Composition and Structure of the solid electrolyte interphase on Na-Ion Anodes Revealed by Exo- and Endogenous Dynamic Nuclear Polarization─NMR Spectroscopy

Abstract

Sodium ion batteries (SIB) are among the most promising devices for large scale energy storage. Their stable and long-term performance depends on the formation of the solid electrolyte interphase (SEI), a nanosized, heterogeneous and disordered layer, formed due to degradation of the electrolyte at the anode surface. The chemical and structural properties of the SEI control the charge transfer process at the electrode-electrolyte interface, thus, there is great interest in determining these properties for understanding, and ultimately controlling, SEI functionality. However, the study of the SEI is notoriously challenging due to its heterogeneous nature and minute quantity. In this work, we present a powerful approach for probing the SEI based on solid state NMR spectroscopy with increased sensitivity from dynamic nuclear polarization (DNP). Utilizing exogenous (organic radicals) and endogenous (paramagnetic metal ion dopants) DNP sources, we obtain not only a detailed compositional map of the SEI but also, for the first time for the native SEI, determine the spatial distribution of its constituent phases. Using this approach, we perform a thorough investigation of the SEI formed on Li₄Ti₅O₁₂ used as a SIB anode. We identify a compositional gradient, from organic phases at the electrolyte interface to inorganic phases toward the anode surface. We find that the use of fluoroethylene carbonate as an electrolyte additive leads to performance degradation which can be attributed to formation of a thicker SEI, rich in NaF and carbonates. We expect that this methodology can be extended to examine other titanate anodes and new electrolyte compositions, offering a unique tool for SEI investigations to enable the development of effective and long-lasting SIBs.

Scheme 1. Proposed Approach for Analyzing the Native SEI Formed on LTO Particles: Endogenous DNP with Mn(II) Metal Ions Enhances the Inner SEI Layer (Purple) whereas Exogenous DNP with TEKPol Biradicals Enhances the Outer SEI Surface (Green)

The different phases making the SEI are represented by the different colored shapes on the LTO surface.

Figure 1

Charge–discharge curves of LTO vs Na metal cells prepared with (a) conventional film electrodes (LTO, carbon black and binder) and (b) pressed powder (pp) LTO cycled with 0% FEC, and (f) ppLTO cycled with 2% FEC. All batteries were cycled at C/10. (c) Discharge capacities as a function of cycle number tested at variable rates for film and pressed powder electrodes cycled with 0 and 2% FEC. (d) SEM image of ppLTO after cycling taken with 5 kV accelerating voltage. (e) TEM-EDS image of ppLTO taken after 20 cycles.

Figure 2

(a) ²³Na MAS NMR spectrum collected on LTO in the discharged state after 8 cycles with a recycle delay of 2.5 and 256 scans, along with ¹⁹F–²³Na and ¹H–²³Na CP experiments obtained with a recycle delay, number of scans and contact time of 5.2 s, 3020, 0.25 ms and 2.5 s, 20,480, 1 ms, respectively. (b) ²³Na MAS NMR spectrum of LTO after 1 cycle with 0 and 2% FEC using recycle delay and number of scans 4, 1024 and 7.5 s, 1024 for the 0 and 2% FEC samples, respectively. (c) ¹⁹F spectrum of LTO after 1 cycle with 0 and 2% FEC acquired using rotor synchronized Hahn echo with recycle delay of 225 s, 48 scans and 93 s, 128 scans, respectively. (d) ¹⁹F{⁷Li} REDOR dephasing curve for LTO after 1 cycle with 2% FEC. The REDOR experiment was performed with a recycle delay of 20 s and 48 scans. The inset shows the deconvolution of the spectrum. (e) ¹⁹F RFDR 2D experiment acquired with 10 ms mixing time, 180 increments, a recycle delay of 19 and 16 scans of LTO after 1 cycle with 2% FEC. (f) ²³Na{¹⁹F} REDOR experiment for LTO after 3 cycles with 2% FEC. Recycle delay and number of scans were 5 s, 1664 scans, respectively. All experiments were performed at room temperature and sample spinning at 25 kHz.

Figure 3

(a) ¹H–¹³C CP spectra of ppLTO after 1 cycle with 0 and 2% FEC enhanced by exogenous DNP. The spectra were taken using recycle delay, number of scan and contact time of 6 s, 1024, 0.5 ms and 6 s, 984, 2 ms, respectively. 2D heteronuclear correlation (HETCOR) of LTO cycled once with 0 and 2% using CP transfer for correlating (b) ¹H–²³Na using 5.5 s recycle delay, 4 scans, 2 ms contact time and (c) ¹H–¹³C with recycle delay of 5.5 s and 48 scans, contact time of 0.5 and 2 ms for 0 and 2% FEC, respectively. All spectra were acquired with MAS of 9 kHz.

Figure 4

(a) ²³Na{¹⁹F} REDOR spectra of an LTO anode after 1 cycle, acquired using recycle delay of 20 s, 128 scans and 1 ms recoupling time with (S) and without (S₀) pulses on ¹⁹F. (b) ²³Na MAS NMR spectra of LTO with 0% FEC after one sodiation and desodiation. The fully desodiated spectrum is the same as in Figure 2b, whereas the full discharge sample was obtained using a relaxation delay of 0.375 s and 5120 scans. (c) Quantification of the ⁷Li and ²³Na NMR signal of LTO anodes cycled at C/10 and extracted from the cell at different SoC, specified by hours (see Supporting Information for NMR spectra). All experiments were performed with 25 kHz MAS.

Figure 5

MW on/off spectra of indirect ¹H–²³Na CP enhanced by exogenous DNP for LTO after one cycle with (a) 0% FEC and (c) 2% FEC. (a) and (c) were obtained with 6.35 s polarization time, 8 scans and 2 ms contact time. (b,d) are ²³Na direct excitation exogenous DNP comparing MW on/off spectra of LTO after one cycle with 0 and 2% FEC, respectively. (b) Was acquired using a recycle delay 600 s and 8 scans and (d) using a recycle delay 60 s and 16 scans. All experiments were performed at 100 K and MAS of 9 kHz.

Figure 6

(a) CW-EPR spectra of LTO doped with different Mn(II) concentrations measured on a Q-band spectrometer at 298 K. (b) Q-band CW-EPR spectra of Mn(II) doped LTO before and after cycling with 0 and 2% FEC taken at 298 K. Inset: X-band CW-EPR spectra of Mn(II) doped LTO before and after cycling with 0 and 2% FEC acquired at 100 K. (c) Endogenous DNP-NMR enhancement factors of ⁶Li and ²³Na of LTO doped with varying amount of Mn(II).

Figure 7

Endogenous DNP-NMR ²³Na MW on/off spectra for 40 mM Mn(II) doped LTO with (a) 0 and (b) 2% FEC. Experiments were performed with a recycle delay of 1200 and 240 s, respectively, and 4 scans. Inset shows the deconvolution of the MW off spectrum. (c) ²³Na{⁷Li} REDOR dephasing curve acquired for LTO after 1 cycle with 0% FEC and 20 mM Mn(II) doping. The REDOR experiment was enabled by endogenous DNP with a recycle delay of 140 s and 16 scans. All data acquired at 100 K and MAS of 9 kHz.

Figure 8

Calculated ratio of ²³Na polarization (P_n) and electron polarization (P_e) as a function of the dopant-nucleus distance (r) and the intrinsic nuclear relaxation time (T₁). The dashed line indicates the distance at which the intrinsic T₁ is equal to the relaxation caused by the paramagnetic ions (PRE). For this calculation an electron relaxation time of 1 μs was used.

Scheme 2. Structural Model of Native SEI Formed in LTO During Cycling Based on Data Acquired with Different Polarization Sources and Correlation Experiments

The outer SEI, detected by exogenous DNP, is predominantly organic compounds (green) whereas the inner SEI, detected with endogenous DNP, is more inorganic (dark blue). NaF is the main inorganic species estimated to form a thin layer at the inner SEI. Na_xLTO formed due to Li–Na exchange at the outer layers of the particles is shown in dark purple.