The selectivity problem in DLE is fundamentally a control problem, not just a materials problem. The electrode material sets the theoretical selectivity ceiling. The electrode potential profile you run in practice determines whether you reach it.
Why Mg/Li Selectivity Is a Control Problem
Most of the DLE community conversation around Mg/Li selectivity focuses on electrode material choice. Lambda-MnO2, titanosilicate frameworks, and iron phosphate materials all exhibit different selectivity coefficients for Li+ over Mg2+ in controlled lab conditions. Those selectivity coefficients are real, but they are measured at fixed electrode potentials against standard solutions. A brine feed is not a standard solution, and the electrode potential is not a fixed number in an operating plant.
The published selectivity coefficient K(Li/Mg) for a lambda-MnO2 electrode under standard test conditions typically falls in the range of 3 to 8, depending on test brine composition and temperature. In a field brine with a molar Mg/Li ratio of 10:1 (within the normal range for Atacama or Clayton Valley continental sources), even a selectivity coefficient of 5 still delivers a meaningful Mg2+ load to the electrode during the adsorption phase. That co-intercalated Mg2+ does not deintercalate cleanly during elution, because the Mg-O bond within the spinel lattice is more stable than the Li-O bond at the same structural site. The result is gradual electrode capacity loss and purity degradation in the eluate stream.
The selectivity coefficient is not constant across operating conditions. It varies as a function of the applied electrode potential. Mg2+ intercalation into the spinel lattice becomes thermodynamically more favorable at more negative electrode potentials, because the divalent ion requires greater reducing driving force to be incorporated into the crystal structure. If your electrode potential drifts negative during adsorption (which happens as brine conductivity shifts or as the electrode approaches saturation), you are effectively lowering the practical selectivity of the material, even though the material itself has not changed.
How Voltage Gradients Shape the Intercalation Phase
In this context, "voltage gradient" refers to the shape of the electrode potential profile over the duration of the adsorption phase, not a spatial gradient across the electrode. A fixed-setpoint system holds the electrode at a constant potential (or drives constant current) from the start of adsorption through to the end. An adaptive control system varies the potential dynamically as a function of real-time brine composition and electrode state of charge.
The mechanism behind this matters. During the early part of adsorption, the electrode surface has high available capacity and the near-surface Li+ concentration is close to the bulk brine value. At this stage, even a somewhat more negative electrode potential delivers acceptable Mg/Li selectivity, because the thermodynamic driving force for Li+ intercalation dominates over Mg2+. As adsorption proceeds and the electrode approaches saturation, the near-surface Li+ depletes via local mass-transfer limitations, the effective Mg/Li ratio at the electrode surface rises above the bulk brine value, and continuing to drive the same electrode potential starts forcing Mg2+ onto the remaining vacant sites.
The ideal adsorption potential profile is not a constant hold. It is a time-varying function that tracks electrode state of charge and near-surface brine composition simultaneously. The optimal shape depends on initial brine Mg/Li ratio, Li+ concentration, volumetric flow rate (which sets boundary layer thickness and mass transport), and electrode age (which affects available site density and surface chemistry over time).
Observations from a Nevada Continental Brine Pilot
In a 2024 pilot program with a continental brine project in Nevada, the feed brine showed a regular daily Mg/Li molar ratio swing between 6:1 and 12:1, driven by the mixing dynamics of the evaporation pond feeding the DLE cell. The operators were running the electrode stack on a constant-current protocol set during commissioning against the median brine composition.
During the high-Mg periods, the eluate Li+ purity dropped below the 99.5% threshold required for downstream processing. The operators had attributed this to electrode degradation and were planning a replacement cycle. When we instrumented the feed inlet with real-time Mg/Li monitoring and connected it to an adaptive potential control loop, the purity excursions during high-Mg periods dropped to less than one-third of their prior frequency within the first two weeks of operation. The electrode material was not degraded. It was being operated at the wrong potential for the brine composition it was seeing at those times.
The control change was not large in absolute potential terms: a 40 to 80 mV reduction in the maximum adsorption potential during high-Mg periods, implemented as a continuous function of the inlet Mg/Li ratio rather than a threshold step-change. The graduated response prevented the sharp purity excursions that occurred when the old control scheme's single threshold was crossed.
Selectivity Versus Throughput: The Honest Tradeoff
We should be direct about the cost of adaptive selectivity optimization. Restricting the electrode potential to preserve selectivity means accepting lower electrode utilization per cycle. A shallower maximum adsorption potential leaves some intercalation capacity unused, reducing the Li mass recovered per cycle. To maintain the same total throughput, you either run longer cycles (reducing cycle frequency per day) or deploy more cells in parallel (increasing capital cost).
This is not a reason to avoid adaptive voltage control. It is a reason to calibrate the selectivity versus throughput tradeoff to your specific downstream requirements. A project supplying a battery-grade lithium product with a strict less-than-50-ppm Mg specification will accept the throughput reduction to maintain purity margins. A project supplying an intermediate product to a downstream refinery with its own purification capability may prefer to tolerate wider purity variation and optimize for maximum cycle yield instead.
The problem with fixed-setpoint operation is that the tradeoff is set once at commissioning and held constant regardless of daily brine variation. On a high-Mg day, a fixed setpoint either under-protects purity or over-restricts throughput. An adaptive system adjusts the tradeoff in real time, which is what the process actually requires.
What the Control System Monitors for This Function
The EELI system uses four primary signals for the selectivity control function: Li+ concentration at the feed inlet (ion-selective electrode, approximate 2-minute response time), Mg2+ concentration at the feed inlet (ion-selective electrode, approximate 3-minute response time), volumetric flow rate (electromagnetic flowmeter, continuous), and electrode potential in the cell (reference electrode, continuous). From these inputs, the control model computes the effective Mg/Li molar ratio at the electrode face, accounting for the hydraulic transit time between inlet sensor and electrode bed midpoint.
That transit time calculation is not trivial. In a typical DLE cell configuration with a packed electrode bed operating at 2 to 5 cubic meters per hour, the travel time from inlet sensor to electrode midpoint is roughly 8 to 25 minutes depending on bed depth and flow rate. Setting the electrode potential based on the current inlet reading without accounting for that transit delay means reacting to a brine composition that will not reach the electrode for another 8 to 25 minutes, while the brine already at the electrode corresponds to the inlet reading from that many minutes ago.
The predictive layer of the control loop uses the current inlet measurement as a forward estimate of what the electrode will see after the transit delay, applies the selectivity model to the projected composition, and adjusts the potential profile ahead of the brine front arrival. That predictive adjustment is what makes the loop effective at handling the gradual Mg/Li ramps typical of salar brine systems, rather than simply reacting after purity has already been impacted.
Where Adaptive Voltage Control Has Limits
Adaptive potential control is not a substitute for electrode material selection in extreme Mg/Li environments. In brines with molar Mg/Li ratios above 40:1, which occur in some evaporite basin edge zones and certain geothermal surface expressions, the mass action of Mg2+ at the electrode surface becomes dominant regardless of potential control strategy. At those ratios, the control layer cannot compensate for the thermodynamic driving force on its own. A pre-concentration step to improve the Mg/Li ratio before the DLE cell, or an electrode material with intrinsically higher selectivity under those conditions, is the appropriate engineering response.
The control layer provides the most value in the Mg/Li range of 2:1 to 25:1, which covers the majority of commercially relevant continental brine sources and most operated geothermal lithium projects. Understanding where that range ends is as important as knowing what the system can do within it.