Environ. Sci. Technol. 1993,27, 961-969
Effects of Aqueous Chemistry on the Binding of Polycyclic Aromatic Hydrocarbons by Dissolved Humic Materials Mark A. Schlautman'~tand James J. Morgan
Environmental Engineering Science, California Institute of Technology, Pasadena, California 9 1125 The influence of solution chemistry on the binding of three polycyclic aromatic hydrocarbons (PAHs) by well-characterized humic material (Suwannee River humic and fulvic acid) was examined by using fluorescencequenching techniques. Our experiments show that binding is complete within 3 min and that the fluorescence of PAH compounds associated with the humic substances is fully quenched as evidenced by quantum yields which approached zero for all systems. These observations validate the use of fluorescencequenching in determining partition coefficients. In NaCl solutions, the binding of PAHs by Suwannee River humic material generally decreased with increases in pH (constant ionic strength) and generally decreased with increasing ionic strength (fixed pH). The presence of Ca2+yielded mixed results: at neutral to high pH values, it generally increased the binding of PAHs relative to that in NaCl solutions, while at low pH it generally had little effect on the binding. From the results of this study, it is hypothesized that the binding of a particular PAH compound by Suwannee River humic substances depends not only on the hydrophobicity of the PAH solute but also on the size of the solute molecule and its ability to fit into hydrophobic cavities in humic and fulvic material. This hypothesis is supported by the experimental observations above, as well as the failure of a Flory-Huggins partitioning (i.e., dissolution) model to consistently characterize the hydrophobic environment of the humic substances. Introduction
The distribution or partitioning of nonionic hydrophobic organic compounds (HOCs) between water and surface soils or sediments has been shown to depend primarily on the hydrophobicity of the compound and the fraction of organic carbon (foe) in the sorbent (1-4). Partitioning results because of the "hydrophobic interaction", a combination of relatively small van der Waals bonding forces and a substantial thermodynamic gradient which drives the organic molecules out of aqueous solution (5). The thermodynamic driving force is the increase in entropy which results from breakdown of the highly structured coordination shell of water molecules surrounding hydrophobic solutes. The hydrophobic interaction has been discussed in detail by Tanford (6) and Israelachvili (7). The importance of the organic carbon content of soils and sediments suggests that various components of organic material in solution may also bind nonionic HOCs. Recently, much work has been focused in this area (e.g., refs 8-14). The collective results from these studies suggest that the binding of hydrophobic organic pollutants by dissolved organicmaterial (DOM)depends on the chemical and structural characteristics of the DOM and may depend Present address: Department of Civil and Environmental Engineering, The University of Michigan, Ann Arbor, MI 48109+
2125. 0013-936X/93/0927-0961$04.00/0
0 1993 American Chemical Society
on the aqueous chemistry of the system. In natural water systems, aquatic humic substances are the largest fraction of DOM, constituting 40-60% of the dissolved organic carbon present (15). The aqueous chemistry environment of natural water systems controls to a large extent the polarity and the conformation of aquatic humic substances (16-19). Intermolecular interactions between humic molecules and intramolecular interactions between a molecule's functional groups change the chemical and physical properties of humic material in different aqueous environments. These interactions control the self-association, and ultimately the size, shape, and polarity, of humic substances. Both nonpolar and polar interactions are thought to be responsible for the self-aggregation of humic material, although nonpolar interactions are thought to be insignificant for the humic and fulvic acids studied here (19). However, because both polar intermolecular and intramolecular interactions increase the nonpolar character of humic substances, increased binding of hydrophobic contaminants by humic material may be observed as a result of these interactions. Molecular interactions of DOM are dependent on temperature, pH, ionic strength, and type of ions in solution, as well as other chemical parameters (19). For intermolecular interactions, the concentration of DOM is also important. Because the above parameters are significant for the self-association of humic material, they should also be important for organic pollutant binding by humic substances. Few studies, however, have been made of the effects of aqueous chemistry on the binding of nonpolar organic compounds by DOM. Carter and Suffet (8)observed a decrease in the binding of DDT with humic acid when the pH was raised from 6 to 9.2. At pH 8.3, increased binding was observed in the presence of Ca2+ and also with increasing ionic strength. Kile and Chiou (20) reported decreases in the binding of DDT and two PCBs by a soil humic acid and Suwannee River fulvic acid at pH 8.5 relative to pH 6.5. At pH values below 6.5, however, the binding by fulvic acid did not show any dependence on pH. Traina et al. (13)studied the influence of pH, ionic strength, and cation type on the binding of naphthalene by a poorly characterized muck soil organic matter extract and observed neither pH (1.5-7.3) nor ionic strength (0.05-0.5 M) effects. The presence of AP+, however, appeared to decrease the binding of naphthalene relative to Na+ and Ca2+,although simultaneous variations in pH complicate interpretation of their results. The objective of this research was to investigate systematically the effects of pH, ionic strength, and the presence of bivalent cations on the binding of polycyclic aromatic hydrocarbons (PAHs) by well-characterized aquatic humic substances in order to elucidate the mechanisms by which they interact. Because PAHs are nonpolar, nonionogenic compounds, they minimize adsorption mechanisms other than the effect of hydrophobic interactions in the binding reaction. The binding of PAHs Envlron. Sci. Technol., Vol. 27, No. 5, 1993
@SI
Table I. Chemical and Physical Properties of Polycyclic Aromatic Hydrocarbons (PAHs)
anthracene pyrene perylene
MW
molar ~ 0 1mL ,~
mp, "C
solubility,b nM
KO,
KoCc( K d
Kocd(solubility)
178.23 202.26 252.32
160.2 171.5 202
216
409.6 667.5 1.585
3.5 x 104 e 1.5 x 105 e 3.2 X lo6 [
2.2 x 104 9.2 x 104 2.0 X lo6
1.8 x 104 5.0 x 104
K,,, (lit)
9 000-500 OOOK 150 29 000-550 000" 278 7.8 x 105 14 000-2 400 000' a Reference 25. Reference 26. log KO,= log KO, - 0.21 (1). log KO,= -0.921 log X,- 0.00953 (mp - 25) - 1.405 (36). e Reference 1. [Reference 27. 8 References 9-11. References 1. 2, 11. 12. and 37. L Reference 14.
by humic material was measured with fluorescence quenching methods (11,14). The technique is based on the observation that PAHs fluoresce in aqueous solution but not when associated with DOM. Thus, the fraction of a PAH associated with humic substances can be determined without separating humic material from the aqueous phase. The ability to measure concentrations in situ is ideal because separation difficulties may introduce artifacts in determination of partition coefficients (11,2124). Experimental Section
Materials. The PAHs used were anthracene (Aldrich, 98+% pure), pyrene (Aldrich, 99+ % pure), and perylene (Aldrich, 99+ % pure). Relevant chemical and physical properties of the PAHs are listed in Table I. Rhodamine 110 (lasergrade) was obtained from Kodak. Fluorescence measurements were made on a Shimadzu Model RF-540 recording spectrofluorophotometer. The excitationlemission wavelengths (nminm) used were 2501380, 3341390, 4321470, and 4961520 for anthracene, pyrene, perylene, and rhodamine 110, respectively. The slits were set for bandwidths of 5 nm (anthracene, pyrene) or 10 nm (perylene, rhodamine 110) on both the excitation and emission monochromators. Absorbance measurements were made a t the above wavelengths on a Hewlett Packard 8451A diode array spectrophotometer to correct for the inner-filter effect (11, 28, 29). The largest correction factors were obtained with anthracene-humic acid systems because of the high absorbance a t 250 nm by humic acid. However, the correction factors never exceeded the recommended maximum value of 3.0. Correction factors for pyrene were less than 2.0, and those for perylene approached 1.0. The purity of the PAHs and rhodamine 110was checked by comparing the fluorescence excitation spectra with absorbance spectra (30); because all compounds showed no impurities, they were used without further purification. Concentrated stock solutions of anthracene and pyrene and a combined solution of perylene and rhodamine 110 were prepared in methanol (Fisher, spectroanalyzed). The stock solutions were stored in the dark a t 4 "C in amber borosilicate glass bottles to prevent photodegradation and/ or volatilization. Aliquots of stock solutions were allowed to thermally equilibrate (in the dark) a t room temperature for a t least 24 h before an experiment. The use of methanol as a carrier solvent for the PAHs did not significantly affect the results of this study, as evidenced by calculations based on solvophobic theory (30). Well-characterized humic materials were obtained from the International Humic Substances Society (IHSS) and were used without further purification. Isolation procedures and characterization of these materials have been previously reported (31). Because the humic materials as received are very hygroscopic, they were stored in a desiccator until needed. Concentrated stock solutions were prepared by dissolving weighed amounts of the humic 962
Environ. Sci. Technol., Vol. 27, No. 5, 1993
material in deionized, distilled water. The solutions were mixed and allowed to stand in the dark at room temperature for at least 24 h. The humic stock solutions were filtered through a prewashed 0.2-pm pore polycarbonate Nuclepore filter to remove possible particle contamination and were then stored in the dark at 4 OC in amber borosilicate glass bottles. All other reagents were of analytical grade or better and were used without further treatment. Additional experimental details have been described by Schlautman (30). Procedure. The fluorescence quenching method used for anthracene and pyrene was adapted from Gauthier et al. (11). A typical experiment consisted of pipeting 3 mL of a salt solution of known pH and ionic strength into a fluorescence cell containing a Teflon-covered micro stir bar and recording the absorbance and fluorescence intensity a t the appropriate wavelengths. The sample was then spiked with 10 p L of anthracene or pyrene stock solution and stirred for 10 min before recording an initial fluorescence intensity and absorbance. Nominal PAH concentrations for anthracene and pyrene were 0.23 and 0.41 pM, respectively. Preliminary results showed that the solutions were completely mixed in less than 1 min and that neither anthracene nor pyrene adsorption to cell walls andlor the stir bar could be detected over the time period required for a complete experiment. The anthracene (or pyrene) solutions were titrated with humic or fulvic acid (previously equilibrated a t the same pH and ionic strength) until a minimum of five 10-pL aliquots were added, resulting in a concentration range of humic substances typically 0-25 mg/L. During the course of an experiment, each PAH solution was kept in the dark while mixing and the shutter of the spectrofluorophotometer was opened only for the actual intensity measurements. The cumulative time of PAH exposure to UV radiation was less than 5 min for each experiment; no PAH photodegradation was observed with controls for these conditions. In order to correct for the background fluorescence contribution of the solution components, a blank sample spiked only with methanol was titrated and analyzed using the same process as for the PAH sample. For very hydrophobic PAHs such as perylene, the sorption to glassware is substantial. Therefore, the fluorescence quenching method used for perylene was adapted from Backhus and Gschwend (14). A typical experiment consisted of pipeting 3 mL of a sample of known pH, ionic strength, and humiclfulvic acid concentration (typically 0-10 mg/L) into a fluorescence cell containing a Teflon-covered micro stir bar and recording the absorbance and fluorescence intensities a t the excitation/emission wavelengths for perylene and rhodamine 110. These values reflect the background intensities of solution components. The sample was then spiked with 10 pL of the combined perylene-rhodamine 110 stock solution and was immediately placed in the covered stirring module of the spectrophotometer, Initial perylene concentrations were nominally 1.6 nM.
After 3 min, the cell was placed in the spectrofluorophotometer and a fluorescence measurement for perylene was made. The sample was then returned to the spectrophotometer for an absorbance measurement and further mixing. This procedure was repeated for a total elapsed time of 40 min. Preliminary experiments showed that although perylene adsorption to cell walls was significant during a single experiment, sorption of perylene by the stir bar was small over this short time period. When the rate of change of perylene fluorescenceintensity decreased sufficiently (after 20 min), fluorescence measurements were made for rhodamine 110 as well. No losses of rhodamine 110 were detected during an experiment. In order ;to account for the slight variation in spiking volume among the different samples, the corrected perylene fluorescence intensities were normalized by the mean of the corrected rhodamine 110 fluorescence intensities (14). Data Treatment. (A) Anthraceneand Pyrene. For PAHs which do not sorb to glassware/experimental apparatus (more precisely, which do not sorb appreciably during the time scale of an experiment), Gauthier et al. (11) showed that the Stern-Volmer equation for static fluorescence quenching of a PAH compound by DOM can be written (1) [PAH,l/[PAH,l = Fo/F = (1+ K,,[OCl) where [PA&] is the total PAH concentration, [PA&] is the dissolved PAH concentration, FO and F are the fluorescence intensities in the absence and presence of DOM, respectively, [OCI is the organic carbon concentration of DOM, and KO,is the organic carbon-normalized partition coefficient. Therefore, in a plot of the fluorescence ratio versus concentration of DOM, the partition coefficient is determined from the resulting slope. The treatment above assumes that the fluorescence of PAH molecules associated with DOM is totally quenched (Le., that the quantum yield, 4, of the complex is zero). Backhus and Gschwend (14) showed that if PAH fluorescence is not totally quenched upon association with DOM, the observed fluorescence ratio will be described by
FIFO=
1+ 4K,,[OCl 1 + K,,[OCI
With this formulation, they showed that as [OCI =+ m, the normalized fluorescence asymptotically approaches 4. From a plot of F/Fo versus [OCI, KO,equals the value of [OCI-l at the point where FIFO = (1 4)/2. (B) Perylene. An alternative approach must be used for very hydrophobic PAHs because decreases in observed fluorescence will result not only from DOM quenching but also from sorption of the PAH to glassware. Assuming first-order kinetics for the surface adsorption reaction, Backhus and Gschwend (14)derived the following equation to describe fluorescence versus time:
+
where Fo* = [PAHTI - [PAH-OCI at time zero, and k, and k-, are the first-order forward and reverse rate constants for wall adsorption. By fitting eq 3 to data and back-extrapolating to t = 0, they obtained the fluorescence values needed for determining partition coefficients with eq 2.
'
1 0O 5
6
,
/
7 mg/L HA
1. O ' l j
0
A
5 BO 4.36 2 92
0
A
+
1.46 0 73
0 37
101
*
o
1 00 l 0
5
time 10 (rnin)
I 1s
20
Figure 1. Kinetics of anthracene binding by humic acid in 0.1 M NaCl solution at pH 5. Anthracene concentratlon was 21.5 pglL (0.12 pM).
Results and Discussion Methods Verification. (A) Kinetics of Binding. The binding of anthracene and pyrene by humic substances was observed to be very fast (Figure 1). Quite often, the quenching reactions appeared to be at equilibrium before the first fluorescence measurement could be taken, a time interval of approximately 15-20 s after the addition of humic material. Complete quenching never took longer than 3 min for any of the experiments (30). The reactions were assumed to be equilibrated when no further change in fluorescence intensity was observed. However,because the possibility of additional binding to DOM over much longer periods of time cannot be excluded by the data of this study, the state should be considered an apparent equilibrium only. Fast equilibration times for the association of PAHs with DOM, however, have been reported by other researchers (10, 11, 14). Although the kinetic experiments showed that the binding of anthracene and pyrene was always complete within 3 min, the solutions were stirred for 10 min after each addition of humic material before the measurements were recorded. Using the fluorescence quenching method of Backhus and Gschwend (14),the rate at which perylene partitions between water and DOM is observed indirectly and, in a sense, must be known a priori in order for the loss of perylene from aqueous solution to be resolved into perylene binding by DOM versus the adsorption of perylene to glassware. A key requirement of the method is that the rate of PAH binding to DOM is fast relative to the rate of PAH adsorption to the cell walls. By allowing the PAHDOM reaction to reach equilibrium before the initial measurement is taken, subsequent decreases in fluorescence can then be attributed to adsorption of the PAH to cell walls. In the presence and absence of DOM, measurements of the perylene fluorescence intensity decreased exponentially with time, in accordance with eq 3. The distinct and parallel curves obtained for various concentrations of DOM (e.g., Figure 2 for perylene-humic acid) indicate that the binding of perylene by DOM was complete within 3 min, the time allotted before recording the first measurement. (B) Efficiency of Static Quenching. On the basis of diffusion calculations and temperature studies, fluorescence quenching of PAH compounds by DOM has been hypothesized to be a static quenching process (11, 13). Mechanisms of energy transfer from an excited organic molecule (donor) to a static quenching species (acceptor) have been discussed previously (28,32-341, while a more Environ. Sci. Technol., Vol. 27, No. 5, 1993 983
-
0
3lonk
A
0 3 0 7 mg/L HA 1 5 3 7 rng/L i P
-0
20
4,-
si)
40
30
Time (min)
Flgure 2. Adsorption of peryleneto cell walls versus time in the presence and absence of humic acid. Solutions were at pH 4 and 0.1 M NaCI. Nominal perylene concentratlon for each sample was 0.4 pglL (1.6 nM). 1 .0
0.8
Koc = 8.87 x l o 5 mL/g-C @ = - 3 x 10-11
0.6 LL"
20.4
0.2
0.0 0
2
4
6
8
[Humic A c i d ] , mg-C/L
Flgute 3. Determination of KO,and $ for perylene binding by humic acid In 0.1 M NaCl solutions at pH 4. Nominal perylene concentration for each sample was 0.4 pglL (1.6 nM).
thorough treatment has been developed by Turro (3.5). These concepts have been summarized for systems in which humic substances bind PAHs (30). Use of Stern-Volmer analysis (eq 1)to calculate partition coefficients requires that only the dissolved PAH species is quantified by measured fluorescence; the fluorescence of PAH molecules associated with DOM must be totally quenched (i.e., 4 = 0). Few published results are available for the quantum yields of PAH-DOM complexes. Although a commercial humic acid was shown to fully quench the fluorescence of perylene upon binding, bovine serum albumin quenched only 42% of the associated perylene fluorescence ( 4 = 0.58) (14). The observations of Gauthier et al. (11, 12) suggest that values of zero for 4 are found for a wide variety of humic and fulvic acids. However, as chemical conditions of a system change, the assumption of complete fluorescence quenching may not be valid and must be verified. Perylene fluorescence values (after accounting for wall losses) were analyzed using eq 2 to determine partition coefficients and quantum yields for the perylene-DOM complexes. Results for a typical experiment are shown in Figure 3, in which the normalized fluorescence intensities obtained from back-extrapolation to t = 0 are plotted as a function of humic acid concentration. The curve 084
Environ. Sci. Technol., Vol. 27, No. 5, 1993
obtained for this particular perylene-DOM system asymptotically approaches a quantum yield of zero as [OCI m. The quantum yield of perylene approached zero for all solution chemistry systems examined when it was associated with humic or fulvic acid. Partition coefficients for anthracene and pyrene were determined from experimental data using both eqs 1and 2. The two equations gave similar KO,values, suggesting that the second fitting parameter ( 4 )in eq 2 did not greatly affect the results. All quantum yields for the bound anthracene and pyrene compounds were not significantly different from zero. However, the limited range of FIFO is not a rigorous test of 4 = 0 for the bound anthracene and pyrene molecules. Because a large amount of perylene binds to humic and fulvic acid, a wider range in FIFOis observed, and the quantum yield calculation is therefore more credible. Aqueous Chemistry Effects on PAH Binding by Humic Material. The influence of solution chemistry on the binding of perylene is summarized in Figure 4. Experiments were performed only at pH 4 with fulvic acid because of its lower ability to bind perylene relative to humic acid. With increasing pH at fixed ionic strengths, a decrease in the binding of perylene by humic acid was observed in all systems (Figure 4a); KO,values decreased 50-95% when the pH was raised from 4 to 10, with the biggest decrease occurring at the highest ionic strength. Increases in ionic strength caused a decrease in all partition coefficients regardless of pH (Figure 4b). An increase in ionic strength from 1mM NaCl to 0.1 M NaCl decreased the partition coefficient with humic acid 20-85 % for fixed pH values, with the biggest decrease observed at pH 10. The ability of fulvic acid to bind perylene showed a similar dependence (20% decrease) on increasing ionic strength at pH 4; however, the partition coefficient with fulvic acid was almost an order of magnitude smaller than with humic acid for identical solution conditions. The presence of 1 mM Ca2+had relatively small effects on perylene binding by either humic or fulvic acid when compared to NaCl solutions of the same ionic strength. The presence of Ca2+ at pH 7 and 10 decreased the ability of humic acid to bind perylene, while both humic and fulvic acid showed a slight increase in the partition coefficient at pH 4. Representative results for pyrene and anthracene are shown in Figures 5 and 6, respectively. A t constant pH, increases in NaCl concentration resulted in decreases in the binding of pyrene by humic acid (Figure 5 ) ;the decrease in KO,ranged from 30 to 40 % as the ionic strength increased from 1 mM NaCl to 0.1 M NaC1. A t constant ionic strength, an increase in pH from 4 to 7 resulted in decreases in the partition coefficients for all solutions. A further increase in pH from 7 to 10, however, had relatively little effect on the amount of pyrene binding by humic acid in NaCl solutions, The presence of 1mM Ca2+in 0.1 M total ionic strength solutions yielded results opposite to those observed for perylene; however, only a t pH 7 was the effect significant. Fulvic acid showed smaller, but similar, trends in the binding of pyrene in NaCl solutions. For example, partition coefficients decreased approximately 20 % with an increase in ionic strength from 1 mM NaCl to 0.1 M NaCl (Table 111). A major exception for fulvic acid relative to humic acid was the enhanced binding of pyrene at pH 10, as well as pH 7, in the presence of 1 mM Ca2+. The binding of anthracene by fulvic acid showed even smaller trends than those observed for pyrene: KO,values decreased approximately 10% with an increase in ionic
1.2E+06
n
0
8.OE+05
I
CT
'=. E
0-0 0.001M, HA
W
) .0.01M,
HA k A 0.1M. HA H Colt, HA 0 0.001M, FA 0 0.01M, FA A O . l M , FA 0 Col*, FA
y8 4.OE+05
O.OE+OO
I
I
I
I
I
I
I
4
5
6
7
0
9
10
PH
n
0
I
0,
'=. E
Q-0 pH 4, HA W pH 7, HA ad pH 10, HA Q4 pH 4, FA 0 Co**, pH 4, HA Co2+, pH 7, HA A Ca", pH 10, HA Co2*, pH 4, FA
W
8
Y
+
ionic Strength (M) Flgure 4. Binding of perylene by humic and fulvic acid. Data for 1 mM Ca2+ were at a total ionic strength of 0.1 M. (a) KO, versus pH. Solid symbols connected by lines refer to experimentswith humic acid. Single (open) points at pH 4 refer to experiments with fulvic acid. (b) KO,versus ionic strength. Open symbols connected by lines refer to experiments in NaCl solutions. Single (solid) points at 0.1 M ionic strength refer to experiments with 1 mM Ca2+. 4 OE+04
W pH 4 k z ? pH 7 Q€ pH 10 Co'*. pH 4 A Ca'*, pH 7
30E+04
0
Ca",
pH 10
,--. 0 3 OE+04-
Q€ W
pH 10
Co".
pH 4
Co*',
PH 10
CT
'=-
E
E
v
y'
v
2.OE+04-
1 OE+04
y8 20E+04
~
lo-'
I
, , ,
,,, 10-3
,
, ,,,,
,
, ,
, , ,,,/
lo-'
lo-'
,
, , ,,
,,,I 100
Ionic Strength (M)
10-4
lo-'
10-2
lo-'
1 00
Ionic S t r e n g t h (M)
Figure 5. Binding of pyrene by humic acid. Data for 1 mM Ca2+were at a total Ionic strength of 0.1 M. Open symbols connected by lines refer to NaCl solutions. Single (solid) points refer to experimentswith 1 mM Ca2+ at the appropriate pH values.
Flgure 8. Binding of anthracene by humic acid. Data for 1 mM Ca2+ were at a total ionlc strength of 0.1 M. Open symbols connected by lines refer to NaCl solutions. Single (solid) polnts refer to experiments with 1 mM Ca2+ at the appropriate pH values.
strength from 1mM NaCl to 0.1 M NaCl while variations with pH were relatively insignificant (Table 111). Anthracene binding by humic acid was more sensitive to changes in solution chemistry relative to fulvic acid. For
NaCl solutions at pH 7 and 10,an increase in ionic strength resulted in a decrease in the binding of anthracene by humic acid while the opposite trend was observed for the pH 4 NaCl solutions (Figure 6). The effects of 1mM Ca2+ Environ. Scl. Technol., VoI. 27, No. 5, 1993 g66
Table 11. Experimental Partition Coefficients (fSD) with Humic Acid and Corresponding Solubility Parameters for Humic Acid#-c anthracene (6 = 9.9) K,, x 10-4 0.001 M PH 4 PH 7 pH 10 0.01 M PH 4 PH 7 pH 10 0.1 M PH 4 PH 7 pH 10
Ca2+ PH 4 PH 7 pH 10
wrene (6 = 10.2)
perylene (6 = 10.7)
K,,, x 10-4
6,
K,, x 10-5
6,
2.37 (0.08) 2.64 (0.05) 2.45 (0.07)
11.7 11.6 11.7
3.15 (0.07) 2.33 (0.04) 2.15 (0.03)
13.2 13.3 13.4
11.28 (0.65) 9.58 (0.30) 5.10 (0.14)
12.5 12.6 13.0
2.60 (0.10) 2.37 (0.09) 2.00 (0.12)
11.6 11.7 11.9
2.23 (0.04) 1.58 (0.06) 1.76 (0.01)
13.4 13.5 13.5
10.31 (0.62) 8.99 (0.28) 1.95 (0.06)
12.5 12.7 13.6
3.19 (0.27) 1.74 (0.18) 1.85 (0.24)
11.5 12.1 12.0
2.26 (0.18) 1.41 (0.14) 1.41 (0.07)
13.4 13.6 13.6
8.87 (0.46) 4.69 (0.14) 0.72 (0.05)
12.7 13.1 14.1
2.28 (0.03) 2.51 (0.09) 2.61 (0.13)
11.8 11.7 11.7
2.13 (0.08) 1.94 (0.04) 1.40 (0.06)
13.4 13.5 13.6
9.29 (0.11) 4.10 (0.09) 0.60 (0.01)
12.7 13.2 14.1
a Units of KO,are in mL/g of C. Numbers in parentheses, f standard deviation. * Anthracene solubility parameter from ref 42. Pyrene and perylene solubility parameters were calculated from Small's group molar attraction constants (42) using molar volumes listed in Table I. Units of 6 and 6, are in (cal/mL)1/2. V , = 18.07 mL/mol, R = 1.98 cal/mol.deg, T = 296 K. 1 mM Ca2+and 0.1 M total ionic strength.
Table 111. Experimental Partition Coefficients (fSD) with Fulvic Acid and Corresponding Solubility Parameters for Fulvic Acide-c anthracene (6 = 9.9) K,, x 10-4 6,
K,, x 10-4
6,
1.85 (0.06) 1.51 (0.06) 1.74 (0.05)
12.1 12.2 12.1
1.54 (0.02) 0.77 (0.03) 0.83 (0.03)
13.6 14.0 13.9
1.49 (0.02)
13.8
ND" ND
ND ND
pyrene (6 = 10.2)
perylene (6 = 10.7) K,, x 10-5
6,
0.001 M PH 4 PH 7 pH 10 0.01 M PH 4 PH 7 pH 10 0.1 M PH 4 PH 7 pH 10
1.93 (0.07) 1.41 (0.10) 1.67 (0.11)
12.0 12.3 12.2
1.42 (0.03) 0.60 (0.02) 0.73 (0.03)
13.7 14.1 14.0
1.37 (0.01)
13.8
ND ND
ND ND
1.71 (0.05) 1.31 (0.12) 1.58 (0.09)
12.2 12.4 12.2
1.20 (0.02) 0.65 (0.02) 0.61 (0.03)
13.8 14.1 14.1
1.19 (0.02)
13.9
ND ND
ND ND
Ca2+ PH 4 PH 7 pH 10
1.62 (0.06) 1.55 (0.08) 1.75 (0.06)
12.2 12.3 12.2
1.19 (0.07) 0.89 (0.01) 1.14 (0.02)
13.8 13.9 13.8
1.28 (0.02)
13.9
ND ND
ND ND
Units of KO,are in mL/g of C. Numbers in parentheses, A standard deviation. Anthracene solubility parameter from ref 42. Pyrene and perylene solubility parameters were calculated from Small's group molar attraction constants (42) using molar volumes listed in Table I. Units of 6 and 6, are in (cal/mL)1/2. V , = 18.07 mL/mol, R = 1.98 cal/mol.deg, T = 296 K. 1 mM Ca2+and 0.1 M total ionic strength. e ND, not determined. Q
on the binding of anthracene by humic and fulvic acid were similar to those observed for pyrene. From the results above, several trends of the influence of solution chemistry on PAH binding by Suwannee River humic substances become apparent. Among these are (1) increases in pH and/or ionic strength generally decrease PAH binding, (2) 1 mM Ca2+at pH 4 typically has little effect on PAH binding, while at pH 7 and 10 it generally increases binding, and (3) the magnitude of aqueous chemistry effects depends on the relative sizes of the solute and humic substance. The pH trends observed here agree with the findings of Carter and Suffet (8) and Kile and Chiou (20). Also, the influence of Ca2+at pH 7 and 10 is similar to that observed at pH 8.3 (8). The effects of bivalent cations at lower pH values have not been previously reported. The ionic strength trends observed here, however, are opposite from those of Carter and Suffet (8). The one major exception to the typical ionic strength trends of this study is the binding of anthracene by humic acid at pH 4. This system was the only one to show an increase in KO,with increasing NaCl concentration. We 966
Environ. Sci. Technol., Vol. 27, No. 5, 1993
believe this observation, as well as the results of Traina et al. (13), relates to the third trend listed above (vide infra). Values of the carbon-normalized partition coefficients for the three PAH solutes with humic and fulvic acid are listed in Tables I1 and 111,respectively. These values can be compared with those predicted by correlations based on the KO,and aqueous solubility of each PAH (Table I). Experimental values for anthracene ranged from 13 000 to 32 000; the average value from this range is close to predicted KO,values. Reported KO,values for anthracene binding to dissolved humic materials have ranged from 16 OOO to 64 OOO when determined by fluorescence quench, 9000 to 74 000 when determined by reverseing ( l l ) from phase chromatography (9, I l ) , and from 16 000 to 500 000 when determined by dialysis (9, 10). Experimental KO,values for pyrene ranged from 6000 to 32000 and were from a factor of 2 to an order of magnitude lower than predicted values. Reported KO, values for pyrene determined by fluorescence quenching have ranged from 29 000 to 550 000 for dissolved humic
materials (11,12)and from 52 000 to 111000 for marine porewater colloids (37). Typical KO,values for sediments and soils have been reported to range from 19000 to 170 000 ( 1 , 2 , 37). ExperimentalKocvaluesfor perylene ranged from 60 000 to 1 100 000; these values range from that predicted by the aqueous solubility correlation to more than an order of magnitude lower. Only one previousstudy has examined perylene binding by DOM. Backhus and Gschwend (14) investigated the binding of perylene by Aldrich humic acid,bovine serum albumin, and groundwater colloids and have reported KO,values ranging from 40 000 to 2 400 000 when determined by fluorescence quenching and from 14 000 to 800 000 when determined by reverse-phase chromatography. Examination of a Partitioning (Dissolution) Model. The binding of HOCs by DOM has been previously modeled as dissolution of a solute into an organic polymer phase (37-39). Chiou et al. (40) assumed components of soil humic material were amorphous polymeric substances and used Flory-Huggins theory to estimate solute activity in the proposed humic phase. The compatibility of partition coefficients with Flory-Huggins models has in fact been used as an argument for a dissolution mechanism for the binding of HOCs by organic material (39, 41). However, these models have not been rigorously tested for their consistency in characterizing humic material in well-defined systems. A Flory-Hugginsmodel was applied to our data in order to test its ability to characterize the organic environments of humic and fulvic acid for each PAH under different aqueous chemistry conditions. Characterization of the humic/fulvic environment is obtained from values of the total solubility parameter (6,); Le., increasingly polar DOM will have higher values of 6,. The equation relating partition coefficients to the FloryHuggins interaction parameter (x)is (30, 37) log KO,= log
vw + log - log p , - log f,, Vi 1 + x - Vi/V,
(4) 2.303 where yiwis the PAH aqueous activity coefficient; V,, Vi, and V, are the molar volumes of water, PAH, and humic material, respectively; and pp is the humic material density (-1.5 and -1.4 g/mL for humic and fulvic acid, respectively) (30). The value of x is comprised of both entropy and enthalpy components. The entropy term, xs, is determined empirically and has the value 0.34 (38),while the enthalpy term, Xh, is frequently estimated from the Hildebrand-Scatchard equation (42): = (Vi/RT)(ai - 6pI2 (5) where 6i is the PAH solubility parameter. As observed in eqs 4 and 5, increasing values of 6, are generally indicative of decreases in partition coefficients (37). Total solubility parameters were calculated for humic and fulvic acid for the varying aqueous chemistry conditions examined (Tables I1 and 111). Calculated values for 6, were observed to range from 11.6 to 14.1 (cal/mL)1/2 and fall into the general range observed previously (3740). However, the solubility parameters fail to consistently characterize the humic environment. For identical aqueous chemistry conditions,the values of 6, calculated using different PAH solutes should be similar for a particular humic substance. Each solute, however, gives very difxh
ferent values for 6, as shown in Tables I1 and 111. The Flory-Huggins model at best only qualitatively characterizes the relative polarity of the two humic substances. For example, under identical conditions the solubility parameter for fulvic acid is always larger than that for humic acid, in agreement with the properties of these humic and fulvic acids (31). The inability of the Flory-Huggins model to consistently characterize the hydrophobic environment of the humic substances used in this study probably results because the assumptions utilized in deriving the model do not properly portray Suwannee River humic material. The formation of “microscopic organic environments” similar to micelles would require the presence of long alkyl chains in the humic structure such as those present in many surfactants or possibly the high molecular weight huminkerogen structures such as those originally proposed by Freeman and Cheung (43) to describe natural organic material in sediments. The likelihood of forming a substantial humic (i.e., organic) phase with the relatively polar and low molecular weight Suwannee River humic material is small. Also, the assumption that a humic phase similar to micelles is present requires that partition coefficients for HOCs increase with increasing solution ionic strength (7). However, partition coefficients for the humic substances used in this study generally decreased with increasing ionic strength. Implications for a Molecular Level Mechanism. In order to gain insight into the molecular level association mechanism, the effects of aqueous chemistry on PAH partitioning between water and humic substances must be understood. Changes in aqueous chemistry can affect both PAH compounds and DOM present in the system, as well as their interactions with the solvent and each other. Examples of these changes are the “salting out” of PAHs with increasing salinity and the coagulationof humic acid with decreasing pH. It would be beneficial, therefore, if the aqueous chemistry effects could be quantified separately for each PAH solute and for the humic substances. An example of the separation of aqueous chemistry effects is shown in Figure 4, where partition coefficients have been plotted first as functions of pH and then as functions of ionic strength. Because pH does not affect the solubility of perylene, the dependence of KO,on pH at fixed ionic strength can be attributed entirely to changes in humic acid. At constant pH, however, differences in KO,with varying ionic strength result from variations in perylene’s aqueous activity coefficient, as well as changing humic properties. A better depiction of the binding reactions observed here can be made usingthe molecular level picture of HOCDOM associations proposed by Schnitzer and Khan (44). They suggested that humic material consists of a broken network of poorly condensed aromatic rings with appreciable numbers of disordered aliphatic or alicyclic structures attached around an aromatic core. The building blocks of the humic structure were thought to be phenolic and benzenecarboxylicacids which are joined by hydrogen bonds and van der Waals forces to form relatively stable aggregates. Humic material was thought to have an open structure which allows for a considerable number of voids of varying dimensions, and the arrangement was expected to be sensitive to changes in pH, salt concentration, and valence of cations. Schnitzer and Khan (44)hypothesized that the voids could trap organic compounds, provided the compounds had the proper molecular sizes to fit into Environ. Sci. Technot., Vol. 27, No. 5, 1993
067
the holes and the environment in the voids was conducive to association with the organic compounds. These voids, or hydrophobic cavities, would thus be sites where van der Waals bonding forces between organic molecules and DOM could occur as part of the overall hydrophobic interaction. The effects of aqueous chemistry on PAH binding can be easily explained using Schnitzer and Khan’s (44)model. At constant ionic strength, increases in pH progressively deprotonate humicifulvic molecules but have little effect on the aqueous activity coefficient of nonpolar, nonionogenic organic molecules (Le., y i w = constant). Deprotonation increases the polarity of humic material, alters its structure, and decreases its ability to bind PAHs. At fixed pH values, increases in electrolyte concentrations “salt out” PAH molecules (i.e., yiw increases) but also compress humic substances by electrostriction (19). This compression decreases the size of voids in dissolved humic material into which PAHs can partition. Suwannee River humic material appears to be affected more than PAHs by electrolyte concentrations, because partition coefficients generally decreased with increasing ionic strength. Therefore, although aqueous activity coefficients of PAHs are the most important parameter overall for determining order of magnitude binding, they are not as sensitive to solution chemistry changes as are the conformation and polarity of dissolved humic materials. The one exception to the trends described above is observed for the anthracene-humic acid system. In NaCl solutions at pH 4, an increase in the ionic strength resulted in increased anthracene binding (Figure 6). Calculations show that the salting effect can account for only 5 and 16% of the increase in KO,as the NaCl concentration increases from 1 to 10 mM and from 1 mM to 0.1 M, respectively. The anomaly is interesting because anthracene was the smallest PAH examined (Table I) while humic acid was the larger DOM investigated; the singular behavior of anthracene may be indicative of the size of voids in the humic acid structure. Traina et al. (13) observed very little effect of solution chemistry on the binding of naphthalene by a soil organic matter extract. Because naphthalene is even smaller than anthracene, its access to hydrophobic cavities in the organic matter may not have been hindered. Observations of steric effects and specific interactions have been reported previously by other researchers, but not in the context of Schnitzer and Khan’s (44)proposed model for HOC-organic matter interactions. Karickhoff (36) reported that studies utilizing large HOC (Le., fivering or larger) found dramatic reductions, as much as an order of magnitude, in KO,for particular soils/sediments. Studies utilizing smaller molecules (1-4 rings) showed less than a factor of 2 variability in KO,for the same sediments/ soils. Karickhoff (36) suggested these observations resulted because sorbent availability depended on sorbate size. The inability of the solute to gain access to the sorbent thus resulted in a steric inhibition of sorption. McCarthy et al. (45) investigated the association of benzo[a]pyrene (BaP) with 11 different DOM samples and correlated the values of KO,with structural and chemical properties of the organic matter. They observed a good correlation between the size and hydrophobic acid content of DOM from different sources and the KO,for binding BaP. The affinity of DOM for binding BaP was related not only to the presence of sufficient hydrophobic moiety to promote partitioning but also to the presence 968
Environ. Sci. Technol., Vol. 27, No. 5, 1993
of acidic functional groups. McCarthy et al. (45) hypothesized that the acidic functional groups acted to open the structure of the DOM and that this more open structure contributed to binding of the HOC by making hydrophobic cavities within the DOM structure more accessible to the solute. Gauthier et al. (12) investigated the binding of pyrene by 14 different humic and fulvic acids and observed that the magnitude of the KO,values correlated strongly with three independent measures of the degree of aromaticity in the humic material. They postulated that the binding mechanism was dominated by van der Waals interactions and that the increased affinity resulted because aromatic bonds in humic substances would increase the polarizability of the material, thereby increasing the strength of the van der Waals attraction forces between pyrene and humic material. The ideas proposed by Schnitzer and Khan (44) are remarkably similar to current theories and concepts in the relatively new field of molecular recognition (e.g., refs 46 and 47). This subdiscipline of modern chemistry is concerned with studying “host-guest” interactions in aqueous and organic media. A major focus of the work in this field is the design and synthesis of water-soluble macrocycles with well-defined hydrophobic binding sites which are able to bind a variety of molecular guests of varying size, shape, and polarity. These hosts have been shown to tightly bind highly insoluble guests such as anthracene and pyrene. Major binding forces which have been identified in molecular recognition studies in aqueous media are hydrophobic interactions, donoriacceptor *-stacking interactions, and ion-dipole attractions (47).
Conclusions The binding of anthracene, pyrene, and perylene by Suwannee River humic and fulvic acid is influenced by solution chemistry. In NaCl solutions, the amount of a PAH bound by a particular humic material generally decreased with increasing pH (constant ionic strength) and also generally decreased with increasing ionic strength (fixed pH). The presence of Ca2+yielded mixed results; at pH 4, it typically had little effect on the amount of PAH bound relative to a NaCl solution. A t pH 7 and 10, however, the presence of Ca2+ generally increased the amount of bound PAH relative to NaCl solutions. The one major deviation from these observed trends was the binding of anthracene by humic acid at pH 4; for this particular system, the association was observed to increase with increasing NaCl concentrations. Results obtained from the binding studies were analyzed with a model utilizing Flory-Huggins theory. With this model, however, each PAH solute gave different values of the humic or fulvic solubility parameter for identical solution conditions. The inability of the model to characterize consistently the hydrophobic environment of humic substances results because the assumptions utilized in the derivation do not properly depict Suwannee River humic material. The molecular level representation of HOC-DOM associations proposed by Schnitzer and Khan (44) better describes the trends of the binding reactions observedhere. Their picture of humic material as an open structure with hydrophobic cavities is plausible. The dimensions and hydrophobicity of the voids in these structures would be sensitive to variations in pH, salt concentration, and valence of cations. Because the properties of the voids
change with varying solution chemistry, a subsequent change in the ability to bind HOC would occur, much like the trends observed in this study.
Acknowledgments We greatly appreciate the helpful discussions with Phil Gschwend, Steve Eisenreich, Deb Backhus, and Yu-Ping Chin throughout this investigation, and we also appreciate the insight provided by Beth Carraway and Amy Hoffman on fluorescence quenching and energy transfer. Finally, we wish to acknowledge one anonymous reviewer for the constructive comments on the organization and content of the manuscript. This work was supported by grants from Andrew W. Mellon Foundation, William and Flora Hewlett Foundation, Smith and Louise Lee Memorial Endowment, San Francisco Foundation (Switzer Foundation Environmental Fellowship), and American Water Works Association (Larson Aquatic Research Support Ph.D. Scholarship).
Literature Cited Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Water Res. 1979, 13, 241. Means, J. C.; Wood, S. G.; Hassett, J. J.; Banwart, W. L. Environ. Sci. Technol. 1980, 14, 1524. Chiou, C. T.; Peters, L. J.; Freed, V. H. Science 1979,206, 831. Chiou, C. T.; Peters, L. J.; Freed, V. H. Science 1981,213, 683. Voice, T. C.; Weber, W. J., Jr. Water Res. 1983, 17, 1433. Tanford, C. The HydrophobicEffect:Formation ofMicelles and Biological Membranes,2nd ed.; Wiley: New York, 1980. Israelachvili, J. N. Intermolecular and Surface Forces; Academic Press: London, 1985. Carter, C. W.; Suffet, I. H. Environ. Sci. Technol. 1982,16, 735. Landrum, P. F.; Nihart, S. R.; Eadie, B. J.; Gardner, W. S. Environ. Sci. Technol. 1984, 18, 187. McCarthy, J. F.; Jiminez, B. D. Environ. Sci. Technol. 1985, 18, 187. Gauthier, T. D.; Shane, E. C.; Guerin, W. F.; Seitz, W. R.; Grant, C. L. Environ. Sci. Technol. 1986, 20, 1162. Gauthier, T. D.; Seitz, W. R.; Grant, C. L. Environ. Sci. Technol. 1987, 21, 243. Traina, S. J.; Spontak, D. A.; Logan, T. J. J . Environ. Qual. 1989, 18, 221. Backhus, D. A.; Gschwend, P. M. Environ. Sci. Technol. 1990,24, 1214. Thurman, E. M. Organic Geochemistryof Natural Waters; Martinus Nijhoff/Dr. W. Junk Publishers: Dordrecht, The Netherlands, 1985. Schnitzer, M.; Kodama, H. Geoderma 1975, 13, 279. Ghosh, K.; Schnitzer, M. Soil Sci. 1980, 129, 266. Cornel, P. K.; Summers, R. S.; Roberts, P. V. J . Colloid Interface Sci. 1986, 110, 149. Leenheer, J. A.;Brown, P. A.; Noyes, T. I. In Aquatic Humic Substances: Influence on Fate and Treatment of Pollutants; Suffet, I. H., MacCarthy, P., Eds.; ACS Symposium Series 219; American Chemical Society: Washington, DC, 1989; pp 25-39.
(20) Kile, D. E.; Chiou, C. T. In Aquatic Humic Substances: Influence on Fate and Treatment of Pollutants; Suffet, I. H., MacCarthy, P., Eds.; ACS Symposium Series 219, American ChemicalSociety: Washington, DC, 1989;pp 131157. (21) Voice, T. C.; Rice, C. P.; Weber, W. J., Jr. Environ. Sci. Technol. 1983, 17, 513. (22) Gschwend, P. M.; Wu, S. C. Environ. Sci. Technol. 1985, 19, 90. (23) Voice, T. C.; Weber, W. J., Jr. Enuiron. Sci. Technol. 1985, 19, 789. (24) Morel, F. M. M.; Gschwend, P. M. In Aquatic Surface Chemistry;Stumm, W., Ed.; Wiley-Interscience: New York, 1987; p 405. (25) Davis, H. G.; Gottlieb, S. Fuel 1962, 8, 37. (26) Mackay, D.; Shiu, W. Y. J. Chem. Eng. Data 1977,22,399. (27) Yalkowsky, S. H.; Valvani, S. C. J . Chem. Eng. Data 1979, 24, 127. (28) Parker, C. A. Photoluminescence of Solutions; Elsevier Publishing Co.: Amsterdam, 1968. (29) Lloyd, J. B. In Standards in Fluorescence; Miller, J. N., Ed.; Chapman Hall: New York, 1981; pp 33-35. (30) Schlautman, M. A. Ph.D. Dissertation, California Institute of Technology, 1992. (31) Averett, R. C.; Leenheer, J. A,; McKnight, D. M.; Thorn, K. A. Humic Substances in the Suwannee River, Georgia:
Interactions, Properties, and Proposed Structures, U.S. Geological Survey Open-File Report 87-577, 1989. (32) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic Molecules, 2nd ed.; Academic Press: New York, 1971. (33) Schulman, S. G. Fluorescence and Phosphorescence
Spectroscopy: Physicochemical Principles and Practice; Pergamon Press: Oxford, 1977. (34) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Plenum Press: New York, 1983. (35) Turro, N. J. Modern Molecular Photochemistry; The BenjarniniCummings Publishing Co., Inc.: Menlo Park, CA, 1978. (36) Karickhoff, S. W. J . Hydraul. Eng. 1984, 110, 707. (37) Chin, Y. P.; Gschwend, P. M. Environ. Sci. Technol. 1992, 26, 1621. (38) Chin, Y. P.; Weber, W. J., Jr. Environ. Sci. Technol. 1989, 23, 978. (39) Chin, Y. P.; Weber, W. J., Jr.; Chiou, C. T. In Organic Substances and Sediments in Water;Baker, R. A., Ed.; Lewis Publishers: Chelsea, MI, 1991; Vol. 1, p 251. (40) Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Environ. Sci. Technol. 1983, 17, 227. (41) Rutherford, D. W.; Chiou, C. T.; Kile, D, E. Environ. Sci. Technol. 1992, 26, 336. (42) Barton, A. F. M. Handbook of Solubility Parameter and Other CohesionParameters;CRC Press: Boca Raton, FL, 1983. (43) Freeman, P. H.; Cheung, L. S. Science 1981, 214, 790. (44) Schnitzer, M.; Khan, S. U. Humic Substances in the Environment; Marcel Dekker: New York, 1972. (45) McCarthy, J. F.; Robertson, L. E.; Burrus, L. W. Chemosphere 1989,19,1911. (46) Petti, M. A.; Shepodd, T. J.; Barrans, R. E., Jr.; Dougherty, D. A. J . Am. Chem. SOC.1988,110, 6825. (47) Shepodd, T. J.; Petti, M. A.; Dougherty, D. A. J . Am. Chem. Soc. 1988, 110, 1983.
Received f o r review October 2, 1992. Revised manuscript received February I , 1993. Accepted February 2, 1993.
Environ. Sci. Technol., Vol. 27, No. 5, 1993 069
