Vibrational properties of rare-earth nitrides: Raman spectra and theory

Raman spectra are presented for the rare-earth nitrides SmN, GdN, DyN, ErN, and LuN measured with 633 and 514 nm excitation wavelengths and at temperatures above and below the Curie temperature. Frozen-phonon calculations are presented for the phonons at $\Gamma$, $L$, and $X$ points in the same series of materials plus previously studied YbN using the full-potential linearized muffin-tin orbital method and the $\text{LSDA}+U$ (local spin-density $\text{approximation}+\text{Hubbard}$ $U$ corrections). The method is found to be in good agreement with recent linear-response pseudopotential calculations for the closely related ScN. Comparison with ScN and Eu chalcogenides (EuS, EuO) allows us to conclude that the main spectral line seen in the RE-N is a disorder induced phonon density of states like spectrum heavily weighted by the $\text{LO}\left(L\right)$ modes. Its second harmonic $2\text{LO}\left(L\right)$ is also observed. The increasing frequency trend with atomic number is related to the decreasing trend in lattice constant and hence increase in force constant rather than the mass of the rare-earth ion because this phonon mode is purely a N-vibrational mode. The disorder does not arise from the spin orientations because no changes are observed upon magnetic ordering but may arise from point defects and the size of the crystallites.

Figure 1

(Color online) Raman spectra of GdN with 633 nm excitation (lower red) and 514 nm excitation (upper black).

Figure 2

Raman spectra of SmN, GdN, DyN, ErN, and LuN on an expanded scale to highlight the hardening across the RE series. The spectra for DyN and LuN show gaps where subtraction of the strong Si line was incomplete.

Figure 3

(Color online) Lattice constants of rare-earth nitrides versus atomic number. The calculated results are from Larson et al. (Ref. 2).

Figure 4

(Color online) Experimental (main peak) and calculated $\text{LO}\left(L\right)$ phonon frequencies versus atomic number. The lines are just linear interpolations as guide for the eyes. The experimental data include an error bar.