Volume 13, Issue 6 p. 1203-1225
Review
Open Access

An Overview of Engineered Graphene-Based Cathodes: Boosting Oxygen Reduction and Evolution Reactions in Lithium– and Sodium–Oxygen Batteries

Juan Luis Gómez Urbano

Juan Luis Gómez Urbano

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

Department of Inorganic Chemistry, University of the Basque Country UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain

Search for more papers by this author
Dr. Marina Enterría

Dr. Marina Enterría

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

Search for more papers by this author
Iciar Monterrubio

Iciar Monterrubio

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

Search for more papers by this author
Idoia Ruiz de Larramendi

Idoia Ruiz de Larramendi

Department of Inorganic Chemistry, University of the Basque Country UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain

Search for more papers by this author
Dr. Daniel Carriazo

Dr. Daniel Carriazo

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

IKERBASQUE, Basque Foundation for Science, 48013 Bilbao, Spain

Search for more papers by this author
Dr. Nagore Ortiz Vitoriano

Corresponding Author

Dr. Nagore Ortiz Vitoriano

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

IKERBASQUE, Basque Foundation for Science, 48013 Bilbao, Spain

Search for more papers by this author
Prof. Teófilo Rojo

Corresponding Author

Prof. Teófilo Rojo

CIC energiGUNE, Álava Technology Park, C/ Albert Einstein 48, 01510 Miñano, Spain

Department of Inorganic Chemistry, University of the Basque Country UPV/EHU, P.O. Box 664, 48080 Bilbao, Spain

Search for more papers by this author
First published: 24 December 2019
Citations: 19

Graphical Abstract

Future is flat: Graphene is a very versatile and promising air cathode in M–O2 (M=Li, Na) batteries due to its high electronic conductivity, large specific surface area, high mechanical strength, low density, and intrinsic catalytic activity towards oxygen reduction and reactions.

Abstract

The depletion of fossil fuels, the rapid evolution of the global economy, and high living standards require the development of new energy-storage systems that can meet the needs of the world's population. Metal–oxygen batteries (M=Li, Na) arise, therefore, as promising alternatives to widely used lithium-ion batteries, due to their high theoretical energy density, which approaches that of gasoline. Although significant progress has been made in recent years, there are still several challenges to overcome to reach the final commercialization of this technology. One of the most limiting and challenging factors is the development of bifunctional cathodes towards oxygen reduction and evolution reactions. In this sense, graphene, which is very promising and tunable, has been widely explored by the research community as a key material for this technology. Herein, a wide literature overview is presented and analyzed with the aim of guiding future research in this field.

1 Introduction

Energy storage is a critical challenge for modern society, with batteries being the predominant technology of choice. At the end of the 20th century, lithium-ion batteries (LiBs) revolutionized the portable electronic device industry; however, the continuing growth of the LiB market1 exposes it to global manipulation due to limited lithium production sources (e.g., China's expected 62 % share of the world's LiB production capacity will lead to dependency in Europe).2 Consequently, there has been growing interest in research focused on developing “beyond lithium” battery technologies to augment, or in certain situations replace, LiBs. In this scenario, metal–oxygen (M–O2) batteries arise as a great alternative due to their high theoretical energy density compared with that of current systems (100–265 Wh kg−1 versus 3458 Wh kg−1 based on Li2O2 and 1108 Wh kg−1 based on NaO2).3, 4 M–O2 batteries rely on the electrochemical reduction of molecular oxygen (the oxygen reduction reaction (ORR)) at the cathode surface, with the inherent advantage of a negligible amount of “dead weight”. The cycle life of these batteries is the major bottleneck towards real implementation, in which efficient charging through the oxygen evolution reaction (OER) represents a significant challenge. The reactions taking place on rechargeable M–O2 batteries vary with the metal electrodes, the type of electrolytes, the operating parameters, and the cathode material. Generally, in aprotic electrolytes, the metal anode (e.g., Li or Na) is oxidized and ions migrate to the cathode, whereas molecular oxygen is reduced (ORR) to superoxide (O2) or peroxide (O22−) anions at the same time. These oxygen reactive species generated by the ORR combine with the metal cation and the precipitation of metal oxides occurs at the surface of the “air–cathode” during discharge [Eq. 1]. Upon charging, the metal oxides are redissolved to perform the oxidation of the oxygen anions (OER) and subsequent deposition of the metal in the anode [Eq. 2].
urn:x-wiley:18645631:media:cssc201902972:cssc201902972-math-0001(1)
urn:x-wiley:18645631:media:cssc201902972:cssc201902972-math-0002(2)

At the beginning of the 21st century, Li–O2 batteries emerged because of their highest theoretical energy density among all M–O2 batteries. Although great achievements have been made in recent years, several drawbacks related to battery operation, such as lithium dendrite growth, side reactions, the stability of the electrolyte, and poor kinetics, have impeded their practical application.5 In 2012, the substitution of Li by Na emerged as a promising approach to surpass Li–O2 battery limitations.6 First, the substitution of Li by Na presents several advantages, such as low cost, abundance, and higher ionic conductivities. Second, the Na–O2 battery forms the superoxide discharge product (NaO2), which does not occur in the Li system, for which the peroxide product (Li2O2) is preferentially formed. The one-electron oxygen reduction thus conducted in the Na–O2 battery reduces the energy density of the battery, but greatly enhances the kinetics and reversibility of the process by delivering a suitable charge potential (lower overpotential) and coulombic efficiency. Na–O2 batteries, however, are still novel technology that requires further development before widespread commercialization is possible.

Apart from the overall cell performance, both technologies require the utilization of a suitable cathode material that is able to accommodate the maximum amount of discharge products without pore clogging and subsequent cell failure.7 In this sense, different strategies have been followed for the design of the air electrode. Generally, ORR/OER electrocatalysts have been utilized to improve the efficiency of the cathode, of which carbon-based materials have been most studied.5, 8, 9 Porous carbon materials have been largely used as electrodes in different electrochemical storage technologies due to their low cost, high surface area, chemical stability, and high conductivity.10 Among carbon-based materials, graphene has been appointed as a very versatile and promising air cathode in M–O2 batteries due to its high electronic conductivity (≈2000 S m−1), specific surface area (>2600 m2 g−1), mechanical strength, low density, and intrinsic catalytic activity towards the ORR/OER. Consequently, graphene has found applications in different fields, such as energy conversion and storage, electronics, sensors, adsorption, or water purification.11 Among electrochemical energy-storage systems, graphene has been widely used in LiB anodes.12, 13 Graphene can buffer the volume changes that occur during Li insertion;14 moreover, graphene acts not only as the ideal matrix, but also as the support, thanks to its high electronic conductivity and ability to mitigate oxide pulverization.15, 16 Graphene has also been used to mitigate manganese loss from spinel lithium manganese oxide (LMO) cathodes during cycling in LiBs to improve the battery lifetime.17 In addition, graphene confers suitable electrode channels for O2 diffusion, facilitates the impregnation of the electrolyte, and offers a high number of active sites for the formation/deposition and decomposition of the discharge products. However, in practical terms, effective access to the large and active 2D surface of graphene is subjected to careful engineering of both the morphology and the 3D structure. This is extremely important in M–O2 batteries; an optimum air cathode should maximize oxygen diffusion towards the three-phase boundary (i.e., electrolyte/electrode/oxygen interface) and accommodate a large amount of discharge product during the ORR.18 To improve the catalytic activity of graphene in M–O2 batteries, different routes have been explored, such as the introduction of heteroatoms into the graphene structure, the addition of transition-metal catalysts (pure metals or metal oxides), or the nanostructuring of electrode materials for increased exposure of the active sites.

Although numerous parameters play an important role in the ORR/OER kinetics, this review focuses on those related to the cathode materials and, specifically, graphene-based air cathodes in both Li/Na–O2 batteries. Hence, we summarize the great efforts that have been dedicated within the last decade to the development of catalytic materials for overcoming sluggish kinetics during the ORR and OER.19 Therefore, this review has been divided into sections on pure graphene-based materials, graphene–metal composites, and heteroatom-doped or functionalized graphene.

2 Lithium–Oxygen Batteries

2.1 Pure graphene-based materials

Promising properties, as described in the Introduction, have pushed the research community to dedicate their efforts to study the viability of graphene air cathodes in Li–O2 batteries (Table 1).

Table 1. Summary of relevant materials reported in the literature for pure graphene-based cathodes in Li–O2 batteries.[a]

Material

Synthesis

Electrolyte

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles

(discharge rate)

Ref.

expanded rGO paper

vacuum-promoted expansion

0.1 m LiClO4 in DMSO

19 800 mAh g−1

(300 mA g−1)

1000 mAh g−1

50

(1000 mA g−1)

27

GNPs

commercially available

1 m LiBF4 in NMP

1000 mAh g−1

40

(100 mA g−1)

28

rGO paper

Hummer′s method+chemical reduction

1 m LiTFSI in TEGDME

497 mAh g−1

(0.1 mA cm−2)

none

85

(0.1 mA cm−2)

25

3D rGO membrane

Hummer′s method+coating in CP

+freeze drying+thermal reduction

1 m LiNO3 in DMAc

1425 mAh g−1

100

(2800 mA g−1)

29

3D paperlike graphene

templating with polystyrene particles with

commercial GO+thermal reduction

1 m LiNO3 in DMAc

1000 mAh g−1

78

(2000 mA g−1)

30

rGO/activated carbon

hydrothermal carbonization of glucose

on GO+KOH activation

1 m LiCF3SO3 in TEGDME

3.2 mAh cm−2

(0.05 mA cm−2)

0.1 mAh cm−2

170

(0.05 mA cm−2)

34

  • [a] GNP=graphene nanoplatelet, rGO=reduced graphene oxide, CP=carbon paper, GO=graphene oxide, NMP=N-methyl-2-pyrrolidone, LiTFSI=lithium bis(trifluoromethanesulfonyl)imide, TEGDME=tetraethylene glycol dimethyl ether, DMAc=dimethylacetamide.

In this sense, graphene nanosheets (GNSs) showed a much higher capacity than that of commercial carbon black (e.g., BP-2000 and Vulcan XC-72; Figure 1 a),20 which is widely used for the manufacturing of commercial gas diffusion layers. Further studies revealed that GNS cathodes could considerably reduce the first-cycle overpotential, which could arise from the presence of vacancies and defects on the GNS surface that catalyzed the OER.21, 22

Details are in the caption following the image

a) Discharge–charge performance of lithium-oxygen batteries with GNS, BP-2000, and Vulcan XC-72 cathodes. Reproduced from Ref. 20 with permission. Copyright 2011, Royal Society of Chemistry. b) Deep discharge properties of Li–O2 batteries with holey graphene air cathodes: discharge profiles at areal current densities of 0.1, 0.2, and 0.5 mA cm−2 (mass loading of 5 mg cm−2). c) Discharge profiles for air cathodes with mass loadings of 2, 5, and 10 mg cm−2 at the same areal current density of 0.5 mA cm−2. d) Dependence of specific capacity and areal capacity of the batteries on the areal mass loading. e) A schematic drawing showing the utilization efficiency and discharge-product accumulation at the air cathode with various mass loadings. Reproduced from Ref. 26 with permission. Copyright 2017, ACS publications.

Later, Xiao et al. developed an unusual 3D graphene structure that displayed “broken egg” morphology, which was formed of shells containing numerous nanosized pores and large tunnels.23 The hierarchical 3D architecture with interconnected pore channels delivered an extremely high discharge capacity (15 000 mAh g−1). This behavior can be ascribed to 1) numerous large tunnels, which facilitate continuous oxygen flow into the air electrode; 2) small “pores”, which provide ideal three-phase regions for the ORR, and 3) preferential nucleation of the Li2O2 discharge product in the vicinity of lattice defect sites.

The processing of graphene into paperlike cathodes was explored by Kim et al. through the vacuum-assisted technique by mixing GNPs with graphene oxide (GO) and poly(4-styrenesulfonic acid) stabilizers.24 GNP films delivered a large discharge capacity compared with that of rGO papers or commercial diffusion layers (9760 mAh g−1 at 100 mA g−1).

The thickness of the paperlike cathodes was also revealed as a relevant parameter that affected the electrochemical performance.25 These air cathodes were prepared by vacuum filtering different volumes of GO suspensions and further reduction with chemical agents. A greater thickness (achieved by simply adding larger volumes of GO) led to an increase of the interlayer sheet distance and, consequently, to a larger average pore size and surface area. In addition, this material presented the best electrochemical performance; a higher mass improved the electrode stability, as well as electron conduction and oxygen transport between the layers of graphene. The importance of cathode thickness was further revealed by Lin et al., who used binder-free holey graphene materials.26 Thicker cathodes showed a better cycling performance at a high curtailing areal capacity (2 mAh cm−2) due to the oxygen gradient, which led to preferential accumulation of the discharge products along the cathode thickness (Figure 1 b–e).

Apart from morphology, Zhou et al. demonstrated that porosity also played an important role in the performance of graphene cathodes by fabricating a micron-sized graphene matrix with a hierarchical meso-/macroporous structure by using vacuum-promoted thermal expansion.27 Further deoxygenation treatment was used to adjust the surface chemistry by reducing the amount of oxygen and selectively removing partially unstable groups. With this simple methodology, a large discharge capacity (19 800 mAh g−1) over 50 cycles (1000 mA g−1 to 1000 mAh g−1) was achieved. Such enhancement was mainly attributed to the synergetic effect between the hierarchical structure and stable surface chemistry, which provided numerous reaction sites, strengthened reactant transfer, and reduced the formation of byproducts. The importance of the porous structure was also evaluated by Park et al., who studied the electrochemical performance of two types of commercial GNPs with different morphologies: sheetlike, with thin and wrinkled features, and granular, with disordered solid nanoparticles (NPs) of random aggregates.28 Graphene with wrinkled sheets presented a better electrochemical performance (3840 versus 255 mAh g−1) and was stable for 40 cycles, due to a wider pore size distribution, which provided mesoporous channels for suitable electrolyte, oxygen, and Li+ diffusion. The development of well-defined porous structures delivered the most promising results regarding pure graphene-based materials. Hence, Zhong et al. manufactured 3D rGO-based membranes by depositing a freeze-dried GO aerogel onto CP followed by annealing.29 This work demonstrated that the molecular diffusion rate, O2/H2O selectivity, and moisture-resistive behavior could be tuned by controlling both the density and thickness of the graphene gel. The highly tortuous hydrophobic graphene membrane retards moisture diffusion and, thus, boosts cycling performance under ambient conditions. As observed in Figure 2, this material exhibited excellent cycling performance (>2000 cycles under a capacity limitation regime of 140 mAh g−1 and >100 cycles under a cutoff of 1425 mAh g−1) and high capacity (>5700 mAh g−1 over 20 cycles at a high current density of 2.8 A g−1).

Details are in the caption following the image

Galvanostatic cycling of freeze-dried rGO deposited onto CP under a capacity limitation of a) 2850, b) 5700, and c) 140 mAh g−1; and the corresponding cycling profiles (d–f) at a current density of 2.8 A g−1. Reproduced from Ref. 29 with permission. Copyright 2017, Springer Science.

The influence of larger pore size on performance was studied by the groups of Kim and Yu, who prepared 3D macroporous paperlike graphene by using polystyrene colloidal particles as a sacrificial template.30, 31 More than 100 cycles were achieved at a current density of 500 mA g−1 under a capacity limitation of 1000 mAh g−1 due to the tailored macroporosity.

Three-dimensional graphene mesoporous foams also showed improved performance, as further demonstrated by the groups of Liu and Zhang.32, 33 In the first case, the open network structure was obtained by directly growing graphene in aluminum current collectors, which allowed the coexistence of open cages and honeycomb channels that facilitated O2 gas diffusion, while providing a large surface area for the deposition of discharge products.32 In the second study, the electrochemical exfoliation of highly ordered pyrolytic graphite, followed by thermal reduction at different temperatures, resulted in graphene foam cathodes; the best electrochemical performance was attained at 800 °C.33

Another approach to enhance the cathode porosity includes coating of the graphene surface with porous carbon to provide numerous meso-/micropores that act as nucleation sites for the discharge products. Such carbon–carbon hybrids not only avoid graphene restacking, but also provide a large surface area and enhanced conductivity. In this regard, Xin et al. prepared graphene–hydrothermally activated carbon by using glucose and GO precursors, followed by chemical activation with KOH.34 The hybrid nanostructure suppressed Li2O2 grain growth, leading to small particles, compared with 100–200 μm toroidal particles observed on pristine rGO. The benefits of hybridizing graphene-based materials with tailored carbons was demonstrated by a discharge capacity of 1800 μAh cm−2 and an improved cyclability (170 stable cycles). Wang et al. also prepared graphene cathodes with hierarchical porosity through a facile and effective in situ sol–gel method, using resorcinol/formaldehyde cross-linkers as the carbon source and a nickel foam template.35 The hybrid presented an improved performance relative to that of isolated phases, showing discharge capacities as high as 11 060 mAh g−1 (280 mA g−1).

The abovementioned studies show that both porosity and morphology of the air cathode are key to Li–O2 battery performance; an open meso-/macroporous channel structure facilitates the suitable interconnection of graphene sheets, while enhancing Li+ and O2 diffusion and the accommodation of discharge products. Control of the porosity can also have electrocatalyst-like behavior by increasing the exposure of active sites (defects on graphene sheets) and, consequently, reducing the charge overpotential.

2.2 Graphene–metal and graphene–metal oxide composites

The ORR and OER are considered to be kinetically limiting processes due to their sluggish kinetics. The limited catalytic activity of graphene has inspired researchers to develop active electrocatalysts that can compensate for efficiency loss of the cell. Noble metals, first-row d-block metals, and their corresponding oxides have been identified as potential OER catalysts. The use of graphene as a conductive support for metal-based NPs not only contributes to the stability of the catalysts, but also to charge transfer and mass transport. Moreover, the incorporation of metal-based NPs onto the graphene surface avoids restacking of the sheets, which boosts the outstanding properties of this 2D material. Metal catalysts supported on graphene have been, therefore, widely investigated, of which ruthenium is one of the most studied noble metals.

Jung et al. evaluated the influence of anchoring ruthenium-based NPs (metallic Ru or hydrated Ru) onto rGO on the cell performance.36 The hydrated Ru phase presented a lower charge overpotential than that of the metallic phase and delivered 30 cycles (500 mA g−1 to 5000 mAh g−1) due to a lower bonding energy that better catalyzed the decomposition of discharge products. In a later study, they compared the catalytic activity of Ru with that of Pd and Pt37 by synthesizing rGO/metal NP composites through a one-pot modified polyol synthetic method. NPs loaded on rGO helped to suppress the restacking of rGO sheets and the graphene sheets facilitated both a smaller particle size and uniform distribution of the metal particles. The resulting cathode material presented a higher surface area, and hence, accessible active catalytic sites. Pd/rGO, Pt/rGO, and Ru/rGO composites demonstrated a positive effect towards catalyzing the OER, whereas pristine rGO showed the ability to catalyze the ORR. Specifically, the Ru/rGO composite presented the lowest charge overpotential (Figure 3 a) and longer cycle life (30 vs. 15 cycles). The improved catalytic behavior of the Ru/rGO composite can be ascribed to the ability of Ru to induce morphological changes in the discharge products; a thin film of Li2O2 was formed instead of typical big toroids.

Details are in the caption following the image

a) Discharge–charge voltage curves of lithium-air batteries by using noble-metal catalysts supported on a rGO electrode. Reproduced from Ref. 37 with permission. Copyright 2015, American Chemical Society. b) Voltage profiles of Super P, rGO, and rGO/hydrous amorphous RuO2 composite. Reproduced from Ref. 38 with permission. Copyright 2016, Royal Society of Chemistry.

Wu et al. continued to study hydrated amorphous RuO2 (20–100 nm) supported on rGO (Table 2).38 They confirmed the tailored morphology of the Li2O2 discharge product due to the presence of Ru in the composite, which resulted in a pronounced reduction of the first cycle overpotential (Figure 3 b) and an enhanced cycle life in comparison with that of previous studies (55 cycles). Similar performance was attained by Jiang et al., who used self-standing graphene-based aerogels decorated with Ru NPs, which were synthesized by means of a simple hydrothermal method.39

Table 2. Summary of relevant materials reported in the literature for noble-metal-based catalysts anchored on graphene-based materials in Li–O2 batteries.[a]

Metal

catalyst

Material

Synthesis

Electrolyte

(in TEGDME)

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles

(discharge rate)

Ref.

Ru

hydrous amorphous

RuO2/rGO

Hummer′s method

+Ru incorporation by sol–gel method

1 m LiTFSI

1000 mAh g−1

55

(100 mA g−1)

38

Pd

Pd/GNS

simultaneous reduction of GO and Pd

precursor

1 m LiCF3SO3

7680 mAh g−1

(0.08 mA cm−2)

500 mAh g−1

100

(0.08 mA cm−2)

43

Au

graphene/Au NPs/

Au NS sandwich

graphene deposition on Ni foam by CVD

+Au deposition by solution impregnation

1 m LiClO4

3347 mAh g−1

(400 mA g−1)[b]

500 mAh g−1

300

(400 mA g−1)[b]

47

Au/rGO

Hummer′s method

+simultaneous reduction of GO and Au

precursor

1 m LiP6

5230 mAh g−1

(0.1 mA cm−2)

none

120

(0.6 mA cm−2)

48

Ir

Ir/deoxygenated

graphene

modified Hummer′s method

+Ir incorporation by ethylene glycol

reduction

0.1 m LiClO4[c]

1000 mAh g−1

150

(2000 mA g−1)

51

  • [a] CVD=chemical vapor deposition. [b] Capacity and current density calculated on the basis of catalyst only. [c] In TEGDME/DMSO.

Despite of the reasonable performance of Ru-based catalysts, Pt has been considered as the reference metal catalyst for the ORR and OER due to its outstanding catalytic activity. Yang et al. prepared highly dispersed Pt NP–GNS hybrids by liquid-phase pulsed laser ablation,40 which delivered a discharge capacity of 4800 mA g−1 and reduced considerably the first cycle overpotential. However, the high cost of platinum, together with self-poisoning issues derived from the strong adsorption of carbonaceous intermediates on the platinum surface, restricted its early commercialization.

In this sense, palladium has been studied as a potential replacement and has attracted considerable attention because of its high catalytic activity and relatively low cost.41 Ye et al. reported the preparation of alloyed PtPd NPs anchored on the surface of graphene.42 The poisoning resistance of palladium, combined with the high electroactivity of platinum, resulted in an enhanced electrochemical performance. An adequate ratio of Pt to Pd was stablished to be Pt60Pd40, which resulted in a lower first cycle overpotential and the best cyclability (80 cycles at 200 mA g−1 to 1000 mAh g−1) among different alloying ratios.

Wang et al. achieved one of the highest cyclability values ever reported among noble-metal-based catalysts by anchoring platinum NPs on the surface of graphene.43 Highly homogenously dispersed Pd NPs were supported on rGO sheets (Figure 4 a) through functionalization of GO with Pd precursor and subsequent reduction with ethylene glycol. This composite delivered a discharge capacity of 7690 mAh g−1 (at 80 μAh cm−2) and almost 100 stable charge/discharge cycles. They also observed that palladium NPs had the ability to change the morphology of the discharge products to homogenously distributed NPs (Figure 4 b–g), which were easier to decompose during charging.

Details are in the caption following the image

a) Schematic illustration of a Pd-functionalized GNS cathode catalyst for a Li–O2 battery. PDDA=poly(diallyldimethylammonium chloride). SEM images evaluating the evolution of the discharge products for pristine GNSs (b–d) and Pd/GNS (e–g). Reproduced from Ref. 43 with permission. Copyright 2015, Royal Society of Chemistry.

Ye et al. followed a different approach by using metallic Pd nanodendrites supported on graphene NPs.44 A discharge capacity of 3000 mAh g−1 and stability for 30 cycles were obtained (200 mA g−1), with a considerable reduction in the charge/discharge overpotential, in comparison with that of bare graphene electrodes. The branched structure of nanodendrites improved the electrochemical activity, due to the higher surface area and location of catalytically active sites at the corners, edges, and stepped atoms along their branches.

Despite the positive performance of Pd and Pt metals, their high cost requires an investigation of new approaches. The use of much cheaper 3 d transition metals arose as an economically feasible alternative for the possible mass production of future Li–O2 batteries. In this sense, Sevim et al. reported two different studies to alloy Pd45 and Pt46 with Co, Ni, and Cu. These NPs were synthesized through the chemical reduction of metal precursors, which were further dispersed in rGO sheets by liquid self-assembly. Regarding Pd alloys, the NiPd-based cathode delivered the best performance in terms of cyclability. For Pt alloys, PtCo showed 80 cycles (0.15 mA cm−2 to 0.75 mAh cm−2), whereas PtNi and PtCu were more resistant to carbon corrosion. In accordance with previous studies, the use of different Pt alloys induces morphological changes in the discharge products, and hence, influences the subsequent recharge process.

Gold has also been explored as a bifunctional catalyst in several studies in the field of Li–O2 cathodes. Thus, a complex synthetic approach was presented by Wan et al.,47 in which Au NPs were sandwiched between Au nanosheets (NS) and GNSs (Figure 5 a). The resulting material showed a discharge capacity of 3347 mAh g−1 and was stable for 170 and 300 cycles (400 mA g−1) under a capacity limitation regime of 500 and 1000 mAh g−1, respectively (Figure 5 b and c). These excellent results could be ascribed to the design of the electrode, which delayed deactivation of the electrode and reduced contact of Li2O2 with both graphene and electrolyte.

Details are in the caption following the image

a) Schematic illustration of the GNS/Au NP/Au NS electrode on Ni and its working mechanism. Corresponding cycling performance under a capacity limitation regime of b) 500 and c) 1000 mA g−1. Reproduced from Ref. 47 with permission. Copyright 2016, Wiley-VCH.

Kumar et al. reported a simpler synthetic method based on the anchoring of Au NPs on rGO by simultaneous reduction of Au3+ and GO in water with NaBH4.48 This method was also followed in two subsequent studies to evaluate the performance of Ag49 and Ir50 NPs anchored onto rGO sheets. In the case of Au NPs, the first cycle overpotential was considerably reduced and 120 cycles were obtained without a capacity limitation regime. In the case of Ag NPs, even if a higher discharge capacity was retained (11.29 mAh cm−2 at 0.2 mA cm−2), cyclability was limited to only 30 cycles. In addition, a lower coulombic efficiency was obtained compared with that of Au NPs. Finally, the Ir/rGO composite displayed a high discharge capacity (11.36 mAh cm−2) and retained a good coulombic efficiency during cycling.

The incorporation of Ir NPs into deoxygenated 3D porous graphene frameworks was explored by Zhou et al. through vacuum-promoted exfoliation and heat treatment.51 Iridium was incorporated by means of a wet reduction method, leading to iridium NPs uniformly dispersed on the graphene surface. This composite showed excellent cyclability for 150 cycles (2000 mA g−1 to 1000 mAh g−1). The hierarchical framework provided unimpeded transport of oxygen, exposed catalytic sites, and stable surface for the electrochemical reaction, while Ir NPs increased the electrocatalytic activity towards the formation/decomposition of Li2O2.

Despite the great catalytic activity of noble metals, their scarcity and high cost have tremendously limited their practical feasibility. The exploration of earth-abundant-element catalysts without compromising the catalytic performance is still a big challenge, for which research has focused on developing non-noble-metal and metal oxide catalysts.52 Cobalt is the most explored element due to the good performance demonstrated in other technologies, such as LiBs; however, its use is still restrained by its low conductivity and low surface area. Different strategies have been investigated to overcome these drawbacks, such as controlling the morphology of Co NPs or adding conductive graphene-based materials as a matrix to alleviate the insulating nature of cobalt oxides (Table 3).53

Table 3. Summary of relevant materials reported in the literature for transition-metal-based catalysts anchored on graphene-based materials for Li–O2 batteries.

Metal

catalyst

Material

Synthesis[a]

Electrolyte

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles

(discharge rate)

Ref.

Co

Co3O4 nanofibers/graphene

Co3O4 preparation by electrospinning

+mixing with GNF suspension and drying

1 m LiTFSI in TEGDME

10 500 mAh g−1

(200 mA g−1)

1000 mAh g−1

80

(200 mA g−1)

54

Co3O4 NS/graphene NS

modified Hummer′s method

+simultaneous hydrothermal reduction of GO

and Co3O4 precursor

10 528 mAh g−1

(100 mA g−1)[b]

1000 mAh g−1

113

(100 mA g−1)[b]

56

CoCu yolk–shell/graphene

modified Hummer′s method

+simultaneous hydrothermal reduction of GO

and CoCu precursor

1 m LiTFSI in TEGDME

15 000 mAh g−1

(200 mA g−1)

1000 mAh g−1

204

(200 mA g−1)

57

Cu

CuO NW/graphene

simultaneous chemical reduction of

graphene and CuO NW precursor

1 m LiNO3 in DMAc

1000 mAh g−1

120

(1000 mA g−1)

59

Ni

NiO nanoplatelets/graphene

modified Hummer′s method followed by

thermal treatment to produce GNS

+simultaneous hydrothermal reduction of GNS and NiO precursor

1 m LiTFSI in TEGDME

2400 mAh g−1

(500 mA g−1)

none

60

(100 mA g−1)

67

Mn

flowerlike α-MnO2/graphene

graphene deposition on Ni foam by CVD

+deposition of MnO2 on 3D graphene/Ni by

hydrothermal treatment

0.1 m LiClO4 in DME

5230 mAh g−1

(0.083 mA cm−2)[c]

492 mAh g−1

132

(0.33 mA cm−2)[c]

64

α-MnO2/graphene

modified Hummer′s method

+microwave irradiation to GO

+simultaneous microwave reduction of graphene and KMnO4

1 m LiCF3SO3 in TEGDME

5862 mAh g−1

(100 mA g−1)[b]

1000 mAh g−1

50

(100 mA g−1)[b]

63

Ce

Ce0.8Gd0.2O2−δ/rGO

modified Hummer′s method

+simultaneous hydrothermal reduction of GO

and Gd/Ce precursors

1 m LiTFSI in DMSO

15 069 mAh g−1

(300 mA g−1)

600 mAh g−1

110

(400 mA g−1)

70

spinels

MnCO2O4 nanospheres/rGO

simultaneous sonochemical reduction of

graphene and MnCo2O4 precursors

1 m LiTFSI in TEGDME

1000 mAh g−1

250

(800 mA g−1)[c]

73

NiCo2O4 needlelike/rGO foam

modified Hummer′s method

+incorporation of NiCo2O4 by simultaneous hydrothermal

treatment of NiCo2O4 precursors in the presence of

graphene

1 m LiTFSI in DMSO

6000 mAh g−1

(300 mA g−1)

1000 mAh g−1

110

(400 mA g−1)

74

CuCr2O4 needlelike/rGO foam

hydrothermal treatment to synthesize CuCr2O4

+simultaneous reduction of GO and CuCr2O4

1 m LiTFSI in TEGDME

1000 mAh g−1

100

(200 mA g−1)

75

  • [a] GNF=graphene nanoflake. [b] Capacity and current density calculated on the basis of catalyst only. [c] Capacity and current density calculated on the basis of the composite (graphene+catalyst).

Ryu et al. proposed a GNF/1D Co3O4 nanofiber composite bifunctional cathode.54 Co3O4 nanofibers were prepared by electrospinning and, subsequently, dispersed in a suspension of GNFs. A discharge capacity of 10 500 mAh g−1, a high coulombic efficiency in the first cycle (>90 %), and stability over 80 cycles (1000 mAh g−1 at 200 mA g−1) were achieved. The large surface to volume ratio of this 1D nanostructure facilitated a continuous, one-way, electron-transport pathway with numerous reaction sites. Yuan et al. compared the performance of CoO nanocrystals53 with that of Co3O4 nanorods.55 CoO nanocrystals were obtained by simply dispersing a CoO precursor in GO sheets and heating, whereas Co3O4 nanorods were attained through a hydrothermal synthesis followed by thermal treatment. A considerable decrease in the first cycle overpotential was observed, which indicated efficient catalytic activity of cobalt oxides towards the ORR/OER, of which the OER process was particularly favored. It is also notable that the overpotential was lower if cobalt oxides were anchored on graphene-based materials, due to an increase in conductivity and a better dispersion of NPs within the matrix. Even if the discharge capacity of CoO nanocrystals was higher than that of Co3O4 nanorods (14 450 vs. 7600 mAh g−1), the latter showed an improved cycling performance (42 cycles with a capacity limitation of 1500 mAh g−1). Co3O4 NSs were also investigated by Song et al., who used a simple hydrothermal synthesis.56 The authors claimed that graphene not only contributed to the uniform dispersion of the NS catalyst, which provided more active sites for the ORR and OER, but also to the accommodation of insoluble discharge products due the high surface area available. The discharge capacity of the bare catalyst was considerably improved by the presence of the rGO matrix (10 528 vs. 4841 mAh g−1 at 100 mA g−1). Moreover, the first cycle charge overpotential and cyclability were improved by the utilization of this composite, for which stability over 113 and 28 cycles were attained (1000 and 3000 mAh g−1, respectively). It is worth noting that the carbon content in the electrode was only 6 wt %, which reduced parasitic reactions and poisoning of the catalyst, and thus, enhanced the cycling life.

As an alternative to oxide-based catalysts, cobalt alloyed with other metals has also been explored. Chen et al. prepared copper particles covered by a cobalt shell supported on graphene through a simple hydrothermal method (Figure 6 a).57 The synergistic effects of Co, Cu, and graphene in the final bimetallic composite are observed in Figure 6 b; a high discharge capacity (15 000 mAh g−1 at 200 mA g−1) and good cyclability (Figure 6 c) were achieved (122 and 204 cycles at a cutoff voltage of 2.5 and 2 V, respectively; 1000 mAh g−1 at 200 mA g−1).

Details are in the caption following the image

a) Schematic illustration of the CoCu/graphene electrode and its working mechanism. b) Deep charge–discharge voltage profiles for graphene, Co/graphene, Cu/graphene, and CoCu/graphene. c) Galvanostatic charge/discharge profiles for the cycling performance of CoCu/graphene. Reproduced from Ref. 57 with permission. Copyright 2015, Royal Society of Chemistry.

The group of Wang et al. reported the preparation of 2D nanocomposites by anchoring an Al/Co hybrid layered double hydroxide onto rGO.58 The presence of graphene sheets in the nanocomposite reduced the charge/discharge overpotential by minimizing hydroxide particle agglomeration, while providing sufficient pathways for oxygen/electrolyte/electron transport and enough space for the deposition of the discharge products. The Li2O2 particles nucleated uniformly around the layer oxide sheets, and thus, prevented overgrowth of the discharge product and subsequently favored the recharging process.

Copper metal oxide in the shape of nanoleaves and nanowires was also explored by anchoring CuOH into graphene sheets with final reduction to CuO at low temperature.59 The presence of the catalyst allowed the reduction of the overpotential from 1.58 to 1.39 V and was stable over 120 cycles (1000 mAh g−1 at 1000 mA g−1).

The use of manganese oxides has emerged due to a favorable compromise between electrocatalytic activity, cost, ease of preparation, and environmentally friendliness. Débart and co-workers systematically studied the catalytic performance of Mn2O3, Mn3O4, and MnO2 (α, β, γ, and λ phases).60 Among them, the α-MnO2 catalyst exhibited excellent round-trip efficiency, a high specific capacity, and good cycling stability.

Numerous studies have reported the hybridization of MnO2 with graphene due to the poor conductivity of this oxide; graphene acts as a conducting agent without compromising the effectiveness of the catalysts. For instance, Cao et al. studied the performance of α-MnO2 anchored on rGO with 80 or 92 wt % of the transition-metal oxide catalyst.52 Composites of α-MnO2/rGO, yielding different morphologies of the metal oxide (nanorods, nanowires, and mixtures), were obtained by the spontaneous redox reaction of rGO and KMnO4. The morphology of the catalyst was a key point in the electrochemical performance of the hybrids, since the mixture of nanowires and nanorods delivered the lowest first cycle overpotential and highest discharge capacity. A low fraction of graphene in the composite led to a considerable reduction in parasitic reactions and longer cycle life. α-MnO2 nanowires/GO composites were also synthesized and electrochemically characterized by Cetinkaya et al.61 The catalyst was obtained by a simple hydrothermal synthesis, further mixed with GO, and cast onto a Ni foam. Although a high discharge capacity was obtained (5570 mAh g−1 at 0.1 mA cm−2), only 22 cycles were achieved under a capacity limitation regime of 1 mAh cm−2 and a current density of 0.1 mA cm−2.

Although α-MnO2 is the most studied allotrope (due to its accessibility and easy preparation), other structures have also been investigated. In this regard, Yang et al. anchored γ-MnO2 clusters onto rGO through pyrrole-assisted synthesis, through which a high discharge capacity of 11 235 mAh g−1 (75 mAh g−1) was attained and stability over 30 cycles was delivered.62 The combination of δ-MnO2 (birnessite-type manganese oxide) with graphene demonstrated the best performance in terms of cyclability. In this regard, Wang et al. prepared δ-MnO2 by an in situ, fast, and environmentally friendly microwave route.63 Discharge capacities of 5862 and 3566 mAh g−1 were delivered at 100 and 600 mA g−1, respectively, and stability over 50 cycles was attained (100 mA g−1 to 1000 mAh g−1). However, among all reported studies, the best results were obtained by Liu et al., who used a binder-free air electrode composed of 3D graphene deposited onto Ni foam with flowerlike δ-MnO2.64 Graphene was deposited onto 3D Ni foam by means of CVD for the subsequent growth of ultrathin NSs of flowerlike δ-MnO2 through a facile hydrothermal route. The presence of δ-MnO2 decreased the charge overpotential compared with that of pure graphene (4.2 vs. 4.5 V) and increased the discharge plateau up to 2.8 V. Stability over 130 cycles was obtained (0.333 mA cm−2 to 492 mAh g−1), which could be ascribed to the efficient deposition of graphene on the skeleton of the Ni foam, and thus, favoring oxygen and electron pathways without blocking the pores. The flowerlike NS morphology in δ-MnO2 supplies a large number of efficient catalytic sites and, at the same time, provides enough space to accommodate the insoluble discharge products generated during cycling.

Li et al. prepared well-distributed particles of Mn3O4 catalyst over the surface of graphene sheets by reduction of manganese acetate in the presence of rGO.65 Mn3O4/rGO composite was able to deliver a discharge capacity of 16 200 mAh g−1 (50 mA g−1), but only 20 stable cycles were obtained under a regime without capacity limitation at the same current density.

Other earth-abundant elements with reduced cost and high efficiency, such as Fe, Ni, Ce, Gd, and Zr, have also been investigated as potential replacements for noble-metal catalysts. Zhang et al. proposed an interesting approach in which Fe2O3 nanocluster-decorated graphene composites were obtained by simultaneous electrochemical exfoliation.66 This composite delivered a discharge capacity of 8290 mAh g−1 (100 mA g−1); however, only 30 cycles were achieved (200 mA g−1 to 1000 mAh g−1).

Qiu et al. explored the performance of Ni metal by the in situ growth of NiO nanoplatelets on GNS through the thermal decomposition of Ni(OH)2;67 this cathode delivered a discharge capacity of 2400 mAh g−1 and 60 cycles (50 mA g−1) under a regime without capacity limitation, with a specific capacity of about 1200 mAh g−1 in the 60th cycle. An additional study on Ni was performed by Zhu et al., who reported the preparation of a composite consisting of NiO/Ni–graphene foam by a simple hydrothermal method and subsequent thermal treatment.68 The homogeneous distribution of NiO particles provided a discharge capacity of 25 986 mAh g−1 (100 mAh g−1). Such a high discharge capacity was ascribed to the ability of NiO particles to lower the reaction barrier, together with the 3D network structure of the graphene foam, which exposed more active sites and speeded up electron transfer.

Ceria (CeO2) has also arisen as an attractive catalyst due to its high redox activity and ability to enhance the reactivity of other elements. Cerium-doped with rare-earth elements has demonstrated to have a tremendous effect on increasing the kinetics of the ORR and OER. Hence, the catalytic activity of Zr- and Gd-doped ceria NPs was explored by the groups of Ahn69 and Jiang,70 respectively. Zirconium-doped cathodes had a more pronounced effect on the ORR, whereas gadolinium had a major effect on the OER. Regarding cycling stability, only 14 cycles were achieved for Zr-based catalysts (1 mA cm−2 to 500 mAh g−1), whereas Gd proved to be more effective by delivering 110 stable cycles (400 mA g−1 to 600 mAh g−1). The uniform dispersion of CeO2/Gd particles on the rGO surface promoted the deposition of Li2O2 nuclei and Gd+3 stabilized CeO2 NPs; thus, the overpotential gap is reduced to 0.89 V.

In the last five years, numerous research studies have been devoted to the use of double-metal oxides (specifically spinel ternary oxides). Cao et al. reported the preparation of CoFe2O4/rGO composites by a simple hydrothermal method.71 The resulting composite delivered a discharge specific capacity of 2116 mAh g−1 at a relatively low current density (50 mA g−1). The authors observed that the presence of CoFe2O4 led to a considerable reduction of the first cycle overpotential and an improvement in the coulombic efficiency, delivering 30 cycles under a capacity limitation regime of 173 mAh g−1 at 50 mA g−1.

Among different spinel-based materials, XCo2O4-based (X=Zn, Ni, Mn, etc.) structures have been widely investigated, thanks to their ability to enhance the kinetics of the ORR and OER. Kim et al. developed nanowire-shaped MnCo2O4 anchored on rGO sheets to increase the spinel conductivity.72 The 1D catalyst was obtained from the pyrolysis of MnCo2(C2O4)3 nanowires and subsequently mixed with rGO to achieve the final composite. Despite the designed nanowire morphology, only 35 cycles and a discharge capacity of 2218.4 mAh g−1 at a relatively low current density (50 mAh g−1) were achieved. In contrast to this poor result, Karkera et al. reported one of the highest performances regarding metal-based electrocatalysts by using MnCo2O4 nanospheres.73 This advanced air cathode was prepared by dispersing both manganese and cobalt acetate in a basic suspension of graphene and applying ultrasound frequencies to promote the formation and grafting of the spinel into the graphene support (Figure 7 a). The resulting composite delivered 250 stable cycles under a capacity limitation regime of 1000 mAh g−1 at a relatively high current density (800 mA g−1). It is worth noting that this performance is really close to that of Au NPs47 and the graphene content in the composite is only 5 wt %. The porous channels in MnCo2O4 nanospheres facilitated oxygen intake into the bulk material along the graphene surface, which allowed a larger exposure of active sites for the ORR and OER; thus, the cyclability was improved due to slower saturation of the cathode surface. The cyclability values achieved by Jiang et al. with Ni metal as a substitute element into the spinel lattice were also close to those reported for noble metals.74 A needlelike NiCo2O4/rGO self-standing composite was obtained following the synthetic procedure represented in Figure 7 b.

Details are in the caption following the image

a) Schematic illustration of the preparation of MnCo2O4 nanosphere/graphene sheet composites. Reproduced from Ref. 73 with permission. Copyright 2018, Wiley-VCH. b) Graphene foam coated with NiCo2O4 electrodes. Reproduced from Ref. 74 with permission. Copyright 2017, Wiley-VCH.

The cathode delivered a discharge capacity of 6000 mAh g−1 (300 mA g−1) and was stable for more than 100 cycles under a capacity limitation regime of 1000 mAh g−1 (400 mA g−1). The high catalytic activity of the spinel was confirmed by the reduction of the first cycle overpotential compared with that of pure graphene material and it resulted from the formation of a filmlike discharge product.

Similar cyclability results were obtained by Liu et al., who proposed the use of CuCr2O4 needlelike NPs anchored onto the surface of rGO.75 The composite was obtained by a simple hydrothermal method of the spinel precursor in the presence of GO (10 wt %) followed by thermal treatment. The authors claimed that the use of rGO as a conductive support provided a suitable distribution/connection of Cu-substituted spinel NPs, which led to increased electronic conductivity, provided pathways for Li+ transport, and buffered volume changes during cycling. The composite was stable for 100 cycles under a capacity limitation regime of 1000 mAh g−1 (200 mA g−1).

2.3 Heteroatom-doped or functionalized graphene

Metal-free graphene catalysts have been explored as promising alternatives to the use of precious metals. Doping graphene with heteroatoms, such as nitrogen, sulfur, or oxygen, alters the physicochemical properties of the materials and creates structural defects, which favor the ORR/OER activity (Table 4).

Table 4. Summary of relevant materials reported in the literature for doped/functionalized graphene-based materials.

Material[a]

Synthesis

Electrolyte

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles (discharge rate)

Ref.

N-doped defective GNSs

annealing of commercial Vor-X in an open furnace

+reduction in NH3 or H2 atmospheres at high temperature

1 m LiCF3SO3 in TEGDME

17 400 mAh g−1

(100 mA g−1)

800 mAh g−1

110

(40 A g−1)

80

3D N-doped graphene nanocages

hydrothermal treatment with thiourea and GO as precursors and polystyrene particles as templates

+freeze-drying+high-temperature annealing

1 m LiTFSI in TEGDME

10 081 mAh g−1

(200 mA g−1)

1000 mAh g−1

72

(300 mA g−1)

81

S-GNSs

CVD in Ni foam catalyst with thiophene as a precursor

1 m LiTFSI in TEGDME

10 081 mAh g−1

(200 mA g−1)

1000 mAh g−1

300

(300 mA g−1)

82

N-,S-codoped porous carbon/graphene hybrid

hydrothermal treatment of thiourea and sucrose+KOH activation

1 m LiTFSI in TEGDME

11 431 mAh g−1

(100 mA g−1)

500 mAh g−1

100

(500 mA g−1)

84

PEDOT microflower/GNS composites

surfactant-assisted polymerization

0.5 m LiTFSI/0.5 LiNO3 in TEGDME

1500 mAh g−1

150

(400 mA g−1)

86

  • [a] PEDOT=poly(3,4-ethylenedioxythiophene).

Li and co-workers studied the catalytic behavior of nitrogen-76 (N-GNSs) and sulfur-doped77 GNSs (S-GNSs) in comparison to that of pristine GNS, for the first time. N-GNS presented a higher average discharge plateau and an increase of about 40 % in the discharge capacity (11 660 mAh g−1 at 75 mA g−1), whereas S-GNS presented improved behavior during charging. The discharge product morphology was greatly influenced by chemical doping; aggregated particles were observed on GNS cathodes (Figure 8 a), highly distributed particles were observed on N-GNSs (Figure 8 b), and large nanorods grew on the surface of S-GNSs (Figure 8 c). The authors stated that the strong interaction between the intermediate products and pyridinic/quaternary nitrogen groups on the surface of the graphene promoted the nucleation of Li2O2 particles, while thiophene/thioether-like functionalities promoted better charging behavior.

Details are in the caption following the image

SEM images of a) pristine GNS, b) N-GNS, and c) S-GNS. Reproduced from refs. 76, 77 with permission. Copyright 2012, Elsevier and Royal Society of Chemistry. d) TEM image of the defective GNS and e) a comparison of the discharge/charge curves of defective and nondefective N-GNS. Reproduced from Ref. 80 with permission. Copyright 2016, American Chemical Society.

The influence of N and S doping was also studied by Bae et al.,78 who reported that S-doped electrodes presented a high proportion of sulfonates. The discharge capacity of the doped cathodes was clearly enhanced and the S-doped cathode delivered the best performance (1980 mAh g−1 at 0.1 mA cm−2). The enhanced ORR catalytic behavior of both N- and S-doped electrodes was attributed to additional defect sites that promoted nucleation of the discharge products. The obtained results suggested that inorganic sulfonates and pyridinic groups catalyzed the ORR, whereas organic C−S−C-like functionalities facilitated the OER. The effect of oxygen atoms on the electrocatalytic performance of graphene-derived air electrodes was reported by the group of Storm, who varied the oxidation time of graphite to obtain GO sheets oxidized to different extents and reduced them either with hydrazine or by thermal treatment.79 The use of highly oxidized GO precursor and thermal reduction resulted in the best electrochemical performance. This cathode yielded the highest proportion of carbonyl, carboxyl, and lactol functionalities, as well as the highest structural order, with a specific capacity of 59 792 mAh g−1. Samples reduced with hydrazine, however, yielded the highest proportion of carbonyl/carboxyl species, while providing the best cycling behavior and lowest charging voltage. The presence of polar C=O bonds served as Li2O2 nucleation sites and introduced an increased oxide coverage of the cathode. The impact of chemical doping on the structural order (i.e., creation of defects) of graphene-derived materials was investigated by Shui et al. and further correlated with the performance in LiO2 batteries.80 A highly defective graphene (Figure 8 d) was prepared by the oxidation of commercial graphene powder and further thermal treatment. Both the rGO defective sheets and GO were doped with nitrogen to study the impact of nitrogen atoms on their structures. The importance of defects on the cathode surface during cycling was highlighted by the much lower reduction overpotential of the doped holey carbon compared with that of holey undoped carbon (0.45 vs. 0.99 V; Figure 8 e). The N-doped defective graphene electrodes exhibited a high efficiency (85 %) and long cycling life (>100 cycles, 800 mAh g−1 at 40 mA g−1).

Several groups have also investigated the synergetic effect between the catalytic activity of heteroatoms and mass transport provided by porous structures. Three-dimensional N-doped interconnected graphene nanocages (Figure 9 a) were prepared by Zhao et al. by using polydopamine/polystyrene sphere templates and hydrothermal/freeze-drying self-assembly of GO sheets.81 The electrode, annealed at 1000 °C, delivered a specific capacity of 10 081 mAh g−1 at 200 mA g−1 (Figure 9 b) and 72 cycles at a high current density of 300 mA g−1 with no capacity loss (Figure 9 c). The enhanced kinetics were ascribed to 1) the presence of meso-/macropores, which are beneficial to the rapid transport of O2 and Li+; 2) the high specific area, which contributes to the triphase (solid–liquid–gas) regions; and 3) full exposure of the pyridinic nitrogen active sites.

Details are in the caption following the image

SEM image of a) N-doped graphene nanocages (N-doped graphene aerogel (NPGA)); b) initial discharge/charge curves; and c) discharge capacity/discharge voltage as a function of the number of cycles for the N-doped graphene nanocages, unstructured graphene (graphene aerogel (GA)), and Super P. Reproduced from Ref. 81 with permission. Copyright 2015, Wiley-VCH. SEM image of d) 3D nanoporous electrode prepared by CVD; e) initial discharge/charge curves for pristine GNSs, N-GNSs, and S-GNSs; and f) cycling performance of 3D S-doped cathode. Reproduced from Ref. 82 with permission. Copyright 2016, Wiley-VCH.

Han et al. prepared 3D undoped and N- and S-doped nanostructured graphene by means of CVD (Figure 9 d).82 The capacity delivered by undoped and S-doped electrodes was very similar, about 4700 mAh g−1, but the capacity was as high as 10 400 mAh g−1 (Figure 9 e) for N-doped graphene, as a consequence of surface chemistry rich in graphitic and pyridinic-like groups.

Sulfur-doped graphene not only exhibited the lowest charge overpotential among the three studied samples but was also stable for 300 cycles at 1000 mAh g−1 (Figure 9 f). The monitoring of the surface chemistry by means of X-ray photoelectron spectroscopy (XPS) revealed that the C−S−C species degraded gradually, while inactive sulfates remained after 300 cycles. Hence, organic C−S−C-like species were definitively identified as active species for the OER. He and co-workers also prepared 3D N-doped graphene through a facile hydrothermal process and subsequent freeze-drying/annealing of the resulting hydrogel.83 The N-doped sample presented a well-defined meso-/macroporous structure with surface graphitic, pyrrolic, and pyridinic functionalities. The N-doped aerogel delivered a discharge capacity of 7300 mAh g−1 at 50 mA g−1 and a higher voltage plateau than that of the undoped material. Heteroatom doping provided better cycling stability (21 cycles at 100 mA g−1) with no evident overpotential increase during cycling. Kim et al. coated the surface of GO sheets with porous carbon that was simultaneously doped with N and S through a hydrothermal carbonization approach.84 The specific surface area and mesoporous volume of the codoped graphene-based sample were very large compared with that of the undoped counterpart. Pyridinic-like groups were the prevailing species for nitrogen atoms, whereas sulfur was predominantly bound in aromatic carbon structures. The synergistic effect of the codoping of S and N heteroatoms and the presence of hierarchical porous structures led to a high initial discharge capacity of 11 431 mAh g−1 and good cycling stability (100 cycles at 500 mAh g−1).

The manufacture of graphene/conductive polymer composites is also a versatile alternative to tailor the chemistry of graphene-based electrodes. Hence, Selvaraj et al. grew polypyrrole (PPy) on the surface of rGO sheets;85 the discharge capacity of the composite was considerably improved compared with that of pristine rGO (3358 vs. 911 mAh cm−2 at 0.3 mA cm−2), as well as the charge overpotential (1.06 vs. 1.41 V). The cycling stability of the composite material was tested by subjecting the cell to 25 discharge–charge cycles at 0.5 mA cm−2 to about 2500 mAh g−1. Yoon and Park covered the surface of commercial graphene with PEDOT microflowers by controlled polymerization of the ether-/thiophene-containing polymer on a graphene suspension; however, this material showed insufficient capacity due to the low conductivity of the polymer.86 To increase the conductivity of PEDOT microflowers, different proportions of graphene (5–20 wt %) were added to the polymer. The composite yielding 5 wt % graphene delivered the lowest overpotential and the best cycling performance by retaining 150 cycles with a constant capacity. It was proposed that redox-active conducting polymers presented a high catalytic activity for the formation–dissociation of Li2O2, while suppressing side reactions triggered by high-surface-area carbon materials, such as graphene.

The catalytic performance of heteroatom-doped graphene is summarized in Figure 10. The functional groups that induce positive charge density on the carbon backbone enhance the adsorption of O2 molecules onto the graphene surface. The proximity between the O2 molecules and electrons delocalized throughout the π-conjugated system of graphene increases charge transfer to reduce O2 to O2. Hence, pyridinic-like functionalities induce polarization in adjacent carbon atoms because nitrogen presents higher electronegativity than that of carbon.

Details are in the caption following the image

Mechanisms affecting the catalytic performance of heteroatom-doped graphene compared with that of undoped graphene.

On the other hand, quaternary nitrogen generates positive charges in the π-conjugated system; nitrogen bears one electron less in the valence band than carbon and can easily accommodate in the lattice due to a similar atomic size. Regarding oxygen and sulfur, their atomic size is larger than that of carbon, but they can be incorporated as functional groups at the edge of the carbon network. These terminal groups can either induce positive charge by resonance or favor the formation of defects by bending the planar structure of the carbon structure, which also enhances the catalytic performance of graphene. Finally, conducting polymers, such as PPy and PEDOT, also present a positive charge when oxidized to the bipolaron state, which favors the adsorption of O2 molecules and subsequent ORR during charging. On the contrary, aromatic sulfur groups (such as thiophene and thioethers) present π orbitals with polarizable electron pairs, which can decrease the bonding of O2 species to the cathode surface, leading to a low passivation of the cathode throughout charging.

2.4 Multifunctional O2 electrode: Combining catalysts and heteroatoms in graphene nanostructures

In the previous sections, efforts to combine graphene with metal-based catalysts or heteroatoms have been analyzed. In this section, both strategies are combined to develop bifunctional air cathodes able to surpass the limitations mentioned in previous sections (Table 5). Wu et al. designed a composite consisting of a 3D hollow network support built from graphene nanocages, in which they deposited homogenously ultrasmall Pt particles.87 The use of nanostructures, based on an interconnected network of cages formed by less than five layers of graphene, presented superior cycling performance to that of traditional carbon black. The addition of Pt nanocatalyst decreased the charge overpotential, due to the smaller size of the generated discharge products and its more amorphous nature.

Table 5. Summary of relevant materials reported in the literature of multifunctional catalysts based on graphene for Li–O2 batteries.

Metal

catalyst

Material[a]

Synthesis

Electrolyte

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles

(discharge rate)

Ref.

Noble metals

PdFe alloy

PdFe/N-rGO

modified Hummer′s method

+N doping by annealing in ammonia atmosphere

+simultaneous chemical reduction or N-rGO and Fe/Pd precursors

10 818 mAh g−1

(200 mA g−1)

1000 mAh g−1

400

(400 mA g−1)

41

Ru

Ru–FeCoN/rGO

simultaneous heat treatment of rGO, Fe/Co

precursors and melamine

+Ru incorporation by heat treatment in presence of Ru precursor

1 m LiTFSI in TEGDME

23 905 mAh g−1

(200 mA g−1)

1000 mAh g−1

300

(500 mA g−1)

89

Ir

IrO2/Co–N-rGO

simultaneous heat treatment of rGO, Co

precursor, and melamine

+IrO2 incorporation by hydrothermal treatment

1 m LiTFSI in TEGDME

11 431 mAh g−1

(200 mA g−1)

600 mAh g−1

200

(200 mA g−1)

90

Transition metals

Ti

Fe–TiO2/N-GO

pyrolysis of GO in the presence of melamine

+colloidal precipitation of ferrate and titanate

+annealing at mild temperature

1 m LiTFSI in TEGDME

13 500 mAh g−1

(200 mA g−1)

1000 mAh g−1

45

(300 mA g−1)

96

Fe

Fe–Fe3C/3D porous N-GO

hydrothermal synthesis of Fe–C

+ball milling with melamine

+high-temperature pyrolysis

1 m LiTFSI in TEGDME

7150 mAh g−1

(0.1 mA cm−2)

800 mAh g−1

30

(0.1 mA cm−2)

100

Fe−N−C/graphene sponge

pyrolysis of iron phthalocyanine at 450 °C

1 m LiPF6 in TEGDME

5300 mAh g−1

(50 mA g−1)

1600 mAh g−1

50

(400 mA g−1)

101

Co

Co-,N-MWCNT

thermal treatment of PANI catalyzed by Co

species and MWCNTs

1 m LiPF6 in TEGDME

3700 mAh g−1

(50 mA g−1)

50

(400 mA g−1)

104

biphasic N-doped Co@graphene

thermal treatment of Co 2-methyl imidazole crystals

+thermal treatment in N2/NH3 atmosphere

1 m LiCF3SO3 in TEGDME

5.98 mAh cm−2

(0.1 mA cm−2)

1 mAh cm−2

30

(0.1 mA cm−2)

105

Co3O4/N-graphene

3D graphene grown on Ni foam by thermal treatment

+ammonia evaporation induced method in the presence of Co3O4 precursor

1 m LiTFSI in TEGDME

19 133 mAh g−1

(200 mA g−1)

1000 mAh g−1

70

(200 mA g−1)

107

Ce

CeO2/N-rGO

hydrothermal treatment of GO and ammonium hydroxide

+incorporation of CeO2 by sol–gel method

+hydrothermal process

1 m LiTFSI in TEGDME

18 866 mAh g−1

(200 mA g−1)

1000 mAh g−1

40

(400 mA g−1)

108

Spinels

NiCo2O4

flowerlike NiCo2O4/N-GO

modified Hummer′s method

+simultaneous hydrothermal reduction of GO, ammonium

hydroxide, and Ni/Co precursors

1 m LiTFSI in TEGDME

15 046 mAh g−1

(200 mA g−1)

1000 mAh g−1

50

(200 mA g−1)

112

  • [a] PANI=polyaniline.

As discussed in Section 2.2, the high cost of Pt triggered the exploration of cheaper metal catalysts, such as Pd 3 d transition-metal-based alloys. Leng et al. analyzed the effect of doping with different metals in the PdM alloy (M=Fe, Co, NI) by using a nitrogen-doped rGO support.41 Palladium resulted in a lower charge overpotential, whereas PdCo and PdNi alloys showed a more limited catalytic activity due to metal dissolution upon cycling. PdFe alloy/N-rGO composite showed excellent cyclability (400 cycles at 400 mA g−1), as observed in Figure 11 a and b. This was attributed to the bifunctional nature of this cathode, in which the alloy was active during the OER, while the graphene material catalyzed the ORR. The use of Ru NPs on multielement (N-Fe-Co) codoped graphene results in interesting catalytic properties and excellent cycling behavior (Figure 11 c, d). The introduction of different elements within the graphene network results in a higher number of defects and provides a greater amount of graphitic carbon.89

Details are in the caption following the image

a) Plots of the terminal voltage of the batteries and b) specific capacities versus cycle number with PdM/N-rGO (M=Fe, Co, Ni) catalysts. Reproduced from Ref. 41 with permission. Copyright 2017, Elsevier. The charge–discharge cycles of Li–O2 cells with Ru–FeCoN/rGO (c) and the corresponding specific energy efficiency (d). Reproduced from Ref. 89 with permission. Copyright 2015, Royal Society of Chemistry.

Multielement N–Co graphene matrix has been used as a support for IrO2 NPs. The high surface area of graphene, enhanced ORR activity due to Co and N, and the high dispersion of IrO2 NPs improved the capacity retention (reversible capacity: 11 731 mAh g−1).90 Nazarian-Samani et al. studied Ru quantum dots on N-doped holey graphene; they observed the positive effect of Ru as the OER catalyst and its role in the creation of structural defects.91 In the formation of Ru/graphene composites, it was also possible to add other carbonaceous nanostructures, such as single-walled carbon nanotubes (SWNTs), to stabilize the cathode structure.92 Initially, a cross-linked gel of SWNTs–single-layer graphene was synthesized by using an ionic liquid that was extracted in a controlled manner. This material presented an acceptable cycling stability (75 cycles), although after the 25th cycle the accumulation of Li2CO3 resulted in an increase in the overpotential. By renewing the atmosphere with O2, it was possible to recover the air electrode up to 100 cycles. Another effective strategy exploited by Guo et al. focused on encapsulating RuO2 NPs in nanoporous N-doped graphene.93 The RuO2 catalyst contributed to improving the kinetics of the OER with a low charge potential (<4.05 V), and the fact of being encapsulated favored a uniform dispersion in the composite.

TiO2 has been also proposed due to its lower price, high chemical stability, and high catalytic activity.94, 95 Yang et al. doped this material with Fe3+ cations to modify the electronic state, and thus, increase the electronic conductivity of TiO2. In this way, Fe3+-doped TiO2 became a p-type semiconductor with a greater amount of O vacancies, which translated into a positive effect on the catalytic activity.96 The Fe–TiO2/N-doped graphene composite presented an important catalytic activity that gave rise to a lower overpotential (0.83 V). In addition, the cell presented a good capacity retention during 5 cycles and acceptable cyclability up to 45 cycles. TiO2@graphene composites have also been tested as high-power LiB electrodes with a surface amorphization that facilitates the diffusion of lithium.97 Other authors focused their studies on the development of Fe-based catalysts supported on N-doped graphene. Li et al. used a cobalt-containing metal–organic framework (MOF) as a template to prepare transition-metal/N/C ORR catalysts (Figure 12 a, b).98 The composite included graphene nanotubes produced by means of CVD with Ni nanowire templates. The ratio of pyridinic to quaternary nitrogen is related to the catalytic behavior and could be adjusted by controlling the temperature of the CVD treatment. The sample heated at 1000 °C yielded the highest amount of pyridinic nitrogen and presented the greatest catalytic activity during the ORR.99 The use of MOF templates for producing carbon-based nanocomposites with unique morphologies was also the strategy followed to prepare Fe/Fe3C–graphene composites.100 With this MOF-based template, and melamine as a source of C and N, it was possible to obtain 3D porous N-doped graphene with a very high proportion of meso-/macropores. This morphology avoided the irreversible phase-to-face restacking of graphene and was stable for 30 cycles. Another way to increase the cyclic performance is to assemble Fe−N−C composites on a 3D structure of a graphene sponge (Figure 12 c).101 In this case, graphene formed an optimal framework for the dispersion of catalyst NPs, and thus, a clear effect was observed during the OER that seemed to be related to its special molecular geometry. The [Fe-4 N] species, if bound to the carbonaceous support, maintained a planar coordination, which allowed Fe to present two free orbitals for bonding with oxygen from the LiOOLi molecule, without breaking the O−O bond. Park et al. developed a composite based on N-doped exfoliated graphene, together with α-MnO2 nanotubes,102 for which the introduction of N atoms increased the ORR kinetics (Figure 12 d).

Details are in the caption following the image

Representative TEM images of N–Fe MOF/graphene catalyst, showing a typical graphene tube (a); the open mouth of a graphene tube is marked by a yellow arrow (b). Reproduced from Ref. 98 with permission. Copyright 2014, Wiley-VCH. c) Photograph of the graphene sponge after being moistened with DMSO solvent. Reproduced from Ref. 101 with permission. Copyright 2014, Elsevier. d) SEM image of MnO2 nanotube/N-doped thermally exfoliated graphene composite electrode. Reproduced from Ref. 102 with permission. Copyright 2013, Electrochemical Society. SEM (e) and TEM (f) images of Co N-doped graphene composite. Reproduced from Ref. 103 with permission. Copyright 2015, Wiley-VCH.

Wu et al. analyzed cobalt-based catalysts to obtain highly graphitized carbon nanostructures from PANI heteroatom polymer.104 This cathode presented stable cycling during 20 cycles, although from the 30th cycle a loss of capacity was detected. This loss is caused by the accumulation of discharge products that are not decomposed during charging, which block the active sites. Tan et al. resorted to the use of cobalt-containing MOFs for the production of N-doped Co/graphene multicapsules.105 In this composite, the presence of N and Co atoms in graphene provided a greater number of active sites, together with a large porosity, which improved the electrocatalytic activity. Another strategy included the preparation of graphene decorated by Co@CoO NPs, to which the addition of core–shell particles had a clear effect on increasing the capacity and cyclability (up to 70 cycles at 100 mA g−1).106 Cobalt oxides are the most actively investigated materials in multifunctional graphene-based catalysts. For instance, CoO and Co2O3 hollow NPs have been used to manufacture composites on N-doped graphene (Figure 12 e,f).103 Although the catalytic effect of cobalt is not very clear, the addition of an n-type carbon dopant, such as nitrogen, leads to a certain disorder in the carbon nanostructures, which facilitates the ORR. An investigation of 3D graphene–Co3O4 NSs supported on Ni foam resulted in improved kinetics of the ORR and OER through the presence of Co.107 More than 60 stable cycles were obtained; however, it was important to control the distribution of Co3O4 due to its restricted electronic conductivity. A design based on 1D nanostructures seems to be the most suitable in this type of system because it facilitates a continuous pathway for electron transport, in addition to providing numerous accessible active sites due to the high surface to volume ratio.

CeO2 is an alternative redox catalyst that can promote both the ORR and OER. NPs of this material can be dispersed uniformly over N-doped graphene to give a composite with an acceptable cycling performance during 40 cycles; from cycle 20th, however, the charge potential increases through the accumulation of byproducts (lithium formate, acetate, and carbonate, mainly).108 This material has also been tested in the form of nanoflakes, with the observation of a certain ORR activity, but poor cyclability.109 In contrast, the ORR activity of ZnO nanofibers on graphene was limited and only 10 cycles were delivered at 210 mA g−1 with a small decrease in the overpotential.110

Among other systems studied, molybdenum sulfides in the form of NSs on highly porous rGO aerogel have been investigated.111 By adjusting the component proportion in MoSx/rGO to 1:2, it was possible to considerably increase not only the capacity, but also the cycling stability. However, the addition of MoSx had no noticeable effect on the ORR or OER. In addition, the response of the composite at high current densities (0.5 mA cm−2) was very poor. Spinel compounds have also been deposited on multifunctional matrixes. Thus, nanostructured NiCo2O4 in the form of a chrysanthemum flower was combined with N-doped rGO to obtain a bifunctional catalyst.112 The addition of rGO sheets to the composite favored four-electron transfer during the ORR, as well as the spinel phase promoting greater OER kinetics as a result of the presence of mixed-metal valences. In this way, a greater capacity due to an increase in the number of active sites was obtained, but there was no relevant improvement in the decrease of the cycling overpotential. In addition to the spinel structure, perovskite-type materials are also very attractive in the design of catalytically active compounds. Kim et al. utilized GNPs as a support for Nd0.5Sr0.5CoO3−δ nanorods.113 This composite presented a clear synergy between both materials in which the OER activity, which was superior to that of the Pt/C reference material, was provided by spinel particles, while improved ORR kinetics was achieved by the use of graphene as a support.

3 Sodium–Oxygen Batteries

Research into graphene-based air electrodes has emerged due to the promising features of this tunable material for Na–O2 technology (Table 6). The use of graphene was first examined by Liu et al., who achieved a discharge capacity as high as 9268 mAh g−1 (200 mA g−1) with GNSs; this value was almost three times higher than that of typical gas diffusion layers.114 The performance of N-GNSs was studied and compared with that of pristine GNS by Li et al. They found that the presence of nitrogen in the carbon network resulted in an excellent electrocatalytic activity towards the ORR, delivering a discharge capacity almost two times greater than that of undoped GNS (8600 and 4350 mAh g−1 at 75 mA g−1, respectively).115 As reported for Li–O2 batteries, this improvement can be assigned to the introduction of defective sites by nitrogen doping and the ability of nitrogen atoms to tailor the morphology of discharge products, resulting in the formation of small and uniformly distributed NaO2 particles. Furthermore, the effect of a hierarchical macroporous framework on the electrochemical performance of rGO electrodes was reported by Liu et al.116 The 3D arrangement of cathodes showed an improved areal capacity compared with that of traditional carbon electrodes (12 vs. 3 mAh cm−2) due to a dense and continuous growth of discharge products. The large size of mesopores, relative to that of mesoporous materials, prevented pore clogging and, consequently, early cell failure. Also, Enterría et al. studied the influence of porosity on electrochemical performance by using rGO aerogels prepared by freeze-drying and thermal reduction of GO suspensions.7 The porosity and orientation of the graphene sheets was tuned by adjusting the freezing temperature of the GO suspensions. The best electrochemical performance was achieved for the lowest freezing temperature (−196 °C), delivering a discharge capacity of 6.61 mAh cm−2 (100 mA g−1) and 40 cycles under a capacity limitation regime of 0.5 mA cm−2 (Figure 13 a). Hence, a random orientation of the graphene sheets, originated by sudden freezing, minimized restacking of the graphene sheets during self-assembly. This specific 3D arrangement enhanced the accessibility of the reactants towards the active sites in defects and edges, and provided much better performance than that of other aerogels (Figure 13 b).

Table 6. Summary of relevant materials in the literature for graphene-based materials for Na–O2 batteries.

Material

Synthesis

Electrolyte

Discharge capacity

(discharge rate)

Capacity

limitation

Cycles

(discharge rate)

Ref.

GNS

Hummer′s method

+casting GO on stainless-steel mesh

+freeze-drying

+annealing at mild temperature

0.25 m NaPF6 in DME

9268 mAh g−1

(200 mA g−1)

1200 mAh g−1

10

(300 mA g−1)

114

N-GNS

Hummer′s method

+annealing at high temperature

0.5 mol dm−3 NaSO3CF3 in DEGDME

6000 mAh g−1

(75 mA g−1)

1150 mAh g−1

3

(75 mA g−1)

115

porous rGO

coating Ni foam with graphene colloids

0.25 m NaClO4 in DME

12 mAh cm−2

(0.1 mA cm−2)

1 mAh cm−2

17

(0.1 mA cm−2)

116

Pt/GNS

in situ growth of Pt particles on GO by

hydrothermal treatment

1 m NaClO4 in PC

7574 mAh g−1

(0.1 mA cm−2)

100 mAh g−1

10

(0.1 mA cm−2)

117

Ag-rGO

simultaneous reduction of Ag precursor

and GO using ethylene glycol

1 m NaPF6 in TEGDME

566 mAh g−1

(0.1 mA cm−2)

0.125 mAh cm−2

30

(0.2 mA cm−2)

88

rGO aerogels

freeze-drying of commercial GO

+annealing at high temperature

0.1 m NaClO4 in DME

6.61 mAh cm−2

(100 mA g−1)

0.5 mAh cm−2

40

(0.25 mA cm−2)

7

Details are in the caption following the image

a) Cycling performance of rGO foam (ArGO_N) under a capacity limitation regime of 0.15 mAh cm−2, and b) evolution of discharge capacity (left axis, points denoted as filled symbols) and coulombic efficiency (right axis, points denoted as open symbols) with the number of cycles for graphene materials with different porosity (film, ArGO_U, and ArGO_N). Reproduced from Ref.  7 with permission. Copyright 2018, Royal Society of Chemistry.

The synergetic effect between graphene-based materials and transition-metal catalysts has also been explored for Na–O2 batteries to improve the kinetics of both the ORR and OER. Zhang et al. reported the incorporation of Pt NPs on GNSs by a simple hydrothermal method.117 The high electrocatalytic activity of the Pt/GNS composite delivered a high discharge capacity of 7574 mAh g−1. Kumar et al. anchored Ag NPs onto a graphene matrix by simultaneous chemical reduction of GO and AgNO3.88 The composite yielded discharge capacity values of 566, 136, and 87 mAh g−1 at 0.1, 0.2, and 0.3 mA cm−2, respectively. NaO2, Na2O2, and Na2O were identified as stable discharge products and discharge capacities of around 130 (0.13) and 70 mAh g−1 (0.06 mAh cm−2) were attained over 30 cycles.

4 Summary and Outlook

Metal–O2 batteries have the capability to play an important role in the development of next-generation energy-storage systems, due to their high theoretical energy density relative to that of current systems (e.g., LiBs). To date, the full potential of these batteries has not yet been achieved due to challenges associated with the metal anode, air cathode, and electrolyte. One of the main challenges to make these systems commercially available is the sluggish kinetics of the ORR/OER. The rational design of the air electrode can, therefore, reduce losses due to these reactions.

At this point, graphene stands out to lead a new generation of batteries with enhanced performance, thanks to its outstanding properties, such as large surface area, high electronic conductivity, and high electrochemical stability. In the last 15 years, the majority of research has been conducted at the laboratory scale, with the aim of implementation in real applications. However, the commercialization of graphene materials has increased enormously from the laboratory scale to several companies producing this material. The market leaders are those working in corrosion and wear-resistant coatings; applications in energy storage have yet to be implemented. The next step will come from a need from the commercial sector in a suitable application (e.g., electric vehicles), and graphene producers will have to meet the demands of the graphene material, depending on the use.

Herein, enormous progress on the utilization of graphene-based cathode materials for Li and Na–O2 batteries has been summarized and the limitations regarding these cathodes discussed. Regarding Li–O2 batteries, the use of pure graphene has been widely investigated with the aim of tuning its porosity and morphology to enhance oxygen diffusion and provide active sites for the nucleation of discharge products. However, the OER kinetics are still sluggish and graphene–metal composites with Ru, Pt, Ni, Co, Cu, Pd, Au, and Ir have been investigated. Nevertheless, the high price of these materials has led to research into earth-abundant elements, such as cobalt, aluminum, and manganese oxides and spinel compounds. In addition, the introduction of heteroatoms and further doping with metals and/or oxides into the graphene structure has led to interesting results. Heteroatoms, of which the most studied and efficient systems are those based on nitrogen, act as active centers for the ORR/OER and, at the same time, generate defects in the graphene network. Although modified graphene results in major improvements during the ORR, its effect on OER is very limited. To overcome this drawback, a subsequent strategy was based on the addition of elements (noble metals, oxides, etc.) that favored the OER.

Regarding Na–O2 batteries, research is less extensive due to the novelty of the technology (first reported in 20126). Work has focused on pure graphene materials based on GNSs and rGO, as well as nitrogen and platinum doping, similar to research reported for Li–O2 batteries. However, there is still a long path to explore in this field.

It is important to highlight the optimal stability of graphene during cycling, in contrast to commonly used carbon-based materials, which present poor stability at high current densities. This is ascribed to the 2D structure of graphene and its outstanding electronic conductivity. The formation of parasitic products under these conditions is kinetically limited, which provides graphene with excellent charge/discharge cycling stability. The selection of an optimal electrode design, and the proper formulation of the graphene/catalyst composite, lead to a superior catalytic activity, which results in a substantial decrease of the overpotential and enhanced cycling performance. For future research in this field, we herein present some perspectives to boost the practical application of these devices. The ideal air electrode must present desirable bifunctional catalytic activities towards the ORR and OER. Graphene, thanks to its planar structure, facilitates the impregnation of the electrolyte and the diffusion of oxygen through its channels. In addition, the large number of active sites favors electrochemical reactions, mainly the ORR. Regarding the OER, the decomposition of as-generated discharged products can be promoted by an effective electrocatalyst, with those related to spinel structure being the most promising. In this sense, a deep understanding of the effect of different catalytic phases on the graphene surface will enable a rational design of high-performing air cathodes that address the challenge of rechargeability by improving the kinetics of the ORR/OER and suppressing side reactions. It is necessary to perform advanced characterizations to elucidate the reaction mechanisms by means of ex situ and, in particular, in situ techniques. Thus, structural and physicochemical changes that occur upon charging and discharging can be understood. The engineering of future graphene-based cathodes should be oriented to the control of the growth and decomposition of discharged products and the three-phase boundary reaction areas. For this purpose, a combination of ex situ, in situ, and operando techniques needs to be undertaken to understand how the formation of defects and vacancies in graphene affects the nucleation and subsequent decomposition of the discharge products. Multifunctional graphene cathodes are potential electrocatalysts in M–O2 batteries. A maximized efficiency during the ORR and OER can be achieved by further exploration of the synergistic effect between the 3D structure, porosity, defects, and catalyst particles/atoms on the surface of the graphene-based air cathodes. It is necessary to develop realistic models that correlate the morphology of the material and its formulation with cell performance.

Finally, realistic processing and environmentally friendly and low-cost preparation methods should be considered in the design of graphene-based electrodes with a view to future scalability.

Acknowledgements

This work was financially supported by the European Union (Graphene Flagship, Core 2, grant number 785219), the “Ministerio de Economía y Competitividad” of Spain (under project no. MAT2016-78266-P), the “Fondo Europeo de Desarrollo Regional” (FEDER), and the Eusko Jaurlaritza/Gobierno Vasco (under project no. IT1226-19). J.L.G.U is very thankful to the “Ministerio de Ciencia, Innovación y Universidades” for a FPU (16/03498) grant.

    Conflict of interest

    The authors declare no conflict of interest.

    Biographical Information

    Nagore Ortiz-Vitoriano is an Ikerbasque Research Fellow and metal–air research line manager at CIC energiGUNE (Spain). She obtained her doctorate in 2011 for her work on solid oxide fuel cells (UPV/EHU, Spain), during the course of which she undertook research stays at Risø DTU (Denmark) and Imperial College London (UK). In 2013, she was awarded a Marie Curie IOF from the EU, which enabled her to join the Massachusetts Institute of Technology (USA), where she worked on metal–air batteries. In 2015, she continued at CIC energiGUNE, where she conducted research stays at Oak Ridge National Laboratory (USA) and Deakin University (Australia).

    Biographical Information

    Teófilo Rojo received his PhD in chemistry from the University of the Basque Country (Spain) in 1981. He became Full Professor of Inorganic Chemistry at the UPV/EHU in 1992. His research has been focused on solid-state chemistry and materials science. Since 2010, he has been Scientific Director of the CIC energiGUNE, developing materials research for advanced batteries. In 2015, he was appointed an academic member of the Royal Spanish Academy of Exact, Physical and Natural Sciences, and in 2016 he was named a member of the working party on chemistry and energy of EuCheMS (European Chemical Science).