A customized strategy to design intercalation-type Li-free cathodes for all-solid-state batteries

Abstract Pairing Li-free transition-metal-based cathodes (MX) with Li-metal anodes is an emerging trend to overcome the energy-density limitation of current rechargeable Li-ion technology. However, the development of practical Li-free MX cathodes is plagued by the existing notion of low voltage due to the long-term overlooked voltage-tuning/phase-stability competition. Here, we propose a p-type alloying strategy involving three voltage/phase-evolution stages, of which each of the varying trends are quantitated by two improved ligand-field descriptors to balance the above contradiction. Following this, an intercalation-type 2H-V1.75Cr0.25S4 cathode tuned from layered MX2 family is successfully designed, which possesses an energy density of 554.3 Wh kg−1 at the electrode level accompanied by interfacial compatibility with sulfide solid-state electrolyte. The proposal of this class of materials is expected to break free from scarce or high-cost transition-metal (e.g. Co and Ni) reliance in current commercial cathodes. Our experiments further confirm the voltage and energy-density gains of 2H-V1.75Cr0.25S4. This strategy is not limited to specific Li-free cathodes and offers a solution to achieve high voltage and phase stability simultaneously.


S4
Electrochemical tests: The working electrode was composed of 70 wt.% of VCS-400 (or VCS-450) sample as active material, 20 wt.% of Super P as conductive additive, 10 wt.% of polyvinylidene fluoride (PVDF) as binder. Using N-methyl-2-pyrrolidone (NMP) as solvent, the mixed slurry was pasted onto an aluminum foil followed by drying at 80 °C for 12 h in vacuum oven. Then the electrode was cut into 10 mm diameter round pieces for use. The mass loading of active materials is ~1.6 mg cm −2 . Polypropylene and Li foil were employed as the separator and reference electrode, respectively. The electrolyte was 1.

Section S1: Traditional voltage tuning strategies for intercalation-type cathodes
In this part, we summarize the electrochemical potential tuning strategies commonly used in the transition-metal-based cathodes for lithium-ion batteries (LIBs), as shown in Fig. S2. Essentially, all these strategies are motivated by changing the ionic/covalent nature of the metal and ligand bond (M-X) that controls the Fermi level in systems, viz., the ionic/covalent change of the M-X bond will govern the quantum mechanical repulsion between bonding and antibonding orbitals, shifting the position of the valence-band top with respect to the Li/Li + energy level, and therefore changing the potential required for the removal of one electron (Fig. S2A). We note that these potential tuning strategies can be divided into the following two types (Fig. S2B): (i) The strategies that directly impact the ionic/covalent nature of M-X bond, such as electronegativity and transition-metal coordination tuning strategies. Previous studies suggested that the higher electrochemical potential of LiCoPO4 cathode (~4.8 V) than LiFePO4 cathode (~3.4 V) is primarily caused by the higher electronegativity of Co (in LiCoPO4, 1.9) than Fe (in LiFePO4, 1.8) [1,2].
Besides, Gutierrez et al. [3] proposed that with the increase of coordination number, the steric hindrance increases, which leads to the more stable M-X bond, thus providing a higher voltage. As a result, the LiFePO4 system with a 6-coordinate octahedral structure exhibits the highest potential (3.4 V) than LiFeBO3 (3 V) with 5-coordinate triangularbipyramid structure as well as Li2FeSiO4 (2.8 V) with 4-coordinate tetrahedron structure.
(ii) The strategies that impact ionic/covalent nature of M-X bond by the indirect primary/secondary inductive effect. The primary inductive effect refers to the case where the charge density on M-X bond is separated by the counter cation, thus reducing the covalent of the M-X bond and increasing the potential, and the secondary inductive effect corresponds to the case where the charge density on M-X bond is partitioned by the neighboring Li atoms, thereby increasing the potential of system. For example, Manthiram et al. [4,5] indicated that the inductive effect elicited by Si-O bond leads to the weakening of Fe-O bond in Fe2(SO4)3, resulting in a significant potential-increasing of Fe2(SO4)3 (~3.6 V) compared to the pristine Fe2O3 cathode (~2.5 V). It is important to emphasize that ionic/covalent nature of above M-X bond is usually not determined by a single factor, even so, based on the above discussions, we conclude that all these general strategies cannot achieve a significant increase of Li + -interaction potential.

Section S2: Basic structural/electronic properties of MX2 systems and their Li +intercalation properties
Group-VIB MX2 mainly crystallize in 2H (space group 6 3 / , #194) and 1T (space group 3 1, #164) polymorphs. Other derived phases, such as 1T' and 3R possess similar intercalation environments for Li + to that of the 1T or 2H phase, thus only these two phases are considered in this work. Examination of the dependence of voltage on W fill for Li +intercalation in Group-VIB MX2 shows that both MS2 and MSe2 have different slopes and intercepts within their individual linear relations. This is because the total energy calculated from first-principles calculation contains three parts (total energy of occupied electronic eigenstates, ion-ion Coulomb energy, and Hartree energy), which is expressed as: =  (Table S3).
Take the common octahedron ligand field as an example, the d orbital is divided into two different symmetries (e g and t 2g ) due to the influence of the perturbation potential, and the energy difference between the two orbitals is ∆ 0 = E(e g ) − E(t 2g ) = 10Dq. It is suggested that the strength of Dq can be described by [6]: Dq= 〈 4 〉 2 6 5 . It is seen that the 10Dq split increases with the increasing of the radial integral 〈 4 〉 of central atom. For different central ions, the calculation method of 〈 4 〉 are different. It is believed that the value of 〈 4 〉 is nearly constant for the same transition metal ion in the same ligand environments. However, for this simple processing method, the relationship between the crystal field splitting and the coordination field environment cannot be quantitatively reflected. Then Shi et al. [7] studied the quantitative relationship between the crystal splits of Ce 3+ and Eu 2+ in halide crystals and environmental factors. By comparing the calculated data with theoretical formula, when the central ion is the same, the ion radius can be used instead of 〈 4 〉. By using the ion radius instead of 〈 4 〉, it is possible to quantitatively compare the degree of splitting of transition metal central atoms in the same period.
Here we propose a descriptor to compare the Fermi levels of specified α and β phases: where α and β indicate the M-d orbital splitting coefficients that determine α and β phases Fermi-level, respectively.

S3.2: − descriptor for quantifying relative stability of different phases
The phase stability is closely related to the crystal field splitting of transition-metal center in systems, which contribute not only to the lowering of crystal field stabilization energy (CFSE), but also to the reduction of electronic configurational entropy (SCFS), by removing orbital degeneracy: G CFS = −CFSE − T × S CFS [8,9]. Considering the contribution of entropy in the ground state (0 K) negligible, the relative energy stability of α and β phases ( G CFS α−β ) can be approximated by crystal field stabilization energy Notably, m-factor in front of pairing energy depends on the number of forced pairing electrons (See Table S4 and Table S5 for different values of ligand-field systems, respectively). Here we take octahedron ligand-field system as an example to illustrate the change rule of the pairing energy with the number of d electrons. If there are only two or three d electrons, both the above-mentioned tendencies can be satisfied simultaneously by placing the electrons in different t2g orbitals with their spins parallel. However, when there are more than three d electrons this is no longer possible. For a d 4 system, occupancy of the t2g and eg orbitals will be different in a strong field and weak field complex. Fig. S7 shows distribution of d 4 electron in the strong field and weak field complex. In the strong field complex 0 > P, so the fourth electron pairs up in the t 2g orbital and gives the electron configuration of the metal as t 2g 4 e g 0 . In the weak field complex 0 < P, so the fourth electron occupies e g orbital and gives the electron configuration of the metal as t 2g 3 e g 1 . If there are 4-7 d-electrons, there are different cases of putting as many electrons into the low-energy t 2g orbital, or distributing them so as to maintain a maximum number of parallel spins.
Each electron in t 2g orbital lowers the energy of the system by 0.4Dq, whereas each S9 electron in an e g orbital raises the energy by 0.6Dq. In addition, each electron pair forced to be paired in the same orbital raises the energy of the system by the pairing energy P. For d 4 -d 7 configurations, if Dq < P, the system is more stable if the electrons occupy the e g orbitals rather than being paired in the t 2g orbitals, giving rise to high-spin complexes. If Dq > P, a low-spin complex results in which the electrons are paired in t 2g orbitals rather than occupying higher-energy e g orbitals. For Group-VB/VIB MX2 systems considered in our work, the electronic configuration is d 1 /d 2 , as illustrated in the small inset in Fig. 1D.
Thus, no paired electrons are formed after the d electrons split, resulting in no paired energy being produced in these systems. As a result, we only need to consider the splitting energy to measure crystal field stabilization energy, CFSE (= ( 1 + 2 +•••)Dq).
Notably, for compounds containing different M element periods, the above method cannot be used. This is because the characteristics of the d orbital itself, such as electron Obviously, the crystal splitting is dominated by two parts: one is the central-ion; the other is the ligand environment [10]. We establish an improved crystal field strength descriptor, which divides Dq as a product of a function f of ligands and g of central-ion: Dq = g(central ion) × f(ligands) = g ′ × 2 4 6 5 (where g ′ represents a spectrochemical series of central ions for the same ligand, showing a strict arrangement according to an increasing number of the transition group 3d < 4d < 5d (with relative values of the functions 1: 1.45: 1.75) or an increasing oxidation numbers +2 < +3 < +4 (with relative values of g ′ approximately 1: 1.6: 1.9). Then we can directly calculate the crystal field stabilization energy difference between specified α and β phases: Section S4: Selection of VxCr2−xS4 (x = 0.5, 1, 1.75) using CALYPSO In this work, the particle swarm optimization algorithm implemented in the CALYPSO (Crystal structure AnaLYsis by Particle Swarm Optimization) method [11,12] was used for the searching of stable VxCr2−xS4 (0 ≤ x ≤ 2) compounds. The effectiveness and the efficiency of this crystal search method have been proven by many well-studied systems, including elements and binary and ternary compounds [13]. Any new materials predicted by CALYPSO have been also experimentally confirmed [14,15]. With the aid of this powerful tool, we obtained the most stable structures of the above selected compounds, as illustrated in Fig. 3A. It should be noted that the V-Cr-S phase diagram (T = 0 K, Fig. S8) does not consider the effect of temperature, which may affect the relative phase stability of 2H-vs. 1T-VxCr2−xS4 phases due to the entropic contributions on free energy at elevated temperature.
Section S6: Elastic properties of 2H-VxCr2−xS4 (x = 0.5, 1, 1.75) The elastic moduli, such as bulk modulus B and shear modulus G of 2H-VxCr2−xS4 (x = 0.5, 1, 1.75) systems were derived based on the Voigt-Reuss-Hill (V-R-H) approximation (Table S8) [16]. The elastic tensor, Cij gives the upper limit of B and G. In the Reuss approximation, the compliance tensor, (= −1 ) is based on uniform stress, leading the lower boundary limit. . We also applied the Born elastic stability criterion to check the mechanically stable of 2H-VxCr2−xS4 under zero pressure. This criterion in harmonic approximation states that for a mechanically stable compound, the relevant elastic tensor must be positive definite. The calculated elastic constants demonstrate that all 2H-phase V0.5Cr1.5S4, VCrS4 and V1.75Cr0.25S4 are mechanically stable (Table S8). S12

Section S7: Experimental analysis A) SEM observation:
The morphologies of VCS-400 and VCS-450 were observed by scanning electron microscopy (SEM). In both low and high magnifications, the VCS-400 has 3D flower-like microstructures assembled with dozens of 2D nanosheets, with a lateral size of 2-4 μm as well as good uniformity (Fig. S19A, B). These nanosheets with smooth surface and regular edges connect to each other to form 3D hierarchical structures. The VCS-450 shows similar microstructure with similar size (Fig. S19C, D).

B) Element analysis:
Energy dispersive X-ray spectroscopy (EDS) mapping results confirmed a uniform distribution of V, Cr, and S elements (Fig. S19E). Furthermore, inductively coupled plasma atomic emission spectrometry (ICP-AES) showed that the Cr/V element ratios agreed with the theoretical values, evidencing a successful doping of Cr into V-based matrix.

C) XRD analysis:
The characteristic (001) diffraction peaks at 15.2° for both the VCS-400 and VCS-450 samples indicate a stacked lamellar structure. Especially, the slightly wider peaks of VCS-400 mean lower crystallinity and smaller grain size, being consistent with its lower synthesis temperature [17].

D) Galvanostatic discharge-charge test:
During the discharge-charge cycles, both VCS-400 and VCS-450 delivered initial discharge capacities very close to the theoretical value of ~230 mAh g −1 , and maintained stability with a high Coulombic efficiency (CE) after 50 cycles.

E) Cross-sectional SEM observation:
Based on the average thickness of electrode pieces, the volume expansion degree of VCS-400 after initial lithiation process (to 1.7 V) was estimated to be ∼4.1% (15.1 μm) as compared with the pristine state (14.5 μm), rendering a considerable structural stability that agreed with the above discharge-charge cycling test. Similarly, the volume expansion degree of VCS-450 after initial lithiation process was estimated to be ∼4.7%.

F) XPS analysis:
For the pristine VCS-400 samples, the two peaks at 517.2 and 524.5 eV correspond to the spin-orbit splitting of V 2p3/2 and V 2p1/2 of V 4+ cations, respectively, when two peaks at 516.0 and 523.2 eV correspond to the spin-orbit splitting of V 2p3/2 and V 2p1/2 of V 3+ cations, respectively [18]. Peak-fitting analysis of S 2p XPS confirmed the presence of S 2p3/2 and S 2p1/2 peaks of S − and S 2− [19]. When it was discharged to 1.7 V, the peak area ratio of the S 2− were obviously enhanced, and the whole peak shifted to lower binding energies as indicated in Fig. 4H. Similarly, the decrease of the chemical valence of V is also observed in Fig. 4G.
Section S8: Improved interfacial resistance between 2H-V1.75Cr0.25S4 cathode and typical sulfide solid-state electrolyte Here we systematically calculated and compared the space charge layer effect at the interface between oxide cathode/sulfide solid-state electrolyte (LiCoO2 (LCO)/Li3PS4 (LPS)) and full-discharged sulfide cathode/sulfide electrolyte (Li2V1.75Cr0.25S4 (LVCS)/LPS) in the equilibrium and initial charging states. We selected LCO as cathode for the present investigation because it is the most widely used cathode material for LIBs, and some studies on the interface are available [20]. Besides, LPS is adopted as a typical example of sulfide electrolyte because it has an ionic conductivity comparable to that of organic liquid electrolytes (~10 −4 S cm −1 ) [21]. In addition, the crystal structure of LPS is simpler than other sulfide electrolytes such as Li7P3S12 or Li10GeP2S12, which is important for controlling the atomic size of the interface structure as well as the computational efficiency [22,23]. We build a slab model of LVCS/LPS and calculate the Li vacancy formation energy in the interface model. LVCS (001), LCO (110) [24] and LPS (010) [25] are selected as the initial surfaces when building slab model, although these surfaces are not necessarily the most stable, they are most likely the conduction path of Li + [26]. In order to eliminate the influence of interface stress and lattice periodicity, we guarantee the interface mismatch (< 10%) and add 1.5 nm vacuum layer to supercell. Considering that the vacancy formation energy of Li atom (Ev) with respect to Li metal can be regarded as Li chemical-potential, the detailed Li + migration behavior under equilibrium and at the initial stage of charging are discussed based on the present calculation results. The relaxed structure shows that Li ions tend to move from LVCS side to LPS side. According to the calculation results, the space charge layer effect also exists in LVCS/LPS at equilibrium state, but it is weaker than LCO/LPS (the difference of Li + concentration on both sides of the LVCS/LPS interface is lower, Fig. S32). As a result, Li2V1.75Cr0.25S4/Li3PS4 has an improved interfacial resistance than LiCoO2/Li3PS4, and its interface is relatively stable.  [27] or experimental results [28][29][30][31][32][33][34][35][36][37][38][39]. Electrochemical stability is measured by the remaining capacity ratio after a certain number of cycles (10-50 cycles) during a constant charge/discharge rate in the range of 0.1 C-0.4 C, where "Stable" represents no attenuation (0%) under ideal conditions. Thermodynamic stability can be divided into "Stable" and "Metastable" states according to the calculated phase diagram given in the MP database, and electronic conductivity can also be divided into "Metal" and "Semiconductor" based on the calculated band gaps given in the database.          Compared with the initial state, the main diffraction peaks of VCS-400 at 35.7° and 57.0° were observed to slightly shift toward lower angles during the lithiation process (to 1.7 V). It agrees with its intercalation-type Li-storage mechanism, since the intercalation of Li + slightly enlarges the unit cell.
Subsequently, during the delithiation process (to 2.8 V), the main diffraction peaks shifted toward higher angles until recovering to their initial state. The stable peak intensity indicates that no obvious conversion reaction occurs during the cycling process, confirming the high structural stability and reversibility of VCS-400. In addition, two strong peaks stabilized at 38.    Figure S31. Rate capabilities of VCS-400 and VCS-450 cathodes. It is observed that VCS-400 demonstrates a similar rate performance as VCS-450. When the rate increased to 2 C, their discharge capacities maintained around 110 mAh g −1 , and their capacities recovered to more than 200 mAh g −1 as the rate returned to 0.1C. Obviously, the same intercalation-type Li-storage mechanism together with same theoretical specific capacities give rise to their similar rate capabilities.  Tables   Table S1. Lattice constants (Å), average M-X bond-length (Å), and the energy difference between 1T and 2H phase (ΔE1T-2H) of MX2 (M = Cr, Mo, W, V, Nb, Ta; X = S, Se).