A Comprehensive Liquid-State Heteronuclear and Multidimensional

Jul 29, 2003 - Calgary, Alberta, Canada, T2N 1N4. The Boreal forest fulvic acid known as Laurentian fulvic acid (LFA) has been interrogated by state o...
1 downloads 0 Views 226KB Size
Environ. Sci. Technol. 2003, 37, 3935-3944

A Comprehensive Liquid-State Heteronuclear and Multidimensional NMR Study of Laurentian Fulvic Acid R O B E R T L . C O O K , * ,†,§ DEANE D. MCINTYRE,† COOPER H. LANGFORD,‡ AND HANS J. VOGEL† Departments of Biological Sciences and Chemistry, University of Calgary, 2500 University Drive N.W., Calgary, Alberta, Canada, T2N 1N4

The Boreal forest fulvic acid known as Laurentian fulvic acid (LFA) has been interrogated by state of the art heteronuclear and 2D high resolution NMR techniques. It is shown that one can obtain very highly resolved and informative spectra of a traditionally fractionated material. It was possible to observe a proton coupled system of up to seven bonds in the TOCSY spectrum and 329 peaks in the 1H,13C-HSQC spectrum. It is found that the majority of the nitrogen in this sample is in the form of ammonium cations. From the combination of inverse-gated decoupling, APT, and INEPT 13C spectra of LFA it can be concluded that while the aromatic moieties of LFA are highly unfunctionalized, the carbohydrate moieties are highly functionalized. Proton coupled networks are observed in the TOCSY spectrum between and within the aliphatic, functionalized aliphatic, and heteroatom substituted regions and, to a lesser extent, also between the amine/aromatic and heteroatom substituted regions. The HMBC spectrum confirms that both the aliphatic and heteroatom moieties are highly functionalized with carboxylic and alcoholic functional groups, while the aromatic moieties are very sparsely functionalized with phenolic and carboxylic functionalities. Additionally, specific model molecular structures have been identified which are consistent with experimental evidence and are in full agreement with our previously derived meso-model based on solid-state 13C NMR data. Finally, some of the shortcomings of 2D liquidstate NMR for the characterization of humic materials are addressed.

Introduction Simply put, humic materials are complex heterogeneous polydisperse mixtures whose properties are echoed in their structural diversity as well as their state of aggregation, conformation, and surface charge distribution (1). Thus, unlike biochemical systems, there is no homogeneity in humic materials based on which one can separate, isolate, and purify individual components, nor would this approach * Corresponding author phone: (225)578-2980; fax: 225-578-3458; e-mail: [email protected]. † Department of Biological Sciences. ‡ Department of Chemistry. § Current address: Department of Chemistry, Louisiana State University and Southern University at Baton Rouge, 636 (or 232) Choppin Hall, Baton Rouge, LA, 70803. 10.1021/es026196f CCC: $25.00 Published on Web 07/29/2003

 2003 American Chemical Society

be highly beneficial as it eliminates the properties which emerge only via interactions. However, we wish to be able to derive structural models for humic materials from which to build structure/function relationships to aid our understanding of humic chemistry (2-4). Thus, spectroscopic methods which are capable of monitoring several variables simultaneously in intact samples will be the most successful in characterizing humic materials. NMR is one of the most promising methods as it can provide high-resolution data sets, quantitative functional group analysis, information on covalent and geometrical structure, size, and dynamics, and both liquid and solidstate data can be obtained without perturbing the sample. The majority of NMR studies reported for humic materials to date have utilized solid-state 13C NMR via the CP-MAS technique (5, 6), as it is far more rapid than its liquid-state counterpart. The Ramp-CP-MAS technique overcomes most of the shortcomings (5-10) of the CP-MAS technique when applied to humic materials and has been shown to yield quantitative, structural, and metal binding results (7, 1113). However, traditional 1D 13C NMR spectra obtained via liquid-state or solid-state NMR have been plagued by a lack of resolution due to peak overlap. This low resolution can be overcome by using multidimensional methods which minimize peak overlap by dispersing the data into a second, third, or even fourth dimension. Although multidimensional NMR has long been used, particularly in the field of structural biology, it was not until 1989 that it was first applied to humic materials by Buddrus et al. (14). This pioneering work showed the great promise of 2D NMR in the study of humic materials; however, only recently has there been a concerted effort to study humic materials via multidimensional NMR (15-32). To date the use of 2D solid-state techniques in the study of humic materials has yielded limited data due to broad peaks (12, 25). On the other hand, solution-state multidimensional NMR is beginning to emerge as a highly powerful tool in the study of humic materials (15-24, 26-32). A large number of these studies have used only proton COSY, TOCSY, NOESY, and/ or DOSY spectra (18, 19, 23, 27, 28, 31), while others used a combination of COSY and/or TOCSY and HSQC and/or HMQC spectra (15, 20, 22, 24, 30, 32) to obtain both proton and protonated carbon information. However, none of these studies characterized all important functional groups of humic materials. In fact, only one other study could be found in the literature which involved a full NMR characterization of a humic material by 2D liquid-state techniques, but the sample used was “more highly fractionated and simpler in its nature than is usual” (26). All humic fractions studied to date appear to have been from the Ah horizon or of aquatic origin. As such they were simpler than the traditionally studied Bh horizon material. The one exception is the abovementioned highly simplified and fractionated sample discussed above, which appears to have been isolated from either an Ah or Bh horizon (26, 29). This report wishes to address this void by using both 1D and 2D NMR to fully interrogate a commercially available traditional Bh horizon fulvic acid, known as Laurentian fulvic acid (LFA) which has been the center of ongoing and extensive studies spanning more than a decade (7-11, 33-54).

Experimental Section Materials. All NMR tubes used in this study were Norell 508 tubes, while D2O (99.9%) was purchased from Cambridge Isotopes Laboratories Inc. (Andover, MA). DMSO was purchased from Aldrich. VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3935

TABLE 1. Parameters for the Collection of the 1D 1H Data sample

number of scans

delay (/s)

sweep width (/Hz)

sample A sample B sample C

128 64 256

3.0 2.6 2.6

6009.62 8012.82 8012.82

TABLE 2. Parameters for the Acquisition of the 1D 14N, 31P, and 13C Data experiment

number of scans

delay (/s)

sweep width (/Hz)

1D 14N 1D 31P 1D 13C inverse gated 1D 13C APT 1D 13C INEPT

400000 3361 98000 134000 50000

0.19 2.5 2.5 1.8 2.5

23980 14970 31250 25062 31250

This study exploits LFA, a fulvic acid extracted from the Bh horizon of a boreal forest podzol from an area controlled by Laval University (Quebec, Canada). It was prepared and purified as previously described (55, 56). LFA has been extensively characterized, including elemental analysis, acidbase and metal titration curves, emission fluorescence, synchronous fluorescence, magnetic circular dichroism, and by 1D and 2D solid-state Ramp-CP-MAS 13C NMR (7, 11-13, 23-40). Sample Preparation. Three different LFA samples were prepared. Sample A was prepared by dissolving 50 mg/mL of LFA in water (10% D2O). Sample B was prepared by dissolving 150 mg/mL of LFA in fresh (ampule was sealed until used to make this sample) deuterated DMSO. Finally, sample C was prepared by dissolving 150 mg/mL of LFA in “wet” deuterated DMSO (the bottle had been opened and closed, and hence exposed to the atmosphere, several times over the past years despite being stored in a desiccator). A small amount of TMS was added to sample B for referencing purposes (internally for the 2D work and externally for samples A and C in terms of the 1H data). The fully dissolved samples were then directly transferred to either a 10 mm NMR tube (sample A) or to a 5 mm NMR tube (samples A, B, and C). NMR Spectroscopy. All the 1D spectra, except for the 1H spectra, were acquired for sample A only on a Bruker Avance 400 MHz NMR spectrometer at room temperature using a 10 mm broad band probe. The 13C and 31P results were externally referenced to DSS and phosphoric acid, respectively, while the 14N result was internally referenced to the dissolved nitrogen gas (57). The 13C spectra were collected under the quantitative conditions established before (7). All 2D and 1D 1H spectra were collected on a Bruker Avance 500 MHz NMR spectrometer equipped with a 5 mm inverse geometry triple resonance (1H,13C, 15N) z-gradient cryoprobe and acquired at 300 K. All 2D experiments were collected on sample B and thus 1H and 13C data were directly referenced to the internal TMS, while the 15N data were indirectly referenced to the internal TMS (58). Water suppression was achieved using excitation sculpting (59). A description of all the experiments with the appropriate references has been previously published (60). The parameters used to collect the 1D data sets are summarized in Tables 1 and 2. For the 1D 13C inverse gated decoupled experiment a 45° excitation pulse was used. A JCH coupling of 145 Hz was used in the APT experiment. The TOCSY spectrum was acquired with 2048 and 512 data points in the F2 (directly detected) and the F1 (indirectly detected) dimensions, respectively, with sweep widths of 6009.615 Hz and a mixing time of 50 ms. The data were 3936

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

acquired in the phase sensitive mode using the TPPI method. The delay between scans was 1.6 s, and 8 scans were collected per slice. For the purpose of processing, these data were zero filled to form a 2K by 2K data matrix and multiplied by a shifted sine-bell curve in each dimension. The 1H,13C-HSQC, multiplicity sensitive 1H,13C-HSQC, and 1H,13C-HSQC-TOCSY spectra were collected and processed using the following conditions. Data sets were acquired with 2048 and 320 data points and sweep widths of 7507 and 25 000 Hz for the F2 and F1 dimension, respectively. A JCH coupling of 145 Hz was assumed, and an interscan delay of 1.6 s was used for all three experiments. All spectra were acquired in phase sensitive mode using echo/antiecho-TPPI gradient selection with decoupling during acquisition and trim pulses were used for the double INEPT transfer. For the 1H,13C-HSQCTOCSY experiment a TOCSY mixing time of 50 ms was used. For processing all data sets were zero filled to 2K (F2) and 1K (F1) and multiplied by a shifted sine-bell curve in each dimension. The 1H,13C-HMBC data were collected in the same manner as discussed above for the 1H,13C-HSQC experiment except 400 scans were acquired per slice with an interscan delay of 1 s. A 50 ms delay was used for the evolution of long-range couplings. Also, the data were not collected in phase sensitive mode, and no decoupling was used during acquisition. The 1H,15N-HSQC data were collected as discussed above for the 1H,13C-HSQC data, except a 2048 (F2) by 256 (F1) data matrix was collected with sweep width of 7002 and 10 136 Hz, respectively. This data set was zerofilled to form a 2K by 0.5K data set and multiplied by a shifted sine-bell function in both dimensions. An interscan delay of 1.3 s was used and 200 transients were collected per slice. A JHN coupling constant of 90 Hz was assumed. All data were collected and processed using Bruker XWINNMR software. Spectral regions were assigned from the data already presented in the literature of humic materials and organic compounds (5, 7, 20, 22, 61). Finally, the SDBS integrated spectral database system was used to identify the organic compounds presented below (62).

Results and Discussion 1D NMR Analysis. 1H NMR. Figure 1 shows three proton spectra of LFA, each obtained using a different solvent makeup (see Experimental Section). In all cases very sharp highly resolved peaks are observed overlaying much broader bands. The resolution of the sharp peaks is very high, especially when compared to the majority of the 1H liquidstate spectra of humic materials presented to date in the literature (18, 24, 63-65). The three most plausible explanations for this high resolution are as follows: (1) the fulvic acid used is more highly soluble than most others; (2) there is high regional mobility within macromolecules; and/or (3) the majority of the molecules are small. Solubility data and molecular weight measurements would indicate that LFA is not exceptional in regards to these when compared to other fulvic acids. Thus, it would appear that similar resolution is possible on other soluble humic fractions. This level of resolution is another piece of evidence that humic materials are large macromolecular assemblies rather than large macromolecules which tumble more slowly in solution compared to small molecules and broader lines are observed in the NMR spectrum. However, it should be noted that diffusion experiments indicate (data not shown) the presence of some broad lines. Thus, it appears that LFA is made up of both small and large molecular entities. This finding is in full agreement with previously published solid-state NMR results (13) and other recent liquid-state NMR analysis of humic materials (29). When LFA is dissolved in water the carboxylic acid protons are unobservable due to very rapid proton exchange, and thus no peaks for the carboxylic protons are observed.

FIGURE 1. 1H spectra for sample A (ammonium ion signals are shown in the insert), sample B, and sample C. However, if DMSO is used as the solvent, such exchange is expected to be less prominent by a factor of about 100, as seen for sample B, where one sees a prominent peak centered at ∼6.9 ppm. Sample C is the intermediate case. Turning our attention to the spectrum of sample A, three sharp peaks are noticeable in the aromatic/amide region. Based on their splitting pattern and the 14N as well as the 1H,15N-HSQC results discussed below, these three peaks can be assigned to ammonium ions. The insert in Figure 1 shows that the ammonium ions are in a minimum of three distinct sets of peaks. The reason for these three peaks is the isotope effect (1H and 2H [D]), see Supporting Information. The narrow upfield peaks are due to NH4+, the broader peaks (broadened by scalar relaxation) slightly downfield are due to NH3D+, and the furthest downfield and broadest peaks are due to NH2D2+. The origin of these ammonium ions is puzzling and should not be due to agricultural practices, since this sample comes from an uncultivated soil. There are, however, two possible sources. The first is that they are formed by the reduction of nitrate ions in the reducing environment of the Bh horizon and, thus, represent the steady-state concentration of ammonium ions in the Bh horizon of this soil. Second, ammonium ions may be liberated from amino acids in the cation exchange process used in the isolation and cleaning procedure used for LFA. Although these spectra could be further analyzed to gain deeper insight into these very complex and heterogeneous substances based on the chemical shifts of the peaks, it will be shown below that such an intensive analysis should be done using modern 2D NMR techniques. 13C NMR. Traditionally NMR analysis of humic materials meant obtaining a 1D 13C chemical shift spectrum via a simple inverse gated decoupled pulse sequence for a liquid-state spectrum or a CP-MAS solid-state spectrum. In this study we have obtained three 1D 13C spectra of LFA. These spectra are presented in Figure 2. Figure 2a shows a 13C spectrum obtained via the inverse gated decoupled pulse sequence, thus NOE effects are removed and the spectrum can be

interpreted both quantitatively and qualitatively (7). From this figure it can be seen that the carboxylic carbons (165185 ppm) are the most dominant species of carbon followed by the carbohydrate/aliphatic carbons (0-95 ppm) and then the aromatic carbons (105-145 ppm). In addition, defined ketonic (185-220 ppm), phenolic (145-165 ppm), and anomeric (95-105 ppm) carbons are also visible. This spectrum is very similar to a solid-state spectrum obtained and analyzed by this group via the Ramp-CP-MAS technique, except for the sharp peak at approximately 165 ppm (13). This 165 ppm peak can be found in the 13C NMR spectra of other humic materials in the literature (20, 66-68) and is only present in liquid spectra. However, neither a clear assignment of this peak nor a discussion of it could be found in the literature. Figure 2b shows an APT spectrum of LFA phased so that carbons with an even number of protons attached to them are positively phased and carbons with an odd number of protons attached are negatively phased. From these data it can be seen that the majority of aromatic carbons are protonated. On the other hand, it appears that the carbohydrate/aliphatic carbons (0-95 ppm) carbons are functionalized, with the single heteroatom substituted aliphatic (40-85 ppm) carbons being highly functionalized. The INEPT spectrum shown in Figure 2c fully supports these findings and even brings the protonation of the aromatic carbons into better focus. However, the results from the APT spectrum provide stronger evidence that the carbohydrate and carbohydrate-like carbons are highly functionalized. The positive phasing of the 165 ppm peak in the APT spectrum and its absence in the INEPT spectrum strongly suggest that the entity which gives rise to this peak has no protons attached to it and could be carbonate. The exact nature of this peak is being actively investigated in our laboratory. 14N NMR. Nitrogen only constitutes 0.83% of this sample by mass and the natural abundance of 15N is 0.37%. Thus, 14N NMR was run on the sample due to the higher receptivity of 14N (15N NMR was also run but no signal was observed). VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3937

FIGURE 2. (a) Inverse gated decoupled 13C spectrum of LFA, (b) 13C APT spectrum of LFA, and (c) 13C INEPT spectrum of LFA. The 14N spectrum of LFA is shown in Figure 3a. From this spectrum two N containing species can be seen. The low field species at about -71.5 ppm can be assigned to dissolved nitrogen (52), while the species that lead to a quintet of peaks can be assigned to ammonium cations. It also appears, from the narrowness of the observed peaks (14N is a quadrupolar nucleus), that at least some of the ammonium cations are highly mobile. It is interesting to note that no amino acids were detected in LFA via 1D 14N or 15N NMR analyses, although others have found amino acids in humic biogeopolymers via the 15N NMR experiment (24). This is most probably due to the fact that the LFA sample used in this work was obtained from a Bh horizon (and is approximately 800 years old as determined by 14C dating), while other samples have been isolated from the Ah horizon or a river and, thus, are probably much younger. This subject will be addressed further in the 2D section below. 31P NMR. Figure 3b shows the 31P NMR spectrum of LFA. Based on this spectrum it can be seen that the majority of phosphorus is in the phosphate form. Though informative, the 1D data are less than optimal from the point of view of detailed structural characterization, which is so common in the study of biochemical macromolecules. To enrich the characterization one must resort to 2D NMR techniques which disperse peaks into two dimensions and help resolve overlapping peaks. 2D NMR. Dispersing data in two dimensions offers two advantages. First, it helps to resolve overlapping peaks, which, as the 1D data reveal, is much needed in the study of NOM. Second, an even more powerful advantage of multidimen3938

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

sional spectroscopic techniques is that the data are dispersed into two dimensions in a very controlled and well understood manner. Homonuclear Shift Correlation. TOCSY. Figure 4a shows the full 1H TOCSY spectrum of LFA, while Figures 4b,c show enlargements of the aliphatic/carbohydrate and amine/ aromatic regions, respectively. The number of peaks resolvable at the field strength used in this study is striking. Two hundred thirty-nine distinctly mirrored cross-peaks were found. This pulse sequence gives rise to cross-peaks for all protons within a coupled network of a chosen proton. Crosspeaks relayed between 1 and 6 bonds were found, thus indicating systems with up to 7 coupled protons. This indicates that LFA either has small mobile molecules in its makeup or parts of the large macromolecules are highly mobile. This viewpoint is also supported by the narrow lines observed in the 1D 1H NMR spectrum, vide supra. However, it is important to note that larger coupled networks may be present but remain unobserved due to rapid relaxation, i.e., short T2s: as are found for less mobile sections of macromolecular complexes. From Figure 4b it can be seen that there are many coupled networks within the aliphatic (0-2 ppm), functionalized aliphatic (2-3 ppm), heteroatom substituted (3-5 ppm), and the aromatic/amine regions (6-9 ppm). Also, it can be seen that there is significant coupling between the aliphatic and functionalized aliphatic regions as well as the functionalized aliphatic and heteroatom substituted regions. Specific partial structures can be assigned, such as intraaliphatic chains (2.30.5 ppm [F2]f2.3-0.5 ppm [F1], presented as 2.3-0.5f2.30.5), deoxy sugars, ethers, and esters (4.4-3.0f1.4-1.0), functionalized aliphatic chains with a single heteroatom (4.5-

FIGURE 3. (a) A 14N spectrum of LFA and (b) a 31P spectrum of LFA.

FIGURE 4. (a) A TOCSY spectrum of LFA, (b) the aliphatic/carbohydrate region, and (c) the amine/aromatic region. 3.2f3.0-1.4), intracarbohydrates with anomeric carbons (4.4-1.0f4.8-4.0), methylated alkenes (2.0-3.0f4.0-5.0), and alcohols (1.0-2.0f3.0-4.0) assignments can be made based on the observed cross-peaks. If we now look at Figure 4c, it can be seen that there is coupling in the aromatic/

amine region as well as, to a lesser extent, between the aromatic/amine and heteroatom substituted regions. Due to the observation of amine and amide signals in the much less sensitive 15N HSQC data discussed above the cross-peaks in this region could arise from either amino acids and/or VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3939

FIGURE 5. A 1H,15N-HSQC spectrum of LFA.

FIGURE 6. A 1H,13C-HSQC spectrum of LFA.

peptide moieties. Finally, in Figure 4a, one can also see coupling between the amine/aromatic and the heteroatom substituted regions. These findings are consistent with other humic materials examined by 2D TOCSY NMR, but the exact pattern is distinct (19, 22, 24). Heteronuclear Shift Correlation. As shown above, the TOCSY experiment can be very useful due to its ability to show the covalent bonding networks via proton coupling, though its usefulness is only exploitable if the proton chemical shifts have been assigned. It would be a reasonable approach to use proton chemical shift data tables if we had a good knowledge of the compounds that are in the system, knew their chemical shifts as well as the chemistry which took place, and had good intuition into the situation. A better approach for NOM would be to use heteronuclear data to help in assigning proton data sets. 1H,15N-HSQC. While a 1D 15N spectrum could not be obtained on this sample, after 5 days of collection it was possible to obtain a 1H,15N-HSQC spectrum (Figure 5) due to the sensitivity advantage of indirect detection and cryogenic probe technology. By far, the most intense peak at about 7.15; 25.1 ppm (1H chemical shift; X nucleus chemical shift) can be assigned to ammonium cations based on chemical shift, the 14N data as well as the TOCSY data (vide infra). It appears as a doublet because nitrogen decoupling was not complete due to cryoprobe limitations. Amine (centered at ∼8.23 ppm; 41.0 ppm) and amide (centered at ∼8.12 ppm; 119.6 ppm) species also emerge from Figure 5. The amide species are found in the same chemical shift range as amino acids that are linked via an amide bond. The peak centered at around 11.2; 137 ppm can be assigned to an indole type nitrogen as found in the aromatic ring system of the amino acid tryptophan. It is of great interest that amino acid-like moieties should exist in an old humic sample at all, since one would expect nature to have developed a method of mining this nitrogen source. One possible explanation is that these amino acidlike moieties have been incorporated into the humic material’s structure. However, this seems unlikely as one would expect at least one type of bacterium to have evolved to mine this natural source of nitrogen. Another explanation is that these amino acid moieties are not of the 20 fundamental amino acids, but are different and, thus, of no interest to bacteria. It could be argued from the weakness of the signals that these amino acids may arise from a steady-state

concentration of amino acids that are found in Bh horizon of the soil at any point in time. Finally, these amino acids might arise from bacteria killed and lysed in the isolation of this humic material from the native soil matrix.

3940

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

1H,13C-HSQC. The 1H,13C-HSQC spectrum of LFA is shown in Figure 6. The richness and resolution of the data is striking, especially when compared to the 1D NMR information. Three hundred twenty-nine peaks were identified in this spectrum. A detailed discussion of each peak is well beyond the scope of this paper.

Some general trends are observed. The aliphatic region (0.4-3.4 ppm 1H chemical shift range and 5-40 ppm 13C chemical shift range, which will be presented as (0.4-3.4f540)), and the single heteroatom substituted aliphatic regions (2.6-5.3f40-85) show a series of well-defined peaks overlaying a continuum. It can be seen that the aliphatic carbons are separated into two groups, i.e.: high field and low field in both the 1H and 13C dimensions. However, the anomeric (4.2-5.6f85-105) region is very well resolved and uncrowded. In the aromatic (6.0-9.0f105-145) region one can see a complex envelope of peaks which overlap the proton shifts of the amine and amide protons discussed above. This is one of the most complex HSQC spectrum of a humic material published to date and is consistent with the view that Bh humic materials are more complex than their Ah horizon counterparts. A HSQC variant called a HSQC-TOCSY experiment due to the fact that it exploits both techniques and provides the combined information of both techniques has shown great promise in the study of humic materials (26, 32). This is either a 2D or a 3D experiment. We have used the 2D version, and the resulting spectrum is presented in Figure 7. In Figure 7b it can be clearly seen that the anomeric carbons are coupled to the single heteroatom substituted aliphatic region. Also, it is evident from Figure 7c that the single heteroatom substituted aliphatic region is coupled to both groups of the aliphatic region. On the other hand, no coupling is observed between the aromatic or amino acid region to the single heteroatom substituted aliphatic region. It should be noted, however, that this experiment requires the molecules it detects to have a rather long T2 (small molecules) due to the time needed to execute the pulse sequence on the spin system. Hence, we suggest caution when applying this method to humic materials.

FIGURE 7. (a) A 1H,13C-HSQC-TOCSY spectrum of LFA, (b) the anomeric carbon region, and (c) the carbohydrate region. 1H,13C-HMBC. One of the central issues in the study of humic materials is the functional group distribution, especially that of the carboxylic and alcohol functional groups. Such groups determine much of the reactivity as well as the tertiary and quaternary structure of humic materials. Although the 2D experiments discussed up to this point provide insight into the moieties which contain the amine and amide groups, they do not give insight into carboxylic or alcoholic functionalities. The HMBC experiment is capable of doing this and allows one to take advantage of the higher resolution of liquid-state NMR, the higher data dispersion of 2D NMR, and sensitivity of inverse detection, while still being able to monitor the connectivity of functional groups to the coupled protonated carbon. A HMBC spectrum of LFA is shown in Figure 8. It is very apparent that the majority of the functionality is associated with the aliphatic and single heteroatom substituted aliphatic regions. Some aromatic moieties are functionalized, mainly with phenolic functionalities. The HMBC experiment also allows one to connect ketonic carbons with the single heteroatom substituted aliphatic and aliphatic regions. It is interesting to note that there are two distinctly different types of ketonic carbons. Based on the 13C chemical shifts, the lower field peaks can be assigned to cyclic ketones, while the higher field peaks can be assigned to acyclic ketones. Further upfield one notices a spread of the carboxylic groups connected to different moieties, but the 13C chemical shift range is narrow. Thus, the broad carboxylic band observed in the 1D 13C NMR spectrum (Figure 2a) is due to a complex envelope of overlapping carboxylic peaks rather than few very broad carboxylic peaks. It is also interesting to note that the phenolic carbons do not overlap with their carboxylic counterparts in the 13C dimension, and thus, for this humic material, carboxylic content can be quantitatively analyzed

by 13C NMR. The alcohol-like carbons also show a great diversity in the moieties with which they are associated. Unlike the carboxylic carbons, they show a large 13C chemical shift range. A number of protonated carbons are coupled (via the protons) to more than one functional group. Thus, some functional groups are grouped together in the aliphatic and carbohydrate moieties. These moieties are likely to be strong metal chelators as found via solid-state 13C NMR (11, 12). This spectrum is also much more complex than the other HMBC spectrum published to date (26) and once again shows the complexity of a Bh horizon fulvic acid. It should be pointed out that the HMBC experiment is based on choosing a single coupling constant to measure the connectivities (as for the HSQC experiment). Consequently, signals arising from coupling constants closer to the selected coupling constant are stronger, while some correlations are not observed at all. This could be overcome via the ACCORDION principle by which a range of 2J and 3J coupling constants would be sampled, thus more correlation signals would emerge compared to the experiment used in this study (due to the fixed polarization delay) (69). As can be seen in Figure 8b, the HMBC spectrum is very crowded and difficult to interpret. Thus, the ACCORD-HMBC experiment may not be well suited for the study of NOM. A better approach would be to run a series of HMBC experiments which sample different coupling constants. Combining 2D Experiments. Though each 2D method delivers extremely important information, an even more powerful approach in the study of NOM is to combine the results of the various 2D experiments (26). In this study HMBC data are used as a starting point, as the authors believe this is the most appropriate approach for the study of complex unknown mixtures such as humic VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3941

FIGURE 8. (a) A 1H,13C-HMBC spectrum of LFA and (b) the aliphatic/carbohydrate region. materials. The data are then linked to the TOCSY data which, in turn, may then be connected back to the HMBC data. This cycle is followed until a dead end, i.e., the other end of the molecule, is reached. It should be stressed that the connections so established are covalent bonds. The 1H,13C-HSQC data serve as verifiers as well as provide the 13C chemical shifts for the coupled protons found in the TOCSY experiment. The final step is to search a 1H/13C NMR database to find compounds with the same NMR signature as the covalent networks revealed by the 2D NMR experiments, after which a series of entities is abstracted based upon the experimental data and knowledge of humic material chemistry. Being able to trace out covalent bond systems allows one to determine specific structures rather than simply propose generic classes. Thus, instead of meso-structural (11-13), microstructural models of humic materials can be determined. Using this approach, entities consistent with phloridzin, hesperidin, 2-hydroxy-2-methylsuccinic acid, 2-hydroxy-3-methylbutyric acid, 3-(p-methoxybenzoyl)propionic acid, 3-butene-1, 2,3tricarboxylic acid, menthoxyacetic acid, and/or benzylmalonic acid have been found in LFA. It should be noted that these entities ought to be viewed as models. The majority of the above molecules are rather small and highly functionalized, thus the HMBC experiment is used only as the starting point in their determination. The small size of these molecules is no surprise since they give rise to sharp and well resolved signals which aid in their detection. Also, these molecules have long T2 values and their signals do not decay in time-consuming 2D pulse sequences. Thus, it should be noted that 2D NMR methods strongly bias the data in favor of small molecules over large molecules. In other words, the lack of detectable peaks for large molecules does not prove their absence. Also, one notices that two of the molecules found are of glycoside type. The data presented in this paper support the mesostructural model and the presence of carbohydrate base strong metal binding moieties previously proposed from 3942

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

solid-state 13C NMR studies (11-13). However, the resolution obtainable via liquid-state NMR offers a quantum leap especially when one can disperse the data into 2D compared to 1D solid-state NMR. Cryogenic probe technology has been indispensable in this study, yielding the information rich spectra in a practical time. In future studies the signal overlap will be diminished with the introduction of ultrahigh field NMR instruments (18.5 and 21 T magnets).

Acknowledgments We acknowledge the Canadian Institute of Health Research and the Natural Science and Engineering Research Council of Canada for financial support. Hans J. Vogel is a scientist of the Alberta Heritage Foundation for Medical Research. The 500 MHz NMR spectrometer was recently upgraded through a grant from the Canada Foundation for Innovation.

Nomenclature 1D

one-dimensional

2D

two-dimensional

APT

attached proton test

COSY

correlation spectroscopy

DMSO

dimethyl sulfoxide

DOSY

diffusion ordered spectroscopy

DSS

2,2-dimethyl-2-silapentane-5-sulfonate

HETCOR

heteronuclear correlation

HMBC

heteronuclear multiple bond correlation

HMQC

heteronulear multiple quantum coherence

HSQC

heteronuclear single quantum coherence

IHSS

International Humic Substance Society

INEPT

insensitive nuclei enhanced by polarization transfer

LFA

Laurentian fulvic acid

NOE

nuclear Overhauser effect

NOESY

nuclear overhauser effect spectroscopy

NOM

natural organic matter

rf

radio frequency

TOCSY

total correlation spectroscopy

TMS

tetramethylsilane

TPPI

time proportional phase increment

Supporting Information Available Complete peak assignments, in tabular form, for all 2D data sets presented in this paper and also a 1H NMR spectrum at low pH which shows the isotope effect. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Buffle, J. Complexation Reactions in Aquatic Systems, An Analytical Approach; Ellis Harwood Ltd.: Chester, UK, 1988. (2) Humic Substances in the Global Environment and Impacts on Human Health; Senesi, N., Miano, T. M., Eds.; Elseveir: Amsterdam, 1994; references therein. (3) Humic Substances II: In Search of Structure; Hayes, M. H. B., MacCarthy, P., Malcolm, R., Swift, R. S., Eds.; John Wiley & Sons: Toronto, 1989; reference therein. (4) Stevenson, F. J. Humus Chemistry, Genisis, Composition, Reactions; 2nd ed.; John Wiley & Sons: Toronto, 1994. (5) Wilson, M. A. NMR Techniques and Applications in Geopchemistry and Soil Chemistry; Pergamon Press: Oxford, 1987; references therein. (6) Preston, C. M. Soil Sci. 1996, 161, 144-166 (7) Cook, R. L.; Langford, C. H.; Yamdagni, R.; Preston, C. M. Anal. Chem. 1996, 68, 3979-3986. (8) Frund, R.; Lundmann, H.-D. Sci. Total Environ. 1989, 81/82, 157-168. (9) Kinchesh, P.; Powlson, D. S.; Randall, E. W. Eur. J. Soil Sci. 1995, 46, 125-138. (10) NMR of Humic Substances and Coals: Techniques, problems, and solutions; Wershaw, R. L., Mikita, M. A., Eds.; Chesea, MI, 1987. (11) Cook, R. L.; Langford, C. H. In Understanding Humic Substances; Advanced Methods, Properties and Applications; Ghabbour, E. A., Davies, G., Eds.; Royal Society of Chemistry: Cambridge, UK, 1999; pp 31-48. (12) Cook, R. L.; Langford, C. H. Polymer News 1999, 24, 6-15. (13) Cook, R. L.; Langford, C. H. Environ. Sci. Technol. 1998, 32, 717-725. (14) Buddrus, J.; Burba, P.; Lambert, J.; Herzog, H. Anal. Chem. 1989, 61, 628-631. (15) Haiber, S.; Burba, P.; Herzog, H.; Lambert, J. Fresenius J. Anal. Chem. 1999, 364, 215-218. (16) Haiber, S.; Herzog, H.; Burba, P.; Gosciniak, B.; Lambert, J. Fresenius J. Anal. Chem. 2001, 369, 457-460. (17) Haiber, S.; Herzog, H.; Burba, P.; Gosciniak, B.; Lambert, J. Environ. Sci. Technol. 2001, 35, 4289-4294. (18) Chien, Y.-Y.; Bleam, W. F. Environ. Sci. Technol. 1998, 32, 36533658. (19) Wang, L.; Xi, M.; Yang, Y. Bopuxue Zazhi (Chin. J. Magn. Reson.) 1998, 15, 411-420. (20) Schmitt-Kopplin, P.; Hertkorn, N.; Schulten, H.-R.; Kettrup, A. Environ. Sci. Technol. 1998, 32, 2541. (21) Hertkorn, N.; Claus, H.; Schmitt-Kopplin, Ph.; Perdue, E. M.; Filip Z. Environ. Sci. Technol. 2002, 36, 4334-4345. (22) Hertkorn, N.; Permin, A.; Perminova, I.; Kovalevskii, D.; Yudov, M.; Petrosyan, V.; Kettrup, A. J. Environ. Qual. 2002, 31, 375387. (23) Morris, K. F.; Cutak, B. J.; Dixon, A. M.; Larive, C. K. Anal. Chem. 1999, 71, 5315-5321. (24) Fan, T. W.-M.; Higashi, R.; Lane, A. N. Environ. Sci. Technol. 2000, 34, 1636-1646. (25) Mao, J.-D.; Xing, B.; Schmidt-Rohr, K. Environ. Sci. Technol. 2001, 35, 1928-1934.

(26) Simpson, A. J.; Burdon, C. L.; Hayes, M. H. B.; Spencer, N.; Kingery. W. L. Eur. J. Soil Sci. 2001, 52, 495-509. (27) Simpson, A. J.; Kingery, W. L.; Shaw, D. R.; Spraul, M.; Humpfer, E.; Dvortsak, P. Environ. Sci. Technol. 2001, 35, 33213325. (28) Simpson, A. J.; Kingery, W. L.; Spraul, M.; Humpfer, E.; Dvortsak, P.; Kersserbaum, R. Environ. Sci. Technol. 2001, 35, 44214425. (29) Simpson, A. J.; Kingery, W. L.; Hayes, M. B. H.; Spraul, M.; Humpfer, E.; Dvortsak, P.; Kersserbaum, R.; Godejohann, M.; Hofmann, M. Naturwissenschaften 2002, 89, 84-88. (30) Simpson, A. J.; Salloum, M. J.; Kingery, W. L.; Hatcher, P. G. J. Environ. Qual. 2002, 31, 388-392. (31) Simpson, A. J. Magn. Reson. Chem. 2002, 40, S72-S82. (32) Simpson, A. J.; Kingery, W. L.; Hatcher, P. G. Environ. Sci. Technol. 2003, 37, 337-342. (33) Sojo, L. E.; Gamble, D. S.; Langford, C. H.; Zienius, R. H. J. Environ. Sci. Health 1989, B24, 619-646. (34) Wang, Z. D. Ph.D. Thesis, Concordia University, Montreal, 1989. (35) Wang, Z. D.; Pant, B. C.; Langford, C. H. Anal. Chim. Acta 1990, 232, 43-49. (36) Wang, Z. D.; Gamble, D. S.; Langford, C. H. Anal. Chim. Acta 1990, 232, 181-188. (37) Wang, Z.; Gamble, D. S.; Langford, C. H. Anal. Chim. Acta 1991, 244, 135-143. (38) Wang, Z.; Gamble, D. S.; Langford, C. H. Environ. Sci. Technol. 1992, 26, 560-565. (39) Li, J.; Gamble, D. S.; Pant, B. C.; Langford, C. H. Environ. Technol. 1992, 13, 739-749. (40) Bruccoleri, A.; Pant, B. C.; Sharma, D. K.; Langford, C. H. Environ. Sci. Technol. 1993, 27, 889-894. (41) Cook, R. L.; Langford, C. H. Anal. Chem. 1995, 67, 174180. (42) Mathuthu, A. S.; Ephraim, J. H. Talanta 1993, 40, 521526. (43) Mathuthu, A. S.; Marinsky, J. A.; Ephraim, J. H. Talanta 1995, 42, 441-447. (44) Mathuthu, A. S.; Ephraim, J. H. Talanta 1995, 42, 18031810. (45) Glaus, M. A.; Hummel, W.; Van Loon, L. R. Anal. Chim. Acta 1995, 303, 321-331. (46) Glaus, M. A.; Hummel, W.; Van Loon, L. R. Appl. Geochem. 2000 15, 953-973. (47) Karanfil, T.; Kilduff, J. E.; Schlautman, M. A.; Weber, W. J., Jr. Environ. Sci. Technol. 1996, 30, 2187-2194. (48) Karanfil, T.; Kilduff, J. E.; Schlautman, M. A.; Weber, W. J., Jr. Environ. Sci. Technol. 1996, 30, 2195-2201. (49) Pinheiro, J. P.; Mota, A. M.; Simoes Goncalves, M. L. S.; van Leeuwen, H. P. Colloids Surfaces, A: Physicochem. Eng. Aspects 1998, 137, 165-170. (50) Vermeer, A. W. P.; Koopal, L. K. Langmuir 1998, 14, 42104216. (51) Thorn, K. A.; Mikita, M. A. Soil Sci. Soc. Am. J. 2000, 64, 568582. (52) Murimboh, J.; Lam, M. T.; Hassan, N. M.; Chakrabarti, C. L. Anal. Chim. Acta 2000, 423, 115-126. (53) Sekaly, A. L. R.; Murimboh, J.; Hassan, N. M.; Mandal, R.; Younes, M. E. B.; Chakrabarti, C. L.; Back, M. H.; Gregoire, D. C. Environ. Sci. Technol. 2003, 37, 68-74. (54) da Silva, E.; Joaquim C. G.; Oliveira, Cesar J. S. Portugaliae Electrochim. Acta 2001, 19, 85-97. (55) Griffith, S. M.; Schnitzer, M. Soil Sci. 1975, 120, 126-131. (56) Schnitzer, M.; Skinner, S. I. M. Soil Sci. 1968, 105, 392396. (57) McIntyre, D. D.; Apblett, A. W.; Lundberg, U. P.; Schmidt, K. J.; Vogel, H. J. J. Magn. Reson. 1989, 83, 377-382 (58) Live, D. H.; Davis, D. G.; Agosta, W. C.; Cowburn, D. J. Am. Chem. Soc. 1984, 106, 1939-1943. (59) Callihan, D.; West, J.; Kumar, S.; Schweitzer, B. I.; Logan, T. L. J. Magn. Reson. 1996, 112, 82-85. (60) Braun, S.; Kalinowski, H.-O.; Berger, S. 150 and More Basic NMR Experiments: A Practical Course; Wiley-VCH: Toronto, 1998. (61) Heese, M.; Meier, H.; Zeeh, B. Spectroscopic Methods in Organic Chemistry; Thieme: New York, 1997. (62) http://www.aist.go.jp/RIODB/SDBS/menu-e.html (accessed May 2003) (63) Stepanov, A. A.; Zharkova, L. V.; Stepanova, E. A. Eurasian Soil Sci. 1997, 30, 142-145. (64) Herzog, H.; Haiber, S.; Burba, P.; Buddrus, J. Fresnius J. Anal. Chem. 1997, 359, 167-170. VOL. 37, NO. 17, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3943

(65) Piccolo, A.; Conte, P.; Trivellone, E.; van Lagen, B.; Buurman, P. Environ. Sci. Technol. 2002, 36, 76-84. (66) Wais, A.; Burauel, P.; de Graaf, A. A.; Haider, K.; Fuhr, F. J. Environ. Sci. Health 1996, B31, 1-24. (67) Conte, P.; Piccolo, A.; van Lagen, B.; Buurman, P.; de Jager, P. A. Geoderma 1997, 80, 339-352. (68) Kingery, W. L.; Simpson, A. J.; Hayes, M. H. B.; Boersma, R. E.; Locke, M. A.; Hicks, R. P. Soil. Sci. 2000, 165, 483-494.

3944

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 17, 2003

(69) Wagner, R.; Berger, S. Magn. Reson. Chem. 1998, 36, S44S46.

Received for review September 30, 2002. Revised manuscript received May 21, 2003. Accepted June 2, 2003. ES026196F