Interfaces between organic molecules and transition metals are one of the most mysterious and hotly pursued topics in materials science1,2. Studies of these systems have focused on the use of magnetic transition metals such as iron, cobalt and nickel or their alloys, and have provided intriguing reports of unexpected magnetic behaviour at room temperature3,4, such as induced molecular magnetism at the interfaces. Writing on page 69 of this issue, Al Ma'Mari et al.5 provide a new perspective on this topic by studying non-magnetic metals. Their findings deviate from an 80-year-old theory of ferromagnetism6 developed by the British physicist Edmund Clifton Stoner, which describes how certain materials form permanent magnets.

In the present work, the authors measured the strength of magnetization arising in multilayered planar structures made from thin films of copper or manganese interfaced with buckminsterfullerene molecules (C60; also known as buckyballs). Their data, obtained using magnetometry, indicated that the samples are ferromagnetic at room temperature. To pinpoint where in the layered samples the magnetization arises, Al Ma'Mari et al. used low-energy muon spin rotation7, a sensitive technique that provides a profile of local magnetism as a function of a material's thickness. From an analysis of their samples, the authors confirmed that the magnetism arises near the metal–C60 interface. Their findings may lead to the development of transformative classes of magnetic systems with potential applications in nanoscale sensors and memory devices2.

Understanding the origin of interface magnetism requires an examination of the chemical interactions occurring between the buckyball molecules, which are 'π-conjugated' electronic systems, and the transition-metal surfaces, which have a d-subshell of electrons8. Systems such as these are characterized by strong charge transfer and hybridization of the electron orbitals at the interface, creating new π–d states that drastically modify8,9 the electron energy bands around the Fermi energy of the material (EF); this is the energy corresponding to the highest filled electronic level of a system at absolute zero temperature. In an ideal non-magnetic copper surface, the shape of the function (ρ) that describes how the d-band is populated with electrons at various energies is expected to be smooth and symmetric (Fig. 1) with respect to the energy axis for electron spins aligned either parallel (spin up) or antiparallel (spin down) to an external magnetic field (H). However, in the buckyball–copper interface, which consists of π–d hybrid states, ρ becomes distorted and changes markedly with energy near EF. This gives the interface characteristic properties that can induce magnetism in the molecular layer and copper surface.

Figure 1: Magnetic copper–buckyball interfaces.
figure 1

Karthik V. Raman

The buckyball molecule (C60) and copper are non-magnetic in their isolated form. a, For C60, the function (ρ) that describes how the electron levels are populated at various energies (E) shows discrete values (horizontal lines) for electron spins aligned either parallel (↑, blue bands) or antiparallel (↓, orange bands) to an external magnetic field. In copper, the levels form continuous bands (shaded curves), and the energy of the highest filled band at absolute zero is called the Fermi energy (EF). In both cases, ρ is symmetric about E. b, In materials containing C60–copper interfaces, such as those studied by Al Ma'Mari et al.5, the shape of ρ at the interface is sharply modified and shows a local minimum near EF. When a magnetic field (H) is applied to the interface, the band with spin ↑ becomes populated by electrons at lower energies compared with the spin ↓ band, so that an energy shift occurs between the bands (for simplicity, the fine-level band splitting is not shown). For progressively larger field strengths, this shift increases. Above a critical value Hc, the interface becomes spontaneously magnetized. Al Ma'Mari et al. find that their C60–copper interface has a value for Hc that allows it to become magnetized from near absolute zero up to room temperature.

Full size image

On the basis of this description of the interface's electronic properties, Al Ma'Mari et al. explain their results using Stoner's theory of ferromagnetism. This theory states that, in materials that have a delocalized population of electrons, ferromagnetism arises from an asymmetry between the number of electrons that are found in the spin-up and spin-down states; the asymmetry is manifested as a shift of ρ to lower energies for the spin-up band. This asymmetry often comes at a cost of extra energy, which can be compensated for by an energy exchange between the electrons that favours the alignment of their spins in a spin-up orientation.

Using this formalism, Stoner showed6 that, for a metal to exhibit ferromagnetism, the product (U) of the parameter corresponding to this energy exchange and the value of ρ at EF should be greater than 1 — a rule known as the Stoner criterion. This is why iron, cobalt and nickel are ferromagnetic. However, in systems that do not satisfy this criterion10, similar analyses suggest that if ρ shows a local minimum near EF, the material can undergo a metamagnetic transition to a state of increased magnetization when a sufficiently high magnetic field (typically, several teslas) is applied. For field strengths above a critical value (Hc), the system becomes magnetic. The value of Hc depends sensitively on the value of ρ and its contour (slope and curvature) near EF (ref. 10).

The authors performed computer simulations, assuming zero temperature, to model the shape of ρ at the copper–C60 interface. Their calculations suggest an enhancement in the value of U at the interface, compared with that in the sample's bulk, by a factor of 4. Although Stoner's criterion is not met at the interface, Al Ma'Mari et al. find that, because of the sharp profile of ρ around EF, the value of Hc for a metamagnetic phase transition of the sample is reduced to about 0.1 T — corresponding to a magnetic energy of about 12 microelectronvolts (from a conversion of the magnetic field strength to Zeeman energy). Expressed as a temperature, this corresponds to about 140 millikelvins, which is considerably lower than the system's thermal energy in the temperature range of the authors' measurements. Therefore, the small Hc values returned by the models indicate that the sample is driven into a ferromagnetic state from near absolute zero up to room temperature.

The remarkable observation of magnetism in copper, which in its bulk form has one of the smallest values of U among transition metals11, leaves room for many challenging investigations. This includes the possibility of observing magnetism in other non-magnetic transition-metal surfaces that are closer to satisfying the Stoner criterion and of confirming the existence of metamagnetic transitions in interface systems that have larger Hc values.

Al Ma'Mari and colleagues' work raises questions about the origin of strong, direction-dependent magnetic properties at the metal–C60 interface that can cause the polarity of magnetization to switch for non-zero magnetic fields; such effects are being extensively researched2,3,4. Overall, the ability to tailor the electronic states at interfaces between different materials holds the key to understanding the emergence of ferromagnetism in systems such as those studied in the present work. Further research in this exciting field could lead to the development of a generation of nano-electronic devices that incorporate a wide range of hybrid materials showing unconventional magnetism. Footnote 1