Morphology and facet effects on the charge and discharge mechanisms in FeSe₂-based lithium-ion storage

Abstract

Small and large FeSe₂ spheres with {111}-bound triangular protrusions, plus stars and {011}-exposed bars, have been synthesized and used as an anode material for lithium-ion batteries. Crystal facets regulate electrical conductivity and influence ion diffusion. A long-term cycling test at 1 A g⁻¹ revealed four stages with large electrochemical capacity changes. Small spheres with a combination of high electrical conductivity and a relatively large surface area exhibited the best capacity stability at current densities of 1 and 10 A g⁻¹. Through synchrotron X-ray diffraction (XRD) analysis, a novel reaction mechanism for lithium ion storage with FeSe₂ quickly converted to FeSe and then to Fe and Li₂Se during discharging was identified. At different stages, the Li₂Se lattice constant varies slightly, suggesting a correlation between crystal lattice and lithium-ion insertion and extraction. Cyclic voltammetry analysis showed variations in the reduction peak contributions at different stages. This study illustrates that control of crystal face, electrical conductivity, and surface area are critical to the optimization of battery performance.

---

Introduction

Lithium-ion batteries offer high energy density and high power density for widespread uses.^1,2 Despite the progress made in anode materials for lithium- and sodium-ion batteries, low conductivity, poor thermal stability, and volume change issues still lead to electrode degradation and reduced performance.³ Consequently, extensive efforts have been made to explore novel anode materials, particularly transition metal oxides and chalcogenides with high theoretical capacities.⁴ While transition metal oxides offer high theoretical capacities, their drawbacks include low conductivity, high initial irreversible capacity loss, and poor rate performance.^5,6 On the other hand, transition metal chalcogenides could offer more active sites, enhanced ion diffusion kinetics, and superior conductivity, resulting in longer cycle life and excellent capacity retention under high current density operation.^7–9 For example, iron sulfides and selenides have been suggested to possess high theoretical energy storage capability.¹⁰ In particular, iron diselenide presents abundant active sites, excellent conductivity, high capacity, and controllable nanoscale morphology, making it highly promising for energy storage applications.^11,12 Various strategies have been employed to inhibit FeSe₂ electrode defects produced during charge–discharge cycles. For example, FeSe₂ nanoparticles were embedded in a porous carbon matrix to achieve a high capacity of 798.4 mA h g⁻¹ after 100 cycles at a current density of 0.1 A g⁻¹, without phase transitions during the cycling process.¹³ Ultra-small FeSe₂ nanoparticles embedded in a carbon framework exhibited outstanding cycling stability, delivering 395 mA h g⁻¹ after 200 cycles at 0.5 A g⁻¹.¹⁴ FeSe₂ microspheres showed long-term cycling stability, maintaining a capacity of 372 mA h g⁻¹ after 2,000 cycles at 1.0 A g⁻¹.¹⁵ These studies highlighted that embedding FeSe₂ in a carbon framework or tuning the crystal morphology are strategies to improve battery stability. Moreover, various FeSe₂ morphologies, including nanotubes, sea urchin structure, micro–nanorods, and microspheres, mixed with graphene/carbon nanotubes to form composites, have been applied to lithium- and sodium-ion batteries to improve structural stability and cycling performance.^11,16–19

We consider crystal surface control as another way that can be explored to improve lithium-ion battery performance and stability, as this aspect is rarely examined. The exposed crystal facets should affect ion diffusion kinetics, reversible storage capacity for cations, and electrical conductivity. Crystal facet engineering has been proven effective in mitigating volume changes and improving electrochemical performance. For instance, a crystal growth strategy that promoted rapid potassium ion diffusion kinetics on the {040} facet could achieve excellent capacity retention over 1500 cycles.²⁰ The exposed {111} faces of LiNi_0.5Mn_1.5O₄ facilitated the formation of a protective solid electrolyte interphase layer, thereby enhancing its cycling stability.²¹ Sea-urchin-like LiNi_0.5Mn_1.5O₄ with dominant {111} facets was found to enhance structural stability, reduce Mn dissolution, shorten Li-ion diffusion paths, and improve battery performance.²² Furthermore, semiconductor crystals broadly possess facet-dependent electric conductivity properties.^23–29 The origin of this phenomenon has been found to relate to the presence of a facet-specific surface layer with slight lattice constant variations from that of the crystal bulk, as well as greater lattice disturbance in the surface layer.^30–32 These surface lattice deviations create dissimilar barriers to charge transport across a particular surface. Considering the small size of lithium ions, the intercalation stress generated during the charge–discharge process is relatively low. Thus, even without carbon coating, the anode material is less prone to structural collapse. This inherent stability motivated us to incorporate different FeSe₂ morphologies for lithium ion storage.

Here, three distinct morphologies of FeSe₂ including star-like structures, {011}-exposed rods or bars, and sphere structures with multiple octahedral protrusions were synthesized and used as anode for lithium ion storage. They showed diversely different specific capacity, rate performance, and cyclic stability properties. Among them, {111}-exposed small spheres achieved remarkable specific capacity, good rate performance, and excellent stability up to 21 000 cycles at a high current density of 10.0 A g⁻¹. Synchrotron XRD analysis revealed lattice constant changes and novel electrochemical reactions occurring at different cycling stages. From the refined diffraction data, we found a correlation between the Li₂Se lattice constant in each stage and the capacity at that cycle. Cyclic voltammetry analysis was also conducted for each stage to examine the variation in the reduction peak contribution.

Results and discussion

Synthesis and structural characterization of FeSe₂ crystals

FeSe₂ stars were prepared in 0.1 M hydrochloric acid, while other particle shapes were synthesized in 1 M citric acid. The synthesis conditions adopted report on the growth of FeSe₂ microspheres.^33,34 To obtain small FeSe₂ spheres, FeCl₂ solution was first mixed with triethanolamine (TEOA) and citric acid under sonication for 10 min, followed by the addition of a selenium solution dissolved in hydrazine. After sonication for 10 min, ethylene glycol was introduced as the solvent. The mixture was transferred to a 25 mL Teflon-lined autoclave with stirring and heated to 160 °C in an oven for 12 h. A higher TEOA concentration was used to form large FeSe₂ spheres. For the growth of FeSe₂ bars, TEOA was replaced with sodium dodecyl sulfate, and no additional solvent was introduced (see Fig. S1, SI for the experimental details). Fine adjustments to the reagent concentrations allowed control over particle size and morphology.

Fig. 1 shows scanning electron microscopy (SEM), transmission electron microscopy (TEM), and high-resolution TEM (HR-TEM) images of the synthesized FeSe₂ crystals. SEM images indicate that the FeSe₂ spheres are composed of small and large octahedra on the surface with approximate average sizes of 2 and 7.6 µm, respectively. In contrast to the sharp-faced bars or rods, they can also grow out into a star or multipod shape. HR-TEM images of the spheres display lattice fringes with d-spacings corresponding to the (111) planes of FeSe₂. The TEM result and the octahedral surface structure support the exposure of {111} faces in the spheres. The (101) and (011) planes were observed in a FeSe₂ star. When the electron beam was focused on a FeSe₂ bar, the crystal became disintegrated from the beam energy, so TEM images were not obtained with this sample. Fig. S2, SI presents XRD patterns of these four samples aligning perfectly with the orthorhombic phase of FeSe₂. The FeSe₂ bars exhibited strong (011) and (022) peaks from a preferred particle orientation on the substrate, inferring that the {011} facets are the predominantly exposed surfaces of the bars.

Fig. 1 (a–d) SEM images of the synthesized FeSe₂ (a) small spheres, (b) large spheres, (c) stars, and (d) bars. (e–j) TEM images of the FeSe₂ (e–h) small spheres, (f–i) large spheres, and (g–j) star and HR-TEM images of the corresponding red line regions.

To probe the possible presence of bulk and surface layer lattices in these crystals, synchrotron XRD measurements were performed using an incident wavelength of 0.43503 Å. Rietveld refinement was performed on the collected diffraction patterns to obtain the accurate lattice constants (Fig. S3 and Table S1, SI). Fig. 2a presents their expanded (011), (101), and (111) peaks. In contrast to the narrow peaks seen for the FeSe₂ stars and bars, both sphere samples show unusually broad diffraction peaks. Moreover, while the stars and bars can be deconvoluted to bulk and surface layer lattice components from slight peak asymmetry, the two sphere samples can be fitted with only a single lattice phase. The lattice variations should naturally appear as required by thermodynamics. Fig. 2b summarizes the lattice constants. The cell constants of the small spheres are more different from the other samples. The larger cell constant ratios (b/a and c/a) observed in small and large spheres are associated with higher microstrain. These lattice variations could impact the electron and ion transport properties of FeSe₂, when the crystals are applied as anode materials for sodium-ion and lithium-ion batteries. It has been shown that crystal facets and lattice distortions could enhance electrochemical performance and stability.^35–38

Fig. 2 (a) Expanded (011), (101), and (111) diffraction peaks of different FeSe₂ crystals collected using a synchrotron X-ray wavelength of 0.43503 Å. Dash lines give the resulting lattice components from the Rietveld refinement simulation. (b) The lattice constants of these FeSe₂ crystals from the Rietveld refinement simulation.

Electrical conductivity and surface area analysis

Fig. 3 presents atomic force microscopy (AFM) topography images of the FeSe₂ crystals, showing the surface morphology and the conductive AFM probe locations for electrical conductivity measurements. In the corresponding I–V curves, large spheres exhibited the highest conductivity, followed by small spheres and stars. The somewhat asymmetric I–V curve seen for the small sphere may be due to the electrode material/contact area difference. Interestingly, the bar was insulative. Clear facet-dependent electrical conductivity behaviors are observed again, which may be influential to the performance of FeSe₂-based ion batteries.

Fig. 3 (a–d) Atomic force microscopy images of the different FeSe₂ crystals. The dots represent the contacting point of a conducctive AFM probe for conductivity measurements. (e) The measured I–V curves. Instrumental current limit is 100 nA.

On the other hand, the specific surface areas of FeSe₂ crystals are linked to the particle morphology and size, which in turn determine the extent of contact with the electrolyte and the rate of ion diffusion. Particle surface area measurements have been performed (Fig. S4, SI), indicating that the specific surface area follows the order: stars (2.81 m² g⁻¹) > small spheres (2.24 m² g⁻¹) > bars (0.93 m² g⁻¹) > large spheres (0.22 m² g⁻¹). Stars and small spheres should be expected to provide more active sites and form more ion channels. This allows the electrolyte to penetrate more effectively, enhancing lithium-ion diffusion efficiency and improving energy storage capacity and stability.

Electrochemical performance of FeSe₂ crystals

To check how the different crystals affect the lithium-ion storage performance, FeSe₂ electrodes were assembled into lithium-ion batteries for electrochemical measurements (Fig. S5, SI). When cycling performance was evaluated at a current density of 0.1 A g⁻¹, small spheres exhibited a gradual capacity increase reaching 893 mA h g⁻¹ at the 40th cycle, followed by large spheres, stars, and bars (Fig. 4a). The superior capacity of small spheres can be attributed to the exposure of the {111} facets with good electrical conductivity and their comparatively higher surface area. The rate performance tests showed the same trend, with all four samples displaying generally high specific capacities at low current densities (Fig. 4b). At higher currents, bulk diffusion cannot keep pace with the imposed current, so the electrochemical response becomes dominated by surface faradaic reactions. Once the current density increased beyond 5 A g⁻¹, the contribution of ion transport became more significant. Stars, possessing the largest surface area, surpassed large spheres in capacity. When the current density was reduced back to 0.05 A g⁻¹, the capacity trend returned to its initial state with notably improved capacities. These observations indicate that slow lithiation is governed by ion insertion, whereas fast lithiation is dominated by surface faradaic processes.

Fig. 4 (a) Cycling performance of different FeSe₂ electrodes at 0.1 A g⁻¹. (b) Rate performances of FeSe₂ electrodes. (c) Galvanostatic discharge curves of small spheres for the 10th, 100th, 150th, 400th, and 1000th cycle at a current density of 1000 mA g⁻¹. (d) Capacity contribution during the lithiation reactions 1 and 2 at different cycles, with reaction 1 being the lithiation reaction and reaction 2 being the FeSe₂ conversion reaction. (e) Long-term cycling performance of FeSe₂ electrodes at 1.0 A g⁻¹. (f–i) GCD curves of FeSe₂ stars in the 3rd cycle (stage I) at a current density of 50 mA g⁻¹, and 150th cycle (stage II), 400th cycle (stage III), and 1000th cycle (stage IV) at a current density of 1000 mA g⁻¹. (j) Long-term cycling performance of FeSe₂ electrodes at 10.0 A g⁻¹.

Fig. 4c presents the discharge curves of small spheres at different cycle numbers under a current density of 1 A g⁻¹. The evolution of discharge profiles is similar for other samples (Fig. S6). In the first 100 cycles, two distinct discharge plateaus corresponding to Li⁺ intercalation (reaction 1) and conversion reaction (reaction 2) were observed. As cycling proceeds, intercalation-induced phase evolution emerges. The electron transfer processes associated with the conversion reaction progressively become the primary source of capacity. With increasing cycle numbers, structural degradation caused by the intercalation reactions gradually reduced ion transport channels and intercalation active sites, accompanying a shortening of the first intercalation plateau, leading to a continuous decline in capacity up to the 150th cycle. After 400 cycles, lithiation was dominated by the surface conversion reaction, with no observable intercalation plateau. To quantify this evolution, Fig. 4d compares the relative contributions of the two lithiation mechanisms at different cycles. The conversion reaction increased progressively from the irreversible transformation of FeSe₂. The observed reactions differ notably from the reversible transformation behavior reported in the previous studies, in which FeSe₂ reacts with Li⁺ to form Fe and Li₂Se in the discharging process (FeSe₂ + 4Li⁺ + 4e⁻ → Fe + 2Li₂Se), and the reverse reaction takes place in the charging process.^{11,13,16–18,39–41}

For long-term cycling, a current density of 1 A g⁻¹ was chosen as a milder condition (Fig. 4e). All morphologies experienced deep phase transitions with cycle number. Small spheres exhibited stable cycling performance possibly due to a combination of the exposure of {111} facets with excellent electrical conductivity and a relatively high surface area. Stars maintained stability for over 2000 cycles, benefiting from both good electrical conductivity and the largest surface area. In contrast, while large spheres possessed high electrical conductivity from exposing the {111} faces, their ultrasmall surface area constrains ion diffusion efficiency and ultimately impacts long-term stability. Insulating {011}-bound bars resulted in the worst cycling stability at 1 A g⁻¹. Stars with the best cycling stability at 1 A g⁻¹ were further studied. The cycling curve of stars was not a monotonic line, but rather exhibited an initial decline, followed by a subsequent increase before becoming stabilized. This behavior is very different from the long-cycle performance curves of FeSe₂ in lithium- and sodium-ion batteries reported in the literature.^{14,17,42–45} To probe this phenomenon, the electrochemical curves were divided into four stages: the initial activation stage, the intermediate lithiation stage (involving both intercalation and surface conversion reactions), and the long-term structural change stage. Synchrotron XRD was employed to uncover structural changes at different galvanostatic charge–discharge (GCD) potentials to further investigate the electrochemical evolution (Fig. 4f–i), as seen in the following discussion.

Additionally, long-cycle testing was conducted at an ultrahigh current density of 10 A g⁻¹ (Fig. 4j). Because the capacity at such high current densities primarily originates from surface faradaic reactions, large spheres and bars, which usually degrade after deep charge and discharge cycles under milder conditions, were still able to retain a noticeable capacity during prolonged cycling. In contrast, the radial stars appear less stable under rapid surface reactions at high currents, resulting in the lowest capacity retention in this regime. The small sphere electrode retained stable capacity even after 21 000 cycles, with a coulombic efficiency of 99.9%. The outstanding high-rate performance of small spheres toward rapid lithiation is attributed to the synergy between abundant active sites from their large specific surface area and the exposure of many highly conductive {111} facets. This balance enables long-term cycling stability.

Electrochemical reactions of FeSe₂ in the long-term cycling process

To elucidate the electrochemical reactions of FeSe₂, high-resolution synchrotron XRD measurements were performed at various charge/discharge potentials. After cycling to specific potentials, the cells were disassembled inside an argon-filled glovebox. The electrodes were extracted and vacuum-dried within the glovebox to remove residual electrolytes. The active material was carefully scraped off, finely ground, and loaded into a 0.5 mm-diameter capillary, which was then cut to an appropriate length and sealed with clay at both ends. To prevent exposure to air and moisture, the capillary was placed inside a microcentrifuge tube and hermetically sealed using a heat-sealing device. All the sample preparation steps were done inside the glovebox to minimize any reaction with oxygen and moisture.

The electrochemical lithiation and delithiation processes of the FeSe₂ star electrode were analyzed by examining structural evolution on reversibility. Ex situ XRD patterns were collected over a voltage range of 0.01 V to 3.00 V. Accurate fitting of the diffraction patterns was achieved through Rietveld refinement (Fig. 5, S7 and Tables S2–S4, SI), providing precise lattice parameters. At the initial lithiation stage with a discharging potential of 2.02 V, no FeSe₂ diffraction peaks were observed; instead, the diffraction patterns correspond to that of FeSe with identifiable peaks from the (001), (101), (111), (112), and (200) planes were obtained along with the formation of Se (Fig. S7). As the voltage decreased to 1.53 V, the FeSe diffraction peaks gradually weakened, while new peaks corresponding to the (111), (200), (220), (311), and (222) planes of Li₂Se, and the (110), (200), (211), and (220) planes of Fe, appeared. When discharged to 0.01 V, the FeSe diffraction peaks disappeared entirely, followed by the formation of highly crystalline Li₂Se along with some Fe. To check any difference between sample preparation in ambient air and in an argon-filled glovebox, the sample with the highest lithium content and best crystallinity (D 0.01 V) was selected. Fig. 5 reveals that the diffraction signals of Li₂Se and highly reactive Fe completely disappeared, leaving only Se diffraction peaks. This finding confirms that the materials are highly sensitive to moisture and oxygen, necessitating all sample preparations to be conducted within a glovebox. In the subsequent charging process, the diffraction peaks of Li₂Se and Fe vanished, while FeSe peaks reappeared, indicating a reversible reaction.

Fig. 5 Synchrotron XRD patterns at different potentials in the 3rd cycle (stage I).

Fig. 6a highlights the (001) and (101) diffraction peaks of FeSe crystals observed at different discharging and charging potentials in stage I. Their peak centers deviate slightly from lattice constant variations. Lithium-ion intercalation leads to lattice expansion, with the a- and c-axis parameters of FeSe continuously increasing during the discharging process (Fig. 6b). Then, a phase transition occurs at 0.01 V, forming Fe and Li₂Se. During the charging process, Fe and Li₂Se react to regenerate FeSe, along with some unreacted Se. As lithium ions are removed, the lattice parameters contract from C 1.87 V to C 2.22 V. With further deintercalation, more vacancies form within the lattice. These vacancies accommodate small amounts of Se²⁻ released during the reaction, which subsequently incorporate to return to the tetragonal FeSe lattice at C 3.0 V with increased lattice parameter. By contrast, the Li₂Se diffraction peak positions are more invariant throughout the four stages (Fig. S8, SI). Precise lattice parameter estimations of Li₂Se and Fe are detailed in Tables S5–S8, SI.

Fig. 6 (a) Ex situ high-resolution (001) and (101) diffraction peaks of FeSe at different potentials in the 3rd cycle (stage I). (b) The lattice constants of FeSe crystals at different potentials.

At the 150th cycle, the diffraction patterns of the electrode exhibit distinct peaks corresponding to FeSe and Fe, indicating partial reversibility of the phase transformation after multiple cycles (Fig. S9 and S10a, SI). At 0.01 V (fully discharged state), the FeSe phase disappears, while peaks associated with Li₂Se and Fe emerge, confirming the conversion reaction. Upon charging to 3.0 V, the FeSe phase is partially restored, although some Li₂Se remains, suggesting an incomplete reaction. After 400 cycles, phase reversibility is still evident, but noticeable peak broadening indicates increasing structural disorder and potential amorphization (Fig. S10b, SI). The Fe and Li₂Se peaks persist at 0.01 V along with the emergence of LiOH. The FeSe diffraction peaks at 3.0 V become weaker than in earlier cycles. Due to this decrease in crystallinity, some FeSe₂ diffraction peaks reappear, indicating that portions of the material remained unreacted in previous cycles. By 1000 cycles (Fig. S10c, SI), the diffraction patterns exhibit further intensity reduction, contributing to the capacity fade observed between 400 and 1000 cycles. Meanwhile, the persistent presence of Fe and Li₂Se peaks at 0.01 V indicates the formation of highly stable reaction products.

Reduction peak contribution analysis from cyclic voltammetry curves

Since the number of electrons transferred during the conversion reaction directly affects the electrochemical performance, reduction peak contribution in cyclic voltammetry curves was analyzed. Initially, the CV curves of small spheres (Fig. 7a) and stars (Fig. 7i) exhibit multiple reduction peaks, with peak 3 corresponding to the conversion reaction forming Li₂Se and Fe. The initial reduction contributions of small spheres and stars are 20.2 and 21.8%, respectively. For bars and large spheres, the conversion reaction at peak 3 accounts for 24.7 and 23.0% of the total reduction contributions, respectively (Fig. S11, SI). Small spheres have the lowest initial reduction peak contribution, indicating a more dominant intercalation-type lithiation process with minimal involvement of the conversion reaction. This characteristic mitigates volume expansion, thereby enhancing electrode stability and prolonging cycle life. After 150 cycles (stage II), the CV curves of small spheres and stars retain distinct reduction peaks, with the conversion reaction shifting to peak 2. The reduction contribution of small spheres increases to 57.8%, while that of stars reaches 58.3%, indicating an increasing role of the conversion reaction. Similarly, for bars and large spheres, the reduction contribution at peak 2 rises to 64.1 and 66.4%, respectively. Although the conversion reaction grows for all morphologies due to improved reaction kinetics, small spheres still maintain the lowest reduction peak contribution, meaning that small spheres continue to favor an intercalation-dominated lithiation process, delaying the onset of conversion reaction-induced degradation. As a result, small spheres preserve better structural integrity and cycling stability.

Fig. 7 Reduction peak current analysis (reduction peak percentage = (specific reduction peak current/all reduction peak current) × 100%): (a–d) CV curves of the small FeSe₂ spheres at different stages under various scan rates in the range of 0.2–1.0 mV s⁻¹. (e–h) Summary of the reduction peak percentages of the small FeSe₂ spheres at different stages. (i–l) CV curves of the FeSe₂ stars at different stages under various scan rates in the range of 0.2–1.0 mV s⁻¹. (m–p) Summary of the reduction peak percentages of the FeSe₂ stars at different stages.

By 400 cycles (stage III), the CV curves of small spheres and stars exhibit further shifts in peak intensities, with the reduction contribution at peak 2 increasing to 63.0% for small spheres and 73.3% for stars. Bars and large spheres show even higher conversion contributions of 71.8 and 69.0%, respectively. The conversion reaction becomes the more dominant lithiation process for all morphologies. However, small spheres consistently exhibit the lowest conversion contribution among the samples. This suggests that small spheres undergo a more gradual and controlled lithiation process, reducing the adverse effects of conversion reactions such as excessive volume expansion, material pulverization, and loss of electrical contact. After 1000 cycles (stage IV), the CV curves of small spheres and stars still exhibit distinguishable reduction peaks, with the conversion reaction at peak 2 remaining a dominant process. The reduction contribution of small spheres stabilizes at 58.1%, while stars reach 54.9%. By contrast, severe degradation is observed in bars and large spheres after 1000 cycles, leading to substantial capacity fading. As shown in their CV curves, the electrochemical profiles of bars and large spheres become irregular and non-smooth, indicating structural collapse and loss of electrochemical activity (Fig. S12, SI). This deterioration makes it impossible to accurately quantify the contribution of the conversion reaction. The increase in the conversion reaction contribution for small spheres, despite being the lowest among all morphologies in earlier stages, can be attributed to gradual structural evolution and reaction kinetics changes over extended cycling. In the initial cycles, small spheres favor a more intercalation-dominated lithiation, which mitigates volume expansion and maintains electrode integrity. However, after 1000 cycles, the accumulated cycling-induced phase transformations and surface reconstruction progressively activate the conversion reaction, leading to a higher reduction peak contribution. Unlike bars and large spheres, which experience severe degradation and capacity loss, small spheres maintained its electrochemical activity while allowing a more controlled transition toward the conversion reaction. This balanced lithiation mechanism contributes to the superior long-term cycling stability of small spheres. These results emphasize the crystal facet impact on long-term cycling stability, where small spheres and stars maintain better structural integrity, allowing sustained conversion reaction activity, while bars and large spheres experience severe degradation, ultimately leading to diminished electrochemical performance.

Lithiation mechanism in the FeSe₂ electrode

The lithiation mechanism of FeSe₂ electrodes was accomplished by analyzing the structural changes during the charge and discharge cycles. Fig. 8 illustrates the sequential reactions involved in Li⁺ insertion and extraction. In the initial discharge process, FeSe₂ undergoes a stepwise reduction, first transforming into FeSe before decomposing into Fe and Li₂Se. The crystal orientations are indicated in Fig. S13, SI. The Fe–Se bond length gradually increases from 2.4025(4) Å to 2.4202(19) Å, indicating lattice expansion from Li⁺ intercalation. As the voltage further drops to 0.01 V, FeSe is completely converted into Fe and Li₂Se, with a measured Li–Se bond length of 2.5986(10) Å, confirming the completion of the conversion reaction. This transformation suggests a two-step mechanism in which FeSe, Li₂Se, and Fe form an intermediate phase before the final conversion into iron and lithium selenide. Upon charging, the reverse process occurs, where Fe and Li₂Se recombine to regenerate FeSe. This reformation is evidenced by the stepwise decrease in the Fe–Se bond length from 2.4014(9) Å at 1.87 V to 2.3966(11) Å at 2.22 V, and eventually reaching 2.4218(5) Å at 3.0 V. The reappearance of FeSe at the fully charged state suggests a partially reversible reaction, as selenium is released to form FeSe after the first cycle. In subsequent cycles, the FeSe electrode undergoes repeated lithiation and delithiation, stabilizing its structure over time. Fig. 8b indicates that the Fe–Se bond length is more invariant in stages II to IV. Overall, the structural evolution confirms the FeSe₂ electrode's ability to undergo a reversible conversion reaction with potential as a high-capacity anode material for lithium-ion batteries through control of the crystal facet and surface area.

Fig. 8 (a) Schematic depiction of the Li⁺ incorporation process in the FeSe₂ electrode during the charge and discharge cycles in the early stage. The involved reactions are provided. (b) Schematic depiction of the Li⁺ incorporation process in the FeSe₂ electrode during the charge and discharge cycles at stages II, III, and IV.

Conclusions

This study demonstrates how distinct FeSe₂ morphologies affect their electrochemical performance as a lithium-ion battery anode. Crystal facets, electrical conductivity, and surface area collectively impact the long-term stability of the battery anode. A novel reaction mechanism for FeSe₂ that is clearly different from that in the previous reports was identified. Furthermore, changes in the lattice constants of the components at various lithiation stages have been identified. Cyclic voltammetry analysis further showed variations in the reduction peak contributions at different stages. This research not only reinforces the potential of FeSe₂ for energy storage applications, but also illustrates that the combination of crystal face and electrical conductivity are important considerations for anode materials.

Experimental

Chemicals

Sodium dodecyl sulfate (SDS, ≥99%, J. T. Baker), iron chloride tetrahydrate (FeCl₂·4H₂O, 98%, Alfa Aesar), selenium powder (Se, 99%, Thermo Scientific), triethanolamine (C₆H₁₅NO₃, TEOA, >99%, Sigma), hydrochloric acid (99%, 12 M, Honeywell Fluka), citric acid monohydrate (C₆H₈O₇·H₂O, Sigma-Aldrich), ethylene glycol (C₂H₆O₂, 99+%, ACROS ORGANICS), and ethanol (C₂H₅OH, 100%, Honeywell) were used as received.

Synthesis of FeSe₂ stars, bars, and spheres

The procedure and reagent amounts used to grow FeSe₂ stars, bars, and spheres are shown in Fig. S1, SI. FeSe₂ samples with different morphologies were synthesized via a solvothermal method using FeCl₂ as the iron precursor and 1.6 mmol selenium powder dissolved in 10 mL of hydrazine to prepare the Se solution. To obtain FeSe₂ stars, 2.5 mL of 0.1 M HCl solution, 0.2 mmol FeCl₂ (40.0 mg, 0.2 mmol), and 5 mL of 1 M triethanolamine were added to a reaction bottle and sonicated for 10 min, and then 2.5 mL of the Se solution was slowly added under sonication for 10 min. Next, 5 mL of ethylene glycol was introduced and stirred for 10 min. The mixture was sonicated, transferred to a Teflon-lined autoclave, and heated at 160 °C for 12 h. For FeSe₂ bars, 2.5 mL of 1 M citric acid was used instead of HCl, and sodium dodecyl sulfate replaced triethanolamine as the surfactant, with no additional solvent introduced after the selenium addition. The mixture was similarly subjected to solvothermal treatment.

Large FeSe₂ spheres were synthesized by adding 5 mL of 4 M triethanolamine and 2.5 mL of 1 M citric acid to the FeCl₂ solution before introducing the Se solution. Small FeSe₂ spheres were synthesized by reducing the concentration of triethanolamine to 1 M while keeping other reaction conditions unchanged. All the crystal growth reactions proceeded at 160 °C for 12 h. After that, the sample was purified from the mixture by centrifuging at 11 000 rpm for 5 min, and washed four times each with 95% ethanol and deionized water at 11 000 rpm for 3 min.

Electrochemical measurements

The electrode slurry was prepared by dispersing 70 wt% FeSe₂, 20 wt% multi-walled carbon nanotubes (MWCNTs), and 10 wt% sodium carboxymethyl cellulose (NaCMC) in deionized water. The homogeneous slurry was coated onto copper foil current collectors and dried at 80 °C for 1 h. The mass loading of the active material was controlled within the range of 0.7–1.0 mg cm⁻². Half cells were assembled using lithium metal foil as the counter electrode, with 1 M LiPF₆ dissolved in a 1 : 1 volume mixture of ethylene carbonate (EC) and dimethyl carbonate (DEC) as the electrolyte, and glass fiber (Advantec) as the separator. All cells were assembled in the CR2032 coin-type configuration in an argon-filled glovebox, where the concentrations of both moisture and oxygen were maintained below 1 ppm.

Instrumentation

SEM images were taken using a thermal field emission scanning electron microscope (JEOL JSM-7000F). A JEOL JEM-2100 microscope with a 200 kV operating voltage was used for TEM characterization. Powder XRD patterns were obtained using Cu Kα radiation (1.5418 Å in wavelength) on a D2 PHASER desktop diffractometer. Synchrotron XRD patterns were collected at the Taiwan Photon Source 19A (TPS 19A) high-resolution powder X-ray diffraction beamline with an X-ray energy of 28.5 keV. The sample capillary tube was rotated at 500 rpm during the measurement. A Hitachi AFM100 atomic force microscope was used for conductivity measurements. The Brunauer–Emmett–Teller (BET) surface area was measured on a Micrometritics 3Flex 3500 surface area analyzer.

Conflicts of interest

There are no conflicts to declare.

Data availability

All the data have been presented in the manuscript and supplementary information (SI). Supplementary information: FeSe₂ crystal growth conditions, XRD patterns, crystallographic data, nitrogen adsorption–desorption isotherms, illustration of the components used in lithium ion battery assembly, and cyclic voltammetry curves. See DOI: https://doi.org/10.1039/d5ta05947j.

Acknowledgements

Financial support was provided by the National Science and Technology Council of Taiwan (NSTC 113-2628-E-002-010-MY3, 113-2628-E-007-006, and 113-2811-M-007-071).

References

1. E. Fan, L. Li, Z. Wang, J. Lin, Y. Huang, Y. Yao, R. Chen and F. Wu, Chem. Rev., 2020, 120, 7020–7063.
2. X. Zeng, M. Li, D. Abd El-Hady, W. Alshitari, A. S. Al-Bogami, J. Lu and K. Amine, Adv. Energy Mater., 2019, 9, 1900161.
3. J. Xie and Y.-C. Lu, Nat. Commun., 2020, 11, 2499.
4. K. Ryu, M. J. Lee, K. Lee and S. W. Lee, Energy Environ. Mater., 2023, 6, e12662.
5. H. B. Wu, J. S. Chen, H. H. Hng and X. W. Lou, Nanoscale, 2012, 4, 2526–2542.
6. G. D. Park, J. H. Hong, J. H. Choi, J.-H. Lee, Y. S. Kim and Y. C. Kang, Small, 2019, 15, 1901320.
7. J. Feng, S.-H. Luo, S.-X. Yan, Y. Zhan, Q. Wang, Y.-H. Zhang, X. Liu and L.-J. Chang, J. Mater. Chem. A, 2021, 9, 1610–1622.
8. Y. Xiang, L. Xu, L. Yang, Y. Ye, Z. Ge, J. Wu, W. Deng, G. Zou, H. Hou and X. Ji, Nano-Micro Lett., 2022, 14, 136.
9. D. Wang, Q. Ma, H. He, Z.-Y. Wang, R.-G. Zheng, H.-Y. Sun, Y.-G. Liu and C.-L. Liu, Rare Met., 2024, 43, 2067–2079.
10. Y. Fang, Z. Chen, L. Xiao, X. Ai, Y. Cao and H. Yang, Small, 2018, 14, 1703116.
11. P. Ge, H. Hou, S. Li, L. Yang and X. Ji, Adv. Funct. Mater., 2018, 28, 1801765.
12. F. Zhao, S. Shen, L. Cheng, L. Ma, J. Zhou, H. Ye, N. Han, T. Wu, Y. Li and J. Lu, Nano Lett., 2017, 17, 4137–4142.
13. H. Wang, X. Wang, Q. Li, H. Li, J. Xu, X. Li, H. Zhao, Y. Tang, G. Zhao, H. Li, H. Zhao and S. Li, ACS Appl. Mater. Interfaces, 2018, 10, 38862–38871.
14. M. Yousaf, Z. Wang, Y. Wang, Y. Chen, U. Ali, M. Maqbool, A. Imran, N. Mahmood, P. Gao and R. P. S. Han, Small, 2020, 16, 2002200.
15. K. Zhang, Z. Hu, X. Liu, Z. Tao and J. Chen, Adv. Mater., 2015, 27, 3305–3309.
16. C. An, Y. Yuan, B. Zhang, L. Tang, B. Xiao, Z. He, J. Zheng and J. Lu, Adv. Energy Mater., 2019, 9, 1900356.
17. B. Cong, S. Sun, B. Wang, C. Lv, J. Zhao, F. Jin, J. Jia and G. Chen, Chem. Eng. J., 2022, 435, 135185.
18. S. Zhou, R. Jiang, S. Wang, L. Yu, X. Shi, L. Shao, Z. Sun and L. Hang, J. Mater. Chem. A, 2024, 12, 11028–11037.
19. W. Feng, X. Wen, Y. Peng, Y. Wang, L. Song, X. Li, R. Du, J. Yang, Y. Jiang, H. Li, H. Sun, L. Huang, J. He and J. Shi, Small, 2023, 19, 2302029.
20. L. Wang, Y. Li, B. Wang, Z. Jing, M. Chen, Y. Zhai, Z. Kong, S. Iqbal, S. Zeng, X. Sun, Y. Chen, J. Dou and L. Xu, Adv. Funct. Mater., 2024, 34, 2406988.
21. H. B. Lin, Y. M. Zhang, H. B. Rong, S. W. Mai, J. N. Hu, Y. H. Liao, L. D. Xing, M. Q. Xu, X. P. Li and W. S. Li, J. Mater. Chem. A, 2014, 2, 11987–11995.
22. W. Sun, Y. Li, K. Xie, S. Luo, G. Bai, X. Tan and C. Zheng, Nano Energy, 2018, 54, 175–183.
23. A.-T. Lee, C.-S. Tan and M. H. Huang, ACS Cent. Sci., 2021, 7, 1929–1937.
24. C.-S. Tan, H.-S. Chen, C.-Y. Chiu, S.-C. Wu, L.-J. Chen and M. H. Huang, Chem. Mater., 2016, 28, 1574–1580.
25. C.-S. Tan, Y.-J. Chen, C.-F. Hsia and M. H. Huang, Chem.–Asian J., 2017, 12, 293–297.
26. M.-S. Hsieh, H.-J. Su, P.-L. Hsieh, Y.-W. Chiang and M. H. Huang, ACS Appl. Mater. Interfaces, 2017, 9, 39086–39093.
27. P.-L. Hsieh, M. Madasu, C.-H. Hsiao, Y.-W. Peng, L.-J. Chen and M. H. Huang, J. Phys. Chem. C, 2021, 125, 10051–10056.
28. G. Kumar, C.-R. Chen, B.-H. Chen, J.-W. Chen and M. H. Huang, J. Mater. Chem. C, 2022, 10, 12125–12131.
29. C.-K. Chen, B.-H. Chen and M. H. Huang, Chem. Mater., 2023, 35, 7859–7866.
30. P.-S. Chang, B.-H. Chen, Y.-C. Lin, W.-T. Dai, G. Kumar, Y.-G. Lin and M. H. Huang, Small, 2024, 20, 2401558.
31. B.-H. Chen, G. Kumar, Y.-J. Wei, H.-H. Ma, J.-C. Kao, P.-J. Chou, Y.-C. Chuang, I.-C. Chen, J.-P. Chou, Y.-C. Lo and M. H. Huang, Small, 2023, 19, 2303491.
32. Y.-J. Chuang, A. Pal, B.-H. Chen, S. Jena, S. Suresh, Z.-H. Lin and M. H. Huang, Chem. Sci., 2025, 16, 3285–3295.
33. W. Shi, X. Zhang, G. Che, W. Fan and C. Liu, Chem. Eng. J., 2013, 215–216, 508–516.
34. S. Dong, Q. Su, W. Jiao, S. Ding, M. Zhang, G. Du and B. Xu, J. Alloys Compd., 2020, 842, 155888.
35. B. Wang, L. Wang, D. Ding, Y. Zhai, F. Wang, Z. Jing, X. Yang, Y. Kong, Y. Qian and L. Xu, Adv. Mater., 2022, 34, 2204403.
36. Y. Zhang, X. Shi, S. Zheng, Y. Ouyang, M. Li, C. Meng, Y. Yu and Z.-S. Wu, Energy Environ. Sci., 2023, 16, 5043–5051.
37. S.-S. Li, X.-L. Zhao, Y.-S. Liu, J.-J. Liu, K.-X. Wang and J.-S. Chen, Energy Storage Mater., 2023, 56, 506–514.
38. S. Jiang, Z. Jiang, C. Li, Z. Khanam, F. Wang, T. Ouyang and M. S. Balogun, Small, 2025, 21, 2408011.
39. W. Ye, K. Wang, W. Yin, W. Chai, B. Tang and Y. Rui, Electrochim. Acta, 2019, 323, 134817.
40. J. Luo, S. Bo, S. Park, B.-K. Park and O. L. Li, Carbon Lett., 2024, 34, 2421–2433.
41. Y. Tian, Z. Wang, J. Fu, K. Xia, J. Lu, H. Tang, K. Rabia, H. Chen, Z. Zhu, Q. Zhang, Y. J. Zeng and Z. Ye, Chem. Commun., 2019, 55, 10960–10963.
42. H. Fan, H. Yu, Y. Zhang, J. Guo, Z. Wang, H. Wang, N. Zhao, Y. Zheng, C. Du, Z. Dai, Q. Yan and J. Xu, Energy Storage Mater., 2018, 10, 48–55.
43. X. Wei, C. Tang, Q. An, M. Yan, X. Wang, P. Hu, X. Cai and L. Mai, Nano Res., 2017, 10, 3202–3211.
44. Q. Pan, M. Zhang, L. Zhang, Y. Li, Y. Li, C. Tan, F. Zheng, Y. Huang, H. Wang and Q. Li, ACS Nano, 2020, 14, 17683–17692.
45. T. Liang, H. Wang, R. Wang, B. He, Y. Gong and C. Yan, Electrochim. Acta, 2021, 389, 138686.

---