Characterization of process-induced damage in Cu/low-k interconnect structure by microscopic infrared spectroscopy with polarized infrared light

Abstract

Microscopic Fourier-transform infrared (FT-IR) spectra are measured for a Cu/low-k interconnect structure using polarized IR light for different widths of low-k spaces and Cu lines, and for different heights of Cu lines, on Si substrates. Although the widths of the Cu line and the low-k space are 70 nm each, considerably smaller than the wavelength of the IR light, the FT-IR spectra of the low-k film were obtained for the Cu/low-k interconnect structure. A suitable method was established for measuring the process-induced damage in a low-k film that was not detected by the TEM-EELS (Transmission Electron Microscope-Electron Energy-Loss Spectroscopy) using microscopic IR polarized light. Based on the IR results, it was presumed that the FT-IR spectra mainly reflect the structural changes in the sidewalls of the low-k films for Cu/low-k interconnect structures, and the mechanism of generating process-induced damage involves the generation of Si-OH groups in the low-k film when the Si-CH3 bonds break during the fabrication processes. The Si-OH groups attract moisture and the OH peak intensity increases. It was concluded that the increase in the OH groups in the low-k film is a sensitive indicator of low-k damage. We achieved the characterization of the process-induced damage that was not detected by the TEM-EELS and speculated that the proposed method is applicable to interconnects with line and space widths of 70 nm/70 nm and on shorter scales of leading edge devices. The location of process-induced damage and its mechanism for the Cu/low-k interconnect structure were revealed via the measurement method.

Diagnosis System of Microscopic Hyperspectral Image of Hepatobiliary Tumors Based on Convolutional Neural Network

“…Hyperspectral image acquisition equipment can usually provide continuous band images from visible light to nearinfrared region, and these band images have higher discreteness than RGB images [12], and the band interval is also narrower. As hyperspectral images provide rich information in spatial and spectral dimensions and the resolution is high, it was first applied to remote sensing analysis and later applied to medicine, chemistry, precision agriculture, and environmental engineering.…”