UC Berkeley

Next-generation lithium batteries with high energy densities are desired for powering the future fleet of electric vehicles. The implementation of these batteries hinges upon the development of novel electrolyte materials that exhibit stability against the lithium metal anode in addition to favorable transport properties. Conventional liquid electrolytes, such as the carbonate-based solvents that are standard in lithium-ion batteries, have high ionic conductivities but lack compatibility with lithium metal. In these systems, brittle, unstable interface layers form on the lithium surface leading to uneven plating and rapid cell failure. Polymer electrolytes offer a promising alternative to conventional liquid electrolytes, as they form stable interfaces with lithium metal and exhibit solid-like material properties. However, despite four decades of persistent research, the transport properties of the most promising solvent-free polymer electrolytes remain insufficient for use in commercial batteries. Our ability to design new polymers with improved electrolyte properties is compromised by a lack of understanding of ion transport in these materials. In this work, we employ a wide variety of electrochemical characterization techniques, supplemented with theory and simulations, to identify the factors that govern ion transport in polymer electrolytes.

The performance of battery electrolytes depends on three independent transport properties: ionic conductivity, diffusion coefficient, and transference number. We perform a complete characterization of all three transport properties in mixtures of 5 kg/mol polyethylene oxide (PEO) and lithium bis(trifluoromethanesulfonyl) imide (LiTFSI) salt over a wide range of salt concentrations. Three different approaches were used to measure the transference number: the steady-state current measurement, pulsed-field gradient NMR, and a new approach proposed by Balsara and Newman. The latter approach is rigorous and based on concentrated solution theory, while the other two approaches only yield the true transference number in ideal solutions. The values obtained from the steady-state current method and pulsed-field gradient NMR are positive at all concentrations. In contrast, the transference number obtained by the approach of Balsara and Newman exhibits a complex dependence on the addition of salt, with negative values obtained at intermediate salt concentrations. Negative transference numbers suggest that ion transport is dominated by highly-mobile ionic clusters. These ion-ion interactions are neglected in the approaches that are derived using dilute solution theory.

There are a variety of techniques that can, in theory, be used to measure the transference number of concentrated electrolytes. We perform a comparison of two different electrochemical approaches: the method proposed by Balsara and Newman and a more well-established technique by Ma and coworkers. Both approaches are experimentally intensive and rely on concentrated solution theory. In high molecular weight PEO electrolytes, the data from the two techniques are in perfect agreement. In contrast, in low molecular weight PEO there is a disagreement between the two approaches, which is attributed to the presence of a complex interface layer on the surface of the lithium electrodes. The parameters measured in technique of Ma and coworkers are thought to be inherently sensitive to the nature of the electrode-electrolyte interface, which may not be representative of the bulk electrolyte. For this reason, the Balsara and Newman approach is taken as the more robust measure of transference number.

Complete characterization of ion transport in an electrolyte enables full cell modeling. In this work, the theory is presented for predicting the cycling characteristics of a lithium-polymer-lithium cell containing an electrolyte with known transport properties. Using the ionic conductivity, diffusion coefficient, and transference number of PEO/LiTFSI electrolytes as inputs to our model, we calculate salt concentration and potential profiles in the electrolyte under a constant dc polarization. At steady-state, these profiles are nonlinear due to the strong concentration dependence of the transport properties of the electrolyte. Predictions of the limiting current in PEO/LiTFSI electrolytes were obtained using the model. Experimentally-measured cycling data from a series of symmetric cells with different salt concentrations were used to test the validity of the model. The time-dependence and steady-state value of the potential measured during cycling experiments were in excellent agreement with model predictions, requiring no adjustable parameters or simplifying assumptions.

Our modeling work supports the notion that the transport properties of PEO/LiTFSI electrolytes are insufficient for immediate commercialization. Thus, there is a great deal of interest in developing next-generation polymer electrolytes with improved transport properties. Designing new polymers is hindered by the complex relationship between the transport of ions in the polymer and the structure of the monomer. For example, the ionic conductivity of a polymer electrolyte depends on a variety of interconnected factors: interactions between the polymer chains and the salt, extent of dissociation of the salt, and polymer dynamics in the vicinity of the ions. All of these are affected by the monomer structure. In this work, we attempt to unravel these factors through systematic analysis of electrolytes comprised of newly-designed polymers and LiTFSI salt. In all cases, PEO/LiTFSI electrolytes are used as a baseline for comparison.

A set of aliphatic polyesters with systematic variations to the monomer structure were characterized using ionic conductivity and glass transition temperature measurements over a wide range of salt concentrations. A novel analysis approach was introduced to factor out the effect of segmental motion on conductivity; the parameter calculated in this analysis is referred to as the reduced conductivity. The dependence of the reduced conductivity on salt concentration helps to clarify the relationship between monomer structure and ionic

conductivity, and highlights differences between PEO and the polyesters. This study also demonstrates that polymers, such as polyesters, which are comprised of multiple polar groups are not an ideal choice for fundamental studies due to the complexity of solvation and ion transport in these systems.

Linear polyethers (CxEOy) were synthesized as a systematic set wherein aliphatic linkers were added to a PEO backbone. The carbon linkers change the glass transition temperature and dilute the polar groups relative to PEO; both factors influence ionic conductivity. The analysis introduced in the previous study was used to factor out the effect of glass transition temperature on conductivity; the results show a clear dependence of the reduced conductivity on the mole fraction of oxygen of the polymer. MD simulations were used to study the solvation site around Li+, which were found to be similar in all polymers. A comparison of experimental measurements and simulation results highlights the importance of solvation-site connectivity, a parameter which is thought to affect the hopping rate of Li+. A polymer with a higher solvation-site connectivity than PEO is predicted to exhibit superior transport properties.

A newly-synthesized polymer, P(2EO-MO), was characterized using a variety of electrochemical and NMR experiments in addition to MD simulations. The maximum conductivity of P(2EO-MO) is comparable to that of PEO, but the glass transition temperature exhibits a more precipitous increase with the addition of salt. The transference number measured using the steady-state current method and NMR are about a factor of two higher in P(2EO-MO) compared to PEO. In lieu of complete electrolyte characterization, the product σt+,SS is identified as the most important transport characteristic for comparison of electrolytes. This product is higher in P(2EO-MO), thus it is predicted to be a more efficacious electrolyte than PEO for battery applications. MD simulations reveal that the promising transport properties of P(2EO-MO) are likely attributed to the high solvation-site density which facilitates the transport of Li+ in this material.