We report on high-resolution potential imaging of heterogeneous surfaces by means of Kelvin probe force microscopy, working in frequency modulation mode ͑FM-KPFM͒, performed in ultrahigh vacuum. To study the limits of potential and lateral resolutions in FM-KPFM, we have investigated clean surface of compound semiconductor InSb͑001͒ and the same surface with some submonolayer coverages of KBr and Au. It was found that long-and short-range bias-dependent interactions, acting between the tip and the surface, could be detected and that both interactions contribute to the measured contact potential difference ͑CPD͒ signal. On the one hand, when only the long-range electrostatic interactions between the tip and the surface are active, the CPD map provides the distribution of the local surface potential on the imaged sample with the lateral resolution and the correctness of the measured values depending on the measurement conditions. For this case, the experimental findings were compared with the predictions of theoretical calculations based on a realistic model for the cantilever-sample geometry. On the other hand, when the short-range and bias-dependent interactions are detected, FM-KPFM provides even the sub-nanometer contrast in the CPD signal. In this situation, however, the measured CPD signal is not related to the sample surface potential but reflects the properties of the front tip atom-surface atom interactions.

The bias potential, V bias, is the key quantity for the Kelvin probe force microscopy (KPFM) measurements and interpretation. Using an efficient method for electrostatic force determination, V bias has been calculated for tip–plane systems, with realistic tip geometry and for non-uniform potential distributions on the plane. The considered potential distributions on the plane include a potential step, a quadratic potential island, and two quadratic potential islands with varying separation. V bias has been evaluated along three different schemes, i.e., from the minimization of electrostatic force, from the force gradient, and from the integral formula. We have studied V bias as a function of tip–surface distance, island size, vibration amplitude, and tip sharpness radius (the so called nanotip). We have found that there are substantial differences between the gradient and integral schemes for V bias evaluation. We have determined that the nanotip presence favors an accurate potential mapping, particularly for small potential islands. The implications of the obtained results for KPFM method are also discussed.