*Manuscript
1
1 Sorption Kinetics and Equilibrium of the Herbicide Diuron to
2 Carbon Nanotubes or Soot in Absence and Presence of Algae
3
4 Fabienne Schwab,1,2 Louise Camenzuli,2 Katja Knauer,3 Bernd Nowack,1 Arnaud Magrez,4 Laura
5 Sigg5, Thomas D. Bucheli2*
6 1) Empa – Swiss Federal Laboratories for Materials Science and Technology, CH-9014 St. Gallen,
7 Switzerland
8 2) Agroscope, Institute for Sustainability Sciences ISS, CH-8046 Zurich, Switzerland
9 3) Federal Office for Agriculture FOAG, CH-3003 Bern, Switzerland
10 4) EPFL – Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland
11 5) Eawag, Swiss Federal Institute of Aquatic Science and Technology, CH-8600 Duebendorf,
12 Switzerland
13
14 * Corresponding author: Tel: ++41 443777342, Fax: ++41 443777201,
16
17
18
19
20 Word count: Text + Abstract 4963 (without references, words in table, and figure captions), 3 small
21 figures, 1 large figure, 1 large table.
Graphical Abstract (for review)
Highlights (for review)
We quantified the sorption of an herbicide to CNT in presence of green algae. The herbicide diuron readily and strongly sorbed to CNT. After adding algae, 15-20% of the bound diuron re-dissolved from purified CNT. Unpurified, industrial grade CNT did not release diuron and behaved like soot. Amorphous carbon impurities were the most likely explanation for this behavior.
2
22 Abstract
23 Carbon nanotubes (CNT) are strong sorbents for organic micropollutants, and may change
24 their distribution and bioavailability. To quantify the rarely recognized influence of biota to
25 pollutant-CNT sorption, this work investigated the effect of green algae (Chlorella vulgaris) on
26 sorption of a model pollutant (diuron, synonyms: 3-(3,4-Dichlorophenyl)-1,1-dimethylurea, DCMU)
27 to CNT (multi-walled purified, industrial grade, pristine, and oxidized; reference material: Diesel
28 soot). The sorption of diuron to CNT in absence of algae was fast, strong, and nonlinear (Freundlich
5.79 6.24 -n 29 coefficients: 10 -10 g/kgCNT·( g/L) and 0.62-0.70 for KF and n, respectively). After addition
30 of green algae to equilibrated diuron-CNT mixtures, 15-20% (median) of the sorbed diuron re-
31 dissolved. Desorption kinetics from purified pristine CNT was fast (31-41% within ~4 h),
32 irrespective of the presence of algae. The results suggest that diuron binds readily, but partially
33 reversibly to CNT, because competition of green algae and diuron for sorption sites in algae-diuron-
34 CNT aggregates led to re-dissolution.
35 Capsule
36 The herbicide diuron readily and strongly sorbed to CNT, but 15-20% of the bound diuron
37 re-dissolved after addition of algae.
38
39 Keywords
40 bioavailability 41 sorption kinetics 42 black carbon 43 engineered nanomaterials 44 nanoparticles
45
3
46 Introduction
47 Carbon nanotubes (CNT), a nano-scaled allotrope of carbon, can strongly sorb organic
48 pollutants (OP) with a similar or higher affinity than other carbonaceous sorbents, depending on
49 their high specific surface area (SSA), and their large delocalized electron systems (Pan and Xing,
50 2010b). Once released in the environment (Hendren et al., 2011), CNT may therefore change the fate
51 and bioavailability of OP (Hofmann and von der Kammer, 2009; Pan and Xing, 2010b).
52 Numerous different sorption mechanisms are discussed to explain and predict the fate of
53 chemicals in the presence of CNT in equilibrium (Brooks et al., 2012; Chen et al., 2011; Chen et al.,
54 2008; Pan and Xing, 2008, 2010b; Smith et al., 2009; Smith et al., 2012; Sun et al., 2012; Wang et
55 al., 2010a; Wang et al., 2010b; Wang et al., 2009; Wu et al., 2013; Yang et al., 2006). Less attention
56 has been given to the kinetics of ad-, ab-, and desorption of OP on/from CNT. Kinetics is not only
57 important for engineering or in analytical chemistry (sorbent regeneration, solid phase extraction),
58 but also to understand OP release from CNT as a secondary water contamination source (Wu et al.,
59 2013). First studies suggest that sorption seems to be fast and reversible for many OP (Chen et al.,
60 2009; Oleszczuk et al., 2009; Yan et al., 2008; Zhang et al., 2012). For instance, sulfonylurea
61 herbicides can be solid-phase extracted from water almost instantaneously, using multi-walled CNT
62 as a sorbent (Fang et al., 2010). Similar fast sorption kinetics was found for CNT and other
63 pesticides (Dong et al., 2009; Wang et al., 2007). Sorption kinetics was usually better described by
64 the empirical pseudo-second order model (PSOM, equations S1-S3) than by the empirical pseudo-
65 first order model (PFOM), or the intraparticle diffusion model (Oleszczuk et al., 2009; Pan and
66 Xing, 2010a; Zhang et al., 2012). Rate constants k2*, obtained using the PSOM, indicate that ad- or
67 absorption kinetics occurs on a time scale of hours (Chen et al., 2011; Pan and Xing, 2010a; Yan et
68 al., 2008; Zhang et al., 2012). However, some pharmaceuticals’ (Oleszczuk et al., 2009) and other
69 OP’s (Wu et al., 2013) sorption equilibrium to CNT was found to be reached only within 5 days. It is
70 currently not entirely clear why some CNT sorb OP that much slower than others. One of the first
71 studies on desorption of OP from CNT was performed by Oleszczuk et al. who showed that
72 desorption kinetics of oxytetracycline and carbamazepine from CNT followed a two-phase first
4 73 order model consisting of a rapid and a slow desorbing fraction (Oleszczuk et al., 2009). Slightly
74 slower desorption than sorption, and sorption hysteresis is proposed to be explainable by
75 rearrangement of the CNT bundles, pore condensation, or the exothermic adsorption of OP to CNT
76 (Pan and Xing, 2008). A more recent study suggested irreversible chemical reactions of the sorbate
77 with carboxylic functional groups of the sorbent (Wu et al., 2013).
78 While there is a wealth of literature on OP sorption to CNT in simple equilibrated solute-
79 sorbent systems, much less is known on sorption in more complex, environmentally more relevant
80 setups, such as in presence of natural organic matter (NOM, Pan et al., 2013), or organisms (Xia et
81 al., 2010). It was shown that the presence of competitively sorbed NOM (Shi et al., 2010; Zuttel et
82 al., 2002), or surface oxidation (Pan and Xing, 2010b; Wang et al., 2010b), may strongly reduce the
83 CNT sorption capacity, especially of aromatic or planar OP (Shi et al., 2010; Wang et al., 2010b;
84 Wang et al., 2009; Zhang et al., 2011). Then again, NOM can increase the sorption to CNT for
85 multiple reasons, such as increased available CNT surface area, or oxygen-containing functional
86 groups (Lin et al., 2012; Pan et al., 2013). Regarding organisms, a first study of Xia et al. suggested
87 that the presence of Agrobacterium probably promoted desorption of phenanthrene from CNT in
88 soil, or that the bacterium utilized sorbed phenanthrene. However, it is not clear which of the two
89 explanations led to the increased availability of the OP (Xia et al., 2010). Also, to our best
90 knowledge, information lacks on how OP sorption to CNT is affected by algae, the major primary
91 producers in aqueous environments. Any biological activity in aqueous sorption experiments is
92 usually suppressed by adding a biocide such as NaN3 (Shi et al., 2010). More information on how
93 OP interact with CNT under more environmentally relevant conditions, like in presence of aqueous
94 organisms, would help to improve our understanding of the bioavailability of OP in water. Such
95 information is needed to assess the environmental concentrations, the toxicity, and the risks of both
96 OP and CNT.
97 Hence, the aim of our study was to systematically quantify the sorption processes in a
98 pollutant-CNT-biota model system, and test the hypothesis that algae may attenuate sorption similar
99 to natural organic matter (NOM). Experiments were performed with the herbicide diuron, different
5
100 types of CNT, soot as a reference sorbent, and the freshwater green alga Chlorella vulgaris. The
101 pesticide diuron is one of the most frequently and permanently observed agrochemicals in surface
102 waters and was shown to strongly bind to different kinds of carbonaceous materials (Cornelissen et
103 al., 2005; Knauer et al., 2007), including CNT (Chen et al., 2011). Diuron is used here in a range of
104 environmentally relevant concentrations. Green algae were chosen as a representative important
105 component of the ecosystem. A freshwater environment was simulated to study the sorption
106 processes. Experiments were carried out to a) quantify sorption thermodynamics and kinetics of
107 diuron and CNT, b) compare the sorption of industrial, purified, oxidized and pristine CNT and the
108 reference sorbent Diesel soot, and c) quantify the impact of algae on the ongoing sorption processes
109 in the pollutant-CNT-biota system. Moreover, the present work provides actual aqueous herbicide
110 concentrations and mechanistic explanation for a companion study on the toxicity of the herbicide to
111 the green algae in presence of CNT (Schwab et al., 2013).
112 Experimental Section
113 Chemicals. 3-(3,4-Dichlorophenyl)-1,1-dimethylurea (diuron, DCMU) and 3-(3,4-
114 dichlorophenyl)-1,1-dimethyl(D6)urea (D6-diuron), both >99% pure, were purchased from Labor
115 Dr. Ehrenstorfer (Germany). For structures, see Table S1. Acetonitrile (99.99%) was obtained from
116 SCHARLAU CHEMIE S.A. (Spain). All other chemicals were per analysis grade and obtained from
117 Merck or Sigma-Aldrich, Switzerland. Ultra-pure water (Milli-Q) was produced by a gradient A10
118 water purification system from Millipore (Volketswil, Switzerland). Nitrogen (99.99995%) was
119 obtained from PanGas (Dagmarsellen, Switzerland).
120 Carbonaceous Nanoparticles. Purified pristine and oxidized CNT (ppCNT and poxCNT)
121 were synthesized by catalytic chemical vapor deposition by EPFL Lausanne, Switzerland (Magrez et
122 al., 2010). Industrial pristine CNT (ipCNT) of similar size and shape were purchased from Cheap
123 Tubes Inc., Brattleboro, VT 05301, USA. The average CNT diameter ranged between 8-15 nm, and
124 the length between 0.5-2.0 m. As native black carbon reference sorbent, a forklift Diesel soot
6 125 standard (soot) with spherical primary particles in the range of 35 nm (Gustafsson et al., 2001) was
126 used (Standard Reference Material 2975, National Institute of Standards and Technology, USA).
127 Transmission electron microscopy images and detailed characterization data for all materials, and
128 synthesis details are given in the supporting information (SI), and in (Schwab et al., 2011). The
129 sorbent suspensions were prepared in 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid
130 (HEPES) buffered OECD algal test medium (OECD, 2006)1.00 mM, pH 7.0±0.3).
131 Preparation details are provided in the SI.
132 Green Algae Cultivation: Chlorella vulgaris, strain 211-11b (Culture Collection of Algae,
133 SAG, University of Göttingen, Germany), was cultivated in the OECD algal test medium (OECD,
134 2006). In short, axenic cultures grew at 24±0.5 °C, 100 rpm, an illumination of 80±15 Em-2s-1
135 (daylight lamps, Sylvania Gro-Lux F15W/Gro T8, Infors, Switzerland), and under a light:dark
136 regime of 16:8 h. Algae used for tests were harvested in the exponential growth phase, and the cell
137 density was counted using a haemocytometer chamber. More details on the incubation conditions are
138 available elsewhere (OECD, 2006; Schwab et al., 2011).
139 CNT and Soot Suspensions. The used CNT and soot in the standardized suspensions and
140 their detailed behavior in the algal medium were characterized previously (Schwab et al., 2013;
141 Schwab et al., 2011). The polydispersity of the CNT suspensions was monitored using dynamic light
142 scattering (DLS) in a range of 1 to 10’000 nm (Zetasizer Nano ZS, Malvern, UK) at the same
143 experimental conditions as used for the ternary sorption experiments. Three replicates per treatment
144 were measured. Finally, sorbent-algae suspensions were examined microscopically. More details
145 about dry and suspended CNT characterization are provided in the SI.
146 Binary Sorption Experiments. All sorption isotherms with CNT or soot and diuron were
147 obtained using a batch equilibration technique at 24±2 °C. Screw cap glass vials (22 mL) were used.
148 To each vial, well homogenized sorbent stock suspension was added and diluted with algal growth
149 medium to obtain, after addition of the diuron stock solution, a final nominal concentration of
150 10.0 mgCNT/L and 5.00 mgsoot/L, respectively. The exact sorbent concentration in the stock
7 151 suspensions was calculated in each experiment from the weighed in mass of CNT or soot. Nine to
152 ten different initial diuron concentrations in a range of 0.73-2,994 g/L were prepared by spiking
153 different volumes of a diuron working stock solution to the CNT suspensions. The concentration
154 range covered more than three orders of magnitude to account for the expected nonlinear sorption
155 behavior of CNT. All concentrations were tested at least in duplicates, and control treatments at least
156 in triplicates. These test suspensions were prepared sterilely without biocide, despite of the increase
157 of variability of the data this inflicts, to avoid killing the algae in the subsequent ternary sorption
158 experiments (for details, see SI). Finally, the vials containing the sorbent-diuron mixture were sealed
159 with polytetrafluoroethylene (PTFE) lined screw caps, additionally covered with Parafilm®,
160 wrapped in aluminum foil and then pre-equilibrated on an orbital laboratory shaker. The mixtures
161 pre-equilibrated for 20 h at 175 rpm in the dark at 24±2 °C, and under equal conditions for 28 d to
162 study sorption of diuron under non-equilibrium and equilibrium conditions, respectively.
163 After pre-equilibration of the solute-sorbent mixtures, 2 mL aliquots of each vial were
164 filtered for separation from the CNT through a 0.4 m PTFE syringe filter (Millipore, Switzerland).
165 The filtrate was spiked with D6-diuron, and kept at 4 °C in darkness until analysis within 3 weeks.
166 Details on the preparation of diuron stock solutions, quantification of diuron by high performance
167 liquid chromatography-tandem mass spectrometry, and quality assurance are provided in the SI.
168 Sorbed diuron was calculated by mass balance. Five common isotherm models were evaluated for fit
169 to the data (Pikaar et al., 2006; Xia and Ball, 2000). All details on the modeling, sorption models and
170 goodness of fits are provided in the SI.
171 Ternary Sorption Experiments. To examine the change of sorption after adding living
172 green algae, 7-37×104 cells/mL of C. vulgaris were added to all the vials of the 20 h, or 28 d pre-
173 equilibrated diuron-CNT mixtures (from the binary sorption experiments). The added algae culture
174 volume was controlled to change the total volume in the vials by ≤1.8%. All vials were placed
175 vertically inside of an illuminated sterile bench (15±5 Em−2s−1) at room temperature to prevent
176 contamination of the samples and to provide light for the algae. These conditions were chosen to
8 177 keep the viability (photosynthetic activity) of the green algae constant in the control treatments
178 (Schwab et al., 2013). After incubation the diuron-CNT mixtures with green algae, the subsequent
179 changes of the diuron concentration were measured after 3, 6, 15, and 24 h (sampling alike in binary
180 sorption experiments). The robustness of this setup was verified by repeating some ternary sorption
181 experiments under different incubation conditions (in 50.0 mL Erlenmeyer flasks at 100 rpm and
182 80±15 Em-2s-1). The resulting sorbed diuron concentration by the CNT was again calculated by
183 mass balance to establish perturbed sorption isotherms. These algae-affected “sorption isotherms”
184 are denoted in this study by the term “perturbed isotherms”, where the perturbation was the
185 incubation with C. vulgaris. Details on calculation of the fraction of additionally sorbed or re-
186 dissolved diuron, and quality assurance are provided in the SI.
187 Adsorption Kinetics. Batch adsorption experiments (50.0 mL per replicate) were performed
188 in 50 mL screw-capped glass vials. Pristine purified CNT, ipCNT, and soot were tested (poxCNT
189 behaved in pre-trials identically to ppCNT). In each vial, sorbent stock suspension, algal growth
190 medium, and diuron were combined to result in final concentrations of 10 mgCNT/L (ppCNT and
191 ipCNT), 5.0 mgsoot/L, and 100 g/L diuron. After addition of diuron, all vessels were immediately
192 sealed and shaken on an orbital laboratory shaker at 24±2 °C in the dark. The dissolved diuron
193 concentration was measured in these batches for the first time one min after the addition of diuron,
194 and repeatedly thereafter until all sorbent-diuron mixtures had reached equilibrium. All time points
195 were measured at least in duplicates. Selected time points were re-measured in a second repetition of
196 the experiment to assure the reproducibility of the results. Details on the fitting of the sorption
197 kinetics are provided in the SI.
198 Desorption Kinetics. The desorption kinetics was measured using 10.0 mL of 10.0 mgCNT/L
199 ppCNT in 15.0 mL polycarbonate centrifuge tubes. First, the CNT suspension was spiked with
200 1562 g/L diuron to saturate the CNT with diuron, and then shaken until sorbent-solute equilibrium
201 was reached after 20 h (laboratory shaker, same conditions as in the binary sorption experiment).
202 Then, tubes were centrifuged at 47’000×g for 20 min, 8.50 mL of the supernatant was sampled (the
203 residual supernatant was left in the tube to prevent accidental re-suspension of the CNT pellet),
9 204 replaced by the same volume of fresh algal growth medium, and placed back on the shaker. Like
205 this, eight desorption steps were carried out (the first after 1 h, the last after 32 h).
206 To study diuron desorption in presence of algal cells, two further experiments were
207 conducted in which the supernatant was replaced slightly differently. In the first modified setup, a
208 stock suspension of C. vulgaris in the exponential growth phase was used instead of the growth
209 medium in the first desorption step, to yield a final concentration of 2.0×105 cells/mL. In the second
210 modified setup, algae were added once at a given desorption step (2–8), whereas for all the other
211 preceding and following desorption steps, the supernatant was the growth medium.
212 After each desorption step, the dissolved diuron concentration in the supernatant was
213 analyzed. The fraction of diuron desorbed from the CNT was calculated by subtracting the dissolved
214 diuron before and after desorption. The amount of diluted diuron from the residual 1.50 mL
215 supernatant that was not replaced was subtracted. Three to six replicates per time step and treatment
216 were measured. To evaluate the reproducibility of the experimental setup, the trial with CNT only
217 was repeated twice. Details on the fitting of the desorption kinetics are provided in the SI.
218 Results & Discussion
219 Sorption Kinetics. The results of the sorption kinetics experiments of ppCNT, ipCNT and
220 soot are presented in Figure 1. In agreement with literature (Chen et al., 2011; Zhang et al., 2012),
2 221 the PSOM, both non-linearized and linearized, fitted well for CNT (0.991 2 222 soot (Radj =0.968, Figure 1, Figure S3, Table S4). The variability of the soot data was caused by its 223 low concentration, which was used to avoid obscuration of the fluorescence signal of the 224 photosynthetic activity of algae in the ternary trials (photosynthetic activity measured in companion 225 study (Schwab et al., 2013), details in SI, sections “Fitting of Sorption Kinetics”, and “Sorption 226 Modeling”). 227 Sorption of diuron to the ppCNT reached equilibrium after ~4 h (modified rate constant * -1 228 k2 =11.9 h , Figure 1A, Table S4). The poxCNT were not tested in detail, but behaved in pre-trials 10 229 identically to pristine CNT (see also SI, section “Dry and Suspended CNT Characterization”). The * -1 230 ipCNT took more than ten times longer than the ppCNT to reach equilibrium (~50 h, k2 =1.1 h ). * -1 231 Sorption to soot (k2 of 0.23 h ) was slower than to ipCNT, however, this value has to be interpreted 232 carefully due to the higher uncertainty of the data. Soot took only ~20 h to fully equilibrate, and the 233 variability of the data was relatively high for reasons indicated above. Repetition of the experiment 234 did not further improve the data. The soot sorption kinetics was thus much faster than the sorption 235 kinetics of diuron to soot in Sobek et al., who found an equilibration time for soot and diuron of 236 14 d, probably due to different experimental setup (half cells separated by a dialysis membrane, 237 (Sobek et al., 2009), The sorption kinetics in the range of ppCNT>> ipCNT>soot could be explained 238 by the different accessibility of adsorption sites (e.g. aromatic sheets) for diuron. The amorphous 239 carbon covering the aromatic sheets of the low purity ipCNT and soot (Table S2) may have limited 240 fast diffusion of diuron to these sites. In agreement with that, the sorption kinetics of diuron to 241 ipCNT was ~7 times slower than that of biphenyl and phenanthrene to CNT in (Zhang et al., 2012), 242 and also slower than the one of atrazine to CNT (Chen et al., 2009). Moreover, the sorption kinetics 243 of diuron to the high purity ppCNT was comparable to the values found in the literature for CNT of 244 similar purity (Deng et al., 2012). 245 The fitted PSOM equilibrium sorption capacities cs,t=∞ of 3.3 [2.3, 4.3]95% g/kgsoot, 4.5 [4.2, 246 4.7]95% g/kgCNT for ppCNT, and 9.4 [9.0, 9.8]95% g/kgCNT for ipCNT correlated highly (Pearson 247 correlation coefficient of 0.9996, p<0.018) with the SSA of these materials (Table S2), and were 2 2 248 comparable when normalized to surface area (36.4 g/m soot, 30.5 g/m CNT for ppCNT, and 21.4 2 249 g/m CNT for ipCNT). This finding points again to a sorption mechanism that is mainly dependent on 250 the surface of the materials (see above). The obtained sorption capacities for CNT were comparable 251 to recent findings for biphenyl and phenanthrene sorbed to multi-walled CNT (Zhang et al., 2012). 11 1.2E+07 purified pristine CNT A 1.0E+07 industrial pristine CNT 8.0E+06 ) CNT 6.0E+06 (µg/kg s c 4.0E+06 2.0E+06 0.0E-01 0 10 20 30 40 50 time (h) 1.2E+07 soot 1.0E+07 B 8.0E+06 ) soot 6.0E+06 (µg/kg s c 4.0E+06 2.0E+06 0.0E-01 0 10 20 30 40 50 time (h) 252 253 Figure 1: Sorption kinetics of diuron (100 g/L) to purified pristine (●) and industrial pristine (♦) carbon nanotubes 254 (CNT) and soot (▲). The CNT concentrations were 10.0 mgCNT/L, the soot concentration was 5.0 mgsoot/L. The solid 255 lines are fits of the data to the pseudo-second order kinetics model. Note that the variability of the soot data is due to the 256 low absolute soot concentration to avoid photosynthetic inhibition of the algae in ternary trials. Pooled data from two 257 independent experiments are shown. 258 Sorption in Absence of Algae. The estimated Freundlich parameters for diuron and all four 259 materials equilibrated for 28 d are presented in Table 1 (results for 20 h in SI, Table S7 and Figure 260 S4, detailed discussion of the modeling in SI). The Freundlich constant K of soot was with F 4.55±0.13 -n 261 10 g/kgsoot·( g/L) comparable to the KF from Sobek et al. for diuron sorption to soot 4.33±0.04 -n 262 (10 g/kgsoot·( g/L) ) measured in half cell systems (Sobek et al., 2009). The small difference 263 of 5% between the two KF demonstrates that the present experimental setup was adequate. The 12 6.24±0.10 -n 264 ipCNT sorbed diuron ~0.5 orders of magnitude stronger (KF=10 g/kgCNT·( g/L) ) than 5.84±0.15 -n 5.84±0.14 -n 265 ppCNT (KF=10 g/kgCNT·( g/L) ) and poxCNT (KF=10 g/kgCNT·( g/L) ). The 266 ppCNT and poxCNT again sorbed diuron >1 order of magnitude stronger than soot. The oxidation of 267 poxCNT did, as in the kinetics experiments, not affect the sorption strength significantly: Both 268 ppCNT and poxCNT exhibited similar KF and n. The sorption was for both CNT rather nonlinear 269 (ppCNT: n=0.62±0.11, poxCNT: n=0.67±0.10) as compared to soot (n=0.97±0.08). The different KF 270 of purified and ipCNT were most likely mainly a result of the different SSA of the sorbents, and not 271 by different sorption mechanisms. This interpretation is further supported by a plot of the sorption 272 isotherms normalized by the SSA (Figure 2B), in which all isotherms of CNT collapse on one non- 273 significantly different isotherm. Hydrogen bonding and Lewis acid-base based sorption mechanisms 274 due to the carboxylic groups seem thus rather unlikely. Electrostatic interactions are also not 275 expected to occur, because diuron is neutrally charged at pH 7.00. Therefore, it seems most probable 276 that diuron bound to the CNT via - stacking on the highly condensed aromatic sheet regions, 277 present both in CNT and soot. This binding mechanism is supported by the fact that diuron is a 278 rather polar molecule (dipole moment is 7.55 Debye, (Chen et al., 2011). After surface saturation of 279 the CNT, lower-energy sorption sites for the neutrally charged diuron such as defects in the aromatic 280 sheets, interstitial and groove regions between CNT bundles, and residual amorphous carbon may be 281 occupied, resulting in the observed nonlinear sorption (Figure 2 and Figure S4, Pan and Xing, 2008, 282 2010b; Pikaar et al., 2006). 283 13 Table 1: Freundlich fitting parameters (log KF: Freundlich sorption coefficient, n: nonlinearity constant) for all non-perturbed isotherms (i.e. before addition of algae, highlighted with bold letters) and for all perturbed isotherms (3, 6, 15, and 24 h incubation with algae). As a measure of the goodness of the fit, the adjusted 2 correlation coefficient Radj is shown. All sorbent-sorbate mixtures were equilibrated for 28 d. -n log KF (µg/kgsorbent·(µg/L) ) n (-) 95% 95% rel. R 2 rel. error adj av conf. av conf. error (%) int. int. (%) purified pristine CNT before addition of algae 5.84 ± 0.15 1.3 0.62 ± 0.11 8.5 0.820 3 h incubation with algae 5.79 ± 0.13 1.1 0.65 ± 0.09 6.5 0.919 6 h incubation with algae 5.72 ± 0.18 1.5 0.65 ± 0.12 9.2 0.796 15 h incubation with algae 5.69 ± 0.11 0.9 0.66 ± 0.07 5.0 0.960 24 h incubation with algae 5.76 ± 0.21 1.7 0.61 ± 0.13 10.4 0.780 purified oxidized CNT before addition of algae 5.84 ± 0.14 1.2 0.67 ± 0.10 7.1 0.863 3 h incubation with algae 5.72 ± 0.15 1.3 0.68 ± 0.10 6.9 0.905 6 h incubation with algae 5.51 ± 0.28 2.5 0.80 ± 0.19 11.4 0.725 15 h incubation with algae 5.57 ± 0.15 1.2 0.69 ± 0.09 6.1 0.933 24 h incubation with algae 5.62 ± 0.23 2.0 0.70 ± 0.15 10.4 0.794 industrial pristine CNT before addition of algae 6.24 ± 0.10 0.7 0.64 ± 0.06 4.7 0.956 3 h incubation with algae 6.28 ± 0.11 0.9 0.59 ± 0.07 5.6 0.941 6 h incubation with algae 6.30 ± 0.09 0.7 0.59 ± 0.06 4.6 0.957 15 h incubation with algae 6.25 ± 0.08 0.6 0.62 ± 0.05 4.2 0.963 24 h incubation with algae 6.28 ± 0.09 0.7 0.59 ± 0.06 4.7 0.956 soot before addition of algae 4.55 ± 0.13 1.3 0.97 ± 0.08 3.7 0.973 3 h incubation with algae 4.18 ± 0.16 1.9 1.10 ± 0.09 3.9 0.975 6 h incubation with algae 4.32 ± 0.23 2.6 1.03 ± 0.14 6.3 0.925 15 h incubation with algae 4.34 ± 0.25 2.7 0.99 ± 0.14 6.6 0.934 24 h incubation with algae 4.33 ± 0.17 1.9 1.02 ± 0.10 4.5 0.965 284 285 14 8.5 8.0 A 7.5 ) 7.0 sorbent 6.5 6.0 (µg/kg s 5.5 c log 5.0 4.5 4.0 -1 0 1 2 3 log cw (µg/L) 3.0 2.5 B ) 2 2.0 sorbent 1.5 1.0 0.5 SSA) (µg/m SSA) / s 0.0 c -0.5 log ( -1.0 -1 0 1 2 3 log cw (µg/L) 286 287 Figure 2: Selected sorption isotherms of the herbicide diuron to purified pristine carbon nanotubes (CNT, ●), 288 purified oxidized CNT (○), industrial pristine CNT (♦), and the reference sorbent soot (▲) with Freundlich fits (for 289 fitting parameters, see Table 1), normalized to sorbent mass (A), and specific surface area (SSA, B). All sorbent-sorbate 290 mixtures were equilibrated for 28 d without addition of biocide. 291 292 Likewise, the lower sorption of diuron to soot as compared to the CNT could be explained by 293 the lower SSA of soot, the high content of oxygen (Table S2) on the soot surface, and attenuation of 294 diuron sorption due to the amorphous carbon, similar to the attenuation of the sorption by NOM 15 295 (Cornelissen and Gustafsson, 2006; Shi et al., 2010; Wang et al., 2010b; Wang et al., 2009; Zhang et 296 al., 2011). This explanation is supported by the finding of Knauer et al. (Knauer et al., 2007) who 5.7 -n 297 found for combusted (i.e. amorphous carbon free) soot a very high KF of 10 g/kgsoot·( g/L) . This 298 KF is comparable to those of ppCNT and poxCNT (Table 1), suggesting that the condensed carbon 299 sheets of combusted soot, once accessible, bind diuron in a similar way as the aromatic sheets of 300 CNT. 301 Microscopical Investigations of Algae-CNT suspensions. Brightfield microscopy revealed 302 that the majority of algal cells attached to CNT agglomerates (Table S3, Figure S5B). Mixing of the 303 suspensions enhanced this agglomeration. In agreement with previous work (Schwab et al., 2013), 304 and the results presented above, no noticeable difference was observed between the ppCNT and 305 poxCNT. The ipCNT seemed to form slightly less agglomerates with algae. In contrast, algae did 306 attach only little to soot, even in well stirred suspensions. No adverse effects on size or morphology 307 of the cells exposed to any of the tested particles was observed, as reported in previous studies 308 (Schwab et al., 2011; Wei et al., 2010). 309 Sorption Equilibrium Change in Presence of Algae and Evolution with Time. After 310 addition of algae to the equilibrated diuron and CNT, medians 15-20% (non-outlier range up to 40%) 311 of the sorbed diuron re-dissolved from the ppCNT and poxCNT during 24 h of incubation with algae 312 (Figure 3, absolute concentration changes in Figure S6). Re-dissolution occurred only in presence of 313 algae, as shown in equally treated control experiments with equilibrated diuron-CNT mixtures 314 without algae, in presence of algal exudates only, or algae without exudates (see below). Given that 315 algal cells intensively agglomerated with CNT, even if very low CNT concentrations were present in 316 suspension, and exudates clearly did not affect the sorption (see discussion below), we conclude that 317 the formation of algae-CNT agglomerates led to more competition for sorption sites, and as a 318 consequence a re-dissolution of diuron. Evidence for re-dissolution due to addition of algae is also 319 provided by increased diuron algal toxicity caused by locally elevated diuron concentrations 320 (Schwab et al., 2013). This explanation is supported by the microscopical investigations mentioned 16 321 above, which showed that ppCNT and poxCNT agglomerated most extensively with algal cells, 322 leading to highly competitive sorption, and therefore more diuron release than in ipCNT and soot. 323 The variability for soot was again higher due to the lower absolute soot concentration, but the trends 324 in Figure 3 point to no major diuron re-dissolution. Both microscopical investigations and earlier 325 work (Schwab et al., 2011) confirmed that soot aggregated only little with algae. Confirming the 326 results of the sorption experiments, ppCNT and poxCNT again behaved similarly, and ipCNT 327 behaved more like soot that did not lead to significant amounts of re-dissolved diuron. More details 328 on the temporal change of the sorption equilibrium after addition of algae are given in the SI (Table 329 S7, Figure S7, Figure S8, chapter “Change of KF with time”). 330 331 Figure 3: Re-dissolved diuron (medians and non-outlier range) after different incubation times of 28 d pre-equilibrated 332 diuron-sorbent mixtures with algae for all four sorbents under investigation. Nonparametric comparison of multiple 17 333 independent groups (Kruskal-Wallis, N=84-99), p=0.0002-0.04 for 15 h, p=0.000002-0.02 for 24 h. *: Significantly 334 different from one group. **: Significantly different from two groups. 335 No Effect of Algal Exudates. Diuron uptake/sorption by algae, or algal exudates was in 336 additional control experiments found to be too small to be quantifiable, presumably due to the more 337 than two orders of magnitude stronger sorption of diuron to CNT than to algae/exudates, the relative 338 short duration of the incubation of the mixture with algae (24 h), and the >11 times lower 339 concentration of algae/exudates than CNT/soot. A theoretical worst-case calculation of the maximal 340 percentage of sorbed diuron by algal exudates resulted in a maximal concentration of 0.13% 341 (assuming the algae produced their own weight in exudates with a worst-case assumption for a 342 log(KOC) for diuron of 3.5, (Schwarzenbach et al., 2003). In the present setup, using algae densities 343 normally found in environment, the sorption to exudates turns out to be negligible. This calculation 344 also shows that increasing the algal biomass by a factor of 10-1’000, as it is possible during massive 345 algal blooms, would elevate the maximal percentage of diuron sorbed by exudates to 1.2-55.7%. 346 Results at such high algal densities are more difficult to interpret, because at densities of several 347 million cells/mL, the algae start to inhibit their own growth. 348 Desorption Kinetics in Absence of Algae. Diuron desorption occurred within ~4 hours 349 (Figure 4A). The obtained data fitted well to the two-phase desorption kinetics model (Equation S7). 350 A fraction of 39.6±1.5% (Frapid,±standard error) of the diuron rapidly desorbed from the ppCNT, and 351 Fslow was consequently 60.4±1.5%. This implies that, if equilibrium conditions change, for example, 352 if a solid phase extraction cartridge leaches CNT, or a CNT-containing effluent gets diluted, a 353 considerable fraction of diuron sorbed to CNT may again be released to the aqueous environment on 354 a relevant time scale. However, it should be noted that in the present desorption kinetics 355 experimental setup, the used CNT were saturated with diuron. The observed 39.6±1.5% is therefore 356 the maximal expectable rapid desorbing fraction of diuron sorbed to CNT. Under environmental 357 conditions as in the above shown ternary sorption experiments, CNT are not saturated with diuron - 358 and smaller fractions of diuron therefore release. The rate constants krapid and kslow were 0.66±0.07 h 1, -1 359 and -0.002±0.002 h , respectively, revealing that compared to the k2* of the sorption kinetics (see 18 360 above), the desorption process of diuron from CNT was slower. Compared to literature data, the 361 desorption kinetics of diuron from CNT is still ~4-15 times faster than e.g. the desorption of 362 carbamazepine and oxytetracycline from similar CNT (Oleszczuk et al., 2009). The observed 363 desorption hysteresis is in accordance with results of other studies for compounds of similar 364 structure, polarity or size (Oleszczuk et al., 2009; Pyrzynska et al., 2007). 365 366 367 Figure 4: Fraction of desorbed diuron from purified pristine carbon nanotubes (ppCNT) as a function of time. Fits: Two- 368 phase desorption kinetics model. A: Desorption from CNT alone (●), pooled data from two independent experiments. B: 369 Desorption from CNT in presence of algae added before the first desorption step (○), and added once at a given 370 desorption step (♦). Note that the match of the solid line in B with some of the ○ symbols is coincidence. 19 371 Desorption Kinetics in Presence of Algae. In Figure 4B, the desorption kinetics of diuron 372 from CNT after addition of algae before the first desorption step, and added once at a given 373 desorption step (2–8) is shown. The diuron desorption kinetics in presence of algae was almost -1 -1 374 identical to the one in absence of algae (krapid= 0.7±0.2 h , kslow=-0.002±0.004 h ). With 41±4%, the 375 fraction of rapidly desorbing diuron was also not significantly different from the fraction of diuron 376 desorbed in absence of algae. This shows that under non-equilibrium conditions, the effect of the 377 presence of algae was overridden by the tendency of the system to re-establish sorption equilibrium. 378 The measurement uncertainty was slightly higher, mainly due to a replicate sample in which >60% 379 diuron desorbed (Figure 4B). Outlier tests and repeated measurements of this sample showed that it 380 could not be excluded from the evaluation. We therefore conclude that the variability of desorption 381 kinetics in presence of algae ranges in the order of 20-30%. 382 Notably, the influence of algae on the slow desorbing diuron fraction was positive: ~25% less 383 diuron desorbed in the experiment in which algae were added once at a given desorption step (2–8) -1 384 to the CNT. The fraction of rapidly desorbing diuron here was 31.3±1.4%, krapid was 0.61±0.07 h , -1 385 and kslow was -0.002±0.002 h . Since the overall confidence bands of the two curves are not 386 significantly different, this result should not be over-interpreted. 387 Conclusions 388 Overall, the findings of this study show that changing environmental conditions, such as 389 algal blooms, or effluent dilution, may affect the sorption equilibrium of pollutants bound to CNT 390 significantly. This renders the behavior of pollutant-CNT mixtures in environment less predictable. 391 In this work, the presence of algae led to moderate 15-20% (maximally 40%) of re-dissolution of 392 sorbed micropollutant (represented here by diuron), despite of the strong sorbent properties of CNTs. 393 Notably, industrial grade CNT showed higher sorption capacities and released almost no diuron in 394 contrast to purified CNT, most likely due to their high amorphous carbon content and surface area. 395 Upon dilution of the aqueous phase, ~31-41% of the diuron desorbed within hours from diuron- 396 saturated CNT, which increases the environmental exposure of the herbicide. Without dilution, the 20 397 major part of diuron (80-85%) remains strongly sorbed to the CNT with high KF, which greatly 398 reduces environmental exposure of the herbicide. A common hypothesis is that this part of diuron 399 should further be unavailable for aquatic organisms. We tested this hypothesis in a companion study 400 on the viability of algae in presence of diuron-CNT mixtures (Schwab et al., 2013). 401 Acknowledgment 402 We thank the Swiss National Science Foundation (project no. 200021-118028, 205321- 403 125299/1) for the funding of the study and the Federal Office for the Environment for additional 404 financial support. The group of László Forró is acknowledged for the synthesis and characterization 405 of the dry CNT. Further, we gratefully acknowledge the research group of Hans-Ruedi Forrer (ART) 406 who made their laboratory available for us, and Nicola Schäppi for assistance with the TOC graph. 407 Supporting Information Available 408 Supplementary data related to this article can be found at… 409 Literature Cited 410 Brooks, A.J., Lim, H.-n., Kilduff, J.E., 2012. Adsorption uptake of synthetic organic chemicals by 411 carbon nanotubes and activated carbons. Nanotechnology 23, 294008. 412 Chen, G.C., Shan, X.Q., Pei, Z.G., Wang, H.H., Zheng, L.R., Zhang, J., Xie, Y.N., 2011. Adsorption 413 of diuron and dichlobenil on multiwalled carbon nanotubes as affected by lead. J. Hazard. Mater. 414 188, 156-163. 415 Chen, G.C., Shan, X.Q., Zhou, Y.Q., Shen, X.E., Huang, H.L., Khan, S.U., 2009. Adsorption 416 kinetics, isotherms and thermodynamics of atrazine on surface oxidized multiwalled carbon 417 nanotubes. J. Hazard. Mater. 169, 912-918. 418 Chen, J.Y., Chen, W., Zhu, D., 2008. Adsorption of nonionic aromatic compounds to single-walled 419 carbon nanotubes: Effects of aqueous solution chemistry. Environ. Sci. Technol. 42, 7225-7230. 420 Cornelissen, G., Gustafsson, O., 2006. Effects of added PAHs and precipitated humic acid coatings 421 on phenanthrene sorption to environmental Black carbon. Environ. Pollut. 141, 526-531. 21 422 Cornelissen, G., Haftka, J., Parsons, J., Gustafsson, O., 2005. Sorption to black carbon of organic 423 compounds with varying polarity and planarity. Environ. Sci. Technol. 39, 3688-3694. 424 Deng, J., Shao, Y.S., Gao, N.Y., Deng, Y., Tan, C.Q., Zhou, S.Q., Hu, X.H., 2012. Multiwalled 425 carbon nanotubes as adsorbents for removal of herbicide diuron from aqueous solution. Chem. Eng. 426 J. 193, 339-347. 427 Dong, M.F., Ma, Y.Q., Zhao, E.C., Qian, C.F., Han, L.J., Jiang, S.R., 2009. Using multiwalled 428 carbon nanotubes as solid phase extraction adsorbents for determination of chloroacetanilide 429 herbicides in water. Microchim. Acta 165, 123-128. 430 Fang, G.Z., Wang, X.N., Wang, S., 2010. Multiwalled Carbon Nanotubes as SPE Adsorbents for 431 Simultaneous Determination of Seven Sulfonylurea Herbicides in Environmental Water by LC-MS- 432 MS. Chromatographia 72, 403-409. 433 Gustafsson, O., Bucheli, T.D., Kukulska, Z., Andersson, M., Largeau, C., Rouzaud, J.N., Reddy, 434 C.M., Eglinton, T.I., 2001. Evaluation of a protocol for the quantification of black carbon in 435 sediments. Global Biogeochem. Cycles 15, 881-890. 436 Hendren, C.O., Mesnard, X., Droge, J., Wiesner, M.R., 2011. Estimating production data for five 437 engineered nanomaterials as a basis for exposure assessment. Environ. Sci. Technol. 45, 2562-2569. 438 Hofmann, T., von der Kammer, F., 2009. Estimating the relevance of engineered carbonaceous 439 nanoparticle facilitated transport of hydrophobic organic contaminants in porous media. Environ. 440 Pollut. 157, 1117-1126. 441 Knauer, K., Sobek, A., Bucheli, T.D., 2007. Reduced toxicity of diuron to the freshwater green alga 442 Pseudokirchneriella subcapitata in the presence of black carbon. Aquat. Toxicol. 83, 143-148. 443 Lin, D., Tian, X., Li, T., Zhang, Z., He, X., Xing, B., 2012. Surface-bound humic acid increased 444 Pb2+ sorption on carbon nanotubes. Environ Pollut 167, 138-147. 445 Magrez, A., Seo, J.W., Smajda, R., Korbely, B., Andresen, J.C., Mionić, M., Casimirius, S.p., Forró, 446 L.s., 2010. Low-Temperature, Highly Efficient Growth of Carbon Nanotubes on Functional 447 Materials by an Oxidative Dehydrogenation Reaction. Acs Nano 4, 3702-3708. 22 448 OECD, 2006. Freshwater alga and cyanobacteria, growth inhibition test. Organisation for Economic 449 Co-operation and Development (OECD): Paris, France. 450 Oleszczuk, P., Pan, B., Xing, B.S., 2009. Adsorption and Desorption of Oxytetracycline and 451 Carbamazepine by Multiwalled Carbon Nanotubes. Environ. Sci. Technol. 43, 9167-9173. 452 Pan, B., Xing, B.S., 2008. Adsorption Mechanisms of Organic Chemicals on Carbon Nanotubes. 453 Environ. Sci. Technol. 42, 9005-9013. 454 Pan, B., Xing, B.S., 2010a. Adsorption kinetics of 17 alpha-ethinyl estradiol and bisphenol A on 455 carbon nanomaterials. I. Several concerns regarding pseudo-first order and pseudo-second order 456 models. J. Soils Sediments 10, 838-844. 457 Pan, B., Xing, B.S., 2010b. Manufactured Nanoparticles and their Sorption of Organic Chemicals. 458 Adv. Agron. 108, 137-181. 459 Pan, B., Zhang, D., Li, H., Wu, M., Wang, Z., Xing, B., 2013. Increased Adsorption of 460 Sulfamethoxazole on Suspended Carbon Nanotubes by Dissolved Humic Acid. Environ. Sci. 461 Technol. 47, 7722-7728. 462 Pikaar, I., Koelmans, A.A., van Noort, P.C.M., 2006. Sorption of organic compounds to activated 463 carbons. Evaluation of isotherm models. Chemosphere 65, 2343-2351. 464 Pyrzynska, K., Stafiej, A., Biesaga, M., 2007. Sorption behavior of acidic herbicides on carbon 465 nanotubes. Microchim. Acta 159, 293-298. 466 Schwab, F., Bucheli, T.D., Camenzuli, L., Magrez, A., Knauer, K., Sigg, L., Nowack, B., 2013. 467 Diuron Sorbed to Carbon Nanotubes Exhibits Enhanced Toxicity to Chlorella vulgaris. Environ. Sci. 468 Technol. 47, 7012–7019. 469 Schwab, F., Bucheli, T.D., Lukhele, L., Magrez, A., Nowack, B., Sigg, L., Knauer, K., 2011. Are 470 carbon nanotube effects on green algae caused by shading and agglomeration? Environ. Sci. 471 Technol. 45, 6136-6144. 472 Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry, 473 2nd ed. John Wiley & Sons, Inc., Hoboken, New Jersey. 23 474 Shi, B.Y., Zhuang, X.Y., Yan, X.M., Lu, J.J., Tang, H.X., 2010. Adsorption of atrazine by natural 475 organic matter and surfactant dispersed carbon nanotubes. J. Environ. Sci. (Beijing, China) 22, 1195- 476 1202. 477 Smith, B., Wepasnick, K., Schrote, K.E., Cho, H.H., Ball, W.P., Fairbrother, D.H., 2009. Influence 478 of Surface Oxides on the Colloidal Stability of Multi-Walled Carbon Nanotubes: A Structure- 479 Property Relationship. Langmuir 25, 9767-9776. 480 Smith, B., Yang, J., Bitter, J.L., Ball, W.P., Fairbrother, D.H., 2012. Influence of Surface Oxygen on 481 the Interactions of Carbon Nanotubes with Natural Organic Matter. Environ. Sci. Technol. 46, 482 12839-12847. 483 Sobek, A., Stamm, N., Bucheli, T.D., 2009. Sorption of Phenyl Urea Herbicides to Black Carbon. 484 Environ. Sci. Technol. 43, 8147-8152. 485 Sun, K., Zhang, Z., Gao, B., Wang, Z., Xu, D., Jin, J., Liu, X., 2012. Adsorption of diuron, fluridone 486 and norflurazon on single-walled and multi-walled carbon nanotubes. Science Of The Total 487 Environment 439, 1-7. 488 Wang, L.L., Zhu, D.Q., Duan, L., Chen, W., 2010a. Adsorption of single-ringed N- and S- 489 heterocyclic aromatics on carbon nanotubes. Carbon 48, 3906-3915. 490 Wang, S., Zhao, P., Min, G., Fang, G.Z., 2007. Multi-residue determination of pesticides in water 491 using multi-walled carbon nanotubes solid-phase extraction and gas chromatography-mass 492 spectrometry. J. Chromatogr., A 1165, 166-171. 493 Wang, X.L., Liu, Y., Tao, S., Xing, B.S., 2010b. Relative importance of multiple mechanisms in 494 sorption of organic compounds by multiwalled carbon nanotubes. Carbon 48, 3721-3728. 495 Wang, X.L., Tao, S., Xing, B.S., 2009. Sorption and Competition of Aromatic Compounds and 496 Humic Acid on Multiwalled Carbon Nanotubes. Environ. Sci. Technol. 43, 6214-6219. 497 Wei, L.P., Thakkar, M., Chen, Y.H., Ntim, S.A., Mitra, S., Zhang, X.Y., 2010. Cytotoxicity effects 498 of water dispersible oxidized multiwalled carbon nanotubes on marine alga, Dunaliella tertiolecta. 499 Aquat. Toxicol. 100, 194-201. 24 500 Wu, W., Jiang, W., Zhang, W., Lin, D., Yang, K., 2013. Influence of Functional Groups on 501 Desorption of Organic Compounds from Carbon Nanotubes into Water: An Insight into the 502 Desorption Hysteresis. Environ. Sci. Technol. 47, 8373-8382. 503 Xia, G.S., Ball, W.P., 2000. Polanyi-based models for the competitive sorption of low-polarity 504 organic contaminants on a natural sorbent. Environ. Sci. Technol. 34, 1246-1253. 505 Xia, X.H., Li, Y.R., Zhou, Z.I., Feng, C.L., 2010. Bioavailability of adsorbed phenanthrene by black 506 carbon and multi-walled carbon nanotubes to Agrobacterium. Chemosphere 78, 1329-1336. 507 Yan, X.M., Shi, B.Y., Lu, J.J., Feng, C.H., Wang, D.S., Tang, H.X., 2008. Adsorption and 508 desorption of atrazine on carbon nanotubes. J. Colloid Interface Sci. 321, 30-38. 509 Yang, K., Zhu, L.Z., Xing, B.S., 2006. Adsorption of polycyclic aromatic hydrocarbons by carbon 510 nanomaterials. Environ. Sci. Technol. 40, 1855-1861. 511 Zhang, S.J., Shao, T., Karanfil, T., 2011. The effects of dissolved natural organic matter on the 512 adsorption of synthetic organic chemicals by activated carbons and carbon nanotubes. Water Res. 513 45, 1378-1386. 514 Zhang, S.J., Shao, T., Kose, H.S., Karanfil, T., 2012. Adsorption kinetics of aromatic compounds on 515 carbon nanotubes and activated carbons. Environ. Toxicol. Chem. 31, 79-85. 516 Zuttel, A., Nutzenadel, C., Sudan, P., Mauron, P., Emmenegger, C., Rentsch, S., Schlapbach, L., 517 Weidenkaff, A., Kiyobayashi, T., 2002. Hydrogen sorption by carbon nanotubes and other carbon 518 nanostructures. J. Alloys Compd. 330, 676-682. 519 520 521 25 522 Figure Captions 523 Figure 1: Sorption kinetics of diuron (100 g/L) to purified pristine (●) and industrial pristine (♦) 524 carbon nanotubes (CNT) and soot (▲). The CNT concentrations were 10.0 mgCNT/L, the soot 525 concentration was 5.0 mgsoot/L. The solid lines are fits of the data to the pseudo-second order 526 kinetics model. Note that the variability of the soot data is due to the low absolute soot concentration 527 to avoid photosynthetic inhibition of the algae in ternary trials. Pooled data from two independent 528 experiments are shown. 529 Figure 2: Selected sorption isotherms of the herbicide diuron to purified pristine carbon nanotubes 530 (CNT, ●), purified oxidized CNT (○), industrial pristine CNT (♦), and the reference sorbent 531 soot (▲) with Freundlich fits (for fitting parameters, see Table 1), normalized to sorbent mass (A), 532 and specific surface area (SSA, B). All sorbent-sorbate mixtures were equilibrated for 28 d without 533 addition of biocide. 534 Figure 3: Re-dissolved diuron (medians and non-outlier range) after different incubation times of 28 535 d pre-equilibrated diuron-sorbent mixtures with algae for all four sorbents under investigation. 536 Nonparametric comparison of multiple independent groups (Kruskal-Wallis, N=84-99), p=0.0002- 537 0.04 for 15 h, p=0.000002-0.02 for 24 h. *: Significantly different from one group. **: Significantly 538 different from two groups. 539 Figure 4: Fraction of desorbed diuron from purified pristine carbon nanotubes (ppCNT) as a 540 function of time. Fits: Two-phase desorption kinetics model. A: Desorption from CNT alone (●), 541 pooled data from two independent experiments. B: Desorption from CNT in presence of algae added 542 before the first desorption step (○), and added once at a given desorption step (♦). Note that the 543 match of the solid line in B with some of the ○ symbols is coincidence. 544 Supplementary Material Click here to download Supplementary Material: Schwab et al. SI submitted.docx