There were two reasons why we wanted to explore indenofluorene in the first place. First, theoretical calculations predicted indenofluorene to adopt a magnetic ground state with two unpaired electrons, which interact to form a high-spin triplet state (where the magnetic moments of both unpaired electrons point in the same direction). Second, the synthesis of indenofluorene had so far proven challenging. In 2017, a team led by scientists at the University of Oregon attempted to synthesize the molecule in solution, where it rapidly decomposed, and the hint that indenofluorene was obtained as a short-lived species came from the isolation of its dianion, which is a stable (albeit non-magnetic) species.2
We teamed up with chemists – Diego Peña’s team at Spain’s University of Santiago de Compostela, who synthesized a precursor to indenofluorene, stable under ambient conditions, which contained two extra hydrogen atoms (indenofluorene: C20H12, precursor: C20H14). The precursor was then sublimed under ultra-high vacuum on a crystalline gold surface, which was covered by two-monolayers-thick films of sodium chloride, and housed in a combined scanning tunneling and atomic force microscope operating at a temperature of 5 Kelvin (–268 °C). Using voltage pulses from the tip of the microscope, two hydrogen atoms were dissociated from the precursor to form indenofluorene (Figure 2).
With an indenofluorene molecule in front of us, we set about thoroughly characterizing it down to the single-atom level. Our job, we thought, was simple – theory unanimously predicted a magnetic state for indenofluorene. Accordingly, we expected that the densities of the frontier molecular orbitals (which are the highest-energy occupied orbitals and the lowest-energy unoccupied orbitals) imaged by scanning tunneling microscopy should correspond to two singly-occupied molecular orbitals, each occupied by an unpaired electron, as should be the case for magnetic polycyclic conjugated hydrocarbons.
Correspondingly, for a non-magnetic polycyclic conjugated hydrocarbon, all electrons are paired, and the frontier orbitals correspond to doubly-occupied (the highest occupied molecular orbital) and fully empty (the lowest unoccupied molecular orbital) orbitals. We also expected that the chemical structure of indenofluorene imaged by atomic force microscopy should correspond to the open-shell structure (Figure 1), where the three hexagonal rings should appear similar with equal bond lengths (as each of them is a benzene-like aromatic ring), and the two pentagonal rings should also exhibit equal bond lengths. While we did see that, there was a surprise waiting around the corner.
When measuring several different individual indenofluorene molecules, we also observed molecules with the non-magnetic para state, with roughly the same likelihood as the magnetic open-shell state. For the non-magnetic state, the frontier orbital densities, as mentioned above, corresponded to the densities of the highest occupied and lowest unoccupied molecular orbitals, and strikingly, the atomic force micrograph showed characteristic bond length alternation in line with the non-magnetic para structure, which was notably different from the open-shell structure. We never found the non-magnetic ortho structure in our experiments, indicating that it is less stable than the other two.
Finally, we could also reversibly switch a single indenofluorene molecule between magnetic and non-magnetic states by moving it on the sodium chloride surface, as shown in Figure 3. By performing additional experiments on different metal surfaces than gold — we used silver and copper — and an array of calculations, we learned that bidirectional magnetic switching in indenofluorene arises due to a change in the adsorption site of the molecule on sodium chloride. In effect, a change in the adsorption site can induce sufficient change in the molecular geometry, leading to a corresponding change in the electronic configuration (magnetic or non-magnetic).
In the last 10 years, spectacular progress has been made in the synthetic chemistry of magnetic polycyclic conjugated hydrocarbons, from single molecules that had been predicted for decades,3 to extended nanostructures hosting multiple unpaired electrons.4 But this was the first time that magnetism in such systems could be controlled. The possibility to tune the magnetic state of an organic molecule opens numerous possibilities. A change in the magnetic state of a molecular system should lead to corresponding changes in the frontier orbital and singlet-triplet gaps, influencing electronic transport, luminescence, and photocatalytic activity, making such systems useful as nanoscale switches. Furthermore, the chemical reactivity of a molecule in magnetic and non-magnetic states is expected to be different, which may render such switchable systems useful for sensing applications. The next step (and challenge) for us is to exert a greater control over the switching behavior. Currently, the only way we can achieve switching is by changing the adsorption position of the molecule on the surface. In the future, we would like to explore other means of switching more pertinent from a device perspective, such as local electric fields or strain.