Composite Electrode Ink Formulation for All Solid-State Batteries

A conventional solid-state-reaction method was employed to prepare Al-doped LLZO powder, using LiOH, La₂O₃, ZrO₂, and Al₂O₃ as the starting materials. Briefly, 2.5 g LiOH (pre-dried at 200 C for 6 h), 6.6 g La₂O₃ (pre-dried at 900°C for 12 h), 3.3 g ZrO₂, and 0.15 g Al₂O₃ were mixed and ball milled for 4 h at 500 rpm (Pulverisette 7 premium line, Fritsch). The mixture was then annealed at 800°C for 10 h after drying. The obtained powder was ball milled again for 4 h at 500 rpm and dried overnight under vacuum. Pellets (1 mm thick, 10 mm diameter) were prepared by uniaxial pressing with a pressure of 400 MPa. The green body pellets with density of 50% were covered with the same mother powder and sintered in alumina crucibles at 1065°C for 6 hours.

Electrode inks are composed of 10 wt% carbon black, 10 wt% binder (PVDF or PVB), 40 wt% Li₄Ti₅O₁₂ (LTO) and 40 wt% LLZO powder. LTO was sourced from MTI Corporation and was used as received (D10 = 0.2–0.6 μm). The total solid content in the electrode ink was 30 wt%. All inks ball milled at 500 rpm for 30 min. THF/toluene (4:1 volume ratio) or terpineol was used as a solvent. LLZO pellets were dry polished to a thickness of ≈600 μm, sonicated in alcohol and dried under vacuum. Cathode slurries were cast on LLZO pellets and dried overnight at 80°C in an Ar-filled glove box. The gravimetric density of the active cathode material was controlled to be 2.5 mg cm⁻². Li metal foil was molten on the other side of LLZO pellets on a hot plate with a temperature of 180°C in glove box. ASSBs with configuration of LTO/LLZO/Li were assembled in 2032 coin cells in glove box. The tests were conducted at 95°C. Cells were cycled between 2.5 V and 1.0 V with current densities of 5, 12.5, 25, 5 A kg⁻¹ sequentially for charge-discharge tests. The currents for galvanostatic charging were calculated using the active material loading in the electrode. The capacities are normalized to the active material loading in the electrode.

Stress test was conducted with a mechanical testing system (Instron 5944, Instron, USA). Ceramic substrates were fixed by super glue and testing rods were attached to electrodes by using super tape. The crystal phase was detected by powder X-ray diffraction (XRD, Rigaku Smartlab, Regaku, Japan) with a step of 0.01°. Surface tension of the cathode inks was measured using a bubble pressure tensiometer (Kruss BP50 tensiometer, KRUSS GmbH, Germany). Surface tensions were recorded for bubble ages from 15 ms to 2000 ms at approximately 23°C. The instrument was calibrated with a water reference prior to each measurement. Electrode inks were rheologically characterized using a DHR3 Hybrid Rheometer (TA instruments, USA) in a stress-controlled mode. Inks were pre-sheared at 100 s⁻¹ for 2 minutes and allowed to rest for 10 minutes prior to any tests to remove any mechanical history in the samples. Average of at least three measurements are reported. Suspensions for particle size measurements were made by dilution of ball-milled inks to 0.3% solid loading. Malvern NanoZS Zetasizer instrument was used for dynamic light scattering studies. The physical properties of solvents used for the experiments are listed in Table I. Images of the anodes were taken on an optical microscope. X-ray photo electron spectroscopy methods analyses were performed using an Ulvac-PHI Versaprobe 5000. Different LLZO samples were submersed in each solvent for a period of 1 minute. Samples were vacuum dried and stored in a glove box for transfer to the XPS instrument. In order to minimize air exposure, samples were transferred in a sealed container. Samples were mounted onto a sample holder using BeCu clips. The mounted samples were pumped in the intro chamber until the pressure dropped below 1 × 10⁻⁶ torr. Monochromatic Al α X-rays (1486 eV), a 100 μm diameter X-ray spot rastered over a 600 μm by 400 μm area, and a takeoff angle of 45 degrees off sample normal were used in each acquisition. Pass energies of 187.7 eV and 23.5eV were used for the survey and high-resolution acquisitions, respectively. Charge neutralization was accomplished using 1.1 eV electrons and 10 eV Ar⁺ ions. CasaXPS processing software was used to analyze the resulting data. Binding energies were calibrated to -CH2- type bonding in the carbon 1s spectrum of 284.8 eV.

A lithium titanate composite electrode demonstrates drastically different performances depending on how the ink is formulated. Depending on the binder|solvent formulations, capacities between 1 and 15 mAh/g are observed at 5 A/kg charging rates (Fig. 2a). We observed that PVB gave both the best performance (Terpineol) and worst performance (THF/Toluene) depending on the ink's solvent. Both electrodes demonstrate an interesting behavior of starting with a very low capacity and increasing rapidly in the first 10 cycles. This 'break-in' period may be required to activate the polymer with either ionic carriers or to mechanically alter the polymer. Composite electrode inks formulated with Terpineol (independent of binder) perform better than inks formulated with a mixture of THF/Toluene (Fig. 2b).

Zoom In Zoom Out Reset image size

Engineering inks requires precise control over how the ink flows (rheology/fluidics), how the active material orients and aggregates, and the ink stability.¹⁰ Polymer-solvent interactions can impact the structure the polymer takes in the electrode and how well an ink flows.⁴⁶ Forming stable liquid-air interfaces during solution processing enable uniform coating and limits defects (i.e. coffee ring effect).^47,48 Dynamic surface tension measurements describe how effective a liquid is at forming a stable interface with respect to a property known as surface age.^47,49 During coating, the inks are subjected to a sudden increase in surface area.⁴⁹ Component diffusion occurs within the ink as a result of surface free-energy minimization.⁵⁰ Polymers typically align themselves at the free interfaces to accommodate hydrophilic-hydrophobic interactions within their structure and help in minimizing the surface tension.⁵¹ Dynamic surface tension for a good ink will decrease rapidly and maintain a constant surface tension with surface age (Fig. 3a). The surface tension of neat terpineol, tetrahydrofuran, and toluene are 33, 26, and 29 mN/m (Table I). One of the inks contains THF/Toluene in a 4:1 ratio, so we can extrapolate that the surface tension of the blended solvent (THF/Toluene) is lower than terpineol (Fig. 3a). All polymer solutions except PVB-terpineol system show surface tension behavior very close to that of the pure solvent system. This suggests that polymer solvation within these systems is favorable and the polymer-particle interactions within these systems would be minimal. The increase in surface tension at low surface ages for the PVB-terpineol system indicating that polymer-particle interactions will be higher for this system. Surface tension is also related to the vapor pressure of the system.⁵² Surface tension arises from cohesive forces acting on the molecules at the free interface with the atmosphere. Higher surface tension implies a stronger cohesive energy for the molecules within the surface layer leading to a reduced vapor pressure. When an electrode is cast, a liquid supernatant must form for solvent evaporation. During this step, the polymer diffuses to the free-boundary of the system to reduce the free energy. If the surface tension at this polymer rich 'skin' is low, polymer can be lost due to volatilization.³² It is hypothesized that PVDF (a linear-chain polymer) that can reorient at the free liquid-air interface and thus result in lower surface tensions. This may result in polymer diffusion to the free surface will be preferred in PVDF containing inks that can lead to polymer loss due to volatilization as well as a non-uniform composition within the electrode. In contrast, the larger PVB molecule is harder to re-orient at the free interface that may result in a higher surface tension. This can prevent polymer diffusion and subsequent loss due to volatilization and help maintain a uniform composition of the cathode. In contrast, PVB in THF/Toluene blended solvent shows surface tension similar to that of the neat blended solvent. PVB in this solution is well solvated, potentially restricting the binder-particle interaction.

Zoom In Zoom Out Reset image size

Electrode slurries generally consist of two internal structures: aggregates and agglomerates.²⁴ Aggregates are collection of primary particles (active material, solid electrolyte, carbon) that are bound together by strong (usually electrostatic) forces. Agglomerates are weak clusters formed of the aggregates by weak Van der Waals forces. Ideal ink structures should have the smallest aggregate size while providing mutual arrangement of the individual components to provide electronic and ionic conduction. Smaller aggregate sizes are favored, as this increases the available electrochemical surface area and enhances the capacity of the battery. LTO inks with PVB binder and terpineol solvent show the lowest aggregate size in dispersion (≈0.5 μm) with no agglomerates. The aggregate sizes increase in the order PVB-terpineol<PVDF-terpineol<PVB-THF/Toluene<PVDF-THF/Toluene (Fig. 3b). Agglomerate structures are only identified within the THF/Toluene cathode inks. Cathode inks with the smallest aggregate size results in the largest capacity measured.

Ink aging is another important consideration for processing. Shelf life of inks have significant commercial impact and ideally should remain stable over production cycles. Optical images of the electrode inks show that the PVDF containing systems are unstable, while PVB systems showcase stable behavior irrespective of the solvent. DLVO theory for colloidal systems indicate two primary forces acting on a colloidal system: (a) Van der Waals' attraction (b) electrostatic repulsion. Additionally, polymer in colloidal solutions can offer two kinds of force mechanisms with particulate matter.⁵³ Steric interaction arises from differences in free volume of polymer chains attached to particles that leads to a repulsive force and stabilization of the colloidal system.⁵⁴ Depletion interaction arises from differences in osmotic pressure between solution trapped between two colloidal particles and the free solution.⁵⁵ The lower osmotic pressure in the vicinity of the particles pushes them together leading to an attractive force. PVDF containing inks show an unstable behavior in both the solvents studied (Fig. 3c). Both the solvent have very different surface tension (Table I) and dielectric constant (Toluene = 2.38, THF = 7.48, Terpineol = 2.3) resulting in opposing contributions from Van der Waals and electrostatic interaction. This suggests that the stabilization mechanism in these inks is polymer binder driven and not governed by the conventional DLVO forces. While PVB is explored as a binder, previous studies have suggested that PVB can also act as a steric stabilizer. The effectiveness of PVB as a steric stabilizer is dependent on the solvent quality. PVB systems are likely sterically stabilized for both terpineol and THF/Toluene blended solvent. Good solvents enhance the ability of the stabilizing moieties to adsorb with sufficient thickness on the colloidal particles, while at the same time have adequate polymer-solvent interaction to ensure that unadsorbed polymer segments can remain in solution and provide steric repulsion. On the other hand, poor solvents lead to collapse of the unadsorbed polymer segments to the particle surface due to low polymer-solvent interactions.^56–58

Rheology of electrode inks is an important factor while designing materials for manufacturing.¹⁰ Rheology provides information about the allowable process conditions based on the ink viscosity. Low stress conditions (0.1 to 10 Pa) are of interest for the coating technique employed in this study. Terpineol based inks show distinct yield stress plateaus followed by shear thinning region (Fig. 4a). THF/Toluene inks do not exhibit a yield stress plateau in the stress range studied. Yield stress in the inks indicate the rigidity of the internal ink microstructure under applied shear stress. Above the yield stress, the aggregates in the inks break up and lead to loss of contact between the components of the ink. PVB/Terpineol ink shows the highest yield stress for all the inks measured indicating a rigid, structured ink microstructure that results in a stable suspension that can withstand the shearing forces applied during coating. Even though PVB in THF/Toluene solvent results in a stable dispersion, its interactions with solvent and particles are inadequate and results in a poor dispersion. Optical images of the electrodes show that PVB containing inks result in electrodes with smaller feature sizes as well as pores (Figs. 4c–4f). Good dispersions should have well connected aggregates of active material, solid ion conductor and carbon additive with the binder acting to keep the aggregates from coalescing (Fig. 4b). Poor dispersions do not have stable interactions between the constituents. It is evident from the discussion that ink engineering needs to satisfy multiple criteria to yield good dispersions. Careful consideration of all factors affecting the ternary interactions between solvent, binder and the composite additives should be undertaken while designing the inks.

Zoom In Zoom Out Reset image size

Stability and effect of polymeric binder in composite electrodes can also be explained via structural analysis. XRD analysis of different cathode inks cast on glass substrates revealed strong presence of pure LTO and LLZO phases. Unwanted Li₂CO₃ phases were detected in the PVDF containing systems upon exposure to air (Fig. 5b). The formation of Li₂CO₃ is ascribed to a weak binding between the PVDF and the cathodes which allows LLZO to react with air.

Zoom In Zoom Out Reset image size

The interfacial interactions between the electrode ink dispersions and the solid electrolyte play a significant role in determining the resultant electrode. Contact angle measurements of the four cathode suspensions on the ceramic LLZO electrolyte surface was performed to elucidate the wettability of the inks. Inks containing terpineol have a higher surface tension compared to the THF/Toluene inks (Fig. 3a). PVB-Terpineol containing electrode inks show the lowest contact angle (12.3 ± 0.4°), while the PVB-THF/Toluene containing inks show the largest contact angle (31.7 ± 1.3°) (Fig. 5a). This shows that the LLZO-ink interaction is much stronger for the terpineol inks and results in improved contact with the electrolyte. Improved wettability ensures better contact of the active materials with LLZO ensuring proper ion conduction and thus leads to better capacity performance of the electrodes. It should be noted that the contact angle measurements are carried out on composite inks while the surface tension measurements are carried out on the polymer solutions. Hence, we observe differences in contact angles while the surface tensions of polymer solutions converge to a similar value for PVB and THF/Toluene solvents irrespective of the polymer type.

Electron and ion transport in electrodes and between electrodes and electrolytes is weakened by poor bonding. The selection of binder and solvent plays a major role on electrode bonding. A pop-up stress tests was performed on electrodes containing PVDF and PVB binders. The average adhesion of the electrode prepared with PVDF binder and THF/Toluene solvent was found to be the least (≈0.03 MPa) while the electrode prepared with PVB binder and Terpineol solvent sustained the maximum force of 0.54 MPa (Fig. 6a). Electrodes with PVB binder are more strongly bound than the electrodes with PVDF binder. This is coincident with the ink's stability test, as the aggregation and inhomogeneity in electrodes with PVDF binder can cause delamination leading to low cell capacity. PVB contains non-polar butyral groups and polar hydroxy groups, showing advantages to mixtures. Polar OH-groups adhere to oxides (LLZO, LTO) and nonpolar butyral groups provide more adhesion to conductive carbon additives.⁵⁹ The electrodes prepared with Terpineol solvent sustain higher forces than those with THF/Toluene solvent. As shown in Fig. 3b, the electrode slurries with the THF/Toluene solvent possess more aggregation, which results in greater active material and binder inhomogeneity. The electrodes are almost peeled off from the electrolytes for the PVDF binder system (Figs. 6b and 6d), but generally still hold their form for the PVB binder system (Figs. 6c and 6e).

Zoom In Zoom Out Reset image size

LLZO is known to undergo spontaneous proton exchange with Li⁺ when exposed to water and can form insulating carbonate products in air.^60,61 Both of these interfacial transformations can cause an increase in the interfacial resistance of the cell and a decrease in lifetime. To evaluate the solvent's role on the interfacial chemistry of the solid electrolyte we submerged pristine electrolytes into each solvent for a short period of time (THF/Toluene and Terpineol) and subsequently dried the samples under vacuum. The samples were then sealed and transferred to XPS to avoid air contamination. In the C 1s spectra, four adventitious carbon contamination contributions are observed at 284.6 eV, 286.4 eV, 288.3 eV, and 289.7 eV, corresponding to C-C, C-O-C, O-C=O, and CO₃, respectively (Fig. 7a). Since the pellets were polished in an air atmosphere, the characteristic peak of Li₂CO₃ is present. A noticeable increase in the carbonate is observed for the solvent treated LLZO in comparison with the pristine electrolyte. Furthermore, there was a pronounced difference between the terpineol carbonate layer (10.3%) and the THF/Toluene (4.2%) (Figs. 7b, 7c). Terpineol has a significantly higher boiling point (215°) than the THF/Toluene mixture and thus has a longer solvent evaporation time (Fig. 7c). To avoid carbonate layer growth, it is advantageous to have a fast solvent evaporation process.

Zoom In Zoom Out Reset image size