Environmental Pollution 190 (2014) 101e108
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Mechanism of and relation between the sorption and desorption of nonylphenol on black carbon-inclusive sediment Lou Liping a, Cheng Guanghuan a, Deng Jingyou a, Sun Mingyang a, Chen Huanyu a, Yang Qiang b, Xu Xinhua a, * a b
Department of Environmental Engineering, Zhejiang University, Hangzhou 310029, PR China Hangzhou Research Institute of Environment Science, Hangzhou 310014, PR China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 January 2014 Received in revised form 9 March 2014 Accepted 22 March 2014 Available online xxx
Correlation between the sorption and desorption of nonylphenol (NP) and binary linear regression were conducted to reveal the underlying mechanism of and relation between sorption domains and desorption sites in black carbon (BC)-amended sediment. The sorption and desorption data could be fitted well using dual-mode (R2 ¼ 0.971e0.996) and modified two-domain model (R2 ¼ 0.986e0.995), respectively, and there were good correlations between these two parts of parameters (R2 ¼ 0.884e0.939, P < 0.01). The NP percentage in desorbable fraction was almost equal to that of the partition fraction, suggesting the desorbed NP came from linear partition domain, whereas the resistant desorption NP was segregated in nonlinear adsorption sites, which were dominated by pores in BC-amended sediment. Our investigation refined theory about the relation between sorption domains and desorption sites in sediment and could be used to predict the release risk of NP using sorption data when BC is used for NP pollution control. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Black carbon (BC) Sediment Nonylphenol (NP) Sorption Desorption
1. Introduction Soil/sediment organic matter (SOM) is the principal factor controlling the sorption of hydrophobic organic compounds (HOCs) (Omega, 1968). The mechanism of sorption to SOM had received a great deal of attention because of its fundamental importance to HOC transport, bioavailability, and toxicity (Cornelissen and Gustafsson, 2004; Armitage et al., 2008). Evidence is accumulating from studies on subjects such as nonlinear sorption isotherms (Johnson et al., 2001; Kleineidam et al., 2002), multiphasic desorption kinetics (Pignatello and Xing, 1996; Cornelissen et al., 1999), and strongly elevated TOC-water distribution ratios in the field (Gustafsson et al., 1997; Ehlers and Luthy, 2003; Lohmann et al., 2005; Sun et al., 2012). The sorption of HOCs to SOM can be described as having “dual-mode sorption” (Pignatello and Xing, 1996; Huang et al., 1997), that is, linear absorption in amorphous organic matter (AOM) such as humic/fulvic substances and lignin, and nonlinear adsorption to more condensed materials such as black carbon (BC), coal, and kerogen. SOM has been conceptualized
* Corresponding author. E-mail address: [email protected] (X. Xinhua). http://dx.doi.org/10.1016/j.envpol.2014.03.027 0269-7491/Ó 2014 Elsevier Ltd. All rights reserved.
as an amalgam of rubbery and glassy phases. Each phase has a dissolution domain, but the glassy phase contains additional surface sites and nanopores (Pignatello and Xing, 1996; Huang and Weber, 1997). The sorption of HOCs to the rubbery state occurs by dissolution, and sorption to the glassy state occurs by concurrent dissolution, surface adsorption and hole-filling mechanisms. The sorption isotherms of HOCs to sediment have been shown to consist of at least two components, namely a Langmuir component and a linear component (Xing et al., 1996; Huang et al., 1997; Xing and Pignatello, 1997), and the desorption of HOCs from soils and sediments occurs in several stages, for example, a rapid desorption phase (hours to days), a slow phase (weeks to months) and a very slow phase (months to years) (Pignatello and Xing, 1996; Cornelissen et al., 1997a; Ten Hulscher et al., 1999). There is a question as to whether there are relations between sorption domains and desorption sites? In the late 1990s and early 2000s, several studies on related research were published (Xing et al., 1996; Huang et al., 1997; LeBoeuf and Weber, 1997; Xia and Ball, 1999; Cornelissen et al., 2000). The authors confirmed that compounds at slow and very slow desorbing sites in sediment exhibited Langmuir-like sorption, whereas compounds at rapidly desorbing sites showed linear sorption. However, these experiments were conducted using natural sediment with little grassy phase content,
102
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
and the HOCs sorbed on natural sediment can desorb almost completely, even if the desorption hysteresis is still apparent and requires long time scales. It has been demonstrated that BC, one type of glassy organic matter in sediment, engaged in extraordinarily strong and nonlinear adsorption to HOCs, leading to distribution coefficients that were several orders of magnitude higher than in natural sediments (Bucheli and Gustaffson, 2000; Accardi-dey and Gschwend, 2002; Chun et al., 2004; Cornelissen et al., 2004a). The decline in desorption rates and fractions, and the occurrence of desorption resistance have also been attributed to the presence of BC (Chai et al., 2006, 2007; Zhang et al., 2010; Yang et al., 2012; Marchal et al., 2013). Therefore, applying BC to treat contaminated sediment is a practical approach to reducing the ecological risk of HOCs (Jones et al., 2011; Martin et al., 2012; Teixidó et al., 2013). However, we wondered whether the addition of glassy phase organic matter, namely BC, would change the relation between sorption domains and desorption sites in sediment. In addition, desorption experiments are tedious, so we wanted to determine whether we could predict the NP percentage in the desorption fraction by using adsorption data. In the present study, one typical endocrine disruptor and persistent pollutant called nonylphenol (NP) was used as a representative HOC. Substantial ecotoxicological data have demonstrated that NP has significant estrogenic effects, toxicity and biological accumulation (Granmo et al., 1989). And more than 60% of NP is distributed in sediments. Sorption/desorption to/from sediment is a fundamental process controlling the fate, toxicity and bioavailability of NP. So we investigated the sorption and desorption properties of NP on/from sediment amended with rice straw biochar and fly ash, respectively. The correlation between sorption isotherm and desorption kinetic parameters was analyzed to determine the quantitative relation and mechanism between the sorption and desorption of NP on sediment containing BC. The objective of this study was to reveal the relation between sorption domains and desorption sites in BC-amended sediment and predict NP percentage in the desorption fraction from sorption data when fresh BC is used for NP pollution control. 2. Materials and methods 2.1. Chemicals and materials NP with a purity of >99% was purchased from Aladdin (Shanghai, China) and prepared as a concentrated stock solution with acetonitrile. Tenax TA (60e80 mesh) was obtained from Supelco (Bellefonte, Pennsylvania, USA). Tenax TA was activated or regenerated by ultrasonic washing with methanol, acetone and hexane in order (Cornelissen et al., 1997b). The sediment was obtained by a clam sampler to collect surface sediment of the Qian-tang River, Hangzhou, Zhejiang province, China. The TOC and BC content of the sediment was 0.964% and 0.37%, respectively. For more details, see Luo et al. (2011). Rice straw black carbon (RC) was prepared from air-dried rice straw collected from the Hua-jia-chi farm of Zhejiang University in China. The rice straw was burned on a stainless steel plate in an open field under uncontrolled conditions. Fly ash black carbon (FC) was collected from the electrostatic precipitation division of a thermoelectric plant in Hangzhou, Zhejiang province, China. To obtain purified RC and FC, the BC samples were treated with 2 M HCl and 1 M: 1 M HCleHF solutions, washed with distilled water, and oven-dried overnight at 105 C. The RC and FC properties were detailed in our previous works (Luo et al., 2011). The BC-amended sediments used in this experiment were prepared by mixing the sediment and specific quantities of RC and FC to achieve amendment rates of 0%, 0.5%, 2.0% and 5.0% (w/w) (Chi, 2014). The BC-amended sediments were thoroughly mixed before use in the NP sorption and desorption experiments. 2.2. Sorption of NP on BC and BC-amended sediment The sorption of NP by BC and BC-amended sediment were measured by using the batch equilibration technique (Düring et al., 2002). The sorbents were added to 50-mL glass centrifuge vials, and each vial received 30 mL of NP solution ranging from 0.2 to 4.0 mg/L with an electrolyte matrix containing 200 mg/L NaCl and 200 mg/mL NaN3, which was added to inhibit biodegradation (Chen et al., 2004). All sorption experiments were conducted in triplicate. The vials were shaken at
150 rpm on a horizontal shaker at 25 1 C for 16 h in the dark. Preliminary tests indicated that 16 h was sufficient to reach apparent equilibrium. After the establishment of sorption equilibrium, the NP concentrations in the initial and equilibrated supernatants were analyzed by high performance liquid chromatography (HPLC, Agilent 1100 series) with a diode array fluorescence detector (FLD) detector and a C18 reversed-phase column (ODS, 5 mm, 2.1 mm 250 mm). Acetonitrile and water (70:30, v/v) were used as the mobile phase at a flow rate of 1.0 mL/min, and the injection volume was 20 mL. The FLD wavelength for detecting NP was set at detection and excitation wavelengths of 233 nm and 302 nm, respectively. The NP concentrations were quantified with an external standard method. Blanks loaded with NP solution without adsorbents were also run to assess the solute losses to reactor components during the sorption process. 2.3. Desorption of NP from BC and BC-amended sediment BC and BC-amended sediment were spiked with NP at 5000 mg/kg and 100 mg/ kg (dry weight), respectively. After the carrier solvent was allowed to evaporate for approximately 12 h in a fume hood until dry, the treated BC and BC-amended sediment samples were capped and shaken at 25 C and 80 g in the dark at room temperature for 24 h. The spiked samples were desorbed for different time increments by using a method similar to that described in Xu et al. (2008). In brief, spiked samples were transferred to 50 mL glass centrifuge tubes. Tenax beads (0.1 g) and 30 mL of electrolyte matrix were then added to each container. The samples were shaken on a horizontal shaker at 100 rpm at room temperature for 0.5, 2, 4, 6, 12, 24, 48, 96, 192 and 384 h, three replicates. For each desorption time interval, the Tenax beads were separated from the sediment suspension. The Tenax beads were collected and then transferred to a 30-mL clear glass vial. Another 0.1 g of clean Tenax beads was added to the vial to continue the desorption experiment. The collected Tenax beads were extracted by sonication using 5 mL of dichloromethane-methanol mixture (1:1, v/v) for three consecutive times. The extracts from the same sample were combined, after which the collected solvent was concentrated to 1 mL through nitrogen blowing and then analyzed by HPLC. Quality control measures were carried out in the Tenax-aided desorption kinetics experiment. First, the efficiency of Tenax has been checked and recovery of NP using Tenax absorption was >98%. Furthermore, after Tenax-aided desorption kinetics experiment, methanol was used to extract the residual NP in sorbent and to calculate the loss of NP during desorption process. Losses of NP during desorption process were all 1.0 mg/L). For the BC-sediment system, the adsorption contribution increased and the partition contribution decreased with the BC content. For all sediment samples with BC & 2%, the contribution percentages of the partition (66.18%e100%) were higher than that of the adsorption (0%e33.82%), whereas when the BC content increased to 5%, the contribution percentages of partitioning were similar to the contribution percentages of adsorption in the sediment-BC system. 3.2. Desorption kinetics of NP from BCs and sediment-BC systems The cumulative desorption of NP, which was plotted as the St/S0 versus time, is displayed in Fig. 3. The experimental data were first fitted by TM (Eq. (5)). The fitted parameter of kveryslow was very small (not listed). For example, the kveryslow of 2.0% RC-amended sediment was 9.19 1083/h. The desorption time was less than half a month, so the relatively small kveryslow, kveryslow t was close to 0. Thus, the TM model could be simplified as follows: St =S0 ¼ Frap ekrap t þ Fslow ekslow t þ Fr (Fr ¼ Fveryslow þ Fnon, meaning the fraction resistant to desorption), which was called the modified two domains model (MM). The desorption kinetic data were in agreement with the MM, with R2 values ranging from 0.986 to 0.995 (Table 2). The desorption curves (plotted as St/S0 versus time) were all characterized by an initial rapid drop in the amount of NP sorbed followed by a slower decrease in the sorbed concentrations. The rest of the sorbed NP remained almost the
same. NP on sediment could desorb almost completely (with Fr ¼ 0.0403 0.012) over 16 d, but desorption hysteresis (Fslow ¼ 0.474 0.035) occurred because of the 0.37% glassy phase (BC) in this sediment. The BC-amended sediment was dominated by the glassy phase, and it was regarded as a strong adsorbent for NP because part of the NP molecules could not desorb but instead formed a residue in the sorbent, which is different from a desorption from weak sorbent (natural sediment, dominated by the rubbery phase), from which the sorbate almost desorbed. The desorption kinetics fitted well by two domain model (Rhodes et al., 2010). Desorption kinetics data for all the sorbents showed that the rapidly desorbing fraction generally occurred from 0 to 6 h, then from 6 to 192 h, and >192 h, and the desorption was dominated by the resistant desorbing fraction. More than 80% of the NP on RC and FC was Fr, and Frap and Fslow were both less than 10%. NP desorption data from sediment amended with BCs showed that Frap, Fslow, krap and kslow all decreased when the resistant desorption fraction (Fr) had increased BC content. For example, when the 5% RC was amended, Frap and Fslow decreased by approximately 50.10% and 64.77%, respectively, with a comparable increase in Fr (Table 2). Moreover, the desorption of NP from RC-amended sediments was more difficult than that of FC-amended sediments. Both krap (/h) and kslow (/h) decreased as the BC content in the sediment increased. The values of krap (/h) and kslow (/h) were greater in the FC-sediment system compared to the RC-sediment system. Desorption rates had been used to imply the binding strength of chemicals (Greenberg et al., 2005), so the result indicated that RC-amended sediment had a far higher affinity to NP than FC-amended sediment. The results indicated that the strong sorption of BC limited the desorption of NP. The most important sorption mechanisms
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
1.0
1.0
0.8
0.8
0.6
0.6 St/S0
St/S0
105
0.4
0.4
0.2
0.2
0.0
0.0 0
50 0.0%
100
150 200 250 300 Desorption time, min 0.5% 2.0% 5.0%
350
400
RC
0
50 0.0%
100
150 200 250 300 Desorption time, min 0.5% 2.0% 5.0%
350
400
FC
Fig. 3. Desorption kinetics curves of NP from BCs and sediment-BC systems fitted by MM.
between HOCs and sediments amended with BC were pore adsorption and pep interaction processes (Lou et al., 2011). In this study, the addition of BC with porosity and pore volumes of 72.10 and 21.00 m2/g 0.133 and 0.090 mL/g for RC and FC, respectively (Table 3), increased the surface area available for NP sorption. 4. Discussion 4.1. Correlation between sorption and desorption of NP on BCsediment systems It is known that the higher the sorption capacity, the lower the desorption amount of NP. As Cornelissen et al. (2000) qualitatively stated, the natural sediment sites exhibiting rapid and slow desorption were most likely the same as the sites showing linear (partition) and nonlinear sorption (adsorption). However, the existence of a quantitative relation between sorption and desorption and whether HOC desorption amounts can be predicted from the sorption parameters are still unknown. Therefore, the correlations between Kom (linear partition coefficient) and Frap (rapid desorption fraction), Kom and Frap þ Fslow (representing the total desorbing fraction), Qmax (nonlinear adsorption parameter) and Fr (resistant desorption fraction) were investigated (Fig. 4). In accordance with the theory that the natural sediment sites with rapid and slow desorption were most likely the same as the sites with linear (partition) and nonlinear sorption (adsorption), Frap should negatively correlate with Kom, and Fslow should positively correlate with Qmax. However, both Frap and Frap þ Fslow were significantly and negatively correlated with Kom (P < 0.01) in the present study, and the linear regression between Kom and Frap þ Fslow (R2 ¼ 0.898, P < 0.01) provided a better fit than that of Kom and Frap (R2 ¼ 0.884, P < 0.01). Fr was significantly and positively correlated with Qmax (R2 ¼ 0.989, P < 0.01). This finding indicated not only that an increase in the partition and adsorption capacities decreased the desorption amount and increased the resistant desorption amount of NP but also that there were quantitative relations between sorption domains and desorption fractions. Therefore, we can avoid tedious desorption experiments and directly calculate the NP content of every desorption fraction by using sorption data to assess the ecological risk of NP release when BCs are used for organic pollution remediation. The most important conclusion was these findings reminded us that the theoretical relation of sorption domains and desorption sites in natural sediment was not suitable for BCamended sediment.
4.2. Mechanism between the sorption and desorption of NP on BCsediment systems Fig. 4 showed that the desorbing fraction (Frap, Frap þ Fslow) and resistant desorption fraction (Fr) were significantly correlated to Kom and Qmax, respectively. However, the relation between sorption domains and desorption sites in sediment-amended BC was unknown. On the basis of these results, we compared the desorbing fraction (Frap, Frap þ Fslow) with the partition to total sorption ratio (Kom/Kf) of NP on the RC-sediment system and the FC-sediment system, respectively (Fig. 5). Kf, the sorption coefficient of the Freundlich model, represents the degree of sorption strength and capacity. Kom, the partition coefficient of HOC in the partition domain, represents the degree of partitioning. Therefore, the Kom/Kf in this study represents the ratio of NP concentration in the partition domain to that in the total sorption sites. The linear equations for the correlation between Frap and Kom/Kf were y ¼ 0.503x þ 0.022 (R2 ¼ 0.921, P < 0.01) and y ¼ 0.492x 0.0053 (R2 ¼ 0.958, P < 0.01) for sediment amended with RC and sediment amended with FC, respectively, and Frap þ Fslow and Kom/Kf were represented by y ¼ 0.946x 0.030 (R2 ¼ 0.976, P < 0.01) and y ¼ 0.980x 0.0051 (R2 ¼ 0.997, P < 0.01) for the RCsediment system and FC-sediment system, respectively. The line of the RC-sediment system was close to the line of the FC-sediment system regardless of Frap or Frap þ Fslow, so the same linear equations y ¼ 0.499x þ 0.0008 (R2 ¼ 0.906, P < 0.01) and y ¼ 0.991x 0.0383 (R2 ¼ 0.978, P < 0.01) could be used for Frap and Frap þ Fslow, respectively. Furthermore, the line for Frap was lower than the line of y ¼ x, and the slope was only 0.499, suggesting that NP from the rapid desorption fraction was only half of the NP in the partition fraction, but not equal to the NP in the partition fraction. The Frap þ Fslow line was close to the y ¼ x line, which indicated that the NP concentration of the desorbing fraction was almost equal to the NP in the partition fraction. In other words, the NP in the linear partition fraction can desorb completely, but it does not only containing rapidly desorption. Therefore, the theory that the natural sediment sites exhibiting rapid and slow desorption are most likely the same as the sites showing linear (partition) and nonlinear sorption (adsorption) was not suitable for sediment-amended BC, which would overestimate the HOC release. At the same time, (1 (Frap þ Fslow)) should have the same quantitative correlation with (1 Kom/Kf). In other words, the NP in the resistant desorbing fraction was the same as that in the nonlinear adsorption fraction (dominated by BC). When combined with Section 4.1, the released and residual NP from/in BCsediment system can be obtained from sorption data.
106
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
Table 2 Parameters of desorption kinetics of NP from sediment-BC systems fitted by MM. Sorbents
Fitting parameters Fr
S 0.5% RC 2.0% RC 5.0% RC 0.5% FC 2.0% FC 5.0% FC Pure RC Pure FC
Frap
0.0403 0.242 0.424 0.579 0.0793 0.179 0.371 0.836 0.829
0.012 0.013 0.013 0.0076 0.014 0.022 0.013 0.0042 0.0033
0.485 0.427 0.279 0.242 0.447 0.491 0.376 0.0858 0.0831
0.043 0.075 0.061 0.061 0.057 0.090 0.060 0.020 0.024
krap (/h)
Fslow
2.586 0.262 0.203 0.174 2.872 0.365 0.245 0.307 0.272
0.474 0.311 0.275 0.167 0.473 0.290 0.234 0.0712 0.0819
0.035 0.074 0.059 0.060 0.049 0.088 0.059 0.020 0.024
kslow (103/h)
R2
95.987 34.132 23.975 30.077 120.901 33.959 28.258 34.096 39.851
0.994 0.995 0.993 0.995 0.991 0.986 0.994 0.988 0.993
Table 3 Surface areas and pore structures of BC and BC amended sediment. Property 2
BET surface area (m /g) Pore volume (mL/g)
S
0.5% RC
2.0% RC
5.0% RC
0.5% FC
2.0% FC
5.0% FC
Pure RC
Pure FC
8.73 0.035
10.65 0.040
13.16 0.048
20.46 0.057
9.10 0.046
9.16 0.046
9.25 0.047
72.10 0.133
21.00 0.039
Fig. 4. Correlation between parameters of sorption isotherms and parameters of desorption kinetics.
Sediment amended with RC and sediment amended with FC 1.0 Sediment amended with RC Sediment amended with FC y=x 2 y=0.991x-0.0383; R =0.978; P
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Mechanism of and relation between the sorption and desorption of nonylphenol on black carbon-inclusive sediment Lou Liping a, Cheng Guanghuan a, Deng Jingyou a, Sun Mingyang a, Chen Huanyu a, Yang Qiang b, Xu Xinhua a, * a b
Department of Environmental Engineering, Zhejiang University, Hangzhou 310029, PR China Hangzhou Research Institute of Environment Science, Hangzhou 310014, PR China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 16 January 2014 Received in revised form 9 March 2014 Accepted 22 March 2014 Available online xxx
Correlation between the sorption and desorption of nonylphenol (NP) and binary linear regression were conducted to reveal the underlying mechanism of and relation between sorption domains and desorption sites in black carbon (BC)-amended sediment. The sorption and desorption data could be fitted well using dual-mode (R2 ¼ 0.971e0.996) and modified two-domain model (R2 ¼ 0.986e0.995), respectively, and there were good correlations between these two parts of parameters (R2 ¼ 0.884e0.939, P < 0.01). The NP percentage in desorbable fraction was almost equal to that of the partition fraction, suggesting the desorbed NP came from linear partition domain, whereas the resistant desorption NP was segregated in nonlinear adsorption sites, which were dominated by pores in BC-amended sediment. Our investigation refined theory about the relation between sorption domains and desorption sites in sediment and could be used to predict the release risk of NP using sorption data when BC is used for NP pollution control. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Black carbon (BC) Sediment Nonylphenol (NP) Sorption Desorption
1. Introduction Soil/sediment organic matter (SOM) is the principal factor controlling the sorption of hydrophobic organic compounds (HOCs) (Omega, 1968). The mechanism of sorption to SOM had received a great deal of attention because of its fundamental importance to HOC transport, bioavailability, and toxicity (Cornelissen and Gustafsson, 2004; Armitage et al., 2008). Evidence is accumulating from studies on subjects such as nonlinear sorption isotherms (Johnson et al., 2001; Kleineidam et al., 2002), multiphasic desorption kinetics (Pignatello and Xing, 1996; Cornelissen et al., 1999), and strongly elevated TOC-water distribution ratios in the field (Gustafsson et al., 1997; Ehlers and Luthy, 2003; Lohmann et al., 2005; Sun et al., 2012). The sorption of HOCs to SOM can be described as having “dual-mode sorption” (Pignatello and Xing, 1996; Huang et al., 1997), that is, linear absorption in amorphous organic matter (AOM) such as humic/fulvic substances and lignin, and nonlinear adsorption to more condensed materials such as black carbon (BC), coal, and kerogen. SOM has been conceptualized
* Corresponding author. E-mail address: [email protected] (X. Xinhua). http://dx.doi.org/10.1016/j.envpol.2014.03.027 0269-7491/Ó 2014 Elsevier Ltd. All rights reserved.
as an amalgam of rubbery and glassy phases. Each phase has a dissolution domain, but the glassy phase contains additional surface sites and nanopores (Pignatello and Xing, 1996; Huang and Weber, 1997). The sorption of HOCs to the rubbery state occurs by dissolution, and sorption to the glassy state occurs by concurrent dissolution, surface adsorption and hole-filling mechanisms. The sorption isotherms of HOCs to sediment have been shown to consist of at least two components, namely a Langmuir component and a linear component (Xing et al., 1996; Huang et al., 1997; Xing and Pignatello, 1997), and the desorption of HOCs from soils and sediments occurs in several stages, for example, a rapid desorption phase (hours to days), a slow phase (weeks to months) and a very slow phase (months to years) (Pignatello and Xing, 1996; Cornelissen et al., 1997a; Ten Hulscher et al., 1999). There is a question as to whether there are relations between sorption domains and desorption sites? In the late 1990s and early 2000s, several studies on related research were published (Xing et al., 1996; Huang et al., 1997; LeBoeuf and Weber, 1997; Xia and Ball, 1999; Cornelissen et al., 2000). The authors confirmed that compounds at slow and very slow desorbing sites in sediment exhibited Langmuir-like sorption, whereas compounds at rapidly desorbing sites showed linear sorption. However, these experiments were conducted using natural sediment with little grassy phase content,
102
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
and the HOCs sorbed on natural sediment can desorb almost completely, even if the desorption hysteresis is still apparent and requires long time scales. It has been demonstrated that BC, one type of glassy organic matter in sediment, engaged in extraordinarily strong and nonlinear adsorption to HOCs, leading to distribution coefficients that were several orders of magnitude higher than in natural sediments (Bucheli and Gustaffson, 2000; Accardi-dey and Gschwend, 2002; Chun et al., 2004; Cornelissen et al., 2004a). The decline in desorption rates and fractions, and the occurrence of desorption resistance have also been attributed to the presence of BC (Chai et al., 2006, 2007; Zhang et al., 2010; Yang et al., 2012; Marchal et al., 2013). Therefore, applying BC to treat contaminated sediment is a practical approach to reducing the ecological risk of HOCs (Jones et al., 2011; Martin et al., 2012; Teixidó et al., 2013). However, we wondered whether the addition of glassy phase organic matter, namely BC, would change the relation between sorption domains and desorption sites in sediment. In addition, desorption experiments are tedious, so we wanted to determine whether we could predict the NP percentage in the desorption fraction by using adsorption data. In the present study, one typical endocrine disruptor and persistent pollutant called nonylphenol (NP) was used as a representative HOC. Substantial ecotoxicological data have demonstrated that NP has significant estrogenic effects, toxicity and biological accumulation (Granmo et al., 1989). And more than 60% of NP is distributed in sediments. Sorption/desorption to/from sediment is a fundamental process controlling the fate, toxicity and bioavailability of NP. So we investigated the sorption and desorption properties of NP on/from sediment amended with rice straw biochar and fly ash, respectively. The correlation between sorption isotherm and desorption kinetic parameters was analyzed to determine the quantitative relation and mechanism between the sorption and desorption of NP on sediment containing BC. The objective of this study was to reveal the relation between sorption domains and desorption sites in BC-amended sediment and predict NP percentage in the desorption fraction from sorption data when fresh BC is used for NP pollution control. 2. Materials and methods 2.1. Chemicals and materials NP with a purity of >99% was purchased from Aladdin (Shanghai, China) and prepared as a concentrated stock solution with acetonitrile. Tenax TA (60e80 mesh) was obtained from Supelco (Bellefonte, Pennsylvania, USA). Tenax TA was activated or regenerated by ultrasonic washing with methanol, acetone and hexane in order (Cornelissen et al., 1997b). The sediment was obtained by a clam sampler to collect surface sediment of the Qian-tang River, Hangzhou, Zhejiang province, China. The TOC and BC content of the sediment was 0.964% and 0.37%, respectively. For more details, see Luo et al. (2011). Rice straw black carbon (RC) was prepared from air-dried rice straw collected from the Hua-jia-chi farm of Zhejiang University in China. The rice straw was burned on a stainless steel plate in an open field under uncontrolled conditions. Fly ash black carbon (FC) was collected from the electrostatic precipitation division of a thermoelectric plant in Hangzhou, Zhejiang province, China. To obtain purified RC and FC, the BC samples were treated with 2 M HCl and 1 M: 1 M HCleHF solutions, washed with distilled water, and oven-dried overnight at 105 C. The RC and FC properties were detailed in our previous works (Luo et al., 2011). The BC-amended sediments used in this experiment were prepared by mixing the sediment and specific quantities of RC and FC to achieve amendment rates of 0%, 0.5%, 2.0% and 5.0% (w/w) (Chi, 2014). The BC-amended sediments were thoroughly mixed before use in the NP sorption and desorption experiments. 2.2. Sorption of NP on BC and BC-amended sediment The sorption of NP by BC and BC-amended sediment were measured by using the batch equilibration technique (Düring et al., 2002). The sorbents were added to 50-mL glass centrifuge vials, and each vial received 30 mL of NP solution ranging from 0.2 to 4.0 mg/L with an electrolyte matrix containing 200 mg/L NaCl and 200 mg/mL NaN3, which was added to inhibit biodegradation (Chen et al., 2004). All sorption experiments were conducted in triplicate. The vials were shaken at
150 rpm on a horizontal shaker at 25 1 C for 16 h in the dark. Preliminary tests indicated that 16 h was sufficient to reach apparent equilibrium. After the establishment of sorption equilibrium, the NP concentrations in the initial and equilibrated supernatants were analyzed by high performance liquid chromatography (HPLC, Agilent 1100 series) with a diode array fluorescence detector (FLD) detector and a C18 reversed-phase column (ODS, 5 mm, 2.1 mm 250 mm). Acetonitrile and water (70:30, v/v) were used as the mobile phase at a flow rate of 1.0 mL/min, and the injection volume was 20 mL. The FLD wavelength for detecting NP was set at detection and excitation wavelengths of 233 nm and 302 nm, respectively. The NP concentrations were quantified with an external standard method. Blanks loaded with NP solution without adsorbents were also run to assess the solute losses to reactor components during the sorption process. 2.3. Desorption of NP from BC and BC-amended sediment BC and BC-amended sediment were spiked with NP at 5000 mg/kg and 100 mg/ kg (dry weight), respectively. After the carrier solvent was allowed to evaporate for approximately 12 h in a fume hood until dry, the treated BC and BC-amended sediment samples were capped and shaken at 25 C and 80 g in the dark at room temperature for 24 h. The spiked samples were desorbed for different time increments by using a method similar to that described in Xu et al. (2008). In brief, spiked samples were transferred to 50 mL glass centrifuge tubes. Tenax beads (0.1 g) and 30 mL of electrolyte matrix were then added to each container. The samples were shaken on a horizontal shaker at 100 rpm at room temperature for 0.5, 2, 4, 6, 12, 24, 48, 96, 192 and 384 h, three replicates. For each desorption time interval, the Tenax beads were separated from the sediment suspension. The Tenax beads were collected and then transferred to a 30-mL clear glass vial. Another 0.1 g of clean Tenax beads was added to the vial to continue the desorption experiment. The collected Tenax beads were extracted by sonication using 5 mL of dichloromethane-methanol mixture (1:1, v/v) for three consecutive times. The extracts from the same sample were combined, after which the collected solvent was concentrated to 1 mL through nitrogen blowing and then analyzed by HPLC. Quality control measures were carried out in the Tenax-aided desorption kinetics experiment. First, the efficiency of Tenax has been checked and recovery of NP using Tenax absorption was >98%. Furthermore, after Tenax-aided desorption kinetics experiment, methanol was used to extract the residual NP in sorbent and to calculate the loss of NP during desorption process. Losses of NP during desorption process were all 1.0 mg/L). For the BC-sediment system, the adsorption contribution increased and the partition contribution decreased with the BC content. For all sediment samples with BC & 2%, the contribution percentages of the partition (66.18%e100%) were higher than that of the adsorption (0%e33.82%), whereas when the BC content increased to 5%, the contribution percentages of partitioning were similar to the contribution percentages of adsorption in the sediment-BC system. 3.2. Desorption kinetics of NP from BCs and sediment-BC systems The cumulative desorption of NP, which was plotted as the St/S0 versus time, is displayed in Fig. 3. The experimental data were first fitted by TM (Eq. (5)). The fitted parameter of kveryslow was very small (not listed). For example, the kveryslow of 2.0% RC-amended sediment was 9.19 1083/h. The desorption time was less than half a month, so the relatively small kveryslow, kveryslow t was close to 0. Thus, the TM model could be simplified as follows: St =S0 ¼ Frap ekrap t þ Fslow ekslow t þ Fr (Fr ¼ Fveryslow þ Fnon, meaning the fraction resistant to desorption), which was called the modified two domains model (MM). The desorption kinetic data were in agreement with the MM, with R2 values ranging from 0.986 to 0.995 (Table 2). The desorption curves (plotted as St/S0 versus time) were all characterized by an initial rapid drop in the amount of NP sorbed followed by a slower decrease in the sorbed concentrations. The rest of the sorbed NP remained almost the
same. NP on sediment could desorb almost completely (with Fr ¼ 0.0403 0.012) over 16 d, but desorption hysteresis (Fslow ¼ 0.474 0.035) occurred because of the 0.37% glassy phase (BC) in this sediment. The BC-amended sediment was dominated by the glassy phase, and it was regarded as a strong adsorbent for NP because part of the NP molecules could not desorb but instead formed a residue in the sorbent, which is different from a desorption from weak sorbent (natural sediment, dominated by the rubbery phase), from which the sorbate almost desorbed. The desorption kinetics fitted well by two domain model (Rhodes et al., 2010). Desorption kinetics data for all the sorbents showed that the rapidly desorbing fraction generally occurred from 0 to 6 h, then from 6 to 192 h, and >192 h, and the desorption was dominated by the resistant desorbing fraction. More than 80% of the NP on RC and FC was Fr, and Frap and Fslow were both less than 10%. NP desorption data from sediment amended with BCs showed that Frap, Fslow, krap and kslow all decreased when the resistant desorption fraction (Fr) had increased BC content. For example, when the 5% RC was amended, Frap and Fslow decreased by approximately 50.10% and 64.77%, respectively, with a comparable increase in Fr (Table 2). Moreover, the desorption of NP from RC-amended sediments was more difficult than that of FC-amended sediments. Both krap (/h) and kslow (/h) decreased as the BC content in the sediment increased. The values of krap (/h) and kslow (/h) were greater in the FC-sediment system compared to the RC-sediment system. Desorption rates had been used to imply the binding strength of chemicals (Greenberg et al., 2005), so the result indicated that RC-amended sediment had a far higher affinity to NP than FC-amended sediment. The results indicated that the strong sorption of BC limited the desorption of NP. The most important sorption mechanisms
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
1.0
1.0
0.8
0.8
0.6
0.6 St/S0
St/S0
105
0.4
0.4
0.2
0.2
0.0
0.0 0
50 0.0%
100
150 200 250 300 Desorption time, min 0.5% 2.0% 5.0%
350
400
RC
0
50 0.0%
100
150 200 250 300 Desorption time, min 0.5% 2.0% 5.0%
350
400
FC
Fig. 3. Desorption kinetics curves of NP from BCs and sediment-BC systems fitted by MM.
between HOCs and sediments amended with BC were pore adsorption and pep interaction processes (Lou et al., 2011). In this study, the addition of BC with porosity and pore volumes of 72.10 and 21.00 m2/g 0.133 and 0.090 mL/g for RC and FC, respectively (Table 3), increased the surface area available for NP sorption. 4. Discussion 4.1. Correlation between sorption and desorption of NP on BCsediment systems It is known that the higher the sorption capacity, the lower the desorption amount of NP. As Cornelissen et al. (2000) qualitatively stated, the natural sediment sites exhibiting rapid and slow desorption were most likely the same as the sites showing linear (partition) and nonlinear sorption (adsorption). However, the existence of a quantitative relation between sorption and desorption and whether HOC desorption amounts can be predicted from the sorption parameters are still unknown. Therefore, the correlations between Kom (linear partition coefficient) and Frap (rapid desorption fraction), Kom and Frap þ Fslow (representing the total desorbing fraction), Qmax (nonlinear adsorption parameter) and Fr (resistant desorption fraction) were investigated (Fig. 4). In accordance with the theory that the natural sediment sites with rapid and slow desorption were most likely the same as the sites with linear (partition) and nonlinear sorption (adsorption), Frap should negatively correlate with Kom, and Fslow should positively correlate with Qmax. However, both Frap and Frap þ Fslow were significantly and negatively correlated with Kom (P < 0.01) in the present study, and the linear regression between Kom and Frap þ Fslow (R2 ¼ 0.898, P < 0.01) provided a better fit than that of Kom and Frap (R2 ¼ 0.884, P < 0.01). Fr was significantly and positively correlated with Qmax (R2 ¼ 0.989, P < 0.01). This finding indicated not only that an increase in the partition and adsorption capacities decreased the desorption amount and increased the resistant desorption amount of NP but also that there were quantitative relations between sorption domains and desorption fractions. Therefore, we can avoid tedious desorption experiments and directly calculate the NP content of every desorption fraction by using sorption data to assess the ecological risk of NP release when BCs are used for organic pollution remediation. The most important conclusion was these findings reminded us that the theoretical relation of sorption domains and desorption sites in natural sediment was not suitable for BCamended sediment.
4.2. Mechanism between the sorption and desorption of NP on BCsediment systems Fig. 4 showed that the desorbing fraction (Frap, Frap þ Fslow) and resistant desorption fraction (Fr) were significantly correlated to Kom and Qmax, respectively. However, the relation between sorption domains and desorption sites in sediment-amended BC was unknown. On the basis of these results, we compared the desorbing fraction (Frap, Frap þ Fslow) with the partition to total sorption ratio (Kom/Kf) of NP on the RC-sediment system and the FC-sediment system, respectively (Fig. 5). Kf, the sorption coefficient of the Freundlich model, represents the degree of sorption strength and capacity. Kom, the partition coefficient of HOC in the partition domain, represents the degree of partitioning. Therefore, the Kom/Kf in this study represents the ratio of NP concentration in the partition domain to that in the total sorption sites. The linear equations for the correlation between Frap and Kom/Kf were y ¼ 0.503x þ 0.022 (R2 ¼ 0.921, P < 0.01) and y ¼ 0.492x 0.0053 (R2 ¼ 0.958, P < 0.01) for sediment amended with RC and sediment amended with FC, respectively, and Frap þ Fslow and Kom/Kf were represented by y ¼ 0.946x 0.030 (R2 ¼ 0.976, P < 0.01) and y ¼ 0.980x 0.0051 (R2 ¼ 0.997, P < 0.01) for the RCsediment system and FC-sediment system, respectively. The line of the RC-sediment system was close to the line of the FC-sediment system regardless of Frap or Frap þ Fslow, so the same linear equations y ¼ 0.499x þ 0.0008 (R2 ¼ 0.906, P < 0.01) and y ¼ 0.991x 0.0383 (R2 ¼ 0.978, P < 0.01) could be used for Frap and Frap þ Fslow, respectively. Furthermore, the line for Frap was lower than the line of y ¼ x, and the slope was only 0.499, suggesting that NP from the rapid desorption fraction was only half of the NP in the partition fraction, but not equal to the NP in the partition fraction. The Frap þ Fslow line was close to the y ¼ x line, which indicated that the NP concentration of the desorbing fraction was almost equal to the NP in the partition fraction. In other words, the NP in the linear partition fraction can desorb completely, but it does not only containing rapidly desorption. Therefore, the theory that the natural sediment sites exhibiting rapid and slow desorption are most likely the same as the sites showing linear (partition) and nonlinear sorption (adsorption) was not suitable for sediment-amended BC, which would overestimate the HOC release. At the same time, (1 (Frap þ Fslow)) should have the same quantitative correlation with (1 Kom/Kf). In other words, the NP in the resistant desorbing fraction was the same as that in the nonlinear adsorption fraction (dominated by BC). When combined with Section 4.1, the released and residual NP from/in BCsediment system can be obtained from sorption data.
106
L. Liping et al. / Environmental Pollution 190 (2014) 101e108
Table 2 Parameters of desorption kinetics of NP from sediment-BC systems fitted by MM. Sorbents
Fitting parameters Fr
S 0.5% RC 2.0% RC 5.0% RC 0.5% FC 2.0% FC 5.0% FC Pure RC Pure FC
Frap
0.0403 0.242 0.424 0.579 0.0793 0.179 0.371 0.836 0.829
0.012 0.013 0.013 0.0076 0.014 0.022 0.013 0.0042 0.0033
0.485 0.427 0.279 0.242 0.447 0.491 0.376 0.0858 0.0831
0.043 0.075 0.061 0.061 0.057 0.090 0.060 0.020 0.024
krap (/h)
Fslow
2.586 0.262 0.203 0.174 2.872 0.365 0.245 0.307 0.272
0.474 0.311 0.275 0.167 0.473 0.290 0.234 0.0712 0.0819
0.035 0.074 0.059 0.060 0.049 0.088 0.059 0.020 0.024
kslow (103/h)
R2
95.987 34.132 23.975 30.077 120.901 33.959 28.258 34.096 39.851
0.994 0.995 0.993 0.995 0.991 0.986 0.994 0.988 0.993
Table 3 Surface areas and pore structures of BC and BC amended sediment. Property 2
BET surface area (m /g) Pore volume (mL/g)
S
0.5% RC
2.0% RC
5.0% RC
0.5% FC
2.0% FC
5.0% FC
Pure RC
Pure FC
8.73 0.035
10.65 0.040
13.16 0.048
20.46 0.057
9.10 0.046
9.16 0.046
9.25 0.047
72.10 0.133
21.00 0.039
Fig. 4. Correlation between parameters of sorption isotherms and parameters of desorption kinetics.
Sediment amended with RC and sediment amended with FC 1.0 Sediment amended with RC Sediment amended with FC y=x 2 y=0.991x-0.0383; R =0.978; P
