Critical dimension atomic force microscopy is employed in semiconductor manufacturing as a reference system, used to provide accurate information to calibrate other tools. However, faced with increasingly challenging features, operators of these systems have needed to use multiple specialized tip types and scan modes on a sample in order to acquire a complete data set. To overcome this need, we measure the potential biases introduced within such a hybrid data set with respect to sidewall slope angle. Measurement results are presented for features with varying sidewall slope across a range of angles just above and below 90 deg. This sample was scanned with a variety tips and two different scan modes: critical dimension (CD) mode, an adaptive sidewall scanning mode able to detect limited undercut; and deep trench (DT) mode, a dynamic top-down-only mode commonly employed on small, challenging measurement features. In DT mode, we detected a cutoff point at 89 deg, below which all tips track the surface with a size-dependent bias. In contrast, CD mode produced results that were much more invariant to tip size except for the smallest tips. The smallest tips in CD mode exhibited divergent behavior, compared to both the mid-range CD tips and one another.