Home -> Research Info -> Dynamic energy landscapes for mobile ions in disordered solids

Dynamic energy landscapes for mobile ions in disordered solids

To gain a fundamental understanding of ionic motion in disordered solids, it is indispensable to understand the dependence of mobility on the structural and energetic environment of the mobile ions. In this project it was demonstrated how an augmented bond valence (BV) approach permits to assess the ion transport properties in glassy and crystalline solid electrolytes. The resulting structure property function relationships provide valuable guidelines for a systematic development of new classes of solid electrolytes for high performance batteries, fuel cells or solar cells. In the case of complex disordered phases it is necessary to start the analysis by producing representative local structure models, here either by Reverse Monte Carlo fitting to synchrotron X-ray and neutron diffraction data or by Molecular Dynamics simulations. The latter provides the opportunity to compare the bond valence model with a complete trajectory analysis, and moreover can – as demonstrated in the project - give rise to dynamic pathway models. We could verify that the BV method produces reliable models of the pathways suitable for a statistical analysis. This was achieved both by empirical comparison with the outcome of independent pathway analyses of MD simulation trajectories (in model cases, where the latter is possible) and by a stringent derivation on the causal link between the site energy of a mobile ion and its bond valence mismatch. Moreover we could apply the approach to various practical cases, demonstrating how it contributes to the understanding of ion transport mechanisms in disordered solids and can guide the search for new solid electrolytes that are needed to improve safety and performance of mobile energy storage devices. Our project particularly promoted the understanding of the mixed mobile ion effect in glassy solid electrolytes. In the frame of a collaboration with the group of Jan Swenson at Chalmers University of Technology we established an empirical correlation that permits to predict both the dc ionic conductivity as well as its activation energy directly from structure models and to identify transport pathways for the mobile ions in a wide range of ion conducting glasses. This led to a widely acknowledged progress in the understanding of ion transport mechanism of glassy and other strongly disordered solids. Based on both new synchrotron and neutron diffraction data collected in the project and previous experiments, we could expand the range of solid electrolytes for which the prediction applies. For metaphosphate glasses with several types of mobile ions we could identify the origin of the strongly non-linear composition dependence of the conductivity (often termed “mixed alkali effect”): Mobile ions with sufficiently similar BV parameters do not block each other leading to a highly correlated ion transport mechanism, while the ability for a correlated motion decreases with the differences in the bond valence parameters. Based on this approach we can nearly quantitatively predict the composition dependence of the ionic conductivity in a wide range of disordered solids directly from the structure model. The project also involved the development of new tools to determine the time evolution of bond valence pathways by applying the analysis to series of snapshots from the MD trajectories. Together with our finding that the method also works well for mobile anions, this proved helpful to identify new solid electrolytes with unique transport mechanisms and to explain the enhanced ionic conductivities in various nanostructured ionic conductors.

Figure 1: Top: Slice though of the bond valence model of cation migration pathways in Li_xRb_1-xPO₃ (left) and Ag_xNa_1-xPO₃ (right) demonstrating that the pathways for Li and Rb are distinct, while Ag and Na transport pathways are nearly identical (x= 0.5 in both graphs). Bottom predicted and experimental variation of the conductivity with the Li or Ag content x, highlighting the different extent of the mixed mobile ion effect and the nearly quantitative prediction (blue) of the experimental conductivity data (red) by our model.

10 Li2O – 90 SiO2	20 Li2O – 80 SiO2
30 Li2O – 70 SiO2	50 Li2O – 50 SiO2

Fig. 2 L.h.s. Comparison of Li transport pathways modelled as slices through iso­surfaces (yellow) of constant Li BV energy for snapshots from MD simulations of 4 different lithium silicate glasses Dark blue spheres = Li positions in the underlying glass structure model. R.h.s.: Correlation between the variation of the activation energy and the logarithm of the conductivity s  in xLi₂O-(1-x) SiO₂ glasses and the volume fraction of regions with a Li BV site energy below a fixed threshold in a snapshot of the equilibrated MD simulation. Broken lines are linear regressions.

Contact Person : Assoc Prof S. Adams

E-mail: mseasn@nus.edu.sg

Tel : 6874 6869
Fax: 6776 3604