February_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 21 NO. 1 20 ENSURING ADVANCED SEMICONDUCTOR DEVICE RELIABILITY USING FA AND SUBMICRON DEFECT DETECTION Doug Gray 1 , Dustin Kendig 1 , Andrew A.O. Tay 2 , and Ali Shakouri 3 1 Microsanj LLC, Santa Clara, Calif. 2 Singapore University of Technology and Design 3 Purdue University, West Lafayette, Ind. doug@microsanj.com EDFAAO (2019) 1:20-25 1537-0755/$19.00 ©ASM International ® INTRODUCTION Recent industry trends are placing increased require- ments on the need to fully understand the thermal behavior of today’s advanced semiconductor devices to ensure long-term reliability. Due to the critical nature of applications such as 5G, automotive electronics, artificial intelligence (AI), cloud storage, and military electronic systems, they all demand higher performance while simultaneously placing increased requirements on long- term reliability. Device developments that achieve higher power levels and faster switching speeds with increased functionality are driving device features to submicron levels and increasing complexity. The resulting power densities andpotential for higher operating temperatures, localized hot spots, and unanticipated time-dependent thermal anomalies are compounding the challenges of ensuring adequate reliability. Temperature has adirect impact ondevicemean-time- to-failure (MTTF). This can be assessedwith the Arrhenius equation: MTTF = Ce − E a /kT (Eq 1) where E a is the activation energy, k is the Boltzmann con- stant and T is the absolute temperature. Figure 1, plotted for an activation energy of 1.84 eV, shows the relationship between temperature and pro- jected MTTF for a typical advanced electronic device. In this case, a 20 degree increase in junction temperature lowers the projectedMTTF by an order of magnitude. The higher power densities resulting from shrinking geom- etries in today’s advanceddevice structures caneasily lead to such temperature increases. The key requirement with these devices is the ability to analyze thermal behavior on a scale consistent with their submicron geometries. While traditional thermal analysis techniques such as IR thermography and μ -Raman spectroscopy have been widely used for years, these techniques fall short due to resolution limitations incompatible with today’s advanceddevices. [1,2] Relying on traditional thermal analy- sis techniques risks the possibility of missing important thermal anomalies or small defects that could lead toearly device failure. This article describes a noninvasive thermal imaging approach based on the thermoreflectance principle. This technique can meet the spatial resolution requirements for advanced devices while also providing temporal resolution in the nanosecond range for analyzing time- dependent thermal events. THERMOREFLECTANCE-BASED IMAGING As the temperature of a material changes, the refrac- tive index and therefore the reflectivity of the material Fig. 1 MTTF versus device temperature, plotted for an activation energy of 1.84 eV.