Anthropogenic activities such as combustion of fossil fuels, oil spills, and increased industrialization have resulted in widespread pollution of aquatic sediments with a variety of hydrophobic organic compounds (HOCs). Over the last decade, in situ addition of activated carbon (AC) to polluted sediments has been proposed as an effective remediation strategy to reduce risks of sediment-bound HOCs and to improve the ecological quality of surface waters 1, 2. Sediment remediation with AC has been shown to result in reduced freely dissolved HOC concentrations in sediment porewater in laboratory and field settings 3–7 and in reduced bioaccumulation and toxicity of HOCs in benthic invertebrates 5, 6, 8.
The key to the effectiveness of AC in sediment remediation is its high sorption capacity, which is caused by a high specific surface area, well developed porosity, and dedicated surface chemistry. Consequently, the majority of AC amendment studies use powdered AC (PAC) or fine granular AC (GAC) having particle diameters within the range of 75 to 300 µm. Small particles have cumulatively large external surface area and, therefore, large accessible numbers of sorption sites per unit of weight 9. Recently, risk reduction of sediment-bound HOCs in pilot field studies has been accomplished through addition of AC to the sediment biologically active layer (upper 30 cm) using commercial mixing units, by capping with a thin layer of AC or by mixing AC in a layer of clay 3, 4, 10, 11.
The science underlying AC application to remediate polluted sediments is rapidly evolving, and granular activated carbon (GAC) may be an effective alternative to PAC in in situ soil remediation 12. In The Netherlands, intensive ex situ treatment of contaminated sediments with GAC has been recognized as a promising and novel technique, which might also be relevant in other regions. The treatment includes the following steps: addition of GAC to a sediment slurry or stream; extraction of HOCs from sediment by GAC; separation of contaminated GAC from the cleaned sediment slurry; regeneration of GAC by thermal treatment; and reuse of cleaned sediment in the construction of, for instance, dikes, roads, or highways. Separation of GAC from sediment at high volumes is feasible using industrial sieving units (Supplemental Data, Figs. S1 and S2).
Regeneration of GAC after sediment treatment reduces the treatment costs and contributes to the reduction of the AC in the environment. Furthermore, the reactivation step might modify carbon structural properties in a manner that enhances sorption of organic compounds 13. The number of studies using GAC particles larger than 0.3 mm in the context of sediment remediation is limited. Previous studies have estimated the effectiveness of GAC to immobilize HOCs in bed sediments 6 and slowly mixed systems 7. Under these conditions, GAC has been shown to be less effective than PAC. However, in an intensive addition–removal scenario, in which mixing conditions are more efficient and removal is by particle separation (sieving), GAC is the preferred candidate sorbent. In a recent field-scale pilot study in The Netherlands, 48 h mixing in a batch operational mode, and subsequent removal of GAC (Supplemental Data, Fig. S2) resulted in a reduction of the concentration of total polycyclic aromatic hydrocarbons (PAHs) by a factor of 2 to 4.
The effectiveness of PAC or GAC in binding sediment-associated HOCs is strongly related to the amount and nature of sediment phases that compete with AC for the binding of PAH. These include amorphous organic carbon, mineral particles, oil, black carbon (BC), and wheathered oil residues 14, 15. Recently, Kupryianchyk et al. 15 showed how limitations in AC performance can be quantified based on organic carbon and BC content. Refractory phases such as BC and weathered oil might significantly limit desorption of HOCs to AC 16, 17. However, a substantial fraction of HOCs is known to be associated with light-density organic carbon 18 composed of humic/fulvic substances, lignin, and plant debris or mineral particles 19. This fraction of HOCs can be expected to desorb readily and bind to AC particles in a sediment slurry. However, it is also known that natural organic matter (NOM) and oil might attenuate sorption to AC by pore blockage or sorption competition by coadsorbing HOCs, NOM molecules, and oil 20–23. Because the added GAC would compete for the sorption of HOCs with natural sediment phases, its effectiveness would strongly depend on dosage. Consequently, distribution coefficients for short-term sorption processes and dosage level of GAC are likely to be important parameters controlling the effectiveness of intensive sediment remediations with GAC.
The aim of the present study was to test this hypothesis and further explore the potential of GAC in the context of ex situ sediment remediation technology. To this end, the optimal dosage of GAC in terms of PAH porewater reduction in sediment for ex situ remediation was determined. Thus, the present study evaluated GAC effectiveness in the presence of associated organic matter. A worst-case sediment with very high PAH and oil pollution levels was selected, to explore the upper limits of the capacity of GAC. Initially, a suite of candidate GAC materials was screened for maximum efficiency in extracting PAHs from sediment within 24 h. The effectiveness of GAC was compared with a single-step solid phase extraction (SPE) with Tenax beads, which currently is the most commonly applied material to remove HOCs from sediments under laboratory conditions 24. Sorption of native PAHs to the best performing GAC was studied for mixtures at different GAC–sediment weight ratios and sediment only. For all these experiments, aqueous-phase PAH concentrations were accurately determined using 76-µm polyoxymethylene (POM) passive samplers 8, 25, 26. The sorption to GAC in the sediment–GAC mixtures was assessed by subtracting the contribution of PAH sorption to sediment from total PAH sorption in the mixture. Sorption data were interpreted in terms of aqueous-phase concentration reduction ratios and distribution coefficients.
MATERIALS AND METHODS
Hexane and acetone (Promochem; picograde), methanol (Mallinckrodt Baker; high performance liquid chromatography [HPLC] gradient grade), ethanol (Merck; p.a.), acetonitrile (Lab-Scan; HPLC grade), CaCl2 (Merck; p.a), sodium azide NaN3 (Aldrich; 99%), Na2SO4 (Merck; p.a.), Na2CO3 (Merck; p.a.), NaHCO3 (Merck; p.a.), and aluminum oxide-Super I (ICN Biomedicals) were used in the experiments. Prior to use, aluminum oxide was deactivated with 10% (w/w) Nanopure water (Barnstead). Other water preparations used Milli-Q water (Millipore Corporation). High performance liquid chromatography standard EPA 610 PAH Mix containing acenaphthene (ACE), acenaphthylene (ACY), anthracene (ANT), benzo[a]anthracene (BaA), benzo[a]pyrene (BaP), benzo[b]fluoranthene (BbF), benzo[ghi]perylene (BghiP), benzo[k]fluoranthene (BkF), chrysene (CHR), dibenzo[>a,h]anthracene (DBA), fluoranthene (FLU), fluorene (FL), indeno[1,2,3-cd]pyrene (InP), naphthalene (NA), phenanthrene (PHE), and pyrene (PYR) was obtained from Supelco Analytical, with purities of >98%, except for PYR (purity >96.6%).
Sediment was dredged in 1997 from Amsterdam Petroleum Harbor (PH), The Netherlands, which was constructed for storage and transshipment of petroleum and coal, and stored at 4°C in the dark. The sediment was passed through a 2-mm sieve and well homogenized prior to sorption experiments. The sediment was characterized for dry weight, total organic carbon (TOC), black carbon (BC), total native content of PAHs, and total native petroleum hydrocarbons (TPH). A summary of sediment characteristics is provided as Supplemental Data, Table S1.
Granular ACs (1240W, 840W) and extruded carbons (RB 3W, RB 4W) were kindly provided by Norit. The Brunauer–Emmet–Teller (BET) and Barret–Joyner–Halenda (BJH) surfaces and pore widths were measured for all GAC types (Supplemental Data,Table S2) using Tristar micrometrics apparatus. The total pore volume (Vtot) was determined from the amount of nitrogen vapor adsorbed as a function of relative pressure (p/po = 0.9610). The surface area was derived according to the BET model (SBET). An average pore width, wads (4Vtot/SBET), was derived from the total pore volume and SBET. Prior to the experiments, GACs were washed sequentially with warm (80°C), demineralized water to remove impurities 27. Subsequently, all carbons were dried overnight at 105°C. Surface functional groups were determined with Boehm titrations, following previously published procedures 28, 29. Experimental details and results of the Boehm titrations are provided as Supplemental Data, Table S2.
Polyoxymethylene (POM) sheets (76 µm thickness) were purchased from CS Hyde Company. Polyoxymethylene passive samplers were prepared by cutting the sheets into 300-mg strips and cold extracting them with hexane (30 min) and methanol (3 × 30 min). Sorbent pretreatment was based on earlier reported procedures for POM strips 25, 26. After extraction, POM strips were air-dried and stored in a brown glass bottle until use. Polyoxymethylene samplers were cut into smaller pieces depending on the experiment.
Comparison of extraction efficiencies of four types of GAC
Four types of GAC (GAC 1240W, 840W, RB 3W, RB 4W) were evaluated for their efficiency in extracting PAHs from PH sediment in 24 h. For comparison, a single-step 24-h SPE using Tenax beads was applied. Details of the extraction methodology are provided as Supplemental Data. Extraction efficiency was reported as residual fraction of original PAH concentration in the sediment after extraction.
Sorption experiments were performed in parallel for sediment only and for GAC − sediment mixtures at different GAC − sediment weight ratios.
Sorption to sediment
Sediment–water equilibration tests were performed to determine distribution coefficients KSED for native PAHs. This setup also served as a control for the experiments in which sediment and GAC were mixed (see below). Brown glass bottles, with volumes ranging from 0.05 to 2 L, were filled with Mili-Q water containing 100 mg/L NaN3 and 0.01 M CaCl2 and increasing mass, that is, 0.75, 3.75, 7.50, 18.75, 37.50, and 75.00 g (dry wt) of PH sediment. The bottles differed in total suspension volume, but the solid–liquid ratio was identical (37 g/L) in the systems, except for the batch with the lowest mass of sediment, which had a solid–liquid ratio of 30 g/L. Polyoxymethylene samplers (0.07–6.7 g depending on the batch) were added, and the bottles were horizontally shaken at 160 strokes per min for 28 d. This period is sufficient to reach equilibrium in 76-µm POM strips 25. Subsequently, the POM strips were removed, rinsed, and wiped with a tissue before extraction. Equilibrium aqueous-phase PAH concentrations were calculated from the PAH concentrations in the POM samplers using previously published POM–water partitioning coefficients (KPOM) 25.
Sorption to mixtures of granular activated carbon and sediment
Sorption of native PAHs to GAC–sediment mixtures (KMIX, see Eqn. 2 below) was determined by using the same procedure as for sediment only (see above). In brief, bottles with a volume between 0.05 and 2 L received the previously described aqueous NaN3/CaCl2 solution and the same series of 0.75 to 75.00 g (dry wt) of PH sediment as in the sediment-only sorption test. However, now a constant dose of 0.75 g GAC was added to each batch, prior to addition of sediment. This resulted in an almost identical liquid–solid ratio of the mixture (37.75 g/L) compared with that in the sediment-only experiment (37 g/L) and in highly variable GAC–sediment weight ratios of 1.0, 0.2, 0.1, 0.04, 0.02, and 0.01 corresponding to GAC doses of 50, 17, 9, 4, 2, and 1%, respectively. Subsequently, POM samplers (0.07–6.7 g) were added, and the bottles were shaken and equilibrated as described above, again for 28 d. Previous studies have used 28-d equilibration for a wide range of carbonaceous materials 26, 30. It is well known that complete HOC uptake by GAC may take longer, especially in bed sediments 6. However, for agitated suspensions, 28 d may be considered sufficient from an ex situ remediation perspective, for which equilibration times of months would be less relevant. In the present study, Freundlich and Langmuir sorption isotherms were used as conventional methods 31–34 to describe aqueous-to-sorbed concentration relations. Therefore, the associated constants obtained after 28 d of equilibration were identified as pseudoequilibrium constants (see below).
Sediment samples were dried with Na2SO4 and homogenized. Total PAH concentrations were determined by Soxhlet extraction of dry sediment samples (∼0.75 g) using 70 ml hexane/acetone (1:1) for 16 h 35. Extracts were concentrated to 1 ml, and cleaned up over aluminum oxide columns using 25 ml of hexane. Column eluates were reduced to 1 ml, switched to acetonitrile, and rereduced to 6 ml. Polyoxymethylene samplers were Soxhlet extracted with 70 ml methanol for 3 h 26. Extracts were concentrated to 1 ml and switched to hexane. Further clean-up and analysis were identical to those used for the sediment samples. Extracts from sediment or POM were analyzed via HPLC using a C18 reversed-phase column with length 250 mm and diameter 4.6 mm (Grace). Polycyclic aromatic hydrocarbons were detected with a photodiode array detector (UVD34OU; Dionex) and fluorescence detector (HP 1046A). The mobile phase consisted of acetonitrile and water.
Total native petroleum hydrocarbons extraction was performed according to NEN 5733 36, see the Supplemental Data. Sediment TOC and BC were determined in triplicate, using the chemothermal oxidation method (CTO375) as described by Gustaffson et al. 16. In this method, amorphous organic carbon is removed by thermal oxidation at 375°C followed by addition of HCl to remove possible carbonate residues. Subsequently, total remaining carbon was detected using an elemental analyzer (EA 1110 CHN; CE Instruments).
Sorption experiments were performed in triplicate in brown bottles to prevent PAH photolysis. Mass balances were validated in our previous work using the same procedures and equipment 26, 30. Numerous blank and clean-up recovery samples were included in the sediment and POM extraction procedures, that is, one blank sample per six samples and one clean-up recovery sample per five samples. The clean-up recovery samples were prepared by adding a standard PAH mixture prior to extraction. Polycyclic aromatic hydrocarbon recoveries ranged between 81.21 ± 6.97% (InP) and 108.34 ± 6.23% (BaP; n = 7), depending on the compound.
Sediment–water distribution coefficients (L/kg) for individual PAHs were calculated as
where CS (µg/kg) is the solid-phase concentration (Soxhlet extraction) and CW (µg/L) is the freely dissolved aqueous-phase concentration (passive samplers). The partitioning of PAHs in the mixture of sediment and GAC can be described with the following equation
where KMIX is a conditional GAC/sediment-to-water distribution coefficient, KSED is the sediment–water distribution coefficient (Eqn. 1), and KGAC is the apparent GAC–water distribution coefficient in the presence of sediment. Note that the conditional KMIX and KGAC relate to the same system. Using fSED + fGAC = 1 and rewriting yields the following
Substitution of the relationships for the remaining distribution ratios (i.e., KGAC = CGAC/CWKMIX = CMIX/CW) allows calculation of the concentration of PAHs in GAC
where CMIX and CGAC are the concentrations of PAHs in the sediment–GAC mixture and in GAC only (µg/kg). Where applicable, a Freundlich isotherm model was used to describe the 28-d pseudoequilibrium sorption data
where KF,GAC is a conditional Freundlich affinity pseudoequilibrium constant (µg(1–n) Ln kg−1) and nF,GAC is the Freundlich pseudoequilibrium exponent for sorption to GAC. Isotherms were also evaluated using the Langmuir sorption model
in which, CMAX is the maximum loading of the carbon (µg/kgGAC) and b is the Langmuir sorption pseudoequilibrium constant (L/µg). Experimental data were analyzed with PASW Statistics 17.0 (SPSS).
RESULTS AND DISCUSSION
Total PAH concentration measured in PH sediment (13 EPA PAH) was 1,604 ± 90.2 mg/kg, which illustrates the high level of pollution of the PH sediment. Polycyclic aromatic hydrocarbon concentration ratios ANT/178, FLU/(FLU + PYR), BaA/228, and InP/(InP + BghiP) were 0.66, 0.62, 0.47, and 0.53, respectively, which indicates the PAH contamination was of pyrogenic origin 37. Individual PAH concentrations and total TPH content are summarized in Supplemental Data, Table S1. The FL (6.3%), PHE (19.1%), ANT (7.3%), FLU (20.3%), PYR (12.5%), BaA (6.7%), and CHR (6.1%) dominated the sum PAH concentration in PH sediment. Measurements with POM passive samplers showed that a subset of PAH made up 99.90% of the freely dissolved fraction, with individual contributions of FL (21.3%), PHE (60.2%), ANT (5.7%), FLU (7.8%), PYR (4.5%), BaA (0.3%), and CHR (0.2%). Aqueous-phase concentrations of the aforementioned compounds exceeded Dutch water quality criteria 38. The PH sediment had a high native BC content of 1.85%, which resulted in a BC/TOC ratio of 0.415 (Supplemental Data, Table S1).
Characterization and extraction efficiency of four types of GACs
The characteristics of GACs varied among the four tested types (GAC 1240W, GAC 830W, GAC RB 3W, and GAC RB 4W; Supplemental Data, Table S2). Single-step 24-h SPEs with four types of GAC and Tenax were performed to determine the differences in residual fractions of PAHs in PH sediment. The Tenax extraction served as a control test for the selection of the best performing GAC and allowed a direct comparison of all tested carbons. As shown in Figure 1, PAH extraction efficiency, in decreasing order, was Tenax, GAC 1240W, GAC 830W, GAC 3W, and GAC 4W. The differences in extraction efficiency were most pronounced for four-ring PAHs, with almost a twofold higher residual concentration after GAC 4W compared with Tenax. The significance of the differences between GAC and Tenax extraction results was tested by ANOVA (multiple comparisons, Tukey HSD, p < 0.05; provided as Supplemental Data, Table S3). It appears that Tenax is superior to the best performing GAC 1240W only for four-ring PAHs (p = 0.001), although the difference in residual mass is small (0.43 ± 0.03 vs 0.51 ± 0.02; Fig. 1). For three-ring PAHs, no significant differences were observed after extraction with GAC 1240W, 830W, or Tenax (Supplemental Data, Table S3). Residual fractions of five- and six-ring PAHs were similar after treatment with Tenax and all tested GACs (Fig. 1).
Polycyclic aromatic hydrocarbon extraction efficiency decreased with increasing GAC particle size (Fig. 1 and Supplemental Data, Table S2). This is consistent with a pore sorption mechanism in which smaller particles have larger external surface areas and shorter diffusion path lengths, leading to faster uptake. The GACs had comparable BET total surface areas, ranging from 997 to 1,112 (m2/g). However, the BJH adsorption cumulative surface area of pores between 1 and 25 nm radius were different for the GACs, except for GAC 1240W and 830W (Supplemental Data, Table S2). The higher extraction efficiency of GAC 1240W and 830W may be explained by the factor of two and four larger surface area of their pores compared with GAC RB 3W and RB 4W, respectively. Furthermore, total volume of pores (<25 nm) and cumulative volume of pores between 1 and 25 nm for 1240W and 830W were 1.3 and 2.5 to 3.7 times larger than for the extruded GACs (RB 3W and RB 4W; Supplemental Data, Table S2). Granular ACs 1240W and 830W provided the largest adsorption average pore width, whereas RB3W had the largest BJH adsorption average pore radius, among all studied sorbents. Consequently, the poorest extraction efficiency was obtained for 3- and 4-mm extruded carbons (GAC RB 3W and RB 4W). These extruded C particles require longer diffusional times for PAH uptake and may also be more sensitive to NOM fouling and sorption competition by complex organic molecules than GAC 1240W and 830W. Residual percentages of total (13 EPA PAH) ranged from 51 to 72%; GAC 1240W had the best overall efficiency (lowest residual concentration; Supplemental Data, Fig. S3), which is why GAC 1240W was selected as a superior candidate for further testing in the context of ex situ sediment remediation. Although particle size of GAC 1240W was much greater than that of Tenax beads, GAC 1240W extraction resulted in an almost identical residual fraction of PAH compared with that of Tenax extraction (Fig. 1 and Supplemental Data, Fig. S3). Notably, the single-step 24-h SPE with GAC 1240W may serve as a preliminary and rapid method for determination of PAH removal efficiency from any other sediments prior to ex situ remediation.
For all GACs tested, Boehm titrations indicated a high content of basic functional groups and low contribution of acidic groups, which is in agreement with the C surface reactivity toward nonpolar organic compounds such as PAHs (Supplemental Data, Table S2). However, GAC 1240W had the smallest amount of basic functional groups among all tested carbons. Previous studies have emphasized the importance of C surface chemistry in the adsorption of organic compounds 39, 40. In general, it is believed that high surface acidity reduces adsorption of nonpolar compounds. However, specific types of oxygen-containing functional groups might indicate variations in reactivity, for example, because of direct interactions with neighboring groups of the same or different type, or as a result of interactions between the electron-rich regions located in the graphene layers with the π electrons of more distant functional groups. Apparently, the differences in chemical properties of GAC 1240W and GAC RB 4W (0.083 vs 0.133 mmol/g; Supplemental Data, Table S2) are less important than the particle size effect.
Sorption to sediment
Polycyclic aromatic hydrocarbon partitioning to sediment was measured using the POM-SPE method 25, 26 (Supplemental Data, Table S4). Replication between systems of equal volume was excellent with (n = 3) standard deviation (SD) in log KSED of 0.01 to 0.1 for all PAHs. An exception was DBA, for which SD ranged from 0.14 to 0.18 (Supplemental Data, Table S4). Small errors in replicated log KSED and log KMIX values are crucial for accurate estimation of log KGAC values in the mixture (see below). The log KSED values were measured at almost identical solid-to-liquid ratios (30–37 g/L), so a single mean log KSED value per chemical (n = 18) was calculated (Supplemental Data, Table S4). When plotted against log KOW (Supplemental Data, Fig. S4), log KSED shows a linear relationship (r2 = 0.965), with a slope of 1.37 ± 0.079 and intercept −2.76 ± 0.461. This shows that PAH sorption to PH correlates with hydrophobicity, sorbent surface area, or molecular volume, which all are chemical properties that covary with log KOW. A slope larger than 1 is often observed for PAHs in sediments containing BC and may reflect the stronger sorption of high-molecular-weight PAHs because of stronger π–π interactions with the BC surface. However, at high PAH concentration, organic matter may still constitute the main binding phase, because of sorption saturation of the BC surface. In addition, sorption to oil may contribute to the strong linear partitioning 14, 41.
Although the log KSED values were measured at almost constant liquid–solid ratios, they are not true replicates because they were measured at system volumes varying over a factor of 40 (0.05–2 L). Interestingly, the n = 3 replicated log KSED values were about 0.05 to 0.1 log units higher at the two lowest system volumes (Supplemental Data, Table S4; 0.75 and 3.75 g sediment added). Log KSED values for setups with 7.5 to 75 g sediment added were very close and showed no trend (Supplemental Data, Table S4). Log KSED values obtained from the lowest system volumes, that is, <3.75 g sediment added, were more often significantly higher than those measured in the larger volumes. For instance, the difference was significant for FL (Kruskal–Wallis, χ2(5) = 13.838, p = 0.017) but not for BbF (Kruskal–Wallis, χ2(5) = 3.015, p = 0.698; for other compounds, see Supplemental Data, Table S4).
We have no conclusive explanation for the small deviations observed at low volumes. We hypothesize that either the small sediment mass may be less representative of the sediment and therefore prone to variation or mixing conditions are different at low system volumes. The hypothesis of lower accuracy at low system volume is consistent with the observation that, for most PAHs, the errors in n = 3 replicates are slightly higher at low system volume. Note that the observed trend in KSED does not affect the validity of subtracting sediment sorption from GAC–sediment sorption in the mixtures. After all, this subtraction does not use the averaged KSED but uses the individual KSED values measured at the same solid–liquid ratio as that of the GAC–sediment mixture (Supplemental Data, Table S4).
Sorption to mixtures of granular activated carbon and sediment
Polycyclic aromatic hydrocarbon distribution coefficients for mixtures (log KMIX) of different GAC to sediment ratios (0.01–1) were measured in triplicate. Errors were small (Supplemental Data, Table S5), similar to those for the sediment-only systems (Supplemental Data, Table S4), which again supports the accuracy of the measurements (see also Supplemental Data, Fig. S4–S10). Log KMIXclearly differs for different GAC–sediment ratios and PAH hydrophobicities (Fig. 2, all data pooled). Plots for separate GAC–sediment ratios are provided as Supplemental Data, Figures S4–S10. The presence of GAC in the mixture clearly enhances the overall partition coefficient of the mixture, an effect most clear for PAHs with log KOW < 5.2 (Fig. 2). For FL (lowest log KOW), the log KSED for sediment only of approximately 3.16 increased by three orders of magnitude to a value of 6.4 in a mixture (log KMIX) at a GAC–sediment ratio of 1. The effect gradually declines with lower GAC doses. This suggests that fouling processes are also likely to affect the binding of PAHs with log KOW < 5.2. For PAHs with log KOW > 5.2, there is no considerable effect of adding GAC to sediment up to a mixing ratio as high as GAC–sediment = 0.2. Only if the ratio is increased to 1 (1:1 GAC:sediment), the distribution coefficients increase by 1 log unit (Fig. 2). The observed difference in behavior of low- and high-molecular-weight PAHs might be explained by the fact that porewater concentrations for low-molecular-weight PAHs were much higher (Supplemental Data, Table S1), and their kinetics are relatively fast. This may allow for a considerable redistribution of mobile PAHs from the sediment to the GAC present in the mixture. The limited effect of GAC addition on log KMIX for the high-molecular-weight PAHs could also be explained by strong sorption of these hydrophobic compounds to BC particles. The fact that log KMIX increases with increasing GAC–sediment ratios implies that the affinity of PAHs for GAC is higher than that for the sediment. We have no conclusive explanation for the fact that GAC does not increase log KMIX at a ratio of 0.2, whereas it does at a ratio of 1. It can be speculated that the extra factor of five effectively eliminates any competitive effect of oil or organic matter fouling of the GAC, such that only under these conditions the GAC acts as an infinite sink for the PAH present in the aqueous phase. Furthermore, it is possible that, at high GAC loading, GAC starts to suppress the KSED for the sediment phase as well, for instance, by limiting rapid exchange of PAH to amorphous organic matter fractions at higher particle densities in suspension, further increasing the competitive strength of the GAC. In other words, the distribution coefficients for fGAC/fSED = 0.01 to 0.2 obtained for PAHs with log KOW > 5.2 level off to reach the values reported for sediment without carbon amendment (Fig. 2), probably because of the aforementioned sorption competition or fouling effects of sediment on GAC. Moreover, the log KMIX may level off with increasing hydrophobicity of PAH because of intraparticle retarded diffusion in native BC, which would take much longer than the 28-d period used in the present study.
Apparent sorption to GAC in the presence of sediment was quantified by subtracting the amount of PAH sorbed to sediment without GAC treatment from the amount of PAH sorbed to the GAC–sediment mixtures (Eqns. 2–4). This assumes that PAH partitioning to pH sediment can be quantified by concentration-independent linear partitioning, as was discussed previously, and is not affected by the presence of GAC. It is most plausible that the latter assumption holds true as long as GAC sediment is <0.2. However, at the highest GAC loading, KSED might be suppressed, which, if neglected, would lead to an equal overestimation of KGAC. The apparent distribution ratios (log KGAC) vary between 4.7 and 8.8 depending on PAH hydrophobicity (log KOW) and the GAC/sediment ratio (Fig. 3). The increase in GAC–sediment ratios from 0.01 to 1 leads to an increase in KGAC of two orders of magnitude. Most importantly from a remediation perspective, even at the lowest ratio of 0.01 (1% GAC added), the GAC sorption affinity for the low-molecular-weight PAH is still one to two orders of magnitude higher than that of the sediment. Consequently, the effectiveness of GAC for PAH uptake is highest for FL, PHE, ANT, and PHE that are present in porewater at concentrations of 64, 180, 17, and 24 µg/L (Supplemental Data, see Table S1, CW) and therefore have priority in remediation.
The previous discussion is an interpretation of sorption data on the level of distribution ratios. This may be appropriate for the sediment-only and GAC sediment-mixed systems. However, sorption to GAC only, either pristine or in the presence of sediment, is nonlinear. Consequently, log CGAC was plotted against log CW to obtain isotherms for PAH sorption to GAC in the presence of sediment (Supplemental Data, Fig. S11). Note that these isotherms are conditional in that they relate to different GAC–sediment mixing ratios as well as pseudoequilibrium. The use of pseudoequilibrium and pseudokinetic models is well established in the recent literature on sorption of PAHs to GAC 31, 32, 42. Following Lesage et al. 42 and Valderrama et al. 32, we argue that, within short time intervals, sorption of PAHs takes place mainly in macro- and mesopores. The micropore compartment is only relevant at much longer time scales 42, which are not relevant for our intended GAC application. Consequently, the isotherms for sorption of seven PAHs to GAC were fitted to Freundlich and Langmuir sorption models, (Eqns. 5 and 6; Table 1 and Supplemental Data, Fig. S12), to obtain model parameters for macro- and mesopore sorption only. To our knowledge, these are the first reported Freundlich affinity pseudoequilibrium constants for PAH levels as high as 1,600 mg/kg in GAC–sediment mixtures, so we cannot evaluate these values against literature data. The conditional Freundlich pseudoequilibrium constants (log KF,GAC) ranged from 5.5 to 6.2 depending on the compound (Table 1 and Supplemental Data, Fig. S13). In addition, log KF,GAC with respect to log KOW was not significantly different (Kruskal–Wallis, χ2(6) = 6.000, p = 0.423). No correlation was found between Freundlich affinity pseudoequilibrium constants and PAH molecular volume (data not shown). When fitted to the Langmuir equation, the sorption data resulted in a quality of fit more or less equal to that for the Freundlich equation (equal r2; Table 1). Interestingly, whereas b and CMAX differ enormously among individual PAHs (Table 1), bCMAX is more or less constant for all PAHs. This term is the initial slope of the Langmuir isotherm (Eqn. 6), which agrees with the KGAC at low CW. This suggests that the first mentioned variations for b and CMAX probably are artifacts from the fitting procedure, caused by insufficiently reaching sorption saturation levels. High values of sorption parameters (CMAX), for PHE and FLU, have been reported elsewhere, although these values were obtained in clean PAH–AC systems 33, 34. The inverse correlation between maximal sorption capacity and sorbent–sorbate contact area has been reported by Van Noort et al. 33. The Langmuir sorption parameters for FL, PHE, and FLU to fouled GAC show a similar tendency, which might imply that these compounds access the pore networks more easily 33.
Table 1. Freundlich and Langmuir isotherm pseudoequilibrium constants for sorption of polyaromatic hydrocarbons to granular activated carbon (GAC) 1240W in the presence of sediment
Log KF,GAC (µg/kgGAC)/(µg/L)n
Log CMAX (µg/kgGAC)
Our current results suggest sorption attenuation of HOCs to GAC, as observed in earlier studies 12, 22, 23, 43. Two plausible competitive mechanisms, which might be responsible for the fouling of GAC, are direct site competition and pore blocking by dissolved organic carbon (DOC) 22, 23. Especially at low GAC–sediment ratios (<0.1), excess DOC and/or particulate matter might result in saturation of active sites located on the external surface of the sorbent and blockage of pore pathways for target compounds to the internal structure. Consequently, DOC outcompetes PAHs in terms of mass 22. The majority of adsorption sites available for low-molecular-weight compounds are located in the interior of sorbent pathways, where bulky DOC molecules have limited access 23. Furthermore, deposits of complex organic molecules may hinder sorption of primarily heavy molecular PAHs, which are too large to diffuse into the sorbent pore networks during one month of contact time. Moreover, sorption to added GAC might be hindered by slow release of highly hydrophobic compounds from sediment particles. Therefore, it is more plausible that small molecules (FL, PHE, and ANT) even at low GAC–sediment mixing ratios, will have priority in penetrating the coated sorbent compared with larger compounds (Fig. 4). The results indicate that DOC molecules show little interference in the sorption of FL, PHE, and ANT to GAC. This is also consistent with the pore-filling mechanism, which favors small compounds 44, 45.
The reduction of HOC sorption by DOC depends on the type and concentration of the sorbate, the nature of DOC, and the effective porosity of the sorbent 21–23. Furthermore, direct competition for sorption sites between cosorbing target molecules may occur 46. As shown in Supplemental Data, Table S2, the average pore width of GAC 1240W might be sufficient for compounds with molecular diameters <1 nm to access pores with average diameters >1.7 times their molecular size 46. The effect of fouling on the sorption of PAHs larger than BaA (Fig. 4) might be explained by small pore widths being less accessible for bulky pollutants, which was also observed by Koelmans et al. 22. However, at high GAC–sediment ratios (i.e., >0.2), the excess of GAC reduces the effect of competitive sediment domains and results in effective sequestration of targeted pollutants.
From the porewater concentration data, the reduction of initial aqueous PAH concentrations in percentage can be plotted as a function of GAC dose to the sediment (Fig. 4). It appears that the lowest dose of 1% already reduces the porewater concentrations of the most relevant (available) PAHs such as FL, PHE, ANT by 30 to 80%. This range increases to 50 to 90% upon dosing with 4% GAC. The obtained results imply that, in the short term, as would be relevant for an ex situ remediation approach, the effectiveness of GAC decreases with increasing PAH molecular weight.
The present study provides data on binding of PAH to GAC in the presence of sediment, as would occur in a completely mixed ex situ remediation scenario. The experimental conditions tested account for any effects of organic matter fouling or sediment sorption competition on the binding of PAH. Consequently, these are the relevant values for evaluating GAC efficiency in GAC sorbent applications, especially in nonequilibrium sediment remediation scenarios. As expected, the binding of PAHs and the effectiveness of GAC to reduce sediment porewater concentrations depend on the GAC–sediment mixing ratio, that is, the dose of GAC. At a dose of 4% GAC, 50 to 90% of the most available PAHs would be bound to GAC. These percentages relate to a worst-case sediment with total PAH concentration as high as 1,600 ppm and native BC content about 40% of TOC. Consequently, GAC efficiency can be expected to be higher for sediments with more common levels of pollution.
Figures S1–S12. (1.5 MB DOC).
The present study was funded by the Dutch Technology Foundation STW, project 10030. We acknowledge financial support from Alterra, National Institute for Public Health and the Environment, Deltares, Boskalis Dolman, Norit, and De Vries and Van de Wiel. We thank Norit for supplying activated carbons and performing nitrogen analysis. We thank E. Reichman for practical assistance during extraction procedures and N. Sutton for total native petroleum hydrocarbons analysis.