Modifying Surface Chemistry to Enhance the Electrochemical Stability of Nickel‐Rich Cathode Materials

Residual impurities such as lithium carbonate and hydroxide are a major concern for accelerating parasitic reactions at the cathode electrolyte interface of lithium‐ion batteries. Removal of these lithium‐bearing species becomes a necessity for high‐performance nickel‐rich cathode materials. Instead of directly removing these impurities through washing steps, a wet impregnation process is employed to convert these detrimental surface impurities into beneficial surface coating on nickel‐rich cathode materials. Specifically, the pristine cathode material is treated with Al(H2PO4)3 solution to convert undesired compounds into Li3PO4 and AlPO4, both of which are considered positive surface coating materials for high‐voltage cathodes. It is found that the introduced modification greatly suppresses the interfacial impedance hike and improves the capacity retention of the cathode material after repeating charging/discharging. It is believed that these benefits are realized through the modification of the surface chemistry of the cathode material, which helps to slow down the parasitic reactions and reduce the damage to the cathode material.


Introduction
Lithium transition metal oxides (LiTMO 2 , TM = Ni, Co, Mn) are the dominant cathode material for state-of-the-art lithiumion technologies.The limited availability and price fluctuation of critical materials like cobalt (Co) has become a major concern characteristics of the delithiated cathode material. [1]The accelerated loss of electrochemical performance for nickel-rich cathodes is widely attributed to either the increasing interfacial reactivity of delithiated cathodes and/or the reduced environmental compatibility with the increase of the nickel content.It has been widely accepted that charging a nickel-rich cathode to a relatively high potential will lead to an increase in Ni 3+ /Ni 4+ content in the charged cathode.This is especially true and important at the surface layer of the cathode particles where the chemically/electrochemically highly reactive Ni 3+ /Ni 4+ tends to lead to the eventual release of O 2 (forming rock salt structure on the interface layer) or oxidizing the electrolyte components (impeding the Li + transport across the cathode electrolyte interface).[8] It has been repeatedly demonstrated in the open literature that the interfacial reactivity of nickel-rich cathode can be efficiently mitigated through a functional surface coating, [9][10][11][12][13][14][15][16][17] although the physical role of surface coating is still under debate. [5,18,19]he second intrinsic technical challenge of nickel-rich cathodes is their sensitivity toward the ambient environment, particularly moisture and CO 2 .It was previously reported that Li 2 CO 3 can form and deposit on the surface of fresh nickel-rich cathode materials during the cooling process after sintering. [20]In addition, more Li 2 CO 3 can form during the storage period when exposed to the ambient condition; accelerated loss of electrochemical performance was also observed after exposing to ambient air. [21,22][25][26][27][28] Therefore, the removal of residual Li 2 CO 3 is a crucial step for synthesizing high-performance nickel-rich cathodes. [29]o fundamentally understand the nature of parasitic reactions occurring at the cathode/electrolyte interface, high-precision leakage current measurement was conducted to quantify the rate of electron transfer across the cathode/electrolyte interface. [2,3]t was found that the exposed transition metal cations on the cathode surface are active sites to form coordination bond with carboxyl group (>C═O), promoting the oxidation of the carbonate solvents on the cathode surface. [4]It was also demonstrated that this catalytic effect can be effectively mitigated by poisoning active sites using stronger coordinating molecules, such as organic cyanides. [4]We also reported that PO 4 − is another good candidate to preoccupy the active sites and to enhance the electrochemical performance of nickel-rich cathodes. [30]Bearing the need to get rid of detrimental Li 2 CO 3 residue and the beneficial impact of PO 4 − , a simple wet-impregnation process is adopted in this work to in situ convert detrimental Li 2 CO 3 to a beneficial phosphate coating material.It is demonstrated that the surface modification greatly suppresses the impedance hike and improves the capacity retention of LiNi 0.83 Mn 0.1 Co 0.07 O 2 .

Results and Discussion
The focus of this work is to develop a simple process to accomplish both goals, removing residual Li 2 CO 3 and depositing phosphate-based coating material, to improve the stability of cathode/electrolyte interface for nickel-rich cathode materials.Without losing generality, the model cathode material used for this study is LiNi 0.83 Mn 0.1 Co 0.07 O 2 (Ni83) synthesized in a dry air environment.As we previously reported, synthesis in a pure oxygen environment is not a necessity if the excessive amount of residual Li 2 CO 3 on the cathode surface can be properly mitigated. [28]Our initial trial was executed by soaking the pristine Ni83 powder in a diluted H 3 PO 4 solution and baking the filtered powder at an elevated temperature (750 °C), presuming that the weak acid environment will react with Li 2 CO 3 residual and form Li 3 PO 4 on the surface of Ni83.High energy X-ray diffraction pattern of the modified Ni83 clearly shows the deposition of PO 4 − bearing species on the surface of Ni83 (see Figure S1a, Supporting Information).However, the electrochemical performance of the modified Ni83 is worse than the pristine one (see Figure S1b, Supporting Information).We speculated that the unexpected result can be attributed to the excessive amount of H 3 PO 4 that promotes the dissolution of transition metal in the acidic environment and generates structural defect in the surface layer of Ni83 powder.Figure S1c (Supporting Information) shows images of recovered liquid phase after solid/liquid separation.The liquid phase containing 5 wt.%H 3 PO 4 turned green after soaking Ni83 powder, clearly indicating a dissolution of transition metals out of the Ni83 powder.
To gain precise control over the interaction between the cathode material and the processing solution, a wet impregnation process was adopted.In this process, as shown in Figure 1, the cathode powder was mixed with ≈20 wt.% of Al(H 2 PO 4 ) 3 solution before mixing with an acoustic mixer.After mixing, the wet powder without residual liquid was then baked at 550 °C for 6 h to remove the absorbed water and to deposit the newly formed metal phosphates on the cathode surface.The desired amount of liquid solution is experimentally determined by the weight loss of wet cathode powder after baking at 550 °C overnight.For the specific Ni83 cathode, the maximum absorption capability was ≈15 wt.%; an extra 5 wt.% was used to assure full coverage of the cathode surface.At the same time, an automatic titrator was utilized to determine the content of residual LiOH/Li 2 CO 3 on the cathode surface; ≈0.43 wt.% LiOH and 1.0 wt.% Li 2 CO 3 was found as the residual impurities on the Ni83 powder (see Figure S2, Supporting Information).The obtained impurity content was then used to determine the desired Al(H 2 PO 4 ) 3 concentration in the processing solution.Based on these numbers, a 5 wt.%Al(H 2 PO 4 ) 3 in the processing solution would nominally be just enough to fully react with all residual LiOH/Li 2 CO 3 impurities while a processing solution containing 1 wt.%Al(H 2 PO 4 ) 3 will leave some LiOH/Li 2 CO 3 on the surface layer of Ni83 cathode but covering the surface with a thin layer of Li 3 PO 4 /AlPO 4 to block the active sites for parasitic reactions.
From scanning electron microscopy, no significant difference can be identified between the pristine and surface modified Ni83 (Figure 2a-c) at the secondary particle level with particle sizes ranging from 5 to 25 microns in size.However, at a lower magnification, one could conclude that the pristine sample has a significant agglomeration of secondary particles, which has been previously [31] attributed to the binding effect of residual LiOH/Li 2 CO 3 .When treated with an Al(H 2 PO 4 ) 3 solution, the aggregation mostly disappeared.Synchrotron X-ray diffraction patterns (Figure 2d-f) of the pristine Ni83 and the ones modified with 1 and 5 wt.% solution of Al(H 2 PO 4 ) 3 , also reveals nearly no difference among samples.Further analysis by Rietveld refinement (Table S1, Supporting Information) confirms that the surface modification step does not alter the underlying crystal structure of the Ni83 powder.
Further investigation into the surface coating via transmission electron microscopy (TEM) reveals clear evidence of a coating layer on the surface of a representative Ni83 particle.Figure 3a shows a normalized line scan from surface to bulk (corresponding to Figure 3b) of a pristine Ni83 particle.We found a relatively high oxygen intensity at the surface layer of the Ni83 particle, which could be related to the surface impurities in the form of lithium hydroxide and lithium carbonate.Signals of Ni, Co, and Mn increased when moving toward the bulk of the sample, which is to be expected due to the increasing thickness in the bulk.The right panel of Figure 3b shows the elemental mapping of the various species, indicating a relatively uniform distribution of all species.It is noteworthy that the apparent signals for Al and P for Ni83 are an artifact stemming from signal normalization, which means that the mapping signals for Al and P are  in fact very weak.This is demonstrated in Figure S3 (Supporting Information) where we plotted the absolute intensity along with the line scan from the surface to bulk, it is showing the Al and P peak intensity are near 0, and the spectrum also showing P amount is 0%, while Al is at 0.2%, which could be from adventitious source of Al such as contamination during calcination from Al 2 O 3 crucible.In contrast, the sample was modified with 1 wt.%Al(H 2 PO 4 ) 3 solution revealed a significant amount of P and Al on the surface of the Ni83 particle (Figure 3c), while the absolute intensity of Al and P along the line scan is also observed to be very low, consistent with the relatively small amount coating (Figure S3, Supporting Information).Moving into the bulk, the Al and P signal returned to baseline levels comparable to the pristine sample.Signals of Ni, Co, and Mn remain essentially unchanged in comparison to the untreated sample.In addition to the persistently high intensity of oxygen on the surface of the particles, this provide strong evidence that the Al and Pbased species likely exist in the form of aluminum phosphate and lithium phosphate, respectively, and are coated onto the surface of the Ni83.Interestingly, beyond a thin layer of Al and P-based coating, the P mapping shown in the right panel of Figure 3d, reveals patches of P distributed throughout the surface of the Nirich NMC.No such patches were found for Al.This suggests that the wet impregnation process produces a thin layer of P and Al over the surface of Ni83 in addition to localized Li 3 PO 4 -rich clusters, which can be the result of a localized in situ reaction between LiOH/Li 2 CO 3 with Al(H 2 PO 4 ) 3 solution, converting part of carbonates into phosphates.This was also seen in adjacent particles found under the microscope and can likely be taken as a proxy for the entire particle population.
To determine the efficiency of lithium carbonate removal, we employed X-ray photoelectron spectroscopy to study the surface composition of the Ni83 samples before and after the surface modification.Figure 3e,f reveals that no Al and P are found over the surface of the sample prior to the treatment.Figure 3g shows a significant amount of carbonate species at the C1s binding energy level centered at ≈289.7 eV.The Ni2p 3/2 spectrum shown in Figure 3h reveals a typical profile for NMC materials.With the surface modification, the signal of Al and P increased substantially (Figure 3i,j, respectively), suggesting the existence of aluminum and lithium phosphate species with signals at 74.3 eV at the Al 2p [32] along with 135 and 133.4 eV for the P 2p level, representing AlPO 4 32 and Li 3 PO 4 , [33] respectively.The large peak at ≈73 eV at Al 2p is likely Al 2 O 3 formed during the sintering process after wet treatment. [34]More interesting, the amount of carbonate decreased substantially, with some leftover probably buried under the newly formed phosphate coating layer as expected by the design.As shown in Figure 3k, the magnitude of the peak at ≈289.7 eV decreased substantially in relation to the adventitious carbon peak at ≈284.7 eV.This indicates that the carbonate content has decreased substantially after subjecting the Ni83 to surface modification with 1 wt.%Al(H 2 PO 4 ) 3 solution.
In contrast to other processing techniques, [35] the Ni2p 3/2 peak of the modified sample (Figure 3l) did not reveal any change in the bonding states, confirming the ability to preserve chemical environment of Ni on the surface after the surface modification.
The electrochemical performance of the surface-modified samples reveals great improvements in terms of cyclability.Figure 4a displays a single charge/discharge voltage profile at C/10 of the pristine Ni83 and ones modified with a 1, 3, and 5 wt.% solution of Al(H 2 PO4) 3 in half cells.At the 3 and 5 wt.% level of Al(H 2 PO 4 ) 3 , the specific capacity is slightly reduced to ≈203 mAh g −1 in comparison to the ≈213 mAh g −1 of the pristine and the 1 wt.%Al(H 2 PO 4 ) 3 sample.Furthermore, the overpotential observed in the first charge appears to be higher for the sample modified with higher concentration (3 and 5 wt.%Al(H 2 PO 4 ) 3 ), whereas the one modified with 1 wt.% yielded slightly lower charging over-potentials in comparison to the pristine sample.We speculate that the relatively higher overpotential for the sample modified with higher concentration (3 and 5 wt.%Al(H 2 PO 4 ) 3 ) is likely due to the excessive amount of coating layers present at higher concentration level.The main benefit of our surface modification can be seen after cycled for 50 cycles.All samples modified with 1, 3, and 5 wt.%Al(H 2 PO 4 ) 3 yielded much higher capacity retention (≈82%) in comparison to the pristine sample (≈76%).This implies that while the sample modified with higher concentration (3 and 5 wt.%Al(H 2 PO 4 ) 3 ) has a higher initial interfacial impedance, the addition of the coating layer and removal of lithium carbonate clearly shows positive impact on the cyclability.This aligns well with previous reports that discussed the intricacy of the detrimental effect of lithium carbonate. [25]Reflecting a similar phenomenon, the dq dv −1 of the voltage profiles at cycles 1, 3, 10, 20, and 50 (Figure 4c), reveals that although Ni83 modified with higher concentration (3 and 5 wt.%Al(H 2 PO 4 ) 3 ) has the higher impedance, over cycling, the impedance increases much faster for the pristine sample as demonstrated in the shift in peak position of the H2-H3 transition (4.1-4.3V) which becomes most apparent at the 50th cycle.The sample was modified with 1 wt.%Al(H 2 PO 4 ) 3 solution yielded the best performance.
To evaluate the impact of surface modification on the interfacial impedance and impedance hike, we performed electrochemical impedance spectroscopy (EIS) on cells after 50th charge/discharge cycle and 1st cycle of samples treated with 1 and 5 wt.%.Before we discuss the implications of these findings, we want to clarify our interpretation of each impedance parameter.First, it is important to point out that the R int discussed here is unlikely to be the R SEI that is associated to the Li metal impedance.This is because the impedance of the Li metal decreases with plating/stripping as the surface area of exposed Li increases.Since the R int remains mostly unchanged as a function of SOC even as the Li metal reference/counter electrode is cycled, it suggests that R int is more related to the cathode rather than the Li metal and is shown to be present even in symmetrical coin cells as shown in Figure S4 (Supporting Information).Although it is difficult to precisely determine what R int represents, we will tentatively, for this study, interpret R int as the existence of some surface layer that hinders Li-ion transport.R ct is much more definitive in its interpretation as the time constant is significantly slower than contributions from the Li metal anode.Typically, R ct is considered associated with the exchange current of the cathode active materials, which is related to the available electrochemically active surface area of the Ni-rich NMC.For the cell to exhibit an increase in R ct , the surface area of the cathode active material must be blocked to the point that its exchange current becomes negligible.On the other hand, an increase in R int suggests that the electrochemically active surface areas are blocked but can still slightly contribute to the exchange current.
Figure 4d shows a Nyquist plot of the pristine, and samples modified with 1 and 5 wt.%Al(H 2 PO 4 ) 3 solutions at the 1st and the 50th cycle at fully charged state (4.4 V).From the inset of Figure 4d, the first semi-circle revealed that initially, the freshly made cells had slightly different impedance values.These include a relatively higher interfacial resistance (2nd semicircle, mid frequency, R int ) for Ni83 modified with 5 wt.%Al(H 2 PO 4 ) 3 solution while a larger low-frequency semicircle (charge transfer resistance, R ct ) was observed for the pristine Ni83.These differences were quickly dwarfed by the impedance growth observed later during cycling.At the 50th cycle, the pristine Ni83 showed a significantly larger R ct .Specifically, the pristine Ni83 was estimated to possess an R ct value of ≈170 ohms from EIS fitting whereas the samples modified with 1 and 5 wt.%Al(H 2 PO 4 ) 3 solution only possessed ≈100 and ≈60 ohms.The R int also showed a similar trend with the pristine Ni83 yielding the highest at ≈10 ohms and the modified samples both possessing lower impedance values of ≈6 ohms as shown in Figure 4e.It might be tempting to suggest that the overpotential increase observed in Figure 4c is overwhelmingly dominated by R ct and that R int is insubstantial.
We performed in situ EIS where the cells at the freshly-made state and after 50 cycles were charged with the EIS intermittently measured at designated capacity intervals.From Figure 4f, one can clearly deduce that 1) R int changes significantly from the 1st to the 50th cycle and 2) difference between the cells follows the general trend of pristine Ni83 > Ni83 modified with 5 wt.%Al(H 2 PO 4 ) 3 > Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 in the order of decreasing R int .Both trends align well with the single-point EIS measurement shown in Figure 4d.However, across all samples, the value of R int remains essentially unchanged respective to the state of charge (SOC).The evolution of R ct as a function of SOC (Figure 4g) is more intricate than that of R int and shifts the contribution of R ct as a dominant source of impedance to one that is comparable in magnitude to R int .Initially, the R ct values are large with the pristine and Ni83 modified with 5 wt.%Al(H 2 PO 4 ) 3, showing nearly 300 ohms in resistance, while Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 possessed 230 ohms.Upon delithiation of the Ni83 materials, the R ct decreases rapidly with SOC, which is known to be attributed to the autocatalytic nature of the initial stages of delithiation [36] During this low R ct region (potential of ≈3.8 to 4.2 V), the impedance dropped down significantly to ≈30, 17, and 10 ohms for pristine Ni83, and Ni83 modified with 1 and 5 wt.%Al(H 2 PO 4 ) 3 , which in turn renders the contribution of R int significant (≈16, ≈12.5, and ≈11 ohms, respectively for pristine, 1 and 5 wt.%Al(H 2 PO 4 ) 3 ).At the end of charge, R ct starts to drastically increase again (due to lack of Li content in Ni83 particles) while R int remains mostly the same and thus, the contribution of R int becomes once again, negligible at the end of charge.Since most of the charge capacity resides in the middle voltage region where R ct has decreased and the contribution of R int is large, we can conclude that both R int and R ct are in fact critical in determining the observed impedance observed in the dQdV −1 in Figure 4c.Overall, the cells using Ni83 cathode with and without surface modification had similar initial charge transfer resistance.The cell using pristine Ni83 had the lowest initial interfacial resistance, followed by the one modified with 1 wt.%Al(H 2 PO 4 ) 3 modified, and the one modified with 5 wt.%Al(H 2 PO 4 ) 3 had the highest initial interfacial resistance.It was observed that all cells experienced significant impedance hike after 50 cycles.It is also clear that the cell using 1 wt.%Al(H 2 PO 4 ) 3 modified Ni83 had the slowest hike on both the interfacial resistance and the charge transfer resistance.
To identify the nature of the interfacial layer on the cathode particles, high-resolution transmission electron microscopy (HRTEM) was performed to study the surface condition of both the pristine Ni83 (Figure 5a) and Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 (Figure 5b) after 50 cycles.The inset selected area electron diffraction (SAED) patterns show the different phases of the pristine Ni83 and the one modified with 1 wt.%Al(H 2 PO 4 ) 3 at different locations.A layered structure exists in the bulk of both materials but at the surface, there are clear evidence of the formation of rock salt phases on pristine Ni83 after cycling.More importantly, the thickness for the rock-salt structure in pristine Ni83 is much thicker than that observed in Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 .As shown in Figure 5b, 1 wt.%Al(H 2 PO 4 ) 3 modified Ni83 exhibited a minimal transformation from the pristine layered phase to rock salt structure.It should be noted that the Li/Al phosphate coating is likely amorphous hence the lack of observable structure from the HRTEM.Electron energy loss spectroscopy (EELS) at the Ni K-edge reveals (Figure 5c,d) a shift in the L 3 and L 2 peak toward a lower energy level after 50 cycles for both the pristine Ni83 and 1 wt.%Al(H 2 PO 4 ) 3 modified Ni83 (respectively) sampled from the bulk to the surface of the NMC particles.For the pristine Ni83, we found that the shift was much larger at 1.64 versus 1.29 eV for the sample treated with 1 wt.%Al(H 2 PO 4 ) 3 .This shift to a lower energy is often attributed to the reduction in the Ni oxidation state, an indication of redox reaction between the oxidative cathode and the reducing electrolyte (solvents).No significant peak shifts were observed for Mn and Co L-edges (Figure 5e-h).Post-cycling analysis of Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 confirms that the coating remains very much intact after 50 cycles.Line scan (Figure S5a, Supporting Information) and selected area scans (Figure S5b-d, Supporting Information) clearly suggests that the surface is still rich in P and Al elements derived from the coating.Elemental mapping further corroborates this point with visual confirmation of the coating as shown in the surface concentrated P and Al maps (Figure S5e, Supporting Information).
To further study the reactivity of the electrolyte against the different cathodes at different voltages, differential electrochemical mass spectrometry (DEMS) was utilized to in situ track the evolution of gas species during charging/discharging. Shown in Figure S6a (Supporting Information), the pristine Ni83 produces a significant amount of CO 2 when charged up to 4.7 V. Whereas, the cells using Ni83 modified with 1 wt.%Al(H 2 PO 4 ) 3 produced significantly less amounts of CO 2 (Figure S6b, Supporting Information).There are two possible sources for the release of CO 2 at the high potential.The former is the electrochemical oxidation of carbonate solvents at a relatively high potential, releasing both CO 2 and protons or protons bearing acidic species.The latter is the decomposition of Li 2 CO 3 at a relatively high potential. [37]ince further oxidation of Li 2 CO 3 is thermodynamically prohibited, the CO 2 release should be associated with the chemical reaction between Li 2 CO 3 with acidic species generated from the oxidation of carbonate solvents.With both CO 2 generation sources, the reduction in CO 2 release can be attributed to the suppression of electron transfer reaction between the delithiated cathode and the electrolyte, which can be accomplished by covering the active reaction sites, both exposure transitional metal atoms [4] and residual Li 2 CO 3 , [28] with a thin layer of phosphates.

Conclusion
The Li 2 CO 3 residual has a detrimental impact on the interfacial stability of nickel-rich cathodes.In this work, a wet impregnation process was utilized to partially convert detrimental LiOH/Li 2 CO 3 residuals into a beneficial Li 3 PO 3 /AlPO 4 surface coating layer on LiNi 0.83 Mn 0.1 Co 0.07 O 2 .XPS analysis of the treated cathode revealed a layer of phosphates on the surface of cathode powder, and a significant reduction of Li 2 CO 3 signal.Electrochemical characterization demonstrated that the proposed surface modification greatly improved the capacity retention of the nickel-rich cathode.It is concluded that the coating layer, comprising of Li 3 PO 4 and AlPO 4 , acts as the protection layer to block the chemical interaction between the carbonate solvents and the exposure transition metal atoms, as well as Li 2 CO 3 residual.The protection layer reduces the electron transfer reaction between the electrolyte and the delithiated cathode, releasing less CO 2 and protons or protons-bearing acidic species, and resulting in less detrimental phase transformation from layered structure to rock-salt structure on the surface of nickel-rich cathodes.

Figure 1 .
Figure 1.Schematics of the wet impregnation process for simultaneously removing Li 2 CO 3 and LiOH and the subsequent deposition of a beneficial surface coating layer composing of Li 3 PO 4 and aluminum phosphate compounds.

Figure 4 .
Figure 4. Electrochemical performance data of pristine and modified Ni83, a) 1st cycle voltage profile across all samples (C/10), b) cycling performance (C/3) along with the corresponding c) dQdV −1 at various cycle numbers across the four samples (C/10 for three formation cycles and then C/3 afterward).d) Nyquist plot of cells at freshly made conditions and at the 50th cycle across all three materials.Panel above indicates the circuit used to model the EIS data, inset shows an expanded view at the high-frequency impedance regime.e) Bar graph showing the fitted impedance values for the R int and R ct after the 1st and 50th charge across all samples.Fitted values for f) R int and g) R ct as a function of voltage over the first charging process and the 50th charging across all three samples.

Figure 5 .
Figure 5. High-resolution TEM after 50 cycles of a) untreated and b) 1% Al(H 2 PO 4 ) 3 samples.Included are selected area electron diffraction patterns along with enlarged sub-surface area images.EELS at the O K edge of c) untreated and d) 1% Al(H 2 PO 4 ) 3 samples at various positions from surface to bulk.e) Plot of Intensity ratio of O K-edge pre-peak to peak as a function of position from surface to bulk.EELS at the Ni L-edge of f) untreated sample and g) 1% Al(H 2 PO 4 ) 3 treated samples at various positions from surface to bulk.