The performance of magnetic nanoparticles is intimately entwined with their structure, mean size and magnetic anisotropy. Besides, ensembles offer a unique way of engineering the magnetic response by modifying the strength of the dipolar interactions between particles. Here we report on an experimental and theoretical analysis of magnetic hyperthermia, a rapidly developing technique in medical research and oncology. Experimentally, we demonstrate that single-domain cubic iron oxide particles resembling bacterial magnetosomes have superior magnetic heating efficiency compared to spherical particles of similar sizes. Monte Carlo simulations at the atomic level corroborate the larger anisotropy of the cubic particles in comparison with the spherical ones, thus evidencing the beneficial role of surface anisotropy in the improved heating power. Moreover we establish a quantitative link between the particle assembling, the interactions and the heating properties. This knowledge opens new perspectives for improved hyperthermia, an alternative to conventional cancer therapies.

Heating and hyperthermia efficiency, though closely related, are not equivalent concepts: two different systems may have the same global heating performance but a completely different local heat release distribution, resulting therefore in different overall hyperthermia efficiency.

In the magnetic fluid hyperthermia (MFH) research field, it is usually assumed that achieving a uniform temperature enhancement (ΔT) of the entire tumour is a key-point for treatment. However, various experimental works reported successful cell apoptosis via MFH without a noticeable ΔT of the system. A possible explanation of the success of these negligible-ΔT experiments is that a local ΔT restricted to the particle nanoenvironment (i.e. with no significant effect on the global temperature T) could be enough to trigger cell death. Shedding light on such a possibility requires accurate knowledge of heat dissipation at the local level in relation to the usually investigated global (average) one. Since size polydispersity is inherent to all synthesis techniques and the heat released is proportional to the particle size, heat dissipation spots with different performances – and thus different effects on the cells – will likely exist in every sample. In this work we aim for a double objective: (1) to emphasize the necessity to distinguish between the total dissipated heat and hyperthermia effectiveness, and (2) to suggest a theoretical approach on how to select, for a given size polydispersity, a more adequate average size so that most of the particles dissipate within a desired heating power range. The results are reported in terms of Fe₃O₄ nanoparticles as a representative example.