Sulfur Speciation in Li-S Batteries Determined by Operando X-ray

Transcription

1 Supporting Information for: Sulfur Speciation in Li-S Batteries Determined by Operando X-ray Absorption Spectroscopy Marine Cuisinier, Pierre-Etienne Cabelguen, Scott Evers, Guang He, Mason Kolbeck, Arnd Garsuch, Trudy Bolin, Mahalingam Balasubramanian, and Linda F. Nazar* Department of Chemistry, University of Waterloo, 200 University Ave W, Waterloo, Ontario N2L 3G1, Canada BASF SE, Ludwigshafen, Germany X-ray Science Division, Argonne National Laboratory, Argonne, IL-60439, USA Synthesis of porous hollow carbon nanospheres In a typical synthesis, 11.4 ml ammonium hydroxide was added to a solution of ethanol (244 g) and deionized water (40 g) in a 1 L round bottom flask with stirring at 30 C. After stirring for 30 minutes, 11.2 ml tetraethyl orthosilicate (TEOS) was added under vigorous stirring. In a separate vial, 1.6 g resorcinol, 2.24 ml formalin and 240 μl poly(diallyldimethylammonium chloride) (7 wt% in deionized water) were mixed and added to the previous solution after 10 minutes. The solution was stirred at 30 C overnight. The solution was then transferred to a hydrothermal autoclave and maintained at 100 C overnight under static conditions. The resultant brown precipitate was centrifuged and washed 3 times each with deionized water, ethanol and dried at 70 C overnight. The dry brown polymer was carbonized under flowing argon gas (ramp 3.75 C/min to 750 C and hold 60 minutes). The resultant carbon coated silica was treated in 15% hydrofluoric acid to etch and remove the inner silica core. The PCNS s were impregnated with 68 wt% sulfur by a melt-diffusion method. Typically, 30 mg PCS and 70 mg sulfur were ground together in a mortar and pestle and were then pressed in a pellet die and placed in an oven at 155 C overnight. In the

2 process some sulfur evolves, resulting in the slightly lower sulfur composition in the final PCNS/S-68% composite. Electrochemical studies Positive electrodes were constructed from 80% PCNS/S-68%, 10% Kynar Powerflex binder and 10% Super S carbon. The positive electrode material, ready for electrochemical studies, contained 54 wt% of sulfur as the active mass. The positive electrode material was dispersed in dimethylformamide by sonication and slurry-cast onto a carbon-coated aluminum current collector (Exopack Advanced Coatings). Coin cells (2325 type) were constructed using 2 separator layers of Celgard 3501 soaked with 50 µl of an electrolyte composed of 1.0 M LiTFSI (lithium bis(trifluoromethanesulfonyl) imide) in a mixed solvent of DOL (1,3-dioxolane) and DME (1,2-dimethoxyethane) (1:1 volume ratio) with 2 wt% LiNO 3 as additive. Lithium metal foil was used as the negative electrode. The batteries were cycled between 1.5 and 3 V using an Arbin battery cycler at room temperature. The discharge/charge rate 1C (1672 ma g -1 sulfur) corresponds to a current density of 1.25 ma cm -2. Synchtrotron studies The experiments were carried out at the sector 9-BM-B in the Argonne National Laboratory using a Si(111) crystal monochromator. The instrumental resolution in the energy range near the sulfur K-edge is about 0.35 ev, with a beam size of about 450 x 450 microns. All of the XANES studies were carried out under constant helium flow in the sample chamber and the data were collected in fluorescence mode using a 4-element vortex detector. Energy calibration was carried out using sodium thiosulfate pentahydrate, Na 2 S 2 O 3.5H 2 O with the pre-edge feature at ev. The cell used to perform operando XANES was adapted from the 2325 coin cells using an aluminized Kapton (Sheldahl) window (see Figure S4 and associated discussion). XANES spectra were acquired continuously every 1110 seconds so that the composition change between two spectra is about 50 mah.g -1 sulfur (i.e. 3% of the total capacity) at a cycling rate of C/10. Reference materials Li 2 S and Li 2 S 6 were synthesized using the appropriate ratio of element sulfur and LiEt 3 BH in THF. 1 The solvent was then removed under vacuum. Na 2 S 2 and Na 2 S 4 were 1 Gladysz et al. J. Chem. Soc., Chem. Commun. 1978, 19,

3 prepared in absolute ethanol. 2 Structural characterization by XRD is reported in Figure S3, showing that pure phases were obtained except for a small fraction of Na 2 S 5H 2 O in the case of Na 2 S 2. For each, a laminate made of the reference compound, Super P carbon and PVDF (10wt% sulfur overall) was sealed in a Kapton pouch for ex situ measurements. The reference data were recorded under the same conditions as the in situ measurements. Sulfur speciation Athena software was used to normalize the spectra and correct them from selfabsorption inherent to fluorescence mode by using the atomic composition, in agreement with TEY measurements. 3 The same software was used to perform Gaussian fitting analysis and linear combination fitting (LCF). Figure S1. Determination of sulfur content in the PCNS/S composite. TGA plot showing a weight loss of 68 wt% sulfur. 2 Rosen et al. Acta Chem. Scandinavia 1971, 25, [3] B. Ravel, M. J. Newville, J. Synchrotron Rad. 2005, 12,

4 Figure S2. a Long-term cycling at 1C rate with the specific discharge capacity in mah.g -1 reported on the left y-axis and the coulombic efficiency on the right. The capacity stabilization after few cycles at a higher value than the first discharge capacity indicates that the active material requires some formation cycles to be fully accessible, certainly due to its confined design. b Specific discharge capacity over 45 cycles at various rates: C/2, 1C, 2C, demonstrating the excellent power capability of the PCNS/S material. Figure S3. Structural characterization of solid-state lithium and sodium polysulfides prepared as reference materials. a XRD patterns of lithium (poly)sulfides prepared by reduction of sulfur using LiEt 3 BH in THF. In each pattern, TiO 2 anatase was added as an external standard to quantify the amorphous content. The Li 2 S 2 sample (red pattern) contains

5 significant nanoscrystalline Li 2 S so it was not used. Li 2 S 6 (blue) is purely X-ray amorphous while Li 2 S 8 (green) exhibits weak diffraction peaks of α-s 8. b XRD patterns of sodium (poly)sulfides prepared by reacting sulfur and sodium sulfide in ethanol. The Na 2 S 4 sample exhibits the diffraction peaks of the pure phase, while the Na 2 S 2 sample contains a Na 2 S 5H 2 O impurity, accounting for 35 % of the total sulfur. Its sulfur K-edge XANES spectrum is given in Figure 3 to show the change in the low energy feature intensity, but this sample was not used in the data analysis.

6 Figure S4. Schematic of the operando cell. The design was adapted from a 2325 coin cell, using an aluminized Kapton TM window to allow X-ray beam penetration. Other modifications were limited to the electrode design and the nature of the lithium salt: PCNS/S-58wt% was mixed with Super P and PVDF to reach 10wt% sulfur overall and slurry-cast onto a carbon paper (AvCarb P50, Ballard Material Products). LiTFSI was substituted by LiClO 4 (lithium perchlorate) to avoid any contribution to sulfur absorbance spectra and the operando cells were assembled so that the cathode material faces the Kapton window. Owing to the porous carbon current collector, the density of sulfur in the 80 µm-thick electrode is only 50 mg.cm -3 so that one absorption length at the S K-edge is increased from 1.8 µm (pure sulfur) to 44 µm in the carbon-rich composite electrode.

7 Figure S5. Sulfur K-edge XANES correction from self-absorption. K-edge spectra of a pristine sulfur cathode measured in the electrochemical cell in the fluorescence yield (FY) before (black) and after (red) self-absorption correction (SA) compared to ex situ measurement in total electron yield (TEY, blue). The difference between the SA-corrected FY and TEY spectra is plotted in grey. The 18% attenuation observed here is in good agreement with the 17% absorbance reduction calculated by others for 200 nm spheres of α- S 8 (ref 18). A feature at 2480 ev, indicative of sulfate species, appears only in the in situ data measured by FY. Since TEY is much more surface sensitive than FY, this suggests that exposure to LiNO 3 -containing electrolyte may create a small amount of sulfate, as reported to form on the negative electrode by Aurbach et al. [4] LiNO 3 is commonly used to improve coulombic efficiency by forming a passivating film on the lithium metal which helps inhibit reduction of diffusion-transported polysulfides at the negative electrode. The role of LiNO 3 in sulfur oxidation could be confirmed by comparison with in situ spectra measured in the absence of the additive, ruling out air exposure as the source of sulfates at the cathode surface. As observed in Figure 3a, this sulfate is electrochemically inactive as it does not evolve during the duration of the experiment. Figure 4b shows the electrochemical signature of LiNO 3 reduction (plateau at 1.85 V), after the completion of sulfur reduction. [2] Aurbach, D. et al. On the surface chemical aspects of very high energy density, rechargeable Li sulfur batteries. J. Electrochem. Soc. 156, A694-A702 (2009).

8 Figure S6. Alternative linear combination fits (LCF) of the operando XANES spectra. a) χ² values of the 4-component LCF shown in Figure 4 as a function of the scan number. b) alternative LCF are suggested, for which the χ² are compared to the lowest (i.e. best) values given above for the 4 component fit, here normalized to 1. The 5-component fit (green) gives similarly good agreement with the experimental data. However, the additional Na 2 S 2 component is found to be zero for a large majority of spectra. Among the 3-component fits, the one using only Na 2 S 4 to account for the low energy feature (red) gives the best results, but χ² values remain 20-50% higher than in the presence of Li 2 S 6 in the 4-component fit. In conclusion, the 4 component fit is the best, justified on the basis of minimum χ².

9 Figure S7. Composition of the sulfur-based cathode upon cycling based on the XANES and on the electrochemistry. Lithium content (x in Li x S) from XANES is calculated from the LCF and reported as a function of x calculated by integrating the current measured by electrochemistry. The deviation from the diagonal line represents the difference between the structural evolution probed by XANES and the capacity recorded on each (dis)charge step. Previous operando studies report a delayed structural response of the system compared to the electrochemistry due to inhomogeneous cell stack pressure, so we include a side discussion here. During a biphasic reaction, there is no driving force for all particles to be in absolute equilibrium. The electrode area in contact with cell body might undergo much of the reaction initially in D-III. As there is no driving force for local equilibration, the area that is probed (window region) might not have undergone any change. As the reaction proceeds deep into D-III the reaction front moves into the window region, as the regions in contact with the cell parts have already undergone the maximum reaction possible. However, the processes in charge (C-I to C-III) do not show a corresponding anomaly. The perfect agreement between the two calculations along the charge strongly indicates that the window region under the beam (although it corresponds to only 10-3 of the surface) is representative of the state of the cell. The use of a slow cycling rate (C/10) for

10 such a high-rate cathode material (Figure S2), combined with a carbon paper backing as both an electrolyte diffusion layer and a current collector would alleviate inhomogeneities due to poor current collection usually faced in the window region. Thus we do not believe that the evolution delay in region D-III is the result of an artifact associated with cell stack pressure. The discrepancy observed upon discharge is thus ascribed to a structural delay in the lithiation process on the second plateau region, as discussed in the main article. Figure S8. Linear combination fit (LCF) of the XANES spectra recorded during the first discharge at C/5. The evolution of each component weight (thick line + symbol) is very similar to the second discharge at C/10 shown in Figure 4 (thin line).