A Systematic Study Aiming Toward Voltage Noise Elimination in Viscoelastic Poly(methyl methacrylate)–Poly(ethylene oxide) Polymer Blend Electrolytes in Li Metal Battery Cells

Random appearance of micro short circuits and voltage noise is observed, until now, on initial cycles of solid polymer electrolyte high voltage cathode cells with lithium metal anodes. However, this phenomenon also takes place in lower operation voltages such as those in LiFePO4 cathode cells. Approaching the direct/indirect effect of the cathode, the polymer electrolyte, and the anode parts on this initial dendritic bridging between the electrodes separately, a systematic study is carried out toward determining potential causes of voltage noise and ways of prevention. Electrolytes based on the blend of PMMA–PEO show that relying only on the viscoelastic nature cannot prevent the appearance of voltage noise. Liquid polymers are used for the wetting of the cathode active material that lead to a complete pore infiltration. Having ensured the optimization of the interfacial area between the solid electrolyte and the cathode, studies along the directions of the lithium surface pretreatment and polymer ionic conductivity are examined as a means of dendrite growth suppression and cell cycling capabilities improvement, leading to the observation that methylated PEO chains work in favor of both of these directions.


Introduction
The Lithium metal anode battery is a technology developed in the early 1990s, however the industry shifted quickly to graphitic anodes for electronics and mobility applications as a pareto of low cost, long cycle life and energy density. This trend still holds nowadays, even though significant progress has been made concerning battery materials. The highest capacity offered by lithium metal (≈3860 mAh g −1 ) comes with a major drawback, which DOI: 10.1002/aenm.202301035 is safety. Lithium's blessing is also its curse, standing at the top corner of the periodic table, as the most electropositive element after hydrogen. Each atom releases with ease an electron, contributing to the desired energy release provided by the battery. However, this unhesitating electronic transition makes lithium a notoriously reactive element. Its unstable life inside the battery is the cause of uneven deposition and dissolution, dendrite growth, unstable solid electrolyte interphase (SEI) creation and dead lithium formation, but most notably of potentially catastrophic failure when in contact even with minimum amounts of water. [1] The behavior of a metal upon deposition and stripping depends on its surface free energy, that is the energy needed to create an area of a particular surface. Lithium has a relatively low surface free energy, ≈0.52 J m −2 , [2] and therefore it is thermodynamically favoring the development of high surface area structures such as dendrites. [3,4] The tendency to form dendrites is enhanced by many other factors inside a Li battery, with the most prominent being surface morphology, interphase composition and hydrodynamic instability especially in high current densities. [5,6] Contrary to Li ion insertion anodic materials, lithium deposition sites are not equivalent. The randomness of the geometrical relief of the surface and the local potential create regions of preferable deposition. Furthermore, inhomogeneity in interphase composition results in a respective inhomogeneity in Li ion conductivity, channeling the ions through specific paths and therefore increasing the current density in specific regions of the anode. [7] Polymer electrolytes have been utilized in order to passivate the anode surface and inhibit the consumption of the electrolyte by creating the SEI upon initial contact. [8,9] At the same time a proper SEI would accommodate volume expansion and engineered to suppress dendrite formation. [10] Recently, the sudden failure in metallic lithium anode cells, expressed as randomly occurring fluctuations in cell voltage , [11][12][13] or voltage noise, has been observed in high voltage cells. These were studied and attributed to micro-short circuits developed due to the growth of temporary lithium dendrites , [12][13][14][15][16] while the same behavior has also been observed in graphitic anode high voltage cells. [17] In order to suppress dendrite formation, a considerable amount of effort has been devoted on developing methods and materials for metallic lithium surface treatment or for the formation of an artificial SEI (ASEI), [18][19][20][21] and avoid the implications of Li plating and stripping. Undeniably, the lithium surface roughness plays a significant role in homogeneous deposition and dissolution of Li. Decreasing roughness by roll pressing and consecutively enhance SEI formation by mechanochemical modification has been shown to suppress dendrite formation. [22] Mechanical surface patterning has been also utilized toward the opposite direction of increasing the surface area of lithium, in order to create preferred plating sites and alleviate the unwanted high surface area lithium formation. [23] Given the roughness of lithium, one general approach for the development of an ASEI is focused on the constituent salts, [24][25][26][27] solvents, [21,26,28,29] and additives, [30][31][32][33] of the electrolyte, thereby tailoring the composition in such a way that their reaction with lithium would lead to the formation of a SEI with desirable properties. [34][35][36] In a more controlled manner, an ASEI can be engineered using organic, inorganic or hybrid coatings on top of metallic lithium prior to cell sealing and operation. [37][38][39][40][41][42][43][44][45] The philosophy behind this approach is to restrict the electrolyte from coming into direct contact with the metal anode, while at the same time create an interphase that can limit the formation of dendrites. For example, Brown et al. created a film of waxy low molecular weight PE-b-PEO diblock copolymer on the surface of a PEO film electrolyte, which would mediate the contact with lithium and restrict the passivation by the electrolyte itself. [46] In the same manner, a thin film of high molecular PEO spin-coated on a Cu electrode was found to regulate the flux of lithium and suppress dendrite formation. [47] In this study we follow a systematic approach in order to address the problem of voltage noise and micro short circuiting in lithium metal anode cells containing solid polymer electrolytes and to eliminate the random formation of dendrites by means of lithium surface pretreatment. Starting from the geometric changes in solid polymer electrolytes of PMMA-PEO blends, significant thickness reduction, due to coin cell crimping stress accumulation and operation temperature have been observed, which increase the probability for dendritic bridging of the electrodes. Another factor contributing toward this direction is the percentage of cathode pore filling and consequently the utilization of cathode effective area. Using liquid polymer electrolytes, a complete infiltration throughout the volume of cathode pores was attained optimizing the interface between the solid electrolyte and the cathode. Examining the effect of polymer ionic conductivity and Lithium surface pretreatment method, it was found that voltage noise can be efficiently suppressed and the cell cycling capabilities can be significant improved when PEO chains are terminated in methyl groups.

Electrolyte Film Deformation
PEO based solid polymer electrolytes (SPEs) have been reported to be susceptible to dendrite formation in high voltage operating cells such as those with NMC active material cathodes. [12] In this study, it is observed that this is a general phenomenon which is also evident in lower operating voltage chemistries such as in LFP cathode cells. The probability of dendrite formation depends on electrolyte film geometry, electrolyte elastic modulus and Lithium surface morphology, while the growth rate is connected to the electrical and chemical properties of the various phases (salts, solvents, SEI, etc.). A low SPE film thickness provides a shorter route for a potential dendrite to bridge electrodes and a higher probability of voltage noise.
Based on this knowledge, the 50% PMMA-50% PEO550 blend system is investigated as a candidate solid electrolyte because of the large volume fraction of PMMA. Together with the high glass transition temperature, PMMA has a molecular weight that is well above the entanglement limit (27.5 kg mol −1 [48] ), these two factors provide a viscoelastic behavior with more pronounced the elastic part that could potentially prevent dendrite growth. Before going to the electrolytic behavior, it would be appropriate to test the electrolyte film geometry changes.
During the coin cell preparation, there is small control over compression and the final internal pressure on the materials after crimping, especially when the cell's temperature is raised. In order to quantify the SPE film change in dimensions, the impedance of 50% PMMA-50% PEO550 LiTFSI film was measured at room temperature, at 60°C and then back down to room temperature as shown in Figure 1, in a time span of 7 days. The measurement had been conducted in a stainless-steel blocking electrode configuration, as described in Figure S1 (Supporting Information). It was observed that the initial resistance of the electrolyte was 8 kΩ and after heating and cooling, it was stabilized at 3.5 kΩ. This difference is solely due to the accumulated crimping stress that was partially relieved when the electrolyte was heated and the change of thickness was measured to be 40 μm, which translates to 30% of the initial thickness. Assuming a constant material density, the radius of the film has been calculated to be increased by 23%. For this reason, all the consecutive studies of SPE films were performed using the spacer aforementioned in the cells.

The Role of the Cathode
Even though the cathode does not have a direct effect on dendrite formation during lithium plating, it plays its role in the process through its infiltration ability. Cells with different LFP treatment have been prepared using 50% PMMA-50% PEO550 LiTFSI SPE films sandwiched between Li metal and LFP. The LFP cathode has been treated in four different ways by means of pore infiltration with either PEO550, PEO250-dME, or PEO250-dME LiTFSI. For comparison infiltration with EC/DEC LiTFSI 1 M carbonate electrolyte has been also done. In all cases the liquid electrolyte has been dropped casted on the LFP surface and left in inert atmosphere (Ar) for one day at room temperature before the cell preparation. The maximum capacity acquired during the first cycle of galvanostatic charging, with a rate of C/20 (0.103 mA cm −2 ), until the onset of voltage noise for each case, with respect to the total resistance (bulk plus interfaces) can be seen in Figure 2.
The resistance was calculated from the impedance spectra. The resistance could be related to the wetting capability and/or the conductivity of each infiltrating liquid. The sole effect of conductivity would lead to orders of magnitude of difference between the points in Figure 2, given that the infiltration is complete. Therefore, what is observed is a combination of both wetting capability and conductivity. The resistance provides an indication and not a measure of pore infiltration, and a more direct way to determine the pore filling will be discussed in the following analysis.
The voltage noise appears earlier and the stored capacity decreases as the resistance increases, as indicated by the datapoints designated as film in Figure 2 for the polymer electrolytes. Although the correlation between cathode infiltration and voltage noise is not fully understood, a possible explanation could be that a high resistance sourcing from poor infiltration indicates a lower interfacial area between the electrode and the electrolyte, which translates to high current density, in order for a specific charging rate to be applied, and therefore a greater probability for the formation and growth of dendrites on the lithium side.  We expect that the wetting with the carbonate electrolyte leads to improved LFP pore infiltration and its resistance was calculated at 28 Ω. When a Celgard separator is inserted between the cathode and the SPE, as indicated by the "Film + Celgard" point in Figure 2, we get a higher capacity at voltage noise onset. This behavior complies with the fact that lithium bridging originates from the anode side and that one layer of Celgard separator can only delay the growth of a fully extended dendrite but not prevent it.
To acquire a more accurate view of the nature of dendrite growth, cyclic voltammetry at 60°C was performed on symmetric Li | 50% PMMA-50% PEO550 LiTFSI | Li cells as seen in Figure 3. During the positive potential increase, on the first run, nothing peculiar is observed, while on the negative sweep some current spikes are seen, beginning at −2 V. However, no spikes are seen on increasing the potential back to 0 V. On the second run and increasing potential, two small spikes are barely discerned while, on negative potentials a large number of current spikes, starting as early as −1.5 V, emerge at decreasing as well as increasing potential. Finally, on the third run current spikes appear above around ±2.5 V. Even with the existence of the not so flexible and entangled PMMA, at a volume fraction of 50%, dendrites are able to form and penetrate through the matrix of the SPE. It is interesting that, these short circuits do not appear at a specific absolute value of the potential and are not permanent, since there are regions of the runs where no current spikes are observed.
The blend with 50% w/w PMMA was not capable to restrain dendrites from forming. We attribute this to the lack of transverse cohesion in the electrolyte in order to accommodate local lithium volume changes and prevent Li dendrite penetration. At this point it must be noted that the high elastic modulus that is commonly referred to, when characterizing a solid electrolyte based on the work by C. Monroe and J. Newman, [49] as a means of suppressing dendrite formation is true on the basis of an idealized 2D displacement at the interface. This study applies to the initial stage of dendrites forming at local high curvature points that begin to apply a stress and deform the electrolyte. However, it is possible that when dendrites have passed this stage, they penetrate into the matrix of the electrolyte by breaking the van deer Waals or dispersion bonds and separating the molecules, thus creating new surfaces. In order to account for the whole phenomenon of dendrite formation and growth, and consequently engineer solid electrolytes with the necessary properties, the surface free energy of the electrolyte, as a material parameter, will be revisited at future work.

Electrolyte Stability and Homogeneity
Since the 50% PMMA-50% PEO LiTFSI electrolyte has not the necessary properties to prevent voltage noise, a combination of viscoelastic properties and lithium surface treatment was examined as a next step. Reducing the volume fraction of the glassy phase, the viscosity of the electrolyte decreases and hence the conductivity increases, as it can be seen in Figure 4a. The blend 30% PMMA-70% PEO550 LiTFSI has an order of magnitude higher conductivity while it maintains its viscoelastic character. Even though the objective is to apply a pretreatment on the lithium surface, a crucial point for the operation of the electrolyte would, in any case, be its stability when in contact with the lithium metal. According to our previous work when the PEO chain is capped with methyl groups it forms a more stable interphase. [50] Following the same procedure, the electrolytes were left in contact with the lithium metal in symmetric cells and impedance spectra were collected with time. The resistance was determined by the impedance spectra shown in Figure S2 (Supporting Information). Comparing the two electrolytes, 30% PMMA-70% PEO550 and 30% PMMA-70% PEO1k-dME, in Figure 4b it is observed that the dimethyl ended PEO blend shows a larger variation in aerial resistance which stabilizes at a higher value at which point a more robust SEI seems to be formed.
Concerning the bulk electrolyte conductivity, the higher volume fraction of PEO increases the probability of phase separation between PMMA and PEO which would also result in PEO crystallization. In order to examine this possibility, the impedance spectra have been measured, in a blocking electrode setup, with respect to temperature and the dc conductivity has been determined as plotted in Figure 4c. Both electrolytes present a Vogel-Fulcher-Tammann dependence with no sign of crystallization down to 10°C for the 30% PMMA-70% PEO1k-dME LiTFSI electrolyte and −20°C for the 30% PMMA-70% PEO550 electrolyte and therefore remain in a homogeneous blend. Both electrolytes have almost the same conductivity, 0.017 mS cm −2 , at 20°C while at 60°C the 30% PMMA-70% PEO1k-dME LiTFSI has two times higher conductivity, 0.44 mS cm −2 compared to 0.19 mS cm −2 , since it presents a more sensitive dependence on temperature.

Lithium Surface Pretreatment
Having proved the homogeneity of the electrolytes the next step was to apply different lithium surface treatments. The prepared electrolyte free standing films, Figure 5a, where sandwiched between lithium metal and an LFP cathode without or with the pretreatment of lithium and enclosed in coin cells. In all the cases, the LFP was infiltrated, as described earlier with PEO250 dME LiTFSI for the maximization of the interfacial area with the solid electrolyte. We can safely say that the percentage of pore filling by PEO250-dME LiTFSI is practically complete for all the prepared cathodes. From the galvanostatic cycling, at C/20, of the non-pretreated cells, the characteristic voltage noise is making its appearance, as can be seen for example in the 30% PMMA-70% PEO1k-dME case shown in Figure 5b.
For the pretreatment of the lithium surface either PEO550 LiTFSI or PEO250-dME LiTFSI was used. Each electrolyte was enclosed in a symmetric Li metal electrode coin cell and left for the interphases to form and monitored by impedance spectroscopy ( Figure S3, Supporting Information). When the interphases reached a stable state, after 7 days, the four pretreated lithium electrodes were extracted and used to form four different cells with LFP, i.e., two with 30% PMMA-70% PEO550 LiTFSI film electrolyte and the other two cells with 30% PMMA-70% PEO1k-dME LiTFSI.
For the prepared cells, impedance spectra were acquired at 60°C right before the beginning of cycling, and some qualitative similarities can be observed in the Nyquist plots as seen in Figure 6. The cells where the lithium was surface treated by PEO550 LiTFSI (light and dark blue in Figure 6a) present two clear semicircles corresponding to two different interphases of the SEI. The elongated second semicircle in both cases indicates a wider distribution in characteristic ion transport times and therefore, a wider compositional variation of the specific interphase which, according to our previous studies, has an inorganic origin. [50] The first semicircle with narrower time distributions is related to the organic interphase which is superimposed by the bulk electrolyte resistance.
The cells where the lithium was treated with PEO250 dME LiTFSI (green and orange in Figure 6a) also present qualitative similarities. In this case, only one semicircle is observed due to the organic SEI which, as before, is superimposed by the contribution of the electrolyte. It is also observed that the cells with the 30% PMMA-70% PEO1k-dME LiTFSI electrolyte significantly lower total resistance.
The effectiveness of the lithium metal surface pretreatment was tested by galvanostatic cycling of the cells at a rate of C/20, set at this value due to the low conductivity of the electrolytes. Due to high polarization in the cell the cut off voltage was set at 3.8 V, instead of the usual 3.6 V for LFP. The voltage profiles are presented in Figure 7 for the 5th, 10th, 15 th , and 20th cycle.
Focusing on the conducting phase and the treatment method, the PEO550 cell with the PEO550 treatment, Figure 7a shows a substantial charge accumulation behavior, indicated by the increased slope of the voltage profiles, and a high overpotential, 0.28 V as calculated by the mean value of the difference between the charging and discharging curves. The same also holds for the cell with the PEO250 dME treatment. The quantity of lithium that finds its way inside the LFP, is limited by the conductivity of the phases that traverses, namely the electrolyte and the SEI, and by the wetting of LFP pores. Figure 7c, shows slightly less charge accumulation and an overpotential of 0.16 V. However, the capacity in these two cases is limited by the high polarization of the cell.
This picture changes completely when PEO1k dME is used for the conductive phase and the electrolyte conductivity is increased. When lithium gets the PEO550 treatment, Figure 7b, the voltage profile exhibits a smaller slope approaching the expected LPF charging plateau. Furthermore, the overpotential is further reduced to 0.13 V. Finally, for the PEO1k dME with PEO250 dME treatment, Figure 7d, the overpotential is reduced to 0.11 V with the charging and discharging curves being more consistent between the cycles, and reaching the almost flat voltage profile of LFP. Now, the degree of pore infiltration is the limiting factor.
The observed specific capacity is not very high in absolute values. However, the abrupt increase of the voltage profile at the end of charging in Figure 7b but mostly in (d), indicates the completion of charging. The charging is regarded complete when it reaches a point where there is a large and abrupt increase in cell voltage with practically zero increase in stored capacity, as it is the case in Figure 7d. There, we observe an increase from 3.6 to 3.8 V with a capacity gain of less than 1 mAh g −1 . However, we see that, from the nominal 170 mAh g −1 of LFP, only about 22 mAh g −1 are tapped and that is because only this quantity of charge has access to the electrolyte. In this way we can determine the actual  percentage of LFP wetting by means of the ratio of the experimental capacity to the nominal capacity estimating that 11.8% of the mass of LFP is actually accessible by the lithium ions.

Conclusion
Coin cell type of cells offer an ease of construction, by the simple enclosure of the materials, however, there in a minimal control over the internal stress under crimping which should not be overseen. The accumulated crimping stress could alter the behavior of the electrode materials and also have a significant effect on the electrolyte geometry. Given the variation of elastic properties of PMMA-PEO solid polymer electrolytes with temperature, a change of over 30% in the thickness of the electrolyte was observed upon crimping and heating.
The lower operation potential of the LFP cathode does not automatically reject the possibility for formation of micro-short circuits and voltage noise in lithium metal anode cells. The indirect effect of cathode infiltration has been shown to play a significant role in voltage noise appearance. The percentage of pore infiltration translates to the percentage of the total effective area exploited during charge and discharge. A limited area would result in high current densities through specific paths inside the electrolyte, which in turn can lead to uneven lithium plating and dendrite formation. According to our understanding the most critical parameter, in order to completely suppress dendrite formation, would be the lithium surface pretreatment which would also protect the electrolyte from decomposition. On the other hand, the geometric characteristics would only delay the phenomenon rather than eliminate it.
We speculated that the usage of a high glass transition temperature and elastic modulus phase such as PMMA would prevent voltage noise. However, even at a high concentration of 50% w/w with proven homogeneity of the PMMA-PEO blend and robust SEI formation, this was not the case. Furthermore, it was proved that a single layer of separator on top of the solid polymer electrolyte, could not deter the appearance of micro-short circuits. Investigating toward two different directions, namely ionic conductivity for the electrolyte and different lithium pretreatment methods, we established that in both cases the voltage noise was suppressed. Regarding the effect on cycling capabilities in terms of capacity and polarization, the worst behavior was observed in the case where both the electrolyte and the pretreatment consisted of hydroxy terminated PEO. On the opposite side, dimethyl terminated PEO in the electrolyte as well as for the lithium pretreatment, showed a flat LFP charging and discharging curve, a significantly lower cell polarization and higher capacity overall.
By changing the PEO chain, and therefore the ionic conductivity, and the treatment method, there is a significant improvement toward ionic conductivity, cell polarization and specific capacity, which increased by five times. The low values of specific capacity indicate that the limiting factor is the degree of LFP pore infiltration during wetting. By comparing the expected versus the experimentally acquired specific capacities we concluded that only about 12% of the LFP was accessible to lithium, indicating that special care should be taken even in the case of liquid electrolytes when in contact with porous electrodes.

Experimental Section
Materials: Material preparation and cell assembly were carried out in an inert Ar atmosphere glovebox. The polymer materials were purchased by Merk and listed in Table 1 along with their nomenclature throughout the text. These were separately vacuum dried for over 72 h at 60°C. LiTFSI with a purity of 99% was purchased from thermoScientific and vacuum Table 1. Polymer materials combined with LiTFSI used in the preparation of the electrolytes.

Nomenclature
Material description PEO550 Polyethylene oxide (PEO) with a molecular weight of 550 g mol −1 and -OH chain ends PEO250 dME 250 g mol −1 dimethylated PEO with -CH 3 chain ends PEO1k dME 1000 g mol −1 dimethylated PEO with -CH 3 chain ends 50% PMMA-50% PEO550 Blend of polymethyl methacrylate (Mw 100 kg mol −1 ) (PMMA) with PEO550 in a 50% w/w concentration 30% PMMA-70% PEO550 Blend of 30% w/w PMMA (Mw 100 kg mol −1 ) with 70% w/w PEO550 30% PMMA-70% PEO1k dME Blend of 30% w/w PMMA (Mw 100 kg mol −1 ) with 70% w/w PEO1k dME dried for 72 h at 160°C. All the electrolytic systems had a salt concentration of [EO]:[Li + ] = 18:1 (r = 0.055). The polymers: PEO550, PEO250 dME and PEO1k dME are liquids above 60°C, therefore the electrolytes were prepared by adding the salt, in the above-mentioned concertation, directly into the polymer and magnetically stirred for 2 days at 70°C. The PMMA-PEO blends were first dissolved in tetrahydrofuran (THF) (99.9% Sigma Aldrich) 5% w/v, next the salt was added and the solutions were magnetically stirred for 2 days at room temperature. Cells: Lithium metal (Sigma Aldrich) and LiFePO 4 (LFP) with areal capacity 2 mAh cm −2 and specific capacity 170 mAh g −1 obtained by Customcells, were used for the anode and cathode, respectively. For the measurements, the materials were enclosed in CR2032 coin cell cases, with the electrolyte sandwiched between the electrodes. The electrolyte formulations were prepared by the combination of each polymeric material shown on Table 1 with LiTFSI. In the cases of liquid PEO550 and PEO250 dME electrolytes either a polypropylene fiber separator or Celgard 2500 was impregnated for the formation of the cells. The solid electrolyte films, based on the different blends of PEO with PMMA, were prepared by drop casting from the THF solution on to a stainless-steel plate. The solvent was evaporated and the film was vacuum dried for 2 days at 100°C. Finally, the free-standing films were peeled off the stainless-steel plate. In this case 3d printed Polylactic Acid (PLA) O-rings were used as spacers between the electrode during the preparation of the cells.
Measurements: The impedance spectra acquisition and cyclic voltammetry experiments were conducted on a PalmSense 4. The impedance analyzer mode was operated in the frequency range 10 mHz to 1 MHz and ac voltage amplitude 0.01 V, while a scan rate 0.1 mV s −1 was applied for the voltammetry. Galvanostatic (Constant current and constant voltage) cycling was conducted on a Basytec CTS Lab testing system.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.