High Operating Potential Induces Conversion in Li-Rich Chalcogenides

Abstract

Lithium-ion batteries are central to the renewable energy transition, yet the capacity of commercial cathode materials limits further improvement in performance. Li-rich cathodes offer a promising route to higher capacities by leveraging both cationic and anionic redox. However, the full theoretical capacity is yet to be realized in the Li-rich materials. Here, we evaluate conditions to attempt full delithiation, which requires higher operating potentials. We focus on the Li-rich sulfide and selenide Li₂FeCh₂ (Ch = S, Se), since high potential required to access anion redox in oxides leads to severe degradation of electrolyte which obscures intrinsic cathode processes. Both materials exhibit poor Coulombic efficiency and rapid capacity loss during galvanostatic measurements that reach >3 V vs Li/Li⁺. The materials undergo conversion reactions at these potentials, which is corroborated by ex situ laser ablation inductively coupled plasma mass spectrometry (ICP-MS) and X-ray diffraction (XRD) measurements. For Li₂FeS₂, conversion to polysulfides causes material loss. However, if the upper voltage cutoff is below 3 V vs Li/Li⁺, there is no evidence of dissolved S even though persulfides are formed in the solid Li_2–x FeS₂. In contrast, conversion of Li₂FeSe₂ to γ-Se is limited to the solid-state, though formation of protons via electrolyte decomposition causes active-material dissolution. The conversion reactions occur >1 V positive of the reversible anion redox reactions in both materials which makes it easy to avoid such parasitic side reactions by potential limited cycling.

Introduction

Li-ion batteries (LIBs) are the leading electrochemical energy-storage technology driving the transition to renewables and the electrification of key sectors such as transportation. However, addressing the ever-increasing demand for energy-storage capacity requires continued advancements of LIBs. Among the several battery components, cathode materials present a major bottleneck preventing further improvements in LIB energy density. − Over the past decade, various strategies have been explored to improve the cathode materials, with one promising approach being the development of “Li-rich cathodes.” − Conventional cathodes typically contain one mole of Li-ions per transition metal (e.g., LiCoO₂), with charge compensation during (de)­lithiation occurring primarily through transition-metal redox. In contrast, Li-rich cathodes contain more than one mole of Li-ions per transition metal (e.g., Li₂RuO₃) and enable multielectron redox processes by utilizing both cationic and anionic redox for charge compensation. , Consequently, Li-rich materials offer higher gravimetric capacity, making them a promising candidate for next-generation LIBs.

Among Li-rich cathodes, transition-metal oxides are the most extensively studied class of materials. Higher operating potentials are required for better utilization of their high theoretical capacities by enabling a greater extent of delithiation. However, it is well established that the high potentials necessary for deep delithiation simultaneously trigger parasitic reactions. − A prominent example is the irreversible release of framework oxygen as O₂ gas at high states of charge. − Oxygen loss generates vacancies within the crystal lattice, which in turn drives deleterious structural transformations, including the conversion from the layered structure to a rock-salt–type phase. ,, This phenomenon is not unique to Li-rich cathodes but is also observed in conventional commercial materials such as LiCoO₂, which exhibits irreversible O₂ release at higher potentials. ,,, Consequently, such parasitic processes limit practical capacity by preventing complete, reversible Li-ion (de)­intercalation due to irreversible structural degradation. Additionally, the parasitic processes translate into practical challenges in Li-rich cathodes, including pronounced voltage hysteresis during the first (dis)­charge cycle and progressive voltage fade upon extended cycling. ,

Furthermore, in oxide cathodes, anion redox typically occurs at potentials around 4 V (vs Li/Li⁺), coinciding with the onset of electrolyte breakdown. ,, This overlap convolutes electrochemical and spectroscopic signatures, thereby complicating the isolation of charge-compensation mechanisms and degradation pathways at high states of charge. In contrast, anion redox in Li-rich sulfides and selenides occurs at substantially lower potentials, owing to the higher-energy frontier p orbitals of sulfur and selenium. ,, We will henceforth refer to the sulfides and selenides simply as chalcogenides. The higher-energy p states make Li-rich chalcogenides particularly well-suited for probing redox mechanisms and degradation phenomena for highly delithiated electrodes. Consequently, Li-rich chalcogenides have attracted significant interest for both Li- and Na-ion batteries because of their lower anion redox potentials, greater compositional diversity, − and high gravimetric capacities. ,, The capacities are so high that, despite the modest voltage, Li-rich chalcogenides can reach energy densities of >1000 Wh/kg. Moreover, these materials can use more sustainable, earth-abundant transition metals such as iron while maintaining high energy density, in contrast to the cobalt- and nickel-rich compositions typical for oxide cathodes. ,,

Li₂FeS₂ has emerged as a promising Li-rich chalcogenide for both fundamental investigations of charge-compensation mechanisms and practical energy-storage applications, owing to its high theoretical capacity of 400 mAh/g. First synthesized in 1976, Li₂FeS₂ crystallizes in a layered structure with P3̅m1 space group, consisting of alternating layers of Li in edge-sharing octahedral sites and a layer of edge-sharing tetrahedral sites with mixed Li/Fe occupancy (Figure a). Upon delithiation, charge compensation initially proceeds via iron oxidation followed by sulfur oxidation at higher states of charge. The latter is accompanied by persulfide bond formation, making Li₂FeS₂ a hybrid system bridging conventional intercalation-type and conversion-type cathode chemistry. , Interestingly, the sulfide/persulfide redox does not result in irreversible material loss analogous to the release of O₂ in the oxides.

1.

Crystal structure of (a) Li₂FeS₂ and (b) Li₂FeSe₂ projected along the a-axis.

Despite the ability of Li₂FeS₂ to theoretically deliver two moles of electrons (or Li-ions) per formula unit, experimentally reported capacities consistently fall short of this limit. , Complete delithiation of Li₂FeS₂ is challenging because the fully oxidized phase (FeS₂) is thermodynamically unstable in the original layered structure and instead prefers transformation to the pyrite phase, a process that requires extensive structural rearrangement. Here, we attempt to fully delithiate Li₂FeS₂ and its isostructural selenide counterpart by charging to higher operating potentials. Owing to the larger ionic radius of Se^2– compared to S^2–, Li₂FeSe₂ has an increased interlayer spacing, as shown in Figure b. In this work, we probe the complete delithiation of Li₂FeCh₂ (Ch = S, Se) through a two-electron oxidation process, during which the materials reach potentials around 4 V. Complementary galvanostatic and potentiostatic measurements are employed to assess the electrochemical and chemical reversibility during the initial (dis)­charge cycles. In addition, inductively coupled plasma mass spectrometry (ICP-MS), combined with appropriate control experiments, is used to elucidate charge-compensation mechanism and degradation pathway at high states of delithiation and operating potentials. We find that both materials convert to polysulfides and γ-Se at high operating potentials, whereby the sulfide dissolves by virtue of polysulfide formation and the selenide reacts with protons generated by electrolyte decomposition to form soluble degradation products.

Materials and Methods

Dimethyl carbonate (DMC, anhydrous, ≥99%), ethylene carbonate (EC, anhydrous, 99%), iron sulfide (FeS, trace metal basis, 99.5%), propylene carbonate (PC, anhydrous, 99.7%), poly­(tetrafluoroethylene) (PTFE, powder), and selenium (99.99%) were purchased from Sigma-Aldrich. Lithium hexafluorophosphate (LiPF₆, battery grade) was purchased from Oakwood Chemical. Lithium sulfide (Li₂S, metals basis, 99.5%) was purchased from Thermo Scientific, iron (99.0%) from Acros Organics, and carbon black Super P (99%) and lithium granules (metals basis, 99%) from Alfa Aesar.

The binary precursors (Li₂Se, FeSe₂) and the cathode materials (Li₂FeS₂, Li₂FeSe₂) were all synthesized in an Ar-filled glovebox, where O₂ and H₂O levels were maintained below 0.1 ppm. Li₂Se and FeSe₂ were synthesized in batches of 200 mg each using conventional solid-state methods following previously reported procedures. For Li₂Se, stoichiometric amounts of Li and Se were placed in separate alumina crucibles, with Se crucible placed above the Li crucible in a fused quartz tube. The tube was sealed under vacuum (55–60 mTorr) and the resulting ampule heated to 400 °C at 1 °C/min, held for 12 h, and then ambiently cooled to room temperature. The gray reaction product was obtained from the lower crucible after opening the ampule. For FeSe₂, stoichiometric amounts of Fe and Se were ground using a mortar and pestle, pressed into a pellet, and sealed in evacuated fused quartz ampules under a vacuum of 55–60 mTorr. The ampule was heated to 700 °C at 10 °C/min, held for 12 h, then cooled to 550 °C at 1 °C/min and annealed for 24 h before being ambiently cooled to room temperature to yield a black product. Figure S1 presents the XRD patterns of the binary precursors, Li₂Se and FeSe₂, used in the synthesis of Li₂FeSe₂.

Li₂FeS₂ was synthesized in batches of 200 mg each following the procedure by Patheria et al. Stoichiometric amounts of Li₂S and FeS powders were ground together in an agate mortar and pestle and pressed into a pellet. The pellet was put in a carbon coated fused quartz tube which was sealed under a vacuum of 55–60 mTorr. The ampules were heated to 900 °C at 1 °C/min followed by a 12-h dwell time before being ambiently cooled to room temperature to yield a black product.

Li₂FeSe₂ was synthesized in batches of 200 mg each following the procedure by Martinolich et al. Stoichiometric amounts of Li₂Se, FeSe₂, and Fe powders were ground together in an agate mortar and pestle and pressed into a pellet. The pellet was put in a carbon coated fused quartz tube which was sealed under a vacuum of 55–60 mTorr. The ampules were heated to 500 °C at 5 °C/min followed by a 12-h dwell time, then further heated to 1000 °C at 1 °C/min, followed by a dwell time of 24 h. The ampule was then quenched to room temperature in a water bath to yield a black product.

Electrochemical Measurements

Composite cathodes for electrochemical measurements were prepared as freestanding electrodes inside an Ar-filled glovebox (H₂O and O₂ <0.1 ppm). A 60:20:20 (wt %) mixture of active material, Super P conductive carbon, and PTFE binder was hand-ground in an agate mortar and pestle for 5 min in 40 mg batches. The resulting powder was pressed into 3/8-in. diameter electrodes under 2 tons of pressure for 5 min using a hydraulic press, finally yielding three electrodes of approximately 12.5 mg total weight each.

All coin cells were assembled in an Ar-filled glovebox (H₂O and O₂ <1 ppm). Electrochemical measurements were conducted in 2032 coin cells (MTI) consisting of a spring, a stainless-steel spacer, a Li metal anode, a glass-fiber separator (Whatman GFD), 200 μL of electrolyte, and a freestanding cathode. The electrolyte used was LP100 which is composed of 1 M solution of LiPF₆ in a 60:20:20 (vol %) mixture of dimethyl carbonate (DMC), ethylene carbonate (EC), and propylene carbonate (PC). Electrolyte solution was prepared in dried high-density polyethylene (HDPE) bottles, where the salt and solvent mixture was allowed to dissolve overnight before use. The cathode (dis)­charge rate of C/10 was based on 1 e^– per formula unit. Electrochemical measurements were performed using a BCS 805 battery cycler (Bio-Logic). All voltages reported in this work are against a Li metal anode, which we take to be Li/Li⁺.

X-ray Diffraction (XRD) Measurements

Powder X-ray diffraction (XRD) data were collected using a Rigaku diffractometer equipped with a Cu Kα X-ray source. The sample powder was loaded in an air-free sample holder (Rigaku) inside an Ar-filled glovebox to prevent exposure to air. XRD patterns were collected with a Cu–K_α X-ray source at 5° min^–1 with 0.01° step size. Rietveld refinement was performed using GSAS-II software.

Laser Ablation Inductively Coupled Plasma Mass Spectrometry (ICP-MS)

Laser ablation inductively coupled plasma mass spectrometry (ICP-MS) was performed with an UP-193 laser ablation system from New Wave Research (2007) equipped with a Supercell ablation chamber. The system was coupled with an Agilent 8800 Triple-Quad ICP-MS instrument operated in no-gas single-quad mode. Data were analyzed with the software ICP-MS Masshunter (Agilent, San Jose, CA). In the UP-193, laser radiation from a YAG-pumped OPO is passed through a selectable aperture that is imaged onto the sample through a 15X objective lens. The laser was operated at 10 Hz and attenuated to provide a fluence of 1.21 J cm^–2 and an irradiance of 0.44 GW cm^–2. Ablated aerosol was carried out of the Supercell by a 9 mL min^–1 flow of helium gas and then combined with a 1 L min^–1 flow of argon that carries the aerosol into the injector of the quartz torch. Time-resolved analysis mode was used to record data for sulfur (32 and 34 m/z), iron (56 and 57 m/z), and selenium (77, 78, and 82 m/z) during the laser warmup (15 s), ablation time (45 s), and washout time (40 s). The dwell time for each mass was 10 ms. This length of acquisition provides an initial gas blank before the arrival of the aerosol and confirmation of return to baseline after the ablation is finished.

UV–Visible Spectroscopy Measurements

UV–visible spectroscopy measurements were performed on an Agilent 8453 spectrometer. Solutions were prepared inside an Ar-filled glovebox and placed in screw-capped cuvettes with a 1 cm path length.

Results and Discussion

We first investigate the electrochemical response of Li₂FeS₂ in two different operating potential ranges to determine if we can achieve a full two-electron oxidation per formula unit (f.u.). The successful synthesis of Li₂FeS₂ is confirmed with a quantitative Rietveld refinement on the powder XRD data and shown in Figure S2. Figure a shows the first three (dis)­charge cycles at C/10 rate within the operating potential window of 1.7–3 V typically used in the literature. , The cathode exhibits a capacity of approximately 1.7 mol of electrons per formula unit in the first charge cycle. Additionally, it exhibits high Coulombic efficiency (defined as the ratio of discharge to charge capacity) and a highly reversible (de)­lithation process, as evident by the near-overlapping charge–discharge curves over the subsequent cycles. Previous work shows that the initial sloping region of the charge curve corresponds to mainly transition-metal (Fe^2+/3+) redox, whereas the subsequent plateau is associated with anion (S^2–/S₂ ^2–) redox.

2.

Galvanostatic measurement of Li₂FeS₂ at C/10 rate based one electron per f.u. with a charge cutoff of (a) 3 V and (b) 2 mol e^– per f.u.

Figure b shows the electrochemical response under capacity-limited cycling with an upper cutoff corresponding to a two-electron per f.u. oxidation, which is used to attempt complete delithiation of Li₂FeS₂. Interestingly, above 3 V the material shows a new plateau around 3.8 V followed by voltage decay. When the material is allowed to continue oxidizing without imposing a two-electron cutoff on the first charge, the voltage drops from the 3.8 V plateau to ∼3.1 V, and the cell does not polarize even after an oxidation corresponding to 10 mol of electrons (Figure S3). When the charge cutoff is limited to 2 mol of electron, we observe that the Coulombic efficiency is notably poor during the first cycle, and the capacity undergoes a substantial decline in subsequent cycles. This behavior suggests that the electrochemical process responsible for the enhanced capacity in the first charge cycle is largely irreversible. Moreover, the continued capacity fade with cycling indicates that the processes occurring at high states of charge induce irreversible structural or chemical transformations within the material. We note that the galvanostatic measurement corresponding to 2 mol of electron per f.u. oxidation may not directly correspond to the removal of 2 mol of Li⁺ per f.u. from the electrode. At high operating potentials approaching 4 V, parasitic processes such as electrolyte decomposition can also contribute to the observed oxidation capacity during galvanostatic measurement. Consequently, quantitatively separating the contributions from electrochemical delithiation and parasitic side reactions remains challenging. To further probe the mechanisms underlying these observations and identify the origin of the differences in electrochemical response, we disassemble the coin cells for subsequent ex situ characterization.

We next investigate the structural change in Li₂FeS₂ cycled under a capacity-limited upper cutoff corresponding to a two-electron oxidation. In an ideal scenario, where the capacity arises exclusively from Li₂FeS₂ delithiation without any parasitic side reactions, the expected product at the end of the first charge would be FeS₂. However, this transformation is nontrivial because Li₂FeS₂ adopts a layered structure, whereas FeS₂ crystallizes in the pyrite structure. Fe is tetrahedrally coordinated in Li₂FeS₂ with all S being sulfide anions (S^2–), whereas in FeS₂, Fe occupies octahedral sites and S exists as persulfide species (S₂ ^2–). This substantial structural transformation likely requires a large overpotential, which is consistent with the absence of pyrite formation at potentials below 3 V. In fact, to observe any additional oxidation, the cell must be polarized to nearly 4 V, which is consistent with a large overpotential for conversion. Dahn and co-workers investigated the reverse process, i.e., the lithiation of pyrite FeS₂, and showed that reduction of the persulfide species in pyrite FeS₂ is kinetically controlled. However, Li₂FeS₂ is also not observed as an intermediate during FeS₂ electrochemical lithiation, , although early reports suggested its presence.

To determine whether Li₂FeS₂ undergoes a structural transformation to pyrite-type FeS₂ upon two-electron oxidation, we perform ex situ XRD at the end of the first charge. Figure a shows the XRD patterns of pristine Li₂FeS₂ and the electrode recovered after the two-electron oxidation. The pristine material exhibits the expected diffraction peaks, consistent with the P3̅m1 space group. The pattern of the oxidized material charged to 2 mol of electron shows peaks associated with unreacted Li₂FeS₂ at 14° and 28.3° 2θ, likely due to the freestanding electrode fabrication. A new reflection is observed at 33.5° 2θ which is present in a separate sample charged to 3 V as well. This reflection is seen in the pristine material and appears to remain unchanged when the material was cycled between 1.7 to 3 V in a previous study. The XRD data magnified to observe the peak at 33.5° 2θ as a function of state of charge are shown in Figure S4. Notably, no pyrite FeS₂ formation is seen in the oxidized material as the most intense reflection at 33° 2θ is absent. The remainder of the peaks in the sample charged to 2 mol of electron corresponds to delithiated Li₂FeS₂, which shows a loss in crystallinity and a shift in the (001) peak to slightly higher 2θ, consistent with the operando results reported previously. The delithiated Li₂FeS₂ exhibits a loss of crystallinity and long-range order for two primary reasons. First, the two-phase anion redox reaction, in which S^2– is oxidized to persulfide (S₂ ^2–) species, disrupts the topotactic delithiation pathway that would otherwise preserve the parent crystal framework. Second, formation of persulfide bonds requires substantial distortion and cooperative tilting of the FeS₄ tetrahedra while maintaining nearly unchanged Fe–S bond lengths. These structural rearrangements disrupt long-range order and lead to broadened diffraction features. A more detailed discussion of this phenomenon is reported in a previous work. These results indicate that Li₂FeS₂ does not transform into FeS₂ after a two-electron oxidation, and the system does not undergo a pyrite-type conversion reaction during the first charge.

3.

(a) XRD patterns of the pristine Li₂FeS₂ electrode, electrode charged to 3 V and the electrode recovered at the end of the first charge after a two-electron oxidation. Peak labeled * is from the PTFE binder used in the electrodes. The bottom curve shows the simulated XRD pattern of pyrite FeS₂ plotted using CIF file from Inorganic Crystal Structure Database (ICSD). (b) Electrochemical response of the Li₂FeS₂ electrode at a C/10 rate, for which the corresponding XRD measurement charged to 2 mol e^– is presented.

Having ruled out the conversion of Li₂FeS₂ to pyrite FeS₂ following a two-electron oxidation, we next evaluate the presence of soluble products by characterizing the glass-fiber separators from the coin cells. Figure a,b shows the glass-fiber separators recovered from the cells after electrochemical cycling. The separator after electrochemical cycling between 1.7 and 3.0 V appears unchanged, as shown in Figure a. In contrast, the separator from the cell that is cycled with a two-electron charge cutoff shows noticeable discoloration (Figure b), indicating a change in the electrolyte composition. To determine whether the discoloration arises from electrolyte decomposition or from cathode material dissolution, we perform laser ablation ICP-MS on the separators. Figure c,d shows the intensity of the ICP-MS for the m/z ratio corresponding to ⁵⁶Fe and ³²S isotopes, respectively. Figure c shows the intensity of the m/z ratio corresponding to ⁵⁶Fe, which is negligible before laser ablation for the separator from the two-electron charge-cutoff cell. Upon laser ablation, the intensity increases significantly, confirming the presence of Fe in the discolored separator. The separator from the cell cycled between 1.7 and 3.0 V shows no increase in intensity at that m/z ratio upon laser ablation, suggesting the absence, or very low concentration of ⁵⁶Fe.

4.

Photographs of glass-fiber separators recovered from the cells after 3 galvanostatic cycles with a charge cutoff of (a) 3 V and (b) 2 mol e^– per f.u. Laser ablation ICP-MS response for m/z ratio corresponding to (c) ⁵⁶Fe and (d) ³²S species for each charge condition.

It is important to take precautions while investigating that an element of interest is indeed the source of the signal at a given m/z in mass spectrometry, rather than an isobaric polyatomic ion. In this regard, Fe poses particular analytical challenges. When performing solution-based ICP-MS, ⁵⁶Fe m/z signal is impacted by the formation of ArO⁺ in the wet plasma. The dry plasma used for laser ablation ICP-MS in this work has significantly lower oxygen levels due to absence of water. This allows us to use the no-gas single-quad mode rather than either kinetic energy discrimination mode using helium or reactive hydrogen mode to remove the ArO⁺. Additionally, the ICP-MS data in Figure show the “gas blank” level at the respective m/z before and after the laser ablation which enables separation of background contributions from the ablation-derived signal, allowing confident assignment of the feature observed at m/z of 56 to ⁵⁶Fe. We additionally record data at m/z 57 corresponding to the ⁵⁷Fe isotope to further support our conclusion of iron loss from the cathode at high operating potentials (Figure S5).

Figure d shows the intensity of the m/z ratio for ³²S, which exhibits a higher background and a slight increase upon laser ablation for the two-electron charge-cutoff cell compared to the cell cycled below 3 V. No other species are expected at this m/z; however, the trace looks different before and after the laser is turned on compared to ⁵⁶Fe because the separator is exposed to ambient atmosphere during the measurement causing hydrolysis to form H₂S. This behavior likely accounts for the elevated background signal observed throughout the measurement, independent of laser ablation, in Figure d. However, ³²S is clearly detected from the separator. In contrast, the separator from the cell cycled below 3 V exhibits a substantially lower signal, indicating significantly less sulfur content. Together, these observations suggest the presence of sulfur-containing species in the separator from the cell cycled above 3 V. To further probe the nature of the soluble degradation products, the discolored separator is soaked in a 1,3-dioxolane/1,2-dimethoxyethane solvent (DOL/DME, 1:1 by volume), and the resulting solution is analyzed by UV–visible spectroscopy. A prominent peak centered at 265 nm is observed (Figure S6), consistent with the characteristic signature of polysulfide species in this solvent system. Based on these results, we propose a generalized degradation pathway for cathode material loss in which Li₂FeS₂ is converted to polysulfides

$${\mathrm{Li}}_2{\mathrm{FeS}}_2\to {\mathrm{Li}}_{2-x}{\mathrm{Fe}}_{1-y}{\mathrm{S}}_{2-\mathit{bz}}+y{\mathrm{Fe}}^{2/3+}+b{\mathrm{S}}_z^{n-}+x{\mathrm{Li}}^{+}+x{\mathrm{e}}^{-}$$

These findings elucidate that polysulfide formation is responsible for the low Coulombic efficiency and continuous capacity fade of Li₂FeS₂ during galvanostatic cycling under a two-electron capacity cutoff, as shown in Figure b. Polysulfides shuttle between the cathode and anode continuously during charge, resulting in the long charge plateaus observed if the cell is allowed to charge uninterrupted as observed in Figure S3.

In order to assess whether the observed active-material loss is specific to Li₂FeS₂ or represents a broader degradation phenomenon in Li-rich chalcogenides, we next investigate its isostructural selenide analogue Li₂FeSe₂. Figure shows the first three galvanostatic charge–discharge cycles at a C/10 rate with two different upper cutoff conditions. Within a potential window of 1.7–2.5 V (Figure a), Li₂FeSe₂ delivers a reversible capacity of approximately 1.5 electrons per formula unit, which is in good agreement with literature. Similar to Li₂FeS₂, the selenide exhibits high Coulombic efficiency and excellent reversibility, as evidenced by the near-overlap of the charge and discharge profiles. In contrast, extending the upper cutoff to either 4.5 V or 4 electrons capacity, whichever is reached first, results in a distinctly different electrochemical response as shown in Figure b. The cell does not polarize on the first charge and delivers a substantially higher capacity due to a new oxidation process occurring beyond 4 V. Subsequent cycles are accompanied by poor Coulombic efficiency and pronounced capacity fade. This behavior indicates that the process responsible for the high initial capacity is largely irreversible, mirroring the high-voltage response observed for Li₂FeS₂ in Figure b. The continued capacity loss further suggests that high-potential cycling induces irreversible structural and/or chemical transformations.

5.

Galvanostatic measurement of Li₂FeSe₂ at C/10 rate at an upper cutoff of (a) 2.5 V and (b) 4.5 V or 4 mol e^– per f.u. capacity.

Notably, we see a distinct peak-shaped feature during the first charge at a capacity of approximately two moles of electrons per formula unit. This feature resembles the peak observed in the first charge cycle of Li₂FeS₂ (Figure ); however, in the selenide analogue, it is significantly less pronounced. This feature is reproducible across multiple Li₂FeSe₂ batches as shown in Figure S7 and likely represents the onset of active-material loss. We set a capacity-based cutoff at 4 mol of electron per formula unit during the charge cycle to better capture this feature and subsequent plateau in the galvanostatic response. In the first charge, the capacity-based cutoff is reached before the 4.5 V potential-based cutoff, yielding capacities beyond the theoretical two-electron intercalation limit per formula unit of Li₂FeSe₂. This behavior suggests the presence of a charge-compensation mechanism beyond putative FeSe₂ formation. When the capacity cutoff is extended beyond 4 mol of electron per formula unit, the potential gradually decreases with time (Figure S8), and the material delivers capacities exceeding eight electrons per formula unit, further supporting the emergence of an alternative charge-compensation pathway at high operating potentials.

We next investigate the structural change in Li₂FeSe₂ following the first charge. Figure a compares the XRD pattern of the pristine and charged samples, with the corresponding electrochemical response shown in Figure b. The pristine material exhibits the expected diffraction pattern of Li₂FeSe₂. Upon oxidation, however, all prominent reflections for Li₂FeSe₂ disappear. Additionally, there is no evidence of FeSe₂ formation as indicated by the absence of its most intense reflection at 35° 2θ. The observed reflections in the charged sample align well with those of elemental selenium (γ-Se) and FeSe, as suggested by the presence of their most intense reflections at 29.7° and 28.6° 2θ, respectively. We note that the FeSe peak at 15.8° 2θ is very low in the XRD of the oxidized material. FeSe is a layered material and thus we suggest this peak could be absent due to preferred orientation. The presence of two new products indicates that extensive conversion reactions are taking place upon delithiation of Li₂FeSe₂. Based on these results, we propose a generalized pathway for the conversion reaction

$${\mathrm{Li}}_2{\mathrm{FeSe}}_2\to {\mathrm{Li}}_{2-x}{\mathrm{Fe}}_{1-y}{\mathrm{S}}_{2-\mathit{bz}}+y\mathrm{FeSe}+b{\mathrm{Se}}_z+x{\mathrm{Li}}^{+}+x{\mathrm{e}}^{-}$$

6.

(a) XRD patterns of the pristine Li₂FeSe₂ powder and the electrode recovered at the end of the first charge. Peak labeled * is from the PTFE binder used in the electrode. The bottom two curves show the simulated XRD pattern of FeSe and γ-Se plotted using CIF files from Inorganic Crystal Structure Database (ICSD). , (b) Electrochemical response of the Li₂FeSe₂ electrode (charged) at a C/10 rate, for which the corresponding XRD measurement is presented.

We disassemble the coin cells after cycling under the conditions shown in Figure to assess how potential above 2.5 V affects the degradation of Li₂FeSe₂. Figure a shows a glass-fiber separator recovered from a cell cycled between 1.7–2.5 V and no discoloration is observed. In contrast, the separator from the cell that reached potentials around 4 V exhibits pronounced discoloration, as shown in Figure b. The discoloration ranges from bright orange to deep red depending on how long the cell is cycled (Figure S9). To identify the species, we perform laser ablation ICP-MS analyses on the recovered separator shown in Figure . Figure c, d shows a substantially higher signal for the m/z ratio corresponding to ⁵⁶Fe and ⁷⁸Se, respectively, for the separator from the cell cycled up to 4.5 V compared to the separator from the cell cycled below 2.5 V suggesting Fe and Se loss to the electrolyte. Again, we record data at m/z 57 corresponding to the ⁵⁷Fe isotope to further support our conclusion of iron loss from the cathode at high operating potentials (Figure S10). For Se, the preferred m/z of 78 is also present in any Ar plasma due to ³⁸Ar⁴⁰Ar⁺ species. Therefore, in addition to recording the “gas blank” levels, we also monitor two additional Se isotopes at m/z of 77 and 82 to validate observations made at m/z 78, with all three signals exhibiting consistent behavior (Figure S11). This analysis demonstrates that high operating potentials cause loss of Fe and Se from Li₂FeSe₂ into the electrolyte. Se and Fe dissolution along with irreversible cell cycling is also observed with a charge capacity cutoff of 2 mol of electron per f.u. (Figure S12). Together, these results demonstrate that charging Li₂FeSe₂ to high potentials induces dissolution of the cathode active material and/or generation of a soluble product(s) (vide infra). This degradation mechanism accounts for the low Coulombic efficiency and continuous capacity fading observed during galvanostatic cycling at elevated upper cutoff voltages.

7.

Photographs of glass-fiber separators recovered from the cells after 6 galvanostatic cycles at an upper cutoff of (a) 2.5 V and (b) 4.5 V or 4 mol e^– per f.u. capacity. Laser ablation ICP-MS data from the corresponding recovered separators for the m/z corresponding to (c) ⁵⁶Fe and (d) ⁷⁸Se species.

Unlike Li₂FeS₂, where polysulfide formation is observed, we do not expect soluble polyselenide formation from the oxidation of Li₂FeSe₂. Multiple studies in the literature report the absence of polyselenide formation due to its insolubility in carbonate-based solvents similar to those used in this work. − To confirm the absence of polyselenide species in solution, the colored separator retrieved from cells cycled to high potentials is soaked in tetraglyme, a solvent known to dissolve polyselenides. Subsequent UV–visible spectroscopy shows no characteristic absorption features associated with polyselenide species between 300 and 400 nm, , supporting their absence (Figure S13). However, the spectrum exhibits a higher absorption intensity around 220 nm, indicating the presence of soluble degradation products.

The potential-dependent nature of the active-material loss, occurring around 4 V, provides important insights into its underlying cause. It is well established that at such high potentials, electrolyte decomposition becomes significant. ,, Comprehensive studies have shown that the electrochemical breakdown of carbonate solvents generates a variety of proton sources, such as formic acid and methanol, at the cathode-electrolyte interface. To test whether these protons trigger active-material dissolution, we perform control experiments in which 10 μL of H₂O is added to 0.5 mL of LP100 electrolyte in the presence of Li₂FeSe₂ powder. The resulting solution upon contact with the glass-fiber separator reproduced the distinctive bright-orange coloration (Figure S14) observed in Figure S9. Interestingly, when H₂O is added to a suspension of Li₂FeS₂ in LP100, no discoloration is observed in the liquid phase (Figure S15). This finding provides a unifying explanation for the observations from the galvanostatic measurements, suggesting a mechanism for active-material dissolution in Li₂FeSe₂. In summary, at high operating potentials, electrolyte decomposition at the electrode surface generates protons, which subsequently promotes formation of soluble degradation products that color the separator.

Conclusions

This study reveals similar behavior of Li₂FeS₂ and Li₂FeSe₂ cathodes at high operating potential when full delithiation is attempted. Electrochemically, the materials function similarly with both exhibiting a high-voltage oxidation process which causes poor Coulombic efficiency and rapid capacity fade. However, both the sulfide and the selenide fail to convert to their thermodynamic binary end-member, pyrite FeS₂ and marcasite FeSe₂, to any significant amount. Instead, oxidation to soluble polysulfides and insoluble elemental γ-Se is observed.

Ex situ laser ablation ICP-MS analysis of coin cell separators shows cathode material loss into the electrolyte for both materials, revealing dissolution processes as a dominant degradation pathway at higher potentials. However, the mechanism of dissolution is different for the two materials. In Li₂FeS₂, charging above 3 V leads to polysulfide formation, as confirmed by UV–visible spectroscopy. We note that cathode dissolution is not observed below 3 V, suggesting polysulfides are only formed at high potentials. By contrast, the γ-Se formed during Li₂FeSe₂ oxidation is insoluble, yet Fe and Se loss into the electrolyte is observed indicating a distinct degradation pathway. Control experiments suggest a mechanism in which electrolyte oxidation at high potentials generates protons, which react with the selenide to form a brightly colored soluble product. Overall, this study establishes active-material loss as a fundamental limitation to reversibly accessing the full theoretical capacity of Li₂FeS₂ and Li₂FeSe₂. By identifying the onset potential of dissolution processes in Li-rich chalcogenides, our findings confirm that lower voltage cycling of the Li-rich chalcogenides bypasses formation of any soluble products or conversion reactions. The potential required to form such products is >1 V positive of the potential at which reversible anion redox is observed, which makes it easy to avoid such parasitic side reactions. We contrast this with the Li-rich oxides, which show O₂ evolution at the potentials required to access anion oxidation.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.6c06453.

• XRD pattern of the binary synthesis precursors (Li₂Se, FeSe₂) and cathode materials (Li₂FeS₂, Li₂FeSe₂), galvanostatic measurements, laser ablation inductively coupled plasma mass spectrometry (ICP-MS) analysis, and UV–visible spectroscopy data (PDF)

The authors declare no competing financial interest.