Zakaria Kharbouch, Advancing All-Solid-State Batteries: Design and Ionic Conductivity Analysis of Novel LLZO-Based Solid-State Electrolytes, PhD thesis, Mohammed VI Polytechnic University (UM6P, LIMSET), supervised by Abdessamad Faik and Mohamed Chakir, defended on 3 November 2026. Read here from the manuscript of 17 October 2026 (V4.0, 163 pp.); page numbers are its printed ones.
Garnet \(\mathrm{Li_7La_3Zr_2O_{12}}\) (LLZO) conducts lithium fast only in its cubic form. Gallium on a lithium site stabilises that form at room temperature, opens lithium vacancies (each Ga\(^{3}\)⁺ replaces three Li⁺) and lowers the symmetry from \(Ia\bar{3}d\) to the acentric \(I\bar{4}3d\), which splits the tetrahedral lithium site in two (12a, 12b) beside the octahedral 48e and multiplies the conduction paths (§I.2.3, pp. 18–20). The thesis asks whether a third substituent, on the lanthanum site, can take Ga-stabilised LLZO further, a site it judges under-explored.
Two families on a fixed Ga backbone (0.2 Ga per formula unit), each with a fixed Zr-site co-dopant:
A modified solid-state route (M-SSR), developed in Chapter II after the sol–gel and the conventional solid-state routes gave \(\mathrm{La_{2}Zr_{2}O_{7}}\), lithium loss and inhomogeneous Ga (pp. 46–52): precursors weighed with a 15 % excess of \(\mathrm{Li_{2}CO_{3}}\); wet homogenisation in acetone–water at about 70 °C; drying at 120 °C; intermediate calcination at 450 °C; regrinding, pelletising and calcination at 900 °C. Pellets of 11 mm, 8 t, sintered at 1200 °C for 6 h (5 °C min\(^{-1}\), after 120 °C and 700 °C holds), buried in mother powder of the same composition in covered alumina boats (p. 55).
| Technique | On what | Conditions (§II.4.6, p. 56) | What it gave |
|---|---|---|---|
| X-ray powder diffraction, Rietveld | calcined powders (900 °C) | Panalytical Empyrean, Cu Kα (1.5406 Å), 45 kV, 40 mA, 2\(\theta\) 10–100°, step 0.01°, room temperature | lattice parameters; phase fractions and agreement factors (Ga–M–Ta); software, refined parameters and zero-shift treatment not stated |
| SEM–EDS | fractured sintered pellets | Thermo Fisher Scios 2 | grain or pore size; element maps at the micrometre scale |
| Archimedes density | sintered pellets | ethanol | relative density (theoretical density used not stated) |
| Impedance spectroscopy | sintered pellets, Ag paste | 200 kHz – 0.1 Hz; 20–80 °C (Ga–M–Ta), room temperature (Ga–M–Sc) | total conductivity (bulk and grain boundary unresolved except for the Ga–Y–Sc series); activation energy (Ga–M–Ta) |
Not measured: the lithium content (or any composition after synthesis: no ICP), the aluminium content, hydrogen or carbonate, the phase content of the sintered pellets, any site occupancy, any atomic position.
Ga–M–Ta. The reference is single cubic, a = 12.977 Å. Ba stays in the cubic phase to x = 0.10 with an almost constant cell; \(\mathrm{BaZrO_{3}}\) appears from x = 0.15. Rb expands the cell steadily to 13.001 Å, with 10–21 wt% tetragonal garnet in most compositions. Y expands it slightly (to 12.994 Å), with tetragonal garnet and \(\mathrm{Li_{2}ZrO_{3}}\) (up to 14.2 wt%); the expansion is read as part of the Y on the Zr site (pp. 62–63). Ga–Y–Ta at x = 0.20 is the best electrolyte of the thesis, 1.04 mS \(\mathrm{cm}^{-1}\), 95.3 % dense, the lowest Eₐ (0.32 eV). Ba peaks at 0.24 mS \(\mathrm{cm}^{-1}\) (x = 0.10); Rb stays below the reference.
Ga–M–Sc. The reference contains tetragonal garnet according to the text (p. 88), conducts 0.451 mS \(\mathrm{cm}^{-1}\) and is 91.2 % dense. Rb gives the best of the family at x = 0.03 (0.903 mS \(\mathrm{cm}^{-1}\), 96.2 %), then declines monotonically; \(\mathrm{Sc_{2}O_{3}}\) segregates. Ba gives \(\mathrm{BaZrO_{3}}\) from x = 0.10. The Y series densifies poorly (73–76 %), with \(\mathrm{Y_{2}O_{3}}\) from x = 0.15 and the lowest conductivities; here Y is read as on the La site (p. 89).
The thesis's reading. No single parameter predicts the conductivity; density is an enabling condition; once the pellet is dense, the cubic fraction and the dopant-induced change of the lithium sublattice decide, through the lithium vacancy concentration (pp. 77–78, 123–124). Chapter V adds a machine-learning model of garnet conductivity trained on a curated literature dataset (R\(^{2}\) = 0.85).
"Future studies combining neutron diffraction, synchrotron X-ray diffraction, solid-state NMR spectroscopy, molecular dynamics simulations, and first-principles calculations could provide direct insights into lithium occupancy, vacancy distribution, dopant site preferences, and migration pathways" (Perspectives, p. 125); and, for the machine-learning framework, "structural descriptors obtained from Rietveld refinement" (p. 126). This project is that programme, made concrete.
Next: the crystallography in detail, and the fact base.