Improved electrochemical properties of LiNi0.91Co0.06Mn0.03O2 cathode material via Li-reactive coating with metal phosphates

Ni-rich layered oxides are promising cathode materials due to their high capacities. However, their synthesis process retains a large amount of Li residue on the surface, which is a main source of gas generation during operation of the battery. In this study, combined with simulation and experiment, we propose the optimal metal phosphate coating materials for removing residual Li from the surface of the Ni-rich layered oxide cathode material LiNi0.91Co0.06Mn0.03O2. First-principles-based screening process for 16 metal phosphates is performed to identify an ideal coating material that is highly reactive to Li2O. By constructing the phase diagram, we obtain the equilibrium phases from the reaction of coating materials and Li2O, based on a database using a DFT hybrid functional. Experimental verification for this approach is accomplished with Mn3(PO4)2, Co3(PO4)2, Fe3(PO4)2, and TiPO4. The Li-removing capabilities of these materials are comparable to the calculated results. In addition, electrochemical performances up to 50 charge/discharge cycles show that Mn-, Co-, Fe-phosphate materials are superior to an uncoated sample in terms of preventing capacity fading behavior, while TiPO4 shows poor initial capacity and rapid reduction of capacity during cycling. Finally, Li-containing equilibrium phases examined from XRD analysis are in agreement with the simulation results.

There has been increasing demand for lithium ion batteries (LIBs) for application in electric devices such as mobile phones and electrical vehicles. To achieve high energy density and long-term cyclability in LIBs, the use of transition metal (TM)-based oxide cathode materials could be an ideal option because their optimal composition can provide large capacity, low manufacturing cost, and great rate capability [1][2][3][4] . Nickel-rich nickel-cobalt-manganese oxide (termed Ni-rich NCM) is a class of promising materials that can satisfy those needs, but they suffer from several types of degradation behaviors such as phase transformation and gas generation [5][6][7] .
To mitigate degradation behaviors in layered oxide cathode materials thus enhancing the electrochemical performance, the surface modification method, i.e., surface coating, has been suggested to provide a physical barrier at the surface of the cathode and prevent the direct contact between active materials and electrolytes. For example, many metal phosphate (MP) materials are suggested as effective coating materials such as MPO 4 (M = Al, Fe, Ce, and Sr) 8,9 , Ni 3 (PO 4 ) 2 10 , Mn 3 (PO 4 ) 2 11 , M 3 (PO 4 ) 2 (M = Zn and Mg) 12 , and Zr-phosphate 13 for cathode materials such as LiCoO 2 (LCO), LiNi 0.9 Co 0.1 O 2 , LiNi 0.8 Co 0.15 Al 0.05 O 2 (NCA), LiNi 0.6 Co 0.2 Mn 0.2 O 2 , and LiNi 0.8 Co 0.15 Mn 0.05 O 2 . In spite of their effectiveness in improving the capacity retention rate, this type of coating approach in principle has drawbacks: 1) Li ion diffusion during electrochemical cycling can be impeded, 2) residual Li still needs to be washed, which requires an additional step during synthesis, and 3) this process can degrade battery performance [14][15][16] . Li impurities, which are residues on the surface of the cathode formed after initial synthesis, are a major source of gas generation, resulting in swelling behavior inside the battery pack 14,17 . The excessive amount of Li used as a prerequisite for achieving sufficient capacity with Ni-rich cathode materials is a source of these impurities 17,18 . Previous studies demonstrated that residual Li (such as LiOH and Li 2 CO 3 ) can initiate the decomposition reaction with electrolytes at the interface, leading to the evolution of gas species such as CO 2 , O 2 , N 2 , and CO [14][15][16] .
Application of Li-containing phosphates such as LiNiPO 4 19 , Li 3 V 2 (PO 4 ) 3 20 , and LiAlTi(PO 4 ) 3 21 coating materials has been recommended to alleviate the problem of constricted Li ion diffusion, and they have been shown to improve the rate capability and prevent capacity fading during cycling. Another approach has been suggested to eliminate the two concerns simultaneously, i.e., finding coating materials that can directly react with residual Li so that they can be transformed to Li-containing phases and in parallel the amount of Li residue is reduced. For example, Kim et al. suggested using Co 3 (PO 4 ) 2 as a Li-reacting coating material that can be converted to LiCoPO 4 during annealing 16 . AlPO 4 was also demonstrated to be functional from this perspective by forming Li 3 PO 4 and LiAlO 2 after its reaction with Li residues 22 . Jo et al. proved that phosphoric acid (H 3 PO 4 ) can be changed to Li 3 PO 4 via a Li-reactive mechanism 23 . However, only a few attempts have been made with this concept, which makes it difficult to choose the optimal MP coating materials for this purpose. It is worth noting that a computational approach has been successfully applied by Wolverton's group to search for optimal coating materials to scavenge hydrofluoric acid (HF) 24,25 and interfacial stability between coating and cathode materials 26 . In this regard, the computational aid is promising for examining and screening MP materials that are highly reactive to Li residues and can form subsequent Li-containing equilibrium phases. The amount of Li residue increases for Ni-rich cathode materials; hence, more surface area is covered with residual Li 17 . Therefore, MP materials are likely to first react with residual Li and form Li-containing structures; remaining unreacted materials still can function as coating material by blocking the direct exposure of the cathode material to the electrolyte.
In this study, we implement a computational framework to propose the optimal MP coating material for removing residual Li from the surface of Ni-rich layered oxide cathode material LiNi 0.91 Co 0.6 Mn 0.3 O 2 (NCM) by employing first-principles calculations on 16 MP materials. The results are validated with experiments measuring the reduction in the amount of Li residue after applying the coating materials. We also perform electrochemical cycling tests to clarify which of the coating materials is more effective in preventing capacity fading. Finally, the equilibrium phases obtained from experiment are compared to those from the phase diagrams obtained by calculation.

Results and Discussion
Design chart and analysis. The overall reaction behavior between MP and Li 2 O in terms of reaction enthalpy (ΔH Li-M ). and gravimetric capacity (G C ) is shown in Fig. 1 (The complete list of reaction equations is shown in Table S.2 in SI). Each MP material has several possible reactions depending on the amount used. For example, Fe 3 (PO 4 ) 2 can undergo five reactions at various molar ratios from 0.12 to 2 relative to Li 2 O. Among the possible reactions with one material, it is energetically preferable when a larger weight of coating materials is applied. Hence, this relation complicates finding the optimal coating material holding both strengths, i.e., the largest reaction energy and the least weight.
To provide comprehensive understanding of the performance of each MP material, we further analyzed the design chart by extracting data and sorting them with respect to their competencies. Two types of data analysis were performed, i.e., the descending order of ΔH Li-M and G C values and their corresponding G C and ΔH Li-M values (denoted G C ′ and ΔH Li-M ′), respectively. The descending order means that the results are sorted according to the largest value among several possible reaction equations containing each material. Meanwhile, corresponding values (e.g., ΔH Li-M ′) indicate ΔH Li-M of the reaction when sorted in descending order of G C , and vice versa. The complete list is shown in Table S Fig. 2b), AlPO 4 is calculated to be the most efficient followed by Co 3 (PO 4 ) 2 , Zn 3 (PO 4 ) 2 , BPO 4 , FePO 4 , etc. Several significant findings can be addressed based on the above analysis. First, the MP materials positioned at higher ranks in terms of ΔH Li-M generally exhibit poor efficiency, i.e., they require a larger weight to achieve that reaction. For example, the most Li-reactive material (Mn 3 (PO 4 ) 2 ) is placed at 11 th in order of G C ′ and at 15 th in the descending order of G C . However, it is worthwhile to mention that its ΔH Li-M ′ is still the highest. This result indicates that it is desirable to use Mn 3 (PO 4 ) 2 when its higher reactivity to Li 2 O is required, but the weight of the coating material is not a critical condition to meet. Likewise, W(PO 4 ) 2 is placed 3 rd in the descending order of ΔH Li and its corresponding ΔH Li with respect to the descending order of G C is 2 nd , which indicates that its reaction with Li 2 O is highly preferable. However, as shown with Mn 3 (PO 4 ) 2 , this material requires a large weight because its G C is positioned 8 th from the corresponding order of ΔH Li-M and 11 th in the descending order of G C values (Table S.3).
As discussed earlier, it is difficult to find coating materials satisfying both criteria of reactivity and gravimetric efficiency. Among 16 calculated materials, Co 3 (PO 4 ) 2 and BPO 4 reveal moderate capabilities for both constraints. For example, ΔH Li-M of Co 3 (PO 4 ) 2 is ranked 5 th and its G C is also positioned high (2 nd ). This means that this MP material can have good performance depending on which constraint is more important. BPO 4 exhibits modest performance from all perspectives; all of its capabilities are within the 6 th place of the 16 MP materials and its capability is not sensitive to the amount used. Its ΔH Li-M value is positioned 6 th and its corresponding G C is 5 th while its G C is ranked 4 th and its corresponding ΔH Li-M is 5 th .
Since two of the metal elements (Fe and Co) can take multiple oxidation states (e.g., Co is in the +2 and +3 states in Co 3 (PO 4 ) 2 and CoPO 4 , respectively), it is important to compare the performances of those species, as shown in the Inset of Fig. 2. In the case of Fe-based phosphate, the comparison shows that Fe 3 (PO 4 ) 2 exhibits better performance than FePO 4 . More specifically, it is more energetically preferable than FePO 4 ; its ΔH Li-M value is positioned 2 nd while that of FePO 4 is at the 15 th position in the descending order. However, FePO 4 exhibits better gravimetric efficiency (G C is ranked 5 th ) than that of Fe 3 (PO 4 ) 2 (ranked 8 th ). Hence, one needs to choose Fe-P based on the purpose of use. Reactivities of Co-based phosphates are almost the same (CoPO 4 and Co 3 (PO 4 ) 2 ranked 4 th and 5 th , respectively) but the G C value of Co 3 (PO 4 ) 2 (ranked 2 nd ) is better than that of CoPO 4 (ranked 6 th ); hence using Co 3 (PO 4 ) 2 will be more satisfactory.  First, the amounts of Li removed/reacted due to reactions after the Li-reactive coating process were measured (Table S.4). The amounts of LiOH and Li 2 CO 3 were measured using titration. This is because during the coating process at 720 °C, these Li compounds are transformed to Li 2 O then they revert to LiOH and Li 2 CO 3 due to reaction with H 2 O and CO 2 impurities 17,27,28 . The result indicates that Co 3 (PO 4 ) 2 can remove almost 70% of residual Li of the uncoated cathode, and Fe-and Mn-P are also shown to be effective (67% and 64% of Li residues were reduced, respectively). Meanwhile, TiPO 4 exhibited moderate performance; 51% of Li was removed. We associate this result with the simulations by comparing their ΔH Li-M values from the most preferable reaction of each material (reaction #1, the case for Calc a ) as shown in Fig. 3. We think this comparison is reasonable because a larger ΔH Li-M value from simulations means that this reaction is more likely to happen, so it can remove more residual Li during the experiment. The general trend is reasonably in agreement, i.e., coating materials with better experimental Li removal capacities (Co-, Fe-, and Mn-P) also have larger values of ΔH Li-M , while TiPO 4 exhibited poorer performance and has a lower value of ΔH Li-M .

Li
Effect of coating on the electrochemical performance of cathode. The energy dispersive X-ray spectrometric (EDS) elemental maps of the metal elements in the MPs are shown in Fig. 4a. Unlike the pristine structure shown in Figure S.3, it clearly confirms the presence of these coating materials, which are distributed on the surface of the primary and the secondary particles of the NCM. The HAADF (high-angle annular dark-field imaging)-STEM (scanning transmission electron microscopy) images and the quantitative EDS mapping data for all of elements are presented in the SI.
To understand the effect of coating material on the electrochemical performance of NCM cathodes, the Coulombic efficiency, initial capacity, and capacity retention rate (CRR) during cycling were measured ( Table 1). The Coulombic efficiency values exhibited by all the coating materials were superior to those of the bare material. Their 1 st capacity at 0.2 C rate exhibited that the Co 3 (PO 4 ) 2 -coated sample possessed the largest capacity (219.08 mAh/g), even larger than that of the bare (uncoated) material (217.47 mAh/g). This unusual behavior can be attributed to the formation of LiCoO 2 phase on the surface after reacting with Li compounds, which is suggested from the current simulation result (further discussion will be provided in the following section).
Other coating materials had initial capacity reduced by around 10 mAh/g relative to the uncoated case. The capacities of all coated materials dropped significantly (around 30 mAh/g) due to the faster C-rate after measuring the 1 st capacity at 1 C rate capacity. This behavior was more severe in the case of TiPO 4 ; the capacity decreased from 208.53 to 143.05 mAh/g after changing the rate from 0.2 C to 1 C.
The trend and values of capacity and CRR for all materials up to 50 cycles with 1 C rate are presented in Fig. 4b,c and Table 1. Co 3 (PO 4 ) 2 coating exhibited the best performance considering its 1 st cycle capacity and the CRR value (76.68%); both properties were larger than those of other materials considered in this study. Fe 3 (PO 4 ) 2 and Mn 3 (PO 4 ) 2 coating materials also demonstrated great performance in terms of CRR (77.16% and 73.08%, respectively), which are larger than that of the uncoated case (72.98%). Although the 1 C capacities with Fe 3 (PO 4 ) 2 and Mn 3 (PO 4 ) 2 coatings are initially smaller than that of the uncoated case, their capacities are expected to be conserved better during cycling considering their CRR values and the trend of curves, compared to that from the uncoated case (Fig. 4b,c). The TiPO 4 coating exhibits poor performance on all electrochemical properties; its CRR rate is very low (55.12%) and hence, using this material is not desirable. In summary, Co-, Fe-, and Mn-P materials can reduce significant amounts of Li residue and they also exhibit great electrochemical properties as coating materials, while TiPO 4 is neither effective at removing Li residues nor at conserving the capacity of the cathode during cycling.

Equilibrium phase verification.
To elucidate the equilibrium phase formed on the surface of the cathode after residual Li-removal by coating materials, a simulated experiment was performed by reacting MP materials directly with the Li compounds, followed by comparison of their phases with those obtained in the calculated phase diagram.
First, we constructed the Co 3 (PO 4 ) 2 -Li 2 O -O 2 phase diagram and obtained the reaction product, as shown in Fig. 5a,b. The stable products generated between MP and Li 2 O are important to understand the direct reaction mechanism between the coating materials and residual Li. Since the experiment is usually performed under an O 2 environment, it is also critical to provide an O 2 axis and investigate the products that newly emerge. It should be noted that in general more phases are available under O 2 flow. The complete list of phases available from simulations of other coating materials is tabulated in Table 2.
The calculated equilibrium phase of Co 3 (PO 4 ) 2 when the molar ratio of 2:1 is LiCoPO 4 (reaction #O1), exhibiting good agreement with the experimental result. A previous study also demonstrated that LiCoPO 4 phase can be formed on the surface of NCA cathode material after Li-reactive reaction with Co 3 (PO 4 ) 2 coating material 16 . LiMnP 2 O 7 or Li 2 MnP 2 O 7 phase can be generated from Mn 3 (PO 4 ) 2 coating material after its Li-reactive reaction when the molar ratio is 2:1 or 1:1, respectively; the latter form shows agreement with experiment. Finally, the   22 , which agrees with current calculations. Figure 6 provides further validation for the Fe-and Ti-P coating materials. For Fe 3 (PO 4 ) 2 , it is important to mention that since the molar ratio of coating materials to residual Li used in the simulated experiment for XRD analysis (molar ratio is 2:1) are much larger than in the actual coating (1 wt% of NCM, molar ratio around 0.3:1), the existing phases after coating can be different. For example, when the molar ratio of 2:1 Fe 3 (PO 4 ) 2 to Li 2 O is provided in a simulated experiment, the possible reaction is #O2 or #O3 in Table S.5, whose equilibrium phase is Li 2 Fe 3 (P 2 O 7 ) 2 or LiFeP 2 O 7 , respectively (the experimentally observed phase is LiFeP 2 O 7 , as shown in Fig. 6c). However, when a smaller molar ratio of coating material is provided, the Li-reacted phase could vary (such as LiFePO 4 , Li 3 PO 4 , Li 2 FeO 3 , etc.) depending on the amount of coating material. Similarly, the possible phases for    3 or Li 3 PO 4 provided the molar ratio (2:1) is between #O1 (6:1) and #O2 (0.67:1) in Table S.5 (the former phase matches the simulated experiment in Fig. 6f); when a smaller molar ratio is provided, the phase could be Li 3 PO 4 , LiTi 2 O 4 , etc.
In summary, Li-containing phases formed from the simulated experiment were verified by comparing with the phases predicted by calculations with molar ratios based on the phase diagram. This indicates that the phase on the cathode surface can be tuned by altering the amount of coating material with respect to that of the Li residue.

Conclusions
In this study, we implemented a screening process by constructing phase diagrams based on a first-principles approach to propose the optimal phosphate coating materials that can effectively reduce the amount of residual Li (Li 2 O). Verification of this framework was achieved by performing experiments measuring the remaining Li residue and obtaining electrochemical properties during cycling. Based on the ΔH Li-M values from calculations, the MP coating materials Co 3 (PO 4 ) 2 , Mn 3 (PO 4 ) 2 , Fe 3 (PO 4 ) 2 , and TiPO 4 were chosen for experimental validation. It was found that the order of reactivity of coating materials for removing Li residues was in good agreement between simulations and experiments; Co-, Mn-, and Fe-P materials exhibited great Li-removal capability. To further confirm the functionalities of coating material on the improvement of cycle life, electrochemical cycling tests showed that Co-, Mn-, and Fe-P materials are effective coating materials for the prevention of capacity fading behaviors. Co 3 (PO 4 ) 2 showed the largest initial capacity among the coating materials tested and its capacity was conserved well, indicating that it would be an ideal coating material for NCM cathode material. In addition, experimental products generated after the reaction of MP and Li were confirmed by comparison with predicted phases obtained from the phase diagram based on calculations.

Methods
Computational details: Screening process. We identified the optimal coating materials based on results from first-principles calculations by constructing the phase diagram to obtain Li-containing equilibrium phases. A total of 16 MP materials were calculated with the information of reaction enthalpy (ΔH Li-M ) and gravimetric capacity (G C ) when they react with residual Li. ΔH Li-M indicates the reaction energy when reactants are changed to products while G C is the weight of coating material required for the removal of one mole of Li 2 O. Gaseous products generated during this reaction can be disregarded because the Li-reactive coating process occurs at 700 °C; these gases can be removed during ventilation. The overall screening process, computational details, and the database for the formation energy of all materials used in this study are adopted from our previous work 29 .
Here is a brief overview of the computational approach.  3) The reaction equations were obtained based on step 2) with their reaction products and enthalpies. 4) The design chart containing ΔH Li-M and G C information was constructed to identify the ideal MP coating material for removal of Li residue.
Unlike our preceding study, here we focused on the reaction of MP with Li 2 O instead of LiOH and Li 2 CO 3 . This is because the decomposition reactions of both LiOH and Li 2 CO 3 to Li 2 O could occur beyond the temperature of 700 °C (the coating process in this study was performed at 720 °C), following reactions based on the JANAF-NIMS thermochemistry data table 34 and post-processed data from our previous study 28 ,  Supplementary Information, SI). The coating process was achieved as follows. The stoichiometric metal source was dissolved in DI water. After the NCM powder was suspended in the metal solution, DMP solution was slowly added with a dropper. The solution was then stirred and dried at 120 °C until the solvent evaporated completely. The resulting coated NCM substrate was heated at 720 °C for 5 h in flowing O 2 gas (30 liter/min).
Structural analysis for Li-reactive coating. Simulated experiments were carried out using the same coating materials composed of metal nitrates (the equivalent source as that used for surface coating), DMP, and Li residues (LiOH and Li 2 CO 3 ). It was assumed that during the coating process, the Li residues and coating materials (excluding NCM) participated in the reaction. The molar ratio of coating material to Li residue was 2:1. After each material was mixed, the mixture was heated to 720 °C under O 2 for 5 h.
The material was analyzed after heat treatment using X-ray powder diffraction (XRD) to identify the equilibrium phases in the coating materials. Structural examination of the sample was performed by an X-ray diffractometer using Cu-K α radiation with a scan speed of 0.02° per minute between 10° and 90° at an applied potential of 40 kV and current of 40 mA.
Microstructure analysis was performed using double Cs corrected transmission electron microscopy (TEM, FEI titan cubed 60-300). The composition of the particle surface was confirmed by energy dispersive spectroscopy (EDS, Bruker Super-X).
Electrochemical measurements. Composite positive electrodes containing 92 wt% active material, 4 wt% Denka black, and 4 wt% polyvinylidene difluoride (PVdF) were fabricated and pasted on the current collector (aluminum foil). The electrodes were dried at 120 °C under vacuum and then pressed. Metallic lithium was used as the counter electrode. The electrolyte solution consisted of 1.0 M LiPF 6 dissolved in a solution of fluoroethylene carbonate (FEC) and dimethylene carbonate (DMC). CR2032-type coin cells were assembled in a dry room. The cells were discharged and charged galvanostatically and the measurements were conducted in triplicate at each test condition. The loading level for the active ingredient was 10 mg/cm 2 . The cycling performance of the cells was measured at 25 °C at a charge/discharge rate of 1 C.
The amount of Li residue was estimated by the titration method. Since Li 2 CO 3 and LiOH are soluble in water, most of the Li sources were assumed to originate from these compounds. The total Li (ppm) was calculated from the following equation.
Data availability. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.