Effect of Atomic Layer Fluorination on the Structural and Electrochemical Properties of LiNi_0.5Mn_1.5O₄ High Voltage Electrodes

Abstract

A surge in the need for energy storage with high energy density is evident, especially given the rise of electric vehicles and hand-held electronic devices. This growing demand underscores the importance of innovating in battery technology. In this realm, lithium-ion batteries (LIBs) stand out for their impressive energy density. Notably, the spinel LiNi_0.5Mn_1.5O₄ (LNMO) material has caught the attention of LIBs due to its commendable electrochemical characteristics. This study focuses on the surface modification of disordered LNMO using the Atomic Layer Fluorination (ALF) method and investigates the influence of atomic fluorine on the interface between the LNMO and the electrolyte. Various characterization techniques, including X-ray Photoelectron Spectroscopy (XPS), Auger Electron Spectroscopy-Scanning Auger Microscopy (AES-SAM), Scanning Electron Microscopy (SEM), Nuclear Magnetic Resonance (NMR), and Electron Paramagnetic Resonance (EPR), are used to analyze the modified LNMO samples and understand the surface mechanism. Electrochemical tests reveal significant improvements in the cycling stability of fluorinated LNMO, while the rate capacity remains comparable to that of the pristine LNMO.

1. Introduction

Technological innovation is advancing rapidly, particularly noticeable in the growth of sectors such as electric vehicles (EVs) and handheld devices. This expansion is driving a vital demand for energy storage that combines high power capacity with compact form factors. A variety of battery technologies are being explored to meet this need, including lithium-ion (LIBs), lithium metal (LMBs), and lithium–sulfur (LSBs) batteries. Among these, LIBs stand out as one of the most successfully commercialized technologies. They are particularly notable for their remarkable energy density, attributed to the use of cathode materials with high specific capacities and the ability to operate at high voltages.

For instance, 5 V class spinel LiNi_0.5Mn_1.5O₄ (LNMO), thermally stable LiMn_1–x Fe _x PO₄ (LMFP), and LiNi _x Mn _y Co_1–x–y O₂ (NMC) are particularly promising contenders. Among these, spinel LNMO and its variants stand out due to their specific energy of 650 Wh kg^–1, good theoretical specific capacity (147 mA h g^–1), and high operating potential (4.7 V vs Li⁺/Li). Additionally, LNMO offers the advantages of being Co-free and cost-effective, having excellent rate performance, and having fast three-dimensional lithium-ion diffusion paths within the cubic lattice. LNMO can crystallize in two different structures: the ordered structure exists at temperatures below 700°C, with the P4₃32 space group, while the disordered structure is present at temperatures higher than 700°C, with the Fd3̅m space group. , In the P4₃32 structure, Mn⁴⁺ and Ni²⁺ ions are located at 12d and 4a sites, respectively, with no Mn³⁺ ions present. In contrast, the Fd3̅m structure has randomly distributed Mn⁴⁺ and Ni²⁺ ions at the 16d sites of the lattice. Disordered Fd3̅m shows oxygen vacancies δ, leading to a LiNi_0.5Mn_1.5O_4−δ structure with some Mn⁴⁺ ions being reduced to Mn³⁺ for charge balancing purposes. Additionally, this structure is accompanied by a rock salt impurity phase, which appears concomitantly with oxygen deficiencies within the spinel and depends on the synthesis process. This impurity phase has been attributed to Ni _x O, Li _x Ni_1–x O, or (LiNiMn) _x O. In recent studies, the disordered structure has demonstrated improvements in electronic conductivity and lithium diffusivity. Notably, it has also exhibited superior rate capability compared to the ordered structure. ,, The Mn³⁺ content in LNMO has been identified as a critical factor influencing its performance, and a reduction in Mn³⁺ content through an annealing process was observed to cause a significant decrease in rate capability.

The high-voltage spinel LNMO faces significant challenges due to electrolyte decomposition and concurrent side degradation reactions at the electrode/electrolyte interfaces. Additionally, capacity fading in LNMO/graphite full cells occurs as a result of active Li⁺ loss, which has been suggested to be linked to the Mn dissolution problem. Moreover, transition metals (Mn²⁺/Mn³⁺ and Ni²⁺) migrate toward the anode, causing detrimental effects on the solid-electrolyte interphase (SEI) and depleting available Li⁺ ions, resulting in the rapid deterioration of battery performance. Furthermore, the distribution of surface elements significantly impacts the electrochemical performance of disordered LiNi_0.5Mn_1.5O₄ due to the disproportionation reaction (2Mn³⁺ → Mn²⁺ + Mn⁴⁺) for manganese, and reducing the concentration of Mn³⁺ on the surface decreases the dissolution of Mn²⁺ in the electrolyte. Some studies have shown that the substitution of other cations for Ni²⁺ and Mn⁴⁺ on the surface has improved cycling performance by segregating the Ni ions from the surface to the bulk, resulting in a Ni-deficient surface. , This modification stabilizes the electrode/electrolyte interface and reduces the surface resistance. Surface modification of LNMO using various common materials such as ZnO, Al₂O₃, MgF₂, AlF₃, Li₃PO₄, and LiAlO₂ has been extensively studied. − The results generally indicate that the suggested protective layers applied to the surface effectively inhibit direct contact between the LNMO and the electrolyte, thereby preventing erosion of the active material, reducing the dissolution of transition metal ions, and enhancing the dynamics of Li⁺ diffusion during electrochemical cycling. A comprehensive understanding of the precise mechanisms occurring at the surface and interface of this material with the electrolyte is yet to be firmly established, as these mechanisms depend on the electrolyte formulation and the presence of additives. Most of these studies were conducted using simple electrolytes, and the electrochemical stability window of these electrolytes is usually insufficient to resist decomposition on the surface of the LNMO cathode when operating at high potentials. By adding fluoroethylene carbonate (FEC) as an electrolyte additive or cosolvent, there have been improvements in capacity retention and the formation of a cathode-electrolyte interphase (CEI) with increased inorganic species, leading to better cycle performance. −

The Atomic Layer Fluorination (ALF) method is a fluorination process based on the use of atomic fluorine radicals to modify the surface of materials in a single step at room temperature and has been successfully applied to negative and positive electrode materials to improve their electrochemical properties. − The atomic fluorine radicals are generated through the decomposition of the solid-state fluorinating agent xenon difluoride XeF₂ at room or moderate temperature. In the case of the high-voltage olivine LiCoPO₄, the presence of LiF environments at the surface of the particles proved to be an efficient alternative to the use of fluorine-containing electrolytes.

In the present study, we applied the ALF method to a disordered LNMO to improve its performance as a Li-ion positive electrode material. The main objective of this study is to modify the surface of disordered LNMO through fluorination and measure the impact of the latter on the electrochemical performance of the material. From a fundamental point of view, the surface fluorination of LNMO was investigated by ultrasurface characterization probes to fully understand the interactions between the fluorine atoms and their vicinity on the surface of the lithiated oxide particles. This is instrumental in the development of spinel-type positive electrodes in commercial Li-ion devices.

2. Experimental Section

2.1. Chemicals

The spinel compound LiNi_0.5Mn_1.5O₄ (LNMO, nanopowder, 120 nm crystal size, 7 μm average aggregate size) was obtained through SAFT from an industrial manufacturer; xenon difluoride (XeF₂, 99.99% purity, trace metal basis) and lithium metal (Li, ribbon, 99.9% purity, trace metal basis) were purchased from Sigma-Aldrich. Polyvinylidene difluoride (PVDF, solef 5130) was purchased from Solvay, and conductive carbon additive C-NERGY Super C65 was procured from Imerys Graphite & Carbon, Switzerland Ltd. One M LiPF₆ in ethyl methyl carbonate and fluoroethylene carbonate (EMC/FEC 7:3 vol %) electrolyte with a purity of 99.9% (H₂O < 0.5 ppm, O₂ < 0.5 ppm) was purchased from Solvionic.

2.2. Atomic Layer Fluorination (ALF) Conditions

The fluorination of LiNi_0.5Mn_1.5O₄ (500 mg, 2.73 mmol) was done in a PTFE reactor vessel placed in an argon-filled glovebox (O₂ < 0.5 ppm, H₂O < 0.5 ppm). An 8 wt % (40 mg) amount of XeF₂ was introduced in a small PTFE crucible, itself placed inside the reactor vessel previously filled with LNMO. Subsequently, the reactor was tightly closed, left undisturbed at room temperature for 0.5 h (LNMO-F1), 1.5 h (LNMO-F2), or 3.5 h (LNMO-F3), and finally opened. LNMO-F0 represents the pristine LNMO, not submitted to the ALF process. The fluorination of LNMO was achieved by carefully controlling the decomposition of XeF₂ based on the reactions

$${\mathrm{X}\mathrm{e}\mathrm{F}}_2(\mathrm{s})={\mathrm{X}\mathrm{e}\mathrm{F}}_2(\mathrm{g})$$ 1

$${\mathrm{X}\mathrm{e}\mathrm{F}}_2(\mathrm{g})=\mathrm{X}\mathrm{e}(\mathrm{g})+2{\mathrm{F}}^{·}$$ 2

$${2\mathrm{F}}^{·}={\mathrm{F}}_2(\mathrm{g})$$ 3

XeF₂ has a vapor pressure of 3.8 mmHg at 25 °C and 318 mmHg at 100 °C and can easily be decomposed by heating, UV exposure, or contact with reactive materials, resulting in a constant release of atomic fluorine (F^•) due to the low saturating vapor pressure of XeF₂. ,

2.3. Surface Characterization Techniques

2.3.1. X-ray Photoelectron Spectroscopy (XPS).

XPS studies were conducted by using the ESCALAB 250 Xi spectrometer with a monochromatized Al Kα X-ray source (hν = 1486.6 eV). A spot size of 650 μm was used to analyze the samples by employing 20 eV as the pass energy and 0.1 eV as the energy step for the core peaks. An electron flood gun was used to minimize the surface charging. For both the pristine and fluorinated powders, spectral calibration was carried out using the C–C or C–H component from the surface carbon contamination in the C 1s peaks (285.0 eV). The peak fitting process was carried out using a Shirley-type background.

The peak positions and areas were optimized by the weighted least-squares calculation method using GL (30) (70% Gaussian and 30% Lorentzian line shapes). The quantification was carried out by using CASA XPS software. The apparatus is directly connected to a glovebox, allowing us to transfer the powders under an argon inert atmosphere with oxygen and moisture under 0.5 ppm. The quantification of atoms for surface composition was based on Scofield’s relative sensitivity factors.

2.3.2. Scanning Electron Microscopy (SEM), Auger Electron Spectroscopy (AES), Scanning Auger Microscopy (SAM), and Cross Sections (cp)

Secondary electron microscopy (SEM), Auger electron spectroscopy (AES), and Scanning Auger microscopy (SAM) were performed using a JEOL JAMP 9500 F Nano-Auger spectrometer (JEOL Ltd., Tokyo, Japan) equipped with a Schottky Field Emission gun and a hemispherical analyzer coupled with a high dynamic multichannel detector (7 Chaneltron detector). The operating pressure was <2 × 10^–7 Pa, and Auger analyses were performed at 30° sample tilt or at a higher angle to prevent the charging effect. For Auger analysis, the electron beam energy was fixed at 10 keV, and the current was in the range of 2–10 nA, corresponding to a spatial resolution of about 20–30 nm in terms of probe diameter. The probing depth is about 2–3 nm for the kinetic energy in consideration. AES survey spectra were recorded with a 1 eV step size using a focused probe and a CRR acquisition mode (Constant Retard Ratio) with a constant relative energy resolution dE/E = 0.5% (high sensitivity). Auger spectra are presented using the EN­(E) mode expressed as the output signal of the electron detector using “true” pulse counting versus the kinetic energy. SAM (elemental 2D distribution images) images were recorded by using either the CAE mode (constant analyzer energy) corresponding to a fixed and adapted energy resolution or the CRR mode, depending on the required outlet energy width. SAM images are represented using the “peak minus background” (P–B) Auger intensity definition in the so-called “simultaneous” mode, which allows the user to define any detector position as a peak of a background event. In addition, this mode can extract from any area of a recorded image the spectroscopic information spread over seven detectors. An “auto probe tracking” correction was also applied every sweep of the data accumulation to control and compensate for any potential drift during acquisition. To analyze in depth the fluorinated LNMO particles, the cross sections were obtained using a JEOL IB-09010CP cross polisher dedicated to ion milling processes (90° surface to cross-section angle). The fluorinated particles were dispersed in a silver epoxy resin matrix microparticle slurry deposited on the side of a silicon wafer. The resin is polymerized after some hours to enshrine the LNMO-F3 particles in this solid and conductive matrix and dried for 8 h prior to cross-sectioning. The wafer and the matrix were then cut by an argon ion beam of 6 kV for 4 h with a depth of 100 μm relative to the sample side.

2.3.3. Scanning Electron Microscopy (SEM)

Scanning electron microscopy (SEM) images were acquired on an FEI Nova NanoSEM 450 FE-SEM. Sample preparation was made by placing a small amount of powder sample onto carbon tape.

2.4. Bulk Characterization Techniques

2.4.1. X-ray Diffraction (XRD)

X-ray diffraction (XRD) analyses were carried out by using a PANalytical Empyrean diffractometer and a Cu Kα X-ray source. XRD was performed on powder samples, and the fluorinated powders were prepared in the glovebox (O₂ < 0.5 ppm, H₂O < 0.5 ppm). The samples were handled with a special sample holder for air-sensitive materials, which was made of Kapton foil and was provided by PANalytical. The analysis was performed over a 2-theta range of 10° to 80° with a step size of 0.0167° and a counting time of 58 s per step.

2.4.2. Nuclear Magnetic Resonance Spectroscopy (NMR)

¹⁹F solid-state nuclear magnetic resonance (NMR) spectra were recorded by using a Bruker AVANCE III 700 MHz spectrometer with a 16.4 T superconducting magnet, operating at a frequency of 658.88 MHz for the ¹⁹F nucleus. Approximately 20 mg of the sample was placed in a 3.2 mm zirconia rotor equipped with Vespel caps, which was spun to 20 kHz (or 30 kHz to discriminate spinning sidebands) at the magic angle in a double resonance 3.2 mm HX low-gamma MAS probe head. The ¹⁹F π/2 pulse length of 2.5 μs was used with a recycle delay of 3 s to ensure complete signal relaxation. The ¹⁹F chemical shifts were referenced to those of CFCl₃. The NMR spectra were processed using the DMFIT software.

2.4.3. Electron Paramagnetic Resonance Spectroscopy (EPR)

The samples were evaluated by using Electron Paramagnetic Resonance (EPR) on a Bruker EMX spectrometer at room temperature. The EPR examination was performed using a 5 mm diameter quartz tube and an X-band spectrometer operating at a frequency of 9.653 GHz within a resonant cavity ER 4105 DR.

2.5. Electrochemical Characterization

Electrochemical properties of the electrodes were measured versus lithium metal using a coin cell 316L stainless steel CR2032. The electrodes were prepared with 90 wt % of LNMO-Fx (x = 0, 1, 2, 3), 5 wt % of carbon black C65, and 5 wt % of PVdF binder dispersed in N-methyl-2-pyrrolidone (NMP). First, the powders were dry-mixed with a Kakuhunter planetary mixer (SK-300SII, SHASHIN KAGAKU). Then, NMP is added and mixed with the Kakuhunter. The obtained slurry was then cast on aluminum foil at 80 °C by using a tape casting coater (MTI Corp., MSK-AFA-HC100). The formulated electrode was subsequently dried at 80 °C for 24 h under a dynamic vacuum (KNF LABOPORT N 820 G pump and a France Etuves XFL020 oven). The dried electrodes were cut into disks 12.7 mm in diameter. Coin cells were assembled in an Ar-filled glovebox using Celgard 2325 polypropylene membrane as a separator, LiPF₆ 1 M EMC: FEC 7:3 as the electrolyte, and Li metal as the counter/reference electrode. The cells were cycled between 3.5 and 4.9 V and recorded at C/10 (i.e., a full theoretical discharge in 10 h). The theoretical capacity of LNMO is 147 mA h g^–1. It is worth noting that the calculation of the theoretical capacity of LNMO-Fx (x = 0, 1, 2, 3) does not consider the contribution of fluorine in the chemical formula of LNMO, as its mass is negligible. Therefore, fluorine is not considered to be a significant factor in determining the overall capacity of the battery.

3. Results and Discussion

3.1. Synthesis

The synthesis of LNMO-Fx (x = 1, 2, 3) is performed by placing LNMO in the presence of XeF₂. Xenon difluoride decomposes in several steps to form highly reactive atomic radical fluorine species that will readily react with any chemical element. Eventually, fluorine radicals could react on their own to form molecular fluorine. After a reaction time of 0.5, 1.5, or 3.5 h for what we tagged as the ALF, or ALF, process, it is possible to isolate LNMO-F1, LNMO-F2, and LNMO-F3, respectively. When considering the mass uptake of the LNMO powder at the end of the reaction, the experimental amounts of fluorine per mole of LNMO are 0.011, 0.031, and 0.077 mmol for LNMO-F1, LNMO-F2, and LNMO-F3 (Table ). The fluorine amount increasing with the reaction time is a rather logical feature of the ALF process, as already described previously. ,

1. Results of Mass Uptake after Fluorination of LNMO.

	F wt %		F mol %
	theoretical	experimental	theoretical	experimental
LNMO-F0	0	0	0	0
LNMO-F1	1.76	0.12	17.3	1.15
LNMO-F2	1.76	0.32	17.3	3.07
LNMO-F3	1.76	0.80	17.3	7.70

The amount of fluorine that could theoretically react with LNMO (“Theoretical” columns in Table ) is only limited by the quantity of xenon difluoride introduced in the reactor (8 wt %). Experimentally, the reaction time of the experiment is the limiting step. At that stage, it is not possible to determine the dispersion of the fluorine atoms in the obtained compounds. Indeed, xenon difluoride is known to decompose into atomic fluorine (as described in the Experimental Section of the present article), which is highly reactive and can diffuse easily within the material to fluorinate, in the present case of LNMO. This means that xenon difluoride, hence atomic fluorine, could modify both the surface and the bulk volume of LNMO.

3.2. EDX

The results of the EDX analysis presented in Figure indicate that all of the elements of LiMn_1.50Ni_0.5O₄ (LNMO) are present on the surface of LNMO-F0, except for lithium (Li), which was not detected. The atomic percentages of these elements detected on the surface are presented in Table , revealing a distinct composition on the surface compared to that of the bulk. Notably, there is an observable rise in the atomic percentage of manganese, calculated by comparing its fraction to that of nickel, which has increased from 3 to 3.50. The significant disparity between surface and bulk compositions is underscored by the existence of LiMn_1.5Ni_0.5O₄ in bulk material.

1.

EDX mapping of Ni, Mn, and O elements in (A,C) LNMO-F0 and (B,D) LNMO-F3 samples.

2. Atomic Percentages of Ni, Mn, and F for LNMO-F0 and LNMO-F3 from EDX.

atomic %	LNMO (theoretical)	LNMO-F0	LNMO-F3
O	57.1	69.9	62.6
Mn	21.4	23.4	22.5
Ni	7.1	6.7	6.1
F	-	0	8.7
Mn/Ni	3	3.5	3.7

Upon comparison with the fluorinated sample LNMO-F3 (3 h 30 min), as illustrated in Figure S1 where both spectra are superimposed, an alternative chemical composition becomes apparent. In this specific composition, there is an elevation in the manganese content from 3.50% to 3.69%, accompanied by a reduction in the atomic proportions of nickel. The introduction of fluorine onto the surface of LNMO exerts a noteworthy influence on the manganese concentration at the surface. Furthermore, the results confirm the successful ALF process with well-distributed fluorine on the surface of particles, without severe aggregation and/or agglomeration.

3.3. XPS

XPS analyses were carried out on pristine (LNMO-F0) and fluorinated LNMO particles in powder form before any electrode formulation to characterize the electronic structure of their surfaces and consequently the impact of the fluorination on the different LNMO materials. First, we will discuss the XPS spectra recorded for LNMO-F0. The C 1s spectrum exhibits a main contribution (C–C/C–H) at 285.0 eV. This signal is primarily attributed to surface contamination by adventitious carbon, which is commonly observed in XPS due to unavoidable exposure to air during sample transfer. Oxidized carbon (C–O/O–CO) at 286.5 and 289.0 eV (Figure S2a). The O 1s spectrum is decomposed into three components assigned to the oxygen of the LNMO network (529.8 eV; Figure S2a) and oxidized species at 532.2 and 533.9 eV. The Li 1s core peak presents a unique component at 54.1 eV, which corresponds to the environment of Li in LNMO-F0 (Figure ). The Mn 3p (49.8 eV) and Ni 3p (67.3 eV) binding energies indicate that the manganese is +4 and the nickel is +2 in LNMO-F0. The energy splitting of the spin multiplet in the Mn 3s spectrum (ΔMn 3s) is highly sensitive to the oxidation state and to the environment of the Mn. In our case, the experimental value of ΔMn 3s corresponds to 4.5 eV (Figure S2), confirming the exclusive presence of Mn⁴⁺ on the surface. The lack of satellite peaks in the Mn 2p core peaks (not shown here) confirms that manganese cannot be in the +2-oxidation state. The O/(Mn + Ni) ratio is close to 2.2, instead of 2 in the nominal composition.

2.

XPS spectra Mn 3p (blue), Li 1s (yellow), Li–F (cyan), Ni 3p (purple), Na 2s (gray), and F 1s (green) acquired for LNMO-Fx (x = 0, 1, 2, 3) samples.

The spectra of the LNMO-Fx series are shown in Figures and S2a. Important changes related to fluorination can be identified as changes in B.E. values and the apparition of new environments in F 1s, O 1s, Li 1s, Mn 3p, and Ni 3p core peaks.

Detection of F 1s core peaks on all fluorinated materials confirms the incorporation. All the F 1s spectra present two components attributed to two different fluorinated environments. The major component located at 685.0 eV is due to M-F fluorination (with M = Li). The minor one located at 686.8 eV is related to inserted fluorine in the LNMO network (associated with the oxyfluoride environment) with a constant contribution of 20% of the whole signal of the F 1s whatever the duration of the fluorination. We can underline that the ratio of F_tot/(Ni + Mn) increases significantly for LNMO-F2 and F3 (4.2 and 5.1, respectively) compared to LNMO-F1, indicating an increased concentration of fluorine atoms at the LNMO surface with an ALF duration higher than 1.5 h (Table ).

3. Surface Chemical Composition of the Pristine and Fluorinated LNMO-Fx (x = 0, 1, 2, 3) Samples Extracted from XPS.

	surface composition	O 1s	O 1s	F 1s	Li 1s
sample	(excluding Li–F)	(O2– + O x–)/(Ni + Mn)	O x–/O2‑	Ftot/(Ni + Mn)	% Li – F (Li – F)/Litot
LNMO-F0	Li2.24Ni0.21Mn0.75O2.15	2.2	0	0	0
LNMO-F1	Li2.01Ni0.18Mn0.75O2.14F0.14	2.3	0.1	0.7	31
LNMO-F2	Li1.75Ni0.17Mn0.75O1.53F0.75	1.7	0.4	4.2	70
LNMO-F3	Li1.56Ni0.21Mn0.75O1.59F1.09	1.6	0.4	5.2	80

After fluorination, a new component in the Li 1s spectra appears at 55.7 ± 0.1 eV beside the Li contribution of the LNMO structure. This new peak is attributed to Li–F, and its contribution given by the ratio (Li – F)/(Li – F+ Li_LNMO) increases from 31% for LNMO-F1 to 80% for LNMO-F3 of the whole Li 1s signal. However, the detection of elements associated with LNMO particles even after a long fluorination time implies that this Li–F layer is not homogeneous and/or has a thickness lower than the XPS analyses depth (<10 nm). Furthermore, the decrease of the Li_LNMO/(Mn + Ni) ratio from 2.3 for LNMO-F0 to 1.6 for LNMO-F3 associated with the augmentation of the Li–F confirms the formation of the fluorination layer; the same trend was observed previously.

The fluorination mode and associated electronic structure changes on LNMO-Fx surfaces were studied by following a rigorous decomposition strategy. Indeed, our decomposition strategy is based on the use of the experimental envelope resulting from the desummation of the Mn 3p and Ni 3p spectra of the pristine material LNMO-F0 to decompose the rest of the LNMO-Fx (x = 1, 2, 3) series. The contribution of the experimental envelopes is summarized in Table S1. We observed that the surface fluorination induces changes in the shape of the Mn 3p, Ni 2p, and O 1s spectra, particularly from a fluorination of 90 min.

The decomposition of the Mn 3p spectra consists of inserting the experimental envelope of the Mn 3p for the bare LNMO-F0, and adding the components is necessary to obtain a correct fitting. More specifically, after 1.5 h of fluorination, it is necessary to include a component at the energy of 48.8 eV, characteristic of the Mn³⁺ oxidation state (appears with a shoulder at lower binding energies), and a third component at approximately 52.2 eV. This last contribution can easily be relied on in an oxyfluoride environment of manganese (Mn–O _x ···F _y ), which is consistent with higher binding energies and the environment detected by 686.6 eV of the F 1s core peaks after fluorination. Nevertheless, we cannot exclude the presence of Mn³⁺ in the LNMO-F1, but in too weak a proportion to be detected, as we are identifying an oxyfluoride environment for this material in the F 1s spectrum.

For both LNMO-F2 and LNMO-F3, the ratio Mn³⁺/(Mn³⁺ + Mn⁴⁺) is constant, and Mn³⁺ represents 9% of the whole signal of the Mn 3p. The ΔMn 3s multiplet splitting values are increasing to reach 5.0 eV for the LNMO-F2 and LNMO-F3 (Figure S2), which confirms the partial reduction of the manganese ions and a probable partial alteration of their coordination environment, transitioning from exclusive oxide coordination to an oxyfluoride environment. The same strategy was used to track the impact of fluorination on Ni in LNMO materials. After fluorination, a new contribution at 70.1 eV must be inserted for LNMO-F2 and LNMO-F3 in addition to the envelope of the LNMO-F0 Ni 2p spectrum and may be attributed to Ni–O _x ···F _y . This latter remains nearly constant at 10% (given by the ratio Ni–O _x ···F _y /(Ni²⁺ + Ni – O _x ···F _y )) for higher fluorination time. Alongside the partial reduction of manganese on the LNMO surface, we detect in the O 1s spectra the appearance of a new environment located at 531.2 eV (Figure S2b). This can be assigned to an oxidized oxygen O ^x– (x < 2) resulting from the oxidation of the O^2– ions due to fluorination. Such oxidized oxygen species have been previously observed in Li-rich cathode materials, where oxygen oxidation contributes to charge compensation without necessarily leading to O₂ gas release. The stabilization of these oxygen species is closely linked to the local electronic structure and the interactions with transition metal cations such as Mn⁴⁺ and Ni²⁺, preventing irreversible oxygen loss and structural degradation.

In conclusion, we emphasize a surface redox mechanism that would involve manganese and oxide ions. The ALF method induces on LNMO surfaces the oxidation of the oxide anion, which results in an increase of the O ^x–/O^2– ratio. The contributions of oxidized oxygen (O ^x–) and of Mn³⁺ (9%) are similar for LNMO-F2 and LNMO-F3. The similar contributions of oxidized oxygen (O ^x–) and Mn³⁺ (9%) in LNMO-F2 and LNMO-F3 further confirm that fluorination triggers a surface redox process, leading to the stabilization of oxidized oxygen within the material rather than its release as molecular O₂.

3.4. AES and SAM

To investigate the fluorine diffusion in depth and its spatial distribution, AES (Auger Electron Spectroscopy) and SAM (Scanning Auger Microscopy) analyses were conducted on the LNMO-F3 particles both before (Figure a) and after cross-sectioning (cp) (Figure d). The morphology of the LNMO-F3 particles reveals a pseudospherical shape, consisting of an aggregate of several multifaceted plates (flake-like).

3.

SAM of the LNMO-F3 particle before (a) and after cross section (d). The AES analysis of the LNMO-F3 particle before (b) and after being cut by cross polishing (e); SAM mapping of Mn LMM, Ni LMM, and F KLL Auger transition of LNMO-F3 (c,f).

The AES spectra for two different areas of interest of a single particle are shown in Figure b. The first spectrum was acquired on platelets (the flakes), while the second spectrum was acquired between the platelets. Both AES spectra exhibit the same chemical composition, confirmed by the presence of Auger signals associated with the LNMO-F3: O KLL at K.E. = 482 and 503 eV, the Mn KLL at 528, 582, and 631 eV, and the fluorine KLL transition at 651 eV. The presence of carbon and potassium impurities is confirmed by the Auger signals for K KLL (243 eV) and C KLL (263 eV), respectively. Note that the recorded intensity of both spectra is modulated by the flake orientation. Figure c presents the chemical mappings based on the previous spectra recorded by scanning Auger Microscopy. Chemical maps offer a more convenient means to visualize the distribution of elements. A uniform dispersion of Mn, Ni, and F can be concluded. Indeed, the contrast shown in the mapping on the top of the particle depends on the flakes’ orientation, as previously reported, and the shadowing effect observed on the bottom of the particle is attributed to a geometric effect arising from the positioning of the primary beam direction as regards the hemispherical electron analyzer angle (30°) Auger spectroscopy detector.

The cross-section reveals exposed particles with their initial morphology preserved (Figure S3). An area (highlighted by the yellow square) for scanning spectra inside the half LNMO-F3 particle (Figure d) confirms the presence of the LNMO structure, as indicated by the Auger signals observed across the entire particle (Figure e), except the fluorine spectrum, which is totally absent inside the particle. This confirms that fluorination affects only the extreme surface of our material. Several cut particles were examined, and none of them showed the presence of fluorine within. In addition to the elemental mapping (Figure f) of Ni and Mn, the F mapping in the red square zone provides compelling evidence. The presence of fluorine is exclusively observed at the periphery, forming a shell-like fluorinated layer.

3.5. XRD

Figure illustrates the X-ray diffraction analysis of LNMO-Fx (x = 0, 1, 2, 3), which reveals diffraction peaks that correspond to the cubic spinel crystal structure of disordered LiNi_0.5Mn_1.5O₄ (space group Fd3̅m and lattice parameter 8.16 Å, ICSD 259947, PDF#2 80-2162). The XRD patterns of the fluorinated samples display the same peak positions as the bare material (for instance, hkl 111, 113, and 004 at 17°, 36°, and 44° 2θ). It suggests that the fluorine on the surface of the LNMO particles has not altered the spinel structure pattern. The atomic layer fluorination process using XeF₂ did not noticeably affect the original crystal structure or its structural parameters, as expected from XPS that probes the ultrasurface of a particle, hence a thickness of approximately 10 nm. The structural parameters obtained from profile matching based on the obtained XRD data are listed in Table S2.

4.

X-ray diffraction patterns of LNMO-Fx (x = 0, 1, 2, 3) powder samples.

3.6. SEM

The morphology and particle size distribution of LNMO-F0 and LNMO-F3 analyzed by SEM are shown in Figure . All samples, with or without surface fluorination, exhibit uniformly distributed spherical aggregates with a size of 6.8 μm, as illustrated in Figure a,c. The LNMO-F0 sample consists of particles with octahedral crystal facets comprising a family of (111) planes.

5.

SEM images of LNMO-F0 (a,b) and fluorinated LNMO-F3 (c,d).

After 3.5h of fluorination, we observed that the ALF process did not affect the micromorphology of the LNMO particles, as evidenced by their unchanged appearance. However, the surface of the fluorinated LNMO-F3 particles displays small aggregates, indicating that only the particle surfaces were modified without altering their underlying microstructure. These features have already been spotted in fluorinated LCP materials.

3.7. NMR

Figure displays the solid-state ¹⁹F NMR spectra of different samples: fluorinated LNMO-Fx (x = 0, 1, 2, 3), 9 wt % LiF in diamagnetic NaCl (to avoid paramagnetic interaction with ¹⁹F nuclei), and an empty rotor for comparison. The ALF method results in low fluorine content, and NMR, contrary to XPS, is a bulk-sensitive technique. This explains the low signal-to-noise ratio (S/N) and why the dilution of LiF in paramagnetic NaCl as well as the recording of the empty rotor are necessary for comparison purposes. First, a slight bump is observed between −100 and −300 ppm, originating from the CF₂ environment of the rotor cap. The intense peak of LiF was observed at around −204 ppm, gives the reference for such an environment. For fluorinated LNMO materials, numerous spinning sidebands are present, which complicates the recording in addition to a low F content. The extent of the spinning sidebands is higher for fluorinated LNMO than for diluted LiF in accordance with the presence of paramagnetic transition metals (Mn, Ni). A comparison of the spectra of LiF and the three fluorinated LNMO-Fx (x = 1, 2, 3) reveals the emergence of a small band at around −204 ppm, potentially resulting from the left-shifted Li–F bonding environment due to paramagnetism. Both the low S/N ratio (due to the content of F atoms) and the numerous spinning sidebands exclude any quantification, but the intensities of the isotropic line of LiF and all its spinning sidebands are significantly higher for LNMO-F3.

6.

¹⁹F solid-state NMR spectra of LNMO-Fx (x = 0, 1, 2, 3) samples, LiF, and empty rotor.

3.8. EPR

The EPR spectra of LNMO-F0 and fluorinated samples were recorded at room temperature for varying fluorination durations, as shown in Figure . All samples exhibit a wide signal (with a peak-to-peak line width ΔH_PP of about 1000 G) centered at around 3540 G. The g-factor of close to 2.0 is in good accordance with the presence of Mn⁴⁺ ions ([Ar] 3d³). It should be noted that Mn³⁺ and Ni²⁺ ions ([Ar]­3d⁴ and [Ar]­3d⁸, respectively) are EPR-silent. Minor changes are recorded for the EPR parameters as a function of the fluorination temperature. The Landé g-factor shifts from 2.0046 ± 0.0002 to 2.0016 ± 0.0002 after fluorination, whatever the duration. The changes in ΔH_PP are quite continuous with the fluorine content. However, the intensities increase for LNMO-F1 and LNMO-F2 in comparison with the pristine sample; the spectrum of LNMO-F3 exhibits a decrease because of its narrowing. To quantify the increase of Mn⁴⁺ in the samples, we took LNMO-F0 as the reference. Using the integrated signal areas detailed in the Supporting Information (illustrated in Figure S4), we extracted the percentages of Mn⁴⁺ increase: 4.55% for LNMO-F1, 6.92% for LNMO-F2, and 11.7% for LNMO-F3. Interestingly, when comparing these percentages with the weight percentages of fluorination, a consistent trend emerges, suggesting a very good accordance between fluorination levels and the percentage of Mn⁴⁺ increase due to Mn³⁺ into Mn⁴⁺ conversion. Those changes with fluorination may be related to the surface oxidation of Mn³⁺ into Mn⁴⁺ in accordance with XPS data, which evidence the exclusive presence of Mn⁴⁺ on the surface. The reduction of Jahn–Teller Mn³⁺ ion concentration and the amplified network distortions experienced by Mn⁴⁺ ions bring about a decline in the peak-to-peak width ΔHpp (illustrated in Table ), while the intensity increases, and this might indicate an increase in network rigidity.

7.

EPR spectra of LNMO-Fx (x = 0, 1, 2, 3) samples.

4. Summary of the g-Factor and ΔHpp Values of LNMO Samples.

samples	LNMO-F0	LNMO-F1	LNMO-F2	LNMO-F3
g-factor	2.0046 ± 0.0002	2.0015 ± 0.0002	2.0017 ± 0.0002	2.0017 ± 0.0002
ΔHpp (G)	999 ± 5	986 ± 5	988 ± 5	936 ± 5

The decrease in ΔHpp when the content of Mn⁴⁺ increases (from LNMO-F1 to LNMO-F3) may be explained by two phenomena. The first is related to an increase in the rigidity of the lattice upon decreasing the concentration of the Jahn–Teller Mn³⁺ ions. It was evidenced for defected MnO₂ (defects caused by the presence of Mn³⁺ ions in the analogous positions of the MnO₂ lattice stabilized by OH^– groups or other compensators), that the partial substitution of tetravalent manganese ions for trivalent ones modifies the character of the dipole–dipole interaction and exchange. For defected MnO₂, a linear dependence of the line width on the Mn⁴⁺/Mn³⁺ concentration ratio is observed. ΔHpp decreased with Mn⁴⁺/Mn³⁺ as in our case. During fluorination, Mn⁴⁺ ions substitute Mn³⁺ ones. When a compound contains ions of the same element in equivalent positions of their lattices but with different valence, the so-called Zener “double exchange” occurs. An exchange with an electron takes place between a Mn⁴⁺ cation and a Mn³⁺ cation separated by an O₂ ^– anion. Our data with the decrease of the line width according to the Mn⁴⁺ content seem to reveal the “double exchange” interaction.

The changes in the line width are higher for LNMO-F3, evidencing the stronger effect of the fluorination for the longer duration, as seen by a very sensitive technique such as EPR, especially for the oxidation state and local environment of the paramagnetic species. The oxidation of Mn³⁺ to Mn⁴⁺ during fluorination is not directly caused by lithium extraction but rather results from charge redistribution due to fluorine incorporation. The substitution of oxygen by fluorine induces local modifications in the oxidation state of manganese to maintain charge neutrality. While some Li⁺ extraction occurs to form LiF and stabilize the fluorinated structure, it is not the primary driver of Mn³⁺ oxidation. The content of LiF in LNMO-F3 appears as the highest in the series considering ¹⁹F NMR data.

3.9. Electrochemical Tests

To evaluate the battery performance of fluorine-surface-coated LiNi_0.5Mn_1.5O₄ (LNMO), galvanostatic charge–discharge tests were conducted at a C/10 rate. The initial charge/discharge profile is presented in the capacity versus voltage curves in Figure a, with the fifth cycle also depicted in the same figure. Upon charging, when the potential reaches approximately 4 V, a plateau is observed, characteristic of the Mn^3+/Mn⁴⁺ redox couple in the disordered structure of the LNMO, as noted in previous studies. Subsequently, two additional plateaus emerge at around 4.7 and 4.75 V, corresponding to the Ni²⁺/Ni³⁺/Ni⁴⁺ redox couples, with Ni³⁺ as a transient species.

8.

Galvanostatic voltage profiles for LNMO-F0, LNMO-F1, LNMO-F2 and LNMO-F3 half-cells cycled at C/10 between 3.3 and 4.9 V versus Li⁺/Li; the first cycle (a) and fifth cycle (b) are displayed.

A comparative analysis of the charge curves between the uncoated LNMO-F0 in black and the fluorinated LNMO across different durations for the initial cycles reveals that the Mn³⁺/Mn⁴⁺ redox plateau of fluorinated LNMO, especially for LNMO-F3, is reduced by approximately 8 mA h g^–1. This reduction suggests a decrease in the Mn³⁺ concentration. To understand this phenomenon better, we correlated these findings to our EPR analysis. The integrated EPR signal areas indicated an increase in Mn⁴⁺ concentration of about 11.7% in fluorinated LNMO, notably in LNMO-F3. This increase in the Mn⁴⁺ signal aligns with the electrochemically observed reduction in the Mn³⁺ concentration.

The lower capacities observed for LNMO-F1 and LNMO-F2 compared to those of LNMO-F0 can be explained by the incomplete fluorination at shorter reaction times. In these cases, the insufficient incorporation of fluorine may lead to a nonuniform LiF layer on the surface, which increases interfacial resistance and limits lithium-ion diffusion, ultimately affecting the overall electrochemical performance. On the other hand, LNMO-F3, with a longer fluorination duration, benefits from a more uniform and continuous fluorine-enriched surface. This optimized layer enhances the stability of the electrode–electrolyte interface while preserving efficient lithium-ion transport, leading to improved capacity retention.

Additionally, the discharge curves show a shortened Ni²⁺/Ni³⁺ plateau for LNMO-F3, which exhibits a marginally higher capacity, approximately 135 mA h g^–1, and reduced polarization compared to the unmodified LNMO. This alteration indicates a more efficient Ni²⁺/Ni³⁺ redox process postfluorination. After 5 cycles, as shown in Figure b, a shortening in the Ni³⁺/Ni⁴⁺ plateau is noted for the fluorinated LNMO. However, LNMO-F3 still retains good performance.

To further analyze the Mn oxidation states in LNMO-F0, we calculated the fraction of Mn³⁺ based on the capacity of the plateau at ∼4 V in the first galvanostatic cycle (Figure a). This plateau corresponds to the Mn³⁺/Mn⁴⁺ redox reaction. Using the total theoretical capacity of LNMO (147 mA h g^–1) and the measured capacity of the Mn³⁺/Mn⁴⁺ plateau (40 mA h g^–1), we estimate that approximately 41% of manganese in LNMO-F0 is in the Mn³⁺ oxidation state. This estimation suggests that about 41% of manganese in LNMO-F0 exists as Mn³⁺ in the bulk. These findings align with the expected stoichiometry of disordered LNMO (Fd3̅m), where oxygen vacancies induce a partial reduction of Mn⁴⁺ to Mn³⁺ for charge compensation. However, this bulk estimation contrasts with the XPS analysis, which reveals a surface dominated by Mn⁴⁺, further emphasizing the distinct chemistry of the surface and bulk.

The capacity retention values after 60 cycles were 92.3% for LNMO-F0, 94.4% for LNMO-F1, 94.3% for LNMO-F2, and 91.7% for LNMO-F3, as shown in Figure . Additionally, the Coulombic efficiencies of LNMO-F0, LNMO-F1, LNMO-F2, and LNMO-F3 were found to be 98.6%, 98.5%, 98.6%, and 98.6%, respectively. Despite the slightly lower capacity retention of LNMO-F3 compared to LNMO-F1 and LNMO-F2, its consistently high Coulombic efficiency suggests a stable electrode/electrolyte interface and minimal side reactions. This indicates that the fluorination process contributes to maintaining long-term cycling stability while ensuring efficient lithium-ion transport. It should be noted that the data presented in Figures and were obtained from two different coin cells prepared under the same conditions. Small variations in Coulombic efficiency during the initial cycles may arise due to experimental differences such as electrode loading or cell assembly. Nonetheless, the overall electrochemical trends remain fully consistent and reproducible.

9.

Evolution of the specific charge and discharge capacity and the Coulombic efficiency as a function of the number of cycles of LNMO-F0 (a), LNMO-F1 (b), LNMO-F2 (c), and LNMO-F3 (d).

As shown in Figure , the discharge capacity of the LNMO-F0 decreases slightly from 129 to 128, 127, and 120 mA h g^–1 when the current rate increases from 0.1 to 0.2 and 1C and further decreases to 118 mA h g^–1 at 2C before recovering to 126 mA h g^–1 when the current rate returns to 0.1C, demonstrating good rate performance. The fluorinated samples follow a similar trend, with LNMO-F3 exhibiting a superior rate capability compared to LNMO-F0, LNMO-F1, and LNMO-F2. Specifically, the discharge capacities of LNMO-F3 were 130, 128, 125, and 122 mA h g^–1 at 0.1C, 0.2C, 1C, and 2C, respectively. Notably, the capacity of LNMO-F3 fully recovers to nearly 130 mA h g^–1 upon returning to 0.1C, indicating excellent electrochemical reversibility.

10.

Current rate capabilities of LNMO, LNMO-F1, LNMO-F2, and LNMO-F3 electrodes.

These results suggest that LNMO-F3, which contains distinct fluorine and LiF surface content, demonstrates the best rate performance among the tested samples. Moreover, the Mn³⁺/Mn⁴⁺ ratio and/or the environment of Mn⁴⁺ is the most perturbed by fluorination, further contributing to the improved electrochemical behavior.

4. Conclusions

This study presents a comprehensive investigation of surface modification of disordered LiNi_0.5Mn_1.5O₄ (LNMO) cathode material using Atomic Layer Fluorination (ALF) and its impact on electrochemical performance. Through careful control of the fluorination process using xenon difluoride XeF_2, a uniform fluorine layer on LNMO particles is achieved without alteration of the underlying cubic spinel structure. Characterization techniques such as XPS, AES, NMR, and SEM confirm the successful fluorination and validate the structural and compositional integrity of the material. The fluorinated LNMO exhibits enhanced cycling stability, maintaining high Coulombic efficiency and mitigating side reactions at the electrode/electrolyte interface. However, a slight decrease in rate capability is observed compared to pristine LNMO, which can be attributed to the influence of the fluorine-modified surface on lithium-ion transport kinetics. Despite this, the improved stability suggests that fluorinated LNMO remains a promising candidate for long-term high-energy-density lithium-ion batteries.

While this study demonstrates that surface modification via atomic layer fluorination effectively enhances the electrochemical stability of LNMO, further optimization of the fluorination process is needed. In particular, controlling the thickness and homogeneity of the fluorinated layer could help mitigate any adverse effects on the lithium-ion transport. Additionally, further investigations using techniques such as electrochemical impedance spectroscopy (EIS), operando XRD/XAS, or post-mortem XPS could provide deeper insights into the impact of fluorination on interfacial resistance and lithium-ion diffusion pathways.

These findings highlight the potential of atomic layer fluorination as a strategy to enhance the performance of high-voltage cathode materials. Future work should focus on refining the fluorination parameters to achieve an optimal balance between the structural stability and lithium-ion transport efficiency.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c01717.

• EDX complementary data, XPS complementary spectra and binding energies table, SEM images of AES samples, XRD profile matching results, and integrated EPR spectra (PDF)

The manuscript was written through the contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.