WANG et al.

Abstract

The rational design of large-area exposure, nonagglomeration, and long-range dispersion of metal nanoparticles (NPs) in the catalysts is critical for the development of energy storage and conversion systems. Little attention has been focused on modulating and developing catalyst interface contact engineering between a carbon substrate and dispersed metal. Here, a highly dispersed ultrafine ruthenium (Ru) NP strategy by double spatial confinement is proposed, that is, incorporating directed growth of metal–organic framework crystals into a bacterial cellulose templating substrate to integrate their respective merits as an excellent electrocatalytic cathode catalyst for a quasi-solid-state Li–O₂ battery. The porous carbon matrix with highly dispersed ultrafine Ru NPs is well designed and used as cathode catalysts in a Li–O₂ battery, demonstrating a high discharge areal capacity of 6.82 mAh cm^–2 at 0.02 mA cm^–2, a high-rate capability of 4.93 mAh cm^–2 at 0.2 mA cm^–2, and stable discharge/charge cycling for up to 500 cycles (2000 h) with low overpotentials of ~1.4 V. This fundamental understanding of the structure–performance relationship demonstrates a new and promising approach to optimize highly efficient cathode catalysts for solid-state Li–O₂ batteries.

INTRODUCTION

Li–O₂ batteries as promising candidates are contributing significantly toward the advancement and development of next-generation high-performance batteries due to their high theoretical energy density (3500 Wh kg⁻¹).^1,2 To resolve the battery safety issues caused by lithium dendrite growth in liquid electrolytes, solid polymer electrolytes are introduced into Li–O₂ batteries because of their good interfacial stability with lithium.^3,4 However, quasi-solid-state Li–O₂ systems still have major limitations, including a high charge overpotential, poor rate capability, low round-trip efficiency, and a short life span.^5–7 To overcome the limitation of high overpotential due to the formation and decomposition of Li₂O₂, major attempts have been made to find effective catalysts to unlock the Li–O₂ technology's potential.^8,9 Noble-metal-based catalysts have long been regarded as the most intriguing cathode catalysts for Li–O₂ batteries since their intrinsic half-filled antibonding condition can confer adequate adsorption strength to reaction intermediates, and thus improve oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) reaction kinetics.^10,11 Various noble-metal-based materials, such as Au, Pt, Pd, Ir, and Ru, are used as efficient cathodes.^12–15

Compared with the bulk metal cluster deposited on the carbon host, nanoparticles (NPs) less than 5 nm in size in the carbon matrix are essential for superior catalytic activities and enhanced durability in heterogeneous catalysis.^16,17 However, to date, far too little attention has been focused on modulating and developing interface engineering between carbon substrates and noble metals to minimize the size of particles, which is an efficient strategy to optimize the local electronic structure.¹⁸ In recent years, several attempts have been made to develop Ru carbon-based composite materials from metal–organic frameworks (MOFs).^19,20 Carboxylates have strong affinities with metal cations, and the addition of Ru metal ions in the synthesis of MOFs leads to the formation of metal–carboxylate complexes.²¹ Using this property, Ru can be uniformly distributed on the metal nodes of the zeolitic imidazolate framework-8 (ZIF-8) skeleton to form a bimetallic ZIF.¹³ At the same time, the Ru metal sites are spatially separated by the 2-methylimidazole bond and the zinc atoms, so the spatial distance between Ru metal sites is large. In the subsequent high-temperature heat treatment for the formation of carbon materials, zinc evaporates when the temperature is >906°C, resulting in the formation of nanocrystalline materials of the target metal Ru.²¹

Although ZIF-8 can separate Ru NPs far apart, MOF particles used as loading substrates usually present a mountain-like disordered accumulation after carbonization.²² Due to the lack of good carbon substrate matrix, the MOF carbon particles are hardly dispersed and connected autonomously in an orderly manner to form a loose structure.^23,24 The dense agglomeration among these MOF particles greatly affects the exposure of highly active sites and the transport of mass and electrons. To resolve these issues, a conductive network that can confine MOF agglomeration from the nucleation site stage has been introduced. The rigid “steel-like” nanofibers and the highly interconnected architecture provide strong support for the well-aligned 3D host, which can be accomplished with carbon nanotubes, carbon nanofibers, graphene, and other materials.^25–28 Compared with other assembled or embedded materials, bacterial cellulose (BC) could be a promising candidate to ensure uniform growth of MOF particles due to its high aspect ratio along with good mechanical properties.²⁹ BC is rich in oxygen-containing groups (hydroxyl groups), which can provide a good environment for highly dispersed nucleation of ZIF-8 by adsorption of metal ions on nanofibers through weak interactions.³⁰ This leads to the uniform assembly and growth of ZIF-8 nanocrystals along the nanofibers of BC.³¹ Through the entanglement of BC fibers, this spatial confinement strategy can promote uniform growth of dispersed MOFs to create maximum metal active sites, increasing the connectivity and the isolation distance between MOF particles while also maximizing the exposure of catalytic metal particles.³²

In this work, we proposed and constructed ultrafine and highly stereoscopic dispersed Ru NPs inside carbon frameworks using a facile double spatial confinement method as the oxygen cathode catalysts for quasi-solid-state Li–O₂ batteries. The nanocrystalline Ru was confined within the nodes of ZIF-8 and was independently controlled in BC with good spatial alignment (denoted Ru-NPC@CBC), which not only maximized the intrinsic activity of Ru but also promoted the catalytic activity by regulating the electronic structure between the carbon and metal interface during the reaction process, resulting in the uniform coverage of Li₂O₂ on the designed cathode by optimized growth pathways. Due to the strong coupling effect, Ru-NPC@CBC outperforms the cathode formed by Ru NPs on MOF-derived N-doped carbon (Ru-NPC) and carbonized BC (CBC), exhibiting a significantly reduced overpotential, large discharge capacity, and long cycling life for quasi-solid-state Li–O₂ batteries. This work inspires fundamental understanding for solid-state Li-O₂ batteries and provides further guidance to design efficient Ru-carbon catalysts.

EXPERIMENTAL SECTION Materials

All regents were analytical-grade quality. Ruthenium (III) 2,4-pentanedionate (Ru(acac)₃, Ru 24%) and 2-methylimidazole (97%) were purchased from Macklin Shanghai Co. Zinc nitrate hexahydrate (Zn (NO₃)₂·6H₂O) was purchased from Aladdin Co. BC was purchased from Guilin Qihongkeji CO. All reagents were used as received without further purification.

Synthesis of material

1.1116 g (3.76 mmol) of Zn (NO₃)₂·6H₂O was fully dissolved in 32 mL of methanol to form solution A. 119.6 mg (0.05 mmol) of Ru(acac)₃ was added to solution A and stirred using a magnetic stirrer until the color of the solution changed to dark red. 1.232 g (15 mmol) of 2-Methylimidazole was fully dissolved in 30 mL of methanol to form solution B. Solution A and solution B were mixed well with magnetic stirring at room temperature and then sonicated for 15 min. BC (2 g) was added to the above mixture solution and incubated at room temperature for 24 h to induce static growth of ZIF-8 on the surface of BC. Later, the ZIF-8-covered BCs were gently washed with methanol. After washing, the resulting suspension was centrifuged at 8000 rpm min^–1 and dried in a freeze dryer for 24 h at −70°C. Finally, Ru-ZIF-8@BC was heated to 900°C under a N₂ atmosphere at a heating rate of 5°C min^–1 and held for 3 h. After cooling to room temperature, the Ru-NPC@CBC was obtained. The Ru-NPC preparation method was the same as that for Ru-NPC@CBC, except that there was no step of addition of BC. The CBC was obtained by directly carbonizing BC following the same pyrolysis procedure.

Characterization

A field-emission scanning electron microscope was used to obtain the scanning electron microscopic (SEM) images of the samples obtained (FESEM, Regulus8230 and SUPRA55). Transmission electron microscopy (TEM, JEM-F200) was used to capture the TEM pictures. The N₂ sorption experiments of the samples were conducted using the micromeritics Tristar II 3020 Instrument. X-ray photoelectron spectroscopy (XPS) data with monochromatized Al Kα X-rays radiation were obtained on a PHI QUANTERA-II SXM. Crystal structures of the materials were obtained by X-ray diffraction (XRD) using Empyrean with Cu K radiation. Raman tests were carried out by LabRaM HP Evolution at an excitation wavelength of 532 nm.

Electrochemical measurements Li−O₂ cell cathode preparation and cell assembly

The synthesized Ru-NPC@CBC, Ru-NPC, and CBC were used as cathode catalysts for Li–O₂ batteries. The catalyst materials and the PVDF binder were mixed in an N-Methylpyrrolidone (NMP) solvent at a mass ratio of 8:1 to prepare a catalyst slurry. Then, they were coated on carbon cloth and dried in a vacuum oven at 80°C for 24 h. Finally, the prepared cathodes were cut into wafers of 10 mm diameter. The electrodes were assembled in a Swagelok-type cell with lithium metal as anodes and a quasi-solid gel polymer electrolyte (GPE) (18 mm diameter) as a separator, and all cell assembly operations were carried out in a glovebox.

Preparation of a GPE

First, a transparent solution A was prepared by dissolving 2.4 g of Lithium bis(trifluoromethane sulfonimide) in 1.6 g of tetraglyme. Then, 4.0 g of NMP was mixed with 1.0 g of poly(vinylidene fluoride-co hexafluoropropylene) and continuously swirling until the mixture became a homogeneous and viscous liquid (Solution B). During the assembly of the Li–O₂ cell, the C solution was prepared in situ by stirring 0.01 g of 2-hydroxy-2-methyl-1-phenyl-1-propanone into 3.0 g of trimethylolpropane ethoxylate triacrylate and storing it in a brown glass bottle. Solutions A, B, and C were mixed in a 4:5:3 weight ratio to form the GPE precursor. Finally, the GPE was cured by UV light irradiation for 30 s.

Electrochemical tests

All electrochemical properties were carried out in a dry chamber filled with pure O₂ gas. Cyclic voltammetry (CV) tests were performed on CHI660D electrochemical workstations at a scan rate of 1 mV s^–1 with a potential window of 2.0−5.0 V versus Li⁺/Li. Room-temperature galvanostatic discharge/charge tests were conducted on the LAND CT2001A electrochemical workstation with different test parameters. The current density ranged from 0.02 to 0.2 mA cm⁻²; the discharge/charge voltage platform was set to 2.0–4.5 V.

RESULTS AND DISCUSSION

The schematic illustration of the preparation of the Ru-NPC@CBC catalyst is shown in Figure 1. Here, a typical MOF (ZIF-8) was selected to be combined with a BC matrix. BC with a certain size and shape was exposed to a solution containing Zn metal ions, followed by the addition of an organic ligand solution. The abundant hydroxyl groups of BC nanofibers created a growth environment identical to that of each MOF group, allowing for the absorption of Zn metal ions on BC nanofibers via weak interactions such as electrostatic interactions. At the same time, ZIF-8 with a cavity diameter of 11.6 Å served as the active site host, and the encapsulated Ru(acac)₃ with diameter of 10.4 Å acted as the guest. The preselected Ru metal ions were in situ trapped into metal nodes of the ZIF skeleton during the nucleation growth of the MOFs due to the similar sizes. After carbonization, the N-doped porous carbons (NPCs) dispersed evenly on the CBC nanofibers; meanwhile, the Zn elements in the ligands were removed by heat treatment, resulting in the spatial confinement of Ru NPs in the nodes of the MOFs. Simultaneously, MOFs were spatially confined in the stereoscopic structure of CBC; a dual spatial confinement strategy was successfully used with a tailorable carbon architecture that showed lightweight and flexibility.

Figure 1. Schematic representation of the preparation procedures for Ru-NPC@CBC

The electron photograph of the BC + ZIF-8 composite aerogel is shown in Figure 2A. BC confers high porosity and mechanical flexibility to the composite aerogels and decreases the aggregation of individual MOF NPs. The microstructure and morphology of Ru-NPC@CBC are shown in Figure 2B–D, and clearly, it is composed of CBC nanofibers and carbonized MOF NPs. The morphology of CBC shows a three-dimensional mesh-like nanostructure in Figure S1. A typical rhombic dodecahedral morphology is obtained after direct pyrolysis of the MOF particles (Figure S2) with an average size of about 200 nm. The original polyhedral shapes of MOF and 3D mesh nanostructures of CBC are still retained in Ru-NPC@CBC after pyrolysis, forming a unique bead-like morphology as shown in Figure 2B,C. The surface of the MOF-based carbon growing in situ on the nanofibers is rougher and the polyhedral prisms are more rounded (Figure 2D) compared with the original MOF-based carbon (Figure S2B). Ru NPs are homogeneously distributed on the MOF-derived carbon interface without self-aggregation in TEM images under different scales (Figure 2E,F). The high-resolution transmission electron microscopy (HRTEM) picture of sample Ru-NPC@CBC (Figure 2G) shows that the size of Ru NPs is about 2 nm. Simultaneously, the inductively coupled plasma (ICP) test also demonstrates that Ru content is 0.725%, as shown in Table S1. For the Wulff structure, based on a theoretical thermodynamic stability model, it was found that 2 nm was the active sites where Ru NPs were exposed to the greatest extent and exhibited the best activity in all sizes of particles.³³ Meanwhile, d-spacings of 0.213 and 0.20 nm are revealed, which correspond to the (100) and (101) planes of hexagonal Ru, respectively. The elemental mapping images (Figure 2H) show that N, O, and Ru are evenly distributed on the sample Ru-NPC@CBC. The typical TEM images of the as-synthesized Ru-NPC are shown in Figure S3. The results reveal that the Ru NPs are likewise well disseminated on the MOF carbon matrix. However, in Figure S3F, the size of Ru NPs is about 10 nm. The elemental mapping in Figure S4 shows that Ru NPs homogeneously disseminated on the carbon matrix, leaving N-rich defects to anchor the Ru NCs. The results demonstrate the controllability of the experiment, leading to a dual spatial confinement strategy for Ru NPs.

Figure 2. (A) Electron photograph of the BC + ZIF-8 composite aerogel. (B–D) Representative SEM images of Ru-NPC@CBC. (E,F) TEM images of Ru-NPC@CBC at different magnifications. (G) HRTEM image of Ru-NPC@CBC. (H) HAADF-STEM and corresponding element mapping images of Ru-NPC@CBC for C, O, N, and Ru elements.

The XRD spectra of ZIF-8 and Ru-ZIF-8 in Figure S5 show the same distinctive peaks ascribed to the pure phase of ZIF-8, implying that Ru does not affect the crystalline structure of ZIF-8, while all of the sharp diffraction peaks of Ru-NPC@CBC and Ru-NPC belong to hexagonal Ru (PDF#06-0664), as shown by the XRD patterns in Figure 3A, which are closely matched with HRTEM data.³⁴ In addition, Ru-NPC@CBC also shows typical (002) and (100) amorphous carbon broadened peaks that reflect the combined superimposed effect of CBC and MOF carbon.²² The peaks at 1342 and 1584 cm^–1 in the Raman spectra (Figure 3B) correspond to the D (disordered carbon and sp³ hybridization) and G (sp² disordered carbon) bands of carbon, respectively.³⁵ The ratio of peak intensity (I_D/I_G) is a measure of the degree of structural disorder.³⁶ Ru-NPC@CBC has an intensity ratio of 1.68 in I_D/I_G, which is higher than those of Ru-NPC (1.49) and CBC (1.42).

Figure 3. (A) XRD spectra of MOF, Ru-NPC, and Ru-NPC@CBC. (B) Raman spectra, (C) N2 adsorption and desorption isotherms, and (D) the corresponding pore size distribution of CBC, Ru-NPC, and Ru-NPC@CBC. High-resolution (E) N 1s and (F) Ru 3p spectra of Ru-NPC and Ru-NPC@CBC.

Due to the etching effects of Ru(acac)₃, during calcination, Ru-NPC@CBC and Ru-NPC have larger Brunauer–Emmett–Teller surface areas of 1049.3 m² g^–1 and 1003.1 m² g^–1, respectively, than CBC (568.9 m² g⁻¹), as presented in Table S2. The N₂ sorption isotherms test and the corresponding pore size distribution using the Barrett–Joyner–Halenda method are used to explore the porosity of the micro/nanoreactors (Figures 3C,D). A combined I/IV-type isotherm confirms the hierarchically porous structure of Ru-NPC@CBC and Ru-NPC with a characteristic hysteresis loop at higher pressure, while CBC only shows the pore size distribution of type IV, which, in combination with the pore size distribution, indicates that mesopores and macropores are mainly present.³⁷ The significant N₂ absorption at low pressure and the visible hysteresis loop suggest that Ru-NPC@CBC and Ru-NPC have a large number of micropores and a certain number of mesopores. Notably, a pore size distribution below 4 nm (Figure 3D) is promising for O₂ transmission and electrolyte infiltration.³⁸

XPS further reveals the chemical composition of the samples. The presence of C, N, O, and Ru elements is demonstrated by the wide XPS spectrum (Table S3 and Figures S7A, S8A, and S9A). Notably, in the high-resolution C 1s and O 1s spectra of CBC in Figure S7B,C, the peak of –OH is ascribed to the stretching of oxygen-containing groups (hydroxyl groups), which further confirms the uniform assembly and growth of ZIF-8 nanocrystals along the nanofibers of BC. In contrast, no –OH is found in the XPS spectra shown in Figures S8B,C and S9B,C of Ru-NPC@CBC and Ru-NPC, demonstrating that BC and ZIF-8 are tightly bound in precursors during the in situ synthesis. The high-resolution N 1s spectra (Figure 3E) of Ru-NPC@CBC and Ru-NPC reveal the presence of three types of N species, corresponding to oxidized-N (403–405 eV), graphitic-N (400.9–401.5 eV), and pyridinic-N (398–398.8 eV).^39,40 Many studies have reported that graphitic N and pyridinic N can act as the main catalytic sites to speed up the ORR process.^40,41 The high-resolution spectra of Ru 3p in Ru-NPC and Ru-NPC@CBC (Figure 3F) can be deconvoluted into two species at about 462.0 and 484.5 eV, which can be attributed to Ru 3p_3/2 and Ru 3p_1/2 of Ru⁰, respectively.^42,43 The weak peaks at about 465.3 and 486.7 eV corresponding to Ru⁴⁺ species can be attributed to Ru 3p_3/2 and Ru 3p_1/2 of Ru^IV, respectively.^34,44,45 Surface oxidation in the air is the main source of Ru^IV.⁴⁶

The catalytic performances of Ru-NPC@CBC, Ru-NPC, NPC, and CBC as electrocatalysts in Li–O₂ batteries are shown in Figure 4. The CV behavior of batteries between 2.0 and 4.5 V at a scan rate of 0.2 mV s^–1 is shown in Figure 4A. A pair of distinct oxidation and reduction peaks can be seen in all catalysts, except for CBC. Ru-NPC@CBC shows a much higher ORR triggering onset potential (~2.8 V) with a high peak current. Ru-NPC presents an onset potential of about 2.6 V in the cathodic sweep, revealing that highly dispersed ultra-fine Ru NPs play a more important role in reducing polarization during the ORR process. Additionally, the anodic peak position of Ru-NPC@CBC is slightly more positive than that of Ru-NPC. The onset potential of OER during the anodic scan is lower, demonstrating that highly dispersed ultra-fine size can maximize the efficiency of Ru catalytic activity in both ORR and OER processes. Meanwhile, the onset potential corresponds closely with the voltage platform in the initial galvanostatic discharge/charge profiles shown in Figure 4B at a current density of 0.02 mA cm^–2 and a cutoff voltage window ranging from 2.0 to 4.5 V. Obviously, the battery with the Ru-NPC@CBC cathode shows a decreased overpotential (1.35 V) compared with batteries with Ru-NPC (1.52 V), NPC (1.73 V), and CBC (1.90 V) cathodes. The improved reduction/oxidation kinetics of Ru-NPC@CBC cathode can be attributed to several factors: as electron-deficient centers, the newly added highly distributed ultra-fine Ru NPs enhance the adsorption of intermediates, hence boosting reduction (discharge) kinetics and the transfer of electrons during the oxidation (charge) process. The greater discharge specific capacity of Ru-NPC@CBC (6.82 mAh cm^–2 at the current density of 0.02 mA^–2) is achieved by reducing the overpotential voltage, which is 1.2-fold that of Ru-NPC (5.59 mAh g^–1), 1.92-fold that of Ru-NPC (3.55 mAh g^–1), and 4.35-fold that of Ru-NPC (1.57 mAh g^–1).

Figure 4. (A) CV curves of catalysts between 2.0 and 4.5 V at 0.2 mV s−1. (B) Initial deep discharge–charge curves of CBC, NPC, Ru-NPC, and Ru-NPC@CBC electrodes at a current density of 0.02 mA cm−2. (C) Discharge−charge curves of Ru-NPC@CBC electrode at different current densities. (D) Cycling curves of the Ru-NPC@CBC electrode with a fixed voltage window of 2.7−4.7 V at a current density of 0.2 mA cm−2. (E) Cycling performance comparison of NPC, Ru-NPC, and Ru-NPC@CBC at 0.02 mA cm–2 with a fixed capacity. (F) Selected discharge and charge profiles of NPC, (G) Ru-NPC@CBC, and (H) Ru-NPC@CBC, and (I) their time–voltage comparison curves of cycling performance with a fixed capacity at 0.2 mA cm−2.

The rate capability of Ru-NPC@CBC catalysts with a limited voltage range of 2.0−4.5 V at different current densities is shown in Figure 4C. Correspondingly, the voltage hysteresis increases significantly from 1.35 to 1.55 V with an increase in the current density from 0.02 to 0.04 mA cm^–2. However, as the current density increases from 0.04 to 0.2, the voltage polarization difference increases slightly. Impressive discharge/charge capacities of 6.31, 5.50, and 4.93 mAh cm^–2 could still be achieved by Ru-NPC@CBC catalysts at 0.04, 0.1, and 0.2 mA cm^–2, respectively. This demonstrates that the ultra-fine Ru NPs catalysts still have strong ORR and OER catalytic performance at high current densities. To evaluate the capacity retention of the Ru-NPC@CBC catalyst at a high current density of 0.2 mA cm^–2, the cycling curve limiting the voltage range from 2.4 to 4.7 V is shown in Figure 4D. When the charging voltage reaches 4.7 V in the first cycle, parasitic reactions such as electrolyte decomposition may occur. Batteries' internal charge-transfer resistance may cause charge polarization degradation and rising overpotentials at high current density. Surprisingly, the capacity of 3.38 mAh cm^–2 in the second cycle and 1.22 mAh cm^–2 in the third cycle can still be maintained after a high potential charging process, proving that this catalyst is highly durable and can withstand a certain degree of overcharging and discharging behaviors.

Cyclic stability is a key indicator for assessing the performance of Li–O₂ batteries. Figures 4E and S10 show the terminal discharge/charge voltages of each cycle and the corresponding selected discharge/charge curves with various cathodes at a current density of 0.02 mA cm^–2. The Li–O₂ batteries with Ru-NPC@CBC as the catalyst delivered a stable long cycle life over 200 cycles (2000 h) with a fixed capacity of 0.25 mAh cm⁻². The overpotential difference remained at 1.4 V during the cycle; in contrast, Ru-NPC and NPC cathodes could only show inferior performance with much higher overpotentials for 150 and 130 cycles. Moreover, Ru-NPC@CBC shows excellent performance at a higher current rate of 0.2 mA cm^–2 and capacity limitation of 0.25 mA h cm^–2 under steady terminal voltages over 500 cycles (Figure 4H), which are longer than those of 200 cycles of Ru-NPC (Figure 4G) and 380 cycles of NPC (Figure 4F). Figure 4I shows their corresponding time–voltage profiles. It is shown that batteries with Ru-NPC@CBC cathodes could maintain stable discharge potentials >2.75 V and charge potentials < 4.05 V and function for more than 700 h, which is superior to batteries with Ru-NPC (500 h) and CBC (250 h). The performances of the Ru-NPC@CBC cathode catalyst are compared with those of several previously reported Li–O₂ battery cathode catalysts, as shown in Table S4. Although the overpotential of the Ru-NPC@CBC catalyst is not impressive enough, it still outperforms most of the catalysts in the Li–O₂ battery systems reported to date.

To investigate the discharge mechanisms occurring in Ru-NPC@CBC and Ru-NPC catalysts, microscopic morphologies of the discharge/recharging processes are characterized. The SEM morphology of the Ru-NPC@CBC-based catalyst in the initial stage, and the first complete discharge and recharge stage with the current density of 0.02 mA cm^–2 are shown in Figure 5A–C. Compared with the pristine material, the surface of Ru-NPC@CBC is covered with a layer of fine NPs about 50 nm in size (Figure S11A) after discharge and disappears after recharging (Figure 5C). The discharge product is stacked closely to form a uniform coating on the whole surface of the cathode. Besides, after cycling for 200 cycles, the shape of the discharge products on the catalyst surface is similar to that of the first cycle discharge product NPs, with the exception that the particle size increases as the number of cycles increases (Figure 5D).

Figure 5. (A) Structural and morphological evolution of the pristine state, (B) after the first full discharge, (C) after the first full recharge, (D) after 200 cycles discharged of Ru-NPC@CBC cathodes, and (E) the first discharge of the Ru-NPC cathode. (F) XRD spectra of Ru-NPC@CBC cathodes at pristine, incompletely discharged, fully discharged, incompletely recharged, and fully recharged states. (G) Schematic representation of Li2O2 deposition on Ru-NPC@CBC and Ru-NPC. (H) The discharge and recharge curves in the first cycle at a current density of 0.02 mA cm–2 and a cutoff specific capacity of 3.6 mAh cm−2. (I) High-resolution XPS Li 1s spectra for the Ru-NPC@CBC cathode at different states during the discharge and charge process.

To reveal the role of highly dispersed ultrafine Ru NPs, the discharge product morphology of Ru-NPC is also investigated. Interestingly, the discharge product shows a typical toroid morphology of about 1 μm (Figures 5E and S11B). Extensive research has revealed that the shape and distribution of Li₂O₂ on the cathode are strongly correlated with OER's limits in charge transport.^15,47 Compared with ultrafine Ru NPs on Ru-NPC@CBC, the bulk counterparts on Ru-NPC induced large sizes of Li₂O₂ on the cathode, eventually leading to electrode polarization. The discharge/charge overpotential and morphology of Li₂O₂ are highly dependent on the adsorption energy of the metal surface. If the surface adsorption energy of LiO₂ is weak, some LiO₂ tends to grow gradually and to form larger Li₂O₂ particles. Suppose that the adsorption energy LiO₂ becomes stronger. In that case, the surface is more likely to have small particle-like Li₂O₂ due to the rapid nucleation of LiO₂, which has a smaller transport resistance and a lower overpotential. Therefore, ultrafine ~2 nm Ru NPs can provide appropriate adsorption strength for reaction intermediates and enhance the reaction kinetics of ORR and OER processes.

The discharge products grown on Ru-NPC@CBC are further confirmed to be Li₂O₂ by XRD in Figure 5F. The diffraction peaks after discharge match well with Li₂O₂ PDF cards (09#0035), which become gradually prominent with the depth of discharge and fade away after full recharge, demonstrating the reversibility of Li₂O₂. Based on the above analysis, we infer that the ultrafine Ru NPs in double spatial confinement can promote the smooth growth of the discharge product in three-dimensional space, which allows film-like discharge products to be formed, as shown in Figure 5G. While Ru particles without restricted size show a toroidal-shaped morphology, which is consistent with the results reported previously.^48,49 The toroidal-shaped discharge products are scattered randomly on the surface. As a result, the large interfacial resistance leads to high discharge/charge polarization and quick failure. As the ultrafine nano-Ru particles excite a large number of active sites, a uniformly distributed discharge product is formed on the Ru-NPC@CBC cathode. The uniformly distributed discharge product increases the contact area with the cathode, which is conducive to the transfer of electrons and Li⁺ during the charging process. The high charge overpotential is reduced and the parasitic reaction caused by high voltage is alleviated; additionally, the impact of the cathode volume can be significantly reduced by the homogeneous distribution of the discharge products, which is beneficial to improve its stability.

To further illustrate the good electrochemistry reversibility of the Ru-NPC@CBC cathode catalyst, the electrochemical impedance spectra (EIS) of Li–O₂ batteries were also investigated at different discharge/charge stages with the frequency range from 10⁵ to 0.1 Hz, and results are shown in Figure S12. In the equivalent circuit diagram, the intercept on the horizontal axis represents the natural impedance (R_ohm) of the battery itself, and the semicircular curve in the intermediate frequency range represents the charge-transfer resistance (R_ct) when the electrode is in contact with the electrolyte interface. The self-impedance of the Ru-NPC@CBC electrode is relatively low at about 150 Ω, which indicates its good electronic/ionic conductivity. After the first discharge, the resistive impedance is significantly increased to ~700 Ω due to the deposition of wide-bandgap insulating Li₂O₂, while after the first recharging process, R_ct can recover to ~400 Ω, which implies that the insulating discharged products are almost completely decomposed on the surface of the Ru-NPC@CBC electrode with favorable reversibility, which ties in with the SEM image after the first recharge (Figure 5C).

Ex situ high-resolution XPS of Li 1s during the different discharge/charge processes with selected pivotal states (Figure 5H) was conducted to investigate the chemical compositions on the electrode surface. Compared with the pristine cathode, the characteristic peak in Li 1 s spectra at the binding energy ~54.5 eV gradually becomes stronger with increasing depth during the 1st discharge process from state I to II (Figure 5I), which can be assigned to the Li₂O₂ discharge product.⁵⁰ In contrast with the previously reported result, the characteristic peaks at ~56.4 eV of Li_2−xO₂ due to the intermediate LiO₂ do not appear.⁵¹ It is further illustrated that the side reaction is suppressed. The peak of Li₂O₂ almost disappears after the 1st recharge process of state III, showing that Ru-NPC@CBC exerts a desirable effect on the decomposition of Li₂O₂. Combined with the analysis of the morphology features of the discharge product on Ru-NPC@CBC, highly dispersed ultra-fine Ru could endow the cathode with an extraordinary electrocatalytic activity, which has higher electronic and ionic mobility, and thus also contributes to the remarkable areal capacity and cycling stability of the Li–O₂ battery.

The comparison of the cycle performance in liquid and solid electrolytes is shown in Figure S13. After 20 cycles, the polarization voltage of the battery in the liquid electrolyte increases significantly, and the discharge platform decreases to about 2.2 V. In addition, after a long-term cycle, a porous lithium electrode with some inactive/dead Li and dendrites is formed, in which the long dendrite can easily lead to short circuiting of the battery. To compare the performance of a GPE and a liquid electrolyte, we disassembled the battery after multiple cycles. After disassembly, we found that the battery failure was not caused by the air electrode but by the lithium metal anode. As shown in Figure S14, in a liquid electrolyte, severe side reactions occurred on the metal anode, and the powdered lithium metal directly broke through the glass fiber separator after several cycles, resulting in the fragmentation of the carbon electrode. On the contrary, the polymer gel electrolyte could protect the lithium metal anode well, and the battery remained intact after 200 cycles of cycling. The voltage gap was reduced and the parasitic reactions from electrolyte decomposition were restrained.

CONCLUSION

In summary, sophisticated and highly active ultrafine Ru NPs on a unique 3D bead-like carbon nanofiber architecture were developed as durable catalysts for solid-state Li–O₂ batteries to achieve both efficient ORR/OER activities and superior electrochemical performance. Compared with pyrolyzed Ru NPC and NPC, the electrode of the Ru-NPC@CBC catalyst has a super high areal capacity of 6.82 mAh cm^–2, low overpotential of only 1.35 V, and superior cycling performance of 2000 h at 0.02 mA cm^–2. Due to the excellent interface between the carbon substrate and Ru NPs, the geometric structure, size, and distribution of the discharge products of Li₂O₂ can be regulated. This superior electrocatalyst has the potential for application in other gas electrodes in advanced electrochemical energy storage systems.

ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (NSFC) through Grant 22179005 and the National Key Research and Development Program of China through Grant 2018YFC1900102. The use of the Swagelok cell was supported by Cunzhong Zhang at Beijing Institute of Technology.

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.