Mueller matrix spectroscopy and physics-based machine learning for gate-all-around sheet-specific metrology

Abstract

The complex vertically stacked gate-all-around (GAA) manufacturing process drives the demand for more challenging inline metrology requirements. GAA technology with specific technical requirements starts from the first process step, 1) the superlattice, where the multi-stack Si/SiGe pairs must be grown defect-free with matched Si nanosheet thicknesses, and %Ge per layer, sharp interfaces, and a minimized subsequent thermal Ge diffusion across the stacks. More critical steps, among others, are the 2) partial recess of the sacrificial SiGe layers that precede 3) the inner spacers which prevent a channel to source/drain short circuit and reduce the parasitic capacitance, and 4) the channel release process at the “remove poly gate” module, where the SiGe is selectively removed before the high-k metal gate. Driven by tight performance control, a sheet-specific metrology solution is highly desired at each of the above four critical steps. The ideal solution for such an application is non-destructive, precise, accurate, and highly productive. In this paper, a scatterometry critical dimension (SCD) solution for the GAA sheet-specific measurement from various GAA structures is presented. The SCD solution includes an advanced and optimized full Mueller Matrix spectroscopic ellipsometry in conjunction with a physics-assisted machine learning (ML) algorithm. Additionally, the best methodology to address the solution's robustness to process variation is described and presented. It will be shown that an optimized signal-to-noise ratio combined with ML can provide a superior optical metrology solution to the growing challenge in GAA applications.

Leveraging x-ray scatterometry and machine learning to enable robust and high-throughput optical critical dimension metrology for advanced 3D flash memory contact hole profile

“…Each critical parameter model will be optimized independently, suggesting its advantage in increasing the sensitivity which also reducing the correlation between different parameters of interest, thus providing more freedom of parameter selection compared to the conventional RCWA model with limited floating parameters. Detailed information on this approach applied to both logic and DRAM cases can be found elsewhere [15,16]. The overall workflow of a 3D flash memory channel hole metrology solution using a physics-based ML approach is displayed in Figure 3.…”

“…A set of process-based synthetic spectra that contain the information of the underlayers' process variations and parameters of interest were generated using an RCWA EM solver as well as the geometric model representing the structure dielectric function. This ML algorithm will then leverage both the measured and simulated spectra with multiple angles of incidence (AOI) and a wide range of wavelengths to capture the correlation breaking between parameters and other sensitive features [15].…”

Optical in die NZO benefit for DRAM manufacturing

“…[8][9]. This work introduces a physical-based machine-learning algorithm [10][11][12] recipe that is capable of in-die overlay measurement by training both real spectra collected from SpectraShape 11k and theoretical spectra generated from the scatterometry model against the corresponding reference to predict the overlay value. Spectra from the training set was kept as a reference and compared with spectra under the testing set.…”