Research Separation of Structural Components in Soil Organic Matter by Diffusion Ordered Spectroscopy ANDRE ´ J . S I M P S O N , * ,† WILLIAM L. KINGERY,† MANFRED SPRAUL,‡ EBERHARD HUMPFER,‡ PETER DVORTSAK,‡ AND RAINER KERSSEBAUM‡ Department of Plant and Soil Sciences, Mississippi State University, Mississippi State University, Mississippi, 39762, and Bruker Analytik, GmbH, Silberstreifen, D-76287 Rheinstetten, Germany
Diffusion ordered spectroscopy (DOSY) was applied to two extracts of organic matter from the surface horizon of an oak forest soil. It was possible to identify and confirm the presence of numerous aliphatic, aromatic, sugar, and amino acid components that could be separated on the basis of diffusion in DMSO-d6 and D2O. On average, sugar components were identified as the largest molecules in solution, with molecular masses up to ∼1500 Da followed by the aliphatic and aromatic components. Amino acids with a range of molecular weights were also identified in the mixture. The summation of the individual slices from the DOSY experiment closely resembles the conventional 1H spectra of the material, indicating that the components identified with DOSY represent all the components present in the mixture. The separation of components in the mixture in organic solvent supports new findings that fulvic and humic acids are not cross-linked, high molecular weight macromolecules but are instead aggregates composed of relatively simple molecules that take on colloidal properties in the presence of metal ions in aqueous solution. Using the knowledge that these organic mixtures are combinations of relatively simple entities with welldocumented reactivities and behavior will improve our ability to predict and model their interactions and fate under natural conditions.
Introduction Humic substances are formed by the chemical and biological transformations of plant and animal matter and constitute the major pool of organic carbon at the earth’s surface. They are widely distributed in soils, waters, and sediments and are known to play important roles in the transport of inorganic and organic contaminants (1, 2) as well as in soil formation (3). Traditional thinking has suggested that humic substances are produced by the cross-linking of degradation products to form high molecular weight macromolecules that are * Corresponding author present address: Dept. of Chemistry, The Ohio State University, Box 67, 100 W. 18th Ave., Columbus, OH 43210; fax: (614)688-4906; e-mail:
[email protected]. † Mississippi State University. ‡ Bruker Analytik, GmbH. 10.1021/es0106218 CCC: $20.00 Published on Web 10/06/2001
2001 American Chemical Society
microbially and chemically recalcitrant (4). These concepts have risen from observations of high molecular weights in aqueous solution using a number of techniques including gel permeation, low-angle X-ray scattering, ultracentrifugation, viscometry, and light reflectance spectroscopy (5). However, recent studies have demonstrated that these are only pseudo-molecular weights and result from the aggregation of many distinct categories of components such as sugars, aliphatic esters/ethers, amino acids, and aromatic components (6-9). Therefore, in organic solvents, such as dimethyl sulfoxide (DMSO), or at high pH in aqueous solution, the aggregates will disperse, and it should be possible to separate the individual components on the basis of their diffusion. Diffusion coefficients provide information on molecular dynamics and are directly related to hydrodynamic radii in solution (10). In a pulsed field gradient (PFG) NMR experiment using a bipolar pulse-pair longitudinal encode-decode (BPPLED) sequence (11), the intensity of a resonance in an NMR spectrum (I) is related to the diffusion coefficient of the molecule (D) as described:
I ) Io exp[-D(∆ - δ/3 - τ/2)(gγδ)2]
(1)
where Io is the intensity in absence of a gradient pulse; ∆ is time through which diffusion occurs; g and δ are the amplitude and duration of the gradient pulses, respectively; τ is the delay after each gradient pulse; and γ is the gyromagnetic ratio. A number of spectra are then collected with increasing values of g. The object of the DOSY experiment is to then transform the 2-D data set, which is simply a stack of attenuated 1-D spectra, into 2-D spectra with chemical shifts on one axis and diffusion coefficients on the other. This may be carried out by a number of automated processes that have been reviewed in depth by Johnson (12). Although alternative techniques are available to extract diffusion coefficients from complex mixtures such as humic substances (13), DOSY has the distinct advantage that the measured diffusion coefficient are directly correlated to the chemical shift in the second dimension, and thus information on both the structures present and their diffusivities can be obtained. DOSY is an NMR technique that is rapidly gaining wider use but which has yet to be fully utilized in organic geochemical and environmental studies. Morris et al. (14) used DOSY in an excellent study to asses the distribution of diffusion coefficients in a number of fulvic and humic acids in order to qualitatively describe the polydispersity of the mixtures (14). However, it has yet to be reported where DOSY is used to separate components in mixtures of humic substances on the basis of their diffusion coefficients. Such separation would provide a powerful analytical tool that would help to resolve spectra for these complex mixtures and to determine the individual entities present as well as to permit monitoring of changes in the diffusion of the humic components associated with contaminants or during aggregation/flocculation processes. Such information is vital if we are to better predict the behavior of these ubiquitous and heterogeneous mixtures under natural conditions and in response to ecosystem perturbations.
Experimental Section Humic material was isolated from the Ah (surface horizon) of an oak forest soil located in Uragh Wood, Lough Inchiquin, VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4421
Kenmare, County Kerry, Ireland (grid reference: V.83.62). Detailed descriptions of the soil profile and vegetation cover are given by Little (15). The fulvic acid used in this study was obtained by exhaustive extraction with NaOH at pH 12.6 of the soil residue remaining after previous sequential extractions with of 0.1 M sodium pyrophosphate (Pyro, adjusted to pH 7.0 using HCl) and Pyro (0.1 M) at pH 10.6. The humic (HA) and fulvic (FA) acids were separated by precipitation with HCl and isolated using XAD-8 and XAD-4 resins in tandem as described by Simpson et al. (16). Metals were then removed using IR-1200H-plus cation-exchange resin. The whole-soil extract was isolated with 0.1 NaOH from the bulk soil that had been passed through a 1-mm sieve and treated on XAD-8 resin (the HA and FA acid components were not separated). The whole-soil extract was then repeatedly (×5) passed over IR-1200H-plus cation-exchange resin in order to remove metal species, which may instigate precipitation. In preparation for NMR, samples (50 mg) were dried for 48 h at 30 °C over P2O5 to remove excess moisture and then dissolved in DMSO-d6 (1 mL for the FA) and D2O (1 mL for the whole-soil extract). Solution-state 1H NMR was carried out on a Bruker Avance 500 MHz spectrometer fitted with a 5-mm 1H-13C cryoprobe with an actively shielded Z gradient. The 1H spectrum was acquired using 8 scans and a delay of 2 s between pulses. PFG NMR data were collected using the BPPLED pulse sequence (“ledbpgs2s” in the standard Bruker pulse sequence library). Scans (1024) were collected using 1.5-ms sine-shaped gradient pulses (3 ms bipolar pulse pair) ranging from 6.8 to 323.5 mT m-1 in 16 increments, with a diffusion time of 100 ms and sample temperature at 298 K. The 2-D DOSY spectrum was calculated using the “ilt” plugin for XWIN NMR with 4096 data points in the 1H dimension (F2) and 128 data points in the diffusion dimension (F1). Biexponential decays were fitted to every data point in F2 (1H dimension) to yield the diffusion coefficients (m2 s-1). The resulting diffusion coefficients and their standard deviations were used to generate a 2-D DOSY spectrum with chemical shifts plotted against -log diffusion coefficients (D). As an alternative method, diffusion coefficients were calculated using CONTIN (17, 18). This computer-based analysis assumes a continuous distribution of molecular weights within the data set. Both methods of analysis produced similar results although CONTIN tended to “over smooth” the data, as previously reported for a number of mixtures (12), slightly decreasing the resolution in the diffusion dimension. As a result, the data set treated with biexponential fitting are shown in this paper.
Results and Discussion The materials used in this study have previously undergone extensive analysis using a battery of multidimensional NMR techniques (6, 19, 20) and chemical degradations, including pyrolysis-GC/MS and chemical degradation combined with GC/MS (19). These studies demonstrate that the components fall into four main structural categories: aliphatics, sugars, and aromatics, with a lesser yet significant contribution from amino acid/peptide structures. Fulvic Acid. DMSO-d6 is an excellent solvent for humic substances that can dissipate hydrogen bonds and prevent aggregation of organic structures. As the purpose of the study is to separate the individual molecular species in humic materials based of their diffusivities, DMSO-d6 was the initial solvent of choice. Figure 1 illustrates where these categories appear on the 1H NMR of the FA. When 2-D spectroscopy is applied to complex mixtures, like humic substances, there is a concern that some molecules with fast spin-spin relaxation (T2) will relax during the experiment and are then underestimated or not detected. In a DOSY experiment, the summation of all the horizontal slices can yield a projection 4422
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
FIGURE 1. (A) 1H NMR spectrum of the FA in DMSO-d6; I and II represent CH2 groups adjacent to ester and ether structures, respectively (6). (B) Projection showing the sum of all the positive signals in horizontal planes of the 2-D DOSY spectrum. The strong similarity of the two spectra indicate that few of the signals are lost through relaxation during the 2-D DOSY experiment and that it represents all the components in the mixture. The negative signals superimposed in spectrum B result from the mathematical treatment of the data in the DOSY analysis where a small number of negative signals have been unavoidably included in the summation of the positive projection. These negative signals should not be considered when comparing the overall profiles of spectra A and B. in which all the detected signals are apparent. If all the components in a mixture are detected through DOSY, then the projection from the 2-D should be a detailed reflection of the 1-D spectra. Indeed this is observed (Figure 1), which suggests that little, if any, signal is lost in the experiment from relaxation; hence, all the components in the mixture are detected. Figure 2 displays a 2-D contour plot of the DOSY data for the FA. It is important to note for the sake of clarity that Figure 2 has been plotted so that only the major signals are apparent. The relatively low resolution (128 data points) in the diffusion dimension results in broad peaks, which tend to spread as the contour intensity is increased and mask less intense signals. Therefore weaker signals, from these components present at lower concentrations, are best viewed in individual slices from the 2-D plot (discussed in detail below). From the 2-D contour, it is possible to observe separation of the major structural components on the basis of their diffusion coefficients (D) in solution (Figure 2). The sugar components were observed to have the slowest diffusion in solution with the most intense signals falling in the range of 9.55-9.65, indicating that they have the largest molecular radii of the components in the mixture and are therefore likely to be present in the form of chains. However, providing accurate molecular weights/sizes of these components in the FA is practically impossible due to the fact that the exact structural components are not known and that there is little published information on the diffusion of sugars in DMSOd 6.
FIGURE 2. 2-D DOSY projection of the FA in DMSO-d6; I and II represent CH2 groups adjacent to ester and ether structures, respectively. A-E label the slices referred to in the text and displayed in Figures 3 and 4, respectively. The major aliphatic components have faster diffusivities (-log(D) ) 9.45-9.58) than the sugars and have been identified as fatty acids and fatty esters/ethers of varying lengths that can reach in excess of 25 carbons (19, 6). Peaks I and II (Figures 1 and 2), which have been assigned to aliphatic CH2 units adjacent to ester and ether, respectively, are of particular interest. Figure 2 clearly depicts separation of these signals on the basis of their diffusion from the large mass of sugar residues with approximately the same 1H chemical shifts and consequently helps to confirm the aliphatic assignment of these resonances. As with the sugars, providing an estimate of molecular weights/sizes is very difficult. Literature values indicate that the short-chain mono and dicarboxylic (1-7 carbons) acids exhibit -log(D) in the region of 8.8-9.15 in H2O (21). The slower diffusivities observed for the aliphatic components in the FA (-log(D) ) 9.45-9.58) suggest that the aliphatics are considerably larger than this, but an accurate estimation of chain length is inappropriate without consideration of the effect of DMSOd6 on the diffusivities of the species in this mixture. The aromatics in the FA are predominately lignin-derived, and the sharp line shapes in 1-D (Figure 1A) and 2-D spectra (data not shown) suggest that they are of relatively low molecular weight (6). This is confirmed by the DOSY experiment, which indicates that they are among the fastest diffusing components in the mixture. However, little can be derived in terms of the molecular weights of the species when DOSY is carried out in DMSO-d6 only. Since, as was mentioned above, the more intense regions tend to spread and obscure weaker signals, it is necessary to access information from these weaker signals by viewing individual 1-D projections created from the rows and columns of the 2-D data set. In Figure 3 is an example of four projections created by slicing though the diffusion axis along the planes labeled A-D in Figure 2. Slice A intersects the diffusion axis at -log(D) ) 9.63, which is consistent with the slower diffusing molecules identified in the mixture. The 1-D slice in Figure 3A displays chemical shifts consistent with sugar residues (visible in the 2-D plot in Figure 2 at 3.5-6 ppm) as well as a number of small aliphatic (0-3 ppm) and aromatic signals (6-8 ppm). These results suggest that along with the more abundant aliphatic components, which exhibit faster diffusion, there are also a number of larger systems. This is consistent for a sample that contains a range of molecules/fragments, all at various stages of humification. Furthermore, in addition to identification of weaker signals,
FIGURE 3. 1H chemical shift profiles obtained by taking slices through the diffusion axes at (A) -log(D) ) 9.63, (B) -log(D) ) 9.53, (C) -log(D) ) 9.47, and (D) -log(D) ) 9.33 in Figure 2. The profiles display chemical shifts identified as follows: (A) predominantly sugars, with smaller contributions from aliphatic and aromatic components; (B) aliphatic components; (C) a substituted cinnamyl; and (D) amino acids. the 1-D projections provide increased resolution over the 2-D DOSY plot. From the 2-D plot, a number of resonances at 0-3 ppm are visible, but the 1-D projection provides a more obvious profile from which detailed chemical shift information can be obtained. Slice B intersects the diffusion axis at -log(D) ) 9.53 and represents molecules in the mixture with medium diffusion coefficient values. The 1-D projection from slice B displayes numerous aliphatic signals. Using prior knowledge of the sample and interpretations made from a series of multidimensional experiments (6), the signals can be assigned to CH2 groups in alkanes, fatty acids, esters, and ethers (1-2 ppm), CH2 groups adjacent to COOH groups (2-3 ppm, predominately in fatty acids/esters), and CH2 units VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4423
adjacent to ethers and esters (3-4.2 ppm). Slice C intersects the diffusion axis at -log(D) ) 9.47 and exemplifies molecules that exhibit medium/fast diffusion. The 1-D projection of this region displays aromatic, olefinic, and aliphatic chemical shifts consistent with substituted cinnamyl structures that represent over 3% of the sample by weight (19) and are major structural units of the oak biomass. Slice D intersects the diffusion axis at -log(D) ) 9.33 and corresponds to molecules that exhibit some of the fastest diffusivities in the mixture. The 2-D projection does not show molecules present in this region, yet the 1-D projection clearly indicates signals consistent with amino acids that constitute 5.2 wt % of the sample (19). These weaker signals cannot be observed in the 2-D projection. Therefore, with humic substances that contain a diverse array of molecules, it is necessary to scan through the rows of the 2-D spectrum if signals from minor components are of interest. Slicing through the 1H chemical shift axis and evaluating the information in each column of the 2-D data set can also provide useful structural information but should be undertaken with caution. With compounds and mixtures that produce well-resolved proton spectra, a peak on the chemical shift axis results from protons on a specific molecule. From the DOSY spectrum, the molecular diffusion coefficient can be extracted and utilized to provide a general understanding of the molecular weights of the specific species of interest. However, the situation becomes complicated with complex mixtures, such as humic substances, that have a high degree of overlap in the chemical shift dimension. If a chosen resonance peak results from a number of overlapping protons in different molecules, the diffusion axis will reflect a multitude of diffusion coefficients consistent with the number of molecules present. For instance, if the number of protons contributing to a specific proton peak is unclear, then it is impossible to distinguish whether two correlated separate diffusion values result from monomers or dimers of the same molecule or two entirely different structures. However, as discussed above, slicing through the diffusion axis provides detailed chemical shift information on all the protons in a molecules, which can be used to aid and confirm assignments, and reduces the potential for errors that may arise if only one chemical shift point is considered. Given this general consideration, with humic substances it may be appropriate to first consider the extra structural information afforded in the chemical shift dimension after separation on the basis of diffusion. Then, using these interpretations, correlation of the structures identified with diffusivities in the second dimension permits insights into possible molecular weights and structural associations or aggregation. An exception to this approach arises when there is an area in the 1-D 1H spectrum that does not exhibit considerable chemical shift overlap from different structures and the nature of the resonance is known. In the 1H spectra of humic substances there is one region where this situation occurs. The peaks at ∼8.5 ppm have been shown repeatedly in D2O exchange and TOCSY experiments to result from the amide signals in peptides (6, 19, 22). In this instance, slicing through the chemical shift axis can provide a method that is less time-consuming than scanning through the individual rows in order to gain information on the molecular weights of the peptide components, which are not apparent in the 2-D projection. Slice E in Figure 2 intersects the axis at 8.46, 8.49, and 8.52 ppm, and the three resulting diffusion profiles are displayed in Figure 4. It is interesting to note that in each slice there are a multitude of observed diffusivities. This suggests that the peptide protons are found in a variety of structures with varying molecular weights and chain lengths. As noted above, hypothesizing molecular weights/sizes of compounds in DMSO-d6 is not appropriate. 4424
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 22, 2001
FIGURE 4. Diffusion profiles of peptide components in the fulvic acid mixture obtained by taking slices through the 1H chemical shift axes (see Figure 2) at 8.46 (A), 8.49 (B), and 8.52 ppm (C).
FIGURE 5. 2-D DOSY projection of the whole-soil extract in D2O. Whole-Soil Extract. Although DMSO-d6 is an excellent solvent for soil organic matter, the literature addressing the diffusivities of compounds in this solvent are sparse. Numerous studies, however, have been carried out to measure the diffusivities of a range of compounds in water. Considering this and after the successful separation of the component in the FA, an additional DOSY experiment was carried out using the total organic extract with D2O as the solvent. Prior to NMR analysis, the material was subjected to repeated cation exchange. This had the effect of removing metal cations that could aggregate the individual organic structures and thereby provided a material that was completely soluble in D2O. When applied to the whole-soil extract, the major structural categories showed distinctly separate diffusivities when dissolved in D2O. In Figure 5 is depicted the 2-D DOSY spectrum of the extract on which the different diffusivities of the aliphatic, carbohydrate, and aromatic components can be clearly seen. The separation of the organic components in the whole-soil extract is noteworthy. Traditionally humic substances have been thought to be macromolecular, yet recent thinking suggests that they are aggregates of much smaller species (9, 23). The separation of the components on
basis of the diffusivities tends to support this more recent concept. Diffusivities observed in the whole-soil extract were generally similar to those observed for the FA in DMSO-d6. However, the whole-soil aromatic species exhibit an average diffusivity (centered on -log(dc) ) -9.6) that is lower than that of the aliphatic structures in the same mixture. This is the reverse of the situation observed in the FA. It is likely the HA fraction, which is part of the whole-soil extract, contains a number of slightly larger aromatic species that contribute to the aromatic intensities with lower diffusion coefficients. The carbohydrates had the lowest diffusion coefficients and therefore the highest hydrodynamic radius in solution. The diffusivities of the carbohydrate species are consistent with fragments of 3-8 sugar units in water (∼600-1500 Da; 24), assuming that they are between 5 and 50% by weight of the sample in the NMR tube, which is well within estimates made from integration of 1H resonance peaks. The amino acid signals are relatively weak and not distinguishable in Figure 5. However, these signals are clearly discernible in individual diffusion-axis slices, and their coefficient values are consistent with diffusivities of polypeptides in water of up to 10 units, depending on the side chain substituents (25, 26). The aliphatic components present display lower diffusion coefficient values than those reported for 1-7 carbon monoand dicarboxylic acids in aqueous solution (21). This suggests that the aliphatic structures in our HS samples are likely to be considerably larger. Estimations based on the model by Oelkers (27) indicate that monomers, dimers, trimers, and even tetramers of C16 and C18 fatty esters fall within the observed range. Since the conformation and linkage between the aromatic units is not yet fully determined, identification of the exact aromatic species present is not possible at this time. With reference to the other components, it is reasonable to reckon from the diffusivities observed that lignin dimers, trimers, and tetramers are in abundance and that fragments containing more than 8 units are highly unlikely. In conclusion, DOSY is shown to be a very powerful technique for the separation of components in humic substances. Separation provides increased resolution in the chemical shift dimension, which aids in the interpretation and confirmation of the wide array of structures present in organic matter and also provides vital insight into the associations of the structures. Furthermore, for mixtures in which a contaminant has been introduced, DOSY has the potential to map the changes in diffusion of both the contaminant and the surrounding molecules providing direct evidence of interactions. In this study, DOSY was able to separate the components in the fulvic acid and show them to be relatively small molecules. However, when analyzed by gel permeation chromatography in aqueous media, all the components in the FA have apparent molecular masses in excess of 6000 Da (19). Traditional thinking has suggested that animal and plant degradation products in humic substances are cross-linked to form stable, high molecular weight macromolecules. Such concepts have been fueled by observation of apparent molecular masses that can exceed 1 000 000 Da in aqueous solution. Average structures have even been constructed using molecular modeling packages based on findings from pyrolysis data, although there is no analytical evidence to support cross-linking of the molecules (28). The DOSY data presented here further support recent findings that the components in humic substances are not extensively cross-linked (6-9) by demonstrating separation in solution on a diffusion basis. Using the knowledge that these organic mixtures are combinations of relatively simple entities that have well-documented reactivities and behavior, scientists will be able to better predict their interactions and
fate in natural environments and in response to anthropogenic influences.
Acknowledgments We thank Dr. Michael Hayes, Limerick University, Ireland, for guidance in sample fractionation and isolation. The Mississippi Agricultural and Forestry Experiment station and Remote Sensing Technology Center, Mississippi State University, provided financial support for this research.
Literature Cited (1) Linn, D. M.; Carski, T. H.; Brusseau, M. L.; Chang, F. H. Sorption and Degradation of Pesticides and Organic Chemicals in Soil; SSSA Special Publication 32; Soil Science Society of America: Madison, WI, 1993. (2) Selim, H. M.; Amacher, M. C. Reactivity and Transport of Heavy Metals in Soils; CRC Press Inc: Boca Raton, FL, 1997. (3) Chaney, K.; Swift, R. S. J. Soil Sci. 1984, 35, 223-230. (4) Cameron, R. S.; Thornton, B. K.; Swift, R. S.; Posner, A. M. J. Soil Sci. 1972, 23, 394-408. (5) Swift, R. S. In Humic Substances II: In Search of Structure; Hayes, M. H. B., MacCarthy, P., Malcolm, R. L., Swift, R. S., Eds.; Wiley: Chichester, 1989; pp 449-466. (6) Simpson, A. J.; Burdon, J.; Graham, C. L.; Spencer, N.; Hayes, M. H. B.; Kingery W. L. Eur. J. Soil Sci. 2001, 52, 495-509. (7) Simpson, A. J.; Kingery, W.; Hayes, M. H. B.; Spraul, M.; Humpfer, E.; Dvortsak, P.; Kerssebaum, R.; Godejohann, M.; Hofmann M. Proceedings of 10th Meeting of the International Humic Substances Society, Toulouse, France; 2000; pp 1109-1112. (8) Conte, P.; Piccolo, A. Proceedings of 10th Meeting of the International Humic Substances Society, Toulouse, France; 2000; pp 1249-1250. (9) Piccolo, A.; Conte, P.; Cozzolino, A. Eur. J. Soil Sci. 1999, 50, 687-692. (10) Chen, A.; Wu, D.; Johnson, C. S., Jr. J. Am. Chem. Soc. 1995, 117, 7965-7970. (11) Wu, D.; Chen, A.; Johnson, C. S., Jr. J. Magn. Reson., Ser. A 1995, 115, 260-264. (12) Johnson, C. S., Jr. Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 203-256. (13) Lead, J. R.; Wilkinson, K. J.; Balnois, E.; Cutak, B. J.; Larive, C. K.; Assemi, S.; Beckett, R. Environ. Sci. Technol. 2000, 34, 35083513. (14) Morris, K. F.; Cutak, B. J.; Dixon, A. M.; Larive, C. K. Anal. Chem. 1999, 71, 5315-5321. (15) Little, D. J. Occurrence and Characteristics of Podzols Under Oak Woodland in Ireland. Ph.D. Thesis, The National University of Ireland, 1994. (16) Simpson, A. J.; Watt, B. E.; Graham, C. L.; Hayes, M. H. B. Humic Substances in Soils, Peats, and Waters; Special Publication of the Royal Society of Chemistry 172; Royal Society of Chemistry: Cambridge, 1997; pp 73-82. (17) Provencher, S. W. Comput. Phys. Commun. 1982. 23, 213-217. (18) Provencher S. W. Comput. Phys. Commun. 1982. 23, 229-242. (19) Simpson, A. J. Structural Interpretations of Humic Substances Isolated from Podzols under Varying Vegetation. Ph.D. Thesis, The University of Birmingham, England 1999. (20) Simpson, A. J.; Boersma, R. E.; Kingery, W. L.; Hicks, R. P.; Hayes, M. H. B. Humic Substances in Soils, Peats, and Waters; Special Publication of the Royal Society of Chemistry 172; Royal Society of Chemistry: Cambridge, 1997; pp 46-62. (21) Albery, W. J.; Greenwood, A. R.; Kibble, R. F. Faraday Discuss. Chem. Soc. 1967, 63, 360-368. (22) Kingery, W. L.; Simpson, A. J.; Hayes, M. H. B.; Locke, M. A.; Hicks, R. P. Soil Sci. 2000, 165, 483-494. (23) Wershaw R. L. Soil Sci. 1999, 164, 803-813. (24) Sano, Y.; Yanamoto. S. J. Chem. Eng. Jpn. 1993, 26 (6), 633-636. (25) Gosting, L. J. Adv. Protein Chem. 1956, 11, 429-553. (26) Longsworth, L. G. J. Am. Chem. Soc. 1953, 75, 5705-5709. (27) Oelkers E. H. Geochem. Cosmochim. Acta 1991, 55, 3515-3529. (28) Schulten, H. R.; Schnitzer, M. Naturwissenschaften 1993, 80, 29-30.
Received for review February 8, 2001. Revised manuscript received August 16, 2001. Accepted August 23, 2001. ES0106218
VOL. 35, NO. 22, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4425
