Phase transformation pathways of ultrafast-laser-irradiated ${\mathrm{Ln}}_2{\mathrm{O}}_3\left(\mathrm{Ln}=\mathrm{Er}–\mathrm{Lu}\right)$

Ultrafast laser irradiation causes intense electronic excitations in materials, leading to transient high temperatures and pressures. Here, we show that ultrafast laser irradiation drives an irreversible cubic-to-monoclinic phase transformation in ${\mathrm{Ln}}_2{\mathrm{O}}_3\left(\mathrm{Ln}=\mathrm{Er}–\mathrm{Lu}\right)$, and explore the mechanism by which the phase transformation occurs. A combination of grazing incidence x-ray diffraction and transmission electron microscopy are used to determine the magnitude and depth-dependence of the phase transformation, respectively. Although all compositions undergo the same transformation, their transformation mechanisms differ. The transformation is pressure-driven for $\mathrm{Ln}=\mathrm{Tm}–\mathrm{Lu}$, consistent with the material's phase behavior under equilibrium conditions. However, the transformation is thermally driven for $\mathrm{Ln}=\mathrm{Er}$, revealing that the nonequilibrium conditions of ultrafast laser irradiation can lead to novel transformation pathways. Ab initio molecular-dynamics simulations are used to examine the atomic-scale effects of electronic excitation, showing the production of oxygen Frenkel pairs and the migration of interstitial oxygen to tetrahedrally coordinated constitutional vacancy sites, the first step in a defect-driven phase transformation.

Figure 1

High-temperature (top) and high-pressure (bottom) phase diagrams of ${\mathrm{Ln}}_2{\mathrm{O}}_3$ materials. Phases are: A type (trigonal $P\text{−}3m1)$, B type (monoclinic $C2/m)$, C type (face-centered cubic $Ia-3)$, H type (hexagonal $P{6}_3/mmc)$, and X type (body-centered cubic Im-3m). Dashed lines represent approximate phase boundaries. The high-temperature phase diagram is adapted from Ref. [17]. The high-pressure phase diagram is constructed using data from Refs. [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28].

Figure 2

(a) GIXRD patterns of ${\mathrm{Ln}}_2{\mathrm{O}}_3\left(\mathrm{Ln}=\mathrm{Er}–\mathrm{Lu}\right)$ measured at a $3^{\circ }$ incident angle with $\lambda =0.9762\AA$. Experimental data (green circles) are overlaid on computed diffraction patterns from Rietveld refinement (black lines). Red lines represent the difference between the computed and experimental diffraction patterns. Black dots above the ${\mathrm{Er}}_2{\mathrm{O}}_3$ diffraction pattern identify maxima from the B-type phase. The $y$ axis is logarithmic to better display the low-intensity monoclinic signal. (b) Results of phase-fraction analysis from Rietveld refinement showing the percent of monoclinic phase for an x-ray penetration depth of 200 nm.

Figure 3

Cross-sectional TEM of irradiated ${\mathrm{Er}}_2{\mathrm{O}}_3$ and ${\mathrm{Lu}}_2{\mathrm{O}}_3$. High-resolution bright-field images and SAED patterns are shown for the near-surface (0-nm depth) and bulk (200-nm depth for ${\mathrm{Er}}_2{\mathrm{O}}_3$, 230-nm depth for ${\mathrm{Lu}}_2{\mathrm{O}}_3)$. Observed phases are indicated in the SAED patterns.

Figure 4

Results from ab initio molecular dynamics showing the distribution of the number of oxygen Frenkel pairs produced in each simulation. Purple indicates overlap of the ${\mathrm{Gd}}_2{\mathrm{O}}_3$ (blue) and ${\mathrm{Lu}}_2{\mathrm{O}}_3$ (red) datasets. The results presented here define only a qualitative trend in the effect of changing the Ln cation due to approximations made in the simulations.