Parallel Factor Analysis of Excitation–Emission Matrix Fluorescence

Nov 30, 2007 - Characterizing the Interactions between Trace Metals and Dissolved Organic Matter Using Excitation−Emission Matrix and Parallel Facto...
0 downloads 9 Views 3MB Size
Download PDF
Environ. Sci. Technol. 2008, 42, 186–192

Parallel Factor Analysis of Excitation–Emission Matrix Fluorescence Spectra of Water Soluble Soil Organic Matter as Basis for the Determination of Conditional Metal Binding Parameters T S U T O M U O H N O , * ,† A R I A A M I R B A H M A N , ‡ AND RASMUS BRO§ Department of Plant, Soil, and Environmental Sciences, University of Maine, 5722 Deering Hall, Orono, Maine 04469-5722, Department of Civil and Environmental Engineering, University of Maine, 5711 Boardman Hall, Orono, Maine 04469-5711, and Department of Food Science, Faculty of Life Sciences, University of Copenhagen, Rolighedsvej 30, DK-1958 Frederiksberg C, Denmark

Received July 26, 2007. Revised manuscript received October 14, 2007. Accepted October 25, 2007.

Organic matter-metal complexes in soil solution and aquatic systems are involved in important environmental and ecological processes such as plant nutrient availability and the solubilization and transport of metals. Our work presented here extends the use of fluorescence spectrometry for determining conditional stability constants for such complexes. We combine the use of excitation–emission matrix (EEM) fluorescence spectrometry and parallel factor analysis (PARAFAC) to determine the stability constants of the chemically meaningful components modeled by PARAFAC. Water-soluble organic matter (WSOM) from O-horizon soils of deciduous and coniferous forest stands were extracted and titrated at pH ) 4.7 with iron(III) (Fe) and aluminum (Al) which are important metals in acid soil systems. The EEM spectra were then recorded and PARAFAC analysis showed that the WSOM contained three humicsubstance-like components. Fe titration led to fluorescence quenching of the three components, while Al titration enhanced fluorescence for two components and quenched one of the components. The average Ryan-Weber stability constants at pH 4.7 ranged from log K of 4.28 to 4.91 for Fe and 4.84 to 5.96 for Al. The conditional stability constants were similar for Fe binding for deciduous and coniferous stand-derived WSOM, while they were stronger for Al binding with coniferous stand-derived WSOM. This difference in binding strengths for Al may affect the chemical behavior of Al in soil and aquatic systems. Determining the individual binding parameters of organic matter components with metals represents a significant * Corresponding author telephone: +1 207-581-2975; fax: +1 207581-2999; e-mail: [email protected]. † Department of Plant, Soil, and Environmental Sciences, University of Maine. ‡ Department of Civil and Environmental Engineering, University of Maine. § University of Copenhagen. 186

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

advance over current approaches that utilize fluorescence quenching at a single excitation–emission wavelength pair to characterize organic matter-metal interactions.

Introduction Regional effects of acidic deposition on terrestrial ecosystems include acidification of soils and surface waters, loss of soil base cations, and the mobilization of metals (1–3). Norton et al. (4) have hypothesized a sequential acid neutralization mechanism in soils where accelerated base cation losses occur in the early phase of acidification, followed by decreasing absolute loss of base cations in the midphase, and decreasing soil pH, Al mobilization, and Fe mobilization in the later stages of the soil acid neutralization process. Bear Brook Watershed in Maine (BBWM) is a site of a longterm, paired-watershed experimental acidification study that includes both deciduous and coniferous stands. Undisturbed forest soil profiles are typically characterized by the presence of a surface forest floor (i.e., O-horizon) with a high organic matter content which is present above the underlying mineral soil horizons. Although the water-soluble organic matter (WSOM) fraction of soils typically constitutes less than 1% of total organic matter, it is the most labile and reactive fraction of the multicomponent soil organic matter pool (5). A study of the chemical properties of BBWM soils indicated a strong negative relationship between soil pH and soil Al, Fe, phosphorus (P), and carbon (C). The close clustering of Al, Fe, P, and C loadings of a principal components analysis for the mineral BBWM soils suggested that organic matter is important for Al and Fe mobilization and that P solubility is influenced by mobilization of Al and Fe (6). Fluorescence spectroscopy has been extensively utilized to measure the stability constants of organic matter with metals (7, 8). This technique relies on the quenching of the fluorescence signal commensurate with metal-organic matter complex formation. Due to the broad nature and the lack of fine structure of the organic matter fluorescence spectra, almost all such studies have used metal titrations based on quenching at an excitation-emission wavelength pair corresponding to the strongest peak. Luster et al. (9) used multidimensional excitation–emission matrix (EEM) fluorescence spectroscopy, which collects a complete profile of fluorescence intensity responses along both excitation and emission wavelengths, to investigate metal complexation. This approach demonstrated that quenching (or enhancement) of fluorescence intensity by metal complexation was quantitatively dissimilar in different regions of the fluorescence landscape. Bro and co-workers have shown that combining EEM fluorescence spectroscopy with parallel factor analysis (PARAFAC), a multiway data analysis method, can model a suite of complex EEM landscapes into chemically meaningful spectral and concentration components (10–12). A study of WSOM extracted from soils and plant biomass has shown that the fluorescence landscape of that particular sample set could be modeled using 5 PARAFAC components (13). Luster et al. (9) have noted that calculated stability constants based on fluorescence spectrometry are dependent on the particular excitation-emission wavelength pair used from the fluorescence intensity landscape. The visual “peak picking” process is subjective in nature and may lead to wrong conclusions because peaks are commonly reflecting more than one underlying fluorophore. The goal of our study was to demonstrate the use of PARAFAC to separate and select peaks corresponding to the chemically meaningful PARAFAC 10.1021/es071855f CCC: $40.75

 2008 American Chemical Society

Published on Web 11/30/2007

components (i.e., fluorophores) modeled in WSOM. The specific objectives of this study were to evaluate the use of PARAFAC to determine the binding parameters of WSOM components with Fe and Al and to compare the metal stability constants of WSOM extracted from the O-horizons of deciduous and coniferous forest stands. Insight on the differences in metal binding of WSOM from these contrasting stand types may be critical to understanding mechanisms of metal and nutrient (e.g., P) accumulation and mobilization in forested ecosystems.

Experimental Section Site Description, Sampling, and Extraction. O-horizons of deciduous and coniferous stands were collected from the reference watershed at BBWM. Details about the sampling site can be found elsewhere (6). Soils supporting forests at the BBWM are freely drained Spodosols [FAO classification Podzols], typical of northern New England, and there is no history of physical disturbance or cultivation at this site. Soils are in the Turnbridge and Rawsonville soil series (loamy, mixed, frigid Typic Haplorthods) with well-developed spodic horizons. Organic horizons were excavated using a 15 × 15 cm frame and the material was collected to the surface of the underlying mineral horizon. All samples were air-dried. Two types of vegetation predominate in the watersheds. The upper elevations have coniferous stands, composed primarily of red spruce (Picea rubens Sarg.) and balsam fir (Abies balsamea L. Mill.), while the lower elevations have deciduous stands, composed primarily of American beech (Fagus grandifolia Ehrh.), yellow birch (Betula alleghaniensis Britt.), red maple (Acer rubrum L.), and sugar maple (Acer saccharum Marsh.) (14). Soil pH was measured using distilled water at a 1 g:10 mL ratio. The organic soil horizons were extracted at a 10:1 (m:v) sample/deionized water ratio. The samples were extracted for 16 h at 4 °C and the suspension was vacuum filtered through a glass fiber filter. The eluent was passed through a H+-saturated cation exchange resin (AG 50W-X8; Bio-Rad, Hercules, CA) to remove free metals present in the eluent. The total soluble carbon (CTS) concentrations of the WSOM extracts were determined using a Shimadzu 5000 analyzer. The acid functional group content of each material was determined by potentiometric titration in a glass reaction beaker maintained at 25.0 ( 0.1 °C. Titration solutions were adjusted to 20 mmol L-1 CTS and 20 mmol L-1 ionic strength using 1 M KCl solution. The solution was adjusted to pH 3 using 0.10 mol L-1 HCl, N2 was bubbled through the solutions for 15 min prior to titration to minimize CO2 contamination, and titrated with standardized 0.05 mol L-1 NaOH to pH 10. The moles of titrant consumed between the operational beginning and end points of pH 3 and pH 8 was taken to be equal to the carboxyl group content (15). Phenolic group content was taken to be equal to twice the titrant consumed between pH 8 and 10 (15). Blank corrections were made by subtracting the quantity of base consumed in titrating 0.02 M KCl solutions in from pH 3 to 10. Fluorescence Titration. The metal titrations were conducted in a thermostatic 500-mL beaker at 298 K containing 150 mL of 30 mg CTS L-1 for all WSOM extracts. The pH was adjusted to 4.7 (mean field soil pH of the forest floor) with 0.1 M KOH or HCl. The solutions were titrated with 1000 mg L-1 Fe or Al made from chloride salts. The titration range was 0-97.9 µM Fe and 0-76.6 µM Al. Solutions were mixed with a stir-bar and addition of the metal titrant was made using micropipets. The pH of the solutions was readjusted to 4.7 with 0.05 mol L-1 NaOH after each addition of metal and allowed to equilibrate for 15 min prior to the removal of 1 mL aliquots for fluorescence EEM analysis. The aliquots were added back to the solution prior to the next addition of metal titrant. Fluorescence EEM spectra were obtained using a

Hitachi F-4500 spectrofluorometer with an excitation range set from 240 to 400 nm and an emission range set from 300 to 500 nm in 3 nm increments. Instrumental parameters were excitation and emission slits, 5 nm; response time, 8 s; and scan speed, 1200 nm min-1. PARAFAC Modeling. The PARAFAC modeling approach has been described in detail elsewhere (10–13) and only a brief description will be given here. Essentially a PARAFAC model of a set of EEM landscapes provides an estimate of the number of fluorophores as well as the excitation and emission spectrum of these fluorophores. It also provides the relative concentration of each fluorophore in each sample. Each fluorescence landscape measurement provides an EEM and when several of these are combined, they can be held in a so-called three-way array of size I × J × K, where I is the number of samples, J is the number of emission wavelengths, and K is the number of excitation wavelengths. Such a three-way array can not be directly modeled by standard multivariate analysis tools because these only work on two-way matrices. The PARAFAC model is specifically made to deal with such three-way data and can be viewed as an extension of principal component analysis (16). Unlike principal component analysis, the PARAFAC model is uniquely identified without additional orthogonality constraints (17). This means that if the underlying structure of the three-way data coincides with the PARAFAC model, then the parameters of the PARAFAC model will reflect the true underlying parameters. For fluorescence data, this is ideally the case. Each fluorophore will give rise to one PARAFAC component and each such component consists of an estimated emission spectrum, and estimated excitation spectrum and a score vector where each element is the relative concentration of the fluorophore. The PARAFAC modeling was conducted with MATLAB Release 14 (Mathworks, Natick, MA) using PLS_Toolbox version 4.0 (Eigenvector Research, Manson, WA). A nonnegativity constraint was applied to the parameters to allow only chemically relevant results. The PARAFAC models with two to six components were computed for the pooled (Fe and Al titration EEM and replicates) deciduous and coniferous O-horizon data sets separately. The determination of the correct number of components in the data set was assessed by the core consistency diagnostic score which should be close to 100% for appropriate models. The core consistency provides an estimate of how well the model captures trilinear information, and if the consistency turns low, i.e. toward zero, it is a strong indication that the model is invalid (18). The number of components was further validated by visual inspection of the estimated parameters and additional model diagnostics. Several preprocessing steps were used to minimize the influence of scatter lines and other attributes of the EEM landscape that are due to the background solution matrix prior to PARAFAC modeling. Subtraction of a control DIH2O EEM from sample EEM was used to remove the lowerintensity Raman scatter lines. The higher intensity Rayleigh scatter lines were removed by replacing the fluorescence intensity values with missing values in the region immediately adjacent to where emission wavelength was equal to 1 and 2 times the excitation wavelength. In addition, the EEM spectra had a triangular shaped region where the emission wavelength was less than that of the excitation wavelength. Such a characteristic is a physical impossibility and thus these data pairs were set to zero (19). Complexation Modeling. The Ryan-Weber model (20) was used to determine the binding parameters for Al and Fe for the PARAFAC-derived components. The conditional equilibrium stability constant, Kc, and the complexation capacity, Lt, and the relative fluorescence intensity of the VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

187

TABLE 1. Selected Chemical Properties of the Water-Soluble Organic Matter from the Reference Deciduous and Coniferous Stands at Bear Brook Watershed in Maine Study Site parameter kg-1

water-soluble organic C, g soil 280 nm absorptivity, L mol-1 cm-1 E2/E3 absorbance ratio carboxyl-group content, mmol g-1 C phenolic-group content, mmol g-1 C total acidity content, mmol g-1 C

deciduous

coniferous

2.09 104 5.3 8.0 2.7 10.7

5.48 83 7.1 6.1 1.9 8.0

metal-saturated complex, IMeL/Iref, may be determined using nonlinear fitting of eq 1 (9) I Iref

) 1 +

(

)

IMeL 1 - 1 1 + KcLt + Kc[Me]t Iref 2KcLt

(

√(1 + KcLt + Kc[Me]t)2 - 4K2cLt[Me]t)

(1)

where I is the fluorescence intensity at the metal concentration [Me]t and Iref is the fluorescence intensity in the absence of added metal. The values of I and Iref can be found from the first loading of the PARAFAC model. The use of nonlinear fitting to estimate all three fitting parameters can lead to overly small estimates of Lt (9). The number of fitting parameters of the Ryan-Weber model was reduced from three to two using the approach described by Luster et al. (9). A constant value for IMeL/Iref can be obtained from nonlinear fitting of eq 2

|

I Iref

| |

- 1 )

|

IMeL - 1 (1 - e-R[Me]t) Iref

(2)

where IMeL/Iref - 1 and R are the fitting parameters. This modification to reduce the number of fitting parameters to two leads to reasonable estimates of Kc and Lt.

Results and Discussion WSOM Properties. The coniferous O-horizon had over twice the WSOM content as the deciduous O-horizon (Table 1). Ultraviolet molar absorptivity at 280 nm and E2/E3 ratio (absorbance at 250 nm/absorbance at 365 nm) has been used to probe the aromaticity and humification status of WSOM (22, 23). Both UV spectrometric indices indicate that the deciduous WSOM was about 20% more humified than the coniferous WSOM (Table 1). Carboxyl- and phenolicacidic functional group contents measured by potentiometric titration showed that the deciduous WSOM contained about 20% more acidic functional groups than the coniferous WSOM. These differences between the two WSOM extracts may affect their ability to bind metals since acidic functional group content and aromaticity are important chemical factors controlling metal complexation by organic matter ligands. EEM Fluorescence Landscape of WSOM. The EEM spectra for the deciduous and coniferous stand O-horizon WSOM at a 30 mg CTS L-1 concentration without metal addition are shown in Figures 1 and 2A. They both contain two readily observable peaks which are common in EEM landscapes of terrestrial and aquatic organic matter. The peak centered near 325 nm EX and 440 nm EM has been previously classified as the “C” fluorophore and the peak with 500 nm range has been previously classified as the “A” fluorophore (24, 25). The fluorescence intensity (arbitrary units) of the deciduous WSOM was greater than that of the coniferous WSOM, which was likely attributable to the greater aromatic character of the former, as indicated by the UV absorptivity and the E2/ E3 ratio results (Table 1). 188

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

FIGURE 1. Multidimensional fluorescence spectra of deciduous-stand-derived water-soluble organic matter: (A) without metal addition, (B) with 24 µmol L-1 Fe, and (C) with 24 µmol L-1 Al. Approximate maximum intensities of the prominent peaks are shown. The effect of Fe and Al addition on the deciduous WSOM fluorescence spectra is shown in Figure 1B and C, respectively. The fluorescence intensity was quenched by about 30% in the presence of 25 µM Fe for both A and C fluorophores. The fluorescence response to Al addition differed from that of Fe, with about a 14% intensity enhancement of the A peak, and a 10% quenching of the C peak. Previous studies have reported both enhancement (9) and quenching (26) of fluorescence intensity of WSOM in the presence of Al. Cabaniss (27) reported that the fluorescence intensity of river

FIGURE 2. Multidimensional fluorescence spectra of coniferous-stand-derived water-soluble organic matter: (A) without metal addition, (B) with 24 µmol L-1 Fe, and (C) with 24 µmol L-1 Al. Approximate maximum intensities of the prominent peaks are shown. water fulvic acid was enhanced at pH 5.0 and quenched at pH 7.5 indicating that observed fluorescence intensity was sensitive to the acid–base status of the fulvic acid The effects of metal addition to the coniferous WSOM are shown in Figure 2B and C. The fluorescence response of the coniferous WSOM to Fe was similar to that of the deciduous WSOM, with about a 30% reduction in intensity for both the A and C fluorophores. In contrast, following Al addition to the coniferous WSOM, no observable change was observed in the intensity of the

C fluorophore while the intensity of the A fluorophore was enhanced by approximately 9%. PARAFAC Modeling. PARAFAC models with 2-6 components were computed for the deciduous and coniferous WSOM data sets separately. The appropriate number of components is chosen sufficiently high to fully describe the systematic variation in the data to the point where the assumptions of trilinearity become invalid, i.e. core consistency diagnostic scores becomes low (13). The core consistency diagnostic score was >94% for the three component models for both WSOM data sets. Models with more components yielded core consistencies below or close to zero indicating that the three-component model provided the greatest spectral resolution for WSOM sets. To the extent that quenching and inner-filter effects can be assumed to be approximately negligible, each PARAFAC component describes the contribution from one fluorophore. Thus, much as in a chromatographic analysis, the mixture EEM is split into contributions from individual fluorophores. The PARAFAC emission loadings then provide estimates of the emission spectra of each fluorophore and the excitation loadings of the corresponding excitation spectra. The scores (sample mode loadings) are then the relative concentrations of each fluorophore. That this interpretation is valid is substantiated by the very similar estimates obtained independently from deciduous and coniferous WSOM models (Figure 3). These spectral loadings are characteristic of humic- and fulvic-like molecules (25, 28). The relative fractional distribution of components one, two, and three were 0.59, 0.27, and 0.14, respectively, for the deciduous WSOM, and 0.62, 0.21, and 0.17, respectively, for coniferous WSOM. These PARAFAC results show that O-horizon derived WSOM from the two forest types are very similar. Iron (III) and Aluminum (III) Titration. The maximum quantity of Fe and Al added to the titration vessels was 14.6 and 7.8 µmol, respectively. The total acidic functional group content determined for the samples was 47.7 and 36.0 µmol for the deciduous and coniferous WSOM, respectively. Thus, in all cases there was an excess of functional group concentration as compared to total metal concentration. The fluorescence intensity of the three components as a function of added Fe and Al is shown in Figure 4 for the deciduous and coniferous WSOM. The addition of Fe quenched the fluorescence intensity of all three components for the deciduous WSOM and components 1 and 2 for the coniferous WSOM. Component 3 for coniferous WSOM increased slightly with the initial addition of Fe and then decreased somewhat with further Fe additions. The effect of Al on the fluorescence intensity for the deciduous WSOM was to quench component 1 and enhance components 2 and 3. These component responses clearly demonstrate that PARAFAC modeling can improve our understanding of the interaction of organic matter ligands with metals by showing the independent metal reactivities of separate WSOM components. In a pioneering paper on the use of fluorescence spectrometry to determine metal–ligand stability constants, Ryan and Weber (29) noted that metal ions can precipitate organic matter and used Rayleigh scattering measurements during the titration to monitor potential precipitation. Precipitation of organic matter is especially likely with trivalent metals such as Fe and Al, as used in this study. Fluorescence intensity at the excitation and emission wavelength pair of 350 nm/ 350 nm was used to monitor scattering through the probable formation of precipitates (Figure 5) (30). Metal concentrations up to 40 µmol L-1 did not cause significant increases in scatter for solutions of either deciduous or coniferous WSOM. There were trends suggesting that Al caused greater scatter than Fe, and that scatter occurred to a greater extent in coniferous WSOM than in deciduous WSOM. Interactions of metals with VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

189

FIGURE 3. Excitation and emission spectral loadings of the non-negatively constrained three-component PARAFAC model of the deciduous- and coniferous-stand-derived water-soluble organic matter. precipitates may occur, and lead to the removal of metal ion from solution through adsorption processes (31). Complexation Modeling. The calculated log Kc values using the Ryan-Weber model for the fluorescence quenching and enhancement in Figure 4 are shown in Table 2. It is interesting to note that for both Fe and Al titrations, only two of the three components of the coniferous WSOM exhibit a monotonic quenching or enhancement required for complexation modeling (Figure 4). For Fe complexation, binding by the PARAFAC-derived WSOM components ranged between log Kc of 4.28 and 4.91. Values of Lt for Fe were similar between the deciduous and coniferous WSOM (Table 2). The log Kc for Al complexation ranged from 4.84 to 5.96 indicating a stronger binding of Al than Fe for the two WSOM used in the study. The values observed for Fe(III) and Al complexation constants are in good agreement with values reported by other investigators using fluorescence quenching by metals. Esteves da Silva et al. (7) determined the Ryan and Weber modeled Fe(III) complexation stability constants to range between log K of 5.0 and 5.6 at pH 4.0 for the fulvic acid fraction extracted from composted sewage sludge, municipal wastes, and livestock wastes. Esteves da Silva and Machado (32) report fluorescence enhancement with Al and a conditional log stability constant of 5.65 and 4.64 for Al by fulvic acid extracted from the leaf litter and lower soil horizons of a pine stand, respectively. The stability constant for the Suwannee River Fulvic Acid at pH 4.0 was reported to be log K of 5.09 for Fe(III) and 5.11 for Al using the Ryan and Weber 190

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

analysis of the fluorescence enhancement (33). Browne and Driscoll (34) reported a conditional log stability constant of 4.24 for binding of Al by the Suwannee River Fulvic Acid at pH 4.7. The Al stability constants for WSOM extracted from Chestnut leaf litter were log K of 4.83 at pH 4.5 and 4.66 at pH 5.0 (35). Using discontinuous visible spectrophotometic titration, Fe complexation stability constant for a chestnut leaf litter extract was log K of 5.3 (36). The calculated binding capacities, Lt, normalized to a mmol metal g-1 C basis are shown in Table 2. Luster et al. (9) reported a Lt value of 0.43 mmol Al per g-1 C for a juniper WSOM, which is very similar to the value of 0.52 and 0.63 mmol Al g-1 C that we report for WSOM from a coniferous forest floor. The Lt value of Fe for a chestnut leaf litter WSOM measured using discontinuous spectrophotometric titration was reported to be 1.1 mmol Fe g-1 C (36), which is near the lower end of the Lt range found in this study. Using the relative fractional component distribution values reported in the above text and assuming that PARAFAC-derived component distribution is representative of the whole extract, Lt of the extract reflecting the component composition can be calculated. The complexation capacities for the deciduous WSOM are 2.88 and 1.08 mmol g-1 C for Fe(III) and Al, respectively. The Lt values for the coniferous WSOM are 0.96 and 0.20 mmol g-1C for Fe(III) and Al, respectively (Table 2). For both Fe and Al, the calculated Lt values are lower for coniferous WSOM than deciduous WSOM, which is likely due to the

FIGURE 4. Fluorescence intensity scores of the titration of the deciduous- and coniferous-stand-derived water-soluble organic matter components with Fe(III) and Al(III).

TABLE 2. Conditional stability constant, Kc, and complexation capacity, Lt, for Fe and Al binding to PARAFAC modeled components of water-soluble organic matter from the reference deciduous and coniferous stands at Bear Brook Watershed in Maine study site. Complexation capacity expressed as mmol metal g-1 C deciduous WSOM coniferous WSOM metal PARAFAC component Fe Al

FIGURE 5. Rayleigh scattering at 350 nm for the titration of the deciduous- and coniferous-stand-derived water-soluble organic matter with Fe(III) and Al(III). lower total acidity functional group content of the coniferous WSOM (Table 1). The low Lt values compared to the total acidity group content are likely due to a number of factors including formation of multidentate complexes

1 2 3 1 2 3

Log Kc

Lt

4.91 4.40 4.28 4.97 5.08 4.84

1.37 3.50 8.03 1.06 1.03 1.24

Log Kc

Lt

4.85 1.08 4.72 1.38 not modeled not modeled 5.96 0.52 5.24 0.63

and inaccessibility of metal ions to acidic groups due to steric or electrostatic effects (37). The binding of metals to organic ligands plays a significant role in many environmental and ecological processes. We have demonstrated that multidimensional fluorescence spectroscopy combined with PARAFAC analysis can provide a more detailed characterization of the metal binding process than the traditional method where quenching (or enhanceVOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

191

ment) at a single excitation-emission wavelength pair is used. The multidimensional data approach has the additional advantage of allowing for the in situ monitoring of scattering, enabling the researcher to ensure that any observed quenching or enhancement is due to complexation reactions, rather than adsorption/precipitation processes. This approach may be useful for geochemical modeling of soil solutions. Often, organic matter and metal interactions are characterized by just one constant which is an oversimplification of the complexation process. PARAFAC has been shown to discriminate aquatic and terrestrial organic matter into different meaningful components and estimate the relative concentration of these differing component fractions. With the demonstrated ability of PARAFAC to determine the binding parameters of these component fractions with metals, a more detailed estimation of metal-organic matter speciation should be possible. This would be a significant advance in the study of a variety of important soil processes such as the availability (or toxicity) of metals to plants and the translocation of metals through the soil profile.

Acknowledgments This project was supported by National Research Initiative Competitive Grant 2003-35107-13628 from the USDA Cooperative State Research, Education, and Extension Service, and has also been supported by Hatch funds provided by the Maine Agricultural and Forest Experiment Station. This is MAFES Journal Publication no. 2983.

Supporting Information Available Site description, sampling, and extraction; WSOM properties; PARAFAC modeling. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Driscoll, C. T.; Lawrence, G. B.; Bulger, A. J.; Butler, T. J.; Cronan, C. S.; Eagar, C.; Lambert, K. F.; Likens, G. E.; Stoddard, J. L.; Weathers, K. C. Acidic deposition in the northeastern United States: Sources and inputs, ecosystem effects, and management strategies. Bioscience 2001, 51, 180–198. (2) Fernandez, I. J.; Rustad, L. E. Soil response to S and N treatments in a northern New England low elevation coniferous forest. Water Air Soil Pollut. 1990, 52, 23–39. (3) Larssen, T.; Vogt, R. D.; Seip, H. M.; Furuberg, G.; Liao, B.; Xiao, J.; Xiong, J. Mechanisms for aluminum release in Chinese acid forest soils. Geoderma 1999, 91, 65–86. (4) Norton, S. A.; Fernandez, I. J.; Kahl, J. S.; Reinhardt, R. L. Acidification trends and the evolution of neutralization mechanisms through time at the Bear Brook Watershed in ME (BBWM), U.S.A. Water Air Soil Pollut. Focus 2004, 4, 289–310. (5) Kalbitz, K.; Geyer, S.; Geyer, W. A comparative characterization of dissolved organic matter by means of original aqueous samples and isolated humic substances. Chemosphere 2000, 40, 1305–1312. (6) Ohno, T.; Fernandez, I. J.; Hiradate, S.; Sherman, J. F. Effects of soil acidification and forest type on water soluble soil organic matter properties. Geoderma 2007, 140, 176–187. (7) Esteves da Silva, J. C. G.; Machado, A. A. S. C.; Oliveira, C. J. S.; Pinto, M. S. S. D. S. Fluorescence quenching of anthropogenic fulvic acids by Cu(II), Fe(III) and UO22+. Talanta 1998, 45, 1155– 1165. (8) Plaza, C.; Brunetti, G.; Senesi, N.; Polo, A. Molecular and quantitative analysis of metal ion binding to humic acids from sewage sludge and sludge-amended soils by fluorescence spectroscopy. Environ. Sci. Technol. 2006, 40, 917–923. (9) Luster, J.; Lloyd, T.; Sposito, G. Multi-wavelength molecular fluorescence spectrometery for quantitative characterization of copper(II) and aluminum(III) complexation by dissolved organic matter. Environ. Sci. Technol. 1996, 30, 1565–1574. (10) Bro, R. PARAFAC: Tutorial and applications. Chemom. Intell. Lab. Syst. 1997, 38, 149–171. (11) Andersen, C. M.; Bro, R. Practical aspects of PARAFAC modeling of fluorescence excitation-emission data. J. Chemom. 2003, 17, 200–215.

192

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

(12) Smilde, A. K.; Bro, R.; Geladi, P. Multi-way Analysis. Applications in the Chemical Sciences; Wiley: New York, 2004. (13) Ohno, T.; Bro, R. Dissolved organic matter characterization using multi-way spectral decomposition of fluorescence landscapes. Soil Sci. Soc. Am. J. 2006, 70, 2028–2037. (14) Norton, S.; Kahl, J.; Fernandez, I.; Haines, T.; Rustad, L.; Nodvin, S.; Scofield, J.; Strickland, T.; Erickson, H.; Wigington, P.; Lee, J. The Bear Brook Watershed, ME (BBWM), USA. Environ. Monit. Assess. 1999, 55, 7–51. (15) Bowles, E. C.; Antweiler, R. C.; MacCarthy, P. Acid-base titration and hydrolysis of Suwannee River fulvic acid;Open File Report 87557; U.S. Geological Survey: Denver, CO; 1989; pp 205–230. (16) Harshman, R. A. Foundations of the PARAFAC procedure: Models and conditions for an ‘explanatory’ multi-modal factor analysis ; UCLA Working Papers in Phonetics; 1970,16, 1–84. (17) Sidiropoulos, N. D.; Bro, R. On the uniqueness of multilinear decomposition of n-way arrays. J. Chemom. 2000, 14, 229–239. (18) Bro, R.; Kiers, H. A. L. A new efficient method for determining the number of components in PARAFAC models. J. Chemom. 2003, 17, 274–286. (19) Thygesen, L. G.; Rinnan, A.; Barsberg, S.; Møller, J. K. S. Stabilizing the PARAFAC decomposition of flourescence spectra by insertion of zeros outside the data area. Chemom. Intell. Lab. Syst. 2006, 81, 107–108. (20) Ryan, D. K.; Weber, J. H. Fluorescence quenching titration for determination of complexing capacities and stability constants of fulvic acid. Anal. Chem. 1982, 54, 986–990. (21) Dudal, Y.; Sevenier, G.; Dupont, L.; Guillon, E. Fate of the metalbinding soluble organic matter throughout a soil profile. Soil Sci. 2005, 170, 707–715. (22) Chin, Y. P.; Aiken, G.; O’Laughlin, E. Molecular weight, polydispersivity, and spectroscopic properties of aquatic humic substances. Environ. Sci. Technol. 1994, 28, 1853–1858. (23) Guo, M.; Chorover, J. Transport and fractionation of dissolved organic matter in soil columns. Soil Sci. 2003, 168, 108–118. (24) Coble, P. G.; Green, S. A.; Blough, N. V.; Gagosian, R. B. Characterisation of dissolved organic matter in the Black Sea by fluorescence spectroscopy. Nature 1990, 348, 432–435. (25) Stedmon, C. A.; Markager, S.; Bro, R. Tracing dissolved organic matter in aquatic environments using a new approach to fluorescence spectroscopy. Mar. Chem. 2003, 82, 239–254. (26) Shotyk, W.; Sposito, G. Fluorescence quenching and aluminum complexation by a chestnut leaf litter extract. Soil Sci. Soc. Am. J. 1988, 52, 1293–1297. (27) Cabaniss, S. E. Synchronous fluorescence spectra of metal-fulvic acid complexes. Environ. Sci. Technol. 1992, 26, 1133–1139. (28) Sierra, M. M. D.; Giovanela, M.; Parlante, E.; Soriano-Sierra, E. J. Fluorescence fingerprint of fulvic and humic acids from varied origins as viewed by single-scan and excitation/emission matrix techniques. Chemosphere 2005, 58, 715–733. (29) Ryan, D. K.; Weber, J. H. Fluorescence quenching titration for determination of complexing capacities and stability constants of fulvic acid. Anal. Chem. 1982, 54, 986–990. (30) Ohno, T.; Cronan, C. S. Comparative effects of ionic- and nonionic-resin purification treatments on the chemistry of dissolved organic matter. Int. J. Environ. Anal. Chem. 1997, 66, 119–136. (31) Saar, R. A.; Weber, J. H. Lead(II)-fulvic acid complexes: conditional stability constants, solubility and implications for lead (II) mobility. Anal. Chem. 1980, 52, 2095–2100. (32) Esteves da Silva, J. C. G.; Machado, A. A. S. C. Interaction of fulvic acids with Al(III) studied by self-modeling curve resolution of second-deriviative synchronous fluorescence spectra. Appl. Spectrosc. 1996, 50, 436–443. (33) Zhao, J.; Nelson, D. J. Fluorescence study of the interaction of Suwannee River fulvic acid with metal ions and Al3+ -metal ion competition. J. Inorg. Biochem. 2005, 99, 383–396. (34) Browne, B. A.; Driscoll, C. T. pH-dependent binding of aluminum by a fulvic acid. Environ. Sci. Technol. 1993, 27, 915–922. (35) Shotyk, W.; Sposito, G. Fluorescence quenching and aluminum complexation by a chestnut leaf litter extract. Soil Sci. Soc. Am. J. 1988, 52, 1293–1297. (36) Blaser, P.; Fluhler, H.; Polomski, J. Metal-binding properties of leaf litter extracts. 1. Discontinuous spectrophotometric titration with iron and copper. Soil Sci. Soc. Am. J. 1980, 44, 709–716. (37) Plaza, C.; Brunetti, G.; Senesi, N.; Polo, A. Molecular and quantitative analysis of metal ion binding to humic acids from sewage sludge and sludge-amended soils by fluorescence spectroscopy. Environ. Sci. Technol. 2006, 40, 917–923.

ES071855F