Sulfate-Containing Composite Based on Ni-Rich Layered Oxide LiNi0.8Mn0.1Co0.1O2 as High-Performance Cathode Material for Li-ion Batteries

Composite positive electrode materials (1−x) LiNi0.8Mn0.1Co0.1O2∙xLi2SO4 (x = 0.002–0.005) for Li-ion batteries have been synthesized via conventional hydroxide or carbonate coprecipitation routes with subsequent high-temperature lithiation in either air or oxygen atmosphere. A comparative study of the materials prepared from transition metal sulfates (i.e., containing sulfur) and acetates (i.e., sulfur-free) with powder X-ray diffraction, electron diffraction, high angle annular dark field transmission electron microscopy, energy-dispersive X-ray spectroscopy, and electron energy loss spectroscopy revealed that the sulfur-containing species occur as amorphous Li2SO4 at the grain boundaries and intergranular contacts of the primary NMC811 crystallites. This results in a noticeable enhancement of rate capability and capacity retention over prolonged charge/discharge cycling compared to their sulfur-free analogs. The improvement is attributed to suppressing the high voltage phase transition and the associated accumulation of anti-site disorder upon cycling and improving the secondary agglomerates’ mechanical integrity by increasing interfacial fracture toughness through linking primary NMC811 particles with soft Li2SO4 binder, as demonstrated with nanoindentation experiments. As the synthesis of the (1−x) LiNi0.8Mn0.1Co0.1O2∙xLi2SO4 composites do not require additional operational steps to introduce sulfur, these electrode materials might demonstrate high potential for commercialization.

Integration of amorphous Li 2 SO 4 into Ni-rich NMC secondary particles, predominantly to the boundaries between the primary particles and their intergranular contacts, increases the secondary agglomerates' interfacial fracture toughness. Thus, it suppresses mechanical fracture of the cathode particles during electrochemical cycling. The composite cathode possesses enhanced capacity retention and rate capability compared to single-phase materials due to surface protection from reaction with the electrolyte and suppressing anti-site disorder. Simultaneously, the composite cathode material synthesis does not include additional time-and cost-consuming operational steps, which makes such material attractive for industry needs.

Materials and Methods
The layered LiNi 0.8 Mn 0.1 Co 0.1 O 2 oxide materials were synthesized via a coprecipitation method followed by high-temperature lithiation in either air or oxygen atmosphere. First, hydroxide or carbonate precursors were precipitated from aqueous solutions of either TM sulfates (NiSO 4 ·6H 2 O, MnSO 4 Table 1 were pumped slowly into a continuously stirred batch reactor under an Ar atmosphere. Simultaneously, an aqueous solution of either NaOH or Na 2 CO 3 and NH 3 ·H 2 O with the concentrations provided in Table 1 was pumped into the reactor as precipitating and chelating agents, respectively. All solutions were pumped separately into the reactor at a feeding rate of about 0.6 L h −1 , and continuous stirring was kept at 800 r min −1 . The reactor temperature and pH value temperature were fixed to 50 • C and 11.5 for the hydroxide route and 56 • C and 7.8 for the carbonate route. The precipitated precursors were filtered, carefully washed several times with deionized water, and dried at 110 • C for 12 h under a dynamic vacuum. The obtained hydroxide or carbonate precursors and LiOH·H 2 O were mixed with a molar ratio of 1:1.06 or 1:1.03, respectively, and annealed at 500 • C for 5 h and then at 850 • C for 12 h in air with intermediate grinding between the calcination steps, or at 750 • C for 12 h in flowing oxygen atmosphere. All samples were further indicated with two capital letters and a calcination atmosphere in brackets. The first letter corresponded to the type of precipitated precursor (either hydroxide (H) or carbonate (C)), the second letter corresponded to the TM source used for precipitation (either sulfates (S) or acetates (A)); the calcination atmosphere was denoted as either air (a) or oxygen (o). Therefore, eight samples in total were discussed further:  X-ray diffraction (XRD) patterns of powder samples were obtained by a Huber G670 Guinier diffractometer (Rimsting, Germany) using Co-K α1 radiation (λ = 1.78892Å), a curved Ge (111) monochromator, and an image plate detector. The powder XRD data were collected at room temperature over the 10 • -90 • 2θ range with the step of 0.005 • . The unit cell parameters were calculated using the Le Bail method. Also, crystal structures were refined using the Rietveld method. Le Bail decomposition and Rietveld analysis were performed with the JANA2006 program package (Prague, Czech Republic, Version string 25/10/2015) [20].
Particle morphology and size distribution in the pristine and cycled cathodes were examined with scanning electron microscopy (SEM) using a ThermoFisher Quattro S microscope (Waltham, MA, USA).
Electron diffraction (ED) patterns, high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images, compositional maps, obtained by energy-dispersive X-ray spectroscopy in a STEM mode (EDS-STEM), and electron energy loss spectra (EELS) were acquired with a ThermoFisher Titan Themis Z transmission electron microscope (Eindhoven, The Netherlands) at 200 kV. It was equipped with a Super-X system for energy-dispersive X-ray spectroscopy and a Gatan Quantum ER965 spectrometer (Pleasanton, CA, USA) for EELS. TEM samples were prepared in air by crushing the crystals with an agate mortar and pestle in ethanol and depositing drops of suspension onto a carbon film supported by a copper grid.
Atomic force microscopy (AFM) nanoindentation and imaging were performed on cross-sections of the samples using a Cypher ES AFM (Asylum Research, Oxford Instruments, Santa Barbara, CA, USA) installed in an Ar-filled glovebox. The nanoindentation was made by a single crystal diamond probe with the 4-sided pyramid tip with a 45 ± 10 • half angle at the tip apex (DRP_In). The imaging was made by a single crystal diamond probe (HA_NC/FD). The cross-sections were prepared by embedding the electrodes into epoxy resin and consequent mechanical polishing.
The electrochemical properties were tested in coin-type 2032R cells using MTI BST battery analyzers (Shenzhen, China) with lithium metal as a counter electrode. The active cathode material, conductive additive (Super-P), and binder (polyvinylidene-fluoride, PVDF) were suspended in N-methyl-2-pyrrolidone (NMP) with a weight ratio of 80:10:10 and homogenized to form a slurry. The obtained slurry was applied onto a carbon-coated Al current collector (Xiamen, Fujian, China) using automatic film applicator Zehntner ZAA 2300 (Schwerzenbach, Switzerland), dried, and then cut into 16 mm diameter disks with cathode loading level of 6-7 mg cm −2 . The electrolyte consisting of 1M LiPF 6 (Sigma-Aldrich, St. Louis, MO, USA, ≥99.99%) solution in ethylene carbonate/propylene carbonate/dimethyl carbonate solution (EC:PC:DMC = 1:1:3 vol.) was used. The half-cells were assembled in an Ar-filled glove box, and galvanostatic tests were performed in the potential range of 2.7-4.3 V vs. Li/Li + at room temperature at different current densities, under the following program: 5 cycles at 0.1C, 5 cycles at 0.2C, 5 cycles at 0.5C, 5 cycles at 0.2C, 5 cycles at 1C, 5 cycles at 0.2C, and 100 cycles at 1C, where 1C = 200 mA g −1 .

Results
Powder XRD patterns of the obtained precursors are illustrated in Figure S1 of Supplementary Materials. Pairs of hydroxide precursors, HS and HA, and carbonate precursors, CS and CA, had identical powder XRD patterns. The hydroxide precursors precipitated from either sulfates or acetates demonstrate the β-Ni(OH) 2 -type structure (ICDD #74-2075) for which all diffraction peaks were indexed to a trigonal structure with the P3m1 space group and refined lattice parameters a = 3.121(1), c = 4.612(6) Å ( Figure S1a). The absence of impurity phases indicates that the Co 2+ and Mn 2+ cations partially substituted Ni 2+ in the β-Ni(OH) 2 structure. Although the carbonate precursor's crystallinity was lower, the strongest diffraction peaks matched the powder XRD pattern of NiCO 3 ·H 2 O (ICDD #12-0276). According to EDS-STEM compositional mapping, all precursors possessed homogeneous distribution of Ni, Mn, and Co ( Figure 1) with the metal atomic ratio corresponding to the desired NMC811 stoichiometry. Additionally, EDS-STEM maps demonstrated that the precursors precipitated from sulfates contain sulfur, uniformly distributed throughout the material (Figure 1a,c). The sulfur content and Ni:Mn:Co atomic ratio for the precursors are presented in Table 2.
The effect of TM source and precursor form on the morphology of NMC811 was examined by scanning electron microscopy (SEM). Both carbonate and hydroxide precursors precipitated from sulfates and acetates consisting of secondary particles with irregular shapes with an average size of 8-12 µm ( Figure S2a,c,e,g). These secondary particles were composed of numerous agglomerated flakes forming 3D hierarchical rose-like nanostructures in the hydroxide precursors ( Figure S2b,d), while both carbonate precursor particles were composed of flakes and small sphere-like particles ( Figure S2f,h). Although the differences in microstructural organization of secondary agglomerates in Nanomaterials 2020, 10, 2381 5 of 17 the precursors were clear in the SEM images, they were not fully inherited in the microstructure of the final NMC811 products, as will be demonstrated later.
oxygen annealing atmosphere should suppress Li and TM cation mixing (anti-site disorder) during synthesis, helping maintain a high content of Ni 3+ and, therefore, enhancing the electrochemical characteristics [21]. Thus, to discriminate the impact of the anti-site disorder on the electrochemical properties of the (1−x) LiNi0.8Mn0.1Co0.1O2•xLi2SO4 materials, the samples with different degrees of Li/Ni anti-site disorder were obtained by the variation of the synthesis atmosphere. The powder XRD patterns for the obtained NMC811 samples, prepared from both hydroxide and carbonate precursors and calcined in air ( Figure 2a) and in oxygen (Figure 2b) atmosphere, show well-crystallized singlephase α-NaFeO2-type layered compounds. All peaks were indexed in the rhombohedral space group R-3m with the unit cell parameters provided in Table 3.   In order to obtain the final NMC811 materials, the precursors and LiOH·H 2 O were subjected to heat treatment in two different atmospheres-air and oxygen. It is generally considered that the oxygen annealing atmosphere should suppress Li and TM cation mixing (anti-site disorder) during synthesis, helping maintain a high content of Ni 3+ and, therefore, enhancing the electrochemical characteristics [21]. Thus, to discriminate the impact of the anti-site disorder on the electrochemical properties of the (1−x) LiNi 0.8 Mn 0.1 Co 0.1 O 2 ·xLi 2 SO 4 materials, the samples with different degrees of Li/Ni anti-site disorder were obtained by the variation of the synthesis atmosphere. The powder XRD patterns for the obtained NMC811 samples, prepared from both hydroxide and carbonate precursors and calcined in air ( Figure 2a) and in oxygen (Figure 2b) atmosphere, show well-crystallized single-phase α-NaFeO 2 -type layered compounds. All peaks were indexed in the rhombohedral space group R-3m with the unit cell parameters provided in Table 3.  The lattice parameters, as well as the amount of Ni 2+ /Li + anti-site defects, originated from the partial interchange of the Li + (r = 0.76 Å ) and Ni 2+ (r = 0.69 Å ) cations between the 3b (i.e., Li) and 3a (i.e., Ni) sites [22]. They were calculated from powder XRD patterns for all samples with the Rietveld refinement based on the layered R-3m structure (Table 3). Small values of the profile reliability factor Rp and Goodness of fit (GOF) indicated reasonably good refinement results. The observed and calculated powder XRD patterns are shown in Figures S3 and S4 for the air-and oxygen-annealed samples, respectively. As expected, changing the annealing atmosphere from air to oxygen leads to the relative stabilization of the layered structure, as reflected by the declining amount of the Ni 2+ /Li + anti-site defects for the hydroxide and carbonate precursors lithiated in O2, compared to ones lithiated in air (Table 3).
Additionally, the crystal structure of Ni-rich NMCs was investigated by selected area electron diffraction (SAED) (   The lattice parameters, as well as the amount of Ni 2+ /Li + anti-site defects, originated from the partial interchange of the Li + (r = 0.76 Å) and Ni 2+ (r = 0.69 Å) cations between the 3b (i.e., Li) and 3a (i.e., Ni) sites [22]. They were calculated from powder XRD patterns for all samples with the Rietveld refinement based on the layered R-3m structure (Table 3). Small values of the profile reliability factor R p and Goodness of fit (GOF) indicated reasonably good refinement results. The observed and calculated powder XRD patterns are shown in Figures S3 and S4 for the air-and oxygen-annealed samples, respectively. As expected, changing the annealing atmosphere from air to oxygen leads to the relative stabilization of the layered structure, as reflected by the declining amount of the Ni 2+ /Li + anti-site defects for the hydroxide and carbonate precursors lithiated in O 2 , compared to ones lithiated in air (Table 3).
Additionally, the crystal structure of Ni-rich NMCs was investigated by selected area electron diffraction (SAED) ( Figure 3). All reflections in the SAED patterns of all samples, exemplified with those for the samples HS(o) (Figure 3a,b) and HA(o) (Figure 3c,d), correspond to the layered O3 R-3m structure. No additional reflections were observed in the SAED patterns, meaning that all investigated samples possessed NMC811 as the only crystalline phase.
SEM investigated the morphology of the obtained NMC811 cathode materials. The secondary agglomerates in the NMC811 samples retain the secondary agglomerates' size in their corresponding precursors. However, the precursors' flake-like primary particles are completely changed into rectangular ones, as shown in Figure 4. The size of the primary particles in the oxygen-annealed samples is clearly smaller (100-500 nm, Figure 4a-d) than the air-annealed particles' size (0.8-1 µm, Figure 4e-h). Size distributions of the primary particles in the obtained samples are presented in Figure S5.  SEM investigated the morphology of the obtained NMC811 cathode materials. The secondary agglomerates in the NMC811 samples retain the secondary agglomerates' size in their corresponding precursors. However, the precursors' flake-like primary particles are completely changed into rectangular ones, as shown in Figure 4. The size of the primary particles in the oxygen-annealed samples is clearly smaller (100-500 nm, Figure 4a-d) than the air-annealed particles' size (0.8-1 µ m, Figure 4e-h). Size distributions of the primary particles in the obtained samples are presented in Figure S5.  In order to determine the precise elemental composition and distribution of the elements, EDS-STEM analysis was carried out for all NMC811 samples. EDS-STEM elemental maps of nickel, manganese, cobalt, oxygen, and sulfur show that all NMC811 materials, regardless of the synthesis conditions, possess uniform Ni, Mn, and Co distribution ( Figure 5). Moreover, both hydroxide and carbonate samples obtained from sulfates include impurities of sulfur-based species. However, powder XRD cannot be detected as any side phase, probably because of the low sulfur content and/or In order to determine the precise elemental composition and distribution of the elements, EDS-STEM analysis was carried out for all NMC811 samples. EDS-STEM elemental maps of nickel, manganese, cobalt, oxygen, and sulfur show that all NMC811 materials, regardless of the synthesis conditions, possess uniform Ni, Mn, and Co distribution ( Figure 5). Moreover, both hydroxide and carbonate samples obtained from sulfates include impurities of sulfur-based species. However, powder XRD cannot be detected as any side phase, probably because of the low sulfur content and/or amorphous nature of the sulfur-containing phase, as no other crystalline compounds except NMC811 were observed with the SAED patterns. The atomic percentages of Ni, Mn, Co, and S measured by EDS are listed in Table 4. The cationic composition was in good accordance with the desired NMC811 stoichiometry. Compositional EDS-STEM maps showed that sulfur-containing species were distributed unevenly in the HS(a), CS(a), HS(o), and CS(o) samples (Figure 5a-d), in contrast to very homogeneous sulfur distribution in the co-precipitated precursors (Figure 1). Sulfur appears to be distributed preferentially at the grain boundaries between primary NMC811 crystals and the primary grains' contact points, as revealed by the correlation between the EDS and HAADF signal profiles in Figure 6 and Figure S6. The sulfur EDS signal was picked at the minima of the HAADF signal corresponding to the intergranular contacts.  The sulfur species' chemical nature at the grain boundary region of the HS(o) sample was revealed with EELS L 2,3 sulfur edge (Figure 7). It showed two characteristic sharp pre-peaks at 174 eV, 182 eV, and the broad core edge at 197 eV. According to Reference [23], this type of spectrum corresponds to sulfur in the +6 oxidation state in a tetrahedral oxygen environment, which allows us to conclude that sulfur is presented in the form of sulfate anions. However, EDS did not reveal any enrichment of Ni, Mn or Co cations at the S-rich areas, as well as a presence of any other side cation to counterbalance the negative charge of SO 4     The sulfur species' chemical nature at the grain boundary region of the HS(o) sample was revealed with EELS L2,3 sulfur edge (Figure 7). It showed two characteristic sharp pre-peaks at 174 eV, 182 eV, and the broad core edge at 197 eV. According to Reference [23], this type of spectrum corresponds to sulfur in the +6 oxidation state in a tetrahedral oxygen environment, which allows us to conclude that sulfur is presented in the form of sulfate anions. However, EDS did not reveal any enrichment of Ni, Mn or Co cations at the S-rich areas, as well as a presence of any other side cation to counterbalance the negative charge of SO4 2− . Moreover, sulfates of Mn(II), Co(II), and Ni(II) were thermally unstable at the oxidative atmosphere of the calcination step. Li + is the only cation invisible with EDS, which can be bound to the SO4 2− anion at the given synthesis conditions, particularly Each sample's electrochemical properties were evaluated using coin-type lithium half-cells between 2.7-4.3 V vs. Li/Li + at room temperature. Figure 8a,b presents the rate capabilities of the NMC811 samples at different current densities ranging from 20 mA·g −1 to 200 mA·g −1 (0.1C to 1C) and the capacity recovery when the current density returns to 0.2C. The air-annealed samples with the same primary particle size and morphology demonstrated different electrochemical performance. The HS(a) sample exhibits a much higher capacity compared to the HA(a) sample for the whole range of current densities ( Table 5). The same tendency was observed for the carbonate-precipitated samples, where the CS(a) sample possessed a higher specific capacity compared to the CA(a) sample ( Table 5). The cathode materials precipitated from sulfates always showed higher capacity retention compared to their acetate analogs. Thus, after 130 cycles, HS(a) and CS(a) samples (Figure 7b,d) demonstrated capacity retention of about 91% and 85%, respectively. This was higher than the capacity of the HA(a) and CA(a) samples, which were about 73% and 69%, respectively. The initial discharge capacity was markedly enhanced by switching the calcination atmosphere from air to oxygen due to suppressing the Ni 2+ /Li + anti-site disorder, as revealed with the Rietveld refinement (Table 3). Given Li diffusion limitations, the HS(o) and HA(o) samples with primary particle size about 100-300 nm demonstrated higher discharge capacity of~199 and 193 mA·g −1 , respectively, compared to 186 and 181 mA·g −1 for the CS(o) and CA(o) samples with primary particle size of about 0.5-1 µm (Table 5). Simultaneously, the initial discharge capacity of the oxygen-annealed sulfate samples is always higher than that of the oxygen-annealed acetate samples. The capacity retention and rate capability for the sulfate/acetate pairs were also different. The HS(o) sample demonstrated higher capacity retention of 91%, compared to that of HA(o) (81%) ( Table 5). The same tendency was observed for the samples obtained from the carbonate precursor: the CS cathode materials possessed enlarged stability at different current densities (0.1C, 0.2C, 0.5C, and 1C) and prolonged cycling. Moreover, the rate capability of carbonate-made cathode material precipitated from sulfates was greater than that of the sample obtained from acetates, giving at least 10-15% growth in discharge capacity at elevated current densities. In addition, the comparison to the analogous NMC811 samples obtained in Reference [15] using the same synthesis conditions showed that sulfur-containing samples prepared in the present work demonstrated improved electrochemical properties. Nanomaterials 2020, 10, x FOR PEER REVIEW 10 of 18 considering sufficient thermal stability of Li2SO4 (melting point of 859 °C, decomposition at >1300 °C). Thus, based on the combination of EDS and EELS results and chemical reasonings, the NMC811 materials, synthesized from TM sulfates through both the hydroxide and carbonate precursors, were considered as composite materials with the general formula (1−x) LiNi0.8Mn0.1Co0.1O2•xLi2SO4, where Li2SO4 is present as an amorphous phase according to XRD and SAED.  Each sample's electrochemical properties were evaluated using coin-type lithium half-cells between 2.7-4.3 V vs. Li/Li + at room temperature. Figure 8a,b presents the rate capabilities of the NMC811 samples at different current densities ranging from 20 mA•g −1 to 200 mA•g −1 (0.1C to 1C) and the capacity recovery when the current density returns to 0.2C. The air-annealed samples with the same primary particle size and morphology demonstrated different electrochemical performance. The HS(a) sample exhibits a much higher capacity compared to the HA(a) sample for the whole range of current densities ( Table 5). The same tendency was observed for the carbonate-precipitated samples, where the CS(a) sample possessed a higher specific capacity compared to the CA(a) sample ( Table 5). The cathode materials precipitated from sulfates always showed higher capacity retention compared to their acetate analogs. Thus, after 130 cycles, HS(a) and CS(a) samples (Figure 7b,d) demonstrated capacity retention of about 91% and 85%, respectively. This was higher than the capacity of the HA(a) and CA(a) samples, which were about 73% and 69%, respectively. The initial discharge capacity was markedly enhanced by switching the calcination atmosphere from air to oxygen due to suppressing the Ni 2+ /Li + anti-site disorder, as revealed with the Rietveld refinement (Table 3)  The sulfate-containing samples, obtained in both atmospheres, as well as the acetate-made samples, annealed in oxygen, recovered their original discharge capacity when the current density returned to 0.2C. This result demonstrated that the capacity fade at high current densities was caused by diffusion limitations rather than irreversible structural changes [24] (Figure 7a,b).
The differential capacity (dQ/dV) plots for sulfur-containing HS(o) composite and single-phase NMC811 HA(o) samples are presented in Figure 9. The most important features, which could be observed from differential capacity curves for Ni-rich NMCs, are the phase transformations, appearing throughout the electrochemical cycling. Here, the dQ/dV profiles of HS(o) and HA(o) samples show three peaks in the anodic branch between 3.5 V and 4.3 V, which are related to three phase transitions: hexagonal-to-monoclinic (H1→M), monoclinic-to-hexagonal (M→H2), and hexagonal-to-hexagonal (H2→H3) (Figure 9a,b) [25]. The corresponding reduction peaks are presented in the cathodic branch. It is generally believed that the H2 → H3 phase transition is associated with structural transformation through transition-metal ion migration to the lithium layer, accompanied by a significant decrease of the c lattice parameter. This effect leads to the formation of microcracks in secondary particles, thereby promoting the material's degradation [8,9]. Besides, at a highly charged state, TM ions' active migration to the Li sites leads to irreversible crystal structure changes, blocking of Li-ions migration channels, and following drop of Li diffusivity, resulting in a decrease of the electrochemical performance of the cathode materials [26].
enlarged stability at different current densities (0.1C, 0.2C, 0.5C, and 1C) and prolonged cycling. Moreover, the rate capability of carbonate-made cathode material precipitated from sulfates was greater than that of the sample obtained from acetates, giving at least 10-15% growth in discharge capacity at elevated current densities. In addition, the comparison to the analogous NMC811 samples obtained in Reference [15] using the same synthesis conditions showed that sulfur-containing samples prepared in the present work demonstrated improved electrochemical properties. The sulfate-containing samples, obtained in both atmospheres, as well as the acetate-made samples, annealed in oxygen, recovered their original discharge capacity when the current density returned to 0.2C. This result demonstrated that the capacity fade at high current densities was caused by diffusion limitations rather than irreversible structural changes [24] (Figure 7a,b).
The differential capacity (dQ/dV) plots for sulfur-containing HS(o) composite and single-phase NMC811 HA(o) samples are presented in Figure 9. The most important features, which could be observed from differential capacity curves for Ni-rich NMCs, are the phase transformations, appearing throughout the electrochemical cycling. Here, the dQ/dV profiles of HS(o) and HA(o) samples show three peaks in the anodic branch between 3.5 V and 4.3 V, which are related to three phase transitions: hexagonal-to-monoclinic (H1→M), monoclinic-to-hexagonal (M→H2), and hexagonal-to-hexagonal (H2→H3) (Figure 9a,b) [25]. The corresponding reduction peaks are presented in the cathodic branch. It is generally believed that the H2 → H3 phase transition is associated with structural transformation through transition-metal ion migration to the lithium layer, Figure 8. Rate capability at various charge-discharge current densities (a,b) and cycling performance at 1C rate (c,d) of the NMC811 samples. As reported previously [27,28], sharp oxidation peaks in the dQ/dV plots of Ni-rich NMC cathode materials usually represent abrupt and possibly detrimental structural changes, while broader peaks with low intensity indicate protracted phase transitions. The dQ/dV curves, plotted at 0.1C, 0.2C, 0.5C, and 1C current densities, show that the presence of a sulfur-containing compound in the HS(o) sample reduces the abruptness of the phase transitions at different current densities. This is because the intensities of redox peaks in the dQ/dV plot of the HS(o) cathode (Figure 9a) show a more rapid decline with an increase of current density compared to those of HA(o) (Figure 9b). The most noticeable suppression of the redox peaks is observed for the H2→H3 phase transition, which is believed to be responsible for the capacity loss of Ni-rich NMCs [29]. Thus, the presence of sulfur-containing species helps to suppress the mentioned deleterious phase transition [7]. cathode materials usually represent abrupt and possibly detrimental structural changes, while broader peaks with low intensity indicate protracted phase transitions. The dQ/dV curves, plotted at 0.1C, 0.2C, 0.5C, and 1C current densities, show that the presence of a sulfur-containing compound in the HS(o) sample reduces the abruptness of the phase transitions at different current densities. This is because the intensities of redox peaks in the dQ/dV plot of the HS(o) cathode (Figure 9a) show a more rapid decline with an increase of current density compared to those of HA(o) (Figure 9b). The most noticeable suppression of the redox peaks is observed for the H2→H3 phase transition, which is believed to be responsible for the capacity loss of Ni-rich NMCs [29]. Thus, the presence of sulfurcontaining species helps to suppress the mentioned deleterious phase transition [7].  The half-cells with HS(o)-and HA(o)-based electrodes after 130 cycles were disassembled inside a glovebox, and the electrodes were investigated by powder XRD analysis. The powder XRD patterns can still be fitted using the Rietveld refinement with the rhombohedral space group R-3m ( Figure S7). As for pristine cathode materials, the amount of Ni 2+ /Li + anti-site defects were calculated from powder XRD patterns for both cycled HS(o)-and HA(o)-based electrodes. It was found that the amount of anti-site defects in HA(o) active cathode material increased significantly from 2.12 to 13.96% after cycling, while in HS(o), it barely increased from 2.54 to 4.12% (Table S1).
Given the high voltage H2→H3 phase transition associated with irreversible mechanical degradation of the material, i.e., the formation of microcracks within the secondary particles and detachment of cathode particles from cathode slurry binder, a study of cathode morphology was carried out via comparative SEM of the as-prepared HS(o)-and HA(o)-based electrodes and the corresponding electrodes after 130 charge/discharge cycles ( Figure 10). While the HS(o) electrode retains its integrity after cycling (Figure 10a,b), large cracks across the cathode secondary particles as well as numerous detachments between the secondary particles and binder were observed in the HA(o)-based cathode (Figure 10c,d). These detrimental issues in the HA(o)-based cathode are mainly associated with mechanical stresses induced by repetitive volume expansion and contraction of the primary particles during electrochemical cycling [30]. Still, no statistically relevant changes in primary particle size were detected from the cycled cathodes' SEM images compared to the pristine ones ( Figure S8).
The improved mechanical behavior of the HS(o)-based cathode could be attributed to the fact that amorphous Li 2 SO 4 serves as a soft binder linking harder NMC811 primary particles into a composite. Given the amorphous nature of Li 2 SO 4 , it may help to relieve stresses that emerged during cycling and, therefore, preserve the cathode integrity [31]. This conjecture has been tested by AFM nanoindentation experiments (Figure 11a). Figure 11b,c illustrates typical indents obtained on the secondary particles in the HS(o) and HA(o) samples at 130 µN applied force. The corresponding indentation curves are shown in Figure S9. From the profiles in Figure 11d, one can see that the indents are deeper in the HS(o) sample. Simultaneously, the expelled material around the indents is not fractured on the HS(o) sample and strongly fractured on the HA(o) sample. Crack formation in the HA(o) sample becomes evident from the load-displacement curves ( Figure S7), demonstrating pop-in events caused by a sudden increase of the indenter penetration due to cracking [32]. It points out that the amorphous Li 2 SO 4 species at the boundaries between the primary particles and their intergranular contacts accommodate part of the mechanical stress and hold the primary particles together, thereby effectively preventing the secondary particles' mechanical fracture. In light of the nanoindentation experiments, the (1−x) LiNi 0.8 Mn 0.1 Co 0.1 O 2 ·xLi 2 SO 4 materials behave as composites consisting of primary particles of the NMC811 phase with higher hardness linked together with softer Li 2 SO 4 binder. It is well known that in such composites decreasing hardness is accompanied by increasing interfacial fracture toughness [33,34]. anti-site defects in HA(o) active cathode material increased significantly from 2.12 to 13.96% after cycling, while in HS(o), it barely increased from 2.54 to 4.12% (Table S1).
Given the high voltage H2→H3 phase transition associated with irreversible mechanical degradation of the material, i.e., the formation of microcracks within the secondary particles and detachment of cathode particles from cathode slurry binder, a study of cathode morphology was carried out via comparative SEM of the as-prepared HS(o)-and HA(o)-based electrodes and the corresponding electrodes after 130 charge/discharge cycles ( Figure 10). While the HS(o) electrode retains its integrity after cycling (Figure 10a,b), large cracks across the cathode secondary particles as well as numerous detachments between the secondary particles and binder were observed in the HA(o)-based cathode (Figure 10c,d). These detrimental issues in the HA(o)-based cathode are mainly associated with mechanical stresses induced by repetitive volume expansion and contraction of the primary particles during electrochemical cycling [30]. Still, no statistically relevant changes in primary particle size were detected from the cycled cathodes' SEM images compared to the pristine ones ( Figure S8). The improved mechanical behavior of the HS(o)-based cathode could be attributed to the fact that amorphous Li2SO4 serves as a soft binder linking harder NMC811 primary particles into a composite. Given the amorphous nature of Li2SO4, it may help to relieve stresses that emerged during cycling and, therefore, preserve the cathode integrity [31]. This conjecture has been tested by AFM nanoindentation experiments (Figure 11a). Figure 11b  the HA(o) sample becomes evident from the load-displacement curves ( Figure S7), demonstrating pop-in events caused by a sudden increase of the indenter penetration due to cracking [32]. It points out that the amorphous Li2SO4 species at the boundaries between the primary particles and their intergranular contacts accommodate part of the mechanical stress and hold the primary particles together, thereby effectively preventing the secondary particles' mechanical fracture. In light of the nanoindentation experiments, the (1−x) LiNi0.8Mn0.1Co0.1O2•xLi2SO4 materials behave as composites consisting of primary particles of the NMC811 phase with higher hardness linked together with softer Li2SO4 binder. It is well known that in such composites decreasing hardness is accompanied by increasing interfacial fracture toughness [33,34].

Discussion
We conducted a comparative investigation of the NMC811 samples obtained through the coprecipitation method from different TM sources using either carbonate or hydroxide routes and lithiation at different annealing atmospheres. The investigation allowed us to discriminate the role in electrochemical behavior of minor sulfur impurity from the influence of other important factors, such as crystal structure, homogeneity of the TM distribution, amount of anti-site defects, size of the primary particles, and secondary agglomerates. Sulfur, which is present as amorphous Li2SO4 at the grain boundaries and intergranular contacts, remarkably enhances capacity retention over prolonged electrochemical cycling, improves rate capability and initial discharge capacity of the NMC811 cathodes.

Discussion
We conducted a comparative investigation of the NMC811 samples obtained through the coprecipitation method from different TM sources using either carbonate or hydroxide routes and lithiation at different annealing atmospheres. The investigation allowed us to discriminate the role in electrochemical behavior of minor sulfur impurity from the influence of other important factors, such as crystal structure, homogeneity of the TM distribution, amount of anti-site defects, size of the primary particles, and secondary agglomerates. Sulfur, which is present as amorphous Li 2 SO 4 at the grain boundaries and intergranular contacts, remarkably enhances capacity retention over prolonged electrochemical cycling, improves rate capability and initial discharge capacity of the NMC811 cathodes.
The attractive point is that grain boundary modification with Li 2 SO 4 does not require any additional synthesis step and occurs spontaneously if the hydroxide or carbonate precursors are precipitated from TM sulfates' solutions. The solutions are the most common initial ingredients for the production of the NMC-based cathode materials. Sulfur is distributed homogeneously over the precursor's particles without any sign of segregation, and numerous attempts to eradicate it by repetitive rinsing with water were unsuccessful. Taking into account the absence of other crystalline phases in the precursors, one might suggest that sulfur enters as SO 4  The mechanism through which Li 2 SO 4 affects the electrochemical behavior is multimodal. Interestingly, lithium sulfate is a good Li-ion conductor in its high-temperature α-phase above 575 • C, after undergoing a first-order phase transition. However, at lower temperatures, crystalline Li 2 SO 4 demonstrates relatively poor Li-ion conductivity [35]. However, amorphous Li 2 SO 4 behaves differently, as demonstrated with improved Li-ion transport in the mixtures of the layered, complex oxide cathodes and amorphous Li 2 SO 4 employed in all-solid-state Li-ion batteries [17,36]. Unlike previous research [17,36], in the present work, improving Li-ion transport in the NMC811 secondary agglomerates by enhancing the intergranular contacts with amorphous Li 2 SO 4 layers positively impacted the rate capability of the sulfur-containing cathodes. We achieved this without noticeable reduction of energy density because of a much smaller amount of Li 2 SO 4 . Simultaneously, grain boundary modification smoothens the phase transitions driving the material towards more solid-solution-like behavior that should also increase the rate capability. A very significant difference is observed in the amount of accumulated Ni 2+ /Li + anti-site disorder in the cathodes with and without sulfur. Three times lower anti-site disorder in the sulfur-containing HS(o) cathode is coupled with the partial suppression of the H2→H3 phase transition, which is associated with oxygen loss at high potentials through the reaction with the electrolyte driving the layered structure towards the rock-salt one [7,[37][38][39]. Thus, the amorphous Li 2 SO 4 layer at the grain boundaries can partially protect the surface of the primary NMC811 crystallites from unwanted reaction with the electrolyte at elevated potentials. Finally, the composite (1−x) LiNi 0.8 Mn 0.1 Co 0.1 O 2 ·xLi 2 SO 4 cathode preserves its mechanical integrity after prolonged cycling, unlike its single-phase NMC811 counterparts. Amorphous Li 2 SO 4 acts as a soft binder linking the harder NMC811 primary particles, thereby lowering the hardness of the secondary agglomerates and at the same time improving the interfacial fracture toughness making the agglomerates more resistant to cracking associated with interrupting the charge transfer pathways and compromising the mechanical stability of the cathode [9].

Conclusions
(1−x) LiNi 0.8 Mn 0.1 Co 0.1 O 2 ·xLi 2 SO 4 (x < 0.005) composite cathode materials for Li-ion batteries were prepared through either hydroxide or carbonate coprecipitation techniques followed by high-temperature lithiation. Amorphous Li 2 SO 4 , located at the grain boundaries and intergranular contacts of the primary NMC811 crystallites, enhances the capacity retention and rate capability of the cathodes cycled within the 2.7-4.3 V vs. Li/Li + voltage window. The positive effect is associated with suppressing the high voltage phase transition and accumulated anti-site disorder after prolonged cycling and improving mechanical integrity of the secondary agglomerates by increasing interfacial fracture toughness by linking primary NMC811 particles with soft Li 2 SO 4 binder. The present work highlights a possible route for further improvement of Ni-rich NMCs' electrochemical properties through dedicated engineering of the grain boundaries' structure and composition.