Atomic layer deposition niobium oxide and lithium niobium oxide as a protection technique for anode ‐ free batteries

As demand for extended range in electric vehicles and longer battery lifetimes in consumer electronics has grown, so have the requirements for higher energy densities and longer cycle lifetimes of the cells that power them. One solution to this is the implementation of an “ anode ‐ free ” battery. By removing the anode and plating lithium directly onto the current collector, it is possible to access the same capacities and voltage windows as traditional lithium metal batteries, with the entirety of the lithium source coming from the cathode. Herein, a copper foil current collector coated with niobium oxide or lithium niobium oxide through atomic layer deposition (ALD) is applied to extend the cycling life of the anode ‐ free batteries by reducing dendrite formation and improving the stability of the lithium metal surface throughout cycling. The ALD coatings are able to extend the cycle lifetime in full coin cells from 20 cycles to 80% capacity retained in the bare copper controls to 50 and 115 cycles for the NbO and LiNbO coatings, respectively. Over the lifetime of the cells, the ALD ‐ LiNbO is able to cumulatively offer a staggering improvement of an additional 100 kWh L − 1 compared to the bare copper control.


| INTRODUCTION
With the advent of the lithium-ion battery and increased storage capacities has come the growth of portable consumer electronics and long-range electric vehicles.As demand for these products grows, so do the requirements for higher and higher energy densities and cycle lifetimes of the cells that power them.As such, attention has turned to alternative cell chemistries and geometries.Lithium metal batteries have emerged as one of the most promising technologies to satisfy the increasing demands in energy storage applications across both electric vehicles and portable electronics. 1 With a best-in-class gravimetric capacity of 3860 mAh g −1 , low density (0.59 g cm −3 ), and lowest possible potential (−3.04 V vs. RHE), Li metal presents an appealing option for high energy density lithium metal batteries.Due to the 10-fold improvement compared to traditional carbonaceous anodes, the use of lithium metal batteries offers a tremendous theoretical improvement to energy density at the cell level. 2 However, low Coulombic efficiencies and extensive parasitic side reactions with the liquid electrolyte require the use of extensive inactive Li reservoirs in the majority of lithium metal literature, meaning much of the volume of the cell is wasted, making many of the improvements to cell energy density merely theoretical. 3,4ne solution to this is the implementation of an "anode-free" battery.By removing the excess lithium and plating directly onto the current collector, it is possible to access the same capacities and voltage windows as traditional lithium metal batteries, with the entirety of the lithium source coming from the cathode.For example, using some industrially realistic cell parameters, the projected energy density of a lithium metal-NCA stack is approximately 516 Wh L −1 , which decreases to a projected 424 Wh L −1 in an 18650 cell format. 5,6By removing the Li reservoir and plating directly onto the current collector, the volumetric energy density can rise as high as 1131 Wh L −1 in the stack and 929 Wh L −1 in the projected 18650 cell format.The thickness difference between the two cell geometries can be observed in Figure 1.Aside from the benefits to energy density, the use of an anode-free geometry also offers significant improvements in cell processing with the removal of air and moisture-sensitive lithium metal from the process, as well as completely sidestepping the slurry coating process required for a carbonaceous or siliconbased anode.Similarly, the removal of lithium metal or traditional anode materials offers potential cost savings for fabrication.Using the software BatPac, as developed by Argonne National Laboratories, Salvatierre et al. demonstrated that an anode-free cell geometry could reduce final cell cost in $ kWh −1 by 15%-20%, predominantly as a function of saved material costs, as well as labor and processing costs. 7,8hile anode-free batteries offer some unique solutions, they are not without their issues.The lack of a lithium metal reservoir necessitates much higher Coulombic efficiencies than traditional lithium metal batteries (LMBs), as the entire lithium source comes from the cathode. 9,10As the plated lithium metal suffers from the same dendrite growth, continual side-reactions, and continual solid electrolyte interphase (SEI) growth as it would in traditional LMBs, these remain key issues to solve to improve the performance of these cells.There are generally three main avenues for the improvement of anode-free cells-(i) the use of alternative electrolyte additives and solvents, (ii) modification of the cycling conditions, and (iii) modification of the current collector surface.
In terms of electrolytes, a locally concentrated LiPF 6 salt with the addition of fluoroethylene carbonate (FEC) extends the cycle life from 15 to 50 cycles, according to research from the Hwang group. 11The Dahn group uses a dual-salt approach of LiDFOB and LiBF 4 in a carbonate solvent to reach as high as 80% capacity retained after 90 cycles. 12Likewise, Beyene et al. use a mixture of LiTFSI and LiFSI in an ether-based electrolyte to maintain a high Coulombic efficiency of 98.9% after 100 cycles. 13xamining alternative cycling conditions has also helped improve performance.Using high cycling pressures up to 1725 kPa to improve the lithium plating and stripping morphologies, it was possible to extend the cycling lifetime of FEC:DEC cells to 50% capacity retained after 70 cycles. 14Similarly, the Dahn group examined using asymmetric charge protocols such as C/5-D/2 and were able to increase the cycle lifetime compared to symmetric or fast charge-slow discharge protocolslikewise, they were able to greatly improve the lithium plating and stripping morphology. 15Beyene et al. demonstrated that a prolonged 24 h rest after the first deposition allows for a more uniform SEI formation via the reduction of the fluorinated solvent, resulting in 64% capacity retention after 50 cycles compared to almost zero capacity retention of the control group. 16here have also been numerous attempts to modify the current collector surface to improve performance; Lin et al. explored using a tri-alloy of gallium, indium, and tin to promote epitaxial lithium plating on the current collector surface and a Li-F-rich SEI layer. 17This liquid metal layer is able to improve the average Coulombic efficiency from 98.83% to 99.24% in half cells and extend the capacity retention at 50 cycles from 66% with a bare copper current collector to 84% in 120 mAh pouch cells.Similarly, Chen et al. utilized a commercial gold membrane as an interlayer atop the copper current collector in a Li 2 S solid-state battery. 18The gold layer allowed for significantly more facile Li deposition on the current collector, with the same capacity depositing over 10 μm thinner and visibly denser in scanning electron microscope (SEM) imaging, resulting in an improved initial Coulombic efficiency of 69.5%.Previously, Bing Joe Hwang's group demonstrated a variety of protection strategies at the current collector, including a chemical vapor deposition synthesized multilayer graphene, which improved the capacity retention after 100 cycles from 46% with bare copper to 61% with the graphene film. 19In a similar fashion, the same authors used a spin-coated polyethylene oxide layer on the copper current collector to prevent the formation of harmful lithium dendrites and protect the deposited lithium surface from continued side reactions with the electrolytes, resulting in an improved capacity retention of 30% after 200 cycles. 20reviously, our group has developed solid-state electrolyte-type coatings via ALD utilizing lithium niobium oxide with a high ionic conductivity. 21The prior literature suggests that the ionic conductivity of the SEI plays a critical role in the performance of anodefree cells, indicating a potentially high ionic conductivity artificial SEI could be a useful protection technique for anode-free batteries. 22,23erein, a novel method of modifying the current collector interface is presented using atomic layer deposition (ALD).The copper foil current collector coated with niobium oxide or lithium niobium oxide through ALD is investigated in half-cell configurations to demonstrate baseline performance.Various physical and electrochemical characterization techniques are utilized to demonstrate the ability of the ALD coatings to form a stable in situ-formed lithium interface.Full coin cell tests of the ALD-protected copper foils are also examined.Our design of the ALD-LiNbO coating layer for a stable and ionically conductive protection layer may offer new avenues to further improve the cycling life of anode-free batteries.

| RESULTS AND DISCUSSION
Our prior work investigated the use of ALD-LiNbO as a potential solid-state electrolyte through ALD. 21By inserting lithium oxide ALD cycles into the niobium oxide process in a ratio of 1:4, the ionic conductivity of these coatings was previously optimized at 6 × 10 −8 S cm −1 at room temperature, resulting in a Li:Nb ratio of 1.63:1.SEM imaging before electrochemical testing of the coatings shows no detectable morphology (Supporting Information S1: Figures S2 and S3).After 250 cycles, the LiNbO coating is approximately 65 nm thick (Supporting Information S1: Figure S4), consistent with the 2.5 Å per cycle growth rate from our previous work.Figure 2A-C shows the schematic diagram portraying the effect of the ALD-NbO and ALD-LiNbO coatings on the copper foil as a protective layer and their ability to prevent short-circuiting due to lithium dendrites.Herein, anode-free samples protected with NbO are denoted as ALD-NbO, while samples protected with LiNbO are denoted as ALD-LiNbO.Control samples are labeled as Bare Cu.While we examined the effect of the lithium layers in the ALD protection layer, we also examined the thickness for optimization.The NbO coating layer was deposited in 5, 10, and 25 cycle increments, while the LiNbO was deposited in 1, 5, 10, and 25 cycle increments.Consulting our previous work, these correspond to thicknesses in Supporting Information S1: Table S1.When comparing the half-cell cycling data of Li||Cu cells in Figure 2D,E, it is clear the positive effect that these ALD protection layers have during cycling.The bare copper cells cycle with an initial lithium plating overpotential of 180 mV before dropping to 50 mV during cycling.After approximately 110 h of cycling, the voltage rapidly rises to the limit of 1.5 V before full lithium stripping, indicating cell failure.Comparatively, even the thinnest layer of deposited ALD-NbO at five cycles was able to cycle with a lower overpotential of 20 mV and cycled for almost 300 h before cell failure.A protective layer of 25 cycles of ALD-NbO on top of the copper foil was able to cycle over 350 h before eventually short-circuiting, a threefold improvement over that of the bare copper.However, this beneficial effect was improved even further by the use of ALD-LiNbO; just one cycle of ALD-LiNbO was able to cycle stably for 350 h compared to the bare copper, 50 h longer than its ALD-NbO counterpart.As each LiNbO ALD macrocycle is made up of four NbO subcycles and one Li 2 O subcycle, we can see across equivalent film thicknesses the cycling improvement in the ALD-LiNbO coated copper foil.After optimization, the 10 cycles of ALD-LiNbO are able to cycle nearly 450 h in the same cell configuration compared to bare copper and over 100 h longer than any ALD-NbO cell.This improved cell stability in half cells represents a lifetime improvement of three to four times the original bare copper, a significant step forward for anode-free architectures.Notably, there is a clear relationship between the cycling performance of the anode-free coated current collectors and the thickness of the coating.If the coating is too thin, the protection provided by the coating layer improves upon bare copper foil but eventually fails.If the coating is too thick, the reduced ion diffusion capabilities afforded by a thicker artificial SEI limit the cycle lifetime of the cell.This is consistent with prior studies of ALD coatings for lithium metal protection and Li-Cu cells, in which coatings that are too thin suffer from Coulombic inefficiency, while coatings that are too thick show increased voltage hysteresis before eventual short circuits. 24,25Notably, this increased hysteresis, even in early cycles, is most evident in the 25 cycles of LiNbO ALD in Supporting Information S1: Figure S7 before eventually hitting the upper voltage limit, as evidenced by the colored box in Figure 2. As 25 cycles of NbO ALD and 10 cycles of LiNbO ALD were demonstrated to be the optimized coating thicknesses, these will be referred to as ALD-NbO and ALD-LiNbO in the experiments discussed moving forward.
When examining the first cycle plating and stripping in both the bare copper system as well as the ALD-protected systems in Figure 3A, there are two significant features to note.First, while the onset voltage of lithium plating for bare copper is the largest, the lithium nucleation overpotential is the voltage difference between the onset voltage and the plating plateau, offering the largest nucleation overpotential in the ALD-LiNbO system.These values are displayed graphically in Figure 3B growth region plateau at 45 mV.This largest nucleation overpotential in the ALD-protected current collectors offers large, low surface area lithium formation.The performance improvement driven by the higher resistance of the niobium oxide layer and by the high nucleation overpotential is consistent with the low surface area highdensity lithium deposition in the literature. 26Previous ALD anode-free work has shown that lower nucleation overpotential ALD coatings like SnO 2 or ZnO deposit high surface area lithium, while much more resistive and high nucleation overpotential Al 2 O 3 plates much more densely, as the very high nucleation overpotential declines to a lower potential in the plating regime.Similarly, the prior literature suggests that the nuclei size and porosity of the deposited lithium metal are significantly influenced by the applied potential in the growth region plateau rather than the initial nucleation overpotential. 27As seen in the plating and stripping figures, both ALD-NbO and ALD-LiNbO have significantly lower applied potentials of 40 and 45 mV in the growth region plateau, respectively, which should be indicative of low surface area and low porosity lithium growth.Conversely, bare copper foil maintains a high applied potential in the growth region of 90 mV.Even in subsequent cycles after deposition of the initial lithium layer, the plating potential on bare copper remains elevated compared to those of ALD-NbO and ALD-LiNbO.To observe this visually, SEM images were taken after cycling at 0.5 mA cm −2 -1 mAh after depositing 2.5 mAh on the first cycle, as pictured in Figure 3C-H.Both NbO and LiNbO form larger lithium nucleation sites with less porous lithium deposition, as suggested by the growth region potential data as well as the nucleation overpotential data.The lithium deposits on the bare copper with extremely high surface area and is evidently very porous, leading to excessive SEI formation, lithium inventory loss, and the formation of abundant dead lithium.In the ALD-NbOprotected copper, we see a very dense and low surface area lithium surface marked with small dendritic lithium structures.We suggest that these are small dendritic lithium nucleation sites on the surface of the coating rather than the surface of the lithium underneath the coating due to the lack of ionic conductivity of the NbO coating.Comparatively, for the ALD-LiNbO coating, we see the same low surface area and uniform Li but no dendritic structures on the surface.This is likely due to the comparatively high ionic conductivity of 6 × 10 −8 S cm −1 at room temperature, according to our group's prior work, allowing lithium ions to easily flow through the thin film coating and deposit uniformly on the lithium layer underneath, while the lower ionic conductivity NbO coating (<10 −10 ) is unable to do so. 21Regardless, both ALD-protected copper foils offer significantly lower surface area lithium and a much more uniform deposition surface with low porosity.These morphological improvements are likely a significant contribution to the improved cycling performance of the ALD-NbO and ALD-LiNbO cells.The ALD niobium-based coatings acting as an artificial SEI, facilitating smooth and planar lithium deposition, is supported by the prior literature on ALD coatings in both anode-free and lithium metal environments. 24,26,28,29However, some alternative coatings, such as titanium oxide, reside as a nucleation layer at the Cu-Li interface. 30Further investigation is required to confirm this assessment.
To observe the stability of the lithium surfaces formed and the interfaces of the ALD coating and lithium, as well as the feasibility of these cells for industrial use, a 7-day storage test was performed.NMC | |Cu cells were charged to 4.5 V before being fully discharged to 3.0 V to show a baseline Coulombic efficiency measurement.
Cells are then charged back to 4.5 V at C/20 before being held for 7 days at room temperature and then discharged.Any capacity loss during storage may be the result of parasitic side reactions or continued SEI formation.While the ALD-NbO had an average Coulombic efficiency of 89% after 7 days of storage, the ALD-LiNbO showed an average Coulombic efficiency of 91%, demonstrating a more stable interface at full lithiation storage.Comparatively, the bare copper current collector showed a Coulombic efficiency of 78%, a significant reduction compared to the ALD-protected copper foils, as shown in Figure 4A.Similarly, the ALD-NbO current collector showed a 150 mV voltage loss during storage, while the ALD-LiNbO only lost 140 mV over the 7 days.These are both significant improvements contrasting with the nearly 300 mV lost during storage by the bare copper current collector.
Similarly, EIS testing was conducted after five cycles and after 50 cycles to observe the impedance growth in the cell during cycling.While impedance growth is normally the cause of cell death in cells that cycle much longer, EIS can offer some insight at earlier cycle times, particularly into the stability of the interfaces formed.Here, after the first few cycles we observe the initial impedance of the NbO and LiNbO is similar in magnitude to that of the bare Cu surface.However, after cycling, a few key trends can be seen in Figures 4B,C.The bare Cu cells cycled at 1 mA/cm² show drastically increased impedance throughout cycling without immediate stabilization, jumping from 50 to over 300 Ω in total resistance.Likely, the Li interface is unable to accommodate lithium deposition, resulting in high surface area, high porosity lithium, as well as an abundance of dead or mossy lithium and greatly increased nonconductive SEI formation, and thus significant impedance growth.When compared to the ALD-protected copper foils, we can see that the impedance has continued to reduce during cycling and presents a much more stable EIS profile than their bare Cu counterpart.However, we can see that the ALD-NbO current collector has shown a greater reduction in total impedance compared to the ALD-LiNbO (~35 Ω compared to ~25 Ω).This is likely due to the increased surface area lithium deposition observed in the ALD-NbO SEM compared to its lithiated counterpart.Lower cell impedance can be positive for cell cycling, particularly for rate capability, but with increased surface area lithium deposits, this could be the result of the slow lithium transport through the NbO layer.Conversely, the higher initial impedance in the ALD-protected copper foils is consistent with the higher nucleation overpotential observed in the Li||Cu cell cycling.However, the ALD-protected foils show a clear and stable interface after cycling, further supporting the ability of NbO and LiNbO to stabilize the current collector surface for anode-free battery performance.
To get a better sense of the protection layer, X-ray photoelectron spectroscopy (XPS) was conducted to observe the chemical state of the niobium in the ALD coatings.In Figure 5A,B, we can clearly see the existence of Nb 2 O 5 in both the ALD-NbO and ALD-LiNbO.When we add the lithium subcycles to the ALD coating, a clear Li 2 O peak emerges.Both spectra show a significant Nb-OH peak, which could be the result of adsorbed hydroxide ligands on the surface as ALD functions by regenerating the -OH group to attach the next layer, but it also could be the result of some reactions between the water precursor and deposited Nb layers.In the Nb 3d spectra, we see a clear Nb 2 O 5 peak in both ALD-protected layers, once again confirming successful deposition, but we see a subtle oxidation in the lithiated sample in Figure 5D.This is likely due to the additional oxygens deposited from the Li 2 O subcycle layers drawing some electron density away from the niobium subcycle layers.To better understand the chemical environment of the ALD deposited films, Nb L3-edge XANES of the ALD protection layers can be seen in Figure 5E.In the XANES data, spectral features occur from the excitation of core electrons to bound states, unlike the photoelectrons of XPS.In this case, the L3 edge probes the 2p 3/2 to 4d 5/2 transition.
Here, we observe three main spectral features: peaks A and B are a clear double peak at the white line energy, while peak C is a smaller and broader feature at high energy.As before, the main two peaks are representative of the transition to the d-orbital.The doublet nature of this peak is driven by the interaction between the Nb atoms and their surrounding oxygen ligands, while the higher energy peak C can be attributed to the 2p 3/2 to Nb 5s transition. 31The first peak for both the ALD NbO and ALD LiNbO reference, as well as the ALD Nb 2 O 5 reference, are well aligned at 2373.0 eV.The ALD LiNbO 3 reference first peak is shifted to a slightly lower energy of 2372.7 eV.Regardless, the general agreement of peak A in all four samples is indicative that the ALD coating is depositing with Nb in the +5 oxidation state.Interestingly, the relative peak intensities of peaks A and B can be indicative of an octahedral or tetrahedral environment-where peak A being larger than peak B is an indicator of an octahedral coordination environment. 32As such, the thin film Niobium coatings present the Nb atoms in octahedral environments.However, there are some key differences.The energy splitting in the ALD LiNbO coating is 2.9 eV, while the ALD NbO coating is 1.7 eV.Compared to the 3.2 and 4 eV energy gaps of the Nb 2 O 5 and LiNbO 3 reference spectra, respectively, these are significantly reduced.The reduced separation of the two peaks is indicative of some distortion of the octahedral coordination environment, which is to be expected.Given the amorphous nature of ALD coatings, it is clear that local interactions are more prevalent than long-range order.Notably, after cycling, the same X-ray absorption features remain in good agreement with the pristine coating peak shifts, as observed in Supporting Information S1: Figure S8.Given the amorphous nature of these coatings, it is unlikely that ALD NbO or ALD LiNbO have any propensity to lithiate further, as evidenced in other studies utilizing high annealing or calcination temperatures to obtain a favorable crystal structure and lattice parameters for lithium intercalation or release. 33,34o demonstrate the realistic application of these protective coatings, the protected copper foils were used in full coin cells.As shown in Figure 6A, the bare copper foils remained unable to adequately cycle beyond the first 15-20 cycles and quickly lost any usable capacity due to an extremely low Coulombic efficiency.As evidenced by the SEM in Figure 3, without any excess lithium from a pure lithium metal electrode, the porous and high surface area lithium deposits quickly react with the electrolyte, causing rapid capacity fade and eventual cell death.As such, it is clear that bare copper foil is not a suitable current collector substrate for anode-free full cells.By comparison, both the ALD-NbO and ALD-LiNbO show dramatically increased cycling performance.After 50 cycles, the ALD-NbO-modified current collector is able to retain 80% of its initial capacity, with 20% of its initial capacity retained after 100 cycles.As a further improvement, the ALD-LiNbO is able to retain over 80% of its capacity at 100 cycles and over 65% at 150 cycles.As these cells are cycled in the 3-4.5 V range, a small amount of lithium remains at the end of each cycle after the first initial plating from the cathode.By allowing for a more uniform, low surface area lithium deposition, the ALD protective layers are able to inhibit the loss of lithium inventory and extend cycling significantly by maintaining this thin lithium layer for much longer than is possible on a bare copper current collector, before experiencing capacity decline through Coulombic inefficiency.
By calculating the volume of each of the three single-layer coin cells, we can calculate the volumetric energy density and the total energy cycled over the 80% capacity retained lifetime of these cells per unit volume.Over the course of cycling, the LiNbOprotected current collector is able to deliver almost three times the amount of energy delivered by the NbO-protected current collector and remarkably displays an almost 10-fold improvement over the bare copper foil controls.When we consider the rate of capacity loss from the 80% capacity point onward for all three systems, the improvement in total energy delivered is extended even further.

| CONCLUSIONS
In conclusion, we demonstrate that niobium oxide and lithium niobium oxide coatings deposited via ALD on the copper foil current collector offer a viable strategy to improve the cycle lifetime performance of anode-free cells.By depositing the ALD NbO and ALD LiNbO coatings, we increase the lithium nucleation overpotential and resistance of the current collector, resulting in low surface area, dense lithium deposits compared to the high surface area, and porous lithium that deposits on the bare copper current collector.We propose that the high room temperature ionic conductivity of the ALD-LiNbO coating of 10 −8 S cm −1 allows for ion transport through the film but prevents the formation of high surface area and short circuit causing lithium dendrites.The ALD-coated current collectors are able to drastically reduce the impedance growth throughout cycling compared to the bare copper control.We report anode-free coin cells cycled with NMC811 cathodes that are able to retain 80% of their capacity at 50 cycles with the ALD-NbO modified current collector and 80% of their capacity at 100 cycles with the ALD-LiNbO modified current collector.In comparison with the bare copper foil current collector, these represent threefold and sixfold cycle life improvements, respectively.In terms of total energy output per unit volume, over the lifetime of the cell, the modified ALD-LiNbO cells are able to provide almost 100 kWh/L more than their copper counterpart.This study demonstrates that ALD is a viable strategy for the improvement of anode-free batteries and that thin film current collector modification presents an exciting potential breakthrough in this research direction.

| Fabrication of ALD coatings
Niobium oxide and lithium niobium oxide films were deposited using a Savannah 100 Thermal ALD system (Veeco/carbon nanotube division of Veeco Instruments Inc.).For the deposition of the niobium oxide coatings, niobium ethoxide (Nb(OE t ) 5 , E t = −CH 2 CH 3 ; >99.9%; Strem Chemicals Inc.) was used as the Nb source while lithium tert-butoxide [LiO t Bu, (CH 3 ) 3 COLi, >99.9%;Alfa Aesar] was the Li source in the lithium niobium oxide ternary coatings.Deionized water was used as the oxidant for both coatings.Nitrogen was used as both the carrier and purge gas.The lithium source was heated to 170°C, while the niobium source was heated to 155°C.The manifold was held at 190°C to prevent condensation of the precursors.All depositions were performed with a substrate temperature of 235°C.Copper foil was cleaned using acetone, ethanol, and deionized water before drying with nitrogen gas.Samples were placed in a custom-made copper boat for deposition to prevent oxidation.Precursors were pulsed for 1s, with a 15s purge.Samples were quickly transferred to an Argonfilled glove box for storage after deposition.ALD recipes were prepared according to our group's previous work. 21LD Lithium Niobium Oxide coatings were prepared using the 1:4 Li:Nb subcycle ratio due to the most optimal ionic conductivity.

| Electrode fabrication
NMC811 coatings were acquired from GLABAT Solid State Battery Co. with a 16 mg cm −2 loading and an areal capacity of 2.5 mAh cm −2 before being punched into 12 mm electrodes.The anode current collector was a 12-μm-thick copper foil (3 M), which was punched into 12 mm electrodes.Copper foils were washed in a 0.05 M HCl solution to remove the native oxide layer, then rinsed with methanol and dried in a vacuum oven overnight before ALD coating or use as bare copper.

| Electrochemical measurements
Electrochemical analysis was performed using CR2032 coin cells.These cells were assembled in an ultra-pure argon-filled glove box with a polypropylene separator (2400; Celgard), with 100 μL of 0.6 M LiDFOB (99%; Alfa Aesar) 0.6 M LiBF 4 (99%; Alfa Aesar) in 1:2 FEC/ DEC (>99%; Sigma-Aldrich) electrolyte.The stripping and plating studies were conducted on a Neware battery system at room temperature.Li||ALD-Cu cells were cycled under constant current with a 6 h wetting period and 5 min rest between cycles, with a voltage range of 0-1.5 V.The initial lithium plating was conducted at 0.35 mA cm −2 to 3.5 mAh, then cycled at 0.5 mA cm −2 -1 mAh.NMC811||ALD-Cu cells were cycled under constant current with a 12 h wetting period and 5 min rest between cycles with a voltage range of 3.0-4.5V. Electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV) measurements were conducted on a Bio-Logic multichannel potentiostat 3/Z (VMP3) in the frequency range of 1 MHz to 0.1 Hz.Cells referred to as ALD-NbO were constructed with 25 cycles of NbO ALD, while ALD-LiNbO cells were constructed with 10 cycles of LiNbO ALD as these were the optimal thicknesses determined from half-cell performance testing.

| Characterizations
SEM images were taken using a Hitachi S4800 Scanning Electron Microscope at an acceleration voltage of 5 kV.Characterization of the electrode surfaces by SEM was conducted by disassembling the cells and rinsing the electrode with dimethyl carbonate to remove electrolytes and salts from the surface in an Ar glove box with <0.1 ppm O 2 and H 2 O.The electrodes were then mounted inside the glove box and sealed before being quickly placed inside the sample chamber to minimize exposure.XPS experiments were performed on a Thermo Fisher Scientific Escalab 250Xi (Thermo Fisher Scientific).Air-sensitive samples were loaded onto the sample stub in a glove box attached to the load lock of the instrument under Ar.After the pump down in the load lock, the sample stub was loaded into the instrument.The sampling position was determined via the video system, and the correct height was determined either visually or by the auto height mode of the system.Charge compensation was applied with the system's combined e−/Ar+ floodgun system.The spot size was typically 900 μm, although in some cases 650 μm.A survey spectrum was obtained (PE-100 eV) followed by a higher resolution (typically 20 eV pass energy, although higher pass energy was used for weaker peaks) for the elements of interest.All processing, as well as instrument operation, was carried out using the system's software (Avantage v.5.9925).Nb L3-edge XANES was conducted at the Soft X-Ray Microcharacterization Beamline (SXRMB) at the Canadian Light Source (CLS); ALD spectra are the average of multiple scans to improve signal to noise and smoothed with a boxcar filter with a kernel size of 11.
NMC111 with a 100 μm Li metal anode and (B) anode-free cells with their respective energy densities and thickness of varying components.
. The bare copper foil shows a first cycle plating overpotential of approximately 180 mV before the growth region plateau occurs at 90 mV.Comparatively, the ALD-NbO-protected Cu foil shows an initial overpotential of approximately 140 mV with a growth region plateau at 40 mV, while the ALD-LiNbOprotected Cu shows an initial overpotential of 165 mV and a F I G U R E 2 (A-C) Schematic diagram of Li plating on bare copper, ALD-NbO, and ALD-LiNbO.(D) Cycling of Li||ALD-NbO-Cu cells at different thicknesses of ALD coating.(E) Cycling of Li||ALD-LiNbO-Cu cells at different thicknesses of ALD coating.

F
I G U R E 3 (A) First cycle plating and stripping on bare copper, ALD-NbO, and ALD-LiNbO in Li||Cu cells.(B) Nucleation overpotential of Li||Cu cells with bare copper, ALD-NbO, and ALD-LiNbO.Numbers are the average of a minimum of three cells, and error bars are the standard deviation of the cells.(C, D) Scanning electron microscope (SEM) images of deposited Li on bare copper and higher magnification of image inside the purple square.(E, F) SEM images of deposited Li on ALD-NbO and higher magnification of image inside the purple square.(G, H) SEM images of deposited Li on ALD-LiNBO and higher magnification of image inside the purple square.

F
I G U R E 4 (A) Seven-day storage testing of bare copper, ALD-NbO, and ALD-LiNbO cells.Cells were charged to 4.5 V and discharged once to obtain first cycle CE, then charged to 4.5 V and held for 7 days before discharging to 3.0 V to observe retained capacity.All measurements are the average of at least three cells.(B) Electrochemical impedance spectroscopy (EIS) of bare copper, ALD-NbO, and ALD-LiNbO after five cycles.(C) EIS of bare copper, ALD-NbO, and ALD-LiNbO after 50 cycles.
NMC811||Cu full coin cell cycling from 3.0 to 4.5 V at C/5-D/2.(B) Total energy cycled per unit volume over the course of cell cycling.