1.4 Composition: congruent and stoichiometric

“LiNbO₃” is a range, not a formula. The crystal is stable from roughly 47 to 50 mol% Li₂O, always on the lithium-poor side of the ideal 50% P-comp. When the melt and the growing crystal have the same composition, the crystal grows uniform from seed to tail. For LiNbO₃ that congruent point is about 48.4–48.5 mol% Li₂O, a lithium-to-niobium ratio of 0.94 P-comp, so in nearly every commercial crystal about one lithium site in twenty holds no lithium derived.

Removing that deficit changes the crystal: Curie temperature up by about 60 °C, UV absorption edge about 17 nm toward shorter wavelengths derived, the field needed to flip the polarization several-fold lower, and the built-in bias field gone (1.3, 1.8). Stoichiometric crystal cannot be pulled from a stoichiometric melt; it needs one of the growth routes in L2.

T / °C    1100      1150      1200      1250      1300
          |----+----|----+----|----+----|----+----|
c 48.45   ========##-------------------------------
Tc CLN            ^
Tc nSLN                        ^
Tm CLN                                   ^

λ / nm    300     305     310     315     320     325
          |---+---|---+---|---+---|---+---|---+---|
α>20/cm   ################################|........
α>15/cm   ###################################|.....

................................
................... ......#.... 
................................
.#................. ............
.............#...... .......... 
............................. ..
.............. .................
... .#.............. ...........
. ................. ............
.... .......................... 
......... ......................
 ...............................

::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::

Ratios

Composition ratios in the literature: R1 = [Li]/[Nb], R2 = [Li]/([Li] + [Nb]) = c/100 derived. At c = 48.45 mol%, R1 = 0.940 derived. Volk & Wöhlecke 2008 prints R1 = 0.969 next to R2 = 0.4845 on p. 5, inconsistent with its own 0.94 on p. 10 misprint.

UV absorption edge vs composition

22 °C, E ∥ c, k ⊥ c, x = c − 48.38 mol% P-comp:

λ20 = 320.4 − 1.829 x − 5.485 x2   nm   (absorption 20 cm−1)
λ15 = 321.9 − 1.597 x − 5.745 x2   nm   (absorption 15 cm−1)

Above 49.5 mol%, 0.1 mol% maps to about 2 nm P-comp.

Growth routes to near-stoichiometric LiNbO₃ (all reported 1992–1997)

Methods. Non-optical: chemical analysis, T_c, density, NMR and EPR linewidths, diffraction. Optical: UV edge, OH-band shape, Raman linewidths, phase-matching temperature; optical methods reach 0.1 mol% absolute and about 0.01 mol% relative accuracy P-comp P(abs). The UV-edge method needs a reflection correction of about 1 nm and fails when an impurity such as Fe absorbs near the edge P-comp. For near-stoichiometric crystal the intensity ratio of the OH components at 3480 and 3465 cm⁻¹ decreases linearly with Li₂O from 49.7 to 50 mol% P-comp.

Why 48.38 vs 48.45 vs 48.6. The congruent point depends on how the melt composition was determined, and on the oxygen content of the Nb₂O₅ feedstock, which may be oxygen-deficient P-comp. The L2 table gives each value with its source.

Defect models in the dial. The Li-site vacancy model compensates each Nb_Li with four Li vacancies; the Nb-site vacancy model compensates five Nb_Li with four Nb vacancies. Both predict the same R1 at every c, so composition alone cannot decide between them P-comp; the evidence is in 1.5. The dial's anchors for E_c and E_int come from 1.8.