Adsorption of aromatic compounds on porous covalent triazine-based framework

Journal of Colloid and Interface Science 372 (2012) 99–107

Contents lists available at SciVerse ScienceDirect

Journal of Colloid and Interface Science www.elsevier.com/locate/jcis

Adsorption of aromatic compounds on porous covalent triazine-based framework Jingliang Liu, Enmin Zong, Heyun Fu, Shourong Zheng, Zhaoyi Xu ⇑, Dongqiang Zhu State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210093, PR China Jiangsu Key Laboratory of Vehicle Emissions Control, Nanjing University, Nanjing 210093, PR China

a r t i c l e

i n f o

Article history: Received 26 October 2011 Accepted 9 January 2012 Available online 16 January 2012 Keywords: Covalent triazine-based framework Adsorbent Adsorption Aromatic compounds removal Water treatment

a b s t r a c t Covalent triazine-based frameworks (CTFs) are an emerging class of polymers whose adsorption properties of organic chemicals are not well understood. The main objective of this work was to evaluate combined effects of the functional groups of aromatic solutes and the triazine structure of a synthesized CTF on adsorption in aqueous solutions. Adsorption of the hydroxyl-, amino-, nitro-, and sulfonate-substituted monocyclic and bicyclic aromatic compounds was generally stronger than their non-substituted, nonpolar counterparts (benzene and naphthalene). When compared with Amberlite XAD-4 resin, one of the most common and widely used polymeric adsorbents, the CTF showed much stronger adsorption toward the polar and/or ionic compounds. To explain the adsorption enhancement of CTF, several specific, non-hydrophobic mechanisms were proposed, including hydrogen bonding (hydroxyl- and amino-substituted compounds), electrostatic attraction (anionized compounds), and p–p electron-donor–acceptor (EDA) interaction (nitroaromatic compounds) with the triazine structure of CTF. The hypothesized mechanisms were further supported by the observed pH dependence of adsorption. Resulting from size exclusion, adsorption of large-size dissolved humic acids on the homogeneous, nanopored (1.2 nm in size) CTF was negligible and did not affect adsorption of aromatic solutes. Additional advantages of fast adsorption/ desorption kinetics and complete adsorption reversibility made CTF a superior adsorbent for aromatic compounds. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction Covalent triazine-based frameworks (CTFs, Fig. 1) are an emerging class of polymers first synthesized in 2008 by ionothermal trimerization of aromatic nitriles [1]. Because of the very large specific surface area, homogeneous pore structure, diverse functionality, and high chemical and thermal stability [1–5], CTFs are considered a promising candidate for energy gas storage and catalytic support materials [1,6,7]. Recently, it was reported that CTFs are effective adsorbents for the removal of ionic dyes from aqueous solutions [8,9]. Some of the conventional synthetic porous organic materials such as polymeric resins have been already widely used as adsorbents for organic contaminants in water and wastewater treatments in recent years [10]. Adsorption of organic compounds on polymeric resins may be affected by several mechanisms including hydrophobic effect, hydrogen bonding, and p–p interaction – their relative importance depends on the physicochemical properties of adsorbates and surface chemistry of adsorbents [11–13]. Azanova and Hradil [11] studied adsorption of benzene, phenol, and aniline ⇑ Corresponding author at: State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing 210093, PR China. E-mail address: [email protected] (Z. Xu). 0021-9797/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jcis.2012.01.011

on typical resin adsorbents (macroporous and hypercrosslinked styrene-divinylbenzene copolymers) and concluded that the adsorption was mainly driven by hydrophobic effect. Streat and Sweetland [12] reported that adsorption of polar pesticides such as simazine, chlorotoluron, and isoproturon to hypercrosslinked polymeric resins was controlled by both hydrophobic effect and hydrogen bonding. Investigation on the adsorption of aromatic compounds over hypercrosslinked polystyrene resins as highperformance liquid chromatography stationary phase further revealed that the retention mechanism included not only hydrophobic effect but also significant p–p interaction [13]. Despite the current wide application of polymeric resins as adsorbents for many purposes, there are several critical factors that restrict the dissemination of the technology. One of the main disadvantages is insufficient mechanical rigidity and solvent-sensitivity of the polymer matrix, which would result in swelling of the polymer matrix [13]. Furthermore, polymeric resins in general consist of irregular-shaped pores with a wide pore size distribution, from the microporous region (<2 nm), through the mesoporous region (2–50 nm), and into the macroporous region (>50 nm) [14]. Pore swelling [15] and heterogeneous pore structure [16] could possibly lead to multiphasic and/or slow adsorption/desorption kinetics and impaired size-selective adsorption. It was found in our previous study that the combination of very large specific

100

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

(a) N

N

N

N

N

N N

(b)

N

N

N

N

N

N

N

N

N

N

N

N

N N

N N

N

N

N

N N

N

N N

N

N

N

N

1.2 nm N

N

N

N N N

N N

N N N

N

N

Fig. 1. (a) The chemical structure and pore size of covalent triazine-based framework (CTF). (b) Schematic representation of CTF adopted from Ref. [1], C: gray balls; N: black balls; H atoms are omitted for clarity.

surface area and open, uniform pore structure of zeolite-templated microporous carbon makes it a superior adsorbent for aromatic compounds [17]. Similarly, as the next-generation framework polymers CTFs possess rigid, permanent, and homogeneous porosity, and therefore may open up new opportunities for efficient and selective adsorption of organic contaminants from water in many environmentally relevant applications. The main objective of this study was to systematically investigate the properties and controlling factors of aromatic compound adsorption on a porous CTF prepared by ionothermal trimerization of 1,4-dicyanobenzene. The synthesized adsorbent was characterized by a variety of experimental and spectroscopic techniques with respect to structural, porosity, and surface characteristics. Aqueous-phase batch experiments were performed to examine adsorption of eight monocyclic aromatic compounds (benzene, phenol, aniline, benzenesulfonate, nitrobenzene, 1,3-dinitrobenzene, 1,3,5-trinitrobenzene, and 4-methyl-2,6-dinitrophenol) and four bicyclic aromatic compounds (naphthalene, 2-naphthol, 1naphthylamine, and 2-naphthalenesulfonate). The adsorbate compounds are of great environmental concern and vary pronouncedly in terms of hydrophobicity, hydrogen-bonding ability, and p-electron-donor/acceptor ability. Adsorption of CTF was further compared with that of a commercial Amberlite XAD-4 resin, one of the most efficient polymeric adsorbents that are extensively used in water and wastewater treatment [18–20], environmental analysis [21–23], biotechnology [24], and food processing [25,26]. The effect of pH on adsorption of selected compounds was also evaluated to further understand the mechanisms controlling adsorptive interactions. Changing pH might significantly affect speciation reaction of the polar adsorbate molecules, as well as surface chemistry properties of CTF. Impact of presence of model dissolved humic acids on adsorption, adsorption/desorption kinetics, and adsorption reversibility of selected compounds was also assessed.

(sodium salt, Sigma), and pyrene (Aldrich). NaCl, HCl, NaOH, and ZnCl2 were purchased from Nanjing Chemical Reagents Co. Ltd. All adsorbates and other reagents were of chemical grade or higher and were used without further purification. Water solubility (SW), n-octanol–water partition coefficient (KOW), n-hexadecane–water partition coefficient (KHW), and/or acid dissociation constant (pKa) of the adsorbates are summarized in Supporting Information (SI) Table S1. Chemical structures and molecular sizes of the adsorbates are given in SI Fig. S1.

2. Materials and methods

Elemental analysis of CTF was performed using an elemental analyzer of Vario MICRO (Elementar, Germany) to determine the C, H, and N contents. X-ray diffraction (XRD) pattern of CTF was collected from a Rigaku D/max-RA powder diffraction-meter (Rigaku, Japan) using Cu Ka radiation. Transmission Fourier transform infrared (FTIR) spectrum of CTF was measured using a Nexus 870 spectrometer (Nicolet, USA). Nitrogen adsorption/desorption isotherms of CTF and XAD-4 were obtained on a Micrometrics ASAP 2020 (Micromeritics Instrument Co., USA) apparatus at 196 °C (77 K).

2.1. Adsorbates The tested adsorbates are as follows: benzene (Alfa Aesar), phenol (Sigma–Aldrich), aniline (Guoyao. Co., Ltd.), nitrobenzene (Fluka), 1,3-dinitrobenzene (Aldrich), 1,3,5-trinitrobenzene (Supelco), 4-methyl-2,6-dinitrophenol (Fluka), benzenesulfonate (sodium salt, Fluka), naphthalene (Sigma–Aldrich), 2-naphthol (Aldrich), 1-naphthalenamine (Sigma), 2-naphthalenesulfonate

2.2. Adsorbents The covalent triazine-based framework (CTF) adsorbent was prepared by ionothermal trimerization of 1,4-dicyanobenzene according to literature method [1]. Briefly, 1.06 g of anhydrous ZnCl2 stored in an anaerobic glove box (M. Braun, Germany) was mixed with 1 g of 1,4-dicyanobenzene (Aldrich) in an anaerobic glove box and the mixture was then transferred into a quartz tube. The tube was evacuated using a mechanical pump to a pressure below 102 Pa and was then sealed and heated at 400 °C for 40 h. After cooling to room temperature, the obtained material was repeatedly washed with distilled water and 1.0 M HCl to remove residual ZnCl2. The resulting black powder was filtered, followed by washing with distilled water and tetrahydrofuran and drying in vacuum at 150 °C overnight. The commercial polymeric adsorbent Amberlite XAD-4 was purchased from Rohm & Haas (USA). Prior to use, the XAD-4 resin was extracted with anhydrous ethanol in a Soxhlet apparatus for 8 h, followed by washing with water and drying at 50 °C overnight.

2.3. Characterization of adsorbents

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

2.4. Adsorption/desorption Batch adsorption experiments were performed using 40 mL glass vials equipped with polytetrafluoroethylene-lined screw caps. Vials received 20 mg of adsorbent and a full volume of background solution containing 0.02 M NaCl [27,28], followed by organic solute in a methanol carrier (2-naphthalenesulfonate in distilled water) that was kept below 0.1% (v/v) to minimize possible cosolvent effects on adsorption. The pH of aqueous solution was adjusted with 0.01 M HCl and 0.01 M NaOH. The samples were covered with aluminum foil from light and were mixed end-overend at room temperature for 24 h. This period of time was sufficient enough to reach apparent adsorption equilibrium (no further uptake) based on kinetic measurements. After centrifugation, the solute in an aliquot was analyzed directly by high-performance liquid chromatography (HPLC) with an ultraviolet (UV) detector using a 4.6  150 mm HC-C18 column (Agilent, USA). The initial concentration ranges of adsorbates and the HPLC analytical conditions were listed in SI Table S2. To account for solute loss from processes other than adsorbent sorption (sorption to glassware and septum and/or volatilization), calibration curves were obtained from control samples receiving the same treatment as the adsorption samples but without the adsorbent. Calibration curves included at least 12 standards over the examined concentration range. Based on the obtained calibration curves, the adsorbed mass of solute was calculated by subtracting mass in the aqueous phase from mass added. Separate sets of experiments were conducted to test the effects of dissolved humic acids on single-point adsorption on CTF. The bulk solution of humic acids was prepared by dissolving 100 mg of humic acids (Aldrich) in 5 mL of 0.1 M NaOH, followed by dilution with distilled water to reach an apparent concentration of 100 mg/L. The solution of humic acids was adjusted to pH 6.0 with 0.1 M HCl, and then filtered through a 0.45 lm cellulose acetate membrane (Shanghai Bandaoshiye Co. Ltd.). The obtained solution of humic acids was diluted to concentrations of 10, 20, 30, and 40 mg/L in background solutions containing 0.02 M NaCl. Additional single batch experiments were conducted to assess the adsorption/desorption kinetics with repeated sampling (<0.5 mL each time) using 50 mL vials for nitrobenzene on CTF (spiked at 0.196 mmol/L) and XAD-4 (spiked at 0.159 mmol/L). Calibration curves were obtained separately from controls receiving repeated sampling in the same time intervals. To test adsorption reversibility on CTF, desorption experiments were performed using a single-step, centrifuge-withdraw-refill method for nitrobenzene. About 90% supernatant was replaced by fresh background solution at the end of adsorption experiments, and then the samples were mixed for 24 h (long enough to reach desorption equilibrium based on kinetic measurements). Adsorption ratios for desorption points were within the range of 20–80% to assure measurement data quality. Triplicate samples were prepared for each data point in pH effect experiments, and duplicate samples for all others. The equilibrium pH of all samples with the exception of the pH experiments was 6.3 ± 0.2.

3. Results and discussion 3.1. Characterization of adsorbents The information of elemental analysis, specific surface area, and pore volume of CTF and XAD-4 is listed in Table 1. The transmission FTIR spectrum of CTF is shown in Fig. 2. The strong bands around 1507 and 1352 cm1 are characteristic of the stretching and breathing vibrations of the triazine structure, respectively. The weak band around 2228 cm1 is assigned to the stretching

101

vibration of C„N [5]. The XRD pattern of CTF is presented in Fig. 3. The peak with 2h at 7.2° is indexed to the (1 0 0) diffraction, reflecting the crystalline triazine-based organic framework with hexagonal pores [1]. The broad peak at 26.2° is assigned to the (0 0 1) diffraction, characteristic of the interlayer aromatic sheets. 3.2. Adsorption isotherms Adsorption isotherms of different adsorbates on CTF and XAD-4 are presented in Fig. 4. The adsorption data are fitted to the Freundlich model ðq ¼ K F C nW Þ by nonlinear regression weighed by 1/q, where q (mmol/kg) and CW (mmol/L) are the adsorbed concentration and aqueous concentration, respectively, at adsorption equilibrium; KF (mmol1n Ln/kg) is the Freundlich affinity coefficient; n (unitless) is the Freundlich linearity index. The fitting parameters are summarized in Table 2. The adsorption parameters of 2-naphthalenesulfonate on CTF are not listed in Table 2 due to the Langmuir-type isotherm of this system. Note adsorption of benzenesulfonate on XAD-4 was negligible under the examined experimental conditions. For adsorption on CTF, with the exception of 2-naphthalenesulfonate the Freundlich model fits all adsorption data reasonably. The degrees of adsorption linearity vary significantly with adsorbates, with the n values from 0.22 (2-naphthol) to 0.90 (naphthalene). Additionally, the polar solutes (in particular 2-naphthol and 4methyl-2,6-dinitrophenol) display much higher adsorption nonlinearity (as reflected by smaller n values) than the nonpolar solutes (benzene and naphthalene) within the tested concentration ranges, implying more heterogeneous adsorption sites on CTF for the polar and ionic solutes. A very interesting observation made from Fig. 4a and c is that the adsorption affinity to CTF correlates poorly with solute hydrophobicity. For example, when compared with benzene, benzenesulfonate has much lower hydrophobicity due to its ionic nature, but shows even slightly stronger adsorption. The same trend with even larger difference in adsorption is displayed between benzene and 4-methyl-2,6-dinitrophenol (nearly completely anionized above pH 6). Furthermore, despite the much lower hydrophobicity, the polar bicyclic compounds (2-naphthol, 1-naphthalenamine, and 2-naphthalenesulfonate) all show stronger adsorption than naphthalene, especially at low solute concentrations. The abovementioned enhanced adsorption implies that strong specific interactions exist between the polar/ionic compounds and CTF. Electrostatic interaction is likely responsible for the extraordinarily strong adsorption of the anionic solutes (benzenesulfonate, 2-naphthalenesulfonate, and 4-methyl-2,6-dinitrophenol) on CTF. Acid-base titration results (presented in SI Fig. S2) demonstrate that the triazine structure of CTF has a point of zero charge (pzc) around pH 7.2. Correspondingly, at the tested pH (6.3) a significant portion of the triazine components of CTF is protonated and positively charged and thus capable of inducing strong electrostatic interaction with the negatively charged adsorbate molecules. Similarly, it was reported that CTF materials can strongly adsorb anionic dyes via electrostatic interaction at neutral or slightly acidic pH conditions [8]. Notably, the benzene rings adjacent to the triazine structure would make the cavity microenvironment of CTF very hydrophobic, therefore alleviating the ‘‘desolvation penalty’’ associated with breaking of the hydration shells around the charged species (protonated N atoms and anionized functional groups of solutes) when they form electrostatic complexes. In fact, the energy compensation for the ‘‘desolvation penalty’’ by the hydrophobic microenvironment surrounding cationized species has been recognized as an important factor to thermodynamically favor cation–p interaction with aromatic structures in biological systems [31,32]. However, adsorption of the negatively charged adsorbate on CTF at high concentrations could be suppressed by intermolec-

102

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

Table 1 Elemental composition, specific surface area, and pore volume parameters for the adsorbents. Adsorbent

Elemental composition

Covalent triazine-based framework Amberlite XAD-4 a b c d e

C%

H%

N%

67.85 NDe

3.73

17.58

Specific surface areaa (m2/g)

Vmicb (cm3/g)

Vmesc (cm3/g)

Vtd (cm3/g)

782.4 834.1

0.35 0.41

0.05 0.89

0.40 1.30

Determined by N2 adsorption using the Brunauer–Emmett–Teller (BET) method [29]. Micropore volume, calculated using the Dubinin–Astakhov method [30]. Mesopore volume, calculated by Vt  Vmic. Total pore volume, determined at P/P0 = 0.976. Not determined.

Transmittance (%)

100 80 60 40 20 0 4000

3000

2000

1000

0

-1

Wavenumber (cm )

Intensity (a.u.)

Fig. 2. Transmission Fourier transform infrared (FTIR) spectrum of covalent triazine-based framework.

0

10

20

30

40

50

2 (degrees) Fig. 3. X-ray diffraction (XRD) pattern of covalent triazine-based framework.

ular electrostatic repulsion when the limited pore space is forced to host more than one adsorbate molecule (particularly of a large size). These arguments can be applied to explain why adsorption of 2-naphthalenesulfonate (the bulkiest bicyclic solute) reaches saturation at a very low concentration (approximately 0.02 mM), as characterized by the Langmuir-type isotherm. To better evaluate the specific adsorptive interaction of the mono- and bicyclic hydroxyl- and amino-substituted compounds, their adsorption data are compared with those of benzene and naphthalene, respectively, upon normalization of solute hydrophobicity using the KHW values. n-Hexadecane is dictated by inert nonpolar methylene structures and thus serves as an ideal reference solvent for justifying solute hydrophobicity [33,34]. The hydrophobicity-normalized adsorption (presented in Fig. 5) follows an order of benzene  aniline < phenol for the monocyclic compounds, and

an order of naphthalene  1-naphthalenamine < 2-naphthol for the bicyclic compounds. At the tested pH (6.3) the polar functional groups of the examined hydroxyl- and amino-substituted compounds are protonated, and expectedly form strong hydrogen bonds with the triazine structure (either protonated or deprotonated) of CTF. The triazine structure is known to have a strong ability to form hydrogen bonds. The hydrogen-bonding energy between 1,3,5-triazine and water estimated by the density functional calculation varied from 13.38 to 22.60 kJ/mol dependent on the number of coordination water molecules [35]. In a similar fashion to the electrostatic adsorptive interaction, the hydrogen bonding of organic solutes with the triazine structure of CTF would also be facilitated by the hydrophobic cavity microenvironment, which effectively reduces the competitive hydrogen bonding by water. The more enhanced adsorption of the hydroxyl-substituted compounds relative to the amino-substituted compounds is probably due to the stronger H-bonding ability of the hydroxyl group. It is further shown from Fig. 4a that the three nitrobenzene compounds exhibit much stronger adsorption than benzene. Moreover, the adsorption enhancement increases with increasing nitro group number. Given the decreased hydrophobicity by the substituted nitro group(s) on the benzene ring (justified by KOW and KHW values), the observations suggest that special mechanism(s) exists in adsorption in addition to hydrophobic effect. Mainly two adsorption-enhancement mechanisms have been identified previously for nitroaromatic compounds: (1) cation–polar interaction between the nitro group and weakly hydrated cations exchanged on the mineral surface [36–40], and (2) p–p electron-donor–acceptor (EDA) interaction between the p-electron deficient benzene ring (due to the strong electron-withdrawing effect of the nitro groups) and polarized, p-electron rich graphitized carbon surface [27,28,41–44]. Both mechanisms seem plausible considering the surface chemistry properties of the CTF cavity. First, the protonated triazine structure may interact with the nitro groups that have highly delocalized electrons (oxygen is partially negatively charged) via cation-polar interaction. Additionally, the cavity surface of CTF is composed of a mixture of benzene and triazine rings to form a large conjugated chain structure that is p-electron rich and thus capable of inducing p–p EDA interaction with nitroaromatic compounds (p-electron acceptors). The relative importance of these two adsorption-enhancement mechanisms is further discussed based on the results of pH-mediated adsorption (see below). By adding one additional benzene ring to the monocyclic compound, the adsorption affinity on CTF of the respective bicyclic compound is remarkably increased (Fig. 4c). For example, the adsorbent-to-solution distribution coefficient (Kd) is in the order of 102–103 L/kg for benzene and phenol, but in the order of 105–107 L/kg for naphthalene and 2-naphthol. As discussed above, solute hydrophobicity is only a minor factor affecting adsorption affinity on CTF. The more enhanced adsorption of the bicyclic compounds than the monocyclic compounds is likely caused by a

103

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

10 3

q (mmol/kg)

NB DNB TNB MDNP BS BEN PHOL AN

10 1

10

-4

-3

-2

10 10 C w (mmol/L)

10

-1

10

10 2

2-NAOL 1-NALA NAPH 2-NS

10 2

-3

-2

10 10 C w (mmol/L)

10

-3

10

-2

10

-1

10

0

C w (mmol/kg)

10 3

-4

BEN PHOL AN NB DNB TNB MDNP

10 1

(d) 10

10

3

10 0 -4 10

0

10 4

10 1 -5 10

q (mmol/L)

10 2

10 0 -5 10

(c)

(b) 10

10

-1

10

0

q (mmo/kg)

q (mmo/kg)

(a)

3

10 2

NAPH 2-NAOL 1-NALA 2-NS

10 1

10 0 10 -5

10 -4

10 -3 10 -2 C w (mmol/L)

10 -1

10 0

Fig. 4. Adsorption isotherms plotted as solid-phase concentration (q) vs. aqueous-phase concentration (CW) at equilibrium for different adsorbates on covalent triazine-based framework (CTF) (a and c) and Amberlite XAD-4 (b and d). (a and b) Monocyclic solutes, including benzene (BEN), phenol (PHOL), aniline (AN), benzenesulfonate (BS), nitrobenzene (NB), 1,3-dinitrobenzene (DNB), 1,3,5-trinitrobenzene (TNB), and 4-ethyl-2,6-dinitrophenol (MDNP). (c and d) Bicyclic solutes, including naphthalene (NAPH), 2-naphthol (2-NAOL), 1-naphthalenamine (1-NALA), and 2-naphthalenesulfonate (2-NS). Note: sorption of benzenesulfonate to XAD-4 is negligible and is not presented.

Table 2 Freundlich model parameters, KF and n ± Standard deviation, R2 and Kd value for adsorption on covalent triazine-based framework (CTF) and XAD-4 resin (XAD-4). Compound/adsorbent

KF (mmol1n Ln/kg)

n

R2

Kd (L/kg)

Benzene/CTF Phenol/CTF Aniline/CTF Benzenesulfonate/CTF Nitrobenzene/CTF 1,3-Dinitrobenzene/CTF 1,3,5-Trinitrobenzene 4-Methyl-2,6-dinitrophenol/CTF Naphthalene/CTF 2-Naphthol/CTF 1-Naphthylamine/CTF Benzene/XAD-4 Phenol/XAD-4 Aniline/XAD-4 Nitrobenzene/XAD-4 1,3-Dinitrobenzene/XAD-4 1,3,5-Trinitrobenzene/XAD-4 4-Methyl-2,6-dinitrophenol/XAD-4 Naphthalene/XAD-4 2-Naphthol/XAD-4 1-Naphthylamine/XAD-4 2-Naphthalenesulfonate/XAD-4

910 ± 40 720 ± 20 390 ± 10 410 ± 30 2500 ± 100 4000 ± 300 10,000 ± 3000 810 ± 20 60,000 ± 10,000 2400 ± 100 3100 ± 300 1600 ± 200 170 ± 7 300 ± 10 1800 ± 70 2000 ± 50 1100 ± 50 3900 ± 300 16,000 ± 3000 1400 ± 40 940 ± 80 30 ± 3

0.65 ± 0.02 0.66 ± 0.02 0.60 ± 0.01 0.43 ± 0.03 0.58 ± 0.02 0.58 ± 0.02 0.64 ± 0.04 0.24 ± 0.01 0.90 ± 0.04 0.22 ± 0.01 0.38 ± 0.02 0.58 ± 0.04 0.75 ± 0.03 0.75 ± 0.02 0.62 ± 0.02 0.64 ± 0.01 0.67 ± 0.01 0.94 ± 0.03 0.59 ± 0.03 0.52 ± 0.02 0.40 ± 0.03 0.43 ± 0.03

0.989 0.997 0.998 0.975 0.992 0.993 0.965 0.994 0.978 0.987 0.983 0.961 0.985 0.995 0.995 0.998 0.997 0.986 0.989 0.996 0.980 0.971

103 102–103 102–103 103–104 102–103 103–104 103–104 103–105 105 105–107 104–105 103–104 102 102–103 103–104 103–104 103–104 103–104 105 103 103–104 101–102

combination of two mechanisms. First, compared with the monocyclic compounds, the bicyclic compounds have higher p-electron densities, and therefore have much greater potential for p–p stacking/coupling with the aromatic structures on the cavity surface of CTF. Second, the bicyclic compounds have molecular sizes closer to the cavity size of CTF, leading to stronger additive adsorptive

interaction (micropore-filling effect). The cavity size of CTF is approximately 1.2 nm (Fig. 1); in comparison, the molecular size along the longest dimension varies from 0.57 (benzene) to 0.89 nm (1,3,5-trinitrobenzene) for the monocyclic compounds, while from 0.75 (naphthalene) to 1.24 nm (2-naphthalenesulfonate) for the bicyclic compounds (see SI Fig. S1). It is thus

104

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

(a) 10

3

(b)

10 4

10 2

q (mmol/kg)

q (mmol/kg)

NAPH 2-NAOL 1-NALA

10 1

10 0 -4 10

PHOL AN BEN

10

-3

10

-2

-1

10 10 CH (mmol/L)

0

10

1

10

2

10 3

10 2

10 1 -5 10

10

-4

10

-3

-2

-1

10 10 10 CH (mmol/L)

0

10

1

10

2

Fig. 5. Hydrophobicity-normalized adsorption isotherms plotted as solid-phase concentration (q) vs. concentration in n-hexadecane (CH, CH = CW  KHW; CW = aqueous-phase concentration; KHW is the n-hexadecane–water partition coefficient shown in SI Table S1) at equilibrium for different adsorbates on covalent triazine-based framework. (a) Benzene (BEN), phenol (PHOL), and aniline (AN). (b) Naphthalene (NAPH), 2-naphthol (2-NAOL), and 1-naphthalenamine (1-NALA).

confirmed that prominent micropore-filling effect is generally induced in adsorption of the bicyclic compounds. In previous studies, the micropore-filling mechanism has been explored to explain the enhanced adsorption of low-size adsorbates (benzene, toluene, and naphthalene) to highly microporous carbonaceous adsorbents [33,37]. More recently, De Moor et al. [45] observed increased adsorption of C2AC8 n-alkanes on microporous zeolites with increasing size of adsorbate. Additionally, for a given adsorbate more marked increase of adsorption enthalpy was shown on the zeolite having smaller pore channel, consistent with the micropore-filling mechanism. A strong experimental evidence for the micropore-filling effect of bicyclic compounds is obtained by comparing adsorption affinity between naphthalene and pyrene. The Kd of pyrene on CTF measured from single-point adsorption is (1.7 ± 0.3)  104 L/kg (standard deviation based on five replicates) at an equilibrium concentration of 1.3  104 mmol/L. Given the same equilibrium concentration, the Kd of naphthalene calculated from the Freundlich model is 1.5  105 L/kg. Because these two solutes differ in molecular size (see SI Fig. S1), the reversed adsorption trend against hydrophobicity (see SW, KOW, and KHW values in SI Table S1) suggests size-exclusion effect for pyrene, but micropore-filling effect for naphthalene. In addition, the specific polar and/or ionic adsorptive interactions with the triazine components in CTF are expected to be strengthened by the micropore-filling effect for two reasons: first, occupation of the cavities by adsorbate molecules, in particular the large-sized bicyclic compounds, repels or excludes water molecules from the cavities and thus lowers the ‘‘desolvation penalty’’ of the specific interaction; additionally, filling of adsorbate molecules in match-sized cavities helps them to be reoriented and thus more susceptible to directed specific forces such as hydrogen bonding. Fig. 4 further compares the adsorption affinity between the synthesized CTF and commercial Amberlite XAD-4 resin. With the exception of benzene and naphthalene, for all tested solutes adsorption is markedly stronger on CTF than on XAD-4. Note that CTF and XAD-4 possess comparable surface areas (782 and 834 m2/g, respectively). The observed trends of adsorption can be reasonably explained by the structural disparities of adsorbents. Styrenic XAD-4 resin is made of nonpolar, hydrophobic structural units and thus can strongly adsorb hydrophobic organic compounds (such as benzene and naphthalene) via hydrophobic effect. In contrast, XAD-4 shows much lower adsorption of the polar and/ or ionic solutes (4-methyl-2,6-dinitrophenol, 2-naphthol, benzenesulfonate, and 2-naphthalenesulfonate) having low hydrophobic-

ity. On the other hand, the more enhanced adsorption of the polar and/or ionic solutes on CTF clearly verifies the predominant role played by specific adsorptive interaction. 3.3. Effect of pH on adsorption Fig. 6 displays the pH effect on CTF adsorption of selected monocyclic and bicyclic adsorbate compounds separately. Over the examined pH range of 3–10, adsorption of the two nonpolar compounds (benzene and naphthalene) keeps nearly constant, in line with the nonspecific, hydrophobic adsorption mechanism. The most significant pH effect (more than two orders of magnitude difference in Kd) is shown for the three anionizable compounds, 4methyl-2,6-dinitrophenol, benzenesulfonate, and 2-naphthalenesulfonate. As long as the adsorbate molecules are anionized, increasing the pH would impair the electrostatic adsorption force as the cationized triazine components in CTF are progressively deprotonated and become uncharged. Additionally, due to the facilitated anionization, adsorption of 4-methyl-2,6-dinitrophenol increases with pH up to the pKa value (4.06). For adsorption of 2-naphthol, the pH-dependency curve is bell-shaped with the peak at pH near 8.8. At the lower pH range (3–8.8), 2-naphthol (pKa = 9.51) is dominated by the neutral species; the increased adsorption implies that H-bonding interaction of the neutral species is stronger with the deprotonated neutral triazine structure than with its protonated counterpart. Likewise, the decreased adsorption at the higher pH range (8.8–9.9) demonstrates that the hydrogen-bonding interaction with the deprotonated triazine structure is mitigated when 2-naphthol dissociates to the anionic form. The pH effect on adsorption of 1-naphthalenamine can be jointly explained by the electrostatic mechanism and the hydrogen-bonding mechanism. The adsorption first increases with pH up to the solute’s pKa value (3.92) – deprotonation of the protonated solute relieves the electrostatic repulsive interaction with the cationized triazine components. The adsorption then keeps nearly constant over the pH range from approximately 4 to 6.4, but slightly increases when the pH is further increased. Similar to 2-naphthol, the increased adsorption of neutral 1-naphthalenamine at high pH levels can be attributed to the favored H-bonding interaction with the deprotonated triazine structure. For both 1,3dinitrobenzene and 1,3,5-trinitrobenzene, the adsorption increases slightly but consistently over the whole pH range examined. As the degree of deprotonation of the cationized triazine components increases with pH, N-heterocyclic aromatic structure in CTF

105

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107 6

(b) 10

K d (L/kg)

10 5

DNB MDNP BEN BS TNB

10 4 10 3

K d (L/kg)

(a) 10

6

NAPH 2-NAOL 2-NS 1-NALA

10 5

10 4

10 2 10 1

2

4

6

8

10

10 3

12

2

4

6

pH

8

10

12

pH

Fig. 6. Changes of adsorbent-to-solution distribution coefficients (Kd) with pH at single-point concentrations for adsorption of selected adsorbates on CTF. (a) Monocyclic solutes, including benzene (BEN), benzenesulfonate (BS), 1,3-dinitrobenzene (DNB), 1,3,5-trinitrobenzene (TNB), and 4-methyl-2,6-dinitrophenol (MDNP). (b) Bicyclic solutes, including naphthalene (NAPH), 2-naphthol (2-NAOL), 1-naphthalenamine (1-NALA), and 2-naphthalenesulfonate (2-NS). Error bars, in most cases smaller than the symbols, represent standard deviations calculated from triplicates. Lines are for visual clarity only.

Kd (L/kg)

10

5

10

4

10

3

10

2

BEN PHOL AN BS NB DNB TNB MDNP

0

10

20

30

40

50

C HA (mg/L)

K d (L/kg)

(b)

(a)

106 NAPH 2-NAOL MDNP 2-NS

105

104

103

0

10

20

30

40

50

C HA (mg/L)

Fig. 7. Effect of dissolved humic acids on single-point adsorption of different compounds to CTF plotted as adsorbent-to-solution distribution coefficients (Kd) vs. initial apparent concentrations (CHA) of humic acids. (a) Monocyclic solutes, including benzene (BEN), sodium-benzenesulfonate (BS), 1,3-dinitrobenzene (DNB), 1,3,5trinitrobenzene (TNB), and 4-methyl-2,6-dinitrophenol (MDNP). (b) Bicyclic solutes, including naphthalene (NAPH), 2-naphthol (2-NAOL), 1-naphthylamine (1-NALA), and 2naphthalenesulfonate (2-NS).

gradually changes itself from a strong p-acceptor to a weak p-donor [46,47]. Accordingly, the p–p EDA interaction is favored between the nitrobenzenes (p-electron acceptors) and the conjugated benzene–triazine chain structures (p-electron-donors) on the cavity surface of CTF. In agreement with this hypothesis, our previous study [48] showed that the p–p EDA interaction between nitrobenzenes and multiple carboxyl-substituted aromatic rings was facilitated by deprotonation of the carboxyl groups due to the promoted electron-donor ability (ACOOH is a weak electron-donor and acceptor, while ACOO is a strong electron-donor but not an acceptor). 3.4. Effect of dissolved humic acids (DHA) on adsorption The effect of DHA on CTF adsorption of selected compounds is shown in Fig. 7. In the presence of DHA (up to 40 mg/L), adsorption of all tested solutes stays nearly constant. The observation can be attributed to the size-exclusion effect by which the large-size humic molecules would be blocked from entering small nanopores (1.2 nm in size) of CTF to access the adsorption sites. To be consistent with this hypothesis, adsorption of DHA on CTF was found to be minimal under the tested experimental conditions (results

presented in SI Fig. S3). The negligible effect of DHA on adsorption would provide practical benefits to the use of CTF as effective adsorbents in real waters. 3.5. Adsorption/desorption kinetics and reversibility Fig. 8 compares the adsorption/desorption kinetics of nitrobenzene on CTF and XAD-4. The adsorption and desorption reach equilibrium within 100 min on CTF, but within 600 min on XAD-4. Unlike the irregular-shaped, flexible pore structure of XAD-4 (average pore size is 5.96 nm; pore size distribution ranges from 2 to 30 nm, measured by Micrometrics ASAP 2020), CTF consists of homogeneous, regular-shaped pores with a fixed pore size (1.2 nm). Compared with XAD-4, pore diffusion of solute molecules within CTF is expected to be more rapid, therefore leading to faster adsorption and desorption kinetics. The adsorption and desorption data of nitrobenzene on CTF are presented in Fig. 9. The adsorption of nitrobenzene on CTF seems to be fully reversible as nearly all desorption points fall on the line of adsorption isotherm. To quantitatively estimate the adsorption reversibility, a thermodynamic index of irreversibility (TII) is calculated according to literature methods [49]. The

106

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107

80

80

Adsorption ratio (%)

(b) 100

Adsorption ratio (%)

(a) 100

60 40 20 0

sorption desorption

0

100

200

300

400

500

600

700

60 40 20 0

sorption desorption

0

200

400

T (min)

600

800

1000

1200

T (min)

Fig. 8. Adsorption/desorption kinetics plotted as adsorption ratio (adsorbed amount/total amount) of nitrobenzene vs. time on (a) CTF and (b) Amberlite XAD-4. Error bars, in most cases smaller than the symbols, represent variabilities calculated from duplicates.

theoretical value of TII varies from 0 and 1, corresponding to fully reversible adsorption (no hysteresis) and no desorption at all, respectively. The average TII is calculated to be 0.0 ± 0.1 for nitrobenzene on CTF. It is thus confirmed that adsorption of nitrobenzene on CTF is nearly completely reversible. Reversible adsorption facilitates separation of adsorbed chemicals from adsorbents, and therefore supports a sustainable use of recycled adsorbents. Sorption hysteresis may occur on glassy polymers when sorption and desorption undergo different mechanistic pathways due to metastate formation by irreversible relaxation of the polymer matrix [50–52]. The synthesized CTF has permanent rigid, framework pore structures that cannot be deformed, and consequently no sorption hysteresis is induced. 4. Summary The unique rigid, homogeneous framework nanoporous structure of CTF makes it a superior adsorbent for a wide variety of aromatic organic contaminants with high adsorption capacity, negligible effect of DHA on adsorption, fast adsorption/desorption kinetics, and complete adsorption reversibility. Furthermore, high adsorption selectivity can be achieved by precisely adjusting the cavity size and by functionalizing the cavity surface of CTF specific to the target compounds, as presented in other polymeric

nitrobenzene

q (mmol/kg)

Acknowledgments This work was supported by the National Science Foundation of China (Grants 21077050 and 21077049), the Natural Science Foundation of Jiangsu Province, China (Grant BK2010051), the SinoAmerica Collaborative Research Program (Grant 2010DFA91910), the National Basic Research Program of China (973 Program, Grant 2008CB418102) and the Program for Changjiang Scholars Innovative Research Team in University (Grant IRT1019). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jcis.2012.01.011. References

10 3

10 2

adsorption desorption

10 1 -4 10

frameworks [53–55]. Notably, the combining of the micropore-filling effect with the polar and/or ionic interactions greatly alleviates the ‘‘desolvation penalty’’ associated with these interactions, leading to markedly enhanced adsorption for hydrophilic ionic and/or polar compounds. Findings in this study highlight the potential of using CTF as a highly efficient and selective adsorbent for many adsorption based/coupled environmentally relevant applications such as stationary phase for chromatographic analysis and chemosensors.

10

-3

-2

10 C W (mmol/L)

10

-1

10

0

Fig. 9. Nitrobenzene adsorption and single-step desorption data plotted as solidphase concentration (q) vs. aqueous-phase concentration (CW) at equilibrium on covalent triazine-based framework. Lines represent the Freundlich fitted adsorption isotherm.

[1] P. Kuhn, M. Antonietti, A. Thomas, Angew. Chem. Int. Ed. 47 (2008) 3450. [2] P. Kuhn, A. Forget, D. Su, A. Thomas, M. Antonietti, J. Am. Chem. Soc. 130 (2008) 13333. [3] P. Kuhn, A. Thomas, M. Antonietti, Macromolecules 42 (2009) 319. [4] P. Kuhn, A. Forget, J. Hartmann, A. Thomas, M. Antonietti, Adv. Mater. 21 (2009) 897. [5] M. Bojdys, J. Jeromenok, A. Thomas, M. Antonietti, Adv. Mater. 22 (2010) 2202. [6] C.E. Chan-Thaw, A. Villa, P. Katekomol, D. Su, A. Thmas, L. Prati, Nano Lett. 10 (2010) 537. [7] R. Palkovits, M. Antonietti, P. Kuhn, A. Thomas, F. Schüth, Chem. Ing. Tech. 81 (2009) 1104. [8] P. Kuhn, K. Krüger, A. Thomas, M. Antonietti, Chem. Commun. (2008) 5815. [9] W. Zhang, F. Liang, C. Li, L. Qiu, Y. Yuan, F. Peng, X. Jiang, A. Xie, Y. Shen, J. Zhu, J. Hazard. Mater. 186 (2011) 984. [10] Z. Xu, Q. Zhang, H. Fang, Crit. Rev. Environ. Sci. Technol. 33 (2003) 363. [11] V. Azanova, J. Hradil, React. Funct. Polym. 41 (1999) 163. [12] M. Streat, L. Sweetland, Process Saf. Environ. Prot. 76 (1998) 127. [13] N. Penner, P. Nesterenko, M. Ilyin, M. Tsyurupa, V. Davankov, Chromatographia 50 (1999) 611. [14] M. Buchmeiser, Polymeric Materials in Organic Synthesis and Catalysis, WileyVCH, Weinheim, 2003. [15] S. Kiatkamjornwong, S. Traisaranapong, P. Prasassarakich, J. Porous. Mater. 6 (1999) 205.

J. Liu et al. / Journal of Colloid and Interface Science 372 (2012) 99–107 [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]

D.J. Pietrzyk, C.H. Chu, Anal. Chem. 49 (1977) 860. L. Ji, F. Liu, Z. Xu, S. Zheng, D. Zhu, Environ. Sci. Technol. 43 (2009) 7870. D.C. Kennedy, Environ. Sci. Technol. 7 (1973) 138. J. Huang, C. Yan, K. Huang, J. Colloid Interface Sci. 332 (2009) 60. G. Rexwinkel, B.B.M. Heesink, W.P.M. van Swaaij, J. Chem. Eng. Data 44 (1999) 1139. D.P. Oliveira, P.A. Carneiro, C.M. Rech, M.V. Zanoni, L.D. Claxton, G.A. Umbuzeiro, Environ. Sci. Technol. 40 (2006) 6682. J.P. Croué, M.F. Benedetti, D. Violleau, J.A. Leenheer, Environ. Sci. Technol. 37 (2003) 328. J.M. Wong, N.Y. Kado, P.A. Kuzmicky, H.S. Ning, J.E. Woodrow, D.P.H. Hsieh, J.N. Seiber, Anal. Chem. 63 (1991) 1644. M. Tolonen, P.E.J. Saris, M. Siika-aho, Microbiol. Biotechnol. 63 (2004) 659. P.E. Shaw, B.S. Buslig, J. Agric. Food. Chem. 34 (1986) 837. K. Grohmann, J.A. Manthey, R.G. Cameron, B.S. Buslig, J. Agric. Food. Chem. 47 (1999) 4859. M. Sander, J.J. Pignatello, Environ. Sci. Technol. 39 (2006) 1606. D. Zhu, S. Kwon, J.J. Pignatello, Environ. Sci. Technol. 39 (2005) 3990. S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. M.M. Dubinin, V.A. Astakhov, Adv. Chem. Ser. 102 (1971) 69. J.C. Ma, D.A. Dougherty, Chem. Rev. 97 (1997) 1303. G.W. Gokel, L.J. Barbour, S.L. De Wall, E.S. Meadows, Coord. Chem. Rev. 222 (2001) 127. M.H. Abraham, H.S. Chadha, G.S. Whiting, R.C. Mitchell, J. Pharm. Sci. 83 (1994) 1085. M. Borisover, E.R. Graber, Environ. Sci. Technol. 36 (2002) 4570. Q. Li, J. Hu, K. Zhao, Chin. J. Chem. 25 (2007) 1078.

107

[36] S.A. Boyd, G. Sheng, B.J. Teppen, C.T. Johnston, Environ. Sci. Technol. 35 (2001) 4227. [37] G. Sheng, C.T. Johnston, B.J. Teppen, S.A. Boyd, J. Agric. Food. Chem. 49 (2001) 2899. [38] C.T. Johnston, M.F. De Oliveira, B.J. Teppen, G. Sheng, S.A. Boyd, Environ. Sci. Technol. 35 (2001) 4767. [39] H. Li, B.J. Teppen, C.T. Johnston, S.A. Boyd, Environ. Sci. Technol. 38 (2004) 5433. [40] X. Qu, Y. Zhang, H. Li, S. Zheng, D. Zhu, Environ. Sci. Technol. 45 (2011) 2209. [41] S. Gotovac, J. Honda, H. Hattori, K. Takahashi, H. Kanoh, K. Kaneko, Nano Lett. 7 (2007) 583. [42] F. Tournus, S. Latil, M.I. Heggie, J.C. Charlier, Phys. Rev. B 72 (2005) 75431. [43] D. Lin, B. Xing, Environ. Sci. Technol. 42 (2008) 7254. [44] W. Chen, L. Duan, D. Zhu, Environ. Sci. Technol. 41 (2007) 8295. [45] B.A. De Moor, M.F. Reyniers, O.C. Gobin, J.A. Lercher, G.B. Marin, J. Phys. Chem. C 115 (2011) 1204. [46] D. Zhu, S. Hyun, J.J. Pignatello, L.S. Lee, Environ. Sci. Technol. 38 (2004) 4361. [47] H. Wijnja, J.J. Pignatello, K. Malekani, J. Environ. Qual. 33 (2004) 265. [48] J. Chen, W. Chen, D. Zhu, Environ. Sci. Technol. 42 (2008) 7225. [49] M. Sander, Y.F. Lu, J.J. Pignatello, J. Environ. Qual. 34 (2005) 1063. [50] B. Pan, B. Xing, Environ. Sci. Technol. 42 (2008) 9005. [51] B. Pan, D. Lin, H. Mashayekhi, B. Xing, Environ. Sci. Technol. 42 (2008) 5480. [52] X. Wang, Y. Liu, W. Qiu, D. Zhu, J. Mater. Chem. 12 (2002) 1636. [53] E. Bourgeat-Lami, I. Tissot, F. Lefebvre, Macromolecules 35 (2002) 6185. [54] S. Hasegawa, S. Horike, R. Matsuda, S. Furukawa, K. Mochizuki, Y. Kinoshita, S. Kitagawa, J. Am. Chem. Soc. 129 (2007) 2607. [55] W.J. Rieter, K.M.L. Taylor, W. Lin, J. Am. Chem. Soc. 129 (2007) 9852.