Ion conduction in crystalline polymer electrolytes for lithium ion batteries
Open Access
- Author:
- Chithur Viswanathan, Shankar Ram
- Graduate Program:
- Chemical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- July 22, 2021
- Committee Members:
- Xueyi Zhang, Major Field Member
Scott Milner, Major Field Member
Janna Maranas, Chair & Dissertation Advisor
Ralph Colby, Outside Unit & Field Member
Seong Han Kim, Program Head/Chair
Michael Janik, Major Field Member - Keywords:
- polymer electrolytes
lithium ion conduction
density functional theory
crystalline polymer electrolytes
lithium ion batteries
nanowhiskers
polymer dynamics
X-Ray diffraction - Abstract:
- Polyethylene oxide (PEO) based Solid Polymer Electrolytes (SPEs) are safe and efficient alternatives to liquid/gel-based electrolytes. In addition to improving safety and design flexibility, SPEs could allow the use of lithium metal anode which can theoretically improve energy density 10-folds than commercially used lithium graphite anode. However, SPEs suffer from low Li+ ion conductivity. In most SPEs, the conductivity is linked to PEO segmental motion. Attempts to increase polymer dynamics reduce the mechanical strength of SPEs. Thus, it is necessary to decouple conductivity from the mechanical strength of the polymer. Conduction through the crystalline domain was never considered possible until the discovery of a PEO/salt co-crystal [PEO6], which was found to be more conductive than its amorphous counterpart. In PEO6 [the crystal structure co-crystallizes 6 PEO ether oxygens to one Li-anion pair], two PEO chains fold around Li+ in a non-helical fashion forming an approximate cylindrical “tunnel” with lithium atoms distributed along the cylinder central axis. Each lithium atom coordinates five ether oxygens; the anions are outside the tunnel and there is no direct bonding between the anions and the Li+. Due to its unique tunnel-like structure, PEO6 conducts Li+ based on a mechanism that decouples the conductivity and segmental motion of the polymer. However, these polymer electrolytes have not been used for battery applications because these studies used low molecular weight PEO (1000 g/mol) to achieve high crystallinity of the PEO6 phase. At this low molecular weight, the polymer does not confer the high modulus required. In SPEs with high molecular weight PEO, conduction through PEO6 is unfavorable as the tunnels fold to form lamellar structures and increase the conduction pathway. In this study, we explore conduction in high molecular weight [600000 g/mol] crystalline polymer electrolytes at EO: Li = 6:1. At this molecular weight, although four lithium salts [LiPF6, LiAsF6, LiSbF6, and LiClO4] can form PEO6, we focus our study on PEO6-LiClO4, which displays the highest conductivity than with other salt complexes. But this conductivity of PEO6LiClO4 drops by an order of magnitude after 2 months of thermal annealing. This is also accompanied by the changes in the XRD pattern which is uncharacteristic of any phases of PEO-LiClO4. Thus to explain this change, we explore the possibility of defects like vacancy, extra salt, and interstitial lithium in PEO6. In addition to being enthalpically stable, these defects also display the peculiar peaks of long-time annealed samples. While the change in the XRD pattern of a long-time annealed PEO6 can be explained by a combination of defects, based on their relative stability, it is more likely to be due to the “trapped” PEO6 structure. In this structure, in contrast to PEO6 where lithium atoms are at the center of the tunnel, one of the Li+ is “trapped” in the periphery of the tunnel coordinating with four ether oxygens and one anion, distorting the PEO6 tunnel. Because this crystal transformation is detrimental to conduction in PEO6, we use a percolated network of high aspect ratio fillers (cellulose nanowhiskers) to stabilize PEO6 tunnels over long distances. The patterned arrangement of the –OH surface group, which has a Lewis acidic character allows it to interact with either the anions or ether oxygen on the PEO chain. In addition, the distance between primary alcohol groups on the cellulose surface along the axial direction closely matches with the lattice parameter of PEO6 along the tunnel direction. This results in a low energy penalty [0.08 eV] for constraining PEO6 on the surface of the whisker, making it a suitable nucleation agent for PEO6LiClO4. Although these patterned cellulose nanowhiskers do stabilize PEO6 tunnels resulting in no change in XRD pattern even after a year of annealing, the room temperature conductivity (6 x10-6 S/cm) is still below the target value (10-3 S/cm). To improve the conductivity further, we draw inspiration from crystalline ceramic conductors, which have utilized doping [adding or substituting a small percentage of impurities to the host material] strategies to increase conductivity. By replacing 0.5-10% LiClO4 with NaClO4 in PEO6LiClO4, we demonstrate an order of magnitude increase in room temperature conductivity with the highest effect at 1% doping. This increase is not correlated with the glass transition temperature. Up to 1%, doping disrupts PEO6 crystallization. Above 1 %, diffraction peaks arise between 10-15o which cannot be due to PEO6 but resemble another polymer salt co-crystal, PEO3. A stable structure for PEO6 with NaClO4 is determined computationally, whereas only structures for PEO3 and PEO8 have been observed experimentally. Due to doping, larger sodium cations could either be accommodated into PEO6 or could end up not being part of the PEO6 lattice, resulting in a vacancy in PEO6. From DFT calculations, we determine that it is 0.92 eV more energetically favorable to swap sodium with lithium in PEO6 than to form PEO6 with vacancy. We conclude the increase in conductivity to be a consequence of weaker coordination of sodium to ether oxygen which increases the “bottleneck” size for conduction. In the sodium doping study, the presence of PEO3 peaks in XRD was correlated with increase an increase in conductivity. This is contrary to the popular belief that PEO3 is non-conductive. In contrast to PEO6, in PEO3, only one chain wraps around the Li+ in a helical fashion resulting in three-fold coordination of ether oxygen and two-fold coordination of anions. Due to tighter coordination of the lithium atoms with the neighboring anions, and lack of uncoordinated neighboring sites in PEO3, the activation energy for lithium hop as reported in PEO3LiCF3SO3 was found to be high [~1 eV]1-2, resulting in low ionic conductivity. If this gridlock is reduced by creating more vacancies, lithium atoms in PEO3 could become more mobile. To test this hypothesis, we create vacancies in PEO3 by reducing the concentration of LiClO4 from EO: Li = 3:1. We observe several orders of improvement in conductivity with a 20% reduction in salt concentration from EO: Li = 3:1, with no change in crystal structure or crystallinity up to 30% concentration deviation. Surprisingly, this change in conduction is accompanied by an increase in activation energy, indicating a change in the mechanism of conduction. To explain this change, we use DFT to find the activation energy for several conduction pathways in PEO3 including lithium diffusion: along the strand, across the strand, into a vacancy, and anion diffusion. In contrast to the previously held view that PEO3 has activation energy [~1eV], we conclude that the activation energy of PEO3LiClO4 can vary from 0.42- 1.3 eV depending on the conduction pathway. Thus, using both experiments and simulations we demonstrate the potential of crystalline polymer electrolytes and develop tools to understand the conduction mechanism in them. Although we did not reach the target conductivity, the findings from this work are important to design fast conduction solid polymer electrolytes.