Materials
Table of Contents
The following materials were used: sulfur powder (Sigma-Aldrich, >99.98%), iodine powder (Sigma-Aldrich, >99.99%), vapour grown carbon fibre (VGCF; Sigma-Aldrich, >99.99%, length: approximately 2–10 μm, diameter: about 200 nm), LPS (NEI, ion conductivity (25 °C): approximately 1 × 10−4 S cm−1, size: about 3–5 µm), Li6PS5Cl (LPSCl; NEI, ion conductivity (25 °C): 9.3 × 10−4 S cm−1, size: about 3–5 µm), SP (MTI, about 40 nm), Li foils (China Energy Lithium, >99.95%, about 250 µm and about 50 µm), Li powder (China Energy Lithium, >97%, 25–60 μm), THF (anhydrous, Sigma-Aldrich, >99.99%) and Li2S powder (Sigma-Aldrich, >99.98%).
Preparation of sulfur iodide materials
Sulfur iodide materials with different sulfur to iodine ratios were prepared by a two-step procedure. Typically, 1.0 g of S/I mixtures with different ratios were first prepared by grinding stoichiometric amounts of sulfur powder and iodine powder using a mortar and pestle in the air for 10 min. The sulfur iodide materials were then obtained by melting the preprepared S/I mixtures in a small vial (20 ml) at 80 °C for 3 h with a heating rate of 1 °C min−1 followed by a natural cooling procedure. The density of liquid S9.3I at 80 °C and 100 °C was measured to be approximately 2.35 and approximately 2.32 g cm−3, respectively. The bulk modulus of S9.3I measured by compression test was 0.29 GPa at 25 °C, which is much smaller than 7.7 GPa of α-S (ref. 35). The viscosity of liquid S9.3I increases from approximately 81 to approximately 196 mPa·s when the temperature increases from 80 °C to 100 °C. This behaviour is associated with the increase of the mass fraction of polymeric S chains which can cause more entanglement36.
Preparation of S9.3I–LPS–VGCF cathode composite
The as-synthesized sulfur iodide pellets were crushed into powders using mortar and pestle. A mixture composed of S9.3I powder, LPS and VGCF with a weight ratio of 4:4:2 was loaded into a 100 ml ball-milling jar inside a glovebox (O2 and moisture below 0.5 ppm) and milled at 500 rpm for 10 h. The weight ratio of ball-milling beads to the cathode composite was around 40:1. The S–LPS–VGCF cathode composite was prepared with the same procedure as the S9.3I–LPS–VGCF cathode.
Cell fabrication
Li/SP/LPSCl/SP/Li symmetric cells
First, 200 mg of LPSCl powder was loaded into a poly(ether-ether-ketone) (PEEK) die with a diameter of 13 mm and two titanium electrodes block on both sides. A pressure of approximately 30 MPa was applied to press the loose LPSCl powder into the die. After that, about 4–5 mg cm−2 of SP powder was loaded on both sides of the LPSCl pellet and a high pressure of 360 MPa was applied for densification. Finally, two pieces of Li foils (around 250 μm) were assembled into the symmetric cells with a pressure of approximately 30 MPa.
Li/SP/LPSCl/LPS/cathode full cells
First, 180 mg of LPS powder was loaded into a PEEK die with two titanium electrodes and pressed with a pressure of 30 MPa. Subsequently, 60 mg of LPSCl powder was pressed on one side of the prepressed LPS pellet at 30 MPa. Then, 4–5 mg cm−2 of SP powder and about 3 mg cm−2 of cathode powder were pressed on the LPSCl and LPS sides, respectively, at 30 MPa. A high pressure of 360 MPa was then used to press the SP/LPSCl/LPS/cathode composite together for 5–10 min. A piece of Li foil (around 50 μm) was then attached to the SP surface to complete the assembly of the solid-state full cells. A pressure of 30 MPa was maintained during cycling tests and periodical interface repair by heating. The full cells with high mass loading cathodes were assembled using the same procedure except with a mass loading of 10.5 mg cm−2 of the cathode composite. It should be noted that the viscosity of liquid S9.3I at 80 °C and 100 °C is approximately 81 and approximately 196 mPa·s, respectively, which greatly limits its flowability in a composite cathode. To further prevent any leakage at elevated temperatures during cycling, we designed a Swagelok cell in which the size difference between the titanium piston and the PEEK die cylinder was ≤0.1 mm. To prevent the liquid cathode from flowing into the LPS solid state electrolyte layer, we placed the cathode at the bottom and the anode on top for battery tests when running above the melting point of the cathode. Moreover, LPS becomes much denser than LPSCl after heating at 100 °C. As a result, the Li–S9.3I full cells were heat treated at this temperature before further testing. A dense LPS layer can help prevent any diffusion of the liquefied S9.3I cathode, although its high viscosity makes such diffusion unlikely.
Electrochemical characterization
Before the electrochemical performance tests, all of the symmetric cells and full cells were heated at 100 °C for 3 h with a heating rate of 1 °C min−1 followed by natural cooling procedures in the oven. After that, cells underwent galvanostatic cycling at different temperatures on a LAND battery test system after resting for 3 h. To measure the long-term cycling stability of the SP interlayer for protecting the lithium metal anode, an Li/Li symmetric cell was cycled at constant current density of 0.3 mA cm−2 and a capacity of 0.3 mAh cm−2. The EIS tests were performed on a BioLogic VMP300t electrochemical workstation within a frequency range from 7 MHz to 0.01 Hz, and the corresponding DRT analysis was performed using techniques as previously reported37.
XRD
Small-angle and wide-angle X-ray diffraction of different sulfur iodide materials were performed on a Rigaku Smartlab diffractometer with a Cu anode (K-alpha emission (Kα), wavelength λ = 1.5418 Å). The primary beam power was set to be 1.76 kW. The sample stage was aligned normal to the sample surface. In situ XRD analysis (30–150 °C) of S9.3I powder during heating was performed on a Bruker D8 Discover diffractometer, which was equipped with a rotating Cu anode (Kα λ = 1.5418 Å) and a Vantec 500 area detector. The XRD patterns were collected every 5 °C. Samples are sealed in a borosilicate capillary (0.8 mm diameter, approximately 0.01 mm thin-wall, Charles Supper) by using epoxy in an Ar-filled glovebox before being taken out for tests. The sample capillary with the Ti sample holder slit and an internal cartridge heater was placed on a custom-built programmable XYZ stage. For XRD measurements, 50 kV and 24 mA were applied. The XRD data were first processed with the Bruker-AXS GADDS software and further analysed by MDI JADE XRD software.
Electrical conductivity tests
To measure the electrical conductivity of sulfur iodide materials with different ratios, 180 mg of sample was pressed in a peek die (diameter: 13 mm) with two titanium electrodes at 360 MPa. A constant voltage of 0.2 V, 0.4 V or 0.6 V was applied for 1 h, respectively, to measure the stabilized current on a BioLogic VMP300t electrochemical workstation. The thickness of the sample was confirmed by SEM. On the basis of the voltage, current and the thickness of samples, the electrical conductivity can be calculated.
SEM
SEM (FEI Quanta 250 SEM) with EDX was used to determine the morphology and chemical composition of the S9.3I sample under 30 kV and 0.1 nA. The cross-sectional images of S9.3I/LPS and S/LPS mixtures were collected using cryo-focused ion beam-scanning electron microscopy (FEI Scios, Scios DualBeam) at 5 kV and 0.1 nA. The cross-sections were first cut with an ion beam of 30 nA, followed by a surface cleaning process at 0.1 nA. For the cross-sectional SEM analysis of cathode and anode interfaces in the full cells, the samples were prepared by an Argon-ion beam cross-section polisher (IB−19520CCP) developed by JEOL. An accelerating voltage of 4 kV and a milling rate of around 25 μm h−1 were used to minimize artifacts. During milling, the sample stage was placed perpendicular to the ion beam and swung automatically within ±30° to broaden the polished area and avoid beam strains. An air-free vessel was used to transfer samples between the glovebox and characterization tools. The cross-sectional SEM images of cathode and anode interfaces were conducted in Axia ChemiSEM.
Cryo-TEM
To prepare samples for the tests, a sample suspension in hexane was drop cast on a gold transmission electron microscopy (TEM) grid in a glovebox to prevent air and water exposure. The TEM grid was then loaded on a Gatan cryo-holder in a liquid N2 environment. The holder tip shutter was closed after loading the TEM grid on the holder and during the transfer to the microscope. The sample was kept at liquid N2 temperature in the TEM (FEI Talos F200x microscope) to reduce radiation damage, and characterized using scanning electron microscopy transmission, EDX (FEI Super-X EDX system) and EELS (Gatan Enfinium). The presence of Li in the sample was confirmed by EELS, which showed a peak at approximately 60 eV, an indication of energy loss at the Li K-edge. The presence of S and I elements in the sample was confirmed by EDX. The composite map of S, I was acquired using EDX spectra, and the map of Li was done by EELS spectra. It should be noted that the chemically lithiated S9.3I sample was still very sensitive to the electron beam even in the cryo-TEM. To reduce the beam damage, a low electron dose was applied for EDX and EELS during tests. In addition, a suitable particle size should be selected to balance the needs of reducing the radiation damage and of a thinner sample for EELS signal acquisition.
X-band EPR
X-band EPR measurements were conducted on a Bruker EMXplus EPR spectrometer. All samples were packed into Q-band tubes inside an Ar-filled glovebox and sealed with epoxy. The Q-band tubes were inserted into 4 mm quartz EPR tubes to hold the samples in place inside the resonator cavity. Variable temperature measurements were performed using an X-band Dual Mode EPR resonator (Bruker ER 4119DM). The sample was allowed to equilibrate at the desired temperatures for 10 min before each measurement. Ex situ measurements performed on the S9.3I cathode samples at room temperature used a high sensitivity EPR resonator (Bruker ER 4119HS-LC). All measurements were carried out with 2 mW of power, a 100 kHz modulation frequency and a modulation amplitude of 4 Gauss. Spectral fits were conducted with EasySpin38.
XPS
The cathode layers were separated from the solid state electrolyte layer by cracking the pellet cells and the conductive tapes were used to fix cathode samples on the XPS sample holders. The XPS tests were conducted on a Kratos Ultra DLD to study the chemical composition and valence states of the elements in the samples with a spot size ranging 300 µm to 700 µm. All of the samples were transported into XPS equipment through a connected Ar-filled glovebox to avoid any air exposure. XPS spectra were acquired with a 0.1 eV resolution for C1s, S2p and I3d regions. The analysis of XPS spectra was performed using CasaXPS software.
XAS
The iodine L2-edge absorption spectra were measured at fluorescence mode at beamline 7-BM of the National Synchrotron Light Source II (NSLS II) at Brookhaven National Laboratory (BNL). Each spectrum takes 15 seconds, and 50 spectra were merged to get the final data for improving the signal-to-noise ratio. The sulfur K-edge spectra were collected in fluorescence mode at the 8-BM beamline of the National Synchrotron Light Source II (NSLS II) at Brookhaven National Laboratory (BNL). All the XAS results were analysed using the Athena software package39.
For PDF measurements, the sample powders were packed inside polyimide capillary (Cole-Parmer) tubes sealed by epoxy glue at the end. The PDF data were collected at the 28-ID-2 beamline of the NSLS II at BNL using a photon wavelength of 0.18475 Å. The obtained data were integrated using Fit2D software40. The PDF and atomic pair distribution function (G(r)) values were extracted using PDFgetX3 software41. For the in situ heating PDF experiment, the sample was heated from room temperature (293 K) to 403 K, and cooled back to 293 K with a step size of 5 K.
UV-Vis spectroscopy
Typically, 20 mg of powder samples was added into 4 ml of THF and then the THF solutions were obtained by centrifugation over several minutes. To prepare the standard Li2S4 and Li2S6 THF solutions, stoichiometric amounts of Li2S and S powder were added into 10 ml of THF with a targeted concentration of 0.2 mmol l−1. Afterwards, the solution and powder mixtures were vigorously stirred over 5 days until fully dissolved. The measurements of samples were performed on a Hitachi UH4150 UV-Vis spectrophotometer.
DSC
10 mg of a powder sample was sealed in a 40 μl Al pan with a lid inside an Ar-filled glovebox. The samples were measured on a TA Instruments Discovery Series DSC 2500 within the temperature window of 20–150 °C at a heating rate of 5 °C min−1.
Raman spectroscopy
Raman measurements were performed on a Perkin Elmer RamanStation 400F with a laser wavelength of 785 nm. The air sensitive samples were sealed between two quartz wafers with Kapton film covering the edges.
Mass spectrometry
Laser desorption ionization–time-of-flight mass spectrometry (LDI-TOFMS) analysis was performed at the Molecular MS Facility at UC San Diego, using a Bruker Autoflex Max matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOFMS) instrument. The LDI-TOFMS measurement was operated under negative ion mode. Laser energy was optimized and set at 10% for ‘soft’ ionization and minimal fragmentation. LDI-TOFMS data were acquired using Bruker flexControl software (v.3.4) and analysed using Bruker flexAnalysis software (v.3.4).
Computational details
DFT parameters
The DFT calculations were carried out using the projector augmented wave (PAW) approach42 as implemented in the Vienna Ab initio Simulation Package (VASP)43. Structure relaxation was performed with the SCAN meta-GGA functional. The calculations of the electronic structure (DOS) were performed using the HSE functional44 as well as the SCAN functional. All the calculations were spin polarized with an energy cutoff of 520 eV. The energy and force convergence criteria were set to 10−4 eV and 0.05 eV Å−1, respectively. For the S unit cell and its derived structures, a Γ-centred k-points mesh of 2 × 1 × 1 was used. For DOS calculations a finer k-points mesh of 2 × 2 × 2 was used. The DFT-optimized minimum energy structures were used to perform Bader charge analysis45. All the input generation and output analyses were performed using Python Materials Genomics (pymatgen)46.
SxI candidate structure generation
SxI candidates at different S:I ratios were obtained by replacing S8 rings with I2 or I3 molecules in octasulfur S unit cells (16 S8 rings). The I2 or I3 molecules were placed at the site of S8 rings in two possible orientations: parallel to the a–b plane along the S8 molecular plane or perpendicular to the S8 molecular plane. Structure enumeration was performed to obtain several distinct site orderings per composition. In addition, the effect of orientational disorder was considered by placing I2 or I3 molecules at different combinations of orientations. The dimensions of the supercell size were 10.587 Å × 12.952 Å × 24.567 Å, which is the same as the unit cell parameter.