Applications of Electron Energy Loss Near-Edge Structure Analysis for Defect Characterization in Advanced Semiconductor Devices
Applications of Electron Energy Loss Near-Edge Structure Analysis for Defect Characterization in Advanced Semiconductor Devices
Monday, November 17, 2025: 2:20 PM
3 (Pasadena Convention Center)
Summary:
This paper investigates the application of Electron Energy-Loss Near-Edge Structure (ELNES) analysis as a powerful tool for identifying defects in advanced semiconductor devices. Analysis of the ELNES region of an Electron Energy-Loss Spectroscopy (EELS) spectrum reveals chemical bonding characteristics and oxidation states, providing information beyond basic elemental identification. This methodology includes high-resolution ELNES data acquisition using a 200kV probe-corrected, cold-FEG TEM equipped with a post-column EELS spectrometer. With the use of advanced instrumentation - such as aberration correctors, monochromators, and more recently direct detection cameras - EELS can now be performed with superior energy resolution and sub-nanometer spatial resolution. The technique is demonstrated using the Si L2,3-edge in a FinFET technology. A subsequent case study highlights the differentiation between elemental tantalum (Ta) and tantalum oxide bonding (Ta-O) by analyzing the oxygen K-edge in the barrier layer of a via stack. Results demonstrate the effectiveness of ELNES analysis in distinguishing different oxides via a fingerprinting technique, highlighting its ability to identify potential barrier layer anomalies. ELNES analysis can also be applied to investigations of oxidation-related anomalies in gate stacks, contamination or oxidation during manufacturing processes, interfacial degradation in multilayer structures, among other failure mechanisms. As semiconductor devices continue to increase in complexity, ELNES provides not just complementary insight, but in some cases, the only viable method for identifying bonding-related anomalies at the atomic scale. This work emphasizes the critical role of ELNES in enabling a comprehensive failure analysis workflow, supporting effective root cause finding in current and next-generation integrated circuits.
This paper investigates the application of Electron Energy-Loss Near-Edge Structure (ELNES) analysis as a powerful tool for identifying defects in advanced semiconductor devices. Analysis of the ELNES region of an Electron Energy-Loss Spectroscopy (EELS) spectrum reveals chemical bonding characteristics and oxidation states, providing information beyond basic elemental identification. This methodology includes high-resolution ELNES data acquisition using a 200kV probe-corrected, cold-FEG TEM equipped with a post-column EELS spectrometer. With the use of advanced instrumentation - such as aberration correctors, monochromators, and more recently direct detection cameras - EELS can now be performed with superior energy resolution and sub-nanometer spatial resolution. The technique is demonstrated using the Si L2,3-edge in a FinFET technology. A subsequent case study highlights the differentiation between elemental tantalum (Ta) and tantalum oxide bonding (Ta-O) by analyzing the oxygen K-edge in the barrier layer of a via stack. Results demonstrate the effectiveness of ELNES analysis in distinguishing different oxides via a fingerprinting technique, highlighting its ability to identify potential barrier layer anomalies. ELNES analysis can also be applied to investigations of oxidation-related anomalies in gate stacks, contamination or oxidation during manufacturing processes, interfacial degradation in multilayer structures, among other failure mechanisms. As semiconductor devices continue to increase in complexity, ELNES provides not just complementary insight, but in some cases, the only viable method for identifying bonding-related anomalies at the atomic scale. This work emphasizes the critical role of ELNES in enabling a comprehensive failure analysis workflow, supporting effective root cause finding in current and next-generation integrated circuits.