Quantitative analysis of GITT measurements of Li-S batteries James Dibden, Nina Meddings, Nuria Garcia-Araez, and John R. Owen Acknowledgements to Oxis and EPSRC for EP/M5066X/1 - CASE studentship, EP/P019099/1- MESS project co-funded by Innovate UK. 1
-voltage -current The Galvanostatic Intermittent Titration Technique - as applied to Li-ION Li e Li Host Constant current pulse + relaxation step time Li + c Li e - Cathode. thickness L depth 0 l L E PULSE IR u constant potential after relaxation IR u time E RELAX
- voltage - current The Galvanostatic Intermittent Titration Technique - as applied to Li-S Constant current pulse + relaxation step time Li + c(lis n ) S e - S n - S ne S Li + depth n 0 L O H A R A Li E PULSE IR u constant potential after relaxation IR u time E RELAX 3
GITT in Li-ion battery materials 4 Weppner and Huggins. J. Electrochem. Soc, 1977, 14, 1569-1578. Chemical diffusion coefficient Equilibrium voltage profile 4 PULSE RELAX E E L D =
Thermodynamically enhanced diffusion Enhancement factor: dln a y F Li de d ln c RT d Li Enhancement of the chemical diffusion coefficient Weppner and Huggins. J. Electrochem. Soc, 1977, 14, 1569-1578. 5
GITT in Li-S cells complications Redox reactions in the liquid state Multiple polysulfide species (not fully identified) Polysulfide shuttle can cause self discharge GITT in Li-S cells our approach Polysulfide shuttling avoided using a lithium selective membrane (Ohara) Equations derived for (complicated) solution redox reactions 6
Plan Validate our approach with a model redox system Apply it to cells containing dissolved sulfur Apply it to cells containing dissolved polysulfides Apply it to Li-S cells Aim Use GITT in Li-S cells to obtain quantitative information of: Mass transport rate (diffusion coefficient) Reaction rate (relaxation rate) Composition-dependent activity coefficients 7
Validation Chemical diffusion coefficient determined by: Cyclic voltammetry Square-wave voltammetry Chronopotentiometry GITT Model redox system: EtV + = 8
Cell design 9
Results: Summary 0.01 mm EtV + in 0.1 M LiTFSI in Pyr 14 TFSI Glassy carbon working electrode Method D / cm s -1 Cyclic voltammetry 6.3 x 10-8 Chronopotentiometry 7.7 x 10-8 Square wave voltammetry 7.5 x 10-8 10
Cyclic voltammetry mm EtV + in 0.1 M LiTFSI in Pyr 14 TFSI System is electrochemically reversible (~ 63 mv) D EtV+ = 6.3 x 10-8 cm s -1 11
Chronopotentiometry mm EtV + in 0.1 M LiTFSI in Pyr 14 TFSI Transition time: D EtV+ = 7.7 x 10-8 cm s -1 1
Square wave voltammetry 0.01 mm EtV + in 0.1 M LiTFSI in Pyr 14 TFSI mm D EtV+ = 7.5 x 10-8 cm s -1 13
GITT mm EtV + in 0.1 M LiTFSI in Pyr 14 TFSI D = 4 L E E RELAX PULSE c surface 30% (0.6 mm) c 0.00mM E bulk RELAX 0 mv Next: repeat the experiments with an smaller and thinner separator to decrease the electrolyte volume and thus increase c bulk. 14
Cell design 15
Results: Summary 5 mm EtV + in 1 M LiTFSI in DOL Method D / cm s -1 Cyclic voltammetry. x 10-6 Chronopotentiometry.4 x 10-6 GITT x 10-6 16
Cyclic voltammetry 5 mm EtV + in 1 M LiTFSI in DOL System is electrochemically reversible (~ 60 mv) D EtV+ =. x 10-6 cm s -1 17
Chronopotentiometry 5 mm EtV + in 1 M LiTFSI in DOL Glassy carbon C-coated Al foil Transition time: D EtV+ =.4 x 10-6 cm s -1 18
GITT (1) 19 5 mm EtV + in 1 M LiTFSI in DOL 4 PULSE RELAX E E L D = Unrealistic variation of the chemical diffusion coefficient
GITT () 5 mm EtV + in 1 M LiTFSI in DOL EtV + + e - EtV + D = 4 L E E RELAX PULSE E E 0 RT F c ln EtV c EtV Equilibrium voltage profile in agreement with Nernst equation Evolution of voltage change induced by pulses is unexpected 0
GITT analysis Fick s first law: I c nfd 0 A xx0 EtV e EtV Evolution of surface concentrations with time: I t c c c AnF D x0 initial 0 initial EtV EtV EtV c I t c x AnF D x0 0 EtV 0 Assumption: E PULSE is proportional to c(x=0) E kc( x 0) PULSE E kc and the proportionality constant is RELAX bulk Since c bulk I / 0 nf AL It is concluded that: L t EPULSE E D RELAX And for t=: D = 4 L E E RELAX PULSE 1
GITT analysis Fick s first law: I c nfd 0 A xx0 EtV e EtV Evolution of surface concentrations with time: c c I x0 initial 0 EtV EtV AnF t D c I x0 0 EtV AnF t D Assumption: E PULSE is calculated with the Nernst equation E PULSE E 0 RT nf ln c c x0 EtV x0 EtV and E 0 is obtained from the equilibrium voltage profile (.44V) taking into account: c bulk I / 0 nf AL
GITT (3) 5 mm EtV + in 1 M LiTFSI in DOL Pulse 1 Pulse 5 Pulse 15 The evolution of the voltage change induced by the pulse is in agreement with the Nernst equation 3
Conclusions Theoretical framework to analyze GITT results of Li-S cells Validation of the evaluation of the diffusion coefficient by: Cyclic voltammetry Square wave voltammetry Chronopotentiometry GITT Next: GITT as diagnostic tool of Li-S cells 4