Carbon-free and binder-free Li-Al alloy anode enabling an all-solid-state Li-S battery with high energy and stability.

Incompatibility of electrolytes with Li anode impedes the application of solid-state batteries. Aluminum with appropriate potential, high-capacity, and electronic conductivity can alloy with Li spontaneously and is proposed herein as a carbon-free and binder-free anode of an all-solid-state Li-S battery (LSB). A biphasic lithiation reaction of Al with modest volume change was revealed by in situ characterization. The Li0.8Al alloy anode showed excellent compatibility toward the Li10GeP2S12 (LGPS) electrolyte, as verified by the steady Li0.8Al-LGPS-Li0.8Al cell operation for over 2500 hours at 0.5 mA cm-2. An all-solid-state LSB comprising Li0.8Al alloy anode and melting-coated S composite cathode functioned steadily for over 200 cycles with a capacity retention of 93.29%. Furthermore, a Li-S full cell with a low negative-to-positive ratio of 1.125 delivered a specific energy of 541 Wh kg-1. This work provides an applicable anode selection for all-solid-state LSBs and promotes their practical procedure.

INTRODUCTION

The achievement of carbon neutrality and the development of sustainable society have been the common goals of all human beings under the stress of global climate change. Batteries with high specific energy are one of the key techniques to reduce carbon emission and alleviate environmental pressure. Meanwhile, state-of-the-art Li-ion batteries consisting of intercalating cathodes and graphite anodes cannot meet the market demand for batteries with high specific energy (, ). Thus, advanced energy storage devices with a high capacity and composed of conversion cathode materials paired with Li anode have aroused wide attention. Sulfur, which is abundant, low-cost, and nontoxic, has a high theoretical specific capacity of 1672 mAh g−1 based on the following reaction

Coupling with Li anode, Li-S batteries (LSBs) have a high specific energy of up to 2600 Wh kg−1. The high solubility of intermediate products, namely, polysulfides, in the liquid electrolyte can promote the quick electrochemical reaction for the S cathode. Nevertheless, during discharge, the dissolved polysulfides will uncontrollably diffuse to the anode, causing a severe Li corrosion and a low coulombic efficiency of the battery. This phenomenon is called the “shuttle effect.” Adding to these dilemma are safety concerns, such as leakage, fire, and explosion associated with liquid electrolytes, which pose a huge challenge to the application of liquid LSBs. Substitution of liquid electrolytes with solid-state electrolytes (SSEs) is an effective approach to fundamentally resolve the two issues mentioned above.

SSEs act as a Li-ion pathway within the battery and a separator between the cathode and anode. Sulfide SSEs, such as Li10GeP2S12 (LGPS) and Li6PS5Cl, with high ionic conductivities and moderate Young’s modulus can result in high rechargeability and low internal resistance of batteries (, ). Nevertheless, sulfide electrolytes suffer from their incompatibility with Li. Two strategies have been proposed to solve the instability problem of electrolyte with Li: the protection of Li by a buffer layer and the substitution of Li by another anode with better compatibility to sulfide electrolytes. A buffer layer, such as Li3PS4 () or 0.75Li2S-0.24P2S5-0.1P2O5 (), can effectively enhance the stability between the electrolyte and highly active Li anode. Meanwhile, the internal resistance of batteries will increase with the introduction of more interfaces. The buffer layer can also be the artificial solid electrolyte interface (SEI) formed by the pretreatment of Li or the SSE (, ). However, the complicated pretreatment process will largely increase the difficulty of preparation and raise the cost.

Li alloys with a suitable operating potential are applicable substitutes for Li as anodes in solid-state batteries. This substitution can result in two effects. The degradation of SSEs caused by active Li can be greatly relieved. In addition, the safety of batteries is greatly improved because of the suppression of lithium dendrite growth. Li-In alloy is one of the most popular Li alloys used in sulfide-SSE batteries (–). Zhang’s group successfully designed an all-solid-state LSB consisting of Li-In alloy anode, LGPS electrolyte, and S@carbon nanotube (CNT) cathode (). However, although Li-In alloy is useful as the counter electrode of solid-state batteries for scientific research, it is subject to a low theoretical specific capacity of 233 mAh g−1. Furthermore, the rarity of In raises the cost of the battery and restricts its large-scale application. Li-Si alloy is popular for its high specific capacity () and low cost. Recently, Meng and co-workers reported a microsilicon anode with outstanding performances under large current density in the all-solid-state Li-ion battery (). However, there are concerns about the Si anode’s low electronic conductivity () and the severe volume change during cycling (, ).

Carbon materials are commonly used as a conductive agent in the anode. However, in solid-state batteries, the addition of carbon is detrimental to the anode-SSE interface because carbon may promote the decomposition of SSEs (, ). Furthermore, anode materials in powder form require a binder to prepare the electrode, which will increase the mass of inactive substances and the internal impedance of the battery. In summary, a suitable anode should have great qualities, such as high conductivity, considerable capacity, moderate potential, low volume expansion, and low cost. Aluminum, as the third most abundant element in Earth’s crust, has a high electronic conductivity of up to 3.5 × 107 S m−1. The excellent electronic conductivity enables the design of a carbon-free Al anode, which is beneficial to the stability of the anode-SSE interface. Furthermore, the specific capacity of the Al anode (990 mAh g−1) is notably higher than that of the conventional graphite anode (372 mAh g−1) and is competitive among Li alloy anodes. In addition, the high ductility of Al enables the easy formation of Al foil without a binder, reducing the amount of inactive component in the battery. The volume change rate of the Li-Al alloy during cycling (LiAl: 96%) is remarkably smaller than that of the Li-Si alloy (Li4.4Si: 320%) (). Furthermore, the operating potential of the Li-Al alloy is 0.3 V versus Li/Li+ (), which is lower than that of the Li-Si alloy (0.4 V versus Li/Li+) () and Li-In alloy (0.6 V versus Li/Li+) (). A low anode potential will lead to a high output voltage and specific energy of the battery. In consideration of these advantages, the Li-Al alloy is a promising candidate for the advanced anode material. Li-Al alloy has been studied as the protective coating layer of Li (, ) and the anode material (–) in liquid electrolyte batteries, whereas it has been rarely reported in all-solid-state LSBs with sulfide SSEs.

Here, a promising all-solid-state LSB involving a carbon-free and binder-free Li-Al alloy anode was designed and fabricated (Fig. 1A). In this battery, the composition of the Li-Al alloy anode was carefully regulated to guarantee its compatibility with the LGPS electrolyte. As illustrated in Fig. 1B, the operating potential of the screened-out Li-Al alloy anode with optimized constitution is located within the practical stability window of the LGPS, preventing the reductive decomposition of the electrolyte. In situ x-ray diffraction (XRD) was adopted to investigate the lithiation process of Al and unveiled a two-phase reaction with modest volume change. The appropriate operating potential and temperate volume change of Li-Al alloy resulted in its excellent stability toward LGPS, as proven by the electrochemical test of symmetric cells and postmortem characterizations of the LGPS pellet after cycling. Furthermore, the electrochemical performances of the full cell were demonstrated to verify the feasibility of the proposed battery system. The assembled all-solid-state LSB exhibited an excellent reversibility and a superb cycling stability for over 200 cycles with a high capacity retention of 93.29%. Last, a well-operated Li-S full cell delivering a specific energy of 541 Wh kg−1 (based on the mass of S and Li0.8Al) was achieved with a low negative-to-positive (N/P) ratio of 1.125, which confirms the bright practical prospect of the all-solid-state LSBs.

Fig. 1.

Schematics of the Li-Al alloy anode in an all-solid-state LSB.

(A) Schematic of a rechargeable all-solid-state LSB with Li-Al alloy anode and its reaction mechanism. (B) Schematic of the practical stability window of the LGPS electrolyte and the chemical potential of different electrodes.

RESULTS AND DISCUSSION

Practical stability window of LGPS and characteristics of Li-Al alloy

The electrochemical stability window (ESW) of the SSE is intrinsically determined by the energy level of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of the electrolyte. Ideally, the operating voltage range of the batteries should be within the electrolyte ESW to avoid its decomposition, whereas in practical applications, batteries often work well at the voltage range exceeding the thermodynamic ESW of the electrolytes. For example, although LGPS has a narrow thermodynamic ESW of 1.7 to 2.1 V versus Li/Li+ (Fig. 1B) (), Li-In alloy or Li4Ti5O12 (1.55 V versus Li/Li+) () with an operating potential lower than 1.7 V has been widely used as the anode of LGPS electrolyte batteries and shown great stability. Thus, the practical stability window of an electrolyte may be notably wider than the computed theoretical value. The experimental oxidative limit of LGPS was 2.8 V versus Li/Li+ (), and this value is consistent with the linear sweep voltammetry test results in fig. S2. Nevertheless, the practical reductive stability limit of LGPS, which is a major consideration in selecting the anode, has rarely been studied. Here, cyclic voltammetry was used to experimentally measure the practical reductive stability limit of LGPS. The configuration of the testing cell was LGPS/Pt-LGPS-Li0.5In alloy. In the working electrode, Pt powder as the conductive agent was mixed with LGPS to enlarge the electrical contact area and enhance electronic conductivity. Traditional conductive agent carbon materials were abandoned because the lithiation current of carbon at low potential may be mistaken as the decomposition current of LGPS and interfere with the test results. Li0.5In alloy with a stable potential of −0.62 V versus Li/Li+ (fig. S3) was used as the counter and reference electrodes. The cutoff potential of the cathodic sweep was set to −0.62 V to avoid the disturbance of Li deposition/dissolution current. As shown in Fig. 2A, the apparent reduction current peak arose at a low potential, indicating the decomposition of LGPS. The inset figure in Fig. 2A displays the differential curve of the first cathodic scan to visualize the setoff potential of LGPS decomposition. A sharp rise in the slope occurred at 0.125 V versus Li/Li+ (−0.495 V versus Li0.5In), signifying that the reductive stability limit of LGPS is 0.125 V versus Li/Li+. This result is a practical guidance to the application of LGPS, given that in the working electrode with Pt powder, the charge transfer was promoted, and the interference of carbon was eliminated simultaneously. Stunted oxidation and reduction current peaks were observed during the following scan loops, suggesting that the electrochemical reduction of LGPS was irreversible.

Fig. 2.

The practical ESW of the LGPS electrolyte and the lithiation behaviors of the Al electrode.

(A) Cyclic voltammograms of the LGPS/Pt-LGPS-Li0.5In cell at a scan rate of 0.1 mV s−1. (B) Lithiation profile of Al in the Al-LGPS-Li0.5In cell at a current density of 0.2 mA cm−2. (C) In situ XRD pattern of the lithiation process of Al and the corresponding lattice parameters of Al and LiAl.

The characteristics of Li-Al alloy were studied to estimate its feasibility as the anode of batteries with LGPS electrolyte. First, the equilibrium potential of Li-Al alloy was measured by the galvanostatic intermittent titration technique test. As shown in fig. S4, after the initial alloying process of Al with Li, the Li-Al alloy presented a stable potential of 0.35 V versus Li/Li+. The moderate operating potential of Li-Al alloy is higher than the decomposition potential of LGPS, guaranteeing the stability of the electrolyte-anode interface. Quantitative lithiation test of Al was carried out to ascertain the most reasonable molar ratio of Li and Al in the Li-Al alloy anode. As shown in Fig. 2B, the potential of LiAl declined when x in LiAl was larger than 0.8. The lower potential may lead to the decomposition of LGPS. Thus, Li0.8Al was identified as the most appropriate composition of the anode, of which the theoretical capacity was calculated to be 793 mAh g−1.

The phase transition of Al during lithiation was characterized by in situ XRD. As illustrated in fig. S5A, the Bragg peaks of Al [Joint Committee on Powder Diffraction Standards (JCPDS), #89-4037] can be detected before lithiation. With the lithiation process, the peaks of LiAl (JCPDS, #65-4905) emerged, and the intensity increased gradually, coexisting with the peaks of Al until the end (Fig. 2C and fig. S5B). These results indicate that Li0.8Al is a mixture of Al and LiAl phases with a proportion of Al:LiAl = 1:4. Thus, the lithiation of Al or delithiation of Li0.8Al is a biphasic reaction with a constant potential, consistent with the flat potential profile in Fig. 2B. Calculation of the lattice parameters of Al (Fmm) and LiAl (Fdm) was performed to measure the volume change of Li0.8Al during cycling (rightmost image in Fig. 2C). The LiAl phase appeared as soon as the lithiation began, as proven by the small peak at 24.253° corresponding to the (1 1 1) crystal face of LiAl (arrow indication in Fig. 2C) arising in the first XRD data of the in situ test. However, the computation of the LiAl lattice parameter had been unable to accomplish until sufficient numbers of peaks (three in the case of LiAl using the program package Jade 6.0) were detected. This condition explains the deletion of LiAl lattice parameters at the first three data points. The calculated lattice parameters of Al and LiAl were 4.04 and 6.33 Å, respectively. The corresponding unit cell volumes were 66.41 Å3 (Al) and 257.26 Å3 (LiAl). The large differences in the lattice parameters and unit cell volumes between Al and LiAl were partially due to the different selection modes of the unit cells. In consideration of the macroscopic symmetry of lattice, the Al unit cell contains one Al atom, whereas the LiAl unit cell contains two Al atoms and two Li atoms. Thus, the actual volume expansion from Al to LiAl was 93.69%. Considering that the molar ratio of Al and LiAl was 1:4 in the Li0.8Al, the volume change rate of the anode during cycling was 74.95%, which was notably lower than those of other alloy anodes such as Li4.4Si (320%) and Li4.4Ge (370%) ().

Interface stability of Li0.8Al-LGPS

Symmetric Li0.8Al-LGPS-Li0.8Al cells were assembled and tested to evaluate the compatibility of the Li0.8Al anode toward the LGPS electrolyte. The Li0.8Al electrode was obtained by attaching the Li foil to the Al foil at a specific stoichiometric ratio. After resting for 6 hours, the silver-white Li vanished, and the shiny Al surface turned dark and dull (fig. S6), indicating the completion of the spontaneous alloying process. Galvanostatic Li plating/stripping tests were conducted to assess the stability of the Li0.8Al-LGPS interface. As shown in Fig. 3A, the Li0.8Al-LGPS-Li0.8Al cell exhibited a low initial overpotential of ~100 mV at 0.5 mA cm−2 and 0.5 mAh cm−2. Then, the overpotential slightly increased to 150 mV (inset figure in Fig. 3A) and maintained excellent stability during the subsequent cycling for over 2500 hours. On the contrary, the Li-LGPS-Li cell was overwhelmed by the high current density of 0.5 mA cm−2, which was manifested by the fact that the overpotential exceeded the detection limit at the beginning of the test. The situation was not satisfactory when the current density was reduced to 0.1 mA cm−2. The overpotential was higher than 3 V initially and underwent a sudden increase to 6 V after cycling for 174 hours.

Fig. 3.

Tests for the compatibility of the Li0.8Al electrode with the LGPS electrolyte.

(A) Galvanostatic Li plating/stripping profiles of the Li-LGPS-Li cell at 0.5 mA cm−2, 0.5 mAh cm−2 (blue) and 0.1 mA cm−2, 0.1 mAh cm−2 (gray), and the Li0.8Al-LGPS-Li0.8Al cell at 0.5 mA cm−2, 0.5 mAh cm−2 (red). Evolution of the electrochemical impedance spectroscopy spectra of (B) the Li0.8Al-LGPS-Li0.8Al cell upon cycling at 0.5 mA cm−2 and (C) the Li-LGPS-Li cell during the rest process. CPE, constant phase angle element. (D) Critical current density test of the Li0.8Al-LGPS-Li0.8Al cell. Scanning electron microscopy (SEM) images of the LGPS surface (E) pristine, (F) after contacting with Li0.8Al for 8 hours, (G) after the Li0.8Al-LGPS-Li0.8Al cell cycling for 100 hours, and (H) after contacting with Li for 8 hours. Corresponding optical images of the LGPS pellet are shown in the insets. (I) S 2p and (J) Ge 3d x-ray photoelectron spectroscopy spectra of the LGPS surface. Top to bottom correspond to the pristine, after contacting with Li0.8Al for 8 hours, after the Li0.8Al-LGPS-Li0.8Al cell cycling for 100 hours, and after contacting with Li for 8 hours. a.u., arbitrary units.

To verify the point that Li0.8Al is the optimized configuration of the Li-Al alloy anode toward the LGPS electrolyte, Li0.9Al-LGPS-Li0.9Al and Li0.7Al-LGPS-Li0.7Al symmetric cells were assembled using the same method, except for the different stoichiometric ratios of Li and Al in the electrodes. Galvanostatic Li plating/stripping tests were carried out under the same condition, and the results are shown in fig. S7. As expected, the Li0.9Al-LGPS-Li0.9Al cell exhibited an initial overpotential of ~150 mV at 0.5 mA cm−2 and 0.5 mAh cm−2, higher than that of either Li0.8Al-LGPS-Li0.8Al (~100 mV; Fig. 3A) or Li0.7Al-LGPS-Li0.7Al (~100 mV; fig. S7). Furthermore, the different cycle stabilities of the symmetric cells illustrate the different anode-electrolyte compatibilities. The overpotential of the Li0.9Al-LGPS-Li0.9Al cell underwent sustained growth, indicating a continuously deteriorating anode-electrolyte interface. In contrast, the overpotential of the Li0.7Al-LGPS-Li0.7Al cell slightly increased at first and then maintained excellent stability during the subsequent cycling, resembling that of the Li0.8Al-LGPS-Li0.8Al cell. Despite the similar stability of Li0.7Al with Li0.8Al, the lower lithium content in Li0.7Al leads to a lower specific capacity. Thus, Li0.8Al is considered as the optimal configuration of the anode of the all-solid-state battery with the LGPS electrolyte.

Electrochemical impedance spectroscopy measurements were carried out to probe into the internal impedance variation of Li0.8Al-LGPS and the Li-LGPS symmetric cells. As shown in Fig. 3B, the Nyquist plot of the freshly assembled Li0.8Al-LGPS symmetric cell consists of a high-frequency semicircle and a subsequent finite-length Warburg impedance (inserted enlarged figure and the equivalent circuit), which is typical of mixed ionic-electronic conductors (). An unalloyed Al foil acted as the mixed conducting interphase (MCI) between the Li foil and LGPS. The Li-ion conducting impedance was 55.15 ohm cm2. After cycling at 0.5 mA cm−2, the Nyquist plots of Li0.8Al-LGPS symmetric cells showed the characteristics of Li-ion conducting SEI, of which the impedance enlarged from 198.77 ohm cm2 after cycling for 10 hours to 292.28 ohm cm2 after cycling for 30 hours, and slightly increased to 314.07 ohm cm2 after cycling for 100 hours, consistent with the gradual stabilization of the overpotential in the Li plating/stripping test. In the case of the Li-LGPS symmetric cells (Fig. 3C), the Nyquist plots present the characteristics of MCI throughout, deriving from the decomposition of LGPS caused by Li. The impedance magnified markedly from 31.88 to 1310.16 ohm cm2 after resting for 12 hours, indicating that the Li-LGPS interface underwent a severe deterioration even without applying current. The large interface impedance was responsible for the high Li plating/stripping overpotential at a low current density. Critical current density (CCD) is the maximum current density that a cell can withstand without failure. Figure 3D shows the CCD profile of the Li0.8Al-LGPS-Li0.8Al cell. With the enlargement of the current density, the overpotential of the cell increased gradually. However, no indication of battery failure emerged when the current density was up to 11 mA cm−2. By contrast, the Li-LGPS-Li cell showed a large overpotential of 2.9 V during the first charge process at a current density of 0.06 mA cm−2, and the overpotential increased out of range at 1.9 mA cm−2 (fig. S8).

Scanning electron microscopy (SEM) and x-ray photoelectron spectroscopy (XPS) were used to study the morphology and the component of the LGPS surface after disassembling the symmetric cells, respectively. As shown in Fig. 3E, the pristine LGPS pellet gained by mechanical compressing had a smooth and dense surface. The morphology of the LGPS surface remained almost unchanged after contacting with Li0.8Al for 8 hours (Fig. 3F). Furthermore, no cracks or voids were observed on the LGPS surface after the Li0.8Al-LGPS-Li0.8Al cell cycling for 100 hours (Fig. 3G). However, the surface of LGPS turned rough and uneven after contacting with Li for 8 hours (Fig. 3H). As shown in the inset optical image, dark regions can be observed on the unsmooth surface. The XPS results suggest that LGPS was strongly reduced by Li, generating Li2S and reduced Ge4+ (Fig. 3, I and J). The mixed ionic-electronic conducting property of the reduction products promoted the continuous decomposition of LGPS and resulted in a steady deterioration of the anode-electrolyte interface, which finally led to the poor Li plating/stripping stability and the low CCD of the Li-LGPS-Li symmetric cell.

Electrochemical performances of all-solid-state LSBs

Sulfur suffers from inadequate electronic and ionic conductivity to induce an electrochemical reaction alone. In addition, SSEs cannot infiltrate into the cathode spontaneously similar to liquid electrolytes. Therefore, the solid-state S cathode requires artificial construction of pathways for electrons and Li ions. A compound consisting of S, CNTs, and LGPS was devised as the cathode material and obtained by heat treatment followed by a high-energy ball milling process (fig. S9). First, S was uniformly mixed with CNTs (fig. S10B) by a two-step heat treatment. Transmission electron microscopy was used to view the micromorphology of the S@CNTs. As shown in Fig. 4A(I), the multiwalled CNTs (MWCNTs) had an average external diameter of 15 to 25 nm, and the average spacing between the adjacent carbon layer was 0.346 nm [Fig. 4A(II)], which is typical of MWCNTs (, ). The electron diffraction pattern in Fig. 4A(III) shows two diffraction rings corresponding to the (0 0 2) and (1 0 0) planes of MWCNTs (). After sintering with S, the average external diameter of the tubes was visibly enlarged to around 25 to 35 nm because of the deposition of S outside [Fig. 4B(I) and (II)]. No crystal lattice [Fig. 4B(II)] or diffraction spots [Fig. 4B(III)] of S was distinguishable, and the XRD of S@CNTs showed weak and broad S peaks (fig. S11), implying that the crystal structure of S had been destroyed after the heat treatment. The mass ratio of S in the S@CNTs was 68.78%, as measured by thermogravimetric analysis (fig. S12). Then, S@CNTs were integrated with LGPS by high-energy ball milling. As shown in Fig. 4C, the obtained S cathode was characterized as a homogeneous mixture with uniform distributions of C, P, and S.

Fig. 4.

Characterizations of the S composite cathode and the electrochemical performances of the all-solid-state Li0.8Al-LGPS-S battery.

Transmission electron microscopy images (I and II) and the corresponding electron diffraction patterns (III) of (A) the MWCNTs and (B) the S@CNTs. (C) SEM images of the pristine S cathode and the corresponding energy dispersive spectrometer spectra of C (green), P (yellow), and S (orange). (D) Cycling stability test and (E) the corresponding discharge-charge curve at the 1st, 10th, 50th, 100th, and 200th cycle of the Li0.8Al-LGPS-S battery with an S loading of 1.07 mg cm−2 at 0.2C. (F) Rate performance of the Li0.8Al-LGPS-S battery. The voltage range was set from 0.9 to 2.4 V for all of the electrochemical tests of the Li0.8Al-LGPS-S battery.

Pairing the well-designed S cathode with the Li0.8Al anode, the all-solid-state LSBs showed excellent cycle stability and rate performances. The Li0.8Al-LGPS-S battery delivered a reversible specific capacity of 1237 mAh gS−1 at 0.2C (1C = 1672 mA gS−1; Fig. 4D) with a loading of 1.07 mgS cm−2. The capacity retention was 93.29% after 200 cycles with a high average coulombic efficiency of 99.96%. Figure 4E shows the corresponding charge-discharge profiles at the 1st, 10th, 50th, 100th, and 200th cycles, indicating the high reversibility of the all-solid-state LSB. In addition, the single-plateau discharge profiles indicate that S underwent a one-step solid-solid conversion reaction to the discharge product Li2S (), fundamentally avoiding the shuttle effect of the polysulfides. For comparison, the Li-LGPS-S battery was tested under the same conditions. As displayed in fig. S13, the specific discharge capacity of the Li-LGPS-S battery was 86.7 mAh gS−1 at the first cycle and decayed quickly, probably because of the severe decomposition of LGPS caused by the Li anode. Figure 4F and fig. S14 display the rate performance of the Li0.8Al-LGPS-S battery. The cell delivered specific discharge capacities of 1362, 1239, 1110, 914, and 514 mAh gS−1 at 0.1, 0.2, 0.3, 0.5, and 1C, respectively. When the rate recovered to 0.1C, the specific discharge capacity was restored to 1364 mAh gS−1, demonstrating the outstanding reversibility of the Li0.8Al-LGPS-S battery. A postmortem characterization of the Li0.8Al anode was carried out to understand the outstanding cycling stability of the Li0.8Al-LGPS-S cell. As shown by the digital photograph in fig. S15A, the Li0.8Al electrode maintains excellent integrity after cycling without pulverization. The pristine Li0.8Al is smooth with the naked eye (fig. S6C), but it shows tiny cracks under an electron microscope, which is probably caused by volume expansion during the alloying process. The Li0.8Al electrode after cycling is covered by a thin layer of granular electrolyte LGPS (fig. S15D), which is difficult to be removed because of the intimate contact of the Li0.8Al electrode and the electrolyte. However, neither dendrites nor pulverization is observed (fig. S15E), showing the superior stability of the Li0.8Al anode.

Battery performances must be evaluated under harsh conditions in consideration of the practical significance. The increase in the S loading will improve the areal capacity of the cell. As exhibited in Fig. 5A, with a high S loading of 3 mg cm−2 and only three times excessive anode, the battery exhibited a reversible areal capacity of 3.458 mAh cm−2 (corresponding to 1149 mAh gS−1) and cycled stably for 100 cycles with a high capacity retention of 86.79%. Furthermore, massively excessive anode will reduce the energy density of the whole cell. To further improve the specific energy of the all-solid-state LSB, we adopted a finite amount of anode with a low N/P ratio of 1.125. As shown in Fig. 5B, the fabricated cell had a high initial specific energy of up to 541 Wh kg−1 on the basis of the mass of S and Li0.8Al, and it was maintained at 477 Wh kg−1 after 12 cycles. The good performances of the all-solid-state LSB with the slightly excessive anode are due to the lithiophilic property of the Li-Al alloy and the great stability of the Li0.8Al-LGPS interface.

Fig. 5.

Electrochemical performances of the Li0.8Al-LGPS-S battery under harsh conditions.

(A) Cycling stability test of the Li0.8Al-LGPS-S cell with a high S loading of 3 mg cm−2 and three times excessive anode at 0.05C. (B) Specific energy and the corresponding discharge-charge curve of the Li0.8Al-LGPS-S cell with an S loading of 1.08 mg cm−2 and 0.125 times excessive anode at 0.06C.

In conclusion, an alloy anode strategy is proposed to resolve the incompatibility issue of the anode-SSE interface. We designed an all-solid-state LSB on the basis of a carbon-free and binder-free Li-Al alloy anode, S@CNTs cathode, and LGPS electrolyte. The high electronic conductivity of Al enabled a carbon-free anode configuration, circumventing the carbon-induced decomposition of the sulfide electrolyte. In addition, the outstanding ductility of Al allowed an easy preparation of the anode film without binder. The Li0.8Al alloy anode had a specific capacity of 793 mAh g−1 and an operating potential of 0.35 V versus Li/Li+. As revealed by the in situ XRD, Al underwent a two-phase reaction with constant potential during lithiation. The appropriate (de)lithiation potential of Li0.8Al was higher than the lower limit of the practical ESW of LGPS (0.125 V versus Li/Li+), guaranteeing the stability of the anode-SSE interface. Thus, the Li0.8Al-LGPS-Li0.8Al symmetric cell operated stably for over 2500 hours at 0.5 mA cm−2 and 0.5 mAh cm−2. SEM and XPS tests confirmed that the decomposition of the LGPS electrolyte was inhibited effectively by the substitution of Li metal with Li0.8Al alloy. The cathode was carefully designed and equipped with continuous electronic and Li-ion pathways. S loaded uniformly on the surface of CNTs by a heat treatment method and then mixed with LGPS electrolyte homogeneously, allowing an efficient electron and Li-ion transport. The Li0.8Al-LGPS-S battery delivered a reversible specific capacity of 1237 mAh gS−1 at 0.2C with a loading of 1.07 mgS cm−2 and operated steadily for 200 cycles with a high capacity retention of 93.29%. With a high S loading of 3 mg cm−2, the cell exhibited a reversible areal capacity of 3.458 mAh cm−2 (corresponding to 1149 mAh gS−1) and cycled stably for 100 cycles. To further improve the specific energy of the battery, we reduced the amount of the Li0.8Al anode to the minimum with a low N/P ratio of 1.125. On this condition, the fabricated cell achieved a high specific energy of 541 Wh kg−1, revealing the bright application prospects of the Li0.8Al-LGPS-S battery system. This work addresses the anode-SSE instability issue in solid-state batteries and provides an applicable scheme for the anode selection of the all-solid-state LSBs. We believe that the Li-Al alloy anode with optimized configuration is applicable to other all-solid-state battery systems, owing to its appropriate operating potential, considerable specific capacity, easy processability, and low cost. In addition, the Li-Al alloy may also serve as a feasible lithium-supplement anode to match state-of-the-art commercial cathodes, including lithium cobalt oxide, lithium-nickel-cobalt-manganese oxide, and lithium-nickel-cobalt-aluminum oxide to further prolong the life span of batteries.

MATERIALS AND METHODS

Material preparation

LGPS/Pt mixture

LGPS (Hefei Kejing Material Technology Co. Ltd.) and Pt powder (Macklin) with a weight ratio of 1:1 were ground for 1 hour to obtain a homogeneous mixture.

S cathode

S@CNTs was obtained by a heat treatment method. S and MWCNTs (Macklin) with a weight ratio of 7:3 were ground for 1 hour. Then, the homogeneous mixture was sealed into a glass vial and heated at 155°C for 5 hours under Ar atmosphere. After cooling to room temperature, the powder was ground for another 30 min and heated at 200°C for 2 hours under Ar atmosphere. The S cathode was obtained by a high-energy ball milling method. S@CNTs, CNTs, and LGPS with a weight ratio of 3:1:4 were ball-milled at 500 rpm for 10 hours under Ar atmosphere.

Material characterization

The thermogravimetric analysis was conducted on differential scanning calorimeter (DSC) (TA SDTQ600, TA Instruments) under Ar atmosphere. The temperature range was from room temperature to 600°C at a heating rate of 10°C/min. XRD was carried out on the PHI 5000 VersaProbe (Ulvac-Phi Co.). The scanning angle range was 10° to 70° or 20° to 80° with a step length of 0.02° per step and a scan rate of 0.4 s per step. SEM characterization was conducted on a Hitachi SU8010 scanning electron microscope. The accelerating voltage was set as 1 kV. XPS measurement was carried out on PHI 5000 VersaProbe-II.

Cell assembly and electrochemical tests

Li-Al batteries and Li-In batteries

Standard 2032 coin-type cells were assembled in Ar-filled glove box with oxygen and water content below 0.01 parts per million (ppm). An Al foil (1-μm thickness, 99.99%; Alfa Aesar) or an In foil with a diameter of 12 mm was used as the working electrode. LiPF6 (1 M) in a mixture of ethylene carbonate (EC)/ethyl methyl carbonate (EMC) [3:7 (v/v)] with 2 weight % vinylene carbonate (VC) was used as the electrolyte. An Li foil (China Energy Lithium Co. Ltd) with a diameter of 12 mm was used as the counter electrode. All the fabricated cells rested for 8 hours before testing. Galvanostatic intermittent titration technique measurements consisted of a current pulse (0.2 mA cm−2) for 1 hour, followed by relaxation for 1 hour. The galvanostatic discharge of the Li-Al battery during in situ XRD characterization was carried out at a current density of 0.2 mA cm−2 and a cutoff lithiation capacity of Li0.8Al.

All-solid-state LSBs

The all-solid-state batteries were assembled in cell molds. The cell mold is composed of two conductive die steel bars and a poly(ether-ether-ketone) cylinder with an internal diameter of 10 mm. First, 100 mg of LGPS was added into the mold and mechanically pressed under 300 MPa for 5 min. S cathode with an S loading of ~1.1 or ~3 mg cm−2 (for high loading test) was distributed evenly on one side of the LGPS pellet. Then, an Al foil (100- or 1-μm thickness) with a diameter of 10 mm was attached to the other side of the LGPS pellet. An Li foil with a molar ratio of Li:Al = 0.8:1 was pressed on the Al foil. The fabricated cells were compressed under 300 MPa and rested for 6 hours before testing. Li-LGPS-S batteries were assembled with the same method except that the Li0.8Al anode was changed to the Li anode. The galvanostatic charge-discharge tests were conducted within the voltage range of 0.9 to 2.4 V. For the Li0.8Al-LGPS-S cell with an N/P ratio of 1.125, the S in the cathode was 0.85 mg, and the Li0.8Al anode was 2.43 mg. The N/P ratio was the ratio of negative and positive capacity. The specific energy was calculated on the basis of the total mass of S and the Li0.8Al.

LGPS/Pt-LGPS-Li0.5In battery

The LGPS/Pt-LGPS-Li0.5In batteries were assembled with the same method as all-solid-state LSBs except the change of the electrodes. The LGPS/Pt (10 mg) mixture was adopted as the working electrode and distributed evenly on one side of the LGPS pellet. An In foil was attached to the other side of LGPS pellet. A Li foil with a molar ratio of Li:In = 0.5:1 was pressed on the In foil. The fabricated cells rested for 6 hours before testing. The cyclic voltammetry measurements were performed at a scan rate of 0.1 mV/s within the voltage range of −0.62 to 2.4 V.

Al-LGPS-Li0.5In battery

The Al-LGPS-Li0.5In batteries were assembled with the same method as the LGPS/Pt-LGPS-Li0.5In batteries except that the LGPS/Pt mixture was changed to an Al foil (1-μm thickness) with a diameter of 10 mm. The fabricated cells rested for 6 hours before testing. The galvanostatic discharge was performed at 0.2 mA cm−2 with a cutoff voltage of −0.62 V.

Symmetric cells

The symmetric cells were assembled with the same method as all-solid-state LSBs except that the S cathode was changed to Li0.8Al or Li. The galvanostatic charge-discharge tests were carried out at 0.5 mA cm−2, 0.5 mAh cm−2, or 0.1 mA cm−2, and 0.1 mAh cm−2 (for the Li-LGPS-Li cell). The electrochemical impedance spectroscopy tests were conducted with a frequency range from 1 MHz to 0.1 Hz and an amplitude of 5 mV.