Environ. Sci. Technol. 2004, 38, 3387-3395
Identification of Black Carbon Derived Structures in a Volcanic Ash Soil Humic Acid by Fourier Transform Ion Cyclotron Resonance Mass Spectrometry ROBERT W. KRAMER, ELIZABETH B. KUJAWINSKI,† AND PATRICK G. HATCHER* Department of Chemistry, The Ohio State University, Columbus, Ohio 43210
Electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), coupled with cross-polarization magic angle spinning 13C nuclear magnetic resonance (NMR) spectroscopy and Kendrick mass defect analysis, was used to study the molecular composition of an aromatic carbon-rich humic acid extracted from a dark black soil from Iwata, Japan. Black carbon, produced by the incomplete combustion of fossil fuels and organic matter, has been suggested as a major component of humic acids having intense peaks in the aromatic and carboxyl regions of the 13C NMR spectrum. Taking advantage of the high resolving power of FT-ICR MS to make precise formula assignments, three different types of highly carboxylated polycyclic aromatic compounds were identified in the sample: linearly fused aromatic structures, aromatic structures linked by carbon-carbon single bonds, and highly condensed aromatic structures. These carboxylated aromatic structures have a low mass defect in their mass spectra due to their abundance of oxygen and deficiency of hydrogen. This mass defect is observed in the vast majority of peaks present in the entire mass spectrum, differentiating them from structures that are hydrogenrich (e.g., fatty acids, proteins, carbohydrates). Thus, we conclude that the bulk of the sample analyzed is comprised of these heavily carboxylated, hydrogen-deficient, condensed aromatic structures, features believed to be characteristic of black carbon-like material.
Introduction Black carbon, which is generally regarded as the product of inefficient burning of fuels or natural vegetation, plays a potentially important role in many biogeochemical cycles and processes, yet there are few universally accepted methods, among many proposed, for identifying and quantifying black carbon in the environment. Black carbon has been identified in many different environments such as soils and sediments (1), atmospheric particulates (2), and natural waters (3). In some soils black carbon is thought to constitute up to 80% of the total organic carbon (4). Black carbon may * Corresponding author telephone: (614)688-8799; fax: (614)6884353; e-mail:
[email protected]. † Present address: Dept. of Environmental Science, Barnard College, New York, NY 10027. 10.1021/es030124m CCC: $27.50 Published on Web 05/11/2004
2004 American Chemical Society
play an important role in the sorption/retention of pollutants, affect the earth’s radiative heat balance when associated with airborne particulates, represent an important carbon sink in global biogeochemical cycles, and provide a tracer for earth’s fire history (5). Recently, numerous references have been made to the existence of black carbon derived material in humic substances present in soils. Haumaier and Zech (6) proposed that a relatively sharp aromatic signal observed at 130 ppm in solid-state 13C NMR spectra of humic substances in some soils was most likely derived from black carbon and not from humification of native plant materials. Photooxidation has also been used to obtain a measure of black carbon in some soils (7). Using this method, it was concluded that as much as 30% of some Australian soil carbon was charcoal. Using combustion techniques that have been developed for black carbon determinations, Mannino and Harvey (8) have proposed that black carbon can constitute up to 72% of estuarine ultrafiltered dissolved organic matter. While combustion techniques are inherently prone to artifact generation (9) and indirect in their molecular determination of black carbon structures, there is nonetheless a strong inference that very refractory (to combustion) components exist in humic acids. This combined with the observations made for a sample set of humic acids from some Japanese soils (10) leads us to consider that black carbon can be a significant constituent of humic extracts. Accordingly, we present here the first direct evidence for compounds that are very likely black carbon derived in humic acid from a volcanic ash soil from Japan studied previously (11). We employ a new technique of electrospray ionization mass spectrometry in which we obtain sufficient mass resolution and accuracy to calculate elemental formulas for a large percentage of the data. This allows us to specifically identify peaks for compounds that are possibly black carbon derived and to distinguish them from peaks that are assigned to other natural organic matter. We achieve this high resolution by employing a Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR MS). FT-ICR MS has been widely practiced for over 25 yr. With the significant advances in technology over this period, this technique has become extremely useful in resolving the complex distribution of peaks that comprise the mass spectra of humic substances (12-15). Resolving powers of over 200 000 can be achieved, allowing molecules of the same nominal mass but differing elemental composition to be unambiguously distinguished from one another, an important capability considering the fact that one usually observes a multitude of peaks at each nominal mass (15). Electrospray ionization is a “soft” method of ionization that has been shown not to fragment even large-sized molecules (16); the discussion as to whether humic substances undergo significant fragmentation has been largely, though not completely, resolved (14). This issue is important in keeping the complexity of the mass spectrum to a minimum. Also, electrospray ionization is well-suited for the analysis of humic substances due to the requisite polarity that stems from the abundance of carboxylic acid functionalities typically present in humic acids. Negative ion mode is best suited due to the ease of ionization of carboxylic acid functionalities in this mode (14). In the present study, a humic acid derived from a darkcolored volcanic ash soil from Iwata, Japan (hereafter referred to as the diluvial HA), was examined by FT-ICR MS. The soil from which the sample was extracted is likely to be very rich in black carbon-like structures, because the burning and VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3387
charring process associated with disposal of crop residues in agricultural systems is an ideal process for the creation of significant amounts of black carbon in humic acids (17, 18). The polycondensed aromatic structures in black carbon can become highly carboxylated during burning (10) or through biological oxidation (19). This sample was the focus of a previous series of investigations that demonstrated the high benzenecarboxylic acid nature of the humic acid (11, 20), a characteristic that has previously been attributed to the presence of black carbon (6). While the study focuses only on one sample, it is important to mention that the observed NMR spectral characteristics of this sample are not unusual as there are numerous reports in the literature of humic acid samples having similar characteristics (6, 11).
Materials and Methods Instrument Parameters. (a) FT-ICR MS. FT-ICR MS analysis was conducted using a home-built 9.4-T FT-ICR MS at the National High Magnetic Field Laboratory at Florida State University (Tallahassee, FL). The humic acid solution was prepared in pH 8 NH4OH at a concentration of 6 mg/mL. The HA solution was then diluted 1:4 from the stock solution with reagent grade methanol (Fisher Scientific). Instrument conditions were optimized for broadest molecular weight range and the stability of the ion populations within the ICR cell. The sample was introduced into the instrument using a micro-electrospray capillary at 400-600 nL min-1. Capillary voltage was maintained at -1800 V for negative ion mode. Accumulation of the sample in the external octapole (frequency set at 1.7 MHz) occurred for 12 s before transfer to the ICR mass analyzer cell. All timedomain data (size ) 2 Mword) were acquired with a MIDAS data station. The time-domain signal was averaged (100200 scans) before baseline correction, and one zero-fill, Fourier transformation, and magnitude calculation. It has been proposed that negative ion mode can produce multiply charged species from fulvic acids (21); however, this phenomenon was not observed while analyzing this sample. All isotope peaks were observed at exactly 1 neutron mass from the molecular ion, indicating that multiple charge states were not present. Initial standardization of the m/z scale was conducted by reference to an external standard. From experience, we recognize that this methodology is insufficient to allow unambiguous assignment of a unique elemental formula to each peak. Thus, an internal standard injected coincident with the sample is needed to ensure exact mass calibration. We know from previous studies of this sample (11) that benzenecarboxylic acids are expected components. We arbitrarily selected a six-ring aromatic system (anthanthrene) that we expected to be present as tri-, tetra-, penta-, hexa-, hepta-, and octa-carboxylic acids. While there is no specific evidence for the existence of such structures, we assume their presence based on the NMR data showing mainly condensed aromatic rings with carboxyl substituent groups. Since all of the errors present in the assignments based on the externally calibrated data for this series of compounds seemed to follow a general trend and be shifted from their exact masses by consistent amounts, it seemed that a simple shift of the data relative to the exact masses for this known series of compounds would effectively calibrate the spectrum. Using the closest externally calibrated peaks (11-16 ppm error) to the exact masses of the reference compounds, we shifted the entire mass scale to minimize deviation of these six compounds from their exact masses. Subsequently, we found that nearly all peaks in the spectrum could be matched to within 1-2 ppm of the exact mass for a specific elemental composition, and numerous matches were below 1 ppm accuracy. This provided us with great confidence in our internal calibration methodology. 3388
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
(b) 13C NMR. To repeat and verify the nature of carbon functional groups with modern NMR equipment, a solidstate 13C NMR spectrum was obtained on a Bruker DSX-300 NMR spectrometer. We utilized the ramp-cross-polarization technique and two-pulse phase modulated decoupling with details provided previously (22). Magic angle spinning was conducted at a speed of 13 kHz in a 4-mm ceramic pencil rotor with Kel-F end caps. The contact time was set at 2 ms and the delay time to 1 s. Variable contact time experiments are routinely performed in our laboratory to verify use of optimum contact times. The 90° 1H pulse length was 5.2 µs. In addition, solution-state 13C NMR spectra of the humic acid dissolved in D2O were obtained using a Bruker Avance 400 MHz NMR fitted with a QNP 1H, 13C, 15N, 31P probe. The spectra were run as follows: 100 mg/mL in D2O/NaOD; for normal carbon spectra, 32K data points, 30° flip angle, 2 s recycle delay, and 50 Hz line broadening; for Q-DEPT spectra, 32K data points, 2 s recycle delay, 145 Hz 1H-13C coupling constant, 50 Hz line broadening. Samples. (a) Extraction and Purification of Diluvial HA. The humic acid was extracted with 0.1 M NaOH. Precipitation occurred as the pH was adjusted to 2. The precipitate was then dissolved and precipitated again for purification. The origin and preparation procedures can be found in Matsuda and Schnitzer (23). (b) Mt. Ranier HA. A sample of Douglas fir wood that was degraded by brown rot fungi was collected from the slopes of Mt. Rainier, WA. Humic acid from this sample was extracted with 0.5 M NaOH for 24 h under nitrogen. Acidification to pH 2 with HCl followed by centrifugation and repeated washings with pH 2 water produced a dark precipitate that was subsequently freeze-dried. A weighed amount of dry humic acid was redissolved to constitute a stock solution. This stock solution was prepared by dissolving 10 mg of the humic acid in 2 mL of NH4OH (pH 8). Before analysis, the stock solution was diluted with ultrapure water and methanol so that the concentration of the organic matter was 1.25 mg/mL and the ratio of water to methanol was 50:50. (c) Oxidized Peat HA. This peat sample was purchased from the International Humic Substances Society (IHSS). The peat humic acid was then extracted using the standard procedure of the IHSS. After the humic acid was isolated, bleaching was performed according to a method first described by Ritchie and Purves (24). Briefly, 10 g of sodium chlorite, 10 mL of acetic acid, and 100 mL of ultrapure water were used per gram of humic acid. The mixture was stirred overnight and repeated three times, each time with fresh solution.
Results and Discussion The highly carboxylated nature of molecules whose origins are thought to be from black carbon is well-known. Benzenepolycarboxylic acids (BPCAs) have been used as biomarkers for black carbon in previous studies (25), and high yields of BPCAs have been obtained from alkaline permanganate degradation of highly aromatic humic acids (20, 26). It is not surprising that solubility in dilute alkali requires that condensed aromatic structures contain significant amounts of hydrophilic functional groups such as carboxylic acid, alcohol, ketone, and aldehyde groups. Elemental data for the sample of humic acid examined in our study (20) reveal that the sample contains 61.4% carbon, 3.2% hydrogen, 0.8% nitrogen, and 34.3% oxygen. Permanganate oxidation data (26) indicate that the products contain 4.4% benzenetricarboxylic acids, 7.7% benzenetetracarboxylic acids, 4.3% benzenepentacarboxylic acids, and 2.6% benzenehexacarboxylic acids. The low hydrogen content and the distribution of benzenecarboxylic acids clearly define a sample having a highly aromatic and condensed nature. Subsequent NMR
FIGURE 1.
13C
CPMAS NMR spectrum of diluvial humic acid.
FIGURE 2. Normal (all carbons) and Q-DEPT (nonprotonated carbons) NMR spectra of diluvial humic acid. studies (11) concluded that the aromatic systems in these humic acids were highly carboxylated but not highly condensed. 13C NMR. The high field CPMAS 13C NMR spectrum (shown previously at low field; 11) of this humic acid (Figure 1) is dominated by peaks in the aromatic and carbonyl regions. Studies by Haumaier and Zech (6) propose that such spectra are characteristic of humic acids having a high proportion of black carbon. The aliphatic region of the spectrum is nearly devoid of signal, with the exception of one small peak at 40 ppm, suggesting minimal aliphatic character. Although the NMR data confirm that the sample is predominantly composed of carboxylated aromatic structures, dipolar dephasing experiments (11) were used to obtain estimates of ring size and degree of carboxylation. These studies proposed an average number of 3.5 protons per ring and an average of 1-2 carboxylic acids per ring. This would imply small rings and minimal condensation, structures that would not be consistent with the expected condensed rings of black carbon derived structures. NMR studies using advanced solution NMR methods indicate that this humic acid contains large amounts of nonprotonated carbon sites as revealed by the plots shown in Figure 2. The full spectrum is essentially similar to that observed by solid-state 13C NMR, but spectral editing with quaternary-distortionless enhancement by polarization transfer (Q-DEPT) plots only the quaternary or nonprotonated carbons (shown in Figure 2) and indicates that a substantial portion of the aromatic signal is due to nonprotonated carbons. Note that both plots are normalized to the nonprotonated carboxyl carbons that should exhibit the same intensity in both spectra. Considering the fact that few signals are present in the aliphatic region of the spectrum, the nonprotonated carbons are unlikely attributable to alkyl substitutions and are more likely the result of ring condensation. Thus, the solution NMR data clearly conflict with the previous solids NMR data and point to the presence of
condensed ring systems. We should make clear that, for samples such as the ones investigated here, it is very difficult to establish quantitative spectra. It is also known that Q-DEPT editing is mainly a qualitative tool rather than a quantitative representation of nonprotonated carbons. There are some slight differences between the solution spectrum and the CPMAS spectrum shown here and those presented earlier (11). These likely arise from the fact that the cross-polarization technique does not adequately represent carbons that are remote from protons, like those that would be present in condensed aromatic ring systems. The relative enhancement in aliphatic signals in the CPMAS spectrum, compared with the solution spectrum, is consistent with the view that aliphatic carbons are cross-polarized more efficiently than aromatics, especially if the aromatic carbons are remote from protons. FT-ICR MS. The FT-ICR mass spectrum of the humic acid is shown in Figure 3, with the inset showing an expanded region illustrating the high resolving power obtained. All the peaks in the spectrum are below 1000 m/z, with the majority falling between 350 and 800. The mass range examined in this experiment has a lower limit cutoff at an approximate m/z value of 250. Therefore, peaks may be present below the range suggested above that are not seen due to instrumental limitations. In a previous study employing positive ion ESI with a quadrupole time-of-flight mass spectrometer, a significant number of peaks were observed below a mass of 250 Da (27). Thus, it is apparent that the FT-ICR data do indeed discriminate for the higher mass peaks above 250 Da. Although the upper limitation of FT-ICR MS is theoretically near 120 000 m/z (28), there are some factors, other than inherent instrumental limitations, that may cause molecules to appear only at substantially lower m/z values. Multiple charge states on the ions can cause molecules of high molecular weight to be observed at lower m/z values due to their z value being greater than unity. In the mass spectra of this humic acid, all isotope peaks appear approximately 1 Da above the (M - H)- peak, suggesting that all major ions detected during the experiment are singly charged and not multiply charged. These results are similar to what has been observed in other studies (21). Although fragmentation has not been observed in samples having similar characteristics as humic acids (15, 16), fragmentation, while unlikely, remains possible when ionizing humic substances (29). The trends in the mass spectrum of our humic acid are similar to those seen in previous humic samples (13, 14, 29). Several peaks are present at each nominal mass; however, with the resolving power of the FT-ICR instrument, most are able to be clearly resolved from one another. The ability to resolve the cluster of peaks at each nominal mass along with accurate mass calibration allows compounds with the same nominal mass but differing elemental compositions to be assigned accurately. This provides a great deal of structural information about the sample (29). The clusters of peaks at each nominal mass derive from the fact that a multitude of elemental formulas can account for the molecular weight of those compounds being nominally the same. However, the various contributing atoms impart a slightly different exact mass that is manifested in resolved peaks at each nominal mass, depending upon the number of constituent elements (C, H, N, and O) and the resolving power. Deviation from nominal mass is imparted by the fact that the exact mass of hydrogen is 1.0078 amu, oxygen is 15.9949 amu, and nitrogen is 14.0030 amu. Mass defect is a term used to define deviation from nominal mass. It is logical to recognize that hydrogen-rich molecules, like those of lipids, would show a high mass defect while hydrogen-deficient molecules, like those of condensed aromatic structures associated with black carbon, would show a small mass defect. The presence of significant amounts of VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3389
FIGURE 3. FT-ICR mass spectrum of diluvial humic acid.
FIGURE 4. Expanded-scale spectra for the three diverse humic acids. The Japanese HA is the diluvial HA. The mass scale for the Mt. Rainier HA was calibrated according to an added internal standard procedure (29), and that of the oxidized peat HA was calibrated similarly to the diluvial HA sample except that fatty acids were used as internal calibrants. oxygen in an elemental formula can tend to generate a negative mass defect. To demonstrate the above trends, Figure 4 shows an expanded region of the MS for the Japanese HA examined here and the exact same region for the negative ion MS of two other samples under investigation. One is a humic acid extracted from a brown-rotted Douglas fir wood described 3390
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
previously (29), and the other is a peat HA that has been bleached to remove lignin-like components (30). The former is a sample rich in lignin-like structures, and the latter is composed mostly of long-chain aliphatic fatty acid-like structures. It can be clearly observed that in the Japanese HA under study, peaks are located very close to their respective nominal
FIGURE 5. KMD plot of diluvial humic acid. masses. This effect is not observed in a system with few condensed aromatics, such as the lignin-rich Mt. Rainier HA or in a system made up of largely aliphatic structures, such as the bleached peat humic acid. The long-chain fatty acids that exist in the bleached sample cause many of the m/z values to be shifted significantly from nominal mass because the relative number of hydrogen atoms present in the structures is high. Although the Mt. Rainier HA does consist of oxygenated aromatic structures from lignin, the peaks representing these molecules are not as dramatically shifted toward nominal mass as are those of the Japanese diluvial HA. It is known that lignin is not as hydrogen rich as lipids, and it is likely that the associated oxygen on lignin-derived structures induces a negative mass defect shift. However, the aromatic structures in lignin are not condensed, and the presence of protonated aliphatic side chains explains why some mass defect is observed. The least amount of mass defect is observed for the diluvial HA from Japan, suggesting that these structures are composed of condensed aromatic rings and/or contain a significant degree of oxygen functionalization. Such a pattern is not only observed for the narrow mass range shown in Figure 4 but extends over the entire mass range of the spectra. It is obvious that this effect can be used to quickly assess the types of molecules that comprise a particular sample. Kendrick Mass Defect (KMD) Analysis. KMD analysis (31) can be used to process large amounts of data efficiently,
as is generated in the mass spectra of humic substances. In KMD analysis, a representative functional group, usually -CH2-, is selected based on the nature of the sample under study. This group can vary widely, but common groups include -CH2-, -COO, and -O. This group should also constitute part of a homologous series because it is this feature that renders KMD analysis so useful as a structure-defining methodology. Once this group is chosen, the exact m/z value of each peak is taken though a series of calculations that ensure that all members of a homologous series that differ by the selected functional group will have a similar mass defect. The CPMAS 13C NMR spectrum of the humic acid under study reveals it to be very rich in carboxylated aromatic structures (11). Therefore, the functional group selected for KMD analysis is COO. The m/z value normalized to COO is known as the Kendrick mass and can be calculated using the following equation:
Kendrick mass ) (exact m/z value of peak) × ((nominal mass of COO)/(exact mass of COO)) The Kendrick mass of the peak is then converted to KMD value using the following calculation:
KMD ) (nominal mass - Kendrick mass) × 1000 The KMD value can then be plotted against m/z value to give a Kendrick mass defect plot (Figure 5). In this plot, all structures with the same base formula but differing numbers of carboxyl groups will appear on a horizontal line with zero slope and have the same KMD value. If one were to expand the plot shown in Figure 5, especially in the dense region of data points, one observes that peaks differing only in the number of carboxyl groups (m/z 44) do indeed plot on a horizontal line as has been previously shown for KMD plots of data normalized to CH2 units (14). The overall trend can be described as a large number of peaks along series of horizontal lines, each being separated on the ordinate by a KMD value that gradually increases because structural units other than m/z 44 contribute to the mass defect. Although it is difficult to observe clear horizontal lines with peaks separated by 44 mass units as plotted in Figure 5, they do indeed exist, and related structures from peaks along a horizon trend can be selected based on the similarity of their KMD values. Using this technique, numerous related com-
FIGURE 6. Van Krevelen plot of diluvial humic acid. VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3391
FIGURE 7. Linearly fused aromatic structures representing peaks observed in the FT-ICR-MS spectrum of the diluvial humic acids. The error represents deviation of the observed mass from the theoretical value in ppm.
FIGURE 8. Condensed aromatic structures linked by biphenyl bonds and representing peaks observed in the FT-ICR-MS spectrum of the diluvial humic acids. The error represents deviation of the observed mass from the theoretical value in ppm. pounds, separated by 44 mass units, are identified. Many aromatic ring systems (4-9 aromatic rings) are found as part of a series in which each structure differs only in the number of added carboxyl functionalities (see discussion below). In addition to the main cluster of peaks ranging from a KMD value of 0 to -300, other peaks at different KMD values are observed. Some of these appear to vary regularly in their KMD value, indicating that they constitute a homologous series themselves other than that defined by COO. These, however, fall outside the range observed at higher KMD 3392
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
values, indicating that they are not part of the mainstream components defined by less negative KMD values. Van Krevelen Analysis. To further illustrate the highly carboxylated and condensed nature of the sample, the HA was analyzed with a technique first utilized by van Krevelen in 1950 (32). In the van Krevelen analysis, a plot is generated with the atomic ratio of hydrogen to carbon on the y-axis and the ratio of oxygen to carbon on the x-axis. The unique elemental formulas are calculated from the internally calibrated FT-ICR mass spectral data, as described in a recent
FIGURE 9. Condensed aromatic structures representing peaks observed in the FT-ICR-MS spectrum of the diluvial humic acids. The error represents deviation of the observed mass from the theoretical value in ppm.
FIGURE 10. Aromatic structures for which searches of the FT-ICR MS data were made. paper from our group (33), and molar H/C and O/C ratios are calculated for each peak. The clear advantage of this plot is that compounds with similar structural characteristics will appear in the same region of the van Krevelen plot. This allows specific types of structures such as fatty acids, ligninlike molecules, and condensed aromatics to be clearly distinguished from one another based on their differences in H/C and O/C ratios (33). The van Krevelen plot of the humic acid is shown as Figure 6. It is obvious from this plot that the type of structures present in the sample have very low H/C ratios of approximately 0.5. This confirms that the bulk of structures in the sample are condensed in nature. The absence of hydrogen in the condensed centers of the molecules found in this study leads to an H/C ratio that is much less than that seen in more aliphatic, or less condensed, systems. The O/C ratios of the sample range from approximately 0.25 to 0.5. The rather high O/C ratio is due to the high degree of carboxylation present
in the sample. The wider range of this value is probably a result of the wide variance in the number of carboxyl groups present on each of the structures. Structural Analysis. In addition to this general information, much more structure-specific information can be obtained by coupling FT-ICR MS with KMD analysis. Applying these techniques, several peaks in the mass spectrum of the humic acid have been assigned to specific molecular formulas and often to specific structures. The extremely high resolving power of the instrument (>200 000) has allowed identification of many peaks that are likely to have a structure that is expected from black carbon-like materials even in this very complicated sample. Three general types of structural entities were identified in this experiment: linearly fused aromatics, aromatics linked by carbon-carbon single bonds, and highly condensed (fused) aromatic structures. It is important to point out that the electrospray process can be very selective and that the relative contributions of peaks identified may VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3393
be nonrepresentative of the sample. While this is likely the case, use of other techniques, such as NMR, help ensure that the structures identified are significant components of the sample. The linearly fused aromatic compounds detected in the HA range from four (tetracene) to seven (heptacene) rings in length (Figure 7). The number of carboxyl groups on each structure varies from three to eight. As the number of aromatic rings in the chain increases, relative intensity (abundance) of the peaks with more carboxyl groups increases as well. It is not known at this time if this is a result of greater solubility or a trend present in the sample. It would seem that as the ring system gets larger, the structures with more carboxyl groups would be more common. However, the solubility of the larger ring systems could also become a factor as the molecules become larger. The overall relative intensity of these linearly fused structures is somewhat low, as these structures are not as heavily condensed as would be expected from a sample rich in black carbon-like structures. It should also be noted that these linear molecules often have very similar structural isomers, which cannot be differentiated from each other using this mass spectrometric technique. These changes are minor, however, and will not greatly affect the properties of the molecules. A series of aromatic structures linked by carbon-carbon single bonds is also detected (Figure 8). These compounds vary in structure, but all contain two aromatic ring systems linked by carbon-carbon single bonds. These types of compounds are similar to those predicted in earlier studies of this HA (11). However, these compounds also are somewhat weak in relative intensity. Figure 9 illustrates all of the condensed aromatic systems identified by FT-ICR up to this point. These structures and their isomers are consistent with those one would expect to find in a sample rich in black carbon. As expected, these compounds produce some of the most intense peaks in the mass spectrum. The relative intensities vary, but generally the peaks representing compounds with five or six carboxyl groups are the most intense. These identifications confirm that the sample is rich in condensed, highly carboxylated aromatic structures, as was predicted by solution 13C NMR studies discussed above. Several previous studies have indicated that the subject humic acids were either not polycondensed (11) or were accumulations of linearly linked polycarboxylic acids, linkages being through either hydrogen bonding or carbon-carbon bonds (20). Such molecules, if they exist in significant amounts, should be detected by FT-ICR MS. We constructed a number of possible structures that conform with the suggestions that linearly linked polycarboxylic acids are major components of such humic acids (Figure 10) and searched the FT-ICR MS database for their presence, knowing their exact mass. We could find few, if any, peaks in the spectrum that could be assigned to these structures within the mass precisions observed for all other assignments made, an indication that they are simply not realistic models for this HA. The heavy carboxylation of the structures detected in this sample could have a profound impact on the study of black carbon in soils. It is currently believed that black carbon is highly refractory and insoluble and therefore will remain in soils without any significant depletion. However, if many of the black carbon-like molecules are soluble in water, a mechanism is provided for the black carbon to be mobilized and therefore removed from the soil. It is our intent to study aquatic systems such as rivers, lakes, and oceans to verify that black carbon can be detected in the DOM of these systems. We feel that the identification of these and other 3394
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 12, 2004
specific structures in black carbon-rich systems (along with the study of DOM from rivers, lakes, and oceans) could allow a much better understanding of black carbon and its contribution to many environmental systems. It is clear that the ESI FT-ICR MS technique has the potential to identify these types of compounds, and we anticipate that the approach will eventually allow us to assess their relative contribution to the various environmental compartments.
Acknowledgments We thank the National Science Foundation, Grants CHE0089172 and CHE-0089147 at Ohio State University and CHE9903528 at the National High Magnetic Field Laboratory, Florida State University, for financial assistance. Additionally we thank Dr. Alan Marshall for providing access to the Florida State facility. We also thank Dr. Michael A. Freitas, Ohio State University, for his advice in optimization of FT-ICR instrument parameters and Dr. Andre J. Simpson for the NMR data. This manuscript benefited greatly from several anonymous reviewers, and we thank them for their constructive comments.
Literature Cited (1) Skjemstad, J. O.; Clarke, P.; Taylor, J. A.; Oades, J. M.; McClure, S. G. Aust. J. Soil Res. 1996, 34, 251. (2) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Science 2001, 294, 2119. (3) Mitra, S.; Bianchi, T. S.; McKee, B. A.; Sutula, M. Environ. Sci. Technol. 2002, 36, 2296. (4) Schmidt, M. W. I.; Knicker, H.; Hatcher, P. G.; Kogel-Knabner, I. Org. Geochem. 1996, 25, 29. (5) Schmidt, M. W. I.; Noack, A. G. Global Biogeochem. Cycles 2000, 14, 777. (6) Haumaier, L.; Zech, W. Org. Geochem. 1995, 23, 191. (7) Skjemstad, J. O.; Janik, L. J.; Head, M. J.; McClure, S. G. J. Soil Sci. 1993, 44, 485. (8) Mannino, A.; Harvey, H. R. Limnol. Oceanogr. (in press). (9) Gelinas, Y.; Prentice, K. M.; Baldock, J. A.; Hedges, J. I. Environ. Sci. Technol. 2001, 35, 3519. (10) Shindo, H.; Honma, H. In Humic Substances: Structures, Models, and Functions; Ghabbour, E. A., Davies, G., Eds.; Royal Society of Chemistry: Cambridge, 2001; p 297. (11) Hatcher, P. G.; Schnitzer, M.; Vassallo, A. M.; Wilson, M. A. Geochim. Cosmochim. Acta 1989, 53, 125. (12) Solouki, T.; Freitas, M. A.; Alomary, A. Anal. Chem. 1999, 71, 4719. (13) Brown, T. L.; Rice, J. A. Anal. Chem. 2000, 72, 384. (14) Stenson, A. C.; Landing, W. M.; Marshall, A. G.; Cooper, W. T. Anal. Chem. 2002, 74, 4397. (15) Kujawinski, E. B.; Freitas, M. A.; Zang, X.; Hatcher, P. G.; GreenChurch, K. B.; Jones, R. B. Org. Geochem. 2002, 33, 171. (16) Gaskell, S. J. J. Mass Spectrom. 1997, 32, 677. (17) Shindo, H.; Matsui, Y.; Higashi, T. Soil Sci. Plant Nutr. 1986, 32, 475. (18) Shindo, H.; Matsui, Y.; Higashi, T. Soil Sci. 1986, 141, 84. (19) Willmann, G.; Fakoussa, R. M. Appl. Microbiol. Biotechnol. 1997, 47, 95. (20) Matsuda, K.; Schnitzer, M. Soil Sci. 1972, 114, 125. (21) Pfeifer, T.; Klaus, U.; Hoffmann, R.; Spiteller, M. J. Chromatogr. A 2001, 926, 151. (22) Dria, K. J.; Sachleben, J. R.; Hatcher, P. G. J. Environ. Qual. 2002, 31, 393. (23) Matsuda, K.; Schnitzer, M. Bull. Environ. Contam. Toxicol. 1971, 6, 200. (24) Ritchie, P. F.; Purves, C. B. Pulp Pap. Can. 1947, 48, 74. (25) Glaser, B.; Haumaier, L.; Guggenberger, G.; Zech, W. Org. Geochem. 1998, 29, 811. (26) Schnitzer, M.; Calderoni, G. Chem. Geol. 1985, 53, 175. (27) Kramer, R. W.; Kujawinski, E. B.; Zang, X.; Green-Church, K. B.; Jones, R. B.; Freitas, M. A.; Hatcher, P. G. In Humic Substances: Structures, Models, and Functions; Ghabbour, E. A., Davies, G., Eds.; Royal Society of Chemistry: Cambridge, 2001; p 95.
(28) Marshall, A. G.; Hendrickson, C. L.; Jackson, G. S. Mass Spectrom. Rev. 1998, 17, 1. (29) Kujawinski, E. B.; Hatcher, P. G.; Freitas, M. A. Anal. Chem. 2002, 74, 413. (30) Chefetz, B.; Salloum, M. J.; Deshmukh, A. P.; Hatcher, P. G. Soil Sci. Soc. Am. J. 2002, 66, 1159. (31) Hughey, C. A.; Hendrickson, C. L.; Rodgers, R. P.; Marshall, A. G.; Qian, K. N. Anal. Chem. 2001, 73, 4676.
(32) van Krevelen, D. Fuel 1950, 29, 269. (33) Kim, S.; Kramer, R. W.; Hatcher, P. G. Anal. Chem. 2003, 75, 5336.
Received for review September 11, 2003. Revised manuscript received March 2, 2004. Accepted March 28, 2004. ES030124M
VOL. 38, NO. 12, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3395
