Fig. 1 Smal spot measurment with large numerical aperture (NA)
Fig.2 The effect of large NA on reflectance spectrum for 200nm oxide measurement. Lines represent: 0, +-30deg(NA=0.5), +-50deg(NA=0.76),+-70 deg(NA=94) and +-90 deg(NA=1)
Fig.3 The effect of large NA on reflectance spectrum for 2000nm oxide measurement. Lines represent: 0, +-30deg(NA=0.5), +-50deg(NA=0.76),+-70 deg(NA=94) and +-90 deg(NA=1)
measurement is, typically, done using microscope objectives with low Numerical Apertures (NA). Most standard microscopes are using high -NA objectives i.e. a larger cone angle of light is striking the sample. High-NA objectives have many advantages: higher resolution, better light collection, etc. But they have some significant downsides when the goal is to measure film thickness.
The most serious problem is reduced contrast of interferometric signal that is used to determine film thickness. In a high-NA objective, light rays are refracted in the film at different angles (see Fig. 1), as a result, they have paths of different length in the film material. This means that they have different phase differences. Once different light rays are combined and phases are superimposed on the detector – the contrast between constructive and distructive interference peaks/valleys is diminished. The severity of this effect depends on the specific filmstack and NA. But, in general, effect increases with the thickness.
The effect is easy to see on an example of Si oxide measurement. Simulation of the UVVis reflectance spectrum (200-1000nm) for 200nm oxide is shown on Fig.2 It shows gradual degradation of the spectrum with the increase of the NA. But the changes are small and can be easily corrected. However, the same simulation for 2000nm oxide (Fig. 3 ) shows a more significant effect of the NA.
e has an option to correct for high NA effects (Fig. 4). However, if the interference signal is already weakened due to other artifacts (scattering, thickness non-uniformity, etc.) – the NA effect can make interence fringes practically disappear and, in this case, software correction cannot help. Use of objectives needs to be evaluated based on application requirements. If possible, it is always better to use low NA objectives. Low NA (ideally NA< 0.5) has a very small effect on measured spectra, practically, in all applications.