Graphene oxide: A promising nanomaterial for energy and environmental applications

Graphene oxide (GO), the functionalized graphene with oxygen-containing chemical groups, has recently attracted resurgent interests because of its superior properties such as large surface area, mechanical stability, tunable electrical and optical properties. Moreover, the surface functional groups of hydroxyl, epoxy and carboxyl make GO an excellent candidate in coordinating with other materials or molecules. Owing to the expanded structural diversity and improved overall properties, GO and its composites hold great promise for versatile applications of energy storage/conversion and environment protection, including hydrogen storage materials, photocatalyst for water splitting, removal of air pollutants and water puri ﬁ cation, as well as electrode materials for various lithium batteries and supercapacitors. In this review, we present an overview on the current successes, as well as the challenges, of the GO-based materials for energy and environmental applications. &


Introduction
As the global concerns in the development of human civilization, the scientific and technological issues of energy utilization and environment protection are currently facing challenges. Nowadays, enormous energy demands of the world are mainly met by the nonrenewable and environmental unfriendly fossil fuels. To replace the conventional energy platform, pursuit of the renewable and clean energy sources and carriers, including hydrogen storage, lithium batteries, and supercapacitors, has become extremely urgent. Additionally, long-term industrial and agricultural activities induce serious environmental pollution (such as greenhouse and toxic gases, heavy metal ions and organic species) in air and water to deteriorate the ecological balance and the daily human health.
GO is a monolayer of graphite oxide, which can be obtained by exfoliating graphite oxide into layered sheets through sonicating or mechanical stirring [1]. The graphene-based lattice and existence of various oxygen-containing groups (mainly epoxy and hydroxyl groups) enable GO abundant fascinating properties. First, the functional groups on GO surface act as effective anchoring sites to immobilize various active species. Furthermore, GO possesses tunable electronic properties. Typically, GO is insulating due to the large portion of sp 3 hybridized carbon atoms bonded with the oxygencontaining groups, which results in a sheet resistance of $10 12 Ω sq À 1 or higher [2]. However, after reduction, the sheet resistance of reduced GO (namely, RGO) can be degraded by several orders of magnitude, hence transforming the material into a semiconductor or even into a graphene-like semimetal [3]. It has been demonstrated that band gap of GO can be tailored by controlling coverage, arrangement, and relative ratio of the epoxy and hydroxyl groups [4][5][6][7][8].
Besides, GO also displays excellent optical and mechanical properties for a wide landscape of applications. The optical transmittance of GO films can be continuously tuned by varying the film thickness or the extent of reduction [9]. Generally, a suspension of GO films in water is dark brown to light yellow, depending on the concentration, whereas that of reduced graphene oxide (RGO) thin films (with a thickness less than 30 nm) is semitransparent [10]. The optical absorption of GO is dominated by the π-π n transitions, which typically give rise to an absorption peak between 225 and 275 nm (4.5-5.5 eV). During reduction, the strength of optical absorption increases while the plasmon peak shifts to $ 270 nm, reflecting an increased πelectron concentration and structural ordering [11]. Usually, the mechanical properties of GO rely on the details of sample, such as the oxidation degree (especially coverage of the epoxy and hydroxyl groups) and thickness [12][13][14][15]. The reported Young's modulus and intrinsic strength of GO sheets show a wide range of distributions of 6-42 GPa and 76-293 MPa, respectively [16]. More details about the fundamental physical properties of GO can be found in a recent book and a review article [1,17].
Despite the aforementioned fascinating properties, there are still some drawbacks of GO for practical applications. The combination of structural defects, poor dispersion, restacking and multilayer thickness can affect the electrical properties and high surface area of GO materials [18]. The insulating nature of regular GO also limits its applications in electronic devices and energy storage. Furthermore, the residual defects and holes degrade the electronic quality of RGO [19,20]. Fortunately, the oxygenated groups can largely expand the structural/chemical diversity of GO by further chemical modification or functionalization, which offer an effective way to tailor the physical and chemical properties of GO to expected extents. As a consequence, GO and GO-based composites have shown great potentials in the applications of energy storage/conversion and environment protection. Figure 1 shows the numbers of journal publications searched by ISI with some relevant keywords. One can see that there have been tremendous efforts in developing GO-based materials for various kinds of Li batteries and supercapacitors, whereas there are also certain activities on hydrogen generation/storage as well as purification of water and air by using GO-based materials. This review aims to highlight the recent progresses on the energy and environmental applications of GO/RGO-based materials and their composites.

Applications of GO in energy conversion and storage Photocatalytic water splitting
Hydrogen is considered as a clean and renewable energy carrier in the future. Both production and storage of hydrogen are crucial for the utilization of hydrogen energy. One promising strategy of hydrogen production is water splitting by semiconductor photocatalyst under visible-light irradiation [21,22]. In the practical applications, this technique is limited by the inability to utilize visible light, insufficient quantum efficiency, amount of charge trapping centers, recombination of electronhole pair, improper energy level offsets, and small contact area between catalyst and water, etc. Thus, designing novel catalysts to meet all the industrial requirements remains a big challenge.
Considering the superior electron mobility, high surface area, feasible assembly and tunable electronic band structures [19,23], GO and GO-based materials have been developed as excellent catalysts for light-driven hydrogen generation from water splitting [24][25][26][27][28][29][30]. Yeh and coworker [25] demonstrated photocatalytic activity of GO with a band gap of 2.4-4.3 eV for H 2 evolution under mercury light irradiation, even in absence of the Pt cocatalyst. Since the GO sheets with high oxidation degree had larger band gap and limited absorption of light [28], appropriate reduction level of GO sheets was facilitated for the photoreactions to generate H 2 [30].
Photocatalytic activity of GO elucidated by DFT calculations. Jiang et al. [31] investigated the electronic and optical properties of GO with different coverage and relative ratio of the epoxy and hydroxyl groups using DFT calculations. They found that the flexibility of controlling the amount of -Oand -OH functional groups in a GO sheet allows efficient tailoring of the band gap along with band positions. GO materials with 40-50% (33-67%) coverage and OH:O ratio of 2:1 (1:1) are suitable for both reduction and oxidation reactions for water splitting (Figure 2a). Among these systems, the GO composition with 50% coverage and OH:O (1:1) ratio is a very promising material for visible-light-driven photocatalyst. Simulated optical absorption spectra (Figure 2b) further confirms that those  carefully chosen GO can harvest the major portion of solar light efficiently.
In view of the p-type conductivity which hinders the hole transfer for water oxidation and suppresses O 2 evolution, Yeh and coworkers [24] introduced amino and amide groups on the GO surface and demonstrated that the ammonia-modified GO exhibits n-type conductivity and is able to catalyze the H 2 and O 2 evolutions simultaneously. Furthermore, the p-n diodes configuration of nitrogen-doped graphene oxide quantum dots (NGO-QDs) can fulfill the simultaneous H 2 and O 2 evolutions from pure water under visible light irradiation [29]. Besides, metal deposition can enhance the photocatalytic activity of GO. For example, Agegnehu et al. [32] found that the hydrogen generation rate of Ni/GO composite from aqueous methanol solution under UV-visible light illumination is enhanced by approximately four to seven times compared to that of the bare GO.
GO-semiconductor composites replacing noble metals as catalyst. To date, large numbers of efficient GO-based composites for photocatalytic water splitting have been synthesized, including GO-semiconductor binary systems and even more complicated ternary composites [19,[33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49]. The improvement of H 2 production rate is mainly attributed to the role of GO as an electron acceptor and transporter to separate the photogenerated electron and hole pairs. For instance, H 2 production rate and quantum efficiency of the GO-Zn 0.8 Cd 0.2 S system are 1824 mol h À 1 g À 1 and 23.4% under 420 nm light, respectively [37], which are much better than the Pt-Zn 0.8 Cd 0.2 S photocatalyst ( Figure 3). Another system with high quantum efficiency is the GO-CdS nanocomposite, which reaches a high H 2 production rate of 1.12 mol h À 1 under visible light irradiation, corresponding to an apparent quantum efficiency of 22.5% at wavelength of 420 nm. It was demonstrated that the electron and hole can be efficiently separated by transferring photoinduced electrons from CdS to GO, and recombination of the electron and hole pairs in the excited semiconductor material is simultaneously suppressed [38]. First-principles investigation [39] indicated that the excited electrons in CdS are injected into GO and transport along graphene layer through π n orbital under visible light irradiation to achieve electron and hole separation, consistent with experiment [38]. Similarly, binary composites of GO/ TiO 2 [40,41], GO/AgBr [42], GO/g-C 3 N 4 [43], GO/3C-SiC [44], GO/Cu 2 O [45], GO/Sr 2 Ta 2 O 7 À x [46] also showed higher photocatalytic H 2 evolution than the bare semiconductors.
GO in the Z-scheme photocatalysis system. Alternatively, GO can be used as a solid-state electron mediator for water splitting in the Z-scheme photocatalysis system [22] to overcome the problems with electron recombination and transportability. Iwase's group [35] [37]. Copyright (2012) American Chemical Society. respectively. Such system exhibited higher H 2 production rate than the GO/BiVO 4 and Ru/SrTiO 3 :Rh. The underlying mechanism of electron transfer can be illustrated as the following picture. First, GO serves as an electron mediator to transfer electrons from the conduction band of BiVO 4 to the Ru/SrTiO 3 :Rh. Meanwhile, electrons in Ru/SrTiO 3 :Rh reduce water into H 2 on the Ru cocatalyst, while holes in BiVO 4 oxidize water into O 2 simultaneously, accomplishing a complete water splitting cycle.
Open issues. As promising photocatalysts of water splitting, GO and GO-based composites not only show the ability of separating the photogenerated electron-hole pair, but also exhibit the capability for photocatalytic H 2 evolution by itself. However, research on GO-based materials for H 2 generation from light driven water splitting is still at its early stage and requires further attentions. Explanation of the photocatalytic activity by GO content in these composites is still controversial. In this regard, theoretical calculations are highly desirable to provide critical insights on the mechanism.

Hydrogen storage
Currently, the hydrogen storage strategies can be categorized into chemical storage (hydrogen in the form of hydrides) and physical storage (hydrogen in the form of H 2 molecules). Both of them have their own deficiencies to store necessary amounts of hydrogen under technologically useful conditions. As for chemical storage, the hydrides usually possess high hydrogen contents, but concomitant with the unsatisfactory performance for hydrogen release. Meanwhile, physical storage suffers from the weak adsorption energy of H 2 molecules, which leads to low storage capacity at ambient conditions. Besides, the host sorbents of physical storage should possess large surface areas and be light in weight.

Physical storage of hydrogen
On the pristine two-dimensional (2D) graphene nanosheets, the weak adsorption energy of H 2 molecule (only 1.2 kJ mol À 1 ) [50] usually leads to very low storage capacity of hydrogen (o2 wt%) [51]. However, Guo et al. fabricated a hierarchical graphene material with the micropore of $ 0.8 nm, mesopore of $4 nm and macropore above 50 nm, which achieved a high H 2 storage capacity over 4.0 wt% at atmospheric air pressure [52]. For the GO layers, it was found that modulating the interlayer distance [53] and pore size [54] of multilayered graphene oxide to optimal values can improve the maximum hydrogen storage capacity.
Metal decoration on GO surface. More importantly, the functional groups on GO surface offer feasible ways to further composite with other species to improve the H 2 binding energy. In principle, if the obstacle of metal aggregation [55] can be conquered, metal (especially transition metal) decoration on carbon sorbents is a promising route to enhance H 2 binding energy and thus storage capacity [56]. For example, firstprinciples calculations by Wang and co-workers predicted that Ti atoms can be stably anchored by the hydroxyl groups on GO surface and simultaneously retain sufficient chemical activity to adsorb H 2 molecules [57]. Each Ti atom is able to bind multiple H 2 with a desired average binding energy (14-41 kJ mol À 1 ), corresponding to a theoretical gravimetric density of 4.9 wt% or volume density of 64 g L À 1 (see Figure 4a and b). In addition to Ti, it was theoretically proposed that Mg can reduce the hydroxyl groups on GO and thus result in À(CÀO) x ÀMg (x=1 or 2) [58]. H 2 molecules can be polarized by strong electric field jointly produced by the anchored Mg and O on GO surface, and therefore leading to a hydrogen storage capacity up to 5.6 wt% (at 200 K without any pressure) with an average binding energy of 28 kJ mol À 1 .
These theoretical predictions have been partly confirmed by experiments. For instance, Pd decoration on graphite oxide can significantly increase the hydrogen storage capability by a factor of 3.3 in comparison with that of pristine GO [59]. Moreover, Figure 4c and d show that the GO wrapped TMO composites without any additional agents, which exhibited enhanced hydrogen storage capacities, i.e., 1.36 wt% for GO/V 2 O 5 or 1.26 wt% for GO/TiO 2 , compared to that of 0.16 wt% for V 2 O 5 or 0.58 wt% for TiO 2 alone, respectively [60].
GO-based three-dimensional (3D) porous pillared materials possessing tunable porosity, accessible surface area and versatile electronic properties for hydrogen adsorption. Such materials can be formed by assembling other reactive agents with the functional groups on GO. Froudakis et al. [61] proposed a new class of pillared graphene oxide in conceptual by introducing carbon nanotubes (CNTs) inbetween the GO layer and implemented with Li decoration to increase H 2 accommodation. The Li decorated pillared GO model with a pore dimension of d= 23 Å and an O/C ratio of 1/8, achieved a gravimetric capacity greater than 10 wt% and a volumetric capacity of 55 g L À 1 at 77 K and 100 bar, respectively. Later, experiment by Aboutalebi et al. [62] demonstrated the feasibility of self-aligned GO/MWCNT hybrid frameworks, which reached an appreciable hydrogen uptake of 2.6 wt% at room temperature with proper interlayer distance.
As shown in Figure 5a, graphene oxide frameworks (GOFs) with 3D porosity was synthesized by intercalating the boronic acid (which can react with the hydroxyl groups) into GO layers, which showed superior hydrogen adsorption behavior [63][64][65]. By adopting various linear boronic acid pillaring units, the interlayer spacing between graphene planes can be tuned to an optimum value for H 2 adsorption on both surfaces, which is twice of typical porous carbon materials and comparable to MOFs. The outstanding hydrogen storage properties are attributed to the porous spaces, and the enhanced hydrogen adsorption is mainly caused by the benzenediboronic acid pillars between graphene sheets [65]. By modulating the intercalation via three kinds of diaminoalkanes between GO layers, Kim et al. [66] found an optimum GO interlayer distance of 6.3 Å with maximum H 2 uptake, similar to the predicted distance from thermally modulated GO [53].
Moreover, tunable porosity and surface area are also beneficial for accommodation of hydrogen gas [67,68]. The metallomacrocycle-graphene frameworks (Figure 5b) possess tunable porosity from microporous to hierarchical micro-and mesoporous, which allows the usually unstable square planar Ni(III) species to be stabilized in the solid state [67]. The H 2 adsorption-desorption isotherms shows that the highest hydrogen uptake of the composites achieved 1.54 wt% at 77 K and 1 bar. Recently, by the C-C coupling reaction of RGO and iodobenzene, the porous graphene frameworks displayed a high BET surface area of 825 m 2 g À a and thus achieved a H 2 storage capacity of 1.2 wt% at 77 K and 1 atm [68].
Intercalation of MOFs into GO. MOF Intercalation can enlarge the space for accommodating H 2 molecules, and also increase the number of effective sites to adsorb H 2 . Petit and Bandosz [69] observed enhancement of hydrogen uptake in copper-based GO/MOFs composites, therefore confirming the formation of new small pores. Later, Liu et al. [70] synthesized Cu-BTC/GO composite with nanosized Cu-BTC (copper-benzene-1, 3,5-tricarboxylate), which significantly improved hydrogen storage capacity compared to the pristine Cu-BTC (from 2.81 wt% of Cu-BTC to 3.58 wt% of Cu-BTC/GO at 77 K and 42 atm). Recently, Pt was incorporated into the GO/HKUST-1 composite to adsorb and dissociate H 2 molecules, which yielded nearly twofold enhancement with regard to HKUST-1 itself [71].
Open issues. Overall, GO and GO-based composites exhibit promising hydrogen storage properties, especially the 3D pillared porous GO materials show remarkable potential for the on-broad applications. More efforts should be devoted to designing new 3D architectures on the basis of GO sheets and enhancing the interaction between H 2 molecules and GO-based sorbent materials.
Chemical storage of hydrogen Noble metal catalysts supported by GO/RGO. As for hydrogen storage in the form of hydrates, one major obstacle is to lower the working temperature for release of hydrogen. Because of the large surface area and excellent chemical stability, GO and RGO can serve as appropriate supports to disperse and stabilize the metal nanoparticles (NPs) for catalyzing and dissociating H 2 molecules. Noble metal catalysts supported by GO/RGO have been also investigated [72][73][74][75][76]. A facile route to synthesize Ru/ RGO NPs via methylamine borane (MeAB) as a reducing agent was reported by Cao's group [72]. Their experiments provided the ever reported lowest activation energy of 11.7 kJ mol À 1 during the hydrolytic dehydrogenation of amine borane (AB). To further improve the catalytic activity, they synthesized the RGO supported core-shell NPs of Ag@M (M=Co, Ni, Fe) [73] and Ru@Ni [74] for the hydrolysis of amine borane (AB) and MeAB. Additionally, Pt-CeO 2 NPs supported by RGO exhibited synergistic effect to stabilize the active centers and increase the catalytic activities [75]. Nevertheless, the turnover frequency (TOF) value of the hybrid Pt-CeO 2 /RGO NPs was only 48.0, much lower than the nanocatalysts with core-shell structure [73][74][75][76].
Non-noble metal composites supported by RGO. These composites also showed good catalytic activity of hydrolytic dehydrogenation of AB. Xi et al. [77] demonstrated the catalytic activity of RGO/Pd nanocomposite in hydrogen generation of AB hydrolysis with a hydrolysis completion time (12.5 min) and an activation energy (5171 kJ mol À 1 ). Metin and coworkers [78] further exploited a better catalytic performance of RGO/Pd nanocomposite for AB hydrolysis with lower activation energy of 4072 kJ mol À 1 and higher TOF than Xi's results [77]. In addition, Zhou et al. [79] found that addition of polyethyleneimine can significantly affect the morphology and size of the resulting Fe-Ni NPs, and therefore modulate the catalytic activity of AB hydrolysis. Very recently, RGO supported Cu NPs were used as catalyst for hydrolytic dehydrogenation of AB, which exhibited the highest TOF value of 3.61 among all of the Cu nanocatalysts ever reported for this reaction [80].
GO facilitated AB (MAB) dehydrogenation. For hydrogen release through thermo-dehydrogenation, Tang et al. [81] reported a recyclable dehydrogenation of AB with a GO-based hybrid nanostructure. The hydroxyl groups on GO surface act as proton donors to react with AB and yield H 3 NBH 2 + cation, which is the key to facilitating AB dehydrogenation (see Figure 6a-c). Using first-principles calculations, Li and co-workers [82] predicted a novel composite by GO and lithium amine borane (LiAB), in which the hydroxyl groups on GO surface may interact with LiAB via one molar equivalent of H 2 released (Figure 6d). Compared to the pure LiAB, GO-LiAB hybrid shows better dehydrogenation performance with reduced reaction barriers. Most strikingly, the dehydrogenated products of GO-LiAB complex with lithium well dispersed on the GO support are still capable of adsorbing H 2 molecules with appreciable H 2 binding energy (7.9 kJ mol À 1 ) and maximum capacity of 5 wt% (Figure 6e). In such a way, chemisorption and physisorption can be appropriately combined for superior hydrogen storage, which might be a promising direction for future development of hydrogen storage materials.

Lithium batteries
To meet the growing demand of portable electronic products and electric vehicles, lithium batteries (Li-ion, Li-S and Li-O 2 ) hold great promise for future energy storage and utilization owing to their high specific energy and energy density. In principle, GO with heterogeneous chemical and electronic structures can be used as electrode material for lithium batteries. The functional groups on GO may serve as sites for chemical modification or functionalization, which in turn immobilize various active species and render hybrid architectures for electrode materials. However, applications of pristine GO in lithium batteries suffer from its insulating nature [83]. It is thus necessary to improve the conductivity while keeping the inherent merits of GO.

Lithium ion batteries
As the most popular rechargeable batteries for critical applications, lithium-ion batteries (LIBs) with conventional electrode materials suffer from its low theoretical capacity limits. In this regard, elaborately designed GO/RGO-based composites as both anode and cathode materials exhibit superior electrochemical performances, as summarized in Table 1.
Anode. Direct combination of GO with commercial graphite. Here, GO acts as the binder-free anode material in LIBs to increase the reversible capacity to 690 mA h g À 1 at the rate of 0.5 C (1 C= 372 mA g À 1 ), along with excellent cycle performance and rate capability. In the combined material, the original graphite maintains the conductivity and GO mainly contributes to the enhanced capacity as lithium accommodator [59,84]. Indeed, the conductive RGO itself can replace the conventional graphite in anode. More importantly, the residual functional groups on RGO provide Table 1 Electrochemical properties of LIBs with electrode using GO/RGO composites.

System
Current density (mA cm À 2 or mA g À 1 ) Charge/discharge rate (C) wide range of the porous structures, which are beneficial for the electrochemical performance of LIBs [85,86]. GO/RGO serves as excellent host materials. To disperse anode materials, concomitant with the synergistic effect of improving the configuration and stability of the anode, it has been demonstrated that incorporation of RGO into either metal oxides or sulfides induces remarkably enhanced battery performance with regard to pure oxides or sulfides.
The pulverization process of transition metal oxide (TMO) induced rapid capacity degradation can be effectively suppressed in the RGO/TMO hybrids [87][88][89][90][91][92][93][94]. For instance, two-step synthesized Mn 3 O 4 /RGO hybrid renders a high specific capacity up to $ 900 mA h g À 1 (near the theoretical capacity) as well as good rate capability and cycling stability [87]. Very recently, the synergistic coupling of a ternary metal oxide (Zn 2 GeO 4 ) and GO layers yields extraordinary high specific capacity, superior rate capability and long cycle life [93].
Moreover, RGO can offer enough buffer space for large volume expansion of the metal sulfide during insertion and extraction of the Li ions and alleviate the insulating problems of polysulfide [95][96][97]. The FeS@RGO nanocomposite with robust sheet-wrapped structure exhibited better electrochemical performance than the isolated FeS NPs because of the smaller particle sizes and the synergetic effects between FeS and RGO sheets, i.e., increased conductivity, shortened lithium ion diffusion path, and effective prevention of the polysulfide dissolution [95].
Application of GO/RGO in Si based anode. Compared to carbon, silicon exhibited higher theoretical capacity of $4200 mA h g À 1 and appropriately lower working potential [98]. However, serious pulverization of bulk silicon during cycling limits its cycle life [99]. Compositing silicon with GO/RGO is an effective strategy to improve the cycling performance [85,[100][101][102]. Hybrid of leaf-like GO and Si NPs as anode of LIBs achieved rather high capacity retention over 100 cycles [61], although the surface groups and large surface area resulted in irreversibility on the initial cycling. Moreover, RGO as the overcoats of Si nanowire@graphene (SiNW@G) nanocables exhibited great capability to accommodate the volume change of the embedded SiNW@G nanocables (see Figure 7) [100]. The SiNW@G@RGO composite maintains the structural and electrical integrity of the electrode and yields a high specific capacity of 1600 mA h g À 1 at a current density of 2.1 A g À 1 . Afterwards, it was demonstrated that Si nanowires directly grow on flexible  RGO sheets can also maintain the structural integrity and provide a continuous conductive network of the electrode, which resulted in over 100 cycles serving as anode in half cells at a high lithium storage capacity of 2300 mA h g À 1 [102].
Cathode. In addition to anode, inclusion of GO/RGO in cathode materials also shows significant advantages compared to the conventional polymer cathodes and lithium-TMOs cathodes in LIBs. The epoxide-enriched GO can be lithiated/delithiated as rechargeable cathode with high capacity and good stability [103,104], in which the hydroxyl groups were identified as the lithiation-active species. More efforts have been devoted to the RGO/TMO hybrid cathodes in LIBs [105][106][107][108]. Enwrapping Li 3 V 2 (PO 4 ) 3 on RGO sheets as cathode material can facilitate the charge transfer but concomitant with low initial discharge capacity (177 mA h g À 1 at 0.1 C) [105]. Even with modification of carbon [105] and cetyltrimethyl ammonium bromide (CTAB) [106] in the Li 3 V 2 (PO 4 ) 3 /RGO composite, the storage capacity only slightly increased to about 186 mA h g À 1 . In comparison, vanadium oxides/RGO composite [107,108] exhibits better lithium storage performance than the Li 3 V 2 (PO 4 ) 3 /RGO, as shown in Table 1. Moreover, the PPy/RGO hybrid shows significantly improved electrochemical properties, i.e., high rate capability and excellent cycling stability for being both cathode and anode materials in LIBs [109]. Finally, GO utilized as a protective coat can effectively adhere to aluminum and thus inhibit corrosion of the aluminum current collector used in LIBs [110]. Open issues. Limited by the horizon of theoretical capacity, even the improved lithium storage capacity of LIBs after modification of both anode and cathode materials is still inadequate for the increasing demands of portable electronics and electric vehicles, which inspires further explorations of other kinds of batteries with higher capacity.

Lithium sulfur batteries
Compared to the traditional cathode materials based on transition metal oxides or phosphates in LIBs, Li-S batteries possess higher specific capacity up to about 1675 mA h g À 1 [111]. However, they suffer several operational problems in the commercial applications, such as insulating nature of sulfur [112], dissolution of polysulfides in electrolyte [113], and volume expansion of sulfur during discharge [114][115][116]. Incorporation of GO/RGO into the sulfur cathode of Li-S batteries is an effective way to deter these operational issues. The oxygenated functional groups on the basal planes and edges as well as the structural defects of GO/RGO provide strong anchoring points [117,118], while the 2D sheet-like structure promotes the formation of self-assembled films for improving the electrochemical performance of Li-S batteries. The reactive functional groups on GO also exhibit strong adsorption of sulfur or polysulfides and effectively prevent loss of the active materials (the lithium polysulfides dissolved in the electrolyte during cycling), therefore leading to high reversible capacities [119][120][121][122] (see Table 2). However, the stable charge-discharge process of S/GO composite can only sustain limited number of cycles (e.g., 50 cycles [119]), followed by a dramatic capacity fading.
Addition of effective assistant components into GO/S composites. It has been demonstrated to further improve the electrochemical performance of GO/S composites [123][124][125][126][127]. For instance, the surfactant of polyethylene glycol (PEG) modified S/GO composites (Figure 8a) exhibit high capacity retention after long cycles, indicating an effective shuttle inhibition of polysulfide [123]. Zhou et al. [125] modified the GO/S nanocomposite with amylopectin to suppress the Li polysulfides escaped from the open channels among GO layers, and thus largely improved cyclability. Moreover, Cui et al. [126] demonstrated that wrapping of GO onto the Li 2 S surface via lithium-oxygen interactions can alleviate dissolution of intermediate polysulfides into electrolyte and therefore achieve a high discharge capacity of 782 mA h g À 1 of Li 2 S ($1122 mA h g À 1 of S) with stable cycling performance over 150 chargedischarge cycles.
S/RGO-based sulfur cathode. Compared to the S/GO composites, the sulfur cathode prevails due to better electrical conductivity [127][128][129][130]. RGO provides a conductive network surrounding the sulfur particles, which facilitates both electron conductivity and Li + ion transportation [127]. The saccule-like S@RGO composites shows better electrochemical performance as compared to the common RGO sheets since that the unique saccule-like RGO can provide a more effective electrically conductive network and offer sufficient space to accommodate the stress and volumetric expansion of sulfur during charge-discharge process [128]. Moreover, the Li 2 S/RGO nanocomposite with a unique 3D pocket structure showed a high initial capacity of 982 mA h g À 1 [130]. But the remaining problem of polysulfide dissolution (even with the presence of functional groups on RGO) still led to noticeable capacity fading over time [128][129][130].
Ternary RGO/S-based cathode materials. Further carbon decoration of the RGO/S cathode materials can enhance the Li + ion transportation and electrical transport properties of the cathode. The thermally exfoliated G@S/RGO nanocomposite possesses superior electrical conductivity and thus resulted in a high specific capacity of 1064 mA h g À 1 at the high rate of 1.6 A g À 1 [131]. Moreover, synergistic coupling of RGO with mesoporous CMK-3/sulfur composite provides sufficient buffering space and extra diffusion channel and consequently improved the overall electrochemical performance [132,133]. Enhanced electrochemical performance was also achieved in the MWCNTs@S/RGO cathode [134]. Nevertheless, adding carbon additive into cathode is still insufficient to restrain the polysulfides and alleviate the capacity fading. The synergistic coupling of RGO and PEG with sulfur can keep the sulfur well confined within the PEG-S-RGO nanostructure and therefore effectively suppress dissolution of the polysulfides, accommodate the volume change and enhance the electrical conductivity of sulfur [135]. In addition, nitrogen doped RGO shows enhanced electrical conductivity and stronger polysulfide confinement compared to the common RGO/S composite, as depicted in Figure 8c and d [136]. The pyridinic-N enriched RGO/S exhibits better discharge capacity (1356.8 mA h g À 1 ) and capacity retention (847.4 mA h g À 1 ) than the Pyrrolic-N RGO/S (1298.6 mA h g À 1 and 764.8 mA h g À 1 , respectively).
Open issues. GO/RGO-based materials facilitate the electrochemical performance of Li-S batteries such as rate capacity and cycle stability significantly, which makes the Li-S batteries rather promising in future commercial utilization. Further efforts are still needed to avoid the polysulfide-induced capacity fading. In this regard, the detailed interaction mechanism between polysulfides and GO/RGO composites has to be elucidated both experimentally and theoretically.

Lithium air batteries
Compared to other kinds of Li batteries, Li-O 2 batteries depict super promising prospects due to its extremely high theoretical energy density (up to 13,000 W h kg À 1 , comparable to petroleum) [137]. However, development of Li-O 2 batteries also suffers from some major challenges such as block of the diffusion path induced by discharged products (Li 2 O 2 and Li 2 O) [138], corrosion by atmospheric moisture [139], and attack by highly reactive reduced O 2 À [140]. Therefore, the ideal cathode materials of Li-O 2 batteries require good electrical conductivity, outstanding O 2 reduction performance, high structural stability, and suitable path for fast oxygen diffusion.
Hierarchical GO-based materials. With highly interconnected 3D channels for rapid oxygen diffusion, these materials have been used to improve the electrochemical performance of Li-O 2 batteries [141,142]. With the assistance of binder materials, the hierarchical GO exhibited an exceptionally high capacity of 15,000 mA h g À 1 in Li-O 2 battery, which is attributed to the formation of isolated Li 2 O 2 nanoparticles and alleviation of air blocking in the cathode induced by defects and functional groups [141]. Besides, GO gel acted as a special carbon source and provided the framework of 3D gel to form a highly efficient cathode material for Li-O 2 battery, which achieved a high capacity of 11,060 mA h g À 1 at the current density of 0.2 mA cm À 2 (280 mA g À 1 ) [142]. A recent study of carbon materials (RGO and thermally exfoliated graphite) in a cell with an allsolid-state positive electrode under true operando conditions, however, suggested that the functional groups of epoxy and carbonates generated during the discharge reaction can limit the rechargeability of Li-O 2 cells [140].
Synergistic effect of GO/CNT cathode materials. The leaf-like GO and CNT in the cathode of Li-O 2 batteries has a discharge capacity of 6000 mA h g GO À 1 with a cutoff voltage of 2.0 V and stable cyclic lifetime over 150, which is much better than the 2250 mA g GO À 1 for GO-based, 2500 mA g CNT À 1 for CNT-based, and 3000 mA g CNT/GO À 1 for CNT/GO mixturebased Li-O 2 batteries [85]. Moreover, the nitrogen doped GO as cathode catalyst substantially improved the electrochemical performance in the oxygen reduction reaction (ORR) process [143]. Specifically, the graphitic nitrogen content affected the electrocatalytic activity and the pyridinic nitrogen content improved the onset potential for ORR.
GO/RGO as an ideal host for metal catalysts dispersion. GO/RGO can fulfill the requirement of high dispersion and low aggregation of metal catalysts to reduce the high  overpotential. The Ru and RuO 2 -based nanomaterials hybrid with RGO exhibited superior electrocatalytic activity for the oxygen evolution reaction (OER) reaction in Li-O 2 cells, leading to a high capacity of 5000 mA h g À 1 [144]. However, the stable cycling of the Ru-based materials only sustained up to 30 cycles due to degradation of the Li-metal anode and pore clogging. In addition, Co and N-doped Co oxides [145] and Mn oxides [146,147] supported on RGO also showed high catalytic activity for the ORR of Li-O 2 batteries. Nevertheless, there was also a rapid degradation of capacity during the charge-discharge cycles in the RGO/α-MnO 2 composite, as illustrated in Figure 9.
Open issues. Despite the remarkable progresses summarized above, there are still daunting challenges in the Li-O 2 batteries, including the practical discharge capacities far below the theoretical value and the complex catalytic mechanisms. These difficulties indicate the necessity of continuous exploration to improve the electrochemical performance in the future, especially, to achieve a high capacity and a high rate in the Li-O 2 batteries simultaneously.

Supercapacitors
Compared to the lithium batteries, supercapacitors (SCs) as an alternative energy storage device possess unique merits of high power densities, short charge-discharge durations, bare memory effect, long cycle lifetime and environmentally benign [148] (see Table 3). According to the chargedischarge mechanism, SCs can be classified into two categories: (1) electrochemical double-layer capacitors (EDLCs) where the electrical charge is accumulated at the electrode-electrolyte interface [148]; (2) pseudocapacitors where the fast and reversible Faradaic reactions take place at the electrode material [149].
GO used in SCs. In principle, GO cannot be directly used in SCs due to its intrinsically poor electrical conductivity, although there have been some investigations that demonstrated feasible usage of GO in SCs [150,151]. To improve the electrical conductivity of GO and to retain the pseudo-capacitive behavior, partially reduced GO has been widely adopted [152][153][154][155][156][157][158][159][160][161][162]. The residual oxygenated groups on RGO can induce large pseudocapacitance in SCs [152]. Moreover, the co-contribution of double layer capacitance and pseudocapacitance from the oxygenated groups were observed in the SCs with water-soluble RGO as the electrode materials [153]. The reduction level of the RGO sheets (especially the variation of oxygen-containing groups) plays a significant role in controlling the intrinsic properties such as the interlayer spacing, oxygen content, BET specific surface area, and thus affects the overall performance of SCs [154,155].
Furthermore, it was shown that pH value also affects the performance of supercapacitors [156]. RGO sheets in the acidic and neutral media, which display both electrochemical double-layer (EDL) and pseudocapacitive behavior, have more oxygenated groups and lower surface areas but broader pore size distributions compared with those in the basic medium (mainly EDL). Meanwhile, chemically nitrogen doping of RGO can also tune the electronic properties to enhance the supercapacitor performance [157][158][159]. For instance, the nitrogen doped RGO exhibited much higher specific capacitance than pure RGO [159]. In particular, the content of N atoms in N-RGO has remarkable influence on the capacitance, as displayed in Figure 10a.
Non-staking GO/RGO for SCs. Considering the negative influence of intercalated water molecules on restacking of RGO at the early dried stage of GO, Lee et al. [160] functionalized GO sheets with melamine resin (MR) monomers to prevent the hydrogen bonding with water molecules upon drying and consequently obtained the porous restacking-inhibited GO sheets with a specific capacitance of 210 F g À 1 and superior capacitance retention for 20,000 cycles. Moreover, the anti-solvent (hexane) without any interaction with various oxygenated groups on GO was used to replace the water solution for fabricating the nonstacked RGO, which achieved high specific capacitance up to 236 F g À 1 at the current density of 1 A g À 1 and long cycle life [161]. Meanwhile, incorporation of Mg(OH) 2 nanosheets as the spacer in-between GO sheets effectively inhibited the restacking behavior during heat-treatment of GO, but preserved the stable oxygen-containing groups to enhance the pseudocapacitance at the same time [162].
Coupling of carbon materials with GO/RGO. It has been demonstrated to be a feasible strategy to enhance the supercapacitance [163][164][165][166][167]. For example, the core-shell MWCNT@GO nanoribbons exhibited a high specific capacitance of 252 F g À 1 , higher than 194 F g À 1 for the SWCNT/ GO electrode [163]. The CNT can be used as the spacer to inhibit restacking of RGO and to increase the supercapacitance [165,166]. Meanwhile, addition of electrochemically reduced GO resulted in a unique wrinkled yet porous structure with higher available surface area for accumulating the electrolyte ions to produce excellent electrochemical double-layers [167].
Incorporation of GO/RGO with transition metal oxides (TMOs). This can break some restrictions of TMOs electrode, such as low working voltage, poor stability and unsatisfactory high-rate capabilities [168,169], therefore improving the overall performance of SCs [170][171][172]. For example, GO induced better electrochemical stability (84.1% retention of supercapacitance) in the needle-like MnO 2 /GO composite electrode compared to the pure nano-MnO 2 (69% retention) even with a slightly smaller supercapacitance compensation of 197 F g À 1 (211 F g À 1 for the pure nano-MnO 2 ) [170]. Furthermore, the RGO/TMOs composites with controllable reduction level possess better electrical conductivity than the GO/TMOs composites, making them more suitable for SC applications [173][174][175][176][177][178][179][180][181][182][183][184][185][186][187][188]. For instance, the one-step synthesized Co 3 O 4 nanosheets-RGO (Co 3 O 4 NSs-RGO) hybrid with good redox activity exhibited a supercapacitance of 187 F g À 1 at 1.2 A g À 1 and maintained 91-94% capacitance even after 1000 cycles [176]. Similarly, the MnO 2 /RGO composite as SC electrode material displayed good flexibility and strength [178]. In addition, it facilitates easy access of electrolyte ions into the large surface area of the pseudocapacitive hybrid paper electrode. In a 3D flower-like hierarchical NiO/RGO composite, the NiO NPs not only inhibited the RGO restacking but also functioned as catalyst for GO reduction, which led to a high supercapacitance up to 428 F g À 1 at the discharge current density of 0.38 A g À 1 [179]. Most strikingly, the morphology of TMOs NPs was found as a key factor to determine the overall electrochemical performance of SCs [180,181]. Other TMOs/RGO hybrid composites like Figure 10 (a) The cyclic retentions of the supercapacitors based on N-RGOs at 0.5 A g À 1 . Reprinted with permission from Ref. [159]. Copyright (2014) Elsevier Ltd. (b) Supercapacitance performance of the PANI/GO-based composite. Reprinted with permission from Ref. [202]. Copyright (2012) American Chemical Society. [182][183][184], CuO/RGO [185,186] and TiO 2 /RGO [187,188] also exhibited great electrochemical performance of SCs. Especially, the one-step solvothermal synthesized Fe 3 O 4 /RGO composite achieved an extremely high specific capacitance of 843 F g À 1 at the discharge current density of 1 A g À 1 in concomitant with very stable cycling performance (without any lose in supercapacitance after 1000 chargedischarge cycles) [182].
Addition of ternary carbon material into the GO/RGO-MOs can further improve supercapacitor performance [189,190]. In this case, the low solubility and aggregation tendency of CNTs restricted the growth of Ni(OH) 2 particles and therefore improved the activity of the NiO-CNT/RGO electrode to reach a specific capacitance of 1180 F g À 1 at 1 A g À 1 , along with excellent capacity retention of 95% after 2000 cycles (Figure 11a and b) [190].
Synergistic coupling of metal sulfide with GO/RGO. This has been demonstrated to be an effective strategy to improve the supercapacitor performance [191,192]. The NiS/GO nanocomposite achieved a high specific capacitance of 800 F g À 1 at 1 A g À 1 as well as a long life over 1000 cycles [192]. The hybrid of layered double hydroxides (LDHs) and GO/RGO composites have also been investigated for their extraordinary SCs performances [193][194][195]. Figure 11c shows the high-resolution image of the 3D-RGO nanocup/Ni-Al LDHs composite. This composite exhibits an extremely high specific capacitance of 2712 F g À 1 at the current density of 1 A g À 1 and kept 98.9% of the initial capacitance after 5000 cycles at the current density of 30 A g À 1 (Figure 11d) [194].
GO/RGO with conducting polymers. Compared to the TMOs, conducting polymers [196] exhibit superior conductivity, higher supercapacitance, lower cost; but the relatively poor cycling performance hinders their practical applications. To overcome this shortcoming, GO/RGO are uniformly dispersed in the conducting polymers, e.g., polyaniline (PANI) [197][198][199][200][201][202][203][204] and polypyrrole (PPy) [205][206][207][208][209], and thus form stable hybrid composites for SC electrodes. The compositing mechanisms between GO/ Reprinted with permission from Ref. [190]. Copyright (2014) The Royal Society of Chemistry. (c) High-resolution image of the 3D -ARGON/NiAl-LDH; (d) Specific capacitances of the 3D -ARGON/NiAl-LDH (red) and pure NiAl-LDH (black) capacitor cells in different discharge current density (Inset shows the cyclic performance of the 3D-ARGON/NiAl-LDH at a current density of 30 A g À 1 ). Reprinted with permission from Ref. [194]. Copyright (2013) Elsevier Ltd. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) RGO and conducting polymers (including electrostatic interaction, hydrogen bonding and π-π stacking) influence the supercapacitor performance [198]. For instance, owing to the strong interaction between GO and PANI, the PANI/GO composite reached an extremely high supercapacitance of 1136 F g À 1 , compared to 484 F g À 1 for the PANI alone at the scan rate of 1 mV s À 1 [200]. Moreover, GO can serve as a weak electrolyte as well as act as an effective charge balancing dopant within the PPy, both being beneficial for the SC applications [207].
Chemistry of GO/RGO-conducting polymer surfaces. The surface chemistry has demonstrated significance for controlling the overall electrochemical performance of SCs. Among four kinds of surface functionalized graphene/PANI composites, i.e., GO, RGO, nitrogen-doped RGO (N-RGO), and amine-modified RGO (NH 2 -RGO) in Figure 10b, the NH 2 -RGO/PANI composite shows largest specific capacitance of 500 F g À 1 at 2 mV s À 1 over 680 cycles [202]. Furthermore, it was found that ethanol can facilitate uniform coating of (poly 3,4-ethylenedioxythiophene), PANi, and PPy on RGO sheets [210]. Another polymer of benzimidazole cross-linked with GO formed a porous composite and therefore showed good cyclic stability of SCs [211]. Moreover, RGO/polyamide-66 (PA66) nanofiber (as electrodes) [212] and GO/PPy composite (as organic current collector) also exhibited great electrochemical performance in SCs [213]. In comparison with the pure ion gel and the conventional supercapacitor, the GO-doped ion gel as electrolyte in the all-solidstate supercapacitor possess less internal resistance, higher capacitance performance, and better cycle stability [214,215].
Based on the GO/RGO-polymer composites, other active materials such as gold [216], TMOs [217,218], and carbon materials [219][220][221] have been further introduced to enhance the SCs performances. Especially, the hierarchical free-standing CNF/GO/PANI composite possesses 3D open diffusion channels for electrolyte ions and thus yields a high supercapacitance of 450 F g À 1 at the scan rate of 10 mV s À 1 and good cyclic stability up to 1000 times [220].
Open issues. The GO/RGO-based composites exhibit superior supercapacitive performance by providing the synergistic effects of EDLs and pseudocapacitive behavior. For the commercialization of high performance supercapacitors, however, there are still significant challenges. Further investigations not only improving the performance of electrodes but also illustrating the intrinsic synergistic effects are expected for materials design and applications.

Management of harmful gases
Air pollution by industrial release of greenhouse and toxic gases is one of the biggest threatens to environment. In principle, removal of air pollutants can be divided into three routes: (1) direct reduction of gases emission, (2) gases capture and storage, (3) ultimate utilization. The oxygen-containing functional groups on the basal plane and sheet edges make GO capable to covalently and noncovalently interact with various molecules. The unique electronic structures of GO also endow effective catalyst for converting undesirable gases during the industrial processing. Therefore, GO-based composites are promise materials for capture and conversion of harmful gases such as CO 2 , CO, NO 2 , and NH 3 .

Capture of harmful gases
Capture of greenhouse gas CO 2 . The few-layer GO sheets show superior adsorption behavior, especially under the assistance of water molecules [222][223][224][225]. Molecular dynamics simulation demonstrated that the functional groups of GO is able to enhance CO 2 adsorption [223]. The presence of water can maintain the integration of CO 2 intercalated GO structure [224] and affect CO 2 migration by repulsive interaction between CO 2 and the oxygenated groups attached on the GO sheets [225]. Compared to the MWCNTs-LDHs [226], GO-LDHs [227] exhibit better efficiency of CO 2 adsorption under low GO content, while the poor network forming ability limited the CO 2 adsorption capacity even with further increased GO loading (see Figure 12a and b). For the layered double oxides (LDOs), it has been demonstrated that incorporation of GO can alleviate the capacity decay of CO 2 [228]. Addition of GO (20 wt% content) into the hybrid monolith aerogels of chitosan affected the morphological characteristics and dramatically enhanced the BET surface area for strong interactions with CO 2 ; thus the CO 2 storage capacity increased from 1.92 to 4.15 mol kg À 1 [229]. Besides, GO facilitated the dispersion of the nanoscale Cu-MOF, which provided more active sites for CO 2 adsorption and thus led to a high CO 2 adsorption capacity of 8.26 mmol g À 1 at 273 K under 1 atm [70].
Removal of ammonia. In addition to the greenhouse gases, GO/RGO-based composites have excellent adsorption capability of ammonia [230][231][232][233][234][235][236][237][238]. Bandosz's group found that the oxygenated groups, remained sulfonic groups, and water molecules play important role in the ammonia adsorption of graphite oxides [230][231][232][233]. Further incorporation of additional active materials such as polyoxometalate [234], Al 13 [235] and MnO 2 [236] into graphite oxides can dramatically enhance the ammonia accommodation. Moreover, first-principles calculations showed that the diverse active sites on GO (hydroxyl and epoxy functional groups and their neighboring carbon atoms) are effective for facilitating the charge transfer between NH 3 and GO [237]. Later, the co-contribution of epoxide groups and carbon vacancies to NH 3 dissociation has been confirmed by in situ IR microspectroscopy experiments combined with DFT calculations [238].
Removal of other harmful gases. It has shown that other harmful gases like formaldehyde [239], acetone [240], H 2 S [241,242], SO 2 [243,244], CO [245], nitrogen oxides [244,246] can be also effectively removed via GO/RGO-based composites. DFT calculations predicted that the hydroxyl and carbonyl functional groups as well as the nearby carbon atoms induced strong interactions with NO x (x=1, 2, 3) molecular species, which eventually led to chemisorption of these gaseous molecules [246]. For the extremely dangerous acetone gas, GO foam displayed higher affinity than that of RGO foams and other carbon materials [240]. In addition, it has been demonstrated that a small amount of GO integrated with MOF-5 can increase the available Zn sites in MOF-5 and also create pore space with strong dispersive force for H 2 S capture [241]. A theoretical study by Huang et al. predicted that the carbonyl groups were the active sites for H 2 S dissociation among various functional groups on GO, while the H 2 O molecules behaved as the challenging gas to block the potential active sites for H 2 S decomposition [242]. Moreover, in the graphite oxide/Zr(OH) 4 composite, SO 2 retention is not only accounted by physical adsorption in small pores but also contributed by strong attraction of terminal -OH groups of hydrous zirconia [243].
The atomistic mechanism for selective gas adsorption of GO-metal hybrid materials has been revealed by DFT calculations and grand canonical Monte Carlo (GCMC) simulations by Zhao's group [245]. As shown in Figure 12c, the titanium-decorated GO (Ti-GO) monolayer sheet is an ideal sorbent for carbon monoxide capture with a large binding energy of about 70 kJ mol À 1 , due to stronger hybridization between the empty d orbitals of Ti and the occupied p orbitals of CO. GCMC simulations further demonstrate strong selectivity of Ti-GO sheet for CO adsorption in a mixture with other gases like CH 4 , CO 2 , N 2 (Figure 12d).

Conversion of harmful gases
GO/RGO-based composites as chemical catalysts for CO 2 conversion. Besides gas capture, GO/RGO-based composites have attracted substantial attentions on converting the air pollutants to some useful energy resource for full utilization. With the assistance of tetrabutylammonium bromide (Bu 4 NBr) as co-catalyst to convert CO 2 into propylene oxide, the GO-Bu 4 NBr composite yielded 96% of propylene carbonate under relatively mild conditions [247]. The porous GO foams served as not only oxidant but also catalyst for oxidization process from SO 2 to SO 3 [248]. SO 2 can also be inserted into GO as the oxidized intermediate, which led to functionalization of graphene oxide with sulfur and further decoration by various types of organic moieties [249]. In addition to acting as catalyst itself, GO/RGO may serve as excellent host to support efficient catalyst for conversion of CO [250], CO 2 [251] and NH 3 [252]. For instance, the oxygenated groups on GO can accelerate formation of Co nanocrystals, leading to 3D reticular structure, which can be seen in the TEM image of Figure 13a. This structure displays good catalytic performance in the Fischer-Tropsch CO 2 hydrogenation process (see Figure 13b) [251].
Photocatalytical conversion of CO 2 by GO/RGO-based composites. Benefited from its unique electronic structures, GO/RGO-based composites behave as superior photocatalysts for CO 2 conversion [253][254][255][256][257][258]. The Pt modified RGO and Pt modified TiO 2 nanotubes were combined as cathode and photo-anode catalysts, respectively, which exhibited high efficiency of converting CO 2 into chemical materials (CH 3 OH, C 2 H 5 OH, HCOOH and CH 3 COOH) [254]. Considering the high cost of Pt catalyst, alternative non-noble metal catalysts have been explored for CO 2 conversion [253,[255][256][257][258]. For Cu 2 O, RGO coating also induced retarded electronhole recombination, efficient charge transfer, and protection, which dramatically enhanced the catalytic activity and the system stability for photoreduction from CO 2 to CO (Figure 13c and d) [253]. By modulating the oxygenated groups, the modified GO reached a methanol conversion rate of 0.172 μmol g À 1 cat À 1 h À 1 (under 300-Watt halogen lamp irradiation) [255]. RGO incorporation also caused charge anti-recombination of TiO 2 [257] and CdS [258] and therefore enhanced the photocatalytic activity for reducing CO 2 to useful CH 4 , which is even superior to the optimized Pt-CdS nanorod photocatalyst [258]. Except for CO 2 , colloidal GO can enhance the photocatalytic activity of nanocrystalline TiO 2 for gas-phase oxidation of ethanol and benzene by air oxygen, since the GO sheets were capable to accept photoelectrons from titania nanocrystals and to bound to TiO 2 surface [259].

Water purification
Adsorption of containments in wastewater Parallel to air pollution, water pollution is another worldwide environmental concern. The effective strategies of water purification can be categorized into pollutants adsorption and conversion. For the pollutants (mainly heavy metal ions and organic dyes) in wastewater that strongly threaten human, animals and plants, GO/RGO-based composites typically show strong binding with these pollutant species, as summarized in Table 4.
GO for heavy metal ion adsorption. Even with the oxygencontaining functional groups on GO/RGO surface acting as the active sites, GO exhibits only very low adsorption capacity of Cu 2+ ions because of the aggregation triggered by the interaction between GO and Cu 2+ ions [260]. Similar result for adsorption of Cu 2+ ions was observed in GO aerogels [261]. Later, it was reported that GO nanosheets showed increased adsorption ability for Cd(II), Co(II), Au(III), Pd(II), and Pt(IV) ions than Cu 2+ , due to their good structural stability [262,263]. Most strikingly, the few-layered GO materials (which showed no aggregation even after five months) exhibited excellent adsorption capacities of about 842 mg g À 1 for Pb(II) ions at 293 K [264]. However, further attempts of applying the magnetite-GO composite for Co(II) ions removal only led to very low adsorption capacity of 12.9 mg g À 1 [265].
GO with organic compounds. This can provide more feasible anchoring sites for heavy metal ions [266][267][268][269][270]. The UV-activated 2,6-diamino pyridine-RGO composite created an extra pyridinicnitrogen lone pair, which facilitated the removal efficiency of Cr (VI) by not only adsorption but also partial reduction [267]. The synergistic combination of conducting polymers and GO exhibits   [268][269][270]. Especially, the hierarchical PANI/GO [269], which exhibited extremely high removal capacity of 1149.4 mg g À 1 for Cr(VI), simultaneous with partial reduction from Cr(VI) to Cr(III) by positive nitrogen groups and the assistance of p electrons on the carbocyclic six-member ring. Other chitosan modified GO, e.g., magnetic chitosan-GO for Pb(II) [271], chitosan-gelatin-GO for Cu(II) and Pd(II) [272] and chitosan-GO for Ag(II) and Pd(II) [273], also showed strong attraction to the heavy metal ions. GO for organic dyes adsorption. On the other hand, GO/ RGO itself renders strong ability to remove another kind of harmful containments in wastewater, i.e., organic dyes. Zhang et al. [274] reported that GO exhibited high removal efficiency (499%) of methylene blue (MB) due to the electrostatic interaction and partially contributed by the π-π stacking interaction between GO and MB. In addition, competitive adsorption of various organic dyes on GO surface was also found [275][276][277]. After removing some functional groups, the partially reduced GO still kept strong affinity for MB adsorption owing to the C= O or C= N double bonds and CÀ C/C = C bonds induced conjugate interaction [278]. Na 2 S 2 O 4 in situ reduction of GO can convert carbonyl groups into hydroxyl groups, which acted as the key sites for a sharply adsorption enhancement of acridine orange from 1400 to 3300 mg g À 1 [279].
GO/RGO-iron oxides composites for adsorption of containments in wastewater. For GO/RGO-based composites, it has been demonstrated that addition of iron oxides can effectively enhance the containment removal in wastewater under external magnetic field [280][281][282][283][284][285][286][287][288][289]. For example, the Fe 3 O 4 -RGO hybrid greatly enhanced the thermal stability of GO and also provided additional magnetic field for facilitating the removal of MB and neutral red (NR) [283]. Similarly, the magnetic chitosan-GO composite functioned as efficient adsorbents in removal of MB, which was benefited from the surface chemistry of GO, the hydrophobicity of β-cyclodextrin, the abundant amino and hydroxyl functional groups of chitosan, and the magnetism of Fe 3 O 4 , simultaneously [289]. In addition, it was found that the iron oxide decorated GO only exhibited adsorption ability for Pb (II) but not for 1-naphthol and 1-naphthylamine, whereas the iron oxide-RGO composite behaved in an opposite way [285]. A recent study of magnetic iron oxide/RGO nanohybrid revealed effective adsorption ability for both heavy ions and organic dyes from the wastewater [286].
Besides, there have been other attempts of utilizing the calcium alginate-GO hybrid for MB adsorption [290] and the 3D porous GO-polyethylenimine composite for amaranth adsorption [291], which achieved maximum adsorption capacities of 181.8 and 800 mg g À 1 , respectively. Furthermore, the poly(diallyldimethylammonium chloride)-GO hybrid exhibited high removal efficiency for both ponceau S (PS) and trypan blue (TB) owing to the strong π-π stacking and anion-cation interactions [292]. The poly(acrylamide)-RGO nanocomposite showed enormous attraction for both Pb(II) ions and MB simultaneously (1000 mg g À 1 for Pb(II) and 1530 mg g À 1 for MB, respectively) [293].

Conversion of containments in wastewater
Chemical reduction of containments. In general, the pollutant conversion technology in water purification via GObased materials can be categorized into chemical-and photo-reduction. For chemical reduction of Cr(VI) to low toxic Cr(III) species at low pH, the ethylenediamine coated RGO (ED-RGO) exhibited a relatively high conversion rate with the aid of π electrons on the carbocyclic six-membered ring of ED-RGO [266]. The conversion mechanism and PH affected conversion ratio of Cr(VI) ions have been displayed in Figure 14a and b, respectively. In addition, the zigzag edges of RGO acted as the catalytic active sites, while the basal plane of RGO served as the conductor for electron transfer during the catalytic reduction of nitrobenzene by Na 2 S [294]. Other GO-based composites like Co 3 O 4 coated GO and Co 3 O 4 nanorods coated RGO also displayed high chemical catalytic ability for the Orange II decomposition [295,296] and the methylene blue degradation [297], respectively.
Photoreduction of heavy metal ions. On the other hand, GObased materials as photocatalysts show high conversion rate for both toxic Cr(VI) ions and organic containments. It has been demonstrated that additions of RGO into TiO 2 NPs [298], CdS [299] and α-FeOOH nanorod [300] can significantly enhance the photocatalytic activity of Cr(VI) reduction. In the α-FeOOH nanorod/RGO composite, the extended π-conjugated flat 2D layer of graphene played a crucial role in channelizing the photoexcited electrons on its surface, which led to minimization of the electron-hole recombination for enhancing the photocatalytic activity of the α-FeOOH nanorod (Figure 14c and d) [300].
Photoreduction of RhB. Due to the extraordinarily slow rate of RhB degradation by RGO alone under visible-light irradiation [301], many compounds have been incorporated with RGO to improve the photocatalytic performance for RhB degradation. The CNT-pillared RGO exhibited excellent visible light photocatalytic activity of RhB degradation owing to its unique porous structure and the exceptional electron mobility of graphene [302]. The synergistic combination of RGO with other photocatalysts such as metal oxides [303][304][305], Ag 3 PO 4 [306], BiVO 4 [307] and Ag@AgCl [308] resulted in outstanding photocatalytic performance for the degradation of rhodamine dyes. For example, Figure 15a shows the TEM micrograph of TiO 2 ÀRGO nanocomposites. This composite exhibits high conversion rates of 98.8 wt% after 80 min under visible light (Figure 15b) [303]. Similarly, the ZnO nanowire À RGO [304] also achieved high conversion rates of 98% within 10 min under ultraviolet irradiation. Moreover, the excellent electron-accepting and -transporting properties of RGO obviously enhanced the photocatalytic activity of RHB in the Ag 3 PO 4 À RGO composite [306].
Photoreduction of other organic dyes. Other organic dyes such as MB [309][310][311][312][313], methyl orange (MO) [314,315] and crystal violet (Cv) [316] have also been effectively photodegraded via GO/RGO-based photocatalysts. Recently, the Z-scheme photocatalytic system of the Ag@AgCl/RGO composite displayed much higher photocatalytic activity of MB degradation than the normal Ag@AgCl/RGO composite because of the two parallel photochemical reactions by the electron-hole pairs of the low energy level recombined in space by metallic Ag as a solid-state electron mediator and the remaining electron-hole pairs of the high energy level [309]. Furthermore, the GO nanosheets not only controlled the fabrication of Ag/AgCl nanostructures, but also acted as a promoter for the photocatalytic performances of MO photodegradation [314]. In the nanoscale hybrid of ZnO-GO composite, GO also induced high e À -h + pair separation under light illumination ($ 95% Cv degraded within 80 min) for high rate photodegradation of Cv dye [316].
Besides dyes, other kinds of organic pollutants can be also photodegraded by the GO/RGO-based composites. The typical herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) was nearly 100% photodegraded by Ag/RGO-TiO 2 composite under simulated solar light irradiation with good cyclic stability over ten times [317]. Similarly, the epoxy-GO x hybrid showed photocatalytic activity of phenol degradation [318] and the CdS nanowires/RGO nanocomposites displayed selectively photocatalytic reduction of aromatic nitro organics in water under visible light irradiation [319].

Summary and perspective
Benefited from the oxygenated groups, GO (including RGO) can be chemically modified or functionalized by cooperating with other active species. The GO/RGO composites therefore possess unique surface chemistry and architectures, such as three dimensional network, large surface area, tunable electrical conductivity, satisfactory chemical/electrochemical stability, high flexibility and excellent elasticity. To promote further investigations of such kind of advanced materials, this review comprehensively summarizes recent progresses on the energy storage and conversion (hydrogen storage, photocatalytic water splitting, lithium batteries, and supercapacitors) as well as environmental protections (air pollutants removal and water purification) using GO/RGO and the related composite materials.
By varying the concentrations of surface functional groups, band gap and work function of GO sheets can be tuned to suitable values as efficient catalysts for photocatalytic water splitting, reduction of harmful gases or heavy metal ions, and degradation of organic containments. Also, the high surface area and abundant functional groups of GO offer enough space and active sites for adsorption of gaseous molecules and various species in solutions, such as hydrogen storage, capture of CO 2 , CO, NO 2 , and NH 3 gases for air purification, removal of heavy metal ions and organic containments for water purification. In principle, applications of pristine GO as electrode materials in lithium batteries and supercapacitors are restricted by its insulating nature. However, appropriate reduction can improve the electrical conductivity while keeping the inherent merits of GO. By further compositing with other active materials, such as carbon materials, metals, metal oxides, conducting polymers and organic species, GO/ RGO-based materials show greatly improved performance in the energy storage devices.
As for the environmental applications, metal-decorated GO exhibit satisfactory capability of gas capture and detection, especially the harmful gases like CO 2 , CO, NO 2 , and NH 3 ; meanwhile even GO itself can act as catalyst for oxidation of the harmful gases. In the polluted water, various GO-based composites not only display strong affinity for the adsorption of heavy metal ion and organic containments but also function as efficient chemical-and photocatalysts in converting the toxic metal ions and organic containments into the harmless products.
Despite the great advances in the GO/RGO-based materials for energy utilization and air/water purification described above, development of mature technologies using GO and GO composite materials still remains a long-term target in the future and requires more attentions from different aspects. From the fundamental point of view, it would be crucial to elucidate the interaction mechanism between GO/RGO and the additional material in a composite and to further understand the possible synergistic effects. Such knowledge would provide useful guidance to design GO-based composites with optimal performance. On the technological side, integration of GO/RGO-based materials into a realistic device for energy or environmental applications remains a grand challenge. For instance, the restacking issue of 2D GO sheets must be taken into consideration. The thermal/chemical stability of GObased materials (especially those hybrid ones) under operational conditions is also a critical concern.
Nevertheless, by combining the superior physical/chemical properties of GO/RGO themselves and the versatile nanomaterials that can couple with GO/RGO, GO/RGO-based materials have a bright future in the energy and environmental applications. We anticipate this active field will continue growing rapidly, leading eventually to a variety of mature materials and devices that would benefit the society. He is now director of the key laboratory of materials modification by laser, ion and electron beams (Ministry of Education). His major research field is computational materials science with special interest in clusters, nanostructures, and new energy materials. He has contributed over 300 refereed journal papers and his current H-index is 46.
Shengbai Zhang received his Ph.D. in Physics from the University of California at Berkeley in 1989. He moved to Xerox PARC as a postdoc, before joining the National Renewable Energy Laboratory in 1991. In 2008, he became the Senior Kodosky Constellation Chair and Professor in Physics at Rensselaer Polytechnic Institute. His computational research covers a wide range of materials for bulk properties, defect structures, and surface physics. His recent work involves low-cost photovoltaic materials, phase change memory materials, topological insulators, two-dimensional layered materials, and excited state dynamics. He is a Fellow of the American Physical Society since 2001.