Synthetic Tailoring of Ionic Conductivity in Multicationic Substituted, High-Entropy Lithium Argyrodite Solid Electrolytes

Superionic conductors are key components of solid-state batteries (SSBs). Multicomponent or high-entropy materials, oﬀering a vast compositional space for tailoring properties, have recently attracted attention as novel solid electrolytes (SEs). However, the inﬂuence of synthetic parameters on ionic conductivity in compositionally complex SEs has not yet been investigated. Herein, the eﬀect of cooling rate after high-temperature annealing on charge transport in the multicationic substituted lithium argyrodite Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I is reported. It is demonstrated that a room-temperature ionic conductivity of ∼ 12 mS cm − 1 can be achieved upon cooling at a moderate rate, superior to that of fast-and slow-cooled samples. To rationalize the ﬁndings, the material is probed using powder diﬀraction, nuclear magnetic resonance and X-ray photoelectron spectroscopy combined with electrochemical methods. In the case of moderate cooling rate, favorable structural (bulk) and compositional (surface) characteristics for lithium diﬀusion evolve. Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I is also electrochemically tested in pellet-type SSBs with a layered Ni-rich oxide cathode. Although delivering larger speciﬁc capacities than Li 6 PS 5 Cl-based cells at high current rates, the lower (electro)chemical stability of the high-entropy Li-ion conductor led to pronounced capacity fading. The research data indicate that subtle changes in bulk structure and surface composition strongly aﬀect the electrical conductivity of high-entropy lithium argyrodites.


Introduction
Increasing demands for fast charging, high-energy-density rechargeable batteries have pushed academic and industrial research towards solid-state battery (SSB) design. [1,2]From a materials perspective, this is pursued through the development of improved cathode, anode and solid electrolyte (SE) materials. [3,4]Apart from the challenge of utilizing lithium-metal anodes and designing a stable interface between the cathode active material (CAM) and the superionic SE, a key requirement for achieving advanced SSBs is the use of (electro)chemically stable and mechanically soft ion conductors.In this regard, lithium thiophosphates represent a major class of promising SEs, as they offer high ionic conductivities and are mechanically soft.7] Particularly lithium argyrodites with the general formula Li 6 PS 5 X (with X = Cl, Br or I) were subject of intense SE development.10][11][12][13] Substitutions typically increase the S 2− /X − site inversion and/or modify the lithium sublattice, which can lead to high ion mobility. [10,12,14,15]part from improving ionic conductivity in crystalline solids by iso-or aliovalent substitutions in the host lattice, [14][15][16][17] the temperature profile used in the solid-state synthesis, i.e., the annealing temperature and especially the cooling rate, can also have a profound effect on the resulting charge-transport properties.[27] When it comes to high-entropy (compositionally complex) lithium-ion conductors, the effect of annealing temperature and cooling rate in general has not yet been thoroughly investigated.
[30] Various HEMs including oxides, sulfides and carbides have been synthesized and employed, among others, as battery materials, catalysts or thermoelectrics. [31,32][35][36][37] However, their (electro)chemical assessment is sparse, and very high ionic conductivities could so far only be achieved for multicationic and -anionic substituted sulfidebased SEs. [36,38,39]Depending on the elemental composition, the entropy contribution may compete with (destabilizing) positive mixing enthalpies ΔH mix .This is particularly critical at low temperatures.While a uniform solid solution may be achieved at elevated temperatures during synthesis, the entropy term may become too small during cooling, leading to demixing depending on the cooling conditions.Whether demixing has a negative effect on ionic conductivity is unclear, and it surely depends on the type of demixing, i.e., if spinodal decomposition occurs or nucleation is required during the demixing process.
In view of these considerations, herein we report on the investigation into how the temperature profile, in particular the cooling rate after high-temperature annealing, affects the ionic conductivity in a multicationic substituted (high-entropy) lithium argyrodite.We demonstrate that altering the cooling rate in the synthesis of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I has indeed a profound effect on conductivity.We find that the maximum ionic conductivity is achieved when the sample is cooled at a moderate rate.Using a combination of structure and charge-transport characterization techniques, the crystalline lattice is found to be relatively robust against different cooling rates, however it exhibits J. Janek Institute of Physical Chemistry & Center for Materials Research (ZfM/LaMa) Justus-Liebig-University Giessen Heinrich-Buff-Ring 17, 35392 Giessen, Germany unique characteristics that enable high ionic conductivity.In addition, X-ray photoelectron spectroscopy (XPS) revealed a surface composition that deviates from the original stoichiometry but appears to be beneficial for charge transport.Finally, the assynthesized SE was tested in pelletized SSBs and compared to the commonly used argyrodite Li 6 PS 5 Cl.Because of the much higher ionic conductivity of the high-entropy material, larger cell capacities were obtained, especially at high C-rates.Yet, the lower (electro)chemical stability of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I than Li 6 PS 5 Cl led to a stronger capacity decay during cycling.

Results and Discussion
Inorganic superionic conductors consisting of a compositionally complex rigid host structure and a defective lithium sublattice (with mobile ions) may be highly sensitive to the synthesis procedure, i.e., annealing temperature and cooling rate.Regarding the synthesis of lithium argyrodites, temperatures ranging from ≈300 to 550 °C with dwell times ranging from a few hours to days are usually applied in the annealing step.However, incorporating a high-energy milling step prior to heating can strongly reduce the required annealing time and induce crystallization, and is even more crucial in the case of compositionally complex materials for ensuring uniform mixing of precursors/elements.Here, stoichiometric amounts of precursors were mixed by ball milling at a relatively low speed, and differential scanning calorimetry (DSC) measurements were performed to reveal potential crystallization and decomposition temperatures (Figure S1a, Supporting Information).Based on the DSC results, three different annealing temperatures, namely 400, 500 and 550 °C, were tested.The corresponding X-ray diffraction (XRD) patterns indicated that 500 °C leads to negligible Li 2 S and LiI impurity phase formation (Figure S1b,c, Supporting Information).Moreover, the material obtained at 500 °C exhibited the highest room-temperature ionic conductivity (in cold-pressed state) among the different samples (10.7 mS cm −1 , see Table S1, Supporting Information).Therefore, an annealing temperature of 500 °C was selected for any further experiments.
A schematic presentation of the synthesis protocol used and the respective temperature profile are depicted in Figure 1a.After the first precursor mixing step (1 h at 250 rpm), the milling speed was increased to 450 rpm, and milling was continued for another 10 h.Subsequently, the recovered powder was pelletized, annealed for 10 h and thereafter cooled at three different rates, namely fast cooling via quenching in liquid nitrogen, moderate (medium) cooling over a time of ≈1.5 h (5 °C min −1 ) and slow cooling over ≈48 h (10 °C h −1 ), which is referred to as Q, MC and SC, respectively, hereafter.As can be seen from the XRD patterns in Figure 1b, the high-energy milling step already induced crystallization of the argyrodite phase.[42] A small crystallite size can be expected considering the broad reflections.To improve crystallinity of the material, pelletized samples were annealed in vacuum-sealed quartz ampules for 10 h at 500 °C.At first glance, no major differences in the patterns were noticed (Figure 1b).To probe the crystal structure, the samples were subjected to high-resolution neutron powder diffraction (NPD) at 298 and 10 K (to minimize atomic displacement), followed by Rietveld refinement analysis.The Rietveld profiles for the Q and SC samples measured at T = 10 K are exemplarily shown in Figure 1c,d.The patterns could be indexed in the F−43m space group with very similar lattice parameter of a = 10.25696(6)Å (SC) and 10.25347(4) Å (Q) (a = 10.24932(5)Å for the MC sample, see Table S2, Supporting Information).The structural parameters from Rietveld analysis are given in Table S3-S6 (Supporting Information) and ref. [36].
The calculated crystal structure of Li   distributed throughout the lattice.Moreover, in the argyrodite structure, S 2− (4d) and I − (4a) anions are known to mix over the respective Wyckoff positions (S 2− /X − site inversion).We found a site inversion of ≈13% for the Q and SC samples, compared to 12% for the MC sample, among the highest values reported in the literature for S 2− /I − mixing.Analysis of the NPD data also provided detailed structural information on the lithium sublattice.The Li atoms usually form Frank-Kasper polyhedra around the 4d site, expressed in two Wyckoff positions, namely 48h and 24g, [8] thereby producing 3D lithium diffusion pathways (Figure 2b).For the Q sample, it was found that 70 and 30% of Li is located on 48h and 24g, respectively, at room temperature.Upon reducing the cooling rate, the Li occupancy on the 48h site is lowered to 62 and 60% for MC and SC, respectively (Figure S2a, Supporting Information).We note that the occupancies remained virtually unaltered at 10 K. Surprisingly, a slight increase in Li situated on the 24g site was observed with decreasing cooling rate, which can be regarded as an intermediate/transition site.Therefore, the opposite trend might be expected, meaning quenching the sample should lead to increased Li occupancy on 24g.This counterintuitive finding seems to be a result of the large lattice distortions prevailing in high-entropy lithium argyrodites.The slightly varying Li occupancies also led to somewhat different Li-Li jump distances, with the intercage jump distance being an important structural parameter in determining the long-range lithium-diffusion characteristics (Figure S2b, Supporting Information).At room temperature, this distance was smallest for the MC sample [3.24(2) Å, compared to 3.28(3) and 3.25(3) Å for Q and SC, respectively, see Table S7, Supporting Information).Naturally, all Li-Li distances were lower upon cooling to 10 K.
To further corroborate the presence of [PS 4 ] 3− and [SiS 4 ] 4− tetrahedra within the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I structure and reveal the presence of potential amorphous impurities (invisible from diffraction data), 31 P, 29 Si and 6 Li magic-angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy measurements were carried out.The respective spectra are shown in Figure 3.][45] It should be noted that, compared to Li 6 PS 5 I, the 31 P signal was shifted by ≈5 ppm, which is probably due to the unique chemical environment around the [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 4 3.5− tetrahedra. [8]Regarding the 29 Si signal, a distinct peak located at 11.5 ppm was observed in all cases (Figure 3b), evidencing the presence of [SiS 4 ] 4− species, in agreement with literature results. [36,43]In the 6 Li MAS NMR data (Figure 3c), a peak at 1.49 ppm appeared, similar to other lithium argyrodites. [9,45,46]Overall, comparable spectra with no signs of impurity contributions were obtained for the samples prepared with different cooling rates.This again demonstrates the minor effect that cooling rate has on the average/local crystal structure and further verifies the robustness of the high-entropy argyrodite lattice.
The local structure of the [GeS 4 ] 4− tetrahedra was studied by Ge K-edge transmission X-ray absorption spectroscopy (XAS) measurements conducted on the MC sample and a GeS 2 reference material.Extended X-ray absorption fine structure (EXAFS) spectra [k 2 -weighted (k)] and the corresponding magnitudes of the Fourier transform are shown in Figure S3a-c [47] To gain information on the oxidation state of the elements and possible differences in near-surface composition, XPS measurements were carried out.The respective binding energies from curve fitting are given in Table S8 (Supporting Information).For the Sb 3d core-level region, only the 3d 5/2 components are shown (Figure 4a-c) due to the large spin-orbit splitting (≈9.3 eV).For all samples, the Sb 3d 5/2 spectrum is superimposed by the O 1s signal at a binding energy of ≈532 eV.The asymmetry of the Sb 3d 5/2 line indicates two different oxidation states, with peaks centered at 530.3 and 529.4 eV.][50] The asymmetric O 1s line also points toward at least two chemical states, which can most likely be attributed to oxygenated Si, Sb, P and/or Ge.However, clear assignment is challenging.][53] In addition, a minor S 2p component is evident at 163.5 eV, which is usually ascribed to polysulfide impurities. [54]igure 4g-i shows the P 2p core-level region, revealing a doublet peak at 132.2 eV for all samples characteristic of the [PS 4 ] 3− units. [36,51,55,56]The Si 2p data are centered at ≈101 eV, and again, they are virtually identical for the different samples (Figure 4j-l).The major component can be assigned to [SiS 4 ] 4− , in agreement with reports available in the literature, [57,58] while the minor component at 102.6 eV lies in between the binding energies usually detected for [SiS 4 ] 4− and SiO 2 .][59][60] The I 4d region (≈2.0 eV spin-orbit splitting) is depicted in Figure 4m-o.The presence of I − is apparent from the characteristic contributions, with the major one at ≈49 eV and the minor one centered ≈0.5 eV higher in binding energy.It has been hypothesized that these contributions stem from two different crystallographic environments in the structure (4a and 4d sites). [36]The Ge 3d and Sb 4d regions are shown in Figure 4p-r.The Sb 4d peaks comprise two contributions, at ≈34.3 and 33.2 eV, confirming the presence of Sb 5+ and Sb 3+ , respectively, at the near-surface regions of the samples.In addition, the Ge 3d spectra (≈30.8 eV) seem to indicate the presence of a single component (corresponding to [GeS 4 ] 4− ). [48,61]Taken together, the spectra collected from the samples prepared with different cooling rates were found to be very similar for the S 2p, P 2p, I 4d and Ge 3d core levels.However, the data also revealed increased Sb 3+ (contribution at 529.4 eV) and SiO x (contribution at 102.6 eV) contents for the MC sample.
To visualize the complex surface composition of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I, scanning transmission electron microscopy (STEM) imaging combined with energy-dispersive X-ray spectroscopy (EDX) was used.As shown in Figure 5, particles in the micrometer size range were observed.The corresponding elemental maps revealed uniform distributions on the nanoscale.As expected, oxygen was also found on the particle surface, in agreement with the oxygenated species detected by XPS (see Figure 4a-c,j-l).According to quantitative XPS analysis, the concentration of surface oxygen was much lower than that of sulfur (Table S9, Supporting Information).However,  when the cooling rate was very low, a slight increase in oxygen content was noticed (for the SC sample).Overall, the oxygen signals from XPS and STEM-EDX can be clearly assigned to surface impurity formation, despite the fact that the materials were strictly handled under inert atmosphere.
After having revealed similar bulk and surface structural characteristics for the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I prepared with different cooling rates, the Li-ion dynamics were probed in the temperature range of 30-70 °C using 7 Li pulsed field gradient (PFG) NMR spectroscopy (Figure S4, Supporting Information).In so doing, diffusion coefficients of D Li = 6.58•10 −12 m 2 s −1 (Q), 7.09•10 −12 m 2 s −1 (MC) and 6.77•10 −12 m 2 s −1 (SC) were determined at 30 °C, which are on the same order of magnitude to that of other highly conducting lithium thiophosphates. [9,36,45,46,62,63]rrhenius plots of diffusivity against reciprocal temperature are shown in Figure 6a.Virtually identical activation energies of E A = 0.20, 0.20 and 0.19 eV were determined from the slopes (assuming Arrhenius-type temperature dependence) for Q, MC and SC, respectively.
The electrical conductivities were determined by electrochemical impedance spectroscopy (EIS) measurements conducted on sintered pellets with ion-blocking electrodes between 15 and 65 °C.The corresponding Nyquist plots of the electrochemical impedance only revealed a capacitive tail, indicating high ionic conductivity (Figure S5, Supporting Information).Roomtemperature conductivities of 11.9 (±0.39), 12.3 (±0.70) and 11.0 (±0.57) mS cm −1 were calculated for Q, MC and SC, respectively, and very similar activation energies of E A = 0.19 (Q), 0.19 (MC) and 0.20 eV (SC) were determined from the corresponding Arrhenius plots (Figure 6b).As can be seen, they agree well with the 7 Li PFG NMR results.
Using the Nernst-Einstein equation, the ionic conductivities were also estimated from the diffusion coefficients.A summary of the EIS and 7 Li PFG NMR results is given in Table 1 and further depicted in Figure 7a.From the data, it is evident that the ionic conductivities determined by 7 Li PFG NMR are lower by ≈2 mS cm −1 .This difference between EIS and 7 Li PFG NMR is presumably rooted in the contribution of (favorable) grainboundary conduction, as EIS probes the overall specimen conductivity unlike 7 Li PFG NMR (only bulk transport properties are probed).However, a similar trend in conductivity with cooling rate is apparent from both techniques.This means that the conductivity of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I (MC) is indeed higher than that of the Q and SC samples.No significant influence of the cooling rate on activation energy was found.
To further assess the transport properties of the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I samples, their electronic conductivity was determined by DC polarization measurements to be 3.83•10 −5 , 3.18•10 −7 and 1.23•10 −5 S cm −1 for Q, MC and SC, respectively.Interestingly, MC was found to exhibit two orders of magnitude lower electronic conductivity than the Q and SC samples (Figure 7b).We assume that this is related to the surface composition and presence of impurities.The fact that the electronic conductivity is orders of magnitude higher than that of other Li-ion conductors seems to be due, in part, to the intrinsic properties of high-entropy lithium argyrodites. [9,46,64]SBs with lithium thiophosphate SEs typically make use of cold-pressed materials.For that reason, the conductivity of the different samples was also determined in a cold-pressed state.As expected, the resulting ionic conductivities were somewhat lower than those of sintered pellets.Nevertheless, high roomtemperature conductivities of 7.4 (±0.4), 10.9 (±0.25) and 5.5 (±0.15) mS cm −1 were found by EIS for Q, MC and SC, respectively, following a similar trend to the sintered pellets (Figure 7b).As mentioned above, this means that the MC sample indeed shows superior ionic conductivity over Q and SC and sintering helps to achieve better grain contact, thereby minimizing grainboundary contributions.From a surface composition point of view, it is found that increasing amounts of Sb 3+ and SiO x species correlate with increasing ionic conductivity and decreasing electronic conductivity (Figure 7c).Hence, slight changes in surface composition apparently can exert a substantial impact on the conductivity of polycrystalline materials.In this regard, it is important to note that enhanced interfacial ion transport is a well-known phenomenon [65,66] and might play a role here as well.For example, Sb 2 S 3 (with Sb 3+ ) is a poor electronic conductor, [67,68] which agrees with our finding that an increasing Sb 3+ (surface) fraction leads a strong decrease in electronic conductivity.From a structural perspective, the short Li-Li intercage jump distances (Figure 7d), together with favorable surface/grain-boundary composition, are likely responsible for fast lithium transport in the MC sample.These results collectively show that, in addition to favorable bulk structural features, the surface composition strongly affects the overall ionic conductivity.However, little attention has been paid to this in the context of sulfide-based SEs up until now.
Finally, the electrochemical behavior of the best-conducting SE sample (MC) was tested in pellet-type SSBs with a LiNbO 3 -coated LiNi 0.85 Co 0.1 Mn 0.05 O 2 (NCM851005) cathode and an indiumlithium anode and compared to that of commercially available argyrodite Li 6 PS 5 Cl.Initially, the cells were cycled at a C/10 rate and 25 °C, followed by ex situ EIS measurements.For the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I, first-cycle specific charge and discharge capacities of 237 and 182 mAh g −1 were achieved, which translates to 77% Coulomb efficiency (referred to as CE in the following, see Figure 8a).In the case of Li 6 PS 5 Cl, the cells delivered a lower charge but higher discharge capacity, with q ch = 221 mAh g −1 and q dis = 190 mAh g −1 , corresponding to a CE of 86%.The ≈10% difference in CE upon using the multicomponent SE suggests a lower (electro)chemical stability compared to the Li 6 PS 5 Cl.However, the impedance spectra revealed a larger charge-transfer resistance for Li 6 PS 5 Cl (Figure 8b).Fitting of the data allowed determining the different contributions to the overall impedance (Figure S6, Supporting Information), namely bulk SE (R bulk ), SE grain boundary (R GB ), CAM/SE (R CAM/SE ) and anode/SE (R anode/SE ).The R GB was found to be 15.5 Ω (12.2 Ω cm 2 ) for Li 6 PS 5 Cl and only 3.4 Ω (2.7 Ω cm 2 ) for Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I, presumably due to favorable grainboundary conduction as discussed above.For the R CAM/SE and R anode/SE , values of 78.9 Ω (61.9 Ω cm 2 )/6.6 Ω (5.2 Ω cm 2 ) and 4.7 Ω (3.7 Ω cm 2 )/27.1 Ω (21.3 Ω cm 2 ) were obtained for Li 6 PS 5 Cl and the high-entropy SE, respectively.Interestingly, the R CAM/SE showed the opposite trend to the first-cycle CE, which was significantly lower for Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I.We assume that, despite more severe degradation, the charge transport through the cathode interface(s) is facilitated, which is directly reflected in the impedance spectra.
Next, the SSB cells were subjected to rate capability and longterm performance testing.To this end, the galvanostatic cycling was conducted at 25 °C and C-rates ranging from C/2 to 2C (Figure 8c; Figure S7a, Supporting Information).For Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I, first-cycle specific charge and discharge capacities of 226 and 168 mAh g −1 were achieved at C/2, corresponding to an initial CE of 74% (Figure 8d).In contrast, the Li 6 PS 5 Cl-based cells delivered lower capacities of q ch = 198 mAh g −1 and q dis = 147 mAh g −1 , leading to a similar CE of 74%.Notable is the large overpotential of the Li 6 PS 5 Cl-based cells (increased by ≈250 mV compared to the high-entropy SE, see differential capacity plots in Figure S7b, Supporting Information).This increase in overpotential is probably due to the lower room-temperature ionic conductivity of Li 6 PS 5 Cl (≈2 mS cm −1 versus 11-12 mS cm −1 ).After five cycles, the C-rate was first increased to 1C and then to 2C.As can be seen from the data for the high-entropy SE, specific discharge capacities of ≈135 and 101 mAh g −1 were achieved at 1C and 2C, respectively.For Li 6 PS 5 Cl, the drop in discharge capacity with increasing current rate was much more pronounced, with q dis ≈ 109 mAh g −1 at 1C and 53 mAh g −1 at 2C.This result can be explained by the significantly higher ionic conductivity of the high-entropy SE, which is capable of better accommodating high current densities (2C ≈ 4.4 mA cm −2 ).Subsequently, the cells were cycled at C/2 and found to undergo virtually linear capacity decay, with q dis ≈ 123 mAh g −1 in the 49th cycle, corresponding to an overall capacity loss of 27% and 16% (relative to the initial discharge capacity) for Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I and Li 6 PS 5 Cl, respectively.This difference in capacity retention provides evidence for the lower electrochemical stability of the multicomponent SE, for which possible decomposition phases (interfacial degradation products) have recently been identified by XPS. [36]Although the initial CE (C/2 rate) was similar for both SEs tested, it was lower for the high-entropy material during subsequent cycling (Figure 8e).After an increase over the first five cycles, the CE stabilized below 99% for Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I but above 99.5% for Li 6 PS 5 Cl, corroborating the above conclusions.This further implies that the as-formed decomposition interphase is kinetically unstable, leading to continuous SE degradation, which might be associated with the increased electronic conductivity.Interestingly, a bump in the CE data between cycles 30 and 40 was noticed for Li 6 PS 5 Cl, which is not mirrored in the specific discharge capacities in Figure 8c.This has repeatedly been reported in the literature and assigned to (chemo)mechanically-driven cell degradation. [51,69,70]However, in the case of the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I, it did not occur, and we assume that this is because of the different softness of the SEs (Young's modulus).Note that increasing compositional complexity typically causes a lowering of the Young's modulus. [71]o determine the (electro)chemical stability window of both materials and deconvolute anodic and cathodic stability issues, cyclic voltammetry (CV) measurements were conducted on mixtures of SE and carbon black as working electrode.Figure 8f shows the first-cycle CV curves, which reveal larger currents in the anodic and cathodic scans for Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I.This again confirms the lower stability of the high-entropy SE (exhibiting a relatively narrow electrochemical stability window).With further cycling (Figure S8, Supporting Information), a strong decrease in the absolute reductive (cathodic) current was noticed for both SEs, while the oxidative (anodic) current only changed marginally in the case of the high-entropy SE.
Overall, the CV data agree with the results from galvanostatic cycling, indicating the formation of a kinetically unstable decomposition interphase that does not effectively prevent SE degradation during battery operation.

Conclusion
In summary, we have examined the impact of the synthesis procedure, especially annealing temperature and cooling rate, on structure-charge transport property relationships in the multicationic substituted Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I SE.As reported in the past for ceramic Li-ion conductors, the synthesis procedure/conditions may have a profound effect on ionic conductivity, however clear trends have yet to be established.[27] For the material studied in this work, we found that the cooling rate after a given high-temperature anneal indeed affects the ionic conductivity.For medium cooling, an increase in conductivity over fast-and slow-cooled samples was observed.To understand the origin of this result, the bulk crystal structure was probed using XRD and NPD in combination with MAS NMR spectroscopy.For the medium-cooled sample, the shortest Li-Li intercage jump distances and a high degree of S 2− /I − site inversion (12-13%) were found.Apart from these for ion mobility favorable bulk structural characteristics, the surface composition was also investigated via XPS and STEM-EDX.Small differences among the samples were noticed, in particular the presence of increased amounts of Sb 3+ and SiO x surface impurities for the medium-cooled material.It can be assumed that the unique surface composition also affects, at least to some degree, the conductivity.Overall, it seems that the high ionic conductivity of the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I SE is mainly due to a combination of advantageous bulk and surface structural/compositional features.However, the bulk crystal structure was found to be less susceptible to temperature effects than the surface (composition).The findings further emphasize that classical synthetic approaches to optimize ionic conductivity in ceramic ion conductors may not be transferable to compositionally complex SEs.
Because of the high room-temperature ionic conductivity, the MC sample was also tested in pellet-type SSBs and the performance compared to that achieved with Li 6 PS 5 Cl.The highentropy SE cells were capable of delivering larger specific discharge capacities at high current rates but suffered from more pronounced (electro)chemically-driven degradation, leading to faster performance decay.The data clearly demonstrate that the conductivity in lithium argyrodite SEs can be synthetically tailored, however their stability seems to be lower than that of Li 6 PS 5 Cl.This indicates that achieving high ionic conductivity is not the only challenge but other metrics, such as (electro)chemical stability and mechanical behavior, need to be considered too.Taken together, high-entropy SEs provide a vast compositional space that is not limited to sulfide-based materials, thereby maximizing the opportunity to improve several key performance indicators of SSBs by compositional design.

Experimental Section
General: If not stated otherwise, all work steps were performed under inert atmosphere, and the precursors were used as received.
Differential Scanning Calorimetry: Measurements were performed at a heating rate of 5 °C min −1 using a NETZSCH DSC 204 F1 Phoenix.To this end, the samples were sealed in alumina crucibles under Ar atmosphere.
Laboratory X-ray Diffraction: The samples were sealed in borosilicate capillaries (0.68 mm inner diameter and 0.01 mm wall thickness, Hilgenberg) under Ar atmosphere and subjected to XRD analysis using a STOE Stadi-P diffractometer with a DECTRIS MYTHEN 1K strip detector in Debye-Scherrer geometry.The instrument utilizes a Mo anode to generate X-rays of wavelength  = 0.70926 Å.
Neutron Powder Diffraction: For NPD, cylindrical vanadium containers of diameter 6 mm were filled with ≈2 g of sample.The measurements were performed at the D2B high-resolution powder diffractometer located at the Institut Laue-Langevin (ILL) using a wavelength of  = 1.595072Å at both T = 10 and 298 K.The diffraction data were analyzed via Rietveld refinement using the FullProf Suite software. [72]The peak shape was modeled using the Thompson-Cox-Hastings pseudo-Voigt function, and a pointby-point background was subtracted.The following parameters were refined one by one: Scale factor, peak shape parameters, lattice parameters, atomic coordinates, individual anisotropic atomic displacement parameters and lithium occupancies.The zero-shift parameter was refined last, and any occupancies that resulted in unreasonable values were disregarded.Finally, all parameters were refined simultaneously to ensure stability of the calculated crystal structure.
Magic-Angle Spinning Nuclear Magnetic Resonance Spectroscopy: A Bruker Avance 500 MHz spectrometer was used for MAS NMR spectroscopy measurements with a magnetic field strength of 11.7 T, corresponding to resonance frequencies of 73.6, 99.4 and 202.5 MHz for 6 Li, 29 Si and 31 P, respectively.The samples were inserted into 2.5 mm rotors

Figure 1 .
Figure 1.Synthesis and structural characterization of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I. a) Schematic illustration of the synthesis pathway including milling and annealing, followed by one of the three chosen cooling rates.b) XRD patterns for the ball-milled precursor mixture and samples prepared with different cooling rates.c,d) NPD patterns and corresponding Rietveld plots at T = 10 K for slow-and fast-cooled samples.The open circles and black and gray lines represent the observed, calculated and difference profiles, respectively.Expected Bragg reflections are indicated by vertical ticks.

Figure 2 .
Figure 2. Crystal structure of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I. a) Schematic view with the different Wyckoff positions and their elemental constituents indicated in the legend on the right.b) Bond-valence energy landscape revealing the 3D lithium diffusion pathways (gray trajectories).For clarity, only atoms on the 4a and 4d sites are shown.[P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 4 3.5− polyanions are depicted as pink tetrahedra.

Figure 3 .
Figure 3. Sulfide-based tetrahedral polyanions revealed by MAS NMR spectroscopy.a) 31 P, b)29 Si and c)6 Li spectra collected from the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I samples prepared with different cooling rates (Q, MC and SC refer to fast, medium and slow cooling, respectively).

Figure 4 .
Figure 4. Effect of cooling rate on surface composition of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I probed using XPS.High-resolution a-c) Sb 3d 5/2 and O 1s, d-f) S 2p, g-i) P 2p, j-l) Si 2p, m-o) I 4d and p-r) Sb 4d and Ge 3d photoelectron spectra collected from samples prepared with different cooling rates (Q, MC and SC refer to fast, medium and slow cooling, respectively).Black dots and solid lines represent the experimental data and curve-fitting results, respectively.

Figure 5 .
Figure 5. Results from STEM-EDX imaging/mapping.A representative high-angle annular dark field (HAADF) image and corresponding elemental maps from EDX are shown.

Figure 6 .
Figure 6.Charge-transport properties of Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I determined by a) 7 Li PFG NMR spectroscopy and b) EIS.a) Arrhenius plot of the diffusion coefficient and b) conductivity versus reciprocal temperature for the samples prepared with different cooling rates.The activation energies are indicated.If error bars are not visible, the standard deviation is smaller than the symbol.

Figure 7 .
Figure 7. Bulk structure/surface composition-charge transport property relationships.a) Ionic conductivities and activation energies determined by EIS and 7 Li PFG NMR spectroscopy.b) Comparison of ionic conductivities measured on sintered and cold-pressed pellets and electronic conductivities determined from DC polarization measurements (on sintered pellets) for the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I samples prepared with different cooling rates.c) Normalized Sb 3+ and SiO x contents from XPS and d) Li-Li intercage jump distances and S 2− /I − site inversions shown together with the ionic conductivities.If error bars are not visible, the standard deviation is smaller than the symbol.Dashed lines are for eye guidance only.

Figure 8 .
Figure 8. (Electro)chemical testing of bulk-type SSBs.a) Initial voltage profiles of cells containing either the Li 6.5 [P 0.25 Si 0.25 Ge 0.25 Sb 0.25 ]S 5 I (MC) or commercial argyrodite Li 6 PS 5 Cl SE at C/10 rate and 25 °C.b) Impedance spectra collected after the first cycle shown in a).c) Rate capability and longterm cycling performance.d) Coulomb efficiencies for the first two and e) following cycles.f) First-cycle CV curves for mixtures of SE and carbon black as working electrode.Data are averaged from two identical cells.

Table 1 .
7ummary of the Li-ion conductivities ( ion ) at T = 298 K and corresponding activation energies (E A ) from temperature-dependent EIS and7Li PFG NMR spectroscopy measurements..