Ellipsometry and Standardization: Validation of Models - Key Issue for Reference Coatings and Non-ideal Samples
- Standardisation of ellipsometry is mainly based on three documents: pioneering work at NIST [1] on film standards, a publicly available specification and stan-dard testing procedure [2] for real-world samples by BAM and a standard specification for integrated metrology by SEMI [3]. Other activities refer to inter-laboratory comparisons for selected layer/substrate systems and analytical methods including ellipsometry [4] and the development of combined standards for thickness d and step height h [5].
Generally, precision of ellipsmetry in terms of raw data Ψ and Δ meets the major demands for all ellipsometric principles (RPE, RAE, PME, RCE, NE).
As a result, fingerprint capabilty is al-most perfect for various layer/substrate systems. However, this is not true for accuracy in terms of physical quantities such as thickness d, optical (n, k) and dielectric constans (ε1, ε2,) as modelling is required. Hence, the layer/substrate and material dependent validation of models is nowadays a keyStandardisation of ellipsometry is mainly based on three documents: pioneering work at NIST [1] on film standards, a publicly available specification and stan-dard testing procedure [2] for real-world samples by BAM and a standard specification for integrated metrology by SEMI [3]. Other activities refer to inter-laboratory comparisons for selected layer/substrate systems and analytical methods including ellipsometry [4] and the development of combined standards for thickness d and step height h [5].
Generally, precision of ellipsmetry in terms of raw data Ψ and Δ meets the major demands for all ellipsometric principles (RPE, RAE, PME, RCE, NE).
As a result, fingerprint capabilty is al-most perfect for various layer/substrate systems. However, this is not true for accuracy in terms of physical quantities such as thickness d, optical (n, k) and dielectric constans (ε1, ε2,) as modelling is required. Hence, the layer/substrate and material dependent validation of models is nowadays a key issue.
In contrast to the sub-nanometer sensi-tivity in vertical direction, the effective lateral resolution of almost all ellipso-metric principles except for imaging null
ellipsometry is rather poor, usually be-yond 100 µm.
Hence, nanometer-scaled homogeneity testing and the measurement of micro- and nanopatterned structures is not feasible by standard configurations.…