Development of an Innovative Procedure for Lithium Plating Limitation and Characterization of 18650 Cycle Aged Cells for DCFC Automotive Applications

: Since lithium-ion batteries seem to be the most eligible technology to store energy for e-mobility applications, it is fundamental to focus attention on kilometric ranges and charging times. The optimization of the charging step can provide the appropriate tradeoff between time saving and preserving cell performance over the life cycle. The implementation of new multistage constant current proﬁles and related performances after 1000 cycles are presented and compared with respect to a reference proﬁle. A physicochemical (SEM, XRD, particle size analysis, etc.) and electrochemical (incremental capacity analysis, internal resistance measurements) characterization of the aged cells is shown and their possible implementation on board is discussed.


Introduction
In a wide panorama with several new emerging energy storage approaches [1][2][3][4][5][6] to guide the ecological transition, today lithium-ion batteries are considered among the most promising and mature technologies [7], with an increasing interest coming from the main players in the automotive field. Carmakers are always more and more pushed by governments to switch their fleets from internal combustion engines (ICEs) to hybrid and electric powertrains, with strict deadlines [8,9] and carbon neutrality targets by 2035.
Lithium-ion cells, showing significant energy (150-200 Wh/Kg) and power (300-1500 W/Kg) densities along with affordable durability, seem to cover a key role to trace this route [10]. One of the main topics the researchers are dealing with is related to the implementation of optimized fast charging profiles to decrease charging time and to limit cell degradation, maintaining performances and reducing capacity fade [11]. Among the wide panorama of charging approaches, multistage constant current charging (MCC) represents a good compromise between efficiency and easy implementation for a battery management system (BMS) [12].
Indeed, when working with lithium-ion cells, degradation inside the active materials is inevitable and directly affected by stress factors, such as high depth of discharge (DoD), high charging-rates, and low temperatures. During the life cycle of a lithium-ion cell, aging affects the performances of the electrochemical system, with the risk of also compromising safety [13,14].
Aging detection is challenging because of the interdependency of plenty of complex sets of degradation mechanisms, bringing capacity and power fade with chemical and The aging campaign was performed at 25 • C in a Falc climatic chamber, monitored by a Eurotherm controller (Como, Italy). Cells have been cycled with a BasyTec XCTS, controlled by BasyTec software (Asselfingen, Germany) (6.2.42.0 version).

Electrochemical Testing
Each cell was aged with a specific direct current fast charging (DCFC) profile, as reported in Figure 1, replicating the aging procedure defined in a previous study [11]. For this specific test, the end of life (EOL) was reached at 1000 cycles. The aim of this test was to try to limit lithium plating formation. The discharging step was chosen at 1C for all the cells, from 4.09 to 2.98 V.
Every 50 cycles, a capacity check was performed to control the aging trend. The capacity check consisted of two C/5 charge/discharge cycles between 2.50 V and 4.20 V.
A very slow charge/discharge (C/50) was performed before starting the aging protocol, to calculate the incremental capacity analysis (ICA) in quasi-stationary conditions. The same test was repeated for comparison at 1000 cycle (i.e., at the end of the aging campaign). Every 50 cycles, a capacity check was performed to control the aging trend. The capacity check consisted of two C/5 charge/discharge cycles between 2.50 V and 4.20 V.
A very slow charge/discharge (C/50) was performed before starting the aging protocol, to calculate the incremental capacity analysis (ICA) in quasi-stationary conditions. The same test was repeated for comparison at 1000 cycle (i.e., at the end of the aging campaign).

Incremental Capacity Analysis
As widely reported in the literature [32][33][34], ICA consists in the calculation of the derivative dQ/dV, displayed as a function of the operative voltage. Considering observed peaks into a typical ICA plot ( Figure 2-right side), we focused attention on the main transition phase during charge and discharge [32,33,35]. In fact, when the voltage profile curve ( Figure 2-left side) shows a small slope, with a limited increase of the cell potential, a transition phase in both electrodes is occurring (i.e., H1-M, M-H2 etc.). More details will be provided in Section 4.2. Here, the voltage response profile is shown as a function of the state of discharge (SoD%).

Incremental Capacity Analysis
As widely reported in the literature [32][33][34], ICA consists in the calculation of the derivative dQ/dV, displayed as a function of the operative voltage. Considering observed peaks into a typical ICA plot ( Figure 2-right side), we focused attention on the main transition phase during charge and discharge [32,33,35]. In fact, when the voltage profile curve ( Figure 2-left side) shows a small slope, with a limited increase of the cell potential, a transition phase in both electrodes is occurring (i.e., H 1 -M, M-H 2 etc.). More details will be provided in Section 4.2. Here, the voltage response profile is shown as a function of the state of discharge (SoD%). Every 50 cycles, a capacity check was performed to control the aging trend. The capacity check consisted of two C/5 charge/discharge cycles between 2.50 V and 4.20 V.
A very slow charge/discharge (C/50) was performed before starting the aging protocol, to calculate the incremental capacity analysis (ICA) in quasi-stationary conditions. The same test was repeated for comparison at 1000 cycle (i.e., at the end of the aging campaign).

Incremental Capacity Analysis
As widely reported in the literature [32][33][34], ICA consists in the calculation of the derivative dQ/dV, displayed as a function of the operative voltage. Considering observed peaks into a typical ICA plot ( Figure 2-right side), we focused attention on the main transition phase during charge and discharge [32,33,35]. In fact, when the voltage profile curve ( Figure 2-left side) shows a small slope, with a limited increase of the cell potential, a transition phase in both electrodes is occurring (i.e., H1-M, M-H2 etc.). More details will be provided in Section 4.2. Here, the voltage response profile is shown as a function of the state of discharge (SoD%).

Teardown and Sampling
The post-mortem characterization of the electrodes was performed through the dismantling of the cells. To operate in safe conditions, the cells were discharged to 2.2 V and opened in an Ar-filled MBraun glovebox (<0.1 ppm O 2 , <0.1 ppm H 2 O). The so-called "jellyroll", with a cylindrical shape, was composed by the staking of the anode and the cathode, separated by a polymeric foil. It was extracted from the aluminum external can, using a cutting tool (Dremel ® ) (Mount Prospect, USA) and unwounded.
The electrode samples were prepared inside the glovebox using a disk puncher (18 mm diameter), suitable for all the performed analyses. To increase the sampling statistics, it was decided to cut 10 electrode disks all over the original electrode foil surface, only where the electrode surface appears in good health on visual inspection. The samples were preserved in a specific box and selected randomly for all the electrochemical and XRD analyses.
Cathode and anode active materials were tested in multiple conditions. Samples powders scratched from the electrode foil were analyzed, using the standard full scale sample holder. Calendered samples picked up from the electrode foils in the original format were characterized in inert atmosphere, employing a specific XRD sample holder with a Kapton window. Both the powder samples (obtained by machining a significant amount of material) and the surface conditions (limited only to the material in contact with the separator on the calendered electrode) were objects of study.
A Rietveld refinement was performed on XRD patterns using PDF software in the 15 • < 2θ < 90 • range, considering a single-phase system composed by NMC811, together with a small fraction of carbon phase, associated to the conductive agent present in the electrode formulation. A trigonal structure (s.g.: R3m, 166) was assumed for the NMC811 phase as a model for the refinement.

SEM/EDX
SEM/EDX analyses were performed on a Zeiss Sigma Scanning Electron Microscope (Oberkochen, Germany) (SEM) on the disc samples. The anode images were collected by means of a 20 kV acceleration beam voltage and a working distance (WD) of 18.5 mm. Cathode images were acquired with a 20 kV acceleration beam voltage and a working distance (WD) of 16.0 mm. The Energy Disperse X-ray Analysis (EDX) spectra were collected at 20 kV on a 0.8 mm 2 area. To analyze the cross-section and measure the thickness electrodes, a small piece of sample was incorporated into a resin block that was successively polished and metalized using Ag.

Particle Size Analysis
NMC811 electrodes were composed by particles of active material dispersed in a polyvinylidene fluoride (PVDF) based binder matrix. Since the distribution of the particles dimension is a useful parameter to follow the aging, the progressive active material particles' disaggregation was checked and compared with the BOL (beginning of life) cell condition. An Antopaar 1090 PSA instrument was used to analyze NMC811 cathode particles materials.
The analysis was performed using a particle size analyzer based on the static light scattering principle. A laser beam was scattered by the dispersed particles sample and the intensity of light scattered was linked to the particles' diameter. The dispersion has been performed in a liquid medium (water).
For cathodic materials, the analysis was performed in liquid dispersion to properly spread the particles. Before starting the test, it was necessary to remove the PVDF. NMC811 electrode samples was treated at high temperature (350 • C for 1 h) and then the active material was detached from the Al current collector surface. The analysis was performed without ultrasound dispersion to avoid further particles' disaggregation. The pump and the stirring were set at a high velocity (250 rpm), repeating the test five times, to assess the stability and the reproducibility of the measurement. The particle size distribution was estimated through the Fraunhofer diffraction equation, which predicts the angular location of the scatter maxima and minima as a function of an object's size.

ICP-OES Analysis
For the quantification of the cathode composition, an inductively coupled plasma (ICP) analysis was performed on calcined cathode powder (350 • C-1 h). A total of 0.25 g of powder was weighed on an analytical balance (accuracy 0.0001 g), transferred to a volumetric flask (250 mL), and digested with a hot aqua regia solution on a hot plate; nickel, manganese, cobalt, and lithium were quantified using a method with three calibrated points (0 ppm, 25 ppm, 50 ppm).

Samples Dismantling Procedure
To deeply understand the aging conditions, it was decided to dismantle the original 18650 cell and to select several disks to assemble a lab scale three-electrode cell (Pat cell form EL-Cell ® , Hamburg, Germany) with 18 mm diameter cathode and anode samples.
To be sure to guarantee appropriate electric contact, it was necessary to remove the coating from one side to expose the current collector and close the electric circuit.
Electrode disks were washed using DMC (dimethyl carbonate). Three-electrode cells were assembled in an MBraun glove box.

Samples Selection
Ten samples from each foil were cut and preserved in a box. The cutting of the 18 mm diameter disks was performed by selecting sample areas with no apparent damage or significant lithium plating. It was difficult to use damaged areas to build new lab scale cells, thus they were used only for other characterization techniques.
To avoid preferential samples, three electrode lab scale cells (EL Cell setup, Hamburg, Germany) were assembled by selecting random electrode disks from the prepared samples.

Lab Scale Cell Building
A lithium pseudo-reference electrode was inserted in the cell to measure the anodic (Si-C vs. lithium) and the full cell potential. Cathodic potential was evaluated by difference. After assembling, cells were cycled at C/10 for the formation and then to check the residual capacity.
A C/50 charge-discharge cycle (2.5-4.2 V) was applied to calculate the incremental capacity analysis in three electrode conditions at a quasi-stationary state [35]. Figure 3 shows the aging trend displayed as SoH % as a function of the number of cycles for five 18650 cylindrical cells. Each cell was aged with a specific DCFC profile, tuned and selected as shown in a previous work [11], where several cells from other cycling campaigns were already aged up to 300 cycles with good reproducibility. At Figure S1, supporting information is visible on the updated trend up to 500 cycles of a previous aging campaign, confirming the reproducibility of experimental results.   Figure 4 shows the ICA analysis performed on reference and MCC2 profile cells, both at the beginning of life (BOL) and EOL conditions. See supporting information (S2) for the complete comparison. Further details will be provided in Section 4.2.   (Figure 5a), and two aged cells. Details about the methodology were discussed in a previous work [11]. Figure 5b,c shows the internal resistance evolution as a function of SoC% for the reference and MCC2 profile cells. The plating zone, if present, is marked with a red area. The complete absence of lithium plating conditions for the BOL cell is visible. On the contrary, the red area is present for EOL conditions. Further details will be provided in Section 4.2.  Figure 4 shows the ICA analysis performed on reference and MCC2 profile cells, both at the beginning of life (BOL) and EOL conditions. See supporting information ( Figure S2) for the complete comparison. Further details will be provided in Section 4.2.

Non-Disruptive Characterization Techniques
x FOR PEER REVIEW 6 of 17 Figure S1, supporting information is visible on the updated trend up to 500 cycles of a previous aging campaign, confirming the reproducibility of experimental results.  Figure 4 shows the ICA analysis performed on reference and MCC2 profile cells, both at the beginning of life (BOL) and EOL conditions. See supporting information (S2) for the complete comparison. Further details will be provided in Section 4.2.   (Figure 5a), and two aged cells. Details about the methodology were discussed in a previous work [11]. Figure 5b,c shows the internal resistance evolution as a function of SoC% for the reference and MCC2 profile cells. The plating zone, if present, is marked with a red area. The complete absence of lithium plating conditions for the BOL cell is visible. On the contrary, the red area is present for EOL conditions. Further details will be provided in Section 4.2.

Visual Inspection
After the dismantling of the cells, photography images were collected on all the samples (Figures S3-S7 on supporting information). Figure 6 shows details on anode electrode foils for all the implemented DCFC profiles samples. A visible lithium plating zone is present for samples subjected to reference and MCCFast2 DCFC cycling conditions, marked with a red arrow. Figure 5b,c shows the internal resistance evolution as a function of SoC% for the reference and MCC2 profile cells. The plating zone, if present, is marked with a red area. The complete absence of lithium plating conditions for the BOL cell is visible. On the contrary, the red area is present for EOL conditions. Further details will be provided in Section 4.2.

Visual Inspection
After the dismantling of the cells, photography images were collected on all the samples ( Figures S3-S7 on supporting information). Figure 6 shows details on anode electrode foils for all the implemented DCFC profiles samples. A visible lithium plating zone is present for samples subjected to reference and MCCFast2 DCFC cycling conditions, marked with a red arrow.  Figure 7a depicts a wide area of lithium plating depositions (red zone). Where lithium deposits are absent, it is possible to notice the graphite, even if the typical flake structure was not observable at this magnification. Figure 7b shows the NMC811 cathode surface. The particles appear severely damaged, with a diffused particle cracking and the important presence of disaggregated particles, as stated in the literature [36]. Figure 7c shows a detail from an MCCFast2 DCFC profile, where particles with no severe damage are also observable, notwithstanding the severity of the applied charging profile.
More images on the cathode side are available in the Supporting Information (S8). ICP-OES analysis performed on cathode samples confirm no significant damage. The transition metal ratio (Ni:Mn:Co) remains unchanged after 1000 cycles with respect to the BOL cell conditions. Figure 7d resumes the particle size analysis performed on NMC811 cathodes in all of the tested DCFC conditions, comparing the particle size distribution with a cell in fresh conditions, with no cycle aging. Where lithium deposits are absent, it is possible to notice the graphite, even if the typical flake structure was not observable at this magnification. Figure 7b shows the NMC811 cathode surface. The particles appear severely damaged, with a diffused particle cracking and the important presence of disaggregated particles, as stated in the literature [36]. Figure 7c shows a detail from an MCCFast2 DCFC profile, where particles with no severe damage are also observable, notwithstanding the severity of the applied charging profile.
More images on the cathode side are available in the Supporting Information ( Figure S8). ICP-OES analysis performed on cathode samples confirm no significant damage. The transition metal ratio (Ni:Mn:Co) remains unchanged after 1000 cycles with respect to the BOL cell conditions. Figure 7d resumes the particle size analysis performed on NMC811 cathodes in all of the tested DCFC conditions, comparing the particle size distribution with a cell in fresh conditions, with no cycle aging.

Three Electrode Lab Scale Cell-Incremental Capacity Analysis
Three electrode lab scale cells for all tested DCFC conditions were assembled to highlight the Si-C and NMC811 aging conditions by using a lithium pseudo reference electrode. In Figure 8, reference and MCC2 cells after 1000 cycles are displayed, with the fresh cell conditions (BOL) reported for comparison. More information is available in the Supporting Information (S9).

Three Electrode Lab Scale Cell-Incremental Capacity Analysis
Three electrode lab scale cells for all tested DCFC conditions were assembled to highlight the Si-C and NMC811 aging conditions by using a lithium pseudo reference electrode. In Figure 8, reference and MCC2 cells after 1000 cycles are displayed, with the fresh cell conditions (BOL) reported for comparison. More information is available in the Supporting Information ( Figure S9

Cycle Aging Campaign
The SoH trend vs. n° cycles ( Figure 3) showed a linear trend for MCC1, MCC2, and MCCFast1 samples. The MCCFast2 and the reference showed the worst performances, with a significant change in the trend at approximately cycle n° 700. In accordance with the literature [22,23,34,37], the pronounced change in the SoH capacity trend shown by MCCFast2 and reference samples indicate the beginning of the irreversible lithium plating phenomenon, with the concurrent loss of cyclable lithium. MCCFast2, even if started with the worst capacity retention, arrived at 1000 cycles with a capacity loss comparable to the reference profile.
As expected, MCC1 and MCC2 DCFC samples presented the best improved capacity retention, during the entire duration of the test, with respect to the reference.

Cycle Aging Campaign
The SoH trend vs. n • cycles ( Figure 3) showed a linear trend for MCC1, MCC2, and MCCFast1 samples. The MCCFast2 and the reference showed the worst performances, with a significant change in the trend at approximately cycle n • 700. In accordance with the literature [22,23,34,37], the pronounced change in the SoH capacity trend shown by MCCFast2 and reference samples indicate the beginning of the irreversible lithium plating phenomenon, with the concurrent loss of cyclable lithium. MCCFast2, even if started with the worst capacity retention, arrived at 1000 cycles with a capacity loss comparable to the reference profile.
As expected, MCC1 and MCC2 DCFC samples presented the best improved capacity retention, during the entire duration of the test, with respect to the reference.

Incremental Capacity Analysis (ICA) on Full 18650 Cells
All the ICA trends visible in Figure 4 highlighted the transition phases coming from the full 18650 cells, with contributions coming from both the Si-C anode and the NMC 811 cathode [32,38,39]. Considering one of the displayed curves from the plot, it is possible to identify a peak around 3.4 V, related to the lithiation (lithium intercalation) of the graphite (C 6 → LiC x ). From 3.6 and 4.1 V, there is a group of three main peaks, all attributed to the NMC 811 transition phases, respectively, H 1 → M (3.6 V), M → H 2 (3.9 V), and H 2 → H 3 (4.1 V). Peaks at 3.9 V and 4.1 V are characteristics of high Ni-content NMCs (Ni > 80%) [40]. The peak at 3.6 V indicates the transition phase H 1 → M (first hexagonal to monoclinic); at 3.9 V it is possible to notice the M → H 2 (monoclinic to second hexagonal); and, at 4.1 V, the H 2 → H 3 transition phase (second to third hexagonal phase transition). The damaging effect of the transition H 2 → H 3 is known from the literature because of a significant volume contraction linked to capacity fade [40][41][42].
Having a look at the dQ/dV vs. V trend in Figure 4, both reference and MCC2 samples show, after 1000 cycles, a general peak shift to higher voltages of the aged materials with respect to BOL. This behavior is indicative of a resistance growth. Cycle aging induced a concurrent shift and area reduction of the peaks from I to III of both 1000 cycles samples. The reference sample showed a more pronounced aging effect. The first peak (3.4 V) appears the most shifted and reduced because of the capacity loss, due to electrolyte and lithium losses, to form SEI (solid electrolyte interphase). The second (3.6 V) and the third (3.9 V) peaks exhibit an area decrease, in line with a general capacity loss, along with a small shift due to resistance increase. No evident changes in the initial crystal structure are visible since we do not observe any disappearance of peaks. Peak IV (H 2 → H 3 phase transition) seems to be unaltered at 1000 cycles, with a good reversibility of the transition phase and no detrimental lattice shrinkage along the c-direction [43].
The obtained results are in accordance with the expectations, since the main purpose of the present work was to limit aging by tuning the DCFC step properly. Incremental capacity analysis confirmed a very limited aging trend from the cathode point of view, for the entire cycle aging campaign.

Internal Resistance Evolution
The plating zone in both cases ( Figure 5) is depicted by a red area. Having a look at the BOL conditions, we can notice the total lack of plating evolution even at the highest C rates, within the limits of the test performed, as confirmed by the literature [44].
It is possible to see how, even if the current applied to the aged cells was reduced based on the measured capacity (after 1000 cycles), the inversion in the slope of the internal resistance trend at the highest C-rates (above 1C) takes place at low SoCs, if compared to the BOL conditions. See Supporting Information ( Figure S10) for the complete comparison.
As expected, a simple reduction of the current based on the measured capacity (Ah) is not enough to develop a derating function. To prevent irreversible lithium plating formation, a change in the cut-off voltage (or alternatively a more pronounced current reduction) is needed to protect the cells over their lifetime. Figure 6 shows the visual inspection of the Si-C anodes both from the dismantling of the MCC1, MCC2, MCCFast1, MCCFast2, and reference cells, after 1000 cycles. It is visible how the anode surface seems to be totally unaltered for MCC1 and MCC2 samples, except for a very small portion of the electrode affected by lithium plating, in the case of MCC2.

Visual Inspection
MCCFast1 shows a yellow zone, attributable to an initial lithium plating, even if in small quantities.
MCCFast2 and reference anodes present a significant amount of lithium deposition on both sides of the electrode (about 10 × 3 cm 2 ). Images ( Figure 6) are in good accordance with the electrical results summarized in Table 2, where it is possible to appreciate the SoH%, at the end of the cycling campaign. The presence of lithium deposits in the inner rolls of the cell, where present, indicates the concurrent effect of the high current applied (crucial effect when the cell aged) and a geometrical influence [45][46][47]. In fact, the pressure applied on the electrode foils increases by going from the outer to the inner layers, with a direct influence of a thick element useful to link and wrap the two parts of each electrode and separator foil. See Supporting Information (Figures S3-S7, yellow element).
Having a look at the charging profile ( Figure 9) we can see how MCCFast2 profile restores the highest amount of capacity (Ah) in the first 15 min, 20% more than with the reference, as reported in Table 3.

SoH %
88.48 91.53% 91.74% 90.88% 87.55% Having a look at the charging profile ( Figure 9) we can see how MCCFast2 profile restores the highest amount of capacity (Ah) in the first 15 min, 20% more than with the reference, as reported in Table 3. Because of the obtained results, we can affirm that this profile can be ideal to fast charge (under 30 min) the battery pack when needed, permitting, among the selected profiles, the highest energy throughput in the lowest amount of time. On the contrary, the reference DCFC profile, which aged more in comparison to the other faster DCFC profiles, also restores the lowest amount of energy after 30 min, with no evident advantages in the application. On the contrary, MCCFast2 results in an effective approach for occasional charging.   Because of the obtained results, we can affirm that this profile can be ideal to fast charge (under 30 min) the battery pack when needed, permitting, among the selected profiles, the highest energy throughput in the lowest amount of time. On the contrary, the reference DCFC profile, which aged more in comparison to the other faster DCFC profiles, also restores the lowest amount of energy after 30 min, with no evident advantages in the application. On the contrary, MCCFast2 results in an effective approach for occasional charging.

SEM, ICP-OES, and Particle Size Analysis
In Figure 7a, where the lithium plating is absent, it was expected to see the presence of the graphite structure, which seems to be covered by a uniform layer. The main hypothesis is relative to a thick SEI layer presence all over the electrode.
From Figure 7b we can assume that the reference DCFC condition caused more mechanical stress inside the NMC811 particles, if compared to the MCCFast2 condition, as seen in Figure 7c.
Even though the higher C-rates applied during the first steps of the charging profile, MCCFast2 seems to be a valid alternative to the reference. From a general aging point of view, the results are very similar, with the main difference in the cathode particle cracking point of view. It seems that currents used inside the MCCFast2 profile, permit a better Li + diffusion inside the cathode particles, preventing the formation of cracks [36]. Figure 7d reports the particle size distribution of NMC811 in all the tested conditions. From a statistical point of view, all the implemented DCFC profiles seem to be able to maintain the average particle distribution reported at BOL stage, where the cell did not undergo cycle aging.
Reference DCFC profile shows the worst case condition from the particle size distribution point of view, with the decrease of the main peak (around 10 µm) and a concurrent shift towards smaller diameters. The increasing of a particles' family at smaller diameters (around 1 µm) is also observable, data that is in accordance with the severe cracking observed with the SEM analysis and a result that is in line with other works [48,49] It is possible to notice the general good health of the selected samples (see Material and Methods for details).
Incremental capacity analysis on a three-electrode lab scale cell does not reveal significant damage inside the cell, where no lithium plating was observed at the visual inspection.
Looking at the peaks in both anode (Si-C vs. Li) and cathode (NMC811 vs. Li) voltage trends, a good matching in term of peak position, between EOL and BOL cell samples, was also found. This behavior is indicative of no significant resistance growth inside the cells assembled with the selected samples. We can observe a general decrease of the peak values with respect to the BOL cell, indicating an expected capacity loss due to normal aging.
When lithium plating is present [34], a significant peak shift and area (below the ICA curve) reduction was observed in the anodic voltage response.
Both MCC2 and reference EOL curves are close to be overlapped (conversely to the overall ICA response on 18650 cells at Figure 4), suggesting that localized lithium plating is the driving aging phenomenon of the performed cycle aging campaign.

XRD Measurements
Data collected on cathode samples in both powders and calendered conditions do not present significant differences from the XRD pattern point of view (see Supporting Information ( Figure S11)).
NMC811 material maintained the initial hexagonal crystal structure (R3m space group). From the Rietveld refinement ( Figure S12 in Supporting Information) performed on cathode powders, it was possible to measure the evolution of the crystal lattice parameters a and c. As confirmed by the literature [31,50], aging mechanisms inside the Li-ion cell induce a shrinkage of the lattice along a-axis and a concurrent stretching along c-axis.
The main hypothesis of this behavior is attributable to: The increase of repulsive forces between slabs due to the decrease of the lithium content in the EOL electrode materials [27,51]; The increase of the average oxidation state for nickel, manganese, and cobalt ions [52].
Cathode peak intensity ratio performed on 003 and 104 peaks ( Figure S12) was measured to evaluate the presence of cation mixing. It is known from the literature [53,54] how the relative ratio of the intensities I 003 /I 104 is associated to the presence of this phenomenon, visible in Table 1. Values below 1.2 are indicative of significant aging.
From the data collected for the present work, all the values are above 1.49, indicative of the lack of cation mixing.
Lattice parameter evolution and cation mixing absence are a clear sign of the general good health of the tested materials, confirming that the main aging mechanism in our testing conditions is lithium plating, localized only to small portions of the electrode foil.

Conclusions
The present work presented data and results obtained from a cycle aging campaign performed on 18650 cells, aged using four new implemented custom designed DCFC profiles, and compared to a reference condition. It was seen how some of the new conditions could provide better behavior in terms of aging and time saving (especially for MCCFast2). Among the main results it is possible to state that:

•
The use of ICA in the present study confirms the general good health of the tested cells, indeed no significant degradation on the NMC811 electrode was observed and localized degradation on the Si-C electrode, where lithium plating occurs, was noticed. Physicochemical characterization, along with visual inspection, confirms data obtained from ICA, both in full cells and in lab scale three-electrode cells. Where no evident damage is visible on the electrode samples, ICA responses confirm the absence of significant degradation; • Among the new implemented approaches, internal resistance test (R int vs. SoH%) can be considered as a useful "in situ" methodology to control the current (and voltage) limits on the cell, during the aging cycle.
It was observed how the implementation of a derating function, based only on the reduction of cell capacity over a lifetime, was not exhaustive for avoiding a lithium plating phenomenon. This technique seems to be promising to tune the charging steps continuously, by performing a period check on the battery system. Since it does not require complex instrumentation, it presents a concrete possibility to be implemented as an "on board" application on BMS.