In times of rising demand of Li for use in Li-ion batteries (LIBs), and ensuring its supply for aluminum electrolysis, ultralight alloys, optics, synthesis, nuclear research, and pharmaceuticals, it is essential to simplify the processing of Li sources in a sustainable manner. Current extraction from ores is complex and requires large energy input while generating high amounts of waste. This causes the danger of making Li a limited resource and even more expensive.[1] A new promising way is the electrochemical extraction of Li by desalination batteries (DBs) from brines or seawater[2]. Thereby, selective extraction of ions can be promoted by the chemistry of the electrode material, e.g., by using FePO4. To further enhance the selectivity, potentiostatic application proved to be useful[3]. DBs are superior to other extraction methods in terms of energy efficiency, and waste generation. Still, varying Li+ concentrations in brines[1b], competitive intercalation of other cations with similar size - especially Na+ [2] -, and degradative effects of anions[4], H2O, and O2 [5] are major challenges to overcome for DBs with FePO4.In this contribution, we focus on the selective extraction of Li+ using FePO4 electrodes by specifically-designed multi-step potentiostatic application. Results are shown for salt solutions containing a mixture of Na+/Li+, and artificial seawater. It is vital to understand the atomic-scale processes so that the electrodes, and electrochemical parameters can be optimized. Therefore, we used operando high-energy X-ray diffraction (HEXRD) microscopy to investigate the crystal structure of FePO4 during the electrochemical process. The experiments showed that the energy barrier for Na+ (de)intercalation was higher than that of Li+ during galvanostatic cycling, i.e., FePO4 is selective for Li+. Using a multi-step potentiostatic procedure, the (de)intercalation process could be enhanced, as evident from our HEXRD results and simultaneous conductivity measurements. Our tentative interpretation is that Li+ is highly selectively (de)intercalated, although electrochemical signatures hinted on a two-step process during deintercalation, which may indicate the intercalation of Na+. To selectively deintercalate the remaining Na+ over Li+, we further optimized the applied potentials. During galvanostatic cycling in artificial seawater, the (de)intercalation potentials were at more positive and more negative potentials, respectively, likely caused by the variety of cations. We expect that our results contribute to the design of high-performance DBs and inspire the exploration of further systems based on our proposed multi-step process.