When process engineers ask about Li/Mg selectivity in intercalation DLE, the conversation usually starts with material selection: which electrode chemistry is best? That question has a materials science answer, but it also has a process engineering answer that applies regardless of which material you chose. Understanding the lattice chemistry helps you apply the process engineering answer correctly.
The Intercalation Mechanism and Why It Is Selective
Lambda-MnO2 is the most widely deployed intercalation electrode material in commercial DLE applications. In its delithiated form, it has a spinel crystal structure with a three-dimensional network of tetrahedral and octahedral sites that was created by electrochemical removal of Li+ from LiMn2O4 (which serves as the starting precursor). The selectivity of this structure for Li+ over Mg2+ arises from two factors: ionic radius and charge density.
Li+ has an ionic radius of 0.76 angstroms. The vacated sites in the lambda-MnO2 spinel lattice were sized by the Li+ that occupied them in the LiMn2O4 precursor. Mg2+ has an ionic radius of 0.72 angstroms, which is close enough to permit insertion geometrically, but the charge difference matters. Mg2+ has twice the charge of Li+, meaning it induces a stronger local lattice distortion upon insertion than Li+. The lattice energy cost of accommodating Mg2+ at a Li+ site is higher than for Li+ itself, which is the thermodynamic basis of selectivity.
The selectivity is real but not absolute. At the Gibbs energy level, Li+ intercalation is thermodynamically preferred over Mg2+ intercalation at a lambda-MnO2 electrode in a mixed Li/Mg brine at any reasonable operating potential. But thermodynamic preference and kinetic outcome under process conditions are not the same thing. When the electrode is driven at higher current density (faster adsorption), the system is not at equilibrium. Kinetic factors, particularly the different diffusion coefficients of Li+ and Mg2+ in the brine and in the lattice solid-state, affect the ratio of Li+ to Mg2+ that actually gets incorporated into the electrode over a finite adsorption phase.
How the Lattice Chemistry Constrains the Potential Window
The electrode potential window during adsorption is bounded on both sides for thermodynamic and kinetic reasons. On the positive (less reducing) side, if the applied potential is too high (too little reducing driving force), the rate of Li+ intercalation slows because the driving force for Mn4+ reduction (which accompanies Li+ intercalation) is insufficient. Recovery per unit time drops below acceptable economics. On the negative (more reducing) side, if the applied potential is too low, Mg2+ intercalation becomes more favorable because the additional reducing driving force overcomes more of the thermodynamic selectivity advantage of Li+, and the Mn3+ state of the reduced lattice is also more susceptible to Mn dissolution, which is the primary degradation mechanism.
For a fresh lambda-MnO2 electrode against a representative salar brine at 25 degrees Celsius, the practical operating window typically spans approximately 150 to 250 mV versus SHE. This window represents the range where Li+ intercalation rate is acceptable and Mg2+ co-intercalation is limited to a tolerable fraction. Where exactly within that window you operate determines the selectivity/throughput balance.
The window narrows over the electrode lifetime. As the electrode undergoes capacity fade through Mn dissolution and re-deposition cycling, the lattice develops structural heterogeneity: some sites have been subjected to more cycles and have slightly distorted local geometries that lower the local selectivity coefficient. An aged electrode with 30% capacity fade may show a 15 to 25% reduction in Li/Mg selectivity at the same operating potential compared to the same electrode when new. Accounting for this aging effect in the control logic, by gradually adjusting the operating potential window as the electrode ages, extends effective service life compared to running constant setpoints until the purity specification is violated.
Solid-State Diffusion and Cycling Rate
The rate-limiting step in the intercalation process for practical-scale packed bed electrodes is typically solid-state diffusion of Li+ within the electrode particle after it has crossed the electrode-solution interface. Solid-state diffusion of Li+ in lambda-MnO2 is characterized by diffusion coefficients in the range of 10^-10 to 10^-12 cm2/s, depending on temperature and electrode crystallinity. These values are many orders of magnitude lower than aqueous diffusion coefficients.
The practical implication for process engineering: if you run adsorption cycles too short (trying to maximize cycle frequency for throughput), the Li+ has not had sufficient time to diffuse deep into the electrode particles and the bulk of the intercalation sites within the particle interior are never reached. You are loading only the outer shell of each electrode particle, which reduces effective capacity per unit mass of electrode material and increases the surface-area-to-volume ratio effect that makes purity worse (the outer shell has higher Mg2+ exposure per unit Li+ loaded than the interior would).
The optimal adsorption phase duration balances the solid-state diffusion requirement against the diminishing returns of late-stage adsorption. As intercalation proceeds and the outer particle layers approach saturation, the driving force for diffusion deeper into the particle decreases. The marginal Li+ recovered per additional minute of adsorption follows a declining curve, while the Mg2+ co-adsorption rate per minute in the late stage is not declining as fast. There is a practical economic optimum on that curve, and it is not the same point for all electrode materials, particle sizes, or brine compositions.
Titanosilicate and Iron Phosphate Materials: How the Lattice Chemistry Differs
Lambda-MnO2 is the most studied material, but it is not the only commercially relevant intercalation material for DLE. Titanosilicate (H2TiO3) and iron phosphate (FePO4) materials are used in some deployments, and their lattice selectivity mechanisms differ from the spinel MnO2 in ways that affect process engineering practice.
H2TiO3 operates by an ion-exchange mechanism rather than electrochemical intercalation. Li+ exchanges for H+ in the titanate lattice at the electrode surface, driven by the concentration difference rather than an applied electrical potential. This means H2TiO3-based DLE systems are typically not electrochemically controlled in the same sense as MnO2-based systems: the driving force is chemical potential rather than electrode potential. The selectivity of H2TiO3 for Li+ over Mg2+ arises from the size-specific tunnel structure of the titanate lattice, with a slightly different basis than the spinel site energy argument for MnO2.
This distinction matters for process control architecture. If you are operating H2TiO3 electrodes, the control variables you primarily adjust are pH (which controls the H+/Li+ exchange equilibrium), temperature (which affects the exchange kinetics), and residence time. Electrode potential is not a primary control variable. The EELI control system's electrode potential optimization module is specific to electrochemically-driven intercalation materials (MnO2 and related spinel structures). The process data management, fouling detection, and feedforward brine variability modules apply to both material types, but the electrode potential control layer is only relevant for the electrochemically driven materials.
What Process Engineers Should Take from the Lattice Chemistry
The lattice chemistry of intercalation electrode materials is not just academic context. It defines three practical constraints that process engineers need to build into their control logic: the operating potential window (material-specific, narrows with aging), the solid-state diffusion time constant (sets the minimum useful adsorption cycle duration), and the Mg2+ co-adsorption risk zone (the late-stage adsorption region where the lattice's selectivity advantage is most stressed).
Knowing these constraints allows you to design control logic that works with the material physics rather than against it. A control system that ignores the solid-state diffusion time constant and runs very short cycles to maximize throughput is leaving electrode capacity unused and generating a worse selectivity outcome than the material is capable of. A control system that runs cycles long enough to approach the solid-state diffusion limit, terminates adsorption before the selectivity-degrading late stage becomes dominant, and adapts both timing and potential to the real-time feed composition is the control system that makes a good electrode material perform as well as it can.