Adsorption of aromatic organic contaminants by graphene nanosheets: Comparison with carbon nanotubes and activated carbon

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Adsorption of aromatic organic contaminants by graphene nanosheets: Comparison with carbon nanotubes and activated carbon Onur Guven Apul, Qiliang Wang, Yang Zhou, Tanju Karanfil* Department of Environmental Engineering and Earth Sciences, Clemson University, 342 Computer Court, Anderson, SC 29625, United States

article info

abstract

Article history:

Adsorption of two synthetic organic compounds (SOCs; phenanthrene and biphenyl) by two

Received 11 October 2012

pristine graphene nanosheets (GNS) and one graphene oxide (GO) was examined and

Received in revised form

compared with those of a coal base activated carbon (HD4000), a single-walled carbon

14 December 2012

nanotube (SWCNT), and a multi-walled carbon nanotube (MWCNT) in distilled and deion-

Accepted 18 December 2012

ized water and in the presence of natural organic matter (NOM). Graphenes exhibited

Available online 28 December 2012

comparable or better adsorption capacities than carbon nanotubes (CNTs) and granular activated carbon (GAC) in the presence of NOM. The presence of NOM reduced the SOC

Keywords:

uptake of all adsorbents. However, the impact of NOM on the SOC adsorption was smaller on

Adsorption

graphenes than CNTs and activated carbons. Furthermore, the SOC with its flexible

Graphene

molecular structure was less impacted from NOM preloading than the SOC with planar and

Carbon nanotubes

rigid molecular structure. The results indicated that graphenes can serve as alternative

Activated carbon

adsorbents for removing SOCs from water. However, they will also, if released to environ-

Natural organic matter

ment, adsorb organic contaminants influencing their fate and impact in the environment. ª 2012 Elsevier Ltd. All rights reserved.

Synthetic organic contaminants

1.

Introduction

Graphene nanosheets (GNS) are two-dimensional, planar sheets of sp2 hybridized carbon atoms packed in a hexagonal honeycomb lattice, and constitute basic building block of fullerenes, carbon nanotubes (CNTs), and graphite (Novoselov et al., 2004). Graphene has a wide range of potential applications due to its unique structure and outstanding mechanical, optical, and electronic properties (Geim and Novoselov, 2007; Geim, 2009). Commercial production and industrial scale application of GNS are expected to grow exponentially over next decades (Geim and Novoselov, 2007; Li and Kraner, 2008). GNS are hydrophobic nanomaterials, and possess a large surface area (Stoller et al., 2008). Thus, they are expected to serve as good adsorbents for organic compounds. GNS can be

modified by covalently bonding oxygen containing functional groups to obtain graphene oxides (GO). The oxygen containing functional groups decrease the surface hydrophobicity while increasing the dispersion of GO in water and also affecting its adsorption characteristics. To date, there are only a small number of peer-reviewed articles in literature on the adsorption of synthetic organic contaminants (SOCs) by pristine and functionalized GNS (Zhao et al., 2011; Ramesha et al., 2011; Wu et al., 2011; Gao et al., 2012). Considering the expected rapid growth of graphene production, understanding adsorption of SOCs by graphenes has multiple important implications including assessing (i) the feasibility of using graphenes as adsorbents in engineered treatment systems, (ii) the fate and transport of SOCs with graphene materials in the environment, and (iii)

* Corresponding author. Tel.: þ1 864 656 1005; fax: þ1 864 656 0672. E-mail address: [email protected] (T. Karanfil). 0043-1354/$ e see front matter ª 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2012.12.031

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

the potential toxicological effects of graphenes on plants, animals, and the environment when the graphene materials are released and adsorb organic contaminants. In natural waters, GNS are likely to interact with natural organic matter (NOM), which is ubiquitous in surface and groundwaters. NOM results from external (e.g., degradation of terrestrial biomass, plants, animal residues) and/or internal (e.g., excretion or decay products of photosynthetic organisms) sources to water bodies (Sparks, 1995). NOM carries a net negative charge in fresh waters, and it is a heterogeneous mixture of humic substances, hydrophilic acids, proteins, lipids, carbohydrates, carboxylic acids, amino acids, and hydrocarbons. The presence of NOM may have two opposite effects on SOC adsorption by GNS: an increase in adsorption due to better dispersion of adsorbent in the presence of NOM and/or a decrease in adsorption due to NOM competition. Previously, the impact of NOM on SOC adsorption by granular activated carbons (GACs) and CNTs in aqueous solutions has been widely investigated (Carter et al., 1995; Knappe et al., 1999; Summers et al., 1989; Zhang et al., 2011). However, to the best of our knowledge, no study has examined the effect of NOM on SOCs adsorption by GNS and GO. The main objectives of this study were to (i) examine the adsorption of two selected aromatic SOCs by GNSs and GO in distilled and deionized water (DDW) and in the presence of NOM, and (ii) compare the adsorption behavior of graphene materials with those of other carbonaceous adsorbents, a single-walled CNT (SWCNT), a multi-walled CNT (MWCNT) and a GAC.

2.

Materials and methods

2.1.

Materials

The adsorbents used in this study included two pristine GNS (GNS-A, Angstron Materials Inc. and GNS-B, Graphene Laboratories Inc.), one GO (Graphene Laboratories Inc.), a SWCNT (Chengdu Organic Chemicals Co., Ltd.), a MWCNT (Nanostructured & Amorphous Materials Inc.) and a coal based GAC (HD 4000, Norit Inc.). Graphenes and CNTs were used as received from the manufacturers, while the GAC was ground to 150e180 mm size prior to use. Phenanthrene (PNT, 99.5%) and biphenyl (BP, 99%) were the two adsorbates selected for this study, and they were obtained from Fluka/SigmaeAldrich Chemical Co. Three dimensional molecular configurations and selected properties of PNT and BP are provided in Table S1 and Figure S1 in supporting information, respectively. Two different NOM solutions were used in the experiments. The first one was an NOM isolate that was collected from the influent of a drinking water treatment plant in South Carolina using reverse osmosis and followed by resin fractionation, as described elsewhere (Song et al., 2009). The second was a water sample that was collected from a local reservoir (Beaver Dam, SC). The reservoir water was filtered using a pre-washed 0.45 mm membrane filter immediately after collection and stored in dark at a refrigerator (w4  C) until the experiments. The isolate allowed conducting the experiments with NOM alone in the absence of background inorganic constituents in natural waters. The reservoir water

1649

sample was used to confirm the findings with the NOM isolate experiments. The characteristics of background solutions are provided in Table S2.

2.2.

Characterization of adsorbents

Several techniques have been used for the characterization of adsorbents. The chemical composition (C, H, N, S and O) was determined using a Flash Elemental Analyzer 1112 series (Thermo Electron Corporation). The structural characteristics were examined using a high-resolution transmission electron microscopy (TEM, Hitachi H-9500) with electron diffraction operated at 300 kV. A drop of the dispersed adsorbent in acetone suspension was placed on a copper TEM grid. FT-IR spectroscopy measurements were conducted using a Nicolet iS10 FT-IR Spectrometer (Thermo Scientific). The zeta potential was measured with a Brookhaven Instruments 90Plus particle size analyzer equipped with ZetaPALS. The BET surface areas and pore size distributions were determined from nitrogen physisorption data at 77 K obtained with ASAP 2020 analyzer (Micromeritics Instrument Corp. U.S.), and using the DFT (density functional theory) model. The details for some of these methods have been provided in our previous publications (e.g., Dastgheib et al., 2004).

2.3.

Isotherm experiments

Constant dose bottle point batch adsorption isotherms were conducted using 255 mL amber glass bottles with Teflon lined screw caps. Two types of isotherms were conducted at room temperature (20  3  C): (i) Single-solute experiments: Bottles containing about 1 mg adsorbent were first filled with DDW and no headspace, then spiked with predetermined volumes of PNT or BP from their methanol stock solutions. For the graphene experiments, the bottles with adsorbents were initially half filled with DDW, sonicated for about 10 min, and then completely filled with DDW prior to spiking SOC. Methanol level in the stock solutions was kept below 0.1% (v/v) to minimize the co-solvent effect. The bottles were placed on a rotary tumbler for one week, which was proved to be sufficient to reach equilibrium by preliminary kinetic experiments (Zhang et al., 2009). (ii) Preloading experiments: The influence of NOM on the target SOC adsorption was examined under preloading conditions, giving an advantage to NOM adsorption prior to that of SOC, which represents the most severe NOM competition condition. For the preloading experiments, 1 mg adsorbent was prepared in isotherm bottles as described above for DDW and contacted for four days with NOM solution buffered with 1 mM NaH2PO4$H2O/ Na2HPO4.7H2O and adjusted to pH 7.0  0.1. Thereafter, predetermined volumes of PNT or BP stock solution were spiked into the bottles. The headspace-free bottles were then tumbled for an additional week. The same experimental procedure was used for both the NOM isolate and reservoir water, except 200 mg/L NaN3 was added to the bottles of the filtered reservoir water to minimize any biological activity.

1650

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

At the end of equilibrium period, 10 mL of samples were centrifuged to remove the adsorbents, and the supernatants were analyzed by high performance liquid chromatography (HPLC) equipped with UV and fluorescence detectors. Bottles without adsorbent were used as blanks to monitor the loss of adsorbates during the experiments, which were found to be negligible.

2.4.

Isotherm modeling

All adsorption isotherms obtained in this study were nonlinear. Therefore, four nonlinear isotherm models, Freundlich (FM), Langmuir (LM), LangmuireFreundlich (LFM) and Polanyi-Manes (PMM), were employed to fit the experimental data (Table S3). The residual root-mean-square error (RMSE) and coefficient of determination (r2) values indicated that FM, LFM and PMM exhibited comparable goodness of fit to the experimental data (Table S4), which was in most cases better than LM. Therefore, FM parameter values were used in further data analyses in the paper.

(Figure S3 in supporting information). On the other hand, GO had a net negative charge within the tested pH range (3e11), which was consistent with the presence of acidic functional groups. The FT-IR analysis indicated the presence of CeO group at 1200 cm1, CeC group at 1560 cm1, and C]O group at 1738 cm1 on GO (Figure S4 in supporting information). The characteristics of graphenes were compared with those of other carbonaceous adsorbents (SWCNT, MWCNT, HD4000 GAC) used in the study (Table 1). HD4000 and SWCNT had comparable surface areas with graphenes, whereas MWCNT surface area was considerably lower. The BET surface areas of the adsorbents followed the order of: HD4000 > Graphenes > SWCNT > MWCNT. The graphenes had w3e4 times higher pore volumes than CNTs, which may be attributed to their much less compact aggregate and bundle structures as compared to CNTs. The GNS and GO aggregates had similar pore size distributions that mainly consist of mesopores and macropores. On the other hand, HD4000 and SWCNTs had mainly micro- and mesopores, while MWCNT was dominated by meso- and macropores.

3.2.

3.

Results and discussions

3.1.

Adsorbent characterization

The results of BET surface area, pore volume, pore size distribution and oxygen content measurements of all adsorbents are presented in Table 1. Graphene materials had comparable BET surface areas and pore volumes, where GNSA had a slightly higher surface area than GNS-B and GO. As expected, GO had higher oxygen content (w16%) than two pristine GNS (w1%). The measured BET surface areas were smaller than the theoretically calculated surface area (2630 m2/g) for monolayer carbon structured GNS (Stoller et al., 2008). This was attributed to aggregation and bundle formation of GNS, which resulted in much lower measured surface area values (500e600 m2/g). The values obtained in this study were in agreement with those reported for other graphenes in literature (Ramesha et al., 2011; Wu et al., 2011). Although the aggregation characteristics may change when graphene is in water, the N2 gas adsorption isotherms indicated that three graphenes had similar distribution of mesoand macropores, and no micropores (<2 nm). The TEM images of graphenes are provided in Figure S2 in supporting information. The zeta potentials of two pristine GNS were similar with a net positive charge at pH values below 4

Adsorption of PNT and BP by graphenes in DDW

Adsorption isotherms of PNT and BP by graphenes in DDW are presented in Fig. 1, and the Freundlich isotherm parameters are provided on Table 2. The single point adsorption descriptors, Kd values (qe/Ce) at different equilibrium concentrations (i.e. at 0.1%, 1%, 10% and 25% of SOC aqueous solubilities) were also calculated and provided on Table S5 in supporting information. Two major observations from the isotherms are: (i) there was a higher uptake of PNT than BP by graphenes on a both mass basis and surface area normalized isotherms. This can be attributed to the higher hydrophobicity of PNT than BP, as indicated by the higher Kow value of PNT (Table S1 in supporting information). Both GNS-A and GNS-B exhibited similar uptake, and surface area normalization did not have an impact because of their similar surface areas, and (ii) GO exhibited, in general, lower BP and PNT uptake than GNSs (especially for GNS-A), which was attributed to the more polar surface of GO, and the difference was more prominent for BP. Presence of oxygen containing functional groups could create water clusters on surface, reducing the number of available adsorption sites. Cho et al. (2008) reported a 5.9% adsorption capacity reduction on CNTs for each additional percentage of surface oxides. Similar findings were also reported for adsorption of SOCs by CNTs and activated carbons (MorenoCastilla, 2004; Dabrowski et al., 2005; Zhang et al., 2009).

Table 1 e Selected properties of adsorbents. Adsorbent GNS-A GNS-B GO SWCNTa MWCNTa HD4000a

SABET (m2/g)

VT (cm3/g)

Vmicro (<2 nm) %

Vmeso (2-50 nm) %

Vmacro (>50 nm) %

Oxygen content %

624 533 576 486 164 706

2.687 2.063 2.564 0.722 0.664 0.711

0 0 0 10 2 36

59 69 52 90 51 59

41 31 48 0 47 5

1.21 0.98 15.8 1.80 0.01 3.45

a These adsorbents were available in our laboratory from a previous study (Zhang et al., 2010).

1651

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

in PNT adsorption between graphenes and GO decreased with increasing equilibrium concentrations. On the other hand, the difference in BP adsorption between graphenes and GO remained relatively constant. This was attributed to the stronger hydrophobic interactions and pep bonding between PNT molecules and the graphene surface. As a result, with increasing PNT concentrations, some of the water molecules that were clustered around the surface oxygen functionalities may be replaced with PNT molecules and/or the planar PNT molecules could be better aligned on the GO sheet surfaces. To investigate the contribution of hydrophobic interactions in adsorption, the solubility normalization (i.e., Ce/Sw) and hydrophobicity normalization using n-hexadecane as reference solvent (Brooks et al., 2012) were applied. The separation between PNT and BP isotherms was reduced; however isotherms did not collapse on a single line (data not shown). Therefore, although hydrophobic interactions were influential, it was not the sole factor controlling adsorption. It appears that the planar vs. nonplanar molecular structure of PNT and BP molecules also impacted the adsorption.

qe (mg/g)

1000

100

10

1 0

1

10

100

1000

10000

Ce (µg/L)

qe (mg/m2)

1.00

0.10 PNT / GNS-A

PNT / GNS-B PNT / GO BP / GNS-A BP / GNS-B BP / GO

0.01 0

1

10

100

1000

3.3. Comparison of PNT and BP adsorption by graphenes, CNTs, and GAC in DDW

10000

Ce (µg/L)

Freundlich isotherm parameters of PNT and BP adsorption in DDW by HD4000, SWCNT and MWCNT are listed on Table 2. The isotherm plots were provided in Figure S5 in supporting information. Adsorption capacities, represented by KF at Ce ¼ 1 mg/L, for PNT were higher than BP for all adsorbents. The same observation also was made for Kd values at different equilibrium concentrations (Table S5, supporting information). The order of PNT adsorption capacities represented by KF was HD4000 > SWCNT > GNS-A > GNS-B w GO > MWCNT. However, at lower equilibrium concentrations (e.g., Kd,0.1%) the

Fig. 1 e PNT and BP isotherms of GNS-A, GNS-B and GO in DDW on a mass basis (top) and a surface area normalized basis (bottom).

To further examine the difference between pristine graphene (GNS-A and GNS-B) and GO adsorption, Kd values at different equilibrium concentrations were analyzed (Table S5-A in supporting information). In DDW, the differences

Table 2 e Freundlich isotherm parameters of PNT and BP adsorption in DDW. Adsorbent

KF (mg/g)/ (mg/L)n

KF-SA (mg/g)/ (mg/L)n

n

r2

PNT GNS-A GNS-B GO SWCNT MWCNT HD4000

208.3 (158.4e273.9) 163.6 (127.4e210.1) 174.6 (160.7e189.7) 293.3 (229.5e374.8) 61.5 (57.4e65.9) 422.4 (281.0e635.1)

0.33 0.31 0.30 0.60 0.38 0.60

(0.25e0.44) (0.24e0.39) (0.28e0.33) (0.47e0.77) (0.35e0.40) (0.40e0.90)

0.28 0.23 0.30 0.43 0.33 0.46

(0.22e0.34) (0.17e0.28) (0.26e0.34) (0.37e0.48) (0.30e0.35) (0.38e0.54)

0.943 0.932 0.984 0.971 0.988 0.958

BP GNS-A GNS-B GO SWCNT MWCNT HD4000

102.6 (90.6e116.2) 104.7 (96.6e113.5) 59.0 (52.3e66.5) 126.1 (117.3e135.7) 29.8 (27.8e32.0) 228.8 (170.2e307.7)

0.16 0.20 0.10 0.26 0.18 0.32

(0.15e0.19) (0.18e0.21) (0.09e0.12) (0.24e0.28) (0.17e0.20) (0.24e0.43)

0.45 (0.40e0.50) 0.49(0.46e0.52) 0.46 (0.41e0.50) 0.38 (0.35e0.40) 0.58 (0.50e0.65) 0.36 (0.32e0.40)

0.981 0.994 0.980 0.994 0.993 0.915

The adsorption data for SWCNT, MWCNT, HD4000 were obtained during a recent study in our group (Zhang et al., 2010). The numbers in parentheses are 95% confidence intervals.

1652

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

order of capacities changed to GNS-B w GNSA > GO w HD4000 w SWCNT > MWCNT. In order to account for the impact of surface area, surface area normalized KF values (KFSA) were examined (Table 2). Surface area normalization suppressed the differences in adsorption capacities indicating that specific surface area plays an important role on the adsorption of PNT (Table 2 and Figure S5-C in supporting information). When the isotherms were further normalized with the oxygen content, all PNT isotherms lined up on a single trend line (Figure S5-E in supporting information). On the other hand, similar normalization procedures for BP still resulted in notable differences among the isotherms (Figure S5-F). Different than PNT, two microporous adsorbents (SWCNT and HD4000) showed highest BP uptake after surface area and oxygen content normalization. This was attributed to better filling of micropores by flexible and non-planar BP molecules. These findings indicate that the factors controlling SOC adsorption by graphenes in DDW are similar to those of activated carbon and CNTs. The overall adsorption behavior depended on the surface area, pore size distribution, and oxygen content of adsorbents as well as hydrophobicity and molecular structure of adsorbates.

3.4. Adsorption of PNT and BP by graphenes in the presence of NOM The adsorption isotherms of PNT and BP by graphenes in the natural organic matter (NOM) solution (i.e., NOM isolate) are presented in Figure S6. DDW isotherms of graphenes are shown on the same figure for comparison purposes. Freundlich isotherm parameters under NOM preloading conditions were provided on Table 3. Uptake of PNT and BP by graphenes decreased under NOM preloading conditions. This indicates the competition of NOM molecules with PNT and BP for the available adsorption sites on graphenes. The percent reductions in Kd values under NOM preloading conditions (i.e., as compared to Kd in DDW) were examined at different equilibrium concentrations, expressed as % of SOC solubility (Fig. 2). GO exhibited the least reduction under NOM preloading

conditions for both PNT and BP. The impact of NOM competition increased with decreasing equilibrium concentration. The percent reduction for Kd,0.1% of GO was 39% and 49% for PNT and BP, respectively, while the reductions for pristine graphenes, GNS-A and GNS-B, were w70% and w80% for PNT and BP. These findings suggest two possibilities: (i) The polarity of surface oxides may have better dispersed GO particles in water reducing the influence of NOM on the target compound adsorption, and/or (ii) the presence of surface oxygen group resulted in less NOM coating on the GO surface due to electrical repulsion between negatively charged NOM molecules and acidic GO surfaces. Since the selected SOCs do not contain polar functional groups, polar interactions between oxidized surfaces and SOCs are not expected to be important. The Freundlich n value indicates the heterogeneity of the surface and a higher n value is indicative of a homogeneous surface with narrow adsorption site distributions (Carter et al., 1995). The increase in the n values of GNS-A and GNS-B was 35% for PNT and 23% for BP, while the increase in n values of GO was 22% and 13%, respectively. Small increases in the n values of GO suggest the NOM coating did not significantly change the surface heterogeneity of GO. Since very low levels (1 mg in 250 mL bottles) of adsorbents were used in adsorption experiments due to strong affinity of hydrophobic PNT and BP molecules to carbon surfaces, it was not possible to quantify the extent of NOM uptake by GO and GNS during preloading experiments. The percent reduction in BP adsorption as a result of NOM preloading was higher than the reduction of PNT for all three graphenes. This was attributed to the stronger hydrophobic attraction of PNT molecules than BP molecules to graphene surfaces. In order to further investigate the impact of NOM, preloading experiments were also conducted using the reservoir water under the same preloading conditions that was used for the NOM isolate. Both NOM isolate solution and natural water preloading experiments showed similar trends (Figure S7 in supporting information). GO exhibited the least reduction in uptake among the three graphenes. The percent reduction of Kd,0.1% for GO was 60% for PNT and 74% for BP, respectively,

Table 3 e Freundlich isotherm parameters of PNT and BP adsorption on GNS-A, GNS-B, GO under NOM preloading conditions. Adsorbent

KF (mg/g)/(mg/L)n

KF-SA (mg/g)/(mg/L)n

n

r2

PNT GNS-A GNS-B GO SWCNT MWCNT HD4000

156.6 143.7 163.9 336.8 35.5 153.9

(139.2e176.3) (126.6e163.0) (147.5e182.1) (259.9e436.4) (33.6e37.5) (109.4e216.3)

0.25 (0.22e0.28) 0.27 (0.24e0.31) 0.28 (0.26e0.32) 0.69 (0.53e0.90) 0.22 (0.20e0.23) 0.22 (0.15e0.31)

0.40 (0.36e0.44) 0.39 (0.36e0.42) 0.39 (0.37e0.42) 0.54 (0.46e0.61) 0.45 (0.42e0.47) 0.89 (0.75e1.04)

0.986 0.986 0.992 0.967 0.994 0.952

BP GNS-A GNS-B GO SWCNT MWCNT HD4000

61.4 54.6 44.8 66.6 6.90 86.3

(58.2e64.7) (50.3e59.3) (41.8e47.9) (60.7e73.2) (5.35e8.90) (68.4e108.8)

0.10 (0.09e0.10) 0.10 (0.09e0.11) 0.08 (0.07e0.08) 0.14 (0.12e0.15) 0.04 (0.03e0.05) 0.12 (0.10e0.15)

0.68 (0.66e0.70) 0.65(0.61e0.68) 0.53 (0.51e0.56) 0.59 (0.54e0.63) 0.60 (0.45e0.74) 0.76 (0.65e0.88)

0.998 0.994 0.995 0.991 0.915 0.966

The numbers in parentheses are 95% confidence intervals.

1653

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

1000

PNT

PNT

80 60

qe (mg/g)

Reduction in Kd (%)

100

40

100

10

20 0

1

0

5

10

15

20

25

30

0

Equilibrium Concentration (% of SOC Solubility)

1000

BP

80 60 40

SWCNT

10 Ce ( g/L)

100

1000

1000

10000

BP SWCNT MWCNT HD4000 GNS-A GO

100 qe (mg/g)

Reduction in Kd (%)

100

1

10

MWCNT HD4000

20

1

GNS-A GO

0 0

5

10

15

20

25

0

30

Equilibrium Concentration (% of SOC Solubility)

Fig. 2 e Percent reduction of PNT (top) and BP (bottom) adsorption under NOM preloading conditions when compared to DDW adsorption. Reduction in Kd was calculated using Freundlich parameters at varying levels of equilibrium concentrations (Kd calculated at the equilibrium concentration for the corresponding % solubility of each SOC).

while it was w71% and 83%, respectively, for other two graphenes. The PNT and BP adsorption by graphenes in NOM isolate was slightly lower than their adsorption in the reservoir water. This may be due to the higher DOC (dissolved organic carbon) concentration of NOM isolate solution (3.5 mg DOC/L) than the reservoir water (2.8 mg DOC/L), and/or some difference in the physicochemical characteristics of these two NOMs.

3.5. Comparison of PNT and BP adsorption by graphenes, CNTs, and GAC under NOM preloading conditions Adsorption of PNT and BP by graphenes, SWCNT, MWCNT and HD4000 under NOM preloading was compared (Table 3, Fig. 3). The order of PNT adsorption capacities represented by KF (at Ce ¼ 1 mg/L) was SWCNT > GNS-A w GNSB w GO w HD4000 > MWCNT. At lower equilibrium concentrations (e.g., Kd,0.1%), the order of capacities changed to GNSB w GNS-A w GO w SWCNT > MWCNT w HD4000. Similar observations were also made for BP adsorption, except HD4000 did not show the same level of negative preloading effect as observed for PNT. These observations under NOM preloading conditions indicate that (i) from an engineering application perspective, graphenes exhibited comparable or

1

10

100 Ce ( g/L)

Fig. 3 e Adsorption isotherms under NOM preloading conditions PNT and BP.

better adsorption capacities than CNTs and GAC, thus they can serve as alternative adsorbents for removing organic contaminants from water, and (ii) from an environmental implication perspective, graphenes, if released to environment, will adsorb organic contaminants. Fig. 2 summarizes the percent reduction of PNT and BP uptake under NOM preloading for different equilibrium concentrations. The lowest reduction in PNT and BP uptake was observed for GO followed by pristine graphene, GNS-A. The lower impact of NOM preloading on graphenes than CNTs and GAC was attributed to less hindrance of PNT and BP molecules to adsorption sites on graphene sheets. The total pore volumes of graphenes were much higher than CNTs and HD4000 (Table 1). This suggests a much less compact bundle structure for graphenes than SWCNT or MWCNT aggregates. Fig. 2 also indicated that the reduction in uptake increased with decreasing adsorbate equilibrium concentrations. During preloading, NOM molecules preferentially occupy high energy adsorption sites. As a result displacement of adsorbed NOM molecules will be more difficult with decreasing SOC concentrations, given the fact the NOM concentration of w3.5 mg DOC/L used for preloading was 2e3 orders of magnitude higher than PNT and BP at the low concentration ranges, and/or if irreversible adsorption of some NOM components occurs. Surface heterogeneity of all adsorbents decreased under NOM preloading conditions as reflected by increasing Freundlich n values in Table S6. However, there were no clear trends observed in the increasing n values. Among all

1654

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 1 6 4 8 e1 6 5 4

adsorbents, HD4000 was influenced the most from NOM preloading. The PNT uptake by HD4000 (represented by Kd values in Table S6 in supporting information) was reduced 80e98% at different equilibrium concentrations. Similarly, BP adsorption was reduced in the range of 55e96%. This significant reduction was attributed to the microporous structure of HD4000, where NOM preloading likely resulted in a combination of pore blockage and site competition for SOC adsorption.

4.

Conclusions

Graphenes exhibited comparable or better adsorption capacities than CNTs and GAC for phenanthrene and biphenyl in the presence of NOM for the two SOCs tested in this study. Thus, they may be considered as alternative adsorbents for removing organic contaminants from water in engineering applications. If released to environment, graphenes will also adsorb organic contaminants influencing their fate and impact in the environment. This should be considered while assessing the potential environmental impacts of graphenes. The presence of NOM reduced the SOC uptake by all adsorbents. The impact of NOM on the adsorption of SOC was smaller on graphenes than CNTs and GAC. This was attributed to a much less compact bundle structure for graphenes than SWCNT or MWCNT aggregates and microporous structure of GAC. The molecular structure of the SOCs also influenced the adsorption trends. Adsorption of BP with nonplanar and flexible structure was less impacted from NOM preloading effects, especially on microporous adsorbents (HD4000, SWCNT) as compared to planar and rigid PNT molecules.

Acknowledgments This work was supported by a research grant from the National Science Foundation (CBET 0967425). However the manuscript has not been subjected to the peer and policy review of the agency and therefore does not necessarily reflect its views.

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2012.12.031.

references

Brooks, A.J., Lim, H., Kilduff, J.E., 2012. Adsorption uptake of synthetic organic chemicals by carbon nanotubes and activated carbons. Nanotechnology 23 (29), 1e13. Carter, M.C., Kilduff, J.E., Weber Jr., W.J., 1995. Site energy distribution analysis of preloaded adsorbents. Environmental Science and Technology 29 (7), 1773e1780. Cho, H., Smith, B.A., Wnuk, J.D., Fairbrother, D.H., Ball, W.P.B., 2008. Influence of surface oxides on the adsorption of

naphthalene onto multiwalled carbon nanotubes. Environmental Science and Technology 42 (8), 2899e2905. Dabrowski, A., Podkoscielny, Z., Barczak, H.M., 2005. Adsorption of phenolic compounds by activated carbon e a critical review. Chemosphere 58 (8), 1049e1070. Dastgheib, S.A., Karanfil, T., Cheng, W., 2004. Tailoring activated carbons for enhanced removal of natural organic matter from natural waters. Carbon 42 (3), 547e557. Gao, Y., Li, Y., Zhang, L., Huang, H., Hu, J., Shah, S.M., Su, X., 2012. Adsorption and removal of tetracycline antibiotics from aqueous solution by graphene oxide. Journal of Colloid and Interface Science 368 (1), 540e546. Geim, A.K., 2009. Graphene: status and prospects. Science 324 (5934), 1530e1534. Geim, A.K., Novoselov, K.S., 2007. The rise of graphene. Nature Materials 6 (3), 183e191. Knappe, D.R.U., Snoeyink, V.L., Roche, P., Prados, M.J., Bourbigot, M.M., 1999. Atrazine removal by preloaded GAC. Journal of American Chemical Society 91 (10), 97e109. Li, D., Kaner, R.B., 2008. Graphene-based materials. Science 320 (5880), 1170e1171. Moreno-Castilla, C., 2004. Adsorption of organic molecules from aqueous solutions on carbon materials. Carbon 42 (1), 83e94. Novoselov, K.S., Geim, A.K., Morozov, S.V., Jiang, D., Zhang, Y., Dubonos, S.V., Grigorieva, I.V., Firsov, A.A., 2004. Electric field effect in atomically thin carbon films. Science 306 (5696), 666e669. Ramesha, G.K., Vijaya-Kumara, A., Muralidhara, H.B., Sampath, S., 2011. Graphene and graphene oxide as effective adsorbents toward anionic and cationic dyes. Journal of Colloid and Interface Science 361 (1), 270e277. Song, H., Orr, O., Hong, Y., Karanfil, T., 2009. Isolation and fractionation of natural organic matter: evaluation of reverse osmosis performance and impact of fractionation parameters. Environmental Monitoring and Assessment 153 (1), 307e321. Sparks, D.L., 1995. Enviromental Soil Chemistry. Academic Press, San Diego, California. Stoller, M.D., Park, S.J., Zhu, Y.,W., An, J.,H., Ruoff, R.,S., 2008. Graphene-based ultracapacitors. Nano Letters 8 (10), 3498e3502. Summers, R.S., Haist, B., Koehler, J., Ritz, J., Zimmer, G., Sontheimer, H., 1989. The influence of background organic matter on GAC adsorption. Journal of American Chemical Society 81 (5), 66e74. Wu, T., Cai, X., Tan, S., Li, H., Liu, J., Yang, W., 2011. Adsorption characteristics of acrylonitrile, p-toluenesulfonic acid, 1naphthalenesulfonic acid and methyl blue on graphene in aqueous solutions. Chemical Engineering Journal 173 (1), 144e149. Zhang, S., Shao, T., Bekaroglu, S.S.K., Karanfil, T., 2009. The impacts of aggregation and surface chemistry of CNTs on the adsorption of SOCs. Environmental Science and Technology 43 (15), 5719e5725. Zhang, S., Shao, T., Kose, H.S., Karanfil, T., 2010. Adsorption of aromatic compounds by carbonaceous adsorbents: a comparative study on granular activated carbon, activated carbon fiber, and carbon nanotubes. Environmental Science and Technology 44 (16), 6377e6383. Zhang, S., Shao, T., Karanfil, T., 2011. The effects of dissolved natural organic matter on the adsorption of synthetic organic chemicals by activated carbons and carbon nanotubes. Water Research. 45 (3), 1378e1386. Zhao, G., Jiang, L., He, Y., Li, J., Dong, H., Wang, X., Hu, W., 2011. Sulfonated graphene for persistent aromatic pollutant management. Advanced Materials 23 (34), 3959e3963.