EXAFS evidence for a primary Zn${}_{\mathrm{Li}}$ dopant in LiNbO${}_3$

We present extended x-ray-absorption fine structure (EXAFS) data as a function of temperature (10–300 K) at the Zn and Nb $K$ edges for Zn-doped LiNbO${}_3$. The focus is on higher Zn concentrations (7–11 mol $\%$) for which there is disagreement in the literature as to the substitution site for Zn. Our data show that Zn substitutes only on the Li site; we find no evidence for Zn on the Nb site. However, uncertainties result in an upper bound of at most 5$\%$ of the Zn dopants being Zn${}_{\mathrm{Nb}}$. In addition, the Zn${}_{\mathrm{Li}}$ defect produces a significant distortion in the lattice out to at least 4 Å; the O atoms are attracted toward Zn while the Nb neighbors are repulsed. The Nb EXAFS agree well with the structure from diffraction for the main Nb-X peaks out to about 3.7 Å. However, there appears to be a weak third Nb-O peak in the first O shell, which has a low amplitude and a longer bond length. For Nb, the shortest Nb-O bond is extremely stiff (correlated Debye temperature, ${\theta }_{cD}$ $\sim$ 1100 K), while the longer Nb-O bond is a little weaker (${\theta }_{cD}$ $\sim$ 725 K) and of comparable strength to the shortest Zn-O bond (${\theta }_{cD}$ $\sim$ 600 K); consequently, substituting Zn at the Li site will stiffen the structure as the Li-O bonds are weak. We discuss implications of a dominant Zn${}_{\mathrm{Li}}$ defect.

Figure 1

Top: The environment about the central Li site (large, red) in LiNbO${}_3$. Bonds are shown for the nearest O atoms; other O shells are shown as small (blue) atoms. The four medium (green) atoms (three in a plane) above the central atom are Nb at $\sim$3.07 Å; a second plane of Nb atoms is below the center at 3.36 Å, while the bottom Nb atom is 3.88 Å below center. There are six Li (red) neighbors (three above and three below center) at 3.77 Å. Bottom: the hexagonal unit cell for LiNbO${}_3$. The cluster above covers the top half of this cell but is expanded in the horizontal directions.

Figure 2

The environment about the central Nb site (green) in LiNbO${}_3$. Bonds are shown for the nearest O atoms; again other O shells are shown as small (blue) atoms. The four red atoms (three in a plane) below the central atom are Li at $\sim$3.07 Å; a second plane of Li atoms is above the center at 3.36 Å. There are six Nb (green) neighbors, three above and three below center, at 3.77 Å. The central Nb atom is at the center of the unit cell shown in Fig. 1.

Figure 3

Theoretical EXAFS functions calculated using feff8 for one neighbor of Li, Zn, or Nb, at a distance 3.07 Å from a central Zn atom. The Zn-Li peak is only 8$\%$ of the Zn-Nb peak; the Zn-Zn peak is also large, about 90$\%$ of the Zn-Nb peak. Note that the Zn-Li peak has a larger shift to lower $r$ than the higher atomic number neighbors.

Figure 4

A simulation of the EXAFS $r$-space data [Fast Fourier Transform (FFT)[$k\chi \left(k\right)$]] for Zn on a Li site assuming the known structure of LiNbO${}_3$ and a typical overall broadening (${\sigma }^2$ = 0.0036 Å${}^2$). The FFT window was 3.5–13.7 Å${}^{-1}$, Gaussian broadened by 0.3 Å${}^{-1}$. Here and in other $r$-space plots, the fast oscillation is the real part $R$ of the FFT while the envelope is $\pm \sqrt{R^2+I^2}$ where $I$ is the imaginary part of the FFT. The vertical solid (green) line near 2.8 Å shows the location of the strong first Zn-Nb peak (actual distance about 3.1 Å).

Figure 5

A simulation of the Zn EXAFS $r$-space data for Zn on a Nb site assuming the known structure of LiNbO${}_3$ and a typical overall broadening (${\sigma }^2$ = 0.0036 Å${}^2$). The vertical solid (blue) line shows the location of the first Zn-Nb peak (actual distance above 3.77 Å). Note that the amplitude is low from 2.5 to 3 Å where the strongest Zn-Li peak occurs. This is a result of the very low backscattering from Li, which has only three electrons (or two for a +1 ion).

Figure 6

An example of the $k$-space data $k\chi \left(k\right)$ for the 11.1-mol $\%$ Zn sample at 6 K. This is an average of two scans. The signal-to-noise ratio is good out to $\sim$15 Å${}^{-1}$.

Figure 7

The $r$-space data (FFT[$k\chi \left(k\right)$]) for the 7.3-, 9.6-, and 11.1-mol $\%$ Zn samples as a function of temperature for the Zn $K$ edge. The FFT window is 4–14.2 Å${}^{-1}$, Gaussian broadened by 0.3 Å${}^{-1}$. The vertical line (green) at 2.8 Å shows where the strongest Zn-Nb peak would occur for a Li substitution site while the vertical line (blue) near 3.8 Å indicates where the Zn-Nb peak would occur for a Nb site substitution. Note that the spectra for all concentrations are nearly identical indicating the same local structure, independent of concentration.

Figure 8

A comparison of the Zn $K$-edge $r$-space data at 10 K for the 7.3- and 11.1-mol $\%$ samples. The amplitudes of most of the peaks are slightly smaller for the 11.1-mol $\%$ sample indicating slightly more static disorder. Same FFT range as for Fig. 7.

Figure 9

An example of a fit for the 11.1-mol $\%$ sample; the FFT window is 4–14.2 Å${}^{-1}$, Gaussian broadened by 0.3 Å${}^{-1}$, and the fit range in $r$ was 1.2–4.2 Å.

Figure 10

Plot of the pair distances for the first two O neighbors and the first three Nb shells about Zn as a function of $T$ for 11.1 mol $\%$ Zn; the pair distances at low $T$ for the Li site in LiNbO${}_3$ are shown as dotted lines. The largest changes are for the first Zn-O and Zn-Nb peaks; at high $r$, the distances are very close to the values about a Li site in congruent LiNbO${}_3$.

Figure 11

Plots of ${\sigma }^2$ (points) as a function of $T$ for the first three shells—two Zn-O shells and the Zn-Nb shell; (a) 7.3 mol $\%$, (b) 11.1 mol $\%$. The results are nearly the same for the first Zn-O and the Zn-Nb peaks, only a small increase in static disorder as the Zn concentration increases. The second Zn-O peak has a much stronger $T$ dependence indicating a much weaker spring constant; this peak has a larger increase in the static disorder with increasing Zn concentration. These data were fit to a correlated Debye model, shown by solid or dotted lines.

Figure 12

Comparison of the Nb $K$-edge data for LiNbO${}_3$ samples doped with 7.3 and 11.1 mol $\%$ Zn, together with the corresponding Nb data for another divalent dopant Mg${}^{+2}$ and In${}^{+3}$. The four traces are nearly identical, with the data for the Mg-doped sample slightly higher near 3.3 Å. The FFT range is 4.5–15 Å${}^{-1}$, Gaussian broadened by 0.3 Å${}^{-1}$.

Figure 13

(a) The temperature dependence of the Nb $K$-edge $r$-space data for the 0.6-mol $\%$ In-doped sample. Note that there is little change of the first Nb-O peak near 1.5 Å, while the second Nb-O and Nb-Nb peaks show significant changes. The weak structure between 2.3 and 3 Å is a sum of weak Nb-Li and Nb-O peaks plus multiscattering peaks. (b) An example of a fit of the 10-K data for this sample from 1.2 to 3.7 Å; note the shorter $r$ range for plotting this fit. Same FFT range as in Fig. 12.

Figure 14

The pair distances around Nb from Nb $K$-edge data, for the 0.6-mol $\%$ In sample. Except for the possible interstitial peak, the pair distances agree well with diffraction at low $T$ (dotted lines) to better than 0.01 Å for all pairs, except for the weak Nb-Li peaks (0.02 Å).

Figure 15

An expanded view of the Nb-O region showing the data for the 11.1-mol$\%$ Zn sample (solid black), the fit using three Nb-O peaks (dashed blue), and the fit with two peaks (dotted red)—note the dip in amplitude near 2 Å for the latter. The individual plots for Nb-O1 and Nb-O2 (with three Nb-O peaks) is plotted in the middle; the extra Nb-O3 peak plus the sum of the Nb-O1 and Nb-O2 peaks are plotted at the bottom. The first two peaks (O1 and O2) are at the expected distances for LNO; the extra peak is centered about 2.05 Å in the EXAFS plot—in the $r$ range over which the real parts of the FFT ($R$) for peaks O1 and O2 are out of phase and nearly cancel as shown in second panel. Also note the phase of the real part $R$ over this range for the data and the two fits at top.

Figure 16

Plot of ${\sigma }^2\left(T\right)$ for the three largest peaks, Nb-O1, Nb-O2, Nb-Nb1 (Nb $K$-edge data), for the 0.6-mol $\%$ In sample. All temperature dependencies are weak indicating strong bonding, with ${\sigma }^2$ for the shortest Nb-O peak having the least $T$ dependence.

Figure 17

The environment about the O atoms showing the Nb-O-Li angles and Nb-O and Li-O bonds. Nb1-O-Nb2, 139.9${}^{\circ }$; Li1-O-Li2, 121.3${}^{\circ }$; Li1-O-Nb1, 93.8${}^{\circ }$; Li1-O-Nb2, 117.3${}^{\circ }$; Li2-O-Nb1, 88.1${}^{\circ }$; Li2-O-Nb2, 95.4${}^{\circ }$. Note that the Li-O bonds (and Zn-O bonds around Zn) are roughly perpendicular to the Nb-O bonds.