) False color infrared image of spot size converter at lower power levels. (b) Infrared image after the permanent damage of coupler at high power levels. Fig. 6 (a) False color infrared image of known good PIC die under operation. (b) Infrared image of a yield detractor die at similar location. (a) (a) (b) (b)
RkJQdWJsaXNoZXIy MTYyMzk3NQ==