Volume 10, Issue 21 e202300279
Research Article
Open Access

Exploring the Synergistic Effects of Dual-Layer Electrodes for High Power Li-Ion Batteries

Dr. Jeremy I. G. Dawkins

Dr. Jeremy I. G. Dawkins

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Search for more papers by this author
Dr. Yani Pan

Dr. Yani Pan

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Search for more papers by this author
Dr. Mohammadreza Z. Ghavidel

Dr. Mohammadreza Z. Ghavidel

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Department of Chemistry NanoQAM, Université du Québec à Montréal Québec Center for Functional Materials Case Postale 8888 Succ. Centre-ville, Montréal, Québec, H3C 3P8 Canada

Search for more papers by this author
Johann Geissler

Johann Geissler

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Search for more papers by this author
Dr. Bastian Krueger

Dr. Bastian Krueger

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Department of Chemistry NanoQAM, Université du Québec à Montréal Québec Center for Functional Materials Case Postale 8888 Succ. Centre-ville, Montréal, Québec, H3C 3P8 Canada

Search for more papers by this author
Dr. Danny Chhin

Dr. Danny Chhin

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Search for more papers by this author
Dr. Hui Yuan

Dr. Hui Yuan

Canadian Centre for Electron Microscopy, McMaster University, 1280 Main Street West, Hamilton, Ontario, L8S 4M1 Canada

Department of Materials Science and Engineering, McMaster University, 1280 Main Street West, Hamilton, Ontario, L8S 4L7 Canada

Search for more papers by this author
Victoria Tong

Victoria Tong

Department of Chemistry NanoQAM, Université du Québec à Montréal Québec Center for Functional Materials Case Postale 8888 Succ. Centre-ville, Montréal, Québec, H3C 3P8 Canada

Search for more papers by this author
Brittany Pelletier-Villeneuve

Brittany Pelletier-Villeneuve

Department of Chemistry NanoQAM, Université du Québec à Montréal Québec Center for Functional Materials Case Postale 8888 Succ. Centre-ville, Montréal, Québec, H3C 3P8 Canada

Search for more papers by this author
Dr. Renfei Feng

Dr. Renfei Feng

Canadian Light Source, 44 Innovation Boulevard, Saskatoon, Saskatchewan, S7N 2V3 Canada

Search for more papers by this author
Prof. Dr. Gianluigi A. Botton

Prof. Dr. Gianluigi A. Botton

Department of Materials Science and Engineering, McMaster University, 1280 Main Street West, Hamilton, Ontario, L8S 4L7 Canada

Canadian Light Source, 44 Innovation Boulevard, Saskatoon, Saskatchewan, S7N 2V3 Canada

Search for more papers by this author
Prof. Dr. Karena W. Chapman

Prof. Dr. Karena W. Chapman

Department of Chemistry, Stony Brook University, 100 Nicolls Rd, Stony Brook, NY, 11794 USA

Search for more papers by this author
Prof. Dr. Janine Mauzeroll

Corresponding Author

Prof. Dr. Janine Mauzeroll

Department of Chemistry, McGill University, 801 Sherbrooke Street West, Montreal, Québec, H3A 0B8 Canada

Search for more papers by this author
Prof. Dr. Steen B. Schougaard

Corresponding Author

Prof. Dr. Steen B. Schougaard

Department of Chemistry NanoQAM, Université du Québec à Montréal Québec Center for Functional Materials Case Postale 8888 Succ. Centre-ville, Montréal, Québec, H3C 3P8 Canada

Search for more papers by this author
First published: 28 August 2023

Graphical Abstract

A Li-ion battery electrode architecture which uses two different active materials in a layered configuration is investigated. The results surprisingly show that layered electrodes are superior to their blended (mixed) counterparts during high-rate (dis)charge. The mechanism of this synergistic effect is elucidated using a newly minted synchrotron fluorescence technique to map the concentration of Li+ throughout an operating cell.

Abstract

The electrification of the transport sector has created an increasing demand for lithium-ion batteries that can provide high power intermittently while maintaining a high energy density. Given the difficulty in designing a single redox material with both high power and energy density, electrodes based on composites of several electroactive materials optimized for power or capacity are being studied extensively. Among others, fast-charging LiFePO4 and high energy Li(NixMnyCoz)O2 are commonly employed in industrial cell manufacturing. This study focuses on comparing different approaches to combining these two active materials into a single electrode. These arrangements were compared using standard electrochemical (dis)charge procedures and using synchrotron X-ray fluorescence to identify variations in solution concentration gradient formation. The electrochemical performance of the layered electrodes with the high-power material on top is found to be enhanced relative to its blended electrode counterpart when (dis)charged at the same specific currents. These findings highlight dual-layer lithium-ion batteries as an inexpensive way of increasing energy and power density of lithium-ion batteries as well as a model system to study and exploit the synergistic effects of blended electrodes.

Introduction

Li-ion batteries (LIBs) are used extensively in the electrification of the transport sector due to their many virtues such as high energy density, low self-discharge and long cycle life.1 While current LIB technology meets the requirements for many applications, it falls short in areas which require sustained or intermittent high power, e. g., current bursts required for accelerating electric vehicles or during regenerative breaking. When designing electrodes for high-power performance, micron to sub-micron particle sizes must be used as common active materials such as layered Li(NixMnyCo1-x-y)O2 (NMC) have slow solid state diffusion (e. g., Li(Ni1/3Mn1/3Co1/3)O2 (NMC111) DLi ≈1×10−10 cm2/s).2 Such stringent size limitations are mandatory if considerable (>80 %) capacity retention is targeted at rates exceeding 5 C, where x C refers to the (dis)charge of the theoretical capacity in 1/x hours. Problematically, reducing the particle size also increases the surface area at which parasitic reactions are prone to occur for high-voltage materials such as NMC.3 Furthermore, the stringent size limitation requires unique synthetic strategies, especially at the industrial scale.4, 5

Given the challenges in reducing particle size, few high-power materials with adequate performance and cycle life have been produced industrially; Olivine LiFePO4 (LFP) being a rare exception.6 In experiments where the oxidation of the active material is not limited by the electrode structure, LFP can reach a full state of charge in 10 seconds.7, 8 Unfortunately LFP has a low energy density which is a consequence of the limited redox potential of the Fe2+/Fe3+ redox couple. There is thus an advantage in combining LFP with a high-energy material such as NMC for applications that require high power capabilities as well as sustained currents.9, 10 In this work, NMC111 was chosen as a high energy material compared to LFP, although the results may be extended to other commercial NMC materials like Li(Ni8/10Mn1/10Co1/10)O2 (NMC811) which have even higher energy densities.11 Materials such as LFP and NMC are commonly used in commercial electrodes given their optimal redox properties,12, 13 although other structures such as MOFs,14 mesoporous Li4Ti5O1215 and phosphate polyanion16 materials have also been explored.

A common method of incorporating both materials in a single electrode is homogenous mixing.17 Both materials are incorporated in the slurry from which the films are cast, thus creating a “blended” electrode. A surprising benefit of homogenously mixed blended electrodes is the reported synergistic effects, which include current redistribution, i. e., buffering at high currents leading to a reduction in electrode polarization and structural changes affecting conductivity.18 The buffering effect was shown experimentally19 and from modeling20 to benefit high and low power-density material combinations by reducing the current drawn from the low power-density material at high currents thereby increasing its capacity.20 There is also evidence of spontaneous thermodynamically driven Li redistribution between both active materials in blended electrodes as the temperature changes, which normally occurs during (dis)charge.21 Finally, an increase in energy density relative to the individual constituents of a multi-material electrode was observed in other combinations of high-power and high-energy density materials where it was in part attributed to a reduced polarization of the electrode.22

Dual-layer (DL) electrodes,23 where the active materials are kept in separate layers, have however received less attention as early work reported poor performance of DLs relative to their blended counterparts.24, 25 The poor performance is surprising as the average distance between the positive and the negative electrode particles is expected to become capacity limiting at high (dis)charge rates. This distance can be minimized in DL positive electrodes by placing the high-power material on top (farthest from the current collector) where it will be closest to the negative electrode, thereby reducing the Li+ transport distance during fast (dis)charge.

Here we report new data on DL positive electrode discharge curves measured in coin cells. The analysis is separated in three parts: 1) evaluating the (dis)charge rate above which it is beneficial to use two active materials in a positive electrode, 2) revisiting the electrode configuration's impact on performance (i. e., DL compared to blended), and 3) study the nature of DL synergistic effects using a recently minted synchrotron X-ray fluorescence technique. This work has the potential to increase the high-power capabilities of industrially relevant LIB materials using a simple, accessible and low-cost method which is explored mechanistically and exploited.

Results and Discussion

DL morphology

Cross-sectional SEM images and EDXS maps were gathered from DL electrodes to assess the interface between both layers. A sharp interface separates the NMC/LFP layers (Figure 1a) in DL assemblies, which is further confirmed by the absence of P and Ni signals in the NMC and LFP layers respectively (Figure 1b). The battery material layer thicknesses in the DL (LFP on bottom) are similar (41 μm for LFP and 42 μm for NMC) leading to a mass ratio close to unity (48 : 52 LFP : NMC). If compared to blended electrodes of similar porosities and thicknesses, the DL morphology suggests that key differences in performance will arise from: 1) the difference in contact area between NMC and LFP particles within the layer, 2) the distance of each active material to the negative electrode and to the current collector 3) the contact resistance to the current collector. Combined these elements are expected to have a profound effect on the overall performance of the electrode.

Details are in the caption following the image

a) SEM image and b) EDXS map on cross-section of a DL positive electrode (LFP on bottom) and c) schematic representation of the orientation of multi-material positive electrodes. (i) Dual-layer (DL) with LFP as bottom layer, (ii) DL with LFP as top layer and (iii) blended NMC/LFP electrode.

Performance of DL and SL electrodes

Traditional C-rates cannot be used for positive electrodes containing two or more active materials that have different electrochemical potentials since in absence of significant electrode charge transport limitations, the current will be drawn from each material sequentially.19, 26 Thus at low current densities during discharge, the high-potential NMC material bears the majority of the current until the potential of LFP (≈3.4 V) is reached. Consequently, the effective C-rate is increased, and the performance is underestimated.21, 27 To address these concerns, the DL and SL performances are evaluated using mass specific current densities (mA g−1). The resulting NMC and LFP SLs capacities are consistent with literature values in the chosen potential range (Figure 2).28, 29 The capacity of the DL (LFP on top) was found to be superior to the high-energy NMC material at high current densities despite the NMC outperforming at the slowest rate of 15 mA g−1 (Figure 2a). Similarly, the DL matched the LFP cells at lower current densities, while at higher current densities the DL prevailed (Figure 2b).

Details are in the caption following the image

a, b) DL LFP top (32 : 68 NMC : LFP, solid) discharge curves compared to a) NMC (dashed) and b) LFP (dashed) single layer discharge curves. c) DL LFP bottom (NMC : LFP 66 : 34, solid) vs. added curve of two (one LFP, one NMC) constituent material SL electrodes (dashed). d) DL LFP top, (NMC : LFP 32 : 68) discharge curves (solid) compared to blended (NMC : LFP 30 : 70) electrode (dashed).

It is surprising that DLs outperform SLs at high rates considering the influence of contact resistance and thickness on cell performance.30 In this set of experiments, the contact resistance of the electrode to the current collector should be equivalent, however the DLs have an additional contact resistance at the NMC/LFP interface. The distance from the bottom layer to the counter electrode is also larger (≈35 μm, Table S1) than that of SLs (≈20 μm, Table S1) which exacerbates the formation of strong concentration gradients at high currents. Despite the increased thickness in the DL electrode, it displays an enhanced performance relative to its individual constituent layers. The result of comparing the DL to a SL reference electrode of equivalent thickness should further increase the performance gap based on lower capacity being a result of increasing electrode thickness, especially at higher rates.30 To examine this further, the summed contributions of a SL NMC and a SL LFP film were compared to the performance of a DL electrode.

Capacity of DLs versus the summed SL contributions

A DL (LFP on bottom) performance (Figure 2c) was compared to the summed SL NMC and SL LFP using a fixed current (dis)charge. Employing the same area specific current density makes the data directly comparable as the mass of LFP/NMC in each SL was equal to the mass of LFP/NMC in the DL counterpart within 3 %. The results of this experiment (Figure 2c) show the DL outperformed the summed SL response at every applied current in terms of discharge capacity.

Compared to the previous section, the summed SLs include the contribution of the contact resistance to the current collector twice, whereas the DL contains the contact resistance to the current collector and the contact resistance between the LFP and NMC layers. Depending on the structure of this interface, these contributions may be significantly different. Using custom (dis)charge algorithms to study each layer individually as well as determining the resistance of the interfacial layer using electrochemical impedance spectroscopy (EIS) can shed light on the synergistic effect observed, especially after long term cycling. Furthermore, distribution of relaxation time (DRT) analysis can be employed to gain more information about the parameters which influence DL performance. Although outside the scope of the current work, this data is expected to be helpful to identify the definitive cause of the synergistic effect in DL electrodes. The distance of the active materials in the DL to either the current collector or counter electrode is significantly larger than its SL counterpart (Table S1). For example, when the top layer is active, electronic charge transport through the inactive bottom layer is required to reach the current collector. When the bottom layer is active, ionic transport must pass through the top layer to reach the counter electrode. Combined, it is therefore unexpected that DLs outperform the summed SLs. It is possible that the contact resistance at the current collector is larger than the contact resistance at the LFP/NMC interface, which could in part explain these results.

Capacity of DLs versus blended electrodes

In these experiments, DL electrodes are compared to their blended counterparts. The applied currents were chosen based on total active material mass (mA g−1) which removes the aforementioned concerns related to employing C-rates for multi-material positive electrodes. Comparing DLs with LFP as the top layer (with 68 : 32 NMC : LFP by mass) to blended electrodes (70 : 30 NMC : LFP) in Figure 2d, the DLs matched the blended electrodes at the slowest current of 15 mA g−1 but outperformed the blended electrodes at high currents. The obtained blended electrode capacities were in agreement with reported values for LFP/NMC blends of various ratios in the selected potential window (2.5–4.3 V).31

In a purely thermodynamic analysis, blending vs. separating the active materials in the electrode should yield the same result (see p.7 Supporting Information for complete analysis). As such the advantage of the DL electrode is expected to steam from kinetic or mass transport effects.

The blended and DL electrodes investigated in the experiment have comparable thicknesses and porosities (Table S1) which is expected to yield equivalent ionic transport. Additionally, the mean particle distances from the current collector and counter electrode should be the same in both cases. The current collector contact resistance is likely different given that the mix of small and larger particles in LFP/NMC blends are known to undergo rheological segregation based on particle size, especially near the current collector.32 The IR-drop (proportional to the potential drop at the beginning of discharge) is larger for blended electrodes (Figure 2d) although it is similar for both DLs and blended cells when the LFP layer is on the bottom (Figure S3).

Combined capacity vs. rate analysis

The comparison of DLs, SLs, and blended electrodes in Figure 3a shows the capacities of the DLs to be competitive. Replicate cells used in this analysis display the same performance trend (n=2). The DL (LFP on bottom) electrode displayed an expected performance by maintaining capacities between those of LFP and NMC (grey shaded area, Figure 3a). The blended electrodes (NMC : LFP ratios of 30 : 70 and 50 : 50) show expected and slightly enhanced performance, respectively. In contrast, the DL electrode with LFP on top outperformed all other configurations and had capacities that displayed synergistic effects (green area) with respect to the individual constituent layers at various currents. This effect is more pronounced at higher currents indicating the effect is driven by mass transport.

Details are in the caption following the image

a) Specific capacity (mAh g−1) and b) energy density (mWh g−1) vs. applied specific current density (mA g−1) for LFP (black), NMC (grey), DL LFP on top (30 : 70 NMC : LFP, blue squares), DL LFP on bottom (50 : 50 NMC : LFP, red squares), a blended (30 : 70 NMC : LFP, blue triangles) and a blended (50 : 50 NMC : LFP, red triangles) electrode. The shading indicates the expected (gray), synergistic (green) or detrimental (red) effect of using multiple active materials.

In addition to the influence of the contact area between the LFP/NMC in the bulk of the layers, the distances of the materials relative to the current collector and counter electrodes, and the presence of contact resistances, other structural features of DLs should be considered in this analysis. Given the disparity in redox potential of the LFP and NMC active materials, one layer will bear the majority of the current at any given time during (dis)charge, except at extreme current densities.21 This allows the layer bearing the lower current density to act as an electrolyte reservoir, thereby mitigating concentration gradients and reducing the overall concentration overpotential. It has also previously been found that significant heterogeneity exists in blended electrodes caused by the different properties and sizes of the constituent materials. This was shown for an LFP/NMC blended system similar to the one used here where inhomogeneous pore size and particle size distribution as well as agglomerates were reported.32 Despite the multiple synergistic effects of blended electrodes, this effect could potentially be detrimental to performance when compared to a segregated electrode.

Energy density considerations and stability

A key practical parameter to assess LIB performance is energy density (mWh g−1), given that NMC and LFP have different nominal cell potentials (3.7 V vs. 3.2 V, respectively). At lower specific currents, both the DLs and blended electrodes had energy densities similar to those of pure LFP and NMC (Figure 3b, denoted by a gray shading) as expected. At larger applied currents, the DL (LFP on top) was dominant and outperformed both the LFP and NMC cells, thereby suggesting a synergistic effect of the two components at high rates of (dis)charge (>150 mA g−1). The key finding in Figure 3b is that blended electrodes do not systematically outperform DLs as previous works indicated.24, 25 A degradation analysis was performed by returning to the lowest current of 15 mA g−1 (Figure S4) after cycling to ensure significant capacity fade was not present.

Identifying the source of the synergistic effect

The electrochemical measurements show that DL cells can outperform single layer and blended cells, and that this synergistic effect is exacerbated at high currents. To study the nature of the enhanced performance of DL electrodes, synchrotron X-ray fluorescence (XRF) was employed to obtain the solution phase concentration profiles of Li+. The concentration gradients formed can indicate whether this effect is related to differences in mass transport in blended and DL cells. The [Li+] was quantified in a DL (LFP on top) and a comparable blended electrode with a spatial and temporal resolution of 13 μm and 150 s (total line scan time) respectively. The application of synchrotron XRF to study concentration gradients in LIBs involves tracking a heavy-element anion (such as AsF6) as a proxy for Li+. A complete report on this method was previously published using an LFP electrode as proof-of-concept.33

The components of the cell (current collector, electrode layers etc.) were first identified by tracking the Fe, As and Co intensity at open-circuit voltage (OCV). In Figure 4a and b, it is possible to differentiate the various components of the cell stack using the Fe and Co Kα to track the LFP and NMC layers, respectively. Sharp transitions exist between the current collector and bottom layer (26–36 μm) indicating that the beam is well aligned within the plane of the interface. A more gradual transition is seen between the top layer (or top part of the blended electrode) and the separator which can be attributed to the active material impregnating the separator when assembling the cell. This is also observed visually when disassembling the cells. The absolute As Kα variation in each cell component (i. e., electrode, separator etc.) is largely influenced by X-ray absorption from the matrix and therefore is not a direct indicator of the As/Li concentration in that region. Normalization of the As Kα to the OCV response at each position is employed to quantify the relative change in As (and therefore Li) concentration.

Details are in the caption following the image

a, b) X-ray fluorescence intensity of iron (green), cobalt (blue) and arsenic (orange) Kα peaks corresponding to a) blended electrode and b) DL (LFP on top) electrode at open circuit voltage. Background shading displays various components of the cell which are also represented by a schematic. c, d) Concentration profiles of c) blended electrode and a comparable d) DL (LFP on top) electrode during a constant current/constant voltage (CC/CV) charge at 150 mA g−1. e) Heatmap of [Li+] as a function of depth and time for a 630 μm DL (LFP on top) electrode charged at 150 mA g−1. The bottom plot shows the potential response (E) and applied current (I) as a function of time. Vertical gray lines indicate two points of interest: the beginning of charge and the constant voltage hold. The horizontal dashed line represents the LFP/NMC interface.

Figure 4c and d show the As Kα normalized to the OCV response, which is equivalent to the solution [Li+], for a DL (LFP on top) electrode (630 μm) and a similar blended electrode (559 μm). To emulate the interfacial properties (resistance, poor adhesion etc.) and thickness of the DL electrode, the blended sample was also produced by overlaying two separate freestanding blended electrodes. The obtained concentration profiles were smoothed using a Savtizky–Golay filter for clarity. Whereas the blended electrode produces the expected response (i. e., a steadily increasing concentration with a larger gradient forming near the current collector) the DL gradient firstly forms within the LFP layer, later subsiding and increasing in the NMC layer as the latter is electrochemically engaged. This is difficult to visualize in a series of concentration profiles so a heatmap of the solution [Li+] was produced for the DL data (Figure 4e).

In Figure 4e, the cell is charged galvanostatically at 150 mA g−1 followed by a potential hold at 4.3 V. Given the thickness of the electrode (630 μm) and its high conductive additive content (20 % by mass carbon black) it is expected to be limited by its ability to carry charge in solution at the applied current density. Qualitatively, the DL charging dynamics differ substantially from that of the blended electrode. First, the LFP layer charges and forms a localized concentration gradient (150 mins) while the solution [Li+] within the NMC changes little. Once the LFP is fully charged, the NMC bears most of the current forming a strong gradient close to the current collector (200 mins) while the gradient in the LFP equilibrates. The NMC layer has a lower [Li+] at its interface with the LFP which could be the result of the LFP layer acting as an electrolyte sink. A strong As signal arises at the interface of the current collector and the NMC layer which is unexpected. This could be the result of a physical change, such as partial film delamination or of a chemical process such as corrosion of the current collector. The latter theory is supported by literature reports of the analogous LiPF6 salt hydrolyzing with trace amounts of water to form products that are known to corrode or passivate steel.34-36 Although this effect was not observed in the blended electrode, the measurement performed is localized and the process could be occurring regionally in the probed area. The contrasting concentration profiles produced by both systems supports the theory that the synergistic effect observed for DL electrodes at faster rates is related to mass transport in solution. Specifically, as a DL cell is (dis)charged, 1) the unengaged layer can act as an electrolyte reservoir for the layer bearing the current and 2) the LFP being on top reduces the mean path traveled by Li+ during (dis)charge which mitigates the concentration gradient formed for the kinetically faster material. The former argument is further supported by the lower electrolyte concentration present at the interface of the two layers (Figure 4e, ≈600 μm). The latter is supported by a mitigated solution phase gradient that forms in the LFP layer closest to the separator but would require a comparison to a similar DL (LFP on bottom) electrode to prove quantitatively. The key finding is that qualitatively, DL and blended electrodes inherently form different solution phase concentration gradients and that this could be the source of the synergistic effect reported in this work.

Conclusions

In this work, we investigated three different arrangements of multi-material electrodes with the high-power material as either the top layer, bottom layer, or included in a blend, at multiple specific current densities. This approach was favored as the use of the more common C-rates with mixed-material electrodes exhibiting different potential regimes makes comparison difficult. The key discovery is that overall DL cells can outperform blended electrodes in capacity, especially at high rates of (dis)charge. Moreover, DLs display enhanced performance when compared to both their individual constituent layers, an effect that was only previously known to occur in blended materials at high rates. These results are contradictory to prior work on segregated LFP/NMC electrodes which reported reduced performance relative to blended electrodes.24 We additionally propose that the synergistic effect of DL electrodes with the high power material on top may be a result of 1) increased proximity to the negative electrode and 2) the unengaged layer acting as an electrolyte reservoir creating a unique concentration gradient. The latter argument is supported by experimental synchrotron data displaying the variation in concentration gradients compared to blended electrodes.

This work paves the way for further investigation of DLs that allow for tunability in terms of material choice, porosity, thickness, and orientation of the layers. A particular advantage of the DL is the tunability of the porosity and thickness of both top and bottom layers separately which allows optimal balancing of mass transport and electronic conductivity of the electrode, leading to enhanced performance.37 A DL electrode with only a thin layer of the high-power LFP on top to buffer higher currents is of future interest as it may retain a larger energy density at low currents while still maintaining synergistic behavior at high currents. In this context, the ultimate goal is to find the optimal conformation for the DL LFP/NMC system as well as expanding the DL system with alternative battery material combinations.

Experimental Section

Slurry preparation

Slurries were prepared by dissolving polyvinylidene fluoride (PVDF, Kynar) in methyl-2-pyrrolidinone (NMP, Sigma Aldrich 99 %) on a heated magnetic stirring plate set to 50 °C and 800 RPM. The volume of NMP required for dissolution was 1.4 mL g−1 of the final electrode solid components (carbon black, PVDF and active material) for the NMC slurry and 1.9 mL g−1 with LFP. Carbon black (Alfa Aesar) was firstly added as conductive additive, followed by the active material (either LiFePO4 (Phostech Lithium Inc., P2 grade) and/or Li(Ni1/3Mn1/3Co1/3)O2 (General Motors)). The 4 g (solid mass) slurries had a mass ratio of 85 : 9 : 6 of active material, binder and conductive additive respectively. The blended slurries were prepared with an NMP volume between 1.4 mL g−1 and 1.9 mL g−1 based on the LFP:NMC ratio. Two blended electrode slurries were prepared with LFP:NMC mass ratios of 70 : 30 and 50 : 50. Slurries were then ball milled for 48 hours to ensure complete mixing.

Film preparation

Before coating, the carbon-coated Al foil (MTI) was pre-dried in a vacuum oven at 80 °C to remove moisture and promote a uniform coating. Carbon coated foil reduces the contact resistance between the electrode and the current collector, thereby reducing ohmic voltage drop. Slurries were coated on the Al current collectors using a doctor blade on a vacuum coater. A table with the height adjustments for each electrode can be found in the Supporting Information (Table S1). The single-layer (SL) films were dried for one hour at 100 °C under atmospheric pressure and for 12 hours at 80 °C under vacuum. To ensure the best comparability for porosity, thickness and tortuosity between DL and SL films, the underlying (SL) film was cut into two strips along the long axis of the film (Figure S1). One of the strips was coated with the second active material (DL) and the other strip was used as a SL reference.

Coin cells

Positive electrodes were punched to a diameter of 15 mm. To ensure the DL electrode and its bottom layer reference were as similar as possible, the two half-films were placed one atop the other and aligned before being punched (Figure S1). The mass of the top layer was found by subtracting the mass of the reference SL from the total DL mass. The coin cells were assembled in stainless steel casings using a 0.75 mm thick Li negative electrode (Alfa Aesar 99.9 %) with Celgard 2500 separators. 1 M LiPF6 in ethylene carbonate (EC) and dimethyl carbonate (DMC) 1 : 1 w/w (Gotion) was used as electrolyte. All cells were assembled in an inert argon atmosphere (H2O <1 ppm, O2 <1 ppm).

Electrochemical measurements

The cells were galvanostatically cycled at identical current densities from 2.5 to 4.3 V vs Li/Li+ with an open-circuit rest period of 90 minutes between charging and discharging. Each cell was formed with three cycles at 15 mA g−1 before measurement. Each data series consisted of four cycles at each current of interest. These series were collected in order of increasing current and terminated by repeating the 4 cycles at the lowest rate. The summed curves were obtained by combining the capacity of the SL NMC and SL LFP multiplied by their respective mass fraction in the DL.

SEM

Circular samples were first cut with a single edge blade before attachment to a standard SEM stub. The sample's top-surface was coated with a C + Pt layer and its cross-section was milled with gradually decreased ion currents within a Helios G4 UXe DualBeam PFIB (ThermoScientific, Hillsboro, USA). Energy dispersive X-ray spectroscopy (EDXS) analysis using a X–Max detector (Oxford Instrument, Abington, UK) and back scattered electron imaging was performed with a JEOL 7000F SEM (JEOL Ltd., Tokyo, Japan) operating at an accelerating voltage of 10 kV.

Operando synchrotron X-ray fluorescence

The measurement employed was a recently established operando fluorescence-based methodology to evaluate the Li concentration within the electrode.33 Briefly, freestanding electrodes (electrodes produced without a current collector support) were produced by mixing 65 % active material, 20 % polytetrafluoroethylene (PTFE, Sigma Aldrich, 60 % suspension in H2O) binder and 15 % Super C65 (Timcal) carbon conductive additive, by mass. The mixture was then rolled and pressed into a film with the use of a steel rolling rod. The DL (LFP on top) electrode was prepared by placing the freestanding LFP electrode on top of the NMC electrode in a PEEK cylindrical cell with thin (≈300 μm) walls which served as X-ray transparent windows for the measurements. To obtain a similar total thickness and interfacial resistance, the blended cell was made using two separate blended electrodes and assembled in the same fashion as the DL. Both cells had a net 60 : 40 NMC : LFP mass ratio. The cells used 1 M LiPF6 in 1 : 1 EC : DMC as electrolyte with 0.1 M LiAsF6 added as the fluorescent label. Measurements were made at the VESPERS beamline of the Canadian Light Source which employs a bending magnet source. The depth profiling was performed at open circuit voltage. Vertical scans were made with 13 μm steps at an incident beam energy of 12.3 keV with 1 s acquisition time using a single element Vortex silicon drift detector (Hitachi). The cells were measured during cycling at a rate of 150 mA g−1 with acquisition at 150 s intervals to reduce beam damage and account for motor overhead time. To obtain the absolute [Li+], the As Kα peak is integrated and normalized to its OCV value.33

Acknowledgments

A major part of the research described in this paper was performed at the Canadian Light Source, a national research facility of the University of Saskatchewan, which is supported by the Canada Foundation for Innovation (CFI), the Natural Sciences and Engineering Research Council (NSERC), the National Research Council (NRC), the Government of Saskatchewan, and the University of Saskatchewan. The authors would also like to acknowledge the NSERC Grant No. 326937-2013-RGPIN/RGPIN-2019-07200 (S.B. Schougaard) and CRDPJ 494074-16 (J. Mauzeroll & S.B. Schougaard) for financial support, Fonds de recherche du Québec – Nature et technologies (FRQNT) for doctoral scholarship No. 291365 (J.I.G. Dawkins) as well as funding from General Motors Canada in addition to providing the NMC material used in this work. Electron microscopy was performed at the Canadian Centre for Electron Microscopy (CCEM), a national facility supported by NSERC, the Canada Foundation for Innovation, under the MSI program, and McMaster University.

    Conflict of interest

    The authors declare no conflict of interest.

    Data Availability Statement

    The data that support the findings of this study are available from the corresponding author upon reasonable request.