Colloidal activated carbon for in-situ groundwater remediation — Transport characteristics and adsorption of organic compounds in water-saturated sediment columns

Colloidal activated carbon for in-situ groundwater remediation — Transport characteristics and adsorption of organic compounds in water-saturated sediment columns

Journal of Contaminant Hydrology 179 (2015) 76–88 Contents lists available at ScienceDirect Journal of Contaminant Hydrology journal homepage: www.e...

1012KB Sizes 0 Downloads 6 Views

Journal of Contaminant Hydrology 179 (2015) 76–88

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology journal homepage: www.elsevier.com/locate/jconhyd

Colloidal activated carbon for in-situ groundwater remediation — Transport characteristics and adsorption of organic compounds in water-saturated sediment columns Anett Georgi a,⁎, Ariette Schierz b, Katrin Mackenzie a, Frank-Dieter Kopinke a a b

Helmholtz Centre for Environmental Research, UFZ, Department of Environmental Engineering, Permoserstr. 15, D-04318 Leipzig, Germany Department of Civil and Environmental Engineering, Texas Tech University, 911 Boston Avenue, Lubbock, TX, 79405, USA

a r t i c l e

i n f o

Article history: Received 1 October 2014 Received in revised form 27 April 2015 Accepted 5 May 2015 Available online 2 June 2015 Keywords: Colloidal activated carbon Plume Sorption barrier Particle transport Contaminant retardation Groundwater Nanoremediation

a b s t r a c t Colloidal activated carbon can be considered as a versatile adsorbent and carrier material for insitu groundwater remediation. In analogy to other nanoremediation approaches, activated carbon colloids (ACC) can be injected into the subsurface as aqueous suspensions. Deposition of ACC on the sediment creates a sorption barrier against further spreading of hydrophobic pollutants. This study deals with the optimization of ACC and their suspensions with a focus on suspension stability, ACC mobility in saturated porous media and sorption efficiency towards organic contaminants. ACC with an appropriate particle size range (d50 = 0.8 μm) were obtained from a commercial powdered activated carbon product by means of wet-grinding. Among the various methods tested for stabilization of ACC suspensions, addition of humic acid (HA) and carboxymethyl cellulose (CMC) showed the best results. Due to electrosteric stabilization by adsorption of CMC, suspensions remained stable even at high ACC concentrations (11 g L−1) and conditions typical of very hard water (5 mM divalent cations). Furthermore, CMC-stabilized ACC showed high mobility in a water-saturated sandy sediment column (filter coefficient λ = 0.2 m−1). Such mobility is a prerequisite for in-situ installation of sorption or reaction barriers by simple injection-well or directpush application of ACC suspensions. Column experiments with organic model compounds proved the efficacy of ACC deposits on sediment for contaminant adsorption and retardation under flowthrough conditions. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Application of nanomaterials for in-situ groundwater remediation has recently gained great attention. Nanoscale zerovalent iron (NZVI) is the most extensively studied material which has been already tested in the field (Comba et al., 2011; Crane and Scott, 2012; Zhang and Elliott, 2006). It has been applied mainly for reductive dehalogenation of halogencontaining organic compounds but also for reduction of other organic contaminants as well as immobilization of heavy metals, radionuclides and arsenic compounds as summarized in (Crane ⁎ Corresponding author. Tel.: +49 341 2351760; fax: +49 341 235451760. E-mail address: [email protected] (A. Georgi).

http://dx.doi.org/10.1016/j.jconhyd.2015.05.002 0169-7722/© 2015 Elsevier B.V. All rights reserved.

and Scott, 2012). Other types of reactive colloidal materials under study are bimetallic nanoparticles (e.g. Pd–Fe, (He et al., 2007)) and Fe on various supports, such as hydrophilic carbon (Schrick et al., 2004), poly(acrylic acid) (Schrick et al., 2004), carbon microspheres (Sunkara et al., 2010), activated carbon (Mackenzie et al., 2008, 2012) or porous silica (Zhan et al., 2008). In addition, polymeric nanoparticles have been suggested as adsorbents for enhanced extraction of pollutants from contaminated sediments during soil washing combined with ex-situ biodegradation (Tungittiplakorn et al., 2005). In this paper activated carbon colloids (ACC) are studied as an injectable material for in-situ groundwater remediation. Activated carbon (AC) has excellent adsorption properties for a wide spectrum of organic compounds and can therefore be

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

used to build an effective sorption barrier. State-of-the-art technologies for groundwater treatment apply granular activated carbon in ex-situ reactors or built-in permeable reactive barriers (PRBs) (Plagentz et al., 2006; USEPA, 2002). Remediation of aquifer sediments contaminated with hydrophobic organic compounds (HOC) by in-situ mixing of the sediments with carbonaceous materials (mainly commercial powdered or granulated AC) is receiving increasing interest, due to the fact that significant reduction of pore water HOC concentrations and thus sediment toxicity can be achieved, leading to an improvement of habitat quality for benthic organisms (Rakowska et al., 2012). So-called super powder activated carbon with a significantly lower particle size of about 1 μm, which is significantly smaller than commercial powdered activated carbon (typically N 10 μm) was suggested for application in batch reactors for water treatment with the focus on the acceleration of the adsorption processes (Matsui et al., 2009). With respect to groundwater remediation, AC with a particle size in the lower μm range (i.e. ACC) offers the possibility to produce stable suspensions that can be injected into an aquifer. Deposition of ACC on the sediment grains around the injection well then creates a sorption-active zone in the aquifer which can prevent spreading of pollutants with the groundwater flow. On the other hand, ACC can be used as a support for catalysts or reagents in order to establish a combined sorption/reaction barrier or to attempt direct source removal. The concept of a sorption/reaction barrier has been put into practice by this group with the novel bi-functional colloid material, Carbo-Iron®, which consists of ACC with deposits of zero-valent iron (about 20 wt.%) and can be applied for reductive dehalogenation of halogenated hydrocarbons (Bleyl et al., 2012; Mackenzie et al., 2007, 2008; Mackenzie et al., 2012). Just after finalization of this study we became aware of a new commercial product for in-situ groundwater remediation based on colloidal activated carbon — PlumeStopTM (Regenesis, 2014), which is claimed to combine sorption-based contaminant removal with accelerated biodegradation. Due to its novelty, detailed reports on the performance of this specific material in the scientific literature are yet to come. Our study is focused on optimization of ACC with respect to suspension stability and mobility in porous media combined with verification of their adsorption properties towards organic contaminants. By this means, this study contributes to providing a basis for various approaches using ACC as adsorbent, biomatrix or carrier of reactive materials for in-situ remediation. In-situ application of colloids requires sufficiently stable colloid suspensions for successful subsurface delivery. In addition, colloid transport within the sediment up to a distance of a few meters around the injection well would be optimal in order to provide a sufficiently large area of impact with a low number of individual injections but also to allow application of lowinvasive infiltration techniques (e.g. by injection wells or directpush) without problems of pore clogging. The rate of colloid deposition during transport through aquifer sediments is a function of the frequency of collisions of the colloids with the surfaces of the sediment grains and the probability of attachment in case of collision. For colloids with a density close to or slightly higher than water (1.05–2.6 g cm−3) the size range of 0.5–2 μm is generally associated with the highest mobility (Elimelech et al.,

77

1995; Yao et al., 1971) in aquifers. Thus, in this study dry and wet milling procedures were tested in order to obtain ACC in the optimal size range. Furthermore, the mobility of colloids in the aquifer is promoted by electrostatic repulsion between the colloid and sediment surfaces. However, electrostatic interactions can be screened in the presence of high concentrations of dissolved ions (especially polyvalent ions). The surface of silica, which is the predominant mineral in aquifers, is negatively charged under the near-neutral conditions of the average aquifer. Consequently, a negative colloid surface charge and a low ionic strength of the suspension medium would promote the mobility of colloids after injection into the aquifer. At the same time these colloid and suspension properties would also be favorable for preventing unwanted colloid agglomeration. Depending on the starting material and production conditions, AC can possess different contents of basic and acidic surface sites. However, for most AC materials, basic sites predominate, which leads to a net positive surface charge at near-neutral pH values (Julien et al., 1998; Moreno-Castilla et al., 1998). Acidic and oxidative treatments of AC are known measures to increase the content of acidic functional groups at the AC surfaces (Bhatnagar et al., 2013). On the other hand, surface modification of colloids by adsorption of surfactants or polymers (such as carboxymethylcellulose (CMC), poly(styrene sulfonate), xanthan, guar gum and others) has attained much attention as a means to increase suspension stability and colloid mobility for in-situ remediation applications of NZVI (Dalla Vecchia et al., 2009; Phenrat et al., 2007, 2008; Saleh et al., 2008). Adsorbed polymers and surfactants can electrostatically, sterically or electrosterically stabilize colloid suspensions and prevent their deposition on porous matrices. The travel distance (LT) for uniform particles within a granular filter medium, i.e. the distance over which a certain reduction of the particle concentration in the mobile phase (e.g. cL/c0 = 0.01 for a hundredfold reduction) is achieved, can be described according to Eq. (1): LT ¼ − ln

    cL 1 c 2dc ¼ − ln L c0 λ c0 3ð1−ε Þα η0

1

where λ is the filter coefficient, dc is the diameter of the filter grains, ε is the porosity of the granular filter medium, η0 is the single collector efficiency and α is the attachment efficiency (Elimelech et al., 1995). The parameter η0 is a measure for collision frequency whereas α describes the probability of attachment after collision. α is determined by the sum of attractive and repulsive forces acting between the solid surfaces involved (e.g. van der Waals attraction, electrostatic and steric interactions). For homogeneous filter media, η0 can be predicted from basic properties of the system using equations which describe the contribution of diffusion, interception and gravity on particle deposition (Elimelech et al., 1995). However, for predictions about colloid transport in real aquifer sediments, the filter coefficient λ must be determined experimentally, e.g. in column experiments under representative conditions. Filter ripening and colloid agglomeration phenomena can limit the applicability of Eq. (1). This paper describes studies on suspension stability and mobility of ACC in aquifer sediments. These studies were conducted by means of laboratory batch and column experiments. For surface modification of ACC, oxidative treatment

78

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

and adsorption of various stabilizers were tested, with the aim to facilitate transport of ACC under typical aquifer conditions over distances in the range of 10 m or more. Dodecylbenzene sulfonate (SDBS for the sodium salt), coalderived humic acid (HA) and carboxymethyl cellulose (CMC, [C6H12 -mO6(CH2COOH)m]n) were used as stabilizers, which at neutral pH are negatively charged molecules. Possible impacts of ACC modification on their adsorption properties towards target pollutants were investigated in batch studies with monochlorobenzene (MCB) as model pollutant. In addition, column experiments with aquifer sediment loaded with ACC were conducted in order to investigate the availability and efficiency of the immobilized ACC for sorption of organic contaminants under flow through conditions. The retardation effects observed for two model compounds (dichloromethane (DCM) and chloroform (CF)) on the ACC-loaded sediment column were related to the corresponding equilibrium sorption data.

Table 1 Particle size distributions of ACC samples obtained by different grinding procedures and their mobilities in columns filled with quartz sand (column: l = 18 cm, d = 1.3 cm), cleaned quartz sand (Q): dp = 0.5–1 mm, mobile phase: ACC (25 mg L−1) in NaOH solution (pH 10). Parameter

ACC sample ACC3.2

ACC1.6

ACC0.8

Grinding conditions d50 (μm)a) d90 (μm)a) fmobile (%)

Dry-grinding

3 h wet-grinding

7 h wet-grinding

3.2 6.0 28 ± 9b)

1.6 3.2 50 ± 13

0.8 (0.9) 1.6 (1.8) 86 ± 5

a) Determined by means of DLS except for values in parentheses, which were determined by laser diffraction. d50 and d90 represent the particle size at the 50% and 90% point, respectively, of the cumulative undersize mass-based distribution. b) The ± intervals indicate the mean deviation of 3 to 5 single values from the mean value.

2. Materials and methods

2.3. Sediment samples

2.1. Chemicals

In the column studies, a cleaned quartz sand (Q) and two size fractions of a native aquifer sediment (Z1 and Z2) were used. Pre-washed and size-fractionated quartz sand was purchased from Quarzwerke GmbH (Weferlingen, Germany) and sieved to a final size fraction of 0.5–1 mm. In order to remove adhering fine particles, several washing steps were performed, whereby the sand was suspended in deionized water and treated in an ultrasonic bath (35 kHz). The aquifer sediment was extracted from a depth of 40 m on the site of a former coal-liquefaction plant in Zeitz, Germany. The native sediment is an Eocene fluvial sand classified as a sandy fine to medium gravel. The air-dried sediment was sieved to two size fractions: 0.25–1 mm (Z1) and ≤ 5 mm (Z2).

Unless otherwise noted, chemicals were obtained from Merck (Germany) or Sigma-Aldrich (Germany) and had a purity of ≥ 96%. HA was purchased from Carl Roth (Germany). It is produced by alkaline extraction of leonardite. HA solutions were prepared by dissolving an appropriate amount of the HA in dilute NaOH. Afterwards the solution was diluted to the desired volume with deionized water and the pH was adjusted using HCl. The applied CMC is a technical product (Antisol FL30) from Wolff Cellulosics (Germany). According to the information provided by the supplier this CMC has a molecular weight of about 70 kg mol−1 and a substitution degree of 0.85 (number of etherified hydroxyl groups per glucose unit).

2.4. Analytical methods 2.2. Colloidal activated carbon SA Super (Norit, Germany) was used as raw material for preparation of ACC. It was purchased in the form of a paste-like product (50 wt.% water content) from ETC-Burgau (Germany, product name: Adsorba®-N). According to the data provided by the supplier, the particle size range of this AC product is 4– 70 μm. The main physical properties of the applied AC are summarized in Table S1 (Supplementary Material). Grinding of the AC was done by Hosokawa Alpine AG (Augsburg, Germany). For wet grinding, a suspension with a solids content of 15 wt.% of AC was prepared. This suspension was re-circulated in an agitator bead mill 200 AHM (grinding medium: YrO2-Cer-beads, 0.4–0.7 mm). One sample of the grinding product was retained after 3 h of grinding. In the following, this sample is denoted as ACC3.2 with the subscript indicating the mean particle size (d50) which is presented for this and all other ACC samples in Table 1. The main part was ground for 7 h until the desired particle size distribution was reached (denoted as ACC0.8). Furthermore, dry grinding in a fluidized bed opposed jet mill (100 AFG, Hosokawa) was applied to an oven-dried (105 °C) sample of the AC (denoted as ACC1.6).

Analysis of the total organic carbon content (TOC) was applied in order to determine the concentration of ACC, CMC, SDBS and HA in aqueous samples. The specific carbon contents of these compounds are 98, 33, 62 and 49 wt.%, respectively. Aqueous samples were injected by means of a syringe into a carbon detector, which consisted of a two-zone combustion unit (800 °C combustion zone, 700 °C catalytic post-oxidation zone filled with 0.5 wt.% Pt/Al2O3), membrane dryer for drying of the combustion gas and NDIR detector for CO2 detection. In the case, that TOC of a solution had to be determined in the presence of a significant concentration of inorganic carbon, the latter was removed prior to analysis by purging of the acidified sample (pH ≤ 5) with N2. The organic carbon content of solid samples was determined by means of a C-Mat 5500 carbon analyzer (Stroehlein, Germany). The particle size distribution of the ACC was determined by means of dynamic light scattering (DLS) using a Zetasizer NS (Malvern, Germany) and laser diffraction (Beckman Coulter LS). For determination of the isoelectric point (IEP), the zeta potential of ACC in suspensions with various pH values (2 ≤ pH ≤ 12) and 10 mM KNO3 as background electrolyte was measured using a Zetasizer NS (Malvern, Germany).

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

2.5. Oxidative pre-treatment of ACC 10 g of ACC was treated with 30 g of H2O2 in 200 mL of deionized water under cooling at about 25 °C for 24 h. Afterwards, the ACC was removed by centrifugation and washed several times with deionized water. 2.6. Investigation of the stability of ACC suspensions The suspension under investigation was treated in an ultrasonic bath for 15 min, filled into a 100 mL graduated cylinder, and subsequently placed on a bench. After certain time periods, samples of the suspension were taken at a distance of 10 cm from the water surface by means of a syringe (without affecting the filling level). These samples were analyzed for their TOC contents. Suspensions were considered as stable if the TOC contents of the samples did not significantly decrease within a time period of 8 h. 2.7. Batch studies on adsorption of model contaminants Batch experiments using headspace analysis with GC–MS were conducted in order to study the adsorption of the model contaminants CF, DCM and MCB on ACC in the absence and presence of stabilizers. Details on the experimental procedure can be found in the Supplementary material. 2.8. Column experiments Three types of glass columns were applied: C1 (l = 18 cm, d = 1.3 cm), C2 (l = 100 cm, d = 2.6 cm) and C3 (l = 100 cm, d = 5 cm). C2 was equipped with sampling ports at four different heights (20, 40, 60, 80 cm and outlet at 100 cm). These sampling ports were closed with rubber septa and equipped with needles which were embedded in the column filling for sampling of the mobile phase. PTFE tubing was used for connections. The column entrance was filled with a thin layer of fine gravel in order to prevent drainage of the packing material. The columns were either slurry-packed or filled with the dry sand or sediment material. In the latter case the columns were flushed with CO2 before the mobile phase was pumped through the column in order to accelerate the water saturation of the packing material. The mobile phase was passed through the column in the upward direction using a hydraulic gradient or a pump. If not otherwise stated, the average linear velocity (=flow rate / (cross section area ∙ porosity)) of the mobile phase was u = 2.8 cm min−1. Tracer tests with KNO3 as conservative tracer were conducted by means of pulse injections (1 or 5 mL of 10 mM KNO3) and UV detection (225 nm) of the column effluent. In order to evaluate the effect of ACC deposition on the dispersivity and effective porosity of the sediment columns, KNO3 breakthrough curves were recorded before and after loading of the column with ACC. 2.9. Column experiments on ACC mobility In column experiments with ACC, a suspension containing ACC, background electrolyte and (if intended) stabilizer was filled into the mobile phase reservoir (Schott flask of appropriate volume) and was continuously stirred by magnetic stirring. The mobile-phase flow through the column was either provided

79

by a hydraulic gradient or a piston pump (Ismatec, Germany). No significant differences were observed between the two ACC feeding modes with respect to ACC mobility in the columns. Before and after each ACC feeding step, the column was flushed with at least 4 pore volumes of a solution with a background electrolyte composition identical to that of the ACC suspension. The concentration of ACC in aliquots of the effluent was determined by means of TOC analysis at various time points after feeding of ACC suspensions was started. In the presence of stabilizers (CMC, SDBS, HA), the TOC measured in the column effluent had to be corrected for the contribution of the stabilizers. Generally, the concentration of stabilizer sorbed on ACC was small (≤0.06 g g−1) and did not significantly contribute to the total TOC. The contribution of the freely dissolved fraction was determined exemplarily for selected samples after separation of ACC by centrifugation (30 min at 5000 rpm). 2.10. Column experiments on retardation of contaminants by ACC deposited on sediment Sediment (Z2) was filled in column C2 and the column was pre-conditioned as described above. Afterwards, breakthrough curves for nitrate (inert tracer), CF and toluene (sorptive tracers) were determined. Tracers were injected with a syringe into the mobile phase flow (2 mM CaCl2 in deionized water, pH 7, flow rate = 4 mL min−1, u = 2.8 cm min−1) via the sampling port below the column entrance (5 mL of 10 mM KNO3 or 5 mL of a solution containing 100 mg L−1 of each CF and toluene injected within 1 min). Nitrate analysis was done by UV detection. For determination of CF and toluene concentrations effluent samples (1.4 mL) were collected with a glass syringe via the sampling ports, transferred into a vial with septum and analyzed by Headspace-GC-MS analysis. Afterwards, the sediment was loaded with ACC by flushing with suspensions containing 100– 200 mg L−1 ACC, 50–100 mg L−1 HA and 1–2 mM CaCl2 in deionized water (pH 7) as described above (for details see Table 3 in Results and Discussion section). After deposition of ACC, breakthrough curves for nitrate (inert tracer), DCM and CF (sorptive tracers) were recorded. Tracers were injected with a syringe into the mobile phase flow (2 mM CaCl2 in deion. water, pH 7, flow rate = 4 mL min−1, u = 2.8 cm min−1) via the sampling port below the column entrance (5 mL of 10 mM KNO3 or 5 mL of a solution containing 500 mg L−1 of CF and DCM injected within 1 min). Tracer analysis was conducted as described above. In total about 22 L of mobile phase were passed through the column during the sorption experiment. 3. Results and discussion 3.1. Suspension stability and mobility of native ACC The mobility of ACC in the column experiments was quantified according to Eq. (2), where the mobile fraction of ACC (fmobile) is calculated from the ratio of the concentrations of ACC in the outflow ([ACC]out) and the inflow ([ACC]in) of the column.

f mobile ¼

½ACCout  100% ½ACCin

2

80

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

This ratio was determined when the ACC concentration in the column effluent had reached a constant value, which generally required exchange of 4–5 pore volumes after switching from background electrolyte to ACC suspensions. Table 1 shows the mobility of various ACC samples which differ in their particle size distributions. The particle size distribution of the sample ACC0.8 was measured with two different methods, DLS and laser diffraction, whereby nearly identical results were obtained (Table 1). Fig. 1a) shows a microscopic image (5000-fold magnification) of ACC0.8. By means of the dry-grinding process it was not possible to obtain ACC with a mean particle size close to 1 μm. The mobility of this sample (ACC3.2) in the saturated quartz sand column was low, even under conditions which should provide optimal electrostatic repulsion between ACC and quartz surfaces (high pH and low ionic strength of the mobile phase). The highest mobility was observed for the sample with the smallest particle size (ACC0.8) which was obtained by the 7 h wet-grinding process. A prolonged recirculation of the material in the wetgrinding apparatus did not lead to a further decrease in particle size. The achievable lower size limit is determined by the applied operation conditions but also by the properties of the material to be ground. At a certain size range, dis- and reaggregation of particles reach equilibrium, preventing any further reduction of particle size. A strong dependency of mobility on particle size in the range of d50 = 0.64 to 2.4 μm was also found for iron-loaded (≈20 wt.% Fe) activated carbon colloids (Busch et al., 2014a,b). Based on clean-bed filtration models and experimental observations, the collision frequency as a function of colloid size was reported to have a minimum in the range of 0.5–2 μm for colloids with a density close to or slightly higher than water (Elimelech et al., 1995). The particle size distribution of the sample ACC0.8 obviously coincides best with this optimum size range and thus shows the highest mobility in the saturated quartz sand. Therefore, this material was applied for all further experiments.

For the non-modified, ‘native’ colloids the isoelectric point (IEP) was found at pH = 5.7 (Fig. 1b). However, the zeta potential of the ACC does not reach a value of ≤ −30 mV for pH values of the solution ≤ 8. For many colloidal systems an absolute value of the colloid zeta potential of ≥ 30 mV was found to be a necessary prerequisite for suspension stability (Xu, 2000). Likewise, stable suspensions of the native ACC with a concentration of ≥ 100 mg L−1 were not obtained at pH = 7 but only in alkaline solutions, where the zeta potential is ≤ −30 mV. The mobility of the native ACC in sand-filled columns is likewise strongly influenced by the pH of the aqueous phase (Table S2 in Supplementary Material). In order to achieve a sufficient mobility (fmobile ≥ 80%) of ACC in small-scale columns (l = 18 cm), the mobile phase must have a pH of ≥ 10 and a low ionic strength. The presence of only 1 mM CaCl2 leads to complete immobilization of ACC, due to screening of electrostatic repulsion. Even under conditions promoting suspension stability and colloid mobility (NaOH solution (pH = 10) in deionized water as mobile phase), the formation of agglomerates was observed, which were deposited within the first centimeters of the column filling. The digital microscope image (Fig. S1 in Supplementary Material) showed that colloid deposition within this segment occurred predominantly in the form of agglomerates in the pore volume of the sand matrix. Before entering the column, the suspension was free of agglomerates. Therefore, we presume that the native ACC is susceptible to an orthokinetic agglomeration effect (Elimelech et al., 1995) caused by the specific flow conditions in the column entrance segment. 3.2. Surface modification of ACC Due to the limited suspension stability and mobility of native ACC we followed various approaches of surface modification. By means of a mild oxidative pre-treatment of the ACC with H2O2, the content of acidic functional groups was increased, thus lowering the zeta potential in the neutral pH range (Fig. 1b).

Fig. 1. a) Optical microscopy image of ACC0.8 suspended in deionized water (5000-fold magniification, scale division: 1 μm), b) Zeta potential of native and pre-oxidized ACC0.8 as a function of suspension pH.

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

Furthermore, anionic stabilizers were applied which can create a negative ACC surface charge at nearly neutral pH when being adsorbed. Fig. S2 (Supplementary Material) shows the adsorption isotherms of the applied stabilizers SDBS, HA and CMC on ACC. The maximum loading of the two macromolecular polyelectrolytes HA and CMC is relatively low (≤7 wt.% on ACC). This is in accordance with the hypothesis that the HA and CMC macromolecules are excluded from the inner pore volume and are adsorbed mainly at the external surface of the ACC. In case of SDBS, the maximum loading is about 35 wt.%, indicating adsorption also within the ACC micropore system. The stability of colloid suspensions generally depends on various parameters: colloid concentration, surface charge of the colloids and ionic strength of the suspension medium. For practical application, stable suspensions with high ACC concentrations are desirable, since the latter parameter determines the necessary injection volume and the injection duration. The use of water with moderate ionic strength (tap water or groundwater), rather than deionized water, would be advantageous with respect to process costs and prevention of possible perturbations in the subsurface. Table 2 illustrates the results of stability tests with suspensions of the native and surfacemodified ACC. ACC concentrations at a level of 10 cm below the surface of a non-agitated suspension were measured over time and suspensions were considered as stable if no significant change in ACC concentration (b 10%) was observed during 8 h. Based on a density difference between suspended ACC and water of about 0.3 g cm−3, a particle with a diameter of 1 μm would have a sedimentation velocity of 0.04 cm h−1. A significant depletion of the TOC content at the sampling depth of 10 cm would be caused only in the presence of a significant fraction of particle agglomerates with dp N 10 μm. In case of HA and CMC, the concentration of stabilizer added to the suspension was adjusted in such a way that, according to the adsorption isotherms, the maximum loading on ACC was reached (about 7 wt.%). In case of SDBS the loading was 20 wt.%. Although suspension stability in deionized water was improved in the case of the oxidatively pre-treated and SDBS-stabilized ACC compared to native ACC, it was still very sensitive to the presence of Ca2+. The latter effect remained unchanged also when the SDBS loading of the ACC was further increased (data not shown). Low concentrations of Ca2+ are tolerated in the case of HA- and CMCstabilized ACC. However, suspensions with HA as stabilizer became unstable if the ACC concentration was raised above Table 2 Stability of suspensions of native and oxidatively pre-treated ACC0.8 depending on the concentration of Ca2+ and the presence of various stabilizers in the suspension at pH = 7 (+ stable suspension, − formation of agglomerates)a). Pre-treatment/ type of stabilizer

Stabilizer concentration in g L−1

[ACC] in g L−1

[Ca2+] in mM

Stability

None Pre-oxidation Pre-oxidation SDBS SDBS HA HA CMC CMC

− − − 0.025 0.025 0.4 2.5 2.5 4

0.1 0.1 0.1 0.1 0.1 1 5 11 24

0 0 0.5 0 0.5 2 2 5 5

− + − + − + − + −

a) + stable suspension: no significant depletion in [ACC] at a level of 10 cm below surface within 8 h.

81

1 g L−1. In contrast, suspensions with CMC as stabilizer were stable at least up to concentrations of 11 g L−1 ACC. Among the various options tested, the addition of CMC most effectively prevents the agglomeration of ACC in the presence of divalent cations and at high ACC concentrations. The stabilizing effect of CMC for colloids is considered to be due to the formation of a bulky and negatively charged layer of CMC molecules adsorbed to the particle surface, which results in electrosteric stabilization (He and Zhao, 2007; He et al., 2007; Phenrat et al., 2008). 3.3. Mobility screening for surface-modified ACC Firstly, mobility tests with differently treated ACC samples were conducted using short columns of 18 cm length which were filled with cleaned quartz sand. When deionized water was used as the suspension medium (pH adjusted to 7) and a low ACC concentration (30–100 mg L−1) was applied, mobile fractions of ≥ 80% were obtained in case of the oxidatively pretreated ACC as well as for ACC suspensions with stabilizer (SDBS, HA and CMC, Fig. 2). Obviously, the necessity of an alkaline suspension medium can be overcome by all the tested options. However, the necessity of using desalinated water is still valid in the case of the oxidatively pre-treated ACC (due to suspension instability) and in the case of the SDBS-stabilized ACC suspensions. For the latter, the mobile fraction (l = 18 cm) decreases to ≤ 20% at a Ca2+ concentration of only 0.5 mM. With HA or CMC as stabilizers, the ACC mobility is sufficiently high at moderate Ca2+ concentrations of 2–4 mM, as will be discussed in more detail in the following sections. With respect to homogeneity of the ACC deposition along the column, stabilization by HA or CMC gave the best results. When oxidatively pre-treated or SDBS-stabilized ACC was applied, a preferential deposition of ACC agglomerates occurred within the first few centimeters of the column, as already described for the native ACC. In contrast with CMC- or HAstabilization, direct deposition of the colloids on the mineral surfaces (see Fig. 3) as well as a nearly homogeneous ACC loading along the whole length of the column (18 cm) was observed. This finding is important with respect to the hydrodynamics in the artificial sorption zone: it can be expected to be passed-through by the groundwater flow rather than passed-by. For achieving a travel distance of L50 ≥ 1 m (i.e. mobile fraction of ≥ 50% of ACC particle mass over 1 m transport distance), the filter coefficient λ for ACC calculated based on Eq. (1) should be ≤ 0.7 m−1. This λ value can be converted into a minimum mobile fraction for the 18 cm-column experiments of ≥ 85%. This criterion is met with both types of stabilizers, CMC and HA. 3.4. Impact of surface modification on adsorption potential of ACC towards target contaminants Surface modification of ACC by pre-oxidation (H2O2 treatment) or addition of stabilizers should not significantly interfere with target contaminant adsorption. The latter was studied under worst case conditions, i.e. in the presence of high stabilizer concentrations in the suspension leading to an equilibrium loading of stabilizer on ACC of about 5 wt.%. Such conditions would be typical for the injection suspension, however, desorption of the stabilizer from ACC is expected

82

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

Fig. 2. Mobility of native and modified ACC0.8 in columns (C1, l = 18 cm, u = 2.8 cm min-1) filled with quartz sand. Mobile phase consisted of suspensions of ACC at a concentration of 30 mg L−1 (untreated, oxidized and HA-loaded ACC) or 100 mg L−1 (SDBS- and CMC-loaded ACC) in deionized water adjusted to pH = 7. If stabilizer was added, its concentration was adjusted to reach a loading on the ACC of 2–5 wt.% (ctotal,stabilizer = 5 mg L−1 for SDBS, 10 mg L−1 for HA and 25 mg L−1 for CMC).

after deposition on the sediment and replacement of the injection suspension by groundwater. MCB was used as model contaminant. Within the range of cMCB,w = 0.3–40 mg L−1 MCB adsorption on native ACC can be well fitted by the Freundlich isotherm model. Isotherm parameters of MCB and all other compounds for which adsorption on ACC was studied in this paper are summarized in Table S3 (Supplement. Mat.). Among the various stabilization methods tested the oxidative pre-treatment of ACC showed the strongest impact on ACC adsorption potential over the whole MCB concentration range studied (Fig. 4, left). The reduction in adsorption affinity and capacity is a result of the increased content of O-containing functional groups and thus surface polarity after oxidation which affects not only the outer but also the internal surfaces of the ACC colloids. The impact of the stabilizers was less pronounced and obvious only at low MCB concentrations

(Fig. 4, right). CMC and HA as macromolecules show the lowest effect, whereas competition for adsorption sites with MCB is more pronounced for the low-molecular weight surfactant SDBS. Overall, the impact of HA and CMC on MCB adsorption is tolerable with a reduction in the equilibrium loading of MCB (qMCB,ACC) by less than a factor of 1.6 at cMCB,w = 1 mg L−1 in the presence of a high concentration of the stabilizer (cstabilizer,total = 500 mg L−1, cACC = 1 g L−1). Based on the results of the initial screening tests, neither oxidative pre-treatment of ACC nor stabilization with SDBS were studied in further detail. Both surface modification methods only moderately improved the suspension stability of ACC but deteriorated their sorption properties. HA and CMC were selected as stabilizers for use in further column tests with native aquifer sediment. These negatively charged macromolecules offer electrosteric stabilization, which, especially under conditions of high ionic strength, is more effective than the electrostatic stabilization offered by SDBS. In addition, the effect of HA and CMC on the ACC adsorption properties is lowest.

3.5. Mobility of HA-stabilized ACC in native sediment

Fig. 3. Optical microscopy image of quartz sand column filling after elution of 30 pore volumes of ACC suspensions (ACC0.8, pH = 7) containing 30 mg L−1 ACC and 5 mg L−1 HA at pH 7. Final mean concentration of ACC on the sand was 0.015 wt.%. Scale division: 100 μm.

Mobility tests applying the coal-derived HA as stabilizer and native sediment as porous medium were conducted. A column of 100 cm length (C2) with sampling ports at various heights was filled with size-fractionated native sediment (size fraction: 0.25–1 mm (Z1)) and successively loaded with ACC until a mean sediment loading of 0.1 wt.% AC was reached. Table 3 shows the mobile fractions of ACC in the various phases of the column experiment where cACC was varied between 100 and 200 mg L−1 and cCa2+ between 1 and 2 mM. The increase in ACC concentration from 100 to 200 mg L−1 did not change the ACC mobility at cCa2+ = 1 mM, whereas it decreased the mobility at cCa2+ = 2 mM. This indicates that ACC mobility is at least partly controlled by agglomeration in the presence of elevated Ca2+ concentrations. The mobile fraction was 77 ± 8% after a distance of 1 m under all applied conditions except for the suspensions with 200 mg L−1 ACC and 2 mM Ca2+

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

83

Fig. 4. Left: Isotherms of MCB adsorption on native and oxidized ACC (ACC0.8) in the absence and presence of stabilizers. Right: Freundlich plot for selected isotherms and model fit for isotherm on native ACC in the range of cMCB,w = 0.3–40 mg L−1. Conditions were adjusted in order to reach approximately 5 wt.% loading of stabilizer on ACC (HA and CMC: cstabilizer,total = 500 mg L−1, cACC = 1 g L−1, SDBS: cstabilizer,total = 5 mg L−1, cACC = 0.1 g L−1).

(fmobile = 49%). Fig. 5 shows the mobile fraction of ACC as a function of the transport distance in form of a first-order plot. Clearly, the net effect of filtration of a polydisperse mixture of particles does not necessarily have to follow a first-order process such as is described in Eq. (1) for uniform particles. Depending on the degree of variation of the filter coefficients for the individual particles and the degree of particle depletion from the mobile phase, a more or less strong upward curvature of the first-order plot might be expected. This is due to the fact that, with increasing filtration distance, the overall process will be more and more dominated by the particles with the highest mobility. In contrast, a successive agglomeration of the colloids during transport through the column, would lead to a downward curvature of the first-order plot. This type of behavior was indeed observed for the suspension with 200 mg L−1 ACC and 2 mM Ca2+ (Fig. 5), confirming the above-mentioned hypothesis of an agglomeration effect under these conditions. For all other suspensions, the depletion of ACC in the mobile phase over the entire sediment column is relatively small (≤40%), not allowing reliable conclusions on the validity of Eq. (1) using only one uniform filter coefficient. However, a filter coefficient in the range of λ = (0.2 ± 0.1) m−1 can be determined as a rough estimate. The column experiment showed that a good mobility of the HA-stabilized ACC in aquifer sediment can be obtained if suspensions with low ACC and moderate Ca2+ concentrations (e.g. ≤200 mg L−1 ACC and ≤ 2 mM Ca2+) are used. However, suspension stability under flow-through conditions is obviously

a limiting factor. The ACC feeding was stopped after a total ACC loading of 0.10 wt.% had been reached for the sediment. This value was calculated from the mass balance of ACC, i.e. from the amounts of ACC injected and recovered in the effluent. Visually, a homogeneous blackening of the whole column was observed. Tracer tests with KNO3 (pulse injection) as conservative tracer before and after the deposition of ACC on the sediment revealed identical breakthrough curves, indicating that effective porosity and dispersivity of the sediment column were not significantly affected.

3.6. Mobility of CMC-stabilized ACC in native sediment For hard-water conditions and higher ACC concentration, where HA is not sufficiently effective (instable suspension at cACC = 5 g L−1, cCa2+ = 2 mM and cHA = 2.5 g L−1, see Table 2), CMC is more appropriate: With CMC as stabilizer ACC

Table 3 Loading of size-fractionated sediment (Z1, dp: 0.25–1 mm) with ACC0.8, column: l = 100 cm, d = 2.6 cm, u = 2.3 cm min−1. Composition of mobile phase (pH 7) [Ca 1 2 1 2

2+

] in mM

−1

fmobile in %

[ACC0.8] in mg L

[HA] in mg L

100 100 200 200

50 50 100 100

−1

78 75 78 46

± ± ± ±

8 8 1 7

Fig. 5. Semi-logarithmic plot of the residual fraction of ACC0.8 in the mobile phase as function of the transport distance (L) in a column (C2) filled with aquifer sediment (Z1). Mobile phase (pH 7) contained either 100 mg L−1 ACC0.8 and 50 mg L−1 HA or 200 mg L−1 ACC0.8 and 100 mg L−1 HA and additionally 1–2 mM Ca2+ in deionized water. The solid line shows the trend for λ = 0.2 m−1 in Eq. (1). Error bars are shown exemplarily for two data sets.

84

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

suspensions remained stable even at ACC concentrations up to 11 g L− 1 and high concentrations of divalent cations (5 mM Ca2 +) reflecting conditions for very hard water (according to classification of U.S. Geological Survey (USGS, 2013)). Thus, a column experiment using very hard tap water (cCa2+ + cMg2+ = 3.4 mM) as mobile phase and suspension medium for ACC was conducted in order to study the mobility of CMC-stabilized ACC under conditions unfavorable for particle mobility in field applications. Native aquifer sediment from which only the coarse fraction N 5 mm had been removed (Z2) was used to fill the column (C3, l = 100 cm). After recording the breakthrough curve for nitrate and subsequent flushing with mobile phase, 1.8 L of a suspension of ACC (10 g L−1) and 2.5 g L−1 CMC as stabilizer in the same tap water were flushed through the column followed by flushing with tap water. Fig. 6 shows the breakthrough curves for nitrate and the TOC profile during elution of the CMC-stabilized ACC suspension at the outflow of the 1 m column. In the plateau region of the ACC breakthrough curve, the fraction of mobile ACC was 82%. No significant depletion of ACC at sampling points in the first third of the column was observed. ACC elution reaches its maximum value slightly retarded compared to the nitrate tracer, which can be due to I) contribution of reversible particle attachment or II) a higher filter coefficient in the preceding particle front. The latter can be due to the mixing of CMC-stabilized suspension with the CMC-free mobile phase or preferential sites for particle deposition which are gradually saturated. The column experiment was continued with further steps of ACC injection (1.8 L each), one again under identical conditions and two with tap water additionally enriched with 2 mM Ca2+. In all cases, the mobile fraction of ACC was ≥ 75% and the shape of the TOC breakthrough profile was similar. 3.7. Implications for in-situ barrier formation The obtained results show that colloid stabilization by means of CMC is favorable in cases where requirements for particle mobility are demanding due to one or several of the

following conditions: I) highly concentrated ACC suspensions (up to 1 wt.%) are to be applied, II) the radius of influence of ACC injection into the subsurface is to be maximized and the site groundwater and/or water available for preparation of the suspension has a high hardness. Based on the results of the column test, a filter coefficient of λ = 0.20 m−1 can be estimated for the transport of CMC-stabilized ACC with d50 = 0.8 μm in the native sediment and very hard water as suspension medium. Assuming this filter coefficient to be valid within a real aquifer, 99% depletion from water would be reached within a transport distance of 23 m. This result illustrates the good mobility of ACC which offers prospects for installation of in-situ barriers by a minimal number of injections. However, it should be emphasized that in practice, the transport distance of injected ACC within the aquifer is expected to be significantly lower. This is due to the fact that typical groundwater flow rates are 1–2 orders of magnitude lower than the pore velocity applied in our column experiment (2.2 × 10−4 m s−1), which is more representative for the conditions close to an injection point. Filter coefficients will be higher at lower pore water velocities (Elimelech et al., 1995; Yao et al., 1971). Furthermore, dilution of the injected suspension with groundwater leads to desorption of the stabilizer CMC from ACC and thus to an increasing deposition tendency of the colloids. Predictions on ACC mobility for a specific contaminated site will require detailed laboratory studies with the site sediment and groundwater. Cascading column experiments (Comba and Braun, 2012) are a feasible option to study particle mobility simulating the conditions around an injection well which are characterized by a decrease in flow velocity and particle concentration with increasing distance from the injection well. Even though CMC obviously shows a more robust stabilization effect than HA, other aspects might be considered when choosing the optimal stabilizer for a specific site (e.g. if conditions are already favorable for particle mobility (soft water)). CMC and HA are both environmentally benign materials: HA is a natural polyelectrolyte and CMC is even offered in food-grade quality. Coal-derived HA as used in this study is available at large quantities at low price (about 1000 $/t dry sodium or potassium humate). The prize for CMC (food or technical grade) is somewhat higher (about 2000–3000 $/t). Further studies are needed in order to compare and evaluate the impact of the stabilizer on the long-term behavior of the particles, including interactions with microorganisms and possible long-range transport of particles. In case of functionalization of the ACC colloids, the interference of the stabilizer with the active component is another important factor to be studied. 3.8. Column studies on retardation of contaminants

Fig. 6. Breakthrough curves for nitrate (solid line) and CMC-stabilized ACC0.8 (■) for a column (C3, l = 100 cm) filled with native aquifer sediment (Z2, dp ≤ 5 mm) operated at u = 1.3 cm min−1. Nitrate breakthrough: 30 mL 0.2 M KNO3 injected within 3 min into the mobile phase flow, mobile phase: tap water (cCa2+ + cMg2+ = 3.4 mM), ACC breakthrough: feeding of 1.8 L suspension consisting of 10 g L−1 ACC, 2.5 g L−1 CMC in tap water, afterwards again flushing with tap water. Influent TOC (10.2 g L−1, dotted line) consists of 9.4 g L−1 POC from ACC and 0.83 g L−1 DOC from CMC.

Tracer tests were performed in order to verify the efficiency of contaminant retardation by the immobilized ACC. The chloromethanes CF and DCM were used as sorptive tracers with a relatively low sorption affinity in order to allow monitoring of complete breakthrough curves within shortterm column experiments. For characterization of the retardation effect of the native sediment before ACC loading, the more hydrophobic toluene was used in addition. Firstly, single-component adsorption isotherms of the model compounds on ACC were determined and fitted by the Freundlich equation (Fig. S3, Supplement. Mat.). Freundlich

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

paramaters were KF,DCM = 1900 (mg kg− 1)(L mg− 1)1/n and 1/n = 0.69 for DCM and KF,CF = 5600 (mg kg−1)(L mg−1)1/n with 1/n = 0.72 for the more hydrophopbic CF (log KOW,DCM = 1.2 versus log KOW,CF = 1.97). The column with size-fractionated native sediment (Z1, size fraction: 0.25–1 mm) which was described in Table 3 and Fig. 5 was applied for contaminant retardation studies before and after immobilization of ACC (mean ACC loading on sediment: 0.10 wt.%). The native sediment showed a very low retardation of HOC (see Fig. S4, Supplementary Material). The retardation factors which were determined from the center of gravity elution volume of the sorptive and the non-reactive tracer (KNO3) according to exp

Ri

flow conditions, we wanted to set the observed retardation effect in relation to the equilibrium sorption data determined in batch experiments. Optimal exploitation of the ACC would be achieved if all deposited ACC is accessible for contaminant adsorption and quasi instantaneous achievement of adsorption/ desorption equilibria in relation to transport velocity is given. The retardation of a sorptive tracer during transport through a porous medium can be related to its sorption coefficient on the solid phase. Eq. (4) describes the most simple case of a rapidly established sorption equilibrium characterized by a linear isotherm. The retardation factor (Ri) represents the ratio between the linear velocities of the flowing groundwater (u) and of a compound affected by sorption to the sediment (ui). ρ (kg L− 1) and ε (0…1) are bulk density and porosity of the sediment. Kd,i,sed is the equilibrium sorption coefficient of the sorbate i on the sediment, i.e. the ratio between equilibrium sediment loading qi,sed (mg kg−1) and the aqueous phase concentration ci,w (mg L−1, Eq. (5)).

gr

¼

Vi

3

gr V KNO3

Ri ¼

Ri ¼

qi;sed ci;w

5

u ρ ¼ 1 þ f ACC K d;i;ACC ui ε

6

However, in case of a non-linear sorption isotherm, as it is usually observed for AC, Kd,i,ACC varies with ci,w. In principle, numerical transport models have to be applied in such cases.

1.4

12

1.2

10

1.0

50 cm 100 cm

CF

-1

6

4

Since the sorption potential of the native sediment is very low, the sorption coefficients of the sorptive tracers on the sediment after immobilization of ACC with a mass fraction of fACC = 0.001 can be described by Kd,i,sed = fACC · Kd,i,ACC with Kd,i,ACC = qi,ACC/ci,w leading to Eq. (6).

14

DCM

u ρ ¼ 1 þ K d;i;sed ui ε

K d;i ¼

c in mg L

c in mg L

-1

were 1.2 for CF and still only 1.5 for the more hydrophobic toluene (log KOW,toluene = 2.69). The elution curves of DCM and CF after immobilization of 0.1 wt.% ACC on the sediment are shown in Fig. 7. In addition, Fig. S5 shows the same elution curves but plotted over exchanged pore volumes instead of total volume. Retardation factors of 6.4 and 41 were determined for DCM and CF, respectively, at the sampling port at a column length of 100 cm. The difference in the retardation factors of CF before (1.2) and after the immobilization of ACC (41) illustrates the vastly different sorption potential of the two sorbents — native sediment and ACC. While the mass fraction of ACC introduced into the sediment (0.1 wt.%) is identical with the original OCcontent of the sediment (0.1 wt.%), the increase in contaminant retardation is tremendous. The recoveries of DCM and CF in the effluent were 90% and 75%, respectively. The slightly lower recovery of CF might be caused by an incomplete tracing of the peak tailing (Fig. 7). In order to evaluate to which extent the sorption potential of ACC can be exploited under continuous

8

85

0.8 0.6

4

0.4

2

0.2 0.0

0 0

1000

2000 V in mL

3000

0

10000

20000 V in mL

Fig. 7. Breakthrough curves of DCM (left) and CF (right) detected at sampling ports at 50 cm (○) and 100 cm (Δ) distance from column inlet after pulse injection: Column: C2, l = 100 cm, d = 2.6 cm filled with native sediment (Z1) which was loaded with 0.10 wt.% ACC (see Table 3 and Fig. 5). Mobile phase: 1 mM CaCl2, pH 7, u = 2.3 cm min−1, Injection of CF and DCM: 5 mL of a solution containing 500 mg L−1 of each compound injected within 1 min into the mobile phase flow.

86

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

Nevertheless, in a simplified approach we calculated upper and lower limiting values of Kd,i,ACC based on the equilibrium isotherm data and local concentrations of the sorptive tracer in the column inlet and outlet zone (for details see Supplement. Mat.). From these values the expected range of the retardation was calculated according to Eq. (6) and factor Rimin ≤ Ri ≥ Rmax i the values are shown in Table 4. In addition, the apparent experimental value for Kexp d,i,ACC determined under flow-through conditions was calculated according to Eq. (7). exp

K d;i;ACC ¼ ðRi −1Þ

ε ρ  f ACC

7

In case of CF the experimentally determined retardation max factor of Rexp CF = 41 is very close to RCF = 44 and the apparent exp max Kd,CF,ACC differs from Kd,CF,ACC only by a factor of 0.94, while this factor is 0.68 in case of DCM. Since the two compounds were injected simultaneously competitive adsorption occurring at least in the first part of the column might cause the retardation effect to be lower than predicted from the single-component isotherms. This is especially true for the less hydrophobic DCM which can be displaced from adsorption sites by the more hydrophobic CF. Nevertheless, the fact that the experimentally determined retardation factors converge to the upper limit of the predicted range indicates that the immobilized ACC is used to nearly full capacity. Eq. (6) can also be used in order to estimate the retardation of various classes of HOC and thus get a conception of the performance of an in-situ ACC adsorption barrier. For three model contaminants, i.e. toluene, naphthalene and phenanthrene, retardation factors and useful operation times for a hypothetical barrier with a dimension of l = 10 m in the direction of groundwater flow were estimated for various ACC sediment loadings (0.1 to 1 wt.%) and shown in Table 5. This estimation implies a nearly homogeneous loading of the sediment with ACC over the barrier zone and breakthrough is approximated to occur as a step function (i.e. qi,ACC = Kd,i,ACC ∙ ci,w). Calculation of qi,ACC and Kd,i,ACC for the three model compounds is based on experimental data determined in this study (toluene) and literature data (naphthalene and phenanthrene, Dobbs and Cohen (1980)) for their single solute isotherms on typical AC products. Hypothetical groundwater concentrations (ci,w) were set at the highest aqueous phase concentrations for which isotherm data were available and reflect about 10–20% of the aqueous solubilities (25 °C) of the individual compounds. For porosity (ε) and bulk density (ρ) of the sediment, values of 0.3 and 1.75 kg L−1, respectively, were used, representing typical values for sandy aquifers (Schwarzenbach et al., 1993). For the estimation of barrier operation times according to toperation,i = l / u · Ri, the

Table 4 Experimentally determined retardation factors for CF and DCM in a column experiment with pulse injection (conditions as noted in Fig. 7). Compound

Rexp i

Rmax i

Rmin i

Kexp d,i,ACC [L kg−1]

Kmax d,i,ACC [L kg−1]

Kmin d,i,ACC [L kg−1]

CF DCM

41 6.0

44 8.3

9.5 3.5

9400 1160

10100 1700

2030 580

average linear velocity of the groundwater is assumed as u = 0.5 m d−1 and the barrier length is set to l = 10 m. For the moderately hydrophobic BTEX compounds, as represented by toluene, the lifetime of the ACC sorption barrier in the absence of any (bio)degradation process would be rather low even if a relatively high mass fraction of ACC of 1 wt.% is introduced into the sediment (Table 5). In contrast, phenanthrene from the group of 3-ring PAHs shows a remarkable retardation already for a barrier with an ACC mass fraction on sediment of 0.1 wt.% and the hypothetical lifetime of the barrier is about 170 years. These simple model calculations clearly provide only a rough estimate and cannot cope with possible adverse long-term effects (e.g fouling by NOM, microbial or inorganic deposits). Nevertheless, they give a first impression of the limitations but also the great potential of in-situ ACC barriers especially for PAHs with three and more rings and other very hydrophobic contaminants. In order to facilitate the degradation of adsorbed contaminants, ACC can be loaded with catalysts or reagents as has been realized already with the ACC-iron composite Carbo-Iron® for reductive dehalogenation (Mackenzie et al., 2012). On the other hand, adsorption on activated carbon can promote biodegradation, e.g. by lowering aqueous phase concentrations of toxic pollutants (Annadurai et al., 2002). In water treatment, ACs colonized by biofilms, i.e. biological activated carbon, have proved to be efficient in the biodegradation of various organic contaminants (Aktas and Cecen, 2007). A similar effect is claimed for the new material PlumeStopTM, which in addition to the AC adsorbent contains low solubility / controlled availability matrix nutrients for improving biodegradation (Regenesis, 2014). However, further verification in field tests is needed in order to evaluate whether retardation of pollutants within an aquifer zone enriched by ACC can be a technique for enhanced natural attenuation. 4. Conclusions Colloidal activated carbon with a particle size in the lower μm-range (d50 = 0.8 μm) was obtained by means of a wetgrinding procedure with the aim to obtain an injectable material for in-situ installation of sorption or reaction barriers in contaminated aquifers. The suspension stability of the native ACC and also its mobility in saturated porous media were not sufficient for the intended use. However, these properties could be greatly improved by the addition of coal-derived HA or CMC to the ACC suspension. After adsorption to the outer particle surface of ACC, these negatively charged polyelectrolytes provide electrosteric stabilization without affecting the adsorption potential of ACC towards the target contaminants. HA is a suitable stabilizer if suspensions with low ACC and moderate Ca2+ concentrations (e.g. ≤200 mg L−1 ACC and ≤ 2 mM Ca2+) are used. In contrast, the use of CMC as stabilizer allows the application of ACC at a concentration of 10 g L−1 providing a sufficiently high mobility in watersaturated sediment. For the in-situ installation of sorption barriers, the injection of ACC suspensions stabilized with CMC appears to be the most promising approach. If ACC shall be used as a carrier for catalysts or reagents, the choice of the stabilizer will depend also on possible effects on the reactivity of the active components. The results of column experiments indicate that the ACC deposits on the aquifer sediment are fully

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

87

Table 5 Estimation of retardation factors and useful operation times estimated for a sorption barrier of 10 m length (in direction of groundwater flow) at various amounts of ACC loading on the sediment (qACC = fACC ∗ 100%) for removal of various hydrophobic organic pollutants, single solute sorption isotherm data qi,w = f(ci,w) from this study (toluene) or from Dobbs and Cohen (1980) (naphthalene and phenanthrene). Compound class

Compound

qACC,sed (wt%)

ci,w (mg L−1)

qi,ACC (wt%)

Rtheor,i

toperation (years)

BTEX

Toluene

2-ring PAHs

Naphthalene

3-ring PAHs

Phenanthrene

0.1 1 0.1 0.5 0.1

40 40 7 7 0.2

13 13 6.4 6.4 1.2

20 190 240 1200 3100

1.1 10 13 65 170

available as adsorbent for contaminants which are transported with the groundwater flow.

Acknowledgements Funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement No. 309517 (NanoRem) is acknowledged.

Appendix A. Supplementary data Supplementary material on: I) Physical properties of the applied activated carbon (Table S1), II) Mobility of native ACC in quartz sand columns (Table S2), III) Microscopic images of column filling after loading with native ACC (Fig. S1), IV) Adsorption isotherms of the applied stabilizers on ACC (Fig. S2) including description of experimental procedure, V) Adsorption isotherms of CF and DCM on ACC (Fig. S3) including description of experimental procedure, VI) Breakthrough curves of nitrate, CF and toluene for column filled with native sediment Z1 (Fig. S4), VII) Estimation of upper and lower limiting values for Kd,i,ACC for DCM and CF in the column experiment, VIII) Breakthrough curves of DCM and CF after deposition of ACC on native sediment (Fig. S5). Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j. jconhyd.2015.05.002.

References Aktas, O., Cecen, F., 2007. Bioregeneration of activated carbon: a review. Int. Biodeterior. Biodegrad. 59 (4), 257–272. Annadurai, G., Juang, R.S., Lee, D.J., 2002. Biodegradation and adsorption of phenol using activated carbon immobilized with Pseudomonas putida. J. Environ. Sci. Health, Part A: Tox. Hazard. Subst. Environ. Eng. 37 (6), 1133–1146. Bhatnagar, A., Hogland, W., Marques, M., Sillanpaa, M., 2013. An overview of the modification methods of activated carbon for its water treatment applications. Chem. Eng. J. 219, 499–511. Bleyl, S., Kopinke, F.D., Mackenzie, K., 2012. Carbo-Iron® - synthesis and stabilization of Fe(0)-doped colloidal activated carbon for in situ groundwater treatment. Chem. Eng. J. 191, 588–595. Busch, J., Meißner, T., Potthoff, A., Oswald, S.E., 2014a. Transport of carbon colloid supported nanoscale zero-valent iron in saturated porous media. J. Contam. Hydrol. 164, 25–34. Busch, J., Meissner, T., Potthoff, A., Oswald, S.E., 2014b. Investigations on mobility of carbon colloid supported nanoscale zero-valent iron (nZVI) in a column experiment and a laboratory 2D-aquifer test system. Environ. Sci. Pollut. Res. 21 (18), 10908–10916. Comba, S., Braun, J., 2012. A new physical model based on cascading column experiments to reproduce the radial flow and transport of micro-iron particles. J. Contam. Hydrol. 140, 1–11.

Comba, S., Di Molfetta, A., Sethi, R., 2011. A comparison between field applications of nano-, micro-, and millimetric zero-valent iron for the remediation of contaminated aquifers. Water Air Soil Pollut. 215 (1-4), 595–607. Crane, R.A., Scott, T.B., 2012. Nanoscale zero-valent iron: future prospects for an emerging water treatment technology. J. Hazard. Mater. 211, 112–125. Dalla Vecchia, E., Luna, M., Sethi, R., 2009. Transport in porous media of highly concentrated iron micro- and nanoparticles in the presence of xanthan gum. Environ. Sci. Technol. 43 (23), 8942–8947. Dobbs, R.A., Cohen, J.M., 1980. Carbon adsorption isotherms for toxic organics. U.S. EPA Report EPA-600/8-80-023 (available via http://nepis.epa.gov). Elimelech, M., Jia, X., Gregory, J., Williams, R.A. (Eds.), 1995. Particle Deposition & Aggregation. Butterworth-Heinemann, Woburn, MA. He, F., Zhao, D.Y., 2007. Manipulating the size and dispersibility of zerovalent iron nanoparticles by use of carboxymethyl cellulose stabilizers. Environ. Sci. Technol. 41 (17), 6216–6221. He, F., Zhao, D.Y., Liu, J.C., Roberts, C.B., 2007. Stabilization of Fe–Pd nanoparticles with sodium carboxymethyl cellulose for enhanced transport and dechlorination of trichloroethylene in soil and groundwater. Ind. Eng. Chem. Res. 46 (1), 29–34. Julien, F., Baudu, M., Mazet, M., 1998. Relationship between chemical and physical surface properties of activated carbon. Water Res. 32 (11), 3414–3424. Mackenzie, K., Hildebrand, H., Kopinke, F.-D., 2007. Nano-catalysts and colloidal suspensions of Carbo-Iron for environmental application. NSTI Nanotech 2007, Santa Clara (CA): Technical Proceedings. CRC Press, pp. 639–642. Mackenzie, K., Schierz, A., Georgi, A., Kopinke, F.D., 2008. Colloidal activated carbon and Carbo-Iron — novel materials for in-situ groundwater treatment. Glob. NEST 10 (1), 54–61. Mackenzie, K., Bleyl, S., Georgi, A., Kopinke, F.D., 2012. Carbo-Iron – an Fe/AC composite – as alternative to nano-iron for groundwater treatment. Water Res. 46 (12), 3817–3826. Matsui, Y., Ando, N., Sasaki, H., Matsushita, T., Ohno, K., 2009. Branched pore kinetic model analysis of geosmin adsorption on super-powdered activated carbon. Water Res. 43 (12), 3095–3103. Moreno-Castilla, C., Carrasco-Marin, F., Maldonado-Hodar, F.J., Rivera-Utrilla, J., 1998. Effects of non-oxidant and oxidant acid treatments on the surface properties of an activated carbon with very low ash content. Carbon 36 (1-2), 145–151. Phenrat, T., Saleh, N., Sirk, K., Tilton, R.D., Lowry, G.V., 2007. Aggregation and sedimentation of aqueous nanoscale zerovalent iron dispersions. Environ. Sci. Technol. 41 (1), 284–290. Phenrat, T., Saleh, N., Sirk, K., Kim, H.J., Tilton, R.D., Lowry, G.V., 2008. Stabilization of aqueous nanoscale zerovalent iron dispersions by anionic polyelectrolytes: adsorbed anionic polyelectrolyte layer properties and their effect on aggregation and sedimentation. J. Nanopart. Res. 10 (5), 795–814. Plagentz, V., Ebert, M., Dahmke, A., 2006. Remediation of ground water containing chlorinated and brominated hydrocarbons, benzene and chromate by sequential treatment using ZVI and GAC. Environ. Geol. 49 (5), 684–695. Rakowska, M.I., Kupryianchyk, D., Harmsen, J., Grotenhuis, T., Koelmans, A.A., 2012. In situ remediation of contaminated sediments using carbonaceous materials. Environ. Toxicol. Chem. 31 (4), 693–704. Regenesis, 2014. PlumeStop™ Colloidal Biomatrix Whitepaper. http://www. plumestop.com/. Saleh, N., Kim, H.J., Phenrat, T., Matyjaszewski, K., Tilton, R.D., Lowry, G.V., 2008. Ionic strength and composition affect the mobility of surface-modified Fe-0 nanoparticles in water-saturated sand columns. Environ. Sci. Technol. 42 (9), 3349–3355. Schrick, B., Hydutsky, B.W., Blough, J.L., Mallouk, T.E., 2004. Delivery vehicles for zerovalent metal nanoparticles in soil and groundwater. Chem. Mater. 16 (11), 2187–2193. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 1993. Environmental Organic Chemistry. John Wiley and Sons Inc., New York.

88

A. Georgi et al. / Journal of Contaminant Hydrology 179 (2015) 76–88

Sunkara, B., Zhan, J.J., He, J.B., McPherson, G.L., Piringer, G., John, V.T., 2010. Nanoscale zerovalent iron supported on uniform carbon microspheres for the in situ remediation of chlorinated hydrocarbons. ACS Appl. Mater. Interfaces 2 (10), 2854–2862. Tungittiplakorn, W., Cohen, C., Lion, L.W., 2005. Engineered polymeric nanoparticles for bioremediation of hydrophobic contaminants. Environ. Sci. Technol. 39 (5), 1354–1358. USEPA, 2002. Field Applications of In Situ Remediation Technologies: Permeable Reactive Barriers. United States Environmental Protection Agency, Washington, DC, USA (http://clu-in.org/download/rtdf/fieldapp_prb.pdf). USGS, 2013. USGS Water-Quality Information. http://water.usgs.gov/owq/ hardness-alkalinity.html.

Xu, R., 2000. Particle characterization: light scattering methods. In: Scarlett, B. (Ed.), Particle Technology Series vol. 13. Springer, Netherlands. Yao, K.M., Habibian, M.M., Omelia, C.R., 1971. Water and waste water filtration — concepts and applications. Environ. Sci. Technol. 5 (11), 1105–1112. Zhan, J.J., Zheng, T.H., Piringer, G., Day, C., McPherson, G.L., Lu, Y.F., Papadopoulos, K., John, V.T., 2008. Transport characteristics of nanoscale functional zerovalent iron/silica composites for in situ remediation of trichloroethylene. Environ. Sci. Technol. 42 (23), 8871–8876. Zhang, W., Elliott, D.W., 2006. Applications of iron nanoparticles for groundwater remediation. Remediat. J. 16, 7–21.