Highly efficient, rapid and selective CO2 capture by thermally treated graphene nanosheets

Journal of CO2 Utilization 13 (2016) 50–60

Contents lists available at ScienceDirect

Journal of CO2 Utilization journal homepage: www.elsevier.com/locate/jcou

Highly efficient, rapid and selective CO2 capture by thermally treated graphene nanosheets Shamik Chowdhury, Rajasekhar Balasubramanian* Department of Civil & Environmental Engineering, National University of Singapore, 1 Engineering Drive 2, Singapore 117576, Republic of Singapore

A R T I C L E I N F O

A B S T R A C T

Article history: Received 20 September 2015 Received in revised form 27 November 2015 Accepted 3 December 2015 Available online xxx

Graphene, by virtue of its unique molecular structure and many interesting properties, is receiving considerable attention as an adsorbent for separation and purification of gases. In this study, graphene nanosheets were subjected to heat treatment in the temperature range of 200–800
C under flowing N2 to improve their textural characteristics (surface area, pores size, and total pore volume) for efficient removal of CO2 from flue gases. The resulting graphene materials showed highly ordered structure, large surface area (up to 484 m2 g1) and interconnected hierarchical pore networks with narrow pore size distribution in the large-micropore/small-mesopore range. As a consequence, the heat-treated graphene samples demonstrated significantly greater CO2 uptake capabilities (up to 2.89 mmol g1) compared to pristine graphene (0.81 mmol g1) at 0
C and 1 bar. More importantly, the materials displayed rapid adsorption kinetics with ultrahigh selectivity for CO2 over N2, as well as stable and readily reversible adsorption/desorption cycling behavior. The isosteric heat of adsorption had an unusual dependence on surface loading because of the presence of attractive intermolecular forces between the adsorbed quadrupolar CO2 molecules. These findings demonstrate for the first time that thermal treatment at hightemperatures can have a positive influence on the single component CO2 adsorption characteristics of graphene sheets and should be explored further as an effective strategy in the design and development of graphene-based porous solid adsorbents for CO2 abatement. ã 2015 Elsevier Ltd. All rights reserved.

Keywords: Graphene Thermal treatment CO2 Adsorption Selectivity Isosteric heat

1. Introduction The effective capture of CO2 from fossil-fuelled power plants and large industrial sources (such as steel, cement and fertilizer production) is a top global priority to reduce the impacts of global warming on our climate system and the natural environment [1]. Among the various strategies (i.e., pre-combustion, post-combustion and oxy-fuel combustion) and numerous technologies (e.g., absorption, adsorption and membrane separation) that are currently being explored to curb CO2 emissions [2,3], postcombustion capture using solid adsorbents is an attractive option because of its ease in retrofitting existing CO2 sources, less energy requirements, simplicity of operation, applicability over a wide range of temperature and pressure conditions, facile regeneration of the spent adsorbents, and low capital costs [4]. As a result, several different classes of adsorbent materials have been examined for separation of CO2 from flue gases and other industrial exhaust streams. These include zeolites, porous silica, porous polymers, metal oxides, alumina, activated carbons, carbon

* Corresponding author. Fax: +65 67744202. E-mail address: [email protected] (R. Balasubramanian). http://dx.doi.org/10.1016/j.jcou.2015.12.001 2212-9820/ ã 2015 Elsevier Ltd. All rights reserved.

fibers, ion-exchange resins, and metal organic frameworks (MOFs) [4–8]. While some of those adsorbents have low adsorption capacities and/or require long time to reach saturation due to slow adsorption kinetics, or sometimes poor selectivity relative to other gases, most of them exhibit reduced activity in the presence of moisture, low chemical and thermal stability, or weak mechanical properties that lessen their suitability for on-site application [8]. There is therefore a pressing need to develop advanced solid adsorbents that can adequately address the inherent requirements of real-world CO2 capture systems. Recently, graphene-based materials have received a great deal of attention for the development of next-generation energy efficient and high volume CO2 adsorbents [9–16], due to their exceptional surface area (theoretical value of 2630 m2 g1), excellent thermal conductivity (up to 5000 W m1 K1), outstanding mechanical strength (Young's modulus of about 1 TPa), and remarkable chemical stability [17]. However, graphene as a bulk material has the tendency to form irreversible agglomerates, behaving as particulate graphite platelets, due to strong van der Waals force between the large and planar basal planes, causing a significant decrease of the ultrahigh surface area of twodimensional graphene sheets [18]. Since gas adsorption behavior

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

is mainly governed by the surface area and pore characteristics of the adsorbent material [10], the introduction of nanopores into graphene sheets has been identified as one of the most effective methods for improving the adsorption performance of graphene materials [19]. Because of their porous structure, such holey graphene scaffolds have a higher surface area and much more “space” for capturing and storing CO2 compared to pristine graphene [15]. For example, Srinivas et al. [9] fabricated highly porous graphene-based adsorbents with large specific surface area (1900 m2 g1) and high pore volume (1.65 cm3 g1), via KOH activation of exfoliated graphene oxide (GO) precursors, that could adsorb up to 16.38 mmol CO2 per gram at 27
C and 20 bar. Likewise, several processes have been considered to obtain graphene architectures with porous morphologies for enhanced adsorption and recovery of CO2, including cross-linking [10,11], template directed chemical vapor deposition [12], chemical etching [13], physical activation [14,15], hydrothermal reduction [16], and combinations thereof. Nevertheless, most of these methods are expensive, time-consuming, chemically hostile, complex processes with potentially negative environmental impacts. In addition, the precise control over the surface area, pore structure, pore size and pore size distribution still remains a large scientific and technological challenge [18]. Thermal treatment/reduction is an efficient and convenient alternative for preparation of porous graphene nanostructures because of its (i) simplicity; (ii) scalability; and (iii) sustainability (since it does not involve the use of toxic chemicals or hazardous substances) [20,21]. By varying the process conditions, graphene with various pore size and morphology can be easily obtained. Zhang and co-workers have recently shown that thermal treatment at low-temperature can lead to crumpling rather than stacking of graphene layers, resulting in a biomodal macro/ mesoporous structure with large surface area (429 m2 g1) [22]. As a consequence, these structural defects onto graphene provided a high density of ion diffusion channels, which facilitated rapid charge storage and transport for high-performance energy storage applications. Thermal processing therefore seems to be a promising research direction for the design and synthesis of porous graphene materials with high surface area and pore volume for CO2 capture applications. However, till date, the effects of thermal treatment on the CO2 adsorption performance of graphene remain unexplored. In this work, we present the facile synthesis of robust graphenebased porous adsorbents with large accessible surface area and well-defined tunable pore morphology through direct thermal treatment of graphene nanosheets at high-temperatures. The prepared graphene materials show an excellent CO2 removal ability from flue gases, specifically in terms of high capacity, excellent selectivity, rapid kinetics, low initial isosteric heat, easy regeneration, as well as superior long-term stability over multiple adsorption–desorption cycles. Most importantly, these nanoporous carbons are essentially hydrophobic and hence merit further consideration for the development of post-combustion CO2 adsorption modules. To the best of our knowledge, this is the first report describing the mechanism of surface modification of graphene nanosheets via heat treatment and its effects on pore development and CO2 uptake. 2. Experimental 2.1. Materials Graphite powder (<20 mm) was purchased from Sigma–Aldrich and used as received. Sulfuric acid (H2SO4, 98 wt.%, Merck), phosphoric acid (H3PO4, 85 wt.%, J.T. Baker), potassium permanganate (KMnO4, Acros Organics), hydrogen peroxide (H2O2, 30 wt.%,

51

Sigma–Aldrich), and hydrazine hydrate (N2H4, 50–60 wt.%, Sigma– Aldrich) were used as obtained from the supplier without any further purification. Deionized water was used during the experimental process. 2.2. Synthesis of graphene oxide GO was prepared from natural graphite powder using an improved Hummer’s method [23]. In brief, a 9:1 mixture of concentrated H2SO4/H3PO4 (360:40 mL) was added to a mixture of graphite powder (3 g, 1 wt. equiv.) and KMnO4 (18 g, 6 wt. equiv.), producing a slight increase in the temperature to 35–40
C. The reaction mixture was further heated to 50
C and stirred for 12 h. It was then cooled to room temperature and poured onto ice (400 mL) with 3 mL H2O2. The mixture was finally centrifuged and the supernatant was decanted away. The remaining solid material was rinsed repeatedly with deionized water until the pH of the solution was neutral. After filtration and drying in air at room temperature, GO was obtained. 2.3. Synthesis of graphene Graphene was synthesized by hydrazine reduction of GO. In a typical procedure, GO (0.025 g) was suspended in deionized water (40 mL), yielding an inhomogeneous yellow-brown dispersion. This dispersion was then sonicated until it became clear with no visible particulate matter. To this solution, N2H4 (10 mL) was added and the mixture was transferred into a 100 mL Teflon-lined stainless-steel autoclave and kept at 100
C for 24 h followed by cooling to ambient temperature naturally. The resulting black precipitate was isolated by filtration, washed copiously with deionized water and ethanol, and dried at 100
C. 2.4. Thermal treatment of graphene Thermal treatment of the as-synthesized graphene (GPN) sheets was carried out at four different temperatures (200, 400, 600 and 800
C) in a horizontal tube furnace (TMH12, Elite Thermal Systems Ltd., UK). A weighed amount of GPN was first loaded on a ceramic boat, placed at the center of the tube furnace, and then heated under a continuous flow of N2 (500 mL min1) at a rate of 5
C min1, with a holding time of 2 h at the desired temperature. Following the heat treatment, the sample was allowed to cool down to room temperature in the furnace. The materials thus obtained were labeled GPN-X, where X is the temperature of thermal treatment. 2.5. Characterization methods Wide angle X-ray diffraction (XRD) patterns were recorded on a Bruker D8 ADVANCE (Bruker Co., Germany) X-ray diffractometer equipped with Ni-filtered Cu Ka radiation (l = 0.15 nm) operating at 40 kV and 40 mA. The diffraction patterns were collected over a 2u range of 5–70
. The interlayer spacing was then calculated using the Bragg’s equation. X-ray photoelectron spectroscopy (XPS) data was acquired using a VG ESCA 220i-XL imaging system (Thermo VG Scientific Ltd., UK). Monochromatic Al Ka X-ray (hn = 1486 eV) was employed for analysis with photoelectron take-off angle of 90
to the surface plane. The analysis area was approximately 700 mm in diameter while the maximum analysis depth was in the range of 4–8 nm. Field emission scanning electron microscopy (FESEM) was performed on a JEOL JSM-6700F (JEOL Ltd., Japan) field emission microscope operated at an electron accelerating voltage of 15 kV. The samples were mounted on an aluminium stub with carbon adhesive tape and coated with a thin layer of platinum under high vacuum (103–107 Mbar) conditions using a Hitachi E-1030 ion

52

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

sputter (Hitachi Co., Ltd., Japan) before the FESEM analysis. Transmission electron microscopy (TEM) was carried out on a JEOL JEM 2010F (JEOL Ltd., Japan) transmission electron microscope operated at 200 keV. For TEM measurements, the samples were ultrasonicated in ethanol to form a homogeneous suspension, dropped on a 200 mesh copper TEM grid coated with a thin amorphous carbon film, and then allowed to dry in air. The specific surface area, pore volume and the average pore size of the materials were determined by measuring the N2 adsorption/ desorption isotherms at 196
C in a Micromeritics ASAP 2020 surface area and porosity analyzer (Micromeritics Instrument Co., USA) using the Brunauer–Emmett–Teller (BET) and Barret– Joyner–Halenda (BJH) methods. All samples were out gassed at 150
C under vacuum for 1 h prior to the N2 adsorption measurements. 2.6. CO2 adsorption measurements

(1 0 1)

(0 0 2)

Intensity (a.u.)

(0 0 2)

CO2 adsorption capacity of the prepared graphene materials was determined volumetrically in a Micromeritics ASAP 2020 adsorption apparatus (Micromeritics Instrument Co., USA). The adsorption isotherms were obtained at three different temperatures (0, 25 and 50
C) and pressures up to 1 bar. The adsorption temperature was controlled by using a Dewar bottle with a circulating jacket connected to a thermostatic bath utilizing water as the coolant. About 100 mg of sample was used for the CO2 adsorption studies. Before each adsorption experiment, all the samples were degassed at 150
C under vacuum for 1 h to desorb any moisture and organics. The CO2 adsorption kinetics (adsorption amount as a function of time) was also measured in the Micromeritics ASAP 2020 system using a built-in function (“rate of adsorption”) at the same time when the adsorption equilibrium data were collected. The change in gas pressure and adsorption volume with time, after the CO2 reservoir was connected to the sample chamber, was recorded and then converted into transient adsorption uptakes to generate the adsorption kinetics. The adsorption equilibrium amount was considered as the final adsorption amount at the terminal pressure and temperature. In order to estimate the selectivity of the synthesized materials, N2 adsorption isotherms at 25
C were also measured using an identical procedure. Ultra high purity (99.9%) grade gas sources were used throughout the study.

GO GPN GPN-200 GPN-400 GPN-600 GPN-800

10 15 20 25 30 35 40 45 50 55 60 65 70

2θ (degree) Fig. 1. XRD spectra of GO, GPN and the GPN-X samples.

3. Results and discussion 3.1. Materials characterization A wide variety of diffraction, spectroscopy, and microscopy techniques were used to examine the effect of heat treatment on the structure and morphology of graphene nanosheets. Fig. 1 presents the XRD pattern of the as-prepared GO, GPN, and the GPN-X samples. GO showed a strong peak from (0 0 2) diffraction at 2u of 11
, corresponding to an interlayer distance (d-spacing) of 0.80 nm, which was much larger than that of natural graphite (d = 0.33 nm) [24]. The large interlayer spacing of GO can be ascribed to the intercalation of oxygen functional groups and water molecules during the Hummer’s process [25]. Upon chemical reduction with hydrazine, the distinctive GO peak at 11
completely disappeared and two new peaks centered at around 2u = 24
(d = 0.37 nm) and 44
(d = 0.21 nm), conforming to the graphitic (0 0 2) and (1 0 1) crystal planes, appeared for GPN. The evolution of these peaks is due to restoration of the van der Waals’ interaction between the carbon frameworks on the graphene sheets upon reduction. The substantial decrease in d-spacing suggests the removal of oxygen groups and water molecules from the interlayer of graphene sheets. Nevertheless, the broad nature of the peaks and their weak intensity reveals a highly disordered and random stacking of the graphene layers [26], which could be due to incomplete deoxygenation of GO by hydrazine as reported by previous researchers [22,27]. After the 2 h heat treatment, the (0 0 2) reflection shifted to larger angles (26
), suggesting a significant shrinking in interlayer spacing (d = 0.34 nm) and a dramatic restoration of the pristine graphite structure. The progressive narrowing of the (0 0 2) reflection with an increasing treatment temperature (from 200 to 800
C) indicates a considerable improvement in structural order. In addition, the sharp and intense peak profile for the GPN-X materials implies high graphitization and well-oriented stacking among the graphene sheets, which was possibly because of the removal of almost all oxygen functional groups from the carbon surface during the thermal treatment procedure [28]. In order to get a more in-depth understanding of the effect of thermal treatment on the structural features of graphene nanosheets, the stoichiometry of the GPN-X samples was measured by XPS and compared with that of GO and GPN (Fig. 2). The high resolution C1s XPS spectrum of GO comprised three different chemically shifted components which can be deconvoluted into: carbon sp2 (C C: 285.0 eV); hydroxyl/carbonyl groups (C OH/ C¼O: 287.1 eV); and carboxyl/carboxylate groups (O C¼O: 288.7 eV) (Fig. 2a). The C1s core-level spectrum of GPN also contained the same oxygenated functional groups with greatly reduced peak intensities (Fig. 2b), demonstrating extensive but not complete deoxygenation by hydrazine as envisaged from the XRD measurements. Further decrease in the oxygenated carbon peaks in GPN-200 can be ascribed to the evaporation of physisorbed water molecules (Fig. 2c). The near complete disappearance of the O C¼O peak in GPN-400 (Fig. 2d) was likely due to the disintegration of carboxylic acid groups and carboxylic anhydrides into CO2 and CO at temperatures in the 220–300
C range and >350
C, respectively [29]. However, no significant change in the C¼O peak intensity was noted for both GPN-400 and GPN-600 (Fig. 2d and e). On the contrary, GPN-800 was virtually free of oxygen functionalities (Fig. 2f), which can be attributed to the near complete degeneration of carbonyl moieties to CO at 700–770
C [29]. FESEM and TEM images were also captured to examine the surface morphology of the graphene samples before and after the heating experiments (Fig. 3). As observed by FESEM, graphene without any thermal modification exhibited a smooth planar

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

C-O/C=O

(a)

53

C-C

(b) Intensity (cps)

Intensity (cps)

C-C

O=C-O

292

290

288

286

284

282

C-O/C=O

O=C-O

292

290

Binding Energy (eV) C-C

C=O O=C-O

292

290

288

286

284

290

286

284

282

284

282

C-C

(f) Intensity (cps)

Intensity (cps)

288

288

Binding Energy (eV)

C-C

290

282

C=O

292

282

C=O

292

284

C-C

Binding Energy (eV)

(e)

286

(d) Intensity (cps)

Intensity (cps)

(c)

288

Binding Energy (eV)

286

284

282

Binding Energy (eV)

292

290

288

286

Binding Energy (eV)

Fig. 2. High-resolution deconvoluted C1s XPS spectra of (a) GO, (b) GPN, (c) GPN-200, (d) GPN-400, (e) GPN-600 and (f) GPN-800.

structure with fluffy appearance (Fig. 3a), typical for graphene materials produced via hydrazine reduction of exfoliated GO. In contrast, the heat treated graphene materials developed a distinct interconnected porous structure with a highly wrinkled and crumpled paper-like external morphology (Fig. 3b–e), which arises due to the decomposition of the labile oxygen functional groups (that otherwise remain even after reduction with hydrazine) with release of physisorbed water during thermal processing. The out gassing of CO, CO2, and H2O induce strong forces that attack the carbonaceous surface, leading to the formation of porous graphene frameworks [21], as outlined in Fig. 4. The pore formation by gas evolution implies that the pores are interconnected from the inside to the surface of the material [30], and such interconnected

hierarchical pore networks are crucial to achieve high volume CO2 removal from flue gases [31]. The wrinkled and corrugated nature of the GPN-X samples was also confirmed by TEM (Fig. 3f and g). According to previous studies, the formation of corrugations on graphene sheets can provide larger accessible area in comparison with its conventional counterpart [32]. The selected-area electron diffraction (SAED) pattern of GPN-800 is illustrated in Fig. 3h. According to Joint Committee on Powder Diffraction Standards (JCPDS) card No. 75–1621, the two diffraction rings can be attributed to the diffraction of graphite (0 0 2) and (1 0 1) planes [33], which is in accordance with the XRD data. N2 adsorption isotherms measured at 196
C were further used to characterize the porous structure of the heat treated

54

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

Fig. 3. FESEM images of GPN (a), GPN-200 (b), GPN-400 (c), GPN-600 (d) and GPN-800 (e). TEM images (f, g) and SAED pattern (h) of GPN-800. The scale bars in (a)–(e), (f) and (g) are 1000 nm, 100 nm and 50 nm, respectively.

Fig. 4. Schematic of pore formation on graphene nanosheets through gas evolution during thermal processing under N2 atmosphere.

graphene samples. The N2 adsorption/desorption isotherms and the corresponding pore size distribution (PSD) of the GPN-X materials are displayed in Fig. 5. The initial increase in adsorption capacity at low relative pressures (P/P0 < 0.1) was associated with the filling of large micropores (wherein an adsorbate monolayer is formed on the pore surface) as well as monolayer—initial multilayer adsorption in mesopores with narrow necks and wide bodies (often referred to as ‘ink-bottle’ pores) that are interconnected to each other [34,35]. The upward deviation at higher P/P0 was due to the progressive filling of the ink-bottle shaped mesopores by the process of capillary condensation [36], giving rise to Type IV isotherm according to the classification of porous materials by The International Union of Pure and Applied Chemistry (IUPAC) (Fig. 5a) [37]. In such a case, the pore network empties only when the relative pressure is below a characteristic percolation threshold, associated with the onset of a continuous cluster of pores open to the surface [35]. Consequently, the desorption branch was significantly steeper than the adsorption branch (Fig. 5a), which resulted in a triangular hysteresis loop of Type H2 as per IUPAC specification [37]. The differential PSD curves evaluated from the desorption branches of the N2 isotherms using

the BJH method shows that the GPN-X samples had a narrow PSD with the peak maxima located primarily in the large-micropore (>1.5 nm)/small-mesopore (2–3 nm) range (Fig. 5b). A detailed quantitative estimate of the textural properties of the untreated and treated graphene materials is listed in Table 1. As can be seen, the thermal treatment process caused a significant enhancement in the specific surface area of graphene sheets (SBET = 94 m2 g1 for GPN to SBET = 484 m2 g1 for GPN-800). The total pore volume also improved remarkably upon heat treatment under flowing N2 (Vtot = 0.07 cm3 g1 for GPN to Vtot = 0.68 cm3 g1 for GPN-800), with greater increment in both micropore and mesopore volume with increasing temperature (from Vmicro/Vmeso = 0.04/0.22 cm3 g1 at 200
C to Vmicro/Vmeso = 0.09/0.58 cm3 g1 at 800
C). Among the synthesized graphene samples, GPN-800 registered the largest BET surface area and total pore volume, which is likely due to the creation of greater porosity in the bulk material by the reaction of CO2 (evolved from the decomposition of carboxyl/carboxylate groups) with the graphitic carbon atoms to produce CO at temperatures >700
C according to the Boudouard reaction (CO2 + C ! 2CO) [38]. It thus becomes apparent that thermal treatment at high-temperatures of 800
C in

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

Volume Adsorbed (cm /g)

(a) 3

GPN GPN-200 GPN-400 GPN-600 GPN-800

0.0

0.2

0.4

0.6

0.8

1.0

Relative Pressure (P/P0)

dV/dlog(D)

(b)

GPN GPN-200 GPN-400 GPN-600 GPN-800

0

1

2

3

4

5

6

7

8

9

10

Pore Size (nm) Fig. 5. (a) N2 adsorption/desorption isotherms at 196
C (solid symbols denote adsorption and open symbols denote desorption) and (b) pore size distribution as determined using the BJH method for the as-prepared GPN and GPN-X materials.

an inert environment for 2 h is the most effective for the development of highly ordered graphene sheets with high surface area and hierarchical interconnected nanoporous structure. 3.2. Adsorption isotherms Low pressure static CO2 adsorption measurements were initially conducted at 0
C. Fig. 6a presents the CO2 adsorption isotherms while Table 1 lists the CO2 uptakes at 0
C and 1 bar for

55

all the samples. All the materials showed a steep rise in the low pressure range and reached a maximum at 1 bar. However, as expected, the CO2 capacity increased with the specific surface area and total pore volume in the following order: GPN < GPN200 < GPN-400 < GPN-600 < GPN-800. The GPN-800 sample had significantly higher SBET,Vmicro and Vmeso, which resulted in the best CO2 uptake of 2.89 mmol g1 at 0
C and 1 bar. To further examine the effect of thermal treatment process on the CO2 adsorption behavior of graphene sheets at low pressure, the evolution of the CO2 adsorption capacity (mmol g1) at 0
C and 1 bar for the GPN-X samples as a function of the different textural parameters was studied as shown in Fig. 6b and c. A strong correlation was observed between the adsorption capacity and the total pore volume under the experimental conditions (R2 = 0.976) (Fig. 6c), although there was a better correlation with the specific surface area in terms of statistical analysis (R2 = 0.989) (Fig. 6b), suggesting that a combination of pore filling and surface adsorption was responsible for the large adsorption capacity of GPN-800. However, the precise mechanism of this adsorption process is rather unclear, due to the complex ink-bottle pore geometry and their interconnected nature in the prepared GPN-X samples. Nevertheless, these correlations reinforce the importance of high-temperature treatment in the design and development of high volume graphene-based CO2 adsorbents. GPN-800 was subsequently tested at high temperature conditions of up to 50
C because of its highest CO2 capacity, as shown in Fig. 6d. The amount of CO2 adsorbed on GPN-800 decreased with increasing temperature which was due to increase in the thermal energy of CO2 molecules at elevated temperatures, leading to lower adsorption uptakes. A comparative analysis was also made between the maximum CO2 capacities of GPN-800 and other competing adsorbent materials reported in the literature (see Supporting information, Table S1). GPN-800 showed higher CO2 adsorption performance in comparison to most of the other materials, notably, zeolites, commercial activated carbon, modified activated carbons, some MOFs, mesoporous alumina as well as some other graphene-based adsorbents under similar temperature and pressure conditions. Besides, a key attribute of GPN-800 is that it has nearly no oxygen functionalities on its surface, making it extremely hydrophobic [39]. This novel adsorption material is thus a highly promising candidate for post-combustion CO2 capture applications, where a considerable volume of combustion water vapor in the flue gas is present and can have an adverse impact on the CO2 adsorption capacity and selectivity of the adsorbent. In the design and analysis of an adsorption-based separation process, equilibrium isotherm models can provide valuable insights into the interaction of adsorbate molecules with the adsorbent surface, enabling a more detailed interpretation of the experimental observations. Therefore, in this study, the twoparameter Langmuir and the three-parameter Toth isotherm models were used to correlate the temperature dependent CO2 experimental isotherm data of GPN-800.

Table 1 Textural properties and CO2 adsorption capacity of the GPN-X materials. Sample

SBET (m2 g1)

GPN GPN-200 GPN-400 GPN-600 GPN-800

94 185 302 388 484

a b c d e

a

Vtot (cm3 g1) 0.071 0.268 0.435 0.547 0.682

Specific surface area calculated using the BET method. Total pore volume at relative pressure of P/P0 = 0.99. Micropore volume calculated by t-plot method. Mesopore volume = V tot–Vmicro. CO2 uptake at 0
C and 1 bar.

b

Vmicro (cm3 g1) NA 0.047 0.066 0.079 0.094

c

Vmeso (cm3 g1) NA 0.221 0.369 0.468 0.588

d

CO2 uptake (mmol g1) 0.817 1.491 1.982 2.516 2.894

e

56

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

3.0

GPN GPN-200 GPN-400 GPN-600 GPN-800

2.5 2.0 1.5 1.0 0.5

0.2

0.4

0.6

0.8

2.5

2.0

1.5

(a)

0.0 0.0

2

R =0.989 CO2 Uptake (mmol/g)

CO2 Adsorbed (mmol/g)

3.0

1.0

(b)

150

200

250

300

3.0

3.5

2

CO2 Adsorbed (mmol/g)

CO2 Uptake (mmol/g)

2.5

2.0

1.5

0.4

0.5

450

500

o

0 C o 25 C o 50 C Langmuir Toth

0.6

3.0 2.5 2.0 1.5 1.0 0.5

(c) 0.3

400

SBET (m /g)

R =0.976

0.2

350 2

Pressure (bar)

(d)

0.0 0.0

0.7

0.2

0.4

3

Vtotal (cm /g)

0.6

0.8

1.0

Pressure (bar)

Fig. 6. (a) Pure component CO2 adsorption isotherms for GPN and GPN-X samples at 0
C. Dependence of the CO2 adsorption capacities of the prepared porous GPN-X materials measured at 0
C and 1 bar as a function of (b) BET surface area and (c) total pore volume. (d) Nonlinear fit of Langmuir and Toth isotherm models to the experimental CO2 adsorption equilibrium data of GPN-800 at different temperatures.

The Langmuir isotherm is one of the most widely adopted adsorption equilibrium model. It assumes that there is only one type of binding site with equal affinity and is defined by the following equation [40]: q¼

qm K L P 1 þ KLP

ð1Þ

where q (mmol g1) is the amount adsorbed, qm (mmol g1) is the maximum adsorption capacity, P (bar) is the adsorbate pressure, and KL (bar1) is the Langmuir equilibrium constant. The Toth isotherm is frequently employed to describe the adsorption equilibria of gases and vapors because of its correct behavior at both low (Henry-type limit) and high (saturation limit)

pressure ranges, and has the following formulation [41]: q¼h

qs bP

ð2Þ

i t 1=t

1 þ ðbP Þ

where qs (mmol g1) is the saturation loading, b (bar1) represents the adsorption affinity, and t is a measure of adsorbent heterogeneity. If t equals 1, the Toth equation is transformed to the Langmuir model, suggesting monolayer adsorption on a homogenous surface. If t deviates away from unity, the gas–solid adsorption system is deemed to be heterogeneous [42]. The Toth isotherm parameters qs, b and t are temperaturedependent according to the following equations [43]:    T ð3Þ qs ¼ qs;0 exp x 1  T ref

Table 2 Isotherm parameters and kinetic constants for CO2 adsorption on GPN-800 at different temperatures. T (
C)

Langmuir

Toth 1

qm (mmol g 0 25 50

)

3.504 2.823 2.078

KL (bar

1

)

3.583 2.825 1.667

R

qs (mmol g1)

b (bar1)

t

R2

0.981 0.988 0.995

5.314 4.745 3.456

4.723 2.118 1.223

0.548 0.625 0.756

0.999 0.998 0.998

2

Temperature-dependent Toth isotherm parameters (mmol g1)

Tref (K)

qs,

298

4.745

0

x

b0 (bar1)

DHads (kJ mol1)

t0

a

1.349

2.118

21.697

0.625

0.841

Avrami kinetic constants T (
C)

qe,exp (mmol g1)

qe,cal (mmol g1)

kA (s1)

nA

R2

0 25 50

2.894 2.192 1.344

2.865 2.141 1.315

0.113 0.074 0.069

0.652 0.683 0.741

0.995 0.989 0.991

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

   DHads T ref 1 b ¼ b0 exp Rg T ref T

ð4Þ

  T t ¼ t0 þ a 1  ref T

ð5Þ

where qs,0, b0 and t0 are the corresponding parameters at a selected reference temperature Tref (K), T is the temperature, DHads (kJ mol1) is a measure of the heat of adsorption, Rg is the ideal gas constant (8.314 J mol1 K1), and, x and a are the fitting parameters. The fitting of the models to the single component CO2 adsorption data was performed by a nonlinear regression analysis using the software Origin Pro 8.0 (OriginLab Corp., USA) (Fig. 6d). The model constants and the correlation coefficients (R2) thus obtained are presented in Table 2. Although the analysis of the R2 values suggests that both the Langmuir and Toth models could adequately fit the experimental equilibrium data over the entire temperature and pressure range, the Toth equation showed a better correlation than the Langmuir isotherm at lower temperatures, suggesting that CO2 uptake by GPN-800 involved the formation of successive multimolecular layers [43]. Moreover, the Toth constant t greatly deviated from unity, indicating a high degree of surface heterogeneity for adsorption of CO2 onto GPN-800. However, at higher temperatures, the surface behaved in a somewhat homogeneous manner as evident from the t value at 50
C (see Table 2). This fact is also corroborated by the greater

accuracy of the Langmuir model to describe the experimental data acquired at 50
C. Additionally, taking Tref = 25
C (298 K), the temperature-dependent Toth parameters including the heat of adsorption were determined via a least square analysis and are also listed in Table 2. As heats of adsorption are generally in the range of 10–20 kJ mol1 for physical adsorption and 40–80 kJ mol1 for chemical adsorption [44], it appears that CO2 adsorption onto GPN-800 was a physical adsorption process (DHads = 21.69 kJ mol1). From a practical point of view, this finding implies that GPN-800 does not have an appreciable number of chemically active functional groups that are capable of interacting with the reactive flue gas contaminants such as SOX, NOX, and HCl; these chemical species could otherwise have a considerable impact on the CO2 separation process [45], demonstrating the potential benefits of thermally treated graphene sheets for CO2 capture applications. 3.3. Adsorption affinity It is well known that the Henry’s law constant (KH) is directly related to the affinity between the adsorbate and the adsorbent, since molecule-surface forces are predominant at low pressure [46]. The higher the value of the constant, the stronger is the interaction between the adsorbate–adsorbent pair. Hence the virial model (Eq. (6)) was considered to estimate the Henry’s law constant for adsorption of CO2 onto GPN-800 [47].   P 1 3 4 ¼ exp 2A1 q þ A2 q2 þ A3 q3 þ ::: ð6Þ q KH 2 3

0.0

3.5

CO2 Adsorbed (mmol/g)

(a)

-0.5 -1.0

ln (P/q)

57

-1.5 -2.0 -2.5 -3.0

o

0 C o 25 C o 50 C

-3.5 -4.0 0.0

( b)

3.0 2.5 2.0 1.5

o

1.0

0 C o 25 C o 50 C Avrami

0.5 0.0

0.5

1.0

1.5

2.0

2.5

3.0

0

50

100

3.5

0 C o 25 C o 50 C

(c)

200

250

-45

o

3.0

150

Time (s)

q (mmol/g)

(d) -40

Qst (kJ/mol)

qt (mmol/g)

2.5 2.0 1.5 1.0

-35

-30

-25

0.5 (iii)

(ii)

(i)

0.0 0

2

4

6

8 0.5

10 0.5

t (s )

12

14

16

-20 0.0

0.2

0.4

0.6

0.8

1.0

Fractional Loading (q /qs)

Fig. 7. (a) Virial plots of CO2 for GPN-800 at different temperatures. (b) CO2 adsorption kinetics for GPN-800 at different temperatures and Avrami model fit to the experimental data. (c) Intraparticle diffusion model plots for adsorption of CO2 on GPN-800 at different temperatures. (d) Isosteric heat of adsorption of pure CO2 on GPN800 as a function of fractional loading.

58

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

where A1, A2, A3 are the virial coefficients. A plot of ln (P/q) vs q should approach the axis linearly as q ! 0 with slope 2A1 and intercept ln (KH). Fig. 7a illustrates the virial plots of CO2 on GPN800 at the studied temperatures. KH for CO2 adsorption onto GPN800 was found to be 0.19,0.11, and 0.04 mmol g1 bar1 at 0, 25 and 50
C, respectively. The fairly high KH values at all temperatures suggest that CO2 was strongly physisorbed onto GPN-800, primarily through van der Waals forces (also known as dispersion  repulsion forces) and electrostatic forces (also known as Coulombic interactions), which mainly arise from quadrupole– quadrupole interactions between CO2 and the defective graphene surface [48,49]. Furthermore, the Henry's constant decreased substantially with increasing adsorption temperature. It thus seems that CO2 adsorption at low pressures was sensitive to temperature causing dramatic changes in the CO2 uptake properties of GPN-800. 3.4. Adsorption kinetics Efficient capture of CO2 from flue gases also requires adsorbents with fast CO2 uptake. The rapid adsorption/desorption can considerably drop the amount of an adsorbent required to capture CO2 from a given volume of flue gas, thus reducing the mass of the adsorbent bed and hence the size of the CO2 recovery apparatus [50]. Therefore, in the present study, the CO2 uptake kinetics of GPN-800 was also measured. As can be seen in Fig. 7b, it took less than a minute to saturate GPN-800 to 50% of its maximum CO2 capacity, and a complete saturation was achieved within 3 min at all temperatures, suggesting that GPN-800 can effectively separate CO2 from flue gases while operating with short adsorption cycle times, which would indeed be economically advantageous for practical deployment. The fast kinetic separation can be attributed to the unique interconnected pore network of GPN-800. A quantitative evaluation of the CO2 uptake kinetics of GPN800 based on the Avrami model was subsequently carried out by directly fitting the experimental data to Eq. (7) employing nonlinear regression technique (Fig. 7b) [51]. nA

qt ¼ qe ½1  eðkA tÞ 

ð7Þ

1

where qt (mmol g ) is the amount of CO2 adsorbed at time t, qe (mmol g1) is the amount of CO2 adsorbed at equilibrium, kA (s1) is the Avrami kinetic constant and nA is the Avrami exponent. These parameter estimates and the corresponding R2 values are summarized in Table 2. The qe values obtained from the model fit were consistent with the experimental qe values at the studied temperature range, and also the R2 values were significantly high (>0.99), indicating that the Avrami model could reasonably define the CO2 adsorption process of GPN-800. While the fractional values of the Avrami exponent reflects that more than one pathway was involved in CO2 adsorption, a sharp decrease in the Avrami constant with temperature suggests that physisorption was predominant at lower temperatures. To further investigate the mechanism of CO2 adsorption onto GPN-800, the effect of intraparticle diffusion was considered by analyzing the isothermal kinetic data with the Weber–Morris model, described by the following equation [52]: qt ¼ kid t0:5 þ C

ð8Þ 1 0.5

) is the intraparticle diffusion rate where kid (mmol g s constant evaluated from the slope of the Weber–Morris plot of qt vs t0.5(s0.5), and C is the intercept. If C = 0, the intraparticle diffusion is considered as the rate-limiting step, whereas, if C 6¼ 0, intraparticle diffusion is not the sole rate-controlling step. Fig. 7c shows the Weber–Morris plot for CO2 adsorption on GPN-800 at 0, 25, and 50
C. All the plots were multi-linear with three distinct

regions, implying that more than one kinetic mechanism was involved [53]. The first linear portion (Region I) can be attributed to the diffusion of CO2 through the bulk gas phase to the external surface of GPN-800. The second portion (Region II) represents the gradual adsorption stage, where intraparticle diffusion takes place. The third portion (Region III) is indicative of the final equilibrium step during which intraparticle diffusion slows down due to saturation of the active sites. Based on these findings, it is clear that apart from intraparticle diffusion, boundary layer diffusion also controlled the adsorption of CO2 to a certain extent. 3.5. Adsorption energetics The isosteric heat of adsorption (Qst), defined as the difference in the partial molar enthalpy of the adsorbate between the gas phase and the adsorbed phase [54], is a key thermodynamic variable for the practical design of CO2 capture units, such as pressure swing adsorption systems, because it governs the changes in the adsorbent temperature inside the adsorber during the adsorption/desorption cycle [55]. Such changes can influence the local adsorption equilibria and kinetics, and thus, the overall efficiency of the gas separation process [56]. By using the temperature-dependent Toth isotherm parameters, the isosteric heat of CO2 adsorption for GPN-800 was determined from the van’t Hoff equation as follows [24]: Q st ¼ Q 0;st ð¼ ( D" Hads Þ 

aRg T ref t

q

ln  1=t qts  qt

#

"

lnðq=qs Þ   1=t 1  q=qs

#) ð9Þ

Evidently, at zero fractional loading the isosteric heat is equal to the heat of adsorption, i.e., Qst = DHads = 21.69 kJ mol1 at q/qs = 0. The low Q0,st makes GPN-800 an attractive candidate for removing CO2 from flue gases of fossil-fired power plants because of the lower energy requirements for regeneration. The plot of Qst as a function of fractional loading for GPN-800 is presented in Fig. 7d. The Qst increased steadily with CO2 loading, and reached the value of 42.13 kJ mol1, close to the saturation point. This unusual increase is likely due to the occurrence of lateral interactions between CO2 molecules at higher loadings rather than adsorption onto the graphene surface as was verified by fitting the room temperature CO2 equilibrium data to the Fowler–Guggenheim model (Eq. (10)) [57]:   Pð1  uÞ 2W u ð10Þ ¼ lnK FG þ ln RT u where u is the fractional coverage, KFG (bar mmol1) is the Fowler– Guggenheim equilibrium constant, and W (kJ mol1) is the interaction energy between adsorbed molecules. If the interaction between the adsorbed molecules is attractive (i.e., W is positive), the heat of adsorption increases with surface loading due to increased interaction between the adsorbed molecules. Conversely, if the interaction among adsorbed molecules is repulsive (i.e., W is negative), the heat of adsorption decreases with surface loading. When there is no interaction between the adsorbed molecules, W = 0 [58,59]. The plot of the Fowler–Guggenheim equation, carried out by calculating u using the maximum saturation loading obtained from the Toth model, showed good linearity (R2 = 0.985). The interaction energy turned out to be positive (W = 3.52 kJ mol1), confirming the existence of attractive intermolecular forces between the adsorbed quadrupolar CO2 molecules. Hence, with more CO2 molecules present at higher surface loadings the Qst values were higher than at lower surface loadings. A similar dependence of the isosteric heat on surface loading has also been reported for adsorption of CO2 on other carbon variants such as

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

0.4

2.5

CO2 N2

0.3 S=6

0.2

(a)

0.1

0.0 0.0

Volume Adsorbed (mmol/g)

Volume Adsorbed (mmol/g)

0.5

CO2

2.0

N2

1.5 S = 43

1.0

(b)

0.5 0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

Pressure (bar)

0.6

0.8

1.0

2.5

Adsorption Desorption

2.192

CO2 Adsorbed (mmol/g)

CO2 Adsorbed (mmol/g)

0.4

Pressure (bar)

2.5

2.0

1.5

1.0

0.5

0.0 0.0

59

2.184

2.170

2.179

2.167

2.0

1.5

1.0

1

2

3

4

5

0.5

(c)

(d) 0.0

0.2

0.4

0.6

0.8

1.0

Adsorption Cycle

Pressure (bar) Fig. 8. CO2 and N2 adsorption isotherms for (a) GPN and (b) GPN-800 as measured at 25
C. (c) CO2adsorption–desorption isotherms for GPN-800 at 25
C. (d) Cyclic CO2 adsorption performance of GPN-800 at 25
C.

dahlia-like carbon nanohorn aggregates and single-wall carbon nanotube bundles [60,61]. 3.6. Adsorbent selectivity, regeneration and reuse In addition to high CO2 uptake, an effective CO2 adsorbent must also have high selectivity for CO2 against other gases, especially N2, so that high purity CO2 can be extracted from the flue gas for subsequent sequestration [50]. For that reason, the N2 adsorption isotherms at 25
C for GPN and GPN-800 were also measured and compared with the corresponding CO2 isotherms (Fig. 8a and b). The N2 adsorption capacities at 25
C and 1 bar for GPN and GPN-800 were 0.07 and 0.04 mmol g1, respectively, whereas the CO2 adsorption capacities at identical temperature and pressure were 0.45 and 2.19 mmol g1 for GPN and GPN-800, respectively. This finding can be attributed to the preferential adsorption of CO2 molecules with a larger polarizability and quadrupole moment (29.11 1025 cm3 and 4.30  1026 esu1 cm1, respectively) to that of N2 (17.40  1025 cm3 and 1.52  1026 esu1 cm1, respectively) [4,24]. Consequently, the CO2/N2 selectivity of GPN-800 (43) was sevenfold higher than GPN (6), and was also significantly higher than those recently reported for other carbon-based solid adsorbents, including commercial activated carbon (NORIT1 R2030CO2), measured under quite similar conditions [62–69]. Since exhaust gas emanating from post-combustion fossil-fuelled power plants consists of 75% N2 and 15% CO2 at a pressure of approximately

1 bar, a comparison was also made for the CO2/N2 adsorption selectivity of the two materials normalized to the flue gas composition according to the following equation [16]: S¼

q1 =q2 P1 =P2

ð11Þ

where q1 and q2 are the adsorbed amount of CO2 at 0.15 bar and N2 at 0.85 bar, respectively, and, P1 and P2 represent the equilibrium partial pressure of CO2 (0.15 bar) and N2 (0.85 bar) in the bulk gas phase, respectively. It was found that the CO2/N2 adsorption selectivity increased from 17 for GPN to 112 for GPN-800, further ascertaining the suitability of GPN-800 for deployment within post-combustion CO2 capture systems. The durable cyclic adsorption/regeneration behavior of an adsorbent is also essential for long-term operation, which in turn has a direct impact on the overall economics of the capture process. Five consecutive CO2 adsorption/desorption cycles were carried out by swinging the pressure between vacuum (<0.01 bar) and 1 bar at 25
C to evaluate the stability of GPN-800. In each adsorption/desorption cycle, CO2 adsorption onto GPN-800 was found to be completely reversible and no significant hysteresis was observed (Fig. 8c), suggesting the facile desorption of CO2 during regeneration of the adsorbent. Moreover, the CO2 adsorption performance of GPN-800 was fairly stable, with less than 5% drop in CO2 uptake capacity even after five adsorption/ desorption cycles (Fig. 8d), which is critical to lower the cost of pollution control in practical CO2-capture applications.

60

S. Chowdhury, R. Balasubramanian / Journal of CO2 Utilization 13 (2016) 50–60

4. Conclusion In summary, we have demonstrated a new, sustainable, and environmentally friendly approach, based on heat treatment of graphene nanosheets, for the facile development of advanced solid adsorbents for post-combustion CO2 capture applications. Graphene heated at 800
C under N2 atmosphere for 2 h showed the highest surface area (484 m2 g1) and largest pore volume (0.68 cm3 g1) with a narrow distribution of pores in the largemicropore/small-mesopore range, leading to a high CO2 capacity (2.19 mmol g1), fast kinetic separation (<3 min), and exceptionally high CO2/N2 adsorption selectivity at 25
C and 1 bar. The isosteric heat of adsorption had an unusual dependence on surface loading which could be due to attractive intermolecular forces between the adsorbed quadrupolar CO2 molecules. Most importantly, GPN-800 could be completely regenerated at mild conditions with relatively low energy consumption, retaining more than 95% of its maximum CO2 adsorption ability after five cycles of adsorption/desorption. Overall, our results represent one of the first studies to demonstrate that thermal treatment at hightemperatures of 800
C is an attractive option to synthesize highly ordered graphene layers with interconnected hierarchical pore networks for enhanced CO2 removal from power plant exhausts and other large industrial sources. Acknowledgements The authors thank Madam Loy Gek Luan of the Cryo-Electron Microscopy Facility at Center for Bio-imaging Science, Department of Biological Science, National University of Singapore for her scientific and technical assistance. Shamik Chowdhury gratefully acknowledges the financial support provided by the National University of Singapore for his Doctoral study. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jcou.2015.12.001. References [1] V. Scott, S. Gilfillan, N. Markusson, H. Chalmers, R.S. Haszeldine, Nat. Clim. Change 3 (2013) 105–111. [2] M. Pera-Titus, Chem. Rev. 114 (2014) 1413–1492. [3] J. Wilcox, R. Haghpanah, E.C. Rupp, J. He, K. Lee, Annu. Rev. Chem. Biomol. Eng. 5 (2014) 479–505. [4] R. Balasubramanian, S. Chowdhury, J. Mater. Chem. A 3 (2015) 21968–21989. [5] O. Cheung, N. Hedin, RSC Adv. 4 (2014) 14480–14494. [6] J. Wang, L. Huang, R. Yang, Z. Zhang, J. Wu, Y. Gao, Q. Wang, D. O’Hare, Z. Zhong, Energy Environ. Sci. 7 (2014) 3478–3518. [7] S.-Y. Lee, S.-J. Park, J. Ind. Eng. Chem. 23 (2015) 1–11. [8] S. Jana, S. Das, C. Ghosh, A. Maity, M. Pradhan, Sci. Rep. 5 (2015) 8711. [9] G. Srinivas, J. Burress, T. Yildirim, Energy Environ. Sci. 5 (2012) 6453–6459. [10] R. Kumar, V.M. Suresh, T.K. Maji, C.N.R. Rao, Chem. Commum. 50 (2014) 2015– 2017. [11] Z.-Y. Sui, Y. Cui, J.-H. Zhu, B.-H. Han, ACS Appl. Mater. Interfaces 5 (2013) 9172– 9179. [12] G. Ning, C. Xu, L. Mu, G. Chen, G. Wang, J. Gao, Z. Fan, W. Qian, F. Wei, Chem. Commun. 48 (2012) 6815–6817. [13] H. Seema, K.C. Kemp, N.H. Le, S.-W. Park, V. Chandra, J.W. Lee, K.S. Kim, Carbon 66 (2014) 320–326. [14] K. Xia, X. Tian, S. Fei, K. You, Int. J. Hydrogen Energy 39 (2014) 11047–11054. [15] Z.-Y. Sui, Q.-H. Meng, J.-T. Li, J.-H. Zhu, Y. Cui, B.-H. Han, J. Mater. Chem. A 2 (2014) 9891–9898. [16] Z.-Y. Sui, B.-H. Han, Carbon 82 (2015) 590–598. [17] S. Chowdhury, R. Balasubramanian, Adv. Colloid Interface Sci. 204 (2014) 35– 56. [18] L. Jiang, Z. Fan, Nanoscale 6 (2014) 1922–1945. [19] W. Yuan, J. Chen, G. Shi, Mater. Today 17 (2014) 77–85.

[20] C. Botas, P. Alvarez, C. Blanco, R. Santamaria, M. Granda, M.D. Gutierrez, Carbon 52 (2013) 476–485. [21] N. Xiao, H. Tan, J. Zhu, L. Tan, X. Rui, X. Dong, Q. Yan, ACS Appl. Mater. Interfaces 5 (2013) 9656–9662. [22] H. Zhang, V.V. Bhat, N.C. Gallego, C.I. Contescu, ACS Appl. Mater. Interfaces 4 (2012) 3239–3246. [23] D.C. Marcano, D.V. Kosynkin, J.M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, L.B. Alemany, W. Lu, J.M. Tour, ACS Nano 4 (2010) 4806–4814. [24] S. Chowdhury, G.K. Parshetti, R. Balasubramanian, Chem. Eng. J. 263 (2015) 374–384. [25] S. Park, J. An, J.R. Potts, A. Velamakanni, S. Murali, R.S. Ruoff, Carbon 49 (2011) 3019–3023. [26] X. Wu, D. Yang, C. Wang, Y. Jiang, T. Wei, Z. Fan, Carbon 92 (2015) 26–30. [27] L. Chen, J. Lei, F. Wang, G. Wang, H. Feng, RSC Adv. 5 (2015) 40148–40153. [28] Q. Zhang, X. Xu, H. Li, G. Xiong, H. Hu, T.S. Fisher, Carbon 93 (2015) 659–670. [29] M. Almarri, X. Ma, C. Song, Energy Fuels 23 (2009) 3940–3947. [30] H. Kim, H.-D. Lim, S.-W. Kim, J. Hing, D.-H. Seo, D. -c. Kim, S. Jeon, S. Park, K. Kang, Sci. Rep. 3 (2013) 1506. [31] S. Dutta, A. Bhaumik, K.C.-W. Wu, Energy Environ. Sci. 7 (2014) 3574–3592. [32] J.-Y. Hong, X. Yu, B.M. Bak, C. Pang, H.S. Park, Carbon 83 (2015) 71–78. [33] J. Ma, G.-y. Li, Z.-y. Chu, X.d. Li, Y.-h. Li, T.-j. Hu, Carbon 69 (2014) 634–637. [34] V.-T. Hoang, Q. Huang, M. Eic, T.-O. Do, S. Kaliaguine, Langmuir 21 (2005) 2051– 2057. [35] M. Thommes, B. Smarsly, M. Groenewolt, P.I. Ravikovitch, A.V. Neimark, Langmuir 22 (2006) 756–764. [36] M. Thommes, Chem. Ing. Tech. 82 (2010) 1059–1073. [37] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 57 (1985) 603–619. [38] J. Hunt, A. Ferrari, A. Lita, M. Crosswhite, B. Ashley, A.E. Stiegman, J. Phys. Chem. C 117 (2013) 26871–26880. [39] K. Doudrick, T. Nosaka, P. Herckes, P. Westerhoff, Envrion. Sci. Nano 2 (2015) 60–67. [40] C. Goel, H. Bhunia, P.K. Bajpai, J. Environ. Sci. 32 (2015) 238–248. [41] L.A. Mitchell, T.M. Tovar, M.D. LeVan, Carbon 74 (2014) 120–126. [42] X. Su, P. Tian, D. Fan, Q. Xia, Y. Yang, S. Xu, L. Zhang, Y. Zhang, D. Wang, Z. Liu, ChemSusChem 6 (2013) 911–918. [43] H. Yi, F. Li, P. Ning, X. Tang, J. Peng, Y. Li, H. Deng, Chem. Eng. J. 215–216 (2013) 635–642. [44] H. Brauer, Y.B.G. Varma, Air Pollution Control Equipment, 1st ed., SpringerVerlag, Berlin, 1981. [45] Q. Liu, L. Ning, S. Zheng, M. Tao, Y. Shi, Y. He, Sci. Rep. 3 (2013) 2916. [46] F.N. Ridha, P.A. Webley, Sep. Purif. Technol. 67 (2009) 336–343. [47] H. Yi, H. Deng, X. Tang, Q. Yu, X. Zhou, H. Liu, J. Hazard. Mater. 203–204 (2012) 111–117. [48] Y. Liu, J. Wilcox, Environ. Sci. Technol. 45 (2011) 809–814. [49] Y. Liu, J. Wilcox, Environ. Sci. Technol. 46 (2012) 1940–1947. [50] A. Samanta, A. Zhao, G.K.H. Shimizu, P. Sarkar, R. Gupta, Ind. Eng. Chem. Res. 51 (2012) 1438–1463. [51] J. Wang, L.A. Stevens, T.C. Drage, J. Wood, Chem. Eng. Sci. 68 (2012) 424–431. [52] S. Loganathan, M. Tikmani, S. Edubilli, A. Mishra, A.K. Ghoshal, Chem. Eng. J. 256 (2014) 1–8. [53] X. Wang, Q. Guo, T. Kong, Chem. Eng. J. 273 (2015) 472–480. [54] C.F. Martin, E. Stockel, R. Clowes, D.J. Adams, A.I. Cooper, J.J. Pis, F. Rubiera, C. Pevida, J. Mater. Chem. 21 (2011) 5475–5483. [55] S. Sircar, R. Mohr, C. Ristic, M.B. Rao, J. Phys. Chem. B 103 (1999) 6539–6546. [56] S. Sircar, App. Surf. Sci. 252 (2005) 647–653. [57] O. Hamdaoui, E. Naffrechoux, J. Hazard. Mater. 147 (2007) 381–394. [58] H. Tahermansouri, Z. Dehghan, F. Kiani, RSC Adv. 5 (2015) 44263–44273. [59] S. Rangabhashiyam, N. Anu, M.S. Giri Nandagopal, N.J. Selvaraju, Environ. Chem. Eng. 2 (2014) 398–414. [60] V. Krungleviciute, A.D. Migone, M. Yudasaka, S.J. Iijima, Phys. Chem. C 116 (2012) 306–310. [61] M. Bienfait, P. Zeppenfeld, N. Dupont-Pavlovsky, M. Muris, M.R. Johnson, T. Wilson, M. DePies, O.E. Vilches, Phys. Rev. B 70 (2004) 035410. [62] A. Wahby, J.M. Ramos-Fernandez, M. Martinez-Escandell, A. SepulvedaEscribano, J. Silvestre-Albero, F. Rodriguez-Reinoso, ChemSusChem 3 (2010) 974–981. [63] C. Ducrot-Boisgoniter, J. Parmentier, A. Faour, J. Patarin, G.D. Pringruber, Energy Fuel 24 (2010) 3595–3602. [64] Z. Zhang, K. Wang, J.D. Atkinson, X. Yan, X. Li, M.J. Rood, Z. Yan, J. Hazard. Mater. 229–230 (2012) 183–191. [65] M. Sevilla, A.B. Fuertes, J. Colloid Interface Sci. 366 (2012) 147–154. [66] J.D. Carruthers, M.A. Petruska, E.A. Sturm, S.M. Wilson, Microporous Mesoporous Mater. 154 (2012) 62–67. [67] H. Yi, F. Li, P. Ning, X. Tang, J. Peng, Y. Li, H. Deng, Chem. Eng. J. 215–216 (2013) 635–642. [68] S. Deng, H. Wei, T. Chen, B. Wang, J. Huang, G. Yu, Chem. Eng. J. 253 (2014) 46– 54. [69] M.G. Plaza, S. Garcia, F. Rubiera, J.J. Pis, C. Pevida, Chem. Eng. J. 163 (2010) 41–47.