Environ. Sci. Technol. 2003, 37, 880-887
Using Selected Operational Descriptors to Examine the Heterogeneity within a Bulk Humic Substance JIN HUR† AND M A R K A . S C H L A U T M A N * ,‡,# Department of Environmental Engineering and Science, Clemson University, Anderson, South Carolina 29625-0919, Department of Agricultural and Biological Engineering, Clemson University, Clemson, South Carolina 29634-0357, and Department of Environmental Toxicology and the Clemson Institute of Environmental Toxicology, Clemson University, Pendleton, South Carolina 29670
The heterogeneity within a bulk humic substance (HS) was investigated for ultrafiltration (UF) fractions of purified Aldrich humic acid (PAHA) using specific ultraviolet absorbance (SUVA), weight- and number-average molecular weights (MWw and MWn), and organic carbon-normalized pyrene binding coefficients (Koc). As expected, variations in these selected operational descriptors were found for the PAHA UF fractions, but the variations were smaller than the ranges observed for a small subset of different aquatic and terrestrial HS. In general, correlations among the operational descriptors for the different HS failed to predict the correct trends for the PAHA UF fractions. More variation in Koc was observed for the different HS versus the PAHA UF fractions despite comparable molecular weight ranges, suggesting that the extent of pyrene binding is more strongly influenced by HS source and chemical structure than merely by physical structural features. Sizeexclusion chromatography (SEC) confirmed the structural stability of PAHA UF fractions, as demonstrated by a nearidentical mathematical reconstruction of the original bulk chromatogram from a weighted summation of the individual UF fraction chromatograms. An ideal mixture model accurately predicted some operational descriptors of PAHA properties (i.e., SUVA and MWw), but failed to predict the PAHA reactivity for pyrene binding. Our Koc results at pH 4 and 7 suggest that the relative contribution of two mechanisms, partitioning and specific interactions, may account for the interactions between HS and pyrene.
Introduction Humic substances (HS) are heterogeneous mixtures of acidic, randomly polymerized, polydisperse, high-molecular-weight organic macromolecules having widely different chemical functional groups and other physicochemical characteristics (1). HS comprise the largest proportion of naturally occurring organic matter (NOM) in soils and waters, and their bulk * Corresponding author phone: (864) 656-4059; fax: (864) 6560338; e-mail:
[email protected]. † Department of Environmental Engineering and Science. ‡ Department of Agricultural and Biological Engineering. # Department of Environmental Toxicology and the Clemson Institute of Environmental Toxicology. 880
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003
physicochemical characteristics vary depending on their source and location (1, 2). For example, terrestrial HS are known to have a higher carbon content, molecular weight (MW), and percentage of aromatic carbon than aquatic HS (2). Significant variations in color, spectroscopic features, and chemical composition of different HS have been attributed to different structural and other physicochemical characteristics according to their origin. Numerous studies have been conducted using different environmentally important properties, including ultraviolet (UV) absorptivity, extent of binding with organic chemicals, and sorption affinity to mineral surfaces, to characterize heterogeneities among different HS (3-6). Variations in physicochemical properties comparable to those observed for different HS have also been reported for components within a single bulk HS (7-11). For example, studies have shown that when a HS is separated into different fractions, those fractions often exhibit different structural features based on elemental analyses and spectroscopic measurements (7-11). It is impossible to adequately represent a bulk HS with a single chemical structure because of the inherent heterogeneity of its components. In fact, all of the physicochemical properties used to characterize a bulk HS are actually ensemble-averaged descriptions of its complex mixture of components. HS have been reported as having fractal geometry tendencies due to their heterogeneity, and fractal geometry-based concepts have been applied to studies of HS aggregation patterns for different solution chemistry conditions (12, 13) and for adsorption/partitioning of organic compounds to HS (14, 15). Comparing the fundamental nature of heterogeneity within a bulk HS to that observed for subsets of different HS has been a mostly overlooked research topic, despite its potential usefulness in understanding the fate and transport of contaminants as well as its likely importance to other areas of environmental research. One illustration of such a comparison would be to examine whether correlations existing among properties and reactivities for different HS are the same as those that might exist for the components within a bulk HS. For example, for several bulk aquatic HS and their sorption-fractionated components, Meier et al. (16) and Namjesnik-Dejanovic et al. (17) observed deviations from a trend reported between molar absorptivity and MW for different aquatic HS (3). Although they attributed the deviations to HS complexation by metal cations present from mineral dissolution (16, 17), this speculation neglects the possibility that correlations based on HS from different sources may not be appropriate for predicting the physicochemical characteristics and environmental behaviors of HS components and/or size fractions. To our knowledge, no research has been conducted to examine the validity of extending correlations between certain characteristics and relationships derived from different HS to the components/fractions within a single bulk HS. From a practical standpoint, such a study to examine organic carbon-normalized partition coefficients (Kocs) for HS components/fractions versus different HS would be important in providing better predictions for hydrophobic organic contaminant (HOC) transport in subsurface systems (18). This is because HOC mobilities depend on the ability of dissolved HS, which have been fractionated by adsorption processes with geosorbents, to bind the contaminants. The information would also be useful for subsurface remediation applications using HS (19) because subsurface fractionation of a bulk HS would result in temporal and spatial variations in Koc values and, thus, remediation effectiveness. 10.1021/es0260824 CCC: $25.00
2003 American Chemical Society Published on Web 01/31/2003
TABLE 1. Selected Chemical Characteristics of Four Different Humic Substancesa elemental analysis (mass %) (ash- and moisture-free basis) H N S O
HS type
C
SRFA SRHA SHA PAHA
52.4 54.2 58.1 57.0
4.3 4.1 3.7 4.2
0.7 1.2 4.1 1.4
0.4 0.8 0.4 3.5
42.2 39.0 34.1 33.9
ash (mass %) 0.5 3.2 0.9 7.5
atomic ratios O/C H/C 0.60 0.54 0.44 0.45
0.98 0.91 0.91 0.88
aromaticity (%) 24.8 42.0 50.0 56.0
a Elemental analysis data of PAHA from Karanfil (personal communication). Elemental analysis data and aromaticity of SHA, SRHA, and SRFA supplied by the IHSS. Aromaticity of PAHA from ref 51.
A related topic of importance arising from the study of heterogeneous HS components is whether reconstituted properties and reactivities summed from different components/fractions resemble those of the original bulk HS. If HS components/fractions within a bulk HS behave ideally (e.g., no intermolecular interactions), then the properties and reactivities of a bulk HS should be predictable from a weighted sum of the corresponding properties and reactivities of the individual fractions. Relevant examples of the use of ideal mixture concepts for complex mixtures can be found in several environmental research areas. For example, Lee et al. (20) successfully used Raoult’s law to predict the partitioning of polycyclic aromatic hydrocarbons (PAHs) between water and coal tars (i.e., complex mixtures of hydrocarbonbased materials). Weber et al. (21) advocated the use of a composite distributed reactivity model to characterize the sorption of organic solutes to soils and sediments based on their individual component sorption reactivities. Conversely, Leenheer et al. (22) reported that the density of a fulvic acid measured in solvents of different polarities was concentration-dependent, and speculated that this nonideal behavior resulted from a combination of intramolecular and intermolecular interactions that varied with HS concentration. In studies more closely related to the present work, Jones and Tiller (23) divided HS into two separate fractions (i.e., mineral adsorbed and nonadsorbed fractions) and tested the applicability of using mass balances and ideal mixture concepts to predict Koc values for the adsorbed HS fractions. In general, they found that the predicted Koc values for adsorbed HS fractions did not agree with the observed values. Because the presence of mineral surfaces creates a more complex and “macroscopically heterogeneous” system, however, the validity of utilizing ideal mixture approaches to characterize HS properties and reactivities is more easily evaluated in “macroscopically homogeneous” systems using dissolved HS fractions. For example, Kitis et al. (24) observed that fractionating dissolved organic matter (DOM) from two different source waters using resin adsorption or ultrafiltration (UF) preserved the specific ultraviolet absorbance (SUVA) and disinfection byproduct (DBP) reactivities of the source waters, evidence that the DOM components/fractions were behaving ideally with respect to those two measured parameters. The objectives of this study were to (1) compare the heterogeneity among dissolved UF fractions of a bulk HS versus a small subset of different dissolved aquatic and terrestrial HS by measuring selected operational descriptors (i.e., SUVA, MW, and pyrene Koc), (2) examine correlations among these descriptors for the different HS versus UF fractions, and (3) determine if the UF fractions behave ideally with respect to the measured descriptors.
Experimental Section Materials. Concentrated pyrene (Fluka, 99.5+%) stock solutions were prepared in methanol (EM Science, HPLC grade) and stored in the dark at 4 °C in amber borosilicate bottles. A small subset of two aquatic and two terrestrial HS were
used to represent a range of different HS properties and reactivities. Soil humic acid (SHA), Suwannee River humic acid (SRHA), and Suwannee River fulvic acid (SRFA) were obtained from the International Humic Substances Society (IHSS) and used without further purification. Purified Aldrich humic acid (PAHA), a terrestrial peat humic acid originally obtained in powdered form from Aldrich Chemicals, was purified by repeated pH adjustment, precipitation, and centrifugation to remove ash, humin, and fulvic acid following a reported procedure (25). Briefly, the Aldrich humic acid powder was dissolved in distilled, deionized water (DDW) and then centrifuged at 6000 rpm for 15 min so that the precipitated ash and humin could be discarded. The pH was then adjusted to below 2 with 5 N HCl solution (J. T. Baker), and the solution was centrifuged again so that the remaining solution (presumably containing fulvic acid) could be discarded. The precipitated humic acid fraction was subsequently dissolved with DDW and the pH was adjusted back to ∼8.0 with 5 N NaOH solution (J. T. Baker). The purification steps were repeated until no further removal of fulvic acid occurred, as measured by UV absorbance at 254 nm. All IHSS HS were dissolved in DDW and stored at 4 °C at concentrations of ∼5 g C/L after first adjusting the pH to ∼7 with 0.1 N HCl or NaOH. No precipitate was visually observed for any of the IHSS HS stock solutions. General chemical characteristics of the four different HS are shown in Table 1. Analytical Methods. A total organic carbon analyzer (Shimadzu model 5050) was used to quantify the dissolved organic carbon (DOC) concentrations of aqueous HS samples. External DOC standards were prepared using potassium hydrogen phthalate. The DOC of HS samples was obtained by subtracting the inorganic carbon from the total carbon measurement. In all cases, sample DOC concentrations were kept well above the detection limit (∼ 0.5 mg C/L) in this study. UV absorbance of HS samples was determined at 254 nm with a spectrophotometer (Beckman model DU640). SUVA values were determined for each HS sample by linear regression of UV absorbance versus DOC concentration over an appropriate concentration range. Because of the possible dependency of UV absorbance on solution chemistry, pH values and the ionic strength of all samples were kept constant (pH ∼7.0, ionic strength of 0.1 M NaCl). The spectrophotometer was also used to measure absorbance at the fluorescence excitation and emission wavelengths of pyrene to make inner-filter corrections (26). Concentrations of freely dissolved pyrene were quantified with a luminescence spectrophotometer (Perkin-Elmer, LS-5B) at the excitation/ emission wavelengths of 336/373 nm using slits set for bandwidths of 3/5 nm. Relative precisions of 1% and 3% were routinely obtained for absorbance/fluorescence and DOC measurements, respectively. Size-exclusion chromatography (SEC) was used to determine the apparent MWs of HS samples, generally following the methodology proposed by Zhou et al. (27). The SEC system consisted of a high-pressure liquid chromatography pump (GP50 gradient pump, Dionex), PDA-100 photodiode array detector (Dionex), LC25 chromatography oven (Dionex), and VOL. 37, NO. 5, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
881
TABLE 2. SUVAa, Pyrene Kocb, and Molecular Weightsc of Different HS and PAHA UF Fractionsd HS
SUVA (L/mg C-m)
SRFA SRHA SHA PAHA 100 K
3.65 (0.02) 6.19 (0.01) 7.04 (0.13) 7.71 (0.17) 8.43 (0.08) 8.79 (0.04) 8.54 (0.04) 8.39 (0.04) 6.73 (0.03)
pyrene Koc × 10-5 (L/kg) pH 4 pH 7 0.23 (0.02) 0.53 (0.05) 2.73 (0.27) 2.45 (0.16) 1.85 (0.09) 1.35 (0.09) 1.76 (0.17) 2.85 (0.25) 3.75 (0.26)
0.25 (0.01) 0.49 (0.04) 2.30 (0.14) 2.05 (0.10) 1.20 (0.11) 0.94 (0.07) 1.22 (0.10) 1.51 (0.04) 2.53 (0.17)
MWw (kDa)
MWn (kDa)
polydispersity (Mw/Mn)
2.18 (0.02) 3.67 (0.03) 7.76 (0.07) 6.06 (0.05) 2.67 (0.07) 1.87 (0.11) 2.77 (0.06) 5.80 (0.19) 10.08 (0.15)
1.69 (0.03) 2.21 (0.02) 3.65 (0.05) 2.25 (0.06) 0.61 (0.02) 1.39 (0.06) 2.15 (0.01) 4.18 (0.12) 6.91 (0.06)
1.29 (0.03) 1.66 (0.03) 2.13 (0.06) 2.09 (0.05) 4.37 (0.04) 1.35 (0.01) 1.29 (0.02) 1.32 (0.01) 1.46 (0.01)
a SUVA values at 254 nm in solutions of pH 7 and 0.1 M NaCl. b K c oc values in solutions of 0.1 M NaCl. Mean values of three independent measurements. Nominal DOC concentration of ∼23 mg C/L except for the 0.99). The uncertainty of average MW by SEC was found to be less than 5% in repeated measurements. MWw and number-average MWs (MWn) of HS from the SEC chromatogram were determined using the following equations (30). N
MWw )
N
∑(h ‚MW )/∑h i
i
i)1
i
(1)
i)1
and N
MWn )
N
∑ ∑(h /MW ) hi/
i)1
i
i
(2)
i)1
where hi is the sample UV absorbance and MWi is the equivalent PSS molecular weight, both corresponding to an appropriate retention time, i. Ultrafiltration of PAHA. Fractionation of PAHA was performed by UF following the procedures described by Kitis et al. (24). Briefly, a semi-batch UF system (Amicon) consisting of a reservoir, peristaltic pump, and hydrophilic, neutral, cellulosic-type spiral-wound membrane cartridges having nominal molecular weight cutoffs (MWCO) of 3 K, 10 K, 30 K, and 100 K was used. The pressure drop (20 psig) and recirculation rate (0.833 L/min) were maintained constant throughout the fractionation process. The starting concentration of PAHA was ∼200 mg C/L in a 0.01 M NaCl solution. DOC concentrations and pH were monitored throughout 882
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 5, 2003
the process for all permeate samples. The pH was relatively stable during the operation, ranging from 6.9 to 7.4. Mass balance on organic carbon revealed that system losses were minor (
