Saxitoxin (STX) is the most toxic non-protein substance known. STX-producing cyanobacteria have been identified in most continents and are being detected more widely because of global warming, threatening human drinking water supplies worldwide. Removal of these components can be accomplished by adsorption on granular-activated carbon (GAC) but little is yet known about the kinetics of the process. This research investigated adsorption kinetics and diffusion behaviour of a decarbomoyl saxitoxin (dc-STX) and a carbamate toxin (STX) on four coconut shell-based GAC samples with different pore size distribution. It was observed that equilibrium concentration was reached within 24 h and that a pseudo-second-order model best represented experimental data. Of the four GAC samples tested, the example with the largest volume of mesopores adsorbed more STX and with a faster upload rate, while dc-STX was adsorbed equally in all four GAC samples.

INTRODUCTION

Saxitoxins (STXs), also known as paralytic shellfish poisons (PSPs), have been associated with multiple human intoxications through seafood consumption, resulting in numbness, paralysis and death (Kuiper-Goodman et al. 1999). Besides poisoning by ingestion of food, humans may be exposed to cyanotoxins through the ingestion of contaminated drinking water and accidental ingestion, inhalation or dermal adsorption during recreational activities in waters affected by a toxic bloom (Merel et al. 2013). PSPs can be divided into three groups, the carbamates, the sulfamates and the decarbamoyl toxins (Figure 1 and Table 1). The molecular structure of STX has several amine groups that can gain protons and become cationic depending on the solution pH. As a result, up to 10 different species of STX exist (Hilal et al. 1995).

Table 1

General structure, corresponding R functional groups, and relative toxicity for the most common PSP variants

 R1 R2 R3 Relative toxicity 
R4 = CONH2 (carbamate toxins) 
STX 
Neo-STX OH 0.924 
GTX1 OH OSO3 0.994 
GTX2 OSO3 0.359 
GTX3 OSO3 0.638 
GTX4 OH OSO3 0.726 
R4 = CONHSO3 (N-sulfocarbamoyl (sulfamate) toxins) 
C1 OSO3 0.006 
C2 OSO3 0.096 
R4 = H (decarbamoyl toxins) 
dc-STX 0.513 
dc-neo-STX OH – 
dc-GTX1 OH OSO3 – 
dc-GTX2 OSO3 0.651 
dc-GTX3 OSO3 0.754 
dc-GTX4 OH OSO3 – 
 R1 R2 R3 Relative toxicity 
R4 = CONH2 (carbamate toxins) 
STX 
Neo-STX OH 0.924 
GTX1 OH OSO3 0.994 
GTX2 OSO3 0.359 
GTX3 OSO3 0.638 
GTX4 OH OSO3 0.726 
R4 = CONHSO3 (N-sulfocarbamoyl (sulfamate) toxins) 
C1 OSO3 0.006 
C2 OSO3 0.096 
R4 = H (decarbamoyl toxins) 
dc-STX 0.513 
dc-neo-STX OH – 
dc-GTX1 OH OSO3 – 
dc-GTX2 OSO3 0.651 
dc-GTX3 OSO3 0.754 
dc-GTX4 OH OSO3 – 

Source: Adapted from Ho et al. (2009).

Figure 1

Structural variations and characteristics of the STX class of cyanotoxins (Ho et al. 2009).

Figure 1

Structural variations and characteristics of the STX class of cyanotoxins (Ho et al. 2009).

Paerl & Paul (2012) suggested that because of climate change, the frequency and intensity of harmful cyanobacteria blooms are expected to increase globally. These blooms can directly influence the quality of drinking water through the production of toxic metabolites. PSP-producing cyanobacteria have been increasingly reported in fresh and brackish water in many countries worldwide (Humpage et al. 1994; Kaas & Henriksen 2000; Sevcik et al. 2003; Liu et al. 2006; Ballot et al. 2010; Berry & Lind 2010; Clemente et al. 2010). According to Shi et al. (2012), the concentrations of extracellular STXs in these countries range up to 15 μg STXequiv L–1 in natural waters, with intracellular gravimetric STXs concentrations ranging from 5 to 3,400 μg STXequiv.g–1 of cell dry weight.

Australia adopted a PSP health alert level of 3 μg STXequiv.L–1 for drinking water (NWQMS 2004), while a maximum PSP concentration of 1–3 μg STXequiv. L–1 is currently required by the New Zealand and Brazilian Ministry of Health, respectively (Orr et al. 2004; Portaria 2914 2011).

Because conventional water treatment plants (CWTPs) cannot remove extracellular cyanotoxins and other metabolites, one of the last barriers available to prevent these substances from reaching households are granular-activated carbon (GAC) filters. However, there is no accurate way to predict how effectively a full-scale GAC filter will remove compounds such as STXs (Ho & Newcombe 2010). The equilibrium concentration of dissolved toxins is the maximum reduction level that can be achieved. However, a sufficient contact time has to be ensured to achieve equilibrium. Insufficient contact times may result in non-equilibrium conditions and reduce theoretical removal capacity. Therefore, adsorption kinetics is an important factor, particularly when working with slow adsorbing compounds (Fuerhacker et al. 2001).

Activated carbon surface charge and pore structure are important parameters for removal of organic compounds in water treatment (Li et al. 2002; Quinlivan et al. 2005; Zhang et al. 2010). Non-electrostatic interaction is the most common adsorption mechanism for apolar organic chemicals on activated carbon. On the other hand, organic compound with a cationic or anionic character may be poorly adsorbed via non-specific physisorption, while electrostatic mechanisms are enhanced (Shi et al. 2012). These authors showed that water pH has a large impact on the adsorptive efficiency of powdered-activated carbon (PAC) for STX removal, not only because it will determine STX charge, but also carbon surface charge.

Zhang et al. (2011) demonstrated that natural organic matter (NOM) may decrease removal of synthetic organic chemicals on activated carbon by competing for adsorption sites, and thereby reducing the surface area available, and by blockage or diffusional hindrance of pores. They also found that the minimum NOM hindrance effect occurred on the surface of the carbon dominated by mesopores, while severe NOM effects were observed primarily on the microporous carbons.

A few researchers have studied adsorption of STXs on activated carbons in aqueous solutions (Bailey et al. 1999; Newcombe & Nicholson 2002; Orr et al. 2004; Silva 2005; Ho et al. 2009; Shi et al. 2012). Shi et al. (2012) showed that STX can be effectively controlled at a common pre-disinfection pH of 8.2 with a PAC dose between 10 and 20 mg L–1. Silva (2005) demonstrated that PAC presented high removals of neo-STX and STX, decreasing concentrations by as much as 28 μg L–1. Orr et al. (2004) used GAC packed columns with an empty bed contact time of 15 min and removed 100% of decarbomoyl saxitoxin (dc-STX), STX, GTX-2/3 and GTX-5, 94% of dc-GTX-2/3, but only partially reduced N-sulfocarbamoyl-gonyautoxins 2 & 3: C1 & C2 toxins by 56% and 74%, respectively. Ho et al. (2009) developed experiments with PAC using two Australian raw waters. Happy Valley water had initial STXequiv. concentrations between 3.7 and 4.1 μg L–1 and Myponga between 7.5 and 9.9 μg L–1. In both cases they were able to decrease concentrations below 3.0 μg L–1 with a PAC dose of 10 mg L–1 (CT = 15 min) and 30 mg L–1 (CT = 70 min), respectively. These studies, however, neither presented appropriate kinetic modelling of adsorption data nor evaluated them in the context of carbon characterisation, especially pore size distribution (PSD) and surface charge, which makes it difficult to extend their findings to further GAC investigations or to real treatment situations. The hypothesis tested was PSD can effectively influence the amount and the rate of the removal of STX and dc-STX from water. Therefore, the aim of this study was to investigate the influence of PSDs in the adsorption of STX and dc-STX on four commercially available coconut shell GAC samples, all with a net positive surface charge, to determine the kinetic data and to provide elements for further modelling studies and real treatment design.

MATERIALS AND METHODS

STX production and semi-purification

The strain Cylindrospermopsis raciborskii T3–CR (Lagos et al. 1999) from the collection of the Federal University of Rio de Janeiro, Brazil, was used to produce STXs for the adsorption experiments. CR cultures were grown in ASM-1 medium (Gorham et al. 1964), with a pH of 8.0, under a white light intensity of 75 μ mol m–2 s–1, at a temperature of 24 ± 2 °C, with aeration and a 12:12 light/dark photoperiod until they reached the end of the lag period. Although Carneiro et al. (2009) demonstrated that this strain can produce STX, neo-STX, dc-STX and dc-neo-STX, high performance liquid chromatography (HPLC) analysis demonstrated that our culture produced only STX and dc-STX in a sufficient amount to be used in the adsorption experiments.

To perform the extraction of intracellular STXs, the culture's biomass was concentrated by centrifugation at 2,700 G for 15 min (25 °C), the supernatant was discarded, and the pellets were collected. The pelleted material was subjected to three freeze–thaw cycles, filtered through a 0.45 μm nitrocellulose membrane (Macherey-Nagel, Bethlehem, PA, USA) and the filtrate was then subjected to a semi-purification step by solid phase extraction on C18 cartridges (Supelco, Bellefonte, PA, USA), according to Lawrence et al. (2005). The semi-purified extract was acidified with 0.1 M acetic acid (Merck, Darmstadt, Germany) to a pH of approximately 4.0 and stored at –20 °C to preserve STX stability (Indrasena & Gill 2000). It is important to note that this semi-purified extract contained not only dc-STX and STX, but other intracellular metabolites considered here as dissolved organic carbon (DOC). We believe that this approach more realistically represents the behaviour of the adsorption process since, in the case of an algal bloom, other dissolved intracellular compounds, apart from STXs, would be proportionally present in the water matrix and would not be removed by the CWTP.

Analytical methods

The semi-purified extract was characterised by analysing STX, dc-STX and DOC. DOC was measured with a total organic carbon analyser (Aurora 1030C, OI Analytical Co., Golden, CO, USA). STXs analysis were performed by HPLC using an Agilent 1260 equipped with a quaternary pump, a C18 chromatography column (250 × 4 mm, 5 μm) maintained at 30 °C, a manual injector with a loop of 20 μL, and a fluorescence detector–FLD with excitation of 340 nm and emission of 390 nm. As mobile phase, a 0.05 M ammonium formate aqueous solution with 5% HPLC grade acetonitrile (A) and a 0.1 M ammonium formate aqueous solution (B) with a total flow rate of 1.5 mL min–1 were applied. The process began with 100% mobile phase A. From 0 to 7.5 min, phase B increased from 0 to 20%. From 7.5 to 11 min, phase B increased from 20 to 80%, remaining unchanged until min 13. From 13 to 15 min it returned to 100% of A. The above methodology was adapted from Lawrence et al. (2005) and was validated (Silvino & Capelo-Neto 2014) using pre-column derivatisation method and STX standards from the Institute for Marine Bioscience (National Research Council–Halifax, Canada).

GAC sample characterisation

Coconut shell, an agricultural by-product from renewable resources, is the most abundant and affordable raw material used to produce activated carbon in Northeast Brazil (Jaguaribe et al. 2005) and therefore, was the focus of this research. Four commercially available activated carbons produced with coconut shell using steam activation were used and named as GAC samples C1, C2, C3 and C4.

A porosimetry system (ASAP 2000, Micrometrics, Norcross, GA, USA) was used to measure N2 (at 77 K) adsorption-desorption isotherm over a relative pressure (p/po) range of 10–6–1. The surface area was calculated using the Brunauer–Emmett–Teller (BET) (Brunaer et al. 1938) equation and the total pore volume was estimated by converting the amount of adsorbed N2 (cm3 g–1 STP) to its liquid volume (Guo & Lua 2000). The PSD was obtained using density functional theory (Kowalczyk et al. 2003). The iodine number (mg of iodine adsorbed per g of carbon) was determined according to ASTM D 4607-86 (1994). The point of zero charge (PCZ) pH, i.e. the pH above which the total net surface of the carbon particles are negatively charged (Leon y Leon et al. 1992), was measured by the pH drift method (Newcombe et al. 1993).

Adsorption experiments

A batch system was applied to study the adsorption kinetics of STX and dc-STX in aqueous solution by GAC samples C1, C2, C3 and C4. Virgin GAC samples were wet sieved between Tyler Standard Mesh 60 (0.25 mm) and 65 (0.21 mm). Since these experiments were meant to simulate future short bed adsorber tests in a 20 mm internal diameter (ID) column, sieving the GAC sample to this diameter (particle diameter < 50 × column ID) was necessary in order to minimise channelling and wall effect in the column (Martin 1978). Each GAC sample was washed with 10 times its volume with ultrapure water to remove fine particles, then dried in a 110 °C oven to constant weight and cooled in a desiccator where it was stored prior to use (Summers et al. 1992). The water samples were prepared using ultrapure water, with an electrolyte concentration of 0.01 M NaCl, buffered with 10 mM phosphate to pH of 7.0 and spiked with semi-purified STX extract.

For the batch experiments, 3 mg of each GAC sample were placed in 12-mL amber glass vials along with 10 mL of water sample described previously. In each vial, the initial toxin concentrations (C0) were 10.5 μg L–1 for dc-STX and 60.4 μg L–1 for STX. This can be considered a worst-case scenario for dissolved STXs in treated water. The DOC of the synthesised treated water spiked with STXs was 2.5 mg L–1, between the range (1 and 5 mg L–1) observed in real treated water (Internal report–CAGECE 2014). The vials were then quickly placed into a tumbler, and overturned continuously at 15 rpm, in the dark, and in a temperature controlled chamber at 28 °C. Individual vials were than collected hourly up to 8 h and then with less frequency until 48 h. A subsample of 2 mL was removed from the collected vial, filtered through a 0.45 μm syringe filter (Acrodisc, Pall Corp., Port Washington, NY, USA) and stored at –20 °C until analysis. HPLC analysis occurred 48 h after the first vial was collected. To observe if degradation or any other kind of removal besides carbon adsorption occurred, a control vial (without GAC sample) was submitted to the same conditions.

Adsorption modelling

Pseudo-first-order kinetic model (Equation (1)), also known as the Lagergren equation, was originally developed to describe adsorption of oxalic and malonic acids on carbons. 
formula
1
where q (μg mg–1) is the amount of the component adsorbed in a given time t (min), qe (μg mg–1) is the amount adsorbed at equilibrium and, k1 (min1) is the pseudo-first-order constant. qe and k1 can be calculated from the slope and intercept of the graph log (qeq) versus t. In published literature, it has been shown to effectively describe the absorption of drugs (Van Doorslaer et al. 2011), pesticides (Vulliet et al. 2002; Fresno et al. 2005) and dyes (Shen et al. 2012).
Ho & McKay (2000) developed the expression of pseudo-second-order rate (Equation (2)), which describes, generally, chemisorption involving valence forces by sharing or exchange of electrons between adsorbent and adsorbate 
formula
2
where k2 (mg μg–1 min–1) is the pseudo-second-order constant, qe and q (μg mg–1) are the amounts adsorbed at equilibrium and at a given time t, respectively. The constant k2 is determined by the slope of the graph of t/q versus t. Ho (2006) reported that the pseudo-second-order kinetics has been widely applied to the adsorption of pollutants on aqueous solutions.
Weber & Morris (1963) proposed the intraparticle diffusion model (Equation (3)) to evaluate the relevance of intraparticle diffusion in the adsorption process. In this model, if intraparticle diffusion is involved in adsorption, then a plot of the amount of adsorbate per unit mass of adsorbent (q) against square root of time (t1/2) would be a straight line and, if this line crosses the origin, intraparticle diffusion is the limiting step. 
formula
3
where kdi (mg μg1s–1/2) is the intraparticle diffusion coefficient and C is the intercept. C expresses the magnitude of the boundary layer. The higher the value of C, the greater the effect of the boundary layer, that is, as values of C tends to zero, the boundary layer resistance should decrease its importance in the overall adsorption process.

The kinetic experimental data obtained were mathematically modelled with pseudo-first-order kinetics, pseudo-second-order kinetics, and intraparticle diffusion models. Linear regression was incorporated to judge the adequacy of the models. Statistical differences between GAC sample adsorption were tested using t-tests and chi-square. Significance was tested at the confidence level of po = 0.05.

RESULTS AND DISCUSSIONS

GAC sample characterisation

The net surface charges of GAC samples C1, C2, C3 and C4 were mainly positive at the working solution pH (7.0) since their pHPCZ are 8.7, 8.8, 10 and 9.0, respectively. Examination of Table 2 shows that, except for C1, GAC samples have a high surface area and micropore volume. In terms of mesopores, GAC sample C3 has a relatively high volume (23%), indicative of an ‘open’ pore structure, while the other GAC samples only have a small volume of macropore and thus a much more ‘closed’ pore structure. The GAC samples used in this study therefore had the same surface charge but different surface areas and PSDs. Although Iodine number is reported in the literature to correlate well with micropore volume (Bacaoui et al. 2001; Nunes & Guerreiro 2011), no correlation was found between Iodine number and any other parameter evaluated (data not shown).

Table 2

Characteristics of four coconut shell GAC samples used

GAC samples C1
 
C2
 
C3
 
C4
 
Raw material/activation method Coconut shell/steam Coconut shell/steam Coconut shell/steam Coconut shell/steam 
BET area (m2 g–1487 981 1,001 1,018 
Total pore vol. (mL g–10.215 0.446 0.494 0.525 
Micropore vol. (mL g–1/%) 0.178 83% 0.381 85% 0.374 76% 0.437 83% 
Mesopore vol. (mL g–1/%) 0.024 11% 0.047 11% 0.114 23% 0.071 14% 
Macropore vol. (mL g–1/%) 0.013 6% 0.018 4% 0.006 1% 0.017 3% 
Average pore size (nm) 1.126 1.813 1.971 2.060 
Iodine number (mL g–1397.5 739.4 454.7 662.3 
pHPCZ 8.7 8.8 10.0 9.0 
Carbon charge at pH 7,0 
Average particle size (mm) 0.23 0.23 0.23 0.23 
GAC samples C1
 
C2
 
C3
 
C4
 
Raw material/activation method Coconut shell/steam Coconut shell/steam Coconut shell/steam Coconut shell/steam 
BET area (m2 g–1487 981 1,001 1,018 
Total pore vol. (mL g–10.215 0.446 0.494 0.525 
Micropore vol. (mL g–1/%) 0.178 83% 0.381 85% 0.374 76% 0.437 83% 
Mesopore vol. (mL g–1/%) 0.024 11% 0.047 11% 0.114 23% 0.071 14% 
Macropore vol. (mL g–1/%) 0.013 6% 0.018 4% 0.006 1% 0.017 3% 
Average pore size (nm) 1.126 1.813 1.971 2.060 
Iodine number (mL g–1397.5 739.4 454.7 662.3 
pHPCZ 8.7 8.8 10.0 9.0 
Carbon charge at pH 7,0 
Average particle size (mm) 0.23 0.23 0.23 0.23 

Kinetic modelling

The control sample showed that up to 48 h no significant degradation (po = 0.05) of toxins occurred. By adjusting the pseudo-first-order and pseudo-second-order models to the experimental data, it was possible to calculate k1, k2, qe1 and qe2 for the adsorption of dc-STX and STX (Table 3).

Table 3

Pseudo-first-order and pseudo-second-order kinetic constants and correlation coefficients

  Pseudo-first-order
 
Pseudo-second-order
 
GAC sample Toxin qe1 (μg mg–1k1 (min–1R² qe2 (μg mg–1k2 (mg μg–1 min–1R² 
C1 dc-STX 0.050 0.046 0.711 0.035 8.690 0.974 
STX 0.124 0.013 0.777 0.143 1.604 0.968 
C2 dc-STX 0.026 0.016 0.905 0.033 6.598 0.943 
STX 0.171 0.033 0.921 0.203 1.892 0.970 
C3 dc-STX 0.027 0.013 0.418 0.032 17.412 0.979 
STX 0.127 0.010 0.464 0.252 3.034 0.982 
C4 dc-STX 0.032 0.010 0.179 0.037 12.821 0.969 
STX 0.249 0.031 0.905 0.239 0.887 0.927 
  Pseudo-first-order
 
Pseudo-second-order
 
GAC sample Toxin qe1 (μg mg–1k1 (min–1R² qe2 (μg mg–1k2 (mg μg–1 min–1R² 
C1 dc-STX 0.050 0.046 0.711 0.035 8.690 0.974 
STX 0.124 0.013 0.777 0.143 1.604 0.968 
C2 dc-STX 0.026 0.016 0.905 0.033 6.598 0.943 
STX 0.171 0.033 0.921 0.203 1.892 0.970 
C3 dc-STX 0.027 0.013 0.418 0.032 17.412 0.979 
STX 0.127 0.010 0.464 0.252 3.034 0.982 
C4 dc-STX 0.032 0.010 0.179 0.037 12.821 0.969 
STX 0.249 0.031 0.905 0.239 0.887 0.927 

The correlation coefficients (R2) of the pseudo-first-order kinetics were lower than the pseudo-second-order coefficients for adsorption of dc-STX and STX on all four GAC samples utilised, indicating that the adsorption process followed pseudo-second-order kinetics and that the rate-limiting step may be chemisorption. In addition, the calculated qe for the pseudo-first-order kinetic showed a larger error compared to the experimental values, while for pseudo-second-order, the errors between calculated and experimental qe varied by ± 9%.

GAC sample pore size distribution

GAC sample C3, with a greater amount of mesopores (23%), presented the largest STX adsorption capacity (qe2 = 0.252 μg mg–1) although its BET area and pore volume were approximately the same as GAC samples C2 and C4 (po = 0.05). Ding (2010) performed experiments using PAC samples with different PSD to remove atrazine in different source waters containing NOM. The author found that the greater the volume of mesopores, the smaller the pore blockage by NOM and the larger amount of atrazine was adsorbed. Silva (2005) also found a relatively close relation between STXs adsorption capacity and mesoporous volume in PAC samples.

The maximum adsorption capacity of STX found in this study is still small compared to the one (0.625 μg L–1) found by Shi et al. (2012). GAC sample C1 adsorbed the smallest amount of STX (qe2 = 0.143 μg mg–1) probably due to its smaller BET area and average pore size compared to the other GAC samples tested. This same behaviour was not observed for dc-STX, which showed statistically (po = 0.05) the same adsorption capacity (qe2) for all four GAC samples. Another point observed from Table 3 is that, except for GAC sample C1, the relation between qe2 for STX and qe2 for dc-STX was approximated (po = 0.05) to the initial STXs concentration ratio. This might indicate that there was no significant competitive advantage for occupation of adsorption sites between these two compounds on GAC samples C2, C3 and C4.

Pseudo-second-order constants (k2) were greater for adsorption of dc-STX as compared to STX on all four GAC samples despite the initial concentration of dc-STX being lower than STX. It has to be noted, however, that dc-STX has a smaller molecular weight than STX. GAC sample C3, with the largest volume of mesopores, had the greater k2 both for STX and dc-STX, suggesting that mesopore volume might influence not only the adsorption capacity, but also the rate in which those components are adsorbed.

Using the pseudo-second-order kinetic constants (Table 3) and the method utilised by Unuabonah et al. (2008), it was possible to generate fits for adsorption of STX (Figure 2) and dc-STX (Figure 3) using non-linear second-order kinetics equations. These results demonstrated that dc-STX and STX were slowly adsorbed. Although, 90% of the adsorbed concentrations occurred within 5–10 h, equilibrium concentrations were only reached within 24 h, corroborating with the equilibrium times for adsorption of STXs in PAC found by Silva (2005) and Shi et al. (2012).

Figure 2

Pseudo-second-order kinetic non-linear curves for adsorption of STX on C1, C2, C3 and C4 coconut shell GAC samples.

Figure 2

Pseudo-second-order kinetic non-linear curves for adsorption of STX on C1, C2, C3 and C4 coconut shell GAC samples.

Figure 3

Pseudo-second-order kinetic non-linear curves for adsorption of dc-STX on C1, C2, C3 and C4 coconut shell GAC samples.

Figure 3

Pseudo-second-order kinetic non-linear curves for adsorption of dc-STX on C1, C2, C3 and C4 coconut shell GAC samples.

 Figure 4 shows the adsorption plots of q versus t1/2 for dc-STX and STX on the GAC samples. Two different patterns can be observed in most graphs. Initially, there is a gradual adsorption stage where intraparticle diffusion has a greater or lesser influence in the adsorption process depending on the GAC sample and the STX involved, and after that, adsorption becomes very slow and approaches equilibrium. The relative influence of intraparticle diffusion resistance in the overall adsorption process decreases in this order in GAC samples C1 < C4 < C2 < C3 for STX and C1 < C2 < C4 < C3 for dc-STX, as intercept C increases (Table 4). As the GAC sample mesopore volume increases, the intercept C tends to move further way from the origin, indicating a decreasing influence of intraparticle diffusion and greater pore accessibility.

Table 4

Intraparticle diffusion parameters for adsorption of STX and dc-STX on GAC samples

  STX
 
dc-STX
 
GAC sample kp (mg g−1h−1/2C (mg g−1R2 kp (mg g−1h−1/2C (mg g−1R2 
C1 0,0233 ± 0,0005 0,0021 ± 0,0015 0,9978 0,0101 ± 0,0011 −0,0013 ± 0,0019 0,9674 
C2 0,0653 ± 0,0185 0,0457 ± 0,0324 0,7132 0,0051 ± 0,0004 0,0020 ± 0,0013 0,9519 
C3 0,0354 ± 0,0033 0,0913 ± 0,0091 0,9590 0,0047 ± 0,0003 0,0097 ± 0,0009 0,9819 
C4 0,0414 ± 0,0058 0,0100 ± 0,0143 0,9282 0,0104 ± 0,0023 0,0036 ± 0,0037 0,8698 
  STX
 
dc-STX
 
GAC sample kp (mg g−1h−1/2C (mg g−1R2 kp (mg g−1h−1/2C (mg g−1R2 
C1 0,0233 ± 0,0005 0,0021 ± 0,0015 0,9978 0,0101 ± 0,0011 −0,0013 ± 0,0019 0,9674 
C2 0,0653 ± 0,0185 0,0457 ± 0,0324 0,7132 0,0051 ± 0,0004 0,0020 ± 0,0013 0,9519 
C3 0,0354 ± 0,0033 0,0913 ± 0,0091 0,9590 0,0047 ± 0,0003 0,0097 ± 0,0009 0,9819 
C4 0,0414 ± 0,0058 0,0100 ± 0,0143 0,9282 0,0104 ± 0,0023 0,0036 ± 0,0037 0,8698 
Figure 4

Intraparticle diffusion model for adsorption of STX and dc-STX on GAC samples C1, C2, C3 and C4.

Figure 4

Intraparticle diffusion model for adsorption of STX and dc-STX on GAC samples C1, C2, C3 and C4.

GAC sample surface charge

The molecular structure of dc-STX and STX have several amine groups that can potentially gain protons and, thereby, become cationic depending on the solution pH. At pH 7.1, Shi et al. (2012) observed that STX shifted to a mix of mono-cationic and di-cationic species (approximately 65% and 35%, respectively). This same behaviour can be extrapolated to dc-STX due to their molecular structure similarity. Owing to the cationic nature of the GAC sample surface and to the cationic speciation of dc-STX and STX, electrostatic repulsion appeared to be the dominant mechanism, explaining the small adsorption capacity encountered. In addition, as the solubility of dc-STX and STX increases and becomes more ionic, the hydrophobic driving force pushing STX from the aqueous phase is reduced concurrently, reducing the apparent adsorption capacity of the carbons.

Bjelopavlic et al. (1999) showed that at pH 4 the NOM found in waters is negatively charged, and at pH 7 the magnitude of the negative charge is twice that at pH 4. It would be expected that for a positively charged activated carbon, there should be an attractive electrostatic interaction with NOM, while for the positively charged STXs, toxin–surface interaction should be repulsive. Therefore, it is viable to assume that besides the possible micropore blockage caused by high-molecular weight NOM, a favourable adsorption of NOM would further decrease adsorption capacity of dc-STX and STX, as suggested by Shi et al. 2012.

CONCLUSIONS

Kinetics of adsorption of dc-STX and STX on commercial coconut shell GAC samples followed a pseudo-second-order model, indicating that the rate-limiting step may be chemisorption. An equilibrium contact time of 24 h was identified for both dc-STX and STX, corroborating with results found in the literature. At the working solution pH (7.0) all four GAC samples presented a net positive surface charge which might have hindered adsorption of the positively charged STXs (qe2 = 0.252 μg mg–1 for STX and qe2 = 0.037 μg mg–1 for dc-STX on GAC sample C3). When compared to STX, dc-STX was poorly adsorbed, probably due to its smaller initial concentration, but had larger adsorption uptake rates for all GAC samples (qe2 = 17.412 μg mg–1 for dc-STX and qe2 = 3.034 μg mg–1 for STX on GAC sample C3). The presence of NOM may have presented a strong competition for the adsorption sites, further decreasing STX removal. The low adsorption capacities displayed by the GAC samples may suggest that during the process of selecting an adsorbent for STX removal, a negatively charged activated carbon at the water pH would be preferred, as strongly supported in the literature. GAC sample C3, with the larger amount of mesopores, showed the largest adsorption capacity and uptake rate for STX indicating that, when comparing different GAC samples, PSD should be considered as important a parameter as BET area. The importance of PSD was also supported by the intraparticle diffusion model analysis, which displayed a relative decrease of intraparticle diffusion resistance in the overall adsorption process as mesopore volume increased. Finally, this study supports the idea that the selection of the activated carbon should be closely matched to the target component and to the treatment objectives.

ACKNOWLEDGEMENTS

We thank FINEP and CNPq for their financial support and CAGECE for kindly making available their staff, facilities, and important data to the development of this study. We also thank Rolando Fabris (SA Water) for his proof reading and important contributions.

REFERENCES

REFERENCES
ASTM D 4607-86
1994
Standard Test Method for Determining of Iodine Number of Activated Carbon
.
Annual Book of ASTM Standards
,
Philadelphia, PA
, pp.
5.01
.
Bacaoui
A.
Yaacoubi
A.
Dahbi
A.
Bennouna
C.
Luu
R. P. T.
Maldonado-Hodar
F. J.
Rivera-Utrilla
J.
Moreno-Castilla
C.
2001
Optimization of conditions for the preparation of activated carbons from olive-waste cakes
.
Carbon
39
(
3
),
425
432
.
Bailey
D.
Bowen
B.
Goodwin
L.
Procter
L.
1999
An organic removal strategy for Grahamstown WTP
. In:
Proceedings Australian Water and Wastewater Association
,
Adelaide, Australia, April 11–14
.
Brunauer
S.
Emmett
P. H.
Teller
E.
1938
Adsorption of gases in multimolecular layers
.
J. Am. Chem. Soc.
60
,
309
319
.
Carneiro
R. L.
Santos
M. E. V.
Pacheco
A. B. F.
Azevedo
S. M. F. O.
2009
Effects of light intensity and light quality on growth and circadian rhythm of saxitoxins production in Cylindrospermopsis raciborskii (Cyanobacteria)
.
J. Plankton Res.
1
(
1
),
1
8
.
Clemente
Z.
Busato
R. H.
Oliveira Ribeiro
C. A.
Cestari
M. M.
Ramsdorf
W. A.
Magalhães
V. F.
Wosiack
A. C.
Silva de Assis
H. C.
2010
Analyses of paralytic shellfish toxins and biomarkers in a southern Brazilian reservoir
.
Toxicon
55
,
396
406
.
Ding
L.
2010
Mechanisms of competitive adsorption between trace organic contaminants and natural organic matter on activated carbon
.
PhD Thesis
.
University of Illinois, Illinois, USA
, p.
185
.
Fresno
F.
Guillard
C.
Coronado
J.
Chovelon
J. M.
Tudela
D.
Soria
J.
Herrman
J.
2005
Photocatalytic degradation of a sulfonylurea herbicide over pure and doped TiO2 photocatalysis
.
J. Photochem. Photobiol. A Chem.
173
,
13
20
.
Fuerhacker
M.
Durauer
A.
Jungbauer
A.
2001
Adsorption isotherms of 17 β-estradiol on granular activated carbon (GAC)
.
Chemosphere
44
,
1573
1579
.
Gorham
P. R.
Maclachlav
J. R.
Hammer
V. T.
Kim
W. K.
1964
Isolation and culture of toxic srains of Anabaena flos-aquae (Lyngb.) de Bréb
.
Verh. Int. Verein. Theoret. Angew. Limnol.
15
,
796
804
.
Hilal
S.
Karickhoff
S. W.
Carreira
L. A.
1995
A rigorous test for SPARC's chemical reactivity models: estimation of more than 4300 ionization pKa’s
.
Quant. Struct. Activity Relationships
14
,
348
355
.
Humpage
A. R.
Rositano
J.
Bretag
A. H.
Brown
R.
Baker
P. D.
Nicholson
B. C.
Steffensen
D. A.
1994
Paralytic shellfish poisons from Australian cyanobacterial blooms
.
Aus. J. Marine Freshwater Res.
45
,
761
771
.
Jaguaribe
E. F.
Medeiros
L. L.
Barreto
M. C. S.
Araujo
L. P.
2005
The performance of activated carbons from sugarcane bagasse, babassu, and coconut shells in removing residual chlorine
.
Brazil. J. Chem. Eng.
22
(
1
),
41
47
.
Kaas
H.
Henriksen
P.
2000
Saxitoxins (PSP toxins) in Danish lakes
.
Water Res.
34
,
2089
2097
.
Kuiper-Goodman
T.
Falconer
I.
Fitzgerald
J.
1999
In:
Toxic Cyanobacteria in Water: a Guide to Their Public Health Consequences, Monitoring and Management
(
Chorus
I.
Bartram
J.
, eds).
E & F. N. Spon
,
London
, pp.
113
153
.
Lawrence
J. F.
Niedzwiadek
B.
Menard
C.
2005
Quantitative determination of paralytic shellfish poisoning toxins in shellfish using pre-chromatographic oxidation and liquid chromatography with fluorescence detection: collaborative study
.
J. AOAC Int.
88
(
6
),
1714
1719
.
Leon y Leon
C. A.
Solar
J. M.
Calemma
V.
Radovic
L. R.
1992
Evidence for the protonation of basal plane sites on carbon
.
Carbon
30
,
797
.
Merel
S.
Walker
D.
Chicana
R.
Snyder
S.
Baurès
E.
Thomas
O.
2013
State of knowledge and concerns on cyanobacterial blooms and cyanotoxins
.
Environ. Int.
59
,
303
327
.
Newcombe
G.
Nicholson
B.
2002
Treatment options for the saxitoxin class of cyanotoxins
.
Water Sci. Technol. Water Supply
2
(
5–6
),
271
275
.
NWQMS–National Water Quality Management Strategy
2004
Australian Drinking Water Guidelines 6. Endorsed by National Health and Medical Research Council–NHMRC. Available from
: www.nhmrc.gov.au/_files_nhmrc/publications/attachments/eh34_adwg_11_06.pdf (accessed 9 December 2014).
Portaria 2914
de 12, December of
2011 Brasília
.
Ministry of Health
,
Brazil
.
Shi
H.
Ding
J.
Timmons
T.
Adams
C.
2012
Ph effects on the adsorption of saxitoxin by powdered activated carbon
.
Harmful Algae
19
(
64
),
61
67
.
Silva
A. S.
2005
Assessment of the ability of removal of saxitoxins by different types of powdered activated carbon (PAC) produced in Brazil
.
PhD Dissertation, UNB, DF
,
115
.
Silvino
L. S.
Capelo-Neto
J.
2014
Validation of analytical method for saxitoxin detection (STX and dc-STX) by high performance liquid chromatography with fluorescence detector and pre-column derivation
.
Revista AIDIS
7
(
3
),
243
258
.
Summers
R. S.
Cummings
L.
DeMarco
J.
Hartman
D. J.
Metz
D. H.
Howe
E. W.
MacLeod
J. M. B.
Simpson
M.
1992
Standardized protocol for the evaluation of GAC
.
Report prepared for AWWA Research Foundation
.
American Water Works Association (AWWA)
,
Denver
.
Van Doorslaer
X.
Demeestere
K.
Heynderickx
P. M.
Van Langenhove
H.
Dewulf
J.
2011
UV-A and UV-C induced photolytic and photocatalytic degradation of ciprofloxacin and moxofloxacin: Reaction kinetics and role of adsorption
.
Appl. Catalysis B Environ.
101
,
540
547
.
Vulliet
E.
Emmelin
C.
Chovelon
J.
Guillard
C.
Herrmann
J.
2002
Photocatalytic degradation of sulfonylurea herbicides in aqueous TiO2
.
Appl. Catalysis B Environ.
38
,
127
137
.
Weber
W. J.
Morris
J. C.
1963
Kinetics of adsorption on carbon from solution
.
J. Sanitary Eng. Div.
89
(
6
),
53
55
.