Inferring the Dy-N axis orientation in adsorbed ${\mathrm{DySc}}_2{\mathrm{N}@\mathrm{C}}_{80}$ endofullerenes by linearly polarized x-ray absorption spectroscopy

Endofullerene ${\mathrm{DySc}}_2{\mathrm{N}@\mathrm{C}}_{80}$ is a single-molecule magnet with a large magnetic anisotropy and high blocking temperature, which is promising for nanomagnetic applications. As the easy axis of magnetization coincides with the Dy-N bond direction, it is important to understand the structure of the ${\mathrm{DySc}}_2\mathrm{N}$ unit in the fullerene cage and to control the orientation of the molecules. Here we report on the experimental determination of Dy-N axis by x-ray absorption spectroscopy (XAS) with linear polarized light at the $\mathrm{Dy}\text{−}{\mathrm{M}}_{4,5}$ white lines. ${\mathrm{DySc}}_2{\mathrm{N}@\mathrm{C}}_{80}$ molecules were adsorbed on a Pt(111) surface and XAS was performed as a function of temperature in the range between 35 and 300 K. The ${\mathrm{M}}_5/{\mathrm{M}}_4$ branching ratio shows a clear and reversible variation with temperature which can be explained, on the basis of a thermodynamic model, by a change of average orientation of the molecules with temperature. The XAS spectra are well reproduced by ligand field multiplet calculations. It is shown that the angle between the magnetization (Dy-N) axis and the surface plane can be directly inferred from the XAS spectra with in-plane polarization by comparison with calculated spectra. It is found that the endohedral unit is randomly oriented at room temperature but tends towards orientation parallel to the surface at low temperature, indicating a weak but non-negligible interaction between the endohedral units and the metal surface.

Figure 1

(a) Scheme of the XAS experiment. The sample is ${\mathrm{DySc}}_2{\mathrm{N}@\mathrm{C}}_{80}$ molecules adsorbed on a Pt(111) surface. The p-polarized x rays impinge at an angle of $9.5^{\circ }$ with respect to the surface normal (n) which points along z in the laboratory frame. (b) Ball and stick model of ${\mathrm{DySc}}_2{\mathrm{N}@\mathrm{C}}_{80}$ (orange: Dy, purple: Sc, blue: N, brown: C). The Dy-N axis points along Z in the molecular frame.

Figure 2

(a) XAS spectrum on the Dy-${\mathrm{M}}_{4,5}$ edge at room temperature. The dashed line is the background baseline. (b) Spectral weight of the Dy-${\mathrm{M}}_5$ edge $\left(\frac{\int I_{M_5}dE}{\int I_{M_4+M_5}dE}\right)$. The solid line is a fit of a thermodynamic model [Eq. (2)]. The dashed line is the expectation from a theory with a $J_z=\pm 15/2\to \pm 13/2$ splitting of 47 meV (obtained from Fig. S7 in the Supplemental Materials [19]).

Figure 3

Comparison between experiment and theory. (a) Average of all experimental spectra after background removal. (b) Calculated XAS spectra on the ${\mathrm{M}}_{4,5}$ edges of a Dy atom for linear polarized x rays along X (blue), Y (green), and Z (yellow). The black line is the sum of ${\mathrm{I}}_X$, ${\mathrm{I}}_Y$, and ${\mathrm{I}}_Z$. The inset depicts the ${\mathrm{DySc}}_2\mathrm{N}$ endohedral unit and the molecular reference frame. The close resemblance of the black line and the experiment indicates random orientation of the Dy-N axes.

Figure 4

Experimental XAS spectral groups (red dots) fitted to ${\mathrm{I}}_Z$ and ${\mathrm{I}}_X$ (black lines). Top panels: 300 K and 35 K. Bottom panel: difference between the fits at room temperature and low temperature.

Figure 5

Temperature dependence of the second Euler angle $\beta$ of the Dy-N axes distribution. $\beta$ values as a function of temperature for groups of 5 subsequent spectra. ($▾$ cooling, $▴$ heating). The circles indicate the spectra shown in Fig. 4. The black solid line with the 90% prediction intervals (light gray) represents the model in Fig. 2, now fitted to the second Euler angles $\beta$.