We show that a near-field scanning thermal microscope, which essentially detects the local density of states of the thermally excited electromagnetic modes at nanometer distances from some material, can be employed for nanoscale imaging of structures on that material's surface. This finding is explained theoretically by an approach which treats the surface structure perturbatively

We study the radiative heat transfer between a spheroidal metallic nanoparticle and a planar metallic sample for near-and far-field distances. In particular, we investigate the shape dependence of the heat transfer in the near-field regime. In comparison with spherical particles, the heat transfer typically varies by factors between 1/2 and 2 when the particle is deformed such that its volume is kept constant. These estimates help to quantify the deviation of the actual heat transfer recorded by a near-field scanning thermal microscope from the value provided by a dipole model which assumes a perfectly spherical sensor.

We apply an analytical approach for determining the near-field radiative heat transfer between a metallic nanosphere and a planar semi-infinite medium with some given surface structure. This approach is based on a perturbative expansion, and evaluated to first order in the surface profile. With the help of numerical results obtained for some simple model geometries we discuss typical signatures that should be obtainable with a nearfield scanning thermal microscope operated in either constant-height or constant-distance mode.

We develop a perturbative approach for calculating, within the quasistatic approximation, the shift of surface resonances in response to a deformation of a dielectric volume. Our strategy is based on the conversion of the homogeneous system for the potential which determines the plasmonic eigenvalues into an inhomogeneous system for the potential's derivative with respect to the deformation strength, and on the exploitation of the corresponding compatibility condition. The resulting general expression for the first-order shift is verified for two explicitly solvable cases, and for a realistic example of a deformed nanosphere. It can be used for scanning the huge parameter space of possible shape fluctuations with only quite small computational effort.

We suggest to exploit the shape-dependence of the near-field heat transfer for nanoscale thermal imaging. By utilizing strongly prolate or oblate nanoparticles as sensors one can assess individual components of the correlation tensors characterizing the thermal near field close to a nanostructured surface, and thus obtain directional information beyond the local density of states. Our theoretical considerations are backed by idealized numerical model calculations.