in situ observation of reversible phase transitions in Gd-doped ceria driven by electron beam irradiation

Abstract

Ceria-based oxides are widely utilized in diverse energy-related applications, with attractive functionalities arising from a defective structure due to the formation of mobile oxygen vacancies ( $V_O^{\cdot \cdot }$ ). Notwithstanding its significance, behaviors of the defective structure and $V_O^{\cdot \cdot }$ in response to external stimuli remain incompletely explored. Taking the Gd-doped ceria (Ce_0.88Gd_0.12O_2-δ) as a model system and leveraging state-of-the-art transmission electron microscopy techniques, reversible phase transitions associated with massive $V_O^{\cdot \cdot }$ rearrangement are stimulated and visualized in situ with sub-Å resolution. Electron dose rate is identified as a pivotal factor in modulating the phase transition, and both the $V_O^{\cdot \cdot }$ concentration and the orientation of the newly formed phase can be altered via electron beam. Our results provide indispensable insights for understanding and refining the microscopic pathways of phase transition as well as defect engineering, and could be applied to other similar functional oxides.

Fig. 1. The reversible transition between F- and C-type.

a Schematic of the reversible transition: in TEM mode, e-beam with high and low EDR is used to stimulate the F-to-C and C-to-F transition, respectively. F- and C-type CGO30 models are viewed along < 001 > at the bottom of (a). Single unit cells of each type are outlined by the solid squares. b–c Enlarged models from the dashed rectangles in a. The three sets of parallel lines in (c) indicate split metal positions. Below the models are the simulated HRTEM images. Laterally averaged intensity profile and the estimated $d_{O\_v}$ (virtually extended as the dotted lines) are plotted on the left and right side of the simulated images.

Fig. 2. Experimental visualization of both F- and C-type.

a iDPC image of the F-type along < 001 > , together with a F-type model and the simulated iDPC image. On the left is the laterally averaged intensity profile. b–c The mapped distances between neighboring M and O positions ($d_M$ and $d_O$) based on (a). On the right side of (c) are the laterally averaged $d_{M\_v}$ and $d_{O\_v}$. d–f The corresponding HRTEM results from the C-type. g Diffraction patterns from a transition cycle. Circles are placed at the same position for all the patterns. h–i FFT patterns from (a and d).

Fig. 3. Transitions with different EDRs.

a, b One F-to-C transition with EDR 0.69 (relative value, where the highest EDR 3825 e · Å^-2 · s^-1 listed in Fig. 3 is noted as EDR 1.0). A series of 40 TEM images were recorded with 0.04 s exposure time and 0.8 s interval. The image at the outset and the FFT patterns (absolute value) during the transition are shown. c–d One C-to-F transition with EDR 0.06. A series of 40 TEM images were recorded with 0.04 s exposure time and 2 s interval. The image at the outset and the FFT patterns (absolute value) during the transition are shown. e–g $r$ (intensity ratio between the four {010} and the four {020} spots), $r_A$, and $r_B$ (intensity ratio between the two {010} and the two {020} spots along A and B) as a function of time from each transition. Linear fitting is applied to $r$, and the estimated slopes are listed in (h).

Fig. 4. Pushing the VO⋅⋅ ordering.

A series of 50 HRTEM images was recorded with 0.5 s interval. Two of them are shown in (a). b The determined $d_{O\_v}$ and $d_{M\_v}$ from each labeled layer, as defined in (a). c $A_{O\_v}$ and $A_{M\_v}$ as a function of time, based on the HRTEM image time series. d $A_{O\_v}$ and $A_{M\_v}$ estimated from the simulated HRTEM images. The image simulations are based on C-type models with varying $\delta$ from other studies^,,,. Solid lines suggest linear fits.

Fig. 5. Rotating the C-type structure.

a, b HRTEM images of the C-type structure before and after e-beam irradiation. c, d The same as in (a, b) with the brighter metal layers outlined by yellow/vertical and blue/parallel lines. e, f The enlarged images from the squared regions in (c, d). At the top is the vertically averaged intensity profiles, and atomic layers are labeled at the bottom. g–h The mapped distances between neighboring O positions, based on (e, f). Two regions (#1 and #2) are defined by the lines in magenta. From region #1, vertically averaged $d_{O\_l}$ are plotted at the top. From region #2, vertically averaged $d_{O\_l}$ and laterally averaged $d_{O\_v}$ are plotted at the bottom and on the right, respectively.

Fig. 6. Chemical structure of CGO.

a iDPC image of the CGO along <110> direction. A F-type model and the simulated iDPC image are overlaid. b EELS SI results: simultaneously acquired ADF image and elemental maps plotting the intensity from the Ce M_4,5, Gd M_4,5 and O K edge. At the lower-left corner is the vertically averaged t/λ profile, and on the right is the laterally averaged intensity profiles from the O, Gd and Ce map. c Fine structure of the Ce M_4,5 edge from three different regions (#1 to #3).

Fig. 7. Proposed F-to-C transition.

The C- and F-type are separated by the colored backgrounds, and one unit-cell of the F-type is outlined at the top-left corner. a Early stage of the transition: two small regions ($A_a^{\parallel }$ and $A_a^{\perp }$) take the C-type structure (C-CGO40, 40% of the metal sites are occupied by M³⁺) with a relative rotation of 90°. b Two more regions transfer to the C-type: $B_b^{\parallel }$ (C-CGO50) and $B_b^{\perp }$ (C-CGO40). Meanwhile, the A regions evolve to $A_b^{\parallel }$ (C-CGO80) and $A_b^{\perp }$ (C-CGO60). c Further irradiation raises the $V_O^{\bullet \bullet }$ concentration in the $B$ and $A^{\perp }$ region, where $B_c^{\parallel }$, $B_c^{\perp }$ and $A_c^{\perp }$ become C-CGO60, C-CGO50, and C-CGO80. Besides, the $A_b^{\parallel }$ region experiences a 90° rotation and are combined with $A^{\perp }$. Composition of the F-type substrate changes accordingly, as indicated at the lower-left corners. d–f Simulated HRTEM images corresponding to the regions defined by the rectangles in (a–c).