A Framework for Quantifying Effects of Characterization Error on the. . .
237
Fig. 10 Mismatched volume (MMV) as a function of the interaction volume normalized by the
average feature size, assuming that sample spacing ∼0 so that resolution error is zero
is based on a number of assumptions, it provides a useful approximation that
can be applied to determine a reasonable resolution without undergoing lengthy
simulations.
3.3 Interaction Volume
As discussed in Sect. 2, changing the size of the interaction volume (IV) in the
simulation is a representation of changing beam energy. More realistic predictions
of the relationship between interaction volume and beam energy in a specific
SEM configuration for a given material require calibration or the use of simplified
models such as Eq. 1. The example shown here serves only to illustrate how relative
increases or decreases in interaction volume affect error.
As was done in the resolution study, the size of the interaction volume is
normalized by the average feature size. Figure 10 shows the mismatched volume
MMV as a function of the normalized interaction volume, with zero resolution error.
It is worth mentioning that for large features (>10 μm) interaction volumes on the
order of 10–25% are infeasible in modern SEM; however, for small features they are
certainly possible. The MMV values for the smaller values of normalized interaction
volume are lower than the typical MMV values associated with typical sampling
resolution, suggesting that the error due to interaction volume may be lower than
that due to resolution in many situations.
237
Fig. 10 Mismatched volume (MMV) as a function of the interaction volume normalized by the
average feature size, assuming that sample spacing ∼0 so that resolution error is zero
is based on a number of assumptions, it provides a useful approximation that
can be applied to determine a reasonable resolution without undergoing lengthy
simulations.
3.3 Interaction Volume
As discussed in Sect. 2, changing the size of the interaction volume (IV) in the
simulation is a representation of changing beam energy. More realistic predictions
of the relationship between interaction volume and beam energy in a specific
SEM configuration for a given material require calibration or the use of simplified
models such as Eq. 1. The example shown here serves only to illustrate how relative
increases or decreases in interaction volume affect error.
As was done in the resolution study, the size of the interaction volume is
normalized by the average feature size. Figure 10 shows the mismatched volume
MMV as a function of the normalized interaction volume, with zero resolution error.
It is worth mentioning that for large features (>10 μm) interaction volumes on the
order of 10–25% are infeasible in modern SEM; however, for small features they are
certainly possible. The MMV values for the smaller values of normalized interaction
volume are lower than the typical MMV values associated with typical sampling
resolution, suggesting that the error due to interaction volume may be lower than
that due to resolution in many situations.