ANÁLISIS E INTERPRETACIÓN DE RESULTADOS

2.4.1

Stability and formation of cubic

Li

TiO

particles

The lithium titanates synthesized under hydrothermal conditions showed Li/Ti ratios close to 2 : 1, corresponding to 12.65wt% of lithium, as long the samples remained dry after the syntheses. Generally, bathing experiments led to higher Li-loss than the water-flushing experiments, probably due to the longer time span of the interaction between the water and theα-Li₂TiO₃ powders. With a higher water content, the flushing experiments tended towards a higher Li-loss, but the decline in the lithium content was not constant, which is most likely caused by inhomogeneous washing in the filter. In contrast, the samples treated in the bathing experiments were exponentially de-lithiated with a higher water volume (Figure 2.1). When treating with water, α-Li₂TiO₃ seems to have a maximum Li-loss of ca. 75%. In the water-flushing experiments, Li-losses of ca. 25% occurred frequently, which fortifies the assumption that lithium occupancies of 1/4 and 3/4 are more stable than intermediate values. In the master thesis of Linda Irbe (2010) it was shown that lithium can be totally leached fromα-Li₂TiO₃ in low concentrated acid media like citric acid or hyperclodic acid. The same study demonstrated that from monoclinic

β-Li₂TiO₃ only 25 - 30% of lithium could be leached in strong inorganic acid solutions, which has already been observed (Zainullina et al., 2003). To maintain stoichiometry, the Li-loss requires the substitution of other cations to balance the charge. Diminishing unit cell parameters with decreasing lithium content allows the assumption that ions smaller than Li+ ions are inserted into the structure. The Li+ exchange with H+ _{in the system}

Li-Ti-O-H might be similar to the well-known exchange of D+ _{with H}+ _{under atmospheric}

conditions. In a study on Li₄Ti₅O₁₂ it was shown that Li can be leached with HCl and thereafter the presence of hydrogen was determined by neutron diffraction (Simon, 2007). Following the assumption that Li+ions are exchanged by H+_{, products of the solid solution}

Li₂TiO₃-H₂TiO₃ are obtained. For instance the 13% Li-loss of sample F2 can be written as (Li1−0.13H0.13)2TiO3. The hypothetical formulas of all samples are listed in Table 2.1.

Discrepancies of the Li/Ti ratios in previous studies on cubic α-Li₂TiO₃ can be ex- plained by the water-flushing and bathing experiments in the present study. That Li-loss may be caused by washing had already been speculated by Tabuchi et al. (2003), but in another study the very low Li/Ti ratio of 0.34 : 1 of cubic Li₂TiO₃ was ascribed to a mixture of cubic Li₂TiO₃ and amorphous TiO₂ (Tomiha et al., 2002). In the latter study, after using 7nm diameter anatase as a reactant, the resulting cubic Li₂TiO₃ had a similar primary particle size, which presumably eased the de-lithiation.

2.4.2

Li

TiO

-H

TiO

solid solution

The XRPD patterns of unwashed and of water-flushed cubic Li₂TiO₃ showed a unit cell parameter a between 4.141(1) and 4.150(1) ˚A. This is in good agreement with the results of the neutron and the synchrotron study, where at room temperatures, the unit cell parameter of the sample F6 (ca. 40% Li-deficit) is not significantly smaller than the

2.4 Discussion 23

Figure 2.5: Illustration of the structural transformations of α-Li₂TiO₃ upon heating. The high-temperature modification (γ-Li₂TiO₃) can be produced in a metastable form via hy- drothermal syntheses and might be stabilized by protons.

unit cell of the stoichiometric sample 2.4. The low unit cell parameters of samples with high Li-loss (e.g. sample B6) can be ascribed to the intercalation of smaller hydrogen ions in the structure.

Heat treated samples with varying Li-content reacted to different products after an- nealing at 800 ◦C. Only samples showing Li/Ti ratios close to 2 : 1 transformed to pure monoclinic β-Li₂TiO₃ (Figure 2.2b). Samples with Li-loss resulted in the intermediate compound Li₄Ti₅O₁₂, either mixed withβ-Li₂TiO₃ (sample F6, Figure 2.2c), or with rutile (TiO₂) if the lithium content was low (sample B5, Figure 2.2d). The intermediate spinel Li₄Ti₅O₁₂ contains 52% less lithium than Li₂TiO₃, which corresponds to the theoretical calculated formula of (Li0.48H0.52)2TiO3– this value is in between sample F6 and sample B5.

Similarly Li₂TiO₃ and Li₄Ti₅O₁₂ compositions can be produced by annealing a mixture of 5 - 30wt% TiO₂ with monoclinic Li₂TiO₃(Tsuchiya et al., 2003). By the heat treatment of an intermediate Li-Ti-O compound, produced via hydrothermal synthesis, Li₄Ti₅O₁₂ was formed with minor impurities (Jiang et al., 2008) and a lithium titanate of cubic spinel structure was produced after annealing a ternary Li-Ti-O compound, which was previously synthesized by hydrothermal reaction (Fattakhova and Krtil, 2002b,a). These studies, as well as the results of the present work, show that Li₄Ti₅O₁₂can be obtained after annealing ca. 50% de-lithiated α-Li₂TiO₃ and that α-Li₂TiO₃ without Li-loss can not transform to Li₄Ti₅O₁₂ upon heating.

2.4.3

Cubic structure

The cubic structure of α-Li₂TiO₃ obtained by hydrothermal reaction is isostructural with that ofγ-Li₂TiO₃ at high temperatures. The interlinked data points of the cubic unit cell volume at low temperatures and above 1100◦C indicate a constant thermal expansion (Figure 2.4), thus showing that α- and γ-Li₂TiO₃ belong to the same phase. Accordingly, by hydrothermal reactions, disordered cubic Li₂TiO₃ can be produced at temperatures far below its stability field (Figure 2.5).

In general, structures disorder when rising the temperature, but similar to metastable cubic Li₂TiO₃,α-Li₂SnO₃ fully orders at T>1000 ◦C and below it shows disorder related to stacking faults. Two different explanations are conceivable for this behavior. First, the low temperature cubic structure may be stabilized by protons or water entering the structure during the synthesis, in allusion to quenchable cubic Li₂TiO₃-MgO solid solu- tions (Castellanos and West, 1979), i.e. protons could stabilize the cubic structure in the hydrothermal process, like MgO enables quenching of cubic Li₂TiO₃. Second, the cubic structure may be the preferred modification for syntheses at low temperature, due difficul- ties in forming the complex layered monoclinic structure, and the monoclinic Li₂TiO₃ can be formed under hydrothermal conditions (Laumann et al., 2011c), which will be presented in Chapter 3.

The deviations from the linear trend at 400 and 500 ◦C (Figure 2.4) may be explained by a preferred exsolution of the more mobile lithium into the initial monoclinic phase. Test refinements at these temperatures revealed a higher (negative) scattering density at the Li/Ti site. This can only be explained by a higher titanium content, thus supporting this assumption.