JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 7 June 2005
Published in J Environ Qual 34:1194-1204 (2005)
DOI: 10.2134/jeq2004.0371
© 2005 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Smernik, R. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Smernik, R. J.
Agricola
Right arrow Articles by Smernik, R. J.
Related Collections
Right arrow Organic Compounds
Right arrow Soil Pollution
Right arrow Isotopes
Right arrow Soil Organic Matter

TECHNICAL REPORTS

Organic Compounds in the Environment

A New Way to Use Solid-State Carbon-13 Nuclear Magnetic Resonance Spectroscopy to Study the Sorption of Organic Compounds to Soil Organic Matter

Ronald J. Smernik*

School of Earth and Environmental Sciences, The University of Adelaide, Waite Campus, Urrbrae, South Australia 5064, Australia

* Corresponding author (ronald.smernik{at}adelaide.edu.au)

Received for publication October 1, 2004.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Several solid-state 13C nuclear magnetic resonance (NMR) techniques were used to characterize soil organic matter spiked with 13C-labeled organic compounds spanning a range of hydrophobicities (benzoic acid, benzophenone, naphthalene, phenanthrene, and palmitic acid). The chemical shifts of NMR resonances of the sorbed species were shifted by up to 3 ppm relative to those of the neat compounds. Sorption also resulted in increased resonance linewidth for the compounds containing a single 13C label, indicating the presence of a range of different chemical environments at the sites of sorption. On the other hand, sorption decreased the linewidth of the resonance of naphthalene, which was uniformly 13C-labeled. This was attributed to the removal of intermolecular 13C–13C dipolar coupling. Heterogeneity of the organic matter was demonstrated using the spectral editing technique proton spin relaxation editing (PSRE), which enabled the identification and quantification of charcoal-rich domains characterized by rapid rates of proton spin–lattice relaxation in the static frame (T1H), and humic domains characterized by slow rates of T1H relaxation. Furthermore it was demonstrated that the sorbed 13C-labeled molecules "inherit" the T1H "signature" of the organic matrix in their immediate vicinity. Thus PSRE on the spiked soils enabled evaluation of the relative affinity of the two domain types for the sorbate molecules. The charcoal-rich domains were shown to have a twofold to tenfold greater affinity for the organic compounds, with greater differences found for the more hydrophobic compounds.

Abbreviations: CP, cross polarization • NMR, nuclear magnetic resonance • PSRE, proton spin relaxation editing • T1H, proton spin–lattice relaxation rate in the static frame • T1{rho}H, proton spin–lattice relaxation rate in the rotating frame


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THE BIOAVAILABILITY, toxicity, persistence, and transport of organic molecules in soils are strongly affected by sorption to the solid phase. For hydrophobic, nonionic molecules, organic matter is the most important sorptive phase. Early studies used a partitioning model to describe the sorption of organic molecules to soil organic matter (Chiou et al., 1979, 1983). However, further investigations uncovered phenomena such as nonlinear sorption isotherms (Weber et al., 1992; McGinley et al., 1993; Huang et al., 1997; Kohl and Rice, 1999; Chiou et al., 2000; Yuan and Xing, 2001), competitive sorption (McGinley et al., 1993), and desorption hysteresis (Benoit et al., 1996; Huang and Weber, 1997; Lueking et al., 2000; Yuan and Xing, 2001) that are inconsistent with a purely partitioning interaction.

Nonpartitioning sorption behavior is widely attributed to the heterogeneity of organic matter (Weber et al., 1992; Xing and Pignatello, 1997; Chiou et al., 2000; Huang et al., 2003). A number of researchers have hypothesized that the soil organic matter matrix consists of various components, or contains various domains with different sorption properties. There is particular interest in the identity of the strongly sorbing components or domains. Suggested candidates include black carbon (Chiou et al., 2000; Ran et al., 2002; Cornelissen and Gustafsson, 2004), kerogen (McGinley et al., 1993; Ran et al., 2003), long-chain hydrocarbon biopolymers (Mao et al., 2002; Chefetz, 2003; Khalaf et al., 2003), and nanometer-size voids (Xing and Pignatello, 1997; Gunasekara and Xing, 2003). However, conclusive identification of strongly sorbing organic matter phases has remained elusive, partly due to the lack of techniques that can identify the location of sorbed molecules in the heterogeneous organic matter matrix.

The solid-state 13C NMR technique proton spin relaxation editing (PSRE) can detect, characterize, and quantify chemically and physically distinct domains within an intact organic matter matrix. The PSRE technique has been used to identify charcoal (Smernik et al., 2000) and highly aliphatic (Preston and Newman, 1992, 1995; Smernik and Oades, 1999) domains in soil organic matter, microbial- and plant-derived domains in sewage sludge (Smernik et al., 2003), and aromatic- and aliphatic-rich domains in kerogen (Petsch et al., 2001).

This paper describes new ways in which solid-state 13C NMR spectroscopy can be used to investigate the sorption of organic compounds to organic matter. By using 13C-labeled organic compounds, detailed information about the chemical of environment of sorbed molecules can be determined directly, at environmentally relevant concentrations. In particular, the application of PSRE to soil organic matter spiked with sorbed 13C-labeled compounds offers a direct way to measure the affinity of different organic matter domains for the sorbed species.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Sample Preparation
The soil used in this study was collected from the top 0 to 50 mm of a Xeric Epiaquert from Urrbrae, South Australia. This soil has an organic carbon content of 32.7 g kg–1, a large portion of which is believed to be natural charcoal (Smernik, 2005). The soil was de-ashed using the procedure of Skjemstad et al. (1994) to isolate the organic matter. The organic carbon content of the HF-treated residue was 381.5 g kg–1. The HF treatment consisted of nine successive treatments with 2% hydrofluoric acid solution. Each treatment involved shaking 3-g portions of soil in 50 mL of 2% hydrofluoric acid solution, end-over-end, for periods of 1 h (five times), 16 h (three times), and 64 h (once). Between treatments, samples were centrifuged, and the supernatant was discarded and replaced with fresh 2% hydrofluoric acid solution. Following the final treatment, the residue was rinsed three times with deionized water, then freeze-dried.

The two melanoidins used were supplied by M. Schmidt (University of Zurich) as reference materials for an interlaboratory "Ring Trial" for black carbon analysis. Details of their synthesis can be found at www.geo.unizh.ch/phys/bc/reference.html#melanoidin (verified 8 Mar. 2005). In brief, urea-glucose melanoidin was prepared by heating D-glucose (monohydrate) (100 g L–1) and urea and (10 g L–1) in the ratio 1:1 (v/v) in glass jars at 90°C for 30 d. The solutions were stirred every second day to prevent a skin from forming on the surface of the liquid. After 30 d, the brown residue formed was isolated by centrifugation, washed four times with deionized water, and freeze-dried. This material was prepared by K. Hammes, University of Zurich, Switzerland. Lysine-glucose melanoidin was prepared by combining alkaline (0.1 M Na2CO3) solutions of D-glucose (30 g L–1) and lysine (24 g L–1) in the ratio 9:1 (v/v) and heating at 100°C for 7 d. The mixture was cooled, then acidified to pH 1 using 1 M trifluoroacetic acid (TFA), and allowed to stand for 16 h. The brown residue formed was isolated by centrifugation, washed with deionized water, and freeze-dried. This material was prepared by S. Brodowski, University of Bayreuth, Germany.

Carbon-13-labeled compounds were purchased from Sigma-Aldrich (St. Louis, MI), and were used as received. The benzoic acid and palmitic acid were singly labeled in the carboxyl position, the benzophenone was singly labeled in the carbonyl position, the phenanthrene was doubly labeled in the 9 and 10 positions, and the naphthalene was fully labeled. In all cases the extent of 13C-labeling was 99 atom percent in the designated position(s). The 13C-labeled compounds were chosen to span a range of polarities, as indicated by their solubility and octanol–water partitioning coefficients (Kow) (Table 1).


View this table:
[in this window]
[in a new window]
  		Table 1. Water solubility and octanol–water partitioning coefficients (Kow) for the organic compounds used in this study.

 
Sorption experiments on the soils were performed by adding approximately 500 mg of de-ashed soil to either water or water and 1-propanol solutions of the 13C-labeled compounds. Experimental conditions (mass of 13C-labeled compound used, solvent system, and solvent volumes) are detailed in Table 2. The 1-propanol co-solvent was used where the water solubility of the 13C-labeled compound was too low for complete dissolution of the required quantity of 13C-labeled compound in 50 mL of water. The quantity of 13C-labeled compound used was chosen so that around 10% of the 13C in the spiked samples would be due to the 13C-labeled compound. It was estimated that this concentration was required to ensure reliable quantification of the contribution of signal from the 13C-labeled compounds in the 13C NMR spectra of the spiked soils. Since the extent of 13C-labeling differed between the sorbates, the loadings of 13C-labeled compound varied by a factor of around five (Table 2). Samples were mixed for 16 h on an end-over-end shaker. The sorbate-spiked residues were isolated by filtration and freeze-dried.


View this table:
[in this window]
[in a new window]
  		Table 2. Experimental details of sorption experiments.

 
Sorption experiments on the melanoidins were performed by adding approximately 500 mg of melanoidin to 50 mL of water containing 1.05 mg of 13C-labeled naphthalene. Samples were mixed for 16 h on an end-over-end shaker. The 13C-naphthalene-spiked melanoidins were isolated by filtration and freeze-dried.

Nuclear Magnetic Resonance Spectroscopy
Solid-state 13C magic angle spinning (MAS) NMR spectra were obtained at a 13C frequency of 50.3 MHz on a Varian (Palo Alto, CA) Unity200 spectrometer. Samples were packed in a 7-mm-diameter cylindrical zirconia rotor with Kel-F end-caps and spun at 5000 ± 100 Hz in a Doty Scientific (Columbia, SC) MAS probe. 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 spectra were zero-filled to 8192 data points and processed with a 50-Hz Lorentzian line broadening and a 0.005-s Gaussian broadening. Chemical shifts were externally referenced to the methyl resonance of hexamethylbenzene at 17.36 ppm.

Cross polarization (CP) spectra of the soils were acquired using a 1-ms contact time and a 1-s recycle delay; 4000 transients were collected for each spectrum. Cross polarization spectra of the melanoidins were acquired using a 1-ms contact time. For the urea-glucose melanoidin, 4000 transients were collected and a 2-s recycle delay was used, while for the lysine-glucose melanoidin, 2000 transients were collected and a 4-s recycle delay was used. "Difference" spectra were generated by subtracting the free induction decay (FID) of spiked soils (or melanoidins) from the FID of the control soil (or melanoidins), followed by Fourier transformation. Inversion–recovery experiments (Smernik et al., 2000) were performed on the soil and melanoidin samples to determine the rate and uniformity of T1H relaxation. For the soils, 13 recovery delays of between 0.1 ms and 1 s were used. A 1-ms contact time and a 1-s recycle delay were used; 2000 transients were collected for each spectrum. For the melanoidins, 13 recovery delays of between 0.1 ms and 4 s were used. A 1-ms contact time and a 4-s recycle delay were used; 1100 to 1200 transients were collected for each spectrum. Inversion–recovery experiments were analyzed by statistically comparing one- and two-T1H component fits to the data, using the method of Smernik et al. (2000). For each soil, the two-component fit was found to be superior, and PSRE subspectra were subsequently generated, representing components characterized by the two T1H values (Smernik et al., 2000).

Cross polarization spectra of the 13C-labeled compounds were acquired using a 1-ms contact time. Carbon-13 CP NMR spectra of 13C-labeled benzoic acid, benzophenone, and naphthalene were acquired in a single scan. The 13C CP NMR spectrum of 13C-labeled phenanthrene represents 360 scans, acquired with a recycle delay of 200 s, while that of 13C-labeled palmitic acid represents 16 scans, acquired with a recycle delay of 100 s. The T1H values for the 13C-labeled compounds were determined using the saturation-recovery method.

Stable Carbon Isotope Analysis
Stable carbon isotopes of the control and spiked soils were analyzed in duplicate on a Europa Scientific (Cheshire, UK) Geo 20-20 stable isotope continuous flow mass spectrometer. Stable carbon isotope ratios for the soil organic matter samples are presented in Table 3, expressed in the usual delta notation ({delta}), and also as the ratio (R) of 13C to 12C and the fraction (f) of 13C to total C (13C + 12C). Values of R and f for the soils were calculated using Eq. [1] and [2], respectively:

[1]

[2]
where Rstandard = 0.0112372 (i.e., the ratio of 13C to 12C for Pee Dee Belemnite, the zero standard on the {delta}13C scale). Values of R and f for the 13C-labeled compounds are also shown in Table 3.


View this table:
[in this window]
[in a new window]
  		Table 3. Carbon stable isotope data, expressed in delta notation ({delta}), as the ratio (R) of 13C to 12C, and as the fraction (f) of 13C to total C.

 
The loading of 13C-labeled compounds in the spiked soils was calculated from the data in Table 3 as follows.

Let the fraction of total C in the spiked soil that is from the 13C-labeled compound be X, then the fraction of total C in the spiked soil that is from the organic matter is 1 – X. The ratio of 13C to 12C in the spiked sample (Rspiked) is given by:

[3]

[4]
where flab and fom are the fraction of total C that is 13C for the 13C-labeled compound, and for the organic matter (i.e., the control soil), respectively. Rearranging Eq. [4] gives:

[5]

Since only 13C nuclei are visible by NMR, the proportion of total 13C that the 13C-labeled compound represents is an important parameter and is denoted here as Y. It was calculated from Eq. [6]:

[6]


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Carbon-13 Cross Polarization Nuclear Magnetic Resonance Spectra of the Soils
The 13C CP NMR spectra of the control soil and the soils spiked with 13C-labeled compounds are shown on the left of Fig. 1 (a–f). The 13C CP NMR spectrum of the control soil (Fig. 1a) shows that it contains a wide range of different carbon types, including unsubstituted alkyl C (0–45 ppm), O-and N-substituted alkyl C (45–110 ppm), aromatic C (110–160 ppm), and carboxyl C (160–190 ppm). The presence of sorbed 13C-labeled compounds is indicated in the spiked soils (Fig. 1b–1f) by the presence of new peaks in the 13C spectra as in Fig. 1c (benzophenone spike) and Fig. 1f (palmitic acid spike), or by a change in the distribution of signal relative to that for the control soil; the benzoic acid spike increases the carboxyl signal in Fig. 1b, while the naphthalene and phenanthrene spikes increase the aromatic signal in Fig. 1d and 1e, respectively. The positions of the spike resonances are indicated by arrows in Fig. 1b–1f.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. (a) Carbon-13 cross polarization (CP) nuclear magnetic resonance (NMR) spectra of HF-treated soil (control soil). (b)–(f) Carbon-13 CP NMR spectra of HF-treated soil spiked with: (b) benzoic acid, (c) benzophenone, (d) naphthalene, (e) phenanthrene, and (f) palmitic acid. (g)–(k) Difference spectra generated by subtracting spectrum (a) from spectra (b)–(f), respectively; these difference spectra represent sorbed (g) benzoic acid, (h) benzophenone, (i) naphthalene, (j) phenanthrene, and (k) palmitic acid. (l)–(p) Carbon-13 CP NMR spectra of neat 13C-labeled compounds: (l) benzoic acid, (m) benzophenone, (n) naphthalene, (o) phenanthrene, and (p) palmitic acid. Spinning sidebands are marked with an asterisk (*).

 
To better characterize and quantify the 13C NMR signal of the sorbed species, "difference spectra" were generated by subtracting the 13C CP NMR spectrum of the control soil (Fig. 1a) from the 13C CP NMR spectra of the spiked soils (Fig. 1b–1f). The difference spectra are presented in the middle column of Fig. 1 (g–k). In all cases there is good cancellation of soil organic matter resonances, leaving mostly resonances for the sorbed species. Some small positive and negative resonances attributable to the soil organic matter are visible in some of the difference spectra (e.g., the positive peak at 130 ppm in Fig. 1h, and the negative peaks at 170 ppm in Fig. 1i and at 130 ppm in Fig. 1k). These residual peaks are due to slight variations in the composition of the soil organic matter between samples.

The right-hand column of Fig. 1 (spectra l-p) contains 13C CP NMR spectra of the neat 13C-labeled compounds. The compounds that contain only one labeled carbon per molecule (benzoic acid, benzophenone, and palmitic acid) produced very narrow resonances. Peak widths at half-height were 47 to 63 Hz (0.94–1.26 ppm) for these compounds (Table 4). The 13C CP NMR spectrum of naphthalene (Fig. 1n), which is fully 13C-labeled, appeared as a single, broad resonance at 128.2 ppm, flanked by two broad spinning sidebands. The peak width at half-height for the naphthalene resonance was 760 Hz (15.2 ppm) (Table 4). There are two causes of this apparent broadness. The first is that there are three different 13C chemical environments in naphthalene, which should give rise to three separate resonances. Indeed, unlabeled naphthalene does produce a spectrum with three sharp resonances at 125.1, 128.8, and 134.0 ppm (spectrum not shown). The single, broad resonance apparent for the fully 13C-labeled naphthalene includes contributions centered at each of these chemical shifts, but each resonance is broadened by 13C–13C dipolar coupling to neighboring 13C nuclei. This broadening mechanism does not occur for the unlabeled naphthalene because 99% of the carbons surrounding 13C nuclei are NMR-silent 12C nuclei. The peak width at half-height for the phenanthrene resonance is 350 Hz (7 ppm) (Table 4), that is, broader than for the singly 13C-labeled compounds, but not as broad as for fully 13C-labeled naphthalene. The phenanthrene is doubly 13C-labeled in the chemically equivalent 9 and 10 positions (see Materials and Methods). Therefore, the labeled carbons are only in one chemical environment, and resonate at only one chemical shift. However, the resonance is significantly broadened by 13C–13C dipolar coupling. Small resonances visible at around 130 ppm in the 13C spectra of pure benzoic acid and benzophenone, and between 15 and 35 ppm in the 13C spectrum of pure palmitic acid, are the natural abundance signals of the unlabeled carbons.


View this table:
[in this window]
[in a new window]
  		Table 4. Chemical shifts of resonance widths at half-height for 13C-labeled compounds before and after addition to soil organic matter.

 
In Table 4, the 13C CP NMR spectra of the 13C-labeled molecules in the free and sorbed states are compared in terms of chemical shift and peak width at half-height. Moderately large upfield shifts (to lower ppm values) were found for benzoic acid (2.4 ppm), naphthalene (0.8 ppm), and phenanthrene (0.7 ppm). A much smaller upfield shift was found for palmitic acid (0.2 ppm). A moderately large downfield shift of 2.9 ppm was found for benzophenone. However, these changes in chemical shift provide little direct information about the nature of the interaction of the different sorbate molecules with the soil organic matter matrix, since the chemical shift in the pure state is influenced by specific intermolecular interactions in the crystal lattice. For example, the chemical shift of the carbonyl resonance of two different crystalline forms of glycine differ by 1.9 ppm (Potrzebowski et al., 1998).

Of the three compounds that gave sharp resonances in the free state (peak width at half-height of 47–63 Hz), sorption resulted in large increases in resonance width for benzoic acid (from 47 Hz in the free state to 380 Hz when sorbed) and benzophenone (63 to 390 Hz), but not for palmitic acid (49 to 52 Hz) (Table 4). Sorption to the organic matter also increased peak width for phenanthrene from 350 to 460 Hz, but the peak width for naphthalene decreased from 760 Hz in the free state to 530 Hz when sorbed (Table 4). The broader resonances observed for benzoic acid, benzophenone, and phenanthrene in the sorbed state indicate that these molecules are in a broader range of chemical environments when sorbed than they are in the pure form. Jurkiewicz and Maciel (1995) reported that the carbonyl resonance of 13C-labeled acetone when sorbed to humic acid was 311 Hz, compared with just 42 Hz when sorbed to kaolinite. They reasoned that the sorption sites on kaolinite were much more homogeneous than those on the humic acid, and hence the size of the shift caused by hydrogen bonding was less uniform for humic acid than for kaolinite. Kohl et al. (2000) also found that hexafluorobenzene produced a very broad resonance when sorbed to peat and attributed this broadness to "the large continuum of different local chemical environments into which the [hexafluorobenzene] may strongly sorb." Khalaf et al. (2003) reported a similar effect for hexafluorobenzene sorbed to soil humic acid.

The resonance for palmitic acid was no broader in the "sorbed" state. This, alongside the fact that the chemical shift changed little on addition to the soil organic matter, indicates that the interaction between palmitic acid and the soil organic matter matrix was very weak. Analysis of proton spin–lattice (T1H) relaxation rates for the palmitic acid spiked sample (see below) provides further evidence that the majority of palmitic acid molecules were not sorbed to the organic matter matrix, but were present as an essentially separate phase.

The decrease in broadness of the naphthalene resonance on sorption reflects a decrease in 13C–13C dipolar broadening, in particular, a decrease in intermolecular 13C–13C dipolar broadening. In the pure state, naphthalene 13C nuclei experience 13C–13C dipolar coupling not only to the other 13C nuclei within the molecule, but also to 13C nuclei in adjacent molecules (dipolar coupling is a through-space interaction). The low concentration of naphthalene in the sorbed state removes the intermolecular interactions. Note that even in the sorbed state, the 13C-labeled naphthalene resonance is broader than that of the other sorbed molecules (Table 4).

The contribution of the 13C-labeled compounds to total signal intensity was determined by integrating the relevant resonances (including associated spinning sidebands) of the difference spectra presented in Fig. 1, and ranged from 3.4% for benzophenone to 17.6% for naphthalene (Table 5). The concentration of the 13C-labeled compounds in the spiked samples was determined independently by stable carbon isotope mass spectrometry (see Materials and Methods). These concentrations are also presented in Table 5, both in terms of overall mass loadings and in terms of 13C balance. The 13C-labeled compounds represented between 0.30% (naphthalene) and 1.94% (palmitic acid) of total carbon and between 6.9% (benzophenone) and 20.4% (naphthalene) of 13C in the spiked samples (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 5. Quantification of sorbate concentration by carbon stable isotopes and solid state 13C nuclear magnetic resonance (NMR) spectroscopy.

 
In each sample, the proportion of 13C derived from the 13C-labeled compound (as determined by mass spectrometry) was greater than the proportion of 13C NMR signal assigned to the 13C-labeled compound (Table 5). This indicates that the NMR observability of the 13C-labeled compounds was on average 5 to 64% lower than that of the soil organic matter. The 13C CP NMR observability of the organic matter itself is 55%, as reported previously (Smernik, 2005). This relatively low value was attributed to the high charcoal content of this soil, and the low intrinsic NMR observability of charcoal. There are a number of possible causes for the low and variable relative NMR observabilities of the 13C-labeled compounds. The highest NMR observability was recorded for phenanthrene (95%), in which both labeled carbons are directly protonated. Cross polarization is most efficient for carbons with directly attached protons. The NMR observability of naphthalene, in which 8 of the 10 labeled carbons are directly protonated, was also high (87%). Lower relative observabilities were recorded for benzoic acid (77%), benzophenone (49%), and palmitic acid (36%), the 13C-labeled carbons of which are not directly bound to hydrogen. Part of the problem for the sample spiked with palmitic acid was found to be saturation of the palmitic acid signal when the spectrum was acquired with a recycle delay of 1 s. Increasing the recycle delay to 20 s increased the intensity of the palmitic acid resonance, and hence its relative observability from 36 to 77%.

The low relative observability of 13C-labeled compounds may also be a consequence of nonuniform distribution between organic matter components. It is shown below that charcoal domains in the organic matter matrix contain a disproportionate concentration of the sorbed molecules. The NMR observability of these domains is low, at least in part due to rapid rates of proton spin–lattice relaxation in the rotating frame (T1{rho}H) (Smernik et al., 2002a, 2002b). Since sorbed molecules "inherit" the relaxation rates of the adjacent matrix, the observability of 13C-labeled molecules sorbed to the charcoal domains is also reduced. Therefore, the low (average) NMR observabilities of the 13C-labeled compounds, relative to that of the organic matter itself, are consistent with disproportionately high concentrations of sorbed molecules in domains that impart low observability on sorbed species.

T1H Relaxation Rates and Proton Spin Relaxation Editing
The effects of sorption on the NMR properties of sorbate molecules extend beyond the effects on chemical shift. In this section, the effects of sorption on rates of proton (1H) spin–lattice relaxation (T1H) of sorbate molecules are examined. Subsequently, chemical shift and T1H relaxation information are combined using the spectral editing technique proton spin relaxation editing (PSRE) to probe the relative affinity of different organic matter domains for the sorbate molecules.

In pure organic compounds, spin–lattice relaxation occurs when molecular motion of magnetic nuclei results in fluctuating magnetic fields at frequencies near the Larmor frequency. In crystalline compounds, molecular motions are highly restricted, and hence rates of spin–lattice relaxation are generally slow. The proton spin–lattice (T1H) relaxation rates of the neat 13C-labeled compounds were determined from saturation-recovery experiments, and are presented in Table 6. The T1H values ranged from 16.9 s for palmitic acid to more than 30 min for benzophenone.


View this table:
[in this window]
[in a new window]
  		Table 6. Proton spin–lattice (T1H) relaxation rates of 13C-labeled compounds.

 
Rates of T1H relaxation for the soil organic matter were at least two orders of magnitude faster than for the 13C-labeled organic compounds, with average T1H values ranging from 62.0 to 114 ms for the control and spiked soils (Table 7). Similar T1H values for HF-treated soils have been reported previously (Smernik et al., 2000). There are two reasons why T1H is much shorter for soil organic matter. First, soil organic matter is an amorphous solid and hence exhibits a greater range and degree of molecular motions than crystalline solids do. Second, soil organic matter also contains paramagnetic species, both organic and inorganic. Unpaired electrons provide a very efficient relaxation mechanism for nuclear spins.


View this table:
[in this window]
[in a new window]
  		Table 7. Results of one- and two-component fits to inversion–recovery data for control soil and soils spiked with 13C-labeled compounds.

 
A process known as spin diffusion also plays an important role in determining spin–lattice relaxation rates in solids. For isotopically dilute nuclei, such as 13C at natural abundance, the rate of spin–lattice relaxation (T1C) is a function of the immediate environment of the nucleus. Hence T1C values will vary for different 13C nuclei in a compound. However, the abundant 1H nuclei in small organic compounds all share the same T1H value, because rapid relaxation of one 1H nucleus induces rapid relaxation of neighboring 1H nuclei, a process facilitated by strong 1H–1H dipolar coupling. This process is called spin diffusion. Over the length scale that it is efficient, spin diffusion causes T1H to be uniform and also short, as overall rates of relaxation are controlled by the 1H nuclei for which relaxation is most efficient. This length scale is a function of T1H relaxation rates, and is around 30 to 100 nm in HF-treated soils (Zumbulyadis, 1983; Smernik et al., 2000).

Each of the control and spiked soils exhibited nonuniform T1H relaxation rates, as determined by statistical comparison of one- and two-component exponential fits to data from inversion–recovery experiments (Smernik et al., 2000). The results of the two-component fits are presented in Table 7. The rapidly relaxing component in each case was characterized by a T1H value between 27 and 54 ms, while the slowly relaxing component was characterized by a T1H value between 143 and 257 ms. The proportions of NMR signal provided by the rapidly and slowly relaxing components varied little between the samples, with the rapidly relaxing component representing between 40 and 51% of the signal and the slowly relaxing component representing between 49 and 60%. Variations in T1Hav, T1Hfast, and T1Hslow between the samples are most probably due to slight differences in moisture content, which in turn affect molecular mobility (Newman, 1992; Hatcher and Wilson, 1991). Although all samples were freeze-dried, they were subsequently exposed to atmospheric moisture during storage. Variations in the length of exposure and the humidity of the atmosphere during this exposure may have resulted in different moisture contents.

Proton spin relaxation editing (PSRE) enables the generation of subspectra for the rapidly and slowly relaxing components, by taking appropriate linear combinations of inversion–recovery spectra, based on the T1H values of the two components (Smernik et al., 2000). The PSRE subspectra of the control soil (Fig. 2) show that the rapidly and slowly relaxing components of this soil have very different chemical compositions. The rapidly relaxing component is highly aromatic, and contains very little alkyl or O-alkyl C. It also contains significant carboxyl carbon, the chemical shift of which (169.2 ppm) indicates that it is primarily attached to aromatic rings. The rapidly relaxing PSRE subspectrum is very similar in appearance to that of weathered charcoal isolated from soils using UV-photooxidation (Skjemstad et al., 1996; Smernik et al., 2000). By contrast, the slowly relaxing component contains little aromatic C, but much more O-alkyl and alkyl C. It also contains significant carboxyl C, but at a more positive chemical shift (173.3 ppm), indicating attachment primarily to aliphatic groups. The slowly relaxing PSRE subspectrum is more typical of "humic" material. Proton spin relaxation editing therefore provides compelling evidence that the organic matter in this soil is heterogeneous, and consists of two types of organic domains: highly aromatic domains of weathered charcoal and predominantly aliphatic domains of humified plant material.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 2. Carbon-13 proton spin relaxation editing (PSRE) subspectra of HF-treated soil and HF-treated soil spiked with 13C-labeled compounds.

 
An important consequence of spin diffusion, in the context of this study, is that the T1H relaxation rate of small organic molecules sorbed to the soil organic matter matrix reflects that of the organic matter domain to which they are sorbed (see below). Thus the NMR signal of the sorbed molecules appears in the PSRE subspectrum of the domain to which they are sorbed. It is clear from Fig. 2 that other than for palmitic acid (see below), the 13C-labeled molecules are sorbed to both the rapidly and the slowly relaxing components. For the soil spiked with benzoic acid, both the rapidly and slowly relaxing subspectra contain more carboxyl signal (170–175 ppm) than do the corresponding PSRE subspectra for the control soil (Fig. 2). However, it would appear that there is a greater enhancement of the carboxyl signal for the rapidly relaxing subspectrum than for the slowly relaxing subspectrum, suggesting that the rapidly relaxing component has a greater affinity for benzoic acid. For the soil spiked with benzophenone, both subspectra contain the signal at 195 to 200 ppm, attributable to the sorbate molecules (Fig. 2). Again, this signal appears to be more intense in the rapidly relaxing subspectrum. For the soils spiked with naphthalene and phenanthrene, it is a little difficult to assess the contribution of the sorbate molecules to the rapidly relaxing subspectra, since the chemical shift of the sorbate resonances coincide with the strong aromatic resonance of the organic matter (Fig. 2). However, by comparing the relative sizes of the aromatic and carboxyl signals, it is clear that a substantial proportion of the signal in these subspectra is from the 13C-labeled sorbate molecules. There is also a significant enhancement of aromatic signal in the slowly relaxing PSRE subspectra, though more so for the soil spiked with naphthalene than for the soil spiked with phenanthrene.

The contribution of the sorbed molecules to the NMR signal in each PSRE subspectrum was quantified by subtracting the corresponding PSRE subspectrum for the control soil from each PSRE subspectrum of the spiked soils. The intensity of the subspectrum of the control soil was scaled so as to minimize the size of organic matter resonances in the difference spectra. The resonances of the sorbate molecules in the difference spectra were then integrated. Table 8 shows the percentage contribution of the 13C-labeled compounds to each of the PSRE subspectra of the spiked soils (except the soil spiked with palmitic acid). Clearly, in each case the sorbate contribution is greater for the rapidly relaxing subspectrum than the slowly relaxing subspectrum. In other words, the rapidly relaxing component has a higher affinity for the sorbate molecules. The ratio of Koc for each component is also shown in Table 8. This ratio was calculated using Eq. [7]:

[7]


View this table:
[in this window]
[in a new window]
  		Table 8. Proportions of nuclear magnetic resonance (NMR) signal in proton spin relaxation editing (PSRE) subspectra derived from 13C-labeled compounds and Koc ratios for PSRE components.

 
The ratios of Koc calculated using Eq. [7] may be influenced by the low NMR observability of the sorbed molecules. As discussed above, the observability of the 13C nuclei in the sorbed molecules was 5 to 64% lower than for the 13C nuclei in the organic matter. However, where this reduced observability is due to slow rates of cross polarization of nonprotonated carbons in the sorbed species (i.e for benzoic acid and benzophenone), there should be no bias in the ratio of Koc for the two PSRE-detected components, as the observability of the sorbed species will be equally affected no matter which component it is sorbed to.

Table 8 indicates that the rapidly relaxing component has a twofold to nearly tenfold greater affinity for each of the 13C-labeled compounds than does the slowly relaxing component. Furthermore, the difference in affinity increases with increasing hydrophobicity of the sorbate molecule. These results provide direct support for the hypothesis (Chiou et al., 2000; Ran et al., 2002; Cornelissen and Gustafsson, 2004) that charcoal or black carbon domains sorb nonionic compounds more strongly than do other organic matter components, at least for this soil. However, the rapidly relaxing component does not appear to have the one to two orders of magnitude stronger sorption affinity for hydrophobic molecules that have been found for pure soots and chars (Bucheli and Gustafsson, 2000). There are a number of possible reasons for this. Cornelissen and Gustafsson (2004) found the char component of sediments had a higher sorption affinity when isolated than when other organic matter components are present. They suggested that the char component of sediment might contain sorbed native polyaromatic hydrocarbons or organic matter that compete for sorption sites with added hydrophobic compounds. Another possible reason is that the laboratory-produced chars are quite different from environmental chars. Laboratory chars are usually produced at high temperatures and under controlled conditions that ensure efficient char production, whereas much of the charred material produced in a wild fire is incompletely burned or is heated to relatively low temperatures or for short durations. We have found that chars produced at high temperature are much more sorptive than chars produced at lower temperatures (unpublished data, 2004).

It should be noted that this soil was chosen because of it is known to have a high char content. The results presented here do not preclude other types of organic matter domains playing important or even dominant roles in the sorption of nonionic compounds in other soils. Indeed, polymethylene-rich domains, which have been suggested as another type of organic matter involved in strong sorption, have been identified in other soils using PSRE (Preston and Newman, 1992, 1995; Smernik and Oades, 1999). Such soils will be the subject of future investigations using the techniques described here.

The PSRE subspectra for the soil spiked with palmitic acid are very different from those for the other spiked soils (Fig. 2). The rapidly relaxing subspectrum contains an inverted resonance at the chemical shift of the palmitic acid carboxyl carbon. The corresponding slowly relaxing subspectrum contains an intense (positive) peak at this chemical shift. This indicates that T1H for the palmitic acid is longer than the value of T1Hslow determined for this sample (Smernik et al., 2000), and provides further evidence that the palmitic acid molecules are not sorbed to the organic matter matrix and instead exist as an essentially separate phase. To determine T1H for the palmitic acid in this sample, a second inversion–recovery experiment was performed, this time with a recycle delay of 10 s rather than 1 s, so as to avoid saturation of the palmitic acid signal. The T1H value for the palmitic acid resonance was found to be 2.6 s. That this is considerably shorter than the T1H value determined for neat palmitic acid (16.9 s, Table 6) suggests that there may be some interaction between the palmitic acid phase and the soil organic matter matrix, or that the palmitic acid domains are relatively small. Figure 3 shows that signal attributable to sorbate molecules in 13C CP NMR spectra of the sample spiked with palmitic acid increases as the recycle delay increases from 1 to 20 s. This is a consequence of the relatively long T1H value of the sorbate molecules.



View larger version (11K):
[in this window]
[in a new window]
  		Fig. 3. Carbon-13 cross polarization (CP) nuclear magnetic resonance (NMR) spectra of HF-treated soil spiked with 13C-labeled palmitic acid acquired with recycle delays of (a) 1, (b) 5, and (c) 20 s.

 
T1H Relaxation Rates of Model Materials (Melanoidins) Spiked with Carbon-13-Labeled Naphthalene
The key requirement of the PSRE technique described above is that the T1H relaxation rate of sorbed molecules is same as that of the surrounding organic matter matrix. From our understanding of spin diffusion, this should be the case, and the fact that T1H for the sorbed molecules are at least two orders of magnitude shorter than for the pure compounds, and in the same range as the T1H values of the organic matter supports this. However, definitive confirmation requires sorption experiments with a simpler organic matrix, in particular, a matrix with a homogeneous T1H relaxation rate. Melanoidins, the brown condensation products formed by heating solutions of sugar and urea or an amino acid, have this property.

Figure 4 shows the 13C CP NMR spectra of two melanoidins, with and without sorbed 13C-labeled naphthalene, and the difference spectra generated by subtracting the spectrum of the spiked melanoidin from the corresponding control sample. The two melanoidins have different chemical compositions—the urea-glucose melanoidin is more aromatic, while the lysine-glucose melanoidin contains more alkyl C. In both cases, the extra aromatic signal provided by the 13C-labeled naphthalene is clear in the 13C CP spectra of the spiked sample, and in the difference spectrum.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 4. Carbon-13 cross polarization (CP) nuclear magnetic resonance (NMR) spectra of melanoidins without (control) and with (spiked) sorbed 13C-labeled naphthalene. Difference spectra were generated by subtracting control spectra from corresponding spiked spectra.

 
In contrast to the soil samples, for which the two-component fit to the inversion–recovery data was much better than the one-component fit, for the melanoidins the two-component fit provided no improvement over the one-component fit. This indicates that T1H was uniform for the melanoidins. The T1H values were different for the two melanoidins, 340 ms (urea-glucose) and 550 ms (lysine-glucose), which is probably a reflection of differences in molecular mobility between the two materials. Figure 5 shows a selection of the inversion–recovery spectra for the two melanoidins. In both cases, all resonances invert at the same rate, further emphasizing the uniformity of T1H. Importantly, the aromatic resonance, which contains signal from the sorbed 13C-labeled naphthalene, exhibits identical T1H behavior to all other resonances. The uniformity of T1H for the melanoidins contrasts with the clearly nonuniform T1H for the HF-treated soil spiked with 13C-labeled naphthalene (Fig. 5), as evidenced by the mixture of positive and negative resonances at a recovery delay of 35 ms.



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 5. A selection of 13C inversion–recovery NMR spectra (showing spectra for 8 of the 13 recovery delays used) for the melanoidins and the HF-treated soil, each spiked with 13C-labeled naphthalene.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
This study was designed as a preliminary "proof of concept" investigation of a new method for studying the sorption of organic compounds to organic matter. It shows that sorbed molecules are readily detected by solid-state 13C NMR spectroscopy when they contribute 10% or even less of 13C nuclei in a sample. For fully 13C-labeled molecules, this equates to concentrations of 0.1% on a per carbon basis, or around 40 ppm for a soil containing 4% organic carbon. Proton spin diffusion ensures that sorbed molecules inherit the T1H "signature" of their immediate environment, as proven for the melanoidins spiked with 13C-labeled naphthalene. Where PSRE is successful in identifying distinct organic matter domains, PSRE on samples spiked with 13C-labeled compounds enables the determination of the relative sorption affinities of the domains.

The techniques described here will be most useful as a supplement to, rather than a replacement for, techniques currently used for sorption studies. These new techniques cannot achieve the same level of sensitivity as batch sorption, and are more time-consuming and expensive. However, the information they provide on the affinity of different components within the organic matrix is something neither batch sorption, nor any other technique in which only the solution phase is analyzed, can offer.

In this paper, the NMR techniques have been used to study the sorption of a range of organic molecules to organic matter isolated from a single soil. Each of the sorbate compounds showed greater affinity for the highly aromatic, charcoal-rich domains. The degree of domain selectivity increased with increasing hydrophobicity of the sorbate molecules. Clearly, it would be of interest to repeat these investigations using organic matter of varying chemistries isolated from a range of soils. Of particular interest would be soils containing hydrophobic aliphatic domains, or with significant concentrations of kerogen, two organic matter types that have been ascribed strong sorption properties (McGinley et al., 1993; Chefetz, 2003; Khalaf et al., 2003; Ran et al., 2003). The techniques described here are also equally applicable to studying other nonpartitioning sorption behaviors, such as desorption hysteresis, competitive sorption, and slow sorption kinetics. These areas are the subject of ongoing investigations.


   ACKNOWLEDGMENTS
 
This work was funded by an Australian Research Council (ARC) grant, and with the cooperation of CSIRO Land and Water. In particular, the contributions of Sonia Grocke and Dr. Rai Kookana are gratefully acknowledged.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2005 34: v. [Full Text]  




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Smernik, R. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Smernik, R. J.
Agricola
Right arrow Articles by Smernik, R. J.
Related Collections
Right arrow Organic Compounds
Right arrow Soil Pollution
Right arrow Isotopes
Right arrow Soil Organic Matter

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH