Graphene-based materials: Synthesis and gas sorption, storage and separation

Graphene-based materials have generated tremendous interest in a wide range of research activities. A wide variety of graphene related materials have been synthesised for potential applications in electronics, energy storage, catalysis, and gas sorption, storage, separation and sensing. Recently, gas sorption, storage and separation in porous nanocarbons and metal–organic frameworks have received increasing attention. In particular, the tuneable porosity, surface area and functionality of the lightweight and stable graphene-based materials open up great scope for those applications. Such structural features can be achieved by the design and control of the synthesis routes. Here, we highlight recent progresses and challenges in the syntheses of graphene-based materials with hierarchical pore structures, tuneable high surface area, chemical doping and surface functionalization for gas (H2, CH4, CO2, N2, NH3, NO2, H2S, SO2, etc.) sorption, storage and separation. 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/


Introduction
Graphene is a two-dimensional (2D) sp 2 bonded carbon sheet, arranged in a hexagonal honeycomb lattice [1][2][3][4]. From a fundamental point of view, graphene is nothing but a single layer of graphite, which is an infinite three-dimensional (3D) material made up of stacked layers of graphene. The layers in graphite interact weakly through van der Waals (vdW) forces. In terms of properties graphene is unique; it is a soft membrane and at the same time possesses a high Young's modulus, and good thermal and electrical conductivities [4][5][6]. In addition, a single-layer graphene is a zero band gap material and highly transparent, exhibits optical transmittance of 97.7%. With its high theoretical specific surface area of $2600 m 2 /g graphene provides a rich platform for surface chemistry [7][8][9][10][11][12]. The combined extraordinary physical and chemical properties of graphene, in turn, has ignited extensive research in nanoelectronics, supercapacitors, fuel-cells, batteries, photovoltaics, catalysis, gas sorption, separation and storage, and sensing [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28]. A roadmap of graphene materials is described in a recent review [4]. It is important to note that most of the graphene properties are sensitive to structural defects and the number of layers [4,[29][30][31]. Thus in order to exploit most of the proposed applications, the synthesis routes and conditions is important in tuning the structure and properties of graphene. There are a number of reviews on graphene related materials for possible applications in relation to optical, electronic, photocatalytic and electrochemical properties . However, there is a lack of consideration on their important molecular interactions, e.g. for molecular adsorption and storage. This article attempts to address such issues.
Several large scale processing methods have been involved for different graphene based materials through either graphitic top-down or molecular carbon precursor bottom-up approaches, as summarised in Fig. 1. The flexibility in modification and functionalization of the graphene surface has opened up many possibilities for the development of tailored functional materials. For example, surface modification has been applied to tune the band-gap of single-layer graphene for microelectronic devices. Similarly, the intrinsic non-porous 2D graphene is tuned to highly porous 3D architectures for electrochemical devises (batteries, fuel-cells and supercapacitors) and gas sorption, storage, separation and sensing.
On a relatively large scale, graphene is mostly obtained from graphite precursors through oxidation-exfoliation-reduction, i.e. in the form of graphene oxide (GO) as schematically shown in Fig. 1a [4,33,[49][50][51][52]. The GO structure contains abundant oxygen-rich functional groups; hydroxide and epoxide groups on the basal plane and carbonyl and carboxyl groups on edges of the graphene sheets. Thus the GO is hydrophilic in nature and soluble in water and several solvents. GO with its lamellar water, a largely expanded and tuneable layer structure provides a rich platform for engineering a wide range of functionalities and reaction sites for chemical modifications. Generally, the oxidation and reduction creates many defective sites on the graphene, which offers clear advantage in gas sorption, storage and separation and further functionalization.
Carbon based materials have been considered for promising gas sorption, storage, and separation because of the abundance, robust pore structure, tuneable porosity and surface area, light-weight, high thermal and chemical stability, and easy synthesis in industrial scale. There is a considerable amount of interest in graphene related materials for gas sorption, storage and separation but still a lack of comprehensive review on such a topic [7,34,36,43,[87][88][89]. Here, we present an up-to-date overview of the theoretical predictions and experimental results on the graphene-based materials towards gas sorption, storage, and separation with particular emphasis on H 2 , CO 2 , and CH 4 . In addition, a detailed account is also provided on the adsorptive removal of toxic gas pollutants, such as NH 3 , NO 2 , SO 2 and H 2 S. Fig. 2 represents the overall overview of graphene-based materials and their possible applications in molecular adsorption and storage.

Gas sorption, storage and separation: principles and methodology
Gas sorption, storage and separation in carbon materials are mainly based on physisorption on the surfaces and particularly depend on the electrostatic and dispersion (i.e., vdW) interactions. The former can be tuned by introducing charge variations in the material, and the latter by chemical substitution. The strength of the interaction is determined by the surface characteristics of the adsorbent and the properties of targeted adsorbate molecule, including but not limited to the size and shape of the adsorbate molecule along with its polarizability, magnetic susceptibility, permanent dipole moment, and quadrupole moment. Li et al. summarise the adsorption-related physical parameters of many gas or vapour adsorbates, and herein Table 1 we show a few of those of interest, H 2 , N 2 , CO, CO 2 , CH 4 , NH 3 , SO 2 and H 2 S [90]. For instance, an adsorbent with a high specific surface area is a good candidate for adsorption of a molecule with high polarizability but no polarity. Adsorbents with highly polarised surfaces are good for adsorbate molecules with a high dipole moment. The adsorbents with high electric field gradient surfaces are found to be ideal for the high quadrupole moment adsorbate molecules [91]. Normally, the binding or adsorption strength with a carbon nanostructure is relatively low for H 2 and N 2 ; intermediate for CO, CH 4 and CO 2 ; and relatively high for H 2 S, NH 3 and H 2 O. Thus, surface modifications, such as doping, functionalization and improving the pore structure and specific surface area of nanocarbons, are important to enhance gas adsorption. For this purpose, graphene offers a great scope for tailor-made carbonaceous adsorbents.
Experimentally, at a given temperature, the quantity of adsorbed gas can be determined by an adsorption isotherm, generally carried out by one of the two methods: volumetric or gravimetric. The adsorption isotherm (namely equilibrium isotherm) characterises the adsorption equilibrium. The adsorptive isotherms of individual pure gases are also considered as a promising way to evaluate the adsorptive separation. In case of membrane separation, the kinetic diameter of test gas plays an important role, where gas molecules with a kinetic diameter smaller or larger than the membrane pore diameter are separated by molecular sieving. Molecular separation can also be facilitated by large differences in diffusion kinetics across a relatively thick porous membrane.

Theoretical insights and predictions
3.1. Hydrogen storage 3.1.1. Expanded and pillared graphene layers The planar sheet of graphene with its inherent specific surface area of $2630 m 2 /g [92] in the ideal case has further motivated theoretical calculations for gas adsorption from the well-established carbon nanotubes and graphitic structures. The adsorption of a monolayer of H 2 on a single side of graphene sheet can lead to about 3 wt% of H 2 (H/C = 0.18). However, it is thermodynamically impossible for H 2 molecules to penetrate between the graphene layers of graphite. Deng et al. design a Li-doped pillared graphene sheets (Li-PGS) and single-wall carbon nanotubes (Li-P-SWNT) [93]. Grand canonical Monte Carlo (GCMC) simulations predict a H 2 storage capacity of 6.5 wt% at 20 bar and room temperature in Li-PGS with a doping ratio of Li:C = 1:3 and an interlayer distance of 1.0 nm (Fig. 3). The Li dopants act as positive (acidic) cores to attract H 2 molecules with a binding energy of $0.3 eV. An interlayer spacing of around 0.6 nm with Li:C = 1:6 leads to over 3 wt% of H 2 adsorption. This interlayer spacing [94] or pore width [95] allows H 2 molecules to interact both sides of graphene. Thus the spatial distribution of molecular hydrogen adsorption on the graphene plane is very delocalised (Fig. 4). The H 2 interaction with graphene in bulk graphite is a localised phenomenon therefore adsorption generally restricted to the outermost graphene plane [96]. A maximum of 3.3 wt% of H 2 is achieved with one monolayer adsorption between two graphene layers separated by 0.6 nm ( Fig. 4). At most two H 2 monolayers could fit between the graphene layers when the layers are separated by 0.9 nm, yielding a maximum capacity of 6.6 wt% of H 2 . Up to 16 kJ/mol binding energy is obtained for an interlayer distance of $0.6 nm [94,97,98]. For larger pore widths or interlayer spacing, the binding due to the second graphene layer/surface weakens [95]. For smaller separations, the exchange repulsion reduce the free energy and becomes positive for interlayer separations of <0.5 nm. Thus energetically the H 2 -graphite system at shorter separations is purely repulsive. The gravimetric and volumetric H 2 storage capacities in the C 60 intercalated graphite (CIG) [99] approach the values obtained for Li-doped graphenes [93] at low temperatures. In another case, a capacity of 3 wt% of H 2 is estimated at ambient in $0.7 nm spaced carbon nanoscrolls followed by alkali doping [100]. First-principles calculations predict 3.4 wt% of H 2 and binding energy of (10-22) kJ/mol in the Li and organic molecules (benzene and tetrahydrofuran) co-intercalated graphite with an interlayer graphene distance of $0.77 nm [101]. Furthermore, the charged graphenes show enhanced H 2 binding energy up to $9 kJ/mol compared to $5.8 kJ/mol in a pristine graphene. Similar binding energies are also achieved in Li-doped single-walled carbon nanotube (CNT) pillared graphene structures with intertube and interlayer distance of 1.5 nm and 1.2 nm, respectively [102]. The molecular dynamics simulations estimate H 2 adsorption of up to 6 wt% at 77 K and 3 wt% at 300 K in the CNT pillared graphenes [103].
Furthermore, a detailed analysis and the theoretical limits of H 2 adsorption and storage capacity of slit pores is also presented [104][105][106]. For example, practicable H 2 adsorption capacity at room temperature is possible for graphene carbons with a slit-pore size between (0.8-1.1) nm, adsorption energy of P15 kJ/mol and a specific surface area of $2600 m 2 /g. The porous materials with micropores smaller than 0.7 nm are not useful for storage application because of low delivery efficiency and large pores (>1.2 nm) have difficulty in fulfilling the volumetric goals at room temperature. The optimal volumetric storage is expected when the pore volume accommodate 1 or 2 layers of H 2 .

Pillared graphene oxide structures
The loosely stacked layers in GO similar to graphite but with a much wider interlayer spacing (0.6-0.9) nm, would be more convenient to store H 2 . However, it is very important to consider how accessible is the surface area between graphene layers when the GO layers are stacked together with the presence of surface functional groups and lamellar water molecules [107]. Under normal conditions, GO usually contains different oxy-functional groups, thus not all of the available sp 2 carbons and interlayer spaces are accessible for gas sorption or storage. In recent investigations, some open GO structures are modelled with pillared GO layers, in which further increase in the interlayer distance to  [94]. (d) H 2 absorption as a function of graphite interlayer spacing at 298 K for 10 bar (circles) and 50 bar (squares) pressures [95]. about 1.1 nm is shown as optimal for practicable H 2 storage [107,108]. Burress et al. design a series of idealised pillared-GO model systems with a phenyldiboronic acid linker, called GO frameworks (GOFs) (Fig. 5) [107]. The variation in linker concentration produces different interlayer distances, pore sizes, pore volumes and surface areas. The GOF-32 structure is characterised by one linker per 32 graphene carbons predicted to hold ca. 6.1 wt% of H 2 at 77 K and 1 bar. In another case, Li-doped CNT pillared GO (Li-PGO) model structures with an interlayer distance of 1.1 nm and pore size of 2.3 nm show a gravimetric and volumetric H 2 capacity >10 wt% and 55 g/L, respectively at 77 K and 100 bar [109]. The interaction binding energy of H 2 in the system is $15.5 kJ/mol. First-principles computations on the Ti-grafted GO show gravimetric and volumetric H 2 storage of 4.9 wt% and 64 g/L, respectively [110]. Similar to the Li in Li-PGO, the Ti atoms bind strongly to the oxygen sites on the GO thus preventing them from clustering. It is also estimated that each Ti binds multiple H 2 molecules with suitable binding energy of (14-41) kJ/mol. The interlayer distance between 1.2 and 1.4 nm shows an optimal volumetric density of (64-74) g/L. First-principles calculations on the covalently bonded graphenes (CBGs) show enhanced H 2 binding energy of 20-150% or more compared to the isolated graphene [111]. The stable porous CBGs are formed through sp 3 carbons at edges of the graphenes. CBGs with adsorbed Ti atoms show P4 wt% of H 2 adsorption depending on the pore size and length of the metal chains.

Doped, functionalised and metal dispersed graphenes
Simple graphene-based nanostructures show weak binding energy thus low H 2 adsorption capacity at ambient conditions. Thus chemical doping or structural modifications in the nanostructures has been carried out. A considerable amount of theoretical work has been reported on the modified graphene structures with doping and/or surface metal (alkali: Li, Na, K; alkaline earth: Mg, Ca; simple: B, Al; transition: Sc, Ti, V, Ni, etc.; and noble: Pd, Pt) dispersion. This strategy increases the binding energy and adsorption of H 2 due to the polarisation of H 2 molecules by an electric field and/or hybridization of H 2 r or r * orbitals with transition metal d orbitals (the so-called the Kubas interaction [112]) [112][113][114]. Lochan and Head-Gordon reported interesting theoretical results, which describes that the ionic metal atoms, such as Li + , Na + , Mg 2+ and Al 3+ , hold up to six H 2 molecules with moderate to very strong binding energies between  kJ/mol, which further depends on the cluster size and complexes [115]. Furthermore, to avoid the issue of structural instability and poor reversibility in metal atomic dispersion (because of the large cohesive energy of bulk transition metals, the aggregation/clustering seems to occur instead of being atomistically dispersed) the structural or chemical defects/doping are introduced in the substrate [111,116,117]. It has been showed that the alkali-and alkaline earth-metal adsorption on graphene surface leads to a metallic state due to charge transfer (Fig. 6) [118,119]. For instance, as shown in Fig. 6, up to 16 wt% of H 2 is predicted in the Li and B adsorbed graphene with a binding energy of $24 kJ/mol [119][120][121][122][123]. H 2 adsorption of up to 14.4 wt% per molecule with an average binding energy of $40 kJ/mol in ethylene-Ti complex is estimated [124,125]. A tensile strain of 10% on graphene increases the adsorption energy of Li (Ti) atom by around 75% (71%) and the gravimetric H 2 storage up to 15.4 wt% (9.5 wt%) with an optimal binding energy of about 19 kJ/mol [126]. By controlling the corrugation of individual layers of graphene, the density functional theory (DFT) calculations estimate up to 8 wt% of H 2 [127]. The corrugation of the graphene sheet and the controlled inversion of its curvature show a fast storage (on convex sites) and release (on graphene concave sites) of H 2 at ambient conditions.  6. Top: (a) Various adsorption sites H1, H2 and H3 on the (2 Â 2) cell, (b) charge accumulation, Dq + , calculated for one Li atom adsorbed to a single side specified as H1, (c) same as (b) for one Li atom adsorbed to H1 site, second Li adsorbed to H3 site of the (2 Â 2) cell of graphene. Bottom: Adsorption sites and energetics of Li adsorbed to the (2 Â 2) cell of graphene and absorption of H 2 molecules by Li atoms. E L is the binding energy of Li atom adsorbed to H1 site, which is a minimum energy site. For H1 + H2 or H1 + H3 configuration corresponding to double sided adsorption, E L is the binding energy of second Li atom and EL is the average binding energy. For H1, H1 + H2 and H1 + H3 configurations, E1 is the binding energy of the first H 2 absorbed by each Li atom; E n (n = 2-4) is the binding energy of the last nth H 2 molecule absorbed by each Li atom, and En is the average binding energy of nH 2 molecules absorbed by each Li. Shaded panel indicates the most favourable H 2 absorption configuration [119].
The B-doping is found to form an electron-deficient structure that enhances the H 2 interaction energy [128][129][130][131]. The energy surface seems to extend over several graphene carbon sites, making the surface more heterogeneous. B-doping in graphene also enhances the Li and Ca bonding strength on the graphene [130][131][132][133]. The DFT calculations estimate approximately (13, 9.9 and 7.9) wt% of H 2 adsorption for the adsorbed Li, Al, and Ti atoms, respectively on the B-doped graphenes [133,134]. Lu et al. show a H 2 adsorption of (6.4 and 6.8) wt% (at 298 K and 100 bar) in a Li-and Ca-decorated B doped (pore rim of polyphenylene) porous graphene (Fig. 7) [135]. Up to 15 wt% of H 2 uptake is predicted in Be adsorbed on B-doped graphene with an average adsorption energy of $29 kJ/mol [136]. The H 2 adsorption capacity of 2.4 wt% is also proposed in Si-doped graphene [137].
H 2 adsorption of (6-9) wt% is estimated in the Ca-adsorbed/decorated graphene surfaces [132,[138][139][140][141]. The Ca adsorption on zigzag edge, porous or covalent-bonded graphenes (CBGs) is also reported [142][143][144]. CBGs doped with B and decorated with Ca clusters show up to 6 wt% of H 2 adsorption [144]. The Ca atoms prefer to bind more strongly on the sp 3 vertex sites of the pores than to planar graphene sites. This leads to formation of Ca chain structures and exhibit mixed characteristics of multipole Coulomb and Kubas interactions with H 2 molecules depending on the occupation of Ca s or d orbital-derived states. A similar Mg-doped GO shows 5.6 wt% of H 2 adsorption at 200 K [145]. The hydroxyl groups are reduced and Mg atom is bound to the surface of GO by the ring-opening reaction of epoxy in the form of -(C-O) x -Mg (x = 1, 2). On the surface the positively charged Mg sites and negatively charged O sites produce local electric fields separately to adsorb H 2 molecule with a binding energy of (14.5 and 24) kJ/mol at O and Mg site, respectively. The calculations on Al-doped graphenes also revealed interesting results for H 2 adsorption [146][147][148].
The H 2 uptake by light transition-metal (LTM = Sc, Ti, V) atom decorated graphenes has also been studied. Each of the LTM atom seems to adsorb on both sides of graphene and hold up to four H 2 molecules to give >6 wt% of H 2 adsorption with the average binding energies between (29-48) kJ/mol [149]. The Sc decorated on either armchair or zigzag edges of graphene nanoribbons show >9 wt% of H 2 adsorption [150]. Ni atomic dispersion on graphene yields about 3 wt% of H 2 but is expected to reduce experimentally due to the clusterization and nanoparticle formation [151].

Graphane based materials
Graphane is nothing but the graphene with alternately covalent bonded hydrogen atoms on each carbon atom on both sides of graphene sheet, as shown in Fig. 8 [152]. Structurally graphane is crumpled, rather than planar, because each hydrogen atom bonded to carbon pulls it a small distance out of the plane. First-principles studies show that the chemisorption of hydrogen on both sides of graphene, i.e., with full coverage of hydrogen (C n H n , i.e., 8 wt%) is the lowest energy structure due to the strain associated with the sp 2 to sp 3 conversion [153]. Due to its stable structure and small size, graphane has been also considered as a potential candidate for H 2 storage if doped with alkali-and alkalineearth metal (M) [154]. The H 2 adsorption mechanism on these doped graphanes is explained by considering the fact that positively charged M + ions polarise the H 2 molecules, which are then held to the M + ions by the vdW attractive forces. Ab initio calculations by Antipina et al. show that the graphane with chemically bonded alkali metals adsorbs up to 12 wt% of H 2 , with a binding energy of $19 kJ/mol [154]. The calculations by Hussain and Ahuja's group also predict practicable H 2 adsorption in the Liand Ca-doped graphane with and without strain [152,155,156]. The optimised structure for the case of maximum coverage (5 H 2 molecules, each on both sides of Ca atoms) is shown in Fig. 8.

Hydrogen spillover
The H 2 adsorption by hydrogen spillover is a hydrogen dissociative atomic adsorption, and it has been found to be other alternative to enhance H 2 adsorption in the carbonaceous materials. The spillover is expected in some of the graphene based materials with functionalized/doped structures [157][158][159][160][161][162]. For example, the stability of hydrogenation states of graphene and conditions for hydrogen spillover are described by Han et al. [161]. The mechanism in metal-doped carbon materials and other solid porous media is explained by George et al. [162]. In brief, it can be explained in three elementary steps as shown in Fig. 9; (i) the activation or dissociation of H 2 molecules on catalytic metal sites, (ii) a transition of atomic hydrogen across the metal-receptor interface, and (iii) the migration of atomic hydrogen throughout the receptor surface. The whole process is denoted as 1, 2; 3, 4; and 5, 6, respectively. The receptor, 'R' is considered inert towards the unactivated H 2 and thus only accepts the activated atomic hydrogen that migrates from the catalyst, 'C'. The receptor surface is expected to be the key that simultaneously provides appropriate thermodynamic stability (binding energy) and kinetic mobility (migration). Fig. 9 also shows stability of the hydrogen chemisorption state of graphene in comparison with the chemical potential of H 2 gas.
Effective dissociation of H 2 molecules into H atoms is observed with the firmly bound catalytic Pt 4 metal cluster on B-doped graphene [160]. The H atoms then migrate from the bridge site of the H-saturated Pt 4 metal to the supporting graphene sheet. By DFT calculations, Cabria et al. reveal that the adsorption and dissociation of H 2 on Pd clusters on the graphene is due to either (a) the adsorption of the H 2 in an activated state or (b) the dissociation of the H 2 and the chemisorption of the two hydrogen atoms [163]. In an activated state, the H 2 bond length is slightly stretched but not broken. The binding energies of the activated H 2 are between (0.4-1) eV/molecule, close to the estimated energies required to obtain reversible H 2 storage at room temperature and moderate pressures. Tight-binding model describes that the C-Pd-H 2 interaction is much weaker than the C-Pd-H bonding [164]. The DFT studies also suggest that the graphene-like materials comprising of curved graphene fragments and many edge atoms, such as microporous carbon, strongly interact with the spillover hydrogen atoms [165]. At 300 K, Li anchored on the graphene vacancies show 6.2 wt% of H 2 dissociative adsorption with a strong binding energy of $84 kJ/mol [166]. Lueking et al. show that both the B substitutionally doped graphene and hydroxylated graphene satisfies the constraints for reversible roomtemperature hydrogenation [167]. A significant enhancement in the binding of H 2 on N-doped graphenes is expected through dissociative adsorption and diffusion on the surface with an externally applied field [168,169]. Under an applied electric field, the N-doped graphenes exhibit a relatively low dissociative adsorption energy barrier for H 2 and low diffusion energy barrier of H atoms.

CH 4 and CO 2 : sorption, storage and separation
Similar to H 2 adsorption, the graphenes with ripples, defects, pores, functional groups, metal dispersion or doping and pillaring also show favourable CH 4 and CO 2 adsorptions [170][171][172][173][174][175][176][177][178][179][180][181]. The GCMC simulations on fullerene intercalated graphene nano-containers show enhanced CO 2 adsorption with increasing fullerene concentration. The pore with an effective diameter of ca. 0.5 nm is found to be optimum for the effective CO 2 adsorption at low pressures due to the energetically favourable surface interactions [170]. In a relatively large slit-pore, (1.0-2.4) nm the high CO 2 adsorption occurs at relatively high pressures due to multilayer molecular adsorption, similar to the H 2 adsorption characteristics discussed above. In another study, molecular simulations demonstrate the CH 4 adsorption in multilayer graphene nanostructures with and without Li-doping (MGNs and Li-MGNs) and function of interlayer distance [171]. The MGNs and Li-MGNs with an interlayer distance of (0.34-0.51) nm (0.34 + 0.17) show no CH 4 adsorption due to steric effect of the adsorption space, whereas an interlayer distance between (0.68-2.04) nm yield a good CH 4 adsorption (Fig. 10). A maximum uptake of 310 v(STP)/v of CH 4 is obtained at 35 bar, with an adsorption energy of 27.2 kJ/mol. The structures also show a higher H 2 S-CH 4 adsorption than CO 2 -CH 4 in case of biogas separation. Several main influencing factors for CH 4 storage and acid gas separation in MGNs and Li-MGNs are identified, including the binding energy between gas molecules and substrates, the optimal layer spacing for CH 4 adsorption,  DFT calculations on the graphene layer modified with Ti at high metal coverage (C 2 Ti) show a dissociative and associative adsorptive behaviour for both CO 2 and CH 4 , respectively at 300 K and 1 bar [172]. First-principles calculations on the Ti-decorated GO (Ti-GO) show a selective adsorption of CO over CO 2 , N 2 , CH 4 either in single component or mixture with computed binding energy of $70 kJ/mol for CO, $20 kJ/mol for N 2 and <5 kJ/mol for CO 2 and CH 4 [173]. The strong interaction between CO molecule and Ti is a result of dative bond, i.e., hybridization between an empty d orbital of Ti and an occupied p orbital of CO. The plane-wave electronic DFT calculations suggest that CO 2 binds stronger to the defective site on graphene surface (with one carbon atom missing, i.e., monovacancy) than a perfect graphene with a physisorption energy of $20 kJ/mol, approximately 4 times stronger than a perfect graphene surface [174]. The adsorption and reactivity of CO 2 on a defective graphene sheet shows strong physisorption energy of $136 meV [175]. Chemisorption of CO 2 is expected by forming a lactone group with the C atoms surrounding the vacancy after overcoming an energy barrier of $1 eV. Further dissociation of the CO 2 occurs through the formation of stable epoxy groups with $614 meV. Ab initio and molecular mechanics calculations show that the periodic nature of the vdW potential energy stored between CH 4 and perfect sheet is altered due to the insertion of vacancies (missing lattice atom) and sinusoidal ripples. The CH 4 molecule avoids to be around vacant cites and on top of the peaks [176]. The produced vdW energy surface indicates that around the vacant cites the energy increases. On the perfect graphene, DFT calculations reveal that the adsorption of CH 4 is essentially driven by dispersion interaction with a reasonable binding energy of 0.17 eV [177].
First-principles simulations show the large CO 2 uptake capacities ($0.4-0.6 g CO 2 /g sorbent) by decomposition-mediated adsorption in Ca-decorated graphenes [178]. In the CO 2 uptake process, decomposition into CO and formation of CaO is found to be thermodynamically favourable. Thus enhancement in CO 2 adsorption is attributed to a secondary process, where further absorption of CO 2 in a CaO leads to formation of CaCO 3 . The Ca-decorated graphene also shows an improved CO 2 / N 2 selectivity, because of stronger binding of CO 2 than N 2 , e.g. 2.731 eV/CO 2 against 0.645 eV/N 2 in case of 12.5% Ca-graphene. Ohba and Kanoh show the edge effects of graphene on Ar, CH 4 , N 2 , CO 2 , and H 2 O gas adsorption [179]. The edge sites provide strong Coulombic interactions thus molecules with a strong dipole or quadrupole moment (H 2 O or CO 2 ) are selectively adsorbed on the edge sites. The dispersion interaction-dominated molecules (Ar, CH 4 , and N 2 ) are selectively adsorbed on the basal planes. Furthermore edge functionalized graphenes with the polar groups, including -COOH, -NH 2 , -NO 2 and -H 2 PO 3 , show enhanced CO 2 and CH 4 adsorption due to strong binding of activating exposed edges and terraces [180,181].

Gas separation membranes
Pristine and porous 2D graphene membranes with and without functionalization have been investigated for gas purification [182]. The pristine graphene is impermeable to He and H 2 [183][184][185]. Calculations by Miao et al. show that protons can readily pass through a perfect graphene sheet with a low tunnelling barrier compared to a substantially higher barrier for H 2 [184]. Fig. 11 shows initial and final energy states for the tunnelling of an H atom through the centre of a hexagonal carbon ring (hollow site). H 2 is physisorbed at a distance of 0.286 nm with a high energetic activation barrier of 2.86 eV. The tunnelling of an H atom through the benzene ring causes the ring to expand and then contract between a diameter of 0.285 nm and 0.291 nm in the initial/final and transition states, respectively. The strong orbital overlap between the electrons from the H atom and the graphene at À21.62 eV and 2.08 eV for C-2s and H-1s, and C-2p and H-2s respectively, revealing a strong interaction between the H atom and graphene sheet at the transition state (Fig. 11b). The penetration barrier decreases exponentially with the size of the defects, expressed by the number of carbon atoms involved in formation of defect. The potential energy as a function of the distance of an H atom from graphene indicates that a larger defect size result in a smaller penetration barrier. For example, the barrier is reduced largely to 0.66 eV for DV (divacancies) and 1.17 eV for 555-777 (one reconstruction from a DV) defects compared to the 2.64 eV of a perfect graphene. For this defect sites the stable states of H atom found at different distances between 0.12 nm and 0.15 nm.
First-principles DFT calculations estimate that the H 2 /CH 4 diffusion selectivities are on the order of 10 8 and 10 23 for the N-and all H-functionalized pores, respectively [186]. The pores are created by the removal of two neighbouring benzene rings on a single-layer graphene. A CO 2 /N 2 selectivity of around 300 with a free energy barrier for permeation of 24.7 kJ/mol and 9.6 kJ/mol for CO 2 and N 2 , respectively, is reported for the same porous graphene using classical molecular dynamics simulations [187]. A sandwich-like model of porous graphene with the dimensions of 10 nm Â 10 nm and the nanopore density of 0.04 per nm 2 shows a linear dependence of flux on CO 2 pressure. The smaller pores of graphene with periodically ''one missing'' phenyl rings (the polyphenylene unit cell) and line-defect-graphene with missing octagons show a high selectivity for H 2 (e.g., 10 17 for H 2 /CO 2 and 10 23 for H 2 /NH 3 ) separation from O 2 , N 2 , NH 3 , CH 4 , CO 2 and CO, since CO 2 , CO and CH 4 has larger kinetic diameter and H 2 is about width of the pore [188][189][190]. A controllable selectivity and permeable separation of H 2 /N 2 is demonstrated in a series of porous graphene membranes with different pore sizes and shapes created by removing 11-16 carbons [191]. For example, the pore size of $0.65 nm (pore-16) show best selectivity of N 2 from H 2 because of the vdW interactions with graphene membrane that leads N 2 molecules to distribute within a single molecule layer around the graphene surface. Gas permeation in various limits of gas diffusion, surface adsorption, or pore translocation as the rate-limiting step is described by analytical expressions [192].
First-principles DFT and MD simulations on 0.135 nm pore membrane show a high selectivity for H 2 over other gases with the energy barrier of (0.12, 0.26, 0.25 and 0.82) eV for H 2 , N 2 , CO and CH 4 , respectively [193]. Clearly, it is the electron overlap between the gas molecules and the pore that causes the energy barrier and hinders the molecules from passing through the pores. Thus the energy barrier for gas molecules passing through the membrane generally increases with decreasing pore size or increasing molecule kinetic diameter. Lu et al. demonstrate a high permeable selectivity for H 2 relative to CH 4 , CO, and CO 2 in a B or N doped pore rim of (polyphenylene) of porous graphene (PG, one hexagon missing) [135]. The 4B-PG shows a highest selectivity (1.2 Â 10 78 ) for H 2 /CH 4 much higher than the PG (6.0 Â 10 58 ) with respective energy barrier of 0.91 eV and 0.68 eV. Substitutional doping leads to a change in the original bond length of PG, where stretching and shortening of the bond is observed by B (BH) and N (NH), respectively. Thus the 3B-PG and 4N-PG with change in pore size possess high selectivity (5.0 Â 10 37 and 2.0 Â 10 36 ) for H 2 /CO (20 times) and H 2 /CO 2 (5 orders) respectively, compared to the PG. In the N-doped pore rim graphene (replacement of the unsaturated two C atoms with two N) a Lewis-acid-base type interaction leads to a bound state of CO 2 with a binding energy of about 21 kJ/mol [194], which show no barrier for N 2 . Due to similar kinetic diameters and no electric dipole moments, it is difficult to separate CO 2 from N 2 ; and pore functionalization is needed to enhance the electrostatic interaction. The modified graphene membrane with -CH 3 , -OH, -NH 2 , -COOH shows a higher permeability for both N 2 and CO 2 compared to the unmodified one [195]. The results show that the porous graphene membrane with all-N modified pore-16 exhibits a higher CO 2 selectivity over N 2 ($11) due to the enhanced electrostatic interactions compared to the unmodified graphene membrane. Functional groups also increase the active surface area thereby enhancing the uptake and diffusion onto the membrane surface for more permeation. The CO 2 adsorption ability of functionalized graphene is in the following order: -OH > -COOH > -NH 2 > -CH 3 . The porous membrane with carbon atoms bonding to three carbons and to two carbons with hydrogen generates nonuniform charge surfaces and strongly interacts with polar adsorbate molecules. A high selectivity for SO 2 adsorption (SO 2 /N 2 , SO 2 /CO 2 ) is reported on fluorinated porous graphene due to reduced dispersive interactions, e.g., for CO 2 [196]. Furthermore, in a porous graphene the enhanced diffusion rates of H 2 , O 2 and CO 2 up to 2, 5, 9 and 7, 13, 20 orders of magnitude are obtained by application of 10% of the uniaxial tensile strain and symmetrical stretch strain, respectively [197].
Clearly theoretical simulations have offered further insights into the mechanisms of molecular interactions with graphene-based structures. Pristine graphene only show weak binding with most of the molecules. Enhancement of binding interactions can be achieved by careful design of microporosities, surface doping, defect engineering, and high density of edge/boundary defects. However, the binding energy needs to be retained at a level to cause ''chemical activation'' of the molecules, typically (10-40) kJ/mol, rather than their complete decomposition. Separation of such molecules is another challenge, given the little differences in the kinetic diameters of the molecules of interest. Modifications that can cause preferential binding of a certain molecules are desirable to promote selectivity of the sorbent materials. The following section will examine some of the experimental developments along those directions.

Chemically reduced/thermally exfoliated graphenes
Graphenes in the bulk quantities have been synthesised by liquid phase oxidation-reduction/exfoliation of graphite (Fig. 1a). The degree of oxidation and exfoliation or reduction conditions yield individual single-layer graphenes to loosely packed multi-layer graphenes with a wide range of pore structures, different degree of residual functional groups and accessible specific surface areas (unless otherwise specified, hereafter the term 'specific surface area (SSA)', is generally determined using the Brunauer, Emmett and Teller (BET) [198] method from nitrogen gas adsorption-desorption isotherm measurement at 77 K). In many cases, the reduction of GO and agglomeration limits the microstructure and SSA. For example, in a report by Ma et al., the graphenes obtained by thermal exfoliation of GO followed by hydrogen reduction show a much limited SSA (156 m 2 /g) due to the severe agglomeration of crumpled graphene sheets [199]. The sample adsorbs about (0.4 and 0.2) wt% of H 2 at 77 K and 290 K under 1 bar and 60 bar, respectively. The exfoliated graphenes show an enhanced heat of H 2 adsorption of $6 kJ/mol compared to carbon nanofibers and active carbons [200].
A considerably improved SSA and gas sorption in graphenes have been obtained from different post-synthesis treatments of GO. Srinivas et al. report an enhanced SSA (640 m 2 /g) in a hydrazine hydrate reduced colloidal suspension of exfoliated GO (Fig. 12) [201]. The sample is characterised with interconnected, wrinkled and disordered graphene sheets with regions of agglomerated multiple layers and a partial restoration of graphitic structure. The gravimetric H 2 uptake measurements at  K and up to 10 bar show a heat of adsorption of $6 kJ/mol at low coverage, and the maximum adsorption of 1.2 wt% at 77 K and 0.1 wt% (and up to 0.72 wt% by extrapolating pressure to 100 bar) at 298 K (Fig. 12). A thermally reduced GO by simple out-gassing at 423 K overnight shows an improved H 2 adsorption of about 1.2 wt% at 77 K and 1 bar with a SSA of 477 m 2 /g due to enhanced microporosity and pore volume of $1.0 cm 3 /g [202]. In another case, a mild chemical reduction of GO with glucose yields a highly enhanced SSA up to 1200 m 2 /g with a mesoporous structure [203]. Surprisingly, a high H 2 adsorption of 2.7 wt% at 298 K under 25 bar is obtained, a much higher than for most of the carbon nanostructures. Lee and Park report H 2 uptake of up to 1.8 wt% at 298 K under 100 bar (Fig. 13) in a series of exfoliated GO's by CO 2 pressure swing [204]. GO exfoliation is carried out at 373 K under high pressure of CO 2 up to 40 bar followed by rapid release to vacuum to produce graphenes with a maximum SSA of 547 m 2 /g. Magnesium combustion derived graphene sample shows a more or less linear H 2 uptake with increasing H 2 pressure up to 300 bar with the final capacity of 0.9 wt% at 293 K [205]. Bulk graphene sample is produced via high temperature combustion of magnesium in a CO 2 atmosphere using dry ice black, which exhibits a more ordered carbon lattice than other bulk graphenes derived from GO.
Rao's group prepare various types of graphenes from thermal exfoliation of GO (EG), pyrolysis of camphor (CG), and conversion of nanodiamond (DG) [206,207], which shows a wide range of SSA: 46 m 2 /g, (280-1013) m 2 /g and (639-1550) m 2 /g for CG, DG and EG, respectively. H 2 adsorption at 77 K and 1 bar reveals an average linear trend with the SSA, in good agreement with other porous carbons. From this trend, the single-layer graphene, i.e. for the SSA of 2630 m 2 /g, is estimated to hold about 3 wt% of H 2 , in other words, about 1.2 wt% of H 2 for every 1000 m 2 /g of SSA, at 77 K and 1 bar. In addition, a considerably higher H 2 adsorption up to 3 wt% at 298 K and 100 bar is reported. Guo et al. obtain a highly improved H 2 adsorption of 4.0 wt% at 77 K and 1 bar in a hierarchical pore structure graphene with a SSA of 1305 m 2 /g [208]. Briefly, the sample is synthesised by exfoliation of GO at 423 K under vacuum followed by further heat treatment at 873 K. The pore structure analysis from the N 2 (at 77 K) and CO 2 (at 273 K) adsorption-desorption isotherm measurements reveal a hierarchical mesopore and ultramicropore structure. This ultra-microporosity leads to an enhanced H 2 uptake at low-pressure as well as a high heat of H 2 adsorption, 9.5 kJ/mol.
The graphene nanoribbons (GNRs) with in-plane and edge sp 2 -and sp 3 -hybridized carbons show a strong irreversible adsorption of CO 2 (0.22 mmol/g) at 303 K and 1 bar [209]. GNRs are synthesised through CVD via aerosol pyrolysis using ferrocene/thiophene/ethanol precursors. Adsorption of 6.4 mmol/g of CO 2 at 30 bar and 298 K is obtained in a CO 2 pressure swing exfoliated GO with a SSA of 547 m 2 /g and a total pore volume of 2.47 cm 3 /g [204]. Furthermore, a greatly enhanced CO 2 adsorption of 21.6 mmol/g at 11 bar and 298 K is reported in a hydrogen exfoliated graphene (HEG) with a wide distribution of mesopores and SSA of 443 m 2 /g [210]. Surprisingly even higher CO 2 Fig. 13. H 2 uptake (top) at 298 K and N 2 (bottom) at 77 K adsorption and desorption isotherms of the various exfoliated graphene oxide samples (filled and hollow symbols correspond to the adsorption and desorption profiles). The exfoliated GO samples obtained are designated by the applied pressure (unit: bar), i.e., GO (for the unmodified sample), E-P-10, E-P-20, E-P-30, and E-P-40 [204].
adsorption of up to 56 mmol/g is measured in a vacuum exfoliated GO compared to the maximum of (8, 15 and 39) mmol/g for zeolite-13X, activated carbon and conventional graphene nanosheets, respectively at 298 K and 30 bar (Fig. 14) [211]. The samples exhibit corresponding SSA and total pore volume of (480, 839, 1453 and 701) m 2 /g and (1.730, 0.396, 1.382 and 0.727) cm 3 /g, respectively. The high CO 2 uptake isotherms in graphene samples exhibit unconventional adsorption behaviour than other porous solids. At low or moderate pressures up to 10 bar, the CO 2 adsorption in graphenes is not comparatively better than others. The improvement at high pressures is linked to the higher interior void volume.
The chemical functionalization and surface grafting of graphenes with basic groups is also demonstrated to obtain increased CO 2 binding and adsorption capacities through enhanced Lewis base-acid interactions [212,213]. Amine-grafted GO synthesised from different amines, including ethylenediamine, diethylenetriamine and triethylene tetramine and aqueous GO by vigorous stirring and reflux at 353 K followed by washing shows about 1.2 mmol/g of CO 2 adsorption in the breakthrough experiments [212]. An unprecedented CO 2 uptake of over 70 mmol/g of CO 2 (at 10 bar and 298 K) is obtained in the HEG-polyaniline composites [213]. However due to the very strong chemical binding of CO 2 , the heat treatment under vacuum is necessary to desorb the CO 2 .

Layered graphenes: simple expansion to pillaring with metals and organic ligands
As described in the theoretical predictions there is an optimum interlayer distance between graphene sheets to maximise the H 2 uptake. Such a range of distance reflects the energetically favourable interaction of one H 2 layer between graphene layers. In order to demonstrate the importance of an interlayer distance for effective H 2 storage a variety of hybrid structures have been designed. Below we summarise the experimental results on graphene layer structures synthesised through intercalating various chemical species starting with simple metals/oxides, molecules to organic ligands and combination of these. The best example for this case is the well-known graphite intercalation compounds (GICs) [214][215][216][217]. GICs are highly ordered periodic layer structures, known for their very large graphene-graphene inter-layer spacing, made up of layers of foreign atoms/molecules sandwiched between single to n-graphene layers, depending on the stage-n compound. Stage-n represents the n-number of graphene layers present between two intercalated layers. For example, the stage-1 and stage-2 potassium GICs, KC 8 and KC 24 contains one and two graphene layers between two adjacent potassium intercalated layers (Fig. 15) [218]. This results in increased interlayer distance over 0.540 nm, sufficient to reversibly adsorb H 2 at low temperatures. In particular, the series of stage-2 GICs of stoichiometry MC 24 (M = K, Rb, Cs) are reversibly and rapidly adsorb H 2 up to $2H 2 /M, without dissociation and behaves as a molecular sieve (Fig. 16) [218,219]. The charge transfer to the graphene planes lead to enhanced binding energy for H 2 adsorption (8.4 kJ/mol) in KC 24 that is about twice to bare graphite.
An enhanced H 2 adsorption is observed in the herringbone graphite nanofibers (GNFs) when expanded the graphite lattice using intercalation of concentrated sulphuric/nitric acid mixture R19.11°structure, frequently used as an approximation of the low-temperature in-plane arrangement of KC 24 . This model has a stoichiometry of KC 28 . The sorption site (blue square) and neighbouring sites (red triangles) are depicted for a single trigonal subunit cell. (b) Schematic side view of a stage-2 alkali-metal graphite intercalation compound [218]. followed by a thermal shock [220]. The SSA of 47 m 2 /g in GNFs is increased to (67 and 555) m 2 /g when treated at 973 K and 1273 K, respectively. Thus expanded GNFs show a multifold increase in H 2 adsorption (1.2 wt% and 0.29 wt% at 77 K and 300 K, respectively at 20 bar) compared to unexpanded GNFs. In order to achieve an optimum interlayer distance, Kim et al. obtain the H 2 isotherms in a series of pillared-GO at 77 K and 1 bar [221]. Interlayer distance is controlled using pillaring diaminoalkanes (NH 2 (CH 2 ) n NH 2 , n = 2, 6, 10) and subsequent thermal annealing between (300-598) K. The structures show a maximum H 2 uptake for an optimum interlayer distance of 0.63 nm, in good agreement with the first-principles calculations for graphitic materials. In their earlier work [222], the GO itself used to tune the interlayer distance without inserting other pillaring supports. A careful reduction in a controlled thermal annealing not only retains the layered structure and also shows improvement in SSA, up to 192 m 2 /g and pore volume, 0.689 cm 3 /g, compared to 57 m 2 /g and 0.076 cm 3 /g for the initial GO. Nuclear magnetic resonance (NMR) measurements confirm a considerable reduction of O-and OH-functional groups in the annealed GO. As high as 4.8 wt% of H 2 adsorption at 77 K and 90 bar is measured in the annealed GO with an interlayer distance of 0.64 nm. A considerable reduction in the H 2 uptake is observed with a little increase in layer spacing over 0.64 nm.
In another approach, the porous networks of graphenes engineered by bridging the graphene layers and metal-oxide particles through surface functional groups. For example, the GO-transition metaloxide composites; GO/V 2 O 5 and GO/TiO 2 are developed through binding of C-OH species of GO and the oxygen on the transition metal-oxide via a dehydration reaction [223]. The hydrothermally designed porous graphene-Mn 3 O 4 nanocomposites from the aqueous GO dispersion and MnO(OH) 2 colloids yield an enhanced SSA up to 680 m 2 /g [224]. The composite shows $2.5 mmol/g of CO 2 adsorption at ambient pressure and temperature compared to $1.75 mmol/g in a controlled graphene. The solvothermally synthesised graphene-iron-oxide hybrid structures reveal an enhanced microporosity with SSA ranging between (418 and 901) m 2 /g [225]. The nanoporous graphene-polyoxometalate (GPOM) hybrid structures are synthesised from in-situ hydrazine hydrate reduction of GO aqueous solution and cross-linker phosphomolybdic acid [226]. POM is a polyatomic ion, usually an anion that consists of three or more transition metal oxyanions linked together by shared oxygen atoms to form a large, closed 3D framework. The polynuclear metal-oxo structured POM is a versatile building block cluster for construction of functional hybrid materials. The hierarchical pore GPOM shows a pore volume of 0.57 cm 3 /g and SSA of 680 m 2 /g, considerably larger than SSA of 23 m 2 /g for GO and 8 m 2 /g for POM. A highly increased interlayer spacing of 3.2 nm and 3.6 nm is achieved in GPOM and o-GPOM (oxidation of GPOM by hydrogen peroxide), respectively from 0.4 nm in a reduced GO. H 2 adsorption of 0.8 wt% and 1.3 wt% in GPOM and o-GPOM respectively, is obtained at 77 K and 1 bar. From the results, it is worth noting that the (micro)pore size distribution, adsorption enthalpy, exposed metal-sites and functional groups play a critical role in enhancing H 2 adsorption.
Using a well-known reactivity between boronic acids and hydroxyl groups, Taner's group [107,227] develop a novel pillared GO structures, called GO frameworks (GOFs) by linking the various phenylboronic acids between GO layers (Fig. 17). For the past few years, boronic acids have been considered as versatile building blocks for the construction of a variety of molecular architectures [78,228]. The interaction between boronic acids and diols has been used recently to construct a variety of crystalline porous frameworks, called covalent organic frameworks (COFs) [78]. Thus the rich surface oxy-functional groups on the graphene leads to the development of GOFs by forming strong boronate ester bonds as shown in Fig. 17. The structures are synthesised through a solvothermal method using methanol solution from various concentrations of GO and phenylboronic acid. GOFs show a systematic increase in the interlayer distance from 0.75 nm to 1.05 nm and SSA from 10 m 2 /g to 470 m 2 /g. About 1 wt% of H 2 uptake at 77 K and 1 bar with relatively high ($9 kJ/mol) heat of H 2 adsorption at low coverage is achieved. The H 2 adsorption capacity for a given SSA and heat of adsorption appears twice as large as typical porous carbons ( [82,199,201], typically about 0.5 wt% of H 2 uptake at 77 K and 1 bar for every 500 m 2 /g of SSA and heat of adsorption of 4-6 kJ/mol). GOFs also show a good room temperature CO 2 adsorption of $2.7 mmol/g at 4 bar with again high heat of CO 2 adsorption of 35 kJ/mol. The micropore structure with slit-like pores and strong surface interactions with functional groups are favourable factors enhancing their capacity for gas adsorption and binding. Similarly, cross-linked and functionalized graphene sheets from diazonium (product 1) and tert-butylaniline or 4-chloroaniline (product 2) as reagents in chlorosulphonic acid or oleum are synthesised aiming at improved H 2 adsorption [229]. Products 1 and 2 show a calculated interlayer spacing of $0.96 nm and $0.76 nm and SSA of $440 m 2 /g and 350 m 2 /g, respectively. Enhanced H 2 adsorption of up to 1.6 wt% at 77 K and 2 bar is obtained compared to $1 wt% in the starting exfoliated graphenes. The enhanced H 2 adsorption again suggests importance of tuning the interlayer separation, microporosity, and functionalization. Zhou et al. design a porous graphene structures by cross-linking non-planar terpyridine complexes using azide-alkyne click reaction [230]. The structures show up to 440 m 2 /g of SSA and a CO 2 adsorption of 2.6 mmol/g at 273 K and 1 atm.
The hybrid monolith aerogels of chitosan (CTS, an environmentally-benign biopolymer) with GO, prepared using freeze-drying, show dramatic increase in both the porosity (from 153 m 2 /g to 415 m 2 /g) and cyclic CO 2 uptake (from 1.92 mmol/g to 4.15 mmol/g at 25°C and 1 bar) for a GO loading of 20 wt% [231]. The DAB poly(propyleneimine) dendrimer (DAB) cross-linked GO-based hybrids display good CO 2 adsorption of $2 mmol/g (P/P 0 = 0.35) upon wetting [232]. Under humid conditions the sample exhibits $3 times the adsorption capacity with fast kinetics, compared with the dry state, due to the presence of amino groups of DAB which can form carbamates through zwitterions (primary amines) and bicarbonates (tertiary amines). The formation of these groups is enhanced in the presence of a base (hydroxyl groups of water).
In a work by Matsuo et al., the pillared graphenes are synthesised by pyrolysis of various sylsesquioxane bridged GOs [233]. The structures are highly microporous in nature (calculated pore widths of about 0.7 nm) and show an interlayer spacing between 1.34 nm and 1.6 nm and SSA up to 940 m 2 /g, which leads to an H 2 excess adsorption of 0.6 wt% at ambient temperature and 200 bar Fig. 17. Representations of (a) boronic ester [107] and (b) GOF formation by benzenediboronic acid pillars in GO, and the H 2 uptake in respective structures [227].
with a high heat of H 2 adsorption of (8-11) kJ/mol. A further enhanced H 2 adsorption of 1.4 wt% and 2.6 wt% at room temperature and 50 bar is obtained in a liquid crystal (LC) route prepared macroscopically ordered layer-by-layer 3D framework of GO and GO-MWCNTs composites, respectively (Fig. 18) [234]. MWCNTs act as 1D spacer between graphene sheets that result in large increased interlayer spacing up to $7 nm. The high H 2 adsorption is attributed to the synergistic effect of the well-spaced GO sheets and debundling of MWCNTs to enhance the accessible surface area. It is worth noting that the $1.4 wt% of H 2 in a LC-GO with its interlayer spacing of around 0.8 nm is also comparatively higher than an earlier reported value of 0.2 wt% at 77 K in GO [107].
The enhanced H 2 uptake capacities are also realised in graphene incorporated clay materials [235,236]. The samples are prepared by impregnation of the natural clay (montmorillonite and sepiolite) substrates by caramel in aqueous media, followed by a thermal treatment in the absence of oxygen, gives rise to graphene-like materials, which remain strongly bound to the silicate support. These carbon-clay nanocomposites shows $0.4 wt% of H 2 at 298 K and 200 bar related to the carbon mass. These uptakes are still far from the predicted value of (3-4) wt% from model simulations. A very high isosteric heat of H 2 adsorption of 14.5 kJ/mol shows a strong stabilization of the H 2 molecule within the structure. The controlled clays without graphene, thermally treated under the same conditions did not show any appreciable H 2 adsorption. The samples also show about (0.6 and 1.7) wt% of H 2 uptake related to the total mass of the system and carbon mass, respectively, at 77 K.
In another case the enhanced adsorptive gas storage is achieved by pore engineering of highly porous MOFs with GO [237][238][239][240]. The MOFs are a type of crystalline and highly porous solids made up of metal clusters linked with organic ligands. For example, Petit et al. synthesise the MOF-GO composites by solvothermal method, with the addition of GO up to 50% by weight to the MOF precursors [237]. The successful retention of porosity suggesting that the building process of the composites occur via the reaction/binding of the copper dimers from the HKUST-1 (Cu-MOF, also known as Cu-BTC) with/to the functional groups in GO (epoxy, carboxylic, hydroxylic, sulphonic). At the optimal GO content, the composites show about 15% and 18% increase in a micro-and total porosity, and SSA up to 1000 m 2 /g compared to a 900 m 2 /g in HKSUT-1. While the composites with the higher GO contents show smeared SSA (600-700) m 2 /g due to the formation of nonporous reduced graphene stacks from the excess, unreacted GO. H 2 adsorption of (2.0-2.5) wt% at 77 K and 1 bar in these structures is consistent with the enhanced microporosity. Later, Liu et al. report a further enhancement in SSA and H 2 adsorption as well as CO 2 , CH 4 and N 2 adsorptive separation in a similar Cu-MOF and GO composites, named as CG (Fig. 19) [238]. The authors also reveal more insights in the growth mechanism of the Cu-  1305, 1470, 1532, 500, and total pore volume (cm 3 /g), 0.518, 0.610, 0.645, 0.312 for Cu-MOF, CG-3, CG-9, and CG-15, respectively. In CG-n, n is a weight percent of GO added to the MOF precursors before synthesis [238].
MOF crystals on GO substrate in the composite through DFT calculations (in a VASP). The results show strong interaction energy of Cu to the GO defect of 15.12 eV, more favourable than those on the -OH (3.92 eV) and -O-groups (1.60 eV) of the GO surface. Furthermore, the epoxy groups and the defects of the GO layers act as seed and termination sites, respectively for formation of nanosized and welldispersed Cu-BTC on the layers of GO (Fig. 19). Thus the composites show about a 30% increase in CO 2 and H 2 adsorption (from 6.39 mmol/g of Cu-BTC to 8.26 mmol/g of CG-9 at 273 K and 1 bar for CO 2 ; from 2.81 wt% of Cu-BTC to 3.58 wt% of CG-9 at 77 K and 42 bar for H 2 ). A similar CO 2 adsorption is observed in a GO-Cu-BTC with stronger interaction energies (activation energies of (68.6 and 56.7) kJ/mol for CO 2 on the composite and Cu-BTC, respectively) and enhanced CO 2 /CH 4 adsorption selectivity, twice that of Cu-BTC for an equimolar CO 2 and CH 4 mixture [239]. Furthermore, the composites of Cu-BTC and GO modified by urea show enhanced CO 2 adsorption of 4.23 mmol/g [240]. The incorporation of aminated GO into a MOF structure shows synergistic effect; the porosity modification with new micro-pores at the interface and basic groups on the surface, which are expected to increase the physical adsorption forces and acid-base interactions, respectively. This would not only increase the adsorption capacity but also the selectivity.
Sandwich-like graphene-silica sheets with incorporation of polyethyleneimine (PEI/G-silica, with about 12 wt% graphene and 60 wt% PEI, amine based solids for CO 2 adsorption) show a high CO 2 adsorption of 4.3 mmol/g at 348 K under 7.4 bar with good cyclic stability [241]. The G-silica sheets obtained through the hydrolysis of tetraethylorthosilicate on the surface of GO with the aid of a cationic surfactant (cetyltrimethyl ammonium bromide) followed by pyrolysation at 973 K under Ar. PEI impregnation is made via a solution route using methanol solvent. PEI/G-silica sheets show a pore size of (0.7-2.0) nm with a SSA of 32 m 2 /g, much smaller than that of G-silica sheets ($2.0 nm and 930 m 2 / g). Clear advantage of graphene sheets is observed in an enhanced CO 2 adsorption at low equilibrium pressures (<1 bar), whereas the PEI/G-silica sheets show a significantly higher CO 2 adsorption than PEI/silica sheets. Sui et al. synthesise 3D cross-linked GO hydrogels with a polyethylenimine (PEI) [242]. As shown in Fig. 20, the structures are developed through gel formation from aqueous GO dispersion with PEI solution at 25°C followed by freeze-drying. The monoliths possess a low density in the range of (0.02-0.03) g/cm 3 due to the open-pore structures interpenetrating the skeleton of GO sheets. This is also well observed in the XRD patterns. Compared with GO, an increase in the interlayer spacing up to 1.56 nm, indicating successful cross-linking of the GO sheets. Therefore, the open-structures show enhanced SSA of up to 476 m 2 /g and CO 2 adsorption of $2.4 mmol/g at 273 K and 1 bar compared to 1.8 mmol/g in a controlled GO with a high SSA (876 m 2 /g). Clearly, the functional basic sites enhance the CO 2 adsorption capacity.

Spillover and dissociative atomic hydrogen sorption
As explained in Section 3.1.5, hydrogen spillover is a dissociation of H 2 on the catalytic metal nanoparticles, followed by subsequent atomic adsorption onto the substrate surface sites. It is considered as one of the effective ways to enhance H 2 storage capacity in the carbonaceous structures [80,243]. In the study by Hu et al., the H 2 uptake by low temperature physisorption and room temperature chemisorption is clearly demonstrated in a high Pd (about 24 wt%) dispersed graphene nanocomposite [244]. The chemisorptive H 2 uptake in Pd nanoparticles shows a two-phase behaviour of isotherms. Initially formed solid solution region at low pressures (a-phase), is transformed to a hydride phase (b-phase) at higher pressure through a coexistence of two-phase region (a + b), plateau range, very similar to the typical pressure-concentration-temperature curves of H 2 storage materials [245,246].
The isotherms also show a Pd particle size, (4-24) nm, dependence; a clear transition from ato bphase is observed with increase in a Pd particle size. A high H 2 uptake of 3.2 wt% at 77 K and 1.1 bar for a given SSA of 230 m 2 /g and a pore volume of 0.3 cm 3 /g suggest a clear spillover effect. Furthermore, an exfoliated graphene (with a SSA of 755 m 2 /g) decorated by Pd and Pt nanoparticles, (2)(3)(4)(5) nm, via an electroless deposition technique also shows enhanced room temperature spillover of H 2 uptake [247]. The observed uptake of 4.3 H/Pd is much larger than that of actual chemisorbed hydrogen in the bulk Pd with a stoichiometry of 0.6 H/Pd. The room temperature H 2 spillover capacity in a mixture of the activated carbon (AC, Maxsorb) as ''secondary spillover receptor'' and Pd-exfoliated graphene (in a 9:1 weight ratio) shows reversible H 2 uptake of 0.82 wt%, 49% more (0.55 wt% of H 2 ) than Pd-free AC-exfoliated graphene under 80 bar [248]. The spillover is also revealed by a high isosteric heat of H 2 adsorption of $14 kJ/mol, compared to $8 kJ/mol for AC.
Ramaprabhu's group report up to 3 wt% and 4 wt% of H 2 in a Pd (20 wt%) nanoparticle decorated graphenes before and after N-doping (ca. 7%), respectively at 298 K and 40 bar (Fig. 21) [249][250][251][252]. The values are comparatively higher than $0.7 wt% and $1.5 wt% in a Pd free graphenes before and after N-doping, respectively. A clear influence of N-doping can be observed. A combined physisorption and chemisorption process gives a high heat of H 2 adsorption over 12.5 kJ/mol. The hydrogen exfoliated GO at 523 K is used to synthesise N-doping and Pd-decoration through nitrogen plasma treatment and reduction of PdCl 2 , respectively. The Pd/N-G sample accounts 48% of the spillover capacity enhancement in the total H 2 storage capacity. Wang et al. develop a nonprecious and air stable Ni-B nanoalloy dispersed graphenes to enhance the H 2 adsorption [253,254]. As much as 4.4 wt% of H 2 is achieved at 77 K and 1 bar in the graphene with Ni (0.83 wt%) and B (1.09 wt%) (graphene-Ni 0.83 B 1.09 ) compared to 1.35 wt% of H 2 in the undoped graphene. The composites are obtained via NaBH 4 chemical reduction of aqueous solutions of GO and Ni(CH 3 OOH) 2 . The H 2 capacity seems excellent among above Pd-graphene and all other carbon-based materials. Up to 12 kJ/mol of isosteric heat of H 2 adsorption is reported.
Further higher H 2 storage capacity of over 5 wt% at 298 K and 20 bar and heat of adsorption, >30 kJ/ mol at low coverage and $14 kJ/mol at high surface coverage is reported in the Pd/Hg nanoalloy and Pd nanoparticle loaded GO-like foams (GLFs) [255][256][257]. Porous GLFs (nominal composition C 2 OH) are synthesised by simple calcination of the molecular precursor, sodium chloranilate dihydrate in air at 573 K [255]. GLFs consist of interconnected bundles of turbostratically stacked few-layer graphenes build-up of aromatic and aliphatic domains with oxygen rich functional groups on the surfaces. GLFs with SSA of 510 m 2 /g and pore volume of 0.89 cm 3 /g selectively absorb CO 2 (about 2 mmol/g) over N 2 (0.4 mmol/g), CH 4 (0.7 mmol/g), H 2 (negligible) and CO (0.7 mmol/g) at 10 bar and room temperature. The Pd 4 Hg particle loaded (12 wt%) GLFs (Fig. 22) with its inherent carbon-based radicals near the edge and defect sites yield an enormously enhanced H 2 uptake of about 5 wt% at room temperature and 20 bar compared to a negligible H 2 uptake in undoped GLFs [256]. The Pd-Hg alloy loading is obtained through NaBH 4 reduction of aqueous GLFs, PdCl 2 and HgCl 2 mixture. In other study, Wang et al. show an important comparison of H 2 uptake by spillover in Pd loaded (10 wt%) GLFs (Pd/GLFs) and super-activated carbon (Pd/AX-21), and oxygen modified AX-21 (Pd/AX-21-O) [257]. By comparing H 2 absorption, the authors conclude that the existence of abundant oxygen functional groups on the surface and edges in Pd/GLFs lead to enhanced H 2 uptake with a high heat of adsorption, >30 kJ/ mol at low coverage and $14 kJ/mol at high surface coverage compared to those of Pd/AX-21 and Pd/ AX-21-O. The H 2 dissociative adsorption is also observed in simple modified graphenic structures without metal dispersions. The volumetric H 2 storage of 5 wt% is also reported in exfoliated turbostratic carbon nanofibers (CNFs) at 77 K and 100 bar compared to 1.5 wt% in the as-produced CNFs [258]. This enhancement is attributed to the simultaneous molecular and dissociative H 2 adsorption in combination with more defective surface, increased SSA (from 103 m 2 /g to 227 m 2 /g) and pore volume (0.39-0.52 cm 3 /g). A graphite exfoliation technique using intercalation of acids and thermal shock is employed to expand the CNFs.

Bottom-up solution-phase synthesised graphenes
Using the bottom-up solution-phase method, graphenes of gram scales are produced from simple laboratory reagents of sodium and ethanol in a 1:1 M ratio by solvothermal method [259,260]. The subsequent rapid pyrolysation leads to a porous graphene in relatively large quantities ($0.1 g for 1 ml ethanol). The sample with a high SSA of 2139 m 2 /g shows a high-pressure H 2 uptake of 0.9 wt% at room temperature with a high heat of adsorption of 12 kJ/mol [261]. The adsorption is attributed to dissociation of H 2 at the zigzag edges of the graphene, or combined effects of residual oxygen and sodium functionalities. In other study, the heteroatom (boron, phosphorous, or nitrogen) substituted carbon scaffolds (micrometer sized flake-like few-layered graphene structure assembled by (3-5) nm sized domains) show H 2 capacities of (2.2-2.4) wt% per SSA of 1000 m 2 /g at 77 K and 2 bar relative to 1.8 wt% for the pristine carbon scaffold or $2 wt% for common carbonaceous materials [262]. The heteroatom substituted scaffolds also show H 2 binding energies between (5-9) kJ/mol.

Highly porous graphene carbons from templating, CVD and KOH chemical activation
As demonstrated above, all the methods; functionalization, exfoliation and pillaring hybrid designs generated a promising gas sorption and storage. However the structures have limited accessible surface areas and pore volumes. To obtain a relatively higher SSA with tuneable pore structure, self-or sacrificial-templating from the ordered structures, CVD and chemical activation methods are employed [263][264][265][266][267]. As shown in Fig. 23, the chemical activation of precursor GO with activator KOH yields a very high SSA of up to 3100 m 2 /g graphene-carbons. The exfoliated GO by thermal shock or microwave irradiation is thoroughly mixed with various amounts of KOH either by solution or dry milling followed by activation at temperatures between 873 K and 1173 K. For example, Srinivas et al. report a range of SSA (up to 1900 m 2 /g), pore volume (up to 1.65 cm 3 /g), pore size and size distribution in GODCs (GO derived carbons) by changing the GO/KOH concentration and activation temperature [263]. The large fraction of smaller pores is transformed to large pores when increasing KOH concentration and/or activation temperature. As shown in Fig. 24, the GODCs show favourable gas adsorption for CO 2 and CH 4 , compared with other high surface area carbons for a given SSA. Furthermore, the N-and S-doping in GODCs with pyrrole and poly-thiophene, respectively show an enhancement in ambient CO 2 uptake ($4 mmol/g) and the selectivity over CH 4 and N 2 [264][265][266], The selectivity is an important factor for the application in flue gas capture. For example, as shown in Fig. 25, the S-doped graphene carbons show highly enhanced and stable recycling CO 2 adsorption capacity of >4.0 mmol/g over CH 4 , N 2 and H 2 (which show much lower adsorption of <1 mmol/g) at 298 K and 1 bar. The capacity is twice as high as for the undoped activated graphene carbons ($2 mmol/g) [263]. The CO 2 capture capacity in the doped porous graphene seems comparable with or even higher than the various other porous media; N-doped porous carbons, amine functionalized silica and MOFs. The high microporosity and N-functionality show a high isosteric heat of CO 2 adsorption between (30 and 56) kJ/mol. The CO 2 adsorption of 1.82 mmol/g at 0.2 bar (is a partial pressure of CO 2 in the flue gas stream) in the S-doped sample is also comparatively larger than other porous carbons. The increase in activation temperature yields enhanced microporous samples but same time the N-and S-doping content reduces; reduction from (7 to 2.7) at% is observed at 700°C, whereas the lower activation temperatures lead to a decrease in SSA. Using porous, layer structure MgO as a template, nanomesh of one-to two-layer graphene in a gram-scale is produced by CVD method using methane as a carbon precursor [267]. After acid treatment to remove MgO template, the graphene sheets show a large corrugations with nanopores of $1 nm, total pore volume of 2.35 cm 3 /g and SSA of 2038 m 2 /g. A high H 2 , CH 4 and CO 2 adsorption of $1.5 wt%, 14.5 mmol/g (at 90 bar) and 36.5 mmol/g (at 31 bar), respectively at 274 K is reported.

Electrochemical hydrogen storage
The graphene materials with their high electrical conductivity and chemical stability also investigated for electrochemical H 2 storage via electro-decomposition of water on a cathodically polarised electrode. The adsorption-desorption mechanism during charge/discharge phenomenon is simply written in terms of the following equations: represents the graphene host. In steps; water reduces first (H 2 O + e À ? H + + OH À ) and the nascent hydrogen then adsorbs into the graphene network due to the electric polarisation of the electrode under the external applied potential. In the reversible process, the hydrogen atoms discharge from the graphene and recombine with OH À to make H 2 O. A reversible H 2 storage of $148 mA h/g is reported in a few layer graphene flakes that are synthesised through arc-discharging of pure graphite under hydrogen atmosphere [268]. The graphene working electrodes are prepared on Ni-foam using PVDF binder and press. The reversible H 2 storage seems highly dependent on the structural properties of graphene flakes; crystallite size, wrinkles, defects, and inter layer spacing. Graphene-porous cobalt-oxide (Co 3 O 4 ) nanocomposites show a first discharge capacity of 241.9 mA h/g which is equal to 0.89 wt% of H 2 [269]. The nanocomposite on a sheet of nickel foam, Ni(OH) 2 /NiOOH and Hg/HgO as a working, counter and reference electrode, respectively is employed in a 6 M KOH electrolyte at 298 K. As shown in Fig. 26 a further high discharge capacity of 1415 mA h/g, equivalent to 5.176 wt% of H 2 is observed in the fullerene-like orthorhombic-structured Co 3 C nanoparticles, synthesised by ball-milling of Co and graphene powders [270]. Two obvious plateaus at 1.2 V and 0.4 V, respectively in the discharge curve of Co 3 C nanoparticles reveal an existence of two different H 2 adsorption sites; first adsorbs into the interstitial sites/pores between Co 3 C nanoparticles and then diffuses into the interstitial sites between Co 3 C. The composition and ball-milling conditions show a profound effect on the cyclic stability of the composites. The continuous decay in a cyclic discharge capacity is attributed to the formation of b-Co(OH) 2 . Overall, the Co 3 C nanoparticles display a relatively high discharge capacity when compared with the traditional AB 5 type H 2 storage materials.

Chemical hydrogen storage
The hydrogenated graphene, also called as graphane, a covalently bonded hydrogen atoms to carbon atoms, (sp 3 C-H bonds on basal plane) is essentially different from the physical H 2 absorbed graphene structures. Number of synthesis methods; plasma, CVD, liquid and gas phase reactions are developed to obtain graphanes. Subrahmanyam et al. demonstrate that it can be used as a chemical H 2 storage material [271]. Birch reduction of a few-layer graphene with excess Li in liquid ammonia produce the hydrogenated graphene that contains up to 5 wt% of H 2 (Fig. 27). Spectroscopic studies reveal the presence of sp 3 C-H bonds and the hydrogen can be released up on heating or irradiation with UV or laser. Indeed, the hydrogenation of various carbon structures, fullerenes, nanotubes, and graphites via a dissolved metal reduction method with Li and methanol in liquid ammonia has been demonstrated early in 2001 [272]. These structures release about (4-5.4) wt% of H 2 upon heating to 873 K. A simultaneous release of H 2 and a small amount of methane is observed during the decompo- Capacities and cycle stability of the as-obtained G and Co 3 C electrodes (the composition Co x C, represents the weight ratios of Co:G = x:1; x = 4-7) at a 30 mA/g discharge current density [270].
sition. Recently, a hydrogenated graphene with $6 wt% of H 2 (C 1.3 H) n , is synthesised directly from a Birch reduction of graphite powder [273].
Simplified techniques have also been developed to obtain chemical hydrogenated graphenes in a large scale compared to complexity involved in a hydrogenation of CVD/mechanically cleaved graphene in hydrogen plasma [274] or Birch reduction of graphene/graphite. For example, a gram scale of hydrogenated graphene directly from the GO is synthesised via in-situ generated atomic hydrogen through the H 2 spillover using nickel as an active catalyst under ambient conditions [275]. The hydrogenated graphenes also produced directly by exfoliating GO under high pressures,  bar of hydrogen at high temperatures, (473-773) K [276]. The hydrogen, fluorine, and oxygen gas phase atomic covalent functionalization to graphene is also explained [277].
Earlier, Orimo and his co-works report a hydrogen concentration up to 7.4 wt% (CH 0.95 ) in the ballmilled graphitic nanostructures (Fig. 28) [278][279][280]. Hydrogenation is attributed to the formation of dangling carbon bonds, the CH x covalent bonds. Interestingly the milled graphite also shows $400 m 2 /g of SSA and H 2 physisorption at cryogenic temperature because of random orientation of fragmented small coherent domains that consist of small stacks of 2-3 graphene layers. Very recently, a large scale (5 g for each batch), edge-hydrogenated graphene with 3 wt% of H 2 (which corresponds to C/H = 2.3) is produced by ball-milling of graphite powder in a hydrogen atmosphere [281]. Electron microscopy studies reveal the mechanochemical cracking of graphite into small grain sizes of (0.1-1) lm after milling of 48 h correspondingly, the SSA and pore volume increases respectively, (3-437) m 2 /g and (0.002-0.391) cm 3 /g. The edge-selective hydrogenation during the ball-milling process is expected due to the reaction between reactive carbon species (radicals and ions) at the broken edges of graphenes.

Graphene membranes and protection barriers
A monolayer graphene membrane is impermeable to standard gases including helium [282]. The nanopore graphene membrane as molecular sieves is demonstrated by measuring the transport of a range of gases (H 2 , CO 2 , Ar, N 2 , CH 4 and SF 6 ) through the irradiated pores [283]. As shown in Fig. 29, a membrane fixed over predefined 5 mm diameter wells on SiO 2 and pressurised with H 2 Fig. 27. Change in the weight percentage of hydrogen of EGH and HGH with temperature. (Inset) The evolution of H 2 is recorded by a gas chromatograph [271]. gas is etched with ultraviolet light leads to a rapid leakage of H 2 but preventing N 2 from passing through it. A size selective permeation of gas molecules is demonstrated by creating different-size pores. The submicrometer thick, (0.1-10) lm and area $1 cm 2 of GO membranes are found to be completely impermeable to liquids, vapours and gases, including helium, while allowing unimpeded permeation of water (H 2 O permeated through the membranes at least 10 10 times faster than He) [284]. It is attributed to a low-friction flow of a monolayer of water through 2D capillaries formed by closely spaced graphene sheets. Diffusion of other molecules is blocked due to reversible narrowing of the capillaries in a low humidity and/or by their clogging with water. Mass-spectrometry measurements show that helium is less permeable in dry GO film than a 1-mm-thick glass. No permeation is detected for several polar and nonpolar organic liquids through GO membranes of 1 lm in thickness. Reduced GO membranes with an interlayer spacing of 0.4 nm are much less permeable to water. In recent studies, the graphene membranes are also tested for different solvent filtration (desalination) and separation [285,286].
The single layer graphene grown by a CVD method on Ru, Cu and Cu/Ni is found to protect the underneath metal from air oxidation even after heating them in air [287][288][289][290][291]. The surface oxidation of bare copper and oxidation resistance with surface grown graphene is clearly demonstrated in Fig. 30 by controlled CVD growth of graphene directly on Cu foils with partial to full surface coverage [289]. The colour change of bare copper foil indicates a surface oxidation, whereas the fully covered graphene surface shows no sign of oxidation. In other case, the oxidation resistance of Fe and Cu foils are also demonstrated by coating them with reduced GO layers [290]. Nilsson et al. investigate the limitations of graphene as an effective corrosion-inhibiting coating on Pt(1 0 0) surface for O 2 and CO using scanning tunnelling microscopy measurements and DFT calculations [291]. Graphene layer is found to protect Pt(1 0 0) surface against O 2 and CO exposure up to a certain pressures of 10 À4 mbar  [280]. Right: Thermal desorption spectroscopic plots of H 2 for the nanostructured graphite prepared by milling for 80 h under various hydrogen pressures. Here, chemisorption hydrogen concentration (C CH ) determined by oxygen combustion method and physisorption-like hydrogen concentration (C PH ) at 60 bar (6 MPa) and 298 K determined by a volumetric method are listed as well [278]. and 10 À6 mbar, respectively. At higher pressures, CO is observed to intercalate under the graphene coating layer.
Designing the polymer thin films or membranes to prevent gas or water molecules from permeating through is a major challenge in many applications for food and electronics. The graphene-polymer composites have been investigated as improved gas barriers due to the large aspect ratio of graphenes of H 2 (represented as red circles). Equilibrium was reached (p int = p ext ) by diffusion through the SiO 2 . (b) After p int > p ext the membrane bulged upwards. Etching pore(s) in the graphene membrane bigger than H 2 but smaller than the air molecules (mostly N 2 and O 2 , denoted as green circles) (c) leaks out the H 2 from microchamber, and (d) all the H 2 molecules have leaked out of the microchamber, the membrane deflected downwards. (e) Deflection versus position, measured from 0 min (black) to 8 min (dashed blue) after etching. (f) Maximum deflection d versus t for one membrane separating H 2 from air, measured by AFM. Black symbols represent the H 2 leak rate before etching and red symbols the H 2 leak rate after introducing selective pores into the graphene. Inset: optical image of the bilayer graphene flake used in this study, which covers many cavities in the SiO 2 substrate (scale bar is 60 mm) [283]. [292][293][294][295][296]. For example, a poly(ethylene-2,6-naphthalate)-graphene nanocomposite film with the addition of just 4 wt% of graphene shows 60% decrease of the H 2 permeability [292]. Up to 90% decrease in N 2 permeation is observed in thermoplastic polyurethane films reinforced with isocyanate treated GO of 3 wt% [293]. Whereas, the solution filtration synthesised PEI (polyethylenimine) functionalized reduced GO thin films with a brick and mortar structure show a significant decrease in the H 2 permeation rate with increasing PEI content due to tight packing and filling up of gallery spacings of the film assembly [294]. A considerably reduced O 2 and CO 2 permeation is observed in a layerby-layer assembly of GO and PEI composite (Fig. 31) [295]. A 10 bilayer film ($91 nm thick) with 0.1 wt% PEI and 0.2 wt% GO mixtures deposited on a 179 lm PET shows O 2 transmission rate of 0.12 cm 3 /m 2 /day. A high H 2 /CO 2 gas separation with a selectivity higher than 383 is observed, could be due to strong binding of CO 2 with the amine groups. A solvent blending prepared graphenepoly(ethylene vinyl alcohol) (RGO/EVOH) thin film (a thickness of about 0.2 mm) with only 0.5 wt% of RGO loading shows a significantly decreased permeable coefficient of O 2 nearly 1671 times lower than that of neat EVOH film [296]. The permeability measurements are carried out by a differential volume-variable pressure method at room temperature with 37% relative humidity and O 2 flow at 1 bar on 80 mm diameter film. The interesting gas barrier properties are attributed to a tightly packed nanobrick wall structure that creates super tortuosity and diffusion length for gas molecules.
Ultrathin (up to 20 nm) graphene and GO based membranes also show excellent gas permeation selectivities [297,298]. CVD grown large-area monolayer graphene films with increasing numbers of graphene sheets deposited on a Poly(1-methylsilyl-1-propyne) (PTMSP) substrate show improved O 2 /N 2 selectivity suggesting that the gas diffusion through irregularly aligned defective graphene pores as well as slit-like interlayer spacings [297]. The GO thin-film membranes deposited on polye- thersulphone supports in humidified state show a high selectivity of CO 2 /N 2 , which is ideal for postcombustion CO 2 capture. The gas permeance (in gas permeation units, GPUs) measured at a feed pressure of 1 bar, using the constant-pressure and variable-volume method shows very different gas transport behaviour (Fig. 32). Introduction of humidity in the feed stream enhances the CO 2 permeance leading to enhanced selectivities of CO 2 /CH 4 , CO 2 /H 2 and CO 2 /N 2 . The highly interlocked layer structure of GO membranes shows a gas permeation in the order of CO 2 > H 2 P He > CH 4 > O 2 > N 2 . As shown in Fig. 32d the separation performance of GO membrane in its dry and humidified state is higher than that reported for CO 2 for polymeric membranes, including thermally rearranged (TR) polymers, polymers of intrinsic microporosity (PIM), and inorganic membranes, such as carbons, silicas, and zeolites. A more permeable, H 2 /CO 2 selective is demonstrated by creating thermally generated irreversible and small defective pores on the basal plane due to partly released surface oxygen-containing functional groups by thermal reduction of GO membranes. Overall the enhanced CO 2 selectivities are attributed to the CO 2 -philic permeation behaviour, which is further enhanced by the presence of water. It is predicted that the incorporation of a carboxylic acid group lead to the highest isosteric heat. Although CO 2 is a nonpolar gas, the polarity of the individual C-O bonds in the molecule allows for interaction with polar groups in GO. Thus, CO 2 can act as a Lewis acid or a Lewis base and can participate in hydrogen bonding. Carboxylic acid groups on GO provide a preferential site for CO 2 adsorption, consequently retarding CO 2 transport in the nanopores by strongly trapping CO 2 molecules. This phenomenon is counterintuitive, because strong affinity between penetrant and nanopore walls often leads to surface diffusion of the penetrant preferentially sorbed on the pore wall, resulting in flux enhancement of highly sorbed or condensable penetrants. As such, these membranes are promising materials for industrial CO 2 separation processes related to petrochemical engineering (CO 2 removal from natural gas), the environment (CO 2 capture from flue gas), and biomass energy (CO 2 recovery from landfill gas).
Ultrathin ($1.8-18 nm thickness) GO membranes prepared by vacuum filtration on anodic aluminium oxide (AAO) support from controlled dilution of GO dispersions show unusual H 2 /CO 2 and H 2 /N 2 separation selectivities as high as 3400 and 900 respectively, at 293 K [298]. Normally the micropo- rous membranes show low H 2 /CO 2 selectivity (<10) or are selective to CO 2 over H 2 at temperatures below 373 K due to a strong CO 2 adsorption and blocking of H 2 permeation. GO membranes with increasing thickness from 1.8 nm to 180 nm show exponential decrease in H 2 and He permeances. Here the gas permeance is mainly attributed to the selective structural defects within GO flakes. The GO interlayer spacing has minimal effect on molecular transport. The H 2 /CO 2 separation selectivity decreased with increasing temperature, resulting from the faster increase of CO 2 permeance. A $18 nm thick GO membrane on cellulous acetate support show H 2 /CO 2 and H 2 /N 2 separation selectivities of 1110 and 300, respectively. As shown in Fig. 33, the performance of the ultrathin GO membranes is far above the upper bound for polymeric (black line) and inorganic membranes.

NH 3 , NO 2 , H 2 S and SO 2 sorption
The emissions of NH 3 , NO 2 , SO 2 , and H 2 S from the industrial plants during number of combustion processes, wastewater treatment, and food and compositing process are toxic, corrosive and malodor- ous air pollutants. Due to growing demand for a more economical and improved process, there is always a need of search for new filtrate/adsorptive removal materials. Recently, the porous inorganic-graphene based materials for adsorptive removal of NH 3 , NO 2 , SO 2 and H 2 S pollutants have been in focus of attention. Many theoretical and experimental studies are conducted for molecular adsorptive detection/capturing on graphene surfaces and GO based functionalised/inorganic hybrid porous structures. The DFT calculations suggest that adsorption of nitrogen oxides, NO x : NO, NO 2 , NO 3 on GO is generally stronger than that on pristine graphene due to the presence of active defect sites, such as the hydroxyl and carbonyl functional groups [299]. The interaction of NO x with GO is expected to result in the formation of hydrogen bonds OHÁ Á ÁO (N), weak covalent bonds CÁ Á ÁN and CÁ Á ÁO, as well as the nitrous and nitric acid like moieties. First-principles calculations reveal that the oxygen groups on GO act as strong binding sites and induce dissociation of the NH 3 into the NH 2 or NH species by the H atom abstractions [300]. This further leads to the removal of surface oxygen species through the hydroxyl group hydrogenation and the ring opening of epoxy group. The reaction of NH 3 with the hydroxyl and epoxy groups is generally exothermic. From the first-principles calculations adsorption of H 2 S onto pristine graphene is found to be very weak due to its small binding energy, large bonding distance and small net charge transfer [301]. Whereas, the Pt-decorated graphene bind up to seven H 2 S molecules to one side of a Pt-graphene system due to large binding energy and a short Pt H 2 S bonding length. The adsorption of several acidic gases (CO 2 , NO 2 and SO 2 ) on light metal (Li, Al) decorated graphene oxide (GO) is also investigated with the first-principles calculations [302]. It is found that Li and Al could be anchored stably by hydroxyl and epoxy groups on GO and acts as strong adsorption sites for CO 2 , NO 2 and SO 2 with improved binding energies up to (0.59, 2.29 and 1.08) eV, respectively, compared with Ti.
Experimentally, it is also identified that the acidic nature of GO with its carboxyl, hydroxyl, and epoxy groups act as energetic sites for chemical adsorption of gases, especially for polar molecules. A highly polar NH 3 interaction with the GO through gravimetric uptake measurements is demonstrated back in 1962 [303]. About 180 mg/g of NH 3 adsorption at room temperature and atmospheric pressure is observed. A physical adsorption at defective sites/or by hydrogen bonding with hydroxyl groups and a chemical interaction with the acidic sites and condensation in pores is proposed at lowand high-pressures, respectively. Therefore an increase in graphene inter layer distance proportional to the size of the ammonia molecule is observed.
Recently, there are renewed interests and increased investigations on NH 3 adsorption in various types of GO based materials, due to the development of promising and efficient new synthesis methods for obtaining GO and its composites. Bandosz's group have been actively exploring the NH 3 , NO 2 , H 2 S and SO 2 adsorption and interaction properties in various types of GO and its composites with polymers, metal oxides, and MOFs . The NO 2 , NH 3 and H 2 S molecules have a small kinetic diameter between (0.30-0.36) nm but exhibit different chemical properties (e.g., acid-base and redox properties). The amount of gas adsorption is demonstrated through the breakthrough measurements by passing a gas stream through a sample packed column with the option of either dry or up to 70% relative humidity at room temperature. The simple schematic of experimental setup is shown in Fig. 34 [304]. The effects of the degree of oxidation, reduction and exfoliation of the GO samples are examined for toxic gaseous adsorption. For example, a high NH 3 breakthrough capacity of 61 mg/g is observed in GO, dried at room temperature compared to 28 mg/g in GO dried at 393 K and 17 mg/g in GO exfoliated at 573 K [305][306][307]. The acid-base reactions and intercalations between the GO layers are identified as the main mechanisms of NH 3 retention. The NH 3 adsorption is very sensitive to the additional functional groups present in GO when different synthesis methods are employed. For example, GO prepared by the Brodie method, NH 3 is mainly retained via intercalation in the interlayer space of GO and reaction with the carboxylic groups at the edges of the graphene layers. On the contrary, the NH 3 reaction with the epoxy, carboxylic and sulphonic groups is observed in the Hummers GO [305]. In addition, the role played by surface texture, degree of oxidation and interlayer space/or pore structure on the NH 3 adsorptive retention is explained [306]. NH 3 dissolving in lamellar water as ammonium ions or by forming hydrogen bonding with epoxy and phenolic surface groups is also proposed [307,308]. The interaction of NH 3 with carboxylic groups is proposed as either Brønsted: or Lewis acids: The second reaction seams more likely to happen in the dry conditions.  12 ] 7+ ) in GO layers show a highly increased interlayer spacing (up to 1.5 nm) and NH 3 sorption due to additional Lewis and Brønsted acidic centres from the Keggin cations [310]. The GO-POM composites also show enhanced NH 3 sorption and retention than GO alone due to the interaction with the both bridging and terminal oxygens of POM (hydrogen bonded to the terminal oxygens of the POM could enhance the acidity) [311]. POM is a kind of Keggin polyanion (consists of a central tetrahedron PO 4 surrounded by twelve octahedral MO 6 (M = W or Mo)) and exhibits a high acidity. The enhancement of NH 3 adsorption in clays-graphene composite is attributed to the synergistic effect of structure as well as graphene edge acidic groups and clay origin Brønsted or Lewis acidic centres [312]. The GO with simple metal-oxide (MnO 2 ) nanocomposites are also tested for NH 3 sorption [313].
Interestingly, GO in its dry condition show negligible adsorption for other pollutants; NO 2 , H 2 S and SO 2 . However, an enhanced H 2 S sorption (breakthrough capacity of 30 mg/g compared to zero in dry GO) is observed in a functionalised GO that is treated with NH 3 at 1223 K [314]. The NH 3 treatment results in decomposition and rearrangement of oxygen functional groups and incorporation of nitrogen mainly in pyridinic and quaternary entities. These nitrogen functionalities enhance the basicity and attract HS À ions of dissociated H 2 S via electrostatic interactions. The oxidation to elemental S and a smaller amount of SO 2 is also detected. The GO-metal oxide/hydroxide composites show enhanced sorption of NO 2 , SO 2 and H 2 S by forming nitrates, sulphide, sulphite and sulphates through the reactive paths (1) and (2) shown below [315][316][317][318][319][320]:  [321,328]. Instant reduction of NO 2 to NO is noticed. A favourable adsorption and conversion of NO 2 , and NO retention is also observed on GO-polyaniline (PANI) composites [322]. GO in its dry state show no NO 2 adsorption, however the sorption and retention of NO 2 is observed with the presence of N-methylformamide between GO layers due to polar and acid-base interactions.
The MOF-GO composites show an improved NH 3 , NO 2 , and H 2 S sorption compared to the constituents because of synergistic effect related to the new pore space in the interface between MOF units and GO and a reactive interaction of metal sites and ligands either by hydrogen bonding (MOF-5) or coordination and subsequent complexation (HKUST-1) [323][324][325][326][327]329]. For example, the Lewis interactions of NH 3 and the Cu sites, (70)(71)(72)(73)(74)(75)(76)(77)(78)(79)(80) kJ/mol and the reaction of NH 3 with the ligands,  kJ/mol exhibit stronger binding energy. The NH 3 adsorption in Fe-MIL shows retention of the framework structure and thus could be useful for adsorbent regeneration. As shown in Figs. 35 and 36 initially dark blue HKUST-1 and its GO composites turned to light blue upon adsorption of NH 3 , NO 2 and H 2 S [323,329]. With the progress of the adsorption test, the colour of the sample further changed to a light blue or blackish tint during exposure to NH 3 and H 2 S, respectively. The first colour change is attributed to the coordination to the Cu sites due to the presence of a lone pair of electrons on the adsorbate molecules. The second colour change is likely caused by the formation of new complexes like CuS. The reactive decomposition in HKSUT-1 is attributed to the formation of the Cu(NH 3 ) 4 2+ , CuS and Cu(NO 3 ) 2 for adsorption of NH 3 , H 2 S and NO 2 , respectively. In Zn-MOF-GO composites, NH 3 interact through hydrogen bonding with the oxygen atoms of the ZnO 4 units whereas for Fe-MIL-GO composites, the acid-base reaction thought to be the reactive pathway between the NH 3 and water coordinated Fe sites. Long et al. show an oxidative absorption of SO 2 in the porous GO foams, prepared by freeze-drying technique [330]. As shown in Fig. 37 the GO foams exhibit a high porosity and resemble a sponge-like appearance. The oxidative adsorption of SO 2 to SO 3 by reduction of GO at room temperature is observed in a dry condition. Whereas, the reaction is intense with the GO water suspension; the GO is reduced quite fast and SO 2 becomes SO 4 2À .

Summary and perspectives
As discussed above and shown in Figs. 1 and 2, there exist a wide range of graphene/GO based materials; starting with expanded graphite, intercalation, exfoliation, chemical reduction, pillared layers, self-assembly, thin-film membranes, functionalization, doping, metal dispersion, porous template and chemical activation. Those are investigated for gas sorption, storage and separation. In particular, the experimental H 2 and CO 2 uptake results are comparatively analysed in various graphene based materials, as indicated in Tables 2-5, along with the SSA and the measurement conditions (particu- Fig. 35. Schematic of the adsorption process of NH 3 , H 2 S and NO 2 on the HKUST-1 with evidence of the colour changes and the identification of the reactions products [323]. larly, pressure and temperature). As specific surface area and pore volume are the key porous parameters to determine the physisorption capacity Fig. 38 shows the H 2 uptake in a wide variety of graphenes against the SSA and pore volume. The structure modification effect on the binding, heat of H 2 Fig. 37. Top: SEM image of GO foam, and GO solution before (brown) and after (dark) coming in contact with SO 2 gas, clearly showed a reduction of GO. Bottom: Colour change of GO foams exposed to SO 2 gas in dry form. The SO 2 gas is introduced into the column containing the GO foams, the colour of the foams gradually changed from brown to black. This suggests that the GO reacted with the SO 2 gas and the extent of the reaction gradually increased with time [330].  [204] adsorption is shown Fig. 39. Earlier, the H 2 uptake capacity against the SSA of a wide range of other porous carbons is generalised and is normally accepted that for every 500 m 2 /g of SSA, the H 2 uptake capacity at 77 K is increased by $0.5 wt% at 1 bar and $1 wt% at high pressures [82,331,332]. Similarly, at room temperature and pressures up to 100 bar the H 2 storage in a wide variety of porous carbons is found to be well below 1 wt% [333]. From Fig. 38 one can clearly understand that the modified graphenes with carefully tuned pore structure or pillaring or chemical modifications (doping, functionalization and metal dispersion) performed much better than simply exfoliated graphenes. At 77 K, most of the exfoliated/reduced graphenes seems to exhibit more or less similar uptake capacities compared to the other porous carbons with respect to the porosity. Normally the exfoliation leaves very high mesoporosity due to the highly curved nature of graphenes. However, it is understood that the modified graphenes enhance the microporosity and functional active sites. Thus comparatively a high H 2 uptake is seen in optimised GO structures, with doped, defective, and metal/alloy nanoparticle dispersed graphenes due to the synergistic effect of enhanced microporosity, binding sites or combined molecular and spillover/dissociative atomic hydrogen adsorption. Most importantly these graphene based materials exhibit very promising H 2 uptake capacities between (1-4) wt% at 300 K compared to the maximum of up to 1 wt% in most porous solid adsorbents, i.e. MOFs, zeolites and nanostructured carbons. This is further evidenced by a high heat of H 2 adsorption mostly between   (8-15) kJ/mol in the modified graphenes, compared to a typical (4-6) kJ/mol in other porous carbon adsorbents. These facts are clearly encouraging and help to minimise the gap between DOE targets and experimental results. Similarly, the highest CO 2 capacity is reported in functionalised/doped and highly porous activated graphene materials ( Table 5). The graphene based materials are also found to exhibit enhanced CH 4 storage that surpasses the DOE target and other traditional highly porous carbons and most of the MOFs. The graphene-based composites also show enhanced NH 3 , NO 2 , SO 2 and H 2 S sorption and retention. The theoretical and experimental results on the graphene related membranes for gas separations are very encouraging. Because of multifunctionality of the graphene-based materials many approaches have been adapted to develop various robust interconnected porous network architectures with various functional groups. Clearly, by controlling the synthesis parameters and tuning the pore parameters; increasing SSA and microporosity, the chemical modification of graphenes could be a very promising in obtaining practicable gas sorption, storage and separation. There are many possible routes still available to engineer graphene structures. For example, a number of robust 3D hierarchical porous graphene structures have been developed through self-assembly, sol-gel and cross-linking chemistry [43,[334][335][336][337][338]. A vacuum centrifugal evaporation has been introduced as a strategy for a large scale synthesis of GO sponges with 3D interconnected network hierarchical structure [334]. Subsequent thermal annealing of the GO sponges result in recovery of sp 2 graphene structure, forming graphene sponges with a large SSA and typical porous structure. An autoclaved leavening and steaming of GO layer films lead to reduced GO foams with open porous and continuously cross-linked structures [335]. Compared to reduced GO films, the reduced GO foams show greatly improved performance as selective organic absorbents. Furthermore, the doping of the graphene foams is performed to enhance its functionalizability. These graphene foams and its thin film forms could be ideal candidates for an efficient gaseous separation.
Optimised pillared graphene structures with amine-based functional groups could produce superior CO 2 adsorbents (pillared GOs with NH 2 groups like silicas with NH 2 ). Much stable structures may be developed due to strong nucleophilicity of amines and epoxy groups on GO through a ring opening reaction without any activation. Using sol-gel method low-density graphene aerogels with tuneable SSA, (584-1200) m 2 /g and very high pore volume, (2.96-6.4) cm 3 /g are synthesised [336]. In another method by gelation of a GO suspension under basic conditions, a robust 3D macroassembly of graphene sheets is obtained with SSA > 1300 m 2 /g [337]. Furthermore, the chemical activation of graphene aerogels with KOH results in highly increased microporosity (up to 0.69 cm 3 /g) and SSA (up to 1715 m 2 /g) [338]. Similarly, CO 2 activation of the mechanically robust, centimetre sized selfassembled interconnected networks of single-layer graphene aerogel monolith increases the SSA up to 3000 m 2 /g compared to the typically 400 m 2 /g for the non-activated foam [339]. Thus, adjusting synthetic parameters allow a wide range of control over surface area, pore volume and pore size, as well as the nature of the chemical cross-links (sp 2 versus sp 3 ) making these materials promising candidates for gas sorption and storage applications.
A method for preparing functionalized 3D carbon-based architectures, consisting of high SSA mesoporous carbon spheres (MCS) with a particle size of 60 nm intercalated between graphene sheets (GS) is demonstrated by Lei et al. [340]. The MCS itself consists of graphene sheets in small domains and exhibits a SSA and pore volume of 2396 m 2 /g and 2.9 cm 3 /g, respectively, obtained from silica template and ferrocene by CVD method. MCS-GS composite shows hierarchical porous architecture with SSA of 1496 m 2 /g and pore volume of 3.36 cm 3 /g with enhanced microporosity. Further thermal treatment of GMC-GS in NH 3 yields N-doped product. The N-containing basic groups could be very useful in obtaining improved CO 2 adsorption at low-pressures due to enhanced acid-base interactions. In addition, a wide variety of high and tuneable surface area and pore structure assemblies of graphenes with chemical functionalization are demonstrated through template, pyrolysis and chemical activation methods [341,342]. For example, the silica and zeolite based templates are conveniently used to produce a very high SSA and pore volume graphenes and carbons. In particular, by controlling the pore size of silica one could obtain much more enhanced microporosity than the reported methods of producing mesoporous graphene structures. Other possible methods, such as soft co-polymer or sacrificial salt or nanoparticle templating, could provide desirable structures as well. The chemical activation is another promising method in which the tailor made porous graphene structures are highly possible by simply choosing the right activator; KOH, NaOH, Na 2 CO 3 , CO 2 , steam, etc. together with concentration and temperature [342]. The chemical modification, functionalization and hybrid structures of graphene based highly porous materials  could have attractive gas sorption and storage properties. The boron-and nitrogen co-doped and porous graphene like materials should help in obtaining higher gas sorption and storage. Zhang et al. show that a very efficient and industrially scalable approach of synthesising defect/wrinkle mesopore structure graphene sheets with a very high SSA up to 3523 m 2 /g and total pore volume up to 2.4 cm 3 /g [366]. The material obtained through the following two standard industry steps: (1) in-situ hydrothermal polymerisation/carbonisation of the mixture of cheap biomass or industry carbon sources with GO to get the 3D hybrid precursors; and then (2) a KOH chemical activation step. An enhanced surface area and pore volume is also obtained with the addition of GO to mesoporous carbon. Tuneable surface area and porosity parameters are also demonstrated in mesoporous carbons with the controllable addition of GO.
Recently, graphenic type fragments are synthesised from zeolites/MOFs as templates, with either self-or addition of small molecular carbon precursors [369][370][371][372][373][374][375]. Upon carbonisation, the organic linker ligand in MOF transforms into single layers functionalised graphene fragments with new hierarchical high slit-like micropores and mesopores while retaining the MOF framework porosity. These carbons with new pore types show ultrahigh porosities i.e., simultaneous high surface area (up to 3300 m 2 /g) and micro-and total-pore volumes (up to 5.53 cm 3 /g) [369,372]. These functionalised carbons yield exceptionally high H 2 , CH 4 and CO 2 uptakes compared with the capacities of those benchmark microporous materials under identical conditions. In particular, a gravimetric (volumetric) H 2 uptake of 3.25 wt% and 7.3 wt% (50 g/l) are achieved at 77 K under 1 bar and 20 bar, respectively [369,370]. At 100 bar and 298 K, these carbons show $0.94 wt% of H 2 storage. The functionalised MOF carbons also show record high CH 4 and CO 2 uptakes [372][373][374][375]. For example, the room temperature uptake of up to 4.6 mmol/g at 1 bar and over 27 mmol/g (119 wt%) at 30 bar are again some of the highest values reported in the literature for porous carbons. The overall synthesis and gas uptake mechanism of such graphenic type MOF carbons is schematically shown in Fig. 40.
The novel ultrathin, robust nanocomposite membranes by incorporating GO sheets into silk fibroin matrix through heterogeneous surface interactions in organised layer-by-layer manner show outstanding mechanical properties, with a tensile modulus of 1.45 Â 10 6 bar (145 GPa), an ultimate stress of more than 3000 bar (300 MPa), and a toughness of above 2.2 MJ/m 3 [376]. These nanocomposite membranes in their support and free-standing state could be valuable for potential applications in protective molecular coatings, permeable membranes for separation and delivery. A new avenue for graphene-based ultra-films anchored on solid supports with a high stability and controlled thickness via a layer-by-layer assembly is demonstrated using amino-substituted p-conjugated compounds, including 1,4-diaminobenzene, benzidine, etc., as cross-linkages [377]. A thermal annealing leads to a reduction of the films thus could yield suitable pore space for molecular sieving. Furthermore, the (1) MOF framework/crystal structure (up to 400°C) with well-defined pore structure in the micropore region, (2) decomposition, and metal-oxide formation, ZnO and carbon between 600 and 800°C yielding poor pore development, (3) at >900°C, ZnO reduction and evaporation of Zn and CO yields highly porous carbon with a hierarchical pore structure and (4) surface morphology of HPCs by TEM and SEM. Middle: Mechanism involved in the carbonisation process of MOF-5 through simultaneous thermogravimetric analysis (TGA), mass spectroscopy (MS) and powder X-ray diffraction (PXRD). The first massloss between 400°C and 600°C, is attributed to the framework decomposition leading to a major release of carbon containing gaseous products (mostly CO 2 , CO, C 6 H 6 and a small amount of H 2 and C x H y hydrocarbon mixtures), where a formation of crystalline hexagonal ZnO occurs (sharp XRD peaks). The second mass-loss starting at $900°C is due to further release of CO 2 and CO with Zn through the reduction of ZnO by carbon via ZnO + C/Zn(g) + CO. The carbonisation yields materials with either a featureless XRD pattern or very weak and broad XRD peaks between 20°and 25°, and at $45°of two-theta indicating the disorderly oriented tiny graphenic type fragments in the structures. It is worth noting that the reduction of ZnO by carbon during carbonisation yields a pure porous carbon. Bottom: Porosity and CO 2 uptake characteristics of HPCs. The 77 K N 2 adsorption isotherms on the millimetre-sized MOF-5 derived HPCs show simultaneously high surface area, up to 2734 m 2 /g, and exceptionally high total pore volume, up to 5.53 cm 3 /g. In the HPCs, micropores are mostly retained and meso-and macropores are generated from defects in the individual crystals, which is made possible by structural inheritance from the MOF precursor. The resulting HPCs show a significant amount of CO 2 adsorption, over 27 mmol/g (119 wt%) at 30 bar and 27°C. Inset also show a direct relationship between the CO 2 adsorption capacity and the surface area: a CO 2 uptake of 10 mmol/g for every 1000 m 2 /g increase of surface area under 30 bar and 27°C [372]. amine groups could show selective gas permeations. A simple and effective method of preparing functional nanoporous graphene/carbon composite membranes from MOFs is also demonstrated and reviewed [378,379]. The resulting symmetric dense or asymmetric composite membranes exhibit good performance in gas separation and liquid separation via pervaporation.

Conclusion
In summary, the rich chemistry of graphene nanosheets have offered great potential for the design and development of many combinations of interconnected porous graphene-based materials as promising molecular adsorbents, with high thermal/chemical stability, tuneable specific surface area and pore structure. The other advantages are the relatively low cost and easy scale-up associated with the manufacturing of the porous graphene carbons. Graphene and its composites have been used as adsorbents for low temperature physisorption and high temperature chemisorption of H 2 , CO 2 , CH 4 , etc., showing promising results, especially when compared with conventional high surface area activated carbons. As highlighted in this report, the interest in gas sorption and storage by physical adsorption has grown considerably over the last few years. While many computational studies prove strong potential of graphene and its derivatives, practical developments have advanced markedly in molecular pillaring, metal dispersion, doping, functionalization and 3D interconnected networks to enhance gas sorption and storage. Clearly, there still exist several challenges in practice, such as the effective control of the desirable binding sites, porosities and functionalities; the balance of functional properties with structural integrity; and further tuning of molecular selectivities in gas mixtures for purification and sensing.
More specifically, there have been many attempts to use carbon adsorbents as reversible H 2 storage materials. Most of the measurements so far have been conducted at 77 K, which limits the practicality. Similarly, carbon capture at ambient conditions is not satisfactory. As summarised in Tables 2-5, and Figs. 38 and 39, the remarkable synergistic properties from combining GO hybrid structures showed promising room temperature H 2 storage and high pressure CH 4 and CO 2 adsorption capacities. For the given rich chemistry and functionalizability of graphene, the number of new compounds are continuously being tailor-made. Thus if carefully tuned the microporosity and the high surface area of the chemically modified graphene materials could be very promising for H 2 and CH 4 storage and CO 2 capture. Similarly, the gas permeation separation on graphene and GO based membranes is highly dependent on GO sheet size and thickness, layer structure and assembly, residual lamellar water content, and deformation of thin-layered structures during dewetting can all effect. There is much scope in tuning and controlling the parameters to develop practically useful membranes.