Investigation of the Role of Structural Domains Identified in

Apr 28, 2005 - The role of composition and structure of sedimentary organic matter (SOM) in the sorption of hydrophobic organic compounds (HOCs) was ...
0 downloads 0 Views 396KB Size
Download PDF
Environ. Sci. Technol. 2005, 39, 3925-3932

Investigation of the Role of Structural Domains Identified in Sedimentary Organic Matter in the Sorption of Hydrophobic Organic Compounds C H R I S T O P H E R J . G O L D I N G , * ,† RONALD J. SMERNIK,‡ AND GAVIN F. BIRCH† School of Geosciences, The University of Sydney, New South Wales, Australia 2006, and Soil and Land Systems, School of Earth and Environmental Sciences, The University of Adelaide, Waite Campus, Urrbrae, South Australia, Australia 5064

The role of composition and structure of sedimentary organic matter (SOM) in the sorption of hydrophobic organic compounds (HOCs) was investigated by spiking 13Clabeled phenanthrene onto six estuarine sediments known to vary in SOM content and character. After equilibration and HF treatment, 13C NMR cross polarization and stable carbon isotope analyses indicated that the amount of desorption-resistant phenanthrene was related to aromatic carbon content. Application of the 13C NMR spectral editing technique proton spin relaxation editing (PSRE) demonstrated that all samples consisted of a rapidly relaxing and a slowly relaxing component, further evidence that SOM can be described as a structurally heterogeneous sorbent. Further, comparison of corresponding control and spiked PSRE subspectra revealed that, for each of the six sediments, desorption-resistant phenanthrene had become associated almost exclusively with the rapidly relaxing component. In only two of the sediments were there even small amounts of phenanthrene discernible in the slowly relaxing component, which is signficant as it was not always true that aromatic carbon was concentrated exclusively in the rapidly relaxing phase. The implication of these findings is that not all aromatic fractions have the same affinity for phenanthrene and that some fractions may indeed have little affinity at all. These results were interpreted as indicative that rapidly relaxing aromatic carbon associated with either sediment-associated charcoal or diagenetic organic matter plays a controlling role in the sorption of HOCs. However, the exact manner in which this rapidly relaxing aromatic phase relates to models presented elsewhere remains unclear.

Introduction The physical structure and chemical composition of sedimentary and soil organic matter (SOM) have been the subject of intense environmental scrutiny in recent decades. When hydrophobic organic compounds (HOCs), such as polycyclic * Corresponding author phone: +612-9351 4939; fax: +612 9351 0184; e-mail: [email protected]. † The University of Sydney. ‡ The University of Adelaide. 10.1021/es048171h CCC: $30.25 Published on Web 04/28/2005

 2005 American Chemical Society

aromatic hydrocarbons and organochlorine compounds, interact with sedimentary and soil systems, it is with the organic, rather than mineral, fraction of the sediment mass that they almost exclusively become associated (1, 2). The relative strength of this association has been demonstrated to significantly influence the bioavailability (3, 4) and environmental persistence (5) of the HOCs. Many early studies conceptualized SOM as a structurally homogeneous gel-like material into which HOCs can linearly diffuse according to partitioning theory (6, 7). However, recent studies have reported phenomena such as competitive sorption, concentration-dependent sorption, and sorptiondesorption hysteresis that are inconsistent with SOM as a capacity-unlimited, energetically and structurally homogeneous sorbent (8-11). Various alternative mechanisms have been proposed to account for anomalous sorption observations. Enhanced sorption of HOCs have been explained by the presence of high-surface-area carbonaceous residues, such as soot and charcoal (12-16) and by physical entrapment within micro- and nanopores (17-19). Other studies have described mechanisms driven by structural and compositional heterogeneity in SOM (20-24). Chen and coworkers (11, 20, 25), for example, have proposed that SOM is comprised of two organic phases: a reversible phase with which HOCs sorb freely and without restriction, and an irreversible phase within which a compound, upon sorption, is trapped by structural realignments or reconformations by the organic matrix. Elsewhere, Weber and co-workers (21, 23) have conceptualized SOM as possessing two broad structural phases. With the first, variously labeled as a loosely packed, amorphous, gel-like or “soft” carbon phase, HOCs interact via a partitioning, or absorption, mechanism. With the second, a condensed, rigid, tightly cross-linked or “hard” carbon phase, interaction is via adsorptive, van der Waals processes at a limited number of surface binding sites. Citing the sorption behavior of HOCs to geosorbents at various stages of diagenetic maturity, these researchers have suggested that the presence of a rigid phase is the product of increased structural condensation over the degradation process. Sorption nonlinearity is greatest in diagenetically mature kerogen-like materials and weakest in immature peats. Black carbon was speculated to be more condensed even than diagenetically matured kerogen, and it may thus exhibit completely adsorptive behavior and act as an “endmember” of the model (23). Numerical models that incorporate SOM heterogeneity (26) and the presence of such phases as black carbon (14, 15) have allowed considerable progress toward more accurately predicting SOM-HOC interactions; however, there remain significant limitations. For example, efforts to predict distribution coefficients through correlation with various geosorbent characteristics have been inconclusive. Some studies have found a relationship between sorption and SOM aromaticity (8, 27-30), while others have instead found good correlations with the aliphatic fraction (31). Attempting to resolve this issue, Simpson et al. (32) varied SOM aromaticity and aliphaticity through a set of chemical transformations and assessed the role of these concentrated residues in the sorption of phenanthrene. However, this study was unable to report correlation coefficients greater than 0.39. The implication of the latter work is that while increased SOM aromaticity is often a product of diagenetic maturity, and while there is substantial evidence suggesting that diagenetically mature geosorbents are more sorptive than immature geosorbents, often certain compositional elements may simply be co-occurring and sorption may be controlled VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3925

by other characteristics of mature sedimentary organic matter. Thus, it is unlikely that the sorptive character of a particular soil or sediment can be described by any single bulk measurement of that geosorbent’s properties. Recently, 13C NMR spectroscopy techniques have been incorporated into studies of HOC contamination, typically as a tool for distinguishing the compositional features that may be responsible for differences in sorptive behavior (8, 30-32), but also in more advanced applications (33-35). One advanced technique, proton spin relaxation editing (PSRE), has potential in contaminant studies. PSRE is a spectral editing technique that enables the identification of chemically distinct organic domains within structurally heterogeneous samples on the basis of differences in proton relaxation rates. PSRE has been applied in the characterization of a range of geosorbents including soil OM (36), sedimentary OM (37, 38), and kerogen (39). Importantly, PSRE can identify the structural character of some of the types of organic matter phases believed to be strongly sorbing. In particular, PSRE can identify aromatic-rich charcoal or black carbon domains (38, 40) and predominantly alkyl domains (36, 37, 39). In the current study, PSRE is used to determine directly whether different organic matter types identified by PSRE differ in their affinity for a common HOC compound, phenanthrene. This is possible because the process of spin diffusion, which is responsible for imparting a uniform T1H relaxation rate on all 1H nuclei within the PSRE domain, also imparts the same T1H relaxation rate on molecules sorbed within the domain. Since the minimum PSRE domain size of 30-100 nm is much larger than the size of a phenanthrene molecule (∼0.5-nm diameter), T1H for each phenanthrene molecule is well described by the characteristic domain T1H value. To facilitate the NMR detection of the sorbate molecules against the background of organic matter resonances, 13C-labeled phenanthrene is used. The six sediments investigated in the present study vary in the quantity and nature of sediment organic matter (38). Furthermore, previous PSRE studies have demonstrated that the organic matter in each sediment was structurally heterogeneous, although the nature of the domains identified varied between sediments (38).

Experimental Section Sediments. Six surface sediments were sampled from the terrestrial (upper estuary) and marine zones (lower estuary), respectively, of three southeastern Australian estuaries: Durras Lake, Lake Conjola, and Port Hacking. The nature of the estuaries, exact sample locations, and sampling procedure are described in detail elsewhere (38). The only exception was Hacking (lower), which was taken slightly closer to the estuary mouth (no more than 100 m, although 5-10 m deeper) to collect a sample that was slightly finer in grainsize. The results of contracted chemical analyses have shown that none of the sampling locations has been significantly impacted by agricultural, industrial, or urban pollution (data not shown). Sediments were analyzed for organic carbon (TOC) before and after HF pretreatment using a LECO elemental analyzer. Sample Loading and HF Pretreatment. Contaminated and control treatments were prepared for each sediment. A background aqueous solution containing 0.005 M CaCl2, 5 mg/L NaHCO3, and 0.01 mM HgCl2 (to control biological activity) was prepared using ultrapure water. One gram dry weight equivalent of each sediment was weighed into sealable glass vials, and 25 mL of background solution was added. 13C-labeled phenanthrene (C H , 99% doubly labeled in the 14 10 9 and 10 positions, Isotec Pty Ltd) was dissolved in methanol (HPLC grade, >99.9% purity) to produce a primary stock solution, which was delivered directly into the sediment slurry using a 100 µL glass Hamilton syringe. The concentration of 3926

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005

methanol in the sediment slurries was 0.2% v/v, a level that has been previously demonstrated to produce no cosolvency effects (41). A mass of 726 µg of 13C-labeled phenanthrene was added to each “contaminated” treatment. This loading was estimated as sufficient to ensure 13C NMR observability of the phenanthrene after HF-pretreatment losses. Controls were prepared in the same way, differing only in that no phenanthrene was added. Each vial was sealed and placed on a rotating tumbler. A period of 28 d was allowed for the sediment to approach equilibrium with the labeled phenanthrene. This period has been demonstrated as sufficient for this purpose in several previous studies (8, 13, 42, 43). To concentrate the organic fraction and to remove paramagnetic minerals that interfere with 13C NMR analysis, control and contaminated sediments were pretreated with hydrofluoric acid using the method of Skjemstad et al. (44). As HF dissolves glass, it was necessary to transfer the sediment samples into polyethylene centrifuge vials. To each sample, 50 mL of 2% HF was added and tumbled for 1 h. After tumbling, the samples were centrifuged at 2000 rpm for 20 min and the supernatants passed through a 5-µm sieve to collect the fine suspended fraction. The successive HF addition, tumbling, centrifugation, and sieving procedures were repeated four more times with a shaking time of 1 h, then three times with a shaking time of 16 h, and finally one time with a shaking period of 64 h. After the final extraction, sample residues were rinsed three times with ultrapure water, combined with the light fraction, and freeze-dried using a Labconko freeze-drier. Solid-State 13C NMR Analyses. All samples were analyzed using CP-MAS and PSRE techniques. Approximately 250 mg of HF-treated sediment was packed into a 7-mm diameter cylindrical zirconia rotor with Kel-F endcaps and analyzed using a Varian Unity 200 NMR spectrometer. CP spectra were acquired by a standard CP pulse sequence using the methods of Smernik and Oades (45). A 1-ms contact time and 1-s recycle delay were used for all samples; these values were chosen to avoid signal loss through saturation. CP spectra took about 3 h to acquire and are the sum of approximately 10000 scans. Inversion-recovery experiments, which were used to determine T1H relaxation rates, were performed using 12 recovery delays of between 0.1 ms and 1 s and a recycle delay of 1 s (40). Free induction decays were acquired with a sweep width of 40 kHz; 1216 data points were collected over an acquisition time of 15 ms. All CP and inversionrecovery spectra were zero-filled to 8192 data points and processed with a 50-Hz Lorentzian line broadening and a 0.010-s Gaussian broadening. Chemical shifts were externally referenced to the methyl resonance of hexamethylbenzene at 17.36 ppm. After linear baseline correction between 300 and -100 ppm, spectra were integrated between 300 and -10 ppm. These integrals were used in spin counting and in the inversion-recovery fitting procedures. The inversionrecovery experiments were analyzed by statistically comparing one- and two-T1H component fits to the data using the method of Smernik et al. (40). Two-component fit was found to be superior in each case, and PSRE subspectra representing rapidly (short T1H) and slowly relaxing (long T1H) structural domains were generated. For the purposes of CP quantitation, spin counting was performed using glycine as an external standard. For spin counting calculations, CP signal intensities were corrected for T1FH signal loss during the contact time. T1FH was determined using a variable spin lock experiment (1-ms contact time, 1-s recycle delay). Quantitation is measured as the amount of carbon that is NMR-observable (Cobs), which was calculated as described by Smernik and Oades (45). Stable Carbon Isotope Measurements. Final concentrations of the 13C-labeled compound in each of the HF-treated sediments were determined from the ratio of the two stable

TABLE 1. Carbon Content, Spin Counting, and δ13C Data for Control and Spiked Sediments Cobs Cobs control spiked Cobs (%) %TOC (%)a (from ref 38)a %TOC (from ref 38) (%)a

sample Durras (upper) Durras (lower) Conjola (upper) Conjola (lower) Hacking (upper) Hacking (lower)

2.8 2.3 3.5 4.3 8.8 4.8

3.0 2.4 4.7 5.2 10.8 12.4

69 54 55 60 44 55

64 60 57 57 47 55

60 43 63 60 47 56

δ13C (control)b -22.9(0.6) -21.1(0.1) -25.8(0.3) -21.7(0.1) -26.0(0.1) -22.7(0.1)

δ13C % of total C % of total 13C (spiked)b from spikec from spikec 58.2(52) 34.1(1.1) 80.4(2.4) 49.4(2.1) 47.7(4.6) 70.3(0.4)

0.64 0.44 0.84 0.56 0.59 0.74

8.2 5.7 10.5 7.2 7.5 9.3

% of 13C NMR % spike that signal from survived HF spiked treatment 8.0 (nd)e 9.1 6.7 9.4 9.4

27.0 15.1 44.1 36.1 75.7 52.2

a 13C NMR-observable carbon, calculated using the methods in ref 57. b Standard deviation in parentheses. c Calculated from δ13C (control) and δ13C (spiked) samples using eq 3. d Calculated from the area under the curves in Figure 1(n-r). e (nd) not determined.

isotopes of carbon. 13C/12C ratios are derived from δ13C values, the isotopic proportion of a carbon-containing substance relative to a standard, typically the belemnite fossil from the Pee Dee Fm (PDB). These values were obtained in triplicate for spiked and control sediment samples using a Europa Scientific (Cheshire, UK) Geo 20-20 stable isotope continuous flow mass spectrometer. 13 C/12C ratios can be related to δ13C values by the following equations:

R ) Rstandard[(δ13C/1000) + 1]

(1)

f ) 1/[1 + (1/R)]

(2)

where R is the 13C/12C ratio, and f is the fraction of total C that is 13C. R and f for the spiked and control samples were determined from these equations and the measured δ13C values, and these values were used to calculate the concentration of 13C-labeled phenanthrene in the spiked samples according to eq 3:

Rspiked ) [(1 - q)fSOM + qfPhen]/[(1 - q)(1 - fSOM) + q(1 - fPhen)] (3) where q and (1 - q) are the proportion of total carbon in the spiked samples contributed by the spiked compound and by SOM, respectively, and fSOM and fPhen are the fractions of 13C naturally occurring in the SOM samples and in the pure 13Clabeled compound, respectively. Equation 3 can be rearranged to determine q, which is the concentration of 13Clabeled compound in the sediment sample.

Results and Discussion Cross Polarization and δ13C Data for Control and Spiked Sediments. Carbon content, spin counting, and δ13C data for control and spiked sediments are presented in Table 1. The carbon contents for the six sediments were all slightly lower than the corresponding samples in Golding et al (38), apart from in Hacking (lower), which was substantially lower at 4.8% TOC as compared with 12.4%. This decrease in TOC probably reflects the difference in sampling location and water depth. This difference in organic carbon content did not appear to result in a difference in organic matter character, as all sediments demonstrated similar carbon observabilities (Cobs) to the values reported previously. Cobs values in the range of 47-64% are not uncommon for samples analyzed by cross polarization, as this method is typically less quantititative than the direct polarization or Bloch Decay (BD) technique. CP spectra for the six control and spiked sediments, as well as “difference” spectra, are presented in Figure 1. All control samples (Figure 1a-f) compare well with the samples characterized previously (38). For each pair of samples from the three estuaries, the upper-estuarine samples were characteristically richer in methoxyl (56 ppm), O-alkyl (115 ppm), and aromatic (145 ppm) functional groups than the

FIGURE 1. Control (a-f) and spiked (g-l) CP-MAS spectra for the six sediments, unsorbed phenanthrene (m) and difference spectra for all sites apart from Durras (lower) (n-r). The band 110-140 ppm (shaded) corresponds to the spectral region where phenanthrene is observed. lower-estuarine samples, which were richer in unsubstituted alkyl carbon (33 ppm). In Golding et al. (38), CP and BD spectra were also compared to reveal the presence of substantial amounts of aromatic C species with low CP observability in the Port Hacking samples. This indicated that the Port Hacking samples contain substantial quantities of charcoal, probably a result of frequent bushfires in the adjoining national park in recent years. Given that the Cobs values of samples Hacking (upper) and Hacking (lower) are similar to those reported earlier, it is reasonable to assume without new BD data that the new samples are also rich in charcoal. Figure 1(g-l) depicts CP spectra for the sediments spiked with phenanthrene, which characteristically differ VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3927

from the control samples in that aromatic resonance at 125130 ppm is more intense. Figure 1(n-r) shows the “difference” spectra for five of the six samples, produced by subtracting the corresponding control spectra from the spiked spectra. For all samples apart from Durras (lower), subtraction of these spectra canceled out the 13C contribution of the SOM, leaving an easily discernible peak at approximately 127 ppm. Comparison of the difference spectra with the spectrum produced for neat, unsorbed phenanthrene (Figure 1m) confirms that the peak is attributable to the phenanthrene spike. In Figure 1(n,o,q,r), faint sidebands associated with the strong phenanthrene signal can be identified, although most of the unevenness in the baseline of the difference spectra can be attributed to imperfect cancellation of SOM peaks due to small compositional differences between the control and spiked samples. In the case of sample Durras (lower), the relatively low retention of phenanthrene through HF-pretreatment meant that the peak associated with the spike could not be discerned reliably from this uneven baseline. The integrated areas of the phenanthrene resonances (central band plus sidebands) in the difference spectra represent the contribution of the spike to the NMR signal, and these data are presented in Table 1. Phenanthrene loadings in the HF-treated residues, calculated from δ13C values, were 0.44-0.84% on a per carbon basis (mass of phenanthrene carbon/mass of sediment carbon). To ensure that the phenanthrene spike would contribute roughly 10% of the 13C detected in the spiked samples, an initial loading of 726 µg/g dry weight sediment was selected on the basis of a set of assumptions including the average natural abundance of 13C and the typical postHF residue mass and TOC content. The assumptions were validated by the δ13C measurements of the spiked samples, which were between 7.2 and 10.5% for all samples except sample Durras (lower), which was slightly lower at 5.7%. These values are in good agreement with the 13C contribution of phenanthrene, also shown in Table 1, as deduced from the difference spectra. The δ13C values for the samples imply that the composition of these estuarine SOM is a reflection of the relative contribution of marine and terrestrial source materials. For the pairs of samples taken from each estuary, the upperestuary samples had consistently more negative δ13C values. These are consistent with values of approximately -21 and -27% that can be expected for marine- and terrestrial-derived organic matter, respectively (46), and are similar to values obtained for Dutch estuarine sediments by Megens et al. (47). The δ13C value for the spiked samples are well outside natural values due to the contribution of 13C in the added phenanthrene. The standard deviations in replicate analyses of δ13C for the spiked samples are higher than for the control samples. This is possibly a consequence of heterogeneous distribution of phenanthrene at the particle-scale. Table 1 describes the percentage of the phenanthrene originally added that survived HF-treatment, as determined from the known initial loadings and the final δ13C measurements. HF-treatment involves 10 successive rinses with acid, followed by three rinses with water. Each of these rinses provides an opportunity for phenanthrene desorption, although the hydrophobicity of the sorbate makes it unlikely that equilibrium between solid and solution phases would be reached during the shorter duration (1 h) HF rinses. For the six sediments, phenanthrene retention was between 15.1 and 75.7%. Retention was greatest in the Port Hacking sediments (52.2-75.7%), moderate in the Lake Conjola sediments (36.1-44.1%), and weakest in the Durras Lake sediments (15.1-27.0%). The amount of phenanthrene retained by the sediments to some extent reflects the TOC content of the six sediments. This is unsurprising, since the 3928

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005

TABLE 2. Results of One- and Two-Component Fits to Inversion-Recovery Data for Control and Spiked Sediments sample Durras (upper)

control spiked Durras (lower) control spiked Conjola (upper) control spiked Conjola (lower) control spiked Hacking (upper) control spiked Hacking (lower) control spiked

T1Hava T1Hfastb T1Hslowc (ms) (ms) (ms) 18.5 18.2 20.0 16.4 13.1 12.6 11.2 11.0 29.1 26.5 14.3 13.8

11.7 12.3 12.6 10.9 6.9 7.2 6.5 7.0 11.4 10.9 9.6 8.7

64 70 72 52 50 44 33 32 113 117 70 62

ratio fast/slowd 66:34 70:30 66:34 68:32 61:39 64:36 63:37 67:33 57:43 59:41 71:29 69:31

a T H value from one-component fit b T H value of rapidly relaxing 1 1 component in two-component fit c T1H value of slowly relaxing component in two-component fit d Ratio of NMR signal intensity attributed to the rapidly relaxing component to that attributed to the slowly relaxing component in two-component fit

ratio of solid sorbent phase to solution differed according to the TOC contentsthe mass of sediment was the same for each treatment, but this corresponds to different quantities of organic matter. Therefore, in treatments with lower TOC sediments, the ratio of solid sorbent to solution was lower, and more phenanthrene would be expected to partition into the solution phase during each HF or water rinsing, resulting in greater losses of phenanthrene, even if the sorption properties of the sediment organic matter did not vary between sediments. However, there is also evidence in these data that the upper-estuary sediments consistently provided greater phenanthrene retention. In particular, the trend in phenanthrene retention in the Durras Lake and Port Hacking sediments reflected an increase in TOC content with distance from the estuary mouth, while in the Lake Conjola sediments the same trend was evident despite a decrease in TOC content. This finding, from a limited number of samples, is significant as it encourages speculation that aspects of SOM character which have trends across an estuary may have an effect on sorption processes. While these results are encouraging, it is appropriate to acknowledge two possible sources of artifact. First, the HF pretreatment has been noted previously to potentially result in SOM losses of between 8 and 17% (44). These authors, however, found that HF pretreatment did not result in the loss of particular organic groups from the treated samples, and therefore pretreatment had little effect on overall carbon chemistry. Therefore in the context of the samples described here, this possibility is likely to have affected, at worst, the quantitative conclusions being drawn, but not the qualitative. Second, Gelinas et al. (48) noted the potential for loss of submicrometer-size black carbon particles during liquid pretreatments through sorption to glass or the water-air interface. With respect to the samples presented here, however, losses through such processes may be regarded as minor, since submicrometer BC particles are generally of the graphite/soot variety which can be expected to constitute a small fraction of total black carbon in bushfire-affected (charred biomass-enriched) residues. Gelinas et al. (48) found that losses following pretreatment could be around 12%; however, in this case the material being investigated was a sediment artificially amended with pure soot rather than the typically larger particles of charred material. Therefore, this number probably exaggerates the potential for losses from more BC-heterogeneous samples such as ours. Proton Spin Relaxation Editing. Results of one- and twoT1H component fits to the inversion-recovery data are presented in Table 2. Each of the sediments was better

described by the two-component fit, and therefore it was possible to generate PSRE subspectra of rapidly and slowly relaxing components. Average (one-component fit) T1H values were similar between corresponding pairs of control and spiked sediments (Table 2). Small variations may be due to slight differences in water content, to which T1H values are very sensitive (R. Smernik, unpublished results). The T1H values of the rapidly and slowly relaxing components (T1Hfast and T1Hslow, respectively) are also similar between corresponding pairs of control and spiked sediments, although the ratio of rapidly to slowly relaxing component (fast/slow ratio) is generally slightly higher for the spiked sediments (Table 2). This may reflect the greater concentration of 13Cphenanthrene in the rapidly relaxing components; however the differences are close to the level of precision in the method. Consistently, the rapidly relaxing component accounted for a greater proportion of total NMR signal intensity in the control samples, with fast/slow ratios ranging from as high as 71:29 to as low as 57:43. These ratios are similar to the relative contributions reported in other PSRE studies (36, 37). PSRE subspectra for the six control and spiked samples are presented in Figures 2-4. For the control samples, the rapidly relaxing subspectra for the Durras Lake and Lake Conjola samples depicted a consistently greater aromatic and carboxyl carbon signal than the slowly relaxing subspectra. For the same samples, unsubstituted alkyl carbon was found preferentially in the slowly relaxing subspectra. The two Port Hacking samples were distinct from the other samples in that both the slowly and rapidly relaxing compartments were characterized by tall, broad peaks in the aromatic region. Indeed, in Hacking (upper) there was a higher aromatic peak in the slowly relaxing than in the rapidly relaxing subspectrum. As was the case for the CP spectra, the PSRE subspectra for the control and spiked samples were almost identical, except for the presence of an extra aromatic signal in the PSRE spectra of the spiked samples, which can be attributed to the contribution of phenanthrene (Figures 2-4). Phenanthrene molecules sorbed to SOM “inherit” the “T1H signature” of the surrounding organic matter. This is because spin diffusion among the 1H population ensures uniform T1H values on the 30-100-nm scale. This scale is large compared with the size of the phenanthrene molecule (∼0.5 nm). In the pure, crystalline form, T1H for the 13C-labeled phenanthrene was found to be 440 s, over 4 orders of magnitude longer than the average T1H values for the six sediments (Table 2). The relative affinity for the sorbate molecules of the two components represented by the PSRE subspectra is therefore reflected in the relative sizes of the phenanthrenederived signal in the respective PSRE subspectra. For each of the six sediments, most of this extra signal appeared in the rapidly relaxing subspectra, indicating that most of the 13Clabeled phenanthrene was located in the domains characterized by rapid T1H relaxation. In fact, for four of the sediments (samples Durras (upper), Durras (lower), Conjola (lower), and Hacking (upper)), there was no discernible increase in the aromatic signal in the slowly relaxing subspectrum of the spiked sample compared with the corresponding control. In only two of the six sediments, samples Conjola (upper) and Hacking (lower), was any phenanthrene apparent in the slowly relaxing compartment, and in both of these cases it was much less than was apparent in the corresponding rapidly relaxing component. Therefore, for each of the sediments, most of the strongly held phenanthrene (that is, resistant to multiple washings involved in HF treatment) is sorbed to the rapidly relaxing component. Since the rapidly relaxing components are generally richer in aromatic C and poorer in alkyl C, these

FIGURE 2. PSRE subspectra for the control and spike samples for Durras Lake (upper) (a-d) and (lower) (e-h). The band 110-140 ppm (shaded) corresponds to the spectral region where phenanthrene is observed. findings support the view that there is a relationship between sorption and SOM aromaticity (8, 27-30), but run counter to recent reports that alkyl carbon provides an important sorption phase (31, 49-51). However, it is also clear that not all aromatic C sorbs phenanthrene strongly, since the slowly relaxing component of Hacking (upper) was rich in aromatic C, but appeared to sorb little phenanthrene. Implications for Models of SOM Structure and Sorption Processes. The results contribute further toward understanding the role of particular SOM compositional elements in sequestering HOCs, and they may assist in explaining some of the inconsistencies identified elsewhere in the literature. Research aimed at establishing the role of aromatic carbon can be cited to illustrate the point (8, 27-32). Aromatic carbon has been used with varying success by many authors in attempts to establish whether sorption is controlled by this compositional element; in some cases, strong correlations between total aromatic carbon and sorption have been established, while in other cases these correlations have been poor. The findings presented in the current work show that for all six sediments analyzed, there were a spectrum of carbon types distributed throughout two different structural domains. Aromatic carbon was present preferentially in the rapidly relaxing compartment of four of the sediments, and it was with this compartment, also, that spiked, equilibrated phenanthrene became associated. However, phenanthrene spiked into the two Port Hacking samples also became associated with the rapidly relaxing compartment, despite the total pool of aromatic carbon existing in substantial quantities in both compartments. If it is true that aromatic carbon is the feature with which spiked phenanthrene becomes associated, then it is not total VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3929

FIGURE 3. PSRE subspectra for the control and spike samples for Lake Conjola (upper) (a-d) and (lower) (e-h). The band 110-140 ppm (shaded) corresponds to the spectral region where phenanthrene is observed.

FIGURE 4. PSRE subspectra for the control and spike samples for sites Hacking (upper) (a-d) and Hacking (lower) (e-h). The band 110-140 ppm (shaded) corresponds to the spectral region where phenanthrene is observed.

aromatic carbon, as there is a substantial amount of this carbon type, that is, the slowly relaxing compartment, that appears to play only a minor role.

because they were blocked by other organic material or were already attenuated by other solutes such as indigenous PAHs.

There are several explanations for the sorption behavior observed in the present study. Comparison of CP and BD spectra of Port Hacking sediments have identified the presence of substantial amounts of bushfire charcoal (38). It is not possible, using NMR techniques, to determine quantitatively how much of this charcoal can be ascribed to the rapidly relaxing compartment. Several studies have indicated the importance of charcoal and other black carbon in the binding of HOCs in general and PAHs, such as phenanthrene, in particular (12, 13, 52). Black carbon, in its various forms, is a hydrophobic sorbent composed of parallel aromatic sheets which are separated at distances of 30-40 nm and cross-linked by short aliphatic linkages. It is speculated that HOCs become sorbed tightly between the aromatic sheets, such that sorption is greater for planar compounds, such as PAHs, than for less planar or nonplanar compounds, such as di-ortho-substituted PCBs (13, 53). The high microporosity of black carbon allows for a proportionately higher surface area than humic materials, such that for those compounds that are able to penetrate between the aromatic sheets, association with the matrix is enhanced through strong adsorptive π-π molecular attractions (13). It is not clear why slowly relaxing aromatic groups have little affinity for phenanthrene; however, two broad explanations are possible. First, as discussed earlier, it is possible that phenanthrene became only weakly associated with this structural component and was easily desorbed during HF treatment. Second, it is possible that the phenanthrene mass never had complete access to these aromatic groups either

However, the presence of charcoal, and its concentration within the rapidly relaxing compartment, does not explain the sorptive behavior of the sediments from Durras Lake and Lake Conjola. The same preferential sorption of phenanthrene was observed in these samples, despite none of them having been characterized earlier as particularly rich in charcoal (38). A number of studies, however, have provided convincing evidence for the existence of structural heterogeneity in humic materials. Weber et al. (26) characterized a diverse set of SOM samples on the basis of “hard” carbon (quantified by high-temperature oxidation with pure oxygen) and “soft” carbon (quantified by low-temperature persulfate oxidation). Later, LeBoeuf and Weber (54) claimed to have discovered the glass transition temperature for a humic acid. Glass transition temperatures mark the thermodynamic transition in chemical polymers from a rigid, tightly crosslinked to a loosely packed structural conformation, and they were cited in the latter study as being evidence that SOM can be similarly characterized along such structural distinctions. Other studies, such as Schnitzer et al. (55) who used X-ray diffraction to describe two peaks representing, on one hand, tightly packed aromatic sheets (0.35 nm) and, on the other, less tighly packed aliphatic sheets (0.41-0.47 nm), have presented further evidence of SOM structural heterogeneity.

3930

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005

The findings of the current study broadly support the argument that such structural heterogeneity in SOM plays a key role in the sorption of HOCs. However, without further data it is difficult to relate our findings with other established models, as there are aspects of these data that both support

and contradict those models. For example, Weber and coworkers present a model wherein two structurally distinct SOM phases exist: a condensed, rigid, and tightly crosslinked phase (responsible for enhanced, nonlinear sorption) and an uncondensed, amorphous and loosely packed phase (responsible for linear, partitioning sorption). It is possible that this distinction could be broadly analogous to the rapidly and slowly relaxing components, respectively, identified in the present work. The preferential sorption of phenanthrene and the concentration of aromatic carbon in the rapidly relaxing component is consistent with this model, as the presence and abundance of the structurally condensed domain is a function of diagenetic maturation, and this phase is consequently often enriched with certain degradationresistant elements such as aromatic carbon (56). Not all aromatic organic matter is necessarily structurally condensed, however, which is consistent with our identification of slowly relaxing aromatic carbon that appears to play no role in the sorption process. Condensation occurs to SOM as a consequence of time and of exposure to appropriate environmental conditions. While it is true that diagenesis can result in the concentration of aromatic functional groups in SOM due to the successive removal of less resistant components, the same groups can simply be present in a sample due to the provenance of the original biomolecules from which the SOM developed. This notwithstanding, there are also aspects of our data that contradict the suitability of drawing a neat analogy between the rapidly/slowly relaxing components and the glassy/rubbery phases. If a condensed phase that is concentrated by the diagenetic process is the domain responsible for enhanced, nonlinear HOC sorption phenomena, then it could be expected that it is with this phase that HOC molecules are physically associated. However, the findings in our research describe that phenanthrene molecules were almost exclusively associated with structural domains rich in the carbon types that are generally considered to be the most easily degraded (for example, carbohydrate and amino acid C) during diagenetic breakdown. Also, the slowly relaxing components (with which there was little phenanthrene associated) were characteristically rich in nonhydrolyzable polymethylene groups that, ordinarily, would not be expected to occur in uncondensed, loosely packed structural phases. It is apparent that more research is needed to clarify how the findings from PSRE experiments fit with other established models. Our results provide further support for the existence of structural heterogeneity in sedimentary organic matter. Strong evidence is presented that certain structural compartments have a particularly strong affinity for hydrophobic organic compounds, such as phenanthrene, lending broad support to established models in the literature. This work, however, goes further in demonstrating that particular compositional elements, such as aromatic carbon, within the rapidly relaxing component may play a controlling role in HOC sorption processes. However, without further research, the exact manner in which our findings relate to the models presented elsewhere remains a matter for speculation. Thus, it would be instructive to examine further the nature of the structural phases that PSRE describes and to explore whether these phases are analogous to those presented in the established models, such as those of Weber and coworkers (21, 23), Chen and co-workers (11, 20, 25), and Gschwend and co-workers (14-16). In particular, given that Weber and co-workers base many of their conclusions on similarities between SOM and artificial polymers, it would be useful to apply PSRE to appropriate synthetic substances. With respect to the irreversible sorption model of Chen and co-workers, the structure of the SOM matrix both before and after sorption could be well characterized using 13C-labeled

compounds and the PSRE technique, as has been done by Guthrie et al. (33) using standard 13C NMR techniques. Currently, a substantial shortcoming of techniques such as PSRE is that the data produced is largely qualitative and therefore, unlike, for example, the soot-inclusive model of Gschwend and co-workers (14-16), is not predictive. A significant challenge for further studies with PSRE is the development of methods to allow the rapidly and slowly relaxing domains to be expressed as quantifiable parameters. The incorporation of PSRE techniques into studies characterizing geosorbents could provide a valuable contribution to the effort to understand HOC-SOM interactions and how these interactions affect processes such as contaminant bioavailability and persistence.

Acknowledgments This research was funded by Australian Research Council Discovery Grant DP0346487. We are grateful to Jacqueline Halpin for her assistance during field sampling and to Tom Savage and David Mitchell at the University of Sydney for technical advice.

Literature Cited (1) Huang, W.; Schlautman, M. A.; Weber, W. J. J. A distributed reactivity model for sorption by soils and sediments. 5. The influence of near-surface characteristics in mineral domains. Environ. Sci. Technol. 1996, 30, 2993-3000. (2) Yang, K.; Zhu, L. Z.; Lou, B. F.; Chen, B. L. Significance of natural organic matter in nonlinear sorption of 2,4-dichlorophenol onto soils/ sediments. Water Resour. Res. 2004, 40, W01603. (3) Alexander, M. Aging, bioavailability and overestimation of risk from environmental pollutants. Environ. Sci. Technol. 2000, 34, 4259-4265. (4) Kraaij, R.; Seinen, W.; Tolls, J.; Cornelissen, G.; Belfroid, A. Direct evidence of sequestration in sediments affecting the bioavailability of hydrophobic organic chemicals to benthic deposit feeders. Environ. Sci. Technol. 2002, 36, 3525-3529. (5) Lueking, A. D.; Huang, W.; Soderstrom-Schwarz, S.; Kim, M.; Weber, W. J. J. Relationship of soil organic matter characteristics to organic contaminant sequestration and bioavailability. J. Environ. Qual. 2000, 29, 317-323. (6) Chiou, C. T.; Peters, L. J.; Freed, V. H. A physical concept of soil-water equilibria for nonionic organic compounds. Science 1979, 206, 831-832. (7) Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Sorption of hydrophobic pollutants on natural sediments. Water Res. 1979, 13, 241-248. (8) Huang, W.; Weber, W. J. J. A distributed reactivity model for sorption by soils and sediments. 10. Relationships between desorption, hysteresis, and the chemical characteristics of organic domains. Environ. Sci. Technol. 1997, 31, 2562-2569. (9) Huang, W.; Weber, W. J. J. A distributed reactivity model for sorption by soils and sediments. 11. Slow concentrationdependent sorption rates. Environ. Sci. Technol. 1998, 32, 35493555. (10) Xing, B.; Pignatello, J. J. Competitive sorption between 1,3dichlorobenzene or 2,4-dichlorophenol and natural aromatic acids in soil organic matter. Environ. Sci. Technol. 1998, 32, 614-619. (11) Kan, A. T.; Chen, W.; Tomson, M. B. Desorption kinetics of neutral hydrophobic organic compounds from field-contaminated sediment. Environ. Pollut. 2000, 108, 81-89. (12) Bucheli, T. D.; Gustafsson, O. Quantification of the soot-water distribution coefficient of PAHs provides mechanistic basis for enhanced sorption observations. Environ. Sci. Technol. 2000, 34, 5144-5151. (13) Jonker, M. T. O.; Koelmans, A. A. Sorption of polycyclic aromatic hydrocarbons and polychlorinated biphenyl to soot and sootlike materials in the aqueous environment; mechanistic considerations. Environ. Sci. Technol. 2002, 36, 3725-3734. (14) Accardi-Dey, A.; Gschwend, P. M. Assessing the combined role of partitioning into organic carbon and adsorption onto black carbon. Environ. Sci. Technol. 2002, 36, 21-29. (15) Accardi-Dey, A.; Gschwend, P. M. Reinterpreting literature sorption data considering both partitioning into organic carbon and adsorption onto black carbon. Environ. Sci. Technol. 2003, 37, 99-106. VOL. 39, NO. 11, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3931

(16) Gustafsson, O.; Haghseta, F.; Chan, C.; Macfarlane, J.; Gschwend, P. M. Quantification of the dilute sedimentary soot phase: implications for PAH speciation and bioavailability. Environ. Sci. Technol. 1997, 31, 203-209. (17) Pignatello, J. J.; Xing, B. Mechanisms of slow sorption of organic compounds to natural particles. Environ. Sci. Technol. 1996, 30, 1-11. (18) Ball, W. P.; Roberts, P. V. Long-term sorption of halogenated organic chemicals by aquifer material. 2. Intraparticle diffusion. Environ. Sci. Technol. 1991, 25, 1237-1249. (19) Wu, S.-C.; Gschwend, P. M. Sorption kinetics of hydrophobic organic compounds to natural sediments and soils. Environ. Sci. Technol. 1986, 20, 717-725. (20) Chen, W.; Kan, A. T.; Fu, G.; Tomson, M. B. Factors affecting the release of hydrophobic organic contaminants from natural sediments. Environ. Toxicol. Chem. 2000, 19, 2401-2408. (21) Weber, W. J. J.; Leboeuf, E. J.; Young, T. M.; Huang, W. Contaminant interactions with geosorbent organic matter: insights drawn from polymer sciences. Water Res. 2001, 35, 853-868. (22) Luthy, R. G.; Aiken, G. R.; Brusseau, M. L.; Cunningham, S. D.; Gschwend, P. M.; Pignatello, J. J.; Reinhard, M.; Traina, S. J.; Weber, W. J. J.; Westall, J. C. Sequestration of hydrophobic organic contaminants by geosorbents. Environ. Sci. Technol. 1997, 31, 3341-3347. (23) Huang, W.; Ping’an, P.; Yu, Z.; Fu, J. Effects of organic matter heterogeneity on sorption and desorption of organic contaminants by soils and sediments. Appl. Geochem. 2003, 18, 955972. (24) van Noort, P. C. M.; Cornelissen, G.; ten Hulscher, D. T. E. M.; Vrind, B. A.; Rigterink, H.; Belfroid, A. Slow and very slow desorption of organic compounds from sediment: influence of sorbate planarity. Water Res. 2003, 37, 2317-2322. (25) Chen, W.; Kan, A. T.; Fu, G.; Vignona, L. C.; Tomson, M. B. Adsorption-desorption behaviours of hydrophobic organic compounds in sediments of Lake Charles, Louisiana, USA. Environ. Toxicol. Chem. 1999, 18, 1610-1616. (26) Weber, W. J. J.; McGinley, P. M.; Katz, L. E. A distributed reactivity model for sorption by soils and sediments. 1. Conceptual basis and equilibrium assessments. Environ. Sci. Technol. 1992, 26, 1955-1962. (27) Xing, B.; Chen, Z. Spectroscopic evidence for condensed domains in soil organic matter. Soil Sci. 1999, 164, 40-47. (28) Salloum, M. J.; Dudas, M. J.; McGill, W. B. Variation of 1-naphthol sorption with organic matter fractionation: the role of physical conformation. Org. Geochem. 2001, 32, 709-719. (29) Ahmad, R.; Kookana, R. S.; Alston, A. M.; Skjemstad, J. O. The nature of soil organic matter affects sorption of pesticides. 1. Relationships with carbon chemistry as detemined by 13C CP MAS NMR spectroscopy. Environ. Sci. Technol. 2001, 35, 878884. (30) Perminova, I. V.; Grechishcheva, N. Y.; Kovalevskii, D. V.; Kudryavtsev, A. V.; Petrosyan, V. S.; Matorin, D. N. Quantification and prediction of the detoxifying properties of humic substances related to their chemical binding to polycyclic aromatic hydrocarbons. Environ. Sci. Technol. 2001, 35, 3841-3848. (31) Salloum, M. J.; Chefetz, B.; Hatcher, P. G. Phenanthrene sorption by aliphatic-rich natural organic matter. Environ. Sci. Technol. 2002, 36, 1953-1958. (32) Simpson, M. J.; Chefetz, B.; Hatcher, P. G. Phenanthrene sorption to structurally modified humic acids. J. Environ. Qual. 2003, 32, 1750-1758. (33) Guthrie, E. A.; Bortiatynski, J. M.; Van Heemst, J. D. H.; Richman, J. E.; Hardy, K. S.; Kovach, E. M.; Hatcher, P. G. Determination of [13C]Pyrene sequestration in sediment microcosms using flash-pyrolysis-GC-MS and 13C NMR. Environ. Sci. Technol. 1999, 33, 119-125. (34) Guthrie-Nichols, E.; Grasham, A.; Kazunga, C.; Sangaiah, R.; Gold, A.; Bortiatynski, J. M.; Salloum, M. J.; Hatcher, P. G. The effect of aging on pyrene transformation in sediments. Environ. Sci. Technol. 2003, 22, 40-49. (35) Deshmukh, A. P.; Chefetz, B.; Hatcher, P. G. Characterization of organic matter in pristine and contaminated coastal marine sediments using solid-state 13C NMR, pyrolytic and thermocheolytic methods: a case study in the San Diego harbour area. Chemosphere 2001, 45, 1007-1022. (36) Smernik, R. J.; Oades, J. M. Effects of added paramagnetic ions on the 13C CP/MAS NMR spectrum of a de-ashed soil. Geoderma 1999, 89, 219-248. (37) Gelinas, Y.; Baldock, J. A.; Hedges, J. I. Organic carbon composition of marine sediments: effect of oxygen exposure on oil generation potential. Science 2001, 294, 145-148. 3932

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 11, 2005

(38) Golding, C. J.; Smernik, R. J.; Birch, G. F. Characterisation of sedimentary organic matter from three southeastern Australian estuaries using 13C NMR spectroscopy. Mar. Freshwater Res. 2004, 55, 285-293. (39) Petsch, S. T.; Smernik, R. J.; Eglington, G.; Oades, J. M. A solidstate 13C NMR study of kerogen degradation during black shale weathering. Geochim. Cosmochim. Acta 2001, 65, 1867-1882. (40) Smernik, R. J.; Skjemstad, J. O.; Oades, J. M. Virtual fractionation of charcoal from soil organic matter using solid state 13C NMR spectral editing. Aust. J. Soil Res. 2000, 38, 665-683. (41) Wauchope, R. D.; Savage, K. E.; Koskinen, W. C. Adsorptiondesorption equilibria of herbicides in soil - naphthalene as a model-compound for entropy-enthalpy effects. Weed Sci. 1983, 31, 744-751. (42) Cornelissen, G.; Hassel, K. A.; van Noort, P. C. M.; Kraaij, R.; van Ekeren, P. J.; Dijkema, C.; de Jager, P. A.; Govers, H. A. J. Slow desorption of PCBs and chlorobenzenes from soils and sediments: relations with sorbent and sorbate characteristics. Environ. Pollut. 2000, 108, 69-80. (43) Millward, R. N.; Fleeger, J. W.; Reible, D. D.; Keteler, K. A.; Cunningham, B. D.; Zhang, L. Pyrene bioaccumulation, effects of pyrene exposure on particle-size selection, and fecal pyrene content in the oligochaete Limnostrilus hoffmeisteri (tubificidae oligochaeta). Environ. Toxicol. Chem. 2001, 20, 1359-1366. (44) Skjemstad, J. O.; Clarke, P.; Taylor, J. A.; Oades, J. M.; Newman, R. H. The removal of magnetic materials from surface soils. A solid state 13C CP/MAS NMR study. Aust. J. Soil Res. 1994, 32, 1215-1229. (45) Smernik, R. J.; Oades, J. M. The use of spin counting for determining quantitation in solid state 13C NMR spectra of natural organic matter: 1. Model systems and the effects of paramagnetic impurities. Geoderma 2000, 96, 101-129. (46) Mook, W. G.; Tan, F. C. Stable carbon isotopes in rivers and estuaries. In Biogeochemistry of Major World Rivers; Degens, E. T., Kempe, S., Richie, J. E., Eds.; John Wiley & Sons: New York, 1991; pp 245-264. (47) Megens, L.; van der Plicht, J.; de Leeuw, J. W. Molecular, radioactive and stable carbon isotope characterization of estuarine particulate organic matter. Radiocarbon 1998, 40, 985990. (48) Gelinas, Y.; Prentice, K.; Baldock, J.; Hedges, J. An improved thermal oxidation method for the quantification of soot/ graphitic black carbon in sediments and soils. Environ. Sci. Technol. 2001, 35, 3519-3525. (49) Mao, J.-D.; Hundel, S.; Tomson, M. L.; Schmidt-Rohr, K. Correlation of poly(methylene)-rich amorphous aliphatic domains in humic substances with sorption of a nonpolar organic contaminant, phenanthrene. Environ. Sci. Technol. 2002, 36, 929-936. (50) Chefetz, B. Sorption of phenanthrene and atrazine by plant cuticular fractions. Environ. Toxicol. Chem. 2003, 22, 24922498. (51) Khalaf, M.; Kohl, S. D.; Klumpp, E.; Rice, J. A.; Tomba´cz, E. Comparison of sorption domains in molecular weight fractions of a soil humic acid using solid-state 19F NMR. Environ. Sci. Technol. 2003, 37. (52) Chiou, C. T.; Kile, D. E.; Rutherford, D. W.; Sheng, G.; Boyd, S. A. Sorption of organic compounds from water to a peat soil and its humic acid and humin fractions: potential sources of the sorption nonlinearity. Environ. Sci. Technol. 2000, 34, 12541258. (53) Bucheli, T. D.; Gustafsson, O. Soot sorption of non-ortho and ortho substituted PCBs. Chemosphere 2003, 53, 515-522. (54) Leboeuf, E. J.; Weber, W. J. J. A distributed reactivity model for sorption by soils and sediments. 8. Sorbent organic domains: discovery of a humic acid glass transition and an argument for a polymer-based model. Environ. Sci. Technol. 1997, 31, 16971702. (55) Schnitzer, M.; Kodama, H.; Ripmeester, J. A. Determination of the aromaticity of humic substances by X-ray diffraction analysis. Soil Sci. Soc. Am. J. 1991, 55, 745-750. (56) Young, T. M.; Weber, W. J. J. A distributed reactivity model for sorption by soils and sediments. 3. Effects of diagenetic processes on sorption energetics. Environ. Sci. Technol. 1995, 29, 92-97. (57) Gelinas, Y.; Baldock, J. A.; Hedges, J. I. Demineralisation of marine and freshwater sediments for CP/MAS 13C NMR analysis. Org. Geochem. 2001, 32, 677-693.

Received for review November 21, 2004. Revised manuscript received March 10, 2005. Accepted March 17, 2005. ES048171H